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

-------