Capacity of U-Turn at Median Opening: Ite Journal June 1999

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 6

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/285833065

Capacity of U-turn at median opening

Article  in  Ite Journal · June 1999

CITATIONS READS
29 647

1 author:

H. R Al-Masaeid
Jordan University of Science and Technology
74 PUBLICATIONS   575 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Currently, I am working on geometric consistency of interchange elements View project

All content following this page was uploaded by H. R Al-Masaeid on 09 April 2016.

The user has requested enhancement of the downloaded file.


Capacity of U-Turn at Median Openings

EMPIRICAL AND INTERSECTIONS ARE PROVIDED BACKGROUND


to facilitate traffic turning movements. As As mentioned before, traffic operation
GAP-ACCEPTANCE a part of traffic management to improve at U-turn median openings has not yet
intersection operation, some traffic move- been addressed. However, the operation
APPROACHES WERE ments are not permitted at some intersec- can be considered as an interaction of
tion locations, especially along divided drivers on the minor or stop-controlled
USED IN A STUDY IN arterials.1 In most cases, such minor turn with drivers on the oncoming
movements are accommodated at sepa- approach of the major street. Although
JORDAN TO HELP rate U-turn median openings. Compared the U-turn movement is more complex
with turning movements at intersections, than right- or left-turning movements at
IDENTIFY SPECIFIC U-turn movement at median openings is unsignalized intersections, the general
highly complex and risky. Normally, the concepts and procedures developed for
GUIDELINES FOR speed of conflicting traffic stream is rela- analyzing capacity at priority unsignalized
tively high and the turning vehicle must intersections are very crucial in this
ESTIMATING CAPACITY wait and then turn under low speed level. respect. Using the empirical approach,
Therefore, the turning vehicle needs a Kimber and Coombe3 concluded that the
AND DELAY OF U-TURN large gap in the conflicting stream before capacities of the nonpriority streams at T-
performing the U-turn. intersections depend linearly on the flow
MOVEMENT AT MEDIAN One of the most important tasks of a in the relevant priority streams. Further
traffic engineer is to evaluate traffic oper- empirical studies were conducted in the
OPENINGS. ation of turning movements. In fact, the United States4 and Poland5 to estimate
1994 Highway Capacity Manual 2 capacities of the nonpriority streams at
(HCM), which contains procedures and two-way stop-controlled intersections.
models for estimating capacity and delay The developed regression models indi-
for different movements at unsignalized cated that the capacity of the nonpriority
intersections, does not provide specific stream depends linearly on the traffic flow
guidelines for estimating capacity and on the conflicting priority approach.
delay of U-turn movement at median On the other hand, Chapter 10 of the
openings. For this reason, an effort was 1994 HCM contains a procedure for the
made to estimate capacity of U-turn capacity and level of service analysis of
movement at median openings. unsignalized intersections. The procedure
In this study, both empirical and gap- relies on a gap-acceptance model devel-
acceptance approaches were used to esti- oped and refined in Germany.6 According
mate capacity of U-turn movement at to this procedure, the capacity of each
median openings of divided arterials. nonpriority stream is estimated using
The empirical approach using regression Siegloch’s formula,7,8 which follows:
analysis was adopted to estimate capacity
C = (3,600/tm) × e
–p[tc – (tm/2)]
of U-turn movement and investigate the (1)
effect of different rel-
BY HASHEM R. AL-MASAEID evant factors that where:
might affect the esti-
mated capacity. Parameters of the gap- C = capacity of nonpriority stream [vehicles/hour
acceptance approach, including critical (veh/h)];
gap and move-up time, were estimated p = q/3,600;
and used to calculate capacity on the q = traffic flow of major or conflicting stream;
basis of the 1994 HCM. The results of tc = critical gap [seconds (s)];
both approaches were compared and tm = move-up time (s); and
presented in this study. e = base of the natural logarithm.

28 ITE JOURNAL / JUNE 1999


Compared with the performance of
the empirical approach, several studies Table 1. Characteristics of the collected data.
reported that the theoretical approach Number of Standard
using the gap-acceptance concept overes- Variable observations Mean deviation Minimum Maximum
timates the capacity of nonpriority
First data set:
streams. For example, Kimber9 concluded
Conflicting traffic flow 210 1,127 575.0 150.0 2,340
that the gap-acceptance models are poor
Capacity of U-turn 210 450 187.2 90.0 900
predictors of the capacity of nonpriority
Average total delay 210 19.1 10.3 4.0 43.0
streams in the United Kingdom. Also, Al-
Masaeid10 indicated that the gap-accep-
Second data set:
tance models would provide unrealistic
Speed of conflicting traffic 68 50.1 6.9 38.0 62.0
capacity values even if critical gaps and
Average total delay 68 21.2 10.96 6.0 47.0
move-up times were estimated for local
Critical gap 68 6.72 0.87 4.55 7.9
conditions in Jordan. However, these
Move-up time 68 2 0.52 1.70 3.4
studies did not consider the possible effect
of delay variable on the estimated capaci-
ties or the size of the critical gaps. was collected to establish critical gap and minute intervals. The gap measurements
move-up time models for turning drivers. were used to estimate critical gap. Similar
METHODOLOGY The first set included data on capacity to the HCM definition, the critical gap is
To accomplish the objective of this of turning stream, oncoming traffic flow defined as the median gap size accepted by
study, seven median openings located in (conflicting traffic flow), traffic speed on turning drivers. The move-up time repre-
different cities in Jordan, including the approach of the conflicting traffic and sents the minimum headway between
Amman, Zarqa and Mafraq, were average total delay. These data were turning vehicles entering the same gap in
selected. These median openings are observed at one-minute intervals. The the conflicting stream. The average total
located along divided suburban arterials observations were taken with stable queu- delay was computed as explained in the
and operated at capacity during peak ing in the left-turning lane (at-capacity first data set. However, the queue length
turning periods. At-capacity condition is operation). Instead of measuring the total was observed at 30-s intervals. To achieve
defined as the condition in which there delay for individual turning vehicles, the wide variability in the average total delay,
is a continuous queue of turning vehicles queue length was observed at 10-s inter- the data were collected during peak and
in the approach of the turning lane. All vals, and the average queue length during off-peak periods. Furthermore, all selected
turning movements were performed one-minute intervals was calculated. sites were at least 200 meters (m) from the
from a single left-turn lane. Based on queuing theory, the average nearest signalized intersection. Thus, no
In this study, both empirical and gap- total delay was computed as follows: spillback problems or abnormal bunched
acceptance approaches were adopted to arrivals were observed during data collec-
estimate capacity of U-turn at median d = L/λ (2) tion. Table 1 summarizes the characteris-
openings. In the empirical approach, a tics of the collected data.
multiple regression analysis was performed where: The measurements also included the
to develop an empirical relationship for geometric elements of each median open-
estimating capacity and identifying vari- d = average total delay (s/veh); ing. The width of each arterial approach
ables that affect the estimated capacity. L = average queue length (veh); and varied from 8.3 to 9.4 m with median of
While in the gap-acceptance approach λ = arrival rate of turning vehicles (veh/s). 1.8 to 3.0 m. Each selected arterial has
(the 1994 HCM procedure), critical gap four lanes—two through lanes/direction.
and move-up time models were empiri- To account for heavy vehicles, the The width of the left-turn lane is about
cally established. The results of modeling capacity and conflicting traffic flow were 3.0 m. All approaches had a good sight
were incorporated into Siegloch’s formula converted into passenger car units distance with grades less than 2.0 percent.
to calculate the capacity of U-turn. Finally, (PCU). For conversion into PCU, light
performance of both approaches was com- trucks and minibuses were rated as 1.5 DEVELOPMENT OF MODELS
pared and presented in this study. PCU. This rate is considered satisfactory U-Turn Capacity Model
in different studies. 10,11 In Jordan, The major purpose of capacity model-
DATA COLLECTION trucks are not allowed to enter cities. For ing is to develop a useful relationship
For the purpose of this study, two the selected sites, the percentage of light between capacity of U-turn median
independent data sets were collected dur- trucks and minibuses was 2 to 5 percent. opening and a set of traffic and geometric
ing the summer of 1996. The first set was For the second set, a manual technique characteristics. The developed model
collected to develop an empirical rela- was used to collect gap measurements, should be easy for practical applications
tionship for estimating capacity of U- move-up time, speed of conflicting stream and predictive under different traffic
turn at median openings. The second set and average total delay data using 15- conditions. A correlation analysis was

ITE JOURNAL / JUNE 1999 29


undertaken to identify the factors that
1000 affect U-turn capacity. The analysis indi-
cated that U-turn capacity had strong
correlation with the conflicting traffic
flow and the average total delay. For illus-
800 tration, Figure 1 shows the scatter plot of
U-turn capacity and conflicting traffic
flow for all investigated median open-
ings. Also, the conflicting traffic flow had
600 high correlation with the average total
Capacity (pcu/h)

delay. The analysis also showed that the


approach speed and median and conflict-
ing approach widths did not have a
400 strong effect on the U-turn capacity.
Regression analysis was carried out to
determine the best form of the predictive
equation for the U-turn capacity. Since
200 conflicting traffic flow and average total
were found to be strongly correlated, the
average total delay was not included in the
modeling analysis. Based on the analysis,
0 two regression equations were obtained:

0 500 1000 1500 2000 2500 C = 799 – 0.31 qc (3)

Conflicting Traffic Flow (pcu/h) C = 1,545 – 790 eqc/3,600 (4)

Figure 1. Scatter plot of U-turn capacity conflicting traffic flow. where:

50 C = capacity of U-turn movement (PCU/h); and


qc = conflicting traffic flow (PCU/h).

Equations 3 and 4 were found to be


40 significant at a 95 percent confidence
level. Although exponential (N = 210, R 2
= 0.89, F = 1,761 and σ = 61) and linear
(N = 210, R 2 = 0.90, F = 1,981 and σ =
Average Total Delay (s/veh)

58) equations were found to provide a


30
reasonable fit to the empirical data, the
linear form was the best from a statistical
point. Clearly, the linear-regression equa-
tion developed in this study compares
20 favorably with capacity models developed
for nonpriority streams at two-way stop-
controlled intersections in the United
Kingdom,3 United States4 and Poland.5
10 The conflicting traffic flow, qc, in
Equations 3 and 4 represents the oppos-
ing flow. As mentioned before, each arte-
rial direction has two through lanes; and
0 vehicles using these lanes conflict with
the turning vehicle. Thus, it is more log-
0 500 1000 1500 2000 2500
ical to estimate the capacity of U-turn
Conflicting Traffic Flow (pcu/h) movement as a function of conflicting
flow irrespective of the number of the
Figure 2. Scatter plot of average total delay and conflicting traffic flow. through lanes. However, if conflicting

30 ITE JOURNAL / JUNE 1999


flow/lane is needed, the following devel- 1000
oped regression equation can be used:

C = 799 – 0.62 qcp (5)


800
where q cp represents the conflicting
flow/lane. The intercept, the slope and

Estimated Capacity (pcu/h)


the regression model were significant at
95 percent confidence (N = 210, R 2 = 600
0.90, F = 1,981 and σ = 58).
Furthermore, the analysis conducted
in this study revealed that there is a rela-
tionship between the average total delay 400
and the conflicting traffic flow (see Fig-
ure 2). Analysis showed that linear and
exponential forms are suitable to
describe the relationship. However, the 200
exponential form was found to have the
best statistical characteristics. The devel-
oped regression equation was as follows:
0
TD = 6.6 × e qc/1,200
(6) 0 400 800 1200 1600 2000

where TD represents the average total Conflicting Traffic Flow (pcu/h)


delay for the turning vehicles (s/veh). The
relationship was found to be statistically Figure 3. Capacity estimates according to empirical and gap-acceptance approaches.
significant at the 95 percent confident
level (N = 210, R 2 = 0.84, F = 1,066.2 and tc = 6.31 – 0.070 TD + 0.026 sp (7) found to be about 3.5 s. However, real-
σ = 4.16). Al-Omari and Benekohal12 izing the complexity and the risk asso-
developed an exponential model to esti- where sp represents the approach speed ciated with U-turn movement at
mate total delay at congested two-way of the conflicting traffic stream (km/h). median openings, this value is not sur-
stop-controlled intersections. The model All regression parameters in Equation 7 prising. The empirical data indicated
has two input variables: the arrival rate and were significant at a 95 percent confi- that the size of the move-up time
the conflicting traffic volume. Also, Kyte et dence level (N = 68, R 2 = 0.85, F = decreases with the increase in average
al.4 indicated that both conflicting traffic 195.1 and σ = 0.34). Clearly, Equation total delay. Based on regression analy-
flow and arrival rate of subject approach 6 indicates that the critical gap is not sis, the following exponential equation
had linearly influenced the average total constant but varies according to the was obtained:
delay; however, analysis conducted in this average total delay and the speed of
study did not confirm the effect of arrival conflicting traffic flow. This result is tm = 1.76 + 3.77 × e–TD/7.5 (8)
rate of turning traffic on the average total consistent with findings of recent stud-
delay. A higher arrival rate would probably ies, which indicated that the average The regression parameter in Equa-
be needed to explore the effect of arrival total delay does affect the critical gap of tion 8 was significant at a 95 percent
rate on average total delay. nonpriority streams at unsignalized confidence level (N = 68, R 2 = 0.74, F =
intersections. 4,13,14 Moreover, Al- 191.3 and σ = 0.26).
Critical Gap and Move-up Time Models Masaeid 10 and Harders15 concluded Previous studies reported that the size
Multivariate analysis was used to assess that the size of the critical gap was of the move-up time is related to the
the effect of different traffic conditions on influenced by the speed of conflicting type of turning movement and the speed
the critical gap and move-up time. The traffic at unsignalized intersections. of the conflicting traffic at unsignalized
analysis showed that the critical gap is Therefore, the results of this study are intersections.15 Other studies showed
strongly correlated with the average total compatible with the cited literature. that the move-up time is correlated
delay and conflicting traffic speed. Equa- Further analyses were conducted to strongly with the relevant critical gap for
tion 6 indicates the best model resulting develop move-up time models. Field nonpriority streams at unsignalized
from a regression analysis with critical gap observations and analysis performed in intersections.10 However, it is worth
as the dependent variable and the average this study revealed that the move-up mentioning that all previous studies did
total delay and conflicting traffic speed as time is related to delay level. Under not investigate the size of move-up time
the determinate variables. low delay levels, the move-up time was for U-turn movement.

32 ITE JOURNAL / JUNE 1999


PERFORMANCE OF THE GAP- the performance of different capacity-esti- II. Germany: Springer-Verlag, 1991.
ACCEPTANCE APPROACH mation approaches. 8. Brilon, W., N. Wu and K. Lemke.
This study investigated the perfor- “Capacity at Unsignalized Two-Stage Priority
mance of the gap-acceptance approach. CONCLUSIONS Intersections.” Transportation Research Record
Capacity estimates using the empirical The results of both empirical and 1555 (1996): 74–82.
approach were considered as the basis for gap-acceptance approaches were com- 9. Kimber, R. “Gap-Acceptance and Empiri-
comparison. Figure 3 shows the relation- pared. Based on the results of this study, cism in Capacity Prediction.” Transportation Sci-
ship between U-turn capacity and con- the following points were concluded: ence, Vol. 23, No. 2 (1989): 100–111.
flicting traffic flow using the empirical and 1. Capacity and average total delay 10. Al-Masaeid, H.R. “Capacity of One-
gap-acceptance approaches. In the empiri- models for U-turn movements at Way Yield-Controlled Intersections.” Trans-
cal approach, Equation 3 was used to esti- median openings were found to be sig- portation Research Record 1484 (1995): 9–15.
mate capacity values. Siegloch’s formula in nificantly influenced by the conflicting 11. Troutbeck, R.J. “Effect of Heavy Vehi-
Equation 1 was used to estimate the traffic flow. The best predictive capacity cles at Australian Traffic Circles and Unsignal-
capacity according to the gap-acceptance model had a linear form, while the delay ized Intersections.” Transportation Research
approach. For each conflicting traffic-flow relationship had an exponential form. Record 1398 (1993): 54–60.
level, Equation 5 was used to obtain the 2. Both the conflicting traffic speed 12. Al-Omari, B., and R. Benekohal.
estimate of average total delay. The esti- and average total delay were found to be “Delay at Congested Unsignalized Intersec-
mated delay values were substituted in significant for estimating the critical gap tions.” Proceedings of Traffic Congestion and
Equations 7 and 8 to obtain values of crit- for U-turn movement. Also, the results Traffic Safety in the 21st Century. Chicago, Ill.,
ical gap and move-up time, respectively. In indicated that the increase in the average USA, June 8–11, 1997.
Equation 7, the speed of conflicting traffic total delay significantly influenced the 13. Madanat, S.M., M. Cassidy and M.
stream was assumed to be 50 km/h. size of the move-up time. Wang. “Probabilistic Delay Model at Stop-Con-
The empirical approach, as shown in 3. Compared with results of the trolled Intersection.” Journal of Transportation
Figure 3, indicates that the estimated empirical approach, the gap-acceptance Engineering, Vol. 120, No. 1 (1994): 21–36.
capacity is 489 PCU/h at conflicting traf- model provides reasonable results. ■ 14. Teply, S., M. Abou-Henaidy and J.D.
fic flow of 1,000 PCU/h. On the con- Hunt. “Gap Acceptance Behaviour—Aggregate
trary, the gap acceptance model indicates References and Logit Perspectives: Part 1.” Traffic Engi-
that the estimated capacity is 355 1. Hummer, J.E. “Unconventional Left- neering and Control Journal, Vol. 38, No. 9
PCU/h. This difference, 134 PCU/h, in Turn Alternatives for Urban and Suburban (1997): 474–482.
the estimated capacity is relatively large. Arterials—Part One.” ITE Journal (September 15. Harders, J. Grenz- und Folgezitlucken als
This difference is about 2.3 of the stan- 1998): 26–34. Grundlage fur die Leistungsfahigkeit von Land-
dard errors (58 PCU/h) for capacity esti- 2. Transportation Research Board. Highway strassen. Heft 216. Bonn, Germany: Schriftenreihe
mates according to Equation 3. However, Capacity Manual. Special Report 209. Wash- Strassenbau und Strassenverkehrstechnik, 1976.
the difference is not significant at a 95 ington, D.C., USA: National Research Coun- 16. Brilon, W., M. Grossmann and B. Stuwe.
percent confidence level. On the other cil, 1994. “Towards a New German Guideline for Capacity
hand, both empirical and gap-acceptance 3. Kimber, R., and R. Coombe. The Traffic of Unsignalized Intersections.” Transportation
approaches provide comparable capacity Capacity of Major/Minor Priority Junctions. Research Record 1320 (1991): 168–174.
values under low or heavy conflicting TRRL Report SR582. Crowthorne, Berkshire,
traffic-flow levels, as shown in Figure 3. England: Department of the Environment and
Thus, the empirical and gap-accep- Transport, 1980.
tance approaches yield approximately sim- 4. Kyte, M., C. Clemon, N. Mahfood, B. HASHEM R.
ilar capacity estimates. This finding is not Lall and C. Khisty. “Capacity and Delay Char- AL-MASAEID,
consistent with conclusions of previous acteristics of Two-Way Stop-Controlled Inter- P.E., is Associate Profes-
capacity studies, which indicated that the sections.” Transportation Research Record 1320 sor of civil engineering
gap-acceptance approach provides unreal- (1991): 160–167. and Vice Dean of scien-
istic capacity values at both low- and high- 5. Tracz, M. “Research on Traffic Perfor- tific research at Jordan
conflicting traffic-flow levels. 9,10,16 mance of Major/Minor Priority Intersections.” University of Science
Previous studies did not consider the delay Intersections without Traffic Signals. Berlin: and Technology
effect on the gap-acceptance performance. Springer-Verlag, 1988. (JUST), in Irbid, Jordan. He received a B.Sc. in
Furthermore, the HCM still treats the 6. Siegloch, W. Die Leistungsermittlung an civil engineering from JUST and an M.Sc. in
gap-acceptance parameters as constant val- Kontenpunkten ohne Lichtsignalsteurung. Heft civil engineering from Yarmouk University, Jor-
ues. This issue is very important because 154. Bonn, Germany: Schriftenreihe Strassen- dan. He also has a Ph.D. in transportation engi-
drivers may change their behavior as they bau und Strassenverkehrstechnik, 1973. neering from Purdue University. His research
wait for a considerable period of time, and 7. Brilon, W., and M. Grossmann. “The New interests are in the areas of traffic safety, traffic
ultimately they accept smaller gaps. Thus, German Guideline for Capacity of Unsignalized engineering and highway engineering.
delay should be considered in comparing Intersections.” Intersection without Traffic Signals Al-Masaeid is a Member of ITE.

34 ITE JOURNAL / JUNE 1999

View publication stats

You might also like