University of Twente Student Theses


The impact of autonomous vehicles on the capacity of a weaving section : how different scenarios of autonomous vehicles and the penetration level thereof, influence the capacity of a 2 by 2 weaving section

Asbreuk, C.M. (2023) The impact of autonomous vehicles on the capacity of a weaving section : how different scenarios of autonomous vehicles and the penetration level thereof, influence the capacity of a 2 by 2 weaving section.

[img] PDF
Abstract:Vehicles are evolving every day. They for example accelerate faster, drive more fuel efficient or get more supporting features compared to earlier models. These supporting features currently include automatic emergency braking, blind spot warning, lane centering, (adaptive) cruise control, automatic parking and many more. These features support the driver, but do not yet take over the task of driving. This aligns with level 0 to 2 of the SAE levels of automation (J.S. Choksey, 2021). From level 3 to level 5, the vehicles drive autonomously. This means that the systems in the vehicle completely take over and the ‘driver’ does not need to drive anymore. In levels 3 and 4 the vehicle can still request the driver to take over, in level 5 drives autonomous in all conditions. Currently, autonomous vehicles that take over the driving task in all circumstances are not yet allowed on the public road. Restrictions are attempted and expected to be disabled by testing and improving the autonomous vehicles (AVs) to than prove the consistency, reliability and advantages of the AV. One of the advantages of AVs could be an increase in capacity. The capacity is the maximum number of vehicles passing a certain point within a certain time interval, often 1 hour, given certain circumstances. For this research motorway conjunction capacity is considered, which is the amount of vehicles driving away from the conjunction. If the capacity would be increased, current motorways can accommodate more vehicles. To determine if an increase in capacity is a likely cause of AVs, this research focusses on a simulation and comparison of conventional and autonomous vehicles. To do this, 5 types of AVs are modelled based on different behavioural features. The first three types are typified by their aggressiveness. This aggressiveness is embodied in acceleration, headway time and more. The other two types focus on a specific feature: connected AVs and AVs driving in a platoon. Each of these AV types will be simulated with 0%, 20%, 40%, 60%, 80% and 100% AV penetration rate, with the remaining percentage of vehicles driving conventional. The simulation will happen on a weaving section of 2 by 2 which has a length of 750 m. At both lanes the same amount of vehicles enter and for both half of the vehicles switch lanes. A weaving section has been chosen since it is a typical bottleneck on a motorway. As mentioned, the 5 types of AVs are simulated through driving behavioural features. The altered features can be classified based on four different types of settings in the VISSIM traffic simulation software; Wiedemann 99 parameters, Other following parameters, Lane Change parameters and autonomous driving features. Parameter changes have two main causes for the different types of Avs; first being due to the removal of human inconsistency and reaction times, second due to the human error being removed which leads lower risks. As a result of these alterations, vehicles can for example drive closer together. By altering the various parameters the 5 types of AVs were modelled. These AV types together with the penetration rates form the scenarios, a combination of a defined AV type and a penetration rate of that type. These scenarios were run, through which the flow downstream of the conjunction was measured. Moreover, the average speed upstream was calculated for each 5 min interval. If there was conjunction, meaning the speed was lower than 50 km/h, the corresponding flow was taken as a capacity value. Since for a given scenario multiple capacity values are present, the median of this value is determined to be the overall capacity. The cumulative frequencies of the capacity values were also calculated and visualised to get a broader view of the consistency of congestion capacity. The cumulative frequencies of the scenarios and the capacities show an inconsistency for both the 100% cautions and the 100% platoon scenario, the capacity values are drastically lower. This contradicts the trend of the other penetration rates and can be explained by some unrealistic vehicle behaviour within the model. Next to this, two trends are spotted. First, AVs seem to have a positive influence on the capacity. Second, this increase in impact from the penetration rate on the capacity seems not linear but increases exponentially. This trend however cannot be stated with certainty. The cautious AV barely influenced the capacity with only a maximum increase of 4% compared to the base scenario (conventional vehicles). The moderate AV already shows a bigger influence which is similar to the influence of the connected autonomous vehicle (CAV). They both lead to an increase in capacity of 35% at a 100% penetration rate. The aggressive and platoon AV show an even bigger impact with both scenarios having a +31% impact on the capacity at 80% penetration. When increasing the penetration rate further to 100%, the aggressive AV reaches the highest impact, 44%. This gives the general conclusion, that the capacity of conjunctions can be positively impacted by AVs. Important however is that the type of AV determines the extend of this positive impact, which ranges from 4% to 44% for the different types of AVs given a 100% penetration rate.
Item Type:Essay (Bachelor)
Faculty:ET: Engineering Technology
Programme:Civil Engineering BSc (56952)
Link to this item:
Export this item as:BibTeX
HTML Citation
Reference Manager


Repository Staff Only: item control page