Optimal train rescheduling after conflict detection Sofie Van Thielen, Sofie Burggraeve, Pieter Vansteenwegen KU Leuven, Leuven Mobility Research Centre – CIB [email protected]
Jan 18, 2016
PowerPoint-presentatie
Optimal train rescheduling after conflict detectionSofie Van Thielen, Sofie Burggraeve, Pieter Vansteenwegen
KU Leuven, Leuven Mobility Research Centre [email protected] descriptionScope of this workStrategiesResultsConclusion & further work22MotivationNormal situation:
Conflict:
3How primary delay leads to secondary delays3MotivationIn practice:Real-time management is performed manually.
44Real-time traffic managementTrain movement prediction
Conflict detection
Conflict resolutionDecision Support System55Problem descriptionGiven one (or more) conflicts, present different good solutions with the corresponding total delay. In a tool that is directly accessible for the dispatchers. 66Scope of this workStar-shaped network
Both freight and passenger trains
Only small perturbations
Conflict resolution: only rescheduling (retiming and reordering)77StrategiesFirst Come, First Serve (FCFS)
8
AB
AB
892) Priority to Type of Train (PTT)
Every train has a type name, which characterises the speed and stopping pattern. Least amount of stops and high speed highest typeAssumption: economic cost of delaying freight train < economic cost of delaying passengers
A
B10If a conflict occurs, then the train with the highest type gets preference.
THAL
THAL103) Conflict limitation (CL)
Minimising the effect of conflictsBased on examining the progress during the next hourConsider both options and choose the train that induces the least total delay
11
AB
11
AB
CAStep 1a: train A first12
1213
ABStep 1b: progress of train B
B
C14Step 1c: effect of train A on train B
AB
B
A
AB
CBStep 2a: train B first15
15
AB
DAStep 2b: progress of train A16
B1617Step 2c: effect of train B on train A
BA
A
BSimulation182-hour window: 7-9 AM
82 trains
Delay scenarios
Study area
18Study areaBrugge-Gent-Denderleeuw (in Belgium)
19DenderleeuwBruggeGentOostende19Further improvements Prevent deadlock situations
Increase entrance delays
20
20Preliminary results
2121Computation time
2222Further improvements (CL)23Consider total delay instead of conflict duration
Also consider the effect of train A on B (and vice versa)
A
B23Results24
24Computation time25
Conclusion & further workSimulation based on actual data from Infrabel works well.
Decide on the time window to consider.
Include more trains create more conflicts
Include rerouting options even only in station areas this will give good solutions.
2626Questions?
27Optimal train rescheduling after conflict detectionSofie Van Thielen, Sofie Burggraeve, Pieter Vansteenwegen
KU Leuven, Leuven Mobility Research Centre [email protected]