Optimized Skip-Stop Metro Line Operation Using Smart Card Data

Table 1

Existing skip-stop schemes for metro operation.

Publication	Skip-stop mode	Input data type/source	Single/bidirectional line	Objective	Model	Solution approach	Results
Suh et al. [12]	Fixed A/B	Static OD	Single	Maximize the total time saving of a skip-stop schedule in both peak and off-peak period	Train operation simulation model	Calculate the total travel time of five passenger boarding types	Per the specific scenario, the skip-stop operation results in less total travel saving in the peak period compared to the off-peak period
Zheng et al. [16]	Fixed A/B	Static OD with 13 stations	Single	Minimize total passenger travel time	A 0-1 integer programming model by analyzing the passenger traveling time	Tabu search	It is better to assign a type AB train between a type A train and type B train
Freyss et al. [10]	Fixed A/B	Assumed static OD from Chilean metro data	Single	Minimize the cost function that is comprised of total passenger transfer time, waiting time, and total traction and energy cost	Consider five passenger types under different combination of A or B or AB stations	Continuous approximation approach	Short lines with fewer stations are less favorable for skip-stop operation; larger benefit can be obtained in the lines with smaller minimum headway
Lee et al. [13]	Fixed A/B	Manipulated static OD based on boarding and alighting ratio from the Seoul Metro line	Single	Minimize the passenger travel time	Consider three types (A, B, and AB) of passenger OD and collision constraints in four scenarios	Genetic algorithm	17%–20% travel time saving for passengers and reduced operational cost for the operator
Cao et al. [15]	Fixed A/B	Simulated static OD based on Beijing Metro data	Single	Biobjective formulation that minimizes the passenger waiting time and trip time	A 0-1 mixed integer model that considers the constraints of two successive trains; train operation simulation	Tabu search	Smaller headway leads to better passenger waiting time and trip time
Sogin et al. [17]	Dynamic	Simulated static OD using gravity model/Chicago Metro lines	Single	Minimize the total passenger travel time	A mixed integer formulation that considers the tradeoff between passenger travel time and service frequency caused by skip-stop	GAMS and CPLEX for small scale problems; genetic algorithm for large-scale problems	Skip-stop operation can reduce passenger travel time by 9.5%
Wang et al. [18]	Dynamic	Static OD based on real data from the Beijing Metro (Yizhuang) Line	Bidirectional; consider turnaround terminal as an intermediate station without storage capacity	Minimize the total passenger travel time and energy consumption	Consider the impact of boarding passenger number on the train dwell time and train utilization at the departure terminal	Bilevel mixed integer programming model with rolling horizon	Time and energy saving up to 15%
Niu et al. [19]	Predetermined skip-stop	Time-dependent demand for nine stations of the Shanghai-Hangzhou High-Speed Rail Line in China	Single	Minimize the total passenger waiting time	A unified quadratic integer programming model with linear constraints; consider safe operation and train capacity	GAMS solver	The model can be solved by standard commercial optimization packages with good computation time
Jamili and Aghaee [14]	Uncertain skip-stop	Static OD based on an Iranian metro line	Single	Minimize the total passenger travel time	A nonlinear programming model that considers boarding passengers and train capacity	Decomposition-based algorithm and simulated annealing (SA) algorithm	Trip time is reduced from 92.3 minutes to 86.6 minutes; the average speed increases by 3.4 km/h
Our approach	Flexible	Time-dependent arrival rate based on smart card data from the Shenzhen Metro line	Bidirectional; consider departure time optimization and storage capacity at the turnaround terminal	Minimize the average travel time	Three types of headways are considered; train operation simulation; turnaround operation	Genetic algorithm	See numerical section