empirical assessment of turbo roundabout operations …abastos/outputs/congressos...

EMPIRICAL ASSESSMENT OF TURBO-ROUNDABOUT OPERATIONS ON 1 TRAFFIC AND EMISSIONS 2

3 Paulo Fernandes, MSc. 4

PhD Student, Mechanical Engineering 5 University of Aveiro 6

Dept. Mechanical Engineering / Centre for Mechanical Technology and Automation (TEMA) 7 Campus Universitário de Santiago, 3810-193 Aveiro - Portugal 8

Phone: (+351) 234 370 830, E-mail: [email protected] 9 (corresponding author) 10

11 Sérgio Ramos Pereira, MSc. 12

Research Assistant, Mechanical Engineering 13 University of Aveiro 14


Phone: (+351) 234 370 830, Fax: (+351) 234 370 309, 17 E-mail: [email protected] 18

19 Jorge M. Bandeira, PhD. 20

Research Assistant, Mechanical Engineering 21 University of Aveiro 22


Phone: (+351) 234 370 830, Fax: (+351) 234 370 309 25 E-mail: [email protected] 26

27 Luís Vasconcelos, MSc. 28

Adjunct Professor, Department of Civil Engineering 29 Polytechnic Institute of Viseu 30

Campus Politécnico de Repeses, 3504-510 Viseu – Portugal 31 Phone: (+351) 232 480 500, E-mail: [email protected] 32

33 Ana Bastos Silva, PhD. 34

Assistant Professor, Department of Civil Engineering 35 University of Coimbra 36

Dept. Civil Engineering / Centre for Territory, Transport and Environment (CITTA), 37 Rua Luís Reis Santos - Pólo II, 3030-788 Coimbra – Portugal 38

Phone: (+351) 239 797 100, E-mail: [email protected] 39 40

Margarida C. Coelho, Ph.D.. 41 Assistant Professor, Mechanical Engineering 42

University of Aveiro 43 Dept. Mechanical Engineering / Centre for Mechanical Technology and Automation (TEMA) 44

Campus Universitário de Santiago, 3810-193 Aveiro - Portugal 45 Phone: (+351) 234 370 830, E-mail: [email protected] 46

47 July 2014 48

Submitted for consideration for publication and presentation at the 94th Annual Meeting 49 of the Transportation Research Board, January 11-15, 2015. 50

51 Word Count: 4,941 text words (exclude references) plus 2,000 for figures/tables 52

(8*250) = 6,941 (Max: 7,000 words) 53

mailto:[email protected]




http://citta.fe.up.pt/



Fernandes, Pereira, Bandeira, Vasconcelos, Bastos Silva, Coelho 2

ABSTRACT 1 In the past decade there has been a significant increase in the number of turbo-roundabouts 2 constructed in several European countries. This paper investigates the impact of turbo-3 roundabouts located in urban areas on pollutant emissions generated from vehicles through 4 an empirical assessment. The research also compares the emissions of vehicles moving 5 through a turbo-roundabout and a conventional multi-lane roundabout. Based on field 6 measurements taken at turbo-roundabouts and multi-lane roundabouts located in Grado 7 (Spain) and Aveiro (Portugal) three representative speed profiles for each speed trajectory 8 type were identified: no stop (I), stop once (II) and multiple stops (III). This study also 9 developed discrete models for turbo-roundabouts and multi-lane roundabouts in which the 10 relative occurrence of those speed profiles is expressed as a function of the entry and the 11 conflicting traffic flows. The Vehicle Specific Power (VSP) methodology was then 12 employed to estimate second-by-second pollutant emissions. 13

This research tested the hypotheses that emissions were impacted by the 14 differences in: 1) the characteristics of speed profiles in each movement; 2) the volumes of 15 entry and conflicting flows; 3) the overall saturation level; and 4) the transportation facility 16 considered (turbo-roundabout /multi-lane roundabout). 17

It was found that vehicles emitted more pollutants at turbo-roundabouts compared 18 with multi-lane roundabouts regardless of the traffic demands at the entry and the 19 circulating areas. These results suggest that there are advantages on implementing turbo-20 roundabouts only when the main concern is road safety. 21

22 Keywords: Turbo-roundabouts, Multi-lane roundabouts, Speed Profiles, Discrete Models, 23 Traffic performance, Emissions. 24 25


1. INTRODUCTION AND OBJECTIVES 1 Multi-lane roundabouts are associated with higher capacity rates for traffic compared with 2 single-lane roundabouts. Nonetheless, they present some drawbacks such as higher speeds 3 as vehicles negotiate through the roundabout and the possibility of lane changing and 4 weaving maneuvers at the circulating and exit areas, which generate traffic conflicts. 5

In order to address these problems, the turbo-roundabout concept was developed, 6 as a variation of the conventional multi-lane roundabouts where drivers are required to 7 choose their intended destination before entering the roundabout. The carriageway 8 contains continuous spiral paths in which the entry, the circulating and the exit lanes are 9 usually separated by curbs. Such raised curbs allow eliminating the conflicting points 10 caused by weaving maneuvers and reducing vehicle’s speeds (1). 11

The first turbo-roundabouts were constructed in the Netherlands, in 2000 (2, 3). 12 Since then, there has been a significant increase in the deployment of turbo-roundabouts in 13 several European countries such as Germany, Finland, Poland, Norway (4), and most 14 recently in Spain (5). There is also a growing awareness about this layout in the United 15 States (6). Their design features are usually based on the Dutch guidelines (2, 3). 16

A typical concern in the use of turbo-roundabouts is how traffic flows will be 17 impacted. In fact, there are some differences between multi-lane and turbo-roundabouts 18 that affect vehicle operations (1): 19

20 i. On a conventional roundabout, the outer circulatory lane at the major entries is 21

used by part of the through movements; on a turbo-roundabout, the opposing traffic 22 is concentrated in a single lane, which leads to a decrease in capacity; 23

ii. On a conventional roundabout, drivers in the outer lane of the minor entries are 24 affected by all circulating vehicles, even if the trajectories do not actually intersect. 25 On a turbo-roundabout, the outer lane is used only to turn to the right and the 26 opposing traffic is reduced since part of the through traffic is physically separated 27 at the exit; 28

iii. While right-turning traffic must use the outer entry lane on the conventional 29 roundabout, both inner and outer lanes can be used at the minor entries of a turbo-30 roundabout. 31 32 These design considerations have a significant effect on intersection capacity and 33

can also affect pollutant emissions. 34 This work seeks to introduce a methodology that can quantify emissions at turbo-35

roundabouts and explore the effect of turbo-roundabout operations on pollutant emissions 36 and capacity. The methodology described in this study is built on previous research (7, 8) 37 which was dedicated to the environmental impacts of the conventional single-lane and 38 multi-lane roundabouts. It is assumed that emissions and capacity are impacted by the 39 differences in: 1) the characteristics of speed profiles; 2) the volumes of entry and 40 conflicting flows; 3) the overall saturation level; and 4) the considered layout (multi-lane 41 vs. turbo-roundabout). 42

The research uses an approach founded on experimental measurements of traffic 43 characteristics and saturation levels in turbo-roundabouts to predict the relative occurrence 44 of each speed profile that vehicles experience as they travel in the turbo-roundabout. These 45 speed profiles are: no stopping (I), stop once (II) and multiple stops (III). Emission 46 estimating is then carried out using the Vehicle Specific Power (VSP) methodology (9), 47 which is based on on-board measurements in Light Passenger Vehicles (LPV). Using the 48 developed models, it is possible to estimate the footprints of emissions at any turbo-49


roundabout by simply knowing the entry and conflicting flows and identifying the typical 1 speed profile for each trajectory type. Thus, the objectives of this research are threefold: 2 3

1) To quantify emissions generated from vehicles at roundabouts (turbo-roundabout 4 and multi-lane layouts) located in urban areas; 5

2) To develop and validate appropriate models to explain the interaction between 6 operational variables (entry and conflicting traffic flows) and geometry of a turbo-7 roundabout; 8

3) To compare the emissions and capacity impacts of turbo-roundabouts with multi-9 lane roundabouts. 10

11 2. LITERATURE REVIEW 12 Previous studies in the field of transportation capacity, safety and emissions have dealt 13 with the impacts of turbo-roundabouts on traffic operations and its comparison with 14 conventional single-lane and two-lane roundabouts. 15

From the literature, it is clear that there is not a consensus about the benefits of 16 turbo-roundabouts regarding the available capacity of the intersection. The first studies 17 carried out (10, 11) showed that turbo-roundabouts achieved higher capacity rates than 18 traditional roundabouts with similar design features. Other authors recognized that the 19 relative performance of turbo-roundabouts was toughly influenced by the local traffic 20 conditions (12-14) and layout. Corriere and Guierre explain that depending on each site, 21 the pedestrians, the conflicting traffic flows, the lane capacity, the drivers’ behavior, the 22 balance of the traffic demand on each approach, and the combination of traffic flows at the 23 circulating area will affect each approach capacity and vehicle delay at turbo-roundabouts 24 (15). Vasconcelos et al. proposed a new lane-based capacity methodology to assess the 25 capacity of turbo-roundabout based on gap-acceptance theory. The authors demonstrated 26 that the turbo-roundabout only achieved comparable capacity levels to the traditional two-27 lane layout when the proportion of right turns at the minor entries was very high. They also 28 confirmed that turbo-roundabout performed worse, in relative terms, as the traffic demand 29 at the main entries increased (16). 30

Safety benefits of turbo-roundabouts are identified in almost all previous works. 31 Fortuijn (1) identified that the probability of injury crashes was significantly reduced on 32 turbo-roundabouts, and the measured effect of that layout on safety was comparable with 33 that of single-lane roundabouts. Mauro and Cattany (17) concluded that the accident rates 34 can be reduced up to 50% relatively to two-lane roundabouts. Silva et al. (18) pointed out 35 that the benefits of turbo-roundabouts can be attributed to the smallest deflection levels 36 and speed values imposed by the geometric features of that layout. 37

The interpreted assessment of turbo-roundabouts on several impacts is relatively 38 unknown. Vasconcelos et al. (19) used microsimulation models to evaluate and compare 39 the performance of a single-lane roundabout, in Coimbra, Portugal, and modeled a two-40 lane roundabout and a turbo-roundabout in terms of capacity, safety and emissions. The 41 analysis results showed that the turbo-roundabout reached higher saturation levels and 42 delays than two-lane roundabout, especially for high proportions of left turns (over than 43 60%). Concerning emissions, carbon dioxide (CO2) and nitrogen oxides (NOX) were 44 higher in turbo-roundabout, regardless of the proportion of turning movements and/or 45 traffic flows at each approach. 46

From the literature review, some gaps are inferred. First, the analysis of the turbo-47 roundabout has been centered on capacity and safety. Second, the researchers did not find 48 any work with the empirical characterization of the relative occurrence of each speed 49


profile in turbo-roundabouts. Third, there is a lack of emissions quantification in turbo-1 roundabouts, based on real traffic and vehicle dynamics measurements. 2

The innovation of this study is that it uses empirical data collected on real turbo-3 roundabouts (traffic flows and vehicle activity data) to estimate emissions. Furthermore, it 4 compares emissions levels of turbo-roundabouts with the conventional multi-lane 5 roundabouts layout. 6

7 3. METHODOLOGY 8 This study is an empirical approach founded on field measurements of the vehicle 9 dynamics as well as the overall congestion level. The methodology overview is depicted in 10 Figure 1. Input data such as entry and conflicting traffic flows (opposite entry traffic), 11 queue length and stop-and-go cycles have been collected using overhead videos of the 12 roundabouts (turbo-roundabouts and multi-lane roundabouts). Vehicle activity data such as 13 second-by-second instantaneous speed, acceleration/deceleration and road topographic 14 conditions (grade) have been collected using a Global Positioning System (GPS) data 15 logger and On-Board Diagnostic (OBD) system. Based on the analysis, the relationship 16 between the congestion level of the roundabouts and the portion of the vehicles that 17 experience each speed profile has been established using discrete models (20). After that, 18 the VSP methodology was used to estimate CO2, carbon monoxide (CO), NOX and 19 hydrocarbons (HC) emissions. Finally, a comparison between the discrete models obtained 20 from turbo and multi-lane roundabouts was conducted. The aforementioned 21 methodological steps are briefly described herein. 22 23

24

25

FIGURE 1 Methodology Overview. 26

Aggregate emissions estimation

CO2, CO, NOx and HC emissions for Light Passenger Vehicles, estimation for each profile,

using VSP approach

No Stop

Profile I

Stop Once

Profile II

Stop more than once

Profile III

Algorithm to determine the probability of characteristic speed profile occurrence

Second-by-second vehicle dynamics:

instantaneous speed, acceleration, grade

Entry traffic flow, conflicting traffic

flow, queue length, stop-and-go

GPS Data Logger

OBD Reader Video Recordings

Comparison between predictive models for turbo-roundabouts and multi-lane roundabouts


3.1. Characteristic Speed Profiles 1 Based on vehicle activity, and patterns in speed profiles, there are three different speed 2 profiles that vehicles experience as they approach a generic roundabout (multilane or 3 turbo). It should be mentioned that the relative occurrence of each profile is highly 4 dependent of the level of congestion on the approach (7, 8). The three speed profiles 5 represent: 6

7 I. A vehicle starting to decelerate while approaching the roundabout, enters and 8

negotiate the circulating area without stopping and then accelerates as it is leaving 9 the roundabout back to cruise speed; 10

II. A vehicle decelerates while approaching the roundabout, experience a complete 11 stop at the yield lane to enter in the circulating stream and finds a crossable gap, 12 then accelerates to enter the circulating lane and exits the roundabout; 13

III. A vehicle that experiences several stops on the approach as it moves up the queue 14 to reach the yield lane, and then accelerates to enter the circulating lane and leaves 15 the roundabout. 16 17 The main goal of this research is to quantify the relationship between the 18

congestion levels of the roundabouts with the percentage of the vehicles that experience 19 each speed profile. These levels of congestion are expressed indirectly as the sum of the 20 entry (Qin) and conflicting traffic flows (Qconf) at each entry lane. The video cameras are 21 used to capture the vehicle movements at the entry and circulating lane of the selected 22 multi-lane and turbo-roundabouts. Both Qin and Qconf are obtained for every 15 minutes of 23 morning and afternoon peak periods. The proportion of the drivers that experience no 24 stopping at the entry (PI), one complete stop (PII), multiple stopping (PIII), the number of 25 stop and go situations and queue lengths are also extracted from the videotapes. 26 27 3.2. Discrete Models 28 By selecting three speed profiles (PI, PII and PIII), one is intrinsically considering a discrete 29 choice process. Discrete choice models are based on the theory of stochastic utility 30 whereby a choice is made by a decision maker in order to maximize the utility function. 31 This utility function, as shown in Equation (1), is constructed as a combination of known 32 explanatory variables, the systematic part of utility, and a random part which is unknown 33 (20). 34 35

, , ,i n i n i nU V (1) 36

37 Where: 38 Vi,n is the systematic part of utility which is a linear function to predict the probability that 39 decision maker n chooses alternative i (or, more generically, that a given observation n has 40 outcome i). 41 εi,n represents the error between the systematic part of utility and the true utility given by 42 user n to alternative i. 43 44

Assuming that the error term of the utility expression is logistically distributed the 45 following multinomial logit model (MNL) can be obtained from Equation (2) (21): 46 47

,

,, ,( ) Probability

i n

j n

n

V

n i n j n K V

j C

eP i U U

e

(2) 48


1 Where: Cn is the choice set that the decision maker n faces. 2 3

For this application, it was assumed that the speed profile of a given vehicle can be 4 expressed as a function of the sum of the entry and conflicting traffic flows (Qtotal = 5 Qin + Qconf), as an indirect measure of the congestion level. Since only differences in utility 6 mater in influencing the choice, the outcome “PI” (no stopping) was chosen as reference. 7 The MNL probabilities for Profiles I, II and III are given by Equations 3, 4 and 5, 8 respectively: 9

10

2,0 2,1 3,0 3,1I

1(Y " ")

1 Qtotal QtotalP P

e e

(3) 11

12 2,0 2,1

2,0 2,1 3,0 3,1II(Y " ")

1

Qtotal

Qtotal Qtotal

eP P

e e

(4) 13

3,0 3,1

2,0 2,1 3,0 3,1III(Y " ")

1

Qtotal

Qtotal Qtotal

eP P

e e

14

(5) 15 16 Where the different βi,j are the model parameters to be estimated from the sample data: 17

β2,0 and β2,1 = intercept and coefficient for outcome “Profile II” 18 β3,0 and β3,1 = intercept and coefficient for outcome “Profile III” 19

20 3.3. Site Selection 21 Two sets of roundabouts were selected for this study – three multi-lane roundabouts and 22 three turbo-roundabouts. Figure 2 displays the aerial view of the data collection sites 23 investigated in this study as well as the studied approaches. Three turbo-roundabouts along 24 the N-634 national road in the city of Grado, Spain were selected, as exhibited in Figure 2 25 (a-c). These turbo-roundabouts were selected because in Portugal such layout was not yet 26 constructed in current intersections. Thus, it was intended to represent Iberian case studies. 27 These turbo-roundabouts were constructed and opened in 2009 (5). The through movement 28 from the northeast-bound approach was studied by the use of overhead video cameras and 29 GPS runs in different directions. That approach has two entry lanes from 200 meters to the 30 yield lane of turbo-roundabout. The right lane only provides movements to the first exit 31 while the left lane allows turning movements to the remaining exits. Concerning multi-lane 32 roundabouts, as exhibited in Figure 2 (d-f), they are placed in the urban area of Aveiro, 33 Portugal. Such roundabouts layouts have two entry lanes on their approaches and two 34 circulating lanes. 35

The posted speed limit in the studied areas is 40 km/h. Sites characteristics such as 36 location, circulating width, and the entry, the circulating and the exit speeds are 37 summarized in Table 1. The morning entry traffic flow is also provided. 38

It should be mentioned that these turbo-roundabouts, at the moment of field tests, 39 do not have a curb raised divider (only longitudinal marking-double lines). However, 40 almost every vehicle correctly uses the inner and outer lanes to their intended destination. 41 42


a) TR1 b) TR2

c) TR3 d) ML1

e) ML2 f) ML3

1 2

FIGURE 2 Aerial View of the three data collection Turbo-roundabouts, Grado, 3 Spain: a) TR1; b) TR2;c) TR3 and Multi-lane roundabouts, Aveiro, Portugal: d) 4

ML1; e) ML2 and f) ML3. 5 6

TABLE 1 Key characteristics of the selected roundabouts 7

ID Entry

Speed (km/h)

Exit

Speed (km/h)

Circulating

Speed (km/h)

Circulating

Width (m)

Entry Traffic⃰

(vph)

TR1 18.8 21.5 15.8 6.0 275

TR2 21.8 21.7 17.0 6.2 500

TR3 27.5 25.7 26.5 6.1 435

ML1 34.1 40.2 30.1 8.3 585

ML2 32.2 34.2 24.1 8.2 660

ML3 24.0 35.0 26.1 8.1 470

*Average values of traffic flows (right and left lanes) observed at the morning peak period (8-9 a.m.). 8 9

Qconf

Qin

Qconf

Qin

Qconf

Qin

Qconf

Qin

Qconf

Qin

Qconf

Qin

© 2014 Nokia © 2014 Microsoft Corporation © 2014 Nokia © 2014 Microsoft Corporation



50 meters Location of Videotaping


3.4. Data Collection and Emission Estimation 1 This work applied field data collection methodologies to get traffic characteristics of the 2 two roundabout layouts. The research team scouted and collected data at the roundabouts 3 during morning (8:00 – 11:00 a.m.) and afternoon (5:00 – 8:00 p.m.) peak periods in 4 typical weekdays (Tuesday to Wednesday). The following data was collected: 5

6 Entry and conflicting traffic flows; 7 Queue Length; 8 Number of stop-and-go cycles; 9 Vehicle activity data (speed, acceleration/deceleration and grade). 10

11 Entry and conflicting traffic flows, queue lengths and the number of stop-and-go 12

were gathered from overhead videos installed at strategic points of the roundabouts, as 13 illustrated in Figure 2. The vehicle activity data characterization was recorded using a LPV 14 making several turning movements at the roundabouts. The QSTARZ GPS Travel 15 Recorder (22) was used to acquire some of the parameters related to the vehicle activity 16 data (second-by-second speed and acceleration/deceleration) and the selected sites 17 characteristics (road grade, latitude and longitude). 18

A total of 240 GPS travel runs of through movement (approximately 40 at each 19 location) were extracted and identified for this research (approximately 400 km of road 20 coverage over the course of 15 hours). To reduce systematic errors, three different drivers 21 were used, each one performing an identical number of trips (approximately 40) on each 22 roundabout movement. Concurrently, over 21 hours of video data were gathered from the 23 six roundabouts approaches (approximately 3.5 hours at each location). 24

One widely used approach based on regression based models is the pollutant 25 emissions estimation, through the concept of VSP. This is a methodology introduced by 26 Jimenez-Palacios (23), which provides the characterization of the vehicle activity data on a 27 second-by-second basis. VSP values are categorized in 14 modes of engine regime, and an 28 emission factor for each mode is used to estimate CO2, CO, NOX and HC emissions for 29 Light Passenger Gasoline Vehicles (LPGV) (24). Kolak et al. (25) and Coelho et al. (26) 30 recognized that a VSP based emission model leads to a better estimation of vehicle 31 emissions (25) and regional air quality concentrations (26) than an average speed-based 32 emissions model. An extensive body of research has documented the effective use of VSP 33 in analyzing the emissions of vehicles at different roundabouts layouts (7-8, 19, 27). For a 34 typical LPV VSP is estimated as (28): 35 36

3. 1.1. 9.81.sin arctan 0.132 0.000302. VSP v a grade v (6) 37

Where: 38 VSP = Vehicle Specific Power (kW/ton); 39 v = Instantaneous speed (m/s); 40 a = Instantaneous acceleration or deceleration (m/s

2); 41

grade = Terrain gradient (decimal fraction). 42 43 The average emission rates for pollutants CO2, CO, NOX and HC for each VSP 44

mode for LPGV with engine size <1.4l, Light Passenger Diesel Vehicles (LPDV) <1.9l 45 and Light Commercial Diesel Vehicles (LCDV) <2.5l can be found in the following 46 studies (9, 24-29). 47

After that, the pollutant emissions per vehicle due to the three speed profiles are 48 aggregated in order to evaluate the overall impact of a change in the average trajectory 49


through the roundabout. Equation 7 provides the estimation of hourly emissions generated 1 by vehicles entering a roundabout by using VSP methodology: 2

3

TR in I I II II III IIIE Q E P E P E P (7) 4

5 Where: 6

ETR = Hourly emissions at the turbo-roundabout (g); 7 Ei = Emission per vehicle associated with each speed profile i = I, II and III (g); 8 Pi = Proportion of vehicles that experienced each speed profile i = I, II and III; 9 Qin = Entry flow rate (vph). 10

11 The pollutant emission values of CO2, CO, NOX and HC are estimated from the 12

distribution of VSP time spent in modes obtained from the GPS runs. Therefore Ei is given 13 by the Equation 8: 14

i

1

mN

j mj

m

E F

(8) 15

Where: 16 Eij = Total emissions for source pollutant (g); 17 m = Label for second of travel (s); 18 i= speed profile (I, II and III) 19 j = Source pollutant; 20 Fmj = Emission factor for pollutant j in label for second of travel m (g/s); 21 Nm = Number of seconds (s). 22 23 To estimate the pollutant emissions for each speed profile (Ei), second-by-second 24

emission rates for the vehicles which experience that speed profile is obtained from 25 Equation 6. 26

It should be noted that a fixed travel distance across the roundabout must be used to 27 calculate a complete Ei second-by-second dynamics for a given speed profile. Thus, a 28 roundabout influence area was defined as the sum of the deceleration distance that a 29 vehicle decelerated from cruise speed as it approaches the roundabout, enters the 30 circulating lane and acceleration distance as it leaves the roundabout up to the point that 31 attains the cruise speed. For this analysis, an average roundabout influence area of 250 m 32 was considered. Since the case studies are located on relatively flat grades the effect of that 33 parameter was negligible. For both roundabouts layouts the following distribution fleet 34 composition was used (30): 45% of LPGV, 34% of LPDV and 21% of LCDV. 35

36 4. RESULTS AND DISCUSSION 37 This section presents and discusses the main results from discrete models for turbo-38 roundabouts and multi-lane roundabouts, and characteristic speed trajectories from turbo 39 and multi-lane roundabouts. Further, pollutant emission impacts (CO2, CO, NOX and HC) 40 of the two layouts are compared. 41 42 4.1. Predictive Discrete Models 43 Two MNL models were obtained – one for multilane roundabouts, other for turbo-44 roundabouts. The models were calibrated trough maximum likelihood using the SPSS 45 software (see Table 2). The sample is comprised of 3162 observations in three two-lane 46 roundabouts and 2498 observations in three turbo-roundabouts. Each of these cases was 47


recorded in a database with three fields: roundabout type (Multi-lane – ML or Turbo-1 roundabouts – TR), speed profile (PI, PII or PIII) and total traffic flow (15 minutes period). 2 3

TABLE 2 Calibrated coefficients for the MNL model 4

Multi-lane (ML) roundabouts Turbo-roundabouts (TR)

Speed

profile x B

Std.

Error Wald Sig.

Speed

profile x B

Std.

Error Wald Sig.

PII Intercept -2.85 β2,0 .17 276.29 .00

PII Intercept -2.984 β2,0 .16 340.65 .00

Qtotal .003 β2,1 .00 86.31 .00 Qtotal .002 β2,1 .00 208.59 .00

PIII Intercept -7.55 Β3,0 .62 146.43 .00

PIII Intercept -8.619 Β3,0 .45 366.48 .00

Qtotal .008 Β3,1 .00 61.97 .00 Qtotal .006 Β3,1 .00 295.34 .00

5 Figure 3 illustrates the two calibrated MNL models. As expected, the probability of 6

a driver being able to negotiate the roundabout without stopping (PI) decreases as the total 7 traffic flow increases. Moreover, the comparison of the two graphs shows that the 8 probability for one or more stops (PII and PIII), for the same total traffic, is higher in the 9 turbo-roundabouts. This happens essentially because on the two-lane roundabouts the 10 conflicting traffic is divided by two lanes, which increases the number of large gaps 11 available for the vehicles waiting at the yield line (16). 12

13 a) Multi-lane roundabouts b) Turbo-roundabouts

FIGURE 3 Predictive models for the relative occurrence of speed profiles I, II and 14 III: a) multilane roundabouts; b) turbo roundabouts. 15

16 4.2. Vehicles trajectories at turbo-roundabouts and multi-lane roundabouts 17 As mentioned before, the relative occurrence of the speed profiles are found to be 18 dependent on the prevailing levels of the traffic demand levels at the roundabout (Qtotal = 19 Qin + Qconf). From the observation of the overhead cameras taken at the turbo-roundabouts, 20 it was noted that due to low circulating speed values (see Table 1 for those details), the idle 21 times and stop-and-go situations along the approach lane were significant. With this 22 concern in mind, it is clear that the vehicle’s dynamics through the two layouts is quite 23 different. Thus, two sets of speed profiles (I, II and III) for the turbo and multi-lane 24 roundabouts were selected to assess the differences in emissions. These speed profiles are 25 the representative average speed trajectories for turbo-roundabout and a multi-lane 26 roundabout using multiple field data GPS runs collected through the studied cases. To 27 maintain consistency between layouts, similar travel distance, posted speed and traffic 28 flows (entry and conflicting flows) values were used. 29

Figure 4 displays the speed profiles for through movements of the turbo-30 roundabout TR2 and multi-lane roundabout ML2 (from the left entry lane) for each speed 31

0%

20%

40%

60%

80%

100%

0

200

400

600

800

100

0

120

0

140

0

160

0

180

0

200

0

Entry + Conflicting Flow (Qtotal, veh/h)

p(1) p(2) p(3)

0%

20%

40%

60%

80%

100%

0

200

400

600

800

100

0

120

0

140

0

160

0

180

0

200

0Entry + Conflicting Flow (Qtotal, veh/h)

p(1) p(2) p(3)PI

PI

PI

PI

PII

PI

PII

PI

PIII

PI

PIII

PI


profile and the correspondent percent time in each VSP mode (considering all performed 1 runs and all drivers). These speed trajectories are later used to estimate emissions from the 2 turbo-roundabout and multi-lane roundabout using the predictive regression models 3 developed for each case. 4

Based on the raw distributions of VSP modes from the turbo-roundabout for speed 5 profile I, vehicles spent most of the time in VSP modes 1, 2, 3 and 4 which corresponds to 6 decelerations as vehicles approach the turbo-roundabout (modes 1 and 2), enter the 7 circulating lanes at low speeds or stop (mode 3) and accelerations as they exit the turbo-8 roundabout (mode 4). The percent of the time spent in VSP modes higher than 4 and 5 for 9 the turbo-roundabout is on average lower across the three speed profiles compared to 10 multi-lane roundabout on speed profile I. This means that vehicles at the turbo-roundabout 11 experienced moderate speeds (perhaps due to higher deflection angles and low circulating 12 speeds). Moreover, a vehicle travelling in turbo-roundabout faces with higher idle and low 13 speed situations at the downstream and circulating areas compared to a vehicle traveling in 14 multi-lane roundabout, especially in speed profiles II and III. The percent of the time spent 15 in mode 3 confirm these findings. 16 17


a)

b)

c)

Legend: TR – Turbo-Roundabout; ML – Multi-lane Roundabout 1 FIGURE 4 Average speed trajectory over distance and total seconds spent in each 2 VSP model (and standard deviation values) for each speed profile: a) I, b) II, c) III. 3

4 4.3. Emission rates 5 This section employs the predictive discrete models and trajectories of each speed profile 6 I, II and III to calculate and compares the emissions produced by vehicles at turbo-7 roundabout and multi-lane roundabout. According to the different values of the entry and 8 conflicting flows, the percentage of the vehicles that experience any of the three speed 9 profiles at turbo-roundabouts are identified from Figure 3. After that, the total emissions 10 for each speed trajectory in each layout are calculated from Equations 6, 7 and 8, using 11 second-by-second speed trajectory exhibit in Figure 4, and considering the composition 12 fleet distribution presented in the section 3.4. 13

Following the previous results, six scenarios of traffic demand are established with 14 the main goal of compare the CO2, CO, NOX and HC emissions for turbo-roundabout and 15 multi-lane roundabout. The pollutant emission effects of both layouts were explored at two 16

0

10

20

30

40

50

0 5 10 15 20 25 30 35

Sp

eed

[k

m/h

]

Travel time [s]

TR

ML

0%

10%

20%

30%

40%

1 2 3 4 5 6 7 8 9 1011121314

Rel

ati

ve

Fre

nq

uen

cies

VSP Modes

TR

ML

0

10

20

30

40

50

0 5 10 15 20 25 30 35 40 45

Sp

eed

[k

m/h

]

Travel time [s]

TR

ML

0%

10%

20%

30%

40%

1 2 3 4 5 6 7 8 9 1011121314

Rel

ati

ve

Fre

nq

uen

cies

VSP Modes

TR

ML

0

10

20

30

40

50

0 5 10 15 20 25 30 35 40 45 50 55

Sp

eed

[k

m/h

]

Travel time [s]

TR

ML

0%

10%

20%

30%

40%

50%

1 2 3 4 5 6 7 8 9 1011121314

Rel

ati

ve

Fre

nq

uen

cies

VSP Modes

TR

ML


levels: 1) sum of the entry and conflicting flows; and 2) total saturation level. The 1 following scenarios are: 2

Scenario 1: Qin = Qconf = 100 vph (Qtotal =200 vph); 3 Scenario 2: Qin = Qconf = 200 vph (Qtotal =400 vph); 4 Scenario 3: Qin = Qconf = 300 vph (Qtotal =600 vph); 5 Scenario 4: Qin = Qconf = 400 vph (Qtotal =800 vph); 6 Scenario 5: Qin = Qconf = 500 vph (Qtotal =1000 vph); 7 Scenario 6: Qin = Qconf = 600 vph (Qtotal =1200 vph). 8 9 The comparison of hourly emissions (g) for the turbo-roundabout and multi-lane 10

for CO2, CO, NOX and HC are illustrated in Figure 5. The results shows that both in low 11 and moderate congestion levels (scenarios 1-4) the pollutant emissions generated from 12 vehicles at the turbo-roundabout are higher than those verified at the multi-lane 13 roundabout (15%, 20%, 15% and 19% for CO2, CO, NOX and HC, respectively). For high 14 flow rate (scenarios 5-6), turbo-roundabouts yield even more emissions than multi-lane 15 roundabout (21%, 24%, 17% and 28% for CO2, CO, NOX and HC, respectively). This is 16 possible due to the longer stop-and-go cycles that vehicles experienced at the turbo-17 roundabout since they had mostly speed profiles II (>33%) or III (>36%) in high flow rate 18 scenarios. 19

It should be also referred that the relative difference between emissions produced at 20 the turbo and multi-lane roundabouts is not sensitive to any other conditions such as 21 conflicting flow or congestion level. These findings are in line with previous research 22 taken at turbo-roundabouts in which those layouts produced an amount of CO2 and NOX 23 emissions higher than multi-lane roundabouts (19). 24

25

a) b)

c) d)

Legend: TR – Turbo-Roundabout; ML – Multi-lane Roundabout 26 FIGURE 5 Variation of total emission (g) per hour for different traffic scenarios: a) 27

CO2; b) CO; c) NOX and d) HC. 28

0

10,000

20,000

30,000

40,000

50,000

1 2 3 4 5 6

CO

2 E

mis

sio

ns

[g]

Scenarios

TR

ML

0

20

40

60

80

100

1 2 3 4 5 6

CO

Em

issi

on

s [g

]

Scenarios

TR

ML

0

6

12

18

24

30

1 2 3 4 5 6

NO

X E

mis

sio

ns

[g]

Scenarios

TR

ML

0

2

4

6

8

1 2 3 4 5 6

HC

Em

issi

on

s [g

]

Scenarios

TR

ML


Figure 6 depicts the cumulative distribution of CO2, CO, NOX and HC emissions 1 along the influence area of the turbo-roundabout and multilane roundabout, for a given 2 speed (no stopping). The acceleration from exit section to cruise speed contributes to about 3 40% and 54% of total CO2 emissions in turbo-roundabout and multi-lane roundabout, 4 respectively. It is noticed that there is also a sharp increase in CO and NOX emissions at 5 the turbo-roundabout at the entry and circulating areas. These areas correspond to about 6 35% of travel distance and are associated to more than 50% of CO and NOX emissions, 7 which indicates that the deceleration/acceleration events are more relevant for CO 8 emissions in the turbo-roundabout than the multi-lane layout. Thus, the main conclusion 9 from Figure 6 is that the difference between cruise and circulating speeds has more impact 10 on emissions in turbo-roundabout than the deceleration/acceleration rates. 11 12

a) b)

c) d)

Legend: TR – Turbo-Roundabout; ML – Multi-lane Roundabout 13 FIGURE 6 Spatial distributions of emissions in turbo-roundabout and multi-lane 14

roundabout: a) CO2, b) CO, c) NOX and d) HC. 15 16 5. CONCLUSIONS 17 This paper explored the impact of turbo-roundabouts located on urban areas on pollutant 18 emissions, using a methodology based on real measurements of traffic performance of 19 vehicles. This study also compared the emissions for vehicles travelling in a turbo-20 roundabout and multi-lane roundabouts. 21

The proposed methodology first used discrete models to establish a relation 22 between distinct speed profiles (no stop, stop once and several stops) that vehicles are 23 subject at roundabouts (turbo and multi-lane roundabouts) and traffic conditions (entry and 24 conflicting flows). Second, a representative speed profile for each trajectory type 25 (assuming that each layout had similar entry speeds) was selected. Third, the VSP 26

0

20

40

60

80

100

0 50 100 150 200 250

Cu

mu

lati

ve

per

cen

tag

e

of

CO

2 E

mis

sio

ns

Cumulative Distance [m]

TR

ML

0

20

40

60

80

100

0 50 100 150 200 250

Cu

mu

lati

ve

per

cen

tag

e

of

CO

Em

issi

on

s


TR

ML

0

20

40

60

80

100

0 50 100 150 200 250

Cu

mu

lati

ve

per

cen

tag

e

of

NO

X E

mis

sio

ns


TR

ML

0

20

40

60

80

100

0 50 100 150 200 250

Cu

mu

lati

ve

per

cen

tag

e

of

HC

Em

issi

on

s


TR

ML


methodology was used to estimate the pollutant emissions for each speed profiles 1 occurrence. 2

The findings of this paper showed that vehicles circulating in turbo-roundabouts, 3 considering a through movement, produced more emissions (16-22% depending on the 4 pollutant) when compared with circulating in conventional multi-lane roundabouts, thus 5 there is no advantage on the implementation of turbo-roundabouts from an air pollution 6 point of view even the capacity of the intersection is rather below to the maximum. 7

Nonetheless, there is some future work that can be developed, namely data 8 gathering in turbo-roundabouts: 1) with different configurations (these three turbo-9 roundabouts have small inscribed circles and circulating widths); 2) with higher traffic 10 volumes; 3) with higher variability in terms of speeds, approach geometries and traffic 11 flows among turbo-roundabouts; and 4) with the presence of a curb raised divider that may 12 affect vehicles’ maneuver along the turbo-roundabout (in the measured turbo-roundabouts 13 there were longitudinal marking-double lines only). Also as future work, the predictive 14 discrete models for the speed profile should be improved by linking the speed profile to the 15 entry capacity and saturation rate. These performance measures can be effectively 16 estimated using well-established models based on the gap-acceptance theory. The resulting 17 predictive models can then be applied to different geometries or directional splits of the 18 approach traffic. 19

20 ACKNOWLEDGEMENTS 21 This work was partially funded by FEDER Funds through the Operational Program 22 “Factores de Competitividade COMPETE” and by National Funds through FCT within the 23 project PTDC/SEN-TRA/122114/2010 and the Strategic Project PEst-24 C/EME/UI0481/2014. P. Fernandes acknowledges the support of FCT for the Scholarship 25 SFRH/BD/87402/2012. 26 27 REFERENCES 28 1. Fortuijn, L. Turbo Roundabouts - Design Principles and Safety Performance. In 29

Transportation Research Record: Journal of the Transportation Research Board, No. 30 2096, Transportation Research Board of the National Academies, Washington, D.C., 31 2009, pp. 16-24. 32

2. Royal Haskoning DHV. Roundabouts - Application and Design, Ministry of Transport, 33 Public Works and Water management, 2009. 34

3. CROW. Turborotondes, publicatie 257. Ede, april, 2008. 35

4. Tollazzi, T., M. Rencelj, and S. Turnsek. Slovenian Experiences with Alternative Types 36 of Roundabouts - “Turbo” and “Flower” Roundabouts, In 8th International 37 Conference on Environmental Engineering, Vilnius, Lithuania, 2011. 38

5. Bulla, L. A., and W. Castro. Analysis and comparison between two-lane roundabouts 39 and turbo roundabouts based on a road safety audit methodology and 40 microsimulation: a casa study in urban area, In 3rd International Conference on Road 41 Safety and Simulation, Indianapolis, Indiana, 2011. 42

6. Trueblood, M. Retrofitting closely spaced intersections using the Peanut Turbo 43 Roundabout concept, In 3rd International Conference on Roundabouts, Carmel, 44 Indiana, 2011. 45

7. Coelho, M. C., T. L. Farias, and N. M. Rouphail. Effect of roundabout operations on 46 pollutant emissions. Transportation Research Part D: Transport and Environment, 47 Vol. 11, No. 5, 2006, pp. 333-343. 48


8. Salamati, K., M. C. Coelho, P. J. Fernandes, N. M. Rouphail, H. C. Frey, and J. 1 Bandeira. Emission Estimation at Multilane Roundabouts: Effect of Movement and 2 Approach Lane. In Transportation Research Record: Journal of the Transportation 3 Research Board, No. 2266, Transportation Research Board of the National Academies, 4 Washington D.C., 2013, pp. 12-21. 5

9. Anya, A. R., N. M. Rouphail, H. C. Frey, and B. Liu. Method and Case Study for 6 Quantifying Local Emissions Impacts of Transportation Improvement Project 7 Involving Road Realignment and Conversion to Multilane Roundabout. Transportation 8 Research Board 92nd Annual Meeting, Washington, D.C., 2013. 9

10. Yperman, I., and L. H. Immers. Capacity of a turbo-roundabout determined by micro-10 simulation, In 10th World Congress on ITS, Madrid, Spain, 2003. 11

11. Engelsman, J. C., and M. Uken. Turbo roundabouts as an alternative to two lane 12 roundabouts, In 26th Annual Southern African Transport Conference, Pretoria, South 13 Africa, 2007. 14

12. Giuffrè, O., M. Guerrieri, and A. Granà. Evaluating capacity and efficiency of turbo-15 roundabouts. Presented at 88th Annual Meeting of the Transportation Research Board, 16 Washington, D.C., 2009. 17

13. Mauro, R., and F. Branco. Comparative Analysis of Compact Multilane Roundabouts 18 and Turbo-Roundabouts. Journal of Transportation Engineering, Vol. 136, No. 4, 19 2010, pp. 316-322. 20

14. Giuffrè, O., A. Granà, and S. Marino. Turbo-roundabouts vs Roundabouts Performance 21 Level. Procedia - Social and Behavioral Sciences, Vol. 53, 2012, pp. 590-600. 22

15. Corriere, F., and M. Guerrieri. Performance Analysis of Basic Turbo-Roundabouts in 23 Urban Context. Procedia – Social and Behavioral Siences, Vol. 53, No. 3, 2012, pp. 24 622-632. 25

16. Vasconcelos, A. L. P., A. B. Silva, and Á. J. M. Seco. Capacity of normal and turbo-26 roundabouts – comparative analysis. Proceedings of the Institution of Civil Engineers 27 – Transport, Vol. 167, No. 2, 2014, pp. 88-99. 28

17. Mauro, R., and M. Cattani. Potential Accident Rate of Turbo-Roundabouts. In 4th 29 International Symposium on Highway Geometric Design, Valencia, Spain, 2010. 30

18. Silva, A. B., L. Vasconcelos, and S. Santos. Moving from Conventional Roundabouts 31 to Turbo-roundabouts. Procedia - Social and Behavioral Sciences, Vol. 111, 2014, pp. 32 137-146. 33

19. Vasconcelos, L., A. B. Silva, A. B. Seco, P. Fernandes, and M. C. Coelho. Turbo 34 Roundabouts: Multicriterion Assessmnent on Intersection Capacity, Safety, and 35 Emissions. In Transportation Research Record: Journal of the Transportation 36 Research Board, No. 2402, Transportation Research Board of the National Academies, 37 Washington D.C., 2014, pp. 28-37. 38

20. Ben-Akiva, M., and S. Lerman. Discrete Choice Analysis: theory and application to 39 travel demand. MIT Press, Cambridge, MA, 1985. 40

21. Correia, G., and A. B. Silva. Setting Speed Limits on Rural Two-Lane Highways by 41 Modeling the Relationship between Expert Judgment and Measurable Roadside 42 Characteristics. Journal of Transportation Engineering, Vol. 137, No. 3, 2011, pp. 43 184-192. 44

22. Qstarz International Co. Qstarz BT-Q1000XTM User Guide, Qstarz International Co., 45 Ltd., Taiwan, R.O.C., 2012. 46


23. Palacios, J. Understanding and quantifying motor vehicle emissions with vehicle 1 specific power and TILDAS remote sensing. Department of Mechanical Engineering, 2 Massachusetts Institute of Technology, Cambridge, MA, 1999. 3

24. Frey, H. C., K. Zhang, and N. M. Rouphail. Fuel Use and Emissions Comparisons for 4 Alternative Routes, Time of Day, Road Grade, and Vehicles Based on In-Use 5 Measurements. Environmental Science & Technology, Vol. 42, No. 7, 2008, pp. 2483-6 2489. 7

25. Kolak, O. İ., O. Feyzioğlu, Ş. İ. Birbil, N. Noyan, and S. Yalçindağ. Using emission 8 functions in modeling environmentally sustainable traffic assignment policies. Journal 9 of Industrial and Management Optimization, Vol. 9, No. 2, 2013, pp. 341-363. 10

26. Coelho, M. C., T. Fontes, J. M. Bandeira, S. R. Pereira, O. Tchepel, D. Dias, E. Sá, J. 11 H. Amorim, C. Borrego. Assessment of potential improvements on regional air quality 12 modelling related with implementation of a detailed methodology for traffic emission 13 estimation. Science of the Total Environment, Vol. 470–471, 2014, pp. 127–137. 14

27. Mudgal, A., S. Hallmark, A. Carriquiry, and K. Gkritza. Driving behavior at a 15 roundabout: A hierarchical Bayesian regression analysis. Transportation Research 16 Part D: Transport and Environment, Vol. 26, 2014, pp. 20-26. 17

28. US EPA. Methodology for developing modal emission rates for EPA’s multi-scale 18 motor vehicle & equipment emission system. Prepared by North Carolina State 19 University for US EPA, Ann Arbor, MI, August, EPA420, 2002. 20

29. Coelho, M. C., H. C. Frey, N. M. Rouphail, H. Zhai, and L. Pelkmans. Assessing 21 methods for comparing emissions from gasoline and diesel light-duty vehicles based 22 on microscale measurements. Transportation Research Part D: Transport and 23 Environment, Vol. 14, No. 2, 2009, pp. 91-99. 24

30. ACAP. Automobile Industry Statistics 2014 Edition [in Portuguese]. ACAP – 25 Automobile Association of Portugal, 2014. 26

empirical assessment of turbo roundabout operations …abastos/outputs/congressos...

Documents