Operational design for shuttle systems with modular vehicles under oversaturated traffic

Discrete modeling method

Zhiwei Chen, Xiaopeng Li, Xuesong Zhou

Research output: Contribution to journalArticle

1 Citation (Scopus)

Abstract

Existing peak/off-peak based schedules in urban mass transit systems feature two types of dispatch headways and a fixed vehicle capacity across the operational horizon, which lowers their service quality and causes significant energy waste. A promising cure to this challenge is to jointly design the dispatch headways and vehicle capacities in urban transit schedules. Chen et al. (2018) propose a continuous modeling method to solve the near-optimum solutions to and shed analytical insights into this joint design problem. Based on the theoretical properties discovered in the preceding paper, this paper formulates the joint design problem as a mixed integer linear programming model that can yield exact solutions to the optimal design with a discretized time representation. Further, a customized DP algorithm is proposed to solve this model. Similar to other problems that can be solved by DP algorithms, the “curse of dimensionality” also exists in the investigated problem since the queue length, as a state variable, may have a large set of possible values at each stage, which may lead to dramatically increasing computational time in solving the problem. To expedite the solution speed of the DP algorithm, we propose a set of valid inequalities based on the relationship between the queue length and vehicle capacity. These valid inequalities can reduce the unboundedly increasing state space into a narrow band and thus dramatically expedite the DP algorithm. With two sets of numerical experiments, we show that the discrete model can be solved by the customized DP algorithm to optimality with much less computation time compared with a state-of-the-art commercial solver, Gurobi. The analysis also reveals that the input parameters affect the effectiveness of dynamic capacity design under oversaturated and unsaturated traffic systems in a similar way despite some minor differences.

Original languageEnglish (US)
Pages (from-to)1-19
Number of pages19
JournalTransportation Research Part B: Methodological
Volume122
DOIs
StatePublished - Apr 1 2019

Fingerprint

traffic
Linear programming
programming
energy
cause
experiment
time
Values
Experiments

Keywords

  • Dynamic capacity design
  • Dynamic programming
  • Modular vehicles
  • Oversaturated traffic
  • Valid inequalities
  • Vehicle scheduling

ASJC Scopus subject areas

  • Civil and Structural Engineering
  • Transportation

Cite this

Operational design for shuttle systems with modular vehicles under oversaturated traffic : Discrete modeling method. / Chen, Zhiwei; Li, Xiaopeng; Zhou, Xuesong.

In: Transportation Research Part B: Methodological, Vol. 122, 01.04.2019, p. 1-19.

Research output: Contribution to journalArticle

@article{f3a5b1e77a8d4b1db6223c5ba5c7c87f,
title = "Operational design for shuttle systems with modular vehicles under oversaturated traffic: Discrete modeling method",
abstract = "Existing peak/off-peak based schedules in urban mass transit systems feature two types of dispatch headways and a fixed vehicle capacity across the operational horizon, which lowers their service quality and causes significant energy waste. A promising cure to this challenge is to jointly design the dispatch headways and vehicle capacities in urban transit schedules. Chen et al. (2018) propose a continuous modeling method to solve the near-optimum solutions to and shed analytical insights into this joint design problem. Based on the theoretical properties discovered in the preceding paper, this paper formulates the joint design problem as a mixed integer linear programming model that can yield exact solutions to the optimal design with a discretized time representation. Further, a customized DP algorithm is proposed to solve this model. Similar to other problems that can be solved by DP algorithms, the “curse of dimensionality” also exists in the investigated problem since the queue length, as a state variable, may have a large set of possible values at each stage, which may lead to dramatically increasing computational time in solving the problem. To expedite the solution speed of the DP algorithm, we propose a set of valid inequalities based on the relationship between the queue length and vehicle capacity. These valid inequalities can reduce the unboundedly increasing state space into a narrow band and thus dramatically expedite the DP algorithm. With two sets of numerical experiments, we show that the discrete model can be solved by the customized DP algorithm to optimality with much less computation time compared with a state-of-the-art commercial solver, Gurobi. The analysis also reveals that the input parameters affect the effectiveness of dynamic capacity design under oversaturated and unsaturated traffic systems in a similar way despite some minor differences.",
keywords = "Dynamic capacity design, Dynamic programming, Modular vehicles, Oversaturated traffic, Valid inequalities, Vehicle scheduling",
author = "Zhiwei Chen and Xiaopeng Li and Xuesong Zhou",
year = "2019",
month = "4",
day = "1",
doi = "10.1016/j.trb.2019.01.015",
language = "English (US)",
volume = "122",
pages = "1--19",
journal = "Transportation Research, Series B: Methodological",
issn = "0191-2615",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - Operational design for shuttle systems with modular vehicles under oversaturated traffic

T2 - Discrete modeling method

AU - Chen, Zhiwei

AU - Li, Xiaopeng

AU - Zhou, Xuesong

PY - 2019/4/1

Y1 - 2019/4/1

N2 - Existing peak/off-peak based schedules in urban mass transit systems feature two types of dispatch headways and a fixed vehicle capacity across the operational horizon, which lowers their service quality and causes significant energy waste. A promising cure to this challenge is to jointly design the dispatch headways and vehicle capacities in urban transit schedules. Chen et al. (2018) propose a continuous modeling method to solve the near-optimum solutions to and shed analytical insights into this joint design problem. Based on the theoretical properties discovered in the preceding paper, this paper formulates the joint design problem as a mixed integer linear programming model that can yield exact solutions to the optimal design with a discretized time representation. Further, a customized DP algorithm is proposed to solve this model. Similar to other problems that can be solved by DP algorithms, the “curse of dimensionality” also exists in the investigated problem since the queue length, as a state variable, may have a large set of possible values at each stage, which may lead to dramatically increasing computational time in solving the problem. To expedite the solution speed of the DP algorithm, we propose a set of valid inequalities based on the relationship between the queue length and vehicle capacity. These valid inequalities can reduce the unboundedly increasing state space into a narrow band and thus dramatically expedite the DP algorithm. With two sets of numerical experiments, we show that the discrete model can be solved by the customized DP algorithm to optimality with much less computation time compared with a state-of-the-art commercial solver, Gurobi. The analysis also reveals that the input parameters affect the effectiveness of dynamic capacity design under oversaturated and unsaturated traffic systems in a similar way despite some minor differences.

AB - Existing peak/off-peak based schedules in urban mass transit systems feature two types of dispatch headways and a fixed vehicle capacity across the operational horizon, which lowers their service quality and causes significant energy waste. A promising cure to this challenge is to jointly design the dispatch headways and vehicle capacities in urban transit schedules. Chen et al. (2018) propose a continuous modeling method to solve the near-optimum solutions to and shed analytical insights into this joint design problem. Based on the theoretical properties discovered in the preceding paper, this paper formulates the joint design problem as a mixed integer linear programming model that can yield exact solutions to the optimal design with a discretized time representation. Further, a customized DP algorithm is proposed to solve this model. Similar to other problems that can be solved by DP algorithms, the “curse of dimensionality” also exists in the investigated problem since the queue length, as a state variable, may have a large set of possible values at each stage, which may lead to dramatically increasing computational time in solving the problem. To expedite the solution speed of the DP algorithm, we propose a set of valid inequalities based on the relationship between the queue length and vehicle capacity. These valid inequalities can reduce the unboundedly increasing state space into a narrow band and thus dramatically expedite the DP algorithm. With two sets of numerical experiments, we show that the discrete model can be solved by the customized DP algorithm to optimality with much less computation time compared with a state-of-the-art commercial solver, Gurobi. The analysis also reveals that the input parameters affect the effectiveness of dynamic capacity design under oversaturated and unsaturated traffic systems in a similar way despite some minor differences.

KW - Dynamic capacity design

KW - Dynamic programming

KW - Modular vehicles

KW - Oversaturated traffic

KW - Valid inequalities

KW - Vehicle scheduling

UR - http://www.scopus.com/inward/record.url?scp=85061126423&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85061126423&partnerID=8YFLogxK

U2 - 10.1016/j.trb.2019.01.015

DO - 10.1016/j.trb.2019.01.015

M3 - Article

VL - 122

SP - 1

EP - 19

JO - Transportation Research, Series B: Methodological

JF - Transportation Research, Series B: Methodological

SN - 0191-2615

ER -