Towards a Reference Model for Multimodal Transport Networks

Daniel Zöttl

, Alexander Granig

, Benjamin Schwendinger

and Sebastian Schlund

1,2

Fraunhofer Austria Research GmbH, Theresianumgasse 7, Vienna, Austria

Institute of Management Science, TU Wien, Theresianumgasse 27, Vienna, Austria

Keywords: Multimodal Transport Network, Reference Model, Relational Data Model, Transport Planning.

Abstract: A multimodal transport network is a widely used form of transport infrastructure. The ability to describe

different modes of transport, taking into account many different attributes, requires a structured model. This

paper outlines the requirements for the description of a multimodal transport network in the form of a

reference model. To this end, expert interviews were conducted with various expert groups from companies,

researchers from the field of transport planning, traffic management system providers and internationally

active logistics service providers to ensure that the reference model is suitable for practical applications.

Furthermore, an approach to transform the attributes into a relational data model including entities and

cardinalities is described and the challenges encountered are highlighted: different data formats, different

stakeholders and insufficient data availability. Finally, the application of the reference model as data base in

a practical real-world scenario is presented.

1 INTRODUCTION

The availability of efficient transport systems is

essential for the organization of economic processes

at both national and continental levels. A well-

developed transport infrastructure supports the

sustainable growth of economies by facilitating the

movement of goods and meeting society's increasing

demands. One modern approach to transport

infrastructure is the multimodal transport network

(MTN), which comprises international transport

corridors (ITCs). These corridors enhance

interregional and transcontinental economic

relations, contributing to the integration of global

production processes Nesterova et al. (2016a).

However, implementing MTN development

projects faces several significant challenges

Nesterova et al. (2016a); (European Commission

2024; European Commission 2013):

1. Interaction between Different Infrastructure

Objects: Coordination among various transport

infrastructures is often inadequate.

2. Technical Disparities: There are considerable

differences in technical standards, capacity, and

infrastructure quality.

3. Lack of Strategic Coherence: Infrastructure

projects are frequently carried out without an

overarching strategy.

4. Funding Deficits: Insufficient funds are available

to eliminate bottlenecks or build new infrastructure.

5. Diverse Ownership and Interests: The

development process is complicated by multiple

owners and operators with different objectives.

These challenges highlight the complexity and

importance of improving existing tools or developing

new solutions to address these issues effectively

Nesterova et al. (2016b). Reference models provide a

structured approach to analyzing and designing

complex systems like MTNs. They are standardized

models used across various fields to define general

structures and processes applicable to similar

problems (Scheer (1999); Grefen (2010); Wirtz

(2021). The key benefits of reference models include:

• Standardization and Comparability: They offer a

standardized framework that allows for the

comparison of different systems and facilitates the

transfer of best practices.

• Knowledge Management and Communication:

Reference models establish a common language

among stakeholders, such as developers, consultants,

and managers, enhancing communication.

130

Zöttl, D., Granig, A., Schwendinger, B. and Schlund, S.

Towards a Reference Model for Multimodal Transport Networks.

DOI: 10.5220/0013391700003953

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 14th International Conference on Smart Cities and Green ICT Systems (SMARTGREENS 2025), pages 130-138

ISBN: 978-989-758-751-1; ISSN: 2184-4968

• Increased Efficiency: Time and resources can be

saved by utilizing proven models, reducing the need

for repetitive analysis and design work.

• Flexibility and Adaptability: Reference models

can be tailored to the specific needs of an organization

while maintaining general principles.

• Quality Assurance: By relying on tested

methodologies, they help avoid common mistakes

and ensure consistent outcomes.

The primary objective of this work is to develop a

reference model for multimodal transport networks

that provides a holistic view of various transport

modes and their characteristics. This involves

defining, describing, and categorizing the necessary

attributes for modeling MTNs. This approach aims to

reduce complexity and establish a common language

for stakeholders involved in multimodal transport.

The paper is structured as follows:

Section 2, the Literature Review offers an overview

of academic research on reference models of

multimodal transportation networks, addressing key

issues such as digitalization, travel time estimation

and related projects.

Section 3, the Methodology, describes how expert

interviews were conducted to identify relevant

elements and the process of incorporating these

elements into a model.

In Section 4, the Results present the identified and

categorized elements of the developed reference

model, including its design and an potential

application of the reference model.

Lastly, Section 5 provides a conclusion and a

discussion on the next steps.

2 LITERATURE REVIEW

A reference model serves as a standardized

framework that describes the structure, processes and

relationships within multimodal transport systems. It

acts as a comprehensive guide for incorporating

various transportation modes (such as road, rail, sea,

and air) into a unified multimodal network. By

implementing a reference model, all parties involved

can ensure adherence to shared guidelines, thereby

averting potential misunderstandings stemming from

inconsistent models. Furthermore, this approach

facilitates the seamless exchange of data in a

universally accepted format. In the context of

transport networks, graphs are often employed to

depict the infrastructure, with nodes representing

terminals where mode transitions occur, and edges

serving as links between distinct nodes. Disruptions

may impact nodes, routes, or only partially affect

them during specific modes. Furthermore, these

disruptions possess a specific duration and can be

classified based on their severity.

Exploring the corridors of the trans-European

transport network and in particular the multimodal

transport network is crucial for improving transport

efficiency, connectivity and sustainability in Europe.

The identification of bottlenecks, the definition of

investment priorities and the optimization of cross-

border coordination are essential elements that

ultimately promote economic growth and regional

integration across the continent. In this context,

reference should be made to publicly funded research

projects and initiatives that are already addressing the

challenges of multimodal transport networks as well

as related work.

Harris et al. (2015) highlight the crucial role of

Information and Communication Technology (ICT)

as a fundamental aspect of logistics. The

transformation to sustainable transport systems can

be realized on the basis of forecasts for the transport

and logistics sector through the use of ICT. The

exploitation of this potential using ICT is contingent

upon the introduction of a "common platform without

national borders" and uniform standards Giusti et al.

(2019). The EU H2020 project SENATOR aims to

create a multi-collaborative framework and a ‘control

tower’ system in this area.

To facilitate the increased use of intermodal

transport, Altuntaş Vural et al. (2020) examine the

individual potential of various digital tools to

overcome obstacles. The results derived from this

indicate a tendency towards conservative and

resistant behaviour in the transport industry.

In this context, the SHIFT2RAIL project, funded

by the EU as part of the H2020 programme, aims to

develop innovative rail technologies and integrate

them into existing and future rail networks. The aim

is to improve the efficiency, sustainability,

performance and resilience of rail transport.

Several methodologies for the planning of

intermodal freight transport have already been

developed. A variety of approaches to planning

intermodal freight transport can be found in the

existing literature.

Demir et al. (2016) employ a stochastic approach

to describe an optimization problem. In this approach,

the objective is to sample the average travel times

from a set of scenarios, thereby allowing for partial

consideration of unexpected events. Abbasi et al.

(2024) investigate the impact of disruptions on

seaport terminals, employing a mixed-integer linear

programming model. In this intermodal model, the

consideration of unexpected events, specifically those

Towards a Reference Model for Multimodal Transport Networks

131

occurring at the transshipment port, results in a

reduction in the available capacity and performance

at the seaport.

At this point, it should be noted that in the

transport network, unexpected events have an impact

on the transport link in addition to the impact on the

transport node. In their work, Hrušovský et al. (2021)

also formulate a mixed-integer linear programming

model as an optimization model. Balster et al. (2020)

introduce machine learning approaches utilizing

random forest, gradient boosting, linear regression

trees and ordinal trees in the model. Spanninger et al.

(2022) also distinguishes between event-driven and

data-driven approaches. The MOTOS research

initiative is developing a simulation platform in this

area that models transport systems on the basis of

mobility data.

The majority of existing transport models for

MTNs are based on transport networks that use either

nationwide transport networks or a small amount of

data. Only a small number of studies, such as that by

Strelko et al. (2022), include complete corridors of

the Trans-European Transport Network (TEN-T)

Richardson (1997).

Each of the aforementioned modelling

approaches is characterised by a common absence:

none employs a standard reference model to describe

MTNs. The review of the literature indicates that

optimisation models in multimodal transportation

have been considered, however, none of the existing

models provide a detailed description of the data basis

that is employed, or the reference model that is used

as a basis for comparison. Furthermore, none address

the potential of transferring the model to other

transport corridors. The focus of this study is to

develop a data model that can be used as a reference

in any optimisation models used in the field of

multimodal transport planning. To this end, we have

developed an approach to a reference model that

brings together the different components of an MTN.

It is important to emphasise that the reference model

serves as a concept. The elements listed, such as the

modelling of predictability, serve as placeholders for

the individual definition of specific values.

3 METHODOLOGY

The following section presents the methodology of

expert interviews as a means of establishing the

foundation of the data. Furthermore, it explores the

process and visualization of the reference model.

Finally, it concludes with a description of the model

for MTNs.

3.1 Approach of Expert Interview

Ensuring correct and high-quality data collection

requires a strong willingness to communicate on the

part of those involved in the model. Furthermore, in

addition to the aspects already discussed, special

expertise is required to be able to adequately

recognize and classify more complex issues. The

individual experts were questioned using an open

interview technique. This approach allows complex

topics to be analyzed using expert knowledge. A

further advantage of this method is the identification

of new topics that were not apparent at the beginning

of the study. (Mayer 2013)

The expert groups consist of companies and

researchers in the field of transport planning. The

relevant stakeholders therefore include transport

management system providers and internationally

active logistics service providers.

The interviews conducted are based on four

overarching questions for each transport mode (ship,

rail and road):

• Which data is required for planning daily

transport operations?

• How would you classify the data into static

and dynamic categories in the context of

disruptive events?

• What classes would you use to categorize

the required data?

• Which data is necessary for an optimization

model in MTNs?

At the outset, the individual expert groups were

presented with the same overarching questions. The

sub-questions were developed in an open and

individualized approach with the objective of

meeting the specific requirements of the respective

expert groups.

3.2 Process Model

The procedure and the visualization of the

information resulting from the surveys conducted are

based on the CRISP-DM process model Wirth and

Hipp (2000). As this work involves the development

of a reference model in multimodal transport

planning, only the first five steps of the process model

are applied (Figure 1). These steps include the

‘development of a business understanding’, ‘data

understanding’, ‘data preparation’, ‘modelling’ and

‘evaluation’. To create the basis for the reference

model, the process model is iterated through.

The categorization of the relevant attributes is

based on the approach outlined in the previously

described CRISP-DM process model. Furthermore,

SMARTGREENS 2025 - 14th International Conference on Smart Cities and Green ICT Systems

132

input is derived from expert interviews, particularly

with logistics service providers who require the

necessary information for transport planning.

Figure 1: Process model for qualitative data collection

(based on Wirth and Hipp (2000)).

3.3 Model Description

The conversion of a significant amount of

information into a model is carried out with the aid of

database models. Schematic representations, such as

maps and processes, can describe a multimodal

transport network, but they do not conform to the

necessary data structure. Relational data models are

often utilized for complex data models. Their

straightforward structure, typically represented as flat

tables where rows correspond to data objects Kemper

and Eickler (2015), supports their application in

multimodal data models. Additionally, a relational

data model is usually developed using an entity-

relationship diagram (ER diagram).

ER modeling is a technique used in logical design

modeling, where the development of the logical data

model begins after capturing business requirements,

data requirements, and understanding of business

rules. The three basic components of an ER model

are:

• Entities

• Relationships

• Attributes.

Entities in the logistics context can include nodes,

trains, commodities, etc., for which the company

stores data. Relationships show how entities are

related to each other, representing the business rules

or constraints. For example, a hub is linked to

another hub via a transport route section. Attributes

are unique characteristics of entities that are

managed in the data model. For example, attributes

of a hub can be the location, its storage size or

working hours. These attributes are recorded by the

company to manage its business along the

multimodal transport network.

An entity can have key attributes (such as primary

and foreign keys) and non-key attributes. Key

attributes uniquely identify data records and link data

from different entities. Non-key attributes must not

uniquely identify data. Figure provides a symbolic

representation of the structure of an entity.

Figure 2: Symbolic representation of an entity (adapted

from Sherman 2015).

Cardinality refers to the number of instances of an

entity that can be associated with another entity in a

relationship. It specifies whether the association

involves one or many instances. There are four main

cardinality options, as described in Table 1:

“Mandatory”, “1 Optional”, “Many Mandatory”,

“Many Optional” Sherman (2015).

Table 1: Crow’s Foot Notation. cf.Gronwald (2024)).

The methodology of entity-relationship models

frequently serves as a foundational technique in the

conceptualization of data models. The streamlined

structure of these relational models represents a

pivotal factor in their deployment within the domain

of a multimodal data model Kemper and Eickler

(2015).

4 RESULTS

This section explains the categorisation of the

identified attributes resulting from the CRISP-DM

and expert interviews. Furthermore, the specific

model design is explained in detail and a use case is

Towards a Reference Model for Multimodal Transport Networks

133

presented, focussing on the identification of

multimodal transport routes in a TEN-T corridor.

4.1 Categorization of Attributes

To incorporate the attributes into the reference model,

the relevant attributes (listed in Figure 4) were

categorised based on the input from the expert

interviews. This categorisation comprises the four

meta-levels ‘transported goods’, ‘node’, ‘route

section’ and ‘disruptive events’ and is shown

graphically in Figure 3. Transported goods primarily

have attributes that do not change along the supply

chain. Nodes can function both as simple

intersections and as transshipment points. Transport

route sections link the nodes with each other and can

be assigned to a specific mode of transport.

Disruptive events can occur both at nodes and along

the transport route sections.

Figure 3: Categorisation of relevant attributes.

4.2 Reference Model

Based on the attributes required for a reference model

to describe a multimodal transportation network

according to the notation of Section 3.3, the ER

diagram shown in Figure 4 was created.

The main entities in turn correspond to the 4 meta

levels:

• Transported Goods (Primary key: Order_id)

• Nodes (Primary key: Node_id)

• Transport Route Section (Primary key: Transport

route section_id)

• Disruptive Events (Primary key: Disruptive

event_id)

The meta levels are shown as entities in the ER

diagram. Attributes that can be assigned from a

certain pool of values have also been combined into

additional entities. This has the advantage that these

attributes do not have to be documented individually

for each data set from other entities. These are:

• Locations – contains all attributes for the

localization of nodes

• Modes - contains the three different modes of

transport (road, rail, ship)

• Load carriers – contains all attributes that describe

the used or permitted load carrier

• Freight forwarders – contains a list of relevant

freight forwarders

These eight entities have the following relationships

with each other:

Transport route section – Nodes:

• departs_at (Many Mandatory to 1 Mandatory):

Transport route sections always require exactly one

start node. A node can be the start node for several

transport route sections.

• arrives_at (Many Mandatory to 1 Mandatory):

Transport route sections always require exactly one

end node. A node can be the end node for several

transport route sections.

Transport route section – Disruptive Events:

• affects (1 Optional to Many Optional):

Transport route sections can be affected by one or

more disruptive events.

Transport route section – Load carriers:

• allows (Many Mandatory to Many Mandatory):

Transport route sections always require at least one

load carrier type that can be transported along the

route. However, one can also allow several load

carrier types.

Transport route section – Freight forwarders:

• approached_by (Many Mandatory to Many

Mandatory):

Transport route sections always require at least one

freight forwarder to handle transports along the route.

However, several freight forwarders can also serve a

route.

Transport route section – Modes:

• has_mode (Many Mandatory to 1 Mandatory):

Transport route sections use exactly one mode of

transport type. A mode of transport category can be

relevant for several transport route sections.

Nodes – Transported Goods:

• pickup_at (1 Mandatory to Many Mandatory):

SMARTGREENS 2025 - 14th International Conference on Smart Cities and Green ICT Systems

134

Transported goods always require exactly one order

start node. A node can be an order start node for

several transported goods.

• delivery_at (1 Mandatory to Many Mandatory):

Transported goods always require exactly one order

destination node. A node can be an order destination

node for several transported goods.

Nodes – Disruptive events:

• affects (1 Optional to Many Optional):

Nodes can be affected by one or more disruptive

events.

Nodes – Modes:

• has_mode (Many Mandatory to Many Mandatory):

Nodes are connected to at least one mode of transport

type. However, a node can also be connected to

several modes of transport.

Nodes – Locations:

• is_located_at (1 Mandatory to 1 Mandatory):

Nodes are assigned to exactly one location.

Transported Goods – Load carriers:

• transported_with (1 Mandatory to 1 Mandatory):

Transported goods are transported in at least one load

carrier. A load carrier can be used for several

transported goods.

Figure 4: Reference model (ER-Diagram).

Towards a Reference Model for Multimodal Transport Networks

135

4.3 Practical Application: Ten-T Use

Case

This section demonstrates the evaluation of the

reference model through the application in a practical

real-world scenario. For this purpose, an application

was created whose underlying data layer was based

on the entities, attributes and cardinalities of the

developed reference model and fed with specific data

records. We specifically examine the problem of

identifying multi-modal transport routes, with a focus

on the Rhine-Danube TEN-T corridor. Our solution

enables the generation of multiple transport routes

using different modes for a single source-to-

destination relationship, allowing for the avoidance of

potential disruptions by manually selecting

alternative routes. To find a feasible transport route,

we must specify the relevant corridor, the starting

point (source) and the destination (sink) for the

transport. Additionally, we need to specify the types

of transportation modes to be utilized, as well as the

desired number of routes to be identified. Once the

route is generated, it is displayed on the map

alongside the input interface.

Figure 5: Multi-modal transport route.

By considering multiple modes of transportation, it is

possible to identify routes that combine different

modalities, as shown in Figure 6. Clicking on specific

transport nodes provides additional information, such

as the estimated time of arrival or departure (see

Figure 6). Moreover, details regarding the transport

node itself, such as opening hours or the available

handling equipment at the terminal, can also be

accessed.

Figure 6: Additional information at transport nodes.

As previously mentioned, our prototype also supports

the provision of multiple routes for a single

connection. This functionality is depicted in Figure 7.

Figure 7: Provision multiple multi-modal transport routes

for a single connection.

The use case shows the fulfilment of the key benefits

of reference models as follows Scheer (1999); Grefen

(2010); Wirtz (2021):

• Standardization and Comparability: The use case

shows that the reference model reflects the properties

of the transport route sections and nodes as well as

their relations in a standardised manner. This enables

comparability between different transports.

• Knowledge Management and Communication:

The developed reference model transforms logistical

requirements into a relational data base and

establishes the basic for solving planning tasks of

multimodal transports.

• Increased Efficiency: The use of the developed

reference model allows the implementation of further

use cases without the need for repeated design work

for the underlying database.

• Flexibility and Adaptability: In the current version

of our prototype, we primarily focused on finding

multi-modal transport routes in a TEN-T corridor.

SMARTGREENS 2025 - 14th International Conference on Smart Cities and Green ICT Systems

136

However, it could easily be used to describe

multimodal transport networks of other corridors.

• Quality Assurance: Based on the developed

reference model, the use case shows that various

transports are reproduced with consistent information

content and format.

5 CONCLUSION AND NEXT

STEPS

This work presents a reference model for multimodal

transportation networks and outlines the

methodology used to develop it. First key attributes

of MTNs were identified and compiled into a

comprehensive data requirements list, categorized

under four main meta-levels: "Transported Goods,"

"Nodes," "Transport Route Sections," and

"Disruptive Events." Sub-categories were also added

to improve the usability of the list for non-specialists.

Expert interviews provided further insights into

the relationships between these attributes and their

roles in processes, particularly those involving

spontaneous adaptations during disruptions. These

interviews helped determine the required level of

granularity for the model.

The chosen reference model structure allows for

the representation of all identified attributes and their

interconnections. Essential elements of MTNs, such

as transport routes, nodes, and goods and their

relationships were integrated into the model.

In addition, a use case was presented that is based

on a data structure in accordance with the developed

reference model. This proves that the developed

reference model provides comprehensive attributes

and relationships for transport planning within the

multimodal transport network. Furthermore, the

fulfilment of the criteria for a reference model was

evaluated with regard to standardisation and

comparability, knowledge management and

communication, increased efficiency, flexibility and

adaptability and quality assurance.

The acquisition of particular data sets posed the

biggest challenge. The absence of a cooperative

inclination among transportation network

stakeholders is a primary factor contributing to the

nonexistence of reference models for multimodal

transportation networks prior to the emergence of this

paper. Even within the framework of this work, a

substantial portion of the datasets had to undergo

anonymization or simulation in order to generate a

practical data model for the use case presented in this

paper. In conclusion, establishing a Europe-wide

database for MTNs depends on standardized data

collection from all network participants. The

reference model developed, provides a standardized

framework, enabling users to populate it with real

data and describe any transport corridor effectively.

As part of this study, further use cases for

evaluation will be conducted in which a routing

algorithm is used to solve transport planning

problems based on the developed reference model.

After the evaluation, the reference model will be

integrated as the basis for an optimisation model

using reinforcement learning.

ACKNOWLEDGEMENTS

Supported by Horizon Europe (HORIZON)

ReMuNet (GA#101104072)

REFERENCES

Abbasi, Mohammad/Consilvio, Alice/Giglio, Davide

(2024). A multi-objective optimization approach for

disruption management in an intermodal freight

transport network 78.

Altuntaş Vural, Ceren/Roso, Violeta/Halldórsson,

Árni/Ståhle, Gabriella/Yaruta, Marina (2020). Can

digitalization mitigate barriers to intermodal transport?

An exploratory study 37.

Balster, Andreas/Hansen, Ole/Friedrich, Hanno/Ludwig,

André (2020). An ETA Prediction Model for

Intermodal Transport Networks Based on Machine

Learning 62.

Demir, E./Burgholzer, W./Hrušovský, M./Arıkan,

E./Jammernegg, W./Woensel, T. V. (2016). A green

intermodal service network design problem with travel

time uncertainty 93.

European Commission (2013). New EU transport

infrastructure policy – background.

European Commission (2024). Infrastructure – TEN-T –

Connecting Europe.

Giusti, Riccardo/Manerba, Daniele/Bruno, Giorgio/Tadei,

Roberto (2019). Synchromodal logistics: An overview

of critical success factors, enabling technologies, and

open research issues 129.

Grefen, Paul (2010). Mastering e-business.

Gronwald, Klaus-Dieter (2024). Data Management.

Harris, Irina/Wang, Yingli/Wang, Haiyang (2015). ICT in

multimodal transport and technological trends:

Unleashing potential for the future 159.

Hrušovský, Martin/Demir, Emrah/Jammernegg,

Werner/van Woensel, Tom (2021). Real-time

disruption management approach for intermodal freight

transportation 280.

Kemper, Alfons/Eickler, André (2015). Datenbanksysteme.

Eine Einführung. 10th ed.

Towards a Reference Model for Multimodal Transport Networks

137

Mayer, Horst O. (2013). Interview und schriftliche

Befragung. Grundlagen und Methoden empirischer

Sozialforschung. 6th ed.

Nesterova, Natalia/Anisimov, Vladimir/Anisimov,

Aleksandr (2016a). Balanced scorecard for evaluation

of multimodal transportation network development

strategies.

Nesterova, Natalia/Goncharuk, Sergey/Anisimov,

Vladimir/Anisimov, Aleksandr/Shvartcfel, Viacheslav

(2016b). Set-theoretic Model of Strategies of

Development for Objects of Multimodal Transport

Network 165.

Richardson, Tim (1997). The trans-European transport

network: environmental policy integration in the

European Union 4 (4), 333–346.

Scheer, August-Wilhelm (1999). ARIS - business process

frameworks. 3rd ed.

Sherman, Rick (2015). Foundational Data Modeling. In:

Business Intelligence Guidebook, 173–195.

Spanninger, Thomas/Trivella, Alessio/Büchel,

Beda/Corman, Francesco (2022). A review of train

delay prediction approaches 22.

Strelko, Oleh/Yurchenko, Oksana/Vasilova, Halyna/Gusar,

Ievgen/Rudiuk, Maryna (2022). Theoretical

Prerequisites for the Development of Multimodal

Transport-Logistic Technologies in Cooperation With

Ten-T Network.

Wirth, Rüdiger/Hipp, Jochen (2000). CRISP-DM: Towards

a standard process model for data mining.

Wirtz, Bernd W. (2021). Business Model Management.

Design, Instrumente, Erfolgsfaktoren von

Geschäftsmodellen. 5th ed.

SMARTGREENS 2025 - 14th International Conference on Smart Cities and Green ICT Systems

138