Download as pdf or txt
Download as pdf or txt
You are on page 1of 5

Availability Model of Equipment Based on

Three-Levels Maintenance System


Qi Gao, You-Wei Li, Yang Ge, Bing Hao
Department of Management Engineering
Mechanical Engineering College
Shijiazhuang, China
gy090@163.com.cn

Abstract—Targeting at the complex equipment guaranteed by and optimum concept was analyzed, Xu [5] established two
three-levels maintenance institution, on the assume that “after mathematical models of optimum maintenance cycle. Through
minor repair is as good as old and after middle or overhaul repair analyzing an application case, the optimal two-level
is not as good as new”, it presented a switching chart of maintenance cycle that can ensure the availability is maximal
equipment application and repair condition transformation in a was received. Targeted at the complex equipment of three
lifecycle by analyzing equipment application and repair flowing
maintenance levels, Gao [6] presented a switching chart of
procedure. It also established the connection model between
maintenance interval period and availability in the lifecycle, and equipment application and repair condition transformation by
worked out the average availability related to the fixed analyzing equipment application and repair flowing procedure.
maintenance interval period and switch rule. Through analyzing The paper also established the connection model between
an application case, the optimal maintenance interval period that maintenance interval period and availability in a updating
can ensure the availability is maximal was received, and cycle, and also gave an application case, in which the
comparing with “after overhaul repair is as good as new”, in applicability and sensitivity of the model are proved.
which the applicability and sensitivity of the model are proved.
This approach can be used for maintenance decision-making. However, based on the three-levels maintenance support
system, the maintenance decision-making modeling that
Keywords-three-levels maintenance; complex equipment; consider different level maintenance has different effect was
availability; maintenance interval period; maintenance switch rule unfamiliar.

I. INTRODUCTION Targeting at the complex equipment guaranteed by


three-levels maintenance institution, on the assume that
Currently, a few scholars have done some work in different level maintenance has different maintenance effect, it
maintenance decision-making modeling and optimization of presented the connection model between maintenance interval
multi-levels maintenance system. period and availability in the lifecycle, and worked out the
According to the traditional maintenance system, through average availability in a lifecycle. What we consider in this
using Markov process theory, Galina [1] optimized the paper is: on the assume that the maintenance rule is set up,
maintenance rule of airplane based on three maintenance how to change the maintenance interval period can insure the
system, and which be used to improve the availability. average availability is maximal.
Targeted at the thermal power stations supported by two-levels The complex equipment in this paper is: it is guaranteed by
maintenance institution, on the assume that after first-level three-levels maintenance institution, adopt the maintenance
maintenance is as good as old and after second-level mode that combine with periodic maintenance and corrective
maintenance is as good as new, the allowable reliability was maintenance. Usually, if simple failure occurs on the
selected as constraint, Masatoshi [2] established the equipment, the surfaceman will implement minor maintenance
maintenance cost decision-making model that based on the in the local. If the assembly and discreteness of the
dimensional reduction method. Targeted at the equipment that
equipment occur failure or implement complicated periodic
need imperfect maintenance, Cassady [3] established the
maintenance, it will implement middle maintenance. If the
availability simulation model, and analyzed the influence of
important assembly and discreteness of the equipment occur
imperfect maintenance effect that refer to the availability of
equipment. Targeted at the multi-components and series complex failure or implement all-sided periodic
parallel mechanical devices, on the assume that after repair is maintenance, it will implement overhaul maintenance.
between as good as old and as good as new, the allowable According to the current support scheme, the maintenance
availability was selected as constraint, the maintenance cost switch rule is: the equipment implements once maintenance at
was identified as optimization objective, Su Chun [4] most in every maintenance interval( If the working hours is
established the maintenance optimization model. The Monte longer than T , implement preventive maintenance. Otherwise,
Carlo simulation was used to compute the model. In order to corrective maintenance). If the equipment implements twice
accurately decide vehicle optimum maintenance period under minor maintenance, the second times minor maintenance will
availability objective, the renewal function of renewal theory translate into middle maintenance. If the equipment

978-1-4577-1232-6/11/$26.00 ©2011 IEEE


implements twice middle maintenance, the second times organization maintenance, j = 2 denote the intermediate
middle maintenance will translate into overhaul maintenance. maintenance, j =3 denote the depot maintenance)
If the equipment implements twice overhaul maintenance, the
second times overhaul maintenance will translate into reject p ──The overhaul maintenance instance of the equipment
(Because of maintenance cost and stock cost are relatively ( p = 0,1,2), p = 0 denote the instance that the equipment
high, and the need of update, the equipment will be deal with
doesn’t implement overhaul, p =1 denote the instance that the
reject). Namely, according to the definite maintenance interval
period and maintenance rule, the equipment implement equipment has implemented once overhaul, p =2 denote the
maintenance that when and which maintenance deepness are instance that the equipment has been rejected.
selected.
i ──The maintenance states that the i times maintenance
II. MODELING has been implemented in a overhaul cycle ( i =0,1,2,3,4),
specially, i =0 denotes the fire-new states.
A. Hypothesis
Most of the models made the system as good as new after M pi , θ pi ──Respectively denote the average maintenance
maintenance, basing on the hypothesis that replacing time and the average working hours from the i − 1
was the only maintenance way, and ignored maintenance maintenance state to the i maintenance state when the
time. However, system (or equipment, assembly) have equipment has been implemented the p overhaul.
different maintenance ways actually, and different
maintenance ways effects different results, and sometimes M , θ ──The average maintenance time and the average
the maintenance time can not be ignored. So researchers working hours in a lifecycle.
advanced some other maintenance ways besides replacing, C. the switching chart
Minor Maintenance: maintenance man repairs According to hypothesis, if the equipment implements
broken-down system less, makes the system come back to the twice overhaul maintenance, the second times overhaul
state before failure, i.e. the failure rate after maintenance will translate into reject, namely there are two
maintenance is same to before failure; Middle overhaul cycles from new to reject in a lifecycle of the
equipment. The equipment application and repair flowing
Maintenance: maintenance man repairs system less,
procedure in a lifecycle as shown in Fig. 1.
improves the system’s aging less; Overhaul Maintenance:
maintenance man repairs system adequately, improves the p =0 p =1 p =2
system’s aging evidently. Basing on these, we propose
N O R
hypothesis as follows:
new first overhaul reject
(1) The usage and maintenance of equipment both are
continuous, without interruption.
first overhaul cycle second overhaul cycle
(2) After minor repair, the equipment is as good as old,
after middle or overhaul repair, the equipment is not as good as Figure 1. Equipment application and repair flowing procedure in a lifecycle
new.
Because of the comparability of every overhaul cycle, in
B. Notation this paper, it only considered the overhaul cycle of the
T ──maintenance interval; equipment from new to the first overhaul. In the first overhaul
cycle, if the working hours is longer than T, implement
t0 , t1 ──Respectively denote the prolonged time of the preventive maintenance, namely implement the corresponding
equipment that after middle and overhaul maintenance (known, minor maintenance, middle maintenance or overhaul.
t1 > t0 ) Otherwise, implement corrective maintenance, namely
implement the corresponding minor maintenance, middle
R ( t ) , F ( t ) ──Respectively denote the reliability function maintenance or overhaul. In order to illuminate the question,
and fallibility function of the equipment in this paper, it only analyzed the two states from i − 1 state
(known), R ( t ) + F (t ) =1. to i state, the switching chart of equipment between two
states as shown in Fig. 2.
f ( t ) ──The failure density function (known). i ' state─the equipment implemented preventive
maintenance in the j level insititution when there was
M ptj , M ctj ──Respectively denote the average
no-one failure in the maintenance interval,
maintenance time in the j level institution of the equipment
that not failure and failure (known, j = 1 denote the
⎧3 i = 4 So, according to the definition of average availability,
⎪ average availability by formula is:
j = ⎨2 i = 2 (1)
⎪1 the rest θ
⎩ A1 = (4)
''
i state ── the equipment implemented corrective θ +M
maintenance in the j level insititution when there was some If the equipment is as good as new after overhaul, the
application and maintenance flow of the equipment is
failure in the maintenance interval, and the j equate (1). regenerative, in this paper, it only considered the overhaul
cycle of the equipment from new to the first overhaul. The
i' preventive formula of average availability can predigest:
using maintenance 4

i -1 i θ 0i ∑θ 0i
A2 = = i =1
(5)
θ 0i + M 0i 4 4

using
corrective ∑ θ 0i + ∑ M 0i
maintenance i =1 i =1
i ''
III. MODEL CALCULATION
θ 0i M 0i Through above analysis, there were two likelihood in the
first overhaul cycle from new 0 to 1 state: one was 0→1'→1,
Figure 2. The switching chart of equipment between two states specially the probability is R (T ) , the equipment entered into 1 state
In a word, the maintenance state order in the first overhaul after M pt1 . The other was: 0→1''→1, the probability is
cycle is: new→minor maintenance→middle F (T ) , the equipment entered into 1 state after M ct1 .
maintenance→minor maintenance→overhaul, so the switching
chart of equipment application and repair state in the first Therefore, the average working hours and maintenance time of
overhaul cycle as shown in Fig. 3. the equipment from 0 state to 1 state were:
T T

1' 2' 3' 4' θ 01 = R (T ) T + ∫ tf (t )dt = ∫ R(t )dt


0 0

M 01 = R (T ) M pt1 + F (T ) M ct1
0 1 2 3 4
Before entered into 1state, the already average working
hour of the equipment was θ 01 . There were two likelihood
from 1state to 2 state: one was 1→2'→2, the probability is
1'' 2'' 3'' 4'' R (T + θ 01 ) , the equipment entered into 2 state after M pt 2 .

Figure 3. The switching chart of equipment application and repair condition


The other was: 1→2''→2, the probability is F (T + θ 01 ) , the
in the first perfect maintenance lifecycle equipment entered into 2 state after M ct 2 . Therefore, the
From the Fig. 3, we can find the application and average working hours and maintenance time of the equipment
maintenance flow of the equipment in the first overhaul cycle from 1 state to 2 state were:
is: 0→1'(or 1'')→1→2'(or 2'')→2→3'(or 3'')→3→4' T +θ 01

(or 4'')→4. θ 02 = (T + θ 01 ) R (T + θ 01 ) − θ 01 R (θ 01 ) + ∫ tf (t )dt


θ01
D. Availability model T +θ 01

The average working hours θ of the equipment in a = ∫ R(t)dt


lifecycle is the sum of average working hours in every phase, θ01

namely M 02 = R (T + θ 01 ) M pt 2 + F (T + θ 01 ) M ct 2
1 4 According to hypothesis, the prolonged time of the
θ = ∑∑ θ pi (2) equipment that after middle maintenance was t0 , so before
p = 0 i =1

The average not working hours M of the equipment in a entered into 2 state, the already average working hours of the
lifecycle is the sum of average maintenance time in every equipment was θ 01 + θ 02 − t0 . Therefore, the average working
phase, namely hours and maintenance time of the equipment from 2 state to 3
state were:
1 4
M = ∑∑ M pi (3)
p = 0 i =1
T+
0 2

∑∑θ pk − t0 is 100h: A1 =0.8129.


p =0 k =1

θ 03 = ∫ R ( t )dt When the equipment is as good as new after overhaul:


0 2
∑∑θ pk − t0 A2 =0.8245
p =0 k =1

⎛ 0 2 ⎞ ⎛ 0 2 ⎞ TABLE I. THE RELATIONAL TABLE BETWEEN T AND A


M 03 = R ⎜ T + ∑ ∑ θ pk − t0 ⎟ M pt1 + F ⎜ T + ∑ ∑ θ pk − t0 ⎟ M ct1
⎝ p = 0 k =1 ⎠ ⎝ p = 0 k =1 ⎠ the prolonged time is 100h after overhaul
Because of the comparability of every state, so before NO. T (h) A1 NO. T (h) A1
entered into i − 1 state, the already average working hours of
1 120 0.7382 11 320 0.8036
the equipment was (in order to analyze easily, θˆ instead of pi 2 140 0.7663 12 340 0.8001
3 160 0.7877 13 360 0.7968
that) 4 180 0.8032 14 380 0.7934
p i −1 5 200 0.8129 15 400 0.7900
∑ ∑θlk − p ( t0 + t1 ) − ⎢⎣⎢ i −21⎥⎦⎥ t0 6 220 0.8170 16 420 0.7865
l =0 k =1
7 240 0.8167 17 440 0.7830
Specially, ⎢⎣ x ⎥⎦ denotes the value of the real number x 8 260 0.8140 18 460 0.7796
9 280 0.8106 19 480 0.7765
that evaluate the alow integer. 10 300 0.8071 20 500 0.7740
T +θˆpi the equipment is as good as new after overhaul
θ pi = ∫ R ( t )dt NO.
1
T
120
(h) A2
0.7385
NO.
11
T
320
(h) A2
0.8767
θˆpi
2 140 0.7671 12 340 0.8801
(
M pi = R T + θˆpi M ptj + F T + θˆpi M ctj ) ( ) 3
4
160
180
0.7901
0.8089
13
14
360
380
0.8821
0.8827
5 200 0.8245 15 400 0.8820
Through above results, We could worked out the average 6 220 0.8376 16 420 0.8801
working hours θ by formula (2), the average maintenance 7 240 0.8486 17 440 0.8775
time M by formula (3) in a lifecycle and the average 8 260 0.8580 18 460 0.8745
9 280 0.8656 19 480 0.8715
availability A by formula (4) or (5) in a lifecycle. 10 300 0.8719 20 500 0.8690
IV. CASE
The some kind of complex equipment guaranteed by
three-levels maintenance insititution, the life obeys Weibull
distribution that the parameters are 7 and 1750h, the
maintenance interval is 200h, and on the assume that the
prolonged time is 50h after middle maintenance and 100h after
overhaul. Through statisticing and analysing the older
maintenance data, we found: the average preventive
maintenance time and corrective maintenance time of this
equipment were 10h and 20h in organization insititution, 50h
and 100h in middle insititution, 100h and 200h in depot
insititution respectively. On the assume that after overhaul is
not as good as new and as good as new, please try to calculate
the average availability A and analyse the connection in a
lifecycle.
Figure 4. The variational graph of A along with T
From above case we know:
⎛t⎞
m Targeted at two instances: one is the prolonged time of the
−⎜ ⎟
R (t ) = e ⎝η ⎠
, F ( t ) = 1 − R ( t ) , m =7, η =1750h, T = equipment after overhaul is 100h, the other is the equipment is
as good as new after overhaul. The relational table between T
200h, t0 =50h, t1 =100h, M pt1 =10h, M ct1 =20h, M pt 2 = and A as shown in Table I and fig. 4 was received through
50h, M ct 2 =100h, M pt 3 =100h, M ct 3 =200h. calculating, which the maintenance interval T was
120h,140h,160h,…,460h,480h and 500h. Specially, in figure 4,
Through making use of the computer procedure, the graph 1 denote the instance that the prolonged time of the
average availability can be worked out by formula (4) and equipment after overhaul is 100h, graph 2 denote the instance
formula (5): that the equipment is as good as new after overhaul.
When the prolonged time of the equipment after overhaul Form the Table I and Fig. 4, we found that, when the
maintenance interval T <160h, the average availability A
increased gradually along with the increase of the maintenance between maintenance interval period and availability in the
interval T , graph 1 and graph 2 were very similar because of lifecycle, and worked out the average availability. Through
the maintenance interval T was smaller and the prolonged analyzing an application case, the optimal maintenance
time t1 was bigger. When the maintenance interval T >160h, interval (220h) that can ensure the availability is maximal was
received, and comparing with “after overhaul repair is as good
the average working hours θ and the average availability A
as new”, in which the applicability and sensitivity of the model
increased along with the increase of the maintenance interval
are proved. This approach can be used for maintenance
T , specially, because of the difference of the overhaul
decision-making.
instances: one is the prolonged time is 100h after overhaul
(instance 1), the other is the equipment is as good as new after REFERENCES
overhaul (instance 2), graph 1 and graph 2 were very different, [1] G. M. Susova, and A. N. Petrov, “Markov model-based reliability and
and the average availability of the instance 2 was bigger than safety evaluation for aircraft maintenance-system optimization,” In
the instance 1. At the same time, through statisticing and Proceedings of Annual Reliability and Maintainability Symposium,
analysing the older application and maintenance data, the pp.29-36, 1997.
instance 1 was very actual was found. Targeted at instance 1, [2] M. Nakamura, T. Katafuchi, and H. Hatazaki, “Decisions for
maintenance-intervals of equipment in thermal power stations based on
according to the actual maintenance scheme, when the few data,”. IEEE Transactions on Reliability, vol. 4, pp.360-364, 2001.
maintenance interval T =200h, the average availability A [3] C. R. Cassady and I. M. Iyoop, “A generic model of equipment
was 0.8129. But when the maintenance interval T =220h, the availability under imperfect maintenance,” IEEE Transactions on
Reliability, vol. 4, pp.564-571, 2005
average availability A was 0.8170. Therefore, when the
[4] C. Sun, Z. Huang, and Y. Xu, “Modeling and optimization for devices’s
maintenance interval T =220h, the average availability A maintenance based on allowable availability constraint and maintenance
was maximal, and the A was very sensitive along with T . Cost,” China Mechanical Engineering, vol. 9, pp. 1096-1099, 2007.
[5] A. Xu and X. Qiao, “Vehicle optimum maintenance period under
V. CONCLUSIONS availability objective,” Journal of Traffic and Transportation Engineering,
vol. 2, pp. 23-26, 2008.
Targeting at the complex equipment guaranteed by [6] Q. Gao, Y. Li, and B. Hao, “Availability model of equipment based on
three-levels maintenance institution, on the assume that “after three-levels maintenance system,” Fire Control & Command Control,
minor repair is as good as old and after middle or overhaul vol. 5, pp.107-109, 2010.
repair is not as good as new”, it presented a switching chart of [7] M. Gan, J. Kang and Q. Gao, Military Equipment Maintenance
equipment application and repair condition transformation in a Engineerimg Science. Beijing: National Defense Industry Press, 1999.
lifecycle by analyzing equipment application and maintenance [8] J. Cao and K. Cheng, Introduction to Reliability Mathematics, Beijing:
Higher Education Press, 2006.
flowing procedure. It also established the connection model

You might also like