J. Cent. South Univ. (2016) 23: 1817-1822

DOI: 10.1007/s11771-016-3235-5

Schedule optimization to improve trunk-local bus transfer efficiency in small conurbations: A case study of New York’s capital region

ZHANG Da-peng(张大鹏), WANG Xiao-kun (王晓坤)

Department of Civil Engineering, Rensselaer Polytechnic Institute, Troy, 12180, USA

Central South University Press and Springer-Verlag Berlin Heidelberg 2016

Abstract: Fostering the use of transit has been broadly accepted as an effective way to improve social equity and reduce the externalities caused by transportation. In the great body of transit literature, many have focused on the improvement of transfer efficiency. However, investigation on transit transfer efficiency is still lacking for medium sized cities or suburban areas that have sprawled from city centers. The special features associated with such an urban form lead to unique travel patterns and bus operations. This work develops a process to improve bus transfer efficiency for small conurbations considering their special characteristics. A case study of New York’s Capital District is used to illustrate the proposed method. Results show that the transfer waiting time can be remarkably shortened. The proposed method can be widely adapted to other transit systems in small conurbations.

Key words: transfer efficiency; small conurbation; rural transit

1 Introduction

Fostering the use of public transit has been broadly accepted as an effective way to improve social equity and reduce the externalities caused by transportation. In order to encourage the use of public transit, researchers have conducted numerous studies, from various perspectives, aiming at improving the quality of transit services. Among them, many focused on the improvement of transfer efficiency, including transfer frequency [1-2], convenience [3-5], reliability [6-7] and most importantly, time [8-10]. Although literature on transfer time is not lacking, the majority of existing studies focused on the transit systems in metropolitan areas with high population densities. Few have investigated the problem for medium sized cities or suburban areas that have sprawled from city centers [11].

Meanwhile, it is generally agreed that travel patterns and user characteristics differ across different urban forms [12-14]. As STEAD and MARSHALL [14] summarized, the relationship between urban forms and travel patterns have been examined in a great body of research from different perspectives. GIULIANO and NARAYAN [12] claimed that urban forms that have low development densities and dispersed population are usually related to car dependent travel patterns. In contrast, dense urban forms are associated with lower levels of car use due to high population density, high accessibility, and better public transportation service, among others. In order to improve transit transfer efficiency, it is first necessary to consider the features of different urban forms.

This work aims to improve bus transfer efficiency in small conurbations. Conurbation is defined as “a large area consisting of cities or towns that have grown so that there is very little room between them” [15]. It can also be called a polycentric urban agglomeration, in which transportation connect areas and creates a single labor market. In the U.S., several large conurbations are well known, for instance, the New York tri-state area and the Greater Boston area. Most existing literature [9, 16] studied transit transfer mechanisms of these large conurbations or cities within that, which feature densely distributed population, transit-oriented transportation systems, and intensive social and economic activities. Compared to these large conurbations, the small conurbations, or polycentric urban agglomeration that is comprised by small to medium cities are relatively understudied. However, this type of urban form is commonly seen in the U.S., especially in the northeast and the west coast regions, where neighboring cities or towns are close, and the sizes of these cities are not very large. In these small conurbations, each city functions as a center, providing commerce, service, or entertainments. Residents live sparsely in the outskirts of the cities and travel to the centers regularly for jobs, shopping, and leisure, among others. Typical examples of the small conurbations in the U.S. include the New York State’s Capital District, the Buffalo-Niagara Falls area, and the Greater Sacramento area. In these small conurbations, population is distributed sparsely, and so is transportation infrastructure. The major challenges faced by these areas are not congestion, but limited resources and the difficulty to ensure accessibility for residents with limited transportation options.

The bus system in a small conurbation usually operates in a spoke-hub pattern. Many local bus routes serve the city center where users can transfer to trunk buses that connect to other city centers. Such a spoke-hub pattern helps lower the operation costs at the expense of service directness. Compared to local buses, trunk buses usually have higher frequencies, longer service times, and carry larger volumes. In some areas, they are also branded as BRT (bus rapid transit) with limited stops and dedicated right of way. Good service of trunk buses is able to enhance regional connectivity and the attractiveness of the overall transit system. Thus, trunk buses take a more important position in the entire bus system. When shifting bus schedules to improve the transfer efficiency, changes made to trunk buses should be minimized whenever possible. Additionally, transit system in a small conurbation also exhibits the following features: The headway of buses is relatively long, which increases the probability of long transfer waiting time. Capacity of local buses is unlikely to be a constraint and there are always seats available on local buses because transit is largely underutilized. Many methods have been proposed to improve the bus transfer experience and some have been approved to be quite effective, such as the public transit vehicle arrival information system [17], GPS data on mobile devices [18], and comfortable bus shelter constructions [19]. However, these methods also have high investments and maintenance requirements, making them less favorable for transit service providers in small conurbations. In contrast, bus schedule coordination is more favorable as it does not require extensive planning and construction efforts, which is the focus of this work.

The next section reviews literature of bus transfer efficiency, with an emphasis on bus schedule coordination. The empirical data and the methodology are then introduced, followed by the results analysis and conclusions.

2 Literature review

Transfer experience is one of the most important aspects of transit systems’ performance [20]. Passengers’ overall travel experience can be remarkably improved with good transfer experience, which comprises many components. ISEKI and TAYLOR [21] claimed that good transfer experience should have short and predictable waiting time in a safe environment. LITMAN [22] found that transfer experience was more sensitive to people who valued time higher, such as managers and professionals. He suggested that related studies should distinguish the transfer experience of passengers with different characteristics. In particular, transfer time should be treated as a set of penalties when establishing related models of measuring transit service performance.

Given the importance of transfer time, great efforts have been put on the transit schedule coordination problems. For example, and [23] used the fuzzy ant system to minimize the total waiting time of all passengers at transfer nodes in a transit network. CEVALLO and ZHAO [24] presented a genetic algorithm approach to synchronizing bus schedules by shifting existing timetables in Broward County. CEDAR et al [16] attempted to maximize the number of simultaneous bus arrivals at the connection nodes of the network. TING and SCHONFELD [25] used a heuristic algorithm to coordinate the schedules in a multiple hub transit network.

In terms of the trunk-local transfer pattern, several methods have been proposed to improve its efficiency. However, most existing studies focused on transit systems in large cities. For example, SHRIVASTAVA and DHINGRA [26] coordinated the schedules of suburban trains and public buses in Mumbai, India to improve the integration of the two public transport modes. SUN et al [27] proposed measurements to investigate the transfer efficiency at terminals, mostly between subways and buses, in Beijing, China. Among the existing literatur, few have looked into the transfer time problem for small conurbations. Typically, congestion is not a big concern in small conurbations so that the departure and arrival times can be reliably followed. However, the bus service in small conurbations tends to be sparse. Headway is usually long so that passengers are likely to endure a rather long waiting time. Recognizing the special features of the bus systems in small conurbations, this paper aims to contribute to the literature by investigating and improving the trunk-local transfer efficiency in small conurbations.

3 Data description

The bus system in New York’s capital region is used as a case study, with the transfer efficiency improvement in Troy as the main focus. New York’s capital region is of 7228 square miles with a population of 837720 in 2010. It consists of three major cities, Albany, Schenectady, and Troy. Albany and Schenectady are the centers of regional commerce, finance, and entertainments while Troy mainly serves as the residential area. In the evenings, residents of Troy commute back from Albany and Schenectady by first taking the trunk buses that connect Troy with other regional centers then local buses within Troy to get home. Besides, the residents in Troy are sparsely located around the center of Troy, with a population density 4796 people per square miles [28].

The bus system of capital district is provided by Capital District Transportation Authority (CDTA), which is running 59 routes with 306 fleets [29]. Troy is served by 15 bus routes. This work aims to coordinate the bus schedule during weekday evenings, because buses operate on reduced schedules with low frequency and long headways at that time. On weekday evenings, 9 routes provide service at the downtown bus transfer station (Fulton St. & 4th St.). Among them, 3 routes can be considered as trunk buses which are route 22 (Troy – Albany), route 70 (Troy – Schenectady), and route 90 (Troy–Latham–Crossgate Mall) [30]. They connect Troy with other cities in the Capital District and operate with longer distances and higher frequencies. The other 6 routes are local buses which only serve the city of Troy and its outskirts. A map of all routes serving the city of Troy in weekday evenings is shown in Fig. 1. The blue lines indicate the inter-city, or trunk, bus routes and the green lines indicate local routes. It is worth noting that route 85 does not use the downtown Troy station as its terminal so that its north bound and south bound services are treated as two separate routes. Most of these local bus routes run in circle with only one vehicle during the evenings, except for route 85, which is served by two vehicles.

After 6:30 pm on weekdays, the headway of buses becomes larger, especially for local buses. For example, bus 87 operates with a 20 min headway during the day, but a 1-hour headway during the evening. The current timetable is listed in Table 1 and the summary statistics of transfer waiting time is listed in Table 2. With the current schedule, passengers are very likely to wait a long time at the downtown station to transfer from a trunk bus to a local bus. For example, a passenger gets off bus line 70 and wants to take the bus line 289. No matter which bus he takes (7:21 pm., 8:21 pm., or 9:21 pm.), the waiting time will be 54 min, a quite long time.

Fig. 1 Evening bus routes map in Troy, New York, USA

In light of the inefficiency of the current bus schedule, this study targets at coordinating the bus schedules on weekday evenings in Troy.

Considering the limited resources available to transit operators in small conurbations, the improvement strategies need to be easily implementable and economical. That is, any significant cost increasing due the to increased service frequency, additional labor or extended service mileage should not be allowed.

Besides, given the importance of trunk lines, the schedules of trunk buses need to remain unchanged and the adjustment should be made to only local lines. Meanwhile, as the bus service is mainly provided to travelers with limited travel options, adequate service time needs to be ensured. Therefore, the latest service time cannot be earlier than the current time.

Table 1 Current timetable and round-trip durations of bus routes in Troy, New York China

Table 2 Current summary statistics of bus transfer waiting time (min) in Troy, New York, USA

4 Methodology

Thanks to the low congestion level in small conurbations, especially during the weekday evenings, bus service is usually reliable. The objective thus solely focuses on the reduction of transfer time:

                          (1)

where Z is the total waiting time at Troy’s downtown transfer station; denotes the waiting time for transferring from trunk bus route j to local bus route i’s mth service. For each round of local service m, the paired trunk service is selected as the one whose arrival time is closest to the local service’s departure time. Thus, the objective function is the summation of waiting time for all reasonable trunk-local transfers.

As transfer volumes do not differ significantly between pairs, this objective also implies that the target is to shorten the transfer time for all passengers. Although the ridership number of each transfer pair is hard to obtain, the assumption can still hold that weights are assigned evenly on each transfer pair.

The headway of each local route should be in a range that the requirements for vehicles and drivers will remain unchanged. Therefore, the headway of each local bus route should be longer than the travel time of one circle (duration_i) plus the drivers’ rest time between two consecutive services. The drivers’ rest time can be set based on the specific regulations. For example, the EU [31] mandates a 45 minutes’ break requirement after 4.5 hours’ drive. In this work, it is assumed that the break time should be at least 10% of the service time after each service circle. The headway condition is thus

       (2)

Finally, to accommodate possible service delays, dwelling requirements, walking time, and the other uncertainty, the transfering time should be at least 3 min. That is,

                                   (3)

The timetables of local buses are optimized subject to the above conditions.

5 Results analysis

The shifted timetables of local buses in Troy, NY during weekday evenings are listed in Table 3. The total waiting time of all transfer pairs has been reduced to 1318 min from the 1999 min before the timetable shifting. This is a 34.07% overall improvement in terms of time saving for passengers with the trunk-local transfer pattern.

A comparison of the before and after timetables shows that most shifts are within 10 min. Besides, the entire service period is close to the original service period with minimal extensions. Although the headway is no longer absolutely even, there are no extremely long headways for any bus route. Therefore, this timetable shift should have limited impacts on passengers who do not transfer or those who transfer between local buses. In terms of the capacity utilization of these buses, the shift is unlikely to cause any significant changes in passengers’ travel pattern or volume, which implies an unchanged comfort level for riders and an unchanged dwelling time.

Meanwhile, the coordinated timetable will provide the same number of services, use the same amount of labor force, and have the same fuel consumption. In other words, this timetable coordination has limited impacts on the service providers, which is very important for transit service operators in small conurbations who operate with extremely limited resources.

However, many passengers can save significant time at the transfer station with the shifted timetable. As shown by Table 4, passengers transferring to bus 289 from bus 70 will have greatly improved travel experience with an average transfer waiting time of 12 min. For passengers getting off bus 70 at 8:21 pm and 9:21 pm, the waiting time is shortened to 3 min from the current 54 min. The 51-min time saving is an impressive improvement and may encourage more passengers to use transit in the long term.

Table 3 Schedule optimization results

Table 4 Summary statistics of transfer waiting time (min) after coordination

6 Conclusions

1) This work proposes a procedure to optimize bus schedules that will effectively reduce the transfer waiting time in a trunk-local bus system that is typically seen in small conurbations. Results show that the overall passengers’ waiting time can be shortened remarkably. Meanwhile, as the timetable change is slight, impacts on the service providers and the passengers who do not make trunk-local transfers are limited.

2) This straightforward approach can be easily implemented in practice and adjusted based on local conditions. The method can be adapted to other transit systems in small conurbations as it builds on the basic characteristics of small conurbations which include low population density, private car-oriented transportation systems, and low social and economic activity frequencies. The transit system in a conurbation thus tends to present a trunk-local pattern, high reliability, and low service frequency. These features are considered in the proposed method, together with the fact these transit providers are constrained by extremely limited resources, and the transfer efficiency is improved in such a context.

3) Collectively, this work adds value to the existing literature by investigating transit transfer issues in small conurbations, which is important for the local residents yet largely understudied. Although the optimization method is not innovative, the consideration of small conurbation, a typical urban form in the United States, adds important values to the existing literature. For these small conurbations, the method and results of this study will help passengers on time saving and will be beneficial to the bus companies and regional developers who want to improve transfer efficiency, and consequently the overall transit service in an economical way.

References

[1] GUIHAIRE V, HAO J K. Transit network design and scheduling: A global review [J]. Transportation Research Part A: Policy and Practice, 2008, 42(10): 1251-1273.

[2] CEPEDA M, COMINETTI R, FLORIAN M. A frequency-based assignment model for congested transit networks with strict capacity constraints: characterization and computation of equilibria [J]. Transportation Research Part B: Methodological, 2006, 40(6): 437-459.

[3] HE J, LIU Z. Transfer convenience of urban rail transit based on passengers’ physical and conscious resistance [C]// Service Operations and Logistics, and Informatics, IEEE/SOLI 2008. IEEE International Conference on. Beijing: IEEE, 2008: 1536-1540.

[4] ZHANG G, CHEN Y, LI P, FIBBE S. Study on evaluation indicators system of crowd management for transfer stations based on pedestrian simulation [J]. International Journal of Computational Intelligence Systems, 2011, 4(6): 1375-1382.

[5] GUO Z, WILSON N H. Assessing the cost of transfer inconvenience in public transport systems: A case study of the London underground [J]. Transportation Research Part A: Policy and Practice, 2011, 45(2): 91-104.

[6] GOVERDE R M. Improving punctuality and transfer reliability by railway timetable optimization [D]. Delf: Department of Technical Mathematics, Faculty of Information Technology and Systems, Delft University of Technology, 1999.

[7] HADAS Y, CEDER A. Public transit network connectivity [J]. Transportation Research Record: Journal of the Transportation Research Board, 2010, 2143(1): 1-8.

[8] WARDMAN M. Public transport values of time [J]. Transport Policy, 2004, 11(4): 363-377.

[9] CEVALLOS F, ZHAO F. Minimizing transfer times in public transit network with genetic algorithm [J]. Transportation Research Record: Journal of the Transportation Research Board, 2006, 1971(1): 74-79.

[10] HSU S C. Determinants of passenger transfer waiting time at multi-modal connecting stations [J]. Transportation Research Part E: Logistics and Transportation Review, 2010, 46(3): 404-413.

[11] SHRIVASTAV P, DHINGRA S. Development of feeder routes for suburban railway stations using heuristic approach [J]. Journal of Transportation Engineering, 2001, 127(4): 334-341.

[12] GIULIANO G, NARAYAN D. Another look at travel patterns and urban form: the US and Great Britain [J]. Urban Studies, 2003, 40(11): 2295-2312.

[13] KRIZEK K J. Residential relocation and changes in urban travel: Does neighborhood-scale urban form matter? [J]. Journal of the American Planning Association, 2003, 69(3): 265-281.

[14] STEAD D, MARSHALL S. The relationships between urban form and travel patterns. An international review and evaluation [J]. European Journal of Transport and Infrastructure Research, 2001, 1(2): 113-141.

[15] Merriam-Webster. Conurbation. [EB/OL]. [2014-07-31]. http://www. merriam-webster.com/dictionary/conurbation.

[16] CEDER A, GOLANY B, TAL O. Creating bus timetables with maximal synchronization [J]. Transportation Research Part A: Policy and Practice, 2001, 35(10): 913-928.

[17] FREDA P, SCHMIER K J. Public transit vehicle arrival information system. U. S. Patent, No. 6006159[P]. 1999.

[18] DAILEY D, MACIEAN S, CATHEY F, WALL I. Transit vehicle arrival prediction: Algorithm and large-scale implementation [J]. Transportation Research Record: Journal of the Transportation Research Board, 2001, 1771(1): 46-51.

[19] LAW P, TAYLOR B D. Shelter from the storm: Optimizing distribution of bus stop shelters in Los Angeles [J]. Transportation Research Record: Journal of the Transportation Research Board, 2001, 1753(1): 79-85.

[20] TYREE J D. Assessing transit service performance: Recommended performance standards for the Santa Clara Valley transportation authority [R]. San Jose: American Public Transportation Association, 2010.

[21] ISEKI H, TAYLOR B D. Style versus service? An analysis of user perceptions of transit stops and stations [J]. Journal of Public Transportation, 2010, 13(3): 23-48.

[22] LITMAN T. Valuing transit service quality improvements [J]. Journal of Public Transportation, 2008, 11(2): 43-63.

[23] D, P. Schedule synchronization in public transit using the fuzzy ant system [J]. Transportation Planning and Technology, 2005, 28(1): 47-76.

[24] CEVALLOS F, ZHAO F. A genetic algorithm for bus schedule synchronization. in applications of advanced technology in transportation [C]// The Ninth International Conference of ASCE, Chicago: ASCE, 2006: .

[25] TING C J, SCHONFELD P. Schedule coordination in a multiple hub transit network [J]. Journal of Urban Planning and Development, 2005, 131(2): 112-124.

[26] SHRIVASTAVA P, DHINGRA S. Development of coordinated schedules using genetic algorithms [J]. Journal of Transportation Engineering, 2002, 128(1): 89-96.

[27] SUN L, RONG J, YAO L. Measuring transfer efficiency of urban public transportation terminals by data envelopment analysis [J]. Journal of Urban Planning and Development, 2010, 136(4): 314-319.

[28] City-Data. Troy, NY. [EB/OL]. [2014-07-31]. http://www.city-data. com/city/Troy-New-York.html.

[29] CDTA. About CDTA. [EB/OL]. [2014-07-31]. https://www.dta.org/ bout. hp.

[30] Capital District Transportation Authority. Routes and schedule. [EB/OL]. [2012-03-01]. http://www.cdta.org/.

[31] European Union Regulation. Regulation (EC) No 561/2006 of the European parliament and of the Council of 15 March 2006 on the harmonisation of certain social legislation relating to road transport and amending Council Regulation (EEC) No 3821/85 and (EC) No 2135/98 and repealing Council Regulation (EEC) No 3820/85. [EB/OL]. [2014-07-24]. http://eur-lex.europa.eu/legal-ontent/EN/N OT/?uri=CELEX:32006R0561.

(Edited by DENG Lü-xiang)

Received date: 2015-09-20; Accepted date: 2016-03-01

Corresponding author: WANG Xiao-kun, PhD, Professor; Tel: +1-518-276-2098; E-mail: wangx18@rpi.edu