Division of Maintenanceand
1
Reliability Analysis of Switches and Crossings – A Case Study in
Swedish Railway
Behzad Ghodrati, Alireza Ahmadi, Diego Galar
Division of Operation and Maintenance Engineering Luleå University of Technology, Sweden
Division of Maintenanceand
2
Introduction
Railway complexity:
Mix of components with different age
Working together
Increase traffic volume
Higher utilization of capacity
Minimize maintenance
time
Minimize unplanned interruption
Maintenance be performed near capacity limits
Time between asset renewals be long enough
Division of Maintenanceand
3
Introduction
The key goal is to achieve availability target cost effectively.
Availability
Reliability Maintainability Supportability
To conduct reliability analysis:
Detail failure and maintenance recorded data
Detail maintenance action done
Mission profile: duty cycle and environmental
characteristics
Division of Maintenanceand
4
Reliability: Ability of an item to perform a required function under given conditions for a given time interval.
RAMS
(reliability, availability, maintainability and safety)
e t
t
R ( )
Availability: Ability of an item to be in a state to perform a required function under given conditions at a given instant of time or during a given time
interval, assuming that the required external resources are provided.
time Total
repair of
Times time
Total
A
RAMS
Division of Maintenanceand
5
Switches
A railroad switch, turnout or set of points is a mechanical installation enabling railway trains to be guided from one track to another at a railway junction.
Name of switche in Swedish railway system: A-B-C-D (e.g.
EV – SJ50 – 11 – 1:9),
A: type of switch (single, double)Check rail B: type of railpanel
C: radius or length of switch blade D: type of angle
Division of Maintenanceand
6
Ballast
Check rail
Cross over panel
Crossing
Fasteners
Heating system
Locking device
Rail
Rail joint (mostly protected rail joint)
Sleeper (bearer)
Snow protection
Switch blade
Switch blade position detector
Switch device (motor, gearbox, coupling, bars, etc.)
Switch and Crossing Elements
Division of Maintenanceand
7
BODEN
ÅNGE
GÄVLE
NORRKÖPING STOCKHOLM
MALMÖ GÖTEBORG
HALLSBERG
Sweden railway network
Division of Maintenanceand
8
Data collection and evolution
Number of registered failures Jan. 2005 – Dec. 2009
Age and location of turnouts
Switches with
numbers inferior to 50 was eliminated
Raw Data 43528 failures
Raw data without unnecessary types of turnouts 42221 failures
Installation date known
Installation date unknown
Turnouts known
Turnouts unkown
Turnouts known
Turnouts unkwown
- Changed between 05/09:
-in BESSY (1452 failures) -not in BESSY - Not changed -installation date ”0" (2004 failures) -the rest (25006 failures)
- In BESSY (30 failures) - Not in BESSY
- Changed between 05/09:
-in BESSY (31 failures) -not in BESSY - Not changed -installation date ”0" (176 failures) -the rest (977 failures)
- Unkwown (10477 failures)
29676 failures
Division of Maintenanceand
9 24%
68% 8%
Turnouts unkown
Data not found in the different files
Data available for study
29676 failures
10477 failures
3375 failures
Final available data
Take into account the 10 types of turnouts
generating most failures and 60 tracks of interest
16627 failures
Division of Maintenanceand
10
Tracks with more failures with at least 10 individuals asset names and at least 2 types of turnouts
Studied tracks and switches
Track
number Type of track
124 Freight track
410 Commuter trians and some freight 414 Mixed passenger and freight 420 Mixed passenger and freight 512 Mixed passenger and freight 611 Mixed passenger and freight 811 Mixed passenger and freight 813 Mixed passenger and freight 912 Mixed passenger and freight
9 (out of 60) focused tracks
Division of Maintenanceand
11 641
715 867
1214 1745
2890 2997
3224 4291
5301
EV-SJ50-12-1:12 EV-SJ50-12-1:13 EV-UIC60-760-1:14
EV-UIC60-1200-…
EV-UIC60-1200-1:18,5 EV-UIC60-300-1:9
DKV-SJ50-…
EV-SJ50-12-1:15 EV-UIC60-760-1:15 EV-SJ50-11-1:9
0 2000 4000 6000
Number of failures
10 types of turnouts generating more failures
EV-SJ50-11-1:9 EV-SJ50-12-1:15
EV-UIC60-1200-1:18,5
EV-UIC60-1200-1:18,5 BL33 EV-UIC60-300-1:9
EV-UIC60-760-1:14 EV-UIC60-760-1:15 EV-SJ50-11 EV-SJ50-12 EV-UIC60-300 EV-UIC60-760 EV-UIC60-1200
Division of Maintenanceand
12
Dividing into 2 types of tracks
nhsp
main track ahsp
diverging track Dividing into 2 seasons
COLD
from November to March (5 months)
HOT
from April to October (7 months)
55%
45%
Proportion of failures by season
Cold Hot
Data classification
Division of Maintenanceand
13 3
7 10 13 31 47
62 79 98 109
120
710
1616
2057
2520
0 500 1000 1500 2000 2500 3000 Cross over panel
Check rail Ballast Sleeper (bearer) Rail Snow protection Locking device Crossing Fasteners Rail joint (mostly insulated rail joint) Heating system Switch blade Switch device (motor, gearbox,…
(blank) Switch blade position detector
Hot
Subsystems affected by
failures – Hot period
Division of Maintenanceand
14 4
7 8 10
36 57 66 79 80
105
624
1194 1474
2521 2765
0 500 1000 1500 2000 2500 3000 Cross over panel
Sleeper (bearer) Ballast Check rail Rail Locking device Fasteners Crossing Rail joint (mostly insulated rail joint) Snow protection Switch blade Heating system Switch device (motor, gearbox,…
Switch blade position detector (blank)
Cold
Subsystems affected by
failures – Cold period
Division of Maintenanceand
15
1194 624
1474
2765 2521
120
710
1616
2057
2520
0 1000 2000 3000
Heating system Switch blade Switch device (motor, gearbox, coupling, bars, ...)
(blank) Switch blade position
detector
HOT COLD
Comparison of subsystems with more failures during the two
seasons
Division of Maintenanceand
16
RDAT (Reliability Data Analysis Tool) software was developed by Alstom and the University of Bordeaux (France), and deal with highly censored field data which wasn’t taken into account properly with the already existing programs.
Data analysis tool
RDAT was used to estimate the reliability functions and failure rates from field data
Four failure models have been implemented in RDAT:
exponential, Weibull, normal, and lognormal distributions.
To select the best model, a goodness-of-fit test is applied.
The maintenance quality is considered by a parameter denoted Rho:
ρ = 1 means that the maintenance quality is AGAN (the maintenance operation is perfect).
ρ = 0 means that the maintenance quality is ABAO (the mission can continue but leaves the item with a reliability corresponding to the age accumulated so far).
Division of Maintenanceand
17
Intrinsic Reliability Analysis
Is Exponential best Estimator?
Kijima
Rho = 0 0 < Rho < 1 Rho = 1
Work on ABAO
Work on First Failure
Work on AGAN
Work on First Failure
Yes No
Work on First Failure Maintenance
effect analysis POSSIBLE
Maintenance effect analysis NOT POSSIBLE
RDAT software methodology
Division of Maintenanceand
18
Data analysis – RDAT software output
Trafikverket (Swedish Railway Administration) maintenance experts consulting:
70% of cases ρ
=1
30% of cases, ρ
= 0,5-1
AGAN maintenance
ABAO maintenance
ABAO model was considered
Division of Maintenanceand
19
Instantaneous failure rate
λ failure rate
β shape parameter
Instantaneous Mean Time Between Failures
T
n
n
i
Ti
T n
n
1
ln ln
Data analysis – RDAT software output
Division of Maintenanceand
20
β < 1 → MTBF ↗
Maybe the maintenance has improved in these 5 years (Case of infant mortality: many problems at the beginning)
The organisation learned how to deal with failures during 05/09
Other possible explanation:
For SJ50-11 switch point detectors taken out (less failures)
Change of switch point detectors on the other types of turnouts (from mechanical to electrical) > reduces
number of failures in Hot and Cold
RDAT implementation and results
Growth factor Beta as a function of types of turnout and season and type of track
Division of Maintenanceand
21
β > 1 → MTBF ↘
”Old equipment fails more” > Maintenance is not compensating the age of the turnout
RDAT implementation and results
Growth factor Beta as a function of types of turnout and season and type of track
Division of Maintenanceand
22
Comparison between hot/cold
There are much more β < 1 during COLD season, better
maintenance? More effective maintenance during winter time?
There are much more β > 1 during HOT season, worst maintenance?
Is there any link with the number of failures avery year?
RDAT implementation and results
Division of Maintenanceand
23
There is no relationship
between the number of failures every year and the improved or not of the maintenance for
these years.
Comparison between hot/cold
Division of Maintenanceand
24
Values of λ and β for different types of turnouts for the 9 tracks
RDAT results
Example for tracks 124, 410 and 912 for main track and SJ50-11
Division of Maintenanceand
25
Example for tracks 124, 410 and 912 for main track and SJ50-11
β ≈1 β
≥1β ≤1
RDAT results
Division of Maintenanceand
26
0 0,00005 0,0001 0,00015 0,0002 0,00025 0,0003 0,00035 0,0004
3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60
Failure rate (λ)
Months (from January 05 to December 09)
Instantaneous failure rate (SJ50-11 and nhsp- cold)
124 410 912
RDAT results
Division of Maintenanceand
27
0 0,00005 0,0001 0,00015 0,0002 0,00025 0,0003 0,00035 0,0004
3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60
Failure rate (λ)
Months (from January 05 to December 09)
Instantaneous failure rate (SJ50-11 and nhsp-hot)
124 410 912
RDAT results
Division of Maintenanceand
28
Turnout 1
Turnout 2
Turnout 3
Turnout 4
Turnouts are in serie in a track
Availability
Division of Maintenanceand
29
Conclusion
The RAMS analysis confirms the more failure in Cold season than in Hot season
For tracks 124, 410 and 912
Failure rate decreasing during Cold season
Failure rate almost constant during Hot season
Track 512, which has the lowest availability, needs to be focused for improvement
The RDAT software is not taking into account this parameter. However, it is possible to do a covariate analysis including this factor.
On the most important failure contributors, which are the switch blade position detectors, switch devices, heating system in the cold season, and switch blades
Division of Maintenanceand
30