&EPA
            United States
            Environmental Protection
            Agency
           Industrial Environmental Research
           Laboratory
           Research Triangle Park NC 2771 1
EPA-600/9-84-025a
November 1984
            Research and Dev.
Fourth
Symposium on the
Transfer and
Utilization of
Particulate Control
Technology:

Volume I.
Fabric Filtration


-------
                                         EPA-600/9-84-025a
                                         Noveirber 1984
               FOUKIH SYMPOSIUM ON THE
             TRANSFER AND UTILIZATION OF
            PARTICULATE CONTROL TECHNOLOGY:
             VOLUME I.  FABRIC FILTRATION
                     Compiled by:

F. P. Venditti, J. A. Armstrong, and Michael D. Durham

              Denver Research Institute
                   P. 0. Box 10127
               Denver, Colorado  80210
               Grant Number: CR 809301
                   Project Officer

                    Dale L. Harmon
  Office of Environmental Engineering and Technology
     Industrial Environmental Research Laboratory
    Research Triangle Park, North Carolina   27711
     INDUSTRIAL ENVIRONMENTAL RESEARCH LABORATORY
          OFFICE OF RESEARCH AND DEVELOPMENT
        U. S. ENVIRONMENTAL PROTECTION AGENCY
    RESEARCH TRIANGLE PARK, NORTH CAROLINA  27711

-------
     This document has been reviewed in accordance with U.S.
Environmental Protection Agency policy and approved for publication.
Mention of trade names or contnercial products does not constitute
endorsement or recommendation for use.
                                     11

-------
                              ABSTRACT
     The papers  in  these three volumes of Proceedings were presented
at the Fourth Symposium on  the Transfer and Utilization of Paticulate
Control Technology held in Houston, Texas during 11 October through 14
October 1982, sponsored by the Particulate Technology Branch of the
Industrial Environmental Research Laboratory of the Environmental
Protection Agency and coordinated by the  Denver Research Institute of
the University of Denver.

     The purpose of the symposium was to bring together researchers,
manufacturers, users, government  agencies, educators  and  students to
discuss new  technology and to provide  an effective means  for  the
transfer of this technology  out of the laboratories and into the hands
of the users.

     The  three major categories of  control technologies  -
electrostatic precipitators, scrubbers, and  fabric filters  - were  the
major concern of the symposium.  These technologies were discussed
from the perspectives  of economics; new technical advancements in
science and engineering; and applications.   Several papers  dealt with
combinations of devices and technologies, leading  to a  concept of
using a systems approach to particulate control rather than device
control.   Additional topic areas  included  novel control devices,  high
temperature/high pressure  applications,  fugitive emissions,
measurement techniques,  and  economics and cost analysis.

     Each volume  of these proceedings contains a  set of  related
session topics to provide easy access to a unified technology area.

     Since the  spirit  and style of the panel  discussion  are  not
reproducible  in  print,  the initial remarks presented by the panelists
have been included in  the volume to which their input to the panel
pertained,   in  the interest  of providing unified technological
organization.
                               111

-------
                               CONTENTS
VOLUME I   - CONTENTS	     V
VOLUME II  - CONTENTS	viii
VOLUME III - CONTENTS	   xii

          Section A - Fabric Filters:  Fundamentals
THEORY OF THE TEMPORAL DEVELOPMENT OF PRESSURE DROP
ACROSS A FABRIC FILTER DURING CAKE INITIATION  . . .
  E.A. Samuel
PULSE JET FILTRATION THEORY - A STATE-OF-THE-ART ASSESSMENT. . .    22
  R. Dennis, L.S. Hovis

LABORATORY TECHNIQUES FOR DEVELOPING PULSE JET COLLECTORS. ...    37
  R.R. Banks, J.T. Foster

OFF-LINE PULSE-JET CLEANING SYSTEM	    48
  T.C. Sunter
           Section B - Fabric Filters;  Measurement Techniques


FIELD EVALUATION OF THE DRAG OF INDIVIDUAL FILTER BAGS	    62
  W.T. Grubb,  R.R. Banks

A DUAL-DETECTOR BETA-PARTICLE BACKSCATTER GAUGE FOR MEASURING
DUST CAKE  THICKNESS ON OPERATING BAG FILTER AND ESP UNITS.  ...    77
  R.P. Gardner,  R.P. Donovan, L.S. Hovis

MIT FLEX ENDURANCE TESTS AT ELEVATED TEMPERATURE	    91
  J.T. Foster, W.T. Grubb

THE ONE-POINT  IN-SITU CALIBRATION METHOD  FOR USING A BETA-
PARTICLE BACKSCATTER GAUGE FOR  CONTINUOUSLY MEASURING DUST
CAKE THICKNESS ON OPERATING BAG FILTER AND ESP UNITS	   107
  R.P. Gardner,  R.P. Donovan, L.S. Hovis


           Section C - Fabric Filters:  Coal Fired Boilers
 PULSE-JET FABRIC FILTER EXPERIENCE USING NON-GLASS
 MEDIA AT AIR TO CLOTH RATIOS  OF 5 TO 1 ON A PULVERIZED
 COAL FIRED BOILER
   G. Pearson, D.D.  Capps
 START-UP AND OPERATION OF A FABRIC FILTER CONTROLLING
 PARnCULATE EMISSIONS FROM A 250 MW PULVERIZED COAL-FIRED
 BOILER .............................    132
   C.B. Barranger,  N.  Spence, J.  Saibini
                                  v

-------
PERFORMANCE OF A 10 MW FABRIC FILTER PILOT PLANT AND
COMPARISON TO FULL-SCALE UNITS	   148
  W.B. Smith, K.M. Gushing, R.C. Carr

THE DESIGN, INSTALLATION, AND INITIAL OPERATION OF THE W.H.
SAMMIS PLANT, UNIT 3 FABRIC FILTER	   164
  D.R. Ross, J.R. Howard, R.M. Golightley

RESULTS FROM THE FABRIC FILTER EVALUATION PROGRAM AT
COYOTE UNIT #1	   179
  H.J. Peters, A.A. Reisinger, W.T. Grubb, M. Lewis

BAGHOUSE PERFORMANCE AND ASH CHARACTERIZATION AT THE
ARAPAHOE POWER STATION	   192
  R.S. Dahlin, D.R. Sears, G.P. Green

AN EVALUATION OF FULL-SCALE FABRIC FILTERS ON UTILITY
BOILERS	   210
  J.W. Richardson, J.D. McKenna, J.C. Mycock

STATUS OF SPS INVESTIGATION OF HARRINGTON STATION UNIT 2
FABRIC FILTER SYSTEM	   226
  R. Chambers, D. Harmon

UPDATE OF SPS PILOT BAGHOUSE OPERATION	   239
  R. Chambers, S. Kunka, D. Harmon

THE USE OF SONIC AIR HORNS AS AN ASSIST TO REVERSE AIR
CLEANING OF A FABRIC FILTER DUST COLLECTOR	   255
  A. Menard, R.M. Richards
          Section D - Fabric ^liters:  Electrostatic Enhancement
ELECTROSTATIC STIMULATION OF REVERSE-AIR-CLEANED
FABRIC FILTERS	   287
  D.A. Furlong, G.P. Greiner, D.W. VanOsdell, L.S. Hovis

ELECTRICAL STIMULATION OF FABRIC FILTRATION: ENHANCEMENT BY
PARTICLE PRECHARGING 	   303
  G.E.R. Lamb, R. Jones, W. Lee

ESFF AS A FIELD EFFECT	   316
  L.S. Hovis, G.H. Ramsey, R.P. Donovan

ELECTRICAL ENHANCEMENT OF FABRIC FILTRATION:  PRECHARGING
VS. BAG ELECTRODES	   327
  R.P. Donovan, L.S. Hovis, G.H. Ramsey

PERMEABILITY OF DUST CAKES COLLECTED UNDER THE INFLUENCE OF
AN ELECTRIC FIELD	   342
  D.W. VanOsdell, R.P. Donovan, D.A. Furlong, L.S. Hovis
                                 VI

-------
          Section E - Fabric Filters:  Practical Considerations
HIGH VELOCITY FABRIC FILTRATION FOR INDUSTRIAL COAL-FIRED
BOILERS	   357
  G.P. Greiner, s- Delaney, L.S. Hovis

OPTIMIZING THE LOCATION OF ANTI-COLLAPSE RINGS IN FABRIC
BAGS   ;	   382
  J. F4isgrove

PULSE JET ON-LINE CLEANING FILTER FOR FLY ASH	   420
  W.G. Wellan

TOP INLET VERSUS BOTTOM INLET BAGHOUSE DESIGN  	   431
  R.M. Jensen

UPGRADE OF FLY ASH COLLECTION CAPABILITY AT THE CROMBY
STATION	   446
  T.J. Ingram, R.J. Biese, R.O. Jacob

HIGH SULFUR FUEL, FABRIC FILTER STARTUP EXPERIENCE  	   460
  P. Hanson, L. Adair, R.N. Roop, R.B. Moyer

FUNDAMENTAL STRATEGIES FOR CLEANING REVERSE AIR BAGHOUSES.  .  .  .   482
  M. Ketchuck, M.A. Walsh, O.F. Fortune/
  M.L. Miller, M. Whittlesey,
          Section F - Dry Scrubbers


DESIGN CONSIDERATIONS FOR BAGHOUSE - DRY S02 SCRUBBER
SYSTEMS	    494
  O.F. Fortune, R.L. Miller

RESULTS OF BAGHOUSE AND FABRIC TESTING AT  RIVERSIDE	    506
  H.W. Spencer  III, Y.J. Chen, M.T. Quach

REACTIVITY OF FLY ASHES IN A SPRAY DRYER/FABRIC FILTER FGD
PILOT PLANT	    521
  W.T. Davis, R.E. Pudelek,  G.D.  Reed


          Section G - Plenary Session
 FABRIC FILTRATION - AS  IT WAS,  HAS BEEN,  IS  NOW
 AND SHALL BE	    536
   E.R.  Frederick

 AUTHOR INDEX 	    551
                                  VII

-------
                              VOLUME II

                     ELECTROSTATIC PRECIPITATION


          Section A - Industrial Applications
MODELING OF WET BOTTOM AGITATOR SYSTEMS FOR ELECTROSTATIC
PRECIPITATORS ON RECOVERY BOILERS  	     1
  M.A. Sandell, R.R. Crynack

DESIGN AND PERFORMANCE OF ELECTROSTATIC PRECIPITATORS
UTILIZING A NEW RIGID DISCHARGE ELECTRODE DESIGN	    17
  G.R. Gawreluk, R.L. Bump

DEVELOPMENT AND EVALUATION OF NEW PRECIPITATOR EMITTER
ELECTRODE	    35
  R.L. Adams, P. Gelfand,

INDUSTRIAL APPLICATIONS OF TWO STAGE TUBULAR ELECTROSTATIC
PRECIPITATORS  	    51
  H. Surati, M.R. Beltran
          Section B - Advanced Technology


PILOT DEMONSTRATION TWO-STAGE ESP TEST RESULTS	    65
  P. Vann Bush, D.H. Pontius

EVALUATION OF PRECHARGERS FOR TWO-STAGE ELECTROSTATIC
PRECIPITATORS  	    84
  G. Rinard, D. Rugg, M. Durham

INITIAL EXPERIMENTS WITH AN ELECTRON BEAM PRECIPITATOR TEST
SYSTEM	    96
  W.C. Finney, R.H. Davis/ J.S. Clements, E.G. Trexler,
  J.S. Halow, 0. Tokunaga

EXPERIMENTS WITH WIDE DUCTS IN ELECTROSTATIC PRECIPITATORS ...   Ill
  E. Weber

A RECONCILIATION:  WIDE VERSUS NARROW SPACED COLLECTING
PLATES FOR PRECIPITATORS	   126
  D.G. Puttick

PULSE CORONA AS ION SOURCE AND ITS BEHAVIORS IN MDNOPOLAR
CURRENT EMISSION	   139
  S. Masuda, Y. Shishikui
                               viii

-------
VOLUME II CONTENTS  (Cont.)

          Section C - Fundamentals

A NEW CORRECTION METHOD OF MIGRATION VELOCITY IN DEUT3CH
EFFICIENCY EQUATION FOR CONVERSION OF ELECTROSTATIC PRECIPITATOR
SIZING FROM A PILOT-SCALE TO FULL-SCALE	   154
  F. Isahaya

DISTORTION OF PULSE VOLTAGE WAVE FORM ON CORONA WIRES DUE TO
CORONA DISCHARGE    	   169
  S. Masuda, H. Nakatani

ELECTROSTATIC PRECIPITATOR ANALYSIS AND SYNTHESIS  	   184
  T. Chiang, T.W. Lugar

COMPUTER MODEL USE FOR PRECIPITATOR SIZING	   194
  G.W. Driggers, A.A. Arstikaitis, L.A. Hawkins

IMPROVEMENTS IN THE EPA/SRI ESP PERFORMANCE MODEL  	   204
  M.G. Faulkner, R.B.Mosley, J.R. McDonald, L.E. Sparks

NUMERICAL SIMULATION OF THE EFFECTS OF VELOCITY FLUCTUATIONS
ON THE ELECTROSTATIC PRECIPITATOR PERFORMANCE	   218
  E.A. Samuel

CORONA - INDUCED TURBULENCE  	   230
  M. Mitchner, G.L. Leonard, S.A. Self

VELOCITY AND TURBULENCE FIELDS IN NEGATIVE CORONA
WIRE-PLATE PRECIPITATOR   	   243
  H.P. Thomsen, P.S. Larsen, E.M. Christensen,
  J.V. Christiansen

THE EFFECT OF TURBULENCE ON ELECTROSTATIC PRECIPITATOR
PERFORMANCE	   261
  D.E. Stock

FACTORS LEADING TO ELECTRICAL BREAKDOWN OF RESISTIVE DUST
LAYERS AND SUSTAINED BACK CORONA	   271
  P.A. Lawless, L.E. Sparks

ELECTRICAL BREAKDOWN OF PARTICULATE LAYERS 	   288
  G.B. Moslehi, S.A. Self

ELECTROMECHANICS OF PARTICULATE LAYERS 	   306
  G.B. Moslehi, S.A. Self

LATERAL PROPAGATION OF BACK CORONA IN TWIN-EIiECTRODE TYPE
PRECIPITATORS  	   322
  S. Masuda, T. Itagaki

FIRST MEASUREMENTS OF AEROSOL PARTICLE CHARGING
BY FREE ELECTRONS	   337
  J.L. DuBard, M.G. Faulkner, L.E. Sparks

                                 ix

-------
VOLUME II CONTENTS (Cont.)

          Section D - Operation & Maintenance
GAS FLOW DISTRIBUTION MODEL TESTING	   349
  D.R. Cook, J.M. Ebrey, D. Novogoratz

AIR FLOW MDDEL STUDIES	   369
  L.H. Bradley

COLLECTING ELECTRODE RAPPING DESIGNED FOR HIGH EFFICIENCY
ELECTRIC UTILITY BOILER ELECTROSTATIC PRECIPITATORS  	   384
  A. Russell-Jones, A.P. Baylis

ELECTROSTATIC PRECIPITATOR AND FABRIC FILTER OPERATING AND
MAINTENANCE EXPERIENCE	   401
  P.R. Goldbrunner, W. Piulle
          Section E - Conditioning
ECONOMICAL FLY ASH COLLECTION BY FLUE GAS CONDITIONING	   416
  E.L. Coe, Jr.

EXPERIENCES AT DETROIT EDISON COMPANY WITH DECLINING
PERFORMANCE OF SULFUR TRIOXIDE FLUE GAS CONDITIONING
EQUIPMENT	   430
  L.A. Kasik, W.A. Rugenstein, J.L. Gibbs
ESP CONDITIONING WITH AMMONIA AT TFffi M3NRQE POWER PLANT OF
DETROIT EDISON COMPANY	   444
  E.B. Dismukes, J.P. Gooch, G.H. Marchant, Jr.

FLY ASH CHEMISTRY INDICES FOR RESISTIVITY AND EFFECTS ON
ELECTROSTATIC PRECIPITATOR DESIGN MID PERFORMANCE  	   459
  H.J. Hall
          Section F - Control Systems
A NEW ENERGIZATION METHOD FOR ELECTROSTATIC PRECIPITATORS
MITSUBISHI INTERMITTENT ENERGIZATION SYSTEM	   474
  T. Ando, N. Tachibana, Y. Matsumoto

SOME MEASURED CHARACTERISTICS OF AN ELECTROSTATIC
PRECIPITATOR OBTAINED USING A MICROCOMPUTER CONTROLLER	   489
  M.J. Duffy, T.S. Ng, Z. Herceg, K.L. McLean

ELECTROSTATIC PRECIPITATOR ENERGIZATION AND CONTROL SYSTEMS  . .   499
  K.M. Bradburn, K. Darby

-------
VOLUME II CONTENTS  (Cont.)

APPLYING MODULAR MICROCOMPUTER CONTROL ELEMENTS IN A
PRECIPITATOR CONTROL SYSTEM  	   521
  I.M. Wexler
          Section G - Plenary Session
THE CURRENT STATUS, FUTURE DIRECTIONS, AND ECONOMIC
CONDITIONS IN THE APPLICATION OF ESP1 S	   534
  S. Oglesby

AUTHOR INDEX 	   539
                                 XI

-------
                              VOLUME III

                ECONOMICS, MECHANICAL COLLECTORS,
            COAL CHARACTERISTICS, INHALABLE PARTIOJLATES,
                  ADVANCED ENERGY AND NOVEL DEVICES

          Keynote Address

PARTICULATE CONTROL TECHNOLOGY AND WHERE IT IS GOING 	   XV
  K.E. Yeager
          Section A - Economic Comparisons
A COMPARISON OF A BAGHOUSE VS. ESP'S WITH AND WITHOUT GAS
CONDITIONING FOR LOW SULFUR COAL APPLICATIONS	     1
  W.H. Cole

APPLICATION OF THE BUBBLE CONCEPT TO FUEL BURNING SOURCES AT
A NAVAL INDUSTRIAL COMPLEX 	    12
  C.S Thompson
          Section B - Mechanical Collectors
CYCLONE PERFORMANCE:  A COMPARISON OF THEORY WITH
EXPERIMENTS	    26
  J.A. Dirgo, D. Leith

HIGH FLOW CYCLONE DEVELOPMENT	    41
  W.B. Giles

CYCLONE SCALING EXPERIMENTS  	    53
  W.B. Giles

TEST METHODS AND EVALUATION OF MIST ELIMINATOR CARRYOVER ....    66
  V. Boscak, A. Demian
          Section C - Coal Characterization
FILTRATION CHARACTERISTICS OF FLY ASHES FROM VARIOUS COAL
PRODUCING REGIONS	    81
  J.A. Dirgo, M.A. Grant, R. Dennis, L.S. Hovis

FLY ASH FROM TEXAS LIGNITE AND WESTERN SUBBITUMINOUS COAL:
A COMPARATIVE CHARACTERIZATION	    97
  D.R. Sears, S.A. Benson, D.P. McCollor, S.J. Miller

USE OF FUEL DATABANKS FOR THE EFFECTIVE DESIGN OF STEAM
GENERATORS AND AQC EQUIPMENT	   114
  N.W. Frisch, T.P. Dorchak

                                 xii

-------
VOLUME III CONTENTS  (Cont.)

          Section D - Inhalable Particulate Matter
DEVELOPMENT OF INHALABLE PARTICULATE (IP) EMISSION FACTORS ...   131
  D.L. Harmon

INHALABLE PARTICULATE MATTER RESEARCH COMPLETED BY
GCA/TECHNOLOGY DIVISION	   141
  S. Gronberg

RESULTS OF TESTING FOR INHALABLE PARTICULATE MATTER AT
MIDWEST RESEARCH INSTITUTE 	   154
  K. Wiloox, F. Bergman/ J. Kinsey, T. Cuscino

INHALABLE PARTICULATE EMISSION FACTORS TEST PROGRAMS 	   166
  J.W. Davison,

CHARACTERIZATION OF PARTICULATE EMISSION FACTORS FOR
INDUSTRIAL PAVED AND UNPAVED ROADS	   183
  C. Cowherd, Jr., J.P. Reider, P.'J. Englehart

CONDENSIBLE EMISSIONS MEASUREMENTS IN THE INHALABLE
PARTICULATE PROGRAM	   198
  A.D. Williamson, J.D. McCain
          Section E - Advanced Energy Applications
GAS CLEANING AND ENERGY RECOVERY FOR PRESSURIZED FLUIDIZED
BED COMBUSTION	   211
  A. Brinkmann, P.M. Kutemeyer

DEMONSTRATION OF THE FEASIBILITY OF A MAGNETICALLY
STABILIZED BED FOR THE REMOVAL OF PARTICULATE AND ALKALI ....   226
  L.P. Golan, J.L. Goodwin, E.S. Matulevicius

TEST RESULTS OF A HIGH TEMPERATURE, HIGH PRESSURE
ELECTROSTATIC PRECIPITATOR	   241
  D. Rugg, G. Rinard, J. Armstrong, T. Yamamoto, M. Durham

COAL-ASH DEPOSITION IN A HIGH TEMPERATURE CYCLONE	   256
  K.C. Tsao, A. Rehmat, D.M. Mason

DUST FILTRATION USING CERAMIC FIBER FILTER MEDIA — A STATE-
OF-THE-ART SUMMARY —	   271
  R. Chang, J. Sawyer, W. Kuby, M. Shackleton,
  O.J. Tassicker, S. Drenker

HIGH TEMPERATURE AND PRESSURE PARTICULATE FILTERS FOR FLUID
BED COMBUSTION	   282
  D.F. Ciliberti, T.E. Lippert, O.J. Tassicker, S. Drenker


                                xiii

-------
VOLUME III CONTENTS  (Cont.)

MOVING BED-CERAMIC FILTER FOR HIGH EFFICIENCY PARTIOJLATE
AND ALKALI VAPOR RE?DVAL AT HIGH TEMPERATURE AND PRESSURE   ...   300
  D. Stelman, A.L. Kohl, C.A. Trilling

TESTING AND VERIFICATION OF GRANULAR BED FILTERS FOR REMOVAL
OF PAKTICULATES AND ALKALIS	   318
  T.E. Lippert, D.F. Ciliberti, R. O'Rourke

BAGHOUSE OPERATION IN GEORGETOWN UNIVERSITY COAL-FIRED,
FLUIDIZED-BED BOILER PLANT, VJASHINGTON, D.C	   335
  V. Buck, D. Suhre
          Section F - Novel Devices
PARTICLE CAPTURE MECHANISMS ON SINGLE FIBERS IN THE PRESENCE
OF ELECTROSTATIC FIELDS	   347
  M.A. Ranade, F.L. Chen, D.S. Ensor, L.S. Hovis

PILOT DEMONSTRATION OF PARTICULATE REMOVAL USING A CHARGED
FILTER BED   	   362
  P.H. Sorenson

PILOT DEMONSTRATION OF MAGNETIC FILTRATION WITH CONTINUOUS
MEDIA REGENERATION	   370
  C.E. Ball, D.W. Coy
          Section G — Plenary Session
NOVEL PARTICULATE CONTROL TECHNOLOGY 	   386
  S. Masuda

AUTHOR INDEX 	   406
                                 xiv

-------
             THEORY OF THE TEMPORAL DEVELOPMENT OF PRESSURE DROP
                ACROSS A FABRIC FILTER DURING CAKE INITIATION

              by:  Eric A. Samuel
                  General Electric Environmental Services, Inc.
                  Lebanon, Pennsylvania 17042
                                  ABSTRACT
                                      ^
     The time variation of the pressure drop across a fabric filter character-
istic of cake initiation in a new fabric as well as in a seasoned fabric is
derived on the basis of the Carman-Kozeny hypothesis, which is well known for
its prediction of the linear growth of pressure drop with time characteristic
of cake filtration.  The non-linear growth of pressure drop during cake
initiation in a new fabric is associated with the filling of the voids within
the fabric prior to cake establishment.   The non-linear growth of pressure
drop during cake initiation in a seasoned fabric is associated with non-
uniform dust loading on the fabric.  The non-uniform loading may either be
an initial condition following cleaning, or be induced by inertial or exter-
nal forces.  Experimental evidence in support of the theory is presented.

                                INTRODUCTION
     The time development of the pressure drop across a fabric filter, through
which a constant flow is maintained, is usually discussed in terms of two
separate processes: (1) cake initiation, and (2) cake establishment.  The
time variation of the pressure drop across the filter is in general non-
linear during cake initiation, but is invariably linear after cake establish-
ment, when the dust concentration in the gas is constant.  The linear time
variation of the filter pressure drop after cake establishment is consistent
with Darcy's law(l), which expresses the proportionality of the filter
pressure drop to the average superficial flow velocity, v, across the filter
(air-to-cloth ratio) and the areal dust loading, W:

                  AP = K2 W v                                            (1)

The proportionality constant, K?, is called the coefficient of specific
resistance.  There are two well known theories by which the coefficient of
specific resistance of a filter cake may be calculated purely from knowledge

-------
of the size distribution of the particles comprising the filter cake, and the
filter cake porosity.  The two models are: (1) the Carman-Kozeny model, and
(2) the Happel model.  Each model is based on a different notion as to the
origin of the flow resistance of the filter cake.  The Carman-Kozeny model(2)
is based on regarding the filter cake as an assembly of capillaries.  In
contrast, the Happel model(3) is based on regarding the filter cake as an
assembly of independent spherical cells, each cell containing a single dust
particle surrounded by a fluid-filled envelope.

     The Carman-Kozeny and Happel models are widely used to characterize
filter cakes across which the pressure drop increases linearly with time
(filtration after filter cake establishment).  There are currently no theories
which deal with the non-linear time variation of the filter pressure drop
characteristic of cake initiation.  The present paper seeks to fill this need
by proposing a theory, based on the Carman-Kozeny hypothesis, for predicting
the shape of the filter pressure drop vs. time curve during cake initiation,
including the transition to the linear regime characteristic of cake filtra-
tion.  The time evolution of the filter pressure drop for a new fabric will
be discussed separately from that for a seasoned fabric.  The central idea
relating to the development of the pressure drop across a new filter is the
filling of the pores and voids within the fabric as a necessary precursor to
the establishment of a filter cake.  Non-uniform loading, either as an initial
condition following cleaning or as a result of weaving flaws in the fabric;
or induced by the action of inertial or external forces continously or
discontinously present during filtration, is the central idea which will be
seen to be consistent with anomalous time development of the filter pressure
drop in seasoned fabrics.  Three different, experimentally observed causes of
non-uniform dust loadings on seasoned fabrics will be discussed: (1) non-
uniform dust loading as an initial condition arising from non-uniform clean-
ing of the filter cake, (2) pinhole formation on weaving flaws in the fabric,
and (3) influence on inertial forces or external forces (such as a corona
field).  The theoretical predictions of the filter pressure drop vs. time
curves will be compared with corresponding experimental curves to demonstrate
areas of agreement with experiment, and also to point out areas in which
refinements to the proposed theory could be made.

     The objective guiding the present investigation into understanding the
shape of the filter pressure drop vs. time curve is in providing a fundament-
al theoretical basis for improving experimental characterizations of fabric
filtration.  The following benefits are expected to accrue from such a theory:
(1) Characterization of fabrics using flow measurements in addition to the
presently used methods based on weave characteristics, fabric strength tests,
and physical properties.
(2) Characterization of fabric-particle combinations with respect to pressure
drop and cleanability using flow measurements.
(3) Evaluation of fabric finishes and fabric coatings based on flow measure-
ments in the laboratory without recourse to extensive pilot testing.
(4) Using filter pressure drop vs. time curve as a diagnostic tool to detect
abnormal operating conditions in fabric filter operation.
(5) Characterizing anomalous effects in fabric filtration, such as the effect
of electric fields.

-------
     Better criteria for fabric selection, better understanding of fabric
filter operation and better evaluation of schemes for fabric filter enhance-
ments are the ultimate goals of the present work.

                                   THEORY
NEW FABRIC

Introduction

     The area of applicability of the conventional Carman-Kozeny model to
only cake filtration may be extended to include cake initiation also by cons-
idering the filling up of the pores within the fabric as a necessary condit-
ion for the establishment of a filter cake.  In the Carman-Kozeny model, the
filtering medium is regarded as an assembly of identical capillaries of
diameter, dc, and length, Lc, such that: (1) the total internal surface area
of the capillaries is equal to the total surface area of the particles comp-
rising the filter medium, and (2) the total internal volume of the capillar-
ies is equal to the volume of the voids 'in the filter medium.  The flow rate
through the capillaries is governed by the Hagen-Poisuelle law(A).  When
considering cake initiation within the Carman-Kozeny model, the surface area
of the fabrics is added to the particulate surface area.  Taking the instant
when flow is first initiated through the new fabric as time t = 0, cake
initiation will proceed up to a characteristic time,T^, which is the time
required to entirely fill up the pores within the fabric.  The filtration
taking place after time t = T^ may be characterized as cake filtration.  The
theory for the time development of the filter pressure drop during cake ini-
tiation is as follows.

Cake Initiation (t < Tf)

     Assuming that there are n(t) Carman-Kozeny capillaries, each of effect-
ive length, L , and diameter, dc, in an area, A, of the fabric characterized
by a porosity, £f, and a thickness, Lf, formed by N(t) particles of known
size distribution uniformly dispersed throughout the voids within the fabric,
the Carman-Kozeny hypothesis may be expressed by the following equations:

              n TT d., L~ = A Af + N IT d     (surface area relationship)    (2)
                   C  L.      I


              n (TT/4) d^ L  = A Lf e       (volume relationship)          (3)
                       \—  C      •*-


              A Lf (1 - e) = A Vf + N (TT/6) d3   (porosity definition)   (4)


              A Lf (1 - ef) = A Vf               (porosity definition)   (5)


              A nf u c T  = A p  Lf ef (1 - e )  (Tf definition)         (6)

-------
                                       n
             A ri^ u c t = N p   (rr/6) dj        (mass  conservation)         (7)


             Q/n =  [ TT d4 / (128 y LC) ] AP    (Hagen-Poisuelle  equation  (8)
                                                for one  capillary)


             L  = k LJ    (correction for capillary length  due to          (9)
                          tortous path through  capillaries)


             Q = A u      (air-to-cloth ratio  definition)                 (10)

In the above equations, Q is the total gas flow rate through area  A of  the
fabric, u the air-to-cloth ratio, AP the pressure drop  across the  fabric
filter, r|f the average collection efficiency  of the  fabric during  cake  init-
iation, Pp the true particulate density, and  c   the  porosity of the particul-
ate packing within the fabric voids at time t = T£ (at  the end  of  the cake
initiation period).  The parameters, Af and Vf,  are  measurable  characteristics
of the new fabric representing  the fiber surface area and  fiber volume  per
unit cloth area.  The effective capillary length, Lc, will be expected  to be
larger than the fabric thickness, Lf,  because of the winding path  a gas
molecule may take through the filter.  If F(z)  dz is the probability of
finding particles whose diameters lie  between z and  z+dz in the particle
distribution, the average diameters used in Equations (2),  (4), and (7) are
defined as follows:


             dq = / F(z) z^ dz  / / F(z) dz                               (11)

               ?      ^                           2      3
The averages, d  and dj are sometimes written as dg  and dv where ds and d
are called the surface diameter and volume diameter  respectively.   The
diameter, cP/d"^, is sometimes written  as dsv  and called the surface to  volume
diameter.

     Equations (2) to (10) may be solved for n,  N, dc,  e and AP.   The solu-
tions are:
             N = 6 A nf u c t  /  (TT p  d)                                (12)


               = [ 6 A Lf / (TT d3) ]  (ef -  e)                            (13)


             n = 4 A e / (TT d£ k)                                        (14)


             dc = 4 Lf e /  [  (6/dsv)  (ef -  e) +  (Af/Lf)  ]                (15)

-------
            AP = 32 k2 y Lf u /  (d2  e)                                   (16)


             e = ef - nf u c t /  (pp Lf)                                 (17)


               = ef [ 1 - (1 - Ep) t/Tf ]                                (18)


            Tf = pp Lf £f (1 ~ ep> / (nf U c)                            (19)

Equation (18) shows that the effective porosity,  £,  decreases  linearly with
time up to time t = Tf.  Equation  (16) coupled with  Equations  (18)  and (19)
yield the filter pressure drop as  a  function of  time during  cake  initiation.
The present model for cake initiation also shows  the effective capillary
diameter, dc, and the effective number of capillaries,  n,  are  also  functions
of time.  The important feature in the model presented  above is the relevance
of the fabric characteristics, Af  and Lf,, to the  time development of the
filter pressure drop during cake  initiation.  The fabric  thickness,  Lf,  may
be measured with the aid of a micrometer screw gauge.   Equations  (15)  and  (16)
applied for the case of a new fabric without dust particles, suggest one
method of obtaining the fiber area per unit fabric area (Af).   In the limit
£ -* £f, the above two Equations give:


             AP = Sf u                                                   (20)
             Sf = 2 k2 y A2 /  (e-| Lf)                                    (21)

Equation (20) indicates a linear relationship between  the pressure  drop
across a new fabric (without dust particles) and  the air-to-cloth ratio.   The
proportionality constant, Sf,  is usually called the fabric drag.  A linear
relationship between AP and u  is also observed for a new fabric, for small u,
as predicted by Equation (20).  The parameter, Af, could be  obtained from the
slope, Sf,  of the AP vs. u graph, through Equation (21), provided £f and  k
are known.   The porosity, e^ , may be obtained from knowledge of the fabric
thickness,  Lf, the mass per unit area of the fabric, nif, and the density,  Pf,
of the material comprising the fabric fibers, through  the relationship:
             mf = ef Lf (1 ~ £f}                                         (22)

The correction factor, k, has been found to be in the range  /3/2  to  v/5/2  for
cake filtration(5) .  The correction factor, k, applicable  to fabrics in the
absence of dust particles has not been investigated.  Because of  the relative-
ly open pore structure of fabrics in comparison with that  of a filter cake
which is formed from particles having large surface area,  the correction
factor, k, for a new fabric is likely to be closer to unity.  The uncertainty

-------
in obtaining A£ from the filter drag, Sf, is directly  related  to  the uncert-
ainty with which k is known.  Measurement of the  surface  area  of  a  fabric  by
gas adsorption, usin the B.E.T. Equation, is an alternative method  of
obtaining Af (6) .

Cake Filtration (t > Tf)

     The Carman- Kozeny hypothesis corresponding to  cake filtration  including
the effect of the fabric, the pores within which  are filled with  dust part-
icles during the earlier cake initiation period,  may be expressed by the
following equations:

                                      o"
             n TT dc LC = A Af + N IT d     (surface area relationship)    (23)
             n (TT d/A) Lc = A Lf cf cp + A  (L -  Lf ) ep                  (24)
                                         (volume  relationship)


   A L (1 - e) = A Lf  (1 - ef ) + A Lf ef (1  - ep) + A  (L -  Lf )  (1  -  e£)  (25)
                                         (overall porosity  definition)


             A n  u c  t / p  = N  (TT/6) d     (mass conservation)          (26)


             A (L - Lf) (1 - t£) = u c (t -  Tf) / pp                     (27)
                                         (cake porosity definition)


             L  = k L   (correction for capillary length)                (28)

In Equations (23) to (28), (L - L^) is the thickness of the filter cake,  of
porosity, e', growing on the fabric substrate, the pores within which have
been filled with particles to a porosity, e  , during cake initiation.  The
porosities, c  and e1, are not necessarily equal.  The solutions  to  Equations
(23) to  (28) tor the filter pressure drop, AP, and the Carman-Kozeny capillary
characteristics, n, N, and dc, are of identical form to those  corresponding
to cake initiation summarized in Equations  (12) to (16).  The  overall poros-
ity, e, is given for the case of cake filtration  by:


                -.      C' ' £f £p
                   P         U C  (t - Tf)
                         1 +	—T.	rr~
                             P~ LJT  (l - e )
                              pi       D

with the characteristic time for  cake initiation, Tf, given by  Equation  (19).
The effective porosity, e, as given by Equation  (29), also varies with time,

-------
as in the case of cake initiation, starting with the value £ = Cf £  at time
t = if and assymptotically approaching the value e = e* for large times
t » Tf.   The parameter, ric, in Equation (26) represents the average collect-
ion efficiency (by mass) of the filter during cake filtration.  In general,
r|  is higher than r|f.  The pressure drop, AP, for cake filtration is:

            AP = 32 k2 y L u / (d2 £)                                   (30)

with the capillary diameter, dc,  given by Equation (15).  The porosity
applicable in Equation (30) is given by Equation (29).

                                              PK
     The coefficient of specific resistance, K^ , as predicted by the Carman-
Kozeny model may be obtained by recasting Equation (30) in a form consistent
with Darcy's law, and recognizing that the areal dust loading is given by:

             W = L pp (1 - e)                                           (31)

The desired result is:


           K^K = 32 k2 u /  [ e (1 - e) d2 ]                             (32)


In the limit of t » Tf, the Carman-Kozeny specific resistance approaches  the
constant value:


           K£K = 4 k2 (1 - ££) /  (Tp e^3)                               (33)


where T  is the average particle relaxation time, given by,


            TP = dsv Pp /  (18 y)                                        (34)

Equation  (33) represents the conventional Craman-Kozeny result widely used  in
studies of cake filtration.

SEASONED FABRIC

Non-Uniform  Cleaning

     The salient features of the filter pressure drop vs. time characteristic
of a seasoned fabric may be obtained by considering a simple distribution
for the initial non-uniform areal loading resulting from the filter  cleaning
operation.  The filter is regarded as being seasoned in the sense that the
pores within the fabric are assumed to be saturated with particles,  so that
its operation may be described by cake filtration theory.  Filter cleaning
is assumed to result in two distinct areas, each area characterized  by a
constant areal loading.  The above assumption is broadly consistent  with
experimental observations using reverse air -cleaning in an abnormally operat-

-------
ing fabric filter, which leaves patches of relatively clean fabric  (lower
areal loading, W^) on a filter cake (having the higher loading, W2).  The
simple non-uniform initial loading described above is illustrated in Figure
2 (a).  When constant volume flow through the cleaned filter is resumed, the
initial velocity through the relatively cleaner patches will be higher than
the average air-to-cloth ratio.  Because of the above fact, the initial rate
of increase of the filter pressure drop will be higher than that when the
fabric loading is uniform over the entire fabric  (in which case the flow
velocity over the entire cake is the air-to-cloth ratio) .  As filtration
continues, the rate of increase of the filter pressure drop assymptotically
approaches the value characteristic of uniform areal loading over the whole
fabric.  The time development of the filter pressure drop across a seasoned
fabric with a two-segment areal loading as an initial condition may be des-
cribed by the following equations  (see Figure 2 (a)):
AP = K2 vx(t) [ Wj + c  /  v^t1) dt1 ]
                                                                        (35)
AP - K2 v2(t) t W2 + c  /  v2(t') dt' ]
Al vl(t) + A2 V2(t) =
                                u
                                                                        (36)


                                                                        (37)
Equations  (35) and  (36) are expressions of Darcy's law for cake filtration
through areas A. and A2 respectively.  Equation  (37) expresses the constancy
of the total flow rate through the filter at all times.  These Equations do
not appear to be amenable to an exact analytical solution.  An approximate
analytical solution to the above Equations may be obtained using perturbation
theory, as follows:
   AP = K_ F.  [ W, + c u t - c a. T   (1 - e~t/Tc) ]  (u - a, e~t/Tc) +
         211               1  c                       I
        c u t - c
                                                     (u - «  e~t/Tc)     (38)
where
                                                                        (39)
                                                                        (40)
U ~ Vl(0) =
                         V F2 U
                                       W
                                                         F2 V
                 ~ V
                                         W
                                              F2 V
                                                                        (42)

-------
                       - W2) / [ c (dj + C*2) ]
                 = (Fj W2 + F2 W1) / (c u)                              (43)


The time dependent velocities through the areas AI and A  are given by:


  v (t) = \ { (u - a  e~t/Tc) + 	AP	rTT~ }  (44)
                                K2 [ W  + c u t - c a. TC (1 - e    c)  ]
  v (t) = ~ { (u- a  e't/Tc) + 	AP	-J-—  }  (45)
                                K2  [ W2 + c u t - c a2 TC  (1 - e~C/LC) ]


As will be seen below, the approximate analytical solutions to the  time
development of cake filtration initiated from a non-uniform loading, as
summarized in Equations  (38) to  (45), compare reasonably well with  the exact
solution, which may be obtained by  using numerical techniques.  The time
constant, Tc, can be interpreted as being a characteristic time for the
transition of the AP vs. t development from its initially high rate of
increase to the linear rate of increase corresponding to uniform areal load-
ing over the entire filter.  The Equations (43) give a simple relationship for
estimating this characteristic time from the initial areal loading  distribu-
tion, the air-to—cloth ratio and the concentration.

Cake Filtration in the Presence of Pinholes

     Experiments on cake filtration have indicated the presence of  pinholes
in the filter cakes formed on fabrics with weaving flaws.  The pinholes were
always centered around the weaving flaws; not every weaving flaw, however,
had an accompanying pinhole.  Pinhole formation appears to be also  character-
ized by sudden jumps in the filter pressure drop.  These pressure drop spurts
may by due to the dynamics of pinhole rearrangement during filtration, and to
pinhole filling in low velocity regions.  Figure 2(b) shows a simplified
representation of a filter cake containing pinholes.  If the filter cake is
assumed to comprise m identical pinholes, each of diameter, d , and if K£ is
the coefficient of specific resistance corresponding to those areas (A^) of
the filter cake containing no pinholes, the Equations governing the AP vs. t
development are as follows:


              AP = K~ (n  c v. t) v,                                   (46)
                    •^   p    I      i


              AP = K  v0                                               (47)
                    P  2

with          K  = 32 y  (c v, t) /  [ d2 p  (1 - e) ]                   (48)
               p            1         m  p

-------
              Al Vl + A2 V2 =  (A1 + V U                               (A9)
with          A  = m (TT d)/4                                           (50)
As in previous discussions, Equation  (46) is an  expression of Darcy's  law
for filtration through area AI .  Equations  (47)  and  (48) assume  that the  flow
rate through the pinholes is governed by the Hagen-Poisuelle law(4).
Equations (49) and (50) indicate incompressible  flow at a constant  flow rate
through the filter.  In these  Equations, r\  is the collection efficiency  of
the filter cake having no pinholes  (area A,), e  the  porosity of  this filter
cake, and p  the true density  of the  particulate matter comprising  the filter
cake in area Aj.  Since the length  of the capillary  corresponding to a pin-
hole is assumed to be the same as the thickness  of the filter cake  in  area Aj,
the velocities Vi and V2 through area Aj^ and through the pinholes will be
independent of time.  The solutions to Equations (46) to  (50) are:
              AP
                          K2
                           (K2/V
                   u /  [ Fj_ +  (KJ/K  ) F2  ]                              (52)
                    (K2/Kp)
                   u /  [  (Kp/K2)
with          K2 = np K2                                                (55)
                                                                        (56)
              F2 =  1 - F2 = A2  /  (Aj + A2)                              (57)

Equation  (51) shows that the  pressure  drop  acro-s  a filter cake  in the
presence  of pinholes will increase  linearly with time,  under  the present
assumptions.  The slope of the  AP vs.  t  line,  however,  will be different  from
the value, K' u^ c, expected  in the absence of pinholes.   The slope will  be
lower than the value in the absence of pinholes when K  <  K',  as is invariably
the case.
                                      10

-------
Filtration in the Presence of Inertial Forces

     A simple representation of a filter cake formed in the presence of
inertial forces acting on the particles is shown in Figure 2(c).  In this
representation, the filter cake is composed of two discrete areas with diff-
erent coefficients of specific resistance.  The difference in the mean size
of the particles comprising the two discrete areas is attributable to iner-
tial forces in the presence of which larger particles are diverted from the
flow streamlines.  If f,  and f_ are the mass fractions of particles collected
in areas AI and A_ resulting in coefficients of specific resistance, K_, and
K__, and flow velocities, v, and v~ for the two discrete areas, the Equations
governing the time development of the filter pressure drop are:
AP = K21 (cl
                t)
                                                                        (58)
AP
                t)
                                                                        (59)
Al Vl
        A2 V2 =
                      V U
                                                                        (60)
or
                           = u
                                                                        (61)
with F. anf F? as defined in earlier discussions.  In  the above Equations,
the concentrations, c^ and c.^* i-n tne 8as streams entering the areas A^  and
A? are related to the concentration, c, of all particles entering  the  entire
filter, by:
c. A. v. = c f. u
                                                                        (62)
c  A  v  = c f  u
                                                                        (63)
Equations (58) to (63) may be solved for the pressure drop, AP, and  the
velocities,  v1 and v«, as:
AP =
                                                                        (64)
                                     K21) ] u
                                                                        (65)
and similarly for v_.  The term between brackets in Equation  (64)  is  the
effective coefficient of specific resistance for the whole filter, which  is
seen to depend on Aj, A2, fj_, f2> K21 and K22'  Ifc is  informative  to  rewrite
                                      11

-------
Equation (64) in terms of the loadings, W1 and W , on the areas AI and A~, as:


             AP = K21 WL u / [ Fj + F2 (K21 WX)/(K22 W2) ]             (66)



with         W. = (f. c u t) / A.                                      (67)
             W2 = (f2 c u t) / A2                                      (68)

Equation (66) shows that the filter pressure drop will be larger than the
value, K.W.u, corresponding to the entire filter being characterized by a
coefficient of specific resistance equal to K_. and an areal loading W., if
K -W^ > K-.W., i.e., if the filter drag for area ^ is greater than for area
Aj.  The areal loadings W, and W2 are equal if f^/Aj = ^2^2'  For t*ie aDOVe
case, the filter pressure drop will be larger than ^^W^u if ^22 > ^21'  ^e
Carman-Kozeny model for cake filtration shows that the coefficient of specific
resistance, is in general, inversely proportional to d~,v, where dgv is the
surface-to-volume diameter of the particle distribution comprising the filter
cake  (see Equations (33) and (34)).  Because of the above relationship, the
filter pressure drop in the presence of inertial forces could be significantly
different from that in the absence of inertial forces, in which case the
entire filter cake can be characterized by a single coefficient of specific
resistance.

Filtration in the Presence of External Forces

     The theory in this section will be confined to those cases in which the
velocity of the particles in the presence of the external forces is much
greater than the flow velocity.  In this case, the dust particles will be
diverted away from the flow streamlines along the same paths.  The profile of
dust deposition, £(x,t), at time t, x being the lateral distance along the
filter cake with respect to the flow direction (see Figure 2(d)), normalized
by the total mass of dust in the filter cake, will under the above assumption
be independent of time, i.e.,


              C(x,t) /  (c u t) = MX)                                  (69)

The influence of an externally induced non-uniform distribution on the filter
pressure drop has also been discussed elsewhere(7).  The Equations governing
the time variation of the filter pressure drop for a filter of unit dimens-
ions  (unit filter area) are:


              AP = K~ tKx) cut v(x)                                  (70)

               1
              /  v(x) dx = u                                           (71)
             0

                                      12

-------
with the equation defining the normalization of the function describing the
dust deposition profile:,
                1
               / Kx) dx = 1                                          (72)
              0

Solving Equations (70), (71), and (72) yields the expressions for the filter
pressure drop and the velocity across a non-uniform filter cake formed in the
presence of external forces, as:

                                   1
              AP = K2 ( c u t / [ / dxmx) ]  } u                     (73)
                                 0
                                      1
              v(x) =  [ uMx) ] [ I/ / dxWx) 1                       (74)
                                    0

The case of H'(x) = 1 yields the equations characteristic of uniform dust
deposition discussed earlier.  Because of the normalization condition imposed
by Equation  (72), vF(x') £  1, for every x.'In this case, / dx/f(x) ^ 1, the
equality applying to the  case of uniform dust deposition, for which case the
pressure drop is the greatest.  The pressure drop in the case of non-uniform
dust deposition is always lower than  that for the case of uniform deposition,
for the same areal loading of dust and the same coefficient of specific resi-
stance.

               EXPERIMENTAL RESULTS AND COMPARISON WITH THEORY
NEW FABRIC

     Figure  3(a) shows  the observed time development of  the filter pressure
drop for a new woven glass fabric.  The measurement was  carried out  in  a
bench-scale  apparatus comprising a flow straightening section, an expansion
nozzle, a filter holder and an automatic flow controller.  A maximum cloth
area of 0.33 ft2 could be attached to the filter holder  in the above setup.
The sigmoidal shape of the experimental AP vs. t curve indicates the presence
of the two transitions associated with cake initiation and cake establishment.
Figure 3(b)  is the theoretically predicted AP vs.  t curve obtained from
Equations (16) and (30).  Experimentally measured  values for the parameters
u» c, Lf, £f, p , and dgv were used in generating  the curve shown in Figure
3(b).  The air-to-cloth ratio, u, was obtained using a calibrated orifice.
The dust concentration was measured by weighing the dust collected in the
filter over a known filtration time at constant u.  The  cloth thickness was
measured with the aid of a micrometer screw gauge.  The  clean fabric porosity,
£f, was measured with the aid of Equation(22).  The surface to volume diameter,
dsv, was obtained from a measurement of the particle size distribution  of a
filter cake  sample performed with the aid of a "Micromeritics Sedigraph"
instrument which is based on the sedimentation technique.  The parameters,
Af, £ , and  e' were systematically adjusted until  reasonable agreement  with
the experimental curve of Figure 3(a) was obtained.  The theoretical curve of


                                      13

-------
Figure 3(b) also displays two transitions.   The first transition,  in which the
rate of increase of the filter pressure drop progressively increases up to a
maximum value, is characteristic of cake initiation.  The pres within the
fabric are progressively filled during this stage.  Figure (4) indicates the
linear decrease of the effective porosity theoretically predicted  for this
stage.  After reaching a minimum value corresponding to complete filling of
the fabric pores, the effective porosity increses with time during the period
of cake establishment.  As filtration progresses in this stage the effective
porosity assymptotically approaches the value characteristic of only the
filter cake formed on the fabric substrate.  Both the experimental and theor-
etical AP vs. t  curves show a linear rate of increase of the filter pressure
drop when sufficiently long time has elapsed since cake establishment.  This
linear rate is characteristic of cake filtration.  The largest area of dis-
agreement between experiment and theory occurs in the time range over which
the transition between cake establishment and cake filtration takes place.
The sharp transition theoretically predicted for this case is based on cake
initiation ending sharply at time t = Tf and cake establishment commencing
immediately thereafter.  Experimentally, this transition is seen to be much
smoother than theoretically predicted.  Simultaneous deposition of the dust
within the fabric pores and on the fabric surface is an effect not included in
the theory which will lead to a smoother transition.

SEASONED FABRIC

Non-Uniform Cleaning

     Figure 5 compares the approximate solution to the time development of the
pressure drop across a filter with non-uniform initial loadings, as given by
Equation (38), with the exact numerical solution for Equations  (35),  (36), and
(37).  Figure 6 is a comparison between the approximate and exact solutions to
the time development of the superficial velocity through the  filter areas with
relatively high and relatively low dust loadings.  The error  between  the
approximate and exact solutions for the pressure drop time development, which
is the largest in the neighborhood of the knee which is characteristic of the
transition between cake initiation and cake establishment, is seen to increase
with decreasing fraction, Aj, of the relatively clean area following  cleaning.
For fractions, A^ < 0.5, the agreement between the approximate  and exact
solutions to both pressure drop and velocity time development appears to be
reasonably good.

     The shapes of experimental AP vs. t curves are  similar to  those  of  the
theoretical  curves shown in Figure 5.  However, a detailed comparison of
experlmantal data with  theory appears to require more  than a  two-segment model
treated in this paper.  Experiments currently  in progress attempt to  obtain
data relevant to comparison with a multi-sement model.  Results of this
comparison may be reported at a future date.

Filtration in the Presence of Pinholes

      Measurements have been carried  out in bench-scale and pilot-scale
 apparatus  with  fly ash having a log-normal size distribution characterized
 by a mean  diameter and a standard  deviation.   These experiments appear to

                                       14

-------
indicate an increasing density of pinholes (number of pinholes per unit area)
with incraesing air-to-cloth ratio.   The highest specific coefficient resist-
ance characteristic of a given fly ash is obtained with closely woven fabric
on which the filter cake showed no pinhole formation.  Every pinhole has
been found to be associated with a flaw in the fabric weave.  However, not
every fabric weave flaw is associated with a pinhole.  The rate of change of
filter pressure drop in the presence of pinholes is linear, as in the case of
no pinhole formation, but with an anomalously low slope of AP/At. This lower
slope is due to the diversion of an appreciable amount of the filter flow
through the pinholes.  Slight fluctuations observed in the AP vs. t curve may
be due to the dynamic process of pinhole movement across the fabric and of
pinhole filling in the low velocity regions.

Filtration in the Presence of External Forces

     Experimental results on fabric filter pressure drop in the presence of
external forces have been reported elsewhere(7).  The observed AP reduction
in the presence of electric fields has compared favorably with the theory
presented in this paper, which is based on an externally induced non-uniform
dust deposition pattern.  The central phenomenon which reduces the filter
pressure drop of a non-uniform cake relative to that of a uniform cake (both
having the same total mass of dust) is the diversion of a large part of the
flow through the filter areas with the lower dust thickness while the exter-
nal force diverts the particle deposition away from the flow streamlines.

                                 CONCLUSION

     The central theme throughout this paper has been the effect of non-
uniform dust deposition patterns on the fabric filter pressure drop.  Diff-
erent phenomena are shown to result in different dust deposition patterns.
Most observed physical phenomena observed in fabric filtration have been
shown to be amenable to simple physical description.  It is hoped that this
paper will initiate experimental and theoretical investigations which will
improve the models presented herein, and eventually improve our understanding
of fabric filtration.

                               ACKNOWLEDGEMENT

     Thanks are due to Tom Bechtel for his encouragement of the work presented
herein.

                                 REFERENCES

1.  Billings, C. E. and Wilder, J.  Handbook of Fabric Filter Technology.
    PB-200-648.  National Technical Information Service, 1970.

2.  Carman, P. C.  Flow of Gases Through Porous Media.  Academic Press,
    New York, 1956.

3.  Happel, J.  Viscous Flow in Multi-Particle Systems: Slow Motion of
    Fluids Relative to Beds of Spherical Particles. A. I. Ch. E. Journal.
    4: 197 - 201, 1958.

                                      15

-------
4.  Rudnick, S. N. and First, M. W.  Specific Resistance (IG?) of Filter Dust
    Cakes:  Comaprison of Theory and Experiments.  In:  Proceedings of the
    Third Symposium on Fabric Filters for Particulate Collection.
    EPA-600/7-78-087.  U. S. Environmental Protection Agency, 1978.
    p 251 - 288.

5.  Dalla Valle, J. M.  Micromeritics, The Technology of Fine Particles.
    Pitman, New York, 1948.

6.  Brunauer, S.  Physical Adsorption.  Princeton University Press,
    Princeton, New Jersey, 1943.

7.  Chiang, T. K., Samuel, E. A. and Wolpert, K. E.  Theoretical Aspects of
    Pressure Drop Reduction in a Fabric Filter with Charged Particles.
    In:  Proceedings of the Third Symposium on the Transfer and Utilization
    of Particulate Control Technology, Volume III.  Particulate Control
    Devices.  EPA-600/9-82-005c.  U. S. Environmental Protection Agency,
    Research Triangle Park, N. C. 27711, July 1982.  p 250 - 260.
                                      16

-------
           FILTER  CAKE
                                                              FABRIC
Figure 1.  Representation of a fabric substrate.  In the present work, the
           filling up of the voids within the fabric substrate is referred
           to as cake initiation.  Complete filling of the substrate voids,
           referred to as cake establishment, is assumed to be a necessary
           precursor to cake filtration, which is characterized by a linear
           increase with time of the filter pressure drop.
      V}(T)
  v2(r)
           K22
        (a)
                            FILTER
                             CAKE
                                             FABRIC
Figure 2.  Non-uniform dust loadings on seasoned fabrics arising from
           different physical conditions:  (a)  improperly cleaned filter
           resulting in areas of low dust loading and areas of high dust
           loading, (b) filter with pinholes formed on fabric weave flaws,
           (c) filter with externally induced non-uniform dust deposition
           pattern; the external force draws the particles away from the
           flow streamlines and Hows the flow to pass through the filter
           areas with relatively low dust loadings.
                                      17

-------
    *f.  A e
    U  1.7
1.0 —
CO
       0.5 —
       13.0
                        EXPERIMENT
                        V = 5 FT./M'N.
                        C = 5 fiR/FT^
                        TEMPERATURE! AMBIENT
                        FABRIC:  TEFLON COATED
                               FIBERGLASS (WOVEN)
                        PARTICLES:  FLYASH
                               .   (DSV - 20.9 WM)
4      6

 TIME.  (MINUTES)
r-p'
  8
                                              10
                                              12
   Figure  3(a).
            Experimental  AP vs.  t curve
            obtained at ambient  tempera-
            ture  in a bench-scale patch-
            work  filter apparatus.
                                                                          2.0 —
                                                                    1 c
                                                                    1.9 -
                                                                          1.0
                                              IX
                                              LU
                                              rj 0.5-
                                                                   a.o—T
                                                                                           AF - 10-
                                                                 f
                                                                                                    THEORY
                                                                                            AF = EFFECTIVE FIBER SURFACE AREA
                                                                                                 PER UNIT CLOTH AREA
                                                                                      463

                                                                                      TIME,  (MINUTES)
                                                                                                                 10
                                                                                  12
                                          Figure  3(b).
                                             Theoretical  P vs.  t curve
                                             based on Equations  (29) and (30)
                                             corresponding to  the experimental
                                             conditions  of Figure 3(a).

-------
      0.45
      0.40
  in
  a
  oe.
  a
      0.35
      0.25
                                                  = 0,43
                                                       THEORY
I   I  f  I   I  I   I  I  I  I  I  Tl  I  I   I
       2463

                  TIME, (MINUTES)
                                                       I   IT
                                                                       12
Figure 4.   Theoretical prediction of the effective porosity during  cake
           initiation, cake establishment, and cake filtration,  based on
           Equation  (29).  The minimum porosity is reached at  the instant
           when all  the pores within the fabric are filled up  (cake establi-
           shment).  After this time the porosity increases and  assymptotic-
           ally approaches the porosity of the filter cake.
                                     19

-------
N)
O
           4 ^—
           3 —
           2 —
                K2  = 20
                R! = 0.90 j  Rg = ®-l®
                	  - EXACT
                  :•-  - RPPROXIMRTE
   TT'T~p~l I I | 'I I I I  | I I  I I | 'I  I I I  | I I T
0     5     10     15    20     25      30

             TIME- 
-------
N)
           10 —
            8 —
            6
              <2  = 20
              AI = 0.90 5  Ag = 3-10
              	  - EXACT
                «  - APPROXIMATE ANALYTICAL
I I  I | I  I I I  | I  I I I  | I I  I I |  I TTI |  I I
    5      10     15     20     2S
           TIME, 
                                                      30
                                                                       4 —
                                                                       3 —
                                                                       2 —
                                                                       1 —
               K2 = 20
               A! = 0.50 ;  Ag = 0.50
               	 - EXACT
                 * - APPROXIMATE ANALYTICAL
i i  i i i  i i  i i i  i i i  i i i  i i i  i
     5      10     15     20
            TIME,  (MINUTES)
~TT~[ MIT
   25      30
     Figure 6.   Comparison  of the exact (numerical)  and approximate analytical  (Equations  (A3)  and  (44))
                 solutions to the time  development of the filter velocities across  the seasoned  filter
                 having the  pressure drop time development shown in Figure 5.

-------
                         PULSE JET FILTRATION THEORY -
                         A STATE-OF-THE-ART ASSESSMENT

               by:   Richard  Dennis
                    GCA/Technology Division
                    213 Burlington Road
                    Bedford,  MA  01730

                    Louis S.  Hovis
                    Industrial Environmental Research Laboratory
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC  27711
                                    ABSTRACT

     The relevant literature on pulse jet filter theory and applications was
assessed to develop coherent guidelines for designing predictive filter
models.  Observations of several foreign and U.S. researchers not previously
subjected to rigorous comparisons showed that there is a considerable
unanimity in technical perspectives when data and theories are viewed from a
common baseline.  Common misconceptions are clarified such as (a) the "air
bubble" that descends through the bag during pulsing, (b) dust removal by air
flushing rather than mechanical projection from a rapidly decelerating fabric,
or (c) failure to recognize that only a small (1-5 percent) fraction of dust
dislodged by pulse action actually reaches the hopper.  Data from several
sources suggest that not only the compressed air pressure, but also the rate
of pressure rise within a bag and the fabric mobility determine dust removal
effectiveness.  Unique pairings of pressure intensity and dust dislodgeability
(as defined by dust/fabric adhesive properties) appear essential for complete
removal of surface dust layers.  When air jets are correctly placed for
specified pulse jet parameters, the surface dust layer should be completely
dislodged.  Various investigators also suggest that the residual dust
(i.e., the irreversibly embedded dust) is relatively constant for fixed
dust/fabric combinations.

     This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.


                                  INTRODUCTION

     The  pulse  jet cleaning concept  introduced commercially during the late
1950's  (1,2,3)  sees extensive application, particularly where stability in gas
flow rates, space economy, and  a capability to handle high dust  loadings (up
to 25  g/m3) are sought.  Despite many  successful installations,  relatively
few investigators have attempted to describe the cleaning process from a
theoretical perspective.  Furthermore, the high  costs associated with
sophisticated instrumentation,  along with complex  interactions among key
variables contributing to the cleaning process have restricted most  functions
for
                                      22

-------
predicting collector performance  to  semi-empirical  equations.  Their  proper
application usually requires  one  set of  actual  performance data  for the
dust/fabric combination of interest  with all  relevant  design  and operating
parameters clearly specified.  The above constraint applies to available
models for predicting performance of both single  and multi-bag units.  Optimum
application of pulse jet units  has been  hampered  by certain common
misconceptions that still prevail despite the fact  that the experimental
findings of several investigators should have long  since dispelled  them.

     For example, a precursor cleaning technique, "Reverse-Jet Cleaning,"
developed by Hersey and introduced during the early 1950"s,  (2,4) relied on  a
high velocity, 1830 to 2440 m/min (6000 to 8000 ft/min) air  jet  directed
against the clean air (outside) surface of a  bag  to remove  the dust cake.
Dust dislodgement by this process is correctly attributed to  a combination of
aerodynamic forces and fabric flexing at the  point  of  jet impact, the latter
effect causing cake fracture and severance of adhesive bonds. What has been
ignored by some individuals, however, is the  fact that air  velocities produced
during pulse jet cleaning are at least 100 times  smaller than those
encountered with the Hersey collectors.  Hence, dust removal  by  air drag alone
is relatively unimportant with pulse jet devices  (3,5,6).

     The persistence of  false impressions is  partly attributable to a failure
to publish relevant experimental data in readily accessible journals, as well
as to the sparsity of overview documents in which the studies and concepts  of
various investigators are compared and evaluated.  A primary objective of  this
paper is to review various independent studies to determine which concepts
relate best to pulse jet collector performance with specific emphasis on  the
pulse jet cleaning process.  Although most field  and/or performance tests  have
concentrated on application  problems rather than the theoretical aspects of
pulse jet systems, their appraisal  has proven helpful in evaluating current
filtration and cleaning  theories.
            COMMON MISUNDERSTANDINGS  OF PULSE  JET  COLLECTOR OPERATION

PROPOSED BUBBLE CLEANING CONCEPT

     Pulse jet cleaning is still described in some documents, accompanied by
imaginative illustrations, as the rapid passage of an air bubble through the
bag.  During its descent, localized fabric distension is supposed to loosen
the dust layer while the transient reverse flow completes the dust dislodgement
from the upstream  filtering surface.  Although there is some truth in the
above concept, it overlooks the fact that typical  pulse durations of at least
0.06 sec far exceed the time for the shock wave to travel to the bottom of
conventional felt bags, roughly 0.0036 and 0.01 sec, respectively, for 1.22 m
and 3.05 m (4 ft and 10 ft) long bags (3,5,7).  In reality, while the pressure
front is advancing, the continued arrival of pulse air, at successively
increasing rates until the solenoid valve is fully open, maintains the
inflating pressure behind the front.  Because pulse air entry continues well
after the bottom of the bag is fully pressurized,  the concept of an advancing
bubble  is not consistent with the mechanics of the process which can be
inferred analytically and has been confirmed by actual measurements (3,5).

                                      23

-------
Observed Pressure Changes During Pulsing

     If, for analytical purposes, it is assumed that none of the pulse air
injected into a bag escapes over the time span required for the solenoid  valve
to open fully (0.01 sec), the time frame for pressure changes at the top  and
bottom of a 1.22 m (4 ft) bag following pulse initiation can be represented by
Figure 1.  Variations in bag pressure with distance from the top of the bag
and time following pulse initiation are shown in Figure 2.  Note that the
axial, top to bottom, pressure gradient diminishes when the "full-open"
pressure front has reached the bottom of the bag 0.0136 sec after valve
activation.  Simultaneous clean-side pressure profiles determined at two
locations for a 1.83 m (6 ft) long felt bag, Figure 3, display the
characteristic time lag indicated in Figure 1 over the 0.01 sec valve opening
period (3).  Pressure/time traces for Figures 3 and 4, however, also reveal
that once the valve is fully open, the pressure decreases rapidly to a lower
and nearly constant level where it remains until the solenoid valve begins to
close.  Figure 4 depicts more sophisticated pressure measurements on a
1.22 m x 11.4 cm bag under typical filtration conditions (5).  The abrupt
pressure reduction is attributed to the sudden dislodgement of the surface
dust layer that facilitates venting of pulse air in the reverse flow direction.
The fact that pressure gradients remain nearly constant (except for the valve
opening and closing intervals) also suggests that negligible dust removal
takes place once the valve is fully opened (5).  It has also been observed
that pulse durations as long as 30 sec produced no discernible differences in
dust removal and/or operating pressure loss across the baghouse (5).

Postulated Bag Deformation During Pulsing

     In the preceding discussion, the physical response of the bag to pressure
imbalance was not considered except for the erroneous bubble concept.  In the
following paragraphs, the temporal and spatial aspects of bag deformation are
examined for what are considered to be the bounding conditions for bag
motion.  Let us first assume that the fabric is infinitely flexible such that
any part of the bag will be displaced almost immediately when subjected to a
pressure differential.  The extent of the displacement or inflation depends on
how tightly the bag is fitted to the supporting cage; i.e., the degree of
slack available.  Additionally, the greater the combined areal density of the
bag and its dust loading, the slower will be its response to a pressure
change.  Hence, very rapid inflation is associated not only with high
flexibility but also a low system mass, as will be discussed later.  When the
fabric response is nearly instantaneous,  the bag contours at selected times
after pulse initiation should appear as if they had been generated by
thrusting a frustum of an inverted cone into the bag at sonic velocity,
Figure 5.  The maximum diameter  for the hypothetical cone will be
approximately the same as that  for the  fully inflated bag.  Maximum lateral
displacement at the top of the bag occurs when the valve is fully open (after
0.01 sec).  Because of the sonic lag, however, maximum displacement will be
seen 0.0136 sec later at the bottom of the bag.  The purported bubble effect,
Figure 5a, is shown for reference.
                                      24

-------
Predicted Bag Response  to  Pulse  Pressure

     Under actual  filtration  conditions,  inherent  fabric  stiffness  plus  the
inertia of the combined dust/fabric mass  will  cause  an  appreciable  lag in  bag
response to internal pressurization.   The initial  radial  motion  imparted to
the dust-laden fabric,  as  described  in earlier studies  (5,8),  consists of  a
rapid shift or "flip flop" from  the  normal deflated  (filtering)  position to
its inflated state where the  fabric  no longer  presses against  its  supporting
cage.  The approximate  lateral (radial) movement can be estimated  by:

where P is the combined dust/fabric mass per unit area,  G the flexing
parameter, r the radial or lateral displacement,  and Ap  the pressure acting on
the fabric.  By equating the restoring force, r/G, that  opposes fabric flexure
to the opposing acceleration force

                                  d2r     -
                                P — r or Pa,
                                  dt
and by treating the lateral displacement or "flip flop"  as a simple harmonic
motion; i.e. ,


                           a = 4^2f2r = 4Tt2r/t2                             (2)
the characteristic oscillation period, T, for a fabric element, may be
estimated as 2ir\/"G~P.  Therefore, the time required for the fabric to attain
its maximum velocity from a dead start becomes T /4 or roughly l.6\/PG .  When
T/4 is less than or equal to the time required for the pulse to reach its
maximum pressure differential 0^0.01 sec), the fabric tends to move with the
pulse.  Conversely, if T/4 is much greater than the pulse rise time, a
sluggish response should be anticipated with poor cleaning characteristics
until  the pulse is lengthened.  Based on prior determinations of G for wool
and Dacron felts (5), T/4 is expected to range from 0.005 to 0.01 sec, thus
suggesting a minimal delay in bag inflation for many commercial applications.
This paper briefly reviews the roles of the flexing parameter, G, gross fabric
areal  density, P, and fabric tensile properties as they apply to fabric
cleaning intensity.

PROPOSED "DEPTH" FILTRATION CONCEPT

Cake Versus Depth Filtration

     There are many misconceptions and divergent  opinions as to how  pulse  jet
collectors operate.  For example, felt filtration is sometimes incorrectly
described as "depth" filtration, implying that the  fabric  functions  as a
deep-bed  filter in which high bed porosity allows the interstitial  fibers  to
play a commanding role in particle capture.  Although base line efficiencies
for most  felts exceed those for conventional woven  fabrics, 90 and  40 percent,
respectively, with atmospheric dust  (5), it  is the  surface dust layer and  not
                                      25

-------
the felt substrate that is mainly responsible for  the  high  efficiency
collection.  Except for brief periods,  about 0.1  sec,  when  flow direction is
reversed by the pulse action, fabric surface loadings  fall  in the same  range,
400 to 1000 g/m2 (5,6,8), seen for mechanically shaken or reverse-air-
cleaned woven fabrics.  Therefore, aside from a transient recovery period,
primary dust capture should be attributed to sieving by the surface dust
layer, whose presence has been well defined photographically (5,9,10).

     Some descriptions of pulse jet cleaning also suggest that the maximum
surface loading is equal to the amount  of dust deposited on the felt during
the nominal 1 to 2 minute filtration cycle.  If this were so, fabric loadings,
W, immediately before cleaning would be expected to range from 2 to 70  g/m2
rather than the 400 to 800 g/m2 actually determined by experimental
measurement (5,9,11,12,13).  Furthermore, since steady state operation  demands
equal dust arrival and removal rates, the only dust remaining on the fabric
immediately after pulsing must be that  imbedded within the  felt.  Were  this
the actual case, there would be some justification for the  "depth" filtration
concept.

Pressure Loss Considerations

     The classical relationship for expressing total pressure loss, P,  across
a dust-laden fabric, i.e.,

                           P = PR + PD  = PR * K2VW

is often incorrectly applied to pulse jet filter systems.  In fact,
Equation (3) should only be used when total pressure loss can be divided into
two distinct components:  P^, the pressure loss across the  fabric per  se
with or without its residual dust holding, Wg, and PQ, the  pressure loss
across the surface dust layer.  PQ depends upon the specific resistance
coefficient for the dust, K2; the filtration velocity, V; and the areal
density, W, for the dust cake.  Based upon average values noted for PR,'V»O.I
kPa, the surface dust loadings of 2 to  70 g/m2 cited earlier could never be
expected to produce the much higher pressure losses observed during typical
pulse jet filtration, Figure 6.  As discussed later, this problem arises
because Equation (3) fails to account for the actual surface loadings  of 400
to 800 g/m2 that contribute  significantly to total system pressure loss.
                                      26

-------
                     UNIQUE FEATURES OF PULSE  JET OPERATION
DISLODGEMENT FORCES AND CAKE SEPARATION

     Most investigators agree that the impulse imparted to a  fabric,  by
mechanical shaking or high pressure air pulse, plays  a major  role  in  fabric
cleaning.  Additionally, the degree of cleaning has been related to the
quantity of dust on or within the fabric (6-8).  Regardless of fabric type,
the dust separating force, Fg, has been defined as the product of  the local
fabric loading, W, and the maximum deceleration of the supporting  fabric
beneath it; i.e., Fg = W amax.  Typical amax values have been indicated
as 1 g for simple bag collapse systems and 3 to 5 g for mechanically  shaken
bags (7).  In the case of pulse jet collectors, however, the  combined
dust/fabric mass may experience decelerations in the 250 g range,  which  afford
the potential for 100 percent dislodgement of the dust cake as compared  to 10
to 50 percent for reverse flow and mechanically cleaned devices (5,8,9,12).

     Irrespective of cleaning method, dust removal can occur  only when Fg
exceeds the local adhesive force, FA, whose magnitude is established by  (a)
the unique interfacial bonding properties for a specific dust/fabric
combination and  (b), any pre-pulse mechanical disturbance of  the dust layer.
Additionally, past studies have shown that dust separation occurs  principally
at the interface between the dust cake and the fabric where the adhesive bonds
are considerably weaker than those within the dust cake (5,8,9).

PROJECTED DISTRIBUTION FOR ADHESIVE FORCES

     When a completely cleaned or new fabric  is first placed  in service,  the
dust cake and the adhesive forces are expected to be uniformly distributed,
barring significant particle  fractionation.   On the other hand, instantaneous
values for separating forces can vary from point to point depending on the
method of cleaning and type of fabric support.  The first dust to be dislodged
is that from areas where local adhesive forces have been exceeded.  The  fact
that only 10 percent removal takes place does not preclude near uniformity for
initial adhesion  levels because adhesive forces for the uncleaned regions need
only exceed the  separating forces by a small  amount.  The "uniformity" concept
becomes even more tenable based on the observation that repetitive collapsing,
shaking, or pulsing always removes additional dust.  Such secondary
dislodgement may be attributed to partial severance of adhesive bonds during
the first cleaning  that decreases adhesion levels over much of the uncleaned
surface.  After many successive filtration and cleaning cycles, cake adhesion
values are expected to envelop a fairly broad force spectrum such as that
shown in Figure  7 for a fly ash/cotton fabric system.  If it  is assumed that
the napped cotton surface provides about as many contact points as those for a
Dacron felt and hence approximately the same  adhesive properties,  extrapolation
of the Figure 7 adhesion/force curve suggests that separating forces must
exceed 200 N/m2  to  obtain complete dust cake  separation with a single pulse.

     The computation of Fg values  for pulse jet systems, as indicated below,
shows that separating forces well in excess of those essential for complete
cake dislodgement,  roughly 1000 to 2000 N/m2f are attainable.

                                      27

-------
FABRIC DECELERATION AND SEPARATING FORCES
Maximum Projected Dust Dislodgement

     Maximum fabric deceleration,  a
                                   max>
                          which  is  imparted  to  the overlying
dust layer, W, determines the separating force exerted at the dust/fabric
interface when further bag inflation is constrained by bag tensile properties.
Dennis and Wilder have calculated amax from the relationship (5,8),
                                  max
                                                     d (Ap)/dt
                                                              (4)
     Increasing values for the flexibility parameter, G, and the rate of bag
pressure increase, d(Ap)/dt, will increase bag deceleration.  Conversely, an
increase in gross bag weight, P, and stretch properties as defined by the
elongation parameter, Me, will produce the opposite effect.  Additional data
for the terms appearing in Equation (4) are summarized in Table 1.  It is
emphasized that bag deceleration rates of 125 to 250 g's (which should be
easily obtainable in routine field applications), coupled with typical fabric
dust loadings of 400 to 800 g/m^ will generate unit separating forces of 500
to 2000 Pa.  Thus, if the surface fiber dispersion for Dacron felts can be
approximated by that for a sateen weave cotton (as visual inspection seems to
indicate) the force levels should suffice to dislodge completely the surface
dust layer.

  TABLE 1.  FABRIC AND PRESSURE PARAMETERS FOR ESTIMATING MAXIMUM DECELERATION
            OF A LIGHTWEIGHT, MODERATELY STIFF DACRON FELT
Parameter
Typical values3
  (cgs units)
                                                       Definition
G

M
e
P

d(Ap)/dt
a
max
5.51 x 10 cm - s /g
-622
2.04 x 10 cm - s /g
-1 2
1.26 x 10 g/cm
6 3
2.49 x 10 g/cm - s
2.71 x 105 cra/s2
(276 g)
Radial displacement per unit pressure

Elongation per unit pressure
Combined dust/fabric areal density

Rate of pressure increase
Maximum dust (or fabric), deceleration

 aNominal  values  for tests described  in Figure 6.
                                      28

-------
Conditions Preventing Complete  Surface  Cleaning

     It is recognized that complete dislodgement  of  the  dust  cake may  not
always be attainable.  For example,  if  the  time required to establish  a
sufficiently heavy dust layer for removal by  a  fixed set of pulse jet
parameters exceeds that for a normal 8-hour shift, complete dust removal and,
hence, steady state operation may not be obtained.   Additionally, improper  jet
placement may result in secondary air entrainment from the "dirty"  side of  the
bag that can actually increase  dust deposition  on the upper end of  the bag
during the pulse interval.  In  the absence  of cleaning,  a fraction  of  the
upper fabric surface is lost just as effectively  as  if the bag were actually
shortened.  Finally, any dust subject to hydration or other chemical reactions
may undergo a cementing action that reduces cake  porosity and greatly
increases its adherence to the  fabric.   This  problem is  accentuated when a
filter remains idle in atmospheres where moisture condensation, extraneous
reactants, and high temperatures prevail.
                            RECYCLED  DUST  LOADING, Wc

     The transient character of the flow interruption during pulse jet
cleaning, about 0.1 sec, permits a small fraction of the dislodged dust to
reach the hopper.  In contrast, practically all dust removed during cleaning
processes where compartment flows are interrupted for 0.5 to 2 minutes settles
to the hopper or is swept into the inlet ducts of adjoining on-line
compartments.  It has been estimated that 1 to 5 percent of the dust actually
dislodged reaches the hopper (5,6,8) and that the remainder is immediately
redeposited on the pulsed and/or adjacent bags (5,8).  Because the dislodged
dust consists principally of large, 50 to 100 urn agglomerates (5), the
coarseness restricts redeposition to the surface region, and its specific
resistance coefficient has been estimated as 5 to 7 times lower than that for
a freshly deposited fly ash (8).  Low l^ values for the redeposited or
"cycling" layer, Wc, may be inferred from the fact that at steady state
operation an average cycling layer of 600 g/m2 contributes less than 0.6 kPa
to the total pressure loss whereas an overlying, freshly deposited layer of
only 60 g/m^ causes a 0.4 kPa increase, Figure 6 (5,8).  The precise manner
in which WQ contributes to total pressure loss is discussed later.

     The term re-entrainment has been used inadvisedly to describe the
recycling layer, Wc, as well as referring to hopper dust swept into the air
stream when overfilling takes place.  Although the latter condition should
definitely be eliminated, the true Wc layer plays an important role in the
filtration process by providing a barrier that minimizes dust penetration
through the relatively inefficient, cleaned fabric.  Additionally, it reduces
variability in air flow that often cannot be tolerated with many ventilation
and process applications.  Thus, when adopting off-line cleaning as a means of
reducing operating pressure loss, one must also accept the performance
penalties of higher emissions and flow variability.  The preferred method to
enhance performance is to select cleaning parameters that enable complete
surface dust dislodgement.
                                      29

-------
ESTABLISHMENT OF STEADY-STATE FILTRATION

Transition From Conditioning to Steady-State Phase

     When pulse jet filtration is  initiated either  with  unused  or uniformly
cleaned fabrics, the pressure loss increases in accordance  with the  rate  of
dust arrival, the filtration velocity,  and the K2 value  for the freshly
deposited dust.  In Figure 8, which simulates the real pressure/time
relationshps graphed in Figure 6,  successive filtering and  cleaning  cycles are
shown for the first 4 minutes and  final 3 minutes of the 60-minute operating
period required to reach constant  operating pressure (or steady-state
filtration).  The pressure/time trace for the 60th  minute corresponds  roughly
to the D-690 curve of Figure 6 except for the time  scale compression.

     At the end of 1 minute, the first  pulse dislodges a very small  fraction
of the initial deposit such that the average (transient) pressure  loss arising
from unequal flows through the just-cleaned and uncleaned surfaces,
respectively, is that designated by the triangles.   The  almost immediate
redeposition, however, of most of  the dislodged dust, elevates the pressure  to
a much higher level depicted by the x symbols while arrival of fresh dust over
the 1-minute filtering period raises the overall filter  resistance to  an  even
higher level as shown by the circles.  With repetition of filtration-cleaning
cycles, the pressure loss before cleaning will reach its maximum value when
maximum surface dust removal is attained.  In some  cases, surface  dust removal
may not be complete either because of poor nozzle placement or adhesion
problems caused by moisture condensation or chemically reactive dusts. After
60 minutes, the transient pressure loss just before redeposition of  the
cycling load (designated by triangles)  must decrease to  the base level
depicting pressure loss for the uniformly cleaned fabric.  Because the interim
quantities of dust returned to the fabric must also increase during  the
conditioning phase, the pressure loss values immediately following
redeposition of the cycling load,  Wc, also increase until steady-state
operation is attained.  In Figure  8, the "X" values indicate the pressure that
would be obtained if the curves of Figure 6 were extrapolated linearly to zero
time.  The latter value has been described as the "effective" bag  pressure
loss, (Pe)^y, at the resumption of a filtration cycle in which the newly
arriving dust is characterized by  the term, AW.  Its use in predictive
modeling (Equation 6) has been discussed by Dennis  and Klemm (7).

Fundamental Pressure Loss Relationships

     In the absence of a recycling layer, Wc, Equation  (3)  affords a good
estimate of total pressure loss based on the respective  contributions  of  the
cleaned fabric, P^, with its residual dust holding and  the  freshly deposited
dust layer, K2VAW.  It has been pointed out, however, that  the above
approach must be modified to account for the very significant contribution of
the cycling layer (7,8); i.e.,

                          P = P_ + (K0) VW  + K.VAW                        (5)
                               R     2  c  c    2
                                      30

-------
     The additional right-hand term depicts the  pressure  loss  across  the
redeposited or cycling layer in terras of its unique  and much  lower  K2 value
(because of its agglomerate structure) and  the  filtration velocity, V.  Owing
to problems in measuring (K2)c and Wc, Equation  (5)  has been  reduced  to
a simpler form for modeling purposes (7,8),

                              P = (PE}AW *  VAW                           (6)

where (PE)^W can be evaluated for a broad range  of design,  flow,  and  pulse
cleaning parameters.   On the premise that there  does exist a  unique Wc layer
that can be described as discussed in the following  paragraph  and,  furthermore,
that complete dislodgement of this layer is possible under optimum  cleaning
conditions, Equation (6) and its derivatives become  important  modeling tools
(7,8).

STEADY-STATE OPERATING CONDITIONS

Fabric Dust Loadings Versus Dust Concentration and Filtration Velocity

     The basic criterion for 100 percent dust cake removal is that  the separa-
ting force, Fg, exceed the maximum adhesive force, F^, the latter uniquely
determined by the dust/fabric combination of interest.  Therefore,  once  the
pulse characteristics that determine  fabric deceleration are  fixed, there is  a
minimum and unique fabric loading, Wc, required for 100 percent cleaning.
Variations in inlet dust concentration, C^  , and filtration velocity,  V^,
have a negligible impact on Wc as suggested by the loading data shown in
Table 2 (5,7).  Bakke's tests, Figure 9, also reflect the independence  of
fabric loading on inlet concentration (14).  For example, with operating
pressure loss fixed at 875 Pa, filtration velocities leveled  off at nearly
constant values irrespective of the broad  range in inlet concentrations,  0  to
22 g/m3.   Since pressure loss and velocity were constant, it  can be inferred
that the cycling loading, Wc, must also have been constant.  Ellenbecker and
Leith have also reported that steady-state  or "cycled" dust holdings were
essentially independent of filtration velocity over the range 3 to  7.5 m/min
(11).  Additionally, for comparable operating conditions, their tests revealed
not only fabric loadings similar  to those  reported by Dennis  and Wilder  (5),
425 and 406 g/m3y respectively, but also relatively uniform dust deposition
over nearly the full length of the bags.

    TABLE  2.  EFFECT OF PULSE PRESSURE, INLET CONCENTRATION,  AND FILTRATION
              VELOCITY ON STEADY-STATE (Wc) FABRIC LOADING (5)
Pulse3
pressure
kPa
(psig)
276 (40)
483 (70)

C = 2.0
V = 2.6
40!
36i
Fabric dust
g/m3
m/min
5
5
loading, g/n/
C = 22.7
V = 1.89
503
423

g/m3
m/min


 a  Pulse duration = 0.06 sec.; pulse interval = 1 min.


                                      31

-------
Dust Removal Versus Pulse Jet Pressure

     Although we believe that nearly 100 percent dislodgement  of the  surface
dust layer should occur, except for the  special  circumstances  cited earlier  in
this paper, Ellenbecker and Leith have expressed reservations  that  100 percent
cleaning takes place or that the major cleaning  process  results  mainly from  a
surface layer release and not some interstitial  reorientation  of the
dust/fiber structure (6,11).  linoya and Mori, however,  describe tests on  a
three bag collector, Figure 10, where pulse pressures  ranging  from  200 to  700
kPa (29 to 102 psig) resulted in up to 75 percent removal  of the total dust
holding, 600 g/m2,  before pulsing (12).   Because no further removal was
noted beyond 550 kPa, we have assumed that surface removal was essentially
complete and that the remaining 150 g/m2 constituted the interstitial and
difficult-to-dislodge residual loading,  W^.  Although  WR values  for
relatively new felts and woven fabrics may be as low as  36 g/m2> it is
anticipated that gradual plugging after  extended service may lead to  values  in
excess of 200 g/m2 (7).

     Past and current GCA tests (7,10) also suggest, based on  a  combination  of
photographic evidence and fabric weighings before and  after pulsing,  that  100
percent removal is attainable under the  proper  operating conditions,  Figure  4
and Table 3.  For example, the curvilinear extrapolation of the  pressure/time
traces, Figure 4, indicates approximate  residual pressure losses of 0.1 to
0.2 kPa just before re-deposition of the Wc fraction of  the dislodged dust.
The latter values are consistent with PR levels  displayed by used felts with
no surface dust loading.

           TABLE 3.  FRACTIONAL SURFACE  CLEANING OF USED DACRON  FELT
                     VERSUS PULSE PRESSURE—LABORATORY PANEL TESTS
Pulse
pressure
N/m2
(psig)
183 (20)
276 (40)

Surface dust
Before pulse
467
467

loading, g/m2
After pulse
100
25

Surface
cleaned
0.79
0.95
Optimum Performance Criteria

     As indicated in previous studies (5,7,8), attainment of optimum operating
conditions, which includes 100 percent dislodgeraent of the surface dust layer,
depends upon several pulse-related factors; i.e., pulse pressure and volume
relative to bag size, solenoid valve opening time, location of jet relative to
bag exit, pulse frequency, and minimum pulse duration.
                                      32

-------
      The effectiveness  of  the cleaning  action  also  depends on  fabric  mobility
as  defined  by its flexibility,  tensile  properties,  and  inertia,  while service
life  is determined by  resistance  to corrosion, abrasion,  and  thermal  stresses.


      Choice of  certain design or  operating  parameters,  such as top  versus
bottom inlet, air-to-cloth  ratio,  and  (most  importantly)  conditioning
processes  plays  a key  role  in obtaining optimum  system  performance.   For
example,  fabric  life can be extended and its  residual pressure loss  kept  at a
minimum by  avoiding overcleaning  and by initiating  no pulse action  until
pressure  loss limitations  demand  it.

                                         REFERENCES


        1.   Reinauer, T. V.  "No Moving  Parts Inside Hot Dust Collector," Chemical
             Engineering 64:188, August  1957.

        2.   Billings, C. E., and J. E. Wilder.  Handbook of Fabric  Filter Technology,
             Volume I, Fabric Filter Systems Study.   U.S. Environmental Protection
             Agency, Control Systems Laboratory, Research Triangle Park,  North
             Carolina.  EPA-APTD 0690 (NTIS No. PB-200-648).  December 1970.

        3.   Dennis, R., and L- Silverman.  "Fabric  Filter  Cleaning by Intermittent
             Reverse Air Pulse," ASHRAEJ  4(3):43, March  1962.

        4.   Hersey, H. J. Jr.  "New Developments in Reverse Jet Filters," AIHA
             Quarterly 14:3, 1953.

        5.   Dennis, R., and J. E. Wilder.  Fabric Filter Cleaning Studies.  U.S.
             Environmental Protection Agency, Control Systems Laboratory, Research
             Triangle Park, North Carolina.  EPA-650/2-75-009 (NTIS No.
             PB-240-372/3T1).  January 1975.

        6.   Ellenbecker, M. J., and Leith, D.  "Dust Removal from Non-Woven
             Fabrics—Cleaning Methods Need to be Improved," Filtration and  Separation
             18:315, 1981.

        7.   Dennis, R., and H. A. Klemm.  "Modeling Concepts for Pulse Jet
             Filtration," JAPCA 30:38, 1980.

        8.   Dennis, R., J. E. Wilder, and D. L. Harmon. "Predicting Pressure Loss
             for Pulse Jet Filters," JAPCA 31:987, 1981.

        9.   Chudleigh, P. W.  "Fabric Filter Performance:  Effect of and Cleaning
             Efficiency," Filtration and  Separation  19:388, 1982.

        10.  Development and Evaluation  of Improved  Fine Particulate Filter  System.
             EPA Contract 68-02-3151 Mod  No. 2, March 1980  to July 1983,
             GCA/Technology  Division, Bedford, MA  01730.   Report in Preparation.

        11.  Ellenbecker, M. J., and D.  H. Leith, "Dust  Deposit Profiles  in  a High
             Velocity Pulse-Jet  Filter,"  JAPCA 29:1236,  1979.

        12.  linoya, K., and Y. Mori. "Fundamental  Experiments of Fabric Filters."
             In Symposium on the Transfer and Utilization of Particulate Control
             Technology, Volume  2.  Fabric Filters and Current Trends in Control
             Equipment, EPA-600/7-79-044b  (NTIS No.  PB-295-227).  February  1979.

        13.  Lefkowitz, L. R.  "Evaluation of Felted Glass  Media under Simulated Pulse
             Jet Operating Conditions."   In Symposium on the Transfer and Utilization
             of Particulate Control Technology, Volume 2.   Fabric Filters and Current
             Trends in Control Equipment, EPA-600/7-79~044b (NTIS No.  PB-295-227).
             February 1979.

        14.  Bakke, E.  "Optimizing Filtration Parameters," JAPCA 24:1150,  1974.
                                               33

-------
    max
U a
a: "
         TOP TO BOTTOM
         TIME LAG = 0.0036 sec

         VALVE OPENING
         TIME • 0.01 sec
              0.002     0.004     0.006


                          TIME, sec
                                      0.008
                                              0.01
Figure  1.   Presure gradient  vs. location  and

            time after pulse  initiation  for

            1.22 m (4 ft)  Dacron felt bag.
                                                                pmox
                                                              If) o
                                                              Ul >-
                                                              OC -
                                                              a. -o
                                                                Figure  2.
                                                                                        0.0136 sec
                                                                                                •005
                                                                                                    sec
                                                                                   0.5


                                                                                 DISTANCE, m
                                                                                                     1.0
                                                                          Pressure gradient along  1. 22  m
                                                                          (4 ft) bag vs.  solenoid valve

                                                                          open time.
    250
    200 -
 CO
 c/i
 LJ
 ft
 a

 o
                               0.20m from bag exit

                          2    1.71 m  from bag exit
    150 -
     0    0.02   0.04   0.06   0.08   O.I    0.2

                      TIME, sec.


Figure  3.   Concurrent pressure/time traces,  1.83 m

            (nominal 6 ft) Dacron felt bag  621  kPa

            (90 psig) pulse,  0.1 sec duration.
                                                                   •	 0.25 sec

                                                                    o.oe sec PULSE
                                                                                                O.ISsec PULSE
                                                              Figure 4.
                                                                         INITIATE SOLENOID VALVE OPENING

                                                                         VALVE  FULLY OPEN

                                                                         START  OF SOLENOID VALVE CLOSING


                                                                        Pressure/time traces  for 483 kPa

                                                                         (70 psig)  pulses  for  1.22 m x 11.4  cm

                                                                         (4 ft  x 4-1/2 in) Dacron felt bag.

-------
   Q.
   UJ
   U
   z
   o
   o
   m
   o
   (a)
          0.005 sec

             (b)
{

1>

\

^
0.01 sec 0.0 136 see
(c) (d)
 Figure 5.   Bag profiles vs.  time

              after  pulse initiation
              for a  1.22 m (4 ft) long

              bag. Bubble concept  (a)

              vs. theory (b)(c)(d).
    l.0r
o
UJ

O
5
UJ
K

Ul
U.
o
    0.5
0.2
    O.I
   0.05
o

o
   0.02
   0.01
                 CURVE  DESCRIPTION

                   I    WOVEN GLASS/
                       COAL FLY ASH

                   2   COTTON SATEEN/
                       COAL FLY ASH
       (I)  (21
               J_
                                  _L
      I         10    20      50

              ADHESIVE FORCE, N/m
                               100   200

                               2
  Figure 7.   Fractional dust removal

               vs. estimated  adhesive

               force.
                                                    1.6
                                                    1.4
                                                    1.2
                                                  o
                                                 a.
                                                    L0
                                              V)
                                              V)
                                                0.8
                                              UJ
                                              oc
                                              3
                                              cn
                                              in
                                              UJ
                                                    0.6
                                                       Pulse frequency -1 minute
                                                       Pulse duration -0.06 sec
                                                    0.4
                                                    0.2
          Pulse Pressure
           kPa, gauge
                                                   -Pi
             D690

             0483

              276

             D345

             D483

             D4'83

*D Indicates Damped Pulse


'Residual Pressure Loss,
 Cleaned Fabric Areas
                                                                   I
                                                            0.2    0.4    0.6    0.8

                                                            FILTRATION TIME, min
                                                                                 1.0
                                           Figure 6.   Resistance charapteristics for

                                                        fly ash  and Dacron felt  at
                                                        C.  = 27.6 g/m3  and V =  2.6 m/min,
                                                      PRESSURE LOSS

                                                      BEFORE PULSING
                                           Figure 8.   Characteristic  pressures before

                                                        and after steady-state

                                                        filtration is achieved.
                                           35

-------
u>
       fO
        E
          20
        <
        o:
UJ
o
z
o
o
        UJ
        _l
        z
       15 oz.(340q/m2) WOOL FELT

      643kPo OOpsig)
     _PULSES

      O.I sec DURATION
       I pulse/min

      3/4 in.(1.91cm) VALVE
           10
        BOTTOM INLET-
                                                                       i.o
                                         •TOP INLET
                                                          j

                                                           3
                   FILTRATION  VELOCITY, m/min
                                                                    o

                                                                    cf
                                                                    >
                                                                    O
                                                                      0.8
                                                            UJ

                                                            <  0.6
                                                            O
                                                                    V)
                                                                    D
                                                                    O
                                                                      0.4
                                                                    O

                                                                    I-
                                                                    u
                                                                      0.2
                                                              00
      NOTES

     600g


V * 3 m/min


WOOL FELT FABRIC

CoC03 MMO  « 2>i


WR (estimoted)
                                                                                I
                                                                                        I
                                                                       200     400      600

                                                                          PULSE PRESSURE, kPo
                                                                                                      800
       Figure 9.  Talc  filtration velocity vs inlet

                  concentration at  constant pressure

                  loss.   Source:  Reference 14.
                                                           Figure  10.   Fractional dust  removal vs.  pulse

                                                                        jet pressure  at  constant W.   Source;

                                                                        Reference 12.

-------
         LABORATORY TECHNIQUES FOR DEVELOPING PULSE JET COLLECTORS


                    Robert R.  Banks and John T.  Foster
                          Wheelabrator-Frye Inc.
                      Air Pollution Control Division
                            Technical Center
                           5100 Casteel Drive
                     Coraopolis, Pennsylvania  15108
                                ABSTRACT
     A major factor in the performance of a pulse jet collector is its
effectiveness in using compressed air.  The manifold configuration,
venturi design and other factors determine this effectiveness.  To ex-
pedite the study of these factors over a wide range, two sets of apparatus
were built.

     The first, the impact force tester, measures the momentum of the air
pulse from each hole in a manifold and the pressure along the manifold.
The momentum of the jets can be equalized by varying the hole size along
the manifold.

     The second test rig measures the characteristic of the ejector or
jet pump formed by the air jet and venturi so that the geometry of this
part of the system can be optimized.  Jet pump curves were found to
predict relative performance as measured in a laboratory pilot collector.
Testing by these methods takes only a fraction of the time required using
a pilot collector.  Data from tests leading to an optimum pulse jet
cleaning system design will be shown.

                              INTRODUCTION
     At Wheelabrator-Frye pulse jet collectors have been developed through
several generations of designs.  The first collectors were small and per-
formance requirements not demanding.  Changes in design could be based on
observation and modifications to commercial collectors.  As development
continued, each improvement in performance or economy became more
difficult to achieve.  To reduce the cost and time required to generate


                                     37

-------
improved designs, laboratory techniques were developed for generating pre-
liminary performance information.

     The main areas of technical study have been the more effective
utilization of air in the pulse cleaning system, the flow and distribution
of air and dust on the dirty side of the collector, and improvements in
filter media.  This paper confines itself to the cleaning system.

                            DEVELOPMENT TECHNIQUES
     In the early stages of development, laboratory work consisted mainly
of making pressure measurements on the header and in the manifold.  Because
of the speed of these transient phenomena, a light writing oscillograph was
used to record pressures measured by means of pressure transducers.  The
data was principally used to study air consumption per pulse and valve
timing compared with  electrical signals.

THREE BAG COLLECTOR

     At this stage changes in the design of a pulse jet collector were
studied by trying the changes on a full-size operating unit.   Consideration
was given to building a laboratory collector using three full-size mani-
folds, as it was felt that it is necessary to use a full-size manifold in
order to get the proper cleaning pulse for test work and that the minimum
number of rows that would not result in overloading the bags not being
cleaned would be three.  At this point, it was suggested that it was un-
necessary to use a bag at every hole in the manifold but rather, use one
hole per manifold for cleaning and discharge the air from the other holes
into the atmosphere.  In this way, the bags would be cleaned by the pulse
exactly as in a full-scale collector, but the amount of air and dust to be
handled would be much more appropriate for laboratory studies.

     A small collector was built following this principle using three bags
in triangular arrangement inside a circular shell.  The shell is made in
sections so that its length can be adjusted for use with different size
bags.  The clean air plenum is flanged at the cell plate level so that
spacers can be inserted to permit changing the distance between the mani-
fold and venturi for test purposes.  The air header is the same size that
would be used on a full-size collector so that the pressure at the dump
valve follows the same pattern as in a full-size collector.  This collector
is shown in Figure 1.

     The equipment is designed so that the manifold or blow pipe can be
slid through the clean air plenum permitting any hole in it to be centered
over the bags.  This was done because measurements of pressure along the
manifold showed a rise in pressure as the velocity dropped from air being
bled off.  Rough calculations using Bernouilli's theorem give figures
fairly close to the measured values.  Tests were carried out that showed
better cleaning performance at the end of the manifold remote from the
dump valve.  Most developmental testing was done using the jet closest to
the dump valve so that the results would be conservative.

                                     38

-------
l^2 , 	 1
i 1
; i
k
AA 	 .
1+ 	 "«
J





9-




^ — i




7-



• 	 -*









"••






<*

\ / "
\ /
\ i
!..i


1 i




< 	 ^
)r4^













-5 s-

— i


1.
2.
3.
4.
5.
6.
7.
8.
9.
-6
n i



HEADER, COMPRESSED AIR 90 psi
DUMP VALVE
TOP PLENUM
OUTLET DUCT. C^cAN AIR
COMPRESSED AIR MANIFOLD
MANIFOLD EXTENSION
BAG
INLET DUCT, DUST LADEN AIR
COLLECTOR SECTION
                                                  8
                  FIGURE 1.   THREE BAG PULSE-JET COLLECTOR
     In carrying out a test on this unit it is operated at a fixed air-to-
cloth ratio until stable operation is reached and then the ratio is raised
and the procedure repeated until the pressure drop on the collector does
not stabilize but continues to rise.

     Tests on this type of equipment led to the adoption of the double
diaphram dump valve, a new venturi, optimization of manifold to venturi
spacing and other features embodied in our present design of pulse jet
collectors.

     Ideas based on observations made in the course of the development work
described suggested ways in which further improvements could be made in the
performance of the cleaning system.  A program was planned to follow up on
these ideas to improve the pulse jet cleaning system and develop a new
generation of pulse jet collectors.  One area of prime interest was the
manifold design.

PULSE FORCE MEASUREMENT

     As described earlier, the manifold of uniform section and with uniform
size orifices gives progessively more cleaning force along its length.  In
order to compensate for the increase in the pressure as the velocity de-
creases along the length of the manifold or blow pipe, two options are
available; 1) taper the pipe so that the velocity remains constant, or 2)
                                     39

-------
use larger orifices where the pressure is lower.  It was felt that using
graduated hole sizes along the length of the manifold would be less costly
than producing a tapered or stepped manifold.

     Use of  the laboratory three bag pulse-jet collector would obviously be
impractical  for developing the optimum drilling pattern for a variable hole
size manifold because of the time required for a test.  The pulse force
equipment was devised to do this job expeditiously and economically.

     In the  pulse jet cleaning system the jet acts in combination with the
venturi or the open end of the bag as a jet pump to entrain air from the
outlet plenum to get a large volume; low pressure pulse that actually per-
forms the cleaning.  Since the jet pump is a momentum transfer device, a
means for measuring the momentum of the jets could be used as a measure of
the cleaning power available at each orifice.  The simplest means for
measuring the momentum of a stream is to measure the force generated when
it impacts on a fixed plate.  In most of the cases studied the velocity is
sonic and the increased momentum of the jet is due to the increased density
of the air at the higher pressure.

     A device was set up as shown in Figure 2.
                                      LOAD CELL
_i
          MANIFOLD
                                         IMPACT PLATE
                                             PRESSURE
                                            TRANSDUCER
    iniiiiiiiiiniiiii inn a ni
                                                               RAILS
                     FIGURE 2.  PULSE FORCE EQUIPMENT
It consists of circular plate  four inches in diameter mounted  on a load cell
and positioned opposite a hole in the manifold.  The output of the load cell
is recorded on the light writing oscillograph after going through a signal
conditioner.  The support for  the force transducer is mounted  so that it
can be slid along rails and clamped in position over each of the manifold
holes in turn.  A full-size header and standard dump valve are used to feed
the manifold.  In all of the test work reported in this paper, the header
size was held constant and the initial header pressure was 90  psig.  In
                                   40

-------
addition to measurements of pulse force,  measurements of pressure along the
manifold are made using a pressure transducer.

     Using this technique, curves as shown in Figure 3 were developed.   The
pulse force for a manifold with uniform hole size (U) varied from 5.4 Ibf
at the nearest the valve to 8.0 Ibf at the 15th hole.  By progressively
reaming the holes the pattern called "A" was developed.  In these drilling
patterns, the number of different hole sizes was limited to make fabrica-
tion on a production basis simpler.
                   I
             H-   9
             114
             "   a
             CC   8
             O
             O
                                                               50
                                                               40
                                                                CO
                                                                Q.
                                                               30
                                                               20
                                                               10
                                              10
15
                                           HOLE POSITION
       FIGURE 3.  PULSE FORCE AND MANIFOLD PRESSURE VS. HOLE POSITION
                                     41

-------
     When evaluated, the first holes in the "A" drilling did not give the
expected performance.  It was reasoned that the cause of this was that the
pressure in the manifold at the first few holes was too low to give the
critical pressure ratio.  The flow from these holes was thus below sonic
velocity while the remainder of the holes were operating at sonic velocity.
In order to correct this, several other drillings were tested and the
pattern referred to as "D" was selected.  The impact force measurements
with drilling "D" are not quite as uniform as is the case for drilling "A"
but the overall performance is slightly better.  Peak pressure at the first
hole for the "D" drilling is 37 psia so that the pressure ratio is about
2.5 versus 1.9 for the "A" drilling.  The critical pressure ratio for air
is about 2.

JET PUMP CURVE APPARATUS

     The next area of the cleaning system to be studied was the jet pump
itself.  In this part of the program, we wanted to measure the flow and
pressure developed by the jet pump comprising the jet of air from the
manifold and the venturi or similar device at the open end of the bag.  The
first course considered was to measure all of these factors dynamically.
While the results from such a system would be the most realistic, it was
felt that the cost of setting up the equipment and the time to develop the
technique for making that type of measurement would be excessive.  As an
alternative, it was decided to study the jet pump characteristics under
steady state conditions.  It was felt that this approach would be accept-
able, since the final test would be an operating collector in any case.
Figure 4 shows the equipment set up for these measurements.
           ONE HOLE MANIFOLD


              1
                                                        STILLING
                                                        CHAMBER
                                                              FLOW
                                                          tf^~ NOZZLE
      FLEXIBLE ^*
       TUBE
                        FIGURE 4.  JET PUMP EQUIPMENT
                                     42

-------
     The air compressors and receivers used have sufficient capacity to
provide a flow in excess of 130 cfm for the tests.  This flow was calculated
by dividing the volume of air discharged by a dump valve in a single pulse
by the length of time of the pulse,  and by the number of holes in the mani-
fold.  For the tests, a manifold with a single hole to the size being
studied is used.  The pressure and temperature in this manifold are'measured
and its position relative to the rest of the equipment is readily adjusted.
The flow to this manifold is adjusted by means of two valves, one is used to
adjust the flow, and the other is used to provide the 90 psi back-pressure
required by the calibration of the rotameter used to measure the flow of
compressed air.

     The air jet from the manifold is directed into the end of either a five
inch or six inch diameter steel tube representing the bag that would be
cleaned in an operating collector.  Venturis or similar devices can be
mounted at the end of this tube.  Pressure in the tube is measured at
pressure taps about halfway down its length.   At the far end of the tube,
there is a butterfly valve that is used to provide variable back pressure
in determining the jet pump characteristic.  Beyond the butterfly valve,
the air enters a stilling chamber prior to exiting through a standard ASME
flow nozzle used to measure the combined flow of the compressed air and
induced air.  The back half of the equipment is essentially the same
equipment that is used for measuring the characteristics of a fan, and is
designed in accordance with the standards of the AMCA and ASHRAE.

     In carrying out a test, the air flow to the one hole manifold is re-
gulated so that the pressure in the manifold is the same as that found in
an operating manifold on the pulse force equipment.  The compressed air
flow for that condition is measured and that flow rate is used for all
subsequent tests on that orifice size.  To determine the characteristic
of the jet pump, the butterfly valve is locked in various positions and the
pressure and flow are measured.  The data from these measurements are
plotted to give a curve as shown in Figures 5-8.

     Figure 5 shows the effect on the jet pump characteristic of the
differences in pulse force shown in Figure 3 for the uniform hole size mani-
fold (Drilling U).  The performance for the conditions at the first hole are
appreciably lower than for the last hole which produces about 25% higher
flow.

     By using the variable hole manifold (Drilling D), as shown in Figure
6 the performance at the conditions of the first and last holes are quite
close and, in fact, the first hole shows slightly more power.  The per-
formance of the jet from the last hole is also increased slightly.
                                     43

-------
   o
   CM
   HI
   u

   LLJ
   DC
   ill
   Q
   HI
   cc

   I
   111
   cc
   O.
       50 --
40 --
                     400         800
                     TOTAL FLOW CFM
                                     1200
FIGURE 5.  JET PUMP CURVE - VENTURI T DRILLING U
                       400         800

                       TOTAL FLOW CFM
                                       1200
FIGURE 6.   JET PUMP  CURVE - VENTURI T DRILLING D
                       44

-------
     Figure 7 shows more clearly  the difference in performance  between mani-
fold with uniform holes and variable sized holes.   These curves are for the
first holes in the two manifolds  using the same venturi and show almost 50%
improvement in performance by using the variable hole manifold.
              O
               CM
              I
              LU
              O
              z
              LU
              cc
              LU
50 -
40 -
               LU   30 -
               cc
               1
               Ul
               cc
               0.
20


10
                                       lling D
                                   -r
                                              -r
                                  400          800
                                  TOTAL FLOWCFM
                                      1200
               FIGURE 7.  JET PUMP CURVE - VENTURI T HOLE 1
VENTURI DEVELOPMENT

     The next component of the pulse cleaning system to be studied was
the venturi.  Various designs of Venturis were tested using the jet pump
equipment.  In these tests the distance from the manifold to venturi was
varied and the optimum determined.  If too close a spacing is used with
some short Venturis the jet tends to blow through without touching the
walls and poor performance results.

     Figure 8 shows the performance of three of the Venturis tested.
Venturi T is a fairly conventional design and shows a high maximum
pressure that falls off rapidly as the total flow volume increase.
Venturi 0 is of increased throat size and shows a low maximum pressure
but it will handle a larger air flow.  By trying several intermediate
designs Venturi F was developed to give an intermediate maximum pressure
with an air flow even better than the 0 design.
                                     45

-------
                o
                 CM
                X
                   40 --
                o

                ui
                cc 30
                iii
                 £
                 oc
20 •-
                 e/5 in ..
                    IW ^K
                 Ill

                 CC
                 0.
                               VENTURI T
                          VENTURI F
        VENTURI O
                                  400        800

                                  TOTAL FLOW CFM
                                      1200
               FIGURE 8.  JET PUMP CURVE - DRILLING D HOLE 1
EVALUATION


     The performance of  these Venturis was evaluated in the three bag

pulse jet collector.   Table  1 shows the air-to-cloth ratio that could be

attained using the three Venturis whose characteristics are illustrated

in Figure 8.


              TABLE 1 AIR-TO-CLOTH RATIO VS. VENTURI DESIGN
                Venturi
                     Air-To-Cloth Ratio
                   T

                   0

                   F
                             7.0

                             8.0

                            10.0
                                    46

-------
If the results shown in Table 1 are compared with the curves in Figure 8,
it appears that the jet pump in a collector must be operating in the high
flow-low pressure section of the characteristic.  In the flow range
above 600 cfm the sequence of performance in the jet pump equipment is
the same as in the three bag collector.   It would thus appear that high
induced flow is more important than high peak pressure in cleaning pulse-
jet bags.

                                 SUMMARY
     Several pieces of equipment have been built to expedite the develop-
ment of pulse-jet cleaning systems.

     By use of these systems, an improved pulse-jet cleaning system has
been developed and preliminary evaluation performed in the three bag
collector.

     The final test — field evaluation — is in progress.
     The work described in this paper was not funded by the U. S.
Environmental Protection Agency and therefore the contents do not neces-
sarily reflect the views of the Agency and no official endorsement should be
inferred.
                                     47

-------
                    OFF-LINE PULSE-JET CLEANING SYSTEM

                         BY:  Thomas C. Sunter
                              C-E Air Preheater
                              Div. of Combustion Engineering, Inc.
                              Wellsville, New York  14895
                                 ABSTRACT

     The paper discusses off-line pulse-jet cleaning systems by formulating
a set of simplified mathematical equations to describe the cleaning process.
A computer model is developed from the equations and manipulated to simulate
both time-based and differential pressure-based control  of the pulse-jet
system.

     Practical application of this information is discussed as it affects
system control, bag life, energy consumption and system operation.
                                     43

-------
                               INTRODUCTION
     Off-line cleaning of a multi-module pulse-jet dust collection system
offers advantages over the conventional  on-line cleaning system utilized
with felted fabrics.   In terms of bag life and total  cleaning energy, it is
the preferred method for those systems employing woven glass fabric bags.

     This paper discussed off-line cleaning under three main headings:

1.   A simplified mathematical formulation of the cleaning process which
     lends itself to computerized simulation.

2.   Manipulation of this model under both time-based and differential pres-
     sure-based cleaning control systems.

3.   Discussion of the practical application of this information as it af-
     fects system control, bag life, energy consumption and general system
     operation.

                         MATHEMATICAL FORMULATION
ASSUMPTIONS

     The following simplifying assumptions were used in the formulation of
the equation sets which describe the cleaning cycle:

1.   Filter drag for the fabric and dust cake is linear with respect to
     total  dust deposition and within any particular model iteration, in-
     dependent of filtration velocity.   l<2» the dust permeability is meas-
     ured in grains/sq. ft/ins. hLO/fpm and measures the slope of the drag-
     dust load line.

2.   The value of filter drag occurring immediately after the cleaning of
     a module is constant and independent of the drag just before cleaning.

3.   The component of system pressure drop not related to the filter fabric
     and dust cake; i.e. all duct, venturi and inlet/outlet losses obey a
     strictly square law relationship between pressure loss and volume and
     are identical for each module.

4.   Isolation of a compartment for cleaning and subsequent de-isolation for
     return to the on-line condition each occur instantaneously.

5.   The system volumetric flow rate is constant.

6.   The pressure drop contribution of the fabric and dust cake is linear
     with respect to filter drag where filtration velocity is the constant
     of proportionality.
                                     49

-------
     It should be emphasized that the assumptions inherent in the formula-
tion will produce results which differ from measured values in an operating
system.  This study is less concerned with predicting field values than it
is with study of the effects on system operation of changes in the system
parameters and control variables.

FORMULATION

     The formulation covers a system of N modules, each containing A sq. ft.
of fabric with a total fixed system volume of Q ACFM.  Three sets of equa-
tions are used to describe system behavior as follows:
Gross Condition
     pt =
n=l
 I
n=N
                            (vn ),
                         For all N values of n
                                                  (1)
                                                       (2)
Where
         n ;G
Net Condition
     "t ' <»„> *
     n=l
               System pressure drop - ins.  hLO at time t.

               Drag of nth module at time t.

               Filtration velocity.  FPM gross condition for nth
               module at time t.

               Gross filter ratio for which all air losses equal 1 ins,
               H20.
     •  t\      a
     'Vn 'N     A

     Except n = nc
[-
   t
(vn ),

 ~R7~
                         For all n values escept n = n
                                                  (3)
                                                       (4)
Where  (v t)M  =  Filtration velocity FPM net condition for nth module at
        n  N     time t.
                                     50

-------
Drag Calculation
       t+t
D
= D
                        >
      n        n         G/N      2                     (5)

Where g = Dust load grains per acf.

     l<2 = Dust permeability grains/sq. ft. /ins. rLO/fpm.

     t  = Time increment minutes for incremental drag calculation.

     Equation (5) simplifies drag calculations by avoiding a set of  differ-
ential equations.  The degree of accuracy required can  be attained by  ad-
justing the value of t .
     System behavior is es-
sentially described by
stringing together a series
of the basic sequence ABCD
shown in Figure 1.

     This shows the clean-
ing cycle for one module
only and can be described
as follows in Table 1:
                    I
                    «
                    1
                                    net
                           isolate
                                        gross
                                                      deisolate
                                                              gross
                                                   time - t
Point
                     Figure  1.   Cleaning  cycle  for  one  module.

TABLE 1.  CALCULATIONS FOR THE BASIC  CLEANING  CYCLE

                  Mode                      Calculation
  A

  B

 BC

  C

  D
   Gross condition before insolation

   Net condition after isolation

   Net condition during isolation

   Net condition before deisolation

   Gross condition after deisolation
                                               Equations  (1)  &  (2)

                                               Equations  (3)  &  (4)

                                               Equations  (3)  &  (4)  & (5)

                                               Equations  (3)  &  (4)  & (5)

                                               Equations  (1)  &  (2)
     Separate computer programs where developed  from  the formulations  dis-
cussed above for off-line cleaning under time  based and differential  pres-
sure based cleaning system control.
                                      51

-------
     Programs were arranged to provide output for up to 100 complete clean-
ing cycles.

     Output included values of P., D  , (v  )G and (v  )N for all modules
and all cycles so that the basic system pressure-time graph could be
drawn for any particular cleaning cycle number.

     Values of (v  ), both gross and net, were included in the output as they
provide informatiBn on the flow distribution among the modules at any time.

                               MANIPULATION
TIME-BASED CLEANING

     Figure 2 and Tables 2 and 3 show the  input and output for the first 3
cleaning cycles of a four module system with time-based cleaning.

     Both the graphical and mathematical output show that system stability
is near completion after one cycle and complete after three cleaning cycles.

     Under stable conditions during the third cycle, gross ratios vary from
2.1 to 4.1 fpm and net values from 3.2 to  4.8 fpm.

     During the first cycle, gross ratios  vary from 1.1 to 6.3 fpm and net
values from 1.7 to 7.0 fpm.

     Comparison of the drag values for all modules at t = 5 minutes shows
that if there were no air losses (R =°°), the flow distribution would be more
skewed with module no. 1 passing 73% of the total system flow.
                                                            t 2
     It is the contribution of the air losses; i.e., the (v  )  term in
equations (1) and (3), which tends to even out the flow malBistribution
occuring in the system due to the different drag values for each module.

          TABLE 2.  INPUT DATA FOR THE TIME BASED CLEANING CYCLE
          VAR.
         .NAME-
N
.A 	
Q
XK2
.G 	
PO
Tl
T?
TO
RO
DO
NUMBER OF MODULES
Mwtnn f, FTLTE" A^FA 
-------
   8

   7

q 6

   5
M

-------
     It is readily apparent that the variation  in  system  inserted pressure
drop is most marked at the beginning of  the  first  cleaning  cycle (i.e.  4.07
ins. wg in the first ABCD loop) and reduces  rapidly  as  the  overall  system
drag diminishes.

     The cleaning frequency is of course determined  by  the  control  system
setting and in this case, for any particular module,  is once  every 28 min-
utes.
PRESSURE-BASED CLEANING
     Figure 3 shows the differential  pressure-time  plot  for  the same (4)
module system under pressure-based cleaning with  set  points  of 3 and 5 ins.
wg. for the low and high levels respectively.

     Gross ratios vary from 1.3 to 6.3  fpm during the first  cycle,  but stay
in the range 1.4 to 6.1 fpm after stabilization.

     The inserted system differential pressure  varies over  the range of
2.3-6.9 ins wg. in the stable condition.

     The cleaning frequency is determined by  the  system  operating parameters
particularly the rate of pressure rise  in the gross condition.  In  this
case cleaning frequency for any particular module is  once every 101 minutes.
   8 •

   7

q  6


 .  5
 9

 I  '
  Q.
  <  2
                                                                       —high
                                                                        low
                   iQ           80           120
                                time  -  minutes

                    Figure  3.    Pressure  based  cleaning.
                                                       160
                                                                     200
                                      54

-------
                           PRACTICAL  APPLICATION
COMPARISON OF TIME AND PRESSURE  BASED  CLEANING

     The essential  advantage of  a  differential  pressure based cleaning sys-
tem is that it cleans on demand;  i.e.,  provided that the system runs  below
the high set point, no cleaning  occurs  and  all  modules  are on line.   Another
and more effective way of stating  this  advantage is  in  terms  of the auto-
matic accommodation of a system  under  AP cleaning to load changes, both up
and down.  Such is not the case  with  time based cleaning.  In the ^vent of
load reduction, either the control  system must be adjusted to a ".  .iger
cleaning cycle or the system will  overclean down a lower system differential
pressure.

OPTIMIZATION OF THE CLEANING CYCLE

     In discussing possible modifications to cleaning cycle control systems
with a view to optimization, a number  of requirements,  some conflicting,
must be considered.

Pulse Pressure Versus Pulse Frequency

     There is adequate experimental evidence to conclude that generally
speaking, the amount of dust removed  from the bags of an off-line module
will increase both with increase in pulse pressure and  with the number of
pulses to which the bags are subjected.   It can also be speculated that:

-There is some pulse pressure below which little bag cleaning, if any
 occurs.

-The dust removal  rate is highest  for  the first of a number of pulses and
 diminishes with successive pulsing.

     What is not readily apparent  is  precisely what effect variations in
pulse pressure and frequency have  on  bag life; e.g., what are the compara-
tive equivalent effects on bag life of 1 pulse @ 100 psig versus 2 pulses
each at 50 psig?

     Our uninformed opinion in the example quoted is that frequency  contrib-
utes more than pressure to shorter bag life, thus the latter is worse than
the former.

Provision Of A Safety Factory

     In the specification and selection of a pulse-jet system to control
a particular process, the Application Engineer may frequently be faced  with
a lack of complete process definition on some or all of  the  following  items:

-The process is new, therefore no dust sample is available.
                                      55

-------
-The process is different so that the dust load and particle size analysis
 of the dust are approximate.

-The system design volume is imprecise.

-The exact nature and duration of upset conditions are speculative.

-The gap between normal and design volume is very large.

     These conditions are frequently met today in applying fabric filtration
systems to fluid bed boilers.

     Lack of definition for any of the process variables enumerated above
suggest that the wise approach is the incorporation of some safety factor
into the pulse-jet system design.

     Since provision of too large a safety factor is economic suicide,
another approach is a control system design which lends itself to system
operation beyond the norm, but retains the optimizing capability at all
system conditions.

System Versus Module Control

     System control under differential pressure-based cleaning usually
involves module cleaning in response to the total system differential pre-
sure; i.e., flange-to-flange pressure drop through the baghouse.

     While this control method monitors and controls the most important
system variable; i.e., flange-to-flange pressure drop, it falls short of high-
lighting and controlling conditions in the individual system modules.  Thus,
for example, control  by this method will not immediately identify a module
which, for some reason, is not being cleaned.  Failure to do so will even-
tually lead to unacceptable system operation and bag life.

     It should be noted here that comparison of the tube-sheet differential
pressure after module cleaning with some pre-set value is not a good indica-
tion of cleaning performance, since the differential pressure value associ-
ated with good cleaning will vary with the condition of all the other
modules in the system at that time.

Unattended Operation

     There is a predilection among industrial users to run pollution control
systems unattended or monitored periodically.  Alarms and/or enunciators
may be incorporated to signal a problem to a manned station, but unattended
operation must necessarily place greater demands on the control system.

OFF-LINE CLEANING SYSTEM SIZING

     Sizing of a pulse-jet off-line cleaning system for a particular appli-
cation involves the following series of steps:
                                      56

-------
-Determination  of the  overall  allowable  inserted  system pressure  drop.

-Breakdown of this pressure  drop  into  its  (3)  constituent  parts:

     -Pressure  required  to satisfy  the gas  only system losses  (with no
      particulate).
     -Pressure  required  to overcome the  residual  filter drag of the cleaned
      fabric.
     -Pressure  required  to support  the cleaning cycle  as the dust load  on
      each module varies throughout the  cycle.

-Maximization of  this  latter component of  total differential pres^^re by
 component sizing and  fabric selection.

-Filter ratio selection  to provide:

     -Adequate  bag life.
     -Acceptable  cleaning cycle frequency.
     -Differential  pressure  within  the allowable  limits.

-Control  system and instrumentation specification to  provide:

     -Allowance for upset conditions
     -Accommodation to load  changes.
     -Possible  selection errors.

CONTROL SYSTEM  DESIGN  -  AP CONTROL

     The two  basic control methods;  i.e.,  time- and pressure-based cleaning,
have already  been compared.  While  pressure-based cleaning does have the
built in capability of automatic  accommodation to load changes, it suffers
from one inherent disadvantage; namely,  that  the  system reverts to time-based
cleaning if the low differential  pressure  set point cannot be  satisfied by
the pre-set cleaning cycle.  Several other features should be  noted:

  1. If system  conditions vary sufficiently that  not  only  the  low AP setting
     cannot be  satisfied, but  also  the high AP setting cannot  be  maintained
     then the system approaches  instability which is  normally  signaled by
     an alarm on  a high-high AP  setting.   At this point, system recovery
     may be irreversible; i.e.,  the bags are already  plugged and  will not
     clean down.

  2. The setting  of the  low  AP value is  arbitrary in  two ways:

     -It is different  from the high AP setting;  i.e., it automatically
      imposes a wider  variation  in  system AP than for time-based  clean-
      ing, where  effectively the  high  and low AP values are equal.

     -It automatically provides  for a  finite period in  the gross  condition
      between cleaning cycles, which is  required to accommodate  upset
      conditions.
                                      57

-------
  3.  Module Isolation times would normally be selected so that under normal
     load conditions at least 50 - 75% of the modules would require clean-
     ing to satify the low AP setting.

     Note:  The remaining 50 - 25 % uncleaned modules supply a safety  factor
     for high load or upset conditions.

            However, at partial loads, the low AP might well be  satisfied by
     cleaning only one module, which would impose a severe asymmetric  flow
     distribution through the total system.

CONTROL SYSTEM DESIGN - ALTERNATIVES

     In situations where a pulse-jet off-line cleaning system  is operated
at normal design conditions under AP control and it is found that the  low
AP setting is never satisfied, several corrective options are  possible,  both
manual and automatic.

     Manual readjustment of the low AP set point to a higher value will
correct the problem as illustrated in figures 4 and 5.  Figure 6 also  shows
an alternative manual adjustment where the module dwell and isolate times
have been reduced so that the module cleaning frequency is  increased.
 . 6

I
10 e

1
' 4
Q.

-------
   s
 o
 1  4

Q.
                                                                                       low
                      10
 20             30

 time  -  minutes
50
            Figure  5.    Pressure   based   deaning  -  pressure   adjusted.
o
a.
<
                                                                                    •high
                     10
20             30
 time  -  minutes
              Figure  6.    Pressure   based   cleaning  -  time   adjusted
                     I    £    ±   ±    ±    i
./

u
/

u
/

w
/

u
/

v
                   Rgure  7.    S   module   system  with  on-line   trim.
50
                                            59

-------
         Table  4.    On- lint trim sequence  for  6 module  system.
mode
pure off - line 1
first trim /2T
sequence
234

5
5/ST

6
6Xr
         second trim     J/£.3    ^.^  %-SJ  Vs'-ST  ^6-IT  6X-2T


         third trim      J/2-3-4T  2/3-4-5T 3/£s-6T Vs-WI 5/6-1-2T
     A novel way in which this problem can be corrected automatically has
been designated "on-line trim" and is illustrated in figure 7 and Table 4.
The system essentially superimposes on-line cleaning of selected modules
on the normal off-line cleaning sequence.  Thus if under normal off-line
cleaning with AP control, the low AP value is not satisfied at the end of
the first cleaning cycle (all modules), the first trim step in introduced;
i.e., the off-line cleaning of any module during the second cycle occurs
simultaneously with the on-line cleaning of the next module in the clean-
ing sequence.  If the low AP setting is not satisfied at the end of the
second cleaning cycle, then additional on-line trim is introduced where
during off-line cleaning of a particular module, on-line cleaning of the
next 2, 3 or 4 modules in the cleaning sequence is successively initiated.

     The on-line trim is not intended as a cleaning process equivalent to
off-line cleaning, but rather as a suppressor of filter cake formation during
the trim period so that the system reacts in terms of AP as if the number of
modules in the system is reduced by one for each trim increment.

     As a practical matter, it is desirable to invoke the trim condition
at a higher pulse pressure than is used for off-line cleaning.  This in-
volves the provision of two separate compressed air systems, either of which
can supply the pulse valve reservoirs on demand.

     Such a system is presently at the installation stage on a large fluid
bed boiler application.

     One further comment is apropos.  Normal operation of an off-line pulse-
jet system under differential pressure cleaning control involves two AP
settings, namely a high and low value.  It should be noted, however, that
from the point of view of system stability, the low value is superfluous;
i.e., if at the end of a system cleaning cycle the AP is no higher than it
was at the start of the cycle, then the system is stable.
                                      60

-------
BAG LIFE

     Bag life in an off-line pulse-jet system is dependent on many factors,
not the least of which is the frequency and pressure level of the bag
pulsing system.

     The fact that there is some correlation among bag life, pulse pressure
and pulse frequency can scarcely be in doubt, but its exact nature is un-
known to this author.   The advent of the microprocessor may prove helpful
in this regard as it can provide a vast increase in control system operating
flexibility and data accummulation.

                                CONCLUSIONS
     It is hoped that this paper provides some additional insight to the
mechanics of multi-module pulse-jet off-line cleaning systems.

     The author believes that the control system, with automatic response
to changes in operating condtions, is the heart of such a system.
Additional work is required to throw light on the effect on bag life and
system efficiency of variations in pulsing pressure and frequency of the
cleaning system.
               The work described in this paper was not funded
               by the U.S.  Environmental  Protection Agency and
               therefore the contents do not necessarily reflect
               the views of the Agency and no official endorsement
               should be inferred.
                                      61

-------
            FIELD EVALUATION OF THE DRAG OF INDIVIDUAL FILTER BAGS
                        W. T. Grubb and R. R. Banks
                          Wheelabrator-Frye Inc.
                      Air Pollution Control Division
                             Technical Center
                            5100 Casteel Drive
                      Coraopolis, Pennsylvania   15108
                                 ABSTRACT
     Evaluation of the performance of filtration fabrics has heretofore
necessitated a full compartment test, as only general observations could
be made using a few test bags.

     Compact, inexpensive, readily portable equipment was developed to
measure the filter drag of a single bag by blowing a metered flow of
ambient air through the bag and measuring the pressure drop.  The test
apparatus is set up in the baghouse compartment to minimize disturbance to
the bag and its dust cake.

     Test results show reproducibility and statistically significant
differences among bags made of different cloths and at different sites.
Evaluation of the performance of various filter fabrics on the basis of a
limited number of trial bags is possible.  Selected data will be presented
reflecting experience with tests on fiberglass and synthetic bags in five
large  utility collectors.

                               INTRODUCTION
     In evaluating the performance of fabric filter media in a baghouse
filter, one of the primary considerations is the pressure drop across the
operating filter media.  This pressure drop is a function of the filter
drag of the fabric/dust cake system, the superficial velocity of the gas
passing through the fabric, and the viscosity of the gas. (1)  In a con-
stant temperature situation the gas viscosity does not vary, and the
following relationship exists:
                                     62

-------
               v
  where:   S = Filter Drag (in H 0-min/ft)

          AP = Pressure drop across the filter media (in HO)

           V = Superficial velocity (ft/min)

  and   v ^

                                                    3
  where:   Q = Total gas flow through the filter (ft /min)
                                       2
           A = Effective cloth area (ft )

     The filter drag varies during the filtration/cleaning cycle with the
amount of dust cake deposition on the filter.   Filter drag just after
cleaning is referred to as "residual drag" and just prior to cleaning as
"terminal drag."  The drag of the filter at any given time during the
filtration cycle is the sum of the residual drag, S , and the drag of the
dust deposited on the fabric during filtration subsequent to the cleaning
cycle.  In a given baghouse filtration situation, if only one compartment
of the baghouse is considered, the factors which affect the drag of the
filtered dust layer should be the same for each individual bag within the
compartment, assuming uniform flow distribution within that compartment.
These factors include dust properties such as specific surface, particle
size and density, inlet dust concentration, time since cleaning and
filtering velocity. (2) If the filter drag variance with time remains
constant, then the residual drag of the "cleaned" fabric would be the
factor governing the overall filter drag and pressure drop.

     The residual drag of a filter is determined by the properties of the
media used, the characteristics of the dust, the interaction of fabric and
dust in the formation of the residual cake and the effectiveness of the
cleaning method employed.  Again considering one compartment of a single
baghouse and assuming uniform flow distribution, the dust characteristics
and cleaning method should not vary from bag to bag, and if the identical
fabric is used for all bags, the residual drag should be the same for all
bags.

     The determination of actual operating filter drag in a baghouse is a
difficult proposition.  This is especially true in a multicompartmented,
periodically cleaned baghouse where velocity changes occur due to changes
in available filter area resulting from compartment isolation during the
cleaning cycle.  The resultant changes in AP due to increased air-to-
cloth ratio and increased drag due to dust deposition may cause the fan
output to vary, and if reverse-flow cleaning is used, the flow will be
increased by the reverse-gas volume.  All of these factors combine to
make it extremely difficult to obtain simultaneous readings for flow
and pressure drop  (exclusive of inlet and outlet losses) on individual
compartments.  To complicate the matter, few commercial baghouses are
instrumented to obtain such data.  (1)
                                     63

-------
     The comparative evaluation of the filter drag of various fabric filter
media in an operating unit requires that, as nearly as possible, the test
fabrics be subjected to the same process variables and equipment operation.
Testing of a full unit of fabric is not economically feasible, nor is it
possible to control process variables during the course of consecutive
full-unit tests conducted during different time periods.  Single compartment
tests have their deficiencies due to flow and dust load variations among
compartments and due to the above-stated difficulties in measuring flow and
AP.  The evaluation of a small quantity of test bags within a single com-
partment by subjective, physical examination or by permeability testing of
bags removed from service has been a common practice.  Determination of
the relative quantity of residual dust cake either by feel or by weighing
is not indicative of the cleanability of the filter bags because there is
no way of determining the relative flow through the different bags.  For
example, a fabric with zero permeability would accumulate no residual cake
because no dust-laden air would pass through it.  Permeability tests of
used bags are of limited value because the dust cake is disturbed during
bag removal and shipment.

     In order to facilitate the evaluation of a limited quantity of test
filter bags, Wheelabrator-Frye has developed a method for field testing
the filter drag of individual bags in an isolated baghouse compartment,
thus minimizing disturbance of the bag and its dust cake.

                         DESCRIPTION OF TEST EQUIPMENT


     The "single bag filter drag" test apparatus consists of a centri-
fugal blower, 10 5/8 inch wheel diameter, powered by a one horsepower
electric motor.  The blower intake is equipped with a blast gate allowing
the air flow to be controlled.  The blower outlet is connected by an
adapter to a four-foot length of five-inch diameter flexible hose, which
is attached to a two-section metal pipe, five feet long and five inches
in diameter.  An orifice plate is mounted between the flanged ends of the
pipe sections, and felt gaskets are utilized to prevent air leakage.
Pressure taps are located one diameter (5") upstream and one-half diameter
(2.5") downstream of the orifice, and an inclined manometer is used to
measure the AP across the orifice during the the test.  An orifice cali-
bration curve (AP  vs.  Flow) as shown in Figure 1, is prepared by compari-
son to a standard°elliptical flow nozzle.

     The outlet end of the metal pipe is inserted into a metal box, one-
foot square with three sloping sides, which serves as the inlet into the
test bag.  The top of the box is equipped with a flange which allows the
mounting of a flanged thimble of the type used for bottom attachment of
the bags being tested.   The inlet box is also equipped with a pressure
tap allowing the measurement of static pressure across the test bag,
using a vertical manometer.  A schematic of this test apparatus is shown
in Figure 2, and photographs are shown in Figures 3 and 4.  The test
apparatus can be easily disassembled for shipment and is readily portable,
by two persons, while fully assembled.


                                      64

-------
100
    0.1
   0.5         1-0

ORIFICE AP (INCHES H20)
  FIGURE  1.   CALIBRATION CURVE  FOR 3.75  INCH ORIFICE
   FLOW CONTROL
    BLAST GATE
                                         TEST BAG
                 FLOW MEASURING
               INCLINED MANOMETER
                                                 PRESSURE DROP
                                                  MEASURING
                                                  MANOMETER
   FIGURE 2.   "SINGLE BAG FILTER DRAG" TEST APPARATUS
                              65

-------
FIGURE 3.   "SINGLE BAG FILTER DRAG"  TEST APPARATUS:  FULL LENGTH VIEW

        FIGURE  4.   "SINGLE  BAG  FILTER DRAG"  TEST APPARATUS:
 INLET CHAMBER,  THIMBLE,  ORIFICE  PLATE,  AND  MANOMETER  ARRANGEMENT

                               66

-------
                          DESCRIPTION OF TEST METHOD
     The "single bag filter drag" test equipment is assembled and set up
in the aisle of the isolated compartment in which bags are to be tested.
The inlet box/test thimble is located adjacent to the test bag.   The bag
is detensioned at the upper suspension level and relaxed slightly to
allow the bag bottom to be disconnected from its thimble.  The bag cap
is removed from its suspension hook,  and the bag is manually suspended
by any suitable means.  The bag is lowered slightly to allow the bottom
to be attached to the test thimble, then it is raised to remove any
slack.  In this configuration, the bag is at a slight angle from the
vertical in order to avoid contact with the upper suspension level walk-
way.

     The blower motor is started, and the desired air flow is obtained
by utilizing the sliding blast gate to control the orifice AP.  Tests are
usually run at two or more air flows, including the normal operating flow
and the maximum flow obtainable by the test blower.  When a stable flow
condition is reached, manometer readings are taken for both the orifice
AP and the bag AP.

     The test bag is then removed from the test thimble, weighed with a
100-lb. spring scale, reconnected to its thimble, and reset to its pro-
per operating tension.
                                                      3
     Orifice AP readings are converted to air flow (ft /min) using the
orifice calibration curve.  Filter velocity (air-to-cloth ratio)-is
calculated by dividing the flow by the effective filter area  (ft ),
    (J.  Filter drag (S) is calculated for each bag AP reading (AP^) as
    A
    AP
S =   B .  Units of filter drag are inches H 0-min/ft.  Dust loading is
    ~T~                                     2
determined by subtracting the weight of a new bag and cap from the test
bag weight.

                                 TEST RESULTS

     During the past two years, eleven  (11) series of single bag filter drag
tests have been performed on six (6) different coal-fired utility boiler
baghouses.  These test programs have been divided between the following two
approaches:

     1.  Evaluation of different types of test bags installed in a single
         compartment.

     2.  Evaluation of selected bags from full compartment tests.

Results of six series of tests at  three different sites  are discussed here-
in.
                                      67

-------
PP & L HOLTWOOD STATION - BAGHOUSE B

     Pennsylvania Power and  Light's Holtwood  Station had experienced an
operating  pressure drop which was somewhat higher than expected in their
Baghouse B during its  initial six months  of operation.  Sonic horns were
installed  to  augment the reverse flow  cleaning  system, resulting in
significant reduction  in system pressure  drop.  There was however some
concern about the fabric used for the  original  equipment bags, and
PP & L initiated a bag test  program in May 1981.  (3)  Six different types of
finished glass fabric  were used by four different bag suppliers to fabricate
twenty-four test bags  (three to seven  of  each type).  These bags were in-
stalled in Compartment 32, rows C and  D.  Bags  of each type were uniformly
distributed from front to rear  of the  compartment to minimize any posi-
tional differences in  flow.  The sonic horns  were rendered inoperable in
order to duplicate the design reverse  flow cleaning mode in the test com-
partment.   PP & L contracted the services of Wheelabrator-Frye to perform
filter drag tests on the individual test  bags and on originally installed
bags used  as  controls. Three series of tests were performed:  after two
months, six months and fifteen  months  of  test bag operation.

     Four  different glass fabric constructions  were evaluated, but only the
three variations of 10 oz/yd fabric will be  considered herein.  These
three fabrics have many common  properties, differing only in yarn construc-
tion, as given in Table 1.

TABLE 1.   FABRIC SPECIFICATIONS: PP &  L HOLTWOOD BAGHOUSE B TEST BAGS
           Fiber:                       Glass  (ECDE).
                                              2
           Weight:                      10 oz/yd   (nominal)

           Count  (Warp x  Fill):         54 x 30
           Weave:                       3x1  Twill

           Permeability:                35-55  ft3/min/ft2 @ 0.5" HO
                                          Yarn Construction
           Fabric Code                  Warp'               Fill
              A                        75 1/0 -    37  1/0 texturized

              B                        75 1/0   ( 50  1/0 texturized
                                                (.150  1/0 filament
              C                       150 1/2    150  1/4 texturized
Four different finish variations were also evaluated.  Fabrics A and B were
finished with Teflon B  (10% loading) applied to greige (not heat-cleaned)
fabric.  Fabric C was also finished with  10% Teflon B, but was heat-cleaned
prior to finishing.  Fabric B was also obtained with two different com-
mercially available "acid resistant" finishes, both of which are applied
to heat-cleaned fabric.
                                     68

-------
Two Months
0.98+
0.74
0.82
0.70
0.79
0.70
	 z 	
Six Months
1.15+
1.19
1.07
0.83
0,93
0.93
Fifteen Months
0.95+
0.98
0.86
0.64
0.67
0.74
     Significant  differences  were  observed in filter drag of the  various  bag
types as shown in Table 2.

  TABLE 2.   SINGLE BAG FILTER DRAG DATA FOR VARIOUS 10 OZ/YD2 GLASS FABRICS:
                  PP & L HOLTWOOD  STATION BAGHOUSE B
   Fabric      Finish           Filter Drag After Indicated Length of Service
Construction    Type            	(In.  H00-min/ft)
     A        Teflon B

     A        Teflon B*

     B        Teflon B*

     C        Teflon B

     B        Acid Resistant-1

     B        Acid Resistant-2
•
These finishes were applied to greige (not heat-cleaned) fabric.   All
others were applied to heat-cleaned fabric.


 Originally installed bags, in service for seven months longer than the
 test bags, tested as controls.
       The original equipment bag fabric, type A, was observed to have the
highest filter drag.  This is probably the result of the "singles"
texturized filling yarn construction and application of the finish over
greige fabric, both of which may have resulted in increased penetration
of particulate into the filling yarn fiber bundle.  Fabric B, with the more
conventional plied/reinforced filling yarn construction, has a significantly
lower drag than fabric A, but substantially higher drag than the three
fabrics which had been heat-cleaned prior to finishing.  Direct comparisons
between fabric constructions B and C could not be made because they were
finished differently.

     Note the variation in filter drag data among the three test programs
conducted at different times.  This variation is due to the length of
service of the test bags, cleaning cycle variations, boiler operating
conditions during the period prior to testing and the procedure used to
shut down the boiler or isolate the compartment prior to testing.  In spite
of these variable conditions, the filter drag values of the different
fabrics relative to each  other remained virtually the same for the two
tests conducted after the bags had achieved a stable residual dust cake.
The residual dust loading of the various fabrics was determined during the
filter drag testing and is given in Table 3.


                                     69

-------
TABLE 3.  RESIDUAL DUST LOADING ON TEST BAGS PP & L HOLTWOOD BAGHOUSE B
   Fabric      Finish             Residual Dust Loading (Lbs.)
Construction    Type             After Indicated Length of Service

                                 Two Months   Six Months   Fifteen Months

    A         Teflon B              59*          60*            64*

    A         Teflon B              26           57             55
    B         Teflon B              24           55             56
    C         Teflon B              22           54             44
    B         Acid Resistant-1      23           54             42

    B         Acid Resistant-2      22           52             48


   Originally installed bags, in service for seven months longer than the
   test bags, and tested as controls.
It is evident that the test bags had not achieved a stable operating
condition after two months service.  It is also evident that variations
in residual dust mass alone do not account for the substantial difference
in filter drag among the test fabrics.

PP & L BRUNNER ISLAND

     Pennsylvania Power and Light also  experienced high operating pressure
drop during the initial year of operation in their Brunner Island Station
Unit #1 baghouse.  This twenty-four compartment baghouse has a design air-
to cloth ratio of 2.13:1 (net) and utilizes reverse flow cleaning for its
11.5 inch diameter by 35 foot 4 inch long bags. (4)  A bag test program,
similar to the one at Holtwood, was initiated in December 1981.  Forty-four
(44) test bags, fabricated from eight different fabric/finish combinations,
by five different bag suppliers, were installed in compartment 3B of the
baghouse.  Bags of each type were uniformly distributed between the two
aisles and from front to rear of the compartment to minimize positional
differences in flow.  Filter drag tests were performed by Wheelabrator-Frye
on the individual test bags and control bags after two and one-half months
and after eight and one-half months operation.

     Four different glass fabric cbnstructions were utilized, but for
purposes of this paper only the three variations of 10 oz/yd  glass fabrics
will be discussed.  As in the Holtwood test, these fabrics are similar,
differing only in yarn construction.  Variations B and C were again
evaluated, along with the original equipment bag fabric, variation D, as
described in Table 4.
                                     70

-------
Warp
75 1/0
150 1/2
150 1/2
Fill
C 50 1/0 Texturized
|l50 1/0 Filament
150 1/4 Texturized
75 1/2 Texturized
TABLE 4.  FABRIC  SPECIFICATIONS:  PP  &  L  BRUNNER ISLAND UNIT #1 TEST BAGS
     Fiber:                                Glass (ECDE)

     Weight:                               10 oz/yd2 (nominal)
     Count (Warp  x Fill):                  54 x 30

     Weave:                                3x1 Twill

     Permeability:                        35-55 ft3/min/ft2 @ 0.5" HO


                                             Yarn Construction

     Fabric  Code

         B


         C

         D
Five different finish variations were utilized.  Fabric D, the original
equipment bag fabric was finished with Teflon B (10%) over greige goods.
All other finishes were applied to heat-cleaned fabric.  Fabrics B and C
were treated with Teflon B (10%) finish applied by the same finishing
company, and in addition fabric C with Teflon B finish was obtained from
a different finisher.  Fabric B was also obtained with two different "acid
resistant" finishes.  Another variable, introduced for evaluation purposes,
was the positioning of the smooth filament warp face on the inside filtra-
tion surface of the bags made from fabric B with acid resistant finish #3.
The results of filter drag tests on the Brunner Island bags are given in
Table 5.

The most significant result obtained in the Brunner Island testing was
that the bags utilizing the smooth warp face as the filtering surface ex-
hibited a twenty to forty percent lower filter drag than any of the other
test bags.  These bags also retained 65-75% less dust load than any of  the
other test bags.  Fabric variation C appeared  to result in somewhat lower
filter drag than fabric B, however additional  analysis of recently ob-
tained data and additional testing will be necessary to validate  this
conclusion.
                                      71

-------
TABLE 5.  SINGLE BAG FILTER DRAG DATA FOR VARIOUS 10 OZ/YD2 GLASS FABRICS
                           PP & L BRUNNER ISLAND UNIT #1 BAGHOUSE

Fabric
Construction

D
D

C
C
B
B
B*
Finish
Type

Teflon B*
Teflon B*
*
Teflon B
Teflon B§
Teflon B§
Acid Resistant 2
Acid Resistant 3
Filter Drag
Indicated
2 % Months
2.38+
1.52

1.37
1.23
1.54
1.57
1.05
(In H20-Min/Ft) After
Length of Service
8 % Months
1.47+
1.34

1.20
1.04
1.25
1.24
0.85
  This finish was applied over greige fabric.  All others were applied over
  heat-cleaned fabric.

  Originally installed bags, in service for fourteen (14) months longer
  than the test bags, tested as controls.
*
  Supplier I

§ Supplier II

jt
  Bags utilized with warp face as the filtering surface.  All others
  utilized the texturized filling face.
     Filter drag values obtained during the second phase of Brunner Island
testing were substantially lower than those obtained during the first
phase, in spite of greatly increased residual dust loading in the bags
(Table 6).  A possible explanation for this phenomenon is that the initial
testing was performed in the compartment shortly after it was isolated
during normal baghouse operation, but the second phase of testing was per-
formed during a boiler outage with the baghouse cold.
                                     72

-------
   TABLE 6.   RESIDUAL DUST LOADING  ON TEST BAGS
                PP & L BRUNNER ISLAND UNIT #1

Fabric
Construction

D
D

C
C
B
B
B"
Finish
Type

Teflon B*
Teflon B*
*
Teflon B
Teflon B5
Teflon B§
Acid Resistant 2
Acid Resistant 3
Residual Dust
After Indicated
2 h Months
76+
30

24
19
29
29
7
Loading (Lbs.)
Length of Service
8 % Months
>85+
>75

55
60
69
58
20
  This finish was  applied over greige fabric.   All others were applied over
  heat-cleaned fabric.


  Data on original bags,  in service for fourteen (14) months prior to the
  installation of  test  bags.

A
  Supplier I

§
  Supplier II


  Bags utilized with warp face as the filtering surface.  All others
  utilized the texturized filling face.
SOUTHWESTERN PUBLIC SERVICE COMPANY

     Filter drag testing was performed on bags at the Southwestern Public
Service Company's Harrington Station.   Bags were tested in five (5) com-
partments of the Unit 2 baghouse and three (3) compartments of the Unit 3
baghouse.  These Wheelabrator-Frye baghouses collect fly ash from two
identical boilers burning the same western sub-bituminous coal.  The bag-
houses utilize deflate/shake cleaning, and are essentially identical
except for the net air-to-cloth ratio which is 3.4:1 in Unit 2 and 3.0:1 in
Unit 3 at full load. (5)

     The purpose of these tests, in addition to the evaluation of various
fabric types, was to determine if there were any positional flow varia-
tions within a single compartment or among the various compartments,
which might have an effect on the filter drag as measured on individual

                                     73

-------
bags.  Each compartment contained bags of a single fabric/finish type.
In general, controlled comparison of the filter drag of various bags in-
stalled in different compartments of the same baghouse is difficult due
to positional flow variation among compartments.  The selection of ad-
jacent test compartments which contain bags having the same length of
service will minimize any potential flow variation.  Care must also be
taken to isolate the test compartments at the same temporal position in
their filtration/cleaning cycle, preferably immediately after cleaning.

     No significant differences in filter drag or bag weight were ob-
served within any of the test compartments which might indicate flow
variation as a function of bag position (Table 7).  Eight (8) bags were
randomly selected and tested along the two outer aisles of the compartment.
The center aisle, was avoided due to suspected disturbance of the bags
caused by air currents entering the isolated compartment door.

TABLE 7.  SOUTHWESTERN PUBLIC SERVICE COMPANY HARRINGTON STATION:
                      SINGLE BAG FILTER DRAG DATA
                            Filter Drag (In H^O-Min/Ft) Bag Weight (Lbs.)
Unit #
2
2
2
2
2
3
3
3
Compartment #
4
6
16
22
26
2
19
27
Mean
1.38
1.30
1.38
1.08
1.31
1.35
1.42
1.26
Std. Deviation
0.07
0.10
0.09
0.14
0.12
0.17
0.09
0.13
Mean
30
28
30
28
16
35
25
24
Std. Deviation
2.0
1.3
0.68
1.3
0.66
1.6
0.64
1.5
     A comparison was made between compartments 19 and 27 on the east side
of Unit 3, both of which contained bags of the same fabric type which were
installed at the same time.  The filter drag in compartment 19, the seventh
compartment from the inlet, was significantly higher at 1.42 inches
H70-min/ft than the filter drag in compartment 27, the third compartment
from the inlet, which was 1.26 inches H 0-min/ft.  This data coincides with
observations at Harrington Station and many other baghouse installations
that compartmental AP increases with increasing distance from the flue gas
inlet.
                                     74

-------
                        PULSE-JET SINGLE BAG DRAG TEST
     Wheelabrator-Frye has recently designed and built a unit for testing
filter drag of individual pulse-jet type bags.   The test apparatus and
method is based on the principles previously discussed in this paper.
A large number of bags can be tested very rapidly with this method com-
pared to the inside-collecting bag test method.   It is anticipated that
this method will be as useful for evaluating pulse-jet equipment design
features as for evaluating various filter media.  Initial field testing
has been performed, and the results of this and other tests will be the
subject of future technical papers.

                            SUMMARY AND CONCLUSIONS
     A method for field testing of the filter drag of individual bags
has been developed,  and used extensively to gather data in utility
coal-fired boiler baghouses.  This method has proved to be valuable in
evaluating various bag fabric types while requiring that only a small
number of test bags of each type be installed in a single compartment.
Meaningful evaluation requires good experimental design, multiple tests
conducted over a period of time, and careful data analysis.  It should
be stressed that this type of evaluation may be valid only for the parti-
cular operating conditions existing within the baghouse tested, and
generalizations concerning the optimum filtration fabric for any specific
coal-fired boiler should not be made solely on the basis of data ob-
tained by this test method in other collectors.

     The work described in this paper was not funded by the U.S.
Environmental Protection Agency and therefore the contents do not neces-
sarily reflect the views of the Agency and no official endorsement should
be inferred.

                            ACKNOWLEDGEMENT S
     The authors wish to acknowledge the support and assistance of Noel
Wagner, Senior Project Engineer, Pennsylvania Power and Light Company, in
the conduct of the field test program.  The authors also acknowledge Greg
Lear and Harry Smith of Pennsylvania Power and Light and Richard Chambers
and Jerry Hermsmeyer of Southwestern Public Service for their cooperation
in the field test program.
                                     75

-------
                             REFERENCES
1.  "Air Pollution Manual, Part II, Control Equipment", American
    Industrial Hygiene Association, Detroit, Michigan, 1968.

2.  Billings, C. E. and Wilder, J., "Handbook of Fabric Filter
    Technology, Vol. I, Fabric Filter Systems Study", National
    Technical Information Service, PB-200 648, Springfield, Virginia,
    1970.

3.  "Discussion: Report on Operating Experience with Fabric Filter Bags",
    Proceedings: First Conference on Fabric Filter Technology for Coal-
    Fired Power Plants, Denver, Colorado, July 15-17, 1981, EPRI CS-2238
    Contract WS 81-206.

4.  Murray, R. W.  and Lear, G. L. "Design, Start-up and Initial
    Operation of the Brunner Island - Unit 1 Bag Filter."  Fifth
    International Fabric Alternatives Forum Proceedings, Scottsdale,
    Arizona, January 1981.

5.  Chambers R., Ladd, K.  and Kunka, S., "Fabric Wear Studies at
    Harrington Station," Third Symposium on the Transfer and Utilization
    of Particulate Control Technology, Vol. I, EPA-600/9-82-005a, July
    1982.
                                     76

-------
             A DUAL-DETECTOR BETA-PARTICLE EACKSCATTER GAUGE  FOR

            MEASURING DUST CAKE THICKNESS  ON OPERATING BAG FILTER

                                AND ESP UNITS

               By:   Robin P. Gardner
                    Center for Engineering Applications of Radioisotopes
                    P.O.  Box 5636
                    North Carolina State University
                    Raleigh, North Carolina  27650

                    R.P.  Donovan
                    Process and Chemical Engineering Division
                    Research Triangle Institute
                    P.O.  Box 12194
                    Research Triangle Park, North Carolina  27709

                    L.S.  Hovis
                    Utilities and  Industrial Processes Division
                    Industrial Environmental Research Laboratory
                    U.S.  Environmental Protection Agency
                    Research Triangle Park, North Carolina  27711
                                  ABSTRACT
     To avoid the use of moving mechanical parts in the hostile environment  of
dust collection that is necessary with the dual-beam beta-particle backscatter
gauge, an alternative measurement principle has been sought that still allows
the measurement of dust cake density thickness independent of distance from
the device.  • It is found that the use of a dual-detector system, in which each
detector detects the response of a different part of the backscattered beta-
particle spectra from a single radioisotope source, can be used to accomplish
this.

     The system investigated here consists of a Sr-90 source and two thin end-
window GM counters, one covered with a simple paper filter with density thick-
ness of 0.050 g/cm2.  The two responses obtained differ significantly with
dust cake density thickness so that models of each response can be solved
simultaneously for two experimental responses to give dust cake density thick-
ness independent of distance from the device.  Laboratory results on a proto-
type device using paper samples of known density thickness to simulate dust
cake density thickness yield standard errors of 0.0278 and 0.0151 g/cm2 for
bag filter and ESP unit configurations, respectively.  The range of distances

                                      77

-------
and simulated dust cake density thickness investigated were 3.4 to 7.4 cm and
0 to 0.2133 g/cm2, respectively.

     This paper has been reviewed in accordance with the U.S.  Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.

                                INTRODUCTION
     To avoid the use of moving mechanical parts in the hostile dust collec-
tion environment that is necessary when using the dual-beam beta-particle
backscatter gauge for measuring dust cake density thickness on operating bag
filters independent of the distance to the device (1), the alternative
approach of using the intensities of two ranges of beta-particle energies is
investigated here.  This approach requires a detection system capable of beta-
particle spectrometry.  Preliminary feasibility investigations were carried
out with a Ca?2 (Eu) scintillation detector coupled to a photomultiplier tube
and high voltage supply, and connected to a multichannel analyzer through a
preamplifier and amplifier.

     The backscattered beta particles from a Sr-90 source with maximum ener-
gies of 2.26 MeV were detected with this system for simulated dust cake
samples.  It became evident from considering the spectra so obtained that a
number of other possible approaches existed for the measurement of dust cake
density thickness independent of the distance to the device when beta-
particle backscatter spectroscopy is employed.  It was found that the beta-
particle backscatter spectra are unique for each dust cake density thickness
and distance.  Therefore, probably the most general approach consists of
storing backscattered beta-particle spectra for the range of dust cake density
thicknesses and distances from the analyzer system.  Then the dust cake
density thickness and distance for a spectrum obtained for a particular sample
at a particular distance can be obtained by a double interpolation procedure.
Or one might accomplish the same thing by first developing a model to pre-
dict the backscattered spectrum as a function of the dust cake density thick-
ness and distance from the analyzer and then use a nonlinear search to
minimize differences between model and experimental spectra to find these
parameters.  The sensitivity and accuracy of these methods will depend upon
how much the backscattered spectra vary with dust cake density thickness and
distance from the device.

     A number of other simpler methods might also be possible and more practi-
cal.  These include a range of empirical and mathematical model methods.  The
original idea of using the intensities of two energy ranges could be extended
to the use of the intensities of multiple energy ranges.  Another approach is
based on reasoning that the shape of these backscattered spectra may be
reasonably insensitive to the distance to the dust cake.  In this case, simple
measures of the spectrum shape might be correlated directly to the dust cake
density thickness.  Or one might consider any alternate method for measuring
distance from the device to the sample (such as the use of an ultrasonic
method) used with a single beta-particle backscatter gauge (2).
                                      78

-------
     In the present paper a very simple and practical method for accomplishing
the desired measurement is investigated.   The method chosen is called the
dual-detector beta-particle backscatter gauge and is based on using two thin
end-window GM counters with a single radioisotope beta-particle source.  One
of the GM counters is used with a simple paper filter placed over the thin
end window so that the lower energy range of the backscattered beta particles
will be adsorbed in, or deflected by, the filter and not be detected.  This
simple, rugged,  inexpensive system provided the two necessary responses to
different energy ranges of the backscattered beta particles.

                            THE PROTOTYPE DEVICE
     A prototype dual-detector backscatter beta-particle gauge has been con-
structed that consists of two thin end-window GM counters and a shielded Sr-90
radioisotope source as shown in Figure 1.   The GM counters were made by the
Amperex Company and have mica end windows with density thicknesses of 0.0014
g/cm .   The Sr-90 source has a nominal intensity of 0.1 mCi.  It is encapsu-
lated in a welded aluminum cylinder with a diameter of 0.3 cm and a length of
2.2 cm.  The source itself is in the shape of a circular disc and is placed
at one end of the aluminum—encapsulation cylinder with a covering of only
about 0.005 cm.  One of the GM counters is used with a paper filter with
density thickness in the range of 0.0077 to 0.050 g/cm2 placed over the end
window.

                          GAUGE CALIBRATION MODELS
BAG FILTER AND DUST CAKE

     In the case of a bag filter with accumulation of dust cake on the inside,
the model for a single beta-particle backscatter gauge mounted outside the bag
filter can be taken as (3):

          R = Bf(d)[l-exp(-vy)] + Cf(d)exp(-v'y)[l-exp(-ux)] + N          (1)

where R is the gauge response in counts per second (c/s); B, C, N, u, v, and
v1 are model parameters to be determined by a least-squares analysis of cali-
bration data; f(d) represents the effect of sample distance d (in cm) on the
backscatter response; y is the bag density thickness in g/cm2; and x is the
dust cake density thickness in g/cm2.  The parameters B and C are the satura-
tion responses in c/s for infinitely thick samples of the bag filter and dust
cake (by itself), respectively.  The parameter N is the background counting
rate in c/s when neither bag filter nor dust cake is present; v and u are the
backscatter coefficients for the bag filter and dust cake in cm2/g, respective-
ly; and v' is the attenuation coefficient of the backscattered beta particles
through the filter bag in cm2/g.

ESP COLLECTOR PLATE AND DUST CAKE

     In the case of an ESP collector plate with accumulation of dust cake on
the same side as the gauge, the model for a single beta-particle backscatter


                                      79

-------
gauge can be taken as:

          R = Cf(d)[l-exp(-ux)] + Pf(d)exp(-u'x)[l-exp(-wz)] -f N          (2)

where the parameters are the same as in Equation 1; P, w, and u' are the model
parameters to be determined by a least-squares analysis of calibration data;
and z is the density thickness of the collector plate in g/cm2.  The parameters
P and w are the saturation response and backscatter coefficient for the collec-
tor plate in c/s and cm2/g, respectively.  The parameter u' is the attenuation
coefficient for the backscattered beta particles through "the dust cake in
cm2/g.  In the present case, when an infinitely thick collector plate is of
interest, Equation 2 becomes:

          R = Cf(d)[l-exp(-ux)] + Pf(d)exp(-u'x) + N                      (3)

                   GEIGER-MUELLER TUBE DETECTOR DEAD TIME
     It was found that for the Geiger-Mueller (GM) tube detector and the
source used in the beta-particle backscatter gauge, a considerable counting
loss due to detector dead time was often encountered.  To correct for this,
the model for observed (R) and true (R') counting rates, in terms of the
detector dead time (r) for a paralyzable detector (3), was used.  This model
is:

                              R = R'exp(-rR')                             (4)

The two-source method (2) utilizing the exact model of Equation 4 was used to
obtain the detector dead time (r) .  This method consisted of obtaining the
counting rates of the background, Source 1 plus background, Sources 1 and 2
plus background, and Source 2 plus background, in that order.  Then the
appropriate model of Equation 4 was used, and iterations on trial values of r
were made until all the experimental and calculated values of the counting
rates agreed to within a specified amount.

     For example, for the counting rates described of 1.10, 1094.00, 2046.00,
and 1309.00 c/s, respectively, the detector dead time is found to be 114.6
usec.  The average dead time found for several determinations for the GM tube
used here was 115.0 ysec.

                             LABORATORY RESULTS
BAG FILTER APPLICATION

     The prototype gauge was used to obtain results for various paper thick-
nesses on top of a piece of bag filter (density thickness of 0.0375 g/cm2) at
various distances to simulate bag filter dust cake samples.  Results were
obtained for a number of detector filter thicknesses in the range from 0.0077
to 0.050 g/cm2 on one of the detectors while the other detector was used with-
out a filter.  When the bag filter density thickness is constant, as in the
present case, Equation 1 can be reduced to:

                                      80

-------
                    R = B'f(d)  + C'f(d)[l-exp(-ux)]  + N                   (5)

where B'  and C'  are B[l-exp(-vy)]  and Cexp(-v'y), respectively.   The form of
f(d) was  found to be adequately represented by:

                              f(d) = exp(-ad)                             (6)

where the parameter a in cm"1 is found by a least-squares analysis of data.
However,  it was  found that the value of a was not equal in both places that  it
appears when Equation 6 is substituted in Equation 5.  In this case Equation 5
must be taken as:

               R = B'exp(-ad) + C'exp(-a'd)[l-exp(-ux)] + N               (7)

     While a rigorous complete optimization has not been completed, the
results obtained so far indicate that the detector filter thickness of
0.050 g/cm2 is the most promising.  Results for this case are given in Table 1
and plotted in Figures 2 and 3.

     Note that one may use the derived response obtained by subtracting the
filtered  GM tube response from the unfiltered GM tube response as well as the
two direct responses.  From these three possible responses there are three
possible  combinations of the two required responses.  The optimum two responses
for a particular case depend upon the detector filter thickness being used.
However,  in most cases the two optimum responses are the filtered GM tube
response  and the unfiltered minus the filtered GM tube response, since this
combination gives the two responses that represent the extremes in back-
scattered beta-particle source energies.  This was the case for the 0.050 g/cm2
filter.  The choice of the optimum two responses is also affected by the fact
that the  derived response is somewhat correlated to each of the two direct
responses, while the two direct responses are completely uncorrelated.

     The  model parameters obtained by a least-squares analysis of the data
were:  u  = 15.593, a' = 0.29432, a = 0.31036, Cf = 98.477, B' = 244.880, and
N = 13.77 for Rl; and u = 11.059, a' = 0.20905, a = 0.36427, C' = 35.207,
B1 = 62.746, and N = 11.66 for R2.  The reduced x2 values, assuming that the
standard  deviations were 0.01 R, were 2.98 and 5.08, respectively.  When only
dust cake density thicknesses less than 0.11 g/cm2 are considered, the measure-
ment standard deviation of the dust cake density thickness is reduced to
0.0178 g/cm2.  When the exact known distances are used in Equation 7 with
experimental values of R2 and the model parameters for this case, the standard
deviation of the measured values of the dust cake density thickness is reduced
to only 0.0134 g/cm2 for all cases and 0.0036 g/cm2 for only those samples
with dust cake density thickness less than 0.11 g/cm2.

ESP APPLICATION

     The prototype gauge was also used to obtain results for various paper
thicknesses at various distances on a 0.16 cm thick stainless steel plate to
simulate the dust cake and collector plate samples in an operating ESP unit.
The same range of distances  (3.4 to 7.4 cm) and simulated dust cake density
thicknesses  (0 to 0.2133 g/cm2) were investigated as with the bag filter case.

                                      81

-------
The optimum gauge detector filter again appears to be that of 0.050 g/cm2, and
the two optimum responses are again the filtered response and the unfiltered
minus  the  filtered response.  The results are given in Table 2 and shown
plotted in Figures 4 and 5.

     The appropriate model in this case (when there is an infinitely thick
stainless  steel plate) is given by Equation 3 using the exponential form for
f(d) that  appears in Equation 3.  However, it is also found that the same
value  can  be used for u1 and u in Equation 3.  With these changes the model
for the ESP application becomes:

           R = Cexp(-ad)[l-exp(-ux)] + Pexp(-a'd) exp(-ux) + N             (8)

     The model parameters obtained by a least-squares analysis of the data
were:  u = 16.853, a = 0.29881, a1 « 0.25512, C = 210.370, P = 750.847, and
N = 13.84  for Rl; and u = 11.516, a = 0.22239, a' = 0.22056, C = 22.005,
P = 480.422, and N = 11.16 for R2.  The reduced x2 values, assuming that the
standard deviations were 0.01 R, were 2.24 and 2.02, respectively.  When only
dust cake  density thicknesses larger than 0.050 g/cm2 are considered, the
measurement standard deviation of the dust cake density thickness is reduced
to 0.00585 g/cm2.  When the exact known distances are used in Equation 8 with
the experimental values of R2 and the model parameters for this case, the
standard deviation of the measured values of the dust cake density thickness
is reduced to only 0.00255 g/cm2 for all cases and 0.00262 g/cm2 for only
those  samples with dust cake density thickness larger than 0.050 g/cm2.

                               ERROR ANALYSIS
     In addition to determining the standard deviations for the measurement of
dust cake density thickness by comparing the measured to the known values, the
measurement errors are also obtained by a random number method.  The method
consists of first randomly choosing the dual-detector responses according to
Gaussian distributions of specified standard deviation.  Then 10 or more sets
of these randomly chosen responses are used to determine measured values of
the dust cake density thickness.  Finally, the various measured values obtained
by this method are used in the general estimator to obtain the measured value
of the standard deviation or error of the dust cake density thickness measure-
ment for the particular distance and dust cake density thickness of interest.
This has been done for the range of distances and dust cake density thicknes-
ses previously used, and the results are given in Table 3 for Gaussian distri-
buted dual-detector responses with standard deviations of 1 percent.

     Examination of the results in Table 3 indicates that the standard devia-
tions or errors calculated by the random number method appear to be consis-
tent with the actual errors given in Tables 1 and 2.  The random number errors
are approximately two-thirds of the values actually obtained, indicating that
a 1 percent error in the dual-detector responses is a fairly representative
estimate.

     Note that the predicted error increases with increasing dust cake density
thickness and is almost independent of distance for the bag filter case.  The

                                      82

-------
predicted error decreases with increasing dust cake density thickness for the
smallest distance (3.4 cm) for the ESP case,  but then exhibits a minimum at
0.16 g/cm2 for the 4.9 cm distance and at 0.11 g/cm2 for the 7.4 cm distance.

                         DISCUSSION AND CONCLUSIONS
     From the results it is clear that the dual-detector beta-particle back-
scatter gauge described and demonstrated here should prove very useful for the
measurement of dust  cake density thickness in bag filter and ESP units in those
cases where the distance to the device is unknown and variable.  While the
accuracy attainable  is not as good as that for the single gauge (2), it is
significantly better than that for the previously reported dual-beam gauge (1).
For the bag filter case the standard deviation for the dust cake density
thickness is 0.0680  g/cm2 for the dual-beam gauge (1), 0.0196 g/cm2 for the
single gauge used at a known distance (2), and 0.0278 g/cm2 for the present
dual-detector gauge.  For the ESP case the standard deviation for the dust
cake density thickness is 0.00496 g/cm2 for the single gauge used at a known
distance (2) and 0.0151 g/cm2 for the present dual-detector gauge.

     The accuracy obtained with the present dual-detector gauge is intermedi-
ate between that for the single gauge used at a known distance and that for the
previously reported  dual-beam gauge.  There are several compensating advan-
tages for the present dual-detector gauge over the previously reported dual-
beam gauge.  First of all, there are no moving parts in the dual-detector
gauge to become fouled in the harsh environment of the bag filter or ESP
units.  Secondly, there is only one radioisotope source required, which means
that there are no problems of matching source intensities or making more than
one correction for decay.

     It is likely that better accuracy can be obtained for a dual-gauge device
of some other design.  For example, one might use an appropriate ultrasonic
device to make an accurate separate measurement of the distance of the bag
filter or collector  plate from the device and then use a simple single beta-
particle backscatter gauge to measure dust cake density thickness.  The limit-
ing accuracy attainable with any dual-gauge device of this type is 0.0134 g/cm2
and 0.00255 g/cm2 for the bag filter and ESP units, respectively.  These
values are obtained  with the Sr-90 source beta-particle backscatter gauge when
the distance to the  device is known exactly and a 0.050 g/cm2 filter is used
over the GM detector end window.

                                 REFERENCES
     Gardner, R.P., Donovan, R.P., and Hovis, L.S.  A dual-beam backscatter
     beta-particle gauge for measuring the dust cake thickness on operating
     bag filters independent of position.  In:  Third Symposium on the Trans-
     fer and Utilization of Particulate Control Technology:  Volume III.
     Particulate Control Devices, EPA-600/9-82-005c (NTIS PB 83-149609),
     July 1982, pages 128-139.
                                       83

-------
2.    Gardner,  R.P.,  Donovan,  R.P.,  and Hovis,  L.S.   The one-point in-situ
     calibration method for using a beta-particle backscatter gauge for
     continuously measuring dust cake thickness on operating bag filter and
     ESP units.  In these Proceedings.

3.    Gardner,  R.P. and Ely, Jr., R.L.  Radioisotope Measurement Applications
     in Engineering.  Reinhold Publishing Corporation,  New York, 1967.
                                     84

-------
Figure 1.   Schematic drawing of the prototype dual-detector
           beta-particle backscatter gauge.
                              85

-------
   150
      0
 0.050        ^0.100        0.150        0.200



   DUST CAKE DENSITY THICKNESS  M, g/cm''
                                                                       0.250
 Figure 2.   Simulated bag  filter experimental  and model  results for Gauge 1.


    60
CO


o
w
H
o
z
I— I
H
o
u
      0
0.050        0.100        0.150         0.200



   DUST CAKE DENSITY THICKNESS (x), g/cnr"
0.250
Figure  3.  Simulated bag filter experimental  and model  results for Gauge 2.




                                     86

-------
  350
w

u
                 0.050        0.100         0.150          0.200

                    DUST CAKE DENSITY THICKNESS  (x),  g/cm2


  Figure 4.  Simulated ESP  experimental and model results for Gauge 1.

   250
   200 -
   150 -
                                                                       0.250
06


7.
M
H
2
3
O
   100'
                 0.050         0.100         0.150        0.200

                    DUST  CAKE DENSITY THICKNESS (x),  g/cm2


   Figure  5.   Simulated  ESP experimental and model results for Gauge 2.


                                    87
                                                                       0.250

-------
TABLE 1.  LABORATORY CALIBRATION RESULTS  FOR THE DUAL-DETECTOR BETA-PARTICLE
          BACKSCATTER GAUGE RESPONSES  FOR SIMULATED BAG FILTER OPERATION WHEN
          ONE DETECTOR HAS A FILTER OF 0.050 g/cm2


Sample
Distance
(d), cm
CO
3.4
3.4
3.4
3.4
3.4
4.9
4.9
4.9
4.9
4.9
7.4
7.4
7.4
7.4

Dust Cake
Density
Thickness
(x), g/cm2
_ __
0.0
0.0499
0.1062
0.1618
0.2133
0.0
0.0499
0.1062
0.1618
0.2133
0.0
0.0499
0.1062
0.1618
7.4 0.2133
Standard Deviations
Experimental
Unfiltered Minus
Filtered Gauge
Response3
(Rl), c/s
13.77
100.26
118.57
129.33
131.87
134.95
66.37
79.66
85.94
86.94
88.88
38.48
44.40
48.40
48.19
49.77
Calculated
Filtered
Gauge
Response
(R2), c/s
11.66
29.39
37.52
41.50
44.41
45.86
22.47
27.16
30.92
32.98
33.25
15.90
19.22
20.93
22.45
22.50

Sample
Distance
(d), cm
	
2.885
3.460
3.311
3.429
3.375
5.136
4.838
4.914
5.028
4.927
7.385
7.448
7.229
7.494
7.306
0.170
Dust Cake
Density
Thickness
(x) , g/cm2
	
0.01787
0.05784
0.08929
0.17667
0.21948
0.00907
0.04169
0.10790
0.22371
0.19824
-0.00023
0.05536
0.08587
0.21892
0.16842
0.0278
Corrected for dead-time losses using Equation 4 with r = 115 ysec for the
 unfiltered gauge responses and r = 223 psec for the filtered gauge responses.

'Corrected for dead-time losses using Equation 4 with r = 223 ysec.
                                      88

-------
TABLE 2,   LABORATORY  CALIBRATION  RESULTS  FOR  THE  DUAL-DETECTOR BETA-PARTICLE
          BACKSCATTER GAUGE  RESPONSES  FOR SIMULATED  ESP  OPERATION WHEN  ONE
          DETECTOR HAS A  FILTER OF  0.050  g/cm2



Experimental
Dust Cake Unfiltered Minus
Sample
Distance
(d), cm
00
3.4
3.4
3.4
3.4
3.4
4.9
4.9
4.9
4.9
4.9
7.4
7.4
7.4
7.4
Density
Thickness
(x), g/cm2
	
0.0
0.0499
0.1062
0.1618
0.2133
0.0
0.0499
0.1062
0.1618
0.2133
0.0
0.0499
0.1062
0.1618
7.4 0.2133
Standard Deviations
Filtered Gauge
Response3
(Rl), c/s
13.84
335.41
190.58
128.37
105.23
96.47
235.94
138.61
91.58
76.66
66.05
125.38
75.72
54.00
42.68
38.86
Calculated
Filtered
Gauge
Response"
(R2), c/s
11.66
234.78
144.02
83.60
53.75
40.72
173.90
110.32
65.46
41.79
32.15
100.79
66.08
41.77
28.88
23.22

Sample
Distance
(d) , cm
	
5.226
3.723
3.404
3.376
3.421
6.426
5.017
4.871
4.608
4.986
8.904
7.588
7.023
7.377
7.494
0.771
Dust Cake
Density
Thickness
(x) , g/cm2
	
-0.03501
0.04279
0.10838
0.16598
0.20966
-0.03024
0.04346
0.10481
0.17262
0.20916
-0.02576
0.04568
0.11475
0.16516
0.20911
0.0151
3Corrected for dead-time losses using Equation 4 with r = 115 ysec for the
 unfiltered gauge responses and r = 223 ysec for the filtered gauge responses.

 Corrected for dead-time losses using Equation 4 with r = 223 ysec.
                                      89

-------
TABLE 3.  RANDOM NUMBER ERROR ANALYSIS  OF  THE  DUAL-DETECTOR BETA-PARTICLE
          BACKSCATTER GAUGE FOR BAG FILTER AND ESP DUST CAKE MEASUREMENTS
          WHEN ONE DETECTOR HAS A FILTER OF  0.050 g/cm2

Experimental
Calculated
Bag Filter
Sample
Distance
(d), cm
3.4
3.4
3.4
3.4
3.4
4.9
4.9
4.9
4.9
4.9
7.4
7.4
7.4
7.4
Dust Cake
Density
Thickness
(x), g/cm2
0.0
0.0499
0.1062
0.1618
0.2133
0.0
0.0499
0.1062
0.1618
0.2133
0.0
0.0499
0.1062
0.1618
7.4 0.2133
Averages
Sample
Distance
(d), cm
0.260
0.120
0.0762
0.0641
0.0656
0.0961
0.133
0.0754
0.0674
0.0535
0.127
0.102
0.0821
0.0514
0.0382
0.0941
Dust Cake
Density
Thickness
(x), g/cm2
0.0105
0.00995
0.0138
0.0183
0.0498
0.00359
0.0121
0.0131
0.0259
0.0495
0.00459
0.0114
0.0131
0.0301
0.0314
0.0198
ESP
Sample
Distance
(d) , cm
2.31
0.268
0.117
0.0674
0.0348
1.53
0.268
0.108
0.0647
0.0507
2.13
0.296
0.0915
0.105
0.0543
0.499
Dust Cake
Density
Thickness
(x) , g/cm2
0.0476
0.00593
0.00366
0.00282
0.00265
0.0309
0.00660
0.00340
0.00232
0.00296
0.0421
0.00651
0.00258
0.00362
0.00398
0.0112
                                   90

-------
              MIT FLEX ENDURANCE  TESTS  AT ELEVATED TEMPERATURE


                     John  T.  Foster  and W.  Theron Grubb
                           Wheelabrator-Frye Inc.
                      Air  Pollution  Control Division
                              Technical  Center
                           5100 Casteel Drive
                     Coraopolis,  Pennsylvania  15108
                                ABSTRACT

     The MIT flex  endurance test,  originally devised for paper,  is used
to evaluate the  fatigue  life of  woven filtration fabrics.  This  test is
used in predicting the relative  life of fiberglass fabric filter bags and
in the development of finishes.   In the past,  MIT flex data has  been
published  only for tests conducted at ambient  (room) temperature.   Fiber-
glass filter bags, however,  are  used at operating temperatures up to 550 F.
Since the  physical properties of the finishes  which protect the  cloth
change with temperature, MIT flex tests conducted at the operating tempera-
ture of the cloth  should provide more meaningful data.

     An MIT flex tester  designed to meet the specifications of ASTM method
D2176, and to operate at temperatures as high as 600 F, has been built and
tested. A relationship  between  flex endurance and operating temperature
for several commercially finished fiberglass fabrics is presented.

                              INTRODUCTION

     During the  cleaning cycle in the operation of a fabric filter the
filter media is  subjected to some form of flexing regardless of  whether
shaker, reverse  air or pulse jet cleaning is used.  It is this flexing
during cleaning  that is  particularly deleterious to bag life in  the form
of fatigue failure.

     Fiberglass  cloth is most commonly used in boiler applications for its
low cost and high  temperature capabilities.  However, while individual
glass fibers have  relatively high tensile strength, the abrasive effect of
one glass  fiber  rubbing  against  another during flexing of the cloth causes
failure after very few cycles.  In order to reduce the self abrasive effects
of fiberglass cloth, temperature resistant finishes with high lubricity


                                     91

-------
are commonly applied to the cloth during manufacture.

     The MIT flex endurance test, originally developed to evaluate the
fatigue life of paper and plastics, is a widely used method for evaluating
the effectiveness of the finishes used to protect fiberglass cloth from
abrasion.  Published MIT flex data exists for the commercially available
finishes but only at ambient laboratory temperatures (1). Since fiberglass
cloth filter bags are used at temperatures ranging from 300 F to 550 F,
finishes that protect the cloth would be more meaningfully evaluated if done
at baghouse operating temperatures.

     Part of Wheelabrator-Frye's commitment to study bag life and the
factors affecting bag life involved the construction of an MIT flex tester
designed to operate at temperatures up to 600 F.  The hot MIT flex tester
was constructed to operate within the requirements of ASTM test method
D2176 and duplicate the motion of the Tinius Olsen MIT flex tester.  Tests
were conducted between early and mid 1982 comparing the performance of the
hot MIT flex tester operating at ambient laboratory temperature and the
Tinius Olsen MIT flex tester operating at the same temperature.  Once
enough data had been collected to determine that the Wheelabrator-Frye
hot MIT flex tester and the Tinius Olsen MIT flex tester produced the same
results, a program to evaluate several finishes at elevated temperatures
was initiated. Three commonly used finishes on the same fiberglass cloth
substrate were compared at 300 F, 400 F and 500 F.  A discussion of the
construction of the hot MIT flex tester and the results of the fiberglass
cloth finish comparison at elevated temperatures is contained herein.

                    DESIGN OF THE HOT MIT FLEX TESTER

     The Wheelabrator-Frye hot MIT flex tester was designed to conform to
the requirements for the test apparatus described by ASTM method D2176-69
paragraph 2.1.1 to 2.1.4.  The motion requirements for the oscillating
folding head described in the above ASTM designation includes rotary
oscillating motion 135  ± 2  both to the right and left of the unfolded
specimen as shown in Figure I.  The required folding frequency is 175 ± 25
left and right cycles per minute.

     Parameters of the oscillating motion not considered by ASTM D2176-69
are the description of displacement, velocity and acceleration through the
folding cycle.   The universally accepted Tinius Olsen MIT flex tester
makes use of a Scotch yoke oscillating mechanism, like the one shown in
Figure  2,  to impart oscillating motion to the lower sample clamp through
a rack and pinion drive.
                                     92

-------
                  SPECIMEN
135°rfc 2°
135°± 2°
      FIGURE 1.  OSCILLATING LOWER HEAD
                       93

-------
                                  I

FIGURE 2.  SCOTCH YOKE OSCILLATING MECHANISM WITH RACK AND PINION
                               94

-------
        DISPLACEMENT
        VELOCITY
       ACCELERATION
                                                      S = R cos
                      0°    90°   180°  270°   360C
                     FIGURE 3.   SIMPLE HARMONIC MOTION
A mechanism of this type produces  simple harmonic motion as shown in
Figure  3.   In order to maintain  comparable performance between the new
hot MIT flex tester and the  Tinius Olsen machine, an oscillating mechanism
was designed to produce simple harmonic motion.  Instead of machining
the necessary parts to duplicate the  Scotch yoke mechanism, the Ferguson
right angle cam driven oscillating mechanism shown in Figure 4  was pur-
chased.

                             OUTPUT SHAFT
                                                 INPUT
                                                SHAFT
           FIGURE 4.  RIGHT ANGLE CAM DRIVEN OSCILLATING MECHANISM
                                    95

-------
This mechanism was  selected for its low cost,  durability and internal
lubrication.   Design  limitations inherent in the  right angle cam drive
mechanism do  not  permit oscillation at the output shaft to be greater than
90  in each direction.  The drive cam was custom  cut by Ferguson according
to Wheelabrator's specification, resulting in  simple harmonic motion at
the output shaft  of the drive mechanism for a  67  1/2° left and 67 1/2°
right rotation. Left  and right 135  oscillation at  the two output shafts
used by the flex  tester was achieved through 2:1  gearing.
                         ~V    ^~i
                                    ^-1
                   • -Iff-

              FIGURE 5.  GENERAL LAYOUT OF HOT MIT  FLEX TESTER
     The photograph  in Figure 5 shows the general  layout of the drive and
oscillating .mechanisms.  At 175 cycles/minute the  Ferguson oscillating
drive is rated  at  760 in-lbs torque.  Since the  maximum torque required to
flex a sample under  maximum load is 4 in-lbs, the  dual output shaft arrange-
ment shown in the  photograph was chosen to double  the sample testing
capacity of the machine.  A magnetic proximity switch on each individual
shaft is wired  in  series with a limit switch that  is tripped when the
sample breaks.  This serves to separately register left and right sample
flex cycle counts  on two 6 digit electromechanical counters.

     The photograph  in Figure  6 shows the dual  oscillating heads, limit
switch housing, counters and oven controller. Upper and lower clamps,
primary sheaves and  dead weights are Tinius Olsen  parts used in the
design of the new  machine.  In addition to the aforementioned counter limit
switches, a second pole on each of the two switches in the housing is wired
                                    96

-------
in series to the motor switch to stop the oscillation once both samples
have broken.

     The oven is constructed of two layers of one inch thick Johns-Manville
Marinite I separted by a 1" dead air space and rigid frame.  Power to the
oven is supplied by two 300 watt Chromalox strip heater elements and
regulated by a Chromalox 3800 series analog controller using a J type
thermocouple.
                                  t  T
                FIGURE 6.  HOT MIT FLEX TESTER - FRONT VIEW
                                     97

-------
                          HOT MIT FLEX TESTING

     In order to determine the variation in results between the
Wheelabrator-Frye hot MIT flex tester and the Tinius Olsen flex tester, a
series of flex tests were run using .03" lower clamps and a 14 oz/yd
fiberglass cloth.  The tests were all conducted at ambient temperature with
the oven doors of the hot flex tester held open.  Each of the three .03"
clamps was tested on the Tinius Olsen flex tester and in each of the two
positions on the hot flex tester.  Data from 100 sets of specimens (three
per set) showed no variation between the Tinius Olsen machine and either
left or right sides of the hot flex tester.

     The Marinite I material used to fabricate the oven is hygroscopic in
nature due to the sodium silicate binder that it contains.  Initial
warmup of the oven after setting several hours at ambient temperature was
always accompanied by condensation on the adjacent metal surfaces of
the machine.  Trial tests conducted immediately after the oven was warmed
up from ambient temperature consistently showed a lower number of cycles
to failure than tests conducted on the same cloth after an eight hour oven
preheat.  The lower flex count is probably due to higher humidity as-
sociated with driving the excess moisture off of Marinite material during
warmup.  This assumption is consistent with other published data (2).  All
hot flex test data presented in this paper is for samples tested after a
minimum eight hour preheat if the oven had been permitted to cool to
ambient temperature.

     Fiberglass cloth must be protected from the gripping surface of the
upper and lower clamps during MIT flex testing.  At ambient temperatures
ordinary masking tape is sufficient for this purpose.  However, at tempera-
tures in the 300 F - 500 F range the masking tape deteriorates and the test
specimen begins to slip from the clamp.  After trials with several types of
high temperature tape, DuPont Kapton polyimide tape was selected.  The
Kapton tape shows no change in properties over the range of test tempera-
tures and has the ability to be impressed into the interstices of the
weave of the specimen to prevent slipping.  Kapton tape was used on all
samples tested at elevated temperature.

             EVALUATION OF THREE COMMON FIBERGLASS FINISHES

     Three commonly used fiberglass cloth finishes were evaluated at 300 F,
400°F and 500°F using the hot MIT flex tester.  In order to hold all para-
meters except the finish and temperature constant, the same substrate
cloth, given below, was selected for all three finishes:
                                     98

-------
                   TABLE 1.  SUBSTRATE CLOTH CONSTRUCTION
                    Warp

                    Fill

                    Weave

                    Count
       ECDE 37 1/0

       ECDE 75 1/3 (2 texturized + 1 filament)
       3x1 twill

       44 x 24 (warp x fill)
     Samples with the selected finishes were obtained from the manufacturers
and tested for compliance to that manufacturer's specifications with regard
to permeability, weight and LOI.
                  TABLE 2.  FINISHED FABRIC SPECIFICATIONS

Finish
Type
Weight (oz/yd2)
Permeability (CFM/Ft2)
A
Silicone-Graphite-PTFE
*13.09/13.2
58.4/50-65
± 5%
B
Acid Resistant
13. 32/13. 5±5% 14
43.2/35-50 45
C
PTFE
.00/14. 2±5%
.0/35-50
-Avg. of warp and fill
 sides

LOI (%)
1.79/1.40
4.19/3.80+     9.44/9.00"*
  Tested/manufacturer's spec.

  Minimum
     The effect of heat aging while flexing at elevated  temperatures was
eliminated by heat aging the samples at the intended  test  temperature  for
twenty-four hours in a stationary convection oven prior  to testing.  Speci-
mens heat aged at 300°F, 400°F, 500°F and a set of virgin  specimens were
tested at ambient temperature on the Tinius Olsen flex tester.   Another
group of specimens heat aged at 300 F, 400 F and 500  F were tested ?.t  the
 corresponding elevated temperature in the hot MIT  flex  tester.
 fill directions of each sample were evaluated in this manner.
                                     Warp and
     To minimize the effect of normal variations  in  the  fabric,  samples
 were prepared as shown in Figure  7.   This method permits  the comparison
 of one specimen heat aged at each temperature along  the  same warp or fill
 yarns.
                                     99

-------
Ill
o
o
LU
LU
CO
0
0
z
X
i
o
u.
z
III

H
f-
o
2
CO
z
LU

u
LU
ft
4
>
J
1
4

^
J

20 WARP SPECIMENS
HEAT AGED
f
@ 300°F
20 WARP SPECIMENS
HEAT AGED
r
@ 400°F
h
20 WARP SPECIMENS
HEAT AGED

h
PERM TEST
LOI TEST
^ r


20 FILL
£*n^/^ i
SPECI -
MENS
HEAT

AGED
@ 300°F

j k.



@ 500°F

6
VIRGIN
WARP
SPECI-
MENS


20 FILL
c*r*cr*t
SPECI
•
MENS
HEAT


AGED
@ 400°F
















20 FILL
onc^i
SrtCI -
MENS
HEAT

AGED
@ 500°F


6
VIRGIN
SPECI -
MENS





















W
A'
R
P
1
4 	 FILL 	 *^






LU
0
LU
_l
LU
CO










                       FIGURE 7.   SAMPLE PREPARATION
     Table III shows the results of ambient and elevated temperature
testing.  Each data point is the average of; 6 specimens for each virgin
fabric test, 5 specimens for each of the ambient temperature tests on heat
aged fabric, and 10 specimens for each of the elevated temperature tests
on heat aged fabric.  The value for a specimen was not used in the average
if it was outside of the 95% level of certainty range for the group of
specimens (3).
                                    100

-------
                      TABLE 3.  HOT MIT FLEX TEST DATA
Finish    Virgin •

        Specimens
(Cycles to Failure - Warp Direction)

Heat Aged @ 300°F   Heat Aged (jj 400°F   Heat Aged  @  500°F
Ambient   300 F
Ambient
400°F
Ambient   500 F
   A      1900       1550     7780

   B      4050       4800    32900
   C      9800      44300    47100

                    (Cycles to Failure
                     2890        16900

                    10100        42700

                    21600        53100
                    - Fill Direction)
                     4330     14900
                     9620     28600
                     8840     13300
A
B
C
444
1010
1170
323
1190
3230
464
2620
3810
457
1810
1600
453
2220
3250
580
2000
492
510
1730
567
     The data from Table III has been  compared  graphically in Figures  8
a-f.  In each figure the coordinate axis  represents  the heat ag^d tempera-
ture of the specimen.  Comparison of the  absolute  values for flex testing a
heat aged specimen at ambient temperature and its  corresponding elevated
temperature is not possible because the effect  of  the reduced relative
humidity at elevated temperature on flex  life is not presently known.  How-
ever, a difference in the shape of the two curves  would imply that the re-
sponse of a given finish to elevated temperature testing is different from
the measured effect of heat aging alone.   The parallel curves in Figure 8
a and b show that no relative difference  in the response of finish A is
achieved by elevating the test temperature.
50 -
x 40 -
UJ
oc
D
^J
5 30 -
u.
g
S 20 -
J
u
u
10 -









x 	 	 	 *
^
*— "'"^
T 	 *i~~ 	 i
300 400 500
TEMP (F°)
                                           4  -_
                                            3   .-
                                            2   --
                                            1   --
                                                    •  •  • TESTED AT AMBIENT TEMP.

                                                    XXX TESTED AT ELEVATED TEMP.
                                                    300
                                           400
                                      TEMP (F0}
                                  500
            FIGURE  8  a.
 FINISH A - WARP DIRECTION FLEX TESTS
                                FIGURE 8 b.
                    FINISH A - FILL DIRECTION FLEX TESTS
                                      101

-------
    50  T-
  x 40
  IU
  oc
  «* 20  --
  u
  >
  u
     10  --
              300
     400
TEMP (F°)
     5


     4


     3


     2
500
                               •  •   •  TESTED AT AMBIENT TEMP.

                               XXX  TESTED AT ELEVATED TEMP.
300
                                                                 4-
      400
TEMP (F°)
500
 FIGURE    8  c.                            FIGURE    8  d.
FINISH B  - WARP DIRECTION FLEX TESTS   FINISH B  - FILL DIRECTION FLEX TESTS
    50  T
   x 40
   oc
   3
   5 30
   u.
   g
   J2 20
   u
   u
     10  - -
              300
     400
TEMP (F°)
                    15

                    18
     4  _-


     3


     2


     1  -L
                                • •  • TESTED AT AMBIENT TEMP.

                                xxx TESTED AT ELEVATED TEMP.
500
300
                                                                 4-
      400
TEMP (F°)
500
 FIGURE  8   e.                            FIGURE
FINISH C - WARP  DIRECTION FLEX TESTS   FINISH C
                                 8   f.
                                 FILL DIRECTION  FLEX TESTS
                                        102

-------
     The t test is used to determine whether the mean values of two sets of
data belong to the same population (statistically the same value) or dif-
ferent population groups (are statistically different).  The following set
of equations was used to determine the t value between several sets of
points in Figures   8  c through   8   f:
          SW   =l(n1-1)(s1 ) + (n2~l)(S2 )     Xj = mean value of first data

                1	nl + n2 - 2     	          S6t
                                               "S2 = mean value of second data
                                                    set

                                               iL = standard deviation for
                                                    first data set
                                               S_ = standard deviation for
                 	                            second data set
                                               n1 = number of points in first
                                                    set
                                               n~ = number of points in
                                                    second data set

t values between points were then compared to the t value for a 95% level
of confidence that the two means are from different population groups.  The
results of the t test shows that for Figure   8  c-f:

          Point    1^2
                   2 t  3
                   4 1  5
                   5=6
                   8 =  9 = 11 = 12
                   7 1 10
                  13 = 14 1 15
                  16 = 13 4 17 t 18
                  19 = 20 1 21
                  22 1 23 t 24

     The results of the above analysis show that the shapes of the elevated
and ambient temperature test curves are statistically different from each
other for finishes B and C.  These nonparallel curves imply that the re-
sponse of finishes B and C to elevated temperature testing is not due
entirely to the effects of heat aging.  Insufficient test data at elevated
temperature precludes any further conclusions with regard to how the
finishes compare to one another at elevated temperature.

     The fill direction with finish B, Figure   8  d, shows a decrease in
flex life with increasing temperature on the elevated test temperature
curve compared to an increase in flex life with increasing temperature on
the ambient test temperature curve.
                                     103

-------
One theory for this behavior is a loss of the finish's binding strength at
elevated temperature permitting the glass filaments in the fill yarns, many
of which are broken during the texturizing process, to pull apart rather
than breaking cleanly.  The photograph in Figure 9  shows heat aged fill
direction specimens tested at elevated and ambient temperatures.  The fill
yarns of the specimen tested at elevated temperature appear to have pulled
apart to a greater degree than the specimen tested at ambient temperature.
                  FIGURE 9  a.   AMBIENT TEMPERATURE SPECIMEN
                                     104

-------

                FIGURE 9  b.   ELEVATED TEMPERATURE SPECIMEN
                                   SUMMARY

    The hot MIT flex tester successfully duplicates,  at elevated tempera-
ture, the test described by ASTM D2176-69 and commonly performed by the
Tinius Olsen MIT flex tester.

    Heat aged fiberglass cloth with  certain finishes  responds differently
to flex testing at elevated temperature  compared  to  flex testing at ambient
temperature.  Some finishes exhibit a reduction in binding strength at
elevated temperature as shown by the  fill yarns pulling apart during testing
in the fill direction.

    The work described in this paper was not funded by the U.S.
Environmental Protection Agency and therefore the contents do not necessarily
reflect the views of the Agency and no official endorsement should be in-
ferred.
                                     105

-------
                                REFERENCES

1.  Knox, C.E., J. Murray, and J. Schoeck.  Technology of Fiber Glass Filter
    Fabric Design.  EPA Report EPA-600/7-79-044b Proceedings of Symposium on
    the Transfer and Utilization of Particulate Control Technology Vol. 2,
    1979.

2.  Boudreau, R.J., W. Hoffman and S.C. McCluskey.  The Manufacture and
    Quality Assurance Testing of Fiberglass Filter Fabrics.  Proceedings
    of the First Conference on Fabric Filter Technology for Coal-Fired
    Power Plants, Denver, Colorado, February 1982.

3.  Chou, Y.  Statistical Analysis 2nd Ed., Holt, Rinehart, Winston, 1969.
                                     106

-------
     THE ONE-POINT IN-SITU CALIBRATION METHOD FOR USING A BETA-PARTICLE

     BACKSCATTER GAUGE FOR CONTINUOUSLY MEASURING DUST CAKE  THICKNESS  ON

                     OPERATING BAG FILTER AND ESP UNITS

               By:  Robin P.  Gardner
                    Center for Engineering Applications of Radioisotopes
                    P.O.  Box  5636
                    North Carolina State University
                    Raleigh,  North Carolina  27650

                    R.P.  Donovan
                    Process and Chemical Engineering Division
                    Research  Triangle Institute
                    P.O.  Box  12194
                    Research  Triangle Park, North Carolina  27709

                    L.S.  Hovis
                    Utilities and Industrial Processes Division
                    Industrial Environmental Research Laboratory
                    U.S.  Environmental Protection Agency
                    Research  Triangle Park, North Carolina  27711
                                  ABSTRACT
     An an alternative to using the dual-beam beta-particle backscatter gauge
to continuously measure dust cake thickness on operating bag filter or elec-
trostatic precipitator (ESP) units independent of gauge distance from the
dust, a one-point in-situ calibration method for using a simple beta-particle
backscatter gauge is developed and demonstrated.  The method consists of ob-
taining a response model for beta-particle backscatter in terms of  dust cake
density thickness (areal density), distance from the gauge, and either bag
filter or collector plate density thickness.  Then,  assuming that the density
thickness of the bag filter or the collector plate (on the ESP unit) without
dust is either known or (for the ESP) larger than the saturation value for the
beta-particle source being used, the initial gauge response before  dust is
present can be used to establish the gauge distance to the bag filter or
collector plate.

     This approach has been investigated in the laboratory with a system con-
sisting of a Sr-90 beta-particle source and an end-window Geiger-Mueller (GM)
tube detector.  The ranges of distances and simulated dust cake density thick-
nesses investigated were 3.4 to 7.4 cm and 0 to 0.2133 g/cm2, respectively.


                                     107

-------
The accuracies obtained were quite good.   For the bag filter application,  dust
cake measurement accuracies of 0.015 g/cm2 or better were obtained for all
distances up to 7.4 cm and dust cake density thicknesses less than 0.05 g/cm2.
This assumes counting rate errors of 1 percent and an error in the measurement
of the bag filter density thickness of 5  percent.  For the ESP application the
dust cake measurement accuracies for the  equivalent error assumptions were
even better, ranging from 0.0012 to 0.017 g/cm2 for all distances and dust
cake thicknesses investigated.  This excellent accuracy for the ESP applica-
tion is in part due to the basic sensitivity for this case (slope of the
calibration curve) and in part because there is no error introduced by varia-
tions in the ESP collector plate thickness.   The accuracies for both applica-
tions have been significantly improved by removing the rotating shield super-
structure, greatly reducing the background response reported for the predeces-
sor dual-beam gauge.

     This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.

                                INTRODUCTION
     As an alternative to using the dual-beam beta-particle backscatter gauge
to continuously measure dust cake thickness on an operating bag filter inde-
pendent of gauge distance from the bag (1), and for the dust cake thickness
on the collector plate of an electrostatic precipitator (ESP) unit, a one-
point in-situ calibration method for using a simple beta-particle backscatter
gauge is proposed and developed.

     The basic idea is to develop a response model for a simple beta-particle
backscatter gauge in terms of the dust cake and bag filter or collector plate
density thickness and distance from the gauge.  Then, assuming that the bag
or collector plate density thickness is known, the initial gauge response with
only air (no dust) entering the bag or the ESP unit can be used to establish
the gauge distance from the bag or collector plate.  Assuming that this
distance remains constant throughout operation allows one to convert gauge
responses during actual operation to sample density thickness.  The assump-
tion of fixed gauge-collector surface separation is very reasonable in the
ESP application and has been shown to be surprisingly good for the bag filter
as well (1).  One can obviously extend this approach to a scanning gauge if
the response as a function of vertical distance along the bag or collector
plate is taken prior to the introduction of dust.

                             SINGLE GAUGE MODELS
BAG FILTER AND DUST CAKE

     For a bag filter with accumulation of dust cake on  the inside, the model
for a single beta-particle backscatter gauge mounted outside the bag filter
(the growing dust cake is on the side of the collector opposite to that of
the gauge) can be taken as (2):


                                     108

-------
          R = Bf(d)[l-exp(-vy)] + Cf(d)exp(-v'y)[l-exp(-ux)]  + N          (1)

where R is the gauge response in counts per second (c/s);  B,  C, N,  u,  v,  and
v' are model parameters to be determined by a suitable least-squares analysis
of calibration data; f(d) is the effect of sample distance d  (in cm) on the
backscatter response; y is the bag density thickness in g/cm2; and  x is the
dust cake density thickness in g/cm2.   B and C are the saturation responses
in c/s for infinitely thick samples of the bag filter and  dust cake (by it-
self), respectively; N is the background counting rate in  c/s when  neither
bag filter nor dust cake is present; u and v are the backscatter coefficients
for the bag filter and dust cake in cm2/g, respectively; and  v' is  the atten-
uation coefficient of the backscattered beta particles through the  filter bag
in cm2/g.  This equation predicts a response that increases with increasing x.

ESP COLLECTOR PLATE AND DUST CAKE

     For an ESP collector plate with accumulation of dust  cake on the same
side as the gauge, the model for a single beta-particle backscatter gauge can
be taken as:

          R = Cf(d)[l-exp(-ux)] + Pf(d)exp(-u'x)[l-exp(-wz)]  + N          (2)

where the parameters are the same as in Equation 1 with four  additions:  P, w,
and u' are model parameters to be determined by a least-squares analysis  of
calibration data,  and z is the density thickness of the collector plate in
g/cm2.  P and w are the saturation response and backscatter coefficient for
the collector plate in c/s and cm2/g,  respectively; and u' is the attenuation
coefficient for the backscattered beta particles through the  dust cake in
cm2/g.  In this case, when an infinitely thick collector plate is of interest,
Equation 2 becomes:

          R = Cf(d)[l-exp(-ux)] + Pf(d)exp(-u'x) + N                      (3)

Both Equations 2 and 3 predict a response that decreases with increasing  x,
since P (the saturation response to the collector plate) is the dominant  term
where there is no dust cake and x is zero.

                            APPLICATION PROCEDURE
     The procedure proposed for using this single-beam beta-particle back-
scatter gauge follows.

     1.   Calibrate the gauge in the laboratory using a representative piece
          of the filter bag or the collector plate with a known range of
          weights of additional material that is compositionally representa-
          tive of the dust to be collected at various distances.

     2.   Take gauge responses with the gauge in each position of interest
          next to the operating bag filter or collector plate when dust-free
          air is being used, but operation is otherwise normal.  Install the
          gauge on a traverse mechanism if desired.

                                      109

-------
     3.   Determine the distance to the gauge from the bag filter or collector
          plate at each position by using the gauge response with the known
          bag filter or collector plate density thickness.

     4.   Take gauge responses when the filter bag or ESP units are in normal
          operation with dust-laden air, and determine the dust cake density
          thickness from the calibration curve with the previously determined
          distance and known response to bag filter or collector plate density
          thickness.

                   GEIGER-MUELLER TUBE DETECTOR DEAD TIME
     It was found that for the Geiger-Mueller (GM) tube detector and the
source used in the beta-particle backscatter gauge, a considerable counting
loss due to detector dead time was often encountered.  To correct for this,
the model for observed (R) and true counting rates (R1) in terms of the
detector dead time (r) for a paralyze detector (2) was used.  This model is:

                              R = R'exp(-rR')                             (4)

The two-source method (2) utilizing the exact model of Equation 4 was used to
obtain the detector dead time (r).  This method consisted of obtaining the
counting rates of the background, Source 1 plus background, Sources 1 and 2
plus background, and Source 2 plus background, in that order.  Then the
appropriate model of Equation 4 was used, and iterations on trial values of r
were made until all the experimental and calculated values of the counting
rates agreed to within a specified amount.

     For example, for the counting rates described of 1.10, 1094.00, 2046.00,
and 1309.00 c/s, respectively, the detector dead time is found to be 114.6
ysec.  The average dead time found for several determinations for the GM tube
used here was 115.0 ysec.

                                   RESULTS
BAG FILTER APPLICATION

     The dual-beam beta-particle backscatter gauge previously used (1) was
modified by removing the rotating shield and the superstructure necessary to
mount it.  An aluminum-encapsulated Sr-90 beta-particle source with nominal
intensity of 0.1 mCi (3.7*106 Bq) was used as the single source.  The source
was collimated by using a 0.3 cm diameter by 1.1 cm long hole in a lead
shield.  Experimental gauge responses were obtained with this device using a
sample of an actual bag filter and various paper thicknesses at various
distances to simulate bag filter and dust cake samples.  The results are
given in Table 1 and shown plotted in Figure 1.  These results indicate a
vast improvement in the signal-to-noise ratio for the single gauge as com-
pared to the dual-beam beta-particle backscatter gauge for the bag filter
application.  (Note the background counting rate for the infinite sample
                                      110

-------
distance.  This is primarily the result of removing the superstructure neces-
sary for mounting the rotating shield which provided a very large source of
beta-particle backscatter.)

     These results were obtained with the same (constant) bag filter density
thickness so that the appropriate modification of Equation 1 is:

                    R = B'f(d) + C'f(d)[l-exp(-ux)] + N                   (5)

The form of f(d) was found  to be:

                              f(d) = exp(-ad)                             (6)

Substituting Equation 6 into Equation 5 gives:

               R = B'exp(-ad) + C'exp(-ad)[l-exp(-ux)] + N                (7)

The values of the model parameters obtained by a least-squares analysis of
the data were:   B1 of 285.308 c/s, C1 of 155.714 c/s, a of 0.302 cm"1, u of
14.209 cm2/g, and N of 24.74 c/s.  The values of R calculated from Equation 7
with the quoted model parameters and the known values of the density thickness
x are given for comparison  purposes in Table 1.

     Equation 7, solved for the sample density thickness x, is:

               x = -(l/u)ln{l-[R-B'exp(-ad)-N]/[C'exp(-ad)]}              (8)

The values of x obtained from Equation 8 with the quoted model parameters and
the experimental values of  the gauge response R are also given in Table 1.
Note that there are cases when the term inside the logarithm becomes negative
and Equation 8 cannot be solved.  This may occur when ordinary statistical
counting rate fluctuations  cause a value of R that is too large and is
physically unrealistic.  In this case it is recommended that one use a
specified maximum value for x; 0.25 g/cm2 is recommended here.

ESP APPLICATION

     The gauge modified as  previously described with the aluminum encapsulated
Sr-90 source was used to obtain experimental results for various paper density
thicknesses at various distances on a 0.16 cm thick stainless steel plate to
simulate the dust cake and  collector plate samples in an ESP unit.  The
results are given in Table  2 and shown plotted in Figure 2.  These results
also indicate an excellent  signal-to-noise ratio for this application.

     To provide additional  information, gauge responses were taken for the
paper samples both with and without the stainless steel collector plate, and
both are given in Table 2.   These data provided a basis for directly deter-
mining separate geometry factors and backscattering coefficients for the paper
(simulating dust cake) and  stainless steel collector plate.  This revealed,
somewhat surprisingly, that the geometry factors were different for the dust
cake and stainless steel collector plate, but the scattering coefficients were
identical; just the opposite of what was expected.

                                     Ill

-------
     Due to these differences, Equation 3 is modified to provide an appropri-
ate calibration relationship:

          R = Cexp(-ad)[l-exp(-ux)] + Pexp(-a'd) exp(-ux) + N             (9)

The model parameter values for u, a, and N remain the same as for the bag
filter application:  14.209 cm2/g, 0.302 cm"1, and 24.74 c/s, respectively.
When these values are fixed and the value of a' is independently determined to
be 0.245 cm"1, a linear least-squares analysis of the data in Table 2 gives
values for the remaining parameters:  C = 258.390 c/s and P = 1257.714 c/s.

     To obtain measured values of the dust cake density thickness, Equation 9
is solved explicitly for x.  Note that if u were not equal to u' as in
Equation 3, then a transcendental relationship would be obtained in solving
for x.  The relationship for x becomes:

          x = -(l/u)ln{[Cexp(-ad)-R + N]/[Cexp(-a'd)-Pexp(-a'd)] }        (10)

     The values of x obtained from Equation 10 with the quoted model parame-
ters and the experimental values of the gauge response R are also given in
Table 2.  As in the bag filter application, values of R can occur that yield
negative values inside the logarithm so that Equation 10 cannot be solved.
In this case this may occur when ordinary statistical counting rate fluctua-
tions cause a value of R that is too small and is physically unrealistic.  In
this case it is recommended that one use a specified maximum value for x;
0.25 g/cm  is also recommended in this case.

                               ERROR ANALYSIS
     The propagation of error in terms of the standard deviation of the
measured dust cake density thickness s(x) is obtained in the usual way from
the product of the slope of the calibration curve dx/dR and the standard
deviation of the gauge response s(R):

                              s(x) = s(R)dx/dR                           (11)

When two or more errors from uncorrelated independent measurements are propa-
gated by addition or subtraction, the total standard deviation is:

                         s2(R) = m2(R) + n2(R) + 	                    (12)

where m(R), n(R), .... are the individual standard deviations.

     In this case the errors to be considered should include any error in the
measurement of the distance from the gauge to the bag filter or collector
plate and the usual error due to statistical counting rate fluctuations.  The
limiting error in the measurement of distance is also due to statistical
counting rate fluctuations.  In the bag filter application, any error caused
by not knowing the density thickness of the bag filter should also be con-
sidered:  this error is not pertinent for the ESP collector plate application
since the collector plate has a density thickness that well exceeds the


                                      112

-------
saturation thickness for steel for the Sr-90 beta particles and will yield the
same backscatter intensity for any thickness larger than about 0.08 cm.

     The standard deviation error in the distance measurement for the bag
filter application is:

                    s2(d) = (dd/dR)2s2(R) + (dd/dy)2s2(y)                (13)

where R in this case is given by Equation 1 with f(d) given by Equation 6 and
a value of x of zero (since the measurement of the distance d is made when no
dust cake is present).  The s(R) and s(y) are the standard deviation errors
in the measurement of R and the bag density thickness, respectively.  R
becomes:

                         R = Bexp(-ad)[l-exp(-vy)] + N                   (14)

The derivative terms of Equation 13 are obtained by differentiation of
Equation 14:

                         dd/dR = exp(ad)/{aB[l-exp(-vy)]}                (15)

                         dd/dy = (v/a)exp(-vy)/[l-exp(-vy)]              (16)

As a rule-of-thumb (2), the standard deviation error for the counting rate
can be taken as:

                                  s(R) = R/100                           (17)

This assumes that the error due to statistical counting rate fluctuations can
be held to less than 1 percent, and that other errors such as electronic
instabilities that are proportional to the counting rate are controlling.

     The error in the measurement of the bag filter density thickness is less
certain.  For the present it is assumed to be 5 percent of the true value:

                                  s(y) = y/20                            (18)

     The standard deviation error in the dust cake density thickness for the
bag filter application consists of contributions from another measurement of
R and the uncertainty in the distance measurement:

                    s2(x) = (dx/dR)2s2(R) + (dx/dd)2s2(d)                (19)

Here R must include the term containing the dust cake density thickness, so
it is given by Equation 7.  The derivative terms of Equation 19 are obtained
by differentiation of Equations 7 and 8, respectively:

                         dx/dR = exp(ad)exp(ux)/uC'                      (20)

               dx/dd = (a/u)exp(ad)(R-N)/[C'+B'-(R-N)exp(ad)]            (21)
                                     113

-------
 The value  of  s(d)  is  obtained  from Equation  13  using  Equations  14  through  18.
 The value  of  s(R)  is  again  taken  as Equation 17,  but  this time  with  the value
 of R  taken from Equation  7.

      The standard  deviation  errors s(d) and  s(x)  using  these relationships
 have  been  calculated  for  all the  distances and  simulated dust cake density
 thicknesses used in obtaining  the calibration results given in  Table  1.  The
 values of  s(d)  are independent of dust cake  density thickness and are found
 to be 0.132,  0.133, and 0.139  g/cm2  for the  distances 3.4, 4.9, and  7.4 cm,
 respectively.   The values of s(x)  are given  in  Table  3.

      The error  in  the distance measurement for  the collector plate in the ESP
 application is  due only to counting  rate fluctuations.  -If it is assumed as
 before that this error can be  held to within a  standard deviation of  1 per-
 cent  of the counting rate when no dust cake  is  present  and that the error in
 the actual  measurement is 1  percent  of the counting rate with dust, then the
 measurement error  is:

                S2(x) = (dx/dR')2s2(R') + (dx/dR)2s2(R)                   (22)

 where R1 and R  are the counting rates when no dust is present and when dust
 is present, respectively.  The derivative terms of this relationship are
 obtained by differentiation  of Equation 9 when x  is either zero or has some
 value:

                         dx/dR' =  l/[uPexp(-a'd)exp(-ux)]                (23)

                dx/dR = l/[uCexp(-ad)exp(-ux)-uP exp(-a'd)exp(-ux)]       (24)

 The values  of the  standard deviation are:

                              s(R') = R'/100                             (25)

                              s(R) = R/100                               (26)

     The standard  deviation  errors s(x) using these relationships have been
 calculated  for  all the distances and simulated dust cake density thicknesses
 used  in obtaining  the results given in Table 2.   The' values are given in
 Table 3.

                         DISCUSSION AND CONCLUSIONS
     The analysis of the previous section and the resulting predicted measure-
ment standard deviations in Table 3 indicate that the dust cake measurement
for the bag filter application is improved over that for the dual-beam gauge
previously reported (1).  This is primarily due to the large background
reduction which resulted from removal of the superstructure necessary for the
rotating shield.  While the advantage of having a measurement independent of
distance from the bag is lost, better accuracy is obtained with the present
method and one does not have the problem of the extensive maintenance neces-
sary for the dual-beam backscatter gauge.  For systems in which bag distance

                                     114

-------
must be continuously determined, the single-beam, dual-detector technique
described elsewhere in these proceedings by Gardner, Donovan,  and Hovis (3)
is preferable.

     Gauge-collector plate distances in an ESP unit do not vary,  so the
technique is better suited for this application.   And indeed,  the results
obtained for the ESP application are even more encouraging.  The  fact that
the 0.16 cm stainless steel collector plate represents a saturation thickness
for Sr-90 beta particles means that no error is introduced by normal varia-
tions in the thickness of this plate.  In addition the slope of the gauge
response with dust cake density thickness is excellent over an extended range.
The resulting measurement standard deviations are excellent for a wide range
of dust cake density thicknesses and collector plate distances from the
gauge.  The GM tube detector is quite rugged and  should easily withstand the
conditions within an operating ESP unit.

     Note that the beta-particle backscatter response to paper is not the same
as to that of dust cake, since the effective or average atomic number of the
dust is higher than that of paper.  This means that the actual sensitivity to
dust cake will be somewhat higher for the bag filter application and somewhat
lower than for the ESP application predicted here.

                                 REFERENCES
1.   Gardner, R.P., Donovan, R.P., and Hovis, L.S.  A dual-beam backscatter
     beta-particle gauge for measuring the dust cake thickness on operating
     bag filters independent of position.  In:  Third Symposium on the
     Transfer and Utilization of Particulate Control Technology:  Volume III.
     Particulate Control Devices, EPA-600/9-82-005c (NTIS PB 83-149609),
     July 1982, pages 128-139.

2.   Gardner, R.P. and Ely, Jr., R.L.  Radioisotope Measurement Applications
     in Engineering.  Reinhold Publishing Corporation, New York, 1967.

3.   Gardner, R.P., Donovan, R.P., and Hovis, L.S.  A dual-detector beta-
     particle backscatter gauge for measuring dust cake thickness on opera-
     ting bag filter and ESP units.  In these Proceedings.
                                      115

-------
  200
                                     d =  3.4  cm     _ 	Q
                                           	_)'
     0.


Figure 1,
  600
Figure 2.
      0.05         0.10          0.15         0.20
SIMULATED DUST CAKE DENSITY THICKNESS (x), g/cm2
Laboratory calibration results for the single beta-particle
backscatter gauge response to paper samples on a bag filter
           to simulate bag filter operation.
                          0.25
                  0.05
                    0.10
0.15
0.20
0.25
SIMULATED DUST CAKE DENSITY THICKNESS  (x), g/cm2
Laboratory calibration results for the single beta-particle
backscatter gauge response to paper samples on a 0.16-cm thick
stainless steel sheet to simulate ESP unit operation.

                       116

-------
   TABLE  1.  LABORATORY CALIBRATION RESULTS FOR THE SINGLE BETA-PARTICLE
            BACKSCATTER GAUGE RESPONSE USING A BAG FILTER AND
             PAPER SAMPLES TO SIMULATE BAG FILTER OPERATION

Experimental
Dust Cake (Paper)
Sample Distance Density Thickness
(d), cm (x), g/cm2
infinite
3.4
3.4
3.4
3.4
3.4
4.9
4.9
4.9
4.9
4.9
7.4
7.4
7.4
7.4
7.4
0.0
0.0
0.0499
0.1062
0.1618
0.2133
0.0
0.0499
0.1062
0.1618
0.2133
0.0
0.0499
0.1062
0.1618
0.2133
Standard Deviation
Gauge
Response
(R), c/s
25.43
129.65
156.09
170.83
176.28
180.81
88.84
106.82
116.86
119.92
122.13
54.38
63.62
69.33
70.64
72.27
Calculated
Gauge Dust Cake (Paper)
Response Density Thickness
(R), c/s (x), g/cm2
25.42
126.92
155.24
170.36
177.09
180.00
89.70
107.70
117.31
121.59
123.44
55.27
63.73
68.25
70.26
71.13
0.00261
0.00353
0.05210
0.10895
0.15226
0.23859
-0.00168
0.04643
0.10227
0.13468
0.17333
-0.00366
0.04895
0.13068
0.17994
0.25000b
0.0196

Corrected for dead-time losses  by using Equation 4 with r = 115 ysec.

The maximum value allowed and used when the term inside the logarithm  in
Equation 8 is negative, as at present.
                                    117

-------
    TABLE 2.  LABORATORY CALIBRATION RESULTS FOR THE SINGLE BETA-PARTICLE
              BACKSCATTER GAUGE RESPONSE TO PAPER SAMPLES ON A
               0.16-CM THICK STAINLESS STEEL SHEET TO SIMULATE
                             ESP UNIT OPERATION
                          Experimental
                                        Calculated
Sample Distance
    (d), cm
Dust Cake (Paper)    Gauge       Gauge    Dust Cake (Paper)
Density Thickness  Response    Response   Density Thickness
   (x), g/cm2      (R), c/s    (R), c/s      (x), g/cm2
3
3
3
3
3
4
4
4
4
4
7
7
7
7
7

.4
.4
.4
.4
.4
.9
.9
.9
.9
.9
.4
.4
.4
.4
.4

0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
Standard
.0
.0499
.1062
.1618
.2133
.0
.0499
.1062
.1618
.2133
.0
.0499
.1062
.1618
.2133
Deviation
570.
326.
211.
158.
137.
409.
241.
157.
118.
98.
226.
143.
95.
71.
62.

19(
60(
25.
82.
97(106.
98(1
18.
19(123.
84 (
52(
04 (
45(
20(
17(
80(
77(
56(
08 (

25.
59.
75.
84.
87.
24.
39.
50.
51.
54.

08)
78)
3)
2)
3)
48)
93)
81)
17)
59)
45)
83)
93)
92)
34)

571.
340.
217.
162.
139.
403.
240.
154.
115.
99.
229.
139.
91.
70.
60.

52
82
73
87
21
37
95
29
66
01
95
77
66
21
96

0
0
0
0
0
-0
0
0
0
0
0
0
0
0
0
0
.00021
.05453
.11036
.16808
.22011
.00141
.04964
.10351
.15594
.21709
.00152
.04673
.09919
.15666
.20468
.00496
3Corrected for dead-time losses by using Equation 4 with r = 115 psec.  The
 values in parentheses are for dust cake (paper) without the stainless steel
 collector plate.
                                     118

-------
TABLE 3.   PREDICTED MEASUREMENT STANDARD DEVIATIONS  FOR THE  BAG  FILTER
                         AND ESP APPLICATIONS

Sample Distance
(d), cm
3.4
3.4
3.4
3.4
3.4
4.9
4.9
4.9
4.9
4.9
7.4
7.4
7.4
7.4
7.4
AVI
Dust Cake
Density Thickness
(x), g/cm2
0.0
0.0499
0.1062
0.1618
0.2133
0.00
0.0499
0.1062
0.1618
0.2133
0.00
0.0499
0.1062
0.1618
0.2133
erases
Bag Filter
Standard Deviation
[s(x)], g/cm2
0.0054
0.0139
0.0345
0.0795
0.1683
0.0055
0.0142
0.0351
0.0808
0.1712
0.0059
0.0151
0.0372
0.0856
0.1812
0.0622
ESP
Standard Deviation
[s(x)], g/cm2
0.00115
0.00184
0.00366
0.00775
0.01588
0.00116
0.00187
0.00372
0.00789
0.01617
0.00120
0.00196
0.00393
0.00833
0.01708
0.00624
                                  119

-------
       PULSE-JET  FABRIC  FILTER  EXPERIENCE USING  NON-GLASS MEDIAS AT  AIR

          TO CLOTH  RATIOS OF  5  TO  1  ON  A PULVERIZED  COAL FIRED  BOILER
                         By:  G.  L.  Pearson
                             D.  D.  Capps
                             Adolph Coors Company
                             Golden,  Colorado   80401
                                   ABSTRACT
     A 12-module baghouse system  using  the Carter-Day high temperature  module
has been in service since November  1979  controlling emissions  from the 450,000
Ib/hr Boiler No.  5 unit.   Bags  were made  of felted  "Ryton"  and  cleaning  is
accomplished via a  low pressure-moderate  volume,  pulse-jet  technique.

     Experience  to  date with this baghouse system will be presented.  Data  on
pressure drop,  cleaning cycles,  cleaning  frequency,  bag  life experience and
particulate  emissions  will   be  reviewed.   Results  of on-line  evaluation  of
felted PBI  bags  will also be presented.
              Prepared for Presentation  at the  Fourth  Symposium
              on   the  Transfer  and  Utilization  of  Particulate
              Control  Technology  Sponsored  by the  U.S.  Environ-
              mental Protection  Agency  and  held in

                             Houston, Texas
                           October 11-15, 1982
                                      121

-------
INTRODUCTION

     Energy availability considerations and cost projections motivated Coors to
proceed with  their second coal fired  boiler  (Unit No.  5)  in  late  1976.   This
unit was scheduled for start-up and went into service in November 1979.   In the
installation  of  this coal fired  boiler,  it was important  to  the  Company that
not only permit requirements were met,  but that emissions were minimized to the
lowest  levels possible  with  current  technology.   However,   both   space  and
capital dollars were  at  a premium to accomplish these objectives.   After study
and careful consideration, a  pulse-jet type modular system, offered by Carter-
Day  of Minneapolis,  was  selected  to  control  particulate   emissions.   This
modular system was  ordered  in April 1978 and went into service with the boiler
in November 1979.   It has operated successfully for the last 34 months.


BOILER DESCRIPTION

     The boiler  is a  Combustion  Engineering pulverized  coal  VU-40  type unit
rated at 450,000 Ib/hr of steam at 825 PSIG and 850°F.  Coal is pulverized with
three Raymond  RB-613  bowl  mills.   A large fin tube boiler feedwater economizer
is  utilized  to cool the flue  gas  to within the temperature range  of 300°F to
360°F depending  upon boiler  load.   The boiler was  designed  for,  and operates
with, a 3 to 1 turn-down capability.


BAGHOUSE MODULE DESCRIPTION

     The basic configuration  of  the Carter-Day Model 376 RF10 high temperature
unit is illustrated in Figure 1.   Each circular unit has 376 oval pattern bags,
10  feet long  for  a filter area of 4800 square feet per module.  (The 12-module
system  has  4512  bags and a filter  area of 57,600 square  feet.)   Bags  are in-
stalled over  and supported by carbon steel 9-gauge  wire  frame cages.   Flyash
cake is accumulated on the outside of the bags and the filtered flue gas  leaves
at  the  top  of the units and goes to the induced draft fan and stack.  Bags are
installed or  removed through  the top clam shell doors at the top of the units.

     Bags  are made  of  22-ounce  per square yard "Daytex"  felted fabric.   This
fabric  is  essentially felted  and needled "Ryton" fibers  on  a "Cortex" scrim.
"Ryton" is  the trademark name which Phillips Petroleum has given to Polypheny-
lene Sulfide.  This material has both excellent acid  resistance and temperature
resistant characteristics up to 370°F.

     Cleaning  of  the filter cake from  the  outside of the  bags is accomplished
by  opening of the diaphragm  valve  at  the top of  the unit and discharging the
7.5  PSIG  reservoir tank into  the inside of the bags  directly  under the nozzles
on  the  distribution manifold.   Each reservoir tank  is charged to 7.5 PSIG by a
Miehle-Dexter  Model  Number 46-4006  positive displacement  blower.   Since each
module  has its own  blower,  tank, valve, etc., the  reliability of the overall
system  is  quite  good.  Tripping of  the  diaphragm  valve is signaled by a pres-
sure switch mounted on the reservoir tank.  The distribution manifold is turned
at  one rotation per  minute during  cleaning via a small  1/2  HP motor and gear
box  to the side of  the  diaphragm valve.   When in the  cleaning mode, the dia-
phragm  valve  pulses  approximately every 3.7 seconds.

                                       122

-------
        Carter-Day RP376 FILTER
  FOR HIGH-TEMPERATURE FLY ASH
 COLLECTION IN INDUSTRIAL-UTILITY
       COAL-FIRED BOILERS
      DIAPHRAGM
        VALVE
AIR RESERVOIR TANK
FOR REVERSE PULSE
AIR CLEANING SUPPLY
 FILTER TUBE SHEET
 FOR HOLDING FILTER
TUBE CAGE ASSEMBLY
       Carter-Day
 DAYTEX FILTER TUBES
    AIR INLET
   COLLECTED
   PARTICULATE
     OUTLET
                 DRIVE MOTOR
              FOR REVERSE PULSE
             AIR CLEANING SYSTEM
                       FACTORY INSTALLED
                     INSULATED TOP ACCESS
                  Jp"  DOORS FOR FILTER
                    TUBE INSPECTION/SERVICE
                       HIGH TEMPERATURE
                        ACID-RESISTANT
                           GASKETING
                             CLEAN
                           AIR OUTLET
                          CLEANING AIR
                           MANIFOLD
                                                                NOTE:
                                                        FILTER BODY AND HOPPER
                                                           INSULATED IN FIELD
                                                             HOPPER ACCESS
                                                             HOLE NOT SHOWN
                       60° CONICAL HOPPER
         FIGURE 1  MODULE
CONFIGURATION
                                      123

-------
BAGHOUSE INSTALLATION AND AUXILIARIES

     To  insure  a  good,  economical  and  easy to  operate  and maintain  system,
considerable thought and  engineering went into the installation.   Figure 2 and
3  show the  basic module  installation  configuration.   The  inlet and  outlet
plenums were designed  to  balance the distribution to each  module  and keep the
ductwork  free  of ash  drifts.   In order to minimize  troublesome and  expensive
expansion joints, all modules and some flue ducts are supported  by  hanging rods
which allow  adequate movement  due to thermal expansion.  The entire  12-module
system  has  only 10  expansion  joints.   Each module has a minimum  leakage type
outlet damper,  which can  be operated remotely from  the control  room  to either
take a module off-line or to accomplish off-line cleaning.   When this  damper is
closed, a small  valve  at the top  of the unit opens, automatically letting air
in to  purge  that particular module and prevent acid corrosion.   Each  module is
equipped with a manually operated inlet damper for use when  that module is in a
maintenance mode.  There  are two 60-inch diameter bypass  poppet type  valves in
the system which  are used during start-up and oil firing  of the boiler.   These
bypass valves are operated from the control room and can open against  40 inches
W.G.  pressure.

     The conical ash hoppers are insulated with 10 inches of MT-8  mineral wool
insulation and electric blanket heaters are used on the bottom 3'-6" portion of
the cone  to  insure  that the flyash  will  stay dry and convey easily.   The main
portion of  the  module  and all  flue  ducts  are insulated with 4  inches  of MT-8
mineral wool  insulation.   The  modules  and flue  ducts  were metal  lagged with
"Reynolds Rainlock  Rib"  diamond embossed  aluminum  0.0165  inches   thick.   The
conical  hoppers,  and  other difficult  areas, were  lagged  with  flat  diamond
embossed aluminum 0.025  inches  thick.   High ash  level  alarms are  installed in
each hopper.

     An  enclosure was  built for  the  facility  so  that  maintenance would be
easier  during  bad weather  conditions  and  to  satisfy  aesthetic considerations
since  it  is located near  the  main entrance-visitors area  of the  plant.   This
enclosure is illustrated in Figures 2 and 3.

     An opacity meter (Lear Siegler RM-41) is installed in the 13-foot diameter
stack downstream  of  the  baghouse and induced draft fan.  Readout of opacity in
the  control  room  is  located  so  as to  be  easily  visible from  the baghouse
operator panel.

     The operator panel  for the baghouse, located in the control room, has the
following features:

      1.  Pressure drop across  the baghouse (inlet plenum to outlet plenum) is
          displayed on a meter.  The operator can select the high pressure that
          he wants  cleaning to  start at  and  the low  pressure  where he wants
          cleaning to cease on this meter.

      2.  Temperature of  the flue gas going to  the  baghouse is displayed on a
          meter on the control panel.

      3.  Conditions of either  high baghouse pressure drop or high temperature
          are alarmed via red lights and buzzers.

                                      124

-------
tsj
       FIGURE 2   PLAN VIEW OF ITOLE
                  ILLUSTRATING INLET PLENUM

-------
FIGURE 3
SECTIONAL ELEVATION OF MODULE ARRANGEMENT ILLUSTRATING BY-PASS
VALVES, INLET PLENUM, OUTLET PLENUM, fDDULE SUPPORT RODS,
AND CONFIGURATION OF BUILDING ENCLOSURE,
                                                                                        OUTLET
                                                                                        PLENUM

                                                                                        INLET
                                                                                        PLENUM

-------
      4.   The  module  outlet  dampers can  be  either closed or opened  from this
          panel  to  isolate  a module  or accomplish  off-line  cleaning.   Damper
          position is indicated  by status  lights.

      5.   Bypass  valves can  be opened or closed from this panel.   Status lights
          indicate their location at all times.

      6.   The  status  of which modules  are in  the clean mode is  indicated on
          this  panel  to the  operator  via blower  motor  running  status  lights.


BAG CLEANING CYCLE LOGIC

     The  system has  been  operated since initial start-up with a pressure drop
(inlet plenum  to  outlet plenum)  initiated cleaning cycle.   Typically, cleaning
is initiated when  the  pressure  across the baghouse builds up to  5 in. W.G.  and
continues through  the  cycle  until  the pressure drop  is  reduced  to typically 3
in. W.G.    The  entire baghouse is  then  in the passive no-clean  mode  until  the
pressure  builds  back up  to  the higher  set  point  of 5 in.  W.G.  When in  the
clean mode, four  modules  at  a time are cleaned for a duration of five minutes
(the other eight  modules  are in the passive  no-clean mode).   The group of four
modules cleaned  is sequentially  stepped  by  the software of  the PLC  (program-
mable logic computer  unit from  Allen Bradley-Series  1774) so  that all  modules
experience an  equal  amount  of  cleaning.   All  automatic cleaning to date  has
been on-line type cleaning.   Some manually initiated off-line cleaning has been
utilized  infrequently  during  periods  when above average amounts  of  flyash  are
being filtered, such  as M.C.R.  operation of  the boiler with simultaneous soot-
blowing.

     This  method  of  cleaning  has  been  very  successful  in maintaining  the
pressure  drop  within the desired 3  in.  to  5  in.  range and at the  same time
enhancing  bag  life,  blower  life,  and  saving  energy.   Any  one  bag  is pulsed
relatively infrequently.

     Tests  have  been performed  on this  unit  which  demonstrate  that off-line
cleaning  is both  more effective and will  clean the bags at system air to cloth
ratios of  roughly between five  and six.  Optimum off-line clean times for each
module appear to  be  in the range  of  two  to  three minutes.   However,  there are
no immediate plans to automate this unit to off-line cleaning.


EMISSION TESTS

     Emission   tests  were performed at  the stack of this unit on  April 17 and
April  18,  1980  as part of the Compliance Tests required on the permit for this
coal  fired  boiler.   The results of the EPA  Method No.   5 particulate tests are
shown  in  Figure  4.   The average of the three tests via  F Factor method is 0.023
pounds per million  BTU particulate (roughly 1/4  of the permit  requirements).
Since  this  ratio  is  energy rate dependent, data is also presented with respect
to the energy base of steam rate out and coal consumed.

     The  coal  burned during  the compliance tests  had a  heating  value of  10,700
BTU/lb,  a moisture  content  of  13.64%, an  ash content  of 5.89%,  and a  sulfur
content  of 0.53%,  all  on an as-received basis.   This  coal comes  from Routt
County,  Colorado.
                                      127

-------
   COMPLIANCE TESTS ON BOILER NO. 5
        ADOLPH COORS COMPANY
      (DATA PER EPA METHOD NO. 5)

PARTICULATE EMISSION RATE
(BASED ON F FACTOR)
PARTICULATE EMISSION RATE
(BASED ON STEAM RATE)
PARTICULATE EMISSION RATE
(BASED ON COAL RATE)

BOILER STEAM OUTPUT
ACTUAL VOLUME OF
FLUE GAS THRU FILTER
AIR TO CLOTH RATIO
TEST#1
4-17-80
1200-1355
0.019 LB
1Q6BTU
0.01 8 LB
1Q6BTU
0.01 9 LB
1Q6BTU
405,000 LB
HR
276,000
ACFM
4.79
TEST #2
4-18-80
0902-1047
0.031 LB
1Q6BTU
0.029 LB
1Q6BTU
0.021 LB
106BTU
420,000 LB
HR
293,000
ACFM
5.08
TEST #3
4-18-80
1240-1424
0.018 LB
1Q6BTU
0.01 6 LB
106BTU
0.01 6 LB
1Q6BTU
420,000 LB
HR
288,000
ACFM
5.00
.
AVG. OF
3 TESTS
.023 LB
1Q6BTU
.021 LB
1Q6BTU
.019LB
106BTU




FIGURED  EMISSION TEST RESULTS
                128

-------
     Emission data beyond  the  compliance test has not been obtained.  However,
the plume has been  essentially non-visible to a human observer during the last
34 months of operation.

     Formal  certification  of the  on-line  stack monitors  for this  boiler  was
declared August  1,  1982.


"PBI" BAG TESTING

     To evaluate the  very  promising potential of  "PBI"  (polybenzimidazole),  a
product of Celanese Corporation,  376 felted (22 to 26 ounce) bags made of this
material were  installed  in Module  5~S (see  Figure 2)  on  December 18,  1981.
These bags  went  into  service and  performed  very well.  As  of July 1982, they
were holding  up  very well  and showed  no  fabric deterioration as  a result of
either temperature  or  S02 in the boiler flue gas.  Experience to date indicates
that "PBI"  is  at least  as  good  a filter media  as  "RYTON,"  although elaborate
testing to date  has not been accomplished.


SUMMARY OF OVERALL  EXPERIENCE

     The initial  start-up  of this system went quickly,  easily  and without any
problems.   Subsystems  (blowers, plenum turning mechanisms, pulse valves, outlet
dampers  and  all  controls)  were  checked  and operated  several  days  prior to
start-up.   After the  boiler was  on-line with  coal  and operating  properly,  the
baghouse was then put  into service gradually over a period of approximately one
hour.   This  system has  been very easy  to put  into and  take out of service.

     During the 34 months  of operation, this system has operated very well  and
up to expectations.   The unit has required minimal maintenance and very little
operator attention  --  almost to the point where the operators take the baghouse
for granted.   The baghouse pressure drop has been maintained between 3 in. W.G.
and  6   in.  W.G.  throughout  the  load  range  of the  boiler  (150,000  Ib/hr to
450,000  Ib/hr)  even  though a  wide  range of  coals  have been  burned  in  the
boiler.   Colorado-Keenesburg coal  with  a heating value as low as 7,000 BTU/lb,
a  moisture  content of 30% and an  ash  content  of 12%,  has been successfully
handled in this  system.

     Through the  first   32  months  of operation, all  felted  "Ryton" bags main-
tained  their integrity and there were no  bag  failures.   Spot checking of bags
at random  in December 1981 and June 1982  indicated that  considerable bag life
remained.  However, between August  1  and August  4,  1982,  a very unfortunate
incident occurred.   The  boiler had been shut down during  the last part of July
1982 for annual  inspection,  maintenance of boiler components, and modification
of  the  flame  safety logic  for  the   boiler.   Light-off was  attempted with
Colorado-Keenesburg coal  having  a  heating  value  of  roughly 7000  BTU/lb  and
moisture  content  of  30%  (a difficult  coal   to  ignite).    During  attempts at
light-off, the baghouse  was on stream.   Numerous attempts were made to develop
a  coal  fire in  the  boiler  without  success during  this  period  by the boiler
operators.    Apparently   fine  unburned  pulverized  coal  was  deposited  in  the
baghouse during  these attempts  at light-off  and  then  the  inevitable finally
happened -- sparks generated in  the furnace ignited the coal on the surface of
some of the  bags  in the baghouse.   The intense heat from  these localized fires
on the  surface of the  bags destroyed the fabric in certain areas.

                                       129

-------
     So far,  it  appears  that a relatively small percentage of the 4512 bags in
the  system  were  damaged.    As  of  September  24,  1982  (seven weeks after  the
incident),  256  bags  had been  found to  have burned holes  in them (246  were
"Ryton" and  10  were "PBI"); 226  of these failed bags were  identified  and  re-
placed  within a week  after the  incident.  Most burned  bags have  occurred  in
modules 1-S  and 2-S (see Figure  2).   Also,  all burned  holes  have  occurred in
the lower three  feet  of  the bags.   It  appears  that most of the remaining 4256
bags (94%) were not significantly damaged during the incident.

     It is very unlikely that a bag made of glass or "Teflon" would have faired
any better  than the  bags  made of  "Ryton" and  "PBI"  because of  the  apparent
intense heat.  The  lesson  learned by this incident is ttiat the boiler-baghouse
system must  be  operated  so  that a  fire  cannot  occur in the baghouse no matter
what the  bags are  made of or how the baghouse is constructed.  It is important
that a baghouse be operated according to the following guidelines:

  •  Always  bypass  and isolate  the baghouse during  coal  light-off conditions
     and other unstable firing conditions.

  •  Observe  in the strictest sense the temperature limits of the fabric in the
     baghouse.

  •  Maintain 02  levels  at  the  lowest  practical  levels consistent with  good
     combustion during all  operating modes.
                                      130

-------
CONCLUSIONS

     Based on the operating  experience  with this system and  the  testing which
has been done,  the following  can be concluded:

  •  Felted "Ryton"  is  a practical  and viable  filter  media for  boiler stack
     gas flyash filtering applications.

  •  Felted "FBI" is  a  practical  and viable filter media  for boiler stack gas
     flyash filtering applications.

  •  The system discussed in  this  paper is operated  successfully  with  on-line
     cleaning up to  an air to cloth ratio of between 4.5 and 5 to  1.

  •  Operation at air to cloth ratios between 5 and 6 is practical  with  "Ryton"
     and "PBI"  felted media  in this  PC fired boiler using the right off-line
     cleaning sequence.   Operation  at air  to cloth ratios  slightly higher than
     6 has been demonstrated  with  the Boiler No.  3  spreader  stoker fired unit
     which  is  another  boiler-baghouse   installation  at the  Golden,  Colorado,
     facility.

  •  Proper  design  and  selection  of baghouse  inlet dampers, outlet dampers,
     bypass  dampers,   and   baghouse  plenums   is  critical  to   good  overall
     operation.

  •  Any boiler-baghouse  system must be operated with  such  care  that  a fire
     condition  is  not  allowed to  occur  in  the  baghouse.   This  requirement
     applies to glass,  "Teflon"  and "Nomex" bags, as well  as  "Ryton" and "PBI"
     bags.
                                      131

-------
                 START-UP AND OPERATION OF A FABRIC FILTER
                 CONTROLLING PARTICULATE EMISSIONS FROM A
                    250 MW PULVERIZED COAL-FIRED BOILER
               by:  Charles B. Barranger     Norm Spence
                    Flakt, Inc.              Stone and Webster
                    Knoxville, TN  37923     Denver, CO

                    John Saibini
                    Sierra Pacific Power Co.
                    Reno, NE

                                 ABSTRACT

     Summarize and describe the design, construction, start-up and operation
of the fabric filter baghouse controlling the particulate emissions from
Boiler Unit No. 1 located at Sierra Pacific Power, North Valmy, Nevada.   The
basic design specification requirements will be described, i.e., fuel type,
flue gas, fabric filter type, system arrangement and performance require-
ments.  Actual pre-start checkout and system start-up procedure will be
reviewed in detail.  The fabric filter was put on-line in November of 1981.
Resulting equipment operating conditions from start-up through date of the
paper will be presented.

     The work described in this paper was not funded by the U.S. Environ-
mental Protection Agency and therefore the contents do not necessarily
reflect the views of the agency and no official endorsement should be
inferred.
                                      132

-------
                          DESIGN AND PERFORMANCE
INTRODUCTION

     The North Valmy Electric Generating Station, jointly owned by Sierra
Pacific Power Company and Idaho Power Company is located approximately 200
miles east of Reno,  Nevada,  just north of Interstate 80; Sierra Pacific
manages operation of the station.   The station consists of two 250 MW (nom-
inally rated) coal-fired units.  Unit No. 1 began commercial operation in
the fall of 1981 and Unit No. 2 is under construction and scheduled for
commercial operation in June, 1985.   Both Sierra Pacific and Idaho Power
share equally in the plant's cost and energy output, which provides electric
power for customers  in northern Nevada, northeastern California, and Idaho.
Stone and Webster Engineering Corporation is the architect engineer and
construction manager for the project.

UNIT DESIGN

     The Unit No. 1  boiler is a Babcock and Wilcox balance draft furnace,
natural circulation, drum type which produces 1,924,000 Ibs/hr steam at 2600
psig/1000°F, with a  single reheat exiting at 600 psig/1000°F.  Separate
primary and secondary air preheaters are provided of the Rothemuhle regen-
erative design.

     Coal is the base fuel and is delivered to the Valmy site by unit train.
No. 2 fuel oil is used for ignition.  At the time of design/engineering for
Unit No. 1, a single source  of coal had not been determined.  As a result,
the unit was designed to accomodate full load operation firing any of nine
alternate Western coals, including coals from mines in Utah, Colorado,
Wyoming, New Mexico, and Montana.

     These alternate coals yielded a significant impact on the design of the
particulate collection equipment.   Design of the equipment had to compensate
for a 230% variation in grain loadings, 15% variation in flue gas volumes
(velocities), and significant differences in flue gas composition and
flyash constituent analyses.

     Data pertaining to the  nine alternate design coals is attached as
Table 1.  Unit No. 1 is presently firing the C-l coal, which originates
from Salina, Utah.

EVALUATION OF ALTERNATE PARTICULATE COLLECTION EQUIPMENT

     In 1975, Sierra Pacific requested Stone and Webster to perform a study
evaluating alternate particulate collection systems for Unit No. 1.  The
evaluation included  a technical and economic assessment of alternate control
systems including hot and cold-side electrostatic precipitators, and fabric
filters.  At the time, hot-side electrostatic precipitators appeared to be
the predominate technology being applied for particulate control on new low
sulfur western coal-fired units.  However, the North Valmy design require-
ments of burning nine alternate coals presented problems to the precipitator


                                    133

-------
 suppliers.   Significant  variations  in  flue  gas  composition, coal ash
 chemistry and  resistivity  led  to  conservative precipitator designs which
 proved to be physically  large  with  high initial capital and high annual
 operating costs.

      As compared  to  the  precipitator,  the baghouse alternate presented ad-
 vantages of  independence from  fuel  and ash  characteristics, and flue gas
 temperature; less dependence on gas distribution and velocity; and compart-
 mentalized design allowed  on-line maintenance for high availability to the
 boiler.  In  the overall  evaluation,  the baghouse was determined the most
 cost effective means of  particulate control for North Valmy Unit No. 1.

 BAGHOUSE DESIGN

 General

      In April,  1976  Stone  and  Webster  issued design specifications for com-
 petitive bidding  of  the  Unit No.  1  baghouse.  In October, 1976 the Pollution
 Control Division  of  the  Carborundum Company (now Flakt, Inc.) was awarded
 the contract to supply and erect  the baghouse.   Stone and Webster maintained
 responsibility for design  of the  baghouse support steel, interconnecting
 field wiring,  and field  insulation.

 Baghouse Type

      The baghouse is an  all welded,  gas tight,  shop fabricated panel con-
 structed unit,  designed  to operate  under a  pressure of + 26" water gauge.
 An automatic reverse air system provides the necessary energy to clean the
 fiberglass filter bags.  The baghouse  is designed to meet a gross filter
 speed (air-to-cloth  ratio) of  1.99  ft/min when  firing the worst case coal
 and 2.42:1 ft/min (net)  when firing the performance coal (See Table 2).
-Figure 1 shows the baghouse location in the flue gas train.  It is located
 between the  air heater outlets and  the two  induced draft fans.

 Baghouse Arrangement (See  Figure  2)

      Arrangement  of  the  baghouse  is ten compartments total, five on either
 side of a common  central plenum.  Each compartment contains 648 filter bags
 making a total complex of  6,480 filter bags. Each filter bag is 12" in
 diameter by 32*-5" long.  Filter  bags  are arranged such that the maintenance
 personnel need to reach  by only two filter  bags to reach the third  (three
 bag reach).  This three  bag reach arrangement required  the design of three
 lower and upper walkway  systems inside each compartment.  There are 36 rows
 of bags in the depth of  each compartment and 18 bags in each row.  Dimen-
 sionally large compartments resulted,  i.e., 27'-9" wide by 45'-3" deep by
 38'-0" tall compartments.   It  appears this  size compartment is  the  largest
 baghouse compartment installed in the Utility  Industry  in the United States.

      To compensate for  thermal expansion in the baghouse, stub columns and
 low friction slide plates  were designed to  separate  the non-expanding
 structural support structure from the baghouse.  A centrally located point
 was fixed and  the baghouse expands  in all  directions from this point at  the

                                      134

-------
slide plate level.   This design considered air preheater failure and a
resulting upset excursion gas temperature of 720°F.

     Each baghouse  compartment is equipped with a series of damper valves.
All valves are vertical shaft poppet dampers and are operated by pneumatic
air cylinders.  Located at the clean gas side of each compartment there are
three gas outlet valves, two reverse gas valves, and one compartment vent
valve.  At the gas  inlet of each compartment there are two gas inlet valves.

     The reverse gas,  outlet gas, and inlet gas plenums are centrally
located, run the length of the baghouse, and are built integrally within
the baghouse.   Two  separate compartment vent manifolds run the length of
the baghouse at the roof level.  A 1/12 scale model study of all the bag-
house compartments  was completed and the results were incorporated into the
compartment baffling system.

     To facilitate  the maintenance of the equipment, access walkways and
platforms are located at the hopper discharge level, the inlet damper
level, top and bottom section of the fiTter bags, and the baghouse roof.
Access to these maintenance levels is via stairs.

Hoppers

     Ten double pyramid hoppers, one per compartment, provide a minimum of
12 hour flyash storage at the maximum operating condition (16.1 tons per
hour of flyash).  Each hopper has been designed with two separate discharge
points, so that the ten compartment baghouse requires 20 ash collection
points.  These hoppers have been extended to the center of the gas inlet
plenum to promote self-cleaning.  Should particulate fall out in the inlet
plenum, it eventually is carried through the inlet damper opening and is
deposited into the bottom of the ash hoppers.  All hoppers are identical
and have been designed with 55° valley angles.

Model Studies

     To evaluate the aerodynamic design of the particulate control system a
model study was performed at NELS, Inc.  A plexiglass 1/12 scale model of
all the ductwork and the baghouse was constructed.  Parameters such as
velocity and gas flow distribution, pressure losses, and particulate distri-
butions were measured.  Baffling, vaning, and design configurations were
optimized and incorporated into the actual particulate control system
designs.  To further optimize the baghouse hopper design, a separate 1/6
scale model of a hopper, inlet damper system, and inlet plenum was built
and tested.

Bag Filter Media and Construction

      The filter bag material, Filter Media Division of Carborundum, Style No.
601B,  is fiberglass coated with a Teflon "B" finish  (see Table 2 for speci-
fications).   Each  filter bag is constructed with  seven unequally spaced
anti-collapse  rings, top cuff with retainer, and  bottom cuff with cord.
The bottom of  the  bag  slips over an extended thimble  in the  tube sheet  and

                                     135

-------
is sealed with a quick release clamp.  The top of the bag is fitted over a
steel cap and held in place by the retainer band and lip on the cap.  Bags
are suspended from structural frames via a chain and spring, and held in
place by clips inserted around the chain links.

Baghouse Controls and Instrumentation

     The baghouse control panel was placed in the control house located
adjacent to the baghouse.  This panel includes a graphic display, annun-
ciator, and instrumentation.  Along with the graphic display, indication
lights show open/closed positions of all compartment valves, on/off display
of reverse air or vent fans, on/off display of hopper heaters,  and indicates
if high dust level exists in the hopper.  Alarm and annunciation of reverse
air fans, low and high gas temperature, low and high baghouse pressure
drop, hopper heater failure, and high hopper dust level is installed.
Pushbuttons are included to silence, test, and rest the annunciator.  Also
manual activation of the cleaning cycle and by-pass system is possible.
Baghouse inlet and outlet gas temperatures and baghouse differential pres-
sures are both monitored and continuously recorded.

     The baghouse cleaning system is automatically activated by a signal
from the differential pressure sensor.  A stepping switch timer and dial
adjustable timers have been included so that all cleaning cycle times can
be varied and optimized to site specific conditions.

Supplementary Systems

     There are three major baghouse supplementary systems.  They are compart-
ment ventilation, key interlock, and flue gas by-pass.   All three are basic
protective systems to protect either plant personnel or baghouse equipment.
The compartment ventilation system will cool down the baghouse compartment
as maintenance personnel work inside that compartment.   Compartment key
interlocks prevent access into the baghouse compartments before the damper
valves isolate that compartment from the flue gas.  Hopper key interlocks
insure deactivation of the level detectors before personnel enter the
hoppers.  Flue gas by-pass system automatically protects the filter bags
from high gas temperatures or high pressure transients.

BAGHOUSE OPERATION

Normal Filtering

     Flyash laden gas from the boiler is drawn into the inlet manifold of the
baghouse.  The inlet manifold spans the length of the baghouse, and ducts the
gases through inlet poppet valves into all the hoppers of the baghouse.
These gases are directed upward into the filter bags of each compartment and
particulate is deposited on the interior surface of the filter bags.  Gases
are filtered as they pass through the dust cake/bag material and are drawn
upward through poppet valves into the common outlet plenum.  Filtered gases
exit the baghouse through the outlet ductwork, pass through the two Induced
Draft Fans, and are discharged into the atmosphere through a common stack.
                                     136

-------
During this mode of operation only the compartment inlet and outlet poppet
valves are in the open position (See Figure 3).

Automatic Fabric Cleaning

     As particulate is filtered from the boiler  flue gas,  ash deposits gradu-
ally build up on the bag surface.   These deposits or ash cake increase the
fabric's resistance to flow,  thus  increasing the pressure loss across the
bags.  This ash cake must be  periodically removed or the resistance to flow
will increase the fabric's pressure to a level  too high for the I.D. Fans
to overcome.  In the Valmy baghouse, cleaning these ash deposits is auto-
matically accomplished.  The  filter bag cleaning system is automatically
activated by the signal sent  by the differential pressure indicator.
(Sensors are located across the baghouse inlet  and outlet duct.)  The
indicator is set to activate  cleaning once the  baghouse differential pres-
sure is sensed at five inches water gauge (See  Figure 7).   Once the cleaning
system is activated all bags  in all compartments are sequentially cleaned
one compartment at a time - compartment number  one through compartment
number ten.  After the total  baghouse is cleaned (approximately 30 minutes)
and is operating somewhere below five inches water gauge,  the differential
pressure indicator is automatically reset and normal filtering resumes.   At
full boiler load and with one baghouse compartment down for maintenance, the
normal filtering mode between baghouse cleanings continues for approximately
45 minutes.  During the automatic  baghouse cleaning mode,  the only dampers
which are opened and closed are the reverse air  and compartment outlet
dampers, the compartment vent damper remains closed and the compartment
inlet dampers remain open (See Figure 4).  Energy required to reverse gas
clean these filter bags is via two 50% capacity centrifugal fan systems.
These fans pull clean heated  gases from the baghouse outlet duct and push
the gas through the bags in a reverse direction.  The particulate is dis-
lodged from the bag surface and falls to the bottom of the hopper.

On-line Maintenance

     The baghouse has been designed with a redundant maintenance compart-
ment.  This allows plant personnel access to the inside of any one compart-
ment for inspection and/or maintenance without  interrupting boiler load.
To insure complete compartment isolation whenever access is desired, a key
interlock lock system was installed.  When the  first key is turned and
released from the key box in the control room,  all compartment dampers
automatically close in that specific compartment.  When the last or sixth
compartment door is locked open, the compartment vent damper opens and the
vent fan is activated.  Ambient cool air is drawn through all compartment
doors and cools the interior  of the compartment (See Figure 5).  At full
load only one compartment at  any one time is allowed to be in the mainten-
ance mode.  To protect plant  personnel from hot  exposed steel partition
walls, 7'-0" tall, expanded metal  guards have been installed at the tube
sheet and bag top levels inside each compartment.
                                    137

-------
Flue Gas By-Pass

     The by-pass system provides the baghouse with a method to divert the
flue gas around the filter bag compartments without interrupting boiler load.
At the inlet to the baghouse, the outlet plenum is extended over the
inlet duct to the baghouse (See Figure 6).  In this section, nine vertical
shaft poppet valves isolate the inlet duct from the baghouse outlet plenum.
When the flue gas is to be by-passed around the filter bags, these by-pass
poppet dampers open and all the other baghouse dampers are closed.  At
Valmy, the by-pass can be manually or automatically activated.  Manual
activation was used at start-up, and the automatic activation occurs when-
ever high differential pressure (above eight inches water gauge) or high
baghouse inlet temperature (above 450°F) is sensed.

PERFORMANCE

     The North Valmy Station was required to meet the emission level of 0.1
pounds per million BTU's fired and 20% opacity as specified in the Environ-
mental Protection Agency, New Source Standards as adopted December 23,
1971.  The baghouse was designed and guaranteed to perform below that
emission level and opacity requirement.  Further, the baghouse was designed
to operate at a differential pressure loss of less than 6.5" of water and
provide Sierra Pacific with a filter bag life of at least two years.

     In January of 1982 performance tests were run in the discharge stack,
and the unit was accepted and passed by the State of Nevada.  Presently,
the opacity in gas discharge stack is monitored and continuously recorded
in the 2%-4% range.  The baghouse differential pressure loss is also con-
tinuously monitored and recorded.  Figure 7, an overlay traced from an
actual strip recording shows the average operating pressure losses (between
less than 5.0" and 6.5" water gauge).  Since start-up some ten months ago,
approximately eight filter bags have failed of the 6,480 bags installed.

                           BAGHOUSE CONSTRUCTION
SCHEDULE AND SEQUENCE OF ACTIVITIES

          Date                               Activity

          5/79                     Start Foundation Framework
          1/80                     Erect Support Steel
          2/80                     Start Baghouse Ground Assembly
          3/80                     Complete Support Steel
          4/80                     Erect Hoppers
          5/80                     Erect Grid Plates
          6/80                     Erect Vertical Walls and Plenums
          8/80                     Erect Roof
          9/80                     Erect Roof Dampers, Vent Duct, Reverse
                                     Gas Duct, etc.
         10/80                     Start Electrical Cable Tray Installation
                                     Perform Air Leak Test

                                     138

-------
          Date                               Activity

         11/80                     Start Field Insulation
          2/81                     Mechanical Erection Complete
          4/81                     Insulation Complete
          5/81                     Electrical Checkout
          6/81                     Tension Bags,  Align Fans,  Adjust
                                     Poppet Valves

     Before erection,  hoppers,  vertical walls, roof,  and baghouse  internal
trusses were ground assembled to their maximum to minimize the quantity of
crane lifts required.

CONSTRUCTION PROBLEMS

     Normal minor difficulties  were encountered during erection,  e.g.,  some
misalignment problems  with ducts and supports, knee brace interference,
insulation interference with hopper reinforcing members, etc.  One unex-
pected problem that did arise was after the filter bags were installed; it
was found they were too long and proper tension could not be set.   To cure
this problem, Flakt had several fabric sewers from the Filter Media Division
of the Carborundum Company come to the Valmy site to cut the bags  to the
proper length and resew bottom cuffs to the bags.  This work was  all per-
formed inside each baghouse compartment with portable sewing machines as
the bags hung from the top caps.  In a span of approximately three weeks
all the bags were properly resewn and properly tensioned.

                             BAGHOUSE START-UP
SCHEDULE AND SEQUENCE OF ACTIVITIES TO DATE

          Date                               Activity

          6/81                     Baghouse pre-start checkout
          7/81                     Boiler boil-out with number 2 oil
          8/81                     Baghouse ready for start-up
          9/81                     First coal-firing and turbine
                                     synchronization
         11/81                     Baghouse placed on line

PRE-START CHECKOUT

     A visual inspection of the complete system was performed.  Filter bag
tension was checked and bags requiring retensioning were retensioned.  Bag-
house compartments, hoppers, and plenums were made clear of all tools and
or debris.  All mounting bolts were tightened, and all poppet valves were
opened and closed and checked for proper sealing and operation.  Reverse
air and vent fans were checked for proper rotation and vibration.  A thirty
minute cleaning cycle was set at the timers in the control panel and a dry
run of the clean cycle was run.  Minor adjustments were made, and the bag-
house was ready for start-up.


                                     139

-------
START-UP

     At the time of the Baghouse Contract, it was envisioned that  the  bags
would be precoated with lime or flyash prior to the baghouse compartments
accepting flue gas.  However, based on experience with other recent  start-
ups, Flakt recommended on-line coating of the bags upon achieving  100%
coal-firing of the Unit.

     The basic plan to start the baghouse was to run on by-pass until  oil-
firing (used for ignition) was discontinued and the boiler was on  100%
coal-firing.  On-line precoating of the filter media could then begin  at a
50% load condition and be completed when the pressure rise across  the
fabric increased approximately 1.5" W.G.

     During the Unit start-up period, the baghouse was set on by-pass.
This period comprised of all initial oil-firing activities including boil-
out, critical pipe steam blows, and initial turbine synchronization.
Additionally, during the three to four week period following the initial
coal-firing of the boiler, the baghouse could not be placed on-line  because
the Unit was load limited and prevented from operating without oil ignitor
assistance.  This inability to immediately achieve 100% coal-firing  and
greater than 50% load operation was caused by various minor start-up prob-
lems, e.g. primary air fan vibration due to imbalance, high silica levels
in the boiler, etc.

     Once 100% coal-firing and 125 MW (50% load) could be maintained with
stability for an extended period, the baghouse by-pass was closed  and  flue
gas was directed to the compartments and filter bags (November 4,  1981).
Within three to four hours the bags were coated with ash and the stack
cleared from a 100% opacity to a 4% to 8% opacity.  The load was slowly
increased to 255 MW and maintained.  During the next several days, the
stack opacity averaged approximately 6% and the baghouse system pressure
drop ranged from three to six inches water gauge.

     The basic problems experienced during the baghouse start-up were
minor.  The baghouse differential pressure sensor clogged and required
cleaning; several bags came loose from the thimbles and had to be
reinstalled.

     Since start-up one year ago, the most major problem has been  that
expansion joints located on the discharge side of the baghouse reverse gas
fans have developed leaks.  These joints have been redesigned and  were
recently installed during a scheduled unit outage.

                                CONCLUSION
     The design, installation, start-up, and performance of this baghouse
has been a complete success.  This, again, demonstrates that fabric filter
baghouses are viable solutions to filtration problems resulting from burning
low sulfur western coals when generating electric power.
                                     140

-------
                          TABLE 2.   FILTER SPEEDS
Filter Speeds (FPM)*
 Including Max.
Reverse Gas Flow
Excluding Reverse
    Gas Flow
One compartment down for
cleaning and one compart-
ment down for maintenance

One compartment down for
cleaning

All compartments on-line
      2.42
      2.15
      N/A
      2.20
      1.96
      1.76
*Based on flue gas flow rate for the boiler performance coal -
 1,102,000 ACFM.
                   TABLE 3.  FILTER MEDIA SPECIFICATIONS
                             STYLE NO. 601B
          Material

          Finish

          Thread Count

          Average Permeability

          Average Weight (ounces per sq. yd.)

          Construction

          Yarn Type - Warp

          Yarn Type - Fill

          Mullen Burst (PSI)
                  Fiberglass

                  7% Teflon

                  54 X 30

                  55

                  10.0

                  3X1 Twill

                  150   1/2

                  150   1/4 TEX

                  450
                                     141

-------
                                                  TABLE 1.   FUELS

Coal
Designation
Heat Content
(BTU/LB)
% Ash
% Sulfur
% Moisture
Flue Gas
Flow Rate
LB/HR X 103
ACFM X 103
For Boiler
Performance
Cl
12,250
7.5
0.7
6.5
2,841
1,102

Possible Fuels
C2 C3 C4 C5 C6 C7 C8 C9
10,290 9,650 10,000 9,700 9,600 8,175 8,000 8,450
8.1 6.0 10.0 10.8 11.5 5.8 11.0 9.0
0.4 0.5 0.6 0.6 0.8 0.48 1.0 0.7
15.0 21.5 12.0 15.0 17.0 30.0 26.0 25.0
2,889 2,921 2,974 2,883 2,864 3,133 2,830 3,098
1,126 1,145 1,160 1,126 1,117 1,246 1,109 1,227

to
     Typical Part.
     Loading
     Grains/ACF
1.48
1.88
1.48
2.32
2.66
2.90
1.59
3.40     2.40

-------
                   Stack
I.  D. Fans
                               I. D. Fans
By-Pass
* ~A

1 ' ,
i
i
9 ' 10
7 j 8
___.L___| 	 L
3 i 4
1 i 2
i
') '

Kev
Baghouse



                                   Rb
                                   Reverse Air Fans

             From Air Heaters




    1.  System Flow
                   143

-------
                    648 Bags per Compartment
                                         Inlet
    Reverse Air
                                     Outlet
Outlet
                              III
                                                        Vent
                                                        1  per  Compartment
                                                           Outlet
                                                           3  per Compartment
                                                              Reverse Air
                                                              2  per
                                                              Compartment

                                   PO
                                        Inlet
Erl
                                                         X
                                           Inlet  —--'
                                           2 per Compartment
Figure 2.  Baghouse Arrangement
                                     144

-------
                                         Vent Damper Closed
                                              Outlet Damper Open

                                               Reverse Air Damper
                                               Closed
Figure 3.  Normal Filtering

      Vent Damper Closed  —^f]
Outlet Damper Closed
                                                     Reverse Air
                                                     Damper Open
                                                     Inlet Damper
                                                     Open
Figure 4.  Automatic Fabric Cleaning Reverse Air Flow
                            145

-------
                        Damper Open
Compartment
Doors Open
                                                   Inlet
                                                Damper Closed
                                                         Outlet Damper Closed
                                                         Reverse Air Closed
Figure 5.  On-Line Maintenance
                                                       By-Pass  Dampers  Open
                                                       All Others  Closed
  Figure 6.  Flue Gas By-Pass
                                      145

-------
   AP
r—  8
   I
   N
   C
   H
   E
   S
 H20
—  6
—  4
       —  2
                          j  (Normal Filtering)
                     (Cleaning)
                     30  minutes
           0  I
              15  min.
Figure 7.   Typical AP Recordings During Cleaning
           One Compartment Down for Cleaning and One Compartment Down
           for Maintenance.
                                   147

-------
               PERFORMANCE OF A 10 MM FABRIC FILTER PILOT
                PLANT AND COMPARISON TO FULL-SCALE UNITS

               by:  Wallace B. Smith and Kenneth M. Gushing
                    Southern Research Institute
                    Birmingham, AL 35255

                    Robert C. Carr
                    Electric Power Research Institute
                    Palo Alto, CA 94303
                                  ABSTRACT

       This paper provides an update on EPRI's fabric filter research program
currently being conducted by Southern Research Institute at the Arapahoe
test facility in Denver, Colorado, and also several full-scale installations.
Data are reported which suggest a quantitative relationship between tube-
sheet pressure drop and air-to-cloth ratio for reverse-gas cleaned baghouses.
Typical results show that average tube-sheet pressure drop increases from
4.5 to 8 in. H20 (1.12 to 2.0 kPa) as the air-to-cloth ratio increases from
1.7 to 2.5 acfm/ft2 (0.52 to 0.76 am /min).  Large residual dust cakes of
up to 1 Ib/ft  (about 500 mg/cm2) were measured in the bags after cleaning
by reverse gas.  Current tests indicate that approximately a 20 percent re-
duction in operating pressure drop could be achieved by increasing the
reverse-gas velocity to 6 ft/min (3 cm/sec), although such a high value may
be impractical for full-scale units.  These results suggest that reverse-gas
alone is only marginally effective as a method of cleaning bags in utility
applications.  Applying intense sonic energy (horns) during reverse-gas
cleaning results in much better dust cake removal.  Tests now underway in-
dicate that this hybrid bag cleaning mode can reduce the operating pressure
drop by more than a factor of two.
                                     148

-------
                        INTRODUCTION AND BACKGROUND
       This paper contains information obtained during the period of July
1981 through August 1982 from the Electric Power Research Institute's fabric
filter research program.  The program includes an empirical analysis of the
operation of a 10 megawatt pilot-scale baghouse, collecting data from sev-
eral full-scale baghouses, and detailed laboratory analyses.  The purpose of
the program is to develop sufficient understanding of baghouse performance
to allow the electric utilities to predict more accurately their performance
before construction and to develop improved designs and operating procedures.

       A description of the fabric filter pilot plant (FFPP) and results ob-
tained between the startup in June 1980 and July 1981 have been published in
earlier papers (1,2,3,4,5).  The FFPP is a versatile, four compartment unit
with a wide choice of operating modes.  Each compartment can be operated in-
dependently.  It is fully instrumented for controlling and monitoring its
operation and conducting aerosol measurements.

       The scope of the test program includes parametric investigations of
operation with reverse gas and shake/deflate cleaning, and different fabrics.
Since most utility installations are cleaned by reverse-gas, we have chosen
to study that mode of operation first, although, some preliminary studies
indicate that the deflate/shake mode of cleaning is very promising.

       In the previous publications it was noted that the design of the FFPP
provides a uniform distribution of gas flow and fly ash between the four
compartments.  It was found that several weeks were required for the system
pressure drop (AP) to stabilize after a significant alteration in its opera-
tion for reverse-gas operation.  Thus, the test periods are rather long for
each parameter evaluated.  In fact, it is not clear if the system ever truly
stabilizes.  It has been observed qualitatively, for example, that it is
easier to make an adjustment in the FFPP operation (e.g. to reduce the mag-
nitude of reverse gas) that degrades performance (AP) than the opposite (e.g.
an increase in reverse gas, with a corresponding decrease in AP).  This may
indicate that reverse-gas cleaning is a marginal method and that the systems
always tend to drift upwards toward a relatively high operating pressure.  It
was observed that heavy residual dust cakes, on the order of 0.5 Ib/ft ,
accumulated on the bags in compartments cleaned by reverse gas.  In some in-
stances, the bags appeared to clean more effectively (decreased AP) after
being left on line longer to accumulate heavier dust cakes.  During a short
period of operation in a deflate/shake mode, the bags appeared to clean much
more thoroughly.  Although several unplanned shutdowns have occurred and the
compartments inadvertently cooled without being purged of flue gas, no irre-
versible "blinding" of bags has been observed.  No bag failures have been ex-
perienced after more than 10,000 hours of operation (the bags are all Albany
International "Tri-coat").  The collection efficiency for all particle sizes
was greater than 99.9 per cent in all of the tests.  Therefore, most of the
tests have concentrated on a better understanding of factors controlling the
operating pressure drop.
                                     149

-------
       Further tests are described in this paper wherein the dependence of
the operation upon the filtering air/cloth ratio (A/C),  the cleaning or re-
verse gas air/cloth ratio (RA/C), and the physical and chemical nature of the
dust cakes was investigated.  Preliminary results from current tests of sonic
cleaning are also described.

                          EXPERIMENTAL PROCEDURES
       To understand the performance of a fabric filter system it is neces-
sary to measure the physical and chemical properties of the aerosol and the
operating characteristics of the system.  The methods used to acquire these
data at the FFPP have been described in the papers cited above.  Similar
techniques are also being used at the full-scale installations.  Briefly,
the data include:  The baghouse design and mode of operation; the baghouse
gas flow (magnitude and distribution); baghouse pressure losses and particle
collection efficiency; gas temperature; chemical analyses of the ash and
coal; particle sizes in the gas stream and bag dust cake; weight, permea-
bility, and photographs of the dust cakes.

       During the FFPP test program, compartments B, C, and D were operated
in a reverse-gas or reverse-gas with sonic energy cleaning mode.  Compartment
A was operated in a deflate/shake cleaning mode during part of the tests and
in a reverse-gas cleaning mode for others.  In order to avoid confusion in
interpreting the data, compartment C was operated throughout as a control
with a filtering air/cloth ratio of 2.3 acfm/ft2, dwell time (time between
cleaning cycles) of 3 hours, and a reverse-gas air/cloth of 2.5 acfm/ft2.
The cleaning sequence in compartment C was:  off line, 90 sec. null, 30 sec
reverse gas, 60 sec. null, and back on line.  The control data were used to
observe and account for fluctuations in AP due to changes in the fuel and
boiler load and long-term variations due to gradual changes in the dust cake.
Operating parameters of the other compartments were varied as described in
the next section.  In order to take advantage of the FFPP's flexibility,
the compartments were controlled and operated independently at constant gas
flow.  This method of operation facilitates interpretation of the data and
is characterized by two distinct values of pressure drop; the pressure drop
before cleaning the bags (AP]jc) and tne pressure drop after cleaning the bags
(APAC)•  It can be shown that the average pressure drop (APave) calculated
from APfiC and AP^c *-n a constant flow mode is within a few percent of the
average pressure drop of a conventional multicompartment unit where the flow
in each parallel compartment is balanced by the pressure drop.

       All of the full-scale units tested had been on line long enough to be
considered in a "stable" operating mode.
                                     150

-------
FABRIC FILTER PILOT PLANT

       Perhaps the most important single parameter in the operation of a bag-
house is the pressure drop during filtration.  In the trade offs between cap-
ital cost,  maintenance cost,  and operating cost it is generally desirable to
operate at  the highest practical A/C ratio.  A series of tests were perform-
ed to determine the dependence of the operating AP upon the filtering A/C and
the dust loading on the surface of the bags.  For these tests compartment C
was operated as described above and again served as the control.  Compart-
ments B and D were operated at the same RA/C as compartment C, but the fil-
tering A/Cs and dwell times were adjusted periodically to yield a range of
data.  The  results of the A/C and dust loading tests are summarized in Figure
1.

       The  dust accumulation on the bags in each cycle is termed "Areal Load-
ing/Cycle"  and is calculated by assuming that all of the dust entering a com-
partment is evenly distributed on the bags.  It was adjusted by extending the
dwell time.  The data sets taken at each value of A/C are shown in Figure 1
where APave is plotted vs. areal loading/cycle.  For each data set, APgc
always had  a positive slope (AP vs areal loading) and AP^c always had a nega-
tive slope.  However, the average (APave), which most nearly represents con-
ventional baghouse operation,  shows only a weak dependence on areal loading,
increasing  for the higher values of A/C and decreasing for the lower values.
From these  data it is concluded that for typical utility operation, a bene-
fit can be  realized by extending the dwell time to keep more compartments in
service on  the average and reduce the stress of cleaning on the bags without
any increase in AP.

       Although reverse-gas cleaning is the most common mode for power plant
baghouses,  little information is available illustrating the relationship be-
tween bag cleaning effectiveness and the magnitude of the reverse gas.  A
series of tests were run on the FFPP to investigate this relationship.  Com-
partment C  was operated in the mode described above.  Compartments B and D
were operated in the same manner as C except that the dwell time in B was
set at six hours.  The magnitude or RA/C was set at 6, 4 and then 2 acfm/ft2
during the  tests for both B and D.

       The  results of the RA/C tests are illustrated in Figure 2.  Compart-
ments B and D were set first to a RA/C of 6 acfm/ft2, operated for about 10
days, set to 4 acfm/ft2 for about 10 days, and finally, set to 2 acfm/ft2 for
several days.  APAC almost doubled as RA/C was decreased from 5 to 2 acfm/
ft2, while APg£ changed very little.  It does not appear that the operating
pressure drop, APave, is a strong function of RA/C; therefore increasing the
reverse-gas flow is not recommended as a practical method of reducing the
operating pressure drop of baghouses.

FULL SCALE UNITS

       A data set describing the operation and performance of full-scale bag-
houses is being accumulated to supplement that taken at the FFPP.  The fol-
lowing paragraphs contain a brief summary of data taken to characterize five
full-scale units and additional data reported by others (6,7,8).  Figure 3

                                     151

-------
                     0.1                    0.2

                       AREAL LOADING/CYCLE. Ib/ft2
                                                                    4517-255
Figure 1.  FFPP tube sheet IP versus air/cloth ratio and dust load per unit area
          (areal loading) per filtering cycle.
                                   152

-------
    12
LU
LU
I
co

CQ
3
    11
    10
re


I    9
«A

-------
    $
   "o
   _
   u
   C
   Q.
   1
   I
   C/3
   LLJ
   02
   13
   cc
25
24

23

22

21

20

19

18

17

16

15

14
13

12

11

10

9
8

7

6

5
4 -

3 -
2 -

 1
        0
AREAL LOADING PER
FILTERING CYCLE (FFPP)
    «fc 0.05 LB/FT2 \

    $ 0.15 LB/FT2 [REVERSE AIR
    £ 0.25 LB/FT2 )
    O SHAKE

UTILITY FABRIC FILTERS
    O MONTICELLO* (SHAKE)
   O ARAPAHOE NO. 3*
    4 CAMEO NO. 2*
    A CHEROKEE NO. 3*
    D ECOLAIRE*
    * HARRINGTON (SHAKE)*
    B KRAMER*
    V MARTIN DRAKE*
   *. NIXON*
    A NUCLA (SHAKE)**
    * TUBE SHEET AP
   ** FLANGE - FLANGE
                         1.0         1.7   2.0
                       FILTERING AIR TO-CLOTH. acfm/ft2
                                                  3.0
                                                              4517-360
Figure 3. Average operating AP versus A/C for the FFPP and nine baghouses installed at
         coal fired plants.
                                   154

-------
is a summary of the average AP vs. air/cloth ratio for the reverse-gas com-
partments of the FFPP at three values of areal loading (from Figure 1),  and
five full scale reverse-gas units.  Also, similar data are plotted for three
shaker units and from the FFPP deflate/shake tests.  Notice that it was  pos-
sible to measure the actual tube sheet Ap for some units, while at others
only flange-flange data were available, which includes an unknown pressure
drop across the ductwork.  The data are generally similar for the reverse-gas
units, all of which are located in Colorado and installed on boilers burning
low-sulfur coal.

       Good agreement is seen between data taken at the FFPP and the full-
scale units.  Notice the linear dependence of AP upon A/C below A/C = 3 acfm/
ft2, indicating the possibility of operating at higher A/C values than the
traditional 2 acfm/ft2 for savings in capital costs.  The increased slope
above A/C = 3 is not well understood, but appears to result from differences
in permeability of dust cakes formed at different face velocities.

       As expected the units cleaned by shake/deflate tend to operate at sig-
nificantly lower pressure drops than the reverse-gas units, except at the
Monticello station, which burns Texas lignite.  This station has experienced
continuous problems of high pressure drop and dust bleed through, and the
interaction of its unusual ash with fabrics of different types is currently
being studied.

DUST CAKE ANALYSIS

       Samples were cut from the bags at each test site after the compart-
ments were cleaned and shut down.  Rather thick residual dust cakes were
found in each instance.  Figure 4 is a cross sectional photograph illustrat-
ing a typical example from a reverse-gas unit.  The cake is rugged, with
large nodular deposits, deep fissures, and a thinner deposit along the "fold
line" where the bag flexed during reverse-gas cleaning.  The actual weights
of  the dust cakes are summarized in Table 1.  The weight of dust per unit
area ranged from 0.4 to 0.6 lb/ft2 for all of the installations investigated
 (40-60 Ib/bag).  Considering that this is the residual cake left after clean-
 ing and that the calculated accumulation of dust during one hour of filtering
 is  expected to be about 0.05 lb/ft2, it is concluded that only about 10 per
 cent of the dust cake is removed by reverse-gas cleaning—probably the
 freshest part of the cake.  Recall again that the data presented in Figure 4
 and Table 1 are for bags that have been "cleaned" by reverse-gas.  Thus, the
 residual dust cake is the one analyzed.

       Studies were made of the bag swatches to determine how permeability of
 the dust cake, or alternately the drag  (AP vs. A/C ratio) depended upon its
 physical properties.  The swatches were placed in a large filter holder and
 their AP vs A/C characteristics measured.  Dust was then removed from the
 swatches for further analysis.  The distribution of particle sizes in the
 range of 4-80 ym was measured using a  Coulter counter, and  the  results re-
 lated to the measured drag.
                                                                    n
        Figure  5 is a summary of all  the  permeability data from  1 ft   samples
 taken from the  top, middle, and bottom of bags at  the  five  full-scale units


                                      155

-------
ui
 -
                     FABRIC
NODULES
                                                            FISSURES
                                                                                                              FOLD AREA
         i  2345673   =

        IC'-s
       'OOt-s

         1234     6789
 12245S7S3     123455789     123456789      1   2  i
         i-40996
                                                                        234     (5789     1234      6789     123
                                                                                                                                   4617-358
                      Figure 4.  Cross sectional view of fiber glass bag and dust cake from a baghouse cleaned by reverse gas.

-------
    Table 1. Summary of the weights of dust cakes from several baghouse installations.
PLANT
       BAG TYPE
MARTIN DRAKE
NIXON
CHEROKEE
(UNIT 3)
CAMEO
(UNIT 2)

ARAPAHOE
(UNIT 3)

FFPP
SHAKE (1/82)
FFPP
REV. AIR (1/81)
ECOLAI RE
PI LOT PLANT
 MENARDI SOUTHERN
    601-T (9.5 oz/yd)

 MENARDI-SOUTHERN
    601-T (9.5 oz/yd)

 FIBERGLASS WITH ACID
    RESISTANT COAT
       (9.5 oz/yd)

 MENARDI-SOUTHERN
    601-T (9.5 oz/yd)

  MENARDI-SOUTHERN
    601-T (9.5 oz/yd)

ALBANY INTERNATIONAL
   Q53-S3016 TRICOAT
  TYPE 11T (NO RINGS)
      (9.2-9.6 oz/yd)

ALBANY  INTERNATIONAL
   Q53-S3016 TRICOAT
    TYPE  12T (RINGS)
      (9.2-9.6 oz/yd)

  FABRIC FILTER CO.
     ACID-FLEX
    STYLE 504-1AF
SIZE OF BAGS
    (LxD)
   30.5'x 12"
   31.75' x 12'
   33.8'x 12"
   22.1' x 8"
   22.1' x 8"
   34'x 11.5"
   34'x 11.5"
   24' x 8"
AVERAGE RESIDUAL WEIGHT
      OF DUST CAKE
       (After Cleaning)
         LB/FT2
           0.5


           0.4


           0.4



           0.5


           0.6


           0.2
           0.2 (1/81)
           0.6 (1/82)
           0.15(11/81)
           1.0 (2/82)
                                                                                4617-267
                                        157

-------
Li.
co
q
o
o
o
LU
_l
Q
Q
Q.
O
                   • //// TOP

                   	MIDDLE

                            BOTTOM
                0.2     0.4     0.6    0.8     1.0     1.2     1.4     1.6     1.8

                  MEDIAN PARTICLE DIAMETER AT TOP, MIDDLE AND BOTTOM
2.0
                           MEDIAN PARTICLE DIAMETER AT MIDDLE
                                                                         4617-266A
        Figure 5. Relative size of particles and drag of dust cakes found at the top, middle,
                 and bottom of bags from several baghouses.
                                       158

-------
and the FFPP.   The data are normalized by dividing both the values of drag
and median diameter measured at the top and bottom of the bags by the values
obtained at the middle of the bags and dividing by the mass of dust per unit
area.  This method of normalization is intended to eliminate variations im-
posed by differences in the inlet particle size distribution at the differ-
ent power plant installations and variations in thickness (weight) of the
values.

       From Figure 5 it can be seen that there is little correlation between
the drag and the median diameters of the particles comprising the cakes.  The
Coulter analysis indicates that the particles near the top of the bags, how-
ever, were somewhat smaller than those from the middle and bottom.  These
data suggest that the drag is not determined by the thickness of the cake or
particle size,  but by the macroscopic structure; i.e., the cracks, voids, and
fold lines.

       Measurements of the ash chemistry, structure, and resistivity are be-
ing made in order to gain a better understanding of the factors that deter-
mine the drag,  or pressure drop.  Although the measured concentrations are
small, the chemical analyses do show progressively higher concentrations of
sulfate in (1)  the hopper samples, (2) smooth cake samples, and (3) nodular
cake samples,  respectively.  This is thought to confirm the conclusion that
an equilibrium situation is reached wherein a relatively small amount of ash
is deposited and removed during each filtering and cleaning cycle, while the
heavy, residual cake remains relatively undisturbed.  No cementitious struc-
ture has been found, however, in our microscopic analysis.  Thus, there is
no indication that the relatively higher sulfate concentration plays a sig-
nificant role in the formation or tenacity of the dust cake nodules.

CURRENT FFPP TESTS

       At the FFPP, research has been initiated to investigate the effective-
ness of sonic energy enhancement on reverse gas cleaning.  Two types of com-
mercially available sonic energy sources (horns) are currently being evalu-
ated.  Cleaning sequences which are being tested range from simultaneous use
of reverse gas and sonic energy to sonic energy alone.  Early in 1983 test-
ing will begin using a "generic" sonic energy source capable of operation at
a continuum of fundamental frequencies and energy outputs.  This will allow
effective sonic energy density within a baghouse compartment (number of horns
and location).   The use of sonic energy to augment reverse air cleaning is
dramatically illustrated in Figure 6.  Here the sonic horn was activated for
10 seconds in the middle of a 30 second reverse gas period.  The reverse gas
air-to-cloth ratio was 2.5 ft/min.  The pressure drop was reduced by more
than one-half when the horn was used.

                                 CONCLUSIONS
       In the FFPP and the full-scale baghouses, heavy residual dust cakes
 are  left on the bags after reverse-gas cleaning.  Only about 10 per cent of
 the  cake is removed during each cleaning cycle, yet the drag is reduced by
 approximately a factor of two.  Accordingly, it is concluded that the

                                     159

-------
01
o
           c
           E
           o
           o
           I
           u
           c

           CL"
           -1
           LU
           O
                PILOT BAGHOUSE ON LINE
                AFTER POWER PLANT OUTAGE
              10
                                             THE HORN WAS ACTIVATED HERE AND DURING
                                             ALL SUBSEQUENT CLEANING CYCLES
CONTROL COMPARTMENT
                  COMPARTMENT WITH HORN


                    I    I    I    I    I    I    I

         I    I    I
                                   10
                                                       15
20
25
                                                             MAY, 1982
    30

4517-390
              Figure 6. Pressure drop/time history illustrating the effectiveness of sonic bag cleaning.  Both compartments were
                       operated at an A/C ratio of 1.2 ACFM/FT? and cleaned every 3 hours by reverse gas enhanced with
                       sonic energy.

-------
pressure drop is determined as much by the structure of the cake as by the
amount of dust present.  For the sites evaluated, there was no particular
correlation between the pressure drop and particle size or chemistry.

       AP vs. air/cloth ratio appears to be a nearly linear function, with a
sharp increase in slope near 3 acfm/ft2.  All of the reverse-gas units eval-
uated performed similarly,  but at a higher pressure drop than has been re-
ported for most shaker units.  If a method could be found to remove the cake
more effectively, it is expected that the operating pressure drop of reverse-
gas baghouses could be lowered by a large fraction without a significant
loss in collection efficiency.

       The dependence of pressure drop on areal dust loading (time between
cleaning) is complex.  For higher values of air/cloth (>2.2 acfm/ft2) the
average pressure drop in the FFPP increased slightly as the dust loading on
the bags (dwell time) increased.  For lower values of air/cloth, the average
pressure drop decreased as the dwell time increased.

       After 10,000 hours of reverse-gas operation no bags have failed.  The
collection efficiency has consistently exceeded 99.9% after the first few
days of operation.  Several weeks are required for the system to stabilize
for each test condition.

       Over the range tested (RA/C of 2.0 to 6.0 acfm/ft2), the effectiveness
of bag cleaning was not a strong function of reverse-gas volume.  Since a
value of 6.0 acfm/ft  is very high for practical operation, reverse-gas
cleaning appears to be a marginal method of removing dust cakes from fabric
filters.

       Preliminary indications are that a sonic aided reverse-gas cleaning is
very effective in removing residual dust cakes.  Future plans include de-
tailed studies of that cleaning mode with the goal of determining the opti-
mum distribution, intensity, frequency, and timing of the sonic energy.
Extended tests of shake/deflate cleaning are also planned.

       The emphasis in all these tests will be to achieve a reduction in
pressure drop and baghouse size without sacrificing reliability or effi-
ciency.

                              ACKNOWLEDGEMENTS
       We would like to express our appreciation to the staff and management
of the Public Service Company of Colorado and the Colorado Springs Depart-
ment of Public Utilities for their cooperation and assistance in evaluating
the fullscale baghouses.  The data reported here were all taken by staff
members of Southern Research Institute.  The FFPP is operated by Kaiser En-
gineers under the direction of Lou Rettenmaier of EPRI.  We would like to
thank Richard Hooper of EPRI for his continuing support in acquiring and in-
terpreting the data.
                                     161

-------
       The work described in this paper was not funded by the U.S.  Environ-
mental Protection Agency and therefore the contents  do not necessarily  re-
flect the views of the Agency and no official endorsement should  be inferred.
                                     162

-------
                                 REFERENCES
1.   Smith,  W. B., R. C. Carr, K. M. Gushing, and G. B. Gilbert.  Air load
    and startup tests of a 10 MW fabric filter pilot plant.  Presented at
    the Fabric Filter Forum, Phoenix, Arizona, January 1981.

2.   Smith,  W. B., R. C. Carr, and K. M. Gushing.  Characterization of a 10
    MW fabric filter pilot plant.  Presented at the Third Symposium on
    Transfer and Utilization of Particulate Control Technology.  Sponsored
    by the U.S. EPA, Orlando, Florida, March 1981.

3.   Carr, R. C., W. B. Smith, and K. M. Gushing.  Characterization of a 10
    MW fabric filter pilot plant.  Paper 81-9.2.  Presented at the Annual
    Meeting of the APCA, Philadelphia, Pennsylvania, June 1981.

4.   Carr, R. C., W. B. Smith, and K. M. Gushing.  Test results from operat-
    ing fabric filters:  full scale and Arapahoe 10MW pilot plant.  Present-
    ed at the EPRI Conference on Fabric Filter Technology for Coal-Fired
    Power Plants, Denver, Colorado, July,, 1981.

5.   Smith,  W. B., K. M. Cushing, and R. C. Carr.  Measurement procedures and
    supporting research for fabric filters.  Presented at the EPRI Confer-
    ence on Fabric Filter Technology for Coal-Fired Power Plants, Denver,
    Colorado, July, 1981.

6.   Ladd, K. L., R. L. Chambers, 0. C. Plunk, and S. L. Kunka.  Fabric fil-
    ter system study second annual report.  EPA-600/57-81-037, July, 1981.

7.   Ensor,  D. S., S. Cowen, A. Shendrikar, G. Markowski, G. Woffinden,
    R. Pearson, and R. Sheck.  Kramer station fabric filter evaluation.
    EPRI Report No. CS-1669, January, 1981.

8.   Ensor,  D. S., R. G. Hooper, and R. W. Sheck.  Determination of the
    fractional efficiency, opacity characteristics, engineering, and econom-
    ic aspects of a fabric filter operating on a utility boiler, EPRI
    Report No. FP-297, November, 1976.
                                    153

-------
                     THE DESIGN, INSTALLATION, AMD
                      INITIAL OPERATION OF THE
                 W. H. SAMMIS PLANT UNIT 3 FABRIC FILTER
                   by: Dennis R. Ross, P.E.
                       Generating Plant Staff Engineer
                       Ohio Edison Company
                       Akron, Ohio UU308

                       James R. Howard,
                       Plant Engineer
                       W. H. Sammis Plant
                       Ohio Edison Company
                       Stratton, Ohio U3961

                       R. Mark Golightley,
                       Plant Engineer
                       W. H. Sammis Plant
                       Ohio Edison Company
                       Stratton, Ohio
                               ABSTRACT
     This paper will discuss the design philosophy and preliminary
operating experience of the first unit of four units to be retro-
fitted with fabric filters at the W. H. Sammis Plant.  Each unit
is l80 MW net.  The unit was operational late August 1982.

     The paper will identify design criteria initially specified,
design enhancements during the course of the project, construction
highlights, and preliminary operating experience.  Results from the
first months of operation will include pressure drop and opacity.
In addition, start-up, shutdown, and maintenance procedures will
be discussed.
                                  164

-------
                             INTRODUCTION
     The W. H. Sammis Plant, located on the Ohio River north of
Steubenville, Ohio, is the second largest of the Ohio Edison Company
system coal-fired plants.  Currently, the units burn coal mined
locally as well as out-of-state coal.  The first four units have a
Net Demonstrated Capacity (NDC) of 180 MW each, with gas flows of
approximately 755,000 ACFM each at 305°F.  Unit 5 is 300 MW NDC with
a gas flow of approximately 1,300,000 ACFM at 2TO°F.  Units 6 & 7
are 600 MW NDC each, with gas flows of approximately 2,600,000 ACFM
each at 270°F.  Flue gas from Units 1 & 2 are combined into Chimney
No. 1 and Units 3 & ^ are combined into Chimney No. 2.  Each of these
chimneys are 500' high.  Units 5 & 6 share the No. 3 Chimney (850* high)
while Unit 7 has its own 1,000* high chimney.  All seven units have
electrostatic precipitators which were originally installed with the
boilers.  Units 1-^ went into service during the period 1959-62, Unit
5-1967 and Units 6 & 7-1969 and 1971, respectively.  The Company is
currently engaged in a program to retrofit high efficiency particulate
collection equipment on all seven units, as the existing precipitators
can not achieve the level of collection required to satisfy the emission
limits.

     For the purpose of this discussion, primarily the equipment,
fabric filters, which have been purchased for the Units 1-U will be
presented.  Units 5-7 are being equipped with electrostatic pre-
cipitators.

                              BACKGROUND
     All of the units were designed and equipped with the then current
state-of-the art emission control features of electrostatic precipi-
tators and tall stacks.  The precipitators, for a variety of reasons,
have never performed to the highest level expected of them.  With
development of particulate emission limitations as required by the
Clean Air Act of 1970, it became apparent that further steps would
be needed to meet the newly imposed regulations.

     In 1972, Gilbert/Commonwealth, Ohio Edison's Architect/Engineer,
was commissioned to perform detailed studies on all of the Ohio Edison
Company plants to determine a compliance strategy to satisfy the limi-
tations imposed by the new regulations.  These studies were updated
routinely, primarily because of changes in the proposed emission regu-
lations and fuel availability.  Most recently, in mid-1977, the U.S.
EPA finalized S02 standards for Ohio.
                                   165

-------
     The federal standards finalized in 1977 are site-specific and even
stack-specific.  Currently, the standards for the W. H. Sammis Plant are
.1 Ibs/MMBTU for particulate and 2.91 Ibs/MMBTU for SC>2 from each chimney.
Utilizing alternate equations to determine allowable S02 emissions per-
mitted some variability to the emissions from individual chimneys.
Needless to say, some interesting combinations were investigated by
which compliance could be met.  The ultimate goal was to establish a
strategy to meet both particulate and S02 emission regulations through
the selection of the proper combination of coals.

     A predominant factor in the development and final selection of
control strategies was the evaluation of site-specific requirements
which also affected the economics.  Probably the most important con-
sideration was the space available for equipment installation.

     The plant is located between railroad tracks and 500'  high hills
on the west and the Ohio River on the east.   Between the plant and
the Ohio River is Ohio Route 7, a four-lane highway.  Available space
for equipment between the plant and the highway was minimal at best.

     Locating the control equipment on the plant roof and in the
south yard were considered, but these locations, upon some detailed
evaluation, were determined to be not feasible.  Other ideas were
explored, including the location of the equipment on a deck-like
structure built over the highway.  The more this concept was explored,
the more feasible it appeared.  Preliminary contact with the highway
department was encouraging, and the concept was further developed.

     The control strategy ultimately selected for particulate and
S02 emission compliance was the use of low sulfur (less than 1$)
for Units ±-k and coal not to exceed 2.8% sulfur for Units 5-7 and
the installation of new high efficiency particulate collection equip-
ment on all seven units.  This strategy was the least costly to
satisfy the regulations.  In accordance with this control strategy,
it was determined that the particulate collectors had to have an ef-
ficiency of approximately 99.1*1$ in order to meet the 0.1 Ibs/MMBTU
particulate emission regulation.

     In January 198l, a Consent Order was signed with the U.S.
EPA which established the compliance dates for meeting particulate
emission limits, as shown in Table No. 1.
                                  165

-------
             TABLE 1. W. H. SAMMIS PLANT COMPLIANCE SCHEDULE
                      Unit No.     Compliance Date
1
2
3
h
5
6
7
January 31, 1983
August" 31, 1982
June 30, 1982
November 1, 1982
April 30, 1981+
August 31, 198U
December 31, 1982
     The Consent Order incorporated new particulate collection
equipment installation on the seven boilers at the W. H. Sammis
Plant and five boilers at three other plants and upgrading par-
ticulate collection equipment on eleven boilers at three plants.
Non-compliance with particulate emission limitations at the W. H.
Sammis Plant could cost Ohio Edison Company $7,500 per day per
boiler.  Interim deadlines established by the Consent Order which
addressed primarily construction activities could also cost the
Company $5,000 per boiler per day fines, (l)

     The total effect of the Consent Order requires that Ohio Edison
Company expend approximately $550 million to reduce particulate emis-
sions.  At the W. H. Sammis Plant alone, approximately $^50 million
will be expended.

                  CONSTRUCTION AND DESIGN HIGHLIGHTS
     As stated previously, the site constraints posed a major problem
from an engineering and construction standpoint.  The deck concept
became the only viable method of supporting equipment.  The deck spans
a four-lane highway between the plant and the Ohio River.  It is 915
feet long and lUO feet wide and allows 20 feet of clearance above the
highway.  The foundation has 192 caisson piers approximately 6 feet
in diameter that go to bedrock.  Approximately 12,000 tons of structural
steel, breeching, fabric filters and electrostatic precipitators, are
supported by the deck.  The deck construction cost approximately $27
million.^'  The deck is shown on Figure No. 1.

     Access to the deck is limited, so construction is staged such that
work activities progress from the plant toward the Ohio River and from
the middle of the deck toward the ends of the deck.  Material and sub-
assemblies from a nearby marshalling yard were moved by barges to the
deck as needed.  Figure No. 2 shows construction activities on the deck.
                                 167

-------
     Draft losses associated with the new fabric filters and the ex-
tensive ductwork necessitated new ID fans for the Units 1-U "boilers.
Modified radial tip fans with two-speed PAM motors are being used.
The PAM motors were selected to reduce low load power consumption,
to allow easier fan start-up and to reduce the risk of boiler implosions.

     Our design criteria for the fabric filters and the ductwork from
the air heater outlet to the chimney is +_30 IWG.  An economic evaluation
was conducted relative to strengthening the boiler to meet the new de-
sign criteria for implosion protection.  In lieu of boiler strengthening,
it was decided to install additional draft controls.  A Bailey Controls
Network 90 implosion protection package was selected for the Units 1-U
boilers.

     Throughout the course of the project, many types of models were
utilized.  Initially, our consultants developed a table-top scale model
of the immediate plant vicinity.  This was used internally to develop
and present alternative equipment locations influenced by the site
constraints.  This model was subsequently used in discussions with
various regulatory agencies in securing approvals for the deck and
barge unloading facility construction, as well as negotiations
relative to the Consent Order.

     Construction of the deck had to be accomplished while maintaining
highway traffic flow at all times.  A model was constructed which in-
corporated scaled-equipment to aid in planning the deck construction.

     Retrofit projects produce unique problems from a standpoint of
engineering and construction.  Site constraints, especially those
imposed at the W. H. Sammis Plant, at times, make these problems even
more burdensome.  Existing facilities must be moved to accommodate new
equipment and facilities must be altered and strengthened to accommodate
the new design.  Our consultants originally constructed a 3/8" scale
model of the back portion of the plant from Units U through 7 to verify
design layouts.  All equipment, piping, steel and ducting was shown in
detail.  As the project progressed, this model became more of a design
construction and planning tool than a design verification model.  Inter-
ferences were quickly recognized and changes were made during the design
stage.

     With the addition of new ID fans and additional lengths of
breeching, boiler control and implosion protection were of great
concern.  Our consultants mathematically modeled the gas side of the
boiler and its controls.  This model defined problem areas and analyzed
the effectiveness of different control schemes.

     As a standard design requirement, we require the precipitator
vendors to flow model the precipitator and duct system to optimize
flow distribution and pressure drop.  At the onset of the project,
this was not required of the fabric filter vendor.  The industry
had been led to believe that a fabric filter was a flow-balancing,
                                 163

-------
pressure-equalizing device.  However, during the project, it was
decided to conduct a flow model study of the fabric filter.  A more
detailed discussion of the fabric filter model will be addressed later.

                         FABRIC FILTER DESIGN
     Coal being burned in Units 1-h is low sulfur eastern coal with
design specifications of: 0.6 to 0.8$ sulfur; 10-20% ash; and 10,000-
12,000 BTU/lb.  Design gas flow is 75^,000 ACFM each at 305°F.  In
1977, early in the planning for the current particulate control program,
it was decided to solicit bids for electrostatic precipitators and
fabric filters.  In June 1978, American Air Filter was issued a
letter of intent to supply four fabric filters, one each for Units l-"k,
Table No. 2 lists pertinent data of the fabric filter.
             TABLE 2. FABRIC FILTER DESIGN INFORMATION	


             Fabric Filter Supplier        American Air Filter
             Number of Compartments        12
             Type of Cleaning              Reverse air
             Air-to-Cloth Ratio-Design
                 Gross                     1.93:1
                 Net(l)                    2.1:1
                 Net-Net(2)                2.31:1    2
             Cloth area per Compartment    32,632 ft.
             Filter Bags
                 Per Compartment           312
                 Total                     3,1^
             Filter Bag Configuration      3w6w3x26
             (l) One compartment off-line for cleaning
             (2) One compartment off-line for cleaning, and one
                 down for maintenance

     The fabric filter reverse air and outlet dampers are poppet-type
and the inlet damper is a flat-plate butterfly damper.  Due to site
constraints and the desire to utilize existing chimneys, the inlet and
outlet of the fabric filter are on the same side.  Figure No. 3 illu-
strates the confined duct arrangement.  The reverse air system has two
100$ sized fans for each unit.  The reverse air circuit is kept hot by
circulating clean flue gas.  When a compartment requires cleaning the
reverse-air bypass dampers close, and the individual compartment reverse-
air damper opens.  A manually operated compartment ventilation system is
also provided.  Automatic control of the fabric filter is accomplished
by an Allen-Bradley programmable controller.  Pressure drop information
is shown in Figure No. h.
                                 169

-------
     The fabric filter is bypassed via two Andco flap-door type
dampers.  The fabric filter will automatically be bypassed on high
flue gas temperature and high differential pressure across the fabric
filter.  The fabric filter is manually bypassed on start-up and can be
operator initiated for emergency or abnormal situations.

     The filter bags are fiberglass with Teflon B coating.  Additional
information is shown in Table No. 3.  In Unit No. 3, thirty-six acid
resistant bags have been installed for test purposes.  These bags are
style QT8-S3016 with Filter Resources' acid resistant proprietary
finish.
              	TABLE 3. FILTER BAG DATA	

               Bag Manufacturer   Filter Resources
               Bag Style          Q79-S3016
               Bag Finish         10% Teflon B
               Weight             9.5 oz/yd2
               Weave              3x1 Twill
               Count              5U x 30
               Permeability       35-6o(l)
               Yarn Designation
                  Fill            75 1/2
                  Warp            75 1/0
               Bag Dimensions
                  Diameter        11 3A"
                  Length          3Uf 7"
               Anticollapse Rings
                  Number          8 per bag
                  Type            Cadmium plated carbon steel
               (l) CFM/ft.2 @ 0.5 IWG
                    FABRIC FILTER FLOW MODEL STUDY
     When the fabric filter was initially procured, a flow model study
was not a requirement.  After seeing results of the flow model con-
ducted for the EPRI pilot plant located at the Arapahoe Plant, it was
deemed appropriate to flow model the W. H. Sammis Plant fabric filter.
American Air Filter was approached with this request, and they agreed
that a flow model would be beneficial.

     The model work was conducted by NELS Incorporated of St. Catherines,
Ontario, Canada.  The model was 1:12.57 scale and made of plexiglass.
It incorporated all the ductwork from the air heater outlet to the
fabric filter, the fabric filter, and ductwork from the fabric filter
                                 170

-------
to the ID fain inlet.  Structural members that affected gas flow,
dampers, and all thimbles were included in the model.  A second phase
of the model addressed the ID fan outlet duct to and including eight
diameters of the chimney.

     The primary objectives of the modeling program were: minimize
pressure drop; achieve acceptable flow distribution to the ID fans;
determine acceptable flow and particulate distribution to the com-
partments; determine acceptable flow and particulate distribution
through the tubesheets; minimize hopper reentrainment; and evaluate
the system under bypass conditions.  Unlike precipitator modeling which
has acceptance criteria established by the industry, fabric filter
flow modeling had no established industry guidelines.  A target
balance for flow within +10% and particulate distribution within +20%
of theoretically equal flow was established.  Through the tubesheet,
a target balance having an RMS deviation less than 25% was established.

     Vaning was added to minimize the pressure losses and to improve
the flow and dust distributions.  Flow distribution values were
generally within or close to the target values.  Dust distribution
was generally outside the target values.  It was recognized that
dust distribution could not effectively be controlled without going
to massive internal control structures and increased pressure losses.
The system pressure loss is essentially constant for various operating
configurations.

     Perhaps the two most important findings of the program pertained
to hopper reentrainment and system bypass.  Flow patterns in the
hoppers indicated severe reentrainment.  The dust flows entering
the hoppers were initially concentrated at the hopper inlet and
continued to recirculate in the hopper without dispersing.  This
condition was rectified by installing ladder vanes and a false floor
in the inlet duct and flow deflectors in the hopper.

     The original outlet duct configuration with poppet-type bypass
dampers had an unacceptably high pressure drop of IT.7 IWG for a full
load bypass.  The entire outlet duct and bypass duct was modified.
The subsequent pressure drop was near that expected under normal
operating conditions. (3, U;

                 PREOPERATIONAL TESTING AND TRAINING
     As the fabric filter installation for Unit 3 neared completion,
 approximately U months before start-up, a team comprised of plant
 and vendor personnel began checking out the various systems.  Wiring
 was verified, and limit switches were set on all dampers which provided
 inputs to the programmable controller.  The cleaning cycle was then
 operated in both the pressure and time initiated mode using the program-
 mable controller.  All other equipment was checked-out in the same
                                 171

-------
manner.  This attention vas useful in that problems encountered with
the reverse air fans, limit switches and control panels were detected
and corrected before start-up.

     In conjunction with the checkout activities, training was conducted
by various vendors and Gilbert/Commonwealth at the plant site.  This
training was directed toward both operating and maintenance personnel.

     Throughout the *t month period prior to start-up, trips were
taken by various engineering and plant supervisory personnel to
operating installations.  This first-hand observation and discussion
opportunity is extremely beneficial.  Learning from the experience
of others allows valuable insight into potential problem areas and
gives guidance for operating and maintenance personnel.  This also
establishes a line of communication to solve future problems that
may be encountered.

     Unit 3 came off line on June 22, 1982 for a scheduled turbine
overhaul and boiler inspection.  During this outage, final tie-in
of the new ductwork was accomplished.  Initial firing of the boiler
began on August 28, 1982.  Finally, on September 1, 1982, precoating
of the fabric filter was ready to begin.

     It was our intent to precoat the fabric filter on line rather
than precoat with cold flyash.  This was a change from the original
thinking during the project.  With American Air Filter's approval,
the following criteria was established: flue gas temperature to be
305°F or greater; gas flow to be approximately U00,000 ACFM; fluidizing
air and hopper heaters at least 2U hours prior to start-up; the unit
had to be on total coal fire; all 12 compartments in service; and no
cleaning until a 3 IWG pressure drop was experienced across a compart-
ment.  At 1:^+0 a.m., on September 1, 1982, the bypass dampers were
closed and flyash laden flue gas was introduced into the fabric filter.

     At start-up, the opacity monitor was not operable.  The indi-
vidual compartment differential pressure gauges were operable, but
needed to be recalibrated.  The primary indication that the fabric
filter was operating was the flange-to-flange differential pressure
recorder.  Precoating continued until ^:00 a.m., when the first cleaning
cycle was initiated.

     Reverse air damper problems were experienced on several of the
compartments.  The damper shaft was hanging up on the guide bearings.
With 9 of 12 compartments operating, the pressure drop was approxi-
mately 3.9 IWG at lUO MW.  At 11:26 a.m., the fabric filter was cleaned
down from a differential pressure of ^.3 IWG to 2.6 IWG.  We have
elected to maintain a differential pressure of 2.6 to 3.6 IWG.  The
design on the Bailey Network 90 implosion protection system needs to
be reviewed and settings readjusted.  In this interim period, we are
overcleaning the fabric filter to minimize system pressure losses.
                                .172

-------
                        PRELIMINARY OPERATION
     Since September 1, 1982, there have been approximately 15 boiler
trips.  These trips have been caused mainly by the boiler implosion
protection system signals and equipment.  To date, the fabric filter has
not caused a unit trip.

     On September 6, 1982, the fabric filter was inspected internally.
At this time, 16 bags were found to have dislodged from the thimbles.
Twelve of the 16 bags were discovered in Compartment B-l which was the
first compartment bagged.  Some holes in the tubesheet and at the thimbles
were also discovered.  All bags were then retensioned.

     The compartments were again inspected on September 2 it, 1982.  It was
observed that some bags needed retensioning, but on a random basis.
Several holes in the tubesheets were observed.  Leakage between the inlet
plenum and the interior of Compartment 'B-l was observed.  This has been
repaired.

     Since September 1, 1982, thirteen bags have been removed.  Seven of
the bags had failed.  Six of the bags were removed because there appeared
to be problems with the bag.  Of these,  one was an acid resistant coated
bag.  These failures have been concentrated mainly in the cuff and first
ring areas.  The bags are being analyzed.  Currently, we are attributing
these failures to installation problems, normal infant mortality of the
bags,' and problems caused by holes in the tubesheet and back wall.

     The opacity monitors were operable  on September 13, 1982.  Opacity
has generally been running between 9-13$.  The opacity is less than the
20$ required opacity, but it is higher than anticipated.  This has been
attributed to instrument calibration, some instrument debugging problems,
and possible overcleaning of the bags.

                      MAINTENANCE CONSIDERATIONS
     The plant personnel are committed to a comprehensive preventative
maintenance program.  This is based on discussions with American Air
Filter and various utilities with operating fabric filters.  Checklists
have been developed to aid in this program.  Specific equipment and areas
are highlighted, and space is provided to note observations.  This data
will be used to provide historical information to troubleshoot problem
areas.

     Because the filter bags are so critical to the operation of the fabric
filter, a filter bag mapping program has been instituted.  A form has been
developed which shows a compartment with all of the thimbles.  An alpha-
numeric matrix is utilized to locate the filter bags.  When a compartment
is inspected, the form is used to note observations such as retensioning
                                  173

-------
requirements, "bag blousing and bags dislodged from their thimbles.   A
second sheet is provided to note bag failures.  On this  sheet,  a filter
bag profile is shown.  Recorded observations would be bag type  (teflon
or acid resistant coating), location of failure, description of failure
and corrective action taken.

     Expansion targets have been installed at the outside corners of
the fabric filter.  These are located exterior to the fabric filter so
they can be easily observed.  It was decided to install  these targets
for early detection of expansion problems or problems with the  column
sliding plates.

                             FUTURE PLANS
     While it is too early to fully assess the operation of the fabric
filter, we are optimistic.  Experience gained from this unit will be
factored into the other three units.  Several tasks have yet to be ac-
complished.

     As previously discussed, the cleaning cycle will be optimized after
adjustments to the Bailey Network 90 implosion protection system are
completed.  From industry experience, the cleaning cycle optimization
could take several months.

     The American Air Filter test program of the Filter Resources'
QT8-S3016 acid resistant bags is scheduled to cover three years.  Ohio
Edison is contemplating its own filter bag test program on Unit 1.  The
primary purpose is to determine candidates for future filter bag replace-
ment.

     Reprogramming of the fabric filter programmable controller is required
for two reasons.  Currently, the cleaning cycle is programmed to stop at
each compartment.  If a compartment is isolated, the cleaning cycle will
still stop at the compartment for its alloted time even though it is
not physically cleaned.  The program will be changed to skip an isolated
compartment.  Also, during the cleaning cycle, the outlet and reverse air
dampers are moving simultaneously for a given period of time.  The program
will be changed to correct this.

     At American Air Filter's request, a temperature survey was conducted
on a compartment, B-6.  They are concerned about the fabric filter tempera-
ture at low load operation due to low air-to-cloth ratio, low flue gas
temperature, temperature stratification and potential acid dewpoint
problems.  The area of concern is primarily at or near the tubesheet
elevation.  We have already added extra insulation on the lower 8' of
the compartment exterior wall.  Flue stops have also been added in the
insulation to prevent convection currents.  Data from the test is currently
being analyzed.

     The work described in this paper was not funded by the U.S. Environ-


                                   174

-------
mental Protection Agency, and therefore the contents do not necessarily
reflect the view of the Agency and no official endorsement should be
inferred.
                                   175

-------
                              REFERENCES
1.   Fines Overhead, Utility Races to Meet Air Rules; Engineering
     News Record, August 5, 1982, pp. 26-27.
2.   Ibid.

3.   Ross, D. R., and Bowen, C.F.P; 1979.  The Modeling of a Baghouse,
     Proceedings of the Vth International Fabric Alternatives Forum;
     Phoenix, Arizona, December 5-6. pp. (3-l) - (3-31).

U.   Bowen, C.F.P; 1981. Gas Flow Model Study.  Proceedings of the
     5th International Fabric Alternatives Forum. Phoenix, Arizona,
     January 15-16. pp. (l-l) - (l-M).
                                   176

-------
Figure 1.  View or deck taken from the southbound lane of Route 7-
           Unit 1 fabric filter is in the "background.
 Figure  2.   View from  deck  looking  towards Unit  k  fabric  filter.
            Structural steel  and  ductwork in  middle  of  the picture
            is  for  Unit 5 precipitator.

                               177

-------
Figure 5. Typical duct  arrangement on Units 1-1*. Fabric filter inlet duct
          enters from the  left.  Fabric filter outlet duct is at top of
          picture. Duct entering the chimney is the ID fan outlet duct.
          > -
          5 -
        •
        I
        .
        r
                                     6 3
                                   /\t^fi*\ t. /ceo
Figure 1*. Guaranteed and  anticipated fabric filter pressure drop vith
          two corrpartments  out of service.
                                   178

-------
                        RESULTS FROM THE FABRIC FILTER
                     EVALUATION PROGRAM AT COYOTE UNIT #1
           H.  James Peters, Arthur A. Reisinger and W. Theron Grubb
                            Wheelabrator-Frye Inc.
                        Air Pollution Control Division
                               600 Grant Street
                       Pittsburgh, Pennsylvania  15219
                                     and
                                Merrill  Lewis
                           Montana Dakota Utilities
                             Beulah, North Dakota
                                   ABSTRACT

    The  fabric  filter  at  Coyote  Unit #1  has been in  service  since  May 1981
collecting lignite flyash  and  sodium-based spray dryer product  as  the second
stage   of  the  dry  scrubbing  system  installed  by  the  Wheelabrator-Frye
Inc./Rockwell  International Joint Venture.

    A program  is  being  conducted  to evaluate  the  performance of  acrylic,
polyester, Nomex,  and  fiberglass  filtration  fabrics  in  full  compartment
trials.  The  unit  uses  predominantly  acrylic fabric  and operates  at filter
ratios of typically  3:1  using  combination  deflation  air  and  mechanical
shaking  as the fabric cleaning method.

    This  paper  highlights the  results  of  the  fabric evaluation  program  to
date  including  comparisons  of  the   pressure drop,  throughput,  replacement
history,   filtration  parameters,  and  changes  in  physical  properties  of the
various  fabrics over the  first year  of operation.  Discussion of the overall
performance of  the unit  and  economic and  technical considerations  for fabric
selection  is  also  covered.
                                      179

-------
                                 BACKGROUND

     In late 1976, prior to final equipment  selection  for the Coyote project,
an extensive  pilot  program was undertaken to demonstrate the use of nahcolite
and  a  fabric  filter  as  an  SO?  and  particulate  collection  process.   The
program  was  conducted  at the  Leland  Olds  Plant of  Basin  Electric Power
Cooperative by  Wheelabrator-Frye  Inc. and Superior  Oil  under the sponsorship
of the  Coyote  Project  Utilities  and  their  engineer, Bechtel.  The resultant
four months  of tests  proved  to  be very positive  with respect  to 502  rernova^
and  sorbent  utilization as  well  as  particulate collection.   S02 removal  in
the  range  of  70-90  percent was  achieved at  stoichiometric  ratios of  1.0  to
1.5(1).  This  compares favorably with reports  (2,3) describing other  testing
of dry injection systems using nahcolite and trona.

     Early in  1977  it  became obvious that the  lack  of commercial availability
of nahcolite would preclude a commercial installation of the  process.   As  an
alternative  to  the  dry  injection  system  utilizing  nahcolite,  testing  was
conducted by Wheelabrator-Frye and Rockwell  International  using a spray dryer
upstream of  the pilot baghouse  as a means  of  injecting an alkaline material
into  the  flue  gas  stream   for  S02  absorption.   This  two  stage  SOp  and
particulate  removal  system   (4)  proved  to  be  technically   and  economically
favorable and was selected for the Coyote Station.


                              PLANT DESCRIPTION

     The  410  megawatt  Coyote  Station  which   has  been  on  stream  since  the
second quarter  of 1981,  is located in  the  city  of Beulah  in  Mercer  County,
North  Dakota  and is fired with  locally  mined  lignite fuel.   The boiler is  a
balanced  draft  cyclone-fired unit  with  twelve  cyclones  and  steam  driven
forced draft  fans.   The system is designed  for  lignite  fuel having an  average
of 7,046 Btu  per pound,  7% ash and  0.78% sulfur.   During  operation  flue gas
exits  two   air  heaters  and  flows  through   four  46-foot   diameter  spray
dryers  into  a  38-compartment fabric  filter.   Two  axial  flow  induced draft
fans  discharge  the  filtered flue   gas  to  a  single stack.   The  flue  gas
temperature  at  the  inlet  to the  fabric filter  is  in  the  range  of 210°-
220°F.    Gas  temperatures beyond  fabric  capabilities  are alarmed  and  gas
flow is diverted through the  fabric filter bypass  system.


                                PILOT TESTING

     Testing for the Coyote  Station was first  conducted in  1976 and examined
the  use   of   nahcolite   as   a  dry   absorbent  in   a   pilot  fabric  filter
incorporating   twelve  11-1/2"  diameter  by 30   ft.  bags  with  combination
deflation  air  - mechanical shake cleaning.   During the nacholite testing and
in subsequent testing of the  spray dryer dry scrubbing process, fabric filter
data was  gathered on  various fiberglass, acrylic,  and  polyester fabrics and
correlated    with    boiler    and     spray    dryer    operating   conditions.
                                      180

-------
     The  resulting correlations allow estimates  of filter fabric pressure
drop during full  scale operation through a  simple  model.
         AP
         K
         L
         T
    AP = SE  V  +  K  L T V2

fabric pressure drop at time T, in w.g.
effective residual  drag coefficient, in w.g.-min/ft
filter velocity,  ft/min
dustload drag coefficient, in w.g.-ft-min/grain
particulate loading, grains/ftj
elapsed time  since  cleaning, minutes
     The duration  of exposure to flue gas  during pilot testing was limited to
at most a few months for each fabric, therefore  no  assessment of long term
performance or  bag life could be made.   In the course of pilot testing no bag
failures occured which were caused by flue gas conditions.  All of the fabric
types tested were  deemed acceptable for  the service.
          COMPARATIVE PRHBURK DROP
                                           COYOTB STATION FABRIC PII.TBR PBRPOIIMANCB
                                                ACTUAL PERFORMANCE

                                                : 6.1 01. ACRYLIC

                                                • OVERALL AVERAGE
                                                           300    «0
                                                           BOILER LOAD-MW
             1       2

               FILTER VELOCITY FPM
                                      2      3

                                  FILTER VELOCITY FPM
      Figures  1  and  2.   Pilot plant and actual operating pressure drop.


    Figure 1 shows the  predicted fabric pressure  drop as a function of filter
velocity for the  lightweight acrylic, polyester,  and  fiberglass fabrics based
on continuous  cleaning  cycles for the fabric  and  expected grain loading.
                                       131

-------
     Figure  2  shows the  predicted fabric  pressure drop  for  the  lightweight
acrylic  only with  representative  actual  operating  points from  full  scale
operation.    The  approximate   relationship  between  boiler  load  and filter
velocity  is  also  indicated.   A  comparison  of  the pressure drop  versus  the
filter  velocity from pilot  unit  correlations  and full-scale  operation shows
the effectiveness of scale-up from pilot plants.


                              PROGRAM OBJECTIVES

     Although the  pilot work  with various  fabrics had  been  successful,  due
consideration was  given to the process  as  a new  application,  and  a  long-term
program to determine  the best  fabric  for this  system in  terms  of  operating
and replacement  costs  was initiated.   A recommendation was made to  the prime
contractor and  the Coyote  partners to  allow  installation of  several  full-
compartment  trials of  various  candidate fabrics.   This  program would allow
evalution  of the  fabrics  in   this  new  application  and  would  enhance  the
selection  of future replacement  bags.   The  program  began with  the  selection
of fabric types and the number  of full  compartments of each which  were to be
installed.

     In  conjunction  with  the  fabric   selection,  a  monitoring  program  was
established  to  review  the  operation of  the entire collector  for comparison
with the  pilot  unit  and  to perform specific  testing  of  the various  fabrics.
The primary  objectives of  this  comprehensive  program were to  determine  the
fabric dimensional stability, filtration performance, and  baglife.


                             FABRIC ALTERNATIVES

     The  process  of   selecting  the  proper  fabric  for  the  Coyote Station
baghouse involved considerations  of  fabric  durability, filtration  performance
(efficiency, throughput,  drag,  and  cleanability), and  economics.    Since the
design  operating  temperature  was 180°F,  synthetic  fabrics  were  considered
as alternatives to fiberglass fabrics, which were  being  utilized  in  all coal-
fired  boiler baghouses in  the  United  States  at that time.    The  synthetic
fabrics had  a  lower cost, potentially  better  durability in the  deflate/shake
cleaning  system,  and  lower operating pressure  drop than fiberglass fabrics as
indicated  in pilot tests.  Fabrics considered  as alternatives  were those made
of acrylic (Dralon T),  aramid (Nomex),  and polyester fibers.

     Polyester  fabric  was  the  most economical  choice and  performed well  in
the  short   pilot   plant  tests   durations   at  up  to   200°F.    Laboratory
exposure   to  180°F/40fc  RH  with and  without  SOp   and with  and without
alkaline  dust  exposure  resulted  in   no  appreciable  strength  loss  of  the
polyester  fabrics.   However,  there was still  a  concern  that  with  the high
moisture content of the flue gas  and the high  alkalinity of the particulate,
if  the baghouse were  to operate for  extended periods  of time in  the 220-
250°F  range, the  polyester  fiber might  be subject  to  hydrolysis  resulting
in severe  strength loss and premature bag failure.
                                      182

-------
     Oral on T  homopolymer  acrylic fabric had  been  successfully used on coal-
fired boiler  baghouses in  Australia.    Acrylic  fiber  is  formed  by addition
polymerization and  thus  is not susceptible to hydrolysis as are polyester and
Nomex, which are formed by condensation  polymerization.   The design operating
temperature  was  safely  below  the  maximum  temperature rating of  284°F  for
Dralon T acrylic.   The only concern regarding the  use of acrylic  fabric  was
dimensional   stability.       Extensibility   problems   had   been    previously
encountered with  acrylic  fabrics in  other filtration  applications.   These
stability  problems  were  discussed with  fabric and bag suppliers who generally
agreed that a  stable  acrylic fabric could  be  produced  by proper heat-setting
and finishing techniques.

    Nomex  fabric was  considered,  in spite  of  its  high relative cost, because
of its potential for  increased cleanability and  lower  pressure drop compared
to  fiberglass, based  on  experience   in  the  Southwestern Public  Service  Co.
Harrington Station Unit #2 baghouse.   In  addition, Nomex  fabric would be able
to  withstand  the   normal  flue gas  temperatures  at  the  spray  dryer   inlet.
Because of  the alkalinity of  the excess soda  ash  reagent, there  was   little
concern about  acid degradation of the  Nomex fabric.

     Fiberglass fabric  was  included  based  on  successful  operation  in  other
coal-fired  boiler  baghouses  and  for   full  evaluation  as  a replacement  fabric
in  the  unlikely  event  that  spray  dryer  operation  could  not  provide  the
necessary  flue gas cooling  for  the  synthetic fabrics.   Since deflate/shake
cleaning was  to  be used, synthetic fibers  were  expected to  provide superior
bag life to fiberglass which has  inherently poor flex abrasion  resistance.
  Fi ber
  Type
          TABLE 1.  NOMINAL PROPERTIES OF  INITIAL FABRICS  INSTALLED.
Yarn
Type
 Weight                   Finish
(oz/yd2)    Finish Type     Code
   No.
Compartments
  Acrylic    Spun             6.1

  Acrylic    Spun            10.0

  Nomex      Spun             5.5

  Polyester  Filament/Spun    6.1

  Polyester  Woolen Spun     12.0

  Glass      Filament/Text.  10.0

  Glass      Filament/Text.  13.5

  Glass      Filament/Text.  13.5

  Glass      Filament/Text.  13.5
                        H20 Repel!ant      E

                        H20 Repel 1ant      E

                        Acid Resistant     F

                        Heat-Set           G

                        Acid Resistant     F

                        Teflon B (10%)     A

                        Acid Resistant     B

                        Acid Resistant     C

                        Acid Resistant     D
                                         19

                                         12

                                          1

                                          1

                                          1

                                          1

                                          1

                                          1

                                          1
                                      183

-------
     Based  on  all  of  the above  considerations,  acrylic  was  chosen  as  the
primary fabric  for use at  Coyote.  This choice was  based  on the expectancy  of
lower fabric  pressure loss, increased bag  life,  and reduced cost  compared  to
fiberglass  fabric,  without the  potential  risk  of hydrolysis  which  might  be
encountered  with polyester fabric.   Special  fabric  finishing and rigorous
quality control testing were utilized to  minimize any potential stretching  of
the  acrylic  bags.    Two   different  acrylic fabrics  were  selected:    the 6
oz./yd^  fabric which  had  been  used  for all  pilot  plant  and  laboratory
evaluations  and  a  10 oz./yd2  fabric similar  to that  used  successfully for
boiler applications in Australia and in many  industrial  applications.   A  list
of the  nine initial  fabrics selected and their nominal properties  is given  in
Table 1.
                          FABRIC FILTER OPERATION

   The Coyote fabric filter  is cleaned by a  combination  of deflation air and
mechanical  shaking.   During  cleaning, an  individual  compartment is  isolated
from gas  flow  and controlled  reverse  flow  is used to collapse  the  bags to a
partially  open  configuration.   The bags  are then  cleaned by shaking  action
for  10-15  seconds,  time  is   allowed  for dust  to  settle,  bags are  gently
reinflated  and the compartment  is placed back  into filtration service.

     In normal operation,  overall  fabric  pressure drop  (including  inlet and
outlet  manifold   and   housing  losses)   is   maintained  by  pressure  drop
initiation  of  the cleaning  cycle  at  6"  w.g.   At  low  loads,  where  the  6"
setpoint  would  not   be reached  within  several  hours,  cleaning  cycles are
initiated  on 3-6  hour intervals.  This maintains  an average dust  loading  on
the  fabric and allows  the fabric  filter to  easily accommodate rapid  boiler
load  increases.   Continuous  cycles (each  compartment  cleans  after  one hour
filtering  time)   are  used at full  boiler  load to  maintain  approximately 4"
fabric and  6" overall pressure  loss at 3 ft/min  filter velocity.

     The  fabric  filter  contributes  significantly  to  the   sulfur  dioxide
removal  as  the  second  stage  of  the  flue   gas  desulfurization   process.
Residual  sodium  carbonate  leaving  the  spray dryer continues  to  react with
SOg  during filtration.  The  fabric filter typically accounts  for  as much as
20 percent  of the overall S02  removal.

     The  fabric  filter  performance  as a  particulate  collection device  is as
anticipated.  Particulate emissions have been  measured  and  are  well  below the
required new source performance standards.

     In  order  to  properly  evaluate the  overall  performance  of the  fabric
filter   and  the   relative   performance  of  the  various   fabrics,  data
on  individual  compartment pressure  drop  and  relative  flow was  collected  at
periodic intervals.

     Under  steady operating  conditions  the  pressure  loss across  the  fabric
was measured for  each  of  the compartments  immediately before  and  after the
compartment  was   isolated   for  cleaning.    Together  with   the  individual


                                      184

-------
compartment  AP  data,  sufficient  information  on  boiler  and   spray dryer
operation was  taken  to estimate the   total  gas flow  and  participate  loading
to the fabric filter.

    Twelve  compartments,   including  the  nine alternative  fabrics  plus three
additional acrylic compartments  were outfitted with pressure  taps 'to measure
the  pressure  loss across  the outlet valve  opening.   This measurement allows
the flow rate through the compartment to be  determined.   To ensure meaningful
comparisons the  data was  coordinated  with  the fabric filter  cleaning cycle
and taken both  immediately before and after cleaning.   The outlet valve  AP
data  enables  relative flow  comparisons  between any of the test  fabrics.   In
combination with  total  gas flow data, a  weighted  average of the  data allows
the filter velocity for each of the fabrics  to be established.

    This  individual  compartment flow  can also  be compared  with flow rates
calculated directly  from  the outlet valve AP value based  on generalized flow
correlations developed during scale model flow studies of  the  outlet  duct and
valve  air flow  configuration.   Good  agreement was  found  between  the model
study correlation  and  flow rates derived  from  system operation  and  weighted
average data.

     With individual  compartment  AP  and  filter  velocity data available as
well  as  system operating  data,  effective  residual  drag  and  dust  load  drag
coefficients can  be  determined  for  each  individual  fabric as  had  been done
during pilot testing.


                          RESULTS OF FABRIC  TESTING

     During scheduled  boiler outages,  test  compartments  were  inspected  and
sample  bags removed  for  measurement  of  fabric   strength and  other fabric
properties.    In  addition,  control  bags  were measured  to  track  dimensional
stability,  and  several bags  in each test  compartment were tested for filter
drag (5) and residual dust weight.

     The results  of  Mullen Burst testing  are presented  in Figure 3.  Results
from measurements  of M.I.T.  flex  vs.  length of  service  for  the  fiberglass
fabrics  are given  in Figure  4 and  for  the  synthetic  fabrics  in  Figure 5.
Note that except for the  polyester  fabrics,  the strength  as  measured by both
tests stabilized  at  levels indicating good  potential  baglife.   Of particular
interest  is the  M.I.T.  flex tests performed on the synthetic  fabrics.   This
type of  testing  is typical for fiberglass  filtration  fabrics but not usually
performed on synthetic  bags.   The M.I.T.  flex  values  were a  clear   indicator
of  the  progressive  hydrolysis  of  the  polyester  which  eventually  led  to
failure  and replacement.   The absolute  value of  the M.I.T. flex  test rather
than  the percentage  loss  appears  to  be a  better  indicator  of potential bag
failure.

     Fabric  in-situ   drag  measurements  and residual  dust weights  were  for
the most part  in  agreement with pressure  drop  and relative flow measurements
                                      185

-------
taken  during operation,  although the state  of the bags at  the time actually
tested  (elapsed time since cleaning,  length  of time  out of  service,  ambient
conditions,  etc.) was not  necessarily the same for each fabric style.

     The  lightweight acrylic  and  Nomex  fabrics  consistently indicated  the
highest  relative flow  (throughput),   low  drag values,  and  low  residual  dust
weights.   The 10 oz. acrylic and all of the  fiberglass fabrics  operated  at
higher drag,  residual dust, and pressure drop with lower throughput.

     Of  interest is the  low  residual dust  loadings  found  for the  fabrics.
Residual  dust weight for  the lightweight acrylic  was as low  as 7  Ibs.   This
can  be  contrasted  with  as   much  as  50-100   Ibs.  for  fiberglass   fabrics  in
reverse  air collectors filtering boiler  flyash.   The shaking  action  on the
lightweight  fabric  maintains  a  low  residual  dust,   and  at  the  same  time
particulate  collection  efficiency  is maintained  at  a  high   level.   Residual
dust weights for the  10  oz. acrylic and fiberglass fabrics  in   the  Coyote
fabric filter were  roughly 50% greater than that of the lighter acrylic.
                                           1
                                           i
                                           1W
                                                                 MULLEN BURST TESTING
                         200

                LENGTH Of SERVICE (DAYII
100         200

 LENGTH Of SERVICE (DAYS)
 Figure 3.  Results of Mullen  Burst tests for synthetic and fiberglass fabric.


     Overall  performance of the  lightweight acrylic fabric  has  been superior
to the  other  test fabrics.  Although a  substantial  decrease in  Mullen  burst
strength  of 50-60% was  observed during  the initial six months  of operation,
the fabric  physical properties have stabilized at a  level  which  should result
in more  than  adequate bag  service  life.   The few bag failures which did occur
were due to random physical  damage  during installation  and minor construction
modifications.
                                      186

-------
     The Nomex  fabric has, also  performed  very well to  date.   Fabric strength
has  stabilized  at  a  level  which  indicates  good  potential baglife.   A  full
compartment  of  Nomex  has been  in  service  for  27 months  at  the  Southwestern
Public  Service  Harrington  #2  baghouse.     The  Nomex  fabric  has  exhibited
excellent  cleanability,  low pressure drop,  and  low filter drag  as determined
during single bag testing.    The relative  flow  has been  slightly  lower  than
the lightweight acrylic  based  on compartment outlet valveAP  measurements.

     The   10  oz/yd^   acrylic  fabric   has  exhibited   consistently  higher
pressure drop than the  lightweight acrylic  fabric.    Single  bag  filter  drag
was  higher  than  for  the other fabrics  and  compartmental  relative  flow was
lower  than  the    6  oz/yd2   acrylic.     The  bulky   yarn  construction  and
resistance to shake cleaning  of the  heavier  weight  fabric apparently  result
in poor cleanability.

     Fiberglass  fabrics  exhibit  excellent  strength  retention   after  twelve
months  operation,  as  would  be expected   at the  low  operating  temperature
(relative  to the  550°F  capability  of  glass fabrics)  and  moderately alkaline
conditions to which they have  been  exposed.
                          MIT FLEX TESTING
                LENGTH Of SERVICE IDAYSI
                                                       LENGTH OF SERVICE IOAYSI
        Figure  4.   Results of M.I.T. flex tests for fiberglass fabrics.


    Polyester  fabrics experienced  drastic  loss of  strength  and  high  rate of
failure due to chemical  degradation of the polyester fiber.  Exposure to  240-
250°F  operating  temperatures in  the  presence of the  typically high flue gas
moisture  content and  moderate  alkalinity  resulted  in  hydrolysis  of  the
fiber.   If it  had been known that actual operating temperatures  would be  230-
250°F   rather   than  the  180°F  design,  polyester  fabrics  would  likely  have
been omitted  as candidates for evaluation.
                                       187

-------
A
                        MIT FLEX TESTING
                   L
              ACRYLIC FABRICS
                                             O  !"•  -
                                                                              MIT FLEX TESTING
          LENGTH OF SERVICE JOAVSI
                                                                  I              I
                                                                 IN             a*
                                                                 LENGTH OF SERVICE (DAVS*
 I
no
                             POLYESTER ««t
          LENGTH OF SERVICE (DAYS)
                                                                  LENGTH OF SERVICE IDAVSI
Figure  5.   Results  of  M.I.T.  flex  test  for  synthetic  fabrics.
                                         183

-------
     The  overall  performance  of  the  Coyote  filtration  fabrics   has  been
excellent.  Excluding the polyester fabrics only  35  failures have occurred in
the  first  eighteen  months  for  a  failure  rate  of  less  than  0.5%.   These
failures  (detailed  in  Table 2) are all  due  to miscellaneous  physical  damage
which  occurred  during  installation   or  as  a result  of  minor  construction
modifications required  inside the compartments after bag  installation.
                         TABLE 2.  BAG FAILURE SUMMARY.

Start-up through September 8, 1982 (excluding polyester bags*):

         Primary bag failures                                 -  33
         Secondary failures  (damage by adjacent bag failure)  -   2_
         Total over sixteen months                            -  35

Primary failures by fabric type:
         Acrylic
              6 oz.
             10 oz.
         Fiberglass
             10 oz. -  Finish A
             13.5 oz.  Finish B
             13.5 oz.  Finish C
             13.5 oz.  Finish D
         Nomex
              5.5 oz.

         Total
Number failed


     16
     11

      1
      0
      0
      4

      I

     33
                                                   Percent failed
0.41
0.45

0.49
0
0
1.96

0.49

0.45
Causes of bag failure:

         Specific  physical  damage:
              Abrasion  against  internals                18
              Damage to bottom  during  installation       7
              Top  cuff  damaged  by bag  clamp              2
              Cause undetermined                         6
         Secondary failures:                             2
         Total                                          35

*Note:  Both types of polyester bags were  severely  damaged  due  to  hydrolysis of
the fiber.  Prior  to total  replacement 25  12  oz.  bags  and 29  6  oz.  bags failed.
                                      189

-------
                              ACTUAL VS DESIGN

     The fabric filter was designed for a nominal  air-to-cloth  ratio of 3-to-
1  with   two  compartments   cleaning  and   two   compartments   off-line  for
maintenance.  Tests  performed  during the past eighteen   months indicate that
air-to-cloth ratios  in  a range of 2.5 to 3.8-to-l have been achieved.  During
this time  the fabric  filter  has  maintained  pressure losses  of  4-8 inches
(including  manifold and  housing  losses)  with stable pressure drop  at each
flow rate.  During periods of reduced boiler  load,  the fabric  filter pressure
loss  operates  at  the   previous  values  for  that  specific  load.   Although
fabric  air-to-cloth ratios  have  been  higher  than  anticipated   the  fabric
filter  has  consistently operated without  difficulty.   Filtration  performance
(pressure drop,  cleanability,  and  efficiency) has  remained  relatively stable
in  spite of  a  wide variety  of boiler  and  FGD  system  operating  conditions.
Pressure excursions, due to  boiler  load swings,  uneven  gas  distribution from
spray  dryers,  fabric  filter  control  or  equipment malfunctions,  etc., have
only been  temporary, and  when system operation  returned to  normal,  so  did
fabric pressure drop.

    The  higher   temperature  operation  (230-250°F operation  for  much  of the
first  twelve months compared  to 180°F  design)  resulted  in  discoloration of
the acrylic fabric as would  be expected, but had  no serious  effect on fabric
strength,  expected  service   life   or  dimensional  stability.     The  higher
temperature  did  accelerate the  hydrolysis failure  of the  polyester  fabrics
leading  to  their  replacement  with  acrylic and at  this  time polyester is not
considered a viable  alternative for  this application.


                                  ECONOMICS

     The  use of  acrylic  fabric in  lieu  of  fiberglass offers   significant
economic advantages  due  to:

     o    Lower fabric cost.

     o    Lower  pressure drop  at equivalent throughput.

     o    Less possibility of  installation damages.

     o    Potentially longer bag life.

     Bags fabricated of the six ounce acrylic fabric  are priced  20% less  than
a  nominal  ten  ounce fiberglass  bag  with  10% teflon  finish yielding a clear
economic  advantage  for  the  acrylic in capital  costs.   In  addition, for the
Coyote  operation,  assuming  one  inch   of  pressure  loss  is   equivalent  to
$50,000  per  year  operating  costs,  lower  pressure  drop operation  reflects
significant  savings.  Longer fabric  life may  also be realized  for  the  acrylic
fabric.

     Nomex   can  be  considered  a  viable  alternative  fabric.    However, the
economics  indicate that an  additional  year  of  baglife  would  be  required of

                                      190

-------
the  Nomex assuming  a  three-year life  for the  acrylic,  and also  a one  inch
lower   pressure   loss.      This   remains   to   be   substantiated   through
continued operation.
                                   OUTLOOK

     The fabric  filter evaluation program at Coyote will continue through the
life of  the  initial  set of  bags.   Modifications  to  the cleaning  action for
the  fiberglass  bags  are  being considered in order  to  improve their relative
performance.  The  basic objective of  identifying  an  optimum fabric selection
for  eventual   rebagging   of  Coyote  has  been   satisfied  by  the  superior
performance of the lightweight acrylic.

     After eighteen  months of fabric  filter operation at Coyote with superior
performance in  terms of low pressure  drop  at  high filter  velocities,  fabric
replacement experience  and service life expectancy the combination deflation -
mechanical shaker  type fabric filter  using synthetic  bags  is  demonstrated to
be  an  exceptional  selection for the dry scrubbing application.  In a side-by-
side comparison  the  general  serviceability  of fiberglass  bags  with  shaker
cleaning  is  also  demonstrated.   Fabric filter  operating temperatures  can be
maintained  below  the  upper  limits   for  the  synthetic fabric   even  during
periods  of off-normal  operation  by  the  spray dryers.   The  ability  to use
acrylic  instead of  fiberglass  offers  significant  economic   advantages  and
potentially longer fabric  life.

    The  work  described in this paper  was not funded by the U.S. Environmental
Protection Agency  and therefore the contents  do  not necessarily  reflect the
views of the Agency  and no official endorsement  should be inferred.


                                  REFERENCES

1.   Estcourt,  V. F., Grutle,  R.O.M.,  Gehri,  D. C., and Peters, H. J., "Tests
    of   a  Two-Stage Combined  Dry  Scrubber/S02  Absorber  using  Sodium  or
    Calcium",  American Power Conference, Chicago, Illinois, April 1978.

2.  EPRI  Report FP-207,    "Evaluation of  Dry  Alkalies  for  Removing  Sulfur
    Dioxide from Boiler Flue Gases", October, 1976.

3.  Muzio,  L.  J.,   Sonnichsen, T.  W.  et  al.,   "Demonstration  of  S02 Removal
    on a Coal-Fired  Boiler by Injection of Dry  Sodium  Compounds",    EPA/EPRI
    Symposium on Flue Gas  Desulfurization, Hollywood, Florida, May 1982.

4.  U.S. Patent  No.  4,197,298, "Sequential Removal of  Sulfur  Oxides  from Hot
    Gases" issued  to Wheelabrator-Frye Inc. and  Rockwell International.

5.  Grubb,  W.  T.   and Banks,  R.  R.,  "Field  Evaluation  of  the  Drag  of
     Individual  Filter Bags",   4th Symposium on the  Transfer and Utilization
    of Particulate Control Technology, Houston,  Texas, October, 1982.
                                      191

-------
            BAGHOUSE PERFORMANCE AND ASH CHARACTERIZATION AT THE
                           ARAPAHOE POWER STATION

                 by:  Robert S. Dahlin
                      Southern Research Institute
                      Birmingham, Alabama  35255

                      D. Richard Sears
                      U.S. Department of Energy
                      Grand Forks, North Dakota  58202

                      George P. Green
                      Public Service Company of Colorado
                      Denver, Colorado  80202
                                  ABSTRACT

     This paper presents the results of a field test conducted in March, 1981,
on the Unit 3 baghouse of the Arapahoe Station of the Public Service Company
of Colorado.  The unit was burning a subbituminous coal from Routt County,
Colorado, and was retrofitted with a baghouse in 1979.  Baghouse performance
was found to be excellent with an overall mass efficiency of 99.98 percent.
The estimated cumulative collection efficiency of all particles smaller than
two microns was 99.92 percent.

     The average electrical resistivity of the fly ash was found to be 6 x 10
ohm*cm at 266°F.  This was consistent with the low level of SOs found in the
flue gas (^ 0.3 ppm).  The measured resistivity agreed reasonably well with
that predicted by Bickelhaupt's technique using the analysis of the ash ob-
tained by atomic absorption spectrometry.  These results suggest that a high
SCA and possibly flue gas conditioning would be required to obtain the same
collection efficiency from an ESP.

     A microanalytical characterization of ash elemental and mineral composi-
tion has been performed.  Using SEM, detailed major element analyses for ^ 240
individual ash particles have yielded concentration frequency distributions,
composition as function of particle size, and inter-element concentration
correlations.  The  results have been correlated with coal and ash mineralogy.
Suggestions are made concerning possible application of this data to practical
problems of fly ash control and disposal.
                                       192

-------
                                 INTRODUCTION
     The Grand Forks Energy Technology Center of the U.S. Department of Energy,
through its contractor Southern Research Institute, is conducting a series of
field tests to study the characteristics and control of fly ash from the com-
bustion of low-rank Western coals.  Emphasis is being placed on two types of
coal:  Western subbituminous coal and Texas lignite.  The overall objective of
the project is to provide data that will be useful in evaluating particulate
control technologies for these coals.

     To date, two field tests have been performed.  The first field test, the
subject of this paper, was conducted in March 1981, at Unit 3 of the Arapahoe
Power Station of the Public Service Company of Colorado.  The unit, retrofitted
with a baghouse that was placed in service in May 1979, was burning a low-
sulfur subbituminous coal from Routt County, Colorado.  The second field test,
which was jointly sponsored by DOE and EPRI, was completed in May 1982 at the
San Miguel Station of the San Miguel Electric Cooperative located between
San Antonio and Corpus Christi, Texas.  Results of the San Miguel test will be
reported elsewhere.

                            OBJECTIVES AND SCOPE
     The primary objectives of the Arapahoe test were to evaluate the per-
formance of the baghouse and characterize the fly ash.  This information will
be used, along with information from other power plants, in evaluating various
particulate control technologies for low-rank Western fuels.  In this case,
the fuel is a low-sulfur Western subbituminous coal.  Baghouse performance was
evaluated in terms of mass efficiency, fractional efficiency, and pressure
drop.  The fly ash was characterized in terms of its particle size distribution,
chemical composition, and electrical resistivity.

     The scope of the test program encompassed the following elements:

     (1)  Determination of inlet and outlet mass loadings and mass
          efficiency by EPA Method 17,

     (2)  Measurement of inlet and outlet particle size distributions
          and fractional efficiency using calibrated cascade
          impactors,

     (3)  Documentation of boiler and baghouse operating conditions
          during testing,

     (4)  Sampling and analysis of coal by ASTM methods,

     (5)  Sampling and analysis of fly ash using glass fiber thimbles
          and a five stage cyclone assembly for sampling with analysis
          of major elements by Atomic Absorption Spectrometry (AA) and
          minor elements by Spark Source Mass Spectrometry  (SSMS),
                                       193

-------
      (6)  Analysis of major flue gas  components by Orsat apparatus with
          moisture determined by adsorption on Drierite dessicant,

      (7)  Determination of ash resistivity using an in situ point-
          plane probe,

      (8)  Analysis of particulate and gaseous organics by gas
          chromatography (GC) and high performance liquid chromatog-
          raphy (HPLC) of extracts of fly ash and XAD-2 resin exposed
          to flue gas in a SASS train,

      (9)  Measurement of SOa and SO3  (H2SOt») levels in the flue gas by
          the Cheney-Homolya method,

    (10)  Collection of two metric tons of coal for use in the pilot-
          scale Particulate Test Combustor at Grand Forks,

    (11)  Mineral characterization of the coal by x-ray diffraction
          (XRD) of as-received coal and of oxygen plasma low tempera-
          ture ashed (LTA) coal, as well as by chemical fractionation,

    (12)  Major element analysis of coal and size-fraction-
          ated ash by neutron activation analysis (NAA), and

    (13)  Detailed microanalytical characterization of size fraction-
          ated ash using scanning electron microscopy/electron micro-
          probe (SEM).

Space limitations preclude detailed presentation of the entire effort here.
Therefore, several phases of this study will appear elsewhere.  See, for
example, references 13 and 14 for a more complete exposition of the micro-
analytical results.

                              SITE DESCRIPTION
BOILER

     The Unit No. 3 steam generator at Arapahoe is a vertically down-fired, dry-
bottom boiler supplied by Babcock and Wilcox.  The unit was placed in service
in 1951, and now operates as an intermediate peaking unit and used for cycling
duty.  Nominal generating capacity is 46 megawatts.  The design specifications
and nominal operating conditions of the boiler are summarized in Table 1.

FUEL

     At the time of the testing, the Unit No. 3 boiler was burning a low-
sulfur subbituminous coal from Routt County, Colorado.  Proximate and ultimate
analyses of the coal performed after completion of the testing revealed that
the coal was of somewhat higher quality than anticipated, being borderline
between a subbituminous A and a high volatile C bituminous coal.  Proximate
and ultimate analyses of daily composite coal samples are given in Table 2.

                                      194

-------
  TABLE 1.  BOILER DESIGN AND NOMINAL OPERATING PARAMETERS
 Manufacturer
 Generating Capacity
 Year On-Line
 Duty Cycle
 Firing Method
 Slagging Method
 Steam Header Temperature
 Steam Header Pressure
 Fuel
      Heating Value
      Moisture
      Ash
      Volatile Matter
      Fixed Carbon
      Sulfur
 Excess Combustion Air
 Fuel Feed Rate
 Combustion Air Flowrate
      Babcock and Wilcox
      46 Megawatts
      1951
      Cycling*
      p-c vertical (down-fired)
      Dry bottom
      900°F  (482°C)
      870 psig  (6100 kPa)
      Western subbituminous coal
      10,600 BTU/lb (5890 kcal/kg)
      8%
      10%
      40%
      42%
      0.5%
      15%
      46,800 Ib/hr (21,200 kg/hr)
      450,000 Ib/hr (204,000 kg/hr)
*Generally base-loaded Monday through Friday; boiler designed
 for quick start-up and unit taken off line during low-load
 periods.
        TABLE 2.  PROXIMATE AND ULTIMATE  COAL ANALYSES
       	(AS RECEIVED)	
         Date
        Sampled
3/3/81    3/4/81    3/5/81    Average
    Proximate
      %  Ash
      %  Volatile
      %  Fixed Carbon
      BTU/lb
      %  Sulfur
 4.62
 9.31
43.63
42.44
10,738
 0.49
13.38
10.46
34.82
41.33
10,454
 0.48
 6.19
 9.94
41.74
42.13
10,786
 0.48
 8.06
 9.90
40.06
41.97
10,659
 0.48
                            195

-------
               TABLE 2.  PROXIMATE AND ULTIMATE COAL ANALYSES
               	(AS RECEIVED)	(Cont' d)
                Date
               Sampled
3/3/81    3/4/81    3/5/81    Average
            Ultimate
% Moisture
% Carbon
% Hydrogen
% Nitrogen
% Chlorine
% Sulfur
% Ash ,
% Oxygen
4.62
61.02
5.15
1.47
0.07
0.50
9.32
17.85
13.38
59.75
4.19
1.39
0.07
0.49
10.46
10.27
6.19
60.62
4.95
1.38
0.04
0.48
9.93
16.41
8.06
60.46
4.76
1.41
0.06
0.49
9.90
14.84
         //Analyses performed by Commercial Testing and Engineering,
          Inc. using standard ASTM methods.
         tBy difference.
Perhaps the most noteworthy feature is the low sulfur content (^ 0.48%).  The
high moisture content on March 4 was believed to be due to snowfall during the
preceding night.  Otherwise, the composition of the coal appears to have
remained fairly constant during testing.

BAGHOUSE

     Unit No. 3 was retrofitted with a Joy-Western baghouse which was placed
in service in May 1979.  The baghouse is a reverse-air unit with Teflon-coated
fiber glass bags.  The design air-to-cloth ratio (A/C) was 2.0 acfm/ft2 (0.010
m/s) with all compartments on line and 2.16 acfm/ft2 (0.011 m/s) with one
compartment out for cleaning, at a design gas flow of 315,000 acfm (149 m3/s).
Actual gas flows measured during the testing were significantly lower, with an
average of 238,000 acfm (112 n»3/s), resulting in A/C values of 1.51 acfm/ft2
(0.0077 m/s), and 1.63 acfm/ft2 (0.0083 m/s) with all compartments on line and
one compartment out, respectively.  The discrepancy between the design and
measured gas flows is a direct result of the conservative baghouse design.

     The baghouse is divided into 14 compartments with 236 bags in each
compartment.  The bags are 8 in (20 cm) in diameter and 22 ft (6.7 m) in
height.  A schematic drawing of the baghouse is shown in Figure 1.  The
drawing also shows the locations of the inlet and outlet test ports.   Table
3 gives a summary of the design parameters of the baghouse.
                                      196

-------
                 STACK
              OUTLET
              TEST PORTS (8)
           FLUE GAS FROM
           NO. 3 BOILER
                                          BAGHOUSE
        /
WxMAV
                         \^yr
      INLET TEST PORTS (4)
   ID FANS (2)
                Figure 1.  Schematic Side View of Baghouse
                  TABLE 3.  BAGHOUSE DESIGN PARAMETERS
Supplier
Year On-Line
Design Gas Flowrate
Design A/C Ratio
  (With One Compartment Off-Line)
Cleaning Method
Design Pressure Drop
Fabric
No.  of Compartments
No.  of Bags Per Compartment
Bag  Diameter
Bag  Length
Efficiency Guarantee
Joy-Wes tern
1979
315,000 acfm @ 290°F (149 m3/s @ 143°C)
2.0 acfm/ft2 (0.010 m/s)
2.16 acfm/ft2 (0.011 m/s)
Reverse Air
6" H20 (11.2 mm Hg)
Teflon Coated Fiberglass
14
236
8 in (20 cm)
22 ft (6.7 m)
99.8 Percent
                                 197

-------
     During the initial operation of the baghouse, minor discrepancies were
noted such as thimble leaks and valve seatings that required adjustments.
However, these repairs were expeditiously made and initial performance tests
conducted.

     After approximately two months of operation, the flange to flange pres-
sure drop exceeded design to the point that reductions in unit load were
required.  In an effort to keep the baghouse on line, hand cleaning and beat-
ing of the bags became necessary on a regular basis to remove cake build—up
on the inside of the bags.  Several attempts were made by plant personnel to
correct the pressure drop problem, such as replacing the original woven fiber-
glass teflon coated bag with a teflon B coated bag, and burning different
types of coal.  Attempts were also made to optimize cleaning cycles and model
flow distribution patterns.

     All efforts at optimizing the cleaning cycle, including adjusting
reverse air periods, pressure-initiated cleaning cycles and continuous clean-
ing cycles, and installing the teflon B bags, did not improve the problem
associated with the high flange to flange pressure drop.  Ultimately, the
pressure drop problems were resolved by the installation of sonic cleaning
devices (1).

                           PROCEDURES AND RESULTS
MASS LOADING AND MASS EFFICIENCY

     Inlet and outlet mass loadings were determined by isokinetically collect-
ing suspended fly ash on in situ filters using the equipment and procedures
specified in EPA Method 17 (2).  Although Method 5 is usually required for
compliance testing, Method 17 was deemed to be more appropriate for mass
efficiency determinations.  Method 17 gives a better representation of the
true particulate concentration at actual flue gas conditions because it avoids
the possibility of new particle formation by condensation.  Each sampling run
involved a traverse of the entire duct area.

     The measured mass loadings and efficiencies are given in Table A.  Based
on a total of six runs, the mean inlet and outlet loadings were 3.20 gr/dscf
(7330 mg/m3) and 0.00055 gr/dscf (1.26 mg/m3).  The standard deviations on the
inlet and outlet loadings were 15.7% and 23.6% of the mean values, respective-
ly.  The mean mass efficiency was 99.983%, corresponding to a mean penetration
of 0.017%.  The standard deviation on the penetration was 19.0% of the mean
value.  Thus, at the 90% confidence level, the penetration is in the range
of 0.012 to 0.023%, or the efficiency is in the range of 99.977 to 99.988%.
This corresponds fairly well to the range of measured efficiency values.

     The outlet mass loadings were comparable to those reported by Ensor
et al (3) for the Kramer Station fabric filter, which was operating at an A/C
value of 1.67 acfm/ft2 (0.0085 m/s).  The mass efficiencies were well above the
design value, certainly due in part to the conservative design of the baghouse.
The average emission rate was calculated to be approximately 0.001 lb/10  BTU
(0.43 yg/J).  The stack opacity recorded by the plant visiometer was essential-

                                     198

-------
ly zero during all testing.
during all testing.
Unit No. 3 was running at essentially 100% laod
             TABLE 4.  MEASURED MASS LOADINGS AND EFFICIENCIES
                                                              t

Run
No.
1
2
3
4
5
6
Flue gas
Flowrate
acfm (m /s)
229,000 (108)
231,000 (109)
241,000 (114)
241,000 (114)
242,000 (114)
241,000 (114)
Temp
. Mass Loading
Inlet
°F (°C) grains/dscf
262
271
273
268
264
258
(128)
(133)
(134)
(131)
(129)
(126)
3.
3.
3.
2.
3.
2.
13
35
84
45
59
86
(7170)
(7670)
(8790)
(5610)
(8220)
(6550)
Outlet
(mg/m3)
0.00066
0.00044
0.00058
0.00036
0.00071
0.00055
Baghouse
Efficiency

(1.
(1.
(1.
(0.
(1.
(1.

5)
0)
3)
82)
6)
3)

99
99
99
99
99
99
7,
.979
.987
.985
.985
.980
.981
tObtained by EPA Method 17.
PARTICLE SIZE DISTRIBUTION AND FRACTIONAL EFFICIENCY

     Inlet and outlet particle size distributions were measured using modified
Brink impactors at the inlet and University of Washington Mark III impactors
at the outlet.  Eight real impactor runs and two blank runs with filters
ahead of the impactors were performed in separate sampling ports at the inlet.
At the outlet, three complete traverses of duct area were performed, with no
blank runs.  The blank runs at the inlet did not indicate any appreciable
substrate interference problems with the flue gas.  Glass fiber substrates
were used at the inlet and greased metal foils were used at the outlet.

     The raw impactor data were reduced using an updated version of the
Cascade Impactor Data Reduction System (CIDRS) originally described by
Johnson, et al (4).  The resulting inlet and outlet size distributions are
given in Table 5 on the basis of cumulative percentages of particulate mass
contained in particles smaller than various diameters.  Based upon these
results, it appears that submicron particles account for about 0.84% of the
particulate mass at the Inlet.  At the outlet, about 7.7% of the particles by
mass are smaller than 1 ym.  The mass median diameter (mmd) of the inlet dis-
tribution is beyond the upper particle size limit of the modified Brink
impactors.  However, an extrapolation of the data using an osculating poly-
nomial (5) suggests that the inlet mmd is about 20 ym.  The mmd of the outlet
distribution is about 4.5 ym.

     Inlet and outlet differential mass distributions (AM/AlogD versus D) were
also computed using the same data reduction system.  Fractional penetration
values were computed as the ratio of outlet to inlet values of AM/AlogD at
various particle diameters.  Figure 2 shows the fractional efficiency curve
constructed from these calculations.  The calculated fractional efficiency

                                     199

-------
        TABLE  5.   INLET AND OUTLET PARTICLE SIZE DISTRIBUTIONS
 Particle Diameter
 	(van)	

        0.4
        0.6
        0.8
        1
        2
        4
        6
        8
Cumulative Percentage  of Particulate Mass
   Contained in Particles Smaller Than
         Indicated Diameter  (%)
         Inlet           Outlet
          0.36
          0.48
          0.64
          0.84
          3.0
         12
         18
         23
 2.1
 3.8
 5.7
 7.7
19
43
64
79
•fAverage of eight  inlet runs with Brink impactors and  three  outlet
 runs with University  of Washington Mark III impactors.
                99.99
             o
             \L
             u.
             u
                99.9
                                               TT
                   0.5     1.0      2.0       5.0     10
                              DIAMETER (urn)      4.51-4
                Figure 2.  Fractional Efficiency Curve.
                                  200

-------
decreased steadily from 99.988% at 10 ym to 99.87% at 0.5 ym.  The impactor
data are not reliable below about 0.5 ym.  Despite the apparent sensitivity to
particle size, the baghous'e still attained an estimated cumulative collection
efficiency of 99.92% for all particles smaller than 2 ym.  In absolute terms,
this is well in excess of the fine particle collection efficiencies typically
obtained with most conventional control equipment.

FLY ASH ANALYSIS

     Fly ash from an inlet mass train thimble was ignited and then digested by
standard acid attack for analysis by atomic absorption spectrometr    The
results are given in Table 6.  The results are reported as the we_ ..it percent-
ages of the elements as their oxides, but this is not intended to imply  that
they actually occur in this form.  Based on the relatively low alkali metal
content of this fly ash, the ash would be expected to exhibit an inherently
high electrical resistivity in the absence of any conditioning effects from
flue gas species.
    TABLE  6.   CHEMICAL ANALYSIS  OF FLY ASH OBTAINED BY ..ATOMIC ABSORPTION
    	SPECTROMETRY ON A DIGESTED SAMPLE"'	


                         (weight  percentage as oxide)

                             Li20          0.03
                             Na20          0.46
                             K20            1.3
                             MgO            1.4
                             CaO            5.1
                             Fe203         3.6
                             A1203        25.1
                             Si02         57.4
                             Ti02          1.3
                             P205          1.0
                             S03,          0.68
                             LOI            1.8
                                 Total    99.17
    #The  sample was taken from the thimble of a mass train which isokineti-
     cally  sampled the flue gas at the inlet of the baghouse.
    tLoss on ignition.
                                      201

-------
FLUE GAS COMPOSITION

     The major flue gas components were determined by the standard Orsat
method with water vapor determined gravimetrically by adsorption on Drierite
dessicant.  The average composition of the flue gas was as follows:  73% Na,
4.8% Oa, 13.5% COa, and 8.7% HaO.  The oxygen level was somewhat higher than
expected from the amount of excess air reportedly being used in the boiler,
but this difference could be attributable to air inleakage ahead of the
sampling ports.

     Sulfur oxides were determined by the method of Cheney and Homolya (6).
Sulfuric acid, which was originally present as either SOa or HaSOi*, was
collected by selective condensation in a temperature-controlled tube packed
with a Pyrex glass plug.  Sulfur dioxide was captured by absorption in a
solution of 3% HzOa.  The recovered condensate and the spent peroxide solution
were analyzed for sulfate content by titration.  Two of the eight analyses
were performed by ion chromatography as a check on the titration.  The two
analytical techniques gave results that agreed to within an average percent
difference of 15.8%.

     The results of eight sampling runs yielded a mean SO3 (HaSOO concentra-
tion in the flue gas of 0.29 ppmv with a standard deviation of 0.039 ppmv.
The mean SOz concentration was 391 ppmv with a standard deviation of 11.5 ppmv.
At the concentration levels found in these analyses, the SO3 (HzSOi*) would be
expected -to have very little effect on the electrical resistivity of the ash.

ELECTRICAL RESISTIVITY OF ASH

     Although the collector in this case is a baghouse, the electrical resis-
tivity of this ash may be of interest to utilities considering various types
of collectors for a new plant burning a similar coal.  Electrostatic effects
play an important role in current theories of filtration, so it is also
reasonable to expect that the resistivity of the ash may influence its filtra-
tion characteristics.  Indirect evidence of this effect has recently been
obtained in an EPRI-sponsored study of fabric filtration (7).

     An in situ point-plane probe was used to measure ash resistivity by the
sparkover method.  This procedure has been described in detail elsewhere (8).
A series of nine measurements were made at the inlet sampling location.  The
mean value of sparkover resistivity was 6 x 1011 ohm*cm with a standard
deviation of 4 x 1011 ohm»cm.  The considerable scatter in the data is partly
due to temporal and spatial fluctuations of flue gas temperature and SO3 and
moisture levels.

     The measured resistivity was compared to that predicted by the technique
of Bickelhaupt (9) using the ash composition given in Table 6 and assuming no
conditioning effect of the SOa.  At the mean flue gas temperature during the
in situ resistivity measurements (266°F/130°C) the predicted resistivity was
5 x 1011 ohm*cm at a flue gas moisture level of 8.7%.  Thus, the predicted
value of resistivity agreed with the mean in situ measurement to within 20%
despite the considerable scatter in the in situ data.
                                     202

-------
PRESSURE DROP

     The pressure drop  across the baghouse was monitored throughout the test
program.  However,  these pressure drop data were probably not representative
of typical baghouse operating conditions at Arapahoe since the bags were
cleaned during a plant  outage that occurred about two weeks prior to the test
program.  Although  the  measured pressure drops were well below the design
value of 6 in HaO (11.2 mm Hg) , excessive pressure drop has been a problem in
the past as previously  noted.  During the test, the pressure drop typically
varied from about 3.4 in HaO  (6.4 mm Hg) near the end of the cleaning cycle
up to about 5.0 in H20  (9.3 mm Hg ) near the start of the cleaning cycle.
The average pressure drop was estimated to be about 4.3 in HaO (8.0 mm Hg).

                      DISCUSSION OF FIELD TEST RESULTS
     The low-sulfur Routt County coal burned at Arapahoe produces a low-alkali,
inherently resistive ash.  In the absence of flue gas conditioning, a
relatively large precipitator would be required to attain collection effi-
ciencies comparable to those achieved by the Arapahoe baghouse.  A comparable
collection efficiency was reported by Lodge Cottrell (10) for the Dave Johnston
Unit 1 cold-side precipitator without flue gas conditioning.  The measured
collection efficiency was 99.96% at a specific collection area (SCA) of
661 ft2/1000 acfm (130 m2/m3/s).  The sulfur content of the coal was in the
range of 0.4 to 0.7%.  With flue gas conditioning, the Pleasant Prairie Unit 1
precipitator achieved a measured efficiency of 99.96% (11).  The SCA was 440
ft2/1000 acfm (86.6 m2/m3/s) and the fuel was a low-sulfur (0.37%) Powder
River Basin coal.  The Pleasant Prairie precipitator was equipped with a
sulfur burner that supplied roughly 5 ppmv of SO3 to the flue gas ahead of
the ESP.  These results suggest that comparable efficiencies may be attainable
in relatively large precipitators without conditioning and in somewhat smaller
precipitators with conditioning.  Efficiencies of 99.98% for the baghouse and
99.96% for the ESP correspond to penetrations varying by a factor of two.
At the Arapahoe Station, however, baghouse penetrations also varied by a
factor of two (0.012 - 0.023%) at the 90% confidence level.  Consequently
the baghouse and ESP efficiencies quoted above must be regarded as comparable
within experimental uncertainty.

     Excessive pressure drop, although not documented during this test, has
been a problem at Arapahoe as it has at a number of other baghouse installa-
tions.  The causes of this problem are generally not well understood.   This
is clearly an area in which additional research is required.  The effects of
electrostatic phenomena and the interactions of the filter cake with various
flue gas species need to be investigated.

                 LABORATORY CHARACTERIZATION OF ARAPAHOE ASH
     Baghouse inlet ash samples comprising five size-fractionated cuts were
obtained using the 5-stage series multicyclone sampling/sizing system
developed by SoRI (12).
                                     203

-------
     Laboratory investigation of the multicyclone cuts included:

     • Major element composition

     • Particle size distribution within the cuts

     • Scanning electron microscope investigation (including
       electron microprobe) study of particle size, morphology,
       and individual particle composition

     • Statistical analysis of SEM data to develop inter-element
       correlations and composition-size correlations

     • Auger spectrometry study of individual particles including
       application of secondary ion mass spectrometry (SIMS) and
       electron spectroscopy for chemical analysis (ESCA)

Major element analyses of the first four multicyclone stages were performed
by x-ray fluorescence.  No significant departures from bulk composition were
observed (cf. Table 6 for the latter).

     A quite different approach is provided by scanning electron microscopy/
electron microprobe (SEM).*  SEM analysis can be applied to individual ash
particles.  If sufficient particles are analyzed, the data can be subjected to
on-line statistical analysis.  Also, morphological information obtained by
SEM may have relevance to ash collectability.  E.g. Arapahoe ash in the >6.5 y
range is composed predominantly of typical spherical ash particles solidified
from the molten state.  Submicron Arapahoe ash, however, contains frequent
hollow pleurospheres, particles with surface irregularities, and numerous
non-spherical particles.  The density and the surface interactions of such
particles will differ from those of the larger, smooth spheres.  This in turn
may be important in filter cake formation and cohesion.

     SEM determination of particle size and shape and SEM-electron microprobe,,
analysis of major element composition were done on approximately 50 particles
from each of the five multicyclone stages.  The data obtained from all five
stages were combined to form a population of ^ 240 particles and the data were
analyzed using the SEM's on-line computational system.

     Because of space limitations, we can present here only a few examples of
the detailed composition information and correlation analyses.  Additional
*GFETC employs a JEOL JSM 35 scanning electron microscope.   X-rays are detec-
 ted with a Kevex lithium-drifted silicon detector and the  elemental analysis
 is performed by means of a Tracer Northern NS-880 x-ray analyzer.  (Reference
 to specific brand names and models is done to facilitate understanding and
 neither constitutes nor implies endorsement by the Department of Energy.)

 To eliminate human bias in selection of particles to be measured and analyzed,
 photographs of SEM fields are overlaid with a grid.  Random-number-generated
 grid coordinates are used to select particles for analysis.

                                      204

-------
information may be found in Refs 14 and 15 and in a forthcoming detailed
report of the SoRI-GFETC field test results.

     Figure 3 displays frequency distributions for two constituents important
in moderating ash resistivity:  NazO and 80s.  Frequency distributions of this
type can help one understand the range of particle resistivities presented to
an ESP.

     Figure 4 displays the same elemental concentrations as functions of
particle size.  At least in the case of SO 3, there is clearly an increased
concentration in the 0-5 ym size range.  Because this range includes sizes
for which ESP efficiencies often are poorest, studies of this type may be
useful in diagnosing successes and failures of various conditioning strate-
gies.

     An example of a well-defined inter-element correlation is that shown by
AlaOs versus SiOa, Figure 5.  This is the only example of a functional rela-
tionship which is preserved to a significant degree in data for each individual
multicyclone stage.  In this plot, as in all of the intervariable correlation
plots, each datum represents a specific, individual, identifiable, serial-
numbered ash particle.  By tracking it back through the statistical system,
each particle's concentration of other elements may be recovered.  A major
concentration of data in Figure 5 falls in a region bounded by 50-60 pct.SiOz
and 25-35 pct.AlaOa.  This lies within the range of compositions of dehydrated
illite (16).

     Using x-ray diffraction applied to low temperature, oxygen plasma ashed
coal, species identified include a-quartz, calcite, and a clay, kaolinite.
Possible minor constituents which could not be positively confirmed included
dolomite and anorthite,
     The thermal decomposition of kaolinite (and illite) proceeds through a
sequence of stages.  Above 1095°F mullite and cristobalite are formed (16, 17,
18).  Clearly, mullite is found in some coal ash.  Hulett and Weinberger have
isolated spherical mullite skeletons in Eastern coal ash particles by removing
the glass phases using a 1 pct.H2Fz etch (17).  Stinespring and Stewart have
discussed these transformations (18) and their relation to fly ash surface
chemistry.  X-ray diffraction of Arapahoe multicyclone ash confirms the
presence of mullite and a-quartz, but not cristobalite.  No other species are
observed.

     The SEM/microprobe analysis of individual particles does not distinguish
phases — microprobe intensities contain contributions from surface atoms as
well as atoms within the interior; the combined signal cannot be "unfolded"
to reveal the relative contributions from several phases containing a common
element.  Figure 5 can best be understood as representing a population of ash
particles containing dehydrated aluminosilicates (especially kaolinites or
illites) which have been converted thermally at least in part to mixtures of
mullite-like phases and quartz.  The high SiOa, descending leg of Figure 5,
would correspond to ash particles high in SiOa to the exclusion of AlzOs-rich
phases.  In addition, all particles contain phases with measurable concentra-
tions of some of the other eight "major" elements.


                                     205

-------
                 SODIUM OXIDE
   50-
 5 BO-
 . 20-
 o
 £  10-
                                                40-1
                                                 10-
                                                           SULFUR TRIOXIDE
       0  I   23456789  10
             Na90 Concentration Midpoint
                                               if 0-&1
                                               S.    0123456789  10
                                                            SOj Concentration Midpoint

Figure 3.   Examples  of Frequency Distributions  of  Elemental Concentrations
    5-

    4-

    3-
  o
  CM
  5 2H
      I
                No^O vs SIZE
                                                               S0  vs SIZE
                                                 8-


                                                 6-

                                                T



                                                 2-
        £'*'•?'.''•*•';!»'. i-'L.f-—~r
      0     20    40    60     80    100
                Diameter, \un
                                                                      —i—•	1	1
                                                         20     40    60    80     100
                                                                 Diameter, pm

 Figure  4.   Examples of  Concentration-Particle Size  Correlations
   4O-I
   30-
 ?20H
 <
    OH
                 AI203 vs Si02
                                               ••
                                                 lOH
                                                  6-

                                                  4-

                                                  2-

                                                  0
                                                              S03 vs MqO
            20    40     60    80
                     %SiO2
                                      IOO
O   2   4    6   8   IO   12    14
               % MqO
Figure 5.   Examples  of Inter-element Concentration Correlations
                                        206

-------
     Ultimately it may be possible to employ detailed coal mineral analysis,
in advance of pilot plant test burns, to predict the chemical species in the
fly ash.  This would be useful in planning ash disposal strategy, and possibly
in predicting collector performance.  Meanwhile, it is becoming possible to
employ detailed ash characterization studies to predict aspects of ash pro-
perties beyond bulk resistivity.

     For example: by using SEM's data it is possible to produce not only
binary, but also ternary and higher order inter-element correlations and also
to calculate various concentration-dependent properties of individual parti-
cles.  In another paper in this symposium (19) this approach is used to
calculate frequency distributions of resistivity and resistivity-particle size
correlations.

     In summary:  Arapahoe ash characterization studies performed at GFETC
have resulted in:

     • Particle morphology information probably relevant to fabric
       filter performance.

     • Detailed composition data for individual particles.

     • Frequency distributions of individual elements, data maps
       of composition versus particle size, and inter-element
       concentration correlations.

     • Identification of mineral species in coal and ash.

     • Suggestions that these studies may be used to understand
       and predict ash behavior in particle collection devices
       such as ESPs.

                                ACKNOWLEDGMENTS
      The  authors  gratefully  acknowledge  the  contributions  of  the staff of
 Southern  Research Institute  who  participated in  the  field  test  at Arapahoe.
 The  field test  personnel were  supervised by  Mr.  G. H.  Marchant, Jr.
 Mr.  W.  R.  Dickson was  responsible  for the gas sampling and analysis  and ash
 analysis.   Other  members of  the  test crew included:  Messieurs  W. S. Hall,
 T. F. Hammond,  E. C: Landham,  C. V.  Lindsey, J.  S. O'Neal,  Z. A. Peich,
 D. Sanders, and T. A.  White.

      We also wish to acknowledge the contributions of  GFETC staff who per-
 formed  extensive  microanalyses,  microscopy and data  analyses.   Principal
 contributors were S. A. Benson,  D.  P. McColler,  S. J.  Miller, and D. K. Rindt.

      This work  was funded  by the U.S. Department of  Energy, Grand Forks Energy
 Technology Center, Grand Forks,  North Dakota, under  Contract  No. DE-AC18-
 80FC10225.
                                     207

-------
                                 DISCLOSURE
     The work described in  this paper was not funded by the U.S. Environmental
Protection Agency and  therefore the  contents do not necessarily reflect the
views of the Agency and no .official  endorsement should be inferred.

                                 REFERENCES
  1.  Menard, A.  R.  and Richards, R. M.  The Use of Sonic Horns as an Assist to
     Reverse Air Cleaning of a Fabric Filter Dust Collector.  This Symposium,
     Session A-5.

  2. Code  of Federal Regulations.  Title 40, Part 60, Appendix A.  July 1,
    1978.  pp. 220-237.

  3. Ensor, D, S., Cowen, S., Shendrikar, A., Markowski, G., Woffinden, G.,
    Pearson, R.  and Scheck, R.  Kramer Station Fabric Filter Evaluation.
    EPRI-CS-1669.   Electric Power Research Institute, Palo Alto, California,
    1981.

  4. Johnson, J.  W., Clinard, G. I., Felix, L. G. and McCain, J. D.  A Computer-
    Based Cascade Impactor Data Reduction System.  EPA-600/7-78-042.  U.S.
    Environmental Protection Agency, Research Triangle Park, NC, 1978.  592 pp.

  5. Johnson, J.  W., Pyle, B. E. and Smith, W. B.  Extending Precision in a
    Computer-Based  Cascade Impactor Data Reduction System.  In: Proceedings of
    the Second Symposium on Advances in Particle Sampling and Measurement.
    EPA-600/9-80-004.  U.S. Environmental Protection Agency, Research Triangle
    Park, NC, 1980.  pp. 146-166.

 6. Cheney, J. L. and Homolya, J. B.  Sampling Parameters  for Sulfate Measure-
    ment  and Characterization.  Environ. Sci. Technol. 13:584, 1979.

 7. Felix, L. G., Merritt, R. L. and Smith, W. B.  Particulate Emission and
    Operating Characterization of a Fabric Filter Pilot Plant.  SoRI-EAS-82-
    549.  Southern  Research Institute, Birmingham, Alabama, 1982.

 8. Nichols, G.  B.  Techniques for Measuring Fly Ash Resistivity.  EPA-650/2-
    74-079.  U.S. Environmental Protection Agency, Research Triangle Park, NC,
    1974.

 9. Bickelhaupt, R. E.  A Technique for Predicting Fly Ash Resistivity.   EPA-
    600/7-79-204.   U.S. Environmental Protection Agency, Research Triangle
    Park, NC, 1979.  105 pp.

10. Smock, R. W. Cold-Side Precipitators Performing Well on Western Coal
    Power Plants.   Electric Light and Power.   59(5):21,  1981.
                                      208

-------
                                  REFERENCES         (Cont'd)



11,  Eskra, B. J. and McKinney, B. G.  One Year's Operating Experience with S03
     Conditioning on a Large Coal-Fired Unit's Electrostatic Precipitator.
     Paper presented at the 75th Annual Meeting of the Air Pollution Control
     Association, New Orleans, Louisiana.  June 20-25, 1982.

12.  Smith, W. B. and Wilson, R. R., Jr.  Development and Laboratory Evaluation
     of a Five-Stage Cyclone System.  EPA-600/7-78-008.  U.S. Environmental
     Protection Agency, Research Triangle Park, NC, 1978.

13.  Hart, F. C. and DeLaney, B. T.  The Impact of RCRA (PL 94-580) on Utility
     Solid Wastes.  FP-878.  Electric Power Research Institute, Palo Alto,
     California, 1978.

14.  Sears, D. R.  Particulate Control for Low Rank Coals.  In; DOE/GFETC/
     QTR-81/3-4.  U.S. Department of Energy, Grand Forks, North Dakota, 1982.

15.  Sears, D. R.  Hydrocarbon and Tracte Element Emissions from Combustion.
     In; DOE/GFETC/QTR-81/3-4.  U.S. Department of Energy, Grand Forks, North
     Dakota, 1982.

16.  Deer, W. A., Howie, R. A. and Zussman, J.  Rock Forming Minerals.
     Vol. 3 Sheet Silicates.  Longman, London, 1976.

17.  Hulett, L. D. and Weinberger, A. J.  Some Etching Studies of the Micro-
     structure and Composition of Large Aluminosilicate Particles in Fly Ash
     from Coal Burning Power Plants.  Environ. Sci. Techno1. 14:965, 1980.

18.  Stinespring, C. D. and Stewart, G. W.  The Surface Chemistry of Alumino-
     silicate Particles—Application to Combustion Stream Chemistry.  METC/RI-
     79/7.  Morgantown Energy Technology Center Report.  Morgantown, WV,  1979.

19.  Sears, D. R. , Benson, S. A., McCollor, D. P. and Miller, S. J.  Fly Ash
     from Texas Lignite and Western Subbituminous Coal: A Comparative Character-
     ization.  This Symposium, Session C-4.
                                      209

-------
       AN  EVALUATION OF FULL-SCALE FABRIC FILTERS ON UTILITY BOILERS

         by:  John W. Richardson, John D. McKenna, John C. Mycock
              ETS, Inc.
              Roanoke, Virginia  24018
                                ABSTRACT

     The objective of this EPA sponsored program was to determine the
particulate emission concentrations of a coal-fired electric utility of
greater than 100 MW output.  Tests were also conducted to determine
gaseous constituent concentrations of the flue gas and particle size
via a cascade impactor.

     Testing was conducted at Southwestern Public Service's Harrington
Station, Unit 3, between July 6, 1981, and July 11, 1981.  A total of
three (3) outlet tests and one (1) inlet test were performed.  Due to the
absence of inlet test ports, inlet testing was done by bypassing the bag-
house and testing at the outlet ports of the stack.

     Emissions for Unit 3 are controlled by two (2) baghouse systems, an
east and a west, each with its own operating control system and bypass
dampers for start-ups, emergency operation and shutdown.  Each system
incorporates shake/deflate cleaning, and consists of 32 compartments with
204 bags per compartment for a total of 6,528 bags.

     Average outlet concentration resulted in a lower than expected value,
of 0.007 Ibs./lO  Btu.*  The loading of the inlet testing was 2.0 Ibs./lO
Btu giving a 99.65% collection efficiency for the baghouse.  This emission
rate is,significantly lower than the existing Federal standard of 0.03
Ibs./lO  Btu.  Particle sizing tests indicated that the mass geometric
mean diameter for outlet tests 1 through 3 ranged from 7.5 to 13 /urn with
an extrapolated inlet mass diameter of 60 /urn, and a baghouse collection
efficiency of 99.86%.

     This paper has been reviewed in accordance with the U. S. Environ-
mental Protection Agency's peer and administrative review policies and
approved for presentation and publication.	
~C*0 Readers more familiar with the metric system may use the conversion
    factors at the end of this paper.

                                   210

-------
                            I.  INTRODUCTION

     This test series of Southwestern Public Service, Harrington Station
Unit 3, was carried out as part of EFA Contract Number 68-02-3649, titled:
An Evaluation Of Full-Scale Fabric Filters On Utility Boilers.

     The purpose of this project is to evaluate and characterize the per-
formance of three (3) full-scale fabric filter units controlling 100 MW
or larger coal-fired power plants.  Efforts are being made to select sites
that are representative of different boiler and coal types, with control
systems characteristic of modern fabric filters.

     Three outlet particulate emission tests were conducted at SPS's
Harrington Unit 3 during the week of July 6-11, 1981,to determine parti-
culate emissions in Ibs./lO  BTU.

     Baghouse and control room data, along with basic baghouse maintenance
and operational data, were also obtained.

     Test protocol followed that specified in the Federal Register.  Dia-
grams of the stack and sampling ports'may be found in Figures 1 and 2.

     A total of four (4) outlet particulate tests and one inlet particu-
late test were performed.  Run 1 was voided because an incorrect stack
moisture content was assumed which resulted in the wrong size sampling
nozzle being used.  Runs 2, 3, and 5 (outlet tests) are considered valid
tests.  Run 4 was the inlet particulate test which was conducted by by-
passing the baghouse and testing at the stack outlet ports.  A summary of
these test data may be found in Table 1.
            II.  DESCRIPTION OF PROCESS AND CONTROL DEVICES

     SPS's Harrington Station Unit 3 consists of a tangentially fired,
Combustion Engineering steam generator, capable of producing 2,700,000
pounds of steam per hour at 2500 psig, 1005 F superheat, and 1005 F
reheat.

     Pulverized Western coal averaging 8,475 Btu/lb., 0.3% sulfur and
5.5% ash is burned.

     Particulate emissions are controlled by Wheelabrator-Frye, Inc. bag-
houses designed to operate at a flue gas flow of 1,650,000 ACFM at 313°F,
with a minimum design efficiency of 98.6%.

     There are two (2) baghouse systems at Harrington Unit 3, an east and
a west system; each with its own operating control system and bypass dam-
pers for start-up, emergency operation and shutdown.  A schematic of the
boiler and baghouse configuration may be found in Figure 3.
                                  211

-------
                                                              300'
      6" Mam in 1 SanpHim Torts
      Continuous Monitor ing. Port j
          4  @90U
220' 8"
14'  7" Inlet: Brccchins
                                                               Inlet Breeching
                                       Figure  1
                        Harrington Unit 3 Stack Configuration
                                          212

-------
Tr.nvorsc Pt.
1
2
3
Distance in Inches
56.3
81.8
119.2
            Figure 2




Traverse Points Harrington Unit 3
              213

-------
                                                      DATA SUMMARY TABLE
                                        Southwestern Public Service,  Harrington, Unit 3
                                                     OUTLET EMISSION DATA
July 6-11,  1981
Run
&
Date


Run 1

7/8/81

Run 2
7/9/81


Run 3
7/9/81


Run 5
7/10/81

Particulate
Emissions
Outlet
lbs/10 BTl!


	 VOID 	


0.009


0.009


0.003

Particulate
Emissions
Outlet
Ibs/hr





35


34.6


11.7

Particulate
Emissions
Outlet
fcr/ACF





0.0024


0.0022


0.0008

MW
Production
7. Opacity






341
5.5


350
5.5


349
5.6

Flow
Outlet
ACFM
DSCFM





1,710,000
944,000


1,740,000
912,000


1,710,000
938,000

Outlet
Temp.
°F






323


323


323

Or sat
% CO 2
Inlet
Outlet




In:
a
Out:
11.6
In:
Out:
11.6
In:
Out:
11.6
Orsat
% o2
Inlet
Outlet




In:
Out:
6.5
Tn:
Out:
6.5
In:
Out:
6.5
Orsat
% CO
Inlet
Outlet




In:
Out:
0.0
In:
Out:
0.0
In:
Out:
0.0
Baghouse
Differential
Pressure
in. H00
2



East BH:
6.98
West BH:
6.26
East BH:
8.1
West BH:
6.7
East BH:
7.1
West BH:
6.6
Stack
Gas
Moisture, 7
In:
Out:



In:
Out:
10.6
In:
Out:
14.3
In:
Out:
10.2
                                                  INLET  EMISSION  DATA
Run
&
Date



Run 4
7/9/81

a Not t
Particulate
E-nissions
Inlet
lbs/10 ETTU


2.0

ested
Particulate
Emissions
Inlet
Ibs/hr


7960


Particulate
Emissions
Inlet
gr/ACF


0.517


Mr/
Production
7. Opacity



349
98

Flow
Inlet
ACFM
DSCFM


1,780,00
947,00

Inlet
Temp.

-------
                                       Stack
West Boghouse
^ C / •s=     East Baghouse
                                                          North
                                                          o
                Ilarrinj;Lon No.  3 Flue  Gas Flow

                         Figure 3
                               215

-------
     This shake/deflate cleaning system consists of 32 compartments with
204 bags per compartment for a total of 6,528 bags; each bag is 11.5
inches in diameter and 30 feet, six inches long.  Bag spacing is 14.0
inches center-to-center, with a bag reach of two.  Design air-to-cloth
ratio is 2.81:1 gross, 2.90:1 with one compartment down and 3.0:1 with
two compartments down.  Total cloth area is 6.0 X 10  ft.^  The baghouse
is outfitted with 50% 14 ounce fiberglass acid resistant coating Criswell
4491625 bags and 50% Criswell 442 570C-2 10 ounce Teflon coated fiberglass
bags.  All bags are of caps and eye bolts design.  Three and one-half
inches of fiberglass insulation covers the baghouse exterior.

TEST PORT LOCATIONS

     Sampling was conducted at four (4) ports spaced equi-distance around
a 20.8 ft. diameter stack.  Three traverse points were assigned to each of
the four (4) ports , resulting in a twelve (12) point traverse particulate
test.  A diagram of the stack traverse points may be found in Figure 2.

TESTING METHODOLOGY

     Particulate emission tests were conducted according to U.S. EPA Re-
ference Method 5 procedures in conjunction with Methods 1, 2, 3 and 4.
Each test included a 12-point traverse with a ten (10) minute sampling
duration for each point.

     Assembly and use of the imp actor train followed stater-of-the— art
protocol and general Method 5 sampling train procedures.'    Special pre-
caution was taken to avoid rough handling of loaded impactors, overloading
of the impactor and in the performance of hot leak tests.
     Test 4 was an inlet particulate test conducted by bypassing the
house and directing the particulate directly to the outlet stack.  Test 4
incorporated EPA Method 17 using a Norton AN-899 5.0 urn retention Alundum
thimble .

SAMPLING EQUIPMENT AND PROCEDURES

Particulate

     The particulate sampling equipment used is referred to as the "EPA
Particulate Sampling Train", designed and developed by the EPA.  '  The
"train" used in this testing was assembled by.ETS, Inc. in accordance with
construction details published by the EPA.      A schematic of the samp-
ling train is shown in Figure 4.

     The EPA sampling train consists of a stainless steel sampling nozzle,
a Method 5 filter holder containing an 87 millimeter Schleicher and
Scherell #1 HV High Purity Glass Filter, a series of four (4) Greenburg-
Smith impingers, a check valve, a leakless vacuum pump, a dry gas meter,
                                  216

-------
                Temperature  Sensor
    1.8 to 2.5 cm
   (0.75 to 1 In.")

       i -    ^       I  Probe
                                   Impinger Train Optional, May  Be  Replaced
                                         By An Equivalent  Condenser
1.8 cm (0.75 In.)  f
           Pitot Tube
                  Probe
 Temperature
— Sensor
Heated
 Area
Therr.oneter
 Filter Holder
                                                            Thermoizeter
  N)
                                 Stack Wall
                          reverse-ivpe
                           Pitot Tube
                      I
                                                                                                  Check Valve
                                                                                                 Vacuum Line
                                             Pitot
                                           Manometer
                                                           Ir.pingers    By-Pass
                                                            Ice Bath
                                  Orifice
                                                                                    Main     Vacuum Gauge
                                                                                    Valve
                                           Thermometers
                                  Dry Gas
                                   Meter
             Air-Tight
               Pump
                                                   Method 5 Sampling Train
                                                          Figure  4

-------
and a calibrated orifice.  The impingers and connecting tubes are made
of Pyrex glass and are connected with glass ball and socket joints.  The
probe is made of type 316 stainless steel.

     Using the Type "S" pitot tube, a velocity traverse is performed along
all traverse axis during each particulate run.  The velocity pressure at
each sampling point is measured by means of an inclined manometer.  A
thermocouple is also attached to the pitot tube to measure gas stream
temperature.  Thermal potential is measured with an Omega portable poten-
tiometer.

     Prior to, and at the conclusion of, each run the complete sampling
train including probe and nozzle is leak tested by plugging the nozzle
with a rubber stopper and then applying a vacuum of 15 inches of mercury
to the system.  The maximum allowable leakage rate of 0.02 cfm was not
exceeded in any of the leak tests.

     During testing, isokinetic sampling conditions are achieved by main-
taining the velocity of the stack gas entering the sampling nozzle equi-
valent to the stack gas velocity at each sampling point.  The required
velocity of the stack gas entering the sampling nozzle is obtained by
adjusting the pressure drop across the calibrated orifice.  The pressure
differential across the calibrated orifice is measured by an inclined/
vertical manometer with an inclined range of 0-1.0 inches of water and a
vertical range from 1.0 to 10.0 inches of water.

     At the completion of each test, the sampling nozzle,inside of the
probe, inside of the thimble holder and front-half of the glass fiber
filter holder  were washed with acetone.  The washings were collected in
separate storage containers.  The Alundum thimble and the glass fiber
filter were removed from their respective fiber holders and stored in
separate storage containers.  The contents of the first three impingers
were measured and discarded.  The silica gel from the fourth impinger was
transferred to a separate storage container.  All storage containers were
returned to the laboratory for analyses.

     Tests to determine carbon dioxide, oxygen and carbon monoxide were
conducted via an Orsat analysis according to EPA Method 3.  A calibrated
cylinder was run as a blank.

                      III.  DISCUSSION OF RESULTS

     During this test series, there were no deviations from normal opera-
ting conditions for Unit 3 that could be determined from the control room
or baghouse control room data or conferences with the boiler operators.
Control room data monitored during the test period compared closely with
previous data.

     Strip charts monitoring ash discharge, 862* 02 and NO  concentrations,
percent opacity, baghouse outlet temperature and pressure drop were col-
lected and the data were recorded.  Flue gas volumes for the three (3)
                                  218

-------
outlet and one inlet particulate tests were within an expected and con-
sistent range.

      Coal and ash samples were taken for analysis.  Approximately five
pound integrated coal samples were taken during each test,  of which appro-
ximately one-half pound was analyzed for ash, sulfur, moisture and Btu
content (proximate fuel analysis).  Coal parameter concentrations remained
within the expected and design range.

      Outlet particulate emissions at Harrington Unit 3 averaged 0.007
Ibs./lO  Btu.  Emission rates were also expressed in Ibs./hr. and aver-
aged 27.39 Ibs./hr. for the three (3) outlet tests.  The single inlet
emission test performed by bypassing the baghouse resulted  in an emission
rate of 2.0 lbs/10  Btu or 7958.5 Ibs./hr.  Emission rates  for outlet
runs 2 and 3 were very close at 0.009 lbs./106 Btu (35.9 Ibs./hr.), and
0.009 lbs./106 Btu (34.6 Ibs./hr.), respectively, while outlet run 5
resulted in an unexpectedly low emission rate of 0.003 Ibs./lO  Btu or
11.7 Ibs./hr.  The inlet run performed at the outlet stack ports also
resulted in a lower emission rate than expected when compared to previous
test results conducted at SPS Harrington Unit 3.

      Cascade impactor sampling was performed by Research Triangle Insti-
tute, July 8 and 9, 1981.  A total of three  (3) outlet and one (1) inlet
impactor tests were performed.  Figure 5 summarizes the cumulative mass
size distribution results for the three (3) outlet impactor tests.

      All outlet impactor tests including a blank had some stages with
weight losses in spite of extended sampling times.  The weight losses on
the substrates were believed to be due to chemical reactions of the flue
gas with the grease to form volatile materials, although the Apiezon-H
grease has performed satisfactorily at other test sites.  All of the out-
let size distribution data were adjusted using the blank test substrate
weights.  The mass geometric mean diameter for outlet tests 1 through 3
ranged from 7.5 to 13 Mm with an extrapolated inlet mass diameter of 60 /urn,
and a baghouse collection efficiency of 99.86%.

      Southwestern Public Service has expressed concern that the results
obtained during this series of tests do not represent typical operation
of the Harrington Unit 3.   The following paragraph is taken from a letter
sent to EPA.

      "SPS questions the validity of the data because there is substantial
disagreement with previous testing by SPS and GCA on Harrington Unit 2
and by SPS on Unit 3.  Additionally, results of the inlet grain loading
test'(Run 4) by ETS are lower than the theoretical values by 50%.
Finally, it seems unreasonable that the lowest particulate loadings
measured came the day after the bypasses had been opened the night before."

      ETS agreed to include this paragraph in the paper to  make the SPS
reservations known, however,  we feel that the emission levels measured are
representative of the baghouse performance at the time of the test.  Inlet
                                   219

-------
u
    99.9&1

     99.9-
     99.8-

       99-
       98-

       95-
       90-
 70-
 60-
 50-
 40-
 30-
 20-

 10-
  5-

  2-
  1-
 .5-
 .2-
 .1-
.05-
      .01-
         .1
                              Run No.  2 on July 8, 1981;
                              3 hrs.

                           A  Run No.  3 on July 9, 1981;
                              8 hrs.

                           x  Run No.  5 on July 10, 1981;
                              13 hrs 51 min.
TIIIII    i   ITT i 1111    i   iiii
     1              10

 AERODYNAMIC DIAMETER  (pm)
                                               122
       Figure 5   Cumulative mass size distribution for
                  three outlet impactor samples.
                          220

-------
testing may be suspect since the inlet sample was obtained at the stack
outlet sampling port's after bypassing the baghouse.  However, as can be
seen in Table 2, the results of the outlet tests compare with emission
tests conducted at other coal-fired utilities.  This conclusion is also
supported by facts such as the 15 hour sampling time required to obtain a
representative impactor sample, a visual opacity reading between 0 and 5
percent during all the outlet tests, and the presence of an RTI quality
assurance specialist during the entire testing program.

      Bag analyses were performed at ETS* laboratory in Roanoke, Virginia.
A battery of tests were performed including permeability, tensile strength,
MIT flex, and Mullen Burst.  A total of four (4) bags (2 used and 2 new)
were tested.

      Overall, the bag analyses conducted by ETS, Inc. indicate a greater
percentage loss of strength in terms of MIT flex, Mullen Burst and tensile
strength than previous tests performed by another firm.  Permeability as
received is essentially the same when comparing ETS's and the other firms'
results.  The cleaning procedures differ between the two laboratories but,
in general, indicate approximately the same permeability improvement after
cleaning.  A more detailed comparison of the bag analyses may be found in
Table 3.

      It should be noted that ETS used a .04 in. jaw and four (4) pound
weight and the other firm used a .03 in. jaw and four (4)'pound weight
when performing the MIT flex test.  Another possible contributing factor
to the differences is the other firms bag testing was done in April 1981,
and the ETS bag analyses were performed in September 1981; thus, the bags
were on stream longer before the ETS tests were conducted.

      Laboratory analytical procedures incorporated were those mandated by
the Federal Register Part II, Environmental Protection Agency, Revision to
Reference Methods 1-8.
Second Test Location

      Particulate emission tests were also conducted at Pennsylvania Power
and Light's Brunner Island  Station Unit 1 August 10-14, 1981 and August 30-
September 4, 1982.

      Unit 1 consists of a 345 MW Westinghouse turbine generator unit sup-
plied by a continuous operating Combustion Engineering type CL pulverized
coal-fired boiler; with a maximum rating of 2,100,000 Ibs. of steam/hr. at
1005°F and 2750 PSIG at the superheater outlet and reheat rating of
1,947,000 Ibs. steam/hr. at 1005°F and 600 PSIG.

      Particulate control is handled by a Carborundum baghouse with a
designed gas flow of 1,200,000 ACFM at 330°F.

      The average particulate emission rate for five (5) outlet tests


                                    221

-------
                           TABLE 2

  MASS PARTICULATE EMISSIONS OF VARIOUS COAL-FIRED POWER PLANTS



     SITE                                 EMISSIONS.gr/ACF


1.  SPS - Harrington Unit 3                0.002


2.  Colorado Ute, Nucla Station            0.003
3.  Northern States Power
    Riverside Station                      0.003 - 0.006
4.  Nebraska Public Power
    Kramer Station                         0.02
5.  Pennsylvania Power & Light
    Sunbury Station                        0.005 - 0.006
6.  Pennsylvania Power & Light
    Brunner Island                         0.0085
7.  Northern States Power
    Elk River                              0.004
8.  SPS - Harrington Unit 2                0.017
9.  Colorado Ute
    Bullock Station                        0.012
                             222

-------
                                   TABLE 3  - FABRIC TESTING RESULTS COMPARISON
Bag Typo
Acid
Resistant (625)
Used
Tet '.on - 3
L'sed
Acid
Resistant (625)
New
Teflon - 3
New
Fabric Tests
Flex (Fill)a
Top - 48
Middle - 55
Bottom - 68
Top - 58
Middle - 57
Bottom - 71
Top - 59
Middle - 50
Bottom - 49
Top - 84
Middle - 88
Bottom - 89
Fill - 2,500
Warp - 12,500
Fill - 3,013
Warp - 10,400
Fill - 500
Warp - 15,000
Fill - 629
Warp - 9,946
Mullen Burst a Tensile a
13 - 297, loss
32% loss
2 - 20% loss
247. loss
700 PSI
619 PSI
525 PSI
343 PSI
Fill - 0-207,
Warp - 1-177.
Fill - 437, ;
Warp - 317.
Fill - 0-167.'
Warp - 0-7%
Fill - 18%
Warp - 277.
Fill - 300
Warp - 600
Fill - 252
Warp - 465
Fill - 130
Warp - 250
Fill - 87
Warp - 228
Permeabi 1 ii:y a
As received - 1.48-2.75
After wash - 48
As received - 1.2-2.1
After vacuum - 60.4
As received - 2.31-2.75
After wash - 55
As received - 1 .8-2.0
^After wash - 46.4




Testing Co.
Other Lab
ETS, Inc.
Other Lab
ETS, Inc.
Other Lab
ETS, Inc.
Other Lab
ETS, Inc.
ro
ro
OJ
        MIT Flex
        Mullen Burst
        Tensile
        Permeability
Measurement Parameters
      cycles
      Ibs/sq. inch
      Ibs/inch
      CFM/sq. ft.
Other Lab - .03 in. jaws @ 4 Ibs.
ETS, Ince - ,QA in. jaws @ 4 Ibs.

-------
                                    u
performed in 1981 was 0.0364 Ibs./lO  Btu.  Inlet emissions averaged
13.08 Ibs./lO  Btu.

     A total of six (6) cascade impactor tests were performed, two (2)
inlet, two (2) outlet, and two (2) blank runs.  Mean extrapolated particle
size was 54.3 microns at the inlet and 8.72 microns at the outlet.  These
impactor tests indicated that the baghouse collected more than 99.9% of
the particulate matter in the flue gas.

     This fabric filter characterization program continues and efforts con-
tinue to secure a third site for testing.
                            METRIC EQUIVALENTS
     Readers more familiar with metric units may use the following equiva-
lents to convert to that system:
              Nonmetric

              Btu
              in. -
              in.
              °F
              ft
              Ib
              micron
              oz
Times

1055
   2.54
   6.45
5/9(°F-32)
   0.30
   0.09
  28.32
   0.06
   0.45
   1.00
  28.35
Yields Metric

     J
     cm-
     cm
     m
     liter
     g
     kg
     Jim
     g
                                   224

-------
                               REFERENCES
1.  Harris, D. Bruce.  "Procedures For Cascade Impactor Calibration and
    Operation in Process Streams".  EPA 600/2-77-004, (NTIS PB263623),
    January 1977.

2.  "Code of Federal Regulations".  Method 17, Title 40, Part 60,
    Appendix A.

3.  Martin, Robert M.  "Construction Details of Isokinetic Source
    Sampling Equipment".  EPA report APTD-0581 (NTIS PB203060),  April
    1971.
                                   225

-------
                      Status of SPS Investigation of
              Harrington Station Unit 2 Fabric Filter System
                             Richard Chambers
                    Southwestern Public Service Company
                              P. 0. Box 1261
                          Amarillo, Texas  79170
                                Dale Harmon
               Industrial Environmental Research Laboratory
                      Environmental Protection Agency
               Research Triangle Park, North Carolina  27711
                                 ABSTRACT

     This paper describes activities during  the  fourth  year  of  the SPS/EPA
fabric filtration  study  at Harrington  Station.   The scope of  the overall
study and its purposes were outlined in earlier papers.

     During  the  fourth year  of this  project,  sufficient information  was
available  to make  certain  comparisons between  Harrington Station's  high
air-to-cloth ratio  (3.4 ft/min)  shake/deflate  fabric filter  system and low
air-to-cloth  reverse  air  units.   Specifically,  comparison  of  bag  life
economics  and  pressure drop performance  is  addressed.   Sizing of reverse
air and shake/deflate baghouses  is  discussed in  light of current  operating
data from Harrington Station and other installations.

     Operation and  maintenance  costs are shown  for the Harrington  Unit  2
fabric filter system, and suggestions for improved shaker design are given.
Plans to investigate sonic augmentation of bag cleaning  are described.

     This paper has been reviewed in accordance with the U.S.  Environmental
Protection Agency's  peer  and administrative review policies and  approved
for presentation and publication.
                              INTRODUCTION

     Electric  utility  experience  with fabric  filtration for  particulate
control of coal-fired boiler emissions has  increased  dramatically  over the
last 5  years.   Southwestern Public  Service  (SPS)  established itself  as  a
pioneer in utility  fabric filtration with  the construction of  two  shake/
deflate baghouses in 1978 and 1980.

     Since  1977,  SPS,   under  contract  to  the  Environmental  Protection
Agency, has been collecting data at  Harrington Station,  Unit  2;  therefore,
a great deal of information is available on  the  Unit  2  system.   Results of
this study have  become  increasingly important as other  utilities, seeking
solutions to  emission control problems,  look  for  design and  performance
specifications based on  actual operating experience.


                                     226

-------
     This paper  presents  information SPS has  accumulated  that will be  of
value  to  utilities  evaluating the applicability of  shake/deflate  cleaning
for  a  specific installation.  Comparisons  are made between  shake/deflate
and  reverse  air  fabric  filters in the areas of pressure drop,  sizing,  bag
life and fabric replacement economics, and operation and maintenance costs.

     The  EPA study  at Harrington Station is now  into its fifth and  final
year.  The results  of this program are providing much needed  knowledge  of
the  effectiveness of  a  shake/deflate  fabric filter  system  used in  associa-
tion with low sulfur western coal.  A final report will be  published at the
conclusion of the study.
                                BAG LIFE

     There is very limited data available to the utility industry regarding
the bag life of various fabrics in reverse air and shake/deflate collectors
because few  collectors have been  on  line long enough  to  establish fabric
life,  especially  on large  units.   According  to  a  1981  EPRI  report  (1),
"Preliminary results from reporting utilities  indicate  bag lives in excess
of two  years,  with several units  exceeding  three years."  As  more exper-
ience is gained, many of these collectors will undoubtedly achieve bag life
in the  3  to  6 year range, but  as  yet 'no reliable  fabric  life  predictions
are available for reverse air units.

     Harrington Unit 2, however, has  been on line since 1978 and bag life
data  is available from  this shake/deflate  unit.  Full compartment  tests
were  done  on a variety of  fabrics to  determine  both  their  pressure drop
performance and wear characteristics (2).
                                                                  2
     Of  the  fabrics  tested  at  Harrington,  the  plied, 10-oz/yd  Teflon-
coated materials have shown the best bag life and performance.  These tests
demonstrate that bag life in  the 3.0  to 3.5 year range is obtainable in a
shake/deflate collector employing suitable fabrics.

     Recent  testing  on the  Unit  2 filter  has  shown  the possibility  of
extending  bag  life beyond  the 3  to  3.5 year  range.   Fabric  wear occurs
mostly during the cleaning cycle where the fabric is subject to flexing and
abrasion.  Therefore,  reducing  the  shaking  time and increasing  the filtra-
tion time between cleaning cycles should yield gains in bag life.  Prelimi-
nary testing results have shown that reducing the shake times at Harrington
from  20 to  5  seconds does  not adversely  affect pressure drop,  implying
satisfactory cleaning with just a 5-second shake.  In addition,  preliminary
results of attempts to increase filtering time before  cleaning  look promi-
sing.  With these  changes, it  is anticipated that bag  life on Harrington 2
can be extended to 4 to 4.5 years.

     The Harrington Unit 3 fabric  filter operates at a design air-to-cloth
ratio of 3.0 ft/min, compared to 3.4 ft/min for Unit 2.  Although it is too
early to determine what bag life the unit will eventually experience, after
only 2 years of operation (start-up summer  1980), the  fabric  appears to be
in better  shape than  the Unit  2 fabric after 2 years service.  Most likely
this  is  because the Unit 3  filter is  cleaned  less frequently  due  to its
                                     227

-------
lower pressure drop.  On  this basis it is expected  that  the  lower air-to-
cloth ratio  fabric  filter on Unit  3  will exceed 4  years bag life  on  its
first set of bags.
         FABRIC REPLACEMENT COSTS FOR R/A AND S/D FABRIC FILTERS

     In  comparing  fabric  replacement  costs  for  reverse  air  (R/A)  and
shake/deflate (S/D) collectors the following factors must be kept in mind:

     1.   The rather  substantial price difference  between  ringed  R/A  and
          unringed shaker bags.
     2.   The larger number of  R/A bags compared to  a  higher air-to-cloth
          shaker.
     3.   The higher wear  rate  of shaker bags vs.  reverse  air  bags due to
          cleaning forces.

     To quantify  the interrelationship  between  these  factors  the levelized
fabric replacement  costs vs.  bag  life  are considered  for  a  1100  MW unit
equipped as follows:
                                  2.0   ft/min  with  2  of  32  compartments

                                  2.7   ft/min  with  2  of  32  compartments
1.   R/A  baghouse,   a/c
     out-of-service.
2.   S/D  baghouse,   a/c
     out-of-service.
3.   S/D  baghouse,   a/c  =  3.0  ft/min  with  2  of  32  compartments
     out-of-service.
4.   S/D  baghouse,   a/c  =  3.4  ft/min  with  2  of  32  compartments
     out-of-service.
For this analysis:
          Bag installation cost
          Bag dimensions
          R/A bag cost
          S/D bag cost
          Inflation
          Cost of capital
                                   $8.30/bag
                                    33.5 x 1 ft
                                   $61.60/bag (7 rings)
                                   $48.50/bag (no rings)
                                    5%.
                                   11%
The breakdown of replacement costs is shown in Table I.

                         TABLE I.  FABRIC REPLACEMENT COSTS *

A/C
(ft/min)
2.0
2.7
3.0
3.4


Cleaning
R/A
S/D
S/D
S/D

Fabric
Cost ($)
1,345,000
784,700
706,200
623,100
Instal-
lation
Cost ($)
181,300
134,300
120,900
106,600
Total
Replacement
Cost ($)
1,526,300
919,000
827,100
729,700
* at 1100 MW
The results of this analysis are shown in Figure 1.
                                    228

-------
     The most striking, result  is  the  very rapid initial decline of all the
curves.  In the 1 to 5 year range of bag life, dramatic changes in replace-
ment  cost  occur from  year  to year.   These  curves strongly point  out the
magnitude of the savings to be gained by optimizing fabric life.

     The highest curve in Figure  1  is that for the reverse air unit.   This
happens because the reverse air filter has the larger number of bags  (hav-
ing the lowest air-to-cloth ratio) and the highest priced fabric.  Although
it  is difficult  to  compare  directly,  reverse  air  cleaning  is  generally
expected to be  less damaging  to  the fabric and gives better bag life.  The
purpose of these curves is to indicate how much longer bag life needs  to be
to obtain the same fabric replacement cost as the shake/deflate units.  The
smallest of the shake/deflate units is the one  with  an air-to-cloth  ratio
of  3.4  ft/min.   The  pressure  drop  and  resulting  fan  cost  at   this
air-to-cloth  ratio  is estimated  to  be  equivalent  to  the  reverse  air
collector at a 2.0 ft/min air-to-cloth ratio (Figures 2 and 3).  Since this
collector  has  the  same  air-to-cloth ratio   as  Harrington,  it could  be
assumed  that  bag  life would be at  least  in  the  3  to  3.5 year  range
(comparable to  Harrington).   As  shown„ in the  lower portion  of  Figure 1, a
3.0-year bag  life  on  this unit is  roughly equivalent  to  5.5 years life on
the reverse air system.

     Bag life should  increase  on the shake/deflate  units  as  the air-to-
cloth  ratio decreases  since less cleaning will be required  to  maintain a
given pressure drop.  Comparing a 4-year bag life on the 3.0 ft/min air-to-
cloth  ratio  shake/deflate  collector with  a  reverse  air unit  gives  a
6.3-year bag life to achieve equivalent fabric replacement costs.

     The analysis  discussed above makes  a number of assumptions  that may
not  actually  be  realized  in practice.   The  intent,  however,  is  not  a
definitive  analysis,  but  to  illustrate the  interplay  of  the  factors
involved for  a  fairly reasonable set of assumptions.  The  one  point that
can  be drawn from  the analysis  is that  the  shake/deflate design can be
competitive with  reverse  air  designs in  terms  of  fabric replacement  costs
under certain conditions.
                  PRESSURE DROP AT HARRINGTON STATION

     Figure  4  shows  the pressure  drop history  of the  Harrington  Unt  2
baghouse  (beginning  after  the first rebagging of  the  unit.)   The pressure
drops shown  are  flange-to-flange pressure  drops,  calculated at  full load
(a/c=3.4 ft/min) from operational data taken for each month.  (Note:  Pres-
sure drops in Figure 2 are lower since they are tube-sheet pressure drops.)
As Figure 4  shows, much  variation  in pressure drop occurs over an extended
period of time due to variations in operating and filtration conditions.

     The  full-load  flange-to-flange  pressure drop  has  averaged 7.87 in.
w.g. over the past 38 months  of operation.  At a  reduced  load (318 MW) and
an air-to-cloth  ratio of 3.0  ft/min,  the pressure drop  averaged 6.85 in.
w.g. for this same period.

                                     229

-------
Point
X
O
A
D
ft./mln
2.0
2.7
3.0
3.4
Type
R/A
S/D
S/D
S/D
                                            (5.6)
                                                  6
Figure 1,
            Boglife.yr
Levelized fabric  replacement cost  (1100 mw  unit)
                              230

-------
 CT
 *
 C
 •
0.
<3
UJ
UJ
X
co
UJ
03
UJ
e>
cr
UJ
25
24

23

22

2!

20
19

I 8

17

16

15

14

13

12

I I

10

9

8

7

6

5

4

3
2

 1 I—
      0
      ARAPAHOE  PILOT  UNIT
      AREAL LOADING PER
      FILTERING CYCLE  (FFPP)
             0.25 LB/FT2

             O.I5  LB/FT2 > REVERSE AIR
             0.25 LB/FT
          O SHAKE

      UTILITY FABRIC FILTERS
          O ARAPAHOE No. 3
          A CAMEO No. 2
          A CHEROKEE No. 3
          D ECOLAIRE
          • HARRINGTON (SHAKE)
          D KRAMER
          V MARTIN DRAKE
             NIXON

                   HARRINGTON PILOT R/A
—  ADAPTED(with permission) FROM EPRI.
                                             HARRINGTON PILOT S/D
                        1.0
                                I.7   2.0
                                           2.5
3.O   3.3
                      FILTERING AIR-TO-CLOTH , acfm/ft.
                 Figure 2.  Comparison  of EPRI/Harrington data.
                                     231

-------
   6 -r
   5 '
   4  •
X
o
o
o_
O
O
*  3
e
o

Q.
Z
   2- •
                                                            s/o
   1 • •
                                                  I
                 1.5
2.0        2.5        3.O

     Design A/C  Ratio
3.5
                                                                      4.0
        Figure  3.   Fan horsepower  costs for  1100 raw unit.
                                    232

-------
to
OJ
GJ
                      West Boghouse
                         Bughouse
                 JASONDJFMAMJJASONDJFMAMJJASONDJFMAMJJA

                     1979                     1980                            1981                1982
                             Figure 4.  Monthly  average  full-load pressure drop.

-------
     Curve fitting the pressure drops versus air-to-cloth ratio over a long
period of time has led to the following correlation for pressure drop:


                          AP = 1.51V1*35                    (1)

where V is the air-to-cloth ratio in ft/min and A P  is  the pressure drop in
in.  w.g.   The  equation  represents both  of  Harrington Station's  fabric
filters (Units 2  and  3)  very well in predicting  the average  pressure drop
at some given air-to-cloth ratio over a long period of time.

     The  pilot  baghouse  (3,  4)  at Harrington  Station  has  been able  to
essentially duplicate the performance  of  the  prototype units  at Harrington
when operated  in  the  shake/deflate mode.  The  pressure drop correlation
developed for the pilot is almost identical to that for the prototype.


                          AP = 1.60V1'35                    (2)
                     PRESSURE DROPS FROM OTHER UNITS

     Care should be exercised when selecting collectors for comparisons of
pressure drop due  to  the  low capacity factor of some units.  A baghouse on
a small boiler that is used for peaking purposes only cannot be expected to
give the pressure  drops that  a  larger base loaded unit would give.  Figure
5 shows the increase  in pressure  drop at  Harrington Station at constant MW
load with respect  to time over a 38-hour period.  This plot illustrates the
effect that extended periods of high  load can have on pressure drop.

     Pressure drop data  for the following comparison were  taken from four
sources:

     1.   Harrington Station.
     2.   EPRI data on utility baghouses.
     3.   EPRI's fabric filter pilot  plant.
     4.   Harrington Station's pilot  unit.

     Figure 2 shows  the  tube sheet pressure drops  and air-to^cloth ratios
of  these  units.   All  of  the shake/deflate  data seem  to  correlate fairly
well as shown in  the  lower right-hand corner of the graph.  The pilot unit
data from Harrington and EPRI agree well with each other and with operating
experience  from many of the  reverse  air prototype units  below an air-to-
cloth  ratio  of  2.0 ft/min.  Above 2.0 ft/min,  the Harrington pilot gives
higher resulting pressure  drops  than  the  EPRI pilot,  the Ecolare unit, and
the Kramer.
                    ANNUAL I.D. FAN HORSEPOWER COSTS

     To  compare  the  I.D.  fan horsepower costs associated with the pressure
drops  shown in  Figure  2, costs  for  a  1100  MW unit  are  again considered
based  on the Harrington pilot work  shown in Figure  2.   Casing losses for

                                    234

-------
   7.5-

-------
reverse air were  approximated  at  2.0 in. w.g. to obtain  the  total flange-
to-flange pressure loss.  In practice the casing  losses vary  between about
1.5  and  3.0  in.  w.g.   Since  data  was available  for the  kilowatt-hours
associated with pressure drop across Harrington Unit  2, these numbers were
used along  with  Harrington's  busbar electrical  cost of   19.17  mills/kWh.
Strictly speaking, this  means  that Harrington's fan  and  fluid  drive effi-
ciencies are  being assumed;  in addition,  a  capacity factor of  0.70  was
assumed.  While these assumptions and the Harrington  pilot projections are
certainly only  approximations  of  the  real  cases,  they  should serve  the
purpose here  of  a "ballpark" comparison of  I.D.  fan horsepower costs  for
shake/deflate and reverse  air units  at various air-to-cloth  ratios.   The
resulting yearly  costs  of  fan horsepower  versus  air-to-cloth ratio  are
shown in Figure 3.

     The 2.7  ft/min  air-to-cloth  shake/deflate collector  shows  equivalent
fan horsepower costs  to reverse air at an  air-to-cloth ratio of  1.6.   As
expected, the 3.0 ft/min air-to-cloth ratio shake/deflate unit  has  higher
pressure drop and is  equivalent to reverse  air in  fan costs  at  an air-to-
cloth  ratio of approximately   1.75  ft/min.   The  3.4 ft/min  air-to-cloth
ratio shake/deflate case has even  higher pressure  losses  and  is equivalent
to reverse  air in fan costs at an air-to-cloth ratio of  approximately 2.0
ft/min.
                     OPERATION AND MAINTENANCE COSTS

     Operation and maintenance costs for Harrington Station's Unit 2 fabric.
filter system for a  3-year  period are shown in Table II  (in 1982 dollars)
in terms of dollars per installed kilowatt per year.
                    TABLE II.   OPERATION AND MAINTENANCE COSTS
                           FOR HARRINGTON STATION UNIT 2
Year
81-82
80-81
79-80
Materials
0.055
0.248
0.570
Maint. Labor
0.167
0.179
0.170
Oper. Labor
0.192
0.215

          Avg.        0.291          0.172          0.204
     The material  costs  shown  include bag purchases in  1979.   The average
O&M cost over this 3-year period is $0.667/kW/yr.

     No  data was available  for  R/A  collectors  to  compare  with  these
numbers.  R/A O&M costs  should vary  from those of a  S/D  collector  mainly
for the following reasons:
                                    236

-------
      1.   Difference in bag replacement costs.
      2.   R/A fan horsepower pressure, drop and maintenance, compared to the
          much smaller deflation fan on S/D collectors.
      3.   No shaker maiiltenance on R/A baghouses.
                      IMPROVEMENTS IN SHAKER DESIGN

     Operating experience over  the last 5 years at  Harrington  Station has
pointed  to  several areas  of improvement  in shake/deflate  design.   These
improvements,  described below,  are aimed  at minimizing  system  cost  and
annual operation and maintenance costs.

     1.   Amplitude of  shake:  The  distance the top  of the bag  is moved
during a shaking cycle  can profoundly  affect bag  life  since the fabric may
actually fold  over  and  abrade against  itself during cleaning.  Maintaining
shake  amplitude  at  around  1 in.  at   reasonable  shake frequencies  could
lengthen baglife considerably.

     2.   Deflation  control   should be established  on flow  rather  than
pressure drop.  The deflation air-to-cloth ratio  employed  at Harrington is
approximately  0.2  ft/min and produces .-a  pressure  drop  in  the  0.1-0.3 in.
w.g. range.  Control with this small pressure drop is difficult at best.

     3.   Multiple  shakers  should  be   used  on larger  compartments  in  an
effort to  keep the inertia of  the shaking  system to  a minimum and reduce
wear on bearings and supports.

     4.   Two-speed  or  variable-speed  shaker motors  might  be of  use  in
minimizing fabric wear.  Under  conditions where  less cleaning is desirable
the shaker motors could be put on low speed  to lessen fabric damage.

     5.   Shaker  tube  bearings  that  are mounted  in the  compartment wall
must  be  carefully designed  for  adequate  wall  stiffness  to avoid  any
movement  that could  damage  the bearings.   In  addition  the  shaker tube
bearings need  to be protected from overheating.
                        SONIC AUGMENTED CLEANING

     A  number of  utilities are  employing sonic  horns to  augment  fabric
cleaning.  Therefore,  EPA  and  SPS decided that a  study of sonic augmenta-
tion was  needed  to determine its effectiveness compared to shaking  and to
look  at the  effects  it has on  fabric life.  After  consulting  with the
Fuller  Co.,  six  sonic horns were purchased and plans were made to suspend
them evenly spaced beneath  the  two  walkways  in one compartment in the Unit
2 baghouse at Harrington Station.

     During testing  the horns will  be activated during the deflation cycle
for approximately 10 seconds.  Both flow and pressure drop will be measured
on the  compartment  and then compared with adjacent compartments.  Testing
will be done  with and without  shakers  to  determine the relative magnitude
of the  cleaning forces being generated by  the  sonic horns.


                                    237

-------
     Fabric samples will be taken from the compartment at certain intervals
and tested to  determine  if  there is any difference  in  fabric  wear between
the two cleaning techniques.
                               CONCLUSIONS

     Southwestern Public Service has gained  a  great deal of knowledge from
the EPA/SPS  fabric  filter  study that has been of  immediate  benefit  in the
proper  operation of  Southwestern's  fabric  filter systems.  The  research
performed under  this program has  allowed  SPS to  identify  the  performance
and wear characteristics of  the commonly used  fabrics and led to identify-
ing a successful fabric for the unit in terms of both bag life and pressure
drop.   Fabric studies  have  revealed  the  mechanism  of  fabric wear  under
shake cleaning  conditions that has been  invaluable in  understanding bag
wear.

     As useful as this  program  has been, fabric  filtration  in  the utility
industry still has  a need for  more fundamental  research.   Ash filtration
properties are  as yet  very  poorly understood  and are  beyond  prediction,
making  accurate  sizing  of fabric  filters  difficult.  Little  research has
been  done  in the  area  of  designing  fabrics   specifically  for  utility
filtration which would  be a very fruitful  area  of  research.   Additional
work  is needed  in  the  fluid mechanical design  of baghouses  to  minimize
casing  losses and ensure good ash distribution.

     As  electric utilities  convert  to  coal,  the challenge of  balancing
environmental  compliance   with   rising   electric  rates   increases  the
significance  of  research  in  the field of fabric  filtration.   Studies such
as  the  one  conducted at Harrington  Station can  best  yield results that
reflect utility standards and regulatory interests.
                               REFERENCES

1.   Operating History and Current Status of.  Fabric Filters in the Utility
     Industry, EPRI Report 1401.  July 1981.

2.   Chambers,  R.,  Ladd, K.,  and Kunka,  S.    SPS  Experience with  Fabric
     Filtration.   Paper  presented  at  Fifth  International Fabric  Filter
     Forum, Phoenix, Arizona, January 1981.

3.   Ladd,  K.,  Hooks,  W.,  Kunka, S.,and  Harmon,  D.   SPS  Pilot  Baghouse
     Operation.   In:  Third  Symposium  on the  Transfer and  Utilization  of
     Particulate Control  Technology.   Volume  I. Control  of Emissions from
     Boilers, EPA-600/9-82-005a, July  1982.   (NTIS  PB83-149583),  pp  55-64,
     July 1982.

4.   Chambers, R., Kunka, S., and Harmon, D.   Update  of SPS Pilot Baghouse
     Operation.  Paper  presented at Fourth  Symposium on the  Transfer and
     Utilization of Particulate Control Technology,  Houston, Texas, October
     1982.
                                    238

-------
                  UPDATE OF SPS PILOT BAGHOUSE OPERATION

                     Richard Chambers and Sherry Kunka

                    Southwestern Public Service Company
                              P. 0. Box 1261
                          Amarillo, Texas  79170
                                Dale Harmon
               Industrial Environmental Research Laboratory
                      Environmental Protection Agency
               Research Triangle Park, North Carolina  27711
                                 ABSTRACT

     An option  of  the 1977 Environmental  Protection  Agency (EPA) contract
with Southwestern Public Service Company (SPS) to assess the performance of
a  large prototype  fabric filter system provided  for  the installation of a
pilot  baghouse.   The pilot  unit was  placed in  service in October  1979.
This paper describes  the  test  program  being conducted in the pilot unit by
SPS for the  EPA.   These  test programs to  date have primarily  involved the
evaluation of the filtration characteristics of a number of fabrics in both
the shake/deflate and reverse air cleaning modes.

     This paper has been reviewed in accordance with the U.S. Environmental
Protection Agency's  peer and  administrative  review policies  and approved
for presentation and publication.
                        INTRODUCTION AND BACKGROUND

     In 1977, the Environmental Protection Agency (EPA) executed a contract
with Southwestern Public Service Company (SPS) to assess the performance of
a  large  fabric  filter system  used on  a  new utility boiler  burning  low
sulfur Western coal.   One option of this contract provided for the instal-
lation of a  pilot filter system on-stream with  Harrington Station Unit 2.
The EPA elected to exercise this option  in  1979,  and SPS agreed to operate
and maintain a pilot unit at Harrington Station.

     The pilot unit is a Wheelabrator-Frye,  Inc.  Model 366,  Series 11.5 RS
Dustube Dust Collector.  It has two six-bag  compartments and initially was
fitted with a Criswell Teflon-coated fabric; bags were 11.5 in. in diameter
and 366  in.  long.   Cloth  area per compartment   is  549  ft .   The  unit is
operated by a 480-V control panel and an instrumentation panel.

     The pilot unit  is installed at the southeast corner  of  the east bag-
house on Harrington's  Unit  2.   Inlet flue gas to the  pilot  is pulled from
the east inlet duct of the  main baghouse and mixed with hot preheater flue
gas through  a  mixing  valve to  maintain  constant  temperature.   (See Figure
1.)
                                    239

-------










NJ
<->














AIR







GAS DUCT
TO AIR
PREHEATEF
(700°F)

IT
<
1

£
3-
b.


PREHEATEf








HOT INLET
x-BLOCK DAMPER
[s<1 n ^ r


HOT INLET
h PURGE. 	 J

<*

• I
\
*•
COLD INLET
BLOCK DAMPER^











TEMPERATURE
/CONTROL DAMPER
r~~j

A_J -
iFANl
7 V OUTLET
^ ABLOCK VALVP





DUCT FROM AIR
PREHEATER TO
BAGHOUSE
(350°F)




AIR

FLOW "~













!






-J
' 	 V<}—1
COMPARTMENT COMPARTMENT
ON LINE\ (CLEANING

"— ^ 	 ^

/BY PASS THROTTL-
ING DAMPER
" " ' "1
BY PASS DAMPER r —
p




FLOW


1
S
r\
CONTROL 
-------
     In October of 1979,  the pilot baghouse was placed in service.   Several
initial problems bad  to  be  resolved before a  satisfactory  operating  level
could be  attained (1).   By July  of  1980, tbe  pilot unit  was ready  for
testing of particulate and  gaseous emissions.   Tbis first  series  of  tests
confirmed that flue gas  composition and particulate loading for the  pilot
unit were comparable to those of the main baghouse.

     The next  series  of  tests, which  are discussed  in  this  paper,  were
designed to  reflect   the  interests of  both SPS and  EPA.   These  areas  of
testing include  scale-up  studies with  shake/deflate cleaning,  reverse  air
fabric studies,  and  special interest studies,  such as ringed  bag testing
and  design  equations.  The  pilot baghouse is now being  prepared for  an
electrostatic enhancement program to be conducted at Harrington Station.


                   SCALE-UP WITH SHAKE/DEFLATE CLEANING

OVERVIEW

     The first objective of the pilot baghouse study following  start-up and
check-out was to prove that  suitable  scale-up  to the prototype unit exist-
ed.  Data were collected  over  a 5-month period with the pilot unit opera-
ting in various modes considered to be representative of the prototype bag-
house.

     The fabric  chosen  for  this initial program was W.  W.  Criswell's 442-
57DC-2  (10-oz.  Teflon-coated).  Ideally,  for  comparison,  the  same fabric
used in  the  prototype unit  would be used  in the pilot unit; however, this
mode of operation was not possible due  to the large number of fabrics being
tested in  the prototype  (see  Table I).  The  Criswell  fabric  was  selected
for  the  pilot unit because  it  represented the majority of  fabrics in the
prototype unit and had a proven performance record.

     In  the  first  phase  of  the pilot scale-up work,  the unit  was  operated
around  the  clock  for  1  month  at the design  air-to-cloth  ratio of  3.4
ft/min.  Phase  two called  for  alternating operation between  high and low
loads.  Therefore, in June the pilot unit was run at air-to-cloth ratios of
2.8  and  3.4  ft/min,  and in July  at  air-to-cloth  ratios  of  2.2   and  3.4
ft/min.

     Because  none  of  the above  test  methods  attempted to  cycle load with
the  boiler,  an effort was made in the third phase of the scale-up  study to
simulate the  daily load  cycle.   During  August  and  September the pilot unit
was  run  at  an air-to-cloth  ratio of 3.4 ft/min from 8:00 a.m.  to 3:30 p.m.
The  air-to-cloth ratio  was then  reduced  to  1.94  ft/min  until  the next
morning.

OPERATING PARAMETERS

     Cleaning cycle  parameters  and operating  conditions for  the  scale-up
study are:

                                    241

-------
           TABLE I.  PROTOTYPE  FABRIC  SUMMARY
         West  Boghouse
East Baghouse
1
*
3
*
5
*
7 1 Nomex All-
Spun;2 Nomex
Comb.jSCriso-
f lex 446;
4Crisoflex
449;BaL «
9
*
(Warp- In)
II
*
13
*
2
*
4
*
Menardi -
Southern
Teflon
Test Bags
8
*
10
*

12
*
14
*
IS
* *
17
* •»
19 Original bags
equipped with
special
shaker
mechanism
21
Criswell 442
Teflon B
Test Bags
23
Fabric Filters
502 Tri-Treat
Test Bags
25
Criswell 449
Teflon B
Test Bags
27
* *

•*• «•

* *
Fabric Rlters
All-Filament
Teflon
34 Acid Flex;
34 Tri-Treat
Balance :
Original Bags

Globe -
Albany
Nomex


* *

•*• *
16

18

20
22
24


26


28

  * Criswell 442 Teflon B; (10.5 oz.) (Rebagging complete).
** Criswell 449 Tri-Treat;(14oz.) (Rebogging complete).
                              242

-------
          Inlet Temperature        400°F
          Grain Loading            0.75-1.2 gr/acfm
          Filtering Cycle          30 min.
          First Null Period        20 sec.
          Deflate                  40 sec.
          Shake                    20 sec.
          Settle "no-flow"         20 sec.
          Cleaning Cycle           30 min.

DATA REDUCTION

     Pressure drop data  was collected throughout each  filtration cycle by
strip chart recorders.  A typical filtration cycle is shown in Figure 2.

     According to filtration theory  the  pressure drop,  AP,  across a single
compartment fabric filter is given by
                             AP = Ap  „+ K_cV2t                        (1)
                                    e    2
where
     Ap  = effective Ap after cleaning
       e
      K~ = the specific resistance

       c = grain loading

       V = air-to-cloth ratio

       t = filtration time.
     The  effective pressure  drop  after  cleaning, A p , was  determined by
extrapolating  the  linear portion of  the filtration curve shown in Figure 2
to zero time.

     The  specific  resistance  can be  calculated by rearranging the equation
above and setting  Ap equal to the terminal A p.  Grain loading, however, was
not measured on a  regular basis during  the  testing program.  Therefore only
values of K c  can  be calculated from the  data without some assumption about
the absolute value and  constancy of  the grain loading.   K c was calculated
from
                           K0c =  (Apt - Ap )/tV2                       (2)
                            2             e
                                    243

-------
o
k.
Q
                                   Terminal Pressure Drop,APt
               Slopes K2cV



Effective Pressure  Drop ,APe
                          Filtration Time
          Figure  2.  Typical filtration cycle.
 40


 35



 3O


 25



 20


 IS



 10
              I
Range of R/A

  Values
                 2.O   Z.2     2.5     2.8            3.4


                         Air-To-Cloth Ratio, f t/min
             Figure  3.
                                   results.
                             244

-------
     For comparison with the prototype, the terminal and effective pressure
drops were averaged to give an average cell-plate pressure drop.   The pilot
cell-plate pressure  drop  data cannot  be  simply compared to  the  prototype
flange-to-flange  pressure  drops:   one includes "casing" losses,  and  the
other does not.  Casing losses, of course, account for entrance effects  in-
to hoppers,  pressure drops across dampers,  duct losses, and  other losses
not associated with the fabric.

     Casing losses across the  Harrington  fabric  filters  are  given approxi-
mately by:
                              A PC = 0.156V
(3)
     To put the pilot and prototype data on the same basis and allow direct
comparison  of  the numbers,  0.156V  was  added  to  the average  cell-plate
pressure drop obtained from the pilot unit.

RESULTS
     Table  II  shows the average  flange-to-flange pressure drops  from the
pilot and prototype units for the 5 months of the study.  The numbers agree
remarkably  well  overall.   The range of the  flange-to-flange  pressure drop
values measured at each operating condition  is  shown  in the last column of
Table II.   The pressure drops for the pilot and prototype units fall within
a  fairly narrow range  for each  operating condition  except  for  June  and
August.   In June the pressure drop  on  the east side  of  the  prototype was
unusually high,  and in August  the east  side  pressure drop  was unusually
high.   No  reason  for these  differences  have  been  identified  but  it  is
suspected that operational problems  are responsible  rather than filtration
phenomena.

                   TABLE II.  SUMMARY OF SCALE-UP STUDY
                        WITH SHAKE/DEFLATE CLEANING
MONTH
May
June

July

August

September


Average
A/C,
f t/min
3.40
2.80
3.40
2.20
3.40
1.94
3.40
1.94
2.50
3.40

a Flange-to-flange
Prototype
East
7.21
6.23
8.08
4.61
8.00
3.88
7.86
4.11
5.70
8.48

pressure drop
F/Fa
West
7.50
5.73
7.27
4.84
8.27
3.74
7.97
3.99
5.76
8.85


Pilot
East"
7.66
6.23
7.59
4.29
8.71
3.58
9.18
4.34
5.87
8.80


F/Fa
West"
7.58
6.02
7.12
4.57
8.23
3.96
8.26
3.73
5.22
8.81


Range
0.45
0.50
0.96
0.55
0.71
0.38
1.32
0.61
0.65
0.33
0.65

   In. w.g.
                                    245

-------
     One surprising result of  the  study  was that cycling the pilot did not
give any  significant  improvement  in  agreement  of data.  Apparently  it is
not necessary in pilot  studies of  this type to  follow the load cycle since
good results  can  be  obtained  by  simply  operating  at  the  conditions of
interest.
                            REVERSE AIR CLEANING
OVERVIEW
     The next  phase of  the pilot  program  concerned reverse  air cleaning
studies  on a  number  of different  fabric  types.   A  summary of  fabrics
studied in the pilot unit is shown in Table III.
                     TABLE III.  PILOT FABRIC SUMMARY
Manufacturer*
Fabric
Style
Finish
Weighjj
oz/yd
Weave
Ken                  617F
FF                  502-1
GA                    877
FF                  502-1
MS                    601
FF                  504-1
FF                  504-1
MS       MS601T (7 Rings)
MS       MS601T (5 Rings)
MS                    601
MS                    601
MS                    601
MS       509-NC (7 Rings)
                         CS-428
                             AF
                            Q78
                             TC
                             TC
                             AF
                             TC
                             TC
                             TC
                             TC
                             TC
                             TC
                             TC
                  10
                  14
                  14
                  14
                  10
                  10
                  10
                  10
                  10
                  10
                  10
                  10
                   9
             3x1 T
             3x1 T
             3x1 T
             3x1 T
             3x1 T
             3x1 T
             3x1 T
             3x1 T
             3x1 T
             3x2 T
          Crowfoot
             2x2 T
             3x1 T
  MS - Menardi-Southern
  FF - Fabric Filters

  Twill

  Napped
           Ken - Kennecott
           GA - Globe Albany
            TC - Teflon Coating
            AF - Acid Flex (1-625)
                                    24f>

-------
     The reverse  air  study  looked  at the  pressure  drop  performance  of
various fabrics having different weights and  coatings as well as numbers of
rings.   In addition,  several fabrics were examined  with weave  types  that
are  not  commonly  encountered   in  utility filtration  systems.   The  vast
majority of fabric used today  in baghouses has a 3x1  twill weave.   To get
an idea of  how other weaves would react to utility filtration conditions, a
3x2  twill,  a  2x2  twill,  and  a crowfoot weave  fabric were  examined.   In
addition,  a napped fabric was evaluated to  examine the effectiveness of
napping as  compared to texturizing in producing a filtration surface.

OPERATING  CONDITIONS

     The operating  and   cleaning   cycle  parameters  employed  during  the
reverse air testing are:
               Operating

               Inlet temperature
               Grain loading
               Filtering cycle

               Cleaning

               First settle
               R/A settle
               Second settle
               R/A air-to-cloth
 400°F
'0.75-1.2 gr/acfm
 60 rain.
 30 sec.
 45 sec.
 30 sec.
 1.3-1.5
DATA REDUCTION

     The reverse air data were reduced in the same way as the shake/deflate
study, with one exception.  To  estimate  the flange-to-flange pressure drop
of  a  multicompartment prototype,  2.0 in.  w.g. was  added  to  the average
cell-plate pressure drop instead of  0.156V  *   Typically,  the casing losses
in  a  large reverse  air  collector are in the  1.5  to 3.0 in. w.g. range and
2.0 in. w.g. assumes fairly good design.

RESULTS OF REVERSE AIR STUDY ON STANDARD FABRICS

     Results of reverse air on standard fabrics are shown in Table IV.  All
of  the fabrics tested resulted in fairly high pressure drops.  The measured
cell-plate pressure drops ranged from 5.26  to 6.52  in. w.g.  The flange-to-
flange pressure drops for  a multicompartment prototype were estimated from
this  data.   These  pressure drops ranged from 7.26 to 8.52  in.  w.g.  at an
air- to-cloth ratio of 2.0  ft/min.
                                    247

-------
        TABLE IV.  RESULTS OF REVERSE AIR STUDY ON STANDARD FABRICS

Fabric
Style
Ken-617F

FF502-1AF
GA 877-Q78
FF502-1TC
MS 601-T
FF504-1AF
FF504-1TC

No. of
Rings
6

6
6
6
6
6
6

Air-to-
Clotha
2.30
(2.00)
2.00
2.00
2.00
2.00
2.00
2.00

Vb
32.29
(Calculated
31.21
27.36
33.89
27.19
29.24
32.22
AP
After
Clean-
ing
6.31
Values)
5.98
4.81
5.84
4.92
5.10
5.81
AP
Before
Clean-
ing0
7.78

7.05
5.70
7.00
5.80
6.05
6.92

Avg.C
AP
7.05
(5.84)
6.52
5.26
6.42
5.36
5.58
6.37

F/Fd
AP
9.05
(7.84)
8.52
7.26
8.42
7.36
7.58
8.37
  ft/min

b in. w.g. min. ft. gr.
       Ib  cu. ft.
         m

  in. w.g.

  Estimated flange-to-flange
     Unlike the remainder of  the  fabrics,  the first test fabric (Ken 617F)
was run  at  an air-to-cloth ratio of 2.3 ft/min.   The  results  of this test
indicated that  the  higher air-to-cloth ratio was  not  necessary to compen-
sate for a short testing period.  After the Kennecott test the air-to-cloth
was lowered to  2.0  ft/min.   To allow comparison with  the  rest  of  the fab-
rics, the Kennecott results were  projected at an air-to-cloth ratio of 2.0
ft/min.

     The data on  the  MS601-T and GA877-Q78 fabrics  (see  Table  IV) must be
viewed  cautiously since  neither came  to   equilibrium during  the testing
period.  In both cases the pressure drop monotonically increased throughout
the  testing  period.    This   phenomenon  of  slowly  approaching  operating
equilibrium has also been observed  for the MS601-T fabric in full-compart-
ment testing on the Harrington prototype baghouses.

     FF502-1AF  and  FF504-1TC  fabrics  had   the  same pressure  drops.   This
result  is  again  in agreement  with full-compartment  testing  done  on the
shake/deflate  Harrington filter.   In  these  full  compartment  studies  no
difference has  been noted between 10-oz.  Teflon-coated  fabrics and 14-oz.
acid-resistant coated fabrics in terms of pressure drop.
                                    248

-------
     The  FF502-ITC  fabric  is  a  14-oz.   fabric  with  a  10  percent  Teflon
coating.  This combination  represents  the "high-priced spread" in that the
more expensive 14-oz.  fabric  and Teflon  coating  are. combined.  The fabric
showed pressure drops in the same range as fabrics FF502-1AF and 'FF504-1TC.
Testing  of  14-oz.   Teflon-coated  materials  in   the  prototype  indicated
reduced  flow  ( 10 percent), but the  pressure drop was  not substantially
different than the commonly used FF502-1AF and FF504-1TC fabrics.
                                          2
     The  FF504-1AF  fabric  is  a lO-oz./yd   fabric  with  an  aci-'-resistant
coating.  This fabric is essentially the  "low-priced spread," c   jining the
least expensive coating and fabric.  The  fabric  exhibited one of the lower
pressure  drops  observed in the  pilot for  the standard  fabrics,  7.58 in.
w.g.  In  the  prototype this fabric  has  shown no  lower  pressure drop than
the other  fabrics.   Baglife on this fabric  is now approximately 20 months
in two compartments in Unit 2.   Observed  wear  on  the fabric seems to indi-
cate that  inferior bag  life  will  be  obtained under  shake/deflate  condi-
tions.   The exact  economic trade-off  of  bag cost versus bag  life  will be
studied for this  fabric  to see  if  it  offers an  economic advantage despite
its shorter bag life.

RING STUDY

     The  effect  of various numbers of rings  in a  bag  was  studied  using
samples having  five, six,  and seven  rings  (equally  spaced).   The  fabric
chosen for this study was  a lO-oz./yd   Teflon-coated material.  Originally
only MS601-T  was  to  be used  in  the  study,  but its failure to come  to
equilibrium made   it  necessary  to  include  data  previously  obtained  on
another   lO-oz./yd    Teflon-coated  material  (FF504-1TC).    The  results
obtained on these fabrics are shown in Table V.
                      TABLE V.  RESULTS OF RING STUDY


Fabric
Style
MS601-T
FF504-1TC
MS601-T
MS601-T


No. of
Rings
5
6
6
7


va
36.70
32.22
27.19
31.74
AP
After
Clean-
b
ing
5.76
5.81
4.92
5.65
AP
Before
Clean-
ing
6.94
6.92
5.80
6.68


Avt.
AP6
6.35
6.37
5.36
6.16


F/Fh o
A Pb'c
8.35
8.37
7.36
8.16
       »1
          in. w.g. miii. ft. gr.
                 Ib  cu. ft.
                   m
       b  .
          in. w.g.

       Q
          Estimated flange-to-flange
                                    249

-------
     The pressure  drops for  the three  fabrics  that obtained  equilibrium
during the  testing show very  little  difference.   From this data  it  would
appear that  the  number of rings in a bag  do not have profound  effects  on
pressure drop.   Outside the range of five  to seven rings in a  30-ft bag,
however, some effect on pressure drop may occur.

NON-STANDARD FABRICS

     One area  of  fabric  filter research left virtually  untouched in  the
utility industry is the effect of fabric weaves and construction on perfor-
mance of  filtration materials.  To date only texturized 3x1 twill weaves
have been  employed to  any extent on  glass  fabrics.  Certain  experiences
with 3x1  and 3x2  twill Nomex  fabrics  at Harrington pointed  to weave  as
being a potentially major factor in fabric performance.

     Several fabric weaves  and constructions that are uncommon  to utility
application were tested in  the pilot  unit.   Table VI lists  the  results for
these non-standard fabrics.
           TABLE VI.  RESULTS OF TESTING ON NON-STANDARD FABRICS
Fabric
Style
MS 3x2 T
MS-CF
MS 2x2 T
MS509N
No. of
Rings Weave
6 3x2 T
6 Crowfoot
6 2x2 T
7 Napped

32.51
26.93
16.07
32.69
AP
After
Clean-
5.05
5.02
3.09
5.73
AP
Before
Clean-
ing
6.71
5.94
3.57
6.75
Avgfi
APB
5.61
5.48
3.33
6.24
P-
7.61
7.48
5.33
8.24
            in. w.g. min. ft. gr.
                 Ib  cu. ft.
                   m

            in. w.g.
            Estimated flange-to-flange
     The  3x2  twill  and  the crowfoot fabrics have significantly lower pres-
sure drops  than most of the fabrics in  common  use.   No operational diffi-
culties were  encountered during the testing of either of these fabrics, but
the  crowfoot  weave exhibited  a  rather strange  phenomenon.   Initially the
pressure  drop rose  after start-up and  leveled off  at  close to 5.5 in. w.g.
For the last  3 days of the test period, however, the pressure drop de-
                                    250

-------
creased suddenly to,approximately 4.3 in. w.g.   To  be  sure of interpreting
this data   conservatively,  the  last few days of  operation were  ignored in
the data analysis.  Further analysis of the crowfoot fabric would certainly
be wise.

     The 2x2 twill fabric exhibited pressure drops below 2 in. w.g. for the
first 4 days and  increased  slowly  throughout  the testing period.  Only the
last few days of testing were used to evaluate the 2x2 twill fabric.  Since
it did  not  come  to equilibrium by leveling off  in  pressure drop,  the data
are not satisfactory.  Again, more  testing  should have been done,  but time
was not available.

     The seven-ringed napped  fabric  with  a pressure drop  of  8.24  in. w.g.
failed  to  show any  advantage over  the  3x1 twill  seven-ringed  texturized
fabric at a pressure drop of 8.16 in. w.g.
                             ADDITIONAL STUDIES
K c RESULTS
     The K.c  values  obtained during  the  scale-up period are  shown versus
air-to-cloth ratio in Figure 3.

     K c decreases as  the air-to-cloth ratio is  lowered.   At  air-to-cloth
ratios lower  than  2.7  ft/min, however, no  significant  changes in  K c  are
observed.  If grain  loading can be assumed constant, this  implies  tnat K
also decreases  with  air-to-cloth ratio and  does not change  significantly
below an  air-to-cloth  ratio  of  2.7 ft/min.  The  minor dip at an  air-to-
cloth ratio of 2.5 ft/min may be an anomaly.  If not, it is due to settling
out of heavier particles as the  velocity gets lower and  lower.

     The  range  of K c  values  measured for  reverse air  cleaning   on  the
various test bags is also shown in Figure 3.  As  shown,  the K values  for
reverse air cleaning are  substantially larger than  the  K«  values obtained
for  shake/deflate  cleaning at  the same  air-to-cloth  ratio (2.0 ft/min).
Cleaning then  is  shown to  influence  K  as well  as ash  properties.   This
effect could  reflect the length  of time  required to"repair"  the ash cake
after cleaning.  Reverse air cleaning causes fissures in the ash cake along
with some areas where  cake has  sloughed off.  After cleaning,  these areas
are quick to rebuild with the result  that any additional cake  is  generally
distributed and influences the pressure drop increase.   After shake/deflate
cleaning, large areas of  the  cloth are stripped bare to  the surface.   The
time required to repair these larger  areas  strongly influences the  rate at
which pressure drop can increase across the  filter cake.
DESIGN EQUATIONS

     The pilot data obtained in the scale-up and reverse air studies can be
used to develop  pressure  drop relationships for use  in design.   The equa-
tions are:
                                    251

-------
                        AP = 1.60V1'34        Shake/deflate           (A)
                        AP = 3.29V1'34        Reverse air             (5)
     Figure  4 shows  a plot  of  pressure  drop  versus air-to-cloth  ratio
predicted from these equations.

     Strictly  speaking,  these  correlations  apply  only to  one ash  type,
certain fabrics, and a given set  of cleaning  conditions.   The  data used in
developing these correlations, however, represent the cloth used in over 90
percent of the baghouses in operation on utility boilers.   In addition, the
cleaning  conditions employed  are  representative  of  those used  by  many
installations.  These  equations  then  should  be useful  in  modeling filtra-
tion of ashes similar to the Wyoming ash produced at Harrington Station.

     Because  the  operating  conditions  during  testing  were  deliberately
chosen to simulate  operations  at  high  capacity factors,  these  correlations
should not be used for peaking units and other low load applications.
                                CONCLUSIONS

     The  fabric  filter pilot unit  at  Harrington Station  has  successfully
demonstrated the ability of a pilot unit to model large scale fabric filter
systems.  The  pressure drop model developed from the  pilot  unit describes
both the  Harrington  fabric filters (Unit 2 and  Unit  3)  with good accuracy
and should be a useful tool in design of units with similar fly ashes.

     Although  no  direct comparison from  pilot  to prototype was available
for the reverse air  studies,  there is no apparent reason  that  the reverse
air equation should not be accurate for the ash type studied.

     The  experience  gained from Harrington  Station's pilot and prototype
units  indicates  that  the  typical  design goal  of  6.0 in. w.g.  flange-to-
flange pressure drop can be obtained with a properly designed shake/deflate
baghouse at an air-to-cloth ratio of approximately 2.7 ft/min.   The reverse
air study indicates  the  maximum air-to-cloth ratio for  a  reverse air  bag-
house on western ash is in the 1.5-1.7 ft/min range.

     The  reverse air  fabric  study  points to the need  for  more  research in
the area  of  fabric weaves and construction.  Shown below  in Table VII are
all the fabrics tested arranged according to increasing pressure drop.
                                    252

-------
     15

     14

     13

     12



     IO-
_§.£
U-i   9
 1 o
2a   8
 i
 fi»3   7
5

4

3

2

I •

0-
                      Reverse Air
                                              Shake / Deflate
               1.5
2.0          2.5

Air to  Cloth Ratio, ft/min
                                                             3.0
3.5
    Figure  4.   Plot  of pressure drop  vs. air-to-cloth.
                               253

-------
                   TABLE VII.  FABRICS BY PRESSURE DROP
Fabric
Weigh£
oz/yd   Coating   Weave
F/Fd
 AP   Comments
MS601-T
GA 877-Q78
MS601-T
MS601-T
FF504-1AF
MS601-T
Ken-617-F
MS601-T (7 Rings)
MS509N (7 Rings)
MS601-T (5 Rings)
FF504-1TC
FF502-1TC
FF502-1AF
10
14
10
10
10
10
10
10
10
10
10
14
14
Teflon
078
Teflon
Teflon
Acid Flex
Teflon
CS-428
Teflon
Teflon
Teflon
Teflon
Teflon
Acid Flex
2x2
3x1
3x1
Twill
Twill
Twill
Crowfoot
3x1
3x2
3x1
3x1
3x1
3x1
3x1
3x1
3x1
Twill
Twill
Twill
Twill
Twill
Twill
Twill
Twill
Twill
5
7
7
7
7
7
7
8
8
8
8
8
8
.33
.26
.36
.48
.58
.61
.84
.16
.24
.35
.37
.42
.52
Equilibrium not
Equilibrium not
Equilibrium not



Calculated






obtained
obtained
obtained










  Flange-to-flange, in. w.g.
     The highest  pressure  drops recorded during  tests were for  the  three
most commonly used fabrics.  Two of  the  non-standard weaves,  the 3x2  twill
and  the  crowfoot, did  show significantly lower  pressure compared to  the
commonly used types.  Although  the 2x2 twill fabric  did  not  come to  equi-
librium, its behavior during the test program  wpuld indicate that it  may
also be  a  candidate for enhanced  performance.   The  three  lowest pressure
drop fabrics are those that did not  come  to  equilibrium during the testing
period,  and  no clear conclusion can  be  drawn from the data about  their
performance.

     These results demonstrate  that  better performance may be obtained if
further research is aimed at developing fabrics especially suited for  util-
ity filtration systems.   Considering the costs of premature bag replacement
and the  cost of  high pressure  drop,  further studies  could  lead  to greater
cost-effectiveness.
                                REFERENCES

1.   Ladd, K.  L.,  Hooks,  W. , Kunka,  S.  L.,  and Harmon, D.  SPS  Pilot Bag-
     house Operation.   In  Third Symposium on the  Transfer  and Utilization
     of Particulate  Control Technology:  Volume  I.  Control  of Emissions
     From  Coal-Fired  Boilers,   EPA-600/9-82-005a   (NTIS  PB83-149583),  pp
     55-64, July 1982.
                                    254

-------
             THE USE OF SONIC AIR HORNS AS AN ASSIST TO REVERSE

               AIR CLEANING OF A FABRIC  FILTER DUST COLLECTOR
                by:  Alan R. Menard
                     R. Mark Richards
                     Public Service Company of Colorado
                     Arapahoe Station
                     Denver, Colorado 80201
                                  ABSTRACT

     A detailed summary of the installation, test proqram, and performance of
sonic air horns used as an assist to reverse air cleaning of an operating
FFDC is presented.  Four sonic air horns were installed and tested on com-
partment number 11 of Public Service Company of Colorado's Arapahoe Unit #3
FFDC.  The overall objective of the test program was to reduce the pressure
drop across the tube sheet of the test compartment.  Sonic cleaning reduced
the compartment tube sheet differential pressure by approximately 27 percent;
this was sufficient to warrant the purchase and installation of two (2)  sonic
air horns in each of the 14 baghouse compartments.
                                     255

-------
                                INTRODUCTION


     The need to decrease the pressure drop across the filtering elements in
a Fabric Filter Dust Collector (FFDC) is inherent to the reduction of its
operating costs.  Public Service Company of Colorado's Arapahce Unit #3 FFDC
has experienced flange to flange pressure drops on the order of 10 inches VWC
at full load as opposed to the design pressure drop of 6 inches VWC.  Much of
this discrepancy in pressure drop can be attributed to the formation of
nodules (Reference 1) or cake on the inner surface of the filtering elements
which is not removed during the reverse air cleaning cycle (Figure III-l).
The formation of nodules or cake reduces the effective filtering area of the
baghouse, which in turn, increases operating air-to-cloth ratio.

     A solution was sought by Arapahoe Station personnel that would decrease
the overall pressure drop of Arapahoe Unit #3 FFDC.  Because the nodule/cake
buildup on the filtering elements was felt to be directly related to the
pressure drop problem, the use of sonic air horns used in conjunction with
the reverse air cleaning cycle was thought to be a viable solution to the
problem of fly ash removal.  A test program was devised that would determine
the effect of using sonic air horns during the reverse air cleaning cycle as
a means of reducing the pressure drop across the filtering elements.

     Four Fuller air horns were purchased and installed in compartment #11 of
Arapahoe Unit #3 FFDC.  Compartment #4, having the most similar flow
characteristics to compartment #11, was used to collect base data for compar-
ison purposes during the test period.  The use of the sonic air horns made an
appreciable improvement in the performance of the test compartment; signifi-
cant enough to warrant the purchase and installation of horns in all 14 bag-
house compartments.

     This report summarizes the installation, test program and performance
of sonic air horns used as an assist to reverse air cleaning of a FFDC.
                                    256

-------
                           TECHNICAL DISCUSSION
BAGHOUSE DESIGN AND HISTORY

     Public Service Company of Colorado's Arapahoe Unit #3 FFDC was designed
for 315,000 acfm flow with a 6-inch VWC pressure drop flange to flange, and
a 2.16 operating air-to-cloth ratio.  The boiler/turbine combination has a
48 MW capability and operates on low sulphur, sub-bituminous western coal.
The baghouse is a reverse air cleaning type, consisting of two rows of seven
modular compartments with inlet, outlet, and reverse air ducts located
between the two rows of compartments (see Figure II-l).  Each compartment
contains a total of 236 bags which are arranged as shown in Figure II-2.
The bags are 8 inch diameter, 22 feet long, woven fiberglass with teflon B
coating.  The baghouse was started'up on May 18, 1979, and reached initial
full load operation on May 21, 1979.  For complete design specifications on
Arapahoe Unit #3 baghouse, see Table 1.

     After initial operation of the baghouse, several mechanical problem
areas such as valve seating, air cylinder operation, and thimble leaks were
recognized and repaired.  With these problems solved, performance testing of
the baghouse began.  Originally, the baghouse was outfitted with 8-1/4 inch
diameter Acme Mills woven fiberglass bags with teflon coating.  After
approximately two months of operation, the baghouse experienced flange to
flange pressure drops inexcessof the design pressure drop of 6 inches VWC
at full load.  Acceptable pressure drops could only be obtained by hand clap-
ping and/or beating the bags to remove the cake buildup on the inside sur-
face  of the filtering elements.  However, as the cake began to build up on
the bags, the pressure drop would again exceed design levels.  The baghouse
manufacturer felt that the pressure drop problem could be solved by replacing
the O.E.M. bags with Menardi Southern bags with teflon B coating.

     Compartments 1 and 7 were outfitted with the Menardi Southern bags and
a test program was initiated.  The performance of the two compartments out-
fitted with the new bags was significantly better than the other compartments
(Reference 2).  It was decided to install the Menardi Southern bags in all
14 baghouse compartments.  Eight-inch diameter bags were chosen as opposed to
8-1/4 inch diameter bags because of long lead times associated with obtaining
the 8-1/4 inch diameter bags.  This three percent reduction in filtering area
was felt to be outweighed by the performance increase with the new bags.

     During this time period, data acguisition and testing was initiated to
optimize the operation of the baghouse.  Different types of coal were burned.
Although some coals increased operating flange to flange pressure drop, no
coal was found that decreased pressure drop.  Optimization of the cleaning
cycle was also attempted.  Multiple settling/reverse air periods, continuous

                                    257

-------
                                     Table 1

                   ARAPAHOE UNIT 3 FFDC DESIGN SPECIFICATIONS
DESIGN CONDITIONS

Process  .......................................... Power Generation
Suspended Material ............................... Fly Ash
Fuel  ............................................. Pulverized Coal
Gas:
   Source ................................... ..... Coal-Fired Boiler
   Gas Flow (max. acfm) .......................... 315,000
   Temperature (°F) .............................. 290
   Outlet Loading (fr/acf) ....................... 0.007
   Pressure Drop (in. VWC) ....................... 6.0
   Design Pressure (in. VWC) ..................... ±20
   Total Filter Area (sq ft) ..................... 156,995
   Effective Filter Area (sq ft)± ................ 145,781
   Total Filter Ratio ........................ ---- 2:1
   Active Filter Ratio±  ......................... 2.16:1
   Reverse Air Flow:
     Total (max acfm) ............................ 16,700
     Static Pressure (in VWC) .................... 10.6
     Reverse Air Fan (qty) ....................... 2 (1 on standby)

FFDC DATA

Type  . ............................................ Modular
No. of Units ...... .... ........................... 2
No. Compartments/Units ........................... 7
Baqs:
   Total Quantity ................................ 3304
   Per Compartment ............................... 236
   Material ...................................... Menardi Southern, Teflon B
                                                   coating
   Diameter (in.) ................................ 8
   Length (ft) ...... ............................. 22
Power Requirements ................. .. ............ 480V, 3 phase, 60Hz
Air Flow Regulation
   Hopper Inlet (Isolation) ...................... Butterfly °       14
   Reverse Air/Outlet Valve ...................... Double Disc0      14
   Reverse Air Control (Nul 1 ) Damper ............. Butterfly °        1
   Outlet Duct (Isolation) ____ . .................. Butterfly °       14
   By-Pass Damper ................. . .............. Poppet °           2

   ±With one compartment out for cleaning and one compartment out for
    maintenance
   0 Pneumatic operated

                                      258

-------
Figure  II-l   Arapahoe Unit 3 Baghouse Compartment Locations
                           2
                                              /3
                                              10
                                               8
t
                                                          NOKTH
                                           \
                                          XKJUET
                                     259

-------
Figure II-2  Typical Compartment Filter  Element  Location
                       A  8  C
D  C  F
H  1
/
2
3
-f
j^
..}
1C
7
8
3
SO
//
/a
/3
H
tS
Ho
17
16
n
-">,-\
>..-'
•? /
^-/
f-— f~^
23
?--i


OOO
ooo
OOO
ooo
ooo
ooo
ooo
ooo
ooo
ooo
0 0 O
ooo
ooo
ooo
ooo
ooo
O O O
C) O O
ooo
0 O O
O O O
C 0 O
C1 O 0
coo


OOOG
0000
C O 0 O
oooo
oooo
oooo
oooo
oooo
oooo
oooo
oooo
oooo
oooo
oooo
oooo
oooo
O O 0 0
.oooo
O O O O
o o o 6
0 0 0 0
oooo
oooo


	 . !.„.,-. , i 	
ooo
ooo
000
ooo
ooo
ooo
ooo
000
ooo
ooo
0 O O
O O 0
O 0 0
0 0 O
ooo
ooo
0 O 0
ooo
o o o
ooo
ooo
OO ^)
on ) •
00 )
i
!
                                         -- DOOR.
                                      260

-------
cleaning as well as differential pressure  initiation of the cleaning cycle
were also tested.  None of these experiments made  any significant  improve-
ment in flange to flange pressure drop.

     After approximately three months of  operating with the new Menardi
Southern bags, the pressure drop from the inlet  flange to the outlet flange
again exceeded the design specification of 6 inches VWC.  To maintain a
reasonable pressure drop, the bags still  had to  be periodically hand beaten
to  remove the cake/nodule buildup on the  inside  surface of the filtering
elements.  As a result of this, Arapahoe  Station personnel, with assistance
from the staff at EPRI's Emission Control  Test Facility at Arapahoe Station
sought  a solution to the pressure drop problem.  Because all attempts at
optimizing the cleaning cycle and the switching  to a different filtering
element made no appreciable improvement on the overall flange to flange
pressure drop, it was felt that the nodule/cake  buildup on the filtering
elements was directly related to the pressure drop problem.  The use of sonic
air horns as an assist to the reverse air cleaning cycle was felt  to be a
viable  solution to the problem of cake buildup not being removed from the
filtering elements with reverse air cleaning only.  A test program was devel-
oped that would determine the effect of sonic cleaning on the nodule/cake
buildup problem, and ultimately would determine  if sonic cleaning  would re-
duce the operating flange to flange pressure drop.

HORN  INSTALLATION

     Four  sonic  air  horns were obtained on a  trial  basis for installation at
 Arapahoe  Station  from  Fuller Company.  Because of  time limitations, a test
 program was devised  that was somewhat  limited  in scope but would still deter-
mine  the  overall   effectiveness of  sonic  cleaning. Arapahoe Station person-
 nel,  in conjunction  with the staff  at  EPRI's Emission Control Test Facility
at Arapahoe Station, assembled  a  list  of  critical  parameters that  would be
monitored  during  the test  period.   During these  initial discussions, decisions
were made  on horn  placement, boiler  loading, horn  operation and special
equipment  needed  to  control horn  operation.

     Four  Fuller air horns, model 310-64-1-0099-01 were installed  in compart-
ment 11 of  Arapahoe  Unit #3 FFDC.   It was thought  that the compartment should
be  divided  into four equal quadrants, and a  horn placed at the center of each
quadrant.   The location of the  "Doghouse" containing the outlet and compart-
ment isolation valves made this objective impossible and the horns were
placed  as shown in Figure  II-3.   Figure II-4 shows the installation of a
typical horn through the roof of the compartment.

     Each air horn requires 70-90 psig air at 50 scfm.  The control air
system  that operates the existing air cylinders  on top of the baghouse, at
80  psig, was adequate to supply the horns.  However, it was felt that an air
receiver tank was necessary to supply the  four horns with a sufficient volume
of  air.  A  30 cubic foot air receiver tank was purchased and installed on top
of  the  baghouse.  A one (1) inch diameter manifold was run from the receiver
tank to the horns with one-half U) inch  copper  tubing branching from the
manifold to each horn.
                                      261

-------
Figure II - 3  Test Compartment Horn Locations
      Horns
                   "Doghouse" •
                                                 Outlet

                                                 Duct
                                                 Reverse
                                                 Air Duct
                   Side View
c


Horns
3 0
Top View
262
                                                      North

-------
Figure 11-4  Compartment Horn Installation
                                       263

-------
     The question of when to operate the horns and for what duration during
the cleaning cycle was discussed.  Having limited experience with sonic air
horns, three options were considered:

     (1)  Operate the horns during the initial settling period (no reverse
          air);

     (2)  Operate horns during reverse air cleaning;

     (3)  Operate horns during the final settling period.

     Because of time limitations on the Fuller air horn trial period, it was
not feasible to optimize horn operation.  It was decided that option two
would achieve the best results.  The horns would be operated as an assist to
the reverse air fan for a period of ten seconds.  This proved to be the best
method of horn operation and will be discussed later.

     A means of controlling horn operation was devised.  A normally closed
solenoid valve was installed in the line exiting the air receiver tank that
would open when horn operation was desired.  At the baghouse control panel,
reverse air damper movement (initiation of compartment clean) would initiate
a primary timer.  After 40 seconds, the first timer would initiate a second
timer which operated the solenoid valve for a period of ten seconds.  This
system served to operate the horns at the approximate time the reverse air
flow reached the compartment.

TEST PROGRAM

     The objective of the test program was to determine the effect of sonic
cleaning on flow and pressure drop in the test compartment.  An increase in
flow and a decrease in pressure drop in the test compartment was desired.
Four Fuller air horns were installed on compartment #11 of Arapahoe Unit #3
FFDC and it was designated the test compartment.  Compartment #4, having the
most similar flow characteristics to compartment #11,  was used to accumulate
base data for comparison purposes.  The following is a list of parameters
that were monitored during the test period:

     (1)  Pressure drop across the thimble sheet of the test compartments;

     (2)  Exit flow of the test compartments;

     (3)  Flange to flange pressure drop;

     (4)  Inlet, outlet, reverse air, and compartments 4 and 11 exit gas
          temperature;

     (5)  Compartment ash pulling times;

     (6)  Bag weights - before and after cleaning.

     For comparison purposes of data taken on different days, it was decided
                                     264

-------
that a constant boiler load was essential.   Variations in load would cause
fluctuations in flow, pressure drop, etc.,  which would have made data inter-
pretation and correlation more difficult.   The unit was base loaded at 46 MW
(full load)  one (1) day per week during the test period.  Although the horns
continued to operate during periods of varying load, the majority of relev-
ant data was taken during the periods of base loading.  The instrumentation
and recorders were installed and were ready for operation by February 22,
1982.

    Before the test period began,  bags were weighed in each of the test
compartments before and after cleaning.  Ash pulling times were   io record-
ed.  The recorders were placed in service  to accumulate base dat^. before any
horns were operated.

    It was decided to test the horns in successive combinations of 1, 2, or
4 horns to determine the minimum amount of horns required for successful sonic
cleaning.  On February 23, 1982, one horn  was operated.  On February 25,
1982, 2 horns and on March 29, 1982, 4 horns were tested.  Ash pulling times
were recorded each Monday.

     During the test period, the baghouse was on a continuous clean cycle.
There was a ten minute delay between-each compartment clean, and it would
take approximately 2.75 hours to clean all  14 compartments.  Reverse air flow
was held constant during the test program.  For a complete log of the test
program, see Appendix A.
                                     255

-------
                        TEST  RESULTS  AND  DISCUSSION
 Introduction
      The  use of  sonic  air  horns  as  an  assist  to  reverse air cleaning proved
 to be a viable solution  to the problems  of  nodule/cake formation on the
 filtering elements  and increased pressure drop in an operating FFDC.  During
 the test  period,  bag weights, filter element  cake thickness and tube sheet
 differential pressure  were all reduced in the test compartment.  In addition,
 flow through compartment 11 increased  during  the period of horn operation.
 Results indicated that two (2) horns per compartment were sufficient to
 adequately clean  the compartment and maintain the reduction in pressure drop.
 As a result of this test program, Public Service Company of Colorado has
 purchased and installed  two (2)  sonic  air horns  in each of the fourteen (14)
 baghouse  compartments  at Arapahoe Station's Unit #3 FFDC.  This section
 summarizes the performance of sonic air  horns used as an assist to reverse
 air cleaning of  a FFDC.

 Discussion of Results

      The  overall  objective of the test program was to reduce the pressure
 drop across the  thimble  sheet in the test compartment.  The following para-
 meters were monitored  during the test  program:

      (1)  Bag weights  and  samples;

      (2)  Compartment  ash  pulling times;

      (3)  Thimble sheet  differential pressure;

      (4)  Test compartment exit  flows.

      A summary of the  effect of  sonic  cleaning on each of the above listed
 parameters is presented  below.

 Bag Weights and  Samples

      Probably the most significant effect of  sonic cleaning was on the fil-
tering elements (bags)  themselves.  Before the sonic horns were operated, bags
 wereweighed in both compartments 4 and 11.  Bag  weights ranged from 34 to 55
 pounds with the average weight being approximately 46 pounds.  At several
 points during the test program,  the test compartments were isolated and bags
 were weighed.  Compartment 4's filtering elements remained the same in
 weight while the  bag weights in  compartment 11 were dramatically reduced.
                                     266

-------
               kr*w   . VT, , •  %,»TMB^MNfTT1'1? • " <§• '•   •' v • iiy^v •». ^^

        -vBilff                     ^>
Figure III-l  Nodule/Cake Formation on a Typical Filtering Element - Without
           Sonic Cleaning
                              •     .
             ,
        . • -*:.

 Figure III-2 Nodule/Cake Formation on a Typical Filterinn Element - With
            Sonic Cleanina
                             267

-------
 Weights in the  test compartment ranged from 12-1/2 to 25  pounds  with  the
 average weight  being approximately 18 pounds.   A new filtering element  weighs
 approximately 9 pounds.   The  sonic air horns reduced the  amount  of  fly  ash"
 cake buildup on the filtering elements an  average of 76 percent.  This
 reduction  in weight has  an  added benefit.   The  filtering  elements are ten-
 sioned to  50 pounds in the  compartments.   As bag weights approach 50 pounds,
 it is difficult to  maintain proper bag tensioning which affects  the way the
 bags collapse during reverse  air cleaning.   With the reduction in bag weights
 due to sonic cleaning, more uniform filtering element tensioning is possible.
 For a complete  tabulation of  bag weights taken  during the  test program  see
 Table 2.

      Samples of the filtering elements were taken  in each  of the test com-
 partments  before and after  cleaning.   Samples were  cut from the top,  middle
 and bottom of selected bags without disturbing  cake  buildup.  The difference
 in nodule  formation with and  without  sonic  cleaning  can be seen in  Figures
 III-l and  III-2.

 Compartment Ash Pulling  Times

      Another means  of determining  the  effectiveness  of sonic cleaning was to
 monitor the amount  of time  required to  remove the fly ash  from the  compart-
 ment ash hoppers.   An increase  in  the  amount of  time  required to remove the
 ash from the hopper on the  compartment  with  sonic horns would indicate  that
 the cake buildup  on the  filtering  elements was  being  removed.  Ash  pulling
 times before the  test program began were recorded and are  shown in  Figure
 III-3.   Figure  III-4  shows  compartment  ash pulling times during one and two
 horn operation.   Assuming that the curves in Figure  III-4 would be  similar
 in shape to the  curves in Figure III-3  without  sonic  cleaning, ash  pulling
 times in compartment  11  increased  by approximately 90 percent.  It was ex-
 pected  that after the majority of  cake  buildup  in the test compartment  had
 been removed by  sonic cleaning, the ash pulling  times for  compartment 11
 would decrease  from the  ash pulling times recorded when the horns were first
 operated.   Figure III-5  shows the  compartment ash pulling times after two
 weeks of continuous two  hourn operation.  The ash pulling time for compart-
 ment 11 should be slightly  higher  than  normal because of the increase in flow
 through the  compartment  due to sonic cleaning (discussed below).   However,
 the  increased amount  of  fly ash collected in compartment 11 over the 2.75
 hour time period between cleanings is difficult to measure and does not show
 up  readily  in the data.

 Thimble Sheet Differential  Pressure

     The pressure drops across the tube sheet of compartment 4 and  11 were
 recorded on a two pen strip chart  recorder.  Data was recorded at full load
 before any  horns were operated.  Tube sheet differential  pressure measured at
 that time was 6.5 inches  VWC immediately after the compartment cleaned and
would decay to 8.5-9 inches  VWC immediately before cleaning (see  Figure  III-
6).

     On February 23, 1982, with the unit stabilized at 45  MW,  one horn in  the
northeast corner of the compartment was operated.  Pressure drop  across  the
tube sheet in compartment 11 was reduced to 3.2  inches VWC immediately after
                                     268

-------
Bag #
13G
20D
IOC
200
15C
10G
19G
16H
11D
140
18G
8H
5C
50
5G
5H
12C
120
12G
12H
20C
200
20G
20H
7C
200
13G
20D
20G
80
8H
12G
15C
Compartment
4
4
4
11
11
11
11
11
11
4
4
4
11
11
11
11
11
11
11
11
11
11
11
11
11
11
11
4
4
11
11
11
11
Weight
6 Ibs.
4 Ibs.
34 Ibs.
46i Ibs.
49 Ibs.
52 Ibs.
47 Ibs.
49 Ibs.
55 Ibs.
45X Ibs.
45 Ibs.
37 Ibs.
12| Ibs.
14 Ibs.
15 Ibs.
16* Ibs.
15 Ibs.
14 Ibs.
144 Ibs.
15 Ibs.
19 Ibs.
154 Ibs.
15| Ibs.
18 Ibs.
18i Ibs.
18 Ibs.
19 Ibs.
36 Ibs.
34 J Ibs.
20 Ibs.
24 Ibs.
19 Ibs.
22 Ibs.
Table 2
BAG WEIGHTS
Comments
0 horns, before cleaning
0 horns, before cleaning
0 horns, before cleaning
0 horns, before cleaning
0 horns, before cleaning
0 horns, before cleaning
0 horns, after cleaning
0 horns, after cleaning
0 horns, after cleaning
0 horns, after cleaning
0 horns, after cleaning
0 horns, after cleaning
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, after cleaning 3/1/82
2 horns, before cleaning 3/17/82
2 horns, before cleaning 3/17/82
2 horns, before cleaning 3/17/82
0 horns, before cleaning 3/17/82
0 horns, before cleaning 3/17/82
2 horns, after cleaning 3/18/82
2 horns, after cleaning 3/18/82
2 horns, after cleaning 3/18/82
2 horns, after cleaning 3/18/82
269

-------
15D
17G
22D
22G
3C
3H
8C
8D
11G
11H
17C
20G
3C
3H
11D
11G
21C
21H
11
11
11
11
11
11
11
11
11
11
11
11
4
4
4
4
4
4
22i Ibs.
25 Ibs.
24 Ibs.
24i Ibs.
18 Ibs.
22 Ibs.
21i Ibs.
18 Ibs.
171 Ibs.
21 Ibs.
20 Ibs.
24i Ibs.
40 Ibs.
35i Ibs.
35i Ibs.
38 Ibs.
34 Ibs.
35 Ibs.
2 horns, after
2 horns, after
2 horns, after
2 horns, after
4 horns, after
4 horns, after
4 horns, after
4 horns, after
4 horns, after
4 horns, after
4 horns, after
4 horns, after
0 horns, after
0 horns, after
0 horns, after
0 horns, after
0 horns, after
0 horns, after
         cleaning 3/18/82
         cleaning 3/18/82
         cleaning 3/18/82
         cleaning 3/18/82
         cleaning
         cleaning
         cleaning
         cleaning
         cleaning
         cleaning
         cleaning
         cleaning
         cleaning
         cleaning
         cleaning
         cleaning
         cleaning
         cleaning
270

-------
       Figure III - 3  Compartment Ash Pulling Times - Pre Test Data
to
      m
      
-------
              Figure III-4  Compartment Ash Pulling Times During One and Two Horn Operation
        12-
        10-
         8-
to
~j
to
      M

      0)



     1
      CO
         6-
     •s
        k-
Note:  Differences in ash pulling times for one and two horn operation are

       due to boiler load fluctuations and time between ash pulling cycles.

       The intent of the Figure is to show time differences between

       compartment 11 and the rest of the curve.
                                                   2 horns operating
                                                                                horn operating


                               ,
                                                                            _,.

-------
          Figure III - £  Compartment Ash Pulling Times After Two Weeks of Continuous Two Horn Operation
to
        18-



        16-
       CO
       0
       ft
       CQ
         k-
                       T~

                       2
k
~T
 5
7
10    11     12     13
                                               Compartment

-------
FIGURE Itt-6. COMPARTMENT TUBE  SHEET
            PRESSURE  DROP AT FULL LOAD
            NO HORNS OPERATING
85 10-

-------
cleaning (see Figure III-7).  It was assumed that the pressure drop across
the tube sheet would increase at the same rate as in the compartment without
sonic horns.  Pressure drop increased at a more rapid rate until it was equal
to the pressure drop in compartment 4 immediately before compartment 11
cleaned.  This phenomenon can be explained because accompanying the decrease
in pressure drop there was a substantial increase in flow through the test
compartment.  This indicates that a greater amount of fly ash was being col-
lected in the compartment with sonic horns than in the other compartments,
hence, the rapid increase in pressure drop.

     On February 25, 1982, two horns, one in the northeast and one in the
southwest corner of the test compartment were operated.  Pressure drop across
the tube sheet was reduced to 3.0 inches VWC immediately after the compart-
ment cleaned.  Again the increased rate of pressure drop decay was observed.

     During this time period, unit 3 was loaded as P.S.Co. system require-
ments mandated.  This served two purposes; first, to allow the baghouse test
compartment to experience a more "real world" environment, and secondly to
allow the test compartment time to seek its own steady state operating point.
Pressure drop across the tube sheet increased to 4 inches VWC at full load
immediately  after cleaning after about one week of continuous two horn
operation.  This pressure drop remained the same for the remainder of the two
horn test period (approximately 1 month).  It was felt that steady state
operation had been reached.

     Figure II1-8 shows tube sheet pressure drop versus time for two horn
operation after steady state operation had been reached.  The dashed line in
the Figure represents a corrected pressure drop curve which assumes that the
pressure drop decay would be similar to the curve of the compartment with no
horns operating if sonic horns were installed in all fourteen compartments.

     During baghouse operation without sonic horns, flange to flange press-
ure drop measured between 9 and 9.5 inches VWC or approximately i to 1 inch
VWC higher than the tube sheet differential pressure immediately before the
compartment cleaned.  It is assumed that the flange to flange pressure drop
curve would be similar in shape with the use of sonic horns in all fourteen
compartments to the pressure drop curve without sonic horns.  This would
equate to a flange to flange pressure drop of 6.5 to 7 inches VWC if two
sonic horns were installed in all fourteen compartments.  This represents
approximately a 27 percent decrease in flange to flange pressure drop if the
horns were installed in all fourteen compartments.

     We believe that this is a conservative estimate of performance improve-
ment because of the significant increase in flow through the test compart-
ment.  A portion of the pressure drop through the compartment could be due
to the fact that the test compartment flow was higher than the flow in the
remaining 13 compartments (discussed below).  When sonic horns are installed
on all fourteen compartments, the flow will be more evenly distributed
throughout the baghouse and the pressure drop across the tube sheets should
decrease.

     It is interesting to note that the horns were accidentally operated at

                                     275

-------
 Figure  III - 7  Compartment Tube Sheet Pressure Drop at Pull Load -
                  One Horn Operating
0)

1
CM
o
m
,8

I
 8-
 CD
-p
 o>
CO
I
    10 .
8.
6-
2 .
                 Compartment 11
                            Compartment
                              Clean •—v^
                                                   Compartment
                                        1

                                  Time, Hours


                                 276

-------
FIGURE
8. COMPARTMENT TUBE SHEET
   PRESSURE DROP AT  FULL LOAD
   TWO HORN OPERATION
o:
UJ
0
CO
LLJ
X
O
en
UJ
o:
o_
                Theoretical curve assumes the pressure drop curve
                would be similar in shape to the pressure drop curve
                with no horns operating if the sonic horns were
                installed, on all fourteen compartments.
     IO
     6
CL
O
cr
Q

Ld
or
ID
                    r-COMPARTMENT 4
COMPARTMENT l!
                      234

                   TIME. HOURS
                       277

-------
the wrong time during the two horn test period.  Because of a mistaken ad-
justment in the timing sequence that operated the horns, the horns were oper-
ated during the initial settling period of the compartment clean (no reverse
air).  Pressure drop across the tube sheet in compartment 11 returned to its
original no horn levels after two days of operating in this manner.  With the
timing sequence problem solved, the horns were again operated during the re-
verse air period.  Tube sheet pressure drop returned to 4 inches VWC immedi-
ately after cleaning.

     On March 29, 1982, all four horns in the test compartment were operated
with the unit at full load.  Pressure drop across the tube sheet of the
compartment with sonic horns was 5 inches VWC immediately after cleaning as
opposed to 7 inches VWC in the test compartment without sonic horns immedi-
ately after cleaning.  The significance of this data is somewhat questionable
because at the time the boiler was operating on a different coal than that
which was used in the beginning of the test program.  This new coal has been
known to significantly increase the operating flange to flange pressure drop
of the baghouse and on that day, flange to flange pressure drop was recorded
at 9.5 to 10 inches VWC.  It can be seen from Figure III-9 that the sonic
horns did reduce the  pressure drop across the tube sheet in the test com-
partment when compared to the pressure drop across the tube sheet of the
compartment without sonic horns.

Test Compartment Exit Flows

     Exit flows were measured in the outlet ducts of compartment 4 and 11 us-
ing a S-type pitot tube.  Data was recorded at full load before any horns
were operated and flow through both test compartments was from 23,600 acfm
immediately after cleaning to 15,850 acfm immediately before each compart-
ment cleaned (see Figure 111-10).  A summary of test compartment exit flows
is presented in Table 3 and is also shown in Figures III-ll, 111-12, and
111-13.  It can be seen that flow through the  test compartment with sonic
cleaning did increase significantly and conversely, flow through the test
compartment without sonic horns decreased.  It is assumed that flow through
the remaining compartments in the baghouse also decreased due to the 15.3
percent increase in flow through compartment 11.  The greatest proportion of
flow followed the path of least resistance; through the compartment with
sonic cleaning.  This would explain the rapid increase in tube sheet press-
ure drop noted earlier.  With the installation of sonic horns in all four-
teen (14) compartments, flow will be more evenly distributed and will return
to the original no horn levels.

     Because pressure drop across the tube sheet achieved steady state oper-
ation at an acceptable level with the use of two (2) sonic horns, it was
decided that two (2) horns per compartment would be the minimum amount of
horns required for successful sonic cleaning.  Four (4) sonic horns per
compartment would have the advantage of more rapid removal of the initial
cake buildup., Once the initial cake buildup is removed from the filtering
elements, four horns show no appreciable improvement in compartment perfor-
mance.
                                     278

-------
Table 3 - Test Compartments Exit Flows - Immediately Before and After Cleaning

0 Horns
1 Horns
2 Horns
4 Horns
Immediately After Cleaning
Compartment 11
23,600 acfm
29,900 acfm
27,200 acfm
26,500 acfm
. Compartment 4
23,600 acfm
22,400 acfm
20,800 acfm
20,100 acfm
Immediately Before Cleaning
Compartment 11
15,800 acfm
17,500 acfm
15,800 acfm
15,800 acfm
. Compartment 4
15,800 acfm
17,500 acfm
13,500 acfm
15,800 acfm
                                       279

-------
    Figure III- 9  Compartment Tube  Sheet Pressure Drop at Full Load

                   Four Horns Operating
   10-
I
•8  B
03

!
oT
I  6
m
I
•3
•H
(D


§

(D
o>
.Q
    2-
                                  Time, Hours
                               280

-------
  Figure III- 10  Compartments  l± and 11 Exit Flow vs Time - No Horns Operating
    30,ooo-
o
H
I
    20,000-
    10,000-
Compartment 11
                                     	T~

                                               3

                                      Time, Hours
                                         T

                                          k
~r

 5
  Figure 111-11   Test Compartments  Exit Flows vs Time - One Horn Operating
     30,000-
     20,000-
     10,000-
 o
 o
                  -Compartment 11
                                   T	1—

                                    2            3


                                       Time, Hours
                                                     T

                                                      5
                                     281

-------
   Figure 111-12  Test Compartments Exit Flow vs Time - Two Horn Steady
                  State Operation
     30,ooon
8
     20,000-
     10,000-
o
                    Compartment 11
                                                 Compartment
                                        TIME,  HOURS
   Figure 111-13  Test Compartments Exit Flows vs Time - Four Horns Operating
     30,000n
     20,000-
 8
      10,000-
o
                    •Compartment 11
                                    T	1	
                                    2           3

                                        TIME, HOURS
T~
 k
                                      282

-------
CONCLUSION

     Because of boiler instabilities and resultant load limitations, along
with the high costs associated with operating Induced Draft (ID) fans at high
pressure drops, the need to reduce the operating flange to flange pressure
drop of a FFDC is imperative.   Assuming the 27 percent decrease in flange to
flange pressure drop noted earlier, a savings of approximately $52,400 per
year could be realized (Reference 3).  It is estimated that the total cost of
purchasing and installing two  horns in each of the fourteen baghouse compart-
ments would be $53,000.  With  the I.D. fan cost savings alone, this invest-
ment would pay for itself in one year.  Added savings would be realized in
the elimination of the costs associated with the labor required for periodic-
ally hand clapping the filtering elements to remove the cake buildup.

     The use of sonic air horns as an assist to reverse air cleaning is an
inexpensive and effective means of reducing the pressure drop across the
tube sheet of an operating baghouse.  As a result of the information obtained
from this test program, Public Service Company of Colorado has purchased and
installed two sonic air horns  in each of the fourteen compartments on
Arapahoe Unit #3 FFDC.  Preliminary test results indicate a significant re-
duction in flange to flange pressure drop.

     A test program has been initiated that will not only determine the net
reduction in flange to flange  pressure drop of Arapahoe Unit #3 FFDC, but
will determine if higher operating air-to-cloth ratios are obtainable with
the use of sonic cleaning.  If higher air-to-cloth ratios are achievable
while still maintaining acceptable pressure drops, significant capital sav-
ings could be realized on new  baghouse installations through the reduction
of the number of compartments  and/or total filtering area required for the
successful operation of a FFDC.

               The work described in this paper was not funded
               by the U.S. Environmental Protection Agency and
               therefore the contents do not necessarily reflect
               the views of the Agency and no official endorsement
               should be inferred.
                                     283

-------
                                 REFERENCES
1.  "Nodular Deposits in Fabric Filters", Robert F.  Lembach and Gaylord W.
    Penny, Journal of the Air Pollution Control  Association, August, 1979.

2.  Letter, Ronald F. Ross, Joy Industrial Equipment Company, to Gordon
    Schott, P. S.  Co., April 8, 1980.

3.  "Evaluating Baghouse Systems for Energy Efficiency", Plant Energy
    Management, February, 1982.
                                     284

-------
                            APPENDIX A

                             TEST LOG



 Date

 2/22    Load 46 MW.   Final setting up of horns.  Obtained base data.

 2/23    Load 45 MW.   Operated 1 horn in NE corner of compartment.
        Dropped 3 inches inAP.

 2/24    Load 45 MW.   Continued 1 horn test.  Pulled ash and timed.

 2/25    Load 46 MW.   Operated 1 horn NE corner.  At  2:30 p.m. 2 horns
        operated NE  and SW corners.  Dropped 1 inch inAP.

 2/26    Load 46 MW.   Continued 2 horn test.  Pulled ash and timed.

 3/1     Load variable.   Compartment isolated took bag weights.  Com-
        partment back in service.  Continue with 2 horn test (NE and
        SW corners)  for next two weeks.

 3/2     Load variable.   Continued 2 horn test.

 3/3     Load variable.   Continued 2 horn test.

 3/4     Load variable.   Continued 2 horn test.

 3/5     Load variable.   Continued 2 horn test.

 3/8     Load 46 MW.   Continued 2 horn test.  Pulled ash and timed.

 3/9     Load variable.   Continued 2 horn test.

 3/10    Load variable.   Continued 2 horn test.

 3/11    Load variable.   Inspected compartment 4 for bag leaks.   Found
        nothing  wrong.   Stood  in compartment 4 while the horns operated
        in  compartment  11  to see if there was any sound attenuation to
        explain  the  decrease in thimble  sheet AP in |4.  Felt no excess
        vibration while horns  were operating.

3/15    Load 46  MW.   Two horn  test continues.  Pulled ash and timed.
        Switched out wrong Agastat so horns were operating before reverse
        air.  Horns  had no effect on tube sheetAP.   Agastats fixed.

3/16    Load variable.   Continued 2 horn test.

3/17    Load variable.   Continued 2 horn test.  Took bag samples in 4
        and  11.  Weighed bags  in both compartments.   All data taken before
        cleaning.
                                  285

-------
3/18    Load variable.   2 horns operating.  Cut samples and weighed
        bags in compartment 11 after cleaning.  Took coal sample.

3/19    Load variable.   2 horns operating.

3/22    Load 46 MW.  2 horns operating.  Pulled ash and timed.

2/23    Load variable.   Continued 2 horn test.

3/24    Load variable.   Continued 2 horn test.

3/25    Load variable.   Continued 2 horn test.

3/26    Load variable.   Continued 2 horn test.

3/29    Load 46 MW.  Continued 2 horn test.  At approximately 1:00 p,
        operating 4 horns.

3/30    Load variable.   Continued 4 horn test.

3/31    Load variable.   Continued 4 horn test.

4/1     Load variable.   Continued 4 horn test.

4/2     Load variable.   Continued 4 horn test.

4/5     Load 46 MW.  Continued 4 horn test.

4/6     Load variable.   Continued 4 horn test.

4/7     Horns shut off at 7:00 a.m.  Weighed bags in compartment 11.
        No leaks found in compartment 11.

4/8     Load variable.  No horns.

4/9     Load variable.  No horns.

4/10    Load 46 MW.  No horns.  Pulled ash and timed.
                                286

-------
      'ELECTROSTATIC' STIMULATION OF REVERSE-AIR^CLEANED  FABRIC  FILTERS"

                by:   D.  A.  Furlong and  G,  P.  Greiner
                     ETS,  Inc.
                     Suite C-103, 3140  Chaparral  Dr.,  SW
                     Roanoke, VA 24018

                     D.  W.  Van  Osdell
                     Research Triangle  Institute
                     Research Triangle  Park,  NC 27709

                     L.  S.  Hovis
                     U.  S.  Environmental Protection Agency
                     Industrial Environmental Research Laboratory
                     Research Triangle  Park,  NC 27711
                                ABSTRACT

     The concept of electrostatic stimulation of fabric filtration (ESFF)
has been investigated on a slipstream of a pulverized-coal-fired boiler
using reverse-air-cleaned, woven-fiberglass fiber bags.  Operation was
demonstrated using ESFF at a gas-to-cloth ratio (G/C)  of 6 ft/min.   An
un-electrified control house was simultaneously operated at a G/C of
3 ft/min.  Under these conditions, the ESFF house maintained a pressure
drop equal to or less than the control baghouse.  In addition to reducing
the filter cake pressure drop, ESFF was observed to apparently have long
term benefits in preventing irremovable dust buildup in the fabric.

     This paper has been reviewed in accordance with the U. S. Environ-
mental Protection Agency's peer and administrative review policies and
approved for presentation and publication.
  Readers more familiar with metric units may multiply ft/min by 0.305 for
  the equivalent m/min.
                                   287

-------
                               INTRODUCTION

     This paper describes development activities from May 1981 to March
1982 using a pilot scale, slipstream fabric filter employing electro-
static stimulation of fabric filtration (ESFF).

     The ESFF concept is based on laboratory scale tests at the Textile
Research Institute (TRI) as reported by Lamb and Costanza.     The TRI
concept of ESFF consists of applying an electric field parallel to the
surface of the fabric through use of electrodes  located at the upstream
surface of the filter fabric.  Figure 1 is a schematic of a TRI electrode
"harness" located on the outside of a pulse cleaned bag and electrically
connected so that alternate electrode wires are at high electrical poten-
tial, producing an electric field of 2 to 4 kV/cm.

     A team composed of Research Triangle Institute (RTI), ETS, Inc., and
TRI, under contract to the U.S. EPA, is evaluating this concept at pilot
scale.  The pilot scale program was started in October 1979.  The first
year's primary objective was to verify the TRI laboratory results under
actual flue gas conditions.  EPA, as manager, and RTI, as prime contractor,
have directed the technical effort and performed initial electrical hard-
ware development and construction.  ETS, Inc. has designed, built, installed,
and is operating the pilot unit.  ETS and RTI are further developing the
ESFF hardware.  The pilot plant is installed on a slipstream of a pulverized-
coal boiler at the E. I. DuPont de Nemours & Company plant, Waynesboro,
Virginia.

     The pilot unit was designed to operate in either the pulse-jet (outside
collection) or reverse air (inside collection) cleaning mode.  The baghouses
were first operated as pulse-jet collectors.- The principal conclusions of
the pulse-jet work, as reported previously,   '   were:

          1)  At any given face velocity, the pilot ESFF baghouse had a
              reduced residual pressure drop and a reduced rate of pressure
              drop increase when compared to the pilot conventional bag-
              house .

          2)  The pilot ESFF baghouse could be operated in a stable
              fashion at face velocities up to about twice the stable
              operating conditions for the pilot conventional baghouse.

          3)  A reduction in total annualized cost of about 30 percent
              was estimated for an industrial boiler application utilizing
              an ESFF baghouse at twice the conventional face velocity.

          4)  Particulate emissions from the two baghouses were not signi-
              ficantly different.

          5)  The electrode configuration developed in the course of the
              study, a combination pulse-jet cage/electrode, located on
              the downstream side of the* filter fabric, was feasible and
              had potential for commercial use.

                                    288

-------
     In May 1981, the pilot plant was converted to reverse-air cleaning.
This paper summarizes the development activities and reports preliminary
results and conclusions based on the work from May 1981 to March 1982.
THE PILOT SYSTEM
     The pilot system provides the capability of simultaneous monitoring
of the ESFF effect through use of two identical baghouses receiving their
inputs from a common slipstream.  One baghouse is supplemented with electri-
cal enhancement equipment; the other is not.

     Figure 2 is an isometric drawing of the pilot unit, and Figure 3 is
the system schematic.  Each baghouse can accept up to 9 reverse-air cleaned
bags each 20.3 cm (8 in.) in diameter and 224 cm (8 ft) long, or up to
13 pulse cleaned bags, 11.7 cm (4-5/8 in.) in diameter and 224 cm (8 ft)
long.

     The provision of parallel control and experimental units is an impor-
tant feature of the pilot plant in that it permits operation in either of
two modes, parallel or separate.  In parallel operation, both baghouses have
the same inlets, compensating for the cumulative effect of varying inlet
conditions.  Any differences between the two units, either intentional or
unintentional, can be evaluated by parallel operation.  Future work is
planned where separate tests will be conducted in each baghouse after
sufficient experience is in hand to negate the further need of a "control"
baghouse.

     The pilot unit is installed on a slipstream from an industrial boiler
baghouse.  Four pulverized-coal boilers are used with a normal load of
about 160,000 kg/hr (350,000 Ib/hr) of steam.  The coal fed to the boilers
changes frequently, with sulfur content varying from less than 1 percent to
about 2 percent, and ash content ranging from 5 to 15 percent.  The boilers
are sometimes co-fired with No. 2 or No. 6 fuel oil.  Inlet dust loadings
at the pilot plant have varied from 0.6 to 5 g/scm (0.25 to 2 gr/scf);
0.7 g/scm (0.3 gr/scf) is typical.  The mass mean diameter of dust particles
averaged 6.3 /um.
ESFF TEST HARDWARE
     The electrical power requirements of the ESFF system are very low.
The power consumption measured by TRI and confirmed in the pilot system is
about 1 W/m^ (0.1 W/ft ) of cloth.  This ESFF power requirement is equiva-
lent to approximately 0.1 in. of water pressure drop at a G/C of 6 ft/min.

     The electrical system used in the pilot unit consisted of a variable
DC power supply (0 to 20 kV at 5 mA) for each of five power supply networks,
Current and primary voltage were measured separately in each network.


                                    289

-------
Current limiting and meter protection circuits were used.

     The electrode designs utilized to date have been based on the combi-
nation cage/electrode developed during the pulse-jet work.  Again, these
cages were placed on the downstream (clean) side of the fabric, so as to
not restrict cleaning of the bag and to avoid dirt buildup on the cage.
These cages (patterned after the pulse-jet cage/electrode, constructed of
welded 1/8-in. (0.318 cm) rods, and referred to as "rigid cages") exhibited
two problems:  first, fabric wear was observed at the insulator locations;
and second, the wire spacing of 2.5 cm coupled with insulator limitations
restricted field strength to about 4 kV/cm.  New cages were fabricated
using 34 1/8-in. diameter electrodes (1.68 cm spacing).  These electrodes
were held in place by woven fiberglass straps having sewn-in pockets.
These cages are referred to as flexible cages, since the electrode/strap
assembly is quite flexible prior to installation on the bags.

     Fiberglass bags were used for the entire reverse-air test series.  The
bags were 20.3 cm (8 in.) in diameter and 224 cm (8 ft) long; no anti-
collapse rings were used.  These bags were made of J. P. Stevens Style
No. 648, 17 oz (482 g) texturized fiberglass fabric with a 10 percent
Teflon B finish.  Each bag was tensioned to 40 Ib (18 kg) using a conven-
tional tensioning spring.
RESULTS AND OBSERVATIONS
     Electrostatic stimulation of fabric filtration has two primary e.ffects
on filter behavior when compared to conventional filtration at the same G/C:
(1)  the average pressure drop is reduced through a combination of lower
residual pressure drop and a reduced pressure drop rise during a filtering
cycle; and (2)  the ESFF baghouse can operate at a higher stable G/C.

     To define and compare the effects on pressure drop rise during a
filtering cycle, a figure of merit called PDR (pressure drop ratio) is
used.  PDR compares the beginning and ending Ap across one cleaning cycle
using electrostatic enhancement with the comparable AP change without
electrostatic enhancement.

                                (AP    - A P   )  ESFF
     PDR is defined as:                      r
                                (Apf   - Ap^  ) control


whereAP is the pressure drop across the bags,

     f refers to the final state (just prior to cleaning), and

     r refers to the residual state (just after cleaning).
                                    290

-------
     For an idealized filter cycle, in which the bag pressure drop increases
linearly with time and for constant dust loading from AP  to AP , it can
be shown that:                                           r

                       (K )  ESFF
              PDR =      *•
                       (K )  control

where K_ is the specific cake resistance.

     PDR, then, is the ratio of the flow resistance of a dust cake collected
by an ESFF system to that collected by a conventional fabric filter.

     Figure 4 presents typical pressure drop recordings for both the con-
ventional and ESFF pilot baghouses in reverse-air operation.  The rate of
increase of pressure drop in the ESFF baghouse when compared to the con-
ventional pilot baghouse is evident.  PDR can be seen to vary depending on
the length of the filter cycle, and thus,.is most useful for comparison of
two baghouses operating with the same cleaning cycle.  As the two baghouses
were operating at the same face velocity when these data were obtained, the
ESFF baghouse also has a reduced residual pressure drop.  Thus, the ESFF
effect is to produce a lower average pressure drop when operated at the
same filtering velocity.

     The reverse-air baghouse operation was started up with a break-in
period of about 1 week at a G/C of 2 ft/min, with no electrical field in
the ESFF baghouse.  During this period, the two baghouses exhibited essen-
tially identical pressure drop characteristics, indicating that the two
baghouses could be analyzed as identical baghouses in parallel.  The electri-
cal field was then put on the bags, and was left on almost continuously for
the next several months.

     Figures 5 and 6 present a chronological summary of the results during
May and June 1981, using reverse-air bag cleaning.  Field strength and
gas-to-cloth ratio were independent variables.  The PDR and average pressure
drop were measured for each filtering cycle and averaged for the day.  The
gas-to-cloth ratio was progressively increased from 2 to 6 ft/min for the
ESFF baghouse; however, the control house could not operate stably above a
G/C of 4.5 ft/min.  Note also that on May 12 and 13, when the electric field
was removed, the average AP of the ESFF house rapidly approached that of
the control house.  Note particularly, that from June 10 to June 30, a 21-
day period of continuous 24-hour operation, the ESFF house operated at a
G/C of 6 ft/min, with an average AP between 6.4 and 7.1 in. (16.3 and 18.0
cm) of water, while the control house operated at a G/C of only 3 ft/min
and yet had an average AP of 7.4 to 8.3 in. (18.8 to 21.1 cm) of water.
Stated another way, ESFF allowed a doubling of the velocity without a major
change in pressure drop.

     Figure 6 demonstrates the ability of an ESFF baghouse to operate at a
higher stable G/C.  Note that from June 9 to 13, the control baghouse demon-
strated inability to operate stably at a G/C of 4.5 ft/min; subsequently,

                                    291

-------
it was reduced to a G/C of 3 ft/min.  Simultaneously, the ESFF baghouse
operated at a G/C of 6 ft/min from June 10 to 30.

     Based on these encouraging results, a series of tests were undertaken
to evaluate the effects of field strength and gas-to-cloth ratio on pressure
drop characteristics.

     Figure 7 summarizes the evaluation of field strengths of 3 and 4 kV/cm
on average Ap over a G/C ratio range of 1 to 6 ft/min.  This test series
used the rigid cages; hence, fields above 4 kV/cm were not possible.  These
data show graphically the benefit derived from the ESFF effect.  The con-
ventional baghouse had a higher averageAp where it was capable of stable
operation.  Increasing the electric field from 3 to 4 kV/cm does not appear
to offer an advantage.

     By the end of this test series, the disadvantages mentioned previously
for the rigid cage had become evident, and it was necessary to install new
bags in the ESFF pilot baghouse.  Figure 8 is a chronological summary of
testing with new bags and the new flexible cages at field strengths up to
6 kV/cm.  After a break-in period at a G/C of 2 ft/min on July 26 through
28, the G/C was increased to 4 ft/min and an electric field of 4 kV/cm
resulted in an immediate decrease in the average Ap compared to the control
house.  Figure 9 summarizes the results of this initial approximately 1
month test period with the flexible cages.  These results indicate two
things.  First, little or no benefit was derived from field strengths above
3 kV/cm.  Second, note the results at 0 kV/cm.  As the ESFF bags were rela-
tively new while the conventional bags had been in use since the beginning
of the test program, some difference in theAp response of the two bag sets
at zero field was expected.  However, the magnitude of the difference was
surprising, and suggested that something other than the electric field
might be causing at least a portion of theAp advantage being shown by the
ESFF baghouse.  An extensive investigation of the situation was undertaken,
and several factors which served to exaggerate the advantage of the ESFF
baghouse were identified.  A concise explanation is still in the process of
being defined; however, three contributing features stand out so far:

     (1)  As is well known, reverse-air baghouses are extremely sensitive
          to cleaning parameters.  Although efforts had been made to favor
          the control house by cleaning it last, this, in fact, may have
          favored the ESFF house, possibly by giving it a pre-coat after
          cleaning.  In addition, a slight leak below the tubesheet in the
          control baghouse, balanced by a similar sized leak above the tube-
          sheet in the ESFF baghouse, raised the effective gas-to-cloth
          ratio in the conventional baghouse.  These factors were quantified
          by experimentation, and their mathematical inclusion in the data
          analysis indicated that the ESFF baghouse at zero field still had
          a lower average AP than did the conventional baghouse.  The same
          corrections reduce the size of the advantage which the data had
          indicated for the ESFF baghouse at the various field strengths.

     (2)  In May, as shown in Figure 5, the field was removed after 2 weeks


                                     292

-------
          of operation,'then the APS at zero field strength were essen-
          tially equal.  In September, after 2 months of operation,  the
          ESFF bags are better even with no field.   This certainly sug-
          gests the possibility of a long term, cumulative effect of the
          electric field by reducing the residual A? by avoiding the
          buildup of un-removable dust accumulations deep in the fabric
          weave.

     (3)   During the investigation into the zero field behavior, the ESFF
          bags were operated for approximately 2 months without the  electric
          field.  It was noted that, after this zero field operation, the
          low residual Ap results which were obtained during the early
          operation were no longer observed.  The hypothesis is that the
          long term lack of an electric field allows the fabric to build an
          interstitial dust layer which cannot be removed by normal  means,
          leading to the increased residualAPS.  A return to new fabric
          will try to verify this observation.
CONCLUSIONS
     Although the evaluation of electrostatic augmentation of reverse-air-
cleaned bags is not as yet complete,  the observations to date of primary
significance are:

     (1)  Starting with new fabrics,  ESFF operation for an extended period
          (21 days) at a G/C of twice (6 vs.  3 ft/min) that of the control
          house was demonstrated at comparable or lower pressure drops
          (Figure 6).

     (2)  Interruption of the electric field  for periods of 1 to 2 days
          did not eliminate most of the benefits of ESFF.  Removal of the
          field for several months did diminish the benefits of ESFF upon
          reapplication of the field.

     (3)  ESFF appears to maintain the fabric in a "near new" condition
          relative to  entrapped particles within the fabric.  It is, of
          course, these entrapped particles that result in the residual
          Ap after cleaning.  This observation is consistent with earlier
          work, indicating that under ESFF conditions more particulate is
          captured near the upstream side of  the fabric, and penetration
          into the fabric is reduced.  Thus ESFF prevents, or at least
          significantly retards, the buildup  in residual
FUTURE WORK
     Both the rigid and the flexible cages are recognized to have practical
and economic limitations for full-scale filter use; hence, electrode design
                                    293

-------
is continuing.  The most promising concept at this time for reverse-air-
cleaned applications appears to be a "woven" electrode, although this has
not as yet been tested in the pilot system.  Woven electrodes consist of
conductive yars woven into the warp of the fabric at the desired electrode
spacing.  Yarns consisting of stainless steel filaments, graphite filaments,
or a graphitized high temperature nylon are now being investigated.   These
conductive yarns, and the techniques for weaving them, have been previously
used to produce anti-static fabrics for both filtration applications and
for carpets.  Woven electrodes in both glass and Teflon fabrics are  being
investigated.
ACKNOWLEDGEMENTS
     This work was supported by EPA Contract 68-02-3186 from Industrial
Environmental Research Laboratory, Research Triangle Park, NC.   The assis-
tance provided by E. I. duPont de Nemours and Company, Inc. is  gratefully
acknowledged.  JChe pilot unit was located at their Waynesboro,  Virginia,
plant.  Teflon^fabric was donated to the test program by DuPont.
                                     294

-------
                                REFERENCES

1.   Lamb,  G.  E.  R.,  and  P.  A.  Costanza.  A Low-Energy Electrified Filter
    System, Filtration and  Separation  17:319, 1980.

2.   Van Osdell,  D. W., G. P. Greiner,  G. Lamb, and L. S. Hovis, "Electro-
    static Augmentation  of  Fabric Filtration", In proceedings of the Third
    Symposium on the Transfer  and Utilization of Particulate Control Techno-
    logy,  Volume I.   Control of  Emissions from Coal-Fired Boilers,
    EPA-600/9-82-005a, July 1982.

3.   Greiner,  G.  P.,  D. A. Furlong, D.  W. Van Osdell, and L. S. Hovis.
    "Electrostatic Stimulation of Fabric Filtration", JAPCA, 31:1125-1130,
    1981 (presented  at the  74th  Annual Meeting of the APCA, Philadelphia,
    Pennsylvania, June 1981).
                                    295

-------
                                        TO POWER SUPPLY
             Hh
             1.5 cm
HIGH VOLTAGE HARNESS
                        SEPARATE, INSULATED
                        BAG SUPPORT "CAGE"
                          :IBERGLASS YARN
                         -ELECTRODES	
                          STAINLESS STEEL
                          WIRE-0.58 mm
HARNESS INSTALLED ON BAG
                      Figure 1

                    TRI  HARNESS
   Conventional Baghoute
                    Figure 2

                  PILOT  PLANT
                          296

-------
                                               OUTLET (TOP)
TRAILER
( .. 0 J
v x „ s KP J-
V ~ , r *

TEST PORTS

                 Figure  3



             SYSTEM SCHEMATIC
                1:00      12:00      1:00     10:00
9:00    8:00 p.m.
                    Figure 4



REVERSE AIR PILOT  BAGHOUSE PRESSURE  DROP TRACE
                         297

-------
cc
Q
Q.
 o>

 2
1.0


0.8


0.6


0.4


0.2


  0


  8
ll 6
itS 4

'*Z
o   o
.2
ui

     0
  8--
 raX
 0°
it
                                  Conventional
                                                      \
                          11          16
                                Date:  May 1981
                                                     21
26
	j

 31
                                    Figure 5


                     ESFF CHRONOLOGICAL SUMMARY, MAY 1981
                                     298

-------
 cc
 Q
 Q.
OJ
 1.0



 0.8


 0.6



 0.4



 0.2



  0


  8



  6



:  4



  2


  0



 12
  £U>
 _ CM  p
 |d  8





II6
      2


      0


      8
o
to

UJ
                                           ESFF
                                       \
                                            Conventional

                               11          16           21
                                     Date: June 1981
                                                                26
_J

 31
                                      Figure  6


                      ESFF CHRONOLOGICAL SUMMARY, JUNE  1981
                                      299

-------
   10
    8
a.
in
Csj
d

~   6
c

d  5
2
o

£


1   4
a.
0)
4  3
                                                                           ESFF
                                                                          3 kV/cm
                                                      I
                              234

                          Gas-to-Cloth Ratio, ft/min (= 0.5 cm/s)
                                      Figure 7


                             DELTA  P AVG. VS G/C  RATIO

                                     RIGID  CAGE

                                APRIL 24 TO JULY  15
                                    300

-------
  1.0


  0.8
cO.6
Q
a.
  0.4
  0.2
I   8
£
«S  6
•8°
    4

    2

    0
I
Hi
                                                                         \
!l8
£»
I2
*a«
s°*
.s
<'~
    0

,2< 8
SE
ceo
x:10. 6

£4

II 2
     25
                           Conventional
                                                                   v
                                      I
        July 1981
                31
10         15
       August 1981
    Date
20
25
30
                                      Figure 8

                ESFF CHRONOLOGICAL SUMMARY,  JULY 25 - AUGUST 31, 1981
                                         301

-------
    8
£
O
 CM
d  4
o
£
Q.

O
O)





I
                                                     Conventional

                                                      Baghouse
ESFF Baghouse

     Strength


    0 kV/cm

    6 kV/cm


    2 and 4 kV/cm



    3 kV/cm
                   JL
  _L
                    234


                Gas-to-Cloth Ratio, ft/min (= 0.5 cm/s)
                            Figure 9



                  DELTA P AVG. VS G/C  RATIO

                         FLEXIBLE CAGE

                      JULY 27 TO SEPT.  21
                            302

-------
              ELECTRICAL STIMULATION OF FABRIC FILTRATION;
                  ENHANCEMENT BY PARTICLE PRECHARGING

              by:  George E. R. Lamb, Richard I. Jones and William B. Lee
                   Textile Research Institute
                   Princeton, New Jersey 08540
                                 ABSTRACT
     The reductions in pressure drop that accompany the establishment of a
strong electric field near a filter fabric appear to be due to three
separate mechanisms.  One is the formation of a more porous dust cake due
to dust capture in the low packing density regions of the fabric.  A
second mechanism is attraction of particles to the bag wall which causes
the bag to act like a precipitator.  The thickness of dust cake is then
greater near the entrance than at the end of the bag, and this results in
a lower pressure drop.  The third mechanism involves attraction of parti-
cles to the bag electrodes.  The dust is then deposited in bands with
relatively thin deposits in between.  Measurements and visual inspections
of the dust deposits indicate that the second and third effects are en-
hanced when the aerosol is charged.  A particle charger of new design
appears to be particularly suitable for this purpose, and is found to
cause major changes in filtration performance.

               This paper has been reviewed in accordance with
               the U.S. Environmental Protection Agency's peer
               and administrative review policies and approved
               for presentation and publication.
                                   303

-------
                               INTRODUCTION

     A number of  investigations have been directed to the development  of  sys-
 tems for electrically stimulated fabric filtration  (ESFF).  Some have taken
 the approach of establishing an electric field near the filter  (1-4).  Others
 have sought to improve performance by adding charges to the aerosol particles
 (5-7).  In a few.  cases, both effects have been combined (8,9).

     When electric fields are applied in the vicinity of the filter,  perfor-
 mance is improved by virtue of the increased capture efficiency of single
 fibers in the presence of a field (10).  This causes the dust cake to form
 closer to the surface of the fabric.  The cake is then more easily removed
 when the filter is cleaned, and less dust is retained (10).  If, in addition,
 the upstream surface of the filter has a nap, the dust cake will form in the
 nap and be more porous.   (The nap is the fuzzy region on the surface  of  some
 fabrics where the fiber volume fraction is smaller  than in the  interior.)
 Both these effects will allow the filter to run at lower pressure drop than
 without the electric field.  Penetration is usually reduced by ESFF.   This,
 intuitively, is also an expected result of the better single fiber efficiency,
 but an indirect mechanism may involve reduced seepage because of the  upstream
 shift of the dust deposits.

     The results in Table 1 illustrate these effects.  They were obtained
 with a series of bags made of Teflon  felts with increasing degrees of nap.
 It can be seen that,as the nap becomes more pronounced, not only overall pres-
 sure drop levels are reduced, but  also  pressure drop ratios (PDR).   If  APf
 is the pressure drop just before a filter is cleaned, and AP. the pressure
 drop just after, then the PDR is defined as the ratio of (AP--AP.) when  the
 electric field is applied to the corresponding value with no field.   The
 table also shows that the field reduces penetration.

 TABLE 1.  EFFECT OF SURFACE LOW-DENSITY LAYER ON PRESSURE DROP REDUCTION
Av.
field
(kV/cm)
0
2
4
. .,




Control
AP.,APf
(mm H20)
7,23
6,16
7,15
PDR
	
0.63
0.50
Eff
98.
98.
98.
•
03
80
88
fi*
(mm
7,
7,
5,



Low nap
H26)
22
15
12
Eff.
PDR (%)
	 98.04
0.53 98.93
0.47 99.13
APi>
(mm
5,
3,
3,



High nap
H26)
17
9
4
Eff
PDR (%)
	 96.
0.50 98.
0.08 98.
•
83
51
76
     Charging the aerosol upstream of the bag has also been found to result
in lower pressure drops and lower penetrations, but few workers have taken
account of the loss of dust which occurs either in the precharger or in the
duct from precharger to baghouse.  In such cases, it is difficult to assess
the effects of charging on bag performance.  One study (7), in which the dust
collected by the bag was weighed, showed no difference in specific cake resis-
tance with precharging.  The lower pressure drop was due to reduced dust
loading.
                                    304

-------
                              EXPERIMENTAL

     In order to study the effects of precharging without interference from
reduced dust loading, the charger design shown in Figure 1 was adopted.










AE




r
i
(-
h
1-
i
hi

ioSOL
I









T
-s1
PJ










Yr-










— »•-

ESFF
POWER
SUPPLY







PRECHARC
POWEF
SUPPL
                          iN
Figure 1.  Location of "lightning rod" precharger in reverse air baghouse.

The charger, a metal rod provided with wire bristles at the upper end,was
positioned by means of an insulated support, so that its tip was just below
the entrance to the bag in a one-bag laboratory baghouse.  With this ar-
rangement it was expected that all dust would enter the bag because of the
high air velocity at the bag entrance.  The bags used were provided with
electrodes on the upstream surfaces.
                                                 ®
     Table 2 lists results obtained with a Teflon  felt bag provided with
so-called "printed" electrodes, which in this case consisted of narrow
bands of a conducting elastomer composition.  The use of "printed" elec-
trodes is a technique undergoing trial.  If successful, it should permit a
length of fabric to be fitted with an electrode pattern more easily and
cheaply than would be the case with wire electrodes.  The results in Table
2 may be taken as an indication that such electrodes can provide electrical
stimulation as effectively as wires.  The question to be examined in the
future is one of durability.

     The results in Table 2 show that,when aerosol precharging is combined
with an electrified bag,  reductions in PDR of almost an order of magnitude
can be obtained.  It is clear that, when bag electrodes and precharger are
raised to potentials of opposite signs, the pressure drop reductions are
greater than when the signs are the same.   It can also be seen that; with
this bag, activating the precharger and not the bag electrodes causes only a
minor reduction in pressure drop.

                                   305

-------
TABLE  2.  PRESSURE DROP RATIOS WITH ESFF AND PRECHARGED AEROSOL
Teflon®
Bag field
(kV/cm)
0
0
+2
+2
-2
-2
-3
-3
+2
-2
>
bag with printed electrodes. Inlet concentration 4.8
Face velocity 6 ft/min (3 cm/s).
Precharger
voltage (kV)
0
-9
0
-9
0
-9
0
-9
+9
+9
AP±, AP
16,110
16,107
15,83
14,28
15,60
15,41
15,52
15,35
16,39
17,31
g/m3.
PDR
0.97
0.72
0.15
0.48
0.28
0.39
0.21
0.24
0.15
     Figure 2 shows results  obtained with woven glass bags having sewn-in
150 ym bare copper wire electrodes.   It can be seen that, as  before, without
a potential on the precharger,  the  reductions in the pressure drop are mod-
est.  With the precharger activated, PDR values fall considerably lower.

                                         Precharger
                                         Voltage
               a:
               o
               a.
                   1.0
                   0.8
                   0.6
                   0.4
                  O.E
                               2.3    4
                            AVERAGE FIELD (kV/cm)
                                                          	3 cm/s
                                                          	1.5 cm/s
Figure 2.  Dependence  of  PDR on bag field after extended run.
           no precharger.
                                                                 Top curves
                                     306

-------
     In order to study the mechanism by which precharging reduces pressure
drop, the bag support was modified to allow the bag to be removed easily
from the baghouse.   The sewn seam of the bag was replaced with an overlap
held closed by clamps.  With this arrangement, the bag(after being removed
from the baghouse)  could be carefully opened and the dust cake could be
examined.  Visual inspection immediately showed that nonuniforin distribu-
tion of the dust deposit occurred both in the longitudinal and in the tang-
ential direction.  Deposition was heavier at the lower end of the bag and
on the electrodes,  as shown in Figure 3.
Figure 3.  Dust deposition pattern with 4 kV/cm field and -15 kV precharger
           potential.

     Mass distribution with respect to bag height was measured by placing a
mask at a series of positions over the opened bag and aspirating the dust
from the area within the mask.  The dust was collected on a filter and
weighed.  This was done for various combinations of bag fields and pre-
                                   307

-------
 changer potentials.  The results are plotted in Figure 4:  in all cases
 more dust collects at the bottom than at the top of the bag.  When this
 occurs with no voltage applied, the cause is assumed to be settling by
 gravity; but with an applied field, the tendency for early deposition is
 increased.  Values of PDR are listed alongside the curves.  It should be
 noted that these were values obtained from the pressure drop reached
 after 15 minutes, starting from an almost clean bag.  The bag was
 cleaned by vacuuming each time so as to allow measurement of the dust
 deposit areal densities plotted in Figure 4.  The PDRs in Figure 4 must
 therefore not be regarded as typical of values that would be obtained after
 prolonged conditioning.   They nevertheless bear a certain relationship
 to the curves.  It can be seen that curves with greater slopes tend to cor-
                   <8
                   16
               (O
               Q
               fe~:  «°
               El
               in g
               5i   8

                         20    40    60    80 -   100  120
                          DISTANCE FROM BOTTOM OF BAG (cm)
 Figure 4.  Distribution of dust mass with respect to distance from  the
            bottom of the bag.  B and P indicate average field on bag  in
            kV/cm and voltage on precharger in kV, respectively.

respond to lower PDR.  Deviations from this trend must be attributed to
nonuniformity in distribution in a tangential direction, such as shown in
Figure 3,  which is a photograph of the opened bag after being run 15 minutes
with an average +4 kV/cm field between wires and a -15 kV potential on the
precharger (44,-15).  It can be seen that: (1) Most of the dust collects on
the positive wires.  (2)' The amount collected, as plotted in Figure 4, Is
greatest at the bottom and diminishes with distance up the bag. (3) Some
                                    308

-------
collection occurs on the ground wires.  This may be dust that, having depos-
ited on the positive wires,  is reentrained after gaining a positive charge.
(4) Spaces between wires are left essentially clean.

     Photographs taken of  the bag surface and dust deposits are shown in
Figures 5, 6, and 7.  In all these photographs, pieces of adhesive tape had
been stuck to the clean bag  over portions of some of  the electrodes.  The
patterns of dust deposition  on the tape lead to some  interesting conclusions
about  the motion of dust particles during filtration, as discussed below.
Figure 5 shows  the bag surface after vacuuming  (a) and with a dust cake laid
at no applied voltage (b).   No dust has collected on  the pieces of adhesive
tape because  there is no air flow normal to the tape.  Figure 6 shows dust
deposited with  a 4 kV/cm field between bag electrodes and +15 kV  (a) and -15
kV  (b) on the precharger.   In both cases the dust forms visible bands on the
wires, but the  bands are sharper when bag voltage and precharger voltage are
of opposite sign.  An interesting feature of this figure is that dust was
deposited on  the adhesive  tape in an amount apparently equal  to that else-
where  on the band.  This indicates that particle velocities due to the field
near the wires  are much greater than the gas velocity.  The bag is thus acting
.like a precipitator.  Figure 7 shows the appearance of the dust deposits
when only a bag potential  was applied  (a).  Here there are still visible
bands, but they are more diffuse and dust is again deposited  on both the
electrodes and  the adhesive tapes.  When only  the precharger was activated
 (b and c) , the  dust deposit was uniform, and there was some deposition on
the tapes, presumably of dust driven to the wall by space charge effects.

     After each of the 15-minute runs  specified in Figure 4,  the bag was
weighed and the mass of dust determined.  Table 3 shows  that  no matter what
 the combination of bag and  charger potential,  the mass collected in 15 mi-
nutes was always  the  same.  Thus the changes  in pressure drop were not to
any extent due  to reduced  total dust loading.   This appears to disagree with
.the data  in Figure 4.   In  that figure  the  total mass  of  dust  should be pro-
portional  to  the  area under each curve, and  the lower curves  appear to enclose
 smaller areas.  The  explanation must be that the  sharp upswings  to  the left
 of  the lower  curves were  plotted with  too  few points  to  show  accurately  the
 amount in  those regions.   Figure  3  shows  that heavy  deposition occurs with-
 in a  few centimeters  of  the bottom of  the bag; the  telephone  (12 cm high)
 to the right  of the  photograph  gives  an idea of scale.

 TABLE  3.   MASS  OF DUST COLLECTED ON BAG IN  15  MINUTES

 Bag field                     Precharger                      Mass of  dust
  (kV/cm)                        voltage                       collected  (g)
                                                               in two  trials
     0                             0                             30, 31
    +4                             0                             33, 31
    +4                           -15                             27, 29
    +4                           +15                             37, -
     0                           -15                             35, 27
     0                           +15                             32, 31
                                     309

-------
       !
CLEAN I •

          Figure 5.  Appearance  of  the  inner  surface  of  the bag:
            a)   after vacuum cleaning through a wire  screen; and
            b)   with an average  of  7.5  mg/cm2 of  fly  ash
                deposited with no electric  field  or precharging.
            No  fly ash has deposited on the tape  in either case.
            In  Figures 5, 6, and 7, horizontal bands  are due to
            anticollapse rings inside the bag.
                                 310

-------
v..   .;'.&   fc  ''
                                      ^•'.•TO* ' Vv'-.-'vi '      ''-V '
                                      --*•              ''
          Figure 6.  Appearance of the inner surface of  the bag:
            a)   after deposition with 4 kV/cm between bag
                 electrodes and +15 kV on the precharger; and
            b)   after deposition with 4 kV/cm between bag
                 electrodes and -15 kV on the precharger.

            Points to note:  (1)  Dust collects on tape,
            (2)  Bands are narrower with negative precharging.
                                 311

-------
                                       *:.;•.', y,
                                       vv-'.-.
Figure 7.   Appearance of inner surface of bag:

   a)   after deposition with 4 kV/cm between bag electrodes and no pre-
       charging.   Dust deposits in bands on both sets of electrodes and on
       tapes.

   b)   with positive precharging only.  Faint bands are visible on grounded
       electrodes, including area covered by tape.

   c)   with negative precharging only.  No bands are visible.
                                   312

-------
     Any technique capable of reducing pressure drop offers several
possible benefits:  one can choose to operate at a lower energy consumption
level, allow longer intervals between cleanings, or, by raising face velo-
cities, use fewer bags.  A recent study (11) has shown that the third of
these options gives the largest savings in operating costs, and it may be
assumed that in any future commercial development of ESFF, face velocities
will be higher than those in present practice.  However, limits to face
velocity are also imposed by falling efficiency.  The possible increases
in face velocities attainable in the combined bag electrode/precharger
mode were explored by making measurements of efficiency at a series of
velocities.  Figure 8 shows efficiencies obtained with no field  0,0),
with +4 kV/cm average field between electrodes  (+4,0), and with ,4 kV/cm
and -15 kV on the precharger (+4,-15).  There is a clear enhancement of
efficiency at all velocities.  The bag used was the same one in Figures 3
to 7, made of textured woven glass.  Long-term  tests in a pilot plant will
be needed to determine the maximum velocity at which stable operation is
possible.  It is reasonable to assume, however, that with a precharger/
electrified bag combination, considerable increases in face velocity will
be possible.
                                                   B. P
                  o 99
                  z
                  UJ
                  u.
                  LL.
                  UJ


                    98
                                                 0,0
                                3        6
                           FACE VELOCITY (cm/s)
Figure 8.   Dependence of efficiency on face velocity for woven glass bag
           with and without ESFF.   B and P indicate the same quantities as
           in Figure 4.
                                    313

-------
                                CONCLUSIONS

     Use of a precharger of new design  in conjunction with electric fields
generated by electrodes in the bag walls has led to reduced pressure drop
and lower dust penetration.  The reduction in pressure drop is due at least
partly to lessened uniformity of the  dust cake,  which is heavier at the
bottom of the bag and on the electrodes.   The improved performance was not
due to reduced dust loading, since no changes in total mass collected on the
bag were observed.  With this type of precharger/electrif ied bag combination,
it should be possible to operate a baghouse at face velocities several times
higher than those of current practice.
                                   314

-------
                               REFERENCES

 1.    linoya,  K.  and Makino,  K.   Application of  electric  field  effects  to
      dust collection filters.  Aerosol Science  5:  357, 1974.

 2.    Rivers,  R.  D.   Non-ionizing electrostatic  air filters.  ASHRAE
      Journal  37: February 1962.

 3.    Frederick,  E.  R.   Some  effects of electrostatic  charges in  fabric
      filtration.  In Proceedings:   Symposium on Use of Fabric  Filters
      for the  Control of Submicron Particulates,  EPA-650/2-74-043
      (NTIS PB 237629),  May 1974.

 4.    Bergman, W., Hebard, H.,  Taylor,  R.,  and Lum,  B.  Electrostatic filters
      generated by electric fields.   Paper  presented at the Second World
      Filtration  Congress, London,  England,  September  1979.

 5.    Lundgren, D.A.  and Whitby,  K.T.   Effect of  particle electrostatic
      charge on filtration by fibrous filters.  I & E  C Process Design and
      Development 4:345, 1965.

 6.    Helfritch,  D.J.   Performance  of an electrostatically aided  fabric
      filter.   CEP 54:   August  1977.

 7.    Hovis, L.S., Abbott, J.H.,  Donovan, R.P.,  and Pareja, C.A.
      Electrically charged flyash experiments in  a  laboratory shaker bag-
      house. In Third Symposium on  the  Transfer  and Utilization of Particu-
      late Control Technology,  Vol.  I,  EPA-600/9-82-005a  (NTIS  PB83-149583),
      July 1982.

 8.    Penney,  G.W.  Electrostatic effects on fabric filtration;   Vol. I,
      EPA-600/7-78-142a (NTIS PB288576),  September  1978.

 9.    Lamb, G.E.R. and  Costanza,  P.A.   Electrical stimulation of  fabric
      filtration  - Effects of electrode current.  Textile Res.  J. 51:389,
      1981.

10.    Lamb, G.E.R.,  Costanza, P.A.,  and Turner, J.  H.  Role of  filter
      structure and  electrostatics  in dust-cake  formation.  Textile Res. J.
      50:  661,  1980.

11.    Van  Osdell, D.  W., Ranade,  M.  B.,  Greiner,  G.  P. and Furlong, D. F.
      Electrostatic  Augmentation  of  Fabric  Filtration:  Pulse-Jet Pilot
      Unit Experience,  EPA-600/7-82-062 (NTIS PB83-168625), November 1982.
                                     315

-------
                            ESFF AS  A FIELD EFFECT

            By:   L.  S.  Hovis and G.  H.  Ramsey
                 Industrial  Environmental Research Laboratory
                 U.  S.  Environmental Protection Agency
                 Research Triangle Park,  North Carolina  27711

                                  and

                 R.  P.  Donovan
                 Research Triangle Institute
                 P.  0.  Box 12194
                 Research Triangle Park,  North Carolina  27709
                                   ABSTRACT

     Evidence to suggest that the mechanism on which ESFF (electrostatic
stimulation of fabric filtration) depends is an electric-field-only mechanism
(as 'opposed to a Coulomb mechanism that depends on both electric field and
electric charge) includes:

     1.   A room-temperature high-humidity factorial experiment in which
          both external electric field and fly-ash electrical charge were
          independent variables.

     2.   Selected experiments carried out at low relative humidity and room
          temperature.

     3.   Enhanced filtration measurements made with a 60-Hz ac electric
          field applied to the bag electrodes.

     4.   Published precharging data of others in which enhancement vanishes
          at high relative humidity.

     All of the data collected in EPA/IERL's Research Triangle Park laboratory
(items 1-3, above) refer to experiments in which pulverized coal fly ash was
the dust source.  Dusts from other sources were used in the experiments of
item 4.

     This paper has been reviewed in accordance with the U. S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.


                                     316

-------
     ESFF (electrostatic stimulation of fabric filtration), an acronym coined
by reseachers at Textile Research Institute(l), refers to a technique for im-
proving fabric filtration performance by applying an external, nonionizing
electric field parallel to the fabric surface during the filtration process.
The improved performance manifests itself as a reduced drag and reduced
penetration (although not all researchers report a reduced penetration--
Reference 2, for example).

     Typical explanations of the observed enhancement invoke Coulomb capture
forces between the electrically charged dust particles and the externally
applied surface field, expecially on the low-fiber-density upstream surface
of the fabric.(1)  Particles collected on these low-fiber-density surfaces do
not contribute as much to fabric drag as those collected in the more tightly
packed interstices of the bulk fabric.  Qualitatively, this model of ESFF
portrays the action of the electric field as enhanced upstream capture of
electrically charged dust at low-fiber-density 'surface sites.

     In such a model, particle charge is expected to be a variable of first
order significance since the Coulomb force experienced by a particle approach-
ing the fabric depends on both particle charge and the electric field in the
region through which it passes.  Other electrical forces exist, however, such
as polarization forces; and other non-Coulomb-force-field-dependent mechanisms
could be significant.  This paper reviews data that support ESFF effects based
on mechanisms other than Coulomb capture.

                            HIGH HUMIDITY FACTORIAL

     This experimental series used both fly-ash electrical charge and an ex-
ternal electric field at the fabric surface as independent variables in a
factorial experiment measuring the performance of a pulse-jet baghouse.  De-
pendent variables of the experiment included pressure drop across the bags
(AP), outlet fly-ash concentration (C ), and an effective cake resistance
(K2').  The last quantity was calculated from increase in AP during a filtra-
tion cycle and the corresponding increase in fly-ash mass on the bags, de-
termined by carefully demounting and weighing.

     In the context of this factorial experiment, high relative humidity
means 52 to 59 percent.  The specific value itself is not important except
as an indication that the electrical resistivity of the fly ash/fabric filtra-
tion media would be low enough to dissipate the electrical charge deposited
along with the collected fly ash without significant charge accumulation
in/on the dust cake.  Since the maximum current carried to the bags by the
fly ash under any of the test conditions was 150 to 200 nA, having the bag
electrode current in the 150- to 200-jjA range, three orders of magnitude
higher, ensures that the field at the bag surface depends primarily on the
voltage applied to the electrodes.  Thus, keeping the relative humidity
greater than 50 percent kept the electical resistivity of this particular
test fly ash/fabric combination low enough to avoid the complications
brought about by an electric field component attributable to deposited
charge.

                                      317

-------
     The details of the experimental procedure and apparatus appear else-
where. (3)  The key conclusions are summarized in Figures 1 through 4, the
first two of which plot normalized K2' and &P  as a function of precharger
current (Figure 1) and field voltage (Figure 2).  Figure 1 shows that K2'
does not change with increasing fly-ash electrical charge, as measured by the
corona precharger current.  (The average charge/mass of the fly ash varied
over the range ±0.2 [tC/g to -2.0 pC/g as the power supply current varied over
the range plotted.)  The dashed line in Figure 1 plots the normalized AP_
values corresponding to these precharger power settings.  APn is the final
pressure drop (AP^,) minus the residual pressure drop (APR) in one filtration
cycle.  The reduction in APn with increasing precharger current is the electri-
cal enhancement effect attributed to precharging.(4-6)  Since K2* (which depends
on the ratio APn/M_, where fL is the increase in fly-ash mass on the bag that
caused APR to increase to AP^) does not change, Figure 1 implies that the APD
decrease has been compensated by a similar decrease in M^.  Figure 3, which
compares the normalized APn curve with the corresponding normalized M^ curve,
shows that that is exactly what has happened and that the electrical enhance-
ment described by the reduced values of AP  are achieved by reducing the M_^
on the bags.  The precharger acts as a collecting prefilter, a conclusion
reported at the previous symposium for both pulse-jet(7) and shaker baghouse(8)
experiments.

     The field voltage dependencies (Figures 2 and 4) tell a different story.
Figure 2 shows that the normalized K2' and the normalized AP_ decrease together
as a function of external field voltage.  Figure 4 confirms that M._ is inde-
pendent of field voltage meaning that the mass of fly ash reaching the bags
is independent of bag electrode voltage.  These plots show, therefore, that
the permeability of the deposit has increased, resulting in a decrease in
cake airflow resistance, because of the field.

     The mechanism by which this increase in permeability takes place cannot
be further specified other than to say that Coulomb forces do not dominate
the mechanism.  This conclusion follows from the observation that charging
the fly ash to a high charge-to-mass ratio (Q/M) does not alter the magnitude
of the permeability enhancement; that is, the reduction in K2* brought about
by the external field is the same for the lowest Q/M fly ash as for the highest.
A mechanism that depended on Coulomb forces primarily should produce increasing
response with increasing Q/M at constant field.  That the fly-ash deposit does
not exhibit this property leads to the conclusion that ESFF, in this experimental
setup at least, depends upon electrical interactions other than Coulomb attraction.

     While the capability of the precharger to add charge to the fly ash
that reaches the bags is indisputable (the bag electrodes can be used to
measure the increased current contributed by the precharger), the distribution
of this charge by particle size is not known.  Conceivably, all the charge
could be added to a size fraction of the fly ash that proves insignificant in
ESFF.  Two additional experiments are planned to further clarify the conclusions:

     1.   Charge spectrometry by the Millikan apparatus.

     2.   Charge neutralization by operating the precharger in the ac mode.


                                     313

-------
     The first supplementing experiment will measure the distribution of
charge among the various sizes of fly-ash particles; the second will attempt
to neutralize the natural charge now present on the fly ash so as to achieve
a data point of even lower Q/M than when the precharger is OFF.

                           LOW HUMIDITY OBSERVATIONS

     When the high humidity factorial experiment is not carried out at "high
humidity," by definition significant electrical charge can accumulate on the
bags because the charges carried by the fly ash do not leak off as readily
as when operating at high humidity.  The cage voltage(9) is one measure of
accumulating bag charge.  At operating relative humidity of 35 percent or less,
the cage voltage (cage voltage is the voltage on the electrically isolated
support cage of a pulse-jet bag, as read by a high impedance voltmeter such
as an electrometer) generally builds up to values exceeding ±1000 V dc, sug-
gesting that the net electric field at the fabric surface now has a component
attributable to accumulating charge in addition to the applied field that
dominated the high humidity case.

     Furthermore, because these charges will drift in the externally applied
field, they distribute themselves so as to neutralize the external field over
much of the fabric surface similar to the action described by Walkenhorst in
his metallic fiber filter.(10)  Figure 5 sketches an extension of the Walken-
horst model to a fabric filter as envisioned here.

     Laboratory observations confirm that the ESFF effect diminishes at low
relative humidity and that charge accumulation increases.  However, because
experiments correlating cage voltage with K2' and other performance measures
are incomplete, this evidence is only speculative at present.  Should the
interaction depicted in Figure 5 prove valid, then a low frequency field
reversal as discussed by Walkenhorst(lO) should restore the ESFF interaction.

                                  AC ESFF

     A third general experimental approach is to examine ESFF as a function
of field frequency.  The reason such observations are significant is that
Coulomb forces depend on field direction, while other field dependent mecha-
nisms such as dielectrophoresis do not.  Dielectrophoresis, the net attractive
force experienced by a polarized neutral body in an electric field gradient,
depends on the direction of the field gradient alone.  The net force is always
in the direction of increasing field strength; thus, reversing the polarity
of the field does not change the direction of the net polarization force in
the typical ESFF experimental configuration.  The net Coulomb force under ac
operation, however, is zero for time periods long compared to 1/f, where f is
the ac frequency.

     To date, the only frequency investigated has been 60 Hz.  At 60 Hz, the
ESFF mechanism appears intact and operative, although limitations in available
power supplies have restricted the available field strength.  A fly-ash fabric
combination that yielded a pressure drop ratio (PDR) of 0.6 with 8000 V dc
across electrodes 1.8 cm apart produces a PDR of 0.7 with 5000 V rms (60 Hz).
The PDR is the ratio of AP  with an electric field to AP  without a field.
Face velocity during both measurements was approximately 5 cm/s.
                                     319

-------
     While the observation of an ac ESFF effect suggests that polarization
is a significant mechanism, Coulomb forces cannot be ruled out until much
higher frequency data are available.

                     FIBROUS FILTER ELECTRICAL ENHANCEMENT

     Other researchers have reported electrically enhanced fibrous filtra-
tion through the use of prechargers(ll), electric fields across the media
(12), or both (13).  Calculations of the single-fiber collection efficiency
of an electrically charged particle flowing past a charged fiber(l4) predict
that the addition of this Coulomb force to the classical capture mechanisms
influences single-fiber efficiency primarily at Stokes numbers below about 2
(Figure 6).  For typical filter parameters, this conclusion implies that
Coulomb forces are important only for particles less than about 5 pm diameter.

     These Coulomb-force-based predictions agree reasonably well with pub-
lished data, using either a precharger (Figure 7) or an external field across
the filter (Figure 8).  In these two plots, overall filter efficiency is the
ordinate and particle diameter, the abscissa.

     Figures 7 and 8 represent an electrical enhancement of fibrous filtra-
tion attributable to a Coulomb force.  What will be argued next is that these
enhancements are not ESFF-type enhancements as described in References 1, 2,
and 3 and, in fact, exhibit behavior different from the reported ESFF en-
hancement.  Furthermore, the behavior exhibited by the electrically enhanced
fibrous filters is consistent with Coulomb force enhancement, while that ex-
hibited in the ESFF experiments is not.

     No significance is attached here to the fact that the primary enhance-
ment criterion differs between the two types of filters.  For the fibrous
filter, collection efficiency is the primary measure of performance; for fab-
ric filters, it is pressure drop, drag, K2', or some normalized ratio describ-
ing filter energy consumption.  What is crucial is that the enhancement of
each depends differently on certain independent variables.

     Take the data in Figure 8 first.  Bergman et al.(12) report that the
enhancement depicted in Figure 8 largely disappears if the dust source is
passed through a charge neutralizer prior to the filter.  This behavior is
consistent with a Coulomb force which depends on both particle charge and the
electric field at the collector.  With fabric filters, as described in the
high humidity factorial experiment, increasing the particle charge produced
no discernible effect on K2'.  While increasing the particle charge is not
identical to the Bergman et al.(12) experiment, a Coulomb-force-dominated
variable should respond to any change in particle charge.  That it did not do
so in the fabric filter experiment led to the conclusion that Coulomb forces
do not dominate the ESFF enhancement observed in the high humidity factorial
experiment.

     The charge neutralization experiment planned as continuing work is
similar to the experiment reported by Bergman, who concluded that "natural"
particle charge was essential for the electrical enhancement he observed
and as a Coulomb mechanism would require.


                                     320

-------
     The precharger-induced enhancement represented in Figure 7 is reported
by Yu and Teague(ll) to disappear when carried out at high humidity or with
conductive dusts.  This observation agrees with a model picturing charge
accumulation on the filter as the source of the electric field at the filter
surface that is required for the Coulomb capture force to be present.  When
these charges are allowed to leak off, the field disappears, and the Coulomb
force vanishes as does the enhancement.  These observations are consistent
with the hypothesis that this fibrous filtration enhancement is Coulomb-force
dominated.

     The ESFF data presented earlier exhibit just the opposite behavior.  It
is at high humidity—no surface charge accumulation—that the ESFF enhancement
is most pronounced.  Charge accumulation seems to be associated with diminished
ESFF, although this conclusion is very tentative at present.

     What the preceding paragraphs have attempted to do is to review the proper-
ties of a type of electrically enhanced filtration thought to be dominated by
Coulomb forces (i.e., fibrous filtration) and then show that certain dependen-
cies of these filters differ from those observed with ESFF and that, further-
more, those differences can be understood by attributing the electrical enhance-
ment of fibrous filtration to Coulomb forces and the electrical enhancement of
fabric filtration of fly ash to some unspecified but nonCoulomb mechanism.

                                  CONCLUSIONS

     The evidence cited supports the existence of a non-Coulomb mechanism of
ESFF.  While the evidence is neither conclusive nor exclusive (in that some
experiments remain to be done before the conclusion becomes solidly established
and nothing presented rules out the existence of conditions under which Coulomb
forces could be significant in ESFF), the data make a non-Coulomb mechanism or
mechanisms the most likely explanation of the ESFF phenomenon observed in
the EPA/IERL-RTP in-house laboratories.  The data do not distinguish between
an electrical polarization mechanism and other non-Coulomb mechanisms.  They
do suggest that the magnitude of the electrical charge on the incoming fly ash
is not an important variable, a surprising conclusion in view of the recognized
dominance of Coulomb forces in most electrified systems.

                                  REFERENCES

1.   Lamb, G. E.  R. and Costanza, P. A.  A low-energy electrified filter
     system.  Filtration and Separation.  July/Aug.:319-322, 1980.

2.   Greiner, G.  P., Furlong, D. A., VanOsdell, D. W., and Hovis, L. S.
     Electrostatic stimulation of fabric filtration.  J. Air Pollut. Control
     Assoc.  31:1125-1130, 1981.

3.   Donovan, R.  P., Hovis, L. S., Ramsey, G. H., and Ensor, D.  S.   Electric-
     field-enhanced fabric filtration of electrically charged flyash.
     Aerosol Science and Technology.  l_:385-399, 1982.

4.   Ariman, T.,  and Helfritch, D. J.  Pressure drop in electrostatic fabric
     filtration.   In:  Second Symposium on the Transfer and Utilization of
     Particulate Control Technology, Vol.  III.  EPA-600/9-80-039c (NTIS
     No. PB-81-144800) (Sept.), 1980.  pp. 222-236.

                                     321

-------
5.   Chudleigh, P. W. , and Bainbridge, N. W.   Electrostatic effects in fabric
     filters during build-up of the dust cake.  Filtration and Separation.
     17:309-3ir, 1980.

6.   linoya, K., and Mori, Y.   Experimental advances in fabric filtration
     technology in Japan—effects of a corona precharger and relative humid-
     ity on filter performance.  In:  Second Symposium on the Transfer and
     Utilization of Particulate Control Technology, Vol. III.  EPA-600/9-80-039c
     (NTIS No. PB-81-144800) (Sept.), 1980.  pp. 237-250.

7.   Donovan, R. P., Hovis, L.  S., Ramsey, G. H.,  and Abbott, J.  H.  Pulse-
     jet filtration with electrically charged flyash.  In:  Third Symposium
     on the Transfer and Utilization of Particulate Control Technology, Vol. I.
     EPA-600/9-82-005a (NTIS No. PB-83-149583) (July), 1982.  pp. 11-22.

8.   Hovis, L. S., Abbott, J.  H., Donovan, R. P.,  and Pareja, C.  A.  Elec-
     trically charged flyash experiments in a laboratory shaker baghouse.
     In:  Third Symposium on the Transfer and Utilization of Particulate
     Control Technology, Vol.  I.  EPA-600/9-82-005a (NTIS No. PB-83-149583)
     (July), 1982.  pp. 23-34.

9.   Donovan, R. P., Ogan, R.  L., and Turner, J. H.  Electrostatic effects in
     pulse-jet fabric filtration of room temperature flyash.  In:  M. P.
     Freeman and J. A. Fitzpatrick (eds.), Theory, Practice and Process
     Principles for Physical Separations.  Engineering Foundation, New York,
     New York, 1981.  pp. 445-466.

10.  Walkenhorst, W.  Deliberations and studies on the filtration of dust-
     laden gases with special allowance for electric forces.  Staub-Reinholt
     Luft.  29(12):1-13, 1969.

11.  Yu, H. S;, and Teague, R.  K.  Performance of electrostatic fiberbed.   In:
     Extended Abstracts of the 1st Annual Conference of the American Associa-
     tion for Aerosol Research, Santa Monica, California.  February 17-19,
     1982.  pp. 19-3-19-4.

12.  Bergman, W., Hebard, H. D., Taylor, R. D., and Lum, B. Y.  Electrostatic
     filters generated by electric fields.  In:  Proceedings of The Second
     World Filtration Congress, 1979.

13.  Ho, C. P., the Bahnson Co., Winston-Salem, N.C.

14.  Loeffler, F.  The influence of electrostatic forces and of the proba-
     bility of adhesion for particle collection in fibrous filters.  In:
     Proceedings:  Symposium on New Concepts for Fine Particle Control.
     EPA-600/7-78-170 (NTIS No. PB-292-095) (August), 1978.  pp.  206-236.
                                     322

-------
       1.0
      0.5
                    o
                   _o
                        0 13.5 ft/min (6.9 cm/s)

                        •  9.8 ft/min (5.0 cm/s)
   2  Precharger
                                                  EAR
                                                     DPrecharger
                                                  EAR
                                                     DO
                   20             50
                          Precharger Current
        100
Figure 1. Normalized K.'2 and APD as a function of precharger current.
     (Each I, value is summed over three values of field voltage.)
        1.0
        0.5
                                                   Field
                                               EAR
                                               EAR
o Field  (no data
      points)
                                                  DO
                       ° 13.5 ft/min (6.9 cm/s)
                       • 9.8 ft/min (5.O cm/s)
                            Field Voltage (kV)
   Figure 2.  Normalized K£ and AF^, as a function of field voltage.
   (Each D value is summed over three values of precharger current.)
                                 323

-------
 1.0
 0.5
                                                D Precharger
                    0 13.5 ft/min (6.9 cm/s)
                    • 9.8 ft/min (5.0 cm/s)
             20             50
                     Precharger Current (/iA)
                            100
Figure 3. Reduction in dust mass (MD) deposited on bags
             because of particle precharging.
  1.0
  0.5
                                                  D Field
                0  13.5 ft/min (6.9 cm/s)
                •  9.8 ft/min (5.0 cm/s)
                      LM,
                         DO
                      DAR
                                                  D Field
                                               L'AR,
4                   8
   Field Voltage (kV)
                                                    (no data
                                                   "points)
   Figure 4. Independence of dust mass and field voltage.
                            324

-------
         Fabric
                            a)  Starting field lines (efabrjc ~ 1>
         Fabric
                              b) After charge accumulation
          Figure 5. Predicted influence of charge accumulation
                        on the surface electric field.
                1.6  -
                                  R«-0.2
                                  Pp- 2.164 B/cm3
                                              R, - Reynolds No.-

                                              q  - particle charge;
                                              Q « fiber charge/cm;
                                              Pp • particle density;
                                              Op * particle diameter;
                                              U0 » upstream flow velocity;
                                                - gas viscosity;
                                              Of • fiber diameter; and
                                              p  « gas density.
                                                        100
                                 t-***
                                     IBuDp

Figure 6. Theoretical single-fiber collision efficiencies. (Inertia, interception,
  and gravity with electrostatic forces. Lamb's flow field.) (Loeffler, 1978).
                                     325

-------
      IOO
     ,T 80
       60
       40
     e 20
                                   nocho,,.   £ 93%»o,(l. 12-mit f.ber
                                         /Qv   3-inch thick bed


                                         "
                            o--:
        01
                               05         10
                                  Particle Diameter. dp . pm
Figure 7.  Fractional efficiency of electrostatic fiberbed. (Yu and Teague, 1982).
     100


       80


       60


       40


       20
                                                   'M      I
O Electrical Mobility
    Analyzer
 *  Laser Particle Counter
                                                      it	L
        0.01
0.05    0.1                 0.5   1.0
          Diameter, /zm
 Figure 8. Filter efficiency as a function of particle size with and without an
    electric field using naturally charged aerosols. (Bergman et al., 1979).
                                      326

-------
ELECTRICAL ENHANCEMENT OF FABRIC FILTRATION:  PRECHARGING VS. BAG ELECTRODES

               by:  R.P. Donovan
                    Research Triangle Institute
                    Research Triangle Park, NC  27709

                                     and

                    L.S. Hovis and G.H. Ramsey
                    Industrial Environmental Research Laboratory
                    U.S. Environmental Protection Agency
                    Research Triangle Park, NC ,27711
                                  ABSTRACT
     Two distinct approaches to achieving electrically enhanced fabric fil-
tration are:  1) upstream precharging in a corona discharge; and 2) dust
filtration in an external electric field applied by electrodes woven into or
positioned adjacent to the bag.  Both techniques have demonstrated perfor-
mance enhancement by some measure, but the mechanisms of enhancement may
differ; clearly the hardware does.

     This paper reviews the hardware options and configurations available
with each basic approach and summarizes various mechanisms whereby each
approach brings about an electrical enhancement, including those which domi-
nate in combined precharger/bag electrode systems.  From this background,
guidelines for matching configurations to source properties are suggested, as
well as a hybrid electrostatic precipitator "(ESP)/fabric filter design incor-
porating features that enable it to operate as either an electric-field aided
baghouse or a hybrid ESP/baghouse.

     This paper has been reviewed in accordance with the U.S. Environmental
.Protection Agency's peer and administrative review policies and approved for
presentation and publication.
                                     327

-------
     As used in this paper, electrical enhancement of fabric filtration
refers to any technique for improving filter performance by the use of high
voltage electrodes, in either an ionizing or non-ionizing mode, at the fabric
surface or some upstream location.  Even this restricted definition of elec-
trical enhancement now encompasses a growing number of configurations and
mechanisms.  Rather than simplify and clarify choices, on-going research is
uncovering more mechanisms and options by which the performance of fabric
filters can be improved electrically.  This paper reviews the present state
of understanding and design options now available.

     The two basic variables that high voltage can control in particle col-
lection by a fabric filter are dust particle electrical charge and fabric
electric field (collector electrical charge).

     Early electrical enhancement work concentrated on directly controlling
only one or the other of these quantities, while more recent research in-
cludes arrangements for controlling each independently.  With the more elabo-
rate control capability has come not only more variation in control options,
but also improved understanding of the full range of electrical enhancement
mechanisms.

CORONA PRECHARGING

     Typical hardware for controlling dust charge consists of a corona pre-
charger located upstream of the bags.  Conventional wire-duct geometry—a
high voltage wire positioned along the centerline of a circular duct which
also serves as the low voltage electrode—lends itself to easy incorporation
in most installations, new or old.  The role of the precharger is to add
electrical charge to the dust which certain researchers imply enhances fabric
filtration, although no plots of dust drag or dust cake resistance versus
collected dust charge/mass are in the literature.  What typically appears is
a plot of AP, drag, or change of drag with dust load versus a measure of
corona power supply output or, more commonly, a comparison of precharger ON
and precharger OFF performance characteristics.  Such data alone leave un-
answered questions regarding the enhancement mechanisms.

     One obvious collection mechanism is electrostatic precipitation in the
precharging section—the precharger is a single-stage electrostatic precipi-
tator, albeit a low efficiency unit in most designs.  Thus, dust charged by
the wire can be trapped in the same applied electric field that created the
corona conditions.

     Even charged dust that is not trapped in the precharger section can be
precipitated on the duct walls or the baghouse walls prior to reaching the
bags.  The electrical force driving the charged dust to the walls can be the
mutual repulsion between species with charge of the same sign (space charge
precipitation) or can be an image force created in the wall by each charged
dust particle.  The latter is likely to be important only at low charge
densities, while the former dominates in high charge density regions more
typical of particulate control equipment design.
                                     328

-------
     Regardless of capture mechanism, the effect is to reduce the rate at
which dust arrives at the fabric, thus slowing the rate at which AP or drag
increases but not necessarily changing the dust cake specific resistance
(K2)-  Thus, this enhancement mechanism is that of a prefilter.  From a mass
viewpoint alone, a cyclone upstream of the bag does the same.  (A cyclone
prefilter would affect the size distribution of the dust reaching the bags
differently than an ESP prefilter.)

     A drawback of any prefilter concept is that two control sections must be
operated and maintained, including cleaning.  Dust collected by the pre-
charger must also be accounted for and removed.  Some clever sch  :es exist
for minimizing this extra cost.  For example, in the Apitron® (i, design
(Figure 1), the same air pulse that cleans the fabric also cleans the corona
wire and duct.  The discharge from both sections is to a common hopper so
that, in principle, one cleaning action regenerates both sections.
                               STAST
                               CLEANING
 RESUME
FILTRATION
                Figure  1.  Apitron® design and operation  (1).

     An alternative approach described by Lamb (2)  is to  locate the pre-
 charger at the bag mouth so that the precipitated dust  is  collected on the
 bag  surface.  In this arrangement, bag cleaning action  is  all  that is neces-
 sary since both the precipitated dust and the filtered  dust  are collected on
 the  bag surface.
                                     329

-------
     At the opposite extreme is the use of a conventional electrostatic pre-
cipitator (ESP) in series with a conventional baghouse.  This option exists
in the field where regulations dictate upgrading of existing ESP equipment.
Simply adding a downstream baghouse while continuing to operate the ESP
would, in the context of electrical enhancement of fabric filtration, cause
the ESP to serve as a precharger for the dust entering the baghouse in ad-
dition to performing its usual collection function.  Here, however, particu-
late control is clearly a tandem operation; both the ESP and the baghouse
would be operated and maintained as independent units.  Because of the re-
duced load, the AP buildup across the fabric would be very slow, independent
of any charge-dependent increase in dust-cake permeability.

     At least three cases of precharger/baghouse coupling thus exist.  Table
1 classifies prechargers according to location with respect to the fabric
collection surface of the downstream baghouse.  The role of a precharger is
to add electrical charge to the dust.  All prechargers are assumed to perform
this function satisfactorily and to succeed in having most of the dust parti-
cles at near-saturation charge when they leave the precharger section.  Then,
depending on the length of the run between the precharger and the geometry of
the intervening duct work, more or less charge is lost in transit.

     For the close-coupled case, virtually all charge added to the dust by
the precharger ends up on the bag surfaces as the dust is collected.  For the
independent units, considerable loss of charge and dust will occur by colli-
sion with the walls.  Dust that contacts the walls but is subsequently re-
entrained will probably have lost most of its charge.  The quantity of charge
reaching the fabric surface is likely to be a small fraction of that carried
away from the precharger by the dust.

     No charge data exist for these two extreme cases.  For the second case
listed in Table 1—the remote coupled—some measurements of charge deposited
on the bag do exist.  By comparing these measured values with the total
charge that would be collected if all the particles collected carried the
saturation values of charge corresponding to the operating conditions of the
precharger, the observed charge is typically 2 to 5 percent of what is calcu-
lated to leave the precharger.  (The procedure for calculating expected
charge per particle follows conventional ESP procedures and is outlined in
the Appendix of Ref. 3.)

     Field coupling (Table 1) refers to a field component being present at
the fabric surface because of the location of the precharger.  While a field
component at the fabric can be created not only by applying a voltage to the
corona wire but also by the charges added to the dust leaving the precharger,
the field coupling in Table 1 refers only to the direct coupling—that com-
ponent that is present when corona power is ON but no dust is flowing—and
only the close-coupled case produces such a fabric surface field component.

     Common cleaning (Column 2, Table 1) implies that an additional separate
cleaning action is not required to remove the dust collected by the pre-
charger.  For the separate units case (No. 1, Table 1), the ESP hoppers must
be emptied and the collection plates rapped.  Common cleaning is not possi-
ble.  For the close-coupled unit, there is no collection surface other than


                                     330

-------
              TABLE 1.  TYPES OF PRECHARGER-BAGHOUSE COUPLINGS
3.
                              Electrical
                               Coupling

                            Charge    Field
                                          Common
                                         Cleaning
                      Electrical
                    Enhancement of*
                                                    S; AS/At
                              K2'
1.

Two independent
separate units
(ESP upstream of the
baghouse)
Yes No No
(small)

Yes ?
(No data
available)
    Remote-coupled units      Yes

    (Precharger in the
     hopper inlet)
Close-coupled units       Yes

(Precharger at the      (high)
 bag mouth)
                                   No    Yes/No

                                        (Geometry
                                         dependent)
Yes
Yes
Yes    Yes/No

       (Dust con-
        ductivity
        and size
        dependent)

Yes    Yes...butt
 S = drag = AP/V; K2 = measured specific cake resistance = AS/AM (where
 M = dust mass/unit area).
 Yes, assuming uniform dust distribution, but dust distribution is known to
 be highly non-uniform.
the fabric.  When the fabric is cleaned, both the baghouse and the ESP col-
lectors are being cleaned, since they are the same.

     The remote-coupled unit (Case 2, Table 1), as operated at EPA/RTP, does
not require additional cleaning because virtually all of the precipitated
dust collects in the hopper or the baghouse walls.  No additional cleaning is
required as these surfaces slough off their dust loads in normal operation
without a special trigger.  The precharger walls do not accumulate any dust
because of the high gas flow-through.  Other larger remote installations may
not operate this way, and aperiodic rapping might be required.

     The evidence showing filtration to be electrically enhanced consists of
reduced S and AS/At in all cases.  Only for the close-coupled case is AS/AM
unambiguously reduced.  This reduction in AS/AM appears to result primarily
from a highly non-uniform dust distribution along the bag length, the great-
est dust areal density being adjacent to the precharger end of the bag, where
the precharger contribution to the fabric surface electric field is concen-
trated (2).
                                     331

-------
     In the remote-coupled configurations, reports of K% enhancement conflict.
Ariman.and Helfritch (4), Chudleigh and Bainbridge (5), and linoya and Mori
(6) present data which show enhanced K2, while Donovan et al. (3) and Hovis
et al. (7) fail to find such a result.  The key to this difference apparently
lies in the capability of certain collecting systems to accumulate electrical
charges as they collect dust, thereby building up an electric field on the
fabric surface.  Filtration of charged dust in the presence of this self-
induced electric field produces a reduced K^.  If> however, the collecting
surface is conductive, so that no significant electric field builds up, the
K.2 enhancement does not occur.  At room temperature, increases in the rela-
tive humidity from 48 percent to 58 percent can convert a K£ enhancing system
to a non-enhancing system as reported in the data published by Chudleigh and
Bainbridge (5).  Because of high flyash resistivity, utility baghouses oper-
ate under conditions that favor a K2 reduction in either the independent or
remote-coupled configurations, but this result also depends on particle size
being small (see Refs. 8 and 9).

     Corona wires in all precharger versions must be cleaned, especially
those which are located between the bags and the hopper, such as the Apitron®
or a bottom-feed close-coupled case.  During cleaning, the dust removed from
the bag cascades through the electric field on its way to the hopper, adding
an extra opportunity for dust collection on the corona wires.  A top-feed
design avoids this potentially deleterious interaction.

FIELD ELECTRODES

     The role of field electrodes is to create a strong external electric
field throughout the volume occupied by the filtration media and the near up-
stream volume of the fabric.  Orientation of the electric field can be either
in the direction of gas flow through the filter--roughly perpendicular to the
fabric surface—or in the plane of the fabric surface—roughly perpendicular
to the direction of gas flow (Table 2).  Different authors have preferred
different orientations based on analyses of a single fiber in an electric
field.  For example, Havlicek (10) argues for the electric field in the
direction of the gas flow; Lamb et al. (11) prefer the electric field perpen-
dicular to the gas flow.  Without discussion, Zebel (12) solves only the case
of electric field parallel to the gas flow.

     In practice, both orientations have been used successfully.  The practi-
cal advantages of building the electrodes into a cage or the fabric itself
have resulted in more experience with the electric field in the plane of the
fabric than along the direction of gas flow, but limitations in this arrange-
ment, chiefly with respect to power consumption with low resistivity dusts or
fabrics, may lead to more research with the alternative orientation and
arrangement.  Primary emphasis here will be on configurations that orient the
field in the plane of the fabric—the ESFF (electrical stimulation of fabric
filtration) orientation developed at Textile Research Institute (11, 13).
                                     332

-------
                      TABLE 2.  FIELD ELECTRODE OPTIONS
          Electric Field Perpendicular to Fabric Surface

               a)   center-line wire electrode (inside-out flow)

               b)   enclosing cage or housing wall (single bags)

          Electric Field in the Plane of the Fabric Surface

               a)   electrodes upstream (dirty side)

               b)   electrodes downstream (clean side)
               c)   electrodes woven into the fabric
     Even after the selection of the field orientation in the plane of the
fabric, options exist.  The electrodes can be upstream (wholly on the dirty-
gas side of the fabric),  downstream (wholly on the clean-gas side of the
fabric), or incorporated into the fabric itself (part upstream, part down-
stream, and part within).

     The option recommended by Lamb and Costanza (14) (and still the prefer-
red from a field strength viewpoint) is the upstream location in which the
electrodes, an array of parallel wires of alternating polarity, are positioned
immediately in front of the fabric but not quite touching it.  This location
creates a high electric field in the near-upstream region of the fabric just
prior to dust collection.  The electrical enhancement is greater when the
fabric surface has a heavy nap finish which causes many surface fibers to
protrude into the upstream region.  In the presence of an electric field
these fibers become more efficient capture centers than without the electric
field.  Since dust collected in this low fiber density region makes a smaller
incremental contribution to fabric drag than dust collected deeper within the
fabric, the net result of turning ON the field is a lower K^.  This model of
ESFF has been dubbed "the tall trees" model (15), and represents a depth
redistribution of collected dust mass.  Areal mass density, averaged over
dimensions large with respect to fiber-to-fiber separation, is assumed con-
stant in this model.  Dust mass distribution as a function of depth into the
filter is postulated to account for the improved performance.

     A major advantage of the upstream location is that the electric field in
the near-upstream volume from the fabric is not attenuated by fabric elec-
trical properties.  When the electrodes are shifted to the downstream clean
side, a layer of fabric shields this critical near-upstream volume from the
electrostatic field.  With conductive fabrics (or conductive dusts which
become an important part of the filtration media as the dust accumulates in
and on the fabric), a significant reduction in the electrical field in the
near-upstream volume could occur.  The "tall trees" model predicts that such
a reduction in field in this region should reduce the magnitude of the en-
hancement per unit applied voltage between the electrodes.

                                     333

-------
     In some reports switching from a high resistivity fabric such as Teflon
to a lower resistivity material such as polyester does in fact degrade the
observed electrical field enhancement, confirming qualitatively this predic-
tion of the tall trees model.  By other accounts, however,  this fabric depen-
dence is not observed.  And, in the initial pilot-scale work on an industrial
boiler (16), switching from upstream electrodes to downstream electrodes
(Teflon fabric) had little effect on the electrical enhancement.   Thus,
attenuation of the electrical field appeared unimportant here.

     The major advantage of downstream electrodes is that they can be easily
incorporated into the support cages of outside-in bag filters, allowing ESFF
to be implemented with only minor modifications in the conventional hardware.
Rather than require the design of a special electrode harness as  in the up-
stream option, the downstream option can be implemented by the simple addi-
tion of electrical insulators between appropriate cage members.  This arrange-
ment has been shown to perform with impressive economic advantage on a pul-
verized-coal-fired industrial boiler (16), even though the fabric used in
this demonstration was the relatively expensive Teflon felt.

     Electrodes woven into the fabric are a third option seemingly well
matched to the requirements of reverse-air cleaned baghouses.  In this con-
figuration yarns made of conductive fibers become part of the fabric weave.
Choice of suitable weave pattern allows the conductive yarn to face primarily
upstream if that is deemed desirable or to be incorporated in any of the
variety of weaves available to textile processing.  A major advantage of this
configuration is that it minimizes user preparation and demands.   In routine
service the plan is to install and handle the ESFF bags much the  same as
conventional bags.  The only differences would be that the fabric is special
and an electrical contact, perhaps no different from conventional mechanical
attachment, however, must be made at each end of the bag.  In operation the
ESFF power supplies must be monitored.

     Experience with this option is limited to laboratory operation at pre-
sent but already some striking differences are apparent, the most pronounced
of which is preferred deposition of the collected dust on the electrodes, as
reported by Lamb (2).  This mechanism of enhancement is a surface redis-
tribution phenomenon—the deposited dust does not coat the fabric surface
uniformly as in conventional filtration but piles up on the electrodes cre-
ating a distinct striped deposition pattern in the early stages of filtra-
tion.  The dust pattern mirrors the electrode geometry, although one elec-
trode polarity generally appears to collect more dust than the other.

     This kind of deposition pattern leads to an electrical enhancement
attributable to non-uniformity in dust deposition between electrodes, similar
to that previously reported by Penney (17) and by Chiang et al. (18).  This
type of deposition non-uniformity is also similar to that previously associ-
ated with the close-coupled precharger.  When the woven-in electrodes are in
the warp direction (along the bag axis) of a vertically suspended bag, the
deposition non-uniformity is in the lateral direction (the concentration
gradient is maximum in the horizontal direction).  When combined with a
                                     334

-------
close-coupled precharger (deposition non-uniformity in the vertical  direc-
tion), as  in Lamb's configuration, the  resultant two-directional  non-unifor-
mity produces dust deposits of drastically reduced K£ as compared to either
single-direction non-uniformity alone or  the uniform deposition case (2).

     Because of the parallel flow paths through a bag, the dust distribution
that results in the highest net drag is the uniform distribution.  Any depar-
ture from  uniformity produces a lower net drag because the flow through the
region of  reduced drag increases with respect to the flow through all other
regions.   This  argument assumes that drag is proportional to  areal mass
density.   Decreasing the mass density in  one area decreases that  area's drag
and correspondingly increases the drag  of all those areas to  which the dis-
placed mass has been added.

     Figure 2 illustrates the parallel  flow concept.  These curves,  adapted
from Dennis et  al.  (19), plot net filter  drag as a function of dust  loading
for two distributions (Figure 3):  1) the uniform loading case, and  2) a
highly non-uniform loading case.  The non-uniform case assumes that  the dust
deposits on the fabric in increments of areal density equal to 700 g/m2 so
that the fabric consists of two areas:  a loaded area and an  unloaded area.
As deposition continues, the size of the  loaded area (dust load always equals
700 g/m2)  grows uniformly with time.  In  the uniformly loaded case,  the dust
load per unit area increases uniformly  with time.
                         10
                              Fraction of Filter Surface Covered
                              (Non-uniform loading curve only)
                                0.2   0.4   0.6    0.8    1.0
                               fraction of surface loaded to 700 g/m*
                                            Nmin
                                           15
                               dean surface drag
                            Sd = drag with 700 g/m2 loading
                                   N min
                                     Uniform
                                  Loading Curve
                                           Highly Non-uniform
                                              Loading Curve
                            6	100  200 300 400  500  600  700
                                Average Fabric Loading, g/m2
         Figure 2.  Impact of  non-uniform dust loading  upon drag (19)
                                      335

-------
         Uniform Loading
       Highly
Non-uniform Loading
                              a) Zero load
                              b) Initial deposition
                              c) Half final load
                              d) Final loading
     — Fabric surface
  Deposited dust
                        Figure 3.  Loading schematic.
     The curves of Figure 2 show that, for the drags assumed for the loaded
and unloaded areas, the net filter drag of the highly non-uniform case is not
only lower than that of the uniformly loaded case but (over much of the range
of dust loads) dramatically so.  Not until the non-uniform case begins to run
out of area does its dS/dM exceed that of the uniform case.  This rapid rise
in S brings the two curves together at the same final loaded condition (the
non-uniform case becomes uniform at 700 g/m2 loading when all its area is
loaded), but the route over which that same end point is reached differs
dramatically.

     Non-uniform loading via electrical forces has the effect of changing the
conventional uniform loading drag characteristic into one more like the
non-uniform loading curve illustrated in Figure 2.  Chiang et al. (18) have
mathematically modeled pressure drop reduction for both step changes in dust
cake areal density and bell-shaped distributions in support of non-uniform
deposition as a mechanism of electrically enhanced fabric filtration.

     This non-uniform loading mechanism is distinct and independent from the
tall trees mechanism.  Both the tall trees model and the non-uniform loading
model can predict the observed results, and some experimental confirmation of
each now exists in the literature.  Which dominates when is important in
choosing hardware for electrical enhancement.

     One further option should be resurrected—that of the electric field
aligned parallel to the gas flow (perpendicular to the fabric surface).  An
                                     336

-------
embodiment that is both easy to implement and compatible with reverse-air
baghouse design is the suspension of a corona wire down the center of a bag
which is itself either conductive or else incorporates conductive electrodes
woven into its fabric.  This configuration can operate in either an ionizing
or non-ionizing mode; its main advantage, however, is that it is compatible
with dusts which otherwise, with electrodes in the plane of the fabric, would
draw prohibitively large currents.  Note, however, that such a configuration
creates the highest field and the highest field gradient in the vicinity of
the corona wire.  This feature means that this configuration will probably be
most effective when operated as an electrostatic precipitator (ESP), gener-
ating its own charges for the dust and drifting this charged dust by Coulomb
forces to the fabric surface for capture.  In this mode the equipment then
operates as a conventional wire-duct ESP in which the collector surface is a
fabric filter and the air flow is through the collector plate rather than
parallel to it.  It is in effect a hybrid ESP/fabric filter.

SUMMARY OF MECHANISMS

     Mechanisms capable of predicting electrically enhanced filtration in-
clude:

     1.   Prefiltering, because of electrostatic precipitation.
     2.   The "tall trees" model, a depth distribution effect.

     3.   Non-uniform deposition, a surface distribution effect.

Other candidate mechanisms, such as electroclamping  (20), exist.

     Comparing these mechanisms with the choices of precharger configurations
(Table 1) and field electrode configurations (Table 2) leads to certain
general conclusions:

     1.   Prefiltering applies only to precharger location; the close-coupled
          precharger minimizes its significance.
     2.   The tall trees model depends strongly on fabric properties and
          electrode location.  It postulates enhanced dust capture by up-
          stream fibers, so depends on any variable that contributes an
          electrical capture force such as fiber-scale electric field in the
          upstream region, and dust electrical charge.  This mechanism is
          compatible with either Coulomb forces or polarization forces.

     3.   Non-uniform deposition also depends on electrical forces but re-
          sponds over a bigger volume—the electrical forces in the upstream
          regions remote from the near-fabric-surface volume, most likely
          Coulomb forces, are important.  Distinct from the tall trees mecha-
          nism, it does not depend on fabric surface finish and should be
          less dependent on fabric properties, a distinct advantage.

     Any specific hardware design for electrical enhancement depends on
understanding the mechanism by which the electrical enhancement occurs.  In
general this understanding does not yet exist.  What typically is known is
                                     337

-------
that electrifying a given conventional operation produces certain improve-
ments in performance.  Reports of such improvements are frequently accom-
panied by speculation as to the enhancement mechanism, such as those previ-
ously listed, but established proof is scarce.

     Against such a background and with the uncertainties it implies, a few
tentative guidelines can be suggested (Tables 3 and 4).
                  TABLE 3.  PRECHARGER DESIGN CONSEQUENCES
     Configuration
     Dust
     Remarks
   Reduction
     Close coupled

     Any
Any
Small (<3-5 |Jm),
nonconductive
No K2 Reduction
     Remote or separate  Any size, conductive

     Remote or separate  Large (>5 Mm) >
                         nonconductive
Non-uniform deposition

Coulomb force aided (field
self-created by collected
particles)
                         No self-created field

                         Inertial capture forces
                         dominate Coulomb forces
                   TABLE 4.  ELECTRODE DESIGN CONCLUSIONS
      1.   Downstream electrodes, incorporated into the support cage, match
          up best with outside-in filters.

      2.   Woven-in electrodes are the preferred configuration for inside-out
          filtration.

      3.   Upstream filters, while electrically advantageous, do not justify
          their added cost at present.
      1) While any of the precharger locations described can serve as a pre-
filter, the close-coupled configuration is the only one guaranteed to decrease
the dust-cake resistance because it is the only one which precipitates dust
on the bag surface and capitalizes on the non-uniform deposition mechanism
(2, 17, 18).  For the special category of non-conductive small particles
                                     338

-------
 (less  than  say  3  to  5  (Jm),  the  self-induced  electric  field  at  the  fabric
 surface  could also produce  a  reduced K^  using  any  of  the precharger  configu-
 rations  (8,  9,  21).

     2)  With remote  or separate precharger configurations„  the only  guaran-
 teed mechanism  of electrical  enhancement is  that of a prefilter.   Only when
 filtering small non-conductive  dust particles  is the  electrical enhancement
 likely to include a  K2 reduction (9).  The prefilter  function  is itself often
 worthwhile,  especially if it  entails no  additional cleaning hardware, hoppers,
 or  dust  removal,  as  is true in  the EPA/RTP pulse-jet  baghouse.

     3)  When corona  prechargers are used, they will be  subject to  operating
 limitations  similar  to those  of existing ESPs  with regards  to  back corona.
 Note,  however,  that  poor particle charging and low-efficiency  collection  can
 be  better tolerated  in a precharger than in  an ESP.   Thus,  a precharger
 should not  be ruled  out simply  because an ESP  has  been  ruled out.

     4)  Bag electrodes that create an electric field  in the plane  of the
 fabric are  simple to build  and  use when  they are incorporated  into the bag or
 as  part  of  the  cage.   This  consideration favors the electrode/cage assembly
 for pulse-jet operations and  woven-in electrodes for  reverse air (printed
 electrodes,  not yet  successfully developed,  also share  this advantage [2]).

     The performance advantages cited for upstream electrode placement do not
 justify  their added  expense.  Ruggedness and long  field life would need to be
 included in any demonstration to alter this  conclusion.

     5)  Configurations that favor non-uniform  deposition  such  as the woven-in
 electrode fabric  and the close-coupled precharger  described by Lamb  (2)
 appear less  dependent  on fabric properties and electrode  location  than those
 relying  on  the  tall  trees mechanism; that is,  dust non-uniformities  (large
 with respect to fiber  or dust particle dimensions) are  less dependent on
 fiber  or dust properties than those that depend on dust-cake porosity changes.
 The fields  and  electrical charges needed to  create gross  non-uniformities are
 easier to reproducibly create in the laboratory than  the  fiber-scale fields
 and charges.

     6)  A practical  field-electrode configuration  compatible with  conductive
 dust ESFF,  while  not yet demonstrated, appears to  be  available based on a
 variation of one  embodiment contained in Sharlit's patent  (22) or  as discus-
 sed by Frederick  (23).  This  concept is  the  suspension  of a corona wire down
''the centerline  of a  reverse-air bag.  The counter  electrode is the bag itself
! (coated  with conductive particles) or electrodes woven  into the bag. No
 conductive  path forms  between the high voltage and the  bag  electrodes.  When
 in  corona,  this configuration operates as a  hybrid ESP/fabric  filter.

                              ACKNOWLEDGEMENTS
      It  is  a pleasure  to  acknowledge  the  contribution of  George  E.R.  Lamb  and
 his  coworkers  at  Textile  Research Institute  who hosted one  of the  authors
 (R.P.  Donovan)  for  an  extended visit  during  which certain of the ideas  ex-
 pressed  in  this paper  were  crystallized or refined.

                                     339

-------
                                 REFERENCES
1.   Helfritch, D.J.  Apitron®.   Product Brochure, Apitron Division, American
     Precision Industries, 12037 Goodrich Drive, Charlotte, NC  28217, 1979.

2.   Lamb, G.E.R.  Influence of particulate precharging on the performance of
     an electrically stimulated fabric filter (ESFF).   Paper presented at
     World Filtration Congress III, The Filtration Society, Downingtown,  PA.
     September 14-16, 1982.

3.   Donovan, R.P., Hovis, L.S., Ramsey, G.H.,  and Abbott, J.H.  Pulse-jet
     filtration with electrically charged flyash.  In:  Third Symposium on
     the Transfer and Utilization of Particulate Control Technology, Vol. I.
     EPA-600/9-82-005a (NTIS PB 83-149583), July 1982.

4.   Ariman, T. and Helfritch, D.J.  Pressure drop in electrostatic fabric
     filtration.  In:  Second Symposium on the  Transfer and Utilization of
     Particulate Control Technology, Vol. III.   EPA-600/9-80-039c (NTIS PB
     81-144800), September 1980.

5.   Chudleigh, P.W. and Bainbridge, N.W.  Electrostatic effects in fabric
     filters during build-up of the dust cake.   Filtration and Separation.
     309-311, July/August 1980.

6.   linoya, K. and Mori, Y.  Experimental advances in fabric filtration
     technology in Japan—effects of a corona precharger and relative humidity
     on filter performance.  In:  Second Symposium on the Transfer and Utili-
     zation of Particulate Control Technology,  Vol. III.  EPA-600/9-80-039c
     (NTIS PB 81-144800), September 1980.

7.   Hovis, L.S., Abbott, J.H.,  Donovan, R.P.,  and Pareja, C.A.  Electrically
     charged flyash experiments in a laboratory shaker baghouse.  In:  Third
     Symposium on the Transfer and Utilization  of Particulate Control Tech-
     nology, Vol. I.  EPA-600/9-82-005a (NTIS PB 83-149583), July 1982.

8.   Loeffler, F.  The influence of electrostatic forces and of the proba-
     bility of adhesion for particle collection in fibrous filters.  In:
     T. Ariman (ed.), Novel Concepts, Methods and Advanced Technology in
     Particulate-Gas Separation.  University of Notre Dame, 1978.

9.   Yu, H.S. and Teague, R.K.  Performance of  electrostatic fiberbed.  Paper
     presented at 1st Annual Meeting of the American Association for Aerosol
     Research, Santa Monica, CA.  February 1982.

10.  Havlicek, V.  The improvement of efficiency of fibrous dielectric fil-
     ters by application of an external electric field.  Int. J. Air and
     Water Poll.  4(3/4): 225-236, 1961.

11.  Lamb, G.E.R.,  Costanza, P.A., and O'Meara, D.J.  Electrical stimulation
     of fabric filtration.  Part II:  Mechanism of particle capture and
     trials with a  laboratory baghouse.  Textile Research Journal.  48(10):
     566-573, October 1978.

                                     340

-------
12.   Zebel,  G.   Deposition of aerosol flowing past cylindrical fiber in a
     uniform electric field.   J.  Colloid Sci.   20, 1965.

13.   Lamb,  G.E.R.  and Costanza,  P.A.   Electrical stimulation of fabric  fil-
     tration.   Textile Research  Journal.  47(5): 372-380,  May 1977.

14.   Lamb,  G.E.R.  and Costanza,  P.A.   A low-energy electrified filter system.
     Filtration and Separation.   319-322,  July/August 1980.

15.   Greiner,  G.P.   Electrical stimulation of fabric filtration.   Paper pre-
     sented at APCA 74th Annual  Meeting, Philadelphia,  PA.  June 1981.

16.   Greiner,  G.P., Furlong,  D.A.,  VanOsdell,  D.W., and Hovis, L.S.   Elec-
     trostatic stimulation of fabric  filtration.  J. Air  Pollut.  Control Assn.
     31:10,  1125-1130, October 1981.

17.   Penney, G.W.   Using electrostatic forces to reduce pressure drop in
     fabric filters.   Powder  Technology.  18:  111-116,  1977.
                                                /•
18.   Chiang, T-K.,  Samuel, E.A.,  and  Wolpert,  K.E.  Theoretical aspects of
     pressure  drop  reduction  in  a fabric filter with charged particles.  In:
     Third  Symposium on the Transfer  and Utilization of Particulate  Control
     Technology, Vol. III. EPA-600/9-82-005c (NTIS PB 83-149609), July 1982.

19.   Dennis, R., et al.   Filtration model for coal flyash with glass fabrics.
     EPA-600/7-77-084 (NTIS PB 276489), August 1977.

20.   VanOsdell, D.W., Donovan, R.P.,  Furlong,  D.A., and Hovis, L.S.   Permea-
     bility of dust cakes collected under the influence of an electric  field.
     Paper  presented at The Fourth  Symposium on the Transfer and Utilization
     of Particulate Control Technology, U.S.  Environmental Protection Agency,
     Houston,  Texas.   October 11-15,  1982.

21.   Bergman,  W., Hebard, H.D.,  Taylor, R.D.,  and Lum,  B.Y.   Electrostatic
     filters generated by electric  fields.  Proceedings of The Second World
     Filtration Congress.  1979.

22.   Sharlit,  I.E.   Filter system.  U.S. Patent 3,577,705 (to Hitco), May 4,
     1971.

23.   Frederick, E.R.   Some effects  of electrostatic charges in 'fabric filtra-
     tion.   J.  Air  Pollut. Control  Assn.  24(12): 1164-1168,  December 1974.
                                     341

-------
                  PERMEABILITY OF DUST CAKES COLLECTED UNDER

                      THE INFLUENCE OF AN ELECTRIC FIELD

            By:  D. W. VanOsdell and R. P. Donovan
                 Research Triangle Institute
                 P. 0. Box 12194
                 Research Triangle Park, North Carolina  27709

                 D. A. Furlong
                 ETS, Inc.
                 Roanoke, Virginia  24018

                 L. S. Hovis
                 Industrial Environmental Research Laboratory
                 U. S. Environmental Protection Agency
                 Research Triangle Park, North Carolina  27711
                                   ABSTRACT

     The reductions in dust cake flow resistance and residual pressure drop
that can be achieved by electrostatic stimulation of fabric filtration (ESFF)
have been documented.  The changes in the dust deposition characteristics
which cause these improvements are not well understood.  This paper presents
results from three investigations into the nature of dust deposits collected
with ESFF.  Permeability data collected at various axial positions along
ESFF and conventional bags were the basis of the first study.  The second
study concerned the deposition patterns of dust collected on a laboratory
filter when ESFF was in use, and the third, the resistance to compression of
the collected dust with and without an electric field.

     These investigations gave evidence that ESFF caused the formation of a
residual dust deposit which was not axially uniform (the permeability at the
bottom of the bag was half that at the top) and was more easily removed than
a conventional dust deposit.  In the laboratory, it was shown that ESFF caused
newly collected dust to form very nonuniform deposits.  The dust tended to
collect near the electrodes, in patterns which have not been observed in the
pilot units.  In the third study, the ESFF electric field was found to improve
the resistance to consolidation of the dust cake collected in a pulse-jet
pilot unit.

     This paper has been reviewed in accordance with the U. S. Environmental
Protection Agency's peer and administrative review policies and approved for
presentation and publication.


                                     342

-------
                                 INTRODUCTION

     The use of electrostatic fields to improve the performance of fabric
filters is a topic of considerable research interest today.  Numerous re-
searchers have investigated the topic, applying electrostatics in different
ways.  Particle charging, accumulation of charge on the fabric filter, and
the application of external electric fields (either perpendicular or paral-
lel to the gas flow) have all been used singly or in combination during
this research.  In all cases, it has been found that the performance of the
filter has been improved in the sense that the drag due to the collected
particles was reduced.  Generally, an improvement in collection efficiency
was also noted.  Although much has been learned about the beneficial effects
of ESFF, the mechanismsjby which it operates in a baghouse are not well
understood.  The experimental evidence is difficult to reproduce and some-
times contradictory.  One important question is whether the ESFF filter de-
posit is nonuniform at the size scale of the bag, at the scale of the elec-
trode spacing, or at the scale of the particles.  Treating the filter as an
assembly of resistances in parallel, at any scale', can account for the ESFF
effect mathematically.(1)  Another question is whether the reduced dust cake
drag attributed to ESFF is due to the structure of the dust cake as col-
lected, or to some other phenomenon related to the electric field.

     This paper describes three experiments with electrostatically stimulated
fabric filtration (ESFF) which cast light on the mechanisms by which ESFF
causes reduced filter drag.  All three experiments utilized the same concept
of ESFF--no particle charging and an electric  field applied at the filter
surface and oriented parallel to the surface.  This concept was initially
developed by  the Textile Research Institute(2,3,4) and was extended to pilot
scale by the  Research Triangle Institute  (RTI) and ETS, Inc.(5,6)

     The first experiment to be described was  conducted on bags from the
Waynesboro ESFF pilot unit.  The permeability  of the cleaned  bags  (both ESFF
and  conventional) was determined at the top, bottom, and middle areas of the
bags.  In addition, the particle size  distribution of the  dust collected at
the  top and bottom  of an ESFF bag was  determined.  The second experiment con-
sidered the appearance of dust deposits collected with and without electric
fields.  A  small  laboratory  filter was used to remove  resuspended  fly ash
from an air stream, the pressure drop  characteristics were measured, and the
filter deposits photographed.  The  third  experiment investigated the  com-
pressibility  of dust  collected with and without an electric field.  This ex-
periment was  conducted on a  pilot-scale pulse-jet  fabric  filter operating  on
 fly ash  resuspended in ambient air.

                  FILTER BAG AXIAL  PERMEABILITY MEASUREMENTS

EXPERIMENTAL  ARRANGEMENT

      The  experiments  reported  in this section  were  conducted  on bags  from  the
 pilot-scale ESFF  baghouse  located  at  Waynesboro, Virginia. This baghouse  has
 been described in earlier papers(5,6), and only a  brief  description  is  pre-
 sented here.   The pilot  unit included two identical baghouses operated  as
 parallel  units with identical  inlets.   One baghouse utilized  ESFF;  the  other


                                      343

-------
was conventional.  The pilot unit was operated on a slipstream from an indus-
trial bo.iler house.  A schematic diagram of the pilot unit is given in
Figure 1.  The boiler burned pulverized coal, and the coal supply was highly
variable; the average sulfur content was about 1.3 percent and the ash con-
tent was about 13 percent.  Boiler operation was continuous.  The average
inlet temperature to the baghouse was about 155° C (310° F), with an average
dust loading of about 0.9 g/sm3 (0.4 gr/scf).

     The pilot baghouses (both conventional and ESFF) each contained four
bags in a dirty-gas-inside, reverse-air-cleaned configuration.  The filter
bags were made of J. P. Stevens 648 fabric, a 570 g/m2 (16.8 oz/yd2) woven
fiberglass cloth with a 10-percent Teflon  B coating.  The initial permea-
bility of the fabric was 18 to 28 cm/s at 0.125 kPa (35 to 55 ft/min at
0.5 in. H20).  The electrostatic augmentation of the ESFF bags was provided
by a vertical array of 3.2-mm (0.125-in.) wires, connected to provide alter-
nating grounded and high negative potential electrodes (Figure 2).

TEST PROGRAM

     The pilot baghouses were used to test the long-term behavior of reverse-
air-cleaned fiberglass bags under ESFF conditions.  The operational results
of this test have been reported(5), and this paper discusses permeability
results obtained after the bags had been removed from the baghouses.  Re-
sults from two bag sets are reported.  One bag set (to be referred to as JF)
was in use during January and February 1982.  The second set (June) was in
use during June 1982.  All bags were conditioned at 1 cm/s (2 ft/min) and
were operated at about 2 cm/s (A ft/min).  The electrical field was applied
to the ESFF bags from start-up and was maintained continuously at about
3 kV/cm.  During the JF test period, the conventional baghouse was used as a
control on the ESFF experiment.  During the June test, both baghouses were
in use in different ESFF arrangements, and control bags were not available.

     The permeability tests reported here were run on a standard bag permea-
bility test apparatus.(7)  The JF bags were in a "just-cleaned" state when
tested.  The June bags were removed before cleaning, taken to the laboratory
to remove dust samples, then extended to allow the remaining dust cake to
fall off prior to permeability testing.  Permeability tests were run on the
fabric in an "as-received" condition and following vacuum cleaning at 3.75 kPa
(15 in. H20) vacuum and at 7.5 kPa (30 in. H20) vacuum.  These tests were
done for both the conventional and ESFF bags.

     The particle size distribution tests were run on dust samples from the
top and bottom of the June bags by resuspending and analyzing the dust in a
Bahco apparatus.

RESULTS

     The results of the permeability tests are presented in Table 1, and the
results of the Bahco size analysis in Figure 3.  Features of these results
which merit special attention are discussed below.
                                     344

-------
UJ
                                                                               \ OUTLET (TOP)
                                                                            CONTROL
                                                                             HOUSE
                                                                               )  INLET (BOTTOM)
                                                                               ,, INLET
                                                                                 (BOTTOM)
                              Figure  1.  Schematic diagram of ESFF pilot unit,
                                                                           OUTLET
                                                                           (TOP)

-------
      Top Electrodes at
   High Negative Potential
   Clamp
                                    Insulating Block
Standard
Top Cap
                                                    Glass Fabric
                                                  Tape Electrode
                                                     Support
Approximate
   Top of
  Thimble
     Bottom Electrodes
         Grounded
   Figure  2.  Pilot-unit reverse-air bag  electrode  array.
                              346

-------
       TABLE 1. PERMEABILITY RESULTS FOR ESFF AND CONVENTIONAL BAGS*
Pet-mobility Resurts-ft/min e 0.5 in. H^ AP
(cm/ie0.125kPaAP)
Simple Descriptor
As received:
Top
Middle
Bottom
Vacuumed @ 3.75 kPa
(15 in. H20):
Top
Middle
Bottom
Vacuumed @ 7.5 kPa
(30 in. H2O):
Top
Middle
Bottom
January- February (JF)
Conventional Bag

3.8(1.9)
2.6(1.3)
3.3(1.7)


3.8(1.9)
5.5 (2.8)
2.9 (1.5)


7.4 (3.8)
8.3 (4.2)
5.7 (2.9)
January-February (JF)*
ESFF Bag

5.5 (2.8)
3.6(1.8)
1.6(0.8)


11.8(6.0)
6.7 (3.4)
4.4 (2.2)


13.6(6.9)
9.7 (4.9)
7.5 (3.8)
June*
ESFF Bag

4.5 (2.3)
IM.M.I
3.6(1.8)


11.5(5.8)
N.M.§
10.5 (5.3)


17.5(8.9)
N.M.§
12.5(6.4)
     * All test work performed by ETS, Inc.
     * January-February results from one bag.
     * June results are average from two bags.
     §N.M. = not measured.
             S
             35
   99

   98


   95


   90


   80
I-
1  »
I  40
|  30
5
   20

   10
    5
                                Ash Sample from
                                Top of Bag    /     O
                                               Ash Sample from
                                               Bottom of Bag
                                   3    45678
                                    Particle Size, jim
                                                     10
                                                                20
Figure  3.   Bahco size  distribution  of dust  from  top  and  bottom of
              June ESFF bag.
                                    347

-------
Axial Variation of Permeability of ESFF Bag

     The permeability at the top of the ESFF bag was consistently greater
than at the bottom, both before and after vacuuming and for both ESFF bags.
This is especially evident for the JF ESFF bag.   The permeability of the JF
conventional bag was lowest at the bottom, highest in the middle or top, and
did not vary as much between positions as did the JF ESFF bag.

Magnitude of Permeability

     The average of the three permeability measurements for the ESFF bag in
the as-received state was 11 percent greater than that of the conventional
bag.  It was 87 percent above the conventional bag permeability after the
first vacuuming and 44 percent higher after the second.

Ease of Cleaning

     The JF ESFF bag had an average permeability before vacuuming of
3.6 ft/min (1.8 cm/s).  Following vacuuming at 3.75 kPa, the average perme-
ability had risen to 7.6 ft/min (3.8 cm/s), an improvement of 11 percent.
For the JF conventional bag, the same treatment increased the permeability
from 3.2 ft/min (1.6 cm/s) to 4.1 ft/min (2.0 cm/s), an increase of
28 percent.

Axial Variation of Dust Deposit Size Distribution

     The size distribution of the dust at the top of the ESFF bag differs
from that at the bottom (Figure 3).  Mass mean diameter at the bag top was
2.6 JJm and at the bottom, 3.7 pm.  As no corresponding data are available for
the conventional pilot plant bags and others(8) report similar trends for
conventional baghouses, this variation cannot be ascribed to the ESFF effect.

                   DUST COLLECTIONS ON A LABORATORY FILTER

EXPERIMENTAL APPARATUS

     The laboratory filter deposition study was conducted on the apparatus
shown in Figure 4.  As shown, the filter is oriented face downward and is
10 cm square.  The electric field was provided by six parallel electrodes
spaced 2 cm apart and located on the clean side of the filter.   The test dust
was resuspended fly ash, fed by a screw feeder into a sonic air jet, blown
against an irapaction plate, and then into the inlet duct.  The tests were
conducted with ambient air as the carrier gas.

TEST PROGRAM

     Fly ash from the Waynesboro pilot unit, resuspended in a sonic air jet
at about 4 g/m3 (1.8 gr/ft3), was fed to the filter at a face velocity of
2 cm/s (4 ft/min).  Cascade impactor size analysis indicated a mass median
aerodynamic diameter of about 3.1 pro for the resuspended ash.  An extractive
charge-to-mass measurement based on a filter within a Faraday cup indicated
an average net positive charge on the resuspended ash of 0.57 MC/8-  The


                                     348

-------
Sample ,
Point
——

Samp
Poin
r
$
'I
•f 	
?
1-1

,tefe
T
Dust
Injection





1
i
j[

•9
\
}
Ji





— r
i
i
i
ij
Test
Filter
Holder
Normal
. Gas
/^^ How
Valve
|
i
Ik






sss
ss
0
d
J IK -'II LL
\
r Valve |f
AF°I
Valve
l
]
l
L«_J Clean
in Air
" 1 Inlet
^
J Valve




L

II-' )
ji



nf=
i
= *t|— { Pump and
— 11— J Pinal Piltar
Valve
Figure 4.  Laboratory filtration apparatus,
                  349

-------
filter was of glass fiber, and a new filter was used for each test.   The
filter was weighed before and after the run to get an approximate measure
of the cake mass.  Run time was 100 min for these experiments, and the total
dust cake mass was between 4 and 5 g.

RESULTS

     The dust accumulations on the filter were nonuniform when the electric
field was ON, becoming increasingly nonuniform as the field increased.  Fig-
ure 5 is a photograph of the dust cake when collected with the field OFF.
The faint surface lines were produced by turning the field ON after the dust
cake was collected, with clean gas flow through the filter.  Figure 6 shows a
deposit collected with an electric field of 1 kV/cm, and Figure 7, a deposit
collected with a 4-kV/cm electric field.  The pressure drop behavior of each
deposit is shown in Figure 8; as expected, the dust cakes with the field
ON have a lower pressure drop than with the field OFF at any given time.

     Dust samples from the positive and negative electrodes of a 4-kV/cm
run were analyzed by a Coulter counter, along with a sample from between the
electrodes.  The size distribution curves from the three sites were identi-
cal; the mean diameter was 2.37 pm from the Coulter analysis.

          INFLUENCE OF AN ELECTRIC FIELD ON DUST CAKE COMPRESSIBILITY

INTRODUCTION

     The compaction of filter dust cakes due to the increasing pressure drop
across the cake has been described.(9)  Increased interparticle adhesion
within a particle bed due to an external electric field has also been docu-
mented and discussed.(10,11)  Increased interparticle adhesion has the po-
tential of being an important ESFF mechanism.  If a large fraction of the
pressure drop across a filter cake is caused by the continual compression of
the cake, then interparticle forces which prevent or reduce cake compression
would reduce the net pressure drop much as is observed with ESFF.  Some evi-
dence for a mechanism of this sort can be observed in a pilot unit pressure
drop characteristic curve.  Figure 9 compares the AP for the buildup of a dust
cake collected with the electric field ON to a conventional field-OFF curve.
While the reduced AP of the field-ON curve is evident throughout, the field-ON
curve exhibits a continually increasing slope which includes a series of steps
at high pressure drops—indicating dust cake consolidation by discrete cake
collapses.  The field-OFF curve also shows these steps but not as many and
they are reduced in magnitude.  This difference between the two curves sug-
gests that the field may  increase interparticle adhesion sufficiently to
support a more porous dust cake initially but that at least part of this ad-
vantage can be lost through cake compression.  The discussion below reports a
test series which investigates the drag of dust cakes collected with and
without an electric field when subjected to stress from increased filter AP.

EXPERIMENTAL APPARATUS

     The experiments were conducted on a small pulse-jet fabric filter lo-
cated at the Industrial Environmental Research Laboratory of the U. S.


                                      350

-------
Figure 5.  Patch filter dust cake collected at 0 kV/cm.

                         'i *'
Figure 6.  Patch filter dust cake collected at  1 kV/cm.
                           351

-------
                                       ,      ,
                                       • -<-
                                            *U-
       Figure  7.  Patch filter dust  cake collected  at 4 kV/cm.
  1.25 r-
  i.oo
  0.75
fe 0.50
  0.25
                                  OkV/cm
                                                    4 kV/cm
                         I    i   I
     0   5  10   15  20  25  30  35  40  45  50   55  60  65  70  75  80  85  90  95  100
                            Time From Start of Filtration Run, min

     Figure 8.   Pressure drop  characteristics of laboratory  filter
                 at constant dust rate.
                                    352

-------
          1.00
           0.75 -
         £
         eC
          0.50  -
           0.25 -
                  10
                        20
                              30     40    50    60

                                  Filtration Time, min
                                                     70
80
      90
          Figure 9.  Long filtration time pressure  drop  curve.
Environmental  Protection Agency in Research Triangle Park,  North Carolina.
This  particular pilot baghouse has been described previously.(12)  In sum-
mary, the  baghouse used 11.4-cm-diameter (4.5-in.) bags,  122  cm (4 ft)  long,
made  of  Teflon  felt.  The test dust was resuspended fly  ash  at a concentra-
tion  of  about  6.4 g/ra3 (2.8 gr/ft3).  Three bags were  in  use.   All of the
work  was at  ambient conditions using air as the carrier gas.
TEST PROGRAM

     The  test program was a three-step procedure.  First,  a  dust  cake  was
formed on the bags—with or without an electric field depending on the test.
The dust  cake was  built until a predetermined pressure drop  had been reached
at the initial face velocity of 8.5 ft/min.  The ESFF effect caused the dust
cake in the  field-ON condition to require more time  (and more fly ash) to
build than was required for an equivalent AP dust cake in  the field-OFF condi-
tion.  Once  the dust layer was in place, the dust feed was stopped.  The dust
cake was  then stressed by increasing the flow rate to reach  a new,  higher,
predetermined AP.   The flow rate at the increased stress level varied  from
test to test; only the AP was fixed.  Following a fixed time period at the
high AP level, the flow rate was reduced to the initial flow rate (the flow
rate at which the  cake was built) and the AP recorded.  The  change in  AP at
the initial  flow rate is then indicative of the amount of  cake compression
which took place during the stress period.
                                     353

-------
RESULTS

     The  results of this test program are presented in Table  2.   The Aug-
ust 6th data,  which followed a thorough bag cleaning by vacuuming,  permits
a fairly  clear interpretation.  Depositing a dust layer with  the electric
field ON  results in a layer which proves more compressible  under a  fixed
field-OFF AP loading (Run 5) than a similarly stressed layer  deposited with
the field OFF (Run 6).  In fact, the post-stress AP across  a  layer  treated as
in Run 5  (deposit field ON, stress field OFF) is very close to what would be
predicted for depositing the same quantity of dust with the field OFF.  It is
as though the air stress step eliminated the drag differences between the
field-ON  and field-OFF depositions.  However, it is notable that leaving the
field ON  during the air stress step largely preserves the reduced drag of the
field-ON  deposit.

                                   CONCLUSIONS

     In summary, the results of these three experiments are:

           1.   The residual permeability of an ESFF bag varied axially to a
                greater extent than that of a conventional bag.  The ESFF bag
                was more easily cleaned by vacuuming and had a greater average
                permeability.

           2.   The dust deposits collected with ESFF can  be very nonuniform;
                nonuniformities of this scale have not been  observed in pilot
                unit operation.

           3.   A given pressure drop caused less dust cake  consolidation with
                ESFF than without; the reduced drag due to ESFF can  be elimi-
                nated by sufficient dust cake compression.
                       TABLE Z STRESS DATA FOR FIXED AP LEVELS
Run
1
2
3
Date
7/28
7/28
7/29
Deposit AP
kP»
(deposit time)
0.91
(~14min)
0.91
(~28 min)
0.85
(~28 min)
Strati AP
kPa
1.74
1.74
1.87
Flow at
high stress
nrVrnin
4.87
5.64
Unstable
Field status
Deposition
Off
On
On
Stress
Off
On
On
Change in AP
after stress
kPa Remarks
0.31
0.17
0.30
0.25 kPa overshoot on high stress level.

Unstable layer; repeated stress cycling
produced continuously increasing AP.
4

5

6

7

8/6

8/6

8/6

8/6

0.67
(28 min)
0.62
(23 min)
0.62
(12 min)
0.62
(14 min)
                      1.50

                      1.50

                      150

                      1.50
                                 {Bags Vacuumed, 7/30/82]
                             6.06     On     On
6.20

630

6.20
On

OH

Off
OH

Off

On
024

0.40

0.21

0.19
Bag voltage read as 0 by MOO COUP
but had appropriate bag current for
field On.
                                      354

-------
     The  conclusions which can be drawn from these experiments are tentative
because the  range of applicability is limited and the number of experiments
small.  Nevertheless, these experiments do increase the understanding of
the  effect of ESFF on dust deposits and indicate fruitful subjects for
furthe r s tudy.

BAG  PERMEABILITY VARIATION

     The  vertical variation in "cleaned-bag" permeability evidenced by the
ESFF bag--and the lack of variation for the conventional bag—can be ex-
plained in several ways.   First,  the difference could be due simply to a
smaller residual deposit  of fly ash at the top of the ESFF bag—the electric
field causes the dust to  be collected at a lower average height in this bot-
tom-feed  unit.   A second  explanation would be that the same amount of fly ash
was  present  but that it was collected and held in a more porous structure at
the  top than at the bottom.  Another explanation would be that the dust frac-
tionates  within the bag due to the electric field and that the fraction at
the  top has  characteristics which differ from those of the bottom fraction.
The  top-to-bottom size difference noted supports this third mechanism, but
the  phenomenon occurs in  conventional bags as well, and the present test
data are  very limited. The low permeability at the bottom of the ESFF bag in
the  as-received condition is a strong indication that more of the residual
dust is at the bottom of  the ESFF bag than for the conventional bag.  This
implies that most of the  dust collects at the bottom of the bag. The improved
cleanability of the ESFF  bag is consistent with the premise that ESFF causes
the  dust  to  collect on the upstream side of the fabric and not penetrate as
deeply into  the fabric.  This combination of the first and second hypotheses--
more dust at the bottom of the bag and collected in deposits more on the
upstream  side of the fabric when compared to the conventional bag—seems best
supported by the data.

NONUNIFORM DUST DEPOSITS

     The  extreme nonuniformity shown by the laboratory dust cake deposited in
the  presence of an electric field would certainly be adequate to account for
the  observed ESFF effect.  However, ESFF has also been observed with dust
deposits  having no clear  nonuniformity.  This has been true for the pilot
units, and it may be that bag cleaning and the redeposition of the cake pre-
vents the extreme pattern shown by the laboratory filter.  The size distribu-
tion analysis indicates that the nonuniformity of the deposits is probably
not  due to fractionation  bv particle size.

INTERPARTICLE ADHESION

     The  experiment on increased interparticle adhesion due to the electric
field demonstrates clearly that the reduced drag of an ESFF dust cake can be
eliminated by compressing the layer.  That compression removes the reduced
drag advantage for ESFF argues that this ESFF effect is more of a bulk struc-
tural effect than just nonuniform deposition.  The fact that the compression
is reduced by keeping the electric field ON is an indication that electro-
static  forces are increasing interparticle adhesion.  The role of the
                                     355

-------
clamping in producing the initial reduced drag—the ESFF effect itself—was
not addressed by the experiment.

                                  REFERENCES

 1.  Chiang, T- K., Samuel, E. A., and Wolpert, K. E.  Theoretical aspects
     of pressure drop reduction in a fabric filter with charged particles.
     In:  Third Symposium on the Transfer and Utilization of Particulate
     Control Technology, Volume III, EPA-600/9-82-005c (NTIS PB 83-149609),
     July 1982.  pp. 250-260.

 2.  Lamb, G. E. R., and Costanza, P. A.  Electrical stimulation of fabric
     filtration.  Textile Research Journal.  47:372, 1977.

 3.  Lamb, G. E. R., Costanza, P. A., and O'Meara, D. J.   Electrical stimula-
     tion of fabric filtration, Part II.  Textile Research Journal.  48:566,
     1978.

 4.  Lamb, G. E; R., and Costanza, P. A.  A low-energy electrified filter
     system.  Filtration and Separation.  17:319, 1980.

 5.  VanOsdell, D. W., Greiner, G. P.,  Lamb, G. E. R., and Hovis,  L. S.
     Electrostatic augmentation of fabric filtration.  In:   Third  Symposium on
     the Transfer and Utilization of Particulate Control  Technology, Volume I,
     EPA-600/9-82-005a (NTIS PB 83-149583), July 1982. pp.  35-44.

 6.  Furlong, D. A., Greiner, G. P., VanOsdell, D. W., and  Hovis,  L. S.
     Electrostatic stimulation of fabric filtration—an update.  Paper
     (No. 82-32.2) presented at the 75th Annual Meeting of  the Air Pollution
     Control Association, New Orleans,  Louisiana, June 20-25,  1982.

 7.  ASTM D737-66.

 8.  Smith,  W.  B., Gushing, K.  M., and  Carr, R. C.  Performance of a 10  MW
     fabric filter pilot plant and comparison to full-scale  units.   Paper
     No. 82-59.6 presented at the 75th  Annual Meeting of  the Air Pollution
     Control Association, New Orleans,  Louisiana, June 20-25,  1982.

 9.  Rudnick, S. N., and First, M. W.  Dust cake compaction  in fabric filtra-
     tion.   Paper No.  78-62.7 presented at the 71st Annual Meeting of the
     Air Pollution Control Association, Houston, Texas, June 25-30, 1978.

10.  Dietz,  P.  W., and Melcher, J. R.  Momentum transfer  in  electrofluidized
     beds.   Air Pollutants—NO  and Particulate Emissions.   AIChE  Symposium
     Series, No. 174,  74:166.

11.  Dietz,  P.  W.  Cohesive force  and resistivity between electrostatically
     precipitated particles.  Journal of Electrostatics.  6:273, 1979.

12.  Turner, J.  H.   EPA research in fabric filtration:  annual report on
     IERL-RTP in-house  program.  EPA-600/7-77-042 (NTIS PB 267441).  U.  S.  En-
     vironmental Protection Agency,  Research Triangle Park,  North  Carolina,
     May 1977.   39 pp.

                                    356

-------
     "HIGH VELOCITY FABRIC FILTRATION FOR INDUSTRIAL COAL-FIRED BOILERS"

              by:   Gary P. Greiner and Shannon Delaney
                   ETS, Inc.
                   Suite C-103,  3140 Chaparral Dr.,  SW
                   Roanoke, VA 24018

                   Lou S. Hovis
                   U.S. EPA
                   IERL, MD-61
                   Research Triangle Park, NC 27711
                                ABSTRACT

     Two parameters which dramatically affect the technical/economic
performance of a fabric filter system are gas/cloth (G/C)  ratio (or
filtering velocity) and bag life.

     Under an EPA sponsored contract, ETS, Inc. has been conducting a
study of state-of-the-art and experimental fabrics and finishes in
full-scale baghouses operating on coal-fired boilers at the Kerr
Finishing Plant, Travelers Rest, South Carolina.  The objective of the
study is to operate, test, and evaluate performance at G/C ratios up
to 10/1 and screen various fabrics with respect to pressure drop, col-
lection efficiency, and fabric strength characteristics necessary for
technical/economic analysis.  Included in the program are some revolu-
tionary fabrics which hold promise of technological breakthroughs.

     This paper will discuss evaluation methodology, performance results,
and future program plans.

     This paper has been reviewed in accordance with the U. S. Environ-
mental Protection Agency's peer and administrative review policies and
approved for presentation and publication.
                                    357

-------
                              INTRODUCTION

     ETS, Inc., through the sponsorship of EPA's Industrial Environmental
Research Laboratory, has been conducting a pioneering program for the appli-
cation of fabric filters to industrial coal-fired boilers since 1973.  Many
of the early lessons concerning this application were learned during the
pilot program and subsequent full-scale implementation.   Fabric compati-
bility, cleaning modes, component requirements and specifications, achie-
vable filtering velocities, and emission capabilities all were evaluated,
defined, and submitted to the industry.  This information and the success
of the full-scale systems did much to accelerate the application of fabric
filters to industrial coal-fired boilers.

     Now the program, continuing its pioneering approach, is seeking to
extend the capabilities of the fabric filter system in this application
through higher filtering velocities (G/C ratios) and longer fabric life,
while maintaining required emission levels.

                      SITE AND EQUIPMENT DESCRIPTION

     The present program is being conducted at the Kerr Finishing Plant
in Travelers Rest, South Carolina.  Figure 1 shows the general site lay-
out.  Baghouse No. 1 controls emissions from a 60,000 Ib /hr, Babcock &
Wilcox (B&W) stoker boiler.; Baghouse No. 2 controls emissions from a
bank of four boilers (one 30,000 Ib/hr and three 20,000 Ib/hr).  The coal
used at this site is typical Eastern low sulfur coal.  The operation is
cyclic depending on production requirements, and is shut down each weekend.

     The current project consists of two programs.  The first, referred to
as the High G/C Program, utilizes Baghouse No. 1 arid seeks to raise the
stable filtering velocity operating point to provide lower overall cost.
The second, referred to as the Fabric Evaluation Program, utilizes Baghouse
No. 2 and seeks to evaluate the bag strength characteristics versus expo-
sure time of various fabric and finish candidates that could be applicable
to both pulse jet and reverse air collectors.

                             HIGH G/C PROGRAM

Boiler Description
     The No. 1 baghouse is applied to a Babcock & Wilcox coal-fired boiler
rated at 60,000 Ib/hr steam.  A stoker grade coal (1% X 1/4 in.) is fed
to a spreader type Detroit stoker from a central weighing scale into three
stoker sections and onto a riser type dump grate.  Normal maximum boiler
operation provides 52,000 Ib/hr steam at 160 psig and saturation tempera-
ture.  A multi-cyclone is located on the boiler outlet, but no economizer
or fly ash reinjection is provided.  The combustion air flow is provided by
a combination of forced draft, overfire air, and induced draft fans.  Nor-
mal operation produces 25-40,000 acfm and 400-450°F flue gas.
(*) Readers more familiar with metric units may use the conversion factors
    at the end of this paper.
                                   358

-------
Baghouse Description

     The No. 1 baghouse accommodates up to 648 bags with a nominal size
of 5 in. diameter X 104 in. long, having a total cloth area of 6800 sq ft
for a maximum filtering velocity of 5.8 fpm with all bags installed.  Its
cleaning energy is supplied by 80-100 psi compressed air pulses assisted
by low-pressure reverse-flushing air.  The original design had 18 separate
bag plenums and outlet dampers, but with common inlet and hopper areas.
This provided a semi-off-line cleaning mode, but not of the type seen in
today's modular collectors.  It also lacked the ability to monitor and
control the flow through each section.  Modifications were performed to
sectionalize the inlet and hoppers.  The sectional inlet dampars, test
ports, flow monitors, and differential pressure (A?) manometers permit
the control and monitoring of the gas flow into each section.  The resul-
tant system provides a high degree of cleaning mode flexibility.  Cleaning
can be on-line, pulse jet only; on-line, pulse jet with varying amounts of
reverse-flush assist; or off-line variations of the above.  Normal opera-
tion is to take an individual cell off-line by closing the outlet damper,
and allowing a minimum volume of reverse gas to flow, primarily to block
the forward leakage flow am  pulse the bags.  This most closely simulates
today's modular off-line cleaning pulse jet collectors.

Fabric Description
     All fabric in baghouse No. 1 is a form of Teflon   fiber felt,  sup-
plied by the E. I. Dupont de Nemours Company.  A description of the  fabrics
and bags utilized is shown .in Table 1.  The prior EPA program utilized a
full house of 23 oz TefIon^felt bags.  Section C, containing these ori-
ginal bags, has been retained to provide reference with past operation.
Section F contains an experimental 17 oz TefIon ^ felt bag set that  is
constructed in an asymmetrical manner, such that most of the felt is on
the outside o£-sthe bag.  Sections A, B, D and E contain bag sets of  experi-
mental Teflon^ fiber felt of varying weights and construction.  A charac-
teristic of Teflon^ is that it physically shrinks with increasing tempera-
ture.  This results in not only a smaller bag,, but also a lower in-use
permeability as shown in Figure 2.  Heat setting at sufficient time  and
temperature will minimize this problem; however, these characteristics
must be accounted for in Jthe initial fabric and bag specifications.   Note
that the original Teflon^bags were not oversized as much as the newer
bags (4-7/8 in dia. versus 5-1/4 in.).  This was due to a lack of under-
standing of the actual shrinkage rate at the time (in 1976) that these
bags were constructed.  Unfortunately, all of these fabric styles were
not obtainable at the beginning of the test program; they were installed
as they became available.  The installation dates are included in Table 1.
   Dupont TFE Fluorocarbon fiber
                                   359

-------
                   TABLE 1.  BAG AND FABRIC DESCRIPTION
Desdrlption
Baghouse
Section3
A
B
C X'
D
E
F
Fabric^
(Teflon^)
Exp. (Yl)
Exp. (X)
: 2363 (Std)
Exp. (Y2)
Exp. (X)
Exp . (Asym)
Wt.
oz/yd
21
23
27
19
23
17
Perm
fpm
36
21
27
42
21
59
Mullen
Burst
psi
354
340
306
324
340
231
Tensile St.
Ibs/in Width
Warp
87
88
115
84
88
83
Fill
78
70.5
93
74
70.5
61
Bag Size
dia. X length,
in.
5% X 108
5% X 108
4-7/8 X 108
5k X 108
5% X 108
5k X 108
Initial
Installation
Date
10/81
07/81
01/77
10/81
07/81
05/81
Location of section shown in Figure 4.

-------
TECHNICAL OBJECTIVES

     The program's primary objective was to develop methods of raising
the stable filtering velocity operating level and, after demonstrating
that stability, detine the effect on drag, efficiency, and bap life.
The mechanisms evaluated for raising the G/C ratio were:

     Cleaning energy; i.e., pulse jet versus reverse air
       versus combinations;
     Cleaning mode - on-line versus off-line;
     Cleaning frequency; and
     Fabric type and style.

PARAMETERS MONITORED AND APPROACH

     The following describes the evaluation methods employed and the
parameters monitored.

Section Drag

     The parameter used to define stable operation was average section
drag, which is the average tube sheet A? during a filtering cycle divided
by the section G/C:
                       S (Avg) = AP (Avg)
                                  G/C

     A section was defined as having short term stability if no consis-
tent upward movement in the average drag was observed over a minimum 2-3
day evaluation period.

     The three cells of each section have pressure taps, so all 18 cells
of the baghouse may be monitored for pressure drop.  Generally, the middle
section is monitored, and the AP is registered on a 15 in. W.G. Magnehe-
lic gauge.  There is one Magnehelic for each section.  The G/C ratio is
monitored by an Annubar located at each section inlet, and the flow is
registered on an incline manometer.  The instruments are located on a
panel at the inlet test level.

     The initial evaluation approach was to establish the cleaning vari-
able to be tested, adjust each section inlet damper to produce equal sec-
tion G/C, and monitor section AP.  A method which proved more reliable
was to adjust for equal section AP and monitor section flows.  Since the
total gas volume fluctuated with steam load demand, the average G/C typi-
cally varied between 5 and 8.  The drag values were recorded several times
each day.  Each condition was evaluated long enough to establish only short
term stability or instability.  At various times during, the test program,
snapshots of AP versus G/C were obtained by varying the G/C from 6 to 10
for short periods and monitoring AP.  This aided in verifying long term
data, and also gave a fabric drag versus exposure time profile.

Efficiency

     Collection efficiency was to be determined at regular intervals dur-
ing the program.  Due to the common outlet, testing of individual sections


                                   361

-------
 requires  taking all other sections off-line.  This has proved somewhat
 difficult and provided questionable results.  Due to budget constraints,
 additional testing was deferred until optimum high G/C operation is
 achieved  and additional funds  are available.  No individual fabric emis-
 sion testing has been done.  Both baghouses have been tested and have met
 compliance.

 Fabric Characteristics

      Bags were periodically removed from the baghouse for laboratory
 testing.   Mullen Burst was monitored to define fabric strength retention.
 Permeability was monitored to  define retention of flow characteristics.

 RESULTS AND CONCLUSIONS

      Time and funds dictated a fairly rapid initial evaluation of the
 magnitude of the effect of the available options.  Having the fabrics
 enter the evaluation at different times further complicated the evalua-
 tion; however, certain trends were consistently found.

 Effect of Cleaning Energy  on Fabric Drag

      The  original baghouse utilized two 1% in. diameter pulse jet valves
 per cell  to pulse 18 bags  each.  In order to achieve the higher G/C de-
 sired without incurring high damper pressure drop losses, bags were re-
 moved as  shown in Figure 3.  Since all active bags were now located in
 four rows per cell,  the cleaning energy supplied to each bag could easily
 be increased by 50% by removing one blow pipe per valve.  The operating
 mode with four pipes is referred to as the High Pulse Jet mode, and with
 six pipes, the Medium Pulse Jet mode.  A much lower level of cleaning
 energy was obtained by eliminating the pulse jet and cleaning with reverse
 air only.

      Table 2 shows the relative drag values observed during the three
 cleaning  energy levels.  A significant effect was seen on all fabric types.

             TABLE 2.   EFFECT OF CLEANING ENERGY ON FABRIC DRAG

                                          Typical Drag
     Fabric Type _ R.A. Only        Med. P.J.        High P.J.
Std. Teflon IS
Asym. Teflon*
Exp. Teflon^5
L Unstable
I) Unstable
' 1.1
1.5
1.2
0.8
1.0
0.8
0.5
     The experimental felts are grouped as shown, since no significant
difference was observed between styles during this experiment.  Note that
both the standard and asymmetrical felts did not produce a stable opera-
ting point with reverse air only.  This may be due to their longer bag set
exposure or differences in construction.

Effect of Cleaning Mode on Fabric Drag

     Off-line cleaning is accomplished by activating a cell's outlet damper,
which closes to block the forward gas flow, pulsing all the bags in that
cell at one time,  and maintaining the cell off-line for 15 seconds before
reopening the outlet damper and resuming normal flow.   On-line cleaning

                                  362

-------
was established by allowing the pulse jet cleaning to continue, but
deactivating the outlet dampers from the cleaning cycle.  High energy
cleaning was employed throughout this experiment.

     Table 3 shows the relative drag values observed during the two
cleaning modes and at high and low filtering velocities.

             TABLE 3.  EFFECT OF CLEANING MODE ON FABRIC DRAG
G/C Ratio
Cleaning Mode
Fabric Type/^
Std. Teflon^k
Asym. Teflon-Sj
Exp. Teflon^
Off -Line
1.0
0.8
0.5
Typical Drag
On-Line Off-Line
1.0 1.0
1.0 0.8
1.0 0.5
>6
On-Line
1.5 (Unstable)
1.5 (Unstable)
1.2 (Unstable)
     This evaluation confirmed past operating experience at Kerr in that
stable operation at high G/C and with on-line cleaning is not possible.
The low G/C evaluation was not run over a long enough time to determine
if long term stability could be maintained.  The dramatic effect on drag
is assumed to be due to the reduction of re-entrainment in the off-line
mode.

Effect of Cleaning Frequency on Fabric Drag

     Theoretically, one tries to establish cleaning frequenty at the mini-
mum level that produces a reasonable stable pressure drop.  This then
should minimize compressed air usage and extend bag, pulse-Aet, and damper
component life.  Prior experience with the standard Teflon^ felt and asy-
mmetric felt was that a 6-minute cleaning frequency is required with the
medium cleaning energy level;"  most data with those fabrics are at that
cleaning frequency.  Initial evaluation of the new experimental fabrics
indicated that a lower cleaning frequency could be used:  most data on
those fabrics was collected at the 20-minute cleaning frequency.  With all
sections employing high energy cleaning, the results of Table 4 were
obtained.
             TABLE 4.  EFFECT OF CLEANING FREQUENCY ON FABRIC DRAG
              Cleaning Frequency
  Typical Drag
6 Min.       20 Min.
Fabric Type/gv
Std. TeflonWjv
Asym. Teflon-S^
Exp. Teflon^

1.0
0.8
0.5
^^•••fc.*-**^""^*— ^»»i»™*^"^B-™»^"^"""^^^^-"

1.2
1.0
0.5
     Note that the fabric drag was not reduced by increasing the cleaning
 frequency on  the experimental felts.  The pressure drop rise between clean-
 ing has been  both very low and extremely linear, indicating the possibi-
 lity of further extending the cleaning frequency.  The short term increase
 in fabric drag of the standard and asymmetric felts under high cleaning

                                  363

-------
 energy  and  20-minute  cleaning was quite modest.  Note that sufficient  test
 duration was  not  allowed to  verify long term stability and that the more
 rapid rise  of pressure  between cleaning indicates that the optimum clean-
 ing frequency may be  somewhere between 6 and 20 minutes.

 Effect  of Fabric  Construction on Fabric Drag

      Due to the variables of bag history, cleaning energyr^fnd cleaning
 cycles  employed,  direct comparisons of the various Teflon ^-'felt con-^
 structions  have been  difficult.  Earlier tables have shown the Teflon^
 styles  for  reference  to indicate the relative effect of the cleaning vari-
 able under  study.   The  indicated trends are both real and significant.

      Table  5  provides relative drag values for the four major felt con-
 structions  now under  evaluation.  Experimental construction _X differs
 from construction Y_ in  that  Y^ was designed to have a 50% higher initial
 permeability.  Two different weight bags of construction Y_ were fabricated
 (styles No. 402 and 403).  No significant difference was seen between  the
 two in  the  drag results.

      All of the prior evaluations were run with the system inlet having a
 high degree of dilution due  to ambient air in-leakage in the boiler stack,
 expansion joints,  and emergency dilution damper.  This produced an inlet
 grain loading of  approximately 0.2 gr/acf.  In-leakage was minimized
 through most  of these areas.  While this has reduced the average G/C,  the
 boiler  load fluctuates  sufficiently over the course of a day that the  same
 G/C range of  6 to  10 can still be evaluated with the reduced cloth area.
 Also, boiler  operation  has provided some variation in grain loading with
 some data being obtained at  0.5 to 0.8 gr/acf.  With this, the standard
 fabric  drags  are now more in line with industrial experience at 4-5 G/C.
 The short term A P  versus G/C evaluations show a constant drag over this
 range.

    TABLE 5.   EFFECT OF FABRIC TYPE ON FABRIC DRAG (short term evaluation)

Inlet Loading (gr/acf)
Std. Felt
Asym. Felt
Exp. Construction X
Exp. Construction Y

0.2 - 0
1.0
0.8
0.5
0.5
Typical
.3




Drag
0.5 - 0.
1.6
1.4
1.1
0.8

8




     The drag data previously shown were a result of short term evalua-
tions which did not allow full stabilization of fabric condition and were
intended only for relative comparisons of fabric types.  From May to
August 1982, an evaluation at fixed conditions revealed the results con-
tained in Table 6.  While these drags are higher than previously reported,
they were consistent under normal boiler operation and represent what
might typically be expected.   To further evaluate the drag values, the
system was operated at relatively constant conditions and modes over an
extended period of time.  The baghouse was operated with the stack cap
bypass damper closed, the Vortex damper control on automatic off-line

                                  364

-------
cleaning, a cleaning frequency of 6 minutes, and a typical grain loading
of 0.5 to 0.6 gr/acf.   This evaluation produced the data of Table 6.

    TABLE 6.  EFFECT OF FABRIC TYPE ON FABRIC DRAG (long term evaluation)
Fabric Type
Standard Felt
Asym. Felt*
Exp.
Exp.
Exp .
Teflon
Teflon
Teflon
Const.
Const .
Const.
X
Yl
Y2
Typical
1
0
0
1
.6 -
.95-
.9 -
.1 -
1
1
0
1
Drag
.7
.0
.95
.2
CONCLUSIONS AS TO HIGH G/C CAPABILITY

     The primary thrust of the evaluation to date has been to identify
parameters that produce lower drag values or improve pressure drop stabi-
lity so that higher G/C levels may be achieved with reasonable and stable
pressure drops.  By necessity, the G/C levels have varied with the boiler
load, and specific evaluations have been short.  Future program activity
is to select the optimum system and fabric parameters and evaluate them
over a longer period.

     The results to date indicate that the optimum system operation is
with higher energy, off-line pulse jet-cleaning, a cleaning frequency of
6-20 minutes, and experimental Teflon^ felt Construction Y.  Short term
stability up to a G/C of 10 has been demonstrated.  During the long term
evaluation, the system was operated in a manner that allowed the section
pressure drops to self balance and the section gas volume flows to be
dictated by their fabric drags.  This resulted in the experimental felt
Construction Y running typically at 7-9 G/C with no instability observed
and a drag typically less than 1.0 even at the highest inlet grain load-
ings during high boiler loads.  Mullen Burst tests on samples from the
above test bags showed only a modest drop in strength for all fabrics
except experimental Construction Yl.

     A longer term evaluation of the above conditions is recommended and,
if successful, could double the present filtering velocity now common in
this application.

                        FABRIC EVALUATION PROGRAM

Boiler Description

     The No. 2 baghouse is applied to a bank of four coal-fired boilers
with a common manifold supplying a natural draft stack.  The first three
boilers were built in 1928 by Walsh & Weidner  Boiler Company and are
equipped with Combustion Engineering underfeed retort spreader type
 * Note:   Asymmetric  felt was not  included  in  this  evaluation.


                                   365

-------
stokers and side dump grates.  The boilers are rated at 22,500 Ib/hr
steam at 250 psig and saturation temperature.   The fourth boiler was
built by Combustion Engineering in 1941 and is equipped with Combustion
Engineering underfeed retort spreader type stoker and side dump grates.
It is rated at 130,000 Ib/hr steam at 250 psig  and saturation temperature.
All four boilers are provided with force draft fans.,  and generally two
boilers at a time are operated.  Outlet flue gas is directed to the bag-
house from the manifold.  Normal operation produces 40,000 acfm and 400 -
450°F flue gas.

Baghouse Description

     The No. 2 baghouse is identical to No. 1, except its cleaning method
is reverse air plenum pulse instead of pulse jet.  The collector incor-
porates 648 bags with a nominal size of 5 in.  diameter and 104 in. long,
having a total cloth area of 6800 sq ft.  Normal filtering velocity is
5-6 fpm.  Its cleaning energy is supplied by a 4000 acfm fan at 10 in.
W.6. pressure through cell outlet poppet dampers that simultaneously stop
the normal forward gas flow through one cell,  and introduce the reverse
pulse of heated ambient air.  A typical cleaning cycle is 10 seconds of
reverse flushing followed by 5 seconds of delay before moving to the next
cell.  This produces a bag cleaning frequency  of 4.5 minutes.  The house
was divided into six sections and is instrumented in the same manner as
house No. 1.

Fabric Description

     The fabric being evaluated consists ojk four bag sets:  three woven
glass with varying finishes and one Nomex ^ felt with an acid resistant
finish.  The woven 'glass bags all have the same basic fabric construction,
one that is presently being used in both pulse jet and reverse air collec-
tors on coal-fired boilers.  Three different, finishes are being evaluated
on the glass bags:  an experimental Teflon Abased finish developed by
Dupont, aad two state-of-the-art finishes, Burlington's style I6L&. 10%
Teflon B^, and Burlington's style ^|?5, Acid  Flex^.  ThagNomex ^bag set
utilized Globe Albany's 16 oz Nomex^felt with Permagard^ finish that
has shown success in Europe.

     Fabric specifications are shown in Table  7.  The locations of the
various fabrics in the baghouse are shown in Figure 4.  A test cell incor-
porating all fabric types was established to normalize any biased effects
of section location.

TECHNICAL OBJECTIVE

     This program's primary objective was to evaluate the operational
characteristics and service life of the various fabrics while exposed to
normal operating conditions of the Kerr coal-fired boiler and baghouse
system.
                                  366

-------
   TABLE 7.   WOVEN GLASS  FABRIC ANALYSIS
   Fiber:
   Construction:
                    ALL FABRICS
Class
        Weave      3X1 Twill
        Count      W44 X F24
        Warp Yarn  DE 37-1/0 - (Singles Construction)
        Fill Yarn  DE 75-1/0 - (Plied,  2 Texturized,
                                1 Untexturized)
        Virgin Cloth Perms 90 fpm @ 0.5 in.  W.G.
Fabric
Finisher
Perm, Avg.
LOI, wt%
Oz/yd2
- Treated
- Untreated
Thickness, in.
A,
Acid/-v
Flex^
Burlington
77.8
4.6
12.6
12.0
0.0144
B,
Teflon^
Burlington
56.1
9.5
14.3
12.9
0.0142
x,
Exp'tal
Clark-Swabel
47.5
9.6
13.6
12.3
0.0158
Virgin Cloth Perm zSQ fpm
                       367

-------
PARAMETERS MONITORED AND APPROACH

     The  following describes the evaluation methods employed and the
parameters followed.

Fabric Drag

     Filtering velocities were maintained as equal as possible between
test sections throughout the evaluation by monitoring the inlet Annubars
and manually adjusting the section inlet dampers.   Normal filtering velo-
cities were maintained in the 4-6 fpm range.  Section tube sheet AP was
recorded  daily to compute average fabric drag values.  Fabric permeability
was monitored on bags that were periodically removed for testing.

Fabric Properties

     To evaluate the effect of operational exposure on bag life, the
fabric strength characteristics were monitored.  Test bags were scheduled
for removal after exposure times of 1 hour, 1 day, 1 week, 1 month, 3
months, 5 months and 7 months.   Fabric testing included Tensile, Mullen
Burst, MIT Flex,.Loss on Ignition (LOI) , and visual and microscopic in-
spection.

Efficiency Testing

     Mass efficiency tests by FJ?A Method 5 were scheduled to be run after
stabilization of the system.  Although difficult,  individual section
emission  testing was scheduled to provide comparative data between fabric
types.  Testing currently has been deferred due to budget constraints.

BAG CONDITIONING AND OPERATING HISTORY

     Conditioning of the bags was initiated on May 14, 1981, and continued
for the next 4 weeks.  The flow was maintained at  the low (1.5 - 2.5) G/C
for an extended time to both obtain some data at low G/C and ensure that
system control would prevent biasing of any fabric.  The flow, temperature,
and section pressure drop were carefully monitored, and the bags were in-
spected visually.

     Due  to intermittent boiler operation and long weekend shutdowns, 7
full days of exposure required 1 month of calendar time (early in the pro-
gram).  The plant was then shut down for the month of July.  After resuming
operation, the baghouse was set to operate at 4-5  fprn-  Malfunctions in the
damper system of section C, the experimental Teflon^ finish, resulted in
lower filtering rates in that section for 3 weeks.  Again, due to limited
plant production which reduced the boiler needs, it would take approxi-
mately 3 months to obtain the next 2 months exposure time.  The 3 month
test bags were scheduled for removal on October 16.  Unfortunately, a
severe boil^ upset caused a fire in the baghouse, and all the experimen-
tal Teflon^finished bags and the test bag cell were destroyed on Octec-
ber 12.  The baghouse remained off-line until new  experimental Teflon™
finish bags could be obtained.  The evaluation was restarted on May 5,
1982.

                                  368

-------
RESULTS AND OBSERVATIONS

     The cyclic nature of the baghouse during this evaluation dramatizes
the need for fabrics that can survive many start-ups, shutdowns, and
their resultant dew point excursions.  Normal boiler operation was start-
up on the 2nd shift Sunday, and shutdown on the 2nd shift Friday.   Due to
production limitations, several shorter operating cycles were run.  In
addition, baghouse auxiliary component malfunctions forced the baghouse
off-line at times.  Needless to say, this type of application intensifies
the need for preheat and purge capabilities (which exist) and fabrics that
can resist acid condensation.

 AP Evaluation
     Through June 25, the plant's summer shutdown, the filtering veloci-
ties were held constant and, for all practical purposes, the AP levels
of all sections were the same.  Following the shutdown, Section C (exp.
finish) ran at lower G/C ratios and higher AP levels as a result of
cleaning system problems.  These problems were intermittent, recurring
throughout August.  After correction, its AP was comparable to that of
the other section.
     No significant differences in fabric drag could be seen during the
evaluation between sections.  Average drags ranged from 1.5 to 2.0,
reflecting the low energy cleaning system being employed.

Fabric Strength Tests
     Results to date are plotted on Figures 5 through 8 and Table 8.  The
test plan calls for the removal of three bags of each type after 1 hour,
24 hours, 1 week, 1 month, 3 months, 5 months, and 7 months of exposure
time.  After removal, the top, middle, and bottom of each bag is tested
for its strength characteristics.  The tests include permeability, tensile
strength, flex strength, and burst strength.  Tests have been completed
through 7 months of exposure.  In addition, all but the experimental bags
have seen over 6 months additional off-line exposure.  It has been deter-
mined that there is no significant difference in the strength characteris-
tics of the top, middle, and bottom of any of the bags.  Note that in-
creases in the value of a given parameter versus on-line time are not
expected.  These increases are attributable to variations in the initial
strength characteristics of the bags, finish protection effectiveness, bag
location, or (in the case of permeability) handling.
     The new, 1 hour, 24 hour, 1 week, and 1 month Acid Flex ^ Teflon B,v-'
and Nomex^ data were obtained before the October 12 fire.  The 3-month
data were obtained from bags off-line for 6 months and were obtained in
May 1982, after restart-up.  All the experimental Teflon data were obtained
on replacements for the original experimental Teflon bags.  This evalua-
tion began in May 1982, and the replacement bags have strength characteris-
tics similar to those of the original bags.

Permeability
     As previously stated with regard to the high G/C program, fabric
permeabilities were taken in the laboratory and in-situ cake conditions

                                  369

-------
have been altered by  removal from the baghouse (removal from cages) and
transferred  to  the laboratory.  Again, this type of evaluation relies on
examining the relativity of the data rather than the absolute values.

     As  can  be  seen in Figure 9, the greatest decrease in the permeabili-
ty  of  all four  fabrics occurs in the first hour of operation.  Another
substantial  decrease  takes place from 1 hour to 24 hours.  The Acid
Flex>-}bag set  had the highest initial permeability, starting 35% higher
than the other  woven  glass fabrics.  However, it drops teethe level of
the other fabrics (25-30 fpm) within 1 hour.  The Nomex ^ fabric starts
with the lowest value; however, it exhibits the least decrease in perme-
ability, and is higher than the other fabrics up to 3 months.  All the
fabrics  are  in  the range of 4 to 10 fpm within 1 month.  Dirty as received
permeability values of 4 to 10 are normal for fabrics operating at 4-8 in.
W.G. pressure drops at 4-6/1 G/C.  At 3 months, all the woven fabrics
have roughly the same permeability.  From 5 months on, all fabrics are
essentially  the same and reflect the sameness in fabric drag seen in the
field.   Note that no severe blinding has occurred in any of the fabrics
through  7 months.

Flex Strength

     Figure.  5 is a plot of the flex strengths for the woven glass fabrics.
The Nomex ^  felt is considerably higher than any of the woven glass fab-
rics (initial value around 47,000).
     The Nomex ^ showed its greatest decrease in flex strength between 24
hours and 1 week (from 42,000 to 9,000).  The Nomex bag set showed dra-
matic loss of strength between 5 and 7 months, and several field failures
occurred during this period.  The Teflon B*^fabric's initial superiority
was eliminated within 1 day, and this may be of significance to the
finisher.  Also, the Teflon B^SJhad the lowest flex value by 1 month and
continued Xo have the lowest-at 5 months, although only slightly less than
Acid Flex vJ.  The Acid Flex^fabric showed its greatest decrease to be
between 24 hours and 1 week.  The"experimental Teflon^finished bags,
although starting at a lower value, have shown much less strength deterio-
ration, and bags tested at 5 and 7 months show virtually no reduction in
flex strength.

     When the 1 month old bags were removed for testing, the samples loca-
ted on the outside walls were^ry wet.  Table 8 is a summary of the ^£1 ex
results for the wet Acid Flex®, Teflon 8®, and experimental Teflon^
fabrics.  The experimental. Teflon© showed no significant decrease in flex
strength.  The Acid Flexv&) showed a significant loss of flex strength only
in the middle of the bag, and the Teflon BVS) showed virtually a total loss
of flex strength.  These are good indications of the finish's ability to
encapsulate the fiber and resist moisture and acid attack.
Mullen Burst
     Figure 6 shows the Nomex ^ to have:  the lowest initial burst value,
as is typical of felts; a slow decline in strength; stability between 1 and
5 months; and raoid deterioration between 5 and 7 months.   Of the glass fab-
rics, Acid Flex^and Teflon B^ show similar burst characteristics with
over 40% strength loss at 1 month, and only a slight decline to 7 months.

                                  370

-------
                    , TABLE 8.  WET BAG FLEX DATA FOR
                              1-MONTH OLD BAGS
                Fabric
  Number Of Flexes

Acid Flex^

Teflon B^

Tog
811
1384
31
26
Middle
37
34
9
14
Bottom
467
588
43
24
                Experimental
                Teflon
1025

1587
1242
 912
614
820
The experimental TefIon^started with the same burst value as the other
woven fabrics (as expected, since all were fabricated with the same basic
fabric construction), but has shown superior retention of strength.

Tensile Strength
     The Nomex^-^fabric shows noticeably lower (typical of felts) tensile
strength (Figures 7 and 8) than the Glass fabrics, but significant deterio-
ration after 5 months.  The Glass fabrics show similar trends to each
other, with the greatest-decrease occurring between 24 hours and 1 week.
The experimental Teflon^ again has the highest value after 3 months.

General Observations
     The study, by design, represents a short evaluation period when put
in the context by seeking greater than 2 years of bag life.  The data are
unique in that they are the first documented evaluation of how rapidly
strength characteristics are lost (i.e., hours, days).

     Fabric filters typically show an early strength loss of 50%, at which
time^they reach a plateau until the end of the bag life cycle.  The Acid
Flex ^ and Teflon B ^ appear to be at this level within 1 week, and are
reasonably stable through 7 months on-line exposure plus the extended
off-line.  The experimental Teflon B^ finish was designed to improve fab-
ric strength retention and has been successful through 7 months, indica-
ting that its finish may be protecting the fabric better.
     The Nomex^bag set with its Permagard^6 finish showed superiority
                                  371

-------
over the 1 month life previously obtained on untreated Nomex^-'bur
insufficient life to be practical in this application.

                           CONVERSION FACTORS

     Readers more familiar with metric units may use the following
factors to convert to that system:

            Non-metric          Times           Yields Metric

               °F            5/9 (°F-32)              °C
               ft,              30.48                cm,
               ft^               0.09                 m
               ft               28.32                 L (liter)
               gr                0.06                 g
               in._               2.54                 cm_
               in.                6.45                 cm
               Ib                0.45                 kg
               oz               28.35                 g
               yd               91.44                 cm
                                 372

-------
                                                    o
                                                    25
g
              BOILER NO.  1



              20.000 LB/HR
              BOT.LKR NO.  2



              20,000 LB/HR
              ROH.KR  NO.  3



              20,000  LB/HR
              nOTI.KR  NO.  4


              30,000  LB/HR
                        u
Bnii.ER NO.  5
60.000 LB/HR
                                                   g


                                                   CO
                                  CQ
                                  ts
                                  i
                                 u
                         Figure 1.  Kerr site layout.
                                   373

-------
    60
     50
00
5
e
P.
     30
EM



3
PQ
20
                    o,.
          B&E
                                                        Section A   X



                                                        Section D   Yl



                                                        Sections B A K  Y2



                                                        Section C  XT2363


                                                        Section F   XT1661
     10
     0
                       100             200


                               TEMPERATURE, °F
                                                      300
400
                     Figure 2.   Permeability vs. temperature.
                                        374

-------
                                                                          COVKS

                                                                          PLATE
o
o
o

o     o
o     o
BLOWPIPE
                       PULSE JET

                          VALVE
                          u
          "~ BAG
o






'
1
1
t
1
1
' ' 1
1 :
1 ;
I





.











i i i
i '
I •
i i
i I . _





. -i r ii •- 	 •-
            Flp,uri.> 3.  High gas-to-cloth cell bag layout.
                                  375

-------
BAG POSITION
                                                                     SECTION
 INLET
	>
(f§) BURLI
14 OZ
3
© (iD
tfGTON STYLE
. WOVEN GLAS
2
© ©
»54
5-EXP .TEFLON
1 &0$

3
BURLINGT
14 OZ. WOV

2
3N STYLE 454
2N GLASS-TEF

1
•1615
.ON B
© ©
TEST
CELL
3
MISC.
FABRIC
2

3
BURL INC
14 OZ. V

2
TON STYLE 45
OVEN GLASS -A

MISC.
FABRIC
1

1
i-1625
]ID FLEX

MISC.
FABRIC
3

3
14 OZ

TEST
CELL
2

2
GLOBE ALBA1
NOMEX.FELT

MISC.
FABRIC
1

1
fY
-PERMAGARD

                                                                                                                        CELL
                                    Figure 4.   Kerr Baghouse  -  cell  schematic  -  system  2.

-------
                                          42.166
ui
-vj
         4000  -
     -J   3000  -
          IOC  -
                New
                                                                                 7,616
                                                                             o  Teflon B
                                                                                    \
                                                                                Notaex

                                                                                   - Teflon
1 i!r
24 Kr       1 Wk

       ON-LINE TIME
1 y.
Mo
Mo
7 Mo
                                        5.  MIT Flex (fill) vs, on-line time
                                                  Kerr baghouse No. 2
                                              fabric evaluation program.

-------
            600 -,
to
•vl
00
       CO
       (X
      1
      «
      2:
450-
           300 -
           150 -
               New
               1 Hr
•  Acid Flex

o  Teflon B

   Nomex


   Exp.  Teflon
                                                                                        5 Mo
                                                                                       7 Mo
                                 Figure 6.  Mullen Burst  vs.  on-line  time

                                             Kerr baghouse  No.  2

                                           fabric evaluation program.

-------
          500  -i
          400  -
    c
    •H
    <
    OS
    H
    V3

CO   —
-j   i-:
\o   i—
    tr.
          300  -
200 -
          100  -*
                New
3. Hr
                           24 Hr         I Wk


                                    ON-LINE  TIM::
I Mo
3 Mo
                                                                                •   Acid Flex


                                                                                O   Teflon B


                                                                                    Nomex


                                                                                    Exp. Teflon
5 Mo
                                Figure 7.   Tensile strength  (warp)  vs.  on-line time

                                                  Kerr baghouse No.  2

                                                fabric evaluation program.
                                                                                                     7 Mo

-------
00
o
       4-1
       •u
       •H
       c
       •H
400  _
       o   300 _
       fc-
       C
       ss
       S-
       u:
           200
           IOC -
                                                                         •   Acid Flex

                                                                         O   Teflon B


                                                                             Noaex
                                                                             Exp. Teflon
                New
               1 Hr
                                Figure 8.   Tensile strength (fill) vs. on-line time

                                                   Kerr baghouse No. 2

                                                 fabric evaluation program.

-------
          100  J
           80  _
     6
     o-
           60  .
      cc
oo
           20


           10

            5
                                                   • Acid Flex

                                                   O Teflon B

                                                   V Nomex

                                                     Exp. Teflon
                New
1 Hr
24 Hr       1 Wk
1 Mo
3 Mo
5 Mo
7 Mo
                                               ONLINE TIXZ
                          Figure 9.  As received permeability vs.  on-line time
                                                Kerr baghouse No,  2,
                                            fabric evaluation program.

-------
               OPTIMIZING THE LOCATION OF ANTI-COLLAPSE RINGS

                               IN FABRIC BAGS
                             By:  John G.  Musgrove
                                  Bechtel  Power Corporation
                                  P.  0. Box 2166
                                  Houston, Texas 77252-2166
                                  ABSTRACT
The number and placement of anti-collapse rings  in fabric filter bags are
frequently left to the bag manufacturers.  Bag manufacturers do not analyze
operating requirements for determination of ring location but merely sub-
divide the bag into sections of approximately equal length.   Bags whose
section lengths are equal are suspected during reverse  air cleaning of having
greater collapse (deflection) in lower sections  than in higher sections.

An analysis has been developed to predict the amount of collapse (deflec-
tion) in all sections of a bag undergoing reverse air cleaning.  The
methodology uses the catenary principle of a uniformly  loaded suspension
cable to determine deflection.  A computer program utilizing this methodology
has been developed to analyze the forces on a warp thread in a bag and to
determine the location of anti-collapse rings to assure equal bag collapse in
all sections.  A companion program has been developed to predict bag section
collapse for bags with established ring locations.
                                     382

-------
                              ACKNOWLEDGEMENTS
Special thanks is given to Research-Cottrell for providing the laboratory
space and support personnel for conducting the full scale test program:
Stephen Zierak, Southwest Regional Sales Manager; Stefan Negrea, Vice
President of Engineering for the Air Pollution Control Division; Dennis
Helfritch, Program Manager with Cottrell Environmental Sciences; and Chet
Gorski and Mike Adams, laboratory technicians with Cottrell Environmental
Sciences.

Thanks is also given to Houston Lighting & Power Company for providing spare
filter bags for use in the test program:  Soli Irani, Program Manager.

The assistance provided by Research-Cottrell and Houston Lighting & Power
Company should in no way constitute their endorsement or agreement with the
scope or results of this study.  The scope and conclusions are solely those
of the author.

                                 INTRODUCTION

Fabric filter bags for reverse-air-cleaned baghouses, such as have been
installed on utility steam electric generating stations, have anti-collapse
rings located at intervals along their lengths to prevent excessive collapse
during reverse air cleaning, Figure 1.

During flue gas cleaning or forward flow, the bag flexes outward uniformly,
being restrained by the circumferencial stresses in the fill fibers.  During
the reverse-air cleaning mode, the bag collapses in a multi-lobe pattern,
Figure 2.  The fill fibers are no longer under tension and the stresses are
taken by the warp fibers.  The fibers at the inside diameter of the lobes
undergo tension and form parabolic curves, Figure 3.  Thus, the warp fibers
form a family of hyperbolic curves, which is the basis for this study. A
single section can therefore be presumed to form a hyperbolic shape, Figure A.

The deflection caused by reverse air flow in each section is believed to
increase from the top of the bag to the bottom of the bag as the tension
decreases (1).  If the original bag tension is inadequate and the ash loading
on the inside of the bags is high, the deflection in the bottom section may
be so great as to cause "pancaking", wherein opposite sides of the bag touch
each other, Figure 5.  This excessive deflection decreases bag life through
excessive bag flexures and through abrasion of the bag by ash falling down
through the bag.

The parabolic shape each fiber assumes under the uniform loading of the
reverse air flow is identical to the catenary shape taken by a suspension
cable under uniform loading.  The purpose of this study is to develop a
mathematical model to predict the amount of bag deflection at the initiation
of reverse air cleaning so that the anti-collapse rings may be spaced so as
to permit equal deflection of each bag section during reverse air cleaning.
This study examines the catenary principle and applies it to the warp fibers
of filter bags.

                                      383

-------
                                     Effective Bog Rodluo, Inohee
Top Cuff « 2.5 in.




Section 1 - 56.13 in.




Cuff 1 " 2.0 in.




Section 2 " 54.5 in.




Cuff 2 - 2.0 in.




Section 3 » 51.75 in.




Cuff 3 * 2.0 in.



Section 4 - 49.75 in.




Cuff 4 - 2.0 in.



Section 5 " 48 in.




Cuff 5 •» 2.0 in.



Section 6 - 45.75 in.



Cuff 6 - 2.0 in.



Section 7 - 43.5 in.


Bottom

Cuff • 3.5 in.
                                                     -d
                                                      X.
                                                                   8 £
                                                                   84
                                             BAG TYPE II



                  Figure 1.   Filter  bag with anti-collapse
                                                 rings.
                  Figure 2.   Plan view of collapsed bag section.




0795h/0074h                               394

-------
             Effective Boq Radius. InoUa
                        uuu
oo
en
     L 1 " 57 in.
     L 2 - 58.2
     L 3 - 53.5 in.
     L 4 - 51.5 in.
     L 5 - 49.7 in.
     L 6 - 47.5 in.
     L 7 • 44.3 in.
•"•  •
*•:•
3 -
  in
      Figure  3.  Section view of collapsed bag.
              Figure 4.  Downward view of collapsed bag section.

-------
                          Effaotive Boa
                                r3r2?
                 L 1 « 57 In.
                 L 2 « 56.2 In.
                 L 3 "53.5 in.
                  L 4 - 51.5 in.
                  L5-49.7 in.
                  L 6 - 47.5 in.
                  L 7 " 44.3 in.
                                   BAG TYPE II
             Figure  5.   Section view of "pancaking" bag section.
                              CATENARY SUSPENSION

The equations for determining stress  and sag of a single cable  under
uniformly distributed  load  (2) are derived from the configuration of Figure
6.  The forces and dimensions are  as  follows:

                   No  load  (initial configuration):
                   o   I = horizontal  distance between supports
                   o   L = developed length of cable
                   o   f = sag of cable
                   Loaded Cable:
                   o   T = tension  in  cable at support caused  by superimposed
                          loads
                                     386

-------
                   o  q. = weight of cable per unit length (assumed uniform)
                   o  a = angle between horizontal and tangent to cable at
                          support
                   o  V = vertical component of T at support
                   o  H = horizontal component of T at support
                   o Af = increased sag in cable due to superimposed loads
.\
<_...
q q q
                Figure 6.  Catenary suspension force diagram.


The operational equations are:

                   L - 1*[ 1 + (8/3)*(f2 /I2)], in.

                   T = [(q*!2/8*f)]* SQR [1 + 16*(f2/A2)],oz.

                   tan a = 4*(f +-Af)/JL

                   V = T sin (a), oz.

                   H = T cos (a), oz.
(1)

(2)

(3)

(4)

(5)
The assumptions for these equations are that the initial sag, f, and distance
between supports, I are known, that the cable loading, q, is uniform, and
that the unloaded cable weight and resulting tension are negligible compared
to those of the loaded cable.
                                     387

-------
                          INITIAL MATHEMATICAL MODEL
CATENARY FORCES

The approach to determining fabric deflection is to  assume  the warp fibers
undergo uniformly distributed loading,  but in a vertical  plane rather than a
horizontal plane, Figure 7.  The fiber  is assumed to have a diameter of .0227
inches and therefore an area/inch of .0227 in^/in.   The fabric is assumed
to have a count of 44 fibers per inch.
                                             wl+w2
                   FCI+1)
             Figure  7.   Bag warp  fiber catenary force diagram.
                                    388

-------
The forces that translate from the suspension cable are:

              o  1 = L   = vertical distance between rings, in.
              o  L = S   = developed length of collapsed fiber, in.
              o  f = d   = deflection of fiber from ring diameter, in.
              o  T = F   = tension force in fiber, oz.
              o  q = p   = force of reverse air, oz/in2/in.
              o  a = a   = angle between vertical and tangent to fiber
              o  V = FJJ  = horizontal component of F at ring, oz.
              o  H = Fy  = vertical component of F at ring, oz.

Additional forces to consider include:

              o  Wl  = unit weight of cloth, oz/in.
              o  W2  = unit weight of ash, oz/in.
              o  W3  = unit weight of ring and ring cuff, oz.

Equations 1 through 5 can thus be rewritten as:

              S  = L t 1 + ( 8/3)*(d2/L2)]                                (6)

              F  = [(p*L2)/(8*d  )] * SQR [ 1 + 16*(d2/L2)]               (7)

              tan a = 4 * d/L                                             (8)

              FH - F * Cos (a)                                            (9)

              FV = F * Sin (a) = p * L2/2                                 (10)

Additional equations developed here include:

              FVN - FVN+1 + W3 + SN *(W1 + W2)                            (11)

              Where N denotes the bag section undergoing analysis.

The tension in the thread is thus:

              TN = SQR [(FHN)2 + (Fvn)2]                                  (12)

BAGHOUSE FORCES

The tension in the warp fiber is further influenced by the thermal expansion
of the baghouse casing, the reverse air force in the bag cap and the weight
of the accumulated ash.  The tension due to thermal expansion is:

              T4 = (T8-T6) * L' * K9 * (CE2 - CEi)                        (13)
                                     389

-------
              Where:
              T8   = operating  temperature, F

              T6   = ambient  temperature during bag hanging, F

              L*   » casing length  ( —L + 16), in.

              K9   = tensioning spring constant, Ib/in.

                   " Coefficient of  thermal expansion of steel,  in/in/F

                   - Coefficient of  thermal expansion of glass  (3),  in/in/F

The relaxation  of  tension due to the reverse air force on the cap  is:

              T5 = A *  .5781  *   v *(R9)2/16                                (14)

              Where:
              A = reverse air  pressure, in. wg

              R9 = bag  radius,  in.

The weight  of ash  per inch of fiber  is:

              W6 - 80 * Tl *  * * D9 * (L9-6)/1728                          (15)

              Where:
              Tl   » thickness  of ash layer, in.

              D9   = bag diameter, in.

              L9-6 = effective  bag length (less ring cuffs)

The total fabric tension just prior to reverse air cleaning is:

              T3 = T +  T4 + T2                                             (16)

              Where:
              T2 « upward force  on bag cap

The total fabric tension just after start of reverse air cleaning  is:

              T9 - T + T4 - T5                                             (17)

The tension created by  the weight of the ash itself is not included, in T3  and
T9 because it is a force that acts solely downward from within the bag.  That
is, since the support spring will deflect and relax the tension  at the bottom
as the tension at the top increases, the net effect within the bag is
negligible and assumed  to be zero.
                                      390

-------
OPERATING EQUATIONS

The four fundamental equations for analysis then become

              FH   = F2(I) = p* [L (I)]2/2                                (18)
                N

              FV   = F1(I) = T9 * (16/1660)-K1 * (Wl + W2) -              (19)
                N       W3 * (1-1)
                           SQR [(F1(I))  + (F2(I))]                      (20)

              D(I) = p* L(I)2/(4 * SQR [(2* F(I))2- (p * L(I))2]          (21)

The initial modeling effort, in using the catenary principles, had no method
for relating tension prior to uniform loading of the fiber to tension
following such loading since the catenary principal assumed initial tension
to be negligible.  Development of this relationship was made by the
measurement of actual deflection under loading of two different fabric filter
bags.

                             LABORATORY  TESTING

Research-Cot trell was kind enough to provide a high-bay test structure and
technical support for measuring the deflections of a filter bag undergoing
reverse air pressure.  The bags were provided from spares purchased by
Houston Lighting & Power Company as part of a separate pilot baghouse test
program.  The pilot test program was performed under a contract with Cottrell
Environmental Sciences.

LABORATORY TEST FACILITY

The test facility was erected in Research-Co ttrell's forty-foot high
precipitator electrode test bay.  Individual fabric bags, each 30 '-6" long
and 11.5" in diameter were hung by a spring scale from the upper bay
supports.  The bottom of the bags were attached to a single thimble which was
restrained by chains linked to earth anchors.  The thimble was sealed at the
bottom and connected to a vacuum pump.  An inclined manometer permitted
measurement of the differential pressure across the bag.  A fixed scale was
placed adjacent to the thimble to indicate any bag movement due to
application of the air pressure, and a tape measure was hung beside the bag,
attached only at the top, to indicate the amount of bag shortening due to
application of reverse air pressure.  A standard workman's scaffold was
erected around the bag and boards placed on it to permit convenient reading
of the support scale load and the convenient measurement of the collapsed
diameter of each section.

To assure attainment of reverse air pressures of up to 1 inch water guage
(in. wg) a tube of plastic film (5 mils thick) was installed over the bag as
it was hung to simulate the resistance of the dust cake.  The weight of the
ash was simulated by hanging masses of approximately equal weight onto the
rings to simulate a uniform loading along the bag length.


                                      391

-------
Bag deflection was measured by use of a standard outside diameter caliper and
a measuring rule to obtain the average minimum collapsed diameter of the bag.

Test Program

A total of ten test runs were conducted, as indicated in Table 1.

                          TABLE 1 - TEST PARAMETERS
     Run,      Bag Type,     R.A. Pressure,    Initial Tension,   Ash Load,
    Number      Number           in. wg        	pounds         pounds

       1          2               0.5                100             0
       2          2               1.0                 98             0
       3          2               0.5                 63             0
       42               1.0                 65             0
       5          2               0.5                 51           65.25
       6          2               1.0                 51           65.25
       7          2               0.5                 85           65.25
       8          2               0.85                72           65.25
       9          3               1.0                 63           65.25
      10          3               1.0                 95           65.25
 The data from Runs 1  through 4  indicated greater deflection at the  top  of  the
 bag than at the bottom of  it.   These data were subsequently discarded as being
 the result of anomolies due to  light loading (no ash) and the influence of the
 plastic bag.

 Test Runs 5 through 10 produced acceptable results - deflections  increasing
 down the bag - and were used in the subsequent analysis.  Appendix  I presents
 the summary of test results.

                                DATA CORRELATION
 TENSIONING DUE TO REVERSE AIR

 The test program clearly identified  a  factor missing from equation 17,  the
 tension created by the collapse  and  shortening of  the bag sections.   In order
 to calculate this tension,  the resulting deflections were assumed  to be solely
 restrained by this tension.   Figure  8  presents the relative distance between
 rings as a percentage  of the  fiber length, a strong correlation.   The total
 bag reduction in length was apportioned to each section  in proportion to its
 unloaded length.   Figure 9 presents  the average reduced  lengths  for each test
 reverse air pressure and the  average for all the tests.
                                      392

-------
          LEGEND
          RUN 15
        BAG TYPE 2   8s

          RUN IB
        BAG TYPE 2

          RUN 17
        BAG TYPE 2
   RUN IB
 BAG TYPE 2
- -.

   RUNI9
 BAG TYPE 3

  RUN #10
 BAG TYPE 3
         1O/3/B2
                    OS
                                FILTER BAG REVERSE AIR DEFLECTION TESTS
                              COLLAPSED RING SPACING v. BAG SECTION NUMBER
                   I     1
                                                  i
                                   BAG SECTION NUMBER - TOP TO BOTTOM
       Figure  8.  Bag section  shortening due  R.A.  pressure.
         LEGEND     g

       AVERAGE FOR   £
       ALL 6 RUNS   8s

       AVERAGE FOR   |
       DELTA P-.5

       AVERAGE FOR
       DELTA P». 85
      _       _        •

       AVERAGE FOR   |
       DELTA P- 1   *
                                FILTER BAG REVERSE AIR DEFLECTION TESTS
                              COLLAPSED RING SPACING v» BAG SECTION NUMBER
                                1     I     i     1
                                                      i
                                  BAG SECTION NUMBER - TO? TO BOTTOM
Figure 9.   Average  bag  section shortening due to R.  A.  pressure,
                                    393

-------
The shape of  the  curves suggested an exponential relationship for the
percentage  shortening, one minus the reduced length.  Figure  10 depicts the
exponential curves for each of the three test reverse air  pressures.
Unfortunately,  the data of the 0.85 in. wg case do not fall between those of
0.5 in. wg  and  those of 1.0 in. wg., perhaps due to measurement error or
insufficient  data at that test condition.  A single exponential curve was
then generated, Figure 11, to approximate the shortening of the hag
sections.   The  relationship obtained is:
      L  - (1 - [(.01)  * (.005 * exp ((1-1) * 0.844) +  .5673)])*S
(22)
      Where:
      L  = distance between rings as a factor of fiber length S
      1  = the bag section under analysis, numbered starting at the top
           LEGBO
         REVERSE AIR
          DELTA P-.5

         REVERSE AIR
         DELTA P-.85
         ^^^HHI^^H* • ••••MHBIBHB
         REVERSE AIR
          DELTA M
                               FILTER BAG REVERSE AIR DEFLECTION TESTS
                                 SECTION SHORTENING UNDER DELTA P
           18WQ2
                                  BAG SECTION MMER (MINUS ONE)
Figure 10. Averaged exponential curves for bag section  shortening due to R. A.
           pressure.
                                      394

-------
           LEGEND
       FROM 6 TEST RUNS
      Y-.M5«EXPC.844«X>9 '
                              FILTER BAG REVERSE MR DEFLECTION TESTS
                                 SECTION SHORTENING UNDER DELTA P
                       .*-
                       .*-
                       .£-
                                  BAG SECTION NUMBER (MINUS ONE)
Figure 11.  Overall  average exponential curve for bag section shortening  due
            to R.A.  pressure.

                           FINAL MATHEMATICAL MODEL
By knowing  the section deflections and the shortening caused by them,  the
tensions resisting  the deflection can be calculated using equations  22,  6,  8
and 10.  Equation 10  was  modified to include a pressure factor which was
varied to provide absolute equality.  The pressure factors thus obtained, as
a function  of bag section I,  can be closely approximated by:
                   F9  (I)  = .07833 + .00607 * I
(23)
Once the estimation of  increased tension was obtained by the summation  of  the
values for each  section,  equations 10, 11, and 12 were then iteratively
calculated (with a  new  pressure factor in equation 10) to obtain a deflection
equal to that measured  in the test program.  The new pressure factor  for
equation 10  is 0.673.
                                      395

-------
The computer program  'DEFLEC' of Appendix II performs the computations of  the
equations to calculate deflections of filter bags.when the section lengths
are known.  Table 2 presents a sample printout of program 'DEFLEC1.  Computer
program  'RINGOP' of Appendix III is a modification of 'DEFLEC1 which assumes
equal section lengths to start and then iteratively adjusts them until all
bag sections produce equal deflections for the stated conditions of initial
tension, ambient/operating temperature differential, support spring constant,
and ash  cake thickness and weight.  Table 3 presents a sample printout of
program  'RINGOP'.

Figure 12 presents the deflections of two bag designs, one with equal ring
spacings and one with optimized ring spacings, for an initial tension of 60
pounds and  an ash weight of 57 pounds.  The benefits of equal deflection can
be observed and are the result of designing the ring spacings to assure equal
deflection.

The'work described in this paper was not funded by the U.S. Environmental
Protection  Agency and therefore the contents do not necessarily reflect the
views of the Agency and no official endorsement should be inferred.
       10/12/82
                                        Rodiu^  Inch..
                                         n P t
                                             u. d
Figure 12.
Section view of collapsed bag for both standard and optimum
spacing of anti-collapse rings.
                                     396

-------
                          TABLE 2.   SAMPLE OUTPUT OF  PROGRAM "RINGOP".
          This is prograM RINGOP to  deterwine the optiwoM location of anti-collapse rings
          in Filter bags to assure equal deflection in each section during reverse? air  cleaning
          11/2/1982
             Bag
           Length,
           inches
            396.0
                Bag
              Di a Meter,
               inches
               11,50
                          Operating
                          Temperature,
                          Degrees  F
                              300
  Ambient
Tenperature,
 Degrees F
     80
                                                         Dia, of
                                                        Ring wire,
                                                          inches
                                                          .1875
                                                                        Tension
                                                                      Prior to RA,
                                                                         pounds
                                                                          95.7
u>
vo
  Weight
 of cloth,
oz./sq.yd,
    14
                        Initial
                        Tension,
                         pounds
                          60,0
                          Forward Gas
                           Pressure,
                            in.w.c,
                             4,00
 Total Ash
  Weight ,
  pounds
   40,8
                                                       Reverse Air
                                                        Pressure,
                                                         in ,w,c ,
                                                            .50
Number of Rings -•  1
Deflection greater than 1/3 diameter)  further analysis terminated.
                                                                        Tension
                                                                      Start of RA,
                                                                         pounds
                                                                         139,5
  Height       Initial
 of cloth,     Tension,
oz./sq.yd.      pounds
    14          60,0
                                    Forward Gas
                                     Pressure,
                                       in.w.c.
                                       4.00
                                          Total Ash
                                           weight,
                                           pounds
                                            40.8
                                                       Reverse Air
                                                        Pressure,
                                                          in,w.c.
                                                            .50
           Number  of  Rings  --   2
           Deflection greater  than  1/3 diameter; further analysis terminated,
                                Tension
                              Start of RA,
                                 pounds
                                 HI,2
  Weight
 of cloth,
oz./sq.yd.
    14
                         Initial
                         Tension,
                          pounds
                           60.0
                          Forward Gas
                           Pressure,
                            in.w.c.
                             4.00
 Total Ash
  Weight,
  pounds
   40.8
                                                        Reverse Air
                                                         Pressure,
                                                          in,w,c.
                                                            ,50
           Nuttber of Rings •   3
           Deflection greater  than 1/3 dianeter;  further  analysis  terminated,
                                                                         Tension
                                                                      Start  of RA,
                                                                          pounds
                                                                          142,6

-------
                         TABLE 2.  SAMPLE  OUTPUT OF PROGRAM "RINGOP"  (CONT.)
Weight
of cloth,
oz./sq.yd,
14
Initial
Tension,
pounds
60.0
Forward. Gas
Pressure,
in.w.c.
4.00
Total Ash..
Weight,
pounds
40.8
Reverse Air
Pressure,
in .w.c .
.50
Tension
Start of RA,
pounds
143.4
          NuMber of Rings -  4
          Section 1 .« 84.05 inches long  (effectively);  its  deflection * 2.705 inches
          Section 2 = 80,55 inches long  (effectively);  its  deflection'•« 2.698 inchao
          Section 3 = 77,8 inches long  (effectively);  its deflection - 2.742 inches
          Section 4 •*. 75,05 inches long  (effectively);  its  deflection ™ 2.795 inches
          Section 5 = 71,5S inches long  (effectively);  its  deflection =2.797 inches
          Tension at the bottow of the bag  (at  the  start  of cleaning) -   91.3 pounds.

          Measured (fabricated)  section  lengths are as  follows  (astuMing  2  inch ring covers)
          Section 1- = 83,175 inches long
          Section 2> 78,8 inches long
          Section 3 •= 76,05 inches long
          Section 4 = 73,3 inches long
w         Section 5 ~ 70.675 inches long
CO
Weight
of cloth,
02 ,/sq.yd,
14
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
Weight,
pounds
40.8
R&verse Air
Pressure,
in.w.c.
.30
Tension
Start of RA,
pounds
143.2
         NuMber  of  Rings =  5

         Section  1  = 7ft.04) inches long  (effectively); its deflection - 1.881 inches
         Section  2  - 67.666 inches long  (effectively)J its deflection - V,SB inches
         Section  3  = 65.291 inches long  (effectively); its deflection = 1.88 inches
         Section  4  - 64.416 inches long  (effectively); its deflection - 1.971 inches
         Section  5  = 61.916 inches long  (effectively); its deflection = 1.971 inches
         Section  6  ~ 57,416 inches long  (effectively); its deflection - 1.972 inches
         Tension  at the bottoM  of the bag  (at the start of cleaning) -  90,6 pounds.

-------
                         TABLE  2.   SAMPLE OUTPUT OF PROGRAM "RINGOP" (CONT.)

            Measured  (fabricated)  section  lengths are as  follows  (assuning  2  inch  ring  covers)
            Section  1  =  69,166  inches  long
            Section  2 =  65.916  inches  long
            Section  3 =  63.541  inches  long
            Section  4 =  6S.666  inches  long
            Section  5 =  60.166  inches  long
            Section  6 =  58.541  inches  long
u>
VO
10
Weight
of cloth,
oz ,/sq.yd.
14
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in ,w.c .
4.00
                                                     Total Ash
                                                      Weight,
                                                      pounds
                                                       40.8
                                                        Reverse Air
                                                         Pressure,
                                                          in,w,c.
                                                            .50
                                                               Tension
                                                             Start of RA,
                                                                pounds
                                                                141.9
NuMber of
Section 1
Section 2
Section 3
Section 4
Section 5
Section 6
Section 7
Tension at
Rings =  6
= 60 inches, long (effectively); its deflection = 1.393 inches
- 58.25 inches long (effectively); its deflecfion - 1.393 inches
= 56.5 inches long (effectively); its deflection - 1.394 inches
~ 55.5 inches long (effectively); its deflection ~ 1,434 inches
= 54.625 inches long (effectively); its deflection^ 1.486 inches
- 52.75 inches long (effectively); its deflection = 1.487 inches
= 50.875 inches long (effectively); its deflection = 1.489 inches
 the botton of the bag (at the start of cleaning) =  88.9 pound'-;.
           Measured  (fabricated) section lengths are as follows (assuwing 2 inch ring covers)
           Section 1 = 59.125 inches long
           Section 2 = 56.5 inches long
           Section 3 = 54.75 inches long
           Section 4 = 53.75 inches long
           Section 5 == 52.875 inches long
           Section 6 = 51 inches long
           Section 7 = 50 inches long

-------
This it, pr-ogran DEFLEC to deternine the filter  bag
deflection in each section during reverse air  cleaning

Niwber of rings =  6
   Bag
 Length,
 inches
  396.0
Bag
Dianeter,
inches
11.50
Operating
Tenperature,
Degrees F
300
Anbient
Temperature,
Degrees F
80
Dia. of
Ring wire,
inches
.1875
Tension
Prior to RA,
pounds
95.7
Weight
of cloth,
oz./sq.yd.
14
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
Weight,
pounds
10.2
Reverse Air
Pressure,
in.w.c .
.50
Tension
Start of RA,
pounds
141.5
Section 1 = 54.375 inches long (effectively)
Section 2 = 55.75 inches long (effectively))
Section 3 = 55.75 inches long (effectively);
Section 4 ~ 55.75 inches long (effectively);
Section 5 = 55.75 inches long (effectively);
Section 6 = 55.75 inches long (effectively);
Section 7 = 52.375 inches long (effectively)
Tension at the botton of the bag (just prior
; its deflection -
 its deflection = 1
 its deflection = 1
 its deflection = 1
 its deflection = 1
 its deflection ~ 1
; its deflection =
 to cleaning) =119.
                  1.147 inches
                   .234 inches
                   .264 inches
                   .296 inches
                   .329 inches
                   .365 inches
                  1.237 inches
                  5 pounds.
   Bag
 Length,
 inches
  396.0
Bag
Dianeter,
inches
11.50
Operating
Tenperature,
Degrees F
300
Anbient
Tenperature,
Degrees F
80
Dia. of
Ring wire,
inches
.1875
                           Tension
                         Prior to RA,
                            pounds
                             95.7
Weight
of cloth,
oz ,/sq.yd.
14
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
Weight,
pounds
20.4
Reverse Air
Pressure,
in.w.c.
.50
Tension
Start of RA,
pounds
141.5
Section 1 - 54.375 inches long  (effectively);
Section 2 - 55.75 inches long (effectively);
Section 3 = 55.75 inches long (effectively);
Section 4 - 55.75 inches long (effectively);
Section 5 = 55.75 inches long (effectively);
Section 6 - 55.75 inches long (effectively);
Section 7 = 52.375 inches long  (effectively);
Tension at the botton of the bag (just prior
 its deflection =
its deflection = 1
its deflection = 1
its deflection = 1
its deflection = 1
its deflection = 1
 its deflection =
to cleaning) =109.
                   1.147 inches
                    .247 inches
                    .292 inches
                    .34 inches
                    .392 inches
                    .449 inches
                   1.332 inches
                    4 pounds.
                                         400

-------
     Bag
   Length,
   inches
    396.0
     Bag
   Dianeter,
    inches
    11.50
  Operating
 Tenperature,
  Degrees F
     300
  Awbient
Tenperature,
 Degrees F
     80
 Dia.  of
Ring wire,
  inches
  .1875
  Tension
Prior to RA>
   pounds
    95.7
Weight
of cloth,
oz./sq.yd.
14
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
Weight,
pounds
30.6
Reverse Air
Pressure,
in.w.c.
.50
                                                                          Tension
                                                                        Start of RA,
                                                                           pounds
                                                                           141.5
  Section 1  -
  Section 2  =
/•N Section 3  =
H Section 4  =
§ Section 5  =
^ Section 6  =
:  Section 7  =
w Tension at
 54.375 inches
 55.75 inches
 55.75 inches
 55.75 inches
 55.75 inches
 55.75 inches
 52.375 inches
the bottoM of
 long (effectively);
long (effectively);
long (effectively);
long (effectively);
long (effectively);
long (effectively);
 long (effectively);
the bag (just prior
     its deflection =
    its deflection =: 1
    its deflection = 1
    its deflection •- 1
    its deflection = 1
    its deflection •- 1
     its deflection =
    to cleaning) = 99.
      1.147 inches
      .26 inches
      ,321 inches
      .387 inches
      .461 inches
      .544 inches
      1.444 inches
      3 pounds.
f
5
o
9
$
Bag
Length,
inches
396.0
Bag
DiaMeter,
inches
11.50
                             Operating
                            Tenperature,
                             Degrees F
                                300
                                Ambient
                              fenperature,
                               Degrees F
                                   80
                                 Dia. of
                                Ring wire,
                                  inches
                                  .1875
                               Tension
                             Prior to RA,
                                pounds
                                 95.7
w Weight
« of cloth,
H oz,/sq,yd.
H
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in.w.c.
4.00
Total Ash
Weight,
pounds
40.8
Reverse Air
Pressure,
in.w.c.
.50
                                                                          Tension
                                                                        Start of RA,
                                                                           pounds
                                                                           141.5
  Section 1  = 54.375 inches long  (effectively);
  Section 2 = 55.75 inches long (effectively);
  Section 3= 55.75 inches long (effectively);
  Section 4 = 55.75 inches long (effectively);
  Section 5= 55.75 inches long (effectively);
  Section 6 = 55.75 inches long (effectively);
  Section 7- 52.375 inches long  (effectively);
  Tension at the bottow of the bag (just prior
                                   its deflection =
                                  its deflection = 1
                                  its deflection = 1
                                  its deflection ~ 1
                                  its deflection = 1
                                  its deflection ~ 1
                                   its deflection -
                                  to cleaning) = 89,
                                      1.147 inches
                                       .273 inches
                                      .351 inches
                                       ,438 inches
                                      .538 inches
                                       .652 inches
                                      1.575 inches
                                      2 pounds.
                                           401

-------
     Bag
   Length,
   inches
     396.0
Bag
Dianeter,
inches,
11.50
Operating
Tewperature,
Degrees F
300
Anbient
Tenperature,
Degrees F
80
Dia. of
Ring wire,
inches
.1875
                                               Tension
                                             Prior to RA,
                                                pounds
                                                 95.7
Weight
of cloth,
oz./sq.yd.
14
Initial
Tension,
pounds
60.0
Forward Gas
Pressure,
in.w.c.
4.00
                                            Total Ash
                                             Weight,
                                             pounds
                                              51.0
                               Reverse Air
                                Pressure,
                                 in.w.c.
                                   .50
                              Tension
                            Start of RA,
                               pounds
                               141 .5
  Section 1 - 54.375 inches long (effectively);
  Section 2 - 55.75 inches long (effectively);
^Section 3 = 55.75 inches long (effectively);
^•"Section 4 = 55.75 inches long (effectively);
g Section 5 = 55.75 inches long (effectively);
Q Section 6 =55.75 inches long (effectively);
^Section 7 = 52.375 inches long (effectively);
o Tension at the bottom of the bag (just prior
                     its deflection =
                    its deflection = 1
                    its deflection = 1
                    its deflection ~ 1
                    its deflection = 1
                    its deflection = 1
                     its deflection =
                    to cleaning) = 79.
                     1.147 inches
                      .287 inches
                     .382 inches
                      .493 inches
                     .622 inches
                      .777 inches
                     1.733 inches
                     1 pounds.
|    Bag
o  Length,
w  inches
£   396.0
Bag
Diane ter,
inches
11.50
Operating
TeHperature,
Degrees F
300
Anbient
Tewperature,
Degrees F
80
Dia. of
Ring wire,
inches
.1875
                                               Tension
                                             Prior to RA,
                                                pounds
                                                 95.7
CO
M   Weight
SJ  of cloth,
jjjj oz./sq.yd.
      14

  Section 1 =
  Section 2 -
  Section 3 =
  Section 4 -
  Section 5 =
  Section 6 -
  Section 7 =
  Tension at
                Initial
                Tension,
                 pounds
                  60.0

              54.375 inches
              55.75 inches
              55.75 inches
              55.75 inches
              55.75 inches
              55.75 inches
              52.375 inches
             the botton of
 Forward Gas
  Pressure,
   in.w.c.
    4.00
Total Ash
 Weight,
 pounds
  61.2
Reverse Air
 Pressure,
  in.w.c.
    .50
  Tension
Start of RA,
   pounds
   141.5
 long (effectively);
long (effectively);
long (effectively);
long (effectively);
long (effectively);
long (effectively);
 long (effectively);
the bag (just prior
    its deflection =
   its deflection = 1
   its deflection = 1
   its deflection = 1
   its deflection = 1
   its deflection = 1
    its deflection =
   to cleaning) = 69.
       1.147 inches
        .301 inches
       .415 inches
        .552 inches
       .717 inches
        .922 inches
       1.926 inches
       1 pounds.
                                            402

-------
                                 BEFERENCES
1.       Jensen, R.M., "Potential Improvements in Baghouse Design,"
        Proceedings  of the Fifth International Fabric Alternatives Forum,
        Scottsdale,  Arizona, January 15-16, 1981.

2.       Gaylord and  Gaylord, "Structural Engineering Handbook", McGraw-Hill
        Book Co., New York, 1979.

3.       Personal telephone call to Steve McCluskey of Burlington Mills,
        3-16-81.
                                     403

-------
                            APPENDIX I - TEST RESULTS
This is data for Run  5  of the bag deflection testing.
The Bay Type is *  2
The bag is 366.4 in. long, 11.38 in. in diaweter and  weighs 252  oz
The initial bag tension was  51 pounds.
The bag tension at the start of R.A. cleaning was 215 pounds.
The R.A. pressure was  .5 inches w.g.
The support spring constant was 90 pounds/inch.
The ash weight was 65.40 pounds.
The bag length prior to R.A.  application  was 31.4479  feet,
The bag length after R.A.  application  was 31.3646 feet.
                                                    Effective(straight)

                                                        56.99 in. long.

                                                        56.24 in. long.

                                                        53.49 in. long.

                                                        51.49 in. long.

                                                        49.74 in. long.

                                                        47.49 in. long.

                                                        44.36 in. long.
The bag
Ring 1
Section
Ring 2
Section
Ring 3
Section
»*» - .
o Ring 4
Section
Ring 5
Section
Ring 6
Section
Ring 7
Section
Ring 8
section dimensions are as follows;
Measured Effectiwe(arc) E
1,50 in, long, 1.50 in. long.
1 56.13 in. long, 57.00 in. long,
2,00 in. long, .25 in. lonci.
2

3

4

5

6

7

54
2
51
2
49
2
48
2
45
2
43
3
,50
.00
,75
.00
,75
.00
.00
.00
.75
.00
.50
,50
in,
long,
56
in, long,
in,
in,
in,
in,
in.
in.
in.
in.
in.
in.
Bag deflections are
Section
Section
Section
Section
Section
Section
Section
Section
Dia,
i;
2;
3;
4;
5)
6)
7;







6.
6.
6.
6.
6.
5.
5.
at
625
438
540
688
188
875
438
long,
long,
long,
long,
long,
long,
long,
long,
long,
long,
as follows:
deflection
in .
in.
in.
in.
in .
in.
in .
53

51

49

47

44
3

.25
,25
,50
,25
,50
.25
.75
,25
.50
,25
.38
.50

in. long.
l
in. long.
in, long
in, long
in, long
in. long
in, long
in. long
in, long
in. long
in. long
in. long


i

i

i
;
i
)
i

Catenary deflection














2.375 in
2.469 in
2.418 in
2.344 in
2.594 in
2.750 in
2.969 in
i

i
i
i


                                                          Equalizing deflection
                                                                1.783 in.
                                                                1.846 in,
                                                                1.811 in.
                                                                1,761 in,
                                                                1.929 in,
                                                                2,031 in,
                                                                2.169 in,

-------
                      APPENDIX  I -  TEST RESULTS (CONT.)
of the bag deflection testing,

  11.38 in. in dianeter and weighs 252 oz,
This is data for Ron
The Bay .Type is t  2
The bag is 366,4 in, long,
The initial bag tension was  51 pounds.
The bag tension at the start of R'.A,  cleaning was, 261 pounds
The R.A. pressure was 1.0 inches w.g.
The support spring constant was 90 pounds/inch,
The ash weight was 65.25 pounds.
The bag length prior to R,A. application was 31.4583 feet.
The bag length after R.A. application was 31.3229 feet.
The bag
Ring 1
Section
Ring 2
Section
Ring 3
Section
Ring 4
o Section
01 Ring 5
Section
Ring 6
Section
Ring 7
Section
Ring 8
section dinensions are as follows:
Measured Effectiwe(arc) E

1

2

3

4

5

6

7

1.
56.
2.
54.
2.
51,
2.
49,
2.
48.
2,
45,
2.
43.
3,
50
13
00
50
00
75
00
75
00
00
00
75
00
50
50
in .
in,
in,
in,
in.
in,
in,
in,
in .
in.
in,
in.
in ,
in,
in ,
long,
long,
long,
long,
long,
long,
long,
long,
long,
long,
long,
long,
long,
1 orig ,
long,
1
57

56

53

51

49

47

44
3
.50
,00
.25
.25
.25
.50
,25.
.50
.25
.75
.25
,50
,25
,38
,50
in
in
in
in
in
in
in
in
in
in
in
in
in
in
in
. long
, long
, long
. long
, long
. long
. long
. long
, long
, long
. long
, long
. long
, long
. long
t

i

.
,
.
,
•
,
4
,
.
,
t
Bag deflections are as follows;
Section
Section
Section
Section
Section
Section
Section
Section
Dia.
1}
2}
3;
4j
5;

7<







5
5
5
5
5
5
5
at deflection
.250
.438
,604
,688
.313
,063
.125
in,
in.
in.
in.
in,
in,
in .
Catenary deflection














3,
2,
2.
2.
3,
3,
3.
063 in
969 in
886 in
844 in
031 in
156 in
125 in
.
t
1
,
f
t
1
                                                     Effective(straiglit)

                                                         56,98 in, long,

                                                         56,23 in. long,

                                                         53.48 in, long.

                                                         51,48 in. long.

                                                         49,73 in. long.

                                                         47,48 in. long,

                                                         44,36 in. long.
                                                           Equalizing deflection
                                                                 2.226 in.
                                                                 2.169 in.
                                                                 2.117 in,
                                                                 2,090 in,
                                                                 2,207 in,
                                                                 2.282 in,
                                                                 2.264 in.

-------
8
                                  APPENDIX I - TEST BESULTS  (CONT.)

            This is data for Run  7  of the bag deflection testing,
            The Bag Type is I  2
            The bag is 366,4 in.  long, 11,38 in, in dianeter and weighs 252 oz
            The initial bag tension was  85 pounds,
            The bag tension at the start of R.A, cleaning was 203 pounds.
            The R.A, pressure was ,5 inches w,g.
            The support spring constant was 90 pounds/inch,
            The ash weight was 65.25 pounds,
            The bag length prior  to R.A. application was 31,4583 feet,
            The bag length after  R,A, application was 31.4323 feet.
The bag section dimensions
                 Measured
Ring 1      1.50 in. long,
Section 1  56.13 in, long.
Ring 2      2,00 in. long,
Section 2  54,50 in, long,
Ring 3      2,00 in. long,
Section 3  51,75 in. long,
Ring 4      2.00 in. long,
Section 4  49.75 in. long,
Ring 5      2.00 in. long,
Section 5  48.00 in, long,
Ring 6      2,00 in. long,
Section 6  45,75 in, long,
Ring 7      2,00 in. long,
Section 7  43,50 in, long,
Ring 8      3.50 in. long,
are as follows:
       Effective(arc)
        1.50 in, long.
       57,00 in. long,
         .25 in. long.
       56.25 in, long,
         ,25 in. long.
       53.50 in. long,
         .25 in. long.
       51.50 in. long,
         .25 in, long,
       49,75 in. long,
         .25 in. long.
       47,50 in, long,
         .25 in. long.
       44.38 in. long,
        3.50 in. long.
Effective(straight)

    57.00 in. long,

    56,25 in, long,

    53.50 in, long.

    51.50 in. long.

    49,75 in. long,

    47.50 in. long.

    44,37 in, long,
            Bag deflections are as follows:
            Section
            Section 1;
            Section 2}
            Section 3;
            Section 4;
            Section 5;
            Section 6}
            Section 7;
            Dia.  at deflection
               6.938 in.
               7.438 in.
               7.646 in.
               7.436 in,
               7.244 in.
               7.063 in.
               6.813 in.
        Catenary deflection
             2.219 in.
             1.969 in.
             1.865 in.
             1.970 in.
             2.066 in.
             2.156 in.
             2.281 in.
      Equalizing deflection
            1,676 in.
            1.500 in.
            1.426 in,
            1.501 in.
            1.569 in.
            1.632 in.
            1.719 in.

-------
3
                                   APPENDIX I - TEST RESULTS  (CONT.)
              This is data for Run  8  of the bag deflection  testing.
              The Bag Type is 4  2
              The bag is 366.4 in. long,  11.38 in,  in  diameter and  weighs 252  oz,
              The initial bag tension was  72 pounds.1
              The bag tension at the start of R.A.  cleaning was 211  pounds..
              The R,A.  pressure was  .9 inches w,g,
              The support spring constant was 90 pounds/inch,
              The ash weight was 65,25 pounds,
              The bag length prior to R.A. application  was 31.4792  feet,
              The bag length after R.A. application was 31.4063 feet.
The bag section dimensions
                 Measured
Ring 1      1,50 in, long,
Section 1  56.13 in, long,
Ring 2      2.00 in, long,
Section 2  54.50 in, long,
Ring 3      2,00 in. long,
Section 3  51,75 in, long,
Ring 4      2,00 in, long,
Section 4  49.75 in. long,
Ring 5      2.00 in. long,
Section 5  48.00 in. Ipng,
Ring 6      2.00 in, long,
Section 6  45.75 in. long,
Ring 7      2.00 in. long,
Section 7  43.50 in, long,
Ring 8      3.50 in, long,
are as follows:
       Effective(arc)
        1,50 in.  long
       57.00 in,  long,
         .25 in,  long
       56,25 in,  long,
         ,25 in,  long,
       53,50 in,  long,
         ,25 in.  long
       51.50 in.  long,
         .25 in.  long
       49.75 in.  long,
         .25 in.  long.
       47.50 in.  long,
         .25 in.  long.
       44,38 in,  long,
        3,50 in,  long.
              Bag deflections are as>  follows:
              Section
            Dia,  at deflection
Section
Section
Section
Section
Section
Section
Section
1}
2}
3;
4}
5}
6}
7)
6.
6,
6,
6,
6,
6,
6,
250
500
646
667
330
438
229
in.
in.
in .
in.
in,
in,
in.
Effective(straight)

    56,99 in.  long.

    56.24 in.  long,
   \,
    53.49 in,  long.

    51.49 in.  long.

    49.74 in.  long,

    47,49 in.  long.

    44,36 in,  long,
        Catenary deflection
             2.563 in.
             2,438 in.
             2,365 in,
             2,354 in.
             2,523 in,
             2,469 in,
             2,573 in,
      Equalizing deflection
            1,908 in.
            1,825 in,
            1.776 in.
            1.768 in.
            1,882 in.
            1,846 in,
            1,915 in.

-------
o
00
                                  APPENDIX I - TEST RESULTS (CONT.)

             This is data  for  Run   9  of  the bag deflection  testing,
             The Bag Type  is *  3
             The bag is 366.4  in.  long,  11.25 in.  in  diameter arid weight, 256 oz
             The initial bag tension was   63 pounds.
             The bag tension at  the  start  of R.A.  cleaning was, 229 pounds.
             The R.A,  pressure was 1,0 inches w.g.
             The support spring  constant  was 90 pounds/inch.
             The ash weight was  65.25 pounds.
             The bag length prior  to R.A.  application was 31,3646 feet,
             The bay length after  R,A. application was 31.2813 feet.
The bag section dinensions
                 Measured
Ring 1      1.50 in. long,
Section 1  54.25 in, long,
Ring 2      2.00 in. long,
Section 2  51.50 in. long,
Ring 3      2,00 in. long,
Section 3  50.00 in, long,
Ring 4      2.00 in. long,
Section 4  50.00 in. long,
Ring 5      2.00 in. long,
Section 5  49.00 in. long,
Ring 6      2.00 in. long,
Section 6  47.63 in. long,
Ring 7      2.00 in. long,
Section 7  47.80 in. long,
Ring 3      3.50 in. long,
are as follows:
       Effective(arc)
        1.50 in. long.
       55.13 in. long,
         ,25 in, long,
       53.25 in, long,
         .25 in. long.
       51.75 in. long,
         .25 in, long.
       51.75 in. long,
         ,25 in. long.
       50.75 in. long,
         .25 in. long.
       49,38 in. long,
         .25 in. long.
       47.88 in. long,
        3.50 in, long.
Effective(straight)

    55.11 in, long.

    53.24 in. long.

    51.74 in. long.

    51,74 in, long,

    50,74 in. long,

    49.36 in. long.

    47.86 in. long.
                 deflections  are  as  follows;
             Section
             Section  Ij
             Section  2;
             Section  3)
             Section  4;
             Section  5;
             Section  6}
             Section  7j
            Dia,  at deflection
               6.250 in.
               6.021 in.
               5.962 in.
               6.313 in.
               5.896 in.
               5.771 in.
               4,808 in,
        Catenary  deflection
             2.500 in.
             2.615 in.
             2.644 in,
             2.469 in.
             2.677 in.
             2.740 in,
             3,221 in,
      Equalizing deflection
            1.865 in .
            1,940 in.
            1.959 in,
            1.844 in.
            1.981 in,
            2.021 in.
            2.316 in,

-------
o
vo
                       APPENDIX I - TEST RESULTS  (CONT.)
     is data for Run  10  of the bag deflection testing.
The Bay Type is 3  3
The bag is 366.4 in. long, 11.25 in. in diacieier arid weight, 256 02
The initial bag tension was  95 pounds.
The bag tension at the start of R.A, cleaning was 241 pound*.
The R.A, pressure was 1,0 inches w,g,
The support spring constant was 90 pounds/inch.
The ash weight was 65.25 pounds.
The bag length prior to R.A. application was 31,3854 feet.
The bay length after R.A. application was 31.3438 feet,
                                                     Effecliue(straight)

                                                         55,12 in.  long,

                                                         53,24 in.  long,

                                                         51,74 in ,  long,

                                                         51,74 in.  long.

                                                         50.74 in.  long.

                                                         49.37 in,  long.

                                                         47.87 in.  long.
The bag
secti
on difiensions are
Measured
Ring 1
Section
Ring 2
Section
Ring 3
Section
Ring 4
Section
Ring 5
Section
Ring 6
Section
Ring 7
Section
Ring 8

1

2

3

4

5

6

7

1
54
2
51
2
50
2
50
2
49
2
47
2
47
3
.50
.25
,00
.50
.00
.00
.00
.00
.00
.00
.00
.63
.00
.00
.50
in .
in.
in ,
in.
in ,
in,
in,
in.
in,
in,
in.
in.
in .
in.
in ,
long,
long,
long,
long,
long,
long,
long,
long,
long,
long,
long-,
long,
long,
long,
long,
as follows:
Effective(arc)
1
55

53

51

51

50

49

47
3
.50
,13
,25
,25
.25
,75
,25
,75
,25
.75
.25
.38
.25
.88
.50
in .
in ,
in ,
in.
in ,
in,
in ,
in,
in .
in .
in .
in.
in .
in.
in .
long
long
long
long
long
long
long
long
long
long
long
long
long
long
long
.
>
,
;
i
>
,
)
•
)
.
>
>
>
,
Bag deflections are ai, follows;
Section
                       Dia. at deflection
Section
Section
Section
Section
Section
Section
Section
1
2
3
4
5
6
7
i
;
i
i
)
]
i
6.
6,
6.
6.
6.
6.
5.
750
417
475
708
313
359
641
in,
in,
in.
in.
in.
in.
in.
Catenary deflection
     2,250 in,
     2.417 in.
     2.388 in,
     2,271 in.
     2.469 in.
     2.446 in.
     2.805 in,
Equalising deflection
      1.696 in.
      1.809 in,
      1.789 in,
      1.710 in.
      1.844 in.
      1.828 in.
      2.062 in,

-------
                          APPENDIX II - COMPUTER PROGRAM "DEFLEC"


10 REM This is prograM DEFLEC developed  by  J.  Musgrove  of Bechtel Power Corporation, Hou&ton, TX
20 REM This prograM was developed as  part  of  'OptiMizing  the Location of Anti-Collapse Rings
30 REM in Fabric Filter Bags,' Paper  89, presented  to The Fourth Synposiun on the
40 REM Transfer and Utilization of Participate Control  Technology
50 REM Houston, Texas, October 11-15,  1982
60 REM This prograM listing is current  as  of  Nove«ber 2,  1982
70 PRINTER IS 701,140
80 OPTION BASE 1
90 DIM L(9>,F<9),F1(9>,F2(9>,D<9),L1(9),DK9>,F9(9>,P9(9),T(9),T1(1>
100 MAT L=ZER@ MAT F=ZER8 MAT F1=ZER8 MAT  F2=ZER
110 MAT D'ZERS MAT LITERS MAT D1=ZER6  MAT  F9=^ZER
120 MAT T=ZER@ MAT Tl-ZERS MAT P9*ZER
130 PRINT "This is prograM DEFLEC to  deternine the filter bag"
140 PRINT "deflection in each section during  reverse  air  cleaning"
150 PRINT
160 CLEAR
170 DISP "What is the bag dianeter, inches"
180 INPUT D9
190 R9=D9/2
200 DISP "What is the total bag length,  inches?"
210 INPUT L9
220 DISP "How nany anti-collapse rings  are in the bag?"
230 INPUT 12
240 PRINT "NoMber of rings = ")I2
250 L8=L9-6~.25*12
260 K=I2-H
270 REDIM L(K),F(K),F1(K),F2(K),D(K),L1(K),D1(K),F9(K),P9(K),T(K)
280 DISP USING 290 ; K
290 IMAGE "Input all ",2D," individual  section lengths  (excluding ring  cover  dimensions)"
300 DISP "top to bottoM.  Be sure to  separate data by coMnas  (i.e.  56,52,55,60)"
310 MAT INPUT L
320 LU)=LUH.875
330 L(K)=L(K)+.875
340 FOR Z^2 TO 12
350 L(Z)=L(Z>+1.75
360 NEXT Z

-------
                    APPENDIX II - COMPUTER PROGRAM "DEFLEC" (CONT.)


370 PRINT g PRINT
380 DISP "What is the bag support spring constant,  Ib/irt?"
390 INPUT K9
400 DISP "What is the operating te«p,,  F?"
410 INPUT T8
420 DISP "What is the awbient tenp., F, during initial  bag tensioning?"
430 INPUT T6
440 T7-T8-T6
450 DISP "What is the weight of the bag Material,  oz./sq.yd,?"
460 INPUT C
470 DISP "What is the nuMber of fill yarns per inch of  fabric  (44 is typical)?1
480 INPUT Cl
490 C2=C1*D9*PI
500 C3-1/C1
510 DISP "What is the dianeter of the ring wire,  inches?"
520 INPUT R
530 DISP "What is the reverse air pressure,  in,  H20?"
540 INPUT A
550 T4=T7*(L9-H6)*K9*. 0000038
560 T5-Ax.5781*PI*R9A2/16
570 CLEAR
580 DISP e DISP e DISP  e DISP e DISP
590 DISP "DO NOT DISTURB!!!"
600 DISP "PROGRAM IN PROGRESS"
610 FOR II•=! TO 3
620 FOR 13=1 TO 6
630 ASM
640 T2=A2*.5781*PI*R9A2/16
650 P=A,*,5781*C3
660 T8(l)=0
670 FOR 1=1 TO K
680 Y6=.OQ5*EXP«I-1)*.844>
690 Y7=Y6+.5673
700 Y8=Y7/100
710 Y9=1-Y8
720 L2(I)=Y9#L(I)

-------
                     APPENDIX II - COMPUTER PROGRAM "DEFLECT (CONT.)
,?3D D9a>=SOR<,375*L2~L2U»)
740 A9"ATN<4KD9/L2(I))
750 F9(I>*,07833+,00407«I
760 T«F9+T(I>
780 NEXT I
790 T*6Q*(Il+3>/4
800 Ai-00
810 T1=.015625KI3
820 U2=A1*T1*C3*16/1728
830 W5*RA2*PIA2*,284*U*12/<4*C2>
840 W1*C/(9*C1*144)
850 W4=B*W1
860 W3^W4+W5
870 W6=T1*PI*D9*a9-6>*Al/l728
880 T3*T+T4+T2
890 TS(1)=T8U)*C2/16
900 T9"T+T4~T5+T8<1>
910 PRINT "   Bag          Bag
920 PRINT " Length,      Dianeter,
930 PRINT " inches        inchet
940 PRINT USING 950
 Operating
Teciperature,
 Degree?, F
  Ambient
Tenperature,
 Degrees F
                      L9,D9,T8>T6>RJ3
950 IMAGE 2X,3D.D,8X,D».DD,10X,3D,12X,3D,10X,D,4D,9X,3D,D
960 PRINT g PRINT & PRINT
970 PRINT "  Weight
980 PRINT " of cloth,
990 PRINT "o:./sq.yd.
1000 PRINT USING 1010
                         Initial
                         Tension,
                          pound
                                     Forward  Ga
                                      Pressure,
                                       in.w.c.
Total Ash
 Weight,
 pounds
                        C,T,A2,W6,A,T9
1010 IMAGE 4X,2D,9X,3D.D,BX,2D.2D,10X,3D,D,11X,D,2D,1QX,3D.D
1020 GOSUB 1220
1030 PRINT USING
1040 NEXT 13
10SO PRINT USING
1060 NEXT II
•J07D PRINT USING
1080 CLEAR
                 "K"  ;  CHR*(27) ; "
                 "K
                 "K
                       CHR$(11)
                       CHR$(11)
                                                                    Dia. of
                                                                   Ring wire,
                                                                     inches
                              Reverse Air
                               Pressure>
                                in.w.c.
                                                                                  Tension"
                                                                                Prior to RA,"
                                                                                   poondc"
                               lension"
                             Start  of RA,"
                               pounds"

-------
                    APPENDIX II  - COMPUTER PROGRAM "DEFLEC"  (CONT.)
10?0 D1SP "     J   000   BBEB"
1100 DISP '     J  0   0  E   B"
1110 DISP "     J  0   0  BBBB"
1120 DISP " J   JO   OB   B"
1130 DISP "  JJJ    000   BBBB"
1140 DISP •"
1150 DISP ""
1160 DISP " EEEEE  N   N  DDDD"
1170 DISP ' E      NN  N  D   D"
1180 DISP " EEE    N N N  D   D"
1190 DISP "' E      N  NN  D   D"
1200 DISP " EEEEE  N   N  DDDD"
1210 END
1220 PRINT
1230 FOR 1=1 TO K
1240 P9-^P*,673
1250 F2/2
1260 ON I GOTO 1270,1280,1290,1300,1310,1320,1330
1270 Kl'O @ GOTO 1340
1280 Kl-L(l) 6? GOTO
1290 K1=L(1)+L(2) 6 GOTO 1340
1300 K1-L(1)+L(2)+L(3)  8 GOTO 1340
1310 K1-L(1)+L(2)+L(3)+L<4) 8 GOTO 1340
1320 K1=L(1)+L(2)+L(3)+L(4)+L(5)  P GOTO 1340
1330 K1=L(1)+L(2)+L(3)-+L<4)+L(5)+L(6)  9 GOTO  1340
1340 F1(I)*T9*(16/C2)-K1*(W1+W2)-U3*(I-1)
1350 F1-T9*<16/C2)-L8*(WHW2)-W3K(1-1)
1360
1370
1380 IF FKO THEN F=-F
1390 D(I)=P9-x-L(I)A2/(4-x-SQR((2'X-F(I))A2-(P9*L(D)ft2))
1400 NEXT I
1410 HAT Ll^ZER
1420 MAT D1=ZER
1430 FOR J=l TO K
1440 LKJ)«INT(100flKL/1000

-------
                     APPENDIX II - COMPUTER PROGRAM  "DEFLEC"  (CONT.)
145& DHJ)
!1£! PRINT "Seeti°nV;HrlljL1R9 THEN GOTO 1480  ELSE 1490
1480 PRINT "               This  section is  pancaking!!"
1490 NEXT J
1500 PRINT USING 1510 ;  F
1510 IMAGE "Tension at  the botton  of the  bag  (just prio^  to cleaning) =",3D.D," pounds."
1520 IF F<0 THEN GOTO 1530 ELSE  1540
1530 PRINT "BAG IS BLOWING INTO  THIMBLE DUE TO  EXCESSIVE  ASH OR INSUFFICIENT TENSION!!"
1540 RETURN

-------
Ul
                           APPENDIX III - COMPUTER PROGRAM "RINGOP"
10 RE.M This is progran RINGOP developed by J,  Musgrove of Bechtel  Power  Corporation, Houston, TX
20 REM This prograM was developed as part of "Optimizing  the Location  of Anti-Collapse  Rings
30 REM in Fabric Filter Bags,' Paper 89, presented to  The Fourth  SyMposiuM on  the
40 REM Transfer and Utilization of Particulate Control Technology
50 REM Houston, Texas, October 11-15, 1982
60 REM This prograM listing is currerrt as of NoveMber  2,1982
70 REM "This prograM RINGOP determines the optiMUM location  of anti-collapse rings"
80 REM "in filter bags to assure equal deflection  in each section  during reverse air cleaning"
90 PRINTER IS 701,140
100 OPTION BASE 1
110 DIM L(9),F(9),F1<9),F2<9>,D(9),P(9),F9<9>,T<9>,T8(1),L2(9)
120 PRINT "This is prograM RINGOP to deternine the optiMUM location  of anti-collapse rings"
130 PRINT "in filter bags to assure equal deflection  in each section during  reverse air cleaning"
140 MAT L-ZERQ MAT F~ZER@ MAT Fl^ZERP MAT F2~ZER
150 MAT D=ZER@ MAT P^ZERS MAT F9=ZER
160 MAT T=ZER(? MAT L2=ZERP MAT T8=ZER
170 CLEAR
180 DISP  "What is today's date? iWdd/yyyy"
190 INPUT At
200 PRINT A$
210 PRINT 6 PRINT
220 CLEAR
230 DISP  "What is the bag dianeter, inches"
240 INPUT D9
250 R9=D7/2
260 DISP  "What is the total bag length,   inches?"
270 INPUT L9
280 DISP  "What is the bag support spring
290 INPUT K9
300 DISP  "What is the operating teMp., F?
310 INPUT T8
          "What is the artbient teMp., F,
                                               constant,  lb/iri?'
     320 DISP
     330 INPUT T6
     340 T7=T9-T6
     350 DISP "What is the weight of the bag
     360 INPUT C
during initial bag tensioning?"
                                          Material,  oz./sq,yd.?"

-------
                     APPENDIX III - COMPUTER PROGRAM "RINGOP" (CONT.)
370 DISP "What is the number of fill yarns per inch of fabr-ic (44 is typical)?"
380 INPUT Cl
390 C2=C1*D9KPI
400 C3=1/C1
410 DISP "What is the diameter of the ring wire,  in.?"
420 INPUT R
430 DISP "What is the reverse air         pressure, in.  H20?"
440 INPUT A
430 P=A*,5781XC3
460 T4~T7*
610 W1=C/(9*C1*144)
620 U4=8*U1
630 U3=U4+W5
640 W6=T1»PI»D9K-(L9-6)*A1/1728
650 T3==T+T4fT2
660 PRINT "   Bag          Bag
670 PRINT " Length,       Dianeter,
680 PRINT " inches.        inches
690 PRINT USING 700
                                      Operating
                                     Temperature,
                                      Degrees F
  Artbient
Temperature,
 Degree's, F
 Dia.  of
Ring wire,
  inches
                      L9,D9,T8,T6,R,T3
700 IMAGE 2X,3D.D,8X,DD.DD,10X,3D,12X,3D,10X,D.4D,10X,3D.D
710 PRINT B PRINT 8 PRINT
720 FOR 12=1  TO 6
  Tension"
Prior to RA,"
   pounds"

-------
                     APPENDIX III - COMPUTER PROGRAM "RINGOP"  (CONT.)

730 LB=L9-6-,25*12
740 T8(l )=0
750 K=I2+1
760 REDIM L(K),F(K)>F1(K),F2(K),D(K),P(K)>F9(K),T(K),L2(K)
770 FOR J=l TO K
780 L(J)=L8/K
790 NEXT J
800 FOR Zl=l TO K
810 Y6=,005*EXP«Z1--1)*.844>
820 Y7=Y6+.5673
830 Y8-Y7/100
840 Y9=1-Y8
850 L2*Y9*L(Z1)
860 D9*-L2=F9(Z1)*L
910 NEXT Zl
920 T8U)=T8<1)*C2/16
930 T9-T+T4-T5+T8U )
940 GOSUB 1160
958 NEXT 12
960 PRINT USING  "K" ; CHR*U1)
970 NEXT 13
980 PRINT USING  "K" ; CHR*(11)
990 NEXT II
1000 PRINT USING "K" ;  CHR$(11)
1010 PRINT  A*
1020 CLEAR
1030 DISP "     J   000   BBBB"
1040 DISP "     J  0   0  B   B"
1050 DISP "     J  0   0  BBBB"
1060 DISP *  J   JO   OB   B"
1070 DISP "  JJJ    000   BBBB"
1080 DISP ""

-------
                     APPENDIX III  -  COMPUTER PROGRAM "RINGOP"  (CONT.)

1090 DISP ""
1100 DISP " EEEEE  N   N  DDDD"
1110 DISP • E      NN  N  D   D"
1120 DISP " EEE    N N N  D   D"
1130 DISP " E      N  NN  D   D"
1140 DISP " EEEEE  N   N  DDDD"
1150 END
1160 PRINT 6 PRINT
1170 PRINT "  Height      Initial      Forward Gas     Total At,h     Reverse Air      Tension"
1180 PRINT " of cloth,    Tension,      Pressure,       Weight,       Pressure,     Start  of RA,"
1190 PRINT "oz./£,q,yd.     pounds        in.w.c.        pounds          in.w.c.         pounds"
1200 PRINT USING 1220 j C,T,A2,W6,A,T9
1210 PRINT
1220 IMAGE 4X,2D,9X,3D.D,8X,2D.2D,10X,3D.D,11X>D.2D,11X,3D,D
1230 J9=0
1240 PRINT "Nuwbei" of Rings = ";I2
1250 FOR 1=1 TO K
1260 P9«P«.673
1270 F2(I)=P9*L(I>/2
1280 ON I GOTO 1290,1300,1310,1320,1330,1340,1350
1290 Kl-=0 S GOTO 1360
1300 K1-~L(1) 8 GOTO 1360
1310 K1«L(1)+L(2) 6 GOTO 1360
1320 K1~LU)+L(2)+L(3> § GOTO 1360
1330 Kl==L(l)-H-<2)+L<3>+L<4) 8 GOTO 1360
1340 K1=L(1)H.(2)-H.<3)+L(4)-H.<5) 6 GOTO 1360
1350 Kl-L(mL(2)+L<3)H.<4>+L<5>+L<6)
1360 FKI)=T9*(16/C2>-K1»(W1+W2)-W3*(I-1)
1370 Fl=T9*(16/C2)-L8*(Wl+W2)-U3x(I-l)
1380 F(I)=SQRA2->
1430 IF D(I»D9/3 THEN GOTO 1650
1440 NEXT I

-------
                     APPENDIX III - COMPUTER PROGRAM "RINGOP"  (CONT.)

1450 Xl-AMAX(D) S X2=*AMAXROW
1460 Y1=AMIN(D) S Y2-AMINROW
1470 Z=X1-Y1
1480 IF Z<.1 THEN 1540
1490 L(X2)=L-.125 8 L(Y2)=L(Y2)+,125
1500 J9=J9+1
1510 IF J9>200 THEN 1530
1520 GOTO 1250
1530 PRINT "Iterative lirtit  reached at  200"
1540 FOR J=l TO K
1550 L(J)=INT(100Q*L(J))/1000
1560 D/1000
1570 PRINT "Section";Jj"=";L(J);"inches long  (effectively); its  deflection =";D(J);"inches"
1580 NEXT J
1590 PRINT USING 1600 ; Fl
1600 IMAGE "Tension at the bottoM  of the  bag  (at  the start of cleaning) = ",3D,D," pounds,,"
1610 IF FKO THEN GOTO 1620  ELSE 1630
1620 PRINT "BAG IS BLOWING INTO THE THIMBLE DUE TO EXCESSIVE ASH OR INSUFFICIENT TENSION!!"
1630 GOSUB 1670
1640 RETURN
1650 PRINT "Deflection greater than 1/3 dianeterj further analysis terminated."
1660 RETURN
1670 L(l)=L(l)-.875
1680 L
-------
         PULSE JET ON-LINE CLEANING FILTER FOR FLY ASH

              by: Wayne G. Wellan
                  Carter-Day Company
                  Minneapolis,  Minnesota 55432
                           ABSTRACT

     This paper describes the development and performance of
on-line cleaning by pulse jet filters with needled felt.  This
pulse jet filter design is in operation on various types of
coal-fired boilers.  Considerable research and testing was in-
volved in finding a fiber that would endure continuous pulsing,
high temperatures, and a sulfur environment.  Additional re-
search was required to develop the proper filter media in the
areas of needling and scrim design.  Also, analysis of various
types of fly ash was conducted to determine particle size and
shape, and what effect it has on the performance of the filter.
                              420

-------
                         INTRODUCTION


     Eight years ago it was decided by our management that
Carter-Day should 'develop a filter for the flue gas filtration
market,  specifically industrial and utility coal-fired boilers.
The market trends at that time showed a tremendous growth due
to the oil embargos and increase in prices of oil and gas.
Also, other people in the baghouse manufacturing industry had
tested a reverse air fiberglass baghouse with success.

     Our first test site was a power plant at Granite Falls,
Minnesota operated by Northern States Power.  This power plant
generated 50 megawatts of electricity from a pulverized coal-
fired boiler.  They burned a blend of western and Kentucky coal.
This plant is a peaking station.  They operate 8 to 16 hours
per day.  Therefore, they pass through the acid dew point on
start-up and shut-down, which is a severe test for the fiber.
We installed two slip-stream filters.  One was a 24RJ96 filter,
the other was a high velocity filter.  The volume on the RJ
filter was 2,500 cfm and the volume for the high velocity filter
was 3,000 cfm.  These units were put on stream in February,
1974.

     Some of the first fibers we evaluated were Nomex, Polyi-
mide, Kynol, DrayIon T, Orion, CS treated Nomex, acrylics, woven
fiberglass, which all failed.  We also tried teflon felt.  This
withstood the high temperature gas and sulfur conditions, but
had very poor collection efficiency due to the on-line cleaning.
Extensive testing was conducted on stainless steel fibers in an
air laid web and also in a sintered construction, and none of
these were successful.  We also evaluated a woven fiberglass
bag in the RJ filter which was in the round bag configuration.
This did withstand the operation for a few months, but the bags
eventually failed.  Tests were also conducted on the high
velocity filter using teflon in a felt and pile construction.
None of  these were successful.

     We  then installed a 48RF96 filter and a standard Pulse Jet
filter,  14PJ96, at the Dairyland Co-op Power Generating Station
at Alma, Wisconsin.  This is an 80 megawatt power generating
system that has a pulverized coal-fired boiler which burned high
sulfur eastern coal, low sulfur western coal, and petroleum
.coke. This plant is a base loaded utility station which
operates 24 hours a day.  The RF filter was designed to handle
3,500 cfm and the volume for the Pulse Jet is 600 cfm.  The
slip-stream filters are designed to operate around the clock.
We have  our own ash conveying system which is tied into their
system.   These filters are operated unattended, other than
having the filter pressure drop, air flow and cleaning pressure
recorded and mailed to us once a week.  Both of these systems

                              421

-------
are designed for on-line cleaning.  We did not have a backup
filter for off-line cleaning.  Our design and premise of this
whole program is with on-line cleaning.  So, the bags are pulsed
while the flue gas is entering the filter.  We again tried some
of the same felt we had at Granite Falls which failed in the
same manner.  We also evaluated other medias such as Kermel,
carbon, graphite, ceramics, and Hyglass, which all failed.  We
have evaluated a newer Hyglass with a 3/4 inch mesh round cage
design, and failed after 9 months of operation.

     During this testing interval we came upon our new fiber
called Ryton and we installed 3 bags in our Pulse Jet unit and
proceeded to life test the bags.  After 9 months we found that
there was no deterioration or failures.

     We acquired additional fiber to completely fit out the
Pulse Jet filter and the RF filter at Alma.  These bags were
evaluated initially for collection efficiency and also for bag
life.  The collection efficiency was not satisfactory on the
first set of bags and were replaced with a heavier weight Ryton
in order to meet the new proposed EPA code, 0.03 to 0.05 Ibs
per million BTU.  We also changed the bag frame by adding addi-
tional vertical and hoop wires to reduce the flexing of the
media, which would cause the media to fail.  We tested the new
media and our collection efficiency did improve.  Our results
were .018 pounds per million BTU.

     We continued testing the bags to determine bag life, and
after 2 years without any bag failures, we proceeded to work
with our fiber supplier to get a commitment to supply us with
fiber for future jobs.

     When we first contacted Phillips they were in the develop-
ment stages of their fiber grade Ryton resin; therefore, they
were not ready to release their fiber for commercial use.
Carter-Day got involved and developed a method to spin the Ryton
resin into fiber for the baghouse systems that are currently in
operation.

     You probably have seen Ryton advertised in various maga-
zines.  There are 2 different types of Ryton resin.  One is
their molding grade resin which is solely used for injection
molding, and the other is fiber grade resin.  The molded grade
resin cannot be spun into fiber.  It is a highly cross-linked
resin which makes the fiber brittle.


                         FIBER TESTING


     During the course of this program we develope'd a labora-
tory accelerated acid and temperature test to evaluate fibers
for the flue gas environment.  The fibers are wicked in an acid

                              422

-------
solution  of  75%  H^SO^ at 350°F for 3 hours.  Then tensile tests
are conducted  on individual fibers and are compared to fibers
before the acid  treatment.

     This test procedure has worked out quite well to screen out
new fibers,  and  also serves as a quality control check on pro-
duction lots of  Ryton fiber.  If the Ryton fibers are not pro-
cessed properly  they will deteriorate in a flue gas environment
in a very short  time.

     Another procedure we use to evaluate fibers is our labora-
tory microscope  with a special lens.  This enables us to see any
deterioration  caused by acid attack and also shows us high stress
concentration  in the fiber caused by over crimping the fiber
during the fiber spinning process.

     In 'addition to the lab tests, bags are removed from the
slipstream filters, and from various installations in the field.
Physical  tests are  conducted on these bags such as tensile,
tear, and mullen burst.  See Figure 1.  From this data we can
project the  bag  life in various flue gas conditions.  Also, all
new production lots of Daytex felt are checked for physical
strength, and  one or two bags from each production lot is placed
in our slip-stream  filter.


                 NEEDLING AND SCRIM TECHNOLOGY

     During  the  test program on our slip-stream filters, we were
able to evaluate the long term effect on the filter bags.  After
2 1/2 years  in a flue gas environment, we have seen evidence of
delamination of  the fiber from the scrim which is related to how
well the  felt  is needled.  So, our next step was to evaluate the
needling  technology and develop ways to improve the process.
After a long series of tests we were able to come up with an
optimum procedure that doubled the tensile and mullen burst
strength  of  the  felt.  The Ryton felt that is currently being
produced  is  needled with the latest technology.

     The  main  mode  of failure is caused by flexing which is
accelerated  by the  sulfuric acid and high temperature environ-
ment.  In order  to  exceed the 2 year bag life, we had to improve
the needling and scrim design.  The scrim is a woven screen made
of a fluorocarbon base fiber.  This is placed in the center of
the felt  to  give it additional strength for flexing.  A series
of tests  were  conducted on various scrim designs to improve the
strength  of  the  felt.  These series of tests were conducted at
the same  time  we evaluated the needling techniques.  So, with
the combination  of  better needling and a stronger scrim we will
exceed our 2 year guarantee.
                               423

-------
100
80
   £ 60

   I-
s  o



     40
20
       (A) TENSILE  (FILL)


       (B) MULLEN


       (C) TEAR (WARP)
              12
                                            4-
30    36    42
                         18    24

                         MONTHS

  FIGURE I- AGING EFFECTS OF OAYTEX IN A FLUE GAS ENVIRONMENT

-------
     "What  is  Daytex"

     This is the  filter media developed by Carter-Day for fly
ash applications,  and  consists of:

     1.  Ryton  fiber -  (Polyphenylene Sulfide) resin developed
        by  Phillips Chemical.

     2.  Scrim  -  fluorocarbon support grid optimized by Carter-
        Day.

     3.  Needling  technology - optimized by Carter-Day.


              PARTICLE SIZE VS FILTER PERFORMANCE


     One of the other  areas that determine a successful
operating filter  system on fly ash is knowing the particle size
and shape.  We have done extensive work in analyzing various fly
ashes from  boilers around the country, and also reviewed the
operating performance  of the baghouse at these locations.  From
this we  compared  it to the data we have accumulated from our
field installations and slip-streams, and are able to recommend
the proper  baghouse size for future fly ash applications.

     We  classify  the various fly ashes into 3 major groups.  One
is fly ash  from  an inefficient fired stoker, such as dump
grates,  some underfeed stokers, or a stoker that is operating
with high excess  oxygen, or upset conditions.  Under these con-
ditions, the stoker will create a very fine soot or unburned
hydrocarbons which are sub-micron.  See Figure 2.  Therefore,
you have to reduce your air to cloth ratio in order to maintain
2-5 inches of  pressure drop across the baghouse.

     Another type of stoker we refer to is an efficient
operating stoker, such as a traveling grate, where there is no
upset conditions  in the fire bed.  This type of stoker will pro-
duce large  unburned carbon particles, lOOy to 200y MMD.  See
Figure 3.   This  type of fly ash is very easy to filter, there-
fore we can size  our baghouse to operate at a higher air to
cloth ratio.

     The other major type of fly ash is from a pulverized coal-
fired boiler.   The particle size will range from 4p to 8y MMD
when burning western coal and lOy to 20u MMD when burning other
coals.  See Figure 4.   The biggest concern is the spherical
shape of the  particles which can bleed through the media during
the cleaning cycle if the media is not designed properly.  The
Daytex will eliminate  the seepage of these particles during the
cleaning cycle.
                               425

-------
FIGURE  2- FLY ASH FROM INEFFICIENT FIRED STOKER

-------
6
            FIGURE 3-FLY ASH FROM EFFICIENT FIRED  STOKER

-------
to
co
                  FIGURE 4- FLY ASH FROM P-C FIRED

-------
                   BAGHOUSE DESIGN FEATURES

     The  filter used for the industrial and utility boiler
applications  is the RF filter.   See Figure 5.  The gases enter
at the  bottom of the filter, and the fly ash is collected on the
outside of the filter tube.  The dust cake is released from the
filter  tube by a high energy pressure wave.  This pressure wave
is released from a quick opening diaphragm valve, then passes
through special orifices on a rotating manifold.  The cleaning
system  is a high volume, low pressure design (7.5 psi).

     The  weight of the Daytex filter tube is 22-25 oz/yd2 and
has proven to have excellent acid and temperature resistance up
to 400°F.  The length of the filter tubes are 8 to 14 feet and
are removed from the top side of the filter.

                            SUMMARY

     Ryton fiber with proper needling and scrim design is an
excellent felt for the Pulse Jet filters on industrial and
utility coal-fired boilers.

     The  particle size and shape of the various fly ashes has a
major effect  on the performance of the baghouse.

     As for bag life, none of the bags at the Coors facility
have failed,  which have been in service for 33 months.

     The  RF filter system can operate with "on-line cleaning"
or "off-line  cleaning."
     The work described in this paper was not  funded by  the  U.S.
 Environmental Protection Agency and therefore  the  contents do
 not necessarily reflect the view of the Agency and no  official
 endorsement should be inferred.

                               429

-------
                                    SOLENOID
                                      VALVE
   DRIVE MOTOR
 FOR REVERSE PULSE
AIR CLEANING SYSTEM
      DIAPHRAGM
        VALVE
AIR RESERVOIR TANK
FOR REVERSE PULSE
AIR CLEANING SUPPLY
 FILTER TUBE SHEET
 FOR HOLDING FILTER
TUBE CAGE ASSEMBLY
   CEA'Carter-Day
DAYTEX FILTER TUBES
   AIR INLET
   COLLECTED
  PARTICULATE
    OUTLET
          FACTORY INSTALLED
        INSULATED TOP ACCESS
          DOORS FOR FILTER
       TUBE INSPECTION SERVICE
          HIGH TEMPERATURE
           ACID-RESISTANT
             GASKETING
               CLEAN
              AIR OUTLET
             CLEANING AIR
              MANIFOLD
                                                             NOTE:
                                                     FILTER BODY AND HOPPER
                                                        INSULATED IN FIELD
                                                          HOPPER ACCESS
                                                          HOLE NOT SHOWN
          60° CONICAL HOPPER
             FIGURE  5 -  CARTER-DAY  RF FILTER
                            FOR  COAL FIRED  BOILERS
                                   430

-------
                 TOP  INLET  VERSUS  BOTTOM INLET BAGHOUSE DESIGN

                  by:  Robert M.  Jensen
                       Bechtel  Power Corporation
                       San Francisco,  California  94119
                                   ABSTRACT

     This  paper  compares  top  inlet with bottom inlet baghouse designs for bags
that collect  on  the inside.  The  comparison  includes  performance  data for top
inlet baghouses  in  service  with comparable bottom inlet performance.

     The  paper  explains  why  pressure  loss predictions for  top inlet may  be
more reliable than  for bottom  inlet.   Other  advantages claimed for  top  inlet
are higher cloth ratio for  the same  pressure loss  as bottom inlet; use  of
longer bags;  better utilization of unreacted reagent  after a spray dryer;  and
more efficient cleaning with  the reverse air.

     The  most important  advantage of  top inlet  design may be the  apparent
ability for all bags  to  operate  at  a constant  face  velocity in contrast  to
bottom inlet  bags  in  which  the face  velocity diminishes with  time on  line
after each cleaning.
     Until  a few years ago it was assumed  that  no  correlation existed between
the pressure loss of a baghouse  on  coal-fired flue gas and the  properties  of
the coal burned and  its  ash.  We now  have ample evidence that  a  correlation
does exist  but we do not know what  it  is.   One consequence of  this  situation
is  that we cannot  predict  accurately  the pressure  loss for  a baghouse  on
coal-fired  flue gas.

     We do  know that a correlation  exists  between pressure loss and particle
size.  We  know that  pressure  loss increases  as  the median  particle  size
decreases.   Anything that  causes the  median particle  size  to  decrease  will
increase pressure  loss and  should be avoided.

     For baghouses on coal-fired  flue  gas, it is possible that  the  degree  of
unpredictability of pressure  loss  is  the  same, or  nearly the  same, as  the
degree of unpredictability of the amount of  shift  in  particle size toward the
smaller size caused by hopper  fallout  and  reverse gas recirculation.  If  we

                                     431

-------
 eliminate the size shift, we may  also eliminate the uncertainty in predictions
 of pressure loss.  One way  to  avoid,  or reduce,  the shift may be  to use  top
 inlet instead of  bottom inlet baghouse design.

      Bottom inlet  baghouses decrease  the  median  particle size  entering  the
 bags by allowing  coarse particles  to fall  out on entering the hopper and by
 recirculating the  reverse  gas  flow with  its fines  back to  the  inlet duct.
 This shift  toward smaller'  particles can  be  avoided  if we  eliminate hopper
 fallout and if we filter the reverse gas in a  separate system.

      This paper  compares top inlet baghouse design with bottom inlet baghouse
 design  for cylindrical  bags  that collect  on  the   inside.   The  schematic
 arrangements of  these two types of baghouses  are  shown for the filtering mode
 in Figure 1 and  for the cleaning  mode in Figure 2.
                    V
                             hO-
V
                   BOTTOM INLET                   TOP INLET

                            Figure 1.  Filtering mode.
     When reliable application data are available, a bottom inlet baghouse can
be  designed with assurance  that  it will work within the  predicted pressure
loss and  not exceed the allowable emission limit.   Reliable  application data
have  been  available  for  many  years  for  a  great  variety  of  industrial
applications.  Until recently, it has been assumed  that  baghouses can be used
to  remove  fly ash  from coal-fired  flue gas using design  criteria  that have
been  satisfactory  for  similar  materials  and  operating  conditions.   This
assumption  is now  in  question  as a  consequence  of the  inability  of some
baghouses on coal-fired flue gas  to meet predicted  pressure loss and, in some
cases, to stay within allowable emission rates.

     One reason  for  this situation may be  the highly variable  nature of fly
ash.   Another reason  may  be some  undesirable  inherent  characteristics  of
bottom inlet baghouses.
                                      432

-------
                                               V
                                                        o-
                   BOTTOM INLET                   TOP INLET

                            Figure 2.  Cleaning mode.
    Two  of  these undesirable characteristics of  bottom inlet baghouses are,
first,  the  unavoidable fallout  of  coarse material when  the  dirty gas enters
the hopper and,  second, the  unavoidable  dead end at the top of the bag.  They
cannot  be corrected  for  obvious  reasons.   An  explanation  of why  they are
undesirable follows.
     Consider the formula used for the pressure loss across cloth and cake:

                     C.
     AP = k'V +
                  x
                    7000
                         v
                         x
x V
x v
                             k'V
(1)
where
     AP =
     Ci  =

      t  =
      V =
  the pressure loss across the cloth and cake  in  one  compartment  just
  before cleaning,  inches of  water gauge (in. wg)
                                              3
  grain loading at  bag inlet, grains/actual ft   (gr/acf)

  time between cleanings, minutes

  air-to-cloth ratio = face velocity,  ft/min,  fpm
= resistivity of new cloth (negligible),
                                                 in. wg
                                                 (fpm)
                                     433

-------
      k'  = resistivity of filter  cake, 	ln' Wg—=-
                                       (fpm)(lb/ftZ)
                                                        111  W£?
      kl = resistivity of  cloth and cake after cleaning, -T-^—x
       3             "                                   (fpm)

 where

       k = Kozeny-Carman constant, dimensionless

         = 5  (generally accepted value)

      n  = gas viscosity,  lb(mass)/ft • sec
       o
                                          3
      6p = specific particle density, Ib/ft
                                      2
      Ap = surface area of particle, ft
                                3
      Vp = volume of  particle, ft

       e = porosity = void fraction = 1 - a

           bulk density of particle      ,.       ,
       a = 	TT~.—j—.„   e	„. , > dimensionless.
           specific density of particle

      The terms kJV and k'V are usually neglected; k|V because it  is very  small
 and  k'V because it is the pressure loss immediately after cleaning which does
 not  have to  be  guaranteed and is not used  in  evaluations.   Our basic  formula
 is thus:


                Ci         2
      AP  =  k2 X 7000 X  C X V  '
     If  we assume  spherical  particles and  typical values  for  the  terms  in
Equation   (2),  we  can  solve  for  k'  for  different  particle   sizes.    The
relationship  of  k'  and particle size  is  shown in Figure  3.   For the assumed
conditions, Figure  3  shows that k' is less than unity  and therefore reduces
pressure  loss for particles greater than 10 microns and that k!  and pressure
loss increase rapidly as the particle size gets smaller below  10Tmicrons.

     The value of k' used  in  Equation  (3)  is usually the median particle  size
entering  the  baghouse.   It should  be  the median  particle size  for  the  size
distribution  entering  the  bottom of the  bag.  We  do  not know  and we cannot
determine the particle size distribution entering the bag.  This is true for a
number of reasons.
                                     434

-------
                     100.0
                      10.0-
                  5
                  K
                                PARTICLES:
                                 SPHERICAL, d DIAMETER
                                 BULK DENSITY 40 LB/FT3
                                 SPECIFIC DENSITY 160 LB/FT3
                         0        10         20         30
                          SPHERICAL PARTICLE DIAMETER. MICRONS
                       figure 3. Variation of kfc with particle size.
     When  writing a baghouse  specification or when designing a baghouse  it is
customary  to calculate the grain  loading entering  the baghouse.  This  is done
by calculating the total  ash, using the  coal burn  rate  and the percentage of
ash in the coal.   An  estimate  is  then made  for the amount  of total ash that
will become fly  ash.   This gives  us an  approximate rate  for the quantity of
fly ash.

     The   gas  volume  entering  the  baghouse  is  also  an  approximation.   It
includes  estimates of  inleakage,  airheater  leakage,  ash  hopper evaporation,
and margin.  The number of grains per actual cubic foot is thus the  result of
dividing one approximation by another.

     On the other  hand,  if we  are working with a baghouse  in service we can
make a traverse  of the inlet duct  and determine  grain  loading and particle
size distribution  entering the baghouse or  a  compartment.   This information
does not   serve our purpose for several  reasons.   First,  it  differs from the
conditions at the  inlet to the bags.   Second,   it  may  not include the  fly ash
that is recirculated during reverse air cleaning.   And third, it includes the
coarse fraction that will fall  out when entering the hopper.
                                      435

-------
     We  cannot  determine  the  grain  loading  and  particle size  distribution
entering the bags because  we  cannot  prevent nor measure hopper  fallout.   The
gas  flow rate  into a  compartment  varies;  it  is  highest immediately  after
cleaning and lowest immediately before the next  cleaning.   Hopper fallout will
therefore vary because of the variable inlet velocity.

     As  the gas  flow  rate into  a  compartment  diminishes,  so  will the  gas
velocity entering  the  bottom  of  the  bags.  A  lowered velocity  entering  the
bags means  a reduced  ability  to carry  solid particles.   As  a  consequence,
hopper  fallout  increases   and the  mean  particle  size  entering  the  bags
diminishes.

     A bottom inlet design  is thus seen to cause a shift in the particle sizes
entering  the bags  toward   the  smaller  sizes.   As  the median  size  of  the
particles entering  the bags  diminishes, k'  increases and  so does  pressure
loss.  This is an undesirable inherent characteristic of bottom inlet designs.

     Another inherent characteristic  of bottom inlet design is  the dead end at
the top of  the bag.  When  the  dirty gas  enters  the bottom of a bag,  it  has a
certain  vertical  velocity.  As the  gas  passes  through the  bag, the  volume
remaining and  rising  in  the  bag  diminishes  and,  therefore,   so  does  the
vertical velocity.  As the  vertical  velocity diminishes, so does the ability
to carry solid particles.   As  the  vertical velocity approaches  zero at the top
of the bag, only the smallest particles  will be  carried upward.   As  a result,
fines accumulate at the top of  the bag.

     Several investigators  report that  they have determined  experimentally
that there  is no  accumulation  of  fines  in  the  top of bottom  inlet  bags.   In
each of  these experiments  there was  a time lapse between  taking  the bags  out
of service  and inspection  of the  bag.   During that time,  the  compartment  was
opened to the atmosphere, it was cooled and purged by pulling outside air into
it, and  there was  reverse  flow through  the  cloth and down-flow  in  the  bags.
During this time there was a  gradual  reduction  in temperature,  which reduced
the chimney  effect  in  the  bags.  This time was  equivalent  to a  null  period
which some  favor  as part of a  cleaning cycle to  allow dust to  fall.   It is
very likely that some dust  fell out  of the bags during  the  cool-down period.
It is possible that the predicted accumulation of  fines in the top of the bag
fell out during  this period.

     The actual  inspection of  the  bags  necessitated  some  agitation of  the
bottoms of the bags  by  the  people  moving about on the tubesheet.   Sampling the
cake on the inside of  the  bags by cutting swatches with the bags in place or
removing  the  bags  and   cutting  them  open  would   also  dislodge  and/or
redistribute the  dust cake.

     The conclusion  that  these  experiments prove that there is  no accumulation
of fines in  the  tops of bottom inlet  bags is not warranted.  These experiments
are inconclusive;  they  do not prove  that there is  no  accumulation of fines at
the top of bottom inlet bags.

     The value  of  k'  at   the  top of  the  bag  may  be higher  if the  median
particle size is  smaller.   The path  of least resistance for gas flow will then


                                     436

-------
be through another part  of  the  bag where k',  and therefore flow resistance, is
lower.   In  other words,  the  face velocity,  or  cloth ratio,  is  variable over
the length of  the bag.

     None  of the conventional  cleaning methods  is effective  in  removing the
fines from the top  of  the  bag.  During  cleaning,  some of the fines  stick to
the top  of the bag,  some agglomerate and fall to the hopper, and some float in
the motionless gas until the  bag goes on line again, at which time they return
to the bag.

     Shaking the  top of the bag will  cause  some agglomeration, but some fines
will still remain to go  back  on the bag.   One or more deflation flows during a
shake/deflate  cleaning  cycle will  help  to move  the  dislodged material to the
hopper.  However, deflation flow will take the path of least resistance, which
is in the  lower part of  the  bag, and therefore  is not  effective for removing
fines from the top of the bag.

     In  some baghouses,  reverse air  cleaning  has  not been  effective either.
It, too, takes the path  of  least resistance through the lower part of the bag,
bypassing the  top of the bag.

     The dead end  at  the  top  of  the  bag  is  thus seen  to be  a  cause  of
ineffective  cleaning that  entails  pressure   loss,  increased  frequency  of
cleaning,  and reduced bag  life.  There  is  no way to avoid  this  because the
submicron  material  will  not  settle by  gravity.   A  half-micron  size  particle
of fly ash  in still  air will reach  its  terminal velocity in  a  fraction of a
second.  At  that  velocity,  it will take weeks for  it to  settle  by gravity 30
feet from  the top of the bag to the hopper.  It actually will not settle any
distance because its fall will  be opposed  by the  convection  currents,  or
chimney  effect, in the bag.

     We  can  assume  that  the  accumulation of fines in the dead end will cause
some cloth area at  the  top of  the bag  to  go blind.  Loss of  filter  area for
this reason would  necessitate  a higher  cloth  ratio  and therefore  a higher
pressure loss  in the  balance  of  the  bag  if  nothing  else changed.   What
actually happens  can be  inferred from the  curve shown in Figure  4,  which is
typical  for  bottom-inlet bags.

     If  we assume reasonable values  for kl  and  C.  and if  we take values of AP
and  t from  the curve,  we can  solve Equation (3]T  for V,  cloth  ratio.   If we
then plot V versus  the same  time span, we have  the  relationship  shown in
Figure 5.  Although  k*  and  C. are also likely to vary with time, we can assume
that k'  will probably increase  as  the size  of  the  particles entering the bags
becomes  smaller and  that its increase  will be  offset,  to  some extent,  by a
decrease in  C..   It  is probably  safe  to  assume  that the shape of the curve in
Figure  k is caused  mostly by  changing  cloth ratio  and  that  Figure 5  is a
reasonable representation of  the variation of cloth ratio with time on line.

     Figure  5 confirms  the well-known fact  that the gas flow rate through a
bottom  inlet  compartment decreases while on line.  The total flow through a
baghouse is  therefore the sum of  the  variable  flows through the compartments.
This can be  illustrated  by  the  electrical analogy of Figure 6.


                                     437

-------
k'-|V -I-
          Figure 4. Pressure loss versus time on line for bottom
                   inlet baghouses.
            Figure 5.  Cloth ratio versus time on line for bottom
                     inlet baghouses.
                                 438

-------
          Q = TOTAL GAS FLOW

          q = COMPARTMENT GAS FLOW

          d = DAMPER RESISTANCE

          b = CLOTH AND CAKE RESISTANCE

          AP = SYSTEM PRESSURE LOSS
                                             AP-
                           Figure 6.  Electrical analogy.
     In  this diagram,  Q is  the sum  of all  the compartment  flows, q,  which
varies depending  on  how long each compartment has  been on line since its last
cleaning.   The system  pressure  loss,  AP,  will vary with  Q, the  number  of
compartments  in  service, and during  the cleaning cycle.   However, at any  one
moment,  the  sum  of the  resistances  of the two  dampers and the cloth and cake
for any  compartment  will be  the same as the system AP.

     The pressure loss  across  the cloth  and cake,  b, will vary as shown  by
Figure 4.   As b  increases,  q   decreases  and,  as  q  decreases,   so  does  the
resistance  through  the two dampers,  d.   Baghouses  are  said  to  be  "self-
leveling"  in the  sense that the gas  seeks the  compartment  with the  lowest
pressure loss.  This may explain the  initial steepness of the  curve  in  Figure
4 when the bags are  newly  cleaned.   Note that "self-leveling"  for bottom inlet
baghouses is  on a compartment basis.

     Figure  5 also illustrates  the well established fact that  cloth  ratio  for
bottom  inlet baghouses  diminishes with the time  on  line.   The   design  cloth
ratio may be the  average  cloth ratio  of  an on-line  compartment   for the time
span between cleanings.  This  average may be different for each compartment.
Variable cloth ratio over  the length of each bag and  in every  compartment is a
consequence  of this  undesirable inherent  characteristic  of  the   bottom  inlet
design.   Variable cloth ratio  is undesirable because  it  means that  the  cloth
is not  filtering  at a constant  rate.   For bottom inlet baghouses, cloth ratio
can be   thought of  as the baghouse  size;  it  is not  the  face velocity  except
momentarily  . at  various  places  in  some  bags   in  one  compartment  when  the
baghouse is  in service.

     Remember, cloth ratio  is  the  amount of gas to be cleaned divided  by  the
area of  the  cloth used  to  clean it, or:
         CFM   ft
         ft
              min
                   = face velocity.
                                     439

-------
     If  V diminishes  between^ cleanings  as shown  by Figure  5,  it could  be
because  CFM  diminishes or  ft   increases.  We  know that  cloth area does  not
increase  and  therefore CFM must decrease.  However, if  the bag goes blind  at
the  top,  we  are reducing  ft  which would  make V  increase.   If the bag  goes
blind at  the top and if V decreases, we conclude that CFM  diminishes at a  rate
that compensates for  the  reduction in filtering cloth area at the  top of  the
bag.

     We  thus  have  a vicious circle.  As  the  compartment  flow  diminishes,  the
velocity  entering the  bag diminishes and  the  proportion of fines  entering  the
bag  increases.   This  accelerates  the blinding at the  top of  the bag which
accelerates  the  reduction  in  the  rate  of flow  which  reduces  the velocity
entering  the bag, and so on.  This may explain why  the curve in Figure  4 shows
the rate of AP increase slowing down as the compartment  nears  the  time  for  its
next  cleaning.   This vicious   circle  is  unavoidable  with  bottom  inlet
baghouses.

     Top   inlet   baghouses  do   not  have  the   two   undesirable   inherent
characteristics of  bottom inlet  baghouses.  There is  no hopper  fallout  and
there is  no  dead end  in the bag.   As a  result,  the relationship of  pressure
loss versus  time on  line  is as  shown by  the plots  of  data  from top  inlet
baghouses  in service  in  Figure  7.   The  data sources  are described  in  the
references.
               AP,
              IN. WG
                    6-
                    5-
                    4-
                    3-
                    2-
                    1-
                                   -REFERENCE 1
                                    ^REFERENCE 2
                           10
                                                   50
                                                         60
                                 20     30     40
                                    t. MINUTES
                   Figure 7. Pressure loss versus time on line for top
                          inlet baghouses •
                                      440

-------
    Substituting  values from  Figure 7  in Equation  (3)  and  solving for  the
corresponding  cloth ratio,  we have  the relationship  for  cloth ratio  versus
time on line as shown  in Figure 8.

    Figure  8  shows  that  top  inlet  baghouses  probably  operate  at  constant
cloth  ratio.   If  so,  this  would mean  that every  bag and  every  compartment
would  have the same face velocity regardless  of how long it has been on line.
For this  to be so, we have  to  infer several things.   First,  k' does  not  vary
over  the  length  of the  bag.   Second,  the rate  of  gas  flow is the  same  in all
compartments all  the  time  on line.   Third,  top inlet baghouses may  be  self-
leveling  on a  per bag basis  rather  than on a  compartment basis  as  is the  case
for bottom inlet baghouses.
                  Figure 8. Cloth ratio versus time on line for top inlet
                         baghouses.
     These inferences  should  be checked  experimentally.   The bases  for  them
 are as follows:

     For  k'  to  be  constant,  the  permeability,  or  porosity,  of the  filter
 cake must Be  the  same in all  bags,  over  the  length of  each bag, and  in all
 compartments -even though  the  cake  gets thicker  with time  on line.  This is
 possible within  limits.  Contrary to  some  reports in the literature,  there is
 no hopper fallout with top  inlet.   As the vertically downward velocity  in the
 bag diminishes,  we have a  downward velocity component  approaching zero and a
 constant  horizontal  velocity  component  (face  velocity)  with  a  resultant
 component approaching the  horizontal.  In  other  words,  the particles  move
 toward  the  bag  and  reach the  bag before  they fall out  of the bottom  of the
 bag.
                                      441

-------
     With no  hopper fallout,  the  cake will have  the full  range of  particle
sizes and the median size will be larger.  With coarse  material in the  cake,
it will be more porous, k' will be smaller, and pressure loss will be  smaller.
For  a typical  grain loading  leaving  a coal-fired boiler and  for  a  typical
cleaning cycle,  the filter cake will usually be only  two  or three hundredths
of an inch thick.  With a porous cake and a small pressure differential across
it,  there may be a  negligible  increase  in k' as  the  cake builds from  say  0.01
inch  to 0.02  inch.  In  addition,  there is  very little  likelihood  of  cake
collapse,  which would  make  it  less  porous.   The  first  inference  is  thus
plausible.
     The second and third inferences are related.  For the flow to be  the  same
in all  compartments  all the time, the resistance to  flow must be the same  in
all  compartments  all the  time.   This is possible  if k',  C , and  V  are all
constant,  which  may be  the  case  for top inlet  bags.   For  this  to  be so,  we
have to assume that the cake in every bag is of uniform thickness, composition,
or  porosity  over  the  full length  of  the  bag.   This   is  possible  if  cake
deposition in the bag is uniform  as a  consequence of  the gas seeking  the  path
of least resistance in each bag.

     As dirty gas enters the top  of  a clean bag, it  will deposit cake at the
top.  The  path of least  resistance will  then be  below that  cake.  As  the  cake
builds,  the  path  of  least resistance  moves  down  the  bag.   This  process
probably happens very quickly over the length of the bag.  The result  may  be a
cake of uniform porosity for the  full length of the  bag.  Any deviation  from
uniformity may be immediately corrected  as  the gas  seeks the easiest  way  out.
The self-leveling process may thus be on a per bag  basis for top inlet.   This
could be  a very  significant difference  from the  per compartment  basis for
bottom inlet.

     There  is a  possibility that  top  inlet bags  perform  differently  from
bottom  inlet  bags.   Figure 9  may  be evidence   for   this  difference   in
performance.   Both   curves   are   for  baghouses  in  service  on  cement   kiln
exhausts.   The top  inlet baghouse has a nominal cloth ratio of 3 ft/min and
the  bottom inlet baghouse has  a nominal   cloth   ratio  of  2 ft/min.  These
pressure losses were measured across  cloth and cake in one compartment.
                                                                  2.6
                                    TOP INLET
                                                                  7.4
                                   BOTTOM INLET
             IOAM
   O5
IPER OPEN
                           1.0
1.5      2.0
 SECONDS
                                                   2.5
                                                           3.0      3.5
                  Figure 9. Initial Pressure Loss Versus Time (Reference 3)
                                      442

-------
     Note the abrupt  rise  in pressure  loss  for the bottom inlet compartment
and  the  gradual rise  for  the  top  inlet  compartment.   Note,  too,  that the
pressure loss  after the  first three  seconds  on  line  for the  bottom  inlet
compartment  is  about  three  times  the loss  for  the  top  inlet  compartment.
During this  test, the  pressure  loss for the bottom  inlet compartment rose to
8.6 to 9.0 in.  wg when it was on line and one of the other compartments was in
the cleaning"mode.   The pressure loss in the top inlet compartment rose to 3.1
to 4.5 in. wg when  it was on line and one of the other compartments was in the
cleaning mode.

     The bottom  inlet  curve may  be a  confirmation  of  the  theory  that  fines
floating in  the top of a bottom inlet  bag jump back on  the  bag as  soon as it
goes back on line after  cleaning.  The top inlet  curve  may be a confirmation
that the filter cake formation in a top inlet bag is at a uniform rate.

     The most significant  difference between top and  bottom inlet  designs is
the fact that the bottom inlet  design has hopper  fallout and  dead  ended bags
and  that the  top inlet  design has  neither.   Top inlet  has other  advantages
that are certainly  worth consideration.

     Some evidence  exists  that top  inlet  baghouses  will operate at  a higher
cloth ratio  for the same pressure loss  as  bottom inlet baghouses or that they
will operate at a lower pressure loss for the same cloth  ratio as bottom  inlet
baghouses.   This is illustrated by Figure 10.

     Figure  10 shows that  top  inlet provides a  saving in capital cost at the
same pressure loss  or a saving  in operating cost (AP)  at the same cloth  ratio
when compared with  bottom inlet.
                                                 TOP INLET.
               Figure 10. Comparison of top versus bottom inlet baghouses.
                                      443

-------
     Top  inlet  bags  can be longer than bottom inlet bags because  they do not
have a dead end to go blind.  Making bottom  inlet bags  longer accelerates the
rate  of  blinding  at  the top  and aggravates  the  vicious  circle  described
earlier.   Contrary  to  common belief in the  field  of fabric  filtration,  high
entrance  velocity at the bag inlet is  not  necessarily the cause of bag failure
at  the  inlet  end.   Failure of bottom inlet  bags near  the bottom  at  low (240
fpm) inlet velocity  is more likely caused by inadequate bag  tension.   On the
other hand,  top inlet  bags are in  service on cement kiln  exhausts  that have
inlet velocities  of  approximately 680 fpm that do  not fail  from  abrasion at
the inlet.

     One  advantage of  longer  bags is  the  lower cost per square  foot  of cloth
and  lower installation  labor  cost per square  foot of  cloth.  In  addition,
longer bags mean fewer bags and less plan  area.

     If  the baghouse is to be  used downstream of  a spray dryer,  the hopper
fallout  in a bottom inlet design will contain  some unreacted  material that
would remove  additional  sulfur  dioxide  if it got into  the cake on  the bags.
With a top inlet design, this material would  not  be  wasted.

To  sum up:

Bottom inlet baghouses

o    Suffer unnecessary pressure loss  because particle  size entering  the bags
     is reduced in several ways,  and smaller particles  entail higher pressure
     loss.

o    Operate at variable cloth ratio within each  bag and among compartments.

o    Are  difficult to clean because of  the accumulation of fines  in  the dead
     end  top of the bag.

Top inlet baghouses

o    Have lower pressure loss because the  filter cake  contains the full range
     of particle sizes.

o    May  operate  at  a  constant   cloth ratio  within  each  bag  and   in  all
     compartments  all the time.

o    Clean more readily  because  the  cake includes  the coarse  particles and
     because the reverse flow does not bypass the top of the bag.

o    Can operate at higher cloth ratios.

o    Can use longer bags.

o    Improve utilization  of  reagent and  fly ash alkalinity  in dry  flue gas
     desulfurization systems.
                                      444

-------
The work described  in this  paper  was not  funded by  the U.S.  Environmental
Protection  Agency and  therefore  the contents  do  not necessarily reflect  the
views of  the Agency  and no  official endorsement should  be  inferred.
                               ACKNOWLEDGEMENTS
     In addition to the persons  listed  in  the references, the author  is  most
grateful to John Musgrove, William Lane, Ernest Purcell,  and  Betty  Winkelman,
all of the Bechtel Power Corporation, for their assistance in various  ways  in
the preparation  of this  paper.   Their assistance  does  not  imply  that  they
agree or disagree with this paper.
                                  REFERENCES
1.   Estopinal,  E.,  and Noone,  G.,  of Columbian Chemical Co.   Data from carbon
     black baghouse in  Mohave,  California.  Top  inlet  bags are  10-in.  diam-
     eter, 31  ft 6 in.  long.   Private communication.

2.   Graham, H., of  Giant Portlandx& Masonry Cement Co.   Data from cement kiln
     exhaust  baghouse  in Harleyville,  South  Carolina.   Top  inlet  bags  are
     8-in. diameter, 62 ft 10 in.  long.   Private communication.

3.   Brumagin,  C., of Fuller Co. and Foster, R.,  of  Whitehall Cement Manufac-
     turing  Co.   Data  from  cement  kiln  exhaust  baghouses  in  Cementon,
     Pennsylvania.  Top  inlet bags  are 11.5-in.  diameter,  37  ft  10  in.  long.
     Private  communication.
                                      445

-------
        UPGRADE OP PLY ASH COLLECTION  CAPABILITY AT THE CROMBY STATION
                  by: T.J. Ingram
                      R.J. Biese
                      Gilbert/Commonwealth
                      Reading, Pennsylvania

                      R.O. Jacob
                      Philadelphia Electric Company
                      Philadelphia, Pennsylvania
                                  ABSTRACT

      A description is given of measures to upgrade the dust  collection cap-
ability of 28-year old Cromby  Unit  1  in the interim period before  installa-
tion of  an S02  scrubber.  A novel  approach was  taken to  enhance  the  per-
formance of outdated equipment, resulting in continued usage  of  existing me-
chanical collectors,  upgraded  electrostatic precipitators  and  the  addition
of a sidestream  baghouse  to  accomplish  the objective.  The equipment  was in-
stalled  and  started up within a year  of writing  specifications,  and  com-
pliance  was  attained.    The  paper  discusses  some  problems  experienced,
notably  bag  blinding which  has resulted  in a  severe maintenance  problem;
steps in progress to resolve the problem are discussed.

                                INTRODUCTION

      In August  1979,  Philadelphia  Electric Company signed a consent agree-
ment with  the Environmental Protection  Agency  and the Pennsylvania  Depart-
ment of  Environmental Resources to bring Unit 1  at  their  Cromby Station into
compliance with  particulate  emission  limitations of 0.65 Ib  per million Btu
input  pending   installation  of  a  wet  particulate  scrubber  and  an  SC>2
scrubber.   Gilbert/Commonwealth (G/C) was consulted to make  recommendations
for achieving this goal and to do the engineering.

      This  28-year old unit  utilized mechanical  dust collectors  and elec-
trostatic  precipitators  in series to remove dust  from the flue gas, and  it
was  found  that emissions would have  to be reduced by approximately  50 per-
cent to  attain the limitation  of  0.65  Ib/MMBtu. Although this  limit is not
strict by  today's  new source standards, the old  and worn equipment,  typical
of  such units,  was  not  judged  capable of  achieving  this  performance with
simple  reconditioning.    On the  other  hand, extensive  replacement  was not
warranted  because  the  scrubber installation would  be complete within four
years.
                                      446

-------
      The  program instituted was  one  of compromises, which  was necessitated
by a schedule calling  for  compliance by April  1981,  and by  the  usual  space
limitations  in an  older plant.  By the  time  initial tests  could be  run  to
establish  base performance, there  were  only sixteen months  remaining  to the
compliance  date.  To meet  it,  a  gas  flow model  study  was  conducted,  the
existing  equipment  was  reconditioned and  modernized,  and  a  small  baghouse
was added on  a side  stream.

      Overall  success of the  program was  demonstrated by  achieving compli-
ance.  However, there  has  been  a  problem with  early blinding and shortened
bag life. A  number of  measures  have  been  taken  and  the  solution  of  this
problem is well  on the way  to  resolution.

                             PROBLEM DEFINITION

HISTORY

      Cromby  Unit 1 went  into  service  in  1954.   It  is a  165 megawatt  (MW)
unit,  designed to  burn eastern pulverized coal  in  a  front-fired,  radiant,
Babcock & Wilcox boiler.   The  original  design  was for  pressurized operation
and omitted  induced draft  fans.   However,  before construction was  complete
it was recognized  that  there were  operating circumstances  where a  negative
pressure  in  the  boiler would  be  required  for periods  of  time  and reduced
size induced  draft  fans were  included.   Casing leaks  eventually forced con-
tinuous use  of these fans and reduced power  output.   However, full size fans
have recently been installed in conjuction  with the  installation of the S02
scrubber.

      This  unit was one of only  three   in  the  Philadelphia  Electric  system
that operated on coal when  this project was  begun  in 1979.   It  was,  there-
fore,  considered economical to  maintain it  in  service to  take  advantage  of
the cheaper  fuel,  and  to  find means  to comply  with  imposed environmental
regulations.

INTERIM CONSENT DECREE

      As  a result of  discussions with  the  regulating authorities a consent
order  was issued  which  required Cromby  1  to burn lower  sulfur  coal  and  to
reduce particulate emissions during the interim period to  a  maximum of  0.65
ib/MM/Btu  input  while the  particulate   and  862 scrubbing  system was  being
installed.

      A series of tests performed in October  1979  showed  average particulate
emissions  of   1.08 Ib/MM/Btu  input  which was well  in  excess of  the  estab-
lished limit.

      An  upgrade  schedule  for  achieving compliance  by  April  1981  was  sub-
mitted in accordance with  the consent order.

                                THE SITUATION

EXISTING  EQUIPMENT DESCRIPTION

      The  flue gas cleanup  equipment that existed at  the  plant at the begin-
                                     447

-------
ning  of  the modifications  was  unique by  today's  standards.  In  each  of two
parallel  gas streams,  between  two  tubular  air heater  stages,   there  was  a
mechanical  collector  (Figure 1)  to  remove most of  the ash.  The mechanical
collector  was  the horizontal multi-tubular  type.  The  flue gas  entered the
tubes  through vanes  which imparted a spin. The ash, being  heavy, was thrown
toward  the wall"  and  exited along with  some gas  through  an annulus  in the
tube  about half  way  through. It then dropped into a collecting  duct and was
drawn  off through two  sets of  cyclones  in  series by  a conveying  fan. The
conveying  gas  was reintroduced  to  the main gas  stream  just upstream  of  a
precipitator as  shown in  Figure 2.

       The  ash transport  gas from the mechanical collectors  was approximately
eight  percent  of total  gas  flow.   Ash   removed  from  this   stream  by the
cyclones  dropped into a  dust  silo  through  rotary valves.  Ash  collected  in
the  precipitator hoppers was picked  up  by a stream  of flue  gas drawn  from
the  precipitator inlet  plenum and carried to  small cyclones  by a  secondary
conveying  fan.  The clean secondary  stream was  then combined  with  the  clean
primary ash  collection stream before emptying into the precipitator  inlet.

       The  precipitator,  or  actually  two  precipitators in parallel,  were old
units, each  having 31 gas passages 8-3/4  inches wide,  and  two fields of  col-
lecting  plates  20 feet  tall by  9  feet  long.   The  collecting  plates   were
constructed  of  double  sheets  of expanded metal,  and  were rapped  from the
Figure 1.  Mechanical dust collector.
                                     448

-------
                                  MECHANICAL
                                   COLLECTOR
            CYCLONE
          COLLECTORS
 ID FANS
             DUST
             SILO
                              AIR
                            HEATER
            ELECTROSTATIC
             PRECIPITA
  AIR
HEATER
        FROM

     ECONOMIZER
                                       GRADE
Figure 2.  Schematic - fly ash collection original installation.
                             449

-------
side  by magnetic .impulse rappers.  The  discharge  electrodes were  weighted
wire  type, using air  vibrators  for rapping.   Power was supplied  from trans-
formers  through  mechanical rectifiers.   Single  perforated  plates  were  in-
stalled  at both the  inlet  and outlet  to  obtain  satisfactory  gas  distri-
bution.

POTENTIAL FOR PERFORMANCE IMPROVEMENT

Condition of Existing Equipment

      At  the  outset,  the  existing equipment was  examined  to  determine  the
potential  for upgrade.  The mechanical  collectors were  found  to  be  badly
worn:  several  of the  spinners  showed advanced wear, and there  were  holes in
a  number  of   the  tubes.    The precipitator  had  several  wires broken  or
missing,  and  a heavy  ash buildup on those remaining.  The  collecting plates
were  heavily  coated with ash.  The perforated plates front and  rear  had some
buildup,  though  were  not plugged.   The  lower stabilizing frame  was subject
to  sway. There were  many ash  deposits  in the  inlet  and outlet indicating
problems  with gas  distribution, and there were several leaks in  the casing.
The   largest   primary  cyclones  had  inspection  ports,   and  when  examined
appeared   to  be in   satisfactory  condition.  The  smaller  ones, having  no
inspection ports, were not examined.

Improvement Requirements

      The  emission limitation  of  0.65  pounds  of particulate  per  million
Btu's required about  90  percent overall  removal efficiency of  the dust con-
trol   equipment.   The  results   of  the  initial   tests   indicated   overall
efficiency of just under 88 percent,  so  the required improvement,  at  least
on  the  surface,  did  not appear  to  be  too  difficult  to   obtain.  Average
efficiency of the  mechanical collectors was  79  percent  and  of  the  precipi-
tators  57 percent.   The  question  then was,  could  the existing equipment be
suitably  modified  or overhauled;  or should  it  be replaced  with new equip-
ment, in view of  the age of the equipment and the required  performance in-
crease?  Although  only a nominal  overall improvement  was  actually  required,
in  order to  be  sure  that all  tests would show  compliance  and  in  order to
maintain  a desired low  particulate  carryover for the future scrubber down-
stream, a somewhat higher target for overall  efficiency was  set.  In fact, it
was   decided  to try  to  upgrade  the existing  equipment  to  get  the  best
possible  performance  improvement.

      There were three  choices  for  the  mechanical collector. One,  it  could
be  completely overhauled with  new tubes and  low pressure drop  spinners to
minimize  wear. Two,  it  could  be  completely  overhauled  as  originally fur-
nished  with  high pressure drop spinners  to  maximize  performance.  Or three,
it could be gutted and not used.

      The  precipitator  presented  another  problem. It suffered  from a very
low specific  collection  area,  about 89 square feet per thousand  ACFM,  and a
very  high  velocity of  about 9.2  feet  per second. In addition,  the aspect
ratio was  only 0.9.  The deposits of  ash observed in the inlet  and outlet
during  the initial inspection,  and a brief examination of  the  ductwork con-
                                     450

-------
figuration,   indicated  that  a  gas  flow  model  study  should  be  made  to
determine  changes  to  improve  the  velocity  distribution  through  the  pre-
cipitator.  It was also obvious from  the type of power  supplies  and controls
that  much  improvement could be made with modern  electrical  gear. Finally, it
appeared  that the  rappers were  not  doing  an adequate  job of  cleaning the
electrodes.  The  wires that remained in  the unit  were  in many cases  kinked or
bent  so that clearances  to  the  plates  were considerably  decreased. The con-
dition  of  the  precipitator casing  was  also  questionable  because  of  the
number  of  holes  that  were observed during inspection.

Alternatives  for  Improvement

     There  was  a valid  question  as  to  whether  it  might be  more  economical
to install  either new additional  precipitator capacity  or a baghouse, rather
than  upgrade  the old  equipment.  There were also  several other considerations
in determining  an alternative.  The  station suffered  from  a  severe  space
limitation,  typical  of many backfits.  Upon  looking  at  a  photograph  of the
plant before these modifications  were  made  (Figure  3),  it  is  obvious that
there was  no space available to  put  a  longer precipitator  between  the duct-
work  coming  from the air  heater  and  the  ID fan  inlet.   There was  also  a
concrete   deck   extending part  way  over  the   existing  precipitator  which
Figure 3.   Cromby Station  before modification.
                                      451

-------
effectively blocked installation of equipment above.  Space  upriver from the
precipitator was  obstructed by the No.  2 Unit and  the area  downriver  from
the precipitator was  limited due to space  allotment  for  the  new scrubber, as
well  as  to  an existing  substation.  In  addition to  the  space  limitations,
there  were severe  schedule restrictions.  The  consent  agreement  had  sfipu-
lated  that compliance must be  achieved by April 1981, which  left  only about
16 months  to make any improvements,  to test to make  sure  those improvements
were  in  fact  doing the  job intended  and, if not,  to make  further improve-
ments.   There were  also,  of  course, requirements  for  a minimum duration
outage to  make the modifications because  of  the  cost of  replacement  gener-
ation.

                             SELECTED ALTERNATIVE

       As  a  result  of all  these considerations,  G/C  recommended  that  the
mechanical collector  be  refurbished, the precipitator upgraded,  and a pulse-
jet  baghouse added  to polish the dust  conveying gas  stream from the mechani-
cal  collectors (Figure 4).  This option required replacement  of the conveying
fan  to overcome  the  additional  pressure  drop  imposed  by the  baghouse.  G/C
engineers  were confident that  the  difficult  schedule could  be met.  Looking
back from  the  compliance date  of April 1,  1981, testing the  unit was planned
for  the  end of February, which called  for start-up around January 1.   Since
it was already mid-December 1979, only a little over  a  year  remained to com-
plete the  project.

       Several  specific  measures  were  considered  for improving precipitator
performance. Power  could be maximized by  replacing  all wires  with a later
design and using heavier  weights,  replacing the old transformer-rectifiers
with modern types  having solid-state controls,  and installing  stabilizing
insulators for the  lower guide frame.  Rapping could  be  improved by  replacing
all   rappers with new magnetic  impulse  type  having  solid-state  controls.
Effective  gas  velocity  could be  improved  by optimizing  distribution  and
relocating the dust conveying  gas  duct discharge downstream of the precipi-
tator. This would require  installation of a baghouse  to  remove dust passed
by   the  cyclones.   It  would  have  the  effect of  increasing  the  SCA  and
decreasing the velocity through  the  precipitator  by  about  eight percent.
The   latter  was  felt to  be  critical  because  the scalping  velocity  was
probably being exceeded.

                               PROGRAM EXECUTION

SCHEDULE

      The  planned schedule is  shown in Figure  5.  After  the  base performance
test results were analyzed in mid-December 1979, the  specification was writ-
ten  for model  testing.  It was decided that these  test results were  needed
before the precipitator  contract  could  be awarded,  so a date of  March 9,
1980 was requested  for a preliminary report. In  order   to assure  completion
by the end of  the year,   it was  necessary to award  contracts  for all  major
equipment   by April  1. Although  the  initial stages  occurred  about a  week
late, the contracts  were  awarded on time.  Erection was  accomplished during a
                                     452

-------
                                   MECHANICAL
                                   COLLECTOR
             CYCLONE
           COLLECTORS
                  \
 ID FANS
     X
             DUST
             SILO
ELECTROSTATIC
 PRECIPITATOR
                               AIR
                             HEATER
                                     AIR
                                   HEATER
   FROM

ECONOMIZER
Figure 4.  Schematic - fly ash collection modified installation.
                              453

-------
 five-week outage, followed  by  checkout  and test prior to  start-up  at the end
 of the year.

 MODEL STUDY

       The first task was  the model  study  which was necessary  to  determine if
 modifications  would be   required  in  the gas  flow  path.  Accordingly,  the
 specification was issued  on December 24,  1979  and  bids  were  due  January 11,
 1980..  The  study  was subsequently awarded to NELS, Inc.,  of St.  Catharines,
 Ontario.

       Surprisingly,  the velocity distribution  which  had been  anticipated to
 be very  bad turned  out   to  be quite  good,  with  RMS deviation  of  about 12
 percent. Although it appeared  improvement might still be  made,  NELS1 experi-
 ments with  turning  vanes  did not indicate  sufficient improvement  to justify
 the installation.

 EQUIPMENT CONTRACTS

       With  the  short  bid time  allotted,  it  was  expected  that  it  would be
 difficult to  evaluate  all  the  bids and  award  the  contracts  on  schedule.
 However,  with the number  of bids received,  the work was  completed  on  time.
 Philadelphia Electric  internally expedited  its procurement process  and the
 required  approvals for  purchase were immediately prepared.
          1979
                                               1980
      NOV
              DEC
JAN
                               FEB
                 MAR
                                                               NOV
                                                 DEC
                    MODEL STUDY
                                                                      STARTUP
                         PRECIPITATOR UPGRADE
                             M ^•QggygQQQd

                              BAGHOUSE
     (BASE
     (PERFORMANCE
     iTEST RESULTS
       CONVEYING FANS
       • ••raaaaaaSMMM

       ASH REMOVAL SYSTEM
                                      SPECIFICATIONS
                                      BIDDING
                                      EVALUATION
                                    FABRICATION
                                    ERECTION
                                    PRELIMINARY REPORT
Figure 5.  Project schedule.
                                      454

-------
Precipitator

     The  precipitator upgrade  work was awarded  to Belco  Pollution Control
Company.  Their proposal  covered  the work  requested,  plus  some  innovative
ways  to increase sectionalization  and improve  rapping.

     Shortly  after Belco  began design, the  scrubber engineer  informed the
project engineers  that the new  ID fans  which were  to  be  installed to handle
the scrubber pressure  drop would  impose a vacuum  of about 20 inches of water
on the  precipitator.  Accordingly,  it was necessary  to  make an examination to
see if the negative  pressure could be  withstood  especially in view of the
poor condition of  the casing.  Belco made  this examination  and recommended
adding  stiffeners  on  the inside of  the  casing.   This was necessary since the
outside of the casing between the precipitators was not  accessible.   To in-
stall the stiffeners, the  collecting plates next to the  walls  would be slid
or swung  out of the way far enough  to gain  access to the casing wall.  Belco
subsequently was  awarded an  addition to their  contract  for this work, which
was scheduled  to  take place concurrently with the other modifications.

Baghouse

      Enviro-Systems  &  Research  was  awarded the  baghouse.  They  proposed  a
four-module,   pulse-jet  unit  with  a  gross  air-to-cloth  ratio  of   3.86/1.
Design dust loading was 3  grains  per actual cubic  foot.  The bags were to be
nominal 5 inch by 10 foot  fiberglass  having an acid resistant finish, with  a
fabric weight  of  16 ounces per square yard, and a double-beam weave.

Other Equipment

      The new  single-inlet,  airfoil vane,  main conveying  fans  (one  operating
 and  one  spare) were  awarded  to Westinghouse-Sturtevant.   A vacuum  ash  con-
 veying system was awarded  to Allen-Sherman-Hoff.

DELIVERY, ERECTION AND START-UP

      Equipment delivery and installation  turned out  to  be a real challenge
for  both  the  project  expediters  and  the  fiel.d workers.    It is  a credit to
all  concerned   that there  was good  cooperation  among Philadelphia Electric's
engineering, production,  construction and  start-up personnel, G/C engineers,
United Engineers  (the construction manager) and  the  vendors.    There were
many long days worked  through the  1980 Christmas  holidays  in  order for the
equipment to be placed in  service  on schedule.

      Erection was made more difficult  by  the weather. The winter was one of
 the  area's colder  ones,  with several days  during  the erection and  start-up
 periods being  around 0°F.  Besides making  work arduous, moisture condensation
 inside the equipment  was a problem, making  start-up tedious.

      The unit went  back  into  service  on December  31,  1980. Because it had
 been a long cold winter and  the work had been done very  rapidly,  a  number  of
 problems  were  revealed   immediately  after  start-up.  There  was  excessive
                                      455

-------
sparking  in the precipitator  and there were  problems  with  stuck  butterfly
dampers  and malfunctioning controls in  the  baghouse.  There were also prob-
lems with the ash system,  primarily  because the weather was  below freezing
and  there was moisture in the ash lines which resulted in  plugging.   It was
found  that the sparking  in the precipitator was  caused  by bits of  welding
rod  and  other debris which  had been  dropped  inside  the precipitator and
which  lodged in the expanded metal collecting plates.  There  were  also some
warped collecting  plates  next to the outside walls which had  been  bent when
they were swung out  of the way during installation of  the casing stiffeners.
The  debris was easily removed;  however,  it  was  decided  that  the collecting
plates,   since  they  were  expanded  metal,  could  not   be  satisfactorily
straightened.   Therefore,  it  was  necessary  to  remove  a  number  of  wires in
the  affected gas passages  and  to  weld  steel strips to the perforated  plate
ahead  of those gas  passages  which  were  short  on wires in  order to restrict
the  amount  of  dirty gas  that  would  flow through  them.    Although  this was
undesirable because it increased the  velocity through the  remainder  of the
precipitator,  the project  staff felt it  was  necessary  in order  to  obtain
satisfactory performance.

       The sticking baghouse  inlet  and outlet  dampers were  trimmed in  order
to keep  them  from binding  and  much time was spent on  the  controls in  order
to secure proper cleaning sequences.   Although the baghouse is equipped with
hopper heaters, the ash  lines  plugged  repeatedly  because of  moisture and
improperly installed controls.

       Start-up  of  the baghouse was done  carefully. Since  it  was on  a side
stream,  its use was not necessary until the unit  was burning coal,  so flow
was  not established during light-off with No. 2 oil.  The  conveying  fan was
started  and the baghouse  dampers  were opened  after two  of  the four mills
were operating.  Meanwhile, the  baghouse hopper  heaters had been on for over
a  day.

                             COMPLIANCE TESTING

      Problems with  the equipment were corrected to the point that  the unit
could  be  tested  for  compliance.  This  was  done  early  in March  of  1981;
fortunately,  the  results  were  nearly as  anticipated.  The  unit averaged  a
little over  0.5  pounds of particulate  per  million Btu's  during  three  tests.
This worked out to a little over 93  percent overall efficiency.

                        UNIT PERFORMANCE  AND  PROBLEMS

      Concurrently with the compliance  tests,  the  unit  was also tested for
equipment  performance. These  results  are  compared  with the base data tests
in Figure 6.

      Although  the  unit  passed  the  required tests,  all  was  not completely
satisfactory.  As  shown in  Figure  6,  the  mechanical  collector  had actually
decreased  in  performance,  probably because  of  the lower loss  spinners, and
was  performing at  67  percent.  However,  the  precipitator, bent plates and
all, was  performing  at  up  to  85  percent  efficiency.   The  baghouse  was

-------
    100-1
  o
  s
  LU
  cc
     80-
  o
  g  60
  o
  LL
  LL
  LU
      40-
      20-
                87.8
             (>OSYSTEM<
             579.2
tot
               BASE
               LINE
               TEST
               1979
                               93.2
                            SYSTEM
                            66.
SOS
;LU
               SO?
                   85.3
                        98.6
              COMPLIANCE
                  TEST

                   1981
                                     93.4
                                 >SYSTEM
                                                74.1
O
LU:
                                                 o
                                                 UJ
                                                    73.8
                                                     .LU
                                            LU
                     RETEST


                       1982
Figure 6.  Performance test results.
                                4S7

-------
reaching  98.6 percent efficiency,  but  was experiencing early  high pressure
drop  and  it was  difficult  to  remove the cages and bags for  inspection. The
latter  problem was finally traced  to  an internal cuff in  the  bags combined
with  oversized  cage  end caps,  which prevented  cages  from  being extracted
from  the  bags. This delayed an investigation  of exactly what  was causing the
high  pressure drop because a  bag  could  not be removed with  the baghouse  in
service.  By the  time a  bag could  be removed  and sent  to  the laboratory for
investigation, the bags  were completely  blinded,  after  less than four months
use.  A further   investigation by  TexLab,  Inc.,  of  trial  bags  of different
material  indicated that  not only the original  bags but also  two Huyglas and
two Teflon  felt  bags were blinded with  fine  particulate  which  was  pene-
trating the  cloth to the  degree  that  it  could  not  be removed with normal
cleaning.  Tests  for  particle  size distribution were made  indicating a mean
particle  size of about  seven microns, by  Bahco  analysis. There  was also  a
significant quantity of  fines.    The  reason  for this  is  apparent  when one
examines  the system schematic in Figure  4. The baghouse receives the finest
particulate  because  the  gas has   previously  passed  through   cyclone  col-
lectors.   Cyclones, being most efficient on large particles,  leave the  small
particles to be  removed  by the baghouse.  Although  this was anticipated, the
number of  fines was larger  than  expected.    In  order   to  alleviate r.-he
problem,  it  was  decided  that a  partial  bypass around  the  large  cyclones
would be installed,  taking dust  directly from the mechanical  collectors  so
that  more of  the coarse  particulate would  be introduced to the baghouse,  at
the slight  expense of higher dust  loading.

      The  testing  program  had also  disclosed  that  the  secondary  cyclones
which  remove  the ash conveyed from  the  precipitator  hoppers  were  passing  a
large  quantity of  material and  in fact appeared to  be leaking.  Therefore,
when  a 12-week outage was  scheduled  for the  end of  1981  to  install  new  ID
fans,  a number of different tasks  were planned to  improve  the  interim modi-
fications.  The baghouse  had  to be  rebagged,  and  it  was  decided  that  trie
original  fabric  was still the most suitable choice  for this, but  with a re-
vised  bag cuff design. Also, new cages having smaller diameter  end caps were
furnished by  the  vendor,  and the partial bypass around  the cyclones  was   in-
stalled. The  secondary cyclones were scheduled  for  refurbishment.   When  they
were dismantled,  a great deal of pluggage and  erosion  was found.

      Once  again start-up  took  place  around  the first of a  year,  in  late
January  1982. The baghouse  was  precoated using  pulverized  limestone  and
started using  similar procedures to the  previous year.  However,  this  time  a
program was  followed to  establish  the  best cake possible  in order  to  keep
the fines from penetrating  the cloth.  The baghouse  was started  with on-line
cleaning, and  pulse air  pressure was set at  55 psig. Cleaning  was initiated
when overall pressure drop reached about 5 inches of water.

      Initially,  things  looked much better than they  did the  year  before,
and a  new test was performed the end of  February. Overall  performance re-
sults  were  very  good,   indicating  that  the  unit  was  emitting about  0.35
pounds  particulate  per  million Btu's  or again a bit over 93  percent effi-
ciency. However,  the  baghouse,  in  spite  of the work  done  on  it, again began
to  creep  up  in  pressure drop until  the  bags again blinded.   Although the
                                      458

-------
cyclone  bypass appeared to have increased the mean particle  size  to  the bag-
house,  it had apparently done nothing to decrease the  overall  high number of
fines.

     At the  time of  this  writing  the baghouse  has  recently  returned  to
service  (9/20/82) after a  one  week outage  during which new  woven fiberglass
bags were  installed,  Staclean  diffusers   were installed,  the  pulse  pipe
sequence was  changed to random and the  pulse air pressure  setting was raised
to 110  psig.

     Additional measures which may  be implemented,  if found  necessary, are
to enlarge the pulsing system and to use bags made  of a felted fabric.

                                 CONCLUSION

     The modification program was successful  in  achieving compliance with
the required  emission limit.   The blinding  problem,  although  troublesome,
appears  to be  solvable.   The  large  amount  of  work  that  was  performed  in a
short  period  of time demonstrates that  a cooperative effort  can accomplish a
great  deal.   The  fact  that the  unit  is  in compliance at  full  rated Load
points  to the real success of the project.

     The work described  in  this  paper was not funded by  the U.S. Environ-
mental  Protection  Agency  and therefore the  contents do not  necessarily re-
flect  the views  of the Agency and  no official endorsement  should be  infer-
red.
                                      459

-------
     HIGH SULFUR FUEL, FABRIC FILTER STARTUP EXPERIENCE

           by:  Phil Hanson
                Power Production Manager
                City Power and Light Department
                Independence, Missouri 64056

                Larry Adair
                Project Manager/Senior Engineer
                Phelps and Phillips
                Independence, Missouri 60455

                Robert N.  Roop
                Product Manager
                Research-Cottrell,  Inc.
                Somerville, New Jersey 08876

                Robert B.  Moyer
                Manager Fabric Filter Applications
                Research-Cottrell,  Inc.
                Somerville, New Jersey 08876
                          ABSTRACT

     The long term operation of fabric filters on low sul-
fur fuels has been, with a few exceptions, well demonstrated.
Similar experience on high sulfur coals has not yet been
gained.  As utilities continue to convert to coal, reliabil-
ity of fabric filters for high sulfur coal service will be
of paramount concern.

     The City of Independence, Missouri, recently put into
service a fabric filter designed for three to five percent
sulfur coal flue gases.  Details of the design will be
discussed.  Anticipating the opportunity to test in real
                             460

-------
world conditions, a bag evaluation program was initiated and
is in progress.  Physical and mechanical properties of the
bag fabric are being tested to determine the effects of high
sulfur service and a possible relationship to bag life.
Results of that bag test program, as obtained to date, and
its implications will be presented.
                       PLANT HISTORY

     The two unit Missouri City Power Station, now owned and
operated by the City of Independence, Missouri, was orig-
inally owned and operated by the N.W. Electric Power Cooper-
ative, Cameron, Missouri, a member of Associated Electric
Cooperative, Inc., Springfield, Missouri.  Two identical
generating units of 22 MW each were put into operation in
1954 at a site located adjacent to the Missouri River
approximately eight miles northeast of Independence.

     Foster Wheeler boilers each generated 220,000 Ibs/hr of
steam by oil and coal firing to drive Westinghouse turbine
generators.  Mechanical cyclones provided particulate col-
lection.  In addition to the boilers, turbine generators
and their auxiliary equipment, the site included coal and
oil unloading and storage facilities, ash pond, cooling
water intake structure, well water storage reservoir, rail-
road sidings, transmission lines and a substation for power
transformation and distribution.  The substation and trans-
mission lines remain the property of N.W. Electric Power
Cooperative, Inc.

     Air Pollution emission limitations entitled "Air
Quality Standards and Air Pollution Control Regulations for
the Kansas City Metropolitan Area" were adopted by the
Missouri Air Conservation Commission on January 5, 1969,
and were applicable to the Missouri City Plant.  The state
cited the Missouri City Plant for violation of visible
emission limitations on March 25, 1970.

     The N.W.E.P.C. was granted variance to burn coal by the
Air Conservation Commission of the Missouri Department of
Natural Resources (DNR), but soon afterwards the EPA issued
an administrative order requiring compliance with the regu-
lations by the installation of electrostatic precipitators.

     In early 1975, N.W.E.P.C. requested that the state
variance and the federal administrative order for ESP's

                             461

-------
 be withdrawn.  The request was  based on escalating coal
 prices.  While at the time, coal was costing $0.62 per
 million BTU and oil $1.80  per million BTU,  it was projected
 that coal would increase to $1.77 per million BTU by 1978.
 Considering the projected  rapid escalation  of coal cost,
 the cost of ESP's and the  limited use of the Missouri City
 Plant, N.W.E.P.C. wanted to revert  to oil firing to comply
 with the air pollution regulations.   The DNR and EPA amended
 their orders to allow for  such.

      In September,  1975, the Missouri City  station exper-
 ienced an electrical  fire  which destroyed the plant's high
 voltage switchgear  and much of  the  adjacent plumbing, wiring
 and mechanical systems.  The N.W.E.P.C.  decided  not to
 repair the damages  and closed the plant;  Associated Electric
 Cooperative had available  other generating  equipment better
 suited to its system.

      During the middle 70's  rapidly  escalating oil and
 natural gas prices  prompted  a need  for the  City  of Inde-
 pendence to look for additional base and intermediate load
 coal-generated  capacity.   The city was operating three coal-
 fired units at  its  Blue Valley  Station and  several oil and/
 or gas-fired combustion turbines throughout its  system.
 Phelps,  Hogland  & Phillips Engineering Company was hired  to
 study the  City's power needs  and various  possible power
 supplies.   These studies and much negotiating resulted in
 the  City's  purchase of the Missouri  City  Power Station in
 1979.   Engineering  for renovation of the  plant and the
 installation  of necessary particulate  collection equipment
 to permit  coal  firing was awarded to Phelps,  Hogland &
 Phillips.
                  FABRIC FILTER SELECTION

     Three major factors contributed to the selection of  a
fabric filter for controlling particulates from the Missouri
City Power Station:

UNKNOWN COAL SUPPLIES

     Engineering to provide quick reactivation of the plant
was begun as soon as it became apparent Independence would
become the new owner.  The contract to provide air pollution
control equipment was directly on the critical path, but
at that time the City had no long term coal supply with a
well defined quality.  The ability of the boilers to burn
a variety of fuels dictated that the air quality system also
have that capability.
                             462

-------
ALLOWANCE FOR FUTURE REQUIREMENTS TO REDUCE SULFUR EMISSIONS
  WITHOUT HARMING PARTICULATE COLLECTION

     The City's Blue Valley Power Station had been re-
stricted on allowable sulfur emissions.  A need to make
further restrictions to control ambient air quality in the
Kansas City area was a possibility.  Any reductions re-
quiring the burning of low sulfur coal would not inhibit
the collection capability of a fabric filter.

ON-LINE MAINTENANCE

     Since one of the reasons the Missouri City plant was
being renovated by the City was to replace oil and gas-fired
generation, outages that required utilizing combustion tur-
bines would be costly; continuity of service with on-line
maintenance of the air pollution control equipment was an
important criterion.  However, funds for renovation were
not unlimited.  The flexibility of a single multicompartment
baghouse to serve both units while providing the capability
of on-line maintenance was a good fit.

     The specification for the fabric filter was published
for bidding on September 4, 1979, with bids received
November 5, 1979.  The contract was awarded in December,
1979.  Construction began in January, 1981 and was completed
in December, 1981.  Start-up was in late March, 1982;
testing was conducted in July, 1982.
                    FABRIC FILTER DESIGN

     The contract was placed with Research-Cottrell  (R-C)
for complete system responsibility.  R-C work commenced at
the discharge of the existing I.D. fans atop the roof of
the boiler house, included dismantling of the tie-duct
between the fan and the old stub stacks, and terminated at
the discharge of two new booster I.D. fans into a new common
chimney.

     The scope encompassed design, engineering, fabrication
and erection of the inlet ductwork, including modulating
I.D. fan discharge dampers from the existing fans, bypass
system, fabric filter, discharge ductwork, booster I.D.
fans, ash handling equipment, insulation and the erection
of all supplied equipment.

     The gas cleaning equipment for the unit is a Research-
Cottrell reverse air fabric filter consisting of eight  (8)
modules, 4 deep x 2 wide.  Each module contains 336  filter
bags, 23' long x 8" diameter, which result  in a total
                             463

-------
filter area of 126,928 ft  (111,062 ft2 with one module
off).  Gas-to-cloth ratio with 8 modules on line is 1.78
ACFM/ft2 cloth and 2.27 ACFM/ft2 with 7 modules on line
including R/A flow.  Filter bags are constructed of 14 oz /
yd2 woven fiber glass cloth with an acid resistant finish
of polymer/silicone-graphite-teflon.  Table 1 shows the
fabric filter design data and Table 2 describes the filter
bags.


	TABLE 1.  FABRIC FILTER DATA	

                                            Design

   Flue Gas Volume (ACFM)                   226,000

   Temperature (  F)                           310

   Inlet Dust Loading (Gr/ACF)                4.64
   Gas-to-Cloth Ratio (Gross)                1.78:1
     Net (w/o Reverse Air)                  2.035:1
     Net (w/Reverse Air)                     2.27:1

   Outlet Loading (Gr/ACF)                   0.010

   Number of Compartments                     8
   Gross Cloth Area                        126,928

   Total Bags                               2688
               TABLE 2.   FABRIC FILTER BAGS

   TYPE:     Fiberglass (ECDE)  with Acid Resistant Finish
                                       2
   SPEC:     Weight - nominal 13.5 oz/yd
            Permeability - 40-55 CFM/sq. ft. @ 0.5" W.G.
            Count - 44 x 24
            Weave -3x1 twill
            Fiber Thickness - 0.015 in.
            Weave - Warp:  37 1/0(F)
                    Fill:  75 1/2(T)  + 75 1/0(F)

   DESIGN:   Bag Size - 8" dia. x 23'-0" (L/D = 34.5:1)
            Bottom Attachment - Snap Ring
            Top Design - Banded Top With Disposable Cap
            Rings - 4 carbon steel anti-collapse rings/bag
                             464

-------
     Cleaning of the modules is accomplished by a reverse
air system which pulls clean gas from the common outlet
duct and discharges it into an isolated module, collapsing
the filter bags and dislodging the collected particulate.
Reverse air (gas) is supplied by two (2) reverse air  (R/A)
fans (one operating, one standby), each rated at 26,000
ACFM requiring 125 HP motors.

     System booster fans are located downstream of the
fabric filter and are rated at 135,600 ACFM, 300 HP each.
Each fan's inlet louvre damper is positioned and operated
by a Bailey Control System.  The Bailey Control signal is
based on the static pressure at the outlet side of the
boiler I.D. fans.

     Hopper ash removal is accomplished by a Hydro Ash
System.  Ash is pneumatically conveyed from each hopper
in sequence, mixed with the vacuum-producing water and
gravity-flowed to a nearby ash pond.  The controls sequence
the compartment discharge feeder immediately after the
compartment cleaning cycle.

     The fabric filter is designed to treat 226,000 ACFM
of flue gas at 310°F.  The collection efficiency is to be
99.8%  (based on a minimum inlet loading of 4.64 GR/ACF)
with one (1) module out of service.  System pressure drop
is guaranteed to be 8 in. W.C. or less across the baghouse,
ductwork and dampers.
                    FABRIC FILTER STARTUP

PRE-COATING

     Since the plant's flyash is generated from burning
high sulfur fuel, it was Research-Cottrell's recommendation
that the filter bags be pre-coated to add protection for
the bag fabric.  An inert, moderate particle size diameter,
low sulfur ash was chosen for the pre-coat.  The selected
ash was trucked to the site and conveyed via 4" diameter
hoses to individual compartments using a blower on the
truck.

     The pre-coat operation was started with one booster
fan operating, all compartment inlet dampers closed, and the
baghouse bypass dampers closed.  Ash was injected to each
compartment through the hopper access door.  Initially
there was little increase in AP on module photohelic gauges
using this method.  It was obvious some ash was getting onto
the bags because the shutting of the outlet dampers caused
ash to fall back into the hoppers.  This method was continued
until the ash truck was empty.  There was an ash level in

                             465

-------
each hopper up to the door.  Continued operation of the
booster fan and air lancing of the ash in the hoppers re-
entrained it in the air flow to the bags.  The second
booster fan was then operated with minimal inlet damper
opening and 90% on the first fan damper.  One compartment
inlet damper at a time was opened (also one bypass damper)
and air lancing of the hoppers continued.  This operation
was maintained until at least three-quarters of the ash
was held on to the bags.

     After all hopper doors and inlet and outlet dampers
were closed, there was no evidence that a significant amount
of ash fell off the bags.  It took only a very short time
to pull the remaining ash from hoppers through the ash
system.  The maximum AP indicated on any module during this
pre-coat procedure was 1" W.C.

INITIAL BAGHOUSE OPERATION

     The following conditions were required for initial
baghouse operation:  one boiler operating at full load  (20MW),
for a minimum of one (1) hour at steady state conditions.

     Baghouse conditions prior to putting first module on
line are summarized below:

     o  One booster fan operating with damper 60% open
        (Note:   at this time fan damper operation was in
        manual and was not being controlled at desired
        1" W.C. boiler outlet pressure).

     o  Bypass damper open

     o  Outlet dampers open

     o  Inlet dampers closed

     o  Reverse air (R/A) dampers closed

     o  Reverse air (R/A) fans off

     o  Hopper heaters on (24 hrs.)

     o  Module AP = 0" W.C.

     The No. 1 module was placed on-line by opening the
inlet damper.    Indicated module AP increased to 0.5".
After approximately 1-1/2 hours of operation, no obvious
increases in AP were., noted.  At this time Module No.. 5 was
placed in service, a AP of 0.5" was observed.
                             466

-------
     After about one more hour of operation, Modules No. 2
and No. 6 were placed into service.  One bypass damper was
then closed.  The AP across in-service modules rose to 1".
Modules No. 3 and No. 7 were placed in service followed by
the remaining modules No. 4 and No. 8.

     The time taken to bring all modules in service and to
close the second bypass damper was five hours.

     After closing of the last bypass damper the increase
in module AP became more obvious.  Within one hour, the
maximum AP was up to 3".

     At this time, modules with the highest pressure drop
were nulled (outlet damper closed) one at a time and then
placed back on line.  The effect was significant.  When
each compartment was placed back on line its pressure drop
had been reduced to between 0.5" and 1".  The conclusion
was that most of the new filter cake could be removed
during a null without reverse flow and therefore the bags
were not blinding.  No reverse air was used during this
time.

     After approximately 32 hours of operation the baghouse
was placed back into the bypass mode and the modules were
taken out of service (inlet and outlets closed) in order to
begin startup of the second boiler.

     When the second boiler was up to full load and oper-
ating at stable conditions with no oil for one hour, the
baghouse was placed back into service for both boilers.
The normal startup method just described was employed with
minor exceptions.  All inlet and outlet dampers were opened.
Time for modules to stabilize was allowed and then both by-
pass dampers were closed.

     Indicated AP across the modules began increasing at a
faster rate than with just one boiler as would be expected.
After a short period when AP had increased to around 2",
nulling was tried to see what effect it would have.  Little
or no effect was observed.  At this time a reverse air fan
was started.

     Initial operation with the R/A fan, with manual control
cleaning of a module, showed when cleaning was started at
3.5" to 4" AP, cleaning for a few seconds would drop the AP
in the clean module by 1".  During this period great diffi-
culty in opening the reverse air dampers was encountered due
to high static pressure in the closed reverse air duct.
After overcoming the static pressure, R/A poppets would
surge open nearly half way.  Cleaning in this method re-
sulted in puffing even with long settling times.

                             467

-------
REPAIRS AND MODIFICATIONS

     Stack puffing was deduced to be caused by the in-rush
of gas into a module from the R/A system when a R/A poppet
opened.  The expansion of the gas in the system and lag
time before the fan automatically throttled-down would force
gas out the inlet duct into the other modules.  A stack puff
would then occur.  This puff was eliminated when the module
inlet damper was closed at the start of cleaning.  Although
satisfactory, a different solution was sought.

     To reduce the R/A duct static pressure and the initial,
fully opened damper-related surge, modifications were made
to the fan damper control system within the Bailey controls.
Originally one damper set point existed which controlled the
dampers to maintain a given fan motor current.  With all R/A
dampers closed (between cleaning periods), the dampers would
open fully in response to a control signal to increase fan
current.  This philosophy resulted in high static pressure
in the R/A duct,  and full volume condition when cleaning
started.  A device was backfit into the Bailey system to
supersede this setpoint until a R/A poppet was fully open.
The device drives the R/A fan damper fully closed.  When
the R/A poppet is fully open (limit switch contact made),
damper control seeks the motor amp set point and R/A volume
increases.

     Because the fan dampers are not totally gastight,
enough gas is still passed to develop a moderate static
pressure in the R/A duct.  A second modification was made
to bleed off this additional gas and keep the static pressure
low.  Three 4-inch pipes were installed between the R/A and
outlet ducts.  These allowed a small volume to bleed off
but did not affect the cleaning volume.  Experimentation
indicated only one pipe was necessary; caps were screwed
onto the other two.

TUNING

     Initial cleaning adjustments were made June 8-10, 1982
at 75% boiler load.  Final tuning was accomplished during
the week of July 12-16 at 100%.

     Full 100% load operation was observed at this time and
final adjustments were made.  Figure 1 shows operation of
the cleaning system with a 4.5" W.C. start-cleaning set
point, 60% opening of R/A inlet damper and approximately
45 seconds of R/A cleaning time per compartment.  Figure 2
shows the same 4.5" W.C. start point but with longer clean-
ing time  (180 seconds indicated) and R/A inlet damper 100%
open.
                             468

-------
     In figure lr it can be seen that the inverval between
cleaning cycles decreases with time and that the clean cloth
AP increases from 2.8 to 3.8.  After adjustment to the R/A
cleaning duration and flow, clean-down has consistantly held
between 3.0-3.5" AP and the duration between cleaning cycles
has remained the same.

PERFORMANCE

     On July 14, 1982,  the fabric filter system was tested
for compliance and performance.  Fuel analysis for the tests
and design values are noted in Table 3.  The inlet test ports
for AP readings were located just downstream of the boiler
#2 I.D. fan discharge after combining the flows from boilers
1&2.  The outlet pressure taps are located just beyond the
outlet flange of the baghouse prior to the inlet of the new
booster fan inlets.
                   TABLE 3.  FUEL ANALYSIS
                                                   Average
Constituents Range %
Sulfur
Ash
Moisture
Carbon
Hydrogen
Nitrogen
Oxygen
Heat Value
2.
5.
2.
57.
3.
•
5.
(BTU/#) 10,
0
0
0
6
5
7
5
- 4.
-19.
-17.
-70.
- 5.
- 1.
- 9.
500-13
1
0
0
5
0
8
0
,500
Design %
2
12
7
64
4
1
6
11,
.7
.8
.4
.6
.7
.4
.4
700
As Tested %
4
11
8
67
4
0
10
12,
.05
.88
.75
.97
.87
.8
.53
800
     At maximum tested gas flow of 210,000 ACFM, the maximum
system AP was measured at 6.9" W.C., when the first compart-
ment was isolated during the cleaning cycle.  This AP is
dependent on the AP utilized to initiate the cleaning cycle
and was within the guaranteed value.  A summary of perform-
ance  data is included in Table 4.
                             469

-------
          TABLE 4.  FABRIC FILTER PERFORMANCE DATA
                                    Design
Tested*
     Flue Gas Volume  (ACFM)        226,000

     Temperature                    310°F

     Inlet Dust Loading (Gr/ACF)    4.64
     Gas-to-Cloth Ratio (Gross)    1.78:1

                        (Net)      2.035:1

     Outlet Loading (Gr/ACF)        0.010

     Removal Efficiency (%)         .998

     Opacity (%)                     20
     System AP  ("W.C.)               8.0

     *Average of 4 runs
210,000

 342°F

 2.025

1.65:1

1.89:1

0.0081

.996

  3

 6.9
                     BAG TESTING PROGRAM

     For this first baghouse to see high sulfur fuel utiliz-
ing a bag finish of polymer/silicone-graphite-teflon, it
was determined that a bag fabric testing program would be
undertaken.  The acid-resistant-finish fiberglass cloth had
shown excellent life on low sulfur with occasional high
sulfur exposure at United Power Association's Elk River
station, but Independence would be the first high-sulfur-
only exposure.  The testing program was designed to examine
bags in operation for increasing periods of time to determine
the cumulative effect of high sulfur exposure.  Standard
fabric tests including weight, permeability, breaking
strength, Mullen burst, MIT flex and LOI were run.  In
addition, the pH of the dust cake was measured.

     Bag weight and permeability measurements can be indi-
cative of bag blinding.  Breaking strength, Mullen burst and
MIT flex tests are measures of chemical, physical or thermal
degradation of the cloth.  LOI (loss on ignition) shows the
presence or absence of the finish.  For comparison purposes,
new bags are used as the base for determination of any de-
gradation.  Weight and permeability are measured with "the
dust cake on the fabric and after it is removed by vacuuming
and washing.  Figures 3, 4 and 5 graphically summarize the
MIT flex tests and breaking strength results.  All of the
test results are enumerated in Table 5.
                             470

-------
          TABLE 5.  CITY  OF INDEPENDENCE TEST REPORT

Weight (oz/yd2)
Top
Bottom
NEW
13.5
APRIL
18.4/13.6
18.1/13.7
JULY
23.3/13.6
22.2/13.6
AUGUST
23.2/13.6
22.3/13.6
SEPTEMBER
22.2/13.7
22.1/13.7
  Permeability
    (cfm/ft )
    Top
    Bottom
47.0
  Strength (psi)
    (WARP/Fill)  592/292
    Top
    Bottom
  Mullen Burst
   (lb/in2)
    Top
    Bottom
 805
          4.73
          5.20
        601/280
        599/278
          725
          739
  1.79
  2.65
424/191
431/193
  538
  537
  1.73
  2.19
385/162
380/159
  470
  448
  1.93
  2.06
384/149
393/151
  476
  485
  MIT (cycles)
    (WARP/Fill) 40,038/3,517
Top
Bottom
L.O.I.
PH*
35,020/1,625
36,881/1,643
4.6 4.5
9.8
- 6,705/631
6,666/629
4.6
3.72
4,659/584
4,779/589
4.3
3.7
4,785/515
4,465/513
4.4

      •Surface pH on bag after cleaning cycle

      Note:  Values in "new" condition reflects fabric tested to
           determine initial values and not published data.
     The permeability in  the "as received" condition
averaged between 1.8 - 2.65  CFM/sq. -ft.  for bags tested
during the  last three test periods.  These values corres-
pond with values obtained from Research-Cottrell's Elk
River installation employing identical  acid-resistant-finish
bags.  At these permeabilities, the differential across the
cloth is a  maximum of 4"  dirty and approximately 2V
immediately after cleaning.

     After  vacuuming the  "as received"  fabric,-the vacuumed
permeability ranged between  11.2 - 12.6 cfm/ft  .  This is
somewhat lower than expected.  Under microscopic examination,
the cloth interstices show nodules with a high  sulfate con-
tent which  had encapsulated  the surface filaments thus pre-
venting proper dust release.  The low pH of the dust  cake
solution supports this observation.  However, as previously
stated, it  appears that it is not significantly affecting
                               471

-------
the cloth resistance while operating within the baghouse.
The washed permeability returned to the values in the "new
condition".

     The L.O.I, of the acid-resistant finish is specified
at a minimum of 4% by weight of the greige goods.  Since
after seven months of operation, the tested L.O.I, values
exceeded the minimum of 4%, the results clearly indicate
its stability.

     The M.I.T. flex cycle tests are indicative of the
number of flexes-to-failure which provides a comparison with
other fabric flex tests.   This M.I.T. flex endurance test is
probably the most misinterpreted and misunderstood test as
applied to bag life in baghouses today.

     Many variables affect M.I.T. flex-to-failure.  These
variables consist of yarn construction, twist levels, method
of texturizing, finishing, treatment procedures, finish cure
time/temperature, operation time and temperature in a bag-
house, removal efficiency of volatile matter and embedded
abrasive particles within the yarn structure.

     A fabric with a high flex value compared to a fabric
with a low flex value does not necessarily mean that one is
superior to the other.  A good way to compare one fabric to
another is to evaluate the percent loss of M.I.T. flex
cycles based upon the environment to which the fabric is
exposed.

     Fiberglass is the primary fabric in the utility and
industrial fly ash particulate market.  That fabric will
usually stabilize after 4-6 months of operation at approxi-
mately 50% reduction in M.I.T. flex cycles compared to new
fabric.  This reduction is primarily caused by physical and
thermal deterioration upon first exposure to flue gas.

     Upon reaching equilibrium, the fabric M.I.T. cycles will
remain virtually constant until the fiber begins to show
additional fatigue or breakdown due to chemical, thermal, or
additional physical deterioration.  Upon drastic degradation
from equilibrium, the rate of fiber fatigue/breakage due to
physical deterioration may indicate that the fabric is about
to fail and bags should be considered for replacement.

     If chemical or thermal deterioration also occurs, the
equilibrium value of flexes will be lower than fabric flex
values caused by physical deterioration only.  Chemical and
thermal deterioration usually occurs rapidly and can be
easily measured.  Once chemical or thermal adverse conditions
cease to exist, fabric degradation stops.  Physical deterior-
ation generally occurs where movement in the fabric exists.

                             472

-------
 For  example, reverse air cleaning causes  local  physical
 damage at the flex zone.  Chemical deterioration  occurs
 throughout the entire fabric.

     By comparing flex values, between  "flex  fold"  and
 "non-flex fold" areas, levels of chemical vs  physical
 deterioration can be established.  As an  example, consider
 the  flex cycle from the July test.  The new fabric  flex
 cycle was 3517.  Assuming the normal 50%  reduction  to reach
 equilibrium, 1759 flex cycles would have  been expected.
 However, due to chemical, thermal and physical  deterioration,
 the  actual equilibrium norm was at 630  cycles,  (test on flex
 fold position) , which equates to an additional  32%  loss.



             = 82%'" 82% ~ 50% = 32% additional  reduction
     Therefore, we can presume that the  fabric deteriorated
the equivalent of 1129 additional flex cycles due to chemical
and physical deterioration  (1759 - 630 = 1129) .  From  further
testing of the fabric on the "off- fold"  section, we obtained
a flex cycle value of 790 cycles vs the  630 cycles from the
"on fold" section.  Subtracting this "off- fold" value  from
the equilibrium norm we can speculate that the difference in
flex cycle loss was due strictly to chemical attack  (1759 -
790 = 969 flexes) or 86% of the additional loss was due
from chemical deterioration  1 - 1129 -  969 _ Q,.   and 14%
                                - TTTo — ~  6%
was due to physical deterioration (1.00  - .86 = 14%).

BAG FAILURES

     Knowing that most baghouse outlet loadings usually are
below 0.005 GR/ACF on utility applications, it was quite
surprising to observe loadings up to 0.0081 Gr/ACF during
the performance test.  Approximately a week after the  per-
formance tests were conducted, the boilers came down.
During an inspection of the baghouse 34  failed bags, over
    of the total installed quantity, were found.
     To date, a total of 93 bags have been replaced due to
mechanical fiber fractures, as shown in the photomicrograph
Figure 6.   Of the 93 bags replaced, 34 were completely failed
where the  bag had actually separated from the cap, and the
additional 59 appeared ready to separate; the top band was
visible, but was still in contact with the cap.   (Figure 7) .
It appears from numerous microscopic examinations that when
the caps were installed in the bags by the bag manufacturer,
the packers apparently pulled on the top cuff area incor-
rectly when seating the band and fabric against the cap seat,
thus severing the fibers.

                             473

-------
     Further inspection revealed that this top band was not
wrapped with a protective cover nor was the body of the bag
folded over to give a separate layer of fabric between the
band and bag body.

     To date, ten complete examinations have been performed
on used bags and two new spare bags and only four of the ten
did not reveal this fiber fracture.  In one instance,
several yarns had all 1632 filaments severed, all in the
position coinciding with the band top edge.  All fractures
were at the top interior side of the bags, thus making it
impossible to observe the defects without removing the cap
and cutting the fabric at the band.

     Observing the fractured filaments also in the new spare
bags confirms that the damage originated prior to installing
the bags in the baghouse.

     A survey of bag suppliers was undertaken to question
methods of installing these bands in bags.  The majority
responded that their standard construction for 8" or 12"
dia. bags includes a pre-wrapped band or the bag body folded
over to protect the bag itself from the band.  As a recom-
mendation to end users and O.E.M.'s to eliminate problems
such as this, specifications issued to bag manufacturers
should specifically require that all bands and ring covers
be independently wrapped or that multiple layers of
fabric exist between any metal object and the bag body itself.
               CONCLUSIONS AND RECOMMENDATIONS

     1.  The City of Independence fabric filter is sur-
         passing all performance guarantee requirements
         while high sulfur fuel is being burned.  The air
         quality system has not caused any boiler outages.

     2.  Startup was uneventful with only minor modifi-
         cations required.

     3.  Pre-coating the bags with inert flyash appears
         to have reduced the degree of bag blinding from
         chemical attack to a nominal level.

     4.  Reduction in MIT flexes-to-failure and tensile
         strength occurred after startup.  An equilibrium
         condition has been reached.

     5.  Continued monitoring of bag condition is recom-
         mended to predict baglife.
                             474

-------
     6.   Bag construction specifications should include
         provisions to prevent any metal and hardware
         contac-b with the bag cloth.
ACKNOWLEDGEMENT

     Research-Cottrell wishes to acknowledge the Environ-
mental Consultant Company of Phoenix, Arizona for their
assistance in performing physical/mechanical, chemical
tests and for the photo micrographs of bag fabric.
DISCLAIMER

     Findings and projections regarding fabric filter
performance and durability which appear herein are strictly
for the purpose of scientific and technical inquiry and
should not be construed as representations or warranties
regarding characteristics of fabric filter systems marketed
by Research-Cottrell.

     The work described in this paper was not funded by the
U.S. Environmental Protection Agency and therefore the
contents do not necessarily reflect the views of the Agency
and no official endorsement should be inferred.
                             475

-------
                                            Figure 1
                                  City of Independence
                                Interval Between Cleaning Cycles
                      Decreased Down to Near 20 Minutes Between Cycles
cr.
                                                      4.5" AP Initiation of cleaning
                                               7/12/82  60% Inlet damper opening RA fan
                                                      45 Sec. (indicated) RA poppet opening
     Full load
  5 - reached
6    11 AM
      12 N   1 PM  2 PM  3 PM  4 PM  5 PM  6 PM  7 PM  8 PM  9 PM  10 PM  11 PM  12 PM  1 AM

                                               Time

-------
o
   8
                                     Figure 2
                          City of  Independence
                        Cleaning Cycle After Final Tuning
                                                4.5" A P Initiation
                                         7/14/82  100% Inlet damper opening RA fan
                                                180 Sec. (indicated) RA poppet opening
   SAM   9AM 10AM  11AM  12 N  1PM   2PM   3PM   4PM   5PM   6PM  7PM   8PM

                                        Time

-------
                         Figure 3

                 Tensile Strength
                             400 -t
   600- -
   500- -
a.

 I

£  400
o>

I
«*
CO

"3

|  300-
co
£
A,

9


1200-
   100-
                                                     21
                     Warp yams
               Fill yams

I
April
I
June
I
Aug.
I
Oct.

    March     May     July      Sept.



                     1982 (Months)


                           478
Nov.

-------
                                              Figure 4
                                    MIT Flex to Failure
                                          (Filling Yarns)
    4000
o
o
0)
                                                                 Cycles
                                                                 10,000
                                                                  1,000
                                                               100
]?   3000- J
V)
JO)
u
>.
u
X
0)
2000- -
                                                                              Warp
                                                                              Fill
                                                                                           *\
                                                                                    V
                 0 Hours 10  '   ioo1   ' '1000
                   Days     1      7  21 32
     1000- -
Expected physical equilibrium norm	
       (50% degradation)
                                                                Physical & chemical deterioration
                                                                         (on flex fold)
                                          Month

-------
                               Figure 5
                      MIT Flex to  Failure
                            (Warp Yarns)
40,000
5,000--
      ~     »«      B     «      I    ' •      I      |     I
    Mar.  April  May   June   July  Aug.  Sept. Oct.   Nov.  Dec.

                           1982 (Months)
                            480

-------
      FIGURE 6.  GLASS FIBER FRACTURE
FIGURE 7.  FABRIC FAILURE AT TOP  SNAP BAND




                     481

-------
                         FUNDAMENTAL STRATEGIES
                                  FOR
                     CLEANING REVERSE AIR BAGHOUSES
               by:  GENERAL ELECTRIC ENVIRONMENTAL SERVICES. INC.

                    M.G. KETCHUCK
                    M.A. WALSH
                    O.F. FORTUNE
                    M.L. MILLER

                    ADAPCO

                    M.A. WHITTLESEY
                                ABSTRACT
     In order to choose among the various cleaning cycle strategies,
(Batch, Continuous, and Distributed)  used for utility reverse air cleaned
baghouses, it is necessary to understand the fundamentals of the cleaning
process.  This paper describes both analytical and experimental investiga-
tion of fundamental mechanics common to all cleaning strategies.
                                  482

-------
                               INTRODUCTION
     Considerable debate has occured as to what is the best way to clean
reverse air baghouses.  Continuous cleaning, Batch cleaning and Distributed
cleaning are all being used in operating utility baghouses with varying
degrees of success.   GEESI's operating experience indicates that a Batch
cleaning provides both reasonable pressure losses, and makes provision
for normal upsets in plant operating conditions, such as wet coal piles,
steam soot blowing,  low load flame stabilization, oil/gas supplemental
firing, minor boiler tube leaks,  etc.

     Another issue which generates considerable discussion is whether it
is better to allow a very heavy filtercake to build up on fabric filter
bags before cleaning them, or to  clean them, lightly loaded, as frequently
as possible.  Thick  filtercakes seem to produce better cleaning in practice,
but if overdone can  lead to slack bags and fabric damage.

     In order to insure that our  baghouse cleaning cycles are fully
optimized, GEESI has committed to on-going experimental and analytical
programs designed to gain more insight into the fundamentals of reverse
air cleaning, and to get factual  answers to the above questions.

     In this paper we would like  to report the results of two of these
programs:

     (1)  An analysis of the basic mechanics of the reverse air cleaning
process, using the DELTAP computer program.

     (2)  Experimental measurements, using load cells, of the change in
bag tensioning during the cleaning process.
                         THE FINITE ELEMENT MODEL
     Observation of filter bags in reverse gas type fabric filter units
has long established that the phenomenon of filter bag collapse and
suspension spring bottoming occur during cleaning.  To gain further
insight into the interdependent effects of the reverse gas flow and the
static/dynamic response of the total suspension system a Finite Element
model was developed utilizing the ANSYS program.  The following parameters
were studied to determine the effects on filter bag cleaning, filter bag
life, and structural loadings on the fabric filter structure.

                   o   Reverse Air Flow

                   o   Internal Decompression

                   o   Initial Tension

                   o   Dynamic Snap Back
                                    483

-------
     The Finite Element model shown in  Figure  1  reflects  the stiffness  of
the suspension support structure,  the suspension spring,  suspension hard-
ware, and the thimble/tube sheet structure.  The fabric filter bag was
modeled as two dimensional spars spanning  ring nodes.  In the Dynamic
Analysis portion of the study the  ring  mass was  lumped at the ring nodes
and the dust mass was attached to  the ring nodes via breakable links.   The
breakable links offered the ability to  study various dust shedding schemes
ranging from simultaneous release  at all node  points to a cascade  release
as dust cake from upper levels impinged upon lower  regions of the  filter
bags.  A gap element was utilized  to analyze the effect, of slack bags.
         TENSION
          SPRING
             SPAN
               I
             SPAN
             SPAN
               3
             SPAN
               4
                      UPPER BEAM
                      SUBSTRUCTURE
             SPAN
               6
             SPAN
               7
.RING NODES -
  HELD LATERALLY

55

•BAG  NODES

47
                    3 I
                    23
                    16
                     ,  SUBSTRUCTURE
         609 Pressurtzotlon Model
         Used To Determine
         Deflections And  Tenstonlng
         Spring Loads
                                              I g    Q DUST MASS
                   Snap Bock Dynastc
                   Analysis Model
          Figure  I.  Finite Elvient Modelling Schematics
                                   484

-------
                  REVERSE GAS FLOW/INTERNAL DECOMPRESSION
     The effects  of  Reverse Air Flow and Internal Decompression were
studied utilizing the Finite Element model combined with a fluids
approach.  The collapse  of the filter bag under the external pressure
loading of  reverse gas flow was studied using a Static Analysis to deter-
mine the deformed filter bag geometry and the bag spring rate.

     A Dynamic Analysis  was performed to determine the vertical mode
frequency response,  as it was initially felt that the rapid load increases
in the suspension spring was possibly related to dynamic response of the
Bag-Spring  System as the dust cake detached and rapidly unloaded the
filter bag  and springs.   The results of the Modal Analysis are shown in
Table 1.  It  is important to note that even in the case where the filter
bag lower span is collapsed, the fundamental frequencies are well above
the observed  frequency of oscillation of .125 to .167 HZ, and therefore,
would not make significant contributions to the observed peaks in filter
bag loadings.
TABLE 1.                   MODAL ANALYSIS SUMMARY
   Case                     Description             Mode      Freq.(HZ)

a


b


c

Bag assumed to be connected to a
rigid tubesheet and rigid upper
support beam.
Bag supported on a flexible
tubesheet and upper support
beam.
Bag supported by a flexible upper
beam and unsupported at the bottom
(i.e. bottom bag span collapsed).
1
2
3
1
2

1
2
3
5.58
12.77
20.92
5.49
7.58

2.58
9.01
16.69
     A Fluids  Analysis approach was used to study the effects of the
falling dust plug.   The effect is similar to that of a piston and cylinder,
however,  in this  case the cylinder walls (the filter bag) has permeability
and the piston (the dust plug) has porosity.  Figure 2 shows the concept
used in this portion of the study
                                    485

-------
0
1
i
1
1
1
1

|l 	
>
1

©
>
1"
S [_

, , . / ' / ' / ' /
^ — BAG WALL
^f
1 " "~4}
ii- -£'
!i

ir~^— 	 CUNIHUL VULUMb
I. '!
ll
1 ]l
JC l

"7"
^/Z/fa—L

i
i I

i
' ^^ 	 FALLING DUST CAKE
P
r
•
*
r
|-*— D— *•]
Figure 2. Schematic of Pressure Rel oi lonsh (ps
For Fluid Analysts Model
     The solution to  the problem involved establishing equivalent "G"
loadings on the spar  elements,  deformed filter bag geometry behind the
falling dust plug,  and inputing the stress strain properties of the filter
bag material.  Due  to the nonlinear material properties,  large deformations
of the bag structure, and stress stiffening effects,  an iterative procedure
was required to converge on the solution.

     Initially the  reverse gas  pressure loading was applied and the
structure was allowed to deform to  equilibrium geometry.   The dust cake
release was initiated at the top of the filter bag, and as shown in Figure
3 the porous dust plug began to fall towards the filter bag outlet.  As
the dust plug cascaded down the length of the filter bag the incremental
dust mass was released.  Maximum suspension spring loading occurred as
the porous dust plug  cleared the bottom of the filter bag.  The results
of one analysis is  shown in Figure  4.   Clearly, spring loadings of 2 to 3
times the initial filter bag tension can occur due to the internal
decompression behind  the falling porous dust plug.

     Filter bag suspension spring,  suspension hardware, and support
structures must be  designed to  accommodate these increased loadings and
proper design considerations must be made to eliminate structural failure
due to low cycle fatigue.
                                    486

-------
                         INWARD  MOTION  DUE  TO
                         SUCTION BEHIND FALLING
                         DUST  PLUG
                        DUST PLUG
Figure 3. Idealization of ihe Falling
          Dust Cake Upon Cleaning
                   487

-------
    250T
                               DUST PLUG
                               CLEARS BOTTOM
                                 CALCULATED -
                                 BASED ONI 25X
                                 POROSITY PLUG
                          BASED ON FIELD
                          DATA
                     TIME
Figure 4. Comparison Between Calculated and Field
          Measured Peak Spring Tension
                        488

-------
                           INITIAL BAG TENSION
     The effects of Initial Bag preload on  the  filter  bag  deformed geometry
and filter cake fracture was studied using  the  Finite  Element  model
developed.  The case of a fabric bag pretensioned  to 75  Ibs. H^ 15 Ibs.  was
studied.  The results of the analysis are presented in Figure  5.   Nodal
displacements for the upper most and lowest span are presented for both
the 90 Ibs. and the 60 Ibs. tension cases,  thereby bracketing  the initial
target tension of 75 Ibs.

     The effect on filter bag nodal displacement and subsequent cake
cracking appears to be narrowly banded.  This indicates  that the require-
ment for precise initial bag tension is currently  over emphasized by
equipment specifications and in fact may be detrimental  from a maintenance
point of view, as the requirement  for precision filter bag tension breeds
complicated hardware with increased cost.

     However, the case for an adequate magnitude of initial tension should
not be diminished as too low an initial tension can lead to early filter
bag failure.
         5.0
        « 4.0 +
        c
        £
        o
        c
        c 3.0
        o
        o
        o
        a
        s"
        o
        o
        2
          1.0 +
BOTTOM SPAN
                    15  LBS. OVER SPECIFIED
                    BAG TENSION

                    15  LBS. UNDER SPECIFIED
                    BAG TENSION
                              1.0                2.0
                            A Pressure  Cinches  H203

              Figure 5. Maximum Span  Displacements for the
                         Top and Hollo* Spans  Versus A P
                         for  A Range  of Bag Tensions
                                             3.0
                                    489

-------
                            DYNAMIC SNAP BACK
     The effect of  a  low  initial filter bag tension may result  in the  lower
region being untensioned  and slack.  In order to study this condition  the
Finite Element model  was  modified by placing a Gap Element in the bottom
bag span.  The filter bag model was then subjected to an initial  dust  cake
loading with the bottom span in a slack condition.  The dust cake was
released and the filter bag was allowed to respond.

     Figure 6 shows the force-time history response of the bottom span.
The fabric material is subjected to a severe, short duration pulse loading.
The effect on subsequent  filter bag spans diminishes as one moves up the
bag.  At the upper suspension point the peak load is non existent and  the
system oscillates about the new equilibrium without the dust loading.

     As previously discussed, the precise magnitude of initial  bag tension
is unimportant,  but to avoid damaging effects to filter bag material
sufficient minimum tension must be applied to prevent Dynamic Snap Back.
               320
               240--
           1/1
           CO
           -J
           LJ
           U
           s
                ISO--
                BO ••
                                    TIME
            Figure 8.  Snap-Bock  Analysis  -- Force  Time
                        Htslory For the BoilOB  Bag Span
                                   490

-------
                          EXPERIMENTAL FIELD DATA
     Visually observed tension spring and bag movements along with the
Finite Element model predictions of bag, tension spring, bag support
structures and filter cake responses during the reverse air cleaning cycle
prompted field "Insitu" testing to further quantify and support the
proposed theories.

     Testing was performed at a utility flyash baghouse operating at a flue
gas temperature of  270°F., with cleaning cycle initiation at a baghouse
compartment pressure drop of approximately 4 inches I^O.  The three acid
resistent coated filter bags tested were 12 inches diameter and 34 + feet
long.  Testing was  accomplished via the setup schematically shown on Figure
7, where a specifically designed load cell was placed in the bag suspension
system between the  tensioning hardware and the top bag cap.  Therefore,
monitoring of the spring/top of bag forces was enabled.

     A typical measured bag tension history is shown on Figure 8, with the
outlet, reverse air, and reverse air relief valve movements shown.  Several
important items to  note include:

     (1)  The gradual rise in spring tension at the onset of the reverse
air valve opening.
     (2)  The sharp gradient starting at the reverse air valve fully opened
position.
     (3)  The high  (approximately 2.5 times initial bag tension) peak
spring/top of bag tension.
     (4)  The spike-like peak duration most probably due to the force
generated by the falling dust cake.
         r
                STANDARD BAG  	
                TENSIONING  ASSEMBLY
                           LOAD  CELL
               TO OSCILLO6RAPH1C
                         RECORDER
                       12 INCH.  DIA..
                         FILTER  BAG
             Figure  7. Schematic  of  Bag  Tension Measurement  Setup
                                    491

-------
U
2
                                 OUTLET VALVE OPENED
                                 RA VALVE CLOSED
                                 RA RELIEF  VALVE  OPENED
                                 RA RELIEF VALVE CLOSED

                                 RA VALVE OPENED
                                 OUTLET VALVE CLOSED
             100              200

                SPRING  TENSION CLBS.3
300
Figure 8. Typical  Bag Response  During Cleaning
                         492

-------
                                CONCLUSIONS
     Both reverse  gas  flow  and  the  falling  porous  dust  plug  contribute to
filter bag  loadings during  typical  cleaning cycles.

     The internal  decompression effect  contributes significantly to
structural  loadings.   Proper  attention  in the design process is required
to insure against  low  cycle fatigue failure of the suspension structure
and hardware.

     The effect  of variances  in initial tension is narrow banded making
economic justification for  hardware capable of precision adjustment
difficult.  Precision  adjustment hardware breeds maintenance headaches.

     The magnitude of  initial tension is extremely important in reducing
dynamic snap back  of the  fabric filter  bag  and damage to the material
fibers.
              The work described in this paper was not funded
              by the U.S.  Environmental Protection Agency and
              therefore the contents do not necessarily reflect
              the views of the Agency and no official endorsement
              should be inferred.
                                    493

-------
        DESIGN CONSIDERATIONS FOR BAGHOUSE  -  DRY S02  SCRUBBER SYSTEMS

               by:  Owen F.  Fortune
                         and
                    Richard  L.  Miller
                    General  Electric Environmental Services,  Inc.
                    Lebanon, Pennsylvania   17042
                                  ABSTRACT

     The combination of  a reverse  air  cleaned baghouse and a dry sulfur
dioxide scrubber is attractive because of the ease with which the baghouse
can remove over 99.9% of the high  solids loading  in the gas stream exiting
the scrubber, and because sulfur dioxide scrubbing continues to occur in the
filter-cake on'the bags.  However, in  order to avoid having gypsum deposits
shortening the usable life of the  bags, close attention has to be paid to
several design considerations.  Among  them  are system response to boiler tube
leaks, approach to saturation temperature,  reheat system,  avoidance of
condensation in reverse  air cleaning system,  plant maintenance problems,
changing boiler loads, and changing sulfur  dioxide concentrations.  Experi-
ence-based design strategies to deal with these issues are discussed in this
paper.
                                     494

-------
INTRODUCTION

     In less  than  ten  years,  the  reverse-air cleaned,  fiberglass bag fabric
filter has  become  accepted, reliable technology for power plant dust emis-
sions compliance.   In  the  last  few years,  the utility  baghouses1 role has
been expanded to become  part  of unified fly ash/sulfur dioxide removal
systems.  When a "wet" S02 scrubber is  used, the baghouse remains immediately
downstream  of the  air  preheaters,  and design modifications are not needed.
However, when a "dry"  spray absorber S02 removal system is used, the
baghouse(s) are placed between  the scrubber outlets and the induced draft
fans, and must be  designed to function  for the dust loadings and close
approach to saturation temperature of the spray absorber exit flows.

     In 1980, we at General Electric Environmental Services, Inc. (GEESI) and
Anhydro A/S of Copenhagen, contracted to supply the City of Marquette,
Michigan, acting through Lutz,  Daily, and Brain, with  a spray absorber/
baghouse system for their  new 44  Megawatt Shiras No. 3 plant.  Some of the
system design parameters are  given in Table 1, and the completed unit is
shown in Figure 1.   Since  Shiras  No. 3  will be coming  on-line in the next
month, it is  timely to recount  the issues that were considered in the design
of the baghouse, and how they would apply to spray absorber/baghouse systems
for boilers as large as  800 megawatts.

TABLE 1.  CITY OF  MARQUETTE SHIRAS NO.  3

     •  PROCESS

             44 Megawatts
             226,000 CFM at 265°  F to spray absorber
             167,000 CFM at ]75°  F to baghouse
             1500  PPM  S02
             15% Ash
             80% S02 removal  guaranteed

     •  Baghouse

             75' x 45' in  plan; 62* high baghouse compartments
             168 bags  per  compartment
             bags  are  12 inch diameter  x 35 feet long
             140,000 square feet  of filtering surface

     •  Spray Absorber Vessel

             36* diameter  x 71' high
             Central 7800  RPM rotating  wheel atomizer
             200 HP vertical  AC motor
                                     495

-------
Figure 1.  GEESI Spray Absorber/Baghouse  System,  City of  Marquette,  Michigan
           Shiras No.  3 Station
CHOICE BETWEEN BAGHOUSE AND ELECTROSTATIC PRECIPITATOR

     Perhaps the first  issue to  be  considered is why one should use a
baghouse instead of  an  electrostatic  precipitator in the first place.  Since
the gas temperature  and humidity out  of  the spray absorber are essentially
constant regardless  of  boiler load,  it would seem to simplify the process of
sizing of a precipitator for this application.   Also,  the very high gas
moisture content is  beneficial to the electrostatic precipitation process.
Indeed, it seemed that  SCAs in the  range of 400-500 would be sufficient to
insure dust emission levels under 0.03 Ib/MBTU.

     However, in investigating the  issue, we found two strong reasons that
will almost always weigh the decision in favor of a baghouse for a new power
plant.  The first is that the dust  removal equipment must be capable of
maintaining compliance, even if  the S02  removal system is off-line.  For low
sulfur western coals, this means SCAs in the 600-800 range.  Compared to
precipitators in that size range, a baghouse will almost always be the more
economical choice (1).
                                    496

-------
     The second  strong  argument  for  teaming  baghouses  with spray absorbers
is that  about  three  times  as much S02  is  removed  from  the  gas  passing
through  a baghouse as through  a  precipitator operating at  the  same approach
temperature.   In other words, a baghouse will  account for about 12% of  the
total S02 removal across the system, while a precipitator  will only account
for about 4% of  the  reductions in S02  emissions.

     This may  seem like a  small  consideration,  but  present worth evaluation
factors  on the order of ten million  dollars  per ton per hour of lime consumed
are currently  being  used for utility boiler  S02 scrubbing  systems.   For  an
800 MW boiler  burning 1% sulfur  coal,  approximately 6  tons of  lime will  be
consumed each  hour at full load.   Thus, being able  to  operate  with an 8%
reduction in lime consumption  would  result in operational  savings valued in
today's  dollars  at 5 million dollars.

     If  the station  were burning a 3%  sulfur eastern coal, the lime consump-
tion would increase  by  a factor  of six (as required S02 removal efficiency
increases from 70% to 90%).  Thus, the financial  incentives towards using a
baghouse become  even more  pronounced,  even though a smaller precipitator
sizing (i.e.,  450 SCA)  would be  appropriate.

     Some air  pollution equipment vendors (2) maintain that precipitators
can be reliably  operated at lower approach to saturation temperatures than
baghouses, but other vendors,  including GEESI,  who  use central rotating  atom-
izers for fragmenting and  dispersing the  lime slurry into  the  boiler gas flov^
have not had this problem  (3,4,5,6).

     Once the  decision  has been  made to use  a baghouse, the two main baghouse
design factors to be considered  are:

        (a) To properly size the baghouse to handle the higher density,
           higher dust loading,  gas exiting the  spray absorber(s),

                                    and

        (b) To design the  system so  that  operation  will always be
           safely above the saturation temperature.

BAGHOUSE SIZING

     At  first  glance, the  fact that  the gas  entering the baghouse will be
100° to  150° F cooler than the gas exiting the  air  preheaters  - and ,  hence,
seems to have  only 80%  its flow  volume -  would  argue that  no modifications
need to  be made  to the  usual baghouse  sizing procedures for the spray absorb-
er FGD application.  Reality is, however, that

        (a) the  mass flow  rate of the  boiler gas  is increased  by as much
           as 6% by the water evaporated in the  spray absorber(s), and
                                    497

-------
        (b) a considerable amount of new solids are formed in the spray
            absorber, due to the combination of the gaseous sulfur diox-
            ide and calcium from the atomized slurry,  to form calcium
            sulfite and calcium sulfate.  In addition,  as much as 25%
            reductions in lime consumption are realized in a recycle
            process by first slurrying and then reinjecting as much as
            50% of the solids collected in the spray absorber and bag-
            house hoppers.

        (c) Finally, the density of the gas entering the baghouse has
            increased by at least 20%, and will influence the baghouse
            pressure drops proportionately.

These factors can combine to easily double the dust loading to the baghouse
and, thus, result in either substantially higher baghouse pressure losses, or
substantially larger baghouses.

     A positive factor in holding down baghouse sizes  is to design the spray
absorber so that a considerable  portion of the solids  enter and formed in the
vessel are captured in the spray absorber hopper.  The standard Anhydro reac-
tion vessel is shown in Figure 2.  A tapered inlet scroll and guide vanes are
used to set up a cyclonic gas flow pattern which is reinforced by the rotary
momentum of the atomized slurry  spray and a near horizontal flow exit is
positioned at two-thirds the height of the-conical hopper.  The controlled
cyclonic flow pattern, combined  with the Vortex holding effect of the exit
duct, results in

        (a) solids mechanical collection efficiencies  of from 45% to 55%,

                                     and

        (b) several hours of hopper storage time to insure vessel maintain-
            ability.

     The use of this type of spray absorber vessel, rather than a vessel that
passes on all,  or most,  of the solids to the baghouse(s), reduces the bag-
house inlet dust loading to levels similar to that at  the air preheater exit.
For instance,  in the Shiras No.  3 design, the maximum  load dust loadings are:

                          STATION                DUST  LOADING (tons/hour)

                  Air Preheater  Exit                  4.4
                  Spray Absorber Vessel Exit          5.2
                  Baghouse Inlet                      5.3

Hence, sizing the baghouse using arbitrary rules (such as gross air-to-cloth
ratios of 1.5:1 or 2.0:1 net-net with reverse air), that ignore the processes
occurring in the FGD vessels is  poor practice.  It is  necessary to understand
the consequences to the baghouse(s) of the various different types (multi
                                    498

-------
Figure  2.  Anhydro  Spray  Absorber Vessel
                                     499

-------
two-fluid nozzle, multi-rotating atomizer, and central rotating atomizer) of
atomizer and hopper  configurations, worst case coal sulfur  content,  and their
resulting impact on  baghouse design.

REHEAT

     Another factor  which  can have a significant impact  on  baghouse  design is
the practice of adding ,hotter gas to the spray absorber  exit  flow for fear of
sub-dewpoint operations occurring in the baghouse or  induced  draft fans.
Many different-ways  of "reheating" the FGD vessel outlet flows have  been
proposed.  A few of  the more logical ones are:

        •  Bypass  of part of the economizer outlet flow around both the air
           preheaters and  spray absorber vessels.

        •  Bypass of part of the air preheater outlet flow  around the spray
           absorber  vessels.

        •  Reheat of baghouse exit gas, using a "waste"  steam heat exchanger
           to elevate temperature.

     Considering  that air pollution control manufacturers  have been able to
demonstrate stable rotating atomizer spray absorber operation at 20° F
approach temperatures, and that about 10° F of reheating occurs across an
induced draft fan operating at a 20" W.G. head, and that a  well built
baghouse - particularly one whose compartments have been pressure tested
before being  insulated - will have a temperature drop of less than  10° F, it
is difficult to see  why it is necessary to boost the  spray  absorber  exit
temperature before entering the baghouse when the boiler is operating at, or
near, full load.  Particularly since doing so will decrease the amount  of
S02 removed across the baghouse and, hence, the amount of lime consumed in
the spray absorber.

     However, as the mass flow through the system decreases,  thermal losses
through the baghouse will increase and can reach 15 to 20°  F, even in a low
inleakage, well insulated system.  And at low boiler, lime  consumption is
less of an issue.

     These factors argue for selection of a low capital  cost, minimal capa-
city (10e F) reheat  system.  GEESI feels that the best way  of supplying this
reheat capacity is to bypass a few percent of the economizer  outlet  around
the air preheaters.  This  insures that 450° F to 650° F  temperature  differ-
ential is available  to elevate the baghouse gas inlet temperature with little
capital cost beyond  thermocouples, well insulated ductwork, and isolation and
modulation dampers.  Neither steam piping and valves  nor electric heaters
need be purchased and maintained.

PREVENTION OF THERMAL LOSSES

     Using baghouse  gas temperatures 20° F above saturation reinforces the
                                    500

-------
importance of minimizing air inleakage, and external and internal insulation
during construction of the baghouse is emphasized.   Pressure leak testing of
the compartments before they are insulated, durable door gasketing and door
insulation shields, and minimizing the number of baghouse expansion joints,
all become even more important than they are for straight fly ash collection
baghouses.

     In particular, the issue of compartment internal insulation becomes of
greatly increased interest and concern.  If a compartment with uninsulated
internal walls is taken off line for inspection, then the wall temperature in
the adjoining compartments will go below the gas dewpoint and there will be
condensation on the on-line compartment side of the wall.  This argues that
full height internal insulation is needed on compartment walls to prevent
wall and tubesheet corner condensation.

     In a similar manner, heating the bottom of the baghouse hoppers to
prevent condensation and insure free flowing ash removal becomes more
important than in a straight fly ash baghouse, as does the importance of
emptying  the hoppers once a shift.  In areas with severe winter climates,
consideration should be given to insulated hopper enclosures in order to
minimize the worst case sizing of the hopper heaters.

LOW LOAD OPERATION

     For boilers larger than 200 MW, it is probable that more than one struc-
tural baghouse will be used per boiler.  Discussions have appeared in the
literature (7) recommending that at low load, entire baghouses be taken off-
line and isolated in order to minimize thermal losses due to having greatly
reduced mass flow rate per square foot of insulated external wall in each
baghouse.  It is true that a baghouse with, say, a 10° F loss at 100% boiler
load, will have about an 18° F loss at 25% load.  However, this low load
condition can be compensated for by either operating the on-line spray
absorber vessels at a higher approach temperature, or using a minimal capa-
city reheat system.

     The reason that this is preferable to simply, isolating baghouses, or
compartments, is that a typical sized, 4000-bag fabric filter will have about
two tons of water vapor in it when brought off line.  Within a few hours, all
of the water will condense as the baghouse cools down, unless an automatic
auxiliary purge system is added to the baghouse to replace the trapped boiler
gas with ambient air.  Then, when boiler load increases, the baghouse must be
reheated before being brought back on line.

     It is difficult to see what is gained through this elaborate procedure,
which significantly increases the capital cost of the baghouse by adding
massive (i.e., 15' x 12') inlet and outlet isolation guillotine dampers, as
well as the new ductwork and many-compartment dampers of the automated purge
system, while doing nothing to  improve system performance.
                                     501

-------
RESPONSE TO SYSTEM UPSETS

     The presence of a spray absorber upstream of a baghouse protects  the
baghouse against the consequences  of many of the minor upsets that  occur in
boiler operation during the course of a year.  For instance, a properly
designed spray absorber system minimizes the consequences of minor  boiler
tube leaks, since it will reduce the amount of recycle, or dilution, water as
the boiler gas humidity rises in  order for the spray absorber outlet  temper-
ature to stay at the specified approach to saturation temperature.  Also, if
one air preheater wheel stops turning, the ability of the spray absorber
system to reduce the gas temperature by at least 150° F, will keep  the aver-
age gas temperature entering the baghouse below 500° F and,  hence,  not even
require bypass of the baghouse.

     Operational upsets in the spray absorber systems - such as a lime slurry
valve failing open - that could adversely impact the baghouse by causing sub-
dewpoint operating conditions, can be controlled by vessel outlet temperature
and humidity monitoring.  For example, the GEESI Two-Loop Control system,
shown in Figure 3, would respond by closing off the vessel recycle  flow, as
well as simply alarming the failure.
     / u. %
                   DRY FLUE GAS DESULFURIZATION SYSTEM
                       SPRAY ABSORBER/FABRIC FILTER
Figure 3.  GEESI Two-Loop Control  System
                                    502

-------
BAG FABRICS AND COATINGS

     It will still be necessary for the baghouse to be able to reduce dust
emissions to below 0.03 Ibm/MBTU,  regardless of whether or not the'spray
absorber is on-line.   Thus,  the bag fabric must be able to withstand gas
temperatures above 300° F, as well as near saturated gas operation at 140 to
170° F.  These dual requirements lead to fiberglass remaining as the most
cost effective bag material in utility fabric filters.

     Since baghouse dust inlet loadings will be higher in the combined S02/
particulate removal system,  GEESI believes that 13-1/2 ounce, rather than
9-1/2 ounce, fabric should be used because of its greater durability, as
shown by flex testing.

     The value of the bag coating now chiefly lies in its ability to act as
a lubricant between glass fibers.   Thus, even in the combined systems'
highly alkali filter cakes,  the so-called "Acid Resistant" coatings are
preferred, since they flow more evenly during application and, hence, cover
a higher percentage of the glass fiber surface area than do the earlier
coatings, such as silicone-graphite-tefIon, or pure Teflon-B.

BAGHOUSE MAINTENANCE CONSIDERATIONS

     The operation of either a wet or dry S02 scrubbing system imposes a
significant load on a plant's Maintenance and Operations Departments and,
hence, it is of increased importance that baghouse maintenance issues be
minimal.

     The best way to insure this is to invest the time needed to keep the
system automatic temperature, S02, and humidity monitors in good calibration.
In particular, humidity sensors require periodic attention.  GEESI believes
that, at present, extractive sampling monitors give the most reliable infor-
mation from dust laden flows.  Thus, care should be taken that insulation and
heat tracing of the sampling lines is adequate, as well as frequently check-
ing the system's calibration.

     The payoff for doing this will be avoidance of hopper pluggage and bag
damage problems caused by sub-dewpoint operation.

SUMMARY

     One very general conclusion that can be drawn from working out the
design of a 'composite dry spray absorber/baghouse system is that the same
factors which lead to good straight baghouse design, simply become more
important for the composite system:

     •  Sharply evaluate the design point dust loading, mean particle size,
        gas density, etc., for sizing the baghouse, rather than just rely-
        ing on rule-of-thumb formulas.
                                     503

-------
     •  Use and maintain gas temperature and humidity  sensors  to  avoid
        sub-dewpoint operation for  even short  periods  of  time.

     •  Use insulation on compartment  internal walls to prevent local-
        ized condensation.

     •  Heat hopper bottoms  to head off potential pluggage problems.

     •  Pressure test compartments  during  erection  to  eliminate sources
        of ambient air inleakage  due to skipped welds, or poor door
        gasketing.

     •  Instrument the system so  that  a mimi-computer  can control feedback
        loops to avoid operational  problems.   Set up agreed upon calibra-
        tion procedures to insure that the computer inputs are reliable.

     •  Use fiberglass bags  with  protective coatings.

     A baghouse designed along these lines should be, a high-reliability
component in a composite spray absorber SC>2 removal/fabric filter system.
               The work described  in this  paper was  not  funded
               by the U.S.  Environmental Protection  Agency and
               therefore the  contents  do not  necessarily reflect
               the views of the Agency and no official endorsement
               should be inferred.

-------
                               REFERENCES
1.  Campbell, K.S., et al.,  Economics of Fabric Filters Versus Precipita-
    tors.  Electric Power Research Institute  FP-775, June, 1978.

2.  Wilkinson, J. M. and Tonn, D. P., Baghouse vs. Precipitator for Dry
    Scrubber Systems — Pilot Study Results.  4th International Coal Utili-
    zation Exhibition & Conference, Houston, Texas, December 1981,  Vol.  3,
    p. 168.

3.  Parsons, E. L. , Hemenway, L. F., Kragh, 0. T. , and Brna, T. G., SO
    Removal by Dry FGD.  Proceedings:  Symposium on Flue Gas Desul-
    furization — Houston, October 1980, Vol. 2, p. 801.

4.  Parsons, E. L. , Boscak, V., Brna, T. G., and Ostop, R. L.,  S02 Removal
    by Dry Injection and Spray Absorption Techniques.  Third Symposium on
    the Transfer and Utilization of Particulate Control Technology, March
    1981:  Volume 1.  Control of Emissions  from Coal Fired Boilers, p. 303.

5.  Samuel, E. A., Lugar, T. W., Lapp, D. E. , and Fortune, 0. F.,  Dry FGD
    Pilot Plant Results:  Lime Spray Absorption for High Sulfur Coal and
    Dry Injection of Sodium Compounds for Low Sulfur Coals.  Paper pre-
    sented at 1982 Symposium on Flue Gas Desulfurization, Hollywood,
    Florida.

6.  Meyler, J. A. and Felsvang, K., One Year of Operation of the Riverside
    Dry Scrubber.  Paper presented at the 44th American Power Conference,
    April, 1982.

7.  Kaplan, S. M., et al., Dry Scrubbing at Northern States Power Company
    Riverside Generating Plant.  Paper presented at 1982 Symposium on Flue
    Gas Desulfurization, Hollywood, Florida.
                                   505

-------
           RESULTS OF BAGHOUSE AND FABRIC  TESTING AT  RIVERSIDE

            Bys H. W. Spencer III, Y. J. Chen,  M.  T.  Quach
                         Joy Manufacturing Company
                       Western Precipitation Division
                      Los Angeles, California    90039
                                  ABSTRACT

     This paper presents the results  of one year of baghouse and fabric
testing at the Riverside Dry FGD Demonstration Facility during 1981.   Opera-
ting parameters and baghouse performance data are summarized.  The results
of the fabric evaluation test are discussed.  Pressure drop measurements are
reported and pressure drop  predictions  for various fabric filters based on
the experimental data are compared.   Pressure drop predictions are compared
with overall baghouse pressure drop.  Good agreement between the predicted
values and the actual measurements are  reported.  Measurements reported in
the paper cover operation with three  different coals.
                                     506

-------
                                INTRODUCTION
     This paper presents the results of one year of baghouse  and  fabric  tes-
ting at the Riverside dry FGD demonstration facility during 1981.   The
Riverside dry FGD demonstration facility is located in Minneapolis, Minnesota,
at Northern States Power's Riverside Station.   Operating parameters and  bag-
house performance data are summarized for the period January, 1981 to December
1981.  During this period, the Western Precipitation reverse  air  baghouse  was
successfully operated downstream from the NIRO spray dryer.

           EQUIPMENT DESCRIPTION DRY FLUE GAS DESULFURIZATION PLANT


     Figure 1 is a schematic of the JOY/NIRO dry FGD Demonstration Plant.   The
design conditions and technical data for the fabric filter portion of the
plant are given in Table I.

                             PROGRAM OBJECTIVES


     The objectives of the JOY/NIRO test program at Riverside are listed
below:

     1.  Confirming the scale-up factors from pilot to full-scale operation.
     2.  Confirming the sensitivity of various operating parameters in  a
         full-scale system.
     3.  Demonstration of the system's operational flexibility, including
         turn-down, start-up, shut-down, and normal operation.
     4.  Demonstration of long-term, low-temperature operation.
     5.  Evaluation of various fabric filter media with dry FGD application.
     6.  Demonstration of operation with various fuels, including low sulfur
         Western coal and higher sulfur Eastern or Midwestern coals.
     7.  Demonstration of particulate and S02 removal capabilities with either
         a baghouse or an electrostatic precipitator.
     8.  Characterization of particulate matter emitted from the system.
     9.  Determining the characteristics of the dry FGD waste product with
         respect to disposal and utilization.
    10.  Characterization of the system operation and maintenance costs.
    11.  Optimization of energy and reagent consumption.
    12.  The establishment of system availability and reliability data.
    13.  The development of operating and maintenance procedures.
    14.  The training of power plant operators.

     During 1981, a majority of these objectives were met.  The test and oper-
ating record for the facility during 1981 is given in Figure 2.  As shown,
tests were conducted with three fuels having sulfur contents ranging from .8%
to 3.2%.  The characteristics of these fuels are summarized in Table II.
Photomicrographs of isokinetically-collected dust samples at the inlet of  the
SDA and the inlet of the baghouse are displayed in Figures 3 and 4.  Larger
                                     507

-------
particle sizes were found at the baghouse inlet with the high sulfur fuel.
This is most likely explained by the higher alkalinity of the slurry that was
being spray-dried.  Typical operating conditions for each of the coals are
given in Table III.  The Bahco particle size measurements of the baghouse
dust samples contained in this table are consistent with the photomicrograph
data.  Dust loadings given in the Table were measured at the inlet of the
baghouse.

     Typical operating baghouse parameters with and without .the spray dryer
in service are given in Table IV.  When the spray dryer is in operation, the
baghouse runs at a considerably lower temperature than for a non-FGD instal-
lation.  Dust loadings are higher and collection efficiencies are also
higher.  Data in Table IV shows increased emissions with broken bags.  Bag
failures are discussed later in this paper and were confined to certain
types of bags.  Figures 5 and 6 are the plots of the operating parameters
during selected periods which include the high sulfur tests.

                            FABRIC EVALUATION
     One of the major emphases of the JOY test program is to evaluate a
variety of fabric materials in search of  materials with longer life, lower
cost, and lower pressure drop characteristics.  The bag arrangements in the
twelve compartment baghouses for the first phase and second phase of the test
programs are given in Figures 7 and 8.  The design characteristics for each
of these materials are given in Table V.   All the bags tested were fiberglass
except for the ones in Compartment 9, which were polyester.  The polyester
bag type was not in service long enough for full test evaluation due to its
temperature limitation.

     The test program was planned to evaluate the fabric strength and bag
life.  New bags and bags which have been  in service for certain periods of
time were removed for evaluation tests, including weight tests, breaking
strengths, Mullen burst, MIT flex test and permeability tests.  In-situ bag
weights were measured to check the dust build-up trends on the bag surface.
A representative sample of in-situ bag weight of three bag types is plotted
in Figure 9.  Significant variation was found between bag material types with
the texturized fiberglass bags having a much higher residual dust layer than
the non-texturized bags.  Very little variation was seen between the front
and back of compartments.  Measurements of the variation in bag weight from
compartment to compartment was also made  and again, there was no significant
difference.

     An example of decrease in breaking strengths and Mullen burst strengths
with time is plotted in Figure 10.  This  plot is for a non-texturized light-
weight teflon-finished fiberglass bag. As shown in a later plot, this bag
had low failure rates even though there was a measurable decrease in the
strength of the bag.

     In Figure 11, the permeabilities of  bag samples and corresponding bag
materials weights are compared for as received, vacuumed, and washed samples.


                                     508

-------
The data shows that washing restored original permeabilities and weights.
Vacuuming the material only partially restored the permeability.  As expec-
ted, the permeabilities of the material in service are considerably less  than
the new materials.

                                 BAG LIFE
     Bag life is a major factor in determining the economics of fabric fil-
ters for utility applications.  Figure 12 presents the cumulative bag fail-
ure rate as experienced at Riverside.  Two of the bag materials, which have
teflon coatings, showed very low failure rates; failures were apparently
limited to bags damaged during initial installation.   However, two bag
materials with silicone, graphite, teflon tri-coated  finish had unacceptable
failure rates.  The finish apparently did not protect the bags as well as the
teflon coating.  The bags with high failure rates also have lower strength in
terms of MIT flex cycles.  Fabrication errors also contributed to the fail-
ures which occurred at the top and lower bands, ring  cover stitchings and
outer ring cover areas.  The tri-coated bags have now been removed from the
Riverside baghouse.  We are continuing to monitor the life of the teflon-
coated bags and other bag materials.

                      BAGHOUSE PRESSURE MEASUREMENTS
     One of the major factors controlling the design of a fabric filter is
minimizing pressure drop.  The Riverside baghouse has been instrumented to
measure flange-to-flange pressure drop, and pressure drop across the indivi-
dual compartments, outlet valves and inlet dampers.  Pressure drop has been
monitored over a range of air-to-cloth ratios by reducing the number of
compartments in service under various boiler loads.  Pressure drop was
measured during full flow periods, settling periods (when one compartment is
out for cleaning), and reverse air cleaning (when one compartment is out for
cleaning with the reverse air flow), corresponding to the gross, net and net-
net air-to-cloth ratios respectively.  A sample trace of flange-to-flange
pressure drop is given in Figure 13.

     Individual compartment flow rate and pressure drop were measured for
determination of filter drag, which is the ratio of compartment pressure drop
and air-to-cloth ratio.  By measuring the dust concentration to the baghouse
and assuming uniform distribution of the dust between all compartments, the
incremental areal cake density was calculated for each compartment.  The
incremental areal cake density is equal to the quantity of dust collected per
unit area between the cleaning cycles.  A knowledge of the dependence of the
filter drag on areal cake density can be used to determine the effective drag
coefficient, SE, and specific dust-resistant coefficients K2 for each speci-
fic bag material as shown in Figure 14.  Knowledge of these coefficients can
be used to predict pressure drops as a function of air-to-cloth ratio, dust
concentrations, and cleaning cycles.  In Figure 14, filter drag is plotted as
a function of the incremental areal cake density for Type B bag materials at
Riverside.  Residual drag, which is the real drag of a filter after cleaning,
is plotted at the incremental areal cake density just prior to cleaning

                                    509

-------
instead of at the zero incremental areal cake density.   The extrapolation of
the filter drag curve at zero cake density point,  Sg, does  not coincide with
the measured residual drag value, however, it is within the expected range.
The slope of the curve determines the value of K£.   Another method using a
computer model developed by Western Precipitation  was used  to reduce the data
and calculate the effective drag, SE, and specific dust-resistant coefficient
K2-  The pressure drops prediction based on these  coeffficients as a function
of air-to-cloth ratio is presented as an example in Figure  15.  Two pressure
drop comparisons are given.  The solid curves show pressure drops across
different kinds of material without taking into account the effect of clean-
ing process.  Considerable differences are found at higher  air-to-cloth
ratios.  The open symbols compare the pressure drop including the effect of
the cleaning process of a multi-compartment system. In this case, only minor
differences between different materials are found.

     In Figure 16, total baghouse pressure drop at Riverside is plotted as a
function of air-to-cloth ratio and compared with computer predictions.  Data
are given for three coals with and without SDA operation.  There is no appar-
ent trend with respect to the three different coal types.  The data shown
correspond well with pilot plant data and with the computer-predicted pres-
sure drops.

                                 SUMMARY
     In summary, the baghouse at Riverside has operated well.   Baglife for
teflon-coated bags has been good, pressure drops have been predictable and
reproducible.  The baghouse as a particulate collector for dry scrubber
should be considered as a demonstrated technology.

                                 REFERENCES
1.  Kaplan, Steven M., Chen, Yang-Jen,  Sannes,  Carl A.,  Jr., "Dry Scrubbing
    at Northern States Power Company Riverside  Generating Plant," EPA/EPRI
    Symposium on Flue Gas Desulfurization,  May  17-20,  1982, Hollywood,
    Florida.

2.  Spencer, H. W. Ill, Brown, Bert, Chen,  Yang-Jen, "Experience with Bag-
    house for Dry FGD Service," 74th APCA Meeting,  June  21-26,  1981,
    Philadelphia, Pennsylvania, Paper No. 81.95.
The work described in this paper was not funded by the U.S.  Environmental
Protection Agency and therefore the contents do not necessarily reflect the
views of the Agency and no official endorsement should be inferred.
                                    510

-------
                          TABLE  I
                 THERM-0-FLEX FABRIC FILTER DATA SUMMARY
PLANT NAME:  Northern States Power Company
            Riverside Generating Plant

LOCATION:    Minneapolis, Minnesota

DESIGN CONDITIONS (FGD)*

     Gas Volume (max acfm)	420,000 (540,000)
     Design Temperature (°F)	     500
     Outlet Loading (gr/106 Btu)	     210
     Pressure Drop (in. VWC)	       6
     Design Pressure (in. VWC)	     -30 to +20
     Total Filter Area (sq. ft.)	321,200
     Total Filter Ratio	1.3111 (1.68:1)

TECHNICAL DATA

     No. of Fabric Filters	1
     No. of Rows	2
     No. of Compartments /Row . .,	6
     Filter Bags:
         Total	3000
         Per Compartment	250
         Diameter (in.)  	   12
         Length (ft.)  	   35

  * Values in  parenthesis  are  for non-FGD  system operation.  Where  none are
   shown, values can be assumed to be the same.
                            TABLE  II
          FUELS USED IN SDA/BH TESTING





Average
Sulfur %
Aah%
Moisture %
BTU/lb.
Normal
85%CoMrip'
(Rotebud, Montani)
(McKay Seam)
15% Petroleum Coke

1.37
8.33
22.0
9,500
Hi-Sulfur
•HnoteCoal




3.19
10.36
12.76
10,930
Low-Sulfur
Sarpy Creek Coal
(Sarpy Creek, MT)
(Robinson Seam)


0.79
9.20
25.33
8,614
                                511

-------
                 TABLE  III
 TYPICAL OPERATING CONDITIONS
• AT Ad. Sat (T)
- Feed Rate (GPM)
• FeedSottd*
• Dust Loading (gr/ACF)
• Bulk Penalty (tapped) («W)
• Bahco Mast Mam (pm)
• Flow Rate (ACFM)
M-8
19.2
1.17
13&5
35.49
10.2
SOX)
17
390,560
Sarpy Creek
17
1.72
140.33
42.42
11.8
02.5
9.4
347,200
Colstrip
17.3
1.34
121.85
44.77
9.9
56.2
8.4
379,950
                  TABLE  IV
             BH Operating Range
—WHhSDA:
   • Inlet Temperature (T)                 140—ITS
   • MetGMVokmio(ACFM)               219.000 — 444,000
   • Inlet DuttLoadtog (gr/ACF)              4J7 —12.1
   • Outlet Dutl Loading (gr/ACF)*            0.0010 — 0.0192-
   • EMctoncy (%)•                      MM' — MM
 -Without SOA:
   • tale! Temperature fF)
   • mlel Gee Volume (ACFM)
   • Met Oust LoedhtgfT)
270 — 310
182,000 — 458,000
1.1 — IS
•Mgh outlet toadMtg end tow efficiency ere due to broken beg*.
                 TABLE  V
                 FABRIC SPECIFICATIONS
Avg. Finish
Wt.
Oz./Sq. Yd.
FABRIC A


FABRIC B
FABRIC C
FABRIC D
FABRIC E
FABRIC F
FABRIC G
FABRIC H
FABRIC J
FABRIC K
8.4


9.3
8.8
8.4
9.3
9.5
13.5
9.5
9.5
9.5
Silicone
Graphite
Teflon
Teflon
Teflon
Grafosil
—
Teflon B
Acid Res.
Teflon B
Acid Res.
Teflon
Material
Glass


Glass
Glass
Glass
Polyester
Glass
Glass
Glass
Glass
Glass
                      512

-------
                          FIGURE  1
        FLOW  DIAGRAM - DRY SCRUBBING SYSTEM AT RIVERSIDE
                         FIGURE  2
            TESTING & OPERATION RECORD
                                   SARPY CREEK
                                   TRANSITION   O-
                                    HIGH-S  —-
                       SDA-ESP    •»«•
                 DEMO.
      NSP EQUIP. DOWN • -
            PARA. • —
       INTERPOLL O
      FLY ASH -O
O
                       O
        O  O CX">O
         1st  1st  1st   1st   1st  1st  1st  1st  1st  1st  1st   1st
         JAN FEB MAR  APR  MAY JUN  JUL  AUG SEP OCT NOV  DEC 81
Note:
           One day event
                             513

-------
               FIGURE 3
 Hi Sulfur
   BH INLET
DUST SAMPLES

    Colstrlp
                     Sarpy Creek
              FIGURE 4
Hi Sulfur
   SDA INLET
DUST SAMPLES

       Colstrlp
                   Sarpy Creek
                 514

-------
                        FIGURE 5
             FGD OPERATION PARAMETERS

   AUG       SEP             SEP           OCT    NOV
   21  25 26  3  89   10 11 22 23 24 26 27    1 2  23  3  9 10 11  12
160


130


400


200


100


 80


 60


 40


 20


  0

 40
 30
 20
 10
                    BH OUTLET TEMP.


                    BH OUTLET SO,
                    AT AD. SAT.
   Jim
   IS	 II .
                         FIGURE 6

            BH OPERATION PARAMETERS
 JULY    AUG                SEP               OCT
. 2«	30	4	6 	10	13	20	24	3	9	10	21	23	25 _ I 	6	 15
                   INLET GH ACF
                FEED SOLID %
                                               L
                                                          CO
                                                          O
                                                          co
                PARA	
                                     -HIGH SULFUR-
                             515

-------
        FIGURE 7
NSP, RIVERSIDE BAGHOUSE
   BAG CONFIGURATION
      March - Dec., 1981
1
A
3
C
5
A
7
0
9
E
11
A

2
B
4
B
6
B
8
B
10
B
12
B
                          lil
        FIGURE 8
NSP, RIVERSIDE BAGHOUSE
   BAG CONFIGURATION
  Second Start-up — June, 1982
1
H
and
F

2
B
3

C


4
B
5

K


6
B
7

K

9

J


8
B
10
B
11

G


12
B
                          lil
          516

-------
                  FIGURE 9
                BAG WEIGHT
    FRONT VS. REAR OF COMPARTMENT
   50
w^
g  40
i-

ui  3°
O
to  20
     MEW
                     FRONT   REAR
APR 6
    JUN9
    1981
TEST DATE
                                      A (Comp. 1)
SEP 15 FEB11
        1982
 700
  600
  500
  0
                  FIGURE 10

                      TYPE C
                                  FLEX CYCLES

                                    WARP
           BREAKING STRENGTH
               1580        2790
                 BH OPERATING TIME
                         4450 MRS
                      517

-------
                   1500              2790
                BH OPERATING TIME
             MRS
                    FIGURE 12
0
                  BAG FAILURES
 Jan.  Feb.  Mar.  Apr.  May  Jun  ,)ul
                    DISCOVERY DATE
Aug.  Sep.  Oct.  Nov. Dec.
                  1981
                       518

-------
                          FIGURE 13


        TYPICAL BAGHOUSE SYSTEM PRESSURE DROP
O
M
Q

01
W
(Q

-------
o
 CN
a
w
m
10

§
U
O
«
o

w
w
w
w
        FIGURE  15


COMPARISON  OF PRESSURE DROP  PREDICTION
            Fabric A

           Fabric B

          Fabric C
                      A   O


                      B   A


                      C   D
              AIR-TO-CLOTH RATIO


                 FIGURE 16

           AP VS. AIR-TO-CLOTH


O J
(N
X
z
H
0,
o
B
a
i
U)
w
o, •
U
U)
o •
0
FULL
FLOW

COLSTRIP 0
HIGH SULFUR *
SARPY CREEK O

FLY ASH ONLY f


SETTLING


O
A
0

_|_


REVE
AIR

0
A
O

_i_


PREDICTED PRESSURE




DROP

MAX
UT?* 11





         AIR-TO-CLOTH RATIO (FT/MIN.)
                   520

-------
            REACTIVITY OF FLY ASHES IN A SPRAY DRYER/FABRIC FILTER
                                FGD PILOT PLANT

                                      by

            Wayne T. Davis, Randal E. Pudelek, and Gregory D. Reed

                        Department of Civil Engineering
                          The University of Tennessee
                       Knoxville, Tennessee  37996-2010
                                   ABSTRACT
                                         /

     This paper summarizes the results of a study in which the reactivities of
23 fly ashes  were  evaluated in a pilot plant spray dryer/fabric filter sulfur
dioxide removal system.  The primary objective was to determine the ability of
each  fly ash  (including lignite,  subbituminous,  and bituminous  eastern  and
western ashes) to  remove S02 when placed in a water-based slurry and atomized
by a  spinning  disk atomizer into a spray dryer located on a slipstream from a
stoker-fired boiler.

     Data are summarized in both tabular and graphical form including chemical
kinetic data as well  as S02 removal  efficiency  indicating  the enhancement in
efficiency resulting  from  use  of fly ash.  The  S02  removal  efficiency ranged
from 10-50% with only fly ash in the slurry.


                                 INTRODUCTION
     In the last  five  years the use  of  spray dryers to remove sulfur dioxide
from  flue  gas has  moved from  pilot plant  to  full scale  applications.   The
system  consisting  of  a  spray  dryer  and  particulate removal  device  has
generally employed  a Ca(OH)2  slurry which was  atomized in  the drying chamber
via a nozzle or spinning disk atomizer.  It has been observed in these systems
that  the  fly. ash may  also play  an important  role  in  minimizing the reagent
requirements,   particularly  if  recycle is  employed.   The objectives  of  this
study, sponsored  by the  Department of Energy,  were to  1) collect a number of
different  fly ashes  from  various  coal-fired  power plants, 2)  quantify  the
ability of  each  ash to  react with  S02  in a spray  dryer pilot  plant,  and 3)
identify the  parameters  responsible  for  the  reactivity.   To  this  end,  23
different ashes were studied, the results of which are  reported herein.
                                      521

-------
                       DESCRIPTION OF THE TEST FACILITY
     The  spray  dryer/fabric  filter  pilot  plant was  installed  on a  Riley
spreader-stoker coal-fired boiler at  the  University of Tennessee steam plant.
A slipstream  of flue  gas  with a nominal  flow  of 1,000 ACFM,  taken from the
main  ductwork,  entered the  system  which consisted  of a  spray dryer with  a
spinning disc atomizer  followed  by  a fabric filter collector.   The slipstream
was pulled  through  the pilot system by an induced-draft fan.   A layout of the
system is shown in Figure 1.

     The ductwork leading to the spray dryer contained ports for the injection
of  ambient  air and/or  S02.   Other  ports  labeled one, two and three respec-
tively,  enable  the  monitoring of temperature,  static pressure,  and S02 con-
centration  at the inlet and  outlet  of the  spray  dryer find the outlet of the
fabric filter collector, respectively.

     Once the desired inlet S02  concentration and  temperature was achieved the
flue  gas then proceeded  to  the spray  dryer inlet  where  it  encountered the
atomization machinery.   Atomization was  accomplished  by  a Stork-Bowen high
speed  AA-6  Spray Machine and a centrifugal  atomizer.   The water  and fly ash
slurry passed from  the  pumping  system through the  spray machine housing into
the atomizer.  Rotating at a  high  speed (17000-18000  RPM),  the disc atomized
and  distributed  the  slurry  as a  uniform,  fine  mist into  the hot  flue gas
entering the drying  chamber.

     Flue gas exiting  the  boiler entered the drying chamber  through a set of
vanes  concentric with  the spray machine.   These  vanes  imparted  an angular
downward  swirling motion to the air.   The  rotation of the swirling air was
opposite  to that of the  feed leaving  the  atomizer  thus  insuring intimate
mixing of the hot air and the fine particulate mist.   The drying chamber was a
7 foot diameter vessel with a standard conical bottom which contained a rotary
valve for the removal of the spray dryer product.

     The particulate-laden gas stream upon leaving the spray dryer entered the
fabric filter collector.   This  collector removed  both the  suspended fly ash-
S02  product and any other type boiler  ash contained in the flue gas stream.
The bag house consisted of one filtration compartment containing four 32 ft.  x
12 inch fiberglass bags (14.5 oz./sq.  yd.).   A low energy shaker mechanism was
used  to  clean the  bags.   It was operated manually at  the completion  of or
prior to a test.

     The control room contained 1)  a panel for the control  of atomizer rpm and
2)  other devices for  the  monitoring of  the system flow rate and the slurry
feed rate.  Also, monitored in the  control room  were the following:

     1.   S02 concentration (3 ports) Lear-Siegler SM800
     2.   Temperature (thermocouples)
          a.   Inlet/outlet spray dryer
          b.   Inlet bag house
          c.   Ambient temperature
          d.   Slurry feed temperature


                                      522

-------
     3.    Slurry feed rate
     4.    Concentration of additive (i.e.  fly ash percent)
     5.    Static pressure inlet/outlet of spray dryer/bag house

     When the desired range of S02 concentration cannot be achieved,  a manual-
ly controlled  supplemental  S02   injection  system is  used.   This system  was
located about 80 feet  upstream of the spray  dryer  in the inlet duct to allow
for sufficient mixing.   Inlet temperature control was maintained by a manually
operated
port.
ambient air  dilution damper  located upstream  of  the S02  injection
     Monitoring of S02  was  accomplished by extracting  samples  through heated
trace lines into the  control  room where the flue gas sample was drawn through
the  sampling  cell  of  a  Lear-Siegler  SM800.   The  concentration of  S02  was
measured on a wet basis  by this  instrument.   A dry basis  concentration  was
calculated later after the moisture content was determined.

     Slurry preparation in  this  study  consisted of  gravimetrically weighing
and mixing of  a  known quantity of a  fly  ash  into a known  quantity  of water.
The detention  time  was 25-40 minutes prior to  conducting each test.
                           LABORATORY TEST PROCEDURE
     The potential  S02 removal  capability of fly ash  was
twenty-three fly  ashes of  different  rank, i.e.,  lignite,
bituminous eastern and western ashes (see Table 1).
                                                   investigated using
                                                   subbituminous,  and
     Prior to  the actual  testing of each  fly ash,  laboratory analyses were
performed to  quantify both  the  physical and  chemical  characteristics  of the
ash.   Table  2  shows the  physical  and  chemical  properties measured and the
instruments that were used.  Chemical analyses conducted for total alkalinity,
calcium and  sodium content,  pH,  and total  hardness yielded  information re-
garding the behavior  of  each fly ash in a water-based slurry.   These analyses
were conducted at  a  temperature  of 140°F(60°C) and a fly ash concentration of
1.45 pounds  per gallon  of water.   These  conditions were  chosen in order to
simulate the condition to be tested at the spray dryer facility.

     In addition,  laboratory preparation of the fly  ashes,  according to ASTM
procedures (5),  followed by analysis with an atomic absorption spectro photo-
meter determined  the  levels  of  calcium  oxide (CaO),  sodium  oxide (Na20),
magnesium oxide  (MgO),  and  potassium oxide  (K20)  as  received.   Using these
data an estimate of total alkaline metal oxides (TAMO) is obtained as follows:
  Total g-moles Alkaline

  Metal Oxides/lOOg. ash
                   g. CaO/lOOg.  + g. Na^O/lOOg.  +  g. MgO/lOOg.

                  56.1 g./g-mole   62.0 g./g-mole   40.3 g./g-mole

                   q. K?0/100g.
                  94.2 g./g-mole
                                      523

-------
                                                     Table  1.  Origin of the Fly Ashes Investigated
                      Fly Ash
                        I.D.
                        Power Plant
                         of Origin
                                        Type of Fuel
                                        (approximate)
             Source
ui
 6
10
13*
14

ISA

18B
23B
24

27

34
36
37

38B
38C

41

42

43

44

45*

46

49

50
   Belews Creek Steam Station            bituminous
        Bowen Steam Plant                bituminous
       Clay Boswell Unit 4              subbituminous
  Nebraska Public Power Station         subbituminous
    Gerald Gentleman Station
     Pacific Power and Light            subbituminous
          Wyodak Plant
       Hunter Steam Plant                bituminous
           Cherokee #4                   bituminous
      Gallagher Station II               bituminous
    Public Service of Indiana
 Texas Utilities Generating Co.            lignite
            Big Brown
    New Madrid Power Plant #2            bituminous
       Ohio Edison - Gorge               bituminous
       Harrington Station:              subbituminous
 Southwestern Public Service Co.
     Laramie River Station              subbituminous
      Basin Electric Powwer                lignite
      Cooperative - Unit 1
     Milton R. Young Station               lignite
          Center Unit 1
    United Power Association               lignite
  Cooperative Power Association
        Black Hills Power               subbituminous
        and Light Company
        Otter Tail Power                   lignite
    Hoot Lake Station Unit #2
Minnesota Power and Light Company       subbituminous
      Clay Boswell Station
        Monifer Resources               subbituminous
   San Antonio Public Service
    University of Tennessee              bituminous
           Steam Plant
     Marshall Steam Station              bituminous
     Low sulfur eastern coal
        eastern Kentucky
 Big Sky Mine - Col strip Montana
       Black Thunder Mine
    Campbell County, Wyoming
 Wyodak Resources -  Wyodak Mine

Wilberg Mine - Emery County,  Utah
     Colorado western slope
          Amax Aryshlre

        Freestone County

   Southern Illinois, Seam #6
         Ohio strip mine
       Black Thunder Mine
     near Gillette,  Wyoming
          Cordero Mine
      Consolidated Coal Co.
      Stanton, North Dakota
       Bankol-Noonan Mine
      Center, North Dakota
          Falkirk Mine
     Underwood, North Dakota
      Wyodak Mine - Wyoming

   Knife River Coal  Mining Co.
      Beulah, North Dakota
          Big Sky Mine.
        Colstrip, Montana
      Cordero Mine, Wyoming

        eastern Kentucky

     Low sulfur eastern coal
                      *These fly ashes were obtained from the same  source,  however,  their  production  resulted  from  different
                       combustion conditions.

-------
                           Table 2.  Fly Ash Physical and Chemical Characteristics
to
Ln
Fly Ash
I.D.
6
10
13
14
ISA
18B
23B
24
27
34
36
37
38B
38C
41
42
43
44
45
46
49
50
A A M
44M -
MMDfl
Cum)
12.0
14.8
9.4
9.5
15.0
7.8
14.9
12.3
9.5
9.2
15.3
7.4
11.5
15.6
12.0
9.4
18.0
11.0
10.0
8.5
12.9
14.0
C 1 . . -• ~ U
r ly asn
Geometric
Deviation
2.3
2.2
1.9
1.8
1.9
1.9
2.8
2.2
2.5
1.7
2.0
2.5
2.2
2.6
1.8
1.9
2.4
1.9
2.0
1.8
1.9
2.1
.
44 ground in a
Particle
Density
(g/cm3)
2.7
2.1
1.5
2.7
2.5
2.3
2.2
2.7
2.4
2.5
2.6
2.7
2.6
3.2
2.7
2.6
2.5
2.7
2.5
2.5
2.1
2.2
U-»11 m-!!!
oa i I mill
Surface
Area
(m2/g)
1.31
2.14
0.21
2.09
0.73
2.14
2.85
2.88
0.74
1.23
0.22
1.15
1.66
0.76
3.27
0.29
1.03
2.37
3.39
1.48
12.52
1.47


PH
5.2
6.7
11.2
11.0
10.6
11.2
11.3
11.4
10.3
4.4
10.5
11.3
11.2
10.8
10.7
11.4
10.9
10.9
10.7
10.9
3.6
4.8
T f\ Q
iU * y
Alkalinity
mg/1 . as
CaC03
200
500
3800
3500
1200
3400
1800
2900
1700
0
1400
1800
1500
1600
1100
3700
2100
1300
1600
1600
0
0
9i nn

Ca,
mg/1.
472
265
962
132
130
849
264
1151
415
604
245
85
123
274
321
887
179
340
302
128
175
321
3CI9

Na,
mg/1.
40
26
48
299
54
88
51
91
300
300
49
200
54
4200
8000
62
75
8200
58
100
150
52
ocnn
O3UU
Total
Hardness,
mg/1 . as Ca
547
-r
1094
210
160
1000
340
1208
547
717
283
170
245
368
396
981
274
396
396
170
--
358
AT)
*f / £.
       Mass mean diameter and geometric standard deviation  (Coulter Counter Model TAII)


       Helium/air Pyncnometer (Micromeritics Model 1302)
       •*

       "Micromeritics Surface Area Analyzer

-------
This determination  was made based  on  previous  studies in which  it  was shown
that TAMO  was related to sulfur  retention  in  fly ash (Fiedler,  1).   Table 3
contains both the percent  of  each compound in  the  ash  (wet basis)  and the
estimate of TAMO.  Another term used in some of the correlations was  the value
of TAMO less the sulfur in the ash:
This  term was used  to  account for  reaction  that had already occured  in the
as-received ashes.
                      FLY ASH TESTING IN THE SPRAY DRYER


     Fly ash  testing  at  the spray dryer facility was accomplished by mixing a
known mass  of fly ash in a known quantity  of  water with sufficient residence
(mixing) time  (greater than 30  minutes) to allow any  soluble  metal oxides to
go into solution in the slurry.   Figure 2 is a  typical  example  of the dissolu-
tion kinetics  exhibited  by  each fly ash.   This information was used to deter-


                  Table 3.   Fly  Ash  Mineral  Analysis Summary
Fly Ash
I.D.
6
10
13
14
ISA
18B
23B
24
27
34
36
37
38B
38C
41
42
43
44
44M
45
46
49
50
% S
0.43
0.32
0.49
0.73
0.70
0.45
1.87
1.22
0.66
1.04
0.36
0.78
0.72
0.
4.06
1.02
0.66
0.75
0.75
0.68
0.75
2.19
0.41
% CaO
7.79
5.44
13.96
22.91
18.96
9.72
7.88
8.66
14.99
6.76
0.98
24.95
21.81
13.95
18.01
18.96
6.96
23.80
23.80
11.63
28.90
4.02
5.75
% Na20
10.71
9.42
10.96
10.96
10.78
10.89
10.25
9.94
11.19
11.79
12.69
10.98
13.48
35.88
16.68
10.97
10.76
17.71
17.71
10.66
14.94
6.97
12.47
% MgO
0.97
0.91
2.79
3.49
3.79
2.14
1.18
1.91
2.40
0.97
0.39
3.59
2.97
5.58
3.98
3.79
1.00
4.57
4.57
2.71
3.98
0.94
1.44
%K20
2.53
2.36
1.79
1.29
1.50
1.94
1.68
0.76
1.70
2.90
5.56
1.00
1.29
3.69
1.52
1.90
2.49
1.14
1.14
1.45
1.39
1.94
2.78
TAMO
(moles/lOOg. Ash)
0.36
0.30
0.52
0.69
0.62
0.42
0.35
0.35
0.53
0.37
0.29
0.72
0.70
1.01
0.71
0.63
0.35 -
0.84
0.84
0.46
0.87
0.23
0.37
                                      526

-------
 mine the required  duration  of mixing and the effect of slurry preparation and
 temperature.   After  mixing, the  slurry was pumped  to a  final  dilution tank
 where it awaited final transfer into the spray dryer.  Prior to and during the
 slurry production  procedure the following conditions  were  maintained for all
 fly ashes tested at the spray dryer/fabric filter facility:

     1.    Inlet temperature range (300 ± 10°F)
     2.    Inlet S02 concentration (600-800 ppm)
     3.    Gas volumetric flow rate (1,000 acfm)
     4.    Slurry preparation temperature (130-140°F)
     5.    Slurry concentration (1.45 IDS fly ash/gallon H20)
     6.    Approach to saturation in spray dryer (20°F)

     The total  S02  removed by the  system (spray dryer  plus bag  house) was
 determined from inlet  and outlet  concentration  data obtained  from  a Lear
 Siegler  SM800  S02  analyzer.   These  data  were converted  to mass/time and
 reported in  pounds/minute  of  S02  removed as well as  g-moles/sec  of S02.  By
 defining the quantity  of  S02  removed  in this  manner  the existence  of any
 significant correlation with  alkaline metal  oxide content of the ash could be
 identified.

     Table 4 provides  a summary  of the  actual  efficiencies of  S02 removal
 across the  spray  dryer (EFFSD)  and the  system  (EFFSYS) which  included the
 spray dryer  plus  the baghouse.   As can  be seen,  the  system ranged from 0-30%
 for the fly ashes mixed in the slurry.  A ball-milling of fly ash 44  increased
 the  efficiency  from  18% up  to 46%.   It was  interesting  to note  that this
 improvement was a result of improved efficiency in the baghouse.

     Two important checks were conducted periodically  on the system during the
 course of this  investigation  to minimize  the  errors involved in actual test-
 ing.   First,  S02  balances were  conducted  at test conditions  by  measuring
 concentrations at all three ports without any fly ash  injection.   A consistent
 four percent reduction in the level of S02 was observed at the system's outlet
 due to air inleakage.   Secondly,  water  injection  (no  fly ash),  via the spin-
 ning disk atomizer,  resulted  in a two  percent S02  efficiency.  The values in
Table 4  and  following  tables represent values corrected for  air  leakage and
water.
                        RESULTS OF STATISTICAL ANALYSES
PRELIMINARY ANALYSES
     The primary objective  of this study was  to  identify the potential reac-
tivity  of  various types  of  ashes  and  determine their  capability  for  S02
removal  in  the spray  dryer/fabric filter  system.   It was  hoped  that a high
level of correlation could be found between the parameters and the S02 removal
rates in the system.  However, the chemical analyses of the slurries, as shown
in Table 2,  were measured on  the  filtrate from the slurries  and  did not re-
flect  any  reactivity which  might  occur  on the surface  of  the suspended fly
ash.   Figure 3 shows that the actual number of moles of S02 removed per second


                                       527

-------
               Table 4.  S02 Removal Efficiency by a Fly Ash Slurry
Fly Ash
I.D.
6
10
13
14
ISA
18B
23B
24
27
34
36
37
38B
38C
41
42
43
44
45
46
49
50
EFFSD
0
12
7
35
22
26
4
7
3
6
2
27
19
13
19
10
23
17
13
14
13
2
EFFSYS
1
7
17
30
21
23
2
6
5
6
0
26
26
15
21
9
26
18
24
13
29
1
                      44M             17              46
was generally equal to or greater than the calcium plus sodium feed rates into
the spray dryer, (expressed in moles/sec as Ca).

     The ASTM analyses of  total  alkaline metal  oxides  (TAMO-S),  on the other
hand, were  a measure of  the total values  contained in the fly  ashes and as
such over-predicted the amount available for reaction due to the insolubility
of the ashes.  This  is  evident in Figure 4 which shows a graph of S02 removed
in the  system versus TAMO-S.  The amount  of S02 removed was  typically only
one-tenth of the  total  amount  which could  have been  theoretically removed
assuming a complete reaction with TAMO-S.

     A preliminary  statistical  analysis  of sulfur  removed  versus individual
parameters such as TAMO, alkalinity,  % calcium,  % sodium, % calcium & sodium,
hardeness,  hardness  + sodium, and surface  area yielded  correlation coeffi-
cients,  R2,  in  the range of  .03-0.28.   Upon dividing the fly  ashes  into the
three coal  ranks  (lignite,  subbituminous, and bitumminous), the  lignite coal
yielded values  of R2 of  0.61,  0.61,  and 0.85 when  linear  correlations were
conducted of sulfur  retained  in  the  system  versus  slurry   sodium,  slurry
calcium + sodium,  and surface area, respectively.   Thus, further analyses were


                                      528

-------
concentrated on multiple  regressions  in which the fly ashes were separated by
coal rank.   Also  in  an effort to account  for the previous history of the fly
ash it was  decided  to use the parameter total  sulfur (TS) removal (sulfur in
the ash  as  received  plus  sulfur  removed in  the spray  dryer/fabric  filter
system) rather than just the latter.

PREDICTION  OF TOTAL SULFUR RETENTION

     As discussed earlier,  various  sorbents  containing  one  or  all  of  the
alkaline metal  oxide compounds  (CaO,  Na20,  MgO,  and K20)  have demonstrated
success in  the  removal   of  S02.   These  metals  have been  suggested as  the
primary constituents  that occur  in the formation  of sulfate  or  sulfite end
products [Rosenberg,  et.al.  (2)  and Ando,  et.al. (3)].  Davis and Fiedler (4)
used this information to  quantify the ability  of fly ash to retain sulfur in
the boiler  after  the coal was burned.  Figure  4 shows the curve developed by
Davis,  et.al.  (4) with the fly  ashes  used in this study  plotted  in the same
manner.  The ashes  in this investigation exhibited less sulfur retention than
found  in the previous study presumably due to the lack of lignite data in the
previous study.

     It was  this concept  that provided the final basis  for  the modelling of
total   sulfur  retention.   Expressing  the  amount  of  S02  removed by the  spray
dryer/fabric filter  system as grams  of sulfur retained per 100 grams  of ash
and summing  that with  the quantity  of  sulfur already in the  fly ash  as re-
ceived (see Table 2), determined the total sulfur retained.  Table 5 contains
the sulfur retention data for both the fly ash as received and for the fly ash
slurry injected into the system.

     The total  sulfur retention  (the sums of  the two components)  is included
in Table 5  and is referred to as  TSL,  TSSB,  and TSB  for  the  lignite,  subbi-
tuminous,  and bituminous data, respectively.

     Multiple  regression  analyses  were conducted to determine the correlation
and. relationship  between TS  and the three parameters of  surface  area  of fly
ash, slurry alkalinity and TAMO.  The following  equations were found to ade-
quately describe the sulfur  retention  for  the  lignite  and  bituminous  fly
ashes:

    TSL = 0.019 (SAI) + 125.320  (ALKAL) - 0.507 (TAMO) - 1.478     (Eq.  1)

     TSB =0.005  (SAI) +  53.397  (ALKAL) + 9.758 (TAMO) - 3.336     (Eq.  2)

where

     TSL = total  sulfur retained by lignite ash (g/lOOg.  fly ash)

     TSB = total  sulfur retained by bituminous ash (g/lOOg. fly  ash)

     SAI = surface area injected (m2/100g. of fly ash)

     ALKAL = slurry  alkalinity injected (moles/lOOg. of fly ash)
                                      529

-------
                     Table 5.   Sulfur Retention by Fly Ash
Sulfur Retained
Fly Ash by Boiler
I.D. g./100g. fly ash
Lignite
27
38C
41
42
44
44M
Subbituminous
13
14
18A
37
38B
43
45
46
Bituminous
6
10
18B
23B
24
34
36
49
50

0.66
0.00
4.06
1.03
0.75
0.75

0.49
0.73
0.70
0.78
0.72
0.66
0.68
0.75

0.43
0.32
0.45
1.87
1.22
1.04
0.36
2.19
0.41
Sulfur Retained
by SD/BH System
g./lOOg. fly ash

0.35
1.33
2.23
0.88
1.81
4.61

1.44
3.35
2.46
2.76
2.87
2.89
2.48
1.21

0.11
0.53
2.95
0.15
1.05
0.49
0.00
3.12
0.25
Total Sulfur
Retention
g./lOOg. fly ash

1.01
1.33
6.29
1.91
2.56
5.36

1.93
4.08
3.16
3.54
3.59
3.55
3.16
1.96

0.54
0.855
3.40
2.02
2.27
1.53
0.36
5.31
0.66
     TAMO =  total  alkaline  metal  oxides  injected  (moles/lOOg.  of  fly  ash)

     The correlation  coefficients,  R2,  for equation  1 and  2  were  0.90  and
0.95  respectively.    The   R2   for  the  subbituminous  data  was  only  0.23,
suggesting  that   further  analysis  is  needed  to  quantify the  relationship.


DISCUSSION

     The data and results  presented herein  support the concept that reactivity
is controlled  by three factors,  none  of  which  is  totally  independent.   For
example, the TAMO inherently  affe.ts the concentration of calcium  and sodium
ions which are available  to  go into solution.   Likewise,  the ajkalinity is a
measure not  of the  cations,  but  rater the anions (OH  and C03  ).   Thus,  the


                                     530

-------
physical  significance  of the developed equations is somewhat obscurred.   It is
certainly reasonable  that,  as the  alkalinity and/or TAMO  are  increased,  the
potential  for  reaction  in  the  spray dryer  is  enhanced.   Likewise,  as  the
surface  area  is increased,  the  possibilities  of physical  adsorption  and/or
surface  reactions are  increased.  To  achieve a higher  level  of confidence in
the above correlations,  it  is  necessary  to determine the  mechanisms  that are
controlling  the reactivity.

     The  possible effect of surface  area is  best illustrated in  fly  ash 49
which had very  low  values of alkalinity, and  TAMO,  the lowest of  any  of the
ashes.  Yet  it  had the third highest sulfur retention (5.31 g/100 gram).  This
can be attributed to  physical  adsorption due to  the  highest observed surface
area (12.5 m2/g).

     Fly  ash 44 also provides some insight into the mechanisms that are acting
to remove S02-   A light ball milling of this ash resulted in a 17% increase in
surface  area and a  62% increase  in alkalinity.   The  sulfur retention was in-
creased  by 150% due to this action.   It  is  unclear whether the alkalinity or
surface  area enhanced the  removal.   It   is  clear however  that  the effect is
non-linear.

     The  failure to arrive  at  a  suitable" equation  for  the subbituminous coal
(R2=0.23) may be due  in part to the  narrower  range of values of total sulfur
retention for the fly ashes  studied.  The eight values only varied by a factor
of 2 whereas the lignite and bituminous data  varied by factors of  6  and 10
respectively.   The  subbituminous data were  input into equation  1 and  2 to
determine if the measured values fell between the values predicted by the two
extreme coal ranks.   In all  but one  case, the data fell  within or equal to the
values predicted by  equations 1 and  2 within the errors of measurement.


                                  CONCLUSIONS
     The major conclusions of this effort are summarized below:

     1.    It was  demonstrated  that S02  efficiencies  of  up  to 30%  could be
          achieved by  simply mixing  fly ash  in  water followed  by injection
          into a spray  dryer/fabric filter FGD system.  The feed rate of 1.45
          Ibs  per  gallon  is  typical  of  the fly  ash  collected at  a typical
          loading of 3 grains/CF.

     2.    Although of a  preliminary nature, the data suggest that the surface
          area,  slurry  alkalinity  and  total  alkaline  metal   oxides  content
          affect  the  removal.    Equations  were   developed  through  multiple
          regression  analysis  to   illustrate  that these  parameters were of
          value in predicting reactivity.

     3.    The  fly ash  preparation technique (ball mill vs. simple mixing) was
          shown to have a significant effect on reactivity for one lignite fly
          ash.   The removal across  the FGD  system was improved from 18% to 46%
          due to ball milling of the fly ash.


                                      531

-------
     This study indicates that it is possible to develop statistical relation-
ships which  can be  used to determine  directions  to be taken  to  enhance S02
removal  efficiency.  The  alkaline  metal  content and surface area of fly ashes
has  in  many cases  gone  untapped as  a  source of  reactive  reagents.   Further
research is needed to determine the following:

     1.    The optimum fly ash ratio for  S02 removal in a spray dryer.

     2.    Improved  extraction  methods to  extract  alkaline metal  oxides from
          fly ash to increase reaction potential  of the slurry.

     3.    Document the advantages of  grinding  the  fly ash  in  a ball  mill and
          thus, produce corresponding increases in  S02 removal.


                                  REFERENCES
1.    Fiedler, Mark A.   Sulfur Retention in Fly Ash.  Masters  Thesis,  Univer-
     sity of Tennessee,  1980.

2.    Rosenberg,  H.S.  et al.   "The Status  of S02 Control  Systems."  Chemical
     Engineering Progress.  71:5,  May,  1975.

3.    Ando et al.  "Sulfur Dioxide  Removal  from Waste Gases:   A  Status Report
     Japan."   Pollution Engineering and Scientific  Solutions,   Plenum  Press,
     1973.

4.    Davis, W.T., and M.A.  Fiedler.   "The  Retention of Sulfur in Fly Ash from
     Coal-Fired Boilers,"  JAPCA. Vol.  32,  No.  4,  1982,  pp. 395-397.

5.    "Gaseous Fuels;  Coal and  Coke; Atmospheric Analysis." Annual Book of ASTM
     Standards.  Part  26, 1978.
                                ACKNOWLEDGEMENT
     The research conducted  in  this study was funded under a contract with the
Department of Energy located at the Grand Forks Energy Technology Center.  The
contents do not  necessarily  reflect the views of DOE, nor does the mention of
trade names  or  commercial  products  constitute endorsement  or recommendation
for use.

     The work described in this paper was not funded by the U.S.  Environmental
Protection Agency and  therefore the  contents do not  necessarily  reflect the
views of the  Agency  and no official endorsement should be inferred.
                                      532

-------
                                                      Fluegas
Ui
U)
CO
                              Waste
                                                                                                         S02

                                                                                                         P
                                                                                                         T
                                                                                                         W
                PORT  3
                                                                                                    --  OP
                                                                                                        CD
                                                                                                 To Fan
Pure S02 injection
Dilution air  port
Flyash injection
S02/02monStoring port
Static pressure port
Temperature port
Wet bulb port
OrFfice plate
Gas flowrate  control damper
Direction of  gas flow
                                                                            Waste
                               FIGURE  1 I   SCHEMATIC  OF SPRAY DRYER/FABRIC  FILTER  SYSTEM,

-------
U>
t-
<
{£
^
z
ui
tJ

O
u
                      5   10   15   20  25   30  35   40  45



                                     TIME (MIN)
               FIGURE 2;   DISSOLUTION KINETICS OF HOOT LAKE  FLY ASH
                       '   IN DISTILLED WATER AT ?4°c.
                                                                               10
                                                                                 ,-3
s
1
                                                                                10
                                                                                                    1*  M
                                                                                                                               LIGNITE


                                                                                                                               SUBBITUMINOUS


                                                                                                                               BtTUMIMOUS

                                                                                                                                       JL.
                                                                J—L
                  5      lO"*      Z         5      W       2



                    SLURRY CALCIUM + SODIUM (MOLES/SEC AS  CAO)




        FIGURE  3:   S02 REMOVED VERSUS CALCIUM PLUS SODIUM,

-------
Ul
U>
U1
              u
              ui
              CO
              CO
              LU
Q
UJ
i
UJ
ae.
              o
              to
                                                                  LIGNITE
                                                                  SUBBITUMINOUS
                                                                  BITUMINOUS
                                                                  l   i I i  I I I
10"3     2
                                                   '2
                                          5       10'     2
                                     TAMO-S (MOLES/SEC)
                                                                                      >
cc
13

                       FIGURE 4:  S02  REMOVED VERSUS TOTAL AVAILABLE METAL OXIDES
                                  LESS SULFUR (AS CAO),
        FIGURE  5!   SULFUR RETENTION VERSUS  TOTAL ALKALINE
                    METAL CONTENT (REPORTED  AS  CAO),

-------
       FABRIC FILTRATION - AS IT WAS, HAS BEEN, IS NOW AND SHALL BE

                by:  Edward R. Frederick, Technical Director
                     Air Pollution Control Association
                     Pittsburgh, PA 15230
                                  ABSTRACT

     Since the first baghouse patent issued  in 1852,  commercial filtration
technology has progressed significantly with advances in both collector design
and in the performance capability of filter  fiber and fabric.  With the re-
sulting extended and expanded service of fabric filters, critical issues have
evolved concerning electrostatic involvement.   Even though the natural charges
present on gas entrained particles and on the collecting media interact to
play a major role in essentially all filtration operations,  more intense
interest and study is being devoted  to electrical augmentation as a means for
optimizing these features and, thereby, all  collection parameters.  These ob-
servations  have also stimulated further interest in and the production of new
types of electrets, the electrified  fibers that retain charges for extended
service even under adverse conditions.

     Although zinc oxide was first suggested as an absorbant for SC>2 during
the last century, more economic reagents now serve with fabric filters to
control this emission contaminant commercially by dry and wet/dry scrubbing.
Further advances in this technology  through  the addition of special catalysts/
additives and/or with special processing aids, increase S(>2 removal efficiency
and also control NOX effectively.

     Normal use temperature filter fabrics are being improved and even better,
although more exotic high temperature products are available.  Improved pro-
cessing, finishing and treating practices are in use to offer special value
in extending bag life, chemical resistance and cleanability.  The major yet
unheralded problem of nodule formation, deserves special research consideration
in view of the check valve effect that these "dingleberries" have on the fil-
tration process.  Corrective measures for this condition and expansion of
waste heat/product recovery and utilization  will certainly lead to further
expansion of fabric filtration technology for S02 as well as particulate
matter control.
                                      536

-------
                                 INTRODUCTION


     Those of you involved with fabric filtration have had to find the devel-
opments during the recent 8 to 10 years very exciting.  Technology in this
particulate control method has advanced further in this period than during the
preceding decades except, perhaps,at the time when synthetic fibers and changes
in processing methods were introduced during the 40's and 50's.

     The current advancement in the art of fabric filtration, as well as in
other control practices, is due in very large measure to the research support
and publicity provided by the U. S. Environmental Protection Agency through
both in-house and contracted research.  Never before were the air pollution
control methods so thoroughly and extensively examined and so favorably ex-
ploited.

     These and independent R & D efforts in particulate control technology
have led to very favorable predictions.  Market Analysts^ have noted that
"Equipment and media sales in the U. S., totaling $2.7 billion in 1980, were
dominated by particle emissions collectors, which accounted for 23 percent of
the total market.  Electrostatic precipitators and fabric filters comprise
the bulk of this category.

     This dominance is expected to remain through 1995, when particle emis-
sions collectors will total nearly $3 billion, with fabric filters gaining
some market share as the result of advances in fabric technology (especially
for high temperature applications) and innovations such as electrostatically
charging filter fabrics to increase particulate removal efficiency."

                                   HISTORY
     Fabric filtration is as old as the art of textile production; but, in-
dustrial particulate collection in baghouses began only during the middle of
the last century.  Despite this early beginning and more than 130 years of
commercial service, fabric filtration remains an art and will not emerge to
the sophisticated status of a science until more is known about media, par-
ticulates, and their interrelationships.

     The commercial baghouse received stimulus early from the nonferrous
metals industry for two principal reasons:  to recover valuable raw material
and to avoid court actions by neighboring farmers.  It is especially inter-
esting to note that early in this period (late 19th century) of baghouse de-
velopment in a lead smelter operation, H. H. Alexander treated woolen bags
with titanium' chloride to achieve effective acid (sulfuric) resistance.  Even
before this, Sprague applied zinc oxide as an acid neutralizer in smelter
smoke control.  This, whether the advocates of dry scrubbing realize it or not,
was the beginning of what they now advocate as an important new phase of
fabric filtration technology.

     Modern baghouse history began with the installation of massive plant-
built collectors and continued with the transition to "manufactured" bag-


                                      537

-------
houses and the introduction of an array of synthetic fibers.   In between these
events, a major filtration study—the "1929 Investigation" was undertaken.
This is significant as the first extensive field study of filter media,
cleaning methods, particulate variables, and other filtration related para-
meters.  According to Armand Labbe3 who had been responsible  for ASARCO'S
baghouse operations, this commercial scale evaluation demonstrated the sup-
eriority of a fulled, twill weave woolen fabric over a variety of other
natural fiber, smooth surfaced, tightly woven materials;  the  critical impor-
tance of the shaking operation, the advantage of a horizontal shake of suit-
able magnitude, and especially, "that in the collection of two smokes of
identical composition and concentration, one could be filtered at 6 times
the rate of the other."  This difference in performance was ascribed to var-
iations in particle crystallinity and size of the otherwise similar dusts.
Other findings revealed,  as we now know, that the rate of filtration depended
upon the amount and type  of dust collected.  Also noted was the value of low
over high A/C ratios in limiting blinding and for providing often needed
reserve capacity.   While  a primary goal of the investigation  was to find the
key to high ratio operations, the results demonstrated that such could be
realized with substantial initial savings but the end result  was increased
bag wear with shortened bag life, lower efficiency, and no reserve capacity
for unpredictable operational peaks.

     During this same historical period, the pressure versus  vacuum baghouse
contoversy developed.  Despite the use of the fan on the dirty side, Labbe
favored the pressure type for a number of important reasons.

     Subsequently, but much later, studies in our own laboratory demonstrated
that not only did different fibers,  yarns, and fabrics perform differently
but that the more exotic  weaves offered real advantages.   The common plain
weave was found to require tight yarn packing to provide suitable efficiency;
whereas, the same yarn woven in twill or sateen configuration leaked less
dust even at higher permeability.  In the filtration process, particle move-
ment tends to be normal to the fabric surface, passing through visible holes
in plain weave fabrics, but,  not holes visible only at an angle in a twill
or sateen weave medium.  At this same time, inherent fiber properties that
were not normally considered important were found in most cases to have a far
more critical influence on the collection process than fabric construction.
For example,  the electrostatic character of the media eluded  to earlier by
several investigators, was found to be influenced by the same property of the
collected particulate matter.  Electrostatic effects were found to control
the filtration process4.   More will be said about this and similar new devel-
opments .

     Despite repeated attempts to develop universally acceptable high ratio
collectors and the successes realized in some applications, the Labbe guide-
lines of low A/C ratios and good cleanability have been proven to be indis-
pensible for realizing high efficiency and long bag life.  To my knowledge
the 30 to 1 ratios achieved with the Hersey or Hersey type reverse air jet
collectors are the only commercial devices operating successfully at such
levels but only with some dusts and the "right" fabrics.  Likewise, and also
in a number of less critical applications, many pulse jet high ratio collect-
ors are performing well,  generally at less than half this A/C ratio.  Here,


                                      538

-------
too, felt type, depth filtration media with suitable qualities, are required
for effective performance.  The development of, and progressive improvement
made in  these media have led to expanded use of high ratio collectors
whether constructed to. employ tubular or pocket type filters.

    Conventional shaker and reverse air type collectors continue to serve
extensively as the low ratio (up to ^6 cfm/ft2 of cloth) collectors.  Bag
life has been and continues to be a critical operating criteria with 5 years
being about the normal expectance.  Even the fragile glass bags have reached
this level of longevity when optimum conditions prevail but, even much long-
er periods of use have been reported for organically based fiber systems.
The utility industry has adopted baghouse technology to overcome problems
associated with fuels and ash by employing mostly low ratio reverse air type
systems.  By and large the results have been extremely favorable and the
development of novel techniques promises to further expand baghouse usage
here and elsewhere.

                           FABRIC DEVELOPMENTS
    The key to an effective and efficient filtration operation, assuming
the equipment hardware is functioning satisfactorily, is the filter medium.
Fiber suppliers, fabric designers, producers, and finishers alike have
searched desperately for answers by developing new fibers and processes for
fabrics and even for different, better cleaning systems.  As a result, the
applications of fabric filtration have expanded significantly into regions
of higher temperature and conditions of chemical insult never before even
considered possible.
FIBERS


    During the last 40 years an array of new fibers, semisynthetic and
synthetic, have been added to the natural fibers available for fabric fil-
tration.  These materials, and especially those of more recent development,
provide unique .properties and extend performance and operating temperatures
up to and even beyond 500°F.  Actually, more exotic new products extend the
range beyond 1000°F with better chemical resistance.

    The fibers commonly available today cover a wide spectrum of properties
from those offered by the natural materials like wool and cotton to those
given by nylon, polyester, acrylic, polyolefin, aramid, Teflon and glass.
They are prepared in a variety of woven, knitted and needled filter media.
These filter fabrics offer almost unlimited opportunities for conducting the
diverse collection operations needed to recover valuable products, by-
products or waste materials and to  limit harmful or nuisance type emissions.

    Opportunities to vary the base  fiber have also brought about a realiza-
tion among users and suppliers that a given fiber, however desirable other-
wise, is not entirely effective in  collecting different dusts.  This inability
to specify desirable fibers unilaterally, especially in certain critical

                                     539

-------
applications, has led to the acceptance of a gray region in our filtration
theory.  Now, because it is becoming more evident that the electrical re-
lationship between particulate matter and filter media plays a major if not
dominant role in the process, an extension to the theory is evolving.

    Recognizing the critical influence of the nature of the cake formed on
the collecting fabric and the fact that most particulates may be deposited
in a porous structure under the influence of favorable electrostatic con-
ditions, new and/or more rational media prescription practices will evolve.
In the interim, electrical augmentation or articifical charging of particles
and applied electric fields at the fiber surface in pilot scale studies are
indicating clearly the interdependence of these conditions.  Substantially
higher air to cloth ratios, consistently higher efficiencies, better fabric
cleanability, less particle penetration and reduced fabric blinding with
promises of extended bag life are all evident through electrical augmentation.
Couple such developments with the practical experience already gained with
fabric filtration in dry scrubbing and in controlling tarry or corrosive
emissions, and it becomes apparent that this technology has come a very long
way.

    By adding neutralizing, absorbing or adsorbing reagents to the filter
surface or by including appropriate agents with the reactive emissions, the
fabric filter/scrubber process has expanded into critical control operations
of aluminum production, coking, galvanizing and flue gas desulfurization.
Other opportunities in difficult to control processes and for upgrading those
already in use by the ferrous, nonferrous, cement, clay, glass, food pro-
ducts, coal, asphalt, clay, and chemical industries are now more apparent.
Even so, this control technology like all of the others, will become even
more useful as the by-product and energy reduction product recovery and
material utilization processes become more feasible.
                         ECONOMIC CONSIDERATIONS
    An attempt to predict the capital, operating, and maintenance costs of
any operation is risky and always subject to correction.  One issue about
economics, however, seems to be crystal clear even though considerable var-
iation in the predictions occurs here too.  I refer to the cost/benefit
relationship of particulate control.  The analyses carried out by EPA, The
National Wildlife Federation, The National Academy of Sciences, and others
show without question, that air pollution control, for example, pays off
substantially with very significant benefits above costs.  The specific ad-
vantages, costwise, of one control technology over another are quite diffi-
cult to determine without consideration of all specifics.  Nevertheless,
when a choice is possible, a few, very "broad-brush" guidelines are available
from various sources.

    An EPA partially funded elaborate 1978 study of the "Capital and Operat-
ing Costs of Selected Air Pollution Control Systems"5 is available.  Two
years earlier, an extensive review of "The Environmental Control Industry"6
was prepared for CEQ. An attempt is made in these presentations to compare


                                     540

-------
the various costs as well as some other features of fabric filters with elec-
trostatic precipitators and scrubbers.  Quite obviously, it is necessary to
acknowledge the questionable reliability of some methods used to equate costs.

     In general, and this may be the best way to relate the information con-
cisely, it seems evident that fabric filters:

0  are capable of controlling very fine particulate matter at very high ef-
   ficiency,
0  perform at lower cost on a flow throughput basis,
0  operate with less energy input than scrubbers but somewhat more than that
   needed for ESPs,
0  show a trend for lower increases in flange-to-flange prices than indicated
   for the other control methods,
0  show less of an advantage in installation, total operating, and maintenance
   costs than ESPs,
0  are less costly to operate than scrubbers.

     The filter medium is critically important in its influence on baghouse
performance and represents a large part of the capitol, operating, and main-
tenance costs.  In any economic analysis, therefore, a comparison is also
necessary.

     Substantial variations occur between the price for very light weight
filament materials, varying weight woven, spun or combination fabrics, and
heavy needled felts.  These differences and the fact that the different air
to cloth ratios indicate again how difficult and unreliable baghouse cost
generalizations can be.

                DEVELOPING MATERIALS, PROCESSES AND TECHNIQUES

ELECTRETS


     The electrical analog of a magnet, the electret, is a recent commercial
development that also serves to verify the value of electrical involvement
in the fabric collection process.  The Hanson electrostatic filter developed
in the 40's for the war effort in England was an outgrowth of Heaveside's
work in about 1890, Eguchi's in 1920, and Gemants in 1935.  His rosin impreg-
nated wool has since been modified and refined by including acrylic fibers
with wool and by using zinc resinate instead of rosin.

     Within the last decade an entirely new and different approach has been
used.  One product is prepared by imposing an electric  field on heated poly-
olefin film after the drawing/orientation stage of the  process and then
fibrillating the charged film to form durable electrified staple fiber.  Fiber
systems of this type, usually needled, now offer a high level electrostatic
field of long durability even in the presence of moisture and other discharg-
ing environments.  They have found extensive use as efficient respiratory
filters.  As far as I know, electrets have not yet been used extensively in
commercial filtration service.
                                      541

-------
DRY AND WET-DRY SCRUBBING
    Interest in, and commitment .to, powder injection and spray drying for
S02 and particulate control has increased dramatically in view of the claims
for lower costs, dry wastes, lower energy requirements and simplicity as well
as the acceptability of needed lower SOa removal from low sulfur coals.

    Since removal efficiency depends upon many factors such as stoichiometric
ratios, temperature and agent reactivity; studies of this technology are
continuing at an accelerated pace.  For example, considerable effort is being
directed toward improvements  in removal not only of S02 but also of NO^.
More favorable parameters and the use of catalysts', promoters** and other
additives" are being investigated for overall increased control.

NOX and SOz Control
    In addition to conventional current scrubbing applications of fabric
filters for SOa control, recent studies   have  demonstrated the value of
electron beam radiation following the spray dryer for removing NO^ and op-
timizing SOa control.  The high energy electrons  oxidize 862  and NOX to
higher oxidation states which react more readily with excess lime from the
spray dryer or with alkaline fly ash.  The net effect is enhanced S(>2 control
plus highly efficient NOx removal.   The practicability and economics of the
method are being investigated at several laboratories.

    An especially interesting new process offered by the National Bureau of
Standards" employs ozone and propylene to oxidize S02 and NOx f°r easier
removal of the resulting sulfate and nitrate with included ammonia.
ELECTRICAL AUGMENTATION


    Artificial charging, as I prefer to identify the practice of charging
particulate matter and/or applying an electric field to the filter medium,
has received very considerable attention recently at both the bench and
pilot scale level in investigation.  The results have been consistently
favorable.  Lest there be unfavorable results coming from any such study, it
must be made clear that the principle benefit of electrical augmentation
resides in the porous nature of the deposited material that occurs in the
presence of the electric field.  Most types of particulate material respond
in this way, but I believe that a few do not.  Can you imagine the disen-
chantment that would occur by a report of negative data from studies of
electrical augmentation?  Recognizing that not all particulate matter are
created equal, it seems evident that an attempt should also be made to class-
ify dusts according to their ability to form a porous deposit as well as how
they respond to the electrical augmentation process.

    Quite possibly, the most serious issue facing the would-be users of


                                     542

-------
electrical augmentation involves the durability of the circuitry.  The
question Is, "how can an electric field be generated and maintained on
the collecting surface for very extensive periods of use?"  EPA's modifi-
cation of TRI's charging electrodes has converted the outside fine wire
girdle to a more substantial charging cage.  These and future developments
will certainly hasten the commercial adaptation of electrical augmentation
to the filtration operation.

    Gaylord Penney11 has proposed a different approach, one that may offer
a simpler practice.  By insulating all or half of the bags in a collector,
using slightly conductive media (<10  Q/Q) and by applying charges of op-
posite polarity to opposing bags, Penney creates a field at the collecting
surfaces.  He also suggests that it may be advantageous to impose a bi-
polar charge on the particles entering this augmented collector.  But this
approach needs further study.

    Whatever practice is ultimately devised for providing an electrical
field at the filtering surface, it will be apparent that mechanical simplic-
ity and durability will be the keys to its successful commercial utilization.
The incentives for adopting electrical augmentation for collecting appro-
priate particulates are truly significant.  The achievable improvements are
real and substantial.  Energywise, for example, the pressure drop reductions
alone can reduce annual fan energy costs by at least a quarter million dol-
lars for very large collectors.  Add to this the savings in cleaning energy
and bag replacement costs and it will be apparent that electrical effects
deserve serious attention.

NATURAL CHARGES AND ELECTRIC FIELDS

    A significant portion of the improved filtration performance realized
by electrical augmentation or artificial charging is achievable simply by
balancing particulate and fabric charges.  All industrial processes produce
particulate matter with charges, positive and/or negative.  Thus, particles
entering conventional collectors are charged, sometimes far more extensively
than at other times but mostly of mixed polarity.  The type of process gen-
erating particles greatly influences the magnitude  of the charge, with
grinding and other energy intensive operations producing particulate matter
with extremely high levels of charge.

    The electrical field, its polarity and magnitude, that develops on the
filter medium by particle contact is determined by the inherent properties
of the fibers that make up the fabric, by the construction of the fabric, by
the particulate itself, and by operating conditions.  It is possible, there-
fore, under the ideal conditions realized by use of a filter fabric of
favorable electrostatic properties relative to those of the particulate being
collected and, of course, for a particulate capable of alteration, to depos-
it a low air flow resistant cake without electrical augmentation.  In other
words, the natural triboelectrification properties of the medium may be
employed to approach the ultimate level of filtration performance now
achieved by augmentation.

    The opportunities to utilize natural electrostatic effects fully are

                                    543

-------
limited to some extent by the availability  of  appropriate media and more so
by the variability of commercial fabrics.   The most  serious  problem, however,
is the limitation that prevails because of  our restricted knowledge of the
electrical properties of particulates and fabrics.   We  need  to know more
about these characteristics separately and  how they  relate to each other.
Here, then, is an area for further study, the  results of  which could easily
lead to more accurate detection and correction of filtration problems and,
thereby, even more reliance on this technology by reliable prescription.

    In addition to the measurement problems, there are  inherent features of
particulate and media that need to be dealt with. For  example, there tends
to be a variation in fiber qualities, even  among those  of the same or sup-
posedly the same commercial origin.  My studies, for instance, showed very
significant differences in the electrostatic properties of polyester fibers.
Even among those of one type, for example,  Dacron fabrics, I've found tri-
boelectric polarities ranging from the very electropositive  region char-
acteristic on nylon, to the electronegative levels of the polyolefins.  No
explanation has been offered for these very significant differences but
there is considerable evidence, both from experimental  and industrial exper-
ience  to indicate that they are responsible for otherwise unexplained per-
formance variations.

    Other less obscure differences among filter media,  regardless of similar-
ity in fiber make-up, are also evident at times and  these too influence be-
havior.  Consider what happens when a residual amount of the fiber producer's
antistatic finish is allowed to remain on the  fabric, even in a trace amount.
Lowered electrical resistivity occurs and adverse side  effects may be expect-
ed.  Even though the fast charge bleed1 off  features  of  this  antistatic agent
may sometimes be desirable, just as often it tends to be unfavorable since
it reduces the available charge below that  needed to optimize porous cake
formation.  Furthermore, these agents are heat sensitive, fugitive quater-
nary ammonium compounds that degrade to an  amine which conveys electropositive
properties to the fabric before it too burns off.

    These effects and the common problems associated with electrostatic be-
havior have contributed to the uncertainty  and even  doubt about the part
that electrostatics play in the filtration  process.   Nevertheless, filter
media users, producers, suppliers, and researchers alike are all well aware
that electrostatic charges do exist on particulate matter and on the collect-
ing bags.  The electrical augmentation studies have  now rekindled the issue
and seem to make it necessary, to clearly  define the role on natural charges.

    Since some doubt remains, let's adopt  a practice that can be accepted
without bias to provide relevant data.  I  propose that from now on, filter
fabric specifications include two additional parameters.  These are, the
electrical resistivity and triboelectric data.  By offering this information
and by making a comparative analysis of performance  characteristics with
other physical properties of the media, it  should be possible to determine
whether, how, and to what extent the collection process depends upon elec-
trical phenomena.  To be more completely effective in such an analysis,
electrical and electrostatic data should also  be available for the particulate
matter being collected.

                                     544

-------
RESEARCH NEEDS
High Temperature Fabrics

     The growth of the fabric filtration market depends upon current and new
developments in several other important areas.  Better high temperature fab-
rics with more favorable chemical and mechanical properties tend to be on
everybody's wanted list.  While I agree that opportunities exist for media
to be used in the range of 750° to 1000°F, a continuing demand remains for
acid resistant, durable fibers capable of operating in the 300° to 500°
range,  I fail to see a very large market for exotic, high cost fibers al-
though a moderate demand will certainly remain for those super fibers now
being produced.

     For many high temperature processes an alternative to high temperature
filter media is possible within certain limits, even when dewpoint problems
exist.  I refer specifically to the prospects for reducing the temperature
of emissions through improved heat transfer conversion systems.  The more
effective use of conventional heat exchangers and more efficient pyroelec-
trical methods that can make the conversion of heat to other forms of energy
more appealing, would make some moderately high temperature filtration pro-
cesses more viable substitutes for those that are now considered necessary
at the higher temperatures.

     The need for improvements in the currently available commercial high
temperature media will remain.  Superior protective and certain special
custom finishes for glass fabrics, for example, can be expected to increase
performance as well as bag life.  Conceiveably too, aramid fabrics might be
made more acid resistant than permitted by current treating practices.  It
must be appreciated, however, that because the filter media market is some-
what limited, certainly as compared to that for apparel fabrics, the incen-
tives, research budgets and overall efforts to effect improvements in media
will be limited.

Fabric Variations

     Other filter fabric developments deserve mention and appropriate con-
sideration.  Seamless  filter bags made possible by the tubular knitting
technique offer new and different features for a variety of collection oper-
ations.  Further development in the knitting processes, better fabric stabil-
izing methods, improved fibers, yarns, weaving, and/or needling and finishing
and treating practices will surely provide better media for some better fil-
ter service.

      In felts and felt-like materials, a major breakthrough can be expected
with  expanded production of durable and stable scrimless or very low scrim
content needled fabrics.   I doubt that we fully realize the restrictions
imposed by the scrim  component of needled fabrics.   Felts of appropriate
fibers in such new constructions will advance the performance of high ratio
collectors substantially.  No-scrim Nomex felts are  now available.  The real
value of this change  will  be soon be apparent.


                                     545

-------
Filter Cleaning

     More effective cleaning of any filter is a continuing goal of everyone
involved in this technology.  While the introduction of scrimless felts is
one approach, other techniques have been applied and will be used for im-
proving cake arrd plug removal.  The prospect for realizing significantly
better cleaning is not restricted to fabric changes and treatment.  It also
applies to the adoption of improved collection procedures, the rigorous
application of electrostatic principles, and the use of other cleaning as-
sists like those offered by magnetic and acoustic devices and others.

Nodules (Refer to Figure 1)

     For many filtration processes, effective control of nodules would lead
to substantially better overall performance.-  The problem presented by
nodules, although not always evident, was most clearly noted by Penney^
in his description of these so called dingleberries as check valves (note
Figure 2) that restrict air transfer and have a serious influence on the
cleaning process.  The check valves simply swing away from the surface under
the influence of reverse air but close the surface voids during the filtra-
tion cycle.

     Nodule formation, I contend, is in part another electrostatic phenomenon.
The initial accumulation of particles at the ends of fibers is easily ex-
plained in this way.  Conversion of these originally soft deposits to a
hard mass seems to occur as a result of a phase or chemical change with a
cementing action.  Some questions for researchers are, therefore:

     1. What mechanisms account for the accumulation of nodules?
     2. Is moisture alone critical in the hardening process, and in causing
        them to adhere tenaciously to fiber(s)?
     3. What practices may be followed to avoid, eliminate or negate the
        influence of these deposits on the filtration process? and
     4. What type, if any, treatment may be applied to the fibers/fabric
        to control nodules without impairing the filtration process?

Waste Recovery/Utilization

     Last but not least among the ways for the fabric filtration industry as
a whole to improve its usefulness and its image is to promote an energetic
program of waste recovery and/or utilization.  This applies to the utiliza-
tion of waste heat as well as to the recovery of waste or seemingly low
value by-products.  Mountains of fly ash will become another problem unless
the contained minerals and included elements are reclaimed or another use
can be made of the ash itself.  Other, so called waste products are or will
be classified as hazardous and will become a serious liability unless they
too are reclaimed.

     As one example, consider with me the product collected in baghouses on
some electric furnaces.  With a 20% zinc, 10% lead and trace metal content;
reuse of this dust in the furnace was deemed undesirable.  As such, it was
treated as a waste and disposed of, until recently, by burial.  Now in view


                                     546

-------
of the recognized potential hazard of leaching, the burial practice has been
forbidden.  Would it not be safer and more compatible with the conservation
trend to recover the raw materials?  If health effects and economics dictate
policy, this action appears justified, especially in view of the potential
danger of lead and the fact that zinc ore is imported.

     In closing, let me show you how different particulate control methods,
including fabric filtration, have been useful for pollution control in
Pittsburgh.  Figure 3 is a photograph of the city taken in 1945.  The second
was taken in 1956, just eleven years later, from the same location looking
toward the downtown area.  It will be apparent that Pittsburgh was trans-
formed to a clean city during this period from its well established status
as the "Smokey City".

     The work described in this paper was not funded by the U. S. Environ-
mental Protection Agency and therefore the contents do not necessarily re-
flect the views of the Agency and no official endorsement should be inferred.
                                     547

-------
 FIGURE 1.  Nodule  Attached  to  Fiber
                Dust Cake
    Reverse Air Flow
        (Cleaning)
           Dust Cake
                                                  Particulate/Gas
                                                       F10V7
Forward Air Flow
  (Filtering)
  The Check Valve Effect - G. W. Penney, Electrostatic
  Effects in Fabric Filtration — EPA— 600/7-78-142a
  September 1978

FIGURE 2. A Model of Nodule Action

                             548

-------



                                              J*rfu*<:- .... I
FIGURE 3. View of  Downtown Pittsburgh - 1945
 FIGURE A. View of Downtown Pittsburgh - 1956




                              549

-------
                               REFERENCES

 1.   Predicasts,  Inc. News Release, August 1982.

 2.   E.  S.  Godsey, American  Smelting and Refining Company, Inc.,  Salt Lake
     City,  Utal,  Personal Communications, July 1980.

 3.   A.  Labbe,  Consultant, Personal Communication,

 4.'  E.  R.  Frederick, How Dust Filter Selection Depends on Electrostatics,
     Chemical Engineering 68:107, June 26, 1961.

 5.   R.  B.  Neveril, J. U. Price and K. L. Engdahl, Capital and Operating
     Costs  of Selected Air Pollution Control Systems, APCA RS-9 Reprint
     Serie's.  1979 from J. Air Pollution Control Association  28. 1978.

 6.   J.  A.  Klein  and K. C. Leung, The Environmental Control  Industry,
     Allanhold  Osmun & Co., Montclair, N. J. 1976.

 7.   M.  Linne1, J. Klinspor, H. T. Karlsson, and I. Bjerle,  Limestone Based
     Wet-Dry  Scrubbing to Form Gypsum, Chem. Eng. Sci., _5, (37),  807, 1982.

 8.   H.  T.  Karlsson, J. Klingspor, M. Linne1 and I. Bjerle,  Activated Dry
     Scrubbing  of SOa, to be published in the J. Air Pollution Control Assoc.

 9.   H.  T.  Karlsson, J. Klingspor and I. Bjerle, Adsorption  of Hydrochloric
     Acid on Solid Slaked Lime for Flue Gas Clean Up, J. Air Pollution
     Control Assoc. 11, 31, 1981.

10.   P.  L.  Feldman and D. J. Helfritch, Particle Properties  and Feasibility
     of E-Beam  Precharging and R. H. Davis, Agenda Overview  - E-Beam Work-
     shop on The  Combined Removal of S02, NOx and Particulate Matter from
     Stack Gas  by Electron Beam Treatment, Arlington, VA, February 25 and
     26, 1982,  Sponsored by  the U. S. Department of Energy.

11.   G. W.  Penney, Electrical Augmentation of Fabric Filters, U.  S. Patent
     3910779 October  7,  1975 and U.  S. Patent 3966435,  June  29,  1976.

12;   G.  W.  Penney, Electrostatic Effective Fabric Filtration, Volune 1.
     Fields,  Fabrics and Particles,  EPA-600/7-78-142A,  September, 1978.
                                   550

-------
                             AUTHOR INDEX

AUTHOR NAME                                                      PAGE

ADAIR, L	     1-460
ADAMS, R.L	     11-35
ANDO,  T	    11-474
ARMSTRONG, J	   III-241
ARSTIKAITIS, A.A	    11-194
BALL,  C.E	   III-370
BANKS, R.R	    1-37, 1-62
BARRANGER, C.B	     1-132
BAYLIS, A.P	    11-384
BELTRAN, M.R	     11-51
BENSON, S.A	    111-97
BERGMAN, F	.	   III-154
BIESE, R.J	     1-446
BOSCAK, V	    111-66
BRADBURN, K.M	    11-499
BRADLEY, L.H	    11-369
BRINKMANN, A	   III-211
BUCK, V	   III-335
BUMP,  R.L	     11-17
CAPPS, D.D	     1-121
CARR,  R.C.   	     1-148
CHAMBERS, R	    1-226,  1-239
CHANG, R	   III-271
CHEN, F.L	   III-347
CHEN, Y.J	     1-506
CHIANG, T	    11-184
CHRISTENSEN, E.M	    11-243
CHRISTIANSEN, J.V	    11-243
CILIBERTI, D.F	    III-282,  III-318
CLEMENTS, J.S	     11-96
COE,JR, E.L	    11-416
COLE, W.H	     III-l
                                 551

-------
COOK, D.R	     11-349
COWHERD,JR, C	    III-183
OOY, D.W	    III-370
CRYNACK, R.R	       II-l
CUSCINO, T	    III-154
GUSHING, K.M	      1-148
DAHLIN, R.S	      1-192
DABBY, K	     11-499
DAVIS, R.H	      11-96
DAVIS, W.T	      1-521
DAVISON, J.W	    III-166
DELANEY, S	      1-357
DEMEAN, A	     111-66
DENNIS, R	     1-22, 111-81
DIRQO, J.A	111-26, 111-81
DISMDKES, E.B	     11-444
DONOVAN, R.P	1-77,  1-107,  1-316,  1-327, 1-342
DORCHAK, T.P	    III-114
DRENKER, S	     III-271,  III-282
DRIQGERS, G.W	     11-194
DUBARD, J.L	     11-337
DUFFY, M.J	     11-489
DURHAM, M	     11-84,  III-241
EBREY, J.M	     11-349
ENGLEHART, P.J	    III-183
ENSOR, D.S	    III-347
FAULKNER, M.B	     11-204, 11-337
FINNEY, W.C	      11-96
FORTUNE, O.F	1-482, 1-494
FOSTER, J.T	    1-37, 1-91
FREDERICK,  E.R	      1-536
FRISCH, N.W	    III-114
FURLONG, D.A	1-287, 1-342
GARDNER, R.P	   1-77, 1-107
GAWRELUK, G.R	      11-17
GELFAND, P	      11-35
                                 552

-------
GIBBS, J.L	    11-430
GILES, W.B	  111-41, 111-53
GOLAN, L.P	   III-226
GOLDBRUNNER, P.R	    11-401
GOLIGHTLEY, R.M	     1-164
GOOCH, J.P	    11-444
GOODWIN, J.L	   III-226
GRANT, M.A	    111-81
GREEN, G.P	     1-192
GREINER, G.P	1-287, 1-357
GRONBERG, S	   III-141
GRUBB, W.T	   1-62, 1-91, 1-179
HALL, H.J	    11-459
HALOW, J.S	     11-96
HANSON , P	,	     1-460
HARMDN, D	   1-226, 1-239, III-131
HAWKINS, L.A	    11-194
HERCEG, Z	    11-489
HOVIS, L.S	   1-22, 1-77, 1-107, 1-287, 1-316, 1-327,
                                         1-342, 1-357, 111-81, III-347

HOWARD, J.R	     1-164
INGRAM, T.J	     1-446
ISAHAYA, F	    11-154
ITAGAKI, T	    11-322
JACOB, R.0	     1-446
JENSEN, R.M	     1-431
JONES, R	.-	     1-303
KASIK, L.A	    11-430
KETCHUCK, M	     1-482
KINSEY, J	   III-154
KOHL, A.L	   III-300
KUBY, W	   III-271
KUNKA, S	     1-239
KUTEMEYER, P.M	   III-211
LAMB, G.E.R	     1-303
LARSEN, P.S	    11-243
                                 553

-------
LAWLESS, P.A	     11-271
LEE, W	      1-303
LEITH, D	     111-26
LEONARD, G.L	     11-230
LEWIS, M	      1-179
LIPPERT, T.E	III-280, III-318
LOGAR, T.W	     11-184
MARCHAWT,JR, G.H	     11-444
MASON, D.M	    III-256
MASUDA, S	11-139,  11-169, 11-322, III-386
MATSOMDTO, Y	     11-474
MATOLEVICIUS, E.S	    III-226
MXAIN, J.D	    III-198
MOOOLLOR, D.P	     111-97
MCDONALD, J.R	     11-204
MCKENNA, J.D	      1-210
MCLEAN, K.L	     11-489
MENARD, A	      1-255
MILLER, M.L	      1-482
MILLER, R.L	      1-494
MILLER, S.J	     111-97
MITCHNER, M	     11-230
MOSLEM, G.B	11-288, 11-306
MOSLEY, R.B	     11-204
MDYER, R.B	      1-460
MOSGROVE, J	      1-382
MYCOCK, J.C	      1-210
NAKATftNI, H	     11-169
NG, T.S	     11-489
NOVOGORATZ, D	     11-349
OGLESBY, S	     11-534
O'ROORKE, R	    III-318
PEARSON, G	      1-121
PETERS, B.J	      1-179
PIDLLE, W	     11-401
PONTIUS, D.H	      11-65
                                 554

-------
PUDELEK, R.E	     1-521
PUTTICK, D.G	    11-126
QUACH, M.T	     1-506
RAMSEY, G.H	1-316, 1-327
RANADE, M.A	   III-347
REED, G.D	     1-521
REHMAT, A	   III-256
REIDER, J.P	   III-183
REISINGERr A.A	     1-179
RICHARDS, R.M	     1-255
RICHARDSON, J.W	     1-210
RINARD, G	11-84, III-241
ROOP, R.N	     1-460
ROSS, D.R	     1-164
RUGENSTEIN, W.A	,	    11-430
RUQG, D	11-84, III-241
RUSSELL-JONES, A	    11-384
SAIBINI, J	     1-132
SAMUEL, E.A	   1-1, 11-218
SANDELL, M.A	      II-l
SAWYER, J	   III-271
SEARS, D.R	   1-192, 111-97
SELF, S.A	11-230, 11-228, 11-306
SHACKLETON, M	   III-271
SHISHIKUI, Y	    11-139
SMITH, W.B	     1-148
SORENSON, P.H	   III-362
SPARKS, L.E	11-204, 11-271, 11-337
SPENCE, N	     1-132
SPENCER,III, H.W	     1-506
STELMAN, D	   III-300
STOCK, D.E	    11-261
SUHRE, D	   III-335
SUOTER, T.C	      1-48
SURATI, H	     11-51
TACHIBANA, N	    11-474
                                 555

-------
TASSICKER, O.J	III-271,  III-282
THOMPSON, C.S	     111-12
THOMSEN, H.P	     11-243
TOKUNAGA, 0	      11-96
TREXLER, E.C	      11-96
TRILLING, C.A	    III-300
TSAO, K.C	    III-256
VANN BUSH, P	      11-65
VANQSDELL, D.W	1-287,  1-342
WALSH, M.A	      1-482
WEBER, E	     11-111
WELLAN, W.G	      1-420
WEXLER, I.M	     11-521
WHITTLESE?, M	      1-482
WILOOX, K	    III-154
WILLIAMSON, A.D	    III-198
YAMAMDOD, T	    III-241
YEAGER, K.E	     III-xv
                                 556

-------