Operation and maintenance performance

of rail infrastructure:

Model and methods

Christer Stenström

Operation and Maintenance Engineering

Luleå University of Technology, Sweden

2016 – 06 – 01

Division of Operation, Maintenance and Acoustics

Supervisors: Prof. Aditya Parida

Prof. Uday Kumar

Opponent: Prof. Rommert Dekker, Erasmus University, The Netherlands

Committee members:

Prof. Ingrid B. Utne, NTNU, Norway

Prof. João C.E.J. Pombo, Instituto Superior Técnico, Portugal

Prof. Bjarne Bergquist, LTU, Sweden

Salvetti PhD Award Winner

Compressed version of disputation presentation

Luleå

Athens

The Arctic Circle

Luleå University of Technology

Division of Operation, Maintenance and Acoustics

3

• From 2010-2014

• Funded by the Swedish

Transport Administration

(Trafikverket)

• On maintenance

performance measurement

• Includes five journal papers,

three in Web of Science

(compilation thesis)

• Jointly with AUTOMAIN and

BGLC EU-projects

Table of contents

Part I

1. Introduction

2. Basics concepts and definitions

3. Literature review

4. Research methodology

5. Summary of the appended papers

6. Results and discussion

7. Extension of the research

8. Conclusions and contributions

4

Table of contents

Part II

Paper A

Paper B

Paper C

Paper D

Paper E

5

Scope

• Operation and maintenance performance

• In terms of technical performance

6

Availability: The proportion of time a system is

functioning

Reliability: How often items fail

Maintainability: How hard it is to do

maintenance

Supportability: How fast an organisation

can react on failures

A key for capacity

and credibility!

Failure

7

Line 21: Iron Ore Line

Line 7: Upper Norrland Line

Line 5: East Coast Line

Line 1: Western Main LineLine 2: Southern Main LineLine 3: West Coast LineLine 4: Coast-to-Coast Line

Stockholm

Gothenburg

Malmö

Umeå

Luleå

Riksgränsen

65 24 816

352 679 52 854 28 704

Sections Failures Inspections Potential failures (= inspection remarks) Rectified potential failures

1 year data (2013-14):

7 lines 65 sections 2010-2014

Data collection

Programming in Matlab

Table of contents

Part II

Paper A:

Stenström, C., Parida, A., Galar, D. and Kumar, U. (2013)

Link and effect model for performance improvement of railway

infrastructure, Journal of Rail and Rapid Transit

Paper B

Paper C

Paper D

Paper E

8

MPM framework

Link and effect model

• For maintenance policy and maintenance strategy

• Focus on: the components of strategic planning, ease of use, implementation issues and continuous improvements (yearly cycle)

9

IM = Infrastructure manager KRA = Key result area CSF = Critical success factor

Act Plan 1. Breakdown of objectives

2. Updating the measurement system

3. Analysis of data for indicators and

performance

4. Ranking, simulation and implementation

Do Study PI PI

KPI KPI

KPI PI

Mission/Vision

Goals

Objectives

Strategies

KRAs

CSFs PIs PI

1.

2. 3.

4.

EU White Paper on transportation

Transport strategy of Sweden

Indicators: • Availability • Reliability • Maintainability • Supportability

Quality of service

Risk-matrices

0 1 2 3 4 5

x 10 4

0

100

200

300

400

500

Wor

k or

ders

[No.

]

Delay [min]

S&C Fault disappeared

Track Pos. Sys.

Signalling ctrl. Signalling Overhead system

Train ctrl

(MIL-STD-1629A and ISO 31010)

Case study

Probability-consequence matrix

10

Reliability problem

Maintainability problem

Reliability and main-tainability problem

At three levels: System, subsystem and component.

Table of contents

Part II

Paper A

Paper B:

Stenström, C., Parida, A. and Galar, D. (2012) Performance

indicators of railway infrastructure, International Journal of

Railway Technology

Paper C

Paper D

Paper E

11

KPIs

KPIs of rail infrastructure

• ~120 indicators mapped • Comparison with EN 15341: Maintenance key

performance indicators

12

Railway Infrastructure PIs

Infrastructure Condition Indicators

Managerial Indicators

Economic

Organisational

HSE

Technical Substructure

Electrification

Superstructure Rail Yards

Signalling ICT

Structure according to EN 15341: Maintenance KPIs

Structure according to Trafikverket

Table of contents

Part II

Paper A

Paper B

Paper C:

Stenström, C., Parida, A. and Kumar, U. (2016) Measuring and

monitoring operational availability of rail infrastructure, To

appear in: Journal of Rail and Rapid Transit

Paper D

Paper E

13

Availability performance

TTR [minutes] 0 1000 2000 3000 4000 5000 6000 7000

0 1000 2000 3000 4000 5000 6000 70000

1

2

3

x 10-3

Den

sity

TTR = logistic time + repair time

Median = 171 min Log-normal mean = 410 min

Availability of rail infrastructure

14

955 train-delaying failures 2010-14

(TTR = Time to restoration)

Availability of rail infrastructure

15

Median:

Log-normal mean:

Registered down time (TTR):

0

0.5

1

AS

A

64 %

79 %

92 %

0

0.5

1

AS

A

Days (2013.01.01 – 2013.01.30)

0

0.5

1

5 10 15 20 25 30

AS

A

955 train-delaying failures 2010-14

(TTR = Time to restoration)

16

R² = 0,734

0

40000

80000

120000

0 500 1000 1500

Trai

n de

lay

[min

]

Train-delaying failures [No.]

R² = 0,944

0

40000

80000

120000

0,7 0,8 0,9 1

Trai

n de

lay

[min

]

Service affecting availability (ASA)

Reliability Maintainability Maintenance supportability

Track

S&Cs

Track

S&Cs

(S&Cs = Switches and crossings)

Table of contents

Part II

Paper A

Paper B

Paper C

Paper D:

Stenström, C., Parida, A., Lundberg, J. and Kumar, U.,

Development of an integrity index for monitoring rail

infrastructure, International Journal of Rail Transportation

Paper E

17

Composite indicator

Integrity index – A composite indicator

• Provides the big picture, easier to interpret than trying to find a trend in many separate indicators

• Used by World Bank, European Commission, UNESCO and OECD

18

Time to restoration = Logistic time + Repair time TTR = LT + RT

Reliability (R)

Maintainability (M)

Supportability (S)Preventive maintenance (PM)

Failures

Railway availability = f(failures, TTR) = f(R, PM, M, S)

Time to restoration (TTR)

Train delay

Train punctuality = f(failures, TTR) = f(R, PM, M, S)

Train regularity = f(failures, TTR) = f(R, PM, M, S)

Railway infrastructure

= f(R, PM)

= f(M, S)

= f(failures, TTR) = f(R, PM, M, S)

Trains

Theoretical framework:

19

Procedure

Min-Max, Z-score and Rank

Aggregation:

Weighting:

Normalisation of data range:

Additive and geometric

Equal weighting, Correlation weighting, AHP and Reduced CI

S&C failures (per S&Cs)

Linear assets failures [km-1]

S&C delay (per S&Cs)

Linear assets delay [km-1]

Logistic time (LT)

Repair time (RT)

Normalisation to switches & crossings and track length:

𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 𝐶𝐶𝐶𝐶𝑖𝑖 = 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 �𝑤𝑤𝑞𝑞𝐶𝐶𝑞𝑞𝑖𝑖𝑄𝑄=6

𝑞𝑞=1= 𝑅𝑅𝑅𝑅𝑅𝑅𝑅𝑅 �𝑤𝑤𝑞𝑞

𝑥𝑥𝑞𝑞𝑖𝑖 − 𝑚𝑚𝑚𝑚𝑅𝑅 𝑥𝑥𝑞𝑞𝑚𝑚𝑅𝑅𝑥𝑥 𝑥𝑥𝑞𝑞 − 𝑚𝑚𝑚𝑚𝑅𝑅 𝑥𝑥𝑞𝑞

𝑄𝑄=6

𝑞𝑞=1

0

0,1

0,2

0,3

0,4

0,563

041

265

541

942

041

092

081

714

662

6

116

511

824

422

112

910

118

113

823

111

3 1 4 1 1 1 3 2 7 3 21 1 4 2 21 2 21 21 4 21

RC

I [a.

u.]

Mean RT

Mean LT

Linear assets delay

S&Cs delay

Results

20

Line 21 shows poor performance

Sections: Lines:

Line 1 shows good performance

LT = Logistic time = travel time RT = Repair time a.u. = arbitrary unit

Table of contents

Part II

Paper A

Paper B

Paper C

Paper D

Paper E:

Stenström, C., Parida, A., and Kumar, U., Preventive and

corrective maintenance: Cost comparison and cost-benefit

analysis, Structure and Infrastructure Engineering

21

Cost-benefit analysis

Comparison of corrective and preventive

maintenance (CM and PM) costs

22

€53 / min

€100 / item

€100 / h

2 personnel

𝐶𝐶𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑆𝑖𝑖𝑆𝑆𝑆𝑆 = 1𝑁𝑁 𝐶𝐶𝐶𝐶𝐶𝐶𝑆𝑆&𝐶𝐶𝐶𝐶 + 𝐶𝐶𝑃𝑃𝐶𝐶𝑃𝑃𝑆𝑆&𝐶𝐶𝐶𝐶 + 𝐶𝐶𝑃𝑃𝐶𝐶𝑃𝑃𝑆𝑆&𝐶𝐶𝐶𝐶 + 1

𝑀𝑀 𝐶𝐶𝐶𝐶𝐶𝐶𝑇𝑇𝑇𝑇𝑇𝑇𝑆𝑆𝑇𝑇 + 𝐶𝐶𝑃𝑃𝐶𝐶𝑃𝑃𝑇𝑇𝑇𝑇𝑇𝑇𝑆𝑆𝑇𝑇 + 𝐶𝐶𝑃𝑃𝐶𝐶𝑃𝑃𝑇𝑇𝑇𝑇𝑇𝑇𝑆𝑆𝑇𝑇

CM

No. of S&Cs

PM: Inspections

PM: Remarks

(S&Cs = Switches and crossings)

Track length

= � 𝑅𝑅𝑃𝑃,𝑖𝑖𝐶𝐶𝑃𝑃 2𝑡𝑡𝐿𝐿𝑇𝑇,𝑖𝑖 + 𝑡𝑡𝑃𝑃𝑇𝑇,𝑖𝑖 + 𝐶𝐶𝐶𝐶,𝑖𝑖 + 𝑡𝑡𝐷𝐷𝑇𝑇,𝑖𝑖𝐶𝐶𝐷𝐷𝑇𝑇𝑆𝑆

𝑖𝑖=1

65 railway sections

0%

20%

40%

60%

80%

100%

Cos

t pro

port

ion

1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65

Railway sectionFailures cost Inspections cost Remarks cost

0250005000075000

100000

Nor

mal

ised

co

st

23

4x cost 4.5x track failures Difference: 3x tonnage

2x more trains

Higher inspection class

Mean (µ) SD (σ)

= 78 % CM = 9.8 %

R2 = 0,38 Disregarding city tracks, R2 = 0,31

61 % CM when disregarding train-delay cost

1 5 9 13 17 21 25 29 33 37 41 45 49 53 57 61 65

Railway sectionFailures cost Inspections cost Remarks cost

0250005000075000

100000

Nor

mal

ised

co

st

24

City tracks 58: Hässleholm 59: Vännäs 60: Helsingborgs 61: Uppsala 62: Gävle 63: Boden 64: Luleå 65: Kiruna

Mean track length = 8 km

Mean track length = 83 km

25

Cost-benefit analysis

For the 65 railway sections together

Mean saved cost of avoiding failure = € 1 806

Mean cost per inspection = € 11,2

Mean cost of repairing potential failures = € 273

Probability of detection

28 704352 678 = 0,08

Potential to functional failure likelihood = 0,75

= 𝟑𝟑,𝟑𝟑 𝐵𝐵/𝐶𝐶 = 𝐵𝐵𝑃𝑃𝐶𝐶𝐶𝐶𝑃𝑃𝐶𝐶

= 𝛼𝛼𝛼𝛼�̅�𝐶𝐹𝐹�̅�𝐶𝑃𝑃 + 𝛼𝛼�̅�𝐶𝑃𝑃

:

Inspections

Repaired potential failures

(Potential failure = Inspection remark)

(€53 / min )

Table of contents

Part II

Paper A

Paper B

Paper C

Paper D

Paper E

26

1. Breakdown of objectives

2. Updating the measurement system

and aligning of indicators to objectives

3. Analysis of datafor indicators and

performance

4. Identification of improvements through

indicators, ranking and implementation

PIPI

KPI KPI

KPI PI

Mission/Vision

GoalsObjectives

StrategiesKRAs

CSFsPIs PI

1.

2. 3.

4.

Rail infrastructure indicators

Condition indicatorsManagerial indicators

Economic

Organisational

HSE

Technical

Availability indicator

Cost of PM/CM

Composite indicator

Risk matrix

Maintenance between trains

Spatial and temporal representation

Comparison of performance

Paper A

Paper BPaper A

Paper C

Extension

Paper E

Paper D

Related paper

27

Thank you!

Division of Operation, Maintenance and Acoustics