Available online at www.sciencedirect.com
ScienceDirect Procedia Engineering 187 (2017) 77 – 81
10th International Scientific Conference Transbaltica 2017: Transportation Science and Technology
Class Library for Simulations of Passenger Transfer Nodes as Elements of the Public Transport System Vitalii Naumova,*, Ganna Samchukb b
a Cracow University of Technology, Poland Kharkiv National Automobile and Highway University, Ukraine
Abstract Transfer nodes are essential elements of public transport systems which provide door-to-door transport services for passengers. Parameters of the technological processes in public transport systems are stochastic variables, thus, computer simulations are usually used for solving optimization problems of public transport. There is a number of simulation tools supporting decisionmaking in public transportation, but they don’t provide the flexibility for solving the transfer nodes optimization problems. Authors present a library of classes implemented in Python, which could be used for computer simulations of public transfer nodes. The proposed software allows researchers to change technological parameters during simulation procedures and makes possible automatization of simulation experiments in the field of passengers’ transportation. © 2017 2017The TheAuthors. Authors. Published by Elsevier Ltd.is an open access article under the CC BY-NC-ND license © Published by Elsevier Ltd. This Peer-review under responsibility of the organizing committee of the 10th International Scientific Conference Transbaltica 2017: (http://creativecommons.org/licenses/by-nc-nd/4.0/). Transportation and Technology. Peer-review underScience responsibility of the organizing committee of the 10th International Scientific Conference Transbaltica 2017 Keywords: public transportation, transfer nodes, computer simulations, Python code, specialized classes’ library, optimization methods
1. Introduction In the contemporary engineering science, the use of simulation tools is necessary while planning and analyzing systems at the macrolevel, especially while solving the problems of public transport systems. Processes in public transportation are characterized by the influence of a large number of different random parameters on the system
* Corresponding author. E-mail address:
[email protected]
1877-7058 © 2017 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the organizing committee of the 10th International Scientific Conference Transbaltica 2017
doi:10.1016/j.proeng.2017.04.352
78
Vitalii Naumov and Ganna Samchuk / Procedia Engineering 187 (2017) 77 – 81
elements. In a consequence, all the resulting characteristics of these elements and a public transport system as a whole are stochastic variables. Estimations of the resulting parameters as stochastic variables cannot be performed in practice without using of computer simulations. Transfers are unavoidable for the creation of an efficient transit route network as it is impossible to ensure the door-to-door access in all directions without transfer nodes. The passenger transfer nodes development is an essential problem to be solved in order to maximize service coverage, support multimodal transportation and ensure sustainable connectivity. The presence of transfer nodes in a public transport system makes it possible to assign routes for a network with optimal headways, vehicle sizes etc., and, as a result, to meet passenger demand and minimize operation costs. However, uncoordinated transfers decrease the service quality and the attractiveness of public transportation to passengers due to additional waiting time. In a process of decision-making on sustainable development of the public transport systems, transportation engineers often use approaches based on analytical models, but the adequacy of such models is much lower in compare to simulation models. The simulation models of public transport systems allow researchers and engineers to consider stochastic parameters of technological processes. In addition, the inner relations between the system elements and input parameters could be described with high precision, and this makes the simulation models of public transport systems much more adequate than analytical ones. 2. Review
of existing optimization problems in passenger transfer nodes
The main objectives of transfer optimization usually are minimization of the passengers waiting time and maximization of the number of arrivals due to synchronization (simultaneous arrivals of public transport vehicles) at the transport system interchanges. The existing literature considers the trade-off between the passenger waiting time and operating costs [1, 2] as well. There also exist multi-objective optimization approaches, such as a model formulated for the multi-objective re-synchronizing of bus timetable problem by Yinghui Wua et al. [3]. As scheduling problems are recognized to be an NP-hard problem, it is difficult to solve them with exact algorithms, so different technics, such as Tabu search method, simulated annealing, genetic algorithms, iterated local search, branch-and-bound, local search algorithms are applied to obtain some rational solution. The model aimed at minimization of the total waiting time of all passengers at transfer nodes was formulated on the basis of the Fuzzy Ant System by Teodorović & Lućić [4].The authors of this model emphasize that the transfer waiting time depends on schedule synchronization. But it should be noted, that stochastic parameters of the demand for public transportation (such as interval of passengers appearance at public transport stops) influences the waiting time as well. Ceder et al. developed a mixed integer programming (MIP) model for the problem of generating a timetable [5]. It is objected to maximize the number of simultaneous bus arrivals at the connection nodes. The bus travel time is assumed to be deterministic. Schroder & Solchenbach tackled the optimization of transfer quality and modeled the problem of timetable synchronization as a quadratic semi-assignment problem [6]. They distinguish five transfer types, which vary from “Convenience” to “No transfer” and propose a penalty-function for each of them. In order to develop realistic and adequate models it is necessary to take into account randomness of different public transport attributes. The existing literatures also propose a timetable synchronization models considering uncertainties. Stochastic disturbances appear due to the variation of traffic intensity over time, traffic jams, weather condition, driver´s behaviour etc. Moreover, any demand change results in the dwell time deviation. The mentioned uncertainties lead to increasing the variability of the travel time and diminishing service reliability. A stochastic programming model for metro train rescheduling problem was proposed by Yin et al. [7]. The developed model includes uncertain time-variant passenger demands. The rescheduling problem was successfully solved due to the use of an approximate dynamic programming based algorithm. Some published timetable synchronization models consider uncertainties in the travel times. Focusing on transfer optimization, Bookbinder & D’esilets proposed to combine a simulation procedure and an optimization model (relaxation of the Quadratic Assignment Problem) [8]. The developed model considers travel time as a stochastic variable.
Vitalii Naumov and Ganna Samchuk / Procedia Engineering 187 (2017) 77 – 81
At the planning stage the travel time variability is commonly compensated by adding a slack time onto a schedule, while a holding strategy is used in case of dynamic synchronization (real time control). However, the main drawback of the former strategy is that it leads to increase in-vehicle time for through passengers. The latter strategy needs precise real-time data of vehicles arrivals at a transfer node. Lee & Schonfeld studied transfers between one bus route and one rail line. Their research [9] is aimed at optimization of slack time that is added in the schedules to avoid miss connections. Ting & Schonfeld also discuss a timed transfer concept and attempt to optimize the headways and slack times by using heuristic algorithm [10]. The objective is minimization of the total costs of operating a multiple-hub transit network. Nesheli et al. attempted to use real-time control actions to reduce bus-bunching and maximize transfer synchronization [11]. The vehicle travel time and passengers arrive moments are considered to be as a random variables in the simulation model. The network was simulated using the ExtendSim software. The authors reported that the research results are sensitive to the passenger demand variation. A stochastic mixed integer programming model for robust coordination schedule scheme is proposed by Wu et al. The authors applied a branch-and-bound algorithm to solve the problem. Their paper [12] addresses the stochastic travel time at two stages namely planning and operation. As we can conclude, authors of existing approaches consider parameters of technological processes and efficiency criteria as stochastic variables to solve optimization problems in transfer nodes. The described numerical methods and algorithms could be used in some cases, but only if quite specific assumptions are fulfilled. The adequate and versatile way to describe the process of passengers servicing (the process of the public transport system functioning) as a complex stochastic system is to model them using advanced simulation software. In the software market segment of simulation tools for modeling and planning of public transport systems, there are many solutions, among which Aimsun and PTV Visum are the most popular. But we consider that these tools are not enough agile for their use in scientific experiments. Although Aimsun and PTV Visum have quite advanced functions for simulations of the technological processes in transfer nodes, they cannot be used in order to generate random parameters of technological processes and stochastic characteristics of demand for public transport services. Technological parameters of passenger transfer nodes also can’t be changed during simulation procedures, if the mentioned software is applied for simulations. The use of Aimsun and PTV Visum tools is quite troublesome when multiple simulations are needed for varied input data sets: the data for each simulation in this case should be entered manually. Development of the specialized software for simulations of transport processes is more agile solution, but it requires certain qualifications of a researcher. 3. Proposed software for simulations of transfer nodes as elements of a public transport system Public transfer nodes should be simulated as elements of a public transport system. All the technological processes in transfer nodes are defined or influenced by processes of other elements of the public transport system. To model processes of the public transport systems functioning for solving scientific problems, we developed the specialized library of base classes. The classes’ implementation was performed with the use of the Python programming language; it ensures compatibility of the developed software with the most popular environments for modeling of public transport systems (including Aimsun and PTV Visum). The developed code of the mentioned base classes is available in open access and could be downloaded from [13]. As the base classes, on the grounds of which simulation models of public transport systems should be implemented, we consider: • Net: is used in order to develop the software implementation of a transport network model as an oriented weighted graph; • Node: allows researchers to model points of the transport network as the graph nodes; the transport net points could be considered in a simulation model as software implementation of the public transport stops, transfer passenger nodes of intersections of the road network;
79
80
Vitalii Naumov and Ganna Samchuk / Procedia Engineering 187 (2017) 77 – 81
• Link: represents a software implementation of a link in a graph, which is used in order to describe a transport system model; the graph link could be used in simulation models for modeling segments of the road network or spans of the public transport lines; • Line: could be used in order to model a public transport line; is defined for the software implementation of a road network as an object of the Net class; • Vehicle: allows to model a vehicle as an element of the transport system model; is used for developing simulation models of the public transport lines; • Passenger: is an abstraction for implementation of passengers as transport system elements; an object of this type is a unit used for description of demand for services of the public transport within the framework of a simulation model of the transport system. To simulate parameters which describe the environment influence on the transport system, in the proposed library the Stochastic class was developed. Implementation of the Stochastic class objects allows to model a random variable with the defined distribution and given numerical parameters. The structure of the proposed class library in presented as the UML diagram at Fig. 1. ϭ ϭ
ϭ
EĞƚ ϭ
ϭ͘͘Ύ
>ŝŶŬ
Ϯ͘͘Ύ Ϭ͘͘Ύ
Ϭ͘͘Ύ
EŽĚĞ Ϯ͘͘Ύ
ϭ͘͘Ύ
>ŝŶĞ ϭ
ϭ
ϭ
Ϭ͘͘Ύ
Ϭ͘͘Ύ
WĂƐƐĞŶŐĞƌ ϭ͘͘Ύ
sĞŚŝĐůĞ
ϭ
Ϭ͘͘Ύ
Fig. 1. UML-diagram of the proposed library for simulations of public transport systems.
The main class, on the base of which implementation of the transport system simulation model could be performed, is the Net class. An object of this type is presented in a simulation model in a single exemplar, it is used in order to form the road network model, to define the public transport lines, to generate demand for trips in the bounds of the transport network, and to run simulations of the transport system. To run simulations, an object of the Net class should contain at least two objects of the Node type (contain at least two transfer nodes), at least one object of the Link class (at least one connection between the transfer nodes should be defined), and at least one object of the Line class (at least one line of public transport should be added to the transport network), but an empty set of the Passenger objects is allowed in the demand model. Besides implementation of the demand model, the Passenger class objects could be used for implementation of the Node class objects (while demand generation for passenger transfer nodes as elements of the transport network) and the Vehicle type objects (for simulations of the processes of passengers servicing by the vehicle). While creating objects of the Line class, in order to define the public line route, the collection of elements of the Node type is being used (the public line should contain at least two stops); to describe vehicles, servicing the passengers at the public line, objects of the Vehicle type are being implemented (the public line should be serviced at least by one vehicle). In the process of implementation of simulation models for public transport systems, the certain values are being assigned to the class fields of developed objects; to do this, the developed class methods and properties are being
Vitalii Naumov and Ganna Samchuk / Procedia Engineering 187 (2017) 77 – 81
used. The class fields have numeric data type or more advanced types – collections and dictionaries. Class methods are used in order to perform initialization procedures or to simulate the processes of the transport systems functioning. Class properties allow developers to calculate numeric characteristics of the simulated objects on the base of inner class values. The procedures for simulations of the technological processes in passenger transfer nodes are implemented in the Node class. As far as the described library is available as an open code, additional functionality could be added to the proposed simulation tools. 4. Conclusions Processes in passenger transfer nodes should be considered in the bounds of the higher level system – the public transport system. Technological processes of passengers service in transfer nodes are stochastic, they are characterized by a number of random parameters. In order to obtain an adequate model of any passenger transfer node, the advanced simulation software should be used. The proposed library of the base classes allows researchers and transport engineers to develop agile simulation models of passenger transfer nodes, which could be used in experimental studies and for choosing the effective technological schemes of the passengers service. Using the library, presented in this paper, we’ve developed a set of simulation models, namely: a model of transfer nodes functioning, a model of a public transport line and a model of a public transport network. On the grounds of simulations with the use of mentioned models, solutions for complex problems in the area of public transportations were obtained, such as synchronization of bus lines in a transfer node, estimation of the optimal bus number for a public transport line, designing the rational network for a public transport system, etc. References [1] G. Bruno, G. Improta, A. Sgalambro, Optimization of the departure schedule at a public transit terminal with multiple destinations, Multidisciplinary International Conference on Scheduling: Theory and Application (2007) 104–111. [2] P. Shrivastava, S. L. Dhingra, Development of coordinated schedules using genetic algorithms, Journal of Transportation Engineering 128(1) (2002) 89–96. [3] Y. Wua, H. Yangb, J. Tangc, Y. Yud, Multi-objective re-synchronizing of bus timetable: Model, complexity and solution. Transportation Research Part C 67 (2016) 149–168. [4] D. Teodorović, P. Lučić, Schedule synchronization in public transit using the fuzzy ant system, Transportation Planning and Technology, 28(1) (2005) 47−76. [5] A. Ceder, B. Golang, O. Tal, Creating bus timetables with maximal synchronization, Transportation Research Part A 35(10) (2001) 913–928. [6] M. Schroder, I. Solchenbach, Optimization of transfer quality in regional public transit: Technical Report, Fraunhofer ITWM. 2006. [7] J. Yin, T. Tang, L. Yang, Z. Gao, B. Ran, Energy-efficient metro train rescheduling with uncertain time-variant passenger demands: An approximate dynamic programming approach, Transportation Research Part B 91 (2016) 178–210. [8] J. H. Bookbinder, A. Desilets, Transfer optimization in a transit network, Transportation Science 26(2) (1992) 106–118. [9] M. Lee, P. Schonfeld, Optimal slack time for timed transferred at transit terminal, J. Adv. Transp. 25(3) (1991) 281–308. [10] C. Ting, P. Schonfeld, Schedule coordination in a multiple hub transit network, Journal of Urban Planning and Development131(2) (2005) 112–124. [11] M. M. Nesheli, A. A. Ceder, V. A. Gonzalez, Real-time public-transport operational tactics using synchronized transfers to eliminate vehicle bunching, IEEE Transactions on Intelligent Transportation Systems17(11) (2016) 3220–3229. [12] W. Wu, R. Liu , W. Jin, Designing robust schedule coordination scheme for transit networks with safety control margins, Transportation Research Part B 93A (2016) 495–519. [13] V. Naumov, Python code for simulation of the public transport network. Available from Internet: https://www.academia.edu/28749786/ Python_Code_for_Simulation_of_the_Public_Transport_Network.
81
