Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT)...
Transcript of Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT)...
![Page 1: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/1.jpg)
1
Passenger Travel Time Reliability for
Multi-Modal Public Transport Journeys 98th Annual Meeting of the Transportation Research Board 2019
Malvika Dixit
Ties Brands
Niels van Oort
Oded Cats
Serge Hoogendoorn
![Page 2: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/2.jpg)
2
Introduction
• Reliability: ‘Certainty of service aspects compared to the
schedule as perceived by the user’ (van Oort, N., 2016)
• Reliability of travel time – Regularity
– Punctuality
![Page 3: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/3.jpg)
3
Motivation
• Urban transit networks typically multi-modal
• Reliability based on the whole journey experience
including the transfers
Mode 1
In-vehicle
time
Waiting
time
Mode 2
In-vehicle
time
Transfer
time Waiting
time
Two leg journey with two transit modes
![Page 4: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/4.jpg)
4
Motivation
• Urban transit networks typically multi-modal
• Reliability based on the whole journey experience
including the transfers
• Existing indicators – Focus on one mode only, or
– Fail to include all travel time components
Requires passenger travel data at the network-level
![Page 5: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/5.jpg)
5
Objective
• To develop a metric that
– measures reliability for multi-modal transit journeys;
– is sensitive to all travel time components; and
– enables comparison between different transit modes and routes.
• Demonstrate its application to a real life network
(Amsterdam)
![Page 6: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/6.jpg)
6
• Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010)
Difference between the 95th and 50th percentile travel time experienced by travelers between a stop-stop pair using a specific route*
*Route : A combination of public transport services a passenger may choose, where each route may or may not include a transfer.
Methodology - RBT
𝑅𝐵𝑇𝑜,𝑑,𝑟 = 𝑡𝑡95𝑜,𝑑,𝑟 − 𝑡𝑡50
𝑜,𝑑,𝑟
Interpreted as the additional time passengers have to budget for their travel
to ensure on-time arrival one out of twenty times
![Page 7: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/7.jpg)
7
Data sources
• Smartcard data
– Tap-in and tap-out location and times
• Automatic Vehicle Location (AVL) data
– Vehicle number, stop location and time stamps
![Page 8: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/8.jpg)
8
Data pre-processing
Smartcard data
AVL data
Data Cleaning + Fusion
Transfer Inference
Journeys Database
Gordon et al. (2013),
Yap et. al (2017)
![Page 9: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/9.jpg)
9
Travel time using smartcard data
• Where first tap-in at station (eg. Amsterdam Metro) – Total travel time (t5-t0)
• Where first tap-in inside vehicle (eg. Amsterdam buses & trams)
– Total travel time minus waiting time at origin (t5-t1)
Mode 1
In-vehicle time
Waiting
time
Mode 2
In-vehicle time Transfer
time
t0 t1 t2 t3 t4
Waiting
time
t5
![Page 10: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/10.jpg)
10
Waiting time at origin
• For journeys where first tap-in is inside the vehicle – Time passenger arrived at stop is not known
– Headway of services known (from AVL data)
– For short headway services – passengers assumed to arrive randomly
– Continuous random variables generated and sampled over uniform
distribution [0, observed headway] to estimate waiting time for each
journey
![Page 11: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/11.jpg)
11
• RBT calculated for each stop-stop (OD) pair and route
• Weighted average calculated for each mode/line/stop
Travel time reliability for multi-modal journeys
𝑅𝐵𝑇𝑜,𝑑,𝑟 = 𝑡𝑡95𝑜,𝑑,𝑟 − 𝑡𝑡50
𝑜,𝑑,𝑟
![Page 12: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/12.jpg)
12
Case study : Amsterdam
• ~850,000 inhabitants
• 4 metro lines
• 15 tram lines
• 25 bus lines
• ~800,000 transactions/day
• Two weekdays (1st and 2nd
March 2018) used for
analysis
LEGEND
Scale (approx.)
2.5 km
Metro
Tram
Bus
![Page 13: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/13.jpg)
13
Results
![Page 14: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/14.jpg)
14
Reliability per mode
Mode(s) used Number of
journeys Median travel time
(mins) RBT (mins)
Unimodal Journeys
Metro (incl. Metro-Metro)
235,287 14.7 5.9
Tram 315,410 15.4 6.6
Bus 104,495 14.8 6.2
Tram-Tram 1,755 23.2 7.2
Multimodal Journeys
Metro-Tram 7,588 25.0 7.6
Metro-Bus 747 28.8 7.8
Tram-Metro 6,665 26.3 8.3
Bus-Metro 1,336 28.7 8.5
![Page 15: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/15.jpg)
15
Reliability of accessing transit hubs
![Page 16: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/16.jpg)
16
Reliability by route used - Station Sloterdijk to Boelelaan
![Page 17: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/17.jpg)
17
Reliability by route used - Station Sloterdijk to Boelelaan
![Page 18: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/18.jpg)
18
Conclusion
• New metric proposed for travel time reliability measurement
– considering multimodal transit journeys
– including waiting and transfer times for all legs of the journey
– consistent for all journeys - comparable across modes and routes
• Demonstrated application to Amsterdam data but can be applied to
other networks
• Can provide reliability at a very disaggregate level
– flexibility of aggregation (eg. mode, transit stop and route level)
– can be used as an input to behavioral models
![Page 19: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/19.jpg)
19
Limitations and future work
• Assumed that passengers boarded the first vehicle (no denied
boarding)
• Did not consider the impacts of availability of real-time information
• Low sample size
– to be applied to a larger dataset
• Part of the ‘Impact of North-South Metro Line’ project
– Reliability comparison of previous versus current network design
![Page 20: Passenger Travel Time Reliability for Multi-Modal Public ... · 6 • Reliability buffer time (RBT) (Chan, 2007; Uniman et al, 2010) Difference between the 95th thand 50 percentile](https://reader033.fdocuments.in/reader033/viewer/2022042104/5e82ad050d64322f3d12986d/html5/thumbnails/20.jpg)
20
Thank you!
Questions?
Contact details:
Malvika Dixit
http://smartptlab.tudelft.nl/