Guidelines and recommendations for future policy of cooperative and automated freight transport

Abstract

This report summarises the main findings on the expected impacts after introducing CCAMs and cooperative, connected and automated freight transport. The strengths and limitations of the theoretical and empirical work underlying these impacts are discussed, and policy considerations in the broader context of the transition to smart urban mobility are presented.

Full text

This project has received funding from the European Union’s Horizon 2020
research and innovation programme under grant agreement No 824361.
Guidelines and
recommendations for future
policy of cooperative and
automated freight transport
Deliverable D7.5 – WP7 – PU

LEVITATE | Deliverable D7.5 | WP7 | Final
i
Guidelines and recommendations
for future policy of cooperative
and automated freight transport
Work package 7, Deliverable D7.5
Please refer to this report as follows:
Goldenbeld, C., Gebhard, S., Schermers, G.,
Mons, C and Hu, B. (2021). G
uidelines and recommendations for future policy of
cooperative and automated freight transport, Deliverable D7.5
of the H2020 project
LEVITATE.
Project details:
Project start date:
Duration:
Project name:
01/12/2018
42 months
LEVITATE – Societal Level Impacts of Connected and Automated Vehicles
Coordinator:
Andrew Morris, Prof. of Human Factors in Transport Safety
Loughborough University
Ashby Road, LE11 3TU Loughborough, United Kingdom
This project has received funding from the European Union’s Horizon 2020
research and innovation programme under grant agreement No 824361.
Deliverable details:
Version:
Dissemination level:
Due date:
Submission date:
Final
Pu (Public)
30-11-2021
06-12-2021
Lead contractor for this deliverable:
SWOV
Report Author(s):
Goldenbeld, C., Gebhard, S., Schermers, G.,
Mons, C (SWOV, Netherlands) and Hu, B (AIT, Austria).

LEVITATE | Deliverable D7.5 | WP7 | Final
ii
Revision history
Date
Version
Reviewer
Description
26-11-2021
Preliminary draft 1
Wendy Weijermars
Accept with reservation
27-11-2021
Preliminary draft 1
Rune Elvik
Accepted
03-12-2021
Final draft
Govert Schermers
Bin Hu
Vanessa Millar
06-12-2021
Final deliverable
Andrew Morris–
Loughborough University  EC
Legal Disclaimer
All information in this document is provided "as is" and no guarantee or warranty is given
that the information is fit for any particular purpose. The user, therefore, uses the
information at its sole risk and liability. For the avoidance of all doubts, the European
Commission and CINEA has no liability in respect of this document, which is merely
representing the authors' view.
© 2021 by LEVITATE Consortium

LEVITATE | Deliverable D7.5 | WP7 | Final
iii
Table of contents
List of abbreviations .......................................................................................... iv
About LEVITATE ................................................................................................. v
Executive summary ............................................................................................ 1
1 Introduction ................................................................................................. 5
1.1 General Levitate approach ................................................................. 5
1.2 Work Package 7 ................................................................................. 7
1.3 Purpose and structure of report ......................................................... 8
2 Background ................................................................................................ 10
2.1 Urban mobility and transport goals .................................................. 10
2.2 Expected automation impacts ......................................................... 12
2.3 Expected impacts of cooperative and automated freight transport .. 14
2.4 Sub-use cases .................................................................................. 16
2.5 Assessment methods ....................................................................... 20
2.6 Approach to synthesizing results ..................................................... 23
3 Main findings: quantified impacts ............................................................... 26
3.1 Impacts on the environment ............................................................ 26
3.2 Impacts on mobility ......................................................................... 28
3.3 Impacts on society, safety & economy ............................................. 30
3.4 Additional impacts economy: annual fleet costs ............................... 33
3.5 Impacts of truck platooning ............................................................. 36
4 Discussion ................................................................................................. 38
4.1 Main findings.................................................................................... 38
4.2 Strengths and Limitations ................................................................ 42
4.3 Policy considerations and discussion ............................................... 44
4.4 Future Challenges for urban freight transport .................................. 50
5 Conclusions and recommendations ............................................................ 51
5.1 Conclusions ...................................................................................... 51
5.2 Policy recommendations .................................................................. 53
References .............................................................................................. 55
Appendix A Full results .................................................................................. 1
Appendix B Cost assumptions vehicle operating costs ................................... 1
Appendix C Bridge models & technical overview ........................................... 2

LEVITATE | Deliverable D7.5 | WP7 | Final
iv
List of abbreviations
AUSS
Automated Urban Shuttle Service
ADAS
Advanced Driver Assistance Systems
AEB
Autonomous Emergency Braking
AV
Automated Vehicle
CACC
Cooperative Adaptive Cruise Control
CCAM
Cooperative, connected and automated mobility
CATS
Connected and Automated Transport Systems
CCAM
Cooperative, Connected and Automated Mobility
C-ITS
Cooperative Intelligent Transport Systems
CV
Connected Vehicle
DisA
Distraction Alert
DrowA
Drowsiness Alert
ERTRAC
European Road Transport Research Advisory
Council
EU
European Union
FCW
Forward Collision Warning
FHWA
Federal Highway Administration
FORS
Fleet Operation Recognition Scheme
GDPR
General Data Protection Regulation
GLOSA
Green light optimal speed advisory
ISA
Intelligent Speed Assist
IVS
In-vehicle Signage
LCA
Lane Change Assist
LDW
Lane Departure Warning
LKA
MPR
Lane Keeping Assist
Market Penetration Rate
NHTSA
National Highway Traffic Safety Administration
NRC
National Research Council
PST
Policy Support Tool
SAE
Society of Automotive Engineers
SRG
SUC
Stakeholder Reference Group
Sub-Use Case
SSAM
Surrogate Safety Assessment Model
TTC
Time to Collision
V2I
Vehicle to Infrastructure
V2V
Vehicle to Vehicle
V2X
Vehicle to everything
VKT
Vehicle Kilometres Travelled

LEVITATE | Deliverable D7.5 | WP7 | Final
v
About LEVITATE
Societal Level Impacts of Connected and Automated Vehicles (LEVITATE) is a European
Commission supported Horizon 2020 project with the objective to prepare a new impact
assessment framework to enable policymakers to manage the introduction of connected
and automated transport systems, to maximise the benefits and to utilise these
technologies to achieve societal objectives.
Connected and automated transport systems (CATS), or recently the more accepted term
Cooperative, Connected and Automated Mobility (CCAM), are expected to be introduced in
increasing numbers over the next decade. Automated vehicles have attracted the public
imagination and there are high expectations in terms of traffic safety, mobility,
environment, and economic growth. With such systems not yet in widespread use, there
is a lack of data and knowledge about impacts.
The potentially disruptive nature of highly automated vehicles makes it very difficult to
determine future impacts from historic patterns. Estimates of future impacts of automated
and connected mobility systems may be based on forecasting approaches, yet there is no
agreement over the methodologies nor the baselines to be used. The need to measure the
impact of existing systems as well as forecasting the impact of future systems represents
a major challenge. The dimensions for assessment are themselves quite broad ranging
from impacts on traffic safety to the environment and potentially including sub-divisions
within the domains which adds to the complexity of future mobility forecasts.
Specifically LEVITATE has four key objectives:
• To establish a multi-disciplinary methodology to assess the short, medium and
long-term impacts of CCAM on mobility, safety, environment, society and other
impact areas. Several quantitative indicators will be identified for each impact type.
• To develop a range of forecasting and back casting scenarios and baseline
conditions relating to the deployment of one or more mobility technologies that will
be used as the basis of impact assessments and forecasts. These will cover three
primary use cases – automated urban shuttle, passenger cars and freight services.
• To apply the methods and forecast the impact of CCAM over the short, medium
and long term for a range of use cases, operational design domains and
environments and an extensive range of mobility, environmental, safety,
economic and societal indicators. A series of case studies will be conducted to
validate the methodologies and to demonstrate the system.
• To incorporate the methods within a new web-based policy support tool to
enable city and other authorities to forecast impacts of CCAM on urban areas. The
methods developed within LEVITATE will be available within a toolbox allowing the
impact of measures to be assessed individually. A Decision Support System will
enable users to apply back casting methods to identify the sequences of CCAM
measures that will result in their desired policy objectives.

LEVITATE | Deliverable D7.5 | WP7 | Final
1
Executive summary
Goals and impacts
Mobility of people and goods is the lifeline of the modern city. In planning for future urban
mobility cities like Manchester and Vienna have set goals in which future mobility should
contribute to a cleaner city environment, to easier, more comfortable, more cost-effective
travel within the city, and to a better, more inclusive society with equal travel opportunities
for all social groups. ‘Smart mobility’ - where various types of vehicles in the city, such as
passenger cars, urban transport vehicles, freight vehicles, are connected to information
systems that help them to navigate more efficiently and safely through city traffic – is seen
as one of the prime movers of the transition towards smart cities. Within LEVITATE,
important goals for future mobility have been identified for the environment, mobility, and
for society & economy. A literature study has identified the direct, systemic and wider
impacts that smart mobility may have on the city traffic network, and how these impacts
are mutually connected.
In LEVITATE, several methods—including a literature study, microsimulation, meso-
simulation, Delphi survey—have been used to study the expected impacts of the increasing
presence of first- and second-generation automated vehicles in city traffic on the domains
of environment, mobility, and society and economy. Levitate has also estimated the
additional impacts of specific policy interventions (termed ‘sub-use cases’) such as
automated urban shuttle services, or hub-to- hub freight transport, on these domains.
These estimated effects are presented as effects over and above the effect resulting from
the increasing presence of automated vehicles anticipated as part Cooperative, connected
and automated mobility (CCAM).
Given the many uncertainties in prediction, it is obvious that any predicted values are
associated with large uncertainty. For the WP7 results, it was decided not to estimate
confidence intervals based on the standard error derived from repeated trail runs of models
since these intervals would be broad and non-informative. Also, the estimation of these
intervals would tend to be biased in itself since the input variables and assumptions in the
models are very likely much stronger determinants of predicted values than the variability
in sample runs.
Approach to summarizing LEVITATE results
The goal of this Deliverable is to summarize the more detailed results presented in D7.2-
D7.4 and to provide an overview of the main expected trends for each selected impact. To
quantify the impacts expected from an increasing penetration rate of connected and
automated vehicles in the total vehicle fleet as well as the implementation of cooperative
and automated freight transport, three primary methods were used: microsimulation,
Delphi, and operations research. A number of SUCs related to particular developments in
the freight transport sector were defined and these methods were applied to derive
estimates of the impacts that these SUCs would have at different penetration rates of
CAVs. To summarize these results, for each sub-use case an average (where applicable)
is taken of its scenarios to derive an average percentage change for the respective sub-
use case (see Table 3.1).

LEVITATE | Deliverable D7.5 | WP7 | Final
2
The impacts are presented as a percentage change from the Baseline scenario at 0%
penetration of CAVs, where neither automated freight nor automated vehicles have been
implemented in the network. These percentage changes are reported for increasing market
penetration rates of automated vehicles throughout the entire vehicle fleet in the network,
as used throughout LEVITATE.
The Baseline scenario refers to a “no intervention” scenario which is essentially the
expected autonomous development of CAVs from human dependence to human
independence. In the Baseline scenarios there is no cooperative and automated freight
transport added to the network. The impacts of CAVs on network performance can be
established by comparing the Baseline 100-0-0 scenario (100% human-driven/reliant
vehicles) to the Baseline 0-0-100 scenario (0% human-driven/reliant vehicles). The
specific effect or impact of the cooperative and automated freight transport scenario can
be determined by comparing the baseline situation for any given penetration rate with the
specific SUC results; the difference between the baseline and the SUC is the added effect
created by implementing the specific SUC intervention in the simulated network.
Main conclusions
Overall effects of CAVs
Estimating the baseline impacts of an increasing share of connected and automated
vehicles (CAVs) for Work Package 7 revealed the following main findings. The results are
based on simulations run on the network of Vienna and for all vehicles in the network
(including both freight vehicles & private cars).
• The increasing presence of connected and automated vehicles in the urban city area is
estimated to have positive impacts on the city environment (less emissions, higher
energy efficiency), and city society and economy (less parking space, lower freight
vehicle operating cost) and on city mobility (less congestion).
• In Work Package 7, the increasing presence of automated vehicles in the city is
estimated to have a temporary negative impact on road safety when penetration
rates of automated vehicles are low. The negative impact found is primarily due to
interactions between human-driven vehicles and automated vehicles, which are
expected to have different driving styles (e.g. AVs adopting different headways) and
different capabilities (e.g. human drivers’ longer reaction times) which may lead to an
initial increase in risks when many human drivers are still on the road. This result differs
from the baseline results found in the road safety impact study (Weijermars et al., 2021)
and discussed in WP5 and WP6, primarily due to two factors: 1) differences in the
network (Vienna) and 2) the inclusion of freight vehicles. Because less data was
available on the driving behaviour of autonomous freight vehicles, some parameters
assumed the values of 1st generation CAVs and others were based on assumptions. This
led to higher crash rate estimations when freight vehicles were included.
Larger positive impacts on road safety are estimated once human-driven vehicles are
replaced and second-generation automated vehicles make up at least 60% of the city’s
vehicle fleet. More broadly within LEVITATE, most estimates point to a large reduction
in crashes with the introduction of automated vehicles including a small reduction at low
penetration rates. At low penetration rates, the balance between the safety of

LEVITATE | Deliverable D7.5 | WP7 | Final
3
automated vehicles (which are expected to crash less often than human-driven vehicles)
and the potential risks of mixed traffic (when human-driven/less advanced automated
vehicles are still on the road) is a point of attention for further research.
• The increasing presence of automated vehicles in the city is estimated to have a slightly
negative impact on public health when traditional (human-driven) vehicles make up
the majority of vehicles, followed by a slightly positive impact at full automation of
the vehicle fleet.
Effects of SUCs: automated delivery, consolidation and hub-to-hub transport
Estimating the impacts of an increasing share of CAVs in the total vehicle fleet together
with one of the three forms of automated freight transport revealed the following main
findings:
• The automated delivery sub-use case is associated with additional benefits for energy
efficiency, CO2 emissions, congestion, public health and vehicle operating costs. The
night-time-only automated delivery scenarios (see Appendix A) show additional benefits
particularly for the two mobility indicators (travel time and congestion), due to less
interaction with the larger daytime traffic volumes.
• The automated consolidation sub-use case is associated with additional benefits for
energy efficiency, CO2 emissions, congestion, travel time, public health and vehicle
operating costs. Compared to automated delivery without consolidation at city hubs (the
first sub-use case), further improvements in energy efficiency, operating costs, and a
large reduction in total kilometres travelled are expected. This suggests that centrally
located city-hubs can help realise a more efficient allocation of resources.
• The hub-to-hub sub-use case is expected to deliver additional benefits for energy
efficiency, CO2 emissions, congestion, travel time, public health, and freight vehicle
operating costs.
• All three automated freight SUCs are predicted to marginally improve road safety
compared to the baseline, particularly at lower penetration rates when less of the
remaining vehicle fleet is automated.
• At the higher-level CAV penetration rates (above 80%), all the automated freight
delivery SUCs require more parking space than the baseline without automated delivery.
The Hub-to-Hub SUC even requires more parking space at 100% CAV penetration
compared to the current situation (with 100% human-driven vehicles).
• The sub-use cases of automated delivery, hub-to-hub and especially automated
consolidation are predicted positively impact public health. This positive expectation is
likely based on the expected additional benefits of these sub-use cases for both road
safety and emissions.
• Using data on freight delivery trips in Vianna, it was estimated that compared to manual
freight delivery, completely automated delivery and automated delivery with city-hubs
will have substantially reduced annual fleet costs (-68%).
Effects of truck platooning on bridges
• The largest effect of truck platooning on simple single span (beam) bridges as
modelled in LEVITATE is observed for the criteria of braking forces. For bridges above

LEVITATE | Deliverable D7.5 | WP7 | Final
4
80m length, it has been estimated that the braking force is at least double of the
baseline scenario.
• According to standard bridge models and standard traffic simulations within LEVITATE,
the need for strengthening structural resistance of bridges arises for many existing
bridge types and brings with it substantial costs
• For bridge strengthening, a model and guidelines for estimating the costs in relation to
the initial construction costs have been developed (D7.3).
• As an alternative to strengthening bridges, intelligent access control can be used to
arrange the increase of inter-vehicle distances for the bridge section to meet the code
level and prevent. Headway have been recommended and these are presented in
LEVITATE D7.3 (Hu et al., 2021b). Forcing an increased inter-vehicle distance by
intelligent access control will not diminish the ecological and economic benefits of truck
platoons.
Recommendations freight transport
For freight transport several recommendations can be given (Hu et al., 2019):
• Passenger transport and freight transport should seek collaboration (e.g., via automated
multi-purpose vehicles)
• Collaborative transportation, supported by city hubs and consolidation centres, are
necessary to improve operational efficiency. CCAM, especially automated hub-to-hub
transport and automated freight consolidation, will contribute significantly
• Multimodality and synchro modality are important factors to aim towards a sustainable
logistic supply chain.
• All the above points require homogenous and shared data among operators, which is
perhaps the most difficult challenge due to the competition between service providers
and freight operators.
Strengths and limitations of Levitate
The followings observations pertain to strengths and limitations of research within WP7
LEVITATE. A potential strength of the LEVITATE project is that both smart city transport
policy interventions and the associated impacts have been selected by a diverse group of
stakeholders. A wide variety of impacts were studied at the same time and the project
tried to capture interdependencies. The best available methods - microsimulation,
mesosimulation, Delphi, and operations research - were used to study and quantify the
expected impacts of mobility interventions intended to support CAV deployment and
sustainable city goals. Within Levitate project these impacts provide essential input for
developing a practical Policy Support Tool for city policy makers.
Concerning limitations, it should be pointed out there are general scientific difficulties in
predicting impacts of connected and automated mobility due to uncertainties about
propulsion energy, future capacity of power grids, employment, development of costs, and
about the behaviour and acceptance with regard automated vehicles. The results of the
models in LEVITATE are dependent upon specific assumptions. The simulation models used
examined only two CAV profiles (first generation vs. second generation ); future work may
extend the number of profiles. The safety results of the microsimulation did not include
crashes where vulnerable road users are involved.

LEVITATE | Deliverable D7.5 | WP7 | Final
5
1 Introduction
Vehicle automation technology is expected to impact many areas of society.
Highly automated vehicle technologies, complying to SAE levels 4 and higher, are
expected to stimulate new innovations and policy interventions across the
transport sector. These could include, for example, new vehicle types, new
transport services and changes to infrastructure. The LEVITATE project is
directed at studying—and where possible, quantifying—the expected impacts of
vehicle automation on society and in particular on mobility, safety, the
environment, and the economy. This report provides a synthesis of the results
achieved in Work Package 7 which studied the impacts of a number of sub-use
cases within the broader domain of cooperative and automated freight transport.
This specific chapter introduces the general scientific approach and methodology
adopted by LEVITATE. Furthermore, it describes the aims of Work Package 7 and
provides an overview of the structure of the report.
1.1 General Levitate approach
Within LEVITATE, a range of cooperative, connected and automated mobility (CCAM)
applications and interventions are studied under three use cases: automated urban
transport, automated passenger cars and cooperative and automated freight
transport. These correspond to Work Packages 5, 6 and 7 respectively.
In each WP, a stakeholder reference group workshop was organised among city
administrators, industry representatives and transport specialists to gather views on the
future and impacts of CCAM on these three primary use cases. Part of the workshop aimed
at identifying specific developments, applications or policy interventions within each sector
(or use case). These were termed sub-use cases. Within LEVITATE, these lists were
subsequently prioritized and refined subsequent project tasks in order to inform the
interventions and scenarios related to urban transport, passenger cars or freight transport.
The prioritisation of the sub-use cases mainly took three input directions into account: the
scientific literature, roadmaps detailing the deployment of CCAM and the workshop among
stakeholders. This resulted in the 13 sub-use cases listed in Table 1.1.
Table 1.1: Sub-use cases (SUCs) investigated in LEVITATE.
Urban transport (WP5)
Passenger vehicles (WP6) Freight transport (WP7)
Point to point automated urban shuttle
service connecting two modes of
transport
Provision of dedicated lanes
for AVs Automated urban delivery
Point to point automated urban shuttle
service in a large-scale network
Replace on street parking with
other facilities
Automated consolidation
On-demand automated urban shuttle
Road use pricing
Hub-to-hub automated
transport
Last mile automated urban shuttle
Parking price regulation
Truck platooning
Green light optimal speed
advisory (GLOSA)
Automated ride sharing

LEVITATE | Deliverable D7.5 | WP7 | Final
6
Within LEVITATE, the impacts of the cooperative, connected and automated mobility
(CCAM) sub-use cases are evaluated at three impact levels: direct, systemic and wider.
Direct impacts are changes that are noticed by each road user on each trip (Elvik et al.,
2020). These impacts are relatively short-term in nature and can be measured directly
after the introduction of an intervention or technology, such as changes in travel time or
costs. Systemic impacts are system-wide impacts within the transport system which are
typically secondary effects resulting from direct impacts. These include measures such as
congestion or modal split. Wider impacts are those aspects on which transport systems
rely to make mobility possible and also those which are in essence a by-product of mobility.
Examples of wider impacts are changes in land use and employment, energy demand and
public health. These are inferred impacts measured at a larger scale and are the result of
direct and system wide impacts. They are considered long-term impacts (Elvik et al.,
2020). Table 1.2 presents the impacts considered within Levitate, their impact level and
the policy area(s) to which they are most related.
Table 1.2: Overview of (estimated) impacts in relationship to policy, scale, term and method (WP7).
Quantified impacts
see D7.2-7.4 (Hu et al., 2021)
Impact level
see D3.1 (Elvik et al., 2019)
Relevant policy areas
Travel time
Direct
Mobility
Vehicle operating cost
Society, economy
Freight transport cost
Society, economy
Congestion
Systemic
Mobility, Economy, society
Truck platooning
Mobility
Road safety
Wider
Safety
Parking space
Mobility, economy
Energy efficiency
Environment, economy
Emissions
Environment
Public health
Society
In Section 2.4 we further describe how the impacts in Table 1.2 have been operationalised
and studied in various methods.
Scenarios: baseline-only and policy intervention-scenarios
LEVITATE considers the impacts of two simultaneous developments: an expected growth
in the popularity of connected and automated vehicles (CAVs) over time, as well as the
policy intervention scenarios defined in the sub-use cases. These are defined in terms of
scenarios, for which the impacts in Table 1.2 are estimated:
• Baseline scenario: growing penetration of connected and automated vehicles
(CAVs) within the entire vehicle fleet in the network WITHOUT a policy intervention
• Sub-use case scenarios: growing penetration of connected and automated
vehicles (CAVs) within the entire vehicle fleet in the network WITH a policy
intervention implemented in the network (see Table 1.1)
For all scenarios it is assumed that the percentage of CAVs in the vehicle fleet will increase
over time and that CAVs will be SAE level 5. As the exact time scale for the development
and adoption of highly automated vehicles (SAE levels 4&5) is still undefined, this growth
is quantified in so-called “deployment scenarios” at varying market penetration rates of
CAVs (see Table 1.3). These penetration rates reflect the transition from a driver-
dependant vehicle fleet (100% human-driven vehicles) to a driverless vehicle fleet (0%
human-driven vehicles).

LEVITATE | Deliverable D7.5 | WP7 | Final
7
In addition, two types of CAVs are distinguished in the deployment scenarios to represent
an expected evolution in technology (Table 1.3). Within LEVITATE, first first-generation
automated vehicles have been defined as vehicles with limited sensing and cognitive
ability. When compared to human driven vehicles these 1st generation CAVS are assumed
to have longer headways (following gaps), earlier anticipation of lane changes and
reaction times (more time required in give way situations). Second generation automated
vehicles have been defined as having advanced sensing and cognitive ability utilising
data fusion usage allowing greater confidence in taking decisions, shorter headways
(small following gaps), earlier anticipation of lane changes than human driven vehicles
and less time in give way situations (Roussou et al., 2021b).
In WP7 an important difference to the other WPs is that all automated freight vehicles
have been modelled as first generation CAV only. These differences in driving style are
implemented within the microsimulation models used in the impact quantification.
Table 1.3: CAV Baseline deployment scenarios used within LEVITATE
Vehicle type Deployment scenarios
A
B
C
D
E
F
G
H
Human-Driven Vehicle
100%
80%
60%
40%
20%
0%
0%
0%
1st generation CAV
0%
20%
40%
40%
40%
40%
20%
0%
2nd generation CAV
0%
0%
0%
20%
40%
60%
80%
100%
Human-driven freight vehicle
100%
80%
40%
0%
0%
0%
0%
0%
Freight CAV
0%
20%
60%
100%
100%
100%
100%
100%
1.2 Work Package 7
WP7 focuses on the impacts that the deployment of cooperative, connected and
autonomous vehicles may have on freight transport operations. Three major cooperative
and automated freight transport related sub-use cases were formulated:
1. Automated urban delivery: Future parcel delivery by automated vans and
delivery robots.
2. Automated consolidation: Extension of automated urban delivery by applying
consolidation at city-hubs.
3. Hub-to-hub automated transport: Effects of transfer hubs to facilitate
automated trucks.
A fourth sub-use case concerns the impact of platooning on bridges. Since this SUC
is rather unique and does not quite fit the methodology adopted to synthesize the results
of the research on automated freight in Levitate, the results are separately presented and
briefly discussed.
The expected impacts of these three cooperative and automated freight transport sub-use
cases on the environment, economy, mobility, safety and society are described in detail in

LEVITATE | Deliverable D7.5 | WP7 | Final
8
four deliverables (D7.1-7.4). In preparation for the quantitative analysis, the expected
impacts were first evaluated with a literature review and stakeholder workshop and
together with the expected future developments related to freight transport, impacts of
current ADAS and definition of SUCs, described in Deliverable 7.1. Subsequently, the
projected impacts of CAVs and, more specifically, the automated urban transport SUCs
were estimated in a series of quantitative analyses and reported in Deliverables D7.2
(direct impacts), D7.3 (systemic impacts), and D7.4 (wider impacts). The purpose of this
report, Deliverable D7.5, is to summarise the main impacts of the studied sub-use cases
and to provide more general recommendations for policymakers. Based on these results
described in D7.1-7.4 and on literature on the transition to smart mobility in smart cities
and other general guidelines, recommendations are developed to potentially inform future
policy on CAVs and automated urban transport.
Table 1.4: Methods used to evaluate and quantify the expected impacts of automation within the urban transport
sector
Goal Method Explanation Deliverable
Exploration
Literature
review
Existing literature on CCAM/CAVs/ADAS
7.1
Stakeholder
workshop
A group of key stakeholders – international/
twinning partners, international organisations,
road user groups, actors from industry,
insurances and health sector support the
project and participated in workshops
7.1
Quantifi-
cation
Delphi study
The Delphi method was used to determine
those impacts that cannot be defined by the
other quantitative methods
7.2, 7.4
Traffic micro-
simulation
AIMSUM microsimulation of traffic at the city-
district level (based on modelling individual
vehicles)
7.3, 7.4
Operations
research
Operations research was used to calculate the
fastest trip from a given depot to a number of
customers and to upscale microsimulation
results to the city-level
7.2, 7.4
Bridge
modelling
Bridge modelling was used to estimated
effects of truck platooning on bridge wear
7.3
Synthesis &
discussion
Synthesis
Major impacts summarized for the policy areas
Environment, Mobility and Society/ Economy/
Safety
7.5
Policy
considerations
Recommendations & considerations for
policymakers based on the wider literature
7.5
1.3 Purpose and structure of report
The purpose of this synthesis report is to present the expected impacts of a range of
mobility policies in the freight transport domain against the background of increasing CAV

LEVITATE | Deliverable D7.5 | WP7 | Final
9
deployment in the urban vehicle fleet on the environment, mobility, society, safety and
economy.
This report is structured as follows; following this general introduction to the Levitate
project, Chapter 2 provides a more detailed theoretical and empirical background to the
expected impacts of cooperative and automated freight transport, and it describes which
approach was used to summarise the various impact results from earlier Levitate
Deliverables D7.2, D7.3 and D7.4. Chapter 3 presents the main summarised findings of
the quantitative analyses which were reported in deliverables D7.2 to D7.4. In Chapter 4,
strengths and limitations of the Levitate approach are discussed and broader policy
considerations regarding the potential impacts of CCAM further discussed. In Chapter 5,
final conclusions are drawn, and some limitations of the present approach are discussed.

LEVITATE | Deliverable D7.5 | WP7 | Final
10
2 Background
The transition towards cooperative, connected and automated mobility (CCAM)
is expected to contribute to the goals of smart and sustainable cities. In Levitate,
the impacts of CCAM – including those of cooperative and automated freight
transport - on these city goals have been studied by various methods and for
different sub-use cases. This Chapter describes the major policy goals towards
which cooperative and automated freight transport may contribute (Section 2.1)
and how the various distinct impacts on transport system are interrelated and
related to the policy goals (Section 2.2). In Section 2.3, the expected impacts of
cooperative and automated freight transport are described. The sub-use cases of
freight transport which have been studied are further described in Section 2.4.
The methods used are further explained in Section 2.5. The approach taken in
this synthesis to summarise the impact results is explained in section 2.6.
2.1 Urban mobility and transport goals
To date, there is no standard European approach for defining goals and indicators for the
further development of smart cities. Within the Levitate project (WP4), two existing city
transport strategies from Greater Manchester in the UK, and Vienna in Austria have been
looked at in more detail, specifically in terms of high-level goals on transport developments
(Papazikou et al., 2020; D4.4). WP4 covers the effects of autonomous vehicle share on the
goals set out by policymakers of these cities (Papazikou et al., 2020).
The Greater Manchester Transport Strategy 2040 follows the vision “World class
connections that support long-term, sustainable economic growth, and access to
opportunity for all”. The strategy has seven core principles to be applied across their
transport network (City of Manchester, 2017):
1. Integrated – allow individuals to move easily between modes and services
2. Inclusive – provide accessible and affordable transport
3. Healthy – promote walking and cycling for local trips
4. Environmentally responsible – deliver lower emissions, better quality vehicles
5. Reliable – confidence in arrival, departure and journey times
6. Safe and secure – reduce road accidents especially injuries and deaths
7. Well maintained and resilient – able to withstand unexpected events and weather
conditions
Table 2.1 summarizes the Greater Manchester Transport Strategy 2040 goals and a
method to measure the impacts. For example, under the policy field, the goal is to improve
road safety, this will be measured by the number of injury or fatalities, as well as the
perception of personal security by transport mode.

LEVITATE | Deliverable D7.5 | WP7 | Final
11
Table 2.1: Overview of goals of the City of Manchester for a viable transport system of the future and
corresponding impact targets (City of Manchester, 2017).
Policy field
Policy goal
Measured impact
Environment
Reduced greenhouse gas emissions
CO2 and NO2 emissions
Best use of existing infrastructure in order to
reduce environmental impacts
Percentage of new homes having > level 4
accessibility to the public transport network
Mobility
More reliable journey times
departure/arrival time reliability by mode of
transport
Reduced congestion
Journey duration by mode
Increase use of sustainable transport (reduce
negative impact car use)
Modal split of sustainable transport
Share of non-sustainable transport modes
Safety
Improved safety and personal security
Number of killed and seriously injured
Perception of personal security by transport
mode
Society
Greater health
Number of walking and cycling trips
Better access to services
Sustainable transport catchment population
for key locations – town centres/hospitals
The second relevant transport strategy for Levitate WP7 is the Viennese Urban Mobility
Plan, under the “STEP 2025 Urban Development Plan”. It includes the following goals (City
of Vienna, 2015):
1. Fair – street space is allocated fairly to a variety of users and sustainable mobility
must remain affordable for all.
2. Healthy – the share of active mobility in every-day life increases; accident-related
personal injuries decline.
3. Compact – distances covered between work, home, errands and leisure activities
are as short as possible.
4. Eco-Friendly – mobility causes as little pollution as possible, the share of eco-
mobility in the trips made in Vienna and its environs is rising. The relative change
in the modal shift will be largest in bicycle traffic. In absolute figures, the largest
increase in the number of trips will be attributable to public transport.
5. Robust – mobility is as reliable and crisis-proof as possible. Mobility should be
possible without necessarily owning a means of transport.
6. Efficient – resources are used in a more efficient way, helped by innovative
technologies and processes.
The goals for Vienna span four policy domains and were subdivided into specific policy
goals for each domain (Table 2.2), each with its own impact measure.

LEVITATE | Deliverable D7.5 | WP7 | Final
12
Table 2.2: Overview of goals of the City of Vienna for a viable transport system of the future and corresponding
impact targets (WP4).
Policy field
Policy goal
Measured impact
Environment
Mobility causes as little pollution as possible
Modal split changes
Mobility
Resources are used in a more efficient way
Absolute final energy consumption of the Vienna
transport system
Distances covered between work, home, errands
and leisure activities are as short as possible
The share of trips done on foot or by bike to shop
for supplies or accompany someone as well as
distances covered for leisure time activities
Mobility is reliable and crisis-proof
Bicycle availability
Safety
Safe road travel
The number of traffic casualties and persons
injured in traffic accidents
Society
Better health: The share of active mobility in
every-day life increases
The share of people in the Viennese population
who are actively in motion for 30 minutes daily
as they run their daily errands
Fairness: Street space is allocated fairly to a
variety of users and sustainable mobility must
remain affordable for all
The total sum of spaces for cycling, walking and
public transport in all conversion and urban
renewal projects
These two city transport strategies reveal that CCAM could contribute toward achieving
these goals although specific policy will need to be adopted to make that achievable. For
each of the Policy domains described above, one or more key impact indicators have been
defined/operationalized for the Policy Support Tool that is intended to help policy makers’
decision-making concerning interventions that may support automated driving.
2.2 Expected automation impacts
It is expected that CCAM will have substantial impacts on road transport. Deliverable D3.1
(Elvik et al., 2019) presented a taxonomy of potential impacts of CCAM which makes a
distinction between direct, systemic and wider impacts. Direct impacts are changes that
are experienced by each road user on each trip. Systemic impacts are system-wide
impacts within the transport system and wider impacts are changes that occur outside
the transport system, such as changes in land use and employment. Moreover, a distinction
is made between primary impacts and secondary impacts. Primary impacts are
intended impacts that directly result from the automation technology, whereas secondary
impacts (rebound impacts) are generated by a primary impact.
Figure 2.1 presents the various impacts of the taxonomy and their expected interrelations
(based on scientific literature and expert consultation). In the figure, impacts are ordered
from those that are direct, shown at the top, to those that are more indirect or wider,
shown further down in the diagram. The diagram is inspired by the detailed model of
Hibberd et al. (2018)

LEVITATE | Deliverable D7.5 | WP7 | Final
13
Figure 2.1: Taxonomy of impacts generated by transition to connected and automated vehicles
Figure 2.1 shows the different paths by which impacts are generated by automation
technology. Three aspects of it are identified in Figure 2.1: vehicle design, level of
automation (SAE 1 to 5), and connectivity (Elvik et al., 2019). These characteristics of
technology can give rise to different impacts. For example, vehicle design - which includes
aspects such as vehicle size, setup of electronic control units, powertrain (fossil fuel or
electric) and ease of getting in or out the vehicle – will, through the technology built into
connected and automated vehicles, influence both vehicle ownership cost and vehicle
operating cost (Elvik et al., 2019). The choice of powertrain will influence propulsion energy
and energy efficiency of the engine. Vehicle design may also influence infrastructure design
and infrastructure wear, depending on, for example, the mass of the vehicle and its ability
for vehicle to infrastructure communication (Elvik et al., 2019). Finally, vehicle design may
influence travel comfort and individual access to transport. As an example, vehicles with
high ground clearance and no ramps will be difficult to access for wheelchair users.
Another example of pathways in Figure 2.1 concerns the primary impacts of CCAM on road
safety. Road safety is influenced by level of automation, as human operator errors will be
eliminated at the highest level of automation (there may still be software errors in
computer programmes operating the vehicle, but there will be no driver who can make
mistakes) (Elvik et al., 2019). The level of automation may also influence road safety
indirectly, by way of trust in technology, in particular before the highest level of automation
is attained. However, even fully automated vehicles will have to interact with non-
automated road users, who may place excessive trust in the capabilities of the technology
to detect them, brake or make evasive manoeuvres. Connectivity will influence safety by
reducing or eliminating speed variation between vehicles travelling in the same direction
and by shortening reaction times in case of braking (Elvik et al., 2019). Finally, road safety
and in the end public health will be influenced by potential changes in the amount of
Technological
driving forces
Direct
impacts
Systemic
impacts
Vehicle
design
Level of
automation Connectivity
Geographic
accessibility
Land
use
Wider impacts on Environment (green), Economy (blue), Safety (yellow), and Society (orange)
Travel
Comfort
Vehicle
emissions
Energy
efficiency
Public
finances
Propulsion
energy
Infrastructure
design
Air
pollution
Noise
pollution
Primary impacts
Individual
access to
travel
Vehicle
ownership
costs
Traffic data
generation
Individual
route choice
Vehicle
operating
costs
Vehicle
ownership
rate
Travel
Time
Trust in
technology
Inequality in
transport
Vehicle km
of travel
Road
safety
Parking
space
Optimisation
of route choice
Congestion
Modal split of
travel
Use and
valuation of
travel time
Road
capacity
Employment Commuting
distances
Public
health
Vehicle
utilisation
rate
Infrastructure
wear
Use of shared
mobility
services

LEVITATE | Deliverable D7.5 | WP7 | Final
14
congestion, vehicle kilometres of travel, changes in the modal split of travel and
optimisation of route choice (Elvik et al., 2019).
2.3 Expected impacts of cooperative and automated
freight transport
In the previous section (Section 2.2), a taxonomy of the impacts of automated vehicles
was described and also how these impacts are interrelated. In this section, the focus is on
the expected impacts on environment, mobility, society, safety and economy from
automation in the freight transport sector. The findings below are taken from the literature
study by Hu et al. (2019).
The Connected Automated Driving Roadmap (ERTRAC 2019) states that CCAM will
provide the opportunity to revolutionize the operation of freight transport. If used properly,
automated commercial freight vehicles could improve fleet efficiency, flexibility, and the
total cost of ownership. It has also great potential to effectively reduce traffic congestion-
related costs through vehicle platooning, improve driver behaviours, reduce driver costs,
and increase fleet mobility as well as safety.
There is not much research on CCAM in urban freight since this is the most difficult part to
be automated (ERTRAC 2019). The trends of city logistics indicate that the last mile
delivery is one of the more expensive, least efficient and most polluting sections of the
entire logistics chain (Gevaers et al 2014). With the introduction of CCAM, new business
models and operational concepts will emerge that will bring large changes for the road
freight transport sector. One of the major cost factors today is the driver or personnel in
general (Panteia 2015). Although the automation of urban freight transport is substantially
more difficult, and the implementation is not expected in the short or medium-term, it has
more possibilities and opportunities to bring substantial changes to the logistic system.
An essential application of urban freight will be automated parcel delivery. These use much
smaller than conventional delivery vans and operate off electricity. This addresses two
current problems, namely emissions and restrictions of road vehicles in narrow and
crowded areas typically found in older European city centres. On the parcel delivery side,
there are lots of projects on (sidewalk) delivery robots (Hu et al., 2019) but the operation
of delivery robots or micro-vehicles is still an under-researched topic (Baum et al. 2019).
The technical capabilities, limitations, challenges and potential time- and cost-savings of
current technologies are well described in a study by Jennings and Figliozzi (2019).
On the parcel receiving side, there are needs for compatible infrastructure for these
delivery robots. The automated parcel locker system is a natural solution for this (Hu et
al., 2019). These lockers are already commercially used where consumers can either
receive or send a parcel from (Hu et al., 2019).
Within Levitate, WP7 estimated that cooperative and automated freight transport will
impact primarily on the environment, mobility and road safety.
Environmental impacts
For freight transport, vehicle automation does not necessarily lead to direct environmental
impacts. ERTRAC (2019) identifies vehicle design, drivetrain, energy composition, and
operational efficiency as main factors that influence how environment-friendly and

LEVITATE | Deliverable D7.5 | WP7 | Final
15
sustainable future freight transport will be. It should be noted that these factors are not
necessarily directly connected to CCAM. Essentially, there is not much difference between
achieving the freight volume (expressed in tonne-kilometres) by vehicles driven by
conventional drivers or automated transport (Hu et al., 2019).
Although a direct connection between CCAM and positive environmental impacts is
ambitious, it is plausible that CCAM could contribute to environmental impact in a broader
sense:
• For platooning, lots of scientific research has been done and these studies indicate that
it can reduce fuel consumption (e.g., Mello & Bauer, 2019).
• For drivetrain and energy, there is a correlation between E-mobility and CCAM on the
level of technology innovation. Therefore, CCAM indirectly reduce CO2 emissions
provided electric energy is generated in an environmentally friendly way.
• New business models and logistic concepts enabled by CCAM will likely increase the
operational efficiency and therefore reduce energy consumption in general.
Studies have shown that using smaller, electrified vehicles and robots for urban freight
transport may reduce emissions (Jennings et al., 2019, Figliozzi et al., 2020). There are
concepts where the autonomous delivery robots are airborne drones (Dorling et al. 2017),
but the operation of drones especially in crowded urban environment is controversial and
legally challenging. Therefore, this not further considered in the Levitate project nor is it
discussed in this WP7 synthesis report.
Mobility impacts
The Connected Automated Driving Roadmap states that CCAM will provide the opportunity
to effectively reduce traffic congestion of freight transport through vehicle platooning
(ERTRAC, 2019). Also, automated Light Goods Vehicles (LGVs) provide for more efficient
delivery with first and last mile access to consolidation centres which will reduce urban
congestion due to reduction of the number of trips in the city centre (Hu et al., 2019;
2021a). Below further explanation is given of this expected development.
Automated consolidation of freight transport, i.e., parcel delivery companies consolidating
their parcels at city-hubs instead of operating independently and delivering parcels straight
to their final recipients – will likely reduce travel or mileage of freight transport (Hu et al,
2021a). Ideally, the city-hubs and the last-mile delivery operate on a white-label basis,
i.e., the delivery vehicles are not bound to a specific delivery company but operate the
service for all companies. This will remove a lot of redundancy of trips in the delivery
system (Hu et al., 2021a).Furthermore, since city-hubs are closer to the city centre than
the original distribution centres, final delivery routes in a consolidated scenario will be
significantly shorter producing positive impacts on the traffic and the environment (Allen
et al., 2012; Quak et al., 2016).
Safety impacts
Safety is a critical issue since freight vehicles, largely composed of trucks, vans and other
large vehicles, have the potential to cause severe crashes with a high injury rate. The
fatality rate of crashes involving freight vehicles is relatively high compared to the number
of collisions (Eurostat, 2015). This is the main driving factor behind the development of
many ADAS which target improving road safety (see section 3.3 for a detailed description
of these effects). Beyond ADAS, the introduction of level 3 and level 4 automation,
especially in urban areas, still requires substantial research and testing. ERTRAC (2019)
states that technology must be proven to ensure functioning without any problems in

LEVITATE | Deliverable D7.5 | WP7 | Final
16
various climates and traffic conditions and that during the transition phase, trials in a
controlled or specific area at specific times should be encouraged.
In the automated freight delivery scenarios, both non-consolidated and consolidated
parcels are delivered by small delivery robots which generate a new set of interactions
potentially impacting road safety. In an advisory report about an on-road test with delivery
robots in the Netherlands, Van Petegem et al. (2018) identified potential road safety risks
related to the interaction between such robots and other road users. While some of these
risks specifically apply to the on-road test, others are more broadly applicable. The latter
are related to the unpredictability of the robot’s behaviour, its speed in comparison to
pedestrians, its low height (others might not see the robot), and the robot blocking
sidewalks (especially for wheelchairs and mobility scooters).
2.4 Sub-use cases
This section describes the automated urban freight transport sub-use cases that were
studied in WP7.
Automated urban delivery
The automated urban delivery sub-use case compares the performance of manual delivery
(using personnel) and (semi-)automated parcel delivery concepts in urban areas. . While
the automated road-based (delivery) vehicles are well-studied, the operation of delivery
robots or micro-vehicles is still an under-researched topic (Baum et al. 2019). Studies show
that using smaller, electrified vehicles and robots addresses several acute problems:
emissions, navigation in confined inner-city areas and the limitation of working time for
manual parcel delivery (Jennings et al., 2019, Figliozzi et al., 2020).
Based on the current manual delivery process, the envisioned automation technologies and
concepts that will emerge in the next decades, the following scenarios were considered
appropriate for automated urban delivery:
• Manual delivery (status quo) is used as a baseline scenario for comparison.
• Semi-automated delivery assumes that the delivery process is not fully automated
yet. While the delivery van is automated, personnel are still undertaking the delivery
task. However, since they do not need to switch between delivery and driving tasks,
time can be saved during each stop.
• Automated delivery is where so-called robo-vans and small autonomous delivery
robots replace all service personnel and operate beyond the road (to the off-loading
areas using pavement, pedestrian areas, etc.). The automated van functions as a
mobile hub where they perform short delivery trips to end-customers, i.e., a hub-
and-spoke setup with moving hubs. This human-less delivery process can be carried
out during off-peak hours when road traffic volumes are lower and be extended to
evening or night-time delivery. For this concept, we assume that the parcel capacity
of the van will be significantly reduced. The main reason is that it has to carry the
delivery robots and the necessary equipment to load them.
• Automated night delivery is the same as above, but deliveries are limited to night-
time delivery only. Since the delivery time is restricted to night-time only, this
scenario will increase the fleet size since the same volume of deliveries will have to
be made in significantly less time compared to the previous scenario.
The delivery performance and their main limiting factors are shown in Table 2.3. (Hu et al.,
2021a).

LEVITATE | Deliverable D7.5 | WP7 | Final
17
Table 2.3: Performance of the delivery scenarios and their main limiting factors (red).
Delivery scenarios
Sub-use case specific scenarios - Automated urban delivery
Delivery scenario parameters
Delivery shifts
Avg.
parcels
per shift
Avg. parcels
per stop
Service
time per
stop
Delivery
vehicle
Manual delivery 6:30 – 15:00 150 Variable 5 Van
Semi-automated
delivery
6:30 – 15:00 180 Variable 4
Automated
van
Automated delivery
9:00 – 15:00,
18:00 – 24:00, 0:00 – 6:00
100 Variable 10 Robo-Van
Automated night
delivery
18:00 – 24:00, 0:00 – 6:00 100 Variable 10 Robo-Van
Automated freight consolidation
The automated consolidation sub-use case is a continuation of automated urban delivery.
In this setting, the parcel delivery companies will consolidate their parcels at city-hubs
instead of operating independently and delivering parcels straight to their final recipients.
Ideally, the city-hubs and the last-mile delivery operate on a white-label basis, i.e., the
delivery vehicles are not bound to a specific delivery company but operate the service for
all companies. Compared to the current delivery system this significantly improves
efficiency. Furthermore, since these city-hubs are closer to the city centre than the original
distribution centres, final delivery routes in a consolidated scenario are significantly
shorter. This has a positive impact on the traffic and the environment (Allen et al. 2012,
Quak et al. 2016).
For the automated freight consolidation SUC the following delivery scenarios were
considered:
• Manual delivery (status quo) refers to the same baseline scenario as in the
previous SUC
• Automated delivery refers to the automated delivery scenario as in the previous
SUC
• Manual delivery with bundling at city-hubs uses bundled parcel delivery via
city-hubs, but both the servicing of city-hubs and the delivery to end-customers are
done manually.
• Automated delivery with bundling at city-hubs is the final scenario that
combines the automated delivery via robo-vans and the city-hubs for bundling.
In all automated scenarios, it was assumed that the delivery is carried out throughout the
day and night, as was the case with the automated urban delivery SUC above. However,
the transport from distribution centres to city-hubs is done during the night via automated
trucks. Solutions or prototypes for automatic loading and unloading already exist for
packages and pallets (Cramer et al., 2020).
Hub-to-hub
This sub-use case studies the impacts of AV truck terminals functioning as transfer hubs.
The goal of these hubs is to facilitate the transition towards level 5 automation by
supporting the operation of level 4 automated trucks that can operate on highways but not
in urban environment. It is assumed that outbound freight containers from the city are
passed to AV trucks at the terminal, which then take over the long-haul highway segment.

LEVITATE | Deliverable D7.5 | WP7 | Final
18
At an AV truck terminal of the destination city, the container is passed to a manually
operated truck to bring it to the destination. An ideal location for such a terminal is on the
city outskirts with direct or good access to the highway road network. Figure 2.3 shows
how this concept should work.
Figure 2.2: Function of the automated transfer hub. Human-operated trucks deliver the containers to the transfer
hub (yellow arrow) and from there automated trucks carry them on to the highway (blue arrow).
The main benefit of deploying an AV truck terminal is that:
• Long-haul freight transport can relatively easily be automated and this can translate
to significant cost reductions.
• For the urban highway, it is possible to reduce the usage during daytime and shift
the freight transport towards night. This can be achieved by coordinating AV trucks
to only depart during night hours.
A study by Berger (2016) shows that this concept is highly attractive for the long haul,
where the driver wage accounts for one third of the total transport costs. It is also expected
that the hub-to-hub connections will be dominated by autonomous trucks, while hub-to-
delivery will be executed by hybrid and full-electric small to medium sized trucks (Novak,
2016).
For this SUC, a small area around a potential AV truck terminal including an urban highway
segment with ramps was considered. Two scenarios are compared:
• Status quo (Baseline) where manual container trucks operate between their
origin and destinations directly throughout the day.
• Operation via transfer terminal: During the day, manual trucks deliver their
freight from origin to the AV truck terminal. At night, AV trucks ship the containers
from the terminal to the destination terminals. Similarly, AV trucks from other

LEVITATE | Deliverable D7.5 | WP7 | Final
19
terminals arrive throughout the day and night, while the further transport into the
city via manual trucks happen during the day.
Bridge platooning
Truck platooning on urban highway bridges is a special SUC in a sense that the assessment
methods and the obtained impacts are different from the other SUCs. This SUC is for study
purpose and will not be included into the Policy Support Tool (PST) estimator, but
nevertheless it is a very important study subject. Although the damage is not a short-term
effect and the probability of a potential failure is small, the possible damage in case of
failure is enormous (Hu et al, 2021b). In Deliverable 7.3 a full technical description is given
of the models and methods for the truck Platooning on urban highway bridges SUC (Hu et
al., 2021b). In this synthesis a non-technical abbreviated description of methods is
provided in the report and a summarised technical description is included in Appendix C.
The model for traffic loads on bridges is standardized and defined in the EN-1991-2
(Eurocode0F
1). It is representing the effects of vehicle loading and is mainly used in the
design of new bridges and with modifications in the assessment of the load bearing capacity
of existing bridges (which are usually defined individually for each country respective to
the bridge construction date). This traffic load model was derived based on axle-load
measurements performed near Auxerre, France (Braml 2010, Sedlacek 2008) in 1986 and
includes statistical assumptions for the future traffic volumes.
To determine the expected maximum bridge loading, traffic simulations were performed
that calculated the bridge loading caused by many years of simulated traffic. This approach
was used to compare the maxima of bridge loading in different traffic scenarios including
also generic future load assumptions for truck platoons (Hu et al., 2021b).
The following types of traffic scenarios were analysed:
• Current heavy traffic (status quo) used as a baseline scenario for comparison.
• Heavy traffic with truck platoons: different truck-platoons compositions mixed into
the current traffic.
• Intelligent access control: heavy traffic with mixed-in truck platoons and imposed
restrictions of minimum vehicle distances within platoons depending on carrying
capacity of bridges.
Bridge strengthening is an option to deal with increased traffic load requirements (such as
caused by closely spaced trucks in a platoon), but it can be very costly. To avoid these
costs, the option of intelligent access control is a possible alternative. The system of
intelligent access control presumes communication between truck platoons and the road
administration. The basic idea is that platoons dynamically adjust their headways (distance
between vehicles) depending on the load-carrying capacity of bridges ahead of them, this
to prevent overloading of the bridges. In the practical implementation, the road network
should be divided into sections, and one required headway should be prescribed for each
road section. This value should be governed by the most unfavourable bridge structure in
each road section, which is probably the bridge with the largest span. The value of the
1 EN 1991-2: Eurocode 1: Actions on structures - Part 2: Traffic loads on bridges, accessed 10 november 2021
at: https://www.phd.eng.br/wp-content/uploads/2015/12/en.1991.2.2003.pdf

LEVITATE | Deliverable D7.5 | WP7 | Final
20
prescribed vehicle distance valid for current road section must then be communicated to
truck platoons as they are travelling across different road sections. The communication of
this information could be executed in real-time, or alternatively it could be provided prior
to the journey for a selected route or parts of the road network (Hu et al., 2021b).
2.5 Assessment methods
The types of impacts that are presented in Deliverable 3.1 (Elvik et al., 2019) have been
estimated using three main assessment methods, Delphi panel method, traffic
microsimulation, and operations research. In addition to these main methods, for the special
sub-use case of truck platooning on bridges a combination of bridge modelling and traffic
simulations were used (Hu et al., 2021b).
The Delphi method is a process used to arrive at a collective, aggregate group opinion or
decision, by surveying a panel of experts. This concept was developed by the RAND
Corporation for the military in order to forecast the effects of new military technology on
the future of warfare, and then continued to make multiple practical applications of this
method (Dalkey & Helmer, 1963). The Delphi methodology is based on a repetitive interview
process in which the respondent can review his or her initial answers and thus change the
overall information on each topic (Hsu & Sandford, 2007).
Traffic microsimulation was used to forecast short-term impacts to be able to develop
relationships that can infer dose (in terms of introduction of sub-use case) and response
(selected impact). Traffic microsimulation also provides further input to assess medium-
term impacts by processing those results appropriately to infer such impacts.
In WP7, the traffic microsimulation framework AIMSUN was used to assess the traffic
impacts such as congestion and road safety (Hu et al., 2021b) using the network of the city
of Vienna. Compared to LEVITATE’s WP5 (automated urban shuttles) and WP6 (passenger
cars), microsimulation simulation played a smaller role in WP7 for three reasons (Hu et al.,
2021b):
• Freight vehicles only take a small share of the traffic volume in urban areas and their
impact is limited when compared to the overall traffic.
• Parameters of automated freight vehicles are still uncertain compared to automated
passenger cars. Therefore, the results are less reliable.
• Freight operations are plannable; therefore, operations research is more suitable for
assessing the fleet size and mileage.
The limitations in simulating automated freight vehicles are particularly relevant for the
measurement of road safety impacts, due to their dependence on vehicle driving behaviour.
Microsimulation was used to study the expected impacts of the freight SUCs on road safety
on the Vienna city network. The frequency with which vehicles in the microsimulation
entered potentially dangerous interactions (ie. traffic “conflicts”) was measured using the
Surrogate Safety Assessment Model (SSAM) developed by the Federal Highway
Administration (FHWA). A prediction is subsequently made for the share of conflicts which
would result in a crash using the probabilistic method developed by Tarko (2018).
As CAVs exhibit different driving behaviours, their behavioural parameters (eg. time gap,
clearance, maximum deceleration) are adjusted for 1st and 2nd generation automated
vehicles, leading to changes in the number of conflicts. For freight vehicles, less knowledge

LEVITATE | Deliverable D7.5 | WP7 | Final
21
was available on the behavioural parameters of future automated freight vehicles;
therefore, some parameters assumed the values of 1st generation CAVs and others were
based on assumptions, leaving some uncertainty. For this reason, road safety results within
LEVITATE have been estimated both including (Work Package 7) and excluding (Work
Package 5 & 6) freight vehicles. Nevertheless, the road safety impacts of increasing
automation in a mix of both passenger and freight vehicles are an interesting development,
especially for these sub-use cases where the deployment of automated freight vehicles is
varied.
Furthermore, in Work Package 7 microsimulation provides an estimation of traffic impacts
for a reference delivery trip which serves as input for upscaling via operations research in
the hybrid assessment approach (further explained below).
Operations research is widely used in freight optimization and calculates results for freight
transport costs, fleet operation costs, and vehicle mileage (Lagoria et al., 2016). Compared
to private passenger transport, freight transport is less time-critical and plannable on an
operational basis, which makes operations research a viable approach for the automated
delivery and automated consolidation SUCs. Vienna was taken as the basis for analysing
these SUCs due to the availability of high-quality data.
As explained in D7.2, in operations research first the data on delivery addresses in Vienna
were generated, subsequently a method (optimisation algorithm) was applied to assess
route planning, and finally, after all delivery trips were calculated, the number of routes and
the sum of their lengths were used as input for the corresponding cost and distance impact
indicators (Hu et al., 2021a). For the sake of applicability of assessment methods, it was
assumed that for the appropriate level of automation, adequate infrastructure exists (e.g.,
for receiving parcels during night).
A more technical description of method is given below. Based on the estimated market
shares of logistic providers and the reported parcel volumes in Vienna, delivery addresses
were generated and randomly distributed but weighted according to the population density
of the respective districts in the city of Vienna (Hu et al., 2021a). The underlying algorithm
for calculating the delivery scenarios was based on optimising the routing of the delivery
vehicles. In all delivery variants considered, the delivery points were assigned to a depot
from which the parcels are delivered. Depending on the delivery scenario, this depot can
be a logistics centre or a city-hub (in case of consolidated delivery). Subsequently, a
problem instance of the Capacitated Vehicle Routing Problem (CVRP) (Toth and Vigo, 2014)
was generated for each depot, with the delivery addresses acting as so-called customers.
Finally, these instances were solved using the Savings algorithm (Clarke & Wright, 1964).
Finally, the required consolidation trips between the individual depots were calculated. If
the demand for parcels at a delivery address exceeded the capacity of a single delivery
vehicle, it was divided into multiple virtual delivery addresses at the same location, with
each of these having a maximum demand for parcels equal to the capacity of the delivery
vehicle (Hu et. al., 2021a).
For the automated delivery and automated consolidation SUCs, a hybrid assessment
method based on a combination of micro-simulation and operations research was applied.
Micro-simulation was used to capture the traffic impacts of a typical delivery trip of one
delivery vehicle. These impacts were then scaled up using operations research, see Figure
3.2.

LEVITATE | Deliverable D7.5 | WP7 | Final
22
Assessing impacts of bridge platooning
Truck platooning on urban highway bridges is a special SUC in a sense that the assessment
methods and the obtained impacts are different from the other SUCs. This SUC is for study
purpose and will not be included into the PST estimator, but nevertheless it is a very
important subject. For truck platooning, there already exist a good amount of scientific
work, but the impacts on the bridge infrastructure is under-researched (Hu et al., 2021b).
Although the damage is not a short-term effect and the probability of a potential failure is
small, the possible damage in case of failure is enormous.
To determine the expected maximum bridge loading, traffic simulations were performed
that calculated the bridge loading caused by many years of simulated traffic. This approach
was used to compare the maxima of bridge loading in different traffic scenarios including
also generic future load assumptions for truck platoons (Hu et al., 2021b). The following
types of traffic scenarios were analysed:
• Current heavy traffic (status quo) used as a baseline scenario for comparison.
• Heavy traffic with truck platoons: different truck-platoons compositions mixed into the
current traffic.
• Intelligent access control: heavy traffic with mixed-in truck platoons and imposed
restrictions of minimum vehicle distances within platoons depending on carrying capacity
of bridges.
To assess the impacts of bridge platooning the Ultimate Limit States (ULS) of midspan
bending moment, the shear force and the horizontal force from braking were analysed.
Their values in different traffic cases were compared in traffic simulations. To make the
results on different bridges comparable, the impacts were not expressed in absolute values
of bridge internal forces, but relative to the bridge internal forces caused by Eurocode load
model LM1. The forces caused by LM1 load model are deterministic, since the load model
is deterministic (Hu et al., 2021b).
Traffic
simulation
Congestion,
Safety
per vehicle
per km
Congestion,
Safety
upscaled to
city level
Fleet size,
Operating cost,
Freight
transport cost
on city level
Operations
Research
Figure 2.3: Flowchart for the hybrid assessment approach.

LEVITATE | Deliverable D7.5 | WP7 | Final
23
The impact of simulated traffic was evaluated in terms of the probability of exceeding the
effects of load model LM1. Since new bridges are designed for the loads of load model LM1,
it was assumed that they have the respective load-carrying capacity. The definition of load
model LM1 according to EN 1991-1 presumes that its exceedance probability in 50 years
is 5%. This probability - 5% in 50 years - is regarded as the “code level” (Hu et al, 2021b).
The resulting bridge forces were evaluated in terms of the probability, that they exceed
the forces from Eurocode load models (Hu et al., 2021b). If the probability, that a resulting
50-years-extreme-value distribution exceeds the force from a Eurocode load model, is
above 5% the structural safety can be regarded as reduced. Thus, higher exceedance
probabilities mean lower structural safety.
2.6 Approach to synthesizing results
The goal of this Deliverable is to summarise the more detailed results presented in D7.2-
D7.4 (Hu et al., 2021a, b, c). As has been explained in Section 1.2, the impacts expected
from an increasing penetration rate of CAVs in the total vehicle fleet as well as the
implementation of an automated freight services were studied using three primary
methods: microsimulation, operations research and Delphi consultation. Within each
methodology, a baseline and automated freight scenarios were defined and quantified (see
Section 2.4).
For the purposes of this synthesis, the results estimated within Work Package 7 of
LEVITATE have been condensed in order to provide an overall overview (Table 2.4). The
full results, broken down per scenario, can be found in Appendix A. In Chapter 3, the
quantified results of Work Package 7 are summarised per SUC in order to arrive at expected
trends (% change) per impact (see Table 2.4; rightmost column). Given the many
uncertainties in prediction, it is obvious that any predicted values are associated with large
uncertainty. For the WP7 results, it was decided not to estimate confidence intervals based
on the standard error derived from repeated trail runs of models since these intervals
would be broad and non-informative. Also, the estimation of these intervals would tend to
be biased in itself since the input variables and assumptions in the models are very likely
much stronger determinants of predicted values than the variability in sample runs.
The following approach was used in order to summarize and structure the quantified results
for WP7:
• Impacts are presented as a percentage change from the Baseline 100-0-0 scenario,
where neither automated freight transport nor CAVs have been implemented in the
network and all vehicles are human-driven. These percentage changes are reported
across increasing market penetration rates of CAVs throughout the entire vehicle fleet
in the network, as used throughout LEVITATE.
• The Baseline refers to a “no intervention” scenario which is essentially the expected
autonomous development of CAVs from human dependence to human independence
(see Section 1.1). In the Baseline scenarios there is no automated freight transport
added to the network.
• The impacts of CAVs alone on network performance can be established by comparing
the Baseline 0-0-100 scenario (0% human-driven vehicles) to the Baseline 100-0-0
scenario (100% human-driven vehicles).
• The specific effect of an automated freight transport sub-use case can be
determined by comparing the Baseline situation at a given CAV penetration rate with
the respective SUC results; the difference between the baseline and the SUC is the

LEVITATE | Deliverable D7.5 | WP7 | Final
24
added effect created by implementing the specific SUC intervention in the simulated
network.
• For the microsimulation study, several scenarios were estimated for the two urban
delivery sub-use cases: automated urban delivery and automated consolidation. These
involve simulations conducted on both central and periphery networks within Vienna,
varying the delivery time window (daytime vs. night-time), and semi-automated
(staffed by delivery personnel) vs. fully-automated (robotic delivery) vehicles. As
described in Deliverable D7.3 (Hu et al., 2021b), the urban and periphery results were
scaled up to arrive at estimations for the entire city of Vienna. The results presented in
Chapter 3 reflect the combination of urban & periphery scenarios for fully-
automated delivery vans. The full breakdown of results per scenario can be found in
Appendix A.
Table 2.4: Synthesized sub-use case scenarios from Deliverables 7.2-7.4
Method
Sub-use case
Scenarios
Synthesized results and
measured effect
Microsimulation
(Vienna
network)
Baseline
• Manual delivery; Urban
network
• Manual delivery; Periphery
network
• Hub-to-hub network; no
transfer hub
Baseline; manual delivery
(combined urban & periphery):
% change
Baseline; no transfer hub
(combined urban & periphery):
% change
Automated
delivery
• Fully-automated:
o Urban; daytime
o Periphery; daytime
o Urban; night-time
o Periphery; night-time
• Semi-automated
Fully-Automated delivery
(combined urban & periphery):
% change
Automated
urban
consolidation
• Automated delivery with
bundling at city hubs
o Urban; daytime
o Periphery; daytime
o Urban night-time
o Periphery; night-time
• Manual delivery with
bundling at city hubs
Automated delivery with
bundling at city hubs
(combined urban & periphery):
% change
Hub-to-hub
automated
transport
No scenarios
Hub-to-hub automated
transport (with transfer hub):
% change
Delphi study
(expert survey)
Baseline
No scenarios
Baseline:
% change
Automated
delivery
• Fully automated delivery
• Fully automated delivery
with night shifts only
Fully automated delivery:
% change
Automated
consolidation
No scenarios
Automated consolidation:
% change
Hub-to-hub
No scenarios
Hub-to-hub:
% change

LEVITATE | Deliverable D7.5 | WP7 | Final
25
In addition, the results of two quantitative methods of Operations research (D7.2) and
Bridge modelling (D7.3) are treated separately in this synthesis. These methods do not
quite fit in with the approach described above where impacts have been estimated for
different market penetration rates of AVs. We have summarized the calculations and
findings based on these quantitative methods separately in Sections 3.4 and 3.5. To be
clear it should be pointed that the results of operations research, as is the case with
the other methods, have been incorporated in the Policy Support Tool (PST) of
LEVITATE.

LEVITATE | Deliverable D7.5 | WP7 | Final
26
3 Main findings: quantified impacts
This chapter presents a summary description of the impacts that were quantified
in the LEVITATE Deliverables 7.2 to 7.4. The findings are presented for policy
domains Environment (Section 3.1), Mobility (Section 3.2), Society – Safety –
Economy (Sections 3.3 and 3.4). A distinct subject concerns the impact of truck
platooning on bridges – findings on this are presented in Section 3.5. The sections
3.1 to 3.3 describe synthesised results in accordance with the approach described
in Section 2.5. The findings in Sections 3.4 and 3.5 are based on quantitative
analyses that differ from the general approach, in that the impacts are not
estimated for different market penetration rates of automated vehicles. In
addition to the summary in this chapter, a detailed overview of quantified impacts
of D7.2 to 7.4 is added in Appendix A.
3.1 Impacts on the environment
In Work Package 7, two indicators were used to estimate impacts on the environment of
freight transport: carbon dioxide (CO2) emissions and energy efficiency (see Table 3.1).
Their importance for the environment has been widely documented (e.g., EEA, 2020).
Carbon dioxide emissions are the primary driver of global climate change; it is widely
recognised that in order to decrease the negative impacts on climate change, the world
needs to urgently reduce these emissions. Improving the efficiency of services and
technologies in urban transport that use energy from fossil fuels will help reduce emissions.
Table 3.1: Environmental impact definitions
Impact Definition Methodology
Energy efficiency
Average rate (over the vehicle fleet) at which
propulsion energy is converted to movement Delphi
CO2 due to freight
vehicles
Concentration of CO
2
pollutants as grams per
vehicle-kilometre (due to road freight transport
only)
Microscopic simulation
Table 3.2 presents an overview of the estimated effects resulting from an introduction of
a number of automated freight transport services (represented by the SUCs)on energy
efficiency and the CO2 emissions. The sub-use cases considered were automated delivery,
automated consolidation and hub-to-hub delivery and are fully described in D7.1. The
estimates are based on results from the Delphi study and the AIMSUN microsimulation
modelling study on the Vienna road network.

LEVITATE | Deliverable D7.5 | WP7 | Final
27
Table 3.2: Estimated impacts of automated freight transport services on CO2 emissions and energy efficiency:
Delphi and microsimulation results. Measured in terms of percentage change with respect to the Baseline
100-0-0 scenario.
Ma
rket penetration rate: AVs in Background vehicle fleet
(Human-driven vehicle - 1st Generation AV - 2nd Generation AV)
100-
0-0
80-20-
0
60-40-
0
40-40-
20
20-40-
40
0-40-
60
0-20-
80
0-0-
100
Impact Sub-use Case % % % % % % % % Method
Energy
efficiency
of freight
Baseline 0,0 -3,7 6,5 8,2 11,9 16,0 16,0 16,0
Delphi
Automated
delivery
0,0 6,1 7,8 11,1 14,8 20,4 20,4 20,4
Automated
consolidation
0,0 7,4 12,5 16,6 20,7 25,2 25,2 25,2
Hub-to-hub 0,0 5,6 7,8 13,5 18,2 18,2 18,2 18,2
CO2
emissions
of freight*
Baseline; manual
delivery
0 -21 -50 -80 -90 -100 -100 -100
Micro-
simulation
(Vienna)
Automated
delivery
-100
Automated
consolidation -100
Hub-to-hub -100
Note * -CO2 emissions are for freight vehicles only. The contribution of freight to overall emissions is small and
the impact of the SUC too small to be meaningful so only effects on freight transport are modelled. Also, the
emission impacts modelled for the SUCs assumes 100% electric powered freight vehicles from the outset CO2
emissions are eliminated (the baseline scenario assumes that human-driven vehicles are still traditionally fuelled
by fossil fuel derivatives)
Delphi results
According to the experts, the baseline development of the energy efficiency of freight
vehicles (used for road transport) is positive; in the baseline, energy efficiency improves
by 6% to 16% once human-driven vehicles are reduced to 60% or lower of the vehicle
fleet and replaced by first- and second-generation AVs.
The expected impacts on energy efficiency of the three-freight service SUCs, namely
automated delivery, automated consolidation and hub-to-hub, are all positive. Compared
to the baseline development, experts estimate that the introduction of automated delivery,
automated consolidation and hub-to-hub in freight vehicles will further improve energy
efficiency. Especially the estimates for automated consolidation are positive withenergy
efficiency being 1.5 to 2 times higher compared to the baseline.

LEVITATE | Deliverable D7.5 | WP7 | Final
28
Microsimulation results
For the automated delivery and automated consolidation SUCs, the CO2 emissions caused
by the freight vehicles were estimated on the basis of the total driven kilometres presented
in D7.2 (Hu et al., 2021a). The impact on the overall emissions including the background
traffic would be not visible since the share of the freight vehicles is too low. Therefore, we
only consider the freight vehicles here. The microsimulation results in Table 3.2 indicate
the following:
• The baseline results for CO2 emissions of freight vehicles show large reductions (50%)
when the share of human-driven vehicles is at 60%- and first-generation automated
vehicles is at 40%. Larger reductions of 80% to 100% are achievable when share of
human-driven vehicles drops to 20% and below and second-generation vehicles
increase to 100%. This gradual reduction reflects the transition in the microsimulation
from a freight vehicle fleet which is 100% human-driven and diesel-fuelled, to a fleet
which is 100% autonomous and electric (assumed to be emission-free).
• In each of the three sub-use cases, a 100% reduction of emissions occurs once electric
freight vehicles fully replace conventional vehicles. In LEVITATE it is assumed that all
freight AVs will be electric and therefore emission-free while the manual freight vehicles
use internal combustion engines fuelled by diesel, which is the standard at the moment.
As the automated freight sub-use cases are implemented at all penetration rates, even
when all other vehicles are human-driven (100-0-0), this complete reduction in freight
emissions is also predicted at all penetration rates.
3.2 Impacts on mobility
This section presents the main findings of the studied impacts on mobility. For the area of
freight transport, two mobility indicators - average travel times in the network and
congestion experienced by freight transport vehicles - were studied (Table 3.3). The size
of these impacts is estimated from two methodologies: the Delphi expert panel and the
AIMSUN microsimulation modelling using the Vienna road network.
Table 3.3: Mobility impact definitions
Impact Definition Methodology
Travel time
Average duration of a 5Km trip inside the city
centre Delphi
Congestion for freight
vehicles
Average delays to traffic (seconds per
vehicle-kilometre) as a result of high traffic
volume, measured for freight vehicles only.
Microscopic simulation
Table 3.4 presents the estimated impacts on the mobility (expressed for travel time and
congestion) of freight transport vehicles under baseline conditions and for the three sub-
use case conditions. In this table for each impact the % effects are reported in respect to
Baseline 100-0-0. The difference between the baseline effect and the specific SUC under
consideration, and given the penetration rate, is the effect of the SUC itself. In this table
a decrease in travel time and congestion (denoted by a “-“) implies a favourable effect.

LEVITATE | Deliverable D7.5 | WP7 | Final
29
Table 3.4. Estimated impacts of automated freight transport services on travel time and congestion, measured
in terms of percentage change with respect to the Baseline 100-0-0 scenario.
Market penetration rate: AVs in Background vehicle fleet
(Human-driven vehicle - 1st Generation AV - 2nd Generation AV)
100-0-
0
80-20-
0
60-40-
0
40-40-
20
20-40-
40
0-40-
60
0-20-
80
0-0-
100
Impact Sub-use Case % % % % % % % % Method
Travel time
Baseline 0,0 6,1 6,3 5,5 2,2 0,5 0,5 0,5
Delphi
Automated delivery 0,0 1,7 1,7 -3,7 -6,8 -4,4 -4,4 -4,4
Automated
consolidation
0,0 -4,2 -4,2 -8,8 -9,8 -11,3 -11,3 -11,3
Hub-to-hub 0,0 -2,9 -3,2 -4,8 -6,4 -6,4 -6,4 -6,4
Congestion
for freight
vehicles
(delay per
veh-km)
Baseline; no
automated delivery
0,0 -15,5 -6,6 -4,8 -7,9 -17,2 -11,6 -8,7
Micro-
simulation
(Vienna)
Automated
delivery
-42,4 -38,9 -42,2 -49,0 -35,6 -46,1 -49,9 -42,3
Automated
consolidation
-42,4 -38,9 -42,2 -49,0 -35,6 -46,1 -49,9 -42,3
Baseline; no
transfer hub
0,0 -9,3 -11,3 -17,5 -19,6 -22,7 -23,7 -24,7
Hub-to-hub; with
transfer hub
0,0 -11,3 -17,5 -21,6 -23,7 -24,7 -24,7 -26,8
Delphi results
According to the experts participating in the Delphi consultation, in the baseline condition
travel times in the network will increase by 5to 6% when automated vehicles are first
introduced (20-60% of vehicle fleet) before settling back to roughly the starting conditions
(less than 1% increase) once all vehicles are automated. Compared to the partly
unfavourable development of travel under baseline conditions, all three sub-use cases are
associated with more favourable developments for travel time. Under the three SUCs the
estimated travel times are reduced once conventional (human-driven) vehicles are down
to 40% of vehicle fleet. The most positive expectations are for the automated consolidation
SUC, where it is estimated that travel time reductions of between 9% and 11% are possible
once CAVs make up more than 60% of the fleet. The hub-to-hub case is expected to reduce
travel time by 3% to above 6% once second-generation vehicles make up 40% or more of
the fleet. The automated delivery is expected to reduce travel time by about 4%-7% once
second-generation vehicles make up 20% or more of the fleet.
In brief, all three SUC’s are expected to result in more favourable development of travel
time, i.e., less travel time, when compared to the baseline scenario. Automated
consolidation shows the most promising results with regard to reduction of travel time. The
reduction in travel time peaks once human-driven vehicles are reduced to 20% and below
and tends to remain constant after that.
Microsimulation results
According to the AIMSUN microsimulation results using the Vienna road network:
• In the manual delivery baseline scenario, the congestion delays experienced by delivery
vehicles vary between a 5-17% reduction when automated vehicles are introduced into
the network. The reduction in congestion is lowest when the vehicle fleet is roughly
equally split between human-driven and autonomous vehicles (60-40-0 and 40-40-20).

LEVITATE | Deliverable D7.5 | WP7 | Final
30
• The sub-use cases of automated delivery and consolidation are associated with a 36%
to 50% reduction in congestion experienced by delivery vans under the eight
penetration rate scenarios. This is substantially more than in the baseline scenario,
suggesting that both forms of automated delivery will bring additional benefits for
congestion levels in urban environments similar to the one modelled in this study.
• While the automated delivery and consolidation sub-use cases exhibit the same
normalized congestion per vehicle kilometre, the automated consolidation sub-use case
is expected to reduce the total kilometres travelled due to more efficient
routing/logistics. This suggests that the total amount of congestion delays experienced
may further be reduced with consolidation.
• A shift to night-time-only delivery (see nighttime scenarios in Appendix A), which could
be facilitated with driverless delivery vehicles, is expected to result in a large (over
90%) reduction in congestion experienced by the delivery vans due to the much lower
traffic volumes during night-time hours.
• The estimated developments in congestion for the hub-to-hub SUC shows a slight
improvement (less congestion) when a transfer hub is implemented compared to the
baseline condition with no transfer hub.
3.3 Impacts on society, safety & economy
In this section the main findings on the wider impacts of automated freight transport
services in city areas that experience increasing numbers of connected and automated
vehicles are presented. Table 3.5 presents the expected impacts on the interconnected
policy domains of society (health, and access to services), road safety and economy. The
impacts on road safety are based on results from the AIMSUN microsimulations whereas
the remaining impacts are estimates from the results of the Delphi study.
Table 3.5: Society, safety & economy impact definitions
Impact Definition Methodology
Freight vehicle
operating cost*
Direct outlays for operating a vehicle per kilometre of travel
(€/km) Delphi
Parking space Required parking space in the city centre per person
(m2/person) Delphi
Public health Subjective rating of public health state, related to transport
(10 points Likert scale) Delphi
Road safety Number of predicted crashes per vehicle-kilometre driven
Microsimulation**
* In section 3.4 freight vehicle costs have also been estimated using operations research methodology
** Post processing done with SSAM + Tarko (2018) crash prediction method
As we have explained earlier, society, safety and economy are highly interrelated policy
areas (see Section 2.2). For example, both road safety and public health have an important
social dimension as well as a well-established economic dimension. Economic indicators
such as vehicle operating costs and parking space have a direct economic value but will
also have an impact on access to mobility and therefore on various social and cultural
activities, and collective well-being (and will also have effects that extend to other
domains).

LEVITATE | Deliverable D7.5 | WP7 | Final
31
Table 3.6: Estimated impacts of automated freight transport services on society and economy, measured in terms
of percentage change with respect to the Baseline 100-0-0 scenario.
Market penetration rate: AVs in Background vehicle
fleet
(Human-driven vehicle - 1st Generation AV - 2nd Generation
AV)
100-
0-0
80-
20-0
60-
40-0
40-
40-20
20-
40-40
0-40-
60
0-20-
80
0-0-
100
Impact
Sub-use Case
%
%
%
%
%
%
%
%
Method
Vehicle
operating
cost
(freight)
Baseline 0,0 7,5 -0,7 -4,0 -10,0 -10,4 -10,4 -10,4
Delphi
Automated delivery 0,0 2,4 4,4 1,7 -3,7 -7,9 -7,9 -7,9
Automated consolidation 0,0 2,4 0,9 -7,3 -13,7 -17,4 -17,4 -17,4
Hub-to-hub 0,0 -3,3 -6,2 -6,2 -12,3 -15,5 -15,5 -15,5
Parking
space
required
Baseline 0,0 -1,4 -1,3 -5,0 -11,5 -11,6 -11,6 -11,6
Delphi
Automated delivery 0,0 -7,9 -4,6 -6,8 -5,1 -4,0 -4,0 -4,0
Automated consolidation 0,0 -4,3 -2,8 -2,8 -4,2 -3,8 -3,8 -3,8
Hub-to-hub 0,0 -1,6 -1,5 -3,3 0,0 1,4 1,4 1,4
Public
health
Baseline 0,0 -5,3 -2,1 0,0 5,2 4,0 4,0 4,0
Delphi
Automated delivery 0,0 2,9 4,7 8,8 8,8 8,4 8,4 8,4
Automated consolidation 0,0 6,0 7,7 9,4 14,4 18,5 18,5 18,5
Hub-to-hub 0,0 6,6 10,0 12,0 17,8 15,7 15,7 15,7
Road
safety:
crash rate
Baseline;
manual delivery
0,0 7,5 14,0 16,1 4,3 -19,5 -37,0 -48,7
Micro-
simulation
(Vienna)
Automated delivery -2,6 -4,2 10,2 4,9 4,6 -19,8 -41,0 -50,2
Automated consolidation -2,6 -4,2 10,2 4,9 4,6 -19,8 -41,0 -50,2
Baseline;
no transfer hub
0,0 11,5 23,1 11,5 0,0 -11,5 -34,6 -61,5
Hub-to-hub;
with transfer hub
0,0 7,7 11,5 -3,8 -11,5 -23,1 -46,2 -61,5
Delphi results
The Delphi consultation was used to obtain results for expected developments in the area
of vehicle operating costs, parking space and public health (See Table 3.6). As explained
before, for each impact the percentage effects reported are in respect to the Baseline
100-0-0. The difference between the baseline effect and the specific SUC under
consideration, and given the penetration rate, is the effect of the SUC itself. For most
impacts in this table a decrease (denoted by a “-“) implies a positive effect. However, for
public health the opposite holds true.
According to the Delphi consultations, vehicle operating costs will be reduced, especially
when human-driven vehicles are reduced to 20% or less of the entire vehicle fleet. Under
baseline conditions, the results show that vehicle operating costs of freight transport will
be reduced by about 10% when the entire vehicle fleet is automated. The automated
consolidation SUC is associated with larger reductions in vehicle operating costs than
baseline increase in automation alone; cost reductions between 13% and 17% are
expected once human-driven vehicles take up 20% or less of the vehicle fleet. Hub-to-hub
is also associated with stronger reductions than the baseline, 12 to 15% reduction once

LEVITATE | Deliverable D7.5 | WP7 | Final
32
human-driven vehicles represent a fifth or less of the fleet. The automated delivery is
associated with slightly less of a reduction than the baseline.
According to expert consultation, in the baseline scenario parking space requirements will
be reduced by nearly 12% once human-driven vehicles are reduced to 20% or lower.
Looking at the impacts estimated for the three freight sub-use cases, all three are
associated with a reduction of required parking space. However, in all cases the impact is
smaller than in the baseline, implying that the automated delivery van SUCs will require
more parking space than the scenario with automation but without a fully-automated,
unstaffed delivery van system. The hub-to-hub scenario with 100% CAV penetration is
even predicted to require slightly more parking space than the baseline situation with only
human-driven vehicles.
Regarding public health, a negative estimate implies a decline in public health. In the
expected baseline development, a small deterioration in public health is expected when
the presence of CAVs is still low (20% to 40% penetration) followed by a small
improvement in public health (4% to 5%) as the penetration of second generation CAVs
increases. The automated consolidation and hub-to-hub freight transport SUCs are
anticipated to generate substantial added improvements in public health (8% to 10%) once
human-driven vehicles are below 60%, and to further improve (by up to 18%) once the
entire vehicle fleet is automated. The automated delivery sub-use case is expected to
generate a more modest improvement in public health, starting at 3% when human-driven
vehicles are still at 80% and rising to above 8% once automated vehicles are in the
majority.
Microsimulation results
Microsimulation was used to study the expected impacts of the freight SUCs on road safety
for all vehicles (freight and passenger) in the Vienna city network. The estimated crash
rates (predicted crash rates per vehicle kilometer) are affected by behavioural parameters
determined for the microsimulation, which affect how human-driven or automated vehicles
drive in the network. For freight vehicles, less knowledge was available on the behavioural
parameters of future automated freight vehicles; therefore, some parameters assumed the
values of 1st generation CAVs and others were based on assumptions, leaving some
uncertainty. For this reason, road safety results within LEVITATE have been estimated both
including (Work Package 7) and excluding (Work Package 5 & 6) freight vehicles; this
revealed that the inclusion of freight vehicles led to higher crash rates (1-28% higher) at
most penetration rates, depending on the network (Weijermars et al., 2021). Nevertheless,
the road safety impacts of increasing automation in a mix of both passenger and freight
vehicles are an interesting development, especially for these sub-use cases where the
deployment of automated freight vehicles is varied.
Taking into account that the inclusion of automated freight vehicles can somewhat inflate
the estimated crash rates, the results for the city of Vienna show:
• In the baseline situation, road safety is predicted to take a turn for the worse when the
first generation of automated vehicles is introduced and there is a lot of interaction
between human-driven vehicles and (two types of) automated vehicles. Due to different
driving styles of human drivers and automated vehicles, some extra risks in mixed traffic
are an expected development during the transition from human to non-human-driven
vehicles.
• Road safety improves once no human-driven vehicles are left in the simulation (from a
60% penetration of second-generation vehicles and above), resulting in roughly half as

LEVITATE | Deliverable D7.5 | WP7 | Final
33
many crashes per vehicle-kilometre when the entire vehicle fleet is made up of 2nd
generation automated vehicles.
• Compared to the baseline, the introduction of both automated delivery and automated
consolidation shows marginal additional benefits for road safety. Especially at lower
penetration rates of automated vehicles in the entire fleet, the addition of automated
delivery reduces the crash rate. This is likely due to the higher total amount of
automation in the network as delivery vehicles are no longer partially human-driven.
The difference with the baseline becomes minimal at higher penetration rates, when the
entire vehicle fleet is already automated.
• Compared to the no hub-to-hub baseline, the hub-to-hub SUC is also associated with
improved road safety performance. The added benefit is particularly large in the middle
of the transition phase, before all vehicles have become 2nd generation CAVs
In summary, the road safety results of this network, including freight transport, suggest
that mixed human-driven and automated traffic can bring about some extra safety risks.
However, once the entire vehicle fleet is automated, substantial safety improvements are
expected. While the automated freight SUCs showed some marginal additional
improvements compared to the baseline, general levels of vehicle automation appear to
be the largest driver of changes in road safety.
3.4 Additional impacts economy: annual fleet costs
Using data on delivery trips in Vienna and operations research methods, Hu et al. (2021b)
estimated the impacts of automated delivery and automated consolidation on mileage,
annual fleet costs and freight transport costs. For the automated delivery and automated
consolidation SUCs, the primary influencing factors for the economic impacts are the fleet
size and the driven km. These factors are fundamental for freight operations since other
impact indicators are directly based on them (Hu et al., 2021a).
Based on delivery trips in Vienna estimated by the operations research (Hu et al., 2021a),
Table 3.7 compares the delivery variants with respect to their fleet composition and driven
kilometres per day. The columns show the number of delivery trips, fleet size, average
number of stops (parking operations) per trip, average trip length and mileage of all delivery
trips. This is followed by the mileage of the consolidation trips by trucks (i.e., trips for
delivering to parcels to the city-hubs), and finally the total mileage of all vehicles.

LEVITATE | Deliverable D7.5 | WP7 | Final
34
Table 3.7: Results for automated delivery and automated consolidation.
SUC Scenario
Delivery via van / robo-van Consolidation
trips by trucks
Total
driven km
No
of
trips
Fleet
size
Stops
per
trip
Trip
length (km) Total driven km Driven km
No consolidation
Manual delivery 1799 1799 42,3 44,7 80.389 km - 80.389 km
Semi-automated delivery 1440 1440 46,5 49,2 70. 805 km - 70.805 km
Automated delivery 2692 898 28,9 39,4 10. 6177 km - 106.177 km
Automated night delivery
2692
1795
28,9
39,4
10.6177 km
-
106.177 km
Consolidated delivery
Manual delivery with city-
hubs
1806 1806 17,8 13,7 24.675 km 10.445 km 35.120 km
Automated delivery with
city-hubs 2716 906 12,5 11,9 32.347 km 10.445 km 42.792 km
It can be seen that on the one hand, the mileage is significantly shortened by the
consolidated delivery via the centrally located city-hubs. On the other hand, mileage
increases due to the lower capacities of the robo-vans for automated delivery. However,
with automated delivery using smaller vehicles more delivery shifts (three as opposed to
2 in the day and 2 as opposed to 1 at night) can be introduced requiring fewer vehicles in
the fleet at any given time. This has the potential to reduce the operating costs
significantly.
For assessing the impact on vehicle operating cost, Hu et al. (2021a) made a number of
assumptions (fully described in Appendix D), and these are summarized in Table 3.8.
Table 3.8: Vehicle operating costs per delivery vehicle per year (EUR).
manual semi auto full auto (robo-van)
Vehicle 3.000 5.000 7.000
Insurance, maintenance, fuel 5.000 3.000 3.000
Driver / delivery personnel 45.500 45.500 0
Delivery robot fleet 0 0 12.000
Monitoring personnel 0 0 12.000
Annual costs per vehicle 53.500 53.500 34.000
Using these numbers, Hu et al. (2021a) applied them on the results shown in Table 3.7 to
obtain the impacts for the annual fleet cost (expressed in Million EUR), vehicle operating
costs (EUR/km) and freight transport cost (EUR / tonne-km). For the freight transport cost,
they assumed an average parcel weight of 1,37kg per parcel (Wirtschaftskammer Wien,
2020). Table 3.9 and shows the results obtained for the modelled Vienna network based
on the current volume of packages delivered.

LEVITATE | Deliverable D7.5 | WP7 | Final
35
Table 3.9: Vehicle operating cost and freight transport cost given 5 freight CAV implementation scenarios.
SUC Scenario Fleet size Driven km Annual fleet cost
(Million EUR)
Vehicle operating
cost
(EUR / km)
Freight transport cost
(EUR / tonne-km)
Manual delivery 1799 80.389 km 96,2 3,9 18,8
Semi-automated delivery 1440 70.805 km 79,9 3,6 14,8
Automated delivery 898 106.177 km 30,5 0,9 6,8
Automated night delivery 1795 106.177 km 61,0 1,9 13,5
Manual delivery with city-hubs 1806 24.675 km 96,6 12,6 61,5
Automated delivery with city-hubs 906 32.347 km 30,8 3,1 22,4
Based on the data in Table 3.9, Figure 3.1 illustrates the annual fleet costs for the different
SUC considered for freight delivery.
Figure 3.1: Annual fleet cost (Million EUR).
This shows that manual freight delivery has significantly higher costs (96,2 million) and
completely automated delivery significantly lower costs (30,5 million). Automated delivery
with city-hubs can reduce annual fleet costs by up to 68%.
0
20
40
60
80
100
120
Manual delivery Semi-automated
delivery
Automated
delivery
Automated night
delivery
Manual delivery
with city-hubs
Automated
delivery with city-
hubs
Million EUR

LEVITATE | Deliverable D7.5 | WP7 | Final
36
3.5 Impacts of truck platooning
Within WP7 it was decided to include a sub-use case dealing with automated freight
vehicles moving in platoons over bridges, this to establish the impacts on bridge loading
and on internal bridge forces, particularly bridges with long spans and under high loads
resulting from closely spaced platoons of fully loaded heavy goods vehicles. Since this SUC
is unique and does not quite fit the methodology adopted to synthesize the results of the
research on automated freight in Levitate, the results are separately presented and briefly
discussed.
The change in traffic composition due to platoons is expected to lead to higher internal
bridge forces. The baseline scenario include all traffic cases without platooning. As
expected, the traffic cases without traffic congestion produced quite low internal bridge
forces and therefore have limited impact on current bridge structures.
In this section we summarise – in a mostly non-technical way - the impacts of heavy traffic
on bridge internal forces (baseline) and the impacts of introducing truck platoons. The full
technical description of model and assumptions is given in Hu et al. (2021b).
The following general effects were reported under the baseline condition (Hu et al., 2021b):
• The traffic simulations without congestion produced relatively low internal bridge
forces and the exceedance probabilities for bending moment and shear force
remained far below the critical code level
• The congestion events introduced a significant increase of bridge internal forces,
and especially on bridges with longer spans.
Impacts of truck (HGV) platooning
In simulating the effects of platooning on bridge loading a simply-supported single-span
bridges was considered. The bridge was modelled as a single beam supported at both ends,
with free rotation (also see Appendix D for the bridges that were considered).
The main impacts of platooning on bridges can be summarised as follows (Hu et al.,
2021b):
• After the platoons were introduced into to the simulated traffic, the bridge internal
forces increased significantly in bridges with longer spans. Figure 3.2 shows the
increase of probabilities of exceeding the load effects of LM1 load model. The red
curve is the traffic case with platooning (baseline), and four curves representing
results with 20%, 40%, 60% and 80% platooning penetration rate are shown.
• Even at a low penetration rate of 20% truck platooning already shows a large
increase of exceedance probabilities. The increase of exceedance probabilities for
increasing bridge spans does not differ for penetration rates of 20%, 40%, 60% or
80% (see Figure 3.2)
• Starting from a bridge span length of 60 m, the “code level” of exceedance
probabilities is exceeded. In that case , the structural safety of the affected bridges
would be compromised, assuming that their load-carrying capacity is on par with
the Eurocode requirements, i.e., without additional reserves.

LEVITATE | Deliverable D7.5 | WP7 | Final
37
Figure 3.2: Bending moment exceedance probabilities for traffic mix A, constant congestion distances (C5),
Pcong=0.99, Pflow=0.999, traffic volume of 39000 vehicles/day and different platooning penetration rates.
• The largest effect of platooning is observed for the criteria of braking forces. The
extremely short distances within a platoon and the sequence of truck platoons lead
to high forces in case of braking. This is because the trucks in a platoon need to
brake with almost the same deceleration, so that all platoon vehicles decelerate
with approx. 5 m/s². For bridges above 80 m length, the braking force is at least
the double that of the baseline scenario.
• Forcing an increased inter-vehicle distance by intelligent access control will not
diminish the ecological and economic benefits of truck platoons.
• It can be concluded that there may be a need to strengthen existing bridges with
= 1 and with a span length of 55 m for bending moment and 60m for shear force
ULS; for existing bridges with resistance at resistance level of = 0.8,
strengthening needs would arise sooner, starting from bridge spans of 40 m.

LEVITATE | Deliverable D7.5 | WP7 | Final
38
4 Discussion
This chapter summarises the main findings on the expected impacts after
introducing CCAMs and cooperative, connected and automated freight transport.
The strengths and limitations of the theoretical and empirical work underlying
these impacts are discussed, and policy considerations in the broader context of
the transition to smart urban mobility are presented.
4.1 Main findings
Below we summarise main findings for WP7:
• The increasing participation of connected and automated vehicles in the urban city area
is estimated to have positive impacts on the city environment (less emissions, higher
energy efficiency), and city society and economy (less parking space, lower freight
vehicle operating cost) and on city mobility (less congestion).
• The road safety impacts estimated for the baseline condition in WP7 differ from the
baselines in WP5 and 6 as a result of different road networks being applied and the
inclusion of freight vehicles in the estimation. Contrary to the results in WP5 and 6, WP7
results reveal that in the baseline the increasing presence of automated vehicles in the
city is estimated to have a temporary negative impact on road safety when
penetration rates of automated vehicles are low. Positive road safety impacts of the
increasing presence of automated vehicles are estimated once human-driven vehicles
are replaced and second-generation automated vehicles have reached penetration
levels above 60% of the city vehicle fleet. Because less data was available on the driving
behaviour of autonomous freight vehicles, assumptions needed to be made in the
behavioural parameters for autonomous freight vehicles. This led to higher crash rate
estimations when freight vehicles were included. More broadly within LEVITATE, most
estimates point to a large reduction in crashes with the introduction of automated
vehicles including a small reduction at low penetration rates.
• The increasing presence of automated vehicles in the city is estimated to have negative
impact on public health when traditional (human-driven) vehicles still make up the
majority of vehicles (80%-60%). The sub-use cases of automated delivery, hub-to-hub
and especially automated consolidation will positively impact public health. This positive
expectation is likely based on the expected additional benefits of these sub-use cases
for both road safety and emissions.
• The automated delivery SUC is associated with additional benefits for energy
efficiency, CO2 emissions, road safety, public health and vehicle operating costs.
• The automated consolidation sub-use case is associated with additional benefits for
energy efficiency, CO2 emissions, travel time, public health, road safety, and vehicle
operating costs.
• The hub-to-hub use case is expected to deliver additional benefits for energy efficiency,
CO2 emissions, congestion, travel time, road safety, public health, and freight vehicle
operating costs.

LEVITATE | Deliverable D7.5 | WP7 | Final
39
• Given the higher-level CAV penetration rates (above 80%) all the automated freight
delivery SUCs require more parking space than the baseline without automated
delivery. The Hub-to-Hub SUC even requires more parking space at 100% CAV
penetration compared to the current situation (with 100% human-driven vehicles).
• The largest effect of Truck Platooning is observed for the criteria of braking forces.
For bridges above 80 m length, the braking force is at least the double of the baseline
scenario.
• The need for strengthening structural resistance of bridges arises for existing
bridges with = 1 starting from span length of 55 m for bending moment and 60 m for
shear force ULS; for existing bridges with resistance at resistance level of = 0.8,
strengthening needs would arise sooner – starting from bridge spans of 40 m.
• For bridge strengthening, a model and guideline for estimating the costs in relation to
the initial construction costs have been developed (D7.3).
• As an alternative to strengthening bridges, intelligent access control can be used to
arrange the increase the headways between HGV in a platoon in order to meet the code
level and prevent potential failures. The preferred increases of inter-vehicle distances
have been calculated for different bridges and circumstances and are reported in
Levitate D7.3 (Hu et al., 2021b)
• Forcing an increased inter-vehicle distance by intelligent access control will not diminish
the ecological and economic benefits of truck platoons.
• Based on delivery trips in Vienna estimated by the operations research, the mileage of
freight transport was substantially shortened by consolidated delivery via the centrally
located city-hubs.
• Using delivery trips in Vienna estimated by the operations research, compared to
manual freight delivery (96.2 million), completely automated delivery (30,5 million) and
automated delivery with city-hubs (30,8 million) will have much lower annual fleet
costs (-68%).
Mobility
The results confirm the hypothesis that freight traffic will only have a small effect on the
overall congestion in the urban environment since its share of the traffic volume is
relatively low. For the SUCs automated urban delivery and automated consolidation, the
impact on congestion caused by the changes in the delivery procedure was minor, i.e., not
statistically significant, despite the shift from fewer larger vehicles to smaller automated
vehicles. There are however other benefits to consider; an obvious advantage of
automated freight transport is the ability to utilise the off-peak hours and the night-time,
allowing passenger transport more space during the peak hours and thereby reducing the
demand for limited road capacity. This potential benefit is supported by findings in Jennings
et al. (2019) and Figliozzi et al. (2020), where the on-road travel could be significantly
reduced in scenarios where the service areas are near to the depot.

LEVITATE | Deliverable D7.5 | WP7 | Final
40
Wider impacts on Society, Safety and the Economy
The estimates for the wider impact on road safety were less positive than perhaps may
have been expected from the results in WP5, WP6, and LEVITATE’s road safety working
paper (see Weijermars et al., 2021). As mentioned before, these differences are related to
two factors:
1. Differences in the network characteristics (eg. road design, fleet composition)
between Vienna (WP7), Athens (WP5), Manchester (WP6), Santander (WP6) and
Leicester (WP6).
2. The inclusion of freight vehicles in the estimation, about which less is known
regarding their behaviour as automation increases. Some of the parameters
dictating their driving style assumed the values of 1st generation CAVs and others
were based on assumptions. In other networks, the inclusion of freight vehicles in
the crash rate estimations lead to 1-28% more crashes per vehicle-kilometer,
depending on the network and penetration rate (Weijermars et al., 2021)
In WP7 baseline conditions it was estimated that road safety was negatively affected when
first generation CAVs are introduced. The low penetration levels of CAVs result in
unfavourable interactions between human-driven vehicles and CAVs, a phenomenon that
is supported in some literature which has found that during the early transition phase to a
fully automated traffic system crash rates may well increase at first. In earlier simulation
studies it has been found that the introduction of automated vehicles in mixed traffic
conditions may increase risk (Shi et al., 2020), especially when the market penetration of
these vehicles is lower than 40% compared to traffic flow consisting of human drivers only
(Yu et al., 2019). The Levitate WP7 results show that road safety steadily and significantly
improves once human-driven vehicles are reduced and finally omitted from the
simulations. The automated freight SUCs of automated delivery, consolidation, and hub-
to-hub improved traffic safety further compared to the baseline development.
Within LEVITATE more positive estimates for road safety were derived in WP5 (urban
transport) and WP6 (passenger cars), where crash rates also decreased slightly at low
penetration rates. At low penetration rates, the balance between the safety of automated
vehicles (which are expected to crash less often than human-driven vehicles) and the
potential risks of mixed traffic (when human-driven/less advanced automated vehicles are
still on the road) is a point of attention for further research.
As has been reported in a special working paper on road safety impacts within LEVITATE
(Weijermars, 2021), the estimated road safety impacts differ between city network, and
differ dependent upon the presence or absence of freight transport in simulation models.
The presence or absence of freight vehicles strongly influences model crash results
(Weijermars et al., 2021). The roads safety results in WP7 are solely based on the Vienna
network, and include freight transport vehicles in the simulation models, whereas freight
vehicles were excluded in the models used in LEVITATE Work packages 5 (passenger cars)
and 6 (urban transport). As Weijermars et al. (2021) noted, there can be some doubt as
to whether there is sufficient knowledge about automation of freight transport to enable
valid simulation of this category of vehicles. Despite this concern and in the absence of
immediate alternatives, it was decided to include freight vehicles (light & heavy goods
vehicles) in the microsimulation models of WP7 which focusses on the effects of changes
to freight transport. Since the model outputs allow for comparisons of relative differences
between baseline and (SUC) penetration scenarios that are all based on the same traffic
input parameters, it was felt that this was preferable to having no estimates at all.

LEVITATE | Deliverable D7.5 | WP7 | Final
41
Important however, is that the absolute values of the tested indicators (travel time, delay
etc) for any given WP7 scenario can only be seen as indicative and these values have not
been validated nor calibrated and must be treated with caution.
The wider impacts on parking space, energy efficiency and public health were based on a
two-round Delphi panel process. While the experts expected energy efficiency and public
health to improve with the increasing AV penetration rate, the situation on parking space
was mixed. From the Delphi results, the baseline scenario would decrease the demand for
parking space with more AVs on the street. However, the automated freight transport
measures such as automated delivery or hub-to-hub automated transport are expected to
require more parking space than the baseline. As observed by Hu et al. (2021c) this finding
is rather surprising since freight consolidation and night-time delivery are expected to
increase the efficiency and remove the redundancy of the freight system (Hu et al., 2021c).
There are several reasons for the positive expert expectations about improvement in public
health though automated freight transport. First of all, CCAM in general has the potential
to improve public health if proper policies and regulatory frameworks are implemented.
AVs are likely to improve road safety and may help reshape cities to promote healthy urban
environments (Rojas-Rueda et al. 2020). In addition, the local emissions caused by freight
transport will be reduced to zero due to the assumption that AVs will be driven by fossil-
free fuels. This might not be a direct contribution of vehicle automation since manual
electric freight vehicles would have the same effect. However, the significant reduction of
fleet operation costs by CCAM as shown by Hu et al. (2021) will accelerate the transition
towards emission-free automated freight transport and logistics.
A particular SUC that was dealt with in WP7 was the effect of truck platooning on urban
(highway) bridges. This is a special SUC in that the assessment methods and the obtained
impacts are different from the other SUCs. This SUC is for study purposes and will not be
included in the PST estimator, but nevertheless it has an eminent importance. For truck
platooning, there already exist a good amount of scientific work, but the impacts on the
bridge infrastructure is under-researched. Although the potential damage is not a short-
term effect and the probability of a potential bridge failure resulting from truck platooning
is not high, we have to be aware that if a failure occurs, the consequences are disastrous
(c.f. Caprioglio bridge collapse, 2020). Therefore, two measures for dealing with the
upcoming truck platoons enabled by CCAM are discussed. The results indicate that
intelligent access control and an associated increase of the headways between trucks will
be required to meet the EU code levels on certain bridge spans. For bridge strengthening,
a model and guideline for estimating the costs in relation to the initial construction costs
are given. Note that the economic and environmental impacts by truck platooning such as
fuel savings are well-researched topics (Humphreys et al. 2016) and were not dealt with
in Levitate (Hu et al., 2021b).

LEVITATE | Deliverable D7.5 | WP7 | Final
42
4.2 Strengths and Limitations
The Levitate project has strengths and limitations. A potential strength of the Levitate
project is that the future development of urban smart cities policy interventions and policy
impacts have been selected by a diverse group of stakeholders. The best available methods
– microsimulation, mesosimulation, Delphi, and other methods – were used to study and
quantify expected impacts of mobility interventions to support connected and automated
vehicles and sustainable city goals and to deliver input for a practical Policy Support Tool
for city policy makers. The knowledge of Levitate is above all intended to contribute to
policy-making for smart city traffic development.
Below we describe some limitations of the Levitate studies, first some general limitations
or difficulties concerning predicting future trends and second some limitations that are
more specifically related to specific methods used.
Limitations in predicting future trends
Research evidence is not available for all potential impacts of connected and automated
vehicles identified in Levitate. The Levitate research can inform policy makers about a
number of potential impacts of connected and automated vehicles. Specific potential
impacts of connected and automated vehicles that are difficult to predict with any
confidence are the following (Elvik et al., 2020):
• Whether there will be a widespread transition from individual to shared mobility. There
is no consensus on whether individual use of motor vehicles will continue at present
levels or be replaced by various forms of shared mobility. This will largely be impacted
by the policy measures of the cities and national authorities. Therefore, the LEVITATE
project aims to support the authorities finding the most beneficial policies on the way
towards an automated transport system.
• It is not clear what type of propulsion energy connected and automated vehicles will
use. Some researchers expect the introduction of connected and automated vehicles to
be associated with a transition to electric propulsion.
• Connected and automated vehicles are vulnerable to cyber-attacks. However, the risk
of such attacks cannot be quantified, and it would go beyond the scope of LEVITATE.
Only potential scenarios can be described.
• The costs of connected and automated vehicles are highly uncertain. It is not clear that
connected and automated vehicles will be as affordable as current motor vehicles. The
costs of automation technology may influence the level of inequality in access to
transport. However, there is evidence that it will result in a significant cost reduction for
operators once drivers are no longer needed.
• Behavioural adaptation to connected and automated vehicles, in particular during the
transition period before full market penetration. While some studies suggest various
forms of behavioural adaptation, predicting its form and impacts is impossible or
speculative.
• Changes in employment are difficult to predict. While full automation will eliminate the
need for drivers, other potential changes in employment are less known.
Given the many uncertainties in prediction it is obvious that any predicted values are
associated with large uncertainty. It was decided not to estimate confidence intervals for
the results in Appendix A based on the standard error derived from repeated trail runs of
models since these intervals would be broad and non-informative. Also, the estimation of

LEVITATE | Deliverable D7.5 | WP7 | Final
43
these intervals would tend to be biased in itself since the input variables and assumptions
in the models are very likely much stronger determinants of predicted values than the
variability in sample runs.
Specific method-related limitations
There are some remarks to be made about the possible limitations or nuances of modelling
results in WP7:
• The impacts estimated using AIMSUM microsimulation for WP7 are based on the
road network of Vienna. The simulation modelled both passenger vehicles and and
freight vehicles (light goods vehicles & heavy goods vehicles). The parameters for
vehicle performance and driver behaviour are derived, where possible, from
literature. The parameters for automated freight vehicles, however, lacked
sufficient basis in the literature and are therefore largely left at their human-driven
levels with a few exceptions. For this reason, the freight vehicles in the simulation
may behave more similarity to human-driven freight vehicles than would be
expected of truly automated freight vehicles. As discussed earlier, this limitation is
particularly relevant for the road safety estimations.
• The results of the microsimulation models and bridge modelling depend upon
specific assumptions. For example, in LEVITATE the simulation models used
examined two CAV driving style profiles were assumed (first vs. second
generation); future work may extend the number of driving style profiles.
• The assumptions on CAV parameters and their values were based on a
comprehensive literature review, including both empirical and simulation-based
studies as well as discussions in meetings with experts, conducted as part of
LEVITATE project.
• At the time of modelling the AIMSUN micro-simulation software used in LEVITATE
was limited to the simulation of motor vehicles on the road network, so pedestrians
and cyclists are not included in the simulations. Road safety impacts were however
estimated separately for VRUs in a Baseline development.
• In LEVITATE it was assumed that all freight AVs will be electric and therefore
emission-free while the manual freight vehicles use internal combustion engines
fuelled by diesel, which is the standard at the moment
• The microsimulation modelling in WP7 was based partly on the Vienna city network,
which means that results are most transferable to those urban conglomerates which
have structural and dynamic characteristics that are similar to these.
• The micro-simulation was only applied on the model of a small network area, a full
city model could be used in future work to verify the upscaled results.
• In the study on bridge platooning impacts the results were evaluated for simply-
supported single-span bridges with dimensions that were regarded as typical for
particular bridge types. These results are intended to provide an indication only.
Results for actual bridge structures may deviate from these, as each bridge is
different.

LEVITATE | Deliverable D7.5 | WP7 | Final
44
4.3 Policy considerations and discussion
The sub-use cases or policy interventions studied in the LEVITATE project are part of a
wider transition to smart mobility and smart cities. In this section we will reflect on a
number of relevant broader policy issues surrounding the introduction of automated
transport systems in urban areas since it is clear that these wider developments towards
smart mobility will also affect the specific use cases. The text in Section 4.3 is identical to
that described in D5.5 and 6.5 and the reason for that is that these issues are not only
relevant to public and private transport but also to freight transport.
Planning and governance of automated mobility in urban environments
Implementing new forms of connected and automated mobility is a highly complex process,
particularly in the urban environment. Many different actors in city governance, industry
and the general population will need to come together to deal with these challenges.
Although there may be a strong push from industry to implement new smart mobility
services, there are still many uncertainties that lie beyond the powers or competence of
any one single stakeholder to fully control or address. Adequate legislation and technical
standards are expected to lag behind CAV deployment trials and pilots (in other words,
technology develops faster and legislation and standards etc. have to follow). It is
important to anticipate these developments and to start the processes necessary for
adopting standards and legislation that will be necessary to regulate large scale CAV
deployment. An example we can learn from is the advent of the motor car. This occurred
in a largely unregulated transport environment, and which introduced many negative
impacts which in time, and to this day, needed mitigation. Safe systems are about
prevention and this pleads for a pro-active approach, also with respect to standards,
legislation and regulation.
There is enthusiasm about the transition towards smart mobility, but not surprisingly
opinions vary. Fraedrich et al (2018) carried out a survey among city planners in 24
German cities. Half of the respondents believed that shared autonomous vehicles could
positively contribute to urban planning objectives, but only 10% reported that private
autonomous cars could contribute to those objectives. According to the respondents,
implementation of automated vehicles would require preparatory action in the fields of
transportation planning, traffic control, road infrastructure, urban planning, citizen
participation, test fields and data standards and requirements. Additional interviews with
city planning experts led to four major insights namely:
• Cities themselves are a major driving force;
• for city renewal or redevelopment, public transport is a major goal;
• there is concern about the possibility of an increase of private car use in cities;
• city goals are not always directly aligned with other stakeholders seeking to push
automated vehicle technology.
In the USA, McAslan et al. (2021) have looked at plans for autonomous vehicles amongst
Metropolitan Planning Organizations (MPOs). One key area that requires attention is public
engagement in the management of emerging technologies. This element seems critical to
advancing CAVs in a way that addresses issues of equity and mobility justice (and others).
Equity, accessibility, and other such goals are often promoted by industry, but ultimately
the realisation of these is ultimately a planning and policy decision (McAslan et al., 2021).
Several of the studied Regional Transportation Plans had policies to address issues of
equity and accessibility. However, MPOs need to engage stakeholders (e.g., the public,
industry, etc.) and make issues such as equity or other valued public goals a priority. Left

LEVITATE | Deliverable D7.5 | WP7 | Final
45
to market forces alone, it is likely that these potential benefits will not be realised and
could even worsen (McAslan et al., 2021).
Many authors have stressed that industry and economy forces tend to push towards
implementation of automated driving, but this technology push should be balanced by an
equally strong orientation on the social-ethical (or the non-technical) dimension of the new
technology. In other words, how it is governed, how it is perceived by citizens from various
social strata, whether it complies with ethical guidelines and whether it really provides the
expected benefits for the city (Fraedrich et al., 2019; McAslan et al, 2021; Habibzadeh et
al, 2019, Milakis & Muller, 2021). In recognition of this, authors have suggested that new
types of national, local or city governance (or management) are needed to steer the
transition towards automated mobility in a responsible way (e.g., Aoyama & Leon, 2021;
McAslan et al., 2021; Milakis & Muller, 2021).
Milakis & Muller (2021) suggest that policy makers need new tools for long term planning
to accommodate uncertain urban futures. They argue in favour of new participative
anticipatory governance instead of traditional governance which is typically supported by
forward looking exploratory deployment scenarios with short term implications. They
suggest a research agenda that is more oriented on citizens than consumers, more focused
on long term than only short term and more based on citizen participation than traditional
short-sighted scenario analysis. Their emphasis on normative scenario analysis (i.e.,
backcasting) aligns well with the LEVITATE project.
McAslan et al. (2021) argue for anticipatory governance looking at future scenarios, using
flexible planning mechanisms, and where monitoring and learning are built in the planning
process, and the public is actively engaged.
Aoyama & Leon (2021) conclude that cities are part of multi-scalar governance frame-
works where new rules, regulations, strategies, and standards are negotiated and enacted.
They identified four key roles for cities in the governance of the emerging autonomous
vehicle economy: regulator, promoter, mediator, and data-catalyst. They cite the example
of the city of Pittsburgh which, in recent years, has shifted away from a role of being
promotor to a new role of being mediator. The initial emphasis of the city government on
the promotion of the autonomous vehicle economy has decreased and has given way to
an acknowledgment of the need to build more equitable relationships between various
stakeholders in the city area. Another example of a city taking up a different governance
role is Boston. In recent years, Boston's city government has become very active as a
data- catalyst; the city takes an active approach in exploring partnerships on data
collection and developing a shared research agenda that includes not only vehicle testing,
but also business model exploration, experiments with connected transportation
infrastructure, and research on autonomous mobility and its implications on Boston's
workforce.
Planning for future urban city mobility: four types of readiness
At the city level, policy makers and planners face four major areas where preparation is
needed to enable future use of CAVs (Alawadhi et al., 2020).
1. Infrastructure readiness - the road infrastructure needs to be adapted in order to
facilitate proper functioning of automated vehicle systems.
2. Digital readiness - the digital infrastructure needs to be set in place, including a
framework, technical standards and procedures for cybersecurity and data privacy.

LEVITATE | Deliverable D7.5 | WP7 | Final
46
3. Legal readiness - there needs to be clarity about how legal responsibilities and
liabilities may be solved and how problems in this area may be avoided.
4. Social readiness - the social understanding, acceptance and approval of the new
forms of mobility amongst various citizen groups and stakeholders in the urban
area seems critical.
Road infrastructure readiness
Road infrastructure will have to be adapted in order to be ready, readable, and cooperative
in all situations and weather conditions (Gruyer, 2021). CAVs require highly visible road
edges, curves, speed limit and other signage (Liu et al.,2019). For the EU it is important
to have uniform road markings. The roadside digital infrastructure needs to meet various
connectivity requirements.
The lack of sufficiently visible road markings is at the moment an obstacle for some
manufacturers for the reliable functioning of autonomous vehicles (Rendant & Geelen,
2020). The reliability of systems such as ISA and LDWA, are dependent on these for reliable
functioning. Other infrastructural aspects have to do with harmonisation of the road
infrastructure (colour, reflective materials, etc.). In Europe this will likely have a positive
influence on the roll-out of CAVs (Rendant & Geelen, 2020). The development of camera
technology and image processing algorithms is so fast that future systems will likely be
able to deal with lower quality markings. Upgrading road markings to support self-driving
vehicles may not be necessary (Rendant & Geelen, 2020).
In the Inframix project, so-called ISAD levels (“Infrastructure Support Levels for
Automated Driving”) were developed in which an impetus is given to define the minimum
infrastructure (physical and digital) required to enable certain self-driving functions. Such
an approach makes sense to clarify what level of automation is possible on a given road
section (Rendant & Geelen, 2020).
A special infrastructure topic concerns the extra burden on bridges caused by automated
truck platooning. In order to minimize the failure probability of structural bridge integrity,
medium-term measures such as structural strengthening and intelligent access control
should be considered. Within LEVITATE a start has been made to develop methods to
assess the impacts of these measures (Hu et al. 2021b).
Readiness to address cybersecurity and data privacy concerns
The successful operation of CAVs and their expected impact depend significantly on their
management and addressing risks associated with them (Lim & Taeihagh, 2018). Two of
these risks are privacy and cybersecurity. The ability of CAVs to store and communicate
personal data may conflict with data privacy laws. Cybersecurity is at stake when
communication networks crucial for safe operation of CAVs can be hacked. Lim & Taeihagh
(2018) conclude that within the EU a proper implementation of the General Data Protection
Regulation (GDPR) can ensure privacy protection. These researchers argue that CAVs are
especially vulnerable to cyber-attacks due to their ability to store highly sensitive data and
transmit such data on external communication networks. The GDPR also provides guidance
on how organisations can comply with legal requirements (Mulder & Vellinga, 2021). These
authors emphasize a three-step approach to cyber-security based on GDPR: first a data
protection impact assessment (DPIA), secondly data protection by design, and finally data
protection by default. Data protection by design and by default are legal obligations set in
Article 25 of the GDPR. A DPIA can contribute to, amongst others, complying with these
two obligations.

LEVITATE | Deliverable D7.5 | WP7 | Final
47
To address cybersecurity the EU enacted the first EU-wide legislation on cybersecurity, the
NIS directive in August 2016 and has also released voluntary cybersecurity guidelines. In
December 2016 the EU agency for Network and Information Security released best
practices guidelines for the cybersecurity of connected vehicles. Cybersecurity and security
concerning private data are important for building trust in and social acceptance of AVs
(Lim & Taeihagh, 2018, also Seetharaman et al, 2020). The GDPR also provides guidance
on how organisations can comply with legal requirements (Mulder and Vellinga, 2021).
Vitunskaite et al. (2019) studied practices of cybersecurity in the cities of Barcelona,
London and Singapore. They observe the following: “The real difficulty for observing
security stems from the complexity of the smart city ecosystem and involvement of a high
number of competing actors and stakeholders. As the cities are still developing, many fail
to take these risks into account and develop an appropriate third-party management
approach. One of the key symptoms of this deficiency is lack of appropriate standards and
guidance, clearly defined roles and responsibilities and a common understanding of key
security requirements. The case studies of Barcelona, Singapore and London has
emphasised and corroborated the importance of technical standards, cyber security
measures and an effective third-party management approach.” (Vitunskaite et al., 2019;
p. 23)
In another paper on cybersecurity in the smart city, Habibzadeh et al. (2019) observe that
it is common knowledge in the literature about public administration that information
technology implementation projects are often derailed by non-technical challenges; issues
of politics, bureaucracy, liability and other non-technical factors slow down the
implementation of technology that is available. Also, with respect to security in the smart
city it is often the case that new technologies have arrived and are deployed whereas
personnel practices, security policies, and other agency practices and municipal practices
tend to lag behind - resulting in a so called “security debt” (Habibzaheh et al., 2019; p. 4).
These authors recommended that cities unambiguously define security roles of individuals
in city administration, that they actively value security leadership, and that the cities form
and maintains specialised security teams to carry out routine security measures such as
training, firmware updates, developing emergency response plans, maintaining
communications with different vendors and service provider
Khan et al. (2020) have studied the various cyber-attacks on automated vehicles and
possible mitigation strategies from a perspective of the communication framework of CAVs.
Based on the literature review, the leading automotive company reports, and the study of
relevant government research bodies, Khan et al. (2020) have described the CAVs
communication framework for all possible interfaces in the form of a flow-chart. The
authors argue that this description has a three-fold value: first, it is imperative to have a
systematic understanding of the CAVs communication framework; second, it is beneficial
for monitoring, assessing, tracking, and combating potential cyber-attacks on various
communication interfaces; third, it will facilitate the development of a robust CAVs
cybersecurity- by-design paradigm by application developers. Important recommendations
from their analysis are (Khan et al., 2020):
• CAVs and connected infrastructure require a continuous surveillance system to alert
relevant operation centres immediately about any data or vehicle breaches
• system designers need to stay up to date with the advances in attacks on the CAV-
embedded system
• manufacturers need to integrate security into every part of their designs

LEVITATE | Deliverable D7.5 | WP7 | Final
48
• in a coordinated approach to CAV cybersecurity ideally a shared problem-solving
approach involves both road operators (as customers) and suppliers such as automotive
manufacturers, equipment manufacturers, data aggregators and data processors
Using risk analysis, Meyer et al. (2021) looked at 6 scenarios of cyberattacks on
autonomous and connected vehicles. They recommend prevention measures to make it
more difficult to manipulate vehicles’ speeds and to protect individual data about travel
patterns. To stimulate vehicle developers to invest in prevention of cyber-attacks,
developers must have sufficient incentives and potentially be held liable for successful
cyberattacks. However, it is probably impossible to prevent all such attacks. Therefore,
measures that limit the consequences of such attacks will be necessary. Such measures
include safety measures in vehicles to protect the occupants in traffic accidents and
measures that make vehicles easier to remove in case they do not function. The last
category of measures includes installing kill switches that make it possible to turn of the
vehicle manually, thus overriding the autopilot and making the vehicle possible to move
by four adults when it is turned off manually (Meyer et al., 2021).
Legal readiness
The EU has not yet amended its legal framework to incorporate AV-related liability and
insurance risks, but it is exploring solutions to these issues. In 2016 the European
Commission launched GEAR 2030 in order to explore solutions to AV-related liability issues.
In May 2016 European Parliament Members recommended that the EC should create a
mandatory insurance scheme and an accompanying fund to safeguard full compensation
for victims of AV accidents and a legal status should be created for all robots to determine
liability in accidents (Taeihagh & Lim, 2019).
Looking at recent developments in the five major areas for legal reform the following
conclusions can be drawn:
• Admission and testing: various countries and states have applied different legal
rules for admission and testing of automated vehicles1F
2; in the future comparative
review of these regulations and associated experiences and outcomes should lay
the groundwork for a more uniform approach in the EU and internationally (Lee &
Hess, 2020);
• Liability: the possible theoretical and legal solutions to liability and insurance have
been outlined by various authors (Evas, 2018; Mardirossian 2020; Bertolini &
Ricaboni, 2021; Vellinga, 2019) and further discussion between stakeholders and
the development of specific cases of litigation will determine the legal option that is
chosen;
• Human-machine interaction: in this particular area a lot of research is still needed
to answer questions on which design of the human-machine interface will allow safe
and reliable control of the vehicle, in all possible circumstances and involving
different traffic situations and different internal states of the driver. Uniform
standards can only be formulated once this research has been carried out and main
conclusions have been agreed upon by all stakeholders involved (Kyriakidis et al.,
2019; Morales-Alvarez et al., 2020; Carsten & Martens, 2018);
2. Published/collected on websites like: https://globalavindex.thedriverlesscommute.com/;
https://www.ncsl.org/research/transportation/autonomous-vehicles-legislative-database.aspx

LEVITATE | Deliverable D7.5 | WP7 | Final
49
• Road infrastructure: both within EU and USA work has been done to formulate
general definitions of the new road classes that are needed to support automated
and autonomous vehicles (Rendant & Geelen, 2020; Liu et al., 2019; Saeed et al.,
2020). ISAD levels (Infrastructure Support Levels for Automated Driving) give an
impetus to define the minimum infrastructure (physical and digital) required to
enable certain self-driving functions (Rendant & Geelen, 2020); for conventional
road infrastructure recognition of road geometry and signs is important and
maintenance is crucial for this; as yet there are no norm or standards in EU referring
to traffic sign machine readability (Lyvritis et al., 2019);
• Digital infrastructure: connected cars require that every vehicle’s location and
journey history be recorded and saved, but the current level of IT security cannot
prevent yet that data may be accessed by unwanted third parties. Thus, the
development of cybersecurity is of the utmost importance for the development of
connected and autonomous driving (Medina et al., 2017). At the moment the
automotive industry lacks a standard approach for dealing with cybersecurity
(Burkacky et al./McKinsey, 2020, 2020). The EU, through the European Union
Agency for Network and Information Security (ENISA) had proposed good practices
that should be considered (Medina et al., 2017); and
• Specific issues concerning electric vehicles: The costs of battery technology, the
number of charging stations and the charging wait time are main variables that will
influence electrification of vehicle fleet (Mahdavian et al., 2021). It has been
estimated that converting all passenger cars in the USA to electric vehicles would
consume 28% more power than the US currently produces (Mahdavian et al.,
2021).
Readiness to engage social and ethical concerns
Introducing automated mobility will raise important social and ethical questions. In many
publications on smart mobility in the smart city it has been emphasised that active
education and engagement of citizens in policy development and decision making is crucial
for the successful implementation of CAVs, CCAM (e.g., Alawadhi et al., 2020; Bezai et al.,
2021; Briyik et al., 2021; Chng et al., 2021; Horizon 2020 Commission Expert Group,
2020; McAslan et al. 2021; Milakis & Muller, 2021; Ayoma & Leon, 2021). User acceptance
of automated vehicles will depend upon how the new automated mobility is perceived, how
it will be used (shared or not, handling of privacy etc.) and what it will cost (Bezai et al.,
2021). Worldwide city management will have to provide and manage new technology that
serves the needs of the city, i.e., the needs of its citizens: “New technologies are not ends
in themselves but have to adapt to what serves the city. In the end, it is the municipalities
that have to implement it” (Freadrich et al., 2018; p. 8).
The Horizon 2020 report on Ethics of connected and automated vehicles gives the following
recommendations for preparing and engaging the public for CAVs (Horizon Commission
Expert Group, 2020; p. 68):
• inform and equip the public with the capacity to claim and exercise their rights and
freedoms in relationship to AI in the context of CAVs
• ensure the development and deployment of methods for communication of information
to all stakeholders, facilitating training, AI literacy, as well as wider public deliberation
• investigate the cognitive and technical challenges users face in CAV interactions and the
tools to help them surmount these changes
Interestingly, Chng and colleagues (2021) investigated citizen perceptions on driverless
mobility by performing Citizen Dialogues, these are structured discussion meetings using

LEVITATE | Deliverable D7.5 | WP7 | Final
50
both qualitative and quantitative methods, designed to be informative, deliberative and
neutral to generate critical but unbiased insights. These dialogues were attended by more
than 900 citizens in 15 cities across North America, Europe and Asia with the following
outcomes:
• public transport was the preferred implementation model for driverless mobility,
followed by ride-sharing and private car ownership
• the levels of trust and acceptance of automated vehicles tended to be lower at higher
levels of vehicle automation
• citizens have reservations about whether industry will sufficiently safeguard citizens’
interests; government should seek to support trust in industrial developments through
regulation and oversight
• the citizens prefer their government to take active roles in driverless mobility and to set
standards and regulations that safeguard and promote their interests
4.4 Future Challenges for urban freight transport
The growing importance of urban freight transport is linked to the growth of the urban
population or urbanisation, a major phenomenon of the 21st century (Hu et al.; 2019).
More than half of the world's population now lives in cities, with one in five people living in
a city with a population exceeding 1 million inhabitants. The UN estimates that by 2030
the world will have 41 megacities with more than 10 million inhabitants and about 70% of
world’s population will live in urban areas by 2050 (United Nations, 2015). Together with
the growing e-commerce sector, this will lead to an increasing demand for freight transport
services and create new challenges for the supporting infrastructure and associated
logistics.
The ERTRAC roadmap on urban freight states that topics related to freight traffic, and to
the exploration of potential synergies between passenger and freight transport at the urban
level are major focal points (ERTRAC, ALICE, 2015). There are important challenges related
to the use of land for urban freight, and the location of logistics activity in and around the
urban environment. Further exploitation of the potential of integrating urban freight and
passenger transport systems will optimise the use of road, rail and inland waterways
infrastructures in space and time, and contribute to healthier cities in terms of less traffic
and congestion. This requires a paradigm shift towards integrated freight/passenger
mobility planning and exploring more opportunities and new business models for the
integration of urban freight with private or public transport at infrastructure and vehicle
levels.
To achieve the best possible future and best outcome for urban freight transport the
following developments are crucial (Hu et al., 2019):
• Passenger transport and freight transport should seek collaboration (e.g., via
automated multi-purpose vehicles)
• Collaborative transportation, supported by city hubs and consolidation centres, are
necessary to improve operational efficiency. CCAM, especially automated hub-to-
hub transport and automated freight consolidation, will contribute significantly
• Multimodality and synchro modality are important factors to aim towards a
sustainable logistic supply chain.
• All the above points require homogenous and shared data among operators, which
is perhaps the most difficult challenge due to the competition between service
providers and freight operators.

LEVITATE | Deliverable D7.5 | WP7 | Final
51
5 Conclusions and recommendations
5.1 Conclusions
The section provides an overview of the primary conclusions that can be drawn from WP7.
Overall effects of CAVs
Estimating the baseline impacts of an increasing share of connected and automated
vehicles (CAVs) for Work Package 7 revealed the following main findings, estimated by
simulations run for the city of Vienna and a Delphi study using experts in the field:
• The increasing presence of connected and automated vehicles in the urban city area is
estimated to have positive impacts on the city environment (less emissions, higher
energy efficiency), and city society and economy (less parking space demand, lower
freight vehicle operating cost) and on urban mobility (less congestion).
• In Work Package 7, the increasing presence of automated vehicles in the city is
estimated to have a temporary negative impact on road safety when penetration
rates of automated vehicles are low. The negative impact found is primarily due to
interactions between human-driven vehicles and automated vehicles, which are
expected to have different driving styles (e.g. AVs adopting different headways) and
different capabilities (e.g. human drivers’ longer reaction times) which may lead to an
initial increase in risks when many human drivers are still on the road. This result differs
from the baseline results found in the road safety impact study (Weijermars et al., 2021)
and discussed in WP5 and WP6, primarily due to two factors: 1) differences in the
network (Vienna) and 2) the inclusion of freight vehicles. Because less data was
available on the driving behaviour of autonomous freight vehicles, less behavioural
parameters were adjusted and autonomous freight vehicles may behave more similarly
to human-driven freight vehicles. This led to higher crash rate estimations when freight
vehicles were included.
Larger positive impacts on road safety are estimated once human-driven vehicles are
replaced and second-generation automated vehicles make up at least 60% of the city’s
vehicle fleet. More broadly within LEVITATE, most estimates point to a large reduction
in crashes with the introduction of automated vehicles including a small reduction at low
penetration rates. At low penetration rates, the balance between the safety of
automated vehicles (which are expected to crash less often than human-driven vehicles)
and the potential risks of mixed traffic (when human-driven/less advanced automated
vehicles are still on the road) is a point of attention for further research.
• The increasing presence of automated vehicles in the city is estimated to have a slightly
negative impact on public health when traditional (human-driven) vehicles make up
the majority of vehicles, followed by a slightly positive impact at full automation of
the vehicle fleet.

LEVITATE | Deliverable D7.5 | WP7 | Final
52
Effects of SUCs: automated delivery, consolidation and hub-to-hub transport
On top of the baseline impacts of increasing CAV penetration, the automated freight sub-
use cases yielded some additional effects:
• The automated delivery sub-use case is associated with additional benefits for energy
efficiency, CO2 emissions, road safety, congestion, public health and vehicle operating
costs. The night-time-only automated delivery scenarios (see Appendix A) show
additional benefits particularly for the two mobility indicators (travel time and
congestion), due to less interaction with the larger daytime traffic volumes.
• The automated consolidation sub-use case is associated with additional benefits for
energy efficiency, CO2 emissions, road safety, congestion, travel time, public health and
vehicle operating costs. Compared to automated delivery without consolidation at city
hubs (the first sub-use case), further improvements in energy efficiency, operating
costs, and a large reduction in total kilometres travelled are expected. This suggests
that centrally located city-hubs can help realise a more efficient allocation of resources.
• The hub-to-hub sub-use case is expected to deliver additional benefits for energy
efficiency, CO2 emissions, road safety, congestion, travel time, public health, and freight
vehicle operating costs.
• All three automated freight SUCs are predicted to (marginally) improve road safety
compared to the baseline, particularly at lower penetration rates when less of the
remaining vehicle fleet is automated.
• At the higher-level CAV penetration rates (above 80%), all three automated freight
SUCs are expected to require slightly more parking space (less reduction) than in the
baseline without automated delivery. The hub-to-hub SUC is even expected to slightly
increase parking space requirements at 100% CAV penetration compared to the current
situation (with 100% human-driven vehicles).
• The sub-use cases of automated delivery, hub-to-hub and especially automated
consolidation are predicted positively impact public health compared to the baseline.
This positive expectation is likely based on the expected additional benefits of these
sub-use cases for both road safety and emissions.
• Using data on freight delivery trips in Vianna, it was estimated that compared to manual
freight delivery, completely automated delivery and automated delivery with city-hubs
will have substantially reduced annual fleet costs (-68%).
Effects of truck platooning on bridges
Connected and automated freight vehicles are expected to facilitate truck platooning, and
as a result potentially test the strength of bridges. The study of truck platooning on bridges
yielded the following main conclusions:
• The largest effect of truck platooning on simple single span (beam) bridges as
modelled in LEVITATE is observed for the criteria of braking forces. For bridges above
80m length, it has been estimated that the braking force is at least double of the
baseline scenario.
• According to standard bridge models and standard traffic simulations within LEVITATE,
the need for strengthening structural resistance of bridges arises for existing
bridges with = 1 starting from span length of 55 m for bending moment and 60 m for
shear force ULS; for existing bridges with resistance at resistance level of = 0.8,
strengthening needs would arise sooner – starting from bridge spans of 40 m.

LEVITATE | Deliverable D7.5 | WP7 | Final
53
• For bridge strengthening, a model and guidelines for estimating the costs in relation to
the initial construction costs have been developed (D7.3).
• As an alternative to strengthening bridges, intelligent access control can be used to
arrange the increase of inter-vehicle distances for the bridge section to meet the code
level and prevent. Headway have been recommended and these are presented in
LEVITATE D7.3 (Hu et al., 2021b). Forcing an increased inter-vehicle distance by
intelligent access control will not diminish the ecological and economic benefits of truck
platoons.
5.2 Policy recommendations
The introduction of CCAM and the implementation of interventions (sub-use cases) in the
area of public transport is part of a wider transition towards smart and sustainable cities
(Alawadhi et al., 2020; Aoyama & Leon, 2021; Bezai et al., 2020; Chng et al., 2021; Lim
& Taeihagh, 2018; Vitunskaite et al., 2019; Mahdavian et al., 2021; McAslan et al., 2021;
Medina et al., 2017; Milakis & Müller, 2021; Seuwou et al., 2019; Taeihagh & Lim, 2019).
A successful transition will largely be impacted by the policy measures of the city, local
and national authorities. Therefore, the LEVITATE project aims to support the authorities
finding the most beneficial policies on the way towards an automated transport system.
Based on recent literature dealing with the transition from a 100% human driver vehicle
population to a 100% autonomous system without any human drivers, and the results of
the LEVITATE project, in particular WP7, the following recommendations can be suggested
to make city managers and policy makers aware of what is to be done to support this
transition and the overall success of Cooperative, Connected and Automated Mobility
(CCAM) and use cases:
• City managers and policy makers should take into account four major areas of readiness
for CCAM (autonomous driving): technology readiness, infrastructure readiness, legal
readiness and the readiness to address social acceptance and ethical/social value issues
(e.g., Alawadhi et al., 2020; Bezai et al., 2020)
• Commercial (technology) push alone will not safeguard the expected social benefits of
CCAM (cooperative, connected and automated vehicles); new types of governance and
planning are called for with a stronger engagement of citizen groups and city
stakeholders, a stronger focus on long term implications, and lesser reliance on
traditional forecasting and traffic models (e.g., McAslan et al, 2021; Milakis & Müller,
2021)
• More anticipatory engaging styles of governance will not spontaneously develop; an
anticipatory governance capacity has to be built (e.g., McAslan et al., 2021)
• Good legislation, guidance and guidelines for CCAM in Europe is already partly available
(e.g., the GDPR, White Paper, Horzon Group report on Ethical guidelines). Authorities
need to be aware of these and use these to survey what implications they have for
planning and policy making at the city level (e.g., Mulder & Vellinga, 2021)
• There are many regulatory gaps for CCAM; using their own experiences and policy and
planning orientations city managers, policy makers and planners should cooperate and
contribute to the national and international debate about how these gaps should be
resolved (e.g., Aoyama & Leon, 2021)
• Automation in freight transport causes its unique problems such as the additional
burden on bridges caused by truck platoons with short headways. In order to minimize
the failure probability, measures such as structural strengthening of single span beam
bridges and intelligent access control should be considered (Hu et al., 2021b).

LEVITATE | Deliverable D7.5 | WP7 | Final
54
• The transition towards CCAM is as much a social and cultural phenomenon as a
technological phenomenon; ultimately a lot if not all depends upon trust in new
technology and trust will be easier to build if citizens have an active voice in what
happens in their neighbourhoods (e.g., Chng et al., 2021, Medina et al., 2017; McAslan
et al., 2021)
• The transition towards CCAM requires building of and participation in new broad
alliances and platforms where many different actors from industry, and interest and
citizen groups are present
• The risks concerning cybersecurity need a full understanding of the total digital
communication framework and all interfaces of connected and automated vehicles;
security-by-design is one of the most general and important principles to follow (e.g.,
Khan et al. 2020)
• The risks concerning cybersecurity cannot be solely managed by legislation and
technocratic controlling strategies but demand social awareness, social education and
cultural change in companies and citizens and third-party management (e.g., Khan et
al., 2020; Vitunskaite et al., 2019)
• Backcasting is one of the analytic methods that can help policy makers to make better
informed decisions about how new technology can be implemented to achieve the
expected benefits (e.g., Milakis & Müller, 2021). For this reason, Levitate has included
a backcasting capability as part of the PST.
Research in these various areas – new governance style, cybersecurity measures and
culture, cooperation between varied stakeholder groups, regulatory gaps, citizen
engagement, ethical concerns - can help develop a better understanding of problems and
issues, possible solutions, and to better informed policy decisions.
For freight transport a number of recommendations can be given (Hu et al., 2019):
• Passenger transport and freight transport should seek collaboration (e.g., via automated
multi-purpose vehicles)
• Collaborative transportation, supported by city hubs and consolidation centres, are
necessary to improve operational efficiency. CCAM, especially automated hub-to-hub
transport and automated freight consolidation, will contribute significantly
• Multimodality and synchro modality are important factors to aim towards a sustainable
logistic supply chain.
• All the above points require homogenous and shared data among operators, which is
perhaps the most difficult challenge due to the competition between service providers
and freight operators.

LEVITATE | Deliverable D7.5 | WP7 | Final
55
References
Alawadhi, M., Almazrouie, J., Kamil, M., & Khalil, K.A. (2020). A systematic literature
review of the factors influencing the adoption of autonomous driving. International Journal
of System Assurance Engineering and Management 11, 1065–1082.
https://doi.org/10.1007/s13198-020-00961-4
Allen, J., Browne, M., Woodburn, A., & Leonardi, J. (2012) The Role of Urban Consolidation
Centres in Sustainable Freight Transport. Transport Reviews, 32 (4), 473-490.
Aoyama, Y., & Leon, L.F.A. (2021). Urban governance and autonomous vehicles. Cities,
Volume 119, 103410. https://doi.org/10.1016/j.cities.2021.103410
Baum, L., Assmann, T., & Strubelt, H. (2019). State of the art - Automated micro-vehicles
for urban logistics. IFAC-PapersOnLine, 52 (13), 2455-2462.
Bertolini, A. & Riccaboni, M. (2021). Grounding the case for a European approach to the
regulation of automated driving: the technology-selection effect of liability rules. European
Journal of Law and Economics. http://doi.10.1007/s10657-020-09671-5
Bezai, N.E., Medjdoub, B., Al-Habaibeh, A., Chalal, M.L., & Fadli, F. (2021). Future cities
and autonomous vehicles: analysis of the barriers to full adoption, Energy and Built
Environment, 2(1), 65-81. https://doi.org/10.1016/j.enbenv.2020.05.002.
Bıyık, C., Abareshi, A., Paz, A., Ruiz, R.A., Battarra, R., Rogers, C.D.F., Lizarraga, C.
(2021). Smart Mobility Adoption: A Review of the Literature. Journal of Open Innovation:
Technology, Market and Complexity, 7, 146. https://doi.org/10.3390/joitmc7020146
Burkacky, O., Deichmann, J., Klein, B., Pototzky, K., & Scherf, G. (2020). Cybersecurity
in automotive: Mastering the challenge. Munich, McKinsey.
Cafiso, S., Di Graziano, A., & Pappalardo, G. (2013). Road safety issues for bus transport
management. Accident Analysis and Prevention, 60, 324-333.
https://doi:10.1016/j.aap.2013.06.010
Cao, Z. & Ceder, A. (2019). Autonomous shuttle bus service timetabling and vehicle
scheduling using skip-stop tactic. Transportation Research Part C: Emerging Technologies,
102, 370-395.
Carsten, O., & Martens, M.H. (2019). How can humans understand their automated cars?
HMI principles, problems and solutions. Cognition, Technology & Work, 21, 3–20.
https://doi.org/10.1007/s10111-018-0484-0
Cerrudo, C., Hasbini, H., & Russell, B. (2015). Cyber Security Guidelines for Smart City
Technology Adoption. Cloud Security Alliance.
Chng, S., Kong, P., Lim, P.Y., Cornet, H., Cheah, L. (2021). Engaging citizens in driverless
mobility: Insights from a global dialogue for research, design and policy, Transportation

LEVITATE | Deliverable D7.5 | WP7 | Final
56
Research Interdisciplinary Perspectives, 11, 100443,
https://doi.org/10.1016/j.trip.2021.100443
Charisis, A., Spana, S., Kaisar, E., & Du, L. (2020). Logistics hub location-scheduling model
for inner-city last mile deliveries. International Journal for Traffic & Transport Engineering,
10(2), 169-186.
City of Manchester (2017). The Greater Manchester Transport Strategy 2040. First
published February 2017.
City of Vienna (2015). Urban Mobility Plan Vienna. Available at
https://www.wien.gv.at/stadtentwicklung/studien/pdf/b008443.pdf
Clarke, G., & Wright, J. W. (1964). Scheduling of vehicles from a central depot to a number
of delivery points. Operations research, 12(4), 568-581.
Cramer, W., Mang, C., McDonnell, A., Nefores, S., & Weisman, L. (2020). The impact of
automation on shipping and receiving. Proceedings of the International Annual Conference
of the American Society for Engineering Management.
Dalkey, N., & Helmer, O. (1963). An experimental application of the Delphi method to the
use of experts. Management Science, 9, 458- 467. https://doi:10.1287/mnsc.9.3.458
Dorling, K., Heinrichs, J., Messier, G. G., & Magierowski, S. (2017). Vehicle Routing
Problems for Drone Delivery. In IEEE Transactions on Systems, Man, and Cybernetics:
Systems, 47 (1), 70-85.
EEA (2020). Air quality in Europe — 2020 report. European Environment Agency.
https://www.europarl.europa.eu/meetdocs/2014_2019/plmrep/COMMITTEES/ENVI/DV/2
021/01-14/Air_quality_in_Europe-2020_report_EN.pdf
Elvik, R., Quddus, M., Papadoulis, A., Cleij, D., Weijermars, W., Millonig, A., Vorwagner,
A., Hu, B., & Nitsche, P. (2019). A taxonomy of potential impacts of connected and
automated vehicles at different levels of implementation. Deliverable D3.1 of the H2020
project LEVITATE.
Elvik, R., Meyer, S. F., Hu, B., Ralbovsky, M., Vorwagner, A., Boghani, H. (2020).
Methods for forecasting the impacts of connected and automated vehicles, Deliverable
D3.2 of the H2020 project LEVITATE.
ERTRAC (2019). Connected Automated Driving Roadmap. Retrieved from:
https://www.ertrac.org/index.php?page=ertrac-roadmap
ERTRAC (2019). Long Distance Freight Transport. Retrieved from:
https://www.ertrac.org/index.php?page=ertrac-roadmap
ERTRAC, ALICE. (2015). Roadmap on Urban Freight. Retrieved from:
https://www.ertrac.org/index.php?page=ertrac-roadmap

LEVITATE | Deliverable D7.5 | WP7 | Final
57
European Commission (2017). Towards clean, competitive and connected mobility: the
contribution of Transport Research and Innovation to the Mobility package, SWD (2017)
223 final.
Evas, T. (2018). A Common EU Approach to Liability Rules and Insurance for Connected
and Autonomous Vehicles: European Added Value Assessment: Accompanying the
European Parliament's legislative own-initiative report. Brussels, European Parliamentary
Research Service. Retrieved from:
https://www.europarl.europa.eu/RegData/etudes/STUD/2018/615635/EPRS_STU(2018)6
15635_EN.pdf
Figliozzi, M. and Jennings, D. (2020). Autonomous delivery robots and their potential
impacts on urban freight energy consumption and emissions. Transportation Research
Procedia, 46, 21-28.
Firnkorn, J. and Müller, M., (2015). Free-Floating Electric Carsharing-Fleets in Smart Cities:
The Dawning of a Post-Private Car Era in Urban Environments? Environmental Science &
Policy, 45, 30-40.
Fraedrich, E., Heinrichs, D., Bahamonde-Birke, F. J., & Cyganski, R. (2019). Autonomous
driving, the built environment and policy implications. Transportation Research Part A:
Policy and Practice, 122(March 2018), 162–172.
https://doi.org/10.1016/j.tra.2018.02.018
Freundt, U., Böning, S. (2011). Anpassung von DIN-Fachberichten „Brücken“a Eurocodes
(Adaptation of DIN technical reports “Bridges” to Eurocodes), Berichte der Bundesanstalt
für Straßenwesen (Reports of the Federal Highway Research Institute), Brücken- und
Ingenieurbau Heft B 77 (Bridges and Engineering Construction issue B 77), ISBN 978-3-
86918-108-0, Bergisch Gladbach, Germany.
Gevaers R., Van de Voorde E., & Vanelslander T. (2014). Cost Modelling and Simulation of
Last-mile Characteristics in an Innovative B2C Supply Chain Environment with Implications
on Urban Areas and Cities. Procedia - Social and Behavioral Sciences,125, 398-411.
Habibzadeh, H., Nussbaum, B.H., Anjomshoa, F., Kantarci, B., Soyata, T. (2019). A survey
on cybersecurity, data privacy, and policy issues in cyber-physical system deployments in
smart cities, Sustainable Cities and Society, 50.
https://doi.org/10.1016/j.scs.2019.101660
Hagenzieker, M. P., Commandeur, J. J. F., & Bijleveld, F. D. (2014). The history of road
safety research: A quantitative approach. Transportation Research Part F, 25, 150-162.
Hibberd, D., Louw, T., et al. (2018). From research questions to logging requirements.
Deliverable D3.1. L3 Pilot Driving Automation. University of Leeds.
Horizon 2020 Commission Expert Group on ethics of driverless mobility – E03659 (2020).
Ethics of Connected and Automated Vehicles: recommendations on road safety, privacy,
fairness, explainability and responsibil8ity. Luxembourg, Publication Office of the European
Union.

LEVITATE | Deliverable D7.5 | WP7 | Final
58
Hsu, C., & Sandford, B. (2007) The Delphi Technique: Making Sense of Consensus.
Practical Assessment, Research & Evaluation, 12, 1-8.
http://pareonline.net/pdf/v12n10.pdf
Hu, B., Zwart, R.d., Papazikou, E., Boghani, H.C., Filtness, A., & Roussou, J., (2019).
Defining the future of freight transport, Deliverable D7.1 of the H2020 project
LEVITATE.
Hu, B., Brandstätter, G., Ralbovsky, M., Kwapisz, M., Vorwagner, A., Zwart, R.d., Mons,
C., Weijermars, W., Roussou, J., Oikonomou, M., Ziakopoulos, Chaudhry, A., Sha, S.,
Haouari, R., & Boghani, H.C., (2021a). Short-term impacts of CCAM on freight transport,
Deliverable D7.2 of the H2020 project LEVITATE.
Hu, B., Brandstätter, G., Ralbovsky, M., Kwapisz, M., Vorwagner, A., Zwart, R.d., Mons,
C., Weijermars, W., Roussou, J., Oikonomou, M., Ziakopoulos, Chaudhry, A., Sha, S.,
Haouari, R., & Boghani, H.C. (2021b). Medium-term impacts of CCAM on freight transport,
Deliverable D7.3 of the H2020 project LEVITATE.
Hu, B., Brandstätter, G., Gebhard, S., A., Zwart, R.d., Mons, C., Weijermars, W., Roussou,
J., Oikonomou, M., Ziakopoulos, Chaudhry, A., Sha, S., Haouari, R., & Boghani, H.C.
(2021c). Long term impacts of CCAM on freight transport, Deliverable D7.4 of the H2020
project LEVITATE.
Jennings, D., & Figliozzi, M. (2019). Study of Sidewalk Autonomous Delivery Robots and
Their Potential Impacts on Freight Efficiency and Travel. Transportation Research Record:
Journal of the Transportation Research Board, 2673(6), 317–326.
Khan, S.K., Shiwakoti, N., Stasinopoulos, P. & Chen, Y. (2020). Cyber-attacks in the next-
generation cars, mitigation techniques, anticipated readiness and future directions.
Accident Analysis & Prevention, 148. https://doi.org/10.1016/j.aap.2020.105837
Lagorio, A., Pinto, R., & Golini, R. (2016). Research in urban logistics: A systematic
literature review. International Journal of Physical Distribution & Logistics Management,
46, 908–931.
Lee, D., & Hess, D.J (2020). Regulations for on-road testing of connected and
automated vehicles: Assessing the potential for global safety harmonization.
Transportation Research Part A: Policy and Practice, 136, 85-98.
https://doi.org/10.1016/j.tra.2020.03.026
Lim, H.S.M. & Taeihagh, A. (2018). Autonomous Vehicles for Smart and Sustainable Cities:
An In-Depth Exploration of Privacy and Cybersecurity Implications. Energies, 11, 1062.
Liu, Y.; Tight, M.; Sun, Q.; Kang, R. (2019). A systematic review: Road infrastructure
requirement for Connected and Autonomous Vehicles (CAVs). Journal of Physics:
Conference Series, 1187, 042073.

LEVITATE | Deliverable D7.5 | WP7 | Final
59
Liu, N., Nikitas, A., Parkinson, S. (2020). Exploring expert perceptions about the cyber
security and privacy of Connected and Autonomous Vehicles: A thematic analysis
approach. Transportation Research Part F,75, 66-86.
Lutin, J., Kornhauser, A., Spears, J., & Sanders, L. (2016). A Research Roadmap for
Substantially Improving Safety for Transit Buses through Autonomous Braking Assistance
for Operators. Paper presented at the Transportation Research Board 95th Annual Meeting,
Washington DC, United States.
Lytrivis, A., Manganiaris, S., Reckenzaun, J., Solmaz, S., Protzmann, R., Adaktylos, A.-M.,
Wimmer, Y., Atasayar, H., Daura, X., & Porcuna, D. (2019). Deliverable. D.5.4
Infrastructure Classification Scheme. INFRAMIX – Road INFRAstructure ready for MIXed
vehicle traffic flows
Mahdavian, A., Shojaei, A., Mccormick, S., Papandreou, T., Eluru, N., & Oloufa, A.A.
(2021). Drivers and Barriers to Implementation of Connected, Automated, Shared, and
Electric Vehicles: An Agenda for Future Research. IEEE Access 9, 22195-22213.
Mardirossian, V. (2020). Will Autonomous Cars Put an End to the Traditional Third-Party
Liability Insurance Coverage? In: P. Marano & K. Noussia (Eds.), InInsurTech: A Legal and
Regulatory View (pp. 271-290). Switzerland: Springer-Verlag.
McAslan, D., Gabriele, M. & Miller, T.R. (2021) Planning and Policy Directions for
Autonomous Vehicles in Metropolitan Planning Organizations (MPOs) in the United States,
Journal of Urban Technology. https://doi.org/10.1080/10630732.2021.1944751
Medina, A., Maulana, A., Thompson, D., Shandilya N., Almeida, S., Aapaoka A., Kutila, M.
(2017). Public Support Measures for Connected and Automated Driving: Final Report.
GROW-SME-15-C-N102. European Commission EC. EU Publications, No. EA-01-17-634-
EN-N. https://ec.europa.e
Meyer, S.F., Elvik, R. & Johnsson, E. (2021). Risk analysis for forecasting cyberattacks
against connected and autonomous vehicles. Journal of Transportation Security
https://doi.org/10.1007/s12198-021-00236-
Mello, E.F., & Bauer, P.H. (2019). Energy Benefits of Urban Platooning with Self-Driving
Vehicles. International Journal of Transport and Vehicle Engineering, 13(2), 94-
100.
Milakis, D & Müller, S. (2021). The societal dimension of the automated vehicles transition:
Towards a research agenda. Cities, 113, 103144,
https://doi.org/10.1016/j.cities.2021.103144
Panteia (2015). Analysis is of the trends and prospects of jobs and working conditions in
transport. Zoetermeer, Panteia.
Petegem, J.W.H. van, Nes, C.N. van, Boele, M.J., & Eenink, R.G. (2018). Advies
praktijkproef: Starship bezorgrobot. R-2018-4. SWOV, The Hague, The Netherlands.
Pakusch, C., & Bossauer, P. (2017). User Acceptance of Fully Autonomous Public Transport
Mittelstand 4.0-Kompetenzzentrum Usability View project Einfach Teilen (Easy P2P

LEVITATE | Deliverable D7.5 | WP7 | Final
60
Carsharing) View project User Acceptance of Fully Autonomous Public Transport. 2(Icete),
52–60. https://doi.org/10.5220/0006472900520060
Papazikou, E., Zach, M., Boghani, H.C., Elvik, R., Tympakianaki, A., Nogues, L., Hu, B.
(2020). Detailed list of sub-use cases, applicable forecasting methodologies and necessary
output variables, Deliverable D4.4 of the H2020 project LEVITATE.
Roussou, J., Papazikou, E., Zwart, R.d., Hu, B., Boghani, H.C., Yannis, G., (2019). Defining
the future of urban transport, Deliverable D5.1 of the H2020 project LEVITATE.
Pruis, J.O. (2000). Evaluatie proefproject parkshuttle: Eindrapport exploitative
(vertrouwelijk) (evaluation of the pilot project parkshuttle: Final report operation.
Technical report. ANT, Rotterdam.
Quak, H., Nesterova, N., van Rooijen, T., & Dong, Y. (2016). Zero emission city logistics:
current practices in freight electromobility and feasibility in the near future. Transportation
Research Procedia, 14, 1506-1515.
Rendant, K & Geelen, van (2020). Connected & Autonomous Vehicles and road
infrastructure State of play and outlook. Brussels, Belgian Road Research Centre.
Ritter, K. (2017). Driverless electric shuttle being tested in downtown Vegas. Available:
https://phys.org/news/2017-01- driverless-shuttle-thrill-downtown-las.html.
Rojas-Rueda, D., Nieuwenhuijsen, M. J., Khreis, H., & Frumkin, H. (2020). Autonomous
vehicles and public health. Annual review of public health, 41, 329-345.
Roussou, J., Oikonomou, M., Mourtakos, V., Müller, J., Vlahogianni, E., Ziakopoulos, A.,
Hu, B., Chaudhry, A., & Yannis, G., (2021). Medium-term impacts of CCAM on urban
transport, Deliverable D5.3 of the H2020 project LEVITATE.
Saeed, T.U., Alabi, B.N.T., & Labi, S. (2020). Preparing Road Infrastructure to
Accommodate Connected and Automated Vehicles: System-Level Perspective. Journal of
Infrastructure Systems, https://doi.1061/(ASCE)IS.1943-555X.0000593
Seuwou, P., Banissi, E., & Ubakanma, G. (2019). The Future of Mobility with Connected
and Autonomous Vehicles in Smart Cities. In The Future of Mobility with Connected and
Autonomous Vehicles in Smart Cities (pp. 37-52). Springer Nature.
https://doi.org/10.1007/978-3-030-18732-3_3
Sha, H., Boghani, H., Chaudhry, A., Quddus, M., Morris, A., Thomas, P. (2021). LEVITATE:
Passenger Cars Microsimulation Sub-use Cases Findings. LEVITATE (Horizon 2020),
January 2021
Shi, Y., Li, Y., Cai, Q., Zhang, H., & Wu, D. (2020). How Does Heterogeneity Affect Freeway
Safety? A Simulation-Based Exploration Considering Sustainable Intelligent Connected
Vehicles. Sustainability, 12(21), 8941. doi:10.3390/su12218941
Strömberg, H., Ramos, É.M.S., Karlsson, M. et al. (2021). A future without drivers?
Comparing users', urban planners' and developers' assumptions, hopes, and concerns
about autonomous vehicles. European Transport Research Review, 13, 44.
https://doi.org/10.1186/s12544-021-00503-4

LEVITATE | Deliverable D7.5 | WP7 | Final
61
Taeihagh, A., & Lim, H.S.M. (2019). Governing autonomous vehicles: emerging responses
for safety, liability, privacy, cybersecurity, and industry risks. Transport Reviews, 39 (1),
103-128.
Tarko, A. P. (2018). Estimating the expected number of crashes with traffic conflicts and
the Lomax Distribution—A theoretical and numerical exploration. Accident Analysis &
Prevention, 113, 63-73.
Toth, P., & Vigo, D. (2014). Vehicle routing: problems, methods, and applications. Society
for Industrial and Applied Mathematics.
United Nations, Department of Economic and Social Affairs, Population Division (2015).
Population 2030: Demographic challenges and opportunities for sustainable
development planning.
Vellinga, N.E. (2019) Automated driving and its challenges to international traffic law:
which way to go? Law, Innovation and Technology, 11(2), 257-278.
https://doi.org/10.1080/17579961.2019.1665798
Vitunskaite, M., He, Y., Brandstetter, T., & Janicke, H. (2019). Smart cities and cyber
security: Are we there yet? A comparative study on the role of standards, third party risk
management and security ownership. Computers & Security, 83, 313-331.
Weijermars, W. et al. (2021). Road safety related impacts within the Levitate project.
Working paper of the road safety working group of the H2020 project LEVITATE.
Wirtschaftskammer Wien. KEP - Branchenreport 2020.
Yu, H., Tak, S., Park, M., & Yeo, H. (2019). Impact of Autonomous-Vehicle-Only Lanes in
Mixed Traffic Conditions. Transportation Research Record: Journal of the Transportation
Research Board, 2673(9), 430--439. doi:10.1177/0361198119847475

LEVITATE | Deliverable D7.5 | WP7 | Final
1
Appendix A Full results
A.1 Environmental impacts
Market penetration rate: AVs in Background vehicle fleet
(Human-driven vehicle - 1st Generation AV - 2nd Generation AV)
Impact Sub-use Case
100-0-
0
80-20-
0
60-40-
0
40-40-
20
20-40-
40
0-40-
60
0-20-
80
0-0-100 Method
Energy
efficiency
Baseline 0,0% -3,7% 6,5% 8,2% 11,9% 16,0% 16,0% 16,0%
Delphi
Automated delivery 0,0% 6,1% 7,8% 11,1% 14,8% 20,4% 20,4% 20,4%
Automated delivery during
night-time only
0,0% 7,7% 5,7% 10,1% 11,8% 11,8% 11,8% 11,8%
Automated consolidation 0,0% 7,4% 12,5% 16,6% 20,7% 25,2% 25,2% 25,2%
Hub-to-hub 0,0% 5,6% 7,8% 13,5% 18,2% 18,2% 18,2% 18,2%
CO2
emissions
Baseline 0% -21% -50% -80% -90% -100% -100% -100%
Micro-
simulation
(Vienna)
Automated delivery -100%
Automated consolidation -100%
Hub-to-hub -100%

LEVITATE | Deliverable D7.5 | WP7 | Final
2
A.2 Mobility impacts
Market penetration rate: AVs in Background vehicle fleet
(Human-driven vehicle - 1st Generation AV - 2nd Generation AV)
Impact
Sub-use
Case
Scenario
100-0-0
80-20-0
60-40-0
40-40-20
20-40-40
0-40-60
0-20-80
0-0-100
Method
Travel
time
Baseline
0,0% 6,1% 6,3% 5,5% 2,2% 0,5% 0,5% 0,5%
Delphi
Automated delivery 0,0% 1,7% 1,7% -3,7% -6,8% -4,4% -4,4% -4,4%
Automated delivery during night-time only 0,0% -7,7% -7,0% -14,2% -16,4% -12,4% -12,4% -12,4%
Automated consolidation 0,0% -4,2% -4,2% -8,8% -9,8% -11,3% -11,3% -11,3%
Hub-to-hub 0,0% -2,9% -3,2% -4,8% -6,4% -6,4% -6,4% -6,4%
Congesti
on of
freight
vehicles
Baseline;
manual
delivery
Combined urban & periphery 0,0% -15,5% -6,5% -4,7% -7,8% -17,2% -11,5% -8,6%
Micro-
simulation
(Vienna)
Urban network 0,0% -18,7% -8,3% -7,5% -11,5% -22,4% -14,4% -11,9%
Periphery network 0,0% 2,1% 3,1% 10,1% 12,6% 11,5% 3,8% 9,1%
Semi-
automated
delivery
Combined urban & periphery;
daytime -17,4% -12,5% -16,4% -27,4% -7,4% -22,9% -28,0% -17,2%
Automated
delivery
Combined urban & periphery;
daytime
-42,4% -38,9% -42,2% -49,0% -35,6% -46,1% -49,9% -42,3%
Urban network; daytime -19,0% -13,0% -17,4% -30,6% -6,9% -25,2% -31,3% -18,1%
Periphery network; daytime -9,1% -9,8% -10,1% -10,8% -9,1% -9,4% -10,8% -11,5%
Combined urban & periphery;
night-time
-92,0% -91,8% -93,9% -92,0% -92,2% -92,4% -93,5% -92,6%
Urban network; night-time -92,4% -92,2% -94,5% -92,5% -92,5% -92,9% -94,0% -92,9%
Periphery network; night-time -90,2% -89,2% -90,6% -90,2% -90,9% -90,6% -89,9% -90,9%
Automated
consolidation
Combined urban & periphery;
daytime
-42,4% -38,9% -42,2% -49,0% -35,6% -46,1% -49,9% -42,3%
Hub-to-hub
transport
Baseline; no transfer hub 0,0% -9,3% -11,3% -17,5% -19,6% -22,7% -23,7% -24,7%
With transfer hub 0,0% -11,3% -17,5% -21,6% -23,7% -24,7% -24,7% -26,8%

LEVITATE | Deliverable D7.5 | WP7 | Final
3
A.3 Societal impacts
Market penetration rate: AVs in Background vehicle fleet
(Human-driven vehicle - 1st Generation AV - 2nd Generation AV)
Impact Sub-use Case Scenario
100-0-
0
80-20-
0
60-40-
0
40-40-
20
20-40-
40
0-40-
60
0-20-
80
0-0-
100
Method
Vehicle
operating
cost
Baseline
0,0%
7,5%
-0,7%
-4,0%
-10,0%
-10,4%
-10,4%
-10,4%
Delphi
Automated delivery 0,0% 2,4% 4,4% 1,7% -3,7% -7,9% -7,9% -7,9%
Automated delivery during night-time only 0,0% -3,1% -1,2% -11,6% -17,0% -17,7% -17,7% -17,7%
Automated consolidation
0,0%
2,4%
0,9%
-7,3%
-13,7%
-17,4%
-17,4%
-17,4%
Hub-to-hub 0,0% -3,3% -6,2% -6,2% -12,3% -15,5% -15,5% -15,5%
Vehicle
operating
cost
Automated delivery -77% Operations
research
Automated consolidation -21%
Parking
space
required
Baseline 0,0% -1,4% -1,3% -5,0% -11,5% -11,6% -11,6% -11,6%
Automated delivery 0,0% -7,9% -4,6% -6,8% -5,1% -4,0% -4,0% -4,0%
Automated delivery during night-time only
0,0%
-4,0%
-5,4%
-8,7%
-9,4%
-7,9%
-7,9%
-7,9%
Micro-
simulation
(Vienna)
Automated consolidation
0,0%
-4,3%
-2,8%
-2,8%
-4,2%
-3,8%
-3,8%
-3,8%
Hub-to-hub 0,0% -1,6% -1,5% -3,3% 0,0% 1,4% 1,4% 1,4%
Road safety:
crash rate
Baseline; manual delivery
Combined urban & periphery 0,0% 7,5% 14,0% 16,1% 4,3% -19,5% -37,0% -48,7%
Urban network
0,0%
5,5%
12,7%
14,5%
0,0%
-20,0%
-36,4%
-49,1%
Periphery network
0,0%
11,5%
15,4%
19,2%
15,4%
-19,2%
-38,5%
-50,0%
Automated delivery
Combined urban & periphery -2,6% -4,2% 10,2% 4,9% 4,6% -19,8% -41,0% -50,2%
Urban network -3,6% -9,1% 7,3% 0,0% 5,5% -18,2% -41,8% -50,9%
Periphery network
0,0%
3,8%
15,4%
15,4%
3,8%
-23,1%
-38,5%
-46,2%
Automated consolidation
Combined urban & periphery; daytime
-2,6%
-4,2%
10,2%
4,9%
4,6%
-19,8%
-41,0%
-50,2%
Hub-to-hub transport Baseline; no transfer hub 0,0% 11,5% 23,1% 11,5% 0,0% -11,5% -34,6% -61,5%
With transfer hub 0,0% 7,7% 11,5% -3,8% -11,5% -23,1% -46,2% -61,5%
Public
health
Baseline
0,0%
-5,3%
-2,1%
0,0%
5,2%
4,0%
4,0%
4,0%
Delphi
Automated delivery 0,0% 2,9% 4,7% 8,8% 8,8% 8,4% 8,4% 8,4%
Automated delivery during night-time only
0,0%
2,3%
3,8%
4,9%
8,2%
11,8%
11,8%
11,8%
Automated consolidation
0,0%
6,0%
7,7%
9,4%
14,4%
18,5%
18,5%
18,5%
Hub-to-hub
0,0%
6,6%
10,0%
12,0%
17,8%
15,7%
15,7%
15,7%

LEVITATE | Deliverable D7.5 | WP7 | Final
1
Appendix B Cost assumptions
vehicle operating costs
For assessing the vehicle operating cost (Section 3.4), Hu et al. (2021a) made the following
assumptions.
Manual delivery:
• For a conventional delivery transporter, we assume acquisition costs of EUR 30,000
(model of Mercedes Vito). With a linear deprecation over 10 years, the costs are
EUR 3,000 per year.
• Costs for insurance, maintenance and fuel are assumed to cost EUR 5,000 per year.
• The average salary of a driver for parcel delivery is around EUR 35,000 per year2F
3,
and the employer pays EUR 45,500 per year due to additional tax and insurance.
• The total costs for a conventional delivery vehicle are therefore EUR 53,500 per
year.
Semi-automated delivery:
• Based on LEVITATE deliverable D3.2 (Elvik et al., 2020), we assume the costs for
a level 5 automated van to be EUR 50,000. With a linear deprecation over 10 years,
the costs are EUR 5,000 per year.
• Costs for insurance, maintenance and energy will be cheaper than a conventional
vehicle. We assume a cost of EUR 3,000 per year.
• The salary of delivery staff / backup driver for emergency remains the same at EUR
45,500 per year.
• The total costs for vehicle in the semi-automated scenario are therefore EUR 53,500
per year, which is the same as for the manual delivery.
Fully automated delivery:
• For the robo-van which needs further equipment for handling the delivery robots,
we assume the costs to be 70,000. With a linear deprecation over 10 years, the
costs are EUR 7,000 per year.
• Costs for insurance, maintenance and energy are the same as the automated van
in the previous scenario. We assume a cost of EUR 3,000 per year.
• The costs for the delivery robots (e.g., Starship) are highly speculative. According
to Starship’s Head of Data, one robot might cost around USD 5,5003F
4. Adding service
costs and assuming a linear depreciation over 3-4 years, we come to a cost basis
of EUR 2,000 per year. We assume that one robo-van operates with six robots,
therefore the total costs for the delivery robot fleet is EUR 12,000 per year.
• The robo-van operates completely without driver or delivery personnel. However,
remote monitoring personnel will be necessary where it is assumed that one person
can cover five delivery vans (ITF 2017). With an estimated annual salary of EUR
60,000, we obtain EUR 12,000 per year per robo-van.
• Applying these costs, we get EUR 34,000 per robo-van per year.
3 https://www.stepstone.at/gehalt/Paketzusteller-in.html
4 https://sifted.eu/articles/starship-robot-delivery/

LEVITATE | Deliverable D7.5 | WP7 | Final
2
Appendix C Bridge models &
technical overview
Given the bridge models used in bridge modelling in D7.3 (simply-supported beam), the
quasi-static traffic load effects are determined only by the bridge span; they do not depend
on the type of bridge structure (Hu et al., 2021b). On the other hand, the effects of
permanent loads (bridge self-weight) depend very much on the bridge type. In the sub-
use case of truck platooning, the following bridge types were considered (the short
notations are used in the remaining document) (see Figure A, Hu et al., 2021b):
• RCS: reinforced concrete slab,
• PCT: prestressed concrete T-beam bridge,
• PCB: prestressed concrete box-girder,
• CBG: composite bridge: steel girders + concrete slab,
• CBB: composite bridge: steel box-girders + concrete slab,
• SGO: steel bridge: steel girders with steel orthotropic deck,
• SBG: steel box-girder bridge.
a) RCS: Reinforced concrete slab
b) PCT: Prestressed concrete T-beam
bridge
c) PCB: Prestressed concrete box-girder
d) CBG: Composite bridge with steel
girders
e) CBB: Composite bridge with steel box-
girders
f) SGO: Steel bridge with girders
g) SBG: Steel box-girder bridge
Figure A: Cross-section schemes of the considered bridge types.

LEVITATE | Deliverable D7.5 | WP7 | Final
3
Given the simple bridge models used in D7.3 (Hu et al, 2021b), the consideration of
different bridge types is reduced to the evaluation of the permanent load (self-weight). In
all cases, apart from the self-weight of the load-bearing elements, additional permanent
loads (road surface, edge beams) are considered with 300 kg per m² of the bridge surface.
All bridges were modelled with a bridge deck width of 10.5 m (incl. edge beams), carrying
two lanes. Table A shows the basic properties of the analysed bridge models. Besides
bridge type and span length, the permanent load µ is listed, as well as the fundamental
resonant frequency f0.
Table A: Basic properties of analysed bridge models
Type
Span [m]
µ [t/m]
f0 [Hz]
Type
Span [m]
µ [t/m]
f0 [Hz]
RCS
15
29.4
5.65
CBG
30
11.6
2.70
RCS
15
22.8
4.17
CBG
35
11.8
2.40
RCS
20
38.2
4.30
CBG
40
12.0
2.15
RCS
20
29.4
3.18
CBG
50
12.5
1.80
PCT
20
16.8
5.83
CBB
40
13.4
2.05
PCT
20
14.9
3.77
CBB
50
14.0
1.75
PCT
25
19.2
4.78
CBB
60
14.5
1.52
PCT
25
16.7
3.17
CBB
70
15.0
1.35
PCT
30
21.6
4.04
SGO
35
6.1
2.61
PCT
30
18.4
2.72
SGO
40
6.3
2.40
PCT
35
24.0
3.49
SGO
50
6.6
2.08
PCT
35
20.1
2.38
SGO
60
6.9
1.84
PCT
40
26.4
3.07
SBG
70
7.0
1.73
PCT
40
21.9
2.10
SBG
90
7.7
1.44
PCB
40
21.1
3.20
SBG
120
9.0
1.14
PCB
50
23.7
2.57
SBG
150
10.5
0.94
PCB
60
26.8
2.12
PCB
70
30.4
1.80
PCB
90
37.1
1.38
Technical overview of modelling HGV platooning effects
Access control
The basic purpose of intelligent access control is to increase the inter-vehicle distance
of truck platoons before entering the bridge (Hu et al., 2021b). The congestion caused by
intelligent access control was evaluated based on the time required to break and reform a
truck platoon, i.e., extending the inter-vehicle distances before the bridge and reclaiming
the platooning distance afterwards. This process takes time and causes delay to the traffic
on the lane where the platoon operates. The delay mainly depends on the length of the
platoon, the change of the inter-vehicle distance and the cruising speed of the platoon.
The process for extending the distance can be regarded as follows. The first truck in the
platoon maintains the cruising speed and all follower trucks decelerate until distance
between the first and second truck reaches the desired distance. Then the second truck
regains the original cruising speed. After the distance between the second and third truck
reaches the desired distance, the third truck regains the original cruising speed, and so
forth. The process of reforming the platooning is analogous. The first truck decelerates
until the gap to the second truck is reduced to platooning distance. Then the second truck
decelerates, and so forth.
Traffic Model

LEVITATE | Deliverable D7.5 | WP7 | Final
4
A traffic model was adopted that was used for evaluation of traffic loads on bridges
(Freundt et.al. 2011) and consecutively for adjustment of load models on bridges. This
model includes 5 truck types (one 2-axle truck type, two 4-axle and two 5-axle truck
types), a crane and a personal car. The intended application of this model is the description
of heavy traffic on intercity highways.
Since the sub-use case intends to give a general analysis of the potential impact of truck
platooning on urban bridges, it is sufficient to use simplified bridge models. In the
simulation, simply-supported single-span bridges were considered. The bridge is modelled
as a single beam supported at both ends, with free rotation. Given the simple bridge
models that were used, the consideration of different bridge types was reduced to the
evaluation of the permanent load (self-weight). In all cases, apart from the self-weight of
the load-bearing elements, additional permanent loads (road surface, edge beams) are
considered with 300 kg per m² of the bridge surface. All bridges were modelled with a
bridge deck width of 10.5 m (incl. edge beams), carrying two lanes (Hu et al., 2021b)
Measured impacts
The traffic flow exerts different forces on the bridge, which must be transferred by the
bridge structure into the subgrade. Usually, the engineers divide the traffic forces on road
bridges into vertical (weight of vehicles) and horizontal (braking, acceleration, centrifugal
force) forces, which are also so defined in the different standardisations like EN 1991-2
(Hu et al, 2021b). The change in traffic composition due to platoons is expected to lead to
higher bridge internal forces, as described in section 3.2 in D7.3 (Hu et al., 2021b).
Three main impacts of these basic forces were measured: the midspan bending moment
and the shear force at the support(s) and the braking force. The Ultimate Limit States
(ULS) of midspan bending moment and shear force and the horizontal force from braking
are the main impacts measured in traffic simulation models. Their values in different traffic
cases are compared.
EN 1991-2 prescribes the consideration of braking and acceleration forces, centrifugal
forces, and lateral forces from skew braking and skidding. Among these forces, the braking
force is the most relevant one in most cases. Therefore, the study focused on the
evaluation of braking forces.
If bridge strengthening is needed, the limit states of bending moment and shear force are
expected to determine the overall strengthening cost in the most cases and the cost
estimates can be used as a first estimate in decision making (Hu et al., 2021b). The
EuroCode recommends the use of load model LM1 in assessment of existing bridges but
allows its reduction using the α_Q factors to account for less demanding traffic
compositions. Assuming that existing bridges fulfil the requirements on their positive
assessment, three cases of bridge resistance levels were considered for the calculation of
rough estimates of strengthening needs (Hu et al., 2021b):
• = 1: Bridge is able to carry exactly 100% of the LM1 load model
• = 0.9: Bridge is able to carry exactly 90% of LM1 load model
• = 0.8: Bridge is able to carry exactly 80% of LM1 load model
Exceedance probability
The impact of simulated traffic is evaluated in terms of the probability of exceeding the
effects of load model LM1. Since new bridges are designed for the loads of load model LM1,
it is assumed that they have the respective load-carrying capacity. The definition of load

LEVITATE | Deliverable D7.5 | WP7 | Final
5
model LM1 according to EN 1991-1 presumes that its exceedance probability in 50 years
is 5%. Therefore, this probability (5% in 50 years) is regarded as the “code level”. The
resulting bridge forces are evaluated in terms of the probability, that they exceed the forces
from Eurocode load models (Hu et al., 2021b). If the probability, that a resulting 50-years-
extreme-value distribution exceeds the force from a Eurocode load model, is above 5%
the structural safety can be regarded as reduced. Higher exceedance probabilities mean
lower structural safety Hu et al., 2021b).
Assumptions
The following assumptions were made in the modelling of traffic flow, in the bridge
assessment, and in the cost estimates (Hu et al., 2021b, p. 21):
Traffic flow on bridges
• Traffic flow is a random stationary process; evolution of the traffic flow over time is
not considered.
• Vehicle speed is constant and all vehicles in one lane share the same speed.
• Vehicles do not change lanes while on the bridge.
• Most vehicles comply with the prescribed limits of gross vehicle weight. Vehicles
that violate the prescribed limit do so in an appropriate manner – the excess weight
is not very large. That means, a certain percentage of vehicles with gross weight
slightly over 40 tons occurs, but for example a single vehicle with 60 tons does not
(except for special vehicles that have the permit).
• Traffic composition and congestion properties as discussed in Hu et al (2021b).The
distribution of the number of vehicles between lanes is assumed as 80%-20%
(Freundt et.al. 2011) for a two-lane urban highway in the case of low traffic
intensity.
• Braking scenarios occur always in one lane only; the case that an obstacle spanning
more than one lane occurs, is not considered, similarly to Eurocode.
• When a vehicle starts braking, the vehicles behind it start braking at the same time
(driver reaction time is neglected).
• Each vehicle brakes with constant deceleration and the distance to previous vehicle
at the end of braking manoeuvre is close to 0.
• First vehicle decelerates with 1= 5.04 /²