Received 26 June 2024, accepted 26 July 2024, date of publication 6 August 2024, date of current version 15 August 2024.

Digital Object Identifier 10.1109/ACCESS.2024.3439471

Digital Twins for Railway Sector: Current

State and Future Directions
EVELIN KRMAC 1 AND BOBAN DJORDJEVIC 2
1 Facultyof Maritime Studies and Transport, University of Ljubljana, 6320 Portorož, Slovenia
2 School of Architecture and the Built Environment, Division of Transport Planning, Department of Civil and Architectural Engineering, KTH Royal Institute of

Technology, 100 44 Stockholm, Sweden

Corresponding author: Boban Djordjevic ([email protected])
This work was supported in part by the Swedish Transport Administration, Trafikverket (TRV) Grant No 2023/30667.

ABSTRACT Today, digitalization is a ‘‘must’’ for all industries and sectors. One of the most promising
and popular concepts of digitalization in the last decade is certainly the technology of the digital twin (DT).
As rail transport ranks very high among transport modes in terms of sustainability, resilience and reliability,
its digitalization is of great importance. Since any disruption or interruption of normal rail operations with
the aim of improving conditions could have many negative consequences, the technology of DT is perfect
in this sense, as it allows testing changes on a virtual copy of the real system without directly affecting its
operation. Combined with machine learning and other artificial intelligence algorithms, it can also be used
to predict and forecast operational characteristics, events and failures that can trigger rapid management
and action and prevent disruptions and cost increases. This article, therefore, aims firstly to provide a
comprehensive and overarching overview of the application of DT technology in the rail sector using a
newly proposed multi-layered classification, and secondly to highlight the research gaps that need to be
addressed in the near future to enable the transition to Rail 4.0. The result of this research, which examined
58 research articles published in scientific journals, shows that the application of DT in the railway sector
has increased in recent years, that investigations in infrastructure for maintenance purposes receive the most
attention in the scientific community, and that there is still much room for research and development of DT
in the railway sector.

INDEX TERMS Digital twin, DT, rail, railway, rail system, rail industry, DT application, rail DT classifi-
cation, rail DT domains.

I. INTRODUCTION European transport system. Based on Commission data [2],

This section explains the background and motivation for this rail is the least used mode of passenger transport in the EU-
literature review, describes the concept of the digital twin, 27 countries in 2020. In terms of passenger kilometers, rail
introduces related previous review studies and provides a accounts for 5.5%, cars, buses and coaches for 93.3% and
summary description of the research aims of this review and trams and metros for 1.3%. However, this status was largely
the structure of the article. impacted by COVID-19 and in 2022 the EU rail passenger
transport performance significantly recovered from the sharp
A. THE BACKGROUND AND MOTIVATION fall in 20201 . The same conclusion can be drawn for freight
According to ERRAC, The European Rail Research Advisory transport. 74.4 ton-kilometers in % of freight transport is by
Council [1], rail is seen as the backbone of mobility by road, and only 16.1% by rail. In addition, in 2022 EU rail
2030. As rail is the cleanest mode of transport, a shift to freight transport performance slightly decreased in compared
rail is recognized as a measure leading to a more sustainable
1 https://ec.europa.eu/eurostat/statistics-explained/index.php?title=
Railway_passenger_transport_statistics_-_quarterly_and_annual_data#
The associate editor coordinating the review of this manuscript and In_2022.2C_the_EU_rail_passenger_transport_performance_
approving it for publication was Jesus Felez . significantly_recovered_from_the_sharp_fall_in_2020

2024 The Authors. This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.
VOLUME 12, 2024 For more information, see https://creativecommons.org/licenses/by-nc-nd/4.0/ 108597
 E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

with the 2021 year.2 Therefore, in order to achieve the goal • There is an urgent need for the digital transformation
of promoting a modal shift to rail and to improve the role of railway systems, which means ‘‘synchronization’’
of rail in the transport market, more investment should be with the requirements of Industry 4.0. According to [7],
made in technical innovation in rail transport. It can also digitalization is ‘‘an ongoing process of convergence
be concluded from OECD data [3] that investment in road of the physical and virtual worlds’’ that is responsi-
infrastructure still dominates compared to rail. This is the ble for innovation and change in the economy. The
main reason for the stagnation of rail’s share of the trans- concept of Industry 4.0 means the implementation of
port market. Therefore, the only opportunity and the most new information and communication technologies in
important effort that should be made in rail transport is the industry with the aim of enabling data collection
to continue investing in the digitalization and automation through sensors installed on physical equipment, the
of the rail system so that ‘‘the digital railway drives the storage of this data (usually in clouds) and its analysis,
integration of the overall mobility-digital ecosystem’’ [1], the exchange of information between physical assets
[4], [5]. To further improve the position in the transport and, above all, the possibility of real-time decision-
market, technologies such as advanced interoperable train making based on real data from the physical world,
control systems, real-time operations management, virtual thus creating the cyber-physical system. To achieve this,
coupling and platooning, autonomous trains, smart commu- many technologies are available today: artificial intel-
nication between trains and other concepts and elements also ligence (AI), machine learning (ML), the Internet of
pointed out by UNIFE [1] and [4] will increase capacity and Things (IoT), Big Data Analytics (BDA), Blockchain
resilience, user satisfaction through higher punctuality and (BC) technology, cloud computing (CC), sixth genera-
comfort, and safety of rail transport without major infrastruc- tion networks (6G), augmented reality (AR), digital twin
ture investments. technology (DT) [8] and others. When combined, they
To achieve these milestones in the predicted time frame, can meet very demanding and high-level stakeholder
radical and rapid changes need to be made in the rail indus- requirements. In the context of railways, there is a wide
try. According to the findings from the railway symposia, range of data on the condition of railway assets (e.g.,
conferences, debates in the railway community, participation temperatures, light, vibration, humidity, GPS) [6] and
in Shift2Rail and EUrail projects and the literature, the most their operational status, which, if collected and pro-
important challenges on the way to the railway system are the cessed with the right technology or set of technologies,
following: could bring great benefits to the railway industry. In this
way, the concept of Railway 4.0 or Digital Railway,
• The rail system is very complex [6], comprising a large
which is a digital transformation or digital technological
number of assets and subsystems of different types with
evolution of railways, can be realized. According to [7],
varying maintenance needs. These assets are intercon-
the key areas of railway digitalization are ‘‘mobile appli-
nected and interdependent and must work together to
cations, e-ticketing, digital train control, signaling and
ensure that goods and people reach their destinations
traffic management, and digital platforms for predictive
safely and on time.
maintenance’’.
• The rail system is dynamic and its behavior in terms of
real-time organization and management is complex,
Most of the above challenges could be overcome if the
• As the population grows and the need for efficient
railway system was always in good condition. To be in good
and effective transport of goods and people increases,
condition, all railway components and equipment should be
so does the importance of reliable and sustainable rail
properly maintained. Otherwise, various problems may occur
transport. Furthermore, rail transport should be cost and
that may affect the safety, reliability and stability of railway
environmentally efficient and effective, while the satis-
operations.
faction of rail users (passengers or shippers/receivers of
To this end, constant monitoring of the condition of infras-
transport goods) should be high.
tructure and operations, detection and prediction of faults and
• The rail network is territorially dispersed, the infras-
possible failures, testing of new or modified equipment or
tructure and equipment in each area or locality vary
modification of operating conditions should be carried out
in age, quality and functionality, so there is a different
with the least possible negative impact on timetables and
need for modernization and maintenance. This makes
train safety. All this has a significant impact on rapid mainte-
the management of maintenance very complex.
nance, reliable operation and high performance. Expenditure
• Any change in the rail network, infrastructure or other
on railway infrastructure maintenance is made on a regular
systems means high investment and disruption to normal
basis. Figure 1a and [3] and [9] show that some countries
operations.
(e.g., France, Hungary, Poland) actually spend more on rail
infrastructure maintenance than on road infrastructure, while
2 https://ec.europa.eu/eurostat/statistics-explained/index.php?title= Rail- Figure 1b shows that in the majority of countries, the cost of
way_freight_transport_statistics#EU_rail_freight_transport_performance rail infrastructure maintenance is higher or even significantly
_slightly_decreased_in_2022_compared_with_the_previous_year higher than the investment in rail infrastructure. However,

108598 VOLUME 12, 2024

E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

TABLE 1. Cumulative numbers of articles regarding DT, DT in transport

and DT in rail transport.

whole life cycle of assets and enables the tracking of railway

assets from design to recycling phase.
Although DT technology is very popular in many indus-
tries, it has only recently been applied in the railway
sector [14], [15]. From the conclusions of some previous
studies [16], [17], DT in the railway sector is mainly focused
on infrastructure and traffic management. According to [6]
and [18], the larger number of DTs as proprietary solutions
and prototypes are mainly realized in the manufacturing and
production industry and thus at an early stage.
The popularity of DT technology in the scientific commu-
nity is evident: to show the increase in publications related to
the use of DTs in recent years, regardless of the industry or
field of application, we first performed a simple search (with-
out restrictions) in several representative databases using the
string (‘‘digital twin’’ OR ‘‘DT’’). The number of publica-
FIGURE 1. (a) Comparison of investments in road and rail infrastructure tions shows that the interest of the scientific community in the
(2021). (b) Comparison of rail infrastructure investments and
maintenance cost for European countries (2021). DT technology is very high. We also wanted to show that only
a small proportion of these articles deal with the application
of DT in transport, and even fewer with rail transport. There-
fore, we repeated the search in the same databases with the
the fact is that maintenance and fault detection, which are additional keywords ‘‘transport or transportation’’ and ‘‘rail
still mostly performed manually today, are among the most or railway’’ and included these results in Table 1.
critical points in railway systems [8]. This is why they have To see the distribution of these publications over the years,
received and continue to receive so much attention in the Figure 2 was created. Since the number of publications before
scientific literature [10], [11], [12], to name but a few. 2017 was quite small, the time frame shown is limited to
In recent years, digital twins (DTs) technology has been 2017 to 2023. It is clear that DT is not a new concept, but its
highlighted as ‘‘the one’’ promising technology that can sig- popularity and researchers’ interest in this technology have
nificantly improve many rail sector operations and contribute steadily increased after 2017. From Table 1 and Figure 2,
significantly to achieving the goal of Rail 4.0 [13]. The use of it can be seen that the research of DT in rail transport is very
DT technology in combination with other disruptive informa- recent and has increased by at least 30% in the last 2-3 years.
tion and communication technologies, in particular artificial
intelligence (AI), blockchain technology (BC) and 5G com- B. THE DEFINITION OF DIGITAL TWIN
munications [5], could also digitalized asset management The concept of DT has been extensively studied and
and enable the planning of infrastructure and rolling stock described in a large number of articles. Many discussions
use and maintenance activities at the right time [1]. As DTs of various names that appear in the vicinity of DT tech-
provide a highly accurate digital copy of the physical asset, nology, e.g. digital model, digital shadow, digital mirror,
process or system [5] that can be observed and tested in the virtual mirror, BIM (building information model), virtual
real environment under different conditions, independently copy, etc., have been presented in [11], [15], [16], [17], [19],
of its physical (real) twin and without interrupting ongoing [20], and [21], among others. Some detailed descriptions
processes, DT can also be of great help in planning, design- of the architecture and the different types of DT can be
ing, building and manufacturing railway assets, monitoring, found in [22]. A review of definitions of DT can be found
detecting and predicting infrastructure faults and failures, in many articles, including [11], [15], [17], [23], [24], [25],
as well as proper train operation, real-time scheduling and and [26] where no less than 46 definitions were systemat-
rescheduling. Another major advantage of DT, highlighted by ically reviewed. In [27], the difference between a standard
ERRAC [1] and UNIFE [5], is that DTs technology covers the train DT and an augmented one is also given. Different types

VOLUME 12, 2024 108599

 E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

FIGURE 2. Distribution of publications regarding DT, DT in transport and DT in rail transport.

of DTs, the structure of DT, the development of DTs, the deployment in rail and road networks, but focusing only on
models used and the milestones of development have also rail and road infrastructure (i.e., immovable assets on road
been described in detail by [10], [18], [25], and [28], to name and rail) and the resilience and sustainability of road and
a few. rail infrastructure. This systematic literature review was con-
To understand the classification framework, we have ducted using 20 journal articles found in the Scopus database
defined for the purposes of this literature review and how the and published up to November 2021. They concluded that
classification and subsequent analysis have been carried out 46% of the studies reviewed were in the field of engineering
in this review, we present the definition of DT, which we have and information technology and that only 32% of all studies
used as a reference: were related to rail transport. Finally, they found that most
A digital twin is a dynamic virtual model or representation studies focus on the use of DT in the operation and mainte-
of a physical entity (this can be a specific object or system nance phase and that due to the growing trend of DT, more
element, a whole system, a network of components or systems, studies can be expected in the rail and road transport sector
a process, or even a human or group of humans) together as well. The scope of other review articles is much narrower.
with its context that automatically collects raw data from McDonald et al. [30] gave an overview of developments in 3D
the real environment in real time and processes this data for visualization of underground inspection of railway tunnels
the purpose for which it was created (i.e., online visualiza- for maintenance purposes, mentioning only the potential of
tion/simulation of an object or process in real time, making Digital Twin Tunnel (DTT) BIMs as a means for two-way
predictions or forecasts, detecting anomalies, assisting in information exchange available at any time, where changes
decision making) to improve the performance of or increase in the physical tunnel led to changes in the digital tunnel
knowledge about the physical entity. providing information for future maintenance. Spiryagin et
al. [17] review the application of vehicle system dynamics
C. RELATED REVIEW STUDIES theories that could serve as a basis for the development of
At the time of writing, our literature search found no pre- DT. They concluded that most work examines the devel-
vious systematic literature review on the use of DT in rail opments of DT ‘‘at the top of the pyramid’’ and therefore
transport that had been published as a journal article. The more comprehensive and in-depth research is needed in the
article that came closest to this topic is a review article [29], application of vehicle system dynamics in the processes of
in which the authors aimed to present the current extent of DT design and construction of DT. Feng et al. [31] investigated

108600 VOLUME 12, 2024

E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

the application of DT in intelligent transport systems (ITS) as complementary to our literature review (or vice versa) and
with the aim of exploring the role of DT resilience in the both are a good basis for future upgrade.
applicability of ITS. They found that current research on
DT technology has varying degrees of success in different D. RESEARCH AIMS AND CONTRIBUTIONS
aspects of ITS and that the system performance of DT should
The aim of this study is not to provide an overview of the
be optimized. They also concluded that the data exchange
general concept of DT, descriptions of the development pro-
between vehicles and infrastructure in the transport network
cess, models and key technologies used for the development
can be solved well if blockchain technology is integrated with
of DT and an overview of the industrial applications of DT,
DT. They confirm the assumption that the improved resilience
but to provide a comprehensive and overarching overview
of DT has an impact on the adaptability of the transport
of the popularity of the development of DT for the railway
system. Errandonea et al. [32] conducted a literature review
sector in the research community in terms of the number
on DT for maintenance in general, which is one of the most
of publications, the railway domains and the DT purposes
researched applications of DT. Of the 167 studies found (arti-
that have attracted the most attention so far. The novelty of
cles, conference papers and other publications), only 68 were
this paper lies in the completely new and unique scheme of
relevant as they investigated the use of DT in maintenance.
an overview of the perspective of DT development in the
The railway sector was only studied in two cases, while the
railway sector. Moreover, the novelty of the paper reflects
leading sector was manufacturing with 29 cases.
the finding that DTs have been developed mainly for railway
In our backward snowballing process, we came across the
infrastructure so far. Furthermore, the novelty in the research
project [33] in which a systematic literature review of all pub-
brings additional knowledge to the research community. DTs
lications published in English between 2000 and 2022 was
have so far mainly been developed using simulations and
conducted with the aim of producing a ‘‘guideline to support
prediction tools. The aims and main contributions of this
the design of a DT for predictive maintenance in the railway
systematic literature review are therefore to
sector’’. This review examined and classified 60 publications
mentioning challenges, open issues or future opportunities • assess the extent to which the technology of DT has
and dealing with predictive maintenance in the railway sector, influenced scientific publications in the field of rail
including the railway manufacturing process. First, they were transport so far;
classified into 8 subsectors of the railway sector based on the • provide a comprehensive classification framework that
classification of [34] and it was found that of all the subsec- allows a deep insight into the role of DT in rail transport;
tors, maintenance and inspection were covered in the most • collect information on already realized implementations
publications. The second classification was made according of DT in rail transport;
to which of the DT enabling technologies were discussed, • identify current research gaps (i.e., aspects of rail trans-
and it was found that in most publications more than one port that have not yet been studied from the perspective
technology was discussed and that artificial intelligence (AI) of DT and concepts related to the development of DT
was most often discussed (in 25 out of 60 publications) in that could be (re)used in another area of rail transport or
combination with other technologies. The third classification beyond);
made by [33] was to sort these 25 ‘‘AI’’ publications into • find the most promising trends and opportunities for
8 subcategories, of which machine learning algorithms and future research.
techniques were most frequently used in the development of
The results and conclusions in the form of an overview
DTs, mainly for fault detection, failure prediction, automated
of realized implementations of DTs and existing gaps as
decision making and health monitoring. Dirnfeld [33] also
well as possible future research directions gained from this
concluded that only half of the publications studied discussed
review can be used by both the research communities and
challenges, open questions and future opportunities. In the
practitioners involved in rail transport to take advantage of
end, the guideline to support the design of a DT for predictive
the promising technology of DT when it comes to improving
maintenance was created in the form of a flowchart.
the rail system and promoting the digital transformation of
While our article was under review, the survey by
railways.
Ghaboura et al. [35] was published, which focused on clar-
ifying ‘‘how DT can serve the railway twin system designers
and developers’’ by presenting a taxonomy for them and E. ARTICLE STRUCTURE
giving a rough and detailed overview of the emerging tech- The article is structured as follows: Section II is devoted to
nologies and techniques used for the application in the rail the presentation of the research questions, the search process
sector. This survey was conducted on 80 articles (45% of and the newly defined classification scheme and classifica-
these were conference papers, book chapters and books, and tion criteria for the classification of the reviewed studies.
55% journal articles, of which only two were published in Section III contains the analyses of the articles reviewed in
2023) found in the Scopus and WoS databases and published relation to the classification used and the presentation of the
between 2018 and January 2023 (including 31 that are also correlations between each classifier or aspect, highlighting
included in our systematic review). The survey can be seen the most important ones. Section IV is devoted to concluding

VOLUME 12, 2024 108601

 E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

reflections and a summary of possible future research areas TABLE 2. Search results in the identification stage.
and directions.

II. RESEARCH METHODOLOGY

The entire research process described in this article comprises
four main steps. In the first step, the research questions and
the scope of the literature review were defined. In the second
step, the search was conducted, while in the third step the
classification criteria were established and the articles were
classified according to these criteria. Finally, the analysis if any relevant article was missed in the first two searches).
of the search results was conducted and research gaps were We started by formulating the search queries. Since the
identified. selected databases use different search engines, two different
search strings were formulated. For the search in Scopus and
A. RESEARCH QUESTIONS TRID we used ((‘‘digital twin∗ ’’ OR DT) AND rail∗ ), and
The research questions that have guided all the research in for ScienceDirect and WoS the string ((‘‘digital twin’’ or DT)
this work are in line with the main objectives of this review and (rail or railroad or railway)) was used. With these strings,
(see 1.4) and are as follows: the database searches were carried out in 4 iterations (first
RQ1: What is the state of the art of DT technology in the Scopus was searched, then TRID and ScienceDirect, finally
rail industry? WoS). At the end of the 4 iterations, 125 articles were selected
RQ2: In which domains of railway systems have DTs for screening (see Table 2).
been developed most frequently and which domains offer the In the screening phase of each iteration, we first read
greatest scope for future developments of DT? titles, abstracts and keywords. Since we used the End-
RQ3: What are the most common uses of the developed Note library for a collection of selected articles, we removed
DT? duplicates after each iteration. At the end of the screening
phase, we had 78 articles. In the eligibility phase, we first
B. LITERATURE REVIEW APPROACH checked whether the specific inclusion criterion was met
In order to conduct a general (i.e., overarching) and com- and then read the remaining 68 articles in full text. During
prehensive literature review, we systematically searched the the reading, data were extracted for each article (year of
literature using the PRISMA approach [36]. The main reason publication, scope/aim of the article, purpose of a DT, scope
why we chose this approach is that the PRISMA model can of DT, railway aspect studied, rail assets considered and
help researchers to write a literature review. In practice, this other details). Backward snowballing was applied to the
model can help to find numerous answers to the research remaining 52 articles.
questions. Moreover, it can be used to find new aspects in Ten new articles were found and steps 2 to 5 were repeated
the respective research topics, and it can be further expanded for these articles (see Figure 3).
depending on how many more references are added to each After the snowballing, 5 of these articles were included
criterion, or the criteria can be divided into some sub-criteria. in the final group of articles for classification. Based on
The entire search process is illustrated in Figure 3. the data collected, the classification categories and criteria
In the identification phase, we first established the or categories were first established (see 2.3) and then the
inclusion or eligibility criteria in relation to the research articles were classified. Data collection began on 2 May
objectives and questions. We decided on four general inclu- 2023 and classification ended on 28 June 2023, on which
sion criteria and one specific inclusion criterion. The general day an additional search cycle was conducted to take into
inclusion criteria are the requirements that the article (1) is account articles that had been published in the meantime
about a development of DT in railways; (2) has already been before the analyses began. From two articles found, one more
published or accepted for publication in a scientific journal was added, so that the final number of articles for the analysis
(i.e., has already been peer-reviewed); (2) is freely in full text was 58.
(i.e., open access); and (4) is written in English. A specific The identification phase was carried out jointly by both
inclusion criterion was the requirement that both keywords authors, while the other steps were first carried out by one
‘‘digital twin’’ and ‘‘rail’’ appear at least once in the text of author and then repeated by another. Any disagreements or
the article and not only in the abstract, title or keyword. There differences were discussed immediately and consensus was
were no restrictions on the year of publication, the source reached before proceeding.
of the publication, the number of citations or other criteria.
Then we made the selection of databases. Four of the most C. CLASSIFICATION CRITERIA
reliable databases with sufficient coverage were selected, The articles studied are classified according to a new classifi-
which were then searched in the following order: Scopus (the cation that we have created for the purposes of this literature
‘‘broader’’ one), TRID (the specific database for transport), review. We have selected five different representative criteria
ScienceDirect and WoS (the last two were included to check or classifiers that allow us to roughly assess the state of DT

108602 VOLUME 12, 2024

E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

FIGURE 3. PRISMA diagram for searching process.

development in rail transport up to June 2023 that has been maintenance and inspection, safety, planning, management
published in scientific journals and to provide an insight into and policy, passenger services, optimization and monitor-
research gaps that could or should be filled by future research. ing). This division was also used by [33] who provided
The selected classifiers - rail domain, DT purpose, asset life an overview of DT in the railway system with a focus on
cycle phase, DT development phase and DT segment - are predictive maintenance. In [11], DT was investigated in the
described in detail in the following subsections. field of maintenance and for this purpose the authors divided
In the ‘‘DT railway literature’’ some classifications and the applications of DT into five groups, namely mainte-
categorizations have already been used for the purpose of nance, PHM (Prognostic Health Management), life cycle
the literature review. For example, in [34] a division of optimization, process/logistics/production and design. Broo
the railway system into seven main areas is used (i.e., and Schooling [10] stated that the full life cycle of assets

VOLUME 12, 2024 108603

 E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

includes all activities from identifying the need, planning and and artificial intelligence algorithms); simulation and mod-
designing to construction, operation, integration and mainte- elling.
nance.
In [29], where the focus of the study was on immovable 3) LIFE CYCLE PHASE
assets in road and rail transport, sectors such as buildings, Each specific DT also has a scope covering the life cycle
transport, general, energy and telecommunications were used phase (LC) or stage of the railway asset under consideration.
for classification. In all cases, the categories were defined The phases introduced describe for which stage of the rail
based on the literature review, while our classification is asset a DT has been developed or in which stage it has the
based on the composition and characteristics of the railway greatest impact or its benefits are greatest. DTs can support all
system on the one hand, and on the characteristics of DTs the initial phases of asset development, i.e., planning, design
and their development on the other. and construction, their main phase, i.e., operation, assist
in decision-making regarding upgrades, replacements and
1) RAIL DOMAIN
capacity expansions, and their decline at the end of their life
cycle. To this end, we address the following seven possible
In creating the new classification, we have assumed that
phases: planning (development of goals and tasks); design
the railway domain or subdomain or area or subsystem is a
(conception and definition of the asset, identification of
set of related ‘‘elements’’ or ‘‘components’’ of the railway
requirements); construction (manufacturing or fabrication,
system to which the ‘‘twinned’’ component belongs, while
building, construction, testing of the asset and its connections
the description of the scope of DT for this component falls
with the environment); operation (function or use of the
into the category DT purpose, and the phase of the life cycle
asset); maintenance (repair and remove of faults or damages,
of the component for which DT is used falls into the category
reconstruction or replacement of components and upgrading
Life cycle phase.
of systems); management (asset, operations management
We therefore assume that the main railway areas or
or maintenance management and all supporting activities);
domains are as follows: infrastructure, including track
whole asset LC.
infrastructure (track, rail sleepers and ties, ballast, rail
joints, level crossings) and infrastructure facilities (includ-
4) DT LEVEL
ing bridges, viaducts, tunnels, stations, maintenance and
To assess the maturity of the development of DT, we intro-
repair facilities); rolling stock (locomotives, passenger and
duced a four-point scale with the following categories:
freight wagons and special vehicles); power supply (power
exploratory (only an idea or a study of the realization possi-
distribution system, overhead lines, third rails and trans-
bilities with the concept of DT); preliminary step(s) (some
formers that supplying power to trains); telecommunications
modelling and simulation activities that can be further devel-
(communication devices and media for data and informa-
oped into DT have already been carried out, usually also
tion transmission); ticketing and fare collection (vending
tested in a real use case); development (DT is fully devel-
machines, ticket offices and systems for collecting fares from
oped and validated against a test data set); implementation
passengers); signaling, operation and train control (sig-
(developed DT is validated by use in a real use case).
naling systems, signals, track circuits and other equipment
providing information to train operators and controlling train
5) SEGMENT
movements, control or dispatch centers); data and informa-
As a final classifier, we have used the description of the
tion (databases and data management applications); system
segment or part of the rail system covered by DT. Thus,
as a whole.
DT can either be individual, meaning that only one aspect
All these components make up the railway system. They
or component is ‘‘twinned’’, or collaborative, meaning that
work together to create a functioning railway base that sup-
the integration of two or more individual DT is set up or
ports the safe and efficient operation of trains.
discussed.

2) DT PURPOSE III. CURRENT RESEARCH ON DIGITAL TWIN IN RAILWAY

DT purpose describes the reason for which a particular DT is TRANSPORT - ANALYSIS OF LITERATURE
used, or the scope of a particular DT, or even what activities a In the spirit of a comprehensive and overarching litera-
particular DT supports. In order to cover the widest possible ture review (the summary of reviewed studies is presented
range of possible uses in rail transport, we have divided the in Figure 4), we first conducted some basic analyses, fol-
purposes into five groups as follows: investigation (real-time lowed by the critical analysis of the classification results
monitoring, inspection, diagnosis, measurement, detection, and concluded with the presentation of the most important
evaluation); improvement and optimization (optimizing correlations between classifiers from different categories.
safety, operations, efficiency, asset management, resilience,
sustainability, maintenance, life cycle or a combination of A. BASIC ANALYSES
these aspects); decision-making (analytical processes); pre- Figure 5 shows that most articles examine the infrastructure
diction (analytical processes involving machine learning aspect of the railway system. Investigation is the fundamental

108604 VOLUME 12, 2024

E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

FIGURE 4. Classification of articles studied.

VOLUME 12, 2024 108605

 E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

FIGURE 4. (Continued.) Classification of articles studied.

activity in maintenance, which of course was most often that can change their position, such as turnouts, switching
highlighted as the phase that will benefit most from a DT. points and level crossings. Therefore, the behavior of such
Half of all DT presented were actually developed and most infrastructure components can be considered as static or
of them were validated in a real railway use case. Almost all dynamic components [38]. Based on the definition of DT,
articles presented the development of an individual DT. it can be developed to mimic a physical component or system.
Figure 6 shows the number of publications per year for However, even though there are static and movable infras-
the rail domain and the categories with the highest number tructure components, the results of the review show that the
of studies. Figure 6a shows the large difference between developed DTs mainly refer to static components such as
2020 and 2021, indicating that DT in rail transport has only tracks and rails and infrastructure facilities (see Figure 7).
recently become a popular research topic. As this overview Furthermore, the developed DTs mainly refer to the third
only includes articles published in the first half of 2023, it can part, which is generally used for the purpose of their condi-
be assumed that the number will be highest at the end of tion assessment. For the same purposes, DTs have also been
2023 compared to the other years. DT publications from DT developed for turnouts, the only studied case of dynamic com-
in the field of rail infrastructure follow the increasing trend ponents [39]. Nevertheless, it should be noted that DT has not
of all publications, while publications from in the field of been developed to plan and schedule maintenance and moni-
rolling stock are constant over the last two and a half years tor real-time operation of movable infrastructure components
(see Figure 6b). such as switching points, level crossings and yard braking
systems (i.e., retarders). Furthermore, our review did not find
B. ANALYSES OF CLASSIFICATION RESULTS DT capable of monitoring, for example, the transmission of
In the last year and a half, the number of studies in the areas of electrical power and the stability of the catenary network.
investigation, simulation and modelling has decreased, with Figure 8b again shows that DTs for the rolling stock
a slight decrease in the number of studies in the area of inves- domain were developed mainly to monitoring states, condi-
tigation. As for maintenance and development, Figures 6d tions and damage to vehicles and locomotives and some of
and 6e show an increase over the years. Finally, Figure 6f their parts. As for behavior, Sun et al. [40] developed DT to
shows the number of studies dealing with the integration of evaluate the stability of the vehicle. However, the behavior of
different DTs into collaborative DTs, which has only recently other movable components of vehicles such as bogies, axles
become of interest to the research community. and couplers and their impact on the environment and society
were not found. The important aspect of rolling stock is fleet
1) RAIL DOMAIN planning to minimize vehicle usage and empty kilometers for
Railway infrastructure consists mainly of static physical which DTs are missing as well. A locomotive can be further
components and includes all components from the overhead divided into different components. However, DT does not
line to the track bed [37]. However, there are also components exist for individual components of the locomotive from the

108606 VOLUME 12, 2024

E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

pantograph to the power axles [41]. Furthermore, there are 2) DT PURPOSE

no DT models to monitor and improve of energy consump- Looking at the results of the review, the purpose of the
tion or emission production of diesel locomotives as well investigation in DT studies can be divided into two groups.
as noise. In the first group of papers, the use of the DT model was
Some studies were found in the domain of railway sig- investigated, while in the second group, different models
naling (see Figure 8c). However, these studies dealt with the or approaches to investigation make up the part of the DT
DT for European Traffic Management Systems (ERTMS). system. In both cases, the investigation is mainly related
A signaling system with all connected components operating to the maintenance of tracks, rails and/or wheels. This DT
in a synchronized manner [42] presents a unique opportu- purpose is related to maintenance, as good maintenance and
nity for the development of individual or cooperative DT. fault analysis are prerequisites for safe and proper railway
Such DTs can improve and optimize train operations and operation.
real-time decisions on train movement authority especially For improvement and optimization, only one DT was
for advanced signaling concepts such as moving block, virtual developed, which focuses on movable components such as
coupling and driverless trains. Only a small amount of work turnouts.
was found that relates to the control of train operations. The aim of railway digitalization and automation is to
DTs for these purposes are useful as they can facilitate the reduce the workload and improve the working environment.
decision-making of traffic management systems. In addition, For example, digital automatic coupling (DAC) is being intro-
DTs in this domain can help predict failures based on past duced in rail freight to reduce the number of work accidents.
experience and provide appropriate solutions in a timely man- Therefore, DT has been proposed for improving the working
ner. Moreover, DTs can simultaneously improve real-time environment, the environment for passengers in a railway
capacity and reduce delays without compromising service coach, for the construction, maintenance and structural health
safety and stability [43]. monitoring and condition assessment of infrastructure assets.
Figure 8d shows some studies that have been conducted However, almost all studies focused on the perspectives of
in the context of power supply. The studies have developed maintenance and construction. Few or no studies dealt with
models for the design and maintenance of the power system. railway components to improve or optimize the real-time
It can be concluded that there is a lack of DTs to simulate and operation of trains. With the introduction of new signaling
optimize the power distribution in the network and the use by concepts, optimizing the location and arrangement of balises
the locomotive using DT models. is the crucial issue to be addressed.
Only the article by Kochan [21], which develops a DT for Only one study was found on this, while communication
ETCS, classified and described under the signaling, operation between vehicles, trains and infrastructure facilities has not
and train control, falls under telecommunications, while no yet been researched.
article deals with the domain of ticketing and fare collec- Preliminary steps to develop DT or to fully develop DT
tion. have been presented for the purpose of decision-making.
The exchange of data and information has been an essen- Decision-making is an important aspect for all railway phases
tial part of railway operations since the early days of the from construction to operation, i.e., for the whole life cycle.
railway and is the most important component of railway oper- Among the results of this study are asset management, deci-
ations today. Some of the studies shown in Figure 8e were sions related to manufacturing and design, and necessary
concerned with the development of DTs for the exchange, investigations and maintenance. Apart from the DTs for the
storage and management of data. This is particularly impor- above decisions, no real-time decision support was found for
tant for railways, a well-known transport system with the train rescheduling, rolling stock management on the network,
worst data assets. Collecting and sorting data is crucial for train movement authority or other decisions related to real-
all components to ensure smooth railway operations. Never- time operation.
theless, the representation of data sharing through DT would With the aim of creating appropriate strategies in planning
be useful for the purposes, especially between physical and and management, predictions are the basis. Therefore, DTs
movable rail components and within the physical and mov- have also been developed for forecasting purposes in the
able components themselves [44]. railway sector. Similar to the previous purposes, DTs for
In addition to the studies that looked at DT for specific prediction were mainly concerned with maintenance issues,
components or purposes, there are also studies that provide more specifically prediction of surface damage to rails,
DT initiative for the entire railway system (Figure 8f). From maintenance of switching points, derailment risks and pre-
the perspective of the whole system, DT is more useful diction of wheel and rail life. In addition, DT was developed
than a separate system as it incorporates data from different for predicting the thermal performance of electric motors.
sources and also historical data to reflect the real-time state Although maintenance is one of the most common topics,
and behavior of the railway assets. In addition, DT provides the development of DTs to predict damage that can cause
synchronization of the many properties and behaviors of each high expenditure and loss of life has been omitted. These
real object for the whole system. damages can be related not only to the damage of rails, but

VOLUME 12, 2024 108607

 E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

FIGURE 5. Coverage of classifiers and categories.

FIGURE 6. Distribution of publications by years (a) of DT in rail; (b) of DT in rail infrastructure and rolling stock; (c) of DT for the investigation
and simulation and modeling in rail; (d) of DT for rail maintenance; (e) of developed rail DT; (f) of collaborative DT for rail.

also to the damage of axles, the heart of switching points or no train delay prediction models were found at DT, which
other rail elements, passing the red signal, etc. In addition, may affect service quality and are related to decision making.

108608 VOLUME 12, 2024

E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

FIGURE 7. Detailed classification of overviewed studies – Part I (Infrastructure).

DTs for predictive purposes can be useful to detect incidents, the relationship between rail traffic characteristics, to support
human and technical failures and errors that can lead to earlier the design model for locomotives and to improve the safety
damage in railway systems. and stability of wagons.
Simulation and modelling are used as a central part of
the development of DT or to investigate the feasibility of DT. 3) DT LEVEL
In the railway sector, simulation is a common approach to The results of the review have shown that different levels
explore new concepts or possibilities. of DT development have been achieved. Some studies only
Nevertheless, simulation and modelling are used to explore discussed the possibility of DT for certain rail sections and/or
DT application and validation. Mainly simulation and mod- components, while there are also studies that initiated pilot
elling in DTs were developed for earlier purposes. For studies of DTs. In most of these cases, DTs were not built
example, simulation in DTs has been used to investigate due to the lack of data and information. From the results,
fatigue damage, semantic segmentation tasks, track diagno- it appears that most DTs were also developed and imple-
sis, optimization of the electrical side of medium voltage mented in a real case.
lines DC and to decide on control equipment in ERTMS.
However, simulations are also used in the railway sector when 4) ASSET LIFECYCLE
there is a lack of data. Therefore, the question arises why the Maintenance, ‘‘a complex and complicated task in the rail
development of DT for this purpose is limited, although the industry’’ [45] and beyond, is a life cycle phase that can
automation and digitalization of rail transport is a hot topic in incur huge costs, take up a lot of time, disrupt normal oper-
the last decade. Modelling in DTs has been used to improve ations and, as mentioned in [46], affect safety and passenger

VOLUME 12, 2024 108609

 E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

FIGURE 8. Detailed classification of overviewed studies – Part II (other domains).

108610 VOLUME 12, 2024

E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

FIGURE 9. Correlations between categories and classifiers.

comfort. The negative impact can be even greater if main- ‘‘simulate’’ the basic units or assets of the railway system
tenance is not supported by timely management of spare and connect them with data and information flows in such
parts or the tools and machinery required for maintenance. a way that they form a ‘‘whole’’. For such a system, [18]
As outlined in [11], [47], and [48], among others, it is impor- have used the term ‘‘supra-system’’ and emphasized that its
tant to take advantage of digital technology that enables the realization requires an ‘‘open and comprehensive networked
transition from traditional, mostly manual maintenance that platform’’, which in turn means that it enables the avoidance
occurs after a failure, to condition-based maintenance that of data silos and the integration of additional services from
is demand-driven depending on the condition or state of the which all stakeholders can benefit. The importance of such
asset, and to predictive and preventive maintenance. The latter collaborative DTs is also highlighted by [49], who states that
two are based on predictions and anticipation of future condi- by 2027, more than 40% of large organizations worldwide
tions based on continuous monitoring and diagnosis. If until will use a combination of Web3, spatial computing and digital
today DTs have mainly focused on supporting the design or twins in metaverse-based projects to increase revenue, where
manufacture of railway assets and condition-based mainte- metaverse means a combination of multiple technologies that
nance by monitoring and inspecting the current state [10], corresponds to DT technology and the concept of Industry
[11], [12], today we are moving to a higher level of main- 4.0, especially that of tomorrow.
tenance and operations by combining data analytics with
machine learning and artificial intelligence to support the C. CORRELATION BETWEEN CLASSIFIERS
whole life cycle of railway assets, which is also confirmed Figure 9 shows the correlations between the categories of the
by the above analyses of the reviewed publications. classifiers used for this report. The graph was created using
the corrrplot function, which plots a matrix of Pearson corre-
5) SEGMENT lation coefficients calculated using the cor function, both in
A digital twin of a system such as a railway system is only R. The strength of the correlation between the variables (i.e.,
possible if individual, specific DTs are first developed that the categories) is shown with the intensity of blue (i.e., for

VOLUME 12, 2024 108611

 E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

positive values) and orange (i.e., for negative values), while Although the infrastructure components and facilities for
the white colour is used to visualize the value 0 (i.e., in our passenger and freight transport are the same, the results of
case for the variable Ticketing_and_fare_collection all values the study show no difference between passenger and freight
are 0, as we did not find any article for this category). rail transport. Rail freight can benefit from DT as it is
The key finding from these correlations is that half of all characterized by challenges such as heterogeneous traffic,
infrastructure DTs were dedicated to the investigation, mostly ongoing innovations and multimodal opportunities. In creat-
in support of the maintenance lifecycle phase. Furthermore, ing a European multimodal transport system with rail as its
infrastructure DTs focused on only one asset or aspect and backbone [1], the connectivity and interoperability of digital
were mainly developed and implemented for a real use case. technologies to support each mode will be of great impor-
The same is true for DTs for rolling stock, with a slightly tance. Again, the concept and interconnectivity of DTs could
higher proportion of DTs for the purposes of prediction and be the challenge that the research community should focus
simulation and modelling. Investigation and prediction are on. Therefore, the use of DTs could enable the railways to
strongly related to maintenance, yet DTs are developed for further meet the requirements of the circular economy under
this area. The focus is on individual DTs, regardless of the the European Green Deal [5].
domain, DT purpose or supporting life cycle phase. The num- As the development of collaborative DTs increases, we can
ber of DTs developed and implemented is also high. Except expect DTs to also support asset and system planning as the
for signaling, operations and train control domain and data need for assets from other DTs from the ‘‘Rail DT’’ or DToC,
and information, at least half of all DTs have been developed Digital Twin of the Customer, ‘‘a dynamic virtual represen-
(and most of them also implemented). tation of a customer that simulates and learns to emulate and
anticipate behavior’’ [49], which is slowly gaining traction in
the economy and will be part of our business in 5 to 10 years,
IV. CONCLUSION AND FUTURE RESEARCH DIRECTIONS according to Gartner [50] and DHL [51]. DHL [51] also
This study provides a comprehensive overview of DTs in emphasizes in its forecasts for DT that the main application
the rail sector, focusing on the domain of DTs development, areas for DT will be single assets and closed systems for the
the level of development, the life cycle phase in which DT time being, and that broad adoption in the next 5 to 10 years
is used, and the purposes of DTs application in the rail will focus on DT for whole systems and networks as an
sector. The results show that for the purposes of the study, integration of thousands of assets across different actors.
DTs have been mainly limited to infrastructure. The main Over the last two decades, there has been much discus-
reason for this is that railways are complex and consist of sion about the full automation of railway operations and the
various components that need regular maintenance to ensure upgrading of railway infrastructure and rolling stock with
proper operation of trains. Therefore, the most DTs have been intelligent software and hardware. Therefore, the railway sec-
provided for monitoring and maintenance of assets. The study tor is expected to witness high demand for the development of
has also highlighted promising applications of DTs, as well DT and its application to monitor proper operations, control
as potential challenges and future opportunities. risks, reduce costs and improve predictions. However, along
The results of this research have several particular impli- with the growth in the development and application of DTs,
cations. First of all, the study covers the knowledge gap in railways also face many challenges. The biggest challenge
terms of where DT development in railways has gone so far is the timing of the application of DTs, as there is still no
and where the potential lies. In addition, the study presented general consensus on the concept of a DTs, while there are
DTs for specific railway components that can be used as good different understandings and definitions of DTs in different
practice in the future. The study is useful for the research industries. This is also related to the fact that models are often
community to investigate the state of the art of DTs in rail created separately by operators, infrastructure managers and
sector. Given that DTs are promising technologies for the manufacturers, who rarely collaborate on sharing concepts or
future, the study provides a basis for encouraging and raising models.
awareness of the potential of DTs in the rail sector among Therefore, the rail industry should follow the ideas of the
stakeholders and policy makers in the rail sector. core theory and the concepts of DTs. Another challenge for
Rail services are being deducted to passengers and freight the rail industry is to identify the scope and scale of DTs,
customers and the potential of DT in passenger rail transport develop use cases and drive large-scale adoption of DTs. The
is recognized as the option to improve the overall passenger challenge is also to demonstrate the DT concepts developed
experience and ensure a switch to rail. Nevertheless, it can and the ability of DTs to operate in real time. Therefore, the
be concluded from the results that a small number of DT concept of DTs in the rail industry can be developed and
studies can be found on the quality of service for passengers. applied on a larger scale in terms of capacity, operational
By monitoring and collecting data from various sources such efficiency and reliability, structural and component issues,
as ticketing systems, traffic flow sensors and train timetables, and safety issues. DTs in the rail sector can be produced at
operators can predict overcrowding, inform passengers in real component level or based on the whole rail system with dif-
time about delays or disruptions and work on better utilization ferent levels of detail for all phases of the life cycle. However,
of their fleet. the creation of trustworthy, validated and less costly concepts

108612 VOLUME 12, 2024

E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

and models is the main requirement as the improvement of all REFERENCES

mentioned rail services is key for a shift to rail. [1] Rail 2030—Research and Innovation Priorities, ERRAC, Brussels,
It should also be noted that the research community has Belgium, 2030.
[2] European Commission. (2022). Statistical Pocketbook 2022. [Online].
only just begun to develop and explore DT collaboration Available: https://transport.ec.europa.eu/facts-funding/studies-data/eu-
in/for the rail sector, which is in line with other sectors, so this transport-figures-statistical-pocketbook/statistical-pocketbook-
is certainly one of the biggest research gaps and a fairly 2022_en16
[3] OECD Data on Infrastructure Investments, OECD, Paris, France, 2021.
certain future research direction.
[4] Digital Trend in the Rail Sector, UNIFE, New York, NY, USA, 2019.
From the results of the literature review, DTs are use- [5] Rail Fit for the Digital Age, UNIFE, New York, NY, USA, 2020.
ful concepts that combine real-time modelling, simulation, [6] R. Sahal, S. H. Alsamhi, K. N. Brown, D. O’Shea, C. McCarthy, and
agent-based modelling, machine learning, prototyping, opti- M. Guizani, ‘‘Blockchain-empowered digital twins collaboration: Smart
transportation use case,’’ Machines, vol. 9, no. 9, p. 193, Sep. 2021, doi:
mization and Big Data. Consequently, research directions for 10.3390/machines9090193.
DTs in the rail industry should focus on developing models [7] J. Pieriegud. (2018). Digital Transformation of Railways. [Online]. Avail-
based on different techniques for the virtual component of able: https://www.sgh.waw.pl/www.mobility.siemens.pl
DTs. These models should be created and tested using various [8] C. Gao, J. Wang, S. Dong, Z. Liu, Z. Cui, N. Ma, and X. Zhao, ‘‘Application
of digital twins and building information modeling in the digitization of
unique use cases that meet the needs and requirements of the transportation: A bibliometric review,’’ Appl. Sci., vol. 12, no. 21, p. 11203,
rail industry, as well as real-world scenarios. In addition, the Nov. 2022, doi: 10.3390/app122111203.
benefits of improving railway functions should be quantified [9] OECD Data on Infrastructure Maintenance, OECD, Paris, France, 2021.
[10] D. G. Broo and J. Schooling, ‘‘A framework for using data as an engi-
by exploring the end products and solutions. Consequently, neering tool for sustainable cyber-physical systems,’’ IEEE Access, vol. 9,
the first phase of DT development should be the definition of pp. 22876–22882, 2021, doi: 10.1109/ACCESS.2021.3055652.
quantitative indices. These are also important to understand [11] I. Errandonea, S. Beltrán, and S. Arrizabalaga, ‘‘Digital twin for main-
the accuracy and effectiveness of a DT. Such metrics can help tenance: A literature review,’’ Comput. Ind., vol. 123, Dec. 2020,
Art. no. 103316, doi: 10.1016/j.compind.2020.103316.
to identify potential errors to assess the stability of the DT, [12] S. Kaewunruen and N. Xu, ‘‘Digital twin for sustainability evaluation of
and they can provide a quantification of uncertainty to assess railway station buildings,’’ Frontiers Built Environ., vol. 4, Dec. 2018,
the confidence level of DT results and recommendations. Art. no. 430624, doi: 10.3389/fbuil.2018.00077.
[13] In-Depth Focus: Digital Twins, Global Railway Rev., Brasted, U.K., 2021.
However, from a research perspective, there are many chal- [14] A. Johnson, J. Heaton, S. Yule, S. Luke, D. Pocock, A. K. Parlikad, and
lenges in the railway industry for DT that require attention, J. Schooling, ‘‘Informing the information requirements of a digital twin:
especially related to DT setup, grouping and integration of A rail industry case study,’’ Proc. Inst. Civil Eng., Smart Infrastruct. Con-
struct., vol. 174, no. 2, pp. 33–45, Jun. 2021, doi: 10.1680/jsmic.20.00017.
layers or DTs if required, experimentation, calibration and
[15] J. Vieira, J. P. Martins, N. M. de Almeida, H. Patrício, and J. G. Morgado,
management, which is very difficult, including how to collect, ‘‘Towards resilient and sustainable rail and road networks: A systematic
transmit, process and use the DT data. literature review on digital twins,’’ Sustainability, vol. 14, no. 12, p. 7060,
Data for the rail sector at the macro level are mainly Jun. 2022, doi: 10.3390/su14127060.
[16] A. Bustos, H. Rubio, E. Soriano-Heras, and C. Castejon, ‘‘Methodology for
protected for use as commercial or trade secrets, while there the integration of a high-speed train in maintenance 4.0,’’ J. Comput. Des.
are databases with some basic rail performance data and data Eng., vol. 8, no. 6, pp. 1605–1621, Nov. 2021, doi: 10.1093/jcde/qwab064.
collection initiatives. Nevertheless, there is no micro-level [17] M. Spiryagin, J. Edelmann, F. Klinger, and C. Cole, ‘‘Vehicle sys-
tem dynamics in digital twin studies in rail and road domains,’’
data related to the condition and operation of a single sys- Vehicle Syst. Dyn., vol. 61, no. 7, pp. 1737–1786, Jul. 2023, doi:
tem [52]. The main constraint to the development of DTs in 10.1080/00423114.2023.2188228.
the rail sector is the availability of data. [18] S. Jeschke and R. Grassmann, ‘‘Development of a generic implementation
strategy of digital twins in logistics systems under consideration of the
In summary, to further assist researchers and practition- German rail transport,’’ Appl. Sci., vol. 11, no. 21, p. 10289, Nov. 2021,
ers, this study provides an initial overview of the creation doi: 10.3390/app112110289.
of DTs in the rail sector and serves as a basic framework [19] M. Attaran and B. G. Celik, ‘‘Digital twin: Benefits, use cases, challenges,
and opportunities,’’ Decis. Anal. J., vol. 6, Mar. 2023, Art. no. 100165, doi:
for the further development of DTs, with the aim of filling
10.1016/j.dajour.2023.100165.
the research gap. This study also found that the design of [20] S. Kaewunruen, M. AbdelHadi, M. Kongpuang, W. Pansuk, and
DTs in previous literature mainly focuses on the physical A. M. Remennikov, ‘‘Digital twins for managing railway bridge mainte-
infrastructure components. Since the movable component of nance, resilience, and climate change adaptation,’’ Sensors, vol. 23, no. 1,
p. 252, Dec. 2022, doi: 10.3390/s23010252.
railway infrastructure is not only switches, DTs for other [21] A. Kochan, ‘‘Digital twin of the ECTS application,’’ WUT J. Transp. Eng.,
movable devices can be part of future work. Movable devices vol. 131, pp. 87–98, Dec. 2020, doi: 10.5604/01.3001.0014.8199.
in railway infrastructure are mainly automated due to their [22] C. Qian, X. Liu, C. Ripley, M. Qian, F. Liang, and W. Yu, ‘‘Digital twin—
Cyber replica of physical things: Architecture, applications and future
limited functions and positions. However, future work can research directions,’’ Future Internet, vol. 14, no. 2, p. 64, Feb. 2022, doi:
focus on developing models and techniques for DTs that can 10.3390/fi14020064.
calculate and optimize the operation of movable devices in [23] B. R. Barricelli, E. Casiraghi, and D. Fogli, ‘‘A survey on
real time. Besides the main role of DTs to simulate, monitor digital twin: Definitions, characteristics, applications, and design
implications,’’ IEEE Access, vol. 7, pp. 167653–167671, 2019, doi:
and testing physical systems, they can also be developed in 10.1109/ACCESS.2019.2953499.
the future as a basis for managing devices in real time and as a [24] A. Consilvio, J. Solís-Hernández, N. Jiménez-Redondo, P. Sanetti, F. Papa,
data collector for providing policies and monitoring physical and I. Mingolarra-Garaizar, ‘‘On applying machine learning and simulative
approaches to railway asset management: The earthworks and track circuits
systems. In addition, DTs can be developed as a tool for case studies,’’ Sustainability, vol. 12, no. 6, p. 2544, Mar. 2020, doi:
testing the implementation of new technologies. 10.3390/su12062544.

VOLUME 12, 2024 108613

 E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

[25] S. Deng, L. Ling, C. Zhang, C. Li, T. Zeng, K. Zhang, and G. Guo, [43] L. De Donato, R. Dirnfeld, A. Somma, A. De Benedictis, F. Flammini,
‘‘A systematic review on the current research of digital twin in automo- S. Marrone, M. S. Azari, and V. Vittorini, ‘‘Towards AI-assisted digital
tive application,’’ Internet Things Cyber-Phys. Syst., vol. 3, pp. 180–191, twins for smart railways: Preliminary guideline and reference architec-
Jan. 2023, doi: 10.1016/j.iotcps.2023.04.004. ture,’’ J. Reliable Intell. Environ., vol. 9, no. 3, pp. 303–317, Sep. 2023,
[26] E. VanDerHorn and S. Mahadevan, ‘‘Digital twin: Generalization, charac- doi: 10.1007/s40860-023-00208-6.
terization and implementation,’’ Decis. Support Syst., vol. 145, Jun. 2021, [44] A. Pool, ‘‘Digital twins in rail freight. The foundations
Art. no. 113524, doi: 10.1016/j.dss.2021.113524. of a future innovation,’’ Dept. Elect. Eng., Math. Comput.
[27] E. Bernal, M. Spiryagin, E. Vollebregt, K. Oldknow, S. Stichel, S. Shrestha, Sci., Univ. Twente, Enschede, The Netherlands, 2021.
S. Ahmad, Q. Wu, Y. Sun, and C. Cole, ‘‘Prediction of rail surface damage [Online]. Available: https://purl.utwente.nl/essays/88902 and
in locomotive traction operations using laboratory-field measured and http://essay.utwente.nl/88902/1/Pool_MA_EEMCS.pdf
calibrated data,’’ Eng. Failure Anal., vol. 135, May 2022, Art. no. 106165, [45] J. Sresakoolchai and S. Kaewunruen, ‘‘Railway infrastructure maintenance
doi: 10.1016/j.engfailanal.2022.106165. efficiency improvement using deep reinforcement learning integrated with
[28] M. Kosacka-Olejnik, M. Kostrzewski, M. Marczewska, B. Mrówczyńska, digital twin based on track geometry and component defects,’’ Sci. Rep.,
and P. Pawlewski, ‘‘How digital twin concept supports internal trans- vol. 13, no. 1, Feb. 2023, Art. no. 2439, doi: 10.1038/s41598-023-29526-8.
port systems?—Literature review,’’ Energies, vol. 14, no. 16, p. 4919, [46] J. Sresakoolchai and S. Kaewunruen, ‘‘Track geometry prediction
Aug. 2021, doi: 10.3390/en14164919. using three-dimensional recurrent neural network-based models cross-
[29] J. Yang, Y. Sun, Y. Cao, and X. Hu, ‘‘Predictive maintenance for switch functionally co-simulated with BIM,’’ Sensors, vol. 23, no. 1, p. 391,
machine based on digital twins,’’ Information, vol. 12, no. 11, p. 485, Dec. 2022, doi: 10.3390/s23010391.
Nov. 2021. [47] S. H-Nia, J. Flodin, C. Casanueva, M. Asplund, and S. Stichel, ‘‘Predictive
[30] T. McDonald, M. Robinson, and G. Y. Tian, ‘‘Developments in 3D maintenance in railway systems: MBS-based wheel and rail life prediction
visualisation of the rail tunnel subsurface for inspection and monitor- exemplified for the Swedish iron-ore line,’’ Vehicle Syst. Dyn., vol. 62,
ing,’’ Appl. Sci., vol. 12, no. 22, p. 11310, Nov. 2022, doi: 10.3390/ no. 1, pp. 3–20, Jan. 2024, doi: 10.1080/00423114.2022.2161920.
app122211310. [48] I. Pombo, L. Godino, J. A. Sánchez, and R. Lizarralde, ‘‘Expectations and
[31] H. Feng, H. Lv, and Z. Lv, ‘‘Resilience towarded digital twins to improve limitations of cyber-physical systems (CPS) for advanced manufacturing:
the adaptability of transportation systems,’’ Transp. Res. A, Policy Pract., A view from the grinding industry,’’ Future Internet, vol. 12, no. 9, p. 159,
vol. 173, Jul. 2023, Art. no. 103686, doi: 10.1016/j.tra.2023.103686. Sep. 2020, doi: 10.3390/fi12090159.
[32] I. Errandonea, P. Ciáurriz, U. Alvarado, S. Beltrán, and S. Arrizabalaga, [49] Gartner. (2023). Top 10 Strategic Technology Trends. [Online]. Available:
‘‘Edge intelligence-based proposal for onboard catenary stagger ampli- https://www.gartner.com/en/newsroom/press-releases/2022-10-17-
tude diagnosis,’’ Comput. Ind., vol. 144, Jan. 2023, Art. no. 103781, doi: gartner-identifies-the-top-10-strategic-technology-trends-for-2023
10.1016/j.compind.2022.103781. [50] Gartner. (2022). What’s New in the 2022 Gartner HypeCycle for
[33] R. Dirnfeld, ‘‘Digital twins in railways: State of the art, opportuni- Emerging Technologies. [Online]. Available: https://www.gartner.
ties, and guidelines,’’ Bachelor Degree Project, Linneuniversitetet Kalmar com/en/articles/what-s-new-in-the-2022-gartner-hype-cycle-for-
Vaxjo, Växjö, Sweden, 2022. [Online]. Available: http://www.diva- emerging-technologies
portal.org/smash/record.jsf?pid=diva2%3A1674046&dswid=-2834 [51] The Logistics Trend Radar 6.0, DHL, Bonn, Germany, 2023.
[34] N. Bešinović, L. De Donato, F. Flammini, R. M. P. Goverde, Z. Lin, R. Liu, [52] B. Djordjević, A. S. Mane, and E. Krmac, ‘‘Analysis of dependency
S. Marrone, R. Nardone, T. Tang, and V. Vittorini, ‘‘Artificial intelligence and importance of key indicators for railway sustainability monitor-
in railway transport: Taxonomy, regulations, and applications,’’ IEEE ing: A new integrated approach with DEA and Pearson correlation,’’
Trans. Intell. Transp. Syst., vol. 23, no. 9, pp. 14011–14024, Sep. 2022, Res. Transp. Bus. Manage., vol. 41, Dec. 2021, Art. no. 100650, doi:
doi: 10.1109/TITS.2021.3131637. 10.1016/j.rtbm.2021.100650.
[35] S. Ghaboura, R. Ferdousi, F. Laamarti, C. Yang, and A. E. Saddik, ‘‘Dig- [53] V. Aksenov, A. Semochkin, A. Bendik, and A. Reviakin, ‘‘Utilizing dig-
ital twin for railway: A comprehensive survey,’’ IEEE Access, vol. 11, ital twin for maintaining safe working environment among railway track
pp. 120237–120257, 2023. tamping brigade,’’ Transp. Res. Proc., vol. 61, pp. 600–608, Jan. 2022, doi:
[36] A. Liberati, D. G. Altman, J. Tetzlaff, C. Mulrow, P. C. Gotzsche, 10.1016/j.trpro.2022.01.097.
J. P. A. Ioannidis, M. Clarke, P. J. Devereaux, J. Kleijnen, and D. Moher, [54] P. Armand and J. Tâche, ‘‘3D modelling and simulation of the impact of
‘‘The PRISMA statement for reporting systematic reviews and meta- wearing a mask on the dispersion of particles carrying the SARS-CoV-2
analyses of studies that evaluate healthcare interventions: Explanation virus in a railway transport coach,’’ Sci. Rep., vol. 13, no. 1, Jun. 2023,
and elaboration,’’ BMJ, vol. 339, Dec. 2009, Art. no. b2700, doi: Art. no. 8929, doi: 10.1038/s41598-023-35025-7.
10.1136/bmj.b2700. [55] C. A. Avizzano, G. Scivoletto, and P. Tripicchio, ‘‘Robust image stitch-
[37] H. Stripple, and S. Uppenberg. (2010). Life Cycle Assessment of ing and reconstruction of rolling stocks using a novel Kalman filter
Railways and Rail Transport—Application in Environmental Product with a multiple-hypothesis measurement model,’’ IEEE Access, vol. 9,
Declarations (EPDs) for the Bothnia Line. [Online]. Available: pp. 154011–154021, 2021, doi: 10.1109/ACCESS.2021.3128564.
https://www.ivl.se/download/18.694ca0617a1de98f473456/162841618 [56] A. Avsievich, V. Avsievich, N. Avsievich, D. Ovchinnikov, and
3076/FULLTEXT01.pdf A. Ivaschenko, ‘‘Railway track stress–strain analysis using high-precision
[38] B. Parbst, E. Dolmans, and T. F. Rasmussen. (2020). Digital Twin and accelerometers,’’ Appl. Sci., vol. 11, no. 24, p. 11908, Dec. 2021, doi:
Railway System Safety. [Online]. Available: https://safety.productions/wp- 10.3390/app112411908.
content/uploads/2020/02/Group_07.pdf [57] P. Bauer and W. Lienhart, ‘‘3D concept creation of permanent geodetic
[39] A. Kampczyk and K. Dybeł, ‘‘The fundamental approach of the digi- monitoring installations and the a priori assessment of systematic effects
tal twin application in railway turnouts with innovative monitoring of using virtual reality,’’ J. Appl. Geodesy, vol. 17, no. 1, pp. 1–13, Jan. 2023,
weather conditions,’’ Sensors, vol. 21, no. 17, p. 5757, Aug. 2021, doi: doi: 10.1515/jag-2022-0020.
10.3390/s21175757. [58] E. Bernal, Q. Wu, M. Spiryagin, and C. Cole, ‘‘Augmented digital twin for
[40] Y. Q. Sun, D. Nielsen, Q. Wu, W. Chua, and M. Spiryagin, ‘‘Design railway systems,’’ Vehicle Syst. Dyn., vol. 62, no. 1, pp. 67–83, Jan. 2024,
and safety analysis of a 11-waggon consist for transporting rails,’’ Aus- doi: 10.1080/00423114.2023.2194543.
tral. J. Mech. Eng., vol. 21, no. 5, pp. 1474–1488, Oct. 2023, doi: [59] A. O. Borjigin, J. Sresakoolchai, S. Kaewunruen, and J. Hammond,
10.1080/14484846.2021.2022576. ‘‘Digital twin aided sustainability assessment of modern light rail infras-
[41] M. Spiryagin, Q. Wu, O. Polach, J. Thorburn, W. Chua, V. Spiryagin, tructures,’’ Frontiers Built Environ., vol. 8, Jul. 2022, Art. no. 796388, doi:
S. Stichel, S. Shrestha, E. Bernal, S. Ahmad, C. Cole, and T. McSweeney, 10.3389/fbuil.2022.796388.
‘‘Problems, assumptions and solutions in locomotive design, traction and [60] L. Boscaglia, A. Boglietti, S. Nategh, F. Bonsanto, and C. Scema,
operational studies,’’ Railway Eng. Sci., vol. 30, no. 3, pp. 265–288, ‘‘Numerically based reduced-order thermal modeling of traction motors,’’
Sep. 2022, doi: 10.1007/s40534-021-00263-w. IEEE Trans. Ind. Appl., vol. 57, no. 4, pp. 4118–4129, Jul. 2021, doi:
[42] E. Krmac and B. Djordjević, ‘‘An evaluation of train control information 10.1109/TIA.2021.3077553.
systems for sustainable railway using the analytic hierarchy process (AHP) [61] A. Ekberg, E. Kabo, and R. Lundén, ‘‘Rail and wheel health man-
model,’’ Eur. Transp. Res. Rev., vol. 9, no. 3, pp. 1–7, Sep. 2017, doi: agement,’’ Wear, vols. 526–527, Aug. 2023, Art. no. 204891, doi:
10.1007/s12544-017-0253-9. 10.1016/j.wear.2023.204891.

108614 VOLUME 12, 2024

E. Krmac, B. Djordjevic: Digital Twins for Railway Sector: Current State and Future Directions

[62] E. Febrianto, L. Butler, M. Girolami, and F. Cirak, ‘‘Digital twinning of [83] M. V. Shevlyugin, A. A. Korolev, A. E. Golitsyna, and D. S. Pletnev,
self-sensing structures using the statistical finite element method,’’ Data- ‘‘Electric stock digital twin in a subway traction power system,’’
Centric Eng., vol. 3, Jan. 2022, Art. no. e31, doi: 10.1017/dce.2022.28. Russian Electr. Eng., vol. 90, no. 9, pp. 647–652, Sep. 2019, doi:
[63] M. M. Futai, T. N. Bittencourt, H. Carvalho, and D. M. Ribeiro, ‘‘Chal- 10.3103/s1068371219090098.
lenges in the application of digital transformation to inspection and [84] M. V. Shevlyugin, A. A. Korolev, A. O. Korolev, and I. A. Aleksandrov,
maintenance of bridges,’’ Struct. Infrastruct. Eng., vol. 18, nos. 10–11, ‘‘A digital model of a traction substation with two types of current,’’
pp. 1581–1600, Nov. 2022, doi: 10.1080/15732479.2022.2063908. Russian Electr. Eng., vol. 89, no. 9, pp. 540–545, Sep. 2018, doi:
[64] Q. Geng, Y. Wen, D. Zhang, J. Xiao, Y. Zhu, and L. Zhu, ‘‘Analysis 10.3103/s1068371218090134.
of electromagnetic coupling characteristics of balise transmission system [85] B. Ton, F. Ahmed, and J. Linssen, ‘‘Semantic segmentation of terrestrial
based on digital twin,’’ Appl. Sci., vol. 11, no. 13, p. 6002, Jun. 2021, doi: laser scans of railway catenary arches: A use case perspective,’’ Sensors,
10.3390/app11136002. vol. 23, no. 1, p. 222, Dec. 2022, doi: 10.3390/s23010222.
[65] M. Hamarat, M. Papaelias, and S. Kaewunruen, ‘‘Fatigue damage assess- [86] F. Uehan, ‘‘Technology for experiment/measurement to clarify phenomena
ment of complex railway turnout crossings via peridynamics-based specific to railway system,’’ Quart. Rep. RTRI, vol. 63, no. 2, pp. 84–87,
digital twin,’’ Sci. Rep., vol. 12, no. 1, Aug. 2022, Art. no. 14377, doi: 2022, doi: 10.2219/rtriqr.63.2_84.
10.1038/s41598-022-18452-w.
[66] C. Hong, J. Zhang, and W. Chen, ‘‘An integrated intelligent approach for
monitoring and management of a deep foundation pit in a subway station,’’
Sensors, vol. 22, no. 22, p. 8737, Nov. 2022, doi: 10.3390/s22228737.
[67] M. Ikeda, ‘‘Recent research and development activities in maintenance
technologies for electric railway power supply systems,’’ Quart. Rep.
RTRI, vol. 61, no. 1, pp. 6–10, 2020, doi: 10.2219/rtriqr.61.1_6.
[68] S. Kaewunruen and Q. Lian, ‘‘Digital twin aided sustainability-based life-
cycle management for railway turnout systems,’’ J. Cleaner Prod., vol. 228,
pp. 1537–1551, Aug. 2019, doi: 10.1016/j.jclepro.2019.04.156. EVELIN KRMAC was born in Koper, Slovenia,
[69] S. Kaewunruen, S. Peng, and O. Phil-Ebosie, ‘‘Digital twin aided sustain- in 1967. She received the B.S., M.S., and Ph.D.
ability and vulnerability audit for subway stations,’’ Sustainability, vol. 12, degrees in computer science from the Faculty of
no. 19, p. 7873, Sep. 2020, doi: 10.3390/su12197873. Computer and Information Science, University of
[70] S. Kaewunruen, J. Sresakoolchai, and Y.-H. Lin, ‘‘Digital twins for man- Ljubljana, in 1991, 1997, and 2005, respectively.
aging railway maintenance and resilience,’’ Open Res. Eur., vol. 1, p. 91, From 1990 to 1994, she was a Teacher
Nov. 2021, doi: 10.12688/openreseurope.13806.2. in mathematics, physics and informatics with
[71] S. Kaewunruen, J. Sresakoolchai, W. Ma, and O. Phil-Ebosie, ‘‘Digital the State Scientific Secondary School—Lyceum
twin aided vulnerability assessment and risk-based maintenance planning ‘‘France Prešeren,’’ State College of Education
of bridge infrastructures exposed to extreme conditions,’’ Sustainability, ‘‘Anton Martin Slomšek,’’ and Vocational School
vol. 13, no. 4, p. 2051, Feb. 2021, doi: 10.3390/su13042051. ‘‘J.Stefan,’’ Trieste, Italy. Since 1994, she has been with the Faculty of Mar-
[72] J. Karolak, ‘‘Interface and connection model in the railway traffic con-
itime Studies and Transport, University of Ljubljana, Slovenia. She started
trol system,’’ Arch. Transp., vol. 58, no. 2, pp. 137–147, Jun. 2021, doi:
as an Assistant Professor. From 1997 to 2013, she was a Senior Lecturer,
10.5604/01.3001.0014.9086.
[73] V. V. Krishna, Q. Wu, S. Hossein-Nia, M. Spiryagin, and S. Stichel, from 2013 to 2022, an Assistant Professor, and an Associate Professor, since
‘‘Long freight trains & long-term rail surface damage—A systems per- 2022. She has been teaching and researching in the fields of computer and
spective,’’ Vehicle Syst. Dyn., vol. 61, no. 6, pp. 1500–1523, Jun. 2023, information sciences and intelligent transport systems and their application
doi: 10.1080/00423114.2022.2085584. in transport and logistics. She is the author/co-author of more than 60 sci-
[74] H. Li, Q. Zhu, L. Zhang, Y. Ding, Y. Guo, H. Wu, Q. Wang, R. Zhou, entific and professional articles, two books, and five book chapters. Her
M. Liu, and Y. Zhou, ‘‘Integrated representation of geospatial data, model, research interests include application of computer and information science
and knowledge for digital twin railway,’’ Int. J. Digit. Earth, vol. 15, no. 1, technologies and the field of transport and logistics, use of multi-criteria
pp. 1657–1675, Dec. 2022, doi: 10.1080/17538947.2022.2127949. decision making techniques for evaluation and optimization purposes in
[75] M. C. Magnanini and T. A. M. Tolio, ‘‘A model-based digital twin to transport, and intelligent transport systems and services.
support responsive manufacturing systems,’’ CIRP Ann., vol. 70, no. 1,
pp. 353–356, 2021, doi: 10.1016/j.cirp.2021.04.043.
[76] A. Marra, I. Trizio, G. Fabbrocino, and F. Savini, ‘‘Digital models for e-
conservation: The HBrIM of a bridge along the Aterno river,’’ Sci. Res. Inf.
Technol., vol. 11, no. 2, pp. 83–96, 2021.
[77] A. Meixedo, J. Santos, D. Ribeiro, R. Calçada, and M. D. Todd,
‘‘Online unsupervised detection of structural changes using train–induced
dynamic responses,’’ Mech. Syst. Signal Process., vol. 165, Feb. 2022,
Art. no. 108268, doi: 10.1016/j.ymssp.2021.108268.
[78] S. Milovanovic, S. Strobl, P. Ladoux, and D. Dujic, ‘‘Hardware- BOBAN DJORDJEVIC was born in Nis, Serbia,
in-the-loop modeling of an actively fed MVDC railway systems of in 1988. He received the B.S. and M.S. degrees
the future,’’ IEEE Access, vol. 9, pp. 151493–151506, 2021, doi: in railway engineering from the University of Bel-
10.1109/ACCESS.2021.3125050. grade, in 2011 and 2012, respectively, and the
[79] N. Pillai, J. Y. Shih, and C. Roberts, ‘‘Sensor selection and placement for Ph.D. degree in rail transportation from the Uni-
track switch condition monitoring through validated structural digital twin versity of Ljubljana, in 2018, on the topic of
models of train–track interactions,’’ Eng. Proc., vol. 10, no. 1, p. 49, 2021, evaluating the impact of train control systems on
doi: 10.3390/ecsa-8-11297. railway capacity. He is currently a Researcher with
[80] S. Qiu, C. Mias, W. Guo, and X. Geng, ‘‘HS2 railway embankment the Division of Transport Planning, Royal Institute
monitoring: Effect of soil condition on underground signals,’’ Social Netw. of Technology (KTH), Stockholm. Before joining
Appl. Sci., vol. 1, no. 6, Jun. 2019, Art. no. 537, doi: 10.1007/s42452-019-
KTH, he was a Postdoctoral Researcher with the Trinity College Dublin,
0552-y.
working on projects related to monitoring the environmental impact of
[81] M. Rahman, H. Liu, M. Masri, I. Durazo-Cardenas, and A. Starr,
‘‘A railway track reconstruction method using robotic vision on a mobile heavy-duty vehicles. His research interests include the automation and
manipulator: A proposed strategy,’’ Comput. Ind., vol. 148, Jun. 2023, digitalization of railways in terms of planning and managing sustainable
Art. no. 103900, doi: 10.1016/j.compind.2023.103900. rail transport. His current research interests include sustainability in rail
[82] D. A. Ramatlo, D. N. Wilke, and P. W. Loveday, ‘‘Digital twin hybrid transport, automated train operation, innovative signaling technologies,
modeling for enhancing guided wave ultrasound inspection signals in multi-criteria decision making in rail transport, and optimization and infras-
welded rails,’’ Math. Comput. Appl., vol. 28, no. 2, p. 58, Apr. 2023, doi: tructure optimization and infrastructure.
10.3390/mca28020058.

VOLUME 12, 2024 108615