Professional Documents
Culture Documents
Reliability Availability and Maintainability Analysis of The Conveyor System in Mechanized Tunneling
Reliability Availability and Maintainability Analysis of The Conveyor System in Mechanized Tunneling
Measurement
journal homepage: www.elsevier.com/locate/measurement
a r t i c l e i n f o a b s t r a c t
Article history: Evaluation of reliability, maintainability and availability (RAM) plays an important role in performance
Received 24 July 2018 estimation of tunneling systems. Reliability predictions, maintenance planning, reliability assessment,
Received in revised form 15 May 2019 failure detection and risk evaluation are necessary for tunneling machines. The RAM analyses of tunnel-
Accepted 4 June 2019
ing equipments have been widely investigated in recent years. The aim of this study is to RAM analysis of
Available online 10 June 2019
the material hauling system in an earth pressure balance tunnel boring machine (EPB-TBM). For this pur-
pose, failures and repairs data from main conveyor system of Tabriz Metro Line 2 in Iran were collected.
Keywords:
Performing trend and serial correlation tests showed that the data are independent and identically dis-
Reliability
Availability
tributed (iid) and therefore the statistical techniques can be used for modeling. To carry out the analysis,
Maintainability the main conveyor system was divided into three sub-systems including conveyors 1, 2 and 3, which are
Tunnel conveyor system located on the TBM machine, inside the tunnel and station. Analysis results indicated that reliability of
Mechanized Tunneling conveyor subsystems 1, 2 and 3 after 267, 58 and 300 h reaches to zero, respectively. Moreover, availabil-
ity results showed that all subsystems are available for more than 89 percent of the time. To improve sys-
tem reliability, preventive maintenance is required for each subsystem. Based on the obtained results,
reliability-based preventive maintenance time intervals for 80 percent of reliability level for subsystem
1, 2 and 3 are 43, 3 and 6 h, respectively.
Ó 2019 Elsevier Ltd. All rights reserved.
https://doi.org/10.1016/j.measurement.2019.06.009
0263-2241/Ó 2019 Elsevier Ltd. All rights reserved.
S. Ahmadi et al. / Measurement 145 (2019) 756–764 757
identifying the critical subsystem that needs improvement or distribution should be investigated. In order to analyze this
repair, and ultimately providing an optimal maintenance plan for assumption, two common methods of trend test and correlation
the conveyor system are the objectives of this paper. test are used [5].
As features of long-term system operation, reliability, availabil- The trend test actually determines whether the disruption of
ity, and maintainability analysis are significant approaches to the failure data is significantly (degraded or improved) changed
reduce maintenance costs and improve system function and over the time interval or not. Cumulative data is used to
operation. describe the existence or absence of trends. Military, Laplace,
and Graphic tests are commonly used to determine the trend
2.1. Reliability in data [22].
In the Military test [23], the evaluation of the trend existence in
Reliability is the ability of a system to perform a task under the the data is done using the calculation of a statistical index, which
given conditions and at a specified time interval. When a system defines as follows:
can perform the specified task in more time, it is more reliable X
n1
Tn
[15]. Therefore, identifying the different predictable conditions U¼2 ln ð4Þ
and operating modes, and also the use and non-use of sub- i¼1
Ti
systems including system, equipment, components and etc., are
where, n is the number of failures, Tn is the last failure time and Ti is
essential in the specification phase required for system design
the time of failure.
[16]. If the failure probability of an item in the time interval from
In this method, the null hypothesis is that the data are free from
zero to t is shown by F(t), given that the function F(t) is the unre-
the trend. Based on this assumption, the U index is Chi-squared
liability function, reliability is expressed as:
distributed with 2(n-1) degrees of freedom. The Chi-squared distri-
Z 1
bution table is used to control the accuracy of the null hypothesis,
Rðt Þ ¼ f ðt Þdt ð1Þ
t after calculating U for data. If U is greater than the critical number
from the standard table, the null hypothesis is accepted. If this
where, R(t) is the reliability at time t and f ðtÞ represents the prob-
assumption is rejected, the data has trend.
ability density function of the failure.
In the Laplace test [22], evaluation of the trend in the data is
Systems are generally divided into two groups of parallel and
done using the following statistics index:
serial (or sequential) systems [17]. The system is series if the sys-
Pn1
tem fails with the failure of only one of the components of the sys- ð i¼1 t i =n 1Þ ðt n =2Þ
LA ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð5Þ
tem and its reliability is defined as follows: tn ð1=12ðn 1ÞÞ
Y
n
Rsys ¼ R1 R2 Rn ¼ Ri ð2Þ Given the critical value range for the 95% confidence level,
i¼1 which is between (1.96, +1.96), the critical value of the obtained
LA is determined from the standard table. If this critical number is
where, Ri is the reliability of the different subsystems (compo-
within the range, the null hypothesis is rejected, show that the
nents). In a parallel system, the function of the system requires
data follow a trend. But if it is not within this range, the null
the operation of at least one of its units [18]. In other words, the sys-
hypothesis is accepted, show that the data are trendless.
tem fails when all units fall off. In this case, the reliability of the sys-
In the Graphical test, the cumulative TBFs and TTRs data are
tem is defined as follow:
plotted against the cumulative number of failures and repairs. If
Y
n a line drawn through the data points is shrinking upwards or
Rsys ¼ 1 F sys ¼ 1 ½1 Ri ð3Þ downwards, the data have trend and the system is an improve-
i¼1
ment or breakdown system, respectively. If the line drawn through
the data is a straight line, the data are trendless, which means that
2.1.1. Statistical analysis method the data sets are distributed equally.
Generally, this method is used to evaluate the reliability of
repairable systems by the mean time between failures (MTBF) Correlation test
and for non-repairable systems with average time to failure (MTTF)
[19]. Failure is due to the inability of a piece of system or equip- In order to determine the correlation, the ith TBF (or TTR) is
ment to operate under certain conditions [16]. The major causes plotted against (i-1)th TBF (or TTR), if the data are independent
of mechanical failure include poor or incomplete design, manufac- and correlated, points are located along one line [20].
turing defects, incorrect use, improper placement and installation,
wear out, failures in other parts of the system, and gradual degra- Data analysis
dation in operation. The general algorithm of the statistical analy-
sis method is shown in Fig. 1. In the statistical analysis method, a system is modeled by ana-
lyzing TBF and TTR data. Determination of the best probability dis-
Pareto analysis tribution functions for different sub-systems is accomplished by
two methods of non-homogeneous Poisson process (NHPP) and a
Pareto analysis is the first step in reliability statistical analysis renewal process (RP). If the data have trend, the power law process
of a system. In this analysis, a system is divided into several model (PLP) is used. Power law process model has same advan-
sub-systems. Pareto analysis provides the possibility to identify tages of Weibull distribution and so flexible in describing the time
troubled areas that should be analyzed as widely as possible using data [24]. In the absence of trend and serial correlation, the
statistical methods [20,21]. Renewal Process is used.
After data collecting and before finding the best fit distribu- The reliability function and failure rate functions are as the rela-
tion, the assumption of failure data independence and identically tions (6) and (7), respectively.
758 S. Ahmadi et al. / Measurement 145 (2019) 756–764
b !
t In the absence of trend and serial correlation, the RP is used.
Rðt Þ ¼ exp ð6Þ Classic methods are used to find the best fit distribution in the
h
RP. For this purpose, various distribution functions are tested on
b1 the data to select the best distribution. The Kolmogorov-Smirnov,
b t Anderson-Darling and Chi-square methods are used to select the
kðt Þ ¼ ð7Þ
h h best probability density function. According to the rating of good-
where, h is scale parameter, b is shape parameter and t is the run- ness of fit test (GOF) of these methods, the best distribution func-
time. The scale and shape parameters are defined as follow [25]: tion for each sub-system is selected [13]. The goodness of fit tests
for the above-mentioned methods is tabulated in Table 1.
n
b¼P ð8Þ
n1 tn
i¼1 ln ti
2.2. Availability
tn
h ¼ 1=b ð9Þ Availability is the probability that a piece of equipment/system
n
is functioning satisfactorily at time t when used according to spec-
S. Ahmadi et al. / Measurement 145 (2019) 756–764 759
Test The statistics Accepted of H0 The failure and repairable data of the material hauling system
Chi-square test [28] Pk ðni ei Þ2 2
X < X 2c
were recorded for continuous analysis. In this regard, various
X 2 ð/; k 1 dÞ ¼ i¼1 ei
sources such as operational and maintenance information, daily
Kolmogorov-Smirnov max DN DacðNÞ
DN ¼ jF ðxÞ SN ðxÞj maintenance reports and data from sensors for reliability modeling
test (K-S) x
Anderson-Darling P ð2i1Þ have been used. To carry out the analysis, the main conveyor sys-
A2 ¼ N N i¼1 N ½lnF ðBi Þ
A2 < A2c
test (A-D) þlnðð1 F ðBNþ1i ÞÞ tem was divided into three sub-systems including conveyors 1, 2
and 3, which are located on the TBM machine, inside the tunnel
ified conditions and in an ideal environment. Inherent availability and station, respectively. As the basis of RAM analysis, the TBF
is the most common definition used in reliability literature [26]. and TTR data were determined as in Fig. 2. The items of each
The average availability of an item can be determined from the fol- sub-system include belt, propulsion, and traction-support systems.
lowing equation. The duration of excavation was used as a time unit in analysis.
Figs. 3 and 4 show the Pareto diagrams of various conveyor sys-
MTBF tems in terms of failures number and failures times. These figures
A¼ ð10Þ
MTBF þ MTTR illustrate that the conveyor 2 is the most critical sub-system in
terms of the number of failures (244 failures) and the failure time
where, MTBF and MTTR are the mean time between failures and the (9345 min of failure).
mean time to repair, respectively. To determine the main reasons of failures in each sub-system,
the Pareto charts for each item was plotted based on the number
2.3. Maintainability of failures and breakdowns that are shown in Figs. 5 and 6. Accord-
3. Case study
Table 2
Specifications of the conveyor system used inside the tunnel.
Fig. 5. Pareto chart of the conveyor sub-system items based on the number of
failures.
1:118 !
t
R2 ðt Þ ¼ exp ð13Þ
563:411
Fig. 7. Frequency failures of conveyor 1 components.
S. Ahmadi et al. / Measurement 145 (2019) 756–764 761
Table 3
Failure data for conveyor sub-system 1.
No. TBF TTR CTBF CTTR No. TBF TTR CTBF CTTR
1 1199.22 15 1199.22 15 28 109.02 70 48350.41 1020
2 381.57 30 1580.79 45 29 218.04 195 48568.45 1215
3 2507.46 60 4088.25 105 30 218.04 390 48786.49 1605
4 4905.90 60 8994.16 165 31 54.51 350 48841.00 1955
5 5396.49 30 14390.65 195 32 1580.79 460 50421.79 2415
6 7849.45 130 22240.10 325 33 54.51 460 50476.30 2875
7 3270.60 5 25510.70 330 34 54.51 150 50530.81 3025
8 599.61 30 26110.31 360 35 763.14 60 51293.95 3085
9 490.59 100 26600.90 460 36 163.53 305 51457.48 3390
10 3815.70 30 30416.61 490 37 2616.48 15 54073.97 3405
11 163.53 55 30580.14 545 38 3107.07 35 57181.04 3440
12 545.10 50 31125.24 595 39 163.53 10 57344.57 3450
13 1144.71 60 32269.95 655 40 2616.48 30 59961.05 3480
14 2125.89 15 34395.84 670 41 109.02 40 60070.07 3520
15 654.12 15 35049.96 685 42 109.02 40 60179.09 3560
16 872.16 35 35922.12 720 43 545.10 10 60724.19 3570
17 6432.19 15 42354.31 735 44 1090.20 15 61814.39 3585
18 3488.64 15 45842.95 750 45 1853.34 340 63667.73 3925
19 381.57 30 46224.52 780 46 436.08 25 64103.81 3950
20 708.63 15 46933.15 795 47 545.10 245 64648.91 4195
21 54.51 10 46987.66 805 48 218.04 20 64866.96 4215
22 708.63 10 47696.29 815 49 327.06 15 65194.02 4230
23 109.02 30 47805.31 845 50 54.51 30 65248.53 4260
24 109.02 20 47914.33 865 51 54.51 440 65303.04 4700
25 54.51 30 47968.84 895 52 4088.25 80 69391.29 4780
26 54.51 40 48023.35 935 53 327.06 40 69718.35 4820
27 218.04 15 48241.39 950 54 109.02 65 69827.37 4885
5.2. Availability
ln ðt þ 20:842Þ 6:9288 The availability of each sub-system was calculated by using Eq.
R3 ð t Þ ¼ 1 £ ð14Þ (9). The MTBF value for each sub-system was obtained through
1:2106
dividing the cumulative time between failures (CTBFs) by the total
The reliability of each sub-system was determined by the best number of sub-system failures. Similarly, the MTTR value for each
probabilistic distribution function. The reliability of each sub- sub-system was determined through dividing the cumulative time
system for TBFs data was calculated individually and their reliabil- to repair (CTTR) by the total number of failures. The corresponding
Table 4
Trend test results.
Table 5
Existence or absence of trend.
Fig. 11. Scatter plot of TBF and TTR data for different sub-systems.
values of MTBF and MTTR for each sub-system with the number of 5.3. Maintainability
failures are tabulated in Table 8.
The results show that conveyors 1, 2 and 3 are available more Similar to the process performed in the reliability analysis on
than 93, 89, and 97 percent of the time, respectively. Also the total TBFs data, results from the TTRs data were used for maintainability
conveyor system will be available at 80% of the time. analysis. Regarding the results of goodness of fit tests, log-normal
S. Ahmadi et al. / Measurement 145 (2019) 756–764 763
Table 6
Best distribution functions for TBF data.
Table 7
Best distribution functions for TTR data.
Fig. 13. Maintainability chart of the conveyor sub-systems and the overall hauling
Fig. 12. Reliability diagram for sub-systems and general conveyor system. system.
distribution function was determined as the best distribution func- ure repair time of the conveyor 1 is about 180 min with 90%
tion for all sub-systems. The Maintainability functions for sub- probability while this repair time for the conveyors 2 and 3 is
systems 1, 2 and 3 for the TTRs data are as follow: 80 min. In addition, the total failure repair time of the whole sys-
tem is 200 min with a probability of 90%. Furthermore, the con-
ln t 3:7717
M 1 ðt Þ ¼ 1 £ ð15Þ veyor 1 has less maintainability than the other sub-systems, and
1:1475
the maintainability of conveyors 2 and 3 are approximately equal.
To improve the reliability of the system, preventive mainte-
ln t 3:1612
M 2 ðt Þ ¼ 1 £ ð16Þ nance is required for each sub-system. Since reliability is different
0:87222
for various sub-systems, maintenance intervals also will vary for
different sub-systems. The maintenance intervals for each sub-
ln t 3:2638
M 3 ðt Þ ¼ 1 £ ð17Þ system of the main hauling system were calculated and deter-
0:79181
mined at the reliability levels of 90, 80, 70, 60 and 50 percent
Using this distribution function, the maintainability of all three through the relevant reliability functions. Table 9 represents relia-
sub-systems was determined by Eqs. (15)–(17) and their associ- bility maintenance intervals for three sub-systems and the total
ated graphs are plotted in Fig. 13. According to this figure, the fail- system.
Table 8
Availability of sub-systems.
Table 9 Acknowledgment
Maintenance intervals based on reliability.
Reliability level Conveyor Conveyor Conveyor General The authors would like to extend their appreciation to the Tab-
(%) 1 2 3 system riz Urban Railway Organization for providing the required facilities
90 26.66 h 1h 4h 1h and data that made this research possible.
80 43 h 3h 6h 2h
70 54 h 4h 9h 3h
References
60 68 h 5h 13 h 4h
50 82 h 7h 18 h 5h
[1] B.K. Low, H.H. Einstein, Reliability analysis of roof wedges and rockbolt forces
in tunnels, Tunn. Undergr. Space Technol. 38 (2013) 1–10.
[2] E. Topal, S. Ramazan, A new MIP model for mine equipment scheduling by
minimizing maintenance cost, Eur. J. Oper. Res. 207 (2) (2010) 1065–1071.
To design a preventive maintenance program, 80% reliability [3] S. Moosazadeh, H. Aghababaie, S.H. Hoseinie, B. Ghodrati, Simulation of tunnel
level was chosen. Based on the selected reliability level, conveyors boring machine utilization: a case study, J. Min. Environ. 9 (1) (2018) 53–60.
[4] M. Koopialipoor, S.S. Nikouei, A. Marto, A. Fahimifar, D.J. Armaghani, E.T.
1, 2 and 3 require 2580, 180 and 360 min maintenance intervals, Mohamad, Predicting tunnel boring machine performance through a new
respectively. This means that the conveyor sub-system 2 needs model based on the group method of data handling, Bull. Eng. Geol. Environ.
to be maintained every 180 min, and the conveyor sub-system 1 (2018) 1–15.
[5] J. Barabady, U. Kumar, Reliability analysis of mining equipment: a case study of
needs maintenance every 2580 min.
a crushing plant at Jajarm Bauxite Mine in Iran, Reliab. Eng. Syst. Saf. 93 (4)
In this project 4 ring installation is carried out per shift and the (2008) 647–653.
average time for ring installation is 3 h. Therefore, for conveyor 2 it [6] H. Amini Khoshalan, S.R. Torabi, S.H. Hoseinie, B. Ghodrati, RAM analysis of
is essential to perform an inspection after each working shift, espe- earth pressure balance tunnel boring machines: a case study, Int. J. Min. Geo-
Eng. 49 (2) (2015) 173–185.
cially rollers and the belt. For the conveyor 3, inspections at the [7] H.A. Khoshalan, S. Torabi, D. Maleki, RAM analysis of hydraulic system of earth
end of each two shifts would be the best preventive action where pressure balance tunnel boring machine, Indian J. Sci. Technol. 8 (28) (2015).
the rollers should be considered as critical components in this sub- [8] M.J. Rahimdel, M. Ataei, R. Khalokakaei, S.H. Hoseinie, Reliability-based
maintenance scheduling of hydraulic system of rotary drilling machines, Int.
system. In the case of the conveyor 1, within 43 h of maintenance J. Min. Sci. Technol. 23 (5) (2013) 771–775.
interval at 80% reliability level, it is recommended that an inspec- [9] J. Barabady, X. Gao, T. Markeset, Criticality analysis from maintainability point
tion be carried out in each 14 shifts and rollers and the belt’s func- of view, in: 2017 IEEE International Conference on Industrial Engineering and
Engineering Management, IEEE, 2017, pp. 2140–2143.
tion should be considered. [10] B. Samanta, B. Sarkar, S. Mukherjee, Reliability analysis of shovel machines
used in an open cast coal mine, Min. Resour. Eng. 10 (02) (2001) 219–231.
[11] N. Vayenas, X. Wu, Maintenance and reliability analysis of a fleet of load-haul-
6. Conclusion
dump vehicles in an underground hard rock mine, Int. J. Min. Reclam. Environ.
23 (3) (2009) 227–238.
This paper describes the reliability, availability and maintain- [12] A. Gustafson, H. Schunnesson, U. Kumar, Reliability analysis and comparison
ability evaluation of materials hauling system in mechanized tun- between automatic and manual load haul dump machines, Qual. Reliab. Eng.
Int. 31 (3) (2015) 523–531.
neling. In this regard, failure data from Tabriz Metro Line 2 in Iran [13] S.H. Hoseinie, M. Ataei, R. Khalokakaie, B. Ghodrati, U. Kumar, Reliability
were used. The material hauling system was divided into three analysis of drum shearer machine at mechanized longwall mines, J. Qual.
sub-systems 1, 2 and 3 and the results of each sub-system were Maint. Eng. 18 (1) (2012) 98–119.
[14] D.B. Kishorilal, A.K. Mukhopadhyay, Reliability analysis of overhauled engines
obtained along with the main system. of dumpers, Appl. Mech. Mater. 592 (2014) 756–1760.
The reliability of each sub-system was defined by the best prob- [15] M. Modarres, M. Kaminskiy, V. Krivtsov, Reliability Engineering and Risk
ability distribution function. According to the results, the sub- Analysis: A Practical Guide, CRC Press, 1999.
[16] B.S. Dhillon, Mining Equipment Reliability, Maintainability, and Safety,
system 1 and 2 showed the highest and lowest level of reliability, Springer Science & Business Media, 2008.
respectively. Moreover, availability results indicated that all sub- [17] B.S. Dhillon, Maintainability, Maintenance, and Reliability for Engineers, CRC
systems are available more than 89 percent of the time. Besides, Press, 2006.
[18] M. Catelani, L. Ciani, M. Venzi, Sensitivity analysis with MC simulation for the
the results of the maintainability illustrated that the sub-system
failure rate evaluation and reliability assessment, Measurement 07 (2015) 003.
1 has less maintainability than the other sub-systems, and the [19] S. Kumar, Performance Analysis and Optimization of Some Operating Systems
sub-systems 2 and 3 have an equivalent maintainability. The main- of a Fertilizer Plant, National Institute of Technology Kurukshetra, 2010.
[20] R.A. Hall, L.K. Daneshmend, Reliability modelling of surface mining
tenance time of the conveyor 1 is about 180 and this time for the
equipment: data gathering and analysis methodologies, Int. J. Surf. Min.
conveyors 2 and 3 is 80 min. Reclam. Environ. 17 (3) (2003) 139–155.
In reliability level of 80%, the maintenance interval estimated [21] M. Mohammadi, P. Rai, S. Gupta, Improving productivity of dragline through
for conveyor 1, conveyors 2 and 3 are 43, 6 and 6 h, respectively. enhancement of reliability, inherent availability and maintainability, Acta
Montanistica Slov. 21 (1) (2016) 1–8.
The maintenance includes inspection, repair, servicing, or replace- [22] J.I. Ansell, M.J. Phillips, Practical Methods for Reliability Data Analysis, Oxford
ment in different subsystems. Based on the obtained time intervals University Press, 1994.
and the average excavation time per shift, it can be concluded that [23] Standard, M., 1981. MIL-HDBK-189. Reliability growth management.
[24] H. Ascher, H. Feingold, Repairable System Reliability, Dekker, New York, 1984.
for the conveyor 2, maintenance operations should be performed [25] P. Kumar, Reliability, Availability and Maintainability Analysis of the Main
after each shift. Furthermore, for conveyor 3, the best time interval Conveyor System in Underground Coal Mine: A Case Study of Churcha (RO)
for maintenance inspections is after each two shifts. Similarly, for a Mine, National Institute of Technology Rourkela, 2016.
[26] F. Simon, B. Javad, B. Abbas, Availability analysis of the main conveyor in the
conveyor 1, the best time interval for carrying out maintenance is Svea Coal Mine in Norway, Int. J. Min. Sci. Technol. 24 (5) (2014) 587–591.
every 14 shifts. Failures frequency showed that the major failures [27] U. Kumar, Reliability Analysis of Load-haul-dump Machines, Luleå tekniska
in all sub-systems are related to rollers and belt. In the case of universitet, 1990.
[28] E. Zio, An Introduction to the Basics of Reliability and Risk Analysis, World
sub-system 2, most failures happened in the arched sections of
Scientific, 2007.
the tunnel, because of a lot of pressure on the system. Therefore,
this extra pressure can be reduced by adding the chassis in these
sections and controlling the propulsion.