TY - JOUR
T1 - Solving large-scale train timetable adjustment problems under infrastructure maintenance possessions
AU - Van Aken, Sander
AU - Bešinović, Nikola
AU - Goverde, Rob M.P.
PY - 2017/12/1
Y1 - 2017/12/1
N2 - During infrastructure maintenance possessions, commonly not all trains can operate, and the original timetable may have to be adjusted accordingly. To deliver the best service to passengers, operators have to coordinate adjustment measures dealing with multiple possessions at the network level. In this paper, we consider the Train Timetable Adjustment Problem (TTAP) and present a mixed integer programming (MIP) model for solving TTAP. In order to solve large-scale problems, such as national Dutch network, and design high-quality solutions, modelling extensions are needed. First, we apply three network aggregation techniques to decrease the problem size, which enables to solve instances on the complete Dutch network within satisfactory computation times. Second, we model turnaround activities for short-turned trains and test different strategies. Third, we introduce flexible short-turning possibilities to the MIP to possibly reduce the number of cancelled train lines. We test the proposed model on real-life cases of Netherlands Railways (NS) and assess the effect on computation times and solution quality. Also, we identify differences with current planners’ practice. Planners were positive about the quality of generated solutions and the computation speed. The current model can also be used to decide on combinations of time windows for possessions.
AB - During infrastructure maintenance possessions, commonly not all trains can operate, and the original timetable may have to be adjusted accordingly. To deliver the best service to passengers, operators have to coordinate adjustment measures dealing with multiple possessions at the network level. In this paper, we consider the Train Timetable Adjustment Problem (TTAP) and present a mixed integer programming (MIP) model for solving TTAP. In order to solve large-scale problems, such as national Dutch network, and design high-quality solutions, modelling extensions are needed. First, we apply three network aggregation techniques to decrease the problem size, which enables to solve instances on the complete Dutch network within satisfactory computation times. Second, we model turnaround activities for short-turned trains and test different strategies. Third, we introduce flexible short-turning possibilities to the MIP to possibly reduce the number of cancelled train lines. We test the proposed model on real-life cases of Netherlands Railways (NS) and assess the effect on computation times and solution quality. Also, we identify differences with current planners’ practice. Planners were positive about the quality of generated solutions and the computation speed. The current model can also be used to decide on combinations of time windows for possessions.
KW - Maintenance
KW - Periodic Event Scheduling Problem (PESP)
KW - Possessions
KW - Railway timetable
KW - Train Timetable Adjustment Problem (TTAP)
UR - http://www.scopus.com/inward/record.url?scp=85025467434&partnerID=8YFLogxK
UR - http://resolver.tudelft.nl/uuid:e87e5323-9e1a-4ddb-9561-22c7f3735324
U2 - 10.1016/j.jrtpm.2017.06.003
DO - 10.1016/j.jrtpm.2017.06.003
M3 - Article
AN - SCOPUS:85025467434
SN - 2210-9706
VL - 7
SP - 141
EP - 156
JO - Journal of Rail Transport Planning & Management
JF - Journal of Rail Transport Planning & Management
IS - 3
ER -