interpreted modeling for safety and emissions at ...abastos/outputs/teses/projeto de tese...

UNIVERSIDADE

DE AVEIRO

Interpreted modeling for safety and emissions at roundabouts in

corridors

Projeto de Tese II

Aluno: Paulo Jorge Teixeira Fernandes – [email protected]

Orientadores: Prof.ª Doutora Margarida Coelho, DEM-UA

([email protected]); Prof. Doutor Nagui Rouphail, North

Carolina State University ([email protected])

Docente: Prof. Doutor Vítor Costa – [email protected]

mailto:[email protected]




Interpreted modeling for safety and emissions at roundabouts in corridors

Thesis Project II 2

Table of Contents

Index of Figures .................................................................................................. 4

Summary ............................................................................................................ 5

1. Introduction ..................................................................................................... 6

1.1. General Statement of the Problem .......................................................... 6

1.2. Objectives ................................................................................................ 6

1.3. Structure of the thesis .............................................................................. 7

2. Review of the Technical Literature ................................................................. 8

2.1. Introduction .............................................................................................. 8

2.2. Roundabouts in corridors ......................................................................... 8

2.3. Roundabouts at isolated intersections ..................................................... 9

2.3.1. Capacity ............................................................................................ 9

2.3.2. Emissions ........................................................................................ 11

2.3.3. Safety .............................................................................................. 13

2.4. Summary of literature review and motivations ....................................... 14

3. Methodology and Methods ........................................................................... 16

3.1. Proposed Tasks .................................................................................. 18

3.1.1. Numerical model for emissions assessment in corridors .............. 19

3.1.1.1. Definition of sub-segments ..................................................... 19

3.1.1.2. Development of models ......................................................... 20

3.1.2. Microscopic simulation platform of traffic, emissions and safety ... 22

3.1.2.1. Traffic Modeling ...................................................................... 24

3.1.2.2. Data collection ........................................................................ 25

3.1.2.3. Emissions Estimation ............................................................. 25

3.1.2.4. Safety ..................................................................................... 27

3.1.2.5. Calibration and Validation ...................................................... 28

3.1.3. Scenarios evaluation .................................................................... 30

3.1.3. Multi-objective analysis ................................................................. 30

3.1.3.1. Micro simulation optimization formulation............................... 30

3.1.3.2. Genetic Algorithms ................................................................. 31

4. Results ...................................................................................................... 35

4.1. Introduction and objectives .................................................................. 35

4.2. Selected corridors and field measurements ........................................ 35

4.3. Discussion and Results ....................................................................... 38


Thesis Project II 3

4.3.1. Characteristic Speed Trajectories ................................................. 38

4.3.2. Characteristic Acceleration and deceleration Trajectories ............ 40

4.3.3. Segments and Sub-segments Emissions ..................................... 42

4.3.4. Emissions per unit distance .......................................................... 43

4.4. Conclusions ......................................................................................... 45

5. Papers in Publication ................................................................................ 46

5.1. Papers published (or accepted for publication) in scientific journals ... 46

5.2. Book Chapters .................................................................................... 46

5.3. Conference proceedings ..................................................................... 47

6. Future Tasks ............................................................................................. 48

6.1. Papers in preparation .......................................................................... 48

Acknowledgments ............................................................................................ 48

References ....................................................................................................... 49


Thesis Project II 4

Index of Figures

Figure 1: Tasks of the work plan proposed for this Doctoral Research. .......... 18

Figure 2: Segments and sub-segments definition of a roundabout corridor and

illustrative speed profile. ................................................................................... 20

Figure 3: Methodological overview for the characterization of traffic and

emissions in corridors. ...................................................................................... 22

Figure 4: Methodological simulation framework of the modelling platform. ..... 23

Figure 5: Basic structure of NSGA-II algorithm. ............................................... 34

Figure 6: Aerial View of the Two Data Collection Roundabouts Corridors and

segments identification (black color-North-South; red color-South-North): a) La

Jolla; b) Avon (Source: http://www.arcgis.com). ............................................... 36

Figure 7: Speed trajectories by distance travelled for each through movement by

corridor: a) La Jolla (North-South); b) La Jolla (South-North); c) Avon (North-

South); d) Avon (South-North). ......................................................................... 39

Figure 8: Acceleration/Deceleration cycles by distance travelled for each through

movement by corridor: a) La Jolla (North-South); b) La Jolla (South-North); c)

Avon (North-South); d) Avon (South-North)...................................................... 41

Figure 9: CO2 Emissions (g) per vehicle for each sub-segment across the

roundabouts corridors: a) La Jolla (North-South), b) La Jolla (South-North); c)

Avon (North-South); d) Avon (South-North)...................................................... 43

Figure 10: CO2 Emissions (g/km) per unit distance for each sub-segment across

the roundabouts corridors and overall corridor: a) La Jolla (North-South), b) La

Jolla (South-North); c) Avon (North-South); d) Avon (South-North). ................. 44


Thesis Project II 5

Summary

During the past few years, roundabouts have experienced a significant increase

and are now widely used in many countries. Scientific research has shown that

the operational performance of roundabouts installed at isolated intersections

depends on its design features and also of the traffic streams characteristics,

which lead to a trade-off among capacity, safety and emissions. Furthermore,

there is a lack of evidence about the vehicle’s performance of roundabouts in a

series along a corridor in terms of emissions and fuel use.

The goal of this Doctoral research will be focused on a comparison of energetic

and environmental performance of different roundabouts in corridors. To achieve

the proposed objectives, several roundabouts corridors will be analysed using

empirical approaches, as well as traffic and emission models. Using multi-

objective analysis, this work will expect to identify the most optimized traffic

system both in terms of energy use, emissions and traffic performance.

This report is an ongoing work from the course “Thesis Project I” in which a review

of the technical literature, the main motivations, the main objectives as well as

the research tasks were developed.

In this course, Thesis Project II, a more comprehensive description of the

methodology used on this thesis topic is performed. More specifically, it is

focused on the description of the selected methods to accomplish the tasks of

this thesis. Moreover, some preliminary results and future tasks are also

provided.

Section 1 highlights the objectives and the structure of this research, followed by

the update of the technical literature review (see Section 2). The presentation of

the selected models is made in the section 3. This report finishes outlining the

developed work in Section 4.


Thesis Project II 6

1. Introduction

1.1. General Statement of the Problem

Roundabouts are proven to be a safety countermeasure for intersections (1).

Roundabouts reduce the unnecessary number of stops for vehicles and improve

the capacity, compared to stop-controlled intersections. Roundabout design

imposes different driver behaviours compared to signalized or stop-controlled

intersections. Roundabout design causes drivers to decrease vehicle’s speed

and decelerate as they approach the roundabout, and enter the circulating traffic

and accelerate as they exit the circulating traffic. Roundabout operation is

affected by different levels of vehicle demand at entry approach and in circulating

lanes. During congested periods long queues at the entrance or blockage in the

circulating and exiting lanes may occur. In order to take advantage of operational

benefits of roundabouts, there has been an increasing interest among engineers

and designers to build multi-lane roundabouts for higher flow rates or higher

speed corridors.

During the last three decades, roundabouts have gained increased popularity and

are now widely used in many countries. Scientific research has shown that the

operational performance of individual roundabouts depends of its geometry and

also of the traffic streams characteristics, which leads to a trade-off between

safety and emissions. Furthermore, there is a lack of evidence about the vehicles’

performance in roundabouts systems installed in corridors in terms of emissions.

1.2. Objectives

The main objective of this Doctoral research is to assess the impact of the

presence of roundabouts systems at corridors on traffic performance, fuel use

and emissions.

To accomplish the established objectives, a wide range of geometrics, traffic and

crashes data will be obtained for multi-lane roundabouts located in corridors. The

study will be focused on the effects of different traffic parameters (pedestrians

influence, geometry configuration, distance between each roundabout within the

corridor, and traffic volumes) on energy consumption and emissions.

Thus, the main objectives of this research:

1. To gain an understanding of the effect between different specific geometric

features of corridors with roundabouts (e.g. distance between consecutive

roundabouts, circulating areas, number of approach lanes, deflection angle,

land use parameters) with the environmental variables, in particular, carbon

dioxide (CO2), carbon monoxide (CO), nitrogen oxides (NOX) and

hydrocarbons (HC) emissions;

2. To compare corridors with roundabouts emissions from well-characterized

scenarios according to the Portuguese reality;

3. To perform a multi-objective analysis to identify the most optimized traffic

system both in terms of traffic performance, energy use and emissions.


Thesis Project II 7

1.3. Structure of the thesis

In this section the structure of the thesis is suggested. It is proposed a layout

composed of seven chapters:

1. Introduction and Objectives

2. Review of technical literature

3. Safety impact on roundabouts corridors

3.1. Methods

3.2. Main Results

3.3. Concluding remarks

4. Energy and emissions impact on roundabouts corridors

4.1. Methods

4.2. Main Results

4.3. Concluding remarks

5. Multi-objective analysis

6. Conclusions and Future Work

7. References

The introduction will set the problem statement, state the objectives and will

explain the contribution of the research.

The review of the technical literature will provide a critical assessment of previous

research efforts and its relationship with the present work.

Chapters 3 and 4 will describe the developed methodology and results obtained,

for each impact. First, Chapter 3 will describe the safety impact analysis (namely

the crash data inventory as well as the development of crash prediction models

for corridors with roundabouts) and its results. Chapter 4 will be dedicated to the

description of numerical models to assess the emissions impact.

Chapter 5 will combine both domains (safety and emissions) and describe the

development of an algorithm to perform a multi-objective evaluation with the main

goal of optimizing roundabout corridors systems in the field of emissions and

safety.

Chapter 6 will be a critical analysis of the work performed, a summary of the contributions and will point out future directions of research. Finally, Chapter 7 will be the list of references used.


Thesis Project II 8

2. Review of the Technical Literature

This chapter offers a review of the most relevant studies focused on the general

analysis of roundabouts. The literature review is divided into four main sections.

Section 2.1 describes a brief overview of roundabouts topic. Then, section 2.2

presents the most significant studies in corridors with roundabouts. In turn,

section 2.3 focuses on the existing research on roundabouts installed at isolated

intersections in terms of capacity, emissions and safety. Finally, the most

important conclusions of the literature review and the major motivations of this

thesis topic are explained in the section 2.4.

2.1. Introduction

One-way circular intersections were created by Eugene Henard, in 1877 (2).

Since the adoption of a yield-at-entry regulation in 1966 in Great Britain (3),

roundabout design has evolved from the use of larger circles with emphasis on

merging and weaving to small compact roundabouts.

There has been a significant increase in the deployment of roundabouts in the

past years in the North-America and in Europe. Currently there are approximately

2,300 and 130 roundabouts in place across the United States and Canada,

respectively (4). Similarly, the number of roundabouts constructed in several

countries such as Germany, Italy, Portugal, Spain and United Kingdom has been

relevant. In France, for instance, it is estimated that there are more than 30,000

roundabouts (5), and in Sweden over than 2,000 roundabouts have been

implemented since the mid 1980s (6).

From that point on, there has been high interest and significant research on

roundabouts because of the simplicity in their design, operations and safety (1).

These traffic facilities provide higher capacity levels in intersections compared to

stop-controlled intersections (7) and are most robust concerning to the reduction

of the unnecessary number of vehicle stops (8). The empirical delay formula for

roundabouts in the Highway Capacity Manual (8) demonstrated that the delay at

roundabouts can be either small or the same as verified in all-way-stop-controlled

intersections. Regarding the safety, there is a greater consensus to the benefits

of roundabouts for motor vehicles and for pedestrians compared to other

intersection forms (1). According to the Federal Highway Administration (FHWA),

there was a significant reduction in injury crashes of converting signalized and

two-way-stop-controlled intersections to roundabouts (7).

2.2. Roundabouts in corridors

In this context, roundabouts in corridors represent a traffic calming technique with

the main aim of improving road safety by reducing vehicles speed. Moreover,

they enable U-turns on access restricted roadways (divided roadways with turn

restrictions) and allow flexibility in maximizing intersection capacity without the

need for excess turn lane storage or additional receiving lanes (9).


Thesis Project II 9

A typical concern for use of roundabouts in a series along a corridor is how flow

will be impacted. Along a main corridor with traffic signals, an effort is usually

made to coordinate the signals or provide for progression on so that a significant

number of vehicles can proceed through the arterial without being stopped. A

series of roundabouts forces all vehicles to slow at every roundabout (9). Another

concern about corridors with roundabouts is that how vehicle emissions will vary

according to the geometrical characteristics of roundabouts along the corridor.

These specifics design features could have a significant effect on pollutant

emissions, namely in urban and rural areas.

Several researchers have investigated and developed algorithms for signalized

corridors to improve safety (10, 11) or to minimize fuel consumption and

emissions (12, 13). The few studies carried out in corridors with roundabouts

raise some uncertainties about their effectiveness. Hallmark et al. (14) recorded

marginal benefits in improved traffic flow of roundabouts within signalized

corridors over stop and signal control intersections. In (15), an on-road

assessment of the emission impacts of roundabouts compared with stop

intersection and roundabout with signal control intersection along two corridors

was conducted. The findings suggested that, under uncongested conditions,

roundabouts did not perform better than four-way, or signal controlled

intersections in the same corridor. Nonetheless, each studied corridor (14, 15)

only contained one roundabout throughout its length. Just recently, a lane-based

micro-analytical network model was developed and implemented in aaSIDRA

(aaTraffic Signalised and Unsignalised Intersection Design and Research Aid)

Intersection Model (16). This model allows estimating the impacts of lane-

utilization on network performance and can be applied in roundabouts systems,

but is not able to assess emission impacts of those facilities. Krogscheepers and

Watters (17) assessed the average speeds, delays and travel times of six

roundabouts along a rural corridor in South Africa and compared that with fixed-

cycle traffic signals. The authors concluded that roundabouts offered operation

advantages over traffic signals, but they recognized that roundabouts were

inefficient for high demand scenarios.

2.3. Roundabouts at isolated intersections

Roundabouts are usually analyzed at three levels: capacity, emissions and

safety. This section provides a brief summary of the most important studies with

respect to the operational performance of roundabouts in these fields.

2.3.1. Capacity

The majority of the existing roundabouts capacity models are based in three

methodologies: fully-empirical, gap acceptance and simulation (18). Any of

methodologies cannot fully describe the complex behavioral and physical

processes involved in roundabout approaches movements.

Empirical regression models were created through statistical multivariate

regression analyses with the main aim of establishing relationships between


Thesis Project II 10

observed entry capacity, circulating traffic flows and other variables with a

meaningful impact on capacity. However, they are subjected to statistical and

sampling constraints, namely: a) reduced transferability among different case

studies; b) only included oversaturated traffic flow conditions, and c) the large

amount of data collection to perform its calibration (18). The LR942 Linear

Regression Model (6), French Girabase Model (19) or based on Neural Networks

are the well-known empirical regression models (20).

Gap acceptance modes focused on the distribution of gaps and also on the

usefulness of these gaps to the approaching vehicles, but are limited by the

relatively weak relationships between these models and design features (18).

The estimation of the critical gap (minimum time gap in the circulating stream

which an entering driver will accept) and the follow-on headway (time between

two consecutive queued vehicles) parameters is fundamental during the process.

The best-known gap acceptance model for roundabouts was created by

Troutbeck, in 1989 (21). The circulating headway distribution was the Cowan M3

distribution, in which a proportion of vehicles was assumed to be aggregated with

a fixed headway while remaining vehicles had exponentially distributed

headways. Wu (22) introduced a universal procedure for calculating the capacity

at non signalized intersections by using a queuing theory. The High Capacity

Manual (HCM) included a gap acceptance model (8) based on the number of

entry and circulating lanes, and whether the entry lane was nearside or offside.

In summary, the main differences of above approaches lies in the assumed

headway distributions, and the formulation of the relevant parameters such as

the proportion of clustering (18). Nonetheless, it is recognized that they still

require more improvements. First, the limited priority process in roundabouts is

rarely included (23). Second, it is demonstrated that pedestrian’s crossings have

effects in the capacity of entry and exit sections of the roundabouts (24). Third,

there is a lack of reliable estimations about some of parameters that are present

in most gap-acceptance formulas (25). Fourth, existing capacity models are not

able to assess the overall performance of unconventional roundabouts and

innovative solutions, such as the turbo-roundabouts.

The stochastic microscopic simulation is an alternative approach that provides a

good flexibility in terms of the assessment of capacity models in roundabouts.

Their validation and reliability is highly dependent of an accurate representation

of vehicle–vehicle interactions, which can be difficult to replicate, even with actual

observed data (18).

In (26), the traffic delay of roundabouts was compared to yield control, signal

control, two-way stop control, and four-way control intersections by using

aaSIDRA model without any field data. The scenarios for roundabouts included

different direction left turns (10%, 20% and 30%) under different traffic flows while

two-way stop control, four-way stop control, signal control, and yield control were

modeled with a 10% left turn. Roundabout showed as the most effective for heavy

traffic intersections with two-lane approaches. Alternatively, Meredith and Rakha

demonstrated that, under over-saturated conditions in the approach lane (e.g.

traffic flows below 500vph), signal control was not necessarily worse than



roundabout (27). Bastos Silva and Vasconcelos (24) studied the effect of exit

pedestrian crossing in the upstream entry capacity and average speed, under

different levels of motorized and pedestrian demand. The findings suggested that

the crosswalk near to the exit provided a decrease in terms of traffic performance.

For high traffic flows, the authors also recommend the possibility to install the

crosswalks further away from the exit. In the research of Zheng and Qin (28), it

was demonstrated that roundabouts recorded lower traffic delays than pre-timed

traffic signals at intersections with moderate queue lengths. They also predicted

that roundabouts efficiency decreased as queues length increased.

2.3.2. Emissions

In spite of having an extensive body of research in macroscopic (e.g. COPERT

(29), MOBILE6 (30) or TREM (31)), mesoscopic (e.g. TEDS (32)), and

microscopic (VT-MICRO (33), CNEM (33), VSP (34) and MOVES (35)) emission

models, their applications on roundabouts case studies are very limited.

COPERT (29) and MOBILE6 (30) models, for instance, are not recommended in

micro scale impact of corridors with traffic interruptions (e.g. pay tools,

roundabouts) (36, 37) since they assumed that emission rates are constant for

all speed ranges. The mesoscopic aaSIDRA intersection model includes an

emission module that uses a four-mode elemental model (cruise, deceleration,

idle and acceleration to cruise) to estimate fuel consumptions and pollutant

emissions. Nevertheless, the impact of each stop and go cycles are not taken

into account (16). Traffic and Emission Decision Support (TEDS) model can be

applied in urban corridors, but only single lane roundabouts are included in this

analysis (32).

The microscopic models aim to provide accurate emissions estimates at the

roundabouts operation levels. One of the most representative and widely

accepted micro-scale models is the Comprehensive Modal Emission Model

(CNEM) (33). This model provides accurate emissions estimation of Light-Duty

Vehicles (LDV) under different vehicle’s operating modes. However, it lacks

accurate emissions predictions at specific situations. VT-Micro is able to estimate

emission rates and fuel consumptions per vehicles based on their instantaneous

speeds and accelerations. VT-micro’s main disadvantage is the not direct

inclusion of road grade parameter on emissions calculations (33). The Motor

Vehicle Emission Simulator (MOVES) (35) allows estimating emission rates that

vary with speed for particulate matter (PM) and greenhouse gases (GHG), but is

most designed for the US reality.

Frey et al. (30, 34) developed the “Vehicle Specific Power” (VSP) methodology

to estimate emissions for a single vehicle. On-board vehicle activity and

emissions measurement using portable emissions measurement systems

(PEMS) enable data collection under real-world conditions at any location

traveled by vehicles on a second-by-second basis. This microscopic model is

based on vehicle speed, acceleration/deceleration and road grade and has

proven to be very effective in estimating instantaneous emissions for both light-



duty gasoline and diesel vehicles (38, 39) as well as in transit buses (40). A

drawback of this approach is the little detail on fleet categories. Some previous

studies have documented the effective use of VSP in analyzing emission impacts

of roundabouts footprints in urban corridors both in the United States (e.g. (41,

42)) and in Europe (e.g. (41, 43-45)).

In the following paragraphs, the most relevant studies on roundabouts operations

on vehicular emissions are reported. These studies are divided in two main

groups: in the first group (41-44, 46, 47), only field data was used; in the second

group (45, 47-50), a microscopic traffic model was linked with an external

emission model.

The research of Coelho et al. (41) and Salamati et al. (44) should be highlighted.

Coelho et al. (41) identified three characteristic speed profiles for a vehicle

approaching both a single and multi-lane roundabouts: 1) no stop; 2) stop once

and 3) multiple stops. They also found that the relative occurrence of these

profiles were dependent of the entry traffic and conflict traffic flows. Based on

these findings, the same authors developed regression models for approaching

vehicles in single-lane roundabouts in urban areas. Built on above research,

Salamati et al. (44) developed similar regression models in each approaching

lane (right vs. left) at multi-lane roundabouts. Whilst methodologies of above

studies are generalized to estimate emissions of roundabout footprints, its

application cannot be extended to corridors. This is due that only traffic data

related to the downstream and the circulating Areas of the roundabouts is

considered. Even if proposed methodologies were applied for each roundabout

along a corridor, the contribution of mid-block sub-segments, that is, within two

consecutive roundabouts, would not be included.

In a Swedish study (46), an environmental assessment was conducted in order

to compare signalized and yielded traditional intersections on arterials which

were rebuilt as small single-lane roundabouts. The authors found that

roundabouts can reduce emissions up to 29% and 21% of CO and NOX,

respectively. Rakha et al. (51) demonstrated that both single and two-lane

roundabouts yielded reduce traffic delays and CO emissions than one-way stop

controlled in a three-way intersection. Anya et al. (43) showed that the

environmental benefits posed by the conversion of a signalized intersection to a

two-lane roundabout in an urban corridor was only relevant, at the intersection-

level, in the right turns movements from the minor street to the main street. They

also found that, at the corridor-level, turning movements from the main street

produced higher total emissions while turning movements from the minor street

produced lower total emissions after the roundabout implementation. More

recently, Mudgal et al. (42) demonstrated that acceleration events at the

circulating and exiting areas of roundabout contributed to more than 25% of

emissions for a given speed profile.

The majority of the studies that used a micro-simulation approach focused on the

comparison between single e multi-lane roundabouts with others intersections

forms.



Mandavilli et al. (48) showed that both CO2 and CO were reduced by 42% and

59%, respectively, on stop-controlled intersections which were replaced by

roundabouts. Ahn et al. (49) used VISSIM and INTEGRATION in conjunction with

air quality models (CMEM and VT-Micro) to compare emissions at a signalized,

stop controlled intersection and high-speed roundabout. The research showed

that roundabouts yielded fewer delays and queue lengths, but produced more

emissions than other alternatives. Chamberlin et al. (50) examined CO and NOX

emissions produced by a three-leg intersection and a single-lane roundabout.

The authors applied PARAMICS traffic model with MOVES and CNEM emission

models. Both models estimated higher emissions for the roundabout when

compared to the three-leg intersection. Likewise, Rakha and Jackson (47)

employed INTEGRATION and VT-Micro to assess the environmental impacts of

an isolated intersection served by a single-lane roundabout, traffic signal, all-way

and two-way stop controlled intersections. The roundabout predicted to have

lower emissions of CO2, CO, NOX and HC, compared to other alternatives.

Vasconcelos et al. (45) compared capacity, safety and emissions levels from a

turbo-roundabout compared with a conventional single-lane and two-lane

roundabouts. AIMSUN traffic model and VSP emission model were used.

Assuming equal traffic demands, turbo-roundabout produced more CO2 and NOX

emissions than two-lane roundabout. Turbo-roundabout also showed as less

robust concerning the directional splits of the entry traffic, namely when most of

the vehicles turned left.

In summary, the candidate did not find a definitive conclusion about the

environmental benefits from roundabouts operations. It should be noteworthy that

the majority of them did not assess the capability of the traffic models to capture

real-world vehicle power distributions. Previous studies demonstrated that

microscopic traffic models tend to overestimate acceleration and deceleration

values in relation to the observed data (52, 53).

2.3.3. Safety

Although the mostly of crashes at roundabouts results in property damage only

(54), there are an extensive body of research focus on demand and improvement

of crash models (7, 55-57) applied to roundabouts and intersections.

Nonetheless, the limited ability of above safety models to properly reflect crash

causality has its source in cross-sectional analysis applied to the estimation of

the intrinsically complex safety factors with highly aggregated and frequently poor

quality of data. In this context, a clear theoretical approach to understand the

theoretical assumptions that are behind of that models is needed. Moreover, the

analysis of the different surrogate indicators associated to such models (58-60)

and their potential for crashes estimation must be considered (61).

Dijkstra et al. (61) applied PARAMICS traffic model to predict the number of

crashes, studying the possibility of the relationship between calculated traffic

conflicts and crashes. Their findings suggested a Poisson log-linear statistical

relationship between the number of simulated conflicts and observed crashes.

Zheng et al. (28) performed an exhaustive analysis of roundabout crash patterns



in the United States (US) and developed a method to calculate crash type

percentages. In (62), four crash prediction models were developed to evaluate

the potential of speed as a measure of the road safety.

The micro simulation models have been played an important role in the field of

safety analysis of roundabouts as occurred in emissions and capacity fields.

FHWA developed a trajectory-based post-processing tool in its Surrogate Safety

Assessment Model (SSAM) (63) to analyze and evaluate conflict points from

some traffic models (e.g. VISSIM, AIMSUN). This tool processes trajectory files

that store the speed, acceleration and position of each simulated vehicle during

each simulation time-step and calculates surrogate safety measures and also

classifies traffic conflicts based on conflict angle. Al-Ghandour et al. (64)

developed conflict prediction models for five zones of single lane roundabout by

using a Poisson regression, and compared them with simulated traffic conflicts

obtained from SSAM model. The authors found good correlations between SSAM

predicted conflicts and crashes, but recognized the need for additional research

focus on the comparison between SSAM outputs and real crash data.

Huang et al. (60) proposed a two-stage procedure to calibrate and validate

VISSIM to be used in conjunction with SSAM model to assess safety at signalized

intersections. The authors concluded that the Mean Absolute Percent Error

(MAPE) for total conflicts are predicted to be reduced from 43% to 24%. Data

analysis results showed that there was a reasonable consistency between the

simulated and the observed rear-end and crossing conflicts. However, they

recognized that the simulated conflicts generated by VISSIM and SSAM were not

good indicators for traffic conflicts which were generated by unexpected driving

maneuvers such as illegal lane-changes in the real world. Vasconcelos et al. (65)

recognized that despite some limitations regarding the nature of traffic models,

SSAM is capable of evaluating the safety of different roundabouts layouts. Some

authors suggested specific procedures to calibrate and validate simulation

models for safety assessment but this is still an ongoing research field (66, 67).

2.4. Summary of literature review and motivations

From above-mentioned studies, two main relevant gaps in the research were

observed:

The assessment of the environmental and safety impacts were found in

isolated roundabouts rather than a sequence along a corridor;

It is noted that vehicle aspects of safety and fuel efficiency / emissions are

treated independently and not in an integrated manner.

With these concerns in mind, one of the main motivations of this research is to

investigate the impact of the different sub-segments of each pair of roundabouts

that are installed in corridors on speed enforcement and pollutant emissions. The

second motivation is to develop an advanced statistical model to predict crashes

as a function of parameters such as traffic volumes, geometry and driver behavior

conditions appropriate for Portuguese facilities.



This research seeks to contribute to a methodology that can incorporate the

corridor geometric features and traffic stream characteristics to assess traffic

performance, energy use and emissions. This methodology can be useful to the

local authorities to the support of decisions in the field of capacity and emissions.

The originality and scientific quality of the proposed research lie in the integration

between these various fields of inquiry in order to understand their relationship

and in the application of the developed methodology to any corridor with

roundabouts. The integration of three areas on those uninterrupted flow facilities

is worthy of research and analysis at this stage.



3. Methodology and Methods

The research work plan consisted of six interrelated tasks:

Task 1: Literature Review.

Task 2: Development of numerical models to assess the emissions

impact of vehicles in roundabouts installed at corridors.

The numerical models include different sub-models to estimate vehicular

emissions in urban corridors taking into account traffic characteristics and

design geometric features.

Task 3: Development of a microscopic simulation platform of traffic, emissions and safety.

This approach attempts to represent traffic, emissions and road safety in

order to evaluate traffic performance evaluation in world study cases.

Task 4: Application of the microscopic platform to examine alternative scenarios.

Both traffic systems will be compared, in order to study several alternative scenarios, namely, different traffic flow demands (peak and off-peak periods), change in pedestrian flows, and change in driver’s behavior (reduction of vehicle’s speed). This task also intends to compare overall corridor performance with individual roundabout and intersection.

Task 5: Application of a multi-objective analysis.

This task focused on the development of an algorithm to perform a multi-objective evaluation with the main goal of optimizing roundabouts corridors systems in the field of emissions and safety.

Task 6: Thesis Writing.

A summary of the proposed tasks is depicted in Figure 1. The expected duration

of this research is four years (1st of June 2013 – 31st of May 2017). The major

tasks in the first year and second years will be to collect the data different

roundabouts corridors as well as to investigate the main gaps of the current

literature with respect of this thesis topic (&task1). Additionally, the development

of numerical emission models for corridors with roundabouts will be made

(&task2). In the second and third years (1st of September 2015 – 31st of August

2016), the modelling simulation platform, the integration of the emission module

with the traffic model, as well as the calibration and validation procedures will be

developed (&task3). After that, traffic scenarios will be implemented and analysed

(&task4). The modelling platform will be supported by VISSIM model for traffic

simulation, and Vehicle Specific Power methodology for emissions estimation.

Next, a multi-objective analysis (&task5) will be made in order to compare the

results obtained for different systems of roundabouts. In the last year, the written

of this thesis will be done (&task6). The evolution of activities is described in

Table 1.



Table 1: Evolution of the proposed activities.

Year Month Task1 Task2 Task3 Task4 Task5 Task6

2013

VI

VII

VIII

IX

X

XI

XII

2014

I

II

III

IV

V

VI

VII

VIII

IX

X

XI

XII

2015

I

II

III

IV

V

VI

VII

VIII

IX

X

XI

XII

2016

I

II

III

IV

V

VI

VII

VIII

IX

X

XI

XII

2017

I

II

III

IV

V



Figure 1: Tasks of the work plan proposed for this Doctoral Research.

3.1. Proposed Tasks

In the following sections, the above tasks are presented and characterized. Since

the literature review (task 1) was done during the course “Project Thesis I”, only

the tasks 2, 3, 4 and 5 are explained.

TASK 1: Literature Review

TASK 2: Numerical emissions model

for roundabouts in corridors

TASK 3: Traffic, emission and safety

modelling

TASK 4: Scenarios Evaluation

TASK 5: Multi-objective optimization

TASK 6: Thesis Writing



3.1.1. Numerical model for emissions assessment in corridors

This section quantifies emissions impacts of corridors with roundabouts which includes a characterization of the different sub-segments along the corridor as well as the data needed to develop the numerical model.

3.1.1.1. Definition of sub-segments

Each roundabout corridor is divided into multiple sub-segments in order to

quantify the emissions impacts and identify the segments with consistently high

emissions. This level of segmentation is motivated by changes in speed as

drivers decelerate while approaching the roundabout, transverse through the

circulating lanes and accelerate while exiting the roundabout.

In order to be consistent with Highway Capacity Manual (HCM) Urban Street

(Chapter 17) the corresponding Urban Street segment is defined from

downstream yield lane from one roundabout to upstream of yield lane of the

adjacent roundabout (in the direction of traveling) (68).

The proposed segmentation is illustrated in Figure 2 for a specific through

movement. The exhibit shows two pairs of roundabouts RBT1/RBT2 and

RBT2/RTB3, separated by (Urban Segments) B and C, respectively. Based on

vehicle activity data and patterns in speed profiles, the four different sub-

segments are identified. For each of these sub-segments the speed and

acceleration patterns are different. In the case of Segment A (remaining

segments have similar descriptions):

Sub-segment A1: Vehicle decelerates to negotiate the circulating area of RBT1 and then accelerates as it is leaving the roundabout (deceleration/acceleration pattern);

Sub-segment A2: Vehicle accelerates after exiting the roundabout back to cruise speed (acceleration only);

Sub-segment A3: Vehicle drives at cruise speed without significant acceleration or deceleration rates (constant speed);

Sub-segment A4: Vehicle starting to decelerate while approaching the next roundabout (deceleration only).

The sub-segments A1, B1 and C1 corresponded to the downstream influence of

each roundabout, and B4 and C4 corresponded to the upstream influence of each

roundabout. Midblock sub-segments are associated to the sub-segments A3 and

B3. For the purpose of analysis and supported by empirical measurements, the

downstream, midblock and upstream sub-segments will be assumed to be equal

in length.

Roundabouts Influence Area (RIAs) is defined as the roundabout circulating area

and any upstream and downstream distance needed for

deceleration/acceleration from/to free-flow speed (unimpeded speed) through the

corridor. It is important to note that in the case of closely spaced roundabouts,

the RIA from an upstream roundabout might overlap with an RIA from a

downstream roundabout and therefore the vehicle would not get a chance to

reach free-flow or cruise speed (sub-segment A3). Therefore sub-segment A3



might not be available for some parts of a roundabout corridor and such design

characteristic has an impact on vehicle speed profile and therefore emissions.

Figure 2: Segments and sub-segments definition of a roundabout corridor and illustrative speed profile.

3.1.1.2. Development of models

From the literature, it is clear that there is a lack of numerical models that are able

to quantify emissions at corridors with roundabouts. The existing models (41, 44)

were developed for single-lane and multi-lane roundabouts at isolated

intersections. Although methodologies of above studies are generalized to

estimate the footprints of emissions at roundabouts, its application cannot be

extended to corridors. This is mostly due that only traffic data related to the sub-

segments B4 (upstream), C1 (circulating area) and C2 (downstream) is

considered. Even if proposed methodologies were applied for each roundabout

in the corridor, the contribution of sub-segment B3 (mid-block) would not be

included.

It must be also taking into account that above methodologies assumed that

vehicles reached cruise speed as they exit the roundabouts. However, the

vehicle’s speed in several mid-blocks sub-segments can be lower than free-flow

speed (e.g. C3), especially in the situation of closely spaced roundabouts.

Another consideration that must be mentioned regards the number of circulating

lanes of roundabouts and the number of approach lanes of approach roadways

between adjacent roundabouts. There are two possible corridors layouts: 1) one

Time

A4 B1 B2 B3 B4 C1 C2 C3 C4 D1 D2

Circulating

Speed Acceleration Constant

Speed Deceleration

RBT2 RBT1 RBT3

Segment A Segment B Segment C Segment D

Sp

ee

d



with only single lane roundabouts and one approach lane between connection

roads and 2) one with a mix of multi-lane roundabouts and single lane

roundabouts and approach roads with one or more lanes. Accordingly, two

different numerical models are needed to characterize those situations.

This task will be focused on the development of such models that can estimate

vehicular emissions for vehicles that drive through the corridor based on speed

profiles occurrences (no stop, stop once and multiple stops) and traffic operation.

It is well known that these proportions are found to be dependent on the prevailing

levels of congestion, simply expressed as the sum of the approach and conflicting

flow rates (41). Considering to the roundabouts installed along corridors, one

must be considered not only all movements from remaining entry legs but also all

conflicting flow rates from each one of approach lanes.

In summary, this task will be composed by four main steps:

1. To quantify traffic and emissions impacts of corridors with only single-lane

roundabouts in urban and rural corridors;

2. To identify the different speeds profiles that vehicles can experiences as

they drive through the corridor;

3. To explain the interaction between roundabout system operational

variable, mainly the approach entry volumes, conflicting or circulating

volume and geometry of the roundabout, and the resulting vehicle

emissions;

4. Repeat the steps 1, 2 and 3 for corridors with multi-lane roundabouts and

more than one approach lanes.

In summary, this task is an empirical approach founded on field measurements

of driver and vehicle characteristics. A review of the methodological steps is

exhibited in Figure 3.

A wide range (between 5 and 10 corridors) of geometrics and traffic data must be

obtained for different roundabouts corridors in order to cover different geometric

and operation features. Furthermore, the effect of pedestrians, presence of

striping on the roadway, lane configuration, vertical geometry and spacing

between adjacent roundabouts along the corridor should be noted. The

observational data will be obtained from video files. Each vehicle will be tagged

using specific software and the following information will be captured: vehicle

type, entry lane, destination lane and timestamp at different instants. In the case

of minor vehicles, parameters like arrival and entering times (to obtain gap

acceptance and follow-up data) will be recorded and also when crossing a

specific section on the approach lane (to obtain delays). Concerning vehicles on

major streams, times on conflicts points and upstream and downstream areas will

be obtained in order to study the yielding process in the circle.

It should be emphasized that a similar methodology will be also carried out at

corridors with traffic signals.



Figure 3: Methodological overview for the characterization of traffic and

emissions in corridors.

3.1.2. Microscopic simulation platform of traffic, emissions and safety

The core of the proposed methodology is to develop a microscopic simulation

platform of traffic, emissions and safety. This platform allows assessing corridors

with roundabouts impacts on traffic performance, emissions and safety. Figure 4

illustrates the modelling framework.

Firstly, the traffic model is presented (see section 3.1.2.1.) followed by the data

collection on each selected study case (see section 3.1.2.2.). After that, the

emission (see section 3.1.2.3.) and safety (see section 3.1.2.4.) models are

described. Lastly, detailed description of the calibration and validation procedures

is made (see section 3.1.2.5.).

Segments

Definition

- Speed

- Acceleration

- Slope

- Entry traffic

- Conflict traffic

- Geometry

Per roundabout

along the

corridor

Sites Selection

Identification of the speed profiles in roundabouts

with corridors

Algorithms to determine the probability of speed

profiles occurrence

CO2, CO, NOX and HC Emission Estimation for each

speed profile

Upstream;

Circulating;

Downstream;

Midblock.



Figure 4: Methodological simulation framework of the modelling platform.

VSP Modes distribution

Emission rates (CO2, CO,

NOX and HC)

Traffic volumes;

Speed

Travel times;

Gap acceptance data;

Simulated traffic conflicts.

YES

NO

Microscopic emission model

(VSP)

Data collection

Traffic flows;

Fleet composition;

O/D matrices;

Speed

Acceleration/deceleration

Road grade;

Gap acceptance parameters.

Microscopic traffic simulation

(VISSIM)

Driver Behaviours

Parameters of Traffic Model

Optimization of selected

parameters

Traffic flows;

Acceleration;

Speed.

Stopping Criteria

Optimization of selected

parameters

Comparison between

observed and estimated data

CALIBRATION

VALIDATION



3.1.2.1. Traffic Modeling

The increasing of computing performance has yielded an increased use of

microscopic traffic models in both environmental and safety fields. Specifically,

these models allow exporting the vehicle position and vehicle dynamics data

(acceleration/deceleration and speed) second-by-second that provides accurate

emissions and relevant safety measures estimation. Note that their simulation

outputs as speed and acceleration/deceleration can be employed in

instantaneous emission and safety models. These models can be used to assess

the environmental and safety impacts of different traffic management strategies

applied to the road network traffic, such as, route diversion, variable speed limits

or traffic signal coordination (69). In Fontes et al. (70) a review of important

research on this topic is presented. Two procedures are usually adopted: 1) a

traffic model linked with an external emission model or 2) the emission model

which are included in the traffic model. Despite the fact microscopic simulation

models have been identified as having potential for emissions evaluation, there

are few studies reported their capabilities to capture the real-world vehicle power

distributions.

The VISSIM traffic model will be used to simulate traffic operations at corridors

with roundabouts and traffic signals (71). This microscopic traffic model is chosen

because of the possibility to define and use different road-user behavior

parameters and sub-models (car-following, gap-acceptance, lane change) for

different traffic controls modelling. VISSIM is widely recognized as a powerful tool

for roundabout operational analysis since it can be calibrated to match

deterministic capacity relationships (e.g. (72, 73)). VISSIM also allows exporting

full disaggregated trajectory files that can be used by external applications to

assess environmental and safety impacts, as described on the following sections.

By using a vehicle actuated programming, this traffic model provided an accurate

representation of the gap acceptance model in intersections. VAP tool also

enables VISSIM to simulate programmable traffic actuated signal controls, both

phase or stage based. Moreover, other explanations have supported the

application of VISSIM in this research:

The possibility to define different road-user behavior parameters and sub-

models such as car-following and gap-acceptance that are essential for

roundabouts modeling;

The possibility to define vehicle performance parameters such as desired

and maximum acceleration per vehicle type and class;

The possibility to record some traffic performance parameters such as

delays or vehicle stops per vehicle type and class;

A Dynamic Traffic Assignment (DTA) that enables users to define routes

and origin-destination routes more efficiently;

The possibility to store different vehicle dynamics data (speed,

acceleration) at a high time resolution (until one tenth of a second);

An Application Programmer Interface (API) that allows users to define

added external applications functionality and access objects and data in

the model during the simulation process.



3.1.2.2. Data collection

Several inputs are required to the integrate platform of microsimulation. For this

process, the following data collection must be provided:

Traffic counts by vehicle class;

Pedestrian counts;

Speed counts;

Time-dependent origin/destination (O/D) matrices;

Vehicle dynamics (speed, acceleration/deceleration and road grade);

Follow-up times by vehicle class;

Time gap distributions (acceptance and rejection gaps);

Cycle length and green times (corridors with traffic signals).

Traffic and pedestrian volumes, as well as time-dependent O/D matrices, will be

gathered from two video cameras installed at strategic points along the corridors.

Time-gap distributions data (gap-acceptance and gap-rejection) and follow-up

times for all turning maneuvers will be also extracted from the videotapes. The

critical gap is estimated by applying the Raff’s method (74).

Data will be collected at morning and evening peak, and off-peak periods over

several typical weekdays (Tuesday to Thursday), under dry weather conditions

and using different drivers.

A QSTARZ GPS Travel Recorder (75) will be installed at the test vehicle to obtain

some of parameters related to the vehicle activity data (instantaneous speed and

acceleration/deceleration) and the selected sites characteristics (latitude and

longitude). This equipment supports a Smart Log function and includes a vibration

sensor to start and stop automatically logging with a preset time schedule, and

smartly manage power saving and waypoint “saving”. Moreover, the software

supports providing multi-condition settings to customize specific travel record at

high time resolutions (second-by-second basis) (75).

Along with the GPS, it is also used a CarChip Pro by Davis Instruments. This On-

Board Diagnostic reader is compatible with passenger cars and light duty trucks

(76). The car chip records second-by second vehicle dynamics such as vehicle

speeds, distance traveled, acceleration and braking rates. It also records

additional engine performance measures such as coolant temperature, intake air

temperature or fuel pressure (76).

Over than 100 GPS travel runs for each movement will be extracted and identified

for this research. This number of GPS data samples provided a suitable

representation of the local traffic conditions.

3.1.2.3. Emissions Estimation

Regarding to the emission calculation, the “Vehicle Specific Power” (VSP)

methodology is employed. This microscopic model is based on vehicle speed,

acceleration/deceleration and road grade and has proven to be very effective in

estimating instantaneous emissions for both gasoline and diesel (38, 39) as well

as for transit bused (40). Several motivations have supported the use of VSP



methodology in this thesis. First, VSP can be applied for both US and European

car fleet because it includes a wide range of engine displacement values.

Second, VSP-based emission model not only has a better estimation of vehicle

emissions than a speed-based emission model, but also is capable of reflecting

the emission changes under different operating modes (77). Some previous

studies have documented the effective use of VSP in analyzing emission impacts

at roundabouts (41, 43-45). The VSP values are categorized in 14 modes, and

an emission factor for each mode is used to estimate CO2, CO, NOX and HC

emissions from Light Duty Diesel Vehicles (LDDV<1.8 L), Light Duty Gasoline

Vehicles (LDGV<3.5L) and Light Commercial Diesel Vehicles ((LCDV<2.5L) (30).

Equation 1 provides the VSP calculation for Light Duty Vehicles (LDV) (78):

31.1 9.81 sin arctan 0.132 0.000302VSP v a grade v (1)

Where:

v = Instantaneous speed (m/s);

a = Instantaneous acceleration or deceleration (m/s2);

grade = Road grade (decimal fraction).

These terms represent the engine power required in terms of kinetic energy, road

grade, friction and aerodynamic drag (38). VSP values are usually grouped in

combinations of 1 kW/ton from -50 to +50. Then, these values are categorized in

modes so that each mode generates an average emission rate. Concerning

transit buses, VSP is estimated using typical coefficient values and expressed by

equation 2 (40):

39.81 sin 0.092 0.00021VSP v a grade v (2)

where:

v = Instantaneous speed (m/s);

a = Instantaneous acceleration or deceleration (m/s2);

grade = Road grade (decimal fraction).

In this case, VSP combinations are grouped in 8 modes that are corresponding

to range values from -30 to 30 kW/ton (40). After the VSP value is calculated, the

VSP mode is determined by the VSP range. The corresponding modal emission

CO2, CO, NOX and HC are then obtained.

To estimate the amount of emissions generated by Heavy Duty Vehicles (HDV),

passenger car equivalents (PCEs) values for HDV at roundabouts (79) will be

used in this thesis.

An integrated traffic-emission micro simulation platform (built in C#) will be

developed by integrating the microscopic traffic simulator VISSIM and the

instantaneous emission model VSP.



3.1.2.4. Safety

The main goal of this new safety assessment approach is a software package

developed by the Federal Highway Administration (FHWA) (Surrogate Safety

Assessment Model – SSAM) (80) that automates traffic conflict analysis by

processing vehicle and pedestrian trajectories on a second-by-second analysis.

This approach has all the common advantages of simulation (e.g. safety

assessment of new facilities before the occurrence of crashes), but also has

some drawbacks: current microscopic traffic models are not able to model some

crashes types such as sideswipe collisions, head-on collisions or U-turn related

collisions (81).

For each interaction, SSAM stores the trajectories of vehicles from the traffic

model and records surrogate measures of safety and determines whether or not

that interaction satisfies the condition to be deemed a conflict. The research team

used the Time-to-Collision (TTC) and Post-Encroachment Time (PET) as

thresholds to define if a given vehicle interaction is a conflict and the Relative

Speed (DeltaS) and Maximum Speed (MaxS) as proxies for the crash severity.

Their definition is posed as follows (81):

TTC is the minimum time-to-collision for a vehicle pair of both maintained their instantaneous heading and speed. If at any time step the TTC drops below a given threshold (1.5 seconds and 2 seconds, as suggested for vehicle-vehicle (63) and vehicle-pedestrian events (82), respectively), the interaction is tagged as a conflict;

PET is the time lapse between end of encroachment of turning vehicle and the time that the through vehicle actually arrives at the potential point of collision. If at any time step the PET drops below a given threshold (5 seconds, as suggested for vehicles (83), the interaction is tagged as a conflict;

MaxS is Maximum of the speeds of the two vehicles involved in the conflict event (MaxS);

DR is the rate at which crossing vehicle must decelerate to avoid collision;

DeltaS is the difference in vehicles (or pedestrians) speeds as observed at the instant of the minimum TTC (81).

The size of the surrogates TTC, PET and DR are intended to indicate the severity

of the conflict event. This measures how likely a collision would result from a

conflict, such that:

Lower TTC indicates higher probability of collision;

Lower PET indicates higher probability of collision;

Higher DR indicates higher probability of collision.

MaxS and DeltaS are used to indicate the likely severity of the (potential) resulting

collision if the conflict event had resulted in a collision instead of a near-miss,

such that:

Higher MaxS indicates higher severity of the resulting collision;

Higher DeltaS indicates higher severity of the resulting collision resulting

collision (58)



3.1.2.5. Calibration and Validation

Calibration and Validation procedures are very crucial to make simulation model

looks real. Model calibration is defined as the process in which parameters of the

traffic model are tuned and fitted so the model will accurately represent field-

measured traffic data. Model validation is defined as the process of comparison

between traffic data from the model and data collected from the field (84).

In the early research, many authors have used single-criterion methodologies in

which the extent of errors between estimated and observed measures such as

traffic flows (83), travel time (85) or capacity (85), were minimized. Others have

developed microscopic simulation platforms by using multiple criteria measures

in which weight summation of each single-criteria was minimized (86, 87). Both

single-criteria and weighted summation were shown as inadequate for calibrating

microscopic traffic platforms (67) since traffic dynamics is a multi-faceted problem

and some trade-offs between criteria can be found. Another method to calibrate

traffic models is to regard model calibration as an optimization problem in which

several parameters are tested to satisfy the objective function. Nonetheless, the

computational complexity arise from this procedure can be exponential (87). In

this context, Punzo et al. (88) recommend a selection of a subset of parameters

to perform a sensitivity analysis and also to narrow the range of possible values

of each parameter. In the context of roundabouts operations, Vasconcelos et al.

(89) proposed a method to calibrate the car-following model for the effect of driver

variability in the speed-flow relationship. In (90), VISSIM was calibrated using

field gaps in two-lane roundabouts.

In this research, the model calibration is focused on the driver behavior

parameters of the traffic model, while model validation is related with O-D

matrices, travel times and VSP mode cumulative distributions, CO2, CO, NOX and

HC emissions, and traffic conflicts. Note that different data samples will be used

for calibration and validation procedures.

Calibration of the VISSIM parameters is made by modifying driver behavior and

vehicle performance parameters of the traffic model and examined their effect on

traffic volumes and speed for each link. The main driver behavior parameters are

divided into car-following parameters (average standstill distance, additive and

multiple part of safety distance), lane-change parameters, gap acceptance

parameters (minimal gap time and minimal headway) and simulation resolution

(91).

To optimize the aforementioned parameters, a procedure based on the

Simultaneous Perturbation Stochastic Approximation (SPSA) genetic algorithm

is used. The objective function, the minimization of Normalized Root Mean

Square (NRMS), is denoted by Equation (3).

NMRS is defined as the sum over all calibration periods of the average of the sum

over all links of the root square of the square of the normalized differences

between observed and estimated parameters (92). The normalization enables

the consideration of multiple performance measures, in this case, link volumes,

speeds and acceleration/deceleration. The calibration procedure is formulated as

follows:



1 1 2 2 1 2 31

11

T

t

NRMS W C W C W W CN

(3)

2

1 1

I i i

ii

v vC

v (4)

2

2 1

I i i

ii

s sC

s (5)

2

3 1

I i i

ii

a aC

a (6)

Subject to: Lower bound ≤ θ ≤ Upper bound Where: vi = Observed link volumes for link i; v (θ)i = Estimated link volumes for

link i; si = Observed speeds for link i; s (θ)i = Estimated speeds for link i; ai =

Observed accelerations for link i; a (θ)i = Estimated accelerations for link i; N =

Total number of links in the coded network; T = Total number of time periods t;

and W1, W2 = Weights factors used to assign more or less value to volumes,

speeds or acceleration based on the trustworthiness of the corresponding data

(note 0 ≤ W1, W2 ≤ 1).

Considering the calibration criteria, the current accepted practice, as

recommended by the FHWA is to rely the Geoffrey E. Havers (GEH) statistic for

assessing goodness-of-fit. The difference between observed and estimated link

volumes should be less than 5% for at least 85% of the coded links (93).

The model validation is focused on the comparison between estimated and

observed O/D matrices, pedestrian volumes and travel times for a preliminary

number of runs (between 10 and 20, as suggested by Hale (94)). GEH statistic

(93) and Root-mean-square percentage error (RMSPE) (95) will be used as

measures of the goodness-of-fit. Along the validation of previous measures, the

observed and estimated VSP cumulative probability distributions will be also

compared. This validation step allows assessing differences in the

acceleration/deceleration profiles between field measurements and simulation.

Since the number of data sets (number of seconds) is roughly higher than 40,

two-sample Kolmogorov-Smirnov test (K-S) for a 95% confidence is deemed

appropriate to assess if the probability distributions of two samples are different

(96). K-S test is recognized to be one of the most useful and nonparametric

methods for comparing two samples, as it is sensitive to differences in both

location and shape of the empirical cumulative distribution functions of the two

data samples. Moreover, it is also suggested when there is a natural ordering of

the modes (bins) (97). The resulting estimated CO2, CO, NOX and HC emissions

are compared with observed values. Lastly, the simulated conflicts that can be



recorded in SSAM model will be compared by real crash data from a crashes and

conflict model for roundabouts.

3.1.3. Scenarios evaluation

After developed the modeling platform of traffic and emissions in corridors and

traffic signal corridors, several scenarios will be implemented. The main purpose

of this task is to understand how emissions and safety performance vary in

according to the different traffic conditions.

Thus, the following scenarios will be compared:

1. Different traffic demands (peak and off-peak periods);

2. Different pedestrian flows (peak and off-peak periods);

3. Different traffic signal settings (fixed versus actuated);

4. Reduction of vehicle’s speed.

The emissions and safety performance of each scenario will be evaluated in

terms of: a) Overall roundabouts corridor vs. Overall traffic signal corridor, and b)

Overall roundabouts corridor vs. individual roundabout along the corridor.

3.1.3. Multi-objective analysis

This section gives a brief description of the multi-objective optimization task

proposed in this research, which includes the formulation of the multi-objective

optimization (see section 3.1.3.1.) and the selected algorithm (see section

3.1.3.2.) to solve its solution.

3.1.3.1. Micro simulation optimization formulation

Multi-objective optimization is concerned with the solution of problems with

multiple, often conflicting, criteria. In particular, the optimization problem consists

of maximizing or minimizing a function by choosing different input values from a

set of available alternative. The challenge regarding the application of this

approach is that, instead of a unique optimal solution, multi-objective optimization

results in several solutions with different trade-offs among criteria. Thus, a

Decision Maker is needed to provide useful information and also to identify the

most acceptable solution.

The general multi-objective optimization problem is posed as follows (see

Equations 7, 8 and 9)

1 2min , ,...... mF f x f x f x x (7)

Subject to

0, 1,2,...,ig i I x

(8)

0, 1,2,...,jh i Jx (9)



In which: 1 2, ,...... mf x f x f x are the m objective functions, i is the number of

inequality constraints, j is the number of equality constraints, 1 2, ,..., nx x xx is

a n-dimensional design variables vector in the solution of space X.

For a multi-objective optimization problem, any two solutions x1 and x2 can have

one of two possibilities: one dominates the other or none dominates the other. In

a minimization problem, without loss of generality, a solution x1 dominates x2 if

the following two conditions are satisfied (see Equations10 and 11):

1 21,2,..., : i ii i m f x f x (10)

1 21,2,..., : i ii i m f x f x (11)

If both conditions are satisfied, the solution x1 does not dominate the solution x2.

If x1 dominates the solution x2, x1 is called the nondominated solution within the

set {x1, x2}1. This means that there exists no feasible solution x that can decrease

some objective functions without causing at least one other objective function to

increase.

The set of all possible solutions that are nondominated within the entire search

space is denoted as Pareto-optimal set. Concerning the respective objective

function values in the objective space, they are defined as the Pareto front. It

should be mentioned that, in some situations, the number of Pareto optimal

solutions can be infinite. Hence, the multi-objective optimization must include a

practical approach to identify a set of solutions, called as the best-known Pareto

set so that it can be suitably represented as the Pareto optimal set.

In summary, the multi-objective optimization should be achieve the following

goals (98):

i. The best-known Pareto set must be as near as possible to the true Pareto

set;

ii. The solutions in the best-know Pareto set must be diverse in relation to

the Pareto front as well as uniformly distributed in order to assure a faithful

representation of trade-offs;

The best-known Pareto front should be extended at both ends in order to explore

extreme solutions.

3.1.3.2. Genetic Algorithms

The issue of finding an optimal solution design is frequently discussed in the

transportation research area. Among the set of search and stochastic

optimization techniques, Evolutionary Algorithms (EAs) have recently received

increased interest. Several authors recognized merits of their applications in

solving real-world multi-objective problems (99, 100). There have been three

main independent implementation instances of EAs (101): Genetic Algorithms

(GAs) (102, 103), Evolution Strategies (ESs) (104) and Evolutionary



Programming (EP) (105). Each approach is inspired by the same basic

mechanisms of natural evolution: selection, mutation and crossover.

The GAs are often used to represent EAs. These adaptive meta-heuristic search

techniques are based on the mechanism of natural selection and natural genetics

that potentially guarantees a robust search, and also a near-optimal. This means

that neighborhood solutions from an initial incumbent solution (value of the

objective function for the problem being solve that is worse than the optimal

objective function value) is allowed to move in the direction of the worse objective

(103). The GAs can deal successfully in a wide range of problem areas, which

include multi-modal objective problems, differentiable and non-differentiable

functions, discrete and continuous parameters, as well as problems with convex

and non-convex feasible regions (102, 106). GAs have been also widely

employed in multi-objective optimization studies applied in the transportation field

such as traffic signal optimization (107-111), scheduling problems (112), traffic

routes assignment (113), network design problems (114-118), road-bicycle lanes

network problems (119), and crash safety design (120).

The first multi-objective GA, called Vector Evaluated Genetic Algorithms (VEGA),

was proposed by Schaffer, in 1985 (121). Afterwards, various multi-objective

genetic algorithms were developed: Multi-Objective Genetic Algorithms (MOGA)

(122), Weight-Based Genetic Algorithms (WBGA) (123), Random weighted

genetic algorithms (RWGA) (124), Niched Pareto Genetic Algorithms (NPGA)

(125), Nondominated Sorting Genetic Algorithm (NSGA) (126), Fast non-

dominated sorting genetic algorithms (NSGA-II) (127), and Rank-Density Based

Genetic Algorithm (RDGA) (128). The main differences of above-mentioned GAs

are related to their fitness assignment method, elitism and diversification

approaches.

Among previous multi-objective GAs, the NSGA-II will be adopted in this

research. Several explanations have supported NSGA-II used: a) diversity in

optimal solutions by incorporating the crowding distance into the fitness function

and b) a binary tournament approach that provides accommodate the selection

process (127). In contrast to the conventional GA(s) based methods, NSGA-II not

require weighting factors for conversion of multi-objective function into an

equivalent single objective function. Moreover, NSGA-II has been tested many

times in the research field of evolutionary optimization so that it has well-known

to find a good approximation of an optimal Pareto front (129). The main

disadvantage of NSGA-II is that the crowding distance only works in the objective

space (129). A good deal of research can be found in the transportation field in

which NSGA-II has been used, namely in network design problems (114, 118),

traffic signal optimization (108), crash safety design (120) and in the calibration

of microscopic traffic models (67).

The main feature of NSGA-II is the implementation of a fast non-dominated

sorting procedure. Specifically, this approach incorporates elitism, and then the

fitness sharing function is replaced by a crowded-comparison approach that

eliminates, either the dependence of fitness value to yield performance of sharing

function, and the overall complexity of the process (127).



In order to sort a population according to the level of non-domination, each

solution is compared with every other solution in the population. This requires

n×m2 comparisons for each solution while, for NSGA, the complexity of

nondominated sorting is n×m3. NSGA-II also implements elitism for multi-

objective search by storing all non-dominated solutions found so far, beginning

from the initial population. Accordingly, the convergence properties towards the

true Pareto-optimal set are improved (127).

The diversity and the spread of solutions are guaranteed without the use of

sharing parameters since NSGA-II adopts a suitable parameter-less niching

approach. It is then used a crowding distance, which estimates the density of

solutions in the objective space, and also a crowded comparison operator, which

conducts the selection process towards an uniform spread Pareto front. Note that

both discrete (binary-coded) and continuous (real-coded) are allowed to be used

by NSGA-II (127). More information about NSGA-II procedure and its basic

theories, such as selection, crossover, and mutation can be found elsewhere

(127).

Figure 5 shows the NSGA-II main steps.

NSGA-II will be implemented in Matlab. A user pre-specified maximum number

of generations is defined as the stopping (convergence) criteria of the NSGA-II

procedure. Unlike single objective optimization, where a simple convergence

criterion is sufficient to assess the performance, the multi-objective optimization

results must be assure both the convergence to Pareto Optimal Front (POF) and

the diversity of the solutions (130).

The convergence to POF is based on the comparison among the sets of non-

dominated solutions from various generations. The smaller the number of

dominated solutions, the better is the convergence. Concerning the diversity of

the solutions, it is measured by the estimating of the Spread and the Uniformity

Measure metrics. The spread of the front is calculated as the diagonal of the

largest hypercube in the function space that encompasses all solutions points. A

large spread is desired to find diverse trade-off solutions. The Uniformity Measure

identifies the presence of poorly distributed solutions by estimating how uniformly

the points are distributed in the POF. This measure it is estimated from the

standard deviation of the crowding distance (130).

Initially, the maximal number of generations will be set to 3000 for all the test

instances while the crossover and mutation rates will be set at 90% and 10%,

respectively. Each scenario will be run 20 times in the NSGA-II code. For each

repetition, above performance metrics (Number of dominated solutions, Spread

and Uniformity Measure) are computed. Once guaranteed the convergence to

POF and the diversity of the solutions in all scenarios, an equal maximum number

of generations will be employed in each scenario.



Figure 5: Basic structure of NSGA-II algorithm.

Stopping Criteria

Pareto-optimal Solution

Initiation of Population with size N

t = 0

Rank the population according to non-dominating criteria

Binary Tournament Selection

Calculate the objective function of the new population

Parents – Pt

Intermediate Crossover

Gaussian Mutation

Offsprings – Qt

Combine old and new population

Non-dominating ranking on the combined population

Calculate crowding distance of all the solutions

Get the N from the combined population

Replace parent population by the better members of the combined

population

Calculate all the objective functions

YES

NO

R t = Pt U Qt

Parents of new

generation – Pt+1



4. Results

This section presents some preliminary results regarding the data collection and

emissions calculations taken at two corridors with roundabouts located in US.

First, the introduction and main objectives are presented (see section 4.1.). After

that, the selected corridors and data collection procedures are introduced (see

section 4.2.), followed by the main results and its discussion (see section 4.3.).

Finally, the main conclusions are exposed (see section 4.4.)

4.1. Introduction and objectives

This study provides a methodology to quantify and characterize the vehicular

emissions of roundabouts at a corridor level by dividing the corridor into multiple

segments (upstream of each roundabout, circulating area, downstream of the

roundabout as well as midblock sub-segments between adjacent roundabouts).

The main goal of the methodology is to identify the locations along the corridors

where the emissions tend to be consistently high. These locations are called

emissions hotspots. The methodology is then applied to two roundabout corridors

located in San Diego (California) and Avon (Colorado) in the United States.

Thus, the main objectives are:

To analyze the vehicular emissions for each pair of roundabouts throughout the corridor;

To identify locations with the highest amount of emissions generated (hotspots) as the vehicle drives through the corridor.

4.2. Selected corridors and field measurements

Figure 6 displays an aerial view of the data collection sites investigated in this

study as well as segments and sub-segments identification for north-south and

south-north directions, based on definitions shown in Figure 2.

One of the study locations is in San Diego, California, along La Jolla Boulevard

Figure 6-a). The free-flow speed is fairly constant along this corridor, and the

spacing is approximately equal between the five roundabouts (coefficient of

variability of average spacing is 0.10). The second corridor is located in Avon,

Colorado, along Avon Road. The spacing between the roundabouts is not uniform

(coefficient of variability of average spacing is 0.46).

La Jolla corridor is 966 meters long, including five single-lane roundabouts in an

urban environment. The Avon corridor is 805 meters long, and it includes three

two-lane roundabouts and a tear drop interchange with two single-lane

roundabouts. The RBT3 and RBT4 roundabouts are located in a close proximity

to each other and therefore make the case for overlapping RIAs for this corridor.

The posted speed limit is 25 mph for both corridors.

Additionally, sites characteristics such as location, inscribed circle diameters,

entry deflection angle, distance between two adjacent roundabouts (measure



from the upstream yield lane) and some descriptive statistics are summarized in

Table 2. Free-flow speeds and traffic data information are also provided.

a)

b)

Figure 6: Aerial View of the Two Data Collection Roundabouts Corridors and segments identification (black color-North-South; red color-South-North): a) La

Jolla; b) Avon (Source: http://www.arcgis.com).

For vehicle activity characterization, the authors used second-by-second vehicle

dynamics data. A Light Duty Gasoline Vehicle (1.5<LDGV<2.5L) was used in

order to conduct several runs through the corridors. These runs were performed

during morning, evening and off-peak periods (from 7 AM to 7 PM). The research

team scouted and collected data at the two selected corridors in the fall of 2013.

A QSTARZ GPS Travel Recorder (75) was employed to obtain some of the

parameters related to the vehicle activity data (such as instantaneous speed) and

the selected sites characteristics (road grade, latitude and longitude). 70 GPS

travel runs for the two directions of through movements are analyzed for this

research (total of 280 speed profiles).

RBT1

0.2 km

RBT2

RBT3

RBT4

RBT5

Tear drop

interchang

e

0.2 km

RBT2

RBT3

RBT4

RBT5

RBT1

NORTH

SOUTH

SOUTH

NORTH

B {B1 B2 B3 B4}

C {C1 C2 C3 C4}

D {D1 D2 D3 D4}

A {A1 A2 A3 A4}

B {B1 B2 B3 B4}

C {C1 C2 C3 C4}

D {D1 D2 D3 D4}

A {A1 A2

A3 A4}

B {B1 B2 B3 B4}

C {C1 C2

C3 C4}

D {D1 D2 D3 D4}

A {A1 A2

A3 A4}

SOUTH

NORTH

B {B1 B2 B3

B4}

C {C1 C2 C3 C4}

D {D1 D2 D3 D4}

A {A1 A2 A3 A4}

NORTH

SOUTH

http://www.arcgis.com/



Table 2: Key characteristics of selected corridors

Site ID Number of circulating

lanes

Inscribed Diameter

[m]

Central Island

[m] Legs

Entry deflection

angle (North-

South/South-North)

Distance from

upstream Roundabout

[m]

Average roundabout spacing [m]

CV

Measured Arterial Volume (7 AM - 7 PM)

North South

South North

La Jolla

RBT1 One 25 16 4 30º/31º –

218.8 0.10 5,531 5,688

RBT2 One 24 15 4 29º/30º 248

RBT3 One 25 14 4 34º/29º 227

RBT4* One 22/25 15/19 4 22º/31º 192

RBT5 One 24 15 3 32º/31º 208

Avon

RBT1** Two 42 24 3 20º/- –

197.0 (-10%)

0.46 (356%)

6,240 6,201

RBT2** Two 48 28 3 -/28º 144

RBT3 Two 46 23 4 23º/30º 170

RBT4 Two 45 28 4 34º/42º 124

RBT5 Two 44 26 4 39º/38º 350

*Oval roundabout which has two values for Inscribed Diameter and Central Island.

**Roundabouts 1 and 2 are tear drop interchange roundabouts.

Notes: CV – Coefficient of Variability (ratio between average roundabout spacing and standard deviation of average roundabout spacing)



4.3. Discussion and Results

In this section, the main results from characteristics speed trajectories (see

section 4.3.1.), acceleration/deceleration profiles (see section 4.3.2.), sub-

segments emissions (see section 4.3.3.) and emissions per unit distance (see

section 4.3.4.) were presented and discussed.

4.3.1. Characteristic Speed Trajectories

Figure 7 (a-d) depicts the speed trajectories through La Jolla and Avon corridors.

The speed profiles across La Jolla corridor were similar as verified in Figure 2.

This mean that vehicles accelerate to free-flow upon exiting each roundabout

(sub-segments B2, C2, D3 and E2) as they approach each mid-block areas (sub-

segments B3, C3, D3 and E3). Nonetheless, vehicles speeds in sub-segments

B3, C3, D3 and E3 are lower than free-flow speeds. Specifically, vehicles reached

a 40km/h of maximum speed on those sub-segments while outside corridor the

speed increased up to 55km/h. It was also verified that, in mid-block sub-

segments, vehicles tend to circulate at nearly constant speeds since adjacent

roundabouts were fairly spaced (see Table 2). Similar results were found in the

opposite through movement (see Figure 7-b).

The results from the Avon corridor analysis (see Figure 7c-d) showed that the

highest average speeds occurred at the RBT4 and RBT5 roundabouts (sub-

segments E2 and E3 in North-South direction, and B2 and B3 in South-North

direction). This was explained by the distance between those roundabouts which

was over than 300 meters and enables vehicles to attain free-flow speeds. On

the remaining roundabouts, the speed values were not so sharper. This was

particularly true between RBT3 and RBT4 whose distance was about 59 meters.

It was also found that the overlapping of RIA was notably expressive between

RBT3 and RBT4. Therefore, vehicles maximum speed at D3 (see Figure 7-c)

and C3 (see Figure 7-d) was half of than the observed at the sub-segments E3

(see Figure 7-c) and B3 (see Figure 7-d).



Figure 7: Speed trajectories by distance travelled for each through movement by corridor: a) La Jolla (North-South); b) La Jolla (South-North); c) Avon (North-

South); d) Avon (South-North).

a)

b)

c)

d)

0

10

20

30

40

50

60

0 1100S

pe

ed

[km

/h]

Distance Travelled [meters]

0

10

20

30

40

50

60

0 1100

Sp

ee

d [km

/h]


0

10

20

30

40

50

60

0 900

Sp

ee

d [km

/h]

Distance travelled [meters]

0

10

20

30

40

50

60

0 900

Sp

ee

d [km

/h]


RBT1 RBT2 RBT3 RBT4 RBT5


1 2 3 4

5





4.3.2. Characteristic Acceleration and deceleration Trajectories

The acceleration and deceleration cycles for vehicles that are crossing the

selected studied corridor are depicted in Figure 8 (a-b) for both through

movements. The higher acceleration rates in La Jolla corridor were recorded over

the circulating area of roundabouts (sub-segments B1, C1, D1 and E1). The

experimental measurements also indicated that the deceleration of vehicles

increased when vehicles approaching each one of the roundabouts (B4, C4, D4

and E4). From the Figure 8 (a-b), and as suspected, the lowest acceleration and

deceleration rates were achieved at the mid-block sub-segments in which

vehicles drove near the cruise speeds. In summary, the acceleration/deceleration

cycles did not significant different between each pair of roundabouts. Several

explanations supported these results. First, the distance between consecutive

roundabouts is similar across the corridor. Second, specific geometric design

aspects of each roundabout, such as inscribed diameter or the number of

circulating lanes, do not vary (see Table 2).

Concerning the Avon corridor, it was noted that, between RBT3 and RBT4

(overlapping of RIAs), vehicles suffered the highest acceleration and deceleration

through the corridor as seen in Figure 8 c-d. In particular, both accelerations and

decelerations range from -3 m/s2 to 3 m/s2. It should be mentioned that vehicles

also tend to decelerate faster before RBT2 (North-South direction) that is placed

in Rain drop interchange area. This means that after B3 sub-segment vehicles

still driving on free-flow speed and decelerate approximately 50 meters before C1

(circulating area of RBT2) as shown in Figure 8-c. The values of deceleration

obtained for vehicles approaching RBT3 and RBT4 were also expressive. Note

that these roundabouts are the fairest spaced along the corridor.



Figure 8: Acceleration/Deceleration cycles by distance travelled for each through movement by corridor: a) La Jolla (North-South); b) La Jolla (South-

North); c) Avon (North-South); d) Avon (South-North).

-3

-2

-1

0

1

2

3

0 1100

Accele

ratio

n [m

/s2]


-3

-2

-1

0

1

2

3

0 1100

Acce

lera

tio

n [m

/s2]


-4

-2

0

2

4

0 900

Acce

lera

tio

n [m

/s2]


a)

b)

c)

d)

-4

-2

0

2

4

0 900

Acce

lera

tio

n [m

/s2]

Travelled distance [meters]




1 2 3 4

5




4.3.3. Segments and Sub-segments Emissions

Figure 9 a-b illustrates that the highest amount of CO2 emissions per vehicle,

about 36% of the total emissions in the corridor, is produced in sub-segments A2,

B2, C2 and D2 (downstream). These sub-segments correspond to about 29% of

the travel distance and travel time across the corridor. This is mainly due to higher

acceleration rates that vehicles experience as they exit each roundabout.

Because of the large distance between the roundabouts, the vehicles often reach

the cruise speed in the midblock section (A3, B3, C3 and D3) as they leave one

roundabout RIA and travel toward the adjacent RIA. Consequently, sub-

segments A4, B4, C4 and D4 (upstream) contribute to more than 27% of total

emissions in the corridor (they include about 29% and 30% of travel distance and

travel time, respectively). These analyses suggest that the deceleration from

cruise speed to the circulating speed is especially relevant for CO2 emissions.

Albeit high, vehicles speeds in mid-block sub-segments (A3, B3, C3 and D3), did

not result in a substantial impact on emissions (overall contribution on total

emission is about 24% for both through movements). This is due to the smooth

speed profiles at those locations. Regarding the assessment of each pair of

roundabouts across the corridor, that is, segments A, B, C and D, the findings

point to small differences in emissions. Specifically, CO2 emissions contributions

of each segment range from 23% to 25% for segments C and A, respectively.

Several explanations support these results. First, the distance between

consecutive roundabouts is similar along the corridor. Second, specific geometric

design aspects of each roundabout, such as inscribed diameter or the number of

circulating lanes, do not vary. See Table 2 for those details.

Sub-segments A2, B2, C2 and D2 also have a major impact on CO2 emissions

across the Avon corridor. According to Figure 9 c-d, vehicles emit about 30% of

CO2 in 25% and 23% of travel distance and travel time respectively on both

directions. Similarly, sub-segments A1, B1, C1 D1 and E1 have a significant

effect on CO2 emissions along that corridor. Their contribution is about 26%, on

23% of travel distance. This is mostly because of the high inscribed diameter at

those roundabouts. The results also show that due to different lengths of each

segment there is a difference in emissions for each pair of roundabouts. From

Figure 9 c-d, and as suspected, the highest amount of overall CO2 emissions is

produced between RBT4 and RBT5 roundabouts (Segments D and A on north-

south and south-north movements, respectively). For that location, a contribution

of more than 43% of total emissions represents 42% of the travel distance is

observed, which is consistent. Along the tear drop interchange, on segment A

(north-south movement), there is no substantial impact of CO2 emissions when

compared with Segments B and D. CO2 emissions only represent 19% of total

emissions of the corridor. Note that vehicles travel 18% of distance on that

segment. Similar results are found on the opposite through movement.



a) b)

c) d)

Figure 9: CO2 Emissions (g) per vehicle for each sub-segment across the roundabouts corridors: a) La Jolla (North-South), b) La Jolla (South-North); c)

Avon (North-South); d) Avon (South-North).

4.3.4. Emissions per unit distance

Figure 10 a-d illustrates the CO2 emissions per unit distance by sub-segment

and regarding the overall corridor. The figure shows that the sub-segments

associated to the downstream (A2, B2, C2 and D2) produce the highest amount

of emissions per kilometer travelled across La Jolla corridor. These results are

verified on both through movements. In particular, above sub-segments generate

an average CO2 emissions per unit distance higher than 27% (210 g/km – north-

south) and 33% (218 g/km – south-north) to the averages values of the corridor

(165 g/km and 164 g/km on north-south and south-north directions, respectively).

It is also observed that the CO2 emissions per unit distance in some circulating

sub-segments are higher than averages values which are recorded in both

movements. On average, CO2 emissions per unit distance at the circulating sub-

segments (A1, B1, C1, D1 and E1) are higher by 3% (170 g/km) and 10% (179

g/km) over the average value which is recorded in north-south and south-north

movements, respectively. It should be emphasized that when vehicles are

crossing each pair of roundabouts, emissions per unit distance follow a similar

trend. Taking as an example the Segment A, it is observed that emissions per

unit distance increase from sub-segment A1 to A2. After that, they decrease on

sub-segment A3 and yet again, they increase on sub-segment A4 (see Figure

10 a-b). A2 and A3 are related to the highest and lowest emissions per unit

0

4

8

12

16

A1

A2

A3

A4

B1

B2

B3

B4

C1

C2

C3

C4

D1

D2

D3

D4

E1

CO

2E

mis

sio

ns p

er

ve

hic

le [g

]

0

4

8

12

16

A1

A2

A3

A4

B1

B2

B3

B4

C1

C2

C3

C4

D1

D2

D3

D4

E1

CO

2E

mis

sio

ns p

er

ve

hic

le [g

]

0

5

10

15

20

25

A1

A2

A3

A4

B1

B2

B3

B4

C1

C2

C3

C4

D1

D2

D3

D4

E1

CO

2E

mis

sio

ns p

er

ve

hic

le [g

]

0

5

10

15

20

25

A1

A2

A3

A4

B1

B2

B3

B4

C1

C2

C3

C4

D1

D2

D3

D4

E1

CO

2E

mis

sio

ns p

er

ve

hic

le [g

]



distance sub-segments, respectively, within Segment A. In some segments, the

differences between above sub-segments exceed the 95 g/km.

The analysis results show that the emissions hotspots in the Avon corridor, 23%

higher than the average corridor value (≈167g/km), are generated in the

circulating areas of the roundabouts (A1, B1, C1 and D1). Concomitantly,

vehicles at any downstream sub-segments generate an amount of CO2 emissions

per unit distance higher than the average value of the corridor with 19%. The

effect of tear drop interchange on the north-south direction (Segment A) is also

relevant (average CO2 emissions per unit distance of 185 g/km in 18% and 17%

of the total distance and time, respectively) as shown in Figure 10-c. This is

mostly due to the higher speeds (cruise speed) in almost Segment A length

between RBT1 and RBT2. Unlike tear drop interchange, Segments C and D on

north-south and south-north, respectively, generate an amount of CO2 emissions

per unit distance 6% and 3% below the average value of the Avon corridor. The

findings for closely spaced roundabouts (RBT3 and RBT4), as illustrated in

Figure 10 c-d, prove that CO2 emissions per kilometer travelled at the

downstream and mid-block sub-segments are much smaller than those which are

observed in equivalent sub-segments Since vehicles have not enough distance

to attain free-flow speeds, lower acceleration rates are observed at the

downstream sub-segment in comparison to the equivalent segments.

a) b)

c) d)

Figure 10: CO2 Emissions (g/km) per unit distance for each sub-segment across the roundabouts corridors and overall corridor: a) La Jolla (North-South),

b) La Jolla (South-North); c) Avon (North-South); d) Avon (South-North).

165

0

60

120

180

240

A1

A2

A3

A4

B1

B2

B3

B4

C1

C2

C3

C4

D1

D2

D3

D4

E1

CO

2E

mis

sio

ns p

er

un

it

dis

tan

ce

[g

/km

] 164

0

60

120

180

240

A1

A2

A3

A4

B1

B2

B3

B4

C1

C2

C3

C4

D1

D2

D3

D4

E1

CO

2E

mis

sio

ns p

er

un

it

dis

tan

ce

[g

/km

]

166

0

65

130

195

260

A1

A2

A3

A4

B1

B2

B3

B4

C1

C2

C3

C4

D1

D2

D3

D4

E1

CO

2E

mis

sio

ns p

er

un

ti

dis

tan

ce

[g

/km

]

166

0

65

130

195

260

A1

A2

A3

A4

B1

B2

B3

B4

C1

C2

C3

C4

D1

D2

D3

D4

E1

CO

2E

mis

sio

ns p

er

un

it

dis

tan

ce

[g

/km

]



4.4. Conclusions

It was found that the emissions distributions along the corridor with fairly spaced

roundabouts and enough spacing to attain cruise speed over the mid-block was

similar in each pair of adjacent roundabouts. In such cases, the downstream sub-

segments were identified as the emissions hotspots (overall contribution on total

emissions exceeded the 35 %) both in absolute terms and per unit distance.

Considering the corridor with evenly spaced roundabouts, the emissions hot

spots per unit distance, about 23% higher than the average corridor value, were

located at the circulating areas sub-segments. This was particularly true for

closely spaced roundabouts (~100 meters of spacing) in which vehicles

decelerated after the roundabout exit section (at the downstream sub-segment).



5. Papers in Publication

During the first year of Doctoral grant the candidate was co-author of several papers

5.1. Papers published (or accepted for publication) in scientific journals

1. Salamati, K., M.C. Coelho, P.J. Fernandes, N.M. Rouphail, H.C. Frey, and J. Bandeira. Emission Estimation at Multilane Roundabouts: Effect of Movement and Approach Lane. Transportation Research Record: Journal of the Transportation Research Board, 2014. 2266 (-1): p. 12-21. DOI: http://dx.doi.org/10.3141/2389-02

2. Fontes, T., Fernandes, P., Rodrigues, H., Bandeira, J. M., Pereira, S.R., Coelho, M.C., Khattak, A.J. Are eco-lanes a sustainable option to reducing emissions in a medium-sized European city?, Journal of Transportation Research A: Policy and Practice, Volume 63, May 2014, Pages 93–106. DOI: http://dx.doi.org/10.1016/j.tra.2014.03.002

3. Jorge Bandeira, Paulo Fernandes, Tânia Fontes, Sérgio Ramos Pereira, Asad J. Khattak & Margarida C. Coelho, Assessment of Eco Traffic Assignment Strategies in an Urban Corridor, accepted to be published in the Special Issue for Transportation Research – Part C on Advanced Road Traffic Control, in press.

4. Paulo Fernandes, Jorge Bandeira, Tânia Fontes, Sérgio Pereira, Bastian J. Schroeder, Nagui M. Rouphail, Margarida C. Coelho, Traffic restriction policies in an urban avenue: A methodological overview for a trade-off analysis of traffic and emission impacts using microsimulation, accepted to be published in the International Journal of Sustainable Transportation, in press. (DOI:10.1080/15568318.2014.885622)

5. Bandeira, J., Carvalho, D., Khattak, A., Rouphail, N., Fontes, T., Fernandes, P., Pereira, S., Coelho, M. C. Empirical assessment of route choice impact on emissions over different congestion scenarios, accepted to be published in the International Journal of Sustainable Transportation, in press. (DOI: 10.1080/15568318.2014.901447)

5.2. Book Chapters

1. Bandeira, J.M., Pereira, S.R., Fontes, T., Fernandes, P., Khattak, A., Coelho, M.C. (2014). An “Eco-traffic” assignment tool In Computer-based Modelling and Optimization in Transportation. Advances in Intelligent Systems and Computing, J. F. de Sousa and R. Rossi (eds.), Springer International Publishing Switzerland, Vol. 262, pp. 41-56. (ISBN: 978-3-319-04629-7, DOI: http://dx.doi.org/10.1007/978-3-319-04630-3_4).

http://dx.doi.org/10.3141/2389-02

http://dx.doi.org/10.1016/j.tra.2014.03.002

http://dx.doi.org/10.1007/978-3-319-04630-3_4



5.3. Conference proceedings

1. Fernandes, P., Pereira, S.R., Bandeira, J.M., Neto, L., Coelho, Vasconcelos, L., Silva, A.B., Seco, A.B, Coelho, M.C. Driving around turbo-roundabouts: are there advantages over conventional roundabouts?. Research Day, University of Aveiro, June 3, 2014.

2. Fontes, T., Lemos, A., Fernandes P., Pereira, S.R., Bandeira, J.M., Coelho, M.C. Emissions impact of road traffic incidents using Advanced Traveller Information Systems in a regional scale, Accepted to be presented in the 17th meeting of the EURO Working Group on Transportation - EWGT 2014, to be held in Seville, July 2014.

3. Bandeira, J.M., Fontes, T., Pereira, S.R., Fernandes, P., Khattak, A.J., Coelho, M.C. Assessing the importance of vehicle type for the implementation of eco-routing systems, Accepted to be presented in the 17th meeting of the EURO Working Group on Transportation - EWGT 2014, to be held in Seville, July 2014.

4. T. Fontes, J. Bandeira, S.R. Pereira, P. Fernandes, and M.C. Coelho, Urban road transportation planning: the trade-off between noise and atmospheric emissions, 2014 Transportation Land Use and Air Quality Conference, Charlotte, March 2014.

5. Daniela Dias, Jorge H. Amorim, Carlos Borrego, Tânia Fontes, Paulo Fernandes, Sérgio R. Pereira, Jorge M. Bandeira, Margarida C. Coelho, Oxana Tchepel. Impact of Road Transport on Urban Air Quality: GIS and GPS as a Support for a Modelling Framework, Conference GIS Ostrava 2014 – Geoinformatics for Intelligent Transportation, Ostrava, January 2014.

URL: http://gis.vsb.cz/GIS_Ostrava/GIS_Ova_2014/proceedings/papers/gis2014526a81620de5b.pdf

6. Luís Vasconcelos, Ana Bastos Silva, Álvaro Seco, Paulo J. Fernandes, Margarida C. Coelho, Turboroundabouts: A Multi-criteria Assessment on Intersection Capacity, Safety and Emissions, 93rd Transportation Research Board Annual Meeting, Washington D.C., January 2014.

URL: http://pressamp.trb.org/aminteractiveprogram/EventDetails.aspx?ID=29216

7. T. Fontes, P. Fernandes, H. Rodrigues, J. Bandeira, S.R. Pereira, M.C. Coelho, Existência de painéis de mensagem variável para gestão de incidentes de tráfego urbano: Impacte nas emissões de poluentes. 10.ª Conferência Nacional do Ambiente, University of Aveiro, November 2013.

URL: http://10cna.web.ua.pt/

8. Bandeira, J., Pereira, S.R., Fontes, T., Fernandes, P., Khattak, A.J., Coelho, M.C., An “Eco-traffic” assignment tool, 16th Euro Working Group on Transportation, Porto, September 2013. Paper 4268.

URL: http://www.ewgt2013.com/1%2009%202013_Programa%20formatado.pdf

http://gis.vsb.cz/GIS_Ostrava/GIS_Ova_2014/proceedings/papers/gis2014526a81620de5b.pdf

http://gis.vsb.cz/GIS_Ostrava/GIS_Ova_2014/proceedings/papers/gis2014526a81620de5b.pdf

http://pressamp.trb.org/aminteractiveprogram/EventDetails.aspx?ID=29216

http://10cna.web.ua.pt/

http://www.ewgt2013.com/1%2009%202013_Programa%20formatado.pdf



6. Future Tasks

The following tasks are expected to be developed during the second year of this doctoral research (1st of June 2014 – 31st of May 2015).

Compare the results obtained in US corridors with Portuguese corridors with similar design features (equally spacing vs. unequally spacing; traffic flows);

To explore pedestrian crosswalk locations effect in some of these corridors on traffic performance and emissions using a microsimulation approach as well as a multi-objective optimization;

To developed regression models to predict the relative occurrence of speed profiles (no stop, stop once and several stops) in corridors based on traffic flows data (corridors);

To compare performance and emissions impacts of corridors and traffic signals in corridors using a microsimulation approach.

6.1. Papers in preparation

1. Fernandes, P., Fontes, T., Pereira, S.R., Coelho, M.C., Rouphail, N.M. Multi-criteria assessment of crosswalks located in urban corridors of roundabouts. To be submitted to the Transportation Research Part A: Journal of Transportation Research A: Policy and Practice.

2. Fernandes, P., Pereira, S.R., Bandeira, J.M., Coelho, Vasconcelos, L., Silva, A.B., Seco, A.B, Coelho, M.C. Empirical Assessment of Turbo-Roundabout Operations on Intersection Capacity and Emissions. To be submitted (before 31st May, 2014) to the 94th Transportation Research Board Annual Meeting. January 11-15, 2015.

3. Fernandes, P., Neves, M., Fontes, T., Pereira, S.R., Bandeira, J.M., Coelho, M.C. Comparison between roundabouts and traffic signals in series in a corridor on traffic performance and emissions: a case study. To be submitted to the 94th Transportation Research Board Annual Meeting. January 11-15, 2015.

4. Fernandes, P., Salamati, K., Coelho, M.C., Rouphail, N.M. Identification of the Emission Hotspots in Roundabouts Corridors. To be submitted to the Transportation Research part D – Transport and Environment.

Acknowledgments

My research work is supported by a PhD scholarship (SFRH/BD/87402/2012) of

the Portuguese Science and Technology Foundation (FCT) and the R&D project

PTDC/SEN-TRA/122114/2010 (AROUND – Improving Capacity and Emission

Models of Roundabouts).



References

1. Rodegerdts L, Bansen J, Tiesler C, Knudsen J, Myers E, Johnson M, et al. Roundabouts: An Informational Guide (Second Edition). Washington, DC: National Cooperative Highway Research Program (NCHRP) Issue 672; 2010.

2. Wolf P. Eugene Henard and the Beginning of Urbanism in Paris, 1900–1914, International Federation for Housing and Planning, The Hague, 1969, cited by Ben Hamilton-Baillie & Phil Jones, Improving traffic behaviour and safety through urban design. Proceedings of ICE – Civil Engineering; 2005, 180 (5). p. 39-47.

3. Thieken S. Implementing Multi-Lane Roundabouts in Urban Areas. Proceedings of the Transportation Land Use, Planning, and Air Quality; 2007, July 9-11; Orlando, Florida. Reston, VA: American Society of Civil Engineers; 2008. p. 41-51.

4. Kittelson & Associates Inc. Modern Roundabouts http://roundabout.kittelson.com/Roundabouts/List/US, Acessed May 20142014 [cited 2014 26 April].

5. Guichet B. French Roundabouts Today. Proceedings of 2nd Transportation Research Board National Roundabout Conference; 2008 May 18-21; Kansas City, Missouri, United States.2008.

6. Roundabouts USA. History of the Modern Roundabout. http://www.roundaboutsusa.com/history.html, Acessed May 20142014 [cited 2014 26 April].

7. Rodegerdts L, Blogg M, Wemple E, Myers E, Kyte M, Dixon M, et al. Roundabouts in the United States. Washington, DC: National Cooperative Highway Research Program (NCHRP) Issue 572; 2007.

8. TRB. Highway Capacity Manual, Vol. 3: Interrupted Flow. Washington, DC: Transportation Research Board; 2010. Chapter 21, Roundabouts; p. 21-44.

9. Isebrands H., Hallmark S, Fitzsimmons E, Stroda J. Toolbox to Evaluate the Impacts of Roundabouts on a Corridor or Roadway Network. Minnesota Department of Transportation, Research Services Section, Publication MN/RC 2008-24: 2008.

10. Myhrberg S. Saving fuel and environment with intelligent speed adaptation.; Proceedings of the 15th World Congress on Intelligent Transport Systems and ITS America's 2008 Annual Meeting; 2008November 16-20; New York, NK, United States.

11. Pierre G, Ehrlich J. Impact of intelligent speed adaptation systems on fuel consumption and driver behavior. Proceedings of the 15th World Congress on Intelligent Transport Systems and ITS America's 2008 Annual Meeting; 2008November 16-20; New York, NK, United States.

12. Xia H, Boriboonsomsin K, Barth M. Dynamic Eco-Driving for Signalized Arterial Corridors and Its Indirect Network-Wide Energy/Emissions Benefits. Journal of Intelligent Transportation Systems. 2012 2013/01/01;17(1):31-41.

13. Barth M, Boriboonsomsin K. ECO-ITS: Intelligent Transportation System Applications to Improve Environmental Performance. Washington, DC: Federal Highway Administration, U.S. Department of Transportation, FHWA-JPO-12-042; : 2012.

14. Hallmark S, Fitzsimmons E, Isebrands H, Giese K. Roundabouts in Signalized Corridors. Transportation Research Record: Journal of the Transportation Research Board. 2010 12/01/;2182(-1):139-47.

15. Hallmark S, Wang B, Mudgal A, Isebrands H. On-Road Evaluation of Emission Impacts of Roundabouts. Transportation Research Record: Journal of the Transportation Research Board. 2011 12/01/;2265(-1):226-33.

16. Akcelik and Associates. Sidra Intersection User Guide (for Version 6). Melbourne: Australia; 2013, Akcelik and Associates Pty Ltd.

17. Krogscheepers J, Watters M. Roundabouts along Rural Arterials in South Africa. Transportation Research Board 93nd Annual Meeting, Washington, DC. 2014.

18. Yap YH, Gibson HM, Waterson BJ. An International Review of Roundabout Capacity Modelling. Transport Reviews. 2013 2013/09/01;33(5):593-616.

http://roundabout.kittelson.com/Roundabouts/List/US

http://www.roundaboutsusa.com/history.html



19. Guichet B, editor Roundabouts in France: Development, safety, design, and capacity. 1997; Proceedings of the third international symposium on intersections without traffic signals (pp.100-105); 1997, July 21-23, Portland, Oregon. USA.

20. Karlaftis MG, Vlahogianni EI. Statistical methods versus neural networks in transportation research: Differences, similarities and some insights. Transportation Research Part C: Emerging Technologies. 2011 6//;19(3):387-99.

21. Troutbeck RJ. Evaluating the performance of a roundabout. Vermont South: Australian Road Research Board (ARRB); 1989: Report No.: SR45.

22. Wu N. A universal procedure for capacity determination at unsignalized (priority-controlled) intersections. Transportation Research Part B: Methodological. 2001 7//;35(6):593-623.

23. Chevallier E, Leclercq L. Microscopic Dual-Regime Model for Single-Lane Roundabouts. Journal of Transportation Engineering. 2009;135(6):386-94.

24. Bastos Silva AMC, Vasconcelos L. Microsimulation applied to roundabout performance analysis: the effect of pedestrian crossings. Presented at: European Transport Conference; 2009, Oct 5; Noordwijkerhout: Netherlands2009.

25. Vasconcelos L, Bastos Silva AMC, Seco AM, editors. A sensitivity analysis of Cowan’s M3 capacity model applied to roundabouts. Paper 718-020, Proceedings of the IASTED International Conference on Modelling, Identification, and Control - MIC; 2011 Feb 14-16; Innsbruck, Austria.

26. Sisiopiku V, Oh H. Evaluation of Roundabout Performance Using SIDRA. Journal of Transportation Engineering. 2001;127(2):143-50.

27. Meredith J, Rakha H, editors. Do Roundabouts Work? Evaluation for Uniform Approach Demands2012; Paper 12-0804, Proceedings of the 91st Transportation Research Board; 2012 January 22-26, Washington, DC, USA.

28. Zheng D, Qin X, editors. Negotiation-Based Conflict Exposure Methodology in Roundabout Crash Pattern Analysis. Paper 10-1616, Proceedings of the 89th Annual Meeting of Transportation Research Board; 2010; January 10-14; Washington DC, United States.

29. Katsis P, Ntziachristos L, G. M. Description of new elements in COPERT 4 v10.0. Thessaloniki (Greece): EMISIA SA;Dec. 2012 71 p. Report No: 12.RE.012.V1 2012.

30. USEPA. Methodology for developing modal emission rates for EPA’s multi-scale motor vehicle & equipment emission system. Prepared by North Carolina State University for USEPA, Ann Arbor, MI, August, EPA420. 2002.

31. Borrego C, Tchepel O, Costa AM, Amorim JH, Miranda AI. Emission and dispersion modelling of Lisbon air quality at local scale. Atmospheric Environment. 2003 12//;37(37):5197-205.

32. Coelho MC, Farias TL, Rouphail NM. A Numerical Tool for Estimating Pollutant Emissions and Vehicles Performance in Traffic Interruptions on Urban Corridors. International Journal of Sustainable Transportation. 2009 2009/06/23;3(4):246-62.

33. Ahn K, Rakha H, Trani A, Van Aerde M. Estimating Vehicle Fuel Consumption and Emissions based on Instantaneous Speed and Acceleration Levels. Journal of Transportation Engineering. 2002;128(2):182-90.

34. Frey HC, Unal A, Rouphail NM, Colyar JD. On-Road Measurement of Vehicle Tailpipe Emissions Using a Portable Instrument. Journal of the Air & Waste Management Association. 2003 2003/08/01;53(8):992-1002.

35. EPA. Motor Vehicle Emission Simulator (MOVES) User Guide for MOVES2010b . Washington, DC: United States Environmental Protection Agency (EPA); 2012, June 2012; p. 202. Report No: EPA-420-B-12-001b 2012.

36. Coelho MC, Farias TL, Rouphail NM. A methodology for modelling and measuring traffic and emission performance of speed control traffic signals. Atmospheric Environment. 2005 4//;39(13):2367-76.

37. Coelho M, Farias T, Rouphail N. Measuring and Modeling Emission Effects for Toll Facilities. Transportation Research Record: Journal of the Transportation Research Board. 2005 01/01/;1941(-1):136-44.



38. Frey HC, Zhang K, Rouphail NM. Fuel Use and Emissions Comparisons for Alternative Routes, Time of Day, Road Grade, and Vehicles Based on In-Use Measurements. Environmental Science & Technology. 2008 2008/04/01;42(7):2483-9.

39. Coelho MC, Frey HC, Rouphail NM, Zhai H, Pelkmans L. Assessing methods for comparing emissions from gasoline and diesel light-duty vehicles based on microscale measurements. Transportation Research Part D: Transport and Environment. 2009;14(2):91-9.

40. Zhai H, Frey HC, Rouphail NM. A Vehicle-Specific Power Approach to Speed- and Facility-Specific Emissions Estimates for Diesel Transit Buses. Environmental Science & Technology. 2008 2008/11/01;42(21):7985-91.

41. Coelho MC, Farias TL, Rouphail NM. Effect of roundabout operations on pollutant emissions. Transportation Research Part D: Transport and Environment. 2006 9//;11(5):333-43.

42. Mudgal A, Hallmark S, Carriquiry A, Gkritza K. Driving behavior at a roundabout: A hierarchical Bayesian regression analysis. Transportation Research Part D: Transport and Environment. 2014 1//;26(0):20-6.

43. Anya AR, Rouphail NM, Frey HC, Liu B, editors. Method and Case Study for Quantifying Local Emissions Impacts of a Transportation Improvement Project Involving Road Re-Alignment and Conversion to a Multi-Lane Roundabout. Paper 13-5243, Proceedings of 92nd Annual Meeting of the Transportation Research Board; 2013 January 13-17; Washington, DC, United States.

44. Salamati K, Coelho MC, Fernandes PJ, Rouphail NM, Frey HC, Bandeira J. Emission Estimation at Multilane Roundabouts: Effect of Movement and Approach Lane. Transportation Research Record: Journal of the Transportation Research Board. 2013;2389(-1):12-21.

45. Vasconcelos L, Silva AB, Seco AM, Fernandes P, Coelho MC. Turbo-Roundabouts: A multi-criteria assessment on intersection capacity, safety and emissions. Accepted for presentation in the 93rd Annual Conference of the Transportation Research Board; 2014 January 12-16; Washington, D.C., United States.

46. Várhelyi A. The effects of small roundabouts on emissions and fuel consumption: a case study. Transportation Research Part D: Transport and Environment. 2002 1//;7(1):65-71.

47. Rakha HA, Jackson M, editors. Are Roundabout Environmentally Friendly? An Evaluation for Uniform Approach Demands. Paper 12-0789, Proceedings of 91st Annual Meeting of the Transportation Research Board; 2012 January 22-26; Washington, DC, United States.

48. Mandavilli S, Rys MJ, Russell ER. Environmental impact of modern roundabouts. International Journal of Industrial Ergonomics. 2008;38(2):135-42.

49. Ahn K, Kronprasert N, Rakha H. Energy and Environmental Assessment of High-Speed Roundabouts. Transportation Research Record: Journal of the Transportation Research Board. 2009;2123(-1):54-65.

50. Chamberlin R, Swanson B, Talbot E, editors. Analysis of MOVES and CMEM for Evaluating the Emissions Impact of an Intersection Control Change. Paper 11-0673, Proceedings of 90th Annual Meeting of the Transportation Research Board; 2011 January 23-27; Washington, DC, United States.

51. Rakha H, Wang Z, Ong BT, editors. Roundabout versus Traffic Signal Control: A Comparative Analysis. Paper 13-4422, Proceedings of 92nd Annual Meeting of the Transportation Research Board; 2013 January 13-17; Washington, DC, United States.

52. Fellendorf M, Vortisch P. Microscopic traffic flow simulator vissim. In Barceló, J. ed. Fundamentals of traffic simulation, New York, NY: Springer; 2010; p. 63-93.

53. Song G, Yu L, Zhang Y. Applicability of Traffic Microsimulation Models in Vehicle Emissions Estimates. Transportation Research Record: Journal of the Transportation Research Board. 2012 12/01/;2270(-1):132-41.

54. Mandavilli S, McCartt AT, Retting RA. Crash Patterns and Potential Engineering Countermeasures at Maryland Roundabouts. Traffic Injury Prevention. 2009 2009/02/27;10(1):44-50.

55. Maycock G, Hall RD. Accidents at 4-Arm Roundabouts. Transport and Road Research Laboratory. Crowthorne, Berkshire, United Kingdom; 1984 January. 60 p. Report No.: TLR 1120.



56. Arndt OK. Relationship Between Roundabout Geometry and Accident Rates [M.E. Thesis]. Brisbane, Queensland: Queensland University of Technology; 1994.

57. Brüde U, Larsson J. What roundabout design provides the highest possible safety? Nordic Road and Transport Research,No. 2. Swedish National Road and Transport Research Institute; 2000.

58. Gettman D, Head, L. Surrogate Safety Measures from Traffic Simulation Models. Transportation Research Record: Journal of the Transportation Research Board 2003;1840:104-15.

59. Tarko AP. Use of crash surrogates and exceedance statistics to estimate road safety. Accident Analysis & Prevention. 2012 3//;45(0):230-40.

60. Huang F, Liu P, Yu H, Wang W. Identifying if VISSIM simulation model and SSAM provide reasonable estimates for field measured traffic conflicts at signalized intersections. Accident Analysis & Prevention. 2013 1//;50(0):1014-24.

61. Dijkstra A, Marchesini P, Bijleveld F, Kars V, Drolenga H, van Maarseveen M. Do Calculated Conflicts in Microsimulation Model Predict Number of Crashes? Transportation Research Record: Journal of the Transportation Research Board. 2010 12/01/;2147(-1):105-12.

62. Chen Y, Persuad B, Lyon C, editors. Effect of Speed on Roundabout Safety Performance: Implications for Use of Speed as Surrogate Measure. Paper 11-2846, Proceedings of 90th Annual Meeting of the Transportation Research Board; 2011; January 22-26; Washington, DC, United States.

63. Pu L, Joshi R. Surrogate Safety Assessment Model (SSAM) - Software User Manual. Federal Highway Administration, U.S. Department of Transportation McLean, VA, United States; 2008 May. 100 p. Report No.: FHWA-HRT-08-050: 2008.

64. Al-Ghandour M, Schroeder B, Williams B, Rasdorf W. Development and Validation of Conflict Models for Single-Lane Roundabout Slip Lanes from Microsimulation. Transportation Research Record: Journal of the Transportation Research Board. 2011;2237(1):92-101.

65. Vasconcelos L, Neto L, Seco AM, Silva AB. Validation of the SSAM technique for the assessment of intersections safety. Paper 14-0229, Proceedings of 93rd Annual Meeting of the Transportation Research Board; 2014 January 12-16; Washington, DC, United States.2014.

66. Cunto F, Saccomanno FF. Calibration and validation of simulated vehicle safety performance at signalized intersections. Accident Analysis & Prevention. 2008 5//;40(3):1171-9.

67. Duong DD, FF S, BR H, editors. Multi-objective calibration/validation of a microscopic simulation platform and the effect of goodness-of-fit form on calibration results2011; Paper 12-0465, Proceedings of the 91st Transportation Research Board, January 22-26, Washington, DC, USA.

68. HCM. Highway Capacity Manual. Transportation Research Board, Washington, DC.2010.

69. Abdul Aziz HM, Ukkusuri S, editors. Environmental Objectives within a Dynamic Traffic Assignment Framework: A Step towards Green Transportation. Paper 11-3893, Proceedings of the 90th Annual Meeting of Transportation Research Board; 2011; January 23-27; Washington DC, United States.

70. Fontes T, Fernandes P, Rodrigues H, Bandeira JM, Pereira SR, Khattak AJ, et al. Are HOV/eco-lanes a sustainable option to reducing emissions in a medium-sized European city? Transportation Research Part A: Policy and Practice. 2014 5//;63(0):93-106.

71. PTV. Vissim 5.30-05 user manual. Planung Transport Verkehr AG, Karlsruhe, Germany. 2011.

72. Bared J, Afshar A. Using Simulation to Plan Capacity Models by Lane for Two- and Three-Lane Roundabouts. Transportation Research Record: Journal of the Transportation Research Board. 2009 12/01/;2096(-1):8-15.

73. Schroeder B, Rouphail N, Salamati K, Bugg Z. Effect of Pedestrian Impedance on Vehicular Capacity at Multilane Roundabouts with Consideration of Crossing Treatments. Transportation Research Record: Journal of the Transportation Research Board. 2012 12/01/;2312(-1):14-24.

74. Raff MS, Hart JW. A volume warrant to urban stop signs. Eno Foundation for Highway Traffic Control, Saugatuck, Conn.1950.



75. Qstarz. Qstarz BT-Q1000XT Travel Recorder XT – User Guide. Qstarz International Co., Ltd. Taiwan, R.O.C.2012.

76. Davis, Instruments. 8226B CarChip Pro User Manual. Vernon Hills, IL2012.

77. Kolak OI, Feyzioglu O, Birbil SI, Noyan N, Yalcindag S. Using emission functions in modeling environmentally sustainable traffic assignment policies. Journal of Industrial and Management Optimization. 2013;9(2).

78. USEPA. Methodology for Developing Modal Emission Rates for EPA’s Multi-Scale Motor Vehicle & Equipment Emission System 2002 Publication EPA420-R-02-027. Prepared by North Carolina State University for US Environmental Protection. Agency Ann Arbor.

79. Lee C. Passenger Car Equivalents for Heavy Vehicles at Roundabouts: Estimation and Application to Capacity Prediction. Accepted for presentation in the 93rd Annual Conference of the Transportation Research Board; 2014 January 12-16; Washington, D.C., United States.

80. Gettman D, Pu L, Sayed T, Shelby S. Surrogate Safety Assessment Model and Validation: Final Report. Federal Highway Administration, Publication FHWA-HRT-08-0512008.

81. Gettman D, Head L. Surrogate Safety Measures from Traffic Simulation Models. Transportation Research Record: Journal of the Transportation Research Board. 2003 01/01/;1840(-1):104-15.

82. Salamati K, Schroeder B, Rouphail N, Cunningham C, Long R, Barlow J. Development and Implementation of Conflict-Based Assessment of Pedestrian Safety to Evaluate Accessibility of Complex Intersections. Transportation Research Record: Journal of the Transportation Research Board. 2011 12/01/;2264(-1):148-55.

83. Hourdakis J, Michalopoulos P, Kottommannil J. Practical Procedure for Calibrating Microscopic Traffic Simulation Models. Transportation Research Record: Journal of the Transportation Research Board. 2003 01/01/;1852(-1):130-9.

84. Spiegelman C, Park ES, Rilett LR. Transportation Statistics and Microsimulation: Taylor & Francis; 2010.

85. Park B, Qi H, Council VTR, Transportation VDo, Administration USFH. Development and evaluation of a calibration and validation procedure for microscopic simulation models: Virginia Transportation Research Council; 2004.

86. Toledo T, Ben-Akiva M, Darda D, Jha M, Koutsopoulos H. Calibration of Microscopic Traffic Simulation Models with Aggregate Data. Transportation Research Record: Journal of the Transportation Research Board. 2004 01/01/;1876(-1):10-9.

87. Ciuffo B, Punzo V, Torrieri V. Comparison of Simulation-Based and Model-Based Calibrations of Traffic-Flow Microsimulation Models. Transportation Research Record: Journal of the Transportation Research Board. 2008 12/01/;2088(-1):36-44.

88. Punzo V, Ciuffo B. How Parameters of Microscopic Traffic Flow Models Relate to Traffic Dynamics in Simulation. Transportation Research Record: Journal of the Transportation Research Board. 2009 12/01/;2124(-1):249-56.

89. Vasconcelos A, Seco AJM, Silva AB, editors. Hybrid calibration of microscopic simulation models.2013; Paper 4227, Proceedings of the 16th Meeting of the EURO Working Group on Transportation; 2013, 4-6 September; Porto, Portugal.

90. Cicu F, Illotta PF, Bared J, Isebrands H, editors. VISSIM Calibration of Roundabout Traffic Performance2011; Paper 11-3637, Proceedings of the 91st Transportation Research Board, January 22-26, Washington, DC, USA.

91. PTV. Vissim 5.30-05 user manual. Planung Transport Verkehr AG, Karlsruhe, Germany.2011.

92. Paz A, Molano V, Khan A. Calibration of Microscopic Traffic Flow Models Considering all Parameters Simultaneously. Paper 14-2283, Proceedings of 93rd Annual Meeting of the Transportation Research Board; 2014 January 12-16; Washington, DC, United States.2014.

93. Dowling R, Skabadonis A, Alexiadis V. Traffic analysis toolbox, volume III: Guidelines for applying traffic microsimulation software. Federal Highway Administration, FHWA-HRT-04-040, U.S. Department of Transportation, Washington, DC2004.

94. Hale D. “How many netsim runs are enough?”. McTrans, 11, 1-9. 1997.



95. Cambridge Systematics Inc. Travel Model Validation and Reasonableness Checking Manual, Second Edition Federal Highway Administration, FHWA-HEP-10-042, U.S. Department of Transportation, Washington, DC2010.

96. Sheskin DJ. Handbook of Parametric and Nonparametric Statistical Procedures. Fifth Edition. Chapman & Hall/CRC.2011.

97. Horn SD. Goodness-of-fit tests for discrete data: a review and an application to a health impairment scale. Biometrics. 1977;33(1):237-47.

98. Zitzler E, Deb K, Thiele L. Comparison of Multiobjective Evolutionary Algorithms: Empirical Results. Evol Comput. 2000;8(2):173-95.

99. Yin Y. Genetic-Algorithms-Based Approach for Bilevel Programming Models. Journal of Transportation Engineering. 2000;126(2):115-20.

100. Ghosh A, Dehuri S. Evolutionary Algorithms for Multi-Criterion Optimization: A Survey International Journal of Computing & Information Sciences. 2004;2(1):38-57.

101. Schwefels H-PP. Evolution and optimum seeking: The Sixth Generation. New York, NY: Wiley; 1993.

102. Goldberg DE. Genetic Algorithms in Search, Optimization, and Machine Learning. Indianapolis, Indiana: Addison-Wesley Professional; 1989.

103. Holland JH. Adaptation in Natural and Artificial Systems: An Introductory Analysis with Applications to Biology, Control and Artificial Intelligence. Cambridge, MA: MIT Press; 1992.

104. Schwefel HP. Numerical Optimization of Computer Models. New York, NY: Wiley; 1989.

105. Fogel LJ, Owens AJ, Walsh MJ. Artificial Intelligence through Simulated Evolution. New York, NY: Wiley; 1966.

106. Chen A, Kim J, Lee S, Kim Y. Stochastic multi-objective models for network design problem. Expert Systems with Applications. 2010 3//;37(2):1608-19.

107. Kesur KB. 17 - Multiobjective Optimization of Delay and Stops in Traffic Signal Networks. In: Yang X-S, Gandomi AH, Talatahari S, Alavi AH, editors. Metaheuristics in Water, Geotechnical and Transport Engineering. Oxford: Elsevier; 2013. p. 385-416.

108. Wanging Ma, Yue Liu, Larry Head, editors. Optimization of Pedestrian Phase Patterns at Signalized Intersections: A Multi-objective Approach. Paper 13-4364, Proceedings of 92nd Annual Meeting of the Transportation Research Board; 2013 January 13-17; Washington, DC, United States.

109. Day C, Bullock D. Computational Efficiency of Alternative Algorithms for Arterial Offset Optimization. Transportation Research Record: Journal of the Transportation Research Board. 2011 12/01/;2259(-1):37-47.

110. Ma W, Liu Y, Xie H, Yang X. Multiobjective Optimization of Signal Timings for Two-Stage, Midblock Pedestrian Crosswalk. Transportation Research Record: Journal of the Transportation Research Board. 2011 12/01/;2264(-1):34-43.

111. Hu H, Gao Y, Yang X, editors. Multi-objective Optimization Method of Fixed-Time Signal Control of Isolated Intersections. Computational and Information Sciences (ICCIS), 2010 International Conference on; 2010 17-19 Dec. 2010.

112. Shafahi Y, Khani A. A practical model for transfer optimization in a transit network: Model formulations and solutions. Transportation Research Part A: Policy and Practice. 2010 7//;44(6):377-89.

113. Kumar AJS, Arunadevi J, Mohan V. Intelligent Transport Route Planning Using Genetic Algorithms in Path Computation Algorithms European Journal of Scientific Research 1999;25(3):463-8.

114. Sohn K. Multi-objective optimization of a road diet network design. Transportation Research Part A: Policy and Practice. 2011 7//;45(6):499-511.

115. Fan W, Machemehl R. Bi-Level Optimization Model for Public Transportation Network Redesign Problem. Transportation Research Record: Journal of the Transportation Research Board. 2011 12/01/;2263(-1):151-62.

116. Mathew T, Sharma S. Capacity Expansion Problem for Large Urban Transportation Networks. Journal of Transportation Engineering. 2009;135(7):406-15.



117. Sharma S, Ukkusuri S, Mathew T. Pareto Optimal Multiobjective Optimization for Robust Transportation Network Design Problem. Transportation Research Record: Journal of the Transportation Research Board. 2009 12/01/;2090(-1):95-104.

118. Duell M, Gardner LM, Waller ST, editors. Multi-Objective Traffic Network Design Accounting for Plug-in Electric Vehicle Energy Consumption. Paper 13-5203, Proceedings of 92nd Annual Meeting of the Transportation Research Board; 2013 January 13-17; Washington, DC, United States.

119. Mesbah M, Thompson R, Moridpour S. Bilevel Optimization Approach to Design of Network of Bike Lanes. Transportation Research Record: Journal of the Transportation Research Board. 2012 12/01/;2284(-1):21-8.

120. Liao X, Li Q, Yang X, Zhang W, Li W. Multiobjective optimization for crash safety design of vehicles using stepwise regression model. Structural and Multidisciplinary Optimization. 2008 2008/06/01;35(6):561-9. English.

121. Schaffer JD. Multiple Objective Optimization with Vector Evaluated Genetic Algorithms. Proceedings of the 1st International Conference on Genetic Algorithms. 657079: L. Erlbaum Associates Inc.; 1985. p. 93-100.

122. Fonseca C, Fleming P, editors. Genetic Algorithms for Multiobjective Optimization: Formulation, Discussion and Generalization.; Proceedings of the 5th International Conference on Genetic Algorithms; 1993 June, Urbana-Champaign, IL. San Francisco, CA, USA: Morgan Kaufmann Publishers Inc.; 1993. P. 416-423.

123. Hajela P, Lin CY. Genetic search strategies in multicriterion optimal design. Structural optimization. 1992 1992/06/01;4(2):99-107. English.

124. Murata T, Ishibuchi H, editors. MOGA: multi-objective genetic algorithms. Evolutionary Computation, 1995, IEEE International Conference on; 1995 Nov. 29 1995-Dec. 1 1995.

125. Horn J, Nafpliotis N, Goldberg DE, editors. A niched Pareto genetic algorithm for multiobjective optimization. Evolutionary Computation, 1994 IEEE World Congress on Computational Intelligence, Proceedings of the First IEEE Conference on; 1994 27-29 Jun 1994.

126. Srinivas N, Deb K. Muiltiobjective optimization using nondominated sorting in genetic algorithms. Evol Comput. 1994;2(3):221-48.

127. Deb K, Pratap A, Agarwal S, Meyarivan T. A fast and elitist multiobjective genetic algorithm: NSGA-II. Evolutionary Computation, IEEE Transactions on. 2002;6(2):182-97.

128. Haiming L, Yen GG. Rank-density-based multiobjective genetic algorithm and benchmark test function study. Evolutionary Computation, IEEE Transactions on. 2003;7(4):325-43.

129. Konak A, Coit DW, Smith AE. Multi-objective optimization using genetic algorithms: A tutorial. Reliability Engineering & System Safety. 2006 9//;91(9):992-1007.

130. Deb K. Multi-Objective Optimization Using Evolutionary Algorithms: Wiley; 2001.

interpreted modeling for safety and emissions at ...abastos/outputs/teses/projeto de tese...

Documents