Beruflich Dokumente
Kultur Dokumente
RAMS 2005
410
RAMS 2005
3. QUASI-RENEWAL THEORY
3.1 Quasi-Renewal Process
FX n (t ) = FX 1 u ( n 1)t
and
f X n (t ) =
f X 1 u ( n 1)t
u
n 1
).
(1)
n(F(n) (t ) F(n+1) (t )) =
n =1
F (t )
( n)
(2)
n =1
1 *
1 ( n )*
*
*
n 1
f (s ) = f X (s ) f X (us )... f X (u s ) .
s
s
1
(3)
1 *
*
*
*
*
Q* (s) = L F(n) (t ) = f X1 (s ) + f X1 (s ) f X 2 (s ) f X 3 (s )... f X n (s ) (4)
n 1=1
n =1
s
f X*n (s) = f X*1 un1s = f X*1 un2us = f X*2 un3us = " = f X* j 1 (us) (5)
1 *
f X (s ) + f X* (s )q * (us )
s
1
(6)
411
f X*1 (s) =
Q ( s)
sQ* (s) , or equivalently *
FX1 (s) =
*
1
+
sQ* (us)
1 + sQ (us)
(7)
4. MODEL DEVELOPMENT
A(t ) = F (t ) +
F (t x )dM (x )
t
(8)
F (t u )h (u )du
n =1
X n +1
(10)
d
H (u ) .
du
Taking Laplace transform on both sides of eq (10), we derive
(11)
A * (s ) = (1 f X* (s )) + (1 f X* (s ))h * (s )
s
n =1
,
where h (u ) =
j
*
n
where h
(s ) =
*
n
n +1
(s )g (s ) .
*
n
C (t p ) =
expressed as:
(13)
(14)
(16)
C p R (t p ) + C f 1 R (t p )
(17)
(18)
(19)
RAMS 2005
Thus, C (t p ) =
1
1 (u ) n
(12)
A* (s ) =
+
n
i 1
i 1
+ s n =1 + u s i =1 ( + u s )( + s )
C (t p ) =
(9)
A(t ) = F X (t ) +
( f (s )) = L ( f (u s )) and
1
*
Xn
R (T ) = P (T > t ) =
*
X1
n 1
L1 f X* (s ) du .
n
5. CASE STUDY
In recent years the Taiwan Air Force has completed
introducing the new line of jet fighters: F-16A/BMLU, Mirage
2000-5, and IDF. In a certain squadron, a certain planes
engine with an ABNC (after burner nozzle control) that had
been used for a long time. It was discovered during routine
flying training that the engine issued a fault code signal which
caused the after burner nozzle to be disenabled. When this
failure occurred, it caused the maximum thrust to suddenly die
off or otherwise influenced operation, but the engine could
still maintain thrust and normal operation. Therefore from a
safety standpoint this failure presented no immediate
412
significant risk. The only loss was that of cost, because the
problem prevented the execution of routine training. However,
if this failure had occurred during combat and resulted in a
failed mission, human life could have been lost. There was a
vast discrepancy between the actual failure occurrence timing
and that predicted in the technical guidelines. The failure
necessitated corrective maintenance, adversely affected
squadron availability rate, and raised the cost of continued
operation. When consulted, the supplier conducted a failure
investigation and offered an engineering change plan.
However, the changes called for would demand great
expenditure of money and time and involve risk. We collect
the failure data of ABNC from January 1st 1998 to December
31st 2003 with 528 observations. We solve the optimal PM
policy problem by the following procedure.
(1) Fitting general distributions to the ith operating times
{X i } and the ith repair times {Yi } to estimate the scale
(21)
f X* (s ) = L{f X (t )} =
1
0.0035
0.0035 + s
(22)
0.0035
0.0035 + 0.6235 n1 s
(23)
f X* (s ) = f X* (u n 1 s ) =
1
0.0035
Q* (s ) = 1 i 1 = 1
(24)
s n =1 i =1 + u s s n =1 i =1 0.0035 + 0.6235n 1 s
The point availability function can be obtained:
n
1
1 (u ) n
A* =
+
n
i 1
+ s n =1 + u s i =1 ( + u s )
1
0.0035 + s
n
0.6235 0.0035
(
)
+
n
n =1 0.0035 + 0.6235 s
i =1
(25)
0.0035 + 0.6235 s
i 1
Rel i abi l i ty vs T i m e
1. 00
Mult i-Plot
D ata 1
E1 R R X - SR M M ED
0.8
F =326 / S =0
D ata 2
0. 80
E1 R R X - SR M M ED
F =166 / S =0
D ata 3
0.6
E1 R R X - SR M M ED
F =36 / S =0
0.4
Reliability, R(t)=1-F(t)
0. 60
0.2
0. 40
T
500
1000
1500
2000
0. 20
S A N G-C H I N Y A N G
C C IT
2004/ 08/ 12 20: 12
0
0
400. 00
800. 00
1200. 00
Tim e, (t )
1600. 00
2000. 00
1 =0 .0 0 3 5 , =0 .7 8 8 2
2 =0 .0 0 5 8 , =0 .8 3 2 7
3 =0 .0 0 8 1 , =0 .8 5 4 0
RAMS 2005
413
T = 200
n
1
2
3
4
5
6
7
8
9
10
11
12
An
0.78007
0.90050
0.94993
0.97126
0.98120
0.98619
0.98887
0.99037
An+1-An
0.12043
0.04942
0.02133
0.00993
0.00499
0.00267
0.00150
0.00087
T = 300
An
0.62205
0.77483
0.85472
0.89688
0.91989
0.93291
0.9405
0.94502
0.94776
0.94944
0.95047
An+1-An
0.15278
0.07989
0.04215
0.02300
0.01301
0.00758
0.00452
0.00274
0.00167
0.00103
0.00063
T = 500
An
0.19665
0.29044
0.35861
0.40471
0.43479
0.45404
0.46623
0.47390
0.47872
0.48173
0.48361
0.48478
An+1-An
0.09379
0.06816
0.04610
0.03007
0.01925
0.01219
0.00767
0.00481
0.00301
0.00188
0.00117
0.00073
T = 700
An
0.10322
0.15839
0.20060
0.23017
0.24991
0.26274
0.27095
0.27615
0.27942
0.28147
0.28275
An+1-An
0.05516
0.04220
0.02956
0.01974
0.01283
0.00820
0.00519
0.00327
0.00205
0.00128
0.00080
T = 80 ~ 240
T
80
100
120
140
160
180
200
220
240
RAMS 2005
CT
7109.94
6549.32
6231.81
6064.73
6002.55
6019.72
6099.88
6231.13
6403.84
T = 140 ~ 180
T
140
145
150
155
160
165
170
175
180
QT
0.59736
0.62502
0.65336
0.68240
0.71218
0.74272
0.77405
0.86019
0.83916
T = 161 ~ 169
CT
6064.73
6040.57
6022.42
6009.86
6002.55
6000.15
6002.38
6008.98
6019.72
T
161
162
163
164
165
166
167
168
169
QT
0.71823
0.72403
0.73041
0.73655
0.74272
0.74982
0.75515
0.76142
0.76772
CT
6001.68
6001.01
6000.53
6000.25
6000.15
6000.23
6000.5
6000.95
6001.57
(A) Equipment:
The devices, testing stands, and
peripheral equipment used in maintenance were
purchased for 52,259,000NT. If we assume that the
life of this equipment is 20 years and that every year
120 repairs will be conducted, then the consumable
equipment cost of repairing the ABNC was
21,775NT.
(B) Materials: Bearings (1 set of 4) and the disposable
materials used in their repair required cost a total of
206,080NT.
(C) Labor: The time consumed by the squadrons part
change operation was 72.5 hrs./person. As the per
hour cost of one technicians service is 578NT
(monthly salary of 40,785 x 14.5 = 591,382NT
divided by the annual standard work time of 1024
hours), the cost of the operation was 72.5 x 578 =
41,905NT.
Thus, PM Maintenance Cost = (A) + (B) + (C) =
269,760NT. Failure correction maintenance cost C f
refers to the cost loss incurred because flight training
could not be performed plus the cost of PM. Each mission
flown by the aircraft in question costs a total of
700,000NT (including the flight and ground crew salaries,
fuel and ammunition cost). Therefore, the cost of
corrective maintenance was 969,760NT. The numerical
results of BPM are shown in Table 2 for renewal numbers
and maintenance cost rates at different times. The
minimal BPM cost rate is at T = 165 cycles with
maintenance cost rate C(T) = 6000.15 NT. Thus, the
optimal BPM policy is to preventively replace the ABNC
upon reaching every 165 cycles. The numerical results are
shown in Table 1.
(7) Substitute the value of n into eq. (20) to derive the
minimal maintenance cost rate, optimal CM number,
optimal PM replacement time, and the corresponding
system point availability of the APM policy. Based on eq
(20), the numerical analysis of the maintenance cost rate
for APM is shown in Table 3. The minimal APM cost
rate is at T = 176 cycles with maintenance cost rate
C(T)=4247.78 NT. Thus, the optimal APM policy is to
preventively replace the ABNC upon reaching the age of
176 cycles. Compare the results of the BPM and APM
policies to obtain the optimal PM policy. The comparison
results are summarized in Table 4. The APM policy is
better than BPM policy for PM cycle at 165 or 176. In
QT
0.30836
0.39718
0.49296
0.59736
0.71218
0.83916
0.97984
1.13542
1.30667
T = 100 ~ 260
T
100
120
140
160
180
200
220
240
260
QT
0.39718
0.49296
0.59736
0.71218
0.83916
0.97984
1.13542
1.30667
1.49393
CT
4911.1
4568.11
4369.08
4270.04
4249.29
4296.8
4409.4
4588.57
4839.29
T = 160 ~ 200
T
160
165
170
175
180
185
190
195
200
QT
0.71218
0.74272
0.77405
0.80619
0.83916
0.87299
0.90770
0.94331
0.97984
T = 171 ~ 179
CT
4270.04
4258.05
4250.71
4247.83
4249.29
4254.99
4264.84
4278.79
4296.8
T
171
172
173
174
175
176
177
178
179
QT
0.78041
0.78680
0.79323
0.79969
0.80619
0.81271
0.81927
0.82587
0.83250
CT
4249.78
4249.03
4248.45
4248.06
4247.83
4247.78
4247.9
4248.2
4248.66
T
165
176
BPM
Point
Maintenance
Availability
Cost Rate
99.5582
6000.15
99.3925
6010.8
APM
Point
Maintenance
Availability
Cost Rate
99.5582
4258.05
99.3925
4247.78
414
ACKNOWLEDGEMENT
BIOGRAPHIES
We are pleased to thank the National Science Council of
Taiwan for partially support under grant No: NSC 92-2213-E014 -006. We are extremely grateful to Dr. Suprasad V. Amari
and Dr. Joel A. Nachlas for their extraordinary care and
patience in reviewing and improving earlier versions of our
manuscript.
REFERENCES
e-mail: scyang@ccit.edu.tw
1.
2.
3.
4.
5.
6.
7.
8.
9.
RAMS 2005
415