Department of Science and Technology
Institutionen för teknik och naturvetenskap
Linköping University
Linköpings universitet
LiU-ITN-TEK-A--20/018--SE
Curbside Management and
Routing Strategies that
Incorporates Curbside
Availability Information
Richard Blixt
Carl Lindgren
LiU-ITN-TEK-A--20/018--SE
Curbside Management and
Routing Strategies that
Incorporates Curbside
Availability Information
Examensarbete utfört i Transportsystem
vid Tekniska högskolan vid
Linköpings universitet
Richard Blixt
Carl Lindgren
Handledare Ivan Postigo
Examinator Johan Olstam
Upphovsrätt
Detta dokument hålls tillgängligt på Internet – eller dess framtida ersättare –
under en längre tid från publiceringsdatum under förutsättning att inga
extra-ordinära omständigheter uppstår.
Tillgång till dokumentet innebär tillstånd för var och en att läsa, ladda ner,
skriva ut enstaka kopior för enskilt bruk och att använda det oförändrat för
ickekommersiell forskning och för undervisning. Överföring av upphovsrätten
vid en senare tidpunkt kan inte upphäva detta tillstånd. All annan användning av
dokumentet kräver upphovsmannens medgivande. För att garantera äktheten,
säkerheten och tillgängligheten finns det lösningar av teknisk och administrativ
art.
Upphovsmannens ideella rätt innefattar rätt att bli nämnd som upphovsman i
den omfattning som god sed kräver vid användning av dokumentet på ovan
beskrivna sätt samt skydd mot att dokumentet ändras eller presenteras i sådan
form eller i sådant sammanhang som är kränkande för upphovsmannens litterära
eller konstnärliga anseende eller egenart.
För ytterligare information om Linköping University Electronic Press se
förlagets hemsida
http://www.ep.liu.se/
Copyright
The publishers will keep this document online on the Internet - or its possible
replacement - for a considerable time from the date of publication barring
exceptional circumstances.
The online availability of the document implies a permanent permission for
anyone to read, to download, to print out single copies for your own use and to
use it unchanged for any non-commercial research and educational purpose.
Subsequent transfers of copyright cannot revoke this permission. All other uses
of the document are conditional on the consent of the copyright owner. The
publisher has taken technical and administrative measures to assure authenticity,
security and accessibility.
According to intellectual property law the author has the right to be
mentioned when his/her work is accessed as described above and to be protected
against infringement.
For additional information about the Linköping University Electronic Press
and its procedures for publication and for assurance of document integrity,
please refer to its WWW home page:
http://www.ep.liu.se/
Abstract
Vehicles that today are searching for a place to stop impacts other vehicles in cities. It can also be seen that the number of vehicles that desire to conduct a pick-up or drop-off increases with an increased number of ride-hailing services. New technology routing ad-vises for such vehicles could improve the overall performance of a traffic network. This thesis analyses therefore how a routing strategy, that incorporates curbside availability in-formation, can impact the performance.
To analyse the effects of how curbside availability information can impact a network, a mi-croscopic traffic simulation model was constructed in PTV Vissim and two different routing strategies were developed and implemented in the model. One strategy that represents the scenario of today where vehicles searches the traffic network while attempting to make a stop at a pick-up and drop-off slot. The second strategy routes vehicles to a slot based on curbside availability information. This strategy directs vehicles to an available slot and therefore reduces the time a vehicle is cruising before a stop has been made. A simula-tion experiment was set-up to compare the strategies that were developed with different penetration rates of vehicles that desired to stop.
The results shows that the average travel time can be reduced with up to 25.2% when ve-hicles have information compared to the scenario with no information. Similar findings is identified for average delay per vehicle which is reduced with up to 49.0% and average traveled distance decreased with up to 15.5%. The results of this thesis needs however to be studied in a wider context in order to draw reliable conclusions. The thesis propose further investigations whether a strategy that incorporates availability information can be imple-mented in a real world scenario and further investigations whether an implementation of a strategy like this would be socioeconomic beneficial.
Acknowledgments
Five years of studying at Linköping University have come to an end. This master thesis project is the last of work to complete our degree in Master of Science in Communication and Transport Engineering.
We would like to thank our supervisors at both Linköping University and WSP. They have provided us with great knowledge which have helped us to realise our ambitions with this master thesis. Thank you Ivan for our back and forth communication and the continues feedback of the report writing. Thank you Alexander for providing expertise in the software PTV Vissim and with report advises. Thank you Dirk for the idea of this master thesis and for contributing with new suggestions during the proceedings of it. At last, we want to thank our examiner Johan for concise guidelines and advises through this thesis.
We would also like to thank PTV for lending us licenses to the software PTV Vissim. Also, a great appreciation and thanks to WSP and the office employees in Norrköping for the dispose of space and technical equipment making this master thesis feasible during the pandemic of Covid-19.
Norrköping, June 2020
Contents
Abstract iii
Acknowledgments iv
Contents v
List of Figures vii
List of Tables viii
1 Introduction 2
1.1 Background . . . 2
1.2 Aim . . . 3
1.3 Methodology . . . 3
1.4 Delimitations and Limitations . . . 4
1.5 Structure of the Report . . . 4
2 Curbside Management 6 2.1 General Curbside Management Aspects . . . 6
2.2 Curbside Pricing . . . 7
2.3 Stopping Behavior at the Curbside . . . 8
2.4 Curbside Productivity and Management Strategies . . . 9
2.5 Curbside Loading Space Demand . . . 10
2.6 Information Sharing - Vehicles and Infrastructure . . . 11
2.7 Future Mobility in Urban Development . . . 12
3 Traffic Simulation 15 3.1 Definition of Traffic Simulation . . . 15
3.2 Microscopic Traffic Simulation . . . 16
3.3 Verification, Calibration and Validation . . . 17
3.4 Simulation of Parking and Stopping . . . 17
3.5 Simulation of Information Sharing In Traffic Networks . . . 18
3.6 Vissim - Microscopic Simulation Software . . . 19
3.6.1 Driving Behavior . . . 19
3.6.2 Constructing a Traffic Simulation Model . . . 20
3.6.3 COM-interface . . . 21
4 Implementation of the Simulation Model 23 4.1 Network Design . . . 23
4.1.1 Intersections in the Network . . . 24
4.1.2 Curbside Zone Configuration . . . 25
4.2 Traffic and Driving Behavior Configurations . . . 26
4.2.2 Vehicle Desired Speeds . . . 27
4.2.3 Input Flow . . . 27
4.2.4 Vehicle Routing Demand . . . 28
4.2.5 Stopping Behavior and Time Distributions . . . 30
4.3 Model Verification . . . 30
5 Implementation of Curbside Routing Strategies 32 5.1 Strategy 1 - Vehicles Without Information . . . 32
5.2 Strategy 2 - Vehicles With Information . . . 35
5.3 Routing Strategies Verification . . . 38
6 Simulation Experimental Design 39 6.1 Simulation Scenario Description . . . 39
6.1.1 Base Scenario . . . 39
6.1.2 Scenario Where Strategy 1 is Implemented . . . 39
6.1.3 Scenario Where Strategy 2 is Implemented . . . 40
6.2 Description of the Network Performance Metrics . . . 40
6.3 Simulation Parameters . . . 41
7 Simulation Results and Analysis 44 7.1 Average Travel Time . . . 45
7.2 Average Delay . . . 47
7.3 Average Traveled Distance . . . 49
7.4 Delay Over Time . . . 51
7.5 Ratio of Searching Vehicles per Available Slot . . . 52
7.6 Number of Stops per Curbside Zone . . . 53
7.7 Summary of the Results . . . 57
8 Discussion 58 8.1 Aim . . . 58
8.2 Method . . . 59
8.3 Impact of Delimitations . . . 60
8.4 The Work in a Wider Context . . . 61
9 Conclusion 63 9.1 Aim and Results . . . 63
9.2 Future Research . . . 64
Bibliography 66
A CPI Example 70
List of Figures
1.1 Two types of curbside utilization (highlighted in yellow) . . . 2
2.1 Single vehicle space requirement for loading activity . . . 10
2.2 Multiple vehicles space requirement for loading activity . . . 11
3.1 Static and Partial Routing example . . . 21
3.2 Hierarchy of Vissim-COM objects . . . 22
4.1 Network Design . . . 24
4.2 Four-way intersection . . . 25
4.3 Three-way intersection . . . 25
4.4 Curbside layout . . . 25
4.5 Reduced speed area at a curbside . . . 27
4.6 Vehicle inputs and vehicle exits in the network . . . 28
4.7 Vehicle Link Flow . . . 29
5.1 Flow chart of Strategy 1 . . . 33
5.2 Flow chart of Strategy 2 . . . 35
7.1 Average travel time per vehicle - All sub-scenarios . . . 45
7.2 Comparison Strategy 1 and Strategy 2 - Average travel time per vehicle . . . 46
7.3 Comparison Strategy 1 and Strategy 2 - Average travel time per vehicle - 25% of stopping vehicles . . . 46
7.4 Average delay per vehicle - All sub-scenarios . . . 47
7.5 Comparison Strategy 1 and Strategy 2 - Average delay per vehicle . . . 48
7.6 Comparison Strategy 1 and Strategy 2 - Average delay per vehicle - 25% of stop-ping vehicles . . . 48
7.7 Average traveled distance per vehicle - All sub-scenarios . . . 49
7.8 Comparison Strategy 1 and Strategy 2 - Average traveled distance per vehicle . . . 50
7.9 Comparison Strategy 1 and Strategy 2 Average traveled distance per vehicle -25% of stopping vehicles . . . 50
7.10 Comparison Strategy 1 and Strategy 2 - Average delay per vehicle over time . . . . 51
7.11 Comparison Strategy 1 and Strategy 2 - Average delay per vehicle over time - 25% of stopping vehicles . . . 52
7.12 Ratio of average number of vehicles per available pick-up and drop-off slot . . . . 53
7.13 Curbside zones and their priority level . . . 53
7.14 Vehicle distribution of the curbside zones per slot - 5% of stopping vehicles . . . . 54
7.15 Vehicle distribution of the curbside zones per slot - 10% of stopping vehicles . . . . 55
7.16 Vehicle distribution of the curbside zones per slot - 15% of stopping vehicles . . . . 55
7.17 Vehicle distribution of the curbside zones per slot - 20% of stopping vehicles . . . . 56
List of Tables
4.1 Number of pick-up and drop-off slots per curbside zone . . . 26
4.2 Vehicle composition . . . 27
4.3 Relative Origin-Destination flow . . . 29
6.1 Sub-scenarios with Strategy 1 . . . 40
6.2 Sub-scenarios with Strategy 2 . . . 40
6.3 Required number of runs per sub-scenario . . . 42
6.4 Simulation parameters’ value . . . 43
7.1 Base scenario results . . . 44
7.2 Number of pick-up and drop-off slots per curbside zone . . . 54
B.1 Used standard deviation values, when calculating the required number of runs . . 71
Abbreviations
API Application Programming Interface
AV Automated Vehicle
COM Component Object Model
CPI Curb Productivity Index
CV Connected Vehicle
GPS Global Positioning Systems
ITS Intelligent Traffic Systems
MaaS Mobility as a Service
OD Origin-Destination
TNC Transport Network Company
V2I Vehicle to Infrastructure
V2V Vehicle to Vehicle
V2X Vehicle to Everything
1
|
Introduction
Curbside management is of great interest for traffic and transportation planning due to its dif-ferent utilization and increasing demand. How the curbside can be managed with new tech-nology will be studied in this thesis. This chapter includes a background, aim and method of the study, delimitations and limitations. The last section in this chapter presents a description of the structure and content of this report.
1.1
Background
The curbside is a public space located between the road and the sidewalk. It can be used in various ways: parking, bus stops, pick-up and drop-off zones, bike lanes or even as space for restaurant patios. From the road to the curbside, there is a demand of access from actors such as parking vehicles, maintenance vehicles, delivery services and vehicles that conducts passenger pick-up/drop-off activities. It is therefore required to balance the utilization of the curbside space amongst all actors. Figure 1.1 shows an example of how the curbside can be utilized.
Figure (1.1) Two types of curbside utilization (highlighted in yellow)
How city planners have divided the curbside space amongst actors has historically been with static strategies (fixed parking fees or fixed spaces for certain activities). With the rise of ride-hailing services and an increased number of delivery services, the curbside has become more crowded and how the curbside is managed needs to be changed (Fehr & Peers 2018). A crowded curbside indicates that the access from the road to the curbside may be limited. Vehicles that are denied to stop at their desired location due to an occupied curbside have to find another location. Consequently, this creates a large number of circulating vehicles in a traffic network. Beyond congestion, other complications that arises are illegal double parking and traffic safety problems.
To address the problem of a growing demand of ride-hailing services, a possible solution is to shift away from an on-street parking focused set-up of the curbside. This set-up could instead include more zones designated for pick-up and drop-off activities. However, the problem of
1.2. Aim an increased number of searching vehicles still exists. An increased number of searching vehicles may increase both the total travel time and the total distance traveled in a network, which in extent contributes to negative environmental impacts.
A new mobility concept is information sharing between infrastructure and vehicles which can impact the utilization and change the management of the curbside. A vehicle that desires to make a stop at a curbside, can with access to information, be aware of whether the curbside is available or not in an earlier state. Instead of having vehicles cruising for available spots, they can drive directly to designated locations which can ease congestion in a traffic network. By implementing sharing of information in a traffic simulation tool, it can be possible to see how this new mobility concept can impact a traffic network. This is the focus of this thesis. Curbsides are today managed by regulations which restrict vehicles where to park and where to make stops for loading and unloading. The questions are however, what is considered an effective way to manage a curbside and how can the management be evaluated? Can the management of the curbside be done more efficiently with vehicles having access to informa-tion?
1.2
Aim
The aim of this master thesis is to investigate how sharing of curbside availability informa-tion between vehicles and infrastructure can impact the performance of a traffic network (in terms of travel time, delay and traveled distance). To achieve this, a model will be developed that describes searching behavior of vehicles that desires to make a stop in a network, with and without access to information about availability at curbsides. The searching behavior of vehicles that desires to make a stop is related to how curbsides are managed. Therefore, the thesis will also investigate how curbsides are managed today and how future technology can impact the management of curbsides.
Research questions that are expected to be answered during this master thesis are: • How are curbsides managed today?
• How can vehicle searching behavior be modelled in a microscopic traffic simulation?
– For vehicles that have no access to information about curbside availability.
– For vehicles that have access to information about curbside availability.
• How can sharing of curbside availability information between vehicles and infrastruc-ture impact the performance of a network?
1.3
Methodology
The methodology consists of four main parts, beginning with a literature review. The areas of curbside management, information sharing between vehicles and infrastructure and traffic simulation will be studied. The knowledge gained from the literature review is the basis of the development of the microsimulation model. The simulation model will consist of a base network in which two routing algorithms is developed and implemented. The two algorithms represents two types of curbside management strategies, in which the routing of vehicles differ. One strategy is for vehicles with information about curbside availability and another for vehicles without information.
To extract metrics and analyse how the performance of a traffic network in a simulation is im-pacted, a scenario with each strategy is created in one network. The metrics will be extracted
1.4. Delimitations and Limitations and used to evaluate the scenarios are the average vehicle travel time, the average delay per vehicle and the average traveled distance per vehicle. Both strategies will be simulated with a different number of vehicles that desire to stop.
1.4
Delimitations and Limitations
This study considers only pick-up and drop-off activities, which is a stop at a curbside. These activities are conducted by passenger cars and are assumed to have the same duration (mean value of 40 s and standard deviation of 10 s). The network used is limited to a small net-work with one central node that attracts vehicles that desire to make a stop. The study do not consider vehicles stopping at several places in the network, e.g. one stop for dropping pas-sengers off or one stop for picking paspas-sengers up. Further on is the number of curbside zones fixed and their location is never adjusted in the simulation model. The simulation study only look at effects at one travel demand level and the gains from the used metrics might be larger or smaller at other demand levels.
Metrics used in the output analysis is delimited to the average vehicle travel time, the average delay per vehicle and average traveled distance per vehicle. In the simulation experiments, the study considers a time period of 60 minutes.
In both of the developed simulation model and strategies, neither of calibration or valida-tion was conducted since the investigavalida-tion was done for a fictitious network and no data for calibration and validation were available.
The implementation of the two curbside routing strategies is done in a centralized way. I.e. when the routing strategy of vehicles that have information is implemented, all vehicles that desires to make a stop have information. Each strategy is implemented separately and does not work in a combination with each other, meaning that a vehicle that doesn’t have infor-mation will not interfere with vehicles that have inforinfor-mation about curbside availability. Because the study is carried out using the simulation software PTV Vissim, the result of the study is limited to Vissims capacity and ability to produce valid results.
1.5
Structure of the Report
In Chapter 2 Curbside Management previous literature in the subject of curbside management is summarized. In this chapter subjects of curbside pricing, productivity, space demand are discussed as well as information sharing systems and future mobility in urban developments. Chapter 3 Traffic Simulation looks into the concept of traffic simulation where the focus is on microsimulation within the simulation software PTV Vissim. This chapter also reviews previous studies in simulation of parking and simulation of information sharing in traffic networks.
Chapter 4 Implementation of the Simulation Model describes the part of the simulation model methodology of the thesis. This chapter presents the development of the network design which describes the physical structure of the network.
Chapter 5 Implementation of Curbside Routing Strategies describes how the routing strategies works and the algorithms that were developed in the two strategies.
1.5. Structure of the Report Chapter 6 Simulation Experimental Design contains the experiments set-up and the construc-tion of comparable scenarios made in this thesis covering the developed simulaconstruc-tion model and the routing strategies. This chapter also presents the metrics used to evaluate the ex-plained experiments.
Chapter 7 Simulation Results and Analysis presents the results for the different sub-scenarios with analysis and a summary of the results is given.
Chapter 8 Discussion contains discussion of aim, method, delimitations and the work in a wider context.
Chapter 9 Conclusion presents the conclusions given from the thesis in terms of the research questions. Some ideas for future research is also given.
2
|
Curbside Management
This chapter presents previous work done in the area of curbside management and a sum-mary is given on the topic to answer the first research question. The literature review assess the management aspects of the curbside, both as it is today and how it can be manage in future scenarios. Further on, a literature review on stopping behavior of vehicles at the curb-side is presented as well as information sharing between vehicles and infrastructure. Lastly, a review on future mobility in urban development including curbsides is presented.
2.1
General Curbside Management Aspects
The curbside is a place where several stakeholders demand space, which can be divided into two groups. One group is those who are demanding access from the roadside of the curbside and the other group is those that are demanding access from the sidewalk. The curbside is both the deliminator and connection point for the two groups. There need to be a way for travelers on the road to reach businesses and residents that are located adjacent to the street and a way for travelers to reach the road (ITF 2018). With the rise of demand for using the curbside, an alternative more efficient way of managing it is required (Marsden, Docherty, and R. Dowling 2020). New technologies can change how the curbside is used, one could for example believe that the introduction of automated vehicles can shift the demand for on-street parking towards a demand for loading/unloading zones.
When managing the curbside today, the objectives can differ between cities, municipalities or regions (Barter 2016). The clear objective for policymakers would be to ensure that there is a reliable connection between the road and the sidewalk and that the curbside resource is fairly divided amongst the users. A well-managed curbside means that travelers with ease can get to and from their desired location. (Debow and Drow 2019)
The fulfillment of this objective would decrease the travel time for people in the network, both in terms of finding a parking space close to the desired destination and in terms of not congesting the network while searching for a free space (Shoup 2004). On average, vehicles in major cities (worldwide) are cruising for 3.5-13.9 minutes to find a free parking space at the curbside (Shoup 2004), which contributes to congestion and air pollution. On-street parking can cause congestion in other ways. For example vehicles can make illegal stops (double parking or near intersections), they can make abruptly stops in the traffic while waiting for spots, or vehicles can disturb the traffic flow while maneuvering into parking spots (Barter 2016).
On-street parking management is a highly local matter, in some places it is not an issue, be-cause of low demand. The management in such places is about communicating instructions on where and when drivers can park. In other places, parking is a real issue, for example in
2.2. Curbside Pricing areas with a high density of commercial buildings and in growing/larger cities or districts. The issues that could arise from poor management, other than congestion, can be occupation of the curbside across whole days, traffic safety problems and underuse of off-street park-ing spaces. Accordpark-ing to Barter (2016), parkpark-ing shortages should be addressed uspark-ing proper management rather than expanding the parking supply. An expansion of the supply (build off-street parking spaces) is often costly and might not give desired effects, as travelers still desire a parking space close to their destination.
2.2
Curbside Pricing
Travellers are commonly searching for on-street parking as this is the most convenient way to reach their destination. Parking fees of on-street parking can motivate travelers to instead find off-street parking spaces. For example, employees in a shopping area could be nudge to an off-street parking instead of using the on-street parking spaces using different prices. The more convenient parking spaces at the curbside could instead be used for serving the shopping areas customers (Barter 2016). The "right" parking fee is based on the demand for parking and can therefore be either high or low. Traffic engineers recommends that there should be an available space between every seven cars (85% occupancy level), which could be achieved using different pricing strategies (Shoup 2004).
Since the introduction of charged on-street parking, when the parking meter was introduced, the pricing strategies have not changed much. Typically a static hourly price is used for pricing the curbside during whole days. The concept of performance parking prices, which is used in the pilot program SFpark (in San Francisco), is reviewed by (Pierce and Shoup 2013). Compared to static pricing strategies, performance based strategies could increase the effectiveness of how the curbside is managed. The authors implies that parking prices can be used to moderate the traffic in a city. According to Pierce and Shoup (2013) the curbside prices should be set to reflect the public goal:
"The parking price that achieves one or two open spaces per block is not a free-market price; it is instead a public price for a public service, and it should be set to achieve the public goal of effectively managing the parking supply."
In the pilot program SFpark the parking prices were adjusted depending on the occupancy rate at the curbside in certain blocks in the city of San Francisco. The occupancy rate refers to the number of occupying vehicles at the curbside in relation to the capacity. If the rate (at one block) was below 30% the price was reduced with 50 cents per hour, if it was above 80% the price was increased with 25 cents per hour. The aim in the study was to have the occupancy rate between 60% and 80%. If this rate was too high the number of searching vehicles was assumed to increase, when the rate was too low businesses adjacent to the curbside was assumed to lose potential customers.
The result of SFpark showed that by using a dynamic pricing strategy, parked vehicles could be more evenly spread out over adjacent blocks. Before the program was implemented, the occupancy rate of the blocks could vary between under-occupied to crowded, whereas after the program was conducted the occupancy rates were more evenly distributed amongst the blocks. The study also showed that when the total number of parked vehicles (in the whole area) increased, the prices became on average lower and the revenue for the government got higher.
2.3. Stopping Behavior at the Curbside Even though the management strategy and regulations can be adjusted based on the demand, there are still uncertainties on how well a system could work without a reliable way of com-municating changes to the users. In the SFpark program real time information about prices and occupancy levels was communicated via a website, how users responded to the pub-lished information is however unclear. If many drivers miss or neglect the dynamic pricing program, the goal of an evenly distributed occupancy rate is hard to reach. The utility gain for cheaper parking (but a couple of blocks away from the destination) might also not apply for every driver.
The regulations have to be adjusted to manage an increasing number of goods deliveries and Transport Network Companies (TNCs), like Uber and Lyft. Historically the regulations for managing the curbside, have evolved over the time (Marsden, Docherty, and R. Dowling 2020). How the regulations will continue to evolve, and what stakeholders that in the future will have most advantages are not yet determined. Stakeholders that in the future are likely to demand more space at the curbside are for example goods vehicles and TNCs (Debow and Drow 2019).
The optimal parking and stopping fees may be hard to determine. Dynamic pricing strategies may be an alternative way of getting the prices "right". By managing the prices at curbsides with high demand the traffic flow in a city could be managed as a higher stopping fee can nudge travelers on where to stop.
2.3
Stopping Behavior at the Curbside
Vehicles stopping behavior depends on the type of stopping activity, on the type of vehicle it involves, on the type of errand to be performed and how long time it expects to use the curbside spot.
Parking has dominated many curbside areas since motorists’ desire to travel door-to-door (Marsden, Docherty, and R. Dowling 2020). However, pick-up and drop-offs are becoming more and more common with TNCs. Another increase of curbside use is the loading and unloading of goods by delivery companies that use the curbside because of the convenient and most accessible point to end delivery (Marsden, Docherty, and R. Dowling 2020). Private cars, car-sharing companies and taxi companies use the curbside as the primary point of transaction between passengers in vehicle and the sidewalk. Another type of vehicle that desire to use the curbside for pick-up and drop-off passengers is buses when the curbside have designated formal bus stops.
Stopping behavior at a curbside is also dependent on how flexible a curbside user is. This is consecutively, often determined by what group the user belongs to. An employee in a shopping facility is more likely to change parking place than a traveler stopping for a short errand. The flexibility of motorists can, when managing long term parking and pick-up and drop-off zones, be used in an effective way. TNCs and private vehicles that are stopping for a pick-up or drop-off are however less flexible, on where they can make stops (or park for a shorter time). (Barter 2016)
The time that a vehicle remains stopped for is denoted as dwell time. The dwell time is dependent on the number of passengers that a vehicle carries and which type of errand a vehicle is conducting. There is however limited literature on the dwell time for vehicles that performs pick-up or drop-off activities.
2.4. Curbside Productivity and Management Strategies
2.4
Curbside Productivity and Management Strategies
The dwell time can be used for evaluating the productivity of a curbside. When studying curbside activities in the San Francisco area, Fehr & Peers (2018) have developed a metric call “Curb Productivity Index” (CPI). The CPI is used to evaluate the curbsides productivity when it is used by various modes, a CPI-value for a bus stop could be compared with a parking space at the curbside. In Equation 2.1 Activity is the number of passenger activities that are taking place during a certain period. Time refers the total dwell time (all vehicles) at the curbside during the period. Space is the how much space the specific mode requires at the curbside.
CPI= Activity
Time ˚ Space (2.1) The outcome of the metric can be factored using a distance, for example 20 meters, which gives the unit # Passengers per time period per 20 meters. The unit can be seen as the service the curbside gives per used curbside length. This metric can be used to compare a curbside for loading/unloading passenger to a curbside used for on-street parking. An evaluation example is presented in Appendix A.
Another way of measuring the productivity of a curbside is to simply look at the occupancy rate of the curbside. The CPI and occupancy rate of a curbside has no connection to how an oversaturated curbside impacts the traffic network. Two similar occupancy rates at two separate streets may have different effects on the traffic flow (C. Dowling, Fiez, Ratliff, and Zhang 2017).
Each curbside user has their own needs and trade-off considerations (convenience, avail-ability, price and utility). By understanding these, it is possible to implement a curbside management program to improve the curbside productivity (Debow and Drow 2019). Such a program could include components as: Allocating the curbside access to various users, in-tegrating fragmented data, monitor and measuring the curbside use, communicate the curb-side rules and access to users and the general public, enforcement of the curb to see that the rules are followed and at last, reporting and analysing against its objectives. Another component that can be used in a management program is curbside strategies. Fehr & Peers (2018) presents three strategies that could improve the curbside productivity and are here summarized.
The first strategy is relocation, which basically implies that curbside spaces, dedicated to a certain user along a block, gets relocated to another space. Relocation merges separated spots into a large spot. The overall amount of space dedicated to each user will be kept constant, i.e. the same amount and sizes of parking spaces etc. Dedicated spaces (such as parking spaces) on a block are often spread out which becomes a problem in the productivity of a curbside. Relocation can reduce common problems such as double-parking, which blocks the traffic flow of the road. For instance, a curbside block has set aside certain spots for loading activities which, at some locations, can be similar in size to the single on-street parking spot sizes. The problem arises when trucks, that are larger than the dedicated curbside spot, wants to use the curbside for loading. By relocating loading spaces to be adjacent each other, more allowed space can be accessed by trucks and thus also increase the flow on the road.
The second strategy is conversion which means converting curbside spaces along a block to adjust the amount of dedicated space each type of user is allowed. Conversion assigns spaces for each activity. Commonly this means removal of some on-street parking spaces. Dedicated
2.5. Curbside Loading Space Demand passenger loading/unloading zones, like a bus or taxi stop, results in better productivity (a higher CPI value) than what on-street parking do. This strategy can improve curbside productivity in already existing blocks where the demand for a certain mode is missing or inadequate by the existing curbside allocation.
The third strategy, flexibility, is based on the time of the day. This strategy includes imple-mentation of technology and modification of the infrastructure. As the demand of a curb-side oscillates throughout the day, implementing this strategy one could increase the supply. However, this strategy needs to be monitored to fulfill its maximum productivity. A flexible curbside system should be particularly effective when a mix in land use around the block consists of overlapping demand (different kinds of users) over the day. In an area with em-ployment centers and commercial uses, passenger loading would typically have the high-est demand during the commuter hours (morning and afternoon) and in the evening hours (when customers go shopping). Commercial goods loading and unloading demand is usu-ally the highest in the early morning and afternoon hours. This strategy requires installations to properly enforce the time of the day permission, to function correctly. This may include displays, curbside meters, markings etc.
By considering these three strategies when managing the curbside it is possible to improve the productivity and achieve a higher CPI value. However, it is important to consider the existing curbside as various curbside locations require different solutions and strategies in management.
2.5
Curbside Loading Space Demand
A formula was developed by Fehr & Peers (2019) to calculate how much space a vehicle requires to make a stop at a curbside for loading activities. The formula is based on various behaviors of vehicles, sizes and the space needed to steer to the curb and back to the travel lane. Figure 2.1 illustrates the space needed.
Figure (2.1) Single vehicle space requirement for loading activity, A = entry distance, B = vehicle length and C = exit distance. (Fehr & Peers 2019)
The total distance required is equal to A + B + C for a single loading space in a midblock, if the allocated curbside space is between parked cars or other obstructions. The calculation would be different if the curbside loading space was placed in the beginning of a block or at the end of a block. One of the findings in (Fehr & Peers 2019) is that approximately 6 meters is a good value for A, B and C each which would offer suitable space (meaning 18 meters in total).
For more than one vehicle, the methodology needs to be adjusted slightly. Figure 2.2 illus-trates multiple vehicles and how much space is required.
2.6. Information Sharing - Vehicles and Infrastructure
Figure (2.2) Multiple vehicles space requirement for loading activity, A = entry distance, B = vehicle length and C = exit distance. (Fehr & Peers 2019)
The calculation of curbside space in this case is calculated according to equation 2.2.
Curbside loading space=A+n ˚(B+C) (2.2) Where n is the number of simultaneous events and A, B, and C are equal to the last case. By using 6 meters as a value in a midblock situation, the one passenger vehicle loading zone becomes (n=1) 18 meters, the two passenger vehicle loading zone becomes (n=2) 30 meters and the three passenger vehicle loading zone becomes (n=3) 42 meters. This calculation can be used to determine the curbside space required and how much allocated space should be given to passenger loading.
2.6
Information Sharing - Vehicles and Infrastructure
Recent advantages in Intelligent Transportation Systems (ITS) can impact the management of the curbside as well as change how the utilization of the curbside affects the performance of a traffic network. In this section the concept of ITS, more specific, information sharing will be reviewed and how it could impact curbside management.
ITS is the inclusion of new information and communication technologies within transporta-tion and traffic management systems. Some improvements of ITS are traffic safety, enhance-ment of mobility and reduction of emissions (M. Zhao, Ang, B. Zhao, and Ng 2019). One ITS application is the connected vehicle (CV) system that use connectivity trough wireless com-munications, positioning by Global Positioning Systems (GPS) and data processing to enable vehicles, infrastructure and other online devices to share information.
From a traffic operations perspective, the key focus of CV systems is to coordinate strategies to improve the flow in intersections and on highways (Mahmassani 2016). This includes sta-bilization of speed, cruise control and queue warnings. The higher the number of vehicles that are connected in a traffic network, the better the opportunity for coordinated interven-tions to ensure that the quality and reliability of the traffic flow improves (Mahmassani 2016). There are three main types of communication which diversifies the CV environment and options, Vehicle-to-vehicle (V2V), Vehicle-to-Infrastructure (V2I) and Vehicle-to-everything (V2X) (Shladover 2017). The latter is at the most general of them which could imply that every device could connect to any other device including vehicles. V2X can be used between vehicles and pedestrian crossings, roadworks or cell phones etc.
V2V can enable applications that are time-critical and safety focused. Information that is shared is typically speed, location and heading. According to Shladover (2017), a V2V appli-cation is supportive collision and hazards warnings. Vehicles can detect a possible oncoming
2.7. Future Mobility in Urban Development collision and actively use the breaks to intervene and avoid the collision. Another applica-tion is cooperative cruise control. This allows vehicles to detect the presence of surrounding vehicles’ speeds and headings and use this information to achieve smother traffic flow as the cruise control adapts to vehicles in front. This application also refers to automated platooning where vehicles can lower the distance between each other and thus increase the road capacity. V2V is also useful at merging lanes or highways, where vehicles can negotiate and be aware of other vehicles when the desire is to shift lane or merge.
V2I enables many different applications, where some of them are safety related while others improves mobility. Example of such applications is the red light violation that uses an inter-face traffic signal timing, the vehicles position and speed to confirm whether the vehicle will have to decelerate or not before reaching an intersection with a traffic signal. An alert is sent to the vehicle if the vehicle will not be able to pass before the light turns to red. Vehicles can also get information regarding green wave speed advises and signal warnings like bridge open-ings. Another V2I application enables access to detailed real time traffic information such as travel times, current speeds, volumes, queue lengths, etc. Information can be collected and shared by both infrastructure and vehicles to minimize the total travel time. Lastly, an application that can be used for vehicles with information is parking area occupancy levels and parking spot availability. This application can be used for vehicles that desire to park or load/unload goods or passengers. Instead of circulating near a featured area, they can get the information on available spots on beforehand. (Shladover 2017)
The traditional way of communicating parking availability is done by signs and road mark-ings and the fact that the driver itself must witness the available spot. Kharrazi and Atif (2020) states that information sharing between vehicles and infrastructure is a viable solution to the problem of finding available spots to park or stop at. A more efficient use of the exist-ing parkexist-ing spot in the infrastructure is required for information sharexist-ing to function properly. Magnetic and ultrasonic sensors might be implemented in parking spots to detect availabil-ity. By transferring the parking spot availability received from the sensors with real-time information to a parking system, it is possible to display available spots in, for instance, an entrance to a parking garage.
Further on, information of parking spot availability could be communicated to a cloud-based service. The technology could then further be implemented in a larger area, such as an down-town area. Information could be spread by real-time signs or even with smartphones and app-based parking guides that provides knowledge of vacant parking options nearby (Barter 2016). Within this, smart parking guidance could help road users to directly go to an avail-able parking spot. Information sharing could help other services by finding free loading and unloading zones close to their desired destination.
A positive effect that could be achieved from smart parking systems in congested downtown areas might be to shift traffic to designated lanes, that leads to more distant off-street parking spots/complexes. This could lower the traffic volumes in city centers and also distribute the overall occupancy levels of parking spots. (Kharrazi and Atif 2020)
2.7
Future Mobility in Urban Development
There is new pressure on the curbside from the recent changes in mobility and transportation. How curbside management will be done in the future in uncertain. It is however clear that changes in the mobility system need to be considered when balancing new movements, pick-up and drop-off, and parking because this could be fundamentally different in the future (Marsden, Docherty, and R. Dowling 2020).
2.7. Future Mobility in Urban Development Nowadays, transportation of individuals is cheaper per traveled distance than it was two centuries ago (Wegener 2013). During the period the price has continuously decreased as people have increased their travel volume. This trend will however not proceed endlessly due to finite sources of fossil fuel and the high costs of alternative vehicle transport regardless of new technology innovation (Wegener 2013).
Mobility has been dominated by private vehicles powered by combustion engines. Private vehicles have transported people to tons of desired locations due their adaptability. This has emerged as a problem as the mass implementation also led to congestion on the roads (JRC 2019). Some factors that are negatively affected by congestion are the environmental, human health and safety factors. The urbanization of cities includes increased number of inhabitants and more mobility requirements which could further aggravate these factors. Transporta-tion and mobility are major factors which contributes to the fact that many countries keep growing their greenhouse gas emissions. Greenhouse gas reduction targets have been set but it requires both technology innovations and price incentives to persuade mobility behavior deviations (Wegener 2013).
In later years, policies and new mobility trends have arisen. To reduce congestion, some strategies that have been established are congestion charges, higher parking fees and restric-tion of vehicles in city centers. Campaigns to promote public transport and encouragement of non-motorized transportation modes such as bicycles and walking are already in progress (JRC 2019). New mobility trends include automated vehicles (AVs), electric vehicles, shared use vehicles and connected vehicles (vehicles that are capable of sharing and receiving infor-mation).
Another new trending mobility concept is Mobility as a Service (MaaS). MaaS is defined by Heikkilä (2014) as:
"A system, in which comprehensive range of mobility services are provided to customers by mobility operators."
The MaaS concept aims to have a future with seamless mobility, making sure customers to have the same freedom and convenience as they would with privately owned cars. By using smartphones, customers should be able to personalize their own multimodal trips and to pay for the whole journey with a single payment (Hirschhorn, Paulsson, Sørensen, and Veeneman 2019). The current transportation approach is made by letting the people plan and pay for all their trips separately. With MaaS, getting from origin to destination is a more seamless and integrated mobility experience.
Ride-hailing services like Uber and Lyft have increased in popularity around the world in the last years. They capture a new significant rate of how people have changed their travel pattern in cities. It is however important for city planners and policymakers to capture these travel patterns to understand how curbside management should be done and to design and shape MaaS models. With the current progress of vehicle automation technology, the esti-mated adoption of ride-hailing services is expected to increase. These services will affect transportation and without any clear understanding on how this affects the urban environ-ment, cities will have difficulties to make valid and intelligent policies. (Clewlow and Mishra 2017)
If alternative transportation modes such as ride-hailing services increases in popularity or if the introduction of fully developed AVs becomes reality, one could assume that future cities would require fewer vehicles (in total) to cover the transportation demand. According to Boesch, Ciari, and Axhausen (2016) the vehicle fleet size could be reduced up to 90% with
2.7. Future Mobility in Urban Development AVs and still cover the travel demand of today. As transportation would be cheaper in a setup with fewer privately own vehicles there will still be a need to control the number of vehicles at curbsides.
One measure to control the number of vehicles at the curbside is curbside pricing. This also makes it possible to control the transition of privately owned vehicles to ride-hailing services. Shoup (2006) states that it is important to get the prices right as it can, in the extreme, elim-inate parking desire at curbsides fully or generate plentiful of cruising vehicles on the roads due to occupied curbside areas. Likewise, curbside pricing can allow an increased demand for ride-hailing services which only requires a short time to stop at the curbside to pick-up or drop-off passengers.
An important factor of the curbside intensification is the uptake of ride-hailing services, like the rapid growth in San Francisco, where 15% of all trips are made by these services (Mars-den, Docherty, and R. Dowling 2020). Small levels of growth of ride-hailing services have a limited impact on the curbside, but when the demand is concentrated geographically it might be problematic in some places. How much these systems and services will disrupt the future mobility and curbside utilization is uncertain. Clewlow and Mishra (2017) discusses that the future changes might be so substantial that society is moving from a city where parking is dominant to a complete pick-up and drop-off environment.
3
|
Traffic Simulation
In this chapter the concept of simulation will be addressed, more exact: traffic simulation. Firstly, a definition of traffic simulation is made. Then a more in depth description on micro-scopic traffic simulation is given. The concepts of verification, calibration and validation are presented. Lastly, the chapter focuses on applied modelling and simulation of vehicle parking and information sharing to end with an description of some objects in the microsimulation tool PTV Vissim, that can be used to build a simulation model.
3.1
Definition of Traffic Simulation
A simulation is often a representation of a real system, that can either be static or dynamic. A static model simulates a system at a point in time, whereas a dynamic system is changing over time. (Robinson 2014)
A traffic simulation model can either be time-based or event-based, which defines the way the model is updated during a simulation. A time-based model updates repeatedly in prede-fined intervals and an event-based model updates when the state of the model is changed. One can also differentiate simulation models whether they are deterministic or stochastic. A deterministic model has no variation or randomness, where all decisions are predetermined. A stochastic model can produce random outcomes due to its use of statistical distributions. (Tapani 2005)
Traffic simulation is a great tool when evaluating road designs, testing changes in rules and regulations and other traffic management approaches (Olstam and Tapani 2004). Experi-ments done in a real system may be costly, hard to control and time consuming, whereas ex-periments done in a simulation environment can be controlled and observed in detail faster (Robinson 2014). Typically can what-if scenarios be evaluated using simulation models. A what-if scenario can evaluate what the effects of e.g. a new lane, a new off-ramp or a new traffic signal scheme can have on a traffic network. What if a new lane is added to the road? What if a new traffic signal scheme is implemented?
Traffic models can be classified with respect to their level of detail in how the traffic state are described. A simulation model at a macroscopic level utilizes aggregated input data and the results are represented in aggregated forms. Traffic models at this level is based on the fundamental relationship between flow, average speed and density. When modelling traffic at a macroscopic level the traffic is assumed to flow as streams of vehicles throughout sections in a network (Tapani 2005). Macroscopic traffic models can be used to analyse larger networks due to its low model resolution. A traffic model at a microscopic level (higher resolution) is more detailed in its way to model the traffic and the network. Entities in a microscopic model are driven by behavioral sub models that controls how individual vehicles
3.2. Microscopic Traffic Simulation reacts to each other. Microscopic traffic simulation models are reviewed more detailed in Section 3.2. Traffic models at the mesoscopic level is a mix of the high and the low resolution level models. It allows larger networks to be simulated significantly faster than networks at the microscopic level. The mesoscopic driving behavior is based on individual vehicles but updates for instance each tenth second. Individual vehicles are examined after an event has occurred, such as when the vehicle reaches the end of a route or passes through a node. The main advantages of mesoscopic simulation models are the increased simulation speed and the lesser time to create and calibrate the network (PTV GROUP 2019).
3.2
Microscopic Traffic Simulation
A higher level of detailed simulation can be represented in microscopic traffic models where individual vehicles and their interactions with other vehicles are studied (Olstam and Tapani 2004). Due to this detailed level, microscopic simulation models tend to be better suited in smaller networks such as an intersection. In general, microsimulation is used when conduct-ing evaluations for road capacity, level-of-service analysis or various intersection designs etc. (Olstam and Tapani 2004).
A microscopic traffic simulation model utilizes different sub-models, so called behavioral models. Behavioral models describes how individual vehicles are behaving. Heterogeneity amongst vehicles can be modelled with different behavioral models which gives a realistic representation of a real traffic system. Factors that influence drivers (which are captured in behavioral models) can be divided into sub-parts, which by Miska, Muller, and Zuylen (2006) are described as individual differences, situational factors and other traffic. The other traffic-factor determines how much influence the other factors has. In a congested state of a network, indi-vidual factors (e.g. stress level of a driver or trip purpose) and situational factors (e.g. weather or road conditions) has less impact of how a vehicle is behaving. If the traffic of a network is in a free flow state, the other traffic-factor has less impact of how a vehicle is behaving. A vehicle can either be in a free state or a restrained state, models that describe this are called Car-following models.
Car-following models describes the behavior of a follower when its traveling in the same lane as a leading vehicle. A car-following model can describe how, and when a follower will decel-erate if a vehicle in front of it has a lower desired speed. Commonly a car following model is set-up using three regimes, one free, a normal following and one regime for emergency break-ing. In these regimes the speed and acceleration/deceleration are adjusted. Car-following models can be classified depending on what logic they are using. Typically, there are three classes, Gaiz-Herman-Rothery-models (GHR), safety distance models and psycho-physical models (Olstam and Tapani 2004).
Another common sub-model used in microscopic traffic simulation models is lane-changing models, that describe vehicles lane-changing maneuvers. Traffic flow oscillations and break-downs are often according to Moridpour, Sarvi, and Rose (2010) depending on the lane chang-ing maneuvers. Before a driver decides to make a lane changchang-ing maneuver, he or she needs to answer following questions (Gipps 1986); Can I change lane? Do I need to change lane? Do I want to change lane?
Given that the answer to these questions are affirmative, a lane changing maneuver can occur. The possibility to change lane (which relates to the first question) are associated to the oncom-ing traffic in the adjacent lane or if its physically possible to change lane. The second question has to do with the desired route of a driver, a lane-changing maneuver may be necessary to make a lawful turn at an upcoming intersection or any other path associated conditions. The
3.3. Verification, Calibration and Validation outcome of the third question is related to the drivers desired speed and the speed of vehicles in the current or adjacent lane.
3.3
Verification, Calibration and Validation
As it is difficult to mimic a real system exactly on a computer, it is necessary that verification, calibration and validation of the simulation models are done in order for the output to be reliable.
The verification process reviews the implementation and simulation model development to make sure that it is built and working properly (Zenina and Merkuryev 2019). Olstam and Tapani (2011) states that verification of the simulation models should include software error checking, input coding error checking and an animation review. Software error checking means that all the known bugs in the software are considered. Input coding error checking reviews the deviation between input data and the output of a simulation model. Animation review is based on observation of vehicles in the simulation model to discover errors that the previous two methods could not find.
The calibration is the process of adjusting parameters in the simulation model until it gen-erates output values that are close enough to the observed values in a real transportation system (Zenina and Merkuryev 2019). The calibration procedure can be divided into three steps, traffic capacity calibration, route choice calibration and traffic performance calibration (Olstam and Tapani 2011). Traffic capacity calibration includes adjusting global and local links capacity in order for the simulation model to reproduce the same capacities as studied in the reality. Route choice calibration means adjustments of route choice parameters to simu-late the traffic flow on links correctly. Calibration of the traffic performance means parameter tuning in the simulation model to make it fit traffic performance indicators such as speed, travel times, queues, etc. (Olstam and Tapani 2011).
The model validation task is to approve that the simulation model is representative to the real-world traffic system that is modelled. This means however that it should not only be able to generate the output values included in the calibration data set, but also measurements collected outside of the calibration data set (Olstam and Tapani 2011). The validation is meant to be carried out after the calibration process. This is because of testing the logic and behavior of the simulation model and to see if the imitations of entities are accurate. If this would not be the case, the calibration step needs to be done again.
3.4
Simulation of Parking and Stopping
Parking availability in urban areas is essential for moderating congestion levels in traffic sys-tems. The time spent for vehicles searching for parking impacts the total travel time in a network. The number of searching vehicles sequentially leads to road congestion (Gallo, D’Acierno, and Montella 2011). Several parking equilibrium models have been formulated to catch different parking activity components and their relationship, as for instance the price of parking and the driving distance. According to Nourinejad, Wenneman, Habib, and Ro-orda (2014), it is however a drawback that these types of models lacks the parameters walking time (distance) and parking availability at different times throughout the day. The walking distance from a parking spot to the actual destination is important to consider, specifically for those that have scarce time. Further on, is cruising time one major parameter of the park-ing behavior that is unprocessed and incomplete when included in the equilibrium models.
3.5. Simulation of Information Sharing In Traffic Networks Benenson, Martens, and Birfir (2008) have tried to fill this gap by developing an adopted agent-based model of parking in the city. Their model, called PARKAGENT, simulates the behavioral and parking decision making of individual drivers in a specific spatial environ-ment. In the city of Tel Aviv, in Israel, the model has been tested with two groups of agents (residential and visitors). Travellers enter the constructed simulation model 250 m before their desired location and decrease their velocities to 20-25 km/h when they estimate their parking needs. The research model is structured and used to evaluate further increased num-ber of parking spots and its impact in a residential area.
To simulate traffic with various parking management conditions, Steenberghen, Dieussaert, Maerivoet, and Spitaels (2012) has developed another agent-based simulation model called SUSTAPARK. The model is a spatiotemporal tool (considers both time and space) to simulate both traffic and parking space searching behavior. The parking search behavior of agents (different car drivers) is determined by factors such as driving behavior, walking distance from the parking spot to the final destination (egress time) and parking availability. When the driver shifts from "driving" mode to the "parking search" mode, the model will suggest empty parking spots that is within reach of one simulation time step of the driver. Based on an utility function that considers the search time, egress time and expected parking fee, the driver decides whether to park or not at each suggested parking lot. The researchers have with various simulation scenarios evaluated parameters as search times for parking, parking occupancies and egress times for passengers. Steenberghen, Dieussaert, Maerivoet, and Spitaels (2012) found for instance that by implementing their base scenario model in the city of Leuven, Belgium, many vehicles parked on-street instead of using off-street garages.
3.5
Simulation of Information Sharing In Traffic Networks
In an environment where vehicles are connected, each vehicle would act as a traffic informa-tion collector and distributor. Vehicles behavior in the real world would then be changed and thus also bring changes to a traffic simulation model, which objective is often to mimic the real-world behavior (Dai, Y. Lu, Ding, G. Lu, and Wang 2019).
With newly introduced systems in the real world, traffic engineers also need new updated simulation tools that can handle this. Do, Rouhani, and Miranda-Moreno (2019) have re-viewed new developed car-following models for intelligent vehicles to be able to simu-late these new characteristics of vehicles. Conventional car-following models are based on human-driving characteristics and does not apply that well to new types of vehicles. Vehi-cles with access to information are for instance able to "see" beyond human driver visibility and thus make earlier decisions regarding the traffic state.
Information sharing vehicles have been evaluated in various simulation studies previously. For instance have Lee and Park (2008) used the microsimulation tool Vissim to evaluate a variety of route guidance strategies. The authors made a sensitivity analysis, studying the influence of factors such as the share of vehicles that enables information sharing, congestion levels of specific roads in a network, updating intervals of route guidance information and drivers’ compliance rates. One of the findings were that vehicles based on information shar-ing with route guidance significantly reduced the travel times over the case with no guidance. Vehicles that have information about the current road conditions and possible roads with ac-cidents ahead of them, can choose to re-route and select an alternative path leading to their destination. This means avoiding the present accident and decreasing their travel time. Another study by Kharrazi and Atif (2020) has used information sharing between infras-tructure and vehicles to develop an algorithm that uses cloud-based parking service data to
3.6. Vissim - Microscopic Simulation Software select the best parking spot in an area for vehicles. The proposed algorithm uses a discrete Markov-chain model to find future states of parking spot availability. The algorithm consid-ers historical parking occupancy data, current state of parking spots and congestion rates in an area. The outcome of their study showed that vehicles with access to information of park-ing spot availability had a clear time advantage of successfully findpark-ing a free parkpark-ing spot. This, compared to vehicles that did not have any information and just drove to the nearest parking spot in ambition to reduce their travel time.
3.6
Vissim - Microscopic Simulation Software
PTV Vissim is a widely used microscopic simulation tool, due to its complexity and ability to carefully simulate traffic conditions (Chen 2019). Vissim is a time-based simulation soft-ware that uses behavior models for microsimulation. It handles multi-modal transportation and can be applied for rural as well as urban areas or pedestrian flows (PTV GROUP 2019). Multi-modal simulation means that the simulation can handle more than one type of traffic simultaneously. Vehicles, pedestrians, bicycles, public transport, etc. can interact with each other. Complex and thorough traffic conditions can be simulated in various situations and it is possible to attain a variety of output evaluation data (Chen 2019). Users can conduct several analyses about intersection geometry, planning infrastructure development, capacity management, development of traffic signal control systems and the develop of public trans-port (Joelianto, Sutarto, and Antariksa 2019). How large the network is together with the traffic volume, complexity of intersections, signal controls etc. will affect the speed of the simulation (Chen 2019).
3.6.1
Driving Behavior
Vissim is according to its developer (PTV GROUP 2019) based on a traffic flow model and signal control models which simulates the movements of the vehicles in a network. The signal controllers are either fixed-time controls or traffic-actuated and are generally controlled and set up outside of the Vissim software. The behavioral sub-models that are used in Vissim to simulate the traffic flow are a car-following model and a lane-changing model.
The car-following model used in Vissim is a psycho-physical car following model developed by Wiedemann (1974). In this model there are four states a driver can be in (PTV GROUP 2019):
Free driving: The driver is not impacted on other vehicles. The speed of a vehicle in this state oscillates around the drivers desired speed.
Approaching:The driver is decelerating to reach, the drivers desired safe distance to a proceeding vehicle.
Following: The driver tries to keep its speed constant so that the desired safe distance towards the proceeding vehicle is kept.
Breaking:The driver is breaking and decelerating in a medium to hard phase.
In Vissim the entities are called driving-vehicle-units. For every unit the behavior of a driver, such as desired speed, desired acceleration/deceleration, desired safe distance etc. is con-nected to characteristics of a vehicle. Technical attributes that impact a unit, associated with a vehicle is for example the vehicles length, maximum speed and maximum acceler-ation/deceleration, but also the current speed and position.
3.6. Vissim - Microscopic Simulation Software Two types of lane changes can occur in Vissim, either is a lane change required by the route or it is a free lane change. A free lane change is triggered according to two factors. (1) If there is more space in the new lane and (2) if the driver assumes to be able to keep its desired speed for a longer time, compared to what the driver can have in the current lane. For both types of lane changes, a minimum headway distance between the lane changing vehicle and the new preceding vehicle needs to be respected.
3.6.2
Constructing a Traffic Simulation Model
This subsection is written in order to give a background to objects that can be used in Vissim to construct a traffic simulation model. Technical information about the objects are described in the software’s user manual (PTV GROUP 2019) and are here summarized.
To model a traffic network in Visism, Links and Connectors are used to represent roads. Links can consist of one or more lanes and by using points to divide a link, curved road can be mod-elled. With the use of Connectors, links can be connected to each other and by connecting Links, intersections can be constructed. Intersections with different priorities for approaching vehi-cles can in Vissim be modelled using Conflict Areas, which is automatically generated when two links or connectors are overlapping. The priority of the conflicting links/connectors can be set in four ways. (1) One of the links is set to have priority over the other which have to give way. (2) Vise verse. (3) Both links in a conflict can be set to give way. Then, both on-coming vehicles waits until sight of range and will continue in the order they arrived to the intersection. (4) Both of the links in a conflict can be set as passive meaning the vehicles will not give way at all. By using Conflict Areas for intersections, main roads can be modelled. In order to simulate a stop or a parking at one link in Vissim, the Parking Lot-object can be used together with a Parking Routing Decision. The Parking Lot-object controls the behavior of a vehicle that has decided to stop. Within the Vissim network object Parking Lots, elements such as blocking time distributions and parking directions can be set. The blocking time distributions refers to the time a vehicle is blocking the travel lane when conducting a parking maneuver. The parking direction refers to the direction of a vehicle while entering the space and leaving it. In Vissim the parking direction can be set to either "Forward > forward" or "Forward > reverse". Therefore, it is not possible to simulate a vehicle conducting a parallel park reversing into a parking space using the standard settings of Vissim.
The Parking Routing Decision connects a parking decision point at a link to one or more Parking Lot-objects and creates a route to the parking lot. Within the decision object the parking rate and the parking duration can be modified. The parking rate refers to a percent of how many vehicles that passes this decision point, actually decides to stop (follows the decision). This rate can be set differently depending on the Vehicle Class. The parking duration refers to the time a vehicle is at standstill at the parking space.
Another way to route vehicles in a Vissim network is by using Static Routing Decisions and Partial Routing Decisions. A route consist of a sequence of links and connectors which vehicles follows until the route ends. When a vehicle has left a route, it is considered free and can begin a new route or leave the network. Routing decisions can take the Vehicle Class into consideration, which makes it possible to route specific vehicles in a network differently. To simulate vehicles making a lane change towards, for instance, a curbside of a constructed network, Partial Routes can be used. These types of routes must be placed within a Static Route and can be controlled with formulas based on states of other objects element in Vissim, for example the occupancy of a Parking Lot. How a partial route can be implemented to simulate a lane change behavior is illustrated in Figure 3.1.
