EPA-600/2-77-207
October 1977
Environmental Protection Technology Series
HIGH TEMPERATURE PARTICIPATE
CONTROL WITH CERAMIC FILTERS
Industrial Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental Protec-
tion Agency, have been grouped into nine series. These nine broad categories were
established to facilitate further development and application of environmental tech-
nology. Elimination of traditional grouping was consciously planned to foster technology
transfer and a maximum interface in related fields. The nine series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
6. Scientific and Technical Assessment Reports (STAR)
7. Interagency Energy-Environment Research and Development
8. "Special" Reports
9. Miscellaneous Reports
This report has been assigned to the ENVIRONMENTAL PROTECTION TECHNOLOGY
series. This series describes research performed to develop and demonstrate instrumen-
tation, equipment, and methodology to repair or prevent environmental degradation from
point and non-point sources of pollution. This work provides the new or improved tech-
nology required for the control and treatment of pollution sources to meet environmental
quality standards.
REVIEW NOTICE
This report has been reviewed by the participating Federal Agencies, and approved for
publication. Approval does not signify that the contents necessarily reflect the views and
policies of the Government, nor does mention of trade names or commercial products
constitute endorsement or recommendation for use.
This document is available to the public through the National Technical Information
Service, Springfield, Virginia 22161.
-------
EPA-600/2-77-207
October 1977
HIGH TEMPERATURE
PARTICULATE CONTROL
WITH CERAMIC FILTERS
by
D.F. Ciliberti
Westinghouse Research Laboratory
Pittsburgh, Pennsylvania 15235
Contract No. 68-02-1887
ROAP No. 21AOL-029
Program Element No. 1AB012
EPA Project Officer: Dennis C. Drehmel
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, N.C. 27711
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, D.C. 20460
-------
DISCLAIMER
This report has been reviewed by the Office of Research and
Development, U. S. Environmental Protection Agency, and approved for
publication; Mention of trade names or commercial products does not
constitute endorement or recommendation for use.
-------
CONTENTS
Abstract ii
Figures iv
Tables vii
1. Introduction 1
2. Conclusions 3
3. Recommendations 5
4. Discussion 6
Phase I
4.1 Theoretical Assessment 7
4.2 Materials Fabrication and Testing 32
Phase II
4.3 Procurement of Materials 57
4.4 Screening Tests 63
4.5 Low Flow High Temperature Tests 91
4.6 High Flow High Temperature Tests 118
4.7 Conceptual Design and Cost Comparison 153
-------
ABSTRACT
This research project was intended to assess the possibility of using
ceramic materials as filters for fine particulate removal in high
temperature applications. The program has evolved into a two phase effort
with the first phase effort directed toward the development of a porous
alumina membrane filter. This effort was limited in success due to the
fragile nature of the membranes formed and because it proved to be
difficult to control the pore size distribution of the filters.
Phase II abandoned the concept of developing an entirely new filter
media and concentrated on screening available materials. The major
objective of this second phase was to identify materials with the good
filtration potential, select one or two of the most promising candidates,
and as rapidly as possible demonstrate them as hot gas fine particle
3
filters on a several hundred m /hr hot test.
The initial screening of materials revealed that the most promising
candidate was a thin walled ceramic, cross flow monolith, which was
originally produced as a catylist support for automotive exhaust systems.
Screening tests indicated that it was possible to achieve virtually 100%
removal of even submicron limestone test dust at face velocities and
pressure drops not dissimilar from those typical of fabric filtration.
Subsequent bench scale tests at temperatures around 1000°K confirmed the
ability of the material to perform well at elevated temperatures. The
final stages of experimentation were conducted in a larger facility where
3
flows of 4.8 m /min at 950°K were achieved. Although time limitations
did not allow optimization of the system, these larger scale tests
indicate that this configuration of ceramic filter offers great potential
as a hot gas filter.
ii
-------
An economic assessment of this filter system was carried out by
costing out a conceptual design for a commercial scale unit and comparing
costs with a comparable granular bed filter. This calculation indicated
that the ceramic filter system could have a capital cost approximately
one-third that of a granular bed system.
This report was submitted in partial fulfilment of contract
number 68-02-1887 to the Westinghouse Research Laboratories under the
sponsorship of the Environmental Protection Agency. The work covers
the period from July 1975 to August 1977.
iii
-------
LIST OF FIGURES
Number Pag£
1 Process Steps in Production of. Ceramic Filters 9
2 Ceramic Filter Production Facility 11
3 Filter Plant Layout 20
4 Westinghouse-ERDA Gasification Plant 27
t
5 Porous Aluminum Oxide Geometry. (Showing the oxide
thickness or channel length £, pore diameter d,
barrier layer thickness b, and cell size, c) 33
6 High Voltage Anodizing Cell 34
7 Edge View Showing Channels of Sample W-66 at 1250X
(1 cm=8 microns). This sample section is 65 microns
thick 50
8 Sample W-61A, Barrier Layer Side of Oxide Formed at
800 V. Cell size ^ 1.3 y; Pore Diameter 0.84 y (5200X) . 51
9 Structural Shapes for 3M ThermaComb 61
10 3M Crossflow Ceramic Monolith 62
11 Size Distribution of Test Dust 64
12 Element Low Temperature Holder 66
13 3M Element and Holder Mounted Inside Pipe 67
14 Face Velocity Vs. Pressure Drop for Thick Walled Filters . 68
15 AP Vs. U for Zircar 135C Material 70
16 Flow Vs. Pressure Drop for 3M ThermaComb 71
17 Pressure Drop Vs. Flow for Clean 15.25 cm ThermaComb Cube. 72
18 AP Vs. Time for Selas-XFF 73
19 AP Vs. Time for Selas-01 74
20 AP Vs. Time for Selas-10 75
21 AP Vs. Time for FMI Material 77
22 AP Vs. Time for Zircar Alumina Board (ZAL-15) 78
23 AP Vs. Flow for Zircar Alumina Board (ZAL-15) 79
24 AP Vs. Time for Zircar 135C Material 80
25 Pressure Drop Vs. Time for ThermaComb 83
iv
-------
List of Figures
Number Page
26 Pressure Drop Vs. Time for ThermaComb 84
27 Determination of Filter Cake Constant 86
28 Before and After 103 kPa Pulse. 87
29 Before and After 17 m3/hr Flush for 2 Sec 88
30 Pulsed Purge=137.9 kPa 3.0 Seconds Duration 90
31 Hot ThermaComb Test Jig 92
32 Flow Schematic and Cleaning Sequence for Low Flow-High
Temperature System 93
33 High Temperature Ceramic Test Facility 96
34 High Temperature Ceramic Filter Assembly 97
35 Pressure Drop Vs. Time at (1090°K) 99
36 Hot ThermaComb Test 101
37 Ambient Temperature Test (ThermaComb) 102
38 Hot ThermaComb Test 104
39 Hot ThermaComb Test 105
40 ThermaComb Filter Test-Constant Flow 107
41 ThermaComb Filter Test-Constant Flow 108
42 ThermaComb Filter Test-Constant Flow 109
43 Hot ThermaComb Test (High Dust Loading) Ill
44 Hot ThermaComb Test Variation of Filter Pulse Purge ..... 112
45 Hot ThermaComb Test Variation of Pulsed Purge ....... 113
46 Hot ThermaComb Test Variation of Pulsed Purge 114
47 ThermaComb Filter Test-Pulse Times 115
48 Hot-Low Flow Test of W. R. Grace & Co. Ceramic Filter . . . 116
49 High Flow Test Facility 119
50 6 Inch ThermaComb Test Assembly 121
51 Schematic of Pulsing Modification 122
52 High Temperature High Flow Test Facility 124
53 High Temperature Ceramic Test Facility 125
54 Detail of Ceramic Filter Assembly 126
55 High Flow Test Starting with Clean Filters 129
-------
List of Figures
Number Page
56 Pulse Intensity Tests 130
57 Pulse Intensity Tests 131
58 Filters and Vessel Precleaned 133
59 Initial Tests with ThermaComb Cut in Half 135
60 Manual Dirty Side Pulse Test • • • • 136
61 Hot Test with Initial Dirty Side Pulse 138
62 Hot Test with Initial Dirty Side Pulsed Continued 139
63 Hot High Flow Test After Individual Dirty and Clean Pulses. 140
64 Cold Test with Modified Pulse 141
65 Hot Test with Modified Pulse 142
66 Hot ThermaComb Test with Cleaning Pulse to Clean and
Dirty Sides 143
67 Hot ThermaConb Test with Cleaning Pulse to Clean and
Dirty Sides 145
68 Hot ThermaComb Test with Combined Pulse 146
69 Hot ThermaComb Test with Pulse Variations 147
70 Hot ThermaComb Test with Pulse Variation 148
71 Final Hot Test with ThermaComb 150
72 Initial Hot Test with 15cm Cube of W. R. Grace Material . . 151
73 Hot Test with 15cm Cube of W. R. Grace Material 152
74 Sketch of Possible Arrangement of Commercial Use of
Ceramic Filters 155
VI
-------
LIST OF TABLES
Number Page
1 Capital Costs for Ceramic Filter Production Facility .... 12
2 Materials and Power Per M2 of Oxide Filter 13
3, Operating Costs in Thousands of Dollars for Producing
2000 MZ/Year of Cylindrical Oxide Membranes 2.5 cm x
100 cm 15
4 Laboi Force for Ceramic Filter Production 16
5 Production Costs - Ceramic Membrane Filters 17
6 Filter Element Costs 19
7 Cost Data for Ceramic Membrane Filter Plant 22
8 Capital Charges 23
9 Costs for Granular Bed Filter - Bag Filter Plant 25
10 Coal Gasification-Combined Cycle Plant Costs $/kW,
250 MW Plant 28
11 Capital Costs for 635 MW Coal Fired Steam Plant 30
12 Annual Costs for Power Plants Per kW of Capacity 31
13 Average Cell Size and (pore diameter) in A/V for Various
Fabrication Conditions 36
14 Pressure Pore Enlargement 39
15 Summary of Pressurized Etched Oxide Samples 41
16 Pressure Etched Samples 45
17 Pressure Etched Samples, Barrier Layer Up 47
18 Mechanical Testing Results 53
19 Survey Results 58
20 Ceramic Filter and Granular Bed Filter Costs 160
vii
-------
SECTION 1
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.
Another promising coal conversion technology that is currently
under investigation is that of pressurized fluidized bed combustion.
This process has particle removal requirements that are essentially
the same as those required by the gasification scheme with the exception
that the gas to be cleaned is a flue gas rather than a reducing fuel
gas. This may allow a wider choice of materials of construction since
it is anticipated that flue gas will be less corrosive to standard
materials.
Current experimentation with inertial collectors indicates that
they will not give adequate cleaning to meet either environmental or
turbine requirements on a mass loading or size distribution basis.
Consequently it appears that some novel, final stage of positive
filtration will be required.
-------
Two methods of filtering hot corrosive gases that have received
attention recently are granular bed filtration and ceramic filters.
There are several schemes for granular bed filtration which exist in
varying degrees of commercialization. The work carried out here was
intended to advance ceramic filtration technology to the stage where
some reasonable comparisons of performance and economics could be made
between the systems.
This study has identified some promising ceramic filter media, and
both the economic and performance comparisons with granular bed
filtration look promising.
-------
SECTION 2
CONCLUSIONS
This experimental work has been carried out in two phases, the first
investigating the development and use of membrane type ceramic filters
and the second phase focusing on screening and testing available porous
ceramic materials as hot gas filters.
Some conclusions that can be drawn from the Phase I effort are
listed below:
• The economics of ceramic membration filtration appear to
be favorable when compared with other hot gas filtration
concepts.
• The delicate nature of the ceramic membrane filters makes
commercial use of them unlikely in large electric power
generating applications.
• No suitable means for supporting the membranes was
discovered although this does not seem to be an
insoluble problem.
• The most troublesome technical problem was the lack of
control of pore size and size distribution resulting
from the electrolytic pore forming process.
Phase II examined several commercially available types of ceramic
materials which can be broadly grouped into two categories: porous thick
walled filters and thin walled monolithic honeycomb structures.
Some conclusions drawn from this work apply to both categories of
materials:
• Filters of this nature are suitable for operation at
temperatures exceeding any current coal conversion
process requirements.
• The filters exhibit good resistance to corrosion with
the possible exception of alkali metal attack.
-------
• The materials withstand thermal cycling well.
• The filters generally exhibit virtually 100% effective
particle removal of a submicron test dust.
• Cleaning methods can be devised to allow continuous
operation.
Conclusions with regard to the thick walled filters tested are as
follows:
• These filters generally had a relatively high resistance
to flow.
• They were relatively difficult to clean, although most
of them could be cleaned by vigorous back washing.
• The physical embodiment of these filters generally did
not lend itself to easy, efficient incorporation into
a viable commercial unit.
Conclusions reached concerning the thin walled monolithic structures are
as follows:
• Effective cleaning for continuous can be achieved
relatively easily with back washing pulses.
• The filters have relatively low pressure drops at
moderate face velocities.
• The filters have very high surface area to volume ratios
which makes efficient use of pressure vessel containments.
• The preliminary economic analysis of a commercial module
of using this type of ceramic filter indicates capital
costs could be as low as one-third those of a similar
granular bed system.
« A system based on the use of this type of filter
material appears to be as viable as a hot gas cleaning
system.
-------
SECTION 3
RECOMMENDATIONS
The work conducted under Phase I of this effort was successful to
only a limited extent. Since the probability of developing a commercially
viable system based on the electrolytic oxidation of alumina membrane
filters is remote, it is recommended that no further work be carried out
in this direction.
The effort in Phase II of this project has in large part been
successful in that it has been possible to demonstrate continuous high
temperature filtration of fine particulates on a fairly large scale.
These encouraging results have been obtained using filter materials that
have not been designed specifically for use as filters, and in spite of
the fact that the high flow facility was not originally designed to
accommodate these materials. The following suggestions can therefore be
strongly recommended.
• An effort should be made to explore filter fabrication
options in an attempt to optimize parameters such as
porosity, pore size, cell size, wall thickness, etc.
F
• This effort should be conducted in parallel with a bench
scale hot filtration test facility so that the effects
on' filter performance can be quickly measured and
correlated.
• A third effort should focus on the design problems
associated with the incorporation of this type of filter
into a viable piece of equipment. This effort should
have as its goal the design and fabrication of a pilot
unit capable of operation at 1000°C and at 10 atm with
flows in the range of 750-1000 m3/hr.
-------
SECTION 4
DISCUSSION
SUMMARY - PHASE I
In the initial phase of this program we have attempted to develop
and test ceramic filter elements suitable for high temperature high
pressure use. The work has concentrated specifically on alumina
membranes fabricated by Horizons, Inc. using their electrolytic
oxidation technology.
Success has been limited. Two problems have proven unsurmountable
within the time frame available for this development effort. The
fragile nature of the membranes has made handling and testing of the
material difficult. No commercially viable support techniques could
be devised. Additionally, wide variations in pore size from specimen
to specimen were not brought under control. Consequently it was not
possible to reliably reproduce filter elements with the required
pressure drop/flow rate/filtration characteristics.
-------
4.1 THEORETICAL ASSESSMENT
Objective. The objective was to establish the technical and
economic feasibility of using a porous ceramic membrane for fine
particle control in advanced power systems.
Collection Efficiency. Techniques for predicting collection
efficiency of ceramic filters by impaction and interception mechanisms
have been considered. In general, these techniques may only be applied
to simplified models of the flow through a clean filter, and do not
account for the effect of deposited particulate. An examination of the
literature failed to furnish any satisfactory technique for predicting
the efficiency of filtration on an established filter cake.
Experimental information on devices with similar operating
characteristics to the proposed ceramic membrane showed essentially
complete particle retention for particle sizes down to 0.1 ym diameter.
A simplified method of calculating Brownian deposition was
used to predict the extent to which particles entering the filter pores
would be collected on the pore walls. This showed deposition to be
complete when
where
D is Brownian diffusivity
t is residence time of flow through the pore
d is pore diameter
dp is particle diameter
For a ceramic filter element which would have 1 ym pores and
a thickness of 250 ym, complete collection is predicted at all gas
velocities up to 30 cm/sec.
-------
Based on these considerations, a porous ceramic membrane filter
should be capable of achieving 90% collection of the submicron particles
produced in advanced power systems.
Assessment of Filter Costs. We have assessed filter fabrica-
tion costs, and filter operating costs as a function of filter character-
istics such as thickness, pore size, free cross-section and operating
pressure loss. The results indicate that the porous ceramic filter can
be competitive with alternate fine particle control systems on a cost-
effectiveness basis.
Production of Filter Elements. Filter element production
requires four major processing steps.
• Electrolysis of aluminum to form porous alumina
• Stripping of the formed alumina from the base metal
• Chemical processing to remove barrier layers and to
etch pores
0 Firing to convert the alumina to the stable a form
In addition several service functions are required. A flow
sheet of the steps involved in the commercial production of ceramic
membrane filters is shown in Figure 1.
Aluminum stock is mechanically formed as required (cut to size,
swaged, etc.) and assembled into racks for electrolysis. The racks pass
to the electrolysis cells where the oxide is formed. From the cells,
the assemblies are sent, after washing, to an aluminum stripping bath.
Here the elements are immersed in a bromine-methanol solution which strips
the oxide from the metal.
Individual elements are removed, washed and immersed in an
etch solution which removes barrier layers and enlarges pore diameters.
The elements are then dried, fired to convert the alumina to the a form,
inspected and packed for dispatch.
-------
Aluminum1
Stock
Wash
Aluminum
Stripping
Dwg. 6365A43
Cut and Formed —
As Necessary
Heat
Assembled Into
Electrolysis Racks
Electrolysis
(0°C)
Power
Wash
Cap Removal Etch
Fire
Dry
Wash
Quality
Control
Pack and Dispatch
Fig. 1- Process steps in production of ceramic filters
-------
Capital Equipment. An estimate of capital equipment require-
ments and capital equipment costs has been made for a production facility
2
with the capacity to produce 1 M /hr of filter surface. This plant would
2
have a nominal capacity of 2000 M per year when operating on a one shift
2
basis, or 8000 M per year when operating on a continuous (4 shift) basis.
For the purposes of this exercise, a base case has been assumed
for filter characteristics so that equipment sizes can be estimated. The
assumptions are that the filter elements have a thickness of 250 urn, and
a pore size of 1 pm. Estimates for filters with differing characteristics
have been obtained by applying factors to the base case.
A plant layout showing the major items of capital equipment is
shown in Figure 2; while Table 1 lists the costs of the individual
items and of the completed facility.
Operating Costs, Annual Costs
Raw Materials. Raw materials costs for the fabrication of
ceramic membrane filter elements have been estimated by Horizons, Inc.
Their estimates indicate a materials cost of approximately $17.50 per M
(Tables 2 and 3).
Power. Power for electrolysis and auxiliary use has been
costed at the rate of 3c per kW hr.
Labor. Labor estimates provided by Horizons have been used
as a basis for establishing a minimum labor force for a production
facility (Table 4).
Capital Charges. Capital costs are estimated at 25% P.A.
Combining these cost estimates for the two production rates
previously cited (2000 M P.A. and 8000 M P.A.) establishes a cost range
o o
for ceramic membrane filters of $150 per M to $250 per M (Table 5).
The major costs are power, labor and capital charges.
10
-------
Dwg. 6365A42
Metal Store
T
D. C.
Power
Supply
Refrigeration
Metal Forming
MH^V ^^M ^^ tm^m ^^K ^MB ^••B ^^ ^B ^
Storage (Chemicals
Spares, Maintenance
Materials)
Storage and
Dispatch
-
"
Packing
Quality Control
Assembly
Area
Electrolysis
\
T
Wash
Strip
Wash
Kiln
Drier
Wash
Etch
Maintenance Area
Wash/
Lxker
Room
Office
Office
Office
Office
c ,
Scale
Fig. 2-Ceramic filter production facility
-------
Table 1
Capital Costs for Ceramic Filter Production Facility
Base Case - 1 M2/Hr., 250 ym Thick, 1 ym Pores
Equipment Installed Cost
Metal Forming Equipment (Budget Estimate) $ 20,000
Power Supply (G.E. Rectiformer, 2000 kW) 85,000
Drier 10,000
Kiln 25,000
Tanks, Racks, Mechanical Transfer 50,000
Building 540 M2 @ $200/M2 108,000
SUBTOTAL $298,000
Engineering 40,000
Contingency 40,000
Interest During Construction 25,000
TOTAL $403,000
12
-------
Table 2
Materials and Power Per M of Oxide Filter
Materials
(1) + (2) Aluminum 1199 formed to proper shape and anodizing chemicals
Membrane Thickness
(ym) (mils)
125 5
250 10
Aluminum Thickness
(ym) (mils)
100 4
200 8
Aluminum Weight
Kg
0.3
0.6
Min. Cost*
(S)
1.2
2.4
Anodizing
Chemicals
($)
2.70
5.40
(3) Chemicals for removal of unused aluminum
Bromine 0.25 Kg O^l.lO/Kg) = $0.275
Methanol 8.0 litra (^$0.10/litre) = $0.80
(4) Chemicals for removal of barrier layer and pore enlargment $2.50
(5) Materials for supports - estimate at < $10.00
Power for Anodizing for Pore Diameters to 0.8 urn
Membrane Average Total Approximate Power Costs
Thickness Amperes Approximate Ampere-Hours for Anodizing and Cooling
(ym) at 600V D.C. Hours at 600V D.C. ($0.03/KW-hr)
125 31 20 620 16.60
*1199 Al formed to cylindrical units 2.5 cm x 100 cm will probably cost at least $4/Kg.
-------
Table 2 (continued)
Membrane Average Total Approximate Approximate Power Costs
Thickness Amperes Approximate Ampere-hours Joule Heating (KW) for Anodizing and Cooling
(ym) at 600V B.C. Hours at 600V D.C. Requiring Cooling ($0.03/KW-hr)
250 31 40 1240 11 35.50
Power for Converting to Alpha Form
High temperature furnace capable of holding 1200°C for 15 minutes.
This may not be needed if conversion is effected in the hot gas stream.
-------
Table 3
Operating Costs in Thousands of Dollars for Producing
2000 M2/Year of Cylindrical Oxide Membranes 2.5 cm x 100 cm
Membrane Thickness (ym) Materials Power Direct Labor* Total
125 £ 0.8 ym diameter
250 < 0.8 ym diameter
15.0 (+<20)
22.8 (+<20)
33.2
71
12.0 (+<20)
16.0 (+<20)
60.2 (+<40)
93.8 (+<40)
*Labor
Direct labor for producing at rate of 100 2.5 cm x 100 cm cylindrical elements
250 ym thick per day.
2 Men at $4/hour = $64/day/8 M2 = $8/M2.
Estimate 125 ym thick at $6/M2 and 500 ym thick at $10/M2.
2
Estimate labor for supports at <$10/M .
-------
Table 4
Labor Force for Ceramic Filter Production
Function
Engineering
Supervision
Process Work
Maintenance
Metal Forming
Quality Control
Packing & Storage
TOTAL ($)
One Shift
Number
1
1
3
1
1
1
1
Operation
Cost($)
20,000
15,000
27,000
11,000
9,000
9,000
9,000
100,000
Four Shift
Number
1
4
9
1
2
2
2
Operation
Cost($)
20,000
65,000
90,000
11,000
18,000
18,000
18,000
240,000
With Overheads @ 100%
200,000
480,000
16
-------
Table 5
Production Costs - Ceramic Membrane Filters
Production Production
+ 2000 M2/Year + 8000 M2/Year
Raw Materials 35,000 140,000
Power
Electrolysis 120,000 480,000
Driers, Kilns, Etc. 1,000 5,000
Labor 200,000 480,000
Capital Charges @
25% P.A. 100,000 100.000
TOTAL ANNUAL COST 456,000' 1,205,000
Cost per M2 228" ' 151
17
-------
Two areas of uncertainty exist - labor costs and capital costs.
The labor force estimates could be significantly lower than actual require-
ments in an operating plant, while capital cost estimates are at best
speculative. If labor costs were double our estimate, filter costs would
2
increase by $60 to $100 per M . In addition, if capital costs were
twice our estimates, filter costs would increase by $15 to $60 per M .
If these escalators are added to the original estimates to give
a worst case, the estimated range for the cost of ceramic filters
becomes $230 to $400 per M .
Projections of filter costs for membranes 125 pm thick and
500 ym thick are listed in Table 4.6. The costs range from $112 per M
2 2
for 125 urn material produced at a rate of 8000 M P.A. to $350 per M
2
for 500 ym material produced at a rate of 2000 M P.A. If capital and
labor charges are doubled to give worst case estimates, these limits
are adjusted to $182/M and $536/M respectively.
Filter Plant. Ceramic membrane filters will efficiently
remove submicron particles from gas streams. However, due to the small
pore diameter, flow rates must be relatively low and pressure drops will
be relatively high. Consequently they cannot be economically employed
in conventional gas cleaning applications.
In high pressure processes (such as coal gasification) where
gas volumes are kept low by pressurized operation and pressure drops up
to 100 kPa may be tolerated, the ceramic membrane has possible
applications. The resistance of the filters to high temperature chemical
attack make them well suited for use under extreme conditions.
A typical plant would consist of several pressure vessels con-
taining manifolded ceramic membrane filter candles. The vessels would
be arranged so that they could be sequentially removed from the
filtration cycle and cleaned by a reverse flow of clean gas. A schematic
layout of a complete plant consisting of four vessels is shown in
Figure 3. Three vessels are shown filtering the raw gas while the
fourth vessel (#4) is cleaned by a reverse gas flow.
18
-------
Table 6
Filter Element Costs
Thickness 125 ym 500 yim
Production Rate 2000 M2 PA 8000 M2 PA 2000 M2 PA 8000 M2 PA
Basic Estimate $/M2 184 112 349 232
Escalated Estimate 321 182 536 314
(Capital & Labor Charges
Increased by 2x)
19
-------
Dwg. 1678B36
Gas Outlet
N>
O
Purge Gas
Outlet
Filter Element
Baffle
AAanway
•Gas Inlet
•Manway
Walkway
Dust Lock
Filter Vessel (Conceptual)
Purge
Gas-*
Return
to Plant
Raw Gas In
Plant Layout Plan (Conceptual)
C - Closed 0 - Open
Vessel #4 in Cleanup Mode
Fig.3 -Filter plant layout
-------
A filter plant of the size shown would accommodate approxi-
2
mately 200 M of filter surface, assuming filter candles 1 M long by
2.5 cm diameter.
Filter Plant Costs. Capital costs for a filter plant can be
assessed once the required filter area is established. This will
determine the cost of the filter elements required, and also the size
and cost of the containing vessels. To establish operating costs it is
additionally necessary to know the operating pressure loss through the
filters.
For the purpose of this evaluation we have estimated costs for
a filter plant which treats approximately 30 M /S of gas at 870°C and
1620 kPa. This is equivalent to the output of a gasifier which supplies
fuel to a 250 MW combined cycle plant.
The estimates have been made for various filter characteristics
pore diameter, thickness and free cross-section. In addition, gas
velocity/pressure drop effects have been estimated. Typical results are
shown in Table 7.
Referring to the table, filter costs are estimated at two
levels; the normal level (N) based on the filter production cost
estimates and the maximum level (M) based on the "worst case" estimates.
Vessel costs were estimated for a 200 JT filter plant based on
a four vessel arrangement, and costs for other installations were
factored by a 0.7 power law.
Annual costs for owning the filter plant are also tabulated
at the normal and maximum level. For this estimate, the plant life has
been assumed to be 20 years, while the filters are assumed to have a
5 year life. The breakdown of capital charges is given in Table 8.
21
-------
Table 7
Cost Data For Ceramic Membrane Filter Plant
Pore
Diameter
um
0.5
1.0
Memo rane
Thickness
ym
125
250
500
125
250
500
Pore Gas
Velocity
cm/s
3
6
9
3
6
3
9
18
36
9
18
9
Free
Cross
Section
%
12
25
50
12
25
50
12
25
50
12
25
50
12
25
50
12
25
50
12
25
50
12
25
50
12
25
50
12
25
50
12
25
50
12
25
50
Superficial
Velocity
cm/s
0.375
0.75
1.5
0.75
1.5
3.0
1.12
2.25
4.5
0.375
0.75
1.5
0.75
1.5
3.0
0.375
0.75
1.5
1.12
2.25
4.5
2.25
4.5
9.0
4.5
9
18
1.12
2.25
4.5
2.25
4.5
9
1.12
2.25
4.5
ip
KPa
20
40
60
40
80
80
15
30
60
30
60
60
Filter
Area
M2
9866
4933
2467
4933
2467
1233
3304
1644
822
9866
4933
2467
4933
2467
1233
9466
4933
2467
3303
1644
822
1644
822
411
822
411
206
3303
644
822
1644
822
411
3303
1644
822
Filter
Cost
$ x 106
N
1.697
0.849
0.424
0.849
0.424
0.212
0.568
0.283
0.141
2.170
1.085
0.542
1.085
0.542
0.271
3.068
1.534
0.767
0.607
0.304
0.152
0.304
0.152
0.076
0.152
0.076
0.038
0.789
0.395
0.197
0.395
0.197
0.099
1.552
0.576
0.288
Filter
Cost
$ x 106
M
3.058
1.529
0.765
1.529
0.765
0.382
1.024
0.510
0.255
3.749
1.875
0.937
1.875
0.937
0.469
4.913
2.456
1.228
1.060
3.530
D.265
0.530
0.265
0.132
0.265
0.132
0.066
1.321
0.660
0.330
0.660
0.330
0.165
1.770
0.885
0.443
Vessel
Cost
(Installed)
$ x 10&
9.168
5.643
3.474
5.643
3.474
2.139
4.261
2.615
1.610
9.168
5.643
3.474
5.643
3.474
2.139
9.168
5.643
3.474
4.262
2.615
1.610
2.615
1.610
0.992
1.616
0.992
0.611
4.262
2.615
1.610
2.615
1.610
0.992
4.262
2.615
1.610
Filter
Cost
Installed
$ x 106
(N)
3.394
1.698
0.848
1.698
0.848
0.424
1.136
0.566
0.282
4.340
2.170
1.084
2.170
1.084
0.542
6.136
3.068
1.534
1.214
0.608
0.304
0.608
0.304
0.152
0.304
0.152
0.076
1.578
0.790
0.394
0.790
0.394
0.198
2.304
1.152
0.576
Filter
Cost
Installed
$ x 10&
(M)
6.116
3.058
1.530
3.058
1.530
0.764
2.048
1.020
0.510
7.498
3.750
1.874
3.750
1.874
0.938
9.826
4.912
2.456
2.120
1.060
0.530
1.060
0.530
0.264
0.530
0.264
0.132
2.642
1.320
0.660
1.320
0.660
0.330
3.540
1.770
0.886
Total
Filter
Plant
Cost
$ x 106
(N)
12.562
7.341
4.322
7.341
4.322
2.563
5.397
3.181
1.892
13.508
7.813
4.558
7.813
4.558
2.681
15 . 304 *
8.711
5.008
5.478
3.223
1.914
3.223
1.914
1.144
1.914
1.144
0.687
5.840
3.405
2.004
3.405
2.004
1.190
6.566
3.767
2.186
Total
Filter
Plant
Cost
$ x 106
(M)
15.284
8.701
7.173
8.701
5.004
2.903
6.308
3.635
2.120
16.666
9.393
5.348
9.393
5.348
3.077
18.994
10.555
5.930
6.382
3.675
2.140
3.675
2.140
1.256
2.140
1.256
0.743
6.904
3.935
2.270
3.935
2.270
1.322
7.802
4.385
2.496
Annual
Cost
$ x 106
(N)
3.097
1.766
1.017
.1.766
1.017
0.542
1.282
0.740
0.432
3.434
1.935
1.101
1.935
1.101
0.634
4.073
2.254
1.261
1.310
0.755
D.440
0.755
0.440
0.258
0.440
0.258
0.153
1.439
0.820
0.472
0.820
0.472
0.274
1.698
0.949
0.537
Annual
Cost
$ x 106
(M)
4.066
2.251
1.260
2.251
1.260
0.713
1.607
0.902
0.513
4.558
2.497
1.382
2.477
1.382
0.775
5.387
2.911
1.589
1.632
0.916
0.521
0.916
0.521
0.298
0.521
0.298
0.173
1.819
1.009
0.567
1.009
0.567
0.321
2.137
1.169
0.647
(0
to
-------
Table 8
Capital Charges
Taxes and Insurance 2.0%
Interest on 65% Borrowed Capital @ 6-3/4% 4.4%
Profit on 35% Equity @ 7-1/2% 2.6%
Federal Income Tax 2.6%
Maintenance 4.0%
Depreciation
Plant - 20 Year Linear 5.0%
Filters - 5 Year Linear 20.0%
23
-------
Alternative Gas Cleaning Plant
Granular Bed Filter Plant. An alternative gas cleaning system
for fine particulate control in a coal gasification combined cycle system
would, based on available data, require two particle collection units to
achieve the same efficiency as the ceramic membrane plant. One unit, a
granular bed filter, would be installed in the high pressure gas line for
turbine protection, while a second unit would be required to remove
submicron particles from the exhaust gases.
To establish the comparative costs of this system, capital and
operating cost estimates have been made for comparison with those of a
ceramic membrane plant.
Granular bed filter costs have been factored from estimates
made by the Ducon Go. On this basis the installed cost for a 250 MW
filter unit is $16.20 per kW. Budget estimates by baghouse manufacturers
suggest that an installed baghouse for the control of fine particle
emissions will cost $16.80 per kW. Consequently a particle control
system based on conventional collection equipment will have an installed
capita
plant.
capital cost of approximately $33 per kW, or $8.25 x 10 for a 250 MW
Annual costs for owning and operating this plant are $1.786 x 10
(Table 9).
Consider now an equivalent ceramic membrane plant which operates
with a maximum pressure loss of 100 kPa. The operating cost for this
unit would be $428,000 based on a power cost of 2
-------
Table 9
Costs for Granular Bed Filter - Bag Filter Plant
Installed Capital Cost - $8.25 x 10
Annual Costs
Fixed Costs $957,000
Depreciation (20 Year Linear) 412,500
Maintenance 330,000
Power (@ 2<: per kWhr) 86,000
TOTAL $1,785,000
25
-------
Conventional Power Plant. A second alternative to the ceramic
membrane-combined cycle power plant would be a conventional coal-fired
power plant.
Such a plant would have to incorporate sulfur and particulate
control processes. For the purposes of this evaluation, it will be
assumed that limestone-wet scrubbing is sufficient for adequate emission
control.
To make a comparative evaluation it is necessary to establish
complete costs for both the coal gasification-combined cycle plant, and
for the conventional steam plant. The coal gasification-combined cycle
system is evaluated on the basis of a 250 MW Westinghouse-ERDA coal
gasification unit (Fig. 4) while for the conventional plant, a 635 MW
unit is considered.
In evaluating the comparative costs for owning and operating
these systems, it is necessary to assume costs for capital, fuel, power,
and sulfur sorbents. For this purpose capital charges have been set at
20.6% (Table 9); power is charged at 2
-------
RAI..CAR DJMPER
S-SMED M-6
fPAYLOAOERS M-3f M-9 TO RECOVER"!
^SOLIDS £ DELIVER TO HOPPERS J
T-3 A|B "REDRIED
COAL SURGE- M-S PREDRIED
9FW_fROM
TO IPACEI
PLANT
MR FROM
FEED ELEVATOR
V-n AJB
DOLOM.TE
FEED HOPPE
INACTIVE COAL
SUPPLY 3QOOOTON(2
PROMISED
JOO PSiG STEAM
FROMEAiEJ
AIR FROM C-4B
Fig. 4® -ERDA Gasification Plant
P-4A{B ASH
SLURRY PUMPS
WESTINGHOUSE COAL GAS FICAT'ON 'FOCES3
FOR £5O MW POWER =LA.NT
V*C.* U-1-7Z
-------
Table 10
Coal Gasification - Combined Cycle Plant Costs
$/kW, 250 MW Plant
Coal Gasification Plant and Auxiliaries $135
Combined Cycle Plant 180
Land 1
Electric Plant 15
Cyclone Dust Collectors 12
Misc. Plant Equipment 5
Interest During Construction 50
TOTAL ($/kW) $398
28
-------
A cost account for a conventional 635 MW power plant is listed
in Table 11. This is based on a published 1971 cost account and has
been factored by the M&S index to update the costs to late 1974. An
additional $50 per kW is added to cover the capital costs associated
with a limestone-wet scrubbing system for particle and sulfur removal.
The final cost is $475 per kW. The heat rate for the plant is 9122
Btu/kWhr (based on Hammond #4 station of Georgia Power and Light).
Annual Costs. Annual costs for owning and operating these
two plants have been assessed per kW of installed capacity to give
numbers which may be readily compared. The results (Table 12) show
the combined cycle plant to be approximately $15 per kW per year cheaper,
before fine particle control equipment is accounted for.
If $10 per kW per year is committed to fine particle collection,
the combined cycle plant maintains its competitive position. This allows
a tota.
plant.
a total annual expenditure of $2.5 x 10 on a ceramic membrane filter
Reference to Table 7 indicates that a ceramic membrane filter
plant can be operated for less than this $2.5 x 10 annual total.
Discussion. This preliminary estimate of ceramic filter costs
and of filter plant costs indicates that such units can be economically
competitive with alternative systems for fine particle control in power
plant applications. Indications are that the required filter properties
are well within the scope of available and developing technology.
The review of filter costs indicates some minimum requirements
for filter properties, based purely on economic considerations. With
additional information on allowable pressure drop-membrane thickness
limits (i.e., mechanical properties) the field of acceptable filter
characteristics may be further defined. This will allow clear objectives
for filter development to be established.
29
-------
Table 11
Capital Costs for 635 MW Coal Fired Steam Plant
$/kW
Land and Land Rights 1.51
Structures and Improvements 36.59
Boiler Plant Equipment 83.57
Turbine-Generator 74.79
Electric Plant 20.10
Misc. Plant Equipment 5.65
Undistributed Costs 40.26
Other Plant Costs 3.90
Contingency 17.58
Escalation 69.91
Interest During Construction 64.23
General Items, Engineering 7,38
425.47
Limestone/Wet Scrubbing System 50.00
TOTAL ($/kW) 475.47
30
-------
Table 12
Annual Costs for Power Plants Per kW
of Capacity Load Factor 0.7
Combined Cycle
Conventional (No Fine Particle Collection)
Fixed Costs (11.6% $55.10 $46.05
Depreciation (5%) 23.75 19.85
Maintenance (4%) 19.00 15.88
Dolomite/Limestone 2.40 4.20
Power (for gas cleaning) 0.34 1.70
Fuel 43.98 42.20
TOTAL $144.57 $129-88
31
-------
4.2 MATERIALS FABRICATION AND TESTING
Objectives. To develop, fabricate, and test small, flat
aluminum oxide filter membranes having satisfactory pore size, open
area, strength, and filtration characteristics.
Fabrication Approach and Results
Fabrication Summary. Gamma aluminum oxide was electro-
chemically fabricated as flat 4.5 cm diameter discs 100 to 250 microns
thick, chemically treated for barrier layer removal and pore enlargement,
and fired to 1200°C for conversion to the alpha form. The oxide was
examined for structural integrity, internal geometry, and bend strength.
Over 150 samples were fabricated to varying stages of completion, with
43 samples between about 60 and 190 microns thick sent to Westinghouse
for filter testing. The better samples had average pore diameters near
each surface of generally 0.7 to 0.9 micron with internal pore diameters
probably closer to 0.4 to 0.6 micron, particularly for the thicker
samples.
Formation of Oxide. Electrochemical oxidation of aluminum
under controlled conditions will form an oxide with parallel channels
extending through its thickness. These channels are open at one end
and capped at the other by a removable continuous oxide layer called
the barrier layer. Figure 5 shows the oxide's structure. When
electrochemically fabricated, the oxide is in the gamma form.
The aluminum oxide was fabricated in 4.5 cm diameter discs,
100 to 250 microns thick, from high purity aluminum 1199 (99.99%) in
an oxalic acid electrolyte. The fabrication cell is shown in
Figure 6. In operation, the cell is tilted so that the electrolyte
covers the specimen even with the circulation pumps off. The pumps
deliver fresh, cooled electrolyte to the face of the oxide specimen
at the rate of 600 gph, and cooled glycol to the back aluminum side
of the specimen at the rate of 350 gph. The 1/2 ton heat exchanger
32
-------
Figure 5. Porous aluminum oxide geometry (showing the oxide
thickness or channel length £, pore diameter d,
barrier layer thickness b, and cell size c).
33
-------
0-800V
DC
To Heat
Exchanger
t
Electrolyte
Out
Coolent
In
Cooling
Chamber
Coolent
Out
From Heat
Exchanger
Electrolyte
In
Specimen
Electrolyte
T
Cathode
'Spray
Nozzle
Figure 6. High voltage anodizing cell
34
-------
Is capable of maintaining electrolyte temperature to within 0.5°C at 2°C
when both fabrication cells are in operation. The equipment is capable
of providing up to 800 V D.C., with appropriate current, voltage, and
temperature monitoring. ;
Initially the electrolyte used was 0.5, 1.0, or 1.5% oxalic
acid containing 0.1% titanium (IV) oxalate at 3°C. The resulting cell
size and pore diameters are given in Table 13.
Since a titanium (IV) salt is needed only to minimize the
lower voltage section thickness and to achieve the higher voltages
necessary to produce large pores, but is not appropriate for a reasonable
oxide growth rate, the titanium (IV) salt containing solution should be
used initially for the first part of the anodizing when the voltage is
increased to 600 V or 800 V, with a non-titanium (IV) salt containing
oxalic acid used for the remainder of the run. The simplest approach is
to have two cells, one containing each electrolyte, and to move the
sample from one cell to another at the appropriate time. However, since
the sample could never be placed in the second cell identical to its
positioning in the first cell, thus allowing for uniform oxide growth,
it was necessary to arrange for a change of electrolyte in each cell
with the sample remaining in position. Consequently, the plumbing for
each of two fabrication cells was modified to permit the following
operation.
• Fabrication of the oxide in an aqueous solution containing
1% oxalic acid and 0.1% titanium (IV) salt while the
voltage is being increased from 0 to 600 or 800 V;
• Power turned off;
• Titanium containing solution pumped out of cell;
cell and lines rinsed with deionized water;
e 5% oxalic acid solution pumped into cell;
• Power turned on and desired high voltage section grown;
• Power turned off, sample removed, and new sample put on;
• 5% oxalic acid solution pumped out, cell and lines
rinsed with deionized water, and titanium containing
solution returned to cell for a new run.
35
-------
Table 13
Average Cell Size and (pore diameter)
in A/V for Various Fabrication Conditions
Oxalic Acid
Concentration
(%) 400V 500V 600V 800V
0.5 16.5 (8.3) 14.0 (7.8) 15-3 (6.8)
1.0 16.2 (7.0) - 13.0 (7.5) 13.7 (8.1)
1.5 15.8 (8.0) - 12.9 (7.4)
36
-------
To increase the current density more, the electrolyte temperature was
raised to 27°C for the entire fabrication process. An anticipated
favorable consequence of increasing electrolyte concentration and tempera-
ture was the growth of larger pores within larger oxide cells. With the
o
5% oxalic acid solution at 27°C, the cell size is 27 A/V. This means
o
the as-formed pore diameters at 600 V will be up to 3000 A or 0.3 micron.
Barrier Layer and Low Voltage Section Removal. After the
excess aluminum attached to the oxide is removed by a 5% bromine in
methanol solution, the barrier layer on one side of the membrane and
the low voltage section (formed as the voltage was increased from 0 to
600 or 800 V) on the opposite side of the membrane were removed by
simple treatment with an aqueous nitric acid-hydrofluoric acid solution.
Successful removal of these two unwanted oxide sections was verified by
scanning electron microscope examination. As an additional check that
the unwanted sections had been removed, the calculated thickness of the
high voltage section (determined from time-current-efficiency data) was
compared to the final oxide thickness.
The decrease in oxide thickness after removal of the unwanted
sections was generally about 50 to 75 microns.
Pore Enlargement. This aspect of the program received con-
siderable attention. Both alkaline and acid etching solutions which
had shown some success in the past were systematically studied using
different procedures for exposing the oxide to the etchant. These
procedures included simple immersion of the oxide in the etchant,
forcing the etchant with mild pressure through the oxide pores, and
vacuum impregnating of the oxide with the etchant. The same acid
solution which was used to remove the barrier layer was found to be
the most effective. This solution's effectiveness was further examined
at various concentration strengths using combinations of the following
procedures:
a Soaking of the oxide in the etchant for 1/2 to 11
minutes after a water presoak.
9 Soaking of the oxide in the etchant for 1/4 to 5
minutes without a water presoak.
37
-------
• Ultrasonic treatment of the oxide in the etchant
for 1/2 to 4 minutes after a water soak.
• Ultrasonic treatment of the oxide in the etchant for
1/2 to 3 minutes without a water presoak.
• Treatment of the unstripped oxide with the etchant
for 1/2 to 3 minutes.
• Treatment of the oxide with etchant under
filtration pressure.
After the oxide samples were thinned by ion milling to permit SEM
examination of the interior geometry, only two of the pore enlargement
techniques were partially effective for the interior of the oxide.
These techniques were ultrasonic treatment of the oxide in the etchant
after the oxide has been dried at about 100°C, and treatment of the oxide
with the etchant under filtration pressure. Both techniques needed
improvement because the interior section pore diameters were not as
large as the exterior section pore diameters. Tables 14 to 17 show
how the pore etching evolved to a final procedure of multiple pressure
etchings, generally with a 25%, 10%, 5% etchant concentration sequence
and the loss of about 12 to 20 microns of oxide thickness. Samples
treated in this manner had larger internal pores than their predecessors,
particularly if they were less than 4 mils thick. A brief study was
made of an acetone based etchant, but no significant advantage was gained
by its use.
Even after many months of examining pore enlargement, the
pore diameters at the center of the membrane were probably no larger than
0.4-0.6 micron, even though the diameters closer to the two surfaces
were frequently 0.8 micron of greater. Figure 7 is a cross-section
micrograph of an etched sample and Figure 8 is a surface micrograph
of an etched sample.
Samples returned from Westinghouse were examined with the SEM.
Some showed structural irregularities which could have aided in their
fracturing. Some of the samples which broke in processing prior to
being sent to Westinghouse had the same irregularities, but others
which broke in processing had no apparent irregularities. Consequently,
with such inconclusive data, no fabrication procedural changes were made.
38
-------
VO
Table 14
Pressure Pore Enlargement
Sample
W-43B-2
W-60-1
W-6
W-l-1
Etchant
Thickness Cone. Pressure
(u) Support (%) (psi)
163 None 5 Electrolyte
side up
4
5.5
Barrier side
up
4
150 None 5 Electrolyte
side up
4
5
Barrier side
up
4
73 None 5 1
1.5
100 Yes* 5 One side
4
Other side
8
Time at
Pressure
(min.)
5
12
2.5
5
1
<1
3
5
21
33
Etchant
Flow Rate
(drops/sec.)
0.18
0.22 increasing
to 0.4
2.9
1.1
liquid poured
out
0
1.3 decreasing
to 0.16
0.28 decreasing
to 0.13
0.03
Observations
small effect
no effect
no effect
sample broke
sample gouged
-------
Table 14 (continued)
.£-
O
jjample
W-l-2
W-19-1
W-23-1
W-32-1
W-33-1
W-34-1
Etchant
Thickness Cone. Pressure
(y) Support (%) (Psi)
100 Yes 5 One side
4
Other side
4
112 Yes 25 4 to 6
7
65 Yes 10 4
5
Time at
Pressure
(min. )
25
12.5
6
2
5
20
133 Yes 10 electrolyte side up
4 5
5 7.5
barrier side up
4 7.5
160 Yes 25 One side only
4
120 Yes 25 One side only
4 to 9
10
8
6
4
Etchant
Flow Rate
(drops/sec. )
0.29 decreasing
to 0.5
0.67
0
0.28
0
0.28 decreasing
to 0.22
0
0.28
1 decreasing
to 0.67
0.06 increasing
to 0.14
0
0.67
Observations
small effect
Interior pores
about 0.35-0.4 p;
etching severe
too severe
slight effect
interior pores
about 0 . 4 p
interior pores
about 0.4 p
*Milipore filter used as support
-------
Sample
W-65-1
W-66-1
W-69
*W-70
W-71
*W-73
Calculated
High Voltage
Thickness (y )
Table 15
Summary of Pressurized Etched Oxide Samples
Etching
Oxide Pressure Time Final
Side Up (psi) (min.) Thickness (y)
160
160
157
154
153
160
electro.
barrier
electro.
electro.
barrier
electro.
electro.
electro.
barrier
4-10
4-7
4-6
4
4-5
4-5
4
4-7
4-7
11
8
6
4
7
3.5
6
5
103
98
135
127
100
127
Observations
Holes in center
Perimeter
damaged. Holes
in center.
Channels opened
to 2i 0.5 y
Few holes at
perimeter.
Channels opened
to >_ 0.5 y
Good appearance
except for chip
at perimeter.
Diameters at
surface ^0.6 y
Broke, Pores
^0.6 y
Good appearance.
Diameters at
surface ^0.65 y
-------
Table 15 (continued)
Calculated
Etching
•P-
t-o
Sample
W-74
W-76
High Voltage
Thickness (p )
irregular
155
Oxide
Side Up
electro.
barrier
Pressure
(psi)
4-7
6-8
Time
(min . )
6
6.5
Final
Thickness (p )
132
Observations
Cracked upon
etching
Holes, Diame
at surface
W-77
155
rough barrier
electro.
4-7
*W-78
*W-81
*W-82
W-84
*W-83
157 electro.
barrier
155 electro.
barrier
149 electro.
barrier
152 electro.
barrier
159 electro.
barrier
4-7
4-7
4-7
4-7
4-7
4-9
4-7
4-8
4-7
4-9
6
5
6
6
6
7.5
6
6
6
7
130
124
147
125
127
Interior fair
Holes
Good appearance
Good appearance
Good appearance
Small holes in
one section.
Diameters at
surface ^0.55 u•
Internal looks
good
Good appearance
-------
Table 15 (continued)
Calculated
Etching
Sample
*W-85
*W-86
W-89
W-90
*W-92
*W-93
*W-94
*W-95
High Voltage
Thickness (y)
160
120
150
160
160
170
170
Oxide
Side Up
electro.
electro.
barrier
electro.
barrier
electro.
barrier
electro.
barrier
electro.
electro.
electro.
barrier
Pressure
(psi)
4-7
4-7
4-7
4-7
4-7
4-6
4-7
4-7
4-7
4-7
4-7
4-7
7
Time
(min.)
7
7
7
6
6
6
6
6
7
6
6
6
5.5
Final
Thickness (y )
124
122
112
124
124
122
124
112
Observations
Good appearance.
Diameters at
surface ^0.55 y
Small holes
Small holes and
large holes in
center
Holes and crack
in one area.
Diameters at
surface ^0.8 y
Holes around edge
Good appearance
Good appearance.
Diameters at
surface ^0.54 y
Many small holes
-------
Table 15 (continued)
Calculated
Etching
Sample
W-96
W-97
W-98
High Voltage
Thickness (u)
250
222
165
Oxide
Side Up
electro.
electro.
barrier
electro.
barrier
Pressure
(psi)
4
4-7
4-7
4
4-7
Time
(min . )
1
6
5.5
2
7
Final
Thickness (u)
223
129
113
Observations
Good appearance
Many small holes
Large and small
holes
W-99
*W-100
*W-101
166
180
245
electro.
barrier
barrier
4-7
4-5
107
135
157
Hole where
cracked after
fabricated
Good appearance
Diameters at
surface ^0.5 u
Chip near edge
Good appearance.
Diameters at
surface M).6 p
*W-102
160
barrier
4-6
4.5
102
Good appearance
-------
Calculated
Table 16
Pressure Etched Samples
Etching
Sample
W-103*
W-105
W-107*
W-108
W-109*
W-110*
W-112*
W-113*
W-114
W-115*
W-116
High Voltage ' Oxide Pressure
Thickness (p) Side Up (psi)
157
210
185
217
217
125
200
195
120
125
195
Barrier side
Barrier side
Barrier side
Barrier side
Barrier side
Barrier side
Barrier side
Barrier side
Barrier side
Barrier side
Barrier side
4-6
4-6
4-5
4-6
4-6
4-6
4-5
4-5
4-5
4-5
4
Time
(min.)
5
6
6.5
6
6.5
5.5
6
5.5
3.5
3
2.5
Final
Thickness (P) Observations
135
185
150
100
167
100
160
150
85
100
135
Good appearance
Cracked in center
upon removal from
etching apparatus;
otherwise , good
appearance
Good appearance
Holes
Good appearance
One pinhole
Good appearance
Good appearance
Holes
Good appearance
Broke
*Sent to Westinghouse
-------
Table 16 (continued)
Calculated Etching
Sample
W-118*
W-120*
High Voltage
Thickness (p )
177
205
Oxide
Side Up
Barrier side
Barrier side
Pressure
(psi)
4-6
4-6
4
Time
(rain. )
6
5.5
2.5
Final
Thickness (y )
150
157
155
Observations
Good appearance
Good appearance
Good appearance
-------
130
Table 17
Pressure Etched Samples, Barrier Layer Up
Sample
121
122
123*
124
125
126
127
128*
129
25% Etching 10% Etching
Pressure Time Pressure Time
(psi) (min. ) (psi) (min.)
4 2.5
4-5
4
4 3.0 4
4
(5%) 4
4 3.0 4
4 4.0 4
(5%) 4
4-5
4
(5%) 4
(5%) 4
3.5
2.5
4.0
2.5
5.0
4.0
5.5
3.0
4.0
1.5
3.5
6.5
Final
Thickness
(mils)
3.0
3.5
~ 6.1
2.8
3.6
7.3
5.8
4.2
5.1
Observations
Has holes
Broke
Good
Broke
Good. Cracked with
cermet support
Broke
Hole in sample
Good
Good. Cracked with
4.0
5.0
6.4
cermet support
Good. Cracked with
cermet support
-------
Table 17 (continued)
oo
Sample
131
132*t
134
135*t
137*t
138*t
139*t
140*t
141*t
142*t
143*t
144
I46*t
149* t
25% Etching
Pressure Time
(psi) (min.)
4
4
5
4
4
4
4
5
4
5
7
4
3.5
5.0
4.5
4.0
4.5
5.0
5.0
5.0
4.5
5.0
6.0
3.0
10% Etching
Pressure Time
I (psi) (min.)
4
4
4
4
4
4
4
5
4
4
7
5
4
4
(5%) 4
5.0
5.0
2.5
1.75
5.0
5.0
5.0
4.5
2.5
1.5
5.0
5.0
3.0
3.0
1
Final
Thickness
(mils)
5.1
6.5
5.1
4.7
6.0
6.7
6.4
6.3
5.9
3.7
6.5
5.3
4.8
2.9
Observations
Pin holes
Good
Pin holes
Good
Good
Good
Good
Good
Good
Good
Good
Broke upon removal
from apparatus
Good
Good
-------
Table 17 (continued)
Sample
150
151t
156*t
157*t
158*t
168*t
172*t
25% Etching
Pressure Time
(psi) (min.)
4 3
4 5
4-6 6
10%
Pressure
(psi)
4
4
4
4-6
4
(5%) 4
0
(5%) 4
Etching
Time
(min.)
3.0
3.0
3.0
5.0
3.0
0.5
1.0
0.5
Final
Thickness
(mils)
3.9
3.8
3.7
4.6
3.3
2.7
3.3
Pin
Good
Good
Good
Good
Good
Good
Observations
holes
*Sent to Westinghouse
tCeramic disc supports
-------
••;•
,- •*•**•• ir •*;
. - - ,,«- «fc% •
*«„ * ^»,
;:^^P^; ^ 1^>:i>&*
;><^Tir. . -fc^i-11
, "-xg^^I^E,^t:%-\£>5^
k ,* •• . S >*:sSr ^^^s<*. i. *•>••
,
.*. »^s2*"Vr *~S -""TS^Sfc-i': *
* ~
Figure 7. Edge view showing channels of
sample W-66 at 1250X (1 cm =
8 microns). This sample section
is 65 microns thick.
•
RM-72206
-------
• W--" *• *-^t
*^ 4t ., J( 11^^ ifc"
• *wls^ fiP-*"
Figure 8. Sample W-61A. Barrier layer side
of oxide formed at 800 V. Cell size
-\/l.3 y; pore diameter 0.84 p (5200X),
;
RM-72207
-------
Conversion to the Alpha Form. After removal of the barrier
layer and low voltage section from the oxide and enlargement of the pores,
the oxide was heated in air at 1200"C for 20-30 minutes in order to
effect the change from the gamma to the alpha form. The membranes
converted to the a-alumina form showed no damage to the channels from
the treatment. The pore diameters were slightly larger as the pore
walls shrank. The membrane thickness also shrank about 3%. The alpha
form seemed less compilable than the gamma. Since the oxide tended to
curl during heating, a confining pressure to keep the oxide flat was
applied by use of thin zircon slabs. Applying this confining pressure
is an art. However, bend strength measurements on the pore enlarged
alpha and gamma forms showed them to be comparable in strength, as
reported in Table 18.
Sample 2267-11 was not pore enlarged, as indicated by the
low porosity. The 7.3 mil thick sample was about 5 times stronger
than the 5.3 mil thick samples and 3 times stronger than the 6.3 mil
thick sample. Measurements were done as described by T. R. Wilshaw's
method for measuring fracture stress. Wilshaw's method is a symmetrical
central bending test in which the disc shaped sample is placed on a
support ring which is approximately half the diameter of the sample.
The load is put onto the top of the sample by a loading ball which is
approximately half the diameter of the support ring. The fracture stress
o~p, is then given by
a2 - r2
2U To r
where P is the load, t the sample thickness, v Poisson's ratio, a
the support ring radius to its inner diai
section radius, and b the sample radius.
the support ring radius to its inner diameter, r the loading ball flat
52
-------
Sample
W-94
2267-11
W-99
¥-100
W-101
W-116
W-126
Thickness
(mils)
4.9
5.0
4.2
5.3
6.3
5.2
7.3
Table 18
Mechanical Testing Results
Porosity
Form
15
3
15
15
16
15
15
Alpha
Gamma
Gamma
Alpha
Gamma
Gamma
Gamma
Gamma
Bend Strength
(103 psi)
11
350
7.3
7.3
8.7
15.1
9.3
47
53
-------
Data acquired some years ago indicated that at 800°C the
approximate temperature of the gasifier gas stream, the gamma oxide
would convert to the alpha in about 20 hours. Since it appeared
possible that the oxide could be mounted in the gamma form, to
facilitate sample production, samples in the gamma form were used for
the pressure testing.
Supports for the Filter Membranes. During the filter testing,
it became apparent that the membranes needed to be supported. The
rationale in designing the supports was to make them mechanically
sufficient to permit room temperature filter testing of the membranes,
and to optimize their design for high temperature application in
Phase II. Two different mechanical supports were examined. The first
consisted of a mullite ring around the circumference on each side of
the membrane with a reinforcing lattice on one side of the membrane
of an alumina cermet. The second consisted of a highly porous solid
ceramic disc about 2 mm thick placed on one side of the filter membrane
with a 6 mm hole latticed 2 mm thick disc of the same ceramic material
placed on the other side of the filter membrane, and the entire assembly
held together at the circumference with a silicone rubber, sufficient
for room temperature testing.
The mullite ring-cermet lattice support was used for 4
samples. The first was tested at Westinghouse, the remaining 3 never
left Horizons, as it was found that the alumina cermet shrank upon
curing and cracked the membranes attached to it.
The second mounting procedure - backing each side of the
membrane with a porous ceramic disc - was most successful in enabling
testing of the membranes to 10 psi.
Filter Tests. The filter test program had three major
objectives:
• To establish the gas flow/pressure drop characteristics
of the membrane filter.
54
-------
• To establish that the filters could achieve high (>90%)
collection efficiency on a polydisperse submicron dust.
f To demonstrate an operable cleanup technique which would
allow the filters to be used in a cyclic manner for an
indefinite period of time.
To achieve any of these ends it was first necessary to mount
and seal the test specimens in a holder. This proved to be difficult.
Initial specimens had been fired to convert the alumina to
the a form. This distorted the filters, producing a "wrinkled" edge
which was virtually impossible to seal, without cracking the specimen.
By avoiding the firing (i.e., leaving the specimens in the as fabricated
Y form) this problem was overcome.
From the economic analysis, an acceptable flow rate/pressure
drop criteria of >750 cc/min flow at a pressure drop of 100 kPa was
adopted. Initial tests showed that the filter specimens would break
under test before this criteria could be met. Several support systems
were used without success.
Finally, the filters were sandwiched between two heavy
porous ceramic discs to provide support while flow characteristics
were determined. Of twelve samples tested in this manner, only two
showed acceptable flow rates, and one of those two broke in spite of
the heavy supports.
Two filtration tests were performed at reduced flow rates using
filters with 0.5 pm pores. Filtration efficiency was greater than 90%
for submicron particles, but both samples were broken before any
definitive tests could be made.
Because of the poor mechanical properties of the material, the
poor reproducibility of the specimens, and the lack of any practical
support system, no further filtration tests were attempted.
55
-------
SUMMARY - PHASE II
In Phase II of this program the emphasis shifted from attempting
to develop and test a new ceramic membrane filter, toward the identi-
fication and testing of existing forms of ceramic materials that held
the possibility of being utilized as hot gas filters. The second
phase effort evolved into several distinct tasks: procurement of
materials, screening tests, hot bench scale tests, hot high flow tests
and conceptual design with cost comparison.
56
-------
4.3 PROCUREMENT OF MATERIALS
Objective. The goal of this task was to search for and identify
existing ceramic materials that offered the possibility of being utilized
as filters in application where elevated temperatures and corrosive
atmospheres would be encountered.
Procedure. This task was an ongoing effort throughout the entire
second phase, however, the most concerted activity occurred in the
initial stage during the transition from Phase I to Phase II. Personnel
from Horrizons Research, Inc., Acurex/Aerotherm and Westinghouse R&D
participated in this effort. Personnel from Horrizons Research, Inc.
identified several promising forms of zirconia and alumina papers, felts
and boards. Samples of several forms of these materials were obtained
from Zircar Products, Inc. through Horrizons.
Acurex Corporation/Aerotherm Division had just completed a project
for the EPA under Contract 68-02-1318, and was in the process of
completing a final report, "Evaluation of Ceramic Membrane Filters As a
(2)
Controf for Fine Particulate." Since one task included in this effort
was an industry survey of possible materials, their input was most
helpful. A summary of their survey results is presented in Table 19.
Because of the close association of personnel at Aerotherm with this
effort, a subcontract was negotiated with them in order that they
might continue their efforts in the identification of new materials.
Through contact with the Ceramics Department at Westinghouse R&D
a cross flow, honeycomb, silicon carbide material manufactured by
Norton Company was identified. Finally, contacts made by Westinghouse
at a jointly sponsored, EPA, ERDA Symposium on fine particulate control
led to the identification of a cross-flow monolithic material produced
by the W. R. Grace & Company.
57
-------
Table 19
Survey Results
Company
1. Babcock & Wilcox Co. ,
Refractories Div.
Augusta, Georgia
2. Coors Poreclain Co.
Golden, Colorado
3. Corning Glass Co.
4. Dollinger Corp.
Rochester, New York
5. General Refractories
Bala Cynwyd, Pa.
6. Horizons, Inc.
Cleveland, Ohio
7- 3-M Company
Saint Paul, Minn.
8. Norton Company
Worcester, Mass.
9. Selas Flotronics
Springhouse, Pa.
10. Wisconsin Porcelain
Sun Prairie, Wisconsin
Code*
C
c,f
f,s
f
c,s
C
c,s
c,s
f
c,f
Did not think they had any materials
that would work for the application.
Several small scale filters for this
oarticle size. No high temperature
filter experience. Samples sent to
Aerotherm. Nominal wall thickness
-6 mm (1/4").
Nothing for this application. Celcor
cordierite monoliths a possibility.
Some filter elements that would work
for this application. Relatively
thick wall, large pressure drop
(-100 kPa (1 atm). Elements
available for testing.
Nothing for this application.
Versagrid cordierite honeycomb is a
possibility; however geometrical
constraints a problem.
This membrane (-0.25 mm thick) being
developed for testing in this applica-
tion under Westinghouse - EPA program.
AlSiMag 795 (cordierite) honeycomb
structure should work for this appli-
cation. Elements available for test-
ing. Low Pressure drop. High surface
area to volume ratio. Good process
control over pore size, porosity and
membrane thickness. Samples sent to
Aerotherm.
Two materials suitable for thick
walled filters. Also Spectramic
honeycomb is a possibility.
Micro-porous porcelain element
would be suitable. Pressure drop
-50 kPa. Sample sent to Aerotherm.
Thick walled filter elements
similar to 4 and 9. No membranes.
*c - ceramic manufacturer
f - filter supplier
s = catalyst support manufacturer
58
-------
Results. In general the materials identified could be classified
in two groups - thick walled elements and thin walled honeycomb structures.
The materials that were screened as being representative of the spectrum
of thick walled materials were: Zircar 135C, Zircar ZAL-15, Selas 01, 10,
XFF and a FMI (Fiber Materials, Inc.) material. The Zircar 135C sample
was an alumina paper tube 3.5 cm long by 2.5 cm inside diameter. The
wall thickness was 0.47 cm. The tubular structure appeared to have been
made by wrapping hundreds of layers of the alumina paper around a mandrel
prior to firing. In this form the material was not brittle and could be
handled and worked with ease. It is reasonable to expect that a much
thinner walled form of the material could be successfully used as a
filter. The Zircar ZAL-15 material and the FMI material physically
resembled each other exactly. The specimens were flat boards approxi-
mately 1 cm thick. The materials are very low density, soft and porous.
These materials could be shaped easily which facilitated the screening
tests, but there was some tendency to crumble at compression seals.
It is expected that mechanical problems with sealing might be experienced
in a commercial device using this material because of its low strength.
The samples of filter media from Selas Flotronics were flat discs with
a diameter of 8.75 cm by 1.25 cm thick. The materials were a micro-
porous porcelain with varying porosity. Two of the samples Selas 01 and
10 were much less permeable than the third, Selas XFF, although there
were no easily discernable differences in appearance.
Of the three materials grouped in the thin walled cross flow
honeycomb category samples of the 3M Thermacomb and W. R. Grace & Co.
material were obtained for testing. Efforts to obtain samples of the
Norton Company cross flow material were not successful. The 3M cross
flow structure is made up of several layers in the following pattern:
a thin (0.25-1.5 mm) porous cordierite sheet, a layer of cordierite
corrugations similar in apperance to those used in cardboard, another
flat of sheet of cordierite followed by another layer of corrugations
59
-------
oriented 90° from the corrugations below. The presently available forms
of the material have 1.97, 3.15 or 4.72 corrugations per cm. The cross
flow element is shown with other available configurations in Figure 9.
The material obtained from W. R. Grace & Company is virtually the same
with the exception that straight, perpendicular dividers are used giving
rectangular holes rather than the triangular holes seen in the
Thermacomb. The Grace material tested had approximately 8.5 holes per
cm and an equal number of layers per cm. One suggested method of using
these materials as filters is illustrated in Figure 10.
These materials have many properties that make them attractive as
filters. Among these are (1) working temperature to 1200°C (2) very
good mechanical strength despite thin separators (3) excellent resistance
to thermal shock (4) excellent resistance to corrosive atmospheres and
(5) very high surface area to volume ratios. It is estimated that the
2 3
Thermacomb material tested had 3.27 cm filter area per cm of element
2
while the W. R. Grace & Company material had approximately 6.52 cm of
3
usable area per cm of element. One might expect a ratio on the order
2 3
of 0.1 cm filter area per cm for a typical fabric filtration facility.
60
-------
B) SC SPLIT CELL (Note Separator)
XFSC
CROSS-FLOW,
SPLIT CELL
Note separators
and corrugations
at 90°
XXSC
CRISS-CROSS,
SPLIT CELL
Note separators
and corrugations
at 45°
XXHC
CRISS-CROSS,
HONEYCOMB
with corrugations
at 45°
Note there is no
separator.
Figure 9. Structural shapes for 3M ThermaComb.
61
-------
ho
Figure W. 3M cross-flow ceramic monolith.
-------
4.4 SCREENING TESTS
Objective. Since the overall goal of the Phase II effort was to
demonstrate successful hot gas filtration on a fairly large scale, it
was decided to screen the materials identified and isolate one or two of
the candidates that looked most promising. Once identified, virtually
all further work would be confined to developing techniques for testing
these materials and recording their filtration characteristics.
Procedure. The screening tests used consisted of two parts. First
a measurement of pressure drop as a function of flow at ambient conditions
was made. This was followed by actual filtration tests to determine
overall particle removal efficiencies and possible techniques for on line
cleaning of the filter.
Since the thick walled samples obtained varied greatly in size and
shape it proved necessary to fabricate test jigs for each of the filters
screened. In general these holders were of simple construction from
plexiglass so that the inlet and outlet of the filter could be observed
visually. Provisions for pressure drop taps and dust sampling were
included.
The overall efficiency of the filters was measured by isokinetically
sampling the dust laden stream from which the inlet to the filter was
taken, and simultaneously filtering the entire outlet from the test
specimen with either a milipore or a glass mat filter in the milipore
filter holder. Dispersed dust was supplied to the test specimen by
3
sampling either the inlet or outlet of a 3 m /min cyclone that had been
in use. The test dust was fed to and dispersed in the cyclone system
(3)
by a modified harvard dry dust disperser. This dust was a ground
limestone classified to be all less than lOia. The size distribution
of the cyclone inlet and outlet dust as measured by a Sierra Cascade
impactor is given in Figure 11. Dust loadings at the inlet could be
O
regulated in the range of 1 to 5 gm/m while the finer outlet dust
loadings were generally about one-tenth this range.
63
-------
.01
.5
50
10
.8 20
£ 30
t 40
I 50
1 60
t 70
0 80
90
95
98
99
99.5
Curve 686415-B
1 1 T
99.99
i I I
0.1
1.0
10.0
Fig. 11- Size distribution of test dust
64
-------
It has been assumed throughout this work that the cleaning tech-
nique used must allow continuous operation since' methods that require
cooling and removal of the filter media for cleaning are not thought to
be practical for commercial applications. The cleaning method most
frequently considered for such applications basically employs some form
of back washing from the clean side of the filter. This flushing of
the filter surface can be in the form of either a sharp pulse in the
reverse direction or a gentler continuous flow for a longer period of
time. In applications where the filter can physically withstand the
pulse cleaning option, it is usually preferred since it is quick and
requires less cleaning gas. For the screening tests, continuous flow
cleaning was accomplished by simply isolating the filter outlet from
the exit sampler and then allowing filtered house air to flow back
through the filter from the clean side. The pulsed cleaning option was
tested by charging a 0.385 I vessel to the desired pressure and then
venting this vessel to the clean side of the filter through a solenoid
valve.
The system for mounting the 3M Thermacomb material was somewhat
more complicated because of the unusual configuration of the material.
Acurex/Aerotherm provided a mounting plate and top retainer similar to
the one shown in Figure 12. A plexiglass enclosure was fabricated to
mate with the support plate provided so the filtration process could
be visually observed. The completed test jig is similar to the one
shown in Figure 13. It was necessary to seal the Thermacomb to the
retaining and support plate with RTV to prevent leaking.
Results. The pressure drop vs. flow characteristics for some of
the thick walled elements is summarized in Figure 14. The most porous
Selas filter and the FMI material had relatively low pressure drops
(6.5 and 3.8 kPa) at face velocities of 20 m/min, while the two fine
pore Selas materials had relatively high pressure drops in the range
of 80-100 kPa at the same face velocity. The two Zircar materials
were received and tested after these tests were completed. The
65
-------
ON
_L
Fiqure 12. 3M element low temperature holder.
-------
A
Nl
Figure 13. 3M element and holder mounted inside pipe.
-------
Curve 686421-A
100.0
03
10.0
o_
1.0
1.0
I L
SELAS -10
SELAS-XFF
F.AU. -J
i i i
10.0
U (m/min)
100.0
Fig.14- Face velocity vs pressure drop for thick walled filters
68
-------
Zircar ZAL-15 proved to have virtually the same pressure drop-flow
characteristics as the FMI material. The Zircar 135C alumina paper
tube had a flow resistance slightly higher than the Selas XFF and the
FMI material as is shown in Figure 15.
The screening tests with the 3M Thermacomb were carried out on
a cube of that material roughly 5 cm on a side with approximately
2
0.029 m of surface area for flow. The results of this test are
shown in Figure 16. At a face velocity of 10 m/min the filter had a
pressure drop of approximately 10 cm H_0 or 0.98 kPa which indicates a
resistance to flow that is lower than the FMI material. Much later in
the program, 15 cm cubes of Thermacomb were tested. These elements had
2
an estimated filter area of 0.804 m . At a pressure drop of 4 kPa a
flow of 8 m /min or nearly 10 m/min was measured. This is a slightly
higher pressure drop than observed in the small cube and it corresponds
to a flow resistance that is virtually the same as the FMI material
(Figure 17). The W. R. Grace & Company samples were received late in the
program and were not tested with clean flow.
Upon completion of pressure drop-flow measurements, experiments
sampling the test dust were begun at ambient conditions to characterize
filter efficiency and cleanability- The filtration runs with the thick
walled elements were flow tests with typical inlet mass loadings in
3
the range of 0.05 to 0.2 gm/m . The results from the tests of the
three Selas materials are shown in Figures 18, 19 and 20. Experiments
with the most porous Selas element (XFF) were carried out at two face
velocities, 11.9 m/min and 3.0 m/min. During these runs no penetration
of dust could be measured, and the pressure drop vs. time curves
indicated that the filter could be cleaned repeatedly. The cleaning
cycle consisted of a single pulse from the 0.385 £ vessel charged to
1 atm pressure. This element seemed to suffer a small irreversible
increase in pressure drop after being exposed to the test dust, however,
this increase in pressure drop did not appear to persist in subsequent
cycles. The other two Selas materials were tested at a face velocity
69
-------
100.0
Curve 627'97-;
I I
Zircar 135 C
10.0
Q_
<
1.0
1.0
10.0
U(m/min)
Fig. 15-AP vs U for Zircar 135C material
100.1
70
-------
CVJ
700
600 I
500 -
400 -
Curve 685633-A
5 300
200 -
100 -
0 /i i i i I i
i I i i i i I i i i i i i i i i i i i i i i i i i i
01234567
AP(cmH20)
Fig. 16 - Flow vs. pressure drop for 3M thermacomb
71
-------
10.0
8.0
6.0
4.0
tu
Q_
2.0
Curve 688441-A
1.0
1
1.0
2.0
4.0 , 6.0 8.0 10.0
Q (m /m\n)
fig. 17— Pressure drop vs flow for clean 15. 25 cm thermacomb cube
-------
Curve 686422-A
Cave=0.10gm/
05 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90
Time (mins)
Fig. 18- A P vs time for SELAS - XFF
73
-------
40
Curve 686419-A
U = 4.9m/rnin .
C =0.20gm/m:
20
0 10 20 30 40 50
Time (min)
60
70
80 90
Fig. 19-A P vs time for SELAS - 01
74
-------
Curve 686420-A
Ca =0.18gm/m
40 50
Time (mins
Fig. 20-A P vs time for SELAS - 10
90
75
-------
of 4.9 m/min and had substantially higher pressure drops. Again, no dust
penetration could be measured, however, there was some difficulty in
cleaning the filters and the pulse pressure had to be increased to 2 atm
to obtain to maintain a stable cycle. It was noted that these less
porous materials did not seem to suffer as much irreversible increase
in flow resitance as did the more porous XFF sample. The data from
tests with the FMI material are presented in Figure 21. As with the
other materials virtually no penetration was measured during this test.
This material experienced only a slight increase in flow resistance after
having been exposed to the test dust. This material was readily cleaned
to a repeatable degree by a one atm pulse. Tests with the Zircar ZAL-15
material gave the results presented in Figure 22. As can be seen there
was a small increase in the initial pressure drop after the first cycle,
but this increase leveled out after a few cycles. The magnitude of
the increased resistance to flow is shown in Figure 23 where pressure
drop vs. face velocity of the clean "as received" material is compared
with the material after one cycle. The holder used in these tests did
not allow the use of a high pressure pulse for cleaning so a continuous
back flush at about 1 atm for 2-5 seconds was used to clean the filter.
The tubular Zircar alumina paper (135C) was tested and found to be,
like the other thick walled elements, 100% effective in removal of the
test dust. The pressure drop vs. time data for this test are presented
in Figure 24. The filter holder used did not allow a pulse type of
cleaning so a continuous flow at about 0.7 atm for several seconds was
used. Even without the pulse cleaning this material suffered virtually
no irreversible increase in pressure drop over the few cycles examined.
It is believed that this material could be formed with a wall thickness
considerable less than the 0.47 cm used in these tests and not reduce
its ability to remove particulates. This would reduce pressure drop and
76
-------
Curve 6864I8-A
40 50
Time (min)
Fig.21-A P vb time for FMI material
77
-------
oo
00
24
22
20
U
12
10
8
6
4(
2
0
0
—I 1 1—
U = 35.6 m/min
C =.067gm/m
ave y
1
1.
6 8 10 12
U 16 18 20
Time (Mins)
22 24 26 28 30
Fig. 22~ AP vstime for Zircar alumina board (ZAL-15)
-------
Curve 686733-B
10.0
9.0
8.0
7.0
6.0
5.0
4.0
CO
1 3.0
2.0
1.0
10
Used For One Filter
Cycle and Blown
Back
20
30 40 50 60 70
U(m/min)
90 100
Fig.23-AP vs flow for Zircar alumina board (ZAL-15)
79
-------
Curve 687199-A
oo
o
ro
28
26
24
22
20
18
16
1 14
% 12
10
8
6
4
2
0
0
Zircar 135C
U = 9.51m/min
= 1.28gm/m
10
20 30 40 50 60
Time (min)
70
80
90
100
Fig.24 — A P vs time for Zircar 135C material
-------
make cleaning easier. Such a filter could be very attractive for hot
gas filtration.
Filtering tests with the thin walled honeycomb structures was
initially confined to samples of 3M Thermacomb since the W. R. Grace
material was not received until late in the project. Tests with the
Thermacomb material led to the following observations:
(1) Pressure drop rose rapidly during an initial transient period
to a relatively steady pressure drop, from which it rose much
more slowly at a constant rate.
(2) Some dust could be seen on the filter outlet walls but the
amount did not increase after about a 5 minute period.
(3) Overall efficiencies based on typical sample times of 1 min.
inlet and 20 min. outlet were around 98 to 99%.
(4') The size distributions obtained from inlet and outlet
sampling with cascade impactors appeared to be the same,
although there was considerable uncertainty in the outlet
sample due to its small size.
(5) Reversing the flow of air removed considerable amounts of
dust from the corrugations and returned the initial steady
pressure drop to its previous value. Visual examination of
the filter indicated that relatively few of the corrugations
were swept entirely clean, but this apparently had little
effect on the filter's performance.
Based on the observation that the inlet and outlet size distri-
butions appeared to be identical it was concluded that relatively large
holes existed in the filter. Finding the holes by examining the
outlet side of the filter was not easy because the dust and filter
material are virtually the same color. Eventually two corrugations were
found to be leaking by probing each corrugation with a dark wire and
checking for dust on the wire. Each leak was marked and a visual
observation was made during a subsequent run. For the first several
minutes a large amount of dust was passed through the leaks, but the
81
-------
leaks appeared to heal themselves in 3 or 4 minutes, after which no
dust was seen to escape, for the remainder of the filtering cycle. This
explained both the high overall efficiency observed, and the similar size
distributions of the inlet and outlet. The corrugations that leaked were
then plugged and a series of constant flow tests were run.
3
An attempt to set the dust loadings at 2.0 gm/m was made, but
problems with the dust dispensing apparatus did not allow good control of
the feed rate. This necessitated the taking of several dust samples
throughout the runs so that a reasonable average dust concentration could
be recorded. At the beginning of each series of constant flow runs, the
filter was removed and cleaned as completely as possible. At the end of
subsequent cycles the filter was cleaned only by a pulse of back flush
air. Figures 25 and 26 show the results of a series of runs carried out
at flow rates of .085 and .057 m /min corresponding to face velocities
of 2.92 and 1.95 m/min.
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 contri-
bution due to the media AP , and due to the accumulated cake, AP .
m c
APt _ = AP + AP
tot m c
For given gas properties, the pressure drop across a filter media, AP ,
m
is proportional to the face velocity, U and filter thickness, L .
m
The pressure drop across the cake is similarly proportional to the face
velocity and cake thickness L . If a constant dust concentration, C, is
assumed, the cake thickness is proportional to the produce of the
measured variables, UCt, where t is time. The total pressure drop for
a given gas, temperature and filter is given by:
= K'U + KU2tc,
82
-------
Curve 68602k-B
CX3
Q.
~ 3
Q.
1IIIITI
Constant Flow = 0.085 m /min
Filter Area = 0.0291m2
Cave [gm/m ]
Cave = 2.7
Cave = 2.1-
Cave = 1.8-
- Cave = 1.8-
J I
l
I
I
I
I
I
I
8 12 16 20 24 28 32
36 40 44 48 52 56 60 64 68 72 76 80
Time (min)
Fig. 25- Pressure drop vs time for ThermaComb
-------
4.8
4.4
4.0
3.6
3.2
15 2'8
Q_
I 2'4
< 2.0
1.6
1.2
.8
.4
0.0
Curve 686022-B
1 1 1 I-TT-
Constant Flow = 0.057 m /min
Filter Area = 0.0291 m2
Cave [gm/m ]
• Cave = 2.1 —»
*— Cave = 1.1 —>
-Cave = 1.5
-Cave
= 2.4 -J*— Cave = 5.0 —4
J L
0 8 16 24 32 40 48 % 64 72 80 88 96 104 112 120 128 136 144 152 160
Time (min)
Fig. 26- Pressure drop vs time for ThermaComb
-------
where K1 and K are constants for a given filter and incompressible cake.
The linear dependence of AP^.. 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 &PtQt vs t curves, normalized by division by U2 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 27 presents
these data and indicates that the assumed form of AP is confirmed by
c
all the data.
Two further observations of major importance were made concerning
the initial steady pressure drop.
(1) The magnitude of the initial steady pressure drop was not
prohibitively high.
(2) The initial steady pressure drop remained constant with the
\
adopted cleaning technique.
At the relatively high face velocity of 3.25 cm/sec this ceramic
filter had an initial steady pressure drop of about 2 kPa. A typical
fabric filter at this face velocity could be expected to have a pressure
drop in the range of 1 to 1.75 kPa.
The fact that the initial steady pressure drop does not appear
to increase as the filter is cycled is most important as it indicates
that no discernable deterioration of performance had occurred due to
irreversible pore blockage. Figures 28 and 29 show the filter before
and after cleaning by a pulsed back flush and a back flush accomplished
by a more gradual flow. The pulse was accomplished by pressurizing a
.385 liter vessel to 2 atm total pressure, and subsequently venting
this vessel using a solenoid valve. The more gradual flush was
accemplished by manually opening a valve allowing a flow of approximately
0.25 m3/min to flow for 2 sec. Both methods return the initial steady
pressure drop to the level obtained when the filter has been removed
and carefully vacuumed clean, although visual examination of the filter
reveals that neither the pulse nor the gradual flush completely clean
the corrugations.
85
-------
oo
10.0 x 105
9.0
8.0
Curve 686021-A
o
CD
l/l
O.
7.0
I 6.0
5.0
4.0
3.0
2.0
1.0
0
1 I ' I ' I ' I
- o 0.057 m3/min Data (U = 1.95 m/min)
0.085m7min Data (U = 2.92 m/min)
0
Slope = 1.68 *105 sec 1 =
I
ave
Fig. 27 — Determination of fifter cake constant
-------
Before
After
Figure 28. Before and after 103 kPa pulse.
8 i
RM-67761
-------
%! 3
Before
After
Figure 29. Before and after 17 m3/hr flush for 2 sec
RM-67762
-------
Upon receipt of the W. R. Grace & Company materials a filtering test
at ambient conditions was carried out. The data from this test are reported
in Figure 30. At a face velocity of 0.483 m/min the initial pressure
drop was about 1 kPa which is about twice that measured for the 3M
Thermacomb material. The barrier thickness for the Grace material was
measured at 0.38 mm or about twice that of the Thermacomb (0.2 mm)
which explains the larger resistance to flow. In other respects the
performance of this material quite similar to the Thermacomb, it being
virtually 100% effective in dust removal and able to be cleaned to a
constant level by a 1.36 atm pulse from a 0.385 H container.
As can be seen from the data generated during the screening tests,
several of the ceramic materials appear to have promise as hot gas
filters. The decision to proceed to hot gas testing of the Thermacomb
type materials was based on material's good performance as a filter
and its unique configuration which allows large surface area to
volume ratios.
89
-------
Curve 689454-8
4.0
3.0
< 2.0
1.0
i i i i i i r
Constant Flow =. 028 m /min
Filter Area = .058m2
Cave=2.6gm/m3
Efficiency = 100%
Cold Test
i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i i
i i i i
8 16 24 32 40 48 56 64 72
96 104 112 120 128 136 144 152 160 168 176 184 192 200 208 216 224 232 240
Time (min)
Fig. 30- Pulsed purge = 137.9 KPa 3.0 seconds duration
-------
4.5 LOW FLOW HIGH TEMPERATURE TESTS
Having made the decision to further investigate the properties of
the thin walled, cross flow, honeycomb structures, it was necessary to
construct an apparatus for hot bench scale experiments. The relatively
few samples of Thermacomb available initially were of a size that was
compatable with the roughly 5 cm cubes used in the screening test
apparatus so it was decided to base the design of the hot low flow unit
on this specimen size.
The first effort was directed toward the design of a filter holder
and sealing system suitable for high temperature application. The design
that evolved was similar in many respects to the screening test filter
holder, A sketch of the device is shown in Figure 31. The inlet to the
filter was up through a 3.8 cm square hole in the support plate. The
actual filter element sat on a 0.64 cm groove around the inlet hole.
The top retainer had a matching groove and through hole, however, a
cylindrical containment was welded over the hole in the retaining plate
and a tubing connection through the top plate of the jig was made.
This arrangement allowed a cleaning pulse to be administered straight
down the corrugations in which the collected dust accumulated. The
cylindrical region was packed with 20 mesh alumina chips to distribute
the pulse over the entire 3.8 by 3.8 cm area. The alumina chips were
retained by a stainless steel screen over the hole. During filtration
the cleaned gas emerged from two vertical sides of the Thermacomb and
exited the filter jig through the outlet indicated in the sketch.
During this period a valve on the central, dirty side line was shut
preventing flow straight through. Cleaning was accomplished by
pulsing either or both the dirty side of the filter and the clean side.
The dust which had accumulated in the vertical channels of the filter
was knocked down through the inlet hole and into a collection hopper.
A schematic of the test facility is shown in Figure 32. During
the filtering mode of operation, dusty air flowed through a normally
open solenoid valve V-^ Valves V2 and Vg were closed forcing the gas
91
-------
Dwg. 1684B44
Duct for Air to
Sweep Dirty Side of Channel
Fig. 31- Hot thermacomb test jig
92
-------
Dwg. 6387A60
Cleaning Sequence
0 - Open
«~— >
V,,
4
». — X
1
S~**
\ t
V3
"^_>
>
S
i
N
S
(
i_
O>
*— •
•HMW
jl
h
"v
^
_/
_o;
•—>
^
2,
I
\
i
)
A <
Backflush Pulse^
AT^
VV
Dirty Side Sweep
~©1
n i
V
(V)
V ' J
-Dust
Hopper
Exhaust i
I/ \ HL
j)
I
J
1 \l
N
_^N_
, ^ - uiosea
' Vl V2 V3 V4 V5 V6 V7 V8 V9
tQOCOOCCOCC
tjCOCCCCCOC
t2coccoococ
t3cocccccoo
t5ocooccocc
s
$)
Dusty
Air Inlet
Fig. 32- Flow schematic and cleaning sequence for low flow-high temperature system
-------
up into the filter jig and through the Thermacomb. The cleaned gas
exited the system through V7 into the house exhaust. During filtering
V and V. were open and the pressure drop across the filter was monitored
by a Dwyer pressure gauge-switch. When the pressure drop reached the
set point on the gauge the following sequence was initiated for cleaning.
First V- and V, closed in order to isolate the gauge from the cleaning
pulse. Valves V. and V_. closed while V0 and V0 opened. This directed
1 / o 2.
the dusty air supply into the by-pass leg and directly into the exhaust.
Valve V_ opened the dust hopper. Valves V, and Vc could be opened
z o .)
allowing a high pressure cleaning pulse to vent into either the clean
side of the filter holder or straight down the dirty side. When this
was accomplished the entire sequence was reversed returning the system
to its filtering mode. The final step was to repressurize the pulse
holder with V_ in preparation for the next cleaning cycle. Since this
sequence had to be repeated hundreds of times a solid state automatic
sequencer was designed and fabricated to step the valves through the
proper sequence upon a signal from the pressure gauge-switch. This
system worked very well since it allowed manual override at any point
in the sequence, thus it gave the operator a great deal of flexibility
for special operation while freeing him from tedious routine switching.
Heat was supplied to the system by placing the entire filter jig
in a vertically mounted tube furnace which was 17.8 cm inside diameter
by 38 cm long. In later tests it was discovered that internal
Thermacomb temperatures were not nearly as high as the exit gas
temperature because of air cooling and poor thermal conduction in the
filter. This necessitated the installation of an additional tube
i
furnace in the feed line between V.. and V». This preheated the inlet
air and eliminated the very poor temperature distribution in the filter
element. The cleaned hot air was cooled by water cooled jackets
around the exhaust piping. The cooled air flowed through a fixed
resistance and by measuring the temperature and pressure drop at the
resistance the flow rate was determined.
94
-------
Dust was dispersed and fed to the system using a "Modified" Harvard
dry dust dispenser. Some problems in feeding dust were experienced
because of the relatively high back pressures experienced. This caused
low suction at the ejector and poor dust feeding control. This problem
was eliminated by placing the rotating dust dispenser in a pressurized
containment. By maintaining the pressure at or slightly above the
filtration pressure, fairly uniform trouble free dust feeding was
achieved.
Photographs of the test assembly are presented in Figures 33 and 34.
It was anticipated that sealing the rigid honeycomb structures in
the support and top retaining plates would be a problem. However, it
was discovered that gaskets cut from fiber frax were ideal for sealing
and cushioning the element in the stainless steel plates. Sealing
problems were minimized by smoothing the edges of the filter element
with a ceramic adhesive called Ceramabond. This material could be used
to create an actual bond between the fiber frax and the element,
however, this was not found to be necessary.
95
-------
Fig. 33 - High temperature ceramic test facility
RM-68799
-------
Figure 34. High temperature ceramic filter assembly.
RM-68800
-------
Low Flow High Temperature Test Results
Completion of the fabrication of the low flow high temperature test
apparatus was followed by several shakedown runs at ambient conditions
to establish operating conditions and check equipment out.
A preliminary cold test was run with the 3M material at a flow rate
3 3
of 0.072 m /min and an average concentration of 3.17 gm/m . Although
the filter allowed no penetration, the initial pressure drop rose from
0.-8 kPa in the first cycle to about 6.0 kPa in cycles 1 to 6. After 6
cycles, the initial pressure drop remained relatively constant.
Subsequent tests at a lower flow rate (0.051 m /min) indicated that the
initial pressure drop of 3.6 kPa did not increase with time. During
these tests it was found that some problems existed when feeding dust in
the low flow case because the ejector could not be run at full pressure.
By putting in a by-pass which allowed some of the flow to be exhausted
directly, the pressure in the ejector could be maintained and a low flow
to the filter was also possible.
A series of hot runs were then undertaken, and during these
preliminary hot runs two unexpected phenomena were observed. The first
of these is demonstrated in Figure 35 which shows the filter performance
3 3
at a flow of about 0.023 normal m /min (0.085 actual m /min at 1090°K).
During this run no penetration of dust could be measured, and the initial
pressure drop remained constant at about 3.6 kPa. This pressure drop
was somewhat lower than expected based on previous experience. For a
given ambient flow and pressure drop, one might expect a 2.5 fold
increase in pressure drop due to viscosity effects and an additional
3.5 times increase due to the higher face velocity.
In the next series of tests another unexpected behavior was observed.
The element was brought up to temperature with just the air flowing and
with no dust being injected. The first cycle initial pressure drop was
relatively low - see Figure 36. The subsequent initial pressure drops
increased giving the appearance of very gradual blinding. However, if
98
-------
Curve 687709-B
1 I I I I
3M Co. Thermacomb ,
Constant Flow (Q) =. 085 m /min
Filter Area = 0.227m2
1 I I I l I I I I I I I I T
14 28 42 56 70 84 98
112 126 140 154 168 182 196 210 224 238 252 266 280 294 308 322
Time (Minutes)
Fig.35-Pressure drop vs time at (1090°K) no penetration
-------
the dust was shut off, and the unit was left hot with only the air on
for an hour or so, the filter could be cleaned by normal back pulse and
the low initial pressure drop regained. Figure 37 is the same filter
run the day after the run shown in Figure 36. The mass flow rates in
Figure 36 and 37 are the same, however, the test shown in Figure 37
was conducted at ambient temperatures. It can be noticed that the
problem does not seem to exist at ambient conditions. The possibility
of some chemical reaction between the limestone dust and the filter
material was considered, but it is difficult to propose a suitable
explanation for these observations. Representatives at 3M Company were
consulsted and they could not offer any explanation for this behavior.
This phenomena has been observed frequently throughout the entire test
program and a satisfactory explanation has not been established.
During these preliminary hot tests a series of runs were made to
explore the cleaning cycle. These tests indicate that over the range
of conditions tried the cleaning cycle was relatively insensitive to
length of pulse. Pulse times from 1 to 3 seconds were tried with
pressures from .77 atm to 1.3 atm.
It was also discovered that if the pulse down the dirty side was
administered second, it could lead to slightly poorer cleaning as
evidenced by higher initial pressure drops. It is believed that the
first pulse from the clean side knocks the cake loose from the wall and
the bulk of the collected material out of the channels. When the
second pulse is administered to the dirty side it merely compresses
the remaining dust against the wall rather than sweeping it out of the
filter.
The less than 100% collection efficiencies recorded in these tests
were due to improper sealing of the specimen in the jig. When corrected
no penetration was observed.
In attempt to discover the reason for the lower than expected
pressure drop it was decided to check the internal temperature of the
Thermacomb sample during a run.
100
-------
~~i—i—i—i—i—i—i—r
Constant Flow. 158 ml mm.
Filter Area = . 0227 m2
CAVEgm/m3
Temp. 1088°K
Filter Efficiency = 95.5%
Curve 687806-8
1 1—I—I 1 1—T
nj
0_
I I I
CAVE=2.185gm/m3 — *|
J I L
= 1.226 gm/m
J 1 L
4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76
Time (min)
84 88 156 160 164 168 172 176 180 184 188
Fig. 36- Hot thermacomb test
-------
Curve 687805-B
03
Q-
T i i i i — i — i — i
Constant Flow = . 0425 m3/min.
Filter Area = . 0227 m2
= gm/m3
3.0
Temperature -Ambient
Filter Efficiency -98.3%
ZO
1.4
• CAVE=2.57gm/m:
i — i — r~n — i — i — i — i — i — i — i — i — i — r
l i i i i i i ii i
CAVE=5.62gm/m-
i i i i i i i i i i i i i
3.0
2.0
1.4
8 16 24 32 40 48 56 64 72 80 88 96 104 112 120 128 136 144 152 160 168 176 184192 200
Time (min)
Fig.37 - Ambient temperature test (thermacomb)
-------
Previously it had been assumed that the operating temperature of
the filter was the same as the gas exiting the filter jig, however, when
this was checked by inserting a fine thermocouple into the Thermacomb
it was discovered that portions of the interior of the Thermacomb were
being maintained at about 525°C because of the materials low thermal
conductivity. The only apparent solution to the problem was to preheat
the air, however, because of constraints on the space available in the
existing test apparatus only a space of about 1 m was available. Several
electrical heating devices were tried, but because of the injected dust
fouling equipment little success was achieved with in line heaters.
Combustion of natural gas in the system seemed promising but it was
anticipated that relatively complicated safety equipment would be
required to prevent explosions. The scheme that was used, with reasonable
success, simply consisted of a 70 cm ceramic tube in a muffle furnace.
A concentric alumina tube was placed within the original tube to obtain
an increase in heat transfer area with a minimum of obstruction to the
dust. With this setup, gas temperatures exiting the preheating device
could be raised from ambient to about 825°K and temperatures within the
Thermacomb could be maintained at about 1QOO°K. These temperatures
were for flow rates of approximately 30 normal liters/min. Two test
runs were completed at these new conditions. The results of these runs
are shown in Figures 38 and 39. During the second run, Figure 39,
some problems were observed in the inlet sampler purge, so that the
inlet loading measured is probably not accurate.
It was noted that increase in pressure drop was still not as great
as would be expected. During heat-up, the mass flow rate of air was
maintained at about 37 gm/min. At ambient temperatures the pressure
drop across the filter was measured at about 1.25 kPa. At the final
filter temperature of about 1000°K the pressure drop had only doubled
to 2.5 kPa. It was noted that the filter was leaking during these
runs, but it seems unlikely that this apparently small leak could reduce
the pressure drop to the level observed. A more likely explanation,
but yet unconfirmed, is that the apparent pore size is an increasing
function of temperaturei
103
-------
Curve 6881 14-B
1—I—I—I—I—I—I—I—i—I—I
Constant Flow =. 099 m3/min
Filter Area =. 0227 m2
Cave=gm/m3
Temperature = 1023° K
Efficiency = 85%
1* 4
O-
2 -
I — i — | — i — | — i — | — i — | — i — | — i
3M Co. Thermacomb II
i — \ — r
H—-C =1.99gm/m3
= *• °7
i . i
i
i i i i i i i i i i i i i i i i i i i i i i i
48 16 24 32 40 48 56 64 72
80 88 96
Time (minutes)
104 112 120 128 136 144 152 160 164
Fig. 38- Hot thermacomb test
-------
Curve 688115-B
Constant Flow =. 096 m I mm
Filterarea =. 0227 m
3
4 8
16 24 32 40 48 56 64 72 80
(High Dust Loading)
- 5
i ' i ' i ' I
3M Co. Thermacomb II
Temperature = 993° K
Efficiency = 95%
= lL97gm/m3 J
C = 8.36gm/m
96 104 112 120 132
Fig. 39- Hot thermacomb test
-------
This specimen was removed from the test jig in order to cement
the corrugations that were leaking. During this process the entire
sample split along a barrier plane. It is suspected that this major
break was due to the rather rough treatment required to open the jig.
At this time it was observed that the filter material seemed more
fragile than it had been. This caused some concern since a commercial
application would require tens of thousands of hours of life without
deterioration. Subsequent samples did not appear to suffer this loss
of physical integrity, leading to the conclusion that it was most
likely the severe treatment the sample received when dismounting it
from the jig that caused the apparent weakness.
A new specimen was cut to size and mounted in the test jig. A
series of three runs at constant mass flow rate, but different temperatures
were carried to examine the effect of temperature on the filter
performance.
The results of the constant mass flow at different temperature
tests are shown in Figure 40 through 42. As can be noted from these
data, the initial steady pressure drop increases from about 2.2 kPa to
2.8 kPa to 3.0 kPa as the test temperature increases from 633°K to 793°K
to 983°K. One would expect an increase in pressure drop of about 1.41
times due to gas viscosity and an additional 1.24 times due to increased
face velocity. This should raise the 2.2 kPa figure at 633°K to about
3.86 kPa at 983°K, instead of the observed 3.0 kPa. Technical
representatives at 3M have agreed that the lower than expected pressure
drops may be due to pore expansion at higher temperatures. Since there
is apparently no penetration with the pores expanded, there may be some
possibility of increasing the porosity of the filter material without
loss of efficiency.
Having observed that both filtration efficiency and pressure drop
characteristics remained excellent at high temperature, the next effort
was directed toward better definition of the cleaning technique. It
had previously been established that the second pulse down the dirty side
106
-------
Curve 688721-B
4.0
ia
a.
Z 3.0
2.Q
I I I I I I I I I
\ \ I ! I I I 1 T
I I I I I I I I I I I I I I I I
3M CO. Thermacomb
Constant Flow = .061 ml mm. (actual)
= .028m I mm. (normal)
Filter Area =. 0227 m2
Temperature = 633°K
Efficiency = 100%
4.0
3.0
i i i i I I
Hot Thermacomb Test
i i i I i i i i i i i I i i
Pulsed Purge 137.9 kPa 3.0 sec.
i i i i i i i i i i i i i i i i
2.0
16 24 32 40 48 56 64
72 80 88
Time (min)
%
104 112 120 128 136 144 152
Fig.40 — Therma comb filter test -constant flow
-------
Curve 688722-B
ZO
3M CO. Thermacomb
Constant Flow =. 077 m /min. (actual) Temperature - 793°K
2
.028m /min. (normal) Efficiency = 99.97%
2
Filter Area =. 0227 m
= 5.16gm/m3
T I I I I I I I I I I l
Hot Thermacomb Test
Pulsed Purge -137.9 kPa 3.0 sec.
i i i i i i i i i i i ' i i i i I i l
8 16 24 32
40 48 56 64 72 80 88 96 104 112 120 128 136 144 152
Time (min.)
Fig. 41—Therma comb filter test -constant flow
-------
"I 1 1 T
T
Curve 688719-B
1 1 1 1
3
Constant Flow = . 095 m/min.
Filter Area -. 0227 m2
C =4.50gm/m3
ave
Temperature -983K
Efficiency = 99.6%
1—I—I—I—I—I—I—T
3M CO Thermacomb m Run 6
137.9 kPa 3.0 sec. 137.9 kPa 6/10 sec.
Hot Thermacomb Test Variation of Pulsed Purge kPa
.....
I i l
I _ I _ I — I — I — I — L
J I I I I I L_L
J ' i i '
4.0
3.0
2.0
4 8 12 16 20 24 28 32 36
44 48 52 56 60 64 68 72 76 80 84
Time (min.)
92 % 100 104 108 112 116 120 124 128 132 136 140
Fig. 42—Therma comb filter test -constant flow
-------
did little or no good, so the remaining parameters to be examined were
pulse duration and initial pulse pressure.
The initial run with a "standard" cleaning cycle is shown in
Figure 43. The cleaning cycle used was an initial pulse pressure of
103 kPa from a .385 £ vessel. The pressure virtually instantaneously
dropped to about 60 kPa for the remainder of the 0.6 sec pulse. As can
be seen in Figure 43 there is a small increase in the initial steady
pressure drop for this cleaning cycle. Some of this is because the
element was new and had not yet been "conditioned". Figure 44 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 0.6 sec pulse and the 34.5 kPa pulse dropped to somewhat
below 10 kPa. Figure 45 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. Figure 46 shows the data from another
run in which the pulse time was reduced to 2 sec.
Later tests indicated that a slightly higher initial pulse pressure
(138 kPa) gave somewhat more consistant cleaning than pulses of 103 kPa,
Apparently, 103 kPa pulses are marginal for this system, as occasionally
some deterioration in cleaning was observed. The performance of the
higher pressure cleaning cycle is shown in Figure 47. With the initial
pressure of 138 kPa a very stable cycle was maintained throughout
the test. The negligible effect of pulse duration was confirmed by this
test as there was no apparent difference between the 0.6 second pulse
and the 3 second pulse.
Later in the program, after the W. R. Grace & Company material had
been received and screened at ambient conditions, a hot test in the low
flow facility was carried out.
The results of this experiment are shown in Figure 48. As the
figure indicates this material performs at least as well as the 3M
Thermacomb material giving 100% efficiency and comparable pressure
110
-------
Curve 688Wt2-B
I I I I I I I I I I I I I I I I I I I I I
3A/1 Co. Thermacomb III 1st Run
I I 1 I I I 1 I I
S.
Constant Flow =. 094 m /min
Temperature - 973 K
Efficiency = 99.9%
Filter Area = .0227 m
2
-C = 2.74 gm/m3*]»C = 4.09 gm/ml
ave
1.0
Cave=7,85gm/m-
0 20 40 60
100 120 140 160 180 200 220 240 260 280
Minutes
300 320 340 360 380
Fig. 43- Hot thermacomb test (high dust loading)
-------
Curve 688Vt3-B
I
~i—i—i—i—i—i—i—i—i—i—i—i—i—r
-------
Curve 68857^-B
to
CL.
Constant Flow =. 093 m /min.
Filter Area = . 0227m2
Cave = 3.75gm/m3
Temperature =
Efficiency = 100%
r I
I I I
I I I
16 24 32 40 48 56 64 72 80 88 % 104 112 120 128 136 144 152 160
Minutes
Fig. 45-Hot thermacomb test
-------
Curve 688445-A
CXJ
1 1
3M Co. Thermacomb III
Constant Flow = . 092 m/min
2
Filter Area =. 0227 m
3
C = 3.16gm/m
Temperature = 953 K
Efficiency = 100%
34.5 kPa
2.0 sec
34.5 kPa
2.0 sec
69.0 kPa
2.0 sec
103.4 kPa
2.0 sec
103.4kPa
2.0 sec
i I I
103.4kPa
2.0 sec
1 I
40 48
Minutes
56
64
72
80
Fig. 46- Hot thermacomb test variation of pulsed purge kPa
-------
Curve 688720-8
(C
D-
2.0
I I I I I I I I I I I 1 I I I I I I I I I I
3M CO. Thermacomb En
Constant Flow =. 094 m3/min. Efficiency = 100%
2
Filter Area = . 0227 m
i i I i I I
I i i i I r
401 Temperature=%8K
i i i i i
i j i
i i i
6/10 Seconds 3.0 Seconds
Pulsed Puree -137.9kPa
Hot Thermacomb Test
i I I I I I I I I i I i i i I i I i i
4.0
3.0
2.0
1 I I I I
48 12 16 20 24 28 32 36 40 44 48 52 60 64 68 72 76 80 84 88 92 % 100 108 116 124 132 140 148 156
Time (min.)
Fig. 47-Thermacomb filter test-pulse times
-------
Curve 689823-B
CO
O_
Constant Flow = .085 m /min
Filter Area = .058 m2
3
W. R. Grace & Co. Comp. 49
Efficiency = 100%
Temperature = 887°K
32 40 48 56 64 72 80
Time (min)
Fig.48- Hot low flow test of W. R. Grace & Co. ceramic filter
-------
drop-flow characteristics. The cleaning cycle consisted of a single
138 kPa pulse from a 0.385 £ vessel for 1 second. The pulse was directed
to the clean side of the filter. A very stable, relatively low initial
pressure drop was maintained with this cycle even as the filter was
loaded with larger and larger amounts of dust. The W. R. Grace & Company
material in this configuration seems to be less fragile than the
Thermacomb and easier to handle.
At the time of this test, work on the high flow hot unit with the
large Thermacomb filters was underway so no further testing of this
material in the hot, low flow apparatus was possible.
117
-------
4.6 HIGH FLOW HIGH TEMPERATURE TESTS
The initial tests on the hot-low flow bench facility were promising
so efforts to carry out hot tests on a larger scale were initiated as
early as possible since the major goal of the Phase II effort was the
demonstration of hot gas filtration at a reasonable scale. Because of
constraints on time it was necessary to modify an existing facility for
the high flow tests.
The most appropriate available apparatus was a device previously
used in hot granular bed filtration studies. This facility had a flow
2
capacity of approximately 15 m /min at temperatures up to 1100°K and at
ambient pressure. The containment vessel h'a.d straight sides about 0.6m
in height and a diameter of l.lmoutside with a conical top and bottom.
The vessel and piping arrangement are shown in Figure 49. By proper
arrangement of the four, 10 cm slide valves flow could be directed
either down through the vessel when filtering or up through the bed
when backwashing the granular bed filter. The granular bed was
supported by a perforated distributor plate sealed between the vessel
flanges. Air was supplied to the system by a 1.0 kg/sec compressor
and this air was heated by an in-line material gas combustor. Dust
could be fed to the system in either of two ways. One method was to
feed a relatively coarse, easy flowing limestone powder to a Sturtevant
fluid energy mill via a hopper and controlled vibrating trough. The
fluid energy mill would grind the dust to the desired size and disperse
it into the air stream from the compressor. The second method was to
use a pre-ground test dust and inject and disperse it using a device
similar to the one used in the bench scale work, but larger. This
second option proved to be the easier to control and was used in all
testing.
The system was equipped with a combustible gas analyzer which
shut the apparatus down automatically if a combustible mixture was
sensed in the vessel. The vessel was also equipped with a vent pipe
and a 35 kPa rupture disk to prevent over pressure in the vessel.
118
-------
(D
IO
Ol
o
-------
Several modifications were required to enable this system to be used
for testing the ceramic filters. The major problem centered around
provisions for blowing back the filter elements since this required
pressures in excess of two atmospheres which was more than the vessel
or distributor plate could withstand at temperature. The solution to
this difficulty involved isolating each of the 15 cm cubes of Thermacomb
in a "hat" like container on a new support plate. This support plate
accommodated seven of these modules each enclosing a 15 cm cube of
Thermacomb. The arrangement of the Thermacomb filter and one hot module
is shown in Figure 50. This arrangement allowed high pressure pulses to
be directed to the clean side of the filter since only a small area of
the support plate was pressurized. It also allowed individual elements
to be pulsed independently. The major problem associated with this
configuration was the method of closing the hat while a pulse was being
administered. The simple sliding mechanism shown in Figure 50 was
adopted since it was anticipated that the closure would not need to be
tight, but that it would only be required to direct most of the pulse
, through the filter. Provisions for opening and closing the slides were
made using rigid mechanical linkage which could be operated manually
from the vessel exterior. Preliminary hot testing revealed that the
support plate warped upon heating causing the linkages to bind atid
become inoperative. This problem was overcome by welding an 0.32 cm
diameter wire rope to each slide. The ropes were extended out through
Conax fittings in access doors at opposite sides of the vessel. This
allowed individual slides to be pulled open or pulled closed regardless
of support plate warpage.
The blow back system simply consisted of a pressure vessel with the
same volume as a hat. A group of seven solenoid valves allowed the
vessel to be vented to any of the seven elements mounted on the support
plate. This arrangement only allowed the clean side of the filter to
be pulsed since bench scale tests indicated that the dirty side pulse
was not effective. In later tests it was necessary to try a pulse
from the dirty side in order to improve cleaning. A schematic of this
modification is shown in Figure 51.
120
-------
BBH
Figure 50.
6 inch ThermaComb
test assembly.
121
RM-69518
-------
Dwg. 16SOB01:;
Vessel Outer
Wall
FILTER BLOW BACK SYSTEM
Slide Valve
Hat
Filter
Base Plate
I
Check Valve
Pulse
Holding
Tank
J
Fig. 51— Schematic of pulsing modification
-------
Figure 52 ±s a schematic of the entire test facility. Figure 53
is a photograph showing the vessel, major piping, sampling ovens and
blow back pulse tank and lines. Figure 54 is a photograph of the filter
modules mounted on the support plate. One hat and blow back tube have
been removed to show the manner in which the Thermacomb elements were
mounted.
In summary, the operation of the high flow high temperature system
was planned to proceed in the manner described below. The filtering
mode began as the hot dust laden gas flowed from the natural gas
combustor down the left leg of vertical piping and into the bottom of
the vessel (refer to Figure 52). Flow was maintained in this direction
by closure of slide valve #2 while valve #1 was maintained in the open
position. The dirty gas flowed up through the support plate, was
filtered in the Thermacomb, exited the hat and finally escaped the
system through slide valve #1. The cleaning cycle began by opening
valve #2 and closing valve #1 which allowed the dirty gas to by-pass
the entire filter system. The slides on each hat were then pulled
closed and a back wash pulse was administered to each filter. The
dust that had been accumulated in the filter was knocked down through
the conical bottom of the vessel and was swept out of the system by
the by-pass stream. This procedure was then reversed and the filtering
mode of operation was re-entered.
123
-------
Owg. 16S8B02
Compressor #1
Pressure Relief Disc
2" Gas 1/2" Pressure
Cock Globe Regulator Solenoid
Valve
House
Regulator
Fig.52 - High temperature high flow test facility
-------
Figure 53. High temperature ceramic test facility.
125
RM-70469
-------
I
Figure 54. Detail of ceramic filter assembly.
-------
High Flow High Temperature Results
Upon completion of the mechanical alterations and required instal-
lations, a series of four preliminary runs were made in the high flow
unit which were as follows: (1) cool (95°C) and clean, (2) hot (610°C) and
clean to test mechanism for opening and closing the slides, (3) cool and
dirty to try dust injection system and pulse back cleaning system and
(4) a hot dirty test. During the first test, the pressure drop rose
linearly from 0 to 1.14 kPa as the flow increased from 0 to 3.14 m3/min.
There probably was some accumulation of dust from the piping in the
filter during the run. The second test demonstrated that the stainless
steel ropes work well. The third test was run at 3.64 m /min with dust
injected. The pressure drop rose from an initial 1.49 kPa to 2.07 kPa
over a period of 110 minute with a dust loading of 0.48 gm/m . The
filters were then back flushed with a pulse from a 8.9 liter vessel
charged to a pressure of 3.4 atm. The volume of this vessel is approxi-
mately the same as the filter hat. The initial steady pressure drop
after being flushed was 1.89 kPa, which then rose to 2.49 kPa over a
period of 65 minutes. The elements were then back flushed and the initial
steady pressure drop returned to 1.92 kPa.
After this test a larger dust injection system was installed in
an attempt to deliver more dust and shorten cycle times. A hot (615°C)
test was initiated after this modification. The initial pressure drop
was 5.03 kPa and increased to 5.22 kPa in 10 minutes with a dust loading
3 3
of 0.1 gm/m , and a flow of 9.15 m /min. There is some doubt about
the accuracy of the dust concentration measurement for this run as it
should have been considerably higher. When the filters were subjected
to the 3.4 atm pulse for cleaning the pressure drop only decreased
from 5.22 kPa to 4.98 kPa. There was an obvious leak so the test was
terminated at this point. Inspection of the filters indicated that they
were not packed in a sufficient amount of fiber frax to make a good
seal with the stainless steel support plate. It was also observed that
the bottom of the filters had accumulated a lot of dust which had not
been removed by the cleaning cycle being used.
127
-------
The problem of leaks at the support plate seal was overcome by
simply increasing the thickness of the fiber frax gasket in the support
plate. This allowed more fiber frax to be compressed into the edge
surfaces of the Thermacomb and formed a dust tight seal quite simply.
During a preliminary check of the sealing improvement it was noted
that the back flush cleaning did not return the initial steady pressure
drop to a constant value, rather it increased in slow but steady fashion
as shown in Figure 55. There were a couple of plausible explanations
for this observed behavior. The first is that most of the dust is never
expelled from the filte elements during the cleaning pulse, and the
residual dust in the filter increases continuously until sufficient
blockage to stop testing has occurred. A second possibility is that as
one filter is pulsed the expelled dust is blown into the surrounding
filters. A third possibility is that the dust is expelled from the
filter to the vessel walls and does not get blown out of the vessel into
the bypass stream. Then, when the device is returned from the cleaning
mode to the filtering mode, the once collected dust is blown back into
the filters.
A series of tests were run at compressor outlet temperature in an
attempt to discern which circumstance accounted for the observed
increase in initial steady pressure drop. In the first test the filters
were run until the initial steady pressure drop increased from about
3.5 kPa to about 5 kPa. One filter element was removed and it was
noted that the bottom surface had accumulated a considerable amount of
dust, and that the open area had been greatly reduced. Then three
elements were given back flushes of varying intensity and removed in
sequence. There was an observable increase in dust removal as the
initial pressure of the pulse was increased. It was noted that the
bottom of the element that was subjected to the highest pressure pulse,
appeared to be in a condition similar to the samples from the hot, low
flow unit where cleaning was successful. Figures 56 and 57 show how the
bottom of the filter surface changed with increasing pulse pressure over
the range from 0 to 552 kPa.
128
-------
Curve 689703-A
5.0
4.0
3M CO Thermacomb Bldg. 301
~~ii—i—i—i—i—i—i—i—i—r
Constant Flow = 4.60 m3/min.
Filter Area = 5.63 m
Cave=.193gm/m3
Temperature = 366°K
i—i—i—r
5.0
4.0
3.0
3.0
2.6
12
20 28 36
Time (min)
44
52
Fig. 55-High flow test starting with clean filters
60
129
-------
;;:
i'.Y.-':p".-
»•'•> »?Jh?$$sU?
. •: • • :.i*{»
*iSS$
Hi i
t. < j
1
'1
•• • v!'|'i; If
1 ' * » • . - *' *
• • ' • *
iiliiiii'iiiiiii&i
111
Wl
a
(0
a.
Figure 56. Pulse intensity tests.
130
-------
.;
i •
j
y
•. .
i • .
*
« .
• ;
i >
1 4
J i
i 5
• \
:
i
*
WWWfl
•
.
•
,
,
•
- •
. . . .
\l ::•'.': $
• . • ,'••••
: ' •' . ';•;;:
i . • > ,
• . •'••'•': • •'
• ' ',}
• . , • . • • •
• ' • '.'(
' .' :•:;.: ;;
. • .
•
• :.- :;. :
'''••• • ;
. V'.
:|
•
*
."
•
,
,
•
' •
• " '
. . .- . - .
. .
' ' • • .• • •:•••• • ' , •'
wiMltlll'lliHrt'WJW*
0)
CO
to
Q.
.• ;
I hi! •
'\ 1 II I
1
i( i A(||
;•
; : -
01
01
3
CL
Q.
-i.
CM
in
in
Figure 57. Pulse intensity tests.
131
RM-72208
-------
All of the elements were then vacuumed clean as was the conical wall
of the vessel. The elements were then reinstalled and tested again.
The initial steady pressure drop decreased from the previous value of
about 5 kPa to 3.0 kPa, but it rose steadily again with each cycle as is
shown in Figure 58. This indicated that immediate entrainment of
accumulated dust on the vessel wall was probably not the problem since
there had not been enough time to accumulate much dust in the vessel.
In subsequent tests designed to prevent the possibility of blowing dust
from the element being cleaned directly into the others, the element
being cleaned was given a quick pulse followed by flow of clean air to
all the other elements. This did not reduce the initial steady pressure
drop. The compressors were shut down to eliminate the by-pass flow
during cleaning, but this did not help either.
In spite of the relatively clean appearance, the problem was
apparently a failure to get the dust out of the element with the pulse
being used, and not a problem of reintrainment of collected dust.
At this point it was decided to attempt to get better cleaning
by one of two methods. The first attempt was to reduce the length
of the dirty corrugations by sawing the cubs in half. It was
hypothesized that due to the high aspect ratio there may have been
difficulty in forcing all the dirt from one channel out the relatively
small triangular opening. The second option to be investigated
consisted of administering a pulse directly down the dirty side to
sweep the bulk of material out of the filter prior to the clean side
pulse. The option to do this had been included in the low flow rig,
but since it had proved unnecessary in the small facility no easy
continuous way to give the dirty side pulse existed in the original
high flow unit design.
The first method was relatively easy to check. This was
accomplished by removing all of the elements and sawing them in half,
reducing the dirty side channel length from 15 cm to 7.5 cm. After
the cubes were cut in half, and the edges re-sealed with ceramic cement,
132
-------
Curve 69099^-B
5.0 -
re
Q.
4.0
3.6
I I I
3M Co. Thermacomb
Constant Flow = 2.7 m/min
Filter Area = 2.8 m
Temperature = 893°K
_L
12 20 28 36 44 52 60 68 76 84 92 100
Time (Minutes)
108
116
124
Fig. 66- Hot thermacomb test with cleaning pulse to clean and dirty sides
-------
they were replaced in the support plate and tested at compressor outlet
temperature. The initial pressure drop was quite low (about 2.5 kPa at
a face velocity of 1.64 m/min) but the initial pressure drop rose rapidly
indicating a continued failure to completely clean the Thermacomb
material (Figure 59). A subsequent test at a lower face velocity
(1.1 m/min) was conducted with similar results. It was noted that when
the elements were cut in half, the upper half of the filters were almost
completely plugged with dust that had not been removed by the cleaning
methods used.
Not having achieved satisfactory results by halving the element
length, it was decided to try the concept of administering a dirty side
pulse to the smaller elements. This idea was tested by the following
sequence:
(1) Load filters with dust
(2) Remove hats
(3) Remove top filter retaining plate and fiber frax
(4) Replace hat
(5) Pulse
(6) Remove hat
(7) Replace filter retaining plate and fiber frax
(8) Replace hat
(9) Continue filtering
This method was used on the initial cleaning cycle of a test run
at compressor outlet temperature and a relatively low face velocity of
0.93 m/min. Figure 60 summarizes the result of this run. There were
20 cycles during this test and the cleaning remained adequate as is
reflected in the constant initial pressure drop of about 3 kPa. No
penetration of dust could be detected throughout this test. Having
observed some success in a cold run, it was decided to continue at
temperature. The first of three such runs was carried out at 953°K
with a face velocity of 1.72 m/min. At these conditions the filter
performance was relatively good although a gradual increase in initial
pressure drop was observed. The data from this run are shown in
134
-------
Curve 690099B
3 M CO. Thermacomb
2.4
I i i I I I I i i i I i i i I I i i i i i I i i \
•3
Constant Flow = 4.60 m /min.
Filter Area =2.8m2
Cave=.055gm/m3
i i i
i i 1 I I
i l l I l l 1 1 1
1
4 12 20 28 36 44 52 60 68 76
1
84
1 1
l
l 1
92 100 108 116
Time (minutes)
Fig. 59-Initial test with Thermacomb cut in half
-------
Curve 690390-B
5.0
Constant Flow = (m /min)
Filter Area = 2.8m
Cave=(gm/m3)
Temperature = 366° K
3MCo. ThermaComb
4.0
3.0 -
Constant
C = .l03gm/m
Constant Flow = 2.61 m^/min
i i i
i i
i i i
4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 80 84 88 92 % 104 112 120 128 136 144 152 160 168 176 184 192 200 208 216
Time (minutes)
Fig. 60-Mafiual dirty side pulse test
-------
Figure 61. The subsequent run was merely a continuation of this test
to determine if the gradual increase in initial pressure drop would
persist. As Figure 62 indicates, it did so a decision was made shut
down and repeat the dirty side pulse procedure. Having done this the
initial pressure drop returned to the original value of about 7 kPa for
a couple of cycles, but the initial pressure drop increased rapidly
during subsequent cycles indicating progressive blockage of the filter.
These results are shown in Figure 63.
At this time the equipment which allowed repeated pulsing of the
dirty and clean sides of the filter was returned from the machine shop.
This modified blow back system (Figure 51) was designed for simultaneous
pulsing of both the clean and dirty sides of the filter but, by capping
off the appropriate line, either side could be pulsed alone or in
sequence. The system was checked out during a cold (compressor outlet
temperature) run. Mechanically, everything checked out, but the 150 min.
test (shown in Figure 64), showed a definite increase in initial pressure
drop indicating incomplete cleaning. Since cold tests are always suspect,
due to problems with compressor oil in the air, a hot run was conducted
next. During this run, the first few cycles started at fairly low
pressure drops, but this pressure drop increased rapidly from about
4.5 kPa to 6.2 kPa at a face velocity of 1.38 m/min and 973°K. The
pressure drop and cycle time then seemed to level out and remained
constant. These data are presented in Figure 65.
These preliminary tests with the modified simultaneous pulse were
followed by a series of experiments which were to investigate the
options offered by the modified blow back system. These runs were all
carried out at the relatively low face velocity of about 1 m/min.
Figure 66 shows the "base" case, which is a simultaneous pulsing of
both dirty and clean sides. Due to a problem with the outlet sampler
no overall efficiency was obtained for this run. It can be seen that
a fairly stable cycle has been established after four or five cycles.
The following day an identical run was attempted in order to verify
137
-------
-r,. '90388-A
3
Constant Flow = 4.81 m I mm
3 M Co. ThermaComb
Filter Area = 2. 8 m
2
Temperature = 953° K
Efficiency = 100%
to
a.
oo
J L
j i
j L
j L
J L I 1 1 1 \ L
4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 80 84 88 92
Time (minutes)
AT — wnt tpct
initial dirtv side oulse
-------
Curve 690389-A
3
Constant Flow = 4.65 m I mm
2
Filter Area - 2.8 m
3 M Co. ThermaComb
Temperature = 933° K
Efficiency = 100%
03
D-
1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1
1 1 1
8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 80 84
Time (minutes)
Fig. 62-Hot test with initial dirty side pulsed continued
-------
TO
D_
3 M Co Thermacomb
Constant Flow = 4.60 m /min
Filter Area = 2. 8 m
Temperature = 923° K
Efficiency = 93%
8 12 16 20 24 28 32 36
Time (minutes)
Fig. 63— Hot high ftow test after individual dirty and clean pulses
-------
Curve 690787-B
i I i I i i i I i i i i i i i i i i i i r i I I T I r T I i i r i I r i i i i
Constant Flew = 2.61 m /min.
2
Filter Area = 2.8m
Temperature = 366° K
3M CO THERMACOMB
4 12 20 28 36 44 52 60 68 76 84 92 100 108 116 124 132 140 148 156
Time, minutes
Fig. 64- Cold test with modified pulse
-------
Curve (,90788-B
T
T 1 T
ra
O.
a.
Constant Flow = 3.78m/min
Filter Area = 2.8m?
—I—I 1
3AA CO THERAAACOAAB
i—\—i—i—i—i—r
Ave
Temperature = 973° K
Efficiency =99%
4 8 12 16 20 24
28 32 36 40 44 48 52
Time, minutes
56 60 64 68 72 76 80 84
Fig. 65— Hot test with modified pulse
-------
Curve 689894-B
3M CO Thermacomb
4.8 -
Constant Flow = 4.60 m /min.
Filter Area = 5.63m2
Cave=.057gm/m3
Temperature=366°K
4.0
to
CL.
Q_
<
3.0
2.0
J I I I I L
J I L
16
24
32
40
48
56 64
Time (min.)
72
80
88
96
104
112
120
Fig.58 -Filtersand vessel precleaned
-------
the long term stability of the cycle established. During this run a
recurrent problem arose with the modified pulse system. Due to unequal
thermal expansion in the vertical line supplying the dirty side pulse,
and the retaining studs there was a tendency for the filter element to be
lifted on one side. When this occurred a leak would result. If it was
not anticipated fairly large pieces of the fiber frax gasket could be
blown out by the pulse and this would result in large leaks as
demonstrated by the data shown in Figure 67. Figure 68 shows the result
of a run after the major leak had been repaired. It is not certain how
much of the penetration measured was due to contamination of the clean
side of the vessel and piping by the previous run. After an initial
operating period of about 45 min. an apparently stable cycle had been
established with the initial pressure drop at about 7 kPa and rising to
7.5 kPa over a 20 min. period. The next two runs were conducted to
establish what effect variations from the combined side pulse would
have. Figure 69 shows a test in which a stable or slightly increasing
pressure drop cycle was established using the standard combined pulse.
At the point labeled 1 a standard combined pulse was immediately followed
by a pulse that was forced entirely down the dirty side. This additional
pulse recovered a full 0.4 kPa in pressure drop, lowering the initial
pressure drop to a point below all but the first cycle. During the next
cycle a combined pulse returned the pressure drop from 7.5 kPa to only
about 7.0 kPa, but when this was followed by a pulse directed entirely
down the clean side the pressure drop returned to its previous low value
of about 6.2 kPa. A subsequent test was performed to examine this
phenomena and the results are presented in Figure 70. As is indicated
by these data, a repeatable decrease in initial pressure drop can be
realized by virtually any combination of the combined, clean or dirty
pulse options. A stable cycle between pressure drops 6.5 and 7.5 kPa
was maintained throughout the test with no evidence of the usual gradual
blocking of the filter.
144
-------
Curve 690993-B
i—i—i—i—i—i—r~
3MCo. Thermacomb
1 1 1 1 1 T
1 T
Constant Flow = 2.6 m /mm
Filter Area = 2. 8 m2
Temperature = 883°K
Efficiency 57% (due to blown gasket)
J L
J I L
16 32 48 64
112 128 144 160 176
Time (Minutes)
192 208 224 240 256
Fig. 67- Hot thermacomb test with cleaning pulse to clean and dirty sides
-------
Curve 691198-A
CO
O.
1 1 1 1 1 1
3 AA Co. Thermacomb
3
Constant Flow 2.74m /min
Filter Area = 2. 8 m
Temperature = 903°K
Efficiency = 88%
4 8 12 16 20 24 28 32 36 40 44 48 52 56 60 64 68 72 76 80 84
Time (Minutes)
Fig. 68~ Hot thermacomb test with combined pulse
-------
Curve 691199-B
7.8
0_
^ 7.0
Q_
6.0
T
T
T
I
Constant Flow = 2.51 m I mm
Filter Area = 2.8 m
Cave = .12gm/m3
Temperature = 893°K
Efficiency = 92%
3 M Co Thermacomb
Pulse - 5.4atm
1 - Combined, Dirty Side
2 - Combined, Clean Side
J_
_L
16 24 32 40 48 56 64 72
Time (Minutes)
80
96
104
112
116 120
Fig. 69- Hot thermacomb test with pulse variations
-------
Curve fc91200-B
Q.
<
6.0
I 11^
3 M Co. Thermacomb
Constant Flow = 2.57 m /min
Filter Area = 2.8 m
Temperature = 913°K
I I I
Pulse - 5.4attn
1 - Combined, Dirty
2 - Combined, Dirty
3 - Combined, Clean to 4
4 - Clean, Combined
5 - Combined, Clean
6 - Clean, Combined
7 - Combined, Clean
1 I
1
8
1
16
1 1
24 32
1
40
1.
48
1 1 1
56 64 11
Time (Minutes)
1 1
80 88
1 1
96 100
1
108 11
Fig. 70- Hot thermacomb test with pulse variation
-------
A final test on the high flow unit with the 3M Thermacomb was
performed using only a clean side pulse initially. At a face velocity
of 0.96 m/min the cycle was somewhat irregular and appeared to be
climbing. On the tenth cycle the clean side pulse was followed by a
combination pulse to both sides. As is indicated in Figure 71, this
reduced the pressure drop as had been observed in previous tests. A
continuation of this test was prevented by a complete failure in one
of the element's seals which resulted in gross leakage.
Very late in the program 7 cubes of the W. R. Grace & Company
material were obtained. There was only time to install them and
perform preliminary tests on them using the original clean side pulse
only on the full 15 cm thick elements. The results of the three runs
accomplished are summarized in Figures 72 and 73. It can be seen from
these data that there was a very gradual increase in the initial
pressure drop over the test period, however, it is believed that a
combination of pulses would be at least as effective in preventing
residual pressure drop buildup in this material as it was for the
Thermacomb.
149
-------
Curve 691313-B
7.5
7.0
< 6.0
5.0
J_
_L
J_
_L
Run-A
j J_
•h-
_L
_L
_L
_L
_L
Constant Flow = m /min
2
Filter Area = m
Cgve= gm/m3
Temperature = Degrees K
Efficiency = %
_L
Run-B-
I
_L
1 1
Run-A Run-B
2.69
2.8
81
2.71
2.8
893
50
8 16 24 32 40 48 56 64 72 80 88
96 104 112
Time (Minutes)
120 128 136 144 152 160 168 176 184 192
Fig. 71-Final hot test with thermacomb
-------
Curve 691311-B
5.0
4.0
Run A
Constant Flow = m /min 4.51
Filter Area = m2 7.9
Caye= " gm/m3 .11
Temperature = °K 948
Efficiency % 98
Run A
B
4.40
7.'9
.44
923
99
-*•««-
W.R. Grace
Run B
_L
_L
_L
_L
_L
_L
J_
JL
16 24 32 40 48 56 64
72
88 96 104 112 120 128 136 144 152 160 168 176 184 192
Time (Minutes)
Fig. 72-Initial hot test with 15 cm cube of W. R. Grace material
-------
Curve 69ni2-B
~1 1 1 1 r—
3
Constant Flow = 6.52 m /min
Filter Area = 7.9 m2
~ 7.0
Q-
6.0
J L
n r
Temperature = 913°K
Efficiency = 87%
J I I L
T——i 1 1 1 r
W.R. Grace
j i
i i i i i i i i i i
8 16 24 32 40 48 56 64 72 80 88
96 104 112
Time (Minutes)
120 128 136 144 152 160 168 176 184 192 200
Fig. 73-Hot test with 15cm cube of W.R. Grace material
-------
4.7 COMMERCIAL AND ECONOMIC ASSESSMENT
Objectives. One of the major objectives of this task was to use
the experimental data available from the high and low flow tests to
estimate operating parameters for a commercial application. This data
on face velocity, pressure drop, physical dimensions of filter elements,
efficiency and cleaning requirements would be used to generate a
conceptual design of a module of ceramic filter apparatus suitable for
large scale application. Finally, this conceptual design was to be
costed out and compared with other high temperature fine particulate
removal devices.
Discussion. It has become apparent that the experimental work
with the high and low test units have not advanced to the stage where
all of the operating parameters necessary for a commercial design are
at hand. This is especially true with respect to filter element
fabrication alternatives and cleaning technique. Other problems such
as the nature of the ash in actual use have not been addressed. Recent
(4)
experiments with granular bed filters at the Exxon mini plant have
shown that this can be an important consideration. In spite of
uncertainties, it is possible to proceed with at least a rough
conceptual design, with current estimates of performance.
The basis for this design was taken to be a 150 MWe low BTU coal
gasification plant. The coal gasification application was chosen since
it is generally considered a more difficult task to clean hot, reducing
fuel gases than cleaning hot flue gas from application such as a
pressurized fluidized bed combustion. The size chosen is large enough
to be representative of a commercial unit, as most equipment for a
larger facility would be composed of modules, and data for the 150 MWe
plant were available. The important plant parameters are summarized as
follows:
Plant Size 150 MWe
Temperature 870°C
Pressure 10 atm
3
Flow 1900 actual m /min
153
-------
Dust to Last Stage
3
Loading 0.5 gm/m
Size 100% < lOy
Using these inputs, the ceramic filter module shown in Figure 74
has been offered as a rough conceptual design for a commercial scale
application.
Operation of the unit is as follows: The dirty gas enters the
four modules at the bottom of the pressure vessel through the inlet
manifold. The dust laden gas then flows up into the square enclosures
that support the filter elements and out through the filter elements
into the blow back enclosure, and out the cleaned gas manifold.
Periodically, as the pressure drop across a module rises, the module is
isolated from the outlet manifold by a slide valve. A pulse of cleaned
gas is then introduced on the clean side of the filter elements knocking
the accumulated dust out of the elements and down into the hopper where
it is periodically lock hoppered out of the system. It is anticipated
that one module can be cleaned while the other three modules remain in
the filtering mode. Assumptions that have been made regarding the
filter elements and their performance are:
Face Velocity 1-1.5 m/min
Max. Filter Thickness 15 cm
Overall Efficiency 99+ %
Pressure Drop 5-10 kPa
Cleaning Cycle Time 60-90 min.
Cleaning Pulse Gas Requirement 0.2 kg/min @ 12 atm
The estimate for face velocity is based on the latest data from the
hot, high flow unit where the most stable operation was achieved at
similar velocities. Earlier operation with the low flow unit has
indicated that good performance can be achieved at velocities 3 or 4
times the 1-1.5 m/min range, however, to be conservative the lower
range has been used in sizing the unit. The assumption of a maximum
filter element thickness of 15 cm is based on the increase in cleaning
154
-------
Plan View
Angle to Make Rigid Walls
6.7m
Cleaned
Gas Exit
Dug. 2613C48
- Slide Valves to Isolate For Blow Back
Blow Back Inlet
Elevation
Refractory
Lining
Ceramic Filter
Elements On
Mounting Plates
Enclosure For
Blow Back
Inlet
(1900 itrVmin)
Fig. 74 - Sketch of possible arrangement for commercial use of ceramic filters
-------
problems observed in going from the low flow unit to the high flow unit.
Improved corrugation sizing and filter porosity along with better pulsing
techniques should increase this maximum thickness. The estimate of overall
efficiency and pressure drop are based on experimental observations from
both high and low flow tests. The estimate of cleaning cycle time depends
on the expected concentration of dust to the filter. It has been
assumed that two stages of mechanical collectors have reduced the projected
loading from the bed of about 35 gm/m to 0.5 gm/m at the filter with
only the sub lOy fraction remaining. The cleaning cycle time data varied
a great deal between the high and low flow units. The high flow unit had
3
typical cycle times of 10-20 min. with loadings of about 0.1 gm/m and
face velocity of 1-2 m/min, but the low flow unit had similar cycle times
with loading of 2-5 gm/m and velocities of 2.4 m/min. The low flow unit
typically operated between 2 and 5 kPa while operation with the high
flow unit was usually in the 6-9 kPa range. These observations may be
explained by assuming that some of the high flow filter channels were
not cleaned and were unavilable filter area. For estimation of the
cycle time, 'it has been assumed that the filter performance will be
o
similar to the low flow unit. Using a 10 min. cycle with 2 gm/m
loading at a face velocity of 2 m/min would lead to an estimate of a
40 min. cycle time for the commercial application. If the other end of
3
the range is used, 20 min. cycle, 5 gm/m and 4 m/min, a cycle time of
400 min. is obtained. The cycle time chosen (60-90 min.) corresponds
to a conservative estimate. The cleaned gas requirement for pulsing
the filters was scaled up to maintain the same ratio of flow to filter
area.
The vessel size is on the order of 6.7 m tall and 5.18 m diameter
with an elipsoidal head, 3 m straight sides and a hemispherical bottom
rather than the conical shape indicated in the sketch. Metal thickness
for the head and sides would be about 5 cm and the bottom would be
about one-half this. This vessel would also require about 15 cm of
refractory lining in order to maintain suitable metal temperatures.
156
-------
The valves required for the pulsing sequence do not need to be tight
seal valves since they are only required to direct most of the pulse to
the intended module. The valves should not, therefore, be a high cost
item. The filter element modules are shown in a square configuration
to minimize filter element mounting and sealing problems. It may be
feasible and desirable to consider cylindrical mounting structures in
the future. A major cost item in the ceramic filter system is the
mounting plate-pulse containment structure. Anticipating possible
corrosion problems in the reducing fuel gas atmosphere it was assumed
that the material chosen for these support plates will have to be
something like Haynes 188, which makes these components very costly.
For oxidizing atmospheres, such as for fluidized bed combustion, the
cost of this material will be substantially less.
The device chosen to compare the performance and cost with is a
granular bed filter since this is about the only other high temperature
fine particle device currently available. Ducon has done considerable
development work with granular bed filters and is currently testing a
device for pressurized fluidized bed combustion application. Since
some data on the cost of elements for the Ducon filter are available,
this particular device has been chosen for comparison with the
ceramic filter.
The process assumptions used in the ceramic filter system were
enumerated previously, while those assumed for the granular bed filter
system are described below:
Filter Type Ducon cylindrical
Face Velocity 15.24 m/min
2
Elements/Unit Plan Area 4 elements/m
Overall Efficiency 98+ %
Pressure Drop 20-35 kPa
In this estimate it has been assumed that the maximum pressure
vessel diameter that is practical is about 7.5m diameter. If more plan
area is required multiple modular units would be constructed.
157
-------
On the basis of these assumptions the granular bed filter system
3
would be composed of two modules, each handling about 950 m /min of low
Btu fuel gas from the 150 MWe coal gasification plant. Each module would
be a 7.6 m diameter refractory lined pressure vessel with an ellipsoidal
head and hemispherical bottom and a 3.05 m straight section. Each vessel
would house about 170 elements fabricated from Haynes 188.
Since the Ducon granular bed filter is in a much more advanced stage
of development, it is likely that this device could be costed out in
greater detail than the proposed ceramic filter system, however, as the
EGAS study has shown wide variations in the estimated cost of this
equipment exist. The approach taken here has been to estimate the costs
of the major components of each system using a consistent basis.
Since the systems have several similar features, which have been
costed at the same degree of detail, the results should be comparable
if not correct on an absolute basis. The results given will be adequate
as order of magnitude equipment costs and more accurate as relative
costs.
The results of the cost comparison are presented in Table 19.
Pressure vessel costs have been estimated using a factor of 4.63$/kg
for large field erected vessels. Scrap and nozzles were estimated at
10%. Refractory lining costs were based on suppliers previous quotes
on similar applications. The Ducon filter element cost was based on
an estimate of 400$/element for production quantities. This estimate
from Ducon was based on stainless steel construction and has been
scaled up to 1400$/element to reflect the additional materials and
labor costs associated with using Haynes 188 at 17.62$/kg. The
ceramic filter element costs are small as estimated by the 3M Company
3
at 3531$/m . However, the materials costs for the retaining plate and
blow back enclosure are large since it has been assumed that these
would have to be fabricated from a material like Haynes 188. The valving
and blow back equipment for the systems represent a relatively small
portion of the total cost of the equipment. It has been assumed that
158
-------
high pressure steam will be used for the granular bed filter blow back
system rather than cleaned fuel gas. The ceramic filter system has
assumed that the cleaning gas will be taken from the filter exit stream
and compressed from 10 to 12 atm and held in a 7.25 m3 vessel for
pulsing.
These estimates indicate that the granular bed filter could cost
about three times as much as the ceramic filter system. The largest
cost in the granular bed filter system arises from pressure vessel
construction. It may be feasible to crowd the elements into tighter
spacing and to run them at higher face velocities and reduce the cost of
the pressure vessel. Because the ceramic filter system has a high
surface to volume ratio the containment vessel costs are relatively
lower. The cost of the high alloy retaining plates and pulse containment
for the ceramic filter system would be reduced substantially if more
conventional materials of construction could be used. This could be an
important consideration in using the ceramic filter system in an
oxidizing atmosphere such as one encounters in fluidized bed combustion.
In these cases the amount of gas to be cleaned per unit mass of coal
burned increases by a factor of about 3 over gasification, but more
conventional materials can be used for the retainer, so the costs of
this stage of particulate removal equipment will not increase in
proportion to the gas flow.
On the basis of this admittedly crude cost comparison it appears
that the ceramic filter system is worthy further examination.
159
-------
Table 20
Ceramic Filter and Granular Bed Filter Costs
Granular Ceramic
Bed Filter Filter
Pressure Vessel(s)
Steel costs 867,514 162,171
Refractory lining 124,404 33,561
Filter Element 470,400 25,200
Retainer — 247,849
Back Flush Equipment
Valves 45,000 40,000
Piping 15,000 15,000
Installation 20,000 5,000
Booster compressor . — 8,880
Pulse holding tank — 23,007
$1,542,318 $560,668
160
-------
REFERENCES
1. T. R. Wilshaw, J. Am. Cer. Soc., 51, (1968), p. 111.
2. Poe, G. G. , R. M. Evans, W. S. Bonnett, and L. R. Waterland,
"Evaluation of Ceramic Membrane Filters as a Control for Fine
Particulate," Final Report, EPA Contract 68-02-1313.
3. Ciliberti, D. F. and B. W. Lancaster, Modified "Harvard" Dry Dust
Disperser, Rev. Sci. Instrum., Vol. 46, No. 7, July 1975.
4. Bertrand, R. R., et al., "A Regenerative Limestone Process for
Fluidized Bed Coal Combustion and Desulfurization," Monthly
Reports 80-88, EPA Contract 68-02-1312.
5. Beecher, D. T. , et al., "Energy Conversion Alternatives Studies,"
Westinghouse Phase II Final Report Prepared for NASA Lewis
Research Center, Contract NAS 3-19407 (1976).
161
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing!
2.
1. REPORT NO.
EPA-600/2-77-207
4. TITLE AND SUBTITLE
High Temperature Participate Control with Ceramic
Filters
3. RECIPIENT'S ACCESSION NO.
5. REPORT DATE
October 1977
6. PERFORMING ORGANIZATION CODE
7. AUTHORtS)
D. F. Ciliberti
8. PERFORMING ORGANIZATION REPORT NO.
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Westinghouse Research Laboratory
Pittsburgh, Pennsylvania 15235
10. PROGRAM ELEMENT NO.
1AB012; ROAP 21ADL-029
11. CONTRACT/GRANT NO.
68-02-1887
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT ANC
Final; 7/75-8/77
ND PERIOD COVERED
14. SPONSORING AGENCY CODE
EPA/600/13
15.SUPPLEMENTARY NOTES IERL_RTp project officer for this report is Dennis C. Drehmel,
Mail Drop 61, 918/541-2925.
16. ABSTRACT
repOrt gives results of an assessment of using ceramic materials as
filters for fine particulate removal at high temperatures . The program was in two
phases. Phase I, directed toward the development of a porous alumina membrane
filter, had limited success because of the fragility of the membranes formed, and the
difficulty in controlling the pore size distribution of the filters. The major objective
of Phase II, concentrating on screening other available materials , was to identify
materials with good filtration potential, select one or two of the most promising, and
(as rapidly as possible) demonstrate them as hot gas fine particle filters in a several
hundred cu m/hr hot test. Initial screening indicated that the most promising was a
thin-walled, ceramic, cross-flow monolith, originally produced as a catalyst sup-
port for automotive exhaust systems. Screening tests indicated the possibility of
virtually 100% removal of even submicron limestone test dust at face velocities and
pressure drops not dissimilar from those typical of fabric filtration. Later bench
scale tests at around 1000 K confirmed the material's ability to perform well at high
temperatures. Final testing, at a larger facility where flows of 4. 8 cu m/min at 950
K were achieved, indicated that this ceramic configuration offers great potential as a
hot gas filter.
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
Air Pollution
Dust
Ceramics
Filtration
Dust Filters
High Temperature Tests
Thermal Resistance
Aluminum Oxide
'1. DISTRIBUTION STATEMENT
Unlimited
b.lDENTIFIERS/OPEN ENDED TERMS
Air Pollution Control
Stationary Sources
Particulate
Hot Gases
I'J. SECURITY CLASS i'l his He port I
Jnclasified
70. SECURITY CLfrc,r (This page)
Unclassified
:. COSATl Held/Group
13B
11G
11B
07D
13K
14B
20M
07B
21. NO. OTTAGES
172
EPA Form 2220-1 (9-73)
162
------- |