ASTRA 2018/002

Project details

Duration

09.2018-07.2019

Sponsor

external pageFEDROcall_made Federal Roads Office

Staff

Prof. Dr. Kay W. Axhausen, Frank Bruns, Clarissa Livingston, Sebastian Hörl, Remo Fischer and Bence Tasnády

Sub-project 2

Automated driving has the potential to fundamentally change mobility and traffic. The aim of sub-project 2 of the ASTRA research package "Impacts of Automated Driving" was to identify and quantify the potential positive and negative impacts of automated driving on the Swiss transport system, in particular its road infrastructure, as concretely as possible.

In this study, mode choice, route choice, and departure time choice were simulated. For this purpose, the "eqasim" implementation of the agent-based traffic model MATSims was used. Three scenarios, one without and two with automated vehicles, were created and simulated for the years 2020, 2030, 2040 and 2050 respectively. The data that formed the basis for these scenarios was a series of data sets from the FSO, the ARE, an SVI research project, and publicly available sources, such as OSM. These scenarios included domestic passenger traffic of the population residing in Switzerland and domestic as well as cross-border heavy road-based freight traffic. Population growth was scaled according to the reference scenario of the FSO population forecasts. The growth of freight traffic was scaled according to the ARE forecast for 2040 in terms of ton-kilometers.
The scenario without automated vehicles was used as the reference scenario. The two scenarios with automated vehicles, namely Scenario A and B, differ only with regard to the assumed private vehicle ownership or, more accurately, access to a private car. In Scenario A, the same proportion of agents have access to a private car as in 2020, whereas in Scenario B, access to a private car decreases significantly. This decrease is derived from the assumption that in Scenario B, people decide to give up their private car or not to buy a private car because of the major improvements in collective transport services brought on by automated driving, mainly the introduction of automated taxis. The other assumptions are identical for Scenario A and B, namely the assumptions about the timeline of regulatory approval for the operation of automated vehicles by roadway category, the capacity effects of automated vehicles, the value of time for the various simulated automated transport modes, and the market penetration rates of automated vehicles in private ownership and freight transport. Although not differing between scenarios, the regulatory approval for automated operations and market penetration levels change from year to year according to the gradual integration of automated vehicles into the transportation system and the automotive market. It is also important to note that it is assumed in all three scenarios that public transport will be fully automated from 2050 onwards and that the operational cost savings deriving from automation is passed on to the passengers. Thus, public transit fares are reduced in all three scenarios as of the year 2050. Sensitivity analyses were also performed to examine the sensitivity of the model to the public transport fares as well as the value of time of the automated modes, the road capacity-increasing effect of the automated vehicles, and the market penetration rate of the automated vehicles in the automobile market. Based on the simulations, bottlenecks on national highways were then analyzed to determine whether they change with automated vehicles. The procedure for the bottleneck analysis corresponds to that of the Strategic Development Program National Roads of the Swiss Confederation (STEP-NS).
In general, the results of the simulations show that the capacity-increasing effect of automated vehicles is already at least partially compensated by the demand effects (modal shifts, empty trips of automated taxis). It should be noted that empty trips of private automated cars, changes in residential and working location choice, and destination choice changes as a result of automated driving were not simulated. These effects were not simulated because sufficiently detailed quantitative data on the respective decision processes were not available. Thus, the demand effect is underestimated.
For many bottlenecks, changes in the bottleneck levels are only to be expected when the capacity-increasing effects of automated vehicles are high and the market penetration of the vehicle fleet with automated vehicles is complete. However, even for the most "optimistic" scenario simulated in terms of capacity effects, namely Scenario B with 100% automation of the vehicle fleet and a roadway capacity gain of 50% per automated vehicle, the performance-relevant traffic loads (in PCU) decrease by only about 10% due to the additional demand, which would mean a reduction of at most one bottleneck level. With population growth and similar mobility needs compared to today, this effect would be absorbed after a few years.
Automated vehicles could alleviate traffic bottlenecks, reduce average travel times, and increase travel comfort and benefits for road users, if these rather optimistic assumptions regarding their spatiotemporal requirements and thus roadway capacity-increasing effects hold true. However, this is only true if private vehicle ownership is drastically reduced. Automated vehicles - even with a drastic decrease in private vehicle ownership - will very likely lead to a significant increase in vehicle kilometers travelled, which would likely be harmful for the environment, residents and weaker road users such as pedestrians and bicyclists. Urban areas in particular would stand to benefit in the first sense and suffer in the second. Therefore, it will continue to be important to take measures that strengthen the environmental alliance (walking, cycling, and transit) and make private vehicle ownership less attractive. Since the effects of automated driving will differ by region - the contrast between urban and rural areas is particularly stark - measures will have to be adapted regionally.