Search

This article is concerned with the performance evaluation of connected, autonomous vehicles (CAVs) in a realistic large-scale microsimulation scenario. In particular the question is: how much could a high diffusion of CAVs possibly change (1)~the average travel speeds (2)~the trip times of all traffic participants, including pedestrians, and (3)~the energy/fuel consumption? For this purpose, admittedly favourable assumptions are made: a 100\% diffusion of platooning-capable CAVs as substitution for private cars as well as a high maximum speed of platooned vehiclesin order to enable platoon formation. The morning rush hour scenario of the metropolitan area of Bologna, Italy has been selected for assessment. This scenario, which has been created and validated in previous works, represents an activity based demand model with travel plans for individual citizens, including all relevant transport modes.  The microsimulation is performed by means of the SUMO simulator. The entire demand has been generated with the SUMOPy tool. For the platooning of CAVs, SUMO's SIMPLA module has been used, which controlls the vehicles via the interactive TRACI API.Results show an increased speed and reduced travel time for CAV vehicles, with respect to human driven cars, in particular in the periphery and less in the center with a dense road network.  However, the reason for improved speeds and travel times is predominantly the higher maximum speed  allowed for vehicles trying to catch up and join a platoon. Furthermore these higher speed would also be resposible for an increase in fuel consumption of approximately 5\%.In conclusion, CAVs alone are unlikely to reduce congestion in an urban area. To make the platooning concept work, additional technology and infrastructure is required in order to merge platoons effectively at freeways and at traffic lights. The latter could be simulated with GLOSA.

We simulate the introduction of shared, automated, and electric vehicles (SAEVs) providing on-demand shuttles service in a large-scale transport digital twin of the San Francisco Bay Area region (California, USA) based on transit supply and demand data, and using the mesoscopic agent-based Behavior, Energy, Autonomy, and Mobility beta software (BEAM) developed at the Lawrence Berkeley National Laboratory (LBNL). The main goal of this study is to test the operations of this novel mobility service integrated with existing fixed-route public transportation service in a mesoscopic simulation of a real case scenario, while testing the BEAM beta software capabilities. In particular, we test the introduction of fleets of on-demand vehicles bound to operate within circular catchment areas centered on high-frequency transit stops, with the purpose of extending the reach of fixed-route transit by providing an alternative first- and last-mile connection at high-frequency public transport stations. Results show that on-demand automated shuttles represent the best solution for some users, increasing the overall transit ridership by 3%, and replacing mostly ride-hail trips, especially those connecting to transit stops, but also some walking trips. This type of service has the potential to reduce overall vehicle miles traveled (VMT), increase transit accessibility, and save energy, but future research is needed to optimize this type of service and make it more attractive to travelers.

This paper analyzes the relation between Shannon's information entropy and trips made by cyclists in an urban context. Such entropy indicator depends on people distribution across a territory in a specific time. Its evaluation and temporal evolution have multiple applications in transport planning, from the specification of demand distribution models to the calibration and validation of transport demand models to be used where transport data are scarce. This feature is crucial in transport planning, as the lack of robust data to estimate travel demand is still a common challenge, especially for cycling mobility. The entropy indicator has been applied to the cyclists' case in the Italian city of Bologna, using approximately 165,000 GPS traces recorded from April to September 2017. The results are encouraging because the entropy indicator is able to explain the evolution of the agents' distribution throughout a day across the territory. However, it is suggested that future research may further deepen the concept to test its transferability, and to better establish its sensibility to other case studies. [ABSTRACT FROM AUTHOR]