FURNACE-WALL CORROSION IN REFUSE-FIRED

BOILERS

P. L. DANIEL J. L. BARNA AND J. D. BLUE

Babcock & Wilcox Babcock & Wilcox
Alliance Research Center Domestic Fossil Operations
Alliance, Ohio Barberton, Ohio

ABSTRACf INTRODUCfION

Modern refuse-fired boilers dispose of refuse while Many refuse-fired boilers are in operation through­
extracting energy for generation of steam and elec­ out the world. These units provide an attractive means
tricity. One of the factors that must be considered in for disposal of refuse while extracting energy for gen­
the selection and design of these units is corrosion. eration of steam and electricity. There remains eco­
This paper describes and discusses furnace wall cor­ nomic pressure for continued development toward
rosion, citing specific experience in an RDF (refuse units operating more efficiently and at higher steam
derived fuel) boiler. temperatures and pressures.
Tubing from the RDF-fired boiler was examined in However, this progress is limited by the occurrence
the laboratory to identify corrosive species in the fur­ of fireside corrosion problems. Refuse-fired boilers gen­
nace. High concentrations of chlorine, sodium, potas­ erally suffer more fireside corrosion than do other boil­
sium, and zinc were associated with areas of corrosion. ers. Alkali metals, heavy metals, and chlorides in the
Laboratory tests simulating the furnace environment fuels corrode furnace-wall and convection pass super­
were conducted to establish relative corrosion rates of heater tubes. The problem is exacerbated by the in­
carbon steel and higher alloys. herent heterogeneity of the fuel which makes furnace
It is concluded that HCI and NaCI are not suffi­ control difficult and by high fouling rates which require
ciently corrosive to explain observed furnace-wall cor­ frequent boiler outages. While these problems by no
rosion rates while mixtures of ZnClz and NaCI readily means threaten the viability of the modern boiler de­
do so. Alloy 625 cladding on furnace-wall tubes was sign, they do increase maintenance costs, decrease unit
found to provide good resistance to chloride attack in reliabilities and efficiencies, and limit design and op­
laboratory short-term screening tests, and should also erational flexibilities. Corrosion problems occur on
resist out of service chloride pitting and chloride stress supports, superheater, and furnace water wall tubing.
corrosion cracking. This resistance to chloride attack Corrosion problems on higher temperature compo­
has been tentatively confirmed in a 6 month trial of nents are described in a previous paper [1]. This paper
test overlays in an operating RDF unit. Laboratory uses experience with one particular unit to illustrate
and field tests are continuing to further evaluate this the furnace-wall corrosion problem, how corrosion re­
and other means to address fireside corrosion problems lates to other boiler characteristics, and how the cor­
in refuse-fired boilers. rosion problem can be addressed.

221
 The subject boiler was built by the Babcock & Wil­ desirable unit characteristics while minimizing the po­
cox Company. The unit is designed to burn 950 tons/ tential for fireside corrosion problems.
day of 4800 Btu/lb refuse derived fuel (RDF). The The subject unit was not lined with silicon carbide
boiler is designed to produce 250,000 lb/hr of 750°F because use of silicon carbide impedes heat transfer.
(399°C) steam at a pressure of 650 psig. The saturation For the design fuel characteristics and metal temper­
temperature is approximately 500°F (260°C). The fur­ atures and with good air and fuel distribution in the
nace-wall tubes are 2.5 in. O.D. X 0.165 in. wall (63.5 furnace, the unit was expected to operate within the
mm X 4.2 mm), SA-I78A carbon steel tubing, in a envelope of safe operating conditions with respect to
membrane wall construction. The unit configuration furnace-wall corrosion. However, once the unit went
is shown in Fig. 1. into operation, the heating value of the fuel delivered
to the plant was 1 5-25% above the design heating
value. The unit also experienced water chemistry up­
sets during the period of initial operation. The resultant
high heat release rates and waterside deposition raised
FURNACE-WALL CORROSION
the tube-metal temperature into the range where fire­
side corrosion became a serious problem for localized
Corrosion of furnace-wall panel tubes with fireside areas of the furnace wall in the flame zone. In addition
metal temperatures of 500-600°F (260-316°C) in re­ to raising the tube-metal temperature, the excessive
fuse-fired boilers is frequently localized, occurring heat release rates and resultant steep thermal gradient
above the fuel feed points, along sides, or in corners across fireside wall deposits accelerate diffusion of cor­
'
of the furnace. The severity of the corrosion generally rosive species to the tube-metal surface. Ultrasonic
increases with increasing proximity to the flame zone. thickness measurements indicate that corrosion rates
The corrosion kinetics are usually linear, with the cor­ of 30-50 mpy (0.8-1.3 mm/year) may occur in lo­
rosion rate showing little tendency to decrease with calized sections of the lower furnace walls.
the passage of time. Corrosion rates .increase rapidly
with increasing tube metal temperatures [2].
Experience indiclttes strong correlation of these
wastage problems with a number of boiler operating
conditions. Conditions that aggravate the problems in­ TUBE EXAMINATION

clude: A rear wall tube from the subject unit was examined
(a) refuse fuel rich in PVC plastics and, hence, in in the laboratory. The tube had been in service since
organic chlorides the original unit startup in July, 1984. It was in service
(b) higher fuel Btu-values and higher gas temper­ for approximately 6 months during which time it had
atures suffered a 25%, 40 mil (1.0 mm) loss in thickness or
(c) higher tube temperatures and waterside depo­ a corrosion rate of 80 mpy (2.0 mm/year). When
sition which interferes with heat removal received for laboratory analysis, the tube retained a
(d) flame impingement and stratification of air flow sooty external deposit. When one gram of deposit was
(e) frequent and thorough fireside deposit and scale dissolved in 100 g of water, the resultant solution had
removal a pH of 4. The tube surface was generally rough and
Several approaches have been taken to ameliorate pitted.
the furnace-wall corrosion problems:
(a) covering the wall with silicon carbide
(b) controlling the fuel chemistry (e.g., burning a
Procedure
mixture of coal and refuse)
(c ) lowering the steam pressure and thereby low­ Bulk deposit chemistry of the fireside deposit was
ering the saturation and tube-metal temperatures obtained from x-ray diffraction and spectrographic
(d) decreasing the amount of flame contact with the analyses. Optical metallography was used to determine
furnace wall the deposit and scale morphologies. To retain water­
Each of these ameliorative actions costs a penalty in soluble constituents of the fireside peposits, kerosene
terms of unit operating load, efficiency, and flexibility. was substituted for water during the metallographic
Maximum steam temperatures and pressures, and preparation. Electron microprobe analysis of the met­
boiler efficiencies are limited by the danger of corro­ allographic samples provided microchemical infor­
sion; and boiler designers attempt to maximize these mation about the fireside deposits.

222
FIG. 1 DRAWING OF THE RDF-FIRED BOILER

223
Results TABLE 1 SPECTROGRAPHIC SEMI-QUANTITATIVE
ANALYSIS (%)* OF FIRESIDE DEPOSITS
The chemical analysis of the fireside deposit on the Rear Will1 Tube

rearwall tube are given in Table 1. The deposits were

composed primarily of iron, sodium, and chlorine. Silicon as );02 3.3

Alumlnum as A1Z0 3 1.3

Large amounts of the following elements were also Iron as Fe Z0 Major
3
identified: silicon, aluminum, titanium, calcium, zinc, Titanium as Ti0 2 3.0

ea lcium as CaD 2.5

lead, carbon, sulfur, potassium, and oxygen. X-ray dif­ Magnesium IS MgO 0.2

fraction analysis identified sodium chloride (NaCl) as Sod; urn II S NIZO.... 20.89

the major crystalline constituent and identified he­ Nic.kel as NiD <0.06

Chromium as CrZO <0.06

matite (F�03) and magnetite (Fe30.) as minor con­ )
MolyDdenum as MoG3 <0.06

stituents. Vanadium as VZ0

5
<0.1

Coba 1t as CoO <0.06

Photomicrographs depicting the nature of the fire­ ,,:anganese as I1n::l2 0.06

side corrosion and associated deposit morphology are Copper a 5 CuD 0.4

shown in Fig. 2. The deposits were porous, grey in Zinc as ZnO 2.3

1.2
leaa is P[)(l
color, and multi-layered. A layer of protective iron Tin as �nv2 0.2

oxide scale, adjacent to the tube metal, was not always ZH\,.unium as Zru
Z
<0.06

Caroon IS C··· 5.68

visually apparent. The rearwall tube deposits were gen­ !loultwr as SO
)
erally adherent and were measured from cross sections Clllorine as C1··· 22.7

NA
to be approximately 20 mils (0.5 rom) thick. Phosphorus IS P O ····
Z S
6.14
PoJtassium as K20··
The x-ray scanning capability of the electron mi­
croprobe was used to determine the relative distribu­ )'-Ray L)iffration (Crystalline Constituents)

tion of several elements in the fireside deposits. Major WaCl

Resultant x-ray fluorescence maps, along with electron Minor

images of the area, are shown in Fig. 3.

Examination of the element maps in Figure 3 reveals
-Tne results of spectrographic analysis are reported by the Research Center
several patterns: as tne oxides. Tnls does not necessarily lRean tnat the elelRents Ire present
as sucn in the sample.
(a) Chlorine and zinc are distributed throughout the
--Flal1'II: pnotometer analysis
deposit.
......et chemical analysis
(b) Sodium is more abundant in outer portions of ...... lCP spectrographic analysis
the deposit. Kajor - )201
(c) Potassium is more abundant in the inner portion JrriA - not analyzed

of the deposit near the tube surface.

(d) There is a thin layer at the metal/deposit in­
terface that is iron-rich. This layer does not appear to (316°C), which is approximately the maximum oper­
be associated with oxygen, but there is a definite as­ ating furnace wall tube metal temperature in this unit.
sociation with chlorine. The materials test, which included carbon steel and
(e) Lead, while scarce, is concentrated near the higher alloys, are listed in Table 2. Iron oxide, Fe30.,
metal/deposit interfaces. was included in the test to determine the extent to
(j) Oxygen is present in relatively low concentra­ which the initially protective oxide, once formed on
tions uniformly distributed throughout the deposit. steel, would be attacked. Also tested, was a section of
(g) Relatively small amounts of sulfur are randomly furnace-wall tubing (carbon steel) clad with the com­
distributed through the deposit. mercial nickel-base Alloy 625, an alloy believed to
possess superior corrosion and pitting resistance in the
anticipated furnace environment.
Each test specimen was a rectangular sheet coupon
LABORATORY TESTS
that was initially prepared by rinsing in deionized water
Laboratory tests were undertaken to measure cor­ and acetone followed by ultrasonic cleaning to remove
rosion rates in zinc chloride and sodium chloride salts any foreign material. The specimc;n coupons were
at 600°F (316°C). The tests helped determine the cor­ weighed and placed in separate crucibles and covered
rosion mechanism and served to screen alloys for use with the chloride mixture with air as the cover gas.
as weld overlay materials to protect the wall tubes The chloride mixture used in the tests was of the two
from further attack. The test temperature was 600°F corrosive salts of particular concern, ZnC12 and NaCl.

224
 TABLE 2 CORROSION RATES IN CHLORIDE SALTS AT
600'F <316°C)

Time a4� ZnCl2 + 16 wtl NaCl2 100S NaCI

Material (hrs) Rate. m�x (IlIn/xr) Rate. m�x (�/xr)

Carbon S tee 1 168 1700 (43) 6 (0.15)

Sample A

Carbon S.teel 336 910 (23)

Sample 8
2
Croloy -1/4 168 410 (10) 1 (0.03)

Fe 0 168 750 (19)

3 4
Alloy 600 168 5 (0.13) 0.1 (0.003)

Alloy 625 168 2 (0.05) 0.2 (0.005)

Alloy 625 clad 336 <10 «0.25)

on carbon steel

ALLOY Alloy 625 clad 672 <12 «0.30)

on carbon steel

Alloy 690 168 3 (0.08) 0.0 (0.000)

Alloy 800 168 3 (0.08) 0.0 (0.000)

Alloy 304 168 3 (0.08) 0.1 (0.003)

FIG. 2 PHOTOMICROGRAPHS OF THE FIRESIDE Alloy 309 168 11 (0.28) 0.2 (0.005)
DEPOSIT AND SCALE ON THE REAR WALL TUBE

cause an 0.045 mpy (0.001 mm/year) corrosion rate.

Hence, while HCI in the combustion gas may corrode
The mixture was 84 wt % ZnCI2 and 1 6 wt % NaCI.
furnace-wall tubes, the HCI corrosion rate is probably
At the conclusion of each of the tests, the samples were
not high enough to account for the corrosion rate
rinsed with deionized water and acetone, acid cleaned
observed in the subject unit.
to remove any oxide formation, and reweighed. Re­
sultant corrosion rate data are presented in Table 2.
Corrosion by NaO Deposits on Tube Surfaces

Substantial amounts of sodium and potassium are

CORROSION MECHANISM
present in paper, textiles, wood, and other components
Corrosion in the RDF-fired unit appears to be caused of refuse derived fuel. A portion of these alkali metals
by chlorides which deposit on the furnace-wall tubes. is volatilized, for example, as alkali metal chlorides
Several different modes of chloride corrosion may and hydroxides. The hydroxides in the combustion gas
occur: react with HCI to form the alkali metal chlorides (e.g.,
(a) corrosion by HCI in the combustion gas NaOH + HCI NaCI + H20) which deposit on the
-+

(b) corrosion by NaCI and KCI deposits on tube relatively cool furnace walls.
surfaces Short duration (50 hr) laboratory tests by Miller, et
(c) corrosion by low melting point metal chlorides aI., indicate that, in oxidizing combustion gas atmo­
(mainly ZnCIJ spheres, sodium chloride deposits accelerate steel ox­
(d) out-of-service corrosion by wet salts on the tube idation rates to approximately 35 mpy (0.89 mm/year)
surface at 800T (472°C) [4]. Based on this data point, the
corrosion rate was expected to be substantially less,
but still significant at 600°F (316°C), perhaps several
Corrosion by HO in the Combustion Gas
mpy. Out laboratory test results indicate that the cor­
Chloride is present in the fuel mostly as polyvi­ rosion rate for carbon steel in NaCI at 600°F (316°C)
nylchloride in plastics, and it volatilizes largely as HCI. is only 6 mpy (0.15 mm/year). Hence, NaCI and KCI
HCI reacts with steel to form FeC12• Short duration (1 deposits alone can not account for the corrosion rate
day) tests indicate that the corrosion rate of carbon observed.
steel in dry HCI gas is approximately 45 mpy (1.1
mm/year) at 550°F (288°C) [3]. Lower HCI partial
Corrosion by Zno2
pressures should result in lower corrosion rates. If the
corrosion rate were proportional to the HCI partial Zinc compounds are present in textiles, leather, and
pressure in the combustion gas, then 0.1 % HCI would rubber. Zinc oxide is the most common of the zinc

225
 SCALE
-""""'"

ALLOY
I I
50 lim

(a) ELECTRON IMAGE (b) IRON MAP 50 lim

FIG. 3 ELEMENT MAPS OF REAR WALL TUBE DEPOSIT AT THE METAL/DEPOSIT INTERFACE. THIS AREA IS THE
SAME SHOWN IN THE PHOTOMICROGRAPHS IN FIG. 2.

226
compounds, and it is relatively nonvolatile and in­ face pH is acidic, the austenitic stainless steels may be
nocuous. However, in the flame zinc oxide is readily subject to stress corrosion cracking [5].
reduced to metallic zinc, and it reacts with HCl to
form ZnCI2•
Zinc chloride is volatile, but readily condenses on
Conclusions
the relatively cool furnace wall. Because of its high
volatility, it moves through the surface deposit and Of the chemical species identified on wall tubes re­
condenses at the tube surface. The zinc (and lead) moved from the unit, zinc chloride is the most ag­
chloride may be present on the tube surface as a liquid: gressive. Laboratory tests described in this paper
for example, zinc chloride and sodium chloride form indicate that zinc chloride can account for the partic­
a eutectic which melts at 504°F (262°C). Zinc chloride ularly rapid corrosion occurring in the subject unit.
destroys the protective oxide film on steel (it is com­ This mechanism depends on a number of factors:
monly used as a fluxing agent for soldering and braz­ (a) chloride in the refuse derived fuel
ing) and attacks the underlying metal. Furthermore, (b) zinc (and lead) in the fuel
by disrupting the protective iron oxide on the steel, (c) sodium and potassium in the fuel
the zinc chloride facilitates rapid corrosion by HCl in (d) reducing conditions at the wall
the combustion gas. Laboratory tests by Miller, et aI., (e) high tube metal temperatures
indicate that the addition of 30% ZnCl2 and 15% (j) furnace-wall material susceptible to chloride at­
PbCl2 to sulfate deposits on steel increases the corro­ tack
sion rate to 120 mpy (3.0 mm/year) at 600°F (316°C) Reducing any of these factors is expected to reduce
[4]. The sulfates alone had comparatively little effect the corrosion rate. However, some of the factors may
on the corrosion rate. be beyond practical control. For example, the fuel com­
The aforementioned corrosion process occurs largely position can probably not be altered. Above about
under reducing conditions. Oxidizing conditions con­ 500°F (260°C), corrosion rates generally increase ex­
vert ZnCl2 to ZnO and promotes the formation of a ponentially with temperature [2]. Hence, avoidance of
protective oxide scale on the tube surface. Reducing waterside deposition which increases tube metal tem­
conditions facilitate formation of ZnCl2 and PbCl2 and perature is imperative.
retard formation of the protective scale. Where con­ Promising weld overlay materials for protecting fur­
ditions are intermediate or alternating, some of the nace tubes are Inconel 625 and similar high-chromium
iron chloride corrosion product is subsequently oxi­ nickel-base alloys. Promising iron-base alloys are In­
dized and forms a nonprotective iron oxide within the coloy 800 and 825. The greatest danger to austenitic
deposit. Under these conditions, displaced chlorine stainless steel overlays is out-of-service chloride stress
may cause further corrosion of the steel substrate. Our corrosion cracking. The greatest danger to nickel-base
laboratory tests indicate that under oxidizing condi­ alloys (for furnace-wall application) is out-of-service
tions, a molten zinc chloride, sodium chloride mixture pitting.
at 600°F (316°C) attacks carbon steel at a rate of several The objectives of this investigation were to identify
hundred mpy. a possible furnace-wall corrosion mechanism, measure
the corrosivity of chloride compounds, and evaluate
potential weld overlay materials. Results show that
ZnCl2 can account for the high in-service corrosion
Out-of-Service Corrosion
rates of furnace wall tubing. Other similar low melting
Chloride-rich deposits on tube surfaces are hydro­ point compounds may also be involved. The effects of
scopic and become moist during out-of-service periods. ZnCl2 and NaCl on carbon steel tubing at upper limit
The pH of this moist surface deposit is acidic (1.0 g furnace-wall temperatures is severe. However, weld
of deposit in 100 g of water gave a pH of 4). As a overlay of carbon steel furnace-wall tubing with Alloy
result, corrosion can proceed rapidly during out-of­ 625 may provide a viable solution to the furnace-wall
service periods and may account for a part of the tube corrosion problem.
metal loss. Corrosion of carbon steel in high chloride The viability of Alloy 625 overlay will be determined
acidic environments typically causes general wastage by long-term field tests. Alloy 625 overlay has been
and pitting. applied in the furnace of the RDF-fired boiler discussed
Out-of-service corrosion is particularly dangerous in this report, and after 6 months of operation the
for austenitic stainless steel overlays and cladding. overlay shows no evidence of corrosion (as determined
Where the overlays have residual stresses and the sur- by ultrasonic thickness measurements).

227
REFERENCES [4] Miller, P. D., Krause, H. H., Zupan, J., and Boyd, W. K.
"Corrosion Effects of Various Salt Mixtures Under Combustion Gas
[1] Chou, S. F., Pramik, A. I., Scott, M.E., and Daniel, P. L. Atmospheres." Corrosion 28 (June 1972): 222-225.
"High-Temperature Corrosion of Tube Support and Attachment [5] Krause, H. H., Vaughan, D. A., and Miller, P.D. "Cor­
Materials for Refuse-Fired Boilers." Presented at the 1985 Joint rosion and Deposits from Combustion of Solid Waste, Part 2-
Power Conference in Milwaukee, Wisconsin October, 1985. ASME Chloride Effects on Boiler Tube and Scrubber Metals." Transactions
Paper 85-1PGC-Pwr-41. of the AIME (July 1974): 216-222.
[2] Norwak, F. "Corrosion Problems in Incinerators." Com­
bustion November 1968, 32-40.
[3] Vaughan, D. A., Krause, H. H., and Boyd, W. K. "Chloride
Ash Deposits and
Corrosion and Its Inhibition in Refuse Firing." In
Corrosion Due to Impurities in Combustion Gases. Washington: Key Words: Boiler; Chlorine; Corrosion; Furnace;
Hemisphere Publishing Corporation, 1977, 455-472. Incinerator; Refuse Derived Fuel; Waterwall

228