6th International Congress on Advances in Civil Engineering, 6-8 October 4 Bogazici University, Istanbul, Turkey Toll Plazas at the Bosphorus Bridge K.S. Öüt stanbul Technical University, Civil Engineering Faculty, Department of Transportation, stanbul, Turkey Abstract The Bosphorus Bridge plays an important role in stanbul traffic. It has three lanes at each direction and about 9, vehicles cross the bridge daily. There are tollbooths at the Asian site. In this study, the arrival and the service time to the plazas are examined at the context of the queueing process. The data are gathered during the peak-hour period for 9 min. The statistical analysis of the observed tollbooths shows that the average vehicle headway is 8.48 sec, whereas the service time is 8.6 sec. The average queue length changes from 6 to 3 vehicles between 7:-8: and from 9 to vehicles between 8:-8:3. Introduction Toll facilities include freeway sections and toll plazas integrated within the system. Toll plazas can be considered interrupted flow facilities, and therefore the toll facility cannot be evaluated as a normal freeway segment. Furthermore, the toll plazas cannot be evaluated as signalized intersection because of their unique characteristics. Woo and Hoel (99) examined toll plazas at Virginia Richmond Petersburg Turnpike where the service times at the toll plazas ranged from.87 to 4.88 sec for trucks and from. to.47 sec for automobiles. The capacity of a general booth is found to range from 6 to 7 passenger cars per hour. Since then, several studies have been made for manually operated (Al-Deek et al. 997), electronic (Al-Deek et al, 996), and the level of service of the toll plazas (Lin and Su, 994),(Klodzinski, ),( Klodzinski and Al-Deek, ). Despite these studies, there is no serious studies for the departure and the arrival of the toll plazas in Turkey. In this study, the arrival, the service time and the queue formation at the manually operated tollbooths on the Bosphorus Bridge are examined.
Study Area and Survey The Bosphorus Bridge (BB), built in 973, is the first bridge which ties the Asian and European continents in stanbul. In 988 the second bridge, named Fatih Sultan Mehmet Bridge, was constructed for Asia-Europe crossing. Since then, the BB has only been used by city traffic, which means the crossing of the lorries, TIRs or intercity buses is not allowed. The BB is on the O freeway, which has x3 characteristic. During the morning peak hours (6:-:), an extra lane is opened on the opposite traffic flow section for the Asia- Europe crossing, consequently four lanes are operated along this direction. Similarly, during the evening peak hours (7:-:), an extra lane is used for the Europe-Asia crossing, this time again on the opposite traffic section (Fig. ). Europe Site Asia Site Europe Site Asia Site Europe Site Asia Site (a) (b) (c) Fig. : Traffic flows on the BB during (a) morning peak, (b) evening peak, (c) off-peak On the BB, there are tollbooths, which are placed at the Asian side and used only by Europe-Asia crossers. Vehicles in the Asia-Europe direction do not make toll payments. During the survey, of the tollbooths were operated with manual toll payment, and 4 others (tollbooth-8, 9, and ) had the Electronic Toll Collection System (ETCS). ETCS fitted vehicles do not stop to pay the toll provided that, they have a specific kit for the automatic passage. In this study, the arrivals and the service at tollbooth-, and 3 were recorded by using two handy cams on 8 th May,, a Thursday, between 7.-8.3. The location of tollbooths can be seen in Fig.. From these records, the headway and the service time were measured. For the booth- and booth-, these data were measured during 9 min. However, for the toll booth-3 as the queue behind the booth- interrupted the vision, only 46-minute measurement was taken into account for the statistical analysis.
Fig. : The location of the toll plazas on the BB Arrival At the beginning of the survey, there were queues behind all observed tollbooths. The arrival time of each vehicle to the end of the queue was measured and the duration between two arrivals, which is called as headway, was calculated. The duration of measurements, the number of arrived vehicles, the average and the standard deviation of the headway are given for each tollbooth separately in Table. Table : The observation of the headway at each tollbooth Duration Arrived Headway (sec) Location (min) vehicles Average Standard dev. Booth- 9 88 9.9 8.3 Booth- 9 684 7.89.97 Booth-3 46 37 8.4 6. Total 6,99 8.48 6.9 In the traffic engineering, the distribution of the arrival to a channel is considered as Poisson distribution (May, 99). If the Poisson distribution is accepted for the arrival, then the headway distribution is negative exponential distribution: t P (t) = exp t () where, t is the headway, t is the average headway and P(t) is the probability distribution function of t. 3
In order to determine the distribution of the vehicle headway for booth- and booth- separately, the χ (chi-square) test is used. As the observation time was short for toll booth- 3, the distribution of the arrival was not determined. According to the χ test, at the level of significance. and the degree of freedom 9, the vehicle headway for toll booth- and booth- do not have negative exponential distribution. The χ test statistics are calculated as 49.9, and 9.6 for toll booth- and respectively. Secondly, the shifted negative exponential distribution is examined at the arrival distribution. Shifted negative exponential distribution is (Bayazıt, 996): t α P (x) = exp () t α where, α is the shifting factor. At the shifted negative exponential distribution, the probability of headway being less than or equal to shifting factor (α) is zero. The hypothesis, that the shifted negative exponential distribution fits the data sets, is accepted by using the χ test, when α is selected as.7 for the booth- and.4 for the booth. The χ test statistics are calculated as 7.4, and 3 for toll booth- and respectively. The number of headway less or equal to.7 is 8 for the booth-, and the number of headway less or equal to.4 is 36 for the booth-. The histograms of the headway with -sec interval for each booth and the shifted negative exponential model for toll booth- are given in Fig. 3 separately. 3 39 6 3 3 44 87 63 4 6 6 8 Frequency 6 7 Booth- Booth- Booth-3 Model 4 4 3 6 3 -, 4,-,,-,,-,,-,,-3, 3,-3, 3,-4, 4,-4, 4,-,,-,,-6, Headway (sec) Fig. 3: The histograms of the headway for 3 toll booths Service Time The service time does not only consist of the toll collecting time. It starts when a vehicle enters the tollbooth, and includes the money-giving duration by driver to the toll clerk and 4
the clerk returning the bill and giving the sign to allow the car to pass the booth. Service time ends when the next vehicle enters this tollbooth. As the service time includes the duration of the entrance of the next vehicle to the booth, there must be a queue behind toll booth during service time measurement. Otherwise, the service time would be longer than its real value. During the observation, the queue was always present behind of the each toll booth. The duration of measurement, the number of vehicles, the average and the standard deviation of the service time are presented for each booth in Table. Table : The observation of the service time at each tollbooth Location Duration Arrived Service time (sec) (min) vehicles Average Standard dev. Booth- 9 8 9.34.4 Booth- 9 67 8. 4.7 Booth-3 46 33 8.3 4.6 Total 6,73 8.6 4.6 The average service time of three booths was 8.6 sec during observation; however, this value varied according to the vehicle type. To determine this variation, the vehicles are grouped into 4 categories, such as motorcycle, car, van or small truck, and public transportation bus. Trucks and intercity buses are not allowed on the BB. The number of the vehicle type passing through the observed booth is shown in Tables 3. Table 3: The distribution of the vehicle types at the tollbooths Location Motorcycle Car Van Bus Total Booth- 3 4 64 39 8 Booth- 4 8 6 84 67 Booth-3 6 3 3 33 Total 8 73 39 4,73 As the toll are collected manually, the service time does not only vary for the vehicle type but also for the toll clerks. The service times of the three booths according to vehicle types are given in Table 4. For instance, the service time of cars was about 8 sec. at the booth- and booth-3; however, it was 9.34 sec. at the booth-. Table 4: The service time of the vehicle types at the tollbooths Location Motorcycle Car Van Bus Booth- 4.69 9.34 8.7.9 Booth- 7.9 8. 8. 8.7 Booth-3 6.79 8. 9.3 3. Average 6.6 8.48 8..7 In the traffic engineering, the distribution of the service time is considered as negative exponential distribution (May, 99). The statistical analysis of the service time data of the tollbooth-, and 3 separately shown that, their distributions are not negative exponential
or shifted negative exponential. The most important reason of this result is the variation at the vehicle type. Exemplify, at the tollbooth-, 6% of vehicles was motorcycles, 4% was cars, 8% was vans, 4% was buses and the service time of each vehicle type was different. The histograms of service time with -sec interval could be seen in Fig. 4 for each booth separately. 9 7 7 66 4 Frequency 8 Booth- Booth- Booth-3 7 73 7 4 4 43 43 48 7 9 3 3 3 9 6 7 4 6 8 4 33 4 - -4 4-6 6-8 8- - -4 4-6 6-8 8- - -4 4-6 6-8 8-3 3-3 3-34 34-36 36-38 38-4 Service Time (sec) Fig. 4: The histograms of the service time at the tollbooths Queueing at the Toll Plazas At a toll booth, if - the arrival rate expressed as vehicle per hour (λ) fits to the Poisson distribution, - the departure rate expressed as vehicle per hour (µ) fits to the negative exponential distribution, - λ < µ the queueing properties are determined by formulas (May, 99). However, in this study, the arrival rate is bigger than the departure rate and the departure distribution is not negative exponential distribution. As a result, the analysis of the formed queue at the BB plazas would not be true if the present formulation (May, 99) was used. The change of queue length at each booth is calculated by using cumulative vehicles time series of the arrivals and departures. In Fig., the arrival and the departure time series of the booth- are given. At the beginning of the survey, -vehicle-length queue was observed behind the booth. 6
6 Numbe r of cumulative ve hicle 4 3 6 8 4 3 36 4 48 4 Time (sec) Arrival Departure Fig. : The time series of the arrival and departures at the tollbooth- between 7:-8:3 In Fig., the arrival and departure represent the headway and the service time respectively. The changes of the queue length time series are given in Fig. 6-8 for three tollbooth-,, and 3. The average queue length was vehicles at the tollbooth- (for 9 min), 4 vehicles at the tollbooth- (for 9 min) and 6 vehicles at the tollbooth-3 (for 46 min) during the observation. The average waiting time for one vehicle, without the service time, is calculated as 4 sec. at the booth-, 4 sec. at the booth- and sec. at the booth-3. Between 7:-7:46, the average queue lengths are 3, 9, and 6 vehicles, and average waiting times are 4, 7, and sec for tollbooth-, and 3 respectively. The reason of the extra delay at the booth- is the public buses. The bus stop is located m after the toll plazas, thus bus drivers usually prefer the booth-. During the observation, 9% of the public buses used the booth-. For the tollbooth- after 8: and for the tollbooth- after 7:3 queue length becomes much longer. The changes of the average queue length are given in Table for three tollbooths for -min interval. After 8:, the average queue length is about vehicles at the booth- and booth-. The daily traffic flow surveys show that between 8:-:, the flow on the BB in the Europe-Asia direction is at the capacity level, which means the queue at the toll plazas. 7
3 lenght of queue (vehicle) 7: 7: 7:3 7:4 8: 8: 8:3 Time Fig. 6: The change of the queue length at the tollbooth- 3 lenght of queue (vehicle) 7: 7: 7:3 7:4 8: 8: 8:3 Time Fig. 7: The change of the queue length at the tollbooth - 8
leng ht of queue (vehicle) 7: 7: 7:3 7:4 Time Fig. 8: The change of the queue length at the tollbooth-3 Table : The change of the average queue length at the tollbooths 7:-7: 7:-7:3 7:3-7:4 7:4-8: 8:-8: 8:-8:3 Booth- 4 7 9 9 Booth- 4 7 8 9 Booth-3 7 6 - - - Conclusion In this study, three tollbooths with manual toll collection on the Bosphorus Bridge in Istanbul are examined. The observation time at toll booth- and booth- is between 7: and 8:3; however, due to the queue formation behind the tollbooth- and prevented complete vision, the tollbooth-3 was examined only between 7: and 7:46. The public transport buses, which cross the BB, usually prefer booth- and booth- because of the bus stop location. The average headway of vehicles arrival is measured as follows: 9.9 sec with an 8.3 sec standard deviation, where the arrival rate is 39 veh/hr at the tollbooth-; 7.89 sec with a.97 sec standard deviation, where the arrival rate is 46 veh/hr at the tollbooth-; 8.4 sec with a 6. sec standard deviation, where the arrival rate is 47 veh/hr at the tollbooth-3. 9
The headway distribution is determined as shifted negative exponential distribution with α =.7 for tollbooth- and with α =.4 tollbooth-. The average service time at tollbooths is measured as follows: 9.34 sec with a.4 sec standard deviation and 386 veh/hr the departure rate at the tollbooth-; 8. sec with a 4.7 sec standard deviation and 444 veh/hr the departure rate at the tollbooth-; 8.3 sec with a 4.6 sec standard deviation and 4 veh/hr the departure rate at the tollbooth-3. The average service times were 4.69, 7.9, and 6.79 sec. for motorcycles, 9.33, 8. and 8. sec. for cars, 8.7, 8. and 9.3 sec. for vans,.9, 8.7 and 3. sec. for buses at tollbooths, and3 respectively. The service times at the booth- and booth- do not fit the negative or shifted negative exponential distribution. The average queue length and the waiting time between 7:-8:3 were vehicles and 4 sec. for tollbooth- and 4 vehicles and 4 sec. for tollbooth-. The average queue length becomes longer after 8:, with about vehicles for booth- and booth-. As the service time changed at each tollbooth, the average waiting time was 8 sec. and 6 sec. for the booth- and booth- respectively between 8:-8:3. References Al-Deek, E. A. Radwan, H., A. Mohammed and J. G. Klodzinski, (996). Evaluating the Improvements in Traffic Operations at Real-Life Toll Booth with Electronic Toll Collection, ITS Journal, pp.-3. Al-Deek, H., A. Mohammed and E. A. Radwan, (997). Operational Benefits of Electronic Toll Collection: Case Study, Transportation Engineering Journal, ASCE, pp.467-477. Bayazıt, M., (996). naat Mühendisliinde Olasılık Yöntemleri, TÜ Matbaası, stanbul. Klodzinski, J., (). Methodology for Evaluating the Level of Service of Toll Booths on a Toll Road Facility, Presented at International Bridge, Tunnel and Turnpike Association Technology Committee Workshop, Washington D.C. Klodzinski, J., and Al-Deek, E. A. (). New Methodology for Defining Level of Service at Toll Plazas, Transportation Engineering Journal, ASCE, pp.73-8. Lin F.B. and Su C.W, (994). Level-of-Service Analysis of toll plazas on Freeway Main Lanes, Transportation Engineering Journal, ASCE, pp.46-63 May, A., (99). Traffic Flow Fundamentals, Prentice Hall, New Jersey. Woo, T. H. and Hoel, L. A., (99). Toll Booth Capacity and Level of Service, Transportation Research Record 3, pp. 9-7.