FULLTEXT01

LiU-ITN-TEK-A--20/055-SE
Two different bus stop layout

designs - A traffic simulation
study in Vissim
Mohamed Elsayed
Erik Torstensson
2020-09-25
Department of Science and Technology Institutionen för teknik och naturvetenskap

Linköping University Linköpings universitet
SE-601 74 Norrköping , Sw eden 601 74 Norrköping
LiU-ITN-TEK-A--20/055-SE

designs - A traffic simulation
study in Vissim
The thesis work carried out in Transportsystem
at Tekniska högskolan at
Linköpings universitet
Mohamed Elsayed
Erik Torstensson
Norrköping 2020-09-25
Department of Science and Technology Institutionen för teknik och naturvetenskap

Linköping University Linköpings universitet
SE-601 74 Norrköping , Sw eden 601 74 Norrköping
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/
© Mohamed Elsayed, Erik Torstensson

Master of Science Thesis in Communications and Transport system
Department of Science and Technology, Linköping University, 2020

designs
- A traffic simulation study in Vissim
Erik Torstensson & Mohamed Elsayed

2
Master of Science Thesis
Erik Torstensson and Mohamed Elsayed
LiTH-ISY-EX--ET--YY/XXXX
Supervisors:
Ivan Postigo
ITN, Linkoping University
Mats Sandin
M4Traffic
Examiner:
Johan Olstam
ITN, Linkoping University
Communications and Transport system

Department of Science and Technology
Linköping University
SE-581 83 Linköping, Sweden
Copyright 2020 Erik Torstensson and Mohamed Elsayed

Abstract
Bicycles are one of the most efficient way to travel within short-distance trips due to
its relatively low maintenance and operation costs (Gao, Liu, & Feng, 2012). Besides
to their efficiency, bicycles provide more flexibility for their parking and its ease of
use. With the increasing transportation demand in major cities, transportation
authorities will encourage to use the bicycle more. However, accommodating bicycles
would require physical road geometry modifications such as introducing dedicated
bicycle-lanes. Cycling appears to be a sustainable form of transportation across
virous countries in Europe, as a result the authorities are planning and implementing
upgrades to make the transportation system safer, convenient and sustainable which
is necessary to encourage more people to use bicycles as a form of transportation.
This thesis is a case study which examines current traffic conditions on a bus stop at
Langholmsgatan in the city of Stockholm, Sweden and evaluates the effects of
different designs for bicycles and buses. At this bus stop in Langholmsgatan, the
bicycle lane is located to the right of the traffic road and to the left of the bus stop in
the upstream direction. Buses need to cross the bicycle lane in order to arrive and
departure the bus stop. Consequently, a conflict will also be created between bicycles
and buses that are crossing the bicycle lane.
In this thesis, an alternative design is evaluated in which buses and bicycles are
separated from each other. However, this will result in a new conflict between
bicycles and pedestrians. These two designs are evaluated in terms of travel time and
delay and the analysis was done using micro-simulation software VISSIM.
The study shows that the current design at the bus stop of Langholmsgatan should
be preferred over the alternative design when considering travel time and delay for
bicycles. If buses should be considered, the alternative design should be preferred
over the current design.
Keywords: microscopic, simulation, VISSIM, driving behavior, travel time, delay.

Acknowledgments
Firstly, we would like to extend our deepest gratitude to our supervisors Ivan Postigo
and Johan Olstam at Linkoping University for their support and feedback during this
thesis. And we would also like to thanks M4traffic in Stockholm for giving us the
opportunity to work with this thesis and especially our supervisors Mats Sandin and
Anders Bernhardsson. Without your knowledge, full support and enthusiastic
guidance we would never have been able to complete this thesis.
We also would like to express our very profound gratitude to our families and friends
for providing us with unfailing support and continuous encouragement throughout
our years of study and through the process of researching and writing this thesis.
This accomplishment would not have been possible without them. Thank you.
We would like also to thank PTV for providing us with the VISSIM software license.
Thank you all for making this possible.
Norrkoping, 2020
Erik Torstensson & Mohamed Elsayed
v
Table of content
1. Introduction ....................................................................................................................................... 11
1.1 Aim and research questions ................................................................................................... 3
1.2 Methodology ................................................................................................................................. 3
1.3 Study area ...................................................................................................................................... 4
1.4 Research Limitations and assumptions ............................................................................ 6
1.5 Structure of the report.............................................................................................................. 6
2. Bus stop layout guidelines .............................................................................................................. 8
3. Microscopic traffic simulation .................................................................................................... 11
3.1 Activities in a microscopic traffic simulation.............................................................. 12
3.2 Microscopic traffic simulation in Vissim........................................................................ 13
3.2.1 Car-following model in Vissim .................................................................................. 13
3.2.2 Lane changing .................................................................................................................. 15
3.2.3 Parameters of bicycles in Vissim.............................................................................. 16
3.2.4 Pedestrian behaviour modelling in Vissim.......................................................... 17
3.2.4.1 Pedestrian routing decisions ............................................................................ 19
3.2.5 Conflict modelling in Vissim ...................................................................................... 20
4. Data collection................................................................................................................................... 23
4.1 Speed and flow.......................................................................................................................... 24
4.2 Traffic signals ............................................................................................................................ 27
5. Simulation model development ................................................................................................. 29
5.1 Design A ....................................................................................................................................... 29
5.1.1 Bus Time Scheduling..................................................................................................... 32
5.1.2 Traffic Signals Modelling ............................................................................................. 33
5.1.3 Speed Modelling .............................................................................................................. 34
5.1.4 Vehicle Input ..................................................................................................................... 35
5.2 Design B ....................................................................................................................................... 35
5.4 Model Calibration .................................................................................................................... 38
5.4.1 Number of replications ................................................................................................ 39
vi
5.4.2 Motorized vehicles ......................................................................................................... 40
5.4.3 Bicycles ............................................................................................................................... 42
6. Results .................................................................................................................................................. 52
6.1 Current traffic demand .......................................................................................................... 52
6.1.1 Travel time......................................................................................................................... 53
6.1.2 Delay .................................................................................................................................... 55
6.2 Different bicycle flows and bus frequencies................................................................. 56
6.2.1 Travel time......................................................................................................................... 56
6.2.2 Delay .................................................................................................................................... 58
7. Discussion & Conclusion ............................................................................................................... 59
7.1 Discussion ................................................................................................................................... 59
7.2 Conclusion .................................................................................................................................. 62
Bibliography............................................................................................................................................ 64
Appendix .................................................................................................................................................. 66
Appendix A ......................................................................................................................................... 67
vii
List of Figures
Figure 1 – Bicycle lane on the left side of the bus stop in the upstream direction (Trafikverket,
2020 A). .................................................................................................................................... 2
Figure 2 – Bicycle lane on the right side of the bus stop in the upstream direction
(Trafikverket, 2020 A)............................................................................................................... 3
Figure 3 – Plan view of the study case (Långholmsgatan). ...................................................... 5
Figure 4 – Sideview of the study case (Långholmsgatan). ....................................................... 5
Figure 5 – Bike lane between the bus stop and traffic lane (Trafikverket, 2020 A). ................. 9
Figure 6 – Recommended bus stop for a higher number of passengers (Trafikverket, 2020 A).
............................................................................................................................................... 10
Figure 7 – Building a simulation model steps (Olstam, 2019) ................................................ 12
Figure 8 – Wiedemann 99 model (PTV AG, 2018). ................................................................. 14
Figure 9 – Social force model (Dimakis & Guðmundsdóttir, 2018). ....................................... 18
Figure 10 – An illustration of two different lambda values in the Social force model (Dimakis
& Guðmundsdóttir, 2018). ..................................................................................................... 19
Figure 11 – Example of a partial pedestrian routing (PTV AG, 2018).................................... 20
Figure 12 – Theoretical framework for the priority rule (PTV AG, 2018) ............................... 21
Figure 13 – Example of a roundabout with the use of the priority rule (PTV AG, 2018). ....... 22
Figure 14 – Shows the bus location & the location of video recording position. ................... 23
Figure 15 – Single vehicle data measurements (Treiber & Kesting, 2013). ............................ 24
Figure 16 – Shows the passing points of the speed estimation and distance between them
visualized as an arrow. ........................................................................................................... 25
Figure 17 – The cumulative distribution curve CDF for motorized traffic. ............................. 26
Figure 18 – The cumulative distribution curve CDF for bicycles. ............................................ 26
Figure 19 – Shows the location of three traffic signals. ......................................................... 28
Figure 20 – Vissim for Design A.............................................................................................. 30
Figure 21 – 3D view for Design A. .......................................................................................... 30
Figure 22 – Shows the priority rules location and the conflict between the bikes and the
entering buses from the bus stop. .......................................................................................... 32
Figure 23 – Shows standard bus size...................................................................................... 33
Figure 24 – Shows articulated bus size. ................................................................................. 33
Figure 25 – Shows the signal group for signal 1. ................................................................... 33
Figure 26 – Shows the signal group for signal 2 and 3. ......................................................... 33
Figure 27 – Desired speed distribution for motorized vehicles. ............................................. 34
Figure 28 – Desired speed distribution for Bicycles. ............................................................... 34
Figure 29 – Shows a 3D view for the Vissim scenario-2 model. ............................................. 35
Figure 30 – The conflict between pedestrians and bicyclists in Design B. .............................. 36
Figure 31 – Shows the first crosswalk of design B. ................................................................ 37
Figure 32 – Shows the second crosswalk of Design B. ........................................................... 38
Figure 33 – Speed estimation in Vissim.................................................................................. 39
Figure 34 – Confidence intervals for motorized vehicles on a 95% level. ............................... 41
viii
Figure 35 – Confidence interval for bicycles on a 95% level with default settings from COWI.
............................................................................................................................................... 43
Figure 36 – Confidence interval for bicycles on a 95% level with adjusted driving behaviour
parameters. ........................................................................................................................... 45
Figure 37 – Desired Acceleration, COWI settings. .................................................................. 46
Figure 38 – Desired deceleration, COWI settings. .................................................................. 46
Figure 39 – Desired acceleration, changes. ........................................................................... 46
Figure 40 – Desired deceleration, changes. ........................................................................... 46
Figure 41 – Confidence interval for bicycles on a 95% level with changes in desired
acceleration and deceleration. .............................................................................................. 47
Figure 42 – Speed reduction near the entry of bus stop and near the second traffic signal. . 48
Figure 43 – Speed reduction zones near the entry of bus stop after the modification. ......... 48
Figure 44 – Speed reduction at the first traffic signal and Desired speed decision................ 49
Figure 45 – Confidence interval for bicycles on a 95% level with modification of the desired
speed and speed reduction zones. ......................................................................................... 51
Figure 46 – Data collection for Travel time and delay in Vissim. ........................................... 53
Figure 47 – Average Travel Time per bicycles comparison of Design A and B. ...................... 54
Figure 48 – Average Travel Time per buses comparison of Design A and B. .......................... 54
Figure 49 – Average delay per bicycles comparison of Design A and B. ................................ 55
Figure 50 – Average delay per buses comparison of Design A and B. .................................... 55
Figure 51 – Average Travel Time per bicycle with input changes for bicycles and buses. ...... 57
Figure 52 – Average Travel Time per bus with input changes for bicycles and buses. ........... 57
Figure 53 – Average delay per bicycle with input changes for bicycles and buses................. 58
Figure 54 – Average delay per bus with input changes for bicycles and buses. ..................... 58
ix
List of Tables
Table 1 – Parameter description for the Wiedemann 99 following model in Vissim with the
corresponding values from COWI. ......................................................................................... 17
Table 2 – Mean speed estimation for bicycles and motorized vehicles. ................................. 27
Table 3 – Number of vehicles during 07:30-08:30.................................................................. 27
Table 4 – Shows the traffic signals are the cycle length, green time, red time and the red-
amber time. ........................................................................................................................... 28
Table 5 – Shows the average flow for different vehicles. ....................................................... 35
Table 6 – Number of replications calculation ........................................................................ 40
Table 7 – The average speed of the field measurement and the simulation output for
motorized vehicles. ................................................................................................................ 41
Table 8 – The average speed of the field measurement and the simulation output for bicycles
with default settings from COWI. .......................................................................................... 42
Table 9 – Adjusted driving behaviour parameters compared to COWI settings. ................... 44
bicycles with driving behaviour parameter adjustments. ...................................................... 45
bicycles with changes in acceleration and deceleration ........................................................ 47
bicycles with changes in desired speed and speed reduction zones. ..................................... 50
Table 13 – Shows the different input for Design A and B of Bicycles and Buses. ................... 56
x
1. Introduction
Bicycles are considered to be the most efficient transportation mode for limited-
distance trips due to its relatively low price and maintenance costs and no
running cost, as well as its flexibility in terms of paperwork to obtain and use a
bicycle, and easy to drive and its variable parking options. Bicycles are also
known for their zero-emission on road compared to other vehicular modes,
which makes it one of the most environmental-friendly transportation modes
(UN Environment, 2016). Over the past two decades, the use of bicycles as a mode
of transportation has increased worldwide and become more and more popular
(Aldred, Best, & Jones, 2016). Therefore, planning and building adequate
bicycles-related infrastructure has become a growing need in most cities, aiming
to meet the growing demand, and to encourage the use of bicycles for longer
distances.
As in most capitals and major cities , the population of Stockholm is increasing,
and so is the demand for transportation (Stockholms stad , 2019 B). The steadily
increasing road upgrade projects and transportation infrastructure such as
subways structures and buses facilities are insufficient to meet the increasing
transportation demand (Stockholms stad, 2020). With Stockholm’s limited space
for new roads and inability to provide additional lanes due to width of the road,
one of the option authorities across Europe are opting for is to provide dedicated
bicycle lanes. However, even with dedicate lanes, possible conflicts between
bicycles, motorized vehicles, and pedestrians might still exist, especially at
intersections and bus stops.
A common conflict between buses and bicycles occur when buses need to cross
through the dedicated bicycle lane to reach the bus stop as illustrated in Figure
1, where the red circles represent the points where the conflicts occur. In this
xi
design, the bicycle lane is located on the left side of the bus stop in the upstream
direction.
Figure 1 – Bicycle lane on the left side of the bus stop in the upstream direction (Trafikverket, 2020 A).
The buses stop frequently at specific locations with different acceleration and
deceleration. The frequent pulling in and pulling out of the buses from stops
create various impacts on the road users. The most noticeable impact is the
conflict between the buses and bicycles when bicycles use the outer lane and
buses need to cross the bicycle lane, which might delay the movements of
bicycles and create queues of bicycles behind the buses.
The conflict points, which are the red circles between the buses and bicycles in
Figure 1, are also hazardous spots that may increase the risk of accidents,
especially when the number of buses and bicycles are too high.
One way of dealing with this conflict is to locate the bicycle lane to the right side
of the bus stop in the upstream direction as shown in Figure 2. Moving the bicycle
to the right side in the upstream direction would create a new conflict between
bicycles and pedestrians who need to cross the bicycle lanes to reach the bus
stop.
However, each layout has its advantages and disadvantages. One way to measure
the performance and evaluate a bus stop layout design is by the travel time and
delay of different modes of transportation and pedestrian.
2
Figure 2 – Bicycle lane on the right side of the bus stop in the upstream direction (Trafikverket,
2020 A).
1.1 Aim and research questions

The aim of this thesis is to study and analyze two road layout designs of bus stop.
Both designs deal with the conflict between the position of the bus stop and the
dedicated bicycle lane. These layouts are presented in Figure 1 and Figure 2 and
will be referred to as Design A and B. The analysis will include comparisons of
the impacts caused by each design quantified in terms of travel time and delay
and delay for both bicycles and buses. The following research questions will be
investigated in order to achieve the aim of this thesis:
• What impact do the two layout designs have on the bus stop, bicycles and
buses in terms of travel time and delay?
• How can a conflict between bicycles, buses and pedestrians be modelled in a
micro-simulation software program?
• Which layout design should be preferred if the number of bicycles and buses
are increased?
1.2 Methodology
A literature review on bus stop layout design is first studied. The investigated
designs are as recommended by (Trafikverket, 2020 A). To assess the impact of
each design, a microscopic traffic simulation study is carried out. Both designs
are modelled in a microscopic traffic simulation software, in this case Vissim.
To carry out the microscopic traffic simulation study, a location with a similar
layout as those recommended in the literature review is selected as a study case.
3
Data from the traffic on the selected location is collected using video cameras and
the same is used to calibrate the model in Vissim. The result from this calibrated
model will then be compared to an alternative bus stop design.
The theory required to carry out a microscopic traffic simulation is studied and
presented as part of the theoretical background, which includes the calibration
method and statistical analysis for hypothesis tests.
The modelling of conflicts between bicycles and vehicles in Vissim is investigated,
and since one of the bus stops designs deals with conflicts with pedestrians, a
method to model this particular conflict is proposed.
Lastly, a prognosis for higher traffic volumes for both bicycles and vehicles is
performed to verify if one design is preferable over the other under different
traffic conditions. This will be referred to as experiments in this thesis.
1.3 Study area

The chosen area to study is Langholmsgatan Street, which is located in
Sodermalm area in Stockholm. The street includes a bicycle lane stretching from
south of Sodermalm to Kungsholmen and has the second largest volume of daily
cyclists in Stockholm, with approximately 15 220 passengers per day in 2018
(Stockholms stad, 2019 A). Studying a location that has a high number of bicycles
per day will have impact on travel time and delay. The layout of the bicycle lane
in this location is similar to Figure 1 where the conflict arises when buses need
to cross through the bicycle lane to reach the bus stop at Hornstull.
Figure 3 and Figure 4 show where the bus stop and the bicycle lane are located
in the study area. The red line in Figure 3 represents the bicycle lane, while the
green area marks the bus stop. There are also three traffic lights shown as red
circles in Figure 3.
4
Figure 3 – Plan view of the study case (Långholmsgatan).
Figure 4 – Sideview of the study case (Långholmsgatan).
5
1.4 Research Limitations and assumptions
For modeling and analysis simplification, cars, trucks and motorbikes will be
modelled and simulated but they will not be counted separately. Cars, trucks and
motorbikes will be counted together, and they will be referred to as motorized
vehicles. This is due to a low influence from motorized vehicles on the
performance of bicycles. Another assumption is that three weekdays of data
collection will be enough to show the morning peak hour and it is assumed to be
representative for the peak hour.
Also, the research limitations include the assumption that buses will only exit the
bus stop at the end of the platform while in reality, buses may leave the bus stop
at any available section of the platform. Another limitation is that the number of
pedestrians and their speed are not included in the model due to complications
of measuring and extracting these data from a video recording. An assumption is
that the number of pedestrians trying to board the bus will be set to 500
passengers per hour. This figure has been considered to represent the
pedestrians for the best possible results. Bus sizes considered in the model are
standard size and articulated bus size with fixed capacity set to 110 and 180
passengers respectively which is default setting in Vissim. The two bus sizes are
considered to reflect the actual two main types of buses observed on site. The
number of boarding and alighting passengers was assumed to be 30 and 40 for
standard and articulated size respectively, as measuring such activity was
impossible to be extracted from the recorded videos.
Another limitation is that the bicycles will only go in one direction which is
referred to as the upstream direction.
Finally, the study was limited to the modelling analysis and assessment, and no
full traffic safety assessment is considered.
1.5 Structure of the report

The work in this thesis is presented as follows:
The thesis starts with a literature review part which contains of chapter 2 and 3.
Chapter 2 is about the bus stop layout guidelines including the recommended
design from The Swedish Transport Administration for bus stops. Chapter 3
covers traffic simulation with focus on microscopic traffic simulation in Vissim
including driving behavior, lane changing and parameters of bicycles in Vissim.
6
This chapter includes pedestrian behavior and conflict modelling in Vissim.
Chapter 4 describes how the data was collected. Chapter 5 illustrates how the
model is developed in Vissim along with its calibration. Chapter 6 includes result
and analysis about Design B and the experiments of Design A and B. Chapter 7
contains the discussion and conclusion.
7
2. Bus stop layout guidelines
In this chapter, relevant literature for this thesis is presented. The chapter
provides information about bus stop layout guidelines.
Krykewycz (2010) states that conflicts between buses and bicycles particularly
occur at points where buses cross bicycle lanes to pull into or pull out from bus
stops. Bicycles can be forced to sway unsafely into motorised lanes. Keeling, Glick
& Miguel (2019) also states that, it is complicated when buses and bicycles share
the same space, as this not only magnifies the bicycles safety risks, but also it
creates significant delays for buses beside the hassle on bus drivers to interact
with the surrounding bicyclists. On city streets, bicycles and buses are in several
ways natural enemies and they often operate in the same space (the right side of
the street) and at roughly the same speeds over significant stretches of road and
this could lead to a variety of conflicts between bicycles and buses ( Delaware
Valley Regional Planning Commission(DVRPC), 2009). Either bicycles or the
buses could have right of way on another and law might be different from each
country.
In USA, the law says that neither vehicle has universal priority and the vehicle
being overtaken has (Krykewycz, 2010). This means that if a bus is attempting to
cross a bicycle lane, it needs to yield for a potential bicycle or vice versa.
Buses should not accelerate around a bicyclist and then cut them off while
curbing. On the other hand, bicyclists should not overtake a bus as it approaches
an intersection and expect the bus to yield. According to Swedish traffic
regulations, buses driving in cities have priority when pulling out from a bus stop
and other traffic must give way for the bus to reach the far-right lane of the road
(NTF, 2020).
There have been a variety of strategies implemented worldwide to address the
specific bike and bus conflict but DVRCP (2009) mentions some various
strategies in order to solve the conflict.
8
One strategy is referred to as ‘’Coloured bike lanes in conflict hotspots, including
transit stop areas’’. This strategy has been used by a number of cities with
coloured lane treatments in bicycle lanes and especially where vehicles cross and
conflicts are common. Another strategy was called ‘’Physical re-routing of bike
lane around stop location’’. This strategy involves the bicycle lane to be routed
onto sidewalk, outside the bus stop. However, this strategy would have limited
applicability to cities with typical narrow and small streets. But the strategy
might be appropriate in cities where the rights-of-way are wider and heavy
bicycle traffic might justify the expense.
The Swedish transport administration has a unit called VGU (vagar och gators
uformning) which develops rules and recommendations for a street or a road
(Trafikverket, 2020 A). The rules are compulsory when working with the state
roads. For the municipalities, VGU is voluntary and provide advisory publications
(Trafikverket, 2019 B).
The choice of bus stop type in an urban environment depends according to
Trafivkerket (2020 A) on the amount of traffic, but also on the reference speed,
pedestrian and bicycle traffic, local priorities, the nature of the city, etc. One way
to design the bus stop with passing bicycle traffic is to place the bicycle lane to
the left of the bus stop in the upstream direction and merged with traffic. If this
design is applied, the buses must cross the bicycle lane in order to reach and exit
the bus stop, see Figure 5. The bicycles must yield for crossing buses so that the
buses can change lanes.
This design is mainly used on the main network in urban areas when car traffic
is prioritized, and bicycle traffic is extensive. Advantages with this solution is that
buses do not block cars, trucks etc. when it is at the bus stop and there is
relatively good safety and comfort for the passengers waiting for the bus . One
disadvantage is uncomfortable driving to get to the bus stop for the buses.
Figure 5 – Bike lane between the bus stop and traffic lane (Trafikverket, 2020 A).
9
VGU (2020 A) states that car traffic should be separated from pedestrians and
bicycles along bus stops to improve the level of safety and to avoid the risk of
accidents, given the relatively high traffic speed. Higher flow and speed increase
the importance of the separation between them.
One way to separate the bicycles and buses is to place the bicycle lane to the right
of the bus stop in the upstream direction if enough safety devices (preferably in
the form of fences) are available to reduce the conflict with pedestrians. The
pedestrians need to cross the bicycle lane in order to reach the bus stop. In this
way, a new conflict is created between pedestrians and bicycles where bicycles
must yield for crossing pedestrians at the crosswalks. This is illustrated in Figure
6. Cross walk is provided on either side of the stop where the railing starts and
ends (Trafikverket, 2020 A).
Placing the bicycle lane to the right of the bus stop would eliminate all bus-bicycle
conflicts. Unfortunately, this configuration is considered to be the best for wide
roads, as it requires a huge amount of right-of-way (Miguel , Keeling, & Glick,
2019).
Figure 6 – Recommended bus stop for a higher number of passengers (Trafikverket, 2020 A).
10
3. Microscopic traffic
simulation
In this chapter, relevant literature for this thesis is presented. This chapter
consists of two sections, the first section gives a brief summary of the activities
in a microscopic traffic simulation and how it can be carried out. The second
section will include information about how microscopic traffic simulation can be
modelled in Vissim.
A traffic micro-simulation model consists of sub-models and various parameters
which describe the driving behavior of humans. Examples of sub-models are
lane-changing and car-following.
Lane-changing models describe the behavior of drivers when it is critical to
decide whether to change lane or not. The car-following model describes the
interactions with preceding vehicles in the same lane (Janson Olstam & Tapani,
2004).
11
3.1 Activities in a microscopic traffic simulation
When building a microscopic traffic simulation model, a few steps can be
formulated in order to finish and make conclusions. Figure 7 shows an example
from Olstam (2019) of the steps in a traffic simulation.
Figure 7 – Building a simulation model steps (Olstam, 2019)

The development of the base model will be further discussed in chapter 5. The
adjusted parameters will be also presented in model calibration section 5.4.
Verification is a process to ensure that no mistakes have been made, e.g.
programming errors, encoding errors, input data errors (Olstam, 2019).
Compare model to data is a comparing of data set where three situations are
needed according to Olstam (2019); calibration, alternative analysis and
12
validation. Calibration is the comparison of a set of simulation output data with
a set of measurements to improve the accuracy of the model. Alternative analysis
is the comparison of two (or more) sets of simulation output data from different
scenarios to find the “best”. Validation is the comparison of a set of simulation
output data with a set of measurements to test the calibration. In order to
compare the data, statistical methods can be applied for comparison of data sets
like confidence intervals and statistical hypothesis test which will be presented
in section 5.4. Another part of the calibration is to determine the number of
replications which are needed to run the model. The process of determining
number of replications will be further discussed in section 5.4.
3.2 Microscopic traffic simulation in Vissim

In this section, necessary information regarding how a traffic simulation of a bus
stop in Vissim is presented. The first and second subsection are sub-models
which are used to describe driving behavior. The first sub-model is referred to
as the car-following model. A car-following model controls driver’s behaviour
with respect to the preceding vehicle in the same lane. The second sub-model
includes information about the lane changing which describes drivers’ behaviour
when deciding whether to change lane or not (Janson Olstam & Tapani, 2004).
The third subsection illustrates previous studies of different bicycle parameters
in Vissim. The fourth section consists of how pedestrian can be modelled in
Vissim. The fifth subsection includes information about how a conflict between
bicycles, buses and pedestrians can be modelled.
3.2.1 Car-following model in Vissim
In Vissim, two car-following models can be used which are Wiedemann 74 and
Wiedemann 99. The main difference between the two car following models
according to Fransson (2018), are that Vissim tries to create a more diverse
driver population. The Wiedemann 99 model is also more up to date and popular
(Gao, Liu, & Feng, 2012).
Wiedemann 99 shows the processes of human perception and decision making.
The model avoids sudden changes in speed and contains a threshold for
perception to identify realistic behavior of road users (Hamm, 2016).
Wiedemann 99 describes the psycho-physiological aspects of the driving
behavior in terms of four discrete driving regimes: free flow, approaching slower
13
vehicles, car-following near the steady-state equilibrium, and critical situations
requiring stronger braking actions (Treiber & Kesting, 2013). The regimes can be
defined by different thresholds which can be seen in Figure 8 from PTV (2018).
Figure 8 – Wiedemann 99 model (PTV AG, 2018).

Ax is the desired distance between two standstill vehicles. ABX is the minimum
distance between two vehicles which are traveling at equal speed. SDX is the
maximum following distance. SDV is a point when a driver realizes he is close to
a vehicle. CLDV is the point where a driver is aware of small differences in speed
when the distance between the vehicle Infront decreases. OPDV is the point when
a driver realizes that his speed is slower than the vehicle Infront (Franson, 2018).
The thresholds can also be summarized according to Palmqvist (2015) as:
• Following
When the driver catches up a slower-moving vehicle, he adapts the driver
speed and brakes to finally reach the safety distance the driver is aiming for.
• Free driving
The vehicle is not affected by the vehicle ahead but the driver trying to
reach their target speed. In fact, this velocity cannot be kept completely
constant but oscillates, which is built into the model.
• Closing in
When the driver catches up a slower-moving vehicle adapts the driver
14
speeds up this and slows in to eventually end up on the safety distance that
the driver is trying to reach.
• Braking
Braking occurs if the vehicle comes closer to the front vehicle than the
desired safety distance. This happens when the vehicle in front is slowing
down or if file changes are made in front of the driver, which in turn affects
the speed of the driver.
The Wiedemann 99 model, according to Fransson, expressed mathematically as
Equation (1).
𝐶𝐶8 − 𝐶𝐶9
𝑢𝑛 (𝑡) + 3.6(CC8 + 𝑢𝑛 (𝑡))∆𝑡
80
𝑢𝑛 (𝑡 + ∆𝑡) = 𝑚𝑖𝑛 𝑠𝑛 (𝑡) − 𝐶𝐶0 − 𝐿𝑛−1 (1)
3.6
{ 𝑢𝑛 (𝑡)
𝑢𝑛 (𝑡 + ∆𝑡) is equal to the minimum of two speeds. The upper speed considers
the restriction of the acceleration and the difference between CC8 and CC9. CC9
is the maximum acceleration when the vehicle is driving at 80 km/h and CC8
maximum acceleration when the vehicle is driving at 0 km/h. The other speed is
calculated when the model is in the condition of steady state with CC0 and the
distance(L) between two vehicles. CC8, CC0 and CC9 are model parameters
among other parameters that are used in the Wiedemann 99 model.
3.2.2 Lane changing
The decision of making a lane change for a vehicle is quite complex. The first lane
change model was developed by Gipps in 1986 where three questions can be
formulated:
• Is a lane change possible?
• Is a lane change necessary?
• Is a lane change desirable?
The first question must undoubtedly always be answered with yes because in
order to make a lane change since collision might otherwise occur. The other two
questions are not necessarily easy to answer and might not be relevant in all
cases. If a lane change is necessary or desirable depends mainly on factors like
the current speed and lane. Some lane types might require specific decisions or
maneuvers when making a lane change.
Lane changing in Vissim consists of two different approaches: Necessary lane
change and Free lane change. For the Necessary lane change, Vissim checks the
15
desired safety distance to the trailing vehicle on the new lane. The desired safety
distance depends on the speed of the vehicle that wants to change the lane and
on the speed of the vehicle preceding it. For a necessary lane change, the driving
behavior parameters contain the maximum acceptable deceleration for a vehicle
and its trailing vehicle on the new lane. The deceleration depends on the distance
to the emergency stop position of the next route connector (entrance ramp).
Regardless of which lane change model that are used, each model first needs to
find a suitable gap in the direction of travel. The gap size depends either on the
speed of the vehicle changing the lane or the speed of the vehicle approaching
from behind on the lane (PTV AG, 2018).
3.2.3 Parameters of bicycles in Vissim
Microscopic traffic simulation has most been used to study the ability to model
road user but most studies have not achieved this for bicycles (COWI, 2013).
COWI (2013) did a bicycle study in Copenhagen and the report was according to
Best, Aldred and Jones (2016):’’ A successful application to model cyclist flows’’.
The goal from the study was to translate the results from the data collection into
updated and validated parameters that can be used to simulate cyclists in VISSIM.
The study has shown that the spread of the speed distribution is larger than the
default settings. Acceleration seemed to be too high compared with cars and was
reduced.
COWI (2013) identifies ten key parameter groups/areas for the microsimulation
of bicycles; vehicle characteristics, speed distributions, acceleration (and
deceleration) distribution, following parameters, overtaking parameters,
behaviour at narrowing sections, behavior at bus stops, behaviour in waiting
zones, behaviour at stop lines and behaviour at right turns.
The group ‘’Behavior at bus stops’’ shows different rules and parameters which
should be used at bus stops. The parameters include desired speed and desired
speed reductions for bicycles. It also shows that there was a difference in the
behavior of bicycles at small and large bus stops. Most bicycles reduce their speed
when they approach small bus stops and make weaving maneuvers when
passengers enter or leave a bus, while the rest of bicycles make a full stop.
Bicycles at large bus stop are more likely to make full stops
The following includes different parameters in the car-following model as shown
in Table 1 with the corresponding bicycle values from the COWI (2013) study.
16
Table 1 – Parameter description for the Wiedemann 99 following model in Vissim with the
corresponding values from COWI.
Parameter Description COWI values
The standstill distances. This parameter is adjusting
CC0 0.20 m
the distance from vehicle to vehicle at zero speed.
Headway time. The parameter adjusts the time
CC1 0.5 s
headway ( time between two following vehicles ).
Restricts the distance difference a driver allows for
CC2 before he intentionally moves closer to the car 2.0 m
preceding him.
Threshold for entering “following”. The value of this
parameter explains in what distance the vehicle
CC3 moves into the state of following another vehicle. -20
When the distance exceeds this threshold. the
vehicle is affected by the vehicle in front.
Negative “following” threshold. Defines negative
speed difference during the following process. Low
CC4 values result in a more sensitive driver reaction to -0.25
the acceleration or deceleration of the preceding
vehicle.
CC5 The same as CC4 but positive. 0.25
This parameter decides how much a vehicles speed
CC6 is varied depending on the distance to the vehicle in 1 m/s2
front.
This parameter describes how much the acceleration
CC7 0.2 m/s2
is depending on the distance to the vehicle in front.
Standstill acceleration. The acceleration a vehicle
CC8 1.8 m/s2
has when starting from standstill.
Acceleration 80 km/h. The acceleration of a vehicle
CC9 when it is traveling in 80 km/h. 0.01 m/s2
3.2.4 Pedestrian behaviour modelling in Vissim
Vehicles and pedestrians move on links in VISSIM. Since pedestrians can move
sideways compared to vehicles, it is not appropriate to model them on links.
Instead they can be modelled on area elements, which have the ability for
movements in all directions (EIDMAR & HULTMAN, 2014). In 1995, Helbing and
Molnar developed a model for the behavior of pedestrians called Social force
model. Each pedestrian is appointed with a desired destination and a desired
speed which is called driving force. The desired destination demands a route for
each pedestrian. This route involves a possible obstacle or other pedestrians
which will impact each pedestrian with a force called repulsive force. The forces
can be visualized by Figure 9 from Diamakis and Guðmundsdottir (2018).
17
Figure 9 – Social force model (Dimakis & Guðmundsdóttir, 2018).
Diamakis and Guðmundsdottir (2018) also include the behavior of pedestrian to
be based on three levels of detail:
• Strategic
The strategic level includes a pedestrian plan which operates on the level of
minutes and hours where pedestrians are predefined with origin,
destination and desired speed.
• Tactical
The tactical level where pedestrian decide their route on the seconds to
minute level. This level also takes into consideration of the routing conditions
which include obstacles and another pedestrian.
• Operational
The operational level decides the route by millisecond to seconds where it

tries to avoid obstacles. The calculation is made constantly during the
simulation and by every interaction with obstacles and pedestrian.
The social force model controls the operational level and parts of the tactical level
in Vissim. The social force model consists, according to Diamakis and
Guðmundsdottir (2018) of two parameters; local and global. The global
parameters affect each pedestrian while the local parameters affect individually.
18
Especially tau and lambda are important parameters for the behavior of
pedestrians. Tau represents the relaxation time or inertia that can be related to a
response time, as it couples the difference between desired speed and desired
direction. Lambda governs the amount of anisotropy of the forces from the fact
that events and phenomena in the back of a pedestrian do not influence him (PTV
AG, 2018).
An illustration of two different lambda values can be seen in Figure 10 from
Diamakis and Guðmundsdottir (2018). When lambda is equal to zero, something
that happens in the back, has no impact at all on the pedestrian. When lambda is
equal to one, every action from all surrounding pedestrian or obstacle has an
effect.
Figure 10 – An illustration of two different lambda values in the Social force model (Dimakis &
Guðmundsdóttir, 2018).
3.2.4.1 Pedestrian routing decisions

Pedestrians can walk in the same direction and on every area of the pedestrian
link and area. This can lead to problem when all pedestrians want to walk on the
same place of the pedestrian link. One way to force the pedestrians to walk in a
specific order is to use partial pedestrian routing decisions (PTV AG, 2018).
Partial pedestrian routing decisions start from a certain point and ends with a
destination point. Partial pedestrian routing serves the local distribution of
pedestrians without changing the pedestrian OD matrix. It is possible to increase
the destination points in order to force the pedestrians to walk exactly as
preferred (PTV AG, 2018). In Figure 11, an example from PTV (2018) of a partial
pedestrian routing is illustrated. The route course is shown as a yellow line and
the blue dot is an intermediate point which traverse the pedestrians.
19
Figure 11 – Example of a partial pedestrian routing (PTV AG, 2018).
3.2.5 Conflict modelling in Vissim
A conflict between two vehicles or a vehicle and a pedestrian can be modelled in

Vissim either using the priority rules or conflict area. The priority rule is not
controlled by signals and is used in situation when vehicles in different links or
connectors need to consider each other (PTV AG, 2018). Figure 12 from PTV
(2018), shows the theoretical framework of the priority rule. A headway distance
and a gap time is used to control the vehicles. If a vehicle has entered the headway
distance or if the gap time is not satisfied, vehicle on the other link will stop at
the stop line. Gap time is the required time for a vehicle to reach time conflict
marker with its current speed (PTV AG, 2018). The priority rule according to
Virginia Department of Transportation (2020), is often used when geometric
complexities can make conflict areas difficult or impossible to apply (Virginia
Department of Transportation , 2020).
20
Figure 12 – Theoretical framework for the priority rule (PTV AG, 2018)
Conflict areas are easier to be applied to a network and are enough for dealing
with a lot of vehicle types. Conflict areas can be located wherever two links are
emerging. Vissim will automatically decide and determine the priority for each
vehicle depending on the link that has right of way. The status of each conflict
area will be showed by a color. Green indicates main flow (right of way) and red
indicates minor flow(yield). If both are red, there is no right of way, as vehicles
simply remain in their original sequence. If both are yellow, it is a passive conflict
area without right of way. The conflict area is be based on some attributes such
as speed, look ahead distance, time gap, safety distance etc. Each calculation
consists of a calculation where a vehicle from a minor traffic stream tries to enter
a mainstream. In this case, safety distance is important and if this distance is too
small, the will vehicle will decelerate if it must stop in front of the conflict area.
Each time a vehicle tries to enter the main flow, braking is either cancelled or the
driver continues driving and might even accelerate, e.g. when finding a gap in the
traffic stream to enter. In Figure 13 from PTV (2018) , the conflict area is applied
to a roundabout with difference status of each conflict area.
21
Figure 13 – Example of a roundabout with the use of the priority rule (PTV AG, 2018).
22
4. Data collection
The field observations were obtained from video cameras and were used as input
for the analysis using VISSIM. Data was collected by extracting the relevant data
from the cameras that were installed on top of a pedestrian bridge across
Langholmsgatan (the road on which the studied bus stop is located), presented by
the red circle in Figure 14. The data extraction was carried out for three different
days during the morning peak hour between 07:00 AM and 08:30 AM at 4th, 24th
and 25th of March 2020. Pavement markings, road signs,, geometry and other
features were obtained from Google Earth. This chapter gives summary of the
collected data which includes traffic flow (volume), speed and traffic signals
timing.
Figure 14 – Shows the bus location & the location of video recording position.
23
4.1 Speed and flow
One of the parameters that is to be estimated was the speed. The speed can be
estimated according to Treiber and Kesting (2013) by the help of two quantities:
• The time 𝑡𝛼0 from where a vehicle is passing a point.
• The time 𝑡𝛼1 from where a vehicle is passing the next point.
By assuming an average vehicle length L, it is possible to estimate a speed for an

individual vehicle by using the time difference between the first single point and
the next point. The vehicle length L can also be interpreted as the distance
between two points. Equation (2) defines the relationship between these
parameters.
𝐿
𝐿𝛼 = 𝑣𝛼 (𝑡𝛼1 − 𝑡𝛼0 ) −→ 𝑣𝛼 = (𝑡 1 −𝑡
𝛼
0) (2)
𝛼 𝛼
The relationship can also be visualized when a single vehicle is passing two points, see
Figure 15 from Treiber and Kesting (2013).
x
𝑡𝛼0 𝐿𝛼 𝑡𝛼1
𝑣𝛼
Figure 15 – Single vehicle data measurements (Treiber & Kesting, 2013).

The field speed observation was estimated using Equation (2) by considering two
identified points.
24
The green circles in Figure 16 show the location of the two points which are used
to estimate the speed for a single vehicle. The time needed to pass through the two
points gives the travel time used in Equation (2), with a total distance of 75
meters.
Figure 16 – Shows the passing points of the speed estimation and distance between them visualized
as an arrow.
This process was made for 100 vehicles each day for three days and for vehicle
type (bicycles and the motorized vehicles) as a sample test under the assumptions
that most of the motorized traffic share identical speed. In total, the speed of
300 bicycles and motorized traffic were estimated and can be visualized in a
cumulative distribution curve (CDF). A CDF curve shows a distribution of a certain
quantity and a probability that the quantity will take a value less than or equal to
zero. Also notice that the speed for buses could not be measured due to difficulties
of measuring it in this area separately. Figure 17 shows the CDF curve for the
observed speed of the 300 estimated motorized traffic and Figure 18 shows the
CDF curve for the observed speed of the 300 estimated bicycles.
25
CDF for the speed of 300 motorized traffic
1
0.9
0.8
0.7
Percentage (%)
0.6
0.5
0.4
0.3
0.2
0.1
0
13.00 18.00 23.00 28.00 33.00 38.00 43.00 48.00 53.00
Speed (km/hr)
Figure 17 – The cumulative distribution curve CDF for motorized traffic.
CDF for the speed of 300 bicycles

1
0.9
0.8
0.7
Percentage (%)
0.6
0.5
0.4
0.3
0.2
0.1
0
10.00 12.50 15.00 17.50 20.00 22.50 25.00 27.50 30.00 32.50 35.00 37.50
Speed (km/hr)
Figure 18 – The cumulative distribution curve CDF for bicycles.
The estimation of the mean speed for bicycles and motorized vehicles can be
visualized in Table 2. The number of bicycles (volume), motorized vehicles and
26
buses between 07:30 and 08:30 for each day can be seen in Table 3. It should be
noted that in Table 3, buses are not included in motorized vehicles and are
measured separately.
Table 2 – Mean speed estimation for bicycles and motorized vehicles.
Mean speed estimation
Mean speed bicycles Mean speed for

Day (km/h) motorized vehicles
(km/h)
4 of mars 2020 19.93 35.16
24 of mars 2020 20.83 36.20
25 of mars 2020 21.58 35.60
Table 3 – Number of vehicles during 07:30-08:30.

Time Bicycles Motorized vehicles Buses
4 of mars 2020 479 1044 37
24 of mars 2020 497 822 16
25 of mars 2020 525 948 18
Average 500 938 25
4.2 Traffic signals

The study area has three traffic signals, the first one is for the intersection in the
beginning of the study area while the other two are midblock signalized
pedestrian crossing signals. Figure 19 shows the different traffic signals, which
are referred to as signal 1, 2 and 3. Considering the traffic signal is important
because it dictates when bicycles and buses might interact. Due to the traffic
signal, bicycles will always arrive in fleets to the bus stop which may delay a big
proportion of the bicycles when they interact with a bus. The measured time from
the traffic signals are the cycle length, green time, red time and the red-amber
time. The measured time for cycle length and green time can be seen in Table 4.
Red-amber time corresponds to the remaining time of the flashing signal.
27
Table 4 – Shows the traffic signals are the cycle length, green time, red time and the red-amber time.
Signal 1 Signal 2 Signal 3
Cycle Length (sec) 100 100 100

Green Time (sec) 26 68 68
Red Time (sec) 68 26 26
Amber (sec) 3 3 3
Red-Amber (sec) 1 1 1
Figure 19 – Shows the location of three traffic signals.
28
5. Simulation model
development
This chapter describes how the model was developed using Vissim along with
related calibration for the model. The chapter starts with the development of
Design A and B and ends with a calibration.
5.1 Design A
The layout of Design A can be seen in Figure 20 while Figure 21 shows a 3D view
of the model. The bicycle path is provided to the left of the bus stop in the
upstream direction of traffic. The lanes for cars, trucks and buses are located to
the left of the bicycle path in the upstream direction. The geometry of the baseline
model is based on estimation from google maps and google earth. In some cases,
the lane widths, length of lanes and crosswalk widths were measured directly
from site using a meterstick. The green area represents walking area for the
pedestrians which also consists of an entry and exit area. The blue area represents
waiting area where pedestrians wait for the arriving bus. The pink area is
representing an area where pedestrians can enter a bus. The yellow area
represents speed reduction zone in where vehicles will reduce or their speed
while maneuvering a curve or a making a turn.
Red and green area in the model represents conflict areas and are used to dictate
the right of way when two links merges. Objects on the red colored link yields to
the object on the green colored link. In the case of ‘’Design A’’, conflict areas are
only applied to the conflict between buses and cars where buses will have the right
of way when entering the traffic lane.
29
Figure 20 – Vissim for Design A.
Figure 21 – 3D view for Design A.

30
Design A is modelled with default settings from Vissim (2018) regarding driving
behavior parameters, acceleration and deceleration for motorized vehicles.
Driving behavior parameters, acceleration and deceleration for bicycles are set
according to COWI (2013).The pedestrian behavior parameters will also be set to
default according to Vissim (2018).
As mentioned in section 3.2.5, modelling conflicts in Vissim can either be done
with priority rules or with conflict areas. The conflict between bicycles and buses
is modelled using the priority rule. Site observations show that bicycles always
yield for buses when a bus will try to cross the bicycle lane. Site observations show
that buses only give way to bikes if the bicycles are right next to the bus, however,
if the bicycles are behind or in front, the bus will not give any priority to the bikes.
This two phenomena or observations are easier to model with the priority rule
since it is possible to use markers and lines in order to dictate where buses and
bicycles must stop. Figure 22 shows the priority rules when buses need to enter
the transit stop. Notice different stop lines, headway and conflict markers which
will govern that bicycles or buses yield for each other.
A problem occurred when buses had left the conflict marker and a bicycle was
close to the stop line. Bicycles would continue to drive even though it can lead to
a crash with the oncoming buses. This was solved by implementing an additional
priority rule for bicycles where buses will give way for bicycles. This priority rule
can be seen as the blue text boxes in Figure 22. The distance headway is 5.12
meters for the priority rule for buses (black boxes) and 2.5 meters for the priority
rule for the bicycles (blue text boxes). The time gap is also implemented, which is
set to 1 second for each priority rule.
31
Figure 22 – Shows the priority rules location and the conflict between the bikes and the entering
buses from the bus stop.
5.1.1 Bus Time Scheduling
The bus time scheduling in the model was based on the extracted data from the
recorded videos. Two types of buses were observed and considered in the Vissim
model. A standard bus size (red color) with a 110-seat capacity and an articulated
bus size (blue color) with 180-seat capacity. The purpose of using two bus sizes is
that the articulated bus is longer which will affect the time when bicycles must
yield and wait for the bus to cross the bicycle lane. A longer bus needs more time
than a shorter bus to cross the bicycle lane. The layout of the two buses is
represented in Figure 24 and Figure 23 respectively. The buses are scheduled to
arrive 13 times per hour for the standard bus and 12 times per hour for the
articulated bus in a frequency of 5 minutes per bus type.
32
Figure 24 – Shows standard bus size. Figure 23 – Shows articulated bus size.
5.1.2 Traffic Signals Modelling
The traffic signals in VISSIM are modelled according to the information in section
4.2. Signal 1 has a different set of parameters for the cycle length, green time, red
time and the red-amber time. Signal 2 and 3 have identical measured time
property regarding green time, red time, amber time and red-amber time. All the
three signals start at the same time but signal 1 has longer green time based on
the observation data as shown in Figure 25 and Figure 26 respectively. All the
signals were assumed to operate as fixed time signals.
Figure 25 – Shows the signal group for signal 1.
Figure 26 – Shows the signal group for signal 2 and 3.

33
5.1.3 Speed Modelling
The speed that was assigned in the model is modelled as a cumulative distribution
function (CDF). The CDF for bicycles and motorized vehicles will be modelled in
such a way to mimic the CDF of bicycles and motorized vehicles found in figure 17
and 18 in data collection. The CDF for motorized vehicles and bicycles can be
seen in Figure 27 and Figure 28. The y-axis represents the probability while the x-
axis represents the speed. Each vehicle will be assigned to a speed based on a
probability when they enter the network.
Figure 27 – Desired speed distribution for motorized vehicles.
Figure 28 – Desired speed distribution for Bicycles.

34
5.1.4 Vehicle Input
For Design A, the hourly traffic volume considered for the vehicle input is based
on the average of the collected data. The flow for bicycles, buses and motorized
vehicles (cars and trucks) can be seen in Table 5. Buses, bicycles and motorized
vehicles in the network will be generated 70 meters before the first traffic
light(intersection Langholmsgatan and Horsnagatan).
Table 5 – Shows the average flow for different vehicles.

Motorized
Bicycles Buses
vehicles
Flow(vehicles/hour) 500 26 938
5.2 Design B
In Design B, the bicycle lane is shifted to the right side of the bus stop and to the
left of the walking area (green area) in the upstream direction as shown in Figure
29. Pedestrians who are entering or leaving the bus stop must pass the bicycle
lane. In order to pass the bicycle lane, pedestrians must use a crosswalk which is
located in the beginning and at the end of the bus stop in the upstream direction.
This new design is based on the recommendations from VGU (2020 A) when there
is a high pedestrian flow, see Figure 6 in chapter 2.
Figure 29 – Shows a 3D view for the Vissim scenario-2 model.
Design B does not have any conflict between bicycles and buses, though it has an
additional conflict between pedestrians and bicycles. The conflict occurs at the
35
two crosswalks located in the beginning and at the end of the bus stop. The conflict
point is illustrated by the red circles in Figure 30, where pedestrians will
encounter oncoming bicyclists. Pedestrians will, according to trafikverket (2019
B) ,always have the right of way when passing a crosswalk. This means that
bicycles in this design, always must yield for pedestrians.
Figure 30 – The conflict between pedestrians and bicyclists in Design B.

In Design B, the bicycle lane needs to be relocated. The current pedestrian area
(Green and blue area) and the walking area needs to be reduced, however, it is
assumed that the walking area (green area) is still sufficient to maintain the
pedestrian volumes. The new conflict in this design will also be modelled with a
priority rule at the crosswalks. Speed reduction zones will also be included close
to the crosswalks since the bicycle lane is curved before and after the crosswalks,
this is necessary to slow down the cyclists in order to enable them to safely
maneuver the curve. Signal heads for bicycles will be ignored in this design since
the bicycle lane is thought to be a part of the walking area.
Figure 31 shows the first crosswalk in the upstream direction (beginning of the
transit stop) and Figure 32 is the second crosswalk in the upstream direction (end
of the transit stops). Notice the pedestrian routing decisions, which will force the
arriving pedestrians to walk on the right side of the crosswalk and the departing
pedestrian will walk on the left side of the crosswalk, see Figure 31 and Figure
32.
The pedestrians will also be prioritized when bicycles are approaching. The
bicycles will give priority and wait for pedestrians if a bicycle is 2.5 meters from a
36
pedestrian, which will try to cross the crosswalk. The priority rule for the second
crosswalk shares the same rules and details as the priority rule for the first
crosswalk. Notice that Design B will use the same driving behavior parameter
settings, desired speed, bus time scheduling table, flow and desired acceleration
and deceleration as Design A.
Figure 31 – Shows the first crosswalk of design B.
37
Figure 32 – Shows the second crosswalk of Design B.
5.4 Model Calibration

The calibration process consists of several different steps. First the number of
replications is presented. Then, the main goal is to achieve a valid t-test and
confidence interval for the measured speed of motorized vehicles and bicycles.
This is done by changing some parameter values in the car-following model for
bicycles, changing the desired acceleration and deceleration for bicycles and also
modification of the speed reduction zones. For the t-test, the absolute value of the
t-value |T| and the critical t-value will be compared with the hypothesis test
investigation if there is a statistical significant difference between the mean value
from the simulation and the mean value from measurement observations. The
result of the test is either that the null hypothesis should be rejected(the absolute
value |T| is larger than the critical value t) or that it cannot be rejected(the
absolute value |T| is smaller than the critical value t). For the calibration this imply
that if the null hypothesis should be rejected the model is not calibrated and
further calibration efforts in terms of adjusting parameters are required. If the null
hypothesis cannot be rejected the model can be said to be able to represent the
observations, i.e. that the model is calibrated with respect to the metrics used in
38
the calibration process. There will also be a confidence level comparison between
the simulation output means and field measurements means on a 95 % level.
The speed is estimated in Vissim at a particular point in the network, see Figure
33. The brown line is the speed estimation for motorized vehicles and the yellow
line is the speed estimation for bicycles. This point is chosen since it is in the
middle of the two passing points, which are used for measuring speed in the field
data collection.
Figure 33 – Speed estimation in Vissim
5.4.1 Number of replications
The number of replications can be calculated according Olstam (2019) as

Equation (3).
𝑠𝑚 ∗ 𝑡𝑚−1 (𝛼/2) 2 (3)
𝑛= ( )
𝑋𝑚 ∗∈
Where n is the required number of replications, Sm is the standard deviation for

the investigated performance indicator, Xm is the mean for the investigated
performance indicator, ∈ the accepted error rate in terms of percent of the mean
value and 𝑡𝑚−1 (𝛼/2) is the value from a student t-distribution for the confidence
level (𝛼/2). The standard deviation s and the mean speed X are unknown but can
be estimated by running a limited set of m simulations.
39
Standard error rate was set to 3 and t-value 𝑡0.05/2 was 4.3. Standard error rate
was set to 3 since it should be between 2 and 5% for traffic according to
Bernhardsson (2017) . Three simulation runs were made with the corresponding
mean time 20 km/h and standard deviation 0,44, which resulted in 9.94
replications and was rounded up to 10. The calculation can be seen in Table 6.
Table 6 – Number of replications calculation
Mean speed Standard Accepted error
t-value Number of replications
(km/h) deviation rate
20.0 4.3 0.44 3% 9.94 ≈ 10
5.4.2 Motorized vehicles
The calibration of the speed for motorized vehicles is made by a t-test and
confidence interval on a 95 % level. Motorized vehicles use default settings in
Vissim regarding car-following model parameters and desired acceleration and
deceleration. Table 7 shows the average speed for the measured sample data
(Field measurement) for three days and the average speed from simulation for 10
replications. It also shows the statistical hypothesis test (t-value) on a 95% level
which can be seen at the bottom of the table. The t-value from the T-Distribution
table in this case will be 2.201 and absolute |T| is 1.7. For the calibration, this
imply that the null hypothesis cannot be rejected and there is significant
difference and the model can be said to be able to represent the observations, i.e.
that the model is calibrated with respect to the metrics used in the calibration
process. A confidence interval test on a 95% level is also made between the field
measurement and the simulation output which indicates overlapping and this can
be seen in Figure 34.
40
Table 7 – The average speed of the field measurement and the simulation output for motorized
vehicles.
Field measurement
Replication Simulation km/h
km/h
1 35,16 35,78
2 36,2 35,75
3 35,6 35,58
4 35,88
5 35,89
6 35,27
7 35,62
8 36,04
9 36,07
10 36,00
Average 35,65 35,79
|T| 1,7
t 2,201
Figure 34 – Confidence intervals for motorized vehicles on a 95% level.
41
5.4.3 Bicycles
This section will describe the calibration process for bicycles. Initially, bicycles
will be modelled with driving behavior parameters, desired acceleration and
deceleration from COWI (2013).The corresponding confidence interval and t-test
on a 95% level can be seen in Table 8 and
Figure 35 below. Since absolute T is larger than the t-value, the null hypothesis
that there is no difference can be rejected, hence there is a statistical difference
between the field measurements and the simulation.
Table 8 – The average speed of the field measurement and the simulation output for bicycles with
default settings from COWI.
Replication Field measurement Simulation km/h
km/h
1 19,93 18,74
2 20,83 18,84
3 21,58 18,39
4 18,40
5 18,91
6 18,63
7 18,61
8 18,79
9 18,42
10 18,66
Average 20,78 18,64
|T| 17,1219
t 2,2010
42
Figure 35 – Confidence interval for bicycles on a 95% level with default settings from COWI.
Adjustment of Driving behavior parameter

The bicycle speed from the first calibration step did not correspond to the speed
measured on the field which required the parameters to be subsequently
adjusted. First modification was to change the driving behavior parameters of
bicycles. This is done manually with the intention to increase the speed. Table 9
shows the parameters for bicyclists, which have been adjusted and the value for
each parameter of the default model from COWI (2013) and the adjusted
parameters.
Parameters which have been adjusted are Min. Look ahead distance, Min. Look
Back distance and CC1(Following time). The Min. Look ahead distance is the
minimum distance that a vehicle can see ahead in order to react to other vehicles
in front of or next to it on the same link. Min. Look Back distance is the minimum
distance a vehicle can see behind it in order to react to other vehicles behind it on
the same link (PTV AG, 2018). The Min. Look ahead distance and Min. Look Back
distance are reduced to 1 meter since it is believed that the bicycles will reduce
their speed if they in an early stage, can see other vehicles in front and ahead of
them.
CC1(Following time) is the time distribution of speed-dependent part of desired
safety distance. This is the distance in seconds which a driver wants to maintain
at a certain speed. The higher the value, the more cautious the driver is (PTV AG,
43
2018). CC1 was reduced from 0.3 s to 0.1 s in order to make the bicycles less
cautious.
Table 9 – Adjusted driving behaviour parameters compared to COWI settings.
Adjusted set of
COWI
parameters
Following
Min. Look ahead distance (m) 20 1
Min. Look back distance (m) 20 1
Smooth closeup behaviour Inactive
Car-following model (Wiedemann 99)
CC0(Standstill distance) 0.2 m 0.2 m
CC1(Following time) 0.3 s 0.1 s
CC2(Following ‘variation’) 2m 2m
CC3(Threshold for entering) -20.0 -20.0
CC4(Negative threshold) -0.25 -0,25
CC5(Positive threshold) 0.25 0.25
CC6(Speed dependency of Oscillation) 1.00 1.00
CC7(Oscillation Acceleration) 0.20 m/s2 0.20 m/s2
CC8(Standstill Acceleration) 1.8 m/s2 1.8 m/s2
CC9(Acceleration with 80km/h) 0.01 m/s2 0.01 m/s2
Lane changing behaviour
Maximum deceleration (m/s2) -3.0 m/s2 -3.0 m/s2
1 m/s2 own and trailing (m) -50 m -50 m
Table 10 shows the average speed of the field measurement and the simulation
output for bicycles with adjusted driving behavior parameters. Notice a small
increase in speed (16.64 km/h to 16.69 km/h) compared to the default settings
from COWI (2013). The null hypothesis that there is no difference between the
field measurement and the simulation is rejected, since the absolute (|T|) is larger
than t. Figure 36 shows the corresponding confidence interval on a 95% level.
44
driving behaviour parameter adjustments.
Field measurement
Replication Simulation km/h
km/h
1 19,93 18,69
2 20,83 19,15
3 21,58 18,53
4 18,52
5 18,93
6 18,49
7 18,62
8 18,66
9 18,67
10 18,66
Average 20,78 18,69
|T| 16.4290
t 2.1788
Figure 36 – Confidence interval for bicycles on a 95% level with adjusted driving behaviour
parameters.
Change of acceleration and deceleration

Since the changes in driving behavior parameter values did not have any large
effect on the difference between simulated and observed speed, another approach
45
was tested in which the desired acceleration and deceleration settings were
changed and the driving behavior parameters were adjusted to COWI (2013).
COWI (2013) settings for acceleration and deceleration were first used which can
be seen in Figure 37 and Figure 38. Figure 39and Figure 40 shows the adjusted
change in desired maximum acceleration and deceleration. In Figure 37, bicycles
with a low speed will be appointed a higher acceleration. In Figure 38, bicycles
with a bit higher speed will also be appointed a higher acceleration. The purpose
of the adjustment was to increase the acceleration and deceleration for bicycles
who are driving with a lower speed. Especially it is desirable to get bicycles to
accelerate and decelerate more intensive when interacting with buses in the
conflict zones.
.
Figure 37 – Desired Acceleration, COWI Figure 38 – Desired deceleration, COWI
settings settings.
Figure 39 – Desired acceleration, changes. Figure 40 – Desired deceleration, changes.

output for bicycles with changes in acceleration and deceleration and a statistical
hypothesis test on a 95% level and Figure 41 shows the corresponding confidence
46
interval. The absolute T is larger than t and the null hypothesis that there is no
difference between the field measurement and the simulation means are equal
should be rejected and the model is not calibrated and further calibration efforts
in terms of adjusting parameters are required.
Table 11 – The average speed of the field measurement and the simulation output for bicycles
with changes in acceleration and deceleration
Replication Field measurement (Km/h) Simulation (Km/h)
1 19,93 18,71
2 20,83 19,03
3 21,58 18,44
4 18,63
5 18,6
6 18,82
7 18,56
8 18,53
9 18,91
10 18,34
Average 20,78 18,71
|T| 16,5206
t 2.1788
Figure 41 – Confidence interval for bicycles on a 95% level with changes in desired acceleration
and deceleration.
47
Desired Speed and Speed Reduction Zones
As mentioned earlier in section 5.1, speed reduction zones were introduced in

order to adapt the bicycles to curves and unexpected turns. Speed reduction zones
will reduce the speed of bicycles independently of their desired speed. As soon as
they have left the speed reduction zone, they will try to reach their desired speed.
Two speed reduction zones before and after the entry of the bus stop will be
removed(see figure 42 and 43 for before the entry of the bus stop) as a part of the
calibration since it is believed that they are reducing the speed too much at this
specific place due to not so sharp curves.
Figure 42 – Speed reduction near the entry Figure 43 – Speed reduction zones near the
of bus stop and near the second traffic entry of bus stop after the modification.
signal.
The speed reduction zones at the first intersection will be kept since the curve is
sharp enough to imply influence on the speed of bicycles, see speed reduction
zones for bicycles in Figure 44. As soon as the bicycles have reached these speed
reduction zones, they will reduce their speed to be around 15 km/h mark. The
bicycles will then accelerate to their desired speed but it will not be sufficient for
the calibration. The speed reduction zones before the bus stop see Figure 44, will
be kept since the bicycles should drive pretty slow in the curves. Because of the
low speed in the speed reduction zones, a speed decision will be introduced after
48
the curve in order increase the speed for bicycles. An alternative is to modify the
desired speed distribution but this is not desirable since the speed reduction
zones will force the bicycles to always drive 15 km/h and we want the bicycles to
drive faster after the speed reduction zones which is not accomplished at the
moment. Desired speed decisions are used as a permanent change, for example
on motorways, for traffic signs which limit the speed, or at the entry and exit of a
town. Desired speed decisions are placed at the position where the change is to
begin (PTV AG, 2018). When a bicycle is passing the speed decision marker in
Figure 44, the speed will increase to around 20 km/h, which is the mean speed
from the field measurements. Parameter values regarding driving behavior,
acceleration and deceleration are from COWI (2013).
Desired speed
decision
Speed reduction
zones for bicycles
Figure 44 – Speed reduction at the first traffic signal and Desired speed decision.
output for bicycles with changes in desired speed and speed reduction zones and
a statistical hypothesis test on a 95% level. Figure 45 shows the corresponding
confidence interval on a 95% level. Notice that the hypothesis that there is no
difference between the field measurement and the simulation cannot be rejected
49
as the absolute t(|T|) is smaller than t-value. For the calibration, this imply that
the null hypothesis cannot be rejected and there is significant difference and the
model can be said to be able to represent the observations, i.e. that the model is
calibrated with respect to the metrics used in the calibration process.
changes in desired speed and speed reduction zones.
Replication Field measurement (Km/h) Simulation (Km/h)
1 19,93 20,60
2 20,83 20,62
3 21,58 20,55
4 20,57
5 20,58
6 20,56
7 20,62
8 20,57
9 20,55
10 20,59
Average 20,78 20,58
|T| 1.7618
t 2.2010
50
Figure 45 – Confidence interval for bicycles on a 95% level with modification of the desired speed
and speed reduction zones.
51
6. Results
This chapter presents and compares the results for the two different design with
respect to travel time and speed for bicycles and buses. The results are presented
in respect to travel time and delay for bicycles and buses. The chapter starts with
the results of Design B compared to the calibrated model of Design A. Then, the
results for bicycles and buses with different inputs in Design A and B will be
presented.
Notice that the minimum required number of replications for all studied metrics
for all cases were calculated and found to be less than ten (as presented in
Appendix A), so that the number of replications considered in the analysis is 10
runs.
6.1 Current traffic demand

This section shows the comparison between design A and B is based on travel time
and delay for bicycles and buses. Note that the results in this section denote the
travel demand in the current situation.
Travel time and delay for bicycles and buses are measured in Vissim from a
starting point from around 70 meters before Langholmsgatan and Horsnagatan
intersection, until an end point close to the pedestrian’s bridge (where the video
recordings were made), which can be seen in Figure 46. The starting and end
points are applied for both Design A and B.
52
Figure 46 – Data collection for Travel time and delay in Vissim.
6.1.1 Travel time
Figure 47 shows the average travel time(seconds) per bicycle of Design A and B.
Figure 48 shows the average travel time(seconds) per bus of Design A and B.
Notice a higher travel time for Design B compared to Design A for both Bicycles
and buses.
53
Average travel time per bicycle
120.00
110.18
100.00 96.10
80.00
Travel time (sec)
60.00 Design A
Design B
40.00
20.00
0.00
Design A Design B
Figure 47 – Average Travel Time per bicycles comparison of Design A and B.
Average travel time per buses

160.00
140.56
140.00
122.51
120.00
Travel time (sec)
100.00
80.00 Design A
Design B
60.00
40.00
20.00
0.00
Design A Design B
Figure 48 – Average Travel Time per buses comparison of Design A and B.
54
6.1.2 Delay
The average delays for bicycles and buses for Design A and B are presented in
Figure 49 and Figure 50. The results indicate that Design A provided less delay for
bicycle, while Design B provides less delay for buses.
Average Delay per bicycle

45.00
40.00 37.24
35.00
30.00
Delay (sec)
25.00 22.77
Design A
20.00
Design B
15.00
10.00
5.00
0.00
Design A Design B
Figure 49 – Average delay per bicycles comparison of Design A and B.
Average Delay per buses

20.00 18.26 17.70
18.00
16.00
14.00
Delay (sec)
12.00
10.00 Design A
8.00 Design B
6.00
4.00
2.00
0.00
Design A Design B
Figure 50 – Average delay per buses comparison of Design A and B.
55
6.2 Different bicycle flows and bus frequencies
As stated in the introduction, the traffic demand of transportation in Stockholm is
increasing. Design A and B will be tested with an increase in number of bicycles
and buses in order to see which design can handle an increase in traffic volumes
and maintain or reduce the travel time and delay. The number of bicycles buses
and pedestrians for Design A and B can be seen in Table 13. 500 bicycles/h is the
current flow and the buses are increase to 30 and 60 buses/h. Number of
pedestrians are 300/h for 30 buses and 600/h for 60 buses.
Table 13 – Shows the different input for Design A and B of Bicycles and Buses.
Number of bicycles/h 500 750 1000
Number of Buses/h 30 60
Number of pedestrians/h 300 600
6.2.1 Travel time
The number of bicycles varies from 500 bicycles per hour (representing the
current flow) and increases to 750 and then finally to 1000 bicycles per hour for
both Design A and B. The number of buses vary from 30 buses per hour to 60
buses per hour. The number of bicycles and buses per hour can be seen in each
figure as the number within the brackets.
Figure 51 shows the average travel time per bicycles and Figure 52 shows the
average travel time per bus with different input of bicycles and buses. Green and
grey color represent 30 buses per hour and orange and blue represent 60 buses
per hour.
56
Figure 51 – Average Travel Time per bicycle with input changes for bicycles and buses.
Figure 52 – Average Travel Time per bus with input changes for bicycles and buses.
57
6.2.2 Delay
Figure 53 shows the average delay per bicycle and Figure 54 shows the average
delay per bus with different input of bicycles and buses. Notice a significantly
higher delay for bicycles in Design B compared to Design A. Also notice a slightly
higher delay for buses when the number of buses increase.
Figure 53 – Average delay per bicycle with input changes for bicycles and buses.
Figure 54 – Average delay per bus with input changes for bicycles and buses.
58
7. Discussion & Conclusion
7.1 Discussion
For the current traffic demand, design B seems to have higher travel time and
delay for bicycles compared to Design A. This might be due to the fact that bicycles
need to give way for pedestrians who will try to cross the bicycle lane. In Design
A, bicycles need to yield for buses while in Design B, bicycles need to yield for
pedestrians. The number of pedestrians is higher than the number of buses which
means that the probability to wait for a pedestrian is higher than to wait for a bus.
The time it takes for a pedestrian to cross the bicycle lane might also be more than
the time it takes for buses to pass the bicycle lane. This might be the reason for a
higher travel time and delay in Design B. For buses, the travel time and delay are
less in Design B compared to Design A due to no interaction between buses and
bicycles in Design B.
For different bicycle flows and bus frequencies, it was noticed that there was
always higher travel time for bicycles in Design B compared to Design A
independently of the number of bicycles or buses. The travel time was
significantly higher when the number of bicycles increases gradually from 500 to
750 bicycles per hour. However, when increasing more than 750 bicycles per hour
and 1000 bicycles per hour, there is no significant difference in the travel time.
The maximum capacity result is mainly due to the width of the bicycle lane. The
travel time for bicycles will also increase more in Design B than Design A specially
when the number of buses increases. This is due to a higher number of
pedestrians who attempt to reach or leave the bus stop which subsequently
increase the interactions between pedestrians and bicycles. The travel time for
buses is almost the same in both Design A and B even when the number of buses
and bicycles increases. This is due to the fact that bicycles must yield for buses in
59
Design A and that there are no interactions between bicycles and buses in Design
B.
The delay for bicycles in Design A is almost the same when the number of bicycles
and buses increase due to the fact that bicycles are not yielding for buses. But in
Design B, the delay will be significantly higher for 60 buses compared to 30 buses.
The increment of travel time is once again due to the increasing number of
interactions between bus users and bicycles as the number of pedestrians will
increase when the number of buses increase. The delay for buses will only
increase slightly when the number of buses increase. This might also be due to a
higher number of pedestrians who are trying to enter the bus.
Another source of delay is the traffic light. In Design A, both bicycles and buses
will be delayed, partially due to the traffic light. In Design B, the second traffic light
is ignored for bicycles and according to the results, there is a 50 to 70 % increase
in delay for bicycles which imply that that the traffic light might not delay the
bicycles that much. Instead, it is probably or might be the pedestrians who will try
to cross the bicycle lane that are the main source of delay for bicycles in design B.
The data was collected during the morning peak hour for three weekdays. The
speed was collected from various randomly chosen vehicles during morning peak
hours. The calibration was also entirely based on these random vehicles due to
extensive amount of time required to extract the data. In this way, the calibrated
model includes uncertainties since the data used for calibration was limited.
Another issue was that the second and third day of data collection took place
during outbreak of the coronavirus. This also meant that the number of vehicles
and pedestrians were reduced compared to the first day. And this issue caused a
variance in the data between the first day and the two other days which should
not be neglected.
The calibration was based on the t-test and the confidence interval of the speed.
There were various changes in parameters and the parameter affecting the speed
the most, was the desired speed decision. The desired speed decisions included
particularly to improve the speed of bicycles on a specific point in the network.
This also means that the behavior of bicycles on this specific point might not
reflect the actual conditions.
The calibration was based on the speed at a certain position but the analysis and
comparison of the two designs were based on travel time and delay for a different
stretch. This implies that there might be some uncertainties within this stretch
and that the results should be interpreted with these uncertainties in mind. The
travel time and delay might not reflect how bicycles behave and interact with
60
buses and pedestrians i.e. if the distance headway and time gap are sufficient or if
the queueing time is too long for bicycles when they interact with buses or
pedestrians. Other uncertainties might be driving behavior parameters overall,
which should be taken with a ‘grain of salt’ for this specific simulation since a lot
of the parameters are hard to calibrate and measure from the study location. How
bicycles behave and drive in curves (speed reduction zones) and intersections
might also be uncertain for this specific simulation since the calibration for the
speed was collected after and before the intersections and speed reduction zones
in this specific simulation. The stopping time for buses might be too long or short
to stop at the bus stop. Other uncertainties might be that overtaking and lane
changing could be accomplished more times and subsequently more bicycles
would be able to drive in the lane.
The speed reduction areas were first introduced in curves to reflect real behavior.
Some of the speed reduction areas were then later partially removed in order to
improve the speed for bicycles. This modification should also be considered when
comparing with the reality.
The priority rule is partially made up by a distance headway and time gap in
Vissim. Initially, time gap and distance headway were also supposed to be
collected from the video recording. This seemed to be quite challenging, and the
main goal of the priority rule was rather to avoid accidents and conflict between
buses and bicycles and thereafter adjust the time gap and distance headway with
the intention to only avoid accidents. It was not realized from the beginning that
parameters regarding speed, time and distances would be difficult to collect from
video recordings. Probably it would have been more advantageous to use another
data collection method such as road tubes which would collect speed and volume
more accurate. Data like distance headway, time gap and acceleration from
different vehicles would also be interesting. Perhaps pedestrian data would also
have been interesting to study and process.
Another problem was that buses could only leave the bus stop at the end of the
platform in the simulation. In reality, the buses would stop in a row, but if a bus is
ready for departure but is further behind in the row, they will choose to leave the
bus stop from the particular place where it stopped. This phenomenon is not
possible to capture in the simulation and should be considered.
Furthermore, the conflict areas were used in the model, but only for motorized
vehicles and buses. Conflict areas are used in the same way as the priority rule but
without specific lines to stop for bikes, vehicles or pedestrian. Conflict areas were
first used to manage the conflict between bicycles and buses, but later it was
61
understood that they were not appropriate for dictating that bicycles must yield
for buses when buses try to cross bicycle lane except if the bicycle is in front of the
bus.
The number of pedestrians might also have been lower than in reality due to
results and estimations since there was never any queues of buses in the
simulation. At maximum it could only stop three buses at the bus stop due to
length of the bus stop. More pedestrians mean more time for the pedestrians to
enter and leave and the bus and subsequently longer waiting time for buses. If the
waiting time is too long, a bus would potentially not have time to leave the bus
stop until the next red light and subsequently a new bus would not have space to
stop for passengers.
Maybe It would have been better or more interesting to increase the number of
pedestrians to see how it would affect the bicycles and how the two designs would
handle it. If the bicycles would standstill too long, it might be appropriate to
consider a traffic light instead of crosswalks.
Another problem is the conflict zones between bicycles, pedestrians and buses
overall. It was difficult to measure how the conflict areas/zones were performing.
The result is only showing travel time and delay, but it might be interesting to
evaluate the conflict areas/zones for specific parameters besides the safety
probability and severity of accidents in design A or B.
The probability of accidents might be higher in Design A since the speed of
bicycles is higher in Design A compared to Design B. The stop distance is the sum
of the reaction distance and breaking distance which is mainly affected by the
speed. Consequently, accidents between bicycles and buses might be more likely
which is also unfortunate since it probably can lead to more severe accidents
compared to accidents between pedestrians and bicycles. However, pedestrians
might behave strangely with improbable and rapid movements, which might
cause more or sudden accidents but not as severe as accidents between buses and
bicycles.
7.2 Conclusion
The most effective and optimum road layout design depends on the purpose and
what type of vehicle that uses the road. Under the assumption that travel time and
delay is sufficient to decide the outcome of the two designs, Design A should be
preferred over Design B for bicycles. This is also under the assumption that
62
bicycles need to yield for pedestrians and buses. Increasing number of bicycles
and buses and maximize the capacity of the network also support that design A is
to be preferred for bicycles. Design B should be preferred over Design A for buses
when considering travel time and delay.
Design B might also be better from the safety perspective since accidents between
bicycles and buses are more likely to result in serious and fatal accidents
compared to an accident between bicycles and pedestrian.
Further conclusion is that traffic signals have a major impact on all road users.
Traffic signals affect all vehicles in terms of travel time and delay, and it might be
interesting for further investigation to analyze how the network is affected by
ignoring traffic signals.
63
Bibliography
Delaware Valley Regional Planning Commission(DVRPC). (2009). Bicycle-Bus
Conflict Area Study. Philadelphia: SEPTA.
Aldred, Best, & Jones. (2016). Cyclists in shared bus lanes: could there be
unrecognised impacts on bus journey times? ICE.
Bernhardsson, V. (2017). Trafiksimulering av E20 Ribbingsberg – Vara: En
simuleringsstudie av framkomlighet för en mötesfri utformning av E20.
Linkoping: VTI rapport 942.
COWI. (2013). MICRO SIMULATION OF CYCLISTS IN PEAK HOUR TRAFFIC.
Copenhagen: COWI.
Dimakis, N., & Guðmundsdottir, B. (2018). Sustainable traffic planning in an
urban area. Goteborg: CHALMERS UNIVERSITY OF TECHNOLOGY.
EIDMAR, E., & HULTMAN, J. (2014). Traffic Network Evaluation using Microscopic
Simulation and Analytical Modelling. 2014: CHALMERS UNIVERSITY OF
TECHNOLOGY.
Franson, E. (2018). Driving behavior modeling and evaluation of merging control
strategies - A microscopic simulation study on Sirat Expressway.
Norrkoping: Linkoping University.
Gao, L., Liu, M., & Feng, J. (2012). Delay Modeling of Ped-Veh System Based on
Pedestrian Crossing at Signalized Intersection. SciVerse ScienceDirect ,
530 – 539.
Guðmundsdottir, B., & Dimakis, N. (2018). Sustainable traffic planning in an
urban area. Gothenburg: Chalmers University Technology.
Janson Olstam, J., & Tapani, A. (2004). Comparison of Car-following models.
Linkoping: VTI.
Krykewycz, G. (2010). Bicycle/Bus Conflict Area Study. Delaware valley regional
planning commision .
Miguel , F., Keeling, K., & Glick, T. (2019). Buses, Bikes, and Right-Turning Cars:.
Portland State University.
NTF. (2020, 07 07). Retrieved from Lagar och regler :Vajningsplikt:
https://ntf.se/konsumentupplysning/lagar-och-
regler/fordonsforare/vajningsplikt/
Olstam, J. (2019). Traffic Theory and Simulation, lecture slides from lec 10:
Calibration [internal material] . Norrkoping: Linkoping University.
Palmqvist, D. (2015). Parametrar for mikrosimulering av cykeltrafik. Lunds
universitet, LTH. Instituionen för Teknik och samhälle. Traffik och väg
2015.
PTV AG. (2018). PTV VISSIM 11 USER MANUAL. Karlsruhe: PTV AG.
Stockholms stad . (2019 B). En del av Framkomlighetsstrategin: Cykelplan.
Stockholm: Stockholms stad.
Stockholms stad. (2019 A). Stockholms Trafikutveckling: Cykel och fotgängare
2018. Stockholm: Stockholms Trafikkontor.
Stockholms stad. (2020). vision2O3O A GUIDE TO THE FUTURE. Stockholm:
Stockholm stad.
Trafikverket. (2019 B). VGU fram till 31 december 2019. Retrieved 02 06, 2020,
from https://www.trafikverket.se/for-dig-i-branschen/vag/Utformning-
av-vagar-och-gator/vgu/aldre/vgu-fram-till-december-2019/
Trafikverket. (2020 A). Råd - VGU, Vägars och gators utformning. Borlange:
Trafikverket.
Treiber, M., & Kesting, A. (2013). Traffic Flow Dynamics: Data, Models and
Simulation. New York: Springer.
UN Environment. (2016). Global Outlook on Walking and Cycling: Policies &
realities from around the world. Nairobi: UN Environment.
Virginia Department of Transportation . (2020). VDOT VISSIM User Guide Version
2.0. Virginia: Virginia Department of Transportation.
65
Appendix
Appendix A
1- Results for travel time and delay in design A and B for the travel demand in the
current situation.
Table 1 – Shows the travel time for design A and B and standard deviation, confidence interval and
the minimum required replications runs.
Travel Time (sec)
Design A Design B
1 97,79 108,75
2 96,39 110,06
3 95,50 110,43
4 99,17 112,03
5 95,86 109,25
6 96,11 115,13
7 94,21 107,31
8 96,76 111,57
9 95,09 106,95
10 94,07 110,35
Average 96,10 110,18
Size 10,00 10,00
Standard deviation 1,56 2,39
alpha 0,05 0,05
Confidence interval 1,12 1,71
Number of replications 1,5004 2,6825
Table 2 – Shows the travel time for design A and B and standard deviation, confidence interval and
the minimum required replications runs.
Delay (sec)
Design A Design B
1 26,32 36,17
2 26,08 38,91
3 25,43 38,64
4 26,54 37,51
5 26,54 35,05
6 25,65 39,24
7 25,04 36,30
8 26,51 38,23
9 26,56 36,42
10 26,03 35,89
Average 26,07 37,24
Size 10,00 10,00
Standard deviation 0,54 1,46
alpha 0,05 0,05
Confidence interval 0,38 1,04
Number of replications 2,4173 8,7163
2- Results for travel time and delay in design A and B for Different bike flow and
bus frequencies.
Table 3 – Shows the bicycles’ travel time for different input in design A and B and standard deviation,
confidence interval and the minimum required replications runs.
Travel Time (sec)
total number of Standard Confidence in- Number of

Average bicycles in the deviation terval replications
network
Bicycles Design A 95,56 492 3,27 2,34 6,647
(500) Design B 112,79 497 3,11 2,22 4,318
Bus Bicycles Design A 155,96 704 6,81 4,87 10,848
(30) (750) Design B 218,65 676 8,82 6,31 9,240
(1000) Design B 233,00 679 2,63 1,88 0,723
(500) Design B 128,90 493 2,57 1,84 2,254
Bus Bicycles Design A 158,55 704 5,85 4,18 7,738
(60) (750) Design B 239,38 674 9,11 6,52 8,232
(1000) Design B 253,18 672 3,06 2,19 0,829
Table 4 – Shows the buses’ travel time for different input in design A and B and standard deviation,
Travel Time (sec)
Standard Confidence Number of

Average
deviation interval replications
Bicycles Design A 140,56 5,36 3,83 8,2629

(500) Design B 122,51 5,10 3,65 9,8478
Bus Bicycles Design A 143,08 5,83 4,17 9,4474
(30) (750) Design B 122,51 5,10 3,65 9,8478
(1000) Design B 122,51 5,10 3,65 9,8478
(500) Design B 129,33 4,26 3,05 6,1718
(60) (750) Design B 129,33 4,26 3,05 6,1718
(1000) Design B 129,33 4,26 3,05 6,1718
Table 5 – Shows the bicycles’ delay for different input in design A and B and standard deviation,
Travel Time (sec)

Average

(500) Design B 39,48 1,39 1,00 7,09
(30) (750) Design B 41,78 1,15 0,82 4,31
(1000) Design B 42,38 0,90 0,65 2,59
(500) Design B 56,34 1,25 0,89 2,78
(60) (750) Design B 61,97 1,42 1,01 2,96
(1000) Design B 62,12 1,06 0,76 1,67
Table 6 – Shows the buses’ delay for different input in design A and B and standard deviation,
Travel Time (sec)

Average

(500) Design B 17,70 0,38 0,27 2,6019
(30) (750) Design B 17,70 0,38 0,27 2,6019
(1000) Design B 17,70 0,38 0,27 2,6019
(500) Design B 21,33 0,35 0,25 1,5161
(60) (750) Design B 21,33 0,35 0,25 1,5161
(1000) Design B 21,33 0,35 0,25 1,5161

FULLTEXT01

Uploaded by

Document Information

Copyright

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

FULLTEXT01

Uploaded by

Copyright:

LiU-ITN-TEK-A--20/055-SE

Two different bus stop layout

Department of Science and Technology Institutionen för teknik och naturvetenskap

Two different bus stop layout

Department of Science and Technology Institutionen för teknik och naturvetenskap

Detta dokument hålls tillgängligt på Internet – eller dess framtida ersättare –

© Mohamed Elsayed, Erik Torstensson

Two different bus stop layout

Erik Torstensson & Mohamed Elsayed

Communications and Transport system

Copyright 2020 Erik Torstensson and Mohamed Elsayed

Keywords: microscopic, simulation, VISSIM, driving behavior, travel time, delay.

1.1 Aim and research questions

1.3 Study area

Figure 4 – Sideview of the study case (Långholmsgatan).

1.5 Structure of the report

Figure 7 – Building a simulation model steps (Olstam, 2019)

3.2 Microscopic traffic simulation in Vissim

3.2.1 Car-following model in Vissim

Figure 8 – Wiedemann 99 model (PTV AG, 2018).

3.2.2 Lane changing

3.2.3 Parameters of bicycles in Vissim

3.2.4 Pedestrian behaviour modelling in Vissim

The operational level decides the route by millisecond to seconds where it

3.2.4.1 Pedestrian routing decisions

3.2.5 Conflict modelling in Vissim

A conflict between two vehicles or a vehicle and a pedestrian can be modelled in

By assuming an average vehicle length L, it is possible to estimate a speed for an

Figure 15 – Single vehicle data measurements (Treiber & Kesting, 2013).

CDF for the speed of 300 bicycles

Mean speed bicycles Mean speed for

Table 3 – Number of vehicles during 07:30-08:30.

4 of mars 2020 479 1044 37

24 of mars 2020 497 822 16

25 of mars 2020 525 948 18

Average 500 938 25

4.2 Traffic signals

Signal 1 Signal 2 Signal 3

Cycle Length (sec) 100 100 100

Figure 19 – Shows the location of three traffic signals.

Figure 21 – 3D view for Design A.

5.1.1 Bus Time Scheduling

5.1.2 Traffic Signals Modelling

Figure 25 – Shows the signal group for signal 1.

Figure 26 – Shows the signal group for signal 2 and 3.

Figure 27 – Desired speed distribution for motorized vehicles.

Figure 28 – Desired speed distribution for Bicycles.

Table 5 – Shows the average flow for different vehicles.

Figure 29 – Shows a 3D view for the Vissim scenario-2 model.

Figure 30 – The conflict between pedestrians and bicyclists in Design B.

Figure 31 – Shows the first crosswalk of design B.

5.4 Model Calibration

Figure 33 – Speed estimation in Vissim

5.4.1 Number of replications

The number of replications can be calculated according Olstam (2019) as

Where n is the required number of replications, Sm is the standard deviation for

20.0 4.3 0.44 3% 9.94 ≈ 10

5.4.2 Motorized vehicles

Figure 34 – Confidence intervals for motorized vehicles on a 95% level.

Adjustment of Driving behavior parameter

Change of acceleration and deceleration

Figure 39 – Desired acceleration, changes. Figure 40 – Desired deceleration, changes.

As mentioned earlier in section 5.1, speed reduction zones were introduced in

Average 20,78 20,58

6.1 Current traffic demand

6.1.1 Travel time

Figure 47 – Average Travel Time per bicycles comparison of Design A and B.