Simulated Capacity of Roundabouts and Impact of

Roundabout Within a Progressed Signalized Road

Joe G. Bared
Federal Highway Administration,
Turner-Fairbank Highway Research Center,
McLean, VA
Joe.Bared@fhwa.dot.gov

Praveen K. Edara*
Department of Civil and Environmental Engineering,
Virginia Polytechnic Institute and State University,
Falls Church, VA
praveen@vt.edu

Word Count = 2549 (text) + 4000 (figures and tables) = 6549

*
Corresponding author
Simulated Capacity of Roundabouts and Impact of Roundabout Within a
Progressed Signalized Road
Joe G. Bared, FHWA
Praveen K. Edara, Virginia Tech

Abstract
Many intersections in the urban areas are signalized. As roundabouts are beginning to multiply, they are
being considered adjacent to signalized intersections and for replacing some signalized intersections.
Traffic simulation has been used to study the performance of both signalized and un-signalized
intersections. This research uses simulation to study the traffic impacts of roundabouts. In this paper, two
problems are studied. Firstly, urban single lane and dual lane roundabouts are modeled in VISSIM traffic
simulation software. Simulation results are compared with the results of RODEL (empirical model) and
aaSIDRA (analytical model). Comparison with real data collected from various sites in United States shows
that VISSIM results are closer to the real data than the RODEL and aaSIDRA results. Secondly, the impact
of signalized intersection proximity to roundabouts is studied using the developed model. More
specifically, the impact of coordinated signalized arterial when a roundabout is inserted within an arterial
corridor is studied. Results of average delay measures are comparable to the signalization alternative when
the roundabout is operating below capacity. However, at heavy volumes, when the roundabout is operating
at capacity, then the performance of signalization is slightly better.

INTRODUCTION
Roundabouts are beginning to multiply in the United States; they are being considered adjacent to
signalized intersections and are replacing some of the signalized intersections. Roundabouts have proved to
be a safer alternative to at-grade signalized intersection for both motor vehicles and pedestrians (1). As
roundabouts are becoming increasingly popular in the United States, it is of utmost importance to conduct
research on the traffic performance at roundabouts. Traditionally, empirical (e.g. RODEL (2)) and
analytical (e.g. aaSIDRA (3)) models have been developed to study the traffic performance. Empirical
models are developed based on regression using data collected at currently operating roundabouts. The
analytical models consider geometry, time gaps, and follow-up time among other variables while modeling
the roundabouts.
Traffic simulation has been used to study the performance of both signalized and un-signalized
intersections. However, simulation has not been used much in the past to study the roundabout
performance. One major reason for this is the difficulty to model roundabouts using simulation software.
Not many simulation software are flexible enough to allow the user to model roundabouts. VISSIM (4) is
one of the few simulation software that can be used to model roundabouts. In this paper, two problems are
studied. Firstly, urban single lane and dual lane roundabouts are modeled in VISSIM traffic simulation
software. Simulation results are compared with the results of RODEL (empirical model) (2), aaSIDRA
(analytical model) (3), and field data. Secondly, the impact of signalized intersection proximity to
roundabouts is studied using the developed model. More specifically, the impact of coordinated signalized
arterial when roundabouts are inserted within an arterial corridor is studied.

ORGANIZATION OF THE PAPER

The paper is organized in the following way – first section explains the simulation modeling of typical
single lane and dual lane roundabouts, second section presents the results of capacity analysis, third section
starts the second topic of the paper - roundabout within a signalized arterial, simulation results and
conclusions are presented in the final section.

TYPICAL ROUNDABOUTS AND SIMULATION MODELING

The geometry of single lane and dual lane roundabouts modeled in this paper are shown in Table 1.
VISSIM simulation software is used to model the traffic operations at roundabouts. VISSIM is a
microscopic, time-step and behavior based simulation model developed to model urban traffic operations.

2
Simulation in VISSIM involves three modules – input module, simulator module, and output module. The
input module has a Windows-based graphic user interface. The simulator is used for generation, movement,
system status update, and collection of statistical data. The output module is usually either a text file or an
animation file (4). Figure 1 shows the VISSIM screen shot of single lane and dual lane roundabouts.
A CAD layout of the roundabout is imported into the VISSIM software and set as background on
which the VISSIM links are drawn. Appropriate scale and units are entered so that all the measurements are
in the same units. While drawing the links, number of lanes, lane widths, and gradients are specified.
Desired speeds on the approaches are set between 30mph to 36mph for cars and 25mph to 28mph for
trucks; and on entries, circulating, and exiting curves between 15mph to 18mph for cars and 12mph to
15mph for trucks. The most important aspect of modeling a roundabout in VISSIM lies in setting the
priority rules for entering, and exiting traffic movements. VISSIM priority rules check for two basic
parameters – minimum gap time and minimum headway. Vehicles enter the roundabout only when the time
gap and headway as measured from the conflict marker are greater than the respective minimum values.
The values for these parameters are set partly based on field experience and partly based on viewing the
simulation animation to have no visible collisions between vehicles. But, appropriate caution needs to be
taken while setting priority rules, as higher values of these parameters could decrease the capacity of a
roundabout considerably.
In VISSIM, a priority rule consists of one stop line and one or more conflict markers that are
associated with the stop line (see Figure 2). The stop line decides whether to allow or not to allow the
vehicles to cross depending on the current gap time and headway available at the conflict marker (by
checking with minimum headway and the minimum gap time set by the user).
First, during the simulation, the current gap time is determined every time step by the time an
approaching vehicle will require to reach the conflict marker, assuming it continues to travel at its current
speed. If the current gap time is less than the minimum gap time defined for the conflict marker, the
corresponding stop line stops any approaching vehicle (see Figure 2) (4). Second, the minimum headway
can be typically defined as the length of the conflict area. During the simulation, the distance between the
conflict marker and the first vehicle approaching it determines current headway. Whenever the current
headway is less than the minimum headway, the corresponding stop line stops any approaching vehicle (see
Figure 2).
Setting priority rules for a single lane roundabout is straightforward. A minimum time gap of 3s
for cars and 3.5s for trucks, a minimum headway of 16ft are used (4, 5, 6) (these values are arrived at only
after numerous iterations or suggested by the manual/literature). We can see that the time gap used for
trucks is higher than the gap used for cars, this is due to the fact that trucks entering the roundabout have
lower acceleration capabilities as compared to the acceleration capabilities of cars, and hence would require
greater amount of time gap to safely enter the roundabout.
In a single lane roundabout (Figure 3), one stop line is used at each roundabout approach. Two
conflict markers are defined for this stop line as shown in the figure (1&2 in the figure). Conflict marker 1
sets the conditions for normal traffic conditions (time gap and headway), while conflict marker 2 secures
the conflict area during slow moving traffic and congestion inside the roundabout (headway is the only
criteria). Conditions at both the markers must be satisfied for a vehicle to enter the roundabout. As
mentioned before, different set of stop lines and the corresponding conflict markers are used for trucks (in
the figure the markers for cars and trucks overlap each other). The time gap and headway parameter values
are shown in Table 2.
Priority rules are set in a similar way for a dual lane roundabout also, however, the procedure is
quite complicated as it involves interactions between two entering lanes and two circulating lanes. Several
priority rules are necessary to model the entry of dual lane roundabout. Each priority rule serves a different
purpose. Due to the difference in acceleration capabilities and the vehicle lengths, cars and trucks are
modeled separately. In Figure 4, the two lanes of roundabout approach are numbered. Outer lane is
numbered 1, and the inside lane is numbered 2. There are 12 priority rules that are used in VISSIM to
completely define the roundabout entry traffic behavior, 5 rules for lane 1, and 7 rules for lane 2. Due to the
space constraint of the paper, it is not possible to show all priority rules. In the following paragraph, three
major priority rules are explained.
Entering traffic using lane 1 should satisfy the following conditions to enter the roundabout - 1)
look for minimum distance headway (16ft) during traffic conditions where circulating traffic is moving
slowly (during congestion within roundabout), 2) look for minimum time headway of approaching
circulating vehicles during traffic conditions where circulating traffic is moving at higher speeds (3s for

3
passenger cars and 3.5s for trucks), 3) vehicles entering should also look out for circulating vehicles in the
inner lane of the roundabout, by checking for a minimum time gap for the approaching vehicles (2s for both
cars and trucks).
Other than the priority rules at the roundabout entrances, priority rules are set at exits also. In
Figure 4, inner lane is numbered 3, and outer lane is numbered 4. Vehicles exiting the roundabout from
lane 3 should yield for circulating vehicles in lane 4, and vice versa. The priority rules set for these lanes
look only for the minimum headway violation (minimum gap time is set to 0s). Minimum headway of 40ft
for cars and 60ft for trucks is used on both lanes. Again, these values are obtained only after numerous
iterations checking for any potential vehicle incidents.

RESULTS AND CAPACITY COMPARISON

The modeled roundabouts are used to determine capacities by flooding an entry at a time and facilitating a
wide variation of circulating volumes. For each approach, the maximum entry capacity and the
corresponding conflicting flow are determined from simulation. In VISSIM, we assumed that an approach
has reached capacity when the throughput is less than the input volume for that approach by more than
100vph and average delay for that approach exceeds 70s. The same procedure is repeated for several traffic
scenarios and the ‘maximum entry capacity vs conflicting flow’ plot is obtained. Capacity estimates using
RODEL, and aaSIDRA are also computed for corresponding circulating volumes.
Figures 5 and 6 show the capacity plots for single lane and dual lane roundabouts respectively. In
Figure 6, VISSIM results for two different minimum time gaps are plotted (2.5s and 3s for cars at
roundabout entry). For the single lane, we can infer that the VISSIM capacity values are less than both
RODEL and aaSIDRA predictions (for most of the cases). However, the behavior of VISSIM plot is similar
to the aaSIDRA plot (displaced by about 200 veh/hr).
For the dual lane roundabout, again, VISSIM capacity values are lower than the aaSIDRA and
RODEL predictions. Also, the VISSIM plot and aaSIDRA plot are parallel to each other displaced by about
500veh/hr. VISSIM capacity estimates do not change considerably when the minimum time gap at the
roundabout entry is changed from 3s to 2.5s for cars and 3.5s to 3s for trucks.
Figures 7 and 8 are the plots of real data collected at different roundabout sites in United States
(6). Data collected at the sites included ADT volumes, Crash data, Geometry data, Video data, and Speed
data. Fifteen different sites of single lane roundabouts in U.S were selected for data collection, which
resulted in generating 434 1-min data values that can be used for capacity analysis. For dual lane
roundabouts, data was collected at seven different U.S sites, resulting in 252 1-min data values. In the plots
shown in Figures 7 and 8, ‘qe’ denotes the Entry traffic flow rate, and ‘x’ denotes the Degree of saturation.
Different types of regression models (linear and exponential) are used to fit this collected data. Details
regarding the data fitting can be obtained from the ongoing NCHRP Project No. 3-65 (6).
In Tables 3 and 4, we compare the VISSIM capacity estimates with the real data (actually one of
the regression equation of real data – Tanner-Wu fitted equation). The 6 data points for single and dual
lane roundabouts are surprisingly comparable except at low circulating traffic volume indicated as
observation 1.

ROUNDABOUT WITHIN A SIGNALIZED ROAD

Description of the Design and Simulation

The second part of this paper deals with the study of the impact of signalized intersection proximity to
roundabouts. It is our hypothesis that when a roundabout replaces a signalized intersection within an
arterial, the overall traffic performance would not be worse than the fully signalized design (Figures 9, and
10). To check this hypothesis, the middle signalized intersection is replaced with a dual lane roundabout.
Both alternatives are simulated in VISSIM and the results were comparable.
A section of an arterial consisting of three signalized intersections is analyzed. The intersections
are separated by a 1/4 mile each. The arterial consists of two through lanes and one exclusive left turn and
right turn lanes at each intersection (Figure 9). The signals are coordinated and have short cycle lengths (60
sec). Signal coordination is achieved by using signal optimization software, TRANSYT-7F. The arterial is
simulated in VISSIM for three hypothetical traffic cases (Table 5a, b, c), and average delays per vehicle
were recorded from the simulation.

4
 The second intersection is now replaced with a dual lane roundabout (Figure 10). The network is
simulated for the same traffic flow cases (Table 5). Simulation results are shown in Table 6. For Case 2 and
Case 3 flows, the hypothesis proves to be true, i.e., roundabout performance is better than signalization.
However, when the roundabout approaches are operating near capacity (Case 1), the fully signalized design
has slightly lower overall delay.
CONCLUSIONS
In this paper, two problems were studied - Firstly, urban single lane and dual lane roundabouts are modeled
in VISSIM traffic simulation software. Simulation results are compared with the results of RODEL
(empirical model) and aaSIDRA (analytical model). Secondly, the impact of signalized intersection
proximity to roundabouts is studied using the developed models. Specifically, the impact of coordinated
signalized arterial when a roundabout is inserted within an arterial corridor is studied. The following
conclusions can be made from the analysis and results:

• Simulated capacities of Single-lane roundabouts are noticeably lower than RODEL and aaSIDRA,
however, they are comparable to fitted U.S field capacity data.
• Similarly, capacities of dual-lane roundabouts as simulated by VISSIM are significantly lower than
RODEL and aaSIDRA, and are comparable to U.S field capacity data for a certain fitted regression.
• A roundabout placed within a signalized, coordinated arterial placed quarter mile from adjacent signals
showed comparable delays to a fully signalized arterial. This finding is true when the roundabout is
operating at or below capacity.

RECOMMENDATION FOR FUTURE RESEARCH

In this paper, traffic performance of the roundabouts was studied using simulation. In the future, safety
impacts of roundabouts would be studied using simulation. Surrogate safety assessment model is currently
under development at FHWA, and after its completion, we expect to use it to compare the safety aspects of
roundabouts and signalized intersections (isolated and within an arterial). The proposed safety model aims
at extracting the safety features from traffic simulation models (VISSIM, AIMSUN, and TEXAS Model)
by analyzing the trajectory of vehicles and estimating their proximity in terms of time, speed differentials
and deceleration rates. Another recommendation would be to study the pedestrian performance at
roundabouts.

REFERENCES (format)

1. FHWA. Roundabouts: An Informational Guide. Publication FHWA-RD-00-067. U.S. Department of

Transportation, 2000.
2. RODEL User’s Manual. Rodel Software Ltd and Staffordshire County Council, U.K, 2002.
3. aaSIDRA User’s Manual. Akcelik and Associates Pty Ltd, PO Box 1075G, Greythorn, Vic 3104,
AUSTRALIA, 2000
4. VISSIM 3.70 User Manual. PTV Planung Transport Verkehr AG: Karlsruhe, Germany, 2003.
5. Akcelik, R., A Roundabout Case Study Comparing Capacity Estimates from Alternative Analytical
Models. Presented at 2nd Urban Street Symposium, Anaheim, California., 2003.
6. Kittleson & Associates., Applying Roundabouts in the United States. NCHRP 3-65 (January-March
2004 DRAFT Quarterly Progress Report).

5
LIST OF TABLES AND FIGURES
Table 1 Geometry of the modeled roundabouts
Table 2 Priority rules for a single lane roundabout in VISSIM
Table 3 Single lane roundabout – Comparison of VISSIM results with Real Data
Table 4 Dual lane roundabout – Comparison of VISSIM results with Real Data
Table 5 a), b), c) Traffic flows at each intersection
Table 6 Comparison of traffic performance
Figure 1 a) Layout of a Single Lane Roundabout in VISSIM, b) Layout of a Dual Lane Roundabout in VISSIM.
Figure 2 Definition of Priority Rules.
Figure 3 Single Lane Roundabout – Priority rules in VISSIM.
Figure 4 Dual Lane Roundabout – Priority rules in VISSIM.
Figure 5 Single Lane Roundabout Capacity Analysis.
Figure 6 Dual Lane Roundabout Capacity Analysis.
Figure 7 Single Lane Roundabout – Real data (6).
Figure 8 Dual Lane Roundabout – Real data (6).
Figure 9 VISSIM screenshot of three coordinated signalized intersections.
Figure 10 Second signal replaced by a roundabout.
TABLE 1 Geometry of the modeled roundabouts

Single Lane Dual Lane

Inscribed circle diameter 35m 55m
Entry radius 20m 40m
Exit radius 20m 40m
Entry width 4.5m 8.5m
Approach width 4m 7.3m
Departure width 4m 7.3m
Exit width 4.5m 8.5m
Circulatory road width 6m 9.5m

7
TABLE 2 Priority rules for a Single lane roundabout in VISSIM

Marker 1 - Cars Marker 2 - Cars Marker 1 - Trucks Marker 2 - Trucks

Time Gap 3s 0s 3.5s 0s
Headway 16ft 16ft 16ft 16ft

8
TABLE 3 Single lane roundabout - Comparison of VISSIM results with Real Data

Observation No. Conflicting Flow (veh/hr) Maximum Entry Flow (veh/hr)

Real Data (veh/hr) VISSIM (veh/hr)
1 120 1020 1250
2 300 852 930
3 480 690 700
4 600 588 550
5 720 480 400
6 900 312 290

9
TABLE 4 Dual lane roundabout - Comparison of VISSIM results with Real Data

Observation No. Conflicting Flow (veh/hr) Maximum Entry Flow (veh/hr)

Real Data (veh/hr) VISSIM (veh/hr)
1 300 1620 1800
2 600 1290 1350
3 900 990 1000
4 1200 750 700
5 1500 552 450
6 1800 372 300

10
TABLE 5a Traffic flows at each intersection

Directional Flows (Veh/hr)

CASE 1 Int 1 Int 2 Int 3
EB Left 200 150 100
EB Thru 850 850 850
EB Right 100 100 150
NB Left 100 150 100
NB Thru 400 600 400
NB Right 150 150 100
SB Left 100 100 100
SB Thru 500 500 500
SB Right 150 150 100
WB Left 100 150 200
WB Thru 700 650 700
WB Right 100 100 150

TABLE 5b Traffic flows at each intersection

Directional Flows (Veh/hr)

CASE 2 Int 1 Int 2 Int 3
EB Left 150 110 88
EB Thru 638 628 618
EB Right 75 88 110
NB Left 75 113 75
NB Thru 300 450 300
NB Right 113 113 75
SB Left 75 75 75
SB Thru 375 375 375
SB Right 113 113 75
WB Left 100 126 150
WB Thru 513 474 525
WB Right 87 75 113

11
TABLE 5c Traffic flows at each intersection

Directional Flows
CASE 3 Int 1 Int 2 Int 3
EB Left 100 75 50
EB Thru 425 425 425
EB Right 50 50 75
NB Left 50 75 50
NB Thru 200 s300 200
NB Right 75 75 50
SB Left 50 50 50
SB Thru 250 250 250
SB Right 75 75 50
WB Left 75 50 100
WB Thru 350 325 350
WB Right 50 75 75

12
TABLE 6 Comparison of traffic performance

Average Delay (sec/veh) Average Queue (ft)

VISSIM-Signalized VISSIM-Signalized
Intersection VISSIM-Roundabout Intersection VISSIM-Roundabout
CASE 1 35 42 53 72
CASE 2 28 24 18 15
CASE 3 27 25 28 23

13
FIGURE 1 (a) Layout of a Single Lane Roundabout in VISSIM, (b) Layout of a Dual Lane
Roundabout in VISSIM.

14
 Relevant Time gap and Headway

Conflict Marker

Stop Line

Roundabout Entry

FIGURE 2 Definition of Priority Rules.

15
 Conflict Markers
2
1

Stop Line

FIGURE 3 Single Lane Roundabout – Priority rules in VISSIM.

16
 3

2 1

FIGURE 4 Dual Lane Roundabout – Priority rules in VISSIM.

17
 Roundabout Capacity Analysis
Comparison of VISSIM, RODEL, and aaSIDRA

2000

1800
Maximum Entry Capacity (veh/hr)

1600

1400

1200

1000

800

600

400

200

0
0 200 400 600 800 1000 1200 1400 1600 1800
Conflicting Flow (veh/hr)
VISSIM aaSIDRA RODEL
Poly. (VISSIM) Poly. (aaSIDRA) Linear (RODEL)

FIGURE 5 Single Lane Roundabout Capacity Analysis.

18
 Roundabout Capacity Analysis
Comparison of VISSIM, RODEL, and aaSIDRA

3000
Roundabout Entry Capacity

2500

2000
(veh/hr)

1500

1000

500

0
0 500 1000 1500 2000
Conflicting Flow (veh/hr)

VISSIM_3s aaSIDRA
VISSIM_2.5s Rodel
Best Fit (VISSIM_3s) Best Fit (aaSIDRA)
Best Fit (VISSIM_2.5s) Best Fit (Rodel)

FIGURE 6 Dual Lane Roundabout Capacity Analysis.

19
FIGURE 7 Single Lane Roundabout – Real data (6).

20
FIGURE 8 Dual Lane Roundabout – Real data (6)

21
FIGURE 9 VISSIM screenshot of three coordinated signalized intersections.

22
FIGURE 10 Second signal replaced by a roundabout.

23