xvEPA
United States
Environmental Protection
Agency
Office of Water
& Waste Management
Washington, DC 20460
SW 175C.4
June 1979
Solid Waste
A Technical and
Economic Evaluation
Of the Project
In Baltimore, Maryland
Volume IV
-------
Prepublication issue for EPA libraries
and State Solid Waste Management Agencies
A TECHNICAL AND ECONOMIC EVALUATION OF THE
PROJECT IN BALTIMORE, MARYLAND
Volume IV
This report (SWl75c) describes work performed
for the Office of Solid Waste under contract no. 68-01-4359
and is reproduced in four volumes as received from the contractor.
The findings should be attributed to the contractor
and not to the Office of Solid Waste.
Volume I of this report is the executive summary
and is available from the Office of Solid Waste (order no. 719).
Volumes II, III, and IV, will be available from the
National Technical Information Service
U.S. Department of Commerce
Springfield, VA 22161
U.S. ENVIRONMENTAL PROTECTION AGENCY
1979
-------
This report was prepared by Systems Technology Corporation, Xenia, Ohio,
under Contract No. 68-01-4359.
Publication does not signify that the contents necessarily reflect the views
and policies of the U.S. Environmental Protection Agency, nor does mention
of commercial products constitute endorsement by the U.S. Government.
An environmental protection publication (SWl75c) in the solid waste
management series.
-------
PREFACE
This report is a complete technical, economic, and environmental
evaluation of the Landgard® Demonstration Plant at Baltimore, Maryland.
Because of its bulk and to serve a twofold purpose, the report is presented
in four volumes: an executive summary, the report proper, an analysis of the
problems, and the appendices. Intended particularly for resource recovery
planners and administrators, the executive summary briefly and succinctly
describes the Landgard® concept and Baltimore application for the state-of-
the-art advancement in the processing of municipal mixed solid waste. In
addition, it presents an introductory problem analysis of most of the major
innovations that proved ineffective, caused serious shutdowns, and required
redesign or abandonment. As the second, third, and fourth volumes are
detailed in-depth accounts of the evaluation, they were prepared primarily
for the designer. Of the four volumes, only the executive summary has been
prepared for wide distribution in a paper copy format. The second, third,
and fourth volumes are reproduced on microfiche, which is readily available
through NTIS.
-------
ABSTRACT
One of the first efforts in this country to demonstrate solid waste
resource recovery technology was the Baltimore Landgard® project which was
a joint venture between the City of Baltimore, the U.S. Environmental
Protection Agency (EPA), the Maryland Environmental Service, and Monsanto
EnviroChem. The Baltimore plant was designed and built by Monsanto
EnviroChem to thermally process (pyrolyze) 907 Mg (1000 tons) per day of
mixed municipal solid waste, convert it to energy (in the form of steam), and
recover magnetic metals and glassy aggregate. Although the plant has never
been fully operational in its original design configuration, considerable
knowledge has been'gained from it concerning resource recovery from municipal
solid waste. The numerous equipment breakdowns and the inability of the
plant to comply with air pollution standards accounted for the major diffi-
culties encountered during the project. Major equipment problems were
encountered with the storage and recovery unit, the refractory in the thermal
processing vessels, the main induced-draft fan, the residue discharge drag
conveyor, and the slag discharge screw conveyor. Despite the fact that the
designer recommended converting the plant to a conventional incinerator,
plant performance has been sufficiently encouraging to warrant continued
investment and operation by the City of Baltimore. One of the primary
reasons for this attitude by the City is that the rotary processing kiln has
been demonstrated to be an excellent primary reaction vessel. Although the
present plant is not environmentally acceptable because of high particulate
emissions, this problem will be resolved by the installation of two electro-
static precipitators.
The thermal efficiency of the plant was determined to be approximately
56 percent for an average feed rate of 454 kg per minute (30 tph). The plant
has a capital cost of approximately $22 million, an annual operating and
maintenance cost of $3 million, and an annual steam revenue of $1 million.
The net operating cost, based on historical operating data, is $64.10 per Mg
($58.20 per ton) of refuse processed. However, if the annual throughput of
67,000 Mg (74,000 tons) could be substantially increased to 270,000 Mg
(300,000 tons), operating cost could be reduced to $7.80 per Mg ($7.10 per
ton) of refuse processed.
This report is submitted in fulfillment of Contract No. 68-01-4359 by
Systems Technology Corporation (SYSTECH) under the sponsorship of the
U.S. Environmental Protection Agency. This report covers a period from
October 1, 1975 to April 30, 1978.
ii
-------
CONTENTS
Preface 1
Abstract ii
Figures iv
Tables vi
List of Unit Conversions vii
Acknowledgment viii
1. Introduction 1
2. Refractory 3
Kiln Refractory 3
Gas Purifier Refractory 5
Duct Refractory 40
3. Slagging 43
Slag Hole Plugging 43
Boiler Tube Slagging 48
4. Conveyors 55
Storage Pit and Shredder Feed Conveyors 55
Screw Conveyors 56
Residue Quench Tank Drag Conveyor 57
5. Ram Feeders 59
Ram Jams 59
Snout Failure 60
Hydraulic System 61
6. Storage and Recovery Unit 63
7. Residue Separation Module 66
8. System Pressure Oscillations 68
9. Kiln Process Control 72
10. 02/C02 Analyzer System 74
11. Induced Draft Fan 77
12. Scrubber Erosion and Corrosion 80
Scrubber Pumps 80
Scrubber Lances and Nozzles 81
13. Summary ..... 82
iii
-------
FIGURES
Number Page
1 Board-mounted assembly of all slag-refractory block
specimens tested for slag penetration 7
2 SEM photographs of Alusite-D block specimen tested
at 1425°C 9
3 SEM photographs of Altex-85 block specimen tested
at 1375°C 12
4 SEM photographs of Altex-85 block specimen tested
at 1425°C 13
5 SEM photographs of DV-38 block specimen tested at 1375°C ... 16
6 SEM photographs of DV-38 block specimen tested at 1425°C ... 17
7 SEM photographs of Lofero-HM block specimen tested
at 1375°C 21
8 SEM photographs of Lofero-HM block specimen tested
at 1425°C 22
9 SEM photographs of Nucon-60 block specimen tested at
1375°C 25
10 SEM photograph of Nucon-60 block specimen tested at
J425°C 26
11 SEM photographs of Oxibak-H block specimen tested
at 1375°C ' 29
12 SEM photographs of Oxibak-H block specimen tested
at 1425°C 30
13 SEM photographs of Korundal-XD block specimen tested
at 1375°C 34
iv
-------
FIGURES (contd)
Number gage
14 SEM photograph of Refrax-50A block specimen tested
at 1375°C 37
15 A plot of slag viscosity based upon temperature and
chemical composition 45
16 Slag temperature characteristics for satisfactory
slag tapping 46
17 Slag viscosity based upon chemical composition 47
18 Scanning electron microscope pictures of fly ash 52
19 Optical microscope pictures of fly ash 53
20 Data trace of sinusoidal pressure pulsation in the kiln ... 68
21 Data trace of random frequency pressure oscillations .... 69
22 Data trace of random frequency pressure oscillations .... 69
23 Data trace of dangerous pressure oscillations 69
24 Data traces of pressure at various locations 71
-------
TABLES
Number Page
1 Typical Slag Composition 5
2 Alusite-D Slag Penetration Test at 1375°C 10
3 Alusite-D Slag Penetration Test at 1425°C 11
4 Altex-85 Slag Penetration Test at 1375°C 14
5 Altex-85 Slag Penetration Test at 1425°C 15
6 DV-38 Slag Penetration Test at 1375°C 18
7 DV-38 Slag Penetration Test at 1425°C 19
8 Lofero Slag Penetration Test at 1375°C 23
9 Lofero Slag Penetration Test at 1425°C 24
10 Nucon-60 Slag Penetration Test at 1375°C 27
11 Nucon-60 Slag Penetration Test at 1425°C 28
12 Oxibak-H Slag Penetration Test at 1375°C 31
13 Oxibak-H Slag Penetration Test at 1425°C 32
14 Korkundal-XD Slag Penetration Test at 1375°C 35
15 Korkundal-XD Slag Penetration Test at 1425°C 36
16 Refractory Chemical Composition 36
17 Refrax-50A Slag Penetration Test at 1375°C 38
vi
-------
LIST OF UNIT CONVERSIONS
DESCRIPTION
SI
ENGLISH EQUIVALENTS
*
LENGTH
AREA
-------
ACKNOWLEDGMENT
This evaluation program was performed under EPA Contract No. 68-01-4359,
"Technical and Economic Evaluation of the EPA Demonstration Resource Recovery
Project in Baltimore, Maryland."
The EPA Project Officer was Mr. David B. Sussman of the Office of Solid
Waste, Washington, D.C.
Testing was carried out at the demonstration facility in Baltimore,
Maryland with the cooperation of the City plant staff and the Monsanto on-
site engineering staff. The contribution of both of these groups has been
greatly appreciated. The contribution of Dr. H. G. Rigo and Richard Eckels,
along with other staff members, is also acknowledged.
Systems Technology Corporation would like to express its gratitude to
the above named individuals and all others associated with this evaluation.
viii
-------
SECTION 1
INTRODUCTION
This last volume of the four-volume report on the Baltimore Pyrolysis
demonstration project documents the investigation of the problems encountered
during demonstration start-up and operation. The investigation was limited
to critical problems; that is, failures, malfunctions, or discrepancies which
either directly or indirectly caused operation shutdown. The reason for this
limitation was twofold: first, there were too many minor problems to evaluate;
and second, plant reliability, that is, continual operation, was the prime
criterion in evaluating the plant's performance.
The problem investigation consisted of describing modifications imple-
mented to resolve problems, of analyzing the problems, and of offering
solutions to the unresolved problems. In addition, each of the critical
problems was classified under one of the following three categories: (1) unique
to the Landgard process, (2) unique to the Baltimore facility, and (3) typical
of solid waste processing in general. To distinguish these categories, the
"Landgard process" was defined to include the specific process of pyrolysis
or starved-air incineration with subsequent combustion of the product gas and
the basic equipment items to perform the process; the "Baltimore facility"
was defined to include the construction materials and practices, the operating
procedures, and the module and component arrangements peculiar to the Balti-
more plant; and "general solid waste processing" was defined to include the
equipment and operations common to a number, of solid waste processing plants.
Each of the following sections addresses the problems related to a
particular type of equipment, operation, or phenomenon. For each item in
each section, the problem investigation consists generally of three parts.
As a general introduction to the problem, the first part explains why the
failure led to a plant shutdown along with a history of the failure and the
modifications implemented to resolve the problem. The second part analyzes
the problem and proposes solutions and recommendations. Finally, the third
part classifies the problem according to the foregoing three categories.
A final section summarizes the analyses and recommendations and reviews
the component requirements as related to the total system. This relationship
is critical since some of the downtime was due to the incompatibility of the
components, processes, and operators. While no single design was necessarily .
incorrect for the component itself, some components were designed without due
consideration of the total system. Moreover, some components were incorrectly
-------
designed because of erroneous estimates of the operating environment and/or
the process streams.
Since much of the information for the problem analysis was covered in
Volume II, this volume summarizes that information to minimize repetition.
-------
SECTION 2
REFRACTORY
The refractory lining the interiors of the kiln, crossover duct, gas
purifier, and ductwork protects the structure metals from excessive heat. A
refractory failure at any point along this protective surface is cause in
itself to shut down the system. A refractory failure may also indirectly
prompt a system shutdown since the refractory fragments can damage or jam the
residue drag conveyor, the slag screw conveyors, or block the slag flow into
the residue outlet of the gas purifier.
A shutdown for refractory repair considerably increases the operating
costs per throughput ton since the placement and proper cure of new castable
refractory generally requires several days.
KILN REFRACTORY
Before any waste was fired, some of the original kiln refractory spalled
off. Shortly after refuse processing was started, substantial amounts of the
kiln refractory in the feed end of the kiln had to be replaced because of
severe damage. Monsanto attributed this failure to the thermal shock caused
by cooling water leaking out of the damaged ram snouts. To solve this problem,
the water to the snouts was turned off, and consequently, the snouts eventually
had to be replaced. In addition, the refractory at both ends of the kiln
fell out regularly until the shell was slotted at both ends to allow for the
movement of the kiln shell resulting from thermal expansion and contraction.
Also, the refractory in the area of the kiln flights and spikes had to be
frequently reinstalled because of failures attributed to patching the refrac-
tory during initial placement. After the entire kiln was relined, the
refractory remained intact during the rest of the demonstration period.
As evidenced by the refractory spalling before any refuse was fired and
by the need to replace the first 8 meters of refractory in the feed end
shortly after the start of the refuse processing, the original refractory was
probably worked excessively and improperly cured. The installation con-
tractor's willingness to make financial adjustments would support this
conclusion. Normally, after castable refractory has been installed, the heat
for the refractory curing should not be applied for at least 24 hours. Then
for a unit with the size of the kiln, a low-fire heat should be applied for a
week to ten days to gradually dry out the refractory lining. When heat is
applied too rapidly, it generates steam which is frequently trapped and can
only be released by an explosive force which spalls the refractory. The
-------
steam is more likely to be entrapped when the surface has been worked
excessively by troweling the surface very smooth. The resulting smooth
surface is too dense for the moisture to escape.
The kiln design criteria called for constant operating gas temperatures
at 650°C (1200°F) to prevent slagging and residue fusing into large balls.
Since slagging and residue fusing both occurred, the actual operating
temperatures are acknowledged to have exceeded design temperatures
considerably. The high temperatures probably caused anchor failure with the
consequent loss of large pieces of refractory at the fire end of the kiln.
While the cold mechanical strength of castable refractory is sufficient to
maintain structural integrity at temperatures up to about 1093°C (2000°F), it
falls off rapidly above this temperature. With the refractory strength so
reduced, the pounding of the large masses of fused residue rolling in the
kiln could disintegrate the refractory.
Metal refractory anchors have definite temperature limits: 927°C (1700°F)
for 304 stainless steel and 1038°C (1900°F) for 310 stainless steel. Above
these limits, castable refractories have a tendency to spall to the depth of
the anchor tips. If the higher than designed operating temperatures could
have been anticipated, the anchors would not have been made of metal. The
actual operating conditions with metal anchors left no margin for processing
temperature control error.
The problem of supporting the refractory shell within a rotating vessel
was aggravated by constructing the kiln with two metals: mild or carbon
steel for the shell, and stainless steel "end rings" as discussed in
Volume II of this report. Therefore, the refractory for the kiln shell and
that for the kiln end sections should have had expansion joints to compensate
for the differential between the carbon steel and stainless steel.
The refractory spalling was probably due to a combination of thermal
shock spalling and structural spalling. Thermal shock spalling is caused by
thermal cycling during shutdown and start-up or by temperature variations
during operation. While spalling can occur at any temperature, the degree of
spalling is a function of the severity of the temperature change.
Structural spalling is caused by impurities within the refractory. As
the kiln pyrolysis gas passes in and out of the permeable refractory, iron
impurities in the refractory frequently dissociate carbon from carbon
dioxide and hydrocarbons, especially mathane, in the gas. The carbon so
released deposits in the outer layers of the refractory. This process is
very active in the range of 540°C to 1100°C (1000°F to 2012°F). Then as the
refractory cools, the imbedded carbon produces stresses that cause spalling.
The extent and frequency of such spalling is a function of the quantity of
iron impurities in the refractory.
While the failure of the original refractory is considered unique to the
Baltimore plant since it was due primarily to improper refractory installation
and curing, the spalling of the present refractory is considered unique to
-------
the Landgard process since it has been due to the kiln processing conditions.
Improved process stability should contribute substantially to minimizing
this spalling.
GAS PURIFIER REFRACTORY
The refractory failure in the gas purifier was less dramatic than that
in the kiln. Rather than falling out in large pieces as in the kiln, the
refractory in the gas purifier diminished by the constant decrease of its
thickness at a rate of aobut 1 mm (0.04 in.) an operational day. The
only dramatic refractory failures in the gas purifier occurred in the
baffle and orifice walls. As the refractory thickness decreased, the gas
purifier was first water-cooled by spraying water on the outside shell
and ultimately required rebricking. The original gas purifier is currently
being replaced by a nonslagging afterburner which will likely minimize
refractory failure by reducing both the operating temperature and gas velocity.
The refractory problems in the gas purifier were due basically to the design
specifications that did not anticipate the actual operating conditions. In
particular, the slag composition and the vessel temperatures differed
considerably from those in the design data.
The alumina-silica refractory originally installed in the gas purifier
is typically used for acid steel slag. The slag in the gas purifier was
similar to a coal slag but with slightly lower iron and higher alkali
metal concentrations (Table 1).
TABLE 1. TYPICAL SLAG COMPOSITION
. U.S. Coal Gas Purifier
Constituent (Percent by Weight) (Percent by Weight)
Si02
A1203
Fe203
Ti02
CaO
MgO
Na20
K20
S03
P203
20-60
10-35
5-35
0.5-2.5
1-20
0.3-4
0.1-2.4
0.3-3.5
0.1-1.2
57
18
4
3
9
2
4
1
<1
2
-------
Slag-Refractory Interactions
As a service to the City of Baltimore in the attempt to resolve the
gas purifier refractory problem, the Bethlehem Steel Corporation conducted
penetration tests with the gas purifier slag and the following 10 refractories
to determine which of these refractories would serve best for the gas
purifier rebricking:
ALUSITE D (Harbison-Walker Refractories)
ALTEX 85 (General Refractories Company) (original refractory)
DV-38 (A.P. Green Refractories Co.)
LOFERO HM (Kaiser Refractories)
NUCON 60 (Harbison-Walker Refractories)
OXIBAK H (Harbison-Walker Refractories)
KORUNDAL XD (Harbison-Walker) Refractories)
REFRAX 50A (Carborundum Refractory Division)
HARBIDE (Harbison-Walker Refractories)
REFRAX 20 (Carborumdum Refractory Division)
The test specimens were prepared as follows: For each of the 10 refrac-
tories, a hole or "cup" was bored into each of two blocks and the two holes
were each loaded with a black gas purifier slag. Next, one block of each
pair was heated in a furnace to 1375°C (2500°F), and the other block of the
pair was heated to 1425°C (2600°F). Then the blocks were each cut in half
through the cup axis to yield the specimens which were mounted on a board,
as shown in Figure 1. Next, SYSTECH used the scanning electron microscopy
(SEM) technique to collect two types of data for the assessment of the slag
penetration into, and the slag attack on, the refractory.
One data type consisted of two differently magnified SEM photographs
of each block specimen. The center point sighting for each photograph was
at the midpoint of the bottom slag-refractory interface. Of the two SEM
photographs for each block specimen, one is a 10x magnification, and the
other is a 25* magnification. Also, for each block specimen, the second
data type consisted of relative constituent percentages at each of several
discrete points along the line perpendicular to the slag-refractory interface.
In the following side-by-side presentations of the magnified SEM
photographs of each specimen block, the 10* magnification is at the left
and the 25X magnification is at the right. The lines on each side of the
individual photographs indicate the reaction zone at the slag-refractory
interface where the slag and refractory are above and below the interface,
respectively. In the 25X magnification, the light dots aligned vertically
to the interface are the SEM probe points for the relative constituent
percentage measurements. These points were numbered consecutively from the
lowermost to the uppermost to denote their relative locations.
-------
COHOUCTID |Y HTKltMtM *T£tl COM.
Figure 1. Board-mounted assembly of all slag-refractory block specimens tested
for slag penetration.
-------
Alusite D (Harbison-Walker Refractories), a fairly high alumina refrac-
tory which showed some evidence of phosphate bonding in the test specimen,
has a maximum service temperature of 1650°C (1800°F) and the following
chemical composition:
A1203 71.8%
SiOa 22.9%
Fe203 1.5%
MgO 0.2%
Lime 0.2%
Alkalies 0.5%
Figure 2 shows the SEM photographs of the block specimen tested at
1425°C (2600°F). Tables 2 and 3 list the relative constituent percentages
for the tests at 1375°C (2500°F) and 1425°C (2600°F), respectively.
At 1375°C (2500°F) the slag penetration was very slight. The reaction
interface is sharp1 with no degradation of the refractory surface. At
1425°C (2600°F) the penetration ranged from 1 to 3 mm. The cup wall above the
slag pool was wetted. The reaction zone is 0.5 mm wide with evidence of
Si02 attack.
Altex-85 (General Refractories Co.) is a phosphate-bonded, unburned,
high-alumina refractory with an optimal service temperature of 1750°C
(3200°F) and the following chemical composition:
A1203 84.8%
Si02 5.8%
Ti02 2.8%
Fe203 2.0%
CaO 0.2%
MgO 0.4%
P205 2.6%
Alkalies 0.2%
The SEM photographs (Figure 3 and 4) and constituent data (Tables 4
and 5) show that at 1375°C (2500°F) the slag dissolved 1.5 mm of the cup
wall and penetrated 1 mm farther into the remaining wall. There is evidence
of strong attack on Si02 in the reaction zone which is 1 mm wide. At
1425°C (2600°F) the results are the same except that the slag penetrated
deeper, to approximately 3 mm.
DV-38 (A. P. Green Refractories Co.) is a phosphate-bonded, dry-pressed,
high-alumina refractory containing a nominal 85 percent alumina with a
fusion temperature of 1875°C (3425°F).
The SEM photographs (Figures 5 and 6) and the constituent data
(Tables 6 and 7) reveal that at 1375°C (2500°F) the slag slightly penetrated
the refractory. At this temperature the cup wall above the slag pool was
wetted. The reaction interface is quite sharp but evidences some Si02
attack. At 1425°C (2600°F) the slag penetrated the refractory up to 1 mm
-------
(IOX)
(25X)
Figure 2. SEM photographs of Alusite-D block specimen tests at 1425°C.
-------
TABLE 2. ALUSITE-D SLAG PENETRATION TEST AT 1375°C
Location
3
4
5*
6t
7
i n
XU
1 0
Xj
*S - 5.26
tP - 0.92
Al
22.08
22.84
15.79
21.26
24.95
77 i n
/ / . XU
7s. nn
/ 3 . UU
cf. on
3D . 07
AR nn
Si
48.05
48.35
38.25
52.68
51.90
on AT
zu . ox
T.C, 71
JJ . / X
oc nn
Z J . UU
AQ fi?
H7 . OZ
Al 79
HX . -3Z
co nn
JZ . UU
Constituent (%)
Ti Ca K
1.30 27.27 1.30
2.06 23.66 2.06
1.75 36.84 1.75
4.07 21.63 4.07
1.00 21.96
o on
0 CC _ _
1 P.n
Fe
1.03
0.35
0.92
0.20
10
-------
TABLE 3. ALUSITE-D SLAG PENETRATION TEST AT 1425°C
Location
1
2
3
4
5
f.
7
ft
0
i n
12
13
14
1 s
Al
27.07
28.25
29.09
75.68
27.03
sn on
SI 81
SS ^8
77 Oft
fin i ^
55.33
fiQ 79
fiQ 7?
fiR IS
C
Si
49.62
53.11
38.18
18.02
40.54
sn nn
AO 07
^n nn
ni o
^6 fin
37.33
?S fiQ
9? fiA
99 79
lonstituent (%)
Ti Ca Fe
1.50 22.56
1.13 16.95
30.91
1.80 3.60
2.70 29.73
0 ft!
Q na _
n 09 _
197 —
3.33 4.00
L. SQ —
7 cs i on
^ in _ •* TQ
11
-------
I
(IOX)
(25X)
Figure 3. SEM photographs of Al tex-35 block specimen losted at 1J75°C.
-------
(IOX)
(25X)
Figure 4. SEM photographs of Altex-85 block specimen tested at 1425°C.
-------
TABLE 4. ALTEX-85 SLAG PENETRATION TEST AT 1375°C
Location
1
2
3
4
5*
6
7
Q
0
10
11
i 9
13
1 L
1 ^
Al
78.00
28.39
87.67
90.71
46.03
81.82
83.10
RT sn
68.09
75.96
57.14
7n i 8
/ U . -LO
Q=; n«
Si
0.40
40.55
5.48
3.83
34.13
9.09
8.10
S 71
23.83
17.80
34.29
3C1
. J-L
Constituent
Ca
0.40
24.33
3.77
2.40
7.94
4.55
2.11
0.85
2.86
(%)
Ti
1.00
3.65
2.05
1.97
5.56
2.27
4.23
fi 9S
2.13
1.78
2,86
Q 77
L i n
Fe
13.00
3.08
1.03
1.09
3.17
,2.27
1.06
i n/.
1.70
0.89
1.90
17 SA
X/ . JH
n 89
p
0.79
1.41
3.40
3.56
0.95
*K - 2.38
14
-------
TABLE 5. ALTEX-85 SLAG PENETRATION TEST AT 1425°C
Constituent (%)
Location Al Si Ca Ti Fe
1
2
4
f.
7+
Q
g
i ns
i ?
1 0
14
1 S
23.91 41.30 26.80 5.80 2.17
22.86 32.38 32.38 8.57 3.81
97.17 1.77 1.06
QQ ^9 — _. n 7A n i/i
91 QQ 19 SO L 6Q —
Qfl 11 ^ c:q o on
Of) 90 _ c: 9^1 9 /.o
79.36 18.35 1.38 0.92
* Ag - 1.0
t S - 17.97,'C1 - 35.94, K - 6.25, Cu - 0.78
§ P - 2.02
15
-------
(IOX)
(25X)
Cv' 3. SEM p'uo tc>t;raphs oi LA'-i-S block spi'c imen ti--sLed at l.i;';>°C
-------
(IOX)
(25X)
Figure 6. SEM photographs of DV-38 block specimen tested at 1425°C.
-------
TABLE 6. DV-38 SLAG PENETRATION TEST AT 1375°C
Location
1
2*
3t
4J
5
6
7
8
9§
10
11
12
13
Al
79.25
67.69
58.85
46.98
86.67
94.12
1.00
76.19
54.68
84.91
78.69
80.00
Constituent (%)
Si Ca Ti
9A 0
20.51 3.59
15.10 1.56
24.16 1.34
5.56
1.31
11.11
35.97
4.72
4 . 92
10.59
7.55
2.56
1.04
3.36
22.22
1.96 l
3.17
2.16
3.30
4.92
2.35
Fe P
o 77
j . / / ~
0.51 3.08
0.52 7.29
0.67 6.04
5.56
2.61
9.52
0.72 2.88
2.36 4.72
1.64 9.84
7.06
* K - 2.05
t F - 15.63
t K - 2.01, F - 15.44
§ K - 3.60
18
-------
TABLE 7. DV-38 SLAG PENETRATION TEST AT 1425°C
Location
1
9
0
L
5
6
7
8
g
i n
12
1 3
14
Al
66.00
ftn 77
QO QC
o j . yj
QQ £9
82.61
QO ?n
7Q "^1
fin no
81.72
Q8 9?
68.13
Constituent (%)
Si Ca Ti Fe
14.00 4.00 6.00
_ S 77
~ _ j . / / — — —
1 9 T? 9 "^fl 1 9 "3
-LZ.JJ — — Z.JU ±.ZJ
1 1 AR
4.35
n Q8
17 9/. __ ^AS
•an nn _ in r\r\
6.45 3.23
1 "I ^ 161
10.99 3.30
P
10.00
1 1 Afi
13.04
8Q9
8.60
17.58
19
-------
with the cup wall above the slag pool being wetted. The reaction interface
is quite sharp at 1425°C (2600°F) but the attack was worse than that at the
lower temperature.
Lofero HM (Kaiser Refractories) is a hard-burned, high-magnesium
refractory formed from priclase with the following chemical composition:
MgO 98.0%
Si02 0.7%
CaO 0.6%
Fe203 0.4%
A1203 0.3%
The SEM photographs (Figures 7 and 8) and the constituent data (Tables 8
and 9) show that the refractory totally absorbed the slag to a depth of 1 cm
with a resultant severe attack of the material matrix and no clearly defined
reaction interface.
Nucon 60 (Harbison-Walker Refractories) is a high-magnesia refractory
with the following chemical composition:
MgO 63.8%
Si02 1.5%
A1203 13.2%
Fe20 6.3%
Ca203 14.5%
Lime 0.2%
As evidenced by the SEM photographs (Figures 9 and 10) and the consti-
tuent data (Tables 10 and 11) the slag was almost completely absorbed by the
refractory at both temperatures to a depth of 1 cm with resultant moderate
attack of the refractory surface and no clearly defined reaction interface.
Oxibak H (Harbison-Walker Refractories) is a very high-magnesia refrac-
tory with course aggregate and the following chemical composition:
MgO 95.7%
Si02 1.3%
A1203 0.3%
Fe203 0.3%
Lime 2.4%
The SEM photographs (Figures 11 and 12) and the constituent data
(Tables 12 and 13) indicate that the slag was completely absorbed by the
refractory at both temperatures to a depth of 1 cm. The slag attacked the
efractory matrix so severely that refractory aggregate particles intermingle
with the slag material in the reaction zone. Consequently, there is no
clearly defined reaction interface.
20
-------
I 4 **, t *f
(IOX)
rigure 7. SEM photographs of Lofero-HM block specimen tested at 1375°C.
-------
(IOX)
(25X)
Figure 8. SEM photographs of Lofero-HM block specimen tested at 1425°C.
-------
TABLE 8. LOFERO SLAG PENETRATION TEST AT 1375°C
Location
i
7
3
A
fi
7
q
i n
12
13
i s
Mg Si Ca
40 88 4"} 80 14 T^
L."\ S9 40 RS ^ ^9
59.34 30.77 3.30
Sfi 1 0 ^171 610
81 1 S 11 Aft A Q9
Q Q C Q
51.85 14.81 3.70
QO no -
Q O O O
AQ QC:
AA A1^ ^ H7 A OA
61.29 12.90 8.87
71.43 5.71 11.43
4fi fiO 7 8S 4 1 q
qo S7 l 8q 1 77
Constituent (%)
Al Ti Fe S
07fl 1 41
. / U ~ J- . t J-
1.10 3.30
— ft 1 0
- 9 4ft
- ft 41
22.22 7.41
_ 7 c:n
— S OH
— A ft/.
8.06 3.23 1.61
4.29 7.14
9 ft9
1 77
K
2.20
9 90
9 SO
1 67
9 A9
4.03
1 OS
23
-------
TABLE 9. LOFERO SLAG PENETRATION TEST AT 1425°C
Constituent (%)
Location Mg Si Ca Al Ti Fe S K
1 50.54 46.24 3.22
2 94.23 5.77
3 89.29 10.71
4 94.12 5.88
5 88.57 5.71 2.86 2.86
6 95.45 4.55
7 58.62 24.14 17.24
8 50.00 16.67 16.67 16.67
9 97.09 2.91
10 27.78 38.89 33.33
11 54.55 9.09 11.36 15.91 9.09
12 46.43 41.07 8.93 3.57
13 43.48 21.74 26.09 8.70
14 24.19 19.35 32.26 8.06 16.13 —
15 24.54 26.38 49.08
24
-------
K
(IOX)
(25X)
Figure 9. SEM photographs of Nucon-60 block specimen tested at 1375°C.
-------
25X)
Figure 10. SEM photograph of Nucon-60 block specimen tested at 1425°C.
-------
TABLE 10. NUCON-60 SLAG PENETRATION TEST AT 1375°C
Constituent (%)
Location Mg Al Si Cr Fe
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
1 7
99 09 — 10 SR
32.50 18.75 7.50 23.75
U11 7/i1 ^ Q 9 6
7 60 1091 — 61 SA
S7 19 6 17 99 9Q
66 A6 116 IS R9
^179 O / 1 /
81 AH S fll
87 1 9 - — 7 SR
Rn nn _ __ i n nn
A6 SR 1A 9S
77 11 _ 1970
7S 68 - 1 1 SI
n9R - S7 AS
23.91 10.87 26.09 10.51
76 09 in nn
8.75
22.22
5.77
10.83
59.46
11.39
24.14
6.98
5.30
8.13
19.18
7.56
10.81
21.28
4.71
6 09
Ca S
17 sn
8.75
118
116
1 RR
-L . Oo
23.91
6 i •;
27
-------
TABLE 11. NUCON-60 SLAG PENETRATION TESTS AT 1425°C
Location
i
9
0
j
A
Q
A
7
/
i n
XU
1 9
J.£
1 °.
XJ
1 A
XH
1 C
X J
1 A
XO
i 7
1 A
Mg Al
AA AT 7 R1
1 S An T.1 Ql
A1 99 in 9n
OX . t-L XU . £U
AA 1 n -
1 0. 1 f. -1 7 C A
U79 10 S7
. / £ X:7 . JJ
noc i Q 71:
. f.J XO . / J
9 S QA 91 1 "^
Lj . 70 Z-L . ±3
on An 1 T Q1^
JJ. . tu J. j . y j
SI 9R
90 on 9R Qn
i.o . OU /.O . OU
07 0.7
A7 "^7
fi/i fi9
1:7 n/. i A on
J / . U4* XH . OJ.
90 0.0.
sn nn
Si
0 Ql
J . J J.
9 RA
sin
J . J.U
no. f.
9 A nA
ZD . Ut
1 A T.S
J.O . J J
-i o. qc
X j . :7 J
00 O.T
1 n A7
J.U . t /
S9 S°,
A9 1 1
Afi 67
9T. S7
Constituent
Cr
1 Ql fi
j . yjL u
AA in ^
to . xu j
T c 0.1 a
XJ . JX o
8c c i
. J J X
A9 9R 7
A9 Sn A
DZ . JU O
1771 A
X/ . / X D
9 A nA 1 9
it . Ut Xi
0.1 An Q
jX . tu :»
1 CL 00
97 77, A
/. / . / 1 t
90. nR 1 9
/. j . uo xz.
9999 S
ZZ . ZZ 3
1 9 1 A Q
(%)
Fe Ca] Ti
oc;
c:c:
i ^
nA 97
fl9 _
9 c _
9 c
on _
. JU _— — — —
i q
i n i n —
i r» co. _ _
01 _
qo.
-— 9A A7 "\ "\1
9Q s nn
28
-------
N>
VO
(IOX)
(25X)
Figure 11. SEM photographs of Oxibak-H block specimen tested at 1375°C.
-------
u>
o
(25X)
Figure 12. SEM photograph of Oxibak-H block specimen tested at 1425°C.
-------
TABLE 12. OXIBAK-H SLAG PENETRATION TEST AT 1375°C
Location
i
9
Q
/,
^
f.
7
0
q
i n
1 9
1 1
1 L
i c
1 ft
17
Constituent (%)
Mg Si Ca S Al
i r»n no - -
•31 9 c 91; nn A0. 7 R — — —
JJ../.J £J , UU tj./J
71 A^ 1 9 QQ 1 ^ ^8
/J..H-3 xZ.33 J.J.JO — ~
QO 00 C (L-J
7J.JJ — — — — _ _ _ j . O /
qi -jn _ _ Q 7A
0191; __ 187^
OJ../.J — — J.O . / J
f.-\ 99 i c; 0,1 90 /,7
qc QO. /, 1 7 _
;? J i oj H . J./ ~
Qn AS q ^9
yu . to y . jz —
i nn nn —
1 f) A O, 9ft 00 tT, T ft
?s nn "^n nn ^s nn
57.47 11.49 25.29 5.75
31
-------
TABLE 13. OXIBAK-H SLAG PENETRATION TEST AT 1425°C
Location
1
2
Q.
4
5
7
q
7
i n
J.U
1 9
1 A
J.H
15
16
17
Mg
68.11
90.91
89 9A
38.67
61.54
Q/. 01
ftQ "}£
Q7 QQ
O / . OO
qn 9n
R7 97
O 1 , LI
QQ AQ
oo . to
Ql ST
7J. . J J
qn nn
84.48
80.88
80.29
Si
11.15
1.82
30 Q
• £m J
33.33
4.62
7 n9
A 96
1 R9
-L . OZ
29C
. Z J
3oq
. J7
3.45
5.88
7.30
Constituent
Ca
6.81
3.64
Q O Q
10.67
18.46
c 9ft
3n°.
. U J
A 91
1 . Z.L
5 no
..uo
3.45
8.82
3.65
S
3.10
3.64
A £1
-7.69
•3 C1
A sS
H . 3 J
7 8A
7 97
/ . /. /
5n^\>
. Uu
i n nn
5.17
4.41
8.76
Al
10.84
17.33
7.70
3.45
32
-------
Korundal XD (HarbisonWalker Refractories) is a very high alumina
refractory with the following chemical composition:
A1203 90.8 %
Si02 8.5 %
Fe203 0.2 %
MgO 0.07%
Lime 0.07%
Alkalies 0.15%
Samples of the refractory at both temperatures exhibited no slag
penetration (Figure 13) and no discernible refractory degradation (Tables 14
and 15). The cup wall above the slag pool was wetted and glazed. However,
the wetting and glazing apparently had only a surface effect with no
refractory attack, with the slag-refractory interface remaining sharp.
Refrax 20 and 50A (Carborundum Refractory Division) are high silicon
carbide refractories with the checmical compositions listed in Table 16.
Samples of both of these refractories at both test temperatures shows slight
wetting of the cup wall above the slag pool and no penetration with slag
separation from the cup after cooling (Figure 14). The attack on the
refractory surface was very slight, and the slag-refractory interface remained
very sharp. The constituent data in Table 17 confirms the visual observations.
Harbide (Harbison-Walker Refractories) is a silicon carbide refractory
with the following chemical composition:
SiC 89.2 %
Si02 7.3 %
A1203 2.1 %
Fe203 1.1 %
Alkalies 0.08%
Samples at both temperatures exhibited no slag penetration and slag
separation from the refractory after cooling. The cup wall above the slag
pool was wetted with a slight attack of the refractory surface. While there
was evidence of some Si02 degradation, the slag-refractory interface remained
sharp.
The slag penetration tests demonstrate that no single refractory can
permanently serve as the lining for combustion chanbers handling molten
slag since the molten slag at high temperature will actively attack brick-
work by taking it into solution. Because of its high alkali metal oxide
concentrations, gas purifier slag will attack refractory more than will coal
ash slag. The current design practice for facilities handling molten coal
ash sag calls for water-cooled floors and walls to maintain low refractory
temperatures.
33
-------
(IOX)
(25X)
Figure 13. SEM photographs of Korundal-XD block specimen tested at 1375°C.
-------
TABLE 14. KORUNDAL-XD SLAG PENETRATION TEST AT 1375°C
Location
1
2
4
5
6
7
q
i o
i 9
1 T
-LJ
1 C
j. j
i A
Al
27.17
30.30
100 00
62.92
83.58
80.00
71.87
70 CM
7S 71
76 Tfi
07 91;
7 / . ^ J
Ql AC
07 QA
O / . ;7t
Q£ CA
7C qo
/ J . 7 J
HH nn
Constituent (%)
Si Ca Ti Fe
51.09 20.65 1.09
50.22 19.48
31.46 3.37 2.25
12.69 3.73
16.84 3.16
23.61 2.46 2.05
9A 9Q —
90 f./, _
9 7 S —
noi -7 o/
• -5-1 7.24
1 9 Of\
1 0 At _ _ _
i 9 nn
35
-------
TABLE 15. KORUNDAL-XD SLAG PENETRATION TEST AT 1425°C
Location Al
1 21.78
2 23.70
•J SS "}"}
L i on no
5Q1 QO
Oi. . ji.
6QO CO
Of. . JO
7 17 AA
/ J / . Ot
8 43.10
9 21.45
TABLE 16.
SiC
Si02
A1303
Fe203
CaO
Constituent (%)
Si Ca Ti
51.32 21.94 3.84
27.08 20.57 11.72
OA 7ft Q 01
i Q no
1 7 A9
AO OC
36.20 14.65 2.60
40.48 22.61 3.22
REFRACTORY CHEMICAL COMPOSITION
Refrax 20
79.15% 85.4%
0.26 0.18
0.64 0.64
0.66 0.66
0.30 0.30
Fe
1.12
7.03
1.04
1.79
36
-------
U)
•si
(25X)
Figure 14. SEM. photograph of Refrax-50A block specimen tested at 1375°C,
-------
TABLE 17. REFRAX-50A SLAG PENETRATION TEST AT 1375°C
Location
1
2
3
4
5
7
/
i n
_LU
1 1
Constituent (%)
Al Si Ca Ti
19.47 48.66 26.79 3.21
18.74 51.79 25.16 2.74
. 20.67 55.01 18.24 2.62
21.40 54.29 18.68 2.38
20.67 57.63 17.32 2.00
_ QQ 9Q n LL
— i on nn —
RI n^ — —
QQ 1 (\ n RA — — —
i nn _ _
Fe
1.87
1.30
1.06
0.96
i f. 91
J-D . i-i.
38
-------
Other Refractory Failure Factors
In addition to slag-refractory reactions, other major factors con-
tributing to the failures of the refractory walls and baffles in the gas
purifier were the high operating temperatures, constituents in the
combustion gases, excessive thermal cycling and chocking, severe operating
conditions, and excessive erosion.
While the gas purifier was designed to operate at an average tempera-
ture of 1370°C (2500°F) with a maximum of 1450°C (2800°F), the temperature
frequently exceed the temperature recording limit of 1650°C (3000°F) as
evidenced by the thermocouple failures. However, since the fluid tempera-
ture of the gas purifier slag was 1360°C (2490°F) and the combustion
temperature must be a few hundred degrees higher than that temperature to
keep the slag fluid, the average combustion temperature shoudl have been
about 1650°C (3000°F) to maintain the slag at a temperature high enough
to allow the fluid slag tapping process to occur. In addition, the slag-
handling experience has established that to maintain a good slag flow the
slag viscosity must be 250 poise or less, which, in this case, requires a
temperature of at least 1482°C (2700°F).
The refractory-damaging constituents in the combustion gases included
chlorine, hydrogen sulfide, sulfur dioxide, fluorine, and alkalies.
In vapor form, chlorine, fluorine, and alkalies will reduce the life of
many refractories at temperatures far below their normal temperature range.
Hydrogen sulfide and sulfur dioxide are harder on refractories than the
normal products of combustion.
The excessive thermal cycling and thermal shocking was due to inadequate
control of the combustion air blown into the gas purifier. This inadequate
control was in turn due to the improper monitoring and measuring of the
combustion temperature and 02/C02 concentrations.
The operating conditions were too severe for ordinary high alumina
refractories. To have ensured reliable long-term operation in the slagging
mode, the original gas purifier should have been lined with a pure alumina
(99% alumina) refractory with a PCE of 42 (2015°C-or 3659°F) and equipped
with a water-wall bottom. Pure alumina refractories have a high resistance
to the fluxing action of basic oxide and maintain a high module of rupture,
crush strength, and load bearing capacity.
The particulate erosion was excessive because of the high particulate
levels in the combustion gases and the high gas velocities. The erosion
was most severe in the area where the gases in the crossover duct enter the
gas purifier. In this area, the gas stream has a directional change and the
gas temperatures are the highest since this is immediately downstream of the
combustion air inlets.
39
-------
Since the prototype gas purifier in the Landgard process operated in
a nonslagging mode and all factors contributing to the failure of the gas
purifier refractory were related at least indirectly to the slagging require-
ment, the failure of the gas purifier refractory is categorized as unique
to the Baltimore plant.
DUCT REFRACTORY
The duct refractory includes the refractory for the crossover duct,
the gas purifier exit duct, the boiler inlet ducts, and the scrubber inlet
duct. In general, the refractory failures of these ducts were due to the
following factors which are also common to the kiln and gas purifier
refractory failures: severe operating conditions, excessive combustion
gas temperatures, thermal cycling, material damaging gas constituents,
thermal shock and structural spalling, and particulate erosion.
While most of the duct refractory has deteriorated substantially,
only the gas purifier exit duct and some portions of the scrubber inlet
duct refractory have been replaced. The original refractory in the gas
purifier exit duct was a castable refractory anchored with metal hairpin
anchors which was replaced with a high alumina brick refractory similar
to the gas purifier refractory.
The crossover duct has two zones: the first extends from the feed
hood to the combusion air inlet, and the second extends from the combustion
air inlet to the gas purifier. In the first zone, refractory failures were
minimal because the temperatures were too low for slagging and severe
thermal shock. The refractory degradation in this zone was due to the
erosion/abrasion from entrained particulates (over 45 kg/min), thermal
cycling effects, and attacked by carbon monoxide, hydrocarbons, alkalies,
and chlorides in the kiln-off gases.
In the second zone, the kiln-off gases have much higher temperatures
and a greater turbulence. These conditions, coupled with the start-up
and shutdown effects, cause severe thermal cycling and shock. Also, slag
accumulates and liquifies in this zone, and the molten slag attacks and
dissolves the refractory at the bottom of the duct. In addition, as the
gases are discharged from the duct into the gas purifier, they undergo a
directional change with the consequent increased turbulence which makes the
particulate erosion/abrasion more severe.
If practical, the best means for minimizing the refractory failure
in the second zone would be the installation of a water-cooled shell with a
thin coating of 90 percent alumina plastic refractory on the inner surface.
If the foregoing is not practical, the next best means for reducing the
refractory failure is believed to be relining the duct bottom with pure
fused alumina (99% alumina) and the duct top with a plastic 90 percent
alumina material.
40
-------
Gas Purifier Exit Duct
Excessively high temperatures caused the original castable refractory
in this duct to sheet off at the tips of the anchors. Then the degraded
refractory became highly susceptible to slagging and spalling. The
•refractory degrading was further intensified by particulate erosion/abrasion
particularly at the turn in the duct.
When the castable refractory was replaced with high alumina brick,
the sheeting off was eliminated and the slagging, spalling, and erosion
were greatly reduced except in the duct bottom where the flowing molten
slag eroded the brick by dissolving it. If a pure fused alumina (99%) were
substituted for the high alumina bricks, it would reduce the slag effects
to more acceptable levels and better resist the erosion/abrasion and
spalling effects. Here again, the best means for minimi zing the refractory
failure would be to water cool the duct walls.
Boiler Inlet Ducts
Excessive temperature due to the lack of effective quench air control
was the primary cause of the refractory failures in the boiler inlet ducts.
To prevent slagging and fouling in the boiler requires that the temperature
in the inlet ducts be carefully controlled by metering the quench air
precisely and thoroughly mixing the quench air with the combustion gas. If
the recommended water-wall construction in the crossover duct, gas purifier,
and gas purifier exit duct were implemented, the amount of molten particulate
entering the boiler would be greatly reduced, and consequently, the boiler
slagging and fouling would be minimized.
Scrubber Inlet Duct
Refractory failures in the scrubber inlet duct were due to two causes:
the first and the most damaging was the effect of the caustic and the water
in the scrubber quench spary, and the second was the erosion/abrasion of
particulates suspended in the combustion gas stream. The caustic reaction
with the refractory caused the surface layers to swell and then structurally
disintegrate. As water impinged on the hot refractory it caused severe
thermal shocking. In addition, water penetrated the permeable refractory
and cased spalling of the refractory. As the refractory degraded because
of the foregoining damaging effects, it became more susceptible to the
erosion/abrasion of the suspended particulates.
The refractory damage was reduced by eliminating the caustic in the
spray and by controlling the spary so that water would not impinge on the
refractory. The elimination of the spary itself, by either allowing the
gas to enter the scrubber hot or quenching the gas with air, would likely
have improved the situation but would have required increasing the capacity
of the induced draft fan.
41
-------
If the spary were eliminated with the consequent lesser susceptibility
of the refractory to the particulate erosion/abrasion, periodic refractory
patching with a very high abrasion-resistant plastic, such as Greenpak-90-PF,
could have sufficed to maintain the structural integrity of the refractory.
The failures of the various duct refractories are considered to be
unique to the Landgard process because they resulted from the conditions
existing in that specific process.
42
-------
SECTION 3
SLAGGING
The two major slagging problems at the Baltimore plant were slag hole
pluggage and the slagging over of the boiler tubes at the boiler inlet.
Pluggage of the gas purifier slag hole does not require immediate shutdown,
but continuing slag build up over a long period in the gas purifier would
eventually cause structural failure of the gas purifier. Boiler tube
slagging does not require immediate shutdown but can damage boiler tubes,
cause low energy recovery efficiency, and require excessive maintenance to
restore the boilers to operation. Slagged tubes can also cause excessive
pressure drop through the boiler and overload the induced draft fan.
The discussion in Section 2 of this report established that the gas
purifier slag is very similar to bituminous coal slag. There is a large
body of coal slag operating experience and data which will be referred to
extensively in examining the problems of slag hole pluggage and boiler inlet
slagging.
SLAG HOLE PLUGGING
Initially, the slag tap hole plugged frequently. A fuel oil burner was
installed in the hole, but it did not prevent plugging and damaged the
refractory in the slag tap hole. To minimize the heat loss, the size of the
slag hole was reduced, and to form a concentrated slag flow, a dam with a V-
notch on two sides was installed. At the same time, the operating tempera-
ture of the gas purifier was increased to 1370°C (2500°F). When these
modifications failed to prevent slag hole plugging, another fuel oil burner
and two propane burners were installed. These burners helped to keep the
slag hole open, but they did not prevent slag hole plugging. To eliminate
this problem, the city is replacing the gas purifier with a nonslagging
afterburner.
It has been noted in Section 2 that the gas purifier was originally
designed to operate at a temperature of 1100°C (2000°F). This should be
compared to the average gas purifier slag softening and fluid temperatures
of 1081°C (1978°F) and 1187°C (2169°F) respectively. Obviously, at the
original design operating temperature the slag will not flow effectively.
43
-------
Determination of Slag Tapping Characteristics
The two methods used to determine the slag tapping properties of a
particular slag are the ash fusion temperature and slag viscosity measure-
ments.
Ash Fusion Temperatures—
The melting temperature of slag is, due to its complex composition, not
a single sharply defined point. As the temperature of a slag is increased,
the ash begins to soften and becomes less viscous until a completely liquid
state is reached. By definition, the various stages are called the "initial
deformation temperature," "softening temperature," and "fluid temperature."
The standard measure of 'initial deformation temperature is the temperature
at which a prepared standard cone of free, individual ash particles begins
to fuse or show evidence of deformation at the top of or apex of the cone as
it is heated in a reducing atmosphere. The softening temperature (also
known as fusion temperature) corresponds to an observed temperature at which
the apex of the test cone of ash has deformed to the point where it assumes
a spherical shape. This is the temperature at which the ash demonstrates a
greatly increased tendency to agglomerate and adhere to heat transfer
surfaces. The fluid temperature is determined by further heating of the ash
test cone until the cone flattens and assumes a pancake shape. At this
temperature, the slag can be expected to flow in streams and drip from heat
transfer surfaces.
The standardized procedure for determining fusion temperatures calls
for a reducing atmosphere in the testing furnaces; that is, the ash sample
is heated through the temperature range from initial deformation to fluid
with a slight deficiency of oxygen present—not an excess. In such an
atmosphere, lower oxides are formed with some of the elements because of a
lack of excess oxygen.
It is also important to note that the slag fluid point temperature with
a reducing atmosphere is approximately 100 to 200°C lower than that obtained
with a slag tested in an oxidizing atmosphere.
Slag Viscosity—
The Bureau of Mines has used both synthetic coal ash as well as actual
coal ash to develop slag tapping characteristics prediction procedures from
the slag chemical composition. In Figure 15 the chemical composition of the
slag is used to determine the 250 poise temperature. This is a viscosity
determined through experience to be the maximum viscosity for satisfactory
slag tapping. Experience has also shown that this viscosity should not
occur at a temperature in excess of 1400°C as seen in Figure 16.
Using this technique, the gas purifier slag has a viscosity of 250 poise
at a temperature of approximately 1400°C (2575°F). According to Figure 16,
the point for the gas purifier slag lies well within the satisfactory slag
tapping region.
44
-------
1000
2000
20
50
100
VISCOSITY (POISE)
200
500
1000
Figure 15. A plot of slag viscosity based upon temperature and chemical composition.
-------
1500
CO
01
CO
o
Q.
Og 1400
o W
1300
CO
a:
9 1200
l!00
1100
1200
1300
1400
ASH-SOFTENING TEMPERATURE, °C
(REDUCING ATMOSPHERE)
Figure 16. Slag temperature characteristics for satisfactory slag tapping.
46
-------
A somewhat similar method for calculating slag viscosity based on slag
composition has been developed by the U.S. Bureau of Mines and the.Babock-
Wilcox Corporation. The slag viscosity calculations for bituminous type
coal-ash slag, which is similar to the type found in the gas purifier, are
predicted on the base to acid ratio, the silica/alumina ratio, and the
ferric percentage of the slag using the following equations:
Fe203
Base to acid ratio =
Si0
CaO
a.
NA20 + K20
A1203 + Ti02
Silica/alumina ratio =
Si02
A120;
and;
Ferric percentage = Fe2o3 + ^u'^Tl^S Fe
The gas purifier slag has a base to acid ratio of 0.254 and a silica/
alumina ratio of 3.2. In the absence of analytical data for Fe203, FeO,
and Fe, it is standard practice to assume a ferric percentage of 20. Using
Figure 17 a slag temperature for 250 poise viscosity of approximately 1468 °C
(2675°F) is obtained. This is approximately 40°C (70°F) higher than the
results of the preceding technique.
CO
8
CO
LJ
CO
£
6
ID
CVI
oc
£
LJ
QL
I
a:
ui
a.
to
<
to
1600
1500
1400
1300
1200
1100
0.2 0.4 0.6 0.8
BASE-TO-ACID RATIO
1.0
Figure 17. Slag viscosity based upon chemical compos-ftion.
47
-------
In order to ensure a slag temperature of at least 1410°C (2570°F), the
average operating temperature in the combustion chamber must be 1650°C
(3000°F).
The fluid temperature (reducing basis) and the viscosity of the ash are
design criteria for predicting the slag tapping characteristics. Since the
tendency is to operate slag-tap combustors with low excess air and at high
temperature in the vicinity of the slag bed, more of the iron content of the
ash is in the ferrous form, and hence, the ash is more fluid. However, the
ash viscosity, quite distinct from the ash-fusion temperature may be the
determining criterion in predicting satisfactory tapping of a given ash.
The gas purifier should be operated in such a manner that slag conditions
can be controlled by the oxygen-fuel ratio. Use of the initial deformation
temperature (oxidizing basis) seems most appropriate as a design criterion
for a dry-ash combustion chamber since the ash particles leaving the combustion
chamber should be in a highly oxidized state. The combustion chamber should
be operated to avoid reducing conditions anywhere in proximity to heat-
absorbing surfaces.
If glass and noncombustibles are removed prior to entering the com-
bustion process, the need for slag removal is minimized and more classical
incinerator combustion techniques are appropriate. In this case, the
combustion chamber should utilize the lowest possible gas velocities allowing
the small unremoved noncombustible carry-over fraction to settle out prior
to entering the boiler tube banks. Without question, the approach to be
used must be the result of the total system requirements.
Since the slagging gas purifier is unique to the Baltimore plant, the
plugging of the slag tap hole must also be considered unique to the Baltimore
plant.
BOILER TUBE SLAGGING
Originally, slagging of the boiler was a significant problem. Slag
coating of the boiler tubes thick enough to bridge the tube gaps occurred at
the bottom of the tubes. The slag coating resulted when the spheroids of
semi-molten slag stuck after impinging upon the tubes. As the coating would
become thick enough to insulate the tubes, the slag surface would take on a
slightly glazed appearance. No samples of this type slag deposit were
available for examination at the time of this study.
Monsanto installed quench air ports in the gas purifier discharge duct
to form a more friable deposit. The intent was to cool the slag pa'rticles
so that the majority of the particles would not adhere to the boiler tubes
and the remainder would form an easily removed deposit. The amount of air
admitted did little to prevent adherence of the slag particles, but the slag
deposits were rough as opposed to the glazed appearance seen previously.
48
-------
Two approaches can be taken to prevent the formation of fly ash deposits
on the boiler tubes. The first, is to remove the fly ash through the cyclone
action of the gas purifier. The second approach is to cool the fly ash to
a point where it will not adhere to the boiler tubes.
Fly Ash Removal
The gas purifier design was intended, with its tangential gas inlet and
orifice-baffle wall, to operate as a cyclone and remove a high percentage of
entrained slag cenospheres. The geometric design appears to have been good;
although for cyclone operation, the tangential gas velocity was somewhat
low, approximately 90 to 125 m/sec (300 to 400 fps). The following will
show the inapplicability of the gas purifier configuration to remove the
slag cenospheres without a reentrant type orifice-baffle wall exhaust nozzle
and high gas rotational velocities.
Using average mass flow through the gas purifier of 2423 kg/min
(5353 Ibs/min), a gas purifier chamber cross-sectional area of 19.2 m2
(214 ft2), and a mean gas temperature of 1165°C (2130°F) and a mean gas
molecular weight of 27.824, the mean gas density is 0.2355 kg/m3
(0.0147 lb/ft*). The drift velocity through the gas purifier chamber is
518 m/min (1700 ft/min). Therefore, the mean residence time of a particle
is approximately 1.7 sec. Initial tangential velocity of gas through the
crossover duct can be calculated in a similar manner.
m, = 1411 kg/min
Crossover duct cross section exit area is * 3.08m2
Gas mean molecular weight = 28.28
Gas mean temperature = 798°C
Then the gas density (p) is
p = 0.32 kg/m3
and the initial tangential velocity is
Vy = 1462 m/min
Flow helix angle is approximately 19.5° from a plane perpendicular to
the gas purifier axis.
The gas viscosity, assuming that air approximates the gas sufficiently,
is 0.051 centipoise or 5.1 x 10"1* poise. From the mean tangential velocity,
the angular velocity of the gas'is 958 radians/min. The normal acceleration
field of a particle entrained in the gas is 255.1 x R m/sec2. The settling
49
-------
velocity for a particle, based upon Stokes law, is:
V = 2/9 r2g Sl " S
2.
V
where Sx (the specific gravity of the solids) equals 2.36 and
S2 (the specific gravity of the gas) equals 3.24 x 10 **
Then
V = 2.623 x 10~3 r2 R
where r is the particle radius in microns, V is the settling velocity in
m/sec and R is the particle location in meters from the rotational axis.
Then the particle acceleration A = dV/dr or A = 2.623 x 10 3r2. Consider-
ing a 20 micron diameter particle, A = 0.2623 for a residence time of
1.7 sec. From D = 1 A t2 + V0t, whire D is the initial particle distance
from the gas purifier wall, the result is 0.379 m. This indicates that a
particle 20 microns in diameter must be within 0.379 m of the gas purifier
wall or it will not be centrifuged from the gas stream.
However, all particles 52 microns in diameter or larger should be
removed from the gas stream.
Proper design of a cyclone type combustion chamber would provide an
inlet and tangential velocity of at least 107 m/sec (350 fps). Assuming a
chamber having a 3.05 m (10 ft) diameter and a 7.63 m (25 ft) length which is
water cooled, then the maximum "g" field would be 7513 m/sec2 (24,650 f/sec2).
The "g" field equation is defined as
A = 4.923 x 1Q3R
n
Assuming the same mass flow as presently used, the axial drift velocity
for a gas temperature of 1650°C (3000°F) and a resulting gas density of
0.973 kg/m3 (0.011 lbs/ft3) is 31.9 m/sec (3-05 fps). A residence time
period of 0.24 seconds results from the assumed chamber length of 7.63 m
(25 ft). In a cyclone combustion chamber such as described, the flow helix
angle would be approximately 16.6 degrees from a plane normal to the chamber
axis. For the temperature noted, the gas viscosity would be approximately
5.86. x 10 ** poise. From Stokes law, the mean velocity is:
V = 4.4 x 10~2 r2 R
from
A = dV dR
then P
A = 4.4 x 10 2 r2
which yields
D = 2.2 x 1Q~2 r2 (t2 + 1.159t)
50
-------
If the residence time of 0.24 seconds is used, the D, the distance from
the wall to be traveled by a particle of size r, is
D = 7.388 x 10~3 r2
Evaluating D for particles having an r = 10 micron then
D = 0.739 m
This means that any 20 micron particle entering the cyclone within
0.74 m (2.4 ft) of the wall will impact the wall without the influence of an
orifice exit nozzle. With the reentrant effect on an orifice type exit
nozzle, the actual particle residence time is increased, temperature
stratification effects of the high "g" field will cause a lower gas viscosity
closer to the walls, and as a result of the higher settling velocities,
particles down to approximately 5 micron will be eliminated from the cyclone
exhaust. Additionally, the cyclone combustor as described previously is a
lower maintenance item typically having an availability of better than
98 percent on line with power utilities including regularly scheduled annual
maintenance.
An example of the accuracy of this analysis is seen in Figure 18 which
is a pair of electron microscope photographs of agglomerated fly ash samples
removed from the boilers. In Figure 18, one cenosphere is seen to have a
diameter of approximately 70 micron with the remaining particles ranging from
5 micron to 45 micron with the most numerous size being approximately
20 micron. Figure 19 is a pair of optical microscope photographs of a loose
fly ash sample at 10* and 100*. The particles are all cenospheres showing
some agglomeration with the most common particle size being 20 micron.
The remaining fly ash not removed must be prevented from adhering to the
boiler tube. Two obvious methods are: (1) remove the slag fraction from the
process stream at some point before it reaches the boiler tubes or (2) cool
the slag cenospheres to a temperature below their initial deformation point
prior to reaching the boiler tubes. Initially, when slag accumulated on the
boiler tubes, the gas purifier duct was modified with quench air ports to
cool the slag particles. This was partially successful, but at the present
time slag still tends to accumulate. Therefore, since at the present time it
is impractical to remove the slag fraction from the process stream, more
effective cooling techniques are needed.
Fly Ash Cooling
Surface cooling modes for fly ash particles are radiation and convection.
There is only a relatively short distance of approximately 9 m (30 ft)
from the gas purifier discharge duct quench ports to the first set of boiler
tubes. The slag particles do impinge on the boiler tubes but do not fuse.
This is indicative of a slag particle temperature approximating the initial
deformation point. Heat transfer calculations indicate that radiative heat
transer to the duct walls exceeds conductive or convective transfer to the
cooled or "quenched" gas stream by several orders of magnitude. The
51
-------
21 X
1O5 X
Figure 18. Scanning electron microscope pictures of fly ash.
-------
-
1O X
1OO X
Figure 19. Optical microscope pictures of fly ash.
-------
conclusion drawn from this is that the quench air admitted to the gas purifier
duct would be more effectively utilized if it were used to cool the duct
walls. If the duct walls were cooled to some 50°C below the initial de-
formation temperature the slag particles should not adhere to the boiler
tubes or could be removed from the boiler and/or ducts through the use of ash
blowers. Another technique used by the power industry to protect boiler
tubes from collecting ash deposits is to install slag screens in front of the
first boiler tube rows. The screens are not screens in the normal sense of
the term but instead consist of up to six rows of water-cooled tubes.
Generally, the slag screen tubes across the boiler inlet are on 25 cm (10 in.)
centers, and from front to rear on 12 cm (5 in.) centers. Alternate tube
rows are staggered with a resulting minimum spacing of 19 cm (7.5 in.).
As the slag particles pass the cool tubes, their temperature is reduced
by radiative heat transfer to the point that they will not adhere to the
downstream boiler tubes.
Soot blowers must be used to remove the resulting fly ash from areas of
repose since, if this is not done, the fly ash will, over a short period of
time, slowly sinter into a consolidated mass of considerable strength,
depending upon sintering time, temperature, and the ash characteristics.
The decision of the City to use a nonslagging afterburner should alleviate
the slag build-up problem since slag softening temperatures will never exist.
An analysis of the problem reveals that this problem is unique to the
Baltimore plant since the prototype Landgard unit had a nonslagging after-
burner .
54
-------
SECTION 4
CONVEYORS
Since the plant operation requires that material flow continuously
through all processing equipment, the plant must be shut down whenever any
one of the principal conveyors fails. The principal conveyors are the
storage pit and the shredder feed conveyors, the screw conveyors, and the
residue quench tank drag conveyor.
STORAGE PIT AND SHREDDER FEED CONVEYORS
Because these conveyors initiate the flow of refuse to all the down-
stream processing equipment, the shutdown of these conveyors would cause a
total plant shutdown.
Shear pin failures, link failures, and the shredder interlock have
caused the numerous shutdowns of these conveyors.
In the original equipment design there were several regions in which the
material could accumulate, build up, and eventually jam, causing shear pin
failure. This problem was particularly severe at the transfer point between
the horizontal pit conveyor and the shredder feed conveyor. The conveyors
are so close that the pusher bars on the horizontal conveyor tend to capture
material that has just fallen onto the shredder feed conveyors and push it
backward into the recessed pit below the conveyors where it can accumulate
and eventually cause jamming. A better approach for any new system, but one
which would be impossible to retrofit, would be to increase the vertical
separation of the hopper pit and shredder feed conveyors with a slide trough
between them to direct the waste onto the shredder conveyors without spillage.
It would be necessary to have sufficient clearance between this trough and
the hopper pit conveyors so that the returning pusher bars could not force
material into the recessed pit area.
Jamming also occurred at the return run of the shredder feed conveyor
where waste would accumulate. This problem was solved by installing a drop-
out chute on the return run of each conveyor which discharged onto the
shredder discharge conveyor.
On one occasion, improper shear pin installation resulted in major
system failure when a solid steel object jammed in a pinch point of a unit.
The shear pin did not fail and when one of the conveyor links failed, the
drive motor continued to pull the pans through the gallery so that major
55
-------
damage to the pans, links, and shafts occurred. This could have been pre-
vented if the manufacturers had allowed for sufficient access for installing
cotter pins on both sides of the shear pin.
The accumulation and wedging of noncompressible materials between
sprockets and drive chains caused a cantilevering of chain links as they
passed over the sprocket. Any link design would ultimately overstress and
fail as a result of such a cantilevered motion. The links of this particular
conveyor system are particularly susceptible to such stress failure. Link
failure of this sort on one side of the conveyor does not result in sufficient
force to cause shear pin failure. Instead, when one link fails it tends to
cause other links down the line to become overstressed and also fail so that
a simple failure cascades into a major system failure with extensive loss of
operating time.
The problem appears to have been solved by enclosing the head sprocket
and chain within a steel guard to prevent material from accumulating between
the sprocket and chain. Conceivably, material could accumulate in the links
and be carried into the sprocket area under the steel sprocket guard, but the
rubber aprons now installed seem to suffice for preventing material side
spillage and accumulation.
These conveyors have failed to uniformly load the shredders causing
frequent overloading of the shredders with the subsequent shutdown of the
conveyors. Leveling bars were added to both pit conveyors in an attempt to
maintain a uniform waste load on the conveyors. These bars caused jams and
subsequently resulted in shear pin failure.
The storage pit and shredded feed conveyor failure are considered to be
typical of solid waste processing since the equipment is typical of solid
waste processing systems and the failures were due to equipment design.
System vendors are currently modifying .their designs to reduce these problems.
SCREW CONVEYORS
When the screw conveyor system or any'part of it fails, the entire plant
must be shut down to prevent the accumulation of slag frit in the seal tank
which would quickly plug the slag tap hole.
Originally, the spillback conveyor and the slag frit conveyor were one
unit which ran the entire length of the seal tank bottom and were driven by
a single motor. When the screw flexed severely during operation, the screw
was divided at the intersection of the transfer frit conveyor, and hanger
bearings and drive motors were installed. The hanger bearings were not
located on the end of the conveyor and blocked the flow of the large un-
expected slag balls. The hanger bearings were subsequently moved to the end
of the conveyor.
The screw conveyors sheared pins regularly, and the strength of the
shear pins was greatly increased to keep the system operating. However, the
increased shear pin strength prevented the performance of a motor protection
56
-------
function and the system was often inoperable due to motor overload. After
the motor size of each conveyor was increased, the screws of the slag frit
and the transfer conveyors were severely damaged and were replaced.
The spiral edges of the screw conveyors were thickened and the screw
pitch was decreased from 45 degrees to approximately 10 degrees. With the
decrease in pitch of the screws, a more dense, more compact material could be
moved without damage to the screws or motors or constant shear pin failure.
Overloading still occurs, and it is expected that a larger motor is required
for the transfer conveyor. Since the design changes, pin shearing and motor
overloading continued but occurred less frequently. The City of Baltimore is
planning to eliminate these conveyors by using a nonslagging afterburner.
Most of the failures were caused by the large slag balls discharged from
the slag tap hole and the extreme underestimation of the slag bulk density.
Large, unfritted slag balls conveyed along the bottom of the tank would jam
against the side causing the tank to flex slightly and the shear pins to
shear. It was then necessary to clean out the tank and replace the shear
pins. Another reason that the original screw conveyors failed was because
the original design bulk density was half of the actual bulk density.
Since the slagging gas purifier was unique to this specific installa-
tion, the screw conveyor failures are unique to the Baltimore plant.
RESIDUE QUENCH TANK DRAG CONVEYOR
Whenever the single drag conveyor to discharge the solid residue from
the quench tank in the kiln fire hood failed, the entire plant had to be shut
down because of the single-line operation of the thermal processing area.
In the early operating period of the plant, the drag conveyor frequently
failed, jammed, or overloaded when large pieces of fused slag were discharged
from the kiln. Although not related to the design conditions for the conveyor,
large pieces of fallen kiln refractory also jammed and overloaded the conveyor.
In the original design, the return run of the drag conveyor was within
the quench tank with all slag particles and other residue expected to fall
through the apertures between the drag flights of the return run and then
onto the tank floor where the flights of the drag run would push the residue
up a dewatering incline to a discharge point. Consequently, oversized slag
balls jammed the conveyor by lodging between the return and drag run flights
or by lodging in a return run aperture to be carried along by the flights and
chains to the tail sprocket where it would wedge at the quench tank wall. In
one instance, a slag ball lodged in a return run aperture and pulled off the
entire tail sprocket assembly. In addition, large balls falling freely onto
the return run damaged the drag flights and overloaded the conveyor system,
particularly, as mentioned above, the shear pins and the chain links. As the
chain links stretched with the constant overloading they had to be adjusted
regularly to prevent them from jumping the head sprockets and to facilitate
smooth start-up.
57
-------
Consequently, the drag conveyor system was redesigned so that the
return run would pass over the kiln fire hood and then down to the original
tail sprocket. In addition, the chains and drag flights were strengthened
and the motor size was increased. Then, after residue began accumulating on
top of the kiln fire hood as the wet residue adhering to the flights and
chains dried and fell off during the overhead pass, notches were cut into the
drag flights to minimize the subsequent spillage and to prevent cans from
wedging between the chains and the head sprockets.
The foregoing redesign and subsequent modifications remedied most of the
conveyor malfunctions. While pin shearing was minimized and the chain links
generally remained intact, the drag flights still required extensive repair
and replacement. Although the drag conveyor tension still had to be adjusted
regularly, the adjustment was a minor effort.
Some of the conveyor failures could have been caused by the high bulk
density of the residue which was twice the design value. However, most of
the failures occurred when large residue slag balls or other large objects
such as kiln refractory were discharged from the kiln. One mode of failure
due to these large slag balls was described earlier. The other mode of
failure, which continues to cause this conveyor to fail, results from the
large concentrated loads that these slag balls put on the conveyor. The drag
conveyor was designed for a uniform loading of.all the flights rather than a
concentrated load on one flight. While individual slag particles were
expected to have an average size of about 10 centimeters (4 in.) with some
occassionally up to about 60 centimeters (24 in.) in length, slag balls as
large as 1 by 2 meters (3 by 6 ft) were seen in the residue quench tank.
Since the drag conveyor failures were caused primarily by the larger-
than-expected slag balls and a facility could be designed to handle such
residue, these failures were considered to be unique to the Baltimore plant.
58
-------
SECTION 5
RAM FEEDERS
The ram feeders advance refuse into the kiln for thermal processing and,
if the ram feeders are inoperative, the entire plant must be placed on
standby or shutdown. The ram feeders have experienced ram jams, snout
deformation and failure, and hydraulic hose damage in their short operating
history.
RAM JAMS
Refuse feeder jams are an extremely critical problem because if a jam
cannot be cleared the facility must be shut down to allow the jam to be
cleared manually from inside the kiln. In addition, when feeding of refuse
through the tubes is halted for any reason, there is the possibility of fire
in the refuse plug since the feed tubes are not cooled and the refuse plug is
in contact with the hot tube walls.
Ram jams have typically occurred in three locations: the end of the
feeder tube snouts, the tube air seal deflection plates, and the feeder tube
entrance.
Ram jams are the result of some obstruction to the smooth flow of
refuse in the tubes and the inherent compactability of refuse which allows
refuse flow to be stopped at some point and further refuse feeding to compact
the initial plug to the point where the feed tube is jammed. Ram jams which
originate at the tube discharge ends are generally believed to be the result
of tube end deformation.
Some ram jamming occurred at the internal deflection plates which were
installed in the feed tubes to prevent air leakage through the refuse plug
into the kiln. Each tube originally contained two metal wedges which were
intended to compress the refuse plug as it progressed through the feed tube.
The intent was to provide a progressively denser refuse plug by decreasing
tube area as the refuse passed the wedge apex thereby reducing the leakage of
air through the refuse plug. Although the wedges did perform the assigned
task, they provided an obstruction which would periodically cause a jam in
the feed tubes. Removal of the wedges naturally removed this source of
jamming but also allowed a higher leakage air flow rate.
Jamming of the rams at the inlet to the feed tubes also occurred. This
type of jamming was originally caused by refuse overloading one ram hopper at
the inlet. Subsequently, a flow splitter was installed to evenly distribute
59
-------
the refuse between the two ram hoppers. This reduced the frequency of jamming
but did not eliminate it. The remaining cause of ram jamming at the inlet to
the feeder tubes can be attributed to a combination of the compactible nature
of the refuse feed stock and the feed tube inlet geometry. The inlet geometry
must be changed to remove the short, steep angle of the entry ramp where most
jamming occurrs. It is felt that improvement in operation could result from
changing the entrance configuration to one where the large piston or ram
enters the tube a short distance rather than stopping even with the forward
wall of the hopper. The entry should be a shear configuration with clearance
around the ram as it enters. This configuration would appear as a piston
entering a cylinder. At the end of the short cylindrical tube section, the
feeder tube would have a gently tapered cone shape transitioning from the
inlet cylinder to the final feed tube cylinder. A modification entailing
considerable expense and difficulty would be to provide access to the feeder
tubes for clearing ram tube jams which will, in all likelihood, continue to
occur although the inlet configuration may substantially reduce the frequency.
Because rams have effectively been used to feed solid waste to other
systems, the Landgard prototype being one of them, this problem is believed
to be specific to the geometry of the equipment installed at the Baltimore
plant.
SNOUT FAILURE
Failure of the ram feeder snouts is considered to be a critical problem
because of the resulting ram jams and equipment damage which require a
complete shutdown of the process. Only when the kiln can be entered can the
ram jam be cleared and the%feeder tubes be repaired.
The earliest ram feeder tube or snout failures were the most damaging.
The original feeder tubes or snouts were constructed of stainless steel and
were water cooled. During the first heat cycle, a heavy hammering was heard
coming from the feeder tubes. The tubes had ruptured at the tube wall seams
and spilled the cooling water onto the refuse and hot refractory of the kiln.
The tubes were replaced with new uncooled tubes of 316 stainless steel. The
uncooled tubes rapidly deformed in the high temperature environment of the
kiln resulting in ram jams on numerous occasions.
The cooled tube failure experience appears, by the manner in which the
equipment behaved and nature of the failure, to have been the result of
several problems. Most obviously, there was a manufacturing defect which
probably had two elements. The first was one or more misplaced baffles short
circuiting the available water supply. The second was a cooling passage
cavity with an obstructed inlet or outlet effectively becoming a partially
sealed area. As a result, water was allowed to fill the cavity and be turned
to steam. When the steam pressure subsided, the cavity would again fill with
water to restart the cycle. An operating problem was also found to exist
complicating the situation. The cooling water flow was supplied from an
undersized main, but the extent of contribution of this situation cannot be
evaluated.
60
-------
Problems encountered with the uncooled feed tubes had their origin in
two areas. First, the material specified had only marginal strength and
second, the kiln temperatures experienced were higher than anticipated
resulting from process control problems. The use of a rolled alloy did
provide some-strength margin as did the ribs and stiffeners.
The discharge ends of the tubes were eventually cut to a 45° angle to
reduce the occurrence of ram jamming. The internal deflection plates used to
produce an air seal with the extruded refuse were also removed for the same
reason. The feeder tubes or snouts presently operate in spite of severely
deformed discharge ends but do not produce good air seals and will require
eventual replacement. The result of the high air leakage rate is excessively
high feed hood temperatures which contribute to rapid refractory deterior-
ation and continued feeder tube distortion as well as volatilization of heavy
metal fractions and poor process control.
Returning to a watercooled set feeder tubes will accomplish two
objectives. The first is to provide a feeder tube capable of surviving the
high temperature environment and, the second is to provide a duct cool
enough to prevent heating of the refuse plug to the point where it can
ignite when refuse feeding is stopped for any reason. There is no reason
that a water-cooled tube cannot successfully survive in the kiln environment.
Water-cooled exhaust nozzles and combustion chambers have been success-
fully operated for many years in environments where the heat flux is at least
an order to magnitude higher than that experienced in the kiln. Care in
design and fabrication should prevent failures of this nature from occurring.
The tube failures experienced are considered to be a characteristic
unique to the Baltimore Plant. The reasoning for this is that by replacing
the tubes with water-cooled tubes having proper design, accurate manufacture,
and an adequate coolant supply, failures should not occur.
HYDRAULIC SYSTEM
The ram feeder hydraulic system is the heart of the kiln refuse feeder
system. Any disabling problem encountered by the hydraulic system will stop
operation of the ram affected. Normally, a disabling problem can occur to
one ram and still permit continued processing operatings for a short period
of time with the second ram while repairs to the first are made. A failure
in the entire hydraulic system would precipitate a plant shutdown.
The hydraulic fluid was originally routed to the hydraulic cylinders
through hard wall tubing having fitting type joints between tube sections.
The frequent shift from normal pressure to high pressure (10,350 kPa to
20,700 kPa) to clear ram jam represented a shock loading causing leaks.
Frequent leaks from split tubing and ruptured fittings were continuous
problems during system start-up. Changing the fluid lines to flexible hose
prevented tubing fatigue but the hoses and connecting fittings still remain a
problem. Frequent leakage at the pump and actuator cylinder seals is a
source of high'cost for lost hydraulic fluid and seal replacement maintenance.
61
-------
The hydraulic fluid used in the system is Monsantofs Pydraul 50-E. This is a
high-temperature capability, synthetic, ester base fluid. This hydraulic
fluid is typical of ester base fluids, requiring stringent control of the
materials in contact with it. The aviation turbine industry has a long
history of successful experience with these types of hydraulic fluids and,
hence, their use cannot be considered unique or experimental. Material
compatability is definitely acknowledged to be a concern, but seal materials
such as teflon, viton, and other florinated hydrocarbons have been used with
complete success. Comments by technicians and maintenance men to the effect
that the fluid "eats up the seals" is typical of the reaction occurring when
normally available commercial seals, which are noncompatiable type materials,
are used in contact with ester base oils and hydraulic fluids.
It seems highly likely that the ruptured hoses are the result of non-
compatible materials, since the use of such hoses with similar hydraulic
fluids is standard operating procedure in the aviation turbine industry. No
information was readily available to determine the flexible hose material
composition, therefore, evaluation of the compatibility of the hose and fluid
was not possible.
The situation is compounded by the ready access of glassy grit to the
seals and seal surfaces. Such materials enter the actuator areas with ease
and, when in contact with the wetted seal surfaces, stick and rapidly abrade
the seals.
Several alternatives are possible means of improving service life.
Clearly, only compatible seal and hose materials should be used during
maintenance, and special care should be taken to ensure installation of only
those type materials.
Abrasion of hoses and seals can only be prevented by eliminating the
glassy grit from contact with those parts by either protective devices or by
separating the glassy fractions from the refuse before it has entered the
feeder.
Failure of hydraulic hoses and seals as well as pumps and other com-
ponents of the hydraulic system must be expected at least on an infrequent
basis. Because of this, it is only realistic to provide safe access to all
parts of the hydraulic system for repair and maintenance while operation of
the second system continues. At the present time, this is not possible.
Elimination of the causes of feeder tube jamming would greatly extend the
life expectancy of all components by eliminating the need for the high
pressure override to clear jams which cause shock loading.
Problems which have occurred concerning incompatability of materials can
be considered typical of any hydraulic system where sufficient care has not
been taken to assure use of compatiable materials. This problem can occur in
any industry or usage. Although the component damaging hydraulic shock
loading used to clear jams results from the basic characteristic of refuse,
the specific feeder design enhances this problem in the Baltimore Plant. The
presence of glassy grit in municipal refuse should be expected unless the
refuse is processed to eliminate this fraction.
62
-------
SECTION 6
STORAGE AND RECOVERY UNIT
While failure of the storage and recovery unit will not immediately
cause a plant shutdown because refuse processing can continue using the
bypass chutes, these chutes have a tendency to jam and are not considered
reliable. More importantly, the storage and recovery unit has not been able
to sustain the rated output capacity, thereby decreasing the throughput and
increasing the net processing costs.
Shortly after start-up it became apparent that it would be difficult to
recover the refuse stored in the storage and recovery unit. Many modifications
were made in attempting to solve this problem: (1) the bucket sweeps were
operated at high speeds, (2) teeth were added to the buckets to dig into the
pile better, (3) the four 10-bucket chains were changed to three 20-bucket
chains to allow more buckets to contact the pile, (4) the size of the center
cone was also increased to allow more buckets to contact the pile, and (5) a
rotary distribution chute was added to the top of the unit to distribute the
waste evenly around the center cone.
While these modifications improved the recovery somewhat, the unit did
not operate satisfactorily. In addition, the high operating speed caused
excess wear of the bucket shoes and the floor of the unit. Because of these
problems, the city made some major modifications: (1) a large door was
installed to allow access by a bulldozer, (2) the bucket chains and the
center cone were removed, and (3) the distribution chute was set to discharge
directly to the outfeed conveyor. After numerous jams occurred on the out-
feed conveyor, the city terminated the use of this equipment.
The basic cause of these problems was a lack of understanding during
system design of actual service conditions which led to improper choices of
design parameters and materials specifications. The design density of the
shredded refuse was set at too high a value which led to the selection of a
live bottom design that was not capable of providing the desired feed rate at
the design operating parameters. Attempts to compensate for this by running
at four times the design bucket speeds greatly increased equipment wear.
The decision to store and process unclassified, unseparated refuse
contributed to the operation problems by (1) introducing large quantities of
fine glass particles into the unit where they tended to filter through to
the floor causing extreme abrasive conditions and excessive wear on the floor
and the bucket wear shoes, (2) allowing wire and large pieces of textiles
which went through the shredders to act as reinforcements tying the refuse
63
-------
together into a mass that was difficult to tear apart for feeding, and
(3) preventing any control of moisture content in the shredded refuse.
The installation of the large cone in the center of the silo to move the
pile out from the center for better bucket contact was a good decision.
However, the,\choice of the cone shape was poor since the sloping sides of the
cone tend to support the matted pile instead of causing it to drop. This is
especially true when wire and textile pieces tie the pile together. A
cylindrical shape with a short cone section at the top would be more effective
in causing the pile to drop.
The floor was constructed of concrete with standard limestone aggregate
with surface dusting of metal particles. The concrete had a compresion
strength of 27,600 kPa (4000 psi) instead of the specified 41,400 kPA
(6000 psi). This type of floor has proven itself to be completely inadequate
for the abrasive conditions and high bucket speeds encountered in this
operation.
The original design specified a refuse bulk density of 400 kg/m3
(25 lbs/ft3) and provided four chains of 2-ft buckets, a bucket speed of
approximately 13.7 m/min (45 fpm), 2 sweep drive motors, and a single dis-
charge conveyor across the diameter of the floor. The measured bulk density
of the refuse has run 200 kg/m3 (12.5 lbs/ft3) and less. To solve the
problem of an inadequate recovery rate, the silo manufacturer has recommended
the installation of from four to six radial collection conveyors, bucket
chains of 4-ft buckets, bucket speeds of 13 m/min (43 fpm) or less, and
4 sweep drive motors. Adoption of these recommendations should improve the
feed rate and reduce the wear.
To solve the floor wear problem, the equipment vendor recommended a
surface of flint aggregate in an epoxy binder which would probably not be
adequate in this operation. Experience with this topping at the Ames, Iowa,
facility shows that the epoxy is quickly picked out by the abrasive glass
particles and the brittle aggregate is easily crushed.
*
The wear resistance of the floor can be improved by restoring the
concrete to a level 2.5 cm (1 in.) below the working level and topping it
with at 2.5 cm (1 in.) layer of Anvil Top®, a product of the Master Builders
division af Martin-Marietta Corporation. This material has been in use in
industry since 1952 and has demonstrated exceptional resistance to abrasion
and impact. Anvil Top 929, which can be poured, finished, and cured the same
as concrete, consists of 90 percent cast iron aggregate and develops
82,800 kPa (12,000 psi) compression strength.
Tests conducted at the Baltimore facility showed that wear shoes made of
Astralloy provided greatly improved service life for bucket wear shoes.
These should be used on all buckets.
A very critical item in reducing bucket shoe and floor wear is the
removal of glass from the refuse prior to storage of the refuse in the
storage and recovery unit. Since much of the glass is pulverized into small
particles by the shredders, it makes an ideal grinding medium when allowed to
64
-------
collect on the floor. The fine glass particles also cause high wear rates in
the bearings of the ring wheels and other moving parts.
The fact that much of the glass is so small makes removal a fairly
simple matter. A trommel, with screen openings of 19 to 25 mm (0.75 -
1.0 in.) would remove a large percentage of the glass particles when located
between the shredders and the storage and recovery unit. Additional benefits
of glass removal would be a large reduction in slagging problems in the kiln,
gas purifier, ducts, and boiler inlets and a possible reduction in the fly
ash particles in the stack gases.
Introduction of a counter-current flow of air through the trommel would
facilitate dust collection and reduce the moisture content of refuse.
Moisture measurements, taken on the feed, varied from 8 percent to
40 percent. A moisture content of 40 percent is approaching that of some
dewatered sludges and is a condition which tends to promote packing or
matting and creates problems in moving material easily. On the other hand,
a low moisture content usually results in a material that tends to stay loose
and move easily.
Using a trommel before shredding is another alternative which has the
additional advantage of reducing shredder wear and the amount of waste
requiring shredding. The operation of the trommel may, however, increase the
net processing costs.
In summary, the performance of the storage and recovery unit can be
improved provided:
1. The live bottom is reworked to incorporate the recommendations
of the manufacturer with at least four conveyors and a 2.5 cm
(1 in.) topping of Anvil Top.
2. The shredded refuse is trommeled prior to storage to remove
abrasive materials and to control moisture.
This unit is used in many other refuse handling facilities, but since
the Landgard process is the only process where this unit is used to store
primary shredded, mixed municipal solid waste, the problem must be considered
typical of the Landgard process.
65
-------
SECTION 7
RESIDUE SEPARATION MODULE
A shutdown of the residue separation module does not have a direct
effect on the operation of the rest of the system since a bypass gate allows
the residue to go directly from the residue drag conveyor into trucks for
disposal. However, the problems encounted in this module were so numerous
and extreme that a discussion of them is justified.
The operating history of this module is very limited due to the short
period of time that it was operated. In general, the equipment did not
separate the residue cleanly and required considerable modification. This
module has been dismantled by the city and the residue will be be sent to the
landfill without separation.
Design deficiencies caused a number of operational problems; most of
these deficiencies have been corrected. The chain drag in the flotation unit
was designed to push rather than pull the bottom material. Adequate access
was not provided for cleaning and maintaining the flotation unit, and
additional access doors had to be cut in the side of the tank. Many of the
flumes, overflow wiers, and conveyors were undersized. Inadequate drains
were provided for various tanks and no clean-out provision was provided in
pipes carrying slurries. The sludge line from the thickener frequently
plugged until a recycle line was installed. A nonweatherproof motor was
mounted outdoors. A grit settling box had no provision for removing collected
grit.
The main cause of problems in the residue separation module was the lack
of satisfactory process control in the kiln which results in the residue
being either overprocessed or underprocessed. Overprocessing results in high
temperatures and the formation of refuse slag balls which jam or damage the
equipment. Underprocessing allows materials such as rags and plastics to jam
and blind pump screens. Indications are that when the pyrolysis process is
operating properly the residue can be handled by the separation module with a
minimum of problems. Clumps of wire and crushed cans, which survived the
shredder, still present problems. The materials which constitute the major
portion of the residue contribute heavily to the problems of good process
control in the kiln which, in turn, cause many of the residue separation
problems.
The tendency of the glass in the trash to soften and form slag balls at
relatively low temperatures contributes to process control problems by
making it more difficult to achieve good pyrolysis without slagging. In
66
-------
addition, the pyrolysis process lowers the resale value of separated metals
significantly, reducing revenue potentials proportionately. Fine glass and
metal particles in the residue are highly abrasive and cause wear problems
throughout the system. Wire, crushed cans, and slag balls cause jamming and
damage to equipment and plugging of pumps and pipes.
The best solution to the problem is to separate metals and glass from
the trash prior to pyrolysis. Pre-pyrolysis separation would necessitate a
redesign of the residue quench tank to provide removal of floating char which
presently sinks due to the weight of the glass and metal in the residue. The
benefits of pre-pyrolysis separation would include:
1. Improved process control in the kiln.
2. Greatly reduced erosion/abrasion in pumps, pipes, duct work,
the storage and recovery unit, the induced draft fran, and the
refractories.
3. Greatly reduced slagging and slag build-up throughout the
thermal processing and energy recovery modules.
4. Potential reduction in particulate emission.
5. Increased revenue from separated materials.
Because the residue separation module is unique to the Baltimore plant,
so are the problems of operating the module.
67
-------
SECTION 8
SYSTEM PRESSURE OSCILLATIONS
The thermal processing system ,is designed to operate at sub-atmospheric
pressure to prevent the escape of process gases to the atmosphere. Because
of the design and the size of the system, internal positive pressure is
dangerous to both nearby personnel and equipment. Thus, when pulsating or
oscillating pressures occur in the system it must be shut down rapidly to
stop the oscillations. For this reason, the periodic instability of the
pressure in the system is considered a critical problem. Additionally, the
resultant rapid decrease in temperature during an emergency shutdown damages
the refractory linings of the kiln and gas purifier.
During the facility shakedown, dramatic pressure oscillations were
experienced with spectacular exhibitions of flame from the many ports and
seals. A separate study of the pressure oscillations was instituted by
Monsanto and the City to discover the origin of the oscillations. Results of
the study are considered proprietary by Monsanto, but several pressure
transducer data traces were supplied for this study. Figure 20 is a data
trace of a sinusoidal pressure pulsation in the kiln.
[«—/ stc.—*\
<-
-t'tMC
Figure 20. Data trace of sinusoidal pressure pulsation in the kiln.
Random frequency pressure oscillations of low magnitude (Figures 21
and 22 are of such high frequency as to be uncontrollable in a system having
the internal volume of the Baltimore plant. At the same time it might be
noticed that the magnitude is of little concern other than to indicate basic
instability or at least neutral stability. Figure 23 is typical of the
pressure oscillations which could be considered dangerous. The steep pressure
68
-------
^i.l.J-_l-
!• T k i-
Y -_"yv- UTL _LV_!_T|! '_^.
•--;-•:--: -,-|; .|.:;;n-:-:
I::M I
—]— rrr—
see.
Figure 21. Data trace of random frequency pressure oscillations.
Figure 22. Data trace of random frequency pressure oscillations.
Figure 23. Data trace of dangerous pressure oscillations.
69
-------
ramps indicate that control of this phenomena after the fact is unlikely and
the only cure is to remove the source. The data also indicate that the kiln
is probably not acting as an organ pipe resonator. Calculations of the kiln-
feed/fire hood configuration indicate a resonant Helmholtz cavity frequency
of approximately 5 Hz. The resonance frequency of the pressure traces
appears to be approximately 2.2 Hz. The Helmholtz resonant frequency is a
very rough approximation since in reality it is difficult to determine what
part of the kiln volume might act as part of the resonator volume. Since the
volume of the resonator enters the resonant frequency calculation in the form
V *£, the greater the apparent volume, the lower the calculated resonant
frequency.
Figure 24 shows traces of pressure at various locations in response to
changes in the crossover combustion air and induced draft fan damper positions.
These are the only pressure traces available from Monsanto which document
pressure response time between the various locations in the system. As can
be seen, there is no apparent response delay. The reason for this is the
trace speed. Assuming a 425 to 500 mps (1400 to 1600 ft/sec) acoustic
velocity, a pressure wave should require only 0.06 to 0.07 sec to pass
through the system. The data resolution is not sufficient to accurately
identify such time delays.
Monsanto concluded that this type oscillation is not controllable and is
related to operation of the oversized crossover combustion air fan which is
no longer used.
It has been noted that the crossover combustion air fan was oversized
for its task. Typically, centrifugal fans under partial load conditions can
develop instabilities. This generally occurs at 1/6 to 1/3 of design volume
flow. When a fan discharges into a duct where the volume content is 5 percent
or more of the design fan discharge rate, then severe pulsations can occur
when the fan is operating at any portion of the characteristic curve where
there is a decreasing pressure with decreasing volume flow. This appears to
be one cause or source of a forcing function driving a Helmholtz resonator
cavity configuration.
At the present time the severe pressure oscillations appear to be under
control by not operating the crossover combustion air fan. This problem is
not unique to the Landgard process nor is it only typical of refuse processing.
It is a problem typical of fan-driven systems and, as such, could occur in
any facility where fans are employed.
70
-------
71 VLT ~...;.. :•
_750421n. _!_-; /-
-i |- •• i i ' TH ! ;'~T • I
t£mff
m^i(ttf§fa^g^^ i:
• -^ •-'= ~
Figfire 24. Data traces of pressure at various locations.
71
-------
SECTION 9
KILN PROCESS CONTROL
Kiln process or temperature control will not directly cause a plant
shutdown, but poor process control can cause a range of secondary problems
which will precipitate a plant shutdown. High kiln temperatures have caused
kiln refractory degradation and kiln slagging with the formation of oversized
slag balls which can jam or damage the drag conveyor. Incomplete and un-
processed refuse overloaded the discharge conveyor. Proper processing of
refuse in the kiln is 'probably the most important aspect of reliable operation
of the facility.
The thermal processing at the Baltimore plant is actually starved-air
incineration rather than true pyrolysis. This process requires that close
temperature and combustion air control be maintained to produce a consistently
high quality gas product and a trouble-free residue. The original design
specified that an 02/C02 analyzer system be used to determine the proper
fuel-air ratio and thermocouples to measure temperature. Temperature is an
insufficient control parameter to determine proper refuse processing because
the peak temperature occurs at the stoichiometric fuel-air ratio. A fuel-
air ratio either greater than or less than stoichiometric could result in
lower temperature. Hence, knowledge of the fuel-air ratio is also needed to
properly control the process.
The pulsating flame front and high temperature combustion zone at the
kiln fire/end common at the Baltimore plant were not experienced in the
prototype unit. The difference in combustion behavior was reasoned to be the
result of differences between the prototype and the Baltimore plant. Numerous
calculations and modifications were made and hypothesized to explain the
differing behavior. The use of geometric scaling rather than thermodynamic
and aerometric scaling is a possible answer since the ratio of gaseous
momentum and viscous forces were considerably different between the two
systems.
Although turbulence was finally blamed for the aberrant behavior, the
reason for the turbulence was not ascertained. In the prototype kiln the
refuse combusiton air was injected at the kiln centerline and parallel to the
axis. In the Baltimore plant the refuse combustion air was injected parallel
to the kiln axis but nearly in line with the upper side of the kiln. The
result is that recirculation will take place in such a manner that the air
will attach to the kiln upper surface through a Coanda effect, spread out
over the surface, and force circulation to take place. Numerous solutions
were attempted, but the only successful solution was to eliminate the air jet
72
-------
and distribute the air into the kiln around the kiln perimeter. This modifi-
cation has minimized the recirculation of the combustion air:to eliminate the
high temperature fire ball in the discharge end of the kiln. Basically, the
problem is one of injection air momentum distribution to prevent local
circulation.
Process control requires several phases of measurement and control.
First, the residue temperature must be accurately measured. Second, kiln
discharge gas must be analyzed for 02 and C02 to determine the fuel-air
ratio. Next, the air input flow must be controlled. The present fan inlet
louvers control the flow but the actual flow rate is not known. For this
reason, each of the fan inlets should be fitted with a simply constructed
flow nozzle such as the ISA nozzle for minimum obstruction or clearance
requirements. Nozzle differential pressure measurements can, through a
simple microprocessor, be used to calculate fan air flow and the resulting
calculation used as a feedback signal to an automatic control of the fan
louver position. By careful aerodynamic design of the air injection ports,
jet diffusion can be accomplished with little or no circulation. The result
should be a well-controlled laminar flame front.
This problem must be classified as unique to the Baltimore plant since
the prototype facility was not so plagued. Furthermore, the use of a
different configuration eliminates the problem.
73
-------
SECTION 10
02/C02 ANALYZER SYSTEM
Although no catastrophic failure or shutdown has been directly attrib-
utable to a lack of information regarding the oxygen or carbon dioxide "levels
in the process gas stream, this condition has contributed to a number of
system problems. Reliable operation of the 02/C02 analyzers would allow
automatic control of the afterburner temperature, minimizing the problem of
slag hole pluggage during process disruptions.
Westinghouse instrument technicians indicated that the sampling probe
used was a water-cooled probe which employed a water spray to aspirate the
probe and moisten the sample. The probe was a premanufactured device selected
by Monsanto as meeting specifications and qualifications for application in
the Landgard process. Slag particles captured by the probe were supposed to
be removed by the spray wash. Three to five micron fly ash (glass slag)
particles entered the probe and plugged it about 2 cm (3/4 in.) inside the
probe even though provisions were made for back flushing the probes with high
velocity air.
In the early demonstration period, good gas samples were obtained from
the probes by rodding them on an hourly basis. This practice wasn't within
the realm of normal safe operation.
In addition to slag plugging the probe, there was a problem with the
water spray which caused flooding of the probe. When excessive water spray
was used, it would be drawn back into the' sample tube with the gas sample and
then into the analyzer where it was frozen by the refrigeration-dehydration
system and created a problem in the analyzer. A washer/separator was in-
stalled in each probe to attempt to overcome this problem.
City technicians indicated that the standard gases introduced to the
analytical equipment for instrument calibration invariably gave good results.
Switching to the probe to obtain a gas sample invariably gave no results or
unreliable results. In spite of the favorable experience with the calibra-
tion gas, the technicians felt that even if a good sample could be brought to
the analyzer, the measurement results would be unreliable.
Mislocation of the equipment was blamed by the technicians for all the
sampling problems. They indicated that the analyzers were located in a
position where oil, grease, and water were constantly dripping or pouring on
them and that the analyzer itself was frequently standing in enough water to
drown the heaters and cause short circuits or failure of the heating elements
74
-------
which resulted in the loss of calibration of the equipment. The technicians
also indicated that on at least one occasion the prism in the infrared
analyzer for C02 cracked and had to be replaced, although the exact cause of
the problem was not known.
Ultimately, the problem of exposure of this delicate instrumentation to
the weather was recognized and plans were made to build an enclosure for the
analytical system. A rebuilding effort was undertaken which included replace-
ment of the sample tubes and the installation of sample cleaning water
cyclone devices at each probe. Even after this modification no improvement
in the analytical capabilities was experienced.
One 10-day experimental trial run was made using a ceramic partial
pressure analyzer which was on loan. This device was installed about 25 ft
downstream from the gas purifier where a catwalk gave access to the duct. It
was reported that the ceramic probe worked well during the 10-day test
period. However, after the test probe was returned to the manufacturer, no
follow-up was made to assess the appropriateness of this equipment. There
are no plans at this time to repair or use the 02/C02 analyzers in the
future.
The probe used by Monsanto in the Landgard system at Baltimore is a
standard, off-the-shelf item. The sampling probe appears to have been the
major cause of failure of the 02/C02 Analysis System since the rodding out of
the probes produced good samples.
Westinghouse technicians believed that the probe could have been made
operable by installing a shield to screen the probe from the slag in the gas
stream and by increasing the vacuum of the probe to overcome the reduced
pressure in the sampling position behind the shield. In addition to the slag
build-up problem, the probes were also inadequately cooled. This apparently
was a two part problem. The first was the lack of assured circulation of
cooling water in the probe, and the other was the lack of an indication to
assure that the water was turned on. This was especially critical during
heat up of the kiln.
The gas samples contained considerable particulate matter. In order to
clean the samples of the particulate matter prior to introduction to the
analysis system, a cyclone washer was installed at each probe location.
While this did reduce the particulate matter in the sample, it also added
considerable moisture to the sample stream which, during the winter months,
caused condensate and freezing problems in the sample lines. Additionally,
it is suspected that both 02 and C02 were absorbed by the water, resulting in
a potentially biased analysis. It is not known whether any correction was
made for this bias.
The sample handling system that transported gas samples from the probe
to the analyzer was not properly designed for the environment in which it was
installed. The long lengths of unprotected tubing and numerous tube con-
nections between the probes and the analyzer were subject to abuse, and
resulting damage- caused leakage of ambient air into the transport tubes.
75
-------
Insufficient regard was given to the problem of particulate precipitation in
the transport tubes which led to blockage of sample gases. Furthermore, the
moisture content of sample gases was not adequately considered and freezing
of the tubes also occurred. Duplication of the analyzer to service two
sampling points did not appear economical at the time of the original design.
Events later proved the use of a single analyzer connected to two distant
sampling points to be unreliable.
A study of the engineering drawings for the system indicate that the
analytical efforts and technologies were well thought out and well engineered.
The sophisticated equipment selected appears to be applicable to a controlled
environment rather than the outdoor environment in which it was actually
installed. The instrumentation is delicate and electrical control circuits
are subject to poor performance in ambient weather conditions.
A probe design incorporating the rodding capability would have been
appropriate for this application, but a probe design which will create a
particle-free wake by ballistic separation at the orifice of the probe may be
possible. This would allow extraction of gases from the stream with a low
probability of ingesting particulate matter which will clog the probe.
The analytical equipment probably would have operated satisfactorily had
it been installed in a temperature and humidity-controlled environment such
as inside an office building. However, this being impractical, a housing of
its own should have been provided at the outset whereby the instrumentation
could have been protected from water, oil, and falling debris.
Another possible solution would be to install a set of the ceramic
partial pressure analyzers. The probe for this analyzer is an ion exchange-
type and is calibrated to give a measurement of the amount of oxygen in the
stream prior to its entrance into the gas purifier. The probe is self con-
tained and needs no gas handling system.
The problems with the analyzer must be considered unique to the Baltimore
plant since most of the problems were due to improper installation of the
analyzers and improper probe selection.
76
-------
SECTION 11
INDUCED DRAFT FAN
Since this fan produces the draft through the entire thermal process,
the process cannot function when the fan is inoperable. When increasing
vibration indicates the necessity for shutdown, an orderly, planned shutdown
will minimize the impact on the system. In this case, the penalties are
thermal cycling of the refractories with possible thermal shock damage and
shortened refractory life, loss of operating time, and increased per ton
costs.
An emergency shutdown, due to vibration suddenly exceeding the 0.36 mm
(0.014 in.) bearing displacement limit, power failure, or scrubber pump
failure has additional penalties. If the kiln stack lid does not open, fire
damage can result in the kiln feed area, and hot gases trapped in the system
can cause short-term, excessive heating of refractories and equipment. If
the kiln stack lid does open, damage to wiring and other equipment can occur
from the hot gases escaping the kiln. The rapid cooling results in even
greater thermal shocking of the refractories as well as hardening of slag in
the kiln and gas purifier in quantities greater than would be the case with
an orderly shutdown.
Vibration problems with the induced draft fan started immediately after
start-up requiring frequent fan rebalancing. After a few months of operation,
the rotor was worn and was replaced with a rotor with wear plates, the
vibration and the frequent balancing continued with the new rotor. None of
the numerous modifications attempted by Monsanto had any effect. The new
rotor was soon worn out even though it had wear plates. The city then decided'
to operate the plant with the scrubber off. This stopped the corrosion and
vibration problems until the scrubber was turned on for steam dumping. The
city intends to eliminate the problem by installing a dry ESP system and,
consequently, provide for a dry fan.
The gases passing through the fan are heavily loaded with fine parti-
culates and acidic vapors. An analysis of the particulates shows large
percentages of alumina and silicon dioxide particles with diameters of a few
microns. These particles are highly abrasive and, since steels depend on an
oxide layer for corrosion resistance, make the steel components of the fan
highly susceptible to corrosion by continually abrading the oxide coating.
In addition, since alumina and silica are harder than steel, the particulates
probably abrade the steel itself.
77
-------
When particles are concentrated in a local region, such as the turns
where the gas goes through the fan, erosion/abrasion is a potential problem.
In an induced-draft fan, blade wear is relatively low at fan-blade-tip
speeds under 6700 m/min (22,000 ft/min). Above this tip speed the wear from
erosion/abrasion increases very rapidly and higher speeds should not be
considered in designs for gases having high concentrations of entrained
particles. The rotor of the' induced-draft fan, when operated at the design
maximum speed of 1200 RPM, has a blade tip speed of 7500 m/min (24,500 ft/min).
In order to maintain a tip speed under 6700 m/min (22,000 ft/min) the fan
must be operated at less the 1077 RPM. This may necessitate the use of a fan
wheel with greater blade width to maintain sufficient draft.
The chlorine content of the g4s in conjunction with the moisture in the
gas forms an extremely corrosive environment for steels. The presence of
chlorine is attested to by the high acidity of the dehumidifier condensate
and the analyses of the stack gases. There are also chlorides in the form of
HC1 in the water droplets and vapor carried over from the scrubber whenever
the scrubber liquid swings to the acidic side which occured periodically due
to poor pH control in the' scrubber.
The corrosion/erosion problem arises from the use of steel components in
an environment for which steel is totally unsuited—a very abrasive and
corrosive environment.
The vibration problem is largely one of deposits on the fan rotor
aggravated by corrosioii/erosion. The spray droplets formed by the scrubber
spray nozzles are injected into a relatively hot gas stream. The water in
the droplets partially evaporates, thereby raising the solids content of the
droplets. Some of the droplets probably become, in essence, sludge droplets.
Droplets of this type arfe carried over the gas stream to the induced-draft
fan. Upon striking the surface of the fan blades, these high-solids droplets
tend to stick and spread but. The centrifugal forces imparted by the rotating
fan would tend to further dewater the solids to form a solid deposit. This (
condition would be promoted by any pitting of the blade surfaces by corrosion.
As the deposit builds up, it becomes susceptible to random flaking due to the
impact of large water droplets or centrifugal force. The exposed surface
left by the departing flake is subject to corrosion until deposits reform on
that area which opens the possibility of uneven corrosion of the metal and
uneven build up of deposits with subsequent unbalancing of the wheel.
An analysis of the fan deposits showed a sodium content of 30 percent.
Depending on the ratio of NaOH to NaCl, the sodium component content of the
deposit would be between 47 and 65 percent. This would verify that the
deposits originate in the scrubber water. This analysis and the maintenance
history together make a very strong case for the deposits being the major
cause of vibration problems.
The problems of abrasion and deposits can both be solved by eliminating
the scrubber and substituting a "dry" system for particulate remoyal such as
an ESP or a baghouse. Once the abrasion by particulate matter and erosion of
deposits by water droplets have been eliminated, the corrosion problem can be
solved by using a fan wheel fabricated of a chlorine-resistant material such
as the Hastelloys or some nonmetallic laminate.
78
-------
Since it appears that the fan problems resulted from the use of the
scrubber, the problem must be considered typical of processes using wet
scrubbers applied to particle laden ga
-------
SECTION 12
SCRUBBER EROSION AND CORROSION
Failure or shutdown of either of the scrubber pumps would have little
impact on system operation since the other pump could continue to operate.
Failure or shutdown of both pumps would cause a shutdown of the entire
system because of the interlock with the induced-draft fan. Failure of the
scrubber lances and nozzles does not necessitate a scrubber shutdown so there
would be no impact on system operations other than reducing the spray flow in
the scrubber.
SCRUBBER PUMPS
After severe wear, the cast iron scrubber pump impellers were replaced
with bronze impellers. The pumps continue to wear and a satisfactory packing
to prevent the constant leaking has not been found. The city plans to
eliminate this problem by installing a dry ESP unit to replace the scrubber.
When solids fill the bottom of the scrubber to the level of the pump
suction pipe, a steady state condition is reached in which all the particu-
lates removed by the scrubber are passed through the pumps with the scrubber
water. Monsanto reported that approximately 100 kg/hr (220 Ibs/hr) of
particulates are going to the scrubber. Even if collection efficiency is
low, a large amount of highly abrasive material is passing through these
pumps.
Cavitation in the pumps was suspected during operation, and a plate was
installed in the scrubber to prevent air from being sucked into the pumps and
causing cavitation. However, cavitation can be caused by other factors. If
the particulates in the water and the speed of the impeller combine to give
the right hydrodynamic conditions, localized cavitation can result from the
impact of the particulate against the surface of the impeller.
Cast iron is a very suitable material for pump impellers in an NaOH
environment at neutral or basic pH. The suitability of bronze depends upon
the particular bronze used and the operating conditions. Control of pH was
not very effective and acidic conditions were a common occurrence. During
conditions of low pH, corrosion Would be severe when combined with the
abrasion of the particulates in the water.
The most probable cause of the excessive impeller wear of the scrubber
pumps was particulate abrasion aggravated by some cavitation and corrosion.
80
-------
To solve or at least minimize the impeller wear and to provide good pH
control, the scrubber system would have to be redesigned to remove the solids
from the scrubber water before they reach the pumps. However, the scrubber
is to be replaced with an ESP.
SCRUBBER LANCES AND NOZZLES
The scrubber lances and nozzles have plugged repeatedly and are very
difficult to clean. Various types of materials were tested to prevent the
corrosion and erosion that occurred to the lances and nozzles. The City
intends to solve this problem by installing a dry ESP system and thus eliminate
the scrubber system.
The problems with the lances and nozzles can be divided into two areas;
internal and external. Internally, the problem is the same abrasion and acid
corrosion problem that exists in the pumps except that cavitation is not
significant. Abrasion is particularly severe in the nozzles because of the
constriction of flow and the sudden change in flow direction. The solution
to the problem is the same as for the pumps.
Externally, the problems are caused by the moist acid stream with
entrained, highly-abrasive fine particulates. The particulates continually
abrade any protective oxide coating on the steel components and the metal
itself.
The moist acid vapors are extremely hostile to any steel alloy. This is
even worse in the threaded areas of the fittings since threading increases
the area for corrosive attack per unit length of pipe and provides sharp
ridges and valleys where attack can be concentrated. Where threaded fittings
enter threaded connectors, conditions are perfect for crevice corrosion.
When the proper spray patterns are maintained, some measure of pro-
tection is provided to the lances from both abrasion and acid attack since
the sprays are directed away from them. However,~~as the nozzles wear, the
pattern of the sprays changes and the protection provided by the sprays is
gradually diminished. If the excessive erosion/corrosion of the nozzles can
be minimized by eliminating the abrasive solids in the scrubber waters and
controlling pH, the attack on the external lance surface can be minimized.
Because the excessive wear of the scrubber pumps, lances, and nozzles
was caused by improper design of the equipment and materials selection, the
scrubber erosion and corrosion problem is unique to the Baltimore plant.
81
-------
SECTION 13
SUMMARY
While the City of Baltimore has solved many of the problems at the
Landgard resource recovery plant by eliminating the source or changing the
process configuration, most of the problems could have been solved by equip-
ment modifications.
Refractory failures must be classified as critical since this could
result in serious structural damage to the processing vessel or ductwork.
After continuous and severe kiln refractory failures, only spalling remains
as a problem. The causes of the original kiln refractory failure include:
iproper installation and curing, exposure to high temperatures, differential
expansion between the shell and refractory, thermal shock spalling, and
structural spalling. The original kiln refractory failures are therefore
unique to the Baltimore plant while the present spalling is unique to the
Landgard process. The thickness of the gas purifier refractory has decreased
rapidly. This failure was caused by molten slag eroding and dissolving the
refractory. This failure is unique to the Baltimore plant since the proto-
type had a nonslagging afterburner. Refractory failures in the ductwork
resulted from the lack of understanding of service condition severity and are
considered to be unique to the Landgard process.
Conveyor failure is critical to the operation of the system because of
the required continuous flow of materials-between the unit processes in the
system. Most of the failures with the residue drag conveyor resulted from
inadequate knowledge of the process stream characteristics, especially
maximum particle size. The "kiln residue is typical of the Landgard process,
but since a screw conveyor was used in the prototype, the failures are
unique to the Baltimore plant. The screw conveyors failed for the same
reasons as the residue drag conveyor. These failures are unique to the
Baltimore plant because the prototype had a nonslagging afterburner. The
storage pit and shredder feed conveyor failures are typical of solid waste
processing since the equipment is typical of solid waste handling machinery.
Failure of the ram feeders is a critical problem since the feeders are
the mechanisms for advancing refuse into the kiln. The ram feeder failures
were of three types: ram jams, snout failure, and hydraulic system failure.
The ram jams were caused by snout deformation, the air seal deflection
plates, and material overload at the snout entrance. The initial water-
cooled snouts failed due to improper manufacturing and operating practice.
The present ram snouts failed due to excessive temperature in the kiln. The
hydraulic failures resulted from the use of noncompatible equipment. All the
ram feeder failures are considered to be unique to the Baltimore plant.
82
-------
System pressure oscillations are critical because they are hazardous to
personnel and equipment. The oscillations were probably caused by the
use of the oversized crossover combustion air fan. The problem is unique to
the Baltimore plant since it is due to the specific configuration of that
system.
Slag hole pluggage is critical because the slag build up in the gas
purifier would result in a structural failure of that vessel. Boiler tube
slagging is critical because it may cause boiler tube corrosion, low energy
recovery efficiency, and possibly an excessive pressure drop through the
boilers. Slag hole pluggage occurs because the gas purifier temperature is
not maintained at a high enough temperature to ensure proper slag viscosity
for slag tapping.
The addition of the quench air ducts prevented molten slag impingement
on the boiler tubes but did not prevent a friable slag mass from forming.
Since the main heat transfer mechanisms for cooling particulate is radiation,
the best solution to cool particulates would be either a slag screen or to
cool the duct walls.
Since the slagging gas purifier is unique to the Baltimore plant, the
problem of slag hole plugging and boiler tube slagging are also unique to the
Baltimore plant.
The f.ailures of the storage and recovery unit are considered critical
because they decrease system throughput and increase net processing cost.
The failures of this unit include poor refuse retrieval and excessive floor
and bucket shoe wear rates. These problems were caused by below design
refuse bulk densities, the nature of the refuse (rags, wire, and glass), and
the use of inappropriate construction materials. Since only the Landgard
process stores primary shredded, mixed municipal solid waste in this type of
unit, the failure of this unit is considered typical of the Landgard process.
A failure of the induced draft fan is critical because it must produce
the draft through the entire thermal processing system. The fan was fre-
quently shut down due to excess vibration until the system was operated
without the wet scrubber. The vibration was probably due to an uneven
build-up of solids on the fan impeller and corrosion/erosion of the impeller.
Since the fan failures were probably caused by the scrubber, the failures are
categorized as unique to the Landgard process.
Failure of the scrubber pumps is not critical unless both pumps fail.
The excessive wear of the scrubber pump impeller is probably due to the
abrasion caused by the particulates in the recirculating scrubber water, poor
pH control of the recirculating scrubber water, and pump cavitation. The
excessive wear of the scrubber lances and nozzles was probably caused by
particulate abrasion and corrosion. Since the excessive wear of the scrubber
was caused by improper design of the equipment and improper material selection,
the scrubber problems are considered unique to the Baltimore plant.
83
-------
Poor kiln process control does not directly cause a plant shutdown but
can result in the formation of large slag balls or unprocessed refuse jamming
or overloading the residue drag conveyor. The major cause of the poor kiln
process control was the inability to monitor the process. Additional control
problems were caused by the method of combustion air addition to the kiln in
a jet stream which caused recirculation of the gases in the fire end of the
kiln. The poor kiln process control is considered unique to the Baltimore
plant and has been resolved by system modification.
A shutdown of the residue separation module would not result in a
complete plant shutdown. The major cause of operational problems with this
module has been poor residue quality. Since the residue separation module is
unique to the Baltimore plant, so are its problems. This system has been
dismantled and scrapped.
A failure of the 02/C02 analyzer does not directly cause a plant shut-
down, but the resulting lack of process control can result in a problem
causing a plant shutdown. Most of the analyzer failures resulted from poor
probe design and the location of analytical equipment outdoors. Since
improper installation and poor probe selection caused the failures, the
problem is considered unique to the Baltimore plant.
«.. -281-147/81
84
------- |