KERNFORSCHUNGSANLAGE JULICH GmbH
Proceedings of the First Workshop
of Paniculate Control
16th / 17th March 1978
Kernforschungsanlage Julidi GmbH
JUI - Conf - 28
Januar1979
ISSN 0344-5798
-------
\\Oberhausen Dortmund
WUPPERTAL
DUSSELDORF
Remscheid
O
Solingen
Kerpen
Duren \WKoblenz V
AACHEN
Aken
Autobahn
BundesstraBe
Motorway
Main Road
Aix-La-Cnapelle
Schnellzugstrecke Main Railway Line
Nebenstrecke Branch-Line
Flughafen Airport
Kernforschungsanlage Juelich Nuclear
Jtilich Research Center
Als Manuskript gedruckt
Berichte der Kernforschungsanlage Diilich - Diil - Conf - 28
Zu beziehen durch: ZENTRALBIBLIOTHEK der Kernforschungsanlage 3iilich GmbH,
Dulich, Bundesrepublik Deutschland
-------
Proceedings of the First Workshop
of Participate Control
16th / 17th March 1978
Kernforschungsanlage JUIich GmbH
Prepared by
U.S. Environmental Protection Agency,
Office of Research and Development, Washington, D.C. 20460
and
Energy Research Project Management,
Nuclear Research Establishment Jiilich GmbH,
Box 1913, 5170 Jiilich
-------
— 1 —
P R E FACE
The workshop was opened by Dr. Engelmann (member of the KFA,
Board of Directors), who welcomed the participants and briefe-
ly outlined the objectives of the work performed at the KFA.
He pointed out that although this research center is renowned
for its nuclear program, considerable developments are being
made in the non-nuclear field. This description was followed
by an introduction by Dr. Holighaus (Project Management for
Energy Research) to the workshop's theme: particulate control
in coal-fired power stations. He stressed the need, because of
increased use of coal to find solutions for protecting the
environment and public from the pollution.
Dr. Holighaus then asked Dr. Gage, who represented the American
members of the workshop, and Dr. Ziegler, representating the
federal ministry, to give short opening statements.
-------
- 2 -
Welcome - Stephen J. Gage -
I am pleased to have this opportunity to welcome you to
this "Joint Workshop on Particulate Control". It is very
enjoyable to return here to Jiilich which we visited last
July during the initiation of this important exchange.
This workshop is being carried out under the umbrella of
the US-FRG program for cooperation in the field of pollu-
tion control for coal-fired powerplants. I should say that
our mutual experience with these cooperative efforts has
been an outstanding success in the past year. In fact, it
has been a stunning victory over the normal bureaucratic
barriers which typically confront and delay such inter-
national efforts.
On both sides of the Atlantic Ocean, governmental activities
for energy and environmental research is located in several
agencies. This fact alone could have resulted in inaction.
However, through sincere efforts by both sides -- and I should
take special note of the outstanding assistance given by Dr.
Alois Ziegler of BMFT and Dr. Rolf Holighaus of KFA in this
regard -- we have launched information exchange in numerous
areas.
The subject of particulate control is quite important in
the United States at this time. First, the U.S. Congress is
now considering legislation which will greatly accelerate use
of coal, especially by combustion in industrial and utility
boilers. However, the increased use of coal will not result
in degraded air quality in the U.S. since both the President
and Congress have agreed upon stringent air pollution controls.
To carry out this objective, the U.S. Environmental Protection
Agency will soon propose New Source Performance Standards for
Particulate Control in large power plants. Consequently we are
very interested in learning from the experience here in Germany
relating to the control of Particulates.
-------
- 3 -
The fourteen U.S. papers to be presented here present a good
summary of the state-of-the-art of conventional particulate
control, high temperature/high pressure Particulate Control
and particulate measurement Technology in the U.S. As you
will see, these papers will be presented by a combination of
EPA and private sector engineer scientists. This is typical of
the approaches we have to use to conduct our research and deve-
lopment. I notice in the program that you also are involving
researchers from your universities and private companies.
The broad spectrum of interests and experiences should result
in a very productive meeting.
I am looking forward to participation in this conference. I am
also looking forward to housing you in the United States when
we again review progress in this important area of cooperation,
-------
- 4 -
Opening address
to the Workshop on Particulate Control
by Dr. Alois Ziegler,
Federal Ministry for Research and Technology
Ladies and Gentlemen,
I am glad to have the opportunity to be with you during the
opening session of this workshop. My task here is to put this
workshop into the perspective of government-to-government co-
operation. We have to consider this workshop as an element of
cooperation under the Agreement for Cooperation on Environment
Protection Policy between the governments of the United States
and of the Federal Republic of Germany.
EPA and BMFT started their cooperation under the aforesaid
agreement only 11 months ago. In March 77, Dr. Holighaus and
myself had a short discussion with Dr. Stephan Gage at the
ERDA offices about the new programme issue of the Federal Govern-
ment: environmentally acceptable coal-fired power stations.
This meeting was followed by two days of discussion at Bonn
about the possibilities of cooperation between the tow agen-
cies in July 1977- After that meeting several cooperative tasks
could already be executed. This workshop is now another step for-
ward .
I have been asked to say a few words about my position at the
Federal Ministry for Research and Technology. I manage the section
for non-nuclear energy research and technology- We have two
sections with that title. I am responsible for the development
of technologies for prospection, exploitation, mining, prepara-
tion and conversion of primary energy carriers, (this means of
coal, oil, gas), geothermal energy, wind energy, etc.) Solar
energy and the technologies for the distribution, storage and
application of secondary energy carriers are allocated to the
second section.
-------
As I am not an expert on environment protection technologies,
I will say only a few words on the subject of 'Particulate Control'
which you are dealing with and I will not stop you from starting
to work. I will do this in a rather personal way. When I went
to school late in the forties and in the beginning of the fifties,
I was taught that in our contry the Ruhr area, and in your country
the Pittsburgh area, were the most dirty regions of the world. But
when I came to the Ruhr area first in 1969 and to the Pittsburgh
area in 1975, I found in both cases a lovely and rather clean land-
scape. In that context I remember the election slogan of one of our
great political parties in 1958 - I960 : blue sky over the Ruhr area.
What do I want to say with these indications: Particulate control
is not a new subject. During the last twenty years there has been
tremendous progress in the technology for particulate control and
especially in the extended application of this technology. Much
has already been achieved, but some work remains to be done.
The emission of fine particulates especially still gives grounds
for concern with regard to the health of people living in areas
with higher immission rates. I am sure that this workshop will
deal at least to some extent with the problems of fine particulate
control.
I wish you a fruitful exchange of information and views during the
workshop, the initiation of one or the other scientific relationship
and finally a few hours of relaxation and recreation at Jiilich.
-------
- 6 -
THE WORKSHOP CONSISTED OF FIVE SESSIONS
SESSION I: CONTROL TECHNOLOGY PERFORMANCE
DR, STEVEN GAGE, SESSION CHAIRMAN
SESSION II: ADVANCED SYSTEMS FOR DUST REMOVAL
DR, STEVEN GAGE, SESSION CHAIRMAN
SESSION III: CONCURRENT REMOVAL OF DUST AND GASEOUS CONSTITUENTS
DR, ROLF HOLIGHAUS, SESSION CHAIRMAN
SESSION IV: HIGH TEMPERATURE AND PRESSURE PARTICULATE CONTROL
DR, PETER DAVIDS, SESSION CHAIRMAN
SESSION V: MEASUREMENT TECHNOLOGY
DALE L. HARMON, SESSION CHAIRMAN
-------
TABLE OF CONTENTS
SESSION I: Control Technology Performance
"EPA Studies in Fabric Filtration"
12
R.P- Donovan
"A.P.T. Field Evaluation of Fine Particles Scrubbers"
Dr. S. Calvert =•- 33
"Electrostatic Precipitator Performance"
J. Gooch 58
SESSION II: Advanced Systems for Dust Removal
"Electrostatically Augmented Particulate Collection Devices"
D. L. Harmon
1OO
"How to Raise the Efficiency of Dry Elektrostatic Precipi-
tator s by means of Gas Conditioning"
125
Dr. H. Reiftmann
"Improved Design Method for F/C Scrubbing"
Dr. S. Calvert 14°
"Application of High Gradient Magnetic Separation to Fine
Particle Control"
C.H. Gooding 158
"Advanced Dust Collection Techniques in the Federal Republic
of Germany: Selected Examples and Research Priorities"
G. Glithner 175
-------
- 8 -
Page
SESSION III: Concurrent Removal of Dust and Gaseous
Constituents
"SO0 Removal by a Fabric Filter Using Nahcolite Injection"
2 188
R.P. Donovan
"Performance Tests of the Montana Power Company Colstrip
Station Flue Gas Cleaning System"
210
J.D. McCain
"Simultaneous Separation of Dust and Gaseous Constituents at
High Gas Temperature by the Use of Molten Metals and Salts"
Dr. K. Hiibner, Prof. Dr. E. Weber
"Optimization of Wet and Dry Processes for Simultaneous
Removal of Particulates and Gaseous Air Pollution from
Coal Fired Power Stations"
Dr. P. Davids 252
SESSION IV: High Temperature and Pressure Particulate
Removal
"Fundamentals of Particle Collection at High Temperatures and
Pressure"
Dr. S. Calvert, Dr. R. Parker 269
"Granular Bed Filters and Dry Scrubbers"
Dr. R. Parker, Dr. S. Calvert 297
"Application and Efficiency of Dry Elektrostatic Precipitators"
H.-G. Pape, Prof. Dr. E. Weber 324
-------
- 9 -
Page
"Range of Use for Filtering Dust Collectors"
R. Schulz, Prof. Dr. E. Weber 335
"High-Temperature Filtration"
M.A. Shackelton, Dr. D.C. Drehmel 348
"Problems on the Application of Centrifugal Separators,
Especially of Rotary Flow Collectors"
Dr. P. Walzel, Prof. Dr. P. Schmidt 376
SESSION V: Measurement Technology
"Experience with Continuously Recording Dust Measuring
Instruments"
Dr. D.W. Laufhlitte 383
"Continuous Control of the Dust Content in Stack Gases with
Laser Devices"
H. Wiggers, Prof. Dr. E. Weber 395
"Manuel Methods for the Determination of Particulate Concen-
tration, Resistivity and Particle Size Distribution in In-
dustrial Flue Gases"
403
J.D. McCain
"Instrumental Techniques for Sizing Industrial Source Parti-
culate"
W.B. Kuykendal 425
-------
- 10 -
Page
"Evaluation of Particle Size Distribution by Means of
Particle Counters"
C. Helsper, Prof. Dr. H.J. FiSan 448
"A Particulate Sampling System for Pressurized Fluidized
Bed Combustion"
L. Cooper, W. Masters, B. Larkin 463
-------
- 11 -
SESSION I: CONTROL TECHNOLOGY PERFORMANCE
-------
- 12 -
EPA STUDIES IN FABRIC FILTRATION
by
R. P. Donovan
Research Triangle Institute
Energy & Environmental Research Division
Process Engineering Department
P. 0. Box 12194
Research Triangle Park, N. C. 27709
(Draft text for presentation to be made to the Participate Workshop,
Oulich, Federal Republic of Germany, March 16-17, 1978)
March 1978
-------
- 13 -
EPA STUDIES IN FABRIC FILTRATION
R. P. Donovan
Research Triangle Institute
Energy & Environmental Research Division
Process Engineering Department
P. 0. Box 12194
Research Triangle Park, N. C. 27709
The long term goals of EPA's research and development work in
fabric filtration are to ensure that the full potential of fabric
filtration as a particulate control technology is realized in the United
States and that the United States environment receives the full benefit
of the control capability inherent in this technology. To achieve these
goals EPA sponsors basic studies in fabric filtration, hoping thereby to
acquire better process understanding and subsequently more efficient,
lower cost equipment. EPA also sponsors selected applications of fabric
filters (or, more commonly, documentation of applications) so that the
experiences of the early users of this technology in a given application
area will be readily available to the general public and particularly to
other potential users in that same application area. This paper describes
some ongoing or recent EPA projects from each of these two R&D activities.
Most of these projects relate to the removal of particulates from
the flue gas of coal-fired boilers, an application area of high EPA
interest at present.
-------
- 14 -
1.0 BASIC STUDIES: MATHEMATICAL MODELING
Although fabric filters (baghouses) have been used to control in-
dustrial participate emissions for over a century, the details of their
operation remain inadequately understood. Much of today's baghouse
design is still empirical, since mathematical models are not yet able to
predict fabric filtration performance satisfactorily.
EPA now conducts both internal and contractor/grantee research and
development work with the goal of improved understanding of the fabric
filtration process. The hope here is to eventually develop mathematical
models that will be capable of reliable baghouse design. Such models
will take account of variables not generally recognized at present.
Some illustrative examples follow.
1.1 Modeling of Filter Performance
GCA Corporation's (Dennis, et al. [Ref.l]) description of the dis-
continuous, two-state condition of a flyash-coated, woven glass fabric
during reverse cleaning is a basic insight not previously recognized
widely. Their analysis breaks the bag into two parallel subdivisions:
a bag area from which no dust has been removed and a bag area from which
effectively all dust (except a permanent residual dust component) has
been removed (Figure 1). Variation of cleaning times and cleaning
energies simply transfers bag area from one subdivision to the other.
Total gas flow through the filter is viewed as the sum of the flows
through two parallel paths—one through the cleaned area, the other
through the uncleaned area. This concept enables the GCA researchers to
mathematically model the pressure/time characteristics of woven glass
fabrics filtering flyash more accurately than previously, as has been
demonstrated in both laboratory filtration experiments and in field
installations at Sunbury and Nucla [Ref.l]. The complete filtration
modeling problem is far from solved, however, and the GCA model, as it
exists today, still requires numerous empirical inputs for satisfactory
predictions and then only over a narrow range.
-------
- 15 -
1.2 Particle Penetration Through A Fabric
Mechanisms of particle penetration through a fabric filter are not
adequately modeled by the classical single fiber interactions of in-
terception, impaction and diffusion. Defect processes such as non-
uniformities in the fabric (a. quality control problem) and pinhole plugs
in the dust cake (dust cake collapse or bursting in a discrete region)
clearly dominate under certain operating conditions, as shown by EPA-
sponsored work at the Harvard School of Public Health.
Several such penetration mechanisms are schematically illustrated
in Figure 2. Assuming that both the seepage and pinhole plug mechanisms
of particle penetration are nonfractionating mechanisms (straight-
through penetration is a fractionating process) and by use of chemically
tagged flyash, the particle penetration mechanisms through the fabrics
studied at Harvard have been shown to vary with deposit thickness as
illustrated in Figure 3. The key penetration properties of each mechanism
were assumed to be as follows:
1. Straight-through -- The outlet flyash is of the same chemical
composition as the inlet and changes immediately with inlet
changes.
2. Seepage — Only the initially deposited flyash penetrates so
that the chemical composition of the outlet flyash is the same
as that originally deposited and is independent of subsequent
changes in inlet chemical composition.
3. Pjnhole plugs — The outlet chemical composition represents
the cumulation of all previous inlet dust compositions and
changes continually with time, following any change in the
inlet chemical composition.
The Harvard work shows that the size distribution of the outlet
dust is, within experimental error, the same as that of the inlet dust
and hence reinforces the process model wherein such nonfractionating
penetration mechanisms as pinhole plugs and seepage dominate. (The face
velocity of the Harvard investigations was between 5 cm/sec and 15
cm/sec--well above the velocities used in most field applications.)
-------
- 16 -
Methods and operating modes for minimizing particle penetration by
these, at present, non-modeled mechanisms are also part of the EPA
sponsored program at Harvard. For example the Harvard researchers have
found that modifying the valve sequence of their pulse jet so as to
stretch its closing time over several hundred milliseconds reduces
particle penetration (and probably lengthens bag life). The pulse
modification is simply the isolation of the main air pressure source
after the cleaning pulse has been initiated at the bags. The line
pressure therefore decreases after the initial pulse and when the bag
valve closes it does so against a much lower line pressure than in an
unmodified pulse. This simple change produces dramatic reduction in
filter penetration [Ref.3].
1.3 Electrostatic Effects
The importance of electrostatic forces during fabric filtration
continues to be an elusive concept to quantify. Dust/fabric systems
exist for which electrostatic interactions are of first-order importance
and yet general recognition of the need to consider electrical/electro-
static properties during fabric filtration does not yet exist, as
evidenced by the omission of such information from the specifications of
most users and manufacturers of fabric filters.
What evidence exists to indicate that electrical/electrostatic
forces are significant? Work at Carnegie-Mellon [Ref.4] provides some
answers as plotted in Figure 4. This figure shows the collection
efficiency of a clean woven glass fabric,filtering smoke generated by an
electric arc. Three cases are contrasted: externally charged particles
with an applied field; applied field only; and the nonelectrified case
(no external charge, no field). The influence of external charge and
field shown here are typical of Professor Penney's observations on a
variety of dust/fabric systems—just an applied field enhances collection
efficiency and when the dust particles are in addition charged, the
improvement is enhanced further.
Even more important, perhaps, is the concurrent reduction in pressure
drop. Comparisons of the pressure drops between conventional, nonelectrified
-------
- 17 -
operation and that in which the dust passes through a corona charger and
is filtered with an external electric field applied perpendicular to the
filter surface invariably show that the pressure drop under this electri-
fied operation is less than under nonelectrified operation. Professor
Penney attributes this desirable consequence to a clustering of the dust
on the fabric surface brought about by the electric forces. The details
of how this clustering comes about are not as well understood as the
practical advantages of being able to capitalize upon this effect.
Complementing work at Carnegie-Mellon University under the direction
of E. R. Frederick shows that even under nonelectrified operation electro-
static properties can be significant [Ref. 5]. These effects are those
that are observed in operation with no dust charger and no applied
electric field. Charging effects still occur because of triboelectric
interactions between the dust and the fabric (and probably other surfaces
of the baghouse). Frederick was one of the early investigators of
electrostatic interactions [Ref. 6]. His approach has been to attempt
to classify both fabrics and dusts according to their triboelectric
interaction. Two different materials when brought into intimate contact
(or rubbed together to form this contact—the tribo [frictional] aspect
of the interaction) undergo a charge exchange in order to establish
electronic equilibrium across the interface (to equalize the Fermi level
across the interface). When subsequently separated, one material has an
excess of electrons; the other, a deficiency. Consequently, the first
material is charged negatively with respect to the latter and to ground.
The triboelectric series is an attempt to classify materials with respect
to charge exchange with each other. Ranking one material above another
in this series means that it will become positively charged when rubbed
with that other material. A drastically abbreviated version of Frederick's
triboelectric series appears in Figure 5. This series was prepared by
pressing a test fabric against a spinning wheel of a reference fabric
for a short (1-2 seconds) fixed time. Upon separation the electrostatic
voltage is measured by pressing a probe against the test fabric and
reading with a high impedance electrometer.
-------
- 18 -
The scale in Figure 5 is normalized by Frederick and reflects some
average of the magnitude of the electrostatic voltages generated on the
test fabric after rubbing against two different reference fabrics.
Unfortunately the voltages are not additive—the test fabric voltage
after rubbing against reference fabric A does not equal the test fabric
voltage after rubbing against test fabric B plus or minus the electrostatic
voltage between reference fabrics A and B. Other factors such as surface
texture and coatings or contamination complicate this simple picture.
This complexity can also be seen by the varying location on the scale of
nominally identical materials so it is easy to understand why different
researchers often prepare differently ordered triboelectric series.
Indeed any given researcher is hard pressed to reproduce his own ordering.
Consequently the scale is only qualitative and cannot be used to predict
the magnitude of electrostatic voltages between materials. It represents
only a broad, general classification.
Perhaps the points to remember from a triboelectric series are that
certain materials are electropositive (wool, nylon, glass) and other
electronegative (teflon). Electrostatically these materials look
different to a dust and choosing fabrics from different locations of the
triboelectric series could be expected to change the electrostatic
interactions between a given dust and the fabric filter. Photomicrographs
of a dust-laden fabric composed of wool and acrylic fibers shows dust
particles (charged by a corona, but no field) packed densely around the
wool fiber while the acrylic fibers are bare (see cover of J.A.P.C.A.,
Jan. 1976).
Frederick has explored the use of triboelectric classification in
industrial applications [Ref.5]. A brief sampling of results are
reproduced in Table 1. This industrial dust was matched against fabrics
of varying triboelectrical properties. The preferred fabric turned out
to be a high permeability wool, not an especially obvious choice. Its
electropositive triboelectric position conceivably favored the formation
of an effective dust cake which enabled it to outperform the much more
costly teflon fabric.
-------
- 19 -
Variables other than composition, such as surface texture, nap,
finish and coatings, influence electrostatic properties. Accounting for
all these different influences promises to be a long term activity but
evidence is slowly accumulating to suggest that such activity will be
worthwhile.
In pulse jet filtration of room temperature flyash, too, filtration
performance appears to correlate with electrostatic properties [Ref. 7].
Tables 2 and 3 show that certain fabrics treated so as to favor charge
dissipation perform better than identical fabrics not so treated (Table
2); similarly, fabrics treated so as to retard charge dissipation can
fare poorer than those not so treated (Table 3).
1.4 Fiber Geometry
Research at the Textile Research Institute (TRI) shows that trilobal
or rough fibers can produce practical fabric filters with improved
efficiencies at no increase in pressure drop [Ref.8]. After a dust cake is
formed, efficiency increases in the following order of fiber geometries:
bilobal < round < pentalobal < trilobal. Before cake forms, however, no
differences exist. The effect apparently depends upon the interaction between
the dust cake and the fiber structure—perhaps like the electrostatic effects
just discussed. TRI continues research to classify dust cake structure and
to relate it to its critical variables of formation and to measure its in-
fluence upon filtration performance.
-------
- 20 -
2.0 FIELD DOCUMENTATION
Supplementing the basic studies outlined in Section 1.0 are careful
examination and analysis of certain fabric filters in utility and in-
dustrial service. The utility applications best known in the U. S. are
those at Sunbury, Pa. and Nucla, Col. largely because of EPA support of
extensive documentation of these experiences [Ref. 9,10]. The reports
issued [Ref. 9,10] describe the installation specifications, the costs
(both capital and operating), the performance over more than a two year
period (Figure 6, for example) and a log of maintenance and trouble-
shooting. Both Reference 9 and Reference 10 contain many particle size
distribution curves and differential size distribution curves which
compare the properties of the inlet and outlet dusts and lead to fractional
efficiency curves. Many raw data records are also part of the reports,
including the properties of the input coal and, in the Sunbury report,
flue gas analysis.
The impact of EPA support of these activities goes beyond the
issuance of reports. Workers and management at both installations
responded to their "showcase" status by hosting countless visitors and
responding to the many inquiries. Such courtesies were not part of the
EPA program, of course, but reflected recognition of the spirit and
significance of the venture. Because of EPA's interest and the utilities'
pride in their leading edge position, these two installations are
acknowledged pivotal demonstrations of the contribution fabric filters
can make in helping utilities meet environmental standards.
A similar exercise has been performed on a small industrial boiler
at Kerr Industries in Concord, N.C. [Ref. 11]. This activity has resulted
in an expansion of the fabric filter installation to full size and to
its use as a field test site for different fabrics and/or at higher than
standard air-to-cloth ratio.
A similar program is just beginning at Southwest Public Service,
Amarillo, TX. The Harrington Station there consists of three units
totalling 1,050 to 1,100 MW generating capability. Harrington No. 2
boiler, a 350 MW unit, will have a Wheelabrator-Frye (WF) baghouse come
-------
- 21 -
on line during 1978. This boiler burns low sulfur western coal rated at
19.6 MJ/kg (8,425 Btu/lb); its ash content is 5.5 percent; sulfur, 0.3
percent. The WF baghouse consists of 28 compartments capable, all
total, of handling 757.5 m /sec (1,605,000 acfm) of flue gas at 156°C
(313°F) and an air cloth ratio of 1.7 cm/sec (3.4 fpm). Under their
EPA-sponsored test plan Southwest Public Service will monitor and report
all solids and gases entering and leaving the boiler. Five sampling
ports will be used and the particulate information to be recorded includes:
mass, size-distribution, total carbon in the ash, elemental analyses and
sulfate, sulfite and nitrate analyses.
The collection and dissemination of reliable field measurements has
become even more important now that baghouses have become "acceptable"
as particulate control devices for boilers. The region of greatest
growth may well be the Western U.S. where Southwest Public Services and
Nucla are located so operating information from this region is of
special importance.
Initial results from Harrington are expected in about a year and,
coupled with the WF installation at Texas Utilities' Monticello Station,
(a retrofit, independent of EPA sponsorship) should provide valuable
guidance in regard to future research and applications for furthering
the capability of fabric filters in performing the particulate control
function.
-------
- 22 -
REFERENCES
1 Dennis, R. eta]., "Filtration Model for Coal Fly Ash with Glass
Fabrics," EPA^600/7-77-084, (NTIS No. PB 276-489/AS), August 1977,
GCA Corp., Bedford, MA 01730.
2. Leith, D., S. N. Rudnick and M. W. First, "High Velocity, High-
Efficiency Aerosol Filtration," EPA-600/2-76-020, (PB 249-457/AS),
January 1976, Harvard School of Public Health, 665 Huntington Ave.,
Boston, MA 02115.
3. Leith, D., M. W. First and D. D. Gibson, "Effect of Modified
Cleaning Pulses on Pulse Jet Filter Performance" (presentation to
the Third Symposium on Fabric Filters for Particle Collection,
Tucson, Arizona, Decemeber 5-6, 1977).
4. Penney, G. A., "Electrostatic Effects in Fabric Filtration: Vol. 1.
Fields, Fabrics and Particles, An Annotated Data Book", Final
Report, Grant R803020 (in press).
5. Frederick, E. R., "Electrostatic Effects in Fabric Filtration: Vol.
2. Triboelectric Measurements and Bag Performance, An Annotated
Data Book", Final Report, Grant R803020 (in press).
6. Frederick, E. R., "How Dust Filter Selection Depends on Electrostatics,"
Chemical Engineering, June 26, 1961, pp. 107-114.
7. Donovan, R. P., R. L. Ogan and J. H. Turner, "The Influence of
Electrostatically-Induced Cage Voltage Upon Bag Collection Efficiency
During the Pulse-Jet Fabric Filtration of Room Temperature Flyash"
(presentation to the Third Symposium on Fabric Filters for Particle
Collection, Tucson, AZ, December 5-6, 1977).
8. Miller, B., G. Lamb, P. Costanza and J. Craig, "Nonwoven Fabric
Filters for Particulate Removal in Respirable Dust Range", EPA-600/
7-77-115, October 1977, Textile Research Institute, P- 0. Box 625,
Princeton, N. J. 08540.
9. Cass, R. W. and R. M. Bradway, "Fractional Efficiency of a Utility
Boiler Baghouse: Sunbury Steam-Electric Station", EPA-600/2-76-
077b, (NTIS No. PB 253-943/AS), March 1976, GCA Corp., Bedford, MA
01730.
10. Bradway, R. M. and R. W. Cass, "Fractional Efficiency of a Utility
Boiler Baghouse: Nucla Generating Plant", EPA-600/2-75-013a, (NTIS
PB 246-641/AS), August 1975, GCA Corp., Bedford, MA 01730.
11. McKenna, J. D., J. C. Mycock and W. 0. Lipscomb, "Applying Fabric
Filtration to Coal Fired Industrial Boilers, A Pilot Scale Investi-
gation," EPA-650/2-74-058a, (NTIS PB 245-186/AS), August 1975,
Enviro- Systems and Research Inc., P. 0. Box 658, Roanoke, VA
24004.
-------
- 23 -
Figure 1. Partially cleaned, woven glass bag under inside
illumination [Ref.l].
-------
- 24 -
STRAIGHT
THROUGH
1
1
o
1
1
1
o
i
i
•
i
i
O
•
SEEPAGE
• • I
**88&S&89
PINHOLE
PLUGS
Figure 2. Schematic representation of flyash emission mechanisms [Ref.2].
-------
- 25 -
1.0
0.8
O 0.6
CO
CO
2
UJ
to 0.4
o
u.
o
z
2 0.2
o
(T
U_
PINHOLE
PLUGS
STRAIGHT
THROUGH
I
20
DEPOSIT
I
I
40 60
THICKNESS,
MICROMETERS
Figure 3 Fractional flyash emission according to mechanism of penetration
[Ref.2].
-------
- 26 -
100 i-
99.9%
99.6%
o
z
LLI
O
U_
U-
UJ
I-
o
UJ
O
O
90
80
70
60 1— Curve A: No field; no charging
Curve B: Electric field; no charging
Curve C: Electric field; charged dust
VELOCITY THROUGH FABRIC (cm/sec)
Figure 4: Electrostatic augmentation of the collection efficiency of a clean woven
glass fabric filtering smoke from an electric arc [Ref. 4].
-------
- 27 -
• WOOL/NYLON 1 [20%]
• WOOL 1 [80%]
• WOOL/NYLON 2 [80%]
+ 6 NYLON 800 B (REFERENCE)
WOOL 2 [85%]
+ 5 . DACRON 1 [50%]
+ 4 • WOOL 3 [80%]
POLYESTER 1 [90%]
+ 3 . ACRYLIC, ZC [90%]
+ 2 . DACRON 2 [40%]
DRALON T [30%]
+ 1 . GLASS [77%]
POLYESTER 2 [70%]
0 . ACRYLIC, Z [25%]
ORLON [30%]
- 1 . DRALON T [30%]
- 2 . POLYPROPYLENE [50%]
- 3 .
- 4 — DARLAN S546 (REFERENCE)
. TEFLON [0%]
[ ] = RELATIVE CHARGE RATE [LOSS (%) IN 2 MINUTES] AT 50% RH
Figure 5. Frederick's triboelectric series of selected filter fabrics [Ref. 5].
-------
1000
750
CM
E 500
.
-------
TABLE 1
EXPERIMENTAL FILTRATION OF A FERROMOLYBDENUM BY-PRODUCT DUST [Ref 5]
Permeability,..
cm/sec at QA25~~2
(fpm at 0.5")
woven staple
acrylic
(napped surface)
woven staple
acrylic
(TFE finish)
woven filament
polypropylene
woven filament
Teflon
woven staple
wool
0.25
0.11
0.10
0.11
0.26
(50)
(21)
(20)
(21)
(50.5)
Triboelectric
Position
(Frederick Scale)
-1.1
-2.4
-2.7
— 8.0
+ 5.5
Properties
Rate of
Charge Loss
(% in 2 min)
75
75
10
0
85
Collected
mass (g)
16
12
23
12
16
Cake
Pressure drop. Weight
N/cm2 (in. H20) (g)
56
71
40
50
30
(1.1) -76
(1.4) ~ 56
(0.7) 4.5
(0.9) 4.6
(0.6) ~ 45
Relative
Leakage
low
low
very high
i
NJ
very high
i
low
-------
- 30 -
TABLE 2
PERFORMANCE SERIES
WITH EXPERIMENTAL NOMEX FELTS [Ref 7]
RUN
NO.
1
2
FABRIC
CONTROL WITH ANTI-STATIC COATING
(50% RELATIVE HUMIDITY)
CONTROL ALONE
COLLECT.
EFF. (%)
98.7*
97.1
(50% RELATIVE HUMIDITY)
CONTROL ALONE 99.3
(70% RELATIVE HUMIDITY)
CONTROL ALONE 96.8
(50% RELATIVE HUMIDITY)
* Average of two runs, 2 days apart
-------
- 31 -
TABLE 3
FILTRATION PERFORMANCE
OF TEST POLYESTER FELTS [Ref 7]
ELECTRICAL
RESISTANCE
OF MOUNTED COLLECT.
FABRIC TESTED BAGS (ohms) EFF. (%)
STANDARD POLYESTER FELT 108 - 109 99.8
STANDARD POLYESTER FELT
AFTER MULTIPLE SOAKS > 1010 97.2
IN PERCHLOROETHYLENE
-------
- 32 -
Discussion
The discussions which followed were opened by a question raised by
Dr. Guthner as to the difference between electrostatic precipitators
and fabric filters in economic terms. Mr. Donovan replied that in at
least two recent analyses by major utilities total, baghouse costs
(initial plus operating) were found to be cheaper and hence baghouses were
selected as the preferred control technology. Two important points were
raised: 1) the efficiency of an electrostatic precipitator depends on
the coal properties (the flyash from low sulfur coals is more costly to
remove); 2) performance predictability and especially performance
stability seems to be better for the baghouse than for the electrostatic
precipitator when burning low sulfur coal or coals with variable sulfur
content. In reply to Dr. Kastner's query about the cost of baghouse
maintenance work, Mr. Donovan stated that although baghouse maintenance
costs are typically estimated on the assumption that the bags will be
changed at least every two years, field experience shows that four or
five year bag life can be achieved. His comment that the argument for
fabric filters was very strong was seconded by Mr. Princiotta. Dr. Davids
wished to know which filter materials were preferred in coal fired power
stations, upon which Mr. Donovan replied that the temperature requirements
generally dictated the use of fiberglass.
-------
- 33 -
A.P.T. FIELD EVALUATION
OF FINE PARTICLE SCRUBBERS
by
Seymour Calvert
Air Pollution Technology, Inc
4901 Morena Blvd., Suite 402
San Diego, California 92117
USA
-------
- 34 -
INTRODUCTION
The need for more reliable data on the fine particle
collection efficiency of air pollution control scrubbers has
become increasingly apparent as control requirements have
grown more demanding. Design methods, including mathematical
models, have been developed from basic theory plus whatever
good data were available, but to a large extent they were
untested.
To compare predictions with scrubber performances in
different situations one needs to know efficiency as a func-
tion of particle size, commonly called "grade efficiency."
The program reported here was supported by the US Environ
mental Protection Agency over the past 5 years in response to
the need for additional reliable performance data on fine
particle collection efficiency as a function of particle size
for scrubbers operating on representative industrial emission
sources and to reconcile the performance data with existing
mathematical models.
FIELD SAMPLING METHOD
The method of approach to the program objectives involved
a number of experimental determinations to obtain collection
efficiency data, the acquisition of information on system
characteristics and behavior, and computations which utilized
the performance data and mathematical models.
In the beginning of the program the particle size range
of primary interest was from a few tenths to a few microns
diameter, which is within the measurement ranoe of a cascade
-------
- 35 -
impactor. Later the size range was extended downward by an
order of magnitude and it was necessary to use a diffusion
battery in addition to the cascade impactor. The apparatus
and methods used are outlined below:
1. Gas velocity distribution and parameters had to be
measured at the inlet and outlet of the scrubber in order to
define the following:
a. conditions for isokinetic sampling,
b. particle concentration per unit volume of dry gas, and
c. gas flow rate.
2. Particle size distribution and concentration (loading)
in the inlet and outlet of the scrubber were always made with
cascade impactors and sometimes with diffusion batteries. In
some tests, a "pre-cutter" was used to remove either the
heavy particle loading from inlet samples or the entrained
liquid from outlet samples. A cyclone separator with about a
3 ymA cut diameter was first used but a round jet impactor
with about an 8 ymA cut diameter was found to have better
characteristics and was adopted for use for both inlet and
outlet sampling.
Simultaneous inlet and outlet measurements minimize the
effects of particle size distribution changes caused by
fluctuations in the operation parameters. Since the program
objective was to investigate scrubber performance on fine
particles, the sampler was held at one location in the duct
for the duration of each sampling run. This is an adequate
technique for obtaining good samples of particles smaller
-------
- 36 -
than a few microns in diameter because they generally are well
distributed across the duct.
RESULTS
A summary of the scrubbers tested is given in Table I.
The first ten scrubbers in Table I are classified as conven-
tional scrubbers. The last three are novel devices because
they are different in some manner from conventional technology.
The dashed lines in Figures 1 through 10 are experimental
grade penetration curves. They were computed from simul
taneous inlet and outlet particle size/concentration data.
We use the symbol "ymA" for aerodynamic diameter, which is
defined by:
dpa = dp
DISCUSSION
Comparison of the experimental results with mathematical
models was done wherever models were available. Table II is
a list of the design equations taken from the "Scrubber Hand-
book" (Calvert et al. 1972). All of these equations are based
on particle collection by inertial impaction.
The predicted grade penetration curves are represented
by the solid lines in Figures 1 through 10. By comparing the
predicted and the measured grade penetration curves, we can
either verify or reject the design equations. The results
of this comparison are summarized as follow:
-------
- 37 -
1. Valve tray on urea prilling tower We used the particle
collection model for sieve plates because the gas jets
emerging from the slots between the valve cap and the tray
impinge on a liquid froth similar to the round gas
on a sieve plate. The model compared well with the data
after accounting for particle growth due to water vapor
condensation.
2. Vaned centrifugal on KC1 dryer - The collection effi
ciency of this scrubber could not be accounted for
simply by centrifugal deposition caused by the internal
vanes. A gas atomized spray model gave predictions
which agreed with the data.
3. Mobile bed on coal-fired boiler No satisfactory model
is available. Subsequent work in a carefully controlled
pilot plant has yielded much lower efficiency than
obtained on the boiler flue gas scrubber. Particle
growth due to plant conditions is suspected.
4. Venturi on coal-fired boiler The model for a venturi
in terms of particle cut diameter correlated with pres-
sure drop agrees well with the experimental results.
5. Wetted fiber on NaCl dryer Heated impactors were used
in the field sampling. Thus, the measured particle size
is that of a dry particle. In the scrubber the particles
are wet and therefore larger than the dry particles. We
did not measure the wet particle size in the field so
wet particle diameter was calculated from dry particle
diameter with the theoretical prediction that its physi
cal diameter should double. Calculation results are shown
-------
- 38 -
in Figure 5. The fibers in the filter pad ivere ellip-
soidal in shape with the longer axis normal to the direc-
tion of gas flow. Therefore, the collection efficiency
should lie somewhere between those of a ribbon and a
cylinder.
6. Impingement plate on NaCl dryer A model based on impinge
ment from round jets gives good agreement \tfith the data
after allowing for particle growth due to condensation.
7. Venturi rod on cupola The venturi model gives a good
prediction for particles larger than about 1.0 ymA but
does not account for low penetration for the sub-micron
particles.
8. Venturi on asphalt dryer The venturi model prediction
agrees with the performance data.
9. Venturi on borax fusing furnace - The model agrees with
data for particles with diameter larger than 1 ymA. For
particles smaller than 1 ymA diameter, the model predicted
penetration a few percent higher than measured.
10. Variable rod on cupola The model predicted penetration
higher than that measured.
CUT/POWER RELATIONSHIP
When scrubbers are operated at different pressure drops
it is very difficult to evaluate and to compare their perfor-
mances based only on grade penetration curves. IVe have devel
oped a useful correlation called the cut/power relationship
-------
- 39 -
for this purpose and others. The cut/power relationship is a
plot of the cut diameter given by the scrubber against pres-
sure drop or power input, as illustrated in Figure 11. Cut
diameter is the particle diameter whose collection efficiency
is 50%. The solid lines in this graph were calculated theo-
retically from the design equations presented in the "Scrubber
Handbook."
The performance cut diameters determined experimentally
in this program were plotted against measured pressure drop in
Figure 11. (In cases where penetration curves do not reach
50%, the reported cut diameters were equivalent cut diameters
calculated by the method presented by Calvert, 1974). As can
be seen, predictions for impingement and sieve plates agree
with the experimental data. The solid line for venturi scrubber
from the previous correlation slightly overestimates the pressure
drop. The dashed line fits the experimental data determined in
this study and is based on a revised method for predicting
pressure drop (Ref: Yung et al. 1976).
NOVEL DEVICES
Under the novel device test program, we have tested a
CHEAP, an electrostatic scrubber, and a charged droplet scrubber.
Electrostatically Augmented Scrubber
The electrostatic scrubber by Air Pollution Systems is
essentially a venturi scrubber with a particle charging elec-
trode placed ahead of the throat. The unit we tested is a
pilot scale unit with a capacity of about 28 m3/min (1,000
CFM). The experimental data are shown in Figure 12 for the
-------
- 40 -
scrubber system with the charger off and with the charger on.
In a venturi scrubber the most important mechanism respon-
sible for particle collection is by inertial impaction on drops
When particles are charged, electrostatic deposition forces
augment the inertial force and increase the collection effi
ciency of the drop.
Calvert et al. (1973) calculated the theoretical total
collection efficiency due to particle deposition caused by
flux forces plus inertial force. A plot of single drop collec-
tion efficiency against inertial impaction parameter with flux
deposition number, Npn, as the parameter is presented in
Figure 13. Flux deposition number is defined as:
N _ ^F_ _ Particle deposition velocity ^^
FD u~~ Gas velocity past drop
Assuming Stokes' law holds, the particle deposition velocity is
given by:
C' (1 E
- ~ -
F 3 TT
u p
Single drop collection efficiency, n ,, is related to scrub-
ber penetration by an equation given by Calvert for a venturi:
2 QT pT d^ u^^
~fa
Equation (4) with "n^" obtained from Figure 13 was used to pre-
dict the collection efficiency of the A.P.S. scrubber both with
the charger on and with the charger off. Figure 12 shows the
-------
- 41 -
predicted A.P.S. scrubber performance along with experimental
curves. As can be seen the design equations predict a higher
penetration than actually measured in the sub-mircon region.
Wetted Fibrous Bed
The CHEAP system is primarily a wetted fibrous bed scrub
ber system. It consists of water sprays to wet and clean the
filter medium, a rotary drum containing a fibrous "sponge"
filter medium, and a water reservoir.
The CHEAP unit we tested was on a diatomaceous earth
dryer exhaust. The experimental results are shown in Figure
14. The grade penetration curve for wet particles in the
figure was calculated from the penetration curve for dry
particles (containing NaCl) and particle growth data.
We can mathematically represent the filter as an array
of equally spaced cylinders. The "Scrubber Handbook" gave
the following equation for the prediction of particle pene-
tration of a clean fibrous bed:
(Pt)d = exp [ S nf] = exp [- 4 * Q-*> nf] (5)
where "nf" is the effective collection efficiency of a single
fiber in the bed which was assumed to be the collection effi
ciency of an isolated fiber.
The pressure drop across the fiber bed is the sum of the
drag losses of all fibers. We used the drag coefficient for
an isolated cylinder.
-------
- 42 -
AP = 6.5 x 10'4 , (6)
Equations (5) and (6) were used to predict the performance
of the CHEAP. The result Is shown in Figure 15, a cut/power
plot for various fiber diameters. The cut/power relation is
highly dependent on the diameter of the fiber, but not much on
the solidity factor. The circle in the figure represents the
data we determined experimentally. The fiber diameter and
filter porosity were not disclosed to us so we could not check
the model.
It is possible to compare our model with the data of Rei
and Cooper (1976) for tests on a pilot scale unit of the CHEAP.
They reported the volume fraction void of the filter medium to
be 97% and the fiber diameters to be 64 ym, 44 urn, and 36 ym
for foams with 18, 26, and 33 pores per cum. They also reported
the measured cut diameter was somewhere below 0.5 ymA for pres-
sure drops ranging from 40 to 90 cm W.C. The dashed line in
Figure 15 shows their data, which are consistent with our
predictions.
Charged Droplet Scrubber
The charged droplet scrubber by TRW Systems does not
charge the particles but charges the water drops. The water
flows out of small diameter tubes, which also act as elec-
trodes, and is atomized. Particle collection of this scrubber
-------
- 43 -
is due to inertial impaction and the electrostatic deposition.
The 680 m3/min (24,000 CFM) unit we tested was used to control
emissions from a coke oven.
Figure 16 shows the experimental data. We did not compare
a model prediction with data in this case.
CONCLUSIONS
Except for the design equation for the mobile bed scrubber,
design equations for the scrubber types tested in this program
give reasonable performance predictions. In cases where (PtJ,
was lower than predicted it could be accounted for by particle
growth due to water condensation.
The cut/power relationship has many useful applications.
It can be used to compare and evaluate scrubbers, to make pre-
liminary scrubber selections, and to estimate the minimum pres-
sure drop of a scrubber for it to attain the required perfor-
mance. We have verified and extended this relationship in
this study.
ACKNOWLEDGEMENT
The work upon which this paper is based was performed
pursuant to Contracts 68-02-0285, 68-02-1328, 68-02-1496,
and 68-02-1869 with the Environmental Protection Agency.
-------
- 44 -
NOMENCLATURE
CD = drag coefficient, dimensionless
C = particle concentration, g/cm3
C = total particle concentration, g/cm3
pt
C' = Cunningham slip factor, dimensionless
d = collector diameter, cm
d, = drop diameter, cm
d_p = fiber diameter, cm
d, = diameter of perforation, cm
d = particle diameter, ym
d = aerodynamic particle diameter, ymA
pa
d = cut diameter, ymA
P» 5 0
d = mass median particle diameter, ymA
y &
E = field strength, kV/cm
F = foam density, dimensionless
f = empirical constant, dimensionless
f = empirical constant = 0.5
d.
f(d ) = frequency distribution of particles
K = inertial impaction parameter, dimensionless
Kpt = inertial parameter in the venturi throat,
dimensionless
£ = thickness of filter pad, cm
Npp = flux force deposition number, dimensionless
n = number of stages, dimensionless
Pt - overall penetration, fraction or percent
= penetration for particles with diameter d ,
fraction P
-------
- 45 -
NOMENCLATURE (continued]
Q- = gas volumetric £lo\v rate, m3/s
QL = liquid volumetric flow rate, m3/s or £/s
Q = electrical charge carried by the particle,
coulomb
S = solidity factor, dimensionless
TV = gas temperature, °C
Up = particle drift velocity, cm/s
Up = gas velocity, cm/s
Up. = interstitial gas velocity, cm/s
Up = gas velocity in the venturi throat, cm/s
u, = velocity of gas through perforation, cm/s
u = gas velocity, past drop, cm/s
z = static bed height, cm
Greek
e = porosity, fraction
£ = filter pad thickness, cm
n = single drop (ndJ or single fiber (nf) collection
efficiency, fraction
y,, = gas viscosity, poise
Li
pr = gas density, g/cm
3
JG
p, = liquid density, g/cm3
p = particle density, g/cm3
0 = geometric standard deviation
&
Ap = pressure drop, cm W.C.
-------
Table 1. SUMMARY OF SCRUBBERS TESTED
No.
Control Device
1 Valve tray (Koch Flexitray)
2 Vaned centrifugal (Ducon)
3 Mobile bed (UOP/TCA)
4 Venturi (Chemico)
5 Wetted fiber filter
(Encort Corp)
6 Impingement plate
(Sly Impinjet)
7 Venturi rod (Cnviro-
engineering)
8 Venturi (AAF)
9 Venturi (AAF)
10 Variable-rod venturi
(National Dust Collector)
11 CHEAP (Johns-Manville)
12 Electrostatic scrubber
(Air Pollution Systems)
13 Charged droplet
(TRW)
a. Numbers cited in Figure 11,
CL
O./i
D-
\j ""iy *a
Source Am3/min £/m3
prilling
rer
ir
yer
fired
.1
ity boiler
fired
.1
ity boiler
dryer
dryer
Ir
il
in
c
;i
Ir
>m
y cupola
t batch
.t
calciner $
ng furnace
y cupola
.aceous earth
86
465
18,000
13,400
1,590
238
1,274
800
1,200
1,010
710
0.
0.
6.
1.
0.
0.
2.
1.
1.
2.
0.
7
26
6
8
24
15
4
3
7
1
75
G'
°C
In Out
27
196
143
163
38
85
88
149
SO
35
63
17
78
85
54
32
38
65
57
55
20
60
Pres-
sure
Drop
cm W.C.
30
S
30
25
19
30
273
66
110
ISO
48-53
Inlet
Size Pcrfor-
Distri- mancc
bution Cut Dia-
j * ., Pt me'
d , umA O o ,
Pg g % (Ui
1.1 1.5 54 1
>100 - 3 1
3 2.5 6 0
3.8 5 2.7 0
10 4.8 2.7 0
>100 - 10 1
1 2 0.4 0
10 3 0.3 0
1 3 2.5 0
0.75 1.8 1.5 0
0.85 450
ter
nA)
.0
.3
.48
.67
.64
.3
.25
.37
.2
.18
.8
calciner 5 dryer
Ti02 test dust 23
1.8 16 16
Coke oven
680 0.05 120 60
40 1 2.1 S.5 0.35
10 1.4 2.1 88-94 0,35
-------
Table 2. DESIGN EQUATIONS FOR VARIOUS SCRUBBER TYPES
SCRUBBER TYPE
QnoVP £ 71 fl VfllVf^ "t"Y* 3 V
Mobile bed
Venturi and gas
atomized spray
Wetted fiber filter
Impingement
DESIGN EQUATIONS
r ? i d* u,
•04-1 - PYTI -4(1F V Y - P ., ,
itj, exp 'iur A , ^
dP L PJ " 9*Gdh
0.38 < F < 0.65
Pt3s-.xp[-9.5x10.(|J'(uGPG)'- Kp|.J
r _ dpa2 uGi
P 9 y, dc
(2 QT ur PT d,
rrn _ .„_ XL Gt L d FfT £^
[.rLjj cxp r^p., xj
dp [ 5S QG yc
1 F /KPt f + °'7\ 0 49
nrr fl - 1 4. In 1 1+ u.q-y - ff f + 0 71
1 l^pt» rJ i.t J.:L i IT ^APt u./j
Kpt [ \ 0.7 / 0.7 H. Kpt f
(Pt), = exp (-nf s)
P
o _ 4£ (1-e)
TT d£
_/1.37 yG n d^ \ »•»
paso I QG /
-------
- 48 -
REFERENCES
1. Calvert, S. , J. Golclshmid, D. Leith, and D. Mehta. "Wet
Scrubber System Study, Volume I, Scrubber Handbook."
EPA-R2-72-118a (NTIS PB 213-016), August 1972.
2. Calvert, S., J. Goldshmid, D. Leith, and N.C. Jhaveri,
"Feasibility of Flux Force/Condensation Scrubbing for
Fine Particulate Collection," EPA 650/2-73-036
(NTIS PB 227-307), October 1973.
3. Calvert, S., "Engineering Design of Fine Particle Scrubbers,"
J. of A.P.C.A., 24. No. 10, p. 929, October 1974.
4. Yung, S., S. Calvert, and H. Barbarika, "Venturi Scrubber
Performance Model," Final report to EPA, EPA 650/2 - 75-021b
(NTIS PB 271 515) August 1977.
5. Junge, C.E., "Air Chemistry and Radioactivity," Academic
Press, 1963.
6. ?.ei, M.T., and D.W. Cooper, "Laboratory Evaluation o-f the
Cleanable High Efficiency Air Filter (CHEAP)", EPA-600/2-
76-202 (NTIS PB 256-698/AS), July 1976.
-------
1 .0
0. 1
0.01
EXPERIMENTAL
0. 1
1.0
PARTICLE DIAMETER,
I T rrc
10
Figure 1. Predicted and experimental penetrations
for Koch Flexitrax-
1.0
0.5
0.1
0.1
-PREDICTED
-EXPERIMENTAL
I i i
0.5 1.0 5.0
PARTICLE DIAMETER, pmA
10
Figure 2. Predicted and experimental penetrations
for Ducon Multivane scrubber.
-------
1. 0
0.1
If, 0.05
0.01
\
\
PREDICTED
EXPERIMENTAL
I l I I I I I l I
0.1
0.5 1.0 5.0
PARTICLE DIAMETER, ymA
10
1.0
2 o.iL
0.01
o.i
1.0
PARTICLE DIAMETER,
en
O
10
Figure 3. Predicted and experimental penetration
for mobile bed scrubber.
Figure 4. Predicted and c'.xperi:
for Chenuco venturi.
nptrarions
-------
1.0
0.5 H
0.1 r-
o.os
0.01
0.1
O.S 1.0
PARTICLE DIAMETER, piriA
5.0
1.0
0.5
o.i
0.03
PREDICTED
EXPERIMENTAL
CAI.C. FOR WET
PARTICLES
O.OU
0.1
0.5 1.0
PARTICLE DIAMETER,
I
Ul
_I—!—1111
5.0
10
Figure 5. Predicted and experimental penetrations
for Encort wetted fiber scrubber.
Figure.' 6. Predicted and experimental penetration
for Sly Irnpinjct.
-------
1. 0
0.1
0. 01
0.001
PREDICTED
EXPERIMENTAL _
I
0.1 1.0 10
PARTICLE DIAMETER, ymA
Figure 7. Predicted and experimental grade pene-
tration curves for venturi rod scrubber.
1.0
0.1
0.0)
0. 001
PREDICTED
EXPERIMENTAL
Ul
I\J
PARTICLE DIAMETER, ymA
Figure 8. Predicted and experimental penetrations
for AAF Kinepactor 32.
-------
0. 1
0.01
0,001
PREDICTED
EXPERIMENTAL
i I I I t i i
0.3 1.0 3.0
PA1U LCLE DIAMETER, tmiA
0.1
0.05
0.01
w o.oo:
o.ooi
o.i
0.5 1.0
PARTICLE DIAMETER, gmA
-------
i.a
4.0
3.0
2.0
1.0
0.3
0.2
0.1
A Impingement
G]Sieve
CyVenturi
N'o. Scrubber
A Impingement plate
B Sieve, F=0.4,
C Venturi, f=O.S
-• >-"> = >O/ --S
10 O
Note: Numbers on data points correspond to numbers in Table 1.
10 20 30 40 50
PRESSURE DROP, cm W.C.
100
200 300
Figure II. Theoretical and experimental cut diameters as a function of pressure
drop for several scrubber types.
0.1 j—
O.OlI—
0.2
1.0
PARTICLE DIAMETER, pmA
Figure 12. Predicted and.experimental penetrations
for APS electrostatic scrubber.
-------
- C.8
o
0.6
o
§ C.4
0. 2
0.01
0.1 1.0
INERTIAL PARAMETER, K
10
0.1
.H 0.05
0.01
0.3
WET PARTICLE
DRY PARTICLE
1.0
PARTICLE DIAMETER, pmA
1 .0
I
LTi
Figure 14. Experimental grade penetration curve for CHEAT.
Figure 13. Collection efficiency of single drop versus inertia parameter at
Nj^ =9.6 with Np_ as parameter.
-------
1. 0
3.0 i
1 . 0
0.1
O A . !'. T . DATA
RLI AND COOi'FR'S DATA
I
_L
I i
10 3 00
PRKSSimi; DROP, cm K.C.
300
Figure IS. Cut diameter versus pressure drop for
fibrous bed.
0.3 —
0.01L
0.1
I
01
1. 0
PARTICLE DIAMETER, vmA
Figure 16. Experimental penetration curve for charged
droplet scrubber
-------
- 57 -
Discussion
Mr. Giithner drew Dr. Calvert' s attention to three omissions in
table 1: the dust concentration, the solids load and a list of
gas inlet and outlet temperatures. Dr. Calvert explained that
the reason for the first two omissions was that neither the mist
eliminator nor the entrainment seperator were functioning ad-
equately, while the last was missing only owing to lack of space
In reply to Prof. Weber's request for information on measuring
drop diameters and the relationship between these and the theo-
retical values, Dr. Calvert mentioned the work of Richard Ball,
Nuki Yama Tanasawa (whose investigations form the basis of APT's
measurements) and Howard Hesketh, which deals with cloud-type
atomization. Atomization varies according to the length of the
scrubber.
-------
- 58 -
ELECTROSTATIC PRECIPITATOR PERFORMANCE
by
John P. Gooch
Southern Research Institute
2000 Ninth Avenue, South
Birmingham, Alabama 35205
U. S. A.
DRAFT
prepared for
A Particulate Workshop
conducted by
The Federal Republic of Germany
March 16-17, 1978
-------
- 59 -
ACKNOWLEDGMENTS
The principal financial support for the work discussed in this
paper was provided by the Electric Power Research Institute. The
cooperation and financial assistance of the Industrial Environmental
Research Laboratory, Research Triangle Park, N.C., of the Environ-
mental Protection Agency is also gratefully acknowledged.
The field measurements were performed by members of the Envi-
ronmental Engineering Group at Southern Research Institute.
-------
- 60 -
ABSTRACT
Studies of the performance of electrostatic precipitators
were conducted at six coal-fired power plant sites. Overall
collection efficiency and collection efficiency as a function of
particle size were measured with the collecting electrode rappers
energized and deenergized. Chemical analyses were obtained on
samples of coal, fly ash, and flue gas. In situ and/or laboratory
measurements of dust resistivity were performed, and secondary
voltage-current relationships were obtained from the precipitator
transformer-rectifier sets. The measurements of fractional
efficiency with and without electrode rapping indicated that
rapping efficiency losses occur primarily for particle diameters
greater than 2.0 ym diameter. The performance of the electrostatic
precipitators was analyzed using a mathematical model based on the
physical principles of the electrostatic precipitation process.
-------
- 61 -
SECTION I
INTRODUCTION
As emission control requirements become more stringent, the
detailed analysis of existing particulate control device installations
assumes more importance in developing more accurate design techniques and
in improving the performance capabilities of existing designs. This
paper summarizes the techniques employed and the results obtained
from studies of the performance of electrostatic precipitators at
six coal-fired power plant sites.1 The basic objectives of the study
were threefold: (1) determination of fractional and overall collec-
tion efficiency with and without collection electrode rapping for
electrostatic precipitators collecting fly ash under various conditions
(2) utilization of the data in an existing mathematical model of the
electrostatic precipitation process (3) enlargement of the data base
concerning electrostatic precipitator usage on coal-fired boilers.
The test programs were designed to provide the data necessary
to differentiate between previously collected but reentrained
particles resulting from electrode rapping and uncollected particles.
Measurements with and without electrode rapping were performed with
mass trains,inertial impactor systems, an ultrafine extractive
system for particle diameters less than 0.20 ym, and a large particle
sizing system based on a size selective diluter and an optical
particle counter. The installations selected for study were in
relatively good mechanical condition and were characterized by high
-------
- 62 -
collection efficiencies (at least 99%). Boilers using fuels producing
high and low resistivity dusts were selected so that the effect of
widely varying dust properties could be examined. Two hot-side units
were tested to examine the variations in performance that may occur
for precipitators located upstream of the air preheater.
-------
- 63 -
SECTION II
MEASUREMENT TECHNIQUES
Figure 1 illustrates the measurements performed at the six
installations. The major portion of the effort was directed toward
particulate characterization at the precipitator inlet and outlet
using cascade impactors for in situ size determination and mass
trains with in-stack filters for total mass loading measurements.
A point to plane resistivity probe was used for in situ resistivity
measurement at the inlet sampling locations. Since SO concentra-
X.
tions in flue gases influence dust resistivity, emphasis was placed
on obtaining S03-S02 concentrations at the operating conditions of
the precipitators. The accurate determination of SOa concentration
requires extreme care because of the potential interference of the
relatively large concentrations of SO2 which accompany the SO3.
The technique employed for this determination is similar to one
described by Lisle and Sensenbaugh2, and it involves the use of a
condenser maintained below the sulfuric acid dewpoint, but above
the water dewpoint.
Cascade impactor sampling procedures as outlined by Harris3
were followed during the measurement programs to obtain size
distribution data from 0.2 to 10 ym particle diameter. Glass fiber
substrates, which were preconditioned in the flue using filtered
flue gas to minimize weight gains caused by chemical reaction with
the gas, were employed for all outlet impactor runs. Blank impactor
runs were conducted using the preconditioned substrate material
-------
- 64 -
simultaneously with the real runs to determine a correction factor
for weight gain attributable to reaction between the flue gas and
the preconditioned substrate material.
Data reduction procedures for the impactor stage weights consisted
of the following steps:
(1) Stage weights were corrected for blank weight gains.
(2) Cut points for the individual stages for each impactor
were based on calibration studies conducted in the laboratory
using polystyrene latex beads for sizes smaller than 2.0 ym
diameter and ammonium fluorescein particles for particle
diameters from 2.0 to 8.0 ym diameter. Glass fiber
substrates were in place for the calibration studies.
(3) Impactor runs are arranged in groups in an appropriate
manner for the individual test series.
(4) The data are supplied as input to a computer program
which calculates size distributions and fractional efficien-
cies, (see reference 4)
A Thermo-Systems, Inc. Model 3050 Electrical Aerosol Analyzer
(EAA) was used sequentially at single sampling points at the inlet
and outlet sampling location to determine concentration vs size
information in the diameter range of 0.01 to 0.30 ym. The operating
principle of the EAA is based on placing a known charge on the
particles and then precipitating the particles under closely controlled
conditions. Size selectivity is obtained by varying the electric
field in the precipitator section of the instrument. Charged particle
mobility is monotonically related to particle size in the operating
-------
- 65 -
regime of the mobility analyzer. A dilution system is required
because the instrumentation cannot tolerate raw flue gases as sampl-
ing streams nor cope with particle concentrations encountered in
flue gases. A detailed discussion of the dilution system and data
reduction techniques for the mobility analyzer-dilution system is
available in reference 5.
Conventional sampling methods with mass trains and impactor
systems require long integration times which are unsuited for
examining 1 to 5 second duration transient events, such as rapping
puffs, on a real time basis. In order to more clearly define the
mechanisms by which reentrained dust emissions occur, time resolved
data are required on the particulate concentrations and size
distributions across typical portions of precipitator exit planes.
Therefore, a decision was made to construct and employ a real-time
optical particle sizing system to obtain supporting data for the
inertial systems. The optical system consists of a modified ambient
air particle counter and a size selective diluter. The diluter, as
a result of the steep gradient in the fly ash size distributions
on a number basis, dilutes small particles in the sample gas streams
by large factors, but a relatively confined and undiluted stream
containing the lower concentration of large particles is passed
directly to the particle sensor. This system has been designated
the Large Particle Sizing System (LPSS).J
Under ideal sampling conditions, the LPSS would be used as
illustrated in Figure 2, with the aerosol sample extracted through
a vertical probe from below the outlet duct with a single 90° bend
between the sampling point and the particle sensor. In this
-------
- 66 -
configuration, the system could conceivably be calibrated to give
absolute concentrations. For most of the plant locations discussed
in this paper, the ideal configuration was not practical, and a
secondary extractive system as illustrated in Figure 3 was constructed,
This sampling system provided information on relative concentrations
of various particle sizes between and during rapping puffs, but it
did not provide quantitative concentration data because of the
uncertainties in the probe losses and in the degree to which the
secondary sample represented the average concentration in the high
flow rate probe.
The use of inertial sampling systems (mass trains and impactors)
for the measurement of rapping reentrainment requires a sampling
strategy which will differentiate between steady-state particulate
emissions and those which result from electrode rapping. At the
first installation tested in this research program, the strategy
employed consisted of sampling on subsequent days with the rapping
system energized and subsequently deenergized while an attempt was
made to maintain boiler operating parameters as constant as was
practical. The precipitator was characterized by high collection
efficiency (99.9%), which required extended sampling times to obtain
meaningful mass measurements. However- it was found that the
sensitivity of the electrostatic precipitator to changes in resisti-
vity and other process variables could mask the differences in
total emissions caused by energizing and deenergizing the rappers.
In order to minimize this difficulty, a revised sampling strategy
was adopted for the remaining installations. The revised strategy
consisted of sampling with mass trains and impactors dedicated to
designated "rap" and "no-rap" periods. Data with a rapping system
-------
- 67 -
energized and deenergized were obtained by traversing selected
ports with dedicated sampling systems in subsequent periods on the
same day. This procedure, while necessarily distorting the
frequency of the rapping program being examined, minimized the
effects of resistivity and other process variable changes.
The use of the alternating sampling strategy leads to at
least three possible procedures for calculating the fraction of
losses attributable to rapping reentrainment. The first procedure
consists of the calculation of the ratio of emissions obtained
with rappers off to rappers on and subtracting from unity- The
emissions data utilized in this procedure were obtained during the
time in which alternating sampling periods for rap and no rap sampling
trains were employed. The second procedure consists of subtracting
the mass emissions obtained with the rappers deenergized from those
of the previous day with normal rapping, and dividing by the emissions
obtained with the rappers operating normally. The data obtained from
the "rap" period will be approximately equal to that obtained during
other test periods in which the rappers are operating in a normal
fashion if: (1) the distortion of the rapping frequency does not
significantly influence emissions during the "rap" period and (2)
there are no other variations in parameters affecting the precipi-
tator performance.
A third possible procedure consists of the use of a weighted
time average emission during the rap-no rap periods as an approxi-
mation to the normal emission rates, substracting the no rap emission
from the weighted time average, and dividing the difference by the
weighted time average to obtain the fraction of emissions attributable
-------
- 68 -
to rapping. This procedure provides an estimate of rapping reentrain-
ment with the effective intervals which result from the alternating
sampling periods. All of the above calculation procedures are used
when applicable to analyze emissions data from the six installations
tested.
-------
- 69 -
SECTION III
RESULTS
In terms of location in the power plant system and type of
fuel burned in the boiler, the installations studied in this program
may be classified as follows:
Plants 1 and 5 - Cold-side ESP's collecting ash from low-sulfur
Western coals
Plant 6 Hot-side ESP collecting ash from low sulfur
Western coal
Plant 4 Hot-side ESP collecting ash from low sulfur
Eastern coal
Plant 2 and 3 - Cold-side ESP's collecting ash from high sulfur
Eastern coals
Figure 4 through 10 illustrate the configuration and electrical
sectionalization of the precipitator installations. A mechanical
collector precedes the ESP at Plant 1. Table 1 summarizes the
important design parameters and the results obtained for the six
installations. The installations were characterized by relatively
high overall mass efficiency- Rapping losses as a percentage of
total mass emission ranged from over 80% for one of the hot-side
units to 30% for the cold-side units. The high rapping losses at
Plant 4 are probably due both to reduced dust adhesivity at high
temperatures and the relatively short rapping intervals.
-------
- 70 -
Table 2 lists the rapping intervals for each field at the various
installations. Also shown are the effective rapping intervals result-
ing from the alternating sampling schedules which were used to obtain
the rap-no rap data. To the extent allowed by process variations, the
range of emissions attributable to rapping should be established by the
calculations using (rap-no rap) and (normal-no rap) data sets. How-
ever, the time weighted average (TWA) calculation is of interest in
that it indicates the change in rapping emissions caused by the effec-
tive increase in time intervals between raps. With the exception of
the normal current density data set at Plant 2, the time weighted
average calculation gives the lowest percentage emissions due to rap-
ping of the three calculation methods. Table 3 provides typical flue
gas and fly ash compositions obtained at the test sites.
Figure 11 shows the time variations over the test period at
Plant 1 in boiler load, precipitator power, dust resistivity and
relative particle concentrations in two size bands (0.6 to 1.8 ym
and 1.5 to 3 ym). August 5 and 6 were "normal" rapper operation
test periods, whereas August 7 and 8 were "no-rap" test periods. It
is readily apparent that, on August 7, changes in variables other
than rapper energization caused exit particulate concentration
changes which masked the effect of rapping system deenergization.
The LPSS system, however, was able to detect rapping puffs, as
described below.
Figures 12 and 13 show the number of 6-12 and 12-24 ym diameter
particles counted in 10 minute intervals through one day of testing
with rapping and one day of testing without rapping, respectively.
Cyclic concentration variations with a period of one hour were expected
when the rappers were on and are fairly apparent in the data shown
in Figure 12. No such cyclic pattern is apparent in the data shown
-------
- 71 -
in Figure 13 which were obtained with the rappers deenergized. Note
the obvious effect of losing power to one of the T.R. sets. The
average counting rate was much reduced in the 6-12 and 12-24 ym
channels with the rappers turned off as can be seen by comparison
of Figures 12 and 13.
As indicated previously, the attempt to determine rapping losses
at Plant 1 by comparison of mass train and impactor data sets from
normal and no rap periods was not successful due to other factors
influencing outlet emissions. However, an estimate of the contribu-
tion of rapping losses to total mass emission was made from data
from the LPSS and outlet impactor systems. The estimate is that
30% of total outlet mass emission during normal rapper operation
can be attributed to rapping reentrainment. Figure V-10 shows the
rap-no rap data for the EAA system and the rap and no-rap impactor
derived effieiencies. The estimated no-rap efficiencies are based
on the data from the LPSS system and these are subject to large
uncertainties because of the poor counting statistics for the larger
particles coupled with the limited time span over which the data were
taken. Fifty percent confidence intervals are shown for the impactor
and EAA data. Even with the existence of the indicated uncertainties,
it is apparent that very high collection efficiencies are achieved
in the particle diameter range 0.05 to 20.0 ym. The minimum collection
efficiency is approximately 99.2% at 0.20 ym diameter.
The alternating sampling strategy with impactors and mass trains
was successfully employed at Plant 2 and subsequent test sites to
differentiate between reentrainment resulting from rapping and steady-
state emissions. Figure 15 presents rap and no rap data from Plant
2 from the EAA and the impactor sampling system. The large error
-------
- 72 -
bars (50% confidence intervals) on data obtained from the ultrafine
particle system are a reflection of difficulties encountered with
condensation of sulfuric acid, which created an interferring aerosol
in the ultrafine size range. The data were screened and those results
which were felt to be non-representative were discarded. It is
apparent that rapping losses become significant only for particle
diameters larger than 1 to 2 ym. The presence of significant large
particle emissions in the absence of rapping is also indicated by
Figure 15, and was confirmed by data obtained from the LPSS. These
emissions apparently resulted from sparking or voluntary reentrain-
ment. Plant 2 was operating with a high sulfur Eastern coal which
produced a fly ash with low electrical resistivity.
Figure 16 illustrates the large particle losses (on a relative
basis) measured at Plant 4, which is a hot-side installation, using
the impactor and ultrafine sampling systems with the rap-no rap
sampling sequence. The data obtained with normal rapper operation
(not shown) show reasonable agreement for sizes greater than 1.0 ym
diameter, indicating the alternating sampling strategy did not
significantly distort the results obtained. As with the previously
discussed data, the results indicate that rapping reentrainment
does not cause a significant change in fine particle emissions.
Comparisons were made between measured collection efficiencies
as a function of particle size and those obtained from a mathematical
simulation of the precipitators using a model developed at Southern
Research Institute under the sponsorship of the Environmental
Protection Agency.6 The model predicted with reasonable accuracy
the relative changes in fine particle collection efficiency
resulting from resistivity changes and the resultant input power
-------
- 73 -
SECTION IV
CONCLUSION
Measurements of fractional efficiency with and without electrode
rapping at full scale precipitator installations show that rapping
efficiency losses occur primarily for particle diameters greater
than 2.0 ym diameter. The largest rapping losses were measured on
hot-side installations. Mass emission data suggest a correlation,
for .the installations tested in this research program, between
the dust removal rate in the last field of the precipitator and the
emissions due to rapping. The electrostatic precipitator with the
highest overall mass efficiency exhibited a minimum collection
efficiency of 99.2% at 0.20 ym diameter.
Comparisons were made between measured and calculated fractional
and overall collection efficiencies using a theoretical model
augmented by empirical relationships based on the field test data.
The comparisons indicated that the empirical factors improved the
capability of the model to simulate the operation of full-scale
electrostatic precipitators under field conditions.
-------
- 74 -
variations. The comparisons indicated that the theoretically
calculated collection efficiencies in the fine particle size range
were lower than the measured values as a result of certain unmodeled
effects. Significant large particle penetrations resulting from
sporadic events in the absence of rapping which exceeded theoretical
predictions of penetration were also observed. Empirical correction
factors were incorporated into the model to account for the under-
prediction of fine particle collection efficiencies and to increase
the value of the model for design purposes until adequate theoretical
modeling of fine particle collection efficiecies under field
conditions is accomplished.
Rapping reentrainment losses were represented in the model
using an average apparent size distribution of a rapping puff and
an empirical relationship between the dust removal rate in the last
field and emissions attributable to rapping. Figure 11 contains
rapping emissions for the six installations as a function of the
dust calculated to have been removed in the last field of the precipita-
tor. Note the effect of current reduction at Plant 2. These data
suggest a correlation between rapping losses and dust removal rate
in the last field. Data for the two hot-side installations tested
show higher rapping losses, which is consistent with the reduced
dust adhesivity which is expected at higher temperatures. Obviously,
additional data under various conditions are required to determine
if this approach or a variation thereof may be used to estimate
rapping losses under a range of operating conditions.1
Figure 18 illustrates the manner in which the empirical correction
factors and the representation of rapping change the fractional
-------
- 75 -
efficiency prediction of the model for the normal current density test
series at Plant 2. The solid line illustrates the model's prediction of
efficiency as a function of particle size using only theoretical
relationships, the operating parameters for the test conditions,
and the precipitator geometry as input data. The open circles give
the model prediction with the same input data, but with the inclusion
of empirical relationships concerning fine particle collection,
gas velocity non uniformity, gas by-passage, and rapping reentrain-
ment. It is apparent that the agreement between measured and
predicted results under field conditions is improved by the use of
the empirical relationships. In general, it is expected that the
rapping puff correlation will tend to under predict large particle
emission because the correlation does not represent the large particle
penetrations which were observed due to sporadic events other than
rapping.
-------
- 76 -
Section V
REFERENCES
1. Gooch, John P., and Guillaume H. Marchant, Jr. "Electrostatic
Precipitator Rapping Reentrainment and Computer Model Studies."
Final Draft Report by Southern Research Institute to the Electric
Power Research Institute under EPRI Contract RP413-1. August 1977.
2. Lisle, E. S., and J. D. Sensenbaugh. Combustion 36(1), 12 (1965).
3. Harris, D. Bruce. "Procedures for Cascade Impactor Calibration
and Operation in Process Streams." Environmental Protection
Technology Series, EPA-600/2-77-004. January 1977.
4. Johnson, J., G. I. Clinard, L. G. Felix, and J. D. McCain. "A
Computer-Based Cascade Impactor Data Reduction System." U. S.
Environmental Protection Agency, Research Triangle Park, N. C.
February 1978.
5. Smith, W. B., K. M. Gushing, and J. D. McCain. "Procedures
Manual for Electrostatic Precipitator Evaluation." EPA-600/7-
77-059. June 1977.
6. Gooch, John P., J. R. McDonald, and S. Oglesby, Jr. "A Mathematical
Model of Electrostatic Precipitation." EPA-650/2-75-037, U. S.
Environmental Protection Agency, Research Triangle Park, N. C.
1975.
-------
MEASUREMENTS
MASS TRAIN
IMPACTORS
ULTRAFINE SYSTEM
LARGE PARTICLE SYSTEM
V-l CURVES
ACCELEROMETER
IN SITU RESISTIVITY
GAS ANALYSIS
VELOCITY DISTRIBUTION
LEAR SIEGLER
COAL ANALYSIS
ASH ANALYSIS
PLANT 1
A
A
A
A
A
A
A
A
A
A
PLANT 2
A
A
A
A
A
A
A
A
A
A
A
A
PLANT 3
A1
A
A
A
A
A
A
A
A
A
A
A
PLANT 4
A
A
A
A
A2
A
A
A
A
A
PLANT 5
A
A
A
A
A
A
A
A
A3
A
A
A
PLANT 6
A
A
A
A
A
A
A*
A
A
1. MASS TRAIN MEASUREMENT AT OUTLET ONLY
2. V-l CURVES OBTAINED ONE MONTH PRIOR TO EPRI TEST
3. OBTAINED BY SF-CARBORUNDUM BEFORE START UP
4. OBTAINED BY SALT RIVER PROJECT PERSONNEL
Figure 1. Types of Data Obtained
-------
- 78 -
GAS FLOW
SAMPLE
FLOWRATE
MANOMETER
s /
/ <
<
f
/
™*
•••
)
p
PROCESS EX
ROBE LINE
_ HEATER
HEATER
VERTICAL
OPTICAL
HEAD
ELECTRICAL
LEADS, ETC.
BLEED
VALVE
PUMP
DILUTION AIR
MANOMETER
AND ORIFICE
INSTRUMENTATION
Figure 2. Large Particle Sizing System
-------
- 79 -
BLOWER
EXHAUST
FLOW
REGULATOR
DILUTER
AND COUNTER
JXL
?v>
DUCT TOP
-EXTRACTION PROBE
GAS
'FLOW
Figure 3. Extractive Sampling System
-------
GAS
FLOW
TR« CIL
TR*C7R
(14.5ft)
4.42m
TR**C2L
TR*C8R
(14.5ft)
4.42m
TR*C3L
TR*C9R
(17.5 ft)
5.33 m
TR*C4L
TR*CIOR
(17.5 ft)
5.33 m
TR*C5L
TR*CIIR
(17.5ft)
5. 33m
TR*C6L
TR*CI2R
(17.5ft)
5.33m
J
(60
60
(12
0.2
60 GAS PASSAGES AT
oo
O
DISTANCE BETWEEN EACH FIELD - (2.5 ft) 0.762 m
COLLECTING PLATES IN I AND 2 FIELDS ARE (12 ft) 3.66 m DEEP
FIELDS 3 THRU 6 ARE (15 ft)4.57 m DEEP
ALL COLLECTION PLATES ARE ( 40 ft) 12 .19 m HIGH
Figure 4. Precipitator layout for Plant 1
-------
- 81 -
t
<*>
oooooooooo
COLLECTOR B
LEARSIEGLER
PORT
7\
O O O O O O O
24.28m
(79'-8")
-------
INLET GAS \
DISTRIBUTION)-— r
SCREEN // fiJ ?
/'
1
1
GAS
FLOW*
i
1
//INLET \
£-( SAMPLING) —
\\PORTS /
m n n ^^d^ '
1
1
GAS
FLOW*
1
V
/
"A" Bl
/TF
i Al
"B" Bl
JSHING
M
3v
JSHING
i!
"A" Bl
/TF
\AI
"B" Bl
JSHING
o
32^
JSHING
"A" Bl
(I
"B" Bl
JSHING
R\
3y
JSHING
"A" BUSHING
/TR\
vS/
"B" BUSHING
!l-»
___— *-c
OUTLET GAS \
DISTRIBUTION) — 7
SCREEN / /
"A" Bl
u
"B" Bl
JSHING
"\
4/
JSHING
GAS FLOW ^
"A" B
(j
"B" B
JSHING
FT\
J4/
USHING
!/
GAS
FLOW
DOWN
A
I
CD
RAPPING MOTORS-
Figure 6. Plant 3 precipitator layout
-------
CHAMBER A-1 CHAMBER B-1
V-
CHAMBER B-2
J t T/R \ V
-rfcM V-oo-
tf/7\,W
•«/™\
SECTION 4
SECTION 3
SECTION 2
SECTION 1
GAS FLOW
0 A
GAS FLOW
f>
I
oo
GAS FLOW
Figure 7. Plant 4 precipitator configuration
-------
- 84
GAS
FLOW
CHAMBER
NUMBER
INLET SAMPLING
LOCATIONS
GAS
FLOW
TI n n n r\
ii i' H "i !! v
*\^,
I ! ' l| ' >
i ii n | !«
l/r^M/^\ll/r^l ,
I
ii
1 !! i' !l I'
i@|i©i;©i|@
i h 'i ii ii
i@ii@'i@ii@ii@
i h h
! '.
i i.
@||©H®'l@l|(g)
i1 'i 'l H
l! !' i, i!
OUTLET SAMPLING
LOCATIONS
TRANSFORMER RECTIFIER
Figure 8. Plant 5 precipitator layout
-------
- 85 -
GUILLOTINE
DAMPER
SAMPLING PORTS
SAMPLING PORTS- H -n
SAMPLING PORT
SAMPLING PORTS
Figure 9. Ductwork arrangement for Plant 6
-------
UPPER
/DD 4
U U U U U
UA
CD
I-1
O
O
PI
H
(D
3
0)
3
rt
Hi
O
1-1
P)
3
rt
c
D
E
F
G
H
t^e; Pa=;c;
1 + 35
arrps
1 ^ 35
1 + 35
1 + 35
1 -> 35
1 ^ 35
Gas
1 -> 35
Passage
1 ^ 35
c
D
E
F
G
H
I
00
^
II | II | M t
SOUTH
NORTH
16 15 14 13 .12 11 10
; u u u u u u u
C
D
E
F
G
H
Has Pass
i ->• 35
aaes
1 -> 35
1 -> 35
1 -> 35
1 ->- 35
1 ^ 35
1 -> 35
Gas Pas
1 -»• 35
C
D
E
F
ages G
H
LOWER
-------
Table 1. SUMMARY OF RESULTS FROM EPRI TESTS
Plant 1
Number of Electrical 6
Fields in Direction
of Gas Flow
Plate-to-Plate 30.48
Spacing, cm
27.94
25.4
22.86
24.76
22.86
Mast with Mast with Rigid Barbed Hanging Round Electrode Frame Hanging Round
Emitting Electrode
Design
Rapper Design
Portion of ESP
Tested
Boiler Load During
Test, MW
Gas Flow During
Test, am3/sec
Temperature During
Test, °C
SCA During Test,
m2/(m3/sec)
Measured Efficiency,%
Dust Resistivity at
Operating Temp,n-cm
% of Mass Emissions 31 65-33 30
Attributed to
Rapping
^Indicating range of values from two methods of calculation.
Laboratory measurement.
Square
Twisted
Wires
Drop
Hammer
Total
128
330.2
152.2
113.5
99.92
1.4X1011
Square
Twisted
Wires
Drop
Hammer
1/2
160
155.2
155
47.6
99.55
1.7xl010
Wires
Tumbling
Hammers
1/2
122
117.2
157.2
50.4
99. 80
2xlOio
Wires
Magnetic Drop
Hammer
1/2
271
203.9
321.1
76.8
99.64
3.2xl010
With Spiral
Wires
Tumbling
Hammers
1/6
508
149.4
106.1
117.9
99.85
4.6X1011
Wires
Magnetic Impulse
Hammers
1/16
800
126.8
358.9
55.4
98.98
1.5xl09b
1
00
-0
1
85
36-29
63-44
-------
PENETRATION-EFFICIENCY
100 c
PLANT 2 - EFFICIENCY - RAP 1-14-76.1-15-76
RHO = 2.40
• EXPERIMENTAL
O THEORETICAL RAP + NO-RAP,
ag = 0.25, s = 0.10
— COMPUTED FRACTIONAL EFFICIENCY FOR
NORMAL CURRENT DENSITY-TEST
(THEORETICAL ONLY)
. . . i . ...I
I 1 I I I I ill99.99
0.01
1.0 10.0
PARTICLE DIAMETER, micrometers
100
Figure 18. Comparison of measured and computed efficiencies
from Plant 2 normal current density series.
-------
- 89 -
GROSS LOAD
MW 120
I001—
zo
PWR / TYP. ' 8
SECTION, kW !6
14
155
GAS TEMP I5°
°C 145
140
4
RESISTIVITY 3
10" , ohm-cm
1
400
300
COUNT RATE 20Q
no. /sec
(1.5-3 um ) 100
PARTICLES
0
2000
COUNT RATE
no. /sec 1000
( 0.55- 1.8 urn )
PARTICLES ,
\ o °"\ ° o o /1/f -
- o/0 \ \/\oJ |/° —
L V ° ° -
A
-JSfA>» 1 *S ^ 1^^ 4/A~
/ A^ ~
o
o
0
° 0 0 0 ° 0 -
1 0 ° 1
— 1
- K ~
~ i °°° r°' ^ ~
/ D-0 /tirDa rfP D
~ ^^O| ^D jin D ~
n
-ffl 'I Da ~
- 1 1 1 1 1 1 II 1^° 1 '
3 12 48 8 12 48 8 12 48 8 12 48
1 — AUG 5 — ' 1 AUG 6 — ' ' AUG 7 — ' ' AUG 8 — '
Figure 11. Events for test period Plant 1
-------
- 90 -
7.0
• 6-12 ym
O 12-24 pm
9 = 00 10 = 00 11=00 12=00
1 = 00 2=00 3=00 4=00
TIME , hours
5=00 6=00
Figure 12.
Particles per minute vs. time for large
particle system on August 6, 1975.
Rappers on. (Plant 1)
-------
- 91 -
• 6-12 jum
O l2-24j4.m
10:00 11=00 12=00 1 = 00 2--00
TIME, hours
3 = 00
4=00 5=00 6=00
Figure 13. Particles per minute vs. time for large
particle system on August 7, 1975.
Rappers off. (Plant 1)
-------
- 92 -
PEISETRATIDN-E
—^
a
M
1-
fX
[•
I i
*jr
LJ
Q_
i —
PERCEN'
10s,
101:
10°-
10'S
10"s-
1(
_
•
A Rap
A No rap
-
•
T
: i \
: }{fx A :
{ f f - t
: f %^'' \? :
Estimate from LPSS j I
i
i 1 1 i 1 1 Ml | 1 1 1 1 1 III 1 1 1 1 1 II 1 1 1 1 1 M 1 1
T2 10"1 10° 101 11
- 0-0
f90.0
Z
LJ
H
U
H
L
- 1-
i iii
PERCEN
r 99-9
-99.99
PARTICLE DIAMETER (MICROMETERS)
Figure 14. Plant 1 rap-no rap fractional efficiency
including ultrafine and impactor measurements.
-------
- 93 -
I
§
101-:
10°::
10'H
10
~2
POSErmATION-EFFICIQSCY
*
CD
OPEN SYMBOLS - NO RAP
CLOSED SYMBOLS RAP
A A ULTRAFINE
O • IMPACTOR
5f
T 0.0
::90.0
"s
i i i iiiii|—i i 111MI|—i i i mill—i i i inn
LJ
M
U
M
UL
LJ
CL
-99.9
10" 10"1 10° 101 10s
PARTICLE DIAMETER (MICROMETERS)
99.99
Figure 15. Rap no-rap ultrafine and impactor fractional
efficiency. Normal current density, Plant 2.
-------
- 94 -
a
M
I-
UJ
LJ
Q_
H
a:
LJ
Q_
10%
loH
io°i
icrH
10"5
10
PENETRATION-EFFICIENCY
OPEN SYMBOLS NO RAP
CLOSED SYMBOLS - RAP
AAULTRAFINE
©•IMPACTOR
T 0.0
+ 90.0
LJ
M
U
M
b_
5
U
e
—' i t mill i i i Mini—i i i n>n|—i i i inn) 99.99
'* 10"1 10P 101 102
PARTICLE DIAMETER (MICROMETERS)
Figure 16. Ultrafine and impactor rap-no rap fractional efficiencies,
Duct El., Plant No. 4, with 50% confidence intervals.
-------
- 95 -
100 ,
8
V)
a
~
O
2
LU
a.
a.
1 •
V1 = 0.155X-905
0.1 •-
1
10 100
CALCULATED MASS REMOVAL BY LAST FIELD
mg/DSCM
Figure 17. Rapping emissions vs. dust removal by last field
(Plant 6 and Plant 4 are hot-side installations)
-------
Table 2. SUMMARY OF REENTRAINMENT RESULTS
Plant
Raps/Hr
Field
Rapping
1
2
3
4
5
6
Losses,
Raps/Hr
Normal
6
6
3
3
1
1
Rap - No -Rap
One-Half
Normal Normal Raps/Hr Raps/Hr Raps/Hr
Current Current Rap- Rap- Rap-
Normal Density Density Normal No-Rap Normal No-Rap Normal No-Rap
10 4.29 3.75 10 1.67 30-60 12.5- 10 4.17
25
6 2.57 2.25 10 1.67 30-60 12.5- 5 2.08
25
1 0.43 0.38 5 0.83 30 12.5 5 2.08
5 0.83 30 12.5 2 0.83
- - - 1 0.42
Raps
/73
Min
Normal
8
8
3
3
1
1
Raps/Hr
Rap-
No-Rap
2.
2.
1.
1.
0.
0.
74
74
03
03
34
34
% of Emissions
Rap-No Rap/Rap
Normal-No Rap/
Normal
T.W.A.-No Rap/
T.W.A.
31
65
33
45
55
82
38
30
18
85
85
71
29
36
15
44
63
24
-------
Table 3. TYPICAL FLUE GAS AND ASH COMPOSITIONS
Plant
Date
Flue Gas
Temp., °C
SO 2, ppm by vol ,
SO 3 , ppm by vol ,
H2O, vol. %
Fly Ash
Ash Source
Date
Wt. % of l
Li2O
Na2O
K2O
MgO
CaO
Fe2O3
A1203
SiO2
TiO2
SO
1
8/7/75
164
282
0.3
8.2
Hopper 1
8/7/75
0.02
0.26
1.72
3.61
8.71
5.49
24.64
50.55
1.22
0.50
0.75
2
1/16/76
154
3200
12
7.2
High Vol.
Sample
1/15/76
0.02
0.54
2.49
0.95
4.73
22.72
18.52
45.69
1.45
0.30
2.77
3
2/25/76
155
2430
8.3
8.2
High Vol .
Sample
3/2/76
0.03
0.67
2.12
1.00
4.95
13.13
21.76
50.23
1.96
0.78
2.29
4
4/28/76
333
750
2.7
7.4
High Vol.
Sample
4/27/76
0.04
0.43
3.5
1.3
1.1
7.2
28.4
53.8
1.8
0.23
0.50
5
10/6/76
106
470
<0.5
8.7
High Vol.
Sample
10/5 &
10/6/76
0.02
1.38
0.54
1.1
5.8
6.1
13.2
70.8
0.87
0.05
0.50
6
1/31/77
346
355
<0.5
9.6
High Vol
Sample
1/31/77
0.013
1.52
1.4
1.8
6.0
5.0
24.3
57.6
2.1
0.32
0.54
LOI
0.61
5.72
10.92
3.5
1.0
0.11
1 Chemical analyses obtained from ignited samples
2 Loss on ignition
-------
- 98 -
Discussion
Mr. Gflthner mentioned a rapping loss of up to 85% of the total emissions
(not only of fine particulate) and wondered how this could be reduced,
as minimization of losses is essential for the improvement of precipitators.
Dr. Gooch answered that this value (85%) was from a hot side installation,
which has frequent rapping and was not typical of rapping reentrainment
emissions in general. Emissions below 2 microns are not affected by a
rapping optimization program; to reduce fine particle emissions, more plates
are needed. An optimization program can however reduce overall mass emissions
Mr. Gdthner asked how many particles in the stack gas were larger than
5 microns; the aaswer was not many, and that information regarding size
distribution was available, on a cumulative basis. A decrease in efficiency
in the 6 micron diameter region occurs even without rapping due to sporadic
reentrainment possibly as a result of sparking in the ESP.
The theoretical calculations used for constructing the ESP model were the
subject of a question posed by Prof. Weber. Mr. Gooch said that the model
was referenced in his paper and then described steps in the calculation of
the migration velocity at Prof. Weber's request: the precipitator uses a
mapping technique to calculate field distribution for incremental lengths
through the ESP. Mr. Wiggers asked what assumptions were made in the
calculations of migration velocity. The reply was that particles are
uniformly mixed, and that problems arise because of the following assumptions:
1) The non-uniform particle concentration gradients.
2) An average electric field was used in calculating the
particle charge in spite of the non-uniform field distri-
bution.
Additional work should be performed in modeling the collection of particles
in the diameter range of 0.5 to 4 microns. Dr. Kastner raised the problem
of rapping losses and of cleaning collecting electrodes (which cannot always
be cleaned by rapping), but Mr. Gooch replied that the Southern Research
Institute had made no investigations of rapping system effectiveness to date.
Mr. Gage added that optimization was required to minimize rapping losses.
-------
- 99 -
SESSION II: ADVANCED SYSTEMS FOR DUST REMOVAL
-------
- 100 -
ELECTROSTATICALLY AUGMENTED PARTICIPATE COLLECTION DEVICES
Dale L. Harmon
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park, N. C.
INTRODUCTION
EPA is placing increased emphasis on the control of fine participates
which persist in the atmosphere, comprise a variety of known toxic
substances, and are a major contributor to atmospheric haze and visibility
problems. The Particulate Technology Branch of IERL-RTP under the direction
of James H. Abbott has the objective of developing and demonstrating
control systems capable of effectively removing large fractions of the
under three micron size dust particles from smoke stack effluents. The
conventional systems have been demonstrated to have this capability for
some sources however, the cost of high efficiency collection is generally
high, in large part because the efficiency of most dust collectors
decreases for fine particle size. This performance loss must therefore
be offset by large size or high energy input.
Devices or dust collection systems based on new collection principles
or on radical redesign of conventional collectors are sometimes offered
by private developers. In the fall of 1973 a novel device evaluation
program was initiated to identify, evaluate and develop, where necessary,
those devices or systems which showed the most promise for high efficiency
collection of fine particulate. A novel particulate collection device
is a device or a dust collection system based on new collection principles
or on radical redesign of conventional collectors which is available for
testing as a pilot scale or full scale unit. Dale L. Harmon of the
Particulate Technology Branch has had the primary responsibility for
this program since it was started.
More than 40 novel particulate collectors have been identified.
About half of the devices identified have been of sufficient interest to
justify a technical evaluation. To date 13 devices have been either
field or laboratory tested. These are:
-------
- 101 -
Braxton-Sonic Agglomerator
Lone Star Steel - Steam Hydro Scrubber
R. P. Industries - Dynactor Scrubber
Aronetics - Two-Phase Wet Scrubber
Purity Corporation - Pentapure Impinger
Entoleter - Centrifield Scrubber
Andersen 2000 - CHEAP
Rexnord - Granular Filter Bed
Air Pollution Systems - Electrostatic Scrubber
Air Pollution Systems - Electro-Tube
Century Industrial Products - FRP-100 Low Energy Wet Scrubber
American Precision Industries - Apitron
Particulate Control Systems - EFB-Electrified Bed
In addition to these evaluations, a pilot scale TRW Charged Droplet
Scrubber was designed, built and demonstrated on a steel mill coke oven
and a mobile University of Washington electrostatic scrubber has been
built for tests on a variety of industrial sources.
Future plans include testing of the following devices if satisfactory
test sites can be located:
Combustion Power - Dry Scrubber
United McGill - NAFCO ESP
Dart Industries - "Hydro-Precipitrol" wetted wall ESP
Ceil cote Company - Ionizing Wet Scrubber
DuPont Company - DuPont Scrubber
Most of the novel devices which have been tested are scrubber
types. The only scrubber types tested which have demonstrated a major
improvement over conventional scrubbers are the electrostatically augmented
scrubbers. Two of the other non-scrubber types of novel devices tested
have also been electrostatically augmented. This paper will present
results of the electrostatically augmented novel particulate collection
devices which have been tested by EPA.
-------
- 102 -
NOVEL DEVICE TEST RESULTS
TRU Charged Droplet Scrubber
The TRW Charged Droplet Scrubber (CDS) applies electrohydrodynamically
sprayed water droplets to remove particulate material from a gas stream.
The droplets have a size in the range of 60 to 250 ym in diameter and
have a surface charge density near that allowed by surface tension
forces. The charged droplets are accelerated through the gas stream by
an applied electrostatic field. EPA's involvement with the CDS began
with laboratory and bench scale studies for the application of the CDS
to fine particle control. These studies included an analysis of the
particle removal interactions between particulate material and charged
droplets. The laboratory scale studies included the determination of
charged droplet characteristics under system operating conditions. The
results of these studies were used to verify some of the models used in
the fine particle removal analysis. The particle removal efficiencies
of a small size CDS operating under simulated process conditions were
measured during the bench scale studies. The results of these tests
indicated that the CDS should be effective for fine particle control and
was sufficiently developed for a pilot demonstration test.
Following the laboratory and bench scale tests, a pilot scale
demonstration was funded by EPA. The objectives of the pilot demonstration
were to verify the applicability for removing fine particles from an
industrial effluent stack, to determine the influence of CDS operating
variables on performance and behavior of the CDS under long term operation.
The demonstration unit was to be of sufficient size to adequately describe
the behavior of a full size unit. The emission source was to be characteristic
of those requiring control with a relatively large fraction of particulate
material in the submicron range.
The source selected for the demonstration was the flue gas emissions
from a coke oven battery. Heretofore, it was the general concensus of
the industry that there was no suitable control technology for this
process because of the wide process fluctuations. Emissions consisted
of varying relative concentrations of submicron sticky hydrocarbon and
micron sized high conductivity carbon black.
-------
- 103 -
The CDS unit used in the demonstration had a capacity of 51,000
3
m /hr at 1.83 m/s gas flow rate. An isometric sketch of the CDS is
shown in Figure I. A design summary is shown in Table I.
The scrubber contained three electrostatic spraying stages arranged
in series with parallel collecting plates on 0.127 m centers. It had 19
collecting modules. The scrubber structural members and collector
plates were fabricated from mild steel. Although the compatibility
problem of mild steel in the stack gas environment was recognized, it
was felt that the material would maintain its integrity during the test
period. The electrodes which distribute high voltage and water were
fabricated from type 316 stainless steel tubing.
Table I. CDS Design Summary
Three high voltage scrubbing stages with 0.127 m collector
plate spacing.
2
Flow cross sectional area, 7.36 m
High voltage electrode, type 316 stainless steel tubing
19 mm diameter, flattened to 12.7 mm.
High voltage electrodes contained 67 spray tubes each on
44.5 mm centers.
Spray tubes, titanium with a 1.27 mm O.D. by 0.15 mm wall
and protruding 25.4 mm from the electrode.
Collector plates 3.05 m long by 1.83 m high by 2.0 mm thick
mild steel.
Wall wash system covering each collecting surface.
The CDS demonstrated effective control of the emissions from the
coke oven battery over widely fluctuating process conditions. Particle
removal efficiencies up to 95% were measured and were an increasing
function of the time averaged particle loading. Improvements in the gas
distribution internal to the equipment should result in additional
improvements in collecting efficiency.
-------
- 104 -
The average inlet particulate load varied between 155 and 755
mg/Nm3, associated aerodynamic mean diameters varied between 0.4 pm
(hydrocarbon aerosol) and 1.5 ym( carbon black). The most sensitive
design variable affecting efficiency was the gas volume flow rate through
the equipment. Low total energy and water consumptions, 0.7 - 1 watts/hr/Nm
and 0.11-0.3 a/Mm3, respectively, were demonstrated over most of the
test conditions. Operation of the CDS with intermittent (8 hour cycle)
collector plate over sprays was adequate for deposit control.
Since the pilot scale demonstration funded by EPA was completed a
full scale CDS unit was purchased and installed on this source. Collection
efficiency has been equal or better than that observed during the pilot
demonstration but severe corrosion problems have developed. EPA is
supporting work to solve these problems.
Air Pollution Systems Scrub-E
The Air Pollution Systems (A.P.S.) electrostatic scrubber (Scrub-E)
is basically an electrostatic charger (or ionizer) followed by a venturi
scrubber. Figure II is a schematic diagram of the pilot system. An
electrode is placed upstream of the venturi to charge the inlet particles,
which then enter the venturi throat. The gas stream atomizes the central
water spray in the venturi throat and the charged particles, according
to A.P.S., are then attracted and collected by the highly polarized
water molecules. The charged particles are also collected on the walls
of the ionizer section prior to the throat of the venturi. A thin film
of water is run down the inclined surfaces to keep the walls clear and
prevent high voltage arcing. The particle laden water droplets are then
collected by a cyclonic separator and sent into a settling tank (clarifier).
The water can then be recycled back into the scrubber system. However,
during the test program, fresh water was used.
The ionizer consists of an electrode supported in the inlet of the
venturi section. According to A.P.S., a stable electrical discharge of
high intensity is maintained across the venturi throat between the
center electrode and the wall. A.P.S. claims that the average field
-------
- 105 -
that can be maintained across the "electrode gap (space between the
electrode probe and the wall) is substantially higher, 14-16 kV/cm, than
that of a standard electrostatic precipitator, 4 kV/cm.
The pilot scale laboratory test of the Scrub-E showed this system
to be equal to a conventional venturi scrubber with a power requirement
1-1/2 to 2-1/2 times as great.
Last year a competitive procurement was issued to demonstrate at
pilot or small full scale the technical and economic feasibility for the
most promising existing novel particulate collection system for control
of fine particulate emissions from industrial sources. This competitive
procurement was won by A.P.S. for demonstration of the Scrub-E. A
contract was funded in September, 1977 with A.P.S. to demonstrate a
300 to 600 m /min Scrub-E on a fine particulate source. A.P.S. is now
looking for a site for this demonstration. Emphasis is being placed on
locating a primary smelter. Final site selection will be made and
equipment design will be underway within the next few days.
Air Pollution Systems Electro-Tube
The pilot scale A.P.S. Electro-Tube is basically a tube electrostatic
precipitator with a central rod electrode and wetted wall collector.
Figure III is a schematic diagram of the pilot system. The inlet particles
are charged in a high energy field (12 kV/cm) by a high intensity ionizer
at the base of the electrode. The charged particles then migrate to the
wetted wall in the body of the device in a field of 5-10 kV/cm. A.P.S.
indicates that initial saturation charge on the particles is higher than
the usual 4-5 kV/cm for a conventional ESP and facilitates increased
migration in the collecting electric field.
The A.P.S. Electro-Tube, which is similar to a wet wall electrostatic
precipitator, gave some very high efficiencies on fine particulates--as
high as 98.9% on 0.5 micron particles. This performance is similar to
that which can be achieved in small wet electrostatic precipitators with
the same ratio of plate area to volumetric flow rate.
-------
- 106 -
University of Washington Electrostatic Scrubber
The University of Washington (UW) Electrostatic Scrubber involves
the use of electrostatically charged water droplets to collect air
pollutant particles electrostatically charged to a polarity opposite to
that of the droplets. A schematic illustration of the UW Electrostatic
Scrubber system is presented in Figure IV. The particles are electrostatically
charged (negative polarity) in the corona section.
From the corona section the gases and charged particles flow into a
scrubber chamber into which electrostatically charged water droplets
(positive polarity) are sprayed. The gases and some entrained water
droplets flow out of the spray chamber into a mist eliminator consisting
of a positively charged corona section in which the positively charged
water droplets are removed from the gaseous stream.
The Particulate Technology Branch has funded construction of a
mobile UW Electrostatic Scrubber for tests on a variety of industrial
sources. Tests have been completed with the unit on the emissions from
an electric arc steel furnace. The tests illustrated the system's
capability for high efficiency fine particle collection at a relatively
low energy consumption. Measured overall particle collection efficiencies
ranged from 79.7% to 99.6% depending on electrostatic scrubber operating
conditions and upon the inlet particle size distribution. Figure V
illustrates the effect of specific plate area (SCA) and liquid/gas ratio
(L/G) on the particle collection efficiency as a function of particle
size.
The mobile unit is now installed on a coal-fired power plant and
preliminary tests showed collection efficiencies as high as 98% for a
0.5 micron diameter particle.
Apitron Electrostatically Augmented Fabric Filter
American Precisions Industries, Inc. has developed the Apitron unit
which is an electrostatically augmented fabric filter. Figure VI provides
a cutaway view of the pilot plant unit which was tested by EPA.
-------
- 107 -
The device is divided into two separate compartments, which share a
common inlet, hopper, and power supply but each has its own exit duct
and flow metering capability. Only one of the two compartments was used
in these tests. Pertinent dimensions and operating data are given in
Table II.
Table II. Apitron Operating Data
Precipitator tube inside diameter
Precipitator tube length
Number of tubes per compartment
Number of independent compartments
Number of bags per tube
Filter area per bag
Operating voltage
Operating current per compartment
Cleaning pulse duration
Cleaning interval
Nominal air flow per compartment
Operating pressure drop
Bag material
12.7 cm
83.8 cm
3
2
4
0.293 m2
30 kv
7.5 mA
50 msec
6 minutes
0.118 am3/s
3.3 cm w.c.
Teflon or Nomex
Incoming air flow enters the precipitator section from below with
the upper portion of the hopper serving as an inlet plenum. The flow
then passes upward, through the tubes of a set of parallel wire-pipe
type precipitators in which the particulate is charged and much is
precipitated. Flow continues upward, past the tubes, into and through
the bags where the final filtration takes place. Clean air exits the
unit at an exhaust located in the side of the bag housing section. In
the pilot plant operation each tube (and associated set of bags) is
cleaned one at a time with a six minute interval between successive
cleaning of any one tube and bag set. In the full scale system, six bag
and tube sets are cleaned at a time. Cleaning is initiated by an electrical
-------
- 108 -
pulse from the control system which opens a diaphragm valve for several
tens of milliseconds. A blow pipe connected to the valve is then pressurized
which results in a compressed air jet downward from a nozzle directly
above and concentric with the corona wire of the tube being cleaned.
The jet of air flowing downward through the tube entrains and mixes with
a secondary air flow sweeping the tube clean of deposited dust by the
mixture of high velocity air. The secondary air flow, passing from the
outside to the inside of the bags, snaps the bags inward and dislodges
the dust deposits from the bags. Vertical height constraints in the
mobile pilot plant required the use of four short bags in parallel over
each precipitator tube rather than one longer bag over each tube as
would be used in full scale systems.
The operating conditions for the Apitron device during the seven
test days are given in Table III. The first two days of testing of the
Apitron system was done with relatively old Nomex bags in use. These
bags had been subjected to sulfuric acid attack during an earlier test
program and were tested only because a new set of Teflon bag which were
scheduled to be used were not immediately available. After two days of
testing with the Nomex bags the Teflon bags arrived and were installed.
One day of tests were performed with the new, unconditioned Teflon
material, after which the bags were run continuously for two more days
over a weekend for conditioning before testing was resumed.
After testing was resumed two days of data were obtained with
electrostatic augmentation at a face velocity of about 35 mm/sec, followed
by two days of testing without electrostatic augmentation. One of the
tests without electrostatic augmentation was run at a face velocity of
33 mm/sec and the other at a face velocity of 14 mm/sec.
The results of the total paniculate tests are given in Table IV.
The power consumption figures do not include the required compressor
power for the cleaning pulses nor any conversion efficiencies for fans,
motors, power supplies, etc. The estimated total energy usage including
losses in the latter items is approximately 40% greater.
-------
Table III. Apitron Operating Conditions
Date
11/30/77
12/1/77
12/2/77
12/5/77
12/6/77
12/7/77
12/8/77
Bag
Material
Nomex
Noruex
Teflon
Teflon
Teflon
Teflon
Teflon
Outlet gas flow
am3/min
7.31
7.00
8.47
7.42
7.00
6.91
2.92
acfm
258
247
299
262
247
244
103
Outlet
temp.
°C
74
77
71
74
74
74
54
Bag face
mm/sec
34.7
33.2
40.2
35.2
33.2
32.8
13.8
Velocity
fpm
6.83
6.53
7.91
6.93
6.53
6.46
2.72
Voltage
kv
31
31
30.5
30
29
0
0
Current
mA
7.5
7.5
7.5
8.0
8.5
0
0
ESP
conditions
Specific collecting area
mVamVs t^/lQOO acfm
8.25
8.60
7.11
8.11
8.60
0
0
41.9
43.7
36.1
41.2
43.7
0
0
Current
density
nA/cm2
750
750
750
800
850
0
0
Energy
usage
joules/m
1910
2000
1620
1940
2120
0
0
Apitron
pressure
drop
era w.c.
5.3
5.3
0.5
3.0
3.8
8.9
3.3
Energy
usage,
air
moving
joules/ra'
530
520
50
300
380
868
323
1
o
10
1
3 New, clean Teflon bags installed overnight of 12/1 - 12/2
-------
- 110 -
Table IV. Apitron Efficiency Results
3
Collection Efficiency. % Energy Usage, joules/am
UCL V,C
12/1/77
12/2/77
12/5/77
12/6/77
12/7/77
12/8/77
99.59
99.84
99.90
99.94
99.905
99.93
2520
1670
2240
2500
868
323
Inertia! sizing of the inlet and outlet particulates was accomplished
using modified Brink impactors and University of Washington impactors.
Collection efficiencies in excess of 99% were consistently obtained in
the fine particle size range (< 3 ym) when either the Teflon or Nomex
bags were used. The efficiency curves for both materials showed efficiency
minima near a diameter of 7 ym which may have resulted from agglomerates
bleeding through the fabric of the bags. The efficiencies were higher
on the two days of testing with the ESP section de-energized than
during the tests with power on. The highest efficiencies during the
entire test series were obtained on December 8 when the device was
operated as a conventional baghouse at a low face velocity. There is
some uncertainty in the significance of the differences among the results
for the various test conditions with the Teflon bags because the bags
were being tested immediately after installation and for a few days
thereafter. The efficiency of the device showed a constant improvement
with each day of testing after the new bags were installed. Thus, the
efficiencies which were obtained during the last two days of testing
with the ESP de-energized may have been obtained had it been on as well.
Fractional efficiencies are shown on an aerodynamic diameter basis for
two days in Figures VII and VIII.
Particulate Control Systems Electrified Bed
Particulate Control Systems, EFB, Inc., has developed an electrified
granular bed (EFB). A schematic diagram of the EFB is shown in Figure
IX.
-------
- Ill -
A corona charging section charges the participate before collection in a
charged granular bed. A pilot scale unit installed on an asphalt roofing
plant asphalt saturator is currently being tested by EPA. In this
application, rather than clean the dirty granular material, it is used
in the process on the asphalt shingles.
Ceil cote Ionizing Wet Scrubber
The Ceilcote Company is marketing an Ionizing Wet Scrubber. A
schematic diagram of this device is shown in Figure X. In operation,
contaminated gases pass through a high-voltage ionizer section that
houses negative-polarity discharge electrodes and wetted plates that act
as positively grounded electrodes. Electric corona discharge from the
electrodes produces ions that intercept the fine contaminants and relinquish
their charge to the particles. Then, the charged particles enter the
crossflow wet scrubber section that is packed with Tellerettes where the
larger particles (3-5 y and larger) collect through inertial impaction.
If the smaller particles (less than 3-5 y) do not impact at some point
along their journey, the probability is high that they will be captured
by electrostatic attraction.
An EPA sponsored test of this device is planned in the near future.
R. P. Industries Electro-Dynactor
The Electro-Dynactor System (EDS) is marketed by R. P. Industries,
Inc. In the EDS, the influent gas is ionized and particulate contaminants
are electrically charged by an electrostatic ionizer prior to each stage
of aspiration and wet scrubbing. The ionizers use about 15,000 volts
and are self-protected against short circuits due to sparks that occasionally
jump the ionizers components.
A three-stage EDS is shown schematically in Figure XI. Particulate-
laden gas is aspirated through the EDS by the diffusion of the scrubbing
water as shown in each of the three scrubbing stages. As the gas passes
through each ionizer, the sub-micron particulate is charged negatively
-------
- 112 -
by a combination of diffusion and field charging. Upon being aspirated
into each scrubbing section, the gas mixes intimately with a dense cloud
of water droplets. The negatively charged particles induce positive
charges called "image charges" in nearby neutral water droplets. The
electrostatic force of attraction between the charged particle and its
image in a water droplet causes the particle to be attracted and collected
by the droplet. The liquid is separated from the gas in the reservoir/separator,
The gas passes through a mist eliminator, and then is recharged in the
succeeding ionizer. Normally three similar units (stages) are used in
series; thus, the process is repeated three times, after which the gas
passes to the stack through a final mist eliminator.
The EDS is being considered for an EPA sponsored test.
-------
COLLECTOR PLATES
FLANGED
GAS EXIT
ELECTRICAL
UPPER SECTION
HIGH TENSION
SUPPORT
HOUSING
GAS
DISTRIBUTION
LOWER
SECTION
HIGH TENSION
CONNECTOR
PANEL
FLANGED
GAS INLET
OVERFLOW
HIGH TENSION
WATER HEADERS
SLURRY
DISCHARGE
MAINTENANCE
PLATFORM
OJ
I
Figure I. TRW charged droplet scrubber.
-------
- 114 -
FLUE
GAS
ELECTRODE-
IONIZER
SECTION
VLNTURI
SPRAY-
WATER
TO WASH
IONIZER
WALL
A—£:
HIGH
VOLTAGE
POWER
SUPPLY
TO INDUCED
DRAFT FAN
CLEAN
GAS
OUT
CYCLOXE
ENTRAIXME.VT
SEPARATOR
CLARIFIER
Figure II. Air Pollution Systems Scrub-E
-------
- 115 -
FROM HIGH
VOLTAGE SOURCE
AEROSOL
INLET
OPTIONAL
SECONDARY
AND TERTIARY
I ONI Z AT ION-
ZONES
AEROSOL OUTLET
TO BLOWER
U
OUTLET
WATER DRAIN
{L, WATER
INLET
ELECTRODE
BODY
HIGH INTENSITY
IONIZER SECTION
•TANK
Figure III. Air Pollution Systems Electro-Tube
-------
- 116 -
GAS INLET
CORONA
(PARTICLE CHARGING)
GAS OUTLET
MIST ELIMINATOR
SCRUBBER
CHARGED WATER SPRAYS
(COLLECTION OF CHARGED PARTICLES
BY OPPOSITELY CHARGED WATER DROPLETS)
Figure IV. University of Washington Electrostatic Scrubber
-------
- 117 -
Test Corona V, Spray V, Overall SCA, L/G.gal./
No. kV kV Col! Eff. % ft2/cfm 1000 c:
99.9
PARTICLE AERODYNAMIC DIAMETER, D50
Figure V. Influence of SCA and L/G on Particle Collection Efficiencies,
-------
- 118 -
COMPRESSED AIR
MANIFOLD
CLEAN GAS
OUTLETS
BAGS
DIRTY GAS
INLET
- JET PULSE NOZZLE
INSULATOR
TUBE SURFACE
COOLING WATER
MANIFOLD
CORONA WIRE
HOPPER
DUST DISCHARGE
WATER OUTLET
Figure VI. Electrostatically Augmented Fabric Filter
-------
- 119 -
a
M
a:
LJ
LJ
a.
h-
LJ
U
Q£
LJ
Q_
lO2^
lO1^
10°-: r
10
"30.0
LJ
qqn M
%. J *- —* * V-/ f »
LJ
M
LJ
U
a:
LJ
a.
1 "i ~T 1 1 1—i—r~~i—r~j 1 1 1 1—< ' _
10"1 10° 101
PARTICLE DIAMETER (MICROMETERS)
S3.
Figure VII. Apitron Fractional Efficiency - 12/6/77
-------
120 -
§
H
LJ
h-
LJ
U
Q£
U
Q_
10
-3
^90.0
I x
Z
LI
H
U
H
U.
Lu
LJ
h-
LJ
U
Of
LJ
Q_
pq qq
»
iaa.
10
rl
1O'
,0
H4- 99.933
ID1
PARTICLE DIAMETER (MICROMETERS)
Figure VIII. Apitron Fractional Efficiency - 12/7/77
-------
FRESH BED MATERIAL
CORONA CHARGER
POWER SUPPLY
POLLUTANT COATED MATERIAL
POWER SUPPLY (-)
Figure IX. Particulate Control Systems electrified bed.
-------
SPRAY HEADER
ELECTRODE
WIRES
GROUNDED PLATES
t\J
ro
Figure X. Ceilcote ionizing wet scrubber.
-------
FIGURE XI
A SCHEMATIC VIEW OF A 3 STAGE
ELECTRODYNACTOR SYSTEM
GAS TO
STACK
t
RADIAL
IMPEDANCE
TRANSFORM SECT,
PLENUM
CHAMBER
SCRUBBING
COLUMN
GAS TO
BE SCRUBBED
MIST ELIMINATOR
RESERVOIR/SEPARATOR
PUMP
-------
- 124 -
Discussion
Dr. Giithner asked which of the new techniques described in the
talk was the most promising when compared with conventional
techniques in term of energy consumption. Mr. Harmon replied
that the API Electrostatic Scrubber which had been demonstra-
ted showed great promise although other units were comparable.
-------
- 125 -
How to raise the efficiency of dry electrostatic precipitators
by means of gas conditioning
Dr. H. ReiBmann
Lurgi Umwelt und Chemotechnik GmbH
Frankfurt/Main
Gentlemen,
1. Introduction
The dust properties have a great influence on the collection
efficiency or, in the case of equal efficiency, on the
required precipitator size. Depending on the different
conditions, the precipitator size can vary according to the
relation 1 : 6.
The dust collection conditions must be regarded as a complex
of many factors such as fuel properties, combustion
mechanisms, gas composition, dust quantity, chemical
properties of the dust, grain size and surface characteristics
of the dust particles.
Many of these influences cannot be measured nor calculated.
The only factor that can be represented is the electrical
dust resistivity, which is determined by the majority of
the other factors, thus forming the most important criterion
for the assessment of precipitator conditions.
-------
- 126 -
The following figure
Figure 1
10s 10'° 10" 10'! 10° S«10"
Stoubwiderstand p in n cm »-
Wanderungsgeschwindigkeit.wirksamer Strom und
Filtersponnung als Funktion des Stoubwiderstandes
T77-1209
shows the dependence of the flash-over voltage Ug, the
possible discharge current I and the attainable migration
velocity w on the dust resistivity; it has to be pointed out
in this connection that, with equal efficiency, the
collection area is proportional to the w value.
By- far the greatest problems arise in large pit coal fired
blocks. The following figure
Figure 2
0 100 200 300 i.00
Gostemperotur / in°C ^
Stoubwiderilondols Funklion
der Gostemperolur fur unter -
schiedlfChpTaupurMstemperaluren
T77-1210
shows the electrical resistivity of a pit coal fly ash as a
function of the gas temperature and the water vapour content
at constant SO^ content of the flue gas.
-------
- 127 -
It can be seen that the resistivity passes, at normal operating
temperatures and moisture contents, through a range far
beyond 10 ohms x cm, which is the approximate critical limit
for trouble-free precipitator operation. Below the maximum,
the resistivity is reduced by the condensation of electrolytes -
primarily sulphuric acid - on dust particles, preferably on
the rugged surfaces and in the recesses of the agglomerates.
Above the maximum, the dust resistivity decreases due to the
increasing conductivity of the material. A surface
condensation at temperatures which are considerably above the
acid dew point is no longer possible. It is at this point
that the dust resistivity curves corresponding to the
different water vapour contents meet. In the majority of
modern pit coal fired boilers, we find many of these
unfavourable factors, such as:
low sulphur content of the coal,
low conversion rates of 302 an(* S°3 due to high
combustion chamber temperatures and to the small
amount of air in excess,
flue gas temperatures which are mostly in the
range of the maximum dust resistivity after the
air preheaters,
low content of unburnt particles, i.e. of
conductive dust particles,
fine, spherical particles as shown in the following
electron micrograph.
The dust shown comes from a 340 M¥ boiler with dry ash
removal; the magnification ratio is 1 : 2000.
-------
- 128 -
Figure 3
LURGI I Kornform ,,nur Kugeln"
7563
The following figure shows dust resistivities in several
South African power plants. We find highly differing dust
resistivities, although the coal analyses are almost
identical. This means that the dust resistivity is very
much influenced by the processes taking place in the boiler.
Therefore the dust resistivity cannot be precalculated on
the basis of the coal analysis alone, but it has to be
measured in the plant itself.
Figure
With a view to coping with these unfavourable factors which,
in extreme cases, cannot be controlled without influencing
them, we investigated the possibilities of improving the
operating conditions. Flue gas conditioning, that means
-------
- 129 -
influencing of the dust resistivity by improving the surface
conductivity, is a solution to this problem. Flue gas
conditioning can be carried out by increasing the water
vapour or SO, content in the flue gas, by reducing the gas
temperature or by adding other chemicals such as ammonia or
triethylamine.
2. Conditioning by the addition of
By adding SO-z or sulphuric acid vapours, the conductivity of
high resistivity dusts can easily be improved. The tests
carried out on a precipitator plant working under unfavourable
conditions in a German 300 M¥ block show the following
results:
Figure
80 100 150 200 250 300
Gostcmperolur / in°C ^*
I LURCIJ
Stoubwiderstandols Funk lion
der Gaslemperalur fur unler-
schiedliche SO^-Konzenlralioner
T77-1213
From the diagram it can be seen that the addition of
20 ppm by volume of SO-? would suffice for reducing the dust
resistivity to a more favourable value.
-------
- 130
The following figure
Figure 6
t
"E
E
S
J
I
1
1
a:
( LURCI )
\
i
'
\
\
1
I <
1
^ ^
\ ' I N 1
109 2 i 6 8 1010 2 i 6 8 1011 2
Staubwiderstand q n fi cm —
Reingasstaubbeladung in Abhangigkeit
vom Staubwiderstand
1
T
J
468 1012
>"
F77-12U
shows the dependence of the clean gas content on the dust
resistivity and thus on the SO^ content for the case mentioned.
I should like to point out again that, according to the
diagram shown, a dust resistivity of 10-^ ohms x cm can
already be achieved by adding approximately 20 ppm of SO-r.
As can be seen, the clean gas dust content is reduced by
Q% of the initial value, that means from 500 mg/m^ to
approximately ^tO mg/m3.
Tests carried out in an Australian power station showed the
following relative increase in the migration velocity as
a function of the 803 addition.
Figure 7
3? A
c
^
•61
c
•° g, t0-
2 c1
S 2 ?n
.0 c
/
/
/
/
/
/
/
0 10 20 30
SOD- Zugobe in ppm ^~
Verbesserunq der Wanderunqsqeschwindiqkeit
LURGI. ^ 3 ^ Tyy 1?1S
n Abhanqigkeit von der S03- Zugobe
With an addition of 20 ppm of SO-r the increase amounts to
100%, that means that the collection area necessary is only
50% of the area required in systems without conditioning.
-------
- 131 -
Naturally, such vital improvements can only be expected in
cases where the flue gas at the boiler outlet contains
virtually no free SO^; this may be the case with coal having
a very low sulphur content so that no 863, or only a small
amount, is formed in the firing system, and that almost
complete absorption of the 803 by the alkaline dust particles
is ensured.
863 conditioning is already carried out on an industrial
scale in a few power stations in the United States of America,
in Australia and in Great Britain. Due to the fact that the
handling of ready-made 803 involves problems, it is advisable
to produce the gas on the spot.
The following figure
Figure 8
Flieflschema fur S03-Erzeugung
T 7 7-1216
shows the flow diagram of an 863 production plant. Liquid
sulphur is burned to form S02, which is oxidized to SO^ with
the help of a catalyst. The 863 is mixed with air and
injected into the flue gas flow. 80^ production plants
working according to this principle are frequently used in
the chemical industry. A flue gas conditioning plant in
a German power station with 1 x 300 M¥ and 2 x 150 MW blocks
will go into operation about the middle of this year.
-------
- 132 -
There is no reason to fear that conditioning of the flue
gas by means of SO^ will cause an additional impact on the
environment due to the higher sulphur dioxide emission.
The SO-r injected is almost completely absorbed by the dust.
A higher SO-? concentration in the flue gas after the
electrostatic precipitator could not be ascertained. It is
only in the case of slag tap firing systems with dust
recirculation that the sulphur compounds are decomposed
again in the firing system, so that additional S02 is set
free. However, for a 750 M¥ block, the quantity involved
by the addition, of 20 ppm of SO-j is only approximately
140 kg/h, which corresponds to an increase in the sulphur
content of the coal by 0.02%.
When comparing 600 MW block dust collection plants with and
without SO^ conditioning for operation conditions with dust
resistivities as shown for the South African power stations
at the beginning, we obtain the following values:
without with
SO-? conditioning
Gas volume m^/s 913
Collection efficiency % 99.2
Precipitator size % 100 60
Investment costs % 100 83
The additional operating costs amount only to $40,000 for
8,000 operating hours per year.
-------
- 133 -
Apart from considerable cost savings, the electrostatic
precipitator plant with SO^ conditioning system has the
advantage of operating with constant collection efficiencies
even in the case of fluctuating operating conditions and
therefore varying dust resistivities.
The sizes of precipitator plants with and without SO-?
conditioning are compared in the following figure:
Figure 9
Prectpitator with 503- Conditioning
Precipitator without S03-Conditioning
Comparison
of Size
T77-1015
3. Conditioning by means of water
A relatively simple way of conditioning, which is
occasionally used also for the improvement of existing
plants, is the injection and evaporation of water in the
boiler in order to increase the moisture content of the gas
However, if a high dust resistivity is to be considerably
reduced, the water vapour content has to be raised to at
least 10% by volume. In the case of a 750 MW block, this
-------
- 134 -
corresponds to an injection water volume of approximately
95 tonnes/hour. The evaporation of this volume is, like
the addition of steam, unfeasible for economical reasons.
But there is another solution available and that is the
injection of water after the air preheaters. In this way
the flue gas is cooled and at the same time the dust
resistivity is reduced by the two components, namely the
higher water content and the lower gas temperature. The
following figure
Figure 10
100 150 200
Gastemperotur / in °C ^-
Staubwiderstand als
Funktion
der Gdstemperatur
lurunterschiedliche
Taupunklstemperaturen
T77-1211
shows the test results obtained in a 350 MW block with pit
coal slag tap firing system. While the influence of the
higher water dew point alone is still small at operating
temperature (Line A-B), it is considerably increased as the
temperature drops (Line A-C). By simultaneous cooling and
moistening, the dust resistivity can be reduced by
approximately one power of ten. The reduction of the gas
temperature is the more important factor in this connection.
In this special case, the injection of only 8 g water / m^ gas
brought about a decrease of the clean gas dust content from
180 to 60 mg/m3.
-------
- 135 -
With this type of conditioning, the water Tolumes are
comparatively small. In order to reduce the flue gas
temperature by 20°C while having a simultaneous increase of
the dew point by approximately ^°C, approximately 25 tonnes/h
water are required for a 750 MW block.
Water injection constitutes a relatively simple way for
improving the collection conditions. However, complete
evaporation of the liquid injected must be ensured in order
to avoid accretions of the moistened dust. In spite of the
fine atomization, this demands, at the low temperature
level involved, relatively long retention times so that the
use of special eAraporation coolers becomes necessary.
Figure 11
{ LURGf)
Elektrofilter
mit vorgeschaltetem Verdampfungskuhler
T77-1212
This method of conditioning the gas by means of water in an
evaporation cooler has been applied in the cement sector
before electrostatic precipitators arranged after suspension
preheater kilns for many years.
-------
- 136 -
Figure 12
Staubwiderstond bei Direkl-
und Mahltrocknungsbelneb
T78-105i
Here we have two types of dust with extremely different
resistivity curves, namely curve a, which corresponds to the
so-called direct operation, and curve b, which corresponds
to the operation with a raw material grinding and drying
unit. If either method is applied in continuous operation,
the dust resistivities will not be super critical (items 1
and 3). However, in the majority of cases, a change-over
from direct to compound operation is carried out once a
week, and the dust resistivity resulting for the electrostatic
precipitator operation will follow items 1-4-5-2. It
will only be after an extended continuous operation that the
resistivity will drop to 3. In this case, it is advisable
to use an evaporation cooler before the precipitator and
thus to change the operating temperature prevailing during
direct operation from 310°C to approximately 150°C. Due to
the simultaneous increase of the dew point, the dust
resistivity will always remain below the critical value.
(item 6)
-------
- 137 -
Figure 13
Verdampfungskuhler • Regelung i nco
T 76 - 1 069
The water to be evaporated is injected into the cooler head.
The volume is controlled in dependence on the precipitator
inlet temperature. Since the cooler volume required for
complete evaporation is proportional to the square of the
maximum droplet diameter, greatest importance has to be
attached to the finest possible atomization. The so-called
return flow nozzles have proved to be the best design for
this application.
While there are many evaporation coolers in operation in the
cement sector, in the steel industry, after waste incinerators
and in other industrial fields, the application of evaporation
coolers for the collection of dust from power station waste
gases is new. Therefore, an industrial scale pilot plant is
being constructed at the moment in a German power station.
Figure
RouchgasenlslQubung durch
Eleklroltller mit vorgeschallelen
Verdomplungskuhler
T78-1055
-------
- 138
The power station concerned is a 150 MW block. The evaporation
cooler has been designed to lower the temperature from 150°C
to less than 100°C. The evaporation cooler is followed by
the fan, the electrostatic precipitator and the stack. By
cutting in or cutting out one of the preceding electrostatic
precipitators, the dust content of the raw gas reaching the
pilot plant can be altered in order to allow testing of the
separation of very fine dusts with and without conditioning
by water. It is expected that very low clean gas dust
contents, such as achieved after bag filters, will be reached.
The first results will be available in about a year.
Preliminary tests which have already been carried out give us
grounds to be optimistic.
Other conditioning means
Conditioning by means of the admission of ammonia is also
carried out in some plants. Its effect on the collection
conditions however is differing and rather small compared
to SO-?. Tests were carrfed out and it was found that,
although equal quantities had been added, the ammonia raised
the migration velocity by 30%, whereas the SO-? increased
the migration velocity in the plant by 100%.
Tests which were primarily carried out in Australia have
shown that triethylamine is, in principle, a good conditioning
agent. Application of triethylamine in practical operation
is, however, unfeasible at the moment due to the high
purchase costs.
Drlim/Shdt/Bltt
7th March 1978
-------
- 139 -
Discussion
Mr. McCain asked whether S03 was created during the electrostatic process.
Dr. Reissmann replied that there is, whether conversion from S02 to SCL
in the boiler or in the precipitator (this was confirmed by Mr. Gooch).*
He added that it is difficult to differentiate S03 from S02 by measurement.
Mr. Parker inquired as to the effect of conditioning on particle size;
unfortunately it was not yet possible for Dr. Reissman to answer this
type of question. Mr. Princiotta raised the question of the economic
viability of S03 when compared with humidification and water addition.
The answer was that the actual running costs at a plant were low and
could vary according to conditions. Mr. Gage then wished to know whether
possibilities existed for cooperation between the U.S. and W. Germany in
humification studies. Dr. Reissmann said that this was not feasible at
the time; however, at Mannheim a plant was under construction which
would test a considerable variety of coals including overseas. He
expressed the hope that future cooperation would take place, and was
supported by Dr. Holighaus. Dr. Laufhutte pointed out that conversion
of SCL to SCL is minimized in coals having a low sulphur content.
Dr. Reissmann stated that normally all boilers were fired with coal; the
German boilers however, were of a slag hole type and produced only small
quantities of SCL. He added that SCL contents less than 1% were sufficient
for the conditioning process.
The question of sulphur trioxide content in fly ash was raised by Dr. Davids,
who was informed that there was no increase at the stack of S03 in the
gas. S03 passed into the dust (which was recycled into the furnace) and
split into SCL.
*Mr. Gooch does not recall what he "confirmed" but he would not expect
significant amounts of oxidation of SC^ in the ESP.
-------
- 140 -
IMPROVED DESIGN METHOD FOR F/C SCRUBBING
by
Seymour Calvert
Air Pollution Technology, Inc.
4901 Morena Blvd., Suite 402
San Diego, California 92117
USA
-------
- 141 -
IMPROVED DESIGN METHOD FOR F/C SCRUBBING
Flux force/condensation (F/C) scrubbing provides a means
for the enhancement of fine particle collection efficiency
through the effects of condensing water vapor. Several phenomena
are simultaneously involved and a detailed mathematical model is
complex. In a series of previous studies l '2 '3 ' ** we have
developed a design method (i.e., performance prediction) based
on a model which required the simultaneous solution of several
differental equations. While it produced useful results, it
was a time-consuming procedure, even with the aid of an elec-
tronic computer.
In the course of refining our design method we arrived at
the conclusion that the flux force effects could be treated
separately from the condensation effects in many scrubber situa-
tions. While the condensation-induced improvement in inertial
impaction efficiency could be handled conveniently, the flux
force deposition prediction remained cumbersome. Recently the
senior author became aware of a concept developed by Whitmore5
which provided the key to simplifying the prediction of the flux
force effects in F/C scrubbing. By incorporating Whitmore's
concept into our model we have developed a much simpler design
method which is convenient to use. This revised model will be
described below.
BASIC CONCEPTS Before proceeding to the details of the mathe-
matical model the basic concepts and outline of the approach
will be discussed. If we consider a typical F/C scrubbing system,
it might have the features shown in Figure 1. The gas leaving
the source is hot and has a water vapor content which depends on
the source process. The first step is to saturate the gas by
quenching it with water. This will cause no condensation if the
-------
- 142 -
particles are insoluble, but will if they are soluble. There
will be a diffusiophoretic force directed away from the liquid
surface.
Condensation is required in order to have diffusiophoretic
deposition, any growth on insoluble particles, and extensive
growth on soluble particles. Contacting with cold water or a
cold solid surface is done next to cause condensation. While
condensation occurs there will be diffusiophoretic and thermo-
phoretic deposition as well as some inertial impaction (and,
perhaps, Brownian diffusion). The particles in the gas leaving
the condenser will have grown in mass due to the layer of water
they carry.
Subsequent scrubbing of the gas will result in more particle
collection by inertial impaction. This will be more efficient
than impaction before particle growth because of the greater
inertia of the particles. There may be additional condensation,
depending on water and gas temperatures, and its effects can be
accounted for as discussed above.
One can apply this general outline of F/C scrubbing to a
variety of scrubber types and Figure 2 shows a multi-plate
F/C scrubber system. It can be seen that the gas is saturated
before entering the plate column, although this is not always
necessary. The first plate can serve as the saturator and par-
tial condenser. Generally, the efficiency of heat and mass
transfer is so high (say, 80%+) on a well designed plate that
most of the condensation occurs on the first plate.
In subsequent plates the gas is scrubbed by inertial impac-
tion and there will be a minor amount of additional condensation.
We have shown a simple counter-current column but other varia-
tions are possible.
The mathematical model is based on the process just des-
cribed for a plate-type F/C scrubber. It is outlined on the
following page.
-------
- 143 -
I. Saturate the gas before plate 1
A. Particles are collected at size "d ".
PI
B. No condensation occurs on the particles
(they are assumed insoluble).
II. Contact on plate 1
A. Particles are collected by impaction in the
bubble formation zone, still at size "d ".
Pi
B. Condensation occurs and particles grow to
"d, •'.
P2
C. Diffusiophoretic deposition removes some
particles from the gas in the froth layer
on the plate. Thermophoresis and centrifugal
deposition are neglected.
III. Contact on plate 2
A. Particles are collected by impaction in the
bubble formation zone at size "di
B. Negligible condensation occurs.
IV. Contact on subsequent plates has same characteristics
as plate 2.
Although the model is somewhat idealized, it is well within
the bounds of engineering accuracy and the precision of experi
mental measurements. One can always revise any stage of the
model with a closer approximation of parameters if he wishes
to examine the sensitivity of the method to parametric values.
In our evaluations of the model we found that further refinements
did not produce significant changes in predictions.
"
Pa
-------
- 144 -
D1FFUSIOPHORETIC DEPOSITION
Particle deposition by diffusiophoresis was described by
the following equation in our previous models 2 ' 3 ' 4 :
„
(1-7) Md-y) \dr
(1)
or,
where D~ = diffusivity of water vapor in carrier gas, cm2/s
b
MI = molecular weight of water, g/mol
M2 = molecular weight of non- transferring gas, g/mol
y = mole fraction water vapor, dimensionless
r = distance in the direction of diffusion, cm
The molecular weight and composition function represented
by "Ci", describes the effect of molecular weight gradient on the
deposition velocity corresponding to the net motion of the gas
due to diffusion (the "sweep velocity"). For water mole fraction
in air ranging from 0.1 to 0.5, "Ci", varies from 0.8 to 0.88.
We used a rough average of 0.85 for "Ci", for computing "u "
and consequent particle collection efficiency by integrating
over the period of condensation.
Whitmore5 concludes that the fraction of particles removed
from the gas by dif fus iophoresis is equal to either the mass
fraction or the mole fraction condensing, depending on what
theory is used for deposition velocity. In other words, it
is not necessary to follow the detailed course of the conden-
sation process, computing instantaneous values of deposition
velocity, and integrating over the entire time to compute the
fraction of particles collected. One can simply observe that
if some fraction of the gas is transferred to the liquid phase
it will carry along its load of suspended particles.
-------
- 145 -
We have used Whitmore's general concept but with two modifi-
cations. First, one can see from equation (1) that Whitmore's
theory would be comparable to assuming that the particles move
with the same velocity as the gas phase. We have chosen to
retain the correction for molecular weight gradient, which
means that we will compute the particle collection efficiency
as 85% of the volume fraction of gas condensing on the cold sur-
face.
The second modification concerns how to compute the proper
value of the volume fraction of gas condensing. The problem is
that not all of the condensate goes to the heat transfer surface;
some of it goes to the suspended particles. As will be shown in
detail later, the fraction of the condensate which causes particle
growth depends on several factors and ranged from about 0.1 to 0.4
of the total condensate for the range of parameters we explored.
If one is concerned only with diffusiophoretic deposition
the particle collection efficiency would, therefore, be 60% to
901 of that computed without accounting for condensation on
particles. In the case of a scrubber which also employs inertial
impaction the particles would be agglomerated to some extent by
the diffusional sweep, so they would have higher mass and be
easier to collect.
Without going into a detailed model of this phenomenon one
could use either of two simplifying assumptions ;
1. Assume that the condensation on particles causes no
agglomeration.
2. Assume that the condensation on particles causes
agglomeration and that the inertial impaction
efficiency is sufficiently high that all of the
particles swept to other particles will be collected
by impaction.
The first assumption will lead to too low an efficiency and the
second to too high an efficiency. However, the maximum differ-
ence between the two for a representative case of 25% of the
volume condensing and 25% of that going to the particles would
-------
- 146 -
be 5.3% (i.e., 0.25 x 0.85 x 0»25 x 100). This is a relatively
small effect compared to the other uncertainties.
PARTICLE GROWTH
Particle growth is dependent on how well the particles can
compete with the cold surface for the condensing water. There
are several transport processes at work simultaneously in the
condenser section of an F/C scrubber:
1. Heat transfer
a. From the gas to the cold surface
b. From the particles to the gas
2. Mass transfer
a. From the gas to the cold surface
b. From the gas to the particles
A mathematical model which accounted for these transport
processes in addition to particle deposition has been described
in EPA reports 2 ' 3 ' "* . The portions of that model relating to
particle deposition were deleted to provide a model which would
describe particle growth in the absence of deposition. The
basic relationships involved are as follow:
The rate of change of particle radius is given by a mass
balance ,
- k>PG (PG Ppi}. cm/s (3)
~dt ~ P
where:
2 D P
k' r = ^f—pr^—— = particle to gas mass transfer coeffi
p G p PBM cient, gmol/cm2-s-atm
pG = water vapor partial pressure in bulk of gas bubble, atm
PBM = mean Partial pressure to non-transferring gas, atm
r = particle radius, cm
TG = gas bulk temperature, °K
p.. = molar density of water, gmol/cm3
p . = water vapor partial pressure at vapor-liquid inter-
pl face, atm
-------
- 147 -
Particle temperature can be computed from an energy balance
where :
hpG
*
2V
P PP
-2-
M
(PG - Pp.)
PartlcJe to §as heat transfer coefficient,
cal/cm -s-°K
(6)
v;here C = heat capacity of particle, cal/g-°K
k = thermal conductivity of gas, cal/cm2 -s - °K/cm
LA. = latent heat of vaporization for water, cal/gmol
t = time, s
The overall energy balance for the gas-liquid interface is
given by :
at LM
i- TL>
PLi} Ap dZ
where k' = mass transfer coefficient, gas to liquid,
gmoI/cm2-s-atm
a = interfacial area for transfer
volume of scrubber, cm2/cm3
A = cross-sectional area of scrubber, cm2
P
= heat transfer coefficient, gas to liquid,
cal/cm2-s-°K
°K
= temperature
liquid hulk,
The equations given above are used along with enthalpy and
material balances for the total system of gas, liquid, and sus-
pended particles to form a mathematical model for condensation
and growth. The model was solved through a finite difference
method on an electronic computer for several situations which
are discussed below.
-------
- 148 -
PREDICTION OF CONDENSATION
The condensation model was used to predict the particle
condensation ratio, £ , (which is the fraction of the total
condensate which goes'to the particles) as a function of several
parameters. The conditions investigated are as follow:
1. Scrubber type - one sieve plate
2. Inlet gas - saturated from 310°K to 350°K
3. Water - uniform bulk temperature from 310°K to 325°K
4o Particle number concentration - 107 to 109/cm3
5. Particle diameter - 0.1 to 1.0 ym
6. Liquid phase heat transfer coefficient - 0.01 to 0.1 cal/cm2-s-°I
7. Condensation can occur when the gas is saturated
The computed values are plotted on Figures 3 through 7. As
can be seen, the figures show the folloitfing:
(Figure) 3. "f " does not depend much on "d "
4. "f " decreases significantly with "T, "
P L
5. "f " decreases with "Tp" to an extent which
P G
depends on "TT"
6. "fp" increases slightly with "n " above 107/cm3
7. "f " increases with "hL" up to "h «
* 0.1 cal/s-cm2-°K
It was found in computations not reported here that "f "
decreases significantly with "n " below about 106 particles/cm3.
Industrial emissions generally have particle number concentrations
on the order of 107, and greater. The liquid phase heat transfer
coefficient is an important parameter but unfortunately, predic-
tions of its magnitude vary considerably depending on which
correlation is used. The value of 0.1 appears to be the best
supported by the literature for mass transfer on perforated plates.
For a combination of parameters such as might be encountered
in a practical situation a value of "f " ~ 0.25 appears to be
reasonable. Given this one can compute the amount of particle
growth that will result from a given condensation ratio
(i.e., g water condensed/g dry gas = q" = condensation ratio).
If the particle size distribution and the scrubber character-
istics are known one can predict the overall penetration that
will be achieved in the scrubber.
-------
- 149 -
INERTIAL IMPACTION DURING BUBBLE FORMATION
During the formation of bubbles on a sieve plate the jets
of gas emerging from the perforations impact on the liquid.
Particles are thus deposited on the liquid surface by inertial
impaction. Particle collection can be determined from:
40 F2d2 p C1 u,
Pf =
where F = foam density, volume fraction liquid
d = particle diameter, cm
p = particle density, g/cm3
C1 = Cunningham slip correction factor, dimensionless
u, = gas velocity in the perforation, cm/s
vy = gas viscosity, poise
d, = diameter of perforation, cm
Pt - = penetration of particles of diameter d , fraction
PERFORMANCE PREDICTION METHOD
The sequence of steps to be followed in predicting the perfor-
mance of an F/C scrubber system involving a sieve plate column
is as follow:
1. Determine the initial particle size distribution.
2. Compute particle penetration from the saturator based
on the saturator collection efficiency characteristics and the
initial particle size distribution. No growth occurs in the
saturator .
3. Compute particle penetration due to inertial impaction
during bubble formation on the first plate. Use the
particle size distribution leaving the saturator and the collec-
tion efficiency relationship for sieve plate given in equation
(8).
4. Calculate the condensation ratio corresponding to the
scrubber operating conditions, from this compute "fv"j the
volume fraction of gas condensing, and then calculate the
penetration due to dif fus iophores is according to equation (9)
-------
- 150 -
for a conservative estimate or equation (10) for an optimistic
estimate
1 PtD = 0.85 (fy) (l-fp) (9)
1 PtD = 0.85 £y (10)
The diffusiophoretic penetration applies equally to all
particle sizes so it will not change the size distribution but
will decrease the particle concentration.
5. Determine the particle size distribution leaving the
condenser from the values of "q1" and "f ". Figure 8 is a size
distribution plot showing lines for the particles before and
after condensation. The conditions used for this plot were:
Initial d =0.75 ymA = dry mass median diameter
r &
a = 2.5 = geometric std. deviation
n = lOVDNcm3 (Dry Normal cm3, @ 0°C)
q' = 0.3 g/g
f =0.25
P
6. Compute the particle penetration function for the
remaining stages of the scrubber, based on the penetration for
one stage given by equation (8). The penetration for a given
particle diameter on one stage is "Pti". For "N" stages of
equal efficiency the penetration for a given particle diameter
"U+. N"
is "Pt-
N"
7. Use the relationship between "Pt. and "d " from step 5
and the grown size distribution from step 5 to compute the
overall penetration due to inertial impaction after growth.
8. To summarize, the total overall penetration for the
F/C scrubber Pt will be the product of the following:
a. "Pt " due to impaction in the saturator
b. ""PtV," due to impaction in the condenser
c. "~PF " due'to diffusiophoresis in the condenser
d. "Pt\" due to impaction in stages after the condenser.
Thus ,
Pt = Pt x PTU x PT x Ft,
abed
-------
- 151 -
COMPARISON WITH DATA
Predictions based on the model described above were made
for a number of experimental determinations on a 100 m3/min
sieve plate F/C scrubber. Some examples of comparisons of
predictions with experimental results are shown in Figures 9
and 10, The correlation shown in Figure 9 is the best obtained
and that in Figure 10 is representative of the average case.
CONCLUSIONS
The revised model is much simpler to use than our previous
version and it appears to give very good predictions. It also
offers the opportunity for easy modification to suit specific
situations. The fraction condensing on the particles is a key
quantity and it needs more study for plates and other contacting
systems.
REFERENCES
1. Calvert, S., J. Goldshmid, D. Leith, and D. Mehta.
Scrubber Handbook. A.P.T., Inc., EPA Contract No.
CPA-70-95. NTIS No. PB 213-016. August, 1972.
2. Calvert, S., J. Goldshmid, D. Leith, and N. Jhaveri.
Feasibility of Flux Force/Condensation Scrubbing for
Fine Particulate Collection. A.P.T., Inc., EPA
Contract No. 68-02-0256, NTIS No. PB 227 307.
October, 1973.
3. Calvert, S., and N. Jhaveri, and T. Husking. Study of
Flux Force/Condensation Scrubbing of Fine Particles.
A.P.T., Inc., EPA Contract No. 68-02-1082. August, 1975.
4. Calvert, S., and S. Yung. Study of Horizontal Spray
Flux Force/Condensation Scrubber. A.P.T., Inc., EPA
Contract No. 68-02-1328, Task No. 10. July, 1976.
5. Whitmore, P.J. Diffusiophoretic Under Turbulent
Conditions, Ph.D. Thesis, University of British
Columbia, 1976.
-------
CLEAN GAS
5S°C
WATER —£>
WATER —J>
WATER
mil
i,ooo"c
O.OlE/g
Figure 1. Generalized F/C Scrubber System.
, CLEAN PAS
i
COLD
IVATE'i — >>
V.'ATUR
1
HOT ^,
c-vs y
. V
SAT. ->
^^r~~^~^-
--s-Jr-
_|l']ate_2
•N-JU-^^-— H-w->— ->^v^-
Pln^e )
-~Jli.
^^--(Imp.)
^— (F/C 5 fi)
— »- ( 1 mp . )
NJ
1
Figure 2. Multiple plate !•/(- s c rubbc r system.
-------
1.0
_j I IIr
_ TC - J45°K (72°C)
n - IO'/DN cm1
_ h. = 0.L cal/s-cm!-°K
r. o.s
0.1
= 51 0°X ( 37°C,
- 32(]°F (47°C,
0.5
PARTICLK niAMETER, urn
1.0 I. I
Figure 3. Effect of particle diameter on condensation
rat io .
o
<
1.0
0.5
0.1 —
n - IO'/DN cm1
T » 34S°K
urn
d - 0.5
h. - 0.1 cal/s-cm'-°K
I
I
I-1
Ln
I
305 310 315 320 320
LIQUID BULK TEMPERATURE, "K
330
Figure 4. Effect of liquid bulk temperature on condensation
ratio.
-------
1 .0
0.5
O.i
I
\
n - lO'/CN cm1
d •= 0 .5 urn
P
)I » 0.1 cal/s-cm2 -°
310°K
320°K
330 335 340 345 350
GAS INLF.T TF.MPERATUIIF.. °K
3SS
Figure' 5. Effect of gas inlet temperature on con Jensat i on
ratio.
1.0
£ 0.5
o.i
I I I I I I I
d - 0.5 pm
• 0.1 cal/s-cm!-0K
i 1 i 1
1 1
i i i i i i i
10'
10' 10'
PARTICLh S'UMllLR CO'JCLMKAT I ON ,
Figure 6. Effect of particle number concentration on
condensation ratio.
I
h-*
(Jl
I
-------
0.1
3.0
- P
TG
n
P
0.5
0. 1
1 I I 1 T
0.5 |.m
345°K
10'/DNcm]
I 1 I I I
0 0.01 0-05 0.1
LIQUID HEAT TRANSFER COEFFICIENT, h^
Figure 7. Effect of liquid heat Transfer coefficient
on condcnsation ratio.
1.0
; o.s _
0.1
— i
Ul
Ln
Figure 8. Particle size distribution before and after
condcnsat ion.
-------
0.5
0.3
0.2
0.1
0.05
0.03
I I I I I
EXPERIMENTAL
PREDICTED
I I I I 1 I
0.3 (1.5 1.0 2.0
PARTICLE DIAMETER, ymA
3.0
Figure 9. Particle penetration versus aerodynamic
diameter for Run 64.
0.5
0.3 —
0.2 -
0.1 —
r 0.05 -
0. 02
0.3
en
I
0.5 1.0
PARTICLH niAMLTKR,
3.0
Figure 10. Particle penetration versus aerodynamic
diameter for Run 69.
-------
- 157 -
Discussion
Prof. Weber said experience indicated that a few particles were used as
a condensation nucleus. Hot and cold streams of air were mixed after
which spontaneous condensation, a disadvantageous process, occurred.
The experiments reported in the paper were not, according to Dr. Calvert,
with hot and cold air. Instead, hot gas came into contact with water,
thus producing a gradual temperature gradient and a liquid film.
Experimental data on condensation scrubbing was consistent with the high
concentration of particles. The model fitted with the data from pilot
studies and laboratory experiments. Dr. Holighaus inquired whether
condensation occurred at all outside the water surface. Dr. Calvert
replied that cold drops of water were vaporized into saturated gas, each
water drop having the same heat capacity, super-saturation and condensation
of the particles took place and air mixed with a boundary layer. Dr.
Holighaus pointed out that this layer consisted of heat and mass.
Mr. Calvert said that the contrary had been proved by laboratory experiments
with plate columns and sprays: condensation was present, and an important
mass transfer coefficient in gas to heat occurred.
-------
- 158 -
APPLICATION OF HIGH GRADIENT MAGNETIC SEPARATION
TO FINE PARTICLE CONTROL
Charles H. Gooding
Research Triangle Institute
Energy and Environmental Research Division
P. 0. Box 12194
Research Triangle Park, NC 27709
Presentation for the
USA/FRG Participate Workshop
Julich, West Germany
March 16-17, 1978
-------
- 159 -
INTRODUCTION
Several widely-used industrial processes, primarily in the iron and
steel and ferroalloy industries, emit large quantities of waste gas
containing magnetic particles. Particulate emissions from these processess
are presently controlled with varying degrees of success by conventional
technologies such as electrostatic precipitation, wet scrubbing, and
fabric filtration. In the last decade research and commercial applications
have demonstrated that high gradient magnetic separation (HGMS) is an
effective and economical method of removing small, weakly-magnetic
particles from selected liquid streams[l], and generalized theory
indicates that the process should also be applicable to the control of
fine, magnetic particle emissions from industrial stacks.
BASIC CONCEPT OF THE PROCESS
In essence HGMS is an enhanced filtration process. The funda-
mental concept is the collection of small particles on ferromagnetic
fibers that are immersed in a uniformly applied magnetic field. The
ferromagnetic fibers induce regions of highly non-uniform magnetic field
intensity, and the particles are attracted to the fibers' surface by
magnetic force.
In its most simple practical form, the high gradient magnetic
separator consists of a canister packed with a fibrous, ferromagnetic
material such as steel wool (Figure 1). The canister is located in a
magnetic field that is normally generated by a solenoid, and the magnetic
traction force provides high-efficiency collection of particles as the
gas passes through the canister. When the collection matrix becomes
fully loaded, the magnetic force is removed and the particles may be
flushed from the matrix with a pulse of pressurized air. This can be
accomplished by using a system of several modules such as the one
depicted in Figure 1 in a parallel-flow, cyclic mode of operation. That
is, each module undergoes a period of filtration. Then the flow is
diverted to other modules; the magnetic field of the loaded module is
-------
MAGNET
COIL
PARTICLE-LADEN
GAS IN
wi>'« *<••••"!;. ••> *'•»'. v- '"(••• fcl -•c-*:-J* >V
M&^t^*$&X;&$
STAINLESS
STEEL WOOL
MATRIX
o
I
CLEAN GAS
OUT
Figure 1. Schematic representation of a hinh gradient magnetic separator.
-------
- 161 -
deenergized; the matrix is cleaned; the field is reenergized; and
filtration resumes. An alternate continuous scheme, which results in
better utilization of the magnetized volume, is to construct the magnet
and matrix so that the dirty matrix can be removed from the magnetized
region into a non-magnetized cleaning region and continuously replaced
by clean matrix material without interrupting the filtration process.
High gradient magnetic separation is widely used on a commercial
scale in the clay industry to remove fine paramagnetic color bodies from
slurries of kaolin[l]. Laboratory and pilot-scale investigations have
been conducted to assess other liquid system applications in mineral
processing[2,3], wastewater treatment[3], and coal cleaning[4,5,6].
Several of these programs are current, and other commercial applications
of HGMS seem likely to occur over the next few years.
POTENTIAL APPLICATIONS FOR FINE PARTICLE CONTROL
With current magnet technology the capital costs and power require-
ments of large, iron-bound solenoids make HGMS potentially competitive
with other particulate control methods. Since the filtration process is
enhanced by the magnetic force, the void volume of the collection
matrix can be much larger than in a conventional filter, allowing very
high gas velocities at a relatively low pressure drop. This combination
translates into a potential reduction in energy requirements compared to
conventional particulate control techniques, even though production of
the magnetic field requires some energy. High operating velocities help
to reduce both the capital costs and space requirements of the equipment.
Furthermore, the process as developed up to this stage is completely dry
so it should avoid the water pollution problems associated with some
scrubber installations. Magnetic stainless steels of the 400 series can
be used as a collection matrix to make the process compatible with high
temperature and corrosive environments, and the absence of any sparking
mechanism in the collection process should allow its application in
combustible gas streams.
-------
162 -
The magnetic susceptibility and size distribution of the dust
particles are the key parameters that will determine the practicality of
fine particle emission control applications although other gas characteristics
could affect economics to a lesser extent. Magnetic susceptibility of
particulate matter cannot be predicted quantitatively from composition
data alone, but the percentage of iron is a qualitative indicator. With
relatively high iron concentrations even submicron particles should be
collected efficiently and economically. Reported data on the particle
size distribution and composition of dusts emitted from several processes
in the iron and steel are shown in Table I. Emissions from ferroalloy
processes are much more diverse, but the production of several alloys
including silico-manganese, ferromanganese, and ferrochrome, results in
the emission of particulate containing significant quantities of iron as
well as other strongly magnetic species. All of these processes should
be considered potential candidates for HGMS fine particle control.
DESCRIPTION OF COMPLETED EXPERIMENTAL WORK
The competitive methods of particulate emission control delineate
practical constraints on the design of an HGMS device for stack gas
applications. Technology is currently available to control particulate
emissions from most industrial sources with a capital investment for
uninstalled primary equipment no greater than $8500 per cubic meter of
gas flow per second ($4/cfm). Power consumption is normally less than
o
3.2 kw/m /s (2.0 hp/lOOOcfm) in precipitators and fabric filters, but
can be many times greater in difficult applications where high-energy
wet scrubbers must be employed. One can deduce from these general
criteria and the cost and power requirements of conventional HGMS equip-
ment that an HGMS control device would have to operate satisfactorily
with a superficial gas velocity of at least 5 m/s (1000 ft/min), a
pressure drop of less than 2.0 kPa (20 cm H20), and an applied magnetic
field of less than 1.0 T in order to be competitive with other control
methods in most applications.
At the beginning of this study bench-scale experiments were conducted
to gain a preliminary evaluation of the practicality of the process.
-------
TABLE I. EXPECTED CHARACTERISTICS OF UNCONTROLLED GAS STREAMS FROM SEVERAL PROCESSES
Process
Sinter Machine
Windbox
Discharge End
Blast Furnace
Basic Oxygen Furnace
Open System
Closed System
Electric Arc Furnace
Open Hearth Furnace
Dust
Concentration
9/m3
1-2
5-12
10-25
10-25
40-70
0.2-7
4-7
Mass Median
Diameter
ym
10
10
100
1
15
1
5
Iron
Composition
% Total Fe
25-50
25-50
35-50
55-70
55-70
15-40
55-70
Noteworthy Gas
Characteristics
5-15% HpO, hydrocarbons,,
flourides, SOX, 120-180°C
120-180°C
20-40% CO, 2-6% H2,
200-300°C
250-300°C
75% CO, 250-300°C
40-1 20°C
7-15% H20, 250-350°C
Scarfing Machine
0.5-1
0.5
50-70
H20 saturated, 50-60 C
-------
- 164 -
Dust from an industrial basic oxygen steelmaking furnace (BOF) was
dispersed in an air stream and passed through a loosely-packed steel
wool matrix contained in an 8.9 cm canister, which was positioned in the
bore of an iron-bound solenoid. The unit was operated with superficial
gas velocities up to 10.6 m/s (2100 ft/min), and high efficiency collection
of dust was achieved with appl-ied fields of 0.3 T or lower. Pressure
drop through the matrix was normally less than 2.5 kPa (25 cm. H20).
Figure 2 shows the dramatic reduction in the penetration of particles
through the matrix that was achieved when relatively low magnetic fields
were applied. These preliminary experiments confirmed that the HGMS
process could be successfully applied to collect gas-borne particles and
provided data from which a larger unit and a more systematic experiment
were designed.
The layout of the second-phase, pilot-scale HGMS system is depicted
schematically in Figure 3. Dusts from a BOF as well as an electric arc
steelmaking furnace (EAF) were dispersed into a wind tunnel, and a 0.8
o
m /s (UOOcfm) slipstream was drawn off and processed through the HGMS
pilot plant. A low efficiency cyclone was placed upstream of the HGMS
to remove uncharacteristically large agglomerates that were not adequately
broken up by the dispersion system. The magnetic separator consisted of
a 30-cm diameter iron-bound solenoid surrounding a canister of loosely-
packed 430 stainless steel wool. The ranges of the experimental operating
parameters are given in Table II. The experiments were systematically
designed so that the effects of individual parameters could
TABLE II. RANGES OF OPERATING PARAMETERS
IN HGMS EXPERIMENTS.
Applied Field 0 - 0.4 T
Matrix Packing Density 0.005 - 0.010
Matrix Length 15 - 30 cm
Superficial Gas Velocity 5.2 - 11.1 m/s
(1020 - 2185 ft/min)
Gas Temperature 25 - 110°C
-------
- 165 -
1.000
0.500
ZERO FIELD
0.200
O O.IOOt-
LJJ
2 0.050
UJ
Q-
0.020
0.010
B0 = 0.094 T
BQ = 0.214 T
OD05_
50.0
80.0
o
LU
90.0 o
u.
Li.
HI
95.0
2
O
H
O
LLJ
O
O
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
PARTICLE DIAMETER,
1.8
98.0
99.0
—'99.5
2.0
Figure 2. Bench-scale collection of BOF dust with a gas
velocity of 8.4 m/s (1650 ft/min) and a
pressure drop of 1.7 kPa (17 cm water).
-------
TO
BAGHOUSE
WIND TUNNEL
FROM
'BAGHOUSE
-C&-
FLOW (T
SAMPLE
CYCLONE
HGMS
SAMPLE
WASTE
DUST
T
AIR
Figure 3. Schematic representation of pilot-scale HGMS facility.
-------
- 167 -
be studied. Experimentally determined magnetization curves for the two
dusts are shown in Figure 4.
After the operating conditions were established for a particular
run, the fractional penetration of dust particles through the HGMS was
determined as a function of particle size by using cascade inertial
impactors to measure the concentration and size distribution of the dust
upstream and downstream of the matrix. Since the impactors required
approximately 90 minutes to collect an adequate sample, an optical
(light-scattering) particle sizing device was also used to ensure that
no significant upsets or transients occurred during the impactor sampling
period.
PILOT-PLANT EXPERIMENTAL RESULTS FOR TWO DUSTS
The experimental efficiency with which the two dusts were collected
under identical operating conditions is shown in Figure 5. As expected,
the more strongly magnetic EOF dust was collected more efficiently than
the EAF dust. The curves drawn in Figure 5 are predictions of a theoretical
model. Basically the model predicts single-fiber collection efficiencies
from a solution of the equations of motion that yields particle trajectories
The single-fiber efficiencies, R , are then extended to predictions of
total matrix penetration, P, using the equation
P = exp[-EFLRc/a(l-F)]. (1)
F is the matrix packing density; L is the matrix length; a is the fiber
radius; and E is an "effectiveness factor," which was included to allow
for deviations from the idealized assumptions of the model. Reasonable
2
assumptions and geometric arguments predict the value of E to be 4/ir ,
but reduction of experimental data from all of the runs showed the BOF
and EAF data to be better fit by E values of 0.09 and 0.07, respectively.
It should be noted that the single-fiber collection efficiency can be
-------
168 -
250
APPLIED FIELD, T
300 350
APPLIED FIELD, A/m x 1(T3
Figure 4. Magnetization curves of two steel industry dusts.
-------
- 169 -
1.0
0.5
2
O
0.2
o.
CC
£ 0.05
UJ
a.
0.02
0.01
0.005
O.I
B0 = 0.40 T
V = 8.2 m/s
F = 0.005
L = 15 cm
EAF
RUN NO. 7181
BOF
RUN NO. 6281
50
80
90
o
y
o
u.
LU
95 O
99
o
UJ
o
u
98
0.2
0.5 1.0 2
PARTICLE DIAMETER,
10
99.5
20
Figure 5. Experimental data and theoretical predictions of HGMS
collection of two dusts under identical conditions.
-------
- 170 -
greater than one because of the magnetic traction force. The develop-
ment of the model is described in more detail elsewhere[7,8].
The effects of individual parameters on particle collection were
found to be in reasonable agreement with theoretical expectations. The
particle size and magnetic susceptibility dependence are illustrated in
Figure 5. At lower fields and higher velocities, the penetration of
larger particles tends to be greater than predicted. The reason for
this observation is not yet fully understood, but it may be due to
detrimental inertial effects that contribute to particle bounce and
reentrainment. Gas velocity has a relatively small effect on the collection
of submicron particles, and higher velocities may actually be beneficial
by enhancing inertial impaction in those cases in which the single-fiber
collection efficiency is less than one.
Higher magnetic fields enhance the collection of particles, but the
effect is diminished as both the particles and matrix approach magnetic
saturation. With dusts exhibiting magnetic properties similar to those
shown in Figure 4, collection efficiency can probably be improved more
economically by increasing the density or length of the collection
matrix rather than increasing the applied field beyond 0.4 to 0.5 T. The
experimental data confirm the effects of matrix density and length
expressed in Equation (1).
Increasing the operating temperature could adversely affect particle
collection since the gas viscosity (and hence the drag force on the
particles) would be increased, and magnetization of the particles and
matrix may be diminished. However, no significant effect of temperature
could be discerned from a few experimental runs that were made at the
higher temperature level used in this study.
Figure 6 demonstrates that both of the dusts studied in this work
can be collected with high efficiency. The collection efficiency of the
larger particles was not as high as the present model predicts (particularly
the high-velocity EOF run) but was still greater than 99 percent. Based
on currently available, continuous-cleaning HGMS equipment, projections
-------
z
LU
Z
1.0
0.5
0.2
0.1
0.05
0.02
o.oi
0.005
o.oo;
0.001
0.0005
0.0002
0.000
BOF RUN NO. 02102
B0 • 0.300 T
V - 9.85 m/t
F =• 0.008
L » 22.5 cm
O.I 0.2 O5 10 2
PARTICLE DIAMETER,
i
10 20 0.1 0.2
EAF RUN NO. 11161
B0 - 0.375 T
V - 7.32 m/i
F = 0.005
L = 30 cm
0.5 I.O 2 5
PARTICLE DIAMETER, fi
50
80
90
95
98
X"
o
z
LU
O
99 u-
LU
99.5
99. e
99.9
99.95
99.98
tu
O
O
10 20
99.99
I
i—>
-J
M
I
Figure 6. High-efficiency collection of BOF and EAF dusts: theory and experimental data,
-------
- 172 -
for a full-scale, high-efficiency EOF dust collection device predict an
3
uninstalled capital cost of $8200/m /s ($3.86/cfm) and power requirements
of 3.2 kW/m3/s (2.0 hp/lOOOcfm). Collection of the EAF dust would be
slightly more expensive. These costs estimates were projected from a
singular design and could possibly be lowered by optimization techniques.
CONCLUSIONS AND CONTINUING WORK
The results of this investigation indicate that HGMS may be an
efficient and economical method of particulate emission control in
selected applications in which relatively high susceptibility dust must
be collected. Several processes in the iron and steel and ferroalloy
industries are potential candidates for fine particle control by HGMS.
The theoretical model provides a valuable tool to screen potential
applications, to evaluate alternative designs, to plan experiments, to
analyze data, and to conduct economic analyses.
Additonal pilot plant data are being obtained to broaden current
experience with the process and to quantify requirements for matrix
cleaning. Refinement of the theoretical model is also continuing. A
mobile pilot plant is being designed so that experiments can be conducted
at industrial sites.
-------
- 173 -
REFERENCES
1. Oder, R. R., High Gradient Magnetic Separation Theory and Applications,
IEEE Trans. Magn., Vol. MA6-12(5), Sept., 1976, pp. 436-443.
2. Murray, H. H., Beneficiation of Selected Industrial Minerals and
Coal by High Intensity Magnetic Separation, IEEE Trans. Magn., Vol.
MAG-12(5), Sept., 1976, pp. 498-502.
3. Oberteuffer, J. A., Engineering Development of High Gradient
Magnetic Separators, IEEE Trans. Magn., Vol. MAG-12(5), Sept.,
1976, pp. 444-449.
4. Oder, R. R., Magnetic Desulfurization of Liquefied Coals: Conceptual
Process Design and Cost Estimation, IEEE Trans. Magn., Vol. MAG-
12(5), Sept., 1976, pp. 532-537.
5. Liu, Y. A., and C. J. Lin, Assessment of Sulfur and Ash Removal
from Coals by Magnetic Separation, IEEE Trans. Magn., Vol. MAG-
12(5), Sept., 1976, pp. 538-550.
6. Maxwell, E., D. R. Kelland, and I. Y Akoto, High Gradient Magnetic
Separation of Mineral Particulates from Solvent Refined Coal, IEEE
Trans. Magn., Vol. MAG-12(5), Sept., 1976, pp. 507-510.
7. Lawson, W. F., W. H. Simons, and R. P. Treat, The Dynamics of a
Particle Attracted by a Magnetized Wire, J. Appl. Phys., Vol.
48(8), pp. 3213-3224, August, 1977.
8. Gooding, C. H., T. W. Sigmon, and L. K. Monteith, Application of
High Gradient Magnetic Separation to Fine Particle Control, EPA-
600/2-77-230. National Technical Information Service, Springfield,
Va., 1978.
-------
- 174 -
Discussion
In answer to Mr. McCain's question regarding methods of cleaning HGMS
filters, Mr. Gooding replied that particles could be blown off the
fibers by pulses of air. This method has been demonstrated in the
laboratory, and work is being undertaken to design a unit to do this in
a practical way. With high concentrations of dust in the gas the
frequency of cleaning might be excessive. However, in the case of lower
dust concentration gas streams, prospects are good.
Dr. Wiggers asked whether the number of cleanings was limited; Mr. Gooding
said that although long term evaluation was needed, it appeared that the
durability of the matrix against corrosion would be good. He clarified,
at the request of Dr. Wiggers, that the matrix is made of inexpensive
stainless steel wool.
Mr. Parker inquired whether fly ash from coal-fired boilers could be
controlled. A few tests had been performed on eastern low sulfur coal,
whose iron content was significant, replied Mr. Gooding. The results
were not as good as those in the case of a steel mill dusts for which
the process described in the paper was more suitable.
Mr. Donovan wondered if adhering magnetic particles could function like
the collection matrix and enhance magnetic collection, or whether it
were essential to remove all particles frequently. Mr.Gooding said that
with weakly adhering magnetic particles the particles will begin to blow
off as the matrix is loaded, resulting in inefficient collection.
However, longterm tests with more strongly magnetic particles indicated
constant collection efficiency: the wire grew and remained strongly
magnetic. However, the pressure drop increased making cleaning necessary
eventually.
-------
- 175 -
Advanced Dust Collection Techniques in the Federal Republic of
Germany: Selected Examples and Research Priorities.
Dipl.-Ing. G. Guthner
Umweltbundesamt
Bismarckplatz
D-1000 Berlin 33
Within the scope of the First General Administrative Regulation
under the Federal Imission Control Law (TA Luft 74) the parti-
culate emissions of 35 stationary sources are restricted to va-
lues ranging from 20 to 300 mg/m . In addition to the source-re-
lated standards for 55 hazardous materials limitations are pro-
vided; these materials are classified into three categories
ranging from 20 to 75 mg/m , respectively (table 1). This material
related provision favours the applicaction of those collection
techniques which yield high collection efficiencies in the fine
particulate range at modest energy consumptions; those techniques
are mainly fabric filters and electrostatic precipitators, not at
least wet electrostatic precipitators (WEP's). The new ruling should
have its major impact in the metallurgical and chemical industry -
The following example, a Wet-EP application in a Tin-refining pro-
cess might be typical: The refining facility consists of crubible
furnaces in which the molten tin is treated with gaseous Chlorine.
The flue gas which escapes from the furnaces contains Chlorides
and Oxides of the impurities (mainly Lead and Zink) and trace of
Chlorine; it used to be blended with the flue gases from a Tin scap
melting furnace containing organic aerosols. The TA Luft restricts
emissions of Lead compounds to 20 mg/m^, of Zink compounds to 50
mg/m and of gaseous inorganic chlorine compounds to 30 mg/m (cal-
culated as CL ). It has been decided to install a WEP because pilot-
Venturi scrubbers and pilot Dry electrostatic precipitators have not
performed satisfactory. It can be gathered from table 2 that ghe dean
gas dust concentration of the full size unit is only 2 to 5 mg/m
and far below the design value of 20 mg/m . The concentration of ga-
seous chlorine has been determined at 1O - 15 mg/m , but this value
seems arbitrary, because it has been measured only once at a ph-value
of some 2,5 (instead of 6 to 8 at regulas conditions). The precipitator
-------
- 176 -
is of the parallel plate type with stainless steel internals and
casing. A major design feature is the possibility to remove the
electrodes easyly, because their life time should not exceed two
years, due to corossion attack by the Chlorides.
Explanation of Slides. Emission control of an anode baking fur-
nace in a primary Aluminum plant is another sucessful application
of the WEP's. The flue gas characteristics of such a furnace are
listet in table 3 (column one for the raw gas and column 5 for the
clean gas). The design requirements have been 90 % collection effi-
ciency for SO-, 95 % for HF and 97 % for tar. Two precipitators in
line, one operating dry for tar collection only, and the second
equipped with continously spraying nozzles to remove gases and tar
simultaneously have been installed. Both precipitators are of the
parallel plate type, with plane collection plates and regid discharge
2 3
frames. The specific collection area is some 70 m /(m /sec) in both
precipitators. SO-- and HF removal is achieved by a double alcaline
absorption process; i.e. the primary (scrubbing) circuit is operating
with Na OH, which is being recovered by Ca (OH)~ in the secondary
(water treatment) circuit.
Heavy built up and corrosion problems have been encountered during
start up of the plant. The built up problem has been solved by
treating the recovered NaOH with CO~-containing flue gas, thus lowe-
ring its Ca-ion contents. To cope with corossion it has been necessa-
ry to line the casing with fiber glass coating and to replace the
mild steel internals by those from stainless steel. The actual price
for such an installation should be some DM 7O/ (Nm /h) for condi-
tioning tower, precipitators, fans and ducts. Inclusion of the water
treatment system will raise the price to DM 100/(m3/h). Energy con-
sumption without water treatment system is some 2,8 kWh/1OOO m . The
water treatment system will add some 0,5 kWh/1000 m3. The benefits
of such an installation are - besides from airpoluttion control -
that no waste water has to be drained, that the sludge can be dumped
and the tar be reused for anode making after some additional treat-
ment by the tar supplier.
Explanation:
Slide 3+4 Anode Baking installation.
Application of fabric filters for control of gaseous and particulate
-------
- 177 -
emission by use of dry agents is a well known techniuqe for Alu-
minum reduction furnaces. The use of additives not for gaseous
emission control but to render possible bag cleaning has been
successfully tested with the emission control system of a glass
melting furnace. The need to develop the new technology has arisen
when dedusting of flue gases from a Lead-Boron-Silicone glass fur-
nace turned out to be impossible because the bags of a conventionally
operated fabric filter have been clogging repidly. The suitability of
Dolomite, Calciumhydroxide and Alumina (Al^O., x H_O) for additives
has been tested with a pilot plant. Each of these materials serve
as raw material for the glass furnace. It turned out that only Alu-
mina could be used, because with Calciumhydroxide the dosing screws
have been plugged and Dolomite has caused rapid wear of the bags.
The operation data of the full size unit are listed in table 4.
Major components of the installation are the plate cooler (to re-
duce the gas temperature from 700 °C to 200 °C)., two pulse jet
filters with fans in parallel and the additive-dosing system. Each
filter and fan are capable to handle the total gas flow. The addi-
tive is fed into the system before the cooler by a metering screw.
Whilst dust concentration of the raw gas is some 0,5 to O,7 g/m
it is encreased by the additive feed to 4 to 6 g/m . The clean gas
dust concentration is less than 5 mg/m^. This result should be com-
pared with the TA-Luft s
and Boron, respectively-
3 3
pared with the TA-Luft standards of 20 mg/m and 75 mg/m for Lead
An explanation for the excellent dust cake removal may be given by
electron microscope photographs, which reveal that the submicron
Lead-Boron sublimates are aggregated to the large 1O to 30 Alu-
mina particles in thin uniform layers, but they do not stick to each
other. It is another advantage of this process that the dust from
the hoppers can be reused for feed of the furnace, which has not
been possible at comparable installations without additives.
Explanantion.
Some TA Luft provisions have strongly influenced the design of
electrostatic precipitators for pov/er stations, too. The TA Luft
rules that Particulate emission from coal fired water tube boilers
-------
- 178 -
be restricted to 150 mg/m , even if one field of each chamber (i.e.
parallel section) is out of service. So far there are two operating
600 - 700 MW utility power stations in the Federal Republic whose
precipitators have been designed by consideration of these require-
ments. Some design and operation data are presented in table 5.
The most outstanding design feature are large specific collection
areas, a fare-going sectionalization (No. of HV-groups) and collec-
tion plate hights with up to 14,7 m. What is not mentioned in the
table are the very tough warranty requirements in terms of corrosion
prevention, rigidness of electrode frames and plates and on life time
of electrodes; e.g.: no more than 8 wire ruptures during the first
year of operation are accepted with installation I and no more than 10
ruptures for 2 years with installation II. (O.O12 % and O.01 %, res-
pectively). But the performance is remarkable too, because the average
clean gas dust concentrations are less than one third of the TA-Luft
standard with both installations.
It should be mentioned that the present 700 MW dry bobbom boilers
have been preceeded in the early 70 ties, by 350 MW wet bottom
boilers. The precipitators to those installations have also been
designed to meet the 150 mg/m standard but with two fields and
2 3
spez. collection areas of 75 m /(m /sec) only. As the emission of
suchan installation prooved to be considerably higher than antici-
pated a third field had been retrofitted, thus increasing the collec-
7 "3
tion area to 1OO m (m /sec);
Explanation
The R&D activities on dust collection technology - under funding by
the Umweltbundesamt - are strongly applicationrelated. The major
objecitves of any of these projectes are extension of the respec-
tive technology to new applications and minimization of overall ex-
penditure related to efficiency.
On electrostatic precipitators a pilot study is being conducted to
evaluate whether farther sectionalization could be a means to in-
crease overall migration velocity. For this purpose the sparkover
Voltage distribution will be measured in a pilot precipitator and
in a full size unit and the feasibility of appropriate design modi-
fications be studied.
-------
- 179 -
Within the scope of a second project on precipitators the influence
of passage width on migration velocity shall be studied. This investi-
gation has mainly been initiated to prove findings of some researchers
and a precipitator manufacturer predicting an increase of migration
velocity with passage width (at constant field strength). Confirmation
of these ancitipations would offer an interesting approach to lower
the overall expenditure of precipitators.
Scrubbers, in particular Venturis shall be optimized by means of
pilot tests a various sources, mainly in the metallurgical industry.
The major design features to be varied will be throat shape and water
supply configuration. Collection efficiency and power requirement in
depencence upon operation mode and design will be studied. In addition
criteria for the transfer of pilot results to full size units shall
be developed.
Another pilot investigation is being conducted with fabric filters.
This project has mainly been conceived to extend the application range
to new sources, in particular in the metallurgical and chemical
industry and to industrial coal-and Oil-fired boilers. Another ob-
jective is the evaluation of collection efficiency in dependance of
bab deterioration and mode of bag cleaning.
Any of these projects has been conceived not only to improve the
technology but also to produce data on fine particle emission by means
of particle size distribution measurements.
-------
- 180 -
Tabli
Emission Standards far Hazardous Materials in the Federal Republic
of Germany
Category I — (m>0.1 kg/h) — c £ 20 mg/m
Category II — (m > 1 kg/h) — c < 50 mg/m
Category III — (m > 3 kg/h) — c < 75 mg/m3
m = ram gas mass-flow
c = clean gas dust concentration
Material: Category:
Aluminum carbide III
Aluminum nitride III
Ammonium compounds III
Antimony and its soluble compounds *) II
Arsenic and its soluble compounds *) I
Asbestos I
Barium sulfate III
Barium compounds if soluble *) II
Beryllium and its soluble compounds *) I
Bitumen III
Boron trifluoride IT
Boron compounds, if soluble *) III
Lead and its soluble compounds *) I
Cadmium and its soluble compounds *) I
Calcium arsenate I
Calcium cyanamide jjj
Calcium fluoride II
Calcium hydroxide III
Calcium oxide m
Chromium compounds, if hexavalent I
Cristobalite with particles smaller than 5 Lim II
Fluorine compounds, if soluble *) I
Fluorspar II
-------
- 181
M_S_lL!L?Li_!Li_i Category :
Iodine and its compounds II
DiatomacGous earth II
Cobalt and its compounds II
Copper and its soluble compounds *) III
Copper fume I
Magnesium hydroxide III
Magnesium oxide III
Molybdenum and its soluble compounds *) III
Nickel I
Nickel carbonate I
Nickel oxide I
Nickel sulfide I
Phosphates III
Phosphorus pentoxide I
Quartz with particles smaller than 5 ,um II
Mercury and its compounds, except cinnabar I
Soot II
Selenium and its soluble compounds *) I
Silver compounds, very soluble, e.g. silver nitrate*) II
Ferrosilicon III
Silicon carbide III
Strontium and its compounds II
Tar II
Cutback pitch II
Tellurium and its soluble compounds *) I
Thallium and its compounds I
Tridymite uiith particles smaller than 5 urn II
Uranium and its compounds I
Vanadium and its compounds I
Bismuth III
Tungsten and its compounds, except tungsten carbide III
Zinc and its compounds II
Dusts of organic compounds, e.g. anthracenes, aro-
matic amines, 1,^-Benzoquinone, naphtalene II
*) Soluble compounds are those materials uhich are soluble in the
respiratory and digestive tracts, on the surface of the skin or
in the absorbing organs of plants to such a degree that they can
cause hazardous effects.
-------
- 182 -
Gas t'faw [rn*lh1
Gas Temperafum L°Cl
Dust Ccnct* ntrotion fay/ml
WEP Inlet
1Q.QQQ
~*6Q
2.000
WEP Outlet
10.000
-JO
ft/far current
N° of fields
Liquid to g&s ratio
Energy consumption
Control by
[ kWhj t
7S
o,2 -Of
2
50
UBA
19®
Wet Electrostatic Przcipitator to a
Tin Refining Process
2
-------
£or flow Lm3ltil
Gos iempzKitum [*C]
S02 fcnglrn*]
HF [mg/ni3]
Tar [rng/rnj
Dust [mgfrr?3!
Tlzsjqn Data
JhtefC&rTfoc)
fa 000
110
1.000
100
tSoo
yf5o
Outlet I
Etidirginst
~*UO
4oo
J
So
SO
Y£P
New insf.
~nf(7ar) and Wei Electrostatic Jrecipitotcrs
-------
- 184 -
f&krz ccoler
~kmp
-------
- 185 -
Capacity i MW]
Suffur Conknt cfcool LKJ
Spec, Cc( feet/on arecr fn*J(fa*/to$
i -i
N° of "sfrvzfs" in psmlfel
N° of fields in tim
A/° erf high Vokge groups
Dust Cone, hefarz E$P j&//7?37
Tfost Cone, offer £SP imslm1]
tffafar gwvrikiKt
^-f- 1 t^j(^j^) — »
rneosmd
^flM^ atKtinrrfeecf
*Jlli*£rHJj ^J*^" "'
measamcf
ricrk hfQhf- Cm]
l&ssaqe Mth [tn]
Installation I
£50
10
] tss
*
*
i€
5*
¥5
< 45
M2
ft,?
0,3
Insfaftcrtionj
72o
0,6
/J&
8
U
J2
/ff"
•MM
/^ -5P
K
^^
fJ
Ot3
UBA
19%
Zkctrodafic Tnzapifator ibr (/ti//fy
Towzr Station Emission Control
r
-------
- 186 -
Discussion
Dr. Donovan asked whether the alumina was added after the cooler,
or if it was added where the gas temperature was 7OO C. Mr. Guthner
replied that it was advantageous to add the alumina before the coo-
ler, and said (in reply to a second question by Mr. Donovan) that
the adhering dust is recycled. Statistically, every particle is re-
cycled 1O times. The fate of the recycled particles was queried by
Mr. Donovan, who was informed that they are fed into the furnace,
and that the alumina additive becomes part of the glass product.
Mr. Gooch wished to know how SO,., removal was accomplished in wet
precipitators„ Dr. Guthner answered that this was achieved by means
of sophisticated, continuously-spraying inlet nozzles, which ensured
saturation of the gas as it entered the electric field.
Mr. Princiotta wondered whether workers hat encountered any sulphur-
scaling or plugging problems, and was told that the calcium ion con-
tents of the recovered sodium hydroxide was up to 2 grams per liter.
Mr. Princiotta added that excess oxygen led to too much sodium sul-
phate and drew attention to double alkali technology which helps
eliminate scaling. Dr. Guthner referred to the pH factor (5 - 6)
in de-scaling, and said that chlorine ion had been proved to be
equally satisfactory.
-------
- 187 -
SESSION III: CONCURRENT REMOVAL OF DUST AND GASEOUS CONSTITUENTS
-------
- 188 -
S02 REMOVAL BY A FABRIC FILTER USING NAHCOLITE INJECTION
by
R. P. Donovan
Research Triangle Institute
Energy & Environmental Research Division
Process Engineering Department
P. 0. Box 12194
Research Triangle Park, N. C. 27709
(Draft text for presentation to be made to the Participate Workshop,
Jiilich, Federal Republic of Germany, March 16-17, 1978)
March 1978
-------
- 189 -
S02 REMOVAL BY A FABRIC FILTER USING NAHCOLITE INJECTION
R. P. Donovan
Research Triangle Institute
Energy & Environmental Research Division
Process Engineering Department
P-. 0. Box 12194
Research Triangle Park, N. C. 27709
Reagent injection is the name applied to the technique whereby a
reagent is injected into a gas flow for the purpose of converting a
gaseous species in that flow into a solid product that can be removed
along with other particulate matter by a downstream fabric filter. The
application of highest EPA interest is the removal of S02 from flue gas
in a dry scrubbing process.
While a variety of reagents have been investigated for this function,
(Table 1), the most promising system now is that based on nahcolite
injection. Nahcolite is a naturally occurring mineral consisting of 60
to 90 percent sodium bicarbonate. In the Piceance Basin of Colorado it
exists in quantities measured in the billions of tons. However, present
production is zero [Ref.l]. Most nahcolite must be recovered by under-
ground deep mining and will involve large startup expenses. Before
undertaking these costs, mine owners are seeking to obtain purchase
commitments from users (primarily utilities) who at present are uncertain
of the value of the product in their process. Given these uncertainties
neither the utilities nor the miners are willing to risk the sizable
investment required for production and it is this impasse that EPA is
now seeking to break by sponsoring both analytic and pilot scale evaluations
of the nahcolite injection process.
PROCESS OUTLINE
The chemical reactions between injected nahcolite and SOp in the
flue gas are thought to consist of the following reactions occurring
simultaneously [Ref.2]:
-------
- 190 -
2NaHC03(s) + S02(g) £ Na2S03(s) + 2C02(g) + H20(g) (1)
2NaHC03(s) £ Na2C03(s) + H20(g) + C02(g) (2)
NaC03(s) + S02(g) + Na2S03(s) + C02(g) (3)
Na2S03(s) + 1/2 02(g) -> Na2$04(s) (4)
In the typical process the nahcolite ore is crushed and ground to a
fine mesh (say, -200 mesh on the Tyler scale [material passes through a
sieve whose openings are 75 vim]). This fine powder is then injected
into the gas flow upstream of the fabric filter. The ground nahcolite
reacts according to the chemistry outlined above, the solid products
(Na2S03, Na2S04 plus unreacted NaHC03 and Na2C03) being caught by the
fabric filter. The reactions take place anytime after injection including
after being collected by the fabric filter since the gas flow passes
through the filter cake on the bags.
Reaction 1 represents a direct interaction between NaMCO, and SO-.
Note that two moles of NaHC03 are required to chemically balance one
mole of S02- These relative quantities define a stoichiometric ratio of
one; i.e., a stoichiometric ratio = 1 implies that two moles of NaHC03
have been injected for every mole of S02 in the gas flow.
An alternative reaction sequence for S02 removal is given by Reactions
(2) and (3). An equilibrium is established between NaHC03 and Na,,C03
(Reaction 2) and a reaction of S02 with Na2C03 takes place to again
produce the Na2$03 product (Reaction 3). Again two moles of NaHC03 are
required for every mole of S02 reacted by this sequence.
BACKGROUND
EPA (actually N.C.A.P.C., a predecessor government agency) involvement
with reagent injection began nearly 10 years ago with field experiments
at the Mercer Generating Staion of New Jersey Public Service Co. [Ref.3].
-------
- 191 -
These tests consisted of feeding a slipstream from a coal-fired boiler
into a four compartment pilot baghouse. Sodium bicarbonate was added to
the flow at various points upstream of the baghouse. Among the variables
investigated, in addition to point of NaHCOo addition, were
(1) stoichiometric ratio [NaHC03/S02J,
(2) temperature,
(3) flow rate,
plus various operating modes (batch feed versus continuous feed, various
cycling times). The primary measures of performance were the percent of
(outlet SOp concentration \
inlet S02 concentration x 100j and the peCcent
/f NaHCO, in spent additive] \
of NaHC03 utilized in the reactionMl - ^ ^^ added Jx looj.
Temperature and stoichiometric ratio proved to have the most significant
impact on S02 removal efficiency (Figure 1) and NaHC03 utilization (Figure 2).
Some dependencies on nahcolite feed mode (continuous feed or batch feed--
either periodically to coincide with the resumption of filtering after bag
cleaning or one shot at the beginning of a test run) may exist; these
dependencies were not as strong as those on temperature or stoichiometric
ratio.
While moving the feed point upstream (and thereby increasing the time
the nahcolite is in the gas stream) did produce increased SC^ removal and
nahcolite utilization, these gains were also small and based on analysis of
fallout before the bags and the dust cake on the bags, much of the reaction
between NaHCCL and S02 (88 percent in this case, although this value varies
widely among different investigators) took place while the gas passed
through the filter cake on the bags. One of the conclusions of this early
work was that increased efficiency and utilization would be favored by a
system that increased the deposition of additive on the bag. Subsequent
experience shows the interaction to be more complex than just precoating
the bags, however. Common practice is to precoat with part of the additive
load and feed the rest into the upstream gas flow.
-------
- 192 -
A second EPA-sponsored study in reagent injection was performed by
researchers at Owens-Corning Fiberglas Corp. [Ref.4j. Most of the
reagents studied required high temperature for optimum reaction (Table
1). The low temperature reagents, alkalyzed alumina and nahcolite, both
were assumed to be impractical without reagent regeneration. American
Air Filter Co. has independently followed up this approach by developing
a regenerative process for nahcolite [Ref.5]. During regeneration
slaked lime is used to recover the sodium salts, the calcium being
disposed of as waste. Figure 3 summarizes the chemical reactions. The
regeneration step reduces the nahcolite consumption by over two orders
of magnitude. It is replaced by the lime and C02 requirements of the
regeneration but lime is cheap and C02 is readily available from the
scrubber flue gas. Economic feasibility of the overall regeneration
cycle is not yet demonstrated, however.
This American Air Filter work also showed the important influences
of temperature and stoichiometric ratio upon S02 removal efficiency.
Moreover a dependence upon water concentration was documented by them as
shown in Figure 4. Increases in water vapor concentration above 5
percent seem unimportant but as the Figure 4 data show, SOo removal
efficiency suffers at say 2 percent water vapor concentration. Happily
the flue gas of interest typically has water vapor concentration above 5
percent.
EPA-SPONSORED STUDY AT TRW, INC. [Ref.6]
The regenerative process is not the approach currently being sponsored
by EPA. EPA is investigating the economics of a nahcolite throwaway
process. The reacted nahcolite is discarded along with the flyash.
Because these sodium salts are highly soluble, their disposal constitutes
an additional environmental problem. Conceivably adequate environmental
safeguards may require insolubilization of the sodium sulfate product
prior to disposal. Coprecipitation with acidic ferric ion to form
insoluble double salts NaFe3(S04)2(OH)6 and Na2Fe(S04)2OH • 3H20 is one
candidate process [Ref.7]. Such environmental considerations depend
heavily on the locale and while clearly important in determining process
feasibility, cannot be specifically assessed independently of a specific
site.
-------
- 193 -
The conditions defined for assessment in the EPA-sponsored study at
TRW are given in Table 2 [Ref.6].
Flue gas at the boiler exit is assumed to consist of the following
constituents and proportions, which are typical for western coal-firing:
N2 - 74%
02 - 4.8%
C02 - 12.3%
S02 - variable
S03 - 1.0% of the S02 value
NO - 0.06%
HC1 - 0.01%
Hf\ n Q°/
y\J ~ O.O/b
The limitation of distance from the coal source (Table 2) immediately
confines the applicability of the process evaluated to the Western
United States as shown in Figure 5. Also listed in this Figure are the
major coal deposits within that area. Coal properties of each of these
sources are tabulated in Table 3.
The state in which each of these deposits lie makes a difference
because the standards governing emissions vary from state to state
(Table 4). By matching the coal sulfur content with the appropriate
state regulations, the required S02 removal for legally burning the coal
in the state in which it is found can be deduced. With 70 percent S02
removal (Table 2) half of the 10 sources would be legal. The highest
required S02 removal is for the Sheridan deposit which needs 91 percent
SO- removal in order to be combusted within the state of its location.
An additional uncertainty of the evaluation is a pending Federal S02
standard calling for 90 percent S02 removal regardless of absolute value
but in no cases to exceed the already existing standard. Table 4 lists only
-------
- 194 -
the existing Federal standard. The impact of this anticipated Federal
change is not yet clear. A 90 percent removal efficiency would require
more nahcolite or a higher reaction temperature or a combination of
both.
Assuming the conditions of Table 2 and estimating the cost of
delivered nahcolite to be $32.50/ton and that of disposing of the
flyash/spent nahcolite in a landfill to be $6/ton, cost comparisons of
the nahcolite injection process with the lowest cost, wet scrubbing
technologies are given in Table 5. These estimates are for conditions
favorable for the nahcolite process. They are for 1 percent sulfur
coal. At higher sulfur concentrations both capital (Figure 6) and
operating (Figure 7) costs rise rapidly. Nahcolite injection is un-
likely to compete favorably with lime or limestone scrubbing on boilers
fired with high sulfur eastern coals. Western coals also typically have
higher ash content than eastern coals so that the added spent nahcolite,
while significant, is not as great an increase in disposal burden. If a
landfill is employed, as in this assessment, having a low water table
lowers the costs of isolating the highly soluble sodium salts from the
environment. Again in the Eastern United States one would not be so
favored.
The estimates in Table 5 ignore the particulate removal requirements,
The fabric filter is primarily a particulate removal device and no added
costs are anticipated. The wet scrubbing process may entail additional
costs for particulate control that are not reflected in Table 5. When
this is true, the dual pollutant control functions of the fabric filter
would become a major economic advantage.
Even without adjustment for particulate control the dry scrubbing
process appears to be competitive in the Western United States. The
economic advantage is not great but EPA now plans to jointly sponsor a
field assessment of the process with the City of Colorado Springs. This
assessment will begin before the end of the year and some preliminary
data should be available in early 1979.
-------
- 195 -
In addition both the ongoing fabric filter evaluations at Kerr
Industries and Southwest Public Services (both sponsored by EPA) in-
clude options to pursue nahcolite injection for S02 control. That
these options will be exercised by the contractors is not yet known
and may depend in part upon the experiences at Colorado Springs and
at Arapahoe/EPRI and Wheelabrator-Frye/Superior Oil where similar work
is being sponsored by other groups.
-------
- 196 -
REFERENCES
Mcllvaine, R.W., "SO? Removal with Fabric Filters," pp.8-1 to 8-31
in the Proceedings of the Second International Fabric Alternatives
Forum, Denver, CO, July 27-28, 1977, American Air Filter Co., Inc.,
215 Central Avenue, Louisville, KY 40277.
Genco, J.M. and H.S. Rosenberg, "Sorption of S0? on Ground Nahcolite
Ore," J. Air Pollution Control Assn. 26., No.10, October 1976, pp.
2.
989-990.
3. Chaffee, R.L. and H. Liu, "Evaluation of Fabric Filter as Chemical
Contactor for Control of Sulfur Dioxide from Flue Gas," Final
Report-Part 1, HEW Contract No. PH 22-68-51 (NTIS PB 194-196), 28
August 28, 1969, the Air Preheater Co., Inc., Wellsville, NY
14895.
4. Veazie, F.M. and W.H. Kielmeyer, "Feasibility of Fabric Filter as
Gas-Solid Contactor to Control Gaseous Pollutants," Final Report,
Contract No. PH-22-68-64 (NTIS PB 195 884), August 1970, Owens-
Corning Fiberglas Corp., Granville, OH 43023.
5. Doyle, D.J., "Fabric and Additive Remove S0?," Electrical World,
February 15, 1977, pp. 32-34. L
6. Christman, R.C., et a!., "Evaluation of Dry Sorbents and Fabric
Filtration for FGD," Final Report, Contract No. 68-02-2165, in
preparation, trw Environmental Engineering Division, 800 Follin
Lane, S.E., Vienna, VA 22180.
7. Genco, J.M., et a!.. "The Use of Nahcolite Ore and Bag Filters for
Sulfur Dioxide Emissions Control," J. Air Pollution Control Assn. 25_,
No.12, December 1975, pp. 1244-1253.
-------
At 17G-100°C
b « batch feed of NaHCG>3 (2 pts)
All other X cither continuous feed
or batch fed by cycle (13 pts)
A/C: 1-2 cm/soc
X
X
80
70
60
M
o
C/5
50
X
HO
Xb
40
X
HO
X
o
At 133-144°C
H = > 1300 ppm (3 pts)
L = ~ 850 ppm (3 pts)
For all other O, (6 pts)
1000
-------
60
50
140
<
N
2
P
3
^ 30,
20:
10 U
X
X
A/C: 1-2 cm/sec
X
00
I
o
O
1.0
2.0
STOICHIOMETRIC RATIO
Fi(jiiro 2. NnHCO3 ulili/ntioii rnto [not 31.
3.0
4.0
-------
- 199 -
CANDIDATE REGENERATION CHEMISTRY [REF 5]
(1) 2NaHCO3 + SO2^-Na2SO3 + 2CO2 + H2O
(3) Na2CO3 + SO2-^-Na2SO3 + C02
(4) Na2S03 + Ca{OH}2-5»2NaOH + CaS04
(5) Na2CO3 + Ca(OH)2-3-2NaOH + CaCO3
(6) NaHC03 + Ca(OH)2-^NaOH + CaCO3 + H2O
(7) NaOH + CO2—
(8) 2NaOH + CO2— >Na2CO3 + H2O
Figure 3.
-------
- 20O -
20
«
Q-
tr
O
)
5 60
CM
O
80
100
LEGEND
T H20
121°C(250°F) 2%
121°C(250°F) 7%
93°C(200°F) 7%
Batch Continuous
FA>3 FA<2 FA>3
a
0
D
O
121° C, 2% H20
93° C. 7% H20
1 2 3
STOICHIOMETRIC RATIO
Figure 4. 862 removal efficiency [Ref. 5].
-------
NORTH DAKOTA } M1NNESOTA
1 - BOULDER
2 - GUNNISON
3 - LAS AMI MAS
4 - CARBON
5 - VALLEY
6 - SOCORRO
7 - PERKINS
8 - CARBON
9 - LINCOLN
10 - SHERIDAN
CO
o
Figure 5. Major coal deposits within 1200 km radius of nahcolite mine [Ref 6].
-------
70-
60
tt
j 60-
O
Q
CD
? 50-
8 40-
o
30-
20-
500 MW BOILER
(TABLE 2 CONDITIONS)
0.5
1.0
l
1.5
I
2.0
PERCENT SULFUR IN COAL
Figure 6. Capital costs versus percent sulfur content [Ref 6].
KJ
O
NJ
-------
30 -4
V)
or
o
Q
CO
CO
O
o
u
cc.
LLJ
Q.
O
20 -I
10 -J
500 MW BOILER
(TABLE 2 CONDITIONS)
I
K)
O
U>
I
I
0.5
I
1.0
PERCENT SULFUR IN COAL
1.5
2.0
Figure 7. Operating costs versus percent sulfur content [Ref 6],
-------
TABLE 1
CANDIDATE REAGENTS FOR DRY S02 SCRUBBING [REFS 1,4,6]
REAGENT
SLAKED LIME
PROMOTED SLAKED
LIME (1% NaCI)
SLAKED DOLOMITE
LIME
PROMOTED DOLOMITE
LIME (1% NaCI)
MANGANESE DIOXIDE
ALKALIZED ALUMINA
NAHCOLITE
AMMONIA GAS
MAGNESIA
FLUE GAS TEMP. RANGE
REMARKS
371-482°C(700-900°F)
OPTIMUM TEMP. (427°C) TOO HIGH
260-371°C(500-700°F)
HIGH COST REAGENT; NOT VERY
TEMP. DEPENDENT OVER THIS
RANGE
149-260°C(300-500°F)
VERY PROMISING
149°C(300°F)
REACTION PRODUCTS HIGHLY
SOLUBLE IN WATER
<88°C«190°F)
VISIBLE PLUME AND LOW
EFFICIENCY ABOVE 88°C; EXCELLENT
EFFICIENCY BELOW 88°C
135°C-900°C
NOT SUFFICIENTLY REACTIVE AT
CONTEMPORARY BAGHOUSE
TEMPERATURES
NJ
O
-------
- 205 -
TABLE 2
CONDITIONS EVALUATED [REF. 6]
500 MW BOILER
WESTERN COAL, 1% SULFUR, 10% ASH
HEATING VALUE: 24.4 MJ/kg (10,500 Btu/lb)
1,200 km RADIUS OF COLORADO DEPOSIT
SEMI-ARID REGION; WATER TABLE 15 m BELOW THE
SURFACE
204° C (400° F) BAGHOUSE TEMPERATURE
70% SO2 REMOVAL
-------
- 206 -
TABLE 3
COAL PROPERTIES [REF 6]
STATE
COLORADO
MONTANA
NEW MEXICO
SOUTH DAKOTA
UTAH
WYOMING
COUNTY
BOULDER
GUNNISOIM
LAS AN i MAS
CARBON
VALLEY
SOCORRO
PERKINS
CARBON
LINCOLN
SHERIDAN
COAL TYPE
SUBBIT.
BIT.
BIT.
SUBBIT.
LIG.
BIT.
LIG.
BIT.
BIT.
SUBBIT.
S %
0.27
0.43
0.70
1.1
1.3
0.82
1.2
0.6
1.0
1.1
ASH %
5.4
3.4
12.8
11.2
9.1
13.8
9.0
5.6
5.5
7.9
HEATING VALUE
•«*• (t)
23.3
31.4
31.4
24.4
15.6
28.6
14.0
29.1
30.9
20.9
(10,000)
(13,500)
(13,000)
(10,500)
(6,700)
(12,300)
(6,000)
(12,500)
(13,300)
(9,000)
-------
- 207 -
TABLE 4
S02 EMISSION REGULATIONS FOR SELECTED STATES [REF 6]
STATE
FEDERAL EMISSION
. REQUIREMENTS,
g SO2 PER MJ HEAT INPUT
(Ib SO2 PER
106 Btu HEAT INPUT)
STATE EMISSION
REQUIREMENTS,
g SO2 PER MJ HEAT INPUT
(Ib S02 PER
106 Btu HEAT INPUT)
ARIZONA
COLORADO
KANSAS
MONTANA
NEBRASKA
NEVADA
NEW MEXICO
NORTH DAKOTA
SOUTH DAKOTA
TEXAS
UTAH
WYOMING
0.52
0.52
0.52
0.52
0.52
0.52
0.52
0.52
0.52
0.52
0.52
0.52
(1.2)
(1.2)
(1.2)
(1.2)
(1.2)
(1.2)
(1-2)
(1-2)
(1.2)
(1.2)
(1.2)
(1.2)
0.34
0.13 (0.3)
0.52
0.52
0.52
0.090
0.15
0.52
0.52
0.52
(0.8)
or 150 ppm
(1.2)
(1.2)
(1.2)
(0.21)
(0.34)
(1.2)
(1.2)
(1.2)
80% REMOVAL
0.09
(0.2)
-------
- 208 -
TABLE 5
FLUE GAS DESULFURIZATION COST COMPARISON [REF 6]
PROCESS
DRY SORBENT
BAGHOUSE
LIME SCRUBBING
LIMESTONE SCRUBBING
CAPITAL
COST
$/kW
46
49
55
ANNUALIZED
COST
Mills/kWh
2.6
2.8
2.8
-------
- 209 -
Discussion
In response to Dr. Holighaus1 request for further information regarding
Figs. 1 and 2 and reaction velocity, Mr. Donovan said that the figures
showed measurements in which sodium bicarbonate had been injected into
the flue gas of the power plant. Two methods of injection were possible:
a single shot or a continuous process. With respect to reaction velocity,
residence time was varied by changing the point of injection. Performance
correlations were the strongest with temperature and stoichiometric ratio,
as shown in Figs. 1 and 2, although more data may show additional, less
pronounced dependencies such as residence time.
Dr. Davids asked:if the nahcolite technology met State emission
standards, and whether compliance depended on the coal's sulphur content.
Mr. Donovan answered that compliance depended on both the sulfur content
of the coal and the SO- standards set by a particular state. He drew
attention to the fact that compliance would be achieved in about half
the Western states assuming the sulfur content of the coal burnt corre-
sponded to that naturally occurring in that state.
Dr. Hubner mentioned the possibility of formation of bi-sulphide
from sodium sulphate; Mr. Donovan replied that the reactions postulated
were semi empirical in that reaction products were sampled at various
points. No bi-sulphide was detected. Dr. Hubner's second inquiry dealt
with differences in the properties of nahcolite and sodium bicarbonate.
Mr. Donovan answered that his analysis treated nahcolite and sodium
bicarbonate as chemically synonymously except for concentration - -
nahcolite was assumed to be 70% sodium bicarbonate and 30% inert
material.
-------
- 210 -
PERFORMANCE TESTS OF THE MONTANA POWER COMPANY COLSTRIP STATION FLUE GAS
CLEANING SYSTEM
Joseph D. McCain
Southern Research Institute
2000 - 9th Avenue South
Birmingham, Alabama 35205
U.S.A.
This paper gives a description of a scrubber used for joint collection of
sulfur oxides and particulate matter produced by coal combustion at a large
coal fired electrical generating station. This novel system utilizes the alka-
linity of the flyash produced by the boiler for the major portion of the sulfur
dioxide removal. The scrubber was designed and constructed by Combustion Equip-
ment Associates (New York, NY, USA) in cooperation with the Bechtel Power Cor-
poration and A. D. Little, Inc., for the Montana Power Company. The system is
currently in commercial operation on units 1 and 2 of the Colstrip Station of
the Montana Power Company. Each unit has a generating capacity of 360 MWe.
The scrubber utilizes a variable throat venturi to permit operation at a
constant pressure drop (nominally 43 cm w.c.) over the full range of possible
boiler gas flow rates. The scrubbing liquor is a recirculating flyash slurry
containing 12% solids by weight. The venturi section is followed in order by
a spray-type absorber, a washtray system for diluting entrainment, and a chevron-
type mist eliminator. Each boiler is equipped with three parallel scrubber
modules, each of which is capable of handling 40% of the total, full load gas
flow from the boiler.
Performance data are given for both SO2 and flyash removal by the system.
-------
- 211 -
PERFORMANCE TESTS OF THE MONTANA POWER COMPANY COLSTRIP STATION
FLUE GAS CLEANING SYSTEM
INTRODUCTION
The flue gas cleaning system (Figure 1) now in operation on the two Col-
strip 360 MWe units is unique in that a wet scrubbing system is used for both
particulate and SO 2 control and that captured ash provides the alkalinity for
the SO 2 removal. This paper provides a description of the operation of the
scrubber and the results of performance testing carried out to determine the
and fly ash cleaning efficiencies achieved by the system.
DESCRIPTION OF THE SCRUBBER*
The system currently installed on the two 360 MWe Units 1 and 2 is illus-
trated in Figures 1 and 2. The hot flue gas leaving the boiler is cooled in
the heat recovery air heater and enters the flue gas scrubbing system at about
300°F. Each scrubber module, as shown in simplified drawing in Figure 2, con-
sists of a downflow venturi scrubber centered within an upflow spray tower con-
tactor. The venturi is equipped with a variable throat to maintain constant
pressure drop at variable loads. In the venturi the scrubbing liquid is finely
dispersed by the high velocity flue gas and serves to efficiently wet and trap
the particulate fly ash. In the spray tower the gas contacts a recycle spray
of absorption slurry. The slurry from the venturi and the spray contactor is
collected and held in the base of the scrubber and recirculated at an L/G ratio
2 H/m3 (15 gal/1000 ft3) for venturi and 2.4 H/m3 (18 gal/1000 ft3) for the
absorber spray. An agitator in the scrubber base serves to maintain suspension
of the fly ash and solid reaction products. Slurry is bled from the recycle to
maintain a 12% suspended solids concentration. Slaked quick lime is added as
lime slurry only if needed to augment the fly ash alkali and maintain the desired
slurry pH.
Each scrubber module is designed to clean 120 MW of equivalent gas flow
under normal conditions and 144 MW under emergency conditions (i.e., when one
module is down, the two in operation will clean the amount of flue gas genera-
ted at 80% of boiler design load.)
*Taken from a paper by C. Grimm, J. Z. Abrams, W. W. Leffmann, I. A. Raben,
and C. Lamatia. Presented at the 1977 National Meeting of the AIChE.
-------
THE MONTANA POWER CO. PUGET SOUND POWER & LIGHT
2 - 360 MW COLSTRIP UNITS 1 & 2
MERGENCY WATER
LUMB BOB
FUYASH POND
FIGURE 1. THE MOTANA POWER CO. - PUGET SOUND AND LIGHT
COLSTRIP UNITS 1 AND 2 -{360 MW EACH) FLUE GAS CLEANING SYSTEM.
-------
- 213 -
RECYCLE HOLD-UP TANK
8 MINUTES TURNOVER
FIGURE 2. COLSTRIP SCRUBBER MODULE.
-------
- 214 -
The treated gas leaving the spray section passes through the water wash-
tray which serves to trap and dilute the entrainment. The gas leaving the wash-
tray passes through a chevron demister followed by a mesh pad demister and leaves
the absorption section water-saturated and cooled to the saturation temperature
of about 120 °F.
To preclude condensation in the fan and stack, and improve the gas buoyancy,
the cooled gas from the scrubber is reheated 50 to 75°F by a steam-heated exchan-
ger . The warmed gas then passes through the dry induced draft fans and is dis-
charged to the atmosphere from the top of a 500 foot stack.
As shown in Figure 1 the slurry discharged from the absorption loop is
passed to an intermediate retention pond where the solids settle and from which
the clarified water is returned to the absorption system. At intermittent in-
tervals (currently only during the warm summer months) , a floating dredge is
used to reclaim the settled solids from the intermediate settling pond and trans-
port them as a 30% slurry by pipeline to the remotely located permanent dispo-
sal pond. Decanted water (supernate) from the disposal pond is returned, also
intermittently, through the same slurry pipeline to the intermediate pond for
recycle to the absorption system. No stabilization of the sludge is required
and a closed water loop is maintained.
Fresh water is added to the absorption system in an amount equivalent to
that evaporated into the warm gas stream plus that retained in the waste sludge.
This fresh makeup water is introduced to the system as dilution water for mini-
mizing the calcium saturation level in the mist eliminator washwater. This
washwater is trapped by and withdrawn from the washtray and circulated to a
small pond where entrained solids are separated. A portion of the water from,
this pond is returned and used to wash the undersurface of the washtray. Another
portion of the flow is diluted with the fresh makeup water, and used for bottom
wash of the mist eliminator.
The scrubber has been free of scale while the pH of the recycle liquid re-
mains in the expected range. Corrosion problems in the reheater and demister
plugging have not been experienced with the installation.
The successful operation of the Colstrip system as described above repre-
sents the culmination of an extensive development program carried out jointly by
the architect engineer, Bechtel Power Corporation, the scrubber system supplier,
Combustion Equipment Associates, Inc. (CEA) , and the power plant owners, Montana
Power Company and Puget Sound Power & Light Company.
A study was made by Bechtel of the possible options for meeting particu-
late and SOa removal standards.
A detailed chemical analysis of the fly ash (see Table 1) revealed that it
contained alkali metal oxides in an amount theoretically sufficient to react
with and adsorb the sulfur dioxide produced by the coal combustion. Laboratory
experiments simulating absorption conditions revealed that this alkalinity was
only usable under low pH absorption conditions (<5.6) . It also revealed that
-------
- 215 -
Table 1
FUEL AND ASH AS DESCRIBED IN SPECIFICATIONS
COAL: Average, As Received
Moisture 23.87%
Volatile Matter 28.59%
Fixed Carbon 38.96%
Ash 8.59% (Max. 12.58%, Min. 6.1%)
Heating Value 8843 Btu/lb. (Min. 8162 Btu/lb.)
Sulfur .777% (Max. 1.0%, Min. 0.4%)
ASH: (Estimated composition, sulfur trioxide-free basis)
Si02 41.60%
A12O3 22.42%
Ti02 0.79%
Fe203 5.44%
CaO 21.90%
MgO 4.95%
Na2O 0.31%
K2O 0.13%
P205 0.41%
(balance unidentified)
Later fly ash data varies slightly from above as follows:
LEACHED IN HZO (1% Fly Ash)
pH 11.8
Conductivity 4.150
Total Dissolved Solids 930 ppm
Calcium 396 ppm
Magnesium 0 ppm
Chloride 15 ppm
Sulfate (30,, = ) 30 ppm
LEACHED IN HC1
% Acid insolubles(SiO2) 57.59
% Calcium as CaO 22.00
% Magnesium as MgO 1.27
% Aluminum as A12O3 15.59
% Iron as Fe2O3 4.97
% Sulfate as SO,, 0.71
% Carbonate as CO3 0.70
-------
- 216 -
absorption under these low pH conditions would result in extensive oxidation of
the absorbed SO2 producing calcium sulfate rather then calcium sulfite as the
predominant reaction product.
Continued laboratory tests were conducted by Bechtel to determine the
process conditions under which the alkalinity of the fly ash could be utilized
while at the same time accommodating the scaling potential of the calcium sul-
fate. The conditions selected were a pH of 5 to 5.6, low enough for alkali uti-
lization and high enough for adequate SO2 absorption capability. The other,
and perhaps the key operating factor, was the use of a high level of suspended
solids in the absorption slurry (12 to 15% by weight, of which some 3-4% is
calcium sulfate formed in the absorption). This provided a high concentration
of calcium sulfate seed crystals to promote desupersaturation. A long residence
time for the recycle slurry in a stirred tank external to the scrubber was also
proposed to ensure alkali utilization and to provide crystallization of calcium
sulfate under controlled and non-scaling conditions. A slurry holdup of 8-10
hours was selected based on bleed rate.
The parameters for the final scrubber design are given in Table 2.
The above two conditions, i.e., low slurry pH and long contact with the
oxygen-containing flue gas, provided substantially complete oxidation. This
high oxidation was shown to improve the disposal characteristics of the waste
sludge produced.
Table 3 compares scrubber availability and plant load for the two units
during the time period September 1975 through December 1976. Note the defini-
tion of scrubber availability below the table. These generating plants have no
bypass capability around the air pollution control system.
PERFORMANCE EVALUATION
The scrubber performance was evaluated by Southern Research Institute
(SoRI) for the Industrial Environmental Research Laboratory of the U.S. Environ-
mental Protection Agency during the month of May 1977.
This evaluation was one of a series of studies being conducted by the In-
dustrial Environmental Research Laboratory of the Environmental Protection Agency
to identify and test novel devices which are capable of high efficiency collection
of particulates. The test methods used may not have been consistent with com-
pliance-type methods, but were state-of-the-art techniques for measuring mass
and fractional efficiency using standard mass train and inertial, electrical,
and optical methods.
The results of previous testing by other organizations and agencies showed
that the scrubber was capable of providing gas cleaning efficiencies substan-
tially in excess of those required to meet both S02 and particulate emission
standards. The results of some of these tests are summarized in Table 4.
-------
- 217 -
Table 2
DESIGN PARAMETERS FOR THE CEA VARIABLE THROAT VENTURI SCRUBBER
(COLSTRIP APPLICATION)
Venturi Pressure Drop 43.2 cm w.c. (17 in.)
Venturi L/G 2 £/m (15 gal/1000 ACF, saturated)
Absorption Spray L/G 2.41 l/m. (18 gal/1000 ACF, saturated)
% suspended solids in recirculating
slurry, by weight 12%
Residence time in the recycle tank 8 minutes
Gas velocity in mist eliminator zone 2.65 m/sec (8.7 ft/sec)
Wash tray pressure drop 9.65 cm w.c. (3.8 in.)
Mist eliminator pressure drop 2.5 cm w.c. (1 in.)
Reheat pressure drop 5.6 cm w.c. (2.2 in.)
Total system pressure drop (including
reheat) 64.8 cm w.c. (25.5 in.)
Total scrubber pressure drop (less
reheat) 55.4 cm w.c. (21.8 in.)
-------
- 218 -
Table 3
SCRUBBER AVAILABILITY VS. PLANT LOAD
Monthly Capacity
Factor %
UNIT 1 2
Sept. 1975 0.5
Oct. 19.4
Nov. 42.2
Dec. 59.9
Jan. 1976 63.8
Feb. 65.4
Mar. 57.0
Apr .
May
June
July
Aug.
Sept.
Oct.
Nov.
Dec.
49.9
26.0
0.0
28.0
37.8
64.5
73.1
55.6
67.2
1.3
23.2
19.5
13.0
64.6
77.0
79.7
82.3
No. Days
On Line
1 2
3
19
24
30
28
26
24
28
14
0
20
23
30
30
30
31
3
16
13
10
30
31
30
31
Avg. MW
for Days
On Line
1 2
50
139
203
239
265
273
277
219
210
0
167
194
239
281
225
249
66
171
180
162
232
298
303
297
Scrubber
Availability %
1 2
90.0
98.0
97.6
74.2
96.8
-
93.2
94.7
88.6
79.9
62.7
73.8
100.0
99.7
98.7
95.8
98.3
90.3
94.7
92.5
Note: Scrubber availability = total module hours available divided by three
times the number of hours in month. May through August base is days in
operation because of extended scheduled outages.
-------
Table 4
Emission Test Results - EPA Method
1. Required by NSPS (358 MW)
2. Scrubber Guarantee (358 MW)
3. Projected from Pilot Plant (358 MW)
a) 0.78%S (760 PPM), 8.19% Ash
b) 1.0%S (965 PPM), 12.58% Ash
4. UNIT 1 TESTS:
COAL AS RECD.
LB/HR
4063
3386
1394
2071
S02
PPM
510
425
185
260
LB/MMBtu
1.2
1.0
0.41
0.61
LB/HR
339
207
130
184
PARTICULATE
LB/MMBtu
0.10
0.06
.038
.054
%OPAC
20
20
20
LB/HR LB/MMBtu
2370
(1)
2370
2370
0.7
(1)
0.7
0.7
I
NJ
5.
Date
2/76
4/76
7/76
9/76
12/76
UNIT 2
10/76
11/76
12/76
Notes: 1.
2.
3.
GEN MW %Sul.
353 0.83
210 0.71
184 0.64
186 0.62
223 0.94
TESTS:
331 0.56
327 0.59
324 0.64
%Ash
9.03
7.79
8.49
7.93
8.54
7.96
7.86
7.87
NO Emissions guaranteed by
Avg. EDC monitor opacity.
Qualified observer.
Btu/LB
8638
8861
8807
8633
8394
8368
8484
8690
boiler s
1464
420
241
255
898
1231
664
780
197
87
52
56
154
178
83
98
0.44
0.21
0.14
0.14
0.43
0.39
0.21
0.25
90.1
57.3
53.6
60.5
67.2
83.4
85.9
105.7
.027
.029
.031
.035
.032
.028
.028
.034
10
14
15
11
15
11
10
16
(3)
(2)
(2)
(2)
(2)
(2)
(2)
(2)
880
738
695
646
662
862
934
784
0.26
0.38
0.40
0.37
0.31
0.28
0.30
0.25
-------
- 220 -
Figure 3 is a schematic of the power boiler and scrubber systems showing
the inlet and outlet sampling locations. The tests were conducted on one of the
three identical scrubber modules which are operated in parallel to control S02
and particulate emissions from the power boiler. The three modules are inde-
pendently controlled with respect to liquor flows and venturi pressure drop.
Pressure drops across the Venturis are regulated by adjusting the position of
the "plumb bob" shown in Figure 3, thereby increasing or decreasing the cross
sectional area of the venturi throat. Throughout these tests, with the excep-
tion of one brief period, the pressure drop across the venturi on the module
being tested was held at 46 ± 2 cm w.c.. Gas temperatures at the scrubber inlet
ranged from 129°C to 137°C. The scrubber exit gas temperatures ranged from
57°C to 60°C and temperatures at the outlet test plane ranged from 94°C to 99°C.
The temperature rise between the scrubber exit and the outlet mass sampling lo-
cation results from a flue gas reheat system and the action of the fan, both of
which are located between the scrubber outlet and the sampling plane. The gas
flow handled by the scrubber throughout the tests was approximately 130 DNCM/sec
(280,000 DSCFM). A complete summary of the scrubber operating conditions during
the tests conducted by SoRI are given in Table 5.
TEST METHODS AND RESULTS
A total of five measurement techniques were used during the tests. These
were: (1) electrical mobility techniques using a Thermosystems Model 3030 Elec-
trical Aerosol Analyzer for determining concentrations and size distributions on a
number basis for particles having sizes between 0.01 urn and 0.3 um, (2) optical
techniques to determine concentrations and size distributions for particles having
diameters between approximately 0.5 um and 2.0 um, (3) inertial techniques using
cascade impactors for determining concentrations and size distributions on a mass
basis for particles giving sizes between approximately 0.5 um and 5.0 um,
(4) standard mass train (Method 17) measurements for determining total inlet and
outlet mass loadings and emission rates, and (5) determinations of S02 concen-
trations by absorption of the SO2 vapor in a solution of hydrogen peroxide fol-
lowed by titration for the sulfuric acid reaction product.
The data obtained by Method 17 are summarized in Tables 6 and 7. The over-
all collection efficiencies for each of the pairs of tests are given in Table 8.
The overall collection efficiency of the scrubber on this source under the
conditions of operation tested is thus found to be approximately 99.4 percent.
Inertial sizing was accomplished using modified Brink impactors for inlet
measurements and University of Washington Mark III impactors for outlet measure-
ments. Sampling was done in both cases at near isokinetic flow rates, thus
errors due to deviations from isokinetic sampling should be of little consequence.
All impactors used in this program were calibrated at SoRI using the methods de-
scribed in EPA publications 600/2-76-280 and 600/2-77-004.
-------
BUTTERFLY ISOLATION DAMPER
-PLUMB BOB DRIVE
STACK
FLUE GAS
FROM AIR
PREHEATER
AND BOILER
FROM PLANT FIRE WATER SYSTEM
EMERGENCY COOLING SPRAY
PLUMB BOB
VENTURI SECTION—^
CLEANING SPRAY
MIST ELIMINATORS
CLEAN FLUE GAS
FROM SEAL WATER SUPPLY
MIST ELIMINATOR UNDERSPRAY
I
WASH TRAY
ABSORPTION SPRAY
GUILLOTINE
SHUT-OFF
TRAY UNDERSP
ALKALI
SYSTEM
Z I- INDUCED
8 O DRAFT FAN
RECYCLE TANK
POND RETURN
RECYCLE PUMPS
SCRUBBER VESSEL SEAL POT K WASH TRAY
Q RECYCLE TANK
SS
CV
WASH TRAY
RECYCLE PUMP
AND SPARE
WASH TRAY
POND
OUTLET
TEST PLANE
WASH TRAY
POND RETURN PUMP
AND SPARE
N)
fO
NOTE: VALVES SHOWN ARE MAJOR CONTROL, BLOCK VALVES IN SYSTEM -
EFFLUENT PUMP V\=^-r-_- A^=. — -
AND SPARE
FLYASH POND
ASH POND PUMP
AND SPARE
Figure 3. Simplified scrubber flow diagram.
-------
Table 5
Scrubber Operating Conditions
Measured Gas Flow,
DNCM/s
Temperatures, C
Liquor Flowa, Upm
Date
Unit Load, Inlet Outlet Plumb Bob Venturi Scrubber Scrubber Reheat Pan Upper
MW Position, P, Inlet Outlet Outlet Outlet Spray
Middle
Spray
Absorption
Spray
% of Travel cm w.o.
5/17
5/18
5/18
5/19
5/19
5/20
5/20
5/17
5/18
5/18
5/19
5/19
5/20
5/20
330
350
355
355
355
290
348
Liquor
pH
4.3
4.7
4.7
4.7
4.6
N.A.
N.A.
110
132
133
130
124
(106):;
(127) i
% Suspended
Solids
11.4
15.2
16.4
14.3
13.4
N.A.
N.A.
128 54
140 58
162 62
137 61
139 61
(113) | 53
(136) 65
44.5 132 60 79 94 15900
46.4 129 58 78 96 15000
46.4 129 57 78 96 18200
46.4 131 59 78 96 17600
47.0 133 57 74 96 17500
45.7 129 56 82 93 17800
45.1 129 52 82 93 17600
10200
9370
11360
10790
10600
10600
11700
22700
24400
19870
24600
24200
25700
25000
Mist Film
Under
Spray
570
570
570
625
625
530
570
Wash Tray Wash Tray
Under Feed
Spray
1170 3600
1060 2900
1170 2800
1170 3220
1190 2840
1170 3220
950 3220
Based on partial traverse and scaling from previous days.
-------
- 223
Table 6
CEA VARIABLE THROAT VENTURI TEST
INLET MASS DATA
Run
Number
Date
Time
Moisture, %
Gas Temperature,
°C
UF
1
5-16-77
1715
10.30
134
274
2
5-17-77
1455
11.62
132
269
3
5-18-77
1235
10.25
129
265
4
5-18-77
1545
10.87
129
264
5
5-19-77
0825
11.86
137
278
6
5-19-77
1245
12.26
133
272
Volumetric Flow,
m /sec
ACFM
Volumetric Flow,
DNCM/s
DSCFM
Concentration,
grams/ACM
Concentration,
grams/DNCM
Isokinetic, %
208.6 201.3 233.9 236.9 238.8 227.6
442,000 426,500 495,500 502,000 506,000 482,300
116.5 110.3 131.9 132.9 130.1 124.4
247,700 233,600 279,500 281,500 275,600 263,500
2.0184 2.8325 3.3097 3.4820 3.5145 3.5829
3.6145 5.1701 5.8663 6.2079 6.4512 6.5546
107.62 105.85 104.23 108.62 103.79 103.56
-------
- 224 -
Table 7
CEA VARIABLE THROAT VENTURI
OUTLET MASS TRAIN DATA
Run
Number
Date
Time
Moisture, %
Gas Temperature,
°C
°F
1
5-16-77
1700
14.01
99.4
211
2
5-17-77
1315
19.45
94.4
202
3 4
5-18-77 5-18-77
1500
17.37 16.53
96.1 96.1
205 205
5
5-19-77
0830
18.70
96.1
205
6
5-19-77
1300
18.15
96.1
205
Volumetric Flow,
m3/s
ACFM
Volumetric Flow,
DNM3/s
DSCFM
Concentration,
mg/ACM
Concentration,
mg/DNCM
Isokinetic, %
194.9 224.8 238.9 273.8 237.8 239.3
413,000 476,200 506,200 579,300 503,800 507,000
118.4 128.4 140.0 162.1 137.3 139.3
250,800 171,100 196,700 343,500 290,900 295,200
26.09 25.86 19.66 21.52 24.23
19.23
42.79 45.31 33.58 36.28 41.90 33.04
105.71 113.99 106.40 106.76 103.91 104.66
-------
- 225 -
Table 8
CEA VARIABLE THROAT VENTURI SCRUBBER EFFICIENCIES
FROM MASS TRAIN DATA
Run No. Date Efficiency (%)
1 5-16-77 98.82
2 5-17-77 99.12
3 5-18-77 99.43
4 5-18-77 99.42
5 5-19-77 99.35
6 5-19-77 99.50
-------
- 226 -
The impactor data are summarized in Figures 4 through 8. Figures 4 and 5
present averaged inlet and outlet size distributions, respectively, on a cumula-
tive percentage (by mass) basis versus aerodynamic particle diameter. Figures
6 and 7 show the same data on a cumulative mass concentration basis. Figure 8
shows the fractional efficiency curve as a function of aerodynamic particle dia-
meter as derived from the inlet and outlet data that were presented in the pre-
vious figures. The fractional efficiency curve is shown later in Figure 11 as a
function of Stokes diameter together with the efficiency curves derived from
the ultrafine particulate data. The scrubber was operating at a venturi pres-
sure drop of about 48 cm w.c. throughout the impactor test periods.
Measurements of the concentration and size distribution of ultrafine particu-
lates were made using a Thermosystems Model 3030 Electrical Aerosol Analyzer (EAA)
and a Royco Model 241 Optical Single Particle Counter.
The EAA provides size distribution and concentration data on a number basis
for particles having diameters between approximately 0.01 urn and 0.3 urn. The
optical counter provides similar data in the range from approximately 0.3 to 2 pm.
Both instruments require extensive sample dilution and conditioning when used to
sample flue gases. The sample extraction and dilution system used in these tests
is described in a forthcoming EPA report on Contract 68-02-2114, Task VIII.
Dilution factors of about 150:1 were used at both the inlet and outlet during
these tests.
In order to insure that condensation effects were minimal, and that the
particles were dry as measured, the diluent air was dried and filtered, and dif-
fusional dryers were utilized in the lines carrying the diluted samples to the
instruments.
Because only one set of instruments and dilution system was available it
was not possible to obtain simultaneous inlet and outlet data for the ultrafine
particulates. The system was first installed at the scrubber inlet and all inlet
data were obtained on May 17. The equipment was then moved to the outlet and
outlet data were obtained on May 19 and 20. For the purposes of calculating
fractional efficiencies the assumption was made that the process was sufficiently
stable that the inlet data, as obtained above, were a valid representation of
that which would have obtained during the time the outlet measurements were made.
Inlet data were obtained with the optical counter in two size channels—
0.35 to 0.60 urn and 0.60 to 2.0 ym. However, an instrument malfunction resulted
in outlet data being obtained only in the 0.6 to 2.0 ym size interval with this
me thod.
Inlet size distributions on a cumulative concentration by number basis are
shown in Figure 9. Outlet size distributions on a similar basis are shown in
Figure 10 for the normal scrubber operating condition (48 cm w.c. venturi pres-
sure drop). Figure 11 shows the fractional efficiencies for ultrafine particles.
Also shown in Figure 11 are the fractional efficiencies as a function of Stokes
diameter, obtained from the impactor data.
-------
- 227 -
h-
u
Ul
Q.
LJ
H
H
3
99.99-
ggie-
99«5 -
99-
98-
95 i
90 \
BO\
60^
50^
301
30 1
10 i
5:
si
1]
o.oi:
1C
L
!
: *
- *
'. •
r
• *
: ******
r1 10° lo1 lo5
AERODYNAMIC DIAMETER (MICROMETERS)
Figure 4, A verage inlet particle size distribution from cascade impactor
data on a cumulative percent by mass basis.
-------
- 228 -
h-
u
Q.
LJ
M
H
H
a
99.99;
99.5^
991
95:
90^
70^
6CH
50i
4Oi
30^
EOi
10^
5-
1-
O.5-
0.2-
0.1-
O.Ob-
o r\* .
'• HI
; J
r X
; X
j- •
~ «
: *
*
" »
^*
r
r
i — i i i 1 1 n4 1 — i i i i ml 1 — i I I I llH
10
,-1
10°
101
102
AERODYNAMIC DIAMETER (MICROMETERS)
Figure 5. Average outlet particle size distribution from cascade impactor
data on a cumulative percent by mass basis.
-------
103,:
a
a
01
01
LJ
10'
- 229 -
a:
a
<
a
-10'
- -i n—4
H—i i i inn 1—i M i nil 1—i i i HII|
10'
10°
101
AERODYNAMIC DIAMETER (MICROMETERS)
Figure 6. Average inlet particle size distribution on a cumulative mass
concentration basis from cascade impactor data.
-------
- 230 -
1C3,:
M
a
a
01
01
M
i—
_j
10°,:
-la
<
*
a
$
:rlCT
-"-10"
1—i i i mil 1—i i i mil 1—i i 11 nil
1CT1 1O° 1O1 102
AERODYNAMIC DIAMETER (MICROMETERS)
Figure 7. Average outlet particle size distribution on a cumulative mass
concentration basis from cascade impactor data.
-------
H
<
- 231 -
FETCTRATIDN-EFFTCIENDr
10?T
IDS:
T O-O
-9O.O
10
rl
H—I I I IIN| 1—I I I HIM 1—I I I Mil
10P
101
Ul
M
U
M
U.
tf
"93-9
.99
AERODYNAMIC DIAMETER (MICROMETERS)
Figure 8. Fractional efficiency curve on an aerodynamic particle diameter basis
for the CEA variable throat venturi scrubber operating at a venturi
pressure drop of 48 cm (19 in.) w.c..
-------
- 232 -
101
10
13
en
E
o
z
O
Z
o
o
EC
UJ
CO
10
12
D
O
10
,10
• EAA
D ROYCO
10T'
10-1
LOWER SIZE LIMIT, micrometers
10U
Figure 9. Scrubber inlet panicle size distribution from electrical
aerosol analyser and Royco optical particle counter data.
-------
233 -
1014
CO
z
Q
o
o
O
o
m
D
Z
LU
D
O
188
10
12
10"
•
o
10°
LOWER SIZE LIMIT, micrometers
Figure 10. Scrubber outlet particle size distribution from electrical
aerosol analyser data.
-------
- 234 -
PENETRATION-!
Iff^
101.
Z :
a
M
i —
1
f\/
f—
LJ
LJ -1 cP.
Q_ -'-^ •
^ :
LJ
u
LJ
Cl-
lO-1:
icrs-
.^ j
-A A A 1
: A A :
.
' * '.
• 1
•
«
: o '•
o
" A EAA
- O IMP ACTORS l /-vi >l /
r U.U
:90.0
: &
2
Ld
M
U
H
U.
- 99 . 0 ^~
:
z
Ld
U
QL
r Ld
Q.
^99.9
-99.99
-^
1O'
PARTICLE DIAMETER (MICROMETERS)
s 11. Fractional efficiencies based on electrical mobility and optical methods
shown on a "physical" diameter basis. Also shown are fractional
efficiencies from the cascade impactor data on a basis of Stokes diameters.
-------
- 235 -
The scrubber was operated at venturi pressure drops of 31, 36, 41, 46, and
51 cm w.c. for a brief period at each condition on May 21, during which time the
outlet concentrations were monitored with the EAA and the optical counter. No
significant concentration changes were noted in the E&A data over this range of
pressure drops, however, the optical counter data did show significant changes.
In the 2 ym to 4 ym size interval, a 50% reduction in concentration was obtained
by increasing the venturi pressure drop from 31 to 51 cm w.c. and a 35% reduc-
tion in concentration occurred in the 0.6 pro to 2.0 ym particle diameter range.
These relative concentration changes are shown in Figure 12.
The results of the SO2 concentration measurements at the scrubber inlet and
outlet are given in Table 9. Table 9 also includes SO2 collection efficiencies
derived from the concentration measurements.
SUMMARY
The overall collection efficiency of the CEA variable throat venturi scrub-
ber, determined by conventional (Method 17) techniques on a pulverized coal fired
power boiler producing particulate having a mass median diameter of about 20 ym,
ranged from 99.12 to 99.50 during three days of testing. The venturi pressure
drop ranged from 44.5 cm w.c. to 48.3 cm w.c.. Measured fractional efficiencies
were about 5% at 0.06 jam, 25% at 0.1 ym, 40% at 0.20 ym, 50% at 0.5 ym, 98.4%
at 1.0 ym, and 99.99% at 2 ym. The system energy usage during the tests was
approximately 7200 joules/DNCM. SOz collection efficiency ranged from 76.5% to
85.6%.
ACKNOWLEDGEMENTS
Appreciation is expressed to B. Knutson, D. Berube, and C. Grimm of The
Montana Power Company for their cooperation during the test program. Apprecia-
tion is also expressed to I. A. Raben of Combustion Equipment Associates, Inc.
for supplying the information on the design and operation of the scrubber. The
test program was conducted under Contract 68-02-2181 for the Industrial Environ-
mental Research Laboratory of the U.S. Environmental Protection Agency. Mr. Dale
L. Harmon was the project officer responsible for the technical effort under this
contract.
-------
- 236 -
14
8
a:
< V
£ 'X
z 10
x
£ V
<
«i— ^
• CH 1 x 100 0.55 - 1.8 Aim \
ACH2 x 10 1.8- 4.1 /im \
25 30 35 40 45 50
VENTURI PRESSURE DROP, C.M. W.C.
Figure 12. Relative outlet paniculate concentrations in two size
ranges as functions of venturi pressure drop.
-------
- 237 -
Table 9
COLSTRIP POWER PLANT
SCRUBBER SO2 REMOVAL EFFICIENCY
Date Inlet SOz Reheater Outlet SO2 Removal
Concentration SO2 Concentration Efficiency
(ppm) (ppm) {%)
5-17-77 658 130 80.2
5-18-77 525 103 80.4
5-19-77 553 130 76.5
5-20-77 625 90 85.6
-------
- 238 -
Discussion
Dr. Holighaus opened the discussion with a question regarding the difference
between using slurry or clean liquid as a cleaning mechanism with view
to particle separation. Dr. Calvert maintained that there was little
difference; more fundamental was the mass ratio of scrubbing liquid gas.
He wondered if the use of a sieve plate would exclude effectively the
large particles; a pressure drop of approximately 3-9 ins. would be able
to cause removal. Prof. Weber asked what the availability of this kind
of plant was, and was told that there was 90% availability.
Dr. Holighaus inquired whether it would separate fly ash and calcium
sulphate from the liquid and what happened to water from the ash pond.
According to Mr. McCain, separation was unnecessary and water was recirculated,
No calcium sulphate was present in the emssions. Mr. Princiotta added
that low sulphur coal was suitable for this process.
-------
239 -
SIMULTANEOUS SEPARATION OF DUST AND GASEOUS CONSTITUENTS AT
HIGH GAS TEMPERATURES BY THE USE OF MOLTEN METALS AND SALTS,
Prof. Dr--Ing. E. WEBER
Institut fur Mechanische Verfahrenstechnik
Universitat Essen GHS
UniversitatsstraBe 2
Postfach 68 43
D 43OO Essen
Tel. 0201-1832795
Dr.rer.nat. K. HUBNER
Institut fur Mechanische Verfahrenstechnik
Universitat Essen GHS
UniversitatsstraBe 2
Postfach 68 43
D 430O Essen
Tel. 0201-1832786
-------
- 240 -
SIMULTANEOUS SEPARATION OF DUST AND GASEOUS CONSTITUENTS AT
HIGH GAS TEMPERATURES BY THE USE OF MOLTEN METALS AND SALTS.
K. Hiibner, E. Weber
Today the cost and the efficiency of power plants strongly
depend on the gas cleaning system. This can be best illustrated
for the example of the carbon gasification process. The usual
way of handling the generated gas is to cool it down by direct
contact to water which is sprayed by nozzles into a spray tower.
After that, any conventional gas cleaning system can be employed.
This procedure, however, results in an irreversible loss of
thermal energy and an unwanted decrease of the power plant's
overall energetic efficiency. This difficulty can be avoided
by use of a heat exchange system. The gas coming from the gas
generator usually contents large amounts of dust and sulfur
and in many cases tarry materials causing sticking effects.
The properties of the solids to be separated and the wanted high
efficiency of the heat exchanger raise great difficulties. It may
be necessary to use at least two parallel heat exchange systems.
Using a wet scrubber, a sophisticated regeneration system for
the washing liquid is needed. The overall cost of invest and the
cost of operation are extremely increased. It seems to be at
least doubtful whether the energy spared by the heat exchange
system can justify these cost.
The application of fibre filter separators or electrostatic
precipitators instead of a Venturi scrubber will raise the cost
too, because of the unavoidable additional cleaning system for
the gaseous constituents.
It can be summarized that until today, there are no gas cleaning
systems allowing the simultaneous separation of gaseous and solid
pollutants at elevated temperatures of more than 400 °C without
loss of the heat content.
-------
- 241 -
A possible solution for this problem is offered by a high
temperature scrubbing process using a suitable washing liquid.
The washing medium should be sprayable at elevated temperatures
just like water or aqueous solutions in the conventional scrubbing
processes. The application of the washing liquid should not lead
to additional problems of emission.
For such a gas cleaning system two types of washing media have been
developed, the composition of which can slightly be modified according
to the given conditions of application. The first type of washing
liquid is a melt of inorganic metal compounds having a density of
about 1.4 kg/1 and a surface tension of about O.1 N/m. These
properties are similar to that of water at room temperature.
The other type of washing medium is a mixture of molten metals
having a density of about 4.5 to 7 kg/1 and a surface tension of
about O.5 N/m.
From the temperature dependence of the vapour pressures ( fig. 1 )
it can be seen that the metal compound melt should not be used at
temperatures of more than 11OO K or 8OO °C. At this temperature
the vapour pressure of this melt is in the 100-Pascals-region,
whereas the metal melt has a vapour pressure as low as 1 to 1O Pascals
even at 13OO K or 1OOO °C. The viscosities of both melts are
sufficiently small ( fig. 2 ) in the total given temperature range
that means both melts are well sprayable. Whereas the metal melt
is only applicable for reductive gases the metal compound melt can
be used independently of the oxygen content of the gas. The metal
melt allows the separation of a single gaseous pollutant :
hydrogen sulfide, which can be removed from the melt by hydrogen
rich gas at elevated temperatures to regenerate the melt. The metal
melt does not react chemically with solid particles.
On the other hand, however, the metal compound melt allows not only
the separation of many gaseous constituents with acidic charcter
but even undergoes chemical reaction with solid particles.
Due to its great reactivity the metal compound melt causes material
and regeneration problems. The material problems can be solved by the
use of nickel alloys. The regeneration of the melt is not necessary
-------
- 242 -
in most cases, since it can operate as a washing medium nearly
to the point of chemical saturation with pollutants. The reaction
products can be led to disposal after suitable treatment.
Figure 3 shows a scheme of a gas cleaning system using melts as a
washing media. After the precipitation of the larger dust particles
by a cyclone the gas is scrubbed in a Venturi using the metal-
compound melt and then passed to a droplet separator. The melt can
be recycled without cleaning to the Venturi for several times
because most of the dust particles react with the melt forming a
solution which is sprayable without blocking the Venturi's nozzles.
To maintain the melt's reactivity it is necessary that the overall
reaction of the melt is basic.
On the other hand it is necessary to neutralize the reacted melt
with acidic oxides like SiO--, or B-O-. in order to obtain waste
products which can be led to disposal.
Due to the metal-compound melt's low cost and due to its high
efficiency this seems the less problematic way, but in principle
the melt can be reactivated in the case of the removal of gaseous
constituents like SO,-, or H~S.
In the case of a metal melt, it is necessary to separate the dust
particles from the bulk liquid because of the high cost for the
metal melt. This can easily be done by the differences in densities
between dust and melt. The melt's density is at about 3 times higher
than that of the solid particles ( e. g. fly ash from a power plant )
From the basic laboratory experiments the use of metal and metal
compound melts seems practicable for the simultanious removal of
gaseous constituents and solid particles from gases.
Measurements at relatively low gas velocities of less than 10 m/s
with a discontinously operating laboratory plant at temperatures
of 40O C have shown that both melts allow the precipitation of
dust with an efficiency of more than 95 percent. For the laboratory
experiments a metal melt was used which consists of pure tin.
-------
- 243 -
Due to the physical properties of the metal melt it needs high
liquid-pressures for spraying when a commercial full cone nozzle
is applicated whereas the metal compound melt which consists
mainly of alkaline materials can be sprayed like water.
The high liquid-pressures can be avoided by the use of pneumatic
atomizers. Figure 4 shows the drop size distribution for a full
cone nozzle obtained at aliquid pressure of 12 bars and that for
a pneumatic atomizer at 2,5 bars. The drop size distributions have
been calculated from sieve analysis of solidified melt droplets
which had been chilled by spraying into atmospherical air.
In both cases mean particle sizes of about 300 tarn are achieved.
As already mentioned, alkaline melts can be used for the removal
of a large scale of gaseous constituents, whereas a tin melt
reacts only with hydrogensulphide.
The table gives a comparison of typical reactions between gases
and melts at a temperature of 800 K. It can be seen that the
reaction of sodium hydroxide, a possible component of an alkaline
melt takes place with a relatively large change of the Gibbs-energy,
that means the thermodynamic equilibrium is dominated bythe removal
of the gaseous constituents. In the case of a tin melt the change
of the Gibbs-energy is somewhat smaller, but it must be taken into
consideration that gases from coal gasitication processes contain
simultaneously hydrogen sultide and hydrogen.
The equilibrium in such gases depends on the ratio of hydrogen
partial pressure and the hydrogen sulfide pressure.
Figure 5 shows the Gibbs-energy for the removal of hydrogen sulfide
by a tin melt depending on the temperature. It can be seen that
for small pu to pTT „ ratios and comparatively low temperatures
H~ H ~S
the removal of H~S is favoured, whereas for a hydrogen rich
gas and high temperatures the decomposition of the tin sulfide is
dominating.
From the viewpoint of the thermodynamic equilibrium both the removal
°f H2S and the generation of the tin melt seem possible. Laboratory
-------
- 244 -
experiments have confirmed these findings showing no severe kinetic
hindering at temperatures of more than 40O C.
3
At present a continuous pilot plant for 4OO m gas/h at temperatures
of more than 400 °C has been constructed and will operate in a few
weeks to complete the experiments of the small discontinuous
laboratory plant. The measurements shall be expanded especially
into the region of higher gas velocities, in order to optimize the
removal of fine particles. Summarizing it can be stated that the
application of molten metals or molten salts seems to offer a low
cost solution for gas cleaning problems in power plants and in any
case were clean gases are needed at high temperatures and pressures.
-------
- 245 -
Table; Standard Gibbs-Energy at 800 K.
Some typical reactions of gases with metal and
alkaline compound melts.
2 NaOH(1) + HS ^ - - NaS + 2 HO - 76,96
NaOH(1) + HCl(g)-^ - * NaCl(s) + HO -133,04
Sn(1) + HS(g)x - * SnS(3) + H - 48,65
-------
70* ^
pi Pa 102 .
100 -
10-2-
70-*-
70-6-
70-*-
metal compound
1300 1200 1100 1000 900
800
700
600 TIK
Mechanische
Verfahrenstechmk
Univers/tat Essen
Vapour pressures of melts
f/g.
1
-------
mPa s
10° -
metal compound
metal
1300 1200 1100 1000 900 850 800 750 700 650
T/K
Mechanische
Verfahrenstechn'ik
Universitdt Essen
Viscosity of melts
fig-
2
-------
cyclone
waste gas
e
1
!
/
Venturi scrubber
droplet separator
\
j
cl
clean gas
separation
melt/dust
Mechanische
Verfahrenstechnik
Universitat Essen
Scheme of the gas cleaning system
fig.
3
-------
0.063 0125 0.25
full-cone nozzle
liquid-pressure. 12 bar
pneumatic atomizer nozzle
liquid - pressure. 2.5 bar
0.5 particle size /mm
Mechan/sche
Verfahrenstechmk
Universitat Essen
Drop Size distribution of molten tin
-------
Gibbs-
energy
-20 H
kJ
mol
20-\
Standard reaction
HyS
10000
line of equilibrium
600
700
800
900
WOO K
temperature
Ul
o
Mechanische
Verfahrenstechnik
Universitdt Essen
H
fig.
5
-------
- 251 -
Discussion
Mr. Shackelton asked whether particle collection had been measured
and was informed that measurements had been made with flyash from
the power plant. A pneumatic atomizer nozzle was used, and the dust
collection efficiency was more than 95%. Dr. Holighaus wished to
know if there was a sulphur outlet, upon which Dr. Hiibner replied
that hydrogen rich gas was used at elevated temperatures. This was
emitted as a hydrogen sulphate rich gas, which was passed through
a cleaning process. The conclusions reached were that the tempe-
rature was sufficiently high, and that cleaning could occur at any
temperature and in all systems with low energy loss. In reply to
Mr. Calvert, Dr. Hiibner described the shape of the Venturi Scrubber,
which is simply a spray scrubber, O,l m in diameter, with an alkali
melt full cone. The melt was composed of potash and sodium hydro-
xide; however, adding a material like limestone or calcium oxide
is more economical. (The other type of scrubber described had a
pneumatic-atomizer, in order to avoid high pressure). The volumetric
flow ratio of liquid: gas was only 1:25 litres m .
-------
- 252 -
OPTIMIZATION OF WET AND DRY PROCESSES FOR SIMULTANEOUS REMOVAL
OF PARTICULATES AND GASEOUS AIR POLLUTANTS FROM COAL FIRED
POWER STATIONS
by
Peter Davids
Federal Environmental Agency
Berlin
prepared for
A Particulate Workshop
Julich, Federal Republic of Germany
March 16-17, 1978
1. Emission tendencies
In the Federal Republic of Germany the typical air pollutant emis-
sions from power stations show different tendencies. Fig. 1 pre-
sents the trends of NO , CH, SO,, and particulates during the de-
X £.
cade from 1965 to 1975 and an estimation for 19BO.
The 50,.,-emission could be stabilized to about k Mio t/a in spite
of increased energy consumption by fuel oil desulfurization, sub-
stitution of coal by low sulfur fuels and - at a still very low
degree - by flue gas desulfurization. An increase of the contribu-
tion of power stations to the total 30,-,-emission from about 30 %
in 1965 to about 5D % in 1980 is expected. The total NO -emission
X
is increasing from 1.3 to 2.3 Mio t/a permanently; the share of
power stations from about 20 % to about one third.
The greatest success in emission control was attained in the
field of particulate removal. The total emission could be reduced
from about 2 Mio t/a to about 0.5 Mio t/a by application of improved
collection techniques. Remarkable is the increasing share of
power station1- from one fourth to one third of the total emission
and the appro) nately unchanged ission of fine particulates.
-------
- 253 -
At last it can be gathered from Fig. 1 that power stations do
not contribute to hydrocarbon emissions significantly.
Fig. 2 presents a map of the Federal Republic of Germany. The
territory was divided into squares according to the geographic
coordinates. For each square the regional emissions were calcula-
ted for 1965, 1970 to an 1975. The results are presented in
Fig. 3 to 6.
The S0_-emission occurs in the industrialized areas mainly, in
particular where power stations are concentrated. Fig. 3 shows
the maximum emission in the area around Julich. In this area we
find a concentration of lignite fired power stations with a total
capacity of about 15.ODD MU. The regional distribution of parti-
culate emissions is similar to the distribution of S0?-emissions
(Fig. <+), according to the distribution of large power stations.
Compared with S09 and particulates the NO -emission is distribu-
£- J\
ted more uniformly because of the higher share of emissions from
motor vehicles (Fig. 5). The most uniform distribution occurs
for hydrocarbon emissions because the main sources are motor ve-
hicles and domestic firings (Fig. 6). As above - mentioned power
stations do not contribute significantly.
2. Air quality trends
Fig. 7 presents the annual average ground level - concentration
of SD? at certain places in the Federal Republic of Germany.
The recommended air quality standard of the LJorld Health Organi-
zation ( 60 ,ug SO /m ) is exceeded in the industrialized areas.
In the main industrialized region, the Rhine-Ruhr-Area, north-
east from Julich, the ambient S09-concentration even exceeds the
t- -$
national air quality standard of 1^0 /ug 30,-,/m .
The trends of the SO^-ground level concentrations from 1970 to
1975 in certain cities are presented in Fig. 8 and 9. The development
is not uniform. The concentration is. decreasing (e.g. Berlin,
Hamburg) or approximately unchanged (e.g. Bochum, Cologne)
-------
- 254 -
or increasing (e.g. Dusseldorf, Saarbrucken).
Fig. 10 presents the development of participate sedimentation
since 1970 in several cities. At most places the values are de-
creasing. The concentration of suspended particulates is decrea-
sing in most of the cities in the Federal Republic of Germany,
too, but at a minor degree than particulate sedimentation (Fig.
11 and 12). In particular the so called "clean air monitoring
stations" of the Federal Environmental Agency, far away from in-
dustrialized areas, show an unchanged or increasing concentration
of suspended particulates due to long range transport of air
pollutants.
Altogether, it's of particular importance that the reduction of
particulate emissions did not lead to an equivalent reduction of
the ambient particulate concentration. The present ground level
concentrations require further control measures. In view of the
development of emissions and ground level concentrations, the
Federal Government has strengthened its efforts to reduce emissions
from power stations by use of best available control technologies.
3. Present emission standards
The First General Administrative Regulation under the Federal Air
Pollution and Noise Control Law from 197*+ contains emission stan-
dards for the main air pollutants from power stations:
- the emission of particulates is restricted to 150 mg/m for
bituminous coal fired power stations and to 100 mg/m for lig-
nite fired power stations; if using an electrostatic precipi-
tator, the standards also mark the maximum allowable emission
even if one electric field of each parallel section is out of
service.
- for power stations with a capacity up to ^20 MLJ the sulfur con-
tent of coal is limited to 1 %; for larger power stations flue
gas desulfurization is required. In 197<+ the new source perfor-
mance standard was set to 3.75 kg SO-XMliJh ; it was reduced to
2.75 kg S02/MUh in August 1977. For comparison: The Federal
US-standard is about 5 kg SO
-------
- 255 -
as to NO^-control the Administrative Regulation only gives a
general recommendation, to reduce emissions as far as possible,
for example by tuo stage combustion or flue gas recirculation.
<+. Improvement of control technologies
In vieu of the emission and air quality - situation in the Fede-
ral Republic of Germany in the last years several activities have
been introduced to improve existing and to develop new control
technologies.
In the field of coal demineralization processes for bituminous
coal have been optimized. The available techniques allow a limi-
tation of the ash content to 5 to 10 % for most of German bitumi-
nous coal. Thereby the sulfur content can be reduced to about
1 %; heavy metals, chlorine and fluorine can be limited to loiu
levels, too. In the field of NO -control in a first step emission
reduction by introduction of combustion modification is intended,
according to the strategy in the US and in Japan. Optimization
of flue gas cleaning processes should offer the largest potential
for improvement of emission control, in particular the optimiza-
tion of flue gas desulfurization processes for simultaneous re-
moval of particulates and gaseous pollutants.
Table 1 presents the range of emission factors for usual fuels
in the Federal Republic of Germany. Burning of coal leads to the
highest SO - and NO -emissions and in addition to considerable
C- X
HC1- and HF-emissions.
Table 1: Emission factors (mg/m ) for pouer stations in the
Federal Republic of Germany
so?
NO
X
HCI
HF
bituminous
coal
1,500-5,000
500-2,000
50- 200
5- 40
lignite
500-3,000
200-1,000
20- 100
0.5- 2
oil
1,000-5,000
200-2,000
-
-
gas
100
100-200
-
-
-------
- 256 -
Actually the simultaneous removal of SD2, NDx, HC1, HF and fine
particulates is being studied with the three desulfurization
systems which so far have been developed in the Federal Republic
of Germany. These systems are
- the Bischoff ( lime/sludge ) process
- the Saarberg-Holter ( lime/gypsum ) process
- the Bergbauforschung ( carbon adsorption ) process.
it. 1 Bischoff-pro cess
The largest flue gas desulfurization installation in the Federal
Republic of Germany is operating at a bituminous coal fired
7DD My power station at Idilhelmshaven. The Bischoff-desulfurizar-
tion unit has a capacity of about 17D MLJ ( Fig. 13 ). The scrub-
ber is a combined spray tower and venturi with adjustable throat;
the scrubbant liquid is lime slurry; after thickening the process
effluent is pumped into a pond.
The inlet particulate concentration to the scrubber is some
50 mg/m , because the boiler flue gas is dedusted first in an
electrostatic precipitator. A clean gas particulate concentration
below 2D mg/m is expected; the current research program in parti-
cular should yield data on the outlet particulate concentration
as a function of energy consumption (i.e. pressure drop) to optimize
scrubber operation. Variation of pressure drop does not influence
SD^ removal significantly; the efficiency is greater than 90 %
independent on pressure drop.
4.2 Saarberg-Haiter process
The particular characteristic of the Saarberg Halter (lime/gypsum)
process is the use of hydrochloric acid and an organic compound
as additives; that renders possible a scrubber operation with a
clear solution (Fig. 14). A demonstration unit with a capacity
of about 4Q MU is operating since 1974.
-------
- 257 -
The flue gas is treated in a venturi scrubber. The calcium sul-
fite in the scrubber effluent is oxidized to sulfate by injec-
tion of air into the oxidizer. After thickening and deuatering
the byproduct gypsum is sold to the gypsum industry. The opera-
ting results have revealed high availability of the system
during the last years; the S0? removal efficiency is greater than
9D %.
In the current research program removal of fine particulates,
HC1, HP and NO is being studied. The particulate inlet concen-
3
tration to the scrubber is about 150 mg/m because the scrubber
is connected in series with an electrostatic precipitator; the
dust outlet concentration of the sen
a pressure drop of about 2DDD Pascal,
dust outlet concentration of the scrubber is about 20 mg/m with
The removal efficiency for NO is poor because more than 90 %
X
of the NO occurs as NO. Improvement of NO -removal by use of an
/\ /\
oxidizing agent in the scrubber is intended. The removal effi-
ciency for HC1 and HF is very high, still higher than for SO,.,;
3 *-
scrubber outlet concentrations of less than 20 mg HCl/m and
5 mg HF/m are achieved.
Objective of the current research program is the further optimi-
zation of simultaneous removal of SO™, fine particulates and the
other gaseous pollutants by systematic variation of the opera-
ting parameters, in particular pressure drop, liquid/gas-ratio
and composition of the scrubber solution.
^.3 Bergbauforschung - process
The Bergbauforschung-process is a dry process using activated
carbon for S0? adsorption (Fig. 15). The flue gas is treated in
a moving granular bed of carbon grains; the grain size is about
10 mm; the carbon is regenerated at a temperature of about 900 K
by mixing with hot sand. The SO^off gas with a concentration of
15 to 20 % SOp is processed to elementary sulfur in a
Claus furnace.
-------
- 258 -
A demonstration unit with a capacity of 5D Mid is operating since
19?it. Successful results have been attained uith the adsorption
and desorption section; problems have occured uith the Claus fur-
nace because the desulfurization unit uas retrofitted to a peak
load power station and the Claus furnace is not very suitable
if load changes occur.
Uith the carbon bed adsorber high removal efficiencies for fine
particulates have been achieved, too. The inlet particulate con-
centration, which is equivalent to the outlet particulate con-
centration of the electrostatic precipitator, is reduced from
3 3
about 15D mg/m to about 3D mg/m . The removal efficiency of
about 50 % for HC1 and HP is poor, compared with wet processes,
but NO -removal is higher than with wet processes.
X
Objective of the current research program is the improvement
of HC1-, HF- and in particular IMO -removal with keeping the
high efficiencies for SO and fine particulates. Considered
measures are optimization of adsorption material and IMO -reduc-
tion by injection of ammonia, using carbon as catalyst.
5. Conclusions
The Federal Government intends to update the new source perfor-
mance standards for power stations within this year. In view of
the operating results with the demonstration units in the Fede-
ral Republic of Germany and the international experiencies, in
particular in the US and in Japan, the Federal Environmental
Agency has proposed to reduce the emission standards for parti-
culates and S02 and to set standards for NO , HC1 and HF.
-------
Emissions-En twick lung
NCb • CH
„„„—.H.,TT« m-mra—c*3isa__ —K«.»a ifpasM—E£33—tfiiai
1965 1970 1975 1980*) 1965 1970 1975 1980*1
|Kraft-u.
| Fernheizwerke
Industrie
HoushoJte u.
Klemverbroucher
Verkehr
StQUb
D
1965 1970~"l975 1980*' 1965 1970 1975 1980*'
*l geschatzt
UBA
1977
Entwicklung tier Johresemissian von SC^.Nt^.CH und
Sloub in der Bundesrepublik Deutschlana
LU-Emi
123
Langengrad
UBA
1977
Emteilungder Bundesrepublik Deulschland in
ErfassungsrbumelFlachenangabe in kml
IU-AII
001
to
ui
Fig. 1: Annual emissions of SO , IMD ,
hydrocarbons and particulates
in the Federal Republic of
Germany
Fig. 2: Map of the Federal Republic
of Germany
-------
33"
c-50
0 ^
02 cio
48°-
6C
fl
fl
B_B
01
ll
DJ3-M
Dli
on
n_s»
UBA
1977
n ra •
n E3 •
n ra •.
n ra «
n m •
S02-Emission
n a m
1965-
1970-
1975-
n PS •
Ffl ffl •
n~*inn
I1S5T
dJ]
7° 8° 9° 10°. 11° 12° 13°
Langengrad
Regionale SO? - Jahresemission
hr
H°
Lu - Emi
003
54° —
52° -|
•^
0
c*
(U
al
» 51°-
nn.
L
HI
6°
UBA
1977
D
P.
P.
7°
n «T» M
n «.
n •• •
Staub- Emission
n M
1965-
1970-
1975-
^
JL^'OO 000 1 / Jahr
fpB-
J^
i
tv>
O
i
8° 9° 10° 11° 12° 13° 14°
Langengrad
Regionale Staub - Jahresemission
LU -Emi
005
Fig. 3: Regional
Fig. l+: Regional particulate emission
-------
33"-
C-JO
Breitengrad
™ I:
o
1 1
nj9_n
flfl
ail
6°
UBA
1977
nil
nil
n'ffl •
n n B
n ra •
n ra •
n BS •
—
55" -
NOz- Emission
ja_nJL
—
19t5-
1970-
197 S-
f
n n •
Jl lOOOOOt/Jo
j.
r
7° 8° 9° 10° 11° 12° 13° 14°
Langengrad
Regionale NO 2 - Jahresemission
Lu - Emi
004
-o 3i
o
O)
c
Q>
?
00 si°-
cno
/QO
1*
ilj
nil
nil
niB
DiH
n H •
n H •
nil
niB
nil
-MM
nil
nil
PS HI MB
all
CH-Emission
P!iB
1955-
1970-
1975-
n m •
—
nfHB
0 IDODOOI/Johr
fflH
=LJ
r
I ^T
go yo go go ]Qo ^o ^° 13° K°
Longengrad
UBA
...„ Regionale CH - Jahresemission
Lu - Emi
006
to
Fig. 5: Regional NO -emission
X
Fig. 6: Regional hydrocarbon emission
-------
1974-1976
erloubt nach lA-luft
© bis 60(jg 50, /m3 Lull
(WHO- Emplehlung)
60 bis 140 pg SOj/m3Lult
unzulassig noch IA -lull
©uber KOpg SO, /m3 Lull
UBA
1977
Jahresdurchschnitt der Konzentrationen
des Schwefeldioxids an Medorten
mderBundesrepublik DeutschlanrJ
LU-Imm
029
S07-Trend
Icrlm-Dohlem
Bollrcp
ig r>
Oorlmund
Essen-Sud
Hnmbuig K H
Ingolsludl©
UBA
1977
'dm - Steglitz
Burghausen
it n
Ousseldorl
Frankluit Z SI
Hamburg H K
Karlsruhe 6
lerlm- Jungf
Coslrop-Rrjuxel
n ;s
Duisburg
Freiburg
Heilbronn
n n
Koln-Buchf
terus WS
Oarmsladt
frlangen ®
Gelsenkirchen Hogen
Deuselbach
Essen - Mitte
Johresmillelwerl
der SO, - Konzenlralion
Jahre
II IA lull
© Zohl att Heflilalionen
lendenz der millleren Scrjwefeldioxid-Jahresbelastung
n mchrjohrig konl.nuierlich uberwochlen Gebielen
(HesMji'Qen der Landesbenorden, des UBA.rJes BGA I
LU-Imm
027
I
K)
Fig. 7: Annual ambient S0?-concentration
at certain cities in the Federal
Republic of Germany
Fig. 8: Trend of ambient SCL-concentration
-------
Koln- Ehrenf
n is
ludwigsholen
Munchen (
Saorlouis
Wanne-Eickel
UBA
1977
Koln-Eilelw
Mainz
Nurnberg©
Schouinsland
Weslerland
S02-Trend
Koln-Merk
Mainz - Kaslell
Oberhausen
Stultgarl ©
I
Wiesboden-Mitte
Koln-Worr.
Mannheim©
Recklmghousen
Volklingen Woldhof
Laulerbach
Mulheim(BW)
« n
Saarbrucken
Jahresmillelwert
der SOj-Konzentration
tig 502
•Jahre
11 IA LuftP " [1
1 1964 J (IA LU||
© low in Hinslalnwn
lendenz der millleren Schwefetdioxid-Jahresbelastung
in mehrjahng konlmuierlidi uberwachlen Gebieten
(Messungen der Londesbehorden.dec UB& I
lU-Imm
028
Bochum
Essen
n n
Kerne
Mainz
Trend Staubniederschlag
Volklmgen
Witten
Bottrop
Castrop-Rauxel
Franklurl
Gelsenkircnen
jo is
Kossel
Mulheim
Wonne-Eickel
UBA
1977
Koln
Neunkirchen
Wattenscheid
Oillingen
Glodbeck
Krefeld
Neuss
Wetzlar
Dortmund
Haaen
Leverkusen
Oberrousen
Wiesbaden
Ousseldorf
Homm
Lunen
10 n B is
Reckunghouseri Saarbrucken
Ouisburg
Honou
.udwigshden
Relative Staubnieder-
schlogsbelastung/
Bezugsjohr
•I.
ISO-
J
IlllAluff
\ 1964 ,
Jahre
'n
TA Luftl
19R I
lendenz der millleren Johresbelastung durcn Slaub-
niederschlag ir langjahrig uberwachten Gebieten
I Messungen der Landesbenorden)
LU-Imm
040
CTv
CO
Fig. 9: Trend of ambient S0?-concentration
Fig. 10: Trend of particulate sedimentation
-------
Schwebstaubtrend
fcchu/n-Stodt1
Deuselbach
torlm.-Nette
n 75
ulsbufg-fiuhrort
rech. Schule
71 B
Hilden
Hurth
Bochum-We'trn
Dmslaken
Ousseld |B6A)
POsseld-Garath Dusseld Stockur
issen-Berres!'
GelsenlcSusr
Hosel
Kasse!-Mitte
UBA
1977
Bochum-Werne
Oormaqen
Esssn-Kamap
Gelser,k.-,Mi!!e
Horrem
Kelslerbach
Borllh-Ossert Sraljacklr'egel
Oarslen
Dorlmund-Derne
Cusseld Wrangeisli
Essen-Kijrertr
Keroen
issen-Steele
Gelsenk.tUBA;
Kellwiq
B'jdberg
DorlmHoesd'
Essen-
Oortm.-Horde
Frechen-Bsch.
Herne
Johresmillelwert
!er xhviEbstrjubkrjnzenlratcn
11 (A Lu!l
\ !96i /
Johre
I1/TA Lull
Ii974 /
fend^nz d?r mittieren Janresbe:as!ung durch Schwetjslaub
in langjahng uberwochlen Gebielen
(Messui^en der LcPdesighcn1'" 'jnd des UBA i
IU- I mm
041
Schwebstaubtrend
Kirch hcllen
II ' 75
ank Lalum
Neuss
Orsoy
71 7S
iodenkirchen
Weslerlond
Koln Br leld
lunen
Seviges
70 15
Porz
Schaumslo/id
W'estoden-M
UBA
1977
Koln-Ceufz
Mannne:rri
Nievenheim
Velberl
'.Siesbcden-5
Marl
OCerh Allst
Rat'ngen
Wdten
Mcers
Oberh-Osler
'Aolrog
reteid Fisjh
70 !5
Monheim
Oberh-Sterk
Rhe-nnausen
Wesel
Gpladen
Jahresmittetasrt
der Schwebstaubkonzeritrctior
ug Sf/m
II (lA LUt'i il/lA Lull^
lendenz der,?.:;! leren .cnreste.astung iurcn Scrwecsioub
in langjdhng uberwochlen Gebielen
(Messurqen der '.c^gog^^er '.inj dss JBi I
lU-Imm
042
I
to
Fig. 11: Trend of suspended particulate
concentration
Fig. 12: Trend of suspended particulate
concentration
-------
Goswdscher und Abscheider
Hohgos-
*" "ji L. x-CaswGSd
T/l-^
t.
Fnschwosser-
Heervasser -
ii
S02- Emission
, Vorlagebehailef
Klarwasserbehalter
Kecopurator
Anselzwosstr
Kolkmilch-Huckiout
Waschdussigkeit Klarphase-Reccpurator
UBA
1977
Flieflschemo der Anlage nach dem Bischoff-Verfohren fur 500.000 m3/h
LU-Emi
026
SO 2-Emission
Rohgas
Gips-
mdustne
Gips
UBA
1977
Flie(5schema d. Anlage noch dem Saarberg-
Holter-Verfahren fur 125.000 m3/h
LU-Emi
025
I
K>
(_n
I
Fig. 13: Bischoff (lime/sludge)
flue gas cleaning process
Fig. 1^: Saarberg-HBlter (lime/gypsum)
process for flue gas cleaning
-------
„ . Rohgas
Remgos 1000 000 mVh
135°C 130°C '"
A
S02-Emission
Abgas
Brennstoff I 385°C
740kg/h IzimKessell
Siebmaschine
Koks
100°C
U BA
1977
Verfnhrensschema einer Abgasenlschwefelungs -
anlage nach dem Bergbau -Forschungs-
Verlohren (300 MW)
LU- Emi
024
Fig. 15: Bergbaufarschung (carbon adsorption)
process for flue gas cleaning
I
NJ
-------
- 267 -
Discussion
The discussion dealt mainly with these new standards. Dr. Holighaus
wished to know whether the standard for S02 would be lower than at
present (2,75 kg per hour); Mr. Davids replied in the affirmative.
He added that new standards would be introduced for HC1 and HF,
but was unable to describe these at length as the full details had
not been finalized at the time. A comparison was then made with
standards in the US, which were 5 kg per hour.
-------
- 268 -
SESSION IV: HIGH TEMPERATURE AND PRESSURE PARTICULATE CONTROL
-------
- 269 -
AIR POLLUTION TECHNOLOGY, INC.
4901 MORENA BLVD., SUITE 402 SAN DIEGO, CA 92117 (714)272-0050
FUNDAMENTALS OF PARTICLE COLLECTION
AT HIGH TEMPERATURE AND PRESSURE
by
Dr. Richard Parker and Dr. Seymour Calvert
Air Pollution Technology, Inc.
Presented at the
US/FRG Particulate Workshop
Julich, Germany
March 16-17, 1978
ENGINEERING • CONSULTING • RESEARCH o DEVELOPMENT • DESIGN « EQUIPMENT APT
-------
- 270 -
ABSTRACT
High temperatures and pressures affect the physical mechanisms
by which particles are removed from gas streams. This presentation
examines the theoretical basis for predicting high temperature and
pressure effects on particle collection mechanisms. In general,
particles larger than a few tenths of a micrometer in diameter
appear to be more difficult to collect at high temperature and
pressure than at standard conditions. Experimental data are needed
to confirm these predictions. A U.S. EPA-sponsored project to ob-
tain experimental data is discussed and the test facility is des-
cribed.
-------
- 271 -
FUNDAMENTALS OF PARTICLE COLLECTION
AT HIGH TEMPERATURE AND PRESSURE
INTRODUCTION
When designing, troubleshooting, or evaluating the performance of par-
ticulate control equipment it is important to have a firm understanding of the
physical mechanisms by which the particles are removed from the gas stream.
This is especially true when the control device is to be used at high tempera-
tures and pressures (HTP) where current design models are unproven. In order
to provide a rational basis for design and scale up, a sound theoretical un-
derstanding of the HTP effects on particle collection mechanisms is essential.
We have conducted a thorough examination of the literature concerned with
HTP effects on particle collection (Calvert and Parker, 1977). Although HTP
particle collection has been of interest for over 30 years no fundamental
evaluation of the theory has been attempted. In general, conventional models
for particle collection (valid at low temperatures and pressures) have been
extrapolated to predict performance in HTP situations. Very few performance
data are available to evaluate these models at HTP conditions, especially as
a function of particle size.
This paper presents a review and evaluation of the theory normally used
to describe particle-collection mechanisms, and a discussion of the EPA-
sponsored experimental program currently under way at A.P.T., Inc.
THEORY
Particle collection devices usually can be characterized by a deposition
velocity, u,, which is related to the particle collection efficiency, n> and
the penetration, Pt, as follows:
Pt = 1 - n = exp I—^—-J (1)
-------
- 272 -
o
where A, = deposition area, cm
0 = volumetric flow rate, cm3/s
G
Particles are kept suspended in a gas stream by the viscous force (drag)
of the gas which resists forces tending to precipitate particles. The depo-
sition velocity for any collection mechanism depends on the balance
between the driving force (precipitating force) and the resistance force of
the gas.
The major difference between the collection of particles at normal con-
ditions and at high temperature and pressure is in the fluid resistance force.
The fluid resistance force is generally approximated by Stokes' law modified
to allow for non-continuum slip flow effects:
Uo
where Fr = fluid resistance force, dynes
\lr - fluid dynamic viscosity, g/cm-s
d^ = particle diameter, cm
UQ = relative velocity between the particle and the gas, cm/s
C1 = Cunningham slip correction factor, dimensionless
The temperature and pressure dependence of Equation 2 -is contained in
the terms y,- and C' . The viscosity of a gas increases with increasing
temperature. At extreme pressures, viscosity also increases with pressure.
This effect is not significant at pressures below about 20 atm. Adequate
theory and experimental data for predicting viscosities at high temperature
and pressure are available in the literature.
The Cunningham slip correction factor may be calculated as:
C1 = 1 + — fl.257 + 0.40 exp (-1.1 dp/2X)| (3)
where X = mean free path of gas molecules, cm
The Cunningham slip correction factor is a function of temperature,
pressure, and particle diameter. It becomes important for small particles,
high temperatures, and low pressures.
-------
- 273 -
Equation 3 is an empirical expression based on Millikan's oil drop ex-
periments (conducted at room temperature and reduced pressure). The constants
are dependent on the momentum transfer (and hence accommodation coefficient)
between the gas molecules and the particle and may not be accurate at extreme
temperatures. Experimental data are needed to resolve this uncertainty.
The particle deposition velocity for most collection mechanisms of in-
terest is inversely proportional to the fluid resistance force, and therefore
proportional to the ratio C'/VU . The effects of high temperature and pres-
sure on this ratio, plotted as a function of particle diameter, are illus-
trated in Figure 1. At atmospheric pressure, the ratio decreases with in-
creasing temperature for particles larger than about 0.4 ym. At 15 atm pres-
sure, the ratio decreases with temperature for all particles larger than 0.1
ym. Therefore, the particle deposition velocity will generally be smaller at
high temperature and pressure than at normal conditions.
Inertial Inipaction
One of the most important mechanisms for the collection of particles
larger than a few tenths of a micrometer in diameter is inertial impaction.
Inertial impaction takes advantage of the difference in mass between the par-
ticles and gas molecules by impinging them on a target. The relative effect
of inertial impaction for different particles and flow conditions may be char-
acterized by the inertial impaction parameter, K , defined as:
C1 p d2u
K = P P ° (4)
P 9 ^G dc
where p = particle density, g/cm3
d = characteristic diameter for collector, cm
The inertial impaction parameter is equivalent to the ratio of the par-
ticle stopping distance, x , to dc/2 . The particle stopping distance is
that distance the particle would travel before coming to rest if injected into
a still gas at a velocity, u , when only the fluid resistance force acts on
the particle. By considering the particle stopping distance divided by UQ ,
the particle's relative inertia can be characterized by a relaxation time,
T, defined as:
-------
- 274 -
x K d C1 p d 2
s pc p p
u ~~ 2 u 18 U- ^
O O Ci
From Equation 5 it can be seen that the effects of high temperature and
pressure on the particle relaxation time come in through the ratio C'/yp
Therefore Figure 1 can be used to illustrate the effects of high temperature
and pressure on the particle relaxation time. A longer relaxation time (lar-
ger C'/yr ) implies that the particle can more easily be removed from the
gas by inertial impaction. For large particles, therefore, inertial impaction
decreases with increasing temperature and pressure.
For small particles (less than about 0.3 ym) at high temperature and at-
mospheric pressure, Figure 1 indicates that inertial impaction begins to im-
prove with temperature. However, high pressure tends to nullify this bene-
ficial effect of high temperature.
Brownian Diffusion
Small particles can undergo significant Brownian motion resulting from
the random bombardment of the particle by gas molecules. The rate of diffu-
sion is characterized by the particle diffusivity, defined as:
D = C' k T (6)
where D = particle diffusivity, cra2/s
k = Boltzman's constant, erg/°K
T = absolute temperature, °K
Figure 2 shows the effects of temperature and pressure on particle diffu-
sivity. Smaller particles undergo higher rates of diffusion. High ••empera-
ture increases the diffusivity for all particle sizes. High pressure de-
creases the beneficial effect of high temperature because of its effect on
the mean free path in C1 .
Electrical Migration
The migration of a particle in an electric potential field is propor-
tional to the field strength, the particle charge, and the fluid resistance
force.
-------
- 275 -
Electrical migration is generally characterized by a deposition velocity which
may be approximated as:
u, = -^—T C10?) (7)
where u = deposition velocity, cm/s
G
qp = particle charge, C
E = electric field strength, V/cm
For a given field strength and particle charge, the effects of tempera-
ture and pressure are contained in the ratio C'/Pg and are illustrated in
Figure 1. The particle charge and electric field strength, however, are also
complicated functions of temperature and pressure.
Gravitational Settling and Centrifugal Separation
Using Equation 2 to describe the fluid resistance force, the gravita-
tional settling velocity and the deposition velocity of a particle in a
centrifugal force field may be approximated as:
18 U
uc = TT — C9)
18 yG
where us = gravitational settling velocity, cm/s
uc = centrifugal force deposition velocity, cm/s
g = acceleration of gravity, cm/s2
ut = tangential particle velocity at radius R , cm/s
R = radial position of particle, cm
pp = density of the gas, g/cm3
In general, even at relatively high pressures (~50 atm), the gas density
is much smaller than the particle density and may be neglected in Equations 8
and 9. Therefore the temperature and pressure dependence of Equations 8 and 9
is contained in the ratio C'/PG anc* ^s illustrated in Figure 1.
-------
- 276 -
Particle Agglomeration
One way to improve the collection efficiency for fine particles is to
cause the fine particles to agglomerate into larger aggregates which can be
collected more easily.
Particles undergoing random Brownian motion will tend to agglomerate
over a period of time. The rate of agglomeration is generally considered to
be proportional to the square of the particle number concentration. That is:
dNP
= -K N 2 (10)
dt op
-3
K =proportionality constant or agglomeration coefficient, cm3/s
where N =particle number concentration, cm
Using Equation 2 for the gas resistance force, Fuchs (1964) presents
the following equation for the agglomeration coefficient of a particle under-
going Brownian motion in a still gas, assuming particles stick together upon
touching:
.
KQ = 4 TT D dp = _
The agglomeration coefficient is shown as a function of temperature,
pressure, and particle diameter in Figure 3. The agglomeration of particles
increases with temperature and decreases with pressure. The net effect of
high temperature and high pressure (20°C, 1 atm to 1,100°C, 15 atm) is to
increase the rate of agglomeration for a 1 urn diameter particle by a factor
of 1.5 (K0 increases from 3. 5 x 10~10 cm3/s to 5.3 x 1CT10 cm3/s) . For a 0.1 urn
diameter particle, the rate of agglomeration remains relatively constant
(KQ = 8.5xlO~10 cm3/s) . Therefore it appears that at high pressure and high
temperature the rate of agglomeration increases for particles larger than
0.1 urn. At high temperature and atmospheric pressure, the rate of agglom-
eration of fine particles should increase more substantially.
Particles may also agglomerate as a result of turbulence, particle
charge, and sonic disturbances. These agglomeration mechanisms were examined
theoretically by Calvert and Parker (1977) and did not appear to offer any
-------
- 277 -
improvement at high temperature and pressure. Sonic agglomeration appeared
to increase with temperature but this was countered by a substantial decrease
at high pressures.
EXPERIMENTAL PROGRAM
Test Facility
An experimental program to study fundamental particle collection mech-
anisms at high temperature and pressure is under way at A.P.T., Inc. under
EPA sponsorship. The experiments will investigate the collection mechanisms
of inertial impaction, Brownian diffusion, and electrical migration at tem-
peratures up to 1,100°C and pressures up to 15 atm. Particles in the general
size range of 0.5 to 10 urn will be considered.
A special high temperature and pressure test facility has been designed
and constructed. This facility is illustrated in Figure 4. All the high
temperature and pressure components are located inside a steel safety barri-
cade. Tests are controlled remotely at the control panel.
High pressure gas is supplied by a manifold of nitrogen gas cylinders.
The gas then passes through a high pressure redispersion fly ash dust gen-
erator and a cyclone precutter. The dust generator is a batch type high
pressure blender. Steady output concentration and size distribution can
be maintained for approximately 2 to 4 hours.
The gas and particles are heated in two stages. The first stage uses
high temperature heating tapes which can raise the gas temperature to 750°C.
The second stage uses resistance heated tube furnaces to increase the gas
temperature to a maximum of 1,100°C. Stainless steel type 316, Inconel 600,
or HastelloyX are used for high pressure piping and flanges depending on the
maximum temperature anticipated at specific locations.
The high temperature and pressure gas passes through one of three
specially designed test sections and is then cooled and returned to the
control panel before being vented. The test sections are designed to iso-
late specific particle collection mechanisms for study at the high tempera-
ture and pressure conditions. Specific test sections are discussed in more
detail later.
Isokinetic samples are taken at the inlet and outlet of the test section.
The samples are collected on sintered metal filters which can be used at
-------
- 278 -
temperatures up to 1,100°C in the nitrogen environment. There is a
provision for adding dilution flow before the filters so that low tempera-
ture filters can also be used.
The filter samples are removed after each test and are analyzed using
an electronic particle counter (Coulter Counter Model TA-II) to determine
the mass and size distribution of the fly ash collected on each filter.
Also the sample probes are washed after each test and analyzed to determine
the amount and size of particles deposited in each probe.
One potential problem using the electronic particle counter is that
particles which may have been agglomerates in the test gas stream will be
analyzed as single particles. To investigate this problem we have run
parallel size distributions using a cascade irnpactor and a filter (analyzed
with the electronic counter). The results were in very close agreement.
Also we have used an optical microscope to observe particles collected on
a glass slide. There appeared to be very few agglomerates. The dilution
line enables us to use cascade impactors at low temperature for com-
parison with the data we obtain from the sintered metal filters. Also we
will examine samples using the microscope as a further check on our analysis.
Inertial Impaction Tests
The inertial impaction test section is illustrated in Figure 5. It is
essentially a single stage impactor placed between two flanges. Five sepa-
rate jet plates are available so that we can look at cut diameters (diameter
corresponding to 50% particle collection) ranging from 0.5 urn to 10 nm.
Particles are collected on a ceramic fiber substrate which is used to
minimize particle bounce at the impaction plate. The substrate will be re-
moved and analyzed after each test in order to complete the mass balance of
particles and to check the efficiency determined from the inlet and outlet
samples. We have calibrated this impactor under controlled conditions in
the laboratory using monodisperse particles impacting on a greased plate
and on the fiber substrates. The agreement between greased and fiber sub-
strates was good.
The data obtained from analysis of the inlet and outlet samples will be
used to determine an experimental penetration curve as shown in Figure 6.
The penetration curve will be used to determine an experimental cut diameter,
-------
- 279 -
dcx- Experiments will be run at temperatures ranging from 100°C to 1,100°C
and pressures from 1 to 15 atm.
Conventional impaction theory will be used to predict a cut diameter,
dcp, based on the impactor calibration. The predicted cut diameter will be
determined from the following equation:
(12)
where K = calibrated value for Kp at 50% collection efficiency
d, = jet diameter, cm
ujj = jet velocity, cm/s
The particle density will be determined by comparing the calibrated cut dia-
meter with the cut diameter measured using fly ash at standard temperature
and pressure.
By comparing the experimental and predicted cut diameters at various
temperatures and pressures we will be able to evaluate the theory as a
function of temperature and pressure. The viscosity of nitrogen gas has
been determined at temperatures up to 1,200°C (Saxena, 1971). Therefore
any discrepancies between the experimental and predicted cut diameters at
high temperatures can be related to the slip correction factor in Stokes'
law (Equations 2 and 3) .
Brownian Diffusion Tests
The diffusion test section is illustrated in Figure 7. It is basically
a screen-type diffusion battery held inside high temperature and pressure
pipe. The screens are 120 mesh and made of 316 stainless steel. There
will be 50 to 100 screens in the test section.
Particle penetration through the screens will be measured as a function
of particle size using the electronic counter for particles down to 0.3 ym
diameter. Cascade impactors will also be used to measure particles as small
as 0.1 urn. The penetrations will be measured at temperatures ranging from
100°C to 1,100°C and pressures from 1 to 15 atm.
-------
- 280 -
Predictions of penetrations for various temperature and pressure condi-
tions are shown in Figure 8. The predictions were based on the theory pre-
sented by Patterson and Calvert (1977). That is:
Pt = exp
{- 6.0 S Np"^67! (13)
where S = geometric solidity factor, dimensionless
N = Peclet number = urd,./D. dimensionless
DC* «J W
u = superficial gas velocity, cm/s
d.. = wire diameter of screen, cm
This theory will be confirmed by experimental penetrations at ambient
conditions and then used to predict experimental particle diffusivities from
penetrations obtained at HTP conditions. The experimental particle diffu-
sivities thus obtained will be compared with theoretical predictions.
Electrical Migration Tests
The electrical migration test section is illustrated in Figure 9. It
is basically a laminar flow, concentric cylinder electrical precipitator.
Particles will be charged to saturation using a corona charging section at
the outlet of the dust generator (before heating). The particle charge col-
lected will be measured using an electrometer. The size distribution col-
lected will be measured using the electronic counter. From the size distri-
bution and total charge collected we will estimate the charge per particle.
These data will then enable us to predict a deposition velocity using the
expression:
(G| qD v
J107 (14)
3 TT y_ d T '
G p o
where V = applied voltage, V
r = radius of outer electrode, cm
The particle penetration through the electrical migration test section
will be measured by taking inlet and outlet samples and analyzing them for
particle mass and size distribution. Experimental migration velocities will
-------
- 281 -
be determined from the basic equation for a laminar flow electrical pre-
cipitator:
u =
nQG
(15)
e TT D L
c
where D = diameter of outer electrode, cm
L - length of cylindrical electrodes, cm
The particle saturation charge can be predicted theoretically (White,
1963). Using a charging field strength of 6 kV/cm, particle charges have
been predicted and used to estimate the collection efficiency of our test
section at various temperatures and pressures. The results are shown in
Figure 10. A field strength of 1 kV/cm was assumed for the precipitator
with an actual flow rate of 472 cm3/s (1 ACFM).
The comparison between experimental and predicted migration velocities
will enable us to evaluate the conventional theory and determine its suit-
ability for use in design models for high temperature electrical precipita-
tion.
CONCLUSIONS
From theoretical considerations it appears that the collection of par-
ticles larger than a few tenths of a micron in diameter will be more diffi-
cult at high temperature and pressure than at standard conditions. This is
largely a consequence of the stronger drag force exerted on particles in
high temperature and pressure gas streams.
The theoretical predictions presented in this paper are based on extra-
polation of current aerosol theory to high temperature and pressure condi-
tions. Satisfactory theory and experimental data exist for predicting the
gas properties at these conditions. However, the available data are insuf-
ficient to validate theoretical predictions for particle motion.
To obtain the necessary data, experimental measurements of the fluid
resistance force, particle diffusivity and electrical deposition velocity at
high temperature and pressure are being made. This experimental research
program is scheduled for completion in May of next year.
ACKNOWLEDGEMENT
This work is supported by the U.S. Environmental Protection Agency.
-------
- 282 -
REFERENCES
1. Calvert, S. and R.D. Parker, "Effects of temperature and pressure on
particle collection mechanisms: theoretical review," A.P.T., Inc.,
EPA-600/7-77-002, NTIS PB-264-203, January 1977-
2. Fuchs, N.A., The Mechanics of Aerosols. Pergamon Press, New York, 1964.
3. Saxena, S.C. Transport properties of Gases and Gaseous Mixtures at
High Temperatures. High Temp. Sci. ,3_:168, 1971.
4. Patterson, R.G. and S. Calvert. Screen Diffusion Battery for Monitoring
Submicron Aerosols in Stack Gases. Presented at AIHA Conference, New
Orleans, Louisiana, May 24, 1977.
5. White, H.J, Industrial Electrostatic Precipitation. Addison-Wesley
Publication Company, Reading, Massachusetts, 1963.
-------
- 283 -
LIST OF SYMBOLS
A, = deposition area, cm2
C1 = slip correction factor- dimensionless
D = particle diffusivity, cm2/s
d = outer electrode diameter, cm
D = collector diameter, cm
d, = jet diameter, cm
d = particle diameter- cm
d = wire diameter, cm
E = electric field strength, V/cm
F = drag force, dynes
g = gravitational acceleration, cm/s2
k = Boltzman's constant, erg/°K
K = agglomeration coefficient, cm3/s
K = inertial impaction parameter, dimensionless
K = K at 50% collection efficiency, dimensionless
Pso P
L = length of electrode, cm
N = particle number concentration, cm"3
Np = Peclet number, dimensionless
Pt = penetration, dimensionless
QP = volumetric flow ratt, cm3/s
q = particle charge, C
R = radial position of particle, cm
r = radius of outer electrode, cm
o
S = solidity factor, dimensionless
T = absolute temperature, °K
t = time, s
-------
- 284 -
u = centrifugal force deposition velocity, cm/s
u, = deposition velocity, cm/s
u = electrical migration deposition velocity, cm/s
c
ur = superficial gas velocity, cm/s
u, = jet velocity, cm/s
u = relative velocity between particle and gas, cm/s
u = gravitational settling velocity, cm/s
u = tangential velocity of gas, cm/s
V = applied voltage, V
x = stopping distance, cm
T) = collection efficiency, dimensionless
X = mean free path, cm
u = gas viscosity, g/cm-s
PG = gas density, g/cm3
p - particle density, g/cm3
T = relaxation time, s
-------
- 285 -
I
E
O
1 atm
- 15 atm
0.5 1.0
PABTICIE
Figure 1. The effect of HTP on the ratio C'/yG
-------
- 286 -
Figure 2. The effect of HTP on particle diffusivity.
-------
- 287 -
X
6
u
H
2
W
n
CJ
t—I
PU
n<
W
o
u
z
o
W
s
o
-J
u
o
<
Particle Diameter,
200
400 600 800
TEMPERATURE, °C
1,000 1,200
Figure 3. The effect of HTP on Brovmian agglomeration.
-------
NJ
00
CD
GAS SUPPLY
Figure 4. Test facility.
-------
GAS FLOW
THERMOCOUPLE
GASKETS
to
00
Figure 5. Impaction test section.
-------
- 290 -
100
o
i—i
H
W
Pu
50
EXPERIMENTAL
PREDICTED
d d
ex cp
PARTICLE DIAMETER
Figure 6. Comparison of Experimental and Predicted
Cut Diameters.
-------
•SCREENS
SUPPORTS
GAS FLOW
Figure 7. Diffusion test section.
-------
- 292 -
100
E-
f-
50 -
0.5
PARTICLE DIAMETER, ym
Figure 8. Predicted penetrations for diffusion test
section.
-------
•HIGH VOLTAGE
SUPPLY
GAS OUTLET
/ ' f A I ' i
f'JJ/Jfff-t t s fJ-Liij^ii i i r.j
\ \ \ \\S--r
INSULATION
COLLECTION
ELECTRODE
TTi
ELECTRODE
ELECTROMETER
O-
GAS INLET
K)
Figure 9. Electrical migration test section.
-------
100
u
U
W
0.1
0.5 1
PARTICLE DIAMETER, ym
10
I
NO
Figure 10. Predicted efficiencies for electrical migration test section.
-------
- 295 -
Discussion
Mr. Gillthner asked whether:
1) APT had started measuring collection efficiency in electro-
static precipitators and fabric filters.
2) It was really essential to use such expensive alloys.
3) Details regarding requirements for new collection techniques
in terms of clean gas dust load and pollution load could be
given.
Dr. Parker's answers were as follows:
1) Work in this field had not been part of the program. U.S.EPA
intended to perform tests on baghouses at Exxon Research
Company in New Jersey. Our work concerns high temperature
and pressure fundamentals as stated in the contract.
2) California law requires that pressure vessel code be observed.
This necessitated good engineering judgment, and use of the
best materials. Steel, nickel and cobalt alloys were chosen,
because of the extremely high temperatures and pressures.
3) A two-fold requirement existed:
a) Similar to that for coal-fired boilers.
b) Tolerance of gas turbines for fine particulates.
Erosion damage to the turbine blades occurred when particles larger than
2-3 microns were present. Alkali metals in ash on the blades could also
prove hazardous; deposits might build up, imparing performance, or
causing further erosion by breaking off. Problems of deposition were
important; an outlet concentration of 0.002 grains per std. cu. ft. was
desirable.
-------
- 296 -
A controversy existed as to whether more stringent requirements should
be introduced to protect the turbine or the environment (Mr. Princiotta).
Mr. Parker pointed out the relationship between turbine protection and
larger micron particles and environmental protection and sub-micron
particles. Further details of U.S. requirements were furnished by Mr. Princiotta.
Dr. Holighaus mentioned work being carried out at NASA on damage to
blades by small particulates; more research in this area was needed. He
also asked whether one could find the best process for separation by
using theoretical methods. Dr. Parker replied that operational problems
presented the most serious questions. Electrostatic precipitators could
be operated at higher voltages in high pressure gases resulting in the
possibility of higher collection efficiency. However, practical problems
have not been identified, nor have electrostatic precipitators been
tested for collecting dust under extreme temperature and pressure conditions.
-------
- 297 -
AIR POLLUTION TECHNOLOGY, INC.
4901 MORENA BLVD., SUITE 402 SAN DIEGO, CA 92117 (714)272-0050
GRANULAR BED FILTERS AND
DRY SCRUBBERS
by
Dr. Richard Parker, Dr. Seymour Calvert,
Mr. Shui Yung, and Dr. Ronald Patterson
Air Pollution Technology, Inc.
Presented at the
US/FRG Particulate Workshop
Julich, Germany
March 16-17, 1978
ENGINEERING « CONSULTING • RESEARCH • DEVELOPMENT • DESIGN • EQUIPMENT APT
-------
- 298
ABSTRACT
This presentation discusses the use of granular bed filters
and dry scrubbers for removing particulate matter from high tem-
perature and pressure gas streams. Engineering models are pre-
sented and performance predictions are made for high temperature
and low temperature applications. Experimental data are presen-
ted for verifying the models at lov,r temperature. The primary
collection efficiency obtainable using granular bed filters or
dry scrubbers is sufficient to meet current environmental regu-
lations. However, there are many operational problems which need
to be resolved before these devices will be sufficiently reliable
for commercial application.
AIR POLLUTION TECHNOLOGY, INC. 4901 MORENA BLVD., SUITE 402- SAN DIEGO, CA9211?
-------
- 299 -
GRANULAR BED FILTERS AND DRY SCRUBBERS
INTRODUCTION
Granular bed filters and dry scrubbers have been proposed
for the removal of particulate matter from high temperature and
pressure gases. One such application would be as the tertiary
collection device in a pressurized fluidized bed boiler power
plant as illustrated schematically in Figure 1. The gas leaves
the boiler at a temperature of approximately 900°C and a pressure
of about 10 atm. First it passes through a primary cyclone which
removes larger particles (including unburnt carbon) and recycles
these particles to the combustor. The gas leaves the primary
cyclone and passes through a secondary cyclone or multiclone
separator. This removes more large particles and reduces the
mass loading of particulate to the order of 1 gr/SCF.
A tertiary cleanup device is necessary to reduce the parti
culate loading sufficiently to protect the gas turbine from ex-
cessive erosion and corrosion damage. It is also desirable for
the gas at this stage to be sufficiently clean to satisfy all
emission regulations. However, if necessary, it is possible to
satisfy the emissions regulations by cleaning the gas downstream
from the turbine using conventional control technology.
The best available information concerning the size and mass
loading of particulate entering the tertiary collection stage of
a pressurized fluidized bed process has been reported by Hoke
et al. (1977a). The mass loading has been found to be typically
about 1 gr/SCF. The size of this particulate is represented in
Figure 2. Approximately 30% of the mass is smaller than 2 vim.
For this size distribution and loading, approximately 90-95%
removal is required to meet U.S. EPA particulate emissions regu-
lations. As much as 99.81 removal may be required to protect the
gas turbine, although there is still much debate regarding turbine
tolerances for fine particles.
AIR POLLUTION TECHNOLOGY, INC. 4901 MORENA BLVD., SUITE 402- SAN DIEGO, CA 92117
-------
- 3OO -
GRANULAR BED FILTERS
Granular bed filters are defined as any filtration system
comprised of a stationary or slowly moving bed of separate re-
latively close-packed granules as the filtration medium. There
are three basic approaches to granular bed filtration currently
being developed. The first class are referred to as "fixed bed"
filters. The bed of granules is kept stationary during filtration.
Usually, it is cleaned by a reverse flow of gas which fluidizes
the bed and entrains the collected particulate. The principal
advantage of the fixed bed approach is that the granules do not
have to be recirculated and thus can be used for many filtration
cycles. For this reason, fixed beds have potentially lower opera-
ting costs.
In a second type of granular bed filter, referred to as the
"moving bed" filter, the gas flows through a slowly moving bed of
granules. The collected dust particles are carried with the bed
and later removed from the granules before the granules are recir-
culated. Moving beds have the advantage that the granules and
dust are separated externally and are free from the problem of
particle buildup and bed plugging. An inexpensive but effective
means of recirculating the granules could significantly lower
operating costs.
The third type of filter, the "intermittently moving" filter,
is really a hybrid of the fixed and moving bed filters. This
filter uses stationary granules for filtration and causes the
granules to move through the system intermittently during the
cleaning cycle.
The major problems with each approach are summarized in
Table 1. Many of these problems can be resolved through further
research and development. Granular bed filters are used success-
fully to control emissions from clinker coolers in the cement
industry and on hog-fuel boilers in the forest products industry.
They operate in the range of 100-200°C and near atmospheric pres-
sure. However, applications for controlling fine particulate at
elevated temperatures and pressures are very scarce. Granular
AIR POLLUTION TECHNOLOGY, INC. 4901 MORENA BLVD., SUITE 402-SAN DIEGO,CA92H?
-------
- 301 -
bed filters are attractive for these applications because they
can be made to withstand high temperature and pressure environments
relatively easily. However, more work is needed to determine
whether granular bed filters are satisfactory from the point of
view of primary collection efficiency, cost, and reliability.
Granular Bed Filter Model
Pressure Drop
Ergun (1952) proposed the following equations to describe
the pressure drop for flow through packed beds:
f Z u* (1-e) Pf
-AP = _H <1 (1)
f = . + 1.75 (2)
NRe
d ur pr
and NR = b (3)
Predictions based on these equations agree very well with experi-
mental data (Figure 3). The agreement is so good, in fact, that
equation (1) has been used with pressure drop data to determine
an experimental "effective bed porosity" for irregular, non-uni
form size bed granules.
Collection Efficiency
No available models were found satisfactory for predicting
collection efficiencies for granular bed filters operating at
velocities likely to be encountered in practice. We have
developed a performance model which has been used to predict
collection efficiency for lab-scale and full-scale filters. The
model is based on the collection efficiency of a clean granular
bed. The collection efficiency would be affected if there were
a significant filter cake on the bed surface and within the bed.
AIR POLLUTION TECHNOLOGY, INC. 4901 MORENA BLVD., SUITE 402-SAN DIEGO, CA 92117
-------
- 302 -
The presence of a surface cake, however, has not been noticed
by many investigators working with large-scale filters. We feel
the clean bed model predicts a conservative estimate of the effi-
ciency attainable by granular filtration and is a satisfactory
model for filters which operate primarily without the presence
of a filter cake.
The granular bed can be envisioned as a great number of im-
paction stages connected in series as illustrated in Figure 4.
Particle collection is by inertial impaction and is similar to
collection in a cascade impactor. The jet openings are the pores
in each layer of granules. It is assumed that the jet diameters
in the granular bed are of uniform size and are equal to the
hydraulic diameter of the void space. The gas velocity in. the jet
is the average superstitial gas velocity.
If 'n' is the collection efficiency of one impaction stage,
the particle penetration for the granular bed will be,
Ptd = (l-n)N (4)
where Pt-, = penetration for particles with diameter, d ,
fraction
r\ = single stage collection efficiency, fraction
N = number of impaction stages
As in some cascade impactors, each layer of granules serves
both as the jet plate and as the collection plate. Therefore,
each layer is an impaction stage and 'N1 is equal to the number
of granular layers in a bed. For a randomly packed bed,
N = - — (5)
2 d
c
where Z = bed depth, cm
d = granule diameter, cm
AIR POLLUTION TECHNOLOGY, INC. 4901 MORENA BLVD.,SUITE 402-SAN DIEGO,CA9211?
-------
- 303 -
and,
Ft, = (1-n)
(1.5Z/dJ
(6)
The impaction collection efficiency, n, is a function of,
K , the inertial impaction parameter. The impaction parameter is
defined as,
where
C' P,
u.
C1
P
d.
Cunningham slip factor, dimensionless
particle density, g/cm3
particle diameter, cm
- — -4- 1 ,-. " -*- /
JCC V C-LVJV-. JL l_y , \-m/ o
gas viscosity, poise
jet diameter, cm
(7)
since,
u. = u,,. =
1 Gi
(8)
and,
d. = 4rR =
J H 3
(9)
where u
Gi
e =
rH =
d =
average interstitial gas velocity, cm/s
bed porosity, fraction
hydraulic radius, cm
granule diameter, cm
we have,
(10)
The relationship between, n, and, K , can be evaluated once
the flow field is defined. Flow fields reported in the literature;
e.g. Ranz and Wong (1952) and Marple (1970) are adequate for
AIR POLLUTION TECHNOLOGY, INC.
4901 MORENA BLVD., SUITE 402- SAN DIEGO, CA 92117
-------
- 304 -
K > 0.15. For K < 0.15, there is no suitable flow field repor-
p P
ted in the literature. Therefore, the relationship between, n,
and K , cannot be calculated analytically from the available
literature.
The single stage collection efficiency has been calculated
as a function of K from equation (6) and experimental data.
Figure 5 shows the results. There is scatter in the lower end of
the curve. For K < 10" , n,is very sensitive to experimental
data. A few percent scatter in data will cause, n , to fluctuate
greatly. Figure 5 compares the experimentally determined, n,
versus, K , curve with those reported by Ranz and Wong (1952),
Stern et al. (1962), and Mercer and Stafford (1969). All reported
curves are for K > 0.15. As can be seen, the present study is
consistent with other researcher's results and is a continuation
of their curves into the range most likely to be important for
f
high temperature and pressure filtration.
Paretsky et al. (1971) and Knettig and Beeckmans (1974)
studied the collection of monodispersed aerosol particles in gran-
ular bed filters. Their data were transformed into, K , versus
n, plots as shown in Figure 6.
Knettig and Beeckmans (1974) used 425 ym glass beads as
granular material. Bed porosity was 0.38. Aerosol particles
were 0.8, 1.6, and 2.9 ym in diameter. As can be seen from Figure
6, their data are in close agreement with the results of the pre-
sent study.
Paretsky et al. (1971) investigated the filtration of 1.1 ym
diameter polystyrene latex aerosols by beds of sand. They used
a bed of 10-14 mesh (1,200-1,700 ym) angular sand and a bed of
20-30 mesh (500-850 ym) sand at superficial gas velocities between
0.3 and 80 cm/s. Bed porosities were 0.41 and 0.43, respectively.
We have calculated single stage collection efficiencies from their
data. In the calculation, the granule diameters were assumed to
be the arithemtical mean of the smallest and the largest granule
size in the bed. The results are plotted in Figure 6. For a
given inertial parameter, Paretsky et al.'s data give a higher
AIR POLLUTION TECHNOLOGY, INC. 4901 MORENA BLVD., SUITE 402-SAN DIEGO,CA92H?
-------
- 305 -
collection efficiency than reported in this study. Their data
would be close to that of the present study if the smallest
granule diameter were used instead of the arithmetic mean.
The design model is based on particle collection by a clean
bed. If there is no filter cake formed on the surface and the
collected particles are uniformly distributed in the bed, the
model should be applicable. The design model has been used to
predict the performance of a Rexnord gravel bed filter and of
Combustion Power Company "dry scrubber." The predictions are com-
pared with field sampling data taken from the literature.
Rexnord Gravel Bed Filter
McCain (1976) conducted a performance test on a Rexnord
gravel bed filter. The gravel bed filter was installed to clean
emissions from a clinker cooler in a Portland cement plant.
Samples were taken simultaneously at the filter inlet and
outlet with cascade impactors. The operating conditions of the
gravel bed were:
Gravel diameter = 4 mm
Face velocity = 73 cm/s
Gas temperature = 175°C
Pressure drop = 25.4 cm W.C.
It was assumed that there was no surface cake and that the
pressure drop across the bed was 80% of the overall pressure
drop. Ergun's equation was used to estimate a porosity of 0.25.
With this bed porosity, the grade penetration curve was calcula-
ted for the operating conditions listed above.
Figure 7 is the predicted grade penetration curve along with
that measured by McCain (1976). The prediction is in good agree-
ment with the data.
Combustion Power Company Moving Gravel Bed
Hood (1976) reported the evaluation of the Combustion Power
Company (CPC) moving gravel bed filter on the control of parti
culate emissions from a hog-fuel fired boiler. The gravel bed
AIR POLLUTION TECHNOLOGY, INC.
4901 MORENA BLVD., SUITE 402-SAN DIEGO, CA92117
-------
- 306 -
filter was a prototype unit with suggested capacity of 1,133 Am3/
min (40,000 ACFM). The bed was packed with an intermediate-size
gravel which was retained on a 3.2 mm (1/8") wire mesh and passed
at 6.4 mm (1/4") mesh screen. The bed was a single do^^rn-flowing
annulus 2.6 m (8.5 ft) O.D. and 1.8 m (6 ft). l.D.
During sampling the unit was operated at a flow rate of 1,558
m3/min (55,000 ACFM). The gas temperature was 177°C (350°F).
Figure 8 shows the penetration curves for three sampling runs.
The average granule diameter was assumed to be 4.6 mm and the
bed porosity was calculated to be 0.25. The predicted grade pene-
tration curve is shown in Figure 8. The predicted penetration is
higher than that measured. Recent data obtained by A.P.T. on the
CPC moving bed filter is shown in Figure 9. These data agree
well with the model's prediction.
PERFORMANCE PREDICTION FOR HTP APPLICATION
Exxon Research and Engineering Company installed a Ducon
granular bed filter at their fluidized bed coal combustor mini-
plant. The bed is packed with Agsco no. 2 quartz (400 ym mean
diameter) to a depth of 3.8 cm (1.5 in.). The temperature of
the flue gas from the combustor is 870°C (1,600°F) and the pres-
sure is 10 atm.
According to Hoke (1977b), there is no surface cake formed
and the fly ash is uniformly distributed in the bed. For this
condition, the clean bed model can be used to predict the per-
formance of the Ducon granular bed in the miniplant.
From the pressure drop data reported by Exxon, the bed poro-
sity was estimated to be 0.24. Figure 10 shows the predicted
performance of the miniplant granular bed filter at high tem-
perature and high pressure and at ambient conditions. A particle
density of 1.5 g/cm3 was used in the calculation.
If the particle size distribution is known, Figure 10 can be
used to estimate the overall collection efficiency of the Ducon
granular bed filter. The equation relating the efficiency for
collecting one particle diameter to the overall efficiency is:
AIR POLLUTION TECHNOLOGY, INC. 4901 MORENA BLVD., SUITE 402-SAN DIEGO,CA9211?
-------
- 307 -
E = 1-Pt =1 / Ptd f(d ) d (d ) (11)
where E = overall collection efficiency, percent or fraction
Pt = overall penetration, fraction
Ptj = penetration for particles with diameter, d _,
fraction
f(d ) = particle size frequency distribution
d = particle diameter, ym or cm
Hoke (1977a) reported data on the particle size distribution
leaving the secondary cyclone (Figure 2). The mass median dia-
meter is 3.5 ym and the geometric standard deviation is 2.9.
The mass loading is approximately 2.3 g/Nm3 (1 gr/SCF). The over-
all collection efficiency for this size distribution was calcu-
lated graphically to be 95.1% (2.9% penetration). Experimental
efficiencies of 95-97% were reported by Hoke (1977b).
As can be seen from Figure 10, the granular bed filter should
be very efficient for all particles larger than 1 to 2 ym in dia-
meter. Whether or not this filter is efficient enough to protect
the gas turbine will depend strongly on the turbine's tolerance
for submicron particles.
DRY SCRUBBERS
Moving bed filters are sometimes referred to as dry scrub-
bers. Another type of dry scrubbing system has been developed
by A.P.T., Inc. and is called the "PxP" system (particle collec-
tion by particles).
The PxP system for fine particle control uses relatively
large particles as collection centers for the fine particles
in the gas stream. The relatively large particles (collector
particles) introduced to the gas stream can collect fine particles
by mechanisms such as diffusion, inertial impaction, interception
and electrophoresis. The larger size of the collector particles
AIR POLLUTION TECHNOLOGY, INC.
4901 MORENA BLVD..SUITE 402-SAN DIEGO, CA 92117
-------
- 308 -
allows easy separation from the gas stream by methods such as
cyclones, and gravitational settling.
Figure 11 is a functional diagram of the process steps for
a representative PxP system. The functions represented
on this diagram could occur concurrently or separately in several
types of equipment.
The first step involves introducing the collectors to the
gas stream. This process can involve pneumatic or mechanical
injection into the gas stream. The second stage involves contac-
ting the collectors with the gas in order to encourage the move-
ment of the fine particles to the collectors. A venturi device
can be used for the contactor which would be analogous to a ven-
turi scrubber except that solid collectors are used instead of
liquid drops. Alternative contactors such as a centrifugal scrub-
ber could be used.
The next process step is to remove the collector particles
after sufficient exposure in the contactor to cause capture of
the initial fine particles present in the gas. At this stage the
large size and mass of the collector particles is utilized to
separate them from the gas. A cyclone separator could be used
for this step. Two streams are shown leaving the separator: the
cleaned gas leaves the process at this point and the second stream
represents the flow of collector particles to the next step. The
final process involves either discarding the collector particles
or cleaning them for recycle and disposing of the material collec-
ted from the gas stream.
Performance Prediction
The particle collection efficiency and pressure drop for an
A.P.T. dry scrubber with cocurrent flow can be predicted with the
same relationships that define cocurrent wet scrubber performance.
The theoretical performance of the PxP scrubber has been determined
based on the venturi scrubber model of Yung et al. (1977). For
particle collection in the venturi throat, the penetration for
a given particle size is:
AIR POLLUTION TECHNOLOGY, INC. 4901 MORENABLVD.,SUITE402-SAN DIEGO,CA92117
-------
- 309 -
Pt = exp
B
K
4 K
po
- 5.02
0.7
KPO
4 K +4.2
po
d 2 u.
where K
po
= 2 I 1 - x2 + (x* -
o -
16
/QC\/PC\ !
and B ={ ~£ ) ( -£ ) 7^-
\QG/\PG/CDO
°'5
0.5
1.5
(12)
(13)
(14)
(15)
(16)
where u
Gt
t
gas velocity at throat, cm/s
throat length, cm
collector density, g/cm3
gas volumetric flow, m3/s
collector volumetric flow, m3/s
drag coefficient for drops at the venturi throat
inlet, dimensionless
AIR POLLUTION TECHNOLOGY, INC.
4901 MORENA BLVD., SUITE 402-SAN DIEGO, CA 92117
-------
- 310 -
Particle collection efficiency was predicted for several
values of parameters in a cocurrent PxP scrubber using 100 ym
diameter collectors and a gas velocity of 57 m/s. Figure 12 is
a plot of particle penetration against particle size with collec-
tor/gas flow rate ratio as a parameter and with a 20°C gas tem-
perature. Figure 13 is a similar plot with an 820°C gas tempera-
ture. To show the effect of temperature on penetration, the
curves for a ratio of 0.002 and temperatures of 20°C and 820°C
are plotted on Figure 14.
The predicted penetration curves have the following charac-
teristics :
1. For a given set of operating conditions, the penetration
decreases with increasing size of fine particles. This is expec-
ted since the collection mechanism is inertial impaction of the
fine particles upon the collectors.
2. For a given size of collector particle and aerodynamic
diameter of fine particle, the penetration decreases with in-
creasing value of (Q Pc/Qp)•
3. A similar dependence upon the gas velocity is apparent
from equation (1).
4. For the 100 ym collectors and a given fine particle
aerodynamic diameter, the penetration increases with increasing
gas temperature. This is the result of an increase in gas vis-
cosity with temperature which reduces the effective inertia of
the fine particles.
It can also be shown that collector particle diameter affects
collection efficiency when other factors are held constant. The
cut diameter (i.e., the diameter of the particle which is collec-
ted at 50% efficiency) decreases as collector diameter decreases.
Collection efficiency for particles larger than several microns
diameter varies in a more complex way, depending on flow and
geometric parameter combinations.
Experimental Program
Experimental work has been performed by A.P.T. to determine
fine particle collection efficiency in a PxP scrubber in order to
AIR POLLUTION TECHNOLOGY, INC. 4901 MORENA BLVD., SUITE 402-SAN DIEGO, CA92117
-------
- 311 -
confirm the predictions obtained from available mathematical models.
A. dibutylphthalate (DBF) aerosol was used in collection efficiency
experiments with 125 ym mean diameter nickel beads and with 100 ym
mean diameter sand as collector particles. The DBF aerosol had
a mass median aerodynamic diameter of 1.3 ymA and standard deviation,
o = 2.0.
c>
The collectors entered the T-shaped contactor through the
branch leg and were entrained by air entering through one of the
"run" legs. The length of the 1.1 cm diameter throat varied from
2.5 to 5.1 cm. The throat velocity was 57 m/s and (Q p /QrJ was
around 0.005 g/cm3 for the nickel beads and 0.0017 g/cm3 for the
sand.
Test aerosol particle cumulative concentration was measured
for each of several diameter increments by means of a Climet light
scattering particle analyzer for the experiments with nickel
collectors. Cascade impactors were used with the sand collectors.
Inlet and outlet cumulative mass distributions were plotted and
the particle collection efficiency was computed from the ratio of
the curve slopes at several particle diameters.
The resulting penetration data are shown in Figure 15 for DBF
collection on sand. The cascade impactor data led to the penetra-
tion relationship labeled "experimental curve." The prediction for
(Q Pc/Qr) = 0.002 is also shown in Figure 15 and compares well
the with experimental curve.
Particle penetration data for all runs with nickel and sand
collectors are represented in Figure 16, a "cut power plot." The
cut diameter is plotted against gas pressure drop in Figure 16.
The line represents the relationship which is predicted and which
has been confirmed by a number of field tests on large wet scrub-
bers. Agreement between the data points and the line is good.
The experimental data on the primary collection efficiency
of the PxP system agree well with predictions based on a mathema-
tical model which was first developed for wet scrubbers. Since
the model was derived for the mechanism of particle collection
by inertial impaction on spheres in a cocurrent scrubber, it is
AIR POLLUTION TECHNOLOGY, INC. 4901 MORENA BLVD., SUITE 402-SAN DIEGO, CA 92117
-------
- 312 -
reasonable to expect it to fit the data. The PxP A.P.T. dry
scrubber system has the same primary collection efficiency/power
relationship as a venturi type wet scrubber.
The overall efficiency of the PxP system will depend on
the reentrainment characteristics of the specific system in addi-
tion to the primary efficiency. Particle and collector proper-
ties, system geometry, flow rate, and other parameters will in-
fluence reentrainment.
Research is continuing on the experimental evaluation of the
PxP system for HTP application.
Table I
Granular Bed Filter Problems
FIXED BED
1. Plugging of retaining
grids and possible par-
ticle buildup in bed.
2. Particle seepage through
bed during cleaning
cycles.
3. Fluidization redisperses
fine dust during clean-
ing.
4. HTP valving required
for reverse air
cleaning.
5. Temperature losses
proportional to vol-
ume of cleaning air.
MOVING BED
1. Particle re-entrainment 1.
in moving bed.
2. Granule recirculation 2.
may cause high opera-
ting cost.
3. Difficult to form a 3.
cake in moving bed.
4. Erosion of retaining 4.
grids and transport
systems.
5. Temperature losses 5.
proportional to heat
capacity of recircu-
lated granules and
recirculation rate.
INTERMITTENT BED
Low gas capacity
can cause high
capital cost.
Granule recircu-
lation may cause
high operating
cost.
Need to form sur-
face cake to avoid
plugging problems.
Erosion of retain-
ing grids and
transport system.
Temperature losses
proportional to
heat capacity of
recirculated
granules and re-
circulation rate.
AIR POLLUTION TECHNOLOGY, INC.
4901 MORENA BLVD., SUITE 402- SAN DIEGO, CA 92117
-------
LIST OF SYMBOLS
C' - Cunningham slip factor, dimensionless
"Do
drag coefficient at throat inlet, dimensionless
d » granule diameter, cm
d. • jet diameter, era
dn * particle diameter, cm
aerodynamic diameter a d (p C1)
vmA
E - overall collection efficiency, fraction
f • friction factor, dimcnsionless
f(d } • particle size frequency distribution
K D inertial impaction parameter, dimensionless
lt - throat length, cm
N - number of impaction stages
NJ, * Reynolds number, dimensionless
PT • overall penetration, fraction
Pt, - penetration for particles with diameter, d , fraction
Q » collector volumetric flow rate, m'/s
Q- • gas volumetric flow rate, m'/s
r., « hydraulic radius, cm
u., - superficial gas velocity, cm/s
average interstitial gas velocity, cm/s
" gas throat velocity, cm/s
JGi
JGt
u. - jet velocity, cm/s
Z " bed depth, cm
c • bed porosity, fraction
n " single stage collection efficiency, fraction
Vj. " gas viscosity, poise
PC • collector density, g/cm1
p_ « gas density, g/cm'
C - particle density, g/cm'
LIST OF REFERENCES
Ergun, S., "Fluid Flow Through Packed Columns,' Chem. Eng. Prog.,
48: 89-94, 1952.
Hoke, R.C., et al., "A Regenerative Limestone Process for Fluidized
Bed Coal Combustion and Desulfurization," Monthly Report 87,
1977a.
Hoke, R.C., "Ducon Gravel Bed Filter Testing, presented at EPA/
E-RDA Symposium on High Temperature and Pressure Particulate
Control, Washington, D.C. September 1977b.
Hoocl, K.T., "Evaluation of the Combustion Power Company Moving
Gravel Bed Dry Scrubber on the Control of Particulate
Emissions from a Hog-Fired Boiler," NCASI Special Report,
September 1976.
Knet.tig, F. and J.M. Becckmans, "Caputre of Monodispersed Aero-
sol Particles in a Fixed and in a Fluidized Bed," Canadian
J. of Chem. ENg., 52: 703-706, 1974.
Marple, V., "The Fundamental Study of Inertial Impactors.'
Ph.D. thesis, University of Minnesota, 1970.
McCain, J.D., "Evaluation of Rcxnord Gravel Bed Filter, EPA
600/2-76-1&4, NT1S PB 225-095, June 1976.
Mercer, T.T. nnd R.G. Stafford, "Inpsction from Round Jets,
Ann. Occupational Hygiene, 12: 41-48, 1969.
ParctsXy, L., L. Theodore, R. Pfiffcr, and A.M. Squires, "Panel
Bed Filters for Simultaneous Removal of Fly Ash and Sulfur
Dioxide: II Filtration of Diluted Aerosol by Sand Beds,"
J. APCA, 21_: 204-209, 1971.
Rani, W.E. and J.B. IVong, "Inpaction of Dust and Smoke Particles,"
Ind. Eng. Chem. 4J_: 1371-1381, 1952.
Stern, A.C., H.W. Zeller, and A.I. Schekman, "Collection Effi-
ciency of Jet Impactors at Reduced Pressures," Ind. and
Eng. Fundamentals, 1: 273, 1962.
Yung, S.C., S. Calvert, and H.F. Barbarika, "Vcnturi Scrubber
Performance Model,-" EPA 600/2-77-172, NTIS PB 271-515/AS,
August, 1977.
I
U>
f-1
CO
I
AIR POLLUTION TECHNOLOGY. INC.
1901 MOHENA BLVD.. SUITE 402- SAN DIEGO. CAS2I17
AIR POLLUTION TECHNOLOGY. INC.
4DO' MOHE'IA BLVD.. SUITE JO?-SAN DIEGO. CA92T7
-------
AIR
I
COMPRESSOR
GAS TURBINE
BOILER
FEED'
•STACK
TERTIARY HEAT
COLLECTOR RECOVERY
CYCLONE
•STEAM
•WATER
Figure 1. Pressurized fluidized bed boiler
50
I 10
I 5
1.0
0.5
1 1 1 1 1 1 r
j i i i i i i \
10 30 50
Wt % undersize
70
90
Figure 2. Particle size distribution from Exxon miniplant
i
UJ
AIR POLLUTION TECHNOLOGY, INC,
SAN DIEGO, CALIFORNIA
-------
10- -
- BED
DHPTII 'C? /
- t--2 ^ / 9
9.2 cm
6.2 cm
3.2 c
0 10 20
40 60
u_, cm/s
b
30
100
Figure 3. Experimental and predicted pressure drops
across granular bed consisting of iron shot.
-
9s9 Q'°
TARGET
Figure 4. Impaction model
AIR POLLUTION TECHNOLOGY, INC,
SAN DIEGO, CALIFORNIA
00
I-1
en
I
-------
O 1.09/z
A 0.76/i
O 0.5/i
0.01
.« 0.001
0.0001
i i i
0.002
0.01
Kr
Figure 5. Experimental data
0.1 0.2
1.0
0.5 -
1.0
- o.os :
cn
CO
0.01
.005
0.001
i—i i i i i r
STERN
MERCER &
STAFFORD
, . l\llt I I IU dllU
20-30 MESH /
-------
1.0
0.1
01
O_
MCCAIN'S
DATA
I 1 1 1 I I I
0.01 I i i i i i.
0.5 1.0 5 10
Particle dia,
Figure 7. Comparison with data for
Rexnord filter
PREDICTION
Pp=2g/cm3 I
0.5 1
Particle dia, /j,m
Figure 8. Comparison with data for
CPC moving bed filter
i
u>
AIR POLLUTION TECHNOLOGY, INC,
SAN DIEGO, CALIFORNIA
-------
0.4 1 5
Particle dia,/zmA
Figure 9. Comparison with CPC
moving bed filter
1.0
0.5
0.1
0.05
0.01
0.1
lOatm
20°C
I I III
nI I I I I i TT
Granule diameter:400^i'
Bed Depth:3.8cm ;
UGr4.5cm/s
p = 1.5 g/crn^
e = 0.25
lOatm
870°C
1 atm
870°C
J 1 i
.5 1.0 5
PARTICLE Di AMETER, pm
Figure 10: Predicted GBF Performance
10
00
I
AIR POLLUTION TECHNOLOGY, INC,
SAN DIEGO, CALIFORNIA
-------
1.0
CONTACTOR
DUSTY GAS
COLLECTOR
PARTICLES
SEPARATOR
COLLECTOR
PARTICLES
CLEAN GAS
DISCARD
DISCARD
Figure 11. Schematic diagram of A.P.T. dry scrubber system.
0.1
o
£
0.01
0.001
0.1
AERODYNAMIC DIAMETER, ynA
I
CO
10
Figure 12. Theoretical particle collection characteristics of
the A.P.T. dry scrubber.
AIR POLLUTION TECHNOLOGY, INC,
SAN DIEGO, CALIFORNIA
-------
1.0
0.1
1-
PJ
Z
0.01
o.noi
i.o
o.i
AERODYNAMIC DIAMETER, ymA
Figure 13 Comparison of the particle collection characteristics
of the A.P.T. dry scrubber at 20°C and 820°C.
0.1
2
O
0.01
0.001
Co
to
O
0 .1
AERODYNAMIC DIAMETER, pmA
Figun; 14. Comparison of the particle collection characteristics
of the A.P.T. dry scrubber at 20°C and SZO°C.
AIR POLLUTION TECHNOLOGY, INC.
SAN DIEGO, CALIFORNIA
-------
- 0.1
0.01
THEORETICAL
EXPERIMENTAL
\
\
0.1
0.5 1.0 2.0
AERODYNAMIC DIAMETER, umA
5.0
Ficure 15 Comparison of experimental with theoretical particle
collection characteristics of the A.P.T. dry scrubber.
3.0
2.0
S 1.0
0.7
0.5
0.3
0.2
0.1
I 1
Q HORIZONTAL FLOW
D VERTICAL FLOW
• THEORETICAL
1
D
I
0.1 0.20.3 O.S0.7 1 2 3 5 7 10
GAS PHASE PRESSURE DROP, Kpa
Figure 16. Comparison of particle collection charncteristics of the A.P.T. dry
scrubber with the A.P.T. cut/power relationship.
I
OJ
AIR POLLUTION TECHNOLOGY, INC,
SAN DIEGO, CALIFORNIA
-------
- 322 -
Discussion
Mr. Cooper mentioned that Acurex had used the same facility but with
different equipment and that similar data had been obtained. He wondered
which type of equipment had been used in the experiments described by
Dr. Parker. The measurements were made at Exxon where a continuous
sampling system was used.
Mr. Princiotta asked whether the equipment was capable of cleaning
gases, sufficiently to protect the turbine or meet the environmental
standards (10-60 mg/m ) and whether it was feasible for these devices to
produce sufficient particulate removal. Dr. Parker confirmed the latter
point, stating that his talk had been based upon the mechanisms of
inertia! impaction, which was a feasible although expensive process.
Electrostatic augmentation or cake formation on granular bed filters
could potentially reduce the expense.
Mr. Wiggers asked for details of the material used in granular filter
beds and proposals for recycling process. Alumina, sand, or any substance
which could withstand high temperatures was used (Dr. Parker). Pneumatic
recirculation took place; particles were fluidized to the top of the bed
and were allowed to fall slowly. This achieved more efficient cleaning
but was more expensive than a fixed bed. Mr. Wiggers asked whether
abrasion occurred; Dr. Parker answered that the moving bed resulted in
the attrition of particles and the degradation of the granules themselves.
Mr. Guthner inquired as to the type of dust for dry scrubbers, what
aerosols were used, and whether Dr. Parker could give details of the
diameter of scrubber particles. The diameter referred to in the talk
was the physical diameter of the collectors, and mono-dispersed aerosol
particles were used. The type of collectors used were nickel powder and
sand, and an oil aerosol was employed, which eliminated particle bouncing.
The venturi scrubber model was validated for the dry scrubber-; flyash
tests were also carried out with the same results, although attrition
problems occurred. The flyash was used as an aerosol, rather than a
scrubber collector.
-------
- 323 -
Mr. Finkh requested details of design limits for granular bed filters,
particularly maximum pressure in relation to gas turbines. In practice,
Dr. Parker answered, the face velocity for granular filters was 40-70 cm/sec
The size needed for a given flow rate could be determined from the velocity.
-------
- 324 -
APPLICATION AND EFFICIENCY OF DRY ELECTROSTATIC PRECIPITATORS
Prof. Dr.-Ing. E. WEBER
Institut fur Mechanische Verfahrenstechnik
Universitat Essen GHS
UniversitatsstraBe 2
Postfach 68 43
D 4300 Essen
Tel. 0201-1832795
Dipl.-Ing. H.-G. PAPE
Institut fur Mechanische Verfahrenstechnik
Universitat Essen GHS
UniversitatsstraBe 2
Postfach 68 43
D 430O Essen
Tel. 0201-1832792
-------
APPLICATION AND EFFICIENCY OF DRY ELECTROSTATIC PRESIPITATORS
H.-G. Pape, E. Weber
One of the main problems introducing new technologies of power
generation is gas cleaning without enregy dissipation. For eco-
nomical use of fluid bed combustion, and other gas steam processes
gas cleaning without great loss of pressure and heat is necessary -
In case of the so called fixed bed gasification for example, the
formed lean gas has a temperature of 60O °C and a pressure up to
20 bars. Expansion of that gas by a gas turbine is only possible,
if dust and pollution gas are removed.
The gas purification has apart from the necessary desulfurisation
to fullfill the following conditions:
Precipitation of the solid components down to dust contents of
less than 5 mg/m gas and
dust separation at gas temperatures up to 1OOO °C and pressures
up to 40 bars.
In addition to fabric filters and wet scrubbers it is possible to
use an electrostatic precipitator to reach high degrees of separation,
Apart from construction problems, the application of electrostatic
precipitators at high pressures and high temperatures is only of
interest, if ihe physical effects of precipitation are not
disadvantageously affected. As a consequence the migration velocity,
defined by DEUTSCH, would become to small.
The main factors of precipitation are
1) the current-voltage-characteristics,
2) the corona-starting voltage,
3) the sparkover voltage
4) the electrical resistivity of the dust.
-------
- 327 -
Up to now electrostatic precipitators are used successfully up to
temperatures of 35O C and pressures up to 3 bars. Publications
about high temperature and high pressure electrostatic precipitation
are rare. Tests in this line are described sporadically.
Essentially it is written about the relations of voltage, pressure,
and temperature, without studying the effectiveness of electrostatic
precipitation and the influence of dust resistivity and distance
between discharge electrode and collecting electrode.
The studies show in accordance, that electrostatic precipitation
at higher temperatures and pressures is possible.
With rising temperatures sparkover voltage and corona-starting
voltage decrease, whereas the sparkover voltage decreases more,
and the voltages reach the same level. An increase of pressure
compensates this decrease, and the difference between sparkover
voltage and corona, starting voltage again reaches a value,
giving guarantee for a steady filtering process.
From very simplified assumptions the theoretical migration velocity
which is at least qualitatively correlated with the effective
migration velocity, can be deduced.
Starting from the equation of motion
mass force = electrical force - power of resistance
it is to obtain
The driving electrical force is
r = ? • EP
with E as collecting field intensity
-------
- 328 -
The power of resistance is to calculate by Stokes law
-^ £ rr
Estimating the time, a single particle needs to reach the final
velocity, it is demonstrated, that this time is very short.
So the process of acceleration is to neglect f9r the calculation
of the theoretical migration velocity.
w
or
"** t> tf Y£ < ?>
Assuming, that the particel gets the maximum charge, from the
derivation of field charging process the factor q is to figure out:
// r— <2- E = charging field intensity
n = number of particle charge
o
e = electronic charge
The multiplier K depends upon the dielectric constant of the dust.
K is equal to 3 for conductiv particles. So theoretical migration
velocity is: ,_ ._
As field intensity is proportional to voltage, the theoretical
migration velocity among other relations is proportional to the
square value of the voltage and inversely proportional to gas
viscosity. Attainable voltage, that means sparkover voltage, and
gas viscosity decrease by rising temperature. The influence of
pressure on gas viscosity is extremly small. But the attainable
voltage increases with rising pressure, as given by H.E. Rose
and A.J. Wood.
-------
- 329 -
A modification of Peek's equation for corona starting voltage
leads to
K-, K2 are fixed by geometry.
£ is the relativ air density.
If this equations hold true, and if in tendency sparkover
voltage is behaving like corona starting voltage, the migration
velocity might be greater at a pressure of 40 bars and a temperature
of 1000 °C than at standard conditions.
But these considerations up to now are not confirmed by experimental
investigations in case of very high pressures and temperatures.
It seems to be shure, that electrostatic precipitation at high
temperatures only is possible, if the pressure at same time is
increased.
Another influence on the characteristics of electrostatic precipitation
is due to the already mentioned resistivity of dust. As known the
resistivity for a good seperation should be between 1O and 10 _0- cm.
At high temperatures low dust resistivity, which has negativ influence
on separation and especially on adhesion of dust particles on the
collecting electrode, is expected. But an extrapolation of test results
made by Schiitz and Winkel ( Abb. 1 ) seems to disprove these expecta-
tions at least for converter dust. Due to these findings the dust
resistivity seems to be between 1O and 1O _fl cm at a temperature
of 1000 C. As this extrapolation is not fused by measurements, and
these results are restricted to converter dust, the precipitation
investigations have to include the determination of dust resistivity
at high temperatures and pressures.
By a research project promoted by the ministery of Research and
Technology is to examine now, if and under which conditions
electrostatic precipitators are to be used succesfully at high
temperatures and pressures.
A flow diagramm of the pilot plant, which is built at present,
shows picture 2. The gas is led by a compressor (2) about a
-------
- 330 -
pressure tank (4) and a pressure reducing valve (5) to the
precipitation tank (1). The compressor permitts a pressure up
to 35 bars. By heating the precipitator to maximum temperature
theoretically the pressure is to increase up to 4.6 fold value.
Unaffected by the flow rate the separation pressure is fixed up
by the pressure reducing valve. The gas is preheated before
reaching the precipitation tank and gas temperature is deter-
mined. Dust is blown in behind the heating by slight super pressure.
The gas is led off through a watercooled heat exchanger, a regulating
valve, permitting a stepless regulation of the flow rate, and a
gas-flow meter.
Because of the high pressure the precipitation tank is constructed
cylindrical. So the dust separation takes place in a one-tube-
electrostatic precipitator. It's intended, to use tubes of different
diameters, to determine the influence of the distanse of electrodes.
The tubes are heated by resistance wires. The temperature of the
wires, being more than 1OOO C is diminished by a heating insulal
between the wires and the wall of the precipitation tank.
Measuring the temperature is done by thermocouples.
The temperatures of head and bottom of the precipitating tube
and of the heating are measured.
The high voltage supply takes place at the top of the precipitation
tank. The electrical insulation consists of a pipe made of ceramic,
where the discharge electrode is inserted.
Because of the high pressure during the tests, very high spark
over voltages are expected. Therefore a transformator permitting
voltages up to 2OO kV is used for generating the high voltage.
Heating experiments and pressure tests took a satisfying course.
After a short time of heating in the precipitation region a temperature
of 1OOO C could bi hold constantly. An increase of gas pressure
up to 45 bars in the precipitation tank didn't show any important
leakage.
-------
- 331 -
After contacting all lines of measurement and vontrol the
precipitation experiments with various pressures and temperatures
can start.
-------
- 332 -
101L
^ow-
f~\J
v_^
X
^
(A
• «^
I/)
Q)
1
10-
J
o
/
o
ol
o
I
o
1
o
o
0
1
/
o
/
J
/
1
1
1
1 • I ' 1 ' 1 1 ' 1 ' 1
1000 800 600 WO 300 200 100
temperature T (
Mechanische Resistivity of Converter Dust
v erionr snsiecnn IH *
Universitat Essen (frOm Wl'tlkel l/HOf Schtitz)
^
50
0 C )
UA
11
-------
©
o
<=>
LH>O
precipitation tank
compressor
pressure meter
pressure tank
pressure reducing
valve
valve
el. heating
dust feeding
thermocouple
recorder
•{£>]—
-------
- 334 -
Discussion
Mr. Cooper said that it was desirable to remove tar in the coal
gasification process. The tar condensed into droplets, which were
collected by an electrostatic precipitator. The use of an electro-
static precipitator in coal gasification could perhaps be dangerous
(Explosion). Mr. Guthner added that O_ in the gas would cause an
explosion regardless of temperature. Mr. McCain drew the workshop's
attention to a non-explosive mixture used at the Morgantown Coal
Gas Plant, which had an electrostatic precipitator. Dr. Gooch asked
whether the dielectric strength of the dust could be measured at high
temperature, and was told by Dr. Parker that this could be tried;
the dielectric strength could be less under certain conditions.
Further research is necessary referred to theoretical migration
velocity -
Precise details concerning the measurements made during the expe-
riments could be obtained at a later date.
-------
- 335 -
RANGE OF USE FOR FILTERING DUST COLLECTORS
Prof. Dr.-Ing. E. WEBER
Institut fur Mechanische Verfahrenstechnik
Universitat Essen GHS
UniversitatsstraBe 2
Postfach 68 43
D 43OO Essen
Tel. 0201-1832795
Dipl.-Ing. R. SCHULZ
Institut fur Mechanische Verfahrenstechnik
Universitat Essen GHS
UniversitatsstraBe 2
Postfach 68 43
D 4300 Essen
Tel. 0201-1832788
-------
- 336 -
RANGE OF USE FOR FILTERING DUST COLLECTORS
R. Schulz, E. Weber
With today known filtration collectors it is possible to
carry out a gas cleaning only at gas temperatures up to
maximal 400 °C. In scope of new energy technologies and energy
saving there is an interest in realizing gas purification at
high temperatures of about 10OO C and high pressures up to 40 bar.
This should take place without large heat and pressure losses.
Among the known precipitators on principle only the centrifugal
separators are applicable for gas cleaning at very high temperatures.
But due to their mode of action even a highgrade cyclone cannot
reach low dust contents in purified gas, which are for instance
necessary for a successful gas turbine process.
At present state of technology very low dust contents in purified
gas can be achieved using filtering dust collectors. The particles
in the dust loaded gas stream are collected passing through a porous
layer of sufficient permeability.
As porous systems are known :
1 . different kinds of textile media
2. packed beds of pebbles or sand
3. sintered ceramics
Of these three systems the textile filters play the major roll in
the industrial application. Mainly fabrics and needle felts are
used of which the needle felts become more and more significant.
The advantage of the felts compared with the fabrics is based on
its tridimensionality. Its fine pore structure and its high pore
volume afford excellent collection grades at high air permeability
and low pressure loss.
-------
- 337 -
Fabrics consist of warp and weft. In order to reach a high
degree of separation the knit of the fabric has to be tighter.
This way the cloth permeability is low and the pressure loss
becomes high. Making the knit looser, small openings appear
which generate free passages for the dust. These passages
will not be closed partly or total before a filter cake is build up.
Whereas for conventional filters the load per filter area in
3 2
general lies between 8O and 15O m /m h, the needle felt filters
3 2
can be stressed up to loads of 180 m /m h. This way you can
reduce the filter area, save space and energy and finally
raise the economy of the filtration plant.
On the other hand filter felts with high collection ability
need a thorough cleaning. A tridimensional filter material easier
runs the risk of clogging with raising the pressure loss if there
is not also a cleaning in the depth of the felt. This can be
managed by a strong combined mechanical and pneumatic filter
cleaning.
As described above, neither fabric filters nor needle felt
filters can be used at gas temperatures above 4OO C. In picture 1a,
taken from a paper of H. Dietrich, you can see the heat resistance
of present used textile filter materials. The natural fibres wool
and cotton do not even resist at temperatures above the vaporization
point of water, while organic fibres can be used between 1OO and
280 °C. At present inorganic fibres can be applicated up to
temperatures of about 40O °C.
On principle also a filtering dust collection at higher temperatures
should be possible because fibres are know which can be employed
at gas temperatures above 1OOO °C ( fig. 16 ). These fibrous
materials can be divided into :
1- metallic fibres
2- graphite fibres
3« quartz fibres
4- ceramic fibres
-------
- 338 -
Asbestos fibres should not be used anymore, because asbestos
particles are respirable.
Metallic fibres made of inconel meet the requirements with
respect to temperature resistance up to 7OO C and mechanic
strength. However there is the danger of corrosion and scaling.
Another disadvantage are the high costs for metallic fibres of
about 80O DM/m2.
However leading producers of metallic fibres refer to the high
loading capacity of metallic fabrics in comparison to those of
meneral materials. It is aspired to get streaming velocities
up to 40 cm/s which is equivalent to a filter area loading of
32 32
about 14OO m /m h. Today only loadings of maximal 2OQ m /m h
are achieved with common textile filters. A high flow rate
would be desirable especially for high temperature filtration
because there are large gas volumina due to the high temperature.
Graphite fibres have a high temperature durability in reducing
atmosphere. In oxidizing atmosphere however they can be applied
only at temperatures up to 35O °C. Therefore these fibres are
inapplicable for most cases of high temperature filtration.
Quartz fibres and ceramic fibres, are stable at temperatures
above 10OO C. Nevertheless the mechanical resistance at high
temperatures and simultaneous mechanical strain is unknown
because the fibres are very brittle. Difficulties must be expected
especially at the fixing points of the filter clothes. Outside
strengthening at the border of the cloth could be achieved by
pasting. Useful adhesives, which are resistant up to temperatures
of 1700 °C are offered by several producers. Another possibility
is the inside armouring of the filter clothes.
For excample ceramic fibre felts are armoured by interlacing
metallic fibres. But it seems questionable if these armour
fibres will withstand the bending stress at high temperatures
-------
- 339 -
which occures during cleaning the filter clothes. At present
one searches for new spinning technologies in order to produce
a strong fibre web system also of the short stapled ceramic
fibres.
Contrary to filtration at high temperatures there are no problems
in fabric filtration at high gas pressure. Independent of the
gas pressure only the difference in pressure at the filter material
is important for the filtration process. In general this difference
amounts between 2OO and 4OOO Pa. The high gas pressure must be
noted only at the construction of the filter chamber.
At present high temperature fibres are offered on the market only
for insulation purposes. Talking with the producers it turned out
that there are multifarious kinds of high temperature resistant
fibres which are aldeady manufactured into fabrics or simular
fibre structures. We got already a number of samples.
An unsophisticated pilot plant was built for a first fundamental
study of the fabric samples. The schema of this plant is shown
in figure 2.
The dust gets over a metering hopper and a funnel into the air
stream which is exhausted by a fan at the end of the apparatus.
In the first filter cell the dust is collected on the fabric
filter. The residual dust content is collected on a paper filter
in a second filter cell and can be determined by a gravimetric
method. The gas flow rate is kept constant manually with the help
of a rotameter and a valve. The change in pressure drop with
increasing filter cake on the fabric is registered over a pressure
gauge and an amplifier on a chart recorder.
We have got two characteristic pressure drop curves as shown in
figure 3. The pressure course according to curve 2 is not suitable
for an economical gas cleaning because the pressure drop increases
-------
- 340 -
rapidly in the beginning of the filtering process when the dust
loading on the filter cloth is still small. This leads to a
higher initial pressure loss after the first filter cleaning.
Fortunately most of the investigated fabric samples showed a
behaviour according to curve 1.
Concerning dust collection we have got very good results for
all fabric samples. For test dust a quartz powder was used
with maximal grain size of 63 jam. The dust concentration of the
waste gas amounted to 6O g/m . The gas velocity was varied
between 3 and 7 cm/s, which corresponds to a load of filter
area between 1OO and 250 m /m h.
In no case degrees of separation below 99.97 % were noted. The
maximal dust load in clean gas was 7 mg/m . Since the used
quartz powder showed the trend to agglomerate, it is planned
to carry out further measurements with a hydrophobe quartz powder.
Then the next step is to investigate the fabrics under extreme
temperature conditions.
Until today the general knowledge about high temperature filtration
using fabrics or felts, which are made of the described fibres is
very low. With financial support of the Ministery for Research and
Technology the prototype of a high temperature fabric filter is
under construction as shown in figure 4. Three, by pairs parallel
filter bags consisting of fleeces, fabrics or mat weaves are used
to study the properties of several materials with respect to :
1 . Efficiency of dust separation
2. Durability
3. Cleaning
To generate a dust-laden waste gas a pulverized coal firing is
used which can be fed additionally with oil in order to obtain
higher temperatures. In continious process waste gas temperatures
-------
- 341
between 400 and 7OO °C can be achieved with this combustion
furnace. Single tests can be carried out also at gas temperatures
up to 1000 °C.
The dust load of the furnace gas will be about 1 g/m . If it
is necessary a definite increase or a variation of the dust
load can be obtained by an additional dust feed. A regulation
of the waste gas composition is possible by a variation of the
oil rate. The cleaning of the filter bags can be done either
by manual knocking, by magnetic generated vibrations or by
pneumatic methods.
It can be summarized that due to the quality of the material
samples and the results of the first preinvestigations fabric
filters should be applicable at low costs for the high temperature
gas purification. Since our investigations have started only a few
months ago it should be possible to achieve more detailed results
in the future.
-------
- 342 -
Wool 70°C
Cotton 70°C
Rayon Staple 70 C
Polypropylene 100 C
Polyacrylonitrile 130 C
Polyester 150 C
aliphatic Polyamid (Nylon) 110 - 120°C
aromatic Polyamid (Nomex) 180 - 200 C
Teflon 260 - 280°C
Glass Fibre 300°C
Mineral Fibre 300 - 350°C
Metallic Fibre AOO°C
Heat Resistance of Textile Filtermedia
Fig.; 1a
-------
- 343 -
ifiulerial
glass
chrysolite
asbestos
'jrnptiibolite-
jsbesios
quartz
basalt
ceramics
stag
rock
graphite
ixidizinq
itmosphere
reducing
Time/sphere
metal
tktnpc-rutui c -
resisting up to
( °C )
550
550
800
1200
1000
1200
700
700
400
>2000
manufactured
into
fabric
fleece
fabric
fleece
fabric
fleece
fabric
fleece
fabric
f leece
fleece
f leece
fleece
.
r
700
fleece
fabric
felt
Icv/rst dia-
metc.r of
the fibres
f/irn)
2-5
2 -10
2 10
2- 5
10-20
/< 20
5 20
15-30
8 15
4-10
chief (.or.stitucnts
Si02 A1203
CaO
B203
V96l(OH)8Sl(tOw]
Ca.Na.K.Mg.Fa.Al.
MnJi.Si,F
Si02
Si02,At203, CaO,
MgO,FeO
Si02,AI203
SiO^,203.C,0.
MgO
Si02lAl203lCaO,
MgO
C.Cr.Ni. Cu,Mo,
Fe
l-ibrous Materials for the High
Tcmnerature Filtration
Fig.: 1b
-------
metering hopper
filter cell
gas meter
Rotameter
valve
fan
u-tube manometer
pressure gauge
amplifier
recorder
I
U)
Experimental Arrangement for Fa brie Dedusting at Normal Tempertures
Fig. : 2
-------
Ul
I
Pressure Drop-Time Characteric for High Temperature Fibre Fabrics
Fig.: 3 :
-------
additional dust feeding
coal dust gas analysis
cooler
oil
combustion chamber
high temperature
test filter
filterfnormal
remperatur)
chimney
air heater
I
U)
Scheme of the Pilot-Plant for Fabric Dedusimg at High Temperatures
Fig.: 4
-------
- 347 -
Discussion
Mr. Princiotta asked for the limitations on gas inlet temperature for a
gas turbine. Dr. Holighaus replied that the present temperature limitation
is 1000°C. New developments in gas turbine technology will increase the
gas inlet temperature to 1250°C.
Mr. Donovan asked if stainless steel fabrics were included in the category
"metallic fabrics", and whether all observations made regarding these
metallic fabrics were also true for stainless steel fabrics. Mr. Schulz
said yes, including the major disadvantage of metallic fabrics—high
cost, which means that their use could only be justified ;n very special
filtration application. Prof. Weber added that they were very good at
high face velocity (up to 40 cm/sec). Corrosion could occur in metallic
fabrics, said Mr. Shackleton. Collection efficiency necessitated a small
weave which resulted in a restricted fabric life.
Dr. GUthner spoke of glass (mineral) fibers, whose life was restricted
owing to brittleness. More emphasis should be placed on the fiber
finish (rather than the fiber itself) since pulses and other mechanical
stresses cause rapid wear.
Glass fibers have a low resistivity to mechanical stress. Mr. Princiotta
pointed out that quick heating and cooling could affect this material;
silicone was resistant to such sudden changes. Mr. Shackleton said that
alumina fibers had been tested at 1800°C and pulsed with unheated air;
no problems resulted. In addition, experiments had been performed which
showed that fibers did not break even when shocked, said Mr. Wiggers.
-------
- 348 -
HIGH-TEMPERATURE FILTRATION
Dr. Dennis C. Drehmel
U.S. Environmental Protection Agency
Research Triangle Park, N.C. 27711
Michael A. Shackleton
Acurex Corporation
Mountain View, Calif. 94042
Abstract
Research on high temperature particle control using ceramic
fiber barrier filtration has shown this technique offers
promise of successful development. Results of testing of rigid
ceramic membrane structures and of ceramic fiber beds including
woven, paper and felt ceramic filters are presented.
-------
- 349 -
HIGH-TEMPERATURE FILTRATION
Introduction
Removal of particles from high temperature gas streams
has been studied for many years. Some of the motivation for
this research was the desire to operate coal fired gas tur-
bines.1 Recently there is renewed interest in the utilization
of coal. The processes most actively being studied are pres-
surized fluid bed combustion (PFBC) and gasification combined
cycle (GCC) plants. Both processes are called combined cycle
since they generate power by means of gas turbines as well as
steam turbines.
In the PFBC, coal is burned in a fluid bed of limestone
(which removes the SO2) and heat is transferred to tubes in the
fluid bed. Up to 80% of the recoverable heat value of the coal
is removed in the fluid bed, and the gas exits at 1500°F and
10 atm pressure. The gas must now be expanded through a gas
turbine to recover the remaining energy. However, previous
investigations showed that large particles erode turbine blades
and small particles cause deposits that choke the turbine. To
protect the turbine, some high-temperature particulate control
is required. Moreover, since it would not be economical to
duplicate particulate control for environmental regulations at
another point of the process, high-temperature control must
also meet new source performance environmental standards for
coal-fired utilities. Currently, this allows emissions to be
no greater than 0.1 #/million Btu.
To meet both the environmental and turbine requirements,
a system consisting of two cyclones and a filter is being
studied. The two cyclones lower the overall particle concen-
tration but fail to remove small particles. Concentrations
leaving the second cyclone can be as high as 1.0 grains/scf
and have a mass median diameter of 5.0 ym. The filter can be
a ceramic bag filter, a ceramic membrane, or a granular bed.
-------
- 350 -
Research to date has concentrated on the granular bed since it
has been considered available technology. However, results of
tests at the Exxon PFBC Miniplant have been disappointing.2
The granular beds tested could barely meet the environmental
standard at the beginning of a run and lost efficiency contin-
uously from 95% to as low as 50% within 24 hours. Although
granular bed filters may still prove to be a solution to high-
temperature particle control, it is now apparent that they will
require more developmental work.
Alternative high-temperature filters, using either a
ceramic bag or ceramic membrane, are being developed. The
remainder of this paper will be devoted to describing ongoing
work by the Environmental Protection Agency to assess these
high-temperature filters as part of environmental control for
the PFBC, although it is expected that results can be extrapo-
lated to the GCC or to high-temperature metallurgical operations.
Ceramic Membrane Filters
Several ceramic materials in many configurations were
screened as possible high-temperature filters. One of the
most promising materials tested was a ceramic cross flow
monolith produced by 3M Company under the trade name of Therma-
Comb. This material is composed of alternate layers of corru-
gations separated by thin filtering barriers. This type of
configuration affords a large amount of filter surface in a
very small volume. Figure 1 shows a piece of this material
illustrating the construction.
Bench side experiments were conducted in the high-
temperature ceramic test facility at 970°K. Provisions were
made to blow back from the clean side and also down the chan-
nels on the dirty side so that various cleaning schemes could
be investigated. A sequencer was designed to automatically
start and operate the cleaning cycle. A 17 cm diameter by
38 cm deep tubular furnace was used to heat the filter. An
additional furnace was added to preheat the dust-laden air.
-------
351 -
Figure 1. 3-M Company Thermacomb.
-------
- 352 -
Cascade impactors were used to determine the size dis-
tribution of the test dust (limestone). The typical mass median
diameter was 1.4 ym and the geometric deviation was 3.0 urn. There
was some difficulty in maintaining constant feed rate, but dust
loadings were maintained at levels from 2 gm/m to 7 gm/m .
Typical results for filtering the limestone test dust
with the 3M ThermaComb are summarized in Table 1. Table 2
shows the effect of varying the initial pressure of a 0.6-
second pulse. Table 2 also shows the result of a similar set
of runs except that the pulse time was increased from 0.6
seconds to 5 seconds. These data show that the length of the
pulse does not have much effect on the cleaning results. In
both runs the collection efficiency was very high (99.6 to
100%) at a linear velocity of 0.41 m/min (1.33 ft/min). Using
the 103.4 kPa pressure pulse for cleaning, it was possible to
return to a stable pressure drop across the filter in spite
of the relatively high dust loadings which in these two runs
were 2.6 and 3.75 gm/m .
Tests using the 3M ThermaComb as a filtering media
showed filtering efficiency to be close to 100% even though
the test dust had a mass median diameter of 1.4 ym and a sig-
nificant fraction of sub micron material. Cleanability of the
media was verified in experiments evaluating the effect of
cleaning pulse intensity and duration. It was determined
that the ceramic filter behaved similarly to fabric filters
in that the pressure drop could be attributed to a residual
pressure drop and that across an incompressible cake.
Ceramic Fiber Barrier Filters
Ceramic fibers are produced by several manufacturers.
In general, these materials are sold for refractory insulation
applications. Many of these ceramic fiber materials are pro-
duced in smaller fiber diameters (3 pm) than are generally
-------
- 353 -
Table 1. Summary of 3M ThermaComb Performance
Flow rate
m /min
Filter area
2
m
Inlet/Concentration
gm/m
Temperature
°K
Efficiency
Percent
Average Range Tested
0.095 0.04 - 0.16
0.0227
3.6
990
96.6
2.2 - 5.4
953 - 1088
85 - 99.6
Table 2. Effect of Changing Cleaning Conditions
Test
Number
1
1
1
2
2
Pulse
Pressure
KPa
34.5
69.0
103.4
103.4
103.4
Pulse
Duration
seconds
0.6
0.6
0.6
0.6
5
Cycle
Time*
minutes
2-4
8
12 - 20
12
12
Residual Pressure
KPa
3.5 increasing to
3.0 increasing to
2.8
2.8
2.8
Drop
limit
limit
*as required to lower pressure drop away from the upper limit
available for filtration applications at room temperature (20 urn)
This smaller fiber diameter, coupled with high temperature and
corrosion resistance characteristics, makes these fibers
intriguing candidates for high-temperature filtration
applications.
-------
- 354 -
Available ceramic liber coniiguraLionu can be cluauJiiod
into the following three groups of materials:
• Woven structures - cloth woven from long-filament
yarns of ceramic fibers
• Papers — Ceramic structures produced from short
lengths of fibers, generally held together with
binders.
• Felts — Structures produced to form mats of relatively
long fibers. These materials are known as blankets in
the insulation industry. They tend to be less tightly
packed than conventional felt materials.
Theory
Filtration theory supports the contention that ceramic
fiber filters should perform adequately at high temperatures
and pressures.
There are three particle collection mechanisms generally
considered to account for the performance of a bed of fibers in
removing particles from gas streams. These mechanisms are:
direct interception, diffusion, and inertial impaction. Exam-
ining these mechanisms under high temperature and high pressure
(800°C and 10 atm) indicates that direct interception and dif-
fusion will be roughly the same as their performance at room
temperatures and ambient pressures, while the inertial impac-
tion mechanism will be slightly less effective at high temper-
atures and pressures. This statement is true when comparing
the performance of a clean filter bed (no dust cake) at low
temperature/pressure and at high temperature/pressure. This
relatively minor performance reduction can be compensated for
in the design of the filter media. For example, using smaller
diameter fibers in the filter bed can increase collection
efficiency far more than the viscosity effect of high temper-
ature reduces it. A fiber bed consisting of 3 ym diameter
-------
- 355 -
fibers, as compared with 20 ym fibers, can be expected to pro-
vide equal collection efficiency with a 3 ym fiber bed weighing
only one tenth as much on a weight per unit area basis. Figure
2 presents a calculated prediction of collection efficiency
for a 3.0 ym fiber bed, consisting of alumina fibers, collect-
ing a 0.5 ym fly ash particle from an air stream at 815°C and
10 atm pressure. The four curves show that for a decrease in
solidity* (a) a reduction in efficiency, for a given basis
weight, should be expected. Similarly, an increase in airflow
velocity causes a small reduction in collection efficiency for
a given filter bed (constant basis weight). Note also that by
adding fibers (increasing basis weight), all of these effects
can be nullified.
The magnitude of predicted efficiency is also interest-
ing. This analysis shows that for a ceramic fiber bed consist-
2
ing of 3.0 ym alumina fibers a basis weight of 500-600 g/m
should provide 80 to 90 percent collection of a 0.5 ym particle
even at the reduced performance levels encountered at high
temperatures and pressures.
For comparison, a standard filter media consisting of
2 2
20 nm fibers and having a basis weight of 540 g/m (16 oz/yd )
can be expected to collect only about 20 percent of 0.5 ym
particles at room ambient conditions. Thus, to provide collec-
tion efficiency performance equal to a conventional filter
requires only about one-tenth the weight of fibers for a 3.0
ym fiber bed. Or, put another way, a ceramic fiber bed of
equal media weight to a conventional filter, but made from 3.0
ym fibers, will be much more efficient even at high tempera-
tures and pressures than is normally sufficient in the filtra-
tion industry.
*Solidity (a) is the fraction of the fiber bed which is solid.
A solidity of 0.02 indicates that 2% of the bed is occupied
by fibers.
-------
- 356 -
3.0M m DIA FIBERS
2.8 g/cm3 FIBER DENSITY
0.5/um DIA PARTICLE
1.5 g/cm3
815°C
10 ATM
90
I 80
O
LU
O
tl
tij 70
z
g
LU 60
o
o
50
40
20
(5 FT/MIN) "\
2.54 cm/secU
(25 FT/MIN)
12.7 cm/sec
00
I
<
16 oz/ydz
I
I
I
100 200 300
400 500 600 700 800
BASIS WEIGHT — g/M*
900 1000 1100 1200
Figure 2. Calculated performs .ce 3.0 pm alumina fiber bed.
-------
- 357 -
Room Ambient Filter Media Tests
A large number of ceramic fiber filter media candidates
have been subjected to a series of filtration tests at room
ambient conditions. These tests included some examples of con-
ventional filter media for comparison. Included among the
tests were:
• Dioctylphtalate smoke (D.O.P) penetration as a
function of air flow velocity
• Determination of maximum pore size (in micrometers)
• Measurement of permeability
• Flat-sheet dust loading tests using A.C. Fine test
dust. Over-all collection efficiency and dust
loading required to develop 3.7 KPa (15 in H_O)
pressure drop are determined from this test which
is operated at 10 cm/sec (20 ft/min) Air-to-cloth
ratio.
Data collected from these tests are summarized on Table 3.
Penetration tests using D.O.P. smoke measure the ability
of the clean fiber bed to stop fine particles. The D.O.P.
smoke generator is adjusted to provide a nominal particle size
of 0.3 ym diameter which is a "most penetrating" particle size
because of the minimal effect of diffusion and inertial impac-
tion at this particle size. The D.O.P. test results should
correlate well to the results predicted by analysis since
particle collection is provided only by the fibers and not
by the dust cake. Figure 3 provides a plot of the DOP effi-
ciency as a function of air flow velocity for all the media
tested. Ceramic media data are plotted in solid lines and con-
ventional media in dotted lines. Numbers on the curves refer
to those on Table 3. Several interesting observations can be
made concerning this data:
• Several of the ceramic materials, especially the
ceramic papers and felts, are capable of higher
-------
- 358 -
100
90
80
70
60
o.
O
O
| 50
to
o
o
c
a)
« 40
30
20
25
10
34
5 10
Airflow Velocity cm/sec
Figure 3. D.O.P. Efficiency fn airflow velocity.
-------
Table 3
SUMMARY ROOM AMBIENT TEST DATA
(W) Woven
(P) Paper
(F) Felt
1. Carborundum Fiberfrax cloth
(W) with nichrome wire insert
2. Zircar Zirconia felt ZFY-100
(F)
3. ICI Saffil alumina paper
(P) with binder
4. ICI Saffil mat
(F)
5. Babcock s Wilcox Kaowool
(F)
6. Carborundum Fiberfrax
(F) durablanket
7. John Mansville Fiberchrome
(F)
8. Stevens Astroquartz
(W) style 581
9. Hitco Refrasil C-100-96
(W) heat cleaned
10. Hitco Refrasil C-100-48
(W) not heat cleaned
11. Stevens Astroquartz cloth
(W) style 570
12. 3M AB-312 basket weave
IW) cloth
Basis
Weight
g/V
1366
615
165
355
746
1363
1297
283
1284
667
677
311
Percent
Efficiency
on ACF @
10 cm/sec
(20 ft. min)
96.55
95.64
99.805
98.74
98.464
99.507
99.654
60.77
81.97
83.37
56.83
51.38
Dust Loading
(g/ft2 to 15" H20)
. 1010G
(22.2912)
Media Fractured
. 0675
(14.81)
Media Fractured
. 0501'
(10.980)
.06326
(13.6523)
.l&Kt
(23.59)
Test Stopped —
Lof Effw
. (WOifl
(.5482)
.-eo-iao
(1.074)
Test Stopped
Low Eff.
Test Stopped
Low Eff.
Permeability
cm-Vsec/cm2 f°r 0.1245 KPa
(ft3/min/ft2 for 0.5" H2° AP)
8.687
(17.1)
10.861
(21.38)
9.307
(18.32)
12.395
(24.4)
8.067
(15.88)
5.583
(10.99)
11.897
(23.42)
37.236
(73.3)
1.240
(2.44)
3.099
(6.1)
22.758
(44.8)
13.553
(26.63)
Maxima*
pore size
Micrometers
248.6
59
43
61.1
66.9
68. 2
112.3
248.6
112.3
133.8
267.7
870
Percent Efficiency on
0.3 pm OOP at cm/sec
2.68 5.35 14.22
45 47 50
75 70 72
82 65 62
79 80 73 |
00
96.5 93.5 86 j"£
1
97.1 94.6 90.5
78 73 74
0 9 12
0 19 34
0 11 10
0 13 32
058
-------
Table 3 (Continued)
13.
(W)
14.
(W)
15.
(W)
16.
(W)
17.
(W)
18.
(W)
19.
(H)
20.
(W)
21.
(W)
22.
(P)
23.
(W)
24.
(W)
(W) Woven Basis
(P) Paper Weight
(F) Felt g/m2
3M AB-312 twill weave 231
cloth
HITCO Refrasil cloth 643
UC-100-48
Zircar Zirconia cloth 60S
ZFY-30A
,-'
-------
Table 3 (Concluded)
(W) Woven Basis
(P) Paper Weight
(F) Felt S/m2
25. Carborundum Fiberfrax 604
(P) paper (with binder) 970J
26. ICI Saffil Zirconia paper 212
(P) (with binder)
27. Carborundum Fiberfrax 152
(P) paper (no binder) 970-AH
28. 3M AB-312 double thick 1035
(W) plain weave
29. FHI crowfoot satin cloth 905
(W) astroquartz
30. 3M AB-312 12 harness satin 675
(W) weave
31. 630 Tuflex fiberglass* 564
(W)
32. 15-011-020 woven filment* 175
(W) polyester
33. 25-200-070 polyester felt* 524
(F)
34. HITCO Refrasil cloth (std) 637
(W) not heat cleaned, med.
thickness
Percent
Efficiency
on ACF @ Dust Loading
10 cm/sec jjfa? fed 3i73*i r.Pi
(20 ft. min) (g/ft2 to 15" H2O
99.99 :0312r
(6.8442)
93.20 .03 'I0'3'
'Probable hole (7.6374)
99.91 .03G>«{-
(7.8369)
43.86 Test Stopped
Low Eff.
40.08 Test Stopped
Low Eff.
53.73 Test Stopped
Low Eff.
93.982 .•Q2aJ5
(4.4187)
96.078 .frllOfr
(2.60163)
99.193 .M73Z
(12.5688)
48.40 . U14b2
(3.2063)
Permeability Maxinua Percent Efficiency on
t~ cm3/sec/cm2 for 0.1245 KPa core size 0.3 urn OOP at cm/sec
) (ft3/min/ft2
26
(52
8
for 0.5" «20 AP) Microneters 2.68 5.35 14.22
.899 47.7 99.5 99.0 97.6
.95)
.692 37.4 83 78 74
(17.11)
12
(24
84
.416 43.5 88 - 73
.44)
.836 too large to 0 10 41
(167) measure with
our equipment
62
(122
75
(148
16
(31
6
(13
11
(23
6
(13
.078 497 0 10 32
.20)
.529 696 0 8 24
.68)
.038 174 10 9 19
.57)
.828 74 604
.44)
.897 128.9 34 24 29
.42)
.934 - -
.65)
1
U)
CTi
1 — l
1
#These raaterLais are conventional (not ceramic) media.
-------
- 362 -
efficiency collection of fine particles than are
media normally used successively in commercial
filter units.
• Many of the woven ceramic materials had zero D.O.P.
efficiency at low velocity and higher D.O.P. effi-
ciency at higher velocity.
This is contrary to what theory suggests and to the
behavior normally seen in tests of conventional
filter materials. A likely explanation for this
performance is that it is caused by the presence of
many large pores in the media. Examination of the
pore size data ,in Table 3 shows that the woven
ceramic materials as a group are .characterized by
larger pore size than are conventional filter materi-
als. Thus, at low airflow velocity, most of the flow
passes through the large pores and little filtration
takes place. As velocity is increased, flow through
the large pores becomes restricted and some of the
flow is caused to pass through smaller pores where
more filtration can take place.
• The D.O.P- data also supports the theoretical analy-
sis. Efficiency as a function of basis weight for
selected ceramic materials is plotted in Figure 4.
The materials selected are ceramic papers and felts.
These materials provide a fiber bed similar to that
for which the analysis summarized in Figure 2 was
based. Figure 4 shows that the nominally 3 pm fibers
do indeed provide higher collection efficiency on a
weight-per-unit area basis than conventional media
produced with larger diameter fibers.
Maximum pore size data shows that many of the woven
ceramic materials had pores larger than those characteristic
-------
- 363 -
99 r
200 400
600 800
Basis Weight — gm/cm2
1000 1200 1400
Figure 4. D.O.P- efficiency fn basis weight.
-------
- 364 -
of filter materials. Also, many of the felt and paper materials
had pore si/es similar Lo those of. conventional filter materials,
Permeability is measured as the flow per unit area at a
constant pressure drop. Thus, a material with low permeability
offers a high restriction to gas flow and one with high perme-
ability allows more gas to penetrate for a given pressure drop.
Table 3 shows that some ceramic materials are available which
have low permeability, while others have high permeability.
Some of the woven materials have low permeability and large
pore size, while others have high permeability and large pore
size. Most of the paper and felt materials have permeability
similar to that of commonly used filter materials.
Flat sheet dust loading tests were performed as follows:
A 7.62 cm (3 inch) diameter disc of media is suspended across
an air stream which is maintained at 10.16 cm/sec (20 ft/min)
velocity through the filter media. In this test the media
supports itself against the pressure drop (no screen is used).
Standard A.C. Fine test dust (0-80 pm silica) was fed to the
media at a nominal rate of 0.883 g/m (0.025 g/ft ) until a
pressure drop of 3.735 KPa (15 in H^O) is reached. Pressure
drop as a function of time is monitored during the test. This
data is presented in Figures 5, 6, and 7 for selected materials.
From the data collected, dust loading (g/m ) necessary to cause
a given pressure drop 3.735 KPa (15 in HO) is determined.
Examination of this data in Table 3 shows that some of the woven
materials reached high pressure drops while collecting only a
small weight per unit area of dust. This is true also of the
commercial woven materials (items 31 and 32). Other woven
ceramics were penetrated so severely that they would not develop
a pressure drop of 3.735 KPa (15 in HO).
Two of the non-woven samples (which were unsupported)
fractured as a result of the pressure drop across them. Several
of the ceramic paper and felt materials exhibited dust loading,
similar to that which is expected from conventional filter
papers and felts.
-------
6 4
CD
Q.
a
o
Q
A.C. Fine Test Dust
0.883 g/M3
A/C 10.16 cm/sec
#33 = Conventional Felt Filter
CO
ON
Ln
10
20
Figure 5.
30 40 50
Time ~ Minutes
Dust loading of ceramic felts.
60
70
-------
- 366 -
A.C. Fine Test Dust
0.883 g/M3
A/C 10.16 cm/sec
20 30
Time - Minutes
40
Figure 6. Dust loading of ceramic paper.
-------
AC Fine Test Dust
0.883 g/m3
A/C 10.16 cm/sec
#31, 32 Conventional Woven Filters
CO
CPi
-J
40 SO
Time — Minutes
80
Figure 7. Dust loading of woven ceramic media.
-------
- 368 -
The flat sheet loading tests also provided overall col-
lection efficiency (mass basis) data for the tested materials.
Dust penetrating the media was collected in an absolute filter
downstream of the test media. Table 3 reveals that most of
the woven ceramic materials did not achieve high collection
efficiency in this test. On the other hand, woven commercial
materials were only moderately efficient. Several of the
ceramic paper and felt materials, however, did provide collec-
tion efficiency of 99 percent or better. The two materials
which fractured would have provided higher efficiency perform-
ance had they not fractured. The test was stopped as soon as
the fracture was detected.
General Conclusions from Room Ambient Tests
• Several of the ceramic paper and felt materials are
capable of removing fine particles at high efficiency
without excessive filter weights.
• The ceramic paper and felt materials have filtration
characteristics and performed similar to paper and
felt commercial filter media in a series of filter
media tests.
• The ceramic woven materials in general were charac-
terized by large pores and poor collection effi-
ciency in the dust loading tests. The range of
parameters exhibited by the various materials, how-
ever, indicate that an acceptable woven ceramic
filter media can probably be fabricated, but such
a filter media would have the same limitations as
currently available woven filters. That is, accept-
able performance is only probable at low air-to-
cloth ratios.
-------
- 369 -
• "Blanket" ceramic fiber materials (felts) consisting
of small diameter fibers (3.0 pm) appear to be the
most promising materials for high temperature and
pressure tests because of their combination of good
filtration performance and relatively high strength.
High Temperature/Pressure Tests
Two major questions concerning the suitability of ceramic
fibers for filtration need to be answered. These are:
1. How durable are ceramic fiber structures when sub-
jected to environmental conditions associated with
filtration applications.
2. How well do ceramic fibers perform as filters in the
HTHP environment.
Some preliminary answers are available concerning the first of
these questions.
Three ceramic filter media configurations have survived
a test during which the filter elements were subjected to 50,000
cleaning pulses. The objective of these tests was to simulate
approximately one year of operation of mechanical loads on the
media at high temperature and pressure. Test conditions were
as follows:
Temperature - 815°C
Pressure — 930 KPa
Air-to-cloth-ratio - 5 to one (2.54 cm/sec)
Cleaning pulse pressure — 1100 KPa
Cleaning pulse interval — ~10 seconds
Cleaning pulse duration — 100 m second
Dust — rccirculated fly ash
The three filter media configurations tested were:
• Saffil alumina mat contained between an inside and
an outside layer 01 304 stainless steel knit wire
screen. Figure 8 shows how easily the residual
-------
- 370 -
Figure 8.
Saffil Alumina - Post Test Dust Cake
(Clean strip using Vacuum Cleaner)
-------
- 371 -
dust cake was removed from this media after the
test.
• Woven Fiberfrax cloth with nichrome wire scrim
insert. Figure 9 shows the dust cake following
the 50,000 Pulse test.
• Fiberfrax blanket contained between an inside and
an outside cylinder of 304 stainless steel square
mesh screen similar to common window screen. The
ceramic fiber blanket was held in position between
the screens with 302ss wire sewn between the screens.
This resulted in about 100 penetrations of the ce-
ramic fiber bed. Figure 10 shows the dust cake fol-
lowing ^:he test.
Pressure drop during the tests was controlled by the rapid clean-
ing pulses and in general remained less than about 5 KPa (20 in
H20).
Dust penetrating the ceramic test media was collected
on a high efficiency filter located downstream (after cooling)
of the test chamber. This data is not reported for the Saffil
Alumina or for the Woven Fiberfrax Cloth because of a leak
discovered in a gasket in the test rig. This problem was cor-
rected before the fiberfrax blanket test was performed. Average
outlet loading during this test was 0.0055 g/ m3 (0.0024 gr/ ft )
a. a.
Figure 10 shows that the dust was concentrated near the places
where wire penetrated the filter element. This concentration
of dust near the wire penetration points could be seen on the
inside of the element also. Thus, most of the dust which pene-
trated apparently did so through the holes made by the wires.
It is reasonable to expect that a filter element without holes
will experience less penetration. Also a less frequent clean-
ing pulse interval will reduce penetration. Therefore even
better performance than that achieved in this test should result
from future tests.
-------
- 372 -
Figure 9 .
Woven Fiberfrax - Post Test
Dust Cake
-------
- 373 -
F i b e r f r a x
Figure 10.
Blanket - Post
Dust Cake
Test
-------
- 374 -
Conclusions
Research on bench scale indicates that fine particle
control at high temperature and pressure can be achieved using
barrier filtration by ceramic filter beds. Evidence in support
of this contention includes the following:
• A theoretical basis exists for it.
• Room temperature tests show that particles including
fine particles are collected at high efficiency.
• Tests at high temperatures and pressures show that
several ceramic filter structures are capable of
surviving in excess of 50,000 cleaning pulses while
maintaining pressure drop at acceptable levels and
providing efficiency close to the lowest reported
turbine requirements.
REFERENCES
Hazard, H. R., "Coal Firing for the Open-Cycle Gas Turbine."
Proceedings of the Joint Conference on Combustion, 1955, ATME
and 1MB.
Hoke, R., "Ducon Gravel Bed Filter Testing," EPA/ERDA Sym-
posium on High Temperature/Pressure Particle Control,
Washington, D.C., Sept. 1977.
-------
- 375 -
Discussion
Mr. Schulz asked whether needle felts were used; Mr. Shackleton replied
in the negative. Neither ceramic nor steel needle felts were used.
Needling creates holes and defects in the filter. This had been shown
by OOP testing. Mr. Guthner inquired whether there had been a steel
support cage below the felts and how this was attached. A cage had been
used comprising wire screens both inside and outside of the filter.
Mr. Shackleton said that the metal screens might have a short life, and
that ceramic or quartz-based screens could be employed. Mr. Guthner
suspected that the use of metal wire cages for supporting the felts in
large installation caused friction and fiber deterioration. Mr. Shackleton
referred to proposed work in order to determine suitable cage size,
structure and optimization of cleaning systems.
New techniques using fleece (lined on both sides with a screen) as
opposed to cloth or needle felts were described by Mr. Shackleton. This
type of construction was applicable to all fibers, including organic
fibers. Acurex would investigate these novel methods and would also aim
to minimize penetration during cleaning.
Mr. Gooding wished to know whether the observations made during the talk
were also applicable to cake filtration. One of the reasons for using
fine fiber fabric filters was to reduce filter material, said Mr. Shackleton.
Fine ceramic fibers occasioned low solidity in the fiber bed and a
largely open space with few points of contact.
Mr. Finkh asked about the pressure drop after longer periods of filtration,
and whether there was a saturation point and a definite period of
operation. Mr. Shackleton answered that the tests carried out were not
concerned with cleaning, but static loading and were designed to measure
filtering capability. A pressure drop did not develop for certain woven
ceramic fabrics owing to the number of large pores. One should bear in
mind that the materials used were originally developed for insulation,
not filtration.
-------
- 376 -
PROBLEMS ON THE APPLICATION OF
CENTRIFUGAL SEPARATORS, ESPECIALLY
OF ROTARY FLOW COLLECTORS
Prof. Dr--Ing. P. Schmidt,
Dr.-Ing. P- Walzel,
Institut fur Apparatetechnik
Universitat Essen
Unionstr. 2
4300 Essen
-------
- 377 -
Reverse flow or radial cyclone
Straight through flow or axial cyclone
1
'
M
dust collection
^
from gas
dust removal
from cyclone
Smallest particle theoretically collected for
laminar flow without agglomeration:
Symbols;
va Axial vel .
a
u Tang. Vel.
h n Viscosity
n
vr Rad. Vel.
V Throughput
Density
reverse flow cyclone straight through cyclone both types of cyclone
, \fl8 n r vr .
d™ = \/ * * r ;
05 u
18 r| r s va
h u2
9 V
University Essen
Instit. process
equipment
Survey of
cyclone dust
collectors
P. Schmidt
1978
1, Characterization of efficiency by deny or overall efficiency rj.
2. In dust collection from hot gases, the largest particle passing through the
cyclone often is-most dangerous because of later erosion. So effectivness,
f.i. the collection of dgg^ -particles, might be more important than effi-
ciency.
3. Efficiency resp. pressure drop or power demand.
4. Efficiency must include removal of dust from the cyclone into the bin.
5. Most important on efficiency, however, is agglomeration.
curve
for large cyclones
O^— Range of curves for
cyclones
O)
O
U
to
s_
O)
Particle diameter d urn
straight through - reserve flow
0 100 200
Pressure drop mm
water gage
University Essen
Instit. process
equipment
Problems on efficiency of cyclone dust
collectors
P. Schmidt
1978
-------
- 378 -
1. Unstable vortex flow in reserve cy-
clones a)
2. No constant velocity in radial di-
rection for reverse flow cyclone a)
3. Dust strands and wall friction for
both types a) and b)
4. Disordered flow in the sedimenta-
tion chamber a)
5. Imperfect dust removal from the
settling area, especially in axial
flow cyclone b)
6. Changing agglomeration of dust
with temperature for both types a)
and b)
7. To cope with this problems the
TORNADO collector has been deve-
loped
University Essen
Instit. process
equipment
Problems in cyclone operation
P. Schmidt
1978
b)
TORNADO with secondary gas c)
jets from nozzles a) or secon-
dary gas flow through guide
vanes b)
c) Fresh air circuit, effi-
ciency relatively good
only, impossible in hot
gas cleaning
d) clean gas circuit, high
cleaning efficiency
e) Flue gas circuit, low
power consumption
f) Internal faning circuit,
high cleaning efficiency
Pressure drop equivalent to
cyclone
Ap = ^LJ_ ^__2
e)
ar
\
l
Instit. process
equipment
TORNADO dust collector, principle and circuit
diagrams
P. Schmidt
1978
-------
a
) 'Axial
flow
b) Axial
c
or straight through collector,
theory
laminar
collector, turbulent flow
) TORNADO, laminar flow theory
d) TORNADO, turbulent flow
a
c
r! d,
R
F?
1
)
X
^ V
'II
X
-«-H
/
V.
I
M ^
*- r
\J
IXI
rr-
WF
>£
University Essen
Instit. process
equipment
v_ sedimentation velocity of
particles
v. mean velocity of turbulence
v axial velocity of gas
a ^T^
i
)
i •<*$.
^~Jl
J
* ii_i ^^ ^^
*r
^^ \
i
c
c
III
5
|
||
\
I
f
IIHIHI
'^ c /
5° /
OJ O
E -
01 -a
in
<4- 'io
0 ••-
X
-C to
4->
(0 C
Q- -r-
"^
>
^
X
*•
+•> /
v/
^/
i
A-y
,h/
S
**
X
v
0 0,33
Radius r/R
Pathes of comparable particles in
direction for axial
\l\\A
4->
>A><
oo /
> M
/'
0,66
axial
and TORNADO collectors
v->
*
jiii.j/
\/
V
\>
0,83. 1
P. Schmidt
1978
c
cu
4-
4-
Q)
C
o
•r—
•P
O
0)
o
o
(O
c
o
•r—
•P
U
(0
LL
0
a Small TORNADO 200 mm 0,
quartz d-Fractions, 5 g/mJ_
b Small TORNADO 200 mm 0,
fly ash, d(-n = 11 urn,
in _ /«"^ ^*-*
ID g/m°
c Large TORNADOS, different
dusts
0 1 2
Particle diameter
Jniversity Essen
Instit. process
Equipment
Typical collection efficiency curves for
TORNADO dust collectors
P. Schmidt
1978
-------
- 380 -
1. In laminar flow it is basically possible to seperate all particles from
gases by cyclone collectors. However, a practical limit seems to be a dia-
meter of d?z5 urn for isolated particles.
2. In turbulent flow particles are totaly separable only if their settling ve-
locity is higher than the average velocity of the turbulence. In practice
the limit might be d«*15 p. The content of smaller particles can be re-
duced gradually by elutriation, i.e. continous removal in longer cyclone.
3, In hot gases, the Reynolds number is lower than in air under normal condi-
tions by the factor of about 8. If the pressure is higher, the Re-number
however increases by a factor of about 1,I/bar. The use of inserts to build
laminar cycones may be of some interest.
4. When cleaning hot gases a sharp separation limit is essential. - Quite
apart from turbulence the overall gas flow should be as homogenious as
possible.
5. Dust collection with cyclones is much more efficient in practical opera-
tion than in theoretical prediction. In most cases the smallest particle
collected with TORNADO collectors really is d« 2 urn compared with the
sayed 15 ^im; this is .due to agglomeration. - We know little about agglome-
ration under normal conditions and less in high temperature gases. Because
of its crucial importance in dust collection, investigations are necessary
here.
University Essen
Instit. process
equipment
Resume of dust cleani
cyclone collectors
ng hot gases with
P. Schmidt
1978
-------
- 381 -
Discussion
Prof. Weber pointed out the relationship between the diameter of the
Tornados and its efficiency, an increase in the former resulting in
inefficiency. Exact values regarding this relationship could not be
provided at the time by the speaker, although an optimum diameter of 1
meter was known. A diagram showed cut size diameter of 50% at 1/2
micron. Mr. Shackelton questioned the source and validity of this
curve. Mr. Princiotta said that cyclones were used only for particulate
collection in the U.S. and asked for Dr. Walzel's opinion as to whether
these could protect turbines.
Information on this could be found in the paper.
Mr. Shackleton described a U.S. modification of a powered cyclone which
failed during tests to prove the predictions. An overall particle
collection of 80% was inadequate, according to Mr. Princiotta.
Mr. GUthner, in reply to his question, was informed that the total
differential pressure in the cyclone mentioned was 200 mm.
Mr. Bonn (Bergbau-Forschung) discussed the performance of cyclones.
General Electric cyclones used in upstream bed combustion failed to
reach the predicted standards of efficiency owing to fly ash brittleness
and dimisting in conventional cyclone size. Investigation was needed
into Tornados, which could be better due to lack of attribution on the
walls. Dr. Holighaus said that more information was also needed on high
temperature and pressures in the same area. Fluid bed tests were being
made at the time in England to compare conventional with Tornado cyclones.
-------
- 382 -
SESSION V: MEASUREMENT TECHNOLOGY
-------
- 383 -
Experience with Continuously Recording Dust Measuring
Instruments
by
Dr. D. W. Laufhiitte +^
In accordance with the Federal Republic's Technical Instructions
for the Prevention of Air Pollution, solid fuel firing equip-
ment with a heating efficiency above 1OO GJ/h must be provi-
ded with measuring instruments to control and record dust
concentration. At the same time, any equipment emitting more
than 15 kg dust/h, must be fitted out with such instruments.
The Federal Minister of the Interior publishes suitable mea-
suring instruments and widelines for the qualification tests,
maintenance, installation and calibration of measuring instru-
ments as well as the evaluation of measuring results.
Accordingly the majority of power stations, consuming solid
fuels, nowadays are equipped with "smoke meters", following
official instructions.
The calibration of these continuously recording measuring
instruments is carried out by comparative gravimetric mea-
suring which must cover as many different conditions of
firing operation as possible to ensure a safe connection
between gravimetric measuring and recording indication. The
recording instruments predominantly work on the principle
of reflection, i.e. the reflection of a light beam radiated
from a lamp. The weakening of the light beam by absorption
and diffusion is described by the Lambert-Beer's law. All
types of instruments are subjected to laboratory tests prior
to admission.
Practical experience with these instruments shows the high
dependence on the grain size spectrum and the mineral compo-
sition of the available dust types. Fig. 1 (Influence of
different fuels at changing coal-fired boiler conditions).
This leads inevitably to the question for the extent of error
which may result from fluctuations of the material composition
of flue dust and the grain spectrum of these dust types.
-------
- 384 -
In order to get an idea on the fluctuation spectrum of the
particle size distribution which may occur, the Institute
for Mechanical Processing Technique of the Karlsruhe Uni-
versity (Tfr') , in an initial test series with approx. 8O
samples of flue dust from different lignite power stations,
carried out particle size analysis. Moreover, the absorbance
was measured on fractions of different solid materials with
a known particle size distribution and predetermined values
of concentration to estimate the influence of the grain spec-
trum on the integral (mean) scattering cross section determined
experimentally. The goal of this test series was to have new
lights on how the particle size distribution and composition
of flue dust may vary in the course of time, for, as already
mentioned, the temporal constancy of the scattering cross
section is a precondition for the future validity of a cali-
bration curve once determined.
In the scope of the first test series, absorbance was measured
on homogenous substances, such as limestone, quartz, graphite
and four samples of flue dust from lignite power stations.
From all samples, fractions were made for absorbance measuring
and their distribution determined by photographic sedimentation
analysis and wet mechnical analysis.
For absorbance measuring, the Sick instrument RM 4 was used.
The test equipment was arranged in such a way that small amounts
of material could be used for measuring at a relatively long
measuring distance of 2.5O m. Not only the position of the
distribution with regard to the median value substantially
influences the absorbance but also the width. The fraction
with the higher content of fine particles, as a rule will
result in a higher specific absorbance, as indicated in Fig. 2.
The distribution curve with a higher content of fine particles
shows a steeper line of absorbance at an almost equal median
value.
-------
- 385 -
The trials with homogenous substances served as an example for
demonstrating the basic connections of the photometric deter-
mination of mass concentration, whereby it could be proven that
a simple physical connection between absorbance and mass concen-
tration does not exist.
The respective connections flue dust are reflected by the ab-
sorbance results of four dust samples in conjunction with the
grain size analyses conducted.
The four dust samples from crude gas dust were first converted
to two fractions each with particles below 18 m and below 5O m
The resultent distribution curves are given in the upper diagram
of Fig. 3. The dependence of the absorbance as a function of dust
concentration can be taken from the lower diagram.
Compared to the fine fractions, the coarse fractions clearly
show the more gently rising streight lines with the lower ab-
sorbance values.
Additional density tests with the 52 pure gas dust samples drawn
in the summer 1976 show the trend that the density increases
with diminishing dust particles.
As random sample analyses showed, normally the density within
a complete fraction can be expected to rise with a reduction
in grain size, too.
On the whole, absorbance tests with the fractionated crude gas
dust show the considerable range of fluctuation of the specific
absorbance by changes in grain size and density. The absorbance
differences at egual dust concentration for the coarse fractions
(below 5O/am) amounted to more than 1OO % and for the fine frac- .
tions (below 18>fctm) still 25 %.
The results of particle size analyses by the Technical University
of Karlsruhe indicate in Fig. 4 the considerable scattering range
which may occur in the distribution of
-------
- 386 -
flue dust samples from lignite fire places. The demonstrated
scattering range reflects the results of 17 pure gas dust
samples from one power plant block. The median values
extend from about 3.5/urn to almost 40yum.
The median values determined in the second test series
from 52 pure gas dust samples even ranged between 3 yum
and 65yum.
Establishing a relation between the considerable fluctu-
ations of the pure gas dust sample distributions and the
results from measuring the absorbance of fractionated crude
gas dust, differences in the specific absorbance of far
more than 100% cannot be excluded.
Related to the use of smoke meters, this means that con-
centration errors to the same extent must be expected,
when basing a medium calibration curve. Strictly speaking,
this is only applicable to the photometric determination
of dust concentration behind lignite fire places. Errors
to the same extent are, however, also likely with coal
fire places and mixed fire places, as for instance imported
coal of different origin and composition is used in power
stations. Even the local coal varies in composition from
mine to mine, resulting in deviations in the grain spectrum
and the physical features of flue dust. Respective absorbance
tests of the Technical University of Munich with flue dust
from coal fire places demonstrated the possibility of errors
occuring to a similar extent.
Practical effects
After the principle clarification of the possible error
extent in the photometric determination of dust concentration,
the question arises whether the aforementioned errors
actually appear.
The first example in Fig. 5 shows three calibrations within
one and a half years behind a lignite fire.
-------
- 387 -
What is noted immediately is the fact that no calibration
result approaches the other; three absorbance values
extremely deviating from one another are coordinated to
the same dust concentration in the range of 1OO mg/m ,
the maximum difference of which is 150%.
A great portion of deviations in the calibration curves
is caused by changes in the material properties of the pure
gas dust, as indicated by the following example.
Fig. 6 also shows two calibration results determined behind
a lignite fire which heavily differ despite identical con-
ditions of operation and the use of the new instrument type,
practically excluding datum errors by instrument shifting.
The measurement of both calibrations did not result in a
statistical connection between absorbance and specific dust
content, the streight regression lines having a vertical
course. The left hand measuring series is a TUV calibration
from 1973, the right hand series an operator's calibration
from 1975. The recalibration was carried out, following the
request of the competent supervisory office.
Although in both calibrations a statistical connection
between absorbance and dust concentrations could not be found
due to an unfavourable arrangement of measuring points, both
results may be considered as a proof for the displacement of
the measuring series by a change in the grain spectrum on
account of the same characteristics (vertical regression line
despite identical fluctuations in the dust content). At the
same medium dust concentration, the change of the corresponding
absorbance values averaging 0.08 to o.12 amounts to neverthe-
less 5O%. For reasons of completeness, it should be mentioned
that an instrumental defect could not be detected by the
supplier.
Both practical examples give an impression to the fact that
a temporal constancy of the specific scattering cross section
substantially determining the calibration curve of the
metering equipment, due to the fluctuations in the compo-
-------
- 388 -
sition of material of the pure gas dust from lignite
fire places cannot on principle be presupposed. Therefore,
even errors in the specific absorbance of more than 100%
cannot be excluded.
Conclusions
The practical control of dust emissions by the photometric
instruments is carried out according to the evaluation
method stipulated by the Federal Minister of the Interior's
circular letter dated 3-3-1975.
The confidence areas of the calibration curves and tolerance
ranges of the individual values calculated in the scope of
the statistical evaluation form the basis for the assessment
of whether a margin has been surpasses or not. These tole-
rance ranges already include measuring errors of more than
1O%, resulting from gravimetric determinations of the dust
content. On principal the operator may only utilize the
closer confidence area for himself because exceeding a mean
absorbance value - formed from 10 to 15 individual values -
beyond the confidence area means that a marginal value
according to the evaluation method has not been met (Fig. 7) .
On account of the incomplete registration of the variation
range of the grain spectrum during calibration , the confi-
dence and tolerance ranges are too close to ensure a suffi-
cient security in the evaluation of the calibration curve.
Under these aspects, the practized evaluation method may
result in misinterpretations regarding a surpassing of a
marginal value inspite of the tolerances allowed.
Particularly in old plants the surpassing of a marginal
value simulated by the absorbance reading cannot be ex-
cluded because their electric filters work in the marginal
range due to the tightening restrictions for dust emission
values. In such cases instructions of the competent super-
visory office may be expected, requiring cost-intensive
reconstruction or new construction of additional plant parts
of the power station on account of tightening legal regu-
lations .
-------
- 389 -
In this connection the use of integrators intended by
authorities must be seen, resulting in a shutting down
of a power station concerned by an automatic cut-off when
a predetermined absorbance value is surpassed. In this way,
integrators are already used in the cement industry, how-
ever, under the prerequisite of calibration results beyond
doubt by completely registering the grain spectrum of dust.
The use of photometric instruments and consequently of inter-
grators in conjunction with the evaluation method practized
at present seems to be rather delicate when the dust properties
show temporally heavy flutuating features.
A trouble-free use of photometric instruments behind lignite
and coal fire places with considerable fuel fluctuations
and without errors in the concentration reading is only
possible by. additional information on the grain spectrum of
pure gas dust, continuously correcting the photometric values,
i.e. the absorbance values. A unique analysis of grain size
distribution even of all types of flue dust occurring in
praxis is of no use if the type of coal used cannot be pre-
determined. In praxis such a predetermination will hardly
be possible, as the operational analyses (e.g. of sulphur,
ashes and water content) normally do not permit conclusions
concerning the material properties and the grain spectrum
of flue dust.
+ ) Dr. rer. nat. D.W. Laufhiitte, Saarbergwerke AG
department Environmental Protection and commissioner
for emission protection, Saarbriicken
-------
Extinktion E
0,35
0,30-
0,25-
0,20-
0,15-
0,10-
0,05-
0-
2 5 10 20 /urn
PartikelgriiBe x
Graphit (x50"9,0^m)
0
50
100
150
200 250 mg
Knn7Rntratim
O
Fia. 1
Fig, 2
-------
- 391 -
D Fig. 3
PartikelgrbBenverteilung der Fraktionen x <50/xm und x <18|iuT|
1
0
10° 2 5 101 2 5 102V"i
PartikelgrbBe x
Extinktion E
Extinktionsmessungen an Fraktionen <5D|am und <18Jnm
0,45
50 100 150 200mg/m3250
Konzentration Cm
50 100 /*m
PartikelgrbBe x
-------
- 392 -
i
onn
600 -
500 '
mg/m3
n,tr
tnn
4UU '
inn
300
onn
luu
inn
1UU •
n .
L
o 1
0
D 2
k
A/
B
A
t
MeBreihe
MeBreihe
MeBreihe,
einZusamn
fj ^
t
nenhang
O
t
1
t
/•/
/ o'°
/]./ ,
\ /
' i
°
D /
/
3 D .0 —
D '
i
i
t
/
/ /
o/
f
0
/
/
»
/
' 0
7
/
/
k.
0 0,1 0,2 0,3 0,4Extink
Fig. 5
200'
mg/m 3
n,tr
100
0
Fig. 6
k
0
o
$
OD
0°
x1
x2
1
x5x:
x /
1
1
o
D
X
h.
3 0,1 Extinktion 0,2^
o ohne Rufiblasen
mit Rufiblasen
Betreibermessung
-------
- 393 -
mg/m3
A
ocn.
&UU *
150-
100-
50-
L
X
X
l>
1
H-
I'l'i
1
r
/
/
^
/
/
i
f
^
)lera
irEi
izgn
izelv
nzer
verte
* Nachreinigung
rechts ausgeschaltet
of
^uBb
asen
. Vertrauensbereich
der Kalibrierkurve
0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 0,8 0,9
Fiq. 7
-------
- 394 -
Discussion
Dr. Holighaus drew attention to the fact that the techniques des-
cribed in the talk were the same as those in the US. Mr. Kuykendal
reported that work had been carried out there with transmisometers.
Good correlations between mass and opacity had been found during
tests carried out in certain North Carolinian industries. The ope-
ration range was limited and there was no regulation requirement.
Questions were posed regarding new regulations, including those
concerning opacity- This was not to exceed 2O%, according to Mr.
Kuykendal, although the limit could be exceeded for brief periods
of time. Continuous monitoring of dust concentration and risible
emission was carried out. Dr. Calvert referred to an opacity study
which had been performed at a coal-fired power plant. Investigations
showed a good relationship between predictions and measurements re-
garding concentration, particle size distribution and refractive
index. In the case of multiple scattering, 4O% extinction was
found and charges occurred (More information on this could be ob-
tained from a paper by Wei et al.) The work was based on 6 m
t ~in^ on Western coal.
-------
- 395 -
CONTINUOUS CONTROL OF THE DUST CONTENT IN STACK GASES
WITH LASER DEVICES.
Prof. Dr.-Ing. E. WEBER
Institut fur Mechanische Verfahrenstechnik
Universitat Essen GHS
UniversitatsstraBe 2
Postfach 68 43
D 430O Essen
Tel. 0201-1832795
Dipl.-Phys. H. WIGGERS
Institut fur Mechanische Verfahrenstechnik
Universitat Essen GHS
UniversitatsstraBe 2
Postfach 68 43
D 43OO Essen
Tel. 0201-1832788
-------
- 396 -
CONTINUOUS CONTROL OF THE DUST CONTENT IN STACK GASES
WITH LASER DEVICES.
H. Wiggers, E. Weber
It is one of the most important demands for any control of the
dust content of stack gases that the measuring datas are
continuously available. For this reason sampling devices are
not usuable. With these procedures an immediate and continuous
measurement is not feasible because each sample has to be
collected during a longer period. Nevertheless the sampling
procedures are up to now the essential basis for any calibration
of the light, x-ray, or p -ray measuring devices.
All these three procedures offer the potentiality of a continuous
measurement. Mostly the transmission ratio is measured to get
informations about the dust content. If a beam with an initially
power S penetrates an aerosol and if the power at the outlet is
S, the ratio of these powers is defined as the transmission ratio
If the dust concentration is not to high, the law of " Lambert-Beer
predicts the power at the outlet :
S =
Here z is the optical path length of the beam through the aerosol
and c is the value of the dust concentration, of. depends upon the
type of the dust and it es also affected by the particle size
distribution if the problematic x-rays are not taken into account.
This law is indeed valid only for low dust concentrations, but also
at higher values the emerging power is always a strict monotonic
-------
- 397 -
function of c. That is, the measurement of the transmission ratio
is by all means an appropriate method for controlling the dust
content, at least so long as one can assure..a constant type and
particle size distribution of the dust.
For monitoring the emission of dust , for instance that of fossil
fueled power stations, it is up to now customary to use devices
that measure the transmission ratios of white light. These
measurements are done at relative low dust concentrations.
Higher dust contents/ for instance those in the flue duct before
the precipitator, cannot be determined with these devices. But
for a research project such a control has become necessary. The
extreme conditions, a dust concentration of about 30 g m at
normal conditions and a duct width of 4.5 m, seemed to require
complex measurements. ( A measurement across the total width of
the duct has been aspired ) The problem was solved utilizing
consequently the characteristics of a laser beam. The successfully
tested device is a surprising plain construction and it recorded
immediately each variation of the operating conditions in the
combustion chamber, so that a permanent application is considered
now.
For a better understanding the typical characteristics of a laser
beam should be mentioned :
1. The light of the used Helium-Neon-Laser is monochromatic,
i. e. the bandwidth of the light is less than 1 nm.
2. The radiant intensity of the Laser beam is high.
3. The beam divergence is extremely low. Even in a distance
of many meters the beam diameter is practically the same.
4. The light of a laser is highly coherent.
-------
- 398 -
The principle assembly is plain :
As the light transmitter a laser L is used, any system of
lenses can be discarded.
/
L
' ' LL'. •• ."•'
' 1 ' ' • •
. .
. • * .
n
r
]/
V
a
/(
' .'K '••
The laser light LL penetrates the flue duct K. As the pressure
of the flue gas is lower than normal two little holes are
sufficient as lead - in and outlet. Before the laser - light,
more or less weakened by the dust in the duct, can reach a
photo-sensible cell P it has to pass a laserline-filter F.
This linefilter can be transmitted only by such a radiation
that has precisely the wavelength of the laser light. That means
the photo-sensible cell is blocked from any other light that
could disturb. The photocell converts the striking laser-light
to a corresponding electrical potential. Appropriately amplified
the voltage can be recorded.
With this device the first three of the previously mentioned
characteristics of a laser beam are utilized. Because of the low
beam divergence a single photoelement is enough and no focusing
system of lenses or mirrors is necessary. The effects are a simple
procedure for adjustment and also a good resolution of low light
intensities. The monochromatic light of the laser permits the use
of the linefilter. With its bandwith of 1 nm the filter blockades
about 99.7 % of any accidentally incoming white light. The high
-------
- 399 -
energy density of the laser beam provides that also under the
mentioned extreme conditions enough light attains the photocell.
To give some figure data: The output of the used Helium-Neon
laser is 5 mW. Under the extreme conditiones of a dust concentration
of about 30 g m and a duct width of 4.5 m the radiant flux was
weakened with a factor of about 1O~ . The remaining radiant power
could be determined perfectly. As modern photocells have a
sensitivity range of 7 and more magnitudes the limitation of the
described measuring instrument was certainly not reached. It should
be added that the instrument is also appropriate for the measure-
ment of low dust contents. At present it is used to record dust
concentrations of le
static precipitator.
concentrations of less than 1O mg m in the duct of an electro-
The measurements of the transmission ratio with light have a
cardinal disadvantage that should be emphasized also:
The transmission ratio depends upon the particle size distribution,
too. At the same concentration a dust with smaller particles
entails a lower outgoing radiant power of the light beam.
Also therefor one exerts since some years on the problem how to get
continuously particle size distributions with optical methods.
Two different ways for the solution are known:
The first method consists in determining the size of the single
dust particles. The measured and then classified particles are tc
count. To measure their size the particles are let indiviuually
into a light beam and their scattered light is observed, rne
scattering volume, where the single particles are exposed to the
light, can be fixed mechanically or optically. To localized the
scattering volume V mechanically the dust particles are led
o
through the light beam at one location. This can be done for
instance with a nozzle D :
-------
- 4OO -
D
:\
To localize the scattering volume optically one has to use a
system of lenses and slits. The intensity of a light beam is made
constant relativ to its width and only from a certain location
the scattered light can get to a photodetector. One proceeds from
the assumption that the particles come accidentally into the
scattering volume and that they are therefor representative:
scattering
volume
light beam
to the photodetector
Indeed, at both methods of fixing a scattering volume an upper
limit of the measurable dust concentration exists. For to avoid
an obstruction the mentioned nozzle should have no diameter
smaller than O.2 mm, respectively the extension of the optically
fixed scattering volume should not be lower than 0.15 mm due to
statistical reasons. Both methods demand therefor a scattering
-------
- 401 -
volume of at least 3 • 1O mm . From that one gets the maximum
3 11
of the particle number per m as 3 • 10 . Taking the mean particle
diameter as 3 pn the highest dust volume per m is about
The upper limit of the dust concentration is therefor at a few
grams per m . As a result the optical
are applicable mainly at clean gases.
grams per m . As a result the optical methods of particle counting
As light both, white and laser light, can be used. But with this
application of a laser the typical characteristics of the laser
light are hardly used. Only the high power density brings an
adventage that enables to determine also particles in the range
below 1 pm.
For the second optical method to get particle size distributions
the laser light should be used. Because here not only a high
power density but also monochromatic light is necessary. If a
laser beam strikes on a particle that is in front of a screen,
a diffraction patters depending on the size of the particle is
generated on the screen. If this screen is a multi-element-
photodetector then it is possible to compute from many simultaneous
existing patterns the particle size distribution.
-------
- 402 -
Discussion
Discussion was opened by Dr. Cooper, who asked whether the techniques
described offered promising prospects for process stream measurements.
Mr. Wiggers answered that these measurements could be performed with a
Brewster window, cleaned by a jet. No interference by the luminescence
of the high temperature stream occurred. Nor was there any window
surface on which obscuration could occur. The problem of dust building
up on the optic filter was raised by Mr. Kuykendal. Experiments in this
had been performed, but further work was necessary, according to Dr. Wiggers
It was important for the laser output to be constant. Two mirrors with
low transmittance could be employed and the beam emitted could be used
with a photocell to correct charges in output. Dr. Holighaus asked
whether results obtained from tests with this equipment were comparable
with other measurements made. Mr.Wiggers replied in the affirmative,
that experiments had been carried out with white light at power stations.
It had been shown that lasers unlike other devices, could measure dust
concentrations before electrostatic precipitators.
-------
- 403 -
MANUAL METHODS FOR THE DETERMINATION OF PARTICULATE CONCENTRATION, RESISTIVITY,
AND PARTICLE SIZE DISTRIBUTIONS IN INDUSTRIAL FLUE GASES
Joseph D. McCain
Southern Research Institute
2000 - 9th Avenue South
Birmingham, Alabama 35205
U.S.A.
Descriptions are given of manual sampling and analytical techniques used
for determining particulate concentrations and size distributions in the U.S.A..
The methods described include those used for verifying compliance with appli-
cable particulate emission standards, for establishing attainment of control
device performance guarantees, and for research purposes on control device per-
formance .
The techniques used for particulate concentration and emission rate deter-
minations are all based on filtration, either internal or external to the duct.
The actual sampling protocol depends on the gas conditions and the intended use
of the data. Particle size distributions are obtained using a variety of iner-
tial methods for particles having diameters between about 0.5 and 30 jam. Vari-
ous optical, diffusional, and electrical mobility methods are employed for de-
terminations of particle size distribution and concentration in near real time.
The optical techniques provide information over the size range from about 0.3 ym
to 50 ym while the diffusional and electrical mobility techniques provide con-
centration and size distribution information over the size range from about
0.01 ym to 0.3 ym.
In addition, methods for the determination of dust resistivity will be
described. One is a field method which provides in situ data, while the second
is a laboratory method.
-------
- 404 -
MANUAL METHODS FOR THE DETERMINATION OF PARTICULATE CONCENTRATION, RESISTIVITY,
AND PARTICLE SIZE DISTRIBUTIONS IN INDUSTRIAL FLUE GASES
Design and assessment of the performance of industrial participate emis-
sion control equipment require the measurement of a number of properties of the
particulate emissions. Among these are mass emission rates, mass concentrations,
particle size distributions, and, in some cases, particulate resistivity. Meth-
ods for determining these properties can be divided into two broad classes - man-
ual and automatic. In the context of this discussion automatic methods are those
which are used for unattended, continuous, on-line data acquisition while manual
methods are those which require continuous or periodic operator intervention and
are normally used only on a sporadic basis. This presentation will cover only
manual methods.
MASS CONCENTRATION AND MASS EMISSION RATE
Gravimetric methods using filtration for obtaining samples are used for
determinations of mass concentration and mass emission rates. Samples are ob-
tained by withdrawing measured gas volumes from the process stream for equal
time intervals at a series of points across the duct. These traverse points are
located at the centers of zones which divide the duct into a number of zones of
equal area. The number of zones into which the duct is divided depends on the
distances to upstream and downstream flow disturbances and the duct dimensions.
All samples are taken under isokinetic conditions.
One of two general types of sampling systems is used for these determina-
tions depending on the test conditions and the purposes for which the data are
being acquired. The simpler system, illustrated in Figure 1, utilizes a filter
holder and attached sampling nozzle which are introduced directly into the gas
stream. The filtration medium may be a flat glass fiber filter, a glass fiber
thimble or an alundum (ceramic) thimble. The choice depends on gas temperature,
the concentration of the particulate matter, and the intended use of the data.
A pitot tube and thermocouple located adjacent to the sampling nozzle make it
possible to continuously monitor the gas velocity and permit correction of the
sample flow rate as required to maintain isokinetic sampling conditions. On
those occasions when the velocity field within the duct is stable over long
periods of time separate traverses may be made for determination of velocity
and dust sampling, in which case the pitot tube on the sampling probe may be
omitted. This system is frequently used for determining whether a control de-
vice is meeting performance guarantees and for some research purposes, but in
most instances cannot be used for determining compliance with applicable air
pollution regulations in the United States of America.
Testing for compliance with air pollution regulations, with few exceptions,
must be done with the system shown in Figure 2. This system, known as the U.S.
Environmental Protection Agency's Method 5 sampling train, uses a nozzle and
heated probe to withdraw the sample from the duct to a filter located within a
temperature controlled oven.
-------
SAMPLING
NOZZLE
GLASS FIBER THIMBLE FILTER
HOLDER AND PROBE(HEATED)
REVERSE-TYPE
PITOT TUBE
CHECK
VALVE
1
DRY TEST METER
o
Ol
AIR-TIGHT PUMP
Figure 1. American Society For Testing and Materials' Particulate
Sampling Train.
-------
- 406 -
HEATED PROBE
IMPINGER TRAIN OPTIONAL:
MAY BE REPLACED BY AN
EQUIVALENT CONDENSER
CHECK
VALVE
ORIFICE
9^.9
« •— _ «• '
MAIN
VALVE
MANOMETER DRY TEST METER AIR TIGHT PUMP
VACUUM LINE
Figure 2. U.S. Environmental Protection Agency's Method 5 Particulate
Sampling Train.2
-------
- 407 -
The filter may be followed by either a condenser and moisture absorbent
or by a series of impingers and an absorbent. A leakless pump is used to provide
the necessary suction downstream of which are located a dry gas meter and an ori-
fice-type flow meter. The latter is used for monitoring and setting the sampling
rate during a test. This system provides for simultaneous determinations of volu-
metric gas flows within the duct being tested, mass emission rate, mass concen-
tration, and moisture content. The use of appropriate reagents in the impinger
portion of the train permits simultaneous determination of one or more vapor
phase components of the flue gas as well (e.g., SOa, SOa). The probe connecting
the nozzle and filter housing must be glass lined or, if longer than 2.5m, may
be stainless steel or Incalloy. With the exception of tests of fossil fuel util-
ity boilers the temperatures may be raised to 165 C to prevent or minimize the
addition of sulfuric acid condensate to the particulate catch. Sampling rates
range from 20 ipm to 200 £pm.
PARTICLE SIZE DISTRIBUTIONS
The methods most widely used in the U.S.A. for the determination of par-
ticle size distributions of industrial flue gas particulate matter are based on
inertial separation. The separators include cascade impactors, cyclone separa-
tors, and other forms of air elutriators, including the Bahco "microparticle
classifier." Because of the lack of resolution in the fine particle end of the
size spectrum and the problems inherent in the redispersion of collected aero-
sols, the use of laboratory devices such as the Bahco are generally being dis-
carded in favor of other devices which perform the size separation during the
sampling operation.
Th2 cascade impactor is the device most commonly used at this time for de-
termining particle size distributions. The operation of an impactor stage is
illustrated in Figure 3. An aerosol sample is withdrawn isokinetically from the
gas stream and passed sequentially through a series of impingement stages like
that shown in Figure 3. Each succeeding stage operates at higher jet velocities
than the previous stage thereby removing successively smaller particles at each
stage. Determinations of the amount of particulate matter collected on each
stage is then made either by gravimetric or chemical analysis. The particle
size range spanned by an impactor is usually from about 0.50 urn to 10 urn. The
number of stages ranges from about 6 to 13 and flow rates range from about
0.6 £pm to 30 £pm. Prototypes of impactors providing size distribution informa-
tion to diameters as small as 0.02 urn have been constructed and successfully
tested at industrial sites. Figures 4 and 5 show two of the more popular con-
ventional (0.5 urn to 10 urn) impactors in use today.
The cyclone separator is a second type of inertial separator which is in
fairly wide use at the present time. Small cyclones have been demonstrated to
provide rather sharp particle size selection characteristics as illustrated in
Figure 6. This shows calibration data obtained using monodisperse aerosols for
a "cascade cyclone" designed and constructed by Southern Research Institute for
the U.S. Environmental Protection Agency. The systemoperates at a nominal flow
rate of 28 fcpm and provides particle size information from about 0.3 ym to
7.0 ym. Figure 7 illustrates the cyclone system itself. Cyclonic type collectors
-------
- 408 -
JET
TRAJECTORY OF
LARGE PARTICLES
TRAJECTORY OF
SMALL PARTICLES
O
H
O
O
cj
GAS STREAMLINES
IMPACTION SURFACE
o
z
UJ
A. TYPICAL IMPACTOR JET AND COLLECTION PLATE
50
50
PARTICLE DIAMETER
B. GENERALIZED STAGE COLLECTION EFFICIENCY CURVE
Figure 3. Operation principle and typical performance for a cascade
impactor.
-------
- 409 -
NOZZLE
PRECOLLECTION
CYCLONE
JET STAGE
(7 TOTAL)
COLLECTION
PLATE
SPRING
0700-14.1
Figure 4. Modified Brink Model BMS-11 Cascade Impactor.
-------
- 410 -
JET STAGE 0-RING
COLLECTION PLATE
oirg)
INLET
7 \
FILTER HOLDER
COLLECTION
PLATE (7 TOTAL)
JET STAGE
(7 TOTAL)
0700-14.2
Figure 5. University of Washington Mark III Source Test Cascade
Impactor.
-------
- 411 -
"0.2 0.3 0.40.50.60.8 1.0 2 3 4 5 6
PARTICLE DIAMETER, micrometers
• FIRST STAGE CYCLONE
• SECOND STAGE CYCLONE
£ THIRD STAGE CYCLONE
T FOURTH STAGE CYCLONE
O FIFTH STAGE CYCLONE
8 10
Figure 6. Laboratory Calibration for the Five Stage Series
Cyclone System. (472 cm3/sec, particle density—1.0
gm/cm3) .3
-------
- 412 -
CYCLONE 1
CYCLONE 4
CYCLONE 5
CYCLONE 2
CYCLONE 3
OUTLET
INLET NOZZLE
Figure 7. Five Stage Series Cyclone System.'
-------
- 413 -
are particularly useful when sampling gas streams having high dust concentrations,
which tend to lead to rapid overloading of impactors, and when large samples are
needed for chemical or toxicological studies. They may also be a valuable tool
for determining size distributions in high temperature process streams (600°C to
1100°C) such as those associated with coal gasifiers and fluidized bed combus-
tors for combined cycle electrical generating stations. Cascade impactors may
not be suitable for high temperature studies because of a lack of a suitable
substrate on which to collect the particles. Small cyclonic separators are also
frequently used as precollectors upstream of cascade impactors to eliminate
rapid overloading of the first impaction stage. Figure 4 shows one application
of a cyclone precollector.
A three-stage "cascade cyclone" operating at a flow rate of 140 s&pm is in
routine use today as a part of the Source Assessment Sampling System (SASS).
The SASS, illustrated in Figure 8, is used for providing semiquantitative infor-
mation on the form, nature, and composition of pollutants emitted from industrial
processes for the purpose of assessing their potential impact on the environment.
The system provides size segregated samples of particulate matter in four size
ranges (greater than 10 urn, 3 urn to 10 \im, 1 urn to 3 urn, and smaller than 1 urn) ,
usually in sufficient quantities for chemical analysis and for toxicological
studies to determine their potential impact on human health and terrestrial and
aquatic ecosystems. In addition to the particulate samples, the system also pro-
vides samples of organic and inorganic vapors for similar analyses.
Optical and electron microscopy are also used occasionally for determining
particle size distributions but difficulties in obtaining unagglomerated samples
at suitable surface densities on the collection media while still collecting a
truly representative sample make the use of microscopy infrequent.
Size distributions are measured rather frequently using inertial techniques
to provide information for particles larger than 0.5 urn. Details of particle
size distributions from about 0.01 \im to about 0.5 ym (ultrafine particles) are
desired or needed occasionally, but far less frequently than for the larger par-
ticles. Electron microscopy is sometimes used to provide information on the
ultrafine particulate size distribution. Here again, difficulties in obtaining
suitable representative, uncontaminated, unagglomerated samples make this an
infrequently chosen method. The methods most frequently used for the determina-
tion of the concentrations and size distributions of ultrafine particles are
diffusiona] and electrical mobility analyses.
Diffusional methods generally use condensation nuclei counters for measure-
ment of particle concentrations and diffusion batteries for providing the neces-
sary particle size discrimination. The condensation nuclei counter, illustrated
in Figure 9, operates by causing a sample of the aerosol being measured to be-
come supersaturated in the concentration of some condensible vapor—usually
water or alcohol. The supersaturated vapor will condense on any particles pre-
sent in the sample having diameters larger than some critical diameter which is
determined by the particle solubility and the amount of supersaturation achieved.
-------
HEATER
CONTROLLER
CONVECTION OVEN
FILTER
DRY GAS METER
ORIFICE METER
CENTRALIZED TEMPERATURE
AND PRESSURE READOUT
CONTROL MODULE
-A
3u
ofj
.
-. 1u
•A*
\J
\J
f]
I ]
1
I
I
I
_J
T.C.
XAD-2
CARTRIDGE \^/
CONDENSATE
COLLECTOR
(
GAS COOLER
GAS
TEMPERATURE
T.C.
IMP/COOLER
TRACE ELEMENT
COLLECTOR
\
1MPINGER
T.C.
10 CFM VACUUM PUMP
Figure 8. Schematic of the Source Assessment Sampling System."
-------
- 415 -
PHOTO DETECTOR
RANGE
VACUUM
PUMP
r INNER LIGHT STOP
OUTER LIGHT STOP
a
SOURCE LAMP
i—
6H
GEAR
MOTOR
3630-248
Figure 9. Diagram of a condensation nuclei counter. After Haberl and
Fusco.5
-------
- 416 -
This critical diameter is typically about 0.002 urn. The condensation of the
vapor results in the formation of a rather homogeneous, monodisperse fog con-
taining one fog droplet for every particle in the sample which had a diameter
larger than the critical nucleation diameter. Light scattering techniques are
then used to determine the concentration of the fog droplets and hence the ori-
ginal concentration of the particles whose diameters were larger than the criti-
cal size. Size distributions are then determined by measuring the concentrations
of a sample gas stream upstream and downstream of a series of tubes, narrow par-
allel rectangular channels, or screens for which losses resulting from particle
diffusion (resulting from Brownian motion) to the internal surfaces can be pre-
dicted as functions of particle size. Figure 10 illustrates a typical diffusion
battery of the parallel rectangular channel type and penetration curves for a
particular set of flow conditions. The condensation nuclei counters require
samples which are essentially at ambient pressure and temperature and which have
much lower particle concentrations than those typically found in industrial flue
gases. Therefore extractive sampling is required followed by extensive dilution
and sample conditioning to remove undesirable condensible vapors. A schematic
diagram of a complete system used for such analyses by Southern Research Institute
is shown in Figure 11. Dilutions from 10:1 to 4000:1 can be provided with the
system. Diffusional analyses typically provide data over the size range from
about 0.01 jam to 0.2 Urn.
Electrical mobility techniques provide an alternative to the diffusional
techniques for making measurements of ultrafine particles. In this technique
the particles are charged to predictable levels whereupon they are passed through
a mobility analyzer in which particles having progressively lower mobilities are
removed by making step increases in the applied electric field in the mobility
analyzer section of the instrument. Figure 12 illustrates a commercial instru-
ment of this type. Because there is a monotonic relationship between particle
mobility and size in the size range of interest, sorting the particles by mo-
bility leads to size separation of the particles. Concentrations are determined
by measuring the current carried by the particles which are not collected in the
mobility analyzer portion of the device. Thes current coupled with the known
(predicted) charge per particle permits the calculation of the particle concen-
tration at the exit end of the analyzer. This device, like the condensation
nuclei counter, also requires that the sample be at or near ambient temperature
and pressure and that the particle concentrations be much lower than those found
in most flue gases. Thus a dilution and conditioning system like that used for
diffusional analyses is required. The electrical mobility technique, like the
diffusional technique, is capable of providing near real time output for monitor-
ing process variations.
Optical single particle counters such as those illustrated in Figure 13 are
sometimes used to provide real time data on the concentration and size distri-
butions of particles larger than about 0.3 ym (a new counter has recently been
introduced with detection and sizing capabilities down to a diameter of about
0.08 ym but this device has not been utilized in flue gas sampling to date).
These counters operate by sensing the light scattered by individual particles
-------
- 417 -
Figure lOa. Parallel plate diffusion battery.
20
0.01
PARTICLE DIAMETER, /urn
Figure lOb. Parallel plate diffusion battery penetration curves for
monodisperse aerosols (12 channels, 0.1 x 10 x 48 cm).
-------
CYCLONE f V DUMP
PUMP I "
vy
PROCESS EXHAUST LINE
CHARGE NEUTRALIZER
CYCLONE
ORIFICE WITH BALL AND SOCKET
JOINTS FOR QUICK RELEASE
TIME
AVERAGING
CHAMBER
BLEED DILUTION DEVICE
CHARGE NEUTRALIZER
---------.--.-.hii
SOX ABSORBERS (OPTIONAL)
HEATED INSULATED BOX
RECIRCULATED CLEAN, DRY, DILUTION AIR '
FILTER BLEED NO. 2
MANOMETER
COOLING COIL
3630-036
PRESSURE
BALANCING
LINE
BLEED NO. 1
00
I
Figure 11. Sample Extraction-Dilution System (SEDS).'
-------
CONTROL MODULE
ANALYZER OUTPUT SIGNAL
DATA READ COMMAND
CYCLE START COMMAND •
CYCLE RESET COMMAND
AEROSOL FLOWMETER READOUT
-CHARTER CURRENT READOUT
-- ----CHARGER VOLTAGE READOUT
AUTOMATIC HIGH VOLTAGE CONTROL AND READOUT
ELECTROMETER (ANALYZER CURRENT ) READOUT
- TOTAL FLOWMETER READOUT
EXTERNAL [
DATA
~ ACQUISITION'
- SYSTEM
TO VACUUM PUMP
Figure 12. Flow schematic and electronic block diagram of the Electrical
Aerosol Analyzer. Liu and Pui.6
-------
- 420 -
SCNSOA
CHAM
* "*«^J
IBEH . x \ T^^
>xT\\
view VOLUME
CALIBRATOR
SCATTERING
PHOTODETECTOR >
MODULE
CURVED MIRROR
90.9* REFLECTIVITY
5 mm F.L.
PARABOLIC UIRROO
90» REFLECTIVITY
•O" RING SEAL
DUMP WINDOW
AEROOYNAMICALLY
FOCUSING INLET
REFERENCE
PMOTOOETECTOft
UOOULE
PMS LAS-200
^r SAMPLE AIR
COLLECTION PUPIL
LIGHT LEMS LENS
TRAP
PHOTOMULTIPLIEH
AEROSOL
FLOW
DEFINING RELA
OPERTURE LENS
HOYCO 220
LAMP CONDENSER
HOYCO 225
/ PHOTOMULTIPLIEI
Y / TUBE
_/ COLLECTING \
II LENSES \
_ U ,-A W. \
LIGHT TRAP ABSORBS
MAIN LIGhfT BEAM
PHOTOMULTIPLIER
TUBE
CONDENSER LENSES
COLLECTING
MIRROR
PHOTOMULTIPLIEfl
LIGHT
TRAP
ROYCO 215
B AND L 40-1
Figure 13. Optical configurations for six commercial particle
counters.
-------
- 421 -
as they pass through a small sensing zone. The intensity of the scattered light
from the particle provides a measure of its size while the rate at which the
particles pass through the sensing zone provides a measure of particle concentra-
tion. Typical counting rates are 5000 to 10000 particles per second. These de-
vices also require extensive dilution and sample conditioning for use in measur-
ing industrial process streams. They can provide very useful information re-
garding process variations and for control device diagnostics (e.g., characteri-
zing emissions from dry electrostatic precipitators resulting from the cleaning
of the collecting electrodes).
DUST RESISTIVITY
Because of the widespread use of dry electrostatic precipitators for the
control of industrial particulate emissions the resistivity of the dust to be
collected can be of great importance. A dust having a high resistivity
(>2xl010 ohm-centimeters) can tend to limit the maximum operating currents in an
electrostatic precipitator because of electrical breakdown within the collected
dust layer and the resultant back corona. Resistivity determinations can be
made using either in situ or laboratory techniques. Because of possible changes
in the surface chemistry of the dust particles when a sample is extracted and
cooled for transport to the laboratory the _in situ analysis is perferred. The
in situ resistivity probe, illustrated in Figure 14, uses a point-plane precipi-
tator to deposit a dust layer on a collection disk. The voltage and current of
the precipitator are monitored during the sample deposition phase as the dif-
ference between clean plate and dirty plate voltage-current curves provides one
means of resistivity determination. After the dust layer is collected a movable
pad is lowered by means of a micrometer actuated system to simultaneously
measure the thickness of the deposited layer and to provide a second disk elec-
trode. Voltage-current characteristics are then measured across the dust layer
which is now contained between the two disks. This then provides the necessary
information for a second means of the determination of the resistivity. The
second method is more reliable.
A similar, but much larger, cell is used for the laboratory determination.
However, in this case the dust sample is a bulk sample previously taken from the
gas stream which is simply poured and tamped into the cell until a certain sur-
face bearing strength is reached. The cell is then placed in an oven, in simu-
lated flue gas conditions, and the voltage-current characteristics of the dust
layer are measured at the required temperature(s).
-------
- 422 -
PICOAMMETER
CONNECTION
HIGH VOLTAGE
CONNECTION
DIAL INDICATOR
PICOAMMETER
CONNECTION
MOVABLE
SHAFT
STATIONARY
POINT
GROUNDED
RING
(b)
Figure 14. Two types of point-to-plane resistivity probes.4
-------
- 423 -
REFERENCES
1. American Society For Testing and Materials. Standard Method For Sampling
Stacks For Particulate Matter. Designation D2928-71, Annual Book of
ASTM Standards, 1977.
2. Federal Register. Standards of Performance For New Stationary Sources:
Revision to Reference Methods 1-8. Volume 42, Number 160, Thursday,
August 18, 1977, pp. 41753-41789.
3. Smith, Wallace B. and R. R. Wilson, Jr. Development and Laboratory Evalua-
tion Of A Five-Stage Cyclone System. Interagency Energy-Environment
Research and Development Program Report, EPA-600/7-78-008, January, 1978.
U.S. Environmental Protection Agency, Office of Research and Development,
Research Triangle Park, NC 27711.
4. Smith, W. B., K. M. Gushing, and J. D. McCain. Procedures Manual For Elec-
trostatic Precipitator Evaluation. Interagency Energy-Environment
Research and Development Program Report, EPA-600/7-77-059, June, 1977.
U.S. Environmental Protection Agency, Office of Research and Development,
Research Triangle Park, NC 27711.
5. Haberl, J. B. and S. J. Fusco. Condensation Nuclei Counters: Theory and
Principles of Operation. General Electric Technical Information Series,
No. 70-POD 12 (1970).
6. Liu, B. Y. H., and D. Y. H. Pui. On the Performance of the Electrical Aero-
sol Analyzer. J. Aerosol Science, 6, pp. 249-64 (1975).
-------
- 424 -
Discussion
Mr. Guthner asked what the upper dust concentration limit was for cascading
impactors. Mr. McCain answered that a maximum load to each stage was
more significant than the dust concentration. Dust loadings up to 10 mg
per impactor stage are usually acceptable. Reasonable sampling times
(15-30 min) are usually possible with dust concentrations of 2 to 10
g/m . At the precipitator inlet, the use of a small cyclone before the
impactor was usual. Large particles with high momentum could rebound,
biasing the findings. These could be removed by a cyclone however.
(more information on Cascade measurement techniques could be obtained
from EPA; National Technical Information Center).
Mr. Gooding asked whether a test using in situ resistivity had been
done. Mr. McCain said that one very inconclusive preliminary test (at a
high temperature) had been carried out. Further investigations had also
met with little success. Dr. Gooch added that two measurements had been
made at the plant, before this self-destructed. Data corresponded to
laboratory temperature data (300°C).
In reply to a question by Mr. Wiggers, Mr. McCain said that resistivity
was measured by a plate with electrical contact; a small voltage existed
between a pad and a collection plate. Both voltage and current were
measured. The dust package count charged slightly and clean plates
became necessary as dust was collected.
-------
- 425 -
INSTRUMENTAL TECHNIQUES FOR SIZING INDUSTRIAL SOURCE PARTICULATE
William B. Kuykendal
Process Measurements Branch
Industrial Environmental Research Laboratory
Environmental Protection Agency
Research Triangle Park, North Carolina
Introduction
Over the past several years there has been an active interest within
the Environmental Protection Agency, as well as within other organizations,
to develop instrumentation that can measure particle sizes in industrial
sources. This interest stems, for the most part, from a desire to develop
engineering data which will allow for the development, evaluation, and
refinement of particulate control devices.
Idealy an instrument should size the particulate without disturbing
it or its flow field yet obtain a sample of the particulate for subsequent
chemical and toxicological analysis. Such an instrument should have real
time output and two or more sensors so that particulate control devices
could be evaluated on line. The particle size range of interest for con-
trol device evaluation would be 0.01 to 10 micrometers. The preferred
sizing parameter would be the..aerodynamic diameter, although the optical
diameter is also of interest. Clearly some of the items on this ideal
list are very optimistic while others appear to be unachievable. For
example, it would be most difficult to obtain a sized particulate sample
for subsequent analysis and not disturb the particle.
The purpose of this paper is to present recent and current work
in summary form so that the reader may obtain a concise picture of the
direction of EPA's research in the area of particle sizing instrumenta-
tion. Seven instruments will be discussed. The first three are optical
instruments which make use of light scattering techniques to give an
optical particle size. The next three instruments all use inertial
impaction to size segregate the particles. Various schemes are used
to detect the particles to yield the aerodynamic particle diameter.
The final instrument uses an acoustic approach to aerodynamically
size the particulate. Table 1 presents a summary of these instruments.
PILLS IV
The Particulate Instrumentation by Laser L_ight Scattering (PILLS)
family of~instruments~has been developed by Environmental Systems Corp-
oration of Knoxville, Tennessee. The PILLS IV is the fourth instrument
in the series and was designed specificallyoto size particulate in
industrial stacks at temperatures up to 200 C. The instrument schematic
-------
- 426 -
in Figure 1 shows how the instrument operates. The light source of the
PILLS IV is a GaAs diode laser with a lens system that focuses the beam
to a very small cross-section of the order of tens of microns across.
There are three photo detectors at angles of 14 , 7 , and 0 from the
incident beam. The detector at 0 is used to monitor the undeviated
beam to establish a reference for the 14 and 7 detector signals.
When the laser is pulsed (103 times/sec), the 14° and 7 detectors
respond to the scattered light from a particle in the view volume.
If the ratio of these two signals (7 - to 14 -signal) is greater than
1.1, then a particle'size is inferred from its value. If the ratio is
less than 1.1, the .signal from the 14 detector is used to determine
particle size.
The output of the PILLS IV is the number of particles per channel in
two series of channels, one series corresponding to values of the signal
ratio and one series to values of the 14 -signal. The upper and lower
limits of each channel can be varied by the operator, who then relates
particle size to each channel from calibration curves. One curve gives
the 14 detector signal versus particle size, and the other gives the
ratio of the 7 -signal to the 14 -signal versus particle size.
The entire optical assembly is housed in an air cooled probe which
can be inserted through a standard 10 cm sampling port. The cooling air
doubles as purge air to keep the optics clean.
The PILLS IV was evaluated in a wind tunnel test to compare its
results with those from other particle sizing techniques. Figure 2
shows the results from this study. It can be seen that the output from
the PILLS IV differs considerably from the results obtained by cascade
impactors represented by the solid curve. The anomalous response of
the PILLS IV at the low end of the size range is of particular concern.
This problem appears to result in changes in the sizing volume caused by
the two different size ranges used by the instrument. This problem could
perhaps be overcome by modifying the instrument. Future development of
this instrument is uncertain and awaits results from ongoing research on
other techniques discussed below. A report on the evaluation of the PILLS
IV should be published in April 1978.
Fine Particle Size Spectrometer
Work on the Fine Particle Size Spectrometer (FPSS) was initiated in
October 1977 by Particle Measuring Systems, Inc. of Boulder, Colorado.
The instrument uses a He-Ne laser as shown in Figure 3 as the illumination
source for the forward scattering technique. The input beam is directed to
the condensing mirror which focuses the beam to a 150 micrometer diameter
sensing volume at the object plane. The object plane is defined by the
combination of the condensing mirror and the prime objective lens. "The
prime objective gathers the total light scattered over the forward angles
2 to 11 . The secondary objective magnifies the 150 micrometer object
-------
- 427 -
plane by a factor of twenty to increase the object area on the paired photo-
diodes. A beam splitter is used to divide the image on to each photodiode.
One photodiode has a mask which obscures a portion of the object plane.
This is done to limit the response of the instrument to only those particles
that are in the uniformly illuminated portion of the input beam. In order
for a signal to be processed, it must appear on the unmasked photodiode and
not on the masked photodiode. All of the optical components except for the
condensing mirror are housed in a water cooled enclosure. The condensing
mirror protrudes from this housing on two support rods. The condensing
mirror will be cleaned by a purge of dry air or nitrogen. The entire
assembly can be inserted through a 10 cm sampling port.
The scattering signal from the photodiodes is then electronically com-
pared with the theoretical Mie scattering curve shown in Figure 4 to yield
the particle size data. This curve exhibits the characteristic oscillation
in the Mie scattering below about 3 micrometers for the range of index of
refraction considered. However, it should be noted that if a best fit
calibration curve is used at no point is the uncertainty greater than
±20%. Laboratory calibration tests conducted with Latex spheres indicate
that the actual uncertainty is less than the maximum. Present data indi-
cates that sizing accuracy of ±10% seems reasonable.
This instrument is in the field prototype stage and will be tested in
a wind tunnel and then in a coal fired power plant. The current assessment
of the instrument is very positive. It appears that it will function well
and yield reliable real time data. No major development problems are antic-
ipated. A report is expected in October 1978.
Optical Particle Sizer
The Optical Particle Sizer is currently being developed by Leeds and
Northrup Company of North Wales, Pennsylvania. This instrument differs from
the two previous instruments by determining the particle size by measuring
the signal from an ensemble of particles rather than from single particles.
The optical schematic of Figure 5 shows the basic layout of the major com-
ponents. A xenon arc lamp is used as the multiwavelength illumination
source. The 0.3 to 10 micrometer sizing range is divided into two ranges.
Particles larger than 1.0 micrometer are sized by forward scattering.
Three scattering angles (10°, 5°, 3.5°) are measured simultaneously and
the signal is compared to a theoretical response curve to give the particle
size. Particles less than 1.0 micrometer are sized using the depolarization
ratio of the light scattered at 90° and at the wavelengths of 0.421 micro-
meters and 0.850 micrometers. This ratio is electronically compared with a
theoretical response curve to yield particle size. Because this instrument
makes use of the scattered signal from a volume of particles it is limited
to relatively low particle concentrations to avoid multiple scattering
effects. In most applications it is expected that the instrument would
function properly after an efficient particulate control device, but
not before one.
-------
- 428 -
The- instrument is housed in a probe much like the one used by the PILLS
IV. This probe can then be inserted into the stack. Optical fibers are used
to transmit the light from both the forward scatter and side scatter optics
to the photodetectors located outside of the stack as shown in Figure 6.
The assessment at the present time for this instrument is quite good.
It is in the prototype stage and wind tunnel and stack tests are planned.
The final report is expected in July 1978.
Beta Impactor
Work on the Beta Impactor was initiated in 1974 by GCA Corporation of
Bedford, Massachusetts. This prototype instrument uses an inertial cascade
impactor to aerodynamically size classify the particulate coupled with a
beta radiation detector to sense the mass of particulate on each impactor
collection stage. The attenuation of beta radiation has been shown to be
a linear function of the mass of collected particulate without regard to the
chemical composition of the particulate. This useful phenomon had been used
in stack particulate mass instruments and therefore seemed reasonable for
use in the Beta Impactor. It was also desirable to size the particulate
as nearly in situ as possible and the decision was made to locate the
impactor with beta sources and detectors in the stack.
The particle sizing is exactly the same as with a conventional cascade
impactor. The particulate laden gas stream enters the impactor nozzle and
is directed on to the impaction surface. Particles larger than the size
cut point for the stage impact on the stage surface, while the smaller
particles follow the gas stream to the next stage. The Beta Impactor
uses a moving impaction substrate which collectes the sized particulate.
Two source-detector pairs are used to measure the mass of the substrate
before and after impaction and the real time particulate mass per stage
is computed by difference. Carbon 14 was used as the beta source while
geiger-muller tubes were used for detectors.
A very serious development problem was encountered during the program
which was never overcome. Since the device was designed to operate at stack
conditions it was necessary that the beta detectors should be capable of
operation at the 200 C stack temperature. No beta detectors were located
or developed that could achieve this design requirement.
The instrument was tested in a wind tunnel to evaluate its operational
characteristics at ambient temperature. These results are shown in Figure
7. This figure compares the output of the Beta Impactor with the same
reference impactor curve presented in Figure 2. Subsequent analysis of
the wind tunnel test configuration revealed that there was an obstruction
between the reference impactor and the Beta Impactor. This obstruction
had the net effect of diverting approximately 80% of the flow away from
the Beta Impactor. If this correction is applied to the Beta Impactor
data, the data points fall more nearly on the reference curve. However,
-------
- 429 -
the same degree of scatter in the data results and the first impaction
stage remains well below the reference curve.
These discrepancies with the reference impactor could probably be
minimized with a refined design. However, when coupled with the beta
detector problems discussed earlier, it was decided that further devel-
opment of this technique was not warranted. A report on this work was
published in April 1977.
Differential Pressure Impactor
As the development problems of the Beta Impactor became apparent, a
second program was initiated with GCA Corporation to develop an in stack
impactor with real time data presentation. Under this program a simpler
sensor was selected from several candidates. The system schematic is
presented in Figure 8. This instrument utilizes virtual impaction on to
a stationary gas rather than the more conventional impaction on to a surface
employed by the Beta Impactor. With virtual impaction the separated sized
particles settle into a collection chamber where a small portion of the
total flow is extracted. The particles are collected on a filter and the
differential pressure across the filter is measured. The change in the
differential pressure is proportional to the weight gain of the filter
and the real time particle size can be calculated.
Because a sample is collected on a filter several advantages result.
The differential pressure/mass calibration can be verified for each test
by weighing the filters. Likewise, because a sample is collected, subse-
quent chemical and toxicological analysis can be performed.
The experimental results on this technique, however, were disappoint-
ing. Figure 9 shows a typical test using fly ash. It is quite, apparent
from this data that the instrument suffers from a lack of sensitivity.
Minimum time response can be up to 10 minutes depending on source char-
acteristics. Although the technique offers more promise than the Beta
Impactor, further development would be required before a useable instru-
ment could result. EPA has no plans for further work on this instrument.
The program results were published in a report in August 1977.
Light Scattering Impactor
Meteorology Research, Inc. of Altadena, California, was selected in
October 1977 to develop an impactor based instrument utilizing a light
scattering data readout. Although the final design has not been selected,
Figure 10 shows the basic concept. A conventional inertial impactor,
possibly employing virtual impaction, will be used to separate the partic-
ulate by their aerodynamic diameter. Each impactor stage will be followed
by a detector module. The active optical elements (light source, photo-
detectors, and associated electronics) will be located outside of the
stack and connected to the in-stack impactor with optical fibers. The
-------
- 430 -
detection scheme can be seen in Figure 11. It employs forward scattering
over a fairly large angular range from the total ensemble of sized particles.
It is recognized that this method of sensing is dependent on the index of
refraction of the particles. However, because the size of the particles
is known from the impactor characteristics, and because it is only necessary
to measure the particle concentration, the approach appears reasonable.
Although no experimental work has been performed to date, the approach
is quite straightforward and no serious problems are foreseen. Future test-
ing, if warranted, calls for evaluation of the instrument in a wind tunnel
followed by a field evaluation in a coal fired power plant. The Light
Scattering Impactor will be compared with conventional impactors in each
test. A report is expected in May 1979.
Acoustical Particle Sizing Instrument
Work on the Acoustical Particle Sizing Instrument was begun in February
1978 by KLD Associates, Inc. of Huntington Station, New York. The instrument
computes the aerodynamic particle size by measuring the change in the speed
of sound at various audio frequencies. When a particle is exposed to a sound
wave in a gaseous medium, the particle will either oscillate in the gaseous
medium at the frequency of the sound wave or it will be unaffected. The
particles' aerodynamic drag and mass are the parameters that determine
whether or not the particle will oscillate. For each size particle there
is a unique frequency above which the particle will no longer oscillate.
If the particle oscillates with the gas then the apparent density of the
gas is increased by the addition of the mass of the particle to the mass
of the gas. If the particle does not oscillate with the gas then no change
is observed in the gas density. Since the speed of sound is a function of
the density of the transmitting medium, then by measuring a change in the
speed of sound the change in apparent gas density can be calculated.
The application of this useful relationship can be seen in Figure 12.
Two known frequencies Fx and FR are passed through the particulate laden
gas medium. The reference frequency, FR, is chosen sufficiently high so
that none of the particles of interest will oscillate. Therefore, the
velocity of sound at FR will remain constant. When the velocity of sound
at FX is measured it is found to be different than the sonic velocity at
F^. The difference in sonic velocity represents the addition of the total
mass of particles in the gas stream that oscillate between frequencies F^
and FR. A useful instrument would result if the frequency F^ is varied
to cover the full range of particle sizes of interest as shown in Figure 13.
Curve (A) represents the case when no particles are present in the gas
stream. The sonic velocity remains constant across the audio spectrum.
When a monodispersed aerosol is introduced curve (B) results. In this
case sonic velocity is constant until the critical frequency for the
monodispersed particles is reached. The speed of sound changes by an
amount AC as the particles start to oscillate and then assumes a constant
but different value., If a polydispersed aerosol is introduced then curve
-------
- 431 -
(C) will result. There is an initial reduction in the sonic velocity
(ACA) caused by the total mass of particles present in the gas stream
that oscillate above the initial frequency. As the frequency is changed
from point A to point B the sonic velocity changes by AC/,g. This repre-
sents the total mass of particles that oscillate between frequency A and
frequency B. The particle size distribution can be determined by differ-
entiating curve (C) to obtain the mass at any desired size.
Although the concept appears quite promising, no experimental work has
been done. There are a number of potential problems dealing with acoustic
transducers that have not been addressed. This instrument may prove useful,
but considerable research will be required. A technical report is scheduled
to be published in June 1979.
Assessment
Past work to develop in situ real time particle sizing instruments has
been largely unsuccessful. Current work, however, seems to hold considerable
hope for success. It appears that both the Fine Particle Size Spectrometer
and the Optical Particle Sizer have excellent chances of working well and
developing reliable data in industrial sources. Likewise, the Light Scatter-
ing Impactor offers promise of success. The Acoustical Particle Sizing
Instrument will represent a major advance in particle sizing technology
if it is successful. Current research should answer these questions.
It is the author's opinion that within one year practical, if not
optimum, real time particle sizing instruments will be available.
-------
- 432 -
References
1. Harris, D. B. and W. B. Kuykendal. Problems in Stack Sampling and
Measurement. Proceedings of the Symposium on Fine Particles, Minn-
eapolis, MM. EPA-600/2-75-059. October 1975.
2. Shofner, F. M., G. Kreikebaum, H. W. Schmitt, and B. E. Barnhart.
In Situ, Continuous Measurement of Particulate Size Distribution
and Mass Concentration Using Electro-Optical Instrumentation.
Presented at the Fifth Annual Industrial Air Pollution Control
Conference, Knoxville, TN. April 1975.
3. Gooding, C. H. Wind Tunnel Evaluation of Particle Sizing Instruments.
EPA-600/2-76-073. March 1976.
4. Lilienfeld, P., D. P. Anderson and D. W. Cooper. Design, Development,
and Demonstration of a Fine Particulate Measuring Device. EPA-600/2-
77-077. April 1977.
5. Lilienfeld, P., D. P. Anderson, and D. W. Cooper. Study on the
Feasibility and Design of Automatic Particulate Size Distribution
Analyzer for Source Emissions. EPA-600/2-77-050. August 1977.
-------
Table 1
INSTRUMENT SUMMARY
INSTRUMENT
PILLS IV
FINE PARTICLE
SIZE SPECTROMETER
OPTICAL PARTICLE
SIZER
BETA IMPACTOR
DIFFERENTIAL
PRESSURE
IMPACTOR
LIGHT SCATTERING
IMPACTOR
ACOUSTICAL PARTICLE
SIZING INSTRUMENT
OPERATING PRINCIPLE
FORWARD LIGHT
SINGLE PARTICLE
SCATTERING
FORWARD LIGHT
SINGLE PARTICLE
SCATTERING
FORWARD LIGHT VOLUME
SCATTERING/SIDE SCATTERING
POLARIZATION RATIO
INERTIAL IMP ACTION
BETA ATTENUATION SENSING
INERTIAL IMPACTION
PRESSURE BUILD-UP
SENSING
INERTIAL IMPACTION
LIGHT SCATTER SENSING
CHANGE IN SONIC VELOCITY
VS AUDIO FREQUENCY
SIZING
RANGE,
micrometers
0.2- 3
0.4 - 10
0.3 - 10
0.2- 5
0.2- 5
0.5 5.0
0.5- 5.0
DEVELOPMENT
COSTS.
$
UNKNOWN
155,000*
150,000
175,000
90,000
265,000*
140,000*
ESTIMATED
UNIT COST,
$
30,000
35,000
25,000
NOT
APPLICABLE
NOT
APPLICABLE
40,000
30,000
STATUS
EVALUATION
COMPLETE
ACTIVE
DEVELOPMENT
ACTIVE
DEVELOPMENT
RESEARCH
TERMINATED
RESEARCH
TERMINATED
ACTIVE
DEVELOPMENT
ACTIVE
DEVELOPMENT
U)
00
-------
CONTROL LASER AND
ELECTRONICS OPTICS
SAMPLING
VOLUME
OPTICS1'
DETECTOR,
AMPLIFIER
RATIO
SIZE
ANALYSIS
CIRCUIT CIRCUIT* READ-OUT
i* DESIGN- ESC PROPRIETARY
DIGITAL PRINT-OUT
PAPER OR MAGNETIC TAPE
CRT HISTOGRAM
LED DISPLAY OF COUNTS
MINI-COMPUTER INPUT
RATE METER
TYPICAL SIZE RANGE -0.2-3.
Figvire
PILLS IV Optical Particle Counter.
-------
- 435 -
10'
10
E
z
c: 10
T5
^k
to
p
UJ
a:
UJ
10
.63
10
O(2)
8
8
o
o
O PILLS N DATA
8
O
O
0(2)
8
0.1
1.0 10
PARTICLE DIAMETER , Dgeo ,
100
Figure 2. Comparison of PILLS IV data with impactor curve
(concentration = 0.955 g/Nnr5) .
-------
FINE PARTICLE SIZE SPECTROMETER
ADJUSTABLE
BEAM STEERING
MIRROR
DETECTOR _
MODULE / SPLITTER | SECONDARY
20X
OBJECTIVE
BEAM
STOP I
OBJECT
PLANE
AR COATED
WINDOW
OJ
Oi
7cm RADIUS
CONDENSING BEAM
FOLDING MIRROR
(WITH AXIAL
ADJUSTMENT)
Figure 3
-------
FINE PARTICLE SIZE SPECTROMETER
THEORETICAL MIE SCATTERING FOR 2-11° REAL INDICES 1.4, 1.5, 1.6, & 1.7
104
103
Z
E
102
>co
LU
CC.
BOUNDARY OF UNCERTAINTY
BEST FIT CALIBRATION
CURVE
10°
BOUNDARY OF UNCERTAINTY
m = 1.4-1.7
• = LATEX PARTICLES
10-i
I
I
I
I
I
I
I
I
I
I
CO
-o
0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4 2.6 2.8 3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8 5.0
RADIUS (u)
Figure 4
-------
OPTICAL PARTICLE SIZER
AIR CURTAIN . CMO
/.-WINDOW /LENS
SIDE SCATTERING
OPTICS
£-SLOT FIBER OPTIC BUNDLES^
PROBE TIP ASSEMBLY
00
I
Figure 5
-------
- 439 -
OPTICAL PARTICLE SIZER
INSULATING COLLAR
SAMPLE SLOT
FLANGE
PROBE
PROJECTOR
AND
DETECTOR
ASSEMBLY
FILTERED
AIR
SUPPLY
LAMP
POWER
SUPPLY
ELECTRONICS
AND OUTPUT
DEVICES
Figure 6
-------
- 440 -
10'
o»
o
o
•o
CD
cr
in
UJ
a:
UJ
I01
io2
io3
O OCA DATA
O.t 1.0 10
PARTICLE DIAMETER, Dgeo.^m
Figure 7. Comparison of Beta Impactor data with impactor curve
(concentration = 0.955 g/Nm3].
100
-------
- 441 -
DIFFERENTIAL PRESSURE IMPACTOR SCHEMATIC.
NOZZLE
GAS INLET
ORIFICE
STAGE 1
xxxxxxxxxxxxxxx
BACKUP FILTER
TOKEN FLOW
TOKEN FLOW
AP3
MAIN GAS FLOW
Figure 8
-------
- 442 -
DIFFERENTIAL PRESSURE IMPACTOR
O ,-
pg O
I
„ r^
0)
0.5
0.4
0.3
0.2
0.1
BACKUP
• « 9
AP2
1 1 L_
8 16 24
32 40 48
TIME, minutes
56 64 72 80 88
TEST NUMBER 29, FLY ASH.
Figure 9
-------
- 443 -
LIGHT SCATTERING IMPACTOR
QUICK
DISCONNECT
PLUGS
LIGHT
SCATTERING
SENSORS
FILTER OPTICS
BUNDLES
INERTIAL
SEPARATION
STAGES
CASCADE IMPACTOR WITH LIGHT SCATTERING
SENSOR STAGES ADDED
Figure 10
-------
- 444 -
SENSOR PIPE—i
SOURCE PIPE—1
LIGHT SCATTERING IMPACTOR
THREADS FOR PRECEDING STAGE
LIGHT SOURCE PIPE
SENSOR PIPE
FIBER OPTICS TO
QUICK DISCONNECT
PLUG ON PROBE
SENSOR FIELD
OF VIEW
SPACER RING
HOUSING
THREADS FOR
NEXT STAGE
DETECTABLE
SCATTERING VOLUME
FIBER OPTIC LIGHT PIPES
HOUSING
FIBER OPTIC SENSOR STAGE
Figure 11
-------
- 445
TRANSMITTER
FR
AMPLIFIER
AMPLIFIER
FR-/L
FILTER
FILTER
FREQUENCY
MULTIPLIER
n
FREQUENCY
COMPARATOR
nF,
INTEGRATION
TIME
CLOCK
DIGITAL
DISPLAY
COUNTER
SCHEMATIC OF THE ACOUSTICAL PARTICLE SIZING INSTRUMENT
FOR MEASUREMENTS AT FREQUENCY, Fx, WITH RESPECT
TO THE REFERENCE FREQUENCY, FR
Figure 12
-------
(A) - NO PARTICLES
Ac
Q
•z.
i
Q
ID
LU
X
O
z
o
X
/
-(B) - MONODISPERSION
SIZE = 1.5pm
NC) - POLYDISPERSION
t
AcA
AcAB
1.5
90
KHz
FREQUENCY
TYPICAL OUTPUT TRACES FROM THE ACOUSTICAL
PARTICLE SIZING INSTRUMENT
(A) NO PARTICLES IN SIZE RANGE
(B) MONODISPERSION: SIZE = 1.5 ^m
(C) POLYDISPERSION
Figure 13
-------
- 447 -
Discussion
Mr. Kuykendal stated that all devices described required physical insertion
into the gas stream. In reply to Mr. Wiggers, he said that the PILLS IV
instrument functioned well apart from a response problem and that a
trained technician (not necessarily a specialist) was well able to
handle the data obtained.
-------
- 448 -
Evaluation of Particle Size Distributions by Means of
Particle Counters
H.J. FiBan
C. Helsper
Vortrag auf dem Workshop "Particulate Control",
Kernforschungsanlage Julich, 16.03. - 17.03.1978
Bericht Mr. 33
-------
- 449 -
Evaluation of Particle Size Distributions by Means of
Particle Counters
Prof. Dr.-Ing. H. FiBanj Dipl. Ing. C. Helsper
Institut fur AerosolmeBtechnik
Gesamthochschule Duisburg
Introduction
The emission of particulate matter is usually described by
its total mass concentration, or, as a function of particle
size, by its mass distribution.
The mass of a single aerosol particle decreases rapidly with
decreasing particle size. Therefore, the determination of
the particle mass concentration or the particle mass
distribution of an aerosol with particle diameters in the
submicron range by gravimetric methods yields certain
difficulties like long sampling times and a loss in accuracy.
For submicron particles it is easier, and more precise to
count the particles in order to get parameters, describring
the aerosol. Another advantage is, that most of the particle
counters allow near real-time measurement, so that the
aerosol parameters can be obtained in time intervals of a
few minutes. Then it is possible to determine changes in the
aerosol as a function of time.
Mass distributions can be calculated starting from particle
number distributions under the assumption of spherical
particle shape, if the density of the particle material
is known.
Instrument description
In our laboratory a set of two instruments has been used
to determine size distributions of aerosols emitted by
combustion processes in the particle diameter range between
0.01 and 10 microns.
-------
- 450 -
For the diameter range below one micron an Electrical
Aerosol Analyzer (E A A, TSI,Model 303D) was used.
Fig. 1 shows a schematic diagramm of this instrument.
The aerosol particles are charged in the upper section of a
tube by unipolar gaseous ions produced by a corona discharge,
These charged particles enter a cylindrical condenser
through a narrow annular slot near the inner wall of the
outer tube. A flow of clean air near the center electrode
is necessary to obtain a laminar flow. With a negative
voltage at the center electrode the particles are deflected
through the air stream towards the electrode. The particles,
which are not precipitated on the center electrode, are
collected on a filter downstream the precipitator-
The particle charge flow is measured by an electrometer
sensor. If the voltage at the center electrode is increased
by a certain amount, the electrometer current will decrease,
.because particles of less electric mobility are precipitated
with the higher voltage in the condenser. The difference
in current for a defined voltage step is a measure for the
number concentration in a certain particle mobility range.
Thus, by varyingthe precipitating voltage, the electric
mobility distribution of an aerosol can be determined. The
relation between particle size and particle mobility depends
on the number of elementary charges per particle. Because of
the random nature of the charging procedure this relation
can be described only statistically- This is the reason for
a non-ideal instrument behaviour.which causes systematic
errors in the measured size distribution. The instrument
response for monodisperse aerosols has been determined
by Liu and Pui in 1974. Based on their results a correction
of these errors is possible.
For the measurement of the larger particles an Optical
Particle Counter (0 P C, Royco, Model 225) has been used.
Fig. 2 explains the principle of the instrument.
A beam of white light is focussed by a set of lenses and
is then caught by a light trap.
-------
- 451 -
The aerosol stream, focussed by clean sheath air crosses
the focus of the light beam. The amount of light, that is
scattered by a particle going through this volume, is a
function of particle size. The light pulse in forward
direction is led to a photomultiplier. The output pulse
of the photomultiplier is amplified and transformed in a
square shaped pulse of the same amplitude. The pulses
are stored according to their amplitude by a multi-channel-
analyzer-
Automatic instrument control and data reduction is done by
a PDF 11/10 computer. Fig. 3 shows a block diagramm of the
whole system. The EAA and the OPC are complemented by a
Condensation Nuclei Counter (C N C) for the determination
of the total number concentration in the size range below
0.8 micron, and a piezoelectric microbalance, which
determines the total mass concentration. By specific
interface circuits the data are modified and transferred
to the processor unit, from where they can be stored on a
magnetic storage device. Data output is available on a
graphic plotter or on a teletype.
The two instruments were designed for ambient air measurement
and have certain limits concerning the maximum particle
number concentration as well as the temperature and pressure
of the gas.
Emission aerosols exceed these limits in many cases. Therefore
it is necessary to modify these aerosols by dilution and
cooling. For dilution a certain amount of the aerosol flow
is led through a filter. The particle-free gas is then
mixed again with the rest of the aerosol flow.Dilution
ratios of 50 to 1 can be obtained with this method without
altering the shape of the size distribution.
Cooling of aerosols is much more critical because condensation
processes can alter the properties of an aerosol considerably.
-------
- 452 -
Data reduction and correction
The data of the EAA and to some extend that of the OPC,
too, are not free from systematic errors. Fig. 4 and 5
show the effects, caused by these errors. The continuous
curve in Fig. 4 shows a model number distribution similar
to the size distribution of atmospheric aerosols. It was
obtained by the superposition of two log-normal distri-
butions and is represented by the full line. The
instruments' response to that model aerosol was simulated
numerically and is represented by the histogramm.
The solid lined blocks represent the EAA data and the
broken lined the OPC data. Practically all particles lie
in the size range below one micron. A considerable part of
the distribution lies even below O.D1 micron, which means,
that no direct information can be obtained about this part
of the distribution with these instruments. The differences
between the measured data and the model distribution are
not very large at first sight because of the logarithmic
scale which ranges over six orders of magnitude for the
number concentration axis. But in the upper size channels
the values differ by a factor of nearly five.
These differences become more abvious, if the number
distribution is transformed into a volume distribution,
which corresponds to the mass distribution for unit density
of the particles. This transformation has been done under
the assumption of spherical particles. Its results are
shown in Fig. 5.
One can see, that nearly ninety percent of the particle
mass is concentrated in the size range below one micron.
The largest differences between the model distribution
and the measured data lie now near the maximum of the
distribution, so that, for example, their effect for the
total mass concentration is much greater than for the total
number concentration.
-------
- 453 -
Besides the intention to correct the measured data it would
be convenient to describe the whole distribution by a
mathematical model distribution with a limited number of
parameters. This reduces the amount of data necessary
to describe the aerosol and allows an extrapolation over
the limits of the size range covered by the instruments.
A superposition of two or three log-normal distributions
has found to be consistent with most of the experimental
data .
A computer program has been developed, which determines
for such a model distribution a set of parameters that
fits the measured data best, taking into account the
non-ideal instrument behaviour. Based on the measured
data a set of start parameters for the model distribution
is estimated. The measuring procedure is then simulated
numerically for the distribution represented by these
start parameters.
The result of this operation is a set of simulated data,
which is compared with the measured data to estimate the
goodness of fit. An optimization algorithm now varies the
distribution parameters according to a certain strategy
in order to minimize the deviation between the measured
and the simulated data. If the deviations fall within the
range of the instruments accuracy, the iteration stops.
The model distribution, found in this way represents the
measured data within the accuracy range of the measure-
ment, which is about twenty percent.
Experimental results
As an example for the application of the instruments for
the determination of the size distribution of combustion
aerosols the result of a measurement at the exhaust gas
of an internal combustion engine shall be discussed.
In order to compare the results of the particle counting
technique with those of gravimetric methods we used a
-------
- 454 -
six stage cascade impactor.
All instruments took their samples downstream a model
exhaust system which cooled the exhaust gases below
fifty degrees centigrade. To compare the results of
the different methods, mass concentrations were calculated
from the number concentration values, assuming a density
3
of the aerosol material of 1 g/cm .
The resulting mass distribution, which is based on
uncorrected data is shown in Fig. 6. The solid points
represent the data calculated from the EAA measurement.
The open circles show the results of the cascade impactor
and the two squares result from the DPC measurement.
Even with the uncorrected data the fit between the
different methods is reasonably well.
The total mass concentration for that measurement was
3
about 15 mg/m .
Nearly eighty percent of the particle mass lies in the
size range covered by the EAA.
Summary;
In the submicron size range the determination of aerosol
parameters by particle counting techniques has several
advantages compared with gravimetrical methods. The
instruments used cover a size range from O.D1 to 1D
microns. As these instruments were designed for atmospheric
aerosol measurements, emission aerosols have to be
diluted in most cases to fit the limits of the instruments.
Certain systematic errors based on the principle of
operation can be corrected to a great extend if necessary
by a special correction algorithm.
The comparison of the particle counting techniques with
a gravimetric method shows a rather good agreement.
-------
- 455 -
List of figures
1. Electrical aerosol analyzer
2. Optical particle counter
3. Aerosol measuring system
4. Instrument behaviour for a number distribution
5. Instrument behaviour for a volume distribution
6. Mass distribution of an exhaust aerosol
-------
- 456 -
aerosol
clean _
ai r
charger —
clean
a i r
H V
charg. current
charg. voltage
precipitator
voltage
POP 11/10
— start
— reset
— read
— signal
control logic
ampl if ier
— center electrode
-filter
ELECTRICAL AEROSOL ANALYZER
GH DUISBURG
-------
- 457 -
aerosol
clean
air
charger —
clean I
~
air
H V
chara current
charg. voltage
ipitator
voltage
POP 11/10
— start
— reset
— read
— signal
control logic
ampl if ier
— center electrode
US:
-filter
ELECTRICAL AEROSOL ANALYZER
GH DUISBURG
-------
lamp
lens system
mul ti - chan nel
analyzer
lens system /
light trap*
p ul s
t rans former
phot o-
mul t i plier
ampl i f ier
en
00
OPTICAL PARTICLE COUNTER
G H DUISBURG
-------
1. MEASURING
INSTRUMENTS 0 PC *f
2 INTERFACE parallel -serial
CIRCUITS
c onverter
i
3. DATA I/0
PROCESSING interface
0
<
4. DATA OUTPUT
3 AEROSO
p EAA CNC PMM
amplifier/ I — • U f — U
attenuator converter converter
*
floppy disc
ADC .. PD P 11/10
unit
U G 0
D E C - B U S >
U 0
plotter teletype
L MEASURING SYSTEM GH DU SBURG
-------
- 460 -
10
c
o
c
a,
o
c
o
o
01
.0
E
o
Q_
10
10
0,01
0,1
Particle Diameter Dp I |im ]
INSTRUMENT BEHAVIOUR FOR A
NUMBER DISTRIBUTION
G H DUISBURG
-------
- 461 -
c.
o
c.
a>
o
c
o
o
QJ
E
=>
o
10°
0,01
0,1
10
Particle Diameter Dp I
INSTRUMENT BEHAVIOUR FOR A
VOLUME DISTRIBUTION
G H DUISBURG
-------
--462 -
102-
E AA
-o impacfor
D OPC
101 -
o>
O.
O
o>
o
\
10°-
c
o
c
a
o
c
o
o
ut
I/I
a
a
0.
102-
0,01
I
0.1
10
Particle Diamefer Dpl|_lm]
MASS DISTRIBUTION OF AN
EXHAUST AEROSOL
G H DUISBURG
-------
- 463 -
A PARTICULATE SAMPLING SYSTEM
FOR PRESSURIZED FLUIDIZED BED COMBUSTORS
By:
William Masters, Robert Larkin, and Larry Cooper
Acurex Corporation
ABSTRACT
A particulate sampler for high-temperature, high-pressure processes has been
developed and successfully demonstrated. The system uses an extractive approach, re-
moving samples from the process stream for complete analysis of particulate size dis-
tribution, morphology, and chemical composition. System capabilities have been demon-
strated by sampling a pressurized fluidized bed combustor. This paper describes the
extractive sampling approach, the HTHP sampler design, and the data obtained from
sampling operations.
INTRODUCTION
Advanced coal conversion processes present new problems in particulate sampling,
including severe environments beyond the capabilities of conventional equipment. This
paper describes a newly developed sampling system, specifically designed for the high
temperatures and pressures found in pressurized fluidized bed combustors. The system
uses an extractive sampling approach, withdrawing samples from the process stream for
complete analysis of particulate concentration, shape, size, and chemical composition.
The capabilities of the new system have been demonstrated in- two phases of sampling
operations at a pilot-scale fluidized bed combustor owned and operated by Exxon Corpo-
ration. The first phase of testing was performed with the sample probe in its basic
configuration. The second test phase utilized modified probe internals designed to
investigate possible condensation of alkali metals. The system performed successfully
in a variety of operating modes, producing sample data from both test series.
The following sections of this paper discuss the extractive sampling concept,
the HTHF sampler design, and sampling operations that have been performed with the new
system.
Extractive Sampling
In extractive sampling, a quantity of particle-laden product gas is drawn out of
the process for analysis. Once extracted, the sample can be thoroughly examined by
conventional methods. If proper care is taken to obtain and maintain a representative
sample, the extractive approach will provide complete, accurate information on process
constituents.
The sample is typically extracted through a probe inserted into the process
duct. The sample withdrawal rate at the probe nozzle must be matched to the duct velo-
city to avoid biasing particle size distribution measurements (isokinetic sampling).
The error in measured particle content as a function of an isokinetic velocity mismatch
can be estimated analytically (see Figure 1). For fine particles at low velocities the
error is negligible, but for larger particles or high velocities, serious errors re-
sult.
Sample temperature is also a consideration in the extractive approach. Ideally,
the temperature would be maintained at process conditions during particulate separation
and analysis. In practice, however, the sample is usually cooled to temperatures com-
patible with analysis equipment.
The major advantage of the extractive method' is that the sample can be analyzed
by conventional techniques. For example, particulate removal and size classification
*Acurex Corporation has developed the HTHP sampler for the Industrial Environmental
Research Laboratory of the Environmental Protection Agency. The work is part of a
broad program investigating new sampling technology for advanced coal conversion pro-
cesses (Contract 68-02-2153). The EPA Project Officer for the contract is William
Kuykendal.
-------
- 464 -
devices, trace element collectors, and chemical analysis techniques are all highly
developed (References 1 to 5) . Extractive sampling, is commonly used in emissions
measurement and combustion studies.
Access to the pressurized duct is the main difficulty in extending extractive
sampling technology to high-pressure, high-temperature processes. The hardware re-
quirements for entering a pressurized process are much more complex than for ambient
pressure applications. The selected design for the HTHP sampler is described in the
following section.
HTHP SAMPLER DESIGN
The new sampler design adapts conventional sampling technology to high-
temperature, high-pressure environments. Key system components are:
• A traversing sample probe that can be inserted or withdrawn during process
operation
• A probe housing that contains process pressure during sampling
• A cascade impactor to both collect and size particulate -- interchangeable
with a bulk filter (Phase I tests)
• An in-stack scalping cyclone and backup filter followed by a cold" final
filter (Phase II tests)
• Conventional trace element collectors (organics trap and impingers)
• Measurement and control instrumentation to assure isokinetic conditions
A schematic diagram of the Phase I sampler is shown in Figure 2. The sample probe is
mounted within a pressure-containment housing. The probe can be inserted into the
stream through valves that connect the housing assembly to the process duct. Sample
flow in the probe passes through a cooler and particulate collector (cascade impactor) .
Flowrate is controlled by a throttle valve at the probe exit. After leaving the probe,
sample gases are conducted through the trace element collectors, and vented.
The sampling system is shown in Figure 3. In addition to the probe arid housing
assembly, controls, and sample collectors, the system includes a portable hydraulic
pump.
One of the basic decisions in designing the sampler was the choice between fixed
and translating probe configurations. A translating probe (one that is insertable and
removable during process operation) is more complex than a stationary probe, but offers
several operating advantages:
• Particulate deposition losses in the probe can be recovered
• Nozzles can be changed to maintain isokinetic conditions
• The probe can traverse the duct to measure flow variations
• Probe exposure to erosive/corrosive conditions is minimized
• Inspection and maintenance are possible during process operations
Based on these advantages, the translating probe design was selected for the HTHP sam-
pler. The sample probe and particle collector are shown in Figure 4.
Selecting the particulate collection temperature was a second major design deci-
sion. A number of well-characterized devices are available for use below 500°F, but,
high-temperature particulate collectors are in an early stage of development. Based on
this practical limitation, a collection temperature of 450°F was selected with the
awareness that possible changes in particulate composition would have to be considered.
Major changes in composition are not likely above the sulfuric acid dewpoint. However,
changes in trace element concentration are a potential concern. The HTHP sampler has
-------
- 465 -
been used in an experiment investigating the effect of collection temperature on par-
ticulate composition, as described in a later section of this paper. The impactor used
with the HTHP sampler is a Mark III, University of Washington Source Test Cascade Im-
pactor, Model D.
The cascade impactor has several advantages over other particle collectors. The
device provides many stages of size classification in a small volume. Also, impactors
classify particle size based on inertial and aerodynamic properties that relate di-
rectly to the performance of particulate removal devices. Impactor performance is well
characterized for moderate temperature, ambient pressure operation. The effect of high-
pressure on sizing performance can be estimated using theoretical correction factors.
For fine particles at moderate temperatures, even large pressure increases have little
effect on impactor performance. Collection temperature, however, can affect measure-
ments more significantly. Variations in cut size with temperatures have been calcu-
lated by the impactor supplier for temperatures up to 500°F.
In the selected sampler design, the sample probe enters the pressurized process
through 4-inch diameter full-opening valves while process pressure is contained by a
surrounding housing assembly. The housing, shown in Figure 5, consists of two tele-
scoping cylinders which move the probe into and out of the process. Hydraulic cylin-
ders connect the two parts of the housing. Their function is to accurately position
the probe, and withstand the large forces from process pressure. Sealing at the joint
between the housing cylinders is critical, so redundant seals are used. The tele-
scoping housing is the most complex part of the sampling system and consequently re-
quired the most design and development effort.
The HTHP sampling system also includes the instruments and controls necessary
for accurate sampling. Sample flowrate is one of the important parameters that is
monitored and controlled. Flow must be both isokinetic at the probe nozzle and within
the operating limits of the particle collector. For proper control, flow conditions
in both the process stream and sample probe must be measured. A pitot tube and thermo-
couple are mounted on the probe to measure process stream conditions, and a calibrated
orifice and thermocouple check the sample flow. The flowrate is adjusted to particle
collector requirements by a valve near the probe exit. Nozzle entrance velocity is
varied by selecting larger or smaller nozzles. The sampling system includes other con-
trols for sample temperature, probe traverse, trace element collector flow, and other
key operating parameters. Sy.stem controls are housed in two portable enclosures, shown
in Figure 6.
The 'trace element collection equipment included in the sampling system consists
of an organics module and impinger train (see Figure 7). Both units are identical to
those used in the Source Assessment Sampling System that is commercially available from
Acurex. The organic module cools the sample gas to 70°F and traps organic vapors in a
porous polymer granular bed. The polymer used in this test series is Rohm 5. Haas XAD-2
gas chromatographic packing material. The impinger train has four high-volume glass
impingers, three filled with oxidizing solutions and one with silica gel moisture
absorbant. The oxidizing reagents in the impingers collect volatile trace elements by
oxidative dissolution. The reagents are:
Impinger Solution
No. 1 6M H202
No. 2 0.2M (NH4)2S208 + 0.02M AgN03
No. 3 0.2M (NH4)2S2Os + 0.02M AgN03
No. 4 Silica gel
The peroxide solution in Impinger No. 1 collects reducing gases such as sulfur dioxide
which would lessen the oxidative capability of Impingers Nos. 2 and 3. The ammonium
sulfate and silver nitrate solutions serve as the trace element collectors in the
impinger train (Reference 3).
PFBC Facility
The new HTHP sampler has been demonstrated in operations at the Exxon Miniplant
PFBC. The PFBC facility is described in this section.
-------
- 466 -
The Miniplant is a pilot-scale pressurized fluidized bed combustor operated for
the EPA by the Exxon Research and Engineering Company in Linden, New Jersey. The PFBC
process, shown in Figure 8, is a combined-cycle coal combustion process. Combustion
occurs under pressure in a limestone bed that is fluidized by incoming air. Fluidiza-
.tion gives good mixing for efficient combustion, and the limestone bed removes much of
the sulfur released during the combustion process. Added useful energy can be produced
by expanding high-pressure flue gases in a gas turbine, if particulate loading can be
reduced to the levels (0.0002 to 0.002 gr/scf) required to protect turbine blades. The
Exxon Miniplant facility is being used to investigate fluidized bed combustion, gas
cleanup devices, and particulate effects on turbine components. At the time of sam-
pling, the facility did not include a gas turbine or final cleanup device.
For the sampler demonstration and condensation tests, the sampling location is
downstream of the primary cyclone as indicated in Figure 8. At this location, there
is a specially constructed duct section with a sampling port (4-inch, 300-pound pipe
flange) which interfaces with the sampling system access valves. Measured process
conditions were 1350°F and 118 psig.
The Miniplant facility is a four-story structure, with platforms at each level
(see Figure 9). The sampling location is physically located at the top of the combus-
tor tower. When installed, the probe assembly is horizontal, about 4 feet above the
platform (see Figure 10). The coolant console and hydraulic pump are also placed on
the top platform, near the probe assembly. The control consoles and gas train equip-
ment are set.up one floor below, where a partial enclosure gives some weather protec-
tion.
' Systems Operations and Test Data
The new HTHP sampler has been used in two series of operations at the Miniplant
PFBC. One series was a field test of system capabilities, the other an investigation
of the effect of sample handling temperature on particulate collection, in particular,
possible condensation effects. The sampler operated successfully in both tests series.
These operations and some of their results are described in this section.
Phase I -- Demonstration Tests
The first series of-sampling operations investigated system performance under
field conditions. These operations successfully demonstrated a variety of system
capabilities. Three sampling runs were made: one using a filter to collect total
particulate, and two using a cascade impactor for particle sizing into eight fractions.
Trace element and organic collection equipment was operated during the filter run. The
tests produced the following data:
o Particulate size distribution
• Particulate chemical composition
• Particulate shape
• Particulate concentration
• Process temperature and pressure
• Moisture content
• Structure temperatures (valves and probe housing)
• Trace element samples (not yet analyzed)
• Organic samples (not yet analyzed)
Particle size distributions from cascade impactor data are plotted in Figure 11.
The effect of pressure and temperature on impactor size cuts was estimated usina
Reference 5. At the relatively low collection temperatures used, increased pressure
had little effect on impactor performance for particles larger than 1 micrometer.
The impactor substrates are shown in Figure 12. Generally, the patterns are
regular indicating normal impactor operation. Stage 7, however, shows evidence of
-------
- 467 -
several plugged jets. The substrates from Run 2 are lightly loaded. Those from Run 3
show heavier, three-dimensional deposits.
Examples of particulate photomicrographs, showing particle size and shape, are
shown in Figures 13 and 14. These plots were made by a scanning electron microscope.
The irregular appearance is typical of flyash from lower temperature combustion pro-
cesses (Reference 8). The photos show the trend of decreasing physical size from
Stage 1 to Stage 6, although irregular shape and possible agglomeration make visual
interpretation of particle size very difficult.
The chemical composition of the collected particulate was analyzed by dispersive
X-ray fluorescence. Spectra of X-ray emissions from impactor Stage 1 and Stage 6 are
shown in Figure 15. The peaks in the spectra correspond to the number of emissions
detected at characteristic wavelengths of various elements. Results show the presence
of aluminum, silicon, sulfur, potassium, calcium, titanium, iron and copper.
Comparing the relative height of the peaks in two spectra can give a rough indi-
cation of the relative quantities of elements present in two samples. The comparison
of Stage 1 and Stage 6 spectra shows no apparent difference in bulk composition between
the coarse particles collected (050 of about 30 microns) and the finer particles (050
of about 0.6 micron).
Data from the system demonstration tests are discussed more extensively in a
test report submitted to EPA IERL (Reference 9).
Phase 7.1 — Condensation Tests
Following the system demonstration tests, a second series of sampling operations
was conducted at the Exxon Miniplant. The purpose of these tests was to investigate
the effect of sample cooling on measured particulate mass and composition. We were
specifically concerned that trace elements might condense between process temperature
and conventional particulate collection temperature (about 450°F). Of particular
interest were the alkali metals, primarily sodium and potassium. For these tests, the
sampler was set up to collect particulate at process temperature, so trace element con-
densation could be investigated in two ways. First, the trace element content of
particulate collected at 450°F (from the demonstration test series) could be compared
with the content of particulate collected at duct temperature to see if any significant
differences result. Second, after the particulate was removed at process temperature,
the sample gasses could be cooled and filtered to collect condensation products.
The probe configuration for the condensation tests is shown in Figure 16. A
scalping cyclone and a high-temperature filter are mounted on the front of the probe to
remove particulate at process conditions. After a series of choked orifices, used to
gradually reduce sample gas pressure, a final filter removes condensed material as well
as breakthrough particulate. The cyclone is a Southern Research Institute model,
designed for much less severe operating temperatures. This cyclone was readily avail-
able, was small enough for insertion into the duct, and had a very efficient 0.6-micron
cut-point. During the tests, however, the cyclone's protective gold plating blistered
and fell off, leaving titanium surfaces exposed to heavy oxidation. Chemical analysis
of the particulate samples showed significant gold contamination in the cyclone and
front filters but none in the rear filter. Titanium contamination was not evident in
any of the samples.
The high-temperature filter following the scalping cyclone is made of saffil
alumina, a material that Acurex is currently testing for high-temperature baghouse
filters. This material seems to offer excellent temperature resistance and effective
filtration, but its performance hasn't yet been fully characterized. Its performance
in the condensation tests was quite good, particularly with a two-filter "sandwich."
The estimated filter efficiency was well over 90 percent of the fine particulate passed
by the scalping cyclone.
The final filter at the sample probe exit is a standard Gelman "microquartz"
type with high efficiency and low trace element content. It is possible to use con-
ventional filter materials at this location because sample gas temperatures are sub-
stantially reduced by the probe cooler section and by sonic throttling in the orifice
section.
-------
- 468 -
Four sampling runs were completed in the condensation test series. Conclusions
have been drawn on the available data regarding condensation of trace metals, parti-
cularly the more common alkali metals, sodium and potassium.
The original intention of the Phase II tests was to compare the elemental con-
centrations found in the filtered material of the in-stack scalping cyclone and backup
filter with the material caught on the rear filter. If the system worked ideally, one
could assume all the material on the rear filter was in a gaseous state at stream con-
ditions and would thereby pass through the hot cyclone filter combination and thus be
indicative of condensation products produced within the probe. Unfortunately, the rear
(or cold) filter could not be reliably analyzed. Carryover of glass fiber filter
material during sample preparation, the small amount of the sample available on the
cold filter and additional contaminations, yielded poor detection limits with the spark
source mass spectrometer (SSMS). Moreover, the values reported for this filter were in
mass units rather than concentration units because a net filter collection w.eight was
not determined. This made any direct comparisons of cold and hot catches from Phase II
results alone difficult.
Samples of the probe wash were analyzed by Arthur D. Little, Inc., to determine
the source of the suspected contamination of the rear filter. A sample from each test
was analyzed by thermal gravimetric analysis (TGA), infrared analysis (IR), X-ray
fluorescence (XRF), and low resolution mass spectra (LRMS). Results indicated there
were three sources of contamination. They were: (1) approximately 30 percent particu-
late -- attributed to hot, in-stack filter breakthrough, (2) approximately 25 percent
sulfuric acid and sulfate condensate -- attributed to localized cold spots (measured
at 200°F) below the H2SC>4 condensation temperature, and (3) approximately 40 percent
'organics — attributed to the disintegration of packing material from a valve located
just upstream of the rear filter. The evidence of breakthrough particulate contami-
nation was supported by photomicrographs, which showed similar appearance between
material on the front and rear filters, and by dispersive fluorescent X-ray spectrum
which shows a similarity in chemical composition. The sulfur content did, however,
increase on the rear filter, apparently as a result of sulfate condensation.
An alternate approach was taken to resolve the question of trace metal conden-
sables. During the Phase I demonstration tests at Exxon, a test run was made at essen-
tially the same operating and stream conditions as the Phase II tests. A total mass
filter was used in place of the cascade impactor and was maintained near 400°F. North-
rop Services, Inc. performed a SSMS analysis of the bulk filter catch, as they did with
the Phase II cyclone and backup filter samples. Table I presents the results for these
analyses. The measured elemental .content is similar to common flyash. A partial,
nondimensionalized comparison of these two sets of results, Phase I and Phase II, is
presented in Table II. Reference quantities Fe and Mg, have been chosen to nondimen-
sionalize the results because they exist in significant concentrations in each sample
and are not likely to be present as a result of contamination, Nondimensionalizing
was done because there appeared to be a diluent in the Phase I filter catch. The Si
concentrations indicate that the filter material itself may be the diluent in the
sample. In fact, the sample analyst acknowledged some difficulty in separating the
sample from the filtering media.
Aside from the Si results, comparison of the other quantities shown indicates
that there was no detectable change in the concentration of Na or K from the hot to
cold particulate catches. This limited data would indicate that particulate collection
at low-pressure and 400°F yields accurate results for particulates.
CONCLUSIONS
The sampling system described in this paper demonstrates that extractive sam-
pling is a feasible approach for sampling high-temperature, high-pressure processes.
Furthermore, the Phase II condensation test data indicates that sample filtration at
reduced pressure and low temperature (400°F) yields accurate results for particulates.
Technology for sampling pressurized fluidized bed combustors is now developed and
available. Future development also will be required, however, to make useful appli-
cation of this technology and extend it to other advanced coal conversion processes.
One of the remaining issues for PFBC high-temperature, high-pressure sampling,
is system cost/performance trade-offs. Process developers seem to be interested in
both upgraded and downgraded versions of the sampling system. Upgraded versions offer
-------
- 469 -
longer sampling durations, quicker turnaround and better operating convenience. Down-
graded versions, such as fixed-probe designs, are cheaper, but give less information.
The next objective for extractive sampling is to develop technology for coal
gasifiers. Particulate measurement is also important for developing these processes,
and environmental difficulties are even more severe than for FPBC's.
REFERENCES
1. Lundgren and Calvert, "Aerosol Sampling with a Side Port Probe," Amer. Ind. Hyg.
Ass. J., 28:213 (1967).
2. Calvert and Parker, "Collection Mechanisms at High Temperature and Pressure,"
Symposium on Particulate Control in Energy Process, EPA-600/7-76-010, Sept. 1976.
3. Hamersma, et al., IERL-RTP Procedures Manual: Level 1 Environmental Assessment,
EPA-600/2-76-106a, June 1976.
4. Blake, D. E., Operating and Service Manual — Source Assessment Sampling System,
Aerotherm Report UM-77-80, March 1977.
5. Gooding, C. H., Wind Tunnel Evaluation of Particle Sizing Instruments, EPA-600/2-
76-073, March 1976.
6. Operation Manual, Mark III University of Washington Source Test Cascade Impactor
(Model D), Pollution Control Systems Corporation, Renten Washington, March 1974.
7. Hoke, R. C., "FBC Particulate Control Practice and Future Needs: Exxon Miniplant,"
Symposium on Particulate Control in Energy Processes, EPA-600/7-76-010, Sept. 1976.
8. Hoke, R. C., Exxon Research and Engineering Company, Linden, New Jersey, Personal
Communi'cation.
9. Masters, W. Z., "Field Testing of a Sampling System for High-Temperature/High-
Pressure Processes," Annual Report, Measurements of High-Temperature/High-Pressure
Processes, Aerotherm Report TR-77-55, July 1977.
-------
Table I. Concentration of Elements in Flyash SSMS Analysis (Partial)
Element
K
Na
Rb
Cs
Al
Si
Fe
Ca
Mg
Ti
Sr
Ba
Au
P
Cu
Zr
Ni
Cr
Pb
Cyclone
(ppm)
8,200
1,310
<70
6.7
164,000
94,000
30,000
20,000
11,400
2,430
810
710
650
276
248
160
120
<90
85
Test No.
Front
(ppm)
8,850
2,500
<68
0.23
94,000
82,600
13,400
19,000
17,800
1,950
555
694
118
223
165
140
100
<140
75
3 — Phase II
Rear
(ppm)
15.1
<135.0 *
<0.43*
<0.62
64.2
<2510.0
60.0
<6.6 *
38.0
7.1
0.6
<5.0 *
<1.4
<224.0 *
<1.5 *
<2.1 *
5.8
<13.1
<4.0 *
Rear Blank
(ppm)
5.4
91.0
0.13
--
6.4
1210
2.36
7.3
5.9
3.1
1.15
1.5
--
68.0
0.91
0.64
0.29
6.4
1.2
Phase I
Bulk Filter
(ppm)
16,260
3,560
866
8.8
Major
310,000
36,300
44,700
28,600
10,600
1,320
1,080
13
1,880
142
334
348
366
86
Notes: < - Natural background limited detection limit.
<* - Blank limited detection limit.
o
I
-------
- 47 1 -
Table II. Partial Comparison of Front and Rear Particulate Catches
from Exxon Test Series I and II
K
Fe
Na
Fe
Si
Fe
Ca
Fe
Mg
Fe
Sr
Fe
Ba
Fe
K
Mg
Na
Mg
Si
Mg
Fe
Mg
Ca
Mg
Sr
Mg
Ba
Mg
Phase I Tests
Bulk Filter
0.45
0.10
8.54
1.23
0.79
0.04
0.03
0.57
0.12
10.84
1.27
1.56
0.05
0.04
Cyclone
0.26
0.04
3.13
0.67
0.38
0.03
0.02
0.72
0.11
8.25
2.63
1.75
0.07
0.06
Phase II Tests
Front Filter
0.66
0.19
6.16
1.42
1.33
0.04
0.05
0.50
0.14
4.64
0.75
1.07
0.03
0.04
Avg.
0.47
0.12
4.65
1.05
0.86
0.04
0.04
0.61
0.13
6.45
1.69
1.41
0.05
0.05
Ratio
0.96
0.83
1.84
1.17
0.92
1.00
0.75
0.93
0.92
1.68
0.75
1.11
1.00
0.80
-------
- 472 -
1?
O
z
g
i-
-------
.Process Duct
Enclosure
(Pressure Boundary)
Control Valves
Flexible Line
rrnnnnnnnnnnnnnn
1Heater
Heat Tracing
Flow Organic
Controls Collector
Trace
Metals
Impingers
CO
I
Figure 2. System Schematic
-------
Probe Drive
Hydraulic Cylinder
Microswitches For
Transverse Control
Dowtherm Coolant
Systems And Controls
Inner Tubular Housing
Hydraulic Lines —.
Outer Tubular" Control Umbilical
Housing
Dowtherm Coolant
Supply And Return
Sample Line
Hydraulic
Control Console Supply System
Control Valve
And Operator
Figure 3. High-Temperature, High-Pressure Sampling System
-------
- 475 -
Transducers
& controls
Impactor
stacks
Heated
transport
tube
13
n
Pitot
tube
Nozzles
Figure 4. Exploded view of HTHP probe.
-------
a
in
s
J
-
Figure 5. Aerotherm HTHP sampling probe and duct interface valve.
-------
-Q
UD
ro
C
I
I
j
I
Figure 6. Control consoles.
-------
rganic \ Impingers
odule
Oven
.;
r -
-•
:
I
1
c
Figure 7. Flow Control Oven and Gas Train
-------
Coal and'
make-up
sorbent
Boiler
Figure 8. Pressurized Fluidized Bed Coal Combustor System
-------
- 480 -
o
i
Figure 9. Miniplant PFBC.
-------
-O
C\J
9
I
i
0
—
!
Figure 10.
HTHP probe assembly
installed at Exxon miniplant.
-------
- 482 -
PERCENTAGE UNDERSIZE (BY WEIGHT)
20 30 40 50 60 70 80
90
E
a.
LU
t-
LU
5
Q
m
O
oc
Q_
98
90
80
70 60 50 40 30
PERCENTAGE OVERSIZE
20
Figure 11. Particle Size Distribution
-------
- 483 -
EXXON-HTHP
4/1/77
i
•
:
••:••... I.Y.*
v •••••»•
*•...«* -^
STAGE 5
EXXON-HTHP
4/1/77
Figure 12. Impactor substrates.
-------
- 484 -
1000X
10 Microns
3000X
3 microns
Figure 13.
Particle photomicrographs Stage 1
-------
- 485 -
3000X
10000X
1 micron
Figure 14.
Particle photomicrographs Stage 6.
-------
- 486 -
-------
Normal
/-;-.
Cooler
Impactor
fl
Process Flow
Condensation Test
Process Flow
Sample Flow
(Or Filter) Throttling
Valve
Scalping 1350° F
Cyclone Filter
r~
vj~ j —
I
I
I
Cooler
Choked 400° I
Orifices Filtei
i
i
*
—
Sample
Flow
I
00
I
Figure 16. Probe Configuration
-------
- 488 -
Discussion
In reply to Dr. Holighaus, Dr. Cooper said that the difficult techniques
described in his talk could not be used in a fluidized bed, which was
characterized by chaotic motion. Mr. Kuykendal asked if a sample after
the first cyclone were possible, and Dr. Cooper answered that particulate
loadings were heavy and that a series of impactors would become overloaded.
Impactors could be replaced by cyclones, however.
Work on fluidized bed combustion would be transferred in the future from
EPA to the Department of Energy. Inside probes on the fluidized bed
would be an interesting, but expensive, basis for West German/US cooperation.
Such work could eventually encourage the production of new equipment.
Mr. Bonn requested details of the measurement technique's influence on
the fly ash, adding that impactors could break up the granules. Dr. Cooper
said evidence of fracture and agglomeration of particles existed. A
more comprehensive program using various sampling techniques was needed.
Mr. Princiotta mentioned that a two cyclone series owned by EPA came
close to providing the total required protection. In answer to Dr. Holighaus
Dr. Cooper said that the type of measuring equipment to be used with
fluidized beds had not been determined at the time the study was initiated.
-------
- 489 -
Final Discussion
Dr. Cooper, seconded by Mr. Princiotta, thanked the KFA for its
hospitality.
Possibilities regarding West German/US Co-operation were then
discussed. Mr. Parker raised the question of West German involv-
ment in the fluidized bed combustion program. Dr. Holighaus said
that DM 2OO million could be spent on the project, and drew atten-
tion to the advanced investigations on fluid bed combustion and
gas turbines at the Bergbauforschung. Feasibility of combustion
using existing equipment would be tested (end of 1979). The main
interest was in the pressurized bed for power stations. Atmospheric
beds were applicable to smaller industries. Three projects would
take place:
6 kcl p.h. steam generated
3O MW Combination of atmospheric fluid bed and
a normal power station
low quality coal would be used in 1) and high
quality in 2)
An extensive project dealing with 2OO MW of electricity was planned
for 1981. Also, a joint pressure/fluidized bed project was being
carried out in co-operation with the United Kingdom. Other work,
financed by the German government and companions, was also being
carried out. Two-thirds of the funds came from the public, and a
third from the companions. Outside technology would not be used,
nor did a project exist at the time dealing with SO^ additives to
electric precipitators.
Mr. Gtithner mentioned that one power station had a license to in-
troduce SO3. Obtaining agreements with the authorities to carry
out such work constituted a problem. Use of US produced additives
was then discussed. Herr Kastner mentioned the abandonment of a
scheme to add NH3 at two power plants, owing to trouble with fly-
ash recycling.
Dr. Holighaus added that all additives could be tested.
Mr. Princiotta referred to a combined wet and dry scrubber program
which had been discussed six months before the workshop; no further
developments had been made in this direction. Mr. Kastner pointed
out that using a wet scrubber for flyash collection was disad-
vantageous, as it was limited to this function and had few discarding
facilities.
-------
- 490 -
Dr. Holighaus stressed the need for US/West German co-operation
regarding measurement techniques. More information on new deve-
lopments would be available in August 1978.
The Umweltbundesamt will provide reports which would be available
to EPA.
The workshop ended with vote of thanks for these who had read
papers.
-------
- 491 -
List of Participants
1. Mr. Robert Donovan
Research Triangle Institute
Research Triangle Park, N.C.
2. Dr. Richard Parker
Air Pollution Technology, Inc.
San Diego, California
3. Dr. John Gooch
Southern Research Institute
Birmingham, Alabama
4. Mr. Dale L. Harmon
Particulate Technology Branch
Industrial Enviromental Research Laboratory
Enviromental Protection Agency
Research Triangle Park, N.C.
5. Mr. Seymour Calvert
Air Pollution Technology, Inc.
San Diego, California
6. Mr. Charles Gooding
Research Triangle Institute
Research Triangle Park, N.C.
7. Mr. Joseph McCain
Southern Research Institute
Birmingham, Alabama
8. Mr. Michael Shackleton
Aeortherm-Acurex Corporation
Mountain View, California
-------
- 492 -
9. Mr. William Kuykendal
Process Measurements Branch
Industrial Enviromental Research Laboratory
Enviromental Protection Agency
Research Triangle Park, N.C.
1O. Mr. Larry Cooper
Aerotherm-Acurex Corporation
Mountain View, California
11. Dr. Steven Gage
Assistant Administrator
for Research and Development
Enviromental Protection Agency
Washington D.C. 2O46O
12. Mr. Frank Princiotta
Director of Energy Processes Division
Particulate Technology Branch
Research Triangle Park, N.C.
13. Dipl.-Ing. G. Helsper
Institut fur Aerosolmefttechnik
Gesamthochschule Duisburg
Bismarckstr. 81
D-4OOO Duisburg 1
14. Dr. Laufhiitte
Saarbergwerke AG
Postfach 1O 3O
6600 Saarbrucken
15. Dr. Kelleter
Kernforschungsanlage Jiilich GmbH
Postfach 1913
517O Jiilich
-------
- 493 -
16. Dr.-Ing. P. Walzel
Institut fur Apparatetechnik
Universitat Essen
Unionstr. 2
43OO Essen
17. Prof. Dr.-Ing. E. Weber
Institut fur Mechanische Verfahrenstechnik
Universitat Essen
Unionstr. 2
43OO Essen
18. Dr. rer. nat. K. Hiibner
Institut fur Mechansiche Verfahrenstechnik
Universitat Essen
Unionstr. 2
43OO Essen
19. Dipl.-Ing. H.-G. Pape
Institut fur Mechanische Verfahrenstechnik
Universitat Essen
Unionstr.2
43OO Essen
2O. Dipl.-Ing. R. Schulz
Institut fur Mechanische Verfahrenstechnik
Universitat Essen
Unionstr. 2
43OO Essen
21. Dipl.-Phys. H. Wiggers
Institut fur Mechanische Verfahrenstechnik
Universitat Essen
Unionstr. 2
43OO Essen
-------
- 494 -
22. Dr.-Ing. P. Davids
Umwe 11 b un de s amt
Bismarckplatz
1OOO Berlin 33
23. Dipl.-Ing. G. Giithner
Umweltbundesamt
Bismarckplatz
1OOO Berlin 33
24. Dr. Reissmann
Lurgi Umwelt und Chemotechnik GmbH
Postfach 11 91 81
60OO Frankfurt/M. 2
25. Dr. Ziegler
Bundesministerium fur Forschung
und Technologie
Postfach 2O O7 O6
53OO Bonn 2
26. Dr. Dr.-Ing. H.J. Stocker
Kernforschungsanlage Jiilich GmbH
Projektleitung Energieforschung
Postfach 1913
517O Jiilich
27. Dr.-Ing. R. Holiqhaus
Kernforschungsanlage Jiilich GmbH
Projektleitung Energieforschung
Postfach 1913
517O Jiilich
28. Frau D. Ermisch
Kernforschungsanlage Jiilich GmbH
Projektleitung Energieforschung
Postfach 1913
517O Jiilich
-------
- 495 -
29. Dipl.-Ing. Grabenhorst
Kraftwerk Siersdorf GbR
Roermonderstr. 63
512O Herzogenrath-Kohlscheid
3O. Dr.-Ing. Finkh
Kraftwerk Union
Hammerbachstr. 12/14
852O Erlangen
31. Dr. rer. nat. Schilling
Bergbau-Forschung GmbH
Franz-Fischer-Weg 61
430O Essen 13
------- |