J Fail. Anal. and Preven.

(2018) 18:189–198
https://doi.org/10.1007/s11668-018-0400-1

TECHNICAL ARTICLE—PEER-REVIEWED

Analysis of Fatigue–Creep Crack Growth in the Superheater

Header of a Power Plant Boilers and Estimation of Its Remaining
Lifetime
Seyed Ebrahim Moussavi Torshizi . Ali Jahangiri

Submitted: 28 October 2017 / in revised form: 21 December 2017 / Published online: 16 January 2018
Ó ASM International 2018

Abstract As superheater header is exposed to high ef Uniaxial creep rupture ductility (in % strain)
pressure and temperature in power plant boilers, it is one of e_ref Creep strain rate at reference stress (1/s)
the most sensitive parts in a power plant.Cracks may form rref Reference stress (MPa)
where tubes and main steam outlet (nozzle) are connected Kc Stress intensity factor corresponding to the applied
pffiffiffiffi
to reservoir. This article examines a quarter-circle crack at loading (MPa m)
the corner of a nozzle junction and its propagation steps A Norton power law creep constant in Eq 3 (–)
under the influence of simultaneous interaction of creep n Norton stress index in power law creep Eq 3 (–)
and fatigue. Header loading in each cycle includes transient Eamb Young’s modulus at ambient temperature (GPa)
steps (increase and decrease in temperature and pressure) at Eat Young’s modulus at operation temperature (GPa)
the beginning and at the end of a cycle and intermediate pffiffiffiffi
DK Stress intensity coefficient range (MPa m)
steady state (fixed pressure and temperature) during oper- pffiffiffiffi
K Stress intensity coefficient (MPa m)
ation. For crack growth calculations, stress distribution in a pffiffiffiffi
KIC Fracture toughness (MPa m)
track-free part was calculated. Fatigue–creep crack growth S Stress distribution (Pa)
was achieved using crack growth rules and the remaining
lifetime was obtained. Research result shows that creep
phenomenon is responsible for maximum crack growth.
Introduction
Keywords Fatigue  Creep  Header 
Remaining lifetime estimation  Crack propagation Headers are the components that are placed in the entrance
and discharge of different parts of a boiler including
List of symbols superheaters, reheaters, economizers, waterwalls, etc. They
a Creep crack radius (mm) distribute water or steam into different pipes in the input
da Creep crack growth (mm) zone and collect water or steam in the output zone. Headers
a_ Creep crack growth rate (m/h) are cylindrical and they are made by rolling and sheet
da Crack growth rate versus number of cycles welding using an appropriate alloy, to which a series of
dN pipes is either axially or radially connected. After flowing
(mm/cycle)
C 
Steady state creep characterizing parameter steam into header chamber, their mixture makes steam
(MPa m/h) temperature uniform so as to be used in the other portion.

Cref Value of C determined from reference stress Headers are not exposed to the combustion flue gases
methods (MPa m/h) and radiative heat transfer; in fact, they are outside of
combustion chamber. Figure 1 is the overall schematic of
the header with its connected superheaters. There are var-
S. E. Moussavi Torshizi  A. Jahangiri (&) ied headers in a boiler, but the important headers found in
Faculty of Mechanical and Energy Engineering, Shahid Beheshti
University, A.C., Tehran, Iran most boilers are classified into the high- and low-temper-
e-mail: a_jahangiri@sbu.ac.ir ature headers categories.

123
190 J Fail. Anal. and Preven. (2018) 18:189–198

Fig. 2 The cracks formed in the header [1]

results of creep–fatigue crack test can be described by

fracture mechanics parameters.
Nakoneczny and Schultz presented [3] an overall pro-
cedure for inspecting and calculating the lifetime of high-
temperature headers.
Yatomi et al. [4] numerically studied the creep crack
growth in a fracture mechanics specimen and stated that the
creep-elastic modeling is not reliable in carbon–manganese
steels with considerable rate of plastic strains and relatively
Fig. 1 An input–output header with its connected superheaters
large applied load.
Wasmer et al. [5] studied creep crack nucleation and its
High-temperature headers operate in a temperature
growth in thick-walled pipes with inner pressure. They
higher than the creep temperature of the header material.
used the global and local reference stress methods to
Among power plant headers, superheater outlet headers
evaluate C*. The test proved that the results obtained from
and reheater outlet headers are classified as high-tempera-
applying global reference stress formula on uniaxial creep
ture headers. As the operational temperature of headers is
properties were more favorable than those of the local
higher than the creep region, they are made of high-
reference stress [11].
strength alloys whose metallurgical structure includes fer-
Javanroodi and Nikbin [6] conducted some studies on
rite–perlite or ferrite–bainite.
the concepts of fracture mechanics of creep–fatigue crack
Crack formation in power plant parts is one of the
growth in lifetime estimation. They used a model based on
problems, which are usually found in inspections (Fig. 2).
C* and uniaxial creep ductility to predict crack growth and
However, a long period may elapse between crack obser-
proved that the model was consistent with the experimental
vation and its critical moment. Therefore, calculation of
results.
reliable lifetime for operation (remaining lifetime)
Due to the sensitivity and economic value of header
becomes important. The sections where main inlet or outlet
boiler in power plants and high probability of cracks in
steam tubes are connected to a superheater header highly
superheater header, this research discusses estimation and
subject to crack formation due to exposure to high tem-
evaluation of the remaining lifetime of power plant header
perature and pressure and their changes during (starting up
boiler with a crack in its nozzle, which is in fact the same
and shutting down) turning on/off.
as damage-tolerant design discussion. It is assumed in this
Piard et al. [1] studied the numerical simulation of
design that the cracks are formed by production process,
fatigue–creep crack growth in nickel superalloys. They
creep, or fatigue. Then fracture mechanics analyses and
examined the growth of a one-dimensional crack at 750 °C
examination specify whether such cracks will be large
in simulation. The periods they considered for loading
enough to lead to failure before detection in periodical
included the periods encompassing up to 1000 s.
inspections. This method can be used for the materials with
Granacher et al. [2] reported creep–fatigue behavior for
slow crack growth and high fracture toughness, and ‘‘Leak
two highly used steels in power plants. They tested some
Before Break’’ is an application of this philosophy in
CT specimens of bainitic steel for turbine rotor and
designing tanks [7].
martensitic types of steel for pipes and expressed that the

123
J Fail. Anal. and Preven. (2018) 18:189–198 191

Biglari and Nikbin [8] numerically predicted the crack

progression and the environmental acceleration of material
degradation compared with experimental remark.
Narasimhachary and Saxena [9] investigated the crack
growth behavior of 9Cr–1Mo (P91) steel at 625 °C under
fixed load magnitude conditions.
Sakane and Tokura [10] experimentally studied the
effect of stress biaxiality on rupture lifetime and creep
damage for type 304 stainless steel.
As the parts can be used after detection of crack in them
to allow crack length to reach a maximum permissible
value, it is highly important to determine the time the crack
reaches maximum permissible length and to study carefully
the effect of each factor effective in crack growth and
remaining lifetime. Since the basis for the remaining life-
time of each part is the type of damages related to its Fig. 3 View of the header and nozzle
application and environmental conditions and the fact that
creep and fatigue are the major mechanisms for header Due to the high thermal changes in the header, a fixed
failure, remaining lifetime estimation and crack growth value cannot be selected for the properties, and each
will be studied under creep and fatigue mechanisms. A thermal and mechanical property should be defined as a
superheater header is exposed to thermal loading and inner function of temperature. Table 1 shows the mechanical and
pressure during utilization. Given the multiplicity of thermal properties for the header. (The header is made of
operations, shutdown of the boiler, and the relatively long 2.25Cr1Mo, which is called ‘‘P22.’’)
period in which the header is exposed to high temperature, Density and specific thermal capacity do not undergo
it is necessary to analyze header completely and accurately, tangible changes at different temperatures and hence
detect its critical places, and study carefully the effect of assume almost constant values of 7820 kg/m3 and 461 J/
creep and fatigue factors. kg °C, respectively [11].
This research has attempted to provide a procedure that Start-up (increasing temperature and pressure) and
can be used in studying and analyzing the nozzle crack of operation (fixed pressure and temperature) and shutting
the superheater header of power plant boilers. The proce- down (reducing pressure and temperature) processes cause
dure may be used to evaluate the headers subject to cracks fatigue and creep in a boiler components and so the damage
in nozzle connection and to prepare a correct schedule for caused by creep and time is directly related. Therefore, to
inspection, maintenance, reconstruction, or replacement of evaluate the lifetime elapsed in creep, all the operation
headers. time of the boiler in hours was divided into boiler start-up
This research discusses the header with a crack at its frequencies to obtain its average operation time in any
longitudinal cross section on nozzle junction and calculates operation. This way, a complete cycle starts by start-up and
its remaining lifetime. ends by shutting down. Boiler background shows that the
cycle is approximately 276 h. Figures 4 and 5 show the
diagram for changes of temperature and pressure of the
Header Specifications and Loading Conditions steam discharging into superheater in cold start-up. Chan-
ges of temperature and pressure at boiler shutting down
The header and nozzle were made of the cylinders with the time were assumed in the reverse order of the changes of
inner radius and outer radius of 360 and 610 mm, respec- temperature and pressure in start-up mode.
tively (Fig. 3).
As the behavior of each part under varied loading con-
ditions is directly related to the properties of that material, Creep Crack Growth
it is necessary to correctly identify the relevant properties
using analysis and they should be defined in proportion to With respect to the header thickness and the fact that plane
the analysis and loading conditions. As thermal and strain conditions are dominant in the deepest point of the
mechanical loading are performed simultaneously, crack, Eq 1 will be used for calculating creep crack growth
mechanical and thermal properties are needed at the same [12].
time in header and nozzle analysis and they should be
defined.

123
192 J Fail. Anal. and Preven. (2018) 18:189–198

Table 1 Mechanical and thermal properties for P22 at different temperatures [11]
Temperature Modulus of Poisson’s Thermal expansion Thermal conductivity Yield stress
(°C) elasticity (GPa) ratio coefficient (9 106/°C) coefficient (W/m °C) (MPa)

20 214.8 0.29 11.5 35.77 207

100 205.9 0.29 12.57 36.74 188
150 201.8 0.29 13.18 36.44 182
200 197.8 0.29 13.67 36.11 177
250 193.9 0.2927 14.07 35.55 177
300 190 0.2951 14.38 35 177
350 185.1 0.2965 14.72 34.61 177
400 179.7 0.2981 15.01 34.19 177
450 174.5 0.3 15.22 33.36 176
500 169.1 0.3023 15.41 32.45 170
540 165 0.3039 15.58 31.89 162

600 Webster and Ainsworth [13] describe calculation

500 relations. In Eq 2, the creep strain rate is in the second
temprature (Co)

400
zone of creep and it is as Eq 3 by Norton law [5].
300 e_ref ¼ Arnref ðEq 3Þ
200 A and n are material constants whose values for P22 are
100 1060 and 6.38, respectively [14]. C* can be obtained as
0 Eq 4 through simplifying the relations [5] and creep crack
0 2 4 6 8 10 growth is also calculated.
time (hour)
C  ¼ Arn1 2
ref Kc ðEq 4Þ
Fig. 4 Changes of inlet steam temperature in start-up vs. time

18 Fatigue Crack Growth

15
Paris’ law is the commonest law for examining fatigue
pressure (MPa)

12
crack. Figure 5 has been proposed for ferritic steels at
9
temperatures between 100 and 600 °C [12]. In Eq 5, the
6
Eamb and Eat of Young’s modulus are at ambient tem-
3 perature and operation temperature.
0  
0 2 4 6 8 10 12 da Eamb DK 3
me (hour)
¼ 1:65ð1011 Þ ðm/cycleÞ ðEq 5Þ
dN Eat
Fig. 5 Changes of steam pressure in start-up vs. time
Creep–Fatigue Interaction in Crack Growth
15C 0:85
a_ ¼ ðEq 1Þ
ef To consider the simultaneous effect of creep and fatigue in
most metals with industrial application, it is often proposed
The unit for crack growth rate is m/h. ef , creep rupture to use a simple linear sum of creep crack growth and
softness, is uniaxial. Equation 2, which is called the fatigue crack growth. Equation 6 indicates creep–fatigue
reference stress method, can be used for obtaining C*. crack growth [6].
This method has been employed in many practical codes      
da da da
[5]. ¼ þ
dN dN creep dN fatigue
 2     ðEq 6Þ
 Kc 1 da da
Cref ¼ rref e_ref ðEq 2Þ ¼ þ ðm/cycleÞ
rref f dt creep dN fatigue

123
J Fail. Anal. and Preven. (2018) 18:189–198 193

1/f is the loading time duration in hours in the steady state. Maximum vertical stress occurred in the direction per-
pendicular to longitudinal plane of symmetry (x direction)
at the nozzle corner (Fig. 7), which indicates that crack will
Stress Intensity Coefficient start from the inner corner of the nozzle.
The model can be studied as far as different thermal or
Stress intensity coefficient (K) is the most important stress-displacement parameters are concerned. The
parameter used in crack growth discussions. ASME has parameter required for analyzing and examining crack
proposed a procedure for calculating the stress intensity growth is the stress distribution perpendicular to crack
coefficient occurring in thick-wall reservoirs, which will be plane, which is applied in calculating stress intensity
used in this article [14]. The standard states that stress coefficient.
distribution for the reservoirs exposed to internal pressure Figures 8, 9, and 10 show the vertical stress distribution
or thermal stress is calculated at longitudinal cross section along the direction perpendicular to the symmetry plane of
for a crack-free part in the direction of 45° (along Xn in the longitudinal section (x-axis) at the beginning of the
Fig. 6). start-up, in steady state (during utilization), and at the time
of shutdown.
Figures 8 and 10 show that stress distributions at the
Analysis of Finite Elements and Stress Distribution times of start-up and shutdown are completely different.
Calculation Due to the small inner pressure, expansion of inner layers
and resistance of outer layers, which are still cold, header
As shown in Fig. 7, a separate grid was applied to the inner points are exposed to a very high compressive stress
header–nozzle connection after separating the remaining at the time of start-up. However, at the time of shutdown
section. After mesh study stage, there were 55,344 ele- during which the inner layers undergo contraction due to
ments and 11,198 nodes. temperature reduction, the outer layers are still at a high
Transient thermal stress analysis was achieved for the temperature and show resistance against contraction of
header–nozzle using FEM code to obtain stress distribu- inner layers, consequently causing a tensile stress in the
tion. Analysis was performed on the header and nozzle inner layers. As the above figures show, the header inner
model with respect to the plane of symmetry using pressure has the maximum effect on the stress distribution
tetragonal elements. With respect to the start-up conditions, of the header inner points. An inner pressure of 17.4 MPa
the analysis showed that the steady state was established generates a 143 MPa stress at the nozzle–header connec-
almost 11 h after start-up and stress distribution became tion. Nevertheless, the stress distribution caused by inner
fixed with the start of deactivation process of the unit. pressure in thick-walled tanks is such that it is expected
that the circumferential stress in the inner layer would
exceed inner pressure (Fig. 11). However, the multiple

Fig. 6 The longitudinal cross

section of header and its
connected nozzle

123
194 J Fail. Anal. and Preven. (2018) 18:189–198

Fig. 7 Vertical stress

distribution along x-axis at
steady state

Fig. 8 The stress distribution along the direction perpendicular to the

Fig. 9 The stress distribution along the direction perpendicular to the
symmetry plane of the longitudinal section at the beginning of start-
symmetry plane of the longitudinal section in the steady state
up

increase in circumferential stress at the nozzle–holder Therefore, the header–nozzle connection is more sensi-
connection point was unexpected. The severe increase in tive than other points and experiences worse working
environmental stress at the corner of the nozzle may be due conditions. Figures 12 and 13 show the stress diagram
to the concentration of stress and the interaction of header during start-up and shutdown for six points along the 45°
inner pressure and nozzle inner pressure. axis. Point 1 is the inner corner of the nozzle–header
connection and other points are placed at equal distances
from the inner corner to header outer surface.

123
J Fail. Anal. and Preven. (2018) 18:189–198 195

150

point 1
100

point 2

50

stress (MPa)
point 3

0
0 2 4 6 8 10 12 point 4

-50
point 5

-100
point 6

-150
me (hour)

Fig. 12 Changes of stress in nozzle–header connection during start-

up

Fig. 10 The stress distribution along the direction perpendicular to 170

the symmetry plane of the longitudinal section at the time of
150
shutdown
point 1
130
40
110 point 2
stress (MPa)

35 90
point 3
hoop stress (MPa)

70
30
50 point 4

25 30
point 5
10
20
point 6
-10 0 1 2 3 4 5
15
-30
0 25 50 75 100 125 me (hour)
thickness (mm)
Fig. 13 Changes of stress in nozzle–header connection points during
Fig. 11 Stress distribution along header thickness far from nozzle boiler shutdown

The points closer to the inner surface of the header inner layers during operation when temperature and pres-
experience the maximum stress when the boiler shuts down sure are fixed and the outer layers resist against the
(Fig. 13), and the thermal stresses caused by steam tem- contraction of the inner layers and tolerate compressive
perature reduction are added to the stresses caused by inner stress as the shutdown process starts and temperature and
pressure. At the start-up time, the points tolerate com- pressure decrease. As a result, thermal compressive stress
pressive stress due to the resistance of outer layers against in outer areas is dominant over the tensile stress caused by
their expansion. Therefore, the stresses created during inner pressure. The studies on the fatigue of the points
shutdown of the boiler are important in the study of the show that thermal stress is of paramount importance in
fatigue close to inner surface. start-up and it is the controlling agent in these points.
However, maximum tensile stress for the points close to To calculate the stress intensity coefficient of Eq 4, it is
the outer surface of the header occurs in start-up time and required to calculate the stress distribution perpendicular to
stress in the points reaches its maximum during a cycle. As the crack plane (direction of x in Fig. 6) on 45° direction
shown in Fig. 12, it can be expected that circumferential (direction of Xn in Fig. 5) at the steady state. Figure 14
stress in the outer layers is significantly lower than the shows the stress distribution.

123
196 J Fail. Anal. and Preven. (2018) 18:189–198

However, specific stress changes are observed in the Final Crack Length
transient mode in which crack growth is related to fatigue
process. Figure 15 shows stress changes for the inner and The final crack length in calculations of the remaining
outer corners of the nozzle. When putting the boiler into lifetime of the cracked reservoirs is 25% of reservoir
operation, a considerable tensile stress occurs in the outer thickness or 25% of track critical length [14]. A crack
corner of the nozzle; however, the inner corner is subjected becomes critical when stress intensity coefficient (K)
to compressive strength (Fig. 15). With the temperature exceeds KIC. Fracture toughness (KIC) for ferritic steels at
and pressure increasing, the stresses subject to some fluc- high temperatures is 220 MPaHm [12]. When a crack
tuations to establish the steady state. develops as much as 80% of thickness (the method for
As Fig. 15 shows, maximum stress of the inner and calculating stress intensity coefficient in ASME standard is
outer points never occurs simultaneously. To obtain stress valid as much as the crack development of 80%), stress
distribution to calculate stress intensity coefficient of Eq 5, intensity coefficient will be 38 MpaHm. Therefore, as long
this research recommends consideration of maximum stress as the relations to calculate stress intensity coefficient are
on the points in the direction of 45° (direction Xn) at dif- valid, crack length will not reach the critical size. As a
ferent times. The result is shown in Fig. 16. result, it seems that the final crack length can be considered
As two stress distributions were obtained, two coeffi- as much as 25% of nozzle cross section thickness. With
cients of stress intensity will be calculated, one for steady respect to the calculation of crack growth at the angle of
state and studying creep crack growth and the other for
transient mode and studying fatigue crack growth.

200

stress (MPa) 150

100

50
150

120 0
0 30 60 90 120 150 180
thickness (mm)
stress (MPa)

90

60 Fig. 16 Stress distribution in the direction of 45° in transient mode

30

0
a0 , N=1
0 30 60 90 120 150 180
thickness (mm)
a = a + da
a
Fig. 14 Stress distribution in the direction of 45° at the steady state N=N+1

inner corner outer corner Kfatigue , Kcreep

150
da =F(a) dN
100

50
stress (MPa)

Kfatigue , Kcreep < KIC

da/dN =F(a) YES
or
0
0 3 6 9 12 a < afin
-50
NO
-100

N
-150
me (hour) (remained life)

Fig. 15 Changes of stress in the inner and outer corners of the nozzle Fig. 17 The schematic view for calculation of crack growth and
in operation remaining lifetime

123
J Fail. Anal. and Preven. (2018) 18:189–198 197

for a new crack. Figure 17 shows the calculations

schematically.
Header remaining lifetime is then achieved through
obtaining the cycles required for the track to reach the final
lifetime. In the calculations, crack primary radius is
assumed 1 mm and its corresponding stress intensity
coefficient is greater than threshold intensity coefficient.
This way, crack growth will start confidently. After making
necessary analyses, the header remaining lifetime for such
a track is about 420 cycles, which is approximately
13 years. Final radius of the crack will be 44.19 mm whose
43.50 mm is due to creep and only 0.69 mm is caused by
fatigue. Figure 18 shows the changes of crack versus time,
and Fig. 19 shows the changes of crack growth rate versus
number of cycles.

Fig. 18 The diagram of crack radius vs. time (104 h)

Results

With respect to the low number of on/off cycles of the unit

versus operation time and the fact that fatigue track
development solely depends on number of cycles, lifetime
reduction caused by fatigue is negligible as compared with
creep. Therefore, it is proposed to overlook fatigue impacts
in the analysis of cracked nozzles of the thick-wall reser-
voirs subject to high temperature such as nozzle of
superheaters and to analyze creep crack growth only.

References

1. A. Piard, D. Gamby, C. Carbou, J. Mendez, A numerical simu-

lation of creep–fatigue crack growth in nickel-base superalloys.
Eng. Fract. Mech. 71, 2299–2317 (2004)
2. J. Granacher, A. Klenk, M. Tramer, G. Schellenberg, F. Mueller,
J. Ewald, Creep fatigue crack behavior of two power plant steels.
Int. J. Press. Vessels Pip. 78, 909–920 (2001)
3. G.J. Nakoneczny, C.C. Schultz, Life assessment of high tem-
perature headers. American Power Conference, Chicago, IL
(United States) (1995)
Fig. 19 The diagram of crack growth rate vs. number of cycles
4. M. Yatomi, K. Nikbin, N. O’Dowd, Creep crack growth predic-
(mm/cycle)
tion using a damage based approach. Int. J. Press. Vessels Pip. 80,
573–583 (2003)
45°, the final radius of crack is 25% of reservoir thickness 5. K. Wasmer, K. Nikbin, G. Webster, Creep crack initiation and
along 45° direction, which will be 44.19 mm. growth in thick section steel pipes under internal pressure. Int. J.
Press. Vessels Pip. 80, 489–498 (2003)
6. F. Javanroudi, K. Nikbin, The fracture mechanics concept of
creep and creep/fatigue crack growth in life assessment. Int. J.
Remaining Lifetime Calculation Eng. Sci. 17, 1–7 (2006)
7. R.I. Stephens, A. Fatemi, R.R. Stephens, H.O. Fuchs, Metal
Fatigue in Engineering (Wiley, Hoboken, 2000)
For remaining lifetime calculation, a numerical method is
8. F.R. Biglari, K.M. Nikbin, Numerical predictions of carburisation
used for calculating crack growth rate for one cycle. The and crack evolution using a combined diffusion rate and
growth rate is then added to crack radius and stress remaining multi-axial creep ductility damage model. Int. J.
intensity coefficients, and crack growth rate is calculated Damage Mech. 1056789516628317 (2016)

123
198 J Fail. Anal. and Preven. (2018) 18:189–198

9. S. Narasimhachary, A. Saxena, Crack growth behavior of 9Cr– 12. R. Api, 579-1/ASME FFS-1 (American Petroleum Institute,
1Mo (P91) steel under creep–fatigue conditions. Int. J. Fatigue Houston, 2007)
56, 106–113 (2013) 13. G. Webster, R. Ainsworth, High Temperature Component Life
10. M. Sakane, H. Tokura, Experimental study of biaxial creep Predictment (Chapman & Hall, London, 1994)
damage for type 304 stainless steel. Int. J. Damage Mech. 11, 14. A. Boiler, ASME Boiler and Pressure Vessel Code: An Interna-
247–262 (2002) tional Code (American Society of Mechanical Engineers, New
11. J.R. Davis, ASM Specialty Handbook: Heat-Resistant Materials York, 1998)
(ASM International, Materials Park, 1997)

123