L ICE N T IAT E T H E S I S
Department of Civil, Environmental and Natural Resources Engineering
Division of Operation, Maintenance and Acoustics
Luleå University of Technology 2016
Per Norrbin Railway Infrastructure Robustness
ISSN 1402-1757
ISBN 978-91-7583-585-3 (print)
ISBN 978-91-7583-586-0 (pdf)
Railway Infrastructure Robustness
Attributes, evaluation, assurance and improvement
Per Norrbin
Operation and Maintenance Engineering
Railway Infrastructure Robustness
Attributes, evaluation, assurance and improvement
Per Norrbin
Operation and Maintenance Engineering
Luleå University of Technology
Printed by Luleå University of Technology, Graphic Production 2016
ISSN 1402-1757
ISBN 978-91-7583-585-3 (print)
ISBN 978-91-7583-586-0 (pdf)
Luleå 2016
www.ltu.se
ABSTRACT
According to the European program Horizon 2020, multifaceted challenges in transport
infrastructure include 1) making infrastructure more resilient to keep pace with the
increasing mobility needs; 2) reducing the impact of infrastructure on the environment;
and 3) dealing with declining resources to maintain and upgrade transport infrastructure.
New design and maintenance approaches must be developed to handle these issues, as
current methods are inadequate.
In maintenance practices of railway infrastructure, most attention has been placed on
RAMS (Reliability, Availability, Maintainability and Safety) study to meet asset safety and
availability requirements. In Sweden, a popular development of the concept is RAM4S
which incorporates supportability, sustainability, and security. However, the reality is
more complex than these concepts suggest; various natural or operational uncertainties
can cause “unfavourable” conditions requiring a quite different approach.
Resilience studies can improve the ability of an infrastructure to withstand disturbances
caused by uncertainties. To this point, most studies have considered extreme events,
including natural events, like earthquakes or floods, and man-made events, like deliberate
attacks on infrastructures. However, both naturally caused and operationally caused
unfavourable conditions which do not belong to extreme events need to be studied as well.
This represents a significant gap in the research.
Robustness as a part of resilience has attracted much attention in recent years as it may be
able to sufficiently consider those unfavourable conditions. In railway systems, however,
robustness studies have mainly been aimed at timetable management to handle delays
(including secondary delays) within the system. Yet according to statistics on total delays
for the year 2015, from the follow-up system of the Swedish Transport administration
(Trafikverket), LUPP, more than 30% of the root causes of delays are due to non-robust
infrastructures.
To fill the above research gaps and support decision making of a railway infrastructure
manager, the study described in the thesis conducts a holistic examination of railway
infrastructure robustness considering its attributes, evaluation, assurance and
improvement by investigating, exploring and developing new definitions and approaches.
First, it develops a new road map for railway infrastructure robustness, including a novel
definition and a framework. Its overview of robustness related topics results in a unique
definition that identifies attributes of railway infrastructure robustness and clarifies the
I
research boundaries between robustness and reliability, resilience, and risk. The study’s
ground-breaking framework, named house of robustness management (HORM), is based
on continuous improvement to support infrastructure robustness management in railway;
HORM consists of robustness management goals and guidance, a continuous improvement
process, and support systems through which assurance and improvement can be
supported. Second, in addition to the qualitative evaluation of railway infrastructure
robustness enabled by the road map, the study proposes a new quantitative evaluation
approach by considering a dataset of “specified disturbance”, a corresponding “acceptable
functionality”, and a dataset of weights of risk preference. Third, to apply railway
infrastructure robustness to the development of sustainability, it provides a case study on
energy efficiency optimisation in Sweden.
This thesis consists of two parts. The first gives an introductory summary of the subject
and research, followed by a discussion of the appended papers, suggested extension of the
research and conclusions. The second part consists of three appended papers. The first
paper concerns the new road map including the definition and a framework. The second
paper proposes a new quantitative evaluation approach for railway infrastructure
robustness. The third paper is a case study on railway switches & crossings which will be
further pursued under the topic of “green robustness”. The three subsequent papers
develop the dependability improvement of railway infrastructure considering information
logistics and risks, as well as maintenance cost.
Keywords: maintenance, resilience, robustness, railway infrastructure, continuous
improvement
II
LIST OF APPENDED PAPERS
Paper A
Norrbin P., Lin J. & Parida A., (2016) Infrastructure Robustness for Railway systems.
Accepted for publication in: International Journal of Performability Engineering.
Paper B
Norrbin P., Lin J. & Parida A., (2016) A novel approach for measuring Railway
Infrastructure Robustness. Submitted for publication.
Paper C
Norrbin P., Lin J. & Parida A., (2016) Energy efficiency optimization for railway
switches & crossings: a case study in Sweden. Accepted for the proceedings of:
The World Congress of Railway Research, 30 of May-2 of June 2016, Milano, Italy.
III
IV
LIST OF RELATED PAPERS
Journal publications
Söderholm, P. & Norrbin, P. (2014) Information logistics for continuous
dependability improvement. Published in: Journal of Quality in Maintenance
Engineering. 20, 3, p. 249-261 13 p.
Söderholm, P. & Norrbin, P. (2013) Risk-based dependability approach to
maintenance performance measurement. Published in: Journal of Quality in
Maintenance Engineering. 19, 3, p. 316 - 329 14 p.
V
VI
ACKNOWLEDGEMENTS
The research presented in this thesis has been carried out in the subject area of Operation
and Maintenance Engineering at Luleå University of Technology (LTU), Sweden. I would
like to thank Trafikverket (the Swedish Transport Administration) for providing support,
funding and data. Specifically to Vivvianne Karlsson, Anders Gustafsson and Stefan Jonsson
who have been involved in initiating and supporting this project. Support and valuable
discussions along the course of this work was also given by, Lars Wikberg, Arne Nissen and
Per Hurtig. I would like to express my gratitude to my supervisors Prof. Aditya Parida,
Associate Prof. Janet Lin, Adjunct Prof. Per-Olof Larsson-Kråik and Prof. Uday Kumar for
their valuable support, discussions and ideas. I would also like to thank, Dr. Christer
Stenström, Levis Zhang, Dr. Stephen Famurewa, Rajib Gosh, Dr. Iman Arastehkhouy and Dr.
Amparo Morant for contributions to my work and good memories, together with all my
colleagues at the Operations and Maintenance department at Luleå University of
Technology. Thereafter I would also like to thank all my dear colleagues at Sweco Rail
together with my managers Helena Sjaunja and Björn Svanberg who have made it possible
to pursue this work as an industrial PhD candidate.
A special thanks goes to Adjunct Prof. Peter Söderholm and Adjunct Prof. Per-Olof LarssonKråik for ensnaring me into the dark woods of research when being perfect role models
during our time as colleagues at Trafikverket.
Finally my deep gratitude goes to my family and especially to my mother and father for
their eternal trust and support as well as my mother and father in-law who have been
invaluable support during this period. My everlasting love and appreciation goes to my wife
Veronica and our wonderful sons Mio, Colin and Levis, Thank you!
Per Norrbin
Luleå, Sweden
May 2016
VII
VIII
CONTENT
Part I
CHAPTER 1. Introduction .................................................................................................................................... 1
1.1 Background ................................................................................................................................................... 1
1.1.1 Transport infrastructure ................................................................................................................ 1
1.1.2 Maintenance in railway infrastructure .................................................................................... 1
1.1.3 Infrastructure resilience ................................................................................................................. 2
1.1.4 Robustness in the railway.............................................................................................................. 3
1.2 Problem statement .................................................................................................................................... 3
1.3 Purpose and objectives ............................................................................................................................ 5
1.4 Research questions .................................................................................................................................... 5
1.5 Scope and limitations ............................................................................................................................... 6
1.6 Linkage of research questions to appended papers ................................................................... 6
1.7 Outline of thesis .......................................................................................................................................... 6
CHAPTER 2. Theoretical framework .............................................................................................................. 9
2.1 Robustness overview ................................................................................................................................ 9
2.2 Robust infrastructure related studies ............................................................................................ 11
2.2.1 An overview of infrastructure robustness .......................................................................... 11
2.2.2 Current status of robust infrastructure related studies ................................................ 12
2.3 Infrastructure robustness related topics in railway................................................................ 14
2.3.1 Resilience............................................................................................................................................ 14
2.3.2 Rapidity ............................................................................................................................................... 15
2.3.3 Vulnerability ..................................................................................................................................... 15
2.3.4 Functionality ..................................................................................................................................... 15
2.3.5 RAMS .................................................................................................................................................... 15
2.3.6 Risk ........................................................................................................................................................ 16
2.3.7 Sustainability .................................................................................................................................... 16
2.3.8 System assurance............................................................................................................................ 16
IX
2.4 Robustness measurement ................................................................................................................... 17
2.5 Summary of framework........................................................................................................................ 19
CHAPTER 3. Research methodology ........................................................................................................... 21
3.1 Research design ....................................................................................................................................... 21
3.2 Data collection .......................................................................................................................................... 23
3.2.1 Interviews .......................................................................................................................................... 23
3.2.2 Interviews with European infrastructure managers ...................................................... 23
3.2.3 Interviews with Trafikverket, Sweden.................................................................................. 24
3.2.4 Data sources: operation and maintenance data ............................................................... 25
3.3 Literature review ..................................................................................................................................... 26
3.4 Data Analysis ............................................................................................................................................. 26
CHAPTER 4. Summary of appended papers ............................................................................................ 31
4.1 Paper A ......................................................................................................................................................... 31
4.2 Paper B ......................................................................................................................................................... 32
4.3 Paper C ......................................................................................................................................................... 32
CHAPTER 5. Results and discussion ............................................................................................................ 35
5.1 Results and discussion related to RQ1........................................................................................... 35
5.1.1 Ability to perform ........................................................................................................................... 35
5.1.2 Acceptable functionality .............................................................................................................. 36
5.1.3 Specified disturbances.................................................................................................................. 36
5.1.4 Specified time period .................................................................................................................... 36
5.1.5 Definition clarification .................................................................................................................. 36
5.2 Results and discussion related to RQ2........................................................................................... 38
5.2.1 Roof of HORM ................................................................................................................................... 38
5.2.2 Ceiling of HORM............................................................................................................................... 39
5.2.3 Walls of HORM ................................................................................................................................. 40
5.2.4 Floor of HORM .................................................................................................................................. 42
5.2.5 Discussion of HORM ...................................................................................................................... 42
5.3 Results and discussion related to RQ3........................................................................................... 43
5.3.1 A new quantitative measurement approach ...................................................................... 43
5.3.2 An example of a robustness evaluation ................................................................................ 44
X
5.3.3 Discussion .......................................................................................................................................... 46
5.4 Results and discussion related to RQ4........................................................................................... 48
5.4.1 Green robustness ............................................................................................................................ 48
5.4.2 Measuring green robustness for railway infrastructure .............................................. 48
5.4.3 Discussion .......................................................................................................................................... 49
CHAPTER 6. Conclusions, contributions and future research ......................................................... 51
6.1 Conclusions ................................................................................................................................................ 51
6.2 Contributions ............................................................................................................................................ 52
6.3 Future research ........................................................................................................................................ 52
References ............................................................................................................................................................... 53
Part II
Appended papers
Paper A: Infrastructure Robustness for Railway systems
Paper B: A novel approach for measuring Railway Infrastructure Robustness
Paper C: Energy efficiency optimization for Railway switches & crossings.
XI
XII
Part I
XIII
XIV
CHAPTER 1
Introduction
This chapter gives a short description of the research background, along with its purpose,
research questions, scope and limitations. It also provides an outline of the thesis.
1.1 Background
This section gives an overview of the research area, including: transport infrastructure,
maintenance in railway infrastructure, infrastructure resilience, and robustness in the
railway.
1.1.1 Transport infrastructure
Transport refers to the movement of people, animals and between points. It enables trade,
which is essential for the development of civilisations. Transportation can be divided into
air, water and surface transportation (Whaley, 2004). A substantial difference between the
three is that for surface transportation, there is always a need to build up infrastructure
between the points.
A research field in infrastructure can further be divided into infrastructure for vehicles and
operations. Transport infrastructure consists of the fixed installations including roads,
railways, canals and pipelines, and terminals such as airports, railway stations, bus
stations, warehouses, trucking terminals, refuelling depots and even safety systems.
According to the European program Horizon 2020, multifaceted challenges in transport
infrastructure include: 1) making infrastructure more resilient to keep pace with the
increasing mobility needs; 2) reducing the impact of infrastructure on the environment;
and 3) dealing with declining resources to maintain and upgrade transport infrastructure.
New design and maintenance approaches must be developed to handle these issues, as
current methods are inadequate (EC, 2015).
1.1.2 Maintenance in railway infrastructure
Railway transport has increased over the last decade and it is likely to further increase as
passenger and cargo transportation shift from road and air to rail, due to rising energy
1
Introduction
costs, congestion of roads and sky, as well as the demand to reduce emissions (Stenström,
2014).
Safety, reliability, sufficient capacity and availability used to be main requirements for
railway transport (Arasteh Khouy, 2013). Today, taking into account the expected growth
in railway transport demand and the ever-increasing customer expectations on quality of
service, the railway infrastructure assets need to become more productive. To achieve this
goal, these assets need to be managed in a more holistic, intelligent and consistent way
(Shift2Rail, 2015), including more effective maintenance strategies.
In maintenance practices of railway infrastructure, most attention has been placed on
RAMS (Reliability, Availability, Maintainability and Safety) study to meet asset safety and
availability requirements (Barabady, 2007; CEN, 1999). In Sweden, a popular development
of the concept is RAM4S which incorporates supportability, sustainability, and security
(Luleå University of Technology, 2015; Patra, 2009).
The approaches to the maintenance of railway systems have changed significantly over the
last century with a shift from an emphasis on technology to techno-economic
considerations (Ekberg & Paulsson, 2010; Kobbacy & Murthy, 2008). Today, various
maintenance strategies are applied in an integrated way depending on the criticalities and
cost-effectiveness of a particular infrastructure. With new requirements generated by the
increasing complexity of both the infrastructure system and operating conditions, these
strategies continue to be developed (ISO, 2014b) (ISO, 2014b)(ISO, 2014b)(ISO,
2014b)(ISO, 2014).
1.1.3 Infrastructure resilience
Resilience refers to the ability to resist, absorb and adapt to disruptions and return to
normal functionality. A resilient system has 1) reduced probability of failures; 2) reduced
consequences from failures; and 3) reduced time to recovery (Bruneau et al., 2003;
Faturechi & Miller-Hooks, 2014).
Infrastructure resilience studies can improve the ability of an infrastructure to withstand
disturbances caused by various uncertainties (Hollnagel, Woods, & Leveson, 2006; Steen &
Aven, 2011). To this point, most studies have considered extreme events, including natural
events like earthquakes or floods, and man-made events, like deliberate attacks on
infrastructures (Adger, Hughes, Folke, Carpenter, & Rockström, 2005; Bruneau et al., 2003;
Faturechi & Miller-Hooks, 2014).
Some argue studies considering resilience to extreme (natural and man-made) events
should focus on adaptation measures and strategies to ensure seamless transport and user
protection. However, both naturally caused and operationally caused unfavourable
conditions which do not belong to extreme events need to be studied as well. Currently,
2
1.2 Problem statement
research on infrastructure resilience is generally divided into two perspectives: rapidity
and robustness. The former is defined as how quickly the system can return to a high
functional level and the latter expresses the ability to resist specified disturbances so at
least some functionality can be maintained. The two are not mutually exclusive, however,
as rapidity is influenced by robustness.
1.1.4 Robustness in the railway
In general, robustness is used to describe the ability of a system (or a model) to maintain
its function (or performance) despite disturbance (incl. faults or perturbations). All
robustness studies aim to reduce uncertainties (Norrbin, Lin, & Parida, 2016).
Robustness as a part of resilience has attracted much attention in recent years to
sufficiently consider those unfavourable conditions. The approaches to robustness studies
include: 1) statistics driven models from input to output (e.g., control engineering,
statistics, computer science, operations research, decision making theory, and economics);
2) physical characteristics driven design or optimisation of a complex system or an item
(e.g., quality engineering/product development, and biology); and 3) their integration, as
in many situations, both the statistics models and the physical characteristics need to be
considered simultaneously. Considering the complexity of different systems and their
special characteristics, management of robust infrastructure requires knowledge from
different research fields. To date, robustness studies are still being developed in separate
fields, and a more holistic approach would be useful.
In the railway, robustness studies have mainly been aimed at timetable management to
handle delays (including secondary delays) within the system (Andersson, Peterson, &
Törnquist Krasemann, 2013; Fischetti, Salvagnin, & Zanette, 2009).
None of the results from the above fields can be directly used for railway infrastructure
robustness (Norrbin et al., 2016).
1.2 Problem statement
As mentioned above, the new and multifaceted challenges in railway transportation
infrastructure require the development of new maintenance approaches; simply stated, the
current ones are inadequate.
The major drawback in present approaches is that they all assume certain “specified
conditions” (CEN 1999). The reality is more complex; conditions are frequently
“unfavourable,” for example, and may require a different approach. For railway systems,
unfavourable conditions stem from both natural and operational causes. In the case of the
former, for example, in northern Sweden, the infrastructures operate in harsh conditions,
3
Introduction
including snow, ice and extreme temperatures ranging from -40°C to +25°C. Such naturally
occurring unfavourable conditions often lead to time delays and economic losses. As for the
latter, in the same region of Sweden, the regulations governing the axle load have recently
increased from 22.5 tonnes to 30 tonnes because of changing transport needs; an even
higher axle load limit (32.5 tonnes) is being tested. These new operational requirements
demand continuous improvement and upgrades on the original track. Obviously, such
operationally caused conditions are not favourable to the original track system: they will
create more Rolling Contact Fatigue (RCF) problems and increase the degradation of the
track infrastructure (Lewis and Olofsson 2009).
Resilience studies can improve the ability of infrastructure to withstand disturbances
caused by various “uncertainties” [Hollnagel, Woods and Levenson 2006, Steen and Aven
2011, Bruneau et al. 2003). To this point, as mentioned above, research into uncertainty
has concentrated on extreme events. However, naturally caused and operationally caused
unfavourable conditions which are clearly not extreme events need to be studied as well.
This represents a significant gap in the research on uncertainty. In addition, research on
the resilience of infrastructure is generally divided into two perspectives: rapidity and
robustness. But infrastructure robustness must be studied considering unfavourable
conditions.
In railway systems, robustness studies have mainly been aimed at timetable management
to handle delays within the system. Yet statistics from the follow up system of the Swedish
Transport administration (Trafikverket), LUPP, on total delays for the year 2015 show that
more than 30% of the root cause of delays can be traced to non-robust infrastructures. Not
surprisingly, railway infrastructure robustness is now attracting the interest of both
industry and government.
Based on the research gaps, this project’s initial exploratory study identified the following
issues:
 Problem 1: Lack of clarity among robustness, resilience, and reliability, etc. , for
railway infrastructure;
 Problem 2: Lack of a comprehensive approach to develop railway infrastructure
robustness continuously;
 Problem 3: Lack of a quantitative approach to measure railway infrastructure
robustness;
 Problem 4: Lack of integrated study for railway infrastructure robustness and
system assurance & sustainability.
4
1.3 Purpose and objectives
1.3 Purpose and objectives
To deal with these issues and to support the decision making of a railway infrastructure
manager, the main purpose of this research is to conduct a holistic examination of railway
infrastructure robustness from the point of view of its attributes, evaluation, assurance and
improvement by investigating, developing and exploring a new definition and novel
approaches.
More specifically, the research objectives include:
 Objective 1: Identification of attributes of railway infrastructure robustness;
 Objective 2: A new road map to improve railway infrastructure robustness to
achieve system assurance and sustainability;
 Objective 3: A novel method for measuring railway infrastructure robustness.
1.4 Research questions
The following research questions have been formulated to achieve those objectives:
 Research question 1: What are the attributes of railway infrastructure
robustness?
 Research question 2: How can railway infrastructure robustness be enhanced
through a continuous improvement process to achieve system assurance?
 Research question 3: How can railway infrastructure robustness be evaluated
using a quantitative measurement approach?
 Research question 4: How can railway infrastructure robustness be applied to
develop system sustainability?
The connection between the research questions and research objectives is shown in
Table 1.1.
Table 1.1 Connection between RQs and objectives
Research questions (RQs) Objective 1 Objective 2 Objective 3
RQ1
X
RQ2
X
X
X
X
RQ3
X
RQ4
X
5
Introduction
1.5 Scope and limitations
This research considers the railway infrastructure issues only from a technical perspective
(not from a human or organisational perspective) for effective and efficient management
of maintenance. The research includes a literature review, survey, exploratory data
analysis, approach development and case study. Through these, it explores the attributes
of railway infrastructure robustness, promotes new approaches to its evaluation and
continuous improvement, and supports the achievement of system assurance.
Limitations of the work are the following:
 Rolling stock is not part of the study
 Sustainability is only considered from an environmental and energy efficiency
perspective
 Interviews and experience are limited to experts on the European railway network
 Case studies are limited to Sweden and switches and crossings
1.6 Linkage of research questions to appended papers
The linkage of the research questions (RQs) and the appended papers are presented in
Table 1.2. RQ1 is answered in Paper A and then extended in Paper B. RQ2 is explored in
Paper A and Paper B. RQ3 is addressed in Paper B. RQ4 is partly discussed in Paper C and
partly in section 5.4 of this thesis. The latter section also supports Paper A and Paper B.
Table 1.2 Linkage of RQs and appended papers
Research questions Paper A Paper B Paper C
RQ1
X
X
RQ2
X
X
RQ3
RQ4
X
X
X
X
1.7 Outline of thesis
This thesis consists of two parts. The first part, chapters 1 to 5, summarises the thesis and
describes the research. More specifically, chapter 1 provides background information for
understanding the relevance of the research and its contextual perspective. The chapter
introduces the research problem and describes the research purpose, objectives and
questions. The theoretical framework is presented in chapter 2, along with an overview of
robustness and its related areas. Chapter 3 describes the research methodology from the
6
1.7 Outline of thesis
perspectives of research design, data collection, and data analysis. Chapter 4 includes the
results and a discussion of the research. Finally, the findings and contribution of the
research, as well as suggestions for future work, appear in chapter 5.
The second part consists of three appended papers. The first paper gives an overview of
robustness and discusses some relevant studies. It then develops a new road map for
railway infrastructure robustness, including a novel definition and a new framework of
robustness management, based on continuous improvement. It explores the opportunities
of applying the road map to the infrastructure of railway systems and outlines some
practical concerns and remaining challenges for future research. The results provide
guidelines for other research into robust infrastructure in railway maintenance. The
second paper proposes a novel quantitative evaluation approach for railway infrastructure
robustness. In this approach, both acceptable functionality and specified disturbance are
considered, together with various risk preferences for railway maintenance strategy
making. A numerical example illustrates its application using a new parameter, Mean Time
between Robustness (MTBR). The results show the approach is feasible for railway
infrastructure and flexible enough to handle complex situations. The third paper is a case
study of the Iron Ore line, Sweden. It consists of an evaluation of energy efficiency of snow
and ice protection for railway switches & crossings (S&C). This study will be further
developed in a new area called green robustness.
The three subsequent related papers listed in this thesis develop the dependability
improvement of railway infrastructure considering information logistics and risks and
maintenance cost.
7
8
CHAPTER 2
Theoretical framework
This chapter presents the theoretical framework of this research through a literature
review of studies of infrastructure robustness. The goal is to provide the theoretical basis
of the thesis and the appended papers following the RQs specified in Chapter 1. The
literature sources cited herein include conference proceedings, journals, international
standards, and other indexed publications.
2.1 Robustness overview
In the past 20 years, robustness studies have been conducted in various research fields,
including but not limited to: control engineering, statistics, computer science, operations
research, decision making theory, quality engineering/product development, and biology.
As shown in Table 2.1, the definitions and research focuses vary across research fields.
In general, robustness is used to describe the ability of a system (or a model) to maintain
its function (or performance) despite disturbance (incl. faults or perturbations). All
robustness studies aim to reduce uncertainty.
The approaches of robustness studies include: 1) statistics driven models from input to
output (e.g., control engineering, statistics, computer science, operations research, decision
making theory, and economics); 2) physical characteristics driven design or optimisation
of a complex system or an item (e.g., quality engineering/product development, and
biology); and 3) the integration of the two, as in many situations, both the statistics models
and the physical characteristics need to be considered simultaneously.
Considering the complexity of different systems and their special characteristics,
management of robust infrastructure requires knowledge from different research fields, in
other words, the adoption of a holistic approach. To date, however, robustness studies are
still being developed in separate fields. Moreover, none of the results from the above fields
can, in our view, be directly used for infrastructure robustness.
9
Theoretical framework
Table 2.1 An overview of robustness fields
Research fields
Definitions of robustness
Control engineering
Insensitivity to uncertainty of a control system in input and model
assumptions (Monje, Vinagre, Feliu, & Chen, 2008; K. Zhou & Doyle,
1998).
Statistics
Insensitivity to small deviations in the assumptions (Huber, 2011).
Computer science
Fault tolerance (Bhowmik, Tomsovic, & Bose, 2004; Rabinowitz &
Spilker Jr., 2005).
Operations research
Solution that remains functional with changing operational scenario
(Burdett, Kozan, & Strickland, 2012).
Decision making theory
Extent to which a system is able to maintain its function when some
aspect of the system is subject to perturbations (Walsh, Einstein, &
Gluck, 2013).
Economics
Ability of a model to remain valid under different assumptions
(Kuorikoski J., Lehtinen A., & Marchionni C., 2010).
Quality engineering/Product
development
Small variability in a system’s function under various noise conditions
(ISO, 2014a; Montgomery, 2008).
Biology
Ability to maintain performance in the face of perturbation and
uncertainty, a well-established property of living systems (Kitano,
2007; Stelling, Sauer, Szallasi, Doyle III, & Doyle, 2004).
10
2.2 Robust infrastructure related studies
2.2 Robust infrastructure related studies
2.2.1 An overview of infrastructure robustness
This subsection reviews over 350 scientific research papers from the last 20 years.
Stage I
Stage II
Stage III
Figure 2.1 Scientific articles related to infrastructure robustness
As shown in Figure 2.1, we can identify three stages in the research development of
infrastructure robustness:
 Stage I: before 2003;
 Stage II: from 2004 to 2011;
 Stage III: from 2012 to 2015.
Stage I features a few papers per year on computer science and network science, as well as
some on transportation science/timetabling. In stage II, more areas are introduced; the
areas of computer science and network science increase the most, but operations research
and transportation research increase too. Stage III has a peak in 2012, with articles from
more areas and more articles overall for all areas, especially network science and computer
science. In 2015, the number of papers for each area decreases but the number of different
areas with articles on robustness remains high. This stage may continue after 2015.
The review results also show that robustness within infrastructure is mainly studied in six
scientific areas with three common application areas for robustness. In decreasing order,
the six research areas are computer science, network science, transportation research,
infrastructure systems, water resources, control theory and civil engineering. The three
main application areas are robustness (control systems), robust optimisation and robust
decision making. The research and application areas are intertwined with each other and
11
Theoretical framework
across areas; one example is control theory – it is both a scientific area and an application
area.
2.2.2 Current status of robust infrastructure related studies
Currently, six research topics are related to infrastructure robustness: robustness of
structures, robustness of infrastructure in extreme events, robustness in risk management,
robustness in maintenance, robust road networks, and robustness of train timetables. So
far, infrastructure robustness has not been studied in a holistic manner; its development
requires knowledge from different research fields.
In this section, we argue infrastructure robustness requires continuous improvement.
Following the continuous improvement procedure proposed by Dr. W. Edwards Deming
(Deming, 1991), the infrastructure robustness management process can be divided into
four stages: Plan (P), Do (D), Study (S), and Act (A). When the Plan, Do, Study, Act procedure
is applied to robustness, we can derive the following objectives for each stage:
 Plan: identification of the robustness requirements (incl. acceptable functionality
in relation to disturbances, likelihood of unfavourable conditions and their impact)
and data requirements, definition of infrastructure system, and design of the work
process;
 Do: data collection, verification and analysis, execution of the robustness
management process;
 Study: construction of key performance indicators (KPIs) for the management of
infrastructure robustness;
 Act: optimisation of the management of infrastructure robustness.
Table 2.2 shows the current research status of these topics, including the definitions and
research focuses of each, from a continuous improvement perspective.
Present studies on the robustness of structures include solutions for structures to
withstand damage or disturbances without collapsing (CEN, 1994), or for robustness to be
measured and evaluated in a suitable way (Baker, Schubert, & Faber, 2008; U. Starossek &
Haberland, 2011). The former aims to identify the robustness requirements and make a
robustness management plan (P); the latter aims to set up KPIs (S).
12
2.2 Robust infrastructure related studies
Table 2.2 Current status of pivotal robust infrastructure related studies
Topics
Represented definitions
P*
D
S
Ability of a structure not to be damaged by events
like fire, explosions, impact or human errors to an
extent disproportional to the original cause (CEN,
1994).
X
X
Robustness of
infrastructure in
extreme events
Ability to resist specified disturbances or
perturbations so that at least some functionality can
be maintained (Bruneau et al., 2003).
X
X
Robustness in
risk
management
Ability to withstand a certain amount of stress with
respect to the loss of function of the system
(Hokstad, Utne, & Vatn, 2012).
X
Robustness in
maintenance
Ability to handle uncertainty in model input (Kuhn &
Madanat, 2006).
X
Robust road
networks
Extent to which, under pre-specified circumstances,
a network is able to maintain the function for which
it was originally intended (Snelder, Van Zuylen, &
Immers, 2012).
X
Ability of a timetable to withstand design errors,
parameter variations, and changing operational
conditions (ON-TIME Consortium, 2014)
X
Robustness of
structures
Robustness of
train timetables
A
X
X
X
X
X
X
X
X
*Research focuses for Plan, Do, Study, Act are marked with X.
Robustness studies examining the effect of extreme events on infrastructure look at
different types of infrastructure and interdependent infrastructure in pre and post disaster
situations. They consider how the infrastructure can resist and absorb extreme events and
adapt to post disaster circumstances from both a quantitative and a qualitative perspective
(P) (Bocchini, Frangopol, Ummenhofer, & Zinke, 2013; Bruneau et al., 2003). The area
builds on the strengths and weaknesses measured by, for instance, robustness, risk and
reliability. In particular, these studies focus on the development of KPIs (S).
Robustness studies of risk management for infrastructure follow the same principles as
robustness studies of extreme events. The objective is to identify risk and determine how
infrastructure robustness is influenced by it (P) (Hokstad et al., 2012).
Studies of robustness in maintenance aim to construct maintenance plans that can handle
uncertainties in predicting deterioration of infrastructure while balancing structural
performance and cost; thus, the research focus is on evaluating the accuracy of the
maintenance plan (P, S) and creating optimisation decision models (A). Note that the
objective is not the capability of the infrastructure (Kuhn & Madanat, 2006; Ok, Lee, & Park,
2013).
13
Theoretical framework
The study of robust road networks covers all stages of the PDSA process. However, there is
no established process for the continuous improvement of robustness within the area.
Vulnerability is the main focus and is regarded as the antonym of robustness; i.e., a
vulnerable network is not robust and vice versa. In this application area, research
contributions mainly include the identification of vulnerability related factors and their
impact (P), data analysis (D), and vulnerability improvement (A). A network robustness
index (NRI) is used to evaluate the performance of traffic and travel time (S) (Mattsson &
Jenelius, 2015; Scott, Novak, Aultman-Hall, & Guo, 2006; Snelder et al., 2012).
Robustness of train timetables is another common topic. So far, such studies have looked
for multi-objective optimisation solutions by balancing asset efficiency, robustness (delay
resistance) of a timetable, etc. Most studies are simulation-based. Although these studies
cover almost the whole PDSA process, as discussed, there is a lack of research on root
causes of delays, i.e., non-robust infrastructure in railway (Abril et al., 2008; Abril et al.,
2008; Fischetti et al., 2009; ON-TIME Consortium, 2014)
As Table 2.2 shows: 1) in different topics, robustness has different definitions and research
interests; 2) no results from any topic can be directly applied to railway infrastructure; 3)
most studies have limited research on specified stages in a continuous improvement
process, but as we argue, continuous improvement is essential to ensuring the robustness
of railway infrastructure.
2.3 Infrastructure robustness related topics in railway
This section introduces concepts and definitions related to railway infrastructure
robustness, including: resilience, rapidity, vulnerability, functionality, RAMS (reliability,
availability, maintainability, and safety), risk, sustainability, and system assurance.
2.3.1 Resilience
Considered the first to define resilience on a system level, Holling (1973) says it is a
measure of the persistence of systems and their ability to absorb change and disturbance
while maintaining the same relationship between population or state variables (Holling,
1973). Resilience is closely related to robustness and it also has a variety of definitions.
Robustness is mostly described as one part of resilience; rapidity is the other. The
difference between resilience and robustness is often subject-specific. Resilient
infrastructure is popular in the area of ecosystems, natural disasters and climate change;
studies consider how the infrastructure can handle changing and uncertain scenarios.
Recent reviews showing the variety of definitions are offered by Zhou et al. (H. Zhou, Wan,
& Jia, 2010) for infrastructure by Bocchini et al. (Bocchini et al., 2013); definitions of risk
are given by Hokstad et al. (Hokstad et al., 2012); finally, Hansson and Helgesson (Hansson
& Helgesson, 2003) provide an overall review. The majority of the definitions consider two
aspects of resilience: the ability to resist disturbances or shock and the ability to recover
14
2.3 Infrastructure robustness related topics in railway
from a non-functioning state. These are often referred to as robustness and rapidity,
respectively.
2.3.2 Rapidity
In the area of infrastructure, rapidity is seen as one part of resilience; together with
robustness, it forms resilience. Rapidity is the time it takes for a structure or system to
return to high functionality after a disturbance or shock. Robustness is also referred to as
a goal of resilience; see Zhou et al. (2010), Zhou et al. (H. Zhou et al., 2010) and Bocchini et
al. (2013) for infrastructure, Hokstad et al. (2012) for risk and Hansson and Helgesson
(2003) for an overall review.
2.3.3 Vulnerability
Vulnerability is often seen as the antonym of robustness (Snelder et al., 2012; U. Starossek
& Haberland, 2010; Steen & Aven, 2011). A vulnerable state is described in Electropedia as
the state of an electric power system in which a credible event will result in loss of load,
stresses of system components beyond their ratings, bus voltages and system frequency
outside tolerances, cascading, voltage instability, or some other instability (IEC, 1990b).
2.3.4 Functionality
Function refers to what is requested or desired from an asset, for example, dependability.
Therefore, a functional state is the antonym of a failed state (IEC, 1990a).
Functionality is central to the robustness of infrastructure. If an item, for some reason, has
lost its functionality, the asset is not robust for that type of disturbance (Bruneau et al.,
2003).
2.3.5 RAMS
RAMS (reliability, availability, maintainability, safety) can be characterised as a qualitative
and quantitative indicator of the degree to which a system, or its sub-system and
components, can be relied upon to function as specified and to be both available and safe.
The goal of a railway system is to achieve a defined level of rail traffic in a given time, safely.
Railway system RAMS is a combination of reliability, availability, maintainability and safety
(CEN, 1999):
 Reliability: The probability that an item can perform a required function under
given conditions for a given time interval.
 Availability: The ability of a product to be in a state to perform a required function
under given conditions at a given instant of time or over a given time interval
assuming the required external resources are provided.
15
Theoretical framework
 Maintainability: The probability that a given active maintenance action for an item
under given conditions of use can be carried out within a stated time interval when
the maintenance is performed under stated conditions and using stated procedures
and resources.
 Safety: Freedom from unacceptable risk of harm. Note that the letter S is sometimes
used to refer to supportability, sustainability or security (Patra, 2009; Stenström,
2014).
Robustness is the system’s ability to withstand a certain amount of stress before losing
system functionality. Hence, it includes the operational specifications of the RAMS
parameters but can go outside these specified boundaries (Hokstad et al., 2012).
2.3.6 Risk
Risk is defined as the effect of uncertainty on the achievement of objectives, where effect
refers to a deviation from the expected, whether positive or negative (ISO, 2009).
Robustness or its antonym, vulnerability, is an aspect of risk. The robustness of an asset is
what causes an effect after a specific disturbance. Hence, the risk is related to the
robustness of an asset (Hansson & Helgesson, 2003; Steen & Aven, 2011).
2.3.7 Sustainability
Sustainable development was introduced in 1987 as: development that meets the need of
the present, without compromising the ability of future generations to meet their own
needs (Brundtland, 1987). Following this many concepts and definitions have been
proposed were businesses worldwide realised that productivity stems not only from costeffective acquisitions, but also from an eco-friendly, sustainable and morally sound
corporate profile. Globally, companies from every industry have fine-tuned their approach
to corporate social responsibility (CSR) (Van Marrewijk, 2003).
Sustainability has three dimensions, economic, environmental and social. This area is a
major challenge for the transport industry and has received increasing attention over
recent years (The World Conservation Union, 2006; EC, 2015).
2.3.8 System assurance
The aim of system assurance is to verify that a system enforces a desired set of goals
(Jaeger, 2008). To achieve system assurance in railway systems, best practices from all
business sectors around the world must be spread, where feasible, into the rail industry.
System assurance has been widely adopted to assure engineering processes and products
conform to requirements for operational safety, reliability, availability, maintainability,
and national and international standards, procedures and regulations (BMT Group, 2016).
16
2.4 Robustness measurement
Considering the new challenges, railway infrastructure robustness must be included in
railway system assurance.
2.4 Robustness measurement
Infrastructure robustness related studies have focused on several pivotal areas, as shown
in figure 2.2: structure robustness, robustness of infrastructure in extreme events,
robustness in risk management, robustness in maintenance, robust networks, and
robustness of train timetables (Norrbin et al., 2016). To this point, the measurement of
infrastructure robustness has not been studied in a holistic manner.
In structural science, robustness is generally considered the ability of a structure to resist
abnormal circumstances and events without disproportional failure and progressive
collapse. Measurements of structure robustness mainly consider the impact of a damaging
event, such as an explosion, impact, fire, human error or structural deterioration (CEN,
2006; Saydam & Frangopol, 2011; U. Starossek & Haberland, 2010). Normally, the
measurement is the ratio of the damaged to the undamaged structure. Some measurements
evaluate different properties of structures, including: structural strength (ISO, 2007; Maes,
Fritzsons, & Glowienka, 2006), system stiffness (U. Starossek & Haberland, 2011), how
failure can progress to undamaged parts of the structure (Agarwal, Blockley, & Woodman,
2003; U. Starossek & Haberland, 2011), and risk perspective (Ellingwood & Dusenberry,
2005; Maes et al., 2006). Baker et al. (2008) have proposed a general measurement of
robustness based on risk, the most highly cited measure of robustness for structures. They
define robustness as an index, a ratio of direct and indirect risk. Indirect risk is damage not
directly caused by a damaging event; the higher the indirect risk, the less robust the system
(Baker et al., 2008).
Robust networks
Structural
robustness
Robustness of
infrastructure in
extreme events
Robustness
measurement
Robustness in risk
management
Robustness in
maintenance
Robustness of
train timetables
Figure 2.2 Infrastructure robustness measurement review areas
Infrastructure robustness in extreme events is generally referred to as the ability to
withstand a given extreme event and still deliver a service, often measured by the residual
17
Theoretical framework
functionality level after the occurrence of the event (Faturechi & Miller-Hooks, 2014). In
this area, robustness is seen as combining resilience and rapidity, as previously defined
(McDaniels, Chang, Cole, Mikawoz, & Longstaff, 2008). Two special applications are
earthquake hazards science and water management science. In earthquake hazards
science, a typical definition of robustness is the ability to resist specified disturbances or
perturbations so that at least some functionality can be maintained (Bruneau et al., 2003).
Bruneau et al. (2003) describe 24 qualitative and quantitative measures used to determine
the extent of damage after an earthquake. However, the focus is still on effects after a
damaging event; one example is how many houses have electricity after an earthquake.
In water management science, robustness is defined as the ability of a system to remain
functioning under disturbance (Mens, Klijn, de Bruijn, & van Beek, 2011). The focus of
measurement is on how the increasing uncertainty of climate change will affect water
systems. Measurement is based on a response curve for two dimensions: system response
and disturbance magnitude. Another quantitative measurement is proposed by Moody and
Brown (2013), but the focus remains on climate change and how various climate scenarios
affect the performance of the water system (Moody & Brown, 2013).
Robustness in risk management is viewed as the ability to withstand a certain amount of
stress without loss of function of the system (Hokstad et al., 2012). Robustness studies of
risk management for infrastructure follow the same principles as robustness studies of
extreme events and are closely related to resilience (Bruneau et al., 2003). The objective is
to identify risk and determine how infrastructure robustness is influenced by it.
Measurement mainly concerns structural risk and remaining durability (Baker et al., 2008).
Steen and Aven (2011) propose a proactive way to look at risk through resilience and
uncertainty using vulnerability to quantify robustness (Steen & Aven, 2011).
In maintenance, studies use robust optimisation to optimise maintenance plans by
handling uncertainty in model input. The uncertainties in predicting deterioration of
infrastructure are optimised whilst structural performance and cost are balanced; thus, the
research focus is on evaluating the accuracy of the maintenance plan and creating
optimisation decision models (Kuhn & Madanat, 2005; Kuhn & Madanat, 2006; Ok et al.,
2013). Robustness measurements are used to evaluate how well the models of
deterioration are measures of uncertainty and not measures of infrastructure ability.
Robustness in networks refers to extent to which, under pre-specified circumstances, a
network is able to maintain the function for which it was originally intended (Holme, Kim,
Yoon, & Han, 2002). Within network theory, the measures mainly focus on the network
topology, for example, how the nodes and links in the system are connected and how the
interconnection upholds the functionality of the network. A big area of concern is
robustness to attacks on networks, with vulnerability seen as the opposite of robustness.
Robustness is measured for networks in general (Holme et al., 2002; Schneider, Moreira,
Andrade Jr., Havlin, & Herrmann, 2011), as well as transportation networks (Schneider et
18
2.5 Summary of framework
al., 2011; Scott et al., 2006), power grids (Wang & Rong, 2011) and road transportation
networks (Nagurney & Qiang, 2007; Scott et al., 2006). In general, all measures consider
the topology of the network, how the nodes and links are connected and support each other
so the robustness of the network system can be enhanced.
Robustness of train timetables refers to the property of timetables to withstand design
errors, parameter variations, and changing operational conditions (ON-TIME Consortium,
2014). Studies look for multi-objective optimisation solutions by balancing asset efficiency,
timetable robustness (delay resistance) etc. The robustness measurements concern
margins (i.e., extra projects etc.) inserted into the timetables, where they should be placed,
and how their addition affects the costs generated by the extra travelling time. Another
concern is the effect of disruptions and disturbances in the system (Abril et al., 2008;
Fischetti et al., 2009; Goverde & Hansen, 2013; Kroon et al., 2009). The focus is on the
timetable characteristics and how buffer and headway times in the system can be
optimised while keeping the transportation system effective.
The present measures for infrastructure robustness are general, with little specific
information on the actual railway infrastructure. For example, the varying operational
scenarios and requirements of an asset during its lifetime, together with the varying
condition of the asset, are not considered in the current measurements for managing
railway assets. In the railway, RAMS parameters are, to a large extent, decided in the design
stage of an asset.
As the above review suggests, because definitions of robustness differ across research
areas, the measurements are different as well. Unfortunately, none is directly applicable to
railway infrastructure. Recently, however, Norrbin et al. (2016) proposed a new definition
for railway infrastructure robustness and suggested a new framework for its management.
The framework allows robustness to be measured in a qualitative way, but a quantitative
measurement is also needed to support continuous improvement (Norrbin et al. 2016).
2.5 Summary of framework
The chapter begins by providing an overview of robustness related studies. Then it
describes areas related to infrastructure robustness. It goes on to introduce key concepts
and definitions of railway infrastructure robustness, including: resilience, rapidity,
vulnerability, functionality, RAMS (reliability, availability, maintainability, and safety), risk,
sustainability, and system assurance. Finally, it discusses robustness measurement related
studies.
19
20
CHAPTER 3
Research methodology
This chapter presents the research methodology including, the research design, data
collection, and data analysis.
3.1 Research design
As shown in Figure 3.1, the work comprises three stages. The motivation for the research
originated in pressure put on the Swedish Transport Administration, Trafikverket, by the
Swedish government. More specifically, the Ministry of Enterprise and Innovation gave
Trafikverket the mission of developing a framework for the operation and maintenance of
roads and railways, including their infrastructure. The Trafikverket management together
with the Ministry decided that the reporting should be based on six key result areas (KRA):
robustness, punctuality, capacity, usability, safety and environment & health. The
highlighted areas for infrastructure were punctuality, robustness and safety (Sveriges
Riksdag, 2012; Trafikverket, 2012).
Trafikverket, Sweco and Luleå Railway Research Centre (JVTC) jointly decided to conduct
an exploratory pre-study on the subject of robustness for railway infrastructure. This was
done, first, to lay the foundation for the larger study by describing the current research
gaps and, second, to achieve new insight into robustness in four areas; transport
infrastructure, maintenance in railway infrastructure, infrastructure resilience and
robustness in the railway system.
The first stage of the project included a literature review, interviews with five European
infrastructure managers and a project to develop the follow up on robustness at
Trafikverket. The interviews combined with the literature review revealed the research
gaps in railway infrastructure robustness and allowed the formulation of a problem
statement. This, in turn guided the formulation of the research purpose, objectives, and
questions: RQ1, RQ2, RQ3, and RQ4. This represents a second stage of the project,
exploratory research.
In the third stage, the work drew on both descriptive and explanatory research to construct
a novel road map of railway infrastructure robustness. The road map includes a new
definition, a new framework, and a new quantitative measurement approach. Based on
this novel robustness roadmap and new approaches, railway infrastructure robustness can
21
Research methodology
be improved continuously, thus supporting system assurance. To illustrate its application,
the study suggests and develops the concept of “green robustness,” using a real case study
focusing on energy efficiency to support the concept.
Generally speaking, the first stage reveals the research gaps, the second stage analyses
them, and the third answers the research questions and fills the research gaps.
Research Background
Project ideas motivated by
Trafikverket & SWECO
Maintenance in
railway
infrastructure
Infrastructure
resilience
Interview
Robustness in
railway
Exploratory research
Research Objectives and questions formulation
Fill research gaps
Transport
infrastructure
Literature Review
Railway Infrastructure Robustness:
attributes, evaluation, assurance, and improvement
RQ 1: What are the
attributes?
RQ 2: How can it be
enhanced?
RQ 3: How can it be
evaluated?
RQ 4: How can it be
applied for
sustainability?
New definition
System assurance
acceptable
functionality
specified
disturbances
New framework
House of Robustness Management (HORM)
New measurement
Application and green robustness
Figure 3.1 Design of this research
22
specified time period
Answer research questions
Ability to perform
Continous
improvement
Descriptive and explanaory research
A novel Road Map
3.2 Data collection
3.2 Data collection
3.2.1 Interviews
Two groups of people were interviewed: European infrastructure managers, and people
from the Swedish Transportation Administration (Trafikverket). The purpose for the
interviews was to take into account the experience and opinions of the personnel involved
in the maintenance of railway infrastructure to help identify the research gaps and
formulate the research purpose, objectives and questions.
3.2.2 Interviews with European infrastructure managers
European infrastructure managers from six organisations were interviewed to collect
information on their definition and application of robustness in asset management. Since
the interviews represented our first meetings with the respondents, they were semistructured with open-ended questions to yield a better understanding of how each
manager used robustness. The interviews were followed by complementary questions as
required. To increase the quality of the interviews, all interviews were recorded with a
Dictaphone, so that they could be further analysed later, if necessary.
The interviewees’ positions were in asset management and maintenance performance
measurement at their respective national infrastructure managers. In some cases, their
knowledge of and involvement with robustness were minor. If so, they gave us contact
information for more suitable people in their organisation. These interviews were
performed by phone or via email.
The choice of infrastructure managers was based on organisational comprehensiveness;
they had to have good knowledge and ample practice in the asset management area. To be
able to compare the results across organisations, Trafikverket was used as a reference.
We interviewed five infrastructure managers to ensure sufficient variety in the answers,
while honouring time and cost constraints. We chose Europe because of the availability and
proximity of contacts. The infrastructure managers came from Trafikverket in Sweden,
ProRail in the Netherlands, Network Rail in the UK, ÖBB in Austria, the Technical University
in Graz (LCC evaluation for ÖBB and railway research) and SBB in Switzerland; the
interviewees are listed in Table 3.1. The questions comprising the basis of the interviews
include:





Do you have a definition of robustness?
What is your definition of robustness?
How is your follow up on infrastructure robustness conducted?
What parameters do you use to measure robustness?
What do you call having an infrastructure that can withstand and handle different
conditions and disturbances?
23
Research methodology
Table 3.1 List of interviewees
Organisations
No. of
Positions of the interviewees
interviewees
Trafikverket
13
Maintenance strategy, maintenance planning, climate customising,
maintenance analysis, track experts
ProRail
4
Asset management, program manager robust & punctual change
program
Network Rail
6
System analysis team
ÖBB
2
Track asset management
TU Graz
4
LCC, track access charges,
SBB
6
Track asset management, track measurement & evaluation, track
subgrade, track wear & track access charges and strategy department
3.2.3 Interviews with Trafikverket, Sweden
At the time of the study, Swedish Transportation Administration (Trafikverket) was
conducting an asset management project examining six revised key result areas (KRA:s),
selected in 2012 by Trafikverket and the Ministry of Enterprise and Innovation:
robustness, punctuality, capacity, usefulness, safety and environment & health. The project
aimed to find new top-level key performance indicators (KPIs) for the KRA:s. I was a
member of the core group developing the KPIs but also specifically in the group developing
performance indicators for robustness. Because of this work and the fact that they had a
clear and official definition for robustness Trafikverket was set as a reference to enable
comparison of all infrastructure managers.
Table 3.2 Competence of participants in the robustness group
Group
Robustness
Position
Section
Unit
Division
Analyst
Analysis
Planning
Maintenance
Analyst
Analysis
Planning
Maintenance
Planner
Strategic
planning
National long term
planning
Planning
Climate adaption
National
maintenance
National maintenance
Maintenance
LCC analyst
Socio-economy
Short term planning
Planning
Dewatering expert Planning
Road systems
Maintenance
Senior analyst
Transport quality
Planning
Usability
The study of robustness was carried out by an expert group of seven people from different
units of Trafikverket with varying areas of expertise on robustness; see Table 3.2. The work
24
3.2 Data collection
included discussions within the group and the collection of data and information between
the group meetings. The first step, presented herein, was to identify indicators that could
be used with existing data at Trafikverket.
An overarching core group was responsible for heterogeneity across KPIs, i.e. each of the
six KPIs should measure a different aspect of performance, but together they would include
most organisational perspectives. This group also gave feedback on and approval of the
work done in the six different KPI groups; hence, they were part of the development of
robustness for Trafikverket. The participants and their competences are displayed in Table
3.3.
3.2.4 Data sources: operation and maintenance data
Historical data of rail infrastructure operation and maintenance were collected from
Trafikverket. The various systems considered include:
 BASUN: Delay statistics of trains operated under service.
 BIS: Asset information system
 OFELIA: Failure reporting system for Trafikverkets railway infrastructure assets.
About 150 data posts for each failure including information on, failure, geography,
maintenance times and more.
 BESSY: System for safety and maintenance inspection of permanent facilities; also
performs automated ultrasonic inspections to detect cracks and defects in rails,
switches and welds before they cause broken rails.
 Rufus: System wherein operation and maintenance contractors: 1) register
preventive maintenance measures, and 2) re-submit corrected inspection
complaints arising in Bessy.
 VVIS: Weather stations with climate data covering Sweden. Data include
temperature on the ground and in the air, max wind speed, min wind speed, average
wind speed, type and amount of precipitation and humidity. Sweden is also divided
into a grid, Mesan, with prognoses of weather conditions using the respective
weather stations to get weather information for the whole of Sweden.
 ÖVV: Control system for switch and cross (S&C) heating, providing data on energy
consumption, temperature of the rails of the S&C and climate data.
 Lupp: User interface used to aggregate and analyse data from different databases.
In particular, in the case study discussed in Paper C, data were collected from the pilot test
S&Cs at Kirunavaara and then verified. Data included the following: outside temperature,
rail temperature from two different sensors, precipitation indicator, turbo (full power)
indicator and energy consumption. All were hourly data from the ÖVV system.
25
Research methodology
Table 3.3 Competence of the participants in the core group
Group Position
Core
Section
Unit
Division
Project leader
Analysis
Planning
Maintenance
Senior analyst
Analysis
Planning
Maintenance
Qualified investigator Long-term Planning
Long-term Planning
Central functions
Controller
Traffic management
Traffic management
Operations
Developer
Business system
Business development Finance and control
Strategist
Strategic development Strategic development Central functions
Strategist
Strategic development Strategic development Central functions
Project leader
Analysis
Planning
Maintenance
3.3 Literature review
The literature review drew on scientific publications databases, such as Scopus, Web of
Science, Google Scholar, etc. Various types of references were reviewed, including
conference and journal papers, monographs, theses, standards, and technical reports. The
search generally used infrastructure as the keyword in the title or abstract, with robust*
the key word in articles and reviews. Secondary references were reviewed for some
studies.
A summary of the results of the literature review appears in Chapter 2, “Theoretical
framework”, and is applied in Chapter 4, “Results and discussions”, and Chapter 5, “Further
research”. More details are found in the appended papers.
3.4 Data Analysis
A real case study is described in Paper C. In this study, the energy consumption of the S&Cs
(switches and crossings) at Kirunavaara station was analysed using data for the outside
temperature and the temperature of the rail sensor with the lowest value. The analysis
used a temperature span of -12 to -5 degrees Celsius. Data were not used if there were
indications of the “turbo” function being used or snowfall. The temperature range was
based on the expert knowledge of the project group. The aim was to minimise variability
with a sufficient amount of data. The exclusion of data for snowfall and the “turbo” function
were also intended to minimise variability; as the heating is operating on full power in both
scenarios, unwanted effects could result. Methods used to compare the energy
consumption are described below.
 Two-sample t-test: If the variance of two specimens is assumed equal, a two-sample
t-test can be used to compare the means by hypothesis testing. The null hypothesis
26
3.4 Data Analysis
will be no difference in the compared means; the alternative hypothesis will be
bigger, smaller, or both depending on the specific test (Montgomery, 2008).
 Paired t-test: The paired t-test is done using a paired comparison design technique,
a special case of randomised block design. A block refers to a relatively homogenous
experimental unit. This test helps eliminate variability because it compares within a
block. As it compares the difference between samples, the degrees of freedom are
decreased, but since the comparison is done in pairs which are similar, the
assumption is that variability will be minimised because of reduced outside
disturbances (Montgomery, 2008). This is a good method for comparing before and
after scenarios.
These two methods are well known and accepted methods for comparison analysis. They
were decided through discussions with analysis expertise.
Paper B implements a numerical example to illustrate the quantitative measurement of
railway infrastructure robustness. In this illustration, the generation of numerical data
follows the background of the example and approach assumptions. In particular, according
to the new definition given here, railway infrastructure robustness is the ability of railway
infrastructure to maintain an acceptable level of functionality in the presence of specified
disturbances for a specified time period. Data analysis follows the mathematical expression
of this robustness function written as:
R rob (t) = P(T > t|D, A, ω)
where D refers to the dataset of specified disturbances; A refers to the corresponding
acceptable functionality; ω refers to the dataset of weights of risk preference; and
∑ω = 1.
In this research, for further study on system assurance, data were analysed to evaluate the
availability status of switches and crossings (S&Cs) at 23 stations. The failures were
grouped per station. The time period was January 2010 through May 2015. The mean time
between failures and the mean time to repair was calculated and the stations were ranked
from 1-23. The failures were categorized in two ways, snow and ice related failures and
total amount. A ratio of the proportion of snow and ice related failures in comparison to
the total amount of failures was also calculated. The analysis was done to evaluate the
current failure population for each station. The S&Cs with new snow and ice protection
equipment were marked for future study. Pre-study results can be found in Figure 3.2 and
Figure 3.3. The ordinate values are the mean times to repair, MTTR, ranking from 1-23. The
abscissa values are the mean times between failures, MTBF, ranking from 1-23. Figure 3.2
is showing the result only considering snow & ice related failures. Figure 3.3 is showing the
result for all failures. The number corresponding to each point are showing the amount of
snow & ice related failures divided by the total amount of failures. 1 would represents the
27
Research methodology
Mean time to repair, ranking from 1-23
case when all failures are snow & ice related and 0 would represent the case when no
failures were snow & ice related. The same ratio is used in both figures.
Mean time between failure, ranking from 1-23
Mean time to repair, ranking from 1-23
Figure 3.2 MTTR and MTBF ranking for 23 different stations for snow & ice related failures with
snow & ice related failure ratio
Mean time between failures, ranking from 1-23
28
3.4 Data Analysis
Figure 3.3 MTTR and MTBF ranking for 23 different stations for all causes of failures with snow &
ice related failure ratio
An explanation of the quadrants is shown in figure 3.4. The best performing S&Cs are in the
green quadrant; they show a long time between failures and short repair times. Conversely,
those in the red quadrant are the worst performing because of their long repair times and
short time between failures. S&Cs falling into the two yellow quadrants need improvement.
The matrixes were used to visualise the failure outcomes for the different stations to
evaluate the performance of the S&Cs per station.
Figure 3.4 Explanation of the quadrants
In addition, the total number of failures for each S&C from 2010-01-01 to 2015-06-01 were
compiled and the 20 individual S&Cs with the most failures were listed. They are
designated according to the station, as shown in Table 3.4. The number following the
station name refers to the number of the S&C for that station
29
Research methodology
Table 3.4 The 20 S&Cs with most reported failures from 2010-01-01 to 2015-06-01
Station and nr. of S&C Number of failures
Kopparåsen1
122
Peuravaara705
46
Vassijaure2
44
Kaisepakte1
41
Stordalen1
39
Kaisepakte4
36
Krokvik1
35
Björkliden2
34
Rautas1
33
Rensjön1
31
Björkliden1
30
Råtsi781
29
Rensjön2
29
Stenbacken2
27
Abisko östra1
26
Peuravaara704
24
Kirunavaara751
23
Kopparåsen2
23
Stordalen2
20
Krokvik2
20
Torneträsk2
20
30
CHAPTER 4
Summary of appended papers
This chapter summarises the appended papers. The summary includes their titles,
purposes and objectives, methodology, and main contributions. Chapter 1 explains the
linkage of the research questions (RQs) and the appended papers (Table 1.2).
4.1 Paper A
Title: Infrastructure robustness for railway systems
Summary: In the railway industry, most maintenance approaches are based on certain
specified conditions, e.g., RAMS (Reliability, Availability, Maintainability and Safety) and
risk. But the reality is more complex. Instead of the assumed conditions, unfavourable
conditions may occur from either natural or operational causes; in such cases, robustness
can be an effective analytical approach. To adequately consider unfavourable conditions
and to reduce uncertainties in railway maintenance, this study conducts a holistic
examination of railway infrastructure robustness. It gives an overview of robustness and
discusses some relevant studies. It then develops a new road map for railway infrastructure
robustness, including a novel definition and a new framework of robustness management,
based on continuous improvement. It explores the opportunities of applying the road map
to the infrastructure of railway systems and outlines some practical concerns and
remaining challenges for future research. The results provide guidelines for other research
into robust infrastructure in railway maintenance.
Main Contribution: The paper gives an overview of robustness and develops a new road
map for railway infrastructure robustness, including a novel definition and a new
framework for robustness management, based on continuous improvement. The new
definition of railway infrastructure robustness is the ability to maintain an acceptable level
of functionality in the presence of specified disturbance for a specified time period. This
definition is useful, as it clarifies the research boundaries between robustness and
reliability, resilience, and risk. The study develops a ground-breaking framework, named
house of robustness management (HORM), based on continuous improvement to support
infrastructure robustness management in railway systems; HORM consists of robustness
management goals and guidance, a continuous improvement process, and support systems.
Finally, the study considers the opportunities and challenges of applying the road map to
31
Summary of appended papers
railway infrastructure and suggests guidelines for other research into robust infrastructure
in different industries.
4.2 Paper B
Title: A novel approach for measuring railway infrastructure robustness
Summary: To sufficiently consider unfavourable conditions and reduce uncertainty in
maintenance, railway infrastructure robustness was given a new definition and framework
by Per Norrbin et al. in 2016 (Norrbin et al., 2016). Building on this work, the present study
proposes a novel approach to quantitatively measure railway infrastructure robustness. In
this approach, acceptable functionality and specified disturbance are considered together
with various risk preferences for railway maintenance strategies. The study provides a
numerical example to illustrate the application of the suggested approach using a new
parameter, Mean Time between Robustness (MTBR). The results show the approach is
feasible for railway infrastructure and flexible enough to handle complex situations.
Main Contribution: The paper describes a novel quantitative method to evaluate railway
infrastructure robustness. The proposed measurement can provide very good support for
an infrastructure manager. The keys to this robustness measurement are: the
differentiation of the disturbance, the acceptable functionality depending on the
disturbance, and the weighting of the disturbance. The measure provides new support for
a railway infrastructure manager in two ways. Firstly, the measure of robustness will
consider important operational scenarios where the infrastructure manager needs good
knowledge of the performance of the assets. Secondly, the differentiated requirements for
robustness will provide more dynamic evaluations of the infrastructure assets than the
current approaches. Both will reduce the uncertainties related to railway infrastructure.
4.3 Paper C
Title: Energy efficiency optimisation for railway switches & crossings: a case study in
Sweden
Summary: With increasing environmental concerns and mounting financial and legislative
pressures on the railway industry, monitoring and optimising energy consumption are
becoming crucial. The northernmost track in Sweden, the Iron Ore Line (Malmbanan),
operates in a harsh sub-arctic climate. Effective snow and ice protection ensures the
successful operation of this line, especially the railway switches and crossings (S&Cs). One
strategy is to remove snow and ice with electrical heating; however, this consumes a great
deal of energy. According to the Swedish Transport Administration, the total energy
consumption for the 6 800 S&Cs in Sweden equipped with electrical heating is 200-130
GWh/year, at costs of approximately 10-15 M€/year. The next generation of S&Cs now
being introduced is equipped with almost triple the effect of S&C heating, jumping from
32
4.3 Paper C
about 10kW to almost 30kW. The project described in this paper compares the
performance of the new and old equipment to determine which is the most energy efficient.
It tests the currently installed equipment using a two-sample t-test, i.e., comparing
enhanced snow and ice protection equipment and normal equipment for two different
S&Cs at the same station. In addition, it performs a paired t-test to test present and previous
energy consumption. The results show a 30% increase in efficiency with the new type of
snow and ice protection equipment. These are preliminary results from the first of 10
locations using the new snow and ice protection equipment. So far, the results are very
promising. Further studies with more data from other locations will be conducted to verify
the results.
Main Contribution: The paper describes a pre-study and evaluates the possible energy
savings accruing from a new type of snow and ice protection equipment for S&Cs. The study
is an important step in the development of novel ideas on green robustness, a concept
discussed in Chapter 5.4.
33
34
CHAPTER 5
Results and discussion
This chapter discusses the findings for each research question (RQ).
5.1 Results and discussion related to RQ1
RQ1: What are the attributes of railway infrastructure robustness?
The first research question is answered in Paper A with the formulation of a novel
definition of railway infrastructure robustness.
This work proposes the following definition of railway infrastructure robustness: the
ability of railway infrastructure to maintain an acceptable level of functionality in the
presence of specified disturbances for a specified time period. In the following sections, we
break this definition down into its four key attributes, see Figure 5.1: ability to perform,
functionality at an acceptable level, specified disturbances, and specified time period.
infrastructure robustness
for railway systems
Ability to perform
acceptable
functionality
specified
disturbances
specified time
period
Figure 5.1 Attributes of railway infrastructure robustness
5.1.1 Ability to perform
Like other robustness related definitions, those for infrastructure robustness describe
some kind of capability that should be evaluated in both quantitative and qualitative ways
to get a comprehensive measurement; this requires knowledge of statistics, maintenance,
quality, mechanics, etc.
35
5.1 Results and discussion related to RQ1
5.1.2 Acceptable functionality
The most basic function of railway systems is transporting people or goods from one point
to another. Different parts of the system have different contributing functions to fulfil the
service needs of transport. The basic function of the infrastructure of railway systems is to
allow rolling stock to pass safely and in a timely fashion according to plan with acceptable
deviations.
The acceptable level of infrastructure functionality should be decided by stakeholders (incl.
infrastructure manager, railway companies) and customers (incl. railway companies,
rolling stock owners) according to existing agreements. Functionality can take various
forms, e.g., train speed, train delays or punctuality, etc.
5.1.3 Specified disturbances
In this study, disturbances refer to unfavourable conditions which raise uncertainties
about the ability to maintain functionality at an acceptable level. To determine the
robustness of railway infrastructure, relevant disturbances should be identified. The
likelihood of their happening must be minimal in the immediate future (if the probability
is high, this becomes a reliability improvement problem). Note that extreme events like
earthquakes, floods or deliberate attracts on infrastructure which are studied in resilience
are not within the study scope. Finally, the robustness of an asset depends on the type of
disturbance; if it is robust against mechanical wear, it is not necessarily robust against
heavy rainfall.
5.1.4 Specified time period
The time period for robustness is specified but can be changed depending on the scope of
the follow up. It can refer to the winter months, the latest snowfall, or any lifetime
calculation span, e.g. running distance for wheels or calendar time. With changing
calculation spans, today’s specified disturbances can become tomorrow’s required
operational conditions. For instance, 10 years ago, the wagon axle load on the iron ore line
in Sweden was 22.5 tonnes; thus, a load of 25 tonnes for one axle represented a
disturbance. However, as the wagon axle load is now 30 tonnes, a 25-tonne axle load no
longer represents a disturbance. In other words, the requirements of infrastructure
robustness can change with time.
5.1.5 Definition clarification
To clarify the novel definition of the infrastructure robustness of railway systems proposed
in this study, definitions of robustness, reliability, resilience, and risk are compared in
Table 5.1. Specified definitions are following:
 Railway infrastructure robustness (section 3.1 in this paper): the ability of railway
infrastructure systems to maintain an acceptable level of functionality in the
presence of specified disturbances for a specified time period.
36
Results and discussion
 Reliability [2]: the probability that an item can perform a required function under
given conditions for a given time interval.
 Resilience [12, 14]; the ability to resist, absorb and adapt to disruptions and return
to normal functionality. A resilient system has (1) reduced probability of failures; (2)
reduced consequences from failures; and (3) reduced time to recovery.
 Risk [46]: the effect of uncertainty on objectives. Risk is often expressed in terms of
a combination of the consequences of an event (including changes in circumstances)
and the associated likelihood of occurrence.
Table 5.1 Clarification of robustness, reliability, resilience, and risk
Considerations
Ability
Functionality
Disturbances

Robustness
Yes
Acceptable
functionality
Specified
disturbance
Specified
Reliability
Yes
Required
function
Resilience
Yes
Normal
functionality
No
Disruptions
Time period
Specified
Railway
Railway industry
Yes
Infrastructure
e.g., EN-50126
*: a dash means it is not applicable.
Not specified
Transportation
infrastructure
Risk*
Considered as a
consequence
Identified
uncertainties
Not specified
In general
As shown in Table 5.1, there are similarities but also clear differences among the above
concepts. More specifically:
 Ability: All concepts concern an attribute of an asset, except for risk, which
describes the impact and likelihood of an uncertainty.
 Functionality: Unlike robustness, reliability considers required function which
normally has been decided in the design stage. Resilience considers the ability
to stay in and return to normal functionality. Robustness includes the
expectations of stakeholders in setting the standards for acceptable
functionality, making it possible to be altered over time.
 Disturbances: Unlike robustness which considers a specified disturbance, or
reliability which does not consider any disturbances, resilience considers all
types of disturbances with an emphasis on extreme events. Risk reflects
uncertainties which can be positive or negative.
 Time period: Robustness and reliability have a specified time period; resilience
and risk have an unspecified time span.
 Railway Infrastructure: Unlike reliability and resilience, there is no definition of
robustness for the railway industry. The novel definition proposed in this study
aims to fill this gap. Risk has a general definition currently accepted in the
railway industry.
37
5.2 Results and discussion related to RQ2
This novel definition also guides the proposed new measurement approach in Paper B.
5.2 Results and discussion related to RQ2
RQ2: How can railway infrastructure robustness be enhanced through a continuous
improvement process to achieve system assurance?
The second research question is answered by developing a new framework of railway
infrastructure robustness management in Paper A.
To ensure railway infrastructure robustness can be enhanced through a continuous
improvement process and to achieve system assurance, Paper A develops a framework of
robustness management. The pivotal elements of the House of Robustness Management
(HORM) are: Roof, Ceiling, Walls, and Floor. The structure is visualised in Figure 5.2.
Robustness Management Goal
Robustness Management Guidance
PLAN







DO
STUDY
Robustness
requirements
identification
Acceptable
functionality
Specified
disburbances
Specified time period
Data requirements
clarification
Infrastructure criticality
identification
 RCM
 FMECA
 RAMS
 SIL
 HAZOP
 RCA
 FTA
 FRACAS


Data collection and
verification
Robustness
Management process
Execution

KPI for robustness of
railway assets

KPI for data quality
assessment

KPI for Robustness
management resources

KPI for support system

KPI for KPI system

KPI related to LCC
ACT

Benchmarking

New Design & Renewal

Robustness Strategy
Review and
Optimization
 RCM
 FMECA
 RAMS
 SIL
 HAZOP
 RCA
 FTA
 FRACAS

Resources Optimization

BPR

KPI System Optimization
Robustness
management process
construction
Support System
Figure 5.2 House of Robustness Management (HORM)
5.2.1 Roof of HORM
In the proposed definition, the robustness of infrastructure systems can be evaluated in
quantitative and qualitative ways. In the roof of the HORM, the goals of robustness
management should be set to clearly guide the process.
38
Results and discussion
First, robustness management goals will stem from the overall business goals of the
organisation and support their achievement. The Swedish Transport Administration
(Trafikverket) views robustness as a pivotal part of its overall business goals, along with
punctuality, capacity, usability, safety and environment & health. The achievement of
robustness will support the overall business goals. At the same time, it will influence the
goals of the other pivotal parts.
Second, robustness management goals should be hierarchically set, i.e., disaggregated
down through different levels. For instance, when the robustness goal for the entire track
system on level 1 is allocated down to level 2, it can be expressed as the robustness goals
of the track and switches & crossings (S&Cs); when it moves further down to level 3, it
becomes the robustness goals of rails, fastenings, sleepers and ballast, etc., separately. This
is shown in Figure 5.2.
Third, the goals should be clear for the working group on each level. These goals should
relate to and provide guidance for the working group’s role in, and contribution to, the total
achievements of the organisation.
Fourth, they should be disseminated periodically among the stakeholders and improved
gradually.
5.2.2 Ceiling of HORM
The ceiling is one of the basic structures of a house. This understanding carries through to
the HORM framework: the related infrastructure management guidance required to
achieve robustness management goals and guide the overall process comprises the ceiling
of the HORM. This guidance could include standards, work instructions, expert knowledge,
and research reports.
Although there is no specific robustness management standard for railway infrastructure,
related standards include international or regional standards, or those requested by a
specified country/industry. For instance, reliability related standards should be considered
as important references for robustness management; these include international standards
(ISO1, IEEE2, IEC3, etc.), European standards (CEN4, ETSI5, etc.), and railway standards
(UIC6, etc.).
Work instructions present a sequence of steps to execute a task or activity. The Swedish
Transport Administration has several handbooks for performing various maintenance
actions, for example, a handbook on how to clear snow from S&Cs (The Swedish Transport
1
ISO: International Organisation for Standardisation.
IEEE: Institute of Electrical and Electronics Engineers.
IEC: International Electrotechnical Commission.
4 CEN: European Committee for Standardisation.
5
ETSI: European Telecommunications Standards Institute.
6 UIC: International Union of Railways.
2
3
39
5.2 Results and discussion related to RQ2
Administration (Trafikverket), ). These handbooks are important, not only to guide
maintenance but also to assure the correct management of robustness.
Expert knowledge can supply valuable guidance as well. This knowledge can come from
group and/or individual experience. Sometime there is a need to consult experts with
knowledge of special cases to support the creation of robustness strategies.
Research reports may include conclusions and suggestions which could guide robustness
management. For instance, in Sweden many ideas come from research institutes
(universities or research centres) or special reports from other organisations, including
labs or consultant companies.
5.2.3 Walls of HORM
Following the thinking of continuous improvement developed by Edwards Deming, the
four walls of the proposed HORM are Plan (P), Do (D), Study (S), and Act (A) (Deming,
1991). The use of this PDSA cycle ensures robustness management is not only standardised
but also improved gradually. The walls follow the robustness management goals (roof) and
guidance (ceiling) and are supported by Information and Communication Technologies
(ICT) systems (floor).
Plan
Plan is the start of the PDSA cycle. Main tasks involve identifying robustness requirements,
clarifying data requirements, identifying infrastructure criticality and constructing the
process for managing robustness.
In light of the novel definition of railway infrastructure robustness in section 3, acceptable
functionality, specified disturbances, and specified time period need to be decided in the
Plan stage. Data requirements should be identified according to the robustness
requirements. These, in turn, require the identification of infrastructure criticality, as
different criticality may lead to different robustness strategies. Knowledge-driven
approaches can be applied to identify criticality, including but not limited to: RCM
(Reliability Centred Maintenance), FMECA (Failure Mode, Effect and Criticality Analysis),
RAMS (Reliability, Availability, Maintainability and Safety), SIL (Safety Integrity Level),
HAZOP (Hazard and Operability Study), RCA (Root Cause Analysis), FTA (Fault Tree
Analysis), FRACAS (Failure Reporting, Analysis and Corrective Action System), etc. The
work process of robustness management should also be defined in the Plan stage.
Do
In the Do stage, the main tasks include data collection and verification, as well as execution
of the process defined in the Plan stage.
Data collection and verification represent a key element of effective robustness
management. Besides work orders, data could come from engineering design data,
component test data, system test data, operational (and experience) data from similar
40
Results and discussion
systems, field-tracking studies in various environments, computer simulations, related
standards and operation manuals, experience data from similar systems, expert judgment
and personal experience, warranty data, etc. Robustness is concerned with uncertainties
and unfavourable conditions that happen infrequently. Incomplete or unreliable data will
lead to misleading results in data processing (i.e. the conditioning and feature
extraction/selection of acquired data) and decision making (i.e. recommendations for
robustness management actions based on diagnosis and/or prognosis, as well as
recommendations for maintenance). Data verification ensures the data are able to evaluate
the robustness of the railway assets in the Study stage. Finally, the robustness management
process needs to be determined in this stage, so that real situations can be assessed in the
next stage.
Study
In the Study stage, a system of key performance indicators (KPIs) should be set up to
control and monitor robustness management. The suggested KPI system consists of five
different aspects, with KPIs selected for the robustness of railway assets, data quality
assessment, resources, support systems, and the entire robustness management process.
Economic considerations such as the life-cycle cost (LCC) should also be included.
Specific KPIs for measuring the robustness of assets are analysed together with the related
KPIs, for instance, RAMS, risk, and railway-specific KPIs like punctuality and train delays.
The purpose of these KPIs is to evaluate the robustness of the assets. When evaluating data
quality, KPIs for data collection and data processing are included.
Robustness management resources include: technical (hardware or software); human
(personnel and competence) and organisational aspects (commitment, management style
and preferences).
The effectiveness and efficiency of the support systems, including, for example, databases
and computerised maintenance management systems (CMMS), must be periodically
reviewed and controlled.
Finally, the entire robustness management process, the indicators for LCC, and the KPI
system itself must be controlled and monitored.
Act
The Study stage is likely to reveal gaps between the present situation and the robustness
management goals. This leads to the Act stage where improvements in performance are
initiated to achieve the overall robustness management goals. This may include setting new
goals. Benchmarking, review and optimisation of the robustness strategy, resource
optimisation, business process re-engineering (BPR), and KPI system optimisation
represent some possible improvement actions in this stage.
41
5.2 Results and discussion related to RQ2
Benchmarking will shed light on the differences in the robustness of comparable assets in
different operational and organisational scenarios.
The strategies for achieving robustness objectives need to be reviewed and possibly
optimised. This can be done using various maintenance management techniques, for
example, RCM, FMECA, RAMS, SIL, HAZOP, RCA, FTA or FRACAS.
Resource optimisation seeks optimal solutions considering multi-objective criteria. This
includes economic, technical (hardware or software), human (personnel and competency)
and organisational aspects (commitment, management style and preferences).
Business process re-engineering (BPR) can be used to improve processes of robustness
management defined in the Plan stage.
Finally, the KPI system needs to be optimised.
All the walls together form a sequential continuous process; for example, after the Act
stage, the Plan stage is reviewed and improved. In this way, continuous improvements are
ensured.
5.2.4 Floor of HORM
To improve the decision-making process in robustness management, data from various
sources (e.g. product, production, maintenance, and business) must be collected,
integrated, fused, and analysed to transform them from information into knowledge.
Therefore, data form the foundation or floor of the house. Most of the maintenance related
data which will support various maintenance management techniques (for example, RCM,
FMECA, RAMS, SIL, HAZOP, RCA, FTA or FRACAS) should have been recorded and stored in
various management systems, such as CMMS, enterprise resource planning (ERP), special
software for condition monitoring or other industry-specific supporting tools. Another
important support is eMaintenance.
5.2.5 Discussion of HORM
This work develops a road map for railway infrastructure robustness, including a novel
definition and a new framework of robustness management. The definition brings clarity
to a research area where unfavourable conditions may affect asset functionality, as
expressed in the concepts of reliability, resilience and risk. The proposed framework for
robustness management, termed HORM, provides a holistic view of, and a continuous
improvement process for, managing the robustness of railway infrastructure.
While the framework shows great promise, the concept of railway infrastructure
robustness is continuing to develop. Remaining challenges include the following:
 First, compared with the research topics of reliability, resilience and risk,
infrastructure robustness requires further study from both theoretical and practical
perspectives, especially for railway systems. This will include further case studies
using both qualitative and quantitative measures.
42
Results and discussion
 Second, of the four key elements discussed in section 3, accepted functionality can
influence the identification of other elements.
 Third, robustness management needs to be considered in a holistic way; HORM
could be a guide for continuous improvement, but top level management must agree
to take a holistic view.
 Fourth, given the quickly developing areas of ICT, determining how to collect,
integrate, fuse, and analyse data, as well as how to transform them from information
into knowledge, will be an ongoing challenge in future robustness studies.
 Fifth, the reality is complex, and unfavourable conditions are hard to identify and
predict; insufficient knowledge or over-estimation may lead to high operational
risks and costs.
 Sixth, the determined robustness will necessarily be conditional, as it will depend on
the identified conditions. An asset can be robust against one disturbance, for
example, snowfall, but not robust against another, for instance, heavy axle loads.
 Seventh, trusted operation and interoperability with other modes and emergency
preparedness should be considered in railway infrastructure robustness.
Although RQ2 question is answered in Paper A from a qualitative perspective, the new
measurement approach proposed in Paper B also helps answer it. However, because Paper
B mainly contributes to RQ3, it is discussed in the following section.
5.3 Results and discussion related to RQ3
RQ3: How can railway infrastructure robustness be evaluated using a quantitative
measurement approach?
To answer the third research question, Paper B proposes a new quantitative approach for
measuring railway infrastructure robustness.
5.3.1 A new quantitative measurement approach
The new definition of railway infrastructure robustness is the following: The ability of
railway infrastructure to maintain an acceptable level of functionality in the presence of
specified disturbances for a specified time period. It can be expressed mathematically as
R rob (t) = P(T > t|𝐷, 𝐴; 𝜔)
where 𝐷 refers to the dataset of specified disturbances, 𝐴 refers to the corresponding
acceptable functionality, ω refers to the dataset of weights of risk preference, and ∑ ω =
1 , 𝜔 ≥ 0.
Robustness differs from reliability and maintainability in two ways: first, 𝐷 and 𝐴 are
exclusive to robustness; second, the threshold for failure in reliability and maintainability
is normally fixed and implemented in the design phase, whereas for robustness, a
43
5.3 Results and discussion related to RQ3
differentiation of the acceptable level of functionality depends on the disturbance scenario
and condition of the evaluated infrastructure.
Table 1 gives an example of a dataset of 𝐷 and 𝐴.
Table.1 Example of dataset of 𝑫 and 𝑨
𝑫
𝑨
𝝎
𝐷0
𝐴0
𝜔0
𝐷1
𝐴1
𝜔1
𝐷2
𝐴2
𝜔2
𝐷3
𝐴3
𝜔3
𝐷𝑢
𝐴𝑢
𝜔𝑢
The following hold for the dataset of 𝐷 and 𝐴 shown in the table:
 If only 𝐷0 is considered, it becomes a normal reliability problem, and 𝜔0 = 1;
 Extreme events are not considered, as these represent a resilience problem;
 Boundaries between 𝐷0 and others can be changed; therefore, reliability, robustness,
and resilience considerations could be changed;
 Mean Time between Failure (MTBF) and Mean Time to Repair (MTTR) need to be
considered for 𝐷1 , 𝐷2 , 𝐷3 , based on the identification of A; therefore, Mean Time
between Robustness (MTBR) should be developed and calculated according to 𝐷
and 𝐴 ;
 𝐷, A, and 𝜔 can be changed both numerically and depending on asset type, the status
of the infrastructure, seasonal variations, amount of traffic, maintenance strategy,
condition of the infrastructure asset, maintenance budget, etc.;
 Each disturbance event could be treated as a special reliability problem or reliability
+ maintainability problem.
 R rob = ∑i=0,1,2,3 ωi R 𝐷𝑖 ,𝐴𝑖 = ω0 R D0,A0 + ω1 R 𝐷1 ,𝐴1 + ω2 R 𝐷2 ,𝐴2 + ω3 R 𝐷3 ,𝐴3

5.3.2 An example of a robustness evaluation
 Snow and ice protection of S&Cs;
 Two solutions, S1 and S2: S1 refers to normal snow and ice protection equipment for
S&Cs, and S2 refers to new snow and ice protection equipment;
 X = 𝐷, A, or 𝜔;
 Three disturbance scenarios are identified. Undisturbed normal scenarios are
marked X 0 For the last scenario, X u represents uncertainty and uncontrolled factors.
These can be extreme situations or unexpected situations with very long delay times.
The level of uncertainty will vary and will be decided based on the track section or
asset type. Uncertainty can be chosen to be accepted to a certain degree.
This example describes the robustness evaluation of S1 and S2; S1 represents normal snow
and ice protection equipment for S&Cs, and S2 represents new snow and ice protection
equipment. Two UIC 60E S&Cs with a moveable frog, both shown in figure 5.3, are selected
44
Results and discussion
for evaluation. S1 consists of wooden boards placed on the outside of the rails for selected
parts of the S&C to prevent snow from packing in and around it, causing the S&C to
malfunction. In addition, a heating system heats up the rails to melt the snow; snow can
also be cleared manually.
S2, the new and upgraded version, consists of the normal wooden boards but also has a
rubber coating and polyurethane insulation underneath. The edges of the snow protection
equipment are covered to prevent snow from blowing in from the sides and to preclude
heat convection. These new features aim to increase the robustness of the S&Cs in winter
and to increase the energy efficiency by preserving the heat in the rail.
Figure 5.3 Pictures of the two versions of snow and ice protection of S&Cs, S1 and S2.
45
5.3 Results and discussion related to RQ3
Table 5.2 Robustness evaluation of S1 and S2 with data for illustrational purposes only.
𝑫: Snowfall
(mm/day)
𝐷0 :
less than 10mm
𝑨: delay time
(hour)
𝐴0 : no time
delay <5min
𝝎:
∑𝑖=0,1,2,3 𝜔𝑖 =1
𝜔0 = 0.4
𝑅𝑟𝑜𝑏
𝑅𝑟𝑜𝑏 0
consider MTBF
MTBR-MTBF
MTBR=MTBF
MTBR-MTTR
𝑅𝑟𝑜𝑏 S1
𝑅𝑟𝑜𝑏 S2
MTTR, t<5min
0.68
0.7
𝐷1 :
(10 30)mm
𝐴1 : less
than 1
hour
𝜔1 = 0.3
𝐷2 :
(30 50)mm
𝐴2 :
𝐷3 :
(50 100)mm
𝐴3 :
1-3hours
3-12hours
𝜔2 =0.2
𝜔3 =0.1
𝑅𝑟𝑜𝑏 1
𝑅𝑟𝑜𝑏 2
𝑅𝑟𝑜𝑏 3
consider
MTBR1
consider
MTBR2
consider
MTBR3
MTBF
during 𝐷1 :
36hours
< 2h
0.85
0.82
MTBF
during 𝐷2 :
48hours
<4h
0.95
0.98
MTBF
during 𝐷3 :
24hours
<8h
0.98
0.99
𝐷𝑢
> 100mm
𝐴𝑢 :>12hours
0.02
0.01
𝑅𝑟𝑜𝑏 𝑆1=0.4*0.68+0.3*0.85+0.2*0.95+0.1*0.98 = 0.815
𝑅𝑟𝑜𝑏 𝑆2=0.4*0.7+0.3*0.82+0.2*0.98+0.1*0.99 = 0.821
Therefore, R rob S2 is more robust than R rob S1.
5.3.3 Discussion
This measurement of railway infrastructure robustness is specific for the railway and can
be used for either scenarios or time. The acceptable functionality is specific for railway
systems, and possible disturbances are specific for railway infrastructure. However, the
measurement is flexible enough to evaluate the robustness for different disturbances,
requirements and time periods.
The main criterion is acceptable functionality, 𝐴. This can be achieved by conducting a large
number of maintenance activities, but this approach will not enhance the robustness of the
infrastructure. Therefore, MTBF and MTTR are important determinants of the appropriate
criteria. For example, in scenario 𝐷3 , there will be a very heavy snowfall, but delays should
not exceed 3-12 hours. MTBF states that despite the snow, maintenance should not be
required on the S&C for 24 hours. If maintenance is needed, the S&C is not considered
robust. For the same scenario, MTTR considers a repair time of more than 8h should not be
required.
46
Results and discussion
For current methods of asset management, the requirements of infrastructure assets with
respect to level of service are mostly set in the design stage and apply throughout the
asset’s service life for all weather conditions and seasons and for every operational
scenario. However, the present approaches do not consider possible changes in the
operational scenario of an asset over its entire service life. Nor do they consider the
changing condition of infrastructure assets in different seasons of the year or over the life
time. This is especially noticeable for a long lasting asset like railway infrastructure. This
is, to some extent, compensated for by the maintenance strategy and maintenance actions.
Still the demands that can be put on a newly installed track system are not the same as
those after 10 years of service. There is also a possibility of minor or bigger changes in
customer demand. Therefore, the proposed robustness measure uses a differentiated
acceptable functionality. The actual operational scenario will put a certain amount of stress
on the infrastructure, making it more or less likely to perform the acceptable function. The
infrastructure robustness measure, or the acceptable functionality, then, will depend on
the actual operational scenario and provide the possibility for differentiated and more
realistic service levels for passengers and train operators.
This robustness measure can be used to evaluate the current condition of the asset
depending on its life cycle phase. In this case, the disturbance could be a degraded state due
to, for example, budget constraints. The total robustness and, hence, the performance
demands of the evaluated infrastructure can then be communicated to stakeholders. This
will set realistic expectations, which are quantified and even put into contracts. This could
be an important measure to prevent the loss of goodwill due to ageing assets in parts of the
infrastructure system.
The measure should also be able to aggregate for stations and track sections, making it
more related to operational performance.
To this point, the method has been tested with fictitious data. Future research plans include
conducting a case study to verify the results using data from the databases at the Swedish
Transport Administration, Trafikverket.
The asset types and disturbances evaluated for robustness will be decided depending on
their criticality. The asset types, operational scenarios or the criteria for criticality will be
individual to each infrastructure manager. The decision on asset type and typical
disturbances is important for the measure to be applicable and useable. The use of MTBF
for scenarios other than 𝐷0 will be challenging, as it is necessary to define the time period
for a similar disturbance scenario. At present, these data are the hardest to obtain.
47
5.4 Results and discussion related to RQ4
5.4 Results and discussion related to RQ4
RQ4: How can sustainability for railway be developed by considering infrastructure
robustness?
In this research, sustainability of railway infrastructure is mainly considered from
environmental and energy efficiency perspectives, and RQ4 is answered by developing a
new concept: green robustness. Basically, the suggested measurement approach consists
of two parts: first, the measurement of infrastructure robustness, as discussed in section
5.3; second, the measurement of green robustness. Green robustness represents an
important application of railway infrastructure robustness to support system assurance.
5.4.1 Green robustness
In traditional thinking of energy efficiency of manufacturing, efficiency is defined as
Efficiency (E) =benefit/total effort or ρ = Emin,abs /Ereal . Normally, power consumption is
the main consideration. The problem with current research is its lack of consideration of
either infrastructure robustness or energy consumption for recovery (e.g., maintenance
cost).
The purpose of the concept of green robustness is to optimise energy consumption of
infrastructure by considering infrastructure robustness, already discussed in Paper A and
Paper B. The basic idea is to evaluate railway infrastructure energy efficiency (EE) by
considering railway infrastructure robustness and the real energy consumption (including
energy consumption for recovery). Considering the new definition of robustness in this
study, this can also be stated as the following: green robustness refers to energy efficiency
considering energy consumption under operation and requirements for functional
recovery.
5.4.2 Measuring green robustness for railway infrastructure
According to section 5.4.1, in green robustness:
Energy Efficiency (EE) = Robustness/ Energy Consumption (for operation and recovery)
(EC).
The calculation of infrastructure robustness draws on the results of RQ 3. The calculation
of the unit energy consumption (EC) is exemplified in a simple way in the example shown
in Table 5.3, but further study will evaluate and develop its applicability.
In Table 5.3, energy consumption comes from both operation and recovery for both cases.
For simplicity, S1 has an energy consumption of 2 units and S2 has one of 1.5 units. The
energy consumption will depend on the specific requirements for operation and recovery
for each case. A higher energy consumption can come from, for example, more powerful
and energy consuming point machines or more frequent maintenance actions.
48
Results and discussion
Table 5.3 Green robustness evaluation of S1 and S2
EE
𝑅𝑟𝑜𝑏 EC (for operation and recovery)
S1 0.865
2 units
0.4325
S2 0.712
1.5 units
0.4747
From the table, the resulting Green robustness (GR) evaluation is that GR S2 is better than
GR S1.
5.4.3 Discussion
The awareness that we live in a world where we need to save and manage our limited
resources is increasing. Therefore, sustainability is an important aspect in all areas. This
measurement of green robustness evaluates the robustness of infrastructure in relation to
sustainability aspects. It is important for an infrastructure manager to achieve a desired
balance between the two and to be able to verify improvements. The green robustness
measure allows the infrastructure manager to achieve a long lasting sustainable
infrastructure system.
This idea is being developed and needs further study before green robustness can be
evaluated in a precise way. The calculation of energy consumption in both operation and
recovery of functionality is one issue to be resolved. The other is the development of the
concept’s applicability to different asset types and various operational and recovery
scenarios. This will require detailed research using different scenarios, as well as case
studies.
49
50
CHAPTER 6
Conclusions, contributions and future research
This chapter presents conclusions on the research, explains the contributions, and suggests
future research.
6.1 Conclusions
To consider “unfavourable conditions” and to reduce “uncertainties” in railway
maintenance, this research holistically examines railway infrastructure robustness by
means of a new road map, including a novel definition and a framework. It proposes the
following definition of railway infrastructure robustness: the ability of railway
infrastructure to maintain an acceptable level of functionality in the presence of specified
disturbances for a specified time period. This definition is useful, as it clarifies the research
boundaries between robustness and reliability, resilience, and risk. The study develops a
ground-breaking framework, house of robustness management (HORM), based on
continuous improvement to support infrastructure robustness management in railway
systems; HORM consists of robustness management goals and guidelines, a continuous
improvement process, and support systems. Finally, the study considers the opportunities
and challenges of applying the road map to railway infrastructure and provides guidelines
for other research into robust infrastructure in different industries.
In addition, this study describes a novel quantitative method to evaluate railway
infrastructure robustness. The proposed measurement can provide very good support for
an infrastructure manager. The keys to this robustness measurement are: the
differentiation of the disturbance, the acceptable functionality depending on the
disturbance, and the weighting of the disturbance. The measure provides new support for
a railway infrastructure manager in two ways. Firstly, the measure of robustness will
consider important operational scenarios where the infrastructure manager needs good
knowledge of the performance of the assets. Secondly, the differentiated requirements for
51
Conclusions, contributions and future research
robustness will provide more dynamic evaluations of the infrastructure assets than the
current approaches. Both will reduce the uncertainties related to railway infrastructure.
To determine the efficacy of the proposed method, the study tested enhanced snow and ice
protection equipment for railway switches and crossings (S&Cs) and performed a prestudy for a project evaluating and comparing the robustness of two different types of snow
and ice protection equipment (normal and enhanced). This was followed by the proposal
to balance sustainability and robustness in a measure called “green robustness.”
6.2 Contributions
Contributions are defined as: new findings and classifications from the literature review;
development of new theoretical methods; application of existing theoretical methods;
application of existing applied methods to a new dataset; and introduction of existing
methods to a new discipline. Accordingly, the research contributions are considered to be:
 A novel definition of railway infrastructure robustness through which the attributes
are identified. (Paper A)
 A framework called house of robustness management (HORM) provides support and
guidelines to continuously improve the robustness of railway infrastructure assets
and achieve system assurance. (Paper A)
 A quantitative measure for railway infrastructure robustness based on specified
disturbance, acceptable functionality and mean time between robustness. (Paper B)
 A new concept called green robustness that incorporates sustainability with
robustness management. (Paper C with licentiate thesis chapter 5)
6.3 Future research
Further research will include the following:
 More detailed studies with operational data from the Trafikverket databases that
can quantify the robustness of different asset types and disturbances to verify the
railway infrastructure robustness measurement method.
 Further studies to optimise the energy efficiency of railway S&Cs. More data,
including other locations and a larger temperature span, will be used to verify the
railway infrastructure robustness measurement method and the proposed
evaluation method for green robustness.
 Further studies to aggregate the robustness measure to make it suitable for a
managerial level.
52
REFERENCES
Abril, M., Barber, F., Ingolotti, L., Salido, M., Tormos, P., and Lova, A. (2008). An assessment
of railway capacity. Transportation Research Part E: Logistics and Transportation
Review, 44(5), 774-806.
Adger, W., Hughes, T., Folke, C., Carpenter, S., and Rockström, J. (2005). Social-ecological
resilience to coastal disasters. Science, 309(5737), 1036-1039.
Agarwal, J., Blockley, D., and Woodman, N. (2003). Vulnerability of structural systems.
Structural Safety, 25(3), 263-286.
Andersson, E., Peterson, A., and Törnquist Krasemann, J. (2013). Introducing a new
quantitative measure of railway timetable robustness based on critical points. In: The
5th International Seminar on Railway Operations Modelling and Analysis-Rail
Copenhagen 2013. Copenhagen: International Association of Railway Operations
Research.
Arasteh Khouy, I. (2013). Cost-effective maintenance of railway track geometry: A shift from
safety limits to maintenance limits. PhD thesis. Luleå University of Technology.
Baker, J. W., Schubert, M., and Faber, M. H. (2008). On the assessment of robustness.
Structural Safety, 30(3), 253-267.
Barabady, J. (2007). Production assurance : Concept, implementation and improvement. PhD
thesis. Luleå University of Technology.
Bhowmik, S., Tomsovic, K., and Bose, A. (2004). Communication models for third party load
frequency control. IEEE Transactions on Power Systems, 19(1), 543-548.
BMT Group (2016). Rail engineering and system assurance. [online] Available at:
http://www.bmt.org/markets/buildings-infrastructure-and-rail/railengineering/rail-engineering-and-system-assurance/ [Accessed 2 May 2016].
53
REFERENCES
Bocchini, P., Frangopol, D., Ummenhofer, T., and Zinke, T. (2013). Resilience and
sustainability of civil infrastructure: Toward a unified approach. Journal of
Infrastructure Systems, 20(2), 04014004 1-16.
Brundtland, G. (1987). Report of the world commission on environment and development:
"our common future". United Nations.
Bruneau, M., Chang, S., Eguchi, R., Lee, G., O'Rourke, T., Reinhorn, A., von Winterfeldt, D.
(2003). A framework to quantitatively assess and enhance the seismic resilience of
communities. Earthquake Spectra, 19(4), 733-752.
Burdett, R., Kozan, E., and Strickland, C. (2012). A practical approach for identifying
expected solution performance and robustness in operations research applications.
ASOR Bulletin, 31(2), 1-28.
CEN (1994). EN eurocode 1: Basis of design and actions on structures. Geneva: European
Committee for Standardisation.
CEN (1999). EN 50126: Railway specifications – the specification and demonstration of
reliability, availability, maintainability and safety (RAMS). Geneva: European
Committee for Standardisation.
CEN (2006). EN-1991-1-7: Structures- part 1-7: General actions - accidental actions. (No. EN1991-1-7). Brussels: European Committee for Standardisation.
Deming, W. (1991). Out of the crisis, 1986. Cambridge, Mass.: Massachusetts Institute of
Technology Center for Advanced Engineering Study. Xiii, 507
EC (2015). Horizon 2020, 11. smart, green and integrated transport, work programme 20162017. ( No. 11. Smart green and integrated transport). Brussels: European Commission.
Ekberg, A., and Paulsson, B. (2010). INNOTRACK (innovative track systems) concluding
technical report. Paris: International Union of Railways (UIC).
Ellingwood, B., and Dusenberry, D. (2005). Building design for abnormal loads and
progressive collapse. Computer-Aided Civil and Infrastructure Engineering, 20(3), 194205.
Faturechi, R., and Miller-Hooks, E. (2014). Measuring the performance of transportation
infrastructure systems in disasters: A comprehensive review. Journal of Infrastructure
Systems, 21(1), 04014025 1-15.
54
REFERENCES
Fischetti, M., Salvagnin, D., and Zanette, A. (2009). Fast approaches to improve the
robustness of a railway timetable. Transportation Science, 43(3), 321-335.
Goverde, R. M., and Hansen, I. (2013). Performance indicators for railway timetables. In:
the Intelligent Rail Transportation (ICIRT), 2013 IEEE International Conference, 301306.
Hansson, S. O., and Helgesson, G. (2003). What is stability? Synthese, 136(2), 219-235.
Hokstad, P., Utne, I., and Vatn, J., (ed). (2012). Risk and interdependencies in critical
infrastructures: A guideline for analysis. London: Springer.
Holling, C. S. (1973). Resilience and stability of ecological systems. Annual Review of Ecology
and Systematics, 4, 1-23.
Hollnagel, E., Woods, D., and Leveson, N. (2006). In Leveson N., ebrary I. and Woods D. D.
(Eds.), Resilience engineering concepts and precepts. Aldershot, England ; Burlington,
VT; Aldershot: Aldershot, England ; Burlington, VT : Ashgate.
Holme, P., Kim, B. J., Yoon, C., and Han, S. (2002). Attack vulnerability of complex networks.
Physical Review E, 65(5), 056109.
Huber, P. (2011). Robust statistics. New Jersey: Springer.
IEC (1990a). IEC 60050-191: International electrotechnical vocabulary: Chapter 191:
Dependability and quality of service. Geneva: International Electrotechnical
Commission (IEC).
IEC (1990b). IEC 60050-191-22-06: International electrotechnical vocabulary: Chapter
191: Dependability and quality of service; operating states of electric power systems. (No.
191-22-06). Geneva: International Electrotechnical Commission (IEC).
ISO
(2009).
Risk
management
principles
and
guidelines
(ISO 31000:2009, IDT). (No. ISO 31000:2009). Geneva: International Organization for
Standardization (ISO).
ISO (2014a). Applications of statistical and related methods to new technology and product
development process -- robust parameter design (RPD). (No. ISO 16336:2014). Geneva:
International Organization for Standardization (ISO).
ISO (2014b). ISO 55000: Asset management - overview, principles and terminology. (No. ISO
55000:2014). Geneva: International Organization for Standardization (ISO).
ISO (2007). 19902. 2007. “Petroleum and natural gas industries–Fixed steel offshore
structures.”. Geneva: International Organization for Standardization (ISO).
55
REFERENCES
Jaeger, T. (2008). Operating system security. Synthesis Lectures on Information Security,
Privacy and Trust, 1(1), 1-218.
Kitano, H. (2007). Towards a theory of biological robustness. Molecular Systems Biology, 3:
137, 1-7.
Kobbacy, K., and Murthy, P. (2008). Complex system maintenance handbook. London:
Springer Science and Business Media.
Kroon, L., Huisman, D., Abbink, E., Fioole, P., Fischetti, M., Maróti, G., Ybema, R. (2009). The
new dutch timetable: The OR revolution. Interfaces, 39(1), 6-17.
Kuhn, K., and Madanat, S. (2005). Model uncertainty and the management of a system of
infrastructure facilities. Transportation Research Part C: Emerging Technologies, 13(56), 391-404.
Kuhn, K., and Madanat, S. (2006). Robust maintenance policies for markovian systems
under model uncertainty. Computer-Aided Civil and Infrastructure Engineering, 21(3),
171-178.
Kuorikoski J., Lehtinen A., and Marchionni C. (2010). Economic modelling as robustness
analysis. British Journal for the Philosophy of Science, 61(3), 541-567.
Luleå University of Technology. (2015). Robust and reliable transport system - a strategic
agenda for resource pooling, collaboration, innovation and research. (No. 2016). Luleå:
Luleå University of Technology.
Maes, M., Fritzsons, K., and Glowienka, S. (2006). Structural robustness in the light of risk
and consequence analysis. Structural Engineering International, 16(2), 101-107.
Mattsson, L., and Jenelius, E. (2015). Vulnerability and resilience of transport systems - A
discussion of recent research. Transportation Research Part A, 81, 16-34.
McDaniels, T., Chang, S., Cole, D., Mikawoz, J., and Longstaff, H. (2008). Fostering resilience
to extreme events within infrastructure systems: Characterizing decision contexts for
mitigation and adaptation. Global Environmental Change, 18(2), 310-318.
Mens, M., Klijn, F., de Bruijn, K., and van Beek, E. (2011). The meaning of system robustness
for flood risk management. Environmental Science and Policy, 14(8), 1121-1131.
Monje, C., Vinagre, B., Feliu, V., and Chen, Y. (2008). Tuning and auto-tuning of fractional
order controllers for industry applications. Control Engineering Practice, 16(7), 798812.
56
REFERENCES
Montgomery, D. (2008). Design and analysis of experiments. Singapore: John Wiley and
Sons.
Moody, P., and Brown, C. (2013). Robustness indicators for evaluation under climate
change: Application to the upper great lakes. Water Resources Research, 49(6), 35763588.
Nagurney, A., and Qiang, Q. (2007). Robustness of transportation networks subject to
degradable links. EPL (Europhysics Letters), 80(6), 68001.
Norrbin, P., Lin, J., and Parida, A. (2016). Infrastructure robustness for railway systems.
International Journal of Performability Engineering.
Ok, S., Lee, S., and Park, W. (2013). Robust multi-objective maintenance planning of
deteriorating bridges against uncertainty in performance model. Advances in
Engineering Software, 65, 32-42.
ON-TIME Consortium (2014). Methods and algorithms for the development of robust and
resilient timetables, deliverable D3.1. Available at: http://www.ontimeproject.eu/deliverables.aspx [Accessed 03 May 2016].
Patra, A. P. (2009). Maintenance decision support models for railway infrastructure using
RAMS and LCC analyses (Ph.D. dissertation).
Rabinowitz, M., and Spilker J. Jr., (2005). A new positioning system using television
synchronization signals. IEEE Transactions on Broadcasting, 51(1), 51-61.
Saydam, D., and Frangopol, D. (2011). Time-dependent performance indicators of damaged
bridge superstructures. Engineering Structures, 33(9), 2458-2471.
Schneider, C., Moreira, A., Andrade J. Jr., Havlin, S., and Herrmann, H. (2011). Mitigation of
malicious attacks on networks. Proceedings of the National Academy of Sciences of the
United States of America, 108(10), 3838-3841.
Scott, D., Novak, D., Aultman-Hall, L., and Guo, F. (2006). Network robustness index: A new
method for identifying critical links and evaluating the performance of transportation
networks. Journal of Transport Geography, 14(3), 215-227.
Shift2Rail (2015). Shift2Rail, european rail research and inovation initiative. Available
at:http://shift2rail.org/ [Accessed 03 May 2016].
Snelder, M., Van Zuylen, H., and Immers, L. (2012). A framework for robustness analysis of
road networks for short term variations in supply. Transportation Research Part A:
Policy and Practice, 46(5), 828-842.
57
REFERENCES
Starossek, U., and Haberland, M. (2010). Disproportionate collapse: Terminology and
procedures. Journal of Performance of Constructed Facilities, 24(6), 519-528.
Starossek, U., and Haberland, M. (2011). Approaches to measures of structural robustness.
Structure and Infrastructure Engineering, 7(7-8), 625-631.
Steen, R., and Aven, T. (2011). A risk perspective suitable for resilience engineering. Safety
Science, 49(2), 292-297.
Stelling, J., Sauer, U., Szallasi, Z., Doyle III, F., and Doyle, J. (2004). Robustness of cellular
functions. Cell, 118(6), 675-685.
Stenström, C. (2014). Operation and maintenance performance of rail infrastructure model
and methods.
Sveriges Riksdag (The Swedish Parliament) (2012). Utförande av drift och underhåll av
järnväg; skrivelse 2012/13:19 ( execution of operation and maintenance of railway, in
swedish).
Available
at:
http://www.riksdagen.se/sv/DokumentLagar/Forslag/Propositioner-och-skrivelser/Utforande-av-drift-ochunderha_H00319/?text=true [Accessed 03 May 2016].
The Swedish Transport Administration (Trafikverket). TDOK 2013:0655 version 3,
spårväxel vinterhandbok (winter handbook for S&C: in Swedish). Borlänge:
Trafikverket.
The World Conservation Union (2006). The future of sustainability: Re-thinking
environment and development in the twenty-first century. In: Report of the IUCN
Renowned Thinkers Meeting. IUCN.
Trafikverket,
(the
Swedish
Transport
Administration)
(2012).
Införande av ett gemensamt styrramverk för drift och underhåll av väg och järnväg.
slutredovisning regeringsuppdrag (in swedish). (No. TRV 2012/11921). Borlänge,
Sweden: Trafikverket.
Van Marrewijk, M. (2003). Concepts and definitions of CSR and corporate sustainability:
Between agency and communion. Journal of Business Ethics, 44(2-3), 95-105.
Walsh, M., Einstein, E., and Gluck, K. (2013). A quantification of robustness. Journal of
Applied Research in Memory and Cognition, 2(3), 137-148.
Wang, J., and Rong, L. (2011). Robustness of the western United States power grid under
edge attack strategies due to cascading failures. Safety Science, 49(6), 807-812.
Whaley, G. (2004). Transportation management. In M. Kutz (Ed.), Handbook of
transportation engineering, 38.1. New York, NY, USA: McGraw-Hill.
58
REFERENCES
Zhou, H., Wan, J., and Jia, H. (2010). Resilience to natural hazards: A geographic perspective.
Natural Hazards, 53(1), 21-41.
Zhou, K., and Doyle, J. (1998). Essentials of robust control. Upper Saddle River, NJ: Prentice
hall.
59
60
Part II
61
PAPER A
Infrastructure Robustness for Railway systems
Authors:
Norrbin Per, Lin Jing & Parida Aditya
Accepted for publication in:
International Journal of Performability Engineering.
Infrastructure Robustness for Railway systems
PER NORRBIN
1, 2
*
, JING LIN 1 , ADITYA PARIDA 1
1. Division of Operation and Maintenance Engineering, Luleå University of Technology, 97187,
Luleå, Sweden
2. Sweco Rail AB, 97324, Luleå, Sweden
Abstract: In the railway industry, most maintenance approaches are based on certain “specified
conditions”, e.g., RAMS (Reliability, Availability, Maintainability and Safety) and Risk. But the reality is
more complex. Instead of the assumed conditions, “unfavorable conditions” may occur from either natural
or operational causes, where robustness can be an effective approach. To adequately consider “unfavorable
conditions” and to reduce “uncertainties” in railway maintenance, this study conducts a holistic
examination of railway infrastructure robustness. It gives an overview of robustness and discusses some
relevant studies. It then develops a new road map for railway infrastructure robustness, including a novel
definition and a new framework of robustness management, based on continuous improvement. It explores
the opportunities of applying the road map to the infrastructure of railway systems and outlines some
practical concerns and remaining challenges for future research. The results provide guidelines for other
research into robust infrastructure in railway maintenance.
Keywords: maintenance; railway infrastructure; robustness; continuous improvement
1. Introduction
According to the European program Horizon 2020, multifaceted challenges in transport
infrastructure include: 1) growing needs to make infrastructure more resilient to keep pace with the
increasing mobility needs; 2) reducing the impact of infrastructure on the environment; and 3) dealing
with declining resources to maintain and upgrade transport infrastructure. New design and
maintenance approaches must be developed to handle these issues, as current methods are inadequate
[1].
In maintenance practices of railway systems, most attention has been placed on RAMS (Reliability,
Availability, Maintainability and Safety) study, which consists of safety and availability requirements
[2, 3]. In Sweden, a popular area of study is RAM4S which incorporates supportability, sustainability,
and security [4, 5] The approaches to the maintenance of railway systems have changed significantly
over the last century with a shift from an emphasis on technology to techno-economic considerations
[6, 7]. Today, various maintenance strategies are applied in an integrated way depending on the
criticalities and cost-effectiveness of a particular infrastructure. With new requirements generated by
the increasing complexity of both the infrastructure system and operating conditions, these strategies
continue to be developed [8].
The major drawback in current approaches is that they all assume certain “specified conditions”
[2]. The reality is more complex; conditions are frequently “unfavorable” and may require a different
approach. For railway systems, “unfavorable conditions” stem from both natural and operational
causes. In the case of the former, for example, in northern Sweden, the infrastructures operate in harsh
conditions, including snow, ice and extreme temperatures ranging from -40°C to +25°C. Such
naturally occurring “unfavorable conditions” often lead to time delays and economic losses. As for
the latter, in the same region of Sweden, the regulations governing the axle load have recently
increased from 22.5 tons to 30 tons because of changing transport needs; an even higher axle load
* Corresponding author: [email protected]
1
limit (32.5 tons) is being tested. These new operational requirements demand continuous
improvement and upgrades on the original track. Obviously, such operationally caused conditions are
not “favorable” to the original track system: they will create more Rolling Contact Fatigue (RCF)
problems and increase the degradation of the track infrastructure [9].
Resilience studies can improve the ability of an infrastructure to withstand disturbances caused by
various “uncertainties” [10-12]. To this point, most studies have considered extreme events, including
natural events like earthquakes or floods, and man-made events, like deliberate attacks on
infrastructures [13-16]. Some argue studies considering resilience to extreme (natural and man-made)
events should focus on adaptation measures and strategies to ensure seamless transport and user
protection [1]. However, both naturally caused and operationally caused “unfavorable conditions”
which do not belong to extreme events need to be studied as well. This represents a significant gap
in the research. Research on the resilience of infrastructure is generally divided into two perspectives:
rapidity or robustness. The former is defined as how fast the system can return to a high functional
level, and the latter expresses the ability to resist specified disturbances so at least some functionality
can be maintained. The two are not mutually exclusive, however, as rapidity is influenced by
robustness. This represents a second gap in the literature; infrastructure robustness must be studied
considering “unfavorable conditions”.
In railway systems, robustness studies have mainly been aimed at timetable management to handle
delays (including secondary delays) within the system. Yet according to the statistics from the follow
up system of the Swedish Transport administration (Trafikverket), LUPP, on total delays for the year
2015 show that more than 30% of the root cause of delays are due to non-robust infrastructures.
Accordingly, railway infrastructure robustness is now attracting the interest of both industry and
government. In Sweden, railway infrastructure robustness is becoming a pivotal area of government
concern, being reported in the yearly review, regarding the condition of the railway assets [17]. In the
Netherlands, the governmental infrastructure manager ProRail is conducting a project dedicated to
achieving a “more robust and punctual rail network”; the idea is to minimize the impact of
disturbances by making the infrastructure more robust [18]. In a recently published EU project report
“SUSTRAIL”, researchers use a fuzzy way to describe railway infrastructures robustness, noting, for
example, that “sometimes the limits of operation can be exceeded” [19]. However, the area lacks
clarity (no clear research scope, definition, etc.) and the topic is not handled in a holistic manner. This
precludes the achievement of a robust railway infrastructure able to face the multifaceted challenges
of the future transport system.
This study takes a holistic approach to railway infrastructure robustness, considering “unfavorable
conditions” in a bid to reduce “uncertainties” in railway maintenance. It proposes a new road map,
including a new definition and research framework, to guide further research.
The rest of this paper is organized as follows. Section 2 presents an overview of robustness and
mentions pivotal studies taking a continuous improvement point of view. Section 3 proposes a novel
definition of infrastructure robustness with a focus on railway systems. For clarification, it compares
relevant research concepts: robustness and traditional reliability, resilience and risk. Section 4
develops a new framework, which we term the House of Robustness Management (HORM), featuring
continuous improvement in infrastructure robustness. Section 5 explores some opportunities in and
practical concerns of the railway industry. Section 6 offers concluding remarks and suggests areas for
further research.
2
2. Robustness overview and robust infrastructure related studies
This section provides an overview of the robustness field and cites several studies of robust
infrastructure. These form the foundation of the proposed road map described in sections 3 and 4.
2.1 An overview of robustness fields
Table.1 An overview of robustness fields
Research fields
Control engineering
Statistics
Computer science
Operations research
Decision making theory
Economics
Quality
engineering/Product
development
Biology
Definitions of robustness
Insensitivity to uncertainty of a control system in input and model
assumptions [20, 21].
Insensitivity to small deviations in the assumptions [22, 23].
Fault tolerance [24, 25].
Solution that remains functional with changing operational scenario
[26].
Extent to which a system is able to maintain its function when some
aspect of the system is subject to perturbations [27].
Ability of a model to remain valid under different assumptions [28].
Small variability in a system’s function under various noise conditions
[29, 30].
Ability to maintain performance in the face of perturbation and
uncertainty, a well-established property of living systems [31, 32].
In the past 20 years, robustness studies have been conducted in various research fields, including
but not limited to: control engineering, computer science, quality engineering, operations research,
decision making theory, economics, and biology. As shown in Table.1, the definitions and research
focuses vary across research fields.
In general, robustness is used to describe the “ability” of a system (or a model) to maintain its
function (or performance) despite disturbance (incl. faults or perturbations). All robustness studies
aim to reduce uncertainties.
The approaches of robustness studies include: 1) statistics driven models from input to output (e.g.,
control engineering, statistics, computer science, operations research, decision making theory, and
economics); 2) physical characteristics driven design or optimization of a complex system or an item
(e.g., quality engineering/product development, and biology); and 3) their integration (in many
situations, both the statistics models and the physical characteristics need to be considered
simultaneously).
Considering the complexity of different systems and their special characteristics, management of
robust infrastructure requires knowledge from different research fields. To date, robustness studies
are still being developed in separate fields. None of the results from the above fields can be directly
used for infrastructure robustness.
2.2 Current status of robust infrastructure related studies
Currently, six research topics are related to infrastructure robustness: robustness of structures,
robustness of infrastructure in extreme events, robustness in risk management, robustness in
maintenance, robust road networks, and robustness of train timetables.
As mentioned above, infrastructure robustness has not yet been studied in a holistic manner;
therefore, its development requires knowledge from different research fields. In this study, we argue
infrastructure robustness requires continuous improvement. Following the continuous improvement
procedure proposed by Dr. W. Edwards Deming [33], the infrastructure robustness management
3
process can be divided into four stages: Plan (P), Do (D), Study (S), and Act (A). When the Plan, Do,
Study, Act procedure is applied to robustness, we can derive the following objectives for each stage:
 Plan: identification of the robustness requirements (incl. acceptable functionality in relation
to disturbances, likelihood of “unfavorable conditions” and the impact) and data requirements,
definition of infrastructure system, and design of the work process;
 Do: data collection, verification and analysis, execution of the robustness management
process;
 Study: construction of key performance indicators (KPIs) for robust infrastructure
management;
 Act: optimization of robust infrastructure management.
Table.2 shows the current research status of these topics, including the definitions and research
focuses of each, from a continuous improvement perspective.
Table.2 Current status of pivotal robust infrastructure related studies
Topics
P*
Robustness of
structures
Represented definitions
Ability of a structure not to be damaged by events
like fire, explosions, impact or human errors to an
extent disproportional to the original cause [34].
Robustness of
Ability to resist specified disturbances or
infrastructure in perturbations so that at least some functionality can
extreme events be maintained [13].
Robustness in
Ability to withstand a certain amount of stress with
risk management respect to the loss of function of the system [35].
Robustness in
Ability to handle uncertainty in model input [36].
maintenance
Extent to which, under pre-specified circumstances, a
Robust road
network is able to maintain the function for which it
networks
was originally intended [37].
Ability of a timetable to withstand design errors,
Robustness of
parameter variations, and changing operational
train timetables
conditions [38]
*Research focuses for Plan, Do, Study, Act are marked with “X”.
D
S
X
X
X
X
A
X
X
X
X
X
X
X
X
X
X
X
X
Present studies on the robustness of structures include solutions for structures to withstand damage
or disturbances without collapsing [34], or for robustness to be measured and evaluated in a suitable
way [39, 40]. The former aims to identify the robust requirements and make robustness management
plan (P); the latter aims to set up KPIs (S).
Robustness studies examining the effect of extreme events on infrastructure look at different types of
infrastructure and interdependent infrastructure in pre and post disaster situations. They consider how
the infrastructure can resist and absorb extreme events and adapt to post disaster circumstances from
both quantitative and qualitative aspects (P) [13, 41]. The area builds on the strengths and weaknesses
measured by, for instance, robustness, risk and reliability. In particular, these studies focus on the
development of KPIs (S).
Robustness studies of risk management for infrastructure follow the same principles as robustness
studies of extreme events. The objective is to identify risk and determine how infrastructure
robustness is influenced by it (P)[35].
Studies of robustness in maintenance look for maintenance plans able to handle the uncertainties
in predicting deterioration of infrastructure while balancing structural performance and cost; thus, the
research focus is on evaluating the accuracy of the maintenance plan (P, S) and creating optimization
decision models (A). Note that the objective is not the behavior of the infrastructure [36, 42].
4
The study of robust road networks is a well-established industrial application, covering all stages
of the PDSA process. However, there is no established process for the continuous improvement of
robustness within the area. Vulnerability is the main focus and is regarded as the antonym of
robustness; i.e., a vulnerable network is not robust and vice versa. In this application area, research
contributions mainly include the identification of vulnerability related factors and their impact (P),
data analysis (D), and vulnerability improvement (A). A network robustness index (NRI) is used to
evaluate the performance of traffic and travel time (S) [37, 43, 44].
Robustness of train-timetables is another common topic in industry. So far, such studies look for
multi-objective optimization solutions by balancing asset efficiency, robustness (delay resistance) of
a timetable, etc. Most studies are simulation-based. Although these studies cover almost the whole
PDSA process, as discussed, there is a lack of research on root causes of delays, i.e., non-robust
infrastructure in railway [38, 45, 45, 46]
As Table.2 shows: 1) in different topics, robustness has different definitions and research interests; 2)
no results from any topic can be directly applied to railway infrastructure; 3) most studies have limited
research on specified stages in a continuous improvement process, but as we argue, continuous
improvement is essential to the robust infrastructure management of railway systems.
3. Novel definition of railway infrastructure robustness
This section introduces a novel definition of infrastructure robustness. It begins by explaining the
key elements of the definition in section 3.1 and goes on to clarify it in section 3.2.
3.1 Railway infrastructure robustness
As discussed in section 2.1 and 2.2, currently, the definition of infrastructure robustness is not clear
for railway systems; in addition, the research scope and topics are vague. There is an urgent need
to take a holistic approach to infrastructure robustness of railway systems, one that considers
“unfavorable conditions” and seeks to reduce “uncertainties” in railway maintenance.
For the purpose of developing a new road map, we define the infrastructure robustness of railway
systems as the following: the ability of railway infrastructure to maintain an acceptable level of
functionality in the presence of specified disturbances for a specified time period. In the following
sections, we break this definition down into its four key elements (see Fig.1): ability to perform,
functionality at an acceptable level, specified disturbances, and specified time period.
infrastructure robustness
for railway systems
ability
acceptable
functionality
specified
distrubances
specified time period
Fig.1 Key elements of infrastructure robustness of railway systems
3.1.1 Ability
Like other robustness related definitions, those for infrastructure robustness describe some kind of
“capability” that should be evaluated in both quantitative and qualitative ways to get a comprehensive
measurement; this requires knowledge of statistics, maintenance, quality, mechanics, etc.
3.1.2 Acceptable functionality
The most basic function of railway systems is transporting people or goods from one point to
another. Different parts of the system have different contributing functions to fulfill the service needs
of transport. The basic function of the infrastructure of railway systems is to allow rolling stock to
pass safely and in a timely fashion according to plan with acceptable deviations.
5
The acceptable level of infrastructure functionality should be decided by stakeholders (incl.
infrastructure manager, railway companies) and customers (incl. railway companies, rolling stock
owners) according to existing agreements. Functionality can take various forms, e.g., train speed,
train delays or punctuality, etc.
3.1.3 Specified disturbances
In this study, disturbances refer to “unfavorable conditions” which raise uncertainties about the
ability to maintain functionality at an acceptable level. To determine the robustness of railway
infrastructure, relevant disturbances should be identified. The likelihood of their happening must be
minimal in the immediate future (if the probability is high, this becomes a reliability improvement
problem). Note that extreme events like earthquakes, floods or deliberate attracts on infrastructure
which are studied in resilience are not within the study scope. Finally, the robustness of an asset
depends on the type of disturbance; if it is robust against mechanical wear, it is not necessarily robust
against heavy rainfall.
3.1.4 Specified time period
The time period for robustness is specified but can be changed depending on the scope of the follow
up. It can refer to the winter months, the latest snowfall, or any lifetime calculation span, e.g. running
distance for wheels. With changing calculation spans, today’s specified disturbances can become
tomorrow’s required operational conditions. For instance, 10 years ago, the wagon axle load on the
iron ore line in Sweden was 22.5 tons; thus, a load of 25 tons for one axle represented a disturbance.
However, as the wagon axle load is now 30 tons, a 25-tonne axle load no longer represents a
disturbance. In other words, the requirements of infrastructure robustness can change with time.
3.2 Definition clarification
Table.3 Clarification of robustness, reliability, resilience, and risk
Considerations
Ability
Functionality
Disturbances
Robustness
Yes
Acceptable
functionality
Specified
disturbance
Specified
Time period
Railway
Yes
Infrastructure
*: a dash means it is not applicable.
Reliability
Yes
Required function
Resilience
Yes
Normal
functionality
No
Disruptions
Specified
Railway industry
e.g., EN-50126
Not specified
Transportation
infrastructure
Risk*
Considered as a
consequence
Identified
uncertainties
Not specified
In general
To clarify the novel definition of the infrastructure robustness of railway systems proposed in this
study, definitions of robustness, reliability, resilience, and risk have been compared, as shown in
Table.3. Specified definitions are following:
 Railway infrastructure robustness (section 3.1 in this paper): the ability of railway
infrastructure systems to maintain an acceptable level of functionality in the presence of
specified disturbances for a specified time period.
 Reliability [2]: the probability that an item can perform a required function under given
conditions for a given time interval.
 Resilience [13, 15]; the ability to resist, absorb and adapt to disruptions and return to normal
functionality. A resilient system has (1) reduced probability of failures; (2) reduced
consequences from failures; and (3) reduced time to recovery.
6

Risk [8]: the effect of uncertainty on objectives. Risk is often expressed in terms of a
combination of the consequences of an event (including changes in circumstances) and the
associated likelihood of occurrence.
As shown in Table.3, there are similarities but also clear differences among the above concepts.
More specifically:
 Ability: All concepts concern an attribute of an asset, except for risk, which describes the
impact and likelihood of an uncertainty.
 Functionality: Unlike robustness, reliability considers required function which normally has
been decided in the design stage. Resilience considers the ability to stay in and return to
normal functionality. Robustness includes the expectations of stakeholders in setting the
standards for acceptable functionality, making it possible for alterations over time.
 Disturbances: Unlike robustness which considers a specified disturbance, or reliability which
does not consider any disturbances, resilience considers all types of disturbances with an
emphasis on extreme events. Risk reflects uncertainties which can be positive or negative.
 Time period: Robustness and reliability have a specified time period; resilience and risk have
an unspecified time span.
 Railway Infrastructure: Unlike reliability and resilience, there is no definition of robustness
for the railway industry. The novel definition proposed in this study aims to fill this gap. Risk
has a general definition currently accepted in the railway industry.
4. Novel framework of infrastructure robustness
To improve system robustness of railway infrastructure continuously, this section develops a new
framework of robustness management. The pivotal elements of the House of Robustness Management
(HORM) are: “Roof”, “Ceiling”, “Walls”, and “Floor”.
Robustness Management Goal
Robustness Management Guidance
PLAN


Acceptable
functionality

Specified
disburbances




DO
Robustness
requirements
identification


Specified time period
Data requirements
clarification
Infrastructure criticality
identification
 RCM
 FMECA
 RAMS
 SIL
 HAZOP
 RCA
 FTA
 FRACAS
STUDY
Data collection and
verification
Robustness
Management process
Execution

KPI for robustness of
railway assets

KPI for data quality
assessment

KPI for Robustness
management resources

KPI for support system

KPI for KPI system

KPI related to LCC
ACT

Benchmarking

New Design & Renewal

Robustness Strategy
Review and Optimization
 RCM
 FMECA
 RAMS
 SIL
 HAZOP
 RCA
 FTA
 FRACAS

Resources Optimization

BPR

KPI System Optimization
Robustness
management process
construction
Support System
Fig.2 House of Robustness Management (HORM)
7
4.1 Roof of HORM
In the novel definition, the robustness of infrastructure systems can be evaluated in quantitative
and qualitative ways. In the roof of the HORM, the goals of robustness management should be set to
clearly guide the process.
First, robustness management goals will stem from the overall business goals of the organization
and support their achievement. The Swedish Transport Administration (Trafikverket) views
robustness as a pivotal part of its overall business goals, along with punctuality, capacity, usability,
safety and environment & health. The achievement of robustness will support the overall business
goals. At the same time, it will influence the goals of the other pivotal parts.
Second, robustness management goals should be hierarchically set, i.e., disaggregated down
through different levels. For instance, when the robustness goal for the entire track system on level 1
is allocated down to level 2, it can be expressed as the robustness goals of the track and switches &
crossings (S&Cs); when it moves further down to level 3, it becomes the robustness goals of rails,
fastenings, sleepers and ballast, etc., separately. This is shown in Figure 1.
Third, the goals should be clear for the working group on each level. These goals should relate to
and provide guidance for the working group’s role in, and contribution to, the total achievements of
the organization.
Fourth, they should be disseminated periodically among the stakeholders and improved gradually.
4.2 Ceiling of HORM
The ceiling is one of the basic structures of a house. This understanding carries through to the
HORM framework: the related infrastructure management guidance required to achieve robustness
management goals and guide the overall process comprises the ceiling of the HORM. This guidance
could include standards, work instructions, expert knowledge, and research reports.
Although there is no specific robustness management standard for railway infrastructure, related
standards include international or regional standards, or those requested by a specified
country/industry. For instance, reliability related standards should be considered as important
references for robustness management. These include international standards (ISO†, IEEE‡, IEC§,
etc.), European standards (CEN**, ETSI††, etc.), and railway standards (UIC‡‡, etc).
Work instructions present a sequence of steps to execute a task or activity. The Swedish Transport
administration has several handbooks for performing various maintenance actions, for example, a
handbook on how to clear snow from S&Cs [47]. These handbooks are important, not only to guide
maintenance but also for the assurance of the management of robustness.
Expert knowledge can supply valuable guidance as well. This knowledge can come from group
and/or individual experience. Sometime there is a need to consult experts with knowledge of special
cases to support the creation of robustness strategies.
Research reports may include conclusions and suggestions which could guide robustness
management. For instance, in Sweden many ideas come from research institutes (universities or
research centers) or special reports from other organizations, including labs or consultant companies.
†
ISO: International Organization for Standardization.
IEEE: Institute of Electrical and Electronics Engineers.
IEC: International Electrotechnical Commission.
** CEN: European Committee for Standardization.
††
ETSI: European Telecommunications Standards Institute.
‡‡
UIC: International Union of Railways.
‡
§
8
4.3 Walls of HORM
Following the thinking of continuous improvement developed by Edwards Deming, the four walls
of the proposed HORM are Plan (P), Do (D), Study (S), and Act (A) [33]. The use of this PDSA cycle
ensures robustness management is not only standardized but also improved gradually. The walls
follow the robustness management goals (roof) and guidance (ceiling) and are supported by
Information and Communication Technologies (ICT) systems (floor).
4.3.1 Plan
Plan is the start of the PDSA cycle. Main tasks involve robustness requirements identification, data
requirements clarification, infrastructure criticality identification, and process construction for
managing robustness.
In light of the novel definition of railway infrastructure robustness in section 3, “acceptable
functionality”, “specified disturbances”, and “specified time period” need to be decided in the Plan
stage. Data requirements should be identified according to the robustness requirements. These, in turn,
require the identification of infrastructure criticality, as different criticality may lead to different
robustness strategies. Knowledge-driven approaches can be applied to identify criticality, including
but not limited to: RCM (Reliability Centered Maintenance), FMECA (Failure Mode, Effect and
Criticality Analysis), RAMS (Reliability, Availability, Maintainability and Safety), SIL (Safety
Integrity Level), HAZOP (Hazard and Operability study), RCA (Root Cause Analysis), FTA (Fault
Tree Analysis), FRACAS (Failure Reporting, Analysis and Corrective Action System), etc. The work
process of robustness management should also be defined in the Plan stage.
4.3.2 Do
In the Do stage, the main tasks include data collection and verification, as well as execution of the
process defined in the Plan stage.
Data collection and verification represent a key element of effective robustness management.
Besides work orders, data could come from engineering design data, component test data, system test
data, operational (and experience) data from similar systems, field-tracking studies in various
environments, computer simulations, related standard and operation manuals, experience data from
similar systems, expert judgment and personal experience, warranty data, etc. Robustness is
concerned with uncertainties and unfavorable conditions that happen infrequently. However,
incomplete or unreliable data will lead to misleading results in data processing (i.e. the conditioning
and feature extraction/selection of acquired data) and decision making (i.e. the recommendations for
robustness management actions based on diagnosis and/or prognosis, and then for maintenance
prescription). Data verification ensures the data are able to evaluate the robustness of the railway
assets in the Study stage. Finally, the execution of the robustness management process needs to be
followed in this stage, so that the real situations can be assessed in the subsequent stage.
4.3.3 Study
In the Study stage, a system of key performance indicators (KPIs) should be set up to control and
monitor robustness management. The KPI system consists of five different aspects, including KPIs
for the robustness of railway assets, data quality assessment, resources, support systems, and the
entire robustness management process. Economic considerations such as the life-cycle cost (LCC)
should also be included.
Specific KPIs for measuring the robustness of assets are analyzed together with the related KPIs,
for instance, RAMS, risk, and railway-specific KPIs like punctuality and train delays. The purpose
of these KPIs is to evaluate the robustness of the assets. When evaluating data quality, KPIs for data
collection and data processing are included.
9
Robustness management resources include: technical (hardware or software); human (personnel
and competence) and organizational aspects (commitment, management style and preferences).
The effectiveness and efficiency of the support systems, including, for example, databases and
computerized maintenance management systems (CMMS), must be periodically reviewed and
controlled.
Finally, the entire robustness management process, the indicators for LCC, and the KPI system
itself must be controlled and monitored.
4.3.4 Act
The Study stage reveals gaps between the present situation and the robustness management goals.
This leads to the Act stage where improvements of performance are initiated to achieve the overall
robustness management goals. This may include setting new goals. Benchmarking, new design or
renewal, review and optimization of the robustness strategy, resource optimization, business process
re-engineering (BPR), and KPI system optimization represent some possible improvement actions in
this stage.
Benchmarking will shed light on the differences in the robustness of comparable assets in different
operational and organizational scenarios.
The strategies for achieving robustness objectives need to be reviewed and possibly optimized.
This can be done through various maintenance management techniques, for example, RCM, FMECA,
RAMS, SIL, HAZOP, RCA, FTA or FRACAS.
Resource optimization seeks optimal solutions considering multi-objective criteria. This includes
economic, technical (hardware or software), human (personnel and competency) and organizational
aspects (commitment, management style and preferences).
BPR can be used to improve processes of robustness management defined in the Plan stage.
Finally, the KPI system needs to be optimized.
All the walls together form a sequential continuous process; for example, after the Act stage, the
Plan stage is reviewed and improved. In this way, continuous improvements are ensured.
4.4 Floor of HORM
To improve the decision-making process in robustness management, data from various sources
(e.g. product, production, maintenance, and business) must be collected, integrated, fused, and
analyzed to transform them from information into knowledge. Therefore, data form the foundation or
floor of the house. Most of the maintenance related data which will support various maintenance
management techniques (for example, RCM, FMECA, RAMS, SIL, HAZOP, RCA, FTA or
FRACAS) should have been recorded and stored in various management systems, such as CMMS,
enterprise resource planning (ERP), special software for condition monitoring or other industryspecific supporting tools. Another important support area in this floor will be eMaintenance.
5. Discussion
In this article we have proposed a road map for railway infrastructure robustness, including a novel
definition and a new framework of robustness management. Our definition brings clarity to a research
area where “unfavorable conditions” may affect asset functionality, as expressed in the concepts of
reliability, resilience and risk. The proposed framework for robustness management, termed HORM,
provides a holistic view of, and a continuous improvement process for, managing the robustness of
railway infrastructure.
While the framework shows great promise, the concept of railway infrastructure robustness is
continuing to develop. Remaining challenges include the following:
10
First, compared with the research topics of reliability, resilience and risk, infrastructure robustness
requires further study from both theoretical and practical perspectives, especially for railway systems.
This will include further details and case studies of qualitative and quantitative measures;
Second, of the four key elements discussed in section 3, “accepted functionality” can influence the
identification of other elements;
Third, robustness management needs to be considered in a holistic way; HORM could be a guide
for continuous improvement, but this requires top level management to take a holistic view;
Fourth, given the quickly developing areas of ICT, determining how to collect, integrate, fuse, and
analyze data, as well as how to transform them from information into knowledge, will be an ongoing
challenge in future robustness studies.
Fifth, the reality is complex, and “unfavorable conditions” are hard to identify and predict;
insufficient knowledge or over-estimation may lead to high operational risks and costs;
Sixth, the determined robustness will necessarily be conditional, as it will depend on the identified
conditions. An asset can be robust against one disturbance, for example, snowfall, but not robust
against, for instance, heavy axle loads;
Seventh, trusted operation, interoperability with other modes and emergency preparedness should
also be considered in railway infrastructure robustness.
6. Conclusions
To adequately consider “unfavorable conditions” and to reduce “uncertainties” in railway
maintenance, this study conducts a holistic examination of railway infrastructure robustness by
developing a new road map, including a novel definition and a framework. In this study, the novel
definition of railway infrastructure robustness is proposed as: the ability of railway infrastructure to
maintain an acceptable level of functionality in the presence of specified disturbances for a specified
time period. This definition is useful, as it clarifies the research boundaries between robustness and
reliability, resilience, and risk. The study develops a ground-breaking framework, named house of
robustness management (HORM), based on continuous improvement to support infrastructure
robustness management in railway; HORM consists of robustness management goals and guidance,
a continuous improvement process, and support systems. Finally, the study considers the
opportunities and challenges of applying the road map to railway infrastructure and provides
guidelines for other research into robust infrastructure in different industries.
Acknowledgement
The authors thank Luleå Railway Research Centre (Järnvägstekniskt Centrum, Sweden), Sweco Rail
AB and The Swedish Transport Administration (Trafikverket) for initiating the research study and
providing financial support. We also thank Vivianne Karlsson, Stefan Jonsson, Anders Gustafsson,
Professor Uday Kumar, Professor Per-Olof Larsson-Kråik, Dr. Christer Stenström and Dr. Stephen
Famurewa (Luleå University of Technology) for their support and valuable discussions on this study.
11
References
[1] European Commission. Horizon 2020, 11.Smart, green and integrated transport, work programme 20162017. Brussels, European Commission, 2015.
[2] CEN. EN 50126: Railway Specifications – The Specification and Demonstration of Reliability,
Availability, Maintainability and Safety (RAMS). Brussels, European Committee for Standardization,
1999.
[3] Barabady J. Production assurance: concept, implementation and improvement. Ph.D. thesis, Luleå
University of Technology, 2007.
[4] Patra A. P. Maintenance decision support models for railway infrastructure using RAMS & LCC analyses.
Ph.D. thesis, Luleå University of Technology, 2009.
[5] Luleå University of Technology. Robust and Reliable Transport System - a strategic agenda for resource
pooling, collaboration, innovation and research. Luleå University of Technology, Luleå, Sweden, 2015.
Available from: http://www.ltu.se/cms_fs/1.144835!/file/RTS_2015_webb.pdf
[6] Kobbacy K. A. H. and Murthy D. P. Complex system maintenance handbook. Springer Science &
Business Media, London, 2008.
[7] Ekberg A. and Paulsson B. INNOTRACK (Innovative Track Systems) concluding technical report.
International Union of Railways (UIC), Paris, 2010.
[8] ISO. ISO 55000: Asset management - Overview, principles and terminology. Geneva, International
Organization for Standardization (ISO), 2014.
[9] Lewis R. and Olofsson U. Wheel-rail interface handbook. Woodhead publishing, Boca Raton, FL, USA,
2009.
[10] Hollnagel E., Woods D. D. and Leveson N. C. Resilience engineering concepts and precepts. Ashgate,
Aldershot, England, 2006.
[11] Steen R. and Aven T. A risk perspective suitable for resilience engineering. Safety Science. 2011; 49(2):
292-297.
[12] Kaaniche M., Lollini P., Bondavalli A. and Kanoun K., Modeling the resilience of large and evolving
systems. International Journal of Performability Engineering. 2008; 4(2): 153-168.
[13] Bruneau M., Chang S. E., Eguchi R. T., Lee G. C., O'Rourke T. D., Reinhorn A. M., Shinozuka M.,
Tierney K., Wallace W. A. and von Winterfeldt D., A framework to quantitatively assess and enhance
the seismic resilience of communities. Earthquake Spectra. 2003; 19(4): 733-752.
[14] Adger W. N., Hughes T. P., Folke C., Carpenter S. R. and Rockström J. Social-ecological resilience to
coastal disasters. Science. 2005; 309 (5737): 1036-1039.
[15] Faturechi R. and Miller-Hooks E. Measuring the Performance of Transportation Infrastructure Systems
in Disasters: A Comprehensive Review. Journal of Infrastructure Systems. 2014; 21(1): 1-15.
[16] Stoop J. A. and Van Kleef E. A. Reliable or Resilient: Recovery from the Unanticipated. International
Journal of Performability Engineering. 2015; 11 (2): 169-179.
[17] The Swedish Transport Administration (Trafikverket). Yearly review 2014. Borlänge, Sweden, The
Swedish
Transport
Administration
(Trafikverket),
2015.
Available
from:
https://online4.ineko.se/trafikverket/Product/Category/12347.
[18] ProRail, Prorail, Dutch railway infrastructure manager. [internet]. 2016. Available from http://prorail.nl.
[19] Beagles A.SUSTRAIL (The sustainable freight railway) concluding technical report. International union
of railways (UIC), Paris, 2015.
[20] Zhou K. and Doyle J. C. Essentials of robust control. Prentice hall, Upper Saddle River, N.J., 1998.
[21] Monje C. A., Vinagre B. M., Feliu V. and Chen Y. Tuning and auto-tuning of fractional order controllers
for industry applications. Control Engineering in Practice. 2008; 16(7): 798-812.
[22] Huber P. J. Robust statistics. John Wiley & Sons, Hoboken, N.J., 2011.
[23] Hubert M., Rousseeuw P. J. and van Aelst S. High-Breakdown Robust Multivariate Methods. Statistical
Science. 2008; 23 (1): 92-119.
[24] Rabinowitz M. and Spilker Jr. J. J. A new positioning system using television synchronization signals.
IEEE Transactions in Broadcasting. 2005; 51(1): 51-61.
12
[25] Bhowmik S., Tomsovic K. and Bose A. Communication Models for Third Party Load Frequency Control.
IEEE Transactions in Power Systems. 2004; 19(1): 543-548.
[26] Burdett R.L., Kozan E. and Strickland C. A practical approach for identifying expected solution
performance and robustness in Operations Research applications. ASOR Bulletin. 2012; 31 (2): 1-28.
[27] Walsh M. M., Einstein E. H. and Gluck K. A., A quantification of robustness. Journal of Applied Research
in Memory and Cognition. 2013; 2(3): 137-148.
[28] Kuorikoski J., Lehtinen A. and Marchionni C. Economic modelling as robustness analysis. British Journal
of Philosophical Science. 2010; 61(3): 541-567.
[29] ISO. Applications of statistical and related methods to new technology and product development process
-- Robust parameter design (RPD). Geneva, International Organization for Standardization, 2014.
[30] Montgomery D. C. Design and analysis of experiments. John Wiley & Sons, 2008.
[31] Stelling J., Sauer U., Szallasi Z., Doyle III F. J. and Doyle J., Robustness of cellular functions. Cell. 2004;
118(6): 675-685.
[32] Kitano H. Towards a theory of biological robustness. Molecular systems biology. 2007; 3(137): 1-7.
[33] Deming W. E. Out of the Crisis. Massachusetts Institute of Technology, Cambridge, 1991.
[34] CEN. EN Eurocode 1: Basis of design and actions on structures. Geneva, European Committee for
Standardisation, 1994.
[35] Hokstad P., Utne I. B. and Vatn J .(ed). Risk and Interdependencies in Critical Infrastructures: A
Guideline for Analysis. Springer-Verlag, London, 2012.
[36] Kuhn K. D. and Madanat S. M. Robust maintenance policies for Markovian systems under model
uncertainty. Computer-Aided Civil Infrastructure Engineering. 2006; 21(3): 171-178.
[37] Snelder M., Van Zuylen H. and Immers L. A framework for robustness analysis of road networks for
short term variations in supply. Transportation Research Part A: Policy and Practice. 2012; 46(5): 828842.
[38] ON-TIME Consortium, Methods and algorithms for the development of robust and resilient timetables,
Deliverable D3.1. [internet]. 2015. Available from http://www.ontime-project.eu/deliverables.aspx.
[39] Baker J. W., Schubert M. and Faber M. H. On the assessment of robustness. Structural Safety. 2008;
30(3): 253-267.
[40] Starossek U. and Haberland M. Approaches to measures of structural robustness. Structure and
Infrastructure Engineering. 2011; 7(7-8): 625-631.
[41] Bocchini P., Frangopol D. M. Ummenhofer T. and Zinke T. Resilience and sustainability of civil
infrastructure: Toward a unified approach. Journal of Infrastructure Systems. 2013; 20 (2):1-16.
[42] Ok S., Lee S. and Park W. Robust multi-objective maintenance planning of deteriorating bridges against
uncertainty in performance model. Advances in Engineering Software. 2013; 65: 32-42.
[43] Scott D. M., Novak D. C., Aultman-Hall L. and Guo F. Network Robustness Index: A new method for
identifying critical links and evaluating the performance of transportation networks. Journal of
Transportation Geography. 2006; 14 (3): 215-227.
[44] Mattsson L. and Jenelius E. Vulnerability and resilience of transport systems: A discussion of recent
research. Transportation Research Part A. 2015; 81: 16-34.
[45] Abril M., Barber F., Ingolotti L., Salido M. A., Tormos P. and Lova A. An assessment of railway capacity.
Transportation Research Part E: Logistics and Transportation Review. 2008; 44 (5): 774-806.
[46] Fischetti M., Salvagnin D. and Zanette A. Fast approaches to improve the robustness of a railway
timetable. Transportation Science. 2009; 43 (3): 321-335.
[47] The Swedish Transport Administration (Trafikverket), TDOK 2013:0655 version 3, Spårväxel
vinterhandbok (Winter handbook S&Cs in Swedish). [internet]. 2016. Available from
http://trvdokument.trafikverket.se/.
13
PAPER B
A novel approach for measuring Railway
Infrastructure Robustness
Authors:
Norrbin Per, Lin Jing & Parida Aditya
Submitted to journal.
A novel approach for measuring railway
infrastructure robustness
Per Norrbin 1, 2 , Jing Lin 1 , Aditya Parida 1
1. Division of Operation and Maintenance Engineering, Luleå University of Technology,
97187, Luleå, Sweden
2. Sweco Rail AB, 97324, Luleå, Sweden
Abstract: To sufficiently consider unfavorable conditions and reduce uncertainty in
maintenance, railway infrastructure robustness was given a new definition and framework by
Norrbin et al. in 2016 [1]. Building on this work, the present study proposes a novel approach
to quantitatively measure railway infrastructure robustness. In this approach, acceptable
functionality and specified disturbance are considered together with various risk preferences
for railway maintenance strategies. The study provides a numerical example to illustrate the
application of the suggested approach using a new parameter “Mean Time between Robustness
(MTBR)”. The results show the approach is feasible for railway infrastructure and flexible
enough to handle complex situations.
Keywords: maintenance; railway infrastructure; robustness; measurement; continuous
improvement
1. Introduction
According to the European program Horizon 2020, challenges in transport infrastructure
include: 1) the need to make infrastructure more resilient to keep pace with increasing mobility
needs; 2) the need to reduce the impact of infrastructure on the environment; and 3) the need to
deal with declining resources to maintain and upgrade transport infrastructure. New
maintenance approaches must be developed to handle these issues, as the present ones are
inadequate [2].
A traditional maintenance approach to railway infrastructure [3] is RAMS (Reliability,
Availability, Maintainability and Safety). More recently, infrastructure robustness is attracting
the attention of researchers and engineers, to improve the ability of an infrastructure to
withstand disturbances caused by various uncertainties. This is a part of resilience studies. The
appeal of robustness stems from the ability to consider “unfavorable conditions” and reduce
uncertainties in railway infrastructure maintenance [4, 5]. Norrbin, et al. [1] have promoted a
new road map for railway infrastructure robustness including a novel definition and a unique
framework. However, the work [1] is focused on management perspectives and employs
qualitative analysis.
At this point, there is no specific quantitative measure for railway infrastructure robustness. The
current measures for infrastructure robustness are general or area specific and do not provide
or use specific information on railway infrastructure. Further, the various operational scenarios
1
and requirements during the lifetime of an asset together with its changing condition are not
considered sufficiently in present measurements.
To fill the research gap and to continue previous work [1], the present study proposes a novel
approach to quantitatively measure railway infrastructure. This approach will allow the railway
industry to quantify, follow up on, and verify robustness goals, and to achieve both continuous
improvement and system assurance. The approach considers “acceptable functionality” and
“specified disturbance”, together with various risk preferences for railway maintenance strategy
making.
The organization of the paper is as follows. Section 2 presents an overview of the measurement
of infrastructure robustness from qualitative and quantitative perspectives. Section 3 proposes
a novel approach to measure robustness by introducing a new parameter, Mean Time between
Robustness (MTBR). Section 4 discusses the findings, and Section 5 offers conclusions and
comments.
2. Infrastructure robustness measurement: an
overview
Infrastructure robustness related studies have mainly focused on several pivotal areas:
structural robustness, robustness of infrastructure in extreme events, robustness in risk
management, robustness in maintenance, robust networks, and robustness of train timetables
[1]. To this point, the measurement of infrastructure robustness has not been studied in a holistic
manner.
In structural science, robustness is generally considered the ability of a structure to resist
abnormal circumstances and events without disproportional failure and progressive collapse.
Measurements of structural robustness mainly consider the impact of a damaging event, such
as an explosion, impact, fire, human error or structural deterioration [6-8]. Normally, the
measurement is the ratio of the damaged to the undamaged structure. Some measurements
evaluate different properties of structures, including: structural strength [9, 10], system stiffness
[11], how failure can progress to undamaged parts of the structure [11, 12], and risk perspective
[9, 13]. Baker et al. have proposed a general measurement of robustness based on risk, the most
highly cited measure of robustness for structures. They define robustness as an index, a ratio of
direct and indirect risk. Indirect risk is damage not directly caused by a damaging event; the
higher the indirect risk, the less robust the system [14].
Infrastructure robustness in extreme events is generally referred to as the ability to withstand a
given extreme event and still deliver a service, often measured by the residual functionality
level after the occurrence of the event [15]. In this area, robustness is seen as combining
resilience and rapidity [16]. Two special applications are earthquake hazards science and water
management science. In earthquake hazards science, a typical definition of robustness is the
ability to resist specified disturbances or perturbations so that at least some functionality can be
maintained [17]. Bruneau et al. [17] describe 24 qualitative and quantitative measures used to
determine the extent of damage after an earthquake. However, the focus is still on effects after
a damaging event; one example is how many houses have electricity after an earthquake.
2
In water management science, robustness is defined as the ability of a system to remain
functioning under disturbance [18]. The focus of measurement is on how the increasing
uncertainty of climate change will affect water systems. Measurement is based on a response
curve for two dimensions: system response and disturbance magnitude. Another quantitative
measurement is proposed by Moody and Brown, but the focus remains on climate change and
how various climate scenarios affect the performance of the water system. It should also be
pointed out that the interdependency of robustness and resilience for water management differ
from the other areas [19].
Robustness in risk management is viewed as the ability to withstand a certain amount of stress
without loss of function of the system [20]. Robustness studies of risk management for
infrastructure follow the same principles as robustness studies of extreme events and are closely
related to resilience [17]. The objective is to identify risk and determine how infrastructure
robustness is influenced by it. Measurement mainly concerns structural risk and remaining
durability [14]. Steen and Aven [5] propose a proactive way to look at risk through resilience
and uncertainty using vulnerability to quantify robustness.
In maintenance, studies use robust optimization to optimize maintenance plans by handling
uncertainty in model input. The uncertainties in predicting deterioration of infrastructure are
optimized while structural performance and cost are balanced; thus, the research focus is on
evaluating the accuracy of the maintenance plan and creating optimization decision models [2123]. Robustness measurements are used to evaluate how well the models of deterioration are
measures of uncertainty and not measures of infrastructure ability.
Robustness in networks refers to the extent to which, under pre-specified circumstances, a
network is able to maintain the function for which it was originally intended [24]. Within
network theory, the measures mainly focus on the network topology, for example, how the
nodes and links in the system are connected and how the interconnection upholds the
functionality of the network. A big area of concern is robustness to attacks on networks, with
vulnerability seen as the opposite of robustness. Robustness is measured for networks in general
[24, 25], as well as transportation networks [25, 26], power grids [27] and road transportation
networks [26, 28]. In general, all measures consider the topology of the network, how the nodes
and links are connected and support each other so the robustness of the network system can be
enhanced.
Robustness of train timetables refers to the property of timetables to withstand design errors,
parameter variations, and changing operational conditions [29]. Studies look for multi-objective
optimization solutions by balancing asset efficiency, timetable robustness (delay resistance) etc.
The robustness measurements concern margins (i.e., extra projects etc.) inserted into the
timetables, where they should be placed, and how their addition affects the costs generated by
the extra travelling time. Another concern is the effect of disruptions and disturbances in the
system [30-33]. The focus is on the timetable characteristics and how buffer and headway times
in the system can be optimized while keeping the transportation system effective.
The present measures for infrastructure robustness are general, with little specific information
on the actual railway infrastructure. For example, the varying operational scenarios and
3
requirements of an asset during its lifetime, together with the varying condition of the asset, are
not considered in the current measurements for managing railway assets. In the railway, RAMS
parameters are, to a large extent, decided in the design stage of an asset.
As the above review suggests, because definitions of robustness differ across research areas,
the measurements are different as well. Unfortunately, none is directly applicable to railway
infrastructure. Recently, however, Norrbin et al. [1] proposed a new definition for railway
infrastructure robustness and suggested a new framework for its management. The framework
allows robustness to be measured in a qualitative way, but a quantitative measurement must be
developed to support continuous improvement.
3. Measurement of railway infrastructure robustness
This section presents Per Norrbin et al.’s [1] definition of railway infrastructure robustness,
upon which the proposed measurement approach is built. It introduces the suggested
quantitative measurement of robustness and provides an example to show how the approach
works.
3.1 Railway infrastructure robustness
According to Norrbin et al. [1] railway infrastructure robustness refers to the ability of railway
infrastructure to maintain an acceptable level of functionality in the presence of specified
disturbances for a specified time period. This definition can be broke down into four key
elements (see Fig.1): ability to perform, functionality at an acceptable level, specified
disturbances, and specified time period.
infrastructure robustness
for railway systems
Ability to perform
acceptable
functionality
specified
disturbances
specified time period
Figure 1. Key elements of infrastructure robustness of railway systems
3.1.1 Ability to perform
Like other robustness related definitions, those for infrastructure robustness describe some kind
of “capability” that should be evaluated in both quantitative and qualitative ways to get a
comprehensive measurement; this requires knowledge of statistics, maintenance, quality,
mechanics, etc.
3.1.2 Acceptable functionality
The most basic function of railway systems is transporting people or goods from one point to
another. Different parts of the system have different contributing functions to fulfill the service
needs of transport. The basic function of the infrastructure of railway systems is to allow rolling
stock to pass safely and in a timely fashion according to plan with acceptable deviations. The
acceptable level of infrastructure functionality should be decided by stakeholders (incl.
4
infrastructure manager, railway companies) and customers (incl. railway companies, rolling
stock owners) according to existing agreements. Functionality can take various forms, e.g., train
speed, train delays or punctuality, etc.
3.1.3 Specified disturbances
In this study, disturbances refer to “unfavorable conditions” which raise uncertainty about the
ability to maintain functionality at an acceptable level. To determine the robustness of railway
infrastructure, relevant disturbances should be identified. The likelihood of their happening
must be minimal in the immediate future (if the probability is high, this becomes a reliability
improvement problem). Note that extreme events like earthquakes, floods or deliberate attacks
on infrastructure which are studied in resilience are not within the study’s scope. Finally, the
robustness of an asset depends on the type of disturbance; if it is robust against mechanical
wear, it is not necessarily robust against heavy rainfall.
3.1.4 Specified time period
The time period for robustness is specified but can be changed depending on the scope of the
follow up. It can refer to winter months, the latest snowfall, or any lifetime calculation span,
e.g. running distance for wheels. With changing calculation spans, today’s specified
disturbances can become tomorrow’s required operational conditions. For instance, 10 years
ago, the wagon axle load on the iron ore line in Sweden was 22.5 tonnes; thus, a load of 25
tonnes for one axle represented a disturbance. However, as the wagon axle load is now 30
tonnes, a 25-tonne axle load no longer represents a disturbance. In other words, the
requirements of infrastructure robustness can change with time.
3.2 A new quantitative measurement
The new definition of railway infrastructure robustness is the following: “The ability of railway
infrastructure to maintain an acceptable level of functionality in the presence of specified
disturbances for a specified time period.” It can be expressed mathematically as:
𝑅𝑟𝑜𝑏 (𝑡) = 𝑃(𝑇 > 𝑡|𝑫, 𝑨, 𝝎)
where 𝑫 refers to the data set of specified disturbances, 𝑨 refers to the corresponding
acceptable functionality, ω refers to the data set of weights of risk preference, and ∑ 𝛚 = 1 .
Robustness differs from reliability and maintainability in two ways: first, D and A are exclusive
to robustness; second, the threshold for failure in reliability and maintainability is normally
fixed and implemented in the design phase, whereas for robustness, a differentiation of the
acceptable level of functionality depends on the disturbance scenario and condition of the
evaluated infrastructure.
Table 1 gives an example of a data set of 𝑫 and 𝑨.
Table.1 Example of data set of 𝐷 and 𝐴
𝑫
𝑨
𝝎
𝐷0
𝐴0
𝜔0
𝐷1
𝐴1
𝜔1
𝐷2
𝐴2
𝜔2
5
𝐷3
𝐴3
𝜔3
𝐷𝑢
𝐴𝑢
𝜔𝑢
The data set of 𝐷 and 𝐴 from table 1 are explained in the following bullet points:







If only 𝐷0 is considered, it becomes a normal dependability problem, and 𝜔0 = 1;
Extreme events represented by 𝐷𝑢 are only listed in the table but not considered, as
these represent a resilience problem;
Boundaries between 𝐷0 and others can be changed; therefore, reliability, robustness,
and resilience considerations could be changed;
Mean Time between Failure (MTBF) and Mean Time to Repair (MTTR) need to be
considered for 𝐷1 , 𝐷2 , 𝐷3 , based on the identification of 𝐴 ; therefore, Mean Time
between Robustness (MTBR) should be developed and calculated according to 𝐷 and
𝐴;
𝐷, A, and 𝜔 can be changed both numerically and depending on asset type, the status of
the infrastructure, seasonal variations, amount of traffic, maintenance strategy,
condition of the infrastructure asset, maintenance budget, etc.;
Each disturbance event could be treated as a special reliability problem or reliability +
maintainability problem.
𝑅𝑟𝑜𝑏 = ∑𝑖=0,1,2,3 𝜔𝑖 𝑅𝐷𝑖 ,𝐴𝑖 = 𝜔0 𝑅𝐷0 ,𝐴0 + 𝜔1 𝑅𝐷1,𝐴1 + 𝜔2 𝑅𝐷2,𝐴2 + 𝜔3 𝑅𝐷3,𝐴3
3.3
An example of a robustness evaluation
As example an enhancement of the snow and ice protection equipment for switches &
crossings, S&Cs will be used.



Two solutions, S1 and S2. S1: normal snow and ice protection equipment for S&Cs
and S2: new snow and ice protection equipment for S&Cs;
𝑋 = 𝐷, A, or 𝜔;
Three disturbance scenarios are identified. Undisturbed normal scenarios are marked 𝑋0
For the last scenario, 𝑋𝑢 represents uncertainty and uncontrolled factors. These can be
extreme situations or unexpected situations with very long delay times. The level of
uncertainty will vary and will be decided based on the track section or asset type. This
uncertainty can be chosen to be accepted to a certain degree.
This example describes the robustness evaluation of S1 and S2; S1 represents normal snow and
ice protection equipment for S&Cs, and S2 represents new snow and ice protection equipment.
Two UIC 60E S&Cs with a moveable frog, both shown in picture 1, are selected for evaluation.
S1 consists of wooden boards placed on the outside of the rails for selected parts of the S&C to
prevent snow from packing in and around it, causing the S&C to malfunction. In addition, a
heating system heats up the rails to melt the snow; snow can also be cleared manually.
S2, the new and upgraded version, consists of the normal wooden boards but also has a rubber
coating and polyurethane insulation underneath. The edges of the snow protection equipment
are covered to prevent snow from blowing in from the sides and to preclude heat convection.
These new features aim to increase the robustness of the S&Cs in winter and to increase the
energy efficiency by preserving the heat in the rail.
6
Figure 2, Pictures of the two versions of snow and ice protection of S&Cs, S1 and S2.
Table 2 demonstrates the robustness measure data is only for illustration purposes.
Table 2 Robustness evaluation of S1 and S2
𝑫 : Snowfall
(mm/day)
𝑨: delay time
(hour)
𝝎:
∑𝑖=0,1,2,3 𝜔𝑖 =
1
𝑅𝑟𝑜𝑏
MTBRMTBF
MTBRMTTR
𝑅𝑟𝑜𝑏 S1
𝑅𝑟𝑜𝑏 S2



𝐷0 :
less than 10mm
𝐴0 : no time
delay, <5min
𝜔0 = 0.4
𝐷1 :
[10 – 30)mm
𝐴1 : less than
1 hour
𝜔1= 0.3
𝑅𝑟𝑜𝑏 0
𝐷2 :
[30 - 50)mm
𝐴2 :
1-3hours
𝐷3 :
[50 - 100)mm
𝐴3 :
3-12hours
𝜔2 =0.2
𝜔3 =0.1
𝑅𝑟𝑜𝑏 2
𝑅𝑟𝑜𝑏 1
𝐷𝑢 :
≤100mm
𝐴𝑢 :
>12hours
/
𝑅𝑟𝑜𝑏 3
consider MTBF
consider MTBR1
consider
MTBR2
MTBR=MTBF
MTBF during
𝐷1 : 36hours
MTBF during MTBF during
𝐷2 : 48hours
𝐷3 : 24hours
/
MTTR, t<5min
< 2h
<4h
<8h
/
0.68
0.7
0.85
0.82
0.95
0.98
0.98
0.99
𝑅𝑟𝑜𝑏 𝑆1=0.4*0.68+0.3*0.85+0.2*0.95+0.1*0.98 = 0.815
𝑅𝑟𝑜𝑏 𝑆2=0.4*0.7+0.3*0.82+0.2*0.98+0.1*0.99 = 0.821
Therefore, 𝑅𝑟𝑜𝑏 𝑆2 is more robust than 𝑅𝑟𝑜𝑏 𝑆1.
7
consider
MTBR3
0.02
0.01
4. Discussion
This measurement of railway infrastructure robustness is specific for the railway and can be
used for either scenarios or time. The acceptable functionality is specific for railway systems,
and possible disturbances are also specific for railway infrastructure. However, the
measurement is flexible enough to evaluate the robustness for different disturbances,
requirements and time periods.
The main criterion is acceptable functionality, 𝑨 This can be achieved by conducting a large
number of maintenance activities, but this approach will not enhance the robustness of the
infrastructure. Therefore, MTBF and MTTR are important determinants of the appropriate
criteria. For example, in scenario, 𝐷3 , there will be a very heavy snowfall, but delays should
not exceed 3-12 hours. The desired MTBF states that despite the snow, maintenance should not
be required on the S&C for 24 hours. If maintenance is needed, the S&C is not considered
robust. For the same scenario, MTTR considers a repair time of more than 8h should not be
required.
For current methods of asset management, the requirements of infrastructure assets with respect
to level of service are mostly set in the design stage and apply throughout the asset’s service
life for all weather conditions and seasons and for every operational scenario. However, the
present approaches do not consider possible changes in the operational scenario of an asset over
its entire service life. Nor do they consider the changing condition of infrastructure assets in
different seasons of the year or over the life time. This is especially noticeable for a long lasting
asset like railway infrastructure. This is, to some extent, compensated for by the maintenance
strategy and maintenance actions. Still the demands that can be put on a newly installed track
system are not the same as those after 10 years of service. There is also a possibility of minor
or bigger changes in customer demand. Therefore, this robustness measure uses a differentiated
acceptable functionality. The actual operational scenario will put a certain amount of stress on
the infrastructure, making it more or less likely to perform the acceptable function. The
infrastructure robustness measure, or the acceptable functionality, then, will depend on the
actual operational scenario and provide the possibility for differentiated and more realistic
service levels for passengers and train operators.
This robustness measure can be used to evaluate the current condition of the asset depending
on its life cycle phase. In this case, the disturbance could be a degraded state due to, for example,
budget constraints. The total robustness and, hence, the performance demands of the evaluated
infrastructure can then be communicated to stakeholders. This will set realistic expectations,
which are quantified and even put into contracts. This could be an important measure to prevent
the loss of goodwill due to ageing assets in parts of the infrastructure system.
The measure should also be able to aggregate for stations and track sections, making it more
related to operational performance.
To this point, the data is used as illustration only. Future research plans include conducting a
case study to verify the results using data from the databases at the Swedish Transport
Administration, Trafikverket.
The asset types and disturbances that are evaluated for robustness will be decided depending
on their criticality. The asset types, operational scenarios or the criteria for criticality will be
8
specific for each infrastructure manager. The decision on asset type and typical disturbances is
important for the measure to be applicable and useable. The use of MTBF for scenarios other
than 𝐷0 will be challenging, as it is necessary to define the time period for a similar disturbance
scenario. At present, these data are the hardest to obtain.
5. Conclusion
This study describes a novel quantitative method to evaluate railway infrastructure robustness.
The proposed measurement can provide very good support for an infrastructure manager. The
keys to this robustness measurement are: the differentiation of the disturbance, the acceptable
functionality depending on the disturbance, and the weighting of the disturbance. The measure
provides new support for a railway infrastructure manager in two ways. Firstly, the measure of
robustness will consider important operational scenarios where the infrastructure manager
needs good knowledge of the performance of the assets. Secondly, the differentiated
requirements for robustness will provide more dynamic evaluations of the infrastructure assets
than the current approaches. Both will reduce the uncertainties related to railway infrastructure.
Reference
[1] P. Norrbin, J. Lin and A. Parida, Infrastructure Robustness for Railway Systems.
International Journal of Performability Engineering. 2016.
[2] EC, Horizon 2020, 11.Smart, green and integrated transport, work programme 2016-2017.
Brussels, European Commission, 2015, .
[3] CEN, EN 50126: Railway Specifications – The Specification and Demonstration of
Reliability, Availability, Maintainability and Safety (RAMS),1999.
[4] E. Hollnagel, D.D. Woods and N.C. Leveson, Resilience engineering concepts and precepts.
Aldershot, England ; Burlington, VT : Ashgate, Aldershot, England ; Burlington, VT;
Aldershot, 2006.
[5] R. Steen and T. Aven, A risk perspective suitable for resilience engineering. Saf.Sci. 2011;
49 (2): 292-297.
[6] CEN, EN-1991-1-7: Structures- Part 1-7: General actions - Accidental actions. Brussels,
European Committee for Standardisation, 2006.
[7] U. Starossek and M. Haberland, Disproportionate collapse: terminology and procedures.
J.Perform.Constr.Facil. 2010; 24 (6): 519-528.
[8] D. Saydam and D.M. Frangopol, Time-dependent performance indicators of damaged
bridge superstructures. Eng.Struct. 2011; 33 (9): 2458-2471.
[9] M.A. Maes, K.E. Fritzsons and S. Glowienka, Structural robustness in the light of risk and
consequence analysis. Struct.Eng.Int. 2006; 16 (2): 101-107.
[10] ISO, 19902. 2008-07. Petroleum and natural gas industries–Fixed steel offshore
structures”. International Organization for Standardization. 2007.
9
[11] U. Starossek and M. Haberland, Approaches to measures of structural robustness.
Struct.Infrastructure Eng. 2011; 7 (7-8): 625-631.
[12] J. Agarwal, D. Blockley and N. Woodman, Vulnerability of structural systems. Struct.Saf.
2003; 25 (3): 263-286.
[13] B.R. Ellingwood and D.O. Dusenberry, Building design for abnormal loads and
progressive collapse. Comput.-Aided Civ.Infrastruct.Eng. 2005; 20 (3): 194-205.
[14] J.W. Baker, M. Schubert and M.H. Faber, On the assessment of robustness. Struct.Saf.
2008; 30 (3): 253-267.
[15] R. Faturechi and E. Miller-Hooks, Measuring the Performance of Transportation
Infrastructure Systems in Disasters: A Comprehensive Review. J. Infrastruct. Syst. 2014;
21 (1): 04014025.
[16] T. McDaniels, S. Chang, D. Cole, J. Mikawoz and H. Longstaff, Fostering resilience to
extreme events within infrastructure systems: Characterizing decision contexts for
mitigation and adaptation. Global Environ.Change. 2008; 18 (2): 310-318.
[17] M. Bruneau, S.E. Chang, R.T. Eguchi, G.C. Lee, T.D. O'Rourke, A.M. Reinhorn, M.
Shinozuka, K. Tierney, W.A. Wallace and D. von Winterfeldt, A framework to
quantitatively assess and enhance the seismic resilience of communities. Earthq. Spec.
2003; 19 (4): 733-752.
[18] M.J. Mens, F. Klijn, K.M. de Bruijn and E. van Beek, The meaning of system robustness
for flood risk management. Environ.Sci.& Policy. 2011; 14 (8): 1121-1131.
[19] P. Moody and C. Brown, Robustness indicators for evaluation under climate change:
Application to the upper Great Lakes. Water Resour.Res. 2013; 49 (6): 3576-3588.
[20] P. Hokstad, I.B. Utne and J. Vatn (ed), . Risk and Interdependencies in Critical
Infrastructures: A Guideline for Analysis. Springer, 2012.
[21] K.D. Kuhn and S.M. Madanat, Robust maintenance policies for Markovian systems under
model uncertainty. Comput.-Aided Civ.Infrastruct.Eng. 2006; 21 (3): 171-178.
[22] K.D. Kuhn and S.M. Madanat, Model uncertainty and the management of a system of
infrastructure facilities. Transp.Res.Part C Emerg.Technol. 2005; 13 (5-6): 391-404.
[23] S. Ok, S. Lee and W. Park, Robust multi-objective maintenance planning of deteriorating
bridges against uncertainty in performance model. Adv.Eng.Software. 2013; 65 32-42.
[24] P. Holme, B.J. Kim, C.N. Yoon and S.K. Han, Attack vulnerability of complex networks.
Physical Review E. 2002; 65 (5): 056109.
[25] C.M. Schneider, A.A. Moreira, J.S. Andrade Jr., S. Havlin and H.J. Herrmann, Mitigation
of malicious attacks on networks. Proc.Natl.Acad.Sci.U.S.A. 2011; 108 (10): 3838-3841.
10
[26] D.M. Scott, D.C. Novak, L. Aultman-Hall and F. Guo, Network Robustness Index: A new
method for identifying critical links and evaluating the performance of transportation
networks. J.Transp.Geogr. 2006; 14 (3): 215-227.
[27] J. Wang and L. Rong, Robustness of the western United States power grid under edge
attack strategies due to cascading failures. Saf.Sci. 2011; 49 (6): 807-812.
[28] A. Nagurney and Q. Qiang, Robustness of transportation networks subject to degradable
links. EPL (Europhysics Letters). 2007; 80 (6): 68001.
[29] ON-TIME Consortium, Methods and algorithms for the development of robust and
resilient timetables, Deliverable D3.1. [internet]. 2015.
[30] R.M. Goverde and I. Hansen, Performance indicators for railway timetables.Intelligent
Rail Transportation (ICIRT), 2013 IEEE International Conference on, 2013.
[31] L. Kroon, D. Huisman, E. Abbink, P.-. Fioole, M. Fischetti, G. Maróti, A. Schrijver, A.
Steenbeek and R. Ybema, The new Dutch timetable: The OR revolution. Interfaces. 2009;
39 (1): 6-17.
[32] M. Fischetti, D. Salvagnin and A. Zanette, Fast approaches to improve the robustness of
a railway timetable. Transp Sci. 2009; 43 (3): 321-335.
[33] M. Abril, F. Barber, L. Ingolotti, M.A. Salido, P. Tormos and A. Lova, An assessment of
railway capacity. Transportation Research Part E: Logistics and Transportation Review.
2008; 44 (5): 774-806.
11
PAPER C
Energy efficiency optimization for railway
switches & crossings: a case study in Sweden
Authors:
Norrbin Per, Lin Jing & Parida Aditya
Accepted for the proceedings of:
The World Congress of Railway Research, 30 of May-2 of June 2016, Milano, Italy.
Energy efficiency optimization for railway
switches & crossings: a case study in Sweden
1
Per Norrbin 1, 2, * , Jing Lin , Aditya Parida
1
1
Luleå University of Technology, Luleå, Sweden
2
Sweco Rail AB, Luleå, Sweden
*Contact: [email protected] *
Abstract
With increasing environmental concerns and mounting financial and legislative pressures on the railway
industry, monitoring and optimizing energy consumption are becoming crucial. The northernmost track in
Sweden, the Iron Ore Line (Malmbanan), is operating in a sub-arctic climate. Effective snow and ice
protection ensures the successful operation of this line, especially the railway switches and crossings
(S&Cs). One strategy is to remove snow and ice with electrical heating; however, the task consumes a
great deal of energy. According to the Swedish Transport Administration, the total energy consumption
for the 6 800 S&Cs in Sweden equipped with electrical heating is 200-130 GWh/year, at costs of
approximately 10-15 M€/year. The next generation of S&Cs now being introduced are equipped with
almost triple the effect of S&C heating, from about 10kW to almost 30kW. This project tested enhanced
and with the ambition to have a more energy efficient snow and ice protection equipment for the S&Cs.
The currently installed equipment was tested using a two-sample t-test, i.e., comparing enhanced snow
and ice protection equipment and normal equipment for two different S&Cs at the same station. In
addition, a paired t-test was done to test the previous energy consumption. The results show a 30%
increase in efficiency with the new type of snow and ice protection equipment.
1. Introduction
Switches and crossings (S&Cs) are a key part of railway infrastructure. They make it possible for trains
to take different routes on the tracks, thereby making train operation more efficient. Functional failures on
S&Cs often result in delays, with a big impact on the quality of service [1]. Therefore, the transport system
needs robust and reliable S&Cs. However, the asset type with the most failures for 2015 in Sweden is the
S&Cs, accounting for 13% of all the failures. The biggest cause of failure is snow or ice. Electrical heating
is often used to remove snow and ice with electrical heating, but this consumes a great deal of energy.
The Swedish Transport Administration (Trafikverket) have about 12 000 S&Cs; 6 800 are equipped with
electrical heating. The energy consumption is approximately 200 to 130 GWh/year, at calculated costs of
approximately 10 to 15 M€/year. The next generation of S&Cs now being introduced are equipped with
almost triple the effect of S&Cs heating, from about 10kW to almost 30kW [2]. According to the European
1
program Horizon 2020, an important challenge in transport infrastructure is to reduce the impact of
infrastructure on the environment [3]. This makes the need for more effective use of the electrical heating
even more important.
The northernmost track in Sweden, the Iron Ore Line (Malmbanan), has trains operating in a sub-arctic
climate, including heavy snow in the winter and extreme temperatures ranging from - 40 °C in the winter
to + 25 °C in the summer. The main reason for S&Cs failure is snow and ice. Such failures account for
35% of the total failures per year. Hence, effective snow and ice protection has a significant influence on
successful operation. Since it is the northernmost line with the lowest temperatures and most snow, its
use of electrical heating is consuming the most energy.
The energy is transformed to heat in the rails to melt snow. The heat is lost through convection, conduction
and radiation [4]. Personnel at the Swedish Transport Administration (Trafikverket), together with their
maintenance contractors and suppliers, developed a new type of snow and ice protection equipment for
S&Cs. Among other things, the equipment is encapsulating the S&C to lower the convection and insulated
to lower the radiation. A partnership project was started in cooperation with Luleå University of Technology
to test and evaluate the new equipment.
The methods used to compare the energy efficiency of the new and the normal snow protection equipment
included a two-sample t-test for testing two different S&Cs in the same station and a paired comparison
using a paired t-test to compare historical data of an S&C before and after the installation of the new snow
and ice protection equipment.
The rest of this paper is organized as follows. Section 2 presents the background for the project and
provides details of both the normal S&Cs snow and ice protection and the new tested version. Together
with the description of data collection, Section 3 explains the project methodologies, its implementation
and methods. Section 4 discusses the results, and Section 5 offers a conclusion.
2. Background
2.1 Iron ore line (Malmbanan)
Figure 1. Geographical locaion of Iron ore line (Malmbanan)
2
The Iron Ore Line (Malmbanan) is the only existing heavy haul line in Europe; it stretches 473 kilometers
and has been in operation since 1903. It is mainly used to transport iron ore and pellets from the mines
in Kiruna and Malmberget, close to Gällivare, in Sweden to the harbors in Narvik (Norway) in the northwest
and in Luleå (Sweden) in the southeast; see Figure 1. As the map shows, the line is situated well above
the Arctic Circle. The climate is sub-arctic with extremely harsh conditions at times. The track section on
the Swedish side is owned by the Swedish government and managed by the Swedish Transport
Administration (Trafikverket). The trains are part of the government owned company LKAB; the iron ore
line is part of its production line, resulting in high demands for constant flow.
2.2 Snow and ice protection equipment for S&Cs
The new (named Kia 766) and normal (named Kia 751) S&Cs snow protection equipment and their
respective orientations in the S&Cs are shown in Figure 2; the upper ones are the new ones. The
alterations include the following. First, a rubber coating on the top and on the sides of the snow and ice
protection boards, on top a fine gravel is added for friction. The boards are positioned, as can be seen,
on the outer side of the rails for the switch panel, over the point machines and the stretcher bars. Finally,
polyurethane is layered as insulation underneath the boards. Some S&Cs have a moveable crossing;
these are equipped with the new snow and ice protection on the outside of the wing rail.
Figure 1 New and normal S&C protection equipment
Figure 2 Heating element
The S&C heating system is made up of several heating elements ranging range from 100W to 1200W.
The elements are placed on the stock rail, on the switch blade, and around the stretcher bars; if the S&C
has a moveable crossing, it is also heated. A heating element on a switch blade can be seen in Figure 3.
There are two sensors mounted in the S&C to control the energy consumption. For Malmbanan, the
sensors are set so that the lowest temperature should be +20 degrees Celsius. There is also a “Turbo”
function which means all heating elements are on full power. This is done if it starts to snow, or if it is
manually started for some reason.
3
2.3 Data collection
ÖVV is a new system in Swedish Transport Administration (Trafikverket) which is used to control the
switch heating. The system collects data on: outside temperature, temperature of rail for two locations in
each S&C, snowfall, times when “turbo” was used, and amount of energy used. The parameters were
collected every half hour. The data were collected from a cabinet controlling one or multiple S&Cs.
3. Methodologies
3.1 Project implementation
The project began by preparing a prototype to be tested in a pilot test on one S&C at one station. The
location was decided by the group based on the availability of ÖVV data, the existence of a comparable
reference S&C of the same type and heating effect, and the geographical distance to the location. The
selected location was Kirunavaara station, and the S&C with identification number Kia 766. The reference
S&C was Kia 751. After installation, the prototype was tested during operation. This resulted in minor
alterations of the design.
The next phase of the study involved choosing another nine locations for the new snow and ice protection
equipment. The selection criteria were the same as for the pilot test; however, they had to be placed
across the whole Malmbanan line, and S&Cs with more failures had to be chosen. The equipment
installation began near the end of 2015 and will be finalized early in 2016.
Data were collected from the pilot test S&Cs at Kirunavaara and then verified. Data included the following:
outside temperature, rail temperature from two different sensors, and energy consumption. All were hourly
data from the ÖVV system.
Data analysis of the energy consumption of the S&Cs at Kirunavaara was done using data for the outside
temperature and the temperature of the rail sensor with the lowest value. The analysis used a temperature
span of -12 to -5 degrees Celsius. Data were not used if there were indications of the “turbo” function,
and no data were used when there were indications of snowfall. The temperature range was based on
the expert knowledge of the project group. The aim was to minimize variability with a sufficient amount of
data. The exclusion of data for snowfall and the “turbo” function were also intended to minimize variability;
as the heating is operating on full power in both scenarios, unwanted effects could result. Methods used
to compare the energy consumption are described below.
3.2 Methods
In this project, to compare the energy efficiency of the new and the normal snow protection equipment, a
two-sample t-test is used to test two different S&Cs in the same station; while, a paired t-test is used to
compare historical data of an S&C before and after the installation of the new snow and ice protection
equipment. Both methods were used in Minitab [5, 6].
For both the two-sample t-test and the paired t-test, a quota for the degrees of Celsius that could be
changed by one kWh was calculated using this formula:
4
𝐷𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑖𝑛 𝑡𝑒𝑚𝑝𝑒𝑟𝑎𝑡𝑢𝑟𝑒/𝑘𝑊ℎ =
Lowest rail temperature − outside temperature
Kwh
4. Results
4.1 Results from the two-sample t-test
The two-sample t-test was used to compare Kia 766 S&C with Kia 751 S&C; recall that the former had
the new equipment installed. The hypothesis are: H0 : μ1 = μ2 and H1 : μ1 ≠ μ2 .
The data for both S&Cs are normally distributed, and the two-sample t-test finds the difference in the
means of the two S&C is significant, with a p-value of 0. The results show the mean value for the new
equipment is 1.32 and the mean value for the normal one is 1.07. The difference in the means shows the
new snow and ice protection equipment is increasing efficiency by almost 30%. This improvement is
shown in Figure 4.
Figure 3 Plots from the two-sample t-test
4.2 Results from the two-sample t-test
A paired t-test was used to compare the Kia 766 S&C before and after installation of the new snow and
ice protection equipment. The hypothesis are: H0 : μ1 = μ2 and H1 : μ1 ≠ μ2 .
The data for Kia 766 before the installation are normally distributed. The results show a difference in mean
between the two periods, with a p-value of 0. In this case, the mean value for the new equipment is 1.32
and the mean value for the normal one is 1.03. The difference is 0.28; this means an increased efficiency
of almost 30% as shown in Figure 5.
5
Figure 5 Plots from the paired t-test
5. Discussion
The study has a number of possible limitations. First, this is an initial test with just one S&C. More data
are needed from other S&Cs to learn more about the influence of the new snow and ice protection
equipment on energy consumption. In addition, the data used for Kia 766 S&C and Kia 751 S&C are not
from the same time period, and this might influence the results. For example, there could be a difference
in the wind conditions, and this may affect the findings. However, the samples are quite large: 252
samples for the Kia 766 S&C and 187 samples for the Kia 751 S&C.
The paired t-test and the two-sample t-test both conclude the same things. First the mean for Kia 766
with normal snow and ice protection is close to 1 degree/kWh, as is the mean for Kia 751. This makes the
results more reliable. Finally, there is a possibility that the heating effects changed during the test period
due to maintenance issues, but as the heating is never on full effect, this should be a minor problem.
6. Conclusion
The new snow and ice protection equipment has a positive effect on energy consumption. The pilot study
S&C shows 30% increased efficiency. These are preliminary results from the first of 10 locations using
the new snow and ice protection equipment. The results so far are very promising. Further studies with
more data including other locations and a larger temperature span will be conducted to optimize the
energy efficiency of railway S&Cs.
Reference
1.
2.
3.
4.
5.
6.
R. Lewis, U. Olofsson. Wheel-Rail Interface Handbook. Boca Raton (US): Woodhead publishing; 2009.
European Commission, Horizon 2020, 11. Smart, green and integrated transport, work program 2016-2017
[Internet] Brussels: European Commission; 2015 [cited 2016 Feb 8]. 123 p. (2015)6776: Available:
https://ec.europa.eu/programmes/horizon2020/en/h2020-section/smart-green-and-integrated-transport
The Swedish Transport Administration (Trafikverket). Slutrapport - Energieffektiva stationsområden (in swedish).
Available:
http://www.trafikverket.se/contentassets/1319b98c076c4653be8e98a69f126300/slutrapport_energieffektiva_st
ationsomraden.pdf, [Accessed Feb. 15, 2016]
C Borgnakke and R.E. Sonntag. Fundamentals of Thermodynamics. US: John Wiley & Sons; 2012
D. C. Montgomery. Design and Analysis of Experiments. Singapore: John Wiley & Sons; 2009.
S. Arnold. Mathematical Statistics. Prentice-Hall; 1990
6