1 Introduction

The efficiency and effectiveness of the railway system is measured by factors such as its availability, reliability, and safety performance. To increase the competitiveness of the railway sector with other means of transport such as aviation, the railway efficiency should be improved through unified key performance indicators (KPIs) related to increased capacity and reduced operational expenditure (Goya et al. 2018). It is required by railway organisations that parameters such as resilience of the infrastructure are quantifiable through key figures for measuring, optimisation and improvement measures related to the parameter (Frid and Grehn 2016). To take strategic decisions related to operation and maintenance of assets in railway organisation, it is essential to track and improve the performance of current strategies (Al-Douri et al. 2016).

The railway system is a complex technical system-of-systems with multiple interconnected systems (Uzuka 2023). A system is a set of interrelated elements considered in a defined context with a view of achieving a given objective (IEV 2019). A system-of-systems (SoS) is a set of operationally and managerially independent systems that are operated together to achieve a given objective (Jamshidi 2008). The railway system has two primary systems: rolling stock and infrastructure, each interdependently contributing to the overall performance. The infrastructure further includes track system, signalling and control systems, overhead catenary system, and switches and crossings. The rolling stock consists of fleet of vehicles, their components, and subcomponents. Additionally, there are others such as asset management, operational planning and management, financial and administrative, and communication and control systems. Neglecting the interdependencies between these systems can impact the Key Performance Indicators (KPIs) of the railway SoS. Focusing on single system KPIs, might lead to sub-optimisation of overall system performance (Li et al. 2019).

A holistic approach with system thinking can be beneficial in improving the efficiency and effectiveness of the railways through asset management. Asset management is a set of coordinated activities in an organisation to realise value from assets (ISO 2014). The reliability, availability and safety of the railway system relies on the subsequent reliability of its infrastructure and rolling stock assets. In addition to improved effectiveness and efficiency, asset management in railways is expected to provide benefits by demonstrating quantifiable improvements in areas such as financial performance, risk management, services, asset investment decisions, compliance, social responsibility, organisational sustainability, and reputation (ISO 2014). These benefits are crucial for the railway system that has a wide social, economic, and environmental impact in the transport ecosystem all over the world.

In the context of developing an asset management regime for a holistic railway system, the specifications related to the System of Interest (SoI) should be clearly articulated (Nielsen et al. 2015). In system engineering, SoI is the system whose lifecycle is taken into consideration (Ramtahalsing et al. 2020). When specifying a SoI, there are various dimensions that should be considered, such as, the stakeholders, the components,the nature of relationships and interactions between the components, the functions and the variability throughout the system lifecycle (Kinder et al. 2012). An enabling system is a system that complements the SoI during its lifecycle stages but does not necessarily contribute to its function during operation (ISO/IEC/IEEE 2015).

The benefits of asset management can be quantified using system KPIs that must be continuously monitored and improved (Silva et al. 2020). The use of technology related to digitalisation and Industrial Artificial Intelligence (IAI), can enable the analytics required for monitoring of these KPIs. These analytics enabled by IAI and digitalisation, can augment the domain knowledge in the railway system, for an effective and efficient asset management regime that integrates data from diverse sources in the railway system to develop and share information, knowledge, and context models through a common platform, that is designed towards stakeholder expectations (Kumari et al. 2021). In this context, a stakeholder is a person or organisation that can affect, be affected by, or perceive to be affected by a decision or activity (ISO 2014).

With advancement in technology, availability of elegant data acquisition systems, and faster and cheaper data storage and computational capabilities, a popular trend in asset management of railways in the last decade has been related to a data-driven approach towards analytics (Olsson and Pétursson 2021, Aronsson and Pétursson 2022, Alnaggar 2022, Kuusk and Gao 2021). However, it is observed that the analytics that is developed to augment the asset management of railways, typically concentrates on a specific organizational performance indicator, potentially missing the opportunity to adopt a more holistic perspective (Bouraima and Qiu 2023, Merishna Ramtahalsing et al. 2021). However, to achieve optimised performance of the railway SoS, the relationships and interactions of the interdependent systems need to be considered in the development of analytics (Arnold and Wade 2015). Challenges related to consideration of such relationships and interactions between inherent systems in the railway SoS have been highlighted in multiple studies in literature (Jakubeit et al. 2020, Rahman et al. 2022, Santos et al. 2019).

Additionally, the data-driven approaches have been observed to focus on challenges related to data and information management and delivering the right information at the right time to the right people (Karim et al. 2016, Berryman and Cheung 2020). Therefore, the focus of research in a data-driven approach, remains on addressing issues specific to data such as quality and accuracy, integration, privacy and security, volume, velocity, variety, governance, scalability, storage and retrieval, preprocessing, selection of modelling techniques, visualisation and so on. Although these are important issues that must be addressed, when trying to extract useful information from large amounts of asset data, the usability of these studies in real-world applications is limited (Hu et al. 2022). This is mainly due to factors such as the complexity of the railway SoS, and the lack of a holistic perspective. A systematic literature review of data-driven models for predictive maintenance of railway tracks has emphasized the importance of including a cost function, related to system KPIs in analytics, instead of solely focusing on model accuracy (Xie et al. 2020).

Finally, the rapid growth of data in industrial applications, driven by technology development and company investments, presents substantial potential for efficient asset management; however, there is a need to combine data-driven processes with domain expertise (Li et al. 2021, Nie et al. 2023, Schuster et al. 2022). It has been observed that stakeholder involvement in early stages of railway projects improves the project efficiency and effectiveness as it considers variable needs and expectations of stakeholders to track performance (Baharuddin, Azizi, and Piri 2021, KANG et al. 2021).

Problem statement The performance of the railway system can be measured and improved through KPIs related to the expected system performance. Since railways is a complex technical SoS with multiple interconnected systems, a SoS approach is required for the identification of KPIs that are indicators to a holistic system performance rather than to sub-optimisation of individual systems. Asset management is expected to improve the effectiveness and efficiency of the railway system. The SoS approach provides a holistic view to asset management of railway system.

Additionally, the effectiveness and efficiency of asset management itself can be improved through augmenting asset management decision support using data-driven technologies. However, when utilising enabling technologies to augment asset management in railways, the focus of the analytics remains on data-driven approaches, rather than overall system KPIs. Implementing such an augmented asset management regime with a SoS approach for the railway system driven by system KPIs and enabled by technology requires a seamless integration of multidisciplinary concepts such as asset management, operation and maintenance, data-driven technology, and domain expertise. There is a need for a systematic framework that provides clear classifications and relationships between these concepts to facilitate the development of an augmented asset management regime for the railway system.

Contribution This paper provides a performance-driven framework with a SoS approach for augmented asset management in railways. The proposed framework has been developed as a result of literature survey, interviews with railway organisations, and development, and execution of real-world projects related to asset management of railways. This approach is intended to consider the development of an asset management regime that augments the enabling technologies to augment the asset management of railway, while maintaining the focus on railway system KPIs. To adopt such a performance-driven approach, it is imperative to establish a top-down and bottom-up relationship between the KPIs of the railway SoS and its inherent systems. The proposed framework is intended to serve as a guiding tool for railway organizations during the design and implementation of an asset management system. The use of this framework is expected to increase the asset management maturity and digital maturity of the organisation. The proposed framework is also expected to facilitate researchers within asset management of railways to increase the technological readiness level and business readiness level of their developed solutions. This can be done by mapping the output of individual data-driven solutions to the components provided in the framework, to assess their contribution to the system KPIs.

2 Related works

2.1 Holistic approaches in railways

The intended outcome of this section is to identify the research gap in the development and implementation of holistic approaches for the railway system. The search keywords used for this review were ‘railway’ and ‘holistic’ in the title of the article. A list of reviewed articles associated to holistic approaches in railway have been provided in Table 1. As seen in the table, the system of interest (SoI) for most of the reviewed articles, is railway infrastructure, or a single component of the railway infrastructure. There is one article (S. No.2), that considers a holistic approach towards the entire railway system; however, the considered aspect of interconnectedness, is in reference to traffic and transport planning, and not asset management.

Table 1 A list of articles on the topic – ‘a holistic approach in railways’
Full size table

Another article (S.no.8) proposes a holistic asset engineering and decision management framework, for power systems, that considers multiple aspects related to asset management and asset lifecycle to align business objectives and asset objectives. However, this study does not consider the challenges and possibilities posed by the integration of technology within asset management. This shows that there is a lack of available research on a holistic approach for the asset management of railway system, enabled by IAI and digitalisation.

2.2 Data-driven approaches in railways

The intended outcome of this section is to identify the need for a performance-driven approach for asset management in railways. This section presents a review of data-driven approaches for asset management analytics in railways.

The data-driven approaches for asset management focus on challenges and possibilities posed by the availability of asset data. The data related to assets are generally acquired for a specific purpose, however, such data can also be explored and used for a broader perspective that was originally intended (Olsson and Pétursson 2021). Studies on data-driven decision-making approaches for asset management are largely focussed on aspects such as IT/OT integration (Kuusk and Gao 2021), data governance (Azad 2022), accuracy and completeness of data (Okwori et al. 2021), and asset data management and logistics (Alnaggar 2022). A literature review on the interoperability of the Dutch railway system states that there is a demand for an integral approach for decision making that could enhance collaborations between different stakeholders with different interests within the railway system (Jakubeit et al. 2020). At times, cutting-edge technology, developed from a singular perspective, encounters challenges in real-world applications because it isn’t aligned with other interconnected railway subsystems (Rahman et al. 2022, Santos et al. 2019).

State-of-the art technology that is used for augmenting the decision support related to asset management of railway SoS are shown in Fig. 1. The research on data-driven asset management focusses on challenges and possibilities posed by these technologies and their integration for operation and maintenance of assets. However, they do not take a holistic approach where the outcome of the analytics should align with the objectives of the railway system.

Fig. 1
figure 1

State-of-the-art technology related to enablement of asset management of railway SoS

Full size image

2.3 Motivation for a performance-driven approach in railways

The concept of performance-driven asset management, as highlighted by (Evans 2008), suggests that leveraging performance measures that align with transportation assets' requirements can effectively convey engineering needs to both internal and external stakeholder organization. This implies that through performance-driven asset management, organizations can foster better collaboration and ensure that engineering needs are effectively met. Doran (2015), in their work on asset management trends, made a remark, that, as progressive organizations shift their emphasis from compliance to performance, there lies the potential for exponential growth in the application, focus, and business impact of performance-driven asset management. This approach involves effectively stewarding assets, delivering reliable and durable services or production processes, while prioritizing safety, effectiveness, and efficiency. In a more recent study examining the total cost of ownership of industrial assets, researchers have emphasized the potential for developing a decision support system by analysing the technical performance of these assets (Roda and Garetti 2020).

The above-mentioned studies have highlighted the significance of a performance-driven asset management framework.

The international standard for asset management (ISO 2014), states the need for a framework to support the setting up of specifications of an asset management policy, and asset management objectives. Such a framework is facilitating the asset management objectives to,

  1. 1.

    Be consistent to the organisation’s plans and objectives and policies.

  2. 2.

    Be appropriate to the nature of assets in the organisation and their operation.

  3. 3.

    Be able to provide documentation, communication, and availability to stakeholders.

  4. 4.

    Be able to be implemented and periodically reviewed.

To achieve the asset management objectives, it is required to identify the decision-making criteria, and establish processes to manage the assets over their lifecycle. These guidelines from ISO 55000 have been considered while developing the proposed framework.

3 Materials and methods

The proposed framework is based on the best practices in asset management that are included in international standards for asset management, ISO55000, ISO 55001, and ISO 55002. In addition to this, a literature survey of related research areas, interviews from domain experts in railways system, and the knowledge gained from the development of project demonstrators, was used as the knowledgebase for development of the proposed framework.

3.1 Literature survey

This research aimed to pinpoint a significant challenge within data-driven analytics for decision support in railway asset management. To achieve this, a systematic literature review methodology was employed to assess the state of the art in railway asset management and analytics. Given the extensive literature in the railway domain, our focus was confined to the following relevant areas:

  • Holistic approaches in railways

  • Systemic approaches in railways

  • Railway asset management

  • Data-driven approaches in railways

  • Performance-driven approaches in railways

We examined the applied methodology, limitations, and suggestions for future work in the selected papers. Concerning data-driven approaches in railways, we considered papers published from 2019 onwards, as this timeframe aligns with the latest trends and challenges in the field. For performance-driven, holistic, and asset management aspects in railways, we did not impose time restrictions due to limited number of literature in these domains.

3.2 Interviews and project demonstrators

This research was conducted as a part of the AIF/R—(AI Factory for railways) project, at the Division of Operation and Maintenance Engineering at Luleå University of Technology. The AIF/R project aims to provide a common and distributed platform for different railway organisations for data and model sharing (Karim, Galar, and Kumar 2021). Several railway organisations contributed with domain knowledge and data, for development of the platform. Figure 2 shows the sample population, and the structure of the conducted interviews with domain experts within railway organisations.

Fig. 2
figure 2

Data collection through interviews and questionnarire for the proposed framework

Full size image

Additionally, the following research projects conducted as part of the AIF/R platform formed a basis for the identified need and building modules of the framework. This was based on the statement of project requirements, development of analytics, and feedback on the developed analytics from stakeholders in railway organisations.

  1. 1.

    Development of a fleet-based approach to enhance model performance and compensate for lack of data for railway assets.

  2. 2.

    Development of a Metaanalyser platform and toolkit for preliminary data analysis, and model selection for large volumes of incoming data for railway assets.

  3. 3.

    Development of a common and distributed platform for stakeholders in railway organisation for data and model sharing.

4 Results and discussions

The main contribution of this paper is the proposed performance-driven framework for augmented asset management of railway system. This section contains the description of the components of the framework, and the implementation of the framework on a case study on railway rolling stock.

4.1 Performance-driven framework for augmented asset management of railway system

As shown in Fig. 3, the proposed framework has three main components. The first and central component of the framework is based on the key principles of asset management system as described by ISO 55000. The second component is related to the railway system requirements. The third component is related to data and information enabled augmented decision-support for asset management.

Fig. 3
figure 3

A performance-driven framework for augmented asset management of railway system

Full size image

The proposed framework lists and describes all the crucial components of the asset management system. For each component, the framework also describes its interconnection with the railway system module and augmented decision-support module.

4.1.1 Context and leadership

The context and leadership blocks in the framework refer to strategic activities, objectives, and decisions in asset management.

The context of a system is defined by the system requirements and stakeholder information from the railway system. The scope of asset management should be aligned with the external and internal context of the railway organisation. In a system-of-systems (SoS) perspective, the internal and external context can be defined by identifying the system of interest (SoI), the enabling systems and their requirements and relationships. The railways in Sweden, is a multistakeholder system, where the railways are owned operated managed, and maintained by different organisations. Each of these stakeholder organisations have requirements specific to the business objectives of their organisation.

The stakeholder needs and expectations are aligned with the system performance goals. Figure 4 shows different type of internal and external stakeholders identification based on a SoS perspective for railway rolling stock system. Consideration to isolated needs of stakeholders, might lead to development of incomplete analytical solutions that may benefit one organisation but do not have a positive impact on the system KPIs.

Fig. 4
figure 4

Different types of stakeholders for the rolling stock system in railways

Full size image

The asset management system and its objectives are derived from system and stakeholder requirements. Additionally, the asset management objectives must also consider the specified context of the organisation and the scope of asset management in the organisation. This alignment ensures the relevance and validity of the analytics within the requirements, limitations,  and constraints posed by the railway system.

The choice of technology to augment the asset management decision support should, align with the system requirements. This ensures that the selected technology addresses the right problems and fits into the operational, governance, safety, sustainability, and regulatory constraints posed by the railway SoS. While selecting these technologies, the drivers, enablers, and constraints of the railway system must be considered. Additionally, factors such as the technology readiness level, business readiness level, change management, the digital transformation phase of the organisation, and the compatibility of the technologies with each other and the railway SoS must be considered.

The role of leadership is to demonstrate commitment in supporting the implementation and continual improvement of the assets, asset management, and the asset management system. The leadership responsibilities include stakeholder engagement, creating a vision, mission and goal, alignment of asset management objectives to organisational goals, and provide resources, communication and integration required for asset management implementation alongside business processes.

The asset management policy includes the principle that guides asset management in the organisation. Authorisation of the asset management policy is also the responsibility of the leadership. The policy provides guiding principles for legal and regulatory requirements, asset management support, reporting and evaluation, and continual improvement.

The asset management policy and leadership responsibilities are based on the system requirements, and stakeholder requirements. They also serve as critical input parameters for technology selection and enablement in the organisation.

4.1.2 Planning and support

The planning and support blocks in the framework refer to tactical activities, objectives, and decisions in asset management.

Planning includes formulating asset management objectives from the organisational objectives. It also determines the activities and resources to achieve those objectives. Planning, iteratively takes a top-down and bottom-up approach, moving back and forth from strategic decisions to asset performance. An asset management plan also includes identification of risks and opportunities and actions to address them.

The resources required to implement the asset management plan are considered in the support block. The required support is identified against the planned activities for asset portfolio, asset management and asset management system. Support is also related to development of competence, communication, and awareness around asset management and organisational objectives within the organisation and its stakeholders.

The planning and support are based on the nature of railway assets and system KPIs. Augmented decision support is aimed to monitor and optimise several parameters to improve the KPIs, that are a measurement of the extent of fulfilment of system requirements and objectives. These KPIs can be related to requirements that are system-based, stakeholder-based, operational, financial, regulatory, user-based, environmental, compliance-based, risk-based and/or liabilities-based.

KPIs can be monitored through relevant data acquisition. With advanced condition monitoring, data acquisition techniques, and storage capability, huge amount of asset data is continuously generated and stored. To maintain an optimised digital infrastructure, it becomes crucial that data acquisition and analytics development are specifically relevant to these KPIs.

There are some unique features related to nature of assets in the railway system that must be considered during planning and support. Lifecycle management of assets is a key principle in asset management. In the context of railway system, lifecycle management is particularly important due to the system complexity. The railway system has several inherent items that are at different lifecycle stages. The typical lifecycle stages for engineering systems are planning, acquisition, operation, maintenance and decommission. However, when planning an effective and efficient asset management regime, the substages within each of these lifecycle stages of the asset must also be considered.

Within the operation and maintenance phase of an asset, the asset failure rate may vary due to changes throughout the asset life, such as change in ownership or operational contexts, number of maintenance actions performed, quality of maintenance etc. Consideration to these factors contribute to assignment of meaningful maintenance and safety thresholds for KPIs depending on the lifecycle stage.

Railway infrastructure in Sweden is owned operated and managed by government organisations, which puts unique constraints on the infrastructure assets due to government regulations. The availability requirements on railway infrastructure are also high and have limited flexibility in terms of adjusting operation and maintenance schedules. The rolling stock assets in railways are designed specifically, according to the environmental conditions and load requirements for passenger and freight in the region of operation. Railway assets generally have a long lifespan of 20 to 30 years, which makes it challenging to track the long-term operation and maintenance history of assets throughout their lifetime and through multiple contractual chains.

The characteristic lifecycle and nature of railway assets poses unique challenges related to augmented decision support. Some of these challenges are: (1) large-scale data acquisition for widespread railway assets, (2) inability to confirm data quality as multiple organisations generate and own the asset data, (3) lack of sufficient failure data to model the behaviour of railway assets etc. Such specific challenges related to asset management of railways require unique approaches and solution such as the metanalysis approach and the fleet management approach.

The metanalysis approach is based on the preliminary automated data preparation, feature selection, model selection and model evaluation for huge volumes of continuously generated asset data. This approach is expected to increase the agility of the asset management analytics and enable an efficient asset management regime through asset lifecycle.

The fleet management approach to augmented decision support in railways is based on the grouping of assets into a fleet/population, that are put together to achieve an expected outcome. This approach can be useful when there is insufficient failure data to model or predict the asset failures for asset management planning.

4.1.3 Operation, performance evaluation and monitoring

The operation and performance evaluation and monitoring blocks in the framework refer to operational activities, objectives, and decisions in asset management. This includes transforming the asset management plan into activities, implementation, risk assessment, performance evaluation of asset management and improvement.

A risk assessment of the asset management activities should be done before implementation. Risks associated to asset management can include factors such as, (1) organisational culture or leadership that inhibits the implementation of asset management, (2) misalignment between the requirements of different sub systems, (3) financial and non-financial losses due to asset management decisions, and (4) inefficient improvement in processes. The proposed framework is designed to mitigate these associated risks, by considering these factors at an early stage of development of the asset management plan.

The assessment of asset management can be done by assessing the traceability and consistency between the system requirements and asset performance. The currency, accuracy, consistence, and completeness of this traceability are aspects against which the performance is evaluated. Monitoring, internal audits, and management reviews can be conducted to measure the aspects for assessment of asset management.

Improvement in asset management can be done through various means such as (1) adoption of new technologies, (2) training and development to improve resource quality and competence, (3) improved control over risks (4) efficient planning and scheduling of asset management activities (5) efforts to reduce human error and (6) performance monitoring strategies and plans.

4.2 Implementation of the proposed framework on a case study in railways

In this section, the implementation of the proposed framework is demonstrated through a case study on fleet asset management of railway rolling stock.

4.2.1 Case description

The organisation that owns railway vehicles in Sweden, is owned by 20 regions across the country. It provides several different fleets of vehicles to the regional owners. The larger regions purchase their own fleets, while the smaller regions rent the fleet from the vehicle owners. In addition to the vehicles the vehicle owners also own the pool of so-called, high value components (HVC) that are configured in these fleets. A component qualifies for being a HVC based on different factors such as procurement cost, long procurement time, etc. The vehicle owners are responsible for maintenance actions that are marked under the category ‘heavy maintenance actions’ for the vehicles in the fleet as well as the HVCs. For certain fleets when they do not have the contract for heavy maintenance actions for the HVCs, they still own the pool of the HVCs for the fleet. Each fleet has its own pool of HVCs. This is due to the heterogenous nature of different fleets. The HVC used in one fleet is not compatible to be configured in other fleets. The maintenance contract for HVCs is then given to maintenance depots. The vehicle owner has contract with only single depots to repair each HVC type.

4.2.2 Implementation

The requirements of the rolling stock system have been described in Fig. 5. The system requirements, asset management objectives and the KPIs for planning and support were identified after several rounds of iterative interviews with railway vehicle owner organisation in Sweden. The analytics for augmented decision support were driven by the identified KPIs mentioned in Fig. 5. Maintenance planning and provision of maintenance support was prepared based on this augmented decision support.

Fig. 5
figure 5

Figure describing the system requirements of the railway rolling stock system

Full size image

Number of unplanned failures were identified from the number of corrective maintenance actions performed for the HVCs per million kilometres of operation, as shown in Fig. 6. It was then investigated if there are certain vehicle operators or maintenance depots that have unusual number of unplanned failures. Figure 7 shows the distribution of failures for each operator and maintenance depot alongside the kilometres operated by the operators and the maintenance depots. The unplanned failure distribution for the operators and depots are relative to the operated kilometres. This shows that the unplanned failures are evenly distributed across different operators and depots with respect to the number of operating kilometres.

Fig. 6
figure 6

KPI card for HVC showing operated kilometeres versus number of corrective maintenance actions

Full size image
Fig. 7
figure 7

Visualisation of distribution of unplanned failures for different operators and maintenance depots

Full size image

The maintenance process for heavy maintenance actions on HVCs is described in Fig. 8. The HVCs are demounted from different vehicles in the fleet at a train depot managed by the fleet operator. The demounted HVCs are sent to the HVC pool or directly to the maintenance depot, depending on the logistics plan. The repaired components are again sent back to the pool or directly to the train depots for mounting on the vehicles. When a vehicle arrives at the train depot for demounting a HVC, there should already be a repaired HVC available for mounting.

Fig. 8
figure 8

The process for incoming and outgoing high value components in the pool

Full size image

Based the process of heavy maintenance actions on HVCs, a maintenance plan was formulated as shown in Fig. 9 with the help of augmented decision support. The data-driven techniques are used for (1) forecasting the number of backlogs in the planned PM of HVCs for the planning horizon (e.g. 7 years), (2) forecasting on the number of unplanned failures, (3) estimation of the turnaround time, and (4) calculations of the number of planned PM over the planning horizon based on current PM interval. Domain specifications on the required accuracy and precision of the forecasting models, aids to provide an optimised rather than high accuracy, high precision forecasts. This reduces the computational complexity of the required digital infrastructure, at the same time utilising explainable modelling techniques that are aligned with system requirements and are implementable in the real-world scenario.

Fig. 9
figure 9

The provision of augmented data support for asset management planning

Full size image

The outcome of the implemented plan has not been analysed within the context of this paper due to time restrictions. The implementation of the maintenance plan over the planning horizon requires long term monitoring and assessment, which is the future research direction of this work.

5 Conclusion

This paper develops and proposes a performance-driven framework with a system-of-systems approach for augmented asset management of railway system. Asset management is expected to increase the efficiency and effectiveness of the railway system. The system-of-systems approach brings in a holistic perspective that prevents sub-optimisation of individual assets and systems and prioritises the overall system requirements. The concept of augmented asset management is based on the notion of utilising enabling digitalisation and industrial artificial intelligence technologies to augment the domain expertise in railway system and increase the level of automation in the decision making related to asset management. A performance-driven approach ensures the focus of asset management on the system performance. The proposed framework, constitutes and provides the important blocks of artefacts that need to be considered for a performance-driven approach to asset management. This research work has been conducted through (1) following best practices in asset management as stated by the ISO55000 series international standards for asset management, (2) literature survey on performance-driven and data-driven approaches in railways and, (3) as the part of solution to real -world case studies in railways. The case studies were conducted as a part of the AI/Factory for railways project which is a consortium of multiple railway organisations within Sweden, that includes infrastructure managers, vehicle owners, operators, and maintenance service providers.

The provision of a framework that is focussed on the railway system performance, using a system-of-systems approach and augmented decision support for asset management, provides a handrail and a checklist of important aspects that must be considered as well as the relationship between these aspects. Adopting such a framework-based approach for development of an asset management regime for the railway system considers risks, readiness levels, and system interdependencies at an early stage of the development of asset management plan. Additionally, the performance-driven framework contributes to reduced complexity of analytics and requirements on the digital infrastructure such as data acquisition and data storage by directing a top-down approach where analytics is developed to explain system key performance indicators. The suggested future work in this direction are: (1) comparison of implementation of a performance-driven and a data-driven approach in terms of parameters such as computational complexity, resources required in development of analytics, technology readiness level, and business readiness level of the solution and (2) evaluation of the asset management and maintenance plan developed using the proposed framework, through lifecycle costing analysis of assets and identifying areas for improvement.