Session 74 Fredrik Johansson

Mikrosimulering av fotgängare - effekter av attpersoner stannar upp eller står och väntar

Fredrik Johansson12 Anders Peterson1 Andreas Tapani12

1Linköping Universitet

2VTI

January 10, 2013

BackgroundWaiting Pedestrians

Results and Conclusions

Outline

1 BackgroundProjectMotivationMethod

2 Waiting PedestriansMotivation and GoalModels

3 Results and ConclusionsSimulation resultsConclusions

Johansson et. al. Transportforum 2013 2/ 22



ProjectMotivationMethod

Outline








Project: “Simulation of interchange stations”

Goal:Evaluate a proposed design of a multi modal public transportinterchange station using microscopic simulation.

Initiators: Peterson and Tapani (LiU and VTI).Financier: Trafikverket.Beneficiaries: Linköping municipality and Östgötatrafiken.Performed by: LiU and VTI.





Motivation

Why study interchange stations?Stations are important for system performance.An increasing number of people travel by public transport.For efficient transfers small stations are needed.

The ProblemSmall station + lots of people⇒ congestion.

Congestion causesDelayDiscomfort





Motivation








Motivation








Method: Microscopic Simulation

What?Modeling of the individual microscopic entities.Macroscopic flow structures are not explicitly modeled, butemerges from the interaction.

Why?Congested pedestrian traffic is highly dynamic.The pedestrian traffic volumes in a station varies muchboth in space and time.Walkable areas can have almost arbitrary shape.The pedestrian population is diverse.





Method: Microscopic Simulation

What?Modeling of the individual microscopic entities.Macroscopic flow structures are not explicitly modeled, butemerges from the interaction.

Why?Congested pedestrian traffic is highly dynamic.The pedestrian traffic volumes in a station varies muchboth in space and time.Walkable areas can have almost arbitrary shape.The pedestrian population is diverse.





General model structure

Behavior Model

Strategical Activityplanning

Notmodeled

Tactical Route choiceShortest

path

Operational Evasivemaneuvers

Social forcemodel






Behavior Model


Notmodeled


path


Social forcemodel

O-D






Behavior Model


Notmodeled


path


Social forcemodel

vp(x)

O-D




Motivation and GoalModels

Outline








Modeling waiting pedestrians

Why?At interchange stations a significant fraction of thepopulation are waiting.The location of waiting areas can to some extent becontrolled.

GoalDevelop different extensions to the model to includewaiting pedestrians.Characterize and compare the predictions of the differentextensions.





Modeling waiting pedestrians

Why?At interchange stations a significant fraction of thepopulation are waiting.The location of waiting areas can to some extent becontrolled.

GoalDevelop different extensions to the model to includewaiting pedestrians.Characterize and compare the predictions of the differentextensions.





A naive waiting model

Model 0: Stop and stay

vi = 0.

Problem: Only a few waiting pedestrians may cause almostcomplete stop.





A naive waiting model

Model 0: Stop and stay

vi = 0.

Problem: Only a few waiting pedestrians may cause almostcomplete stop.





StructureAt what level should waiting be modeled?

Behavior Model

Activityplanning

Notmodeled

Route choiceShortest

path

Evasivemaneuvers

Social forcemodel

vp(x)

O-D

Waitingmodel

Waiting area

Placement inwaiting area

Interactionswhile waiting





Three waiting models

Model A: Stop

vpi = 0.

Model B: Choose a spot

vpi = (xp

i −xi)/4τ, |xpi −xi |< 4τvp0

i .

Model C: Choose a spot, adjust it

vpi = (xp


i .

Maxpi=−Fp

i −Ffriction






Model A: Stop

vpi = 0.


vpi = (xp


i .


vpi = (xp


i .

Maxpi=−Fp

i −Ffriction






Model A: Stop

vpi = 0.


vpi = (xp


i .


vpi = (xp


i .

Maxpi=−Fp

i −Ffriction




Simulation resultsConclusions

Outline








Total delay distribution

−2 0 2 4 6 8 100

200

400

600

800

Model BModel AModel C

Mean total positive delays:Model A:1.4, Model B: 2.2, Model C:1.8





Density, model A





Density, model B





Density, model C





Delay rate density, model A





Delay rate density, model B





Delay rate density, model C





Adjust SFM for waiters

Behavior Model

Activityplanning

Notmodeled

Route choiceShortest

path

Evasivemaneuvers

Social forcemodel

vp(x)

O-D

Waitingmodel

Waiting area

Placement inwaiting area

Interactionswhile waiting





Conclusions

The models produce reasonable behavior.Probably necessary to interfere with the SFM.Significant differences in the traffic resulting from thedifferent models.

OutlookDataCalibration





Conclusions

The models produce reasonable behavior.Probably necessary to interfere with the SFM.Significant differences in the traffic resulting from thedifferent models.

OutlookDataCalibration


Session 74 Fredrik Johansson

Documents

pedestrians results

background motivation

goal models3 results

lots of people congestion

pedestrian population

problemsmall station

pedestriansmodels results

increasing number of