Presented at WCTR 2016 by PhD candidate Ying Wang www.wctrs-conference.com www.its.leeds.ac.uk/people/y.wang

1. Institute for Transport Studies
FACULTY OF ENVIRONMENT
Railway rescheduling under
adverse weather conditions
on a single-track line
Ying Wang, Dr. Ronghui Liu, Dr. Raymond Kwan
Contact: tsyw@leeds.ac.uk

2. 2
Contents
• Introduction
• The weather impact to a railway line
• Methodology
• Case study
• Summary

3. 3
Introduction
Weather related delay compensation cost between 2006 and 2014.
(source: Network Rail Weather Analysis Report, UK)

4. 4
Introduction
Single track
Outbound
S (1) S (2) S (3) S (4) S (5)
InboundSidings
segment
station
A single-track line
One segment between two stations
Two sidings in a station
Time (min)
Distance(km)
Trains’ operation trajectory/ Schedule

5. 5
How weather is considered
Train timetable under normal condition
Weather impact is
considered the same
as UISDs happened
on specific trains.
PCDDs on tracksDifference
Proposed
method
Present method
Introduction
The Stochastic disturbance
The weather impact
Modeled as speed reduction on tracks
Modeled as delay
on trains
UISDs: unpredictable, independent and static disturbances
PCDDs: predictable, compounded and dynamic disturbances
NB: here we don’t consider the sever condition

6. 6
: UISDs : : PCDDs: Weather
Predictable and Compounded
Space
(b) UISDs Rescheduling approach
1st
R 2nd
R Time R Time
Present time
Space
(a) PCDDs Rescheduling approach
Introduction
UISDs VS PCDDs rescheduling
: : R scale
Present time
R ： Rescheduling
Assume weather
forecasting is reliable

7. 7
The weather impact to a railway line
Rain full level at different time (Background images are from Met Office, UK)
(a) (b)
(d)(c)
Dynamic

8. 8
The weather impact to a railway line
weather impact model
The speed limitation applied on tracks
and one possible rescheduling plane
if

9. 9
Methodology
Optimization object
Optimize total delay (first delay and second
delay); some first delays will also be reduced
from rescheduling.

10. 10
Methodology
Single track problem and resolution
M （
i ，
j ）
=1
M
（
i ，
j ）
=0
potential conflict
time
distance
Time space trajectory

11. 11
Original timetable & Weather effect
C1,3
C2,1
C2,2C1,1
S3
S2
S1
S4
S5
S6
Tj(1) Tj(2) Tj(3)
Ti(1) Ti(2)
Chromosome
Gen
The bold number is the gen expresed
Solve the earliest conflict
Update conflict set
Satisfy end condition
Output
Yes
No
Conflict set is empty
Yes
No
Input
Conflict set is empty
Yes
No
Generate new M
Methodology
single track rescheduling using genetic algorithm
Genetic algorithm

12. 12
Case study:a single track railway line
Cambrian Line
Input: the original time table is shown as follows
the weather affected area is shown by the shadow

13. 13
Case study:
Cambrian Line,UK
M(4,4)=1
M(5,4)=1
M(4,5)=0
M(5,5)=1
Planned Total delay=152min
Traditional method: Rescheduling after the delay showing up
Accrual trajectory Rescheudling plan
Reschedule decision
Traditional algorithm Time tabled delay,
espected delay

14. 14
Case study:
Cambrian Line
Actual trajectory by traditional method
M(4,4)=1
M(5,4)=1
M(4,5)=0
M(5,5)=1
Actuall Total delay=346min
Reschedule decision
Traditional algorithm

15. 15
Rescheduling plan
Case study:
Cambrian Line
Proposed algorithm
Proposed method :Rescheduling before the adverse weather showing up
Accrual trajectory
M(4,4)=0
M(5,4)=1
M(4,5)=1
M(5,5)=1
Total delay=279min
Reschedule decision
Future trajectory

16. 16
Summary
• Alternative method to model the adverse weather
UISDs VS PCDDs rescheduling
• New rescheduling method to deal with the weather
disturbance
• Better performance
Less delay

17. 17
Future
• Mix of UISDs and PUDDs
• Smaller weather resolution ratio
• Section blockage

18. 18
Contact: tsyw@leeds.ac.uk
Thanks