Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

10th International Congress on Advances in Civil Engineering, 17-19 October 2012

Middle East Technical University, Ankara, Turkey

Effect of Dwell Time on Performance of Signalized Intersections

K. Ghasemlou1, M. M. Aydın2, S. Tanyel3, A. Topal4, S. P. Çalışkanelli5


1
Department of Civil Engineering, DEU, Izmir, Turkey, kiarash.ghasemlou@gmail.com
2
The Graduate School of Natural and Applied Sciences, DEU, Izmir, Turkey, metin.mutluaydin@deu.edu.tr
3
Department of Civil Engineering, DEU, Izmir, Turkey, serhan.tanyel@deu.edu.tr
4
Department of Civil Engineering, DEU, Izmir, Turkey, ali.topal@deu.edu.tr
5
Department of Civil Engineering, DEU, Izmir, Turkey, pelin.caliskanelli@deu.edu.tr

Abstract
Bus stops near signalized intersections are known to have a significant impact on the capacity of these
intersections. According to Highway Capacity Manual (HCM) 2010, if there is a bus stop which is close to a
signalized intersection, a correction factor (fbb) is used for calculating the capacity of this signalized intersection.
In the fbb equation of HCM (2010), the constant value 14.4s is a coefficient that was determined according to
dwell time of buses. As known, drivers in different regions have different behavior characteristics. Also dwell
time of buses may vary depending on regional factors and driver behaviors. On the other hand if geometric
factors are added to regional factors, the impact of bus stops on a signalized intersection capacity shows different
values. In this study the impact of bus stops at signalized intersections has been tried to evaluate by using the
traffic flow characteristics and geometric factors of bus stops. Bus Blockage factor (fbb) is obtained by using data
from 2 signalized intersection in İzmir, Turkey. To obtain the capacity of signalized intersections, a new
equation for fbb parameters has been studied to create by using bus station correction equation theory. This study
tried to determine the effect of bus stops near signalized intersections. The results showed that when calculating
the capacity of these types on signalized intersections, it should consider the position of bus stop, green phase of
signal time, driver behaviors, characteristics of roads and regional factor.

Keywords: bus stop, signalized intersection, highway capacity manual, bus blockage factor

1 Introduction
Various methods have been used to determine operational characteristics of buses (including stopping and
moving) in traffic operations. Operational characteristics may adversely affect capacity and the level of service
of roadways, signalized and unsignalized intersections. The U.S. Highway Capacity Manual 2010 (TRB, 2010)
applies some adjustment factors to account these adverse effects. The adverse effects of buses (including
nonstopping buses) in a traffic stream are considered by converting them to passenger car equivalent (PCE)
numbers. To account for the adverse effects of buses that stop for boarding or alighting passengers (called here
stopping buses or transit buses), a bus blockage adjustment factor fbb is used. In this study operational
performance of urban transit buses are tried to analyze at signalized intersections.

When a transit bus stopped at a bus stop during the green phase, it is expected to cause greater lead gap between
bus and the vehicle in front of bus. In the HCM the operational differences were determined by using fbb between
nonstopping buses and stopping buses. The HCM 2010 (TRB, 2010)procedure does not account for all factors
contributing to the adverse effect of stopping buses on the traffic stream. To model and estimate the bus
blockage effects, the total impact of stopping buses must be considered at signalized intersections.(Rodriquez-
Seda&Benekohal, 2006).

The main difference between stopping buses and moving buses is the length of time the stopping bus blocks the
moving traffic lane to serve passengers (dwell time) and delay is caused by acceleration and
deceleration.(Rodriquez-Seda&Benekohal, 2006). The factor fbb is a function of the number of stopping buses,
ACE2012

the number of lanes in the lane group and average blocking time during green period. However it does not
account for all factors caused by stopping buses. According to Rodriquez-Seda&Benekohal (2006), the
following are some of the factors contributing to bus blockage caused by stopping buses:

(a)Dwell time: It is defined as the amount of time a bus spends while stopped to serve passengers by the HCM.
Dwell time is influenced by passenger demand and loading, bus frequency, fare payment procedure, vehicle type
and onboard circulation.
(b) Facility type or bus lane type.
(c) Bus stop location and loading area
(d) Traffic and bus volumes.

The objective of this study is to develop a new methodology for the estimation of effective bus blockage time
caused by a stopping bus to the other vehicles in the traffic stream.

2 Literature Review
HCM 2010 (TRB, 2010) defines bus blockage factor (fbb) as a factor that “accounts for the impacts of local
transit buses that stop to discharge or pick up passengers at a near-side or far-side bus stop within 250 ft of the
stop line (upstream or downstream).”

According to Rodriquez-Seda&Benekohal (2006), “The 1950 edition of the HCM recognized the operational
difference between nonstopping and stopping buses by suggesting that a stopping bus has the same effect on the
intersection capacity as three to five passenger cars.” However, the bus blockage factor was introduced in the
HCM 1965(TRB, 1965).The HCM 1965 (TRB, 1965)introduced the concept of “local bus factor” for near-side
and far side bus stops with and without curb parking. The HCM 1965 (TRB, 1965) considered the adverse effect
of local transit buses at signalized intersections and provided some adjustment procedures for every bus stop
type (Rodriquez-Seda&Benekohal, 2006). Given procedures by HCM 1965(TRB, 1965) includes charts for each
bus stop type with permitted and prohibited parking. Given the number of buses per hour, the area type (number
of lanes), turning movements and adjustment factor was determined. This procedure applied for situations with
no more than 90 buses in downtown areas (Rodriquez-Seda&Benekohal, 2006).

The HCM 1985 (TRB, 1985) was introduced in the bus blockage factor (fbb). For the calculation of this factor
HCM 1985(TRB, 1985) provided two procedures for local transit buses. For cases where the stopping buses
were not a major factor in intersection performance, fbb was a function of the number of lanes in the lane group
(no more than three lanes) and the number of buses stopping per hour (no more than 40 buses stopping per hour),
as shown below: (Rodriquez-Seda&Benekohal, 2006).

(1)

where;
fbb = local bus blockage adjustment factor,
N = number of lanes in the affected lane group, and
NB = number of buses stopping per hour.

According to HCM 1985 (TRB, 1985), following procedure to quantify the lost green time for cases where
transit buses stopping to load and unload passengers had a major reductive effect on the roadway capacity.

( ) (2)

where;
TL = lost time in 1 h (s);
g/c = ratio of green time to cycle length;
N = number of stopping buses in 1 h;
D = average dwell time (s); and
L = additional lost time while starting, stopping, and queuing (L= 6 to 8 s, assuming average conditions).

2
ACE2012

HCM calculated PCE values from the following relationship based on the lost time per hour;

( ) (3)

where;
Eb = passenger car equivalent for a stopping bus and h= 2 s per car.

The PCEs for stopping buses were converted into fbb in the same manner that the PCEs for heavy vehicles were
converted into fhv.

The 1994 update of the HCM (TRB, 1994) and the 1997 update (TRB, 1997) adopted the fbb procedure provided
in the 1985 edition (TRB, 1985). For cases with no more than 40 buses stopping per hour and no more than three
lanes in the lane group, an average blocking time during green of 14.4 s was assumed. The fbb was computed by
the following equation;

(4)

In general, the HCM 2000 (TRB, 2000) procedures do not consider all factors contributing to the bus effects.
Rodriquez-Seda&Benekohal (2006) determined some of the bus situations and factors not considered in the
HCM procedures as follows:

(a) HCM 2000 (TRB, 2000)considers the differences between nearside and far-side stops, neither of the methods
include this factor in the fbb computations. The bus stop type should be considered in the bus blockage factor to
determine whether different values need to be provided for each bus stop type. In summary, the fbb methods
suggested in the HCM 2000 (TRB, 2000) do not accurately calculate the total adverse impact of urban transit
buses stopping to board and alight passengers at signalized intersections.

(b) HCM 2000 (TRB, 2000)considers that when a transit bus stopped near a signalized intersection, bus fully
blocks just the right lane during dwell time. But observations show that during the dwell time the capacity of the
other lanes could change. The reasons for this situation are the width of lanes, anxiety behaviors of drivers on the
other lanes caused by the stopping bus.

In addition to the situations mentioned above, Rodriquez-Seda&Benekohal (2006) studied the following
situations and the factors used in the calculation of fbb :

(a) To change lane when a bus waits at the bus stop (dwell time), the vehicles are behind the bus.
(b) In case of bus stop has a bay, for bus to come close exactly or partially to the bay and as a result of this
increase in the lane widths adjacent to bus bay and depending on this increase in the lane widths, increase in the
speed of the cars who use that road.

(c) In the case of bus station has a bay for stopping or road is wide, the situation that the cars behind the bus
move without changing lane during the time that the bus is exactly or partially stopping at station.

To determine the effect of urban transit buses, Rodriquez-Seda&Benekohal (2006) suggested delay-based
passenger car equivalents taking into account bus position within the queue, bus arrival time within the cycle,
and the additional delay experienced by vehicles due to vehicles changing lanes to prevent the bus from
stopping. They also discussed the fact that although the HCM recognizes near-side and far-side bus stops, their
research were aimed at near-side bus stops and presented a new method to calculate these effects by using
Equation 5.

∑ ∑ ∑ ( ) (5)
where;
= total additional bus delay experienced by a queue of N vehicles,
= average delay for a car in all passenger car queue of N cars,
= probability of bus in position x,
P = probability of n vehicles queued behind the bus,
Ps = probability of bus arrival/departure case s,

3
ACE2012

N = last vehicle in queue created by the bus,


L = maximum number of vehicles in a queue, and
n = total number of vehicles behind bus.

The concept of delay resulting from a near-side bus stop upstream from a signalized intersection was also
examined by Wong et al. (1998) They suggested a simulation based on approach that used factors such as
distance from bus stop to the traffic signal, the frequency of buses, the traffic volume, bus dwell times, and the
signal settings. They suggested Equation 6 to account for the delay on a single lane approach:

(6)

where;
c = cycle length,
λ = g/c, effective green portion of a cycle,
q = flow rate of traffic (veh/sec),
= q/λ, degree of saturation at the signal,
= q/[(1- Ω )+s], degree of saturation at the bus stop,
L = distance between the bus stop and stop line (meters),
f = bus frequency (vehicles/sec),
O = average dwell time (sec/vehicle),
Ω = f*O, proportion of time when the bus stop is blocked,
g = effective green time (sec), and
= unknown coefficients to be calibrated from simulation

It is also important to note that none of the literature examined the effect of lane width, bus stopping position at
the bus stops and partial blockage during dwell time. This shows an obvious gap in the literature and an
opportunity to address it in the research. This research has attempted to evaluate the influence of these factors.

3 Methodology and assumptions


An average bus blockage time during green phase is assumed as 14.4 s. However 14.4 s doesn’t describe
effective blockage time for different regions. In this study, the other factors affecting the blocking time in the
equation of fbb (suggested by HCM) was evaluated. A new model was developed by adding the effect of dwell
time on the roadway while approaching the intersection in the main part of bus blockage time.

3.1 Generalized Bus Blockage Factor Based On Headway Ratio Method


Many factors affect the Driver’s characteristics. These factors can be name as geography, the level of
development, roadway characteristics, existing traffic signs etc. Because of these factors driver’s reactions show
variations. In this study, the bus effects occurring on the roadway during the bus stop, passenger boarding or
alighting and leaving of the bus from the bus stop, the behaviors of the passenger car drivers were tried to be
examined.

In this new method which is named as Generalized Bus Blockage Factor Based on Headway Ratio Method to
examine the vehicle – bus interactions, the following assumptions were made;

 The stopping bus effects on left or right turning are not considered.
 It is assumed that the busses come in the green phase duration by coincidence.
 The place where the buses stop is accepted as the place which fully indicates the capacity of the
road (as bottleneck).
 It is assumed that the buses pass through the intersection.
 It is assumed that Dwell time has a linear effect on saturated flow.
 It is assumed that acceleration and deceleration of the buses have no effect on the other lanes flow.

It is necessary to examine the saturated flow value of the signalized intersection to find out the effects of the
stopping busses on the capacity. The saturated flow value of the existing roadway while no bus at a bus stop,
( ) is obtained by multiplying the ideal saturated flow value ( ) which is defined by Equation 7 and the other

4
ACE2012

correction factors ( which are showed at HCM 2010 (TRB, 2010) and lane number. Meanwhile, the saturated
flow value while there is no bus at a bus stop can be calculated by equation 9 using headway values in the type
of average time and by adding to the fbb factor which is used to define the effect of busses waiting at bus stop.

̅̅̅̅ (7)

̅ (8)
(9)
(10)

where;
N = Lane number
̅̅̅=Time headway in ideal saturated flow.
̅ = Time interval of saturated flow when there is no bus stop.(sec/veh)
=The other correction factors except bus blockage factor.
= Saturated flow value including bus blockage effect.
= Saturated flow value

Between these two situations (existing and non-existing bus blockage effect) the decrease in the roadway
capacity can be calculated according to time with the equation given below;

̅ (11)

On the other hand, the effective blocking time is obtained by multiplying the average effective blockage time by
the numbers of stopped buses at bus stop during the green time in an hour.

(12)

fbb correction factor given below can be obtained by using Equation 13, Equation 14 and Equation 16.

̅ (13)

̅̅̅̅ (14)

̅̅̅̅ ̅ (15)

̅ (16)

̅
(17)

At Single-lane roadway without bus bay, blocking time is shown in Figure 1. In Figure 1 it is assumed that the
bus decreases its speed before it stops for passenger boarding or alighting, after that, it leaves from the bus stop
and then accelerates to reach saturating flow speed. In conclusion, total blockage time of a bus can be calculated
by Equation 18.

(18)

5
ACE2012

Figure 1.Bus blockage time for a single-lane roadway

On a multilane roadway To determine the dwell time effect, dwell time should be multiplied by different
coefficients which are obtained varying numbers of effective blockage times. The equation is defined this way
because the entire right lane and the other lanes have different effects on saturated flow. In dependent to this
effective blockage time for each bus is defined by Equation 19.

(19)

where;
= Delay in the bus movement because of bus acceleration and deceleration. (s)
= The correction coefficient for right lane according to bus stop and road widths, it is calculated by Equation
20.

̅
(20)

where;
= The number of cars pass through the right lane on average bus stop time in the situation of saturated flow.

= The correction factor for right lane capacity change when the bus stopped at the bus stop. It is also
proportional capacity difference when a bus stopped or not stopped at the bus stop for an hour. This proportional
capacity difference is calculated by Equation 21.

(21)

where;
= Saturated flow value when the bus is not at the bus stop which is in the neighbor left lane or right lane
= Saturated flow value when the bus stopped at the bus stop which is in the neighbor left lane or right lane.
= Average waiting time for a bus at bus stop (s).

As it is seen in the Figure 2, in case of lane width and existing special bus bay for transit buses, the gap in front
of the bus is decreased because of the cars moving on the right lane and it is resulted from the decrease in the
blockage time.

Whereas, the situation defined above is opposite to the situation which is foreseen as the bus gets close to the
right lane while waiting at bus stop in HCM 2010 (TRB, 2010). It is necessary to define how many cars pass by

6
ACE2012

the bus without changing the lane at saturated flow condition and during green phase to define this effect
basically.

Figure 2. The case of existing special bus bay for transit buses

4. Field Data Collection and Evaluation


The observations were used to have a better understanding about impact of bus blockage time on signalized
intersections capacity and estimation of coefficient of new model,. The observations were collected from two
crowded bus stops at two signalized intersections in downtown of Izmir, Turkey. Table 1 shows geometric
properties of these intersections.

Table 1. Geometric properties of Balçova kipa and Konak Fevzipaşa signalized intersections

Intersection Type of Number Lane width Signal Time effective green Length of Distance
intersection of lanes (m) (s) time(s) bus stop between
lay by (m) intersection
and bus stop
(m)
Balçova
Signalized 3 3,10 70 40 40 15
kipa
Konak
Signalized 3 3,15 70 40 - 40
Fevzipaşa

These signalized intersections were chosen to determine impact of the different types of bus stops and road
geometries on capacity of signalized intersections. They were all located on the main arterials of Izmir and high
entry flow rates could be observed from these intersections. All observations were made during peak hours
(18:00-19:00) by using video camera records taken from a higher building near the intersections during
weekdays under dry and open weather conditions. To analyze the bus blockage effect from the video records
VideoPad Video Editor and MB-Ruller softwares were used.The following data was collected from the video
records at these intersections:

 Vehicle’s time headway value for each lane (sec);


 Average speed of vehicles.(Km/h);
 Dwell time of buses (s)
 Number of buses that arrived during green phase.
 Types and percentages of bus;

The intersections are consisted of 3 lanes per approach and a fixed traffic signal. The bus stop is on-line type for
Kipa Balçova intersection and has an exclusive lane in Konak Fevzipaşa intersection. (Figure 3a and 3b).

7
ACE2012

(a) (b)

Figure 3.Data collection with the help ofVideopad Editor and MB-Ruller sofware from the field observations

(a) (b)

Figure 4.Bus blockage station at saturated traffic flow.

Table 2. Statistical analyze results of observation data

The number Standard Number of data


Average Number of data when
of coming deviation when there was
Intersection Dwell there was no bus at
buses during the of dwell a bus at bus
time bus stop
green time time stop
Balçova
25 24 20,6 40 51
kipa
Konak
35 27,5 22,5 68 60
Fevzipaşa

From the obtained data only following two cases are evaluated;

a) Bus arrives during red phase but stops at bus stop during green light
b) Bus arrives during green phase and leaves during the same phase.

To define the blockage time on the other lanes capacity, the effect of bus on the traffic during green phase
situations has been taken into consideration. The passing duration of cars between two parts is recorded for each
vehicle travel. Obtained durations are divided to distances and the average speeds of vehicles were calculated.
To define these cars intervals in terms of time, two cars following each other were taken as basis and the
duration when they pass through the parts were recorded and the intervals of those two cars were obtained.
Negligible values were not taken into consideration (for example; too short waiting time for a bus or red light
turn on when the car stops and below 8 seconds). To calculate capacity, firstly time intervals for each car and
harmonically average of their speeds were calculated. Secondly, from average time intervals, volumes belonging
to those periodswere obtained. Volumes were divided to average speeds and the density was obtained from

8
ACE2012

density-capacity values for specified situations. The main reason to choose this method for capacity calculation
is the variation of the stopping time of busses at green duration. Bus movements at bus stop at red phase of the
signal was neglected. Because of this, there is no chance to obtain constant time period from recordings(like 45
seconds or a minute).

The vehicles pass through the defined parts in this study were categorized in two different situations. The first
one for the bus at the bus stop, and the second one for the bus not at the bus stop. Meanwhile, the data groups
less than four were neglected because they do not represent the real traffic flow. The Volume – Density graphs
were plotted for both situations. Graphs for bus at the bus stop or not at the bus stop are shown in figure 5 and 6.

1600 y = -0.3849x2 + 44.558x


1400 R² = 0.6126
1200 Durakta Otobüswhen
The situation Olmathere
Durumu
is a
Volume (v/h)

1000 bus at bus stop


800
y = -0.4512x2 + 49.831x
600 R² = 0.8096
400 The situation when there is
200 no bus atOtobüs
"Durakta bus stop
Olmama
0 Durumu"
0 20 40 60 80
Density (Vehicle/Lane/Km)

Figure 5.Volume-Density graph of adjacent lane to the bus stop lane at Balçova Kipa signalized
intersection

1600 y = -0.3803x2 + 43.274x


1400 R² = 0.7756
Otobüsün Durakta
The situation Olması
when there
1200
Volume (v/h)

Durum
is a bus at bus stop
1000
800
600
400 y = -0.3436x2 + 41.814x Otobüsün Durakta Olmaması
200 The situation when there
R² = 0.7396 Durum
0 is no bus at bus stop
0 10 20 30 40 50 60
Density (Vehicle/Lane/Km)

Figure 6.Volume-Density graph of adjacent lane to the bus stop lane at Konak Fevzipaşa
signalized intersection

As a result of regression analysis, Balçova Kipa intermediate lane capacity was obtained as 1375 vehicle/hour
when there is no bus at the bus stop and 1258 vehicle/hour there is a bus waiting at the bus stop for an hour. In
conclusion of regression analysis, Konak Fevzipaşa intersection capacity was obtained as 1225 vehicle/hour
when there is no bus at the bus stop and 1230 vehicle/hour when there is a bus waiting at the bus stop.

The F1 coefficient in Equation 18was accepted as F1=1 for both intersection in the case of bus stops and no cars
pass through the intersection. For Kipa intersection, the F2 coefficient was obtained 0,065 using q1 and q2
values. This value is calculated for Fevzi Paşa intersection and obtained as 0,004.

Acceleration and declaration time was calculated approximately 5 sec per bus from HCM 2000 (TRB, 2000).
Therefore, the total delay caused by acceleration and deceleration can be taken as the half of the total
acceleration and deceleration time. Effective blockage time can be obtained from Equation 22 and 23.

9
ACE2012

(22)

(23)

5. Conclusion and Recommendations


During this study, the HCM equation for bus blocking factor is thought to be inadequate so HCM model is
developed and a new model is offered. To calibrate this model with observations, required data were collected
and analyzed from Izmir’s important signalized arterials of two different points.

As a result of the analysis, F1 and F2 values are calculated within the offered model. It was seen that at
signalized intersections where the sections are wide, negative effects of bus stops are decreased. It was seen that
in Fevzi Paşa intersection which includes special bus bay for busses, buses do not give the chance for vehicle
drivers to overtake the buses in the intermediate lane, therefore the capacity decreases. However, in Kipa
intersection, where there is no bus bay for bus stops, %6 decrease in capacity was seen in intermediate lane for
an hour. In whatever way, HCM used constant time (14.4 sec) to determine bus blockage factor. Additionally
HCM does not consider the effect of physical properties of bus stop and roadway geometrical properties on
driver behaviors. In suggested model blockage time was taken into account at two different components (F1 and
F2 parameters). Considering the effect of lane and bus stop width, it was tried to determine the change in
capacity with the suggested model. Taking the site observations into account, it was observed that the new
suggested model can reflect the effect of bus stop and lane width.

The results can be developed using the other cities bus blockage factors and dwell time data. This study has
highlighted the importance of bus blockage factor which can be affected by different values such as geometric
shape of lanes, traffic, bus stop types and bus types (single unit bus or articulated bus). During this study, it was
also seen that the dwell time value has a linear effect on saturated flow. Bus acceleration and declaration has no
effects on the other lanes which the bus passes through the intersection.

Developing and exclusive bus blockage factor models for the determination of bus effects on the roadway based
on these influencing parameters is promising future direction for this research. Future studies can potentially
include the reverse situation of the assumptions of this study. The other factors affecting bus blockage can also
be expanded and evaluated.

6. References
Rodriguez-Seda, J.D. and Benekohal, R.F. (2006).Methodology for delay-based passenger car equivalencies for
urban transit buses. Transportation Research Record: Journal of the Transportation Research Board, No.
1988, pp. 127–137.
Transportation Research Board (TRB). 1950. Highway capacity manual, Transportation Research Board,
National Research Council, Washington D.C., U.S.A.
Transportation Research Board (TRB). 1965. Highway capacity manual, Transportation Research Board,
National Research Council, Washington D.C., U.S.A.
Transportation Research Board (TRB). 1985. Highway capacity manual, Transportation Research Board,
National Research Council, Washington D.C., U.S.A.
Transportation Research Board (TRB). 1994. Highway capacity manual, Transportation Research Board,
National Research Council, Washington D.C., U.S.A.
Transportation Research Board (TRB). 1997. Special report 209:Highway capacity manual, Transportation
Research Board, National Research Council, Washington D.C., U.S.A.
Transportation Research Board (TRB). 2000. Highway capacity manual, Transportation Research Board,
National Research Council, Washington D.C., U.S.A.
Transportation Research Board (TRB). 2010. Highway capacity manual, Transportation Research Board,
National Research Council, Washington D.C., U.S.A.
Wong, S.C.,Yang, H.,Au yeung, W.S.,Cheuk, S.L., Lo, M. K. (1998), Delay at signal-controlled intersection
with bus stop upstream,American Society of Civil Engineers, Journal of Transportation Engineering, Vol
124,p. 229

10
SİNYALİZE KAVŞAKLARDAKİ GECİKME ZAMANININ DOYGUN

ÜSTÜ AKIMLAR İÇİN MODELLENMESİ

Kiarash GHASEMLOU1, Metin Mutlu AYDIN2, Serhan TANYEL3,Ali TOPAL4, Peyman

Aghazadeh DOKANDARİ5

Özet

Sinyalize kavşaklarda oluşan gecikmelerin doğru belirlenmesi ulaşım sistemlerinin performansı ve seyahat süresi

açısından çok büyük önem taşımaktadır. Seyahat sürelerinin kavşak gecikmeleri yönünden daha iyi optimize

edilmesi, seyahat sürelerindeki gecikmeden dolayı oluşan fazla yakıt tüketiminin azalmasını sağlamayı

amaçlamaktadır. Gecikme süresi, sinyalize kavşaklarda tıkanıklık ve doygun akım üzeri durumlar için

gözlemlerden elde edilen veriler kullanılarak ve farklı modellerle karşılaştırılarak irdelenmiştir. Ayrıca bu

çalışma kapsamında, Türkiye´de konu ile ilgili yapılmış çalışmalar ile diğer bazı ülkelerdeki modellerin

karşılaştırılması amaçlanmıştır. Çalışma özellikle İran´da yapılmış olan çalışmalara odaklanmış ve böylece iki

ülkedeki akademik çalışmalardaki değişik yaklaşımlar incelenmiş ve iki ülke sürücü davranış farklılıklarının

tanımlanması amaçlanmıştır

Anahtar Kelimeler: Sinyalize Kavşaklar, Gecikme Süresi, Doygun Üstü Akımlar, Modelleme.

Abstract

The accurate determination of the delays at signalized intersections has crucial importance in terms of

performance and travel time of transport systems. The optimization of the travel times in terms of intersection

delays, aims to ensure the reduction of fuel consumption caused by delay times by reducing travel times. Delay

time is examined for congestion and over saturated flow situations with comparing different model using data

retrieved from observations. Furthermore within the scope of this study, it is aimed to compare the current

conducted studies in Turkey with different models in other countries related to this issue. This study is focused

on the conducted studies especially in Iran and therefore the different approaches examined in academic studies

and the driver behavior differences at both countries have been aimed to identify.

Keywords: Signalized Intersection, Delay Time, Over Saturated Flows, Modeling.


1. Giriş:

Gecikme, bir sinyalize kavşağın hizmet düzeyinin (LOS) etkinliğini ölçmek için kullanılan

temel parametrelerden birisi olmaktadır. Gerçek gecikmenin hesaplanması oldukça büyük

önem taşımaktadır. Gecikme, araziden toplanan veriler ya da analitik modeller kullanılarak

hesaplanabilmektedir.

Analitik modellerde, trafik akımının geliş ve çıkış karakteristiklerine bağlı olarak,

deterministik, rastgele kararlı (steady state) bileşen ya da zamana bağlı model (time dependent

delay model) olarak üç durumda tanımlayabilmek mümkün olmaktadır.. İlk iki model

çoğunlukla, doygun altı ve doygun üstü akımların kavşak gecikmelerinin hesaplanması için

kullanılmaktadır [1]. İlk model doygunluk derecesi 1,0 olduğu zaman sonsuz gecikmeyi,

ikinci model ise bu doygunluk derecesi için sıfır gecikmeyi hesaplamaktadır. Bunun

sonucunda, bu modeller sinyalize kavşak kapasitesine yakın durumları modellemekte yetersiz

kalmaktadır (Şekil 1). Bu sebeple zamana bağlı model dönüşüm yöntemi kullanılarak,

durağan ya da değişmeyen ve deterministik modellerin birlikte kullanımı yoluna gidilmesi

araştırmacılar tarafından önerilmiştir [2-4].


Şekil 1. Koordinat dönüşümü

Highway Capacity Manual (HCM) [4]‟de sinyalize kavşaklardaki araçların

gecikmelerinin hesaplanması için zamana bağlı bir model önerilmektedir. Türkiye´de yapılan

çalışmalarda da bu modeller kullanılmıştır [5-7]. Bu çalışmada, zamana bağlı olarak

geliştirilmiş analitik modeller ile İran‟da Nassiri ve Nadernejad [8] tarafından önerilmiş

ampirik bağıntıların karşılaştırılması ve kullanılabilirlikleri üzerinde durulmaya çalışılmıştır.

2. Zamana Bağlı Gecikme Modeli (Time Dependent Delay)

Bu model, doygunluk altı ve üzeri durum gecikmelerinin tahmin edilmesi amacıyla

kullanılmaktadır ve aşağıdaki bağıntıda göründüğü üzere üç parçadan oluşmaktadır [4].

dToplam =düniform + daşırı akım + dbaşlangıç (1)


Burada dtoplam :Toplam ortalama gecikmeyi (sn), düniform: Üniform gecikmeyi, daşırı akım:

Rastgele ve sürekli aşırı akım gecikmelerinin toplamını, dbaşlangıç : Analiz periyodu başladığı

zamanda kavşakta bekleyen araçların sebep olduğu gecikmeyi tanımlamaktadır.

2.1. Uniform Gecikme

Uniform gecikme bir sinyalize kavşakta kesintili bir deterministik (düzgün ve sabit) gelişe

sahip olan trafik akımının her bir devre içinde deterministik kuyruk teorisine bağlı gecikmesi

olarak tanımlanmakta ve Bağıntı (2) ya da (4) yardımıyla hesaplanmaktadır. [9], [4]. [4]‟ de

bu değer aşağıdaki bağıntı ile verilmektedir:

g 2
C (1  )
d uniform  PF C (2)
 g
2 1  min(1, x) 
 C

Burada,

C: Devre süresi (saniye),

g: Yeşil süre (saniye),

x: Doygunluk derecesi,

PF: İlerleme düzeltme faktörü olarak tanımlanmakta ve Bağıntı (3) ile

hesaplanmaktadır.

(1  p) f PA
PF  (3)
g
1
c
Burada,

f PA :Yeşil sürede grup gelişleri için ilave düzeltme faktörü,

P:Yeşil sürede araç gelişlerinin oranı olarak tanımlanmaktadır. [9], üniform

gecikmeyi aşağıdaki bağıntı ile hesaplamaktadırlar.

g 2
C (1  )
d uniform  C (4)
 g
21  x 
 C

Bu bağıntıda doygun üstü akımlarda (x>1), üniform gecikme hesabında doygunluk

derecesi 1 olarak alınır.

2.2. Aşırı Akımdan Kaynaklanan Gecikmeler

Bir sinyalize kavşakta kesintili ve deterministik olmayan gelişe sahip bir trafik akımının her

bir gözlem periyodu içinde oluşan geçici ya da sürekli şekilde devam eden aşırı akımın

sonucunda oluşan gecikme olarak tanımlanmaktadır. [4]‟de “artan gecikme” ismi ile

tanımlanmakta rastgele gecikme ve sürekli devam eden gecikmeden oluşmaktadır. Bağıntı (5)

ile hesaplanabilmektedir. Buradaki bağıntı aşırı akım gecikme modelinin [2] genelleştirilmiş

hali olup ayrıca [10] ve [4]‟deki modeli de kapsamaktadır.

daşırı akım= dRasgele+ dsürekli (5)

dsürekli  1800T(X - 1) (6)


 m( x  x0 ) 
d Rasgele  900T - (X - 1)  (x - 1) 2   (7)
 CT 

 m( x  x0 ) 
daşırı akım  900T  (X - 1)  (x - 1) 2   (8)
 CT 

Burada C:kapasite (araç/saat), x:doygunluk derecesi (q/c), T:analiz süresi (saat),

m:gecikme parametresi ya da ölçümleme parametresi (Tablo 1), x0 :taşan akımın yaklaşık

sıfır olduğu doygunluk derecesidir (Tablo 1).

Tablo 1. m ve x0 değerleri

m X0 k I
Akçelik(1980b) 12 0.67+sg/600 - -

HCM2000 8kI 0 0.5(x>1)* 0.9(x>1)*

Akgüngör & Bullen(2007) 8k 0 0,8x2  1,4 x  1,1 -

Akgungor (2007) 8k 0 0,6923T 0,0844 -


*(HCM 2000) de doygun üstü akımlar için k ve l parametreleri sabit sayı olarak alınmıştır

Başlangıç gecikmesi analiz periyodu başladığı zaman, önceki devre sebebi ile

kavşakta kuyruk mevcut ise, bu kuyruğu temizlemek için gereken zaman olarak

tanımlanmaktadır [4]. Bu gecikme sadece [4]‟deki modelinde yer almakta ve Bağıntı (9)

yardımı ile hesaplanmaktadır. Diğer modeller de ise analiz periyodu başladığı zaman

başlangıçta kuyruk olmadığı durumu ele alınmaktadır.(dbaşlangıç=0).

1800 Qb (1  u)t
dbaşlangıç = (9)
cT
Burada Qb :Başlangıç akımındaki kuyruk miktarı (araç), u :Gecikme parametresi [eğer

t<T ise u=0„dır], C:Düzeltilmiş şeritli grup kapasitesi (araç / saat), T:Analiz Periyodu (saat), t:

T zamanında karşılanmayan talep zamanı (saat) [doygun üstü akımlar için t=T].

3. Ampirik Model:

Çalışma kapsamında, İran‟da [8] tarafından önerilmiş olan doygun üstü akım durumunda,

doygun gecikme amprik modellerinin kullanılabilirliği incelenmeye çalışılmıştır. Modeller,

doygunluk dereceleri birden fazla olan ve sabit devreye sahip farklı kavşaklardan kamera

çekimi ile kavşağa giriş ve çıkış yapan araçların hacimleri ve gecikme verilerinin SPSS

programı kullanılarak regresyon analizine tabi tutulmasıyla elde edilmiştir. Bu modeller ve

regresyon analizinden elde edilen parametrelere ait değerler Tablo (2-5)‟de verilmektedir.

Tablolardan da görüldüğü gibi yedi ayrı model incelenmiştir. Tablo 2, Tablo 3 ve Tablo 4‟de

modeller sırasıyla X ve (X-1) amaç fonksiyonuna bağlı olarak iki tip olarak alınmıştır.

Tablo 2. Tip 1 ve 2 modelleri

Model No Modeller a katsayısı T istatistiği R2

(c  g e )
1  aX 141,15 45,48 Eksi
2
(c  g e )
2  aX 2 96,65 64,13 0,51
2
(c  g e )
3  aX 3 65,39 68,74 0,79
2
(c  g e )
4  aX 4 43,75 51,57 0,77
2
(c  g e )
5  a( X  1) 447,26 64,55 0,78
2
(c  g e )
6  a( X  1) 2 876,04 26,60 0,59
2
(c  g e )
7  a( X  1)3 1609,37 16,52 0,23
2
(c  g e )
8  a( X  1) 4 2748,75 11,65 Eksi
2

Tablo 3. İran H. Nassiri ve M. Nadernejad [8] modelleri Tip 3 ve 4

Model Katsayılar t istatistiği


Modeller R2
no a b c a b c
(c  ge )
9  (aX 2  bX ) 159,35 -92,22 - 7,29 -2,88 - 0,75
2
(c  ge )
10  (aX 3  bX 2  cX ) 54,39 2,18 24,06 7,27 0,35 1,48 0,76
2
(c  ge )
11  (aX 3  cX ) 54,39 - 24,06 7,27 - 1,48 0,75
2
12
(c  ge )
2

 a( X  1)2  b( X  1)  -136 529,81 - -2,8 14,32 - 0,75

13
(c  ge )
2

 a( X  1)3  b( X  1)2  c( x  1)  18,18 -184,5 533,42 0,03 -0,32 4,18 0,76

Tablo 4. İran H.Nassiri ve M.Nadernejad [8] modelleri Tip 5 ve 6

Model no Katsayılar t istatistiği


Modeller
a b a b R’2
(c  ge )
14  aX b 75,63 2,63 10,38 11,43 0,75
2
(c  ge )
15  aebX  14,83 1,79 4,26 11,35 0,75
2
(c  ge )
16  aLog( X ) 1262,60- - 69,84 - 0,75
2
(c  ge )
17  a( X  1)b 290,30 0,815 17,65 11,13 0,75
2
(c  ge )
18  aeb ( X 1)  59,24 1,79 12,65 11,35 0,75
2
(c  ge )
19  aLog( X  1) -513,89 - -13,51 - Eksi
2

Tablo 5. İran H.Nassiri ve M. Nadernejad [8] modelleri Tip 7


Model no Model 7 a Katsayısı t İstatistiği R2

(c - g e )  4x 
20  a  (X - 1)  (x - 1) 2   222,68 66,11 0,79
2  C 

Burada a, b, c parametreleri, verilerin regresyonundan modeller için elde edilen

katsayılardır. Sonraki adımda bu modeller arasından istatiksel açıdan daha anlamlı olan 7

model seçilerek, 4 farklı kavşaktan toplanan gecikme verileri ile karşılaştırılması yapılmış ve

bunun sonucunda aşağıdaki 3 bağıntı önerilmiştir [11].

(c - g e )  4x 
d toplam   a  (X - 1)  (x - 1) 2   (9)
2  C 

(c - ge )
dtoplam   a(X - 1) (10)
2

(c - ge )
dtoplam   aX4 (11)
2

Burada uniform gecikme, modellerin ilk bölümünü oluşturmakta kalan bölümler ise

doygun gecikmeyi tanımlamayı amaçlamaktadırlar. Bağıntı (6) HCM modelinden ilham

alınarak üretilmiştir. Model (7) ve (8) ise amaç fonksiyonu x ve (x-1) olan, sırasıyla Tip 1 ve

Tip 2 modeller arasında yer almaktadır.


4. Verilerin değerlendirilmesi

Şekil 2‟de görüldüğü üzere analiz süresi 0.25 saat ve kapasitesi 500 araç/saat, doygun akımı

1500 araç/saat ve 90 saniyelik devir içerisinde 30 saniyelik yeşil süreye sahip olan modellerin

doygun gecikme miktarı, doygunluk derecesi 1 ve 1,7 olan bir kavşak için çizilmiştir.

400
350
300 Akçelik(1980b)
250 Akgüngör(2007)
200 Akgüngör&Bullen
Nassiri 1
150
Nassiri 2
100 Nassiri 3
50 HCM 2000
0
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7

Şekil 2. Doygun gecikme miktarı (sn)–Duygunluk derecesi grafiği

Doygunluk derecesinin birden fazla olan akımlar için zamana bağlı gecikme ve Nassiri

ve Nadernejad [8] modelleri ortalama uniform gecikmenin hesaplanmasında, araçların efektif

kırmızı sürenin ortasında kavşağa vardıklarını düşünmektedir. Tek farklılık, HCM‟ de ek

olarak PF faktörünün kulanılmasındır. Diğer taraftan başlangıç gecikmesi olmadığı, başka bir

ifade ile analiz başlangıçında hiç kuyruk olmadığı varsayılarak sadece yöntemlerde doygun

gecikme miktarı ele alınmış ve irdelenmiştir [8].


Zamana bağlı modeller analiz süresine bağlı oldukları için sürenin artması ile gecikme

miktarları da zaman bağlı olarak artmaktadır. Burada Akçelik modeli X0 için bir değer

öngörmekte ve aşırı akım gecikmesinin doygunluk derecesine varmadan X0 başladığını

varsaymaktadır. [8]‟in önermiş olduğu Bağıntı (9)‟u, Bağıntı (8) kapsamında incelemek

mümkündür. (m=1,Tf=0.25,X0=0) zira bulunan “a” katsayısı analiz süresi katsayısı 900 T‟ye

çok yakın olmaktadır. Aynı zamanda m/Tf =4 olarak düşünülmektedir.

Bu durumda Nassiri ve Nadernejad aslında İran için “m” değerini “1” olarak aldığını

da söylemek mümkündür. Nassiri ve Nadernejad (10) bağıntısı ise zamana bağlı modeller ile

karşılaştırıldığında, Bağıntı (6) ile benzer değerler verdiğini söylemek mümkündür. Zira

bulunan “a” katsayısı 1800 Tf değerine çok yakın bir değer almaktadır. Yani bu bağıntının

daha çok devam eden doygun gecikmeyi tanımladığı söylenebilmektedir. Nassiri ve

Nadernejad [8], Bağıntı (11) ise fonksiyon yapısı açısından değerlendirdiğinde doygunluk

derecesinin artması ile beraber yüksek değerleri hesaplayacağından 1.7 doygunluk derecesi

üzerindeki gecikmeleri çok doğru tanımlanmayacaktır.

5. Sonuç ve Öneriler

Çalışma kapsamında, sinyalize kavşaklarda doygun üstü akımlar için çeşitli araştırmacılar

tarafından önerilmiş olan gecikme bağıntıları incelenmiş ve karşılaştırılmaya çalışılmıştır.

Çalışma sonucunda İran‟da H.Nassiri ve M.Nadernejad [8] tarafından önerilmiş olan

modellere ait parametrelerin büyük oranda HCM [4] veya Akçelik [3] bağıntıları gibi dünyada

kabul görmüş bağıntılardaki parametrelere çok yakın değerler verdiği görülmüştür. Bu

kıyaslama sonucunda elde edilen bulgular, sürücü davranışlarının ülkeden ülkeye değiştiği

düşünüldüğünde farklı ülke, kültür ve yaşam tarzına sahip sürücülerin bile doygun üstü

akımlarda benzer başarım özellikleri gösterebileceğini ortaya çıkarmaktadır. Bu bağlamda,


sinyal sürelerinin ve kullanılan sistemlerin benzerliklerinin de göz önünde bulundurulmasında

yarar olduğu düşünülmektedir.

Kaynaklar:

1. Hurdle, V. 1984. Signalized Intersection Delay Models: A Primer for the Uninitiated.

Transportation Research Record 971, Transportation Research Board, National Research

Council, Washington, DC, pp. 96-105.

2. Kimber, R.M. and Hollis, E.M. 1978. Peak Period Traffic Delay at Road Junctions and

Other Bottlenecks, Traffic Engineering and Control, 19, 442-446.

3. Akçelik, R. 1981. Traffic signals: capacity and timing analysis. Australian Road Research

Board, Research Report 123.

4. TRB, Highway Capacity Manual 2000.Transportation Research Board, Chapter 16,

National Research Council, Washington, D.C., USA, 2000.

5. Murat Y.Ş., 2006. Comparison of Fuzzy Logic and Artificial Neural Networks Approaches

in Vehicle Delay Modeling, Transportation Research Part C- Emerging Technologies, Vol

14/1 pp 316-334., Pergamon Press.

6. Akgungor, A.P. 1998. Analytical Models of Delay Estimation at Signalized Intersections

for Variable Demandand Time Conditions, School of Engineering, Ph.D. Dissertation,

University of Pittsburgh, USA.

7. Akgungor, A.P. and Bullen, A.G.R. 1999. Analytical Delay Models for Signalized

Intersections, 69th I.T.E. Annual Meeting, Nevada, USA 1999.


8. Nassiri.H, and Nadernejad.M, 2006. Delay Modeling at Over-Saturated Signalized

Intersections, 7th International Conference on Civil Engineering, Tehran, Iran, (In Farsi).

9. Webster, F.V. 1958. Traffic Signal Settings, Road Research, Technical Paper 39, Road

Research Laboratory, Her Majesty‟s Stationery Office, Berkshire, England.

10. Akçelik, R. 1980. Time-Dependent Expressions for Delay, Stop Rate and Queue Length at

Traffic Signals. Australian Calculation for Discrete Adaptive Traffic Signal Control. Road

Research Board, Internal Report, AIR 367-1.

11. Shirazi H.A. 2005.Amerika, Avustralya ve İngiliz yöntemleri ile sinyalize kavşak

tasarımının karşılaştırılması, Elmo Sanat Üniversitesi, Yüksek lisans Tezi, 155 s, İran.

You might also like