Environmental Impact Assessment: Automated Minibuses for Public Transport

Abstract

This chapter examines the environmental impacts of integrating automated minibuses into public transport systems, focusing on the trade-offs between energy savings and increased energy demand due to connectivity and automation features. A comprehensive life cycle assessment (LCA) is used to analyse the potential environmental benefits and problems of automated minibuses. The results show that the environmental benefits of these vehicles depend significantly on the electricity mix for charging, passenger occupancy, and vehicle utilisation throughout their lifetime. In particular, scenarios with renewable energy sources, high passenger occupancy, and optimal vehicle utilisation show the significant potential of automated minibuses to contribute to a more sustainable transport system. This study highlights the importance of holistic considerations, including technological, operational, and infrastructural factors, in maximising the environmental benefits of automated minibuses in urban mobility. It provides a nuanced understanding of the conditions under which automated minibuses can act as a catalyst for the transition to greener public transport.

1 Introduction

The following chapter presents an environmental impact assessment concerning the deployment of automated minibuses in public transport systems. Therefore, we present a study of the potential energy demand and savings of automated driving and an environmental life cycle assessment for automated electric minibuses in two subchapters. Another subchapter summarises the results into environmental indicators, which are used for the sustainability assessment in Chap. 17 of this book.

The main findings of this chapter are:

• Predictive, adaptive, and information sharing through vehicle communication with infrastructure and other vehicles improves vehicle braking performance and consequently energy consumption. However, a highly connected vehicle means more processing required by the infrastructure, remote, or cloud servers which may outweigh the V2X sustainability.

• Based on results from the AVENUE pilots, the automated driving components for an automated minibus driving at 30 km on an 8-hour day require 82.1 Wh km^-1 (304.4 W x 8 h/30 km) or 15.6% of the total energy use of 520 Wh km^-1.

• The data transmission and energy consumption for 3GPP and 5GAA use cases were estimated for automated driving connectivity for 11 scenarios: platooning, sensor and state map sharing, remote driving, lane change, infrastructure-based perception of environment, collision avoidance, collective information sharing, see-through for passing, emergency trajectory alignment, intersection crossing, and cooperative driving.

• Significant potential in energy savings can be achieved in particular from intelligent route optimisation and velocity control.

• Data on energy saving for predictive functions are presented for selected cases based on literature. Among the functions, the eco route planning and traffic light assistant are cited for being urban scenarios that require little exchanged information between the vehicles and the infrastructure. This makes them very promising candidates for real energy savings achieved through the implementation of automated urban mobility.

• The energy efficiency for exchanging data within the automated minibuses ecosystem depends on the number of connections and the advancement of the deployed technologies.

• The cooperative V2X is undoubtedly the key sustainable communication mode and plays an important role in energy demand.

• A fixed route or a mature on-demand service would have different energy consumption due to different numbers of involved servers.

The life cycle assessment (LCA) of the automated minibus shows that the automated technologies in the automated minibuses, as deployed in the AVENUE pilots, are around 5% of the total energy used. When considering the near-future use case, the study points that 59% of the automated minibuses impact stems from the use phase, while component production accounts for 39%. The use phase climate impacts are mostly due to the burning of fossil fuels to produce the electricity required for driving the automated minibus. The global warming potential for each pkm is 78 g CO2eq.

The assessment of the automated minibus based on the environmental indicators shows that at the current stage, the automated minibuses face challenges to be deployed as an environmentally friendly mode of transport. These results are confirmed by the LCA study, pointing that the automated minibus at the current deployment does not show significant environmental benefits, but future use cases are likely to improve substantially. In addition, the automated minibuses qualification as environmentally friendly depends on many factors such as occupancy, vehicle speed, mileage, and lifetime. Taking into consideration the perspective on the mobility system, the automated minibuses are seen as a complementary service in public transport. In combination with door-to-door, on-demand, and driverless services, the automated minibuses are expected to improve and strengthen public transport, hence bring benefits by reinforcing shared, multi- and intermodal mobility as well.

2 Assessment of Energy Demand of CAV/AV

Automated driving technologies are likely to reduce energy demand for driving compared to traditional vehicles, e.g. by functions such as platooning and eco-driving. However, the increased demand for data transmission and processing might increase energy demand. This part elaborates on both sides, the energy demand side of automated driving technologies and the potential energy savings of automated driving. Such an assessment is crucial from an environmental perspective as energy increase or decrease effect the overall environmental performance of automated vehicles significantly. In this chapter, we would be mainly referring to automated vehicles as CAV (connected automated vehicles) instead of AV, as we aim to address various aspects of connectivity in automated driving.

2.1 Energy Demand of Automated Driving Technologies

Throughout the scientific literature, researchers debated the energy demand of CAV to be either greedy or efficient depending on the implemented automation units, Internet technologies, and deployed services, though they clearly agreed on the direct impact of the data exchange on the overall energy consumption. This section sheds the light on the key publications that showcased the vehicle connectivity and data transfer impact on energy consumption. It also reported the efforts from literature in translating the exchanged bytes and bits into energy units.

Noroozi et al. (2023) conducted a systematic literature review on the automation impact on energy consumption. They review recent literature focusing on how energy consumption of automated vehicles is influenced by advancements in powertrain operation and driving patterns. The papers they found were organised based on various factors affecting power consumption, but it’s noted that existing studies vary greatly in their design and implementation, suggesting a need for more specific and comprehensive research to accurately benchmark the energy consumption impact of vehicle automation.

Liu, Tan, et al. (2019) provided a quantitative study on the negative effects of smart vehicles on energy consumption. The authors draw attention to the fact that automated and intelligent vehicles are equipped with computing devices, advanced sensors, controllers, and actuators, in combination with connecting communication technologies, resulting in higher energy consumption compared to conventional vehicles. The authors suggest that computing platform performance, connection strength, and radar performance are the three main factors impacting the energy consumption of CAV. Their study led to the assessment of fuel consumption per 100 km for different levels of automation—primary, intermediate, and advanced intelligence (corresponding to SAE levels 3, 4, and 5 accordingly)—and the identification of different factors that potentially influence vehicle’s consumption costs. The study has shown an increase in energy consumption depending on the level of intelligence with 0.78 L/100 km in primary intelligent vehicles, 1.58 L/100 km in intermediate intelligent vehicles, and 1.86 L/100 km in advanced intelligent vehicles.

Chen et al. (2021) explore how vehicle automation affects fuel consumption for electric and gasoline powered vehicles with automation levels 0, 2, and 5 across different scenarios, by analysing design changes and performance impacts. Results suggest energy savings of 4–8% in an optimistic level 5 scenario and a 10–15% increase in fuel consumption in a pessimistic scenario. It also notes that power demand varies between urban and highway driving, with inertial power dominating in urban conditions and aerodynamic power in highway conditions.

Song et al. (2023) analyse the impacts of level 2 automation on traffic efficiency and energy consumption of expressways, considering travel time, road capacity, and energy consumption. The study proposes a benefit evaluation framework and uses microscopic traffic simulation software for experiments. Different market penetration rates and traffic flow statuses are considered, along with dedicated lanes for CAVs. Results show that they generally save travel time and reduce energy consumption per vehicle, with positive economic benefits increasing with larger market penetration rates.

Gawron et al. (2018) present a life cycle assessment (LCA) of CAV sensing and computing hardware with SAE level 4 of automation exploring the potential energy and greenhouse gas (GHG) emission impacts of CAV based on six scenarios. Three of the scenarios simulate sensing and computing hardware configurations of Tesla Model S, Ford Fusion (AV test vehicles), and Waymo’s Chrysler Pacifica respectively integrated into an internal combustion engine vehicle (ICEV), and the other three scenarios simulate the hardware configuration on a battery electric vehicle (BEV). They reported that the additional hardware resulted in an increase of 3–20% of energy consumption compared to conventional vehicles. However, when considering the automated driving functions (e.g. eco-driving, platooning, and intersection connectivity) facilitated by the additional hardware, the net result is up to 9% of energy (and emission) reduction based on the Tesla and Ford hardware configuration. The authors claim that data transmission is one of the four factors contributing to an increase of energy consumption. Their research studied data transmission over 4G wireless networks, which was estimated to 1.4 MB/mile and to a consumption of 1.25 MJ/GB.

Figure 13.1 depicts Gawron et al. (2018) life cycle energy estimation for a medium CAV.

Fig. 13.1

Medium CAV life cycle energy consumption (reprinted (adapted) with permission from (Gawron, James H./Keoleian, Gregory A./Kleine, Robert D. de/Wallington, Timothy J./Kim, Hyung Chul (2018). Life Cycle Assessment of Connected and Automated Vehicles: Sensing and Computing Subsystem and Vehicle Level Effects. Environmental science & technology 52 (5), 3249–3256. https://doi.org/10.1021/acs.est.7b04576}. Copyright {2024} American Chemical Society)

Pihkola et al. (2018) evaluated the environmental impact of mobile access networks and sustainability of services within the IoT (Internet of Things) ecosystem using the LCA methodology. In their study, the authors constructed a trend of kWh per transferred gigabyte where they linked the network electricity consumption to the transferred data within the network. However, their computations were limited to the 4G mobile network consumption in Finland that can be extended to any IoT model.

Greenblatt and Shaheen (2015) focused their research on environmental impacts of CAV’s on-demand services, which reduce the vehicle ownership, the number of households owning a car, and the vehicle miles (kilometres) travelled.

Based on some environmental background data from Gawron et al. (2018) and Baxter et al. (2018), we calculated the specific energy usage of fully automated minibuses used within AVENUE. The vehicle components required for automated driving are of particular interest and listed in Table 13.1. For each component, reference technologies and nominal power figures have been derived from the component manufacturer’s information. In total, automated driving components in automated minibuses demand about 300 W. According to Gawron et al. (2018), the additional power required for a medium-sized, automated vehicle sums up to 240 W, while Baxter et al. (2018) state that 200 W is caused by the sensor layout for a mid-sized vehicle. The higher value of this study might be explained by a more detailed list of components in comparison to the studies by Gawron et al. (2018) and Baxter et al. (2018), which focus on primary hardware technology, such as sensors, radars, cameras, LiDARS, computers, and location detection.

Table 13.1 Nominal power of automated driving components installed in one automated minibus

In the current trial mode of the AVENUE project, the distance-weighted average speed of all sites equals 11.4 km/h (this calculation is based on Table 13.4). Assuming that all automated components run at full nominal power, the energy demand for 11.4 km of driving is 304.4 Wh, which equals 26.7 Wh/km. The distance-weighted average of the trial site’s energy demand is 554 Wh/km. Hence only 4.8% of total energy demand is caused by the use of components required for automated driving.

2.2 Energy Demand of CAV Connectivity

To compute CAV’s energy consumption, it is important to consider the connection operations and strength. Such calculations depend on the connection hardware and its related power and time, the vehicle automation level, and the amount of exchanged data (Liu, Tan, et al., 2019). The following subsections present an in-depth classification of the different sources of consumption related to the vehicle data exchange. It addresses the energy consumption of the vehicle communication to external servers or devices, including the vehicle communication to other vehicles (V2V), to the infrastructure (V2I), to the cloud (V2C), to pedestrians (V2P), and to the grid (V2G).

2.3 CAV Connectivity Technologies

The CAV’s connectivity is built through multiple channels (El-Rewini et al., 2020): radio (AM/FM/DAB/RFID), Wi-F- (IEEE 802.11), Bluetooth, cellular (3/4/5G), and bidirectional communication (IEEE 802.11p, DSRC, WAVE), or using IoT networks (IEEE 802.15.4, Zigbee). With the presence of wireless connections, virtual ad hoc networks (VANET) can be spontaneously created among CAVs, leading to V2V communication. With the increase of modern concepts, infrastructure (V2I) and additional devices (V2X) are required to assist the VANETs for data storage and data transmission for long distances (El-Rewini et al., 2020; Lee & Atkison, 2021). V2X also compasses cloud (V2C) and grid (V2G) communication in addition to any further devices or peripherals interacting with the vehicle such as smartphones (V2P), car keys or Bluetooth devices (Domínguez et al., 2019). Being hyper-connected by nature, CAVs sustainability embeds the inherited energy consumption of data transfer technologies through their communication networks. Furthermore, the V2X network topology requires further processing either at the RSU (roadside unit) or at a remote server, which would cost additional computation resources, cause delays, and hence increase the energy consumption (Belogaev et al., 2020).

In academic literature, multiple approximations and estimations are used to quantify wireless cellular networks’ energy consumption. In 2020, the 4G energy consumption was assessed to be around 0.1 kWh/GB (Andrae & Edler, 2015; El-Rewini et al., 2020; Pihkola et al., 2018), including the network, data centre computations, and data storage. Masoudi et al. (2019) added that 5G networks promise higher efficiencies (up 1 Mbit/J) to the energy consumption within the IoT ecosystem. Further research recommends new IoT technologies such as Zigbee for efficient energy consumption (Gheorghiu & Minea, 2016).

To improve the vehicular energy consumption related to data transmission, researchers have studied some protocols for higher energy efficiency. Pihkola et al. (2018) highlight that new efficiency measures that have been deployed within the last decade lower the energy consumption of Internet data transmission to 0.1 kWh/GB in 2020 instead of 12.35 kWh/GB in 2010. Dong et al. (2016) proposed an optimum cluster management method to reduce the V2V transmission power while using DSRC and LTE. Passafiume et al. (2020) proposed a battery-less transponder plugged to an RSU supporting the V2I communication.

2.4 Data Transmissions of CAV

In the last decades, various consortia have been active in defining and studying various V2X data transmission technologies and protocols and their application in real-world use cases of automated driving with 3GPP and 5GAA alliances have been the most active in this domain. An early 3GPP report (3GPP, 2015) defined data transmission use cases using 4G-based long-term evolution-vehicle (LTE-V) and 3GPP (2018), 3GPP (2019), as well as 5GAA (2020) and defined data transmission use cases using 5G based C-V2X technologies. More advanced real-world scenarios with real traffic situations are being evaluated in various EU-funded telecommunication projects such as METIS 2020 and METIS 2020-II and V2X projects such as 5GCAR, 5GCroCo, and 5G CARMEN. It is not clear, though, whether these projects measure data transmission levels and whether such measures are accessible. Furthermore, simulation platforms such as Fraunhofer’s Simulation Platform for Cellular V2X Fraunhofer IIS (2021) may also be useful as tools to collect data about real-world traffic situations.

Based on 3GPP and 5GAA use cases and data and the summarisation of the use cases as conducted by Kanavos et al. (2021), we calculate the energy consumption for V2X for various automated driving functions. We employed the 4G energy consumption measure of 0.1 kWh/GB (Andrae & Edler, 2015; El-Rewini et al., 2020; Pihkola et al., 2018) to calculate the average energy usage of automated driving connectivity for each of the eleven scenarios.

Table 13.2 presents the estimated data transmission rates for each use case based on 5GAA and 3GPP estimations along with our calculated energy consumption for each use case. Kanavos et al. (2021) point out that 3GPP data transmission estimations are on the higher level to prepare the telecom networks to support extreme cases of each use case, while 5GAA has more modest values based on more efficient implementation of the services. We believe more in the modest values of 5GAA as we believe that every market-conscious car manufacturer would strive to lower their vehicle’s energy consumption and is more inclined to develop energy-efficient implementation of their automated driving services as long as it does not affect the driving security.

Table 13.2 Data transmission and energy consumption for different automated functions

2.5 Potential Energy Savings Through Predictive Driving Functions

Automated and connected driving functions do not only control the perception, decision-making, and driving command execution to move the vehicle in a safe and convenient way, but they also enable energy savings by optimising the driving route selection, motion planning, and powertrain operation. This is an important aspect when considering the connectivity options to be implemented since a few additional data provided by the infrastructure plus a few megabytes of additional bandwidth needed might pay off significantly in terms of energy saved.

Hu et al. (2017) and Connor et al. (2021) discussed the direct correlation between the vehicle connectivity, velocity, and the battery consumption and their impacts to the environment. The authors studied real-world driving scenarios for electric buses using V2I and V2V technologies. According to their findings, the V2I and V2V communications provide energy savings that are up to 27% of battery cost reduction. Bo et al. (2019) also asserted the beneficial impact of V2I to have an optimal energy control. The US Department of Energy, through the NREL study (Stephens, 2016), reported 2–6% fuel savings by adopting the V2I smart intersections.

The predictive functions typically combine models of the vehicle and its powertrain with external data such as the upcoming driving route characteristics and traffic conditions to predictively control the vehicle. Examples of such functions are provided in Table 13.3 below, including published data on corresponding energy savings. Considering the literature for energy-saving potentials indicated in Table 13.3, it is noteworthy that:

• The achievable energy savings are generally heavily dependent on the defined vehicle and use cases, resulting in wide ranges of savings typically being published for similar functions by different authors.

• The energy savings also strongly depend on the particular baseline to which they are calculated, which often consists of different types of human drivers or non-predictive control algorithms.

Table 13.3 Predictive automated driving (AD) functions and their energy-saving potential

Nevertheless, the available published results demonstrate that a significant potential in energy savings can be achieved in particular from intelligent route optimisation and velocity control.

2.6 Discussion and Implications for Environmental Impact Assessment

The nexus of data processing and exchange within the automated driving landscape raises challenges to consider while assessing CAV’s sustainability. To this end, the energy consumption related to CAV’s data transfer depends on wireless technologies, the cooperative communication modes, and the implemented services. Automated minibuses may support different types of internet connections which result in large differences in energy consumption. The driverless wireless network can vary from 4G and 5G to DSRC, which definitely impacts the amount of exchanged data and hence the vehicle energy consumption (Masoudi et al. (2019). Cooperative V2X is expected to be the favourable communication mode with regard to energy use, according to Bo et al. (2019) and Stephens (2016). Predictive, adaptive information sharing provided through vehicle communication with infrastructure and other vehicles improved vehicle’s braking performance and consequently its energy consumption. However, a highly connected vehicle also requires more processing within its infrastructure, remote, or cloud servers. This may outweigh the V2X sustainability. A fixed route or a mature on-demand service would not have comparable energy consumption as the number of involved servers and processing will not be proportionate (Greenblatt & Shaheen, 2015). As with every exchanged data within the automated minibus ecosystem, the energy efficiency can fall over to either high or low energy demand depending on the number of connections and the advancement of the deployed technologies.

Although the reported potential energy savings through predictive driving functions differ between studies, it seems evident that the savings are likely to counterbalance and even overcompensate the energy costs associated with the communication modules. It is ‘common sense’ in the European research community that it will be impossible to implement large-scale automated urban mobility in a safe way without infrastructure support; both in-vehicle and infrastructure communication equipment will already be there regardless of the use of predictive functionalities. Consequently, the additional cost for employing predictive functionalities is merely the additional bandwidth used by the additional data which need to be transmitted. Especially the first two functions mentioned in Table 13.3, eco route planning and traffic light assistant, apply to urban scenarios and require only a little information to be exchanged between the vehicles and the infrastructure, which makes them very promising candidates for real energy savings achieved through the implementation of automated urban mobility.

3 LCA (Life Cycle Assessment) Model

This part summarises the final results of a life cycle assessment (LCA) study on automated minibuses, which was also published as an analysis in Transportation Research Part D (Huber et al., 2022).

The LCA study investigated the environmental impacts of automated minibuses to be integrated into the public transport of cities, guided by the following research questions: (1) Which environmental impacts are associated with the operation of an automated minibus? (2) What are the main drivers of these impacts, and how can these be reduced? (3) What conclusions can be drawn from these findings for the role of automated minibuses in future public transportation systems?

3.1 Goal and Scope of the LCA Study

The common functional unit of a passenger kilometre (pkm) enables the comparison between automated minibuses and other modes of transport.

The automated minibus under investigation is 4.75 m long, 2.11 m wide, and 2.65 m high, weighs 2400 kg, and can carry 15 passengers (11 seated and 4 standing) with a maximum speed of 25 km/h.

The automated minibuses operation involved fixed routes for public transport, with trials on on-demand, door-to-door, and pooling options for passengers to request the vehicle to come to a designated pick-up location through the use of a mobile application (Navya, 2018). The automated minibuses are intended for use in a public transport system, not as a replacement for individual vehicles.

The assessment adopts a cradle-to-grave approach, encompassing the main life cycle phases recommended by Duce et al. (2013), namely, component production, vehicle assembly, use, and end-of-life treatment. The production of components has been divided into three categories: battery manufacturing, the production of automated driving components, and the production of other bus components. In order to accurately assess the environmental impact of these components throughout their life cycle, it is necessary to account for the material and energy inputs and outputs at each stage of the LCA. The following six impact categories were employed: acidification, climate change, eutrophication, ozone depletion, photochemical ozone formation, and resource depletion. A control calculation showed that the five remaining environmental impact categories presented similar results in terms of their impact.

3.2 Life Cycle Inventory and Data Collection

The life cycle inventory encompasses all environmental impacts associated with the system under investigation, including the inflow and outflow of materials and energy. A preliminary generic automated minibus model has been constructed using existing literature data (Gawron et al., 2018; Hawkins et al., 2013; Majeau-Bettez et al., 2011). This model was then refined by incorporating primary data obtained from the automated minibus manufacturer and public transport operators involved in the AVENUE project.

The primary data for this study were obtained from the demonstrator sites of the AVENUE project, where an automated minibus manufacturer provided information on vehicle components (such as weight, functions, and nominal power) and transport operators provided data on the use of these vehicles in public transportation. The data collection took place between 2019 and early 2021 in an iterative manner.

The sources of information regarding component production, including batteries and automated driving components, are Majeau-Bettez et al. (2011), Hawkins et al. (2013), Gawron et al. (2018), and Moreno Ruiz et al. (2020). The manufacturer supplied all the components and their respective weights for the vehicles in this study. The total weight of all the automated minibus components used is over 99% of the total automated minibus weight, which meets the requirements of standard LCA. The assembly of these components was based on data from Majeau-Bettez et al. (2011), Hawkins et al. (2013), and Gawron et al. (2018) which represents industry-scale assembly of electric passenger cars. Additional secondary data was sourced from well-known LCA databases, chiefly ecoinvent 3.7 (Moreno Ruiz et al., 2020). A comprehensive overview of all components and materials required for vehicle assembly, along with their life cycle inventories (LCI) and details on primary and secondary sources, is provided in the study’s supplementary material.

Table 13.4 presents data on the use phase of automated minibus service on fixed-route buses in five trial sites from September 2018 to January 2021. The data reveals differences in the average speed, expected annual mileage per shuttle, average vehicle occupancy, and average and minimum energy demand. It should be noted that the extremely low average occupancy is due to the experimental nature of the trial sites, where vehicles are also used for functional and technical testing purposes.

Table 13.4 Automated minibus use phase data

Umberto^® software was utilised to create a product life cycle model and analyse the results. The comprehensive automated minibus LCA model comprises 198 processes and 42 subnets across 4 hierarchical levels. A portion of the overall model is depicted in Fig. 13.2.

Fig. 13.2

Automated minibus life cycle model main model with subnets for battery production (a), production of all further components (b), and production of automated components (c). Sankey diagrams depict global warming potential in g CO2eq per pkm for the automated minibus near-future use case. In the Petri-net-based material flow network approach underlying the Umberto LCA software, blue squares represent processes or subnetworks, and circles represent input points (green), output points (red), and connection points (yellow). All values have been rounded to one digit

3.3 Scenario Setting

Using primary data gathered from trial sites, we have generated a near-future use case and worst- and best-case values for scenario analysis. These values can also be combined to form an ideal use case. The relevant parameters for these scenarios include the automated minibuses expected lifetime, annual mileage, average passenger occupancy, energy demand, energy source, and the used battery LCA data. The parameter settings are described in detail and summarised in Table 13.5.

Table 13.5 Parameter setting for near-future use case and scenario analysis

According to the manufacturer and transport operators involved in this study, the battery lifetime can be used as a proxy for the overall automated minibus lifetime. If one charging process occurs per day and the automated minibus operates for 5 days per week, the estimated battery lifetime is 7.7 years. This has been rounded up to 7 years to account for probable losses and reduced efficiency as the battery ages (Hadjipaschalis et al., 2009; Oliveira et al., 2015). The high cost of batteries for electric vehicles has led to the belief that the lifespan of an automated minibus is the same as the lifespan of its battery, although it is acknowledged that the rapid advancement of automated minibus technology may result in obsolescence and decommissioning before the end of the battery’s lifespan. Conversely, some studies on batteries for electric vehicles suggest longer lifetimes, such as 10 years (Deng et al., 2017). The average lifetime of 7 years is obtained through scenario analysis, which ranges from 3 to 10 years in some cases.

The near-future use case is based on an annual mileage of 20,000 km, while the scenario analysis considers a range of mileages from 5000 km (which is roughly the average of other trial sites in Table 13.5) to 36,500 km, assuming a daily operating distance of 100 km.

The expected average occupancy for the near-future use case is five passengers, which is above the current trial data but aligns with transport operators’ economic feasibility expectations. The scenario analysis considers a worst-case value of one passenger on average and a best-case value of ten passengers on average.

The energy demand for the near-future use case is 554 Wh/km, which is the distance-weighted average of the trial site’s typical energy consumption. This is within the manufacturer’s specifications of 520 Wh/km, which was measured with one person on board, traveling at an average speed of 6.6 km/h, and an outside temperature of 30 degrees Celsius, while the vehicle’s interior was cooled down to 16 °C. The energy demand for the autonomous mobility system encompasses all automated components, passenger interaction components, and the electric driving components. Given that various factors such as speed, temperature, weight, and others can impact automated minibus energy consumption, it is crucial to conduct scenario analysis. The distance-weighted average of the minimum energy demand at the trial site (332 Wh/km) represents the best-case scenario, while the highest average energy demand at a trial site (780 Wh/km) represents the worst-case scenario for scenario analysis.

For the near-future use case, we have assumed a European electricity mix with 418 grams of CO2 equivalent per kilowatt-hour (kWh) (ecoinvent 3.7 databases with low voltage electricity market datasets for Europe, Poland, and Norway). In contrast, the worst-case scenario for scenario analysis is a mostly fossil electricity mix with 1037 grams of CO2 equivalent per kWh, while the best-case scenario is an almost entirely renewable electricity mix with 23 grams of CO2 equivalent per kWh. Battery production has been modelled using detailed data from the literature.

4 Results

This section will provide an analysis of the environmental impacts of additive manufacturing, as well as the results of scenario simulations and assessments of the automated components. Additionally, a comparison of automated minibuses with other modes of transportation will be presented to provide a more comprehensive understanding of the findings.

4.1 Life Cycle Impacts of Automated Minibuses

Table 13.6 displays the environmental impact per passenger kilometre for the selected impact categories, disaggregated into the life cycle phase components of production (separated into battery, automated, and all other bus components), vehicle assembly, use, and end of life.

Table 13.6 Environmental impacts for 1 pkm of automated minibuses in a near-future use case and in comparison to an ideal use case; climate change (global warming potential 100 years) measured in g CO2eq (carbon dioxide equivalents); acidification in mol H + eq (proton equivalents); eutrophication in kg Peq (phosphorous equivalents), ozone depletion in kg CFC-11 eq (trichlorofluoromethane equivalents), photochemical ozone formation (POCP) in NMVOCeq (non-methane volatile organic compound equivalents), resource depletion in kg Sbeq (antimony equivalents)

In the near-future use case, each pkm has a global warming potential of 78 g CO₂eq (as shown in Fig. 13.2 with a Sankey visualisation of the global warming potential within the automated minibus life cycle model). The majority of this impact, at 59%, stems from the use phase, while component production accounts for 39%. The primary climate impacts resulting from the use phase are primarily caused by the combustion of fossil fuels to power the operation of the automated minibus. For this reason, the use phase also accounts for 54% of the overall acidification potential. In all other environmental impact categories, the production of components either dominates moderately (eutrophication potential and photochemical ozone depletion potential) or by a wide margin (ozone layer depletion potential and resource depletion potential). The assembly and end-of-life phases of the product have no significant impact in any of the chosen environmental impact categories, which is why their modelling is based on average literature data.

The aforementioned Table 13.5 outlines the optimal parameter settings for the best possible scenario. By utilising these settings, a practical example with high automated minibus lifetime mileage, high passenger occupancy, low energy demand, and renewable energy supply can be modelled and evaluated. The total environmental impacts of this ideal scenario are presented in Table 13.6, revealing substantial reductions in environmental impact across all categories, with 80–91% reduction in resource depletion and climate change impact.

4.2 Scenario Analysis

The scenario analyses have been executed for all the parameters listed in Table 13.5 and for every environmental impact category under consideration. The findings of the scenario analyses, depicted in Fig. 13.3, demonstrate the paramount importance of non-technical parameters in determining the overall environmental impact of automated minibuses in terms of resource depletion and global warming potential. The depletion of resources and the impact of climate change are significantly affected by low passenger occupancies and low annual mileage, whereas the energy demand of the vehicle has only a limited impact. Additionally, the energy source used to charge the automated minibus battery is the most important factor in reducing the climate impact. Operating the automated minibus within a fully renewable energy system can reduce the global warming potential by 58%.

Fig. 13.3

Scenario analyses for resource depletion potential and global warming potential (near-future use case = 0%)

The environmental impact category of climate change demonstrates the disparate effects of different parameters on the production and use phases (see Fig. 13.4). While the vehicle’s lifetime and mileage only affect the production phase, the energy demand and energy mix influence the use phase. Occupancy affects both the production and use phases equally. The relationship between the use phase and the production phase, as shown in Fig. 13.4, exhibits strong fluctuations. In extreme cases, such as very high mileage and a high fossil energy mix, the use phase dominates. Conversely, when renewable energies are used, the importance of the use phase decreases from a climate perspective, and the importance of production for the overall result increases.

Fig. 13.4

Scenario analysis (global warming potential) comparing automated minibus production and use phase (near-future use case = 100%)

4.3 Impact of Automated Components

The focus of the study lies in the components necessary for automated driving. Table 13.6 indicates that the production of these components has a minimal impact on environmental performance, accounting for less than 2% in all impact categories. Additionally, the energy consumption of these components is of importance, despite the shift towards renewable electricity sources. The use phase’s significance is reduced in the long run, but energy demand and energy mix remain critical factors in overall performance in the present. Table 13.7 provides a list of reference technologies and their corresponding nominal power figures for all the components used in automated manual driving (for a more detailed table that includes manufacturers, models, and Internet sources, please refer to the supplementary material). In summary, the components used in automated manual driving require approximately 300 W of power in total.

Table 13.7 Nominal power of automated driving components installed in one automated minibus (Light detection and ranging sensors (LiDARS), GNSS, GPS)

According to Gawron et al. (2018) and Baxter et al. (2018) the power requirements for a medium-sized, automated vehicle are estimated to be 240 W and 200 W, respectively. The higher value of the present study might be explained by a more comprehensive list of components compared to Gawron et al. (2018) and Baxter et al. (2018), which primarily focus on primary hardware technology, such as sensors, radars, cameras, light detection and ranging sensors (LiDARS), computers, and location detection.

Table 13.7 presents the average speed at the various trial sites. The distance-weighted average speed of all sites is 11.4 km/h. If all automated components are running at their full nominal power, the energy consumption for a distance of 11.4 km is 304.5 Wh, which translates to 26.7 Wh/km. The average energy demand of the trial sites is 554 Wh/km. Therefore, only 4.8% of the total energy demand is due to the use of components required for autonomous driving.

4.4 Contextualisation

With a high degree of confidence, the environmental impact per pkm of near-future and ideal future applications of automated minibuses is significantly lower than the current trial cases. To compare its performance with other modes of transportation, the automated minibuses’ climate change impact per km is contrasted against literature values of other vehicles (Table 13.8). For all vehicles, the impact is calculated for off-peak, average, and peak operation. The average number of passengers per vehicle used in the calculation is 1.58, based on a study by Chester and Horvath (2009). Figure 13.5 shows the climate change impacts of all transportation modes, including the automated minibuses near-future and ideal use case.

Table 13.8 Climate impacts, lifetime mileages, and passenger occupancies for various individual and public transportation vehicles (based on [1] Puig-Samper Naranjo et al., 2021; [2] Gawron et al., 2018; [3] Kemp et al., 2020; [4] Nordelöf et al., 2019; [5] this paper): abbreviations: Ind. individual, ICEV internal combustion engine vehicle, HEV hybrid electric vehicle, BEV battery electric vehicle, BECAV battery electric connected and automated vehicle, ICECAV internal combustion engine connected and automated vehicle, SUV sports utility vehicle, BEB battery electric bus, PHEB plug-in hybrid electric bus, HEB hybrid electric bus, AM NF automated minibus near-future use case, AM ideal automated minibus ideal use case

Fig. 13.5

Climate impact of different transportation modes in g CO₂eq per pkm (own compilation, based on [1] Puig-Samper Naranjo et al., 2021; [2] Gawron et al., 2018; [3] Kemp et al., 2020; [4] Nordelöf et al., 2019; [5] this study, all abbreviations are detailed in Table 13.8)

Compared to other forms of public transportation, the near-future use case of automated minibuses has higher climate impacts per pkm, with the exception of the comparison to the average operation of diesel buses. However, it should be noted that all other forms of public transportation use larger buses with higher passenger numbers for peak, average, and off-peak operation. The ideal use case of automated minibuses performs better than any other form of transportation, demonstrating the significant potential for EMA to improve the environment further and optimise it. It should be viewed with caution, as no ideal use case was calculated for the other forms of transportation, and a renewable energy mix would also have a positive effect on all other battery-electric and hybrid vehicles.

4.5 Discussion and Limitations

The present LCA identifies the key factors that influence the environmental impact of automated minibuses. Similar to electric vehicles in general (as noted in Helmers et al., 2017), the energy consumption and electricity mix used for charging batteries have a significant impact on the vehicle’s climate performance. Moreover, systemic factors, such as the utilisation of the vehicle in terms of annual mileage and passenger occupancy, also play a crucial role. Therefore, whether automated minibuses can be considered environmentally friendly depends on various factors. While an infrequently used automated minibus with few passengers is highly unlikely to have a favourable environmental impact, a heavily used automated minibus that is also fully utilised in terms of its passengers can outperform other transportation methods in terms of environmental benefits. Based on measurement data from operating automated minibuses, a near-future use case has been defined that is considered to be very achievable in the coming years. This use case already shows a very good environmental performance under the aforementioned framework conditions.

The use case calculated in an optimal manner also highlights the potential environmental benefits of automated minibuses, as long as all conditions and parameters are optimal. The most environmentally intensive phases of an automated minibuses life cycle are the production of components and driving, and the components required for automated driving play a minor role, contributing less than 2% to production-related climate impacts and approximately 5% to driving-related climate impacts.

The LCA results presented here are promising, particularly because they indicate the potential for achieving higher mileages in the short term, and there is a positive attitude (goodwill) expressed by potential users towards the automated minibus (Korbee et al., 2024). This study also has some limitations that are important to consider. For example, the data used was collected from pilot sites provided by public transport operators and vehicle manufacturers. These sites have an experimental nature and were affected by COVID restrictions, which limited the amount of data collected and the number of passengers in the vehicles. As data accuracy is a recurrent concern in LCAs of emerging technologies (Arvidsson et al., 2018; Hetherington et al., 2014), future research could focus on utilising longer time-series of data collected from regular operations rather than relying on data from demonstration and trial site operations. To address the potential consequences of innovative technology data, this study suggests utilising scenario development. The purpose of these findings is not to present specific numbers for inclusion in product declarations but rather to offer valuable insights on the key factors affecting the future environmental impact of automated minibuses.

The low relevance of automated driving components is comparable to studies of automated vehicles (Gawron et al., 2018).

Although other studies forecast a much greater effect of automated components on overall energy consumption (Brown et al., 2013; Gawron et al., 2018; Gonder et al., 2016; Grisoni & Madelenat, 2021; inria, 2019; Krail, 2021; Pihkola et al., 2018; Saujot et al., 2017; Wadud et al., 2016).

The deployment of automated vehicles necessitates adaptations in both physical and digital infrastructure, as pointed out by Noussan & Tagliapietra, 2020, and the implementation of vehicle-to-everything (V2X) technologies requires additional technical infrastructure.

The following out-of-vehicle technical infrastructure is necessary for the operation of automated vehicles: roads, sensors to detect special signals, and a long-range wireless network (Liu, Tight, et al., 2019). The examined automated minibus neither transmits large amounts of data to the outside nor requires extensive additional technical infrastructure, which may change in the future. Notably, the study did not consider the long-term benefits of automated driving compared to human driving. Research suggests that connectivity and cooperative technologies will lead to better traffic anticipation, modulated driving, better manoeuvring (Fagnant & Kockelman, 2015), and better ride-matching capacity (Shaheen & Bouzaghrane, 2019), which could further reduce energy consumption. However, the use of automated vehicles may simultaneously reduce energy demand through efficient driving and increase it through additional data processing and transmission (see, e.g. Stephens et al., 2019). The current state of research does not allow for a definite conclusion, and the impact of data processing and transmission on the automated minibus under investigation is currently minor. The implementation of innovation and efficiency improvements in batteries is expected to decrease the negative impact on the environment caused by automated minibuses.

The automated minibus energy demand of 554 Wh per driven km in comparison to other electric vehicles is quite high, as reported by Puig-Samper Naranjo et al., 2021 and Bauer et al., 2015. Since the components for automated driving do not significantly influence this consumption, other factors must be responsible for this high level of energy consumption. Specifically, the heating and cooling of the vehicles, as well as their low speed, are noteworthy. The entire interior of the vehicle is constantly cooled on warm days and warmed up on cold days, which results in high energy consumption. To further reduce the overall energy consumption and environmental impacts, higher speeds, reduced heating and cooling behaviour, and additional energy efficiency measures on the vehicle could be implemented, but these are not within the scope of this study.

4.6 Conclusion and Consequences for Future Mobility Systems

This LCA study highlights the potential of automated minibuses as part of public transport. If automated minibuses are utilised extensively in terms of mileage and regularly used by multiple passengers, they offer significant environmental benefits and perform similarly or better than other public transport vehicles. However, it becomes clear that automated minibuses play a specific role in the overall mobility system and cannot replace all other modes of transport. Currently, the performance of automated minibuses, which features low speed, low passenger capacity, door-to-door service, on-demand service, and driverless operation, is seen as complementary to public transport. For example, automated minibuses can cover the ‘first and last mile’ or take over off-peak operations of regular buses, increasing the availability, flexibility, efficiency, and reliability of local public transport, which brings significant environmental benefits, especially when replacing individual motorised transportation. However, there may be some rebound effects if automated minibuses replace walking or biking or lead to more travel due to their convenience and comfort (Grisoni & Madelenat, 2021; inria, 2019; Saujot et al., 2017).

Taking into account the environmental and sustainability science perspective, this study highlights the limitations of LCA studies that solely concentrate on individual vehicles and vehicle types. The environmental benefits of automated minibuses are influenced by the individual vehicle’s performance but are largely determined by the vehicle’s utilisation, occupancy, and integration within a comprehensive transportation system. As a result, comparing different automated minibus types or brands is relatively inconsequential in this context.

The findings of this research are relevant to decision-makers at various levels of policy and public transport operators. For the latter, the study offers clear understanding of the environmental benefits and drawbacks of the automated minibus implementation. For policymakers, the study underscores the importance of developing plans and frameworks for the deployment of autonomous vehicles in a timely manner to maximise environmental benefits. A public transport system that integrates automated minibuses at strategic points and is multimodal and flexible appears to be a promising approach.

5 Final Environmental Indicators for Sustainability Assessment of Pilot Sites

The assessment of the pilot sites finalises the analysis presented in the second iteration of the environmental deliverable, and it focuses on the environmental indicators. The assessment presents the data collected from the pilot sites as well as the recent updates of the methodology and results. It serves as background information and data for the final sustainability deliverable 8.12.

The objective of this section is to investigate the environmental performance of the automated minibus through mobility indicators. Sustainability indicators are a powerful tool to simplify, quantify, analyse, and communicate complex information (Innamaa & Salla, 2018; KEI, 2005; Singh et al., 2009). In addition, urban sustainability indicators are fundamental to support target setting and performance reviews and enable communication among policymakers, experts, and the general public (Shen et al., 2011; Verbruggen & Kuik, 1991).

The environmental indicators and respective units of assessment are presented in Table 13.9.

Table 13.9 The environmental indicators and units of assessment

In addition to the environmental indicators, the indicators for the sustainability assessment comprehend the social, economic, governance, and system performance of the automated minibus (Nemoto et al., 2021).

Each indicator requires a specific methodology (refer to APPENDIX A), and the value of the indicators is represented on a scale of 1–5, with 5 being considered the best score. For each indicator we:

• 1. Defined a parameter.

• 2. Defined a scale, with minimum and maximum values considering the environmental impacts of main urban modes of transport, e.g. walking, cycling, small and big cars, and bus (freight transport and air transport were not comprehended, for example).

• 3. Calculated the indicator value for the automated minibus according to the demonstrator site.

The results are presented on a spider chart, providing a disaggregated overview of the indicators. This allows for identifying the weaknesses and strengths of each indicator (WBCSD, 2015), also for a comparison between the pilot sites.

The limitations of the assessment concern the innovativeness of the automated minibus. The technology is still in a test and development phase. Hence, the main limitations concern the fact that the pilot projects are restricted to a local/neighbourhood area, and the automated minibuses drive in mixed traffic area at a low average speed (10–18 km/h). The automated minibuses drive on a fixed route (with the exception of ‘Belle Idée’ test site, where on-demand service has been tested), and the safety driver on board the automated minibus is required in case human intervention is required, as well as to report the performance of the automated minibuses in general.

These limitations reduce the performance and usability of the automated minibuses. In addition, the demonstrator sites have been facing constraints due to the COVID-19 pandemic. As a result, there have been interruptions in the pilot tests, and some transport companies have limited the maximum number of passengers to four during certain periods. This factor has a negative impact on the automated minibuses performance assessed by the environmental indicators.

The next section presents the results for the environmental indicators for five AVENUE demonstrator sites: Pfaffenthal and Contern (Luxembourg City), Groupama Stadium (Lyon, France), Ormøya (Oslo, Norway), and Nordhavn (Copenhagen, Denmark). The transport operators provided primary data; therefore, the results and analysis rely on the data presented in Table 13.4 (Chap. 3), which also provided inputs for the LCA study.

An overview of the pilot trials is present in Table 13.10, and the results per site are illustrated in Fig. 13.6, followed by analysis and conclusions.

Table 13.10 Overview of the AVENUE demonstrator sites

Fig. 13.6

Environmental performance of the automated minibuses in the demonstrator sites. The scale ranges from 1 to 5, with 5 as the best score and 1 worst score

The indicators addressing ‘local air pollution’ and ‘local noise pollution’ do not vary from site to site because they are assessed according to the vehicle. As an electric vehicle, the automated minibus has a good score on local air pollution. It is explained by the fact that BEVs in their use phase have zero exhaust emissions, e.g. NOx and PM, and they just emit PM locally from road and tyre and brake wear, like other motor vehicles (European Environment Agency, 2018). The air pollutant emissions for the electricity generation to charge BEV batteries occur in power stations and tend to impact less densely populated areas (ibid). For this reason, the local air pollution emissions are assessed here for the use phase, as they affect cities (more densely populated areas) and consequently cause greater human exposure and potential health damage.

For local noise pollution, the automated minibus as an EV do not differ significantly from ICEV in the usual traffic and from 30 km/h speed. This is due to the fact that ‘the tyre/road noise increases more with increasing speed than the propulsion noise, and therefore the tyre/road noise dominates the propulsion noise at high speeds’ (Marbjerg, 2013). Therefore, the automated minibuses as an EV play a role to avoid local noise pollution for urban traffic during the night in low-speed areas (Jochem et al., 2016). Since the automated minibuses currently run at a low speed of 11–18 km/h, their noise pollution is slightly lower than ICEVs and lower than regular buses.

In all the pilot sites, the automated minibuses scored poorly for ‘low contribution to Climate Change’. This indicator is highly affected by the low occupancy of the automated minibuses, due to the characteristics of the pilots, such as temporary and new services, the newness of the technology, as well as the interruptions of the trials due to the constraints of the COVID-19 pandemic and the reduction in mobility and in the use of public transport. Further, the climate change indicator is affected by the vehicle lifetime, total mileage, and electricity mix, as pointed by the LCA study (Chap. 3). From all the sites, Pfaffenthal (Luxembourg) presents a better performance due to the higher average of vehicle occupancy, while Ormøya (Oslo) and Nordhavn (Copenhagen) present the lowest performance for climate change due to the very low average occupancy and low mileage in the case of Nordhavn.

Likewise, the energy efficiency indicator is directly impacted by the average occupancy of the automated minibus. Therefore Pfaffenthal (Luxembourg) presents a good score (with an average occupancy of 2 or 8 passengers), in contrast to the other sites.

The indicator of renewable energy for the use phase varies according to the share of energy from renewable sources in gross electricity consumption in each country. In this case, Nordhavn (Copenhagen) and Ormøya (Oslo) present a good score since Denmark and Norway have a share of energy from renewable sources in gross electricity consumption of 62% and 100%, respectively, in contrast to 9% in Luxembourg and 21% in France.

In relation to Chap. 2, it is worth noting that the pilot trial at Groupama Stadium (Lyon) comprehends V2I communication, meaning that three traffic light junctions operate in communication with the automated minibus. The V2X communications were not taken into account for the environmental indicators at this stage. However, on a larger scale, vehicle communications and connectivity could contribute to reducing energy impacts in mobility. Lee and Kockelman (2019), for example, pointed out that energy savings resulting from vehicle-to-infrastructure connectivity and smart intersections range from 6% to 30%, thanks to improvements in traffic interactions and better fuel-efficient driving (see more in Chap. 2).

The assessment of automated minibuses based on the environmental indicators point out that the automated minibuses face challenges to be deployed as an environmentally friendly mode of transport at the current stage.

A key factor targeting ‘low contribution to climate change’ is primarily to increase the vehicle occupancy and secondly through technology development, to increase the vehicle speed, mileage, and lifetime. Likewise, by aiming at a better energy efficiency, it is crucial to increase vehicle occupancy. Therefore, it is important that the automated minibuses are deployed in routes in order to cover real gaps in mobility, with more permanent services and good acceptance. And as mentioned previously, the average occupancy of the automated minibuses was also affected by the COVID pandemic, interruptions in the trials, and mobility restrictions.

The automated minibuses, as a BEV, can highly contribute to the reduction in local air pollution, and while targeting the reduction of local noise pollution, the automated minibuses present limited advantages in comparison to regular cars and buses, reducing noise during the night and at low-speed areas.

In the future, the automated minibus has the potential to be deployed as environmentally friendly mobility taking into account technological improvements, better social acceptance and usability, better integration into urban mobility as part of intermodal and MaaS systems, as well as shared and electric mobility.

The sustainability assessment study aims to set goals for the future deployment of the automated minibuses and therefore monitor the progress of the environmental and remaining sustainability indicators towards a more sustainable operation.

6 General Discussion and Conclusion

The different assessments in this deliverable reach interesting results on the current and future performance of automated minibuses. The energy demand analysis of the automated components in the automated minibus shows that the energy efficiency depends on the wireless technologies, the cooperative communication modes, and the implemented services. The automated technologies in the automated minibus, as deployed in the AVENUE pilots, are around 5% of the total energy used. The potential evolution of the energy demand depends on the different types of Internet connections. The driverless wireless network impacts the transmitted data and eventually the overall energy consumption or savings. Moreover, predictive and adaptive driving functions and information sharing in the automated minibus are likely to improve the acceleration and braking processes and hence contribute to overall energy savings. The energy demand also differs if the automated minibus provides an on-demand service. Overall, the energy-saving potential from predictive driving functions is highly likely to outweigh the energy consumption from data transmission energy.

Going beyond the energy analysis related to driving itself, the presented LCA study focuses on environmentally relevant energy flow and materials throughout the life cycle of the vehicle within the public transportation system. The study reiterates findings from other studies that energy consumption and the electricity mix used for charging have significant climate performance impacts. It also shows a low relevance of automated driving components in current deployment circumstances. The LCA study was used to identify factors that contribute to the environmental impact of automated minibuses. The environmental benefits of automated minibuses rely on the utilisation rate and occupancy factor and thus on their integration in the overall transportation system.

The current deployment of automated minibuses does not show significant environmental benefits, but future use cases are likely to improve substantially. The development and assessment of environmental indicators for an overall sustainability assessment corroborate the LCA study’s conclusions. Occupancy, vehicle speed, mileage, and lifetime play an important role in reducing environmental impacts. Automated minibuses are particularly beneficial if they can be deployed to close former public mobility gaps, which would otherwise lead to the use of individual cars.