Download - Jens Parbo PhD student Technical University of …railtokyo2015.cs.it-chiba.ac.jp/pres/084p.pdfAdapting stopping patterns to improve robustness from the users’ perspective Jens Parbo

Transcript

Adapting stopping patterns to improve robustness from the users’ perspective

Jens Parbo

PhD student

Technical University of Denmark

[email protected]

2 DTU Transport, Danmarks Tekniske Universitet

Outline

• Introduction & motivation

• Skip-stop optimisation

• Results

• Summary

3 DTU Transport, Danmarks Tekniske Universitet

Outline

• Introduction & motivation

• Skip-stop optimisation

• Results

• Summary

4 DTU Transport, Danmarks Tekniske Universitet

Introduction & Motivation

• Stopping patterns of railway lines have a large impact on travel time, but also on system-wide delays

• Applying skip-stop optimisation intelligently enables:

– The benefits of skip-stop services (reduced travel time)

– Reduced heterogeneity (i.e. operation less vulnerable to delays)

• Homogeneous (solid lines) vs. heterogeneous (dashed lines) operation

0

10

20

30

40

50

60

70

1 2 3 4 5 6 7 8

Tim

e

Stop

5 DTU Transport, Danmarks Tekniske Universitet

Outline

• Introduction & Motivation

• Skip-stop Optimisation

• Results

• Summary

6 DTU Transport, Danmarks Tekniske Universitet

Skip-stop Optimisation

• Bi-level optimisation

Passengers’ travel behaviour

Railway lines’ stopping patterns

7 DTU Transport, Danmarks Tekniske Universitet

Mathematical Model (Upper level)

• Finding optimal stopping patterns.

8 DTU Transport, Danmarks Tekniske Universitet

Public assignment model (Lower level)

• Deriving passengers’ travel behaviour.

• Utility-based approach.

9 DTU Transport, Danmarks Tekniske Universitet

Heuristic solution algorithm - Framework

1. Run public assignment.

2. Calculate optimisation potential for each skip-stop.

3. Calculate elaborate optimisation potential for the most promising skip-stops (2)

4. Impose skip-stops.

5. Prohibit skipping stops in the sub network for imposed skip-stops

6. If stopping criterion met, stop.

7. Otherwise, go to (1).

Inner lo

op

10 DTU Transport, Danmarks Tekniske Universitet

• When skipping a stop on a line, trips to/from the considered station within the corridor are separated in 3 types.

• Benefit: reduced in-vehicle time (2 min.)

• Trip 1 costs (Waiting time)

• Trip 2 costs (Waiting time + Transfer time)

• Type 3 trips are PROHIBITED!

• Trips from outside the corridor (Waiting time)

Heuristic solution algorithm – Elaborate optimisation potential

11 DTU Transport, Danmarks Tekniske Universitet

Heuristic solution algorithm - Tackling overtaking and heterogeneity

• Limitation on the number of skipped stops between parallel lines when considering the stops sequentially.

• Conflicts between railway lines from different corridors can be handled by adapting depature times or adding buffer time.

• Reduced capacity utilisation as a result of more homogeneous operations.

Line A

Line A

Line B

Line B

12 DTU Transport, Danmarks Tekniske Universitet

Outline

• Introduction & Motivation

• Skip-stop Optimisation

• Results

• Summary

13 DTU Transport, Danmarks Tekniske Universitet

• Passengers are marginally better of compaired to the existing situation.

• The small reduction in Generalised travel cost is misleading

Results - Passengers’ travel cost

Percentage change (Final solution)

Changes In-Vehicle Time Generalised Cost Waiting Time

Relative to existing stoppping pattern -0,18 1,01 0,09 1,66

Relative To All-stop Base -1,51 5,46 1,76 -1,67

-6

-4

-2

0

2

4

6

0 1 2 3 4 5 6

Per

cen

tag

e ch

an

ge

Iteration #

Transfers

In-Vehicle Time

Generalised Cost

Waiting Time

14 DTU Transport, Danmarks Tekniske Universitet

• Existing vs. optimised

– Time-space diagrams

– Line diagrams

• Optimised -> homogeneous

– More buffer time

– Delays are absorbed

Results - Heterogeneity of railway operations

Optimised Existing

Optimised Existing

15 DTU Transport, Danmarks Tekniske Universitet

• Number of skipped stops per line in each corridor.

– Average

– Standard deviation

• SSHR (heterogeneity) values for each corridor.

• All-stop scenario serves as LB on SSHR.

Results - Heterogeneity (Continued)

16 DTU Transport, Danmarks Tekniske Universitet

Outline

• Introduction & Motivation

• Skip-stop Optimisation

• Results

• Summary

17 DTU Transport, Danmarks Tekniske Universitet

Summary

• Skip-stop optimisation solved as a bi-level optimisation problem taking passengers’ travel behaviour explicitly into account.

• Reduction equal to 1.01 % in in-vehicle time and 1.66 % in waiting time was obtained compared to existing network.

• Heterogeneity significantly improved in all corridors (5-55 %).

• Consequently, passengers get faster and more reliable from A to B.

18 DTU Transport, Danmarks Tekniske Universitet

Questions