Corrosion Science 51 (2009) 1022–1029

Contents lists available at ScienceDirect

Corrosion Science
journal homepage: www.elsevier.com/locate/corsci

Prediction of oxide scale growth in superheater and reheater tubes

J. Purbolaksono a,*, A. Khinani a, A.Z. Rashid a, A.A. Ali a, N.F. Nordin b
a
Department of Mechanical Engineering, Universiti Tenaga Nasional, Km 7 Jalan Kajang-Puchong, Kajang 43009, Selangor, Malaysia
b
TNB Research Sdn Bhd, No. 1 Lorong Air Hitam, Kajang 43000, Selangor, Malaysia

a r t i c l e i n f o a b s t r a c t

Article history: In this paper a procedure on how to estimate the oxide scale growth in superheater and reheater tube
Received 20 August 2008 utilizing the empirical formulae and the finite element modeling is proposed. An iterative procedure con-
Accepted 24 February 2009 sisting of empirical formulae and numerical simulation is used to determine scale thickness as both tem-
Available online 6 March 2009
perature and time increase. Results of the scale thickness over period of time for two different design
temperatures of the steam and different heat transfer parameters are presented. The procedures may pro-
Keywords: vide better estimation on the oxide scale growth, provided that all the heat transfer parameters are well
A. Steel
specified.
A. Steam
B. Modeling studies
Ó 2009 Elsevier Ltd. All rights reserved.
B. Heat transfer
C. Oxidation

1. Introduction
Failures resulting from long-term overheating occur in steam-
cooled tubes such as superheaters and reheaters. As reported by
Port and Herro [1], almost 90% of failures caused by long-term
overheating occur in superheaters, reheaters and wall tubes. Tubes
that are especially subjected to overheating often contain signifi-
cant deposits. The deposits will reduce coolant flow, and the tubes
experience excessive fire-side heat input. They also described that
heat transfer is markedly influenced by a thin gas film that nor-
mally exists on external surfaces. A temperature drop commonly
occurs across this film. Scales and other materials on external sur-
faces will slightly reduce metals temperatures. The thermal resis-
tance of the tube wall may cause a very slight drop in
temperature across the wall. When heat transfer through the
steam-side surface is considered, the effect of deposits is reversed.
Steam layers and scales insulate the metal from the cooling effects
of the steam, resulting in reduced heat transfer into the steam and
increased metal temperatures.
When the tube metal is in contact with the steam over period of
time, the oxidation process may begin to form a layer of magnetite
(Fe3O4) scale. In the prolonged exposure this phenomenon will
worsen situation that leads to potential creep rupture problems.
Scales inside the superheater and reheater steam tubes have also
been found to be one of the major contributors to the tube failure.
Heat transfer rate across the tube also decreases due to the accu-
mulated scales inside the tube.
* Corresponding author. Tel.: +60 3 89212213; fax: +60 3 89212116.
E-mail addresses: judha@uniten.edu.my, j.purbolaksono@gmail.com
(J. Purbolaksono).
A further effect of growing scales is that the tube will have high-
er temperatures than those as originally specified. Such exposure
may cause degradation of the tube alloy, and this eventually will
lead to tube rupture. It is estimated that 10% of all power-plant
breakdowns are caused by creep fractures of boiler tubes due to
the scales formation [2].
Clark et al. [3], who were working for Aptech Engineering Ser-
vices Inc., provided a validated procedure in the form of computer
code for predicting the remaining useful life of SA213-T22 super-
heater and reheater tubes. One of the important tasks they per-
formed is acquisition and compilation of oxide growth
information for 2.25%Cr-1Mo steel. The procedure was to be based
on steam-side oxide scale measurements by ultrasonic technique,
tube geometry measurements and other readily available operat-
ing parameters.
Viswanathan et al. [4,5] reported a methodology developed by
Electric Power Research Institute (EPRI) and its contractors to help
utilities make more informed run/replace decisions for tubes by
judiciously combining calculation, nondestructive, and destructive
evaluations. In the methodology, the tubes/tube assemblies at risk
are identified by ultrasonically measuring the thickest steam-side
oxide scale and thinnest wall thickness in the tubes. The research
has further refined the methodology by validating the ultrasonic
technique for scale measurement, identifying the appropriate
stress formula and oxide growth laws.
Babcock & Wilcox Company, USA, has designed and built the
portable, ultrasonic Nondestructive Oxide Thickness Inspection
System (NOTISÒ) for measuring oxide scale on the inner surface
of tubes [6]. The application of NOTISÒ makes it possible to non-
destructively assess a large number of tubes within a
superheater.

0010-938X/$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.corsci.2009.02.025
 J. Purbolaksono et al. / Corrosion Science 51 (2009) 1022–1029 1023

An accurate prediction of the scale growth on the inner surfaces

of the superheater and reheater tubes will aid the power plant
inspectors or engineers in order to evaluate the remaining life of
the boiler tubes. French [7] described the procedures to estimate
the growth of scale thickness on the inner surface of the tubes
using empirical formula correlating scale thickness with Larsen- 100 mm Oxide Steel
Steam
Miller parameter [8] and approximated formulae of the tempera- Scale
Hot gases
ture increase for limited cases.
Ennis and Quadakkers [9] discussed the significance of the for-
mation of thick oxide scales during exposure of Cr steels in steam-
containing environments on the service life of components. The
quantitative effects of the loss of load-bearing cross-section on
the creep rupture life are presented. Potentially much more dam-
aging is, however, the thermal insulation effect of the porous, thick X Remaining metal thickness
Hollow Radius
oxide scales, which leads to overheating of heat transfer tubes. The Initial tube thickness
higher metal temperatures will then lead to early failure by creep.
Fig. 1. Model of the superheater and reheater tubes with scale on the inner surface.
The scale present after 10,000 h exposure in steam may be suffi-
cient to cause a temperature increase which will reduce the rup-
ture time for a constant stress. They also reported several Table 1
consequences of the formation of thick oxide scales for the service Geometries of the tubes.
behaviour of components.
Tube Inner radius (m) Outer radius (m)
Starr et al. [10] proposed an expert system for identifying the
root causes of the failures in superheater tubing made of the P91 1 0.0219 0.0254
2 0.0219 0.0274
and P92 martensitic alloys. The system may encapsulate current
3 0.0199 0.0254
knowledge about superheater problems in the form of ‘‘If-Then”
rules. Root causes of creep failures include furnace design and
operation, overestimation of alloy creep properties, inadequate
heat treatment and a non-optimum content of strengthening ele- of 540 and 605 °C, and the detailed other heat transfer parameters
ments. A characteristic of the P91 and P92 martensitic alloys is that are tabulated in Tables 2–4. Heat transfer along the external sur-
oxidation on the steam side of the tubing can induce premature face between the flue gas and the tube wall is considered as forced
failures due to the insulating effect of the oxide scales raising tube convection heat transfer due to cross flow of the hot flue gas over
temperatures. In addition, scale spallation could also increase tube bare tubes. The material of the seamless ferritic low-alloy steel
temperatures, as spallation debris may collect in the bottom of tube used in this work is SA213-T22 (see Table 2 for its thermal
tubes, blocking steam flow. Attention is drawn to a potential ‘‘run- conductivity). The chemical composition of the material is listed
away affect” in which the tube temperature and rate of oxidation in Table 5. Ferritic low-alloy material such as SA213-T22 generally
increase with time as the oxide builds up. The root cause of this cannot withstand highly oxidizing environment for a long period of
could either be excessive rates of heat transfer or could be due to time. The use of the material is usually limited to locations where
inadequate oxidation resistance caused by low levels of protective the temperatures are relatively lower.
elements. The steam-side scale is usually reported to be duplex (inner spi-
With the respect to the concerns stated in the previous works nel layer and outer magnetite layer) or triplex (inner spinel layer,
[1–7,9,10], the present study confined the analysis in the absence middle magnetite layer and outer hematite layer). In this study
of oxide scale developed on the external surface of the boiler tubes. material of the scale is treated to be all magnetite.
The procedure on how to estimate oxide scale growth in reheater Phenomenon of heat transfer inside the boiler tube is consid-
and superheater tubes utilizing the empirical formula correlating ered as forced convection with turbulent flow. Correlation for fully
scale thickness with Larsen-Miller parameter [8] and the finite ele- developed turbulent flow in tube is expressed as [11]:
ment modeling is proposed in this paper. Finite element models for Nus ¼ 0:023ðRes Þ0:8 ðPrs Þ0:4 ð1Þ
heat transfer analyses, that involve forced convections on the inner
surface due to the turbulent flow of steam and on the outer surface where Res is Reynolds number that may be expressed as
due to cross flow of the hot flue gas over bare tubes, are carried out o
in order to obtain temperature distribution in the tube. An iterative
4 ms
Res ¼ ð2Þ
procedure is used to determine scale thickness as both tempera- pDls
ture and time increase. The scale thickness over period of time
for two different design temperatures for steam and different heat
transfer parameters is presented. The finite element analysis is car- Table 2
ried out using software package of ANSYS. Properties of steam and solid materials.

Temperature, 540 °C Temperature, 605 °C

2. Numerical models Inlet steam properties [11]
Thermal conductivity 0.0604 W/m C 0.0662 W/m C
In modeling of the steady state heat transfer for the problem Specific heat 2161 J/kg C 2205 J/kg C
using ANSYS, the area of the model is divided into two regions, Density 0.2697 kg/m3 0.2489 kg/m3
Dynamic viscosity 2.834 e05 N s/m2 3.071 e05 N s/m2
i.e. scale region and tube region (see Fig. 1). The steam region is ta-
ken into account in determining the convection coefficient of Water wall properties [7]
Tube material SA213-T22
steam film for fully developed turbulent flow in circular tube. Mod-
Thermal conductivity 34.606 W/m C
el of the tube section used is 100 mm in length. Three different
Fe3O4 iron oxide (magnetite) [7]
geometries of the tube as shown in Table 1 are used. Steam flows
Thermal conductivity 0.592 W/m C
through the internal of tube with two different inlet temperatures
1024 J. Purbolaksono et al. / Corrosion Science 51 (2009) 1022–1029

Table 3 Heat transfer outside the boiler tube is considered as forced

Combinations of the studied models. convection due to cross flow of the hot flue gas over bare tubes.
Model Steam Mass flow Steam Flue gas Tube A conservative estimated convection coefficient of flue gas hg on
temperature rate (kg/h) pressure temperature outer surface of bare tube in inline and staggered arrangements
(°C) (MPa) (°C) (see Fig. 4) is given by [13]
1 540 3600 4 800 1
12kg
2 540 720 4 800 1 hg ¼ 0:33 ðReg Þ0:6 ðPrg Þ0:33 ð5Þ
3 540 3600 4 900 1 d0
4 540 3600 4 1000 1
5 540 3600 4 900 2 where kg is flue gas conductivity; d0 is outer diameter of the tube;
6 540 3600 4 900 3 Prg is defined as
7 605 3600 4 800 1
lg Cpg
Prg ¼ ð6Þ
kg

Table 4
in which Cpg and kg are specific heat and thermal conductivity of the
Properties of the flue gas and convection coefficients at different temperatures [11]. flue gas, respectively. The corresponding Reynolds number Reg may
be expressed as
Temperature (°C) 800 900 1000
Dynamics viscosity lg (N s/m2) 0.0418 0.0442 0.0465 Gd0
Reg ¼ ð7Þ
Specific heat Cpg (J/kg °C) 3158 3220 3277 12lg
Thermal conductivity kg (W/m °C) 0.0410 0.0440 0.0469
where G is gas mass velocity and may be defined as
Wg
G ¼ 12 ð8Þ
Table 5 Nw LðSt  d0 Þ
Chemical composition of SA213-T22 [12].
in which Wg is gas flow; Nw is number of tube wide; St is transverse
Code C Si Mn P S Cr Mo pitch (see Fig. 4), and L is the tube length. In this study parameters
(max) (max)
used to determine gas mass velocity are given in Table 7. Composi-
SA213- 0.05– 0.5 0.3– 0.025 0.025 1.90– 0.87– tions of flue gas at 15% excess air as shown in Table 8 is used in this
T22 0.15 0.6 2.60 1.13
study. The convection coefficients hg on external surface of the boi-
ler tube are obtained from Eq. (5) using parameters given in Tables
o
4 and 7. The coefficient values are presented in Table 9.
in which ms is mass flow rate of the steam; D is the inner diameter Superheater and reheater tubes operate at a continually
of the tube; ls is steam viscosity, and Prs is its Prandtl number that increasing temperature, and a prediction must be made of scale
is defined as thickness as a function of time and temperature. In this work in or-
ls Cps der to perform a scale growth prediction, steam-side scale forma-
Prs ¼ ð3Þ tion for ferritic steel of 1–3% chromium correlated with the
ks
Larsen-Miller parameter as reported by Rehn et al. [8] is utilized
in which Cps and ks are specific heat and thermal conductivity of the (see Fig. 2). The data of Fig. 2 may be approximated as
steam, respectively.  
Eq. (1) must comply with the following conditions: X
log ¼ 0:00022P  7:25 ð9Þ
0:0254
– Fluid properties are evaluated at the mean bulk temperature. where X is scale thickness in mm.
– Res > 10; 000. In the Larsen-Miller method, time and temperature are related
– 0:7 < Prs < 100. by the following equation:
– L/D P 10; L is the length of the tube.  
9
T þ 492 ðC þ log tÞ ¼ P ð10Þ
Convection coefficient of steam film for fully developed turbu- 5
lent flow in circular tube is expressed as [11]: where P is the Larsen-Miller parameter; T is the temperature in de-
ks gree Celsius; t is the service time in hours; C is a constant equal to 20.
hs ¼ 0:023 ðRes Þ0:8 ðPrs Þ0:4 ð4Þ The increasing of metal temperature DT for the reheater or
D
superheater tube (SA213-T22) may be obtained from the
where ks is steam conductivity. The convection coefficients hs on the
internal surface of the boiler tube are obtained from Eq. (4) using
parameters given in Tables 2 and 3. The coefficient values are pre- Table 7
sented in Table 6. Parameters used to determine gas mass velocity G.

Gas flow (kg/h) 400,000

Number of tube wide 32
Transverse pitch (m) 0.1016
Table 6
Tube length (m) 10
The convection coefficients hs on internal surface of the boiler tube.

Model hs (W/m2 °C)

1 2053.65
Table 8
2 566.70
Compositions of flue gas at 15% excess air [13].
3 2053.65
4 2053.65 Nitrogen (mole %) 71.08
5 2053.65 Oxygen (mole %) 2.46
6 2440.00 Carbon dioxide (mole %) 8.29
7 2118.21 Water (mole %) 18.17
 J. Purbolaksono et al. / Corrosion Science 51 (2009) 1022–1029 1025

Table 9 (10) are used to calculate the scale thickness of X1a for the service
The convection coefficients hg on external surface of the boiler tube. hours of 1 h and the scale thickness of for the service hours of
Temperature (°C) 800 900 1000 250 h (see Table 10) using the average temperature of Tave1. Next,
Outer diameter of tube (m) by subtracting one from the other, the scale increase of
0.0508 0.0508 0.0548 0.0508 DX1(= X1b  X1a) is determined and a new scale thickness of
Convection coefficient hg (W/m2 °C) 126.01 130.96 133.47 135.62 X1(= X0 + DX1) is obtained.
Step 2. The average temperature of Tave2 is then determined
from the numerical modeling with the new scale thickness on
the inner surface. The average temperature of Tave2 obtained
from the average of the inner surface and the scale/metal inter-
face temperatures is then used to calculate the incremental scale
thickness from 250 to 500 h using Eqs. (9) and (10). For service

ss
Field observations

ne
year’s exposure hours of 500 h, P is calculated using Eq. (10) and X2b is found

ck
thi
from Eq. (9). For service hours of 250 h, P is calculated using
le
a
Eq. (10) and X2a is found from Eq. (9). Subtracting one from
Sc
500
621 0C - 5 yr on the other (X2b  X2a) produces the incremental scale formation
ati

538 0C - 100,000 hr from 250 to 500 h, which is added to X1 to give a new scale
tr
ne

thickness of X2. Repeat Step 2 for further predictions up to the

Pe

250
538 0C - 5 yr maximum of 160,000 h with the steps of time shown in
Scale thickness, µm

Table 10.
150 538 0C - 1 yr Since the initial increment of time determines the further
estimation results, it is proposed to use the steps of time as
100
shown in Table 10. A smaller increment of time might provide
a better estimation, whereas a bigger increment of time for ini-
tial iteration may be resulting in inaccuracy estimation or less
50
conservative prediction. For estimations starting from the service
Laboratory data
hours of 20,000 h, an increment of time may be proposed to be
X 538 0 C/ 2500 hr taken at every 20,000 h. There is general agreement that the
25
Y 621 0 C/ 2500 hr growth of thick scales on iron follows a parabolic-rate law [7]
Z 649 0C/ 2500 hr and [14]. However, in the extreme conditions with very high
steam and flue gas temperatures the scale growth will initially
follow a parabolic-rate and then tends to follow an exponen-
Parameter 34,000 36,000 38,000 40,000 42,000
tial-rate [3].
Finite element models are generated according to the geome-
P = (9/5 T + 492) (20 + log t)
try of the tube, the scale thickness and heat transfer parameters
Fig. 2. Steam-side scale formation for ferritic steels of 1–3% chromium correlated governing the problem. It is important to note that all the geo-
with the Larsen-Miller parameter [8]. metrical units used for modeling are in m. Hence, the meshing
size control of 0.0001 is used to generate the 2D solid triangular
elements in order to allow the model having appropriate size of
numerical simulation corresponding to the given running hours
elements. The properties of the elements are then defined as 2D-
and scale thickness. In this work, the simulations performed for
axisymmetric solid elements. As stated in Step 1, the finite ele-
the predictions are made up to the maximum of 160,000 h with
ment model is produced in the absence of oxide scale. The bulk
an increment of time as shown in Table 10. The iterative proce-
temperature and convection coefficient hg of the flue gas are ap-
dures used to determine scale thickness as a function of time and plied on the right edge of the model (see Fig. 1). Next, the bulk
temperature are as follows: temperature and convection coefficient hs of the steam are ap-
Step 1. The design temperature for the steam is set to Ts at the plied on the left edge of the model. In the presence of the oxide
inlet of reheater or superheater tube. From the numerical simula- scale as stated in Step 2 the model will have two domain areas,
tion in the absence of scale (X0), the average temperature of Tave1 i.e. scale and tube metal. In order to make connectivity of the
is the temperature on the inner surface of the tube. Eqs. (9) and
domain areas at scale/metal interface, a merge-size control of
0.00001 is used. The merge-size control should considerably be
smaller than the meshing size control. The bulk temperature
Table 10
Total time after the steps of time used in the iterative procedure. and convection coefficient hg of the flue gas are applied on the
right edge of the tube metal region and the bulk temperature
Step h
and convection coefficient hs of the steam are applied on the left
1 250 edge of the oxide scale region. Fig. 3 shows the isometric view of
2 500
the 3=4 expansion model from 2D-axisymmetric model and the
3 1000
4 2500 estimated temperature distribution in Tube 1 with the steam
5 5000 temperature of 540 °C and the flue temperature of 900 °C for
6 10,000 the service hours of 60,000 h.
7 20,000
8 40,000
9 60,000
10 80,000 3. Results and discussion
11 100,000
12 120,000 Results of scale thickness predictions over the increasing of ser-
13 140,000
vice hours obtained using the iterative procedures with different
14 160,000
heat transfer parameters as specified in Tables 2–4 are plotted in
1026 J. Purbolaksono et al. / Corrosion Science 51 (2009) 1022–1029
587.384
584.688
581.993
579.297
576.601
573.905
571.209
568.513
565.817
563.131
3/4 expansion 2D axisymmetric
Fig. 3. Temperature distribution in Tube 1 with the steam temperature of 540 °C
and the flue temperature of 900 °C for the service hours of 60,000 h.
Figs. 5–8. The corresponding temperatures at scale/metal interface
are presented in Table 11. Some features may be deduced accord-
ing to the specified parameters as follows:
– Geometries of the tubes. Three different sizes of tube as
shown in Table 1 are used in this study. It can be seen from
Fig. 5, the largest estimated scale thickness at every service
hours is found in Tube 2. According to the geometry of the
tubes shown in Table 1, inner radius of Tube 2 is bigger than
the radius of Tube 3; the tube thickness of Tube 2 is larger
than the thickness of Tube 1. It is clear that the geometry
of tube influences the growth of oxide scale. The thinner
tube has less incremental scale formation. It means that the
metal temperature is also less increasing (see Table 11).
However, the thinner tube causes higher operational hoop
stress. Conversely, the thicker tubes have the greater growth
of scale as a result of higher temperature increase. It may
result in changes of microstructure of the tube metal and
cause material degradation.
– Mass flow rate of steam. Mass flow rate is taken into account
in determining coefficient of the forced convection. The
lower mass flow rate of steam at a design temperature will
increase the growth of oxide scale on the inner surface. It
may indicate that the poor or impaired mass flow rate of
the steam, e.g. blocking steam flow, will cause significant
increase of the scale growth as a result of higher tempera-
ture developed in scale/tube metal interface (see Table 11).
Fig. 4. Inline and staggered arrangements of the bare tubes.
0.45
0.40
Scale thickness, mm
0.35
0.30
0.25
0.20
0.15
Model 3, Tube 1
0.10 Model 5, Tube 2
Model 6, Tube 3
0.05
0.00
0 20000 40000 60000 80000 100000 120000 140000 160000 180000
Service time, h
Fig. 5. Estimated scale thickness as a function of time with the steam temperature
of 540 °C for different tube geometries (see Table 3).
The effect of different mass flow rate of steam on scale
growth is shown in Fig. 6.
– Steam temperature. For design temperature of 605 °C shown
in Fig. 7, the curve tends to have higher increasing of scale
thickness at the same of service hours. This feature is related
to the reduced oxidation resistance if the tube metal has
higher temperature. The increasing of scale thickness conse-
quently becomes larger, and it will result in the greater tem-
perature increase in the tube metal (see Table 11).
– Convection coefficient on the outer surface of tube and flue gas
temperature. It can be deduced from Fig. 8, the higher con-
vection coefficient and higher flue gas temperature lead to
a larger increase of scale thickness. In order to estimate
the growing of scale in a certain tube, the convection coeffi-
cient and flue gas temperature used in finite element model-
ing may be specified according to the particular location of
the tube, for instance, the tube situated facing directly to
the furnace section has different convection coefficient and
flue gas temperature.
– Increments of time. Fig. 9 shows that a bigger increments
results in less conservative estimation. Initial increments
are particularly important to be treated since they will sig-
nificantly affect further estimation during the iterations.
In order to validate the reference data used shown in Fig. 2 and
the results obtained, comparison of the data for a long-term expo-
sure needs to be made. However, the heat transfer parameters and
 J. Purbolaksono et al. / Corrosion Science 51 (2009) 1022–1029 1027

0.80 0.60

Model 1, 3600 kg/h

0.70
Model 2, 720 kg/h 0.50

Scale thickness, mm
Scale thickness, mm

0.60
0.40
0.50

0.30
0.40

0.30 0.20
0
Model 1, 800 C
0.20 0
Model 3, 900 C
0.10
0
0.10 Model 4, 1000 C
0.00
0 20000 40000 60000 80000 100000 120000 140000 160000 180000
0.00
0 20000 40000 60000 80000 100000 120000 140000 160000 180000 Service time, h
Service time, h
Fig. 8. Estimated scale thickness as a function of time for different flue gas
Fig. 6. Estimated scale thickness as a function of time with the steam temperature temperatures (see Table 3).
of 540 °C for different mass flow rates of steam (see Table 3).

The corresponding Larsen-Miller parameter P for service time of

1.60 100,000 h at 538 °C is 36,500. It can be obtained from Fig. 2 that the
0
corresponding scale thickness is around 140–200 lm (0.14–
Model 1, 540 C
1.40
0
0.2 mm). It can be seen from Figs. 5–8 that the estimated results
Model 7, 605 C
Scale thickness, mm

for the models with steam temperature of 540° and mass flow rate
1.20
of 3600 kg/h are ranging from 0.25 to 0.47 mm and showing higher
1.00
than those shown in Fig. 2. The differences might be due to differ-
ent parameters used. According to Figs. 5–8, parameters governing
0.80 the problem such as higher mass flow rate of steam, lower steam
and flue gas temperature and smaller inner diameter and thickness
0.60 of the tube may reduce the scale growth rate.
Further, it is important to compare the estimated values with
0.40
the actual data of the available reports [15] and [16] which were
0.20 obtained from dimensional measurements for the scale thickness
of the as-received reheater tube samples taken from Kapar Power
0.00 Station Malaysia. Two different cases with different tube diameters
0 20000 40000 60000 80000 100000 120000 140000 160000 180000
from two different locations are used. The detailed samples used
Service time, h
are shown in Table 12. Sample for Case 1 was located at the first
Fig. 7. Estimated scale thickness as a function of time for different steam row facing to the burner while the sample for Case 2 was located
temperatures (see Table 3). a relatively further from the burner. Operating steam temperature
of both tubes is 576 °C. The flue gas temperatures were reported
ranging from 800–900 °C. Parameters used to determine gas mass
geometry of the specimens for the reference data (Fig. 2) are not velocity G and the estimated convection coefficients hs and hg for
given. It might give the comparison to be compromised. Scale the internal and external surfaces are shown in Tables 13 and 14,
thickness for service time of 100,000 h at 538 °C is used for com- respectively. The estimated scale thickness and the actual data
parison with the estimated results obtained using the proposed are plotted in Fig. 10. It can be seen that the scale thickness for
technique. the actual data of Case 1 is relatively close to the estimated scale

Table 11
Temperatures (°C) at the scale/metal interface of the tubes.

Service time (h) Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7
0 557.06 592.73 564.47 572.28 566.49 562.89 617.43
250 558.58 595.06 567.01 576.22 569.35 565.30 621.39
500 559.05 595.80 567.81 577.47 570.24 566.13 622.68
1000 560.21 596.73 568.81 579.06 571.37 567.20 624.33
2500 560.68 598.35 570.54 581.83 573.32 569.04 627.19
5000 561.68 599.95 572.27 584.61 575.27 570.87 630.05
10,000 562.95 601.98 574.46 588.17 577.74 573.18 633.66
20,000 564.53 604.52 577.23 592.74 580.88 576.13 638.21
40,000 566.52 607.73 580.76 598.65 584.88 579.85 643.92
60,000 567.92 609.98 583.27 602.96 587.74 582.52 647.96
80,000 569.04 611.78 585.31 606.50 590.05 584.68 651.17
100,000 570.00 613.30 587.04 609.55 592.04 586.53 653.88
120,000 570.82 614.62 588.57 612.27 593.78 588.15 656.25
140,000 571.57 615.80 589.95 614.75 595.35 589.61 658.35
160,000 572.24 616.88 591.21 617.04 596.79 590.95 660.26
1028 J. Purbolaksono et al. / Corrosion Science 51 (2009) 1022–1029

0.35 0.90

0.30 0.80
Scale thickness, mm

0.25
0.70

Scale thickness, mm
0.60
0.20

0.50
0.15

Model 1 with steps of time as specified in Table 10 0.40

0.10 Estimated Case 1, 8000C
Model1 with increment of time at every 20,000 h
0.30 Estimated Case 1, 9000C
0.05 Actual Case 1 [15]
0.20 Estimated Case 2, 8000C
0.00
0.10 Estimated Case 2, 9000C
0 20000 40000 60000 80000 100000 120000 140000 160000 180000
Actual Case 2 [16]
Service time, h
0.00
Fig. 9. Estimated scale thickness as a function of time for different increment of 0 20000 40000 60000 80000 100000 120000 140000
time.
Service time, h

Fig. 10. The estimated scale thickness and the actual data.
Table 12
Oxide scale thickness and geometry of the as-received tubes.

Case Inner radius Tube thickness Service Year of Scale 0.35

(m) (mm) time (h) failure thickness, mm
1 0.0225 4 92,525 2001 0.68 0.30
Scale thickness, mm

2 0.0219 3.5 117,522 2003 0.58

0.25

0.20
Table 13
Parameters used to determine gas mass velocity G for validation of actual data.
Model 1
0.15
B = 0.5
Gas flow (kg/h) 500,000
B = 1.0
Number of tube wide 50
0.10 B = 2.0
Transverse pitch (m) 0.1016
Tube length (m) 8
0.05

0.00
Table 14 0 20000 40000 60000 80000 100000 120000 140000 160000 180000
The estimated convection coefficients hs and hg for internal and external surfaces, Service time, h
respectively.
Fig. 11. The oxide scale growths for different constants B for the approximated
Case hs (W/m2 °C) hg (W/m2 °C)
increases of temperature.
1 1990.59 125.31
2 2053.65 126.01

growths for different B tend to follow parabolic-rate law. For in-

thickness for the flue gas temperature of 900 °C. It is apparently
shown to be in good agreement with data with respect to the loca-
tion of the tube since the higher temperature of the flue gas will
accelerate scale growth over period of time. Bigger inner diameter
and thickness of the tube may also accelerate scale growth. The
scale thickness for Case 2 also fairly agrees with the estimated
scale thickness for the flue gas temperature of 800 °C. It is essential
to make appropriate monitoring of the heat transfer parameters
which may govern the problem from time to time.
In order to show that the growth of thick scales on iron gener-
ally follows a parabolic-rate law as stated by French [7] and Uhlig
[14], a temperature increase DT as a function of a scale thickness
increase DX over a period of time may be used as
DT ¼ B DX
where B is a constant. DT in Eq. (11) may be introduced by adding it
with the temperature on the inner surface stated in Step 1 to re-
place the average temperature in Step 2. Model 1 is used to examine
this procedure. It can be seen from Fig. 11 that the oxide scale
stance, if B is set on trial to be 1 for Model 1, the estimations for
the scale thickness using the given DT over period of time are
shown to be in very good agreement with those obtained using pro-
cedures given in Steps 1 and 2. It means that the approximated va-
lue for B can be well determined for future predictions after
estimations of scale thickness over period of time for the corre-
sponding condition using Steps 1 and 2 have been made.
The oxide scale thickness developed on the inner surface of
reheater and superheater tubes over period of time can be esti-
mated by utilizing the empirical formula correlating scale thick-
ness with Larsen-Miller parameter and the finite element
modeling. In the finite element simulations, the data for tube
geometries and all the heat transfer parameters that might govern
the problem are taken into account in order to determine the tem-
perature distribution in tube metal. The temperature increase is an
ð11Þ
important influence on the scale growth rates. Data for heat trans-
fer parameters according to variations in the operating conditions
in the system may be introduced into the iteration procedure. Bet-
ter estimation of the scale growth could be obtained, provided that
all the heat transfer parameters used are well specified.
 J. Purbolaksono et al. / Corrosion Science 51 (2009) 1022–1029
4. Conclusions
Estimation on the oxide scale growth in superheater and
reheater tube utilizing the empirical formulae and the finite ele-
ment modeling was proposed. An iterative procedure was used
to determine scale thickness as both temperature and time in-
crease. The scale thickness and the corresponding temperatures
at scale/metal interface over period of time for two different design
temperatures for steam and different heat transfer parameters
were presented. The scale growths were influenced by the tube
geometry and heat transfer parameters such as steam tempera-
tures, mass flow rates of steam, flue gas temperatures and convec-
tion coefficients on the outer surface of tube. The procedures may
provide better estimation on the oxide scale growth, provided that
all the heat transfer parameters are well specified.
Acknowledgements
This work is supported by the Ministry of Science Technology
and Innovation, Malaysia through the research projects of IRPA
09-99-03-0033 EA001 and Sciencefund 04-02-03-SF0003. The
authors wish to thank Universiti Tenaga Nasional, Kapar Energy
Ventures Sdn Bhd and TNB Research Sdn. Bhd Malaysia for permis-
sion of utilizing all the facilities and resources during this study.
References
[1] R.D. Port, H.M. Herro, The NALCO Guide to Boiler Failure Analysis, Nalco
Chemical Company, McGraw-Hill Inc., 1991.
1029
[2] D.R.H. Jones, Creep failures of overheated boiler. Superheater and reformer
tubes, Engineering Failure Analysis 11 (2004) 873–893.
[3] K.J. Clark, S.R. Paterson, T.W. Rettig, Remaining Life Assessment of Superheater
and Reheater Tubes, Aptech Engineering Services Inc., Sunnyvale, California,
1988.
[4] R. Viswanathan, S.R. Paterson, H. Grunloh, S. Gehl, Life assessment of
superheater reheater tubes in fossil boilers, Journal of Pressure Vessel
Technology, Transactions of the ASME 116 (1) (1994) 1–16.
[5] R. Viswanathan, H. Grunloh, S.R. Paterson, S. Gehl, Life assessment of
superheater/reheater tubes, American Society of Mechanical Engineers,
Pressure Vessels and Piping Division (Publication) PVP 240 (1992) 103–122.
[6] T.J. Wardle, Creep-rupture assessment of superheater tubes using
nondestructive oxide thickness measurements, in: International Conference
on Life Management and Life Extension of Power Plant, China, May 2000.
[7] D.N. French, Metallurgical Failures in Fossil Fired Boilers, A Wiley-Interscience
Publication, John Wiley and Sons Inc., New York, 2000.
[8] I.M. Rehn, W.R. Apblett Jr., J. Stringer, Controlling steamside oxide exfoliation
in utility boiler superheaters and reheaters, Mater Performance (1981) 27–31.
[9] P.J. Ennis, W.J. Quadakkers, Implications of steam oxidation for the service life
of high-strength martensitic steel components in high-temperature plant,
International Journal of Pressure Vessels and Piping 84 (2007) 82–87.
[10] F. Starr, J. Castle, R. Walker, Potential problems in the identification of the root
cause of superheater tube failures in 9Cr martensitic alloys, Materials at High
Temperatures 21 (3) (2004) 147–160.
[11] F.P. Incropera, D.P. DeWitt, Introduction to Heat Transfer, third ed., John Wiley,
1996.
[12] J.E. Bringas, Handbook of Comparative World Steel Standards ASTM DS67A,
second ed., 2002.
[13] V. Ganapathy, Industrial Boilers and Heat Recovery Steam Generators: Design,
Applications, and Calculations, Marcel Dekker, New York, 2003.
[14] H.H. Uhlig, Corrosion and Corrosion Control, second ed., Wiley, New York,
1971.
[15] J. Ahmad, Technical memorandum of Kapar Power Station Sdn Bhd Malaysia:
Remaining Life Assessment of Reheater Tube, TNB Generation Sdn Bhd, Mei
2001.
[16] J. Ahmad, Technical Memorandum of Kapar Power Station Sdn Bhd Malaysia:
Report of Reheater Tube Failure, TNB Generation Sdn Bhd, January 2004.