Chinese Journal of Aeronautics 24 (2011) 164-170

Contents lists available at ScienceDirect

Chinese Journal of Aeronautics

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

Theory of Economic Life Prediction and Reliability Assessment of

Aircraft Structures
a, a,b
YAN Chuliang *, LIU Kege
a
Beijing Aircraft Structure Institute, Beijing 100083, China

b
School of Aeronautics Science and Engineering, Beihang University, Beijing 100191, China

Received 26 January 2011; revised 24 February 2011; accepted 14 March 2011

Abstract
The theory of economic life prediction and reliability assessment of aircraft structures has a significant effect on safety of air-
craft structures. It is based on the two-stage theory of fatigue process and can guarantee the safety and reliability of structures.
According to the fatigue damage process, the fatigue scatter factors of crack initiation stage and crack propagation stage are
given respectively. At the same time, mathematical models of fatigue life prediction are presented by utilizing the fatigue scatter
factors and full scale test results of aircraft structures. Furthermore, the economic life model is put forward. The model is of sig-
nificant scientific value for products to provide longer economic life, higher reliability and lower cost. The theory of economic
life prediction and reliability assessment of aircraft structures has been successfully applied to determining and extending the
structural life for thousands of airplanes.

Keywords: life prediction; model of economic life; fatigue scatter factor; full scale test; reliability

1. Introduction1 life and inspection periods are the key question of the
theory. In Refs.[5]-[8], the safe life of aircraft struc-
Failure accidents caused by fatigue occasionally tures was determined by probabilistic fracture me-
[1]
occurred on aircraft structures in the past . The fa- chanics, which is complex and cannot be easily used in
[1-3]
tigue of structures has been studied a lot . The the- engineering practice. Meanwhile, the inspection peri-
ory of economic life prediction and reliability assess- ods were given by studying the relation of crack
ment of aircraft structures has a significant effect on growth and crack size. In this method, the safety of
[4]
safety of aircraft structures .
aircraft structures was not taken into full consideration.
With the development of design and technology,
The safe life can be determined by fatigue scatter
aircraft structures of longer economic life, higher reli-
factor and full scale fatigue test. The fatigue scatter
ability and lower cost have been produced by other
[5] factor plays a vital role in life prediction of aircraft
countries . However, little literature can be found structures. But in the past, the fatigue scatter factor
with regard to theory of economic life prediction and was general and the difference of different stages was
reliability assessment of aircraft structures. The safe [1]
not considered . In this paper, according to the
two-stage theory of the fatigue damage process, the
*Corresponding author. Tel.:+86-10-64882655. fatigue scatter factors of crack initiation stage and
E-mail address: [email protected] crack propagation stage are given respectively. The
Foundation item: National Natural Science Foundation of China whole life are divided into crack initiation life and
(50135010)
crack propagation life. At the same time, the mathe-
1000-9361/$ - see front matter © 2011 Elsevier Ltd. All rights reserved. matical models of fatigue life prediction are presented
doi: 10.1016/S1000-9361(11)60020-4 by utilizing the fatigue scatter factors and full scale test
·No.2
165 YANYAN
Chuliang
Chuliang
et al.et/al.
Chinese
/ Chinese
Journal
Journal
of Aeronautics
of Aeronautics
24(2011)
24(2011)
164-164- No.2
· 165
170 170

results of aircraft structures. Furthermore, a new where P and V are the mean value and standard devia-
method, called economic life model, is put forward by tion of population, respectively.
establishing the relation of crack initiation life and If the fatigue life follows two-parameter Weibull
crack propagation life. The economic life model in- distribution, the probability density function (PDF) can
corporates the design of safe life with the damage tol- be expressed as
erance design organically. The model is of significant
D 1
scientific value in providing longer economic life, D§ N ª § N ·D º
f (N ) · exp « » (3)
higher reliability, and lower cost products. E ¨© E ¹¸ ¨E¸
© ¹ »¼
«¬

2. Fatigue Life and Reliability where D is the Weibull slope (shape parameter) and
The fatigue life of aircraft structures is determined E the characteristic life (with 36.8% reliability).
[9]
by the state of dangerous position . Aircraft structures
3.1. Mathematical model of life prediction with nor-
endure cyclic load during every fight. Crack will initi-
ate and propagate in some dangerous areas after many mal distribution
flights. In the end, the aircraft structures will rupture.
The fatigue life and reliability of aircraft structure If crack initiation life and crack propagation life fol-
are relative. The reliability of aircraft structures is de- low lognormal distribution, the whole life is the sum of
fined as the probability that the aircraft structures per- the lives of the two stages. According to the two-stage
form their intended function under specified condi- theory of fatigue process, the method of fatigue scatter
[1]
tions
[10-11]
, at the same time the economy and mainte- factor of the whole life is established :
nance are taken into account. The failure of aircraft [ N50 ]t
Lf (4)
structures caused by fatigue and rupture is minimized. Np
Quantitatively, the reliability can be expressed as
R 1 F where Lf is the scatter factor of fatigue, [N50]t the me-
(1)
dian test life, and Np the safe life when the reliability is
where R is the probability of survival (also called p.
reliability) , and F the probability of failure.
The median test life is the estimator of the fatigue
If R equals 99.9%, and the safe life of some types of life with 50% reliability. When the logarithm of fatigue
aircraft is 3 000 flight hours, then 999 of 1 000 air-
life follows normal distribution, the logarithm of the
planes are safe and will not be destroyed by fatigue median test life is the mean value of sample, i.e.,
after 3 000 flight hours.
lg [N50]t = x . To obtain the mean value of population
Flight hours, flight takeoff-landing number and cal-
by the mean value of sample, confidence interval
endar life can represent fatigue life of aircraft struc-
method is adopted. The fatigue life can be calculated
tures. If any of the above three reaches its expected
by replacing the mean value of population with the
value, the aircraft will be no longer put into service. In
lower limit of confidence interval.
this paper, flight hours and flight takeoff-landing
When the logarithm of fatigue life follows normal
number are investigated.
distribution, we can obtain
3. Mathematical Model of Life Prediction by Ap- x P
plying Fatigue Scatter Factor of Whole Life ~ N (0,1) (5)
V
n
Usually, the fatigue of fighter plane is caused by
maneuver load, and it follows lognormal distribution. where x is the mean value of sample, and n the num-
Meanwhile, the fatigue of transport plane is caused by ble of sample.
gust load, and it follows two-parameter Weibull dis- When the confidence level J is given, Eq.(6) holds:
tribution. Different reliability and confidence levels are
needed for different life distributions. When the fa- § ·
¨x P ¸
tigue life follows lognormal distribution, at least d uJ
P¨ V ¸ J (6)
99.9% reliability and 90% confidence level are re- ¨ ¸
quired. When the fatigue life follows two-parameter ¨ n ¸
© ¹
Weibull distribution, at least 95% reliability and 95%
confidence level are needed. which can be rewritten as
If the fatigue life, denoted by N, follows lognormal § V ·
distribution, it can be explained that the logarithm of P ¨ P t x uJ ¸ J (7)
© n¹
fatigue life, denoted by x , follows normal distribution,
i.e. where uJ is the standard normal variable with confi-
1
(x P )
2
dence level J.
f (x) e 2V 2 The fatigue life with reliability p and confidence
(2)
V 2S level J can be calculated by replacing P with
·No.2
166 YANYAN
Chuliang
Chuliang
et al.et/al.
Chinese
/ Chinese
Journal
Journal
of Aeronautics
of Aeronautics
24(2011)
24(2011)
164-164- No.2
· 166
170 170

V Eqs.(15)-(16) hold:
x uJ .
n § uJ ·
¨ u ¸V *
When the logarithm of fatigue life follows normal * n p ¹ 0
Lf 10© (15)
distribution, we can obtain
N* [ N*50 ]t [ N*50 ]t
xp P u pV (8) p (16)
L*f § uJ
¨
·
u p ¸V 0*
© n ¹
10
where xp and up are the logarithm of fatigue life and
the standard normal variable with reliability p, re-
where Lf and [N 50 ]t are the fatigue scatter factor
* *
spectly. and the median test life of the crack propagation stage,
By replacing P with x uJ V , Eq.(8) can be re- *
and V is the standard deviation of logarithm of
n 0

written into crack propagation life.

The standard deviation of logarithm of crack propa-
gation life for different metal structures is adopted as
xp x uJ V u pV (9)
*
n
V0 0.07-0.10
In the stage of crack initiation, Eq.(4) shows that the *
When V equals 0.09, the scatter factor of the
0
safe life can be obtained by utilizing the median test
crack propagation stage is also shown in Table 1.
life and fatigue scatter factor.
When the standard deviation of population V =V0 is Table 1 Fatigue scatter factors with 90% confidence
known, the following equations can be obtained ac- level and 99. 9% reliability
cording to Eq.(9) and Eq.(4):
Fatigue scatter factor
V Number of sample
Crack initiation stage Crack propagation stage
lg N p lg [N50 ]t uJ 0 u pV 0
n 1 5.54 2.47
[ N 50 ]t § uJ · 2 4.78 2.29
lg Lf lg lg [N50 ]t lg N p ¨ up ¸ V 0
Np © n ¹ 3 4.48 2.21
then Eq.(10) holds: 4 4.31 2.17

§ uJ
¨ u p ·¸V 0 3.2. Mathematical model of life prediction of two-
n
Lf 10© ¹
(10) parameter Weibull distribution
When up, uJ, n and V0 are all known, the fatigue
If crack initiation life and crack propagation life
scatter factor can be calculated.
follow two-parameter Weibull distribution, the fatigue
According to Eq.(4) and Eq.(10), the safe life can be
life factor is defined as
obtained:
E
[ N 50 ]t Lf (17)
(11) Np
Np § uJ ·
u p ¸V 0
¨
© n ¹ When the fatigue life follows Weibull distribution,
10
its cumulative distribution function (CDF) is given by
In China, the standard deviation of logarithm of
crack initiation life for different metal structures is ª § N ·D º
adopted as F N 1 exp « ¨ ¸ » (18)
«¬ © E ¹ »¼
V 0 0.16-0.20
The CDF of fatigue life with reliability p can be ob-
When V0 equals 0.17, the scatter factor of the crack tained by
initiation stage is shown in Table 1.
To study further, the coefficient of fatigue reliability ª §N Dº
·
SLR and the coefficient of confidence level SLC are F Np 1 exp « ¨ p ¸ » 1 p (19)
E
adopted, which are defined as « © ¹ »
Lf S LR SLC (12) ¬ ¼
SLR u pV 0
(13) According to Eq.(19), we can obtain
10
1
V 1
uJ
0 § 1 ·D
N E ln p D E ln (20)
p ¨ ¸
·No.2
167 YANYAN
Chuliang
Chuliang
et al.et/al.
Chinese
/ Chinese
Journal
Journal
of Aeronautics
of Aeronautics
24(2011)
24(2011)
164-164- No.2
· 167
170 170

S LC n
(14) © p¹
10
Similarly, in the stage of crack propagation,
The coefficient of fatigue reliability is defined as
·No.2
168 YANYAN
Chuliang
Chuliang
et al.et/al.
Chinese
/ Chinese
Journal
Journal
of Aeronautics
of Aeronautics
24(2011)
24(2011)
164-164- No.2
· 168
170 170

E level reaches 95%, the fatigue scatter factor is shown

SR Np (21) in Table 2. For two-parameter Weibull distribution, the
fatigue scatter factors of crack initiation stage and
By utilizing Eq.(20) and Eq.(21), we can obtain crack propagation stage are same.
1
Table 2 Fatigue scatter factor with 95% confidence
§ 1· D
level and 95% reliability
(22)
S ln
R ¨ p ¹¸
©
Number of sample 1 2 3 4
The fatigue life Ni (i=1, 2, n , n) is obtained by Fatigue scatter factor 2.75 2.58 2.50 2.46
small sample test, and the estimator of characteristic
life can be obtained by applying the maximum like- 4. Mathematical Models of Reliability of Fatigue
Eˆ
lihood method: Life
1
ˆ The coefficient of fatigue reliability SLR(SR) reflects
§1 n ·D
E ¨ ¦ NDi ¸ (23) the dispersion of fatigue life, and the structure reliabil-
©ni 1 ¹ ity coefficient, FRF, reflects the characteristic of
The shape parameter D can be obtained by experi- structure. With the development of durability and
ence and test. damage tolerance technology, the structure reliability
In fact, the characteristic life cannot be obtained by coefficient is used to predict aircraft structure fatigue
finite test data but it can be calculated by the lower life. When the check of structure is very difficult and
confidence limit of estimator of characteristic life. the cost of maintenance is high, the value of structure
reliability coefficient should be increased.
° Eˆ ½° Consider the reliability and confidence level, the
P ®E t (24)
°¯ SC ¾°¿ C expected design life of aircraft structure NE,C/R is de-
fined as
where C and SC are the confidence level and the coef-
N C/ R
ficient of confidence level with two-parameter Weibull N E,C / R (31)
distribution. FRF
where NC/R is the basic reliability life.
The characteristic life can be given as If the fatigue life accords with two-parameter
Eˆ Weibull distribution, the safe factor is defined as
E (25)
SC FW SC S R S t FRF (32)

When the shape parameter is known, we can obtain where St is the coefficient of test sample size. For full
scale fatigue test, St equals 1.
SC D nn D n 1 D The expected design life of aircraft structure by
³0 * n
y e
ny
dy C (26)
adopting the estimator of characteristic life is defined
as
where * (n) is Gamma function, and y the variable,
then SC can be known by solving Eq.(26). N E,C / R
Eˆ Eˆ (33)
According to Eqs.(20)-(21) and Eq.(25), the safe life FW SC S R S t FRF
can be expressed as According to Eq.(31) and Eq.(33), the basic reliabil-
ity life can be given by
E Eˆ
Eˆ
Np (27)
SR S R SC Lf Eˆ
NC / R NE,C / R FRF (34)
SC S R S t
It can be known that
Lf S R SC (28) When both the reliability and the confidence level
equal 95%, Eq.(33) can be rewritten as
Similarly, in the stage of crack propagation, we can
obtain N E,95 / 95 Eˆ (35)
S95 S95 S t FRF
E
*
E ˆ*
Np* (29) where the FRF equals 1-2 for the structure designed by
SR S RSC
the damage tolerance technology; FRF equals 2-4 for
Lf* S R SC (30) the structure designed by the safe life technology.
The relations of different lives of aircraft structures
When the reliability equals 95% and the confidence with different coefficient are shown in Fig.1.
 ·No.2
169 YANYAN
Chuliang
Chuliang
et al.et/al.
Chinese
/ Chinese
Journal
Journal
of Aeronautics
of Aeronautics
24(2011)
24(2011)
164-164- No.2
· 169
170 170

art suitcase for structural health monitoring.