OCR error (C:\Conversion\JobRoot\00000CD6\tiff\20013I64.tif): Unspecified error
-------
DRAFT DISCLAIMER
The High Temperature/High Pressure Particulate Control
Symposium has been published under the auspices of the Depart-
ment of Energy (DOE). However, the symposium itself was spon-
sored by the Energy Research and Development Administration (ERDA)
in September, 1977, prior to ERDA's reorganization into DOE on
October 1, 1977. It should be recognized that the titles/
positions of the speakers reflect past ERDA organization and
may not reflect the present staffing at DOE.
-------
EPA-600/9-78-004
CONF-770970
UC- 11
c
EPA/DOE Symposium on
High Temperature High Pressure
Particulate Control
Sponsored By
U.S. Department of Energy
Assistant Secretary for
Energy Technology
Fossil Energy Program
U.S. Environmental Protection
Agency
Office of Research and Development
Industrial Environmental Research
Laboratory
Research Triangle Park, N.C.
vvEPA
September 20—21, 1977
Washington, D.C.
-------
INTRODUCTION
Advanced technology methods of using coal include pressurized fluidized
bed combustion and coal gasification. These technologies have been under de-
velopment for several years and have reached a level of development more
advanced than the particle control technology required for its use. Therefore,
a potential barrier to the utilization of advanced coal combustion technologies
is the need for particulate control at high temperatures and pressures.
The development of a particle removal system capable of operating reliably
at high temperatures and high pressures (HTHP), and able to meet the stringent
cleanup requirements needed to assure acceptable turbine life is a formidable
task. To accomplish this task will require high mass removal efficiency by
particle control equipment which must operate in a severe environment. In
addition, this particle control equipment must not be so costly as to destroy
the economic advantages of the new coal combustion processes.
Development of high temperature and pressure cleanup devices in parallel
with the coal conversion process represents an opportunity to integrate this
technology into the process in a manner which can benefit the process economics
and performance while minimizing damage to the environment. In addition, the
research and development efforts needed to accomplish successful development
of HTHP particle control devices will stimulate improvements in conventional
particle control technology which may result in improvements in ambient air
quality.
The "EPA/ERDA Symposium on High Temperature/Pressure Particulate Control"
was held in Washington, D. C. September 20 and 21, 1977. A primary objective
of the symposium was to promote an interchange of ideas among researchers on
the HTHP problem. This proceeding is a compilation of papers presented at the
symposium.
Ill
-------
Sponsors of the symposium were: the Particulate Technology Branch of the
U. S. Environmental Protection Agency (Research Triangle Park, North Carolina)
and the Fossil Energy Division of the U. S. Energy Research and Development
Administration (Washington, D. C.)«
Aerotherm Division of Acurex Corporation (Mountain View, California)
coordinated and hosted the meeting.
Dr. Dennis C. Drehmel from the EPA and Mr. Jack Seigel from ERDA deserve
credit and my thanks for their help in organizing the meeting. I wish to
thank each of the Authors as well for their efforts and cooperation which made
the symposium and their proceedings a successful presentation of the current
state of the art in high temperature and pressure particle control.
Mike Shackleton
Conference Coordinator
Aerotherm Division
Acurex Corporation
IV
-------
TABLE OF CONTENTS
Page
Keynote Address: Dr. Philip C. White, Past Assistant Administrator
For Fossil Energy, ERDA 1
A Review of the Symposium: Alan A. Smith, Babcock & Wilcox, Ltd.,
England 7
SESSION I: FUNDAMENTALS
"The Tolerance of Large Gas Turbines to 'Rocks', 'Dusts', and
Chemical Corrodants," E. F. Sverdrup, D. H. Archer, M. Menguturk ... 13
"Materials For Use in High Temperature/Pressure Hostile Environ-
ments," J. Hull 33
"High Temperature and Pressure Effects on Particle Collection
Mechanisms," R. D. Parker, S. Calvert, D. Drehmel 61
SESSION II: FILTRATION
"Granular Bed Filter For Particle Collection At High Temperature
and Pressure," S. Yung, R. D. Parker, R. G. Patterson, S. Calvert,
D. C. Drehmel 89
"Evaluation of a Granular Bed Filter, For Particulate Control in
Fluidized Bed Combustion," R. C. Hoke, M. W. Gregory Ill
"Performance and Modeling of Moving Granular Bed Filters,"
G. L. Wade i 3j
"Ceramic Fabric Filtration at High Temperatures and Pressures,"
M. Shackleton, J. Kennedy 193
"High Temperature Fine Particle Control Using Ceramic Filters,"
D. C. Drehmel, D. F. Ciliberti 235
"Problems of Gas Purification Occurring in the Use of New Techno-
logies For Power Generation," E. Weber 249
-------
TABLE OF CONTENTS (Concluded)
Page
SESSION III: OTHER COLLECTION DEVICES
"High Temperature, High Pressure Electrostatic Precipitation,"
P. Feldman, J. Bush, M. Robinson 281
Pulse-Jet Acoustic Dust Conditioning in High Temperature/Pressure
Applications," D. S. Scott, W. M. Swift, G. J. Vogel 309
"The Application of Sonic Agglomeration for the Control of Particu-
late Emission," D. T. Shaw, J. Wegrzyn 325
"Cyclocentrifuge Development For Particulate Control, Phase I:
Feasibility Study," J. T. McCabe 355
"Fine Particle Collection Efficiency in the A P.T. Dry Scrubber,"
S. Calvert, R. G. Patterson, D. C. Drehmel 399
"Hot Gas Clean-Up By Particle Entrainment in Coal Slag Based
Glasses," L. R. McCreight, A. Gatti, H. W. Rauch, M. J. Noone 415
"Molten Salt Scrubbing For Removal of Particles and Sulfur From
Producer Gas," R. H. Moore, G. F. Schiefelbein, G. H. Stegen, D. G. Ham 429
SESSION IV: PARTICLE SAMPLING AND MEASUREMENT
"Particulate Sizing in High-Temperature, High-Pressure Combustion
Systems," H. W. Coleman 467
"Particulate Sampling at High-Temperature and High-Pressure —The
Extractive Approach," W. Z. Masters 495
"Particle Field Diagnostic Systems For High Temperature/Pressure
Environments," J. D. Trolinger, W. D. Bachalo 527
"On Line Particulate Analysis on a Fluidized-Bed Combustor,"
E. S. VanValkenburg, H. N. Frock 557
"Characterization of Suspended Flue Gas Particle Systems With On-Line
Light Scattering Particle Analyzers," J. C. Montagna, G. W. Smith,
F. G. Teats, G. J. Vogel, A. A. Jonke 579
-------
KEYNOTE ADDRESS
By:
Dr. P. C. White
Assistant Adminstrator for Fossil Energy
Department of Energy
Washington D.C. 20545
-------
SPEECH PREPARED FOR DELIVERY BY
DR. PHILIP C. WHITE
ASSISTANT ADMINISTRATOR FOR FOSSIL ENERGY
AT THE
SYMPOSIUM ON HIGH TEMPERATURE
HIGH PRESSURE PARTICULATE CONTROL
WASHINGTON, D. C.
SEPTEMBER 20, 1977
Welcome to this Symposium, which is being held to summarize ERDA and EPA-
funded work in hot gas particulate cleanup. Past meetings convinced us that
such an exchange would be worthwhile—because the implications of combined-
cycle development on our nation's energy future are potentially so far-reaching.
As we head into a time of energy shortages and the implied environmental prob-
lems, we especially need technologies that will provide clean, efficient energy.
The President, in his National Energy Plan, has outlined a course that
will help us provide such energy, including the increased use of coal. While
coal makes up 90 percent of conventional U.S. energy reserves, it supplies only
18 percent of the energy we use.
One of coal's most important energy uses is to generate electricity. Yet
in recent years electric power generation from coal has declined. We want to
reverse that trend. As you know, to save dwindling supplies of oil and gas,
FEA, under the authority of the Energy Supply Environmental Coordination Act
of 1974, ordered power plants to burn coal for steam generation. Along those
lines, the NEP, through imposing taxes on the use of oil and natural gas,
encourages industry and utilities, to switch back to coal. Soon new industries
and utilities will be prohibited from burning those two scarce fuels, except
under exceptional circumstances. By 1990, nearly all utilities will be pro-
hibited from burning gas.
While coal is our most abundant fossil resource, it is also our dirtiest,
and its increased use presents environmental problems. To ensure that the new
Department of Energy gives adequate weight to the environment, an Assistant
Secretary for the Environment within the new Department will oversee environ-
mental research and monitoring for Federally-funded energy technology
-------
development programs. In ERDA we have started Environmental Development Plans
(EDPs) to coordinate the technological development of new energy technologies
with related environmental research. The EDP is ERDA's means to ensure that
in decision-making, the environment receives the same consideration as do
economics and technology.
Of course, our role in Fossil Energy will be key to ensuring that coal is
used in an environmentally acceptable way. No fossil energy technology will
be commercialized unless it is environmentally acceptable.
This Symposium focuses on a technical problem of great importance and
great difficulty—one that affects both energy production and environmental
quality: the cleanup of gases from burning and converting coal.
If we can gasify coal and then pass hot off-gases directly into an advanced
turbine to generate electricity, we can gain several percentage points in the
efficiency of the power generation. This means we need less coal, less mining,
and lower cost all along the operation. We can only charge those gases di-
rectly to the turbine however, if we can have a very high efficiency in the
particulate removal of the gases coming from the gasifier.
Hot gas cleanup is also important to certain technologies in which parts
of unburned or unconverted coal exit to the atmosphere with the gas. To have
a more efficient and economical system, we must collect and recycle this coal
or reinject it for more complete conversion or combustion.
In ERDA Fossil Energy we are researching five coal technologies for which
hot gas particulate removal is especially important. These technologies are
coal gasification for both low and high Btu gas, atmospheric and pressurized
fluidized bed combustion (AFBC, PFBC) and advanced power systems.
The pressurized fluidized bed combustion and low Btu gasification tech-
nologies are those in which the hot combustion gas must be cleaned of partic-
ulates before entering the turbine.
In high Btu gasification and atmospheric fluidized bed combustion, the
recycling or reinjection of portions of the coal exiting with the gas insures
more complete combustion or conversion.
-------
The advanced power systems program is researching the turbines themselves,
both open and closed-cycle. The program addresses turbine design and material.
If we are successful, the open-cycle turbine could be fired with a high-
•
temperature, high-pressure coal-derived gas.
For all of those technologies, the objective is to clean the gas of par-
ticulates without lowering its temperature and pressure. By cooling the gas,
there is at least a partial waste of energy that could be used in the turbine
or in the next step in the process, resulting in lowered efficiencies and
increased costs. For instance, low Btu gas can be cooled and cleaned using
conventional devices, but this would not be economical or efficient if the gas
is to be used in an advanced gas turbine.
Our research on gas turbines has reached a state where particulate con-
trol has gained the spotlight as a major impediment to combined cycle develop-
ment .
To clean the hot gas of particulates without lowering its high temperature
and pressure, we can focus our research on three general areas. First, we can
determine the quality requirements of the turbine, that is, the degree of
abrasion and corrosion the turbine can withstand, which tells us to what level
we have to clean up the gas. Second, we can characterize the raw gas intended
for use in the turbines in terms of its physical makeup—by this I mean the
size distribution of the particles and the temperature and pressure of the gas,
and its chemical makeup—what chemicals are present and in what form. Finally,
we can zero in on collection devices to clean the raw gas to the level which
the turbine requires. Here we are researching the various options for gas
scrubbing, collection techniques and component materials. This Conference re-
flects EKDA's and EPA's interest in all three areas.
Research in these areas addresses a number of interrelated unknowns, making
progress particularly challenging. Gas sampling and measurement equipment is
as prone to erosion and corrosion as the turbine itself, making it difficult to
sample or treat the gas to determine what is damaging the turbine. We need
data on the performance of collection systems under the high temperature, high
pressure conditions involved in advanced power systems. Usually the data are
collected during experimental work performed on a bench-scale, or via computer
-------
modeling; thus, it is difficult to scale up results to the larger dimensions
which we will encounter at pilot or commercial scale. Since problems involved
in hot gas cleanup are often interrelated, the results can affect the direc-
tion of work on the entire energy system. That is why a symposium of this sort
is particularly useful.
The advantages of overcoming the problems involved with high temperature,
high pressure particulate control are many. Once developed, we could apply
these controls to other processes, such as metallurgical ones, where heat re-
covery is not now economical because of particulate problems, and where we
need expensive machinery to cool the gas to a temperature compatible with con-
ventional gas cleanup devices. Hot gas cleanup also would improve the effi-
ciency of technologies such as atmospheric fluidized bed combustion and high
Btu gasification.
But especially in electricity production, the advanced gas turbine com-
bined with a steam-powered generator will produce economic and environmentally
clean energy. Open cycle gas turbines are now used as peaking load units in
power plants, but are fired with natural gas or oil, both virtually particulate-
free. The advanced gas turbine, coupled with advanced coal technologies that
on their own are clean and efficient energy systems, will offer an improved
conversion efficiency of up to 45-50 percent, a substantial gain (10-15 per-
cent) over that of conventionally coal-fired utility boilers. This approach
could provide mid- and baseload power in the future. Other technologies, such
as magnetohydrodynamics, could exploit the results of advanced power systems'
work related to the use of topping and bottoming cycles. If these advanced
technologies are developed, the electricity needed in the coming decades could
be produced at higher conversion efficiencies, lower costs, and with fewer
adverse environmental impacts than those offered by today's conventional
utility steam boilers.
EPA AND ERDA PROJECTS
Complementing the ERDA-sponsored research on hot gas cleanup, EPA has
funded extensive research on particulate control devices, focusing on electro-
static precipitators, scrubbers, and filters. EPA is continuing to evaluate
-------
techniques to enhance collection efficiencies at high temperatures, in partic-
ular, through charging and precharging of high resistivity dust.
EPA is working with ERDA on several environmental assessments and is par-
ticipating in EKDA's environmental sampling and monitoring programs for both
the fluidized bed combustion and low Btu gasification programs. EPA is repre-
sented on ERDA task forces and will review the environmental development plans
for ERDA technologies. We will discuss many of these ERDA and EPA studies at
the Symposium.
We are working together at ERDA and EPA to perfect hot gas cleanup tech-
niques applicable to advanced power systems and other technologies. Our joint
efforts will hasten the development of the advanced gas turbine as a major new
method of producing electricity using coal as the energy source. Such advanced
power systems will provide the clean, efficient, economical energy from coal
on which our National Energy Plan is based.
As we strive to continue the availability of dependable, clean energy
supplies, we have to find the answers to these problems. Coal is our most
tangible and abundant energy resource. Finding ways to curb the potentially
severe environmental impacts associated with its use are mandatory. Our work
at ERDA and EPA on high temperature, high pressure particulate removal is a
critical stage in our movement toward combined goals of more efficient energy
use and a quality environment.
-------
A REVIEW OF THE SYMPOSIUM
By:
A. A. Smith
Babcock & Wilcox, Ltd.
England
-------
A REVIEW OF THE SYMPOSIUM
By
Alan A. Smith
Babcock & Wilcox Ltd., England
The subject of particulate control in a high pressure, high temperature
environment has resulted in the presentation at this Symposium of many papers
of a highly technical and scientific content. This was to be expected in the
current atmosphere of appraisal of the utilisation of energy resources. Partic-
ulate control, or clean-up of combustion gases, is of particular significance
in respect of the application of the new processes being developed for electri-
cal power generation.
I believe it opportune to make this unrehearsed and unprepared statement
from the point of view of a major contractor in the energy business.
I have come to this Symposium from my company, Babcock & Wilcox Ltd. , of
the UK, together with our Consultant, Dr. C. J. Stairmand, with the objective
of making a necessary assessment of the state of the art in respect of partic-
ulate collection.
We in BW have a vested interest in this technology since we already have
several contracts in North America in the field of atmospheric and pressurised
fluidised combustion. Some of you may be aware of the work carried out in the
UK by BW and our partners, the National Coal Board, and also by our licensor,
Combustion Systems Ltd.
You may also be aware of the work currently being carried out by BW and
our partners, Stal Laval of Sweden, for American Electric Power. Particulate
collection and control is therefore, for us, very much a live and ongoing re-
quirement .
Whilst this has been a very interesting Seminar and a great deal of know-
hew and experience has been made public, I must confess that it is disappointing
8
-------
to find that although considerable progress is being made, it will evidently
be quite some time before the necessary hardware comes on to the market. That
is, available to a main contractor at a firm price with performance guarantees,
and on a commercial scale. By commercial scale I mean for power generation
plants of say 200 MW'capacity. I think we must accept that the next generation,
say 1,000 MW plus are not likely to appear for another decade.
I think it was the Keynote speaker who told the story of the airline pilot
who announced to his passengers that although lost, he was nevertheless 20
minutes early. I feel bound to question—are we lost and 20 minutes late?
On a constructive note may I suggest that, in addition to continuing the
work on development of high temperature, high pressure collection-equipment,
we should perhaps be encouraging turbine manufacturers to develop a new breed
of turbines.
Of course they would like the same clean gas from the coal based process
that they get from combusting say distillate. This is what you are trying to
give them.
Our studies have indicated that to give them this, we shall need to in-
stall huge complex (and currently unproven at the commercial scale) clean-up
equipment costing many millions of dollars. I think this has been verified at
this Symposium.
The turbine manufacturer could make a big contribution to getting this
process off the ground, and thus allow us to utilise the low grade fuels which
utility companies want to employ, by relaxing their standards. I appreciate
and understand their problems. However, with changing load pattern on a com-
mercial scale plant there will surely be periods anyway when they may have to
accept a relatively dirty gas if only by virtue of the combustion process.
Is blasting nutshells through the turbine the only possible solution? We
hear talk of possible self cleaning turbines. Well where are they? Is anybody
doing anything about it?
Although the subject of particulate collection has many industrial appli-
cations, I believe the one of most current significance is that related to
power generation, in particular applied to gas turbines.
-------
This will still be the case with combined cycle plants where the bulk of
the energy generation is achieved by producing steam.
Protection of the environment can always, as a last resort, be accomplished
by conventional equipment—for example electrostatic precipitators or bag fil-
ters operating at conventional temperatures after the exhaust gases have been
cooled through heat recovery systems.
It would appear then that there are two routes to be followed in realising
the goal of how to burn low grade fuels and recover the energy in power genera-
tion via gas turbines. One is through the subject of this Symposium and the
other through a redevelopment of the gas turbine. No doubt the final solution
will be a compromise because each possibility imposes complex technical prob-
lems.
I have no doubt we will get there, the immediate problem is when.
10
-------
SESSION I:
FUNDAMENTALS
DENNIS C. DREHMEL
CHAIRMAN
11
-------
THE TOLERANCE OF LARGE GAS TURBINES TO "ROCKS",
"DUSTS," AND CHEMICAL CORRODANTS
By:
E. F. Sverdrup, D. H. Archer, M. Menguturk
Westinghouse Research & Development Center
Pittsburgh, PA 15235
13
-------
THE TOLERANCE OF LARGE GAS TURBINES TO "ROCKS",
"DUSTS", AND CHEMICAL CORRODANTS
By:
E. F. Sverdrup, D. H. Archer, M. Menguturk
Westinghouse Research & Development Center
Pittsburgh, Pa. 15235
An order of magnitude estimate of the particle concentra-
tion that can be allowed in the gas flowing through large
industrial gas turbines for combined-cycle electric utility
service can be made. The estimate is based on theoretical
models that predict the rates of arrival of submicron par-
ticles on the blade surfaces and assumptions about the
quantity of material that will stick. The estimate is also
based on calculations of the trajectories of micron sized
particles and estimates of particle impact damage to turbine
blading. Hard data, taken under conditions simulating
turbine operation is needed to calibrate these models.
Cleaning of the turbine expansion gas to 0.002 grains/scf
with particles larger than six microns in diameter effectively
removed is our best current guess as to the turbine expansion
gas cleanliness required. Strict attention must also be
given to controlling sodium compounds to about 50 parts per
billion by volume in the expansion gas.
14
-------
SECTION 1
INTRODUCTION
BACKGROUND
Gas turbine generators used by electric utilities for meeting peak
power demands have been fired with "clean" fuels - distillate fuel oil
or natural gas. Erosion and deposit formation in these turbines has not
been a problem. Hot corrosion has been controlled by limiting alkali
metal concentrations in the fuel oil to between 0.25 and 0.5 ppm by
weight. Formation of sodium sulfate melts which would rapidly corrode
the blading has been prevented by this limitation. In those turbines
burning oils having vanadium contents, hot corrosion is controlled with
fuel additives such as magnesium oxide and silica. These additives
react with the vanadium compounds to form solids and prevent aggressive
vanadium based melts from forming on the blading.
The use of large gas turbines in combined cycle power plants fired
with coal-derived fuels is being considered to provide for electric
utilities' "intermediate-load" requirements. These combined-cycle
plants offer the potential of higher cycle efficiencies at costs com-
petetive with conventional steam plants with stack-gas cleaning to meet
air quality standards. In coal fueled combined-cycle plants, the gas
turbines expand a gas containing a new mix of aggressive chemicals and
particles. Reliable and economic methods must be devised to achieve
sufficient gas cleanliness, without degrading system performance to the
extent that the combined-cycle plant loses its advantages over a conventional
steam plant burning the coal directly.
How Clean is Clean Enough?
We are concerned with three interacting tolerances: (1) to control
corrosion of turbine parts, (2) to control the build-up of deposits in
the flow path, and (3) to control erosion of the turbine parts by large
particles. The gas cleaning system must control the concentrations of
molecular sized chemical contaminants, sub-micron "dusts", and micron
sized "rocks".
15
-------
SECTION 2
CONTROL OF HOT CORROSION
THE TOLERANCE FOR CHEMICAL CONTAMINANTS
Molten sodium sulfate must be prevented from remaining on the tur-
bine blades for any appreciable length of time. These melts corrode
turbine blading and act as glues to cement deposits. Chemical thermo-
dynamics can be applied to establish the concentrations of gaseous
sodium compounds which can be permitted above the blade surface. Fig-
ure 1 shows the reactions controlling the formation of sodium sulfate
which melts at about 880°C (1620°F) and which can form on portions of
the first and second stage blading of a modern turbine. Three processes
are involved: (1) direct volatilization of sodium sulfate - this is
not a very important process because of the sulfates1 low volatility at
blade temperatures, (2) reactions with water vapor and sulfur oxides to
form volatile sodium hydroxide - a process important in oil and coal
fired turbines, and (3) reactions with hydrogen chloride to form
volatile sodium chloride - this reaction is important in turbines expand-
ing gases derived from coal because of the appreciable chlorine contents
of coal.
Figure 2 presents the calculated maximum sodium concentrations
which can be permitted in the turbine expansion gas while preventing a
liquid sodium sulfate deposit from remaining on the blading. The
allowed concentration is a function of the chlorine and sulfur oxide
concentrations in the gas. In an oil fired turbine the chlorine con-
centration in the expansion gas is negligible and this simple chemical
equilibrium model predicts an allowable sodium concentration of about
50 parts per billion by volume in the expansion gas. (An oil fired tur-
bine consuming about 16 pounds per second of fuel oil containing 0.5 ppm
by weight of sodium and burning it with air will produce a turbine
exhaust flow of about 760 Ib/sec of combustion gases having a sodium
concentration of about 80 parts per billion by volume.)
Although the primary concern of this conference is participates, it
is important to recognize that chemical contaminants which can form
aggressive melts must be strictly controlled.
16
-------
SECTION 3
CONTROL OF DEPOSITION
TOLERANCE FOR SUB-MICRON PARTICLES
Among the many factors that affect the rate of arrival of sub-micron
size particles at the turbine blade surface are the concentrations of these
particles in the gas stream and the thickness and turbulence levels in the
boundary layers which are present on the blading. These arrival rates
have been calculated in two dimensional turbine flows where the main stream
gas flow is highly turbulent and where there is both turning of the main
stream and strong pressure gradients in the direction of flow. Three
different mechanisms are controlling the rates of particle arrival:
(1) molecular diffusion, (2) the changes in the intensity of the gases'
turbulent fluctuations caused by the braking action of the blade surface
on the fluid flow, and (3) the particle's size and inertia which control
its response to the turbulent velocity fluctuations in the gas. The struc-
ture of the boundary layer gas flows is determined by the turbine's
flowpath design and is calculated. The deposition program then calcu-
lates the arrival rates of the particles as a function of their size at
the various locations through the turbine stage. By combining the cal-
culated arrival rates with the concentrations of each size fraction of
particles at the turbine inlet, the rate of deposit formation can be
calculated provided one knows how much of what arrives sticks. Because
we do not have either a theoretical model nor experimental data to predict
deposit growth under conditions relevant to a coal fueled turbine to
determine how much of the arriving particulate will stick, we can predict
the rate of deposit growth and the consequent effects on turbine perfor-
mance only by assuming how much of the arriving particles will stick
on the surface.
If we assume that everything that arrives at the blade surface that
is larger than one micron bounces off without removing surface material
or deposition - and everything that is smaller than one micron sticks,
and if we further assume that what sticks consolidates into a film having
the particle density of 2.5 grams/cm^, then Figure 3 presents the rates
of deposit build up through the first stage of a large gas turbine for a
particle concentration of 0.002 grains/scf entering the turbine having
the particle size distribution shown in Figure 4. (This particle size
distribution is similar to that measured after gas cleaning with fabric
filters).
17
-------
If everything smaller than one micron diameter arriving at the tur-
bine blade is sticking, Figures 3 and 5 show that first stage stator
passages constrict with a deposit growing 0.4 cm thick on both its
pressure and suction surfaces in 1000 hours of operation. If only one
tenth of what arrives sticks, it would take 10,000 hours for the same
deposit thickness to develop.
How does the growth of this deposit effect power developed by the
turbine? Preliminary calculations show that the effect of the changing
blade profiles on the efficiency of the turbine stage are small, (assum-
ing that the rapid arrival of material at the vane nose is mostly blown
away) but the choked passage restricts the mass flow through the turbine.
The power developed would be expected to drop off with time as shown in
Figure 6. A five percent reduction in turbine stage power would be
expected after 250 hours of turbine operation with everything sticking
or after 2500 hours of turbine operation with only 10% sticking. Clean-
ing of the turbine would then be required.
It would appear that, depending on the actual sticking fraction,
the concentrations of sub-micron particulate that can be permitted at
the inlet to the turbine to control deposition lie in the range 0.002
to 0.02 grains/scf.
The amount of fine particulate that actually will stick depends on
the gas chemistry, the choice of turbine inlet temperature, and the
blade metal temperatures maintained by the turbine blade cooling system.
18
-------
SECTION 4
CONTROL OF EROSION
THE TOLERANCE TO MICRON SIZE PARTICLES
Just as it is possible to calculate the arrival rate of small par-
ticles on the turbine surfaces, a computer model is available to cal-
culate the trajectories of the particles through a turbine stage and to
determine the number, angle and velocity of impact of larger micron size
particles whose motion is determined by the drag forces exerted by the
gas flows and by the particles' inertia. Figure 7 shows the trajectories
of six micron particles that are flowing through the first stage, midway
between the inner and outer radius of the turbine flow path, in a large
electric utility turbine. The figure shows that 18% of these six micron
diameter particle impact on the pressure surface of the first stage
turbine vane and all of them impact on the pressure surface of the first
stage rotor blade. Figure 8 shows the calculated variation of angle and
velocity of impact with position over the first stage vane surface. For
each impact, the computer calculates the amount of metal removed from
the blade considering both the angle and velocity of impact using the
erosion damage prediction model of Figure 9 to make the estimate. Fig-
ure 10 shows the calculated metal recession rates as a function of posi-
tion through the first stage vane and blade row. Again one can add up
the damage expected from each particle size at the concentration expected
to be present in the gases leaving a particle cleaning system.
Advantage can be taken of the ability of the computer model to
permit us to mathematically vary the erosion damage response of the
blade material from a very ductile to a very brittle material in order
to see the sensitivity of the blading to the character of the blade.
This has been done by calculating the response using the four assumed
erosion damage characteristics of the blade material shown in Figure 11.
If the gas being expanded through the turbine contains 0.002 grains/scf
of particles with essentially no particles larger than 6 microns in
diameter then the expected metal recession rates at the three critical
locations - rotor blade nose - and rotor and stator trailing edges are
shown in Figure 12 as a function of the "ductility" of the blade material.
Figure 12 indicates that for the 0.002 grains/scf grain loading and the
full range of "ductility", the greatest metal recession in 10,000 hrs.
is on the order of 0.1 inch which is about the trailing edge thickness
used in electric utility turbines.
These erosion predictions are based on estimates of peak erosion
rates in the curves in Figure 11 for coal ash and sulfur sorbent particles
impacting superalloy turbine materials under turbine conditions. Experi-
mental erosion data for these materials and turbine conditions are necessar\
for more accurate erosion evaluations.
19
-------
CONCLUSIONS
• Extensive work has been done to understand how much par-
ticulate matter can safely be put through a turbine.
• Theoretical models to predict the arrival rates of sub-
micron particles and the effect of a deposit layer on
turbine performance have been developed.
• Models to predict the growth and removal of deposits are
not available nor is experimental data taken under condi-
tion relevant to a coal fired combined cycle plant avail-
able to calibrate the deposition models.
• Erosion damage prediction models based on "guessed" data
of the response of the materials to particle impact have
been developed.
• Experimental data on the response of turbine materials to
impact by micron size particles under appropriate turbine
conditions is needed.
• An order of magnitude estimate of the gas cleanliness
required to successfully operate a large industrial gas
turbine in combined cycle service has come out of all
this - in the gas flowing through the turbine we are look-
ing for particulate concentrations within an order of mag-
nitude of 0.002 grains/scf with the concentration of
particles larger than 6 microns in diameter reduced to
negligible levels.
• Careful attention must be given to maintaining sodium
chloride and hydroxide concentration in the turbine gases
in the tens of parts per billion by volume range to avoid
corrosion and glueing of deposits.
20
-------
Dwg. 6357A67
NaOH
Gas
H20
so2
HCI
Na
S02+l/202
Na S04(g)
Na2S04U)
Three Processes Remove Liquid Film:
(2)
(3)
884 °C
Turbine
Metal
2 HCI - 2NaCI
Fig. 1 -Theoretical turbine tolerance to alkali metals
21
-------
Curve 678765-A
o>
Concentration of Cl Compounds in Combustion Gases, ppm by Volume
Fig. 2- Tolerance of a gas turbine to Ma contaminants in
coal gas combustion products. The effects of Cl and
variations are indicated
22
-------
Deposition in Turbine
First Stage
Rotor Blade
12 0
Axial Distance (cm)
Fig. 3 - Deposition in first stage of a 60- 70 MW turbine for an inlet particle concentration
of 0. 002 gr/ scf - everything below 1 micron sticking
23
-------
Particle Data
Used in Deposition Calculation
Size
Wt% Oversize
Particle Diameter
Fig. 4 - Particle size distribution used for deposition rate calculations
24
-------
Deposition in
20r First Stage
Scale in centimeters
10
0
Stater
Rotor
0
Blade
Deposit after 1000 hrs.
5
0 5
Fig. 5-Calculated deposit build-up after 1000 hours
10
25
-------
Effect of Deposition
on Power Output of Stage
Operation Time (Hr
Fig. 6 - Effect of deposition on power output of the turbine stage
26
-------
a:
or
T.Ioi
c."1
2:
a.
Z t-
UJ
or
Cr'
I
id E
c' C "IICRL'N
61/CC
VIEW
1C 11 12
RXIPL LENGTH,!N
Fig. 7 - Trajectories of 6 micron particles through a
turbine stage. (18% of 6 micron particles impact the
first stage stator -100% impact in the 1st stage rotor)
27
-------
2400
2000
1600
1200
.5 800
400
0
Curve 681^59-A
_o
>
TS
ro
o.
Stator
p =L5gm/cc
u.
cr>
o>
T3
80
70
60
50
Curve
O)
•5, 40
30
0
Stator
p =1.5gm/cc
1.0
2.0
3.0
Axial Distance- inches
12 urn
4.0
5.0
Fig. 8- Particle impact velocities and impact angles as a function
of position on the first stage stator vane of a 60-70 pu turbine
28
-------
Curve 681611-A
o>
*— »
CO
s
E = £j + Eo, Total Erosion Damage
30 40 50 60
Impact Angle, degrees
Fig. 9 - Erosion damage model
29
-------
0.026
£ 1.0
^
c
S.
o
^
•3 0.8
•&. 0.6 -
-. 0.4 |-
'. 0.2
S
£ 0.0
SUtor
pp = 1.5gm/cc
Particle Concentration =0.0001 grains/scf
0.0
1.0
2.0 3.0
Axial Distance, Inches
• 12 nm
0.020 £ "
r
2.
0.016 8
_0
•g
- 0.012 B
-0.008 g
c
c
- 0.004
0.1 Grain/10 SCF
Ductile Blade Material
Erosion rate in stator as a function
of axial position
1.0 2.0
Rotor Axial Distant*, Inches
Erosion rate in rotor as a function
of axial position
Fig. 10 - Calculated erosion of 1st stage stator vane and rotor blade pressure surface
- electric utility gas turbine
30
-------
Curve 680947-A
Impact Velocity =250 fps
Maximum Erosion Angles
A. 10°
Ductile
30 40 50 60
Impact Angle, degrees
Fig. Ur Erosion curves corresponding to four cases of blade material response
31
-------
Curve 683149-A
(turbine rating-60-70my electric)
(erosivity of coal ash assumed 1/25
SiC abrading Ni-Coalloy)
Rotor Nose
o
m
o
cu
JC
o
^o
'oo
cu
o
cu
One Stage Clean-Up Cyclones
Followed by Granular Bed Filter -
0.002 Grains/scf at Turbine Inlet
Rotor Trailing Edge
Stator Trailing Edge
0
20 30 40 50 60 70 80 90
Angle of Maximum Erosion Damage, (3 ,degrees
max
Fig. 12- Sensitivity of gas turbine first -stage erosion to
angle of maximum erosion damage
32
-------
PRESSURE AND HOSTILE ENVIRONMENTS
By:
J. Hull
Acurex Corporation/Aerotherm Division
Mountain View, CA 94042
33
-------
MATERIALS FOR USE IN HIGH TEMPERATURE/
PRESSURE HOSTILE ENVIRONMENTS
By:
Jacques Hull
Acurex Corporation/Aerotherm Division
Mountain View, California 94040
Particulate control equipment for coal conversion systems
must withstand high temperatures (>1200°F) , high pressures
(>5 atm), and sulfidizing or carburizing atmospheres. In this
paper, limitations of commercial materials used for particulate
control equipment are summarized and causes for material dete-
rioration are outlined. Data are presented on materials that
resist deterioration, and possible trends in the development of
improved materials are indicated.
34
-------
MATERIALS FOR USE IN HIGH TEMPERATURE/
PRESSURE HOSTILE ENVIRONMENTS
Coal conversion equipment requires materials that can withstand environ-
ments with particulates and gases containing carbon and sulfur at high tem-
peratures (above 1200°F (650°C)) and high pressures (above 5 atm). While most
commercially available high-temperature materials perform well at high tem-
peratures in air, they may be subject to severe corrosion in carburizing or
sulfidizing atmospheres.
Materials for coal conversion equipment must be either metallic or
ceramic, since polymeric materials cannot survive temperatures above 800°F
(427°C). Metallic materials are subject to compound formation, and can oxi-
dize, carburize, and sulfidize. Because they are used primarily in air, alloys
that resist oxygen at high temperatures do so by forming protective oxides.
These oxides then act as barriers to further oxidation.
It is not surprising that these oxide barriers do not resist penetration
by sulfur and carbon effectively at high temperatures. Until new alloys are
specifically developed to resist these elements, the few alloys which have
demonstrated some degree of resistance to corrosive attack in coal conversion
environments will have to be used.
In contrast, ceramic materials are made up of compounds such as oxides,
carbides, and sulfides. Since these compounds must dissociate and recombine in
order to form new compounds, they tend to resist corrosive attack more suc-
cessfully than metals. Ceramics also are more resistant to high-temperature
attack because they are more refractory than metals (their melting or dissoci-
ation temperatures are significantly higher). But they do have some undesirable
properties, such as poor thermal shock resistance.
Materials respond to high pressures somewhat differently. Pressure
affects corrosive attack by altering the reactions (and the reaction rates)
35
-------
which occur at atmospheric pressure. Thus, while one reaction may control at
room pressure, a different reaction may govern under high pressure, and the
reaction products may be different. Pressure tends to slow down reaction
rates in the solid state, and consequently, it can slow down corrosive attack.
However, pressure can also slow down the formation of barriers which protect
against corrosive attack. Metallic resistance to high-temperature and high-
pressure corrosive attack can be seen in terms of two competing phenomena:
the formation of protective barriers, and the simultaneous penetration of
these barriers by hostile elements.
Ceramic materials behave somewhat similarly under corrosive attack. For
instance, under the proper conditions, silicon carbide will react in air to
form oxides of silicon. Under other conditions, silica will react with steam
to form silicic acid. Consequently, the fact that the ceramics themselves
can act as protective barriers does not protect them from corrosive attack.
In addition to the properties of the metals and ceramics, the forms that
these materials are used in are important. While scrubbers, cyclones, and
most other equipment for controlling particulates use metals and ceramics of
more-or-less conventional dimensions, equipment such as electrostatic precip-
itators, fabric filters and granular bed filters will require that some of
their components be of fine dimension. Thin metallic wire, ceramic filaments,
or ceramic microspheres are required. This presents an additional problem
since materials of fine size disintegrate more rapidly under corrosive attack
than bulk materials. Consequently, it is even more necessary to obtain
materials which are highly resistant to corrosive attack.
In the remainder of this paper, these problems will be discussed in de-
tail. The corrosive attack of metallic materials will be described first.
Because the presence of particulates also may result in erosive attack, the
erosion resistance of both metals and ceramics will be considered, followed
by a discussion of the corrosive attack of ceramic materials. Finally,
materials of fine dimensions will be looked at in light of the more severe
corrosive and erosive problems inherent in such materials.
36
-------
GASEOUS CORROSION OF METALLIC MATERIALS
Metals useful in high-temperature particulate control equipment include
the iron-base, nickel-base and cobalt-base alloys. The iron-base alloys
include both stainless steels and superalloys. Some examples are shown in
Table I. While iron is the main constituent, chromium is an important alloy-
ing element present in significant quantity (~20 percent), and nickel gene-
rally is present. Chromium forms a tenacious, protective oxide barrier, and
nickel adds significant high-temperature strength to the alloy.
Some nickel- and cobalt-base superalloys are listed in Table II. While
nickel and cobalt generally are the predominant elements in the alloy classes
listed, they are not always the major element present. For example, Inco-
800 is an important nickel-base alloy not listed in Table II, consisting of
47 percent iron, 31 percent nickel, and 21 percent chromium. As in Table I,
these superalloys contain a substantial amount of chromium and nickel.
The nickel in the alloys provides the necessary high-temperature strength,
as illustrated in Figure I.1 This figure also shows that rupture strength
increases with increasing nickel content in the alloy. While strength drops
rapidly with increasing temperature, alloys containing more nickel have higher
rupture strength at higher temperatures. Unlike failure at room temperature,
delayed failure can occur at high temperatures under reduced loads. Materials
often are selected on the basis of the time it takes them to rupture under
stress at temperature. Design criteria generally include the creep rate
(rate of change in dimensions) for the given stress and temperature con-
ditions, since materials tend to creep at high temperatures.
Corrosive attack of a metal is highly dependent on the gaseous environ-
ment. Table III lists the type of gases generally present in coal conversion
processes, and the typical amounts in which the gaseous components are present.
Most of these gases can be present over a very wide range, and consequently,
metals should be selected for the specific environments to which they will
be exposed. In one typical environment (the laboratory gas used for
screening tests at IITRI on the MPC/ERDA program2), a number of alloys
have been tested at 10 atmospheres and 1800°F (982°C). This environment
has an equilibrium composition (in volume percent) as follows: 32.5% HoO,
37
-------
31% H2, 3% CH4, 17% CO, 15% CC>2, 1% NH3, 0.5% H2S. If the temperature is
lowered to 1500°F (815°C) , the equilibrium composition shifts to 36.5% HO,
23% H2, 9% CH4, 11% CO, 19% C02, 1% NH3> 0.5% H2S. Thus, the reaction between
the gas and metal is influenced not only by the temperature, but also by the
change in the gas composition.
If sulfur is present in the environment, the melting temperature of the
sulfides of the elements present in the alloy must be considered. These are
listed in Table IV. It is evident that when nickel combines with sulfur at
temperatures above 1193°F (645°C), the compound nickel sulfide will be in the
form of a molten slag. This means that when sulfur penetrates the chromium
oxide protective barrier and attacks the nickel, a molten slag will result at
high temperature. The slag will flux away the surface material, removing the
protective scale. This can result in catastrophic failure of the material.
However, the sulfur first must combine with the nickel present. Since this is
a diffusion-controlled phenomenon, the rate at which the combining occurs
slows down at an exponential rate as the temperature is lowered. Consequently,
this slagging phenomenon is only critical at very high temperature (over 1600°F
for most alloys).
The formation of molten cobalt sulfide requires temperatures in excess
of 1600°F (870°C); molten iron sulfide requires temperatures above 1800°F
(982°C). Chromium sulfide remains solid to even higher temperatures. Sulfate
formation in an oxidizing atmosphere (in contrast to sulfide formation in a
reducing atmosphere), is not as severe a problem. The kinetics are not as
favorable, and the respective melting temperatures are higher.
If the thermodynamic stability conditions are such that the metal oxide
forms in preference to the sulfide, then sulfidation need not be feared.
Figure 2 illustrates the influence of the gas composition at given tempera-
ture and pressure conditions on sulfide formation.3 The IITRI gas is plotted
for both 0.1 and 1.0% H-S. In both cases, the points will be within the
chromium oxide field of stability and will not form chromium sulfide. (This
can be seen by extending the CrS-Cr 0 line in Figure 2.) The 1.0% H S point
falls within the sulfide stability region of both the nickel and the iron, in
preference to the oxide region. Consequently, these sulfides will form, sub-
ject to the kinetics of the reactions. The 0.1% H2S point falls on the
38
-------
boundary of the Fe-Fe 0 fields and in the nickel field. This means that the
x y
nickel in the alloy will not combine with either sulfur or oxygen, and the
iron may combine with oxygen. At somewhere between 0.1 and 1.0% H S (approxi-
mately 0.3% H,,S by interpolation), there will be enough sulfur present to
form both iron and nickel sulfide. Consequently, the alloys are subject to
sulfide damage if the H S content of the gas is more than about 0.3% tLS. Of
course, if the CO, C0? and H« levels in the gas are changed (or the tempera-
ture or pressure changed), the sulfide formation conditions will also change.
Figure 3 illustrates the effect of changing the pressure and/or the gas
composition on carburization.4 The gas composition at 1 atmosphere and
1800°F (982°C) with a CH4 level of 10% is 37% H20, 23% H£, 17% CO, 12% C00,
1% H~S. When the CH, content is increased to 30%, the composition changes to
29% HO, 18% H2, 13% CO, 9% C02, 1% H2S. As shown in Figure 3, as the CH,
level is increased, the stability of the chromium oxide in equilibrium in the
alloy in the presence of the gas, tends to disequilibriate towards the less
desirable chrome carbide formation at atmospheric pressure and toward the
even less desirable carbon deposition (metal dusting) condition at high
pressure. If the CH, level is kept constant and the pressure increased, equi-
librium tends to move toward the highly undesirable carbon deposition condi-
tion. This example illustrates the effect of pressure and gas composition in
the formation of undesirable carburization.
Figure 4 illustrates the influence of chromium content on the corrosion
rate in the IITRI gases.^ Note that the oxidizing gas composition consists of
68% N2, 1%-H 1% CO, 30% C02, and 0.1% S02- The reducing gas composition is
as listed previously. The corrosion rate shown in Figure 4 is severe for
less than 20 percent chromium, while more than 25 percent chromium signifi-
cantly protects the alloy from corrosion. This strikingly illustrates the
role of chromium oxide in forming a protective barrier which slows down the
corrosion rate.
Pilot plant data are presented in Table V.6 Tests A, B, and C occurred
in the regenerator section of the gasifier at temperatures in excess of 1700°F
(930°C) in the catalyzing presence of the calcium oxide from the dolomite.
The results were rather erratic. For this type of gas, existing commercial
alloys are on the "ragged edge" of corrosion resistance, and ready to corrode
39
-------
catas'trophically. Aluminized alloys (in which an aluminum oxide coating is
formed) provide an additional protective barrier which reduces the corrosion
rate. Test D occurred at lower temperature in the off-gas condition in the
gasifier. Here the corrosion rate is moderate to low for all eight
alloys.
Alloys with superior performance in reducing atmospheres are listed in
Table VI.7 While the 0.5% H S content is in the middle of the range antici-
pated for coal gas, the test temperature is relatively high. Alloys such as
310 stainless steel failed catastrophically at around 2000 hours. Subsequent
testing to 5000 hours indicated that RA-333, Crutemp 25 and Thermalloy 63 WC
deteriorated significantly. In-800, In-671, Haynes 188 and Stellite 6B con-
tinued to resist corrosion.
The following listing summarizes the available data on the corrosion re-
sistance of alloys in sulfidizing atmospheres:
1. In the IITRI oxidizing atmosphere (0.1% S02), the 1000 hour, 10
atmosphere, 1850°F (1010°C) data indicate that the descending order
of preference is 310 aluminized, 310, 800 aluminized and In-671.
2. In the IITRI reducing gas at 70 atmospheres and 1800°F (982°C), the
data indicate the acceptable alloys to be:
a. At 0.1% H2S, 1000 hours:
In-671, In-800, In-800 aluminized, Stellite 6B, Haynes 188,
Hastelloy X, 310 Aluminized, Multimet N-155, Co-Cr-W#l, Al
29-4-2
b. At 0.5% H2S, 5000 hours:
In-671, In-800, In-800 Aluminized, Stellite 6B, Haynes 188
c. At 0.5% H S, 7000 hours:
In-671, In-800
d. At 0.5% H S, 10,000 hours:
In-671
e. At 1.0% H2S, 1000 hours:
800 Aluminized, Stellite 6B, Haynes 188
f. At 1.5% H2S, 1000 hours:
Stellite 6B
40
-------
3. In the C0~ Acceptor Gasifier, in the off-gas location (trace H?S)
at 1600°F (870°C) for 15,000 hours:
In-800
4. In the IITRI reducing gas at 70 atmospheres and 1500°F (815°C) for
1000 hours. (Only In-671 and the more common alloys are listed
here. The corresponding alloys listed for 1800°F (982°C) will
perform well at 1500°F.):
a. At 1.0% H2S:
In-671, 446, 310, 800 Aluminized
b. At 0.5% H2S:
In-671, 446, 310, 800 Aluminized, 310 Aluminized, In-800
c. At 0.1% H2S:
In-671, 446, 310, 800 Aluminized, 310 Aluminized, In-800,
In-793, In-601, In-600, 316, 304, 302
PARTICULATE EROSION OF METALS AND CERAMICS
Up to now we have considered the corrosive effect of gases on metallic
materials available for particulate control. The erosive influence of parti-
culates on both metals and ceramics is discussed below.
Within the range of conditions used, the effect of particulate on
material wear is essentially proportional to the particulate loading. The
influence of particle velocity on wear is approximately equal to the velocity
to the 2.5 power for metals and to the 4.5 power for ceramics (see Table VII8).
Consequently, the effect of particle velocity is greater for ceramics than
for metals. The effect of particle size on erosive wear is essentially pro-
portional to particle size, when particles are between 3 and 30 microns.
When the particle size is below 0.1 micron, the particles flow with the gas
stream. Intermediate sizes tend to deposit rather than impact, and may cause
slagging corrosion.
The effect of the angle of impact is shown in Table VIII. When the im-
pact is perpendicular to the surface, the erosive effect on ceramics is mode-
rate. With a glancing angle of around 20°, the erosive effect on ceramics
becomes very slight. Not so with metals. Glancing angles of around 20° cause
41
-------
severe erosion, while impacts perpendicular to the surface result in signifi-
cantly less erosion (although such erosion is much more severe than the
erosion in ceramics). Consequently, metallic parts should be designed
for perpendicular impact conditions, and should be coated with ceramic
whenever low-angle impacts are anticipated.
Erosion resistance appears to be associated with the Young's modulus, and
is high for high-modulus materials. Hardening a metal by heat treatment or
cold working does not change its erosion resistance. As shown in Table IX,
cemented tungsten carbides are highly erosion resistant (though not as erosion
resistant as compacted diamond). Alumina ceramics are less resistant, but
still are superior to silicon carbide. Among the metals, the cobalt-base
alloys are the most erosion resistant.
GASEOUS CORROSION OF CERAMIC MATERIALS
As noted in Table X, ceramic materials are weak in terms of tension
and are subject to cracking under thermal cycling. Ceramics are subject
to attack by steam in the temperature range from 1000 to 1700°F (540 to
930°C), where silica is leached out to form silicic acid.9 The higher the
pressure, the more severe is this effect. When the temperature exceeds
1700°F (930°C), ceramics are subject to hydrogen attack, leading to forma-
tion of the monoxide. This effect is less severe with increasing pressure.
Carbon monoxide attacks iron impurities in ceramics, dissociates, and re-
sults in carbon deposition.
High-alumina ceramics have been found to be weakened by steam, while 50-
percent alumina ceramics are not affected. High-density ceramics are prefer-
able to low-density ceramics, since the latter abrade more readily. It has
been found that the bond in calcium-bonded ceramics does not hold up well in
the coal gas environment, while phosphate-bonded ceramics seem to have bonds
which do not deteriorate in coal gas. Self-bonded ceramics are the most
resistant to the severe condition of slag erosion. Silicon carbide bonded
ceramics have been found to be subject to oxidation of the bond.
42
-------
MATERIALS FOR HIGH-TEMPERATURE FILTRATION
The materials for high-temperature filtration must be fine-sized, whether
in the form of metallic wire or porous foil, as ceramic cloth or felt, or as
fine granular aggregates such as sand or ceramic microspheres. Corrosive
attack will rapidly disintegrate these finely-sized materials and they will
lose their structural integrity very rapidly.
This problem is particularly severe with metallic materials, whether in
the form of continuous wire, felted wire, or porous metallic foil. As shown in
Table XI, metallic wire must be very fine if it is to be useful for filtration.
However, at temperatures on the order of 1500°F (815°C) , such wire will disin-
tegrate in less than 1 month. Porous foil is subject to both rapid pore en-
largement (making it an ineffective filter) and pore clogging from corrosion
products. But coarse wire can be used economically—both in electrostatic
precipitators and as a supportive wire mesh for ceramic felts which otherwise
would tend to separate.
Ceramic felts include Saffil alumina (produced by ICI), which consists of
3-micron staple filaments 1 to 2 inches long. This is available as mats, blan-
kets or paper of moderate cost. Zirconia felt (produced by Zircar Corp.) is
available with 5-micron filaments—both continuous or staple—with a signifi-
cantly higher cost than the Saffil alumina. Low-cost felts, generally used
for insulation purposes, are readily available. These include Kaowool (Babcock
and Wilcox), Cerafelt and Fiberchrome (J. Manville), and Fiberfrax (Carborundum)
They are essentially 50-percent silica-50-percent alumina staple filaments,
approximately 3 microns in diameter. Fiberfrax is available as blankets,
papers, and moderately high-cost staple-cloth reinforced with nichrome wire or
glass filaments. The Fiberchrome is chromized for better abrasion resist-
ance.
Ceramic cloth woven from continuous filaments can also be obtained.
Leached-out E glass consists almost entirely of silica, and is available woven
at moderate cost as Refrasil (Hitco) or Sil Temp (Haveg).- The filaments are
9 microns in diameter, but of low strength. Refrasil is also available
chromized, which should have superior abrasion resistance. High-strength
silica is available in the form of Astroquartz cloth (J. P. Stevens) , with
43
-------
filaments 6 microns in diameter, but it is a high cost material. Finally,
cloth is produced from continuous filaments of AB-312 (3M Co.), which consist
of alumina-boria-silica. These are coarse, high-strength, high-modulus
filaments, 11 microns in diameter. The cloth also can be obtained with a
chromized finish.
These filtration materials are currently under evaluation, and conclu-
sions as to their effectiveness cannot as yet be drawn.
44
-------
REFERENCES
1. Metals Handbook, Vol. 1, ASM, Metals Park, Ohio, 1961, p. 482.
2. A. 0. Schaefer, "A Program To Discover Materials Suitable For Service
Under Hostile Conditions Obtained In Equipment For the Gasification
of Coal And Other Solid Fuels," Metal Properties Council Program At
1ITRI, ERDA Annual Report FE-1784-24, January 1, 1976 To December 31,
1976, p. 1 - 4.
3. Ibid., ERDA Report FE-1784-21, Supplement To Quarterly Report July 1,
1976 To September 30, 1976, p. 22.
4. I. G. Wright, "Correlation Of The High-Temperature Corrosion Behavior
Of Structural Alloys In Coal Conversion Environments With The Components
Of The Alloys And Of The Corrosive Environments," Battelle, Columbus,
Ohio, ERDA First Quarterly Report, September 3, 1976, p. 11.
5. Ibid., Ref. 2, Based On Data In ERDA Annual Report FE-1784-12, January 1,
1975 To December 31, 1975, p. 108.
6. Ibid., Ref. 2, Based On Data In ERDA Annual Report FE-1784-24, Janua-
1976 To December 31, 1976, p. 2 - 17, And In ERDA Quarterly Report
FE-1784-27, January 1, 1977 To March 31, 1977, p. 2 - 7.
7. Ibid., Ref. 2, Based On Data In ERDA Quarterly Report FE-1784-30,
April 1, 1977 To June 30, 1977, p. 1 - 18.
8. Jack W. Clark, " Conversion To Power — Gas Turbines," Westinghouse
R&D Center, And A. W. Ruff, "High Temperature Erosion in Oxidizing and
Reducing Atmospheres," NBS, Papers Presented At ASM Materials/Design
Conference on Materials for Coal Conversion Systems Design, April 26 - 27,
1976, Pittsburgh, Pa.
9. M. S. Crowley, Amer. Ceramic Soc. Bull., 46, No. 7., 1967, and 49,
No. 5, 1970.
45
-------
TABLE I. IRON BASED METALS FOR SERVICE ABOVE 650°C (1200°F)
Composition, Percent By Weight
Alloy Fe Ni Co Cr Other
Stainless :
446
304
316
310
75
70
65
52
9
14
20
24
19
17
25
Iron Base:
A1-16-5-Y 78 16 5 Al
Al-29-4-4 63 4 29 4 Mo
Armco 22-13-5 57 13 22 5 Mn
Crutemp 25 47 25 25 2 Mn
46
-------
TABLE II. SUPERALLOYS FOR SERVICE ABOVE 650°C (1200°F)
Composition, Percent By Weight
Alloy Fe Ni Co Cr Other
Nickel Alloys:
In-600 7 76 16
Inconel-625 3 62 22 9 Mo
In-671 49 50
Hastelloy X 19 45 3 22 9 Mo
Cobalt Alloys:
Stellite 6B 2 2 56 29 7 W
Haynes 188 1 23 36 23 15 W
Multimet N155 29 20 20 22 4 W
Thermalloy 18 35 15 26 5 W
63WC
47
-------
TABLE III. ENVIRONMENTAL GASES IN COAL CONVERSION PROCESSES
Gas
H20
H2
CH4
CO
C02
N2
C H
x y
NH3
H,S or SO,
Typical Range
Volume Percent
0-50
10-50
0-75
5-70
0-30
0-70
0-2
0-1
0-1
Also Present: Flyash and Char Particulates
48
-------
TABLE IV. MELTING TEMPERATURES FOR METAL SULEIDES
Compound Melt Temperature, °F
Nickel Sulfide 1193
Cobalt Sulfide 1611
Iron Sulfide 1810
49
-------
TABLE V. PILOT PLANT CORROSION TESTS
Consol-CO,., Acceptor Environment
Loss of Sound Metal in Mils Per Year
Alloy
21-6-9
22-13-5
In-800
310
In-793
Hastelloy X
310 Aluminized
In-800 Aluminized
Test A
>1700°F
800 Hrs
78
85
218
64
149
438
65
65
Test B
>1700°F
1600 Hrs
199
> 1000
> 1000
412
317
> 1000
80
79
Test C
>1700°F
1127 Hrs
—
32
145
38
177
—
92
121
Test D
>1500°F
2390 Hrs
34
26
14
2
16
8
30
31
Gas: 70 N-27C02~3CO-Trace H S Except For Test D:
48H,,-23H00-12CH, -8. 5CO-6C00-2. 5N0
50
-------
TABLE VI. ALLOYS OF LOW WEIGHT CHANGE
1 2 ~
Less Than 10 mg/cm in 3000 Hours at
1800°F - 1000 psi - 0.5% H2S
(Corrosion Rates of <20 mils per year)
In-800 Hastelloy X
In-671 Crutemp 25
In-800 Aluminized Stellite 6B
RA-333 Thermalloy 63WC
Haynes 188
51
-------
TABLE VII. INFLUENCE OF PARTICLE PARAMETERS ON EROSION
Particle
Effect on Wear
Loading
Velocity
Size, Microns
>3
0.3-1
Proportional
•2.5 Power For Metals
•4.5 Power For Ceramics
Impact Damage
•Proportional to Size
Particles Deposit,
Cause Slagging Corrosion
Particles Remain in
Gas Stream
52
-------
TABLE VIII. INFLUENCE OF IMPACT ANGLE ON EROSION
Metallic Ceramic
Impact Angle Erosion* Erosion
Low -20° High Low
High -90° Lower Moderate
*To Reduce Metallic Erosion, Coat with
Ceramic
53
-------
TABLE IX. EROSION RESISTANT MATERIALS*
Listed in Descending Order
• Cemented Carbides - K701 and K703
• Ceramics - Alumina
• Cobalt Alloys - Stellite 6B
'
Erosion Resistance Increases With
Intrinsic Hardness
54
-------
TABLE X. PROBLEM AREAS WITH CERAMICS
Problem
Effect
Tensile Stressing
Thermal Cycling
Attack by Steam
Attack by H
Attack by CO
Attack by CO
Attack by Alkali Vapors
Mechanical Failure
Cracking Leading to Hot Spots
Leaching Out of Silica
Less Severe Than Steam
Less Severe, Attacks Iron
Impurities
Less Severe Yet
Least Severe
55
-------
TABLE XI. METALS FOR FILTRATION ABOVE 650°C (1200°F)
• Metallic Wire Cloth or Felt
- Sub-Mil Size Required for Effective
Filtration
- Corrosion Resistant Lifetime Measured
in Days
- Coarse Wire (>20 mils dia.) Can Be Used
As Supportive Wire Mesh For Ceramic
Felts
• Porous Metals
- Subject to Rapid Clogging by Sticky
Particulates
- Pores Rapidly Enlarged by Corrosion
Render Filtration Ineffective
56
-------
O)
IO
O
CD
O)
O)
01
S-
o
o
o
CM
|sd 0001- '
57
-------
.Q
CNJ
8
601
oo
o
(O
QJ
s-
o
M-
rC
i-
Dl
(O
E
"co
CM
O
Q.
-------
CO
8
CO
300d/200d 601
CM
o
I
31 CO 00
I I I
I I I
2Hd/frHOd Bon
CM
O CM
(O 00 r- T-
[
o
CO
o o
] 9 J"1
• °
^•^^^
cT
X
o
O
1 1 1 1
1
CM
1
CO
1
o
-CM
1
-CM
1
CO
-CM
1
CM
- CO
i
-------
CO
Cft
2
0)
Q.
3
CO
13
c
CO
0)
£
"5
00
(0
0)
i°
ji i
CO ^^™
"55 O
"5 °
2 o
T3 ^
II
(/) l^
'S
-------
HIGH TEMPERATURE AND PRESSURE EFFECTS
ON PARTICLE COLLECTION MECHANISMS
By:
R. D. Parker, S. Calvert
Air Pollution Technology, Inc.
San Diego, CA 92117
D. C. Drehmel
Environmental Protection Agency
Industrial Environmental Research Laboratory-RTP
Research Triangle Park, NC 27711
61
-------
HIGH TEMPERATURE AND PRESSURE EFFECTS
ON PARTICLE COLLECTION MECHANISMS
By:
Richard D. Parker and Seymour Calvert
Air Pollution Technology, Inc.
4901 Morena Boulevard, Suite 402
San Diego, CA 92117
Dennis C. Drehmel
Particulate Technology Branch
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
High temperatures and pressures affect the physical
mechanisms by which particles are removed from gas streams.
This paper 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. An EPA-sponsored project to obtain experimental
data is discussed and the test facility is described.
62
-------
HIGH TEMPERATURE AND PRESSURE EFFECTS
ON PARTICLE COLLECTION MECHANISMS
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:
f-u, A,, }
(1)
-------
where A. = deposition area, cm2
Qr = volumetric flow rate, cm3/s
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:
3 Try d u0
F = ^ (2)
r C'
where Fr = fluid resistance force, dynes
yr = fluid dynamic viscosity, g/cm-s
d = particle diameter, cm
UQ = relative velocity between the particle and the gas, cm/s
C' = Cunningham slip correction factor, dimensionless
The temperature and pressure dependence of Equation 2 is contained in
the terms y^ and C1 . 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:
C' = 1+
dp
1.257 + 0.40 exp (-1.1 dp/2X)
where A = mean free path of gas molecules, cm
(3)
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.
64
-------
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. P : xnenment a 1 darn 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'/yf • 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 Impaction
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:
C' p d 2u
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 u ,
the particle's relative inertia can be characterized by a relaxation time,
T, defined as:
-------
Kj /-. i „ j 2
u L p d
u0 ~ 2 UQ 18~^T
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'/Ur •
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'/vif. ) 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 urn) 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:
r' v T
D = 3L * ld (6)
where D = particle diffusivity, cm2/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 tempera-
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 C' .
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.
66
-------
Electrical migration is generally characterized by a deposition velocity which
may be approximated as:
u = H— (107) (7)
e ^ 7i u <\
where u = deposition velocity, cm/s
q = 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:
1 C' dp2 (PP - pG>g m
us = — -- (8)
18
c' V (P
P
— - lyj
18 UG
where UQ = 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
p,, = density of the gas, g/cm3
b
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* is illustrated in Figure 1.
67
-------
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:
d NP
= -K N 2 (10)
dt o p
where N =particle number concentration, cm"3
K =proportionality constant or agglomeration coefficient, cm3/s
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:
4 C' k T
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 yim diameter particle by a factor
of 1.5 (K0 increases from 3. 5 x10"10 cm3/s to 5.3x10"10 cm3/s). For a 0.1 urn
diameter particle, the rate of agglomeration remains relatively constant
(KQ = 8.5 x10~10 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
68
-------
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
69
-------
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 impactor 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 urn.
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,
70
-------
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,
d , based on the impactor calibration. The predicted cut diameter will be
determined from the following equation:
pp c' u
where K = calibrated value for Kp at 50% collection efficiency
d, = jet diameter, cm
u^ = 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 urn
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.
71
-------
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 = expj-6.0 S Np-J-67} (13)
where S = geometric solidity factor, dimensionless
N = Peclet number = Ugdw/D, dimensionless
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:
3 TT \ir d r
G p o
(14)
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
72
-------
be determined from the basic equation for a laminar flow electrical pre
cipitator :
n Q
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.
73
-------
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.
74
-------
LIST OF SYMBOLS
A, = deposition area, cm2
C' = 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
w '
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
L = length of electrode, cm
N = particle number concentration, cm"3
Np = Peclet number, dimensionless
Pt = penetration, dimensionless
Q = volumetric flow rate, cm3/s
q = particle charge, C
R = radial position of particle, cm
r = radius of outer electrode, cm
S = solidity factor, dimensionless
T = absolute temperature, °K
t = time, s
75
-------
u = centrifugal force deposition velocity, cm/s
u, = deposition velocity, cm/s
u = electrical migration deposition velocity, cm/s
C
Up = 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
H = collection efficiency, dimensionless
X = mean free path, cm
y = gas viscosity, g/cm-s
pp = gas density, g/cm3
p = particle density, g/cm3
T = relaxation time, s
76
-------
I
E
10
5
1 atm
15 atm
I II I I I I I
I I I I I 1 I I
0.1
0.5 1.0 5.0 10
PARTICLE DIAMETER,
Figure 1. The effect of HTP on the ratio C'/Ur •
77
-------
10
LU
ri 0.5
0.1
-I II 11 I I I I I M II I I I I I I Ml | I T3
1 atm
15 atm
5 10"7 5 10"6 5 10"5
PARTICLE DIFFUSIVITY, cm^/s
Figure 2. The effect of HTP on particle diffusivity.
78
-------
70
50
I I I I I I I I I I
— latm
- 15atm
30
1/2
u
20
U
o
u
2
c
I—I
E-
UJ
O
10
Particle Diameter, urn
o
4')!1 (VIM 800
TliMPHRATURE, °C
1,000 1,200
Figure 3. The effect of HTP on Brownian agglomeration.
79
-------
X
• H
O
OJ
W)
80
-------
c
o
o
-------
100
H
W
50
J T
EXPERIMENTAL
PREDICTED
d d
ex cp
PARTICLE DIAMETER
Figure 6. Comparison of Experimental and Predicted
Cut Diameters.
-------
o
CO
V
c
o
o
0)
t/1
t/i
-------
100
H
<
Pi
2
W
50
i I I I T
1 T
1,100°C, 1 atm
1,100°C, 15 atm
I I I I
l I I
0.5
PARTICLE DIAMETER, ym
Figure 8. Predicted penetrations for diffusion test
section.
84
-------
CO
<
CJ3
V
c
o
o
CO
c
o
•H
o
•H
fH
O
0)
CD
f-l
3
OO
85
-------
c
o
a;
ty
H
PJ
s
—< <
I—I
Q
tu
J
u
I—I
H
LO Pi
t/1
(LI
C
o
ni
f-i
00
rt
o
•H
S-i
4-)
a
CJ
OJ
S-i
D-.
O
,—I
a>
?-H
3
DC
•H
tu
O
O
O
LT>
86
-------
SESSION II:
FILTRATION
JACK SIEGEL
CHAIRMAN
87
-------
GRANULAR BED FILTER FOR PARTICLE COLLECTION
AT HIGH TEMPERATURE AND PRESSURE
By:
S. Yung, R. D. Parker, R. G. Patterson, S. Calvert
Air Pollution Technology, Inc.
San Diego, CA 92117
D. C. Drehmel
Environmental Protection Agency
Industrial Environmental Research Laboratory-RTP
Research Triangle Park, NC 27711
89
-------
GRANULAR BED FILTER FOR PARTICLE COLLECTION
AT HIGH TEMPERATURE AND PRESSURE
By
Mr. Shui-Chow Yung
Dr. Richard D. Parker
Dr. Ronald G. Patterson
Dr. Seymour Calvert
Air Pollution Technology, Inc,
4901 Morena Blvd., Suite 402
San Diego, California 92117
Dr. Dennis C. Drehmel
Particulate Technology Branch
Utilities and Industrial Power Division
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
This paper presents an evaluation of particulate control using granular
bed filters. Data and design models for collection efficiency and pressure
drop are presented. This information was obtained from the literature,
available performance data, and in-house experiments.
The experimental program considered particles in the 0.1 - 5 urn size range.
The penetration of monodisperse aerosols through the granular bed is deter-
mined with an optical particle counter.
The effect of filter cake on performance is discussed. Applications
for particulate removal from high temperature and pressure gas streams
are emphasized.
90
-------
GRANULAR BED FILTER FOR PARTICLE COLLECTION
AT HIGH TEMPERATURE AND PRESSURE
INTRODUCTION
High temperature and pressure gas streams are encountered in the develop-
ment of advanced energy processes such as coal gasification and fluidized
bed combustion. It is often economically desirable to utilize this gas
stream directly by passing it through a gas turbine. To prevent the
erosion and corrosion of turbine blades and heat exchanger tubes, it is
necessary to remove the particulates before utilization.
There are a number of particulate removal systems under development
that are capable of operating at high gas temperature and pressure. The
granular bed filter is one of these systems. "Granular bed filter" is a
general term describing any filtration system including a bed of dis-
crete granules, or particles, as the filtration medium. By selecting the
proper bed material, granular bed filters offer the possibility of simul-
taneous removal of particulate and gaseous pollutants. One example
suggested by Squires and Graff (19711 is the use of half-calcined dolomite
for the simultaneous removal of fly ash and sulfur dioxide.
LITERATURE SURVEY
Types of Granular Bed Filters
A granular bed filter is defined as any filtration system including
a stationary or slowly moving bed of separate, relatively close-packed
granules or particles as the filter medium. In order to prevent the
particulate matter from plugging the interstices between granules and
causing excessive pressure drop, the device needs some means for either
periodic or continuous removal of the collected particles from the col-
lecting surfaces. This definition then excludes continuously fluidized
or dispersed beds where the granular particles are kept in motion by the
gas being treated. It does include fixed bed or closely packed moving
91
-------
bed systems.
With respect to the cleaning method, granular bed filters may be
classified as continuously moving, intermittently moving, or fixed bed
filters.
The continuously moving bed filter is usually arranged in a cross-flow
configuration. The bed is a vertical layer of granular material held in
place by louvered walls. The gas passes horizontally through the granular
layer while the granules and collected dust continuously move downward
and are removed from the bottom. The dust and granules are separated by
vibration. The cleaned granules are then returned to the overhead hopper
and the panel by a granule circulation system.
The Dorfan Impingo Filter, the Consolidation Coal Company filter, and
the Combusition Power Company's "Dry Scrubber" are continuously moving
bed filters. The "Dry Scrubber" is the only one that is presently marketed.
In the late 1950's, Squires modified the continuous moving bed design
to obtain a fixed bed device with an intermittent movement of granular
solids. The bed is stationary during filtration. The accumulated filter
cake and the surface layer of granules are ejected from the panel by a
sharp backwash pulse and are immediately replaced by downward moving
fresh granules from the overhead hoppers. This design still requires a
granule recirculation system.
As opposed to the continuous and intermittent moving beds, fixed bed
granular filters require no granule circulation. Collected particles in
the bed are removed either mechanically or pneumatically. There are three
fixed-bed devices. The "Lurgi-MB-Filter" and the "Rex-Gravel Bed Filter"
clean the bed mechanically. The "Ducon Granular Bed Filter" uses a
reversed gas flow to clean the bed.
The "Lurgi-MB-Filter", which was discontinued in 1969, uses a shaking
mechanism to clean the bed. The "Rex-Gravel Bed Filter" is a revised
version of the Lurgi. It uses a rake-shaped double-arm stirring device to
loosen the filter cake and a backwash of clean air to blow the dust out
of the bed.
92
-------
In the Ducon design, the bed is cleaned by blowhack air which is
sufficient to momentarily fluidize the bed. This blowback gas flexes the
bed and expels deposited dusts.
Particle Collection Mechanisms
The primary mechanisms for particulate collection in a bed of granular
solids are:
1. Inertial impaction.
2. Flow-line interception.
3. Diffusional collection.
4. Gravity settling.
In finely packed beds operated at low gas velocities, gravity settling
and diffusional deposition will predominate. The collection efficiency will
be expected to decrease as the gas velocity increases. Coarsely packed
beds operating at higher velocities (but still below fluidizing velocities)
provide separation mainly by inertial deposition and interception. Increased
velocity will be expected to increase the collection efficiency, if the
gas velocity is not high enough to re-entrain collected material.
Interception is the mechanism whereby particles are collected on
surfaces by gas convection. Collection by this mechanism is negligible
for a clean granular bed. However, during a filtration cycle, particles
will deposit in the interstices of the bed to form an internal cake and on
the surface to form a surface cake. As the cake builds up the bed porosity
and flow channels decrease and interception becomes an important collection
mechanism. When the flow channels are too small to allow particles to
pass, collection is ensured, being referred to as complete interception
or sieving.
The operation of granular bed filters is similar to that of fabric-
filters even though granular bed filters have larger pore sizes and thick-
nesses. Payatakes (1977) classified the filtration cycle into four different
successive stages.
1. When the filter is new, particles deposit directly on the granule
surfaces. This is referred to as clean bed filtration. The collection
93
-------
efficiency for this stage Of filtration depends primarily upon the granule
size and the depth of the bed.
2. Particles deposit not only directly on granules but also and
preferentially on already deposited particles forming particle dendrites.
.1. 1 he dendrites grow to the extent that they intermesh with their
neighbors,forming a particulate coating around each granule which is non-
uniform in thickness.
4. If the granules of the bed are sufficiently small, particle coatings
of neighboring granules will bridge the gap to form an internal cake. Lee
(1975) called the internal cake a rooting cake, since it is the foundation
which supports the formation of a surface cake. Once a surface cake is
formed, filtration efficiency no longer depends on the depth of the
granular bed but rather on the thickness and structure of the surface cake.
Cake filtration results in a much higher efficiency than the original
granular bed. This is because particle collection by sieving becomes a
more important collection mechanism.
Mathema_tj_cal_Mo_dels — Clean Granular Bed
Particle Collection --
Currently several mathematical models arc available for predicting
particle collection in a granular bed (Table T). All models are for the
prediction of particle collection by clean beds; i.e., stage 1 filtration.
Stage 1 filtration may be very brief compared to the total filtration cycle.
Stage 2 filtration has been observed experimentally by several
researchers. Billings and Wilder (1970) summarized the studies by various
investigators concerned with the cake formation process during the initial
stage of filtration. They concluded that aerosol deposition occurs pri-
marily on previously deposited particles.
Based on this observed dendrite-like growth, Payatakes and Tien (1976)
proposed a model describing the dendrite growth over the entire filtration
period. This model was expanded and revised by Payatakes (1977).
Payatakes and Tien's model described the rate of dendrite growth, but
did not explain the reason for dendrite formation. In a recent paper,
Wang et al. (1977) proposed the shadow-effect concept for dendrite growth.
94
-------
They hypothesized that once a particle is deposited, it creates a shadow
area around itself on the collector surface, within which no subsequent
particle deposition may take place.
The creation of shadow areas by deposited particles has two consequences.
Since there will be no deposition within any shadow area, it means that
particle collection takes place at a discrete position along a collection
surface. The deposited dust cannot be in the form of a smooth coating.
The second consequence arises from the fact that with the creation of a
shadow area, subsequent approaching particles which would have deposited
within the shadow area had there been no deposition now attached them-
selves to the deposited particle.
Currently there is no theory available for stage 3 and stage 4 filtration.
Pressure Drop --
Hrgun (1952) proposed the following equations to describe the pressure
drop for flow through packed beds:
fzu2 (1-e) p
, b u
-AP = - - -
*>•"
Re
dc UG PG
and NRe = yTTT^T (31
b
Mathematical Model -- Filter Cake Filtrat i on
Particle Collection --
There are only a few qualitative studies on filter cake filtration. No
quantitative study has been reported. Miyamoto and Bohn (1975) studied
the effect of particulate load on collection efficiency of granular bed
filters. Particulate load is defined as the weight of particles collected
per unit bed area. They found that collection efficiency increased with
increasing particulate load. Deeper beds had higher initial collection
efficiencies but had little influence on collection efficiency at higher
particulate load. This was because collection occurs primarily in the
95
-------
surface cake rather than the filter medium itself at higher particulate load.
Leith et al. (1976) and Leith and First (1977) studied high velocity
cake filtration of fabric filters. Three mechanisms were described by
them by which particles can pass through a fabric filter or a granular
bed:
1. Straight through penetration.
2. Seepage or bleeding penetration.
3. Pinhole plug penetration.
In straight through penetration, particles pass through the filter
without stopping ; i.e., they are not collected by the filter. Once a
particle lands on or in the filter, it does not necessarily remain at its
point of initial impact. As the dust deposit builds up, the dust may work
its way through from the dirty to the clean side of the filter. Gas flowing
through the bed produces a drag force on the particle deposits which may
move particles through the bed. Penetration of this sort is called seepage
or bleeding. The pinhole plug mechanism postulates that a plug of deposited
particles dislodges from the dust deposit and moves out of it, leaving
behind a pinhole.
They found significant trends in the dust penetration mechanisms.
Straight through penetration is important after a cleaning cycle but
rapidly diminishes in importance after a filter cleaning cycle. The
seepage mechanism is relatively constant during the entire filtration
cycle. The pinhole plug mechanism rapidly rises in importance after
cleaning, passes through a maximum, and then declines as the dust deposit
becomes thicker and pressure drop through the deposit increases.
Pressure Drop --
In granular bed filters, the flow resistance should change little as
long as the large pores are open, but will increase when the large pores
are closed by surface cake. The rate at which the pressure drop increases
depends on whether the filter cake compacts during formation. For laminar
flow and no cake compaction effect, the pressure drop across the filter
cake is:
96
-------
150 yr C . u2
_Ap = ___£^EL_5_Llz£U (41
d2 e3 p
c p
This equation predicts, that at constant inlet conditions, the pressure drop
across the surface cake varies linearly with time as long as the porosity
of the filter cake remains constant. Leith et al. (1976) have studied the
compaction of filter cake. They found that compaction did occur for fly
ash. They also found that the Carman Kozeny equation yields excellent
results for dusts which are compressed to a specified porosity provided
the particles are isometric. They gave the following equation for the
increase in pressure drop.
- e.2 [Ya W] -2e. e- ^ -2 [l-e.
[l-e. e-a(^] - [l-e. e^^] [ze.-e.2 + 2(l-e.) £n (l-e.)]
EXPERIMENTS
For a granular bed filter to be of any practical use the filter should
be operated at high gas flow rates. For superficial gas velocities greater
than 10 cm/s, inertial impaction is the principal particle collection
mechanism. The theoretical prediction equation? listed in Table 1 were com-
pared with experimental data reported by McCain (1976), Knettig and Beeckmans
(1974), Paretsky et al. (1971), Gebhart et al. (1973), Thomas and Yoder
(1956), and Lemezis (1976). These comparisons revealed that none of the
equations listed in Table I can adequately predict the inertial collection
in a granular bed over a wide range of variables.
To obtain further information on the mechanism of particle collection
by impaction and generate additional clean bed performance data, the experi-
mental apparatus shown in Figure 1 was constructed. Filtered room air was
used for the study and all flow rates were monitored with rotameters. Mono-
disperse polystyrene latex aerosol was generated using a Collison atomizer.
The aerosol mist from the generator mixed with a stream of dilution air
and passed through a dryer to vaporize the water. Static charges
97
-------
were removed by passing the aerosol through a charge neutralizing section.
The charge neutralizing section consisted of a Krypton-85 charge
neutralizer.
Following the neutralizing section, the aerosol (further diluted
with filtered room air) flowed into the granular bed test section. Gas
flow through the granular bed was controlled with the bypass vent.
The granular bed test section was made of 15 cm (6 in.) I.D. glass
pipe and the filter was a bed packed with iron shot. The aerosol concen-
trations before and after the bed were measured with an optical counter
(Climet model CI-201). Pressure drop was monitored with calibrated gauges.
Figure 2 shows experimental data on particle penetration for a 1.1 urn
diameter polystyrene latex aerosol. The bed granule was SAE-170 iron shot
with a mass median diameter of 620 ym. The bed porosity determined by
weighing was 0.39. As can be seen, deeper beds give higher collection
efficiency.
Figure 3 shows the effect of granule diameter on collection efficiency.
Higher collection efficiency (or lower penetration) is obtained with
smaller granular diameter.
Figure 4 shows the data obtained on Agsco No. 2 quartz granules. The
granule diameter is -30+50 mesh. This granular material was obtained from
Exxon Research § Engineering Company. Two different aerosols were tested,
collection improved with increased particle size indicating that inertial
impaction is important.
Table II is a list of pressure drops and collection efficiencies of
1.1 ym diameter latex at a superficial gas velocity of 50 cm/s. It
reveals that less pressure drop is required for particle collection with
smaller-granule bed material and shallower beds.
For a granular bed with a bed depth of 3.2 cm and operated at a super-
ficial gas velocity of 50 cm/s, the collection efficiencies for 1.1 ym
diameter particles are 22% and 53%, respectively, for 620 ym and 490 ym
diameter iron shot. The pressure drop increases from 10 cm W.C. for 620
ym diameter granules to 21 cm W.C. for 490 ym granules. The increase is
110%. However, by using the finer grade of granules, the increase in
efficiency is 140%.
98
-------
Table II is a list of pressure drops for various beds whose collection
efficiencies are 50% for I.I ym diameter particles. As can be seen the
pressure drops for shallow beds are less than for deep beds.
For a shallow bed to have the same collection efficiency as a deep bed,
it has to run at a high superficial gas velocity. Therefore, the gas flow
capacity of a shallow bed is higher than that of a deep bed. However,
there is a limit on the gas velocity at which the bed can be safely
operated without causing particle re-entrainment.
Figure 5 shows the pressure drop data along with predictions by Ergun's
equation (1). As can be seen, the predicted and the experimentally
determined pressure drops are in excellent agreement. Thus, Hrgun's
equation can be used to predict the pressure drop across a clean (i.e.,
no filter cake) granular bed.
CONCLUSION
Most of the granular bed filter installations reported in the litera-
ture are for the control of emissions from clinker coolers in cement plants.
The size of particulates from these sources is quite large. Therefore,
granular bed filters have no trouble in meeting the efficiency requirements.
Performance data for removing fine particles with granular bed filters
are lacking for industrial applications. Data obtained on laboratory scale
granular bed filters demonstrate that they have the ability to remove fine
particles if small granules are used.
Operating experience at high temperature and pressure is limited.
Mechanical problems (handling and cleaning) could be much more severe at
high temperature and pressure.
99
-------
REFERENCES
Billings, C.E. and J. Wilder. Handbook of Fabric Filter Technology, Vol 1.
EPA-APTD-0690, NTIS No. PB 200-648, December 1970.
Bohm, L. and S. Jordan. On Filtration of Sodium Oxide Aerosols by Multilayer
Sand Bed Filters. J. of Aerosol Science, 7_: 311-318 (1976].
Ergun, S. Fluid Flow Through Packed Columns. Chemical Eng. Progress, 48:
89-94 (1952).
Gebhart, J. et al. Filtration Properties of Glass Bead Media for Aerosol
Particles in the 0.1.2 urn Size Range. J. of Aerosol Science, 4_: 355-371 (1973).
Jackson, S. and S. Calvert. Entrained Particle Collection in Packed Beds.
AIChE Journal, J_2: 1075-1078 (1966).
Knettig, P. and J.M. Beeckmans. Capture of Monodispersed Aerosol Particles
in a Fixed and in a Fluidized Bed. Canadian J. of Chemical Eng. 52: 703-
706 (1974). ~~
Lee, K.C. Filtration of Redispersed Power-Station Fly Ash by a Panel Bed
Filter with Puffback. Ph.D. Dissertation. The City University of New
York (1975).
Leith, D. et al. High Velocity, High Efficiency Aerosol Filtration. EPA-
600/2-76-020, NTIS No. PB 249-457 (1976).
Leith, D. and M. First. Performance of a Pulse-Jet Filter at High Filtration
Velocity, I. Particle Collection. J. of APCA, 2_7: 534-539 (1977).
Lemezis, S. Advanced Coal Gasification System for Electric Power Generation,'
Research and Development Report No. 81, Interim Report No. 3. Prepared by
Westinghouse Electric Corporation for U.S. Energy Research § Development
Admin. 1975.
McCain, J.D. Evaluation of Rexnord Gravel Bed Filter. EPA 600/2-76-164,
NTIS No. PB 255-095, June 1976.
Miyamoto, S. and H. Bohn. Filtration of Airborne Particulates by Gravel Filters:
I. Initial Collection Efficiency of a Gravel Layer. J. of APCA, 20_: 1051-1054
(1974).
Miyamoto, S. and H. Bohn. Filtration of Airborne Particulates by Gravel Filters:
II. Collection Efficiency and Pressure Drop in Filtering Fume. J. of APCA,
_2_5: 40-43 (1975).
Paretsky, L. et al. Panel Bed Filters for Simultaneous Removal of Fly Ash
and Sulfur Dioxide: II. Filtration of Diluted Aerosol by Sand Beds. J. of
APCA, 21: 204-209 (1971).
100
-------
REFERENCES (continued)
Payatakes, A.C. and C. Tien. Particle Deposition in Fibrous Media with Dendrite-
Like Pattern: A Preliminary Model. J. of Aerosol Science, _!: 85-100 (1976).
Payatakes, A.C. Model of Transient Aerosol Particle Deposition in Fibrous
Media with Dendrite Pattern. AIChE Journal, ,23: 197 (1977).
Squires, A.M. and R.A. Graff. Panel Bed Filters for Simultaneous Removal of
Fly Ash and Sulfur Dioxide: ITT. Reaction of Sulfur Dioxide with Half-CaTcined
Dolomite. J. of APCA, 2J_: 272-276 (1971).
Thomas, J.W. and R.E. Yoder. Aerosol Penetration Through a Lead Shot Column,
A Method of Particle Size Evaluation. AMA Archives Ind. Health, 1_3: 550 (1956).
Wang, C.S. et al. New Concepts of Particle Deposition from Suspensions Flowing
Past a Collector. Paper 18b, presented at the AIChE 83rd. National Meeting,
Houston, Texas, March 20-24, 1977.
101
-------
C . = inlet particle concentration, g/cm3
LIST OF SYMBOLS
a = empirical constant, cm2/dyne
= inlet particle con<
Ci = empirical constant
D = particle diffusivity, cm2/s
d = single collector diameter, cm
d = particle diameter, cm
f = friction factor, dimensionless
f = ratio of collector diameter to initial capillary diameter,
dimensionless
g = gravitational acceleration, cm/s2
K, = D'Arcy permeability, cm2
K = inertial impaction parameter, dimensionless
k = Boltzman's constant = 1.38 x 10~ erg/°K
Pt = particle penetration, percent or fraction
R = collector radius, cm
T = absolute temperature, °K
t = time since last cleaning, s
u~ = superficial gas velocity, cm/s
b
u^. = interstitial gas velocity, cm/s
bl
Z = bed depth, cm
GREEK
Z = bed porosity, fraction
e. = initial bed porosity, fraction
\ir = gas viscosity, poise
b
r\ = total single collector efficiency, dimensionless
pr = gas density, g/cm3
b
p = particle density, g/cm3
Ap = pressure drop, dyne/cm2
DIMENSIONLESS NUMBERS
N0 = Reynolds number
He
N~ = Peclet number
Pe 102
N = Nusselt number
-------
PH
O
2:
o o
rH W
H pq
rH Pi
a <
P4 —1
Pi | >
PH 2
PH Pi
H "
Pi
O Z
PH H
co 2;
2: o
O HH
n E— i
E— i U
< W
1=) -1
CX i — -J
W 0
U
w
i-J W
pq 1-3
< CJ
1 — 1 E-H
< Pi
< PH
rH
0
rH
,0
03
H
CO
W
H
0
2:
2:
0
^H
E — '
Ctj Pi p| pj
PI ^UaJ'-H O O O
O P X) rH -H -H .H
•H 0 ^S P p 4J
P tfirHXtO O U U
U -H 3 U 0 • 0 • CD
Oj p|p|0 rHX rHX rH
PH raj00 rHrH rHrH ,— |
n rbOUMH UO UO O
o
13!
(J
PH
a.
bo
' i
^*" ~— --. rj O.
1 1 ^3 0 r 1 rrj
^f 2: ^O +
O -^^ •• p^ ^
TJ c\X° <4H
II 1X1 p. V"
PH "" "^ O -H tO
$4 (J ^-A CJ3
i a) u u 3 [— '
M U rH 1 H3
^— ' to
U VO ^
I i I
• i i IL i [ i |
P< PH PH PH PL,
x x x x y.
0000 CD
u u u u n
PH PH PH PH P
^— ' S S S S
P P P P +-)
PH PH PH PH (X,
P
rH
0 pi
rH • O p]
OJ rH pq • Oj
U OS rH H3i
•XJ nj 5-H
"^ P" Pi o
Pi 0 Oj P r-j
ni gj
X O rrj
Pi r^4 P P {T|
T3
nj
br
rH
^
' l
p
p
0 '
t/) p!
O
P P
•H U
> ni
cC PL,
bC-H
0
3
M
M cxiO
Xo~> rd cr> ,d en
rtrH OJrH -HrH 0rH ;Q ,
— 1
, — ^
10:
-------
O
LO
O
P^ -5
O W
m
rH *1
CJ<
rH I-H
PH U
P-, rH
W P-.
fX
2 W
O PH
HH ^J
H CO
O
•-J E-
O <
U
CO
CO
co
R
W
H^
PQ
<
H
o
CNJ
.
CNJ
rH
II
x
ft
0
n
nd
0
pq
U
CNI
cn
II
A
0
Q
^
0
PQ
0
CNI
MD
II
-P
ft
0
Q
nd
0
pq
u
CNI
to
II
-P
ft
0
a
"XJ
0
T
U
.
£L| ^
<3
6
u
•
i-H
rH
o\o O'
C_)
„
U
PH S:
<]
6
o
rH
rH
oV3 O
CJ
t — \
U
PH ;s
o
6
o
.
i—i
rH
o\° O
C_)
„
cj
a, «='
<
o
•
rH
rH
o\=> O
U
0 0
1 — 1 P f — \
306
a e ^
nj cd v__/
fH -H
U O
I l>- to to
1 IO CN] CN]
1 CO LO 00
i \o LO to
1 OO OO CO
1 CNI rH CNJ
to LO <5t- r--
cr> LO -^ CN]
OO CD CNJ CNI
tO CN) rH rH
to oo CD cn
OO tO to i-H
^
rH 0 LO M3
CN) rH
tO CNI CNI O
LO CNI rH ,H
O O O O
en CNI cn \o
^- \£) [^^ OO
104
-------
Table III. PRESSURE DROP FOR 502 COLLECTION OF
1.1 ym DIAMETER PARTICLES.
Granule
Diameter
(ym)
490
620
730
790
860
Pressure Drop (cm W.C.)
Bed Depth
3.2 cm
19
17
16
15
15
Bed Depth
6.2 cm
24
23
20.5
23
Bed Depth
9.2 cm
26
22
28
Bed Depth
12.2 cm
22
25
21
30
105
-------
OPTICAL
COUNTER
OPTICAL
COUNTER
GRANULAR
BED
DILUTION AIR
VENT
ROTA-
METER
DRYER
FILTERQ
COMPRESSED
AIR
ROTAMETER
PRESSURE
TAP
CHARGE
NEUTRAL IZER
ROTAMETER
AIR —CJ
BLOWER
FILTER
-IXH*
O
Figure 1. Schematic diagram of the experimental apparatus
106
-------
11
o
o
Q
W
PQ
10
o
(X]
n
oi
w
E-
Pi Q
,-1
O
CO
o
w
-. o
oo
— o
X
CD
0
c
CD
4-i
O
a
aj
S
u
H
i— i
U
O
J
UJ
o en
LT) <£
tL,
c£
w
0,
D
00
„. o
o
(xi
CD
CD
r— I
U
CO
aj
4->
p
0)
J-i
CD
X
(XJ
CD
O
O
o
LO
107
-------
o
(XI
CTi
04
w
Q
O
W
PQ
o
w
o
\o
oo
ii
rt
W
E->
W
<
i—(
O
Pi
o
o
o
CTl
O
oo
o
t---
in
S
o
>-.
H
O
o >
LO
en
u
I— I
PL,
Oi
K)
CO
O
(XI
CD
•M
i—I
•H
m
T3
CD
JH
03
c
03
o
•H
^
03
fn
•P
0)
C
0)
0)
I— I
o
•H
-p
JH
03
03
•M
c
CD
e
•H
!-.
0
P<
X
P4
bfl
•H
O
o
o
LO
'NoiivtfiaNad
108
-------
-P
i— I
•H
QJ
03
O
•H
•P
03
0
C
0)
(D
i—I
U
rt
cx
rt
0)
e
0)
ft
X
in
bo
•H
109
-------
40
u
!s
6
u
PH
30
20
10
BED DEPTH
12.2 cm/
e = 0.39
= PREDICTION
0 10 20
100
Figure 5. Experimental and predicted pressure drops
across granular bed consisting of iron shot
110
-------
EVALUATION OF A GRANULAR BED FILTER FOR
PARTICULATE CONTROL IN FLUIDIZED BED COMBUSTION
By:
R. C. Hoke, M. W. Gregory
Exxon Research and Engineering Company
Linden, NJ 07036
111
-------
EVALUATION OF A GRANULAR BED FILTER FOR
PARTICULATE CONTROL IN FLUIDIZED BED COMBUSTION
By:
R. C. Hoke, M. W. Gregory
Exxon Research and Engineering Company
Linden, New Jersey 07036
A program is underway to evaluate the use of a granular
bed filter to reduce emissions of particulates in the flue gas
from pressurized fluidized bed coal combustion. Since the
filtered flue gas is to be expanded through a gas turbine, the
emissions should be at a level satisfying both the turbine and
environmental requirements.
A granular bed filter of the Ducon Company design was
installed on the EPA/Exxon pressurized FBC Miniplant unit and
tested.' Initial tests were terminated due to plugging of
inlet screens by the particulates. The filter system was
modified and further tests were more successful. A 24 hour
run was completed and in this and other tests, the ability of
the system to remove and collect particulates and maintain low
pressure drops was demonstrated. However, the filter outlet
particulate concentration is still slightly above the upper
limit of the tentative target range set by gas turbine require-
ments. The concentration also appears to increase with time.
Screens used to retain the filter media during the filter
cleaning cycle plug very readily and cannot be used. Reten-
tion of the filter media in the beds therefore, requires close
control of the operation during the filter cleaning cycle.
The filter system has been vulnerable to upsets and it has
not been possible to take corrective action to offset the
effects of the upsets.
The current program is aimed at resolving these problems.
Tests are now underway using dense filter media with no
retaining screens and employing various filter cleaning cycle
conditions. The objective of these studies is to increase the
filtration efficiency further and to prevent the drop in
efficiency with time. If current tests are successful, exten-
ded testing of the system will begin, coupled with a gas
turbine materials test program. Environmental assessment
tests will also be made.
112
-------
EVALUATION OF A GRANULAR BED FILTER FOR
PARTICULATE CONTROL IN FLUIDIZED BED COMBUSTION
INTRODUCTION
The successful development of the pressurized fluidized bed coal combustion
process is dependent on the ability of particulate control devices to remove
particulates from the hot combustor flue gas to very low levels. This must be
done to assure that the expansion of the flue gas through a gas turbine does
not cause damage to the turbine by erosion, corrosion, or deposition of solids
on the turbine blades. Current estimates of the allowable particulate concen-
o
tration in the flue gas entering the turbine range from 45 to 1 mg/m (0.02 to
(2)
0.0004 grains/SCF) . To-meet these estimated requirements, the flue gas
leaving a pressurized combustor must first be precleaned in a two stage cyclone
and then sent to a third stage high efficiency device for final cleaning. The
efficiency of the third stage device must be in the range of 95 to 99.7% to
be within the currently estimated particulate loading target range.
In addition to the gas turbine inlet requirements, the U.S. Environmental
Protection Agency has imposed limits on the emission of particulates from coal
fired installations of 0.1 Ib/MBTU coal fired. This translates to a particulate
3
concentration in the flue gas of about 115 mg/m (0.05 gr/SCF), somewhat
higher than the limit set by turbine requirements. Therefore, at the present
time, removal efficiencies are dictated by the turbine requirements. However,
the environmental standards are currently being reviewed and may be tightened
in the future, especially with regard to the emission of particulates less
than 2 microns in size. This is the size which is more difficult to remove
and also causes the least amount of damage to the gas turbines. Therefore,
environmental considerations may indeed limit the allowable particulate
emission levels in the future.
The present particulate removal program at Exxon Research and Engineering
Co., which is sponsored by the U.S. Environmental Protection Agency, will test
113
-------
two particulate removal devices. The first device is a granular bed filter of
a design developed by the Ducon Company. The choice of the second device will
most likely be from one of those discussed at this conference.
The Ducon-type granular bed filter consists in a number of small beds
packed with suitable granular filter media such as alumina, quartz, etc. A
stack of the filter beds form a single filter element. A number of filter
elements can be used depending on the volume of gas to be filtered. A photo-
graph of a filter element purchased from the Ducon Company for this program
is shown in Figure 1. The photograph is a closeup which shows five individual
filter beds of the total twelve which form this particular element. Dirty gas
passes through the screen sections down into the filter beds immediately below
the screen sections. Clean gas from the beds is collected in a manifold in
the interior of the element and then passes to the clean gas outlet system.
As the filtration step proceeds, the pressure drop across the element increases
and eventually the element must be cleaned by the reverse flow of clean gas.
This "blow back" occurs by flowing clean gas in reverse direction through the
outlet gas manifold, up through each filter bed and out through the screens.
The function of the screens is to retain the filter media during the blow back
step, keeping it inside the filter beds, while allowing the fine particulates
removed from the filter media by the blow back gas to pass through. The fine
particulate then settles outside the filter elements and is collected at the
bottom of the vessel containing the filter elements.
This type of granular bed filter was chosen to be tested in the EPA/Exxon
Miniplant since it was felt to possess certain advantages over other high
temperature particulate removal systems. In the first place, development of
granular bed filters as a class was believed to be further along than the
development of other high temperature particulate removal systems. Of the
granular bed filters under development, the Ducon filter had been previously
tested at relatively high temperatures and pressures and showed some promise
in meeting the high particulate removal efficiencies required by the gas
turbines ^. The Ducon design also had the desirable feature of retaining the
granular filter media in the filter vessel during the cleaning step. In all
other high pressure granular bed filters, the filter media is removed, cleaned
externally and recycled back to the filter vessel.
114
-------
The objectives of the granular bed filter test program at Exxon Research
were to measure the outlet loading from the granular bed filter, determine if
the removal efficiency was maintained with use, measure operational stability
of the filter e.g., can low pressure drop across the filter be maintained, does
the filter plug, is the amount of blow back gas needed to maintain steady
operation within reason, etc. and finally, to measure the long term life of
the filter hardware. The primary operating parameters were the filtered gas
flow rate, usually measured as the gas velocity entering each filter bed, the
blow back gas flow rate, the duration and the frequency of the blow back step,
and the type of filter media used i.e., particle size, shape, density.
DESCRIPTION OF EQUIPMENT
The granular bed filter was installed in a pressure vessel tied into the
flue gas exit system from the EPA/Exxon Miniplant pressurized fluidized bed
combustion unit. The Miniplant has been described in detail in a previous
report . Briefly, it consists of a combustor, 33 cm (13 in) inside diameter
and 10 m (32 ft) high, capable of operating at pressures up to 1000 kPa (10
atm abs), temperatures up to 980°C (1800°F), superficial gas velocities up
to 3 m/s (10 ft/sec) with coal feed rates up to 200 kg/hr (450 Ibs/hr). The
3
maximum flue gas rate is about 34Sm /min (1200 SCFM) but a typical flue gas
3
rate is 18Sm /min (650 SCFM). A photograph of the Miniplant is given in
Figure 2.
The pressure vessel housing the filter elements consists of a refractory
lined vessel approximately 2.4 m (8 ft) in diameter by 3.4 m (11 ft) high.
Figure 3 is a photograph. Access to the interior can be made through a 70 cm
(27 in) manhole. The vessel can hold up to four filter elements installed
through four flanges at the top of the vessel as shown in Figure 4. Each
filter element is contained within a shroud in the inside of the pressure
vessel. Inlet gas is piped to each shroud, passing through a measuring orifice
which determines the flow rate to each filter element. This is shown in
Figure 5. Clean gas exits from each shroud through openings at the top
(Figure 5) and fills the interior of the pressure shell. Blow back air enters
each filter element through the top flanges of the pressure vessel (Figure 4)
and flows in reverse direction through each filter element. Particulates
115
-------
removed from the filter element during blow back impinge on the inside surface
of the shroud, fall to the bottom and are collected in lock hoppers. The blow
back gas leaving a filter element flows in reverse direction through the inlet
gas system into the filter elements which are in the filtration step.
A natural gas preheat burner was installed to heat the interior of the
pressure vessel to a temperature above the dew point of the combustor flue gas
before starting a filtration test. The burner fires into the vessel through
a side port for an 8 to 12 hour period prior to the start of a run. The
filter vessel is at atmospheric pressure during this period. The burner was
installed after initial operation of the filter resulted in condensation of
moisture on the filter elements and inside the pressure vessel during heat up
with flue gas from the coal combustor. Condensation in the presence of flue
gas particulates caused plugging of the filter and particulate removal lines.
A number of filter element designs and blow back methods were tested
during the filter shakedown tests. Each design and the results of the tests
are described in the following Section.
RESULTS
GBF MODEL 1
Initially, three filter elements were purchased from the Ducon Company.
Figure 6 is a photograph of one of the elements and the shroud in which it is
contained when placed in the pressure vessel. Each element was 20 cm (8 in)
in diameter by 1.8 m (6 ft) long and contained twelve beds. The inlet screen
3
size was 50 mesh. The nominal flow capacity of each element was 8.5s m /min
(300 SCFM). One of the Ducon elements was designed to be blown back by short
pulses of high pressure air. The pulse duration was approximately 0.5s. This
blow back method was not tested. The other two Ducon elements were blown back
with a larger volume of air for longer durations. The intent was to fluidize
the filter beds rather than shocking them with a short pulse of high pressure
air. The volume of blow back air was sufficient to fluidize the filter media.
The duration was designed to be less than 10s. Blow back was accomplished by
isolating one end of a filter element by a blow back nozzle and seal plate and
blowing back with air at a pressure slightly higher than filtration pressure.
116
-------
After installation of the filters, shakedown began with ambient temperature
testing of the system.
The objectives of these preliminary tests were to (1) check combustor
pressure control with the filter on line, (2) pressure test the system, (3)
check the alignment of the blow back nozzles, (4) check out the operation of
the blow back flow system, and (5) measure the distribution of flow to each
one of the filter elements. A number of mechanical problems were discovered
(leaks, misalignments, etc.) and had to be corrected before further testing
could be resumed.
A number of high temperature runs were then attempted but the pressure
drops across the filter were extremely high and all attempts at blow back were
unsuccessful. Inspection of the filter elements after each of these runs showed
that a hard filter cake had formed on the inlet retaining screens. This is
shown in Figure 7. The filter medium was usually particulate free indicating
very little penetration through the screens. It was originally thought that
the plugging occurred during startup when moisture was present. The preheat
burner was later installed and a run was made to re-evaluate the Ducon filter.
The same screen plugging problems occurred and the original Ducon filter was
deemed to be unacceptable for our application.
GBF MODEL 2
A fourth filter element designed by Exxon was also fabricated and tested.
The Exxon filter element consisted of ten beds instead of twelve and was
designed to permit easy diassembly and removal of the 50 mesh retaining screens.
The blow back method was also different. The Exxon filter element used a
"positive blow back" technique. It was equipped with shut off valves to allow
it to be completely isolated from the feed and product streams during blow
back. The element was then depressurized and blown back with a larger volume
of low pressure air.
Runs using the Exxon designed filter also proved unsuccessful. Some
screen plugging was in evidence but it was the inability to seal the blow back
nozzles which caused the most problems. Evaluation of the Exxon filter was
discontinued at this point.
117
-------
GBF MODEL 3
Discussions with the Ducon Company led to the design and fabrication of
the third filter system. Ducon indicated that they had encountered the same
screen plugging problem with flyash and bypassed it by removing the screens
and designing the individual beds with more freeboard to prevent entrainment of
the filter media during blow back. It was also recommended that a fluidizing
grid be used at the bottom of the beds to assure good distribution of the blow
back air. The use of an ejector to replace the sometimes troublesome plunger
type blow back nozzles was also suggested. Filter elements incorporating these
suggestions were fabricated and shakedown continued using two of these elements
each of which contained five filter beds. A photograph of one of these
modified filter beds is shown in Figure 8. Figure 9 is a photograph of an
element.
Dirty gas enters the bed through the opening below the top flange and
passes downward through the filter bed and out into the clean gas outlet tube
in the center of the element. During blow back, the blow back air passes up
through the fluidizing grids supporting each bed, fluidizes the beds and blows
the fine particulates out through the inlet slot. A 18 cm (7 in) freeboard
above the filter beds acts as a disengaging section for the filter media and
prevents its entrainment through the outlet slot.
Operability of the modified filter system has been demonstrated. The end
of the initial shakedown phase was signified by the successful completion of a
24 hour demonstration run. This run was preceded by a number of shorter dura-
tion runs used to establish suitable operating conditions for the demonstration
run. The successful use of an ejector was also demonstrated during one of
these runs. These runs were successful in that filtration, ability to blow
back, ability to maintain low pressure drops and collection of particulates
after blow back were demonstrated. Collection efficiencies of about 90% were
3
measured based on outlet particulate concentrations of about 70 mg/m (0.03
gr/SCF) although efforts to optimize filter performance were jaot completed.
Stable operation for up to 24 hours was also demonstrated with no significant
increase in baseline pressure drop across the filter. Blow back was usually
required every 10-20 minutes during which time the filter pressure drop had
increased by 2 psi above its baseline value. A range of blow back conditions
118
-------
were used to restore the baseline pressure drop. Blow back durations ranged
betweeu 2 and 30 seconds and superficial velocity between 0.5 and 2.5 ft/sec.
Filtration velocities generally ranged between 60 and 80 ft/min. Filter media
consisting of 300 to 600 micron quartz particles and alumina particles greater
than 840 microns were tested. The quantity of blow back air used ranged from
1 to 5% of the filtered gas rate.
PROBLEM AREAS
A number of problem areas were defined during the shakedown portion of our
program. Demonstrated particulate outlet concentrations are still higher than
those required to meet the turbine requirements, although the lowest levels
measured to date are only slightly above the upper limit of the tentative tar-
get range. Changes in filter media size could be expected to improve collection
efficiency. However, at times, the filtration efficiency was very poor and the
o
outlet particulate concentrations were as high as 700 to 1200 mg/m (0.3 to 0.5
gr/SCF). It was also observed that the efficiency appeared to decrease with
time in some of the longer runs, dropping from 90% initially to about 50% later
in the run. Loss of filter media during blow back was another reoccurring
problem during shakedown. Further attempts were made to use retaining screens
on the GBF-3 model but failed because of plugging. Figure 10 is a photograph
of a 50 mesh screen plugged during a test. Additional tests made with 10 mesh
screens also resulted in significant screen plugging. Since inlet retaining
screens are susceptible to plugging, a denser filter media will probably have
to be used or better control of the blow back air supply established to
minimize these losses. A significant buildup or particulates in the filter
beds was also observed amounting to about 30% of the weight of the filter
media. A possible steady long term increase in filter pressure drop may result
because of this. However, no significant increase in filter pressure drop was
noted during any of the shakedown runs.
It was also observed that the particulates were not only building up in
the beds, but were uniformly mixed with the filter media. It is possible that
the buildup and mixing of particulates in the bed could be responsible for the
increase in the particulate concentration in the outlet gas with time. The
formation of a hard filter cake on the filter media surface resulting in rat
119
-------
holing and decreased efficiency could possibly be another problem area. This
was noted during one of the shakedown runs in which kerosene was burned for a
prolonged period of time.
Another potential problem with the current design is its vulnerability to
upsets. If upsets occur, such as bed plugging or loss of filter media, the
operating problems caused by such upsets usually require shutdown of the system.
It is usually not possible to take corrective action which restores good
operation. Another problem which may be unique to the Miniplant was the inter-
action of the granular bed filter with the rest of the FBC system during the
blow back cycle. An increase in system pressure was noted during blow back
resulting in problems with the coal feed system which is controlled by the
differential pressure between the coal feed vessel and combustor. This required
modifications to the coal feed control system to minimize the effects.
CURRENT PROGRAM
In the current program it is planned to conduct additional tests or
further modify the filter system to minimize some of the problem areas
uncovered during the shakedown runs. Since plugging of the inlet retaining
screens was a serious problem, testing will be done to determine whether filter
media losses during blow back can be reduced to an acceptable level by examining
the relation between entraining velocity particle size and density. Filter
media of different densities will be tested and acceptable blow back conditions
will be established. During these tests the effect of blow back velocity and
duration on blow back efficiency will be determined. Ways to increase filtra-
tion efficiency will be studied by evaluating the effect of filter media
particle size, bed depth, and filtration velocity. The apparent problem of
efficiency decline with time will also be studied further.
After a practical set of operating conditions has been determined, several
extended runs will be attempted to evaluate long term performance at these
conditions. These runs will allow a better evaluation of the mechanical and
material performance. Extended runs are also planned using the granular bed
filter aimed at determining erosion and corrosion effects on gas turbine
samples located downstream of the filter. This work will be done as part of a
120
-------
separate program sponsored by EKDA. Further EPA sponsored studies will consist
in testing another particulate control device and measuring the concentration
of trace emissions from a fluidized bed combustion system using a high
efficiency particulate removal system.
ACKNOWLEDGEMENT
This work is being carried out by Exxon Research and Engineering Company
under contract to the U.S. Environmental Protection Agency, Industrial
Environmental Research Laboratory. The Contract Number is 68-02-1312. The
EPA project officer is D. B. Henschel.
REFERENCE
1. Hoke, R. C., et al, "Studies of the Pressurized Fluidized-Bed Coal
Combustion Process," EPA-600/7-76-011, September 1976.
2. Keairns, D. L., et al, "Fluidized Bed Combustion Process Evaluation,"
EPA-650/2-75-027-C, September 1975.
3. AIChE Symposium Series No. 137 Vol. 70 pp. 388-396 (1974).
121
-------
IP
" """ ""' *
I' ^ *• ^fp'sf^ilJL ^- ^
•
| ^^ ;s :'
Figure 1. Ducon Granular Bed Filter
122
-------
ant
12;
-------
Figure 3. Filter Pressure Vsssel Side View
124
-------
Figure 4. Filter Pressure Vessel Top View
125
-------
Figure 5. Filter Pressure Vessel Interior
126
-------
Figure 6. Dueon Granular Bed filter and Shroud
127
-------
Figure 7. Filter with Plugged Inlet Retaining Screens
128
-------
Figure 8, Modified Filter Bed
129
-------
Figure 9. Modified Filter Element
130
-------
i
»>*«, *
Figure 10. Filter with Plugged Inlet Retaining Screen
131
-------
PERFORMANCE AND MODELING OF
MOVING GRANULAR BED FILTERS
By:
G. L. Wade
Combustion Power Company, Inc.
Menlo Park, CA 94025
133
-------
PERFORMANCE AND MODELING OF
MOVING GRANULAR-BED FILTERS
By:
Gordon L. Wade
Combustion Power Company, Inc.
Menlo Park, California 94025
ABSTRACT
An ongoing experimental and theoretical effort sponsored
by ERDA on moving granular-bed filters is decribed. Test
apparatus includes a flexible cold-flow facility featuring a
3500-cfm filter where a parametric test program is in pro-
gress. The test facility is described, along with the
completed and planned variations in operating conditions and
configuration geometry.
Initial data on measured collection efficiency and pres-
sure drop are presented; regression analysis indicates that
reasonable operating conditions may be specified for which
total efficiency in excess of 95 percent is readily attain-
able.
A parallel mathematical-modeling effort is discussed,
and key features of the resulting computer simulation program
are identified.
134
-------
SECTION 1
INTRODUCTION
BACKGROUND
Many energy-producing processes based on fossil fuel include a stringent
requirement for separation of particulate materials from hot pressurized gas
streams. Protection of such downstream equipment as combustors, turbines, and
heat exchangers is normally the dominant motivation, although compliance with
emission standards may also pose similar performance requirements. Existing
separation equipment such as cyclones and electrostatic precipitators, though
still evolving in performance and operating costs, are generally held to have
distinct limitations. This is particularly the case in the environment pro-
duced by combustion or gasification of high-sulfur coal.
A promising device for this application is the moving granular-bed filter
(GBF) now in development under ERDA Contract EF-77-C-01-2579 at Combustion
Power Company. Shown conceptually in Figure 1, a GBF features a downward-
moving bed of packed granules (media) through which the dirty gas passes in
a cross-flow pattern. Experience has shown that the process efficiently trans-
fers particulate from the gas stream to the solids stream, where several possi-
bilities for subsequent separation are available to permit circulation and
re-use of the media.
Commercial versions of the concept, restricted to temperatures below
about 800 F and to pressures near atmospheric, have been marketed by Combus-
tion Power Company for the past four years. These devices, known as Dry
Scrubbers, have been supplied for numerous stack-cleaning applications on
existing plants.
135
-------
CURRENT PROGRAM
The current contract includes an intensive program to determine the physi-
cal principles upon which GBF's operate, and to provide a quantitative verifi-
cation of these principles by controlled testing of a sizable unit (up to
3500 cfm) under cold-flow conditions.
The scope of the cold-flow program includes (1) theoretical analysis, in
which the theories and mechanisms of the interaction of particle-laden gas and
granular media are studied and mathematically expressed for computer simulation;
(2) design and construction of the cold-flow model for parametric testing and
correlation with the math model; (3) installation of a particle-injection and
sampling facility to enable controlled dust injection, particle sampling, and
analysis; and (4) carrying out of a parametric test matrix.
The primary objective is to generate information on the cold-flow model
for future application, correlation, performance prediction, and implementa-
tion on a next-generation GBF hot-flow model.
At the time of this writing, the five-month parametric testing effort is
in progress, along with analysis of the experimental data and operation of the
computer simulator. Accordingly, the present paper should be regarded as a
status report containing presentations of early data. Continued analysis,
acquisition of new data, and forthcoming computer results are expected to
modify and improve understanding of the subject material.
136
-------
SECTION 2
APPARATUS FOR COLD-FLOW TESTS
GBF SCHEMATIC
The GBF and its media circulation loop are shown in Figure 2. Pneumatic
lift is employed to transport media exiting the GBF a vertical distance of 36
ft, after which it is disengaged to begin another pass of the filter. The
pneumatic-transport process also serves as a first stage of media-particulate
separation. Second and third stages of media cleaning occur in the counter-
flow "rattler" and fluidized-bed sections atop the GBF. Separated particulate
is thus concentrated in a relatively small air stream directed to a convention-
al baghouse. The fluidized bed also serves as a convenient media reservoir.
Media flow rates are controlled by modulation of ejector air admitted near the
base of the lift pipe. Glass sections in upper and lower seal legs and in the
disengagement vessel permit visual monitoring of media flows at those points.
Many design features of the cold-flow unit look ahead to the high-tempera-
ture, high-pressure environment of subsequent program phases. Thus, for
example, pneumatic processes were selected for media cleaning and transport
instead of vibrating feeder-screeners and bucket conveyors (such as used in
commercial scrubbers) since the latter combination is not suited to high-
temperature applications.
The GBF is housed in a 3/16-inch-thick 304 stainless-steel vessel that is
15.5 ft long and 5 ft in diameter. The vessel is sized to accommodate any one
of three outer panels to form bed depths of 3.8, 7.6, or 15.2 in, based on a
137
-------
fixed inner-panel diameter of 15.4 in. Height of the slotted section of the
inner panel is nominally 53 in, to give an approach area of 17.5 so ft for
air flow. Rubber boots on top of each panel provide a flexible gas seal that
has minimal effect on panel load measurements.
The inner panel for 2-mm media (nominal granule size) contains 1/8-in x
1-in slots (Figure 3). This slot configuration promotes media spillage for
front-face cleaning during operation. The louvers in all outer panels are
based on positive retention of 0.8-mm media.
All panels are suspended from above by load cells mounted on the vessel
cover. Figure 4 shows details of load-cell orientation along with the rela-
tive location of 12 pressure sensors adapted to the measurement of normal
pressures developed by the flowing media. Analysis of these measurements is
expected to improve the understanding of loads imparted to GBF structures by
this unique configuration and process.
DUST INJECTION
Early in the GBF Cold-Flow Program, it was recognized that many factors
which are independent of granular-bed filter design, operation, and performance
would nevertheless have a major impact on the successful conduct and completion
of the program. These vital factors include:
* Selection of artificial dust(s) with properties in the same range
as typical combustion particulate but compatible with special test
requirements such as flowability, non-toxicity, and economy.
* Reliable, consistent dust feed over a range of 1 to 100 Ib/hr.
* Injection of dust against the positive pressure upstream of the GBF.
* Uniform and repeatable deagglomeration and dispersion of dust in the
inlet duct.
* Selection and/or development of equipment and techniques for sampling
particulate concentration and size distribution.
* Efficient laboratory analysis of samples including automated data
reduction.
138
-------
* Development of procedures for efficient subcontractor sampling.
Since these functions are not necessarily dependent upon the availability
of an operating GBF, it was possible to initiate this task during the fabri-
cation and erection of the filter, thereby providing lead time to address and
resolve inevitable problems associated with hardware development.
It was deemed most economical to install the plumbing upstream of the
filter for injection "calibration" as well as for its ultimate GBF service.
Techniques and operating procedures developed during the calibration phase
would thus not require "translation" to a new system since all equipment
selected, modified, and calibrated for this task would naturally be appli-
cable to GBF testing without further alteration in design or operation.
Figure 5 illustrates the facility constructed for use in the GBF Cold-
Flow Program. An existing Sutorbilt positive-displacement blower capable of
continuous operation at about 5 psig and 4000 scfm supplies air to the GBF and
auxiliary systems, including dust-injection air, media-circulation air, and
fluid-bed air (but excluding instrumentation air). This air is branched
ahead of the dust-injection station; the portion of flow which passes through
the main venturi is controlled by a pneumatically-actuated 8-in,butterfly
valve. An "Annubar" integrating pitot-static flow-sensing element is located
in the main flow branch to provide both flow indication and a signal for the
main air-flow control system.
The purpose of the dual-venturi dust-injection system is threefold:
1. It provides a means of feeding dust into the main air line (which
is typically at a gage pressure between 2 IW and 50 IW) without
special rotary airlocks or similar non-continuous feeding devices.
2. It deagglomerates the dust from its compacted storage condition
to stable particles near the intrinsic crystal size of the dust.
3. It removes very large foreign material or particles of dust which
failed to deagglomerate.
139
-------
Dust injection is accomplished in an air stream that passes through a
venturi with an inlet-to-throat ratio of six. A %-in coupling forms a "tee"
with the throat. To this coupling is threaded a hopper into which dust is
introduced by a variable-speed vibrator feeder. Dust so introduced is sub-
jected to high velocity in the venturi throat, which deagglomerates most of
the compacted material. The venturi also produces a depression in static
pressure of about 70 IW (maximum) below atmospheric, sufficient to prevent
backflow from the injection system during minor downstream upsets.
From the feed venturi, the air/dust mixture flows into an 8-inch-
diameter cyclone having a 4-in outlet and three 1-in x 16-in x %-in baffles
in the annulus. This component serves both to further the deagglomeration
process and to remove any remaining large particles. Fallout from the cyclone
is captured in a sealed — and periodically emptied — container at the
bottom of the conical section.
The main venturi (inlet-to-throat area ratio of four) produces a depres-
sion to help overcome losses in the feed venturi and cyclone, as well as a
high-velocity region in which to disperse the air/dust mixture before its
deceleration into the main duct. Downstream from the main venturi are two
sample stations, each consisting of two 3-in threaded couplings at 90°. Each
station is located in accordance with EPA recommendations (8 diameters down-
stream and 2 diameters upstream from the nearest disturbances).
Desirable properties of candidate dusts for the GBF Cold-Flow Program
include:
* Flowability (for feeding purposes) and ease of dispersion.
* Narrow size distribution.
* Particle sizes between 0.5 ym and 50 ym.
* Specific gravity between 1 and 3.
* Non-toxicity.
* Availability and economy.
140
-------
A number of commercially available dusts of suitable size were considered,
including hydrated alumina, magnesium oxide, kaolin clay, zinc, zinc oxide,
and carbon.
Kaolin clay was eliminated on the basis of in-house experience with feed-
ing problems. The zinc materials were much heavier than typical flyash, so
that data obtained with them would be of limited practical utility. Carbon
was thought to represent an unacceptable housekeeping problem.
Actual dust-injection tests were carried out on hydrated alumina (Alcoa
Hydral 710, Alcoa C-333, and Reynolds RH-730) and magnesium oxide (Basic
Chemical Magox 90 and 98-HR). All materials behaved reasonably well during
feeding, although the larger particle sizes (C-333 and Magox 90) required less
attention. The size range of magnesium oxide was quite wide. Magox 98-HR
and Hydral 710 had similar median particle sizes, but the next grade of magne-
sia (Magox 90) had a median particle size approximately twice that of C-333.
As a result of a number of trials, it was found that Hydral 710 and C-333 con-
sistently yielded median sizes of about 3 urn and 20-30 ]_im, respectively.
These results, along with considerations of availability (shelf stock in
Moraga, California), cost (19C to 26% per pound, depending upon grade and
quantity), and safety (the normal use of these grades is as the polishing
compound in toothpaste), led to a decision to use these two materials as the
basic dusts for all tests. Table I lists typical properties. In order to
""•nthesize different mean particle sizes, the materials are used as follows:
* Small Particle Size: Hydral 710 (<5 ym median).
* Nominal Particle Size: Mixture of Hydral 710 and C-333 in the
proportion 2 lb H-710 to 1 Ib C-333
(5-10 um median).
* Large Paricle Size: C-333 (>15 urn).
141
-------
PARTICLE SAMPLING
A number of manufacturers have developed continuous quantitative and semi-
quantitative particle monitors in recent years, but it was thought best in
this program to rely on methods that already have wide acceptance. This meant
that we would deal with discrete samples, would measure total loading by
conventional EPA filtration techniques, and would determine size distribution
by one of several impaction methods. Total loading was thus measured with
an EPA Method-5 sampling train (without impingers, since preliminary runs show-
ed no need for them). CPC's measurements of size distribution were made with
an Andersen 2000 Mark III impaction classifier. Size distribution was measured
only at the central axis of the air line, but experience showed that good
total-loading measurements were possible only with a six-point traverse of
both a horizontal and a vertical diameter.
To establish the consistency and accuracy of our sampling operations, we
engaged as subcontractors a sampling team from Air Pollution Technology (APT)
of San Diego. The APT team also used the EPA Method-5 train for total loading,
but a University of Washington impactor for determining size distribution.
There are three ways available to us for assessing the accuracy and repro-
ducibility of total-loading measurements. We can compare simultaneous EPA-5
measurements; we can compare the sampling results with a material balance on
dust and air flow; and we can compare with results obtained by summing up the
several sizes measured by impingers. Statistical comparisons using the latter
two approaches are shown in Figure 6. Least-squares curve fits of straight
lines through the origin demonstrate quite good agreement. Any particular
sample, of course, represents the time period when it was taken while the
material balance provides an average for the complete run. A special document
giving additional details of the calibration task, sampling procedures, and
data-processing techniques has been published.
Also employed (and shown in Figure 5) is an on-line opacity meter directly
across the GBF outlet stream. This unit, a Lear-Seigler Model PM7A, proved
142
-------
valuable in subsequent GBF operations. A stabilized reading from the semi-
quantitative output signal provided one indication (of several) that steady-
state operating conditions had been achieved in the filter, so that meaningful
particle sampling could begin. The highly responsive signal also was a prime
indicator of transient upset conditions.
DATA ACQUISITION
Aside from particulate sampling upstream and downstream of the GBF, the
bulk of operational measurements were logged by a data acquisition system (DAS)
based on a Tektronix Model 4051 computer (desk-top class).
The DAS 4051 unit, with its internal magnetic-tape cartridge, CRT screen
and 32 K bytes of memory, is shown adjacent to the control console in Figure 7.
An extended BASIC language gives easy access to the 4051 with its high-resolu-
tion CRT (1024 x 780 addressable points).
A special feature of the 4051 used in the DAS application is the general-
purpose interface bus (GPIB) available for plugged connections to an extensive
family of programmable instruments. The four external devices shown in Figure
8, for example, can be plugged to the GPIB to give a highly flexible and power-
ful DAS operating under full control of 4051 software.
Programmed (and therefore adjustable in real time) signals from the timing
generator are processed by the 4051 into convenient elapsed-time scales for
test purposes. The programmable scanner and variable-gain voltmeter are
commanded by the 4051 to read and submit measurements of 40 instrumentation
channels at adjustable periodic intervals. These are transformed by the 4051
into larger sets of performance parameters in engineering units; the latter are
logged on tape cartridges for post-test processing and display. A separate
Tektronix 4051 with hard-copy unit is available for that function.
Software for the DAS 4051 accomplishes many auxiliary functions in addi-
tion to the primary sampling/transformation/logging process outlined above.
143
-------
Most of these utilize the programmable memory and graphical-display capabilities
of the 4051. A periodically refreshed schematic diagram annotated with latest
measurements is generated on the CRT, for example, to provide quantitative
status for test conductors and observers at a glance (Figure 9 shows the
format) .
Another feature of the developed software involves storage in memory of the
most recent 51 time-slices of data from 19 selectable channels. Between sampl-
ing/logging episodes, one or more of these channels may be selected at will for
graphical display. Simple statistical summaries of the selected channel are
calculated and presented along with the plotted data. The intent is to pro-
vide observers with self-consistent trend indicators in a quasi-real-time
environment. One important use of such displays is to permit a determination
that steady-state conditions have been achieved.
144
-------
SECTION 3
PARAMETRIC TEST PROGRAM
SCOPE OF THE PROBLEM
From an academic point of view, the precise description of GBF performance
is an exceedingly complex subject involving many dimensions. The diagram of
Figure 10 is intended to convey some appreciation of this point and to facili-
tate an orderly discussion of the subject.
The purpose of a given GBF design is to remove particulate matter from a
dirty gas characterized by the six somewhat generalized quantities identified
on the left of Figure 10. One set of parameters (among many possible alterna-
tives) for characterizing the clean gas thus produced is shown on the right of
the diagram. Primary interest naturally centers on quantification of the par-
ticulate parameters. The incremental losses suffered by the gas stream are
nearly as important, however, since these represent penalties that must be
suffered if a GBF is to be incorporated in a process. Chief among those list-
ed (for most applications) is pressure loss. Relief of intolerable pressure
drop, of course, provides the original motivation for keeping the granular bed
in motion.
Size and shape of granules is a key factor in any model of granular fil-
tration, but moving the bed adds several other factors. The media flowrate, in
fact, is the parameter ordinarily used for control purposes; an increase acts
to reduce pressure drop, but with some degradation in collection efficiency.
The practical necessity of circulating media introduces a potential modifica-
tion of collection performance owing to imperfect cleaning of the media, as
does the implementation of necessary pressure gradients around a solids trans-
port loop.
For purposes of design synthesis, geometrical parameters that define the
145
-------
filter must also be regarded as independent factors in any functional per-
formance model.
At elevated temperatures, the issue of heat transfer within the GBF and
the associated impact of non-isothermal conditions on performance must be
addressed. If media and gas inadvertently (or intentionally) enter at dif-
ferent temperatures, resultant spatial distributions of temperature will
likely produce a detectable variation in GBF behavior.
The purpose of the foregoing comments is to indicate that the principal
"dependent" performance parameters of a hot, pressurized GBF may be influenced
by as many as 15 "independent" factors. In fact, the reader will have little
difficulty suggesting other candidate factors for addition to those shown in
Figure 10 (gas properties such as molecular weight, for example).
For purposes of the present ERDA cold-flow program, the list of parameters
to be varied experiementally was trimmed to the following seven:
1. Gas flow
2. Media flow
3. Particulate loading
4. Particulate size distribution
5. Media size
6. Element height
7. Element depth
Following satisfactory completion of this effort, and assuming indications
of promising performance over some region of the "independent factor space", the
general plan will be to proceed into a hot-testing program where three more
factors are added:
1. Gas temperature
2. Media temperature
146
-------
3. "Real" particulate properties as generated by combustion of coal
in a fluidized bed at near-atmospheric pressure.
Assuming positive results, a final stage of development would add the
factor of elevated pressure to those already studied.
PLANNED TEST MATRIX
The number of tests required to fully explore a seven-dimensional space in
any statistically satisfying way when each factor has three or four levels of
interest is exceedingly large. Furthermore, the disparity between any such
ideal number and a realizable number is magnified when each test is neces-
sarily tline-consuming and expensive. And unfortunately, GBF testing falls
into this category. First, primary interest is limited to "steady-state" per-
formance — that typically dictates a 1-4 hour dwell time at given operating
conditions before useful data can be recorded. During this period, the entire
inventory of media must be exchanged at least once as equilibrium concentration
profiles of captured dust are established within the bed. Once steady-state
conditions are approached, sampling of particulate concentrations can begin.
That, too, consumes a significant amount of time, particularly at the GBF out-
let, where low loadings translate into extended time intervals if measurable
quantities of particulate are to be extracted.
Time and funding constraints applied to the considerations above dictated
an appreciable compromise in test planning. It was found that the number of
separate test points that could be scheduled on a two-shift basis with allow-
ance for configuration changes, inspections, replications, and contingency re-
runs was just over 100.
Ranges of interest in the independent factors together with the number
of discrete levels in each are presented in Table II.
Test organization was facilitated by partitioning into 9 groups or sub-
experiments and selection of "nominals" for the last four factors to serve as a
147
-------
basis for single-factor excursions (see Table II). Nominal values for the
geometric factors were derived from overall scale considerations, available
air supplies, and previous experience regarding element depth/height ratio.
A plentiful stock of 2-mm Arlcite (a highly spherical, high-alumina ceramic
manufactured as a grinding product) and successful prior experience led to
the choice of nominal media.
The entire set of planned factor combinations is presented in Table III.
The first sub-experiment, a survey of variations in the first three indepen-
dent factors (readily altered without apparatus changes), has been completed
with nominal values for the other factors. Factor combinations for the 25
tests were purposely selected to be an approximate subset of an applicable
"double-latin-squares" design (for 3 factors, each with 4 levels, a full set
calls for 32 combinations).
The short series involving intermittent movement of media constitutes a
side experiment to study possible performance advantages of a mode of opera-
tion featuring variations on this approach.
Two variations in the size distribution of injected particulate were
addressed in the next two sub-experiments of 12 tests each. There followed
three sub-experiments dealing with factor-of-two changes in bed geometry. In
the final two sub-experiments, significant changes in media size will be made.
The larger size, like the nominal, consists of nearly spherical high-alumina
material; use of the thin bed for this series instead of the nominal was dic-
tated by limited supplies of the media. Gopher sand, a relatively pure silica
sand of high sphericity, was selected as a small media for the final sub-
experiment .
148
-------
SECTION 4
PRELIMINARY RESULTS
As previously stated, the test program and related data reduction and
analyses are in progress at the time of this presentation. Accordingly, the
results and discussion contained in this section are necessarily preliminary in
nature.
PARTICLE SAMPLINGS
Certain general results in this area have already been discussed; others
are documented in Reference 1. Satisfactory sampling equipment and procedures
for the application were developed (by both CPC and APT) and are also des-
cribed in the reference. Reasonable agreement between the data produced by
the two teams is demonstrated. Since the subcontractor effort is continuing,
additional comparative data will be generated and reported.
Tests by both CPC and APT teams indicate that both of the dusts used in
the program are bimodally distributed. This characteristic produces an S-
shape in a cumulative-distribution plot such as that displayed in Figure 11
(the "S" is here truncated by the inability of the impactor to characterize
the largest particulate any more precisely than "larger than 19.2 microns").
The fitted regression line is used only to facilitate quick comparisons of
samples; the distribution is clearly not log-normal. The term "median dia-
meter" thus has only a relative meaning, and very little of the material may
actually have sizes near to it.
149
-------
The display in Figure 11 is typical output from a Tektronix 4051 program
written to process CPC measurements made with the Anderson impactor. The
ordinate used is the "aerodynamic diameter" of the particles. Note that com-
pensation is made for the degree of isokinetic sampling during the collection
period.
FLOW SURVEY, LARGE PARTICULATE, AND SMALL PARTICULATE
Tests in these three sub-experiments were recently completed and will be
discussed here as one group. Pending further detailed study, this recombination
into a simple 4-factor experiment appears to be partly justified by difficulties
in test-to-test control of the size distribution of the particulate, particular-
ly in the earliest tests.
A typical display generated by the DAS is shown (reduced size) in the
several traces of Figure 12. Inspection of the traces leads to a number of
observations.
* The plugged-together DAS and Tektronix 4051 software make a highly
satisfactory unit.
* The control system for GBF airflow works well, maintaining essentially
constant levels after the momentary shutdown near ET = 30 min.
* Media circulation rate is likewise well controlled by the scheme
illustrated earlier (Figure 5).
* The initiation of dust injection at ET = 50 min is readily observed in
the opacity signal, and in the positive response of GBF pressure drop
to the event.
* A 7-minute interruption in dust feed centered around ET - 140 min pro-
duced an immediate reaction in opacity of the GBF outlet air stream.
This response, corroborated by visual observations of the exhaust,
illustrates the value of the opacity meter and the fact that GBF col-
lection (or penetration) mechanisms have a short-term component as
well as a long-term one assocaited with equilibrium distributions
within the bed.
150
-------
* GBF pressure drop also responds to the disturbance, and follows with
a transient quite similar to its earlier approach to steady state.
* The temperature traces remind us that the cold-flow facility is in
reality a "warm-flow" configuration where some heat exchange between
warm air (heat of compression) and cooler media (see fluid-bed trace)
occurs.
* Load cells supporting the inner panel show a distinct (and similar)
response to the early air-flow disturbance; response to the dust-flow
interruption is much less distinct. Differing magnitudes and long-
term trends among the three traces have not yet been explained; the
apparent indication of imbalanced lateral loading with long-term drift
will be studied.
Post-test disassemblies and inspections proved to be less of a problem than
originally supposed. Figure 13 shows one such operation in progress. The bed
side of the suspended inner panel is in view (white dust accounts for the
appearance) along with the downstream surface of the outer panel. Note the
relatively long smooth section of bed annulus above and below the active cross-
flow region.
No plugging or deposit problems were encountered with the slotted inner
panel. Some dust deposits on the bed side of the cone section at the base of
the inner panel have been noted but are not yet known to present potential
problems.
Selected steady-state data from 45 tests is given in Table IV. To aid in
characterizing the mass of data for diverse combinations of the independent
factors (i.e., the first four columns), note the following mean values for the
last three columns:
GBF pressure drop 6.95 IW
Overall collection efficiency 0.917
Outlet loading 0.039 gr/sdcf
151
-------
Additional insight is afforded by the histograms for efficiency and
outlet loading, shown in Figure 14. Here, for example, the two low-effi-
ciency cases (C1634R and C1671) are highlighted. Reference to Table IV shows
that both of these tests feature low air velocity, relatively high media flow-
rate, and low inlet loading, a combination expected to yield low efficiency.
It should be remembered that the test program is specifically designed to
impartially explore neighborhoods of suspected low performance as well as
those of anticipated high performance.
A first step in multiple-regression work on the data base of Table IV
has been performed, using the form:
Eff T = 1 - EXP (C1 + C2(VELOC) + C3(MEDIA) 4- C4(L in) + C5(d in)}, (1)
where the symbols are those of Table IV. Results of a Tektronix least-squares
solution give a correlation coefficient of 0.775 (reasonable for first attempt)
with the coefficients:
C1 = -1.400
C2 = -0.005540
C3 = 0.1793
C4 = -0.9434
C = -0.02201
Scans of the regression function with respect to each independent factor
are shown in Figure 15 along with residuals (points are entered as characters
to aid in identification with test numbers). All of the indicated trends are
reasonable and in accord with expectations.
* Efficiency improves with increased air velocity. No downward curva-
ture is evident at higher velocities to suggest any pronounced re-
entrainment, bouncing, etc.
* Efficiency increases as media rate decreases. This can be explained
by the increased concentration of captured particulate residing in
the bed, which itself becomes a collector of other particulate. The
152
-------
slightly higher bed void fraction existing for high media flow may
also produce some reduction in efficiency.
* Efficiency increases as inlet loading increases. This trend is a
corollary to the one just noted. The apparent strength of the trend
is also very encouraging since it tends to reduce the requirements
for cyclone collectors upstream of the GBF.
* Efficiency improves as the mean diameter of injected particles in-
creases. This is to be expected, of course. The surprise is that
the trend is not more emphatic than it is. The data (and theory)
bearing on this relationship will receive special attention.
It is interesting to note that the regression line indicates that' total
efficiencies in excess of 0.95 are readily attainable with reasonable operating
conditions and with the "nominal" geometry factors used. In view of the rela-
tively thin absolute depth of the bed, that is encouraging. The penalty in
pressure drop must be fully assessed, of course, to obtain a more complete
picture.
THICK BED
This sub-experiment was completed on 2 September 1977. A preliminary
set of steady-state data is given in Table V. The general improvement in
efficiency (or reduction in outlet loading) produced by doubling the bed
depth can be seen in mean values. The average outlet loading is lower by a
factor of 2.8 than that achieved with the 7.6-inch-thick bed under similar
conditions. On the other hand, the increase in average pressure drop is less
than 50 percent.
Preliminary regression attempts (with mean size of particulate excluded
as an independent factor) yielded trends in collection efficiency with respect
to inlet loading and media rate similar to those discussed above. The first
trend obtained with respect to air flow, however, was opposite to that expect-
ed, albeit with a very small indicated slope. Realistically, the result pro-
bably constitutes a non-indication; further data analysis will be needed.
153
-------
INTERMITTENT MEDIA FLOW
Addition of timer circuitry to the lift-pipe controls permitted this set
of exploratory tests to be run in an automated fashion. There were no special
surprises in this departure from normal continuous operation. Periodicity is
clearly shown in the three DAS traces of Figure 16. The opacity trace is
distinctly in phase with the programmed waveform in media flow - more pene-
tration of particulate when media is flowing than when it is not. (The
apparent level change in the opacity signal from that recorded in Figure 12
should be ignored - it was later found that a stalled purge pump in the in-
strument had allowed lenses to become dirty.) One way to view an objective of
these tests is that one wishes to discover if the opacity waveform develops a
downward bias for some on-off pattern in media flowrate. If such an indication
of improved collection efficiency can be found, one may be willing to pay the
implementation penalties. A direct penalty is the systematic ripple in GBF
pressure drop so produced since process design must consider the peaks instead
of the mean. Observe the predictable phasing of the near-triangular ripple -
the downward ramp is initiated with media flow, the upward ramp at flow cessa-
tion. Another direct penalty is the need to provide high capacity in the media
lift pipe, again to accomodate peaks instead of means.
Normal sampling operations were conducted for these four tests; total
loading by EPA methods is incorporated in the columns of Table VI. Also
tabulated is a calculated total efficiency based on the regression cited for
the data of Table IV and the measured values of the four independent factors.
The purpose of the calculation is to display an efficiency value for a hypo-
thetical test with all factors identical to the test in question but without
the intermittent media flow. The measured efficiency can then be compared with
the fitted value as one means of assessing whether the on-off pattern produced
an improvement. Interestingly, all four tests exhibit a mild improvement from
expected performance of their continuous counterparts (increments in last
column ranging from 0.8 to 4.1 percentage points). Such enhancements are
judged to be marginally significant rather than dramatic, but they may justify
deeper investigation in the future.
154
-------
With regard to an optimuir. waveform for on-off media flow, the data are
inconclusive. If the concept is to be pursued, it now seems clear that long-
duration tests featuring sequential bursts of several different waveforms
would be the most promising experimental approach. Measured response of
the opacity and pressure-drop waveforms would be the basis for comparison.
Such an approach eliminates particle sampling from the procedure, a very cum-
bersome and error-prone operation in an environment featuring periodic changes
in dust concentration.
155
-------
SECTION 5
MATHEMATICAL MODELING
The purpose of theoretical investigations in the present program is to
develop and implement a computerized model of GBF performance. If success-
ful, such an analytical tool will have a number of potential applications
and benefits including those listed below.
* The understanding of physical principles and mechanisms contri-
buting to observed filter performance will be enhanced. Just as
importantly, those mechanisms that have negligible influence may
be identified.
* Interaction between the model results and an ongoing experimental
program can build confidence and improve accuracy in the former,
and provide useful direction to the latter.
* Empirical correlation efforts for a set of experimental data can
be guided in the choice of promising forms, etc. by reference to
available computer solutions.
* The model can be extended to predict performance for new designs,
conditions, and applications.
* The model can be used to search for optimal operating conditions,
and to define sensitivities.
In addition to the preceding motivations, the transitional justification
for simulators — to generate solutions that simultaneously satisfy a large
number of interacting relationships in a consistent, accurate, and repeatable
manner — applies with special force to the GBF situation.
156
-------
A detailed presentation of model equations, program listings, and the
like is beyond the scope of this paper (a special document addressing the
subject will be published later in the contract). In this section, however,
the effort and approach will be outlined.
COMPONENTS OF THE MATH MODEL
An assumption of axisymmetry makes this a two-dimensional problem, in
which the "control volume" concept illustrated in Figure 17 is used. An
array of contiguous ring elements defined by the sequence 1234 in Figure 17
is assembled to simulate the global volume denoted ABCD. Height of the
global volume, H, exceeds the panel height (53 in) by an arbitrary amount to
account for end effects. Both cross sections are defined by rectangular
areas in a diametral plane of the GBF.
A two-dimensional adaptation of the Ergun correlation equation (Ref-
erence 2) for gas flow through a packed bed of granules is solved. The
standard form for pressure drop, consisting of additive viscous and kinetic
terms (respectively linear and quadratic with regard to local superficial
velocity), is modified to account for the axisymmetric variation of gas
velocity with radius. It is also re-expressed in terms of weight flowrates
to permit convenient application of continuity relationships.
The gas-flow model itself is a comlex problem of some interest, even in
the restricted case of a bed of uniform granules and fixed void fraction
throughout. General closed-form solutions for the two-dimensional problem
are not available since the problem is decidely nonlinear by virtue of
the kinetic term. (In the present application, the contributions of the two
terms to total pressure drop are approximately equal.) In this so-called
"clean-bed" case, the sensitive terms in the coefficients dealing with bed
void fraction (e) and granule sphericity () may be treated as fixed con-
stants .
157
-------
In the GBF application, added complexity comes from the fact that, in
general, each control volume features a unique void fraction and a unique
sphericity for its resident granules depending on the local concentration of
captured particulate. Such a non-uniform distribution of "resistance" to
gas flow produces a related distortion of the clean-bed profiles for gas
flow.
and
Relationships used to calculate the parameter modifications are:
Max
bc P
e = e - 7—— Cs
o Pbp
0.5,
1 "
1
P
4- i- °
PP
d
c
dT cs
p
(2)
(3)
where: p
P
bc
bp
C
bulk density of the media
bulk density of the particulate
ash concentration in weight of particulate
per weight of media
void fraction of clean media
diameter of the granule
diameter of the particulate
density of the ash particulate
density of the media granule
sphericity of the clean granule
A unique feature of the GBF model, however, is the set of expressions
used to describe transfer of particulate from a gas-borne state to a media-
borne (i.e., "captured") state. Neglecting electrostatic and thermophoric
effects, individual granules capture particles by at least four mechanisms:
158
-------
1) Inertial impaction
2) Interception
3) Diffusion
4) Sedimentation
From literature surveys and inputs of program consultants, candidate
expressions for collection by each of the four mechanisms have been obtained.
The expressions are held to apply for an intrinsic collection stage that con-
sists of a layer of granules oriented normal to the gas flow. Along the
direction of gas flow, the layer is thought to consist of a single granule.
Efficiency for each control volume is obtained by cascading (i.e., staging)
a relatively small number of layers.
Indications to date are that the impaction mechanism dominates the
others, particularly for larger particles, since it is proportional to at
least the second power of dp. It is also at least proportional to local gas
flow. Since each of the contributions to collection by a single layer is
much smaller than unity, the net efficiency for the layer is taken as the sum
of the contributions.
One of the least understood phenomina is re-entrainment of captured
particulate within the moving-bed environment. It may develop that empirical
reductions in one or more of the collection mechanisms will suffice to match
experimental findings. Conversely, a separate re-entrainment form may need
to be developed in order to adequately discount the collection mechanisms
over a wider range of conditions.
All of the preceding pertains to a bed of clean granules. The present
approach to describing the enhanced collection efficiency observed for a dirty
(i.e., loaded) bed is to apply the ratio AP/APO to the values obtained above.
Flow of captured particulate is modeled by reference to a concept of
"state variables" wherein the array of particulate concentrations in each
control volume seek equilibria based on continuity restrictions applied to
159
-------
inflow-outflow balances. Thus concentrations are increased by the net
efficiency of removal from entering gas streams (both radial and axial) plus
inflows of already captured particulate, and decreased by outflows of cap-
tured particulate. Concentration of outflowing captured particulate is
presumed to be identical to the existing concentration of resident particu-
late in the control volume.
At the present time, the distribution of media flows through the bed is
treated as a prescribed invariant. Uniformly distributed downward flow is
one limiting assumption but variations and radial profiles may be specified
to simulate spillage at the front face.
All of the preceding aspects of computer modeling are complicated by a
desire to account simultaneously for more than one size of particulate.
Though straightforward in principle, this added dimension adds appreciably
to the programming difficulty as well as to the cost of production runs.
COMPUTER-PROGRAM PROFILE
A diagram giving some of the nomenclature as applied to a control volume
in the interior of the global volume is included as Figure 18. Observe that
32 individual flows are treated: 4 gas flows, 4 granular-media flows, and
24 particulate flows. The larger number of partiuclate flows comes from the
presumed necessity to subdivide the particulate into at least three classes
by size. This is in recognition of the different collection efficiencies
applicable to different-sized particles. Further, it is necessary to dis-
tinguish between gas-borne particulate and media-borne particulate, since
this is the essence of the collection process in a moving GBF.
The choice of the resident weights of caught particulate (each size
class) as "state variables" for each control volume (CV) is natural since
that permits use of the capacitive form of particulate-continuity relations
as a convenient basis for time-domain integration to a convergent equilibrium
solution. Inherent in this mechanization is the assumption that media-borne
160
-------
particulate flow from a CV has the same particulate/media concentration as the
resident weights (i.e., perfect mixing within each CV).
Fixed and Initial Conditions
The array of CV cross sections forms a rectangular pattern that is sub-
divided into an arbitrary number of equal-volume rings arranged in columns
(along the axis) and rows (along the radius).
The top surfaces of the top row and the bottom surfaces of the bottom row
have boundary conditions of zero flow in the radial direction. This is the
portion of the GBF above and below the slotted or louvered panels included
in the model to account for "end effects".
Distributions of media and media flow are specified and presumed to
remain unchanged; allowances are made for the possibility of variable pack-
ing density over the field, and for simulation of spillage at the inner
(front) panel. General non-zero inflows of media-borne particulate can be
specified to simulate imperfect cleanup of circulated granules. Initial
values for the array of captured particulate in each CV are specified. The
initial value for GBF pressure-drop, a state variable, is specified.
Gas-Flow Model
A two-dimensional implementation of the Ergun flow equation is solved.
Uniform bed properties are assumed within each CV, but differing from CV to
CV as a function of the distribution of resident particulate. Because of
the nonlinear kinetic term, iteration is generally required at every time
step.
The effective void fraction for each CV is calculated as a function of
resident particulate. The effective sphericity for granules in each CV is
calculated as a function of resident particulate.
161
-------
The radial and axial components of gas flow are calculated along with the
pressure profile in an iterative scheme accelerated by a multi-dimensional
form of the Newton-Raphson method. These distributions coincide with the
current value of GBF pressure drop.
Partic'ulate-Collection Model
Individual efficiency parameters for removal of particulate from radial
and axial gas streams are calculated. It is recognized that each CV consists
of a number of instrinsic collection stages in series. There are individual
numbers for the radial and axial directions based on the ratio of CV dimensions
to granule diameter. Terms for each instrinsic stage include impaction, inter-
ception, and re-entrainment effects as modified by local void fraction.
Separate collection-efficiency parameters are calculated for each size of
particulate.
Gas-Borne Particulate-Flow Model
Flows at the inlet panel are based on loading factors applied to the
radial gas-flow distribution found above. The flow network of gas-borne
particulate is evaluated in an inlet-to-outlet marching process that pro-
ceeds by columns. Efficiency parameters are applied at the inlet edge of
each CV (conceptually) for the removal of particulate from the gas streams.
Multiple outgoing streams from any CV have identical particulate/gas concen-
trations (i.e., perfect mixing).
Media-Borne Particulate-Flow Model
This distribution depends on the fixed media-flow distribution and the
existing particulate/media concentration profile by hypothesis, so the cal-
culation is straightforward.
162
-------
Equilibrium-Solution Display
When all convergence criteria are satisfied, the program branches to a
display section that includes the following:
Overall GBF collection efficiency, total and by particulate size.
GBF pressure drop.
Particulate/media concentrations at media outlet and throughout the GBF.
Gas-flow profiles through the GBF.
Numerous secondary outputs.
is:
-------
REFERENCES
1. Combustion Power Company, Inc. Particle Sampling Facility Operation.
Report No. FE-2579-10 under ERDA Contract No. EF-77-C-01-2579.
Menlo Park, California, 1977.
2. Ergun, S. Fluid Flow Through Packed Columns, In: Chemical Engineering
Progress, Vol. 48, No. 2, 1952.
164
-------
TABLE I
TYPICAL PROPERTIES OF ALCOA HYDRATED
ALUMINA USED IN GBF COLD-FLOW TESTS
Grades
Hydra 1 710
C-333
Typical Properties
Fe?0 ...................... %
f— O
Na20 (total) .............. %
Bulk density, loose ---- Ib/ft
3
Bulk density, packed ...Ib/ft
Speci f i c gravi ty .............
On 325 mesh ................ %
2
Specific surface area ... m /g
Particle distribution,
cumulative 0
% less than 2 microns ......
% less than 1 micron .......
% less than 0.5 microns ...
Median Particle Size. .microns
64.7
0.04
0.01
0.45
8-14
16-28
2.40
0.04
6-8
100.0
85.0
28.0
0.7
65.0
0.01
0.004
0.15
44.0
77.0
2.42
1.0
I
6.5-9.5
1 As determined by electron microscope on a weight basis,
2 Not available from manufacturer.
165
-------
TABLE II
RANGE SUMMARY FOR INDEPENDENT FACTORS
Independent Factor
Description
Air flow
Media Flow
Parti cul ate loading
Participate size
distribution
Media size
Element height
Element depth
Selected Form
Approach Velocity
Ib/granules/lb air
gr/sdcf
mean diameter
per Andersen
granular diameter
per sieve analysis
panel dimension
bed differential
radius
Range
40-160 fpm
0.4-1.6
0.25-2 gr/sdcf
3-15 microns
0.8-3.5 mm
26.5-53 in
3.8-15.2 in
No.
of
Level s
4
4
4
3
3
2
3
Nominal
Value
7 microns
2 mm
53 in
7.6 in
166
-------
TABLE III
PLANNED ERDA COLD FLOW TESTS
Page 1 of 2
Operating Conditions
Participate Media
Filter
Test No.
Approach
Velocity
(fpm)
Media
Circulation
Rate
(tmedia/lair)
Loading
(gr/sdcf)
Nominal
Size
(vm)
Size
(mm)
Bed
Thickness
(inches)
Panel
Height
(inches)
Comments
FLOW SURVEY
C1610
C1611
C1612
C1613
C1614
C1615
C1616
C1617
C1618
C1619
C1620
C1621
C1622
C1623
C1624
C1625
C1626
C1627
C1628
C1629
C1630
C1631
C1632
C1633
C1634
40
80
40
120
120
160
80
160
80
120
40
160
40
80
40
120
120
160
80
160
80
120
40
160
40
INTERMITTENT MEDIA
C1635
C1636
C1637
C1638
LARGE
C1640
C1641
C1642
C1643
C1644
C1645
C1646
C1647
C1648
C1649
C1650
C1651
SMALL
C1660
C1661
C1662
C1663
C1664
C1665
C1666
C1667
C1668
C1669
C1670
C1671
80
80
80
80
PARTICULATE
160
120
120
40
160
80
80
40
80
40
80
40
PARTICULATE
120
120
120
120
160
80
160
80
80
40
80
40
1.2
0.8
0.8
0.8
1.6
1.6
1.2
0.4
0.4
0.4
0.4
1.2
1.2
1.6
1.6
1.2
0.8
1.2
1.6
0.4
0.4
0.4
0.4
1.6
1.6
MOVEMENT
1.0
1.0
1.0
1.0
1.6
1.6
0.4
0.4
0.4
0.8
1.2
1.2
1.6
1.6
0.4
0.8
1.6
0.4
1.2
0.8
1.2
1.2
0.8
0.8
0.4
0.8
1.6
1.2
1.0
2.0
0.5
2.0
2.0
2.0
1.0
1.0
0.5
0.25
0.4
0.25
0.25
2.0
0.5
2.0
1.0
2.0
0.5
0.25
2.0
0.25
0.25
1.0
1.0
1.0
1.0
0.5
0.25
2.0
0.25
1.0
2.0
0.25
0.5
1.0
2.0
0.5
1.0
1.0
2.0
0.25
UO
1.0
'.ft
1.2s
0.5
0.25
2.0
0.25
0.25
15
15
15
15
15
15
15
15
15
15
15
15
7.6
7.6
7.6
7.6
7.6
7.6
7 6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7 6
7 6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
Mm. between media movement
3
6
9
12
167
-------
TABLE III
PLANNED ERDA COLD FLOW TESTS
Page 2 of 2
Operating Conditions
Test No.
Approach
Velocity
(fpm)
Media
Circulation
Rate
Umedia/lair)
Loading
(qr/sdcf)
Participate
Nominal
Size
(urn)
Media
Size
(m)
Fi
Bed
Thickness
(inches)
Her
Panel
Height Comments
(inches)
REDUCED HEIGHT
C1680
C1681
C1682
C1683
C1684
C1685
C1686
C1687
C1688
THICK BED
C1690
C1691
C1692
C1693
C1694
C1695
C1696
C1697
C1698
THIN BED
C16100
C16101
C16102
C16103
C16104
C16105
C16106
C16107
C16108
120
160
80
80
40
80
160
80
40
120
160
80
80
40
80
160
80
40
120
160
80
160
40
80
40
80
40
1.2
1.6
1.6
0.4
1.2
0.8
0.8
1.2
0.4
0.8
0.8
0.8
0.4
1.2
1.6
0.4
1.2
0.4
1.2
1.2
0.4
0.8
0.8
0.8
1.2
1.6
1.6
1.0
1.0
2.0
1.0
2.0
0.5
0.25
0.25
0.5
0.5
1.0
2.0
1.0
2.0
0.5
0.25
0.25
0.5
0.25
0.5
1.0
0.25
2.0
0.5
1.0
0.25
0.5
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
2
L
2
2
2
2
2
2
2
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
15.2
15.2
15.2
15.2
15.2
15.2
15.2
15.2
15.2
3.8
3.8
3.8
3.8
3.8
3.8
3.8
3.8
3.8
26.5 Blank bottom 1/2 of both panels
26.5
26.5
26.5
26.5
26.5
26.5
26.5
26.5
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
53
LARGE MEDIA
C16110
C16111
C16112
C16113
C16114
C16115
C16116
C16117
C16118
C16119
C16120
C16121
160
40
120
120
160
80
160
80
80
40
80
40
0.4
1.6
0.8
0.4
0.8
0.8
1.2
1.6
1.2
0.4
0.4
1.2
0.25
1.0
0.5
1.0
1.0
2.0
0.25
0.5
1.0
2.0
0.25
0.5
7
7
7
7
7
7
7
7
7
7
7
7
3.5
3.5
3.5
3.5
3.5
3.5
3.5
3.5
3.5
3.5
3.5
3.5
3.8
3.8
3.8
3.8
3.8
3.8
3.8
3.8
3.8
3.8
3.8
3.8
53
53
53
53
53
53
53
53
53
53
53
53
SMALL MEDIA
C16130
C16131
C16132
C16133
C16134
C16185
C16136
C16137
C16138
C16139
C16140
C16141
120
40
80
120
160
80
160
80
40
160
40
80
1.2
1.6
0.4
0.4
0.8
0.8
1.6
1.6
0.8
1.2
0.4
1.2
2.0
0.25
0.5
0.25
0.5
1.0
1.0
2.0
2.0
0,5'
0.5
0.25
7
7
7
7
7
7
7
7
7
7
7
7
0.8
0.8
0.8
0.8
0.8
0.8
0.8
0.8
0.8
0.8
0.8
0.8
7.5
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
7.6
53
53
53
53
53
5j
53
53
53
53
53
53
16U
-------
TABLE IV
SELECTED STEADY-STATE DATA
Note: See Table III for Test Configurations
TEST
UELOC MEDIA L ir, d in Del P Eff T L out
01611
C1613P
C1614
C1615
C1616RB
C1617
C1619
01629
C1623
C1624
01625
01626
01627
01*23
01630
01631
01632
C1633
C1634P
C1640
C1641
01642
01643
C1644P
01645P
C1646
01647
01648P
01649
01650
01651
C1660
01661
ng62
_ . 01. »
01664
01665
01666
C1667R
C1667RA
C1667PB
01663
01669
01679R
01671
Legend
74.
119.
119.
156.
80.
158.
149.
72.
83.
61.
129.
119.
157.
34.
93.
119.
61.
155.
79.
162.
120.
119.
69,
162.
SO.
73.
65.
79.
63.
86.
57.
119.
126.
120.
118.
159.
80.
160.
79.
80.
88.
80.
60.
32.
59.
0
1
3
•}
0
3
5
6
0
£
0
f
f.
9
4
9
0
9
4
er
0
2
6
7
9
3
-j
f
9
9
5
8
4
3
e
i
9
5
6
2
1
1
5
9
9
3
0.
0.
1.
1.
1.
0.
0.
0.
1.
0.
1.
fl.
1.
1.
0,
0.
0.
1.
1.
1.
1.
0.
0.
0.
0.
1.
1.
1.
1.
0.
0.
1.
e.
i.
0.
i.
i.
0.
0.
1.
0.
0.
0.
1.
1.
Veloc..
Medic:
L
d
DC
in '
in:
!l P
869
734
367
436
128
368
442
248
fi^S
196
572
348
149
684
239
503
655
340
082
451
371
599
523
303
724
591
253
472
580
505
685
528
622
090
781
052
103
612
750
083
353
466
719
428
082
Air
2.255 2.87
0.597 9.75
1.103 S.27
0.769 10.94
0.538 4.45
0.536 3.27
0. 129 27.97
0.150 4.22
0.929 35.92
0.725 3,09
1.359 5.34
0.810 2.36
1.920 5.72
0.53^ 1 Qcri
0. 164 44.04
0.320 3.05
1.618 1.58
0.253 2.31
0.190 1.87
0.607 28.01
y.2'13 29.29
1.552 29.61
9.142 50.37
0.655 1.63
0.927 35.65
0.183 39.05
0.411 35.65
0.617 33.45
1.051 43.91
9.920 31.28
0.653 35.93
9.514 1.74
1.232 6.12
0.379 1.97
0.449 1.70
8.479 2.97
1.439 4.94
0.149 1.97
0.614 1.91
0.271 1.96
0.592 1.22
0.199 2.99
1.5S2 4.05
9.126 1.79
3.250 1.51
4
9
11
14
6
10
8
1
c
sj
8
^
=;
ft
-?
4
9
17
12
4
Q
7
3
TI
-?
<
2
5
4
1
1
2
r
4
14
5
5
7
3
6
6
6
8
3
11
6
4
approach velocity,
.25
.06
.22
.35
.04
.33
.33
.99
.70
.05
.09
.24
.73
'•' 1
.63
.03
.35
.21
.62
.59
.29
.49
.64
.85
.80
.39
.89
.99
.65
.82
.03
.74
.91
.17
.83
.33
.30
.36
.35
.67
.79
.86
.28
.02
.07
fpm
Media flowrate, Ib/media/lb
0
e
0
o
9
0
9
e
9
0
9
9
0
o
0
0
0
0
n
0
0
0
9
ft
0
0
9
0
0
e
0
0
0
Q
0
0
e
0
0
0
0
9
0
0
0
.964
.950
.937
.948
.890
.970
.907
.860
.971
.349
.971
.917
. 920
351
.' 902
, 363
. ^S^
.831
.702
.960
. "330
.976
.944
. ^8fi
.988
.967
.973
.950
.921
.954
.940
.887
.976
.988
.906
.915
. ^62
.893
.302
.815
.943
.329
.976
.341
.712
9
0
0
9
0
0
9
0
0
0
0
9
0
(\
6
0
0
0
0
0
0
0
0
e
0
&
9
0
0
0
0
0
e
8
0
0
0
9
0
0
0
0
0
0
0
.981
.030
.079
.940
.059
.016
.012
.021
.027
.049
.054
.067
.832
f! 5 '5
.'016
.044
.063
. 930
.055
.024
.915
.033
. 0tt£
.013
.911
.006
.011
.031
.083
.042
.039
.058
.029
.034
.942
.040
.955
.915
.069
.050
.034
.034
.038
.029
. 072
air
Inlet loading, gr/sdcf
Median diameter of
Eff T:
L
out
GBF
pressure drop,
Overall collection
inlet dust,
IW
microns
efficiency
: Outlet loading, gr/sdcf
-------
TABLE V
SELECTED STEADY STATE DATA
FOR THICK BED SUBEXPERIMENT
Note: See Table III for Test Configurations
: TEST
•
:C1690
:C1691
:C169lH
:C1691B
:C1692
:C1693
:C1694
IC1695
:C1696
IC1693
UELOC
120.3
157.1
81.5
121.5
82.2
81.5
42.5
79.9
161.2
42.5
MEDIA
0.763
0.739
0.719
0.711
0.672
0.341
1.001
1.544
0.347
1.423
L in
0.102
0.922
0.499
0.377
0.796
0.372
0.756
0.370
0.221
0.554
d in
5.50
3.25
3.40
2.68
2.25
1.81
1.84
1.69
1.62
1.54
Del P
4.03
23.09
5.18
15.52
19.01
14.81
3.69
2.85
12.51
1.99
Eff T
0.863
0.989
0.966
0.979
0.994
0.995
0.962
0.935
0.973
0.948
L out
0.014
0.010
0.017
0.008
0.005
0.002
0.029
0.024
0.006
0.029
Mean values:
10.27
0.960
0.014
Legend
Veloc: Air approach velocity, fpm
Media: Media flowrate, Ib/media/lb air
L in: Inlet loading, gr/sdcf
d in: Median diameter of inlet dust, microns
Del P: GBF pressure drop, IW
Eff T: Overall collection efficiency
L out: Outlet loading, gr/sdcf
170
-------
COLLECTION EFFICIENCY FOR
INTERMITTENT FLOW OF MEDIA
Note: See Table III for Test Configuration
Tot. Fitted Inc.
Test Veloc. Media
1635R 80.20 0.663
1636R 79.00 0.827
1637 80.50 0.74
1638R 79.36 0.747
Legend Veloc:
Med i a :
L in:
d in:
Eff T:
Fitted Eff:
L In D In Eff. Eff. Eff.
0.572 6.32 0.951 0.910 0
1.239 12.27 0.964 0.956 0
0.344 2.90 0.892 0.878 0
0.655 4.31 0.940 0.912 0
Air approach velocity, fpm
Media flowrate, Ib media/lb air
Inlet loading, gr/sdcf
Median diameter of inlet dust, microns
Overall collection efficiency
Result of substituting values of inde-
.041
.008
.014
.028
pendent factors into regression function
Inc. Eff:
generated for continuous-flow data of
Table IV
(Tot Eff) - (Fitted Eff)
171
-------
MEDIA INLETS (4)
CLEANED GAS
Figure 1 GBF concept.
172
-------
DISENGAGEMENT VESSEL
PARTICLE-LADEN AIR
(TO BAGHOUSE)
DIRTY AIR
FLUIDIZED BED
/-FLUIDIZING AIR
MEDIA RETURN PIPE
CLEANED AIR
FRONT PANEL
FILTER BED
OUTLET PANEL
MEDIA LIFT PIPE
MEDIA OUTLET PIPE
TRANSPORT AIR
EJECTOR AIR
Figure 2 GBF media loop.
173
-------
END
VIEW
1/8 in
i
) C )
J ( )
: ) c i
^l.OOin.
1.263 in
c
c
c
~5 . 0 ?fi
_J t 1
_ ' T
J
PITCH
(a) Slot Pattern for Inner Panel with 2mm Media
0.02" to 0.03" SLOT OPENING
\
END
VIEW
^1.00 in^
1.714 in
^
k-
0.375
t
'b) Louver Pattern for Outer Panel
Figure 3 Typical panel slot and louver patterns.
174
-------
LOAD CELL
LOCATION
LOAD
CELL-
OUTER
PANEL
MEDIA
PANEL SUPPORT
SHAFT
PRESSURE
SENSORS (12)
GBF SENSOR LOCATIONS
JACK PLATE
SET UP PLATE
LOAJLCELL DETAIL
TRANSDUCER
ADAPTER
PRESSURE _SEN_SO_R DETAIL
Figure 4 Load cells & media pressure sensors.
175
-------
PRESSURE
SET POINT
LEGEND
(V) PRESSURE TRANSDUCER
(,jj .IP TRANSDUCER
(FT) FLOW TRANSDUCER
Figure 5 GBF cold-flow program facility.
176
-------
+ 1.48
3 +1.28
£>
T3
£ +1.00
I-
01
z:
£ +0.80
UJ
o
•ZL
cC
I +0.60
a-
Q
g
^ +0.40
<
h-
O
h-
+6.20
,8.80
Slope = 0.97
Standard error = 0.027
0 +2.00
>> +1.50
+1.00
+ 0.50
0.00
xx
Slope = 1.089
Standard error = 0.09
8.00 +0.50 +1.98 +1.50
TOTAL LOADING BY EPA METHOD-5, gr/std dry cu ft
Figure 6 Comparison of EPA method-5 measurements
with other measurements.
177
-------
Figure ? GBF data acquisition system, & control console.
178
-------
1
KEYBOARD
CATHODE
RAY TUBE
t
PROCESSOR
&
MEMORY
1
MAGNETIC
TAPE
CARTRIDGE
u TEKTRONIX 4051
GENERAL PURPOSE
INTERFACE BUS
DIGITAL
VOLTMETER
I
SCANNER
TIMING
GENERATOR
MAGNETIC
TAPE
CARTRIDGE
40
DATA
CHANNELS
Figure 8 Programmable data-acquisition system configuration.
179
-------
ERDft Cold Flow Model
Test ID
ET= 9
Q.0 psig
0.0 ps ig
__0.8 psid
psiji
Figure 9 Data acquisition system CRT display.
180
-------
LOADING
SIZE
DISTRIBUTION
PHYSICAL
PROPERTIES
PRESSURE
TEMPERATURE
FLOW
MEDIA MEDIA MEC
FLOW SIZE CLEAf
AND SHAPE
1
)IA ME
LINESS TEMPE
DIA GAS SEALING
OTURE PARAMETERS
N. MOVING BED /
PARTICULAirX
DIRTY
SIDE
I GAS ./
GBF
FUNCTIONAL
RELATIONS
( / DESIGN f
1 GEOMETRY
/PARTICULATE
CLEAN
SIDE
\ GAS
\
LOADING
SIZE
DISTRIBUTION
PRESSURE LOSS
TEMPERATURE LOSS
FLOW LOSS
ELEMENT
HEIGHT
ELEMENT
DEPTH
INLET PANEL
RADIUS
MEDIA-RETENTION AND
HEAT-LOSS PARAMETERS
id Diagram of GBF performance dependencies.
181
-------
** ERDA COLD-FLOW ANDERSEN IMPACTOR DATA **
TEST DATE:97/27/77 TEST STARTING TIME: TEST ID:Ci64Z_
50i i i i i i i—7-1 i CUMUL.
STAGE Dp, WT '/.
NO. Micron
/
/
f
J
/
/
'
/O
/
/
V
<
<
y
/
/\
/
^
>
,/
if
'
/
/
/
1 2 5 10 30 50 70 90 95 98 9i
CUMULATIUE WEIGHT PERCENT LESS THAN Dp
19.2
12.0
8.16
5.57
3.59
1.81
1.12
0.74
81.05
80.39
79.74
79.08
77.12
69.93
46.41
25.49
B
C
D
E
F
G
H
I
MEDIAN PARTICULnTE
SIZE FROM INTERCEPT
OF THE FIT,
Microri = 1.641
TOTAL LOADING,
gr.-sdcf = 9.W23 (
STANDARD CONDITIONS
=70 F, 1 atr-i )
PERCENT OF ISO-
KINETIC SAMPLING,
= 151
NOTE: DP=CALCULATED 50* EFFECTIUE CUTOFF DIAMETER OF PARTICULATES WITH
AN ASSUMED DENSITY OF 1 g.'cc ( DUST = ALCOA C-333 >
Figure 11 Andersen impactor data.
182
-------
Figure 12 Typical DAS record.
168.
140.
120.
108.
80.
68.
48.
20.
8.
TEST: ci6«3 DATE:
INLET APPROACH VELOCITY '
Tine:
20 40 68 88 188 128 148 160 188 288 228 248 2
E T i n 1 r. >
TEST: C1663 DATE: 8-5 77 TIME: 1337
MEDIA CIRCULATION RrtTE«' Ib-mn'
380.8
258.0
208.8
158.0
100 0
50.8
0.8
A
b
.
i
.•—-"••
~~*~
rJ~ -f-
uj-li ..
- YVJ
8 28 48 68 88 108 128 146 160188 208 220 248 268
ET 'HiiO
FLUID BED PRESSURE DROP > IW'
18.8
16.8
14.8
1? 0
10.0
8.0
6.8
4.0
2.0
0.0
0
1 I
l\T
V
*^S
»
"*v/\.
f\^^
f^r^
i_^ fc
r- •^'t
ni^v,
•v^s
^/Vi
^'Orf
^^v
20 48 60 80 100 120 140 160 188 200 220 240 2
ET i'nin>
30.0
25.0
20.0
TEST: Cl«63
OUTLET OPACITY
tiftTE: 8-5/77
TIME: 1337
10.0
5.8
b.6
C
P^^
0 28 48 68 38 160 128 148 168 180 280 228 248 268
FILTER PRESSURE DROP .IK .
10.0
9.0
8.0
7.0
6.0
•>.e
4.0
3.0
?.0
1.0
R.ft
fa '
/
•W
^•—v
\
J/'
*
-~***>
11s*-
1
I
29 48 60 80 180 120 148 160 180 200 220 240 260
ET
-------
Figure 12, continued.
C'HTE: 3 •> V
AMBIENT TEMPERMTUPE • F < • 1
FILTER INLET TEHPERnTUPE •F •..•2
FILTER OUTLET TEHPEPHT'JPE • F >. • 3 •
FLUID' PEC- TEMPEPnTURE 'F' '4
TIME: 13V
20 46 60 88 100 l-iO MS ltd ISO i'00 £20 240
650
TEST: Cltt'3
INLET SCREEN LUHt1 M
INLET SCREEN LOAD &
INLET SCREEN L0nt- C
Tint:
ib
Ib
2(j 40 60
S0 100 120 140 lt'0 ISO 200 220 240 260
ET ' Ml Tl'1
184
-------
Figure 13 Post-test disassembly of GBF.
185
-------
Overall Collection Efficiency
6.7 0.75 0.8 0.85 0.? 0.95 1
Eff T
16
14
10
8.04
L out
Outlet Loading
0.08
Figure 14 Histograms for final two columns of Table IV,
186
-------
Media Flowrate = 1 1ta 'lb
Inlet Loading = 1 gr/scf
Inlet [>ian. = 3 nicrons
Eff T
0.3
0.7
100
Approach «elocity.
£00
Approach velocity = 1^0 fpti
«.y
Eff T
0.8
0.7
Media rate = 1 Ita per lb
Inlet dian. = 1 Microns
e.s i
Inlet loading. 31-'sc f
1.5
Figure 15 Regression for GBF total efficiency
Typical DAS record
167
-------
Approach velocity = 120 fpn
6.9
Eff T
0.3
0.7
Inlet Loading = 1 gr/scf
Inlet DIQH. = 3 nicroris
8.5 1 1.5
Media rate. 1b per Ib air
Approach Uelocitu = 12
Eff T
8.8
8.7
Media rate = 1 Ib per Ib
Irilet loading = 1 gr-'sc f
10 >6 30 40
Mean diai-ieter of inlet ctust. Micron-
Figure 15, continued.
1H8
-------
TEST: C1638R DATE: B-'ll-
MEDIA CIRCULATION RrtTE\lb'HjTt>
TIME: 8545
300.0
250.0
203.3
150.0
100.0
50.0
0.0
0
80.0
75.0
70.0
65.0
60.0
55.0
50.0
0
20.0
18.0
16.0
14.0
12.0
10.0
3.0
6.0
4.0
2.0
0.0
J
20
OUTLET
M
40
OPACITY
L
N\t
K
JV/vJ
fctf
A
OP
!
20 40
FILTER PRESSURE
^-^~>
s^"*1
ET -'MIT
Ilk
vj
I
I
'
HI
60
DROP
<"^ --
j^
/I
\j
f
IA
80
UW
•^v^ /
^
120
i
/il
j
V
I
\
100
ET
-------
GLOBAL MODEL (ABCD)
VOLUMETRIC ELEMENT (1234)
Figure 17 Control volumes.
190
-------
INTERIOR CONTROL VOLLW (TOTAL VOtUMt" « V)
ONE OF A RECTANGULAR ARRAY HAVING N ROWS AND M COLUMNS
1*1, i
CV ROW INDEX
CV COLUMN INDEX
PARTICULAR SIZE INDEX
,: RADIAL WEIGHT FLOWRATE OF MEDIA
. AXIAL WEIGHT FLOWRATE OF HEDIA
GAS PRESSURE
RADIAL WEIGHT FLOWRATE OF GAS
AXIAL WEIGHT FLOWRATE OF GAS
RADIAL WEIGHT FLOWRATE OF GAS-BORNE PARTICL'.ATE
AXIAL WEIGHT FLOWRATE OF GAS-BORNE PARTICIPATE
RADIAL WEIGHT FLOWRATE OF KEDIA-BOW PARTICIPATE
AXIAL WEIGHT FLOWRATE OF MEDIA-BORSE PARTICUATE
Figure 18 Nomenclature for computer model
of GBF performance.
191
-------
CERAMIC FABRIC FILTRATION AT HIGH TEMPERATURES AND PRESSURES
By:
M. Shack!eton, J. Kennedy
Acurex Corporation/Aerotherm Division
Mountain View, CA 94042
193
-------
CERAMIC FABRIC FILTRATION AT HIGH TEMPERATURES AND PRESSURES
By:
Michael Shackleton and Jeffrey Kennedy
Acurex Corporation/Aerotherm Division
Mountain View, California 94042
Barrier filtration using ceramic fiber filters offers a promising
solution to the problem of controlling particles in the high-temperature,
high-pressure environment. Industrial experience has proven this technique
is capable of high efficiency particle control, including fine particles,
in near ambient temperatures and pressures. Examining those particle remov-
al mechanisms which apply to barrier filtration indicates that a small
decrease in efficiency should be expected at high temperatures and pres-
sures.
This is primarily caused by a reduction in the effectiveness of the
mechanism of inertial impaction. This reduction can be compensated for in
the design of the filter medium and in the design of the filter system.
Ceramic fibers are available which have smaller diameters (3 ym) than con-
ventional fibers used for filters (10 to 20 ym). Analysis indicates that
using these fine diameter fibers should make it possible to use filter media
having weights less than or, at most, equal to conventional media.
This paper reports on work being performed under EPA Contract 68-02-
2169 to demonstrate the feasibility of high-temperature, high-pressure parti-
cle control by filtration. A description of the high-temperature and pres-
sure media test facility is presented along with test results to date.
194
-------
CERAMIC FABRIC FILTRATION AT HIGH TEMPERATURES AND PRESSURES
INTRODUCTION
Many advanced technology processes currently being developed require
removing particles from high-temperature and pressure gas streams. An objec-
tive of developing these processes is to increase coal use by making it econo-
mically efficient and environmentally safe. These processes, such as pres-
surized fluidized-bed coal combustion, involve expanding the high-temperature
and pressure gases across a turbine to generate power to produce electricity.
Such applications require removing particulate flyash from the gas streams
before expansion across the turbine. Techniques to accomplish the required
particle control have not yet been demonstrated.
Under normal environmental conditions, barrier filtration is an effec-
tive method of achieving the required level of particle control. However, at
high temperature (815 C) and pressure (10 atm), barrier filtration and other
conventional particle control methods are limited by materials capable of sur-
viving in the environment and by effects of changes in gas properties.
Under EPA Contract 68-02-2169, Aerotherm is investigating the suitability
of commercially-available ceramic fiber filters for high-temperature filtra-
tion. This work is sponsored by the Particulate Technology Branch of the
Industrial Environmental Research Laboratory at Research Triangle Park,
North Carolina.
Major goals of this program are to:
• Design and build a filter media test facility capable of operating
at 815 C and 10 atm pressure
• Test available ceramic fiber forms (woven cloths, felted mats) to
determine if any can survive mechanical displacements and accelera-
tions likely to be encountered in online cleaning of high-
temperature filter applications
195
-------
• Develop preliminary performance data for those configurations
which appear most promising for high-temperature filter applications
• Make recommendations based on the experience and data collected
TEST PLAN
To provide the data needed to justify further development of high-
temperature, high-pressure barrier filtration equipment, Aerotherm has out-
lined the following test plan:
• Obtain and classify ceramic media candidates. A survey of available
ceramic filters and materials has been made. The materials avail-
able are generally made for insulation applications, and those that
are not rigid structures can be catagorized into three groups: (1)
Woven fabrics, produced by several companies are made from yarns
produced from a continuous length of ceramic fibers. They are
available in various weaves, are flexible, and exhibit good strength
relative to other ceramic fibrous structures. (2) Ceramic papers
are available which are constructed from short fibers generally held
together with binders. These structures are characterized by a
relatively higher packing density of fibers and poor mechanical
strength compared to other ceramic fibrous structures. (3) Ceramic
felts consisting of relatively long fibers are available. These
materials are relatively porous mats held together by randomly inter-
mixing the fibers. They are flexible and exhibit reasonably good
strength.
• Mechanical Screening Tests — In the first phase of the test program,
the various candidate filter media are subjected to "filter cleaning
loads" to determine their relative strength. These tests are being
performed on samples shaped into tubular bags. The tests are per-
formed at an air-to-cloth ration of 5 to 1 (2.54 cm/sec) at high
temperature and pressure in the presence of flyash. Samples are
initially subjected to reverse flow cleaning cycles, but the primary
test is pulse-cycling. While maintaining forward flow, the sample is
shook using 5 to 10 pulses per minute from a reservoir pressure
196
-------
Theory
of 1100 kPa (160 psig), with a duration of 100 ms. The relative
strength of each sample is determined by the number of cycles it
withstands. If several thousand cycles are required to cause fail-
ure, then it is probable that the filter would last a long time,
provided the cleaning technique is perfected.
Parameter Variation Tests — Filter performance data are being
developed from these tests. Those samples which withstand filter
cleaning loads will be given longer tests to investigate their
operating limits in a filtration application. Flyash will be metered
into the gas stream at a controlled rate, while monitoring pressure
drop of the filter. Reverse flow and/or pulse cleaning will be used
to maintain a reasonable stabilized pressure drop. Pressure drop
and efficiency will be determined as a function of air-to-cloth
ratio, cleaning method and energy level, time and dust-loading.
These tests will provide a basis for selecting near-optimum operat-
ing conditions for each sample tested.
Life Tests — Our tests may identify media samples and operating
limits which appear capable of meeting the proposed requirements
for high-temperature and pressure barrier filtration. If so,
extended tests will be given to attempt to determine their long-term
performance. This work is scheduled for completion in August 1978.
Preliminary results from the mechanical screening tests are included
in this report.
Barrier filtration with available ceramic fibers is likely to be a good
technique for particle control at high temperature and pressure. To illus-
trate why this is true, a short review and discussion of barrier filtration
theory is helpful.
Figure 1 is taken from a report titled "Effects of Temperature and
Pressure on Particle Collection Mechanisms: Theoretical Review" by Seymour
Calvert and Richard Parker (EPA-600/7-77-002), January 1977. This figure
shows a calculated fractional efficiency curve for a fiber bed. Minimum
197
-------
efficiency is indicated for a particle size of about 0.5 ym. The dip in the
curve occurs because of the interaction of the three collection mechanisms
which apply to barrier filtration. These mechanisms are direct interception,
diffusion, and inertial impaction. For particle size less than about 0.5 ym,
collection by diffusion is increased, improving the efficiency of the filter
bed. For particle size larger than about 0.5 ym, collection by inertial
impaction is improved, increasing the collection efficiency of the filter
bed. It should be remembered that this curve applies only to initial per-
formance of a clean fiber bed. That is, it does not include the increased
collection efficiency that results from the filtration of the accumulating
dust cake. Note also that the No. 3 curve indicates that the inertial impac-
tion parameter for high-temperature and pressure conditions should show a
small decrease in performance. To understand the magnitude of this effect
we can compare the performance of standard filter media when tested with
Dioctylphthalate smoke (D.O.P.) to its performance when tested after a stabi-
lized dust cake has been developed. A D.O.P. smoke penetration test is a
standard test to measure the efficiency of high performance filters such as
those used to filter "Clean Room" air or to collect biological contaminants.
This test measures how efficiently a filter removes a 0.3 ym diameter D.O.P.
smoke particle. Woven or felt filter media of the type commonly used for
industrial filters will collect only 10 or 20 percent of 0.3 ym D.O.P. smoke.
Yet, after developing a dust cake, these same filter media will collect sub-
micrometer particulate at an efficiency of greater than 90 percent. Thus,
compared to the changes in performance which take place in a filter media
during the conditioning process, the changes predicted as a result of high-
temperature operation are small.
Available ceramic fibers offer unique advantages for filtration, since
many of these fibers have finer diameters than conventional filter fibers.
Conventional fibers are usually 10 or 20 ym in diameter, while ceramic fibers
are available with average diameters of only 3.0 ym.
Collection efficiency can be improved simply by making a filter bed
thicker, thus increasing the basis weight of the filter (its weight per unit
area). However, to achieve high collection efficiency in this way can lead
to high operating pressure drops. Collection efficiency can also be
198
-------
increased by reducing the fiber diameter, which can result in decreased basis
weight and filter bed thickness. The importance of fiber diameter is illus-
trated in the following equations which describe the three primary particle
collection mechanisms applicable to barrier filtration.
d
Interception parameter K = — ^-
1 df
C,P d 2U
Impaction parameter K = — P °
,
P 9ygdf
C' kT
Diffusion parameter K-, = -= - , • • j
v d 3iry d U d..
g P g f
These equations describe the collection mechanisms, but are not collection
efficiency equations. However, when expressed as above, an increase in any
of the mechanism parameters (K , K , K,) will result in an increase in
efficiency.
The interception parameter is not a function of temperature and pressure,
but it is a function of fiber diameter. Changing from a 20-ym fiber to a 3.0-ym
fiber will increase the interception parameter by a factor of 6.67 times.
The impaction parameter is a function of temperature and pressure,
essentially through changes in the gas viscosity (yg) . For air, increasing
temperature from 20 to 815 C increases viscosity by about 2.5 times. This
reduces the impaction parameter by a factor of 1/2.5 or 0.4. But, the change
in fiber diameter from 20 ym to 3.0 ym increases the impaction parameter by
6.67 times. The net effect of the two changes is to increase the impaction
parameter by 2.7 times.
The diffusion parameter is a function of temperature and pressure through
changes in the ratio of (C'T/yg) . When operating at 815 C and 10 atm pressure,
this ratio tends to remain unchanged or to increase slightly. But, the dif-
fusion parameter is also a function of fiber diameter anr1. a change in fiber
diameter from 20 ym to 3.0 ym will increase the diffusion parameter by
6.67 times.
199
-------
From the above discussion it is evident that if we make a filter using
3.0~ym diameter ceramic fiber (which is commercially available), it is reason-
able to expect that even at high temperature and pressure this filter will
have high collection efficiency without excessive filter bed thicknesses or
basis weights. Using the method developed by Torgeson, it is possible to
calculate collection efficiency for a given particle size and fiber bed para-
meters. This calculation was performed for a 0.5-ym diameter particle with
a density of 1.5 g/cm3 (as measured at the Exxon Miniplant), for gas tempera-
ture of 815 C, and pressure of 10 atm. A fiber bed composed of alumina fibers
with 3.0-ym diameter and fiber density of 2.8 g/cm3 was assumed. Results of
this analysis for two filtration velocities and two solidities (a - the volume
fraction of the fiber bed which is solid) are plotted in Figure 2. This
analysis indicates that a 3.0-yro diameter ceramic tiber filter bed with a basis
weight of 500 to 600 g/m will collect submicrometer particulate with an
initial (clean) efficiency of about 90 percent at high temperature and pres-
sure. Recall that a typical industrial filter media (20-ym fibers) would
collect such particles at only 20 percent efficiency for the same basis
weight. Another way to look at this is to note that to achieve collection
efficiency comparable to commercial industrial media will require a ceramic
fiber filter media with weights only one-tenth that of the commercial media.
Another interesting feature of the analysis is that efficiency decreases for
increasing velocity. However, by adding fibers, the given efficiency can be
maintained as velocity is increased. The quantity of additional fiber
required is relatively small, especially if only 20-percent initial efficiency
is adequate for a 0.5-ym particle.
Most commercially-available ceramic fiber structures are produced for
insulation applications. Consequently, these materials are generally charac-
terized by an open fibrous structure. That is, they have low solidity, with
perhaps only 2 percent of the volume occupied by fibers (a = 0.02), A
solidity of a = 0.10 is more typical of a structure designed for filtration.
Figure 3 shows the effect on fiber bed thickness for changes in solidity and
air-to-cloth ratio. For solidity typical of insulation materials (a = 0.02),
a fiber bed about 1 cm thick should achieve high initial collection efficiency
of submicrometer particulate, while a more compressed media with a = 0.10
200
-------
would pchieve this efficiency with a bed thickness of only 2 mm. If effi-
ciency typical of industrial filters is adequate, very thin layers of the
3.0-Um diameter fibers will suffice. Note also that filter media thickness
is not a strong function of air-to-cloth ratio, indicating that high filtra-
tion velocity should be possible. Of course, higher filtration velocity will
result in increased pressure drop, but this may be acceptable in a PFBC
application.
Ceramic Media Performance at Ambient Conditions
To test the performance predicted analytically for ceramic fiber media,
several ceramic media samples were tested for D.O.P. penetration at two air
flow velocities. The resulting data are plotted in Figure 4. This data shows
that the paper media had the highest efficiency, felts in general had inter-
mediate efficiency, and the woven materials had the lowest efficiency. More
importantly, it shows that many of the samples collect D.O.P. smoke at higher
efficiency than standard industrial-grade filters which usually perform at
less than 20-percent efficiency in a D.O.P. test. Also, several of the woven
samples perform as well as industrial filters (these specific media are identi-
fied in the appendix). In order to compare these results with the calculated
performance shown in Figure 2, the D.O.P. data for several paper and felt
samples are plotted in Figure 5 using the same scale as Figure 2. Woven
samples were not included in the curve because the analysis assumes a random
orientation or bed of fiber. Figure 5 clearly shows that many of the fine
fiber ceramic media collect fine particles at high efficiency with low basis
weight. Two of the paper media performed nearly as predicted from the high-
temperature, high-pressure analysis. The vertical line through the data
points represents the extremes measured for the two velocities tested. The
performance line is drawn through the average of these two data points. All
of the media, except Fiberchrome type 800, are claimed to have 3.0-ym average
fiber diameter. Fiberchrome type 800 is claimed to have a 3.5-ym average fiber
diameter. The performance for the blankat samples was less than predicted.
This may be explained by variations in actual fiber diameter and the poten-
tially nonuniform fiber distribution in the relatively open mat and blanket
materials. However, all of these materials have higher collection efficiency
for 0.3-ym D.O.P particles than an equal weight per unit area of most standard
filter media in industrial applications.
201
-------
Summary Barrier Filtration Theory
We have examined those particle collection mechanisms that apply to
barrier filtration. Our particle collection efficiency predictions indicate
that barrier filtration at high temperature and pressure should be possible.
Furthermore, if we take advantage of the small diameter fibers currently
available in ceramic materials it should be possible to produce acceptable
filtration efficiency using thin, lightweight fiber structures. There is no
theoretical reason why these filters cannot operate at high air-to-cloth
ratio for particle collection. Room temperature tests of ceramic fiber media
using 0.3-Um D.O.P smoke tend to confirm these theories. Thus, we can con-
clude that ceramic fiber filter media should be able to remove particles from
high-temperature and pressure gas streams.
The major unanswered question is, can these media be cleaned adequately
to maintain an acceptable pressure drop? Fundamental to that question is,
can these media, or specially-designed media, survive the stresses they will
be subjected to in operation and cleaning cycles in a high-temperature and
pressure application? Aerotherm is attempting to answer these questions.
As a first step in this effort we have designed and built the high-temperature
and pressure filter media test rig described below.
Ceramic Media Test Facility
The Hot filtration Testing Facility (shown in Figure 6) is capable of
subjecting 46-cm (18-inch) by 9.5-cm (3-3/4-inch) diameter test filters to an
815°C, 10-atmosphere environment with air-to-cloth ratios from 0.5 to
5 cm/sec (1 to 10 ft/min). During high-dust loadings, bags tested in the
facility can be either reverse-flow or pulse-jet cleaned. There are three
test chambers so that three bags can be tested simultaneously, though only
one chamber is presently in use.
The filter media test chamber (Figure 7) is a 244-cm (8 foot) by
30.4-cm (1 foot) diameter length of pipe with inlet and outlet tubes welded
into the blind flanges on the top and bottom. The inside of the chamber
is lined with 5 cm (2 inches) of castable refractory wnich surrounds a 10-
kilowatt, 230-volt electrical resistance heater used to compensate for
refractory and external piping heat losses. Clean, high-temperature, high-
202
-------
pressure air enters the chamber through the bottom, where it is passed
through a dust pot.
The dust pot, an inverted conical container filled with flyash, funnels
the dust into the clean air entering the chamber, and then recaptures the
dust after it is filtered. Except for losses through the bag, the dust pot
is self-recharging and ideal for tests not requiring precise and controllable
dust loadings. The diameter of the dust pot is about 6.3 cm (2-1/2 inches),
opening up to the full 20.3 cm (8 inches) of the inner wall.
The 46-cm (18-inch) long test bag is above the dust pot. The bag-shaped
filter is mounted around a 9.2-cm (3-5/8-inch) diameter support cage screen
which prevents the bag from collapsing during outside-in filtering. The test
bags are open at both ends, requiring clamping at the top around the venturi
and at the bottom around a sealing plate attached to the cage. The filtered
air passes through the venturi, into the pulse-jet plenum and exits the cham-
ber through the outlet tube.
Once the filtered air leaves the test chamber, it is cooled with a
forced air, finned-tube heat exchanger. It is then directed through a
commercial, low-temperature, high-efficiency, tube-type filter (Figure 8).
This filter can be inspected periodically to determine the test bag's filter-
ing effectiveness. Downstream of the low-temperature filter, the high-
pressure air is bled through a sonic-flow metering orifice, through a noise
suppressor, and vented to the atmosphere.
Gas temperatures are measured using thermocouples located at selected
points within the facility. The air temperature measurements are taken at
the test chamber inlet, above and below the test bag, at the low temperature
filter, and at the orifice meter. These temperatures are displayed in the
control room and stored on a 24-channel strip chart recorder. Pressure
measurements using diaphragm pressure transducers are taken above and below
the bag, at the orifice and in the pulse-jet reservoir. This information is
also displayed and recorded in the control room.
There are two methods available for cleaning the test bag during opera-
tion: pulse-jet and reverse-flow. The pulse-jet method cleans the bags
using quick blasts of high-pressure air, introduced at the clean side of the
203
-------
bag. Pulse pressure can be regulated from 1033 kPa (150 psi) to 1723 kPa
(250 psi). Fast-acting solenoid valves provide 100 msec pulses at a
nominal 6 to 7 pulses per minute. Pulse-jet cleaning does not interrupt for-
ward flow filtering.
Reverse-flow cleaning is a milder cleaning technique, which requires
interrupting forward flow. The first step of the reverse-flow cleaning
sequence is to shut off the main flow and bleed the test chamber down to a
lower pressure. Next, the downstream outlet valve is shut and the main flow
redirected in through the outlet tubes at the top of the chamber. The inside-
out flow across the filter blows the dust off the bag. This process stops
when the chamber is once again pressurized. Reverse-flow effectiveness can be
varied by adjusting either the bleed-down pressure or the rate of repressuriza-
tion. To prevent dust from entering the inlet lines during reverse flow, a
small regulated amount of the main flow is introduced into the chamber
through the inlet tubes while repressurizing.
Both cleaning operations can be controlled manually or automatically at
the control room. A digital time sequencer, in combination with a solid-
state logic sequencer, can be programmed to provide automatic cleaning. The
following parameters are adjustable: pulse frequency and duration, or
reverse-flow frequency, bleed-down duration, and repressurization rate. The
number of cleaning cycles in a given time can be either predetermined, or
dependent on the pressure drop buildup across the bag. Automatic shutdown in
the event of bag failure or overtemperatures allows for optional, unattended
operation of the facility.
Before entering the test chambers, the high-pressure air is heated to
815°C (1500°F). This is accomplished by passing the air through 0.95 cm
(3/8-inch) holes in a 152-cm (5-ft) long by 15-cm (6-inch) diameter stainless
steel block with a wall temperature in excess of 870 C (1600 F). The block
is heated using a 50-kW induction coil and power supply operating at 30 kHz.
Additional capabilities are presently being added to the Hot Filtration
Test Facility: a revolving disk, high-pressure dust feeder and a steam injec-
tion system. The dust feeder introduces flyash to the main air upstream of
the block heater at a controlled rate. It is enclosed in a pressurized vessel
204
-------
with a dust hopper filling a small groove in a rotating disk. The pressur-
ized air enters the vessel below the disk and exits, carrying the dust out
through a small tube directly above the dust-filled groove. The feedrate
is adjustable by varying either the disk rotation speed or the groove cross
sectional area.
A 1033-kPa (150-psi) steam generator injects steam into the air stream,
subjecting the test filter media to 30-percent steam in addition to the high
temperature and pressure.
High-Temperature and Pressure Test Data
This paper reports on work being done under EPA Contract 68-02-2169,
scheduled for completion in August 1978. Consequently, at the time of this
writing only limited data is available from our high-temperature/pressure
test facility. The data that are available for reporting here have been
obtained during shakedown tests of the test rig, and should be viewed as
preliminary data. That is, no strong conclusions are warranted by the high-
temperature/pressure data available to date.
Mechanical Screening Test Data
As stated previously, the available ceramic fiber structures can be
generally classed into three groups: woven, paper, and felt. During the
preliminary testing, all three types of media were subjected to survey tests.
The objective of the mechanical screening tests was to subject the media to
typical filter-cleaning loads in the high-temperature and pressure environ-
ment and determine which media can survive these loads. All the tests
reported here have been operated at an air-to-cloth ratio of 2.54 cm/sec
(5 ft/min).
Woven Media Tests
Three woven media have been tested. Results of these tests are summar-
ized in Table I. These tests were performed early in the program before we
had fully proved the 815 C capability of the heater. Since these tests are
fairly qualitative, the following photographs and discussion are offered to
augment the data in Table I.
The AB 312 media test was stopped because of excessive dust penetration
through the relatively large pores around the yarn filaments of the cloth.
This media clearly showed a clean separation of the dust cake from the media
205
-------
surface (Figure 9 illustrates this effect, which has been seen on other tests
as well). This apparent cleanability will be studied in greater detail to
determine its reliability, and whether it can be used to enhance cleaning.
An Irish refrasil bag was tested with pulse cleaning after surviving
195 reverse-flow cycles. This media developed holes during the pulse testing.
Examination after 3,100 pulse cycles revealed the condition shown in Figures
10 and 11. In general, the holes were located near the center of the openings
formed by the support cage. They appear to be a result of bending and shock
at the maximum points of media acceleration during cleaning. This effect
could possibly be controlled by making the support cage openings smaller.
Note that Table I shows an 1100-m/sec pulse duration. This occurred because
of an error in setting the pulse duration and was not discovered until after
the test. This could also have caused the media failure. In addition, it
should be noted that pulse cleaning is probably not required for woven media
since the less rigorous reverse-flow cleaning can remove the dust cake.
Figure 12 shows a seam failure which occurred on the first pulse and
illustrates the force of the pulse. This particular test element had been
installed several times, and therefore had been handled more than normal.
Failure can be caused by excess handling and sharply bending the fiber.
Paper Media Tests
One test of paper media has been performed. Results of this test are
summarized in Table II. No failures were evident after a short reverse-flow
test; however, the media did not survive pulse testing. A bias-cut wire
screen supported this media on both surfaces. The bias-cut allowed the tubu-
lar shape of the element to stretch, contributing to its failure. A straight-
cut screen would provide better support for paper media. Figures 13 and 14
show the bias-cut screen, and several tears in the media.
Felt Media Tests
Test results for felt media are summarized in Table III. In the first
test no damage was evident as a result of reverse-flow testing. After about
5,000 cycles of pulse testing, the pressure-drop monitor indicated a problem.
Visual examination revealed that the lower band clamp had slid upward, allow-
ing a leak under the filter element. The lower end of the media was reclamped,
206
-------
and the tests continued. The lateral seam for this media was accomplished
by rolling a flat sheet of media into a cylinder and overlapping the ends.
This tube shape was supported inside and out by a wire screen. After a total
of 9,405 pulse cycles, visual examination revealed that the seam had sepa-
rated by sliding apart. This separation probably started when the band clamp
moved earlier in the test. The flexible bias-cut support screen also contrib-
uted to this problem. Figures 15 and 16 show the element and a close-up of
the seam separation. The whiteness shown in Figure 16 indicates a good sepa-
ration of the dust cake from the media surface. Cleaning in this region was
accomplished by tapping the screen several times with a pencil!
For the second test, a screen with a fine diameter wire with more strands
per unit length was used to support the media. While this screen was also
cut on the bias, it provided a stiffer support for the media. Extra care was
taken to ensure that the clamps were tight. This media operated for 7,450
pulse cycles with no apparent damage. The test was stopped because of a
failure of the test rig heater element. Figure 17 is a photo of the element
after the test.
Present Program Data Objectives
It should be emphasized that the data presented here is preliminary data.
The test program is continuing and will provide information outlined below
over the next year.
• Mechanical screening tests will continue to develop information on
media durability.
• Filter performance at high temperature and pressure will be measured.
Efficiency and pressure drop will be measured as a function of air-
to-cloth ratio, time, dust loading and cleaning techniques, and
energy. These tests will be performed for various media and media
support techniques. The objective of these tests will be assess both
filtration performance and durability.
• This data will be evaluated to provide a basis for preliminary cost
estimates for a high-temperature-pressure ceramic fabric barrier
filter system.
207
-------
Conclusions
While much remains to be done, the following tentative conclusions
summarize the current state of development for ceramic fiber barrier filtra-
tion:
o Barrier filtration using ceramic fiber filter media offers a promis-
ing approach to high-temperature, high-pressure particle control.
Based on room ambient tests, some of the available ceramic materials
appear to have good filtration properties.
o Innovative cleaning and media support techniques can be designed
which are compatible with the special properties of ceramic media.
Demonstrating these techniques is the major task to be accomplished.
o Theory indicates that high air-to-cloth ratio operation may be poss-
ible for a HTHP filter system. To accomplish operation at high air-
to-cloth ratio will require special cleaning techniques. This is a
desirable objective because smaller less expensive control equipment
may result.
208
-------
REFERENCES
1. Calvert, Seymour, Parker, Richard, "Effects of Temperature and Pressure
on Particle Collection Mechanisms: Theoretical Review," EPA-600/7-
77-002, January 1977.
LIST OF SYMBOLS
d = Particle diameter
P
df = Fiber diameter
C1 = Cunningham correction factor
p = Particle density
U = GAS velocity at media face (flow/area)
o
VI = Gas viscosity
6
k = Boltzmann constant
T = Absolute temperature
209
-------
APPENDIX A
Following is a list of media samples tested for D.O.P. penetration. The
list contains descriptive information taken from advertising literature for
each material as well. In examining these data, the reader should remember
that in general these materials were not produced for filtration applications.
Based on this information none of the materials can be recommended or
rejected. Additional information will be required before the judgment can be
made. The available information does indicate that at least some of the
materials will probably offer good filtration performance.
210
-------
DOP Test Results, High Temperature Filter Media
% Efficient
@ 5.3 cm/sec
Face Velocity
% Efficient
(§ 16.6 cm/sec
Face Velocity
Sample Identification
1. Fiberfrax Cloth with wire
L144TT
2. Fiberfrax Paper
970J 09-306-007
3. Zircar Zirconia Felt
Type ZYF-100
4. ICI Saffil Alumina AL4 Paper
5. ICI Saffil Alumina HT Mat LD
6. Kaowool Jfc thick 6 Ib Density
7. Carburundum
Fiberfax Dura Blanket
8. Johns Mansville (81b/ft3)
Fiberchrome \ in. thk type 800
9. J. P. Stevens and Company
Astro Quartz Style 581/38-4
Finish 9073 Lot 30701
10. Hitco Type C100-96
Roll No. 96-6293
11. Hitco Type C100-48
oil No. 48-5597
12. J. P. Stevens Style 570/38
Finish 9073 Lot 30601
13. 3M Company AB 312
Basket Weave Style 22B
14. 3M Company AB 312
Twill Weave Style 22T
15. UC 100 Hitco Type - 48
16. Zircar Zirconia Cloth
Type ZYW-30A
17. FMI Astroquartz (Custom Weave)
Note: Sample size 100 square cm. The lower face velocity tests were run
at 32 SLPM. This conforms with the military specification for high
efficiency media and represents a face velocity of 10.5 ft/ndn.
The high face velocity tests were run at 100 SLPM (standard liters/
min) .
2%
84%
43%
69%
30%
70%
90%
38%
0%
16%
11%
10%
0%
0%
14%
28%
4%
48%
82%
62%
75%
56%
76%
90%
70%
0%
27%
10%
10%
0%
1%
4%
35%
6%
211
-------
Media Samples — Description
Fiberfrax Cloth
Fiberfrax Paper
Zircar Zirconia
Saffil Alumina
(1C1)
Kaowool
(Babcock & Wilcox)
No organic binder (removed), 2300 F service temperature
L-126TT grade; 1/8" thick, twill weave, 21 pcf,
L-144TT grade;
32 oz/sq. yd.
1/10" thick, twill weave, 28 pcf,
34 oz/sq. yd., has nichrome wire
insert, strong to 2000°F
52% Al 0 , 48% SiO , 2300°F service temperature, up to
1" long, 2-3 microns (mean) fibers, 2.53 gm/cc density
970-AH grade;
970-J grade;
Felt ZYF-100:
1/32" thick, no binder, 12 pcf,
4.5 oz/sq. yd.
1/8" thick, has up to 5% binder,
10 pcf, 15 oz/sq. yd.
19 oz/sq. yd., 0.1" thick, 15 pcf,
96% voids, 600 scfm/ft2 at 0.5 psi,
5.8 gm/cc density, 4.5 ym fibers,
no binders. Breaking strength
1.6 pounds/in. width
Fiber density 3.4 gm/cm3, 3000 F service temperature,
3 micron fibers, 1-2" long, 1.5 cm2/gm
surface area, 95% A120 - 5% SiO
Mat 4 pcf, % in. thick, 2.4 oz/sq. yd.,
no binder
Paper 12 pcf, 0.04 in. thick, 6 oz/sq. yd.,
has binder
47% alumina - 53% silica, good to 2600°F, 2.8 micron
fibers, 2 in. long and up to 4 in. long. Has no binder.
2.6 g/cc density, 6 pcf, %, h, and 1" thick, 18, 36,
72 oz/sq. yd.
Fiberfrax (Carborundum)
Durablanket
48% A1203, 52% SiO , 2300 F service
temperature, long fibers, 2-3 micron
(mean) fiber diameter, density 2.62 gm/cc.
No binder, 6 pcf, \ in., 18 oz/sq. yd.
Fiberchrome
(J. Manville)
Felt
41%
good
8 pcf,
alumina, 55% silica, 4% Cr 0 .
( to 2700°F. 3.5 micron filers,
in. thick, 48 oz/sq. yd.
212
-------
Media Samples — Description (Continued)
Astroquarts
J. P. Stevens)
Cloth (with binder)
Style 570:
Style 581;
Custom Weave;
FMI Astroquartz
5 Harness Satin 300-2/8, 27 mils thick,
19.5 oz/sq. yd., 38 x 24 thread count/
inch, 325 (warp) and 300 (fill) pounds/
inch width breaking strength. Pure
Silica (99.9%), 2000°F service temperature.
Filament diameter ~ 6 microns)
8 Harness Satin 300-2/2, 11 mils thick,
8.4 oz/sq. yd., 57 x 54 thread count/
inch, 175 (warp) and 170 (fill) pounds/
inch width breaking strength. Pure
Silica (99.9%). Good to 2000°F continuous
service. ~ 6 micron filaments
Crow foot Satin 300-2/2 (warp), 4/2 (fill)
54 (warp) x 36 (fill) threadcount per
inch, ~ 12 mils thick, ~ 11 oz/sq. yd.
Pure Silica (99.9%), good to 2000°F
service.
Refrasil
(Hiteco)
Refrasil
(Fibers 8 to 10 microns diameter)
Irish (Chromized)
C 1554-48: 8 Harness Satin, 26 mils thick, 19.2 oz/
sq. yd., 53 x 40 thread count per inch,
96 (warp) and 62 (fill) pounds/inch
breaking strength. Consists of almost
pure Silica with 1 to 3% Chrome Oxide.
Continuous use to 2300°F.
Heat Cleaned (Preshrunk)
C 100-48: 8 Harness Satin, 26 mils thick, 18.6 oz/
sq. yd., 52 x 39 thread count per inch,
86 (warp) and 61 (fill) pounds/inch
breaking strength. Consists of almost
pure Silica (99%+). Good to 2300°F.
C 100-96; 12 Harness Satin, 50 mils thick, 37.1 oz/
sq. yd., 50 x 39 thread count per inch,
130 (warp) and 65 (fill) pounds per inch
breaking strength. Essentially pure
(99%+) Silica. Good to 2300°F
213
-------
Media Samples — Description (Concluded)
Refrasil (continued)
UC 100-48:
AB-312
(3M Co.)
Zircar Cloth
(Zircar Corp)
8 Harness Satin, 26 mils thick, 18 oz/
sq. yd., 46 x 36 thread count per inch,
80 (warp) and 60 (fill) pounds per inch
breaking strength, good to 2300 F.
99%+ Silica
Heat Cleaned. Alumina-Boria-Silica, 390 filament strand,
11 micron diameter filaments, 250,000 psi tensile strength,
22 x 106 psi modulus, 2300 F service temperature,
2.5 gm/cc density
Style 22B:
Style 22T:
ZYW 30A:
50-1/0, 32 x 38 thread count/inch,
9.3 oz/sq. yd, 12 mils thick basket weave.
50-1/0, 32 x 25 thread count/inch, 7.6 oz/
sq. yd., 10 mils thick, twill weave
20 oz/sq. yd., 0.030 in. thick, 5 Harness
Satin, 63 pcf, 83% porosity, 5.7 gm/cc
density, 5 micron continuous filaments.
Breaking strength 4 pounds/in, width
BET surface area 1 m2/gm, 92% Zr 0 -
8% Y 0 . Good to 3300°F, 2-ply x 480
filament yarn, 45 x 34 threads/in.
214
-------
in
H
hJ
£3
Cfl
W
Pi
H
W
H
c/j
rJ
M
Pi
w
H
^
J2
>
0
M
W
rJ
p3
5i
H
Ol
rH
•H
cO
Pn
0)
a
•H
H
CO
01
i-H
CJ
K*1
O
60
c
•H
c
cO
01
rH
u
0)
j_i
^
4J
0)
J_j
0)
ft
e
01
H
CO
0)
1
^H
o
O O J3
!3 £3
•\
en
0)
M M r.
in c*") r~-
rH
00
ro m o
in o> o
rH rH rH
ro
CO
» » a
0 0
rH rH O
pr | pr ^ ^^
> rH
0) 01 3
Pi Pi Pn
0 U U
o o o
OO 00 00
CO C"O CO
in in in
C
O) -H
K* 4-)
CO cO
Oi w
*~y -1
jj> r~ 1
•H 0)
rH en g
rH CO 0
>*, -H }-J M
CO H 01 O
ft Pi ---
6 CM CO
O rH & O
cj m co -H
•H rH
^ m >-i -H
rO rH
Ol 3
Pi P-I
u o
o o
oo m
OO rH
m oo
fi
•H
4-1
CO
c/>
rH
•H 0)
en 8
cd o
r4 M
Ol O
Pi --»
CO
rC O
en -H
•H rH
M -H
M C/)
f — ^
•H t/5
en S
ft
O
O O
^O i — I
rH H
x^/
215
-------
CO
H
hJ
J"-\
H-*
W
M
Pi
H
CO
H
GO
3
H
Pi
W
H
•^
(
0)
Pi
u
o
m
rH
OO
CS
C
•rl
e
3
rH
H
(-1
^2
ON
VO
CT\
^sj"
cd /^^\
A - '
P-t *rt
r^i CO C/D
fX S
O
0 O O
H vD O
rH tH rH
x^-
0)
CO
rH
3
PH
216
-------
CO
H
a
CO
3
H
CO
W
H
co
^
^
H
PJ
W
H
S
H
W
M
M
M
W
PQ
^
H
01
rH
•H
cd
fa
01
•H
H
W
01
rH
a
^
u
W)
c
•H
C
OJ
i — 1
O
0)
3
cd
^
Q^
a
S
ai
H
a
01
O 03 O
53 ^
03
OJ
^ ^
m ox oo
<1- CO
in in o
r~ o m
-cj- rH rH
01 3 3
P^ fX, PL,
O O O
O O O
m in m
i — ) ,-~J rH
00 00 OO
cd cd
fl C
•H -H
a a
3 3
rH rH
rH rH
•H -H
M MH M i4H
U cd CJ cd
H CO M CO
s-**
•H
03 CO
P. B
O 0
\£) O
rH rH
^~s
217
-------
to
to
O)
s_
Q. •
"O
-O O)
C -Q
O)
Q.M-
E O
QJ
+-> >>
U
-C C
CT- O)
•r~ "r—
j= o
0)
O O
O) -r-
M- 4->
M- O
O) O)
O) i—
.c o
I— O
o
o
o
o>
o
oo
o
(O
o
in
o
CO
o
C\J
218
-------
>O/8l - v
ui
g
DC
Si
w
m
IT
Si uJ
u. m
O CO
m cy
8
IS
i-
§ g 8
% — XON3IOIJJ3 NOI103T1OO
-O
O)
S-
(U
"
3
s
3
D
_
—
_
§(0
c
'i
CO fO
M
2 f=
§ °1 p"
° 1 0
£ «
So a-
* 5 S
> c
1 55 ^
10 S I
fi-
cu
a.
-a
o a;
i ^
(0
3
8 o
la
0
CVJ
O)
01
•^
LU
219
-------
1.4
1.3
1.2
1.1
1.0
O 0.9
co 0.8
LLJ
z
*
O 0.7
I
S °-6
CD
£ 0.5
CD
0.4
0.3
0.2
0.1
0
0.5 Mm PARTICLE
815° C
10 ATM
LL
90%EFF « = 0.10
.
20% EFF a = 0.02
.
20% EFF a = 0.10
-©-
10
AIR-TO-CLOTH RATIO — CM/SEC
Figure 3. Fiber bed thickness.
12
14
220
-------
100
90
0.3 urn D.O.P. SMOKE
AMBIENT TEMP, PRESS
o—
20 40 60 80 100 120 140 160 180
AIRFLOW VELOCITY MM/SEC
(P) = PAPER
(W) = WOVEN
Figure 4. D.O.P. efficiency test data.
221
-------
Q.
6
d
E
a
O)
I
h-
(3
i
-------
Figure 6. Hot filtration facility.
223
-------
PULSE IN
CHAMBER HEATER
AIR OUTLET
TEST MEDIA
FLYASH
AIR INLET
Figure 7. Test chamber cross section.
224
-------
68S9I-V
GO
to
a
u
fO
d)
-C
U
10
o
to
cu
00
O)
3
CT.
225
-------
qieO-H
•tf .v&o^. -
-. ^ * _* - ..
_ • u_ •*•• I • *» —»
«?*»_» >" , _J » I «•»
Ol
>
ro
I
C\J
CO
CQ
CO
CO
o
co
to
-------
Figure 10. Irish Refrasil pulse test failure.
227
H-029b
-------
;*»^^*'**yi *-•-»?* ,£*•%»*
•^T:--**Si
Figure 11. Irish Refrasil pulse test failure.
228
-------
Figure 12. Refrasil media seam failure.
229
-------
Figure 13. Saffil alumina paper after pulse testing.
230
-------
r*l
.>• <*• - -
- -
/ ,
Figure 14. Saffil alumina paper after pulse testing.
231
-------
f
\
Figure 15. Saffil alumina felt bag post-test.
232
-------
/fv,
CM
C\J
o
Figure 16. Separated seam.
233
-------
Figure 17. Saffil alumina blanket after pulse tests.
234
-------
HIGH TEMPERATURE FINE PARTICLE CONTROL USING CERAMIC FILTERS
By:
D. C. Drehmel
Environmental Protection Agency
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
D. F. Ciliberti
Westinghouse Electric Corporation
Pittsburgh, PA 15235
235
-------
Introduction
During the next two decades, the use of coal for the generation of
electrical energy in the United States will triple. New coal-fired
power plants will be built whose total capacity will almost equal that
of plants currently installed. There is a real need for a lower cost,
higher efficiency, less polluting means of generating power from coal.
Gasification coupled with combined gas and steam turbine generation is one
promising technique. But if coal gasification with combined cycle
generation is to be completely successful, a system of cleaning and burning
hot fuel gases is needed to meet emission standards on particulates, and
to protect high temperature turbine blading.
Consequently, the objective of this work was to establish the technical
and economic feasibility of using a porous ceramic filter for fine particle
control in advanced power systems.
Theory
Experimental information is available in the literature on various
filtration devices which have similar operating characteristics to the
ceramic filter.
The most appropriate information is on membrane (e.g. "Millipore")
filters which have controlled pore sizes in the sub micron range. For
example, Fitzgerald(1) and Detweiler found complete retention of particles
greater than 0.1 um diameter on Millipore filters with 0.85 urn pores.
Minimum collection was found with particles approximately 0.02 um diameter
and this minimum varied with the superficial gas velocity. (78% at 10
cm/sec., 20% at 40 cm/sec.).
Porous metal filters have been used extensively for particle collection
from high temperature gases. Complete retention down to 1 u is reported,
and good efficiency for sub micron particles is implied.(2)
Fabric filters rely on an accumulated filter cake to provide high
efficiency dust collection. This will also occur with the ceramic membrane
filters. Consequently collection efficienciea with the ceramic membrane
should be at least as high as for fabric filters operating under similar
conditions. 95 to 99% collection of sub micron particles has been
reported in fabric filter installation.
Particles which penetrate the filter cake may also penetrate into the
ceramic pore structure. If a pore size of 1 ym is assumed, all particles
greater than 1 y will be collected, however, there will be a significant
possibility that smaller particles will be collected within the pores, or
alternatively, will penetrate the filter.
A simplified method of calculating Brownian deposition suggested by
Kaufmann^ ' has been used to examine this situation. This assumes that
in deposition from laminar flow, all particles moving along stream lines
located a distance x from a collector surface would be deposited on the
collector, if x was equal to the mean Brownian displacement in the x
direction, i.e.:
x =
* ir
236
-------
(D is Brownian diffusivity and t is time of flow past the collector).
This is modified for collection in a cylindrical pore by allowing for
the possibility of collection in both the x and y directions. As a
consequence
becomes
x =
Complete collection is achieved when:
d - dp
2
where d is the pore diameter and
dp is the particle diameter.
For a ceramic filter with 1 pm pores, and a thickness of 250pm, complete
collection is estimated at gas velocities up to approximately 30 cm/sec.
(Table I).
It should be stressed, however, that not all sub micron particles will
enter the pores, and it is not reasonable to assume that all particles
which meet the pore wall will be captured by it. So while the estimate
implies high efficiency collection it cannot be used in a quantitative
sense.
Table I. Collection by Brownian diffusion in pores.
dp
Mm
0.7
0.5
0.3
0.1
D
cm /sec
1.9 x 10~7
2.8 x 10"7
5.0 x 10"7
2.7 x 10"6
Gas Velocity in Pore
6 cm/sec
/SDt^ d - dp
/ TT 2
pm pm
0.44 0.15
0.53 0.25
0.66 0.35
1.48 0.40
30 cm/sgc
/8 Dt d - dp
/ tr 2
pm urn
0.20 0.15
0.24 0.25
0.30 0.35
0.66 0.40
237
-------
If particles are deposited in the pores, as suggested above, they must
subsequently be removed in the clean up cycle if a useful filter life is
to be achieved. A reverse flow of gas equivalent to the initial forward
flow is unlikely to dislodge such particles and a high velocity surge
may be required.
Pall(2J suggests that a low flow for clean up is preferred, such that
the accumulated cake is removed from the filter surface, but a protective
population of particles is left clustered at each pore opening. These
particles prevent penetration of other particles into the pore.
Experimental Methods
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
tradename of ThermaComb. This material is composed of alternate layers
of corrugations separated by thin filtering barriers. This type of
configuration affords a large amount of filter surface in a very small
volume.
Bench side experiments were conducted in the high temperature ceramic
test facility shown in Figure 1. Part of this facility included a test
jig for the ThermaComb material which is shown in Figure 2. Provisions
were made to blow back from the clean side and also down the channels on
the dirty side so that various cleaning schemes could be investigated.
A schematic of the testing loop is shown along with the sequence to be
used for the cleaning cycle in Figure 3. A sequencer was designed to
automatically step the system through the cleaning cycle so that the
operation of the system after start up was automatic. A 17.8 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.
The size distribution of the test dust (limestone) was determined
using cascade impactors. The typical mass median diameter was 1.4 vim
and the geometric deivation was 3.0. Some difficulty in maintaining
constant feed rate was experienced but dust loadings were maintained at
levels from 2 gm/m to 7 gra/m .
Results
Typical results for the 3M ThermaComb filtering the limestone test
dust are shown in Figures 4 and 5. Figure 4 shows the effect of varying
the initial pressure of a 0.6 sec pulse. The 69 kPa initial pressure
dropped to a steady pressure of 34.5 kPa for the remainder of the .6 sec
pulse and the 34.5 kPa pulse dropped somewhat below 10 kPa. Figure 5
shows the result of a similar set of runs except that the pulse time was
increased to 5 sec. from 0.6 sec. It can be seen from these data 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
238
-------
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^.
Discussion of Results
The behavior of the ceramic filter is remarkably similar to that of
fabric filters, with a short transient of rapid increase in pressure drop
followed by a steady, linear increase in AP with time. The data is
consistent with filteration theory as indicated by the following analysis.
The total pressure drop is considered to consist of a contribution due to
the media APm, and due to the accumulated cake, APC.
AP = AP + AP
tot m c
For given gas properties, the pressure drop across a filter media, APm,
is proportional to the face velocity, U and filter thickness, Lm. The
pressure drop across the cake is similarly proportional to the face
velocity and cake thickness Lc. If a constant dust concentration, C, is
assumed, the cake thickness is proportional to the product of the measured
variables, UCt, where t is time. The total pressure drop for a given
gas, temperature and filter is given by:
AP = K'U + KU2tc,
where K' and K are constants for a given filter and incompressible cake.
The linear dependence of APtot on time is substantiated by the steady
part of the APtot vs. t curves. Confirmation of the assumed form of
dependence of the variables U and C can be achieved by plotting the slope
of the APtot vs. t curves, normalized by division by U vs. C. This plot
should yield a straight line with a slope equal to K which is a property
of the filter cake and gas viscosity only. Figure 6 presents this data
and indicates that the assumed form of APC is confirmed by all the data.
Two further observations of major importance can be made concerning
the initial steady pressure drop.
(1) The magnitude of the initial steady pressure drop is not prohibi-
tively high.
(2) The initial steady pressure drop remained constant with the adopted
cleaning technique.
239
-------
Conclusions
Tests using the 3M ThermaComb as a filtering media were conducted at
temperatures from ambient to 970°K. Filtering efficiency was found to
be close to 100% even though the test dust had a mass median diameter
of 1.4 ym and a significant 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.
240
-------
References
1. J. Fitzgerald and C. Detweller, Arch. Ind. Health 15_, 3, (1957).
2. D. B. Pall, Ind. Eng. Chem., 45, 1197 (1953).
3. A. Kaufmann, Z. Ver Deutsch Ing. 80. 593, (1936).
241
-------
Fig. 1 - High temperature ceramic test facility
KM-687n1
242
-------
Dwg. 1684B44
Duct for Air to
Sweep Dirty Side of Channel
Fig. 2- Hot thermacomb test jig
I
Si
s
243
-------
CD
O
c
CO
IJ T*l
CD •— • CD
0 ^ £ °
J§ c* oo
§ *c 1 1
« g O 0
w 0
>*
oo
r—
^^
NO
^
KJ-
>m
^
i— 1
*-
0
o
o
0
o
o
o
o
0
0
o
o
0
o
o
o
0
o
o
o
0
o
o
0
0
o
0
o
0
o
o
o
o
0
o
o
o
o
o
0
o
o
o
0
0
JP +T* ^ ^ ^
E
o>
CD
o>
Q.
CD
O
c
s*
CO
(O
_o>
o
x>
(O
s
0>
cn
1Z
244
D. Ciliberti
l.t. -k.b. 8/12/76
-------
I
o
e
8
03
en
tr
245
-------
dV
-
*-*
246
-------
03
CM
C
CO
C
o
OJ
o
u.
O>
C
o
to
c
I
OJ
0
I
vO
cn
o
o*
CD
06
X
o
o'
. Ciliberti
t. -r.m. 7-13-76
247
Curve 686021-A
-------
PROBLEMS OF GAS PURIFICATION OCCURRING IN THE USE OF
NEW TECHNOLOGIES FOR POWER GENERATION
By:
E. Weber
University Essen
West Germany
249
-------
Problens of Gar Furlfication Occurring in the Use _of_
New Technoloqj.es for Power Generation
In the Federal Republic of Germany there are some conceptions
for new coal conversion technologies based on the assumption
that the development of conventional steam power plants has in
principle attained perfection. The optimum efficiences which
are usually between 36 and 37 % will decrease somewhat by the
employment of waste gas desulfurination plants due to their
requirement for additional energy.
Among the new technologies expected to increase considerably
the efficiency of power plants are the combined gas steam
turbine processes preceded by a suitable coal conversion system.
For the latter there are coming into consideration
Fluid bed combustion
Fixed bed gasification
Partial dur.t gasification
Fluid bed gasification.
As an example figure 1 shows the appU cat ic :i of the combined
gas slt.(j.m turbine process behind a fluid b; i ed to cause ri'.'J.rii^ji.u losses of he?t ai:.'i pressure.
250
-------
Nearly the same prob]eras arise in the case of the so-called
fixed bed gasification ( Figure 2 ). The coal is gasificatcd
by the use of air. The formed lean gar, has a temperature of about
600 °C and pressure:; up to 2o bars. Before the combustion in the
vessel the gas is partially expanded and the resulting energy is
utilized by a gas turbine.
This and the reaction of the gases in the vessel are only
practicable when the lean gas has been already dedusted to a
far extent c:nd when the hyciiogen sulfic'ie and sulfur dioxide contents
of the g;Lc; have been reduced -- if possible at operating temperatures
and pressures- At present a pilot plant v:i bh a throughput of
60 tons co.il per hour, equivalent to a power output of 170 MW,
is tested at Liinen by the S'i'EAG using five LURGT fixed bed gasifiers
Among others, however, a sufficient solution is missing for the
problems of gas purification at operating temperatures and pressures
Similar problems with respect to the necessary gas cleaning system
for high i.-civ-craturces ar"^ pressures occur also in the case of
the. pari'ia] dust gasification and the fluid bed gasification.
These processes aro not discussed explicitly in this paper.
ft can Lo F,vnpj!!<.;ri%<::d that the gas purification re-presents an
important c~lprae.at of the coal conversion processes. The gr.s
purification has ap;-rL from the necessary dosulfuri nation to
fulfill the follow.i UM conditions:
251
-------
1) Precipitation of the so.1 id components down to dust
3
contents of less than 5 mg/m at S.T.P.
2) Dust separation at gas temperatures up to 1OOO °C and
pressures up to 40 bars.
Low dust contents in the purified gas are the precondition for
a successful gas turbine process whereas the dust separation
at the high gas temperatures and pressures is necessary to
avoid efficiency losses when applicating a steam turbine process.
It may be noticed that an energy sparing high temperature and
high pressure purification is of major interest not only for
power plants but even for other branches of industry e.g. in the
cat;s of remelting processes and refuse and sludge burning.
It .shall be dJscussed whether a dust separation under extreme
c:>jicHtions is possible with known dust separators or whether
ii5w separating systems have to bo developed. To illustrate the
situation figure 3 shows a survey of wel] known dust separation
systems and thoir efficiency. Tlif- gas pur s "Tication systems are
d i v .I d :-j d into f o u r groups:
Gravity and moT.c-.ntum separators
F ab r .i c f i. 1 1 e r s
E IGCTL ro t- 1: at i c precipi r ator &
Wet sc.ii
IV-Jiho'jt furtht.'i discuasioj; of th-:: oeparal.i'- ••> effects of each
si"-.;lo opp;ivatur.: .it cor. be a toted that the .^btain^b] ^ efficiency
3 r. ft? j rerent ff:r u^-.ch r.c;purator r.yr^tora. GT r.\Tity and iii
252
-------
separators defending on the mass of the solid particles usually
attain a poor or average dust separation. Fabric filters and
electrostatic precipitators however reach high degrees of
separation. In principle this aim can also be achieved by
the use of suitable wet scrubbers.
In the last row of figure 3 there are the maximum values of the
gar; temperatures and pressures that have been realized for each
dust separator system til] today. It can be seen that only gravity
and momentum separators can be applicated at high temperatures
whereas there is an upper limit for fabric filters and electro-
.static precipitators at about 350 C. The corresponding value
for wet scrubbers is al about 1OO )C. According to the laws of
physics the gas pressures can have any magnitude for all dust
separator systems, except the electrostatic precipitators. For
thi:' la L tor it seems possible to increase tho gas pressure
co.rrparaJi.le a] though that has not boon confidently proved.
Indu.'.lri G! plants have boon buj It; for prrr?;.u'-res up to 3 bars.
It cjn be suivrnor Lzed that the only wellknov.^ system applicable
withou!: complications at high temperatures and pressures are
tl;f gravity and rao~r.entum fcCuaraLorG. Due to their ruodo of action
CVCTI the best gravity and momentu))! separate-, s do not reach lov;
duni;- contents in the purified ga.s, Furthoi: .-.a their efficiency
is decreasing at hitjh ops temperatures acco: cling to the incj'easing
vi::.c:or,j..ty of the gas. j'/i.ts is i I'l ac;t;.;ateci .;.:< figure 4 for several
pressure drops. rilhe rcitio of the c-olJ ectab'i-• parti c] n diameter
at ?.0 C co thf.it of higher temperature io (.lotted as a fun];hion
of i.'ne qas tc^in'^rr'tuv" up to '1000 °C, Due to t'lose conditions
253
-------
it can not be exspectcd that gravity and momentum force
scparato'rs will fulfill the demands for a sufficient gas
purity at high temperatures. At extreme conditions these
sepr-rators can therefore only be applicatcd as pre-stage
systems.
Very low dusl contents in the purified gas can be achieved,
hovover, by the use of fabric filters. The fibrous materials
eipplicated on an industrial scale are till today suitable
only for gas temperatures up to about 35O C.
In principle the gas purification is also possible at high
temperatures using fibrous materials which resist gas temperatures
of lucre than 10GO °C ( Table 5 ). This f.ibrous materials can be
cLi vido.d into:
1 . minor a3 fibres
2. graphite fibres
?j, UK:t alii c £ ibr es
7bccording to the obtainable minimum fibre thickness all mentioned
fjbyous mate-rials should be applicable for ^^ high-giade gas
pin i fj option. lit is dif ficul t, hcvcver, to process fleeces, fabrics,
and feltr. of mineral fibres, for these matci ials arc frequently
brittle or fragile. Moreover graphite and mctalljc fibres are
extremely expensive caused by the monufactn> Ing processes. At
present thero are sevordl investigations in the FRG concerned
with new technologies for the munvifacturing and processing of
254
-------
fibres with the aim of using these materials for gas cleaning.
Besides this it is necessary to study the possibilities of
cleaning and the durability of the filter materials. 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 6. Two parallel filters
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 l;hjch can be fed additionally with oil in order to obtain
higher hemperaturcs. Each filter has an effective area of about
o
2. n:'. The filters allow to test all kinds of cleaning procedures
Precursory experiments espoc.i ally with respect to new cleaning
procedures seeu l.o offer successful results. Accoiding to these
.finding5; fabric filters should be applicable at low cost for
the high temperature gas purification. Detailed informations
c-boul the expire J.niori ts and their results shall be reported in
the next future.
In contrast, to O.;.bi:ic filters, pacl.ed bed hoparators of adequate
material can al.ct'idy be u~.ec1 at nearly all temperstm es and
pressures. At acceptable-- pror.surc drops pc.c'-od bed separators
cjchicvo low C!UL c concents in the purified <•. •'.:• only i/: special
Co1,1-or:, j n ad-j LI.; on it irv / be cliff ic;-it to i ' -un the par-kcd beus
iic-ni .",t i cl:inc/ p. rl J.cl --s , ••<_• ih^t Lhl1-; scp;^ c-;.or can not solve
255
' ' "J ;. ' t.''~ ' ""ac", i'*r':._ • i'T)Tir,cM i • , -i' i on
-------
Like fabric filters electrostatic precipitators belong to
the most effective gas cleaning devices. Their mode of function
is largely determined by physical and chemical factors.
On an industrial scale these precipitalors are realized
hitherto to gas temperatures of 350 °C and to gas pressures
of 3 barw. At present it is the aim of a research project
supportc-rl by the Ministery for Research and Technology to
investigfrle the feasibility and the operating conditions
of electrostatic precdpitators at high temperatures and pressures.
It shall be examined if they can be successfully applied with
respect to the dust concentration of the purified gas.
Sortie aspects of this problem hcive been studied already by a
number of. investigators ( 1, 2, 3, 4 ), but up to now the subject
has not been treated comprehensively.
Figure 7 shows a laboratory plant which operates as a tube
procipitciLor with variable diameters allowing temperatures
up to 1000 °C and pressures cf 3O bars. The air or gas with
the dispersed dust is hold in a closed circuit.
2-.pb.rt from construction problems, the application of electrostatic
precipitalors at high pressures and high tt!-".pci atures in only of
interer.t, if the physical effects of precipitation are not
disadvantageously affected. As a consequence the migration
velocity, defined by DEUTfiCTl, would become !~o small.
The joain factors are:
256
-------
1. The current-voltage characteristics, that is the correlation
between corona current and applied voltage,
2. the corona starting voltage,
3. the maxim?'1'. permissible voltage, that is the sparkovor
voltage,
4. the electrical resistivity of the dust.
The opera Lion of an. electrostatic precipitator requires a
considerable difference between the corona starting voltage
and the sporkover voltage. Experimental investigations have
to clarify, to what extent there quantities are influenced
by pressure and temperature. The results of previous investigations
on the corona starting voltaoe and the sparkover voltage are
shown in figure 7, The. voll^ocs are outlined as functions
of the gar; temperatures for tvo pressures. Accordingly the
ope rat a no- range of i-r< elect ro,i' Lc'ti c precipitr tor decreases \-'i th
incrcas.uKj teirporntur and cci'Scant prossme. This can be eypJained
above all by the increase of the ion inobility. For instance the
tiicjgraiv, r;.'iO\.'s that at a pressure of 1 bar end. temperatures of BOO C
the difference bstv.-cu the cpr,rkover vo.l r.:ige and the corona
starting ^-oltage does not cyu-i3 an tee the operation of an electro-
f. Liitic prc:c.i.pitator. Hut thJ ?., is possible at a pressure of <1 bars
an-i idc-T) I :i cal tempi?].r?.tore. In ligure fi one .'>( :r..'L consider utions , that an electro-
257
-------
sldtic preccp:tator at high temperatures can be realized only,
if the pressures are raised.
1'rojfl very simplified assumptions like axial electric field,
simple flow conditions, spherical particles, and the validity
of STOKE' s Jaw, a theoretical migration veloucity can be deduced.
It is at least qualitatively correlated with the effective migration
velocity. The shortened derivation is shown in figure 9. It appears
thc;t the theoretical migration velocity depends on the product
of the charging and collecting field intensities.
At constant electrical field intensity the theoretical migration
velocity is inversely proportional to the gas viscosity, which in
its turn depends mainly on the gas temperature. Proceeding from
a linear relationship between the rtean field intensity and the
pressure as wall as the temperature of the gas, individual values of
theoretical migration velocity can be calculated. Figure 1O shows
the variation of the theoretical migration velocity in air with
fce.ivperatui e and pressure. The ordinate is the ratio of the actual
t-icoreticctl migration velocity to that at 1OO °C and 1 bar.
It can be seen that the- theoretical migration velocity decreases
with rising temperatures and increases v;.tr> increasing pressure.
The unfavorable-; influence of high temper<;ii.res can be therefore
l)ulanced by rising the pressure:.
Another influence on the characteristic." of electrosLatic
precipitate;:::1, is due to the already ment.iuhnd resistivity of the
dust. This electrical quantity f dimensioncc ;; '.- era, is significant
for the adijcrcnee of the dust to the collecting electrodes. It is
258
-------
11 /• %
in general accepted that dust resistivities of more than 1O -is c' cm
or less than 1O i'n2 cm are disadvantageously according to the
diminished adherence ot the dust. Theoretical understanding and
present experience suggest that electrical resistivities may
become to low at high gas temperatures. This, however, has not
been confirmed for converter dust by the experimental investigations
of CCHUTZ and WINKEL. Therefore it is necessary to study more
detailed dust resitivities at extremely high pressures and
temperatures.
Based on today's knowledge? the effectiveness of electrostatic
precipitators at high gas pressures and temperatures can be
summarised as follows:
Probably the current voltage charaotoristien should allow at
least a sui1.ab.le precipitation if i.n addition to high temperatures
also high pressures are at hand. It must b? seen wet her this
view wiJl be confirmed and how the electr:' cp';!. resistivity of the
dust- will behove ai these extreme concli tior >,. Besides of the
ifiont.ii'KC.'d physical effect's the industricai realization of tho
electrostatic prccipiLM or at high pressxi LC-"-, and high temperatures
is opposed by severe constructive and tec!v icnl problems.
After al] the present investigation has to -;bow if it j.s possible
ii, f",''j>-^lope an electrostatic pro."i pitator V.- ving a su f-Cj cienir.
off e--l-i voner.-j a.s \vcll as reiiscnablc: cost.
In contract to tho throe already mentioned .^jpeiraticni systems,
wel' scrubbers a'Licnv the f;.i :n.uT tan four, precj , • tation of both solid
pa? '• j :; j C1;: ai;ci o,-.;^<:')us con^i 1 •! nr-Mi.-':. Thj.s c. binatior; of two
go:" i,AC<;ui:iq r-.\ ,--..c:jus 3 ii n :;::ingje o;ic enal-J ; a low cost: gas
259
-------
purification as compared to the use of a two stage system for
the separation of gaseous and solid constituents. It is a severe
disadvantage, however, of all known wet scrubbing systems using
water as a washing medium that the gas is cooled considerably
before or during the separation process. Wet scrubbers although
offering several advantages with respect to other separation systems
cannot be employee! if the gas shall be kept at a constant high
temperature level to avoid energy losses.
To overcome these difficulties it shouJd be possible to use a
suitable washing medium instead of water or aqueous solutions.
Evidently the choise of the washing ]iquid is most important.
Substances for the-: separation of gaseous and solid pollutants at
high gas temperatures have to fulfill the following conditions:
1„ The substances have to be in the liquid state at temperatures
of 6bout 50O C. They should be sprayablt-. and must have a
low surface tension.
2. The vapour preorure of the melt should be negligible at
.le-:;;;r for temperatures up to 800 °C.
3. There should be formed no toxic or harmful compounds.
4. The melt has to be either loss-free reproeessable 01
convertable .into products that can be led to dispopu.'l ..
260
-------
5. The cost for the used melt should be acceptable.
Due to these restricting preconditions orientating studies have
shown that a number of inorganic substances can be used for a
suitable melt. In contrast to other investigated metals that
are toxic, high melting, or very expensive the metallic tin has
a very low vapour pressure and can be well applicated as a
washing liquid.
Due to their low oxidative stability metal melts are only handab'te
in the cane of gases with low oxygen content which are present in
coal gasjfjcation processes. The possibility of separating gaseous
constituents by liquid metals is restricted. So they Eire mainly
appropriate to the separation of solids at high gas temperatures
and pref.r.uror,. Laboratory experiments have pointed out that metals
a^c well f.-prayable and that their ability to remove dust is
compc? rabl o to that of water. Qualitative experiments concerning
the separation showed that the adhesive forces at the surface of
the tin ;:re sufficiently strong to enable the sticking of dust.
CK/ing to tho difference of densities the solid particles ponetrated
inlo the bulk liquid get very fast to I he surface. Due to their
high den;..i i'.y the tin droplet;; dispersed in the gas can be separated
easily E.OU v.JLh high efficiency ucing clf.-.-vihy and r.iomenturi forces,,
Or.i entivL"; ny r>?.r surer,,en ts have shown up that, the separation of solid
PI rti cles f.i.ora the tin melt con by carriod out without losses of
juphaj.. Thj :-> i." a ncr-cssary priv.-jrjridj Lie ts of
I h^ iaa hcrr ; <.:'.\ .
261
-------
In contrast to inetal melts it is possible to use inorganic salt
melts as well in an oxidizing as in a reducing atmosphere. Besides
solids a large number of gaseous constituents like hydrogen chloride,
nitrogen o:ddes, sulfur oxides and hydrogen sulfide can be separated.
Depending on the gas to be cleaned and on the gaseous and solid
pollutants, compounds of sodium, potassium, and calcium can be
appljcated some of which have low melting points and behave at
high temperatures like water of room temperature. The vapour pressure:
of the mixtures are sufficiently low even at high gas temperatures,
especially when the gaseous and solid constituents are reacting
with the nielt.
It js notcv.'orthy that alkaline carbonate and hydroxide melts can
bind a great number of oxidic materials by solving or chemical
reaction. Laboratory experiments .with alkali hydroxides have
pointed out that silica can react with the melt nearly up to the
Ktoicbiomctric ratio without affecting the wanted physical
properties of the melt. This result is important in as much as
fJlica is a moJn constituent of the fly ash of coal power plants.
Experiments hive shown that fly ash is completely soluble in
c'lkalinc jiieJlK. Particles of coal and soul, however, remain
i.ndissolvad a.nd may be separated from the melt by mechanic
methods„
According to experimental results the use of nickel is
recommended ac a material for a laboratory plant because of the
great corrosivity of the alkaline melts. On an industrial scale
it may be possible to use cast iron,
262
-------
For the: investigated salt melts the dust separation is in
principle compilable to the conventional wet scrubbing process.
The alkaline melts show special features due to their chemical
rear: I j vity . The melts can precipitate as well chemical reactive
dupt particles ss nonreaotive substances like coal particles.
In principle it is possible to reprocess the salts former! by
reaction with the gaseous components. Usually this is not necesseiry
because the cost of the melt material are sufficiently low. Commonly
disposal problems are not to be expected for the waste materials,
For in? Lance the waste materials yielded by the purification
of gnses from power plants can be converted into a nearly water-
insoluble substance, at least by addition of fly ash and by
heating up to temperatures of more than 8OO C, The disposal of
the re.c;ulting v/dSto products causes cost between 2 and 4
per ton. These cost arc comparable low with respect to the
erpplo/ed irator i.-.'l s and to other v,Taste substances.
Jnw;-- i.iq -tiour, in this fie.Jd are svpported by the riiinstry of th
Intc. r i or olid by i hc> iJitiv/oH bvitidesaint .
Prc ! l'-»i i,a ry stud.' c-r; have r-lready bten finished. At present a
•3
lyLoix.toyy plant for a cj£s flow roto of 300 iV'/h as r-hov;n
in ±'f, urc 13 j r, u;]6.r:r corif, truct j on . The gas *lov; is led in a
cloi-c.r3 .;: i j'cviit , 'il;e .scrubbing system is cor.'o^irted to a separate
licju!-1'' j'<-'prorei. .---ing circuit. Gas ten,peraturr\s of snore than
i"1
60O c: c ~:M be r.i;i j i; rained, .it ir-. oor.&ibJe to i'-^e vc:.x~ions gc-s
coi';;r-,._ -; t i ._,,; ^ ,-,^0 io out)] :> /, ; tr Ixii'L net al and r.alt me Its for dust
p.. cc .! ; ' 1 1 n ';: on c.jt-'i the r.f.i>=- ration of gciseoun c ,'v;stituenty . The
j/!^; r. C]',;^rr i:c^a r. I. rorrr.Vi ,,':c-~,suvc !..>'".>;aube th;. i^hysical <.:oj»di x.ii.-. i'
263
-------
for a separation are not changed considerably at high pressures.
It must b'e seen if and to what extent the results of the
prelini.uia.cy investigations will be confirmed.
Summarising it can be stated:
The optimum use of new coal conversion technologies for power
generation requires the development of suitable gas purification
system;-, at high tmuperaturep and high pressures. It has been
shown t:hr;t in principle scrubbing processes are suitable. The
separation of solids should bo also feasible with electrostatic
pro dpi ti Lors anc! fabric filters while gravity and momentum
separators are not of interest because of their low efficiency.
It can bo orspeclc.'d that in about tv.'O or three years the first
gas pur ; vacation :-.y.c (.".cm oprrating at high temperatures and
press1,)"i es will be. employed on a technical scrle.
264
-------
Q)
.C
CJ
O
O
6
to
(3
o
Q)
Q)
c*j O
^
Q) C
b; Q)
*- r
- p
0) O
Q- H~
t O
O
M
265
Q)
CO
.0)
tD
Q)
C3
O
D
Q)
-C
O
O
O
C
Q)
QQ
QJ
j5
•- ' ml? tin.li r
•^ C
•C W)
Q) -4-^
-5 2
D
w i •«:
if =
-------
266
V)
QJ -^ ^
-C U)
u c
."> CJ
c ^
I»
" t:
«> fe
^ ^
ersitti
Un
-------
10
I.
0)
-0
-Q
3
0
^
CO
o
•^
2
to
o
o
0)
"S
3
X)
^
QJ
X
-H
10
3
0
l^
0
Xj
c
o
^
' —
a
Bll»ll««».
C
o
"^
N
Q)
O
a
o
-c
— ~—
to
L
O
U,
O
.0.
n
Q)
a
C
o
'«-.
o
Q>
^**
O
W.
O
D
L.
a
a
to
E
_3
r
<;>
Inom
!
^0
3
X)
0)
•C
o
Q)
tn
O
^
c:
u
•_£•
a
•*.,.
0
o
0
as
-c
•-
£
c
.0
•*—
(3
*—
*~
'C
J
Ol
c
-Q
-Q
3
O
to
~ Cj C
&-, c o
v, •-. u J
u.
o
o
*~
-Q
ic
0 to
L,
-~-
>-. QJ *' 0)
O u «J u
X; Q. i C\
C"
o
u
u.i
O V.
o
>*
0) ° , '*-'
i o C;
*- q,. u o
?| =1
--Goo
I) (i U ( i
C*
o
-u.
O
-Sf a.
— u
S o fc
•+- ^ 4—
? — °
^. ::> "o cj
O ^ u q
^ "*"* tr u
^J r "1 *""''
• T < i ~r_, v
.™,,- .-»««««»«„«.„ ..,„
.
— .* '
o n:
-»- o
c ',";
CD U
E 3
X) to
f; c
3 0
•«- 0
267
^
Qi
O
s
. ^
"f~l
^
r, •£*'
*^
v
i,,'
0)
0
E
3
IP
-C.
E '!'
•"
^
QJ
N
0
•4-j
E
3
O Cr>
r- ~~
C -f
._.
o
tu
" r-
0 ^
X) a
C) o
f- l~
*~ t
to
10
0)
c
*t
.
~~
u
'i>
".,
• t^r^- -t*.^= ,f -W-P-!
C3
v~ If)
c-j .^'
Lf) u
O
U L.'
o j"
O f\
LO i
r-, v^
I- U
0 0
Q
O
C3 O
-- (Nj
.i; .^
0 0
,,~.^,^~ ™
v_ G
X) >«~. fi
,. 0)
to g o u
X) X; 0 10
0 P- f!
o o t a>
"!" C.t ^~
1-. -1- Q
w O
•'^ ^b p D
!
5 ^^
'
\
i
i
j
1
i
I
•;
!
|
i
i
i
(n i
v) I
c. :
O
H«^.
»^.
|
^^
Cs
O
' *^N,
*^-»
" —• -
«0
(/)
C3
~~—
( \
^
A- » ~
• ~ 01 1
C 1.; :
'n i"'
• i
O *..'
•c t, :.
t-> C lj
. '-^ ;<
r- '-
U -~j ,
u ''j
-------
o
.
o
-O
R
.8
o
.o
CO
(Ng
UJTl
o
o
c\
£
CD
.c
"S
o
o
o
Q,,
(f}
W
cc
-C
-?•*.
• *~^.
•(,.
lf-h»
(ft
QJ
C
-2
o
V
••~\
o
H~.
O
C,
IE
o
-o
^ 0
«- c
D D
CD
-o
o
QD
268
0
-------
dl
5
._
t
.._,__
material
glass
-hrysotHe -
asbestos
amphiboliie-
jsbestos
quartz
basalt
ceramics
slag
rock
graphite
oxidizina
Jtmor,nhere
reducing
itmctphere
metal
temperature -
resisting up to
( °C )
550
550
800
1200
1000
1200
700
700
400
>?000
.. ..
700
manufactured
in to
fabric
fteect;
fabric
/Veer-?
fabric
fleece
fabric
fleece
fabric
f leecc
fleece
f leece
fleece
felt
fleece
fabric
felt
lowest dia-
meter of
tfis fibres
(p.m)
2-5
2 -10
2 -10
2- 5
10-20
4 -20
5 -20
15-30
8- 15
4 -W
chief constituents
5i02 A1203
CaO
B203
l
Mg5[(OH)8S,,Ow]
*""""" "*'• '
Ca,Na,KlMg,Fe,All
MnJi.Si.F
!
Si 02
Si02,Al203, CaO,
MgO.FeO
S/02JA1203
s:o2,Ai2o3.cao,
MgO
Si02lAl2031CaO,
MgO
C
C ,Cr, Ni, Cu,Mo,
Fe
,1
schp
lk \Fibrous Materials for the High
er:i:ct £sscn]Tc!rnperafare Filtration
UA
269
-------
c
' "s*^
!/)
d
o
-CJ
o
c.
Q)
CD
CD
c;
CD
O g
"c p-'
cu ^;;
CD .Q
a O)
Lu "r
270
u ;•
-------
5 ~
0-1
-sparkover voltage
•corona starting voltage
0
200
600 800
gas temperature (°C)
Hechanische
Vsr fahrcnstcchn i.'<
Operating Ranges of an Electrostatic
itiit Es:^.-: Precipitator versus Gas Temperature
UA
7
-------
75-
0)
•2
6
o
•4—
D
O
CU
10-
5-
• sparkover voltage
'corona starting voltage
gas pressure (bar)
Mcchanische
Vcrfahrenst^chn Ik
Operating Ranges of an Electrostatic
'Jnivcrsitat Csscn\Prec!pitator versus Gas Pressure
UA
8
272
-------
*— c\l
N
c
o
•4"**
O
^
o
c
p
^
3
Cr-
Uj
\
i LU0"
|j_ '
Cr
'' 'i
_q- u.
"bj
CL
6
»-L
O
,L
r
-«a
1
p-) «sj LT) Co tN%
-^
*-^
Q)
V
Q)
^
ci
^ "V*-
^ i °
o- uj
5 . Q. • tD
' ^"* s^~.
^ 'o a ^T
0 Uj ^ b
* • P* "^^ •
K f-». ».!> V) "*^ (\i
»^ C?" CO C" Q-
^ " " - i
11 -C V-
I; c-> -c
^ ^ Cr Uj° i*-
^
(/) "*--
Q) CD
i? ^0
t) ',7 ^
;< - (3 ' ^ -j - cu
i (- ^ "r- >. O
i fJ ( j W — ."*" '-^ I*~
C~ """ '""'*' T~i O ^ O
_ CU--; 10 t Cl ^ O '"~ ^
D t IT N 3 0 -C .Q> ~~ 't
°» ..^ ^ ?, "S a ° ^ ^ §. o
•S! tJ ^r o L' ^ '^ c1 c ^ 2
1 >-J rtY '"'tfrt ft\ " * •" ' * J —
C3 OJ r!' w) W t 2 -d --~ O ~-
, — "o ',-: '> 0 C) i» 0 *- .9-
-i. -•- \'^ l£ -'J v'- O P (/) (j
10 (, (J 10 <-, i"c ^ ^ fj U) ^
3 t.;i ,r Cj O 3 i; C) ••-- O t~
"Cj '•'- o Cj^ r-i c: o> o C G: Q-
1
i , o (o n ,, . a. i: a.
I cr tij is- u1- r: cu Uj > i= ^5
273
2-
(f)
i^.
O
-*i
CJ
-*».
.§.
(J
Q)
^.
CL
.0
-i^;
D
•+»,
(f)
0
t->
H^
(J
QJ
Cu
H^
O
£
•«^.
O
^
Oi
s!
c
.0
•4^.
D
t-.
.c^
S
**^.
C3
.0
•d
QJ
i^.
O
CD
-C
~-br_
--5E £
c 10
-c to
u uj
Cj
0) »-
-c "> • o
bi c °
. "> d) •-.
'? •- IT
|^ fe
1 -'-
. J fi> C
•£ ^ ^
-------
w,
th
W.
thr
100
W
th,
corresponding to 100°C,1bar
200
300
UOO 500
600 700
800
gas temperature (°C)
Theoretical Migration Velocity as a Func-
Mcchanische
Vcrfa!ireni>tech.nir<
Universitdt £~s.s.-/; tion of Gas Temperature and Gas Temperature
274
UA
10
-------
10
1L
6
u
Cj
-V0-
/
~JO
9.
10'
R I r I ' I F T I
1000800 600 WO 300 200
100
50
temperature T ( ° C )
Mcchanische
Vc-n'ahrens technik
Resistivity of Converter Dust
nh'crsitot Essen (from Wi'nkel undSchutz)
n
UA
11
275
-------
r~—
feeding
high voltage supply
insulators
discharge
electrode
collecting
electrode
heat
insulation
pressure vessel
Mcchanische
Verfalircnstcchnih
Univcrsitat Essen
High Pressure and High Temper-
a fare Elec trosto. iic Precipita tor
UA
12
276
-------
c-,
CD
CD
c.,
<-»
CD
2"'
Q)
O
CD
qj
o
277
jc c
• ;t -^
% c -2
.12 «.! •-,
C l~ 10
^•§ o
u t •
«?
-------
SESSION III:
OTHER COLLECTION DEVICES
JIM ABBOTT
CHAIRMAN
279
-------
HIGH TEMPERATURE, HIGH PRESSURE ELECTROSTATIC PRECIPITATION
By:
P. F. Feldman, J. Bush, M. Robinson
Research-CottreTi, Inc.
Bound Brook, NJ 08805
281
-------
HIGH TEMPERATURE, HIGH PRESSURE ELECTROSTATIC PRECIPITATION
By:
Paul Feldman
John Bush
Myron Robinson
Research-Cottrell, Inc.
Bound Brook, New Jersey 08805
ABSTRACT
This paper presents results of work conducted
by Research-Cottrell under EPA Contract 68-02-2104.
The purpose of the work completed to date was to
demonstrate the ability to generate stable corona at
temperatures to 2000°F and pressures to 500 psig,
thus establishing the feasibility of electrostatic
precipitation as a means of particulate removal from
the effluent of fluidized bed combustors or coal gas-
ifiers at high temperature and pressure. The work
was quite successful in demonstrating stable corona
generation and in defining ranges of temperature and
pressure over which the stable discharge can be main-
tained .
Gases investigated were air, flue gas, and simu-
lated (noncombustible) fuel gas in coaxial wire-pipe
electrodes. Pipe diameter was fixed at 3 inches;
wire diameter varied from 0.062 to 0.125 inches.
Results are reported for both polarities in terms of
current-voltage characteristics, corona onset and
sparkover voltages, and critical gas densities above
which sparkover alone, without antecedent corona,
occurs .
282
-------
SUMMARY
The objective of this project was to demonstrate the techni-
cal feasibility of electrostatic precipitation at temperatures
to 2000°F and pressures to 500 psig. By technical feasibility is
meant the ability to generate stable corona over the range of tem-
perature and pressure indicated. This was accomplished in a lab-
oratory scale tubular precipitator in no-flow, particle-free oper-
ation. This method of operation was chosen because it allowed a
well-controlled, economical evaluation of the electrical charac-
teristics of the system over the total range of the variables. A
second phase program is needed to carry the work further into
evaluation of particulate collection characteristics.
The primary variables studied were temperature, pressure, gas
composition, discharge polarity, and discharge electrode geometry.
Data were taken in the form of current-voltage curves to define
ranges of the variables over which electrostatic precipitation is
feasible.
The work was very successful in that stable corona was demon-
strated over the entire range of temperature and pressure of in-
terest to the emerging fluidized-bed combustion and coal gasifica-
tion applications. Furthermore, this work indicates that condi-
tions for electrostatic precipitation actually become more favor-
able as temperature and pressure increase.
283
-------
FUNDAMENTALS OF HIGH TEMPERATURE/PRESSURE PRECIPITATION
The electrostatic precipitator appears to be unique among
conventional particulate collectors such as fabric filters, cy-
clones, and scrubbers, in that elevated gas pressure offers the
promise of increased collecting efficiency. Such improvement re-
sults from the potentially higher sparkover voltages, and hence
greater electric field strength and particle-migration velocities,
attainable at high pressures . The advantage holds for precipi-
tation at high temperature, as long as the condition occurs in
combination with adequate pressure.
There is a fundamental problem encountered in designing a'
precipitator for a given high temperature/pressure service. This
is our incomplete knowledge of i) the range of variables (pressure,
temperature, electrode geometry, gas composition, polarity) over
which a stable corona discharge can be maintained, and ii) the
current-voltage characteristics in that range. In particular,
there exists for the positive discharge a critical pressure above
which sparkover alone, without antecedent corona, prevails. When
the discharge polarity is negative, the critical phenomenon is
not so precisely defined, and a postcritical discharge (often un-
stable) may be found at pressures extending beyond the critical
value .
Two opposing effects are responsible for the phenomenon of
the critical pressure. First, shorter mean-free paths at elevated
pressures impede ionization by collision and so tend to raise the
sparkover level. Second, the denser packing of gas molecules
renders photoionization more likely and reduces ion diffusion.
284
-------
Thus, pressure facilitates streamer propogation from the anode
across the gap and, at the critical pressure, sparkover results.
The likely explanation of the relatively low value attained
by the positive sparkover voltage and its concomitant lower cri-
tical density is as follows: Intense ionization of the gas is
produced in the high-field region in the vicinity of the dis-
charge wire which attracts and removes the highly mobile electrons.
The heavy positive ions are repelled from the wire and move slowly
toward the collecting electrodes. However, on the far side of the
ion cloud (away from the wire), the field, suffering from positive-
ion space-charge distortion, is increased, and with it, the rate
of ionization. Conditions are now favorable for the growth of a
positive streamer to develop toward the cathode. On the other
hand, when the wire is negative, the positive space charge sur-
rounding it in the corona sheath tends to shield the wire from the
anode. This action reduces the field on the anode side and so a
higher voltage is needed to promote a spark.
Thermal ionization is another effect which must be considered
in high temperature precipitation. At normal temperatures ioniza-
tion is governed by i) electronic and molecular collision processes
in the presence of an electric field and ii) photon absorption.
At very high temperatures an additional mechansim enters the pic-
ture: thermal ionization. This introduces the following new pos-
sibilities.
i) Electron-ion production by collision of the gas molecules
with each other. Such ionization may occur even in the absence of
ionized particles and a high-intensity electric field. It is the
high-temperature condition that provides velocities and kinetic
energies high enough to cause ionization.
ii) Photoionization resulting from thermal emission of the
hot gas. When a high-temperature gas flows through a precipita-
tor, it will emit quanta in accord with the laws of black-body
radition.
The electrode surface and the walls of the pressure vessel
will reflect most of this radiation. Thus, photons capable of
285
-------
ionizing are always availabe when the temperature is high enough.
Calculation, however, shows that this condition is probably not
met under present circumstances.
iii) lonization by collision with high-energy electrons that
have been generated by the above two processes.
Apprehensions arise for precipitation at very high tempera-
atures because significant thermal-ionization rates may lead to
catastrophic currents at reduced voltages. Earlier predictions
that this condition might occur at a temperature of about 1500 F
3 4
have been shown experimentally to be untenable . Revised work
now suggests that whereas thermal effects may become noticeable
below 1500 F, practical high-temperature precipitation may be
limited only by temperatures exceeding 2000-2400 F. The lower
limit of this range would apply to gases containing a component of
low ionization potential, say potassium, in quantities as low as
4
one atom in 10 . Potassium is singled out because of the alkali
metals (all of which have characteristically low ionization poten-
tials), it is the most common. Hence, the probability of its
appearance in industrial gases is not remote.
286
-------
EXPERIMENTAL APPARATUS
Figure 1 shows the configuration of the test precipitator
used in this program. It is a wire-pipe design enclosed in a
pressure vessel. The pressure vessel was designed for pressures
to 500 psig and was assembled in three sections, each serving a
specific purpose. The top section contains the feedthrough bushing
for applying high voltage to the discharge electrode and a pres-
sure relief line to protect from over pressurization. The bottom
section has a side access opening for adjustments and observations,
a bottom support insulator to center the discharge electrode, and
the gas inlet. The center section of the vessel holds the preci-
pitator tube surrounded by a three-zone heater used to reach the
desired operating temperature. A layer of Kaowool insulation
separates the heater and the pressure vessel wall.
The collection tube electrode is a 7.26 cm internal diameter,
Inconel 601 tube, sectioned into three electrically isolated
pieces. This arrangement provides a center section with precisely
known area and uniform field for corona current measurements. The
endpieces are flared in order to prevent sparkover at premature
voltage levels. The discharge wire is also of Inconel 601.
The precipitator was powered by a power supply capable of
400 kV and 60 milliamps. Gas was charged to the precipitator from
compressed gas cylinders to obtain the desired gas compositions.
Moisture was adjusted to desired levels by passing the gas through
a humidifier at controlled temperature.
287
-------
TEST CONDITIONS
The experimental test plan was designed to establish ranges
of stable corona generation as a function of temperature, pressure,
gas composition, discharge polarity and discharge electrode geo-
metry. Table I shows the ranges of these variables studied. Tem-
perature was varied in 500 F intervals from 500 F to 2000 F for
all gas compositions, and, with air, ambient temperature data were
also taken.
Pressure was varied in 50 psi intervals from ambient to 500
psig. In moving from one test condition of pressure to a new
condition, at constant temperature, gas was always released from
the vessel to maintain constant composition and thermal stability.
At each pressure level, results for both negative and positive
polarity were obtained.
Data were taken in the environments of three different gas
compositions: air, combustion gas and substitute fuel gas. A
substitute gas mixture was used in place of actual fuel gas be-
cause of laboratory safety requirements. This mixture was similar
to actual fuel gas except helium was substituted for hydrogen and
carbon dioxide for carbon monoxide. The resultant mixture was
similar to fuel gas in important physical properties including
thermal conductivity and ionic mobility. The fuel and combustion
gas mixture used are shown in Table II.
Data were taken for three discharge electrode sizes as shown
in Table I with air. For the combustion gas and and substitute
fuel gas mixtures, only the 2.34 mm wire was used.
The primary data taken were current-voltage curves for each
of the experimental conditions. The current-voltage curves were
288
-------
obtained on an X-Y recorder by recording the curves for both in-
creasing and decreasing voltage levels and repeating each trace
Sparking voltage was determined as the final voltage attained
after being held at sparking for two minutes. Corona starting
voltages were determined by (1) observation of voltage at which
corona pips disappeared on the oscilliscope with decreasing
applied voltage and (2) extrapolation of the current-voltage
curves to zero.
289
-------
RESULTS
The raw experimental data, consisting of curves of linear
current density (mA/m) vs impressed voltage (kV) are reproduced in
Figures 2 through 5 for air, Figure 6 for simulated combustion gas
and Figure 7 for substitute (i.e., noncombustible) fuel gas. Corona-
starting and sparkover voltages, derived from these curves and in-
dependent measurements, are shown as functions of relative gas
density, &*, in Figures 8 to 10 for air, Figure 11 for combustion
gas, and Figure 12 for substitute fuel gas.
The first and most important objective is to examine the data
for the purpose of establishing temperature or pressure limits to
a stable corona discharge. Such limits may be caused by: i) ex-
cessive currents at low voltages resulting from thermal ionization
(where "excessive" and "low" are taken from the point of view of
practical precipitator operation) and ii) the disappearance of
(stable) corona due to the manifestation of the critical pressure.
Examination of the data shows that catastrophic high-tempera-
ture currents are not observed in this study under any conditions.
This significant point is evident over the full range of experi-
mental pressures and temperatures and both polarities.
It might be expected that runaway currents at low voltages
are most likely to occur at the lowest gas densities. However,
for the negative corona at 6 less than about unity, the reverse
is generally true, i.e., the presparkover currents are much less
*The relative gas density 6 is taken with respect to atmospheric
pressure and room temperature (294 K)
290
-------
than at higher densities. For positive polarity, the low-6 pre-
sparkover currents are mixed: in most cases they are not lower
than at higher 6. Still considering the positive discharge, the
data show that combustion gas and fuel gas to a lesser extent,
but not air, reveal a tendency to sparkover without corona at the
lowest densities. But, in any event, a problem of high gas con-
ductivity associated with low densities does not arise.
On the basis of earlier critical-pressure studies ' it might
be supposed, at least at lower temperatures, that the critical-
pressure phenomenon would set an upper-pressure limit to the posi-
tive current-voltage curves and the associated voltage-density
curves — upon converting density to pressure -- that the positive
critical pressure increases with temperature. This comes about
because the greater molecular diffusivity prevailing at the higher
temperatures more effectively suppresses the sparkover streamer
before it completely bridges the interelectrode space . The
effect described has not hitherto been demonstrated over so wide
a span of temperatures.
The negative critical pressure, as already explained, is ex-
pected to be higher than the positive, all other conditions being
held fixed. Indeed, by comparison of families of curves for each
polarity, it is clear that the negative critical pressure always
exceeds the positive. It is further apparent that the negative
critical pressure, like the positive, increases with temperature.
In other words, the higher the temperature, the greater the range
of pressures for stable negative corona.
Comparison of the sparkover-voltage vs gas-density data of
Figures 8-10 reveals a tendency for the positive sparkover voltage
to exceed the negative at temperatures of 533 K and higher and
for low air densities (6 less than about 1 or 2). The data are
not unequivical on this score in each case, but the trend seems
clear, particularly when supported by experimental results of
earlier workers. The significantly higher negative than positive
currents prevailing at a given voltage at the higher temperatures
are, however, unmistakable.
291
-------
Somewhat higher negative than positive currents that may be
observed at the lower temperatures are, in part, to be attributed
to the significant free-electron component of the current present
for relatively long mean free paths (low 6) and narrow interelec-
trode spacing.
Again, it may be generally (though not invariably) seen from
Figures 8-10 that, above an air density of 1 or 2, the negative
sparkover voltage is higher than the positive. Since increased
density reduces the mean free paths and mobilities of the charge
carriers, enhanced electron attachment and increased negative-ion
space-charge density might be expected to lead to higher negative
sparkover voltages. That is, high pressure in combination with
high temperature restores, in a sense, the low-temperature situa-
tion .
In the case of substitute fuel gas (Figure 12) the positive
sparkover voltage exceeds the negative, over the full temperature
range shown, up to a density of 6 or 7. For combustion gas
(Figure 11) the transition occurs at about a density of 4 for
temperatures of, or greater than, 1089 K.
As temperature and pressure are increased together, for all
of the experimental situations, it is clear from the data that
precipitation is possible at significantly higher voltages than at
normal conditions. This is a most important fact when assessing
the viability of electrostatic precipitation for high temperature,
high pressure particulate removal applications, especially in com-
parison to other collection devices. The reason for this is that
the rate of particle collection in a precipitator is roughly pro-
portional to the square of the electric field strength in the pre-
cipitator. The field strength in turn increases with applied vol-
tage. The net effect of an increase in operating voltage is there-
fore a proportionally larger increase in particle collection effi-
ciency, or a decrease in precipitator size. Thus precipitation
becomes more efficient as temperature and pressure increase to-
gether .
292
-------
Other particle collection devices such as filters of various
types, cyclones, etc. do not benefit from increasing temperature
and pressure. In fact, performance deteriorates in these devices
because of increasing gas viscosity and decreasing molecular mean
free path. The higher voltage effect in electrostatic precipita-
tion greatly overrides these adverse effects.
293
-------
CONCLUSIONS
Following are the major conclusions derived from this work:
1. There are no temperature or pressure limitations to elec-
trostatic precipitation over the range studied.
2. Precipitation becomes more efficient with increasing tem-
perature and pressure. This is in direct contrast to the trend
of other particle collection devices.
3. Critical pressure increases with temperature.
4. Negative critical pressure is higher than positive.
5. Negative currents are higher than positive in most cases,
294
-------
REFERENCES
1. Robinson, M., "Electrostatic Precipitation" in Air Pollution
Control, W. Strauss, ed. , Vol. 1, Wiley-Interscience, New York,
1971, pp. 227-335
2. Cooperman, P., "Spontaneous lonization of Gases at High Tem-
perature, "Conference Paper 63-173, Amer. Inst. Elec. Engrs.,
1963 .
3. Brown, R. F, and Walker, A. B., "Feasibility Demonstration of
Electrostatic Precipitation at 1700°F" J. Air Pollution Con-
trol Assoc. 21, 617-620 (1971).
4. Cooperman, P., "Spontaneous lonization of Gases at High Tem-
perature," Paper ES-MON-6, Inst. Electron. Engrs., 1971.
5. Robinson, M., "Critical Pressures of the Positive Corona Be-
tween Concentric Cylinders in Air," J. Appl. Phys. 4 0,
5107-5112 (1969).
6. Howell, A. H., "Breakdown Studies in Compressed Gases,"
Trans. Am. Inst. Elec. Engrs., 58, 193-204 (1939).
295
-------
TABLE I. TEST VARIABLES
Temperature: Ambient to 2000°F
Pressure: Atmospheric to 500 psig
Discharge Electrode
Diameter: 1.58 mm, 2.34 mm, 3.18 mm
Discharge Polarity: +, -
Gas Composition: Air, Combustion Gas, Substitute
Fuel Gas
TABLE II. GAS COMPOSITION
(Volume %)
Component
co2
He
°2
N2
H 0
Substitute Fuel Gas
23.
18 .
--
53.
5.
0
5
5
0
Combustion Gas
9.2
--
2.8
83.0
5.0
296
-------
TO POKER SUHLY
MA
TOP INSULATOR BUSKING
RUPTURE DISC
HIGH VOLTAGE FEED
THROUGH ROD
DISCHARGE ELECTROi
TUBE ELECTRODE
HEATER
ACCESS PORT
BOTTOM SUPPOUT
INSULATOR
GAS INLET
TEST PRECIPITATOR
Figure im Laboratory Precipitator and Pressure Vessel for Test
Program
297
-------
ro
oo
in
CTi
CM
O
cn
0)
0)
s~\
r— -I
s
aj
EH
-P
(0
M 0)
-H T3
i 4->
M U
Q C
I - 1
c: a
•H
C
C M
O -H
^ r;
rC
EH |
cn
0) •=;
U CM
CD rci
Cn
rd S-i
-P O
I
-P O
0) C>
3 C
U 03
*
CN
-------
S-,
o
oo
MH
C
s-i
3
•4-1
2
CD
EH
(13
C
•H
rfl B
EH '-n
a)
>
U O
0) U)
&1 a)
tO Ti
-P O
fH M
O -P
> u
I D
-P -
d w
cu
M D
M H
3 -H
U IS
a)
299
-------
H
O
M-l
cri
CO
o
0)
M
d
e
OJ
E-
rd
vr
*r
CNi
Q)
EH LH
r^
in -
(U •
> -H
CJ O
0) U}
Cn 0)
rt t)
-p o
.H M
O 4->
> 0
I 0
0)
w
G)
S-i
0)
M
-H
300
-------
O
0)
M
3
jJ
nj
a
e
0)
EH
-P
(d
M
cl •*
0) v
>J CO
(tf •
EH CM
Se
> e
M IT)
3 r^
U in
Cn
ni 4-'
-P O
1-1
O tn
> a)
I T3
•4-> O
C M
0) -P
H O
M 0)
^5 H
U t,
in
01
Cn
-H
301
-------
U
g I
(/I *
3 * •*
"*N
°t
«/»«> 'i
S-l
O
(U
-P
X
•H
in *
*ri I-M
c
rd cc
O
C -H
-H
^
C\y
hH
(0 rH
EH CO
t/1 *
0) «
>
S-l m
3 n
O in
G M-i
Cn O
03
-P U)
rH O
O Sn •
> 3
-------
(1;
rci
O
4-J M
fd -P
u
OJ OJ
H 1-1
3 W
-P
X QJ
•H M
O
0) ro
3 •
fo (N
QJ fO
-P
3 H
-P C
•H U-l
-P
cn ^
co
C
(0
CD
A; H
(13 r-l
EH 00
U) -
Q) X
U LO
CJ 4-1
Cn O
m
-p w
rH 0)
O M
> 3
I 4-)
4-1 fC
C M
0) OJ
H D,
•H g
3 (D
U E-
QJ
S-i
3
ijl
-rH
303
-------
G
O
•H
4-1
O
w
fO
oooooooo
AH '39V1~IOA
A»'30»nOA
oooooooo
AH'30V110A
L'
oooooooooo
A»'39VJ.-IOA
. I
(~* *v^
•H
n
QJ ro
4-1
O
4-1
4-1 rtS
cn
O CO
S-l C
O CD
U Q
13 S-l
G --H
Cn
•H -H
0)
V}
0)
00
Q)
M
3
tn
•H
304
-------
A»'30»nOA
X!
-p
m
n
W
CD
0)
tn
M
-H
O
•r\
.Q
A* '39vnoA
oooo oooo
AH'39»iTOA
o '• s
OOOOOO
Al'39»ilOA
u>
«
en
in
-p oa
n5 -P
-a
C-H
O tt -
O CD g
U Q ^
T3 M ro
C -H •
(T3
-------
°
___! J _1_ 1. 1 I J__
0)
o
O 4-1
C 0)
O -H
4-1
AV3DV110A
U .
* o
R' '
0°
oooooooo
0)
r-7
°
-H .
U3
03
-H
-H
U) ro
tn
4J c e
i—i nj £
0 in
£> h^H f^
LO
00 rH
-P
A1-39VDOA
4-1
0!
03 4-1 E
C -H <'
O en "=3*
O 0) •
O Q CN
Cn r-
•H -H r-H
(U
PM
306
-------
, ;-'•;< ' i i
rrt • c
.
1
-' f • o
• " * J > *°
L a>t £ 5 • °
j O% « W "^
L J - J L__ i 1 - L j , J i_ft.
oooooooooo<
AH '39VJL10A
p I 1 1 1 i I i i
rrj 1 0°
1_ * ° -
• 0
| 01 • O ~
>-- O • o
1- 0 6
tK E LU • 0
£* ? > • o
L U 5 13 Z • O_
1 J ! J I 1 i 1 1 1 «
JOOoOOOOOO
AH '39V110A
" r~7 i r i i r~n
i * °
r'*"* • o
•
r °
• o
I" w * °
^
^ LU * °
- 1 * N * " ° ~
1 o — " "J • o
I 111 «°
A«'3S»nOA-
]"
-
-
• .1
* c
a • •>
-1 I • -
r-'s""i i i o ; i-r~
AH ' 3SV1TOA
&
a>
10 >
-
* 5
-o 0
(M
—
c
o
-
(0
"1
•4 UJ
0
_
3
=
'
X
•" K
^ X
LU
"^ O
tf
fO
M
3
¥
q.
•2
V
~
o
'" ^
" z
o
„
~
L
C
! T ' ' r
XJ
• 0
3 " °
0 £ • 0
K Si " t ' "
0 S ' O " °
oooooo^ooc
AH '38V1-IOA
— r "i — [ — r T " r i ~i i
- iy\ \ o
• o
"
< "0
2 E • 0
- E
ro ^ "^ ^ • o
g o ^ ° • o
J ] 1 1 1 I 1 1 !
AH'39V1"IOA
7 " "T 7 ' ^ T 1 '
B
"
•
* t
i- U * >
m ^ * ^ " °
0 - " O «1
L J. |_ | 11 | 1 "
AH ' 3 9V110A
- 0)
•
_
.
• "
!
_— . U 1 1 ! ' ' "-
O" OOCOOO^
Al '39VJ.10A
cn
fd CTl
O oo
« o
» O
tfl >- "^ ^
s -P t4
* S U3
„ Q r-< 1
r— ' ri
~ rQ "H
S 00
: 8 .
_j *
(U ro
4J n
2 nj LO
1
,5 03 -r-1
M C
O QJ •
m Q QJ
OJ W O
0) rtf >H
iji O -H
tO U
n- i ^ n) QJ
^ H > H
O'-H W
> -P
2 tO 0)
» tri iH in
C OJ -H
^ 1 '5 m g
-° m o i
« 4J ^f
CO G ^
O fO
^ fd *H •
S^ 4J CN
.— . r \
2 C 03
0 d
U CM M
o
T3 oi m
! C
td en ^
- en vo
C OJ vo
t? -r-j ^ CO
rX 3 cH
"i r« X Ti
" S &4-H C
* ° CO S t!
_«i
r' ' 1
H
QJ
M
Cn
•H
307
-------
QJ
-P
X
•H
in «•
03 W
o
CTi
rH 00
CD O
u
ATaOVllOA
y.
A< '30»iTOA
OOOOOOOOOO
AH '39V110A
OOOOOOOOOO o
AII 'aovnoA
0)
Ai'aovnoA
CD «
-P
-P r-l
-H 00
-P
m
rd in
C -P
•H re
en >i
CD -P «
rji-H 0)
rd en 13
-P ',1 O
O G -P
> u
in CD
tn ro t—|
cow
•H
-P CD CD
M > M
rO -H -H
-P -P S
rO
C
O
ro
O O
^ O fO
C -H
rfl -P M
U O
DI a m
c s
•H CM «
M fd >~D
rd UD
di en co
CO rd r-<
(U
SH
Cn
Pn
308
-------
PULSE-JET ACOUSTIC DUST CONDITIONING IN HIGH
TEMPERATURE/PRESSURE APPLICATIONS
By:
D. S. Scott
University of Toronto
Toronto, Canada M5S 1A4
W. M. Swift, G. J. Vogel
Argonne National Laboratory
Argonne, IL 60439
309
-------
PULSE-JET ACOUSTIC DUST CONDITIONING
IN HIGH TEMPERATURE/PRESSURE APPLICATIONS
By:
* ff
David S. Scott , William M. Swift , G. John Vogel
*University of Toronto, Toronto, Canada, M5S 1A4
tArgonne National Laboratory, Argonne, Illinois, 60439
ABSTRACT
The results of a completed investigation of pulse-jet
acoustic dust conditioning (PJ-ADC) at conventional gas clean-
ing temperatures and pressures are reviewed with emphasis on
how these performance characteristics might reasonably be
extrapolated to high temperature/pressure operation. These
arguments indicate that PJ-ADC at high pressure/temperatures
might be more effective than at more conventional conditions.
On this basis, a planned experimental program to evaluate PJ-
ADC upstream of high efficiency cyclones in the pressurized
fluidized-bed combusion (PFBC) process development unit (PDU)
at the Argonne National Laboratory, is described. Two promis-
ing features of PJ-ADC in this application are: (i) the heat
of PJ combustion contributes to the PFBC heat, and (ii) the PJ
should operate more effectively at elevated pressures.
310
-------
PULSE-JET ACOUSTIC DUST CONDITIONING
IN HIGH TEMPERATURE/PRESSURE APPLICATIONS
SECTION 1
INTRODUCTION
Acoustic conditioning of fine-particle aerosols is a
process by which the mean size of the particle is increased
and their number density decreased through exposure to finite-
amplitude acoustic fields. This change in the particulate
size distribution is important — for it can allow an increase
in the collection efficiency of a downstream dust collector
and/or a reduction in the overall particulate-gas separation
cost because, with few exceptions, collecting devices retain
higher mass fractions as the particulate size is increased.
The increased coagulation rate which achieves this rapid
change in the size distribution is the result of manyfold
increase in the particle-particle collision frequency — which
in turn results from dynamic acousto-aerosol interactions
(and perhaps to some extent through intermediate processes
such as acoustically enhanced turbulence). Experimental
evidence, and straightforward physical arguments, indicate
that the process can be particularly effective on high dust
load fine particle aerosols.
Although the principle of acoustic dust conditioning
(ADC) has been known for some time and has been shown to be
technically effective, the process has enjoyed little
311
-------
industrial application. This absence of industrial
application has been the consequence — primarily — of
economic limitations. The first of these is high operating
costs due to high specific power requirements, and the
second is high capital costs.
Scott (1975) reviewed studies performed at the Ontario
Research Foundation and the University of Toronto which
examined a new approach to ADC — in which the following two
complementary principle's were explored.
(i) Firstly, the conventional standing sinusoidal acoustic
field, used in most earlier ADC studies was dropped in favour
of progressive saw-tooth fields. This change in acoustic
configuration allowed major simplifications in the agglomera-
tion chamber yielding both capital and operating cost savings.
(ii) Secondly, the normal techniques of sound generation —
such as sirens and mechanically activated pistons — were
discarded in favour of a resonant pulse-jet acoustic source.
This not only allowed a direct conversion of fuel to sound
and the use of a simple no-moving-part device, but also
readily produced high intensity progressive saw-tooth
acoustic fields.
In summary, the practical results of this program showed
that significant — 0(10) — advances can likely be achieved by
this new process in all three areas of (a) capital, (b)
maintenance, and (c) operating costs. At the same time, the
mean particle size was increased 5-7 fold.
The more important aspects of Scott (1975) results for
the research plan described in this paper were that (i)
312
-------
mean particle size increases of 0(6) appears to be fairly
readily achieved, (ii) the process requires a low level of
technical sophistication and (iii) it is possible to make
approximate cost evaluations relating to the process should
it be proven technically feasible. By these criteria, it
appeared reasonable to examine the feasibility of using
Pulse Jet Acoustic Dust Conditioning (PJ-ADC) in the high
pressure/temperature application of pressurized fluidized-
bed combusion. (PFBC).
SECTION 2
BASIC APPROACH OF PLANNED RESEARCH/DEVELOPMENT
Development of PFBC for use in thermal power generating
stations is being carried out at several laboratories. The
process involves coal combustion in a fluidized-bed of CaSOs-
The lime solids react with the combustion released sulphur
oxides to form CaSOt,. which remains in the bed as particulate
matter. Thus, compared with conventional combustion process-
es, the aerosol effluent contains a lower level of S02 — but
the particulate mass loading can be much greater (due to
entrainment of calcium particulate along with the ash
particulate). In a sense, the gaseous pollution problem is
reduced at the expense of increased particulate loading.
In the PFBC configuration being examined by Argonne
National Laboratory (as well as at several other laboratories),
the combustion process is pressurized to approximately 10
atmospheres with the intention that the off-gases be expanded
313
-------
through gas turbines. But hot, particulate laden, gases can-
not pass through gas turbines without massive destruction to
the turbine blades resulting in unacceptably short lifetimes.
And so we have particulate-gas separation requirements at
high temperatures and pressures which are critical not only
for environmental but also process requirements.
At the temperatures 0(1150 K) and pressures 0(10 atm.)
which are anticipated, most conventional particulate removal
techniques are unsuitable. Separation by cyclones seems to
be one of the few systems that can operate under these condi-
tions. Unfortunately, the performance characteristics of
current cyclones do not appear sufficient to meet the gas
cleanliness standards set by turbine manufacturers — although
there remains some uncertainty regarding what these standards
should be. Nevertheless, it appears likely that the ultimate
-3 3
criteria may require that there be no more than 3x10 g/m of
particulate matter above 2 ym diameter.
Results of the PFBC program at ANL indicate that the mass
loading and size distribution of the particulate matter exit-
ing from a two stage high efficiency cyclone should not be
far from that which would be acceptable as turbine inlet
conditions. Moreover, a fairly straightforward evaluation
indicates that if the mean particle size of particulate matter
entering the cyclone was increased by perhaps less than a
factor of 5, the performance of the cyclone would be improved
such that the effluent WOuJLd meet turbine inlet criteria.
For these reasons it is our current plan to initiate a
research program to investigate the effects of using ADC in
314
-------
this application.
The program at ANL will evaluate the effectiveness of
pulse-jet ADC in a bench scale fluidized-bed combustor.
There are several promising aspects of this approach.
Firstly, the heat of combustion from the pulse-jet simply
adds to the overall process heat. Thus the power require-
ments will, in principle, be zero — although the cost per
heating unit for pulse-jet fuel may be expected to be some-
what higher than that of coal. Secondly, because the pulse-
jet will run at higher power levels due to the elevated
pressures it should, in principle, be a more effective sound
generator than if it were running at ambient conditions — the
result of both charge density effects and more efficient
combustion.
Conceivably, there might be deleterious ADC effects.
For instance, very fine submicron particles might be agglome-
rated into a size range below that which would be effectively
removed by the cyclone, but within a range which could do
damage to turbines. Evaluations of the mass loading ratio in
the very fine particle spectrum indicate that this should not
be a problem — but it will clearly be an aspect which shall
be monitored during the experimental program.
Turning to the experimental program itself, one of the
main difficulties of process evaluation shall be "scaling".
A pulse-jet is a device which cannot be scaled up or down
without changing the acoustic characteristic — for a pulse-
jet is a quarter wavelength device. Ultimately, it is expect-
ed that single pulse-jets will generate sufficient sound to
315
-------
treat between 5,000 to 10,000 cfm. But the current test plant
at ANL yields approximately 81 scfm, or on the order of 20
acfm.
In response to these scaling demands, the pulse-jet will
exhaust into a resonant-manifold. The manifold shall be
capable of splitting the acoustic energy into a "waste" stream
and two "process streams". Two process streams are required
in order to examine the effects of ADC on both the primary
cyclone and secondary cyclone performance. Pulse-jet products
of combustion will be exhausted primarily through the "waste"
stream duct (vent). By means of the resonant-manifold a
large total acoustic power output can be reduced to match the
amount of particulate-laden gas being treated. At the same
time, this approach will allow the acoustic field to be
varied in intensity from approximately 165 db downwards,
in order to parametrically evaluate sound intensity require-
ments. By exhausting the pulse-jet into a resonant-manifold,
and then drawing off the acoustic power as required for the
much smaller scale dust treatment streams, we expect to
eliminate both the scaling difficulties as well as allow
a greater degree of parametric flexibility than would be the
case were we to exhaust the pulse-jet directly into the
particulate-laden gas.
SECTION 3
SOME EXPERIMENTAL DETAILS
Although our experimental program is very much in the
316
-------
formulation stage, it is possible to make additional
comments regarding the experimental capability we are
developing.
PULSE-JET (PJ) SOUND GENERATION
A range of literature has been reviewed on pulse-jet
operation. From this, we conclude that the power generated
by a pulse-jet is essentially linearly related to pulse-jet
rating — that is, its ability to consume fuel. On this basis,
it is clear that our expectations that the pulse-jet operate
more effectively at higher pressures should be fulfilled.
Several pulse-jet configurations will be tried. And it is
expected that fundamental operating frequencies between 250-
300 Hz should be achievable. From private communication with
pulse-jet researchers, it appears that units designed to
operate at frequencies much above this range tend to be
temperamental. Aerodynamically valed (i.e. no moving part)
pulse-jets using propane as the fuel are now under develop-
ment for our research program.
RESONANT MANIFOLD SYSTEM (RMS)
As we noted earlier, it is intended to exhaust the pulse-
jet into a resonant manifold which shall act as a sound source
from which acoustic energy can be drawn for aerosol treatment.
A schematic illustration of the RMS is given in Fig. 1. The
manifold will be designed as a "length resonator". In this
configuration, the flat plate ends of the chamber will act as
velocity nodes for standing waves set up within the RMS. It
is from the flat end walls that the acoustic field shall be
317
-------
drawn for ADC processing. Fig. 1 shows two process sound
ducts existing from the same end plate that contains the
pulse jet. Two ducts are intended so that the aerosol
entering both the primary and secondary cyclones may be
treated.
It should be noted that all dimensions involving the
acoustic wave length, are to be adjustable. This flexibility
is required since it is impossible to predict in advance —
except within a fairly broad range — the frequencies and wave
lengths which will be generated.
PRESSURE CONTROL
One of the unique requirements of this research is that
of providing sound at elevated pressures — specifically bet-
ween 1 and 10 atmospheres. This requires that the entire PJ-
EMS be capable of operating over this pressure range. And
this includes the pulse-jet fuel and air manifolds.
To meet these requirements, the design as currently
envisaged incorporates a pressure differential measuring unit
between the PDU and RMS which in turn will operate a valve
controlling the rate of exhaust from the vent shown in Fig. 1.
This control will be critical to the entire experiment. And
it is expected that a very small Ap between the RMS and PDU
will be maintained — with the pressure in the RMS being the
higher.
Fail-safe valves for PDU/RMS isolation in the event of a
disruptive loss of pressure in either, will be installed in
the process sound ducts. These valves will also serve as
318
-------
isolation valves while the PDU and RMS are brought up to
operating conditions.
ACOUSTIC TREATMENT SECTION (ATS)
One of the advantages in examining PJ-ADC by means of
this intermediate RMS, is that alternate acoustic treatment
sections will be particularly straightforward to install. We
intend to use a simple "two flange" procedure, whereby we
interchange various ATS configurations quickly, and with
little cost, in order to cover several alternatives. A
schematic illustration of one "two flange" ATS configuration
is shown in Fig. 2. The spiral is an intentional design
which — from previous research — we believe might be effec-
tive. This conclusion comes from both experimental findings
and the impression that interactions between finite ampli-
tude acoustic fields and basic flow induced turbulence can
enhance the coagulation process. The spiral configuration
will cause additional circulation phenomena. But the use of
grids and other turbulence generators might also be examined.
SECTION 4
CONCLUDING REMARKS
It is our view that earlier work on PJ-ADC has shown that
modifications to the mean size of the particulate matter in a
dusty gas stream can be accomplished to a degree and at a
cost, that would be very attractive if such could be achieved
in the high pressure/temperature off-gases associated with
319
-------
fluidized-bed combustion processes. Since there is no
presently identifiable reason why these basic results
should not be possible to reproduce at high pressures and
temperatures — indeed the high pressure/temperature environ-
ment seems likely to enhance the success of the process —
it appears reasonable to examine the feasibility of PJ-ADC
in this application.
But, the transformation from the operation of PJ-ADC at
approximately atmospheric conditions to these higher
pressures, involves certain non-trivial advances to current
technology. For this reason, what we learn regarding these
supportive aspects of the experimental program may be of as
much interest as will our success or failure at illustrating
the feasibility of PJ-ADC under these conditions.
320
-------
REFERENCES
Scott, D.S., "A New Approach to the Acoustic Conditioning
of Industrial Aerosol Emissions", J. Sound and
Vol. 43, no. 4, pps. 607-619, 1975.
321
-------
UJ
fe
fe
2
O
(O
UJ
QL
U_
O
O
I
LJ
I
O
en
o
322
-------
From
FBC
FIG. 2 FLANGE TO FLANGE ACOUSTIC TREATMENT SECTION
323
-------
THE APPLICATION OF SONIC AGGLOMERATION
FOR THE CONTROL OF PARTICULATE EMISSION
By:
D. T. Shaw, J. Wegrzyn
State University of New York at Buffalo
Buffalo, NY 14226
325
-------
ABSTRACT
The feasibility of using acoustic agglomeration for particulate
emission control in industry is evaluated. The dependence of the agglo-
meration time, which determines the agglomerator size, on various para-
meters - such as the mass loading, particle geometric mean diameter and
standard deviation, acoustic frequency and intensity - is analyzed. Two
agglomeration mechanisms are included: the inertial capture and the hydro-
dynamic collision. Based on the estimated agglomerator size and the
specific energy consumption, possible applications of the acoustic agglom-
erator in industry is discussed, with special emphasis on the possible
use of such a device in an environment in which the combined effects of
high-pressure, temperature and chemical corrosion make it difficult to
use the conventional devices.
326
-------
Introduction
Sonic agglomeration has been a rather controversial subject ever
since the first experiment by Patterson and Cawood in 1931 [1] who dis-
covered that aerosol particles are quickly agglomerated in a standing-wave
sound tube. Shortly afterward independent acoustic agglomeration experiments
were carried out by Brandt, Freund, and Hiedemann [2] in Germany, Andrade
and his co-workers in Great Britain [3] and St. Clair at the U.S. Bureau of
Mines [4]. Their results generated great enthusiasms for acoustic agglom-
eration and eventually the large scale industrial testing of its effective-
ness. During the second World War,serious efforts were made by the U.S.
Navy and Army to use acoustic dissipation of fogs for the improvement of
visibility for landing at airports and on aircraft carriers. Although the
field testings gave a positive result in calm weather, this technique was
proved to be ineffective under moderately high-wind conditions and the
project was abandoned [5]. During the post war years several commercial
models of high power sirens with an acoustic efficiency of the order of 40-
60% were available and this spurred a widespread interest in the application
of sound waves for the control of particulate emissions from power plants,
steel mills, and other process and chemical industries [6-11]. The work in
this period was summarized by Mednikov [12] and more recently by Shirokova
[13]. However, the results of the industrial-type acoustic agglomeration
testings were invariably discouraging, mostly due to the high power consump-
tion for the sound generation. ThLs greatly reduce any interest among the
researchers in the United States and after 1953 the acoustic agglomeration
work was almost completely stopped in this country.
327
-------
Recently, there has been a renewed interest in acoustic agglomera-
tion as related to energy-environmental research. In order to reduce our
national dependence on foreign oil imports numerous new programs have been
established by the Department of Energy to seek answers to problems asso-
ciated with how to generate electricity with high-sulfur coal, to improve
central power plant efficiency and above all, to accomplish these with a
minimum environmental degradation. These new programs lead to new engin-
eering problems such as dust removal in a coal-gasification plant, solid-
seed extractions from a MHD generator, and so forth. These new problems
create a critical need for a nonconventional particulate-emission control
device which can operate effectively in a combined high temperature, pres-
sure and corrosive environment.
As an example, acoustic agglomerators are considered to be one of the
primary candidates for the removal of dust particles before the combustion
gas enters the gas turbine in a pressurized fluidized-bed boiler power plant.
Several concepts of the fluidized in-bed combustors have been considered for
power plant applications. Among tnese the ideas of using a pressurized,
fluidized bed combustor in a combined gas-turbine and steam-turbine cycle
is particularly attractive. Steam for operating a turbine is raised by both
the steam tubes immersed in the fluidized bed and by the heat converter
boiler of the gas turbine. The heat of combustion is removed from the bed to
run directly to the gas-turbine. Typically such a fluidized bed uses crushed
coal of about 1/4" diameter. Most of the fly-ash particles arc rather l^rge,
say 100 urn; however, some smaller particles are also produced which must be
removed from the flue gas to avoid the attack of gas-turbine blades by such
corrosive particulates as sodium salts.
328
-------
Cyclones are used to effectively remove large particles. Hor par-
ticles in the range of 2-20 ym diameter nonconvcntional devices are suit-
able for this application. Filters are difficult to use because of the
high temperature, corrosive environment and because of the inherent high-
pressure drop across the filters. The electrostatic precipitator is not
applicable because under the high pressure (10-20 Atm in the fluidized-bed
boiler) it is difficult to produce the corona discharge without generating
excessive electric arcs. The use of an acoustic agglomerator,. operating
either in standing-wave or traveling-wave mode, appears very attractive.
Evaluation of Sonics for the Control of Particulate Emissions
Under the sponsorship of the F.nvironmental Protection Agency, a
report on the application of acoustic agglomeration for fine particle control
was published by Hegarty and Shannon of the Midwest Research Institute [16].
While the report represents a good review of the work done during the 1950's
it contains some questionable assumption in its evaluation of the system.
Specifically, the effectiveness of an acoustic agglomerator is evaluated on
the basis of the agglomeration tiem t , which according to Mednikov [12] can
be written as
3
t = •• In
o
a
where >
i
°!
a is the particle radius, p is the particle density. The subscript o repre-
sents the conditions at time t = 0, K is the agglomeration constant which
will be shown later to be directly proportional to the square root of the
acoustic intensity fi. In reference [!(>), however, K is assumed to he
1/2
equal to fi ; that is
329
-------
- = F(d ,d ,o ,m,f) fl and F = 1
a g c' g' ' _ (2)
where F is a complex function of the geometric mean diameter of the particle
d , the particle diameter" to be controlled d , the geometric standard devia-
6 C
tion a , the mass concentration m, and the acoustic frequency f. In refer-
&
ence [16J the value of F is assumed to be 1 at all times. The justification
of the assumption is not available in the report.
The objective of the present paper is to carry out a more realistic eval-
uation based on the theoretical model and the experimental data obtained
recently in our laboratory.[15,17-20] To do this, we write
Ka = air (al + a^ e(x^) U12U n! (3)
o
where a is the refill factor,e is the particle collection efficiency,n is the
concentration of large particles, y,0 is the relative entrainment given by
1
J ()2T 2) i
JL £ £. -L j_
where to is the angular frequency to = 2f"f. T is the particle relaxation time
defined by
T - i ^ a2 rsi
9 Cn
where C is the Cunningham correction factor, n is the dynamic viscosity and
U in Rq. (3) is the amplitude of the particle oscillating velocity
o
330
-------
where, c is the sound speed in air and pg is
the gas density. It is clear from Eq.(3) and Oq.(6) that K is proportional
a
1/2
to fi as shown in Eq.(2).
The derivations of the analytical expressions for e are given in ref-
erence [17] for both the case of inertial capture and that of hydrodynamic
collisions and arc not repeated here, only the results are ^ivcn.
For hy_drodynamic interaction we have
i - iuf'M>?r1/6- uj-n'-W-V2"!
co-7 7-, -77 T, , / ;) -I'M /a -10
•_^7 + HiZl + - :VM . _ PXI1 / c \-=- / c
a 7 ; "if,,, CXP
' e a
where T is the large particle
relaxation time. a^ is (lie "ffrriiv<- r.idiii; "I HI .1. i-u in r,|i( i \ i U-iit p,n
I i i-1 e whose i-;rav i t i ona 1 scltlini; ''Mp..-i(\- r I he ,.m< i i I,,. i\ei.i,'.. fhn.,-
.•incind velority- of (he larjM- par I n le of r.ehie. i [17]
l;or inertial capture, we have
7
e
(K ^ 0.25)'
e
where K - -
e a
i1
331
-------
Using .Eqs. (3) to (8), computer calculations have been carried out on
the variation of the particle size distribution function as a function of
time under various acoustic conditions. These theoretical results are then
compared with the experimental data based on the combined use of the elec-
tromagnetic speakers for low-frequency sound and the aerodynamic siren in
our laboratory for the high-frequency sound [18,19]. The good agreements
between the theory and our experimental data give us reasonable confidence
on the validity of our theoretical model, Nbre specifically, the hydrodynamic
interaction and the inertial capture as represented by Eqs. (7) and (8)
respectively, can be considered as the principal mechanisms responsible for
the enhanced agglomeration rate under an acoustic field. Other secondary
mechanisms, such as the parakinetic interaction, attractional interaction and
drift are considered to be primarily responsible for refilling of the con-
trolled volume every time after the particles are swept clean during each
oscillating cycle. Experimental data show that at high intensities
( > 150 db) the refill processes are very effective, Thus the refill factor
a in Eq.(3) is assumed to have a value of 1 in the present paper.
Depending on the magnitude of the interception -number I f= a^/a), the pre-
sent model determines the acoustic agglomeration constant in two regions:
(1) the hydrodynamic collision region(I ~ 1] and (2) the inertial capture
region (I «!") .
In the following sections we will attempt to make some predictions con-
cerning the agglomeration rate and the dwell time based on this model These
results will then be used to compare with the results obtained in rcf. |101.
332
-------
Determination of the Half-Life Decay Time
Consider now a polydisperse aerosol. We define u half-life decay
tiAie t ._ as the time when the aerosol concentration reduces to half of
its original value. Using the rate equation
dN(a )
we have by integration:
N(a
,
= 2
or
= In 2/K
a (11)
where N(a ) ~ * n(a) da is the particle number concentration with
2 o
radius less than a0 , N (a_) is the value of N(a9)at t = t and K is the
2 o 2. ^ o 3.
effective acoustic agglomeration constant for a log-normally distributed
aerosol. According to Eq. (4) we can write
Ka ~ a2 Vwl' ""1 (12)
where a and a? are the radii for large and small particles, respectively.
Using Eq. (3) we have
n(a ) is assumed to be a log-normal distribution function which is iusti-
fied both from our measurements and from direct measurements of the combustion
products from a boiler [21]. Based on Eqs. (11) and (13), the variation of
333
-------
the half-life decay time t ._ can be calculated once the particle mass
loading (yg/cm ), the geometric standard deviation o , the geometric mean
o
2
diameter d (vim), the acoustic intensity ft (Watt/cm ) and acoustic frequency f
are given.
Figure 1 shows some theoretical predictions based on this model.•
Herp, the fraction of the fine particles (d2=2um) remained airborne after
the acoustic field (f = 1 KHz) is turned on.at t = 0. The geometric
mean diameter of the particles, weighed with respect to number concentration,
is fixed at 0.15 urn. The size distribution function is assumed to be log-
normal with the total number concentration N and the geometric standard
5 -3
deviation o as variables. The initial values of N and a are 1.2 x 10 cm
g g
and 3.45 respectively (curves 1,2 and 3). As shown, the t_ ,„ for curve 1
(J2 = 0.01 W/cm ), curve 2 (fi = O.lW/cm ) and curve 3 (ft = IW/cm ) are
respectively 95, 30 and 9.5 sec. This is consistent with the theoretical
model as expressed in Eq,(2) which is an analytical relation for inertial
capture of particles based on Eq.(8).
The effect of reducing the standard deviation value o is demonstrated
g
by curve 4. Here a is reduced from 3.45 (for curves 1, 2 and 3) to 1.6.
O
Such a relatively monodisperse aerosol has a very small value of acoustic
agglomeration constant K and thus a very large value of t , (Eq.ll">.
3- _L / £-
The agglomeration becomes ineffective because the "Ujuber of large particles
(which act as the collecting centers of small particles) are drastically
reduced when a is decreased.
g
334
-------
The fact that the number of available large particles has a dominant
effect on agglomeration rate suggests that for practical applicationst
it may be advantageous to introduce some sort of spray system in the agglom-
eration chamber. The mean size of the spray droplets should be large in
comparison with the fine-particle size to be controlled, but should be, kept
small enough so that the droplets remain airborne for a relatively long
period of time. Also the spray liquid must be chemically inert so that no
additional air-pollution problem is introduced.
The effect of increasing particle concentration is demonstrated by curve?
5 and 6 of Fig. 1. For example, for the case when the initial conditions
2
are those shown in the figure, and fl = O.lW/cm , an increase of concentration
by a factor of 5 reduces the value of t , from 30 sec (curve 2) to 6 sec
(curve 6). In fact, it can be seen from Eqs.(ll; and (13) that
where N is the total number of large-collecting particles.
For convenience, it is useful to define a critical diameter below which
particles can be considered oscillating together with the carrier gas.
Typically, this is specified as the radius at which y = 0.8. From Eq.(2)
we have
2 -A ? 2
2.2 or d f : 4.10 cm /sec = 40 urn KHz (15)
For d^= 2um,f = 10 KHz. Thus, if an acoustic field o<~ 10 KHz is apnlied
to an aerosol, particles with diameter less than 2 pm would oscillate with
nearly full amplitude (y >_0.8) while particles with diameter larger than
335
-------
2 microns are virtually stationary (\i ~ 0) . Such differential motion of
the oscillating particles leads to particle agglomeration.
Using this critical frequency, Fig. 2 shows the variation of t ,,, for
an aerosol (d = 1 ym, N = 10 cm , and P = 1 W/cm ) with respect to the
&
standard deviation a for the removal of particles with diameters less than
2 ym, 4 ym and 10 ym. For each curve the acoustic frequency is set at the
value determined by the critical relation (Eq.15). It is seen here that the
larger the mean particle size to be removed, the higher the values of t ,„
Furthermore, the advantage of a high value of a is obvious. This, together
with the fact that K is inversely proportion to number concentration (Eq.14),
cl
makes it attractive to use the acoustic agglomerator as a preconditioner,
i.e., use it upstream ahead of a conventional device such as a cyclone or aiv
electrostatic precipitator for the best effect.
In most of the practical cases, it is common to use the mass concentra-
tion m(in yg/cm or g/m ), and not the number concentration, to indicate the
amount of particulate emission. Thus, instead of keeping number concentration
N constant, m is maintained constant in Fig. 3. Here t . , reduces initially as
o increases, reaches a minimum, and eventually increase rapidly. This increase
t~*
is due to the fact that a fixed mass concentration, the total number of part-
icles reduces as 0 increases, leads to a reduction in K as shown in Eq.(141.
Figure 4 shows the same calculations but is presented with t. /9 vs.
the mass concentration for various values of a for fine-particle control
&
(d.., = 2 ym) . For a given value of the mass concentration, the optimum value
of a is again shown. The drastic reduction in the required agglomeration
&
time, t ._ for increasing value of the mass concentratin suggests again thit
the device should be used as a preconditioner.
336
-------
Comparative Energy Requirements of the Acoustic Agglomerator and Other
Conventional Control Systems
It is useful to compare the specific energy requirement for the oper-
ation of an acoustic agglomerator to other devices for the control of par-
ticulate emissions. The procedure used is similar to that used in refer-
ence [16] which is based on Mednikov's work [12]. The specific energy re-
quirement for acoustic agglomeration is given by
ft t
E = ° (16)
0.36 e e e M
s c uf
where E is the specific energy in Kw-hr/1000 m' , fi is the sound intensity
2
in W/cm , t is the agglomeration time in sec, e is the acoustic efficiency
of the sound source, e is the overall efficiency of the compressor for the
sound source and e ,, is the utilization factor of the ap",l operation chamber.
uf
H is the height of the agglomeration chamber and can be estimated by requir-
ing that the aerosols spend an average time t in the agglomerator. Thus
we have
H = 21.2 tQ/D2 (17)
where D is the diameter of the agglomerator. Substituting l;q.(17) in Eq.
(16), we have
E = °'13 "°2 (18)
es ec euf
Equation (18) which was used in reference [16] for the evaluation of the
acoustic agglomerators, is also used here. However, instead of using Eqs,
(1) and (2)ft is estimated based on the values shown in Figs. 2 and 4.
337
-------
Following reference [16] we assume that e = 0.5, e =0.65 and e =0.7.
s c uf
The first two rows in Table 1 are the result of the present estimation
of the specific energy requirements and the size for acoustic agglomerators
under traveling-wave conditions, t (in column V) is estimated from t,/?
(in column IV) by assuming a 99% removal, i.e.,
By comparing Eq.(19) with Eq.(ll), we have
t = t = 6 6^ t f?0~l
o 1/100 1/2 '*• '
where t1/100 is the agglomeration time for the reduction of the particle
concentration to 1% of its original value.
It should be noted that in all calculations in Table 1 the aerosol
size distribution function is assumed to be constant, i.e., it is lognormally
distributed with d = 1 urn and a = 2.0. Consequently, t. ,_ is inversely
proportional to the mass loading, m according to Eq.(14). The particles to
be controlled have diameters less than 2 urn (d = 2 urn). For other values
of d similar tables may be constructed. These particular valuesof d (2um)
are picked because it is a reasonable value for applications in the pressur-
ized fluidized-bed boilers, coal gasification plants and MMD solid-seed
separators.
It should be noted that the value of K (column VI), as determined by
cl
1/2
Eq.flS;, is not equal to £7 as was assumed in reference |lb]. In fact,
K increases rapidly as the mass 'loading increases.
n
338
-------
Another interest ing point is that the value of E is independent of t
as shown in Eq.(18). An increase in the effectiveness of acoustic agglom-
eration leads only to a more compact agglomerator (the value of H reduces).
At 160 db and when the mass loading is low (m < 0.5 g/iif ) , the length of
the agglomeration chamber is more than 15 m even if a larger chamber diam-
eter is used ( D = 4m) However, at higher mass loadings, the values of H
become more reasonable; again indicating the advantage of using acoustic
agglomerators as preconditioners under traveling-wave conditions. When Q
is decreased to 150 db, E is reduced by a factor of 10 and the value of H
increases by a factor of about 3. But for most industrial applications,
the mass loading is less than 2 g/m , and the value of 11 is considered to be
unacceptably high.
Standing-wave Acoustic Agglomerators
In a recent paper, the acoustic agglomeration of aerosol particles have
been reported under standing-wave conditions [20]. The half-life decay time,
as determined by Eq.(ll), has been measured for acoustic frequencies ranging
from 640 H?. to 1.7 KHz under both the stationary condition and the flow-
through condition. For the stationary system Fig. 5 shows the vari-
ation of t , as a function of the acoustic intensity at three frequencies.
For agglomeration under the standing-wave condition,
it is found that acoustic turbulence plays a critical role in the agglom-
eration process. At low intensities, a small flow velocity which
reduces the degree of acoustic turbulence in the agglomeration chamber,thus
decreases the effectiveness of the system. But, when the acoustic intensity
at the node-points is larger than 175 db (corresponding to open air inten-
sity of about 140 db) the turbulent eddy velocities of the particles lie in
339
-------
the range between 70 to 110 cm/sec. Thus according to our experimental
data, the acoustic agglomeration rate would not be reduced significantly
by the flow velocity as long as the flow velocity is less than the particle
eddy velocity.
In Table 1 two open-air intensities are used: 140 and 150 db. By extra-
polating the present data, we obtain the values of t ,« as shown in the table.
It is seen that a general reduction of the agglomeration time t is obtained
o
by changing from a traveling-wave system to a standing-wave system. For a
given intensity, the specific energy consumption is the same for either a
traveling-wave or a standing-wave system, the improvement in the required
agglomerating time would reduce the agglomerator length II. The compactness
of the agglomerator is of critical importance since very long chambers are
undesirable from the standpoint of increasing acoustic attenutation losses
in the chamber.
DISCUSSIONS AND CONCLUSIONS
In order to evaluate acoustic agglomeration for particulate-emission
control we have estimated the specific energy consumption and the size of
the agglomerator as shown in Table 1. The calculations have been made for
a set of conditions similar to those used in reference [16] for easy com-
parison of the results. Specifically, it is assumed that:
® Gas flow rate = 1000 m /min
» E =0.65, e =0.5 and e _ = 0.7
c s uf
• Cylindrical chamber with diameter = 2 and 4 in.
340
-------
The major differences between our model and the Mednikov model used in
reference [16] are as follows:
(1) The only parameter in Mednikov's model is the growth ratio df/d.
(final diameter/initial diameter) which is varied in reference [16] between
2 and 20. For practical application df/d. seldon exceeds the value of 10.
All estimates made in the present paper assume df/d. = 2; i.e.,d = 2 urn,
d.
(2) The agglomeration time for a given intensity depends only on df/d.
in reference [16]. Our model shows it is a function of df/d. the geometric
standard deviation o" , the mass loading m and the acoustic frequency f.
&
Assuming the optimum values of the acoustic frequency and the aerosol stan-
dard deviation are used, the agglomeration time t then depends only on the
mass loading for fixed values of d,-/d. and the acoustic intensity as shown
in Table 1.
(3) For a specific va-lue of acoustic intensity, K is constant in refer-
a
ence [16] -.ccording t,o Eq.(2). For traveling-wave and at Q = 160 db.
K = 1 sec when m ~ 1.6 g/m . Since this is not an unreasonable nunber for
mass loading, the major conclusions on the specific energy consumption and
the size of reference [16] are in the. same order of magnitude as those
reached in the present paper.
(4) Thus for traveling-wave, the use of Eq.(2) is close to the optimum
performance of an acoustic agglomerator for moderate growth fd./d. = 2),
medium mass loading ( m ~ 1.6 g/m ) and high intensity (0, = 160 db). However,
the estimates of reference [16] arc overly conservative when the mass loading
341
-------
is higher than 1 . (> g/m" .
It is interesting to compare the estimates for the traveling-wave sys-
tem to those of the standing-wave system. At 150 db, when the system is
changing from a traveling-wave to a standing-wave, the agglomeration time
is increased by a factor 3 to 4. At 150 db open-air intensity,the node-
point intensity is measured at 180 db. It is interesting to note that for
standing-wave, the agglomeration mechanisms are no longer those for the
traveling-wave systems in which the inertial capture and the hydrodynamic
interaction are the dominant mechanisms. As shown in Eq. (2) for traveling-
1/2
wave, K increases in proportion to Q . This relationship is obviously not
3.
valid when switching from traveling-wave to standing-wave in which turbulence
plays a critical part in particle agglomeration.
Typical energy requirements for conventional systems are given in refer-
ence [16] as follows:
• Cyclones - 0.5 to 5 hp/1000 cfm
• Fabric filters - 2 to 3 hp/1000 cfm
• Electrostatic Precipitators - 1 to 1.5 hp/1000 cfm
• Venturi Scrubbers - 10 to 20 hp/1000 cfm
Based on Table 1 the energy consumption for standing-wave acoustic agglo-
merators is estimated to be in the range between 0.5 and 2 hp/1000 cfm, which
is about the same as those for the cyclone, the fabric filter and the electro-
static precipitators and is less than that for the venturi scrubbers.
In conclusion, it is important to point out that caution must be exer-
cized in the use of the results in Table 1. First, the estimates are made on
specific assumptions such as d,./d.=2 and the standard deviation of the aerosol
size distribution function is optimum. The latter assumption implies that
342
-------
the aerosol is properly conditioned for maximum acoustic effectiveness.
Secondly, the results have been tested onJy in the low-intensity region
(less than 150 db for traveling-wave and an open-air intensity less than
130 db for standing-wave). While there are reasons to believe that the
extrapolation should be at least semi-quantitatively-correct, experimental
verifications of the findings at high intensities are urgently needed for
the further development of the acoustic agglomerators. Finally, as was done
both in reference [16] and in Table 1 in the present paper, the agglomera-
tion time t estimated for each case, is determined on the assumption that
K remains constant with respect to time. This, however, is not the case.
3.
During the early phase of acoustic agglomeration, K is high because of
a
the high mass loading. As soon as acoustic precipitation starts, K
3.
decreases accordingly. Thus the actual agglomeration time required for
99% removal is higher than those estimated in Table 1.
The last point- i.e., the advantage of high mass loadings on accoust it-
agglomeration- makes it attractive to use acoustic agglomeration in a hybrid
system. For example, in the pressurized fluidized-bed boiler, the standing-
wave acoustic agglomerator can be followed by a high-pressure cyclone which
removes the relatively large residual particles from the acoustic agglomer-
ator.
The application of acoustic agglomeration for the removal of corrosive
particles at high temperatures and pressures has attracted great interest in
recent years partially because of the limitations encountered with various
conventional devices under these unconventional environments. At high
pressures the gas medium density increases and thus the particle velocity in
the acoustic field decreases. Thus reduction in particle velocities would
343
-------
lead to a lower acoustic agglomeration rate. Therefore, to compensate
for these effects, the acoustic intensity must be increased at high pres-
sures. Acoustic agglomeration of aerosols has been reported by Mednikov
in the early 1960's at pressures as high as 50 atm [12].
The effects of an increasing temperature on acoustic agglomeration
is less certain. Mednikov [12] stated that the only effect of the temper-
ature is to increase the viscosity which also leads to a decrease in the
particle oscillating velocity, thus it is expected that the acoustic agglo-
meration rate would decrease as the temperature increases. But Belenkii,
Timoshenko and Fedorak [22] recently have reported an increase in acoustic
agglomeration rate as the temperature is increased. No explanation has been
given for the increase.
Given the uncertainties of the effects of temperature and pressure
on acoustic agglomeration, the important fact is that there does not exist
any fundamental reason for the acoustic agglomerators not to perform effec-
tively at high temperatures and pressures. To compensate for the effects
of high temperature and pressure the standing-wave agglomerator may have
to increase its open-air intensity from the 150 db in Table 1 to 160 db.
Questions concerning the economic competitiveness of such an increase in
the specific energy consumption cannot be fully answered until more experi-
mental data are obtained in this area.
344
-------
References
1. H.S. Patterson and W. Cawood, "Phenomena in a Sounding Tube," Nature, 127,
067 (1931).
2. 0. Brandt and E, Hidemann,"The Aggregation of Suspended Particles in Gases by
Sonic and Supersonic Waves," Trans.Faraday Soc. 3^ (184):1101-1110 (1936).
3. E.N. da C. Andrade, "The Coagulation of Smoke by Supersonic Vibrations,"
Trans.Faraday Soc. 32_ (184) 30-35 (1936).
4. H.W. St.Clair, "Agglomeration of Smoke, Fog, or Dust Particles by Sonic Waves,1'
Ind. 6 Eng. Chem., 4_1_, 2434 (1949).
5. D. Sinclair, "Coagulation by Sonic and Supersonic Vibrations," Handbook of
Aerosols, Washington, D.C 1950.
6. J. Krebs and R.C. Binder, "Use of Ultrasonic Coagulation with a Cyclone Separ-
ator," Combustion, 23, 12, 45 (1952).
7. H.W. Danser and E.P. Neumann, "Industrial Sonic Agglomeration of Collection
Systems," Ind.fi Eng.Chem., 41_, 2439 (1949).
8. E.P. Neumann and J.L. Norton, "Application of Sonic Energy to Commercial
Aerosol Collection Problems," Chem.Eng.Progr., Symposium- Sec.1 , 47 (1):4-10
(1951).
9. C.R. Soderberg, Jr.."Industrial Applications of Sonic Energy," Iron Steel Engr.
29_, (2): 87-94 (1952).
10. M. Nord, "Sonic Precipitation of Smoke, Fumes and Dust Particles," Chem. Engr.,
pp. 116-119, October 1950.
II. C.A. Stokes and J.E. Vivian, "Application of Sonic Energy in the Process Ind-
ustries," Chem. Engr. Prog. Symp. Series 1, 47_ (1): 11-21 (1951).
12. E.P. Mednikov, Acoustic Coagulation and Precipitation of Aerosols, Translated
from Russian by C.V. Larrick, Consultants Bureau, New York 1965.
345
-------
13. N.L. Shirokova, "Aerosol Coagulation," Physica1 Principles of Ultrasonic
Technology Vol. 2, ed. L.[). Rozenberg, Trans, from Russian by J.S. Wood,
Plenum Press, New York 1970, p 477-541.
14. A.A. Jonke, W.M. Swift and G.J. Vogel, "Fluidized-Bed Combustion Development
and Status," TransTASME, 258, 159 (1975).
15. D.T. Shaw and J. Wegrzyn, "New Applications of Acoustic Agglomerators 'in
Particulate Emission Control," Research Workshop on Novel Concepts, Methods
and Advanced Technology in Particulate-Gas Separation, Univ. of Notre Dame,
April 1977.
16. R. Hegarty and L.J. Shannon, "Evaluation of Sonics for Fine Particle Control,"
Report 600/2-76-001 EPA, Washington, D.C. January 1976.
17. D.T. Shaw and N. Rajendran, "Application of Acoustic Agglomerators for
Emergency Use In LMFBR Plants," submitted for publication to Nuclear Science
and Engineering, November 1977.
18. D.T. Shaw and K.W. Tu, "Acoustic Particle Agglomeration Due to Hydrodynamic
Interaction Between Monodisperse Aerosols," submitted for publication to the
Journal of Aerosol Science, November 1977.
19. D.T. Shaw, S. Patel and N. Rajendran, "The Removal of Airborn Dust Particles
in an Acoustic Field Under Traveling Wave Conditions," submitted to the Journal
of the Acoustic Society of America, November 1977.
20. N. Rajendran, J. Wegrzyn and D.T. Shaw, "Acoustic Precipitation of Aerosol
under Standing-Wave Condition," submitted to the Journal of Aerosol Science,
November 1977.
21. L.J. Shannon, P.G. Gorman and M. Reichel, "Particulate Pollutant System Study,"
Vol. Ill - Fine Particle Emission, Midwest Research Institute, August 1971,
PB-203-521.
22. V.A. Belenkii,V.I. TLmoshenko and T. Ya.Fedorak, "Choosing the Optimum Condi-
tions for the Acoustic Coagulation of Aerosols," Soviet Phys-Acoustics, 19,
2, 181 (1973).
346
-------
: u'-
J 1 u
c~ c
•£ o
o
i ;
x3 UJ
to
! ^
! 1 i
^o 1 ^
•y t- ; •
•~ 0 u
^ II ; i
ii
o , ;
d i >• Q
c LO : ;
J ! .'
j f
! * _T"
•/) ii < i
^ ' ' >j
§ uu ' > S
§ - ! 1 %
'•* O i ; ?^
H • I '
g. ^ f i
D 11 ', '
X D j >j in
S 6 j ^°
CN] ; T
•-> j ^
•HII » (j
;J ™ : 1 ^
1) >• r-i
° i ! — '
6 -M
V-* • — I
S "M ^ "^
•s> -d I-H M
" ' ^
r i
— ' i u
_Y ' k ( 1 ,
r ) r i r i r-i
LO ' — 1 LO i — ' LO i— 1 I-O i — 1
r- 1 r i r- 1 r-i
oj cn <~ \ en r i cr> r i cr>
• -
f to IO to to to t" 1"
LO ^-1 LO •— ' LO • — ILO*— <
-C CTi tO O CO 1^1 ^f i — 1
^H r-- o — i
C-] •'Tf (Nl ^T OJ -"t C~] -"t
t-~ LO r-i
r^ co co to
O t~O O r-<
O C to ~D
cn LO
r- a~i ^ r—
en i — ^ *-H o
LO ^H
LO to
O r-i —t
CO r i —i
O~:
r-H O O
LO
r-j r j
— l O O O
o r- 10 o
LO —1 LO ^H '° i-H '°
O cxi o c-i ° r\i O
r-j cn r j cn r i en r i
O O O O C O p
to to to to to f to
LO — i LO i — i to i — I I o
oc r-- o c: LO \o r i
ci ~i o LO i-g .— i
cri r-i r i
r 1 '-t n -t r i ^ r i
ri r l t - r i
O — I CTi C")
to co r-- to
co r-- *t <^i
co to
r-H
— i 0
•^r r- r-- to
CO 1O O O
r ]
10
ft f- 1
— i -D C C-
o o ,0 -
o o
\O LO
347
-------
1
o
QJ <— •,
E r-
•H
H 0
C II
o
+-> 2
ri OJ
O 13
e c
o ri
i — i
CJ) LO
M
f O
OJ II
XI
OJ
"O
d
ri uo
vO
LU
c
'-0
tJ II
c
01 (J
6 OJ
•H O
cr 01
OJ
X bo
w> D
5-4
cu
d 6
60
!/i -d
1 1 J ^ — '
J*
LU +->|
J C
=Q 01
< ul
X
X
1 — 1
J — 1
1 — I
I-H
^>
t-l
t-H
t— 1
^>
^>
|^>
1— H
, .-
1 — *
h-H
1 — )
KH
i — |
i — |
?*
u
C- ^
-C 0
o
. — 1
m
to__
5-i ^
Si O
^ O
o
,
CJ
t/1
^^
^
0
p i
e t
i 1
n:
, .
c
i — , \
Q
i — *
i — 4
1
rj
D
t/)
i — i
oj
i — i
O
d)
(/l
i — i
O
^_)
u
0>
t/1
1 1
0-1
~~^
,—4
-P
, .
1 1^
s
^^-^
b-J
4 1
e
ri
E
o
^-^
^
.
LO i-H l-O 1 — 1 LO --H LO -— <
O ri o ro o oj c r-j
CM Ot r\i Ol r ] CTi r j 31
ooooooco
to to >o to to to to rr>
>.••••..
LO .—1 l-O i-H LO i— < LO r— 1
\O xc co r^-
d i-~ •— i LO \D i—1 ri o
vc •£; n
OJ
r\i -3- CN rt oj ^ oi ^t
(j\ \O ^O xO
O i — 1 Tf MD
O ^H tO OO
oo
CTl CT^ \O to
Ol LO
O to
LO r— 1 C
oo a> oo
^O ^O r— * O
I-- O O O
LO
OJ
i-H vO
O ^H IO O
1 — 1
o
LO
1 — 1
n A V M - D N
!
LO LO LO LO
O r i o o,j o 01 o ' J
• •••.•,•
OOOOoOcO'
K1 i — i to , i ro • — i to — i
oi CTI r.i cr> 01 cn rs ai
o o c o o o o o
.
C O C O C C'O C
to to to to to to to to
LO.— ILOr-ILO,— ILO— 1
LO — i tO vC O LO O to
O C \D ^H Ol i — '
co r i
r\i ^ 01 TJ- rj ^t rj -t
IO OO r— t — 1
O to OJ -t
O C — ( 01
CT> CO d
OJ •— 1 IO i— (
LO i — 1
, — 1
\D CO
r- co
O OO LO 01
tO •-! O O
OJ
i-O
0]
.-i xD O
o •— i 10 o
• — 1
o
*T{-
r—H
I (1 N V 1 S
348
-------
OJ
E
•H
d
o
•rH
4-1
O
C
3
ca
0)
•H
IT)
M
C
•H
C
•H
10
E
01
E
3
Vl
n)
T3
to
U)
01
r>
E
;3
in
t—i
Q
II
e
o
x
CM
3
O
m
a.1
3
U
vC
0)
3
U
OJ
3NIJ dO
349
-------
1000
FIG. 2
The half-life decay time t, ,_ vs. the standard deviation a for three critical
1/2 8 6 -3
diameters (d =2 Pm, 4 ym, 10 )Jm) , number concentration N = 10 cm , geometr:
C 2
mean diameter d = 1.0 |im, ft = 1 W/cm , critical frequency as determined by
o
Eq. 52 is used for all three values OJL d .
M c
350
-------
100
LJJ
H
10
1,0
2,0
3,0
FIG. 3 - Half-life decay time
vs. the standard deviation o for
a fixed mass concentration (M = O.f,2 Ug/cm ), d? =2 urn, 4 pm, 10 urn,
n = 1 W/cm , d = 1.0pm, critical frequency determined by Eq . 52 is
o
used for all three curves.
351
-------
10
CO
1
0
,25
,5
,75
1,
(y cm )
1,25
1,5
FLG.
. 4- Half-life decay time t^, vs. mass concentration M for five values of the
standard deviation (o = 1.5, 2.0, 2.5, 3.0, 3.5), d., -- 2 pm, H = ] W/cm2,
O t-
d = 1.0 ym, critical radius determined by Eq.52 is used.
O
352
-------
STANDING-WAVE NODAL-POINT INTENSITY (db)
100 110
OPE;I-AIR INTENSITY w
Fig. 5 Half-life decay time vs. intensity for three frequencies in
vertical position with driver at the bottom.
Q- 640 Hz, ^- 85(1 Hz, O-1070 Hz
353
120
-------
CYCLOCENTRIFUGE DEVELOPMENT FOR
PARTICULATE, PHASE I: FEASIBILITY STUDY
By:
J. T. McCabe
Mechanical Technology Incorporated
Latham, NY 12110
355
-------
ABSTRACT
Gas cleanup was determined to be an area in which the special characteristics
of a modified centrifuge offered technical and economic advantages over exist-
ing approaches. A new concept, called a Cyclocentrifuge, was evolved during
an analytical study in which the desirable characteristics of a cyclone and a
centrifuge were combined in a compact design capable of separating fine part-
iculate matter from hot pressurized gas at large flow rates on a continuous
basis. A design example is given which shows the Cyclocentrifuge to be capable
of achieving a purity of 1 ppm of solids with a nominal maximum particle diameter
of one micron when processing 125,000 scfm of low Btu fuel from a coal gasifier.
356
-------
1. INTRODUCTION
Existing techniques for removing fine particulate matter from coal conversion
gases are not considered sufficiently effective or efficient for meeting many
current and projected application demands. The fundamental question addressed
in this study was to determine if gas centrifuges technology could provide the
bases for technical or economic improvements in the fields of gas particulate
cleanup and other related processes. The initial study of the general application
of gas centrifuge technology quickly focused on the special field of particulate
control, which is the subject of this paper. This field is important because
it bears directly upon the efficiency with which the overall coal conversion
process can be carried out, and the efficiency of the conversion process will
determine the extent to which the vast coal reserves that are available will be
exploited.
Cleanup of gasifier-produced low-Btu fuel for direct firing in the gas turbine
of power generating system was selected as a representative study application
because failure to remove fine particulate matter from the fuel gas can cause
severe and expensive erosion and corrosion. In addition to cyclones, there are
presently three approaches to resolving the gas cleanup problem for this appli-
cation: Filters, electrostatic precipitators, and scrubbers. Filters require
large surface areas. Carry-over of filter material is often equally as bad, or
worse, than by-passing the original contaminant. Filters must also be taken
off-line to be purged, and clogging is a problem because it is difficult to re-
move fine particles and tar from a high efficiency bed.
Electrostatic precipitators are bulky and must be contained in large pressure
vessels. They must be shut down to be cleaned, they consume considerable elec-
trical power, they are inefficient when the gas is hot, they are ineffective
with certain types of gas and process upsets can result in carry-over and some-
times fires.
Liquid scrubbing systems are effective in removing fine particulates; however,
corrosion remains a problem, and the quantity of energy lost in the gas due to
cooling during the scrubbing process is very significant. In addition, disposal
of the blow-down water and the need for sludge ponds present the added expense
of resolving the accompanying pollution control problem.
357
-------
A unique and viable alternative to existing methods for gas cleanup was evolved
during the study called a Cyclocentrifuge. The principal difference between the
Cyclocentrifuge and a simple cyclone is the magnitude and method of generating
the swirl velocity. Cyclones are limited in the magnitude of the swirl velocity,
because, as the swirl velocity is increased beyond about 100 feet per second,
secondary effects, such as the strength of eddies shed from the expanding inlet
nozzle jet, eventually become equivalent in strength to the vortex core. This
results in mixing rather than separation. The swirl velocity limit for the
Cyclocentrifuge is expected to be six to seven times greater than a cyclone
because the gas is accelerated gradually by a cascade of swirl augmentation
blades that are designed to minimize the generation of mixing eddies.
Feasibility for the Cyclocentrifuge is established by presenting a design ex-
ample which includes design analyses, conceptual drawings, and cost data for a
Cyclocentrifuge based on delivering 125,000 scfm of high-purity, low-Btu gas
at 250 psi and 1000°F.
The application chosen for this study was typical of those used in a combined-
cycle electrical power system based on the gasification of coal. The arrangement
is shown schematically in Figure 1, where the key gas cleanup element is a
Cyclocentrifuge. This schematic arrangement is similar to that given in Reference
1, which presents the economic analysis of a combined cycle. The economic data
in Reference 1 was also used as a Baseline for the purposes of this study. The
major differences between Figure 1 and the system analyzed in Reference 1 is
that a steam generator has been inserted in the gasifier to limit the gas stream
temperature entering the Cyclocentrifuge and thereby obtain satisfactory long-
term rupture and creep properties in the centrifuge shell. It may be feasible
to operate at a higher gas stream temperature by using the externally cooled
bearing lubricant or process gas to control the centrifuge shell temperature;
however, shell cooling schemes were not considered in this study.
358
-------
2. PERFORMANCE GOALS
Reference 2 indicates that turbine blade erosion will be avoided if the fuel
gas contains less than 30 ppm of solids by weight, and the maximum equivalent
spherical particle diameter is less than 10 microns. There is no equivalent
specification for limiting the number and size of solid particles that contain
corrosion or deposit-causing contaminants. The most troublesome trace contam-
inants are sodium, potassium, lead, vanadium, calcium, and sulfur (References
2, 3, and 4), and the most corrosive compounds are sodium sulfate and vanadium
pentoxide which form during the combusion process and damage the hot gas path
components. However, in addition to appearing as solids, some of the contam-
inants can be in the gaseous or liquid state. In order to establish design
goals for the Cyclocentrifuge so that gas cleanup also includes the elimination
fine solids containing corrosion and deposit producing elements, the following
assumptions are made:
• All of the corrosion and deposit-causing contaminants are contained
in solid particles that have the weight density of ash.
• The gas purity requirement for corrosion and deposit protection
of hot turbine blades is the sum of the allowable trace metal
contaminants.
The second assumption leads to a total purity requirement that depends on the
air/fuel ratio and limits the present cases to approximately 1 ppm of solids by
weight according to recient G.E. gas turbine specifications. Reference 5 shows
typical probability curves, which assume that all particles are less wher the
maximum particle size measurable was limited to 80 microns, for particle size
ranges, and calculations contained in Reference 12 show that the theoretical cut
for the Cyclocentrifuge must be approximately 1 micron to achieve a purity of 1 ppm.
Based on Reference 1, Figure 1, and the above discussion, the design requirements
were taken as follows:
• Volume flow from coal gasifier:
• Inlet pressure:
• Inlet temperature:
• Purity:
• Nominal particle size (cut):
359
125,260 scfm
250 psi
1000°F
1 ppm total solids by weight
1 micron
-------
3. GENERAL ARRANGEMENT OF THE CYCLOCENTRIFUGE
Figure 2 shows that the general arrangement of the Cyclocentrifuge consists of
two major components: a rotating assembly called the "centrifuge" and a sta-
tionary assembly called the "cyclone". The cyclone is both a pressure vessel
and a containment vessel for the rotating centrifuge.
Particulate matter is separated outside of the centrifuge shell. The high
centrifugal field, which is the basis for separation, is generated by the
cyclone swirl velocity caused by the tangentially located inlet duct and,
primarily, by the energy imparted to the gas by the swirl augmentation blade
arrangement that is fixed to the centrifuge.
The rotating gas forms a vortex which originates in the dust collecting vessel
below the cyclone cone. At the bottom of the conical section, the layer of
gas at the.cyclone wall is subjected to an intense centrifugal field because
the angular velocity of the gas increases as the radius decreases in accordance
with the law of conservation of angular momentum. The gas layer at the wall,
which contains the centrifuged particulate matter, is bled-off into the collec-
tion vessel to complete the separation; the remainder of gas feeds the rising
vortex core with clean gas." Re-entrainment is prevented by diffusing the bleed
gas in the expanding cross-section of the dust collected vessel where the
particulate matter settles to the bottom of the vessels. In leaving the
collection vessel, the particles are again fluidized and carried through a
duct to the suspensions pump.
The centrifuge drive power is extracted from the process gas in the form of a
pressure drop across the built in reaction turbine. Figure 3 shows the rotating
assembly, which contains the following design features:
• The swirl augmentation blades are arranged to gradually accelerate
the gas to achieve the centrifugal force field necessary to separate
fine particulate matter without exceeding the maximum throughput
velocity for conventional cyclone design.
• The exit vanes are zero-lift structural support struts that carry the
weight of the rotating assembly.
360
-------
• The upper bearing rotates; the lower bearing is stationary.
• The drive turbine is located at the centrifuge entrance.
• The structural stiffness-to-weight ratio is maximized to place the
first critical speed 25 percent above the operating speed.
Figure 4 shows the upper bearing and shaft assembly. The important features
are:
• Both the journal and thrust bearings are externally pressurized
hyrostatic bearings. The bearings are essentially optimized with
respect to stiffness and consist of a number of deep pockets sur-
rounded by "lands", all of which are vented to the return line.
• Pressure for forcing the lubricant through the return lines is
provided by the process gas.
• The weight of the centrifuge assembly is carried by the upper
surface of the thrust bearing; the lower surface of the thrust
bearer provides the additional stiffness necessary to control
the axial critical frequency.
• The stator guide vanes are canted to direct the flow towards the
exit duct.
• A gas blow-back seal, shown in View A, overcomes the outside-
to-inside pressure gradient and prevents dirty gas from mixing
with clean gas.
• Dust protection for the bearings is provided by a cover surrounding
the upper shaft support and by the labyrinth seal at the top of
the bearing. Pressure seals are not required.
• The upper support blades, which are also exit vanes, are twisted
to a zero-lift configuration.
361
-------
Figure 5 shows the lower bearing and shaft assembly. The important features
are:
• The journal bearing is similar to the upper journal bearing.
• Bearing lubricant is supplied and returned through passages in
the lower bearing support struts. Dust protection is provided
by the labyrinth seal. A pressure seal is not required.
• Power to drive the centrifuge at 5500 rpm comes from a turbine.
The turbine extracts energy from the cyclone swirl component;
however, the primary source of energy is the process gas.
• The lower bearing support struts are twisted clockwise to an
average zero-life angle for the downward flow outside the
centrifuge and then twisted counter-clockwise for the upward
flow entering the centrifuge drive turbine. Transition occurs
at the lower bearing support ring.
• The outboard struts are attached to the support ring in a "semi-
bicycle-spoke" arrangement to accomodate differential expansion
of the struts (see Section AA, Figure 2).
• By virtue of the relative moment stiffness between the strut
assembly and the journal bearing, the strut assembly forms an
elastic gimbal to accommodate angular misalignment of the lower
bearing.
• The lower bearing and mounting assembly is designed to align
to the upper assembly in order to accommodate slow, thermally
induced dimensional changes in the mounting members or support
points.
Figure 6 shows the lubrication schematic which is typical in design for most
similar applications.
362
-------
4. AERODYNAMIC DESIGN
The Aerodynamic analyses were performed iteratively with stress and rotor
dynamic studies to insure that stress limits were not exceeded and that the
first critical speed was at least 25 percent higher than the design speed.
The final compromise in this process was to reduce the operating speed from
5370 rpm to 5500 rpm. This change of 4 percent, which was made to satisfy
rotor dynamic considerations, is not reflected in the subsequent sections
because the effect is too insignificant from an aerodynamic view-point to
warrant another iteration.
A. Principal Dimensions
After consideration of flow areas and structural stiffness requirements, the
centrifuge OD was set at 26 inches.
The centrifuge design speed was determined on the basis of preliminary design
sketches and a maximum allowable stress of 80,000 psi. This stress limitation
is based on the material In 718 at 1000°F. The design speed was set at 5730
rpm, which is equivalent to a centrifuge surface speed of 650 ft/sec.
Based on Reference 6, the throughput velocity between the cyclone and centrifuge
was taken as 24 ft/sec and the required inside diameter of the cylindrical sec-
tion of the cyclone was calculated to be 57 inches.
The axial length of the Cyclocentrifuge was determined by calculating the axial
distance traveled by a 1 micron particle in its spiral path from the wall of
the centrifuge to the wall of the cyclone.
In keeping with Reference 6, the axial length of the cyclindrical section below
the centrifuge inlet was taken equal to one diameter, and the length of the
conical section as 2.5 diameters.
The required distance from the top of the inlet duct to the bottom of the
centrifuge was determined from preliminary design of the inlet duct and swirl
augmentation blade section to be 80 in.
The overall Cyclocentrifuge height also includes the distance from the top of
the inlet duct to the top of the vessel head, plus the length of the particulate
363
-------
discharge flange, plus the height of the particulate collection vessel. These
lengths are taken as 20 inches, 6 inches, and 6'-4 1/4", respectively, making
the overall length of the Cyclocentrifuge vessel was 30'-11 1/2".
B. Separation Efficiency
There is a critical size of particle of given density for which the centrifugal
and drag force exactly balance so that the particle moves neither outward nor
inward. All particles larger than the critical size will be collected; all
smaller will escape. The critical particle size is called the theoretical cut.
The theoretical cut for the Cyclocentrifuge was determined in accordance with
Reference 7, and is calculated to be 1,08 microns for the present case.
Figure 7 shows the predicted grade efficiency for the Cyclocentrifuge and for
a conventional, high efficiency cyclone. This curve anticipates the following:
• Compared to an ordinary cyclone, more particles larger than cne
Cyclocentrifuge cut size will be prevented from entering the exit
duct because:
• The distance from the inlet duct to the centrifuge
entrance is greater than the equivalent distance
in a cyclone.
• The cycloncentrifuge "exit duct" is the spinning
centrifuge which will reject particles that bounce
off the cyclone wall.
• The re-entrainment potential is reduced because the
outward "g" field is larger than a cyclone.
« Compared to a cyclone, more particles smaller than the cut size
will be separated because collisions with larger particles will
increase and smaller size aggregates of fine particles will be
removed by the increased centrifuged field.
C. Swirl Augmentation Blades
The swirl augmentation blades are designed with a constant cross-section and
no individual camber. The blades are arranged in cascade such that the effec-
tive camber of the cascade is 70 degrees, as shown in Figure 8, the required
364
-------
gas deflection of 88 degrees is accomplished using 11 blade rows occupying an
axial length of 45 inches. There are 16 individual blades spaced uniformly in
each of the 11 rows.
Figure 9 shows the velocity triangles which are the bases for the blade design.
The tangential velocity at the hub is seen to be 650.6 ft/sec, while the axial
component is 24.08 ft/sec and the entering gas velocity, due to cyclone swirl,
is 150 ft/sec. At the blade tip, the tangential velocity is seen to be 850.8
ft/sec with the axial and entering velocity components being identical to hub
section conditions. The lower velocity triangle sketch in Figure 9 shows that
the swirl velocity of the gas is increased uniformly through the 11 rows of
blades, without changing the axial velocity.
D. Centriguge Drive Turbine
The centrifuge is rotated by a high-reaction axial turbine driven by energy
extracted from the process gas. The exit gas velocity vector is purely axial
and so that the gas passing through the centrifuge will not tend to centrifuge
particles, and solids buildup on the inside wall of the centrifuge will be
minimized.
The whirl component from the cyclone vortex is used as an energy source for
the turbine, and inlet nozzle guide vanes are not used. The remainder of the
drive power is produced by a pressure drop in the process gas across the rotor
blades.
The power required to be delivered by the turbine can be approximated by
P = WaU (AUQ)
where Wa is the gas mass flow through the annular area, Aa, based on the
centrifuge diameter and the diameter at the tip of the swirl
augmentation blades.
U is the gas tangential velocity at the centrifuge OD = 650.9 ft/sec
from Figure 9.
365
-------
AU is the increase in tangentail velocity due to the swirl augmentation
blades = 500.6 ft/sec from Figure 9.
Using the above data, the required horsepower is approximately 500.
Figure 10 shows the blade cross-sections at the root, mid-cord, and tip, and
the pressure drop across the turbine is calculated to be 6.1 psi.
2
The turbine inlet whirl distribution is obtained from the assumption U R - constant
as suggested in Reference 6. Figure 10 incicates that most of the turbine work
output is produced in the hub to mid-cord region where the prevailing whirl velocity
is of significant magnitude. Thus, the blades are straight near the tip and
highly twisted at the hub. The blade solidity for the cascade is based on avoid-
ing choking everywhere and on a Zweifel coefficient of approximately 1.1 at the
hub. The incidence is taken as zero at the design point, and the deviation angle
is based on the recommendation given by Horlock in Reference 9.
E. Total Pressure Drop
The total pressure drop through the Cyclocentrifuge Ap is
where Ap = drop across the drive turbine =6.1 psi
In accordance with the calculation procedure given in Reference 7,
Ap = 8.5 psi
therefore,
Ap = 14.6 psi
F. Static Pressure Distribution
Figure 11 gives the static pressure at selected points in the Cyclocentrifuge.
The basis for this estimate is References 6 and 7 and the calculation for the
pressure drop across the turbine.
366
-------
G. Particle Buildup
Even though the absolute velocity of the gas entering the centrifuge is purely
axial, the potential for particle build-up on the inside wall of the centrifuge
still exists. Accumulation depends on particle size distribution, radial veloc-
ity, velocity in the boundary layer, and the rate of particle arrival relative
to departure.
Preliminary calculations based on particle concentration derived from spectral
measurements presented in Reference 10 and frictional velocity estimates sug-
gested in Reference 11 indicate, particle buildup should not occur to any
significant degree.
5. SIGNIFICANT STRESSES AND DEFLECTIONS
Nickel-chromium alloy In 718 was selected for all rotational parts because of
its long-life properties at temperatures up to about 1200°F and because it can
be cast and welded. Based creep data, a design temperature of 1000°F was
selected and the maximum stress was set at 80,000 psi. Figure 12 shows the
maximum stress and deflections at and between the swirl augmentation blades,
which are considered acceptable.
6. ROTOR DYNAMICS ANALYSES
A. Rotor Model
Figure 13 shows a schematic of the Cyclocentrifuge rotor assembly and a tab-
ulation of the engineering data used in mathematical modeling for dynamic
analyses.
The design stiffness of each of the journal bearings is 1.0 x 10 lb/in., and
including the bearing mounting stiffness, the total support stiffness at each
bearing station is 0.857 x 106 lb/in.
B. Critical Speeds
The first two undamped lateral critical speeds are plotted in Figure 14 versus
the stiffness of the upper journal bearing, which, for the present case, is the
same as the lower journal bearing. At the design stiffness of 1 x 10 lb/in.,
the first two critical speeds are 6,936 and 11,244 rpm, thus, the first critical
367
-------
is 26 percent above the operating speed. The modes shape for the first critical
speeds is shown in Figure 15.
C. Unbalance Response
The degree of expected residual unbalance as calculated by 4 W/N, gives 0.726
oz-in. of unbalance for the present rotor. This balance specification is often
used by the U.S. Navy for rotors of the present size where the unbalance is
given in inch-ounces and W = weight of rotating assembly, Ib; N = rotor speed,
rpm. The response is tabulated for an unbalance of 1.0 oz-in. at Stations 4
and 29. The amplitude of response at the bearings is plotted versus speed in
Figure 16, and is seen to be acceptable because the expected residual unbalance
can degrade by an order of magnitude before the bearing eccentricity ratio ex-
ceeds 25 percent.
D. Damped Lateral Critical Frequencies and Stability
Damping coefficients of 320 Ib sec/in, were calculated for each journal bearings
and the first two damped critical frequencies were determined to occur at speed
of 7053 and 10,933 rpm. Thus, the lowest damped critical speed is about 28 per-
cent above the operating speed.
Figure 17 gives the mode shape and stability corresponding to the first natural
frequency. The stability is determined from the system log decrement which
indicates if a disturbance to the rotor will grow or decay in amplitude. The
threshold of stability is designated by a zero log decrement. If the log
decrement is positive, a disturbance will decay and the system is stable; if
the log decrement is negative, the opposite is true. In the present case, all
natural frequencies are well within the stable region.
BEARING DESIGN
Hydrostatic oil bearings were selected because they will provide long life with-
out replacement.
A. Journal Bearings
The upper and lower journal bearings are identical and their geometry is shown
in Figure 18. Table 1 gives the design summary for the journal bearing.
Since the rotor is mounted vertically, the only load on the journal bearing
will be due to rotor unbalance. The touch-down load can be estimated with
reasonable accuracy by assuming the stiffness is constant and the eccentricity
368
-------
TABLE 1
DESIGN SUMMARY FOR JOURNAL BEARINGS
Hydrostatic Journal Bearing with Two Rows of Pads Separated by a Circumferential
Groove.
Length = 3.000"
Diameter = 5.000"
Radial Clearance = 0.0025"
_ , „., , f Axial = 0.300"
Land Widths j
CCircumferential = 0.300"
_ . _ ("Axial = 0.200" x .25" deep
Drain Grooves ) v
iCircumferential = 0.400" x .25" deep
Number of Circumferential Rows of Pads = 2
Number of Pads per Row = 5
Number of Pockets per Pad = 1
Depth of Pocket = 0.15"
Number of Restrictors per Pocket = 1
D «. • n- • f Length = 0.30"
Restrictor Dimension )
(.Diameter, d = 0.020"
J Supply, p = 600 psia
3
Pocket, p = 369 psia
Ambient, p = 250 psia
3.
Total Lubricant Flow = 8.74 gpm
Journal Speed = 5500 rpm
„ A- i ct.-ff ("Theoretical = 1,360,000 Ib/in.
Radial Stiffness )
(.Used in Design Cal. = 1,000,000 Ib/in.
rCouette = 2.40 hp
Power Loss < Pumping = 1.79 hp
^-Total = 4.19 hp
369
-------
ratio at touch-down is 0.75. Using these assumptions, the touch-down load is
estimated to be 1875 Ib per bearing or a total radial load of 3750 Ib. The
estimated maximum radial load capacity is a factor of 3.78 times greater than
the weight of the rotor, which should be satisfactory for reasonable
contingencies.
B. Thrust Bearing Design
The thrust bearing is designed to support the centrifuge weight on a hydrostatic
film. Figure 19 shows the geometry of the double-acting thrust bearing and Table
2 summarizes the design. Figure 20 gives the load capacity, stiffness, flow,
and power loss as functions of axial displacement. Figure 20 shows the design
has ample load capacity and stiffness and that the rigid body axial critical
speed is 7201 rpm which is safely above the operating speed.
C. Lubricant Properties and Flow Requirements
Lubricant properties used in the bearing analyses are listed below. These proper-
ties are obtainable from fluorinated oils such as Krytox 143.
Temperature
Property 100°F 400°F
Viscosity p, Ib sec/in.2 8.56 x 10~5 9.0 x 10~7
Density p, lb/in.3 6.76 x 10~2 5.76 x 10~2
Heat Capacity C , BTU/lb °F 0.238 0.290
Thermal Conductivity K Btu/(sec in. °F) 1.16 x 10~ 1.16 x 10~
Flow requirements for the bearings are:
Upper Journal Bearing 8.74 gpm
Lower Journal Bearing 8.74 gpm
Thrust Bearing 8.80 gpm
Total 26.28 gpm
370
-------
TABLE 2
DESIGN SUMMARY FOR THRUST BEARING
Double-Acting Thrust Bearing with 20 Pads/Side
OD
ID
Drain Grooves: Width x Depth
fCircumferential
Land Widths:
[Radial
Pocket Depth
Supply Orifices:
rLength
Diameter
Total Clearance (Both Sides)
Design Weight (Load)
Design Eccentricity (From Centered
Position)
Design Stiffness
Design Speed
Pressure:
• Supply
• Ambient
Flow (GPM)
Power Loss, HP:
• Couette
• Pumping
• Total
Temperature Rise °F
= 7.200"
= 6.475"
= 0.10" x 0.10"
= 0.10"
= 0.10"
- 0.05"
= 0.45"
= 0.0160 (Loaded Side)
= 0.0135 (Unloaded Side)
= 0.0050"
= 991 Ib
= 0.000835"
= 1,460,000 Ib/in.
= 575.96 Rad/Sec = 5500 RPM
= 600
= 250
Loaded
Side
5.21
1.73
1.06
2.79
psia
psia
Unloaded
Side
3.59
0.86
0.73
1.59
Total
8.80
2.59
1.79
4.38
5.90
4.84
5.47
371
-------
D. Temperature Distribution Around Bearing Assemblies
Heat flow rates, heat transfer coefficients, and selected temperatures in the
upper and lower bearing areas are shown sketched in Figures 21 and 22, respec-
tively. The temperature levels are considered acceptable for satisfactory
bearing performance except for the metal area surrounding the journal bearing,
which will require insulation to avoid condinsation of tar.
8. ECONOMIC ANALYSIS SUMMARY
The economic viability of the Cyclocentrifuge can be assumed to be determined
by its effect on the production cost of electricity in a typical combined steam
and gas turbine plant burning low Btu gas from coal. For purposes of this
analysis, the production cost was calculated on the same basis as that used in
ERDA Report 76-49, "Economic Analysis of Westinghouse Low Btu Gasification -
Combined Cycle Power Generating System Producing 134.1 Megawatts", prepared
by U.S. Department of the Interior, Bureau of Mines for the Energy Research
and Development Administration, March 1976.
The production cost, as calculated in ERDA 76-49, includes the cost of coal,
dolomite, water, spent oxidizer disposal, labor and supervision, plant main-
tenance including supervision and materials, payroll overhead, operating supplies,
administration and general overhead, and capital charges at 17 percent of total
capital investment.
With the Cyclocentrifuge concept as presently developed, the temperature of
the low Btu gas must be reduced to about 1000°F from the Baseline temperature
of 1546°F. This requires a higher gas flow for the same gas turbine firing
temperature and corresponding output (i.e., constant gas turbin^ air flow).
The higher gas flow requires a 5.5 percent increase in coal to the gasifier.
To provide the lower gas temperature, additional steam generating surface must
be provided in the gasifier, with a resulting increase in low pressure steam
generated. This steam, over and above that required by the gasification pro-
cess, can be expanded in the steam turbine and will produce additional
electrical output of about 4.4 Megawatts. The net effect is to reduce the
cycle efficiency a little over 2 percent.
372
-------
At the same time, the Cyclocentrifuge is estimated to have only 15 psi pressure
drop through it, which is less than the 30 psi of the cyclone and sand bed
filters used in the Baseline cycle. This reduction in pressure drop reduces
the work required of the booster compressor to maintain the same pressure at
the gas turbine control valve by about 450 kw, or 0.34 percent gain in net
output. This gain in output was neglected in estimating the production cost
as offsetting other losses that might not have been properly calculated or
accounted for.
In order to calculate the capital cost of the complete installation and the
corresponding capital charges to be included in the production cost of elec-
tricity, an estimate was made of the cost of the Cyclocentrifuge based primarily
on the weight of the complete assembly. This estimated installed cost of
$93,573 was substituted in the capital cost estimate for the $86,000 allowed
in the Baseline system for the cyclone and sand bed filters.
Allowance was also made for the increase in the steam turbine-generator and
condenser cost, due to the increased rating, the increase in coal preparation,
dolmite preparation, and gasification cost due to the increased coal flow.
The total capital cost increased from $40,915,400 to $41,654,980 and the
corresponding capital charges from $6,955,600 to $7,081,47.
With the increase in plant output, the net effect is to reduce the production
cost of electricity, with coal at $11.00 per ton, from $.01592 per kw hr to
$.01581 per kw hr, a cost improvement of .7 percent. Considering the effect
of reduced pressure drop in the gas clean-up system, this is probably realistic.
Table 3 summarizes the comparison between the baseline system and the modified
system.
373
-------
TABLE 3
Economic Summary of Comparison Between Baseline System and Cyclocentrifugal
Item
Gas temperature to gas turbine, °F
Output, Mw
Coal fired tph
Steam produced in gasifier Ib/hr
Pressure drop across cleaning
equipment, psi
Gas cleanup, equipment cost, $1,000
Cycle efficiency, %
Capital cost, $1,000,000
Capital charges, $1,000,000
Output cost per kw hr
Baseline
System
1546
134.1
49.6
42,000
30
86.0
37.40
40.92
6.96
0.01592
Modified
System
1000
138.5
49.5
51,926
15
93.6
36.65
41.6
7.08
0.01581
Percent
Increase
-35.3
3.3
5.5
23.6
-50.0
8.8
-2.0
1.7
1.7
-0.7
9 . SUMMARY AND CONCLUSIONS
The mechanical arrangement of the Cyclocentrifuge developed during this study
is shown in Figure 1. It has may intrinsic advantages, some of which are
unique to the concept. These advantages, listed below, implicitly define
areas in which the concept can be employed to aid in improving the effective-
ness and efficiency of coal conversion.
• Fine particulate matter can be separated from a gas stream.
Separation of sub-micron particles is possible.
• High gas purity can be attained. One pound of solids per million
pounds of gas is feasible (1 ppm total solids).
• Hot gas can be treated without special heat transfer design.
A nominal gas stream temperature of 1000°F is practical. With
special heat transfer schemes, higher temperatures are possible.
374
-------
• Pressurized gas can be treated. The centrifuge assembly is
pressure-balanced because it is located inside a modified
conventional cyclone.
• The design concept does not accumulate particulate matter.
Continuous operation is practical.
• Electrical input power is not required. The rotating assembly
is efficiently turbine-driven by the process gas.
• The design is compact and does not require intricate piping or
controls.
• There are a minimum of interface problems.
• The design does not create new secondary pollution problems.
• First cost and operating costs should be equal to or less than
present alternate cleanup approaches.
10. ACKNOWLEDGEMENTS
This work was performed under Contract No. E(49-18)-2428 for the Fossil Energy
Research and Technology Division of ERDA by the Mechanical Equipment Section of
the MTI Engineering Department. The ERDA Program Manager was Mr. W. Fedarko.
Mrs. A. 0. White, L. Folsom, A. Tuzinkiewicz, S. Park, A. Artiles, and P. McGee
made significant contributions to this project in behalf of MTI.
375
-------
8. REFERENCES
1. Process Evaluation Office, ERDA at Morgantown, W. Va., "Economic Analysis
of Westinghouse Low-Btu Gasification Combined-Cycle Power Generating
System Producing 134.1 Megowatts, ERDA Report 76-49.
2. Foster, A. D.; Von E. Doering, H.; Hickey, J. W., "Gas Turbine Fuel",
General Electric Gas Turbine Reference Library, Document GER-2222H with
Supplement "Specifications for Gas Turbine Fuel, Gases", GEI-41040C;
"Liquid Fuels Specification", GEI-41047E.
3. White, A. 0., "20 Years Experience Burning Heavy Fuels in Heavy Duty Gas
Turbines", ASME publication 73-GT-22.
4. Buckland, B. 0., "The Effect of Treated High-Vanadium Fuel on Gas Turbine
Load, Efficiency and Life", ASME publication 58-GTP-17.
5. Weth, G.; Farmer, W. M., "In-Sites Laser Measurements in Coal Processing
Systems", View-Graph presentation at ERDA, Washington, 9 November 1976;
From Spectron Development Laboratories, Inc. SDL Report 77-6801.
6. Stairmand, C. J., "The Design and Performance of Cyclone Separators",
Trans.'Instr. Chem. Engr., Vol. 29, 1951 (Great Britain).
7. Stairmand, C. J., "Pressure Drop in Cyclone Separators", Engineering,
October 21, 1949.
8. Zweifel, )., "The Spacing of Turbomachine Blading, Especially with Large
Angular Deflection", Brown Boveri Rev. 32, December 1945.
9. Harlock, J. H., "Axial Flow Turbines", Butterworths, 1966.
10. In Situ Laser Measurements in Coal Processing Systems, Spectron Development
Laboratories, Inc., Report No. 77-6801, Nov. 9, 1976.
11. Howarth, L., ed., Modern Developments in Fluid Dynamics - High Speed Flow,
Vol. I, Aeronautical Research Council.
12. McCabe, J. T.,'"Centrifuges for Coal Conversion and Related Processes,
Phase I: Feasiblility Study", Final Report MTI 77TR34 Mechanical Technology
Incorporated to ERUA, contact E(49-18)-2428, March 1977.
376
-------
a
•H
u
0) p.
t-H 3
PJ C
ra
0) O>
f-H i—I
u u
x
U CO
ra
T) U
0)
C M
•H O"
r^ '-'--'
e
c a
cj ex
3
V- U-l
O -H
U^ Vj
4-1
O C
•H CJ
JJ O
d c
E r-l
ai u
.n >>
u u
CO
tc
to
w
(U
u
o
^ c
£-, O
PQ J-i
I OJ
3 C
o a)
^ o
c
-H
-------
SHAFT
UPPER HYDROSTATIC JOURNAL
8 THRUST BEARING ASSEMBLY
CENTRIFUGE UPPER SUPPORT STRUTS
DIRTY GAS INLET
DESIGN SPECIFICATIONS
VOLUME FLOW 125.260 SCFM
INLET PRESSURE 250 PSIA
INLET TEMPERATURE lOOO'F
PURITY I PPM TOTAL SOLID
BY WEIGHT
MAXIMUM ALLOWABLE
PARTICLE SIZE I MICRON DIA
CENTRIFUGE
SHELL
SWIRL
AUGMENTATION
BLADES
CENTRIFUGE LOWER
SUPPORT STRUTS
DRIVE
TURBINE
ASSEMBLY
BEARING
SUPPORT-
STRUTS
LOWER HYDROSTATIC
JOURNAL BEARING
ASSEMBLY
PARTICULATE
COLLECTOR
VESSEL
LOWER FLANGE OF
CYCLOCENTRIFUGE
Fig. 2 Cyclocentrifuge, General Arrangement
378
-------
CENTRIFUGE
UPPER SUPPORT
STRUTS
CENTRIFUGE
LOWER SUPPORT
STRUTS
UPPER JOURNAL 8 THRUST
BEARING ASSEMBLY
SEAL
CENTRIFUGE SHELL
SWIRL AUGMENTATION
BLADES
TURBINE
LOWER SHAFT
ASSEMBLY
Fig. 3 Rotating Assembly - Cyclocentrifuge
379
-------
UPPER JOURNAL
THRUST BEARING
HOUSING
,OIL RETURN
..OIL SUPPLY
VESSEL HEAD
UPPER SHAFT
DIRECTION OF
ROTATION
MID
ROOT,
12
REVERSE THRUST
BEARING HOUSING
CENTRIFUGE UPPER
SUPPORT STRUTS (16)
CENTRIFUGE SHELL
AERODYNAMIC CONE
4 Upper Bearing and Shaft Assembly - Cyclocentrifuge
380
-------
TIP
AERODYNAMIC
ROOT
Tip CONE
SUPPORT
TURBINE BLADE _
SECTIONS Cf
LOWER-
SHAFT
HYDROSTATIC
BEARING HOUSING
ASSEMBLY
LOWER SUPPORT
STRUT SECTIONS
.CENTRIFUGE
SHELL
-CENTRIFUGE LOWER
SUPPORT STRUTS (10)
LOWER BEARING
SUPPORT RING A
ft >
( }
— *; —
( ) ^
J/'
1
,.
1 A^ ^
A<^ t 1
'- 1 DWFR
1
J
RF
B<^1
1
r
r i»
B-J1
•ARINR -1
BEARING
OIL IN
AERODYNAMIC
CONE
SUPPORT STRUTS
SECTION A-A
SECTION B-B
Fig. 5 Lower Bearing and Shaft Assembly - Cyclocentrifuge
381
-------
UPPER JOURNAL S
THRUST BEARING
COMPONENTS
ITEM
1
*2
*3
4
5
6
7
*8
9
10
"II
»I2
13
14
15
16
»I7
*I8
DESCRIPTION
RESERVOIR/CLEANOUT ft DRAIN
PUMP/MOTOR ASSEMBLY
SUCTION STRAINER
COOLER
WATER STRAINER
MODULATING VALUE
DIRECTION VALVE
RELIEF VALVE
NEEDLE VALVE
GAGE
FILTER
CHECK VALVE
FILLER/BREATHER
SIGHT GAGE
TEMPERATURE SWITCH
ACCUMULATOR
SUCTION PUMP
PRESSURE REGULATOR
* DENOTES BACKUP UNIT REQUIRED.
LOWER
JOURNAL
BEARING
H20 COOLING
Fig. 6 Lubrication System Schematic - Cyclocentrifuge
382
-------
o
Pi
es
w
H
w
tj
u
M
H
Pi
o
c
01
•H
o
•H
w
01
T3
03
M
O
01
CO
3
C
0)
o
o
NOIlVJIVdHS JO ADN3IDIJJ3
383
-------
\
\
\
I
M O
cfl
C
c
•H
4->
ra
4-1
C
CJ
e
—<
3
C
0)
-------
= 87.3C
a = 80.9C
a2
HUB
RAD. = 13 IN.
V 0 = 24.08
a2
TIP
RAD. = 17 IN.
EXIT A
LAST BLADE EXIT
INTERMEDIATE A's
11 BLADE ROWS
INLET A
1ST BLADE INLET
Fig. 9 Velocity Triangles for Swirl Augmentation Blades
385
-------
1.0
CO
w
g 0.5
1.0
CO
I 0.5
z
ROTATION
ROTATION
0.5
1.0
I
1.5
INCHES
CAMBER LINE
TIP SECTION
0.5 1.0 1.5 2.0 2.5
INCHES
MID SECTION
2.0
2.5
3.0
1.5
1.0
CO
0.5
ROTATION
0.5
1.0
INCHES
1.5
ROOT SECTION
2.0
Fig. 10 Drive Turbine Blade Sections
386
-------
p = Static Pressure, psia
s
p = Total Pressure, psia
p =245.4
S
= 249.5
p = 240.6
Fig.11 Static Pressure at Selected Points in Cyclocentrifuge
387
-------
BETWEEN POINT LOADS
38.4 KSI
0.0201"
(V
57.7 KSI
0.0298"
57.7 KSI
0.0298"
50.4 KSI
0.0127"
AT POINT LOADS
15.3 KSI
50.3 KSI
0.0041"
41.2 KSI
0.0211"
9.3 KSI
0.0027"
76.7 KSI
0.0396"
76.7 KSI
0.0396"
56.8 KSI
0.0027"
Fig. 12 Maximum Stress and Deflection at Selected Points in Centrifuge
388
-------
o
S
n)
a
O
4-1
O
389
-------
100K.
10,000
1 ,000
c
_ c
/">
PERATING
PEED
500 RPM
2nd CRITICAL X^
1st CRITICAL x"""^
^^T
^
_^
^
^
\ i
BEARING PEDESTAL
STIFFNESS
6. y. 106 LB/IN.
PEDESTAL MASSES
BEARING 1 = 42.11 LB
BEARING 2 = 92.67 LB
^
--*-
^~
^~~^\
t
""
26% MARGIN
1
T
._ DESIGN STIFFNESS OF
HYDROSTATIC BEARINGS
1. x 106
I i
10
10 10
BEARING STIFFNESS (LB/IN.)
10
Fi8-
Undamped Lateral Critical Speed Map for Cyclocentrifuge Rotor
390
-------
o —
I
II
*-
X X X X
01
t-l
T4
fa
o
<4-l
<
_i
UJ
a
391
-------
0.5
w
Q
H
z
o
CL,
c/3
w
0.4
0.3
0.2
0.1
BEARING CLEARANCE =3.5 MILS
BEARING #2
STATION 29
SPEED
DESIGN
2000
BEARING //I
STATION 4
CRITICAL
SPEED
4000 6000
SPEED, RPM
8000
Fig. 16 Response Amplitude per In.-Oz of Unbalance at Station 18
392
-------
1 CYCLOCKNTRIFUGE - CONFIGURATION n - SINGLE LEVEL ROTOS - STAHILITY
MAJOR AXIS RFLATIVF AMPLITUDE
o.o
o.noo ..
•l.o
PH..SF ArJGLt.
PS I (DfC)
1 . 0 << <•
BRG. Mr. I 1 .'.M
10.000 .
20.000
30.noo .
FP NATU'MI. CRFOHF
SYSTLM 1.00 OprPtMEMT
BRG. MO.
3 . S 1
IF » WtPkKSFNTS Tht HOTOI' AM^I. ITHHF, * FHfF V1RHATION MAY MT h XPHl SSFn AS -
AH>>(lr) • r«» (Lf MHf)ABTI
THF INITlAl AMPl TTIIDF
TlMf (StCOMHSI
«T • PSI )
ARStX)
T
PSI
L AMHPA
(IMF (j = M1MPFO NAT'lOAL rrjFOUFsirY (RAPIANS/SFC)
-51.9 (I/sec)
Fig. 17 Mode Shape and Stability for First Natural Frequency
393
-------
BEARING
0.15"
C = 0.0025"
0.25"
1
1
3.0(
^0.30TYP
j,, °'80
-*
f
0.20
—
-* — 2.742" (62.83^ — *\
d -"So. 020"
o
0
K1 — 2.142" — »J
(49.08°) '
o
VTKW A
Fig. 18 Journal Bear in p, Geometry
394
-------
Loaded
Side
0.725 U-
3.2375- *•
OCKET DEPTH
0.05"
Unloaded
Side
NOTE: SEE TABLE 10 FOR
OTHER DESIGN
PARAMETERS
Fig. 19 Thrust Bearing Geometry
395
-------
2.5 p.
z
I—I
PQ
00
00
u
H
00
z
M
Pi
H
00
PO
*
H
M
U
cu
u
a
z
M
Pi
Cd
H
05
OS
H
0.5 1.0 1.5 2.0 2.5
AXIAL ECCENTRICITY, MILS
3800
3000
2000
W = 991 LB
1000
e = 0.835 MILS
0.5 1.0 1.5 2.0 2.5
AXIAL ECCENTRICITY, MILS
§
o
O
z
M
PQ
H
00
g
PH
00
00
o
§
p-l
O
(O
H
C/3
I
H
10
Q=8.74GPM
0.5 1.0 1.5 2.0 2.5
AXIAL ECCENTRICITY, MILS
P = 4.38 HP
TOTAL
POWER
COUETTE
SHEAR POWER
2.59 HP
PUMPING POWER
= 1.79
I,
_l_
0.5 1.0 1.5 2.0 2.5
AXIAL ECCENTRICITY, MILS
Fig. 20 Thrust Bearing Performance Characteristics
396
-------
METAL T = 370°F
(h = 1275)
METAL T = 535°F
(h = 1.5)
METAL T = 630°F
METAL T = 880°F
(h = 1.5)
OIL OUT
T = 321°F
\
\
V
^
/*
OIL IN 17.54 GPM @
>/
4
_y
f
^ f /
/
s
T = 299°F
AMBIENT GAS TEMPERATURE = 1000 °F
UalupR nf Hpat Transfer
OIL
FLOW
1
>r
#•
^~
^
,-'-
^
*
^
***
^
\
**•
^
Coefficients, h, given in
Btu
hr ft2 °F
/ METAL T = 485 °F
/ (h = 383)
^ METAL T = 420°F
-^ (h - 58)
.* OIL T = 300°F
^^OIL T = 315°F
METAL T = 565 °F
/
^
METAL T = 850°F
--^^^(h = 1.5)
\
\
OIL T = 320°F
OIL T = 317°F
Q (GAS -> OIL) = 48,000 BTU/IIR
Q J. BEARING = 10,700 BTU/HR
O THR. BEARING = 11,200 BTU/HR
Fig. 21 Temperature Distribution at Upper Bearing
397
-------
"»
fc
PH 0
O U~)
in o
iH r°
ro
II
II
H
H _____
J f
l-l 1-
M 0
o \^^
\
\
c\
0 \
-/
°^ /
<*"> c~~z..,.
11 £
H ^
3 II
«•«
^
V
1
o c
o . -^
S S e
U-4 01
11 S3
f_i ,j « 00
Z C/2 E"1 "
M tn -1-
bJ u
•] ^ Clj **
M 1"^ OJ cn
ij U-< 01 *-•
/,.. 0 -^ CO
/ u
Ol U-l
b 3 ^
o •-< OJ
x~\ Q CB O
en 0 > "
CO o
f> ^H
» II
•C V>
•^ <
o
1 t-l
Z fc
W o
v-( O
9 CN
1
•J "-
* ^
* ^ "J y
< »
/ H
V b ^
3 3
, H H
| CQ CO
1 00
1 O O
1 r-~ o
o en
rH CN
.... ,.,,, J u
— z
I-H
0
^^ Z 1
ro C = 0
00 Z O rJ
c*"i i-i ^ at, u
.. Si nJ Pi >H
N < M ^ CJ
t: o V2
js pq —
^ t t U
a
i^i co on O
a < < a!
1 -o C • o ~
o- o- cf
- u.
3 0
3 CN
CNl
— ro
O
CM
U-l
;<
=s
i:
=)
-<
23
O
— 1
X
J~l
II
^
NT R I CURE VESSEL
5ulation
•Jj ^
c 1 C
U M
0_
00
c
0)
cc
O
o
O
•H
cn
— j
O
0)
l-l
J-l
a
i-i
0)
&
OJ
00
398
-------
FINE PARTICLE COLLECTION EFFICIENCY
IN THE A.P.T. DRY SCRUBBER
By:
S. Calvert, R. Q. Patterson
Air Pollution Technology, Inc.
San Diego, CA 92117
D. C. Drehmel
Environmental Protection Agency
Industrial Environmental Research Laboratory-RTP
Research Triangle Park, NC 27711
399
-------
FINE PARTICLE COLLECTION EFFICIENCY
IN THE A.P.T. DRY SCRUBBER
By
Dr. Seymour Calvert
Dr. Ronald G. Patterson
Air Pollution Technology, Inc.
4901 Morena Blvd., Suite 402
San Diego, California 92117
and
Dr. Dennis C. Drehmel
Particulate Technology Branch
Utilities and Industrial Power Division
Industrial Environmental Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
The Particle-by-Particle (PxP) scrubber is a device which can be used
at high temperature and pressure for the collection of fine particles on
larger particles, which can be cleaned and recycled. Particle collection
is mainly by inertial impaction and to some extent by diffusion for
smaller particles. Experimental data on particle collection are in
agreement with a mathematical model.
400
-------
FINE PARTICLE COLLECTION EFFICIENCY
IN THE A.P.T. DRY SCRUBBER
The development of advanced energy sources such as coal and shale
oil gasification results in high temperature and pressure (HTP) process
gas streams which require removal of particulates before utilization.
For example, the ultimate use of such a process gas could be combus-
tion and expansion in a gas turbine for generation of electric power.
Turbo-machine experience indicates that potential erosion and/or corro-
sion of machine components are due in part to the particles in the gas
stream.
Fine particle removal from gases at the high temperatures and pressures
encountered in fluidized bed combustors and various fuel conversion pro-
cesses places severe requirements on the gas cleaning system. The environ-
ment of reactive gas mixtures of temperatures up to 1,100°C and pressures
up to 15 atmospheres can be withstood by only a few structural materials.
Particle collection efficiency must be high; perhaps 90% for 1.0 urn diameter
particles and 99.5% overall.
The elevated temperature and pressure conditions suggest that new
devices for removal of fine particles may be necessary. Typical particle
collectors used in fossil-fuel-fired power plants (electrostatic precipi-
tators, scrubbers, fabric filters) generally operate at temperatures below
260°C and at low pressures. The suitability of these components at elevated
temperatures and pressures may be limited. The A.P.T. dry scrubbing system,
which we call the "PxP" system (for "particle collection by particles"), is
compatible with the special demands of HTP gas cleaning.
PxP SYSTEM
The PxP system for fine particle control utilizes relatively large
particles as collection centers for the fine particles in the gas stream.
The relatively large particles (collector particles) introduced to the gas
401
-------
stream can collect fine particles by mechanisms such as diffusion, inertial
impaction, interception and electrophoresis. The larger size of the col-
lector particles allows easy separation from the gas stream by methods such
as cyclones, and gravitational settling.
Figure 1 is a functional diagram of the process steps for a represen-
tative PxP system. The functional phenomena 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 contacting the collectors with the gas in order
to encourage the movement of the fine particles to the collectors. A ven-
turi device can be used for the contactor which would be analogous to a
venturi scrubber except that solid collectors are used instead of liquid
drops. Alternative contactors such as a centrifugal scrubber could be used.
The next process step is to remove the collector particles after suffi-
cient exposure in the contactor to cause capture of the initial fine par-
ticles 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 clean-
ing them for recycle and disposing of the material collected 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 perfor-
mance 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:
402
-------
Pt = exp
B
K
po
4 K
0 5
+ 4.2 'l-uS) - 5.02 K
0.5
B
K
po
4
- 5-02
po
po
4-2
d 2
u
where K
Gt
u5 = 2
po
+ &1W1
K
po
1 - x2 + (x1* - x2)
0.5
. -
16
V0.5
po
0.7
CD
(2)
(3)
(4)
J
and B
JDo
where d
u
pa
Gt
P
p
(5)
= aerodynamic particle size, cm
= gas velocity at throat, cm/s
= gas viscosity, g/cm-s (poise)
= collector diameter, cm
= throat length, cm
= gas density, g/cm3
= collector density, g/cm3
= gas volumetric flow, m3/s
= collector volumetric flow, m3/s
= drag coefficient for drops at the venturi throat
inlet, dimensionless
403
-------
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 2 is a plot of particle penetration
against particle size with collector/gas flow rate ratio as a parameter
and with a 20°C gas temperature. Figure 3 is a similar plot with an 820°C
gas temperature. 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 4.
The predicted penetration curves have the following characteristics:
1. For a given set of operating conditions, the penetration decreases
with increasing size of fine particles. This is expected since the- col-
lection 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 increasing value of (Q PC/Q(0 •
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 viscosity with temperature which reduces
the effective inertia of the fine particles.
It can also be shown that collector particle diameter affects collec-
tion efficiency when other factors are held constant. The cut diameter
(i.e., the diameter of the particle which is collected at 50% efficiency)
decreases as collector diameter decreases. Collection efficiency for par-
ticles larger than several microns diameter varies in a more complex way,
depending on flow and geometric parameter combinations.
Experimental Program
Experimental work has been done by A.P.T. to determine fine particle
collection efficiency in a PxP scrubber in order to confirm the predictions
obtained from available mathematical models. A dibutylphthalate (DBP)
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 DBP aerosol had a mass median aerodynamic diameter of
404
-------
1.3 ymA and standard deviation, a =2.0
o
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 /Q ) was around 0.005 g/cm3 for the nickel
beads and 0.0017 g/cm3 for the sand.
The gravity separator used in these experiments is shown in Figures
5 and 6. The system gas flowed either horizontally or vertically downward
flow into the separator. Cleaned gas flowed out of the branch of the
separator Tee.
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 cumula-
tive mass distributions were plotted and the particle collection effi-
ciency was computed from the ratio of the curve slopes at several particle
diameters.
The resulting penetration data are shown in Figure 7 for DBF collec-
tion on sand. The cascade impactor data led to the penetration relation-
ship labeled "experimental curve." The prediction for (Q p /Qr) = 0.002
is also shown in Figure 7 and compares well with the experimental curve.
Particle penetration data for all runs with nickel and sand collectors
are represented in Figure 8, a "cut power plot." The cut diameter is
plotted against gas pressure drop in Figure 8. The line represents the
relationship which is predicted and which has been confirmed by a number
of field tests on large wet scrubbers. Agreement between the data points
and the line is good.
CONCLUSIONS
The experimental data on the primary collection efficiency of the PxP
system agree well with predictions based on a mathematical 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 reasonable to expect it to fit the data. The
PxP A.P.T. dry scrubber system has the same primary collection efficiency/
405
-------
power relationship as a venturi type wet scrubber.
The overall efficiency of the PxP system will depend on the re-entrain-
ment characteristics of the specific system in addition to the primary
efficiency. Particle and collector properties, system geometry, flow rates,
and other parameters will influence re-entrainment.
Research is continuing on the experimental evaluation of the PxP system
for HTP application. The work upon which this paper is based is supported
by the U.S. Environmental Protection Agency.
REFERENCE
Yung, S.C., S, Calvert, and H.F. Barbarika, "Venturi Scrubber Performance
Model", EPA-600/2-77-172, NTIS PB 271515/AS, August 1977.
406
-------
CO
C-D D
fy".
2 <
< U
W CO
J I-H
>
Pi
o
H
Pi
D-,
W
CO
-> O
< >
Pi CO
>
Pi
o
H
O
U
H -J
U U
tq KH
^ H
nj Pi
0 <
U PH
<
\
'
k
Pi CO
H _}
U CJ
W I-H
O <
f ^-3 O-t
Q
Pi
<
U
CO
1— 1
>
CO
W
t— I
o
^
Pi
OPi
E— ' PJ
U 2
fjj
-------
1.0
0.1
o
i—i
H
U
<
&.
H,
Z
O
Pi
H
W
2
W
P-.
0.01
0.001
T = 20°C
u = 57 m/s
o
d = 100 ym
0.1
10
AERODYNAMIC DIAMETER, ymA
Figure 2. Theoretical particle collection characteristics of
the A.P.T. dry scrubber.
408
-------
1.0
= 0.001
0.1
o
I—I
H
U
<
Pi
O
I—I
H
<
Pi
H
W
0.01
T =
u =
s
d =
0.001
820°C
= 57 m/s
100 urn
0.1
Figure 3.
10
AERODYNAMIC DIAMETER,
Comparison of the particle collection characteristics
of the A.P.T. dry scrubber at 20°C and 820°C.
409
-------
1.0
= 0.002
0.1
O
i—i
E-i
U
O
I—I
H
0.01
u = 57 m/s
100 ym
0.001
0.1
Figure 4.
10
AERODYNAMIC DIAMETER, ymA
Comparison of the particle collection characteristics
of the A.P.T. dry scrubber at 20°C and 820°C.
410
-------
rt
f-H
03
X
w
00
re
-p
P!
o
Pi
o
H
U
LT>
-------
FROM CONTACTOR
GAS EXIT
Figure 6. Vertical flow separator,
412
-------
1.0
2
O
i—i
H
U
P-c
O
H
W
2
W
PL,
\
\
0.01
THEORETICAL
EXPERIMENTAL
\
\
\
\
\
\
0.1
Figure 7
0.5 1.0 2.0
AERODYNAMIC DIAMETER, ymA
5.0
Comparison of experimental with theoretical particle
collection characteristics of the A.P.T. dry scrubber
413
-------
D
O
i-J
P-.
_1
<
H
2
O
1X1
I !
Pi
O
O
_1
p-
cx
tu
tq
PC
c
tq
n:
LO
CO
t--
O
LO
O
to
O
03
O
Di
PJ
CO
CO
p-1
CO
<
a:
PI-
CO
On I
X
H
PH
CD
t/)
u •
•H px,
•H (/)
!H C
CD O
•M -H
U -P
re nj
fi rH
03 CD
X ^
O
S-i
C CD
O S
•H O
•P D-
o ^^
CD 4->
O
U
0
^H
U
^ CD
03 X
CXP
O P
•H
•H CD
^ ,C
03 fj
PH 2
6 ^
O U
c_> to
00
CD
O
to
O
Cs!
tO
(XI
C
'H313WVIQ
O
bC
•H
Lt,
414
-------
HOT GAS CLEAN-UP BY PARTICLE ENTRAPMENT IN
COAL SLAGE BASED GLASSES
By:
L. R. McCreight, A. Gatti, H. W. Rauch, M. J. Noone
General Electric Company
Philadelphia, PA 19101
415
-------
HOT GAS CLEAN-UP BY PARTICLE ENTRAINMENT IN COAL SLAG BASED GLASSES
Louis R. McCreight - Arno Gatti - Harry W. Rauch - Michael J. Noone
Space Sciences Laboratory
General Electric Company
Valley Forge Space Center
P. 0. Box 8555
Philadelphia, Pennsylvania 19101
An important problem in the utilization of gases from coal
combustors to fuel gas turbines is the need to remove harmful particulates.
For efficiency this must be done without appreciable cooling of the gases.
Coal slag based glasses are being considered in an analogous manner to
water scrubbers at normal ambient conditions for this purpose. Control of
the viscosity by the use of both chemical and electrical energy additions
is also included in the study to potentially permit operation over a
temperature range of about 1000 to 1600°C. In this initial feasibility
demonstration program, viscosity studies on various glasses including
those made from fly ash and some model air cleaning studies are being
performed in a clear plastic labyrinth unit using glycerine at room
and low temperatures to simulate glasses. Later work will include high
temperature studies and conceptual designs for a full sized system.
416
-------
HOT GAS CLEAN-UP BY PARTICLE ENTRAINMENT IN COAL SLAG BASED GLASSES
INTRODUCTION
The objective of this study program is to demonstrate the
technical and economic feasibility of utilizing coal slag based glasses
as a fluid to entrain particulate matter at high temperatures without
significant carry-over. This will permit the direct utilization of the
effluent from a coal combustor without cooling the gases to effect the
clean-up. The emphasis shall be on providing gas sufficiently free of
particulates to permit turbine operation without undue corrosion or
erosion. The primary benefit from the use of glass coated surfaces to
trap particles is due to the viscosity or stickiness of the glass which
will minimize richochet and fracture of the particles as apparently often
happens in other apparatus. Then since the particles are often of a
siliceous nature they should dissolve in the glass and thereby be less
volatile or likely to carry over into the turbine. Thus in this approach
to hot gas clean up the stickiness of the particles is a virtue rather than
a problem as it is in some other processes.
In addition to this primary benefit, successful accomplishment of
this program objective would provide more compact, easily handled waste
products, since the solid glassy form of the fly ash occupied only about
20-25% of the volume of the powdered fly ash and may therefore have
considerable utility as an aggregate etc. in building materials. These
aspects are being considered but not extensively explored at this time.
417
-------
Instead the primary effort is aimed at demonstrating both qualitatively
and eventually quantitatively the practicality of the idea.
The work is currently being done under two principal tasks.
I - "Fly Ash Melt Studies", and II - "Particle Collection Studies".
The results of these two tasks will then be factored into a third task
toward the end of the program titled "Total Power Generation Cycle
Efficiency Study". Since the program is less than five months old,
emphasis in this paper will be on the primarily qualitative demonstrations
of feasibility being performed under the first two tasks.
FLY ASH MELT STUDIES
Several widely used coals were considered as sources of typical
fly ash for this program. These included Montana Rosebud, Pittsburgh No. 8
and Illinois No. 6, plus a standard soda lime silica and an amber container
glass for model studies at elevated temperatures. Glycerine at various low
temperatures is also used in the particle collection studies to be discussed.
The compositions of these materials is shown in Table I and the viscosity
versus temperature curves for synthetic versions of some of these materials
are shown in Figures la and Ib along with the curve for an average synthetic
slag prepared by the National Bureau of Standards which is designated as
K-884 and is based on an average of Montana Rosebud and Illinois No. 6.
Flow studies on both the synthetic slags and samples of actual fly
ash using a modification of the standard flow button test used for porcelain
enamels have been carried out while work in a model duct using glycerine has
been initiated under Task II, described later.
418
-------
The flow button test has been used throughout the porcelain enamel
industry for many years. It is a commonly used practical, quality control
technique for checking the fluidity and fusibility of enamel frits. The
test is considerably less expensive and simpler than high temperature
viscometry while producing reliable practical data. However, Dekker (1)
has shown that viscosity-temperature relationship can be determined but
with a somewhat more elaborate procedure than the conventional flow button
test.
Several factors must be controlled for reproducibility. These
include particle size of the materials, forming pressure, height and weight
of the pellets, composition of the materials and the temperature. However,
Plankenhorn (2) concludes that the predominant factors are the temperature
and the material composition. Our limited study supports that conclusion.
Fluidity studies are being conducted on the following materials:
Montana Rosebud fly ash
Pittsburgh #8 fly ash
An intermediate Montana Rosebud/Illinois #6
synthetic slag (NBS, K-884)
A standard soda lime silica glass
Amber container glass.
The Montana Rosebud fly ash was obtained from the Montana Power
Company, Billings, Montana; the Philadelphia Electric Company supplied the
Pittsburgh #8 fly ash; and, we synthesized the intermediate Montana Rosebud/
Illinois #6 slag from a composition derived by Capps and Kaufman of NBS
and designated as K-884. The soda lime silica standard glass (Standard
Reference Material-SRM 710) was purchased from the National Bureau of
Standards. The amber container glass was obtained locally and ground in our
laboratory.
-------
Both of the fly ash materials were of sufficiently small particle
size to readily pass through a 100 mesh sieve thus no further particle
size reduction was required. Both the K-884 synthetic slag and the
SRM-710 glass were ball milled until the particles passed through a
100 mesh screen.
Pellets were formed by mixing 25 grams of each material with
2.5 grams of 3.0% PVA solution, placing 1.5 grams of the mixture in a
steel die and pressing at 15,000 psi. Specimens resulting from this
treatment are cylinders 3/8" diameter and 3/8" high.
The fluidity runs were conducted by placing the pellets on tiles
mounted at 30° to the horizontal in a furnace preheated to the desired
temperatures, leaving them for the predetermined time, and then removing
them. Various refractory materials as tiles have been used including
mullite, alumina coated mullite, and dense magnesia-alumina spinel. The
latter has proven to be the most useful because of the lack of porosity
which obscured the results with the other materials.
As a result of a number of trials, it was found that the amber
container glass offered the most potential for operation at about 1000 C
where the current efforts on advanced gasifiers and combustors is focussed.
Figure 2 illustrates the behavior of this glass and various mixtures of it
with the K-884 synthesized coal slag in the flow button tests at
temperatures from 1000 to 1250°C.
Some further tests utilizing this same container glass with
admixtures of Montana Rosebud fly ash in the range of 2-10% indicates that
small amounts (2-5%) of this fly ash noticeably lowers the viscosity of the
420
-------
glass. This is a very important finding in that it portends that an
even broader range of fluidity may be achieveable.
PARTICLE COLLECTION STUDIES
Early feasibility studies of this idea for collecting particulates
were made with glazed wall tiles mounted in a manner to provide a
labyrinth flow path for the gases from a gas burner into which silicon
carbide particles could be entrained. These tests indeed showed the validity
of the idea but were not easily quantified nor modified to test a wide
variety of parameters.
Under this program a general purpose test duct was therefore
designed, built, operated, and modified for gaining design ideas and data
to be used in building a hot gas duct later in the program. The current test
duct is built of clear plastic as shown in Figure 3a and b. It has 10 square
feet of collection area, 0.1 sq. ft. inlet and exit ducts, and an internal
cross-sectional area of 0.05 sq. ft. to give a two fold increase in pressure
and flow. It is operated on a house air line at up to 100 cu. ft. per minute.
Collection of the fly ash particles in this duct is by means of
glycerine which can be cooled to yield viscosities which corresponds to
temperatures of 1000 to 1600 C in various glasses. The glycerine flows
through 1/4" diameter holes in the bottom of the supply tank, down over the
plates in the duct and initially into a common dump, however, this was later
revised to provide 14 separate compartments so that the results from each
collector section could be kept separate. Any particles which are not caught
on the glycerine coated walls are finally trapped in a vacuum cleaner filter
bag at the exit duct.
421
-------
The duct has been operated several times with a manually
operated syringe for injecting the fly ash, but is now equipped with a
vibrator feeder on the Metco flame spray gun used to inject the fly ash.
Another aspect of the operation that should be mentioned is that the flow
of glycerine into the chamber is controlled by a slide valve that is
manually opened to any desired setting at the beginning of the run. For
high viscosity runs, the valve can be fully opened and still permit a
several minute run on a tank of glycerine. A low viscosity run however
will be quite short if the slide valve is completely opened but if it is
only partially opened the glycerine flows in streams over only a portion of
the plastic plates in the duct.
Two further variations of these tests are also underway. One is to
coat the plates with a very viscous hydrocarbon to study the longer time
operation of the duct under conditions in which there would not be a flow of
glycerine (or glass). Another variation is to mount several plastic plates
at a 45 angle to the gas flow such that glycerine flowing over and through
the large number of small holes in the plates would serve as either or both
a bubbler and/or a curtain to perhaps more completely trap the particles.
Test results in the duct to date have been very encouraging with
99% of the injected material trapped in the glycerine and usually mostly
within the first third of the duct. The particles that reached the vacuum
cleaner bag on the exhaust duct were generally less than 2 microns in
diameter.
422
-------
TOTAL POWER GENERATION CYCLE EFFICIENCY STUDY
This aspect of the program is scheduled for later in the year
when more quantitative information is available. However conceptual
designs and operational scenarios are being developed against which to
perform the previously described R & D work.
It is envisioned that the sequence of coal combustion or
gasification to provide gas to drive a gas turbine may utilize a cyclone and
this coal slag based glass process for cleaning the particles from the gas
stream at temperatures of about 1000°C. The glass process is envisioned to
consist of several vertical chambers dn order to provide standby reserve
capacity each)of which would have the gas flow and the glass flow downward
with the gas exiting on the side while the glass could be removed
intermittently through a lock hopper. Means of combining it with the
particulate effluent from the cyclone as well as recovering the sensible heat
from these two sources is contemplated. The combined waste products from
these two clean-up processes could then be either disposed of as a more compact
fill than fly ash or perhaps used in building products as aggregate.
Start-up and operation of the unit are envisioned to be by coating
a large number of square feet of ceramic plates with a fraction of an inch
thick layer of powdered glass which would be fused in place and be
continuously refurbished and replaced primarily by capturing and dissolving
fly ash. If necessary some supplementary glass could be added in the gas
stream or from a lock hopper. It is possible, if not probable, that some local
control of the viscositv of the elass will also be necessarv. It is
423
-------
especially likelv that lowering of the viscosity will be needed. For this,
two methods are possible, one is to add materials that will flux the
glass/fly ash in the unit and the other will be to provide local heating
(primarily electrical) much as is so widely done in the glass industry. The
latter seems most desirable since presumably electrical energy is being
generated in these plants and any usage of energy for this purpose would be
largely recovered in the gas stream. Also the use of electrical energy as
opposed to adding chemicals or low melting glasses for viscosity control
would seem preferable from the standpoint of the potential carry over of low
melting materials into the turbine.
ACKNOWLEDGEMENTS
The authors thank ERDA and especially Mr. William Fedarko who
represents them as the program manager for supporting this work under
contract EF-77-C-01-2608. In addition, Mr. William Laskow of our laboratory
has provided invaluable assistance in preparing and performing the experimental
work.
REFERENCES
1. Dekker, P., "Calculation of Viscosity-Temperature Curves for
Porcelain Enamels from the Flow-Button Test," J.Am.Ceram.Soc.
_48 (6) 319-27 (1965).
2. Plankenhorn, W.J., "Factors Affecting Reproducibility of Flow-
Button Test." J.Am.Ceram.Soc. 31 (12) 338-44 (1948).
424
-------
TABLE I. COMPOSITION IN OXIDE WEIGHT PERCENT OF MATERIALS
EVALUATED FOR FLUIDITY TEMPERATURE RELATIONSHIP
Oxide
Na20
K20
CaO
MgO
A1203
£ Go vJ Q
Amber
Container
SRM-710 Glass**
8.7 13.3
7.7 1.8
11.6 9.8
-
0.18 4.2
0.02 0.4
„ ^ ***
Montana
Rosebud
0.4
0.4
14.9
4.7
21.7
8.0
***
Pittsburgh
#8 K-884
-
1.5
3.3 10.06
0.7 3.3
21.4 21.8
28.3 16.6
Sb203 1.1
P205 - 0.4 - -
Ti02 - 0.008
Si02 70.5 70.6 47.5 44.8 47.8
S03 0.2 0.02 -
100.00 100.12 99.78 100.0 100.1
* Soda-lime-silica glass from NBS
** Typical amber container glass composition
*** Fly ash
**** Synthetic slag developed by NBS-glass Section to represent
intermediate composition between Montana Rosebud and Illinois #6
fly ash materials.
425
-------
in
u
1/1
(0
1200
1300
1400 1500
TEMPERATURE (°t)
UOO
170«
Figure 1A. Viscosity/Temperature Relationship Determined by NBS on Synthetic
Slags Formulated to Represent the Average Compositions of Fly Ash
from Montana Rosebud and Illinois No, 6 Coal Types. The Central
Curve is an average of these Data and may be Considered to
Represent "Typical" U.S. Coal Slag Behavior.
1.0
TYPICAL IVtRAOC COAL SLAG W-IM)
GLYCERINE
I MO
f-IO)
UOO I MO
»> <»>i>:
TEMPERATURE (°C>
1600
1.20)
1700 -» A SLAG
(t») -» O GLYCCRINC
Figure IB. Viscosity/Temperature Relationships for a Typical "Average" Coal
Slag (NBS K-884) and Glycerine which will be Used as a Model System
to Demonstrate Particulate Capture Mechanisms at Low Temperatures.
426
-------
uJ
cc
0°
o
o
0°
o
in
o
O
o
o
o
o
o°
o
o
o
Q
7
U
(D
im
, i
tS
£
•^
oo
CO
U)
<
J
(D
\
o
o
w—
<
o
o
en
m
o
CM
O
co
o
in
CM
in
O
o
co
o
O
o
0
U
o
in
£
u.
o
10
0
(D
0
o
co
I
in
"^
in
CM
I
o
CO
o
CM
-
o
o
-
o
O
1
*
oU
o
in
CM
U
o
o
o
CM
O
°0
o
CM
0°
o
in
CQ
0)
CQ
g
rt
bn o>
JS H
02 T3
O «
^ "5
0) O
w J-t
W 3
Oj O
i— I l-
o
a)
fl
to
43 CQ
0) 0)
1$
P4
-------
f«.
w
* *,
m f
< "fz' cr
, #-, %^* '1_,<
C 4 . ^ .,
«?'
^
"M
'^>^ , ,^».
' '##&?
•
Tfe
Figure 3. Photograph of a plexiglass model of a "sticky-wall" labyrinth
gas scrubber (A) and (B) a close-up view of scrubber plate detail,
428
-------
MOLTEN SALT SCRUBBING FOR REMOVAL OF
PARTICLES AND SULFUR FROM PRODUCER GAS
By:
R. H. Moore, G. F. Schiefelbein, G. E. Stegen, D. G. Ham
Battelle Memorial Institute
Richland, WA 99352
429
-------
MOLTEN SALT SCRUBBING FOR REMOVAL OF
PARTICLES AND SULFUR FROM PRODUCER GAS
By:
R. H. Moore
G. F. Schiefelbein
G. E. Stegen
D. G. Ham
Battelle Memorial Institute
Pacific Northwest Laboratories Division
Richland, WA, 99352
Data are presented which, shows 95 to 99% extraction of
sulfur compounds from producer gas at 1250° to 1500°F has
been achieved using a molten salt as the working fluid in a
vehturi scrubber. The salt was fully regenerable using excess
steam + CO- to reverse the reaction. Efficient removal of
particles was also achieved. Producer gas at a flow of 50 to
75 SCFM was generated in Battelle's fixed bed gasifier
operating on metallurgical coke.
These data together with earlier laboratory data have
enabled design of a Process Demonstration Unit for operation
in fully continuous mode. This PDU is in final phases of
construction. The design basis and a few details of equipment
are presented here.
430
-------
MOLTEN SALT SCRUBBING FOR REMOVAL OF
PARTICLES AND SULFUR FROM PRODUCER GAS
INTRODUCTION
Reversible extraction of H~S from synthetic fuel gas by molten carbonates
was demonstrated in laboratory scale bubble contactors early in 1973. These
tests defined the equilibrium extraction of H^S as a function of temperature
and provided data for subsequent Process Demonstration Unit design. Parallel
corrosion tests disclosed that "alonized" 300 series stainless steels might
be suitable for PDU construction.
On the basis of these experiments it was concluded that a process for hot
fuel gas cleaning could be devised. It would have the following characteristics:
• The process would entail gas-liquid contacting at 500° to 750°C.
• Removal of sulfur compounds, particles, and low volatile tars would
occur.
• The sulfur compound extraction capacity of the salt would be easily
regenerated to yield a regenerator product gas sufficiently high in
H?S to be a suitable Glaus process feed gas.
• Numerous minor impurities (halogens, volatile metals and volatile non-
metals) would be removed from the fuel gas. Ammonia removal, if it
occurred, would occur only as a consequence of dissociation. Water vapor
would not be removed.
Of course, there would be a few problems! Few molten salt processes
with any degree of complexity have been operated. Serious problems have been
431
-------
encountered in the operation of mechanical pumps. This has forced adoption
of processes so simple that no pumps are required. The Atomics International
coal gasification process is an example. In consequence, the process capa-
bility becomes limited.
The hot fuel gas cleaning process employs a venturi scrubber—packed bed
de-entrainer combination. The use of a mechanical pump can be avoided by
operating the venturi in a vertical orientation. Salt fed to the venturi
becomes totally entrained in the gas phase and is carried upward to a
de-entrainer situated well above the level of the venturi feed tank. Upon
de-entrainment the salt returns by gravity induced flow to the feed reservoir.
It only remains to insert a salt regeneration step between the salt reservoir
and the de-entrainer to achieve fully continuous operation in process mode.
This is now being done, but initially, all effort was directed toward design,
construction, and test of a venturi de-entrainer package operating on low Btu
fuel gas produced by Battelle's fixed bed gasifier. Only the extraction
portion of the process has been tested at PDU scale. These tests measured:
• the efficiency of H~S and COS removal
• the efficiency of particle removal
• the effectiveness of salt de-entrainment
• the behavior of materials
• the effectiveness of a vertical venturi scrubber for salt circulation.
To some it may appear the problem of particle removal has been complicated
by introducing a salt dispersion. Certainly it is necessary to separate the
ash-char particles from the salt. This will be done by treating a salt bleed
in a separate aqueous process for salt recovery and recycle. So far as the
gas is concerned, introduction of the salt may facilitate collection of parti-
cles. This is due to the tendency of a liquid to agglomerate and to occlude
the particles in the agglomerating liquid droplets. On the other hand, speci-
fications for alkali metals may be lower than for particles and this may further
complicate the problem.
432
-------
THE VENTURI SCRUBBER-PACKED BED DE-ENTRAINER
The fuel gas will typically be used to fuel a gas turbine, and cleaning
requirements were formulated on this basis. The venturi scrubber was designed
with particle removal efficiency as the design basis. It was assumed that
gas-salt dispersion and mass transfer efficiency would be such as to ensure a
single equilibrium stage for H~S extraction.
The Battelle-Northwest gasifier shown in Figure 1 was originally
designed for steam-air pyrolysis of carbonaceous wastes. It is a ceramic
lined vessel 36 in. I.D. and 12 ft high with a lock hopper above for feed
addition and a fixed grate through which air and steam may enter at its base.
It operated well on coke and is expected to function equally well on non-
caking coal. It will gasify such fuels at rates up to 150 Ib/hr producing
up to 150 SCFM of gas. At the modest rate of 50 to 75 SCFM the gasifier should
operate continuously for 4 to 5 days before shutdown for ash removal becomes
necessary.
At the time design of the venturi scrubber was initiated, nothing was
known about the particle size distribution or particle loading in the gas
produced from coke by this unit. A value of 0.1 lb/1000 SCF was assumed, a
value which proved to be about twice the value measured. With coal, tars and
oil will accompany the particles.
Particles which cause erosion of turbine blades must be larger than about
2.0 microns. This venturi scrubber was designed for 99% removal of particles
4.0 microns or larger. Calculations patterned after those of Seymour Calvert,
et al were used to establish the throat velocity sufficient to achieve this.
The value turned out to be 200 ft/sec at a feed rate of 50 SCFM. For design
conservatism this was increased by 20%.
Figure 2 shows a schematic of the venturi scrubber. The throat diameter
is 1.5 in., the inlet angle is 25° and the exit angle 7°.
433
-------
Considerable thought was given to the design of the salt inlet port.
This venturi is so small no latitude exists to provide a distributor ring
(the holes in the ring would be minute). There was concern about possible
plugging of the inlet port and the decision was reached to provide a single
inlet hole. A scrubber with an inlet of 3/16 in. diameter was used which
allowed a maximum flow of 0.9 gal/min of molten salt under the available head.
The dispersion of salt generated by the atomizing action of the high
velocity gas flowed upward through a heated pipe to a de-entrainer situated
above the salt feed reservoir. The mean droplet size generated by the
venturi at a gas flow of 50 SCFM was calculated from the equation of
(2)
Nukiyama and Tanasawa given in Perry and found to be 116 microns. It
was easy to show that salt droplets of this size would be carried upward
and into the de-entrainer. It was also necessary to show that transfer of
momentum from the gas phase to the film of salt flowing along the pipe wall
would be sufficient to ensure a climbing film. Measurements by Sutey and
(3)
Knudsen on aqueous solutions permitted calculation of the shear stress
in the salt film assuming a simple linear proportionality between the shear
stresses and density for various liquids. The gas velocity required to
produce the calculated shear stress turned out to be 39 ft/sec. For higher
velocities upward flow of the liquid film would occur. For conservatism and
for more economical construction costs (smaller pipe) a design velocity of
100 ft/sec was selected. This led to a required pipe diameter of 2.4 in.
To recover the entrained salt a packed bed de-entrainer was selected.
(4)
Jackson and Calvert have demonstrated >99% collection of 4 micron oil mist
droplets with such beds. A de-entrainer was designed initially for use with
1.5 in. Raschig rings. These proved unobtainable in alumina refractories and
1 in. mullite grinding balls were substituted. These were placed in a bed
12.5 in. in diameter and 25 in. deep. This bed was calculated to have a
flooding velocity of ^1200 ft/min and at 50% of this velocity should have
a removal efficiency of 90% for 3.5 micron particles.
A demister was fabricated from stainless steel screen rolled into coils
about 3/4 in. diameter by 3 in. long. Screen of 50 mesh and 20 mesh was used.
434
-------
A 10 in. deep layer of the rolls of 20 mesh screen was placed above the 1 in.
ceramic balls and covered with a 4 in. deep layer of the 50 mesh rolls. This
arrangement was used during only one run. At other times no demister was used.
All equipment and pipe must be heated to the operating temperature. Trace
heating and furnaces were used as shown in Figure 3. This large pot furnace
heated a 30 gal salt pot from which salt was fed to the venturi. The packed
bed de-entrainer was situated just above this pot and the replacement furnace
elements used for heating are clearly visible.
Figure 4 shows trace heating in the form of tube furnace heating elements
on lines leading to the venturi. The large pot furnace below contains a 30 gal
pot to which the salt can flow in the event gas flow should be interrupted.
This pot can be pressurized to transfer salt back to the upper pot. The com-
pleted PDU configured for extraction in batch mode is shown in Figure 5. Here
insulation is in place and the plant is ready for operation. The gasifier is
at the right, the venturi and de-entrainer at center, and a burner for gas
disposal is at the left. Auxilliaries for gas preheating, aqueous carbonate
scrubbing, and coal conveying are also visible.
EXTRACTION OF SULFUR COMPOUNDS FROM LOW BTU
FUEL GAS AT VARIOUS TEMPERATURES
The extraction section consisting of the venturi scrubber and packed
bed de-entrainer was operated in a series of short runs which confirmed the
laboratory data. Figure 6 shows the chemical reactions of interest here. In
these runs a four component salt mixture was used which was of the composition
shown in Table I. In this salt mixture, CaCO_ is so much more reactive than
any of the accompanying alkali metal carbonates that reaction to form CaS is
almost the sole reaction until most of the CaCO,, is consumed. This is illus-
trated by laboratory data which discloses the temperature dependence of the
reactions of H2S with various molten and solid carbonates (Figure 7). Clearly
CaC03(s) is much more reactive than Na2CO (s) or mixtures of molten alkali
carbonates. Its addition to the latter in amounts of from 15 to 20 mole %
CaCO~ substantially increases the reactivity of the mixture.
435
-------
The schematic flow diagram for operation of the venturi scrubber-packed
bed de-entrainer is illustrated in Figure 8. Hot fuel gas entered the venturi
scrubber where it atomized and dispersed the entering flow of salt and carried
the dispersed salt into the head space of the salt feed tank and into the
packed bed de-entrainer. This was packed to a depth of about 24 in. with 1 in.
balls, but most of the de-entrainment effect achieved occurred in the lower 3 to
4 in. The de-entrained salt returned to the venturi feed tank and the cleaned
gas leaving the top of the de-entrainer passed to a burner for disposal.
Clearly the process operated in batch mode because the extraction of
H«S would cease when the extraction capacity of the available salt was
exceeded. When this occurred the salt would either have to be regenerated
or replaced. The salt could be regenerated by allowing it to flow into a
second pot furnace containing a 30 gal salt pot. Here it could be cooled to,
e.g., 600°C, and treated with a steam-CO mixture introduced via a distributor
ring. The reaction with steam + C0~ is a reversal of the extraction reaction
and was shown in laboratory experiments to proceed readily to 96 to 98% com-
pletion. Here the reaction proceeded at a rate proportional to the rate of
steam addition because the salt containing CaS was in large excess. Time was
taken to regenerate no more than 80% of the theoretical maximum, but there is
no reason to conclude that complete regeneration would not occur. An aqueous
carbonate scrub was used to capture regenerated H S. The resulting NaHS solu-
tion was sewered for disposal.
Three runs of short duration were made using the venturi de-entrainer
extraction section. In the first run (Table II) at 700°C, troubles were
encountered adjusting liquid and gas flows and for a brief period operation of
the venturi was accompanied by flooding. This resulted in loss of about half
of the circulating salt inventory to the lower (regeneration) pot furnace. At
the time, this was not obvious, and it appeared instead that the extraction
capacity of the salt was much less than expected for reasons unknown. The
data in Table II clearly shows the decay in recovery efficiency as the salt
became saturated with sulfide. The unexpected path taken by the salt was dis-
covered following termination of the run.
436
-------
In a second run, this time at 750°C (Table III), operation was normal with
excellent sulfur recovery. During the period from 1600 hours to 1855 hours,
CaCO utilization ranged from 16.7 to 29.5%. The recovery of 94 to 97% of the
sulfur is in agreement with predictions from laboratory data.
In a run at still higher temperature (Table IV) difficulty was experienced
in establishing salt flow during the early part of the run. At such high tem-
peratures, the extraction efficiency of the alkali metal carbonates is fairly
high and the addition of CaCO_ may offer no advantage. CaCO_ tends to dis-
sociate as the temperature is increased forming CaO which has only limited
solubility in the melt. When this solubility is exceeded, precipitation of
CaO occurs accompanied by a marked increase in viscosity of the melt. The
natural tendency to apply heat to lower the viscosity is the wrong thing to
do in this case. Addition of CO is effective. With full salt flow (evidenced
by pressure drop) recovery of sulfur jumped to very high efficiency. The feed
gas sulfur concentration was varied by addition of H«S cylinder gas and the
high H S concentration near the end of the run made it easy to demonstrate 99%
H,.,S recovery.
The addition of supplemental H_S is considered to be responsible for the
anomalous COS behavior. Reaction of added H-S with CO is believed to be the
principal reaction which produces COS. The inlet feed sample is taken just
downstream and around a bend from the point of H_S addition. Even here the
increase in COS is evident in consequence of the variation of its formation
rate with temperature (Table V). Even after entering the venturi scrubber
reaction to form COS continues but as the H S is extracted COS formation slows
and may even reverse. The product sample analysis shows that the final COS
concentration is above the equilibrium value predicted in terms of the final
H~S concentration, but is below the equilibrium value for the inlet H-S con-
centration. This requires that the reaction rate be fast but not so fast that
the reaction rate keeps up with the rate of H?S extraction. It is also required
that the rate of direct COS extraction be less than the rate of H S extraction.
The data indicate that presence of COS in the product gas limits overall sulfur
removal. The solution would appear to be to provide additional residence time
in the extraction section (or additional extraction stages).
437
-------
BEHVAIOR OF PARTICLES AND EFFICIENCY
OF SALT DISPERSION DE-ENTRAINMENT
Two isokineti'c samplers were used to sample inlet and outlet gases for
particles. The inlet sample was collected on a filter, so only total particle
burden was obtained. The gas was also dried by passage through a silica gel
absorption bulb. When this was weighed, the average moisture content of the
gas could be calculated.
The outlet sample was passed through an Anderson-2000 Impinger, so in
addition to total particle burden, the particle size distribution was obtained.
Salt losses could be estimated from analysis of deposits on the plates.
Data obtained from runs at 750°C and >SOO°C are shown in Tables VI and
VII. For the case shown in Table VI, the PDU was operating with no demister.
For the data shown in Table VII, the wire screen demister described earlier
was in place. The demister appears to have reduced the particle burden in the
product gas by approximately a factor of three.
The data of Table VII show a total weight gain of 0.0301 g which corresponds
to 0.0245 grains/SCF or about 40 ppm (by weight) in the product gas. The
appearance of the deposits on the plates of the Anderson head is most informa-
tive. The deposits on plates 2, 3, and 4 are very thin and cast no shadow when
viewed at moderate magnification, e.g., 80 x. On plates 5 through 8, however.
the deposits are piled up in little mounds. These may have been formed by salt
droplets which solidify as they hit. The deposit on plate 8, shown in Figure 9,
is typical. The flat deposit on plate 4 is shown in Figure 10 for comparison.
When viewed at 2000 x under the scanning electron microscope (Figure 11),
the deposit on plate 4 appears as round spherical particles of 4.5 microns diame-
ter. This confirms the calculated cut-off diameter for this plate, but the point
of major interest is the uniformly spherical nature of these particles. These
must surely have been formed by solidification of a liquid.
438
-------
Portions of the candle-shaped mound from plate 8 were transferred to
sticky tape and viewed at 20,000 x under the scanning electron microscope.
Again spherical particles are observed (Figure 12). These are 0.3 to 0.4
microns in diameter, slightly smaller than the calculated cut-off diameter for
this plate.
The scanning electron microscope excites an x-ray fluorescence spectrum
of the elements present in the sample. Figure 13 shows the spectrum of the
deposit on plate number 4. Na is not efficiently detected and elements lighter
in mass than Na are not observed. K and Ca, which are major salt components,
are noted as well as S, which the salt has extracted. In addition Al, P, Si,
Fe, and Zn, which are coke derived, were observed. Some Na, K, and Ca may also
be coke derived. The deposit on plate 8 yields the spectrum shown in Figure 14.
Here Ca, a major salt component is barely detected. Na and K and some S are
evident. A major constituent is P. Si, Cl, Fe, Zn, and Ga are also present.
Comparison of these and other measurements with spectra of the pure starting salt
indicates that on average these deposits are about one third salt with the
balance being material derived from coke. This means the product gases from this
run contained approximately 12 ppm salt. This is slightly more than an order of
magnitude too high to be of acceptable quality for turbine fuel; however, there
are steps which can be taken to reduce salt and particle levels still further so
continued development of this process seems warranted.
DEVELOPMENT OF FULLY CONTINUOUS PROCESS
Operation of the venturi scrubber-packed bed de-entrainer disclosed
some critically important information. It disclosed some severe operating
problems (due to salt freeze-ups) with salts high in CaCO . These would become
serious at temperatures above 750°C. No serious corrosion of the alonized
steel occurred, though the exposure history was short (at most 150 hours at
temperatures >700°C). Some erosion occurred in the throat of the venturi
scrubber and in the future, the shape of the venturi scrubber will be defined
by a ceramic liner (high purity, high density Al-0 ) which will fit inside an
j— O
alonized steel shell. Finally, the data disclosed that regeneration of salts
439
-------
containing CaCO«-CaS might require high steam consumption and might lead to
a regenerant gas too lean in H_S for efficient conversion to S in a Glaus
process. Processes other than the Glaus could, of course, be used.
These considerations led to design of a process which utilized the alkali
carbonate eutectic with no deliberate CaCO,, additions. Small amounts of CaCO~
which enter from coal ash will cause no problem. A vertically oriented venturi
scrubber is again to be used to circulate the salt. As shown in Figure 15, the
salt dispersion will be carried upward to a de-entrainer where the flow of
salt is split. Most of the salt recycles to the venturi feed tank. A smaller
amount containing sulfur in an amount roughly equivalent to that in the feed
gas is cooled and passed through a regeneration column countercurrent to a
flow of steam + C0_. The regenerator is a 6 in. diameter column containing
eight bubble cap trays with two caps per tray. It will operate at 500° to
600°C. It is designed to produce a CO -H S mixture, following steam removal,
containing 30% H S (from fuel gas of 1.3 v/o H S).
Salt fed to the regeneration column is so rich in sulfur (in order to
meet the requirements of the Glaus process) that it is not possible to produce
a fuel gas adequately low in sulfur in a single equilibrium extraction stage.
The regenerated salt will flow to the salt make-up tank, here it will mix with
fresh salt entering sporadically as bleed make-up, and then flow to the top
tray of a two tray, 10 in. diameter bubble cap column. (The trays in this
column contain 4 bubble caps.) On this top tray, the sulfur in the salt and
in the 'product gas are in equilibrium. The level to which sulfur can be
reduced by these additional extraction stages will be determined by the effi-
ciency of regeneration. Removal efficiencies exceeding 95% with a maximum of
98% are anticipated.
The packed bed de-entrainer is situated above the top tray of the 10 in.
extraction column. It is packed with one layer of 1 in. balls and 15 in. of
Al?0 catalyst support cylinders 5/8 in. in diameter by 5/8 in. long containing
£• O
a central 1/4 in. diameter hole. Above this packing, there is a volume 12.5 in.
in diameter by 25 in. high for which the following is planned:
440
-------
1) A thin bed, 2 to 3 in. in thickness, of -4 +8 mesh activated carbon will
be placed above the Al.,0,, cylinders. This bed will partly fludize and
will both absorb submicron salt droplets and will trap them by wake effects
and/or inertial impaction. Other absorbants such as Al~0 and SiO~ may
be tried.
2) Alternatively, this space will be filled with a wire demister of fine
copper wire. Wires of other material may also be tried.
3) An experimental material has been furnished by 3M Company from which a
filter of ceramic felt can be fabricated. Other filter materials such
as graphite felt and cloth may also be tried. Quartz wool may be
feasible.
Figure 16 shows a view of the supporting structure for the new PDU.
The gasifier with its drag chain conveyor and lock hopper appears at the
right. The burner and stack are visible at left-center. The storm cap on
the stack is 50 ft above the concrete pad.
Figure 17 is a view showing the 10 in. extraction column and packed bed
de-entrainer assembly. Also visible are the 2.5 in. venturi off-gas line and
the 4 in. recycle line. The coarse de-entrainer is barely visible. The upper
level grating is 31 ft above grade.
It is planned to have the remodeled PDU operating by October 1, 1977. It
is the intention to demonstrate continuous operation over a period of 5 to
6 weeks. This will provide data necessary for economic and engineering
evaluation of this process and for a management decision to proceed with its
future development.
441
-------
ACKNOWLEDGEMENTS
The authors recognize a debt of gratitude for major assistance in
early design phases involving the efforts of C. H. Allen and Dr. Arlin Postma.
Additionally, the contributions of R. F. Maness in selection of materials and
corrosion testing is gratefully acknowledged. More recently D. H. Mitchell and
Dr. R. J. Robertus have played major roles in remodeled PDU design and
construction and in process modeling.
This work was initiated with support from the Office of Coal Research
under Contract 14-32-0001-1519. Continuing support by ERDA Fossil Energy's
•
Division of Coal Conversion and Utilization and particularly the support of
E. (Zeke) Clark and Dr. Michael Gurevich is gratefully acknowledged.
REFERENCES
S. Calvert, et al., "Scrubber Handbook," Report for EPA Contract
CPA-70-95, July 1972, A.P.T. Inc., Riverside, CA, 92502.
J. H. Perry, Chemical Engineers Handbook, 4th Edition, Chapter 20,
McGraw-Hill Book Company, New York, NY, 1963.
A. M. Sutey and J. G. Knudsen, "Mass Transfer at the Solid-Liquid
Interface for Climbing Film Flow in an Annular Duct," AICHE Journal,
15_:719-726, 1969.
S. Jackson and S. Clavert, "Entrained Particle Collection in Packed
Beds," AICHE Journal, 12:1075-1078, 1966.
442
-------
TABLE I. Salt Composition Used in PDU Demonstration Runs
Component Mole %
CaC03
3
3
18.0
37.3
29.6
15.1
TABLE II. Data Summary Molten Salt
Time
1533
1551
1640
1720
1755
Gas
Flow
SCFM
48
55
60
48
50
Salt
Temp.
°C
700
700
700
700
700
Gas
Temp.
°C
675
525
525
500
500
Gas
Press.
psig
0.9
1.0
1.0
1.0
0.9
v/o
H2S
(in)
0.84
0.84
0.84
0.84
0.84
Weight %
13.0
36.0
37.3
13.8
Pilot Plant Run
ppm
H2S
(out)
5
5
5
840
0.57%
v/o
COS
(in)
0.021
0.021
0.021
0.021
0.021
A
Number 2
v/o
COS
(out)
0.016
0.1
0.049
0.136
0.104
7
/o
Rec.
S
98.9
93.3
96.7
83.4
25.9
* Nominal gas flow 50 SCFM; nominal salt flow 0.9 gal/min. During this
run partial loss of salt to the lower tank and to piping up stream of
this lower tank occurred which reduced the circulating salt inventory.
This caused breakthrough of sulfur compounds earlier than anticipated.
443
-------
TABLE III. Data Summary Molten Salt Pilot Plant Run Number 3
Time
*
1552
1605*
1617
1650
1715
1740
1805
*
1835
Gas
Flow
SCFM
71
68
68
67
67
70
67
67
Salt
Temp.
°C
750
750
750
750
750
750
750
750
Gas
Temp.
°C
750
750
750
750
750
750
750
750
Gas
Press.
psig
0.5
0.5
0.5
1.8
2.3
2.3
3.05
2.3
v/o
H2S
(in)
0.35
0.35
0.35
0.38
0.38
0.45
0.40
0.41
ppm
H2S
(out)
101
103
25
11
10.5
11
13.5
63.5
v/o
COS
(in)
0.026
0.026
0.026
0.026
0.023
0.023
0.026
0.026
v/o
COS
(out)
0.068
0.08
0.04
0.016
0.018
0.024
0.0285
0.05
%
Rec.
S
A
86.6
*
84.7
93.1
97.3
97.1
96.8
95.8
*
91.8
Nupro valve open only 2 turns. All other times valve was fully open.
At fully open position, salt flow may be as much as 0.9 gal/min. At
2 turns, salt flow is estimated to be <0.5 gal/min.
TABLE IV.
Time
1527
1542
1612
1625
1635
1708
1730
1745
1820
Gas
Flow
SCFM
70
70
74
74
74
74
75
60
55
Salt
Temp.
°C
813
823
860
840
830
816
816
816
816
Data Summary Molten Salt
Gas
Temp.
°C
750
750
750
750
750
750
750
750
750
Gas
Press.
psig
0.6
0.6
0.4
0.4
0.4
1.6
1.8
1.5
1.4
v/o
H2S
(in)
0.425
0.425
0.530
0.530
0.530
0.530
0.685
1.30
1.38
Pilot Plant Run Number
v/o
H2S
(out)
0.060
0.060
0.275
0.270
0.285
0.00595
0.0145
0.0136
0.0029
v/o
COS
(in)
0.040
0.040
0.054
0.054
0.054
0.054
0.038
0.061
0.019
v/o
COS
(out)
0.094
0.092
0.125
0.117
0.120
0.0395
0.0085
0.0075
0.0125
*
6
%
Rec.
S
74.3
74.9
38.3
40.0
37.1
94.9
97.4
98.7
99.3
From 1445 until 1708 attempts were being made to establish salt flow.
Apparently a very slight, perhaps sporadic flow existed from 1527 to
1700 hours, but at 1700 hours, a marked increase in pressure accompanied
by increased recovery efficiency gave definite evidence of flow at near
minimum rate M3.9 gal/min.
444
-------
TABLE V. Variation of Inlet COS Contentration
with Temperature Due to the Reaction:
H2S + CO +
Temp.
°C
700
750
816
840
: COS + H2
Inlet COS
0.021
0.026
0.043
0.054
TABLE VI. Particle Behavior; Molten Salt Pilot Plant Run Number 3
Inlet Gas; Total Particle Burden = 0.109 Grains/SCF, % HO =9.1
*
Outlet Gas; Total Particle Burden = 0.0772 Grains/SCF,
Anderson Head Particle Size Distribution
% H20 = 4.5
Plate
No.
1
2
3
4
5
6
7
8
Filter
Total
Wt. Gain
g
0.0048
0.0042
0.0327
0.0498
0.0405
0.0381
0.0289
0.0314
0.0948
0.3253
Calculated
Effective
Cut-off Dia, y
17.6
11.2
7.5
5.3
3.5
1.7
1.0
0.7
0.15 EST
%
1.5
1.3
10.1
15.3
12.5
11.7
8.9
9.7
29.2
Concentration
in Gas
Grains/SCF
0.00114
0.00099
0.00774
0.0118
0.0096
0.0091
0.0069
0.0075
0.0224
Westinghouse
Spec.
0
0
0.001
0.75
* Appearance of deposits on the plates of the Anderson head is indicative
of substantial salt contamination (especially on plates 5, 6, 7 and 8).
445
-------
TABLE VII. Particle Behavior; Molten Salt Pilot Plant Run Number 6
Inlet Gas;
Outlet Gas;
Total Particle Burden = 0.074 Grains/SCF, % H20 =11.2
Plate
No.
1
2
3
4
5
6
7
8
Filter
Total
Total Particle Burden = 0.0245 Grains/SCF,
Anderson Head Particle Size Distribution
H20 = 3.3
Wt. Gain
g
0.0015
0.0013
0.0016
0.0019
0.0026
0.0029
0.0030
0.0047
0.0106
0.0301
Calculated
Effective
Cut-off Dia. yi %_
20.1 5.0
10.5 4.3
7.0 5.3
5.0 6.3
3.1 8.6
1.6 9.6
1.0 10.0
0.6 15.6
0.15 EST 35.2
Concentration
in Gas
Grains/SCF
0.0012
0.0011
0.0013
0.0015
0.0021
0.0021
0.0024
0.0038
0.0086
Westinghouse
Spec.
0
0
0
0.001
0.75
446
-------
Figure 1. The Battelle Fixed Bed Gasifier During
Early Construction of the PDU
447
-------
448
-------
Figure 3. Packed Bed De-entrainer Assembly Showing
Installation of Trace Heaters
449
-------
t-l
4J
C
Q)
Q)
C
O
O)
4-1
CO
0)
PC
O)
O
(fl
1-1
H
cfl
to
C
-3-
-------
Figure 5. Completed Process Demonstration Unit
(Configured in Batch Operating Mode)
451
-------
r-" CM
5?
O
CO
o
CJ
o
o
L -2
O) CM
" 8
CM
T3
C
n)
CO
CN1
£,
jj
•H
5 o
H n
3 S
» o
CM W
z u
+ +
n
O O
o
(0
O
O
-------
1000
100
10
1.0
0.1
DATAOFROSENQVIST,
CaCO, (S)
\
\
\
\
EXPERIMENTAL DATA
RUNS 5, 6, AND 7
QUATERNARY MIXED CARBONATES
\
KOSZEGI -ROSEN
NaC0 (s)
RUN 4
(Li, Na, K)2CO
\
0.01
0.8
0.9
1.0
1.1
°
1/TK x 10
1.2
3
L3
1.4
Figure 7. Temperature Dependence of the Equilibrium
Reac
453
Constants for Reaction with H_S
-------
HI
O
O
O
OC
D
OC
UJ
Q.
0
OJ
CO
Cfl
I
Q)
td
o
3
to
DO
0)
(-1
60
-------
Figure 9. Deposit on Plate Number 8 at 80 X Magnification
455
-------
Figure 10. Composite Photo Showing Deposit on
Plate Number 4 at 80 X Magnification
456
-------
»*%*
tan»«
v >r
%
Figure 11. Scanning Electron Microscope Photo of
Deposit on Plate Number 4 (2000 X)
457
-------
Figure 12. Scanning Electron Microscope Image of Deposit Lifted
from Plate Number 8 Using Sticky Tape (20,000 X)
458
-------
-------
- to-
.S
/
J
.
\
^
^^
™
00
S-SJ
•" N
2 "-
CM (i)
*" N
00
§-£
n «
rs
CO m
N co o
a-*."
s
e>
r^
M **.
«, « ^
S-2
» 52
;_J^§
10 N
I0_§ «
9>
f.
r jj
oo
QJ
nj
I
P-i
c
o
4J
•H
OD
0
O
14-1
0
6
3
4-1
a
ft
w
0)
o
c
1)
en
0)
o
3
i— I
P£-I
^~-i
Pi
1
XJ
-3-
i— H
-------
n
TO HOT AIR EXHAUST
-M—•
TAIR
TO BURNER
H2S-H20-C02
EXHAUST TO BURNER
* M-
*«— CO
r
2
>SALT
BLEED
STEAM
-M-
Figure 15. Process Flow Diagram
461
-------
i .
§
0)
iH
0)
T3
O
e
0)
Pi
C
0)
O
0)
J-4
4J
O
(-1
4J
in
O
ex
0)
n
3
M
•H
462
-------
Figure 17. Support Structure Showing Preliminary
Equipment Installation
463
-------
SESSION IV:
PARTICLE SAMPLING AND MEASUREMENT
JOHN GEFFKIN
CHAIRMAN
465
-------
PARTICULATE SIZING IN HIGH-TEMPERATURE
HIGH-PRESSURE COMBUSTION SYSTEMS
By:
H. W. Coleman
Sandia Laboratories
Livermore, CA 94550
467
-------
PARTICULATE SIZING IN HIGH-TEMPERATURE,
HIGH-PRESSURE COMBUSTION SYSTEMS
By:
Hugh W. Coleman
Combustion Sciences Department
Sandia Laboratories, Livermore, CA 94550
The results of a review of particulate sizing techniques
for application to turbine inlet flows in advanced, open-
cycle gas turbine systems fired with pulverized coal, liquid
and gaseous fuels derived from coal, and heavy residual fuels
are discussed. Emphasis of the study was on optical tech-
niques, which provide nonperturbing, •Ln. £
-------
PARTICULATE SIZING IN HIGH-TEMPERATURE,
HIGH-PRESSURE COMBUSTION SYSTEMS
SECTION 1
INTRODUCTION
A review1 of diagnostic techniques for application to turbine inlet flows
in advanced, open-cycle gas turbine systems fired with pulverized coal, liquid
and gaseous fuels derived from coal, and heavy residual fuels was recently
completed. This review was part of the initial phase of the Diagnostics
Assessment Program conducted by the Combustion Sciences Department at Sandia
Laboratories and sponsored by the Advanced Power Systems Branch of ERDA/Fossil
Energy. The Interim Report discussed the diagnostic requirements for
advanced, open-cycle gas turbine systems and presented an assessment of
current techniques for measurement of particulate size distributions and
loading densities, velocity, temperature, and species concentrations in
"dirty" flows. Emphasis of the study was on optical techniques, which
provide nonperturbing, -Ln &-LX.U. measurements. The present paper discusses the
results of the study which pertain to measurement of particulate size distri-
butions and mass loading densities.
Diagnostic needs and constraints on the measurement systems will vary as
the combustor/turbine program progresses. During component development,
reasonable spatial resolution and accuracy are required, but rapid time
response is not necessarily essential. During prototype evaluation, spatial
and temporal resolution, accuracy, reliability and long lifetimes are
necessary. Measurements required later for on-line system monitoring and
control are not yet defined. During this phase, one would expect that less
complex diagnostic systems than those required for prototype evaluation will
be sufficient; but reliability, long lifetimes, and rapid time response will
be required.
469
-------
In the past, most particulate sizing research programs have been aimed
primarily at characterization of atmospheric aerosols, sprays from nozzles
(droplets in the 10 to 100 microns range), and particulates in various exhausts
in the size range to be respirable health hazards (submicron to five microns).
In general, the environments of interest in these programs are much less
hostile than the high-temperature, high-pressure combustion systems being
considered here.
The particulate sizes of interest in combustor exhausts (immediately
upstream of the turbine inlet) will range from submicron to a maximum of 5 to
50 microns. The smaller particles are of interest from a health hazard stand-
point, while the upper limit on size will be determined by blade erosion and
impaction considerations. This upper size limit, together with permissible
loading densities, will be dictated by the turbine design.
470
-------
SECTION 2
MEASUREMENT OF PARTICULATE SIZE DISTRIBUTION
Conventional techniques (and the commercial instruments utilizing them)2'3
as a general rule require sample withdrawal by a physical probe with subsequent
dilution and/or cooling of the sample prior to sample introduction into the
particulate sizing instrument. Insertion of the probe into the flow may have
significant effects on the flow field in the vicinity of the probe. (For the
high-temperature, particulate-laden and highly oxidative flows expected in
advanced gas turbine/combustor systems, water-cooling of the probe is required
—which leads to a fairly bulky sampling system.) Furthermore, dilution and/or
cooling of the sample may induce substantial particle size changes in the
sample due to agglomeration and condensation of gaseous species, in particular,
water. Cold traps may reduce the magnitude of such errors, but particle size
distribution estimates from cooled samples are not expected to be representa-
tive of the hot gas flow. It was concluded that there are presently no
commercially available instruments for reliable measurement of particulate
sizes in high-temperature, high-pressure flows containing a polydisperse system
of particles of unknown shape, index of refraction and composition.
There is, however, a sizable amount of research underway on various
laser-based optical techniques for particulate sizing, and several of these
appear promising for application to the present problem. These techniques are
nonperturbing in that no physical probe is inserted into the flow; however,
they do require optical access to the flow. The problems associated with
providing adequate optical access are not explicitly considered in the
following discussion.
The optical particulate sizing methods can be broadly categorized into
imaging (photography, holography) and scattered light techniques. The range of
applicability of the second category of techniques considered is related
471
-------
directly to the physics of light scattering processes. In order to discuss
these range limitations it is useful to briefly consider the Mie theory of
light scattering from spherical particles. Then various optical techniques
for particle sizing and their application to the present problem will be
examined.
MIE THEORY
Descriptions of the Mie theory of light scattering are given in the books
by van de Hulst and Kerker.5 Mie theory is an exact solution of Maxwell's
equations for the scattering of electromagnetic radiation. In the discussion
below, which will be restricted to spherical particles, the following defini-
tions will be required:
A = wavelength of incident light
m = n - ik = complex refractive index of the scatterer
d = diameter of scatterer
a = ird/A = optical size of scatterer
I = intensity of light
r = distance from scatterer to detector
The geometry to be used for the discussion of light scattering is shown
in Figure 1. The linearly polarized incident light wave propagates along the
z axis in the positive direction with its electric vector, E, parallel to the
x axis and its magnetic vector, H, parallel to the y axis. The plane of
polarization of the incident light is designated as the plane containing the
electric vector and the propagation vector (the xz plane). The scattering
direction, defined by r in the figure, is taken from the origin of the scatter-
er to the point of observation. The angle 6 is measured from the forward
direction (direction of propagation) to the direction of observation (defined
by r), and the angle
-------
I = (A2/4ff2r2) Sj^Ce, m, a) 2 sin2 <}> (1)
I||= (A2/4^2r2) S0(C, m, a) 2 cos2 <|> (2)
The S and S? functions are the complex amplitude functions calculable using
Mie theory. Particles of general shape require calculation of two additional
amplitude functions.
If an experiment is arranged such that 0, so that m is
complex) exhibit less intensity oscillation with 9, so that the curve of inten-
sity versus 0 is smoother. The variations of Ii and Iii with 0 and a for a
8
nonabsorbing sphere are shown in Figure 2 as an example of scattered light
patterns.
473
-------
For the case of multiple scatterers in a region, the scattered intensities
from each of the particles add in the scalar sense to give the total observed
scattering if the effects of multiple scattering (secondary scattering of the
originally scattered light) are small. As particle loading densities increase,
multiple scattering effects cause increasing deviation from the summation of
individual scattered intensities.
Shape effects on the scattering from nonspherical particles are minimized
if the resultant scattering is from a large number of particles with random
orientation. For example, the Fraunhofer diffraction patterns from individual
nonspherical particles (with a large) vary significantly with shape, but the
resulting forward diffraction lobe for a collection of nonspherical, randomly
oriented particles does not differ greatly from that for a collection of
spheres equal in projected area to the collection of irregular particles.7
DISCUSSION OF TECHNIQUES
The discussion of individual techniques based on various characteristics
of light scattering will not cover all the methods of this class which have
been proposed. A comprehensive discussion of all such work reported in the
literature would fill a volume by itself. The emphasis in this discussion will
be on the methods which have potential application to a high-temperature, high-
pressure flow containing particulates of unknown refractive indices and which
have a polydisperse distribution.
Several techniques should be mentioned in addition to those to be
discussed in detail. The first of these is the light extinction (or
transmittance) technique, which is reviewed in detail by Hodkinson. The basis
of the method is to measure the percent of the incident light scattered and
absorbed or, equivalently, the percent of the incident light transmitted
through a volume containing scatterers. According to Hodkinson, for a poly-
disperse system with particle sizes larger than ^ 1 micron (for visible light)
the total area projected by the particles can usually be determined without
prior knowledge of size, shape or refractive index. The mean particle size
(but not size distribution) can be calculated if particle number density or
mass concentration is known. If the refractive index of the particles is known,
474
-------
both mean particle size and number density can be determined from transmittance
measurements at several wavelengths.
Another approach for large particles (d > 2-5 y) is relating the amplitude
of the scattered signal to particle size. Yule, et al.,9 relate particle
diameter to forward-scattered signal amplitude from a particle traversing the
measuring volume of a crossed-beam laser Doppler anemometer as a function of
collection angle (aperture size). This method requires knowledge of refractive
index. Durst and Umhauer10 use a white light source and 90° collection in a
method relating signal amplitude to particle diameter. They report excellent
comparison of measured size distributions of particles of known refractive
index using the optical method and electron microscope measurements. Their
results extended down to the submicron diameter range.
Penner, et al.,11'12 have reported results from recent work, both theoret-
ical and experimental, relating the frequency distribution of measured
photocurrent power spectra to particle size. This approach, with further
development, may have potential over a fairly wide range of particle sizes.
Specific classes of methods discussed in more detail below are photography,
holography, visibility, ratios of angular intensities, and diffraction.
Photography
This is a fairly standard technique using either a spark or pulsed laser
beam to illuminate the system of interest, as in Figure 3. The collection
system can be designed to magnify the particle images if necessary before
transmission to the recording device. Images may be recorded on film or by a
television-type camera with a videotape recorder. Commercial instruments are
available for semi-automatic or automatic analysis of particle sizes from
recorded images. Cadle provides a description of several of these systems.
An indication of the lower size resolution limit for photographic systems
can be obtained from the Rayleigh criterion
d . = (1.22 A f)/D (3)
mm
where f and D are the focal length and diameter of the collecting^lens,
respectively. For an f/2 lens and taking X = 0.5 y, d . =1.22 y. This value
475
-------
is for an ideal optical system. It is generally agreed that the lower limit on
resolution using visible light is in the d _> 2-5 y range.
The maximum illumination pulse width for a particle of known size and
velocity can be determined as follows: if a permissible movement of the
particle is taken as, say, 10% of its diameter during the pulse duration, then
a 10 y diameter particle moving at 50 m/s would require a pulse of maximum
duration equal 20 ns. The energy per pulse required is determined by the
optical arrangement and the sensitivity of the recording system.
Disadvantages of the photography approach for the present application are
its limited sensitivity to small particles, the low f/D collecting lenses
required, and the narrow depth-of-field when large magnifications are used.
Advantages are its insensitivity to refractive index, applicability to
relatively heavy particle-loading densities, relative simplicity of the optical
arrangement, and relatively unambiguous determination of size and shape for
large particles. This technique can be applied in a "real time" mode by use of
a TV-type camera, subsequent digitization of the output signal, and interfacing
with a properly programmed minicomputer.
Holography
A description of laser holography applied to particle size measurement is
contained in the recent review article by McCreath and Beer.1"1 A schematic of
a holography system is shown in Figure 4. A hologram is formed by splitting a
laser beam into a beam which illuminates the object and a reference beam which
is reflected with phase unchanged. Both beams fall on the holographic emulsion,
with the phase difference between them (caused by the object) forming an inter-
ference pattern on the emulsion. The image is reconstructed by illuminating
the hologram. A wave pattern formed behind the hologram appears as an image
of the original object.
Advantages of holography in par tide-sizingllf are that there is little
depth-of-field restriction and that upon projection of a hologram the magnifica-
tion of particles separated along the original optical axis is uniform.
Holography has been used to measure particle sizes down to the 5-10 y diameter
range. The pulse width requirement for moving particles is the same as that
presented earlier in the photography discussion.
476
-------
Disadvantages of the holography technique are its complexity (as compared
with photography), the time required to obtain size information (as compared
with potential real-time operation of the photography method), and the limita-
tion to large particles.
Visibility
The visibility method relates particle diameter to the degree of modula-
tion of the Doppler-shifted signal from the particle as it traverses a measure-
ment volume formed by two laser beams which intersect with a small angle
between them. The frequency information in the signal used to determine
visibility can be processed by the electronics of a laser Doppler velocimeter
(LDV) system to determine particle velocity.
The system used to provide visibility information is shown schematically
in Figure 5. A laser beam of wavelength A is split into two equal intensity
beams, which are focused and intersect at angle IJJ to form a measurement volume.
Interference of the two coherent beams forms a fringe pattern in the volume
with fringe spacing 6 given by
6 = X/(2 sin i]V2) (4)
As the particle moves through the fringe pattern, a modulated signal results as
illustrated in Figure 5. The visibility of the signal is defined as
I - I .
_ max mm , r\
I + I . U;
max mm
As the technique was originally formulated by Farmer,15 the intersection
angle ijj was small (1-2 ) and the scattered light collected either by on-axis
backscatter or forward scatter. By assuming the particle was spherical,
located at the center of the measurement volume, and using approximations
accurate for d/6 _< 2, Farmer obtained a closed-form solution for the visibility
as
V = 2J1(7T d/6)/(TT d/6) (6)
where J (IT d/6) is a first-order Bessel function. This solution is plotted in
Figure 6. As represented by Equation 6, the visibility is independent of index
of refraction. It is also evident from Figure 6 that particle diameter is an
unambiguous (single-valued) function of visibility only for V ^ 0.2.
477
-------
Since Farmer's original proposal, there has been much additional research,
both experimental and analytical, on the technique. Orloff, et al.,16 and
Yule, et al.,9 have experimentally observed much higher values of visibility
for d/6 > 1 than Equation 6 would predict. Durst and Eliasson8 have published
theoretical calculations which show the same result. Both Robinson and
17 1 ft *
Chu and Hong and Jones have published analytical results showing visibility
to be a function of aperture size. In addition, it was shown18 that visibility
also has a strong dependence on refractive index under certain conditions.
In a recent paper, Roberds19 reported both analytical and experimental
results based on the visibility technique. For large spherical particles
(d » X) and carefully designed collection optics geometry, he concluded that
the visibility method used in the forward scatter configuration could be used
to reliably size single particles. He also concluded that backscattered light
was unreliable for sizing by this method.
A particulate sizing system based on the visibility technique and designed
by Spectron Development Laboratories was recently used to determine particulate
sizes in the flue gas from a fluidized bed combustor at Argonne National
Laboratory. This work is discussed in another paper at this symposium.
The visibility technique is dependent on having only one particle in the
measuring volume at any one time. For measuring volumes with dimensions
typical of most LDV systems, this translates to particle loading densities of
/ C Q
the order of 10 -10 particles/cm . However, one must be careful in using
particle loading density figures without considering the corresponding distri-
bution of sizes. In combustion systems where small particles (d < .1-.2 u) are
present in addition to larger particles (d > 1-2 y), signals from the smaller
particles in the measuring volume will contribute to the background noise when
compared with the Doppler signals from the large particles. The effect of this
background can be eliminated or minimized in many cases by careful signal
analysis and system design.
Ratios of Forward-Scattered Intensities
This method, first proposed by Hodkinson,21 uses the ratio of the
intensities of light scattered at different angles from the forward direction
to determine particle size. A laser beam is focused into the flow system, and
478
-------
the scattered light is collected at desired angles from the forward direction
by use of apertures and light pipes, lens and mirror arrangements, or an array
of photodetectors.
In further developing this technique, Gravatt22 presented results of Mie
calculations for spherical particles over a refractive index range of 1.33 to
2.0 for the real component and 0.0 to 4.0 for the imaginary component.
Figure 7 shows results he obtained for I 0/I,-o (where I. is the intensity of
light scattered at angle 0 from the incident beam), with unpolarized incident
radiation and no detector aperture assumed. The error envelope includes all
ranges of refractive indices used in the calculations. Gravatt suggested a
cut-off based on intensity at ot = 19 for Iino/I[-o measurements to avoid the
multiple-valued function problem at larger a shown in the figure. Based on his
calibrations and experimental results with a variety of particulate materials,
he concluded the technique was relatively insensitive to particle shape and
that within the range of applicability of the method the total sizing error was
no greater than 20%.
Hirleman, et al.,23 extended the ratio technique by using two ratios,
l-1-.o/I^o and I,0/Inoj in order to avoid the ambiguity caused by the increase in
12 o b j
ratio as a becomes large. They developed a data acquisition scheme which
allowed real-time particle size analysis in automobile exhausts.
Results of Mie theory calculations by the author for spheres with a
refractive index typical of soot (m = 1.57 - 0.56 i) are shown in Figure 8.
These calculations assume no collection aperture and incident radiation
linearly polarized perpendicular to the scattering plane. The smoothness of
the curves is characteristic of absorbing particles. If incident radiation
with X = 0.488 u is assumed, an a of 30 corresponds to a particle diameter of
4.7 u. Figure 8 shows that by measuring intensity ratios between I,0/I 0 and
I -0/1,0, particles with diameters between approximately 0.6 }J and 4.7 JJ could
be sized.
Hirleman2** pointed out a significant source of error which affects the
ratio technique and possibly other light scattering methods as well. The
sensitive volume from which valid signals are obtained is a function of both
particle size and index of refraction. Calculations2** of the ratio of the
2
sensitive area A at the beam focal point to the 1/e area are shown
s
479
-------
in Figure 9 versus particle diameter for two values of index of refraction
assuming 12° scattering. If this effect is not considered carefully in both
system design and data reduction, the resulting uncertainties can easily
render the data worthless.
Advantages of the ratio technique are its relative insensitivity to index
of refraction (see preceding paragraph, however), its apparent low sensitivity
to particle shape (though this aspect needs more definition), and the
acquisition of real-time information. A disadvantage is its limitation to
loading densities low enough for acquisition of single-particle signals,
though the comments made in this respect about the visibility method also
apply to the ratio technique.
Diffraction
This technique determines the size distribution of particles within the
cylindrical volume formed by a laser beam passing through a flow system from
the total Fraunhofer diffraction pattern from the particles within the volume.
The size distribution can be calculated by matrix inversion or a least squares
determination of parameters in an assumed size distribution law. The method
is described in the review by McCreath and Beer, and results using the
technique were reported by Swithenbank, et al.25 An instrument based on this
principle and built by Leeds and Northrup was described by Wertheimer.26
Application of this instrument to the fluidized bed combustor at Argonne
National Laboratory is reported in other papers at this symposium.
Using visible laser light, the method is applicable for particles of
diameter greater than about 2 U. Movement of the particles does not alter the
diffraction pattern, since the collection lens focuses parallel (undiffracted)
light on the axis and a given diffraction angle results in a constant radial
displacement in the focal plane. Size determination in real time is possible
by using an array of photo-detectors in the focal plane to measure the
diffracted light energy distribution. Signals from the detector array can be
processed by a minicomputer to provide size distributions.
Advantages of the method are its relatively simple optical arrangement,
real-time operation, insensitivity to particle shape if the particles are
randomly oriented, independence of refractive index and applicability to
480
-------
relatively high loading densities. Disadvantages are the limitation to
particles greater than about 2 y in diameter and the large measurement
volume—it is a line-of-sight rather than point measurement technique.
O -J
Self, et al., have reported on a sizing system under development using
the diffraction pattern from single particles. The system uses an expanded,
truncated laser beam and cylindrical lenses to provide an almost uniform
intensity in the measurement volume. Collection of scattered light at 90° in
a gating-discrimination channel assures a well-defined measurement volume
size. The system has been calibrated with pinholes and polystyrene spheres.
This method appears very promising for single-particle measurements for
diameters greater than approximately 2 p.
481
-------
SECTION 3
MEASUREMENT OF PARTICULATE MASS LOADING DENSITY
As discussed by Coleman, et al.,1 both the absolute value of mass loading
density and its spatial distribution at the turbine inlet are critical factors
in the operation of a coal combustor/gas turbine system.
It seems practical to use optical techniques to measure particulate number
density in given size ranges. Number densities could be determined by using,
say, the single-particle measurements of particle size distribution and the
velocity measurements from an LDV system. By knowing the volumetric gas flow
rate through the measurement volume and the number of single-particle signals
per unit time, the number density within a given size range could be calculated.
The variation of true measurement volume size with particle size23*21* would
have to be taken into account in this procedure.
However, the conversion of number density and size distribution measure-
ments into mass loading density values is of questionable validity in the type
of flow systems under consideration. Fly ash and soot particles may be porous
9 ft
or hollow, and agglomerates will usually contain voids. The photographs in
Figure 10 are a vivid demonstration of the physical characteristics one may
observe in fly ash samples. Similar micrographs have been reported by Fisher,
et al.29 Therefore, a knowledge of the material density and composition of the
particles is insufficient in general for an accurate conversion to mass loading
density.
It appears that the only presently available method of determining mass
loading density for these systems is collection of a sample of the flow with
subsequent determination of the mass of particulates present. The collection
may be with an isokinetic probe used also for species concentration sampling or
with a probe designed especially for collection of particulate matter.
482
-------
SECTION 4
SUMMARY
No technique or instrument is currently available which will meet all the
requirements for measurement of particulate size distribution and/or mass
loading density in the systems under discussion. The photography, holography,
visibility and diffraction techniques show promise for particulate size
measurements over a fairly wide range of flow conditions and for particulate
diameters of 2-5 microns or greater. The intensity ratios technique has
potential, with further research, for measurement of particulate sizes in the
0.2-5 micron range in certain flow conditions of interest. A probe sampling
system appears to be necessary for determination of particulate mass loading
density and its distribution in the flow system.
483
-------
REFERENCES
1. H. W. Coleman, et al., "Interim Report: Diagnostics Assessment for
Advanced Power Systems," Sandia Laboratories, Livermore, Report
SAND77-8216, March 1977.
2. D. A. Lundgren, "Aerosol Measurement Methods," Paper No. TUA1 in NASA
CP-2004, 1976.
3. R. D. Cadle, The Measurement of Airborne Particles, John Wiley and Sons,
1975.
4. H. C. van de Hulst, Light Scattering by Small Particles, John Wiley and
Sons, 1957.
5. M. Kerker, The Scattering of Light and Other Electromagnetic Radiation,
Academic Press, 1969.
6. A. R. Jones and W. Wong, "Direct Optical Evidence for the Presence of
Sooty Agglomerates in Flames," Combustion and Flame 24, p. 139-140, 1975.
7. J. R. Hodkinson, "The Optical Measurement of Aerosols," Aerosol Science,
C. N. Davies, editor, Academic Press, 1966.
8. F. Durst and B. Eliasson, "Properties of Laser Doppler Signals and Their
Exploitation for Particle Size Measurements," Proceedings of the LDA-
Symposium Copenhagen 1975, 1976.
9. A. J. Yule, N. A. Chigier, S. Atakan and A. Ungut, "Particle Size and
Velocity Measurement by Laser Anemometry," AIAA Paper No. 77-214, 1977.
10. F. Durst and H. Umhauer, "Local Measurements of Particle Velocity, Size
Distribution and Concentration with a Combined Laser Doppler Particle
Sizing System," Proceedings of the LDA-Symposium Copenhagen 1975, 1976.
11. S. S. Penner, J. M. Bernard and T. Jerskey, "Power Spectra Observed in
Laser Scattering from Moving, Polydisperse Particle Systems in Flames—
I. Theory," Acta Astronautica 3. p. 69-91, 1976.
12. S. S. Penner, J. M. Bernard and T. Jerskey, "Laser Scattering from Moving
Polydisperse Particles in Flames—II. Preliminary Experiments," Acta
Astronautica 3, p. 93-105, 1976.
484
-------
13. C. S. Williams and 0. A. Becklund, Optics, Wiley-Interscience, p. 198ff,
1972.
14. C. G. McCreath and J. M. Beer, "A Review of Drop Size Measurement in Fuel
Sprays," Applied Energy 2, p. 3-15, 1976.
15. W. M. Farmer, "Measurement of Particle Size, Number Density and Velocity
Using a Laser Interferometer," Applied Optics 11, p. 2603-2612, 1972.
16. K. L. Orloff, F. C. Myer, M. F. Mikasa and J. R. Phillips, "Limitations On
the Use of Laser Velocimeter Signals for Particle Sizing," Proceedings
Minnesota Symposium on Laser Anemometry (1975), p. 359-370, 1976.
17. D. M. Robinson and W. P. Chu, "Diffraction Analysis of Doppler Signal
Characteristics for a Cross-Beam Laser Doppler Velocimeter," Applied
Optics 14, p. 2177-2183, 1975.
18. N. S. Hong and A. R. Jones, "Light Scattering by Particles in Laser
Doppler Velocimeters Using Mie Theory," Applied Optics 15, p. 2951-2953,
1976.
19. D. Roberds, "Particle Sizing Using Laser Interferometry," Applied Optics
16, p. 1861-1868, 1977.
20. Spectron Development Laboratories Report 77-6171, April 1977.
21. J. R. Hodkinson, "Particle Sizing by Means of the Forward Scattering Lobe,"
Applied Optics 5, p. 839-844, 1966.
22. C. C. Gravatt, "Light Scattering Methods for the Characterization of
Particulate Matter in Real Time," Aerosol Measurements, NBS Special
Publication 412, p. 21-32, 1974.
23. E. D. Hirleman, S. L. K. Wittig and J. V. Christiansen, "Development and
Application of an Optical Exhaust Gas Particulate Analyzer," Laboratoriet
for Energiteknik Report RE76-4, Technical University of Denmark, 1976.
24. E. D. Hirleman, "Optical Technique for Particulate Characterization in
Combustion Environments: The Multiple Ratio Single Particle Counter,"
Ph.D. Thesis, Purdue University, August 1977.
25. J. Swithenbank, J. M. Beer, D. S. Taylor, D. Abbot and G. C. McCreath, "A
Laser Diagnostic Technique for the Measurement of Droplet and Particle
Size Distribution," AIAA Paper No. 76-69. 1976.
26. Alan Wertheimer, "Optical Methods for Real Time Particulate Measurements,"
paper presented at Air Pollution Control Association (Ontario Section)
Symposium, April 27, 1977, Toronto, Ontario, Canada.
27. S. A. Self, et al., "Investigation of Novel Laser Anemometer and Particle
Sizing Instrument," Project SQUID Semi-Annual Progress Reports,
October 1, 1976, and April 1, 1977.
485
-------
28. Photographs from research performed at MIT under the direction of
Professor A. F. Sarofim of the Chemical Engineering Department.
29. G. L. Fisher, et al., "Fly Ash Collected from Electrostatic Precipitators:
Microcrystalline Structures and the Mystery of the Spheres," Science 192,
May 7, 1976.
486
-------
Figure 1. Geometry for Light Scattering Analysis
For small particles,
light intensity distributions
are similar in fore ward and
backward direction
For larger particles,
light intensity
distributions are complex:
More light is scattered
in foreword direction
Figure 2. Mie Theory Results: Variation of Scattered Intensity with
6 for Two Sphere Sizes (Figure from Reference 8)
487
-------
Laser
Beam
X
Recording
Device
Flow
Figure 3. Schematic of Photography System for Particle Sizing
MONOCHROMATIC
COHERENT
LASER
LIGHT
MIRROR
Figure 4. Schematic of Holography System
(Figure from Reference 14)
488
-------
,-0
i- O
4-> cr+J
(U
-------
1.0
.8
x
t=
.6
CM
"
.2
0
x = Dp/5
Figure 6. Visibility Function for Spherical Particles
as Determined by Farmer (Reference 15)
Diameter (^unl for X'SWSnm
Intensity
Ratio
Figure 7. Intensity Ratio (10°/5°) for a Wide Range of Indices
of Refraction as Calculated by Gravatt (Reference 22)
490
-------
o
ro
oo
CM
OJ
CM
O
CM
_ 00
OJ
N
o
•r-
s- o
re O
0. 00
M- t.
O O
c
o c
•r- O
U •*->
C O
3 -a
re c
in
c CD
•!-
c to
CO
CD
S-
3
CD
-------
25
20
O)
\
V.
<
\
w
<
15
10
0
12° SCATTERING
n=l.54
n=
.3 .4 .5 .6 .7 .8 3 1.0 1.5 2.0 3
DIAMETER , /J™ (X = 0.488 pm)
Figure 9. Sensitive Area As vs Particle Size for Two Indices
of Refraction, Accounting Only for Intensity Effects
(from Hirleman, Reference 24)
492
-------
Lignite (75-90) Coal, Complete Combustion
(Photograph shows a big ash cenosphere
containing many smaller spheres)
B
Lignite (75-90) Coal, Complete Combustion
(Photograph shows a big broken cenosphere
of ash)
Figure 10. Scanning Electron Micrographs of Fly Ash (Reference 28)
493
-------
PARTICIPATE SAMPLING AT HIGH-TEMPERATURE AND HIGH-PRESSURE
THE EXTRACTIVE APPROACH
By:
W. Z. Masters
Acurex Corporation/Aerotherm Division
Mountain View, CA 94042
495
-------
PARTICULATE SAMPLING AT HIGH-TEMPERATURE AND HIGH-PRESSURE —
THE EXTRACTIVE APPROACH
By:
William Masters
Acurex/Aerotherm
Mountain View, California 94042
ABSTRACT
A participate sampler for high-temperature, high-pressure
processes has been developed and successfully demonstrated. The
system uses an extractive approach, removing samples from the
process stream for complete analysis of particulate size distri-
bution, morphology, and chemical composition. System capabilities
have been demonstrated 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.
496
-------
PARTICULATE SAMPLING AT HIGH-TEMPERATURE AND HIGH-PRESSURE --
THE EXTRACTIVE APPROACH
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 sampling
operations at a pilot-scale fluidized bed combustor. The system performed
successfully in a variety of operating modes, producing sample data.
Acurex/Aerotherm 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 processes (Contract 68-02-2153). The EPA Project
Officer for the contract is William Kuykendal.
The following sections of this paper discuss the extractive sampling
concept, the HTHP 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 velocity to avoid biasing particle size distribution measurements
(isokinetic sampling).
497
-------
The error in measured particle content as a function of anisokinetic 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 result.
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 compatible 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 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 requirements 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
t Conventional trace element collectors (organics trap and
impingers)
• Measurement and control instrumentation to assure isokinetic
conditions
A schematic diagram of the sampler is shown in Figure 2. The sample probe
is mounted within a pressure-containment housing. The probe can be inserted
into stream through valves that connect the housing assembly to the process
498
-------
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 and
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:
t 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
t 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 sampler. The sample probe and particle collector are shown in Figure 4.
Selecting the particulate collection temperature was a second major
design decision. 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 been used in an
experiment investigating the effect of collection temperature on particulate
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
Impactor, Model D.
499
-------
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 directly to the performance of participate
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 measurements more significantly. Variations in cut size with
temperatures have been calculated 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 telescoping cylinders which move the probe into and out of the
process. Hydraulic cylinders 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 telescoping housing is the most
complex part of the sampling system and consequently required 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 steam and sample
probe must be measured. A pi tot tube and thermocouple 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 controls for sample temperature, probe traverse, trace
element collector flow, and other key operating parameters. System controls
are housed in two portable enclosures, shown in Figure 6.
500
-------
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/Aerotherm. 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 & Haas XAD-2 gas chromatographic
packing material. The impinger train has four high-volume glass impingers,
three filled with collector 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 (HN4)2S2Og + 0.02M AgN03
No. 3 0.2M (NH4)2S208 + 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
Mi nip!ant PFBC. The PFBC facility is described in this section.
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. Fluidization 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 load-
ing can be reduced to the levels (0.0002 to 0.002 gr/scf) required to
501
-------
protect turbine blades. The Exxon Mini pi ant facility is being used to
investigate fluidized bed combustion, gas cleanup devices, and particulate
effects on turbine components. At the time of sampling, the facility did not
include a gas turbine or final cleanup device.
For the sampler demonstration 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 Mini pi ant 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 combustor 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 equipment are set up one floor below, where
a partial enclosure gives some weather protection.
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 collection temperature on particulate
compensation. The sampler operated successfully in both test series. These
operations and some of their results are described in this section.
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. Trace element
collection equipment was operated during the filter run. The tests produced
the following data:
• Particulate size distribution
• Particulate chemical composition
• Particulate shape
• Particulate concentration
502
-------
t Process temperature and pressure
• Moisture content
• Structure temperatures (valves and probe housing)
• Trace element 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 using 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 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 processes (Reference 8). The plots show the trend
of decreasing physical size from Stage 1 to State 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 a aluminum, silicon, sulfur,
potassium, calcium, titanium, iron and copper.
Comparing the relative height of the peaks in two spectra can give a
rough indication 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 (D™ of
about 30 microns) and the finer particles (D™ of about 0.6 micron).
503
-------
Data from the system demonstration tests are discussed more extensively
in a test report submitted to EPA IERL (Reference 9).
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
composition. We were specifically concerned that trace elements might condense
between process temperature and conventional particulate collection temperature
(about 450°F). For these tests, the sampler was setup to collect particulate
at process temperature, so trace element condensation 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 filter are mounted at the front of the probe to remove
particulate at process temperature, a series of orifices gradually reduce
sample gas pressure, and a final filter collects condensation products. The
cyclone used is a Southern Research Institute model, designed for much less
severe operating temperature. This cyclone was readily available, 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.
The chemical analyses of particulate samples may show some debris from the
plating failure.
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.
The estimated filter efficiency was well over 90 percent of the fine
particulate passed by the scalping cyclone.
504
-------
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 conventional filter materials at this location because
sample gas temperatures are substantially reduced by the probe cooler
section.
In the condensation test series, we completed four sampling runs.
Analysis of the samples is still in progress, so no conclusions on trace
element condensation have been reached. Some sample data is available,
however. The particulate collected at process temperature has been analyzed
by spark source mass spectrometer. The results are presented in Table 1.
The measured trace element content is similar to common flyash. A better
evaluation will be possible when spark source mass spectrometer analysis of
particulate samples from the demonstration test series is completed. The
comparison will show if the different collection temperatures result in
different chemical composition measurements.
The material collected on the final filter has been analyzed in detail
by the Arthur D. Little Company. The goal of this analysis was to find what
materials were collected on the filter at the probe exit, and determine if
they would have resulted from condensation processes. The analysis showed
that the material collected is partly particulate which penetrated the hot
filter, partly contaminants from a Grafoil packing in a value at the probe
exit, and partly sulfuric acid residue. The acid residue is a condensation
product caused by inadequate heat tracing in the probe valve area. A
thermocouple in this area measured sample gas temperatures as low as 200°F
during the condensation tests. This is well below the 450°F target for
proper sample conditioning. At this point, it appears that more meaning-
ful information on trace element condensation will come from the
comparison of the hot particulate catch and demonstration test particulate
catch than from further examination of the materials collected on the final
filter. However, further analysis of the final filter is planned.
505
-------
Photomicrographs of the participate samples from the condensation
tests are shown in Figures 17 and 18. The particulate is similar in
appearance, supporting the conclusion that part of the material on the final
filter carried through the hot filter. Dispersive fluorescent X-ray spectra
in Figure 19 show a similarity in composition between the filter catches,
except for the increased sulfur contents of the rear filter.
Further data will be available when the analyses of condensation tests
samples are completed.
Conclusions
The sampling system described in this paper demonstrates that
extractive sampling is a feasible approach for sampling high-temperature,
high-pressure processes. Technology for sampling pressurized fluidized bed
combustors is now developed and available. Future development also will be
required, however, to make useful application of this technology and extend
it to other advanced coal conversion processes.
For FBC sampling, remaining issues include collection temperature
selection and system cost/performance trade-offs. The collection temperature
issue will be resolved by further condensation tests or by the development of
high-temperature particulate collectors. Process developers seem to be
interested in both upgraded and downgraded versions of the sampling system.
Upgraded versions offer longer sampling durations, quicker turnaround and
better operating convenience. Downgraded 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 PFBC's.
506
-------
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 Parti cul ate Control in Energy Process,
EPA-600/7-76-010, Sept. 1976.
3. Homersma, 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
^ 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, September, 1976.
8. Hoke, R. C., Exxon Research and Engineering Company, Linden, New Jersey,
Personal Communication.
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.
10. Lee and Lehmden, "Trace Metal P'ollution in the Environment," Journal
of the Air Pollution Control Association, Vol. 23, No. 10, October 1973.
507
-------
TABLE I. CONCENTRATIONS OF METALS IN FLYASH - SSMS DATA
From Condensation Test Samples:
Element
K
Na
V
D£
Si
Fe
Ca
Mg
Ti
Cu
Ni
Mn
Zn
T£
Cyclone
(PPM)
8200
1310
185
16400
94000
30000
20000
11400
2430
248
120
59
< 13
6
Hot
Filter
(PPM)
8850
2500
135
9400
82600
13000
19000
17800
1950
. 165
100
49
< 80
1950
508
-------
o
r~ .
UJ
M
oo
UJ
i
0
1 — 1
1—
o:
Q-
f_i__
o
^fI
o
t— 1
1—
o
^3
u_
et
GO
e£
'
GO
•>-
— 1
U-
•^
co
t
e^
1—
UJ
s:
UJ
0
*~c
C£
1 —
•
*— *
KH
LU
— J
CQ
^C
l_
E
Q.
QL
„
E
O
•_J3
ro
<-.
C
cu
o
c
o
CJ
E
LO
•
•~
E
LO
.
CO
E
O
•—
E
LO
•
CM
i — -
E
LO
CM
fO
c~
CU
E
cu
f~™
UJ
O 0
O 0
CO LO
rv
CTl
LO
O O
o o
LQ LO
*i
OO
l£3
O O
O O
OO LO
r~.
LO
o o
0 O
CO LO
fl
— r— OCMCMCMCM
n
en
LO
CMLOOOOOOOO
iCOLOOOOOO
vli — i — OCMOOCOCM
«
.0
«^J*
cu-as-rscucT-jD
CQOOC-JU-Sr^D-^
c
ro
c
0
•r-
J ^
rO
•r-
•*->
O
(a
c
0
s-
OJ
c
^w
o
ai
N
ro
ro
-£,
E
fO
E
ro
Q.
^_
OJ
2
O
a.
E
ra
cu
-p
CO
t3
0)
S-
1 S-
1 — +->
ro O)
0 E
0 0
rO •)->
E QJ
O CL
t- 00
to
T3 to
OJ ro
-i-0 E
r_)
O) CO
i— CJ
O 3
0 O
to
CU
O. 5-
£ fQ
ro a.
oo to
ro
509
-------
2.0
1.8
o
1.6
1-4
z
O
O
m
Z
O 1.0
z
111
O 0.8
O
O
D
HI
C 0.6
in
OT
m
O
u.
O 0.4
O
0.2
k = 0.005
k = 0.02
k = 0.04
k = 0.08
k = 0.14
L k = 0.32
k = 1.0
I
I
I
0.1 0.3 1-3
VELOCITY RATIO (R)
10
Figure 1. Probe inlet bias (from reference 1)
510
-------
o
(J
to
OJ
oo
CXJ
O)
13
en
511
-------
CD
(ft
I/I
O)
S-
Q.
I
_c
en
OJ
i.
QJ
CL.
O)
4J
I
^:
CT)
ro
o>
13
CD
512
-------
.Q
CO
8
Figure 4. Exploded view of HTHP probe.
513
-------
qgeo-H
o>
(O
-------
qgeo-H
1/5
O)
c
O
O
O
-(->
c:
O
O
O)
3
CTi
515
-------
qzeo-H
!*«
(T3
S-
to
re
en
res
c
O
O
o
o
5-
3
cn
•r-
U-
516
-------
Uk.
OJ
-p
co
>>
CO
S-
o
§
(J
(O
O
O
•o
0)
-O
OJ
N
•5
•I—
^
<+-
T3
CD
s-
=3
CO
CO
01
S-
D-
co
-------
.a
«=}
oo
o
"pi-
9. Miniplant PFBC.
518
-------
QSSO-H
CO
r—
Q-
o
X
X
OT
c:
QJ
CO
LO
(O
O
s_
Q.
OJ
S-
519
-------
100
80
60
10
PERCENTAGE UNDERSIZE (BY WEIGHT)
20 30 40 50 60 70 80
'90
98
fc
vfl
N
40
E
a.
of
UJ
UJ
5
O
cc
<
Q.
20
10
8
1 0
0.8
0.6
0.4
0.2
0.1
90 80 70 60 50 40 30
PERCENTAGE OVERSIZE
20
10
Figure 11. Particle size distribution.
520
-------
Figure 12. Impactor substrates.
521
-------
1000X
10 Microns
3000X
3 microns
Figure 13. Particle photomicrographs impactor Stage 1
522
-------
3000X
10000X
1 micron
Figure 14. Particle photomicronraohs imnactor Staae 6.
523
-------
-------
(Q
O
CO
Q.
o
o
525
-------
PARTICLE FIELD DIAGNOSTIC SYSTEMS
FOR HIGH TEMPERATURE/PRESSURE ENVIRONMENTS
By:
J. D. Trolinger, W. D. Bachalo
Spectron Development Laboratories, Inc,
Costa Mesa, CA 92626
527
-------
PARTICLE FIELD DIAGNOSTIC SYSTEMS
FOR HIGH TEMPERATURE/PRESSURE ENVIRONMENTS
By:
J. D. Trolinger and W. D. Bachalo
Spectron Development Laboratories, Inc.
3303 Harbor Boulevard, Suite G-3
Costa Mesa, California 92626
Particle diagnostic techniques and instrumentation are
reviewed and their relative operational characteristics are
discussed. A brief appraisal of optics techniques based on
imaging and on light scattering methods has been given. The
description of the principles of operation, applicability, and
measuring characteristics of the instrumentation being developed
by Spectron Development Laboratories (SDL) has been reviewed.
These laser-based light scatter detection instruments are based
on the predictable angular scattering characteristics of small
particles and the interference of coherent light. Angular
scattering intensity ratios are used for sizing particles in
the range of 0.5 - 10 ym and particle sizing interferometry
(visibility) is used in the range 2 - 100 ym.
528
-------
PARTICLE FIELD DIAGNOSTIC SYSTEMS
FOR HIGH TEMPERATURE/PRESSURE ENVIRONMENTS
SECTION 1
INTRODUCTION
In view of the shift of emphasis to coal combustion as an energy source,
the monitoring of particulates associated with the utilization of coal has
become a very important area for technological development. Of the particle
sizing systems available, few can withstand the high temperature and high
pressure environments of the power plants. However, monitoring the particle
flux in such systems is imperative to both the design and operation of the
facility. As an example, in fluidized bed combustion (FBC) systems there are
a number of areas in which the knowledge of the particle size and morphology
would be valuable.
In the bed area, the coal particles under combustion could be monitored
for temporal size variation as they pass through the combustion process and
for formation of soot and fly ash. Dense spheres of fly ash formed in a
stoichiometric combustion are borne in the hot gases. These particles, if not
removed by the cyclone separators, can cause excessive erosion (and corrosion)
of the gas turbine blades. Efficient removal of particles in the size range
of 2 to 10 i_im in diameter from the flue gas is especially difficult. Reliable
measurements in this size range are required to evaluate the effectiveness of
particulate removal systems. Only sparse data now exists on particle loading
in this area.
In combined cycle systems where gas and steam turbines are used, water
droplets in the steam system may form at lower superheat temperatures. An
undetected excess of such droplets will cause greatly accelerated erosive
damage to the steam turbine. Reliable droplet detection instrumentation would
allow closer control of the steam turbine cycle efficiency.
529
-------
Environmental concerns require that the levels of emissions from a FBC
plant be acceptable or at least tolerable. Of the particulates released into
the atmosphere, those in the range of about 0.2 to 3.0 ym are perhaps most
damaging to the human respiratory track. Monitoring of these small partic-
ulates in a polydisperse effluent in the presence of larger particles is a
challenging problem. Typically, particle collection systems can remove most
of the larger particles from the effluents. It is a reasonable assumption
that the submicron particulates are, to date, escaping undetected.
At Spectron Development Laboratories we have been involved in extensive
particle diagnostics research in a broad area of applications including all
of the areas mentioned above. For example, holographic particle field diag-
nostic instrumentation is being tested for its feasibility in monitoring the
combustion of coal particles in a laboratory FBC system. Information regard-
ing the particle distribution, size, morphology, and the dynamics of the
process are possible with this instrument. Furthermore, there is a possibil-
ity of obtaining some information about the flame structure engulfing indiv-
idual particles. We are refining our particle interferometer and scattering
intensity ratioing systems for application in FBC plants. These systems will
soon be tested in the ANL Solids/Gas Flow Test Facility. The instrument pro-
vides non-intrusive on-line monitoring of flow velocity and particulate content
(hence, particle flux). Design of the latest instruments has been directed
toward making them operable in the FBC plant environments shown by flags on
Figure 1.
In the spirit of the goals of the meeting, our paper will review some of
the general principles and techniques available that may be applicable in high
temperature/pressure environments as particle diagnostics instruments. The
laser-based light scattering optical systems developed by SDL will be discussed
in detail with emphasis on the systems' performance and range of application.
530
-------
SECTION 2
A REVIEW OF PARTICLE DIAGNOSTIC TECHNIQUES
In previous research and development of particle diagnostic methods, we
compiled and studied more than 500 research publications dealing with this
subject. These reports dealt with a wide variety of techniques. To select
the instrument concepts that are satisfactory for the application, careful
consideration must be given to what characteristics constitute a viable
instrument. Some of the criteria to be considered are the following: The
instrument must be reliable. In the early stages of development, reliability
may mean that the instrument works or does not. The physics concept or con-
cepts involved must be shown to be applicable in the expected particle size
range; that is, the system must be suited to the specific application.
As in most situations, the simplest device that performs the desired task
is usually the best. This is particularly true where the user of the device
sees it as a tool to attain his goals and does not care to or have time to
fully understand the details of its operation. In this vain, ease of opera-
tion of the instrument must be a fundamental concern. How well the instrument
performs may be as dependent on its proper utilization as it is on its concep-
tual and design integrity. A field-ready instrument must be as rugged as
possible and portable, preferably one-man portable. Data retrieval should be
straightforward and conveniently formatted. Modern microprocessor technology
provides the medium to reduce most instrument information with usable data.
The selection of a particle diagnostics concept from which an instrument
may be developed includes four major problem areas: (a) the choice and devel-
opment of the probing head, (2) calibration and verification of the technique,
(3) information processing, and (4) packaging. The instrument development
should follow this same order.
531
-------
The probing element produces the measured parameters for the individual
or a collection of particles passing through the sample volume. Selection
of this element of the system is critical and hence, the previously mentioned
criteria must be fully considered at this stage. In situ instruments typically
have even more stringent requirements as they are subjected to the process
environment. Once selected, the concept must undergo an exacting verification
and calibration with standardized particle fields. Not only does this require
verification of the sizing capabilities but also of the operational limitations
such as particle loading limitations, rejection of spurious signal, and the
system's dynamic range. The system's information storage, retrieval, and dis-
play can be as sophisticated as is necessary in light of the advanced electron-
ics technology available in minicomputers and microprocessors. The output may
entail doing some statistical analysis on the data, it may be simple mean
quantities, or it could be a warning signal transmitted to the plant operator
or directly to the plant electronic controls.
There are many methods currently available for measuring size distribu-
tions of particulates in gaseous streams, each with their advantages and dis-
advantages. In this report we review only those techniques that have a possi-
bility of being applied in high temperatures (1000-2QOO°F) and high pressure
(1-10 atms) plants. Some of the techniques require withdrawal of a sample
from the stream and the sample then must be analyzed in the laboratory. Others
are in situ devices which eliminate some of the uncertainties due to the hand-
ling of the sample. However, data quality from such devices is often question-
able. Most of the techniques are limited to a definite particle size range.
For real particle field measurements, two or more techniques may be required
to cover the range of sizes.
Particle sizing techniques fall into one of the following categories:
• Optical
• Imaging
• Light Scattering
• Aerodynamic
• Impaction
• Centrifugation
• Sedimentation (Gravitational)
532
-------
• Filtration
• Barrier
• Sieving
• Electrostatic
• Condensation
• Diffusion
• Acoustic
NON-OPTICAL TECHNIQUES
Aerodynamic
Particle impactors are currently the standard means of obtaining size
distribution measurements under field conditions. Particle laden air is ex-
tracted by means of a sampling probe and passed through a series of orifices
or slits arranged in order of decreasing size (increasing velocity) and dir-
ected against impaction plates. Particles are thus collected on the plates
in proportion to their size and information can be gained by counting, weigh-
ing or other analytical techniques. Unfortunately, several drawbacks exist
which have prompted the need for other methods. Sampling by means of a probe
has inherent problems and inaccuracies. Great care must be taken to insure
that the sample is extracted isokinetically and that the diameter of the probe
and extraction velocities are chosen correctly (to avoid particle deposition).
Analyzing the impactor stages after a sample has been taken is also very time
consuming. Impactors must be disassembled and careful weighing of samples
must occur under laboratory conditions. Weight gain of or loss from the col-
lection substrates is caused by the presence of some gasses and can be a
serious problem in data analysis.
Centrifugal devices such as cyclones are also used routinely for particle
size determination. Cyclones operate by injecting a particle laden stream
tangentially into a tapered conical enclosure. Centrifugal acceleration af-
forded by the spiral motion determines whether the particle is collected by
the cyclone or remains suspended and continues out the exhaust.
Cyclones can be cascaded like impactors to provide a range of particle
cut points. Their advantages are that they adapt easily to sampling apparatus
533
-------
and they provide a large collection capacity. Unfortunately, they have the
same disadvantages common to impactors.
Centrifugal elutriation devices are also available and this method is the
accepted ASME means for fly ash classification. Disadvantages of this device,
like impactors and cyclones, are the lack of in situ measuring and the time
required for sizing analysis. Other disadvantages are that the sample must
be collected and then redispersed into the instrument, classification is highly
dependent on flow conditions and inlet geometry, and these devices are gener-
ally more expensive than impactors and cyclones.
Gravitational devices depend on particle weight to determine a settling
velocity and trajectory causing a distribution by size on the bottom of a
settling chamber. This method requires the minimization of convective thermal
currents and very long periods of time (several hours) to resolve fine parti-
cles which makes it an unattractive means to size particles.
Filtration
Particles drawn into a sampling probe can be collected by barrier fil-
tration. The barrier filter can be designed to collect essentially 100 percent
of the particles present in the stream. This technique is often used as the
final collection stage of an inertial impactor train or series of cyclones.
This method can also be used as the only collection device. While it is not
an in situ method, it does have the advantage that essentially all the partic-
ulates are collected in one place. As with several of the other techniques, it
has the disadvantage that size distribution is done by time-consuming labora-
tory techniques such as Coulter counter analysis, optical or SEM microscopy.
Accuracy is dependent upon the ability to redisperse the particles. As a rule,
this is increasingly difficult as particle size is reduced.
The size distribution of particles in bulk form can be obtained down to
about 2 ym by sieving. Sieves for fine particle analysis are pans with bottoms
of electroformed grids. Particles are separated into size fractions by means
of a series of sieves and the particle mass retained on each sieve is usually
determined by weighing.
534
-------
Acoustic
A relatively large airborne particle, e.g., 15 urn, passing rapidly
through a gradually tapering tube that expands rapidly after a small throat,
produces an audible click that can be counted or recorded electronically. The
click originates near the end of the throat as a shock wave due to boundary-
layer disturbance in the throat created by the particle. The device is sim-
ple, but no embodiment yet designed responds to small particles, and the re-
sponse that is given is only a rough indication of particle size.
Particle size can be deduced from the sinusoidal paths particles follow
if a slowly rising aerosol stream is dark-field illuminated and exposed to
low-power sound waves. The method is obviously tedious and is to be recom-
mended only if special circumstances dictate its use.
A measure of the particle size and concentration of aerosol can also be
deduced from the extinction of ultrasonic sound waves passing through an
aerosol and through the same gas without particles. Conditions are most fa-
vorable for the application of sound-attenuation techniques with aerosols of
high concentrations. This makes such techniques of dubious practical value in
most instances.
Vibrating Crystals
Piezoelectric crystal mass monitors in combination with an impactor
train would appear to hold some promise for an in situ particle size distri-
bution and concentration device. The piezoelectric mass monitors are intrigu-
ing because of their extreme sensitivity. However, such a device can be
expected to be subject to several sources of error: linearity problems could
require short sampling periods especially in high concentrations, errors could
result from temperature or humidity fluctuations, non-uniform particle density
could be interpreted as changes in concentration and the devices would require
a crystal cleaning device for subsequent readings.
Filter Tape with Beta Absorption Mass Monitor
A device of this type has been sued to measure the mass of atmospheric
fine particle aerosol. Since it is basically a mass measurement device, it
is probably not easily adapted to in situ measurements. Particle sizing
535
-------
would depend upon impactors and the mechanism required for moving the filter
tape would be prone to problems in the severe environment contemplated for
the in situ fine particle analyzer.
Contact Electrification Devices
This is a class of instruments which provides an electrical current
proportional to the particle mass concentration in a gas stream. They work by
transfer of charge from the particulate to the instrument probe through par-
ticulate probe collisions. The sensitivity of the devices depend on the
electrical resistivity of the particulate as well as upon the condition of
the surface of the probe. The theory for operation of these devices is poorly
understood. They can be expected to suffer from the effects of sticky par-
ticles and from particulate streams of varying composition. Furthermore, the
device does not seem well suited to measurement of particle size distribution.
OPTICAL TECHNIQUES
Imaging Methods
A number of considerations and definitions apply in general to imaging
techniques. The most fundamental is the degree of resolution attainable with
the system. For a perfect imaging system, a point source of light will be
imaged to a finite sized spot (impulse response, point spread function, blur
circle) limited in size by the laws of diffraction. This image size repre-
sents the smallest object which can be resolved in the image or the smallest
spot to which a light beam may be focused. The diffraction limited resolution
is given approximately by
RD = 1.22 X S/D (1)
where S is the lens to object distance, D is the effective diameter of the
collecting system (lens or limiting aperture) and X is the wavelength of the
light. Smaller objects than this can be detected in the image if the signal-
to-noise ratio is sufficient, but the dimensions of smaller particle images
can be no smaller than the resolution limit. Except for very limited cases,
resolution is worse than that given by the above relationship because geomet-
rical aberration terms must be added.
536
-------
Another important consideration is optical noise or system signal-to-
noise ratio. Every optical system scatters useless light into the vicinity
of the image being analyzed to the extent that this usually limits the utility
of the imaging technique. In particle field diagnostics, the situation is
amplified by the fact that all particles besides the one being imaged at a
given time are potential sources of optical noise. This means that not only
the optical system but the particle field itself must be considered in eval-
uating ultimate system capability. Ambient light usually represents a noise
source, but methods are now available to effectively eliminate this as a
problem.
Holography
Holography or wavefront reconstruction allows the reproduction of the
scattered light field in the laboratory. With this technique, holograms are
made of the particle field (Figure 2); and after processing, the image is re-
2
constructed and scanned with the aid of microscopy and video techniques.
The narrow depth of field is given approximately by
s " T (2)
which is a problem in photography but is relaxed with holography. The recon-
structed image can be scanned in depth with a microscope. Thus, for a given
resolution, holography expands the image storage capability by three or four
orders of magnitude.
Holography has been shown capable of resolving particle sizes as small as
1 micron in diameter. The advantage of this technique is that a global view
of the particle field is available. Spatial and temporal variations in the
field may be detected and evaluated as they relate to the process at hand.
This is a particularly useful capability when investigating processes such as
the ignition and combustion of coal in a fluidized bed. Shape variations as
well as size change can be observed with this technique. Double pulse holog-
raphy poses the possibility of investigating the local temperature field
simultaneously.
The processing of holograms requires a period of time for recovering the
information making real time diagnostics impossible with the materials
537
-------
currently available. Holography related technology is almost certain to pro-
duce real time recording and reduction materials. When combined with rapid
electronic image analysis, holography would qualify as a valid method for a
much greater range of high temperature/pressure applications.
Scattering Instruments
Instruments based on light scattering can measure particle sizes to at
least an order of magnitude smaller than the resolution limit of the system's
transmitting lenses. This can be done without high quality scattered light
collecting systems. However, the transmitting optics must be of a high
quality.
Particles illuminated by a high intensity light beam will scatter light
by an amount and in a spatial distribution that depends on several factors.
Careful optical selection of conditions can reduce the number of parameters
or provide for factoring out such parameters. When the particle diameter is
greater than the light wavelength and when only forward scattered light is
considered, the Fresnel-Kirkhoff approximations may be used to predict the
diffracted light intensity distributions. The intensity of forward scattered
light is given for a circular disk by:
I (a) - I k2a4 ["-
J. I dy — -L~IV d I
(J [_ _
where I,-. = the incident flux per unit area
k = 2TT/A
w = sin 6 where 6 is the angle relative to the incident beam
a = particle radius.
An instrument that collects the scattered light making a single measure-
ment per particle collects the integrated intensity over the collecting angle.
When the collecting angle is small enough such that J.. (kau) / (kaoo) is constant,
the intensity is approximately:
0 ~ TT , , 2 24
Thus, a particle sizing instrument could be based on the collection of forward
4
scattered light yielding a signal proportional to a . It is important to note
538
-------
that since the scattering is based upon diffraction theory, the scattered
light is not dependent on refractive index.
If light collection is taken over a large scattering angle, the collected
intensity is approximated by
I-Y\
2
IL = f I(o))2irwdo) =irl a . (5)
The important result being that when all the scattered light is collected, a
2
signal proportional to a results.
Scattering Ratio Measurements
In order to overcome the sensitivity of a scatter-based instrument to
refractive index and to incident intensity variation, the rat;Lo of the scat-
tered light intensity at two angles may be measured. In polydisperse particle
fields of unknown materials or with several known materials of different index
of refraction, insensitivity to this parameter is important. Furthermore, the
light intensity incident on the particle in a laser-based system will show
large intensity variation (at least an order of magnitude) depending on where
the particle crosses the focal volume. Ratioing of the light scattered at two
angles eliminates the uncertainty due to the variations in incident light
intensity.
Figure 3 shows a polar plot of angular scattering intensity for represent-
ative particle sizes. These scattering relationships are based on Mie scat-
tering theory. Because the forward scatter lobes become very narrow for larger
particles, this technique is limited to particle diameters between about 0.5
and 5 ym for an incident light wavelength of 0.633 ym (Figure 4). For smaller
particle diameters, ratioing of the forward and backscatter intensities may be
used. This reintroduces at least a weak dependence on index of refraction due
to the backscatter measurement. However, particles in the range of 0.1 to 1.0
ym in diameter have been measured using this principle.
Because this type of measurement does not require single particle scat-
tering to provide useful information, the device is well-suited for making
measurements in the small size range with a relatively high number of particles.
When an ensemble of particles is illuminated, it can be shown that measurements
539
-------
of intensity ratio of light scattered at two angles can produce the mean
diameter of the sample.
We are currently actively involved in developing this technique into a
workable instrument. Incorporation of this system into our particle sizing
interferometer will enhance the accuracy of our small particle measurements
and alleviate the difficulties with multiple particles being in the probe
volume simultaneously.
Particle Sizing Interferometry
When a laser beam is split into two equal intensity beams and the beams
are focused to a crossover, fringes formed of spatially varying light intensity
occur (Figure 5). The fringes fill the small ellipsoidal space that consti-
tutes the focal volume. These fringes are parallel (if the system is properly
designed) and equally spaced a distance 6 apart, given by
x = * (6)
2 sin 6/2
where 8 is the beam intersection angle (Figure 6).
Particles moving through the focal volume will scatter light in propor-
tion to the encountered spatially varying light intensity. The scattered light
signal is a type of convolution of the particle geometry with the probe volume
intensity distribution. As the particle size with respect to the fringe spac-
ing increases, the scattered light signal increases in strength but decreases
in modulation percentage. The measure of the ratio of the minimum to maximum
signal modulation has been termed the visibility (Figure 7).
This technique possesses the desired features of the best scattering
methods; a well-defined and controlled sample volume; a measurement of ratios
of intensity as opposed to absolute magnitude; and perhaps most important,
an independence from refractive index of the particle. Because the optics
arrangement is identical to that of the well-known and powerful laser velocim-
eter technique, the velocity of the particles is simultaneously available.
Hence, particle flux and concentration can be evaluated with this device.
The technique has been demonstrated by several researchers to be
exceptionally reliable when operated in the forward scatter mode of light
540
-------
collection. Particle sizes in the range from 5 - 100 ym have been measured
accurately. Smaller sizes may be measured but care must be taken in properly
designing the collection system. At one fringe setting, a decade in particle
size range can be measured. A simple adjustment is required to change the
probe size and fringe spacing to cover a larger range in decade steps. We
are presently engaged in designing a system to cover two decades in particle
size without any manipulations required by the operator.
Considerable effort has been devoted in the development of the signal
processing electronics (Figure 8). Good accuracy has been achieved with our
visibility processor. The instrument's noise rejection logic has been further
refined to decrease the possibility of spurious signal acceptance.
We are also concentrating on developing a more rugged and compact packag-
ing arrangement (Figures 9 and 10). The package is being planned such that
little or no operator adjustments need be made after the system has been
established. Furthermore, the enclosure for the instrument will protect the
optics and laser from damage or contamination when in the hazardous plant
environment.
Particle sizing interferometer instruments have been developed by Spectron
Development Laboratories and applied in the field to a wide range of situations.
The following is a list of such applications demonstrating the instrument's
versatility:
1. Pollution monitoring from a helicopter platform.
2. Cloud diagnostics from a high altitude aircraft.
3. Particle diagnostics in simulated erosion facilities.
4. Diagnostics of fuel droplet sizes from nozzles.
5. Metrologic measurements in cloud chambers.
6. Analysis of medical nebulizers.
7. Rocket exhaust diagnostics.
8. Fog diagnostics.
541
-------
SECTION 3
CONCLUSIONS
Although a number of techniques may be devised for application to high
temperature/pressure environments, we believe the optical methods will produce
the most desirable instruments. Optical particle sizing instruments are non-
intrusive and hence do not disturb the flow field. Only a window access to
the test region is required. In situ measurements have been made with accept-
able reliability. The data retrieval systems can be automated and measure-
ments can be made in almost real time. Sampling techniques invariably require
significant time periods in handling the sample and recovering the desired
information.
A disadvantage in using light scattering techniques is the inability to
determine the particle shape. Analysis and calibrations are based on spherical
particles. In flow fields where particle shapes may deviate from spherical
errors in the size determination will occur. We are investigating certain
possibilities that may allow the determination of at least gross shape infor-
mation from additional scattering information.
In the future, we anticipate that longer range operations (1-2 meters)
of our scattering instruments will be required. Research is being carried out
on photon correlation processors for our particle sizing interferometer. That
would allow processing of signals wherein only a slightly greater number of
photons reach the detector when the particle is in a bright fringe than when
it is in the dark region (Figure 11). Such techniques have the advantage of
utilizing low-power lightweight lasers for long range measurements.
The data storage and processing systems we have available will be updated
with programmed microprocessors. We expect to have a broad range of data
542
-------
management formats to supply the user's special requirements. In addition,
we have the capability of interfacing our systems to minicomputers and provid-
ing the software for more elaborate data analysis and systems automation.
543
-------
SECTION 4
REFERENCES
1. Brooks, Robert D., "Coal Fired Combined Cycle Development Programs,"
Proceedings of the Fluidized Bed Combustion Technology Exchange Workshop,
April 1977, pp 109-124.
2. Trolinger, J. D., "Particle Field Holography," Optical Engineering,
Vol. 14, No. 5, p 383 (1975).
3. Farmer, W. M., "Measurement of Particle Size, Number Density, and
Velocity Using a Laser Interferometer," Applied Optics, 11, 2603 (1972).
4. Adrian, R. J. and Orloff, K. L., "Laser Anemometer Signals: Visibil-
ity Characteristics and Application to Particle Size," Applied Optics, 16,
No. 3, 677 (1977).
5. Robinson, D. M. and Chu, W. P., "Diffraction Analysis of Doppler
Signal Characteristics for a Cross-Beam Laser Doppler Velocimeter," Applied
Optics, 3.4, No. 9, 2177 (1972).
6. Chu, W. P. and Robinson, D. M., "Scattering from a Moving Spherical
Particle by Two Crossed Coherent Plane Waves," Applied Optics, 16, No. 3, 619
(1977).
544
-------
s
O
^
fa
u
w
X
u
c/3
^
a,
«
w
s
o
p*
w
^1
u
>l
u
Q
W
s
H
§
O
CJ
§
M
fe
O
u
EH
3
W
§
OJ
gag pe E0!;
w
-------
rMTM i
E
c
O)
CT
C
V)
Z5
o
I
C9
u
C9
(/I
ro
i
"05
c
L.
W
+->
X
LU
_!_>
L_
o
Q_
O)
1
W
>
tz
3
O
>
C9
Q.
03
C9
+->
O
£_
Q_
"O
O
O)
L.
Z5
Q_
C9
b
+->
03
O
(J
'"I r— /— -t-1 ,
•§ I S ^ b
CC LL LL CO Q-
o
o
r-(
o
33
o
•H
4-1
M
03
CM
£.
CO
r
O <3
m c
03 03
-" CL
O
I
C9
-*— '
ro i
Q_ (
O
+->
Z3
<
S_
- O
S *-
0)
rH
^2
n)
4J
M
O
P4
546
-------
JS
60
•rl
T)
0)
N
•rl
rl
ed
TH
O
CM
a)
(3
•H
0)
M
00
547
-------
60
•rt
O
to
a
0)
4J
M
60
M
0)
4J
4-1
td
u
60
O
M-l
4J
O
rt
o
•rl
4-1
P-
O
0)
(-1
3
60
•H
-------
•H
P
01
H
O
•H
4-1
rt
0) -H
•H Cfl
O PM
•H
0) 60
O d
M CO
C 3
•H O
M -H
(1) >-i
4-) tfl
4-1 >
tfl
O 4-1
cn n)
60
•H
549
-------
o
0
•
* ><
o
•rl
-------
CCS
o
•H
SO
o
to
W
o
0)
,fl
4J
u-i
o
CO
s
-------
max min
(b) V = 0.47
Figure 7. Oscilloscope Traces of Visibility Signals from the
Particile Sizing Interferometer.
552
-------
Figure 8. Particle Sizing Interferometer System.
553
-------
(-1
o
o) co
N (3
>. O
.H -H
< a
•H
(U H
.H p.
o a.
•H O
4-1
^ 01
td M
Pu 3
CO
T3 CO
a) a)
CO t-t
I 01
CU 3
CO 4J
cfl n)
0)
»-i 0.
o H
r-( 0)
O H
O 00
IS -rl
H 33
ai
M
60
•H
-------
555
-------
(a) Classical Signal, 100% Visibility
(b) Signal Composed of Individual
Photon Bursts
Figure 11. Classical Versus Photon Limited Visibility Signals.
556
-------
ON LINE PARTICULATE ANALYSIS ON A FLUIDIZED-BED COMBUSTOR
By:
E. S. VanValkenburg, H. N. Frock
Leeds & Northrup Company
North Wales, PA 19454
557
-------
ON LINE PARTICIPATE ANALYSIS ON A FLUIDIZED-BED COMBUSTOR
By:
E. S. VanValkenburg
H. N. Frock
Leeds & Northrup Company
North Wales, Pa. 19454
ABSTRACT
An on-line participate analysis instrument has been
developed by Leeds & Northrup Company under ERDA sponsorship
for monitoring particle loadings in the gas clean up stages
of advanced combustion systems. This instrument utilizes low
angle forward scattering of optically illuminated particles
by across-the-duct measurements to determine their size and
concentration.
A prototype instrument has been designed and construc-
ted to evaluate this means of measuring particles on fluidized
bed combustion systems and field tests have been completed
on this unit at the Argonne National Laboratory. This paper
presents a brief description of the instrument and discusses
results obtained from the Argonne tests.
558
-------
ON-LINE PARTICULATE ANALYSIS
ON A FLUIDIZED-BED COMBUSTOR
SECTION 1
INTRODUCTION
An on-line particulate analysis instrument has been developed by Leeds
& Northrup Company under ERDA sponsorship for in situ measurement of particle
loading and size in the product gas streams of advanced combustion systems.
A prototype instrument has been designed and constructed to evaluate this means
of measuring particles on fluidized bed combustion systems and field tests
have been completed on this unit at the Argonne National Laboratory. This
paper presents a brief description of the instrument and discusses results
obtained from the Argonne tests.
This particulate instrumentation is based on Leeds & Northrup's prior
research in low-angle forward scattering of light by micron size particles
suspended in fluid streams. When such particles are optically illuminated,
the scattered light intensity at any given angle is a function of the size,
shape and index of refraction of the particles. In the case where the wave-
length is small in comparison to the size of the particles, the spatial distri-
bution of the scattered light in the far field is dominated by the volume and
size characteristics of the particles. The design of the ERDA instrument is
based on utilization of simple diffraction theory to convert measurements of
the composite Fraunhofer diffraction pattern for a large number of particles
into meaningful particle data which characterizes the size distribution and
concentration.
This type of instrumentation will be used later to evaluate the per-
formance of secondary particle clean up and to measure the size distribution
and concentration of particles at the inlet to gas turbines in direct combus-
tion coal-fired systems. The latter application requires instrumentation
amenable to measurement of particles in high temperature (1500-2000F)
559
-------
pressurized (up to 10 atmospheres) gas streams and be adaptable to rather
large diameter gas ducts. The prototype instrument will accommodate gas
ducts up to one foot internal diameter.
Field tests were completed at ANL on August 19, 1977 and the proto-
type instrument is nowat Leeds & Northrup Company for calibration recheck.
It will be delivered to Curtiss Wright Corporation soon where it will be
installed on the Small Gas Turbine fluidized bed combustion system. This
installation will enable testing the instrument performance for in situ
measurements at seven atmospheres pressure and 1600F range.
I will first describe the particulate analysis instrument and then
present some of the results from the tests at Argonne National Laboratory.
560
-------
SECTION 2
INSTRUMENT DESCRIPTION
A schematic diagram of the optical train of the ERDA instrument is
shown in Figure 1. This instrument consists of optical elements mounted to
an optical bench which extends under a horizontal run of the combustion system
duct. (The duct is not shown in the schematic, only its vertical center line).
The illumination source is a helium cadmium laser mounted to the underside
of the optical bench. Folding optics, attached to the left end of the op-
tical bench, direct the laser beam across the duct where particles in the
gas stream, which are illuminated by the beam, scatter the light. The
scattered flux is collected by a lens and focused on a set of function masks.
The portions of flux, which are transmitted through the function masks, are
focused by a second lens on to the photomultiplier detector.
The output power of the laser is measured by a silicon photodiode
which detects radiation from the rear Brewster window of the laser. This
signal is displayed on a meter for ease of adjusting the laser alignment to
maintain peak power output (10 to 20 mVI).
A photograph of the prototype instrument is shown in Figure 2. This
instrument consists of the optical subsystem assembly, an electronics
package, laser power supply, and a digital line printer. All units, except
the printer, are contained in NEMA 12 class enclosures to meet industrial
environmental requirements. The laser and receiving optics units contain
thermostat controlled heaters so that the optical subsystem may be installed
on a gas duct which is located outside of a building.
561
-------
The electronics package includes a microcomputer, visual display and
all the operating controls. This unit can be located up to 200 feet from
the optical subsystem. The digital data printer is used to log the output
measurements and can be located remotely from the electronics package, if
desired.
The opening on the optical subassembly from the folding optics on the
left to the collection lens bezel is 30 inches. The vertical clearance
above the optical bench to the laser beam center line is 9-1/2 inches.
These dimensions were chosen to accommodate a 12-inch I.D. duct with a tee
section viewing port extending on each side of the duct.
Mechanical adjustments are provided on the folding optics for align-
ment of the optical train to the duct windows after the instrument is
installed. The instrument can be mechanically mounted to the duct through
the load carrying base structure.
Without going into detail on the theory of optically scattering measure-
ments, I will describe the basic fundamentals to enable those who may not be
familiar with our instrument to understand the nature of its data outputs.
All particles illuminated by the collimated laser beam as it traverses
across the gas duct scatter flux off the beam axis. For particles in the
size range 1-100 microns most of the scattered flux is in the near forward
direction. We collect that flux and through an optical process convert that
data into information concerning the particle size distribution and the
particle loading. The particle loading output is calibrated to be direct
reading in parts per million by volume, i.e., ratio of total volume of
particles oer unit volume of gas.
A common means of presenting size distribution is to plot the popula-
tion of particles as a function of their size as shown in Figure 3A. In
order to obtain data more useful for control purposes, it is possible to
describe size distributions in ways other than number density. Three means
of presenting the same particle distribution are shown in Figure 3. The area
distribution D/\(a)da is the fraction of the total surface area of the
particles within the range a to a + da. This is shown in Figure 3B. Similar-
ly, Figure 3C shows the distribution by volume of particles. This cubic
562
-------
response biases the distribution towards the larger size particles. Thus,
the mean volume size av is always larger than a^ and it can be shown that
the width of the area distribution is
The Leeds & Northrup particle instrument provides the following data
outputs:
dV = volume of particles in ppm
MV = mean volume diameter in microns (2cf^)
MA = mean area diameter in microns (2cnr)
HA = width of area distribution in microns (2a)
563
-------
SECTION 3
PARTICLE INSTRUMENT TESTS AT ANL
Photographs of the prototype instrument installed on a one inch i.d.
pipe at the Argonne facility are shown in Figures 4-6. Figure 4 shows the
optical assembly mounted to the gas duct sample cell. After initial align-
ment, the laser beam is enclosed with a rubber boot from the instrument to
the ports. There is no laser radiation danger to personnel as long as the
equipment is maintained in this buttoned up state. No realignment of the
optics was required during the two month test period.
The data processor and control console are located in the enclosure
shown in Figure 5. The data logging printer is shown on the shelf above
the enclosure. There were no electronic malfunctions during the course of
the tests. The only problem encountered was an unexpected reduction in laser
power. However, this didn't inhibit operation of the instrument even though
the power dropped over a period of a few weeks from 20 to 8 mW.
A close-up of the ANL sample cell is shown in Figure 6. Two sets of
air purge lines were provided by ANL to generate air curtains across each
of the viewing port quartz windows. These optical quality windows were
coated to eliminate reflection at the laser wavelength (442 nm). The win-
dows were cleaned weekly but daily background measurements with clean air
flowing in the duct show insignificant particle deposits between cleanings.
The Leeds & Northrup instrument was operated solo by ANL personnel for
the evaluation tests after a one week break-in and training period.
564
-------
The piping arrangement at Argonne permitted measurement of particles
after the secondary cyclone and after the metal filter which is downstream
from the cyclones. Flue gas flow was directed to the particle instrumenta-
tion by valves to enable measurement of particles at either the input to or
output from the metal filter.
Argonne provided an extractive port upstream of the optical windows
that allowed particle size analysis on sampled material with cascade impac-
tors. In addition, steady state grab particle samples were obtained with
membrane filters.
The results of three combustion test runs utilizing Sewickley coal
and Greer limestone are presented in Table I and Figures 7-9. Table I gives
a comparison of size measurements made with the Leeds & Northrup instrument,
called MICROTRAC, with the reduced data from the Anderson Cascade Impactor
samples.
The average mean volume diameters, MV, for the impactor samples were
calculated from truncated log normal distributions obtained via the Anderson
Impactor. Since the MICROTRAC has a linear size response for particles one
micron in diameter and larger, and a highly attenuated response to submicron
size particles, the lowest channel (submicron region) data points from the
impactor were not used. This provides directly comparable data over the
size range 1-20 microns. The average mean area diameters,MA", were similarly
calculated from the Anderson Impactor data. The differences between the
direct reading, on-line observations via MICROTRAC and the impactor data are
tabulated.
The median diameter, 50th percentile for log normal distributions can be
expressed as Median Diameter = / MV x MA. The results of this computation
are shown in the bottom three rows of Table I.
These results indicate good agreement between optical scattering and
cascade impactor methods for particle sizing. In all but one case, the dif-
ference is less than one micron.
Two interesting characteristics are observed, however. The MICROTRAC
size measurement tends to indicate slightly smaller size for the median
565
-------
diameter and the MICROTRAC consistently shows the size of particles coming
out of the final filter to be larger than at the input. In one of the three
tests, the impactor data also shows the output particle size to be greater
than the input. Further studies are needed to determine whether these are
real characteristics of the particle dynamics, characteristics of the
experimental method or a function of the instrument. It will be interesting
to make similar measurements on a broader distribution of particle sizes and
on a different type of particle separator to ascertain whether these observa-
tions are unique to the Argonne tests.
The loading data as functions of time are shown in Figures 7-9 for
the three operating tests. The in situ volumetric loadings, as outputted by
the MICROTRAC instrument, are converted to standard pressure/temperature
conditions by the following equation:
I (Grains/scf) = -^°367T)dV (2)
where D = density of particulate (gm/cc)
T = gas temperature, nominally 160C
P = gas pressure, nominally 3 atms.
dV = instrument output in ppm.
The density of the material samples in these tests was 1.2 gm/cc.
The data from MICROTRAC are shown as dots for unit intervals of time.
The loading is always high at the beginning of each run due to material
loosened in setting the duct valves. It takes about 15 to 20 minutes for this
to be purged out and reach a steady state.
The MICROTRAC data shows a slow oscillatory characteristic. Its means
of measuring particle size is independent of laser power whereas the indica-
ted loading is directly proportional to laser beam intensity. Similar varia-
tions in the dV output were observed when clean air was directed through the
gas duct -- when a steady state zero was expected. Therefore, it is most
likely that the oscillations were caused by laser instability. This condi-
tion is now being investigated by us to identify cause and determine means
of eliminating this condition.
Smooth lines are drawn through the MICROTRAC data points to show the
566
-------
loading trends. In addition, the loading values obtained by the cascade
impactor and the membrane filter samples are shown. The horizontal location
and length of lines for the extracted sample loadings indicate approximate
time and duration for collection of each of those samples. The vertical
location of each line designates its loading measurement.
A significant part of the variance between MICROTRAC and extracted
samples may be due to non-uniformity of loading across the pipe or, probably
more likely, due to problems of achieving isokenetic sampling with the
extractive probe. A particular advantage of the MICROTRAC type of instrument
is that it measures all particles passing through the laser beam and the gas
flow is not influenced in any way by this method of measurement.
For installations on larger ducts, such as on the 10 inch i.d. duct
at Curtiss Wright, the scattered flux from some particles 1 to 3 microns in
size will be vignetted due to the limited size of the collecting optics. The
scatter angle G . to the first minimum in the diffraction pattern is a
rmn ^
function of wavelength (A) and particle diameter (d) as shown in Figure 1.
Thus, very small particles at long distance from the collector lens scatter
flux outside this lens. To compensate for this factor (vignetting), we
assume that the smallest particles are either uniformily distributed or
distributed concentrically about the center of the line of the gas stream.
The data processing routine provides a weighting factor for particles produ-
cing vignetting, to compensate for their lost flux.
In such installations, the MICROTRAC will detect and measure all parti-
cles traversing across the laser beam which are larger than a specified size
(e.g., 3 microns). The amount of scatter flux collected from particles
smaller than that size is determined by the distance of the collecting lens
from the center line of the duct and the size and quantity of these particles.
Thus, once the installation geometry is specified, the appropriate weighting
factors are determined for compensating that flux which is lost. These
constants are then inserted into the microcomputer program to correct the
size distribution and loading data and provide a uniform response over the
total range of particle size (1 to 20 microns).
567
-------
The effectiveness of this method for compensating micron size particles
in large ducts will be evaluated in the next few months at the Curtiss
Wright facility. The instrument will be installed first on a 4 inch duct
just upstream of the turbine. The receiving optics can be moved on its
optical bench to vary the distance of the collector lens from the duct and
thus compare data output for various distances.
This evaluation will complete our work on the existing ERDA contract
and we will be ready to investigate its application to demonstration and
commercial size combustion system installations.
568
-------
TABLE I: Particle Sizing in Microns
Measurement
Calculated MV
Anderson Impactor
Sample
MICROTRAC MV
Difference AMV
Calculated MA
Anderson Impactor
Sample
MICROTRAC MA
Difference AMA
50 Percenti le
Anderson Impactor
Sample
50 Percentile
MICROTRAC
Difference
In
5.77
2.92
+2.85
2.98
2.15
+0.83
4.15
2.50
+1.65
Out
4.64
5.13
-0.49
2.34
2.39
-0.05
3.29
3.50
-0.21
In
5.39
3.91
+1.48
2.08
2.26
-0.18
3.35
2.97
+0.38
Out
7.44
5.45
+1.99
3.04
2.64
+0.40
4.75
3.79
+0.96
In
5.33
3.62
+1.71
2.20
2.19
+0.01
3.42
2.82
+0.60
Out
4.74
4.39
+0.35
2.58
2.71
-0.13
3.50
3.45
+0.05
569
-------
0)
.c:
o
oo
0)
s_
c
»—I
O)
-------
Figure 2: Prototype Participate
Instrument
571
-------
Number
DN(a)da
Area
DA(a)da
Figure 3
Three Methods of Defining a Particle Distribution
(A) by number,
(B) by area, and
(C) by volume
Particle Radius, a
Area
Mean Radius
aT
Area
Std. Deviation
Particle Radius, a
Volume
Dv(a)da
Volume
Mean Radius
Particle Radius, a
572
-------
Figure 4: Optical Assembly
Installation at ANL
573
-------
Figure 5: Electronics Unit
Installation at ANL
574
-------
Figure 6: Close-up View of ANL
Sample Cell
575
-------
S-
O)
C_J
M-
O
4->
13
Q.
O
n_
x
O
CO
O
r--
C
un
O
ro
C
C\J
o
-------
o
01
I/)
O)
c
o
o
>>
C_5
S-
O)
to
•4-J
QJ
o
-t->
13
C-
C
CO
o
r--
or
UJ
0.
X
LlJ
CO
OJ
S-
. I
(O
00
•1
to
"o;
CJ3
CJ
<=c
o
o:
9
\f
\l
II
. /I
/'
/•
OJ
'c.
£
fO
oo
i- •
o
o
CO
tQ
C-
c
°
CO
s_
QJ
T3
C
eC
O
CM
577
ro
o
OJ
o
-------
14-
o
oS I
o
o
o
(Ti
O
co
O)
D-
E
ro
OO
S_
o
o
re
C
c
I/!
S-
cu
TD
o.
x
o
•sa-
O
CM
CM
O
-------
By acceptance of this article, the
publisher or recipient acknowledges
the U S. Government's right to
retain a nonexclusive, royalty-free
license in and to any copyright
covering the article.
CHARACTERIZATION OF SUSPENDED FLUE GAS
PARTICLE SYSTEMS WITH ON-LINE LIGHT
SCATTERING PARTICLE ANALYZERS
By:
J. C. Montagna, G. W. Smith, G. G. Teats
G. J. Vogel, A. A. Jonke
Argonne National Laboratory
Argonne, IL 60439
579
-------
CHARACTERIZATION OF SUSPENDED FLUE GAS PARTICLE
SYSTEMS WITH ON-LINE LIGHT SCATTERING PARTICLE ANALYZERS
by
John C. Montagna, Gregory W. Smith, F. Gale Teats,
G. John Vogel, and Albert A. Jonke
Argonne National Laboratory
Argonne, Illinois 60439
ABSTRACT
Two light-scattering particle size analyzers have been
tested at ANL in the process development unit (PDU) fluidized-
bed combustion system. The analyzers are (1) a single-particle
analyzer developed by Spectron Development Laboratory and (2)
a multiparticle analyzer developed by Leeds and Northrup.
Particle size distributions and mass loadings determined at
different flue gas duct locations with the Spectron and the
Leeds and Northrup instruments have been compared with those
obtained with (1) an Andersen cascade impactor, (2) a Coulter
counter, and (3) positive filters. These comparison were used
to evaluate the two instruments at their present state of
development.
580
-------
CHARACTERIZATION OF SUSPENDED FLUE GAS PARTICLE
SYSTEMS WITH ON-LINE LIGHT SCATTERING PARTICLE ANALYZERS
INTRODUCTION
In the development of pressurized fluidized-bed combustion (PFBC) systems,
an on-line particle analyzer for the flue gas could provide continuous particle
size and loading analysis without disturbing the off-gas stream. These measure-
ments will be used (1) in measuring the efficiency of upstream particulate-
removing devices (cyclones and filters), (2) establish gas turbine performance
at different particulate loadings, and (3) to protect turbines or a test cascade
in the event of sudden system upsets. In a pressurized FBC system, the flue
gas will be at ^900°C and vLOOO kPa (^10 atm) between the boiler and the turbine.
In the absence of an on-line particle analyzer, routine batch sampling of the
hot off-gas using inertial impactors will be necessary. Batch sampling from a
pressurized hot off-gas environment (described in a following section) is
difficult, and the long time lag between sampling and analysis of samples by
the latter technique is also a disadvantage.
Two types of optical analyzers which have not yet been proven for fluidi-
zed-bed application were evaluated. Both instruments use laser light beams.
One is a single-particle analyzer (a particle morphokinetometer, developed by
Spectron Development Laboratory) which characterizes the scattered light from
each particle (laser interferometer method), and the other is a Microtrac
multiparticle analyzer developed by Leeds and Northrup, which characterizes
the entire particle distribution in its optical path.
Single-Particle Analyzer
The single-particle analyzer measures the sizes of particles and their
velocities by measuring the light scattered from each particle as it crosses
581
-------
an interference pattern generated by the intersection of two laser beams. A
schematic drawing of a typical single-particle analyzer is given in Fig. 1.
The laser beams are directed radially into the flue-gas duct of the ANL PDU
combustion system through specially designed windows. The region of measure-
ment of this instrument, called the probe volume or sample space, is at the
center of the duct where the two coherent beams intersect and generate the
interference pattern. The scattered light signal from each particle in the
sample sapce is detected and is changed to an electric signal shown in Fig. 1.
The light may be detected in either a forward or backward observation mode.
In the ANL evaluation, forward-scattered light was detected. Particle size has
been shown to be a function of (1) the ratio of the AC to DC amplitude of this
signal and (2) the shape of the particle. (Farmer, 1972) The particle sizes
that can be measured depend on the interference fringe spacing (6) in the
sample space. The interference fringe periods and the corresponding detectable
spherical particle size ranges used in this evaluation are given in Table I.
The fringe periods were set by adjusting the angle (a) at which the two laser
beams intersected.
A more detailed description of the principles of the single-particle
analyzer is available in the literature. (Farmer, 1972)
Multiparticle Analyzer
In the more conventional light-scattering technique used in multiparticle
analyzers, a single laser beam is directed into the off-gas duct. The particle
size distribution is obtained by examining the Mie scattered light from
particle-laser beam interaction. All of the particles in the beam's path (off-
gas duct cross section) simultaneously scatter incident light. As a result,
the size of an individual particle cannot be determined. The particle size
distribution can be obtained by examining the total scattering intensity as a
function of angle and laser beam polarization and by comparing the experimental
2
data with theoretical calculations for assumed size distributions. (Weiss and
Frock, 1976)
The multiparticle analyzer used in this work utilizes a new measurement
technique in which three measurements of the scattered light are made. These
measurements are accomplished by means of a uniquely shaped spatial filter
582
-------
which, when placed in the Fraunhofer diffraction plane, transmits the proper
amount of light as a function of scattering angle to give the desired responses.
These measurements may be used to determine the mean diameter and variance of
the distribution for particles in the 1-50 ym range. Also, one measured signal
is proportional to the volume of the particles illuminated, and this can be used
to calculate the concentration of particles (loading) in the fluid stream.
If the type of distribution is known (i.e., normal or log-normal) and if
the distribution is unimodal, the information on mean diameter and variance is
sufficient to completely describe the distribution of the suspended particles.
The distribution of the particles leaving the combustor is multimodal (see next
section) and cannot be characterized with this instrument. The distribution of
the particles leaving the cyclones is unimodal and very nearly log-normal; it
is expected that a multiparticle analyzer is capable of characterizing the sus-
pended particles in the flue gas on either side of the filter upstream from
the turbines. The primary objective in this work was to evaluate these instru-
ments for gas-particle streams that would be expected on either side of such a
filter.
583
-------
SYSTEM AND PROCEDURE FOR FLUE-GAS PARTICLE MEASUREMENTS
The flue gas system of the ANL fluidized-bed combustion system (PDU) has
been modified for these evaluations, as shown in Fig. 2. Windows for particle
analyzers have been installed in two locations; one pair is upstream from the
primary cyclone. The other windows are near the system outlet with the capa-
bility of routing the flue gas past these windows, either upstream or downstream
from the sintered metal filters (S4 and S5, Fig. 2). With this arrange, it is
possible to size (1) the coarse entrained particles from the combustor, (2) the
smaller particles escaping the two cyclones, and (3) the smallest particles
passing through the sintered metal filters (representative of particles that
might enter turbines). The coarse particles leaving the combustor were not
sized with the multiparticle analyzer. Downstream from each window location,
sampling ports have been installed that allow (1) particle size analysis of
representative grab samples with cascade impactors and (2) measurement of
particle loading with membrane filters. Also, steady state particle samples
were obtained from the cyclones and test filter.
The particle size measurements were obtained (1) with on-line particle
analyzers, (2) using an Andersen cascade impactor (described below), and (3)
from steady state samples obtained from the cyclones and/or the test filter.
The steady state samples were analyzed by sieve analysis and with a Coulter
counter.
Coulter counter analyses are performed by suspending the particles in an
electrolyte (2 wt % NaCl in H20); surfactants are used to enhance dispersion.
In the Coulter counter, the suspension is passed through an orifice which
isolates two electrodes. As a particle passes through the orifice, it generates
a resistance pulse; the size of the pulse is proportional to the volume of the
particle or the electrolyte that is displaced by the particle. The measurable
584
-------
size range of particles is limited by the size of the orifice (^2-50% of the
orifice diameter). A distribution is calculated by assuming that the particles
are all spherical. This instrument was calibrated with standard particles (Dow
polystyrene latex particles, pollen, and National Bureau of Standards glass
beads).
Particle size distributions on a weight basis were obtained by assuming
that all observed particles were spheres of equal density and that the particles
observed with the on-line instruments (single-particle and multiparticle
analyzers) were identical to those that were mechanically removed from the
system and later analyzed. Since density is assumed to be constant for all
particle diameters, the fractional volume distribution and fractional mass
distribution are equivalent.
Cascade impactors are the devices used most often for obtained size distri-
butions of airborne particles in process or ambient air in the size range 0.3-
20 ym. In this study, an Adersen cascade impactor was used to obtain combustor
flue gas particle size distribution data for flue gas grab samples. Each stage
of the impactor consists of equidiameter orifices followed by a target plate for
collecting the particles. Smaller orifices are used in successive stages, and
thus smaller particles are collected in successive stages, The particle size
distributions are calculated from experimental data by relating the mass that is
collected on each stage to the corresponding stage diameter. Impactor designs
(including the design of the impactor used) are based on the theoretical devel-
3
opment of Ranz and Wong. (Ranz and Wong, 1952) The size measurements obtained
with the cascade impactors are aerodyanmic diameters (based on aerodynamic
behavior of spherical unit-density particles). This particle measurements
should be most appropriate in characterizing airborne particles in relation to
turbine erosion, which is an aerodynamic process. Previous FBC studies have
shown that measurements of combustion particles with cascade impactors, Coulter
4.5
counters, and microscopes are in good agreement. ' (Vogel et al.y 1974) (Hoke
et al.3 1977) Because of previously found agreement between the different
measuring techniques and of the scope of this evaluation, no direct microscopic
comparisons have been made in this evaluation. However, it is planned to perform
such a comparison during the on-going work on the evaluation of particulate
clean-up components at ANL, which is a much more extensive task.
585
-------
The total cumulative mass distributions obtained with the impactor in two
consecutive measurements of suspended particles leaving the secondary cyclone
(SGL-2C) are given in Fig. 3, Aerodynamic mass diameters of 2.7 urn and 3.0 ym
were obtained, it being assumed that the apparent density of all particles was
O v
1.0 g/cm . The sampling conditions for the impactor samples are given in Table
II. The total cumulative mass distribution obtained with the Coulter counter
is also given in Fig. 3. The log^mean diameter was found to be 3.5 pm, which
compares favorably with the distribution means obtained with the cascade
impactor. Because the Coulter counter was calibrated with standard particles
and its measurements of combustion particles agreed well with cascade impactor
measurements, it was assumed that these comparative measurements were represen-
tative of the true particle distributions. The evaluation of the multiparticle
analyzer was performed only with cascade impactor comparative measurements.
The sampling system in the FBC system for the cascade impactor is illus-
trated in Fig. 4; for some samples, a glass fiber membrane filter was substituted
for the impactor. The particle laden-flue gas flowed by the optical windows
(located in the line downstream from the cyclones or metal filter) , where the
particles were sized with a light-scattering particle analyzer. Next,.the off-
gas line was expanded to reduce the velocity to that compatible to isokinetic
sampling with the cascade impactor and membrane filters. The tip of the
sampling probe (0.78-cm ID for single-partcile analyzer evaluation and 1.33-cm
ID for the multiparticle analyzer evaluation) was machined to enhance aerodynamic
stability near the probe entrance. The sample line was electrically heated to
maintain the temperature of the gas sample above its water dew point. The
cascade impactor and the membrane filter holders was contained in a heated
pressure shell to permit sampling from the pressurized (3 to 8 atm) combustion
system. Gas velocities were calculated from measured gas flows and local
temperatures.
586
-------
CHARACTERISTICS OF SUSPENDED PARTICLE DISTRIBUTIONS
AT DIFFERENT FLUE-GAS CLEANUP STAGES
The total size distribution of particles elutriated in the combustor
during a coal combustion experiment (SGL-1) was obtained by combining Coulter
counter measurements with sieve analyses of samples collected in the cyclones
and metal filter. The particles passing through the metal filter were
justifiably assumed not to contribute significantly to the total mass size
distribution. The fractional mass distribution in sucessive half-volume
intervals of all particles between 2 ym and 1000 ym is given in Fig. 5. (The
loading in the flue gas leaving the combustor is ^14 grains/scf or ^20 grains/
acf which is quite high,) This distribution consists of elutriated partially
sulfated limestone (apparent density; pa = 1.9 g/cm3), unburned coal (pa ^1.0
g/cm3), and coal ash (pa ^ 0.6 g/cm3). Since these materials could not be
separated, the distribution could only be obtained by assuming a uniform
density; thus, the smaller-diameter fractions, which contain more of the
lighter ash and carbon, are probably biased high.
Since the distribution of particles leaving the combustor is multimodal,
the detailed characteristics of the distribution can only be obtained with a
single-particle analyzer such as was used in this work. However, because the
distribution extended from 2 ym to 1000 ym, only that part of the distribution
below 'WO ym could be sized with the single-particle analyzer.
The distribution of the particles escaping from the cyclones during a
different coal combustion experiment (SGL-2C) is given in Fig. 6. The mass
loading at this flue gas location is ^0.2-0.5 grain/scf, and the mass
contribution by particles larger than 10 ym is small. The largest mass fraction
consisted of 3.0 to 3.8 ym particles, which contained 22.5 wt % of the total
loading. This distribution can be characterized as a log-normal distribution
and it is expected that particles escaping from the filter have the same type
587
-------
of distribution. Thus, because the type of pa,rti,cle distribution that a gas
turbine will see is expected to be log^-normal, both single-particle analyzers
and multiparticle analyzers can be used in the hot flue gas of a PFBC system.
Because the flue-gas particle distributions are not homogeneous (carbon,
fly ash, and unburned coal), the extent to which particle shapes, apparent
densities, refractive indices, etc. affected the differences between measure-
ments obtained by different measuring principles (techniques) was evaluated
with experiments in which only virgin limestone particles were suspended and
measured in the flue gas. This was accomplished by continually feeding virgin
limestone into the cold combustor, thereby maintaining a steady state fluid bed
of virgin limestone. The particles that elutriated were generated by attrition
and had basically homogeneous properties. The fractional mass distribution of
virgin limestone leaving the cyclones of the PDU system, was determined with
the Coulter counter and is given in Fig. 5 (LASER-IB). It was also log-normal
in nature and the log-mean diameter was approximately equal to that obtained
for the combustion experiment measurement (SGL-2C). Thus, this distribution
can also be characterized by use of both single-particle and multiparticle
light-scattering analyzers.
588
-------
EXPERIMENTAL EVALUATION OF THE SINGLE-PARTICLE ANALYZER
Some results of particle size measurements obtained with the single-
particle analyzer for several combustion experiments have been compared with
size distributions determined by Coulter counter for steady state particle
samples and cascade impactor measurements. In addition to size measurements,
particulate loading measurements were compared. The measurements with this
laser instrument were made in the PDU combustion system's off-gas duct between
the combustor and the primary cyclone, where all particles elutriated in the
combustbr were observed. Also, particle size measurements between the secondary
cyclone and the metal filters are reported.
Measurements of Particle Size Distribution
The conditions for the combustion experiments in this evaluation are given
in Table II. In the first experiment (SGL-1), the distribution of particles
leaving the combustor was measured. Two fringe periods, see Table I, were used
on the single-particle analyzer, 71.4 ym and 22.3 urn. The measured particles
consisted of limestone fragments, coal ash, and unburned coal.
In the comparison, only the mass distributions in the measurable particle
size ranges (1.5-23 ym and 5-74 ym) of the single-particle analyzer were com-
pared with the corresponding distributions obtained with the Coulter counter.
The small (1.5-23 ym) and the large (5-74 ym) particle measurement comparisons
are given in Figs. 7 and 8, respectively. For the small-particle size range
(Fig. 7), the mass log-mean particle diameters obtained were 8.5 ym with the
Coulter counter and 20 ym with the single-particle analyzer. For the large-
particle size range, the mass log-means of the partial distributions were found
to be 26 um with the Coulter counter and 70 yra, with the single-particle analyzer
(Fig. 8). The difference between the two measurements is greater for larger
particles.
589
-------
The conditions for combustion experiments SGL-2C and SGL-1C, in which the
sizes of suspended particles between the secondary cyclone and the metal filters
of the PDU combustion system were measured, are given in Table II. Particles
with diameters of 0.2 to 3.1 ym, and 1.5 to 23 ym were sized with the single-
particle analyzer. The resulting partial (1.5-2.3 ym) mass distributions are
given in Fig. 9. The mass log-means of these distributions are 3.5 ym (Coulter
counter) and 17 ym (Curve A, single-particle analyzer). In the distribution
obtained with the single-particle analyzer, the mass particule population in-
creased sharply for particles larger than 15 ym. Since at this point the flue
gas had passed through two cyclones, most particles larger than 10 ym should
have been removed, as found with the Coulter counter measurements.
The submicron particles were measured with the single-particle analyzer
in experiment SGL-1C (see Table II), using a fringe period of 2.94 ym. To
achieve the recommended velocity of ^3.0 m/s or less (necessary due to response
of electronics), the flue stream was split downstream from the cyclones and
only a metered portion was allowed through the sampling system. Because the
Coulter counter at ANL is not capable of measuring particles smaller than 1.5 ym,
the small particle measurements made with the single-particle analyzer were
compared with the partial mass cumulative distribution obtained with the cas-
cade impactor (Fig. 10). The mass log-means obtained were 0.72 yra (single-
particle analyzer) and 1.6 ym (impactor). The agreement obtained is considered
to be as good as might be expected for the small particle measurements.
Because of the low loadings (^0.05 grain/scf) downstream from the metal
filter, no size measurements could be made with the single-particle analyzer.
Not enough measurable (1.5-23.9 ym) particles passed through the sample space.
However, the present state of the art on turbine technology suggests that this
particle size range will be significant at loadings of ^0.05 grain/scf in causing
turbine damage.
On-line particle analyzers will be most useful downstream from FBC particle-
removal devices to monitor particle distributions and loadings in the flue gas
that enters gas turbines. Particles <10 ym and >1 ym are expected to erode the
turbines. It is encouraging that the characteristics of the fractional distri-
butions obtained by the two different methods (single-particle analyzer vs
Coulter counter) were the same and that the discrepancy between the two
590
-------
measurements became smaller for smaller particles (<20 Mm). However, a signifi-
cant difference remained between the comparative measurements for 1.5 to 23.9 Mm
particles (the mass log-means obtained with the single-particle analyzer were a
factor of at least 3 larger). Possible reasons for the difference are given
below:
a. The mass loading downstream from the combustor was ^14 grains/scf
(^20 grains/acf) , which is quite high. Thus, for the measurements
upstream from the cyclone, the chance that there would be more
than one particle in the sample space of the single-particle
analyzer was high. Over 98% of the signals were rejected by the
single-particle analyzer because of particle coincidence inter-
ference. (Signal rejection rates between 97% and 90% are con-
sidered acceptable by the manufacturer.)
b. The original calibration for spherical particles, which was
obtained with low number densities of mists and aerosols by micro-
scopic measurements, is greatly influenced by particle shape and
orientation in the sample space. The bias becomes more pronounced
for larger particles via visibility function. (Farmer, 1972).
c. Fragile particles might have broken up as the particle samples
were collected in the cascade impactor or in the cyclones and test
filter or as they were prepared for Coulter counter analysis
(dispersed in an electrolyte). This effect could impose a bias
towards small diameters in Coulter counter and impactor measure-
ment.
The agreement in the cold experiments between the single particle analyzer
and impactor was better for the measurements with the lower loading (downstream
from the cyclones). However, because the suspended virgin limestone particles
were relatively homogeneous, similar in shape (not dependent on diameter), and
consistent in scattering properties (refractive indices), a better than the
found agreement was expected between the single-particle analyzer measurements,
the measurements with the cascade impactor (seen in Fig. 11), and the Coulter
counter.
Because of difficulties in evaluating the causes for discrepancies
between the single-particle analyzer and comparative measurements, an empirical
591
-------
correlation of the experimental data wa.s. used to obtain a best fit for the
calibration curve. In this analysis, it was assumed that the Coulter counter
and inertial impaction measurements were correct. The particle distribution
measurements obtained with the Coulter counter and cascade impactor were expres-
sed as histograms, with intervals equivalent to the single-particle analyzer
increments. The diameter-dependent factor necessary to force the fractional
contributions of the histogram intervals from the single-particle analyzer into
agreement with those from the Coulter counter and impactor measurements was
obtained. The ratios of the fractional contributions obtained from the Coulter
counter and impactor measurements to those obtained from the single-particle
analyzer for a considerable number of measurements were correlated with the
reduced diameter of the interval, D/6. (D is the particle diameter and 6 is
the fringe period of the laser probe volume.) By use of least squares
techniques, the following correlation was obtained:
In (K) = -4.91 ln(|) - 2.5 [ln(|)]2 (1)
where K is the ratio of the expected mass concentration in the diameter interval
to the measured number of particles within the analyzer interval. This correl-
ation gives a fair fit (the correlation coefficient is M3.88).
The above discussed experiments were reanalyzed using this empirical cali-
bration and the measurements in these experiments were adjusted using the
calibration function (Eq. 1). The results for SGL-2C, in which the particle
measurements were performed on suspended particles downstream from the cyclones,
are given in Fig. 9 (Curve B). This figure also contains the distribution
which is based on the calibration function originally supplied by the manufac-
turer (Curve A):
K - I/D (2)
The mass log-mean diameters for SGL-2C (Fig. 9) were: 3.5 urn (Coulter
counter), 8.8 ym (Eq. 1 calibration), and 18 ym (supplied analyzer calibration).
Although the adjusted (by Eq. 1) analyzer measurements still deviate from the
Coulter counter measurements, agreement of the measurements improved signifi-
cantly.
592
-------
Measurement of Parj_iculate Loadings
Loading measurements cannot be made with the single-particle analyzer in
its present state of development because the appropriate volume of the sample
space is not readily attainable without a comparison measurement. Comparative
loading measurements by gravimetric means were used to predict the probe volumes,
and it was estimated that at best, the loadings could be estimated within one
order of magnitude with the single-particle analyzer for high loading conditions
(>0.1 g/m3). (Montagna e~t at., 1977) No particle measurements were made with
the analyzer under low loading conditions because too few particles passed
through the probe volume.
593
-------
EXPERIMENTAL EVALUATION OF THE MULTIPARTICLE ANALYZER
Windows for the multiparticle analyzer were installed near the system
outlet with the capability of routing the flue gas past these windows, either
upstream or downstream from the metal filters. The sampling port for repre-
sentative grab samples was installed upstream from the window location. Steady-
state grab particle samples were obtained with: the cascade impactors, membrane
filters, and a total flue gas test filter. (The test filter replaced
the metal filter during sampling periods, but loadings obtained with this
filter are not reported in this paper.)
Measurement of Log-Mean Particle Size
The log-mean aerodynamic particle diameter (D ) measurements obtained
Alj
with the impactor may be compared with the "projected" particle area mean
diameter (D ) obtained with the multiparticle analyzer. The volume mean dia-
1\
meter D , also obtained with the multiparticle analyzer, is larger by
definition:
In DT7 = In li + 0.5 In2 a (3)
V A n
where a is the geometric standard deviation on a number basis; (Irani and
Callis, 1963) D is more sensitive to changes in the distribution. D cor-
V A
responds more closely than does D to impactor diameter measurements, DATn,
V o AE
which are dependent on the particle's aerodynamic drag and inertia. (Orr and
Dalla Valle, 1960) The multiparticle analyzer also measured the arithmetic
deviation of the distribution. The geometric standard deviation, o , can be
calculated from Eq. 1 and the measurements D. and DV. The calculated value
of a and D. were used to characterize the loe-normal distribution of sus-
n A , & •
pended flue gas particles. (Montagna et al., 1977) The resulting distri-
butions from the multiparticle analyzer are slightly narrower (less variance)
than the distribution from the cascade impactor.
594
-------
The conditions, for the combustion experiments a.nd some cold experiments
are given in Table III together with the measured diameters and loadings.
Because the contribution of particles smaller than 1 ym was less than 10%, mass
basis, the impactor distributions were not truncated to accommodate the range
of the multiparticle analyzer (1-20 ym-dia). In coal combustion experiments,
the aerodynamic log-mean diameter (D.g) ranged from 2 to 3.5 ym and the pro-
jected area mean diameter (D ) ranged from 2.2 to 3.4 urn for all suspended
A.
particle measurements when the loading (gravimetrically obtained) was relatively
high (>0.1 g/m3 or >0.044 grain/scf). The volume mean diameters (D ) ranged
from 2.9 to 5.5 ym for the same measurements.
In "cold" experiment LN-9-5 , only virgin limestone particles were measured.
The loading was also relatively high (>0.1 g/m3); D was 2.7 ym, in comparison
r\.
with 1.5 ym for D . It had been expected that the measurements made on the
At
suspended virgin limestone would have agreed better than those made on the
mixture of particles during combustion experiments. However, the (DAT7 and D )
AH* A.
agreement for suspended combustion particles was slightly better than for the
above measurements under high loading (>0.1 g/m ) conditions. In comparison
with the multiparticle analyzer, the mass log-mean diameters of particles
leaving the cyclones (loadings >0.i g/m3) measured with the single-particle
analyzer (only particles between 1.5 and 23.9 yn were compared) were at best
a factor of three greater than those obtained by the Coulter counter and
impactor measurements.
The agreement of D with D was poorer for measurements made at lower
AJtj A.
loading conditions (£0.01 g/m3) in experiments LN-9 and LN-11. D ranged
A£J
from 0.63 to 1.9 ym and D ranged from 2.0 to 5.5 ym. The loading of 'vO.OS g/m3
A-
for particle diameters of 1-10 ym is presently being proposed as the approxi-
mate maximum tolerable loading for turbines. Because the particulate loading
tolerances on turbines remain uncertain, it is desirable to have optical
instruments capable of measuring particle loadings <0.05 g/m3. In these low
loading measurements, D. was approximately three times j.arger than D.p.
Considering the difficulty of obtaining accurate impactor measurements for the
comparison, this agreement can be considered fair at this stage. In comparison,
no particle measurements could be made with the single-particle analyzer at
these low loadings because not enough particles passed through the sample space
of the instrument.
595
-------
Measurement of Particulate Loadings
In these experiments, particulate loadings were, measured in addition to
mean particle sizes. Values for comparison were obtained by three backup
methods. The amount of particulate collected on a total flue gas metal filter
located downstream from the multiparticle analyzer windows and sampling port
provided one backup measurement. The other backup measurements were grab •
samples, and each consisted of (1) the particulate collected in the impactor
including a backup filter to the imp-actor and (2) the material collected on a
membrane filter.
For the measurements during the combustion experiments in which the loadings
were high (>0.1 g/m3, gravimetric), the gravimetric loadings were a factor of
four or less smaller than the optical loading measurements made with the multi-
particle analyzer, see Table III. (The loadings from the multiparticle analyzer
are proportional to the particulate density assumed; 1.0 g/cm3.) -for
many high loading measurements, such as LN-4-2, the gravimetric and multiparticle
analyzer loadings were very close. The gravimetric loadings under low loading
conditions (<0.01 g/m3) were one order of magnitude less than the optical
multiparticle analyzer loadings. The semicontinuous background loading sig-
nal is compared with the background plus sample loading signal in Fig. 12
for low loading (j-jO.Ol g/m3) measurements. The loading measurements signal
is only 4% greater than the background; this can account for the order of
magnitude difference between the multiparticle analyzer measurements and the
gravimetric loading measurements. These low loading measurements were made
on suspended particles that had escaped through the metal filter.
In the cold experiment (LN-9), loadings determined gravimetrically were
one order of magnitude smaller than those obtained in the hot experiments for
the high loading measurements; they were about two orders of magnitude smaller
for the low loading experiments. This large difference may be due to an increase
(over the period of an hour) in the multiparticle analyzer background signal
during these cold experiments (see Fig. 13). The average gas linear velocity
in the flue gas duct near the optical windows was ^30 m/s during the cold
experiments; during the hot combustion experiments, the linear gas velocity
was only ^3 m/s. A higher gas velocity creates more turbulence and a greater
chance for particles to build up in the free space between the windows and
596
-------
and the flue ga,s flow stream. The background, was observed to decay to
normal after flow of limes.tone-laden air was turnded off.
597
-------
CONCLUSIONS
The single-particle analyzer is capable of characterizing a suspended-
particle distribution as a histogram. Thus, the actual shape of the size
distribution can be obtained. The multiparticle analyzer can characterize a
distribution by giving the mean and a nongeomtric variance of the distribution.
Unless the type of distribution is known and it is unimodal, the latter instru-
ment cannot describe the distribution of the particles. The distribution down-
stream from the cyclones of a PFBC system is expected to be log-normal, based
on measurements in this and other studies. Therefore, the distribution can
also be obtained with multiparticle analyzer measurements. However, the single-
particle analyzer has a greater capability to characterize particle distribu-
tions .
For high-loading (>0.1 g/m3) measurements, the projected particle area
mean diameters obtained with the multiparticle analyzer were generally <25%
smaller than the aerodynamic mass log-mean particle diameters determined with
the impactor. In comparison, for the same loading conditions, the mass log-
mean particle diameters obtained with the single-particle analyzer were a
factor of three (or more) greater than those obtained by Coulter counter and
impactor measurements. For low loading conditions (<0.01 g/m3), the mean
diameters obtained with the multiparticle analyzer were approximately three
times larger than log-mean diameters obtained with the impactor. No particle
size measurements could be made at these low loadings with the single-particle
analyzer.
For high loadings, the gravimetric loadings were a factor of four or less
smaller than the optical loading measurements made with the multiparticle
analyzer. However, many measurements made at these relatively high loadings by
598
-------
the two techniques were very close. Under low loading conditions (<^0.01 g/m3),
the gravimetric loadings were one order of magnitude less than the optical
measurements with the multiparticle analyzer. For these low loads, the loading
measurement signal is only 4% greater than the background; this is a potential
source of error. If the background can be reduced electronically or via
cleaner windows and/or optical path, this instrument would be very promising
for on-line PFBC measurements to protect the gas turbines.
For the single-particle analyzer, loading measurements cannot be made at
its present state of development because the appropriate volume of sample space
is not readily calculable. Based on comparative measurements by gravimetric
means, it was estimated that at best, loadings could be estimated within one
order of magntidue with the single-particle analyzer for high loads. No low
loading (<0.01 g/m ) measurements have been made with the single-particle
analyzer.
The multiparticle analyzer is a state of development allowing it to be
used with little operator training. Operation of the single-particle analyzer
requires much more care, and data reduction is presently very time-consuming.
The single-particle analyzer is a first-generation instrument and has much
potential. On the other hand, the multiparticle analyzer is a more advanced
instrument and with some refinement can be useful now in the development of
PFBC technology.
599
-------
ACKNOWLEDGMENTS
The support of this program by the Energy Research and Development Agency
and the Environmental Protection Agency is gratefully acknowledged. This study
was made under the direction of Messrs. A. A. Jonke, D. Webster, and L. Burris
of the Chemical Engineering Division. Experimental data were obtained by Messrs,
H. Lautermilch, S. Smith, J. Stockbar, and A. Ziegler.
600
-------
REFERENCES
1. W. M. Farmer, Measurements of Particle Size, Number Density, and Velocity
Using a Laser Interferometer, Applied Optics 10, 2603 (1972).
2. E. L. Weiss and H. N. Frock, Rapid Analysis of Particle Size Distribution
by Laser Light Scattering, Power Technology 14, 287 (1976).
3. W. E. Ranz and J. B. Wong, Impaction of Dust and Smoke Particles, Ind. Eng.
Chem. 44(6), 1371 (1952).
4. G. J. Vogel, W. M. Swift, J. F. Lenc, P. T. Cunningham, W. I. Wilson, A. F.
Panek, F. G. Teats, and A. A. Jonke, Reduction of Atmospheric Pollution by
the Application of Fluidized-Bed Combustion and Regeneration of Sulfur-
Containing Additives^ Annual Report, July 1973-June 1974, ANL/ES-CEN-1007
(1974).
5. R. C. Hoke, et al., A Regenerative Limestone Process for Fluidized Bed Coal
Combustion and Desulfurization, Monthly progress Report No. 77, July 1977.
6. J. C. Montagna, G. W. Smith, F. G. Teats, G. J. Vogel, and A. A. Jonke,
Evaluation of On-Line Light-Scattering Particle Size Analyzers for
Measurements at High Temperature and Pressure, ANL/CEN/FE-77-7 (in prepara-
tion) .
7. Riyad R. Irani and Clayton F. Callis, Particle Size: Measurement,
Interpretation, and Application, p. 43, John Wiley & Son, Inc., New York,
1963.
8. C. Orr and J. M. Dalla Valle, Fine Particle Measurement, Size, Surface and
Pore Volume, pp 83-91, McMillan, New York, 1960.
601
-------
TABLE I. SELECTED INTERFERENCE FRINGE SPAGINGS AND
THE CORRESPONDING MEASURABLE SPHERICAL
PARTICLE SIZE RANGES
Fringe Min. Particle Max. Particle
Period, Diameter, Diameter,
ym ym ym
71.4 4.9 74
22.3 1.5 23
2.94 0.2 3.1
602
-------
TABLE II. EXPERIMENTAL CONDITIONS FOR A COMBUSTION EXPERIMENT
IN THE EVALUATION OF THE SINGLE-PARTICLE ANALYZER
Location of PM Windows: Between PDU combustor and
cyclones, SGL-1.
Between second cyclone and
filter, SGL-2C.
Sorbent: Greer Limestone
Coal: Sewickley
System Pressure, kPa: 308 (3 atm)
Fluidizing Gas Velocity, m/s: 1.0
Conditions Near Probe
at Sampling Duct
Exp.
SGL-1
SGL-2C
SGL-1C
Combustor
Temp,
°C
850
855
855
Conditions at
PM Windows
Gas
Velocity,
m/s
5.2
11.8
2.6
Temp,
°C
350
123
90
Ratio of Duct
Gas Velocity to
Velocity Probe Gas
Vfg Temp, Velocity
m/s' °C Vfg/Vs
3.26 110 1.16
0.91 60 1.15
603
-------
TABLE III. COMPARISON OF PARTICLE SIZES AND LOADINGS
OBTAINED WITH THE MULTIPARTICLE ANALYZER, THE
ANDERSEN IMPACTOR, AND MEMBRANE FILTER
FBC Combustor Conditions:
Pressure 308 kPa (3 atm)
Temperature 855 °C
Fluidizing Velocity 1 m/sec
Sorbent Greer Limestone
Coal Sewickley
Exp.
LN-4-l-a
LN-4-l-b
LN-5-l-a
LN-5-2-b
LN-6-l-a
LN-6-2-b
LN-10-3-b
LN-10-4-b
LN-11-l-a
LN-ll-2-a
LN-ll-3-a
LN-9-3-ag
LN-9-2-a§
LN-9-5-bS
LN-9-4-b8
MICROTRAC
— c
(ym)
5.1
2.9
5.5
3.9
4.4
3.6
3.8
-
11.9 +.3
-
10.5 +.2
2. 24+. 36
-
3. 77+. 23
-
~" i
DA
(ym)
2.8
2.2
3.4
2.3
2.8
2.2
2.2
-
5.5 +.11
-
4. 97+. 12
2. 01+. 06
-
2. 72+. 07
—
Loading
Andersen
Impactor
DAE*"
L
(g/m3)a (ym) (g/m3)
0.43
0.83
0.18
1.0
0.46
0.64
1.5
0.37
0.12+.01
0.13+.01
O.llf.02
0.21+.04
0.10+.02
1.20+.06
0.52K17
2.2
3.5
6.0f
3.2
3.0
3.5
2.0
-
1.9
-
1.8
0.63
-
1.5
-
0.33
1.1
0.21
0.98
0.12
1.1
0.56
-
0.01
-
0.01
0.001
-
0.115
-
Membrane
Filter
Loading
(g/m3)
0.64
1.3
0.41
0.40
0.54
1.0
_
0.63
0.009
0.003
0.110
al g/m3 = 0.437 grain/scf
^After filter.
Between secondary cyclone and filter.
^Volume mean diameter.
Projected area mean diameter.
Aerodynamic mean diameter.
Suspect, possibly due to unexpected flue gas re-entrainment.
gCold elutriation experiments - only virgin limestone particles measured.
604
-------
INTERFERENCE FRINGES
>/e2 MODULATION CONTOUR
-Z
LASER
ENLARGED VIEW
OF REGION OF
CROSS-FOCUS POINT
FORWARD SCATTER
OBSERVATION MODE /
C
PMT
TIME
Fig. 1. Spectron Development Laboratory's PM Analyzer System
for Velocity and Particle Measurement
605
-------
COAL
AIR-*
am
PREHEATER
L
-*
k.
S
SOF
/
&
}BENTT
i
i
i
>jk
PA CYCLONES FILTERS
-^-^mtw
1
O
SPO SI S2
COMBUSTOR
(S4S5
*ro-[po-
S3 '
-TEST
FILTER
OSP
PA
PA-WINDOWS FOR PARTICLE
ANALYZERS
SP-SAMPLE PORT FOR SAMPLING
WITH SAMPLE CYCLONE AND
CASCADE IMPACTOR
Fig. 2. Schematic of FBC System with Modified Flue-Gas System
606
-------
99.9
99
98
95
90
ID 80
1 70
§ 60
w 50
if\
40
30
20
>
10
O.I
0.5
V COULTER COUNTER ANALYSIS
O O ANDERSON IMPACTOR
5 10
PARTICLE DIAMETER,
50
100
Fig. 3. Cumulative Mass Distribution of Particles
in the Flue Gas Between the Secondary
Cyclone and the Metal Filter (SGL-2C)
607
-------
TC - THERMOCOUPLES
PI -PRESSURE INDICATOR
FROM FLUE-GAS
SYSTEM
TO FLUE -GAS
SYSTEM
AN
LIN
„
1 C
4LY£LR 4
E
SAMPLING
PROBE (3/8 in.K
v.— —
2 in 10-
y
A
PA
— tfc) PROBE PURGE
HL
§-p
,„ PREHEATING
H AIR
M HL_ [^-
W CYCLONE
IMPACTOR
[—PRESSURE SHELL
-1 in. ID
Fig. 4. Flue Gas Particle Sampling System
608
-------
10 20 50 100 200
PARTICLE DIAMETER, ^m
500 1000
Fig. 5. Fractional Mass Distribution
of All Elutriated Particles
during a Combustion Experi-
ment (SGL-1) from the ANL FBC
Combustor
609
-------
tt
UJ
I-
bJ
Q
UJ
_J
O
K
01
en
CO
UJ
CO
o;
4-1
fO
D-
O CO
QJ
C C
O O
4-> O
=3 >1
J3 O
00 >,
co CO
ns
£1 co
(O
i— CD
tO
C. O)
O 3
O
fO C_)
S- CQ
CD
U_
%c(3!flrnoA) SSVIAJ
610
-------
99
98
95
90
*
- 80
U
1 TO
10
5
2
I
0.5
02
O.I
V COULTER COUNTER
O SPECTRON PM
I I I I I
10
PARTICLE DIAMETER ,
30
Fig. 7.
Comparison of the Partial (1.5-23
Cumulative flass Distribution of
Particles Leaving the Combustor
(SGL-1)
611
-------
99
98
99
90
X
~ 80
UJ
I 70
o 60
* 50
X 40
» 30
S zo
s
=J 10
2
I
o.s
02
O.I
1 '' i ; '
COULTER COUNTER
SPECTRON PM
10
PARTICLE DIAMETER,
100
Fig. 8. Comparision of the Partial
(5-74 ytn) Cumulative Mass
Distribution of Particles
Leaving the Combustor (SGL-
1)
612
-------
999
SPECTRON PM
(ANL CALIBRATION)
SPECTRON PM
(SUPPLIED CALIBRATION)
° COULTER COUNTER
, , I , , ,
5 10
PARTICLE DIAMETER,
50 100
Fig. 9. Comparison of the Partial
(1.5-23.8 ym) Cumulative
Mass Distribution of
Particles Leaving the
Secondary Cyclone (SGL-2C)
613
-------
999
99
90
o 70
50
10
01
O ANDERSON IMPACTOR
DSPECTRON PM
01
5 I 2
PARTICLE DIAMETER, Mm
Fig. 10. Comparison of the Partial
(0.2-3.1 ym) Cumulative
Distribution of Particles
Leaving the Secondary
Cyclone (SGL-1C)
614
-------
999
LJ
5
O
<
5
13
O
o COULTER COUNTER
aSPECTRON PM
5 10 20
PARTICLE DIAMETER, /i m
Fig. 11. Comparison of the Partial (1.5-23.4 ym)
Cumulative Distribution of Particles
Leaving the Secondary Cyclone during a
Cold Fluidization Experiment (LASER-4A-2)
615
-------
IB!~_ i i i i r
,gi —o— BACKGROUND PLUS SAMPLE
—*— BACKGROUND
% 1-4
1.0
'•2| SAMPLE SIGNAL
-r-
- '~
15 30 45 60 75 90
TIME, min
Fig. 12. A Comparison of the Semi-
continuous Background Signal
with the Background plus
Sample Signal from MICROTRAC,
Experiment LN-11
616
-------
BACKGROUND PLUS SAMPLE
30 60
TIME , mm
H
90
Fig. 13. The Semi-continuous Partic-
ulate Lading Signal from the
MICROTRAC Particle Size Analyzer
for Measurements LN-9-4 and -5
617
-------
LIST OF PROCEEDINGS RECIPIENTS
EPA/ERDA SYMPOSIUM ON HIGH
TEMPERATURE/PRESSURE PARTICULATE
CONTROL
SEPTEMBER 20-21, 1977
Jim Abbott
EPA
IERL-RTP
Research Triangle Park
North Carolina 27711
N. Abuaf
Brookhaven National Laboratory
Upton Long Island
New York 11973
Richard L. Adams
Wheelabrator-Frye Inc.
600 Grant street
Pittsburgh PA 15219
Jeffery C. Alexander
M.I.T.
Room 36-323
Cambridge, MA 02139
Francis M. Alpiser
Chemical Environmental Engineering
EPA III—AHMD-SIP
Sixth and Walnut
Curtis Building Ms-3A-Hll
Philadelphia, PA 19106
Gerald L. Anderson
Institute of Gas Technology
3424 South State Street
Chicago, IL 60616
Herman B. Anderson, Jr.
General Services Administration
Region 3, PBS
Technical Field Office
7th & D Sts., SW
Washington, DC 20407
Dr. John W. Anderson
Rexnord, Inc.
1914 Albert Street
Racine, WI 53404
Larry W. Anderson
Acurex/Aerotherm
485 Clyde Avenue
Mountain View, CA 94042
Stig Andersson
STAL Laval
S-61320 Finspong
Sweden
D. H. Archer
Westinghouse
Beulah Road
Pittsburgh, PA 15235
Ted Atwood
Process Engineering
Department of Energy
Room 504
20 Massachusetts Avenue
Washington, DC 20545
-B-
Dr. Suresh P. Babu
Institute Gas Technology
3424 South State Street
Chicago, IL 60616
W. D. Bachalo
Spectron Development Labs.
3303 Harbor Blvd.
Costa Mesa, CA 92626
E . T. Barrow
Ministry of the Environment
Air Resources Branch
44880 Bay Street
Toronto, Ontario MS5 128
Canada
Walter A. Baxter
Environmental Elements Corp.
P.O. Box 1318
Baltimore, MD 21230
-------
Roland Beck
Department of Energy
7374 S. Forest
Whittier, CA 90602
Robert W. Bee
Consultant
126 Hopeland Lane
Sterling, VA 21170
K. Bekofske
General Electric Co.
CR&D
P.O. Box 8
Schnectady, NY 12301
Michael Beltran
Beltran Assoc., Inc.
1133 E. 35th Street
Brooklyn, NY 11210
Dr. Michael Benarie
Chef Du Services Pollution
Stmospherique
Institute National De Recherche
Chimique Applique I.R.Ch.A.
Centre De Recherche
91710-Vert-Le-Petit, Prance
Paul A. Berman
Westinghouse Electric Corp.
P.O. Box 9175, Lester Branch
Philadelphia, PA 19113
Dr. Samuel Bernstein
Flow Research Co.
P.O. Box 5040
Kent, WA 98031
Albert J. Bevolc
Associate Physicist
US Department of Energy
Ames Laboratory
A205 Physics
Ames, IA 50011
Prem Sagar Bhardwaja, PhD
Lawrence Berkeley Laboratory
Univ. of California
Berkeley, CA 94720
Charles E. Billings
Environmental Engineering Science
740 Boylston Street
Chestnut, MA 02167
Samuel J. Biondo
Federal Power Commission
825 N Capitol Street, NE
Room 5008
Washington, DC 20426
David Blake
Acurex/Aerotherm
485 Clyde Avenue
Mountain View, CA 94042
Dale Blann
Acurex/Ae rothe rm
3301 Woman's Club Drive
Raleigh, NC 27611
Frederick L. Blum
Research-Cottrell Inc.
P.O. Box 750
Bound Brook, NJ 08805
B. H. Bochenek
U.S. EPA
401 M Street, SW (A-134)
Washington, DC 20460
Dr. R. Boericke
General Electric Co.
Building 2, Room 712
One River Road
Schnectady, NY 12345
Bruce Boggs
Amax
5020 S Atlanta Rd, SE
Smyra, GA 30080
Dr. R. H. Boll
Babcock and Wilcox
Research Center
P.O. Box 835
Alliance, OH 44601
J. Bormans
General Secretary Ecochem
FICB-ECOCHEM
Square Marie-Louise 49
1040 Brussels, Belgium
-------
Jeffery Bradley
Research Associate
University of Wisconsin
at Milwaukee
3200 N Cramer Street
Milwaukee, WI 53201
William L. Brangers
U.S. Army Environmental
Hygiene Agency
Abeerdeen Proving Grounds
MD 21010
Ed Brooks
TRW
One Space Park Drive
Redondo Beach, CA 90278
Beth A. Brown
Aspen Systems Corporation
20010 Century Blvd.
Germantown, MD 20767
Robert F. Brown
Research-Cottrell
P.O. Box 750
Bound Brook, NJ 08805
Warren L. Buck
Argonne National Lab.
9700 S Cass Avenue, Bldg. 308
Argonne, IL 60439
Charles L. Burton
Combustion Engineering
1000 Prospect Hill Road
Windsor, CT 06095
Chris Busch, President
Spectron Dev. Lab.
3303 Harbor Blvd.
Costa Mesa, CA 92606
John R. Bush
Research-Cottrell
P.O. Box 750
Bound Brook, NJ 08805
-C-
Captain Jesse B. Cabellon
US Army Environmental
Hygiene Agency
Abeerdeen Proving Grounds, MD 21010
Dr. Seymour Calvert
Air Pollution Technology, Inc.
4901 Morena Blvd., Suite 402
San Diego, CA 92117
Robert Carlson
Acurex/Aerotherm
485 Clyde Avenue
Mountain View, CA 94042
Dr. R. L. Carpenter
Lovelace Biomedical and
Environmental Research Institute
P.O. Box 5890
Albuquerque, NM 87115
Captain James W. Carroll
US Army Environmental Hygiene Agency
Abeerdeen Proving Grounds, MD 21010
John J. Casey
Sr. Marketing Engineering
Fluidyne Engineering Corp.
5900 Olson Memorial Highway
Minneapolis, MN 55422
Gilbert K. C. Chen
Monsanto
800 N Lindbergh Blvd.
St. Louis, MO 63166
Gregory H.S. Cheng
Environeering, Inc.
4233 N United Parkway
Schiller Park, IL 60176
C.K.H. Choi
TRW Energy Systems Group
Bldg. R4/Room 2136
One Space Park Drive
Redondo Beach, CA 90278
Robert A. Chronowski
MITRE Corporation
Westgate Research Park
McLean, VA 22101
Dr. David F. Ciliberti
Westinghous R&D Center
Beulah Rd, Bldg. 501-3E29
Neil H. Coates
MITRE Corp./Metrek Div.
1820 Dolly Madison Blvd.
McLean, VA 22101
-------
John F. Cobianchi
J.P. Stevens & Company, Inc.
1185 Avenue of the Americas
New York, NY 10036
Dr. Hugh Coleman
Sandia Laboratories
P.O. Box 5800
Albuquerque, NM 87115
Leland Collins
Mech. Eng. Dept.
Stanford University
Stanford, CA 94305
Steven D. Colome
Harvard School of Public Health
665 Huntington Avenue
Boston, MA 02115
John Colton
TRW
7600 Colshire Drive
McLean, VA 22101
James Colyar
Process Engineering
Cogas Development Company
P.O. Box 8
Princeton, NJ 08540
Peter Costanza
Staff Scientist
Textile Research Inst.
P.O. Box 65
Princeton, NJ 08540
Glenn Coury
Coury & Associates, Inc.
7400 W. 14th Avenue
Lakewood, CO 80214
Fred Cox
Menardi Southern
1201 W. Francisco Blvd.
Torrance, CA 90502
Rufus Crawford
Stearns-Roger Eng. Co.
P.O. Box 5888
Denver, CO 80217
Steven R. Cross
Buell-Envirotech
253 N. Fourth Street
Lebanon, PA 17042
T. E. Ctrvtnicek
Monsanto-Dayton Labs
1515 Nicholas Road
Dayton, OH 45407
John S. Cvicker
Foster-Wheeler Energy Corp.
110 S. Orange Avenue
Livingston, NJ 07039
-D-
Alberta M. Dawson
Sala Magnetics, Inc.
247 Third Street
Cambridge, MA 02142
John C. Dempsey
Program Manager
US Department of Energy
Mark N. Director
Head, Gas Physics Section
Atlantic Research Corp.
5390 Cherokee Avenue
Alexandria, VA 22314
Thomas Donnelly
Donaldson Company, Inc.
P.O. Box 1299
Minneapolis, MN 55440
John M. Ducan
Shell Development
P.O. Box 2463—One Shell Plaza
Houston, TX 77001
Frederick J. Dudek
UOP, Inc. Air Correction Div.
P.O. Box 1107, Tokeneke Road
Darien, CT 06820
-E-
Sally Edler
Bendix Environmental & Process
Instruments Div.
1400 Taylor Avenue
Baltimore, MD 21204
Heinz L. Engelbrecht
Wheelabrator-Frye Inc.
600 Grant Street
Pittsburgh, PA 15219
-------
Harold M. Englund
Editor, APCA Journal
Air Pollution Control Association
P.O. Box 2861
Pittsburgh, PA 15230
Joseph Epstein
MHD Division
US Department of Energy
20 Massachusetts Avenue, NW
Washington, DC 20545
George Erskine
MITRE Corp./Metrek Div.
1820 Dolly Madison Blvd.
McLean, VA 22101
-F-
James Facet
Facet Enterprises
434 w. 12 Mile Road
Madison Hts., MI 48071
B. P. Faulkner
Allis-Chalmers Corp.
P.O. Box 512
Milwaukee, WI 53201
William Fedarko
US Department of Energy
20 Massachusetts Avenue, NW
Washington, DC 20545
Paul L. Feldman
Research-Cottrell
P.O. Box 750
Bound Brook, NJ 08805
Lewis K. Felleisen
EPA, Region III
Curtis Building
Sixth and Walnut Streets
Philadelphia, PA 19106
Oliver Foo
MITRE/Metrek Div.
Westgage Research Park
McLean, VA 22101
Paul E. Fredette
Midland-Ross Technical Center
P.O. Box 907
Toledo, OH 43691
Max Friedman
Chemico Air Pollution Control Co.
One Penn Plaza
New York, NY 11801
H. N. Frock
Leeds & Northrup
Dickerson Road
North Wales, PA 19454
Robert Frumerman
Frumerman Associates
5423 Darlington Road
Pittsburgh, PA 15217
Dale Furlong
Buell Emission Control Div.
Envirotech Corp.
200 N. Seventh Street
Lebanon, PA 17042
-G-
John A. Garbak
Montgomery County
Dept. of Environmental Protection
6110 Executive Blvd., Room 340
Rockville, MD 20852
John A. Garbak
Environmental Engineering
Montgomery County
1909 Hyannis Ct. #202
McLean, VA 22101
Betty Gaskins
Library Services
Pollution Abstracts
Data Courier Inc.
620 South 5th Street
Louisville, KY 40202
John Gefkin
Department of Energy
Fossil Energy/Advanced Power Systems
20 Massachusetts Avenue, NW
Washington, DC 20545
Peter Gelfand
Buell Emission Control Div.
Envirocech Corp.
200 N. Seventh Street
Lebanon, PA 17042
-------
Masood Ghassemi
TRW, Bldg. R4/Room 1128
One Space Park Drive
Redondo Beach, CA 90278
James A. Gieseke
Batelle-Columbus Labs.
505 King Avenue
Columbus, OH 43201
W. Giles
General Electric Co.
Corporate R&D
P.O. Box 8
Schenectady, NY 12301
Gerald M. Goblirsch
Mechanical Engineering
Department of Energy
Grand Forks Energy Research Ctr
Box 8213 University Station
Grand Forks, ND 58202
John S. Gordon
TRW
Energy Systems Planning Div.
7600 Colshire Drive
McLean, VA 22101
Robert Gugger
Curtiss-Wright Corporation
One Passaic Street
Wood Ridge, NJ 07075
Nigel Guilford
Ontario Res. Foundation
Sheridan Park
Mississauga, Ontario L5K 1B3
Canada
Michael Gurevick
US Department of Energy
1900 S. Eads Street
Arlington, VA 22202
Karl-Axel Gustavsson
Bahco Ventilation
Bahco Systems, Inc.
P.O. Box 48116
Atlanta, GA 30362
Adi R. Guzdar
Foster-Miller Association
135 Second Avenue
Waltham, MA 02176
-H-
Dr. Simon L. Goren
National Science Foundation
Engineering Divsion
Washington, DC 20550
Eugene Grassel
Donaldson Company
P.O. Box 1299
Minneapolis, MN 55440
Michael W. Gregory
Exxon Res. & Dev.
1600 Linden Avenue
Linden, NJ 07036
Ulrich Grimm
MERC
US Department of Energy
P.O. Box 880
Morgantown, WV 26537
Stanley S. Grossel
Process Design Section
Hoffman-LaRoche, Inc.
Nutley, NJ 07110
Francine Hakimian
Librarian
Mcllvaine Co.
2970 Maria Avenue
Northbrook, IL 60062
H. J. Hall
H. J. Hall Associates
Cherry Valley Road
Princeton, NJ 08540
D. G. Ham
Battelle-Northwest
Battelle Blvd.
Richland, WA 99352
C. Frederick Hansen
Ames Research Center
NASA, Div. 229-3
Moffet Field, CA 94033
Mark S. Hanson
Battelle-Northwest
P.O. Box 999
Richland, WA 99352
-------
Michael J. Hargrove
C-E Power Systems
1000 Prospect Hill Road
Windsor, CT 06095
G. O. Haroldsen
Allied Chemical Corporation
550 Second Street
Idaho Falls, ID 83401
Andrew Harvey
Foster-Miller Associates
135 Second Avenue
Waltham, MA 02154
William J. Havener
Waltz Mill Site Box 158
Westinghouse Madison, PA 15663
J. H. Hedberg
Aeroject Energy Conversion Co.
P.O. Box 13222
Sacramento, CA 95813
K. H. Hemsath
Surface Division
Midland-Ross Corp.
P.O. Box 907
Toledo, OH 43691
Howard E. Hesketh
Southern Illinois University
School of Eng.
Carbondale, IL 62901
Graham Hilder
BP North America
620 - 5th Avenue
New York, NY 10020
R. C. Hoke
Exxon Research and Engineering
P.O. Box 8
Linden, NJ 07036
John D. Holmgren
Westinghouse
Waltz Mill Site Box 158
Madison, PA 15663
Dr. Donald J. Holve
Stanford University
Mechanical Engineering
Stanford, CA 94305
Robert Hoseman
PG&E
77 Beale Street
San Francisco, CA 94106
Dr. Chao-Ming Huang
TVA Energy Research
1320 Commerce Union Bank Bldg.
Chattanooga, TN 37401
Gordon Huddleston
Montana Energy and MHD
Research and Development lnst.,Inc.
P.O. Box 3809
Butte, MT 59701
Jacques Hull
Acurex/Ae roth e rm
485 Clyde Avenue
Mountain View, CA 94042
-J-
I. L. Jashnani
Arthur D. Little, Inc.
One Acorn Park
Cambridge, MA 02140
Albert A. Jonke
Argonne National Lab.
9700 S Cass Avenue
Argonne, IL 60439
-K-
Yale G. Kardish
Peabody Air Resources Equip. Co.
P.O. Box 5202
Princeton, NJ 08540
Harry F. Keller
Carrier Corporation/Research Div.
Carrier Parkway
Syracuse, NY 13201
William M. Kelly
Environmental Elements Corp.
P.O. Box 1318
Baltimore, MD 21203
Richard A. Kennedy
MITRE/Metrek Division
1820 Dolly Madison Blvd.
McLean, VA 22101
-------
Frank J. Kiernan
Aerojet Energy Conversion Co.
99 Clinch Avenue
Garden City, NY11530
Mike Klett
Process Engineering
Gilbert Associates
P.O. Box 1498
Reading, PA 19603
Richard N. Kniseley
Department of Energy
Ames Laboratory, DOE
Iowa State University
Ames, IA 50011
Dr. Charles E. Knox
Uniglass Industries
1440 Broadway
New York, NY 10018
Arthur L. Kohl
Rockwell International/
Atomics International Div.
8900 DeSoto Avenue
Canoga Park, CA 91304
John J. Kovach
MERC
Department of Energy
P.O. Box 880
Morgantown, WV 26505
Dr. A. Krishan Rao
Monsanto Enviro-Chem
P.O. Box 14547
St. Louis, MO 63178
William Krisko
Donaldson Co., Inc.
P.O. Box 1299
Minneapolis, MN 55440
Harry Krokta
The Ducon Company, Inc.
147 E. Second Street
Mineola, NY 11501
Andres Kullendorff
Atal-Laval Turbin AB
S-61220 Finspong, Sweden
T. Kumar
Occidental Research Corporation
1855 Carrion Road
LaVerne, CA 91750
William B. Kuykendal
IERL-RTP
Environmental Protection Agency
Research Triangle,Park, NC 27711
-L-
Norman R. LaMarche
General Electric
1 River Road
Bldg. 23, Room 355
Schenectady, NY 12345
George Lamb
Textile Research Institute
P.O. Box 625
Princeton, NJ 08540
William T. Langan
Buell Emission Control Div.
Envirotech Corporation
200 N. Seventh Street
Lebanon, PA 17042
Robert Langley
Inex Resources, Inc.
7475 W. Fifth Avenue
Lakewood, CO 80226
C. E. Lapple
Chemical Engineering Department
SRI International
333 Ravenswood Avenue
Menlo Park, CA 94025
Benjamin Linsky
A Different Air — Skyline
1360 Anderson
Morgantown, WV 26505
C. E. Lombard!
Teller Environmental Systems, Inc.
10 Faraday Street
Worcester, MA 01605
Bruno Loran
Ralph M. Parsons Co.
100 W Walnut Street
Pasadena, CA 91124
-------
E. T. Losin
Allis-Chalmers Corporation
Milwaukee, WI 53201
F. E. Lukens
Western Arcadia
4115 W. Ogden Avenue
Chicago, IL 60623
-M-
M. MacCafferty
IEA Coal Research
Technical Information Service
14/15 Lower Grosvenor place
London SWIW OEX, England
Andrej Macek
Department of Energy
Fossil Energy/Advanced Power Systems
20 Massachusetts Avenue, NW
Washington, DC 20545
Dr. Ray F. Maddalone
TRW Defense and Space Systems
Group
Bldg. 01/2020
One Space Park Drive
Redondo Beach, CA 90278
Capt. Joseph A. Martone
USAF, BSC
DET 1, HC ADTC
Tyndall AFB, FL 32403
Dr. Lidia Manson
TRW
Bldg. 01/2161
One Space Park Drive
Redondo Beach, CA 90278
William Masters
Acure x/Ae rothe rm
485 Clyde Avenue
Mountain View, CA 94042
Sigvard Mats
Flakt inc.
1500 E Putnam Avenue
Old Greenwich, CT 06870
I. Matsunga
Mitsubishi Heavy Industries
875 North Michigan Avenue
Suite 2100
Chicago, IL 60611
Mike May
Babcock and Wilcox Company
20 S VanBuren
Barberton, OH 44203
J. T. McCabe
Mechanical Technology
968 Albany-Shaker Road
Latham, NY 12110
J. D. McCain
Southern Research Institute
2000 9th Avenue
Birmingham, AL 35205
William McCarthy
Chemical Engineering
US/EPA — OEMI
Waterside Mall
Washington, DC 20460
Joseph E. McGreal
United States Steel Research
"B" Street
Penn Hills, PA 15235
James R. Melcher
MIT
Room 36-313, 50 Vassar Street
Boston, MA 02139
Dr. Arthur G. Metcalfe
Solar Turbines International
Mail Zone R-l, Box 80966
San Diego, CA 92138
Dr. Jim Meyer
Research Engineering
Oak Ridge National Lab.
P.O. Box "X"
Oak Ridge, TN 37830
Larry Michalec
Code 64270 NARF
NAS North Island
San Diego, CA 92135
M. Miller
Fluidyne Engineering Corporation
5240 Port Royal Road
Springfield, VA 22151
Ron Miller
PSM Sales — Rexnord
7675 Maple Avenue
Pennsauken, NJ 08109
-------
Richard Mitchell
Inex Resources, Inc.
7475 W. Fifth Avenue
Lakewood, CO 80226
Henry Modetz
EPA, Region IV
230 S. Dearborn
Chicago, II 60604
John G. Montagna
Argonne National Lab.
9700 S. Cass Avenue
Argonne, IL 60439
R. H. Moore
Battelle-Northwest
Battelle Boulevard
Richland, WA 99352
Samuel J. Moore
Carrier Corp/Research Div.
Carrier Parkway
Syracuse, NY 13201
William E. Moore
Fossil Energy/Adv. Power Systems
Department of Energy
20 Massachusetts Avenue
Washington, DC 20545
John A. Morris
Air Pollution Div.
Rexnord Inc.
P.O. Box 13226
Louisville, KY 40213
Tom Mosure
US/EPA
Environmental Research Inf. Ctr
26 W. St. Clair
Cincinnati, OH 45268
Andy Murphy
Acurex/Aerotherm
3301 Woman's Club Drive
Raleigh, NC 27611
Keshava S. Murthy
Battelie-Columbus Labs.
505 King Avenue
Columbus, OH 43201
-N-
Leonard M. Naphtali
Department of Energy
Fossil Energy/CCN
20 Massachusetts Avenue
Washington, DC 20545
James C. Napier
Solar Turbines International
P.O. Box 80966
2200 Pacific Hwy
San Diego, CA 92138
Charles K. Neulander
General Electric Company
P.O. Box 8, Bldg. K-l
Schnectady, NY 12301
G. S. Newton
Lovelace Biomedical
Environmental Research Institute
P.O. Box 5890
Albuquerque, NM 87115
-O-
Thomas O'Hare
Brookhaven National Laboratory
Upton, Long Island, NW 11973
Dr. Morris S. Ojalvo
National Science Foundation
Engineering Division
Washington, D.C. 20550
John M. Ondov
Lawrence Livermore Lab.
P.O. Box 808
Livermore, CA 94550
Dr. H. H. Osborn
C-E Air Preheater Co.
P.O. Box 372
Wellsville, NY 14895
-P-
James B. Paddan
Facet Filters
434 W. 12 Mile Road
Madison Heights, MI 48071
Dr. Richard D. Parker
Air Pollution Technology, Inc.
4901 Morena Blvd., Suite 402
San Diego, CA 92117
-------
Subhash S. Patel
Hittman Associates, Inc.
9151 Rumsey Road Bldg.
Columbia, MD 21045
Dr. Ronald G. Patterson
Air Pollution Technology, Inc.
4901 Morena Blvd., Suite 402
San Diego, CA 92117
Walter Podolski
Argonne National Lab.
9700 S. Cass Avenue
Argonne, IL 60439
Jospeh R. Polek
Catalytic, Inc.
1500 Market Street, CSW-10
Philadelphia, PA 19102
Duane H. Pondius
Southern Research Inst.
2000 9th Avenue S.
Birmingham, AL 35205
-Q-
Sandra Quinlivan
TRW
Bldg. R4/Room 1120
One Space Park Drive
Redondo Beach, CA 90278
-R-
E. Radhakrishnan
Battelle-Columbus Labs.
505 King Avenue
Columbus, Ohio 43201
Madhav B. Ranade
Research Triangle Inst.
P.O. Box 12194
Research Triangle Park
NC 27709
David L. Raring
Sonic Development Corporation
3 Industrial Avenue
Upper Sa-dle River, NJ 07458
Dr. Richard Razgaitis
Ohio State University
206 W. 18th Avenue
Columbus, Ohio 43221
D. L. Reid
Battelie-Northwest
P.O. Box 999
Richland, WA 99352
R. B. Reif
Battelle-Columbus Labs.
505 King Avenue
Columbus, OH 43201
Jean V. Remillieux
Air Industrie
19, Avenue DuBonnet
92401 Courbevoie
France
Ronald Renko
Development Engineering
C-E Air Preheater
Wellsville, NY 14895
George Rey
Office of R&D (RD-681)
EPA
Washington, DC 20460
George Rinard
Research Engineer
Denver Research Inst.
Denver, CO 80208
Frank G. Rinker
Midland-Ross
P.O. Box 907
Toledo, OH 43691
Thomas J. Robertazzi
Facet Enterprises
6521 Arlington Blvd.
Falls Church, VA 22042
George L. Roberts
Vice-President
Universal Transport Systems, Inc.
2665 Marine Way
Mountain View, CA 94040
David R. Rubin
USDA, Rural Electrification Admin.
Agriculture South Bldg.
14th & Independence Ave., SW
Washington, DC 20250
Stephen N. Rudnick
Harvard School of Public Health
665 Huntington Avenue
Boston, MA 02115
-------
-s-
Gene Schaltenbrand
C-E Preheater Company
P.O. Box 387
Wellsville, NY 14895
Robert K. Schaplowsky
Chemist
Aerojet
7767 LaRivera Drive #219
Sacramento, CA 95826
Robert Scheck
Stearns-Roger
P.O. Box 5888
Denver, CO 80217
G. F. Schiefelbein
Battelie-Northwest
Battelle Blvd.
Richland, WA 99352
Martin Schiller
CSI Engineering
P.O. Box 1515
Fairfield, CT 06430
J.W. Schindeler
John Zink Company
P.O. Box 7388
Tulda, OK 74105
Dr. Schnittgrunt
Consolidated Aluminum
2302 Weldon Way
St. Louis, MO 63178
H. F. Schulte
Rexnord, Inc.
1914 Albert Street
Racine, WI 53404
Richard A. Schwartz
Koch Engineering Co., Inc.
161 East 42nd Street
New York, NY 10017
Gernot Mayer Schwinning
Lurgi Apparate-Technik GmbH
Postfash 1, Gwinner Street
6000 Frankfort am Main-2
West Germany
Dr. David S. Scott
Dept. of Mech Engineering
Univ. of Toronto
Toronto, Ontario M5S lA4
Canada
Stanley J. Selle
Grand Forks Energy Res. Ctr.
Department of Energy
P.O. Box 8213, University Station
Grand Forks, ND 58202
Michael Shackleton
Acurex/Aerotherm
485 Clyde Avenue
Mountain View, CA 94042
J. K. Shah
Surface Div.
Midland-Ross Corporation
P.O. Box 907
Toledo, OH 43691
Jer-Yu Shang, PhD
MITRE Corp./Metrek Div.
Westgate Research Park
McLean, VA 22101
Jer-Yu Shang
Professional Engineering
4524 Andes Drive
Fairfax, VA 22030
David Shaw, Professor
State University of NY/Buffalo
4232 Ridge Lea Road
Buffalo, NY 14226
W. J. Sheeran
General Electric Company
Corporate R&D
P.O. Box 43
Schenectady, NY 12301
Dr. Thomas S. Shevlin
3M Company
3M Center, P.O. Box 33221, Bldg. 230
St. Paul, MN 55138
Kevin Shields
Hittinaw Associates, Inc.
9190 Red Branch Road
Columbia, MD 21045
-------
Dr. Gajendra H. Shroff
Bechtel Power Corporation
15740 Shady Grove Road
Gaithersburg, MD 20760
Jack Siegel
Department of Energy
Fossil Energy/Advanced Power
Systems
20 Massachusetts Avenue
Washington, DC 20545
Dr. A. P. Sikri
US Department of Energy
Gas & Shale Technology
20 Massachusetts Avenue
Washington, DC 20545
Dean Simeroth
California Air Resources Board
P.O. Box 2815
Sacramento, CA 95812
Theodore B. Simpson
Department of Energy
Fossil Energy/Advanced Power S
Systems
20 Massachusetts Avenue
Washington, DC 20545
A. V. Slack
SAS Corporation
Wilson Lake Shores
Sheffield, AL 35660
A. A. Smith
Babcock and Wilcox Ltd.
Woodall-Duckham House
Crawley, Sussex,
United Kingdom
Dr. Donald W. Smith
HAVEG Industries
900 Greenbank Road
Wilmington, DE 19808
Gregory W. Smith
Argonne National Lab.
9700 S. Cass Avenue
Argonne, IL 60439
Wallace B. Smith
Southern Research Inst.
2000 9th Avenue
Birmingham, AL 35205
John H. Smithson
Department of Energy
20 Massachusetts Avenue
Washington, DC 20545
Herbert W.Spencer, III
Western Precipitation Division
Joy Manufacturing
4565 Colorado Blvd.
Los Angeles, CA 90039
John Spriggs
Donaldson Company, Inc.
P.O. Box 1299
Minneapolis, MN 55440
Dr. C. J. Stairmand
Babcock and Wilcox Ltd.
Woodall-Duckham House
Crawley, Susses, United Kingdom
William A. Sandstrom
Institute of Gas Technology
3424 South State Street
Chicago, IL 60616
Walter Steen
Chemical Engineering
US EPA IER
IERL-RTP
Research Triangle Park
NC 27711
G. E. Stegen
Battelie-Northwest
Battelle Blvd.
Richland, WA 99352
David Stelman
Rockwell/Atomics Int. Div.
8900 DeSoto Avenue
Canoga Park, CA 91304
Ed Stenby
Project Engineering
Stearns-Roger
P.O. Box 5888
Denver, CO 80217
Richard C. Stone
Stone & Webster Engineering Co.
P.O. Box 2325
Boston, MA 02107
E. F.Sverdrup
Westinghouse
Beulah Road
Pittsburgh, PA 15235
-------
William M. Swift
Argonne National Laboratory
9700 S. Cass Avenue, Bldg. 205
Argonne, IL 60439
Ronald D. Snyder
Buell Emission Control Division
Envirotech Corporation
200 N. Seventh Street
Lebanon, PA 17042
-T-
S. I. Taub
IU Conversion Systems, Inc.
Joshua Rd & Stenton Ave.
P.O. Box 331
Plymouth Meeting, PA 19462
F. Gale Teats
Argonne National Lab.
9700 S. Cass Avenue
Argonne, IL 60439
Suresh P. Tendulkar
Westinghouse
Waltz Mill Site, Box 158
Madison, PA 15663
C. M. Thoennes
General Electric Company
Bldg. 2, Room 712
One River Road
Schnectady, NY 12345
Yung Kwang Ti
Research Associates
Univ. of Wisconsin at Milwaukee
3200 N Cramer Street
Milwaukee, WI 53201
C. Tien
Syracuse University
Syracuse, NY 13210
R. F. Toro
Recon Systems, Inc.
Cherry Valley Road
Princeton, NJ 08540
Richard H. To in
NY State Energy Research
and Development Authority
230 Park Avenue
New York, NY 10017
Dr. Jim Trolinger
Spectron Development Labs
3303 Harbor Blvd.
Costa Mesa, CA 92626
Norman R. Troxel
Research-Cottrell, Inc.
P.O. Box 750
Bound Brook, NJ 08805
Keh C. Tsao
Univ. of Wisconsin at Milwaukee
College of Engineering
3200 N. Cramer Street
Milwaukee, WI 53211
Alex Turchine
Proctor and Gamble
7162 Reading Road
Cincinnati, OH 45222
-U-
V. S. Underkoffler
Manager, Combustion & Advanced Power
Gilber Associates, Inc.
525 Lancaster Avenue
Reading, PA 19603
-V-
E. S. Van Valkenburg
Leeds and Northrup Company
Dickerson Road
North Wales, PA 19454
V. A. Varady
UOP Process Division
UOP Plaza
Des Plaines, IL
F. Munro Veazie
Owens-Corning Fiberglass Corp.
Technical Center
Granville, OH 43023
John A. Verrant
Donaldson Company, Inc.
P.O. Box 1299
Minneapolis, MN 55440
S. N. Vines
University of VA, Chemical
Eng. Dept.
Thornton Hall
Charlottesville, VA 22901
-------
G. J. Vogel
Argonne National Lab.
9700 S. Cass Avenue
Argonne, IL 60439
D. V. Vukovic
Faculty of Technology and
Metallurgy
Belgrade University
1100 Beogard, Karnegijeva 4,
P.O.B. 494
Yugoslavia
-W-
Gordon L. Wade
Combustion Power Company
1346 Willow Road
Menlo Park, CA 94025
Peter Waldstew
Physical Chemist
RDR
P.O. Box 33128
District Heights, MD 20028
Andrew Wallo
Env. Scientist
1605 Craig Street
Sterling, VA 22170
Walter Wolowodivk
Foster-Wheeler Dev. Corp.
12 Peach Tree Hill Road
Livingston, NJ 07039
Donald E. Wambsgans, II
District of Columbia
Dept. of Environmental Services
Room LL-3
614 "H" Street, NW
Washington, DC 20001
Stephen Wander
Dept. of Energy
400 First Street, NW
Washington, DC 20545
Frank Weiskopf
Environmental Elements Corp.
P.O. Box 1318
Baltimore, MD 21203
Phillip C. White
Dept. of Energy
Fossil Energy/Advanced Power
Systems
20 Massachusetts Avenue
Washington, DC 20545
Clyde L. Witham
SRI International
333 Ravenswood
Menlo Park, CA 94025
-Y-
Dr. H. C. Yeh
Lovelace Biomedical and Environmental
Research Institute
P.O. Box 5890
Albuquerque, NM 87115
-Z-
Dr. Karim Zahedi
President
EFB, Inc.
94 Francis Street
Brookline, MA 02146
John H. Zarnitz
NYC Dept. of Air Resources
7044 Manse Street
Forest Hills, NY 11375
Dr. F. A. Zenz
The Ducon Company
147 East Second Street
Mineola, LI, NY 11501
August H. Zoll
Curtiss-Wright Corporation
Power Systems
Wood Ridge, NJ 07075
Irena M. Zuk
Interdevelopment, Inc.
Rutherford B. Hayes Bldg.,Suite 104
2361 South Jefferson Davis Hwy
Arlington, VA 22202
Stephen H. Zukor
Department of Energy
OGST
20 Massachusetts Avenue
Washington, DC 20545
-------
TECHNICAL REPORT DATA
(Please read J/iitnictions on tlie reverse before completing)
I REPORT NO
EPA-600/9-7 8-004
3. RECIPIENT'S ACCESSION-NO.
4. TITLE A\D SUBTITLE
EPA/DOE Symposium on High Temperature/Pressure
Particulate Control
5. REPORT DATE
March 1978
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
Mike Shackle ton (Compiler)
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Acurex Corporation/Aerotherm Division
485 Clyde Avenue
Mountain View, California 94042
10. PRCiGRAM ELEMENT NO.
EHQg623
11. CONTRACT/GRANT NO.
68-02-2611, Task 2
12. SPONSORING AGENCY NAME AND ADDRESS
EPA. Office of Research and Development*
Industrial Environmental Research Laboratory
13. TYPE OF RE PORT AND PERIOD COVERED
Proceedings: 6-12/77
14. SPONSORING AGENCY CODE
Research Triangle Park. NC
27711
EPA/600/13
15.SUPPLEMENTARY NOTES (*) The U.S. Department of Energy (formerly ERDA)cosponsored
the symposium. IERL-RTP project officer is D.C.Drehmel, Mail Drop 61, 919/541-
2925.
s. ABSTRACT
proceecjjngS are a compilation of papers presented at the EPA/ERDA
(now DOE) Symposium on High Pressure/Temperature Particulate Control, in Wash-
ington, DC September 20-21, -1977. The symposium was sponsored jointly by EPA's
Industrial Environmental Research Laboratory (Research Triangle Park) and ERDA's
(now DOE's) Fossil Energy Division. Session I, on Fundamentals, included turbine
erosion/corrosion effects, materials, and the theory of HT/P particulate control.
Session n, on Filtration, Included granular bed filter test results, tests of ceramic
fiber filtration, tests of rigid ceramic filters , and a summary from the Federal
Republic of Germany. Session HI. on Other Collection Devices, included HT/P elec-
trostatic precipitation, sonic agglomeration, a cyclocentrifuge device, and scrubbing
with dry particles, molten glasses, and molten salts. Session IV, on Particle Sam-
pling and Measurement, included optical and extractive sampling for HT/P parti-
culate analysis.
17.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
Air Pollution, Dust Control, Turbines
High Temperature Tests , Erosion
High Pressure Tests, Corrosion
Sampling, Optics, Extraction
Measurement. Filtration, Ceramics
Electrostatic Precipitation, Acoustics
Agglomeration, Centrifuges , Fused Salts
b.IDENTIFIERS/OPEN ENDED TERMS
Air Pollution Control
Stationary Sources
Particulate
Sonic Agglomeration
Cyclocentrifuge
Molten Glass
c. COSATI Field/Group
14B, --
— \ 20F, 07A
--. 07D, 11B
13H, 20A
1. CJ.3TRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
20. SECURITY CLASS (Tins page/
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
------- |