2.8. RELIABILITY OF A COMPONENT BY THE P-S-N CURVES APPROACH 61
erefore, the new mean for the new distributed cyclic number n
j eq
at the stress level
aj
will
be:
j eq
D n
eqij
C
n
Lj
D n
eqij
C exp
"
ln n
Lj
C
ln n
Lj
2
2
#
: (2.61)
e new distributed cyclic number n
j eq
at the stress level
aj
will still be a lognormal distribu-
tion. e log standard deviation of the new distributed cyclic number n
j eq
is the same as that
of n
Lj
.
ln n
j eq
D
ln L
j
: (2.62)
For a lognormal distribution, the log mean of the new distributed cyclic number n
j eq
is:
j eq
D exp
"
ln n
j eq
C
ln n
j eq
2
2
#
: (2.63)
Rearranging Equation (2.63), the log mean of the new distributed cyclic number
n
j eq
is
ln n
j eq
D ln
j eq
ln n
j eq
2
2
; (2.64)
where
ln n
j
eq
and
ln n
j
eq
are the log mean and the log standard deviation of the lognormally
distributed random variable n
j eq
.
If there are more than two stress levels in the model #5 cyclic loading spectrum, we can
use above procedures to continuously convert cyclic stress from one level to next level per Equa-
tions (2.57)–(2.64) until the equivalent cyclic stress in the last stress level has included all trans-
ferred equivalent cycles. If the stress level j is the last stress level, the reliability of the component
under model #5 cyclic loading spectrum is:
R D ˆ
0
B
@
ln N
Cj
ln n
j eq
q
ln N
Cj
2
C
ln n
j eq
2
1
C
A
; (2.65)
where ˆ
.
/
is the CDF of standard normal distribution.
Example 2.16
e component made of steel is subjected to model #5 cyclic loading spectrum with three fully
reversed bending stress levels. Both the numbers of cycles of the fully reversed bending stress
n
L
and the component fatigue life N
C
at each stress level are lognormal distributions as listed
in Table 2.26. Calculate the reliability of the component under such cyclic loading spectrum.