Analysis and modeling of pedestrian flows in railway stations

Size: px

Start display at page:

Download "Analysis and modeling of pedestrian flows in railway stations"

Shanon Houston
5 years ago
Views:

1 SBB-Beirat Technologie, Methoden und Prozesse Analysis and modeling of pedestrian flows in railway stations Flurin Hänseler, Transport and Mobility Lab, EPFL December 4, /24

2 Pedestrian flows in train stations

3 The PedFlux Project Collaborative EPFL/CFF research project: Development of a comprehensive modeling framework for pedestrian demand estimation in railway stations. 1) extensive data analysis of exemplary train station Gare de Lausanne 2a) development of demand estimation methodology dynamic origin-destination demand 2b) development of traffic assignment model accessory to demand estimation level-of-service assessment 3) application of combined framework to case study 3/24

4 Pedestrian underpasses of Gare de Lausanne #9 Coop PU West South entrance Place de la gare Metro #9 #7/8 #5/6 #3/4 #1/Main hall #7/8 & South entrance Place de la Gare Metro PU East 4/24

5 Coverage of tracking sensors Monitored area in PIO (above) and PIE (below): Figure adapted from VisioSafe 5/24

6 Tracking algorithm Sensor topology: (0) Figure adapted from VisioSafe 6/24

7 Tracking algorithm (1) Detection Figure adapted from VisioSafe 6/24

8 Tracking algorithm (1) Detection (2) Tracklet generation Figure adapted from VisioSafe 6/24

9 Tracking algorithm (1) Detection (2) Tracklet generation (3) Association Figure adapted from VisioSafe 6/24

10 Sample trajectory tracked vs. interpolated periods microscopic vs. macroscopic fidelity 7/24

11 Sample trajectory corresponding (v,t)-map 15 infinitesimal velocity [km/h] tracking point number 7/24

12 Pedestrian movements on January 16, 2013 Animation: Animation: Nicolas Anken, EPFL 8/24

13 750 Visualization of pedestrian demand South #9 #7/ Coop North Metro pedestrian underpasses, Gare de Lausanne busiest 15-min period extracted from tracking data #5/ #3/4 Figure: Nicholas Molyneaux, EPFL 9/24

14 North Metro #3/ #5/ #7/ #9 0 Coop 0 South Gare de Lausanne 07:37 7:52 10-day average

15 Periodic flow patterns Daily pattern (January 16, 2013) pedestrian demand (#/h) :00 12:00 18:00 24:00 10 / 24

16 Heat map of PUs, January 22, 2013 LOS Pedestrian density A < [ped/m 2 ] B < C < D < E < F Table: Pedestrian walkway LoS density threshold values according to NCHRP density as indicator for: comfort performance safety Ref: [Hig00], Exhibit / 24

17 Heat map of PUs, January 22, 2013 Figure: SBB-I-AT-BZU-PFL 11 / 24

18 Heat map of PUs, January 22, 2013 Figure: SBB-I-AT-BZU-PFL 11 / 24

19 Heat map of PUs, January 22, 2013 Figure: SBB-I-AT-BZU-PFL 11 / 24

20 Heat map of PUs, January 22, 2013 Figure: SBB-I-AT-BZU-PFL 11 / 24

21 Heat map of PUs, January 22, 2013 Figure: SBB-I-AT-BZU-PFL 11 / 24

22 Heat map of PUs, January 22, 2013 aggregation: t = 60 s, A = 7.29 m 2 7:40 7:41: Low occupation, no train arrivals/ 11 / 24

23 Heat map of PUs, January 22, 2013 aggregation: t = 60 s, A = 7.29 m 2 7:41 7:42: Arrival of train IR 1606 at 7:40:20 on platform 3/4 11 / 24

24 Heat map of PUs, January 22, 2013 aggregation: t = 60 s, A = 7.29 m 2 7:42 7:43: Arrival of train IR 706 at 7:41:24 on platform 5/6 11 / 24

25 Heat map of PUs, January 22, 2013 aggregation: t = 60 s, A = 7.29 m 2 7:43 7:44: Arrival of train IR 1407 at 7:42:20 on platform 3/4 11 / 24

26 Heat map of PUs, January 22, 2013 aggregation: t = 60 s, A = 7.29 m 2 7:44 7:45: Gradual decrease in pedestrian occupation/ 11 / 24

27 Voronoi-based spatial tessellation finite set of points p 1, p 2,... in space Voronoi cell of point p i defined as V (p i )={p p p i p p j, i j} each point represents a pedestrian Slide: Marija Nikolic, EPFL 12 / 24

28 Empirical fundamental diagram Slide: Marija Nikolic, EPFL 13 / 24

29 Framework for pedestrian flow estimation Train timetable, travel surveys Spatio-temporal observations Layout of walking facilities Demand estimation demand supply Traffic assignment Dynamic trip table Travel time, occupation 14 / 24

30 Pedestrian demand estimation Train timetable, travel surveys Spatio-temporal observations Layout of walking facilities Demand estimation demand supply Traffic assignment Dynamic trip table Travel time, occupation 15 / 24

31 Pedestrian demand estimation: Train timetable flows into pedestrian underpasses sample pedestrian trajectories PU West PU East 16 / 24

32 Pedestrian demand estimation: Train timetable correlation between train schedule and pedestrian flows Total # passengers 3,000 2,000 1,000 Tracks 5/6 Tracks 3/4 0 7:00 7:30 8:00 8:30 Figure: Train unloading flow and train arrivals, April 9, 2013 Ref: [MHB13] 16 / 24

33 Pedestrian demand estimation: Train timetable correlation between train schedule and pedestrian flows unloading flow as superposition of train-induced events Cumulated flow [#] s α Time Q inflow after train arrival dead time: s 46.3 s flow rate: α long = 6.8 ± 1#/s α short = 4.5 ± 1#/s disembarkations per train: Q = Ref: [MHB13] 16 / 24

34 Pedestrian demand estimation: Train timetable correlation between train schedule and pedestrian flows unloading flow as superposition of train-induced events sample prediction (April 9, 2013, based on HOP data) Total # passengers 1,500 1, Observation 150 Estimation :00 8: :30 Residuals [#] Ref: [MHB13] 16 / 24

35 Pedestrian demand estimation: Methodology PU West PU East 7:07 7:02 7:05 7:03 7:01 flow counter sales point tracking system boardings/disembarkations 17 / 24

36 Pedestrian traffic assignment Train timetable, travel surveys Spatio-temporal observations Layout of walking facilities Demand estimation demand supply Traffic assignment Dynamic trip table Travel time, occupation 18 / 24

37 Pedestrian traffic assignment: Desired properties accurate prediction of travel times given demand calibration with trajectory data customizable I/O interface coupling with demand estimation framework high computational performance several times faster than real-time mesoscopic pedestrian flow model 19 / 24

38 Pedestrian traffic assignment: Space representation walkable area entry/exit points route sequence of areas path sequence of cells 20 / 24

39 Pedestrian traffic assignment: Propagation model pedestrian fundamental diagram [Wei93] v (m/s) v f = q (#/ms) q opt = k opt = 1.75 k jam = 5.4 k (#/m 2 ) 21 / 24

40 Pedestrian traffic assignment: PU West, Lausanne Figure: Pedestrian Underpass West, Lausanne railway station c CFF, Sandro Campardo 22 / 24

Pedestrian traffic assignment: PU West, Lausanne pedestrian demand extracted from tracking data prediction of travel times, flows and densities January 22, 2013, 07:40 07:46 LOS [#/m 2 ]

41 Pedestrian traffic assignment: PU West, Lausanne pedestrian demand extracted from tracking data prediction of travel times, flows and densities January 22, 2013, 07:40 07:46 LOS [#/m 2 ] A < B < C < D < E < F #9W #7/8W #5/6W #3/4W #1W South exit North exit #9E #7/8E #5/6E #3/4E #1E/Main hall Animation: 22 / 24

42 Concluding remarks and next steps 1. extensive data analysis for Gare de Lausanne 2. framework for pedestrian flow modeling 2a) demand estimation methodology (primary aim) 2b) traffic assignment model (accessory) 3. application of combined framework to case study prototype tool for integrated demand/supply estimation operationalization of research findings with third party tbd apply knowledge/methodology to further train stations develop decision-aid tools for practitioners Ref: [HMTB13] 23 / 24

43 Thank you SBB-Beirat Technologie, Methoden und Prozesse: Analysis and modeling of pedestrian flows in railway stations Flurin Hänseler, Transport and Mobility Lab, EPFL Many results shown in this presentation are due to Nicolas Anken, Nicholas Molyneaux and Thomas Mühlematter. Support by SBB-I-AT-BZU-PFL, EPFL-TraCE and VisioSafe is gratefully acknowledged. Picture on slide 2 c Michael Buholzer, Reuters. flurin.haenseler@epfl.ch 24 / 24

44 Bibliography I Highway Capacity Manual. Transportation Research Board. Washington, DC, F.S. Hänseler, Nicholas Molyneaux, M. Thémans, and M. Bierlaire. Pedestrian strategies within railway stations: Analysis and modeling of pedestrian flows (PedFlux Mid-Term Report). Technical report, EPFL, Nicholas Molyneaux, F.S. Hänseler, and M. Bierlaire. PedFlux Analysis Report: Train-induced loading and unloading flows in platform access ways. Technical report, EPFL, / 24

45 Bibliography II U. Weidmann. Transporttechnik der Fussgänger. Institute for Transport Planning and Systems, ETH Zürich, / 24

Pedestrian traffic flow operations on a platform: observations and comparison with simulation tool SimPed

Pedestrian traffic flow operations on a platform: observations and comparison with simulation tool SimPed W. Daamen & S. P. Hoogendoorn Department Transport & Planning, Delft University of Technology,