prof. haiying li state key laboratory of rail traffic control and … · g n e i i) b b e n 0 2 4 6...

*

Metro Passenger Flow Management in a Megacity:

Challenges and Experiences in Beijing

International Workshop on High-Speed Rail Planning and OperationsWashington D.C., Oct 30, 2015

Prof. Haiying Li

State Key Laboratory of Rail Traffic Control and Safety

Beijing Jiaotong University

Beijing, China

E-mail: [email protected]

*

OutlineChallenge

• Fast development of urban rail network

• Passenger volume increases rapidly

• Isolated operation→ Network operation

• Single operator→ Multi-operator

Experience• Passenger flow forecast

• Calculation and evaluation of station capacity

• Metro operation simulation

• Passenger flow control

*

Metro map of Beijing in 2015

(18 lines,660km)

*

4

Metro planning of Beijing in 2020(30 lines,1177km)

*

In 2020, the network of Beijing subway will consist of 30 lines, and the

operation mileage will be 1177km.

1969 2000 2008 2011 2012 2013 2014 2015

Number of lines

1 2 8 15 16 17 18 20

Number of stations

16 41 123 215 261 276 325 350

Operationmileage

(km)23 54 200 372 442 465 527 660

Daily passenger

volume(million)

— 1.20 3.33 6.01 6.72 8.65 9.56 10.46

Operator Beijing subway Beijing subway, Beijing MTR

*

6

Metro map of Shanghai in 2015(14lines,548km)

*

7

Metro planning of Shanghai in 2020(18 lines,800km)

*

8

Metro map of Guangzhou in 2015(9 lines, 260km)

*

9

Metro planning of Guangzhou in 2020(17 lines,677km)

10

*

Outline

Challenge• Fast development of urban rail network








*

Load factor

>120%

100%~120%

80%~100%

0%~80%

Passenger Status of Beijing Subway during morning peak hours

(8:15-8:30, Wednesday, May 20, 2015)

13

*

Outline









*

20082007

2009

2003Before 2003

2010

Isolated Lines → Network

*

Outline









*

Beijing MTR Corporation

Beijing Subway

16lines

2lines

Line 4, Daxing Line, Line 14

2 Operators since 2009

*

Outline









*

18

• Project

• “Transport organization key technologies and system development

on urban rail transit network”

• Participant Organization

• State Key Laboratory of Rail Traffic Control and Safety(RCS) of


• Beijing Metro Network Control Center(TCC)

Experience in Beijing

*

19

Objective of the Project

• Analysis of passenger demand

• Deployment of transportation capacity

• Coordination among stations, lines and

network

• Corresponding evaluation method and

standard for operators

*

20

Experience in Beijing• During 2011-2014, we established the “Networked-operation

Decision Center”(NDC)

*

Experience in Beijing

• NDC has been deployed in the information system of

Beijing Metro Network Control Center.

• It can analyze the impact of new lines to existing lines, and

evaluate the coordination of capacity and demand.

• It also can be used as a reference to the design of stations

and lines in urban rail transit.

*

Outline









*

Passenger flow forecast of new lines

The passenger volume impact of new lines to transfer stations

The passenger volume forecast of new lines to an existing line

24

*

24

Multi-resolution modeling and simulation:

Station Simulation

Station simulation

of Pinganli Station

Line 6

Line 4

Space occupancy of

Pinganli Station

A multi-agent-based microscopic pedestrian simulation model in the

station is proposed.

*

25


Facility capacity calculationBased on the concept of the stress test, the facility capacity can

be calculated by the station simulation system.

*

Identification and Relief of Bottlenecks

According to the bottleneck transmission characteristics, we

proposed a bottleneck identification method and the sensitivity

analysis-based bottleneck relief strategy.

Bottleneck

Non-Bottleneck

*

1B*

2B

1v

2v3v

4v5v

6v10v

7v9v

8v

11v15a

45a

32a

57a

65a53a

78a

86a108a

89a 911a

1110a

Entra-

nce

A

Stair1 Walkway1 In-gate A

Escalator1 Walkway2 Out-gate A

In-gate D

Stair2

Escalator3

Walkway3

Entra-

nce

D

Entr-

ance

B

Stair4Walkway5

Escalator2

Walkway4Out-gate B

Walkway5

Walkway3

Walkway6

Entr-

ance

C

Stair3

West

hall

Eest

hallIsland

Platform

Up

Down

Stair5

Stair6

Stair7

Stair8

Out-

gate C

In-gate B

*

Evaluation of Station Capacity

Indexes of station efficiency Capacity utilization rate of facilities

*


Network Simulation

•Network topology model building

•Simulation experimental project

management

•Simulation experiment process

control

•Passenger distribution

•Simulation data acquisition and

processing

•Visual display of simulation process

•Calculation and display of evaluation

indexes

29

*

Outline









*

Metro Passenger flow control:

Boarding-limiting method65 constant boarding-limiting stations in Beijing

58 stations at morning peak hours

19 stations at evening peak hours

*

31

Inbound passenger flow control

Control the number of inbound passengers per unit time,

to reduce the load of facilities in the station.

Fencing in the entrances

Adjust the number and the location of the ticket gates

Change the speed of ticket sale

*

Transfer passenger flow control

Control the transfer passenger volume per unit time from

one line to another, to influence the number of boarding

passengers of another line and passenger route choice.

Adjust the width of transfer walkway

Change the transfer streamline

*

33

Multi-station Collaborative Boarding-Limiting

(MCB) •During rush hours, intensive passenger flow usually appears not

only at a single station but also at several adjacent stations on

the urban rail transit network.

•The organization scheme of an upstream station will influence the

passenger boarding at several downstream stations

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Caofang-Changying

Changying-Huangqu

Huangqu-Dalianpo

Dalianpo-Qingnianlu

Qingnianlu-Shilibao

Shilibao-Jintailu

Jintailu-Hujialou

Hujialou-Dongdaqiao

Secti

on

lo

ad

fa

cto

r

Section

load factor>1.0

load factor<1.0

0

1000

2000

3000

4000

5000

6000

7000

8000

Caofang Changying Huangqu dalianpo Qingnianlu Shilibao Jintailu Hujialou

Inb

ou

nd

Pa

ssen

ger V

olu

me（

pa

ssen

ger p

er h

ou

r）

Station

Inbound passenger volume>5000

Inbound passenger volume<5000

Section load factor of line 6

during morning peak hours

Inbound passenger volume of line 6

during morning peak hours

*

Station A Station B Station C Station D


Station E

Line 1

Line 2 Station A Station B Station C Station D

Station E

Line 1

Line 2

Boarding-LimitTransfer

The influence of the MCB method

on the station congestion

The same line

Before MCB

Before MCB

After MCB

After MCB

The adjacent line


*

The influence of the MCB method

on the route choice

Before MCB After MCB

Station A Station B

Station C Station D

Line 2Line 1

Line3

Line4

Route2

Route1

Station A Station B

Station C

Station D

Line3

Line4

Line 2Line 1 TransferBoarding-Limit

*

MCB Model

Assumption

(1) The OD demand of the network is fixed.

(2) The inbound passengers enter the station at a constant speed.

(3) The headway of each line, running time in the each section and

dwelling time at each station are fixed.

Decision variable

Number of inbound passengers taking line r at station n per unit time

Number of transfer passengers taking line r at station n per unit time

Control the number of passengers boarding at each station

to allocate the train capacity reasonably.

37

*

MCB Model: Object function

(1) Maximize the number of boarding passengers

It can be calculated by inbound and transfer passenger volume

per unit time.

(2) Reduce the congestion in the key stations.

The key stations means the stations that suffer huge passenger

volume. More passengers should be allowed to board the trains

than other stations.

This objective can be expressed by the variable coefficient of the

ratio of boarding passenger volume to passenger demand.

38

*

38

MCB Model: Constraints

(1) Passenger demand constraints

(2) Dwell time constraints

(3) Train capacity constraints

The boarding passenger volume should be less than

passenger demand per unit time.

The passengers on the platform should be able to board

during dwell time.

The number of passengers on the platform should be less

than the residual train capacity.

(4) Transfer walkway capacity constraint

The number of boarding-limiting transfer passengers

should be less than the transfer walkway capacity.

*

MSA Algorithm

Object function(1)Maximum boarding passengers

(2)Reduce the congestion in the key

stations

The MCB Scheme

GA algorithm

Update route choice

Multi-objective

Optimization

Decision Variable(1) Inbound passenger volume per unit time(2) Transfer passenger volume per unit time

Passenger assignment

with multi-path and

varying impedance

MCB Model: Algorithm framework

*

Xi’erqi

Shangdi

Wudaokou

Zhichunlu Xitucheng JiandemenMudanyuan Beitucheng

Olympic Green

Olympic

Sports Center

Lincuiqiao

Yongtaizhuang

Xixiaokou

Huoying

Yuxin

Huilongguandongdajie

PingxifuYuzhiluZhuxinzhuang

Gonghuacheng

Life Science Park

Line 08

Line 10

Line 13

Line 15

Changping

Line

HuilongguanLongze

Qinghuadongluxikou

Liudaokou

Beishatan

Shahe

Non-Boarding-limiting

Station

Boarding-limiting

Station

Shahe University Park

Nanshao

South Gate of

Forest Park

Adjacent boarding-limiting station

Case Study

*

(b) Changping Line

(a)Line 13

The numbers of boarding passengers before and after MCB

(Line 13 and Changping Line)

0

50

100

150

200

250

300

350

400

Huoying Huilongguan longze Xierqi Shangdi

Pa

ssen

ger

Vo

lum

e (p

ass

eng

er p

er h

ou

r) Before MCB

After MCB

Arriving Rate

Station

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Huoying-

Huilongguan

Huilongguan-

Lingze

Longze-

Xierqi

Xierqi-

Shangdi

Shangdi-

Wudaokou

Sec

tio

n L

oa

d F

act

or

Before MCB

After MCB

Section

0

20

40

60

80

100

120

Nanshao Shahe

UniversityPark

Shahe Gonghuacheng Zhuxinzhuang Life

SciencePark

Pa

ssen

ger

Vo

lum

e (p

ass

eng

er p

er h

ou

r) Before MCB

After MCB

Arriving Rate

Station

0

0.2

0.4

0.6

0.8

1

1.2

1.4

Nanshao-

Shahe

University

Park

Shahe

University

Park-

Shahe

Shahe-

Gonghuacheng

Gonghuacheng-

Zhuxinzhuang

Zhuxinzhuang-

Life

Science

Park

Life

Science

Park-

Xierqi

Sec

tio

n L

oa

d F

act

or

Before MCB

After MCB

Section

*The loading factor before and after MCB

(Line 13 and Changping Line)

(b) Line 10(a)Line 13

0

0.2

0.4

0.6

0.8

1

1.2

Beitucheng-Jiandemen Jiandemen-Mudanyuan Mudanyuan-Xitucheng Xitucheng-Zhichunlu

Secti

on

Lo

ad

fa

cto

r

Before MCB

After MCB

Section

0.00

0.20

0.40

0.60

0.80

1.00

1.20

Huilongguandongdajie-Huoying

Huoying-Yuxin

Yuxin-Xixiaokou

Xixiaokou-Yongtaizhuang

Yongtaizhuang-Lincuiqiao

Lincuiqiao-South Gate of Forest Park

South Gate of Forest Park-

Olympic Green

Olympic Green-Olympic Sports

Center

Olympic SportsCenter-

Beitucheng

Sectio

n L

oa

d f

acto

r

Before MCB

After MCB

Section

Xi’erqi

Shangdi

Wudaokou

Zhichunlu Xitucheng JiandemenMudanyuan Beitucheng

Olympic Green

Olympic

Sports Center

Lincuiqiao

Yongtaizhuang

Xixiaokou

Huoying

Yuxin

Huilongguandongdajie

PingxifuYuzhiluZhuxinzhuang

Gonghuacheng

Life Science Park

Line 08

Line 10

Line 13

Line 15

Changping

Line

HuilongguanLongze

Qinghuadongluxikou

Liudaokou

Beishatan

Shahe

Non-Boarding-limiting

Station

Boarding-limiting

Station

Shahe University Park

Nanshao

South Gate of

Forest Park

•The section load factors of Line 8

and Line 10 rise slightly. This

passenger flow is transferred from

heavily jammed Line 13 to less

crowded Line 8 and Line 10.

•Taking line 13 and Changping Line

into consideration, the total number

of boarding passengers increases by

600 passengers per hour after MCB.

*

Thank you for your attention!

Prof. Haiying Li

State Key Laboratory of Rail Traffic Control and Safety


Beijing, China

E-mail: [email protected]

International Workshop on High-Speed Rail Planning and OperationsWashington D.C., Oct 30, 2015

prof. haiying li state key laboratory of rail traffic control and … · g n e i i) b b e n 0 2 4 6...

Documents