Engineering Failure Analysis 106 (2019) 104159

Contents lists available at ScienceDirect

Engineering Failure Analysis

journal homepage: www.elsevier.com/locate/engfailanal

Multi-axis low-cycle creep/fatigue life prediction of high-pressure

T
turbine blades based on a new critical plane damage parameter
Junzhou Huo, Debin Sun , Hanyang Wu, Weizheng Wang, lin Xue
⁎ ⁎

School of Mechanical Engineering, Dalian University of Technology, No. 2 Linggong Road, Ganjingzi District, Dalian, Liaoning 116024, China

ARTICLE INFO ABSTRACT

Keywords: To study the multi-axis low-cycle creep/fatigue life of high-pressure turbine blades, a multi-axis
New critical plane damage parameter low-cycle fatigue life prediction model based on a new critical plane damage parameter is pro-
High-pressure turbine blades posed in this paper. Firstly, based on the thermal-structural coupling analysis of high-pressure
Creep/fatigue interaction turbine blades, the law of stress-strain distribution under the complex load of centrifugal load,
Fatigue damage mechanism
temperature load and the aerodynamic load is obtained. Secondly, the multi-axis low-cycle fa-
Life prediction
tigue life prediction model and the L-M equation are applied to predict the fatigue life and creep
life of blade respectively. Furthermore, different experimental loading schemes are used to study
the fatigue damage mechanism for the creep/fatigue interaction under high temperature con-
dition, and a creep/fatigue interaction life prediction model is simultaneously provided. And
finally, the experiment results demonstrate the rationality of the multi-axis low-cycle fatigue life
prediction model and the creep/fatigue interaction should not be ignored.

1. Introduction

As a paramount part of an aero-engine, the turbine blades directly determine the safety and service life of the entire machine [1],
which suffer from not only low-cycle fatigue damage caused by cyclic loading and temperature generated during start-up and
shutdown, but also from creep damage caused by the centrifugal loading in a high-temperature environment during stable operation
period in the course of actual service. The number of turbine blades also increases the probability of fatigue damage, and the stability
and the reliability of blades operation directly affect the performance, durability and reliability of the aero-engine. Factors, such as
the special material characteristics, structural processes, extremely harsh working environments, and the coupling effect of complex
loads [2–5], make the failure mechanisms of turbine blades fatigue extremely complicated, which greatly increase the difficulty of
turbine blades life prediction. The accuracy of the fatigue life prediction of turbine blades has not been effectively solved, which
seriously restricts the recyclability of the economic cost. In addition, the fatigue damage of them is also one of extremely crucial
factors in the design, manufacture, and operation of aero-engine blades.
There has been an army of damage mechanisms of creep and fatigue and life prediction methods used for turbine blades under
complex stress conditions, such as Ref. [6–10]. The fatigue life prediction methods mainly focus on the equivalent strain method, the
energy method, and the critical plane method. Among them, the equivalent strain method is used to convert the multi-axis into an
equivalent single-axis, and to combine this with uniaxial Manson-Coffin equation to obtain the life prediction equation. Yokobori
[11] first proposed the equivalent strain theory, including the maximum principal strain model, the von Mises equivalent strain
model, and the maximum shear strain model, respectively, which has no clear physical relationship between equivalent strain and

⁎
Corresponding author.
E-mail addresses: [email protected] (D. Sun), [email protected] (l. Xue).

https://doi.org/10.1016/j.engfailanal.2019.104159
Received 20 October 2018; Received in revised form 15 August 2019; Accepted 26 August 2019
Available online 28 August 2019
1350-6307/ © 2019 Elsevier Ltd. All rights reserved.
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

fatigue life, and the interaction of stress-strain and loading factors was ignored. Morrow [12] put forward the energy method, and
Ostergren [13] applied it to uniaxial fatigue life prediction for the first time. Garud [14] and Ellyin [15] further presented a multi-axis
fatigue life prediction model and the plastic strain energy theory based on Garud's research, respectively. And Lee et al. [16] con-
sidered the sum of elastic and plastic strain energy as the fatigue damage parameter. Due to the continuous expansion of the damage
interface inside the material during the cyclic loading process, the total energy consumption and the number of cycles were non-
linear, which could not adequately and comprehensively reflect the multi-axis fatigue failure mechanism and make the energy
method limited in practical engineering applications. Subsequently, Brown and Miller [17,18] proposed critical plane method, as-
suming that crack initiation and propagation occurred on the critical plane,which adopted the maximum shear strain and the normal
strain on maximum shear strain plane as the damage parameters. However, the KBM [19] model used the maximum shear strain
range and the normal strain range as damage parameters and simultaneously verified by experiment. Afterwards, to consider the
effect of non-proportional hardening, Fatemi and Socie [20] presented the FS model based on KBM, in which the maximum shear
strain range and the normal stress on the plane of the maximum shear strain range were used as the damage parameters for the tensile
failure materials. And the SWT [21] model, combined model of Smith, Watson and Topper, took the maximum normal strain range
and the maximum normal stress on maximum normal strain plane as the damage parameters. In this model, the crack propagation
was the cause of fatigue failure, which was perpendicular to the maximum principal strain.
In recent years, a multitude of scholars, such as Shang De-guang [22] and Yu Zheng-yong [23], have further revised and improved
models based on the typical critical plane model and deduced some novel multi-axis fatigue life models. Although a sea of studies
have proposed different forms of life prediction models, a universally applicable fatigue life model has not yet been generated
because of the complexity of fatigue problems. Scholars in a host of countries have established different creep/fatigue life prediction
models for the different failure mechanisms of materials. These theoretical models mainly included traditional parameter relationship
method, ductility exhaustion method, damage mechanics method, fracture mechanics method, multivariate statistical method, and
neural network method and so on. Among them, the ductility exhaustion method was applied to the stress control mode, which
separately calculated creep damage and fatigue damage. The damage mechanics method had considerable difficulty in determining
the damage variable in actual life prediction, which leaded to the complexity of damage evolution equation and directly affected the
accuracy of the life prediction. The fracture mechanics model divided fatigue life into crack initiation life and crack propagation life,
which calculated creep life by introducing the creep rupture parameter C∗ integral. Paradoxically, the multivariate statistical method
and neural network method were new for predicting creep/fatigue life, which were currently limited in practical engineering ap-
plications due to the insufficient material parameter data base. Significantly, the traditional parameter relationship method has been
the most widely used model in aviation field so far, which was divided into linear damage accumulation method and strain control
method. And the former was based on the damage linear superposition method of Miner [24] and Robinson [25], widely utilized in
practical engineering because of the simplicity and practicality.
Unfortunately, although most of the research on turbine blades life has focused on creep-fatigue life and thermomechanical
fatigue life, such as Ref. [26–28], the interaction mechanism between creep and fatigue was indeterminate and ambiguous owing to
the complexity of creep-fatigue interaction. Besides, the method of linear superposition of fatigue damage and creep damage ignored
interaction of creep/fatigue. Moreover, some creep/fatigue life prediction models were generally complex, which needed large
amount of calculation, leading to poorer engineering applicability and the existing material performance data generally cannot be
applicable to the present model satisfactorily. Consequently, the creep/fatigue interaction failure is of great importance to turbine
blades life, which urgently need to be explore the interaction of creep/fatigue to precisely calculate creep/fatigue life.
One of high pressure turbine blades is selected as the research object in this paper. Through theoretical analysis, numerical
simulation, and experimental research, the multiaxial effect on the low-cycle fatigue life of a turbine blade and the creep/fatigue
interaction are fully considered, and the multi-axis low-cycle fatigue based on a new critical plane damage parameter is proposed.
Furthermore, the multi-axis low-cycle fatigue life prediction model and the L-M equation are applied to respectively predict fatigue
life and creep life of the assessment positons selected on blade, and then the creep/fatigue interaction damage mechanism is studies
by different experimental loading schemes, which provides reference and technical support for safety inspections, regular main-
tenance, and optimization improvement during service. Meanwhile, this work has important theoretical and practical engineering
value for the reliability during safe operation period of aero-engine, the overall life of the aircraft, and the recovery of economic costs.

2. Materials and methods

2.1. Theoretical analysis - life prediction model

High-pressure turbine blades serve under the conditions of high-speed, suffering from low-cycle fatigue and high-temperature
creep fatigue and interaction damage of creep/ fatigue. The crack initiation at blade root, caused by the high stress generated during
take-off landing process, frequently leads to blade failure. For reliability and flight safety, aero-engine design manuals clearly define
that fatigue cracks are not allowed on turbine blades which means the low-cycle fatigue life of turbine blades in actual work refers to
the crack initiation life.
The typical working cycle of the aero-engine is divided into one main cycle and four secondary cycles by rain flow counting
method. An analysis of the aero-engine speed spectrum manifests that the amplitude of the main cycle speed fluctuate intensely
during take-off landing process, and the high-level cyclic stress acting on blade causes blades to undergo elastoplastic deformation,
resulting in low-cycle fatigue damage. The variation of speed amplitude is inapparently fluctuant during cruise process. The sec-
ondary cycles are usually negligible because of the slight effect of the aero-engine speed on the fatigue life of blade. However, to

2
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

predict the creep/fatigue life of blade more accurately during actual service, the effect of the above two types of cycles is simulta-
neously considered in this paper.

2.1.1. Low-cycle fatigue life prediction model

In calculating the low-cycle fatigue life of high-pressure turbine blade, the C-M theory is used as the theoretical basis for the life
prediction model of the Mises yield criterion:

t f
= (2Nf )b + (2Nf )c
2 E
f
(1)

2 2 2 2] 12
= [( 1 2) +( 3) +(1 3)
t
3
2
(2)
where εt is the maximum principal strain amplitude, σf′ and b are fatigue strength coefficient and index, respectively, εf′ and c are the
fatigue plasticity coefficient and index, respectively, E is the elastic modulus, and Nf is fatigue life.
The actual turbine blades are serving at high temperature, high pressure, and high speed, plus the effect of their own masses,
which brings about tensile stress, vibration stress, and thermal stress. Unfortunately, Eq. (1) is applicable under the condition of a
symmetrical cycle which leads to considering the effect of mean stress during the actual service period. Finally, the modified Morrow
equation is obtained by modifying mean stress.

t f m
= (2Nf )b + (2Nf )c .
2 E
f
(3)
The modified Morrow equation is applicable to the single axis, but not to the multi-axis non-proportional load of turbine blades.
And Smith, Watson and Topper [21] found that the normal strain range and normal stress on the critical damage plane of turbine
blades were two crucial factors affecting crack initiation, gaining the SWT life prediction model based on the C-M equation for tensile
failure .

n,max ( f )2
= (2Nf ) 2b + (2Nf ) b+ c
n,max
2 E
f f
(4)
where ∆εn, max is the maximum normal strain range and σn, max is the maximum normal stress on the plane of the maximum normal
strain range.
A large number of multi-axis non-proportional loading experiments have shown that the shear and normal strain range on the
maximum critical damage plane lead to multiaxial fatigue cracks. The SWT equation takes no account of shear stress effects, shear
strain range, and the mean stress on the critical plane, which results in an inaccurate blade life prediction. To account for the
comprehensive influence of the shear and normal strain range, the fatigue damage parameter of Shang De-guang is introduced, which
is independent of loading path and considers the maximum shear and normal strain range on the maximum critical damage plane.
The shear strain range ∆γ and the forward strain range εn∗ between the two shear strain points are combined into a new equivalent
strain amplitude, simultaneously considering the effect of the additional reinforcement under the multi-axis non-proportional
loading. The equivalent strain amplitude is finally merged into Eq. (5).
1
2 2
eq 2 1 max
d= = n + .
2 3 2 (5)
Considering the multi-axis non-proportional load condition of the mean stress, the theoretical prediction model of the multi-axis
non-proportional low-cycle fatigue life of turbine blades is obtained, which is combined with the advantages of the modified Morrow
equation and the SWT equation.
1
2 2
2 1 max
2
( f m)
n,max n + = (2Nf )2b + ( f m) f (2Nf ) b+ c .
3 2 E (6)
It is then found that the normal stress, the shear and normal strain range on the maximum failure critical plane affect the multi-
axis non-proportional fatigue life, so the product of σn, max and the equivalent strain amplitude is taken as the new fatigue damage
parameter d.
1
2 2
2 1 max
d= n,max n + .
3 2 (7)
The critical plane is determined by the maximum fatigue damage parameter, and the mean stress effect is considered to deduce
Eq. (8).
( 2
f m)
d= (2Nf ) 2b + ( m) f (2Nf )b + c .
E
f
(8)
The theoretical prediction model for the multi-axis non-proportional low-cycle fatigue life of turbine blades is finally obtained by
further modifying Eqs. (5)–(6) with combining Eqs. (7)–(8).

3
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

1
2 2
2 1 max
2
( f m)
n,max n + = (2Nf )2b + ( f m) f (2Nf ) b+ c .
3 2 E (9)

2.1.2. Creep life prediction model

The blade creep life is determined by temperature, stress level, and holding time. The blade structure is extremely complexed,
which exists a certain distortion between the span direction and the chord direction. The stress-strain appears in a multi-dimensional
direction, which increases the modeling difficulty, and there exists quite a few defects, such as the internal grain boundary holes,
which accurately measure the stress-strain field and temperature fields during blade operation extremely arduous. Therefore, it is
especially important to establish a theoretical prediction model, which can calculate the blade creep life more practically and ac-
curately. The blade creep life prediction is generally performed by the L-M [29] method.
P = T (log tb + C )/100000 (10)

T = (9 × /5 + 32) + 460 (11)

log = a 0 + a1 P + a2 P 2 + a3 P 3 (12)
where C is the material constant, tb is the creep failure time, and T is the absolute temperature. P is the Larson-Miller parameter,
a0,a1,a2 and a3 are material constants, and σ is mean stress.

2.1.3. Creep/fatigue life prediction model

The low-cycle fatigue life and creep life is of importance to the high pressure turbine blades, which adopts linear superposition of
the time-independent low-cycle fatigue damage and time-dependent creep damage.
ni ti
Df + Dc = + = D.
Nf tr (13)
If D ≠ 1, there exists interaction, otherwise there is no interaction. In order to reflect the creep/fatigue interaction and reveal the
mechanism of component failure, Lagneberg et al. [30] proposed a modified linear damage accumulation model, which introduces an
interaction coefficient B to macroscopically express the creep/fatigue interaction strength.
1/2
ni ni ti ti
+B × + = 1.
Nf Nf tr tr (14)

2.2. Numerical simulation-blade thermal-structural coupling simulation

2.2.1. Blade geometry model and finite element model

The turbine blade generally consists of three parts: the blade body, the edge plate, and the tenon, as shown in Fig. 1(a), which can
achieve the conversion from gas internal energy to mechanical energy of aero-engine. The blade body is to achieve energy conversion
and improve the direction of the airflow, the edge plate is to reduce the stress concentration and prevent the high-temperature gas
from entering the temperature-resistant components such as the turbine disk, and the tenon is to connect blade and turbine disk to
reduce the contact stress and increase the heat dissipation.

Fig. 1. Geometric model and finite element model of a turbine blade.

4
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

In this study, a high-order three-dimensional 10-node tetrahedral solid element was used for the mesh. The number of elements
was 303,697 and the number of nodes was 461,788. The final finite element model is shown in Fig. 1(b).

2.2.2. Calculation and analysis of the thermal-structural coupling of blade

2.2.2.1. Control equations of fluid and heat transfer in turbine blades. Assuming the incompressibility of the gas to the turbine blade of
aero-engine, the steady flow of blade in the actual operation follows the conservation laws of mass, momentum and energy,
respectively [31].

(1) The gas in blade fluid domain is incompressible and satisfies the fluid continuity equation.

+ ( U) = 0
t (15)

(2) The component of gas in blade fluid domain in each direction of velocity satisfies the momentum conservation equation.

( U)
+ ( U × U) = ( + µ ( U + ( U )T )) + SM = 0
t (16)

(3) Regardless of the viscous dissipation of gas in blade fluid domain, the energy conservation equation is satisfied.
( h) T 2
+ ( Uh ) = ( T) + µ U + U U U + SE
t t 3 (17)

where SM is momentum source term, SE is energy source term, h is specific enthalpy.

1 2
h =h+ U , h = h ( , T)
2 (18)

Eq. (19)–(20) are respectively state equation and the specific heat capacity equation.
= ( , T) (19)

C = C ( , T) (20)

For the ideal gas:

w( + ref )
C = C ( , T ), =
RT (21)
where w is the gas molecular weight, ρref is the reference pressure, and R is the general gas constant.

2.2.2.2. Turbulence model in turbine blade fluid domain. The gas in outer flow field of turbine blade is complex typical turbulence. The
k − ε turbulence model proposed by Launder and Spalding is a simple and standard turbulence model, and can effectively solve the
problem of high Reynolds number fluid turbulence, which was widely used to describe the turbulent state of gas in industrial frontier
because of its high accuracy and stability, but which is not excellent to predict complex flow with strong pressure gradient. An ocean
of scholars have improved the k − ε model by introducing an additional term Rε to obtain the RNG [32], and the Realizable k − ε
model [33] by modifying the swirl. The SST k − ε model [34] by adding a mixed function of transition is applied to solve the flow
characteristics of turbine blade in this paper, which has favorable simulation results for complex flow such as shock wave and
counter-pressure gradient flow.

2.2.2.3. Coupled heat transfer calculation method. Conjugate heat transfer between solid and fluid domain of the turbine blade is
conducted by the fluid-solid coupling mode after obtaining blade temperature field by the flow field analysis. The energy equation is
used to solve blade temperature resulting from coupling heat transfer when calculating the heat problem. The convection and
diffusion term are not considered owning to existing heat conducting of the solid part of blade, and Eq. (22) is the energy
conservation equation [31].
( c T) T T T
= + + + SE
t x x y y z z (22)

where ρ, cρ, λ, SE are the solid density, the specific capacity, the thermal conductivity and the internal energy source term in solids,
respectively.

2.2.2.4. Material properties of turbine blades. Directionally solidified high temperature nickel base alloy DZ125 is chosen as the
material of turbine blades. It is a nickel base precipitation hardening directionally solidified columnar crystal super-alloy. The
thermal conductivity λ, elastic modulus E, Poisson's ratio μ, linear expansion coefficient α, yield strength σ0.2 and tensile limit σb at
different temperatures can be obtained by consulting China Aviation Materials Manual [35], as shown in Table 1.

5
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

Table 1
Material prosperities of DZ125 alloy at various temperatures.
ρ = 8.48 × 103kg/m3

T/ ° C 200 300 400 500 600 700 800 900

λ/(m ∙ ° C) 10.7 13.1 15.6 18.0 20.3 22.5 24.6 26.4

T/ ° C 20 250 500 600 700 800 900 /
E/GPa 176.5 168 161.5 151.5 145.5 139.5 135 /
μ 0.41 0.405 0.41 0.415 0.43 0.43 0.435 /
T/ ° C 20–300 20–400 20–500 20–600 20–700 20–800 20–900 /
α/10−6 ∗ ° C−1 12.45 12.86 13.26 13.53 14.04 14.55 15.06 /
T/ ° C 20 700 760 850 900 950 / /
σb/MPa 1320 1220 1210 1030 850 740 / /
σ0.2/MPa 985 930 955 905 580 455 / /

2.2.2.5. Analysis of finite element calculation results. For the coupling of the centrifugal load, the aerodynamic load, and the
temperature load, the ANSYS finite element software program was used to analyze the thermal-structural coupling of turbine blade.
The field distributions of temperature and stress-strain of blade of the main cycle were respectively obtained under typical flight
conditions which are shown in Fig. 2(a) and Fig. 2(b).
Fig. 2(a) showed that the highest temperature point was located about three-quarters down the length of blade body, and the
overall temperature change trend gradually rose from the middle to both sides of the blade, which reached a maximum at the leading
edge of blade exhaust port and the bottom edge of the air intake at the bottom. Furthermore, the blade temperature difference was
about 300 °C.
Fig. 2(b) showed that the maximum equivalent stress was 840.23 MPa, located at the root of the suction plane, existing serious
stress concentration, which was the reason that the connection form of tenon and tenon groove made blade become cantilever beam.
According to the DZ125 alloy material properties, the maximum equivalent stress was lower than the yield strength, which meant
that blade did not plastically deform and that it satisfied the design. And the maximum equivalent strain was about 8.536 × 10−3,
located at the trailing edge of 1/2 blade body exhaust port.
Therefore, the blade root and the trailing edge of 1/2 blade body exhaust port are high-risk areas for low-cycle fatigue and creep
failure, which are the key assessment positions for blade strength check and life prediction. Simultaneously, the field distributions of
temperature and stress-strain of the main cycle are basically the same as that of the secondary cycle, including the unchanged high
stress-strain zone and the high-risk areas, and the stress-strain field distribution of the other areas is reasonable.

Fig. 2. Blade simulation cloud map.

6
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

Fig. 3. Blade assessment positions.

2.2.3. Blade assessment position determination

Under the action of complex loads, the turbine blade life is not only related to the stress-strain of the assessment position, but also
to the temperature of the same positon. According to blade field distribution results of temperature and stress-strain of the main cycle,
combining with the external field failure statistics of turbine blades, A and B were respectively selected as the assessment positions to
study the fatigue damage mechanism, as shown in Fig. 3.
For assessment position A, the maximum stress point, was located at the root of the suction plane. And for assessment position B,
the high temperature, the small gradient, and the strain field were comprehensively considered, so the trailing edge of 1/2 blade
exhaust was finally chosen as the assessment position.
According to the analysis process in the main cycle, the stress-strain and temperature of A and B in the secondary cycle were
sequentially obtained, as shown in Table 2.

2.3. The new critical plane damage parameter solution for the multi-axis fatigue analysis of turbine blade

2.3.1. Model solving steps

(1) The stress-strain tensors. Based on the finite element analysis of the thermo-structural coupling simulation, the assessment po-
sition was determined, and the stress- strain tensor were obtained by Eq. (23).

x xy xz x xy xz
= yx y yz = yx y yz
zx zy z zx zy z (23)

(2) Coordinate transformation

The stress in any plane in Fig. 4 and the cosines of x, y, and z axes are as shown in Eq. (24). Matrix (25) was further used to obtain
each stress component of the node.

a11 = cos sin a12 = sin sin a13 = cos

a21 = sin a22 = cos a23 = 0
a31 = cos cos a32 = sin cos a33 = sin (24)

Table 2
Stress-strain, and temperature of assessment positions for different working conditions.
Parameters Assessment positions 30 % n 57 % n 75 % n 82%n 89 % n 96 % n

Stress value/MPa A 420 508 613 671 769 840

B 395 458 591 641 719 772
Strain value% A 0.262 0.312 0.373 0.414 0.505 0.556
B 0.389 0.455 0.593 0.653 0.777 0.853
Temperature values/°C A 440 470 500 540 700 760
B 535 560 590 630 790 850

7
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

Fig. 4. Stress-strain on any plane.

, 2 2 2
x a11 a12 a13 2a11 a12 2a11 a13 2a13 a12
,
y 2 2 2
, a21 a22 a23 2a21 a22 2a21 a23 2a23 a22
z 2 2 2
, = a31 a32 a33 2a31 a32 2a31 a33 2a33 a32
xy
, a11 a21 a12 a22 a13 a23 a11 a22 + a12 a21 a13 a21 + a11 a23 a12 a23 + a13 a22
yz
, a11 a31 a12 a32 a13 a33 a11 a32 + a21 a31 a13 a31 + a11 a33 a13 a32 + a12 a33
xz
a21 a31 a22 a32 a23 a33 a21 a32 + a22 a31 a23 a31 + a21 a33 a22 a33 + a23 a32 (25)

Furthermore, the stress variable in Matrix (25) was replaced by a strain variable to calculate the strain component on an arbitrary
plane of the node.

(3) The plane determined by θ = 0 and φ = 0 at the node of the assessment position was taken as the initial critical plane, the stress
and strain were obtained on the initial critical plane under the typical flight condition, to go a step further to gain the fatigue
damage parameter.
(4) The critical plane was the plane bearing the maximum damage. The double-cycle search of the fatigue damage parameter d was
performed in steps of 2° with θ andφ. The θ represented the angle between the normal strain and x-axis, and φ represented the
angle between the normal strain and z-axis. The maximum damage critical plane determined by (θ, φ) followed with dmax by
MATLAB, so the multi-axis low-cycle fatigue life was finally gained by acquiring other parameters on the critical plane.

2.3.2. Critical plane solution and model correction

From the thermal-structural coupling finite element simulation analysis and blade external field failure statistics, the position
coordinates of A and B were (−35.2, −17.7, 1.7) and B (−10, 2.136, 131.5), respectively. The main cycle was used as an example for
the search calculation, and the stress and strain of assessment positions A and B were exhibited in Table 3.
The 3D maps of different d of A and B were received by different equations in the form of double-cycle search, as shown in Fig. 5.
And the first two were got by Eq. (5), but the latter two were gained by Eq. (7), respectively. Besides, Table 4 showed the results of
searching.

Table 3
Stress and strain components of the assessment positions A and B.
Assessment position Stress component (MPa) Numerical value Strain component (%) Numerical value

A σx −235.24 εx −0.306
σy 302.09 εy −0.184
σz 371.15 εz 0.2906
τxy −118.23 γxy −0.2148
τyz −319.39 γyz −0.3497
τxz 77.001 γxz 0.1387
B σx 2.8861 εx −0.168
σy 11.476 εy −0.326
σz 776.96 εz 0.776
τxy 5.0289 γxy −0.004
τyz −6.8697 γyz −0.0682
τxz −33.432 γxz 0.0572

8
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

Fig. 5. 3D maps of d on different planes of A and B.

Table 4
Search results of different critical planes of A and B.
Assessment position θ(°) φ(°) dmax σn, max(MPa)

A 26 74 0.5302% 532.82
B 144 6 0.3918% 736.34
A 142 32 290.9 /
B 160 2 299.6043 /

2.4. Experimental study – experimental validation

2.4.1. Specimen material

DZ125 is one of the highest levels of directional solidification alloys in China and applied to turbine blades manufacturing due to
the outstanding performance of high-temperature, thermal fatigue and casting. The chemical composition, and the constants of creep
and strain fatigue of DZ125 are shown in Table 5 and Table 6, respectively.

2.4.2. Experiment sample determination

In order to simulate the actual working state of blade in a real environment, the direct and effective method of studying the
damage mechanism of creep/fatigue was used to sample different areas in blade [36], which needed to design a suitable experimental
fixture and blade feature simulation specimens. Generally, the design and manufacture of the specimens had got to abide by the
principles listed below:

Table 5
Chemical composition of DZ125 (%).
C Cr Ni Co W Mo Al Ta Hf Ti

0.07~0.12 8.4~9.4 Bal. 9.5~10.5 6.5~7.5 1.5~2.5 4.8~5.4 3.5~4.1 1.2~1.8 0.8

9
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

Table 6
Constants of creep and strain fatigue.
Creep constants C a0 a1 a2 a3

14.147 4.225449 −8.192647 31.45216 −63.83470

Strain fatigue constants σf′/MPa εf ′ b c E/GPa

1855 3.2% −0.105 −0.587 96

(1) The geometry of blade simulation specimens was similar to the corresponding part of the blade.
(2) Under the experiment conditions, the stress-strain and temperature field distributions of the experiment position of the feature
simulation specimen were the same as the corresponding position for a real working condition.

According to the above principles, the designed blade feature simulation specimens are shown in Fig. 6. Additionally, Fig. 7 shows
the dimensions and the actual blade machining specimens.

2.4.2.1. Edge plate feature simulation specimen. The characteristics of the geometrical mutation near blade edge plate were considered
when designing the edge plate feature simulation specimen. The assessment area of the simulation specimen was the geometrically
abrupt region near the convex platform. Parameters such as the mutation ratio, the height of the mutation area, and the rounding of
the transition part were similar to that of the corresponding areas of blade, which could be reflective of the geometric features of the
transitional connection between the edge plate and blade root.

2.4.2.2. Blade body feature simulation specimen. The cross-section change of blade body specimen was small. For convenience of
contrast, the overall dimension of blade body simulation specimen was similar to that of the edge plate feature simulation specimen,
with extensometer supporting lugs but without the convex platform. Moreover, the specimen is with notches to acquire the same
stress field distribution of the edge plate.

2.4.3. Experiment equipment and loading

The SY72-1 thermal machine coupling fatigue testing machine consists of four parts: a mechanical loading system, a temperature
control system, an electro-hydraulic servo system, and a cooling system, as shown in Fig. 8, which could achieve a high-temperature
environment simulation and be mechanically loaded for the actual service condition of specimens.
The actual load conditions of blade was effectively and significantly realized by using this testing machine to apply cyclic loads,
which was shown in Fig. 9.

2.4.4. Determination of the assessment positions of the feature simulation

Through the finite element simulation analysis, it was found that for the edge plate, the σmax = 842.77MPa when loading 14kN
and for blade body, theσmax = 775.87MPa when loading 10kN, which meant the stress field distribution of blade feature simulation
was extremely homologous to that of the actual blade assessment positions, as shown in Fig. 10. Through concrete analysis above, the
stress value and the stress field distribution achieved a real simulation of blade assessment positions, which simultaneously de-
monstrated the design rationality of blade feature simulation specimen.

Fig. 6. Blade feature simulation specimen.

10
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

Fig. 7. Dimensions and machining specimens of blade.

2.4.5. Low-cycle fatigue experiment at room temperature and high-temperature creep/fatigue experiment
2.4.5.1. Low-cycle fatigue experiment at room temperature. In this study, a rectangular wave was used for experiment loading. The
load-holding time, load cycle time, frequency, stress ratio and experiment load sampling rate were 0.8 s, 1 s, 1 Hz, 0, and 100 points
per second, respectively. The specific loading form was shown in Table 7.

2.4.5.2. High-temperature low-cycle creep/fatigue experiment. The axial loading mode controlled by force was adopted, and the
different spectrums were applied according to the loads determined by the finite element simulation. And then the stress-strain field
distribution of blade feature simulation specimen were of similarity to the reality when loading 12kN used in the load spectrum 1 and
2, 8.5kN used in the load spectrum 3 and 4, respectively. Likewise, the degree of creep/fatigue interaction was further studied by
changing the duty cycle, as shown in Table 8.

3. Results

3.1. Low-cycle fatigue experiment results at room temperature

Table 9 showed the room temperature fatigue experiment results by experiment.

It can be seen from Table 9 that when the loading mode was the same, the smaller the stress was, the smaller the low-cycle fatigue
life was, which explained the root of the suction plane was the extremely hazardous area under the low-cycle fatigue condition.
Table 10 demonstrated that Morrow's modified fatigue life was the largest, followed by Eq. (6) and Eq. (9), while the calculated
results of modified SWT equation were the most conservative, and the predicted results of Eq. (9) were closer to the experimental
results. Simultaneously, it can be seen from Fig. 11 that the accuracy of all above life prediction models was all within 2 times
dispersion zone. In addition, the prediction errors of modified Morrow equation, Eq. (6), and SWT equation were about 18%, 13%,
and 13%, respectively, but Eq. (9) was about 3.5%, which is located in the 1.1 times dispersion zone.
It can be seen from the above calculation results that under the working load cycle spectrum, the main cycle had the greatest
influence on the fatigue damage of turbine blades, as shown in Table 11. Compared with assessment position B, assessment position A
had a higher stress level and a lower multi-axis low-cycle fatigue life, which meant that the low-cycle fatigue life was determined by

11
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

Fig. 8. SY72-1 thermal machine coupling fatigue experiment system.

Fig. 9. Feature loading.

assessment A, and the life was 7396 times. Therefore, the calculated multi-axis low-cycle fatigue life of turbine blade was about
12,316 h by combining the life of assessment position A and a take-off and landing time of 100 min for an aircraft. Concurrently, it
can be seen that the secondary cycle during flight had less fatigue damage.

12
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

Fig. 10. Stress distributions of blade simulation specimen and blade assessment position.

3.2. High-temperature creep/fatigue experiment results

Table 12 showed the high-temperature creep/fatigue experimental results.

From Table 12, it can be known that: (1) At the same assessment position of blade feature simulation specimen, the longer the
load holding time in one cycle was, the smaller the life cycle number was; (2) At different assessment positions of blade, the higher

13
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

Table 7
Load form.
Feature simulation specimen Load size/kN Cycle/s Duty cycle Experiment temperature/°C

Edge plate 14 3 80% 25

Blade body 10

Table 8
Load form.
Load spectrum Load size/kN Cycle/s Duty cycle Load-holding time/s Experiment temperature/°C

1 12 3 80% 2.4 760

2 25% 0.75
3 8.5 80% 2.4 850
4 20% 0.6

Table 9
Results of the room temperature fatigue experiment.
Feature simulation Failure time /min Experiment life/cycles

Edge plate 125.7 7542

Blade body 139.3 8358

Table 10
Low-cycle fatigue life of the assessment position.
Assessment positions Morrow equation life/cycles SWT equation life / cycles Eq. (6) Life/cycles Eq. (9) Life/cycles Experiment life/cycles

A 8932 6554 8538 7471 7542

B 9828 6995 9589 8070 8358

Fig. 11. Comparison of low-cycle fatigue life prediction values and experimental values for four models.

Table 11
Low-cycle fatigue life and single-cycle fatigue damage for blade assessment position.
Range of rotation rate Frequency Assessment Assessment Assessment position A single-cycle Assessment position B single-cycle
position A cycle position B cycle total damage/low-cycle fatigue life total damage/low-cycle fatigue life
life life

0 − 96 % n − 0 1 7471 8070 1.352 × 10−4/7396 1.25 × 10−4/8000

30 % n − 57 % n − 30 % n 2 4.846 × 108 5.728 × 108
75 % n − 82 % n − 75 % n 1 2.916 × 108 3.891 × 108
82 % n − 89 % n − 82 % n 1 1.142 × 108 1.539 × 108
82 % n − 96 % n − 82 % n 1 8.162 × 108 9.963 × 108

14
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

Table 12
Experiment results of the feature simulation specimens for the load spectrum.
Simulation specimens Load spectrum Experiment temperature/°C Failure time/min Cycles

Edge plate 1 760 53.1 1062

2 97.8 1956
Blade body 3 850 46.5 930
4 109.3 2186

the temperature was, the shorter the life was, when the loading form was the same;(3) The load-holding time and the temperature
were two paramount factors affecting the creep/fatigue life, which conclude that when the load-holding time and the temperature
increased, the life obviously decreased.
During the take-off and landing of the aircraft, low-cycle fatigue and high-temperature creep damage occurred simultaneously in
the high-pressure turbine blades. Additionally, the ratios of creep damage and low-cycle fatigue damage at assessment positions A
and B were 2 and 4.72, respectively, and the total damage ratios of creep damage were 66.8% and 82.5%, respectively, which
revealed that creep damage was predominant, and the higher the temperature was, the greater the proportion of creep damage was.
As a matter of fact the service life of turbine blades was determined by the minimum life in each assessment position, which
explicated the creep/fatigue life of assessment position B was the working life of high pressure turbine blades based on the linear
accumulation theory of damage, and the creep/fatigue life was about 2331 h, as shown in Table 13, which was consistent with the
results of the failure statistics of the external field failure of turbine blades.

3.3. Calculation results and analysis considering creep/fatigue interaction

Based on the above comprehensive analysis, the interaction degree between creep and fatigue was further studied by Eq. (14), and
the interaction coefficient was calculated. Finally, the life prediction models considering the interaction between creep and fatigue
were gained. Based on the experimental data obtained in Table 12, the B values at 760 °C and 850 °C were 2.8 and 1.32 respectively,
and then Eqs. (26)–(27) of the non-linear life prediction models at different temperatures were acquired, as shown in Fig. 12.
1
2
ni ni ti ti
D= + 2.8 × + =1
Nf Nf tr tr (26)
1
2
ni ni ti ti
D= + 1.32 × + =1
Nf Nf tr tr (27)

It can be seen from Table 15 that the creep/fatigue life of high-pressure turbine blades was determined by assessment position B
and the service life was about 1552 h, which had a specific reference value for turbine blades maintenance strategy. In addition, the
higher the temperature was, the more serious creep damage was, and the less interaction damage was in the creep/fatigue for blade,
which lead to the nonlinear damage equation was closer to the damage linear accumulation model. In conclusion, the linear damage
accumulation model had a better life prediction effect when pure fatigue or pure creep damage occurred, but when fatigue and creep
damage occurred simultaneously, the creep/fatigue interaction damage dominated, and the interaction life prediction model could
more appropriately describe the creep/fatigue interaction.
Table 16 showed that when the interaction occurred, the creep/fatigue life of turbine blade was significantly shortened, and the
interaction life prediction result was reduced by about 33.4% compared to that of the damage linear accumulation method. Briefly,
the creep/fatigue interaction had a greater impact on the life of turbine blade.

4. Discussion

In this study, Eq. (6) and the modified Eq. (9) are obtained by introducing the Shang De-guang fatigue damage parameter, which
are applied to calculate the multi-axis low-cycle fatigue life of turbine blade. Moreover, the calculated results are compared with the
experimental results by the modified Morrow equation and SWT equation, which supplement the lack of experiments in Ref. [37].
Table 10 demonstrates that when the external load is multi-axis non-proportional, there is a phase difference between the positive
strain and the shear strain by analyzing different the results calculated by equations above, which means the magnitude and direction
of normal and shear strain changed in the same cycle. The von Mises equivalent strain only considers the principal strain of the

Table 13
Creep/fatigue life of blade assessment positions.
Assessment positions Single-cycle fatigue damage Single-cycle creep damage Total damage Actual life /h

−4 −4 −4
A 1.352 × 10 2.71 × 10 4.062 × 10 4103
B 1.25 × 10−4 5.9 × 10−4 7.15 × 10−4 2331

15
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

Table 14
Creep/fatigue damage of blade simulation specimens.
Feature simulation Load spectrum Experiment temperature /°C Fatigue damage Creep damage

Edge plate 1 760 0.143 0.2877

2 0.264 0.1656
Blade body 3 850 0.116 0.5480
4 0.273 0.3220

Fig. 12. Creep/fatigue interaction model and experimental results.

material in three directions, ignoring the shear strain on the critical plane of the maximum failure. The relationship between the
stress and the strain under a multi-axis non-proportion becomes more complicated, and the constitutive relationship is laborious to
determine which makes it unfeasible to simply convert the multi-directional strain into an equivalent strain using the equivalent
principle. The prediction result of the modified Morrow equation is higher without any consideration of the shear strain on the
critical plane, while the SWT equation ignores the effect of the shear strain on the critical plane of the material for fatigue crack
initiation and propagation and the interaction of the stress and the strain under non-proportional loading, only considering the
maximum normal strain and the maximum normal stress, which causes non-proportional loading generated by the rotation of the
spindle not only to make stress and strain analysis difficult, but also to produce non-proportional additional strengthening, and
further makes results smaller. Eq. (6) determines the maximum damage plane using the strain, and the equation is then used to
calculate the maximum normal stress on the maximum damage plane. For Eq. (9), the maximum damage plane is obtained using the
equivalent strain and the normal stress, and in the meantime, Eq. (9) considers the normal stress and shear strain on the critical plane
and the additional strengthening effect for multi-axial non-proportionality. Subsequently, Eq. (9) absorbs the advantages of the
modified Morrow equation and the SWT equation, which brings more accurate prediction results. Additionally, the critical plane

Table 15
Fatigue, creep, and interaction damage for blade assessment positions.
Assessment Temperature /°C B Single-cycle fatigue Single-cycle creep Creep fatigue Single-cycle total Actual life
position damage damage Interactive damage damage /h

A 760 2.8 1.352 × 10−4 2.71 × 10−4 5.36 × 10−4 9.422 × 10−4 1768
B 850 1.32 1.25 × 10−4 5.9 × 10−4 3.585 × 10−4 1.0735 × 10−4 1552

Table 16
Comparison of creep/fatigue life results.
Creep/fatigue life based on a linear accumulation of damage /h Consider creep/fatigue life of interaction /h

2331 1552

16
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

method considers the magnitude and direction of the stress and strain, which is more consistent with the damage during the actual
operation of high-pressure turbine blades.
Based on the high-temperature creep/fatigue experiment, the coupling effect of the creep and fatigue of turbine blade during take-
off and landing is much accounted, which makes up for the lack of the creep damage and creep/fatigue interaction in Ref. [23] and
Ref. [38,39,40,41]. The interaction mechanism of the creep/fatigue is studied in this paper, the interaction coefficient at the cor-
responding temperature is obtained, and the actual creep/fatigue life of the assessment position is calculated, as shown in Table 14.
Compared with the creep/fatigue life model based on a linear accumulation of damage, it can conclude that when only pure fatigue or
pure creep occurred, the damage-based linear cumulative model prediction is better but when creep and fatigue occurred simulta-
neously, the interaction model predicts better. The reasons for the above phenomena are that the creep deformation causes the
fatigue crack initiation rate and the number of creep holes to increase when creep and fatigue occur simultaneously. In simpler terms,
the creep holes act as sources of destruction to promote the initiation and propagation of fatigue cracks, and the fatigue cycle promote
the nucleation and growth of creep holes and aggravate the fatigue fracture process, which explain the importance and non-neg-
ligibility of interaction.

5. Conclusion

In this paper, a multi-axis low-cycle fatigue life prediction model based on a new critical plane damage parameter was proposed.
And the effects of main cycle and secondary cycle on the low-cycle fatigue life of turbine blade under typical flight conditions were
considered. Simultaneously, the stress-strain distribution law under complex load coupling was obtained based on the thermal-
structural coupling analysis. Additionally, the creep/fatigue life of blade assessment positions was calculated by the novel proposed
model, and different experimental loading schemes were used to further study the creep/fatigue interaction damage mechanism. The
specific conclusions can be summarized as follows:

(1) The finite element ANSYS software program was used to calculate the high stress-strain area of high-pressure turbine blade under
a typical flight condition, which revealed that the high stress-strain area of the main cycle and the secondary cycle was located at
the root of blade under the complex load of the temperature, and the centrifugal and aerodynamic forces. The results showed that
the blade's maximum equivalent stress was 840.23 MPa and the maximum equivalent elastic strain was 8.536 × 10−3, which met
the safety design requirements of the aero-engine and explained the rationality of blade modeling.
(2) The modified Morrow's fatigue life was the largest, followed by Eq. (6), while the modified SWT's fatigue life was the most
conservative, and Eq. (9)‘s calculated fatigue life was closer to the experimental results under the condition of considering the
centrifugal load, the temperature load, and the aerodynamic load of high-pressure turbine blades, simultaneously.
(3) Based on the thermal-structural coupling analysis of high-pressure turbine blade, a multi-axis low-cycle fatigue life prediction
model established was applied to calculate the fatigue life of the assessment positions A and B. The creep/ fatigue life of the two
assessment positions were obtained by adopting the linear cumulative damage principle, which was about 2331 h, and by the
interaction model, which was about 1552 h. Besides, the results showed that the life at the root of blade was the smallest,
indicating this weak area should be the key component of the further inspection, maintenance, and design optimization of turbine
blade.
(4) By calculating and comparing the creep/fatigue damage at different assessment positions for high-pressure turbine blade during
take-off and landing of the aircraft, it was found that the damage form based on the damage accumulation was mainly creep
damage. Meanwhile, compared with creep/fatigue life based on linear cumulative damage, that calculated by interaction life
model reduced by 33.4%, which indicated that the interaction was significant and non-negligible.

Funding

This work was supported by National Natural Science Foundation of China (Grant No. 51875076), National Key Research and
Development Project (2018YFB1306701), NSFC-Liaoning United Key fund (Grant No. U1708255).

References

[1] H.F. Gao, C.W. Fei, G.C. Bai, L. Ding, Reliability-based low-cycle fatigue damage analysis for turbine blade with thermo-structural interaction, Aerosp. Sci.
Technol. 49 (2016) 289–300.
[2] A.M. Kolagar, N. Tabrizi, M. Cheraghzadeh, M.S. Shahriari, Failure analysis of gas turbine first stage blade made of nickel-based superalloy, Case Stud. Eng. Fail.
Anal. (Netherlands) 8 (2017) 61–68.
[3] R.K. Mishra, V. Nandi, R. Raghavendra Bhat, Failure analysis of high-pressure compressor blade in an aero gas turbine engine, J. Fail. Anal. Prev. (Germany) 18
(3) (2018) 465–470.
[4] R.K. Mishra, J. Thomas, K. Srinivasan, V. Nandi, R.R. Bhatt, Failure analysis of an un-cooled turbine blade in an aero gas turbine engine, Eng. Fail. Anal. 79
(2017) 836–844.
[5] A. Mokaberi, R. Derakhshandeh-Haghighi, Y. Abbaszadeh, Fatigue fracture analysis of gas turbine compressor blades, Eng. Fail. Anal. 58 (2015) 1–7.
[6] O. Mallet, H. Kaguchi, B. Ilschner, F. Meyerolbersleben, K. Nikbin, F. Rezaiaria, G.A. Webster, Influence of thermal-boundary conditions on stress-strain dis-
tribution generated in blade-shaped samples, Int. J. Fatigue 17 (2) (1995) 129–134.
[7] T. Brendel, E. Affeldt, J. Hammer, C. Rummel, Temperature gradients in TMF specimens. Measurement and influence on TMF life, Int. J. Fatigue 30 (2) (2008)
234–240.
[8] N.G. Bychkov, V.P. Lukash, Y.A. Nozhnitsky, A.V. Perchin, A.D. Rekin, Investigations of thermomechanical fatigue for optimization of design and production
process solutions for gas-turbine engine parts, Int. J. Fatigue 30 (2) (2008) 305–312.

17
J. Huo, et al. Engineering Failure Analysis 106 (2019) 104159

[9] R.Q. Wang, F.L. Jing, D.Y. Hu, In-phase thermal-mechanical fatigue investigation on hollow single crystal turbine blades, Chin. J. Aeronaut. 26 (6) (2013)
1409–1414.
[10] R.Q. Wang, K.H. Jiang, F.L. Jing, D.Y. Hu, Thermomechanical fatigue failure investigation on a single crystal nickel superalloy turbine blade, Eng. Fail. Anal. 66
(2016) 284–295.
[11] T. Yokobori, H. Yamanouchi, S. Yamamoto, Low cycle fatigue of thin-walled hollow cylindrical specimens of mild steel in uni-axial and torsional tests at constant
strain amplitude, Int. J. Fract. Mech. 1 (1) (1965) 3–13.
[12] J. Morrow, Cyclic plastic strain energy and fatigue of metals, Internal Friction, Damping, and Cyclic Plasticity, ASTM International, 1965.
[13] D.M. Ji, M.H.H. Shen, D.X. Wang, J.X. Ren, Creep-fatigue life prediction and reliability analysis of P91 steel based on applied mechanical work density, J. Mater.
Eng. Perform. 24 (1) (2015) 194–201.
[14] S. Suman, A. Kallmeyer, J. Smith, Development of a multiaxial fatigue damage parameter and life prediction methodology for non-proportional loading, Frattera
ed Integrita 10 (38) (2016) 224–230.
[15] F. Ellyin, B. Valaire, High-strain multiaxial fatigue, J. Eng. Mater. Technol.-Trans. ASME 104 (3) (1982) 165–173.
[16] Z.R. Wu, X.T. Hu, Y.D. Song, Multiaxial fatigue life prediction for titanium alloy TC4 under proportional and nonproportional loading, Int. J. Fatigue 59 (3)
(2014) 170–175.
[17] D. Kulawinski, M. Hoffmann, T. Lippmann, G. Lamprecht, A. Weidner, S. Henkel, H. Biermann, Isothermal and thermo-mechanical fatigue behavior of 16Mo3
steel coated with high-velocity oxy-fuel sprayed nickel-base alloy under uniaxial as well as biaxial-planar loading, J. Mater. Res. 32 (23) (2017) 4411–4423.
[18] J.F. Mei, P.S. Dong, A new path-dependent fatigue damage model for non-proportional multi-axial loading, Int. J. Fatigue 90 (2016) 210–221.
[19] N.R. Gates, A. Fatemi, On the consideration of normal and shear stress interaction in multiaxial fatigue damage analysis, Int. J. Fatigue 100 (2017) 322–336.
[20] A. Ince, G. Glinka, Innovative computational modeling of multiaxial fatigue analysis for notched components, Int. J. Fatigue (2016) 82.
[21] F. Öztürk, J.A.F.O. Correia, C. Rebelo, A.M.P.D. Jesus, L.S. Silva, Fatigue assessment of steel half-pipes bolted connections using local approaches, Procedia
Struct. Integrity 1 (2016) 118–125.
[22] X.F. Zhao, D.G. Shang, Y.J. Sun, M.L. Song, X.W. Wang, Performance, multiaxial fatigue life prediction based on short crack propagation model with equivalent
strain parameter, J. Mater. Eng. Perform. 27 (1) (2018) 324–332.
[23] Z.Y. Yu, S.P. Zhu, Q. Liu, Y.H. Liu, A new energy-critical plane damage parameter for multiaxial fatigue life prediction of turbine blades, Materials 10 (5)
(2017) 18.
[24] Y. Wang, L. Susmel, The Modified Manson–Coffin Curve Method to estimate fatigue lifetime under complex constant and variable amplitude multiaxial fatigue
loading, Int. J. Fatigue 83 (2) (2016) 135–149.
[25] R.Z. Wang, J. Wang, J.G. Gong, X.C. Zhang, S.T. Tu, P.V.T. Zhang, Creep-fatigue behaviors and life assessments in two nickel-based superalloys, J. Press. Vessel.
Technol. 140 (3) (2018) 031405.
[26] A. Koster, A. Alam, L. Rémy, A physical-base model for life prediction of single crystal turbine blades under creep-fatigue loading and thermal transient
conditions, European Structural Integrity Society (2002) 203–212 Elsevier.
[27] L.J. Chen, Y.H. Liu, L.Y. Xie, Power-exponent function model for low-cycle fatigue life prediction and its applications - part II: life prediction of turbine blades
under creep-fatigue interaction, Int. J. Fatigue 29 (1) (2007) 10–19.
[28] R.Q. Wang, B. Zhang, D.Y. Hu, K.H. Jian, J.X. Mao, F.L. Jing, A critical-plane-based thermomechanical fatigue lifetime prediction model and its application in
nickel-based single-crystal turbine blades, Mater. High Temp. 36 (4) (2019) 325–334.
[29] A. Loghman, M.J. Moradi, Creep damage and life assessment of thick-walled spherical reactor using Larson–Miller parameter, Int. J. Press. Vessel. Pip. 151
(2017).
[30] R. Lagneborg, R.J.M.T. Attermo, The effect of combined low-cycle fatigue and creep on the life of austenitic stainless steels, Mettal. Trans. 2 (7) (1971)
1821–1827.
[31] E. Kianpour, N.A.C. Sidik, M. Bozorg, Thermodynamic analysis of flow field at the end of combustor simulator, Int. J. Heat Mass Transf. 61 (2013) 389–396.
[32] V. Yakhot, S.A. Orszag, Renormalization-group analysis of turbulence, Phys. Rev. Lett. 57 (14) (1986) 1722–1724.
[33] T.H. Shih, W.W. Liou, A. Shabbir, Z.G. Yang, J. Zhu, A new kappa-epsilon eddy viscosity model for high reynolds-number turbulent flows, Comput. Fluids 24 (3)
(1995) 227–238.
[34] F.R. Menter, 2-equation eddy-viscosity turbulence models for engineering applications, AIAA J. 32 (8) (1994) 1598–1605.
[35] C. Shi, Z. Zhong, D. Feng, China Superalloys Handbook, China Zhijian Publishing House and Standard Press of China, Beijing, 2012.
[36] X. Yan, K. Zhang, Y. Deng, R. Sun, L. Lin, X.J.P. Zhang, P. Research, The effects of DS blade' s geometry features on material' s creep strength, Propul. Power Res.
3 (3) (2014) 143–150.
[37] B.A. Mohamad, A.J. Abdelhussien, Failure analysis of gas turbine blade using finite element analysis, Technology 7 (3) (2016).
[38] W. Maktouf, K. Ammar, I. Ben Naceur, K. Sai, Multiaxial high-cycle fatigue criteria and life prediction: application to gas turbine blade, Int. J. Fatigue 92 (2016)
25–35.
[39] C. Wang, D.Q. Shi, X.G. Yang, S.L. Li, C.L. Dong, An improved viscoplastic constitutive model and its application to creep behavior of turbine blade, Mater. Sci.
Eng. A-Struct. Mater. Prop. Microstruct. Process. 707 (2017) 344–355.
[40] K. Yang, C. He, Q. Huang, Z.Y. Huang, C. Wang, Q.Y. Wang, Y.J. Liu, B. Zhong, Very high cycle fatigue behaviors of a turbine engine blade alloy at various stress
ratios, Int. J. Fatigue 99 (2017) 35–43.
[41] S.P. Zhu, P. Yue, Z.Y. Yu, Q.Y. Wang, A combined high and low cycle fatigue model for life prediction of turbine blades, Materials 10 (7) (2017) 15.

18