JOURNAL OF TRANSPORTATION

SYSTEMS ENGINEERING AND INFORMATION TECHNOLOGY

Volume 14, Issue 5, October 2014
Online English edition of the Chinese language journal

Cite this article as: J Transpn Sys Eng & IT, 2014, 14(5), 15−22. RESEARCH PAPER

Review and Outlook of Roundabout Capacity

QU Zhaowei, DUAN Yuzhou, SONG Xianmin*, XING Yan
College of Transportation, Jilin University, Changchun 130022, China

Abstract: Capacity is an important part of the operation efficiency at roundabouts. Through analyzing existing capacity models of
roundabout, this paper focuses on the problem research principles and methods such as empirical regression model, gap acceptance
model and model based on simulation software. The key technologies of modeling are also analyzed. Then the general approaches of
modeling roundabout capacity are proposed, combined with some problems easily overlooked in previous study. Determining the
interaction mechanism among each traffic flow and considering the significant impact factors, the capacity model is established on
the whole. Finally, combing with the control objectives of roundabouts, an outlook is proposed for future studies on roundabout
capacity.
Key words: traffic engineering; roundabout; capacity model; gap acceptance theory; critical gap

1 Introduction Four typical capacity models of roundabout are analyzed in

the paper from the perspective of system analysis. The pros
As an important parameter of transportation system, and cons of different models have been systematically
capacity is the basis of the analysis of delay, level of service discussed based on previous study. Then the general
and queue length. It is obvious that capacity is closely related approaches and key factors of modeling roundabout capacity
to traffic status. Roundabouts are received more attention are studied. Finally, an outlook is proposed for future studies
increasingly due to the advantages of accident-decreasing and on roundabout capacity with the control objectives of
efficiency-increasing. According to statistics, the number of roundabouts.
roundabouts in U.S. and Canada has been over 2000 by
2010[1]. Roundabout capacity is the maximum possible 2 The models and methods of roundabout
throughput per unit time under given road geometry and capacity
traffic conditions. As the premise of evaluating operating
In the initial stage of constructing roundabouts, Wardrop[5]
efficiency, the capacity has become a hotspot of roundabout
regarded the maximum throughput in weaving sections as the
research.
capacity of roundabout. With “give way” rule presented in
Unsignalized intersection theory provides a solid
modern roundabout, the entrances are more prone to emerge
foundation for the research of roundabout capacity[2, 3]. With
bottlenecks[6]. Thus the weaving section model is not
the rise of modern roundabouts, research methods of two-way-
applicable any more. In addition, three modeling methods are
stop-controlled (TWSC) intersection and two-way-yield-
discussed in this paper in detail, namely, empirical regression
controlled (TWYC) intersection have been successfully
model, gap acceptance model and simulation-based method.
applied into analyzing operational efficiency of roundabouts[4].
Different methods have their own applicable conditions and
However, driving behaviors have been changed at roundabout
will be elaborated in the following subsections.
due to the differences of geometrical characteristics and traffic
2.1 Weaving theory model
conditions. Moreover, the theoretical transferability must meet
Weaving section is the zone where the traffic bottlenecks
with the traffic flow characteristics highly. Therefore, the
probably occur at the traditional roundabout. Regarding the
method of capacity calculation which adequately considered
maximum possible throughput in this area as the capacity of
the operational characteristics of roundabout needs to be
entire roundabout, Wardrop[5] proposed the weaving theory
proposed.
model (Eq.(1)). However, its application should match some

Received date: Dec 19, 2013; Revised date: Mar 17, 2014; Accepted date: Mar 25, 2014
*Corresponding author. E-mail: songxm@jlu.edu.cn
Copyright © 2013, China Association for Science and Technology. Electronic version published by Elsevier Limited. All rights reserved.
DOI: 10.1016/S1570-6672(13)60141-5
 QU Zhaowei et al. / J Transpn Sys Eng & IT, 2014, 14(5), 15−22

certain conditions, such as D>40m, w=6.1~18.0m (where D is
the diameter of the central island, and w is the width of
weaving section), so it is limited. And the circulating vehicles
have priority over entry ones with the “give way” rule
proposed. Therefore, the UK Department of the Environment
put forward an improved model[7] (Eq.(2)).
e p w Empirical regression model，also known as conflicting
Q  280(1  )(1  ) / (1  )
w 3 l volume model, predicts the capacity by means of establishing
160l ( w  e)
Q 
wl
where, Q is roundabout capacity, e is average width of
approach ( e  (e1  e 2 ) / 2 , m), w is width of weaving
section (m), p is the proportion of weaving traffic in weaving
section (%), l is length of weaving section (m), e1 is width of
access approach (m), and e 2 is circulatory width in front of
splitter island (m).
Based on weaving section, weaving theory model only
applies to medium-to-large-scale roundabout with multilane.
The model mainly considers the static feature and ignores
time-variant characteristics of traffic flow. Because of less
weaving behaviors at modern roundabout caused by the
reduction of central island’s diameter, the weaving theory
model may be not suitable for current operational mode[8].
2.2 Empirical regression model
(1)
the regression equation between entry capacity and circulating
(2)
volume. The prediction is significant under saturated flow
condition, besides this methodology could take into account
the “pseudo conflict” caused by exiting vehicles[9]. Many
countries use empirical regression model, such as UK,
Switzerland, Germany and France[6, 10, 11]. In addition, Federal
Highway Administration (FHWA) [12, 13] also proposed this
type of model. In summary, some typical models are listed in
Table 1.

Table 1 Some typical empirical regression models

Country Model
UK（Kimber） Ce  F  fcQc fc  0.29  0.116e F  329e  35u  2.4D  135
， ，
1 8 1 2 Qexit Qexit
Switzerland（Bovy） Ce  [1 500  ( )(  Qc   Qexit )]  = 0 (1  ) and Qt  Qc  Qexit
 9 ，
3 3 Cexit Qt
 BQc
Germany（Stuwe） Ce  Ae10 000
U.S.（FHWA） Ce  1 218  0.74Qc
A  BQc
Jordan Ce  e
10 000
NOTE: Ce is entry capacity, Qc is conflicting volume, Qexit is exiting volume. In Switzerland’s model：γ is effect of number of entry lanes: one lane=1, double
lane=[0.6,0.7]. β is effect of number of circulatory lanes: one lane=[0.9,1.0], double lane =[0.6,0.8]. α is effect of exiting vehicels;  0 is mid-value of α. A, B is
intercept and slope constants, respectively.

Considering the effects of geometry, some scholars have put
forward the improved regression models. Polus et al.[14]
introduced the diameter into regression model (Eq.(3)), but the
data were collected at the small-to-medium-scale single-lane
roundabout, without taking the number of lanes into account.
Through analyzing the integrated effects of circulating flow,
lane width, central island’s diameter and distance between
entrance and exit, Al-masaeid et al.[11] also established a
capacity model (Eq.(4)). And the prediction values are similar
to other models’ under low volume conditions.
Ce  394D0.31e(0.000 95QC )
Ce  168.2D 0.312
S 0.219 0.071EW  0.019 RW
e e
(4)
where, D is diameter of central island, S is distance between
entrance and exit, EW is entry width, and RW is the circulatory
width.
Furthermore, Wei et al.[15] proposed the streamline process
to estimate capacity via video data. Due to the difficulties in
capturing and extracting videos at multi-lane roundabout, this
method is limited to single-lane roundabout. Al-Madani[9] put
forward an empirical model aiming at heavy demand
conditions and compared with other models. Considering the
influences of slow traffic, Martijn et al.[16] proposed the
improved model based on Bovy’s( Eq.(5) - Eq.(7) ).
Ce  Ce, h Fexit pe (5)
Fexit  1  pblocking  1  ( x N 1  0.14 x)
(6)
 qc ,b t0
pe  e 且t0  tcr ,b  0.5t f ,b
(7)
where, x is virtual V/C ratio, N is space between roundabout
(3) and cycle facility expressed in number of cars, pblocking is
5.602QC /10 000 probability of exiting vehicles blocking the roundabout, qc ,b
is volume of circulating cyclists (bic/s), tcr ,b is critical gap to
cyclists (s), and t f ,b is follow-up headway, Ce is entry
capacity (pcu/h) , Ce, h is entry capacity due to the main
conflicts (pcu/h), Fexit is reduction factor caused by
downstream exit , pe is probability that exit is not blocked
by cyclists.
Above all, the empirical regression model is relatively easy
 QU Zhaowei et al. / J Transpn Sys Eng & IT, 2014, 14(5), 15−22
to be built, and has good applicability to local roundabouts.
Nonetheless, data demands are too large, and the
transferability is poor. Besides, the interactions of traffic
streams are not considered in the model[13]. Therefore, there
are some limitations in practicability, and it is hard to uniform
different models for different types of roundabouts[8].
2.3 Gap acceptance model
The circulating streams have the absolute priority at modern
roundabout, while the entry flows are similar to the
minor-streams at TWSC intersection. Therefore, capacity
model can be derived by means of gap acceptance theory[17].
Such model has systematical macro-structure and
comprehensive theoretical connotation, and therefore it can be
widely approved in modeling roundabout capacity. Driving
behaviors at roundabout have been considered in the model.
The model has some key parameters which can describe
microscopic traffic status, such as critical gap, follow-up time.
Thus, the model can reflect the time-dependent characteristics
of traffic flows and has more flexibility.
2.3.1 Gap acceptance model
Gap acceptance model is mainly used in Australia ，
Denmark and U.S. (Highway Capacity Manual, HCM ).
Assuming that the major-stream obeyed the
shifted-negative-exponential distribution, Tanner[2] derived the
capacity of minor-stream at unsignalized intersection (Eq.(8)).
Then Troutbeck[17] and Akcelik18] assumed that the
major-stream was subjected to Cowan’s M3 distribution, and
they modified the entry capacity model[17, 18] (Eq.(9), Eq.(10)).
In connection with the number of lanes of major road,
Hagring[3] put forward a general formula for capacity
estimation based on Tanner’s (Eq.(11)). Moreover, the
calibration procedure was built in HCM 2000 and HCM
2010[19], and capacity models of single-lane roundabout were
shown as Eq.(12) and Eq.(13), respectively. Additionally,
Akçelik[20] also revised the model in aaSIDRA many times.
3 600  qc  (1    qc )  e qc (T  )
Ce 
1  e qc T0
3 600  qc    e  (T  )
Ce 
1  e  T0
3 600(1    qc  0.5    qc  t f )  e  (T  )
Ce 
tf
e
 T
q k k k
Ce   i i i  (1
i e   T )
m m 0m
3 600  qc  e qcT
Ce 
1  eqcT0
T  0.5T
3 600  ( 3 600 0 ) qc
Ce  e (13)
T0
where, qc is traffic flow rate in major road (veh/h), △ is
minimun headway (s), T is ciritical gap (s), T0 is follow-up
time (s), α is proportion of free vehicles,   i i ，and  is
decay constant; k ( m ) is decay constant of lane k(m), Tk is
ciritical gap of lane k (s), T0m is follow-up time of lane m (s).
Other parameters are shown as above.
As a vital parameter in gap acceptance model, critical gap
directly determines the precision of the model. Hence, critical
gap has been estimated by many scholars[21-28]. Estimating the
critical gap with Logit model and modeling the follow-up time,
Al-Masaeid[29] compared the capacity models between
Australia and Germany. Considering the effect of the
proportion of free vehicles and combining the Troutbeck’s
improved formula, Tanyel et al.[30] discussed the capacity
based on Cowan’s M3 distribution. He also found that the
method in HCM 2000 could provide a first estimate of
capacity. When the upper and lower boundaries are
determined, it is effective to single-lane roundabout[31, 32].
Furthermore, the estimated procedure in NCHRP Report 572
was used in HCM 2010 (Eq.(14)). The field data were applied
into verifying the calibration procedure in HCM 2010 by Wei
et al[33]. The results show that the modified model is consistent
with the real data when the flow rate is less than 800 vph,
whereas it would overestimate capacity if more than 800 vph.
Ce  A  exp( B  qc )
T0
T （14）
3 600 2
A B
T0 3 600
There are also some domestic studies on gap acceptance
model at roundabouts. Wang et al.[34] proposed a capacity
model with right-turning rate, applied queuing theory to revise
the effects of slow traffic, and derived the practical capacity
model. Considering the influence of different lanes, Xiang
et al.[8] put forward a model for double-lane roundabout
(Eq.(15)), however it could only be applied to
low-channelized one. Assuming that there was an equilibrium
state when the total entry volume reaches to total capacity, Bo
et al.[35] estimated the capacity by stepwise iteration, whereas
ignored the differences among different lanes. By mean of
(8)
revised Raff’s method, Guo[36] also systematically analyzed
roundabout capacity under different conditions based on gap
(9)
acceptance theory.
qe  (T  )  2q1q2   2 q2 e 2 (T  ) (15)
Ce  3 600 1     3 600
(10) 1  e T0  q1  q2  1  e 2T0
2.3.2 Certain influencing factors
(11) Besides critical gap and circulating volume, there are still
some factors influencing entry capacity, e.g., limited priority,
exiting vehicles or heavy vehicle percentage. Limited priority
(12) is defined as the collaborative adjustment of major vehicle to
meet the minor vehicle crossing the intersection. Assuming
that the headway of major-stream obeyed Cowan’s M3
distribution, Troutbeck et al.[37] derived the capacity model
under limited priority conditions (i.e., Eq.(16) and Eq.(17)).
On this basis, Bunker and Troutbeck[38] derived the delay
 QU Zhaowei et al. / J Transpn Sys Eng & IT, 2014, 14(5), 15−22
formula with queuing theory.
 qc Ce   (T  )
Ce 
1 e  T0
1  e T0
C
[1  e   (T  )
  (T  T0  )e  (T  ) ]
where, C is limited priority term.
Louah held the view that exiting vehicles would have an
effect on entry capacity if the distance between entrance and
exit was less than 15 m. Bovy indicated that the influence was
decided by the distance between entrance and exit and the
number of exiting vehicles[39]. Hagring[39] found that the
capacity would decrease with the increase of exiting vehicles,
and took the proportion of exiting vehicles into capacity
model. Then he also discussed the capacity of double-lane
roundabout with two distributions[40], i.e. M31 and M32.
Additionally, Mereszczak et al.[41] analyzed the effects of
exiting vehicle ratios and the length of splitter island on
capacity. Considering the impact of truck at approach,
Dahl[42,43] re-estimated the critical gaps and follow-up times,
which made the modified capacity model closer to reality.
Moreover, Lee[44] considered the effect of truck in the
circulatory lanes and improved the prediction accuracy in
further. He found that heavy vehicle percentage can influence
the estimated value of capacity.
In summary, the gap acceptance model is used widely, and
can integrate the impacts of certain conditions into model by
modifying some relational parameters, which is more close to
reality. However, the un-consistency of critical gaps has
significant influence on prediction accuracy. The estimation
procedure ignores the effect of geometry, which will bring
about some complication under some conditions.
2.4 Simulation-based method
The analysis tools for roundabout capacity are divided into
two categories. The former is based on calculation models,
such as aaSIDRA and RODEL, in which roundabout is
individual. The latter are some software based on
micro-simulation models, such as PARAMICS, SimTraffic
and VISSIM. They can be used to simulate complex
intersections, and the result will be more realistic and reliable.
Through analyzing single-lane and double-lane roundabouts
with VISSIM and SIDRA, Bared et al.[45] indicated that the
capacity measured from VISSIM is smaller than SIDRA’s.
Based on various scenarios designed with VISSIM and
comparison with the data of NCHRP 572, they also derived
the lane-based capacity models of double-lane and triple-lane
roundabout (i.e., Eq. (18) and Eq. (19)). Through a
simulation-based sensibility analysis, Li et al.[46] proposed a
revised model with the correlation coefficient between
influencing factors and capacity.
 (7.207 9 
1.300 8*c1 1.294 0*c2
 )
 EL  e 1 000 1 000
, R 2  0.960
（18）
（16）  0.925 9*c1 1.012 0*c2
 E  e(7.207 9  1 000  1 000 ) , R 2  0.987
 R
 1.186 4*c1 1.081 3*c2 0.947 9*c3
（17） (7.075 4    )
 EL  e 1 000 1 000 1 000
, R 2  0.955
 0.675 8*c1 1.155 6*c2 0.904 9*c3
 (7.075 4    )
（19）
 EM  e 1 000 1 000 1 000
, R 2  0.980
 0.556 9*c1 0.904 4*c2 1.025 8*c3
 E  e(7.075 4  1 000  1 000  1 000  0.279 5*Rt ) , R 2  0.955
 R
where, EL is capacity of left-turning lane (vph), EM is capacity
of middle lane (vph), ER is capacity of right-turning lane (vph),
c1 is circulating volume of inner lane (vph), c2 is circulating
volume of middle lane, c3 is circulating volume of outer lane,
and Rt is the proportion of right-turning vehicles in the total
entry volume.
The simulation-based method can reflect the operation
characteristics of traffic streams as a whole. Additionally, it
can connect the roundabout with neighborhood roads, which is
more suitable for reality of transportation system. However,
the calibration process is complex, which has a great impact
on calculation precision.
2.5 Other methodologies
Besides traditional methods, some scholars estimated the
capacities of modern roundabout and special roundabout in
other ways. Using the fundamental diagram, Dixit et al.[47]
modeled the relations among volume, speed and density at
roundabout, and obtained the maximum volume as capacity
based on VISSIM. Considering the priority rank of traffic
streams, Wu[48] put forward a general procedure for estimating
capacity. He applied the queuing theory to estimate the
different zone of major road, and obtained the capacity of
minor road. By modeling the critical gaps and follow-up times
of Type2/1 and Type1/1, Lindenmann[49] proposed the capacity
of Siegloch’s type of roundabout. Results showed that the
capacity of Type 2/1 increased by 20%-30% to Type 1/1'
under the same spatial condition. Besides, through introducing
the pseudo-conflict into capacity estimation, Fortujin[50] built
the general capacity model for turbo roundabout with gap
acceptance theory.
Moreover, based on the calculation model of HCM 2000,
Wang et al[51] simulated the unsignalized intersections of 422
type and determined the setting criteria of TWSC intersection,
AWSC intersection and roundabout. Guo et al[52] proposed the
iteration method for capacity on the basis of estimation
method of entry capacity with gap acceptance theory.
2.6 Comparisons of different methods
Scholars at home and abroad have analyzed and compared
these models in different aspects. Fisk[53] deemed that
calibrating the regression model needed a large amount of data
and the gap acceptance model only needed to change the
critical gap according to different vehicle classes. Therefore,
 QU Zhaowei et al. / J Transpn Sys Eng & IT, 2014, 14(5), 15−22
the gap acceptance model has better generality. However,
Al-Masaeid proposed that the gap acceptance model
overestimated the capacity. Stuwe[10] thought that gap
acceptance theory would fail when the follow-up time was
larger than critical gap, because it was difficult to define the
major or minor road. Çalişkanelli et al.[54] proposed that the
gap acceptance model with limited priority was more realistic
than other models, whereas the effects of Ashworth’s method
and regression method were poor.
Summary for the three types of methods is as follows:
(1) Calibrating empirical regression model needs massive
field data, which is hard to meet. However, regression model
has good applicability to local roundabouts.
(2) It is time-consuming to calibrate the simulation
parameters, while the model has spatiotemporal limitation.
However, the simulation-based method has some significant
advantage. Randomness is close to reality and the operational
patterns of roundabout can be grasped overall.
(3) The gap acceptance model has strong theoretical
property, but it oversimplifies the actual situation. Therefore,
its application is restricted in some certain conditions, i.e.
priority reversal or limited priority. However, through
changing the assumed conditions and calibrating the
parameters of driving behaviors, people can extend the
application of the gap acceptance model.
3 Discussions
3.1 Capacity modeling idea
Now main methods of calculating roundabout capacity
mostly start with single approach. However, roundabout has
some remarkable characteristics apart from traditional
intersections. The circulating streams are merged with
different entry streams and the interactions increase the
analytical complexity. Some unsignalized theories may fail
due to the original relations between the two stream groups, so
capacity model should be established based on making clear
the operational characteristics of traffic flows.
Associated with engineering practice, roundabout capacity
prefers the corresponding maximum output on the best
operational performance. According to cybernetics, the
capacity is the optimal solution of the state function when the
OD demands can make the result fit optimal objects. However,
each approach is analyzed independently in current capacity
models, which causes that the relevance and coordination of
different approaches are ignored. Hence, the capacity is not
derived from a systematic perspective. Besides, exiting
vehicles also have effects on entry capacity under certain
conditions[39]. We need think more about the coordination of
the lanes of different approaches, and then obtain the
roundabout capacity from a systemic perspective.
Consequently, roundabout capacity should be made overall
analysis based on grasping the operational characteristics of
traffic flows during modeling process. Considering major
influencing factors together, people can obtain a more realistic
capacity model.
3.2 Capacity modeling method
The regression model depends on large amounts of field
data. By comparison, the gap acceptance model has systematic
and comprehensive theoretical foundation. The parameters can
be revised in time according to actual situations, which is easy
to be used. However, traditional gap acceptance theory is hard
to properly deal with some certain conditions. After building
the impact model of parameters, the capacity should be
corrected by adjusting some parameters. Meanwhile, the
simulation-based method can be used to verify the capacity
because of the randomness, repeatability and the whole
effects.
Therefore, roundabout capacity should be analyzed by
using the modified gap acceptance model together with
simulation-based method. Note that: (1) the model must be
based on the traffic flow characteristics. And (2) simulation
parameters should be calibrated accurately. In summary,
developing the advantages of the two methods and integrating
their own features, people can derive a more realistic model in
the future.
3.3 Influencing factors
Recently, signals have been set at roundabout as an
effective controlled measure [20, 55-57]. However, the majority of
researches are focused on unsignalized roundabout.
Correspondingly, there is less study on signalized
roundabout. The performance character of traffic flow will be
changed because of signal setting, so gap acceptance theory
will fail. The capacity model of signalized roundabout can be
put forward based on the analysis of the performance
characteristics of traffic flow and the combination of the phase
sequence of signals.
In addition, in order to make the model more realistic, some
influencing factors, such as limited priority, exiting vehicles
and heavy vehicle percentage, should be taken into account in
the modeling process.
4 Outlook
Current researches have provided the foundation for
roundabout capacity both in theory and practice. However, the
diversity of theoretical methods and the specificity of
roundabout’s geometrical design make each model have their
own typical characteristics. In order to provide an effective
reference and precise parameters for traffic control and
management, there are four proposals in this paper, which are
elaborated as follows.
(1) The objective of modeling roundabout capacity is to
optimize the transportation system. The study should be
discussed from the perspective of entirety and systemic. Entry
 QU Zhaowei et al. / J Transpn Sys Eng & IT, 2014, 14(5), 15−22
streams and circulating streams should have close connection.
A coordinated roundabout capacity model will be established
on the premise of making clear the mechanisms between the
two. The model also can provide references for traffic control
and optimization of road design in engineering practice.
(2) Combining the macroscopic theories with vehicular
microscopic characteristics, an integrated capacity model
linked macrocosm and microcosm could be derived, which
should not be limited to gap acceptance theory.
(3) Considering the mixed traffic in China, the effects of
slow traffic should be studied. Taking into account the
influence of pedestrians and non-motorized vehicles, some
proper adjustments should be made in order to make the
model fit reality.
(4) By initially exploring the criteria of signal setting and
studying the capacity of signalized roundabout, the capacity
model of signalized roundabout would be proposed. We can
analyze and compare the capacity of roundabout in different
control modes. Through solving the function which the
maximum capacity is optimization objective, we can obtain
the critical volume for signal setting, which will provide a
reference for signal setting at roundabouts.
5 Conclusions
The paper makes analysis and discussion on existing
capacity models of roundabout. On the basis, the modeling
ideas and methods of roundabout capacity are proposed in
detail. Combining the domestic traffic characteristics, some
influencing factors and notices are elaborated. Finally,
combining with the control objects of roundabout, an outlook
is proposed for further studies on roundabout capacity.
Acknowledgements
This research was funded by the National Natural Science
Foundation of China (Nos. 51278520, 51278220).
References
[1] Pochowski A L. An analytical review of statewide roundabout
programs and policies. Civil and Environmental
Engineering ,Georgia Institute of Technology. 2010.
[2] Tanner J C. The capacity of an uncontrolled intersection.
Biometrika. 1967, 54(3-4):657-658.
[3] Hagring O. A further generalization of Tanner’s formula.
Transportation Research Part B: Methodological, 1998, 32(6):
423-429.
[4] Blogg M, Wemple E, Myers E, et al. Appendixes to NCHRP
report 572: Roundabouts in the United States. Transportation
Research Board, 2006.
[5] Wardrop J G. The traffic capacity of weaving sections of
roundabouts. In: Proceedings of the First International
Conference on Operational Research, Oxford English
University Press, 1957.
[6] Kimber R M. The traffic capacity of roundabouts. Crowthorne,
Berkshire, England: U.K. Transport and Road Research
Laboratory, 1980.
[7] Wang D H. Transportation system analysis. China
Communications Press, 2007.
[8] Xiang Q J, Wang W, Chen B, et al. A study of the model
construction for roundabout's capacity. China Journal of
Highway and Transport, 1999, 12(4):69-72. (Chinese)
[9] Al-Madani HMN. Capacity of large dual and triple-lanes
roundabouts during heavy demand conditions. Arabian Journal
for Science and Engineering. 2012, 38(3):491-505.
[10] Stuwe B. Capacity and safety of roundabouts-German results.
Intersections without Traffic Signals II. Springer Berlin
Heidelberg, 1991: 1-12.
[11] Al-Masaeid H R, Faddah M Z. Capacity of roundabouts in
Jordan. Transportation Research Record: Journal of the
Transportation Research Board,1997,1572:76-85.
[12] Robinson B W, Rodegerdts L A. Capacity and performance of
roundabouts: A summary of recommendations in the FHWA
roundabout guide. In: Fourth Int. Symp. on Highway Capacity.
Washington, DC: Transportation Research Board National
Research Council, 2000: 422-433.
[13] Robinson B W, Rodegerdts L, Scarborough W, et al.
Roundabouts: An informational guide. 2000.
[14] Polus A, Shmueli S. Analysis and evaluation of the capacity of
roundabouts. Transportation Research Record: Journal of the
Transportation Research Board, 1997, 1572(1): 99-104.
[15] Wei T, Jeromy L G, Hardik R S. Developing capacity models for
local roundabouts: a streamlined process. Transportation
Research Record: Journal of the Transportation Research Board,
2011, 2257:1-9.
[16] Martijn A M de Leeuw, Hein Botma, Piet H L Bovy. Capacity of
single-lane roundabouts with slow traffic. Transportation
Research Record: Journal of the Transportation Research Board,
1999, 1678: 55-63.
[17] Troutbeck R J. Unsignalized intersections and roundabouts in
Australia: recent developments. In: Intersections Without Traffic
Signals II, Bochum, Germany, 1991: 238-257.
[18] Akcelik R, Chung E, Besley M. Roundabouts: capacity and
performance analysis. 1998.
[19] Transportation research board. Highway Capacity Manual.
National Research Council, Washington, DC, 2000.
[20] Akçelik R. Roundabout metering signals: capacity, performance
and timing. Procedia-Social and Behavioral Sciences, 2011, 16:
686-696.
[21] Polus A. The effect of waiting times on critical gaps and an
improved entry-capacity model at roundabouts. In:81th Annual
Meeting of the Transportation Research Board, Washington, DC.
2002.
[22] Polus A, Lazar S S, Livneh M. Critical gap as a function of
waiting time in determining roundabout capacity. Journal of
 QU Zhaowei et al. / J Transpn Sys Eng & IT, 2014, 14(5), 15−22
Transportation Engineering, 2003, 129(5): 504-509.
[23] Khatib Z K. critical gap and follow-up time at multilane
roundabout case study. In: 83th Annual Meeting of the
Transportation Research Board, Washington, DC. 2004.
[24] Xu F. Driver behavior and gap acceptance studies at roundabouts.
University of Nevada, Reno, 2007.
[25] Xu F, Tian Z Z. Driver behavior and gap-acceptance
characteristics at roundabouts in California. Transportation
Research Record: Journal of the Transportation Research Board,
2008, 2071(1): 117-124.
[26] Abrams D S, Fitzpatrick C D, Tang Y, et al. A spatial and
temporal analysis of driver gap acceptance behavior at modern
roundabouts. In: 92th Annual Meeting of the Transportation
Research Board, Washington, DC. 2013.
[27] Guo R. Estimating critical gap of roundabouts by different
methods. In: 6th Advanced Forum on Transportation of China,
2010, 84-89.
[28] Hainen A M. Roundabout critical headway measurement based
on high-resolution event-based data from wireless
magnetometers. In: 92th Annual Meeting of the Transportation
Research Board, Washington, DC. 2013.
[29] Al-Masaeid H R. Capacity and performance of roundabouts.
Canadian Journal of Civil Engineering, 1999, 26(5): 597-605.
[30] Tanyel S, Yayla N. A discussion on the parameters of Cowan M3
distribution for Turkey. Transportation Research Part A: Policy
and Practice, 2003, 37(2): 129-143.
[31] Tanyel S, Baran T, Özuysal M. Determining the capacity of
single-lane roundabouts in Izmir, Turkey. Journal of
Transportation Engineering, 2005, 131(12): 953-956.
[32] Tanyel S, Baran T, Özuysal M. Applicability of various capacity
models for single-lane roundabouts in Izmir, Turkey. Journal of
Transportation Engineering, 2007, 133(12): 647-653.
[33] Ting Wei, Jeromy L Grenard. Calibration and validation of
highway capacity manual 2010 capacity model for single-lane
roundabouts. Transportation Research Record: Journal of the
Transportation Research Board. 2012, 2286:105-110.
[34] Wang W, Xu J Q. The study of new method of analyzing the
roundabout capacity-The application of the queuing theory inn
the roundabout analysis. Journal of Nanjing Institute of
Technology, 1987, 17(1):119-28.
[35] Bo C Y, Cui H L, Wang W, et al. A study on service capacity of
roundabout. Journal of Huazhong University of Science &
Technology (Natural Science Edition), 2002, 30(5):47-49.
[36] Guo R J. Study on capacity of roundabouts based on gap
acceptance theory. Beijing Jiaotong University. 2013.
[37] Troutbeck R J, Kako S. Limited priority merge at unsignalized
intersections. Transportation Research Part A: Policy and
Practice, 1999, 33(3): 291-304.
[38] Bunker J, Troutbeck R. Prediction of minor stream delays at a
limited priority freeway merge. Transportation Research Part B:
Methodological, 2003, 37(8): 719-735.
[39] Hagring O. Derivation of capacity equation for roundabout entry
with mixed circulating and exiting flows. Transportation
Research Record: Journal of the Transportation Research Board,
2001, 1776(1): 91-99.
[40] Hagring O, Rouphail N M, Sørensen H A. Comparison of
capacity models for two-lane roundabouts. Transportation
Research Record: Journal of the Transportation Research Board,
2003, 1852(1): 114-123.
[41] Mereszczak Y, Dixon M, Kyte M, et al. Including exiting
vehicles in capacity estimation at single-lane US roundabouts.
Transportation Research Record: Journal of the Transportation
Research Board, 2006, 1988(1): 23-30.
[42] Dahl J. Capacity estimation for roundabouts with high truck
volume using gap acceptance theory. Masters Abstracts
International. Japan Association for Earthquake Engineering,
2011, 50(04).
[43] Dahl J, Lee C. Empirical estimation of capacity for roundabouts
using adjusted gap-acceptance parameters for trucks.
Transportation Research Record: Journal of the Transportation
Research Board, 2012, 2312(1): 34-45.
[44] Lee C. Prediction of capacity for roundabouts based on
percentages of trucks in entry and circulating flows. In: 92th
Annual Meeting of the Transportation Research Board,
Washington, DC. 2013.
[45] Bared J G, Afshar A M. Using simulation to plan capacity models
by lane for two-and three-lane roundabouts. Transportation
Research Record: Journal of the Transportation Research Board,
2009, 2096(1): 8-15.
[46] Li H, Li J, Yang Z, et al. Micro-simulation study on capacity of
roundabout. In: Intelligent Transportation Systems (ITSC), 14th
International IEEE Conference on IEEE, 2011: 852-857.
[47] Dixit V V, Radwan E. Roundabout capacity: new perspective. In:
92th Annual Meeting of the Transportation Research Board,
Washington, DC. 2008.
[48] Wu N. A universal procedure for capacity determination at
unsignalized (priority-controlled) intersections. Transportation
Research Part B: Methodological, 2001, 35(6): 593-623.
[49] Lindenmann H P. Capacity of small roundabouts with two-lane
entries. Transportation Research Record: Journal of the
Transportation Research Board, 2006, 1988(1): 119-126.
[50] Fortuijn L G H. Turbo roundabouts: estimation of capacity.
Transportation Research Record: Journal of the Transportation
Research Board. 2009, 2130(1):83-92.
[51] Wang H, Wang W, Chen J.Capacity and delay performance of
unsignalized intersection. Journal of Huazhong University of
Science & Technology. ( Nature Science Edition), 2007,
35(1):114-117.
[52] Guo R J, Lin B L, Wang W X. The iterative calculation of full
capacity of roundabouts. In: Information Science and
Engineering (ISISE), 2009 Second International Symposium on.
IEEE, 2009: 570-573.
[53] Fisk C S. Traffic performance analysis at roundabouts.
Transportation Research Part B: Methodological, 1991, 25(2):
 QU Zhaowei et al. / J Transpn Sys Eng & IT, 2014, 14(5), 15−22

89-102. [56] Yang X, Li X, Xue K. A new traffic-signal control for modern

[54] Çalişkanelli P, Özuysal M, Tanyel S, et al. Comparison of roundabouts: method and application. IEEE Transactions on
different capacity models for traffic circles. Transport. 2009, Intelligent Transportation Systems. 2004, 5(4):282-287.
24(4):257-264. [57] Ma W, Liu Y, Head L, et al. Integrated optimization of lane
[55] Davies P, Jamieson B, Reid D A. Traffic signal control of markings and timings for signalized roundabouts.
roundabouts. Traffic Engineering and Control. 1980, Transportation Research Part C: Emerging Technologies. 2013,
21(7):354-357. 36:307-323.