Analysing Busway Station Capacity under Mixed Stopping and ... Widanapathiranage B… · AIMSUN...

Analysing Busway Station Capacity under Mixed Stopping and Non-

Stopping Operation Rakkitha Widanapathiranage1, Jonathan M Bunker2, and Ashish

Bhaskar3

Paper 15-3949, TRB 94th Annual Meeting

(1) Doctoral Student, (2) Associate Professor, (3) Senior Lecturer Queensland University of Technology, Australia

Bus Rapid Transit Advantages

Dedicated Running-

way

High Station

Inclusions

Premium Bus

Services

Increased Ridership

Speed Reliability Efficiency Identity Customer Satisfaction

Rationale

• Some BRT lines have “non-stopping” buses passing certain stations

• Brisbane’s South East Busway for example

• This study addresses this phenomenon

All buses stop

Capacity through greatest constriction • Usually

busiest stop

Transit Line Service Capacity

Some buses non-stopping

?

BRT Line Capacity Estimation Knowledge Gap

Capacity Definitions

TCQSM Service Capacity • Stipulated repeatable,

safe working conditions • Operating margin avoids

congested operation – Average dwell time – CV of dwell time – Z variate

Potential Capacity • No operating margin • Represents maximum

possible outflow • all other conditions as

for service capacity • Degree of saturation = 1

Study Methodology

compare

verify Base deterministic potential capacity •no operating margin •actual number loading areas

Base simulation capacity •CV dwell time = 0 •train-like throughput

AIMSUN microscopic simulation testbed

•Av dwell time •Av clearance time

•Headway distribution •Dwell time distribution AIMSUN API

Field surveys

Some non-stopping buses

Mixed-Stopping Buses potential

capacity

(TCQSM) potential capacity •no operating margin •effective number loading areas

All-Stopping Buses potential

capacity

Bus-bus interference •CV dwell time ≥ 0 •merging behavior

Buranda Station South East Busway, Brisbane Australia

N

200m (650ft)

Buranda station

Eastern Busway

Eastern Busway

South East Busway

To CBD Cleveland

railway

Buranda Station Simulation Testbed

Inbound platform

Outbound platform

B

A

CBD suburbs

Buranda Station Measured Headway Distributions

0

0.01

0.02

0.03

0.04

0.05

0.06

0 20 40 60 80 100

Prob

abili

ty D

ensit

y

Headway (s)

Exponential(0.055)

Buranda Station Measured Dwell Time Distributions

0

0.01

0.02

0.03

0.04

0.05

0.06

0.07

0 10 20 30 40 50 60 70

Prob

abili

ty D

ensit

y

Average dwell time (s)

Log-normal(2.718,0.612)

AIMSUN Microscopic Simulation Model Development

Feature AIMSUN default AIMSUN modified using API

Arrival distribution Normal Negative exponential Trajectory Car-following = Dwell time distribution Normal Log-normal Merging Gap acceptance =

Driver reaction time 0.75s moving 1.35s stationary =

Simulation time step 0.15s =

Simulation Model Development Scenarios and Experimental Values

Simulation Model

Percentage Non-stopping

Buses

Av Dwell Time (s)

CV of Dwell Time

Base potential capacity (ASB)

0 Incremental 5s to 90s

0

ASB potential capacity

0 Incremental 5s to 90s

0.4, 0.5, 0.6

MSB potential capacity

10, 20, 30, 40 Incremental 5s to 60s

0.4, 0.5, 0.6

Base Deterministic Potential Capacity (Train-like operation)

𝐵𝑝 =3600

𝑡𝑑 + 𝑡𝑐𝑁𝑙𝑙

Where: 𝐵𝑝 = potential capacity (bus/h) 𝑡𝑑 = fixed dwell time (s) 𝑡𝑐 = fixed clearance time (s) e.g. 16s 𝑁𝑙𝑙 = actual number of loading areas e.g. 3

Base Simulation Model and Base Deterministic Potential Capacity

0

100

200

300

400

500

600

0 10 20 30 40 50 60 70 80 90 100

Pote

ntia

l cap

acity

(bus

/h)

Dwell time (s)

deterministic model with zero dwell time coefficentsimulation results with zero dwell time coefficent

(TCQSM) Potential Capacity with No Operating Margin

𝐵𝑝 =3600

𝑡𝑑 + 𝑡𝑐𝑁𝑒𝑙

• Where: 𝐵𝑝 = potential capacity (bus/h) 𝑡𝑑 = average dwell time (s) 𝑡𝑐 = average clearance time (s) e.g. 16s 𝑁𝑒𝑙 = effective number of loading areas e.g. 2.65

Proposed ASB Potential Capacity Model

𝐵𝑙𝑎𝑎|𝑝 =3600

𝑡𝑑 + 𝑡𝑐𝑁𝑙𝑙𝑓𝑎𝑎𝑏

• Where: 𝐵𝑙𝑎𝑎|𝑝 = all-stopping potential capacity (bus/h) 𝑡𝑑 = average dwell time (s) 𝑡𝑐 = average clearance time (s) e.g. 16s 𝑁𝑙𝑙 = actual number of loading areas e.g. 3 𝑓𝑎𝑎𝑏 = station bus-bus interference factor

Bus-bus Interference Factor

• Accounts for loss of capacity: – Varying dwell times causes asynchronous bus

movements, constraining lead LAs’ usage – Shared priority gap acceptance process

• Alternative approach to “effective number of loading areas”

Bus-bus Interference Factor (From Regression on Simulation Data)

00.10.20.30.40.50.60.70.80.9

1

0 10 20 30 40 50 60 70 80 90 100

Bus-

bus i

nter

fere

nce

fact

or

Average dwell time (s)

dwell time coefficent=0.4dwell time coefficent=0.5dwell time coefficent=0.6

Bus-bus Interference Factor (From Regression on Simulation Data)

𝑓𝑎𝑎𝑏 = 0.90 − 0.004 𝑐𝑣𝑡𝑑

• Where: 𝑡𝑑 = average dwell time (s) e.g. 5s to 90s 𝑐𝑣 = coefficient of variation of dwell time e.g. 0.4, 0.5, 0.6

Calibrated ASB Potential Capacity against TCQSM (no operating margin)

050

100150200250300350400450

0 10 20 30 40 50 60 70 80 90 100

ASB

pote

ntia

l cap

acity

(bus

/h)

Average dwell time (s)

TCQSM (CV = 0) CV = 0.4 CV = 0.5 CV = 0.6

Calibrated MSB Potential Capacity Model (CV dwell time = 0.4)

050

100150200250300350400450500

0 10 20 30 40 50 60 70

MSB

pot

entia

l cap

acity

(bus

/h)

Average dwell time (s)

0% (sim)10% (sim)20% (sim)30% (sim)40% (sim)0% (mod)10% (mod)20% (mod)30% (mod)40% (mod)

Proposed MSB Potential Capacity Model (From Regression on Simulation)

𝐵𝑚𝑎𝑎|𝑝 =𝐵𝑙𝑎𝑎|𝑝

1 − 0.48 𝑃𝑛𝑎𝑎

• Where: 𝐵𝑚𝑎𝑎|𝑝 = mixed-stopping potential capacity (bus/h) 𝐵𝑙𝑎𝑎|𝑝 = all-stopping potential capacity (bus/h) 𝑃𝑛𝑎𝑎 = proportion of non-stopping buses

Stopping Buses Potential Capacity from MSB Potential Capacity Model

𝐵𝑎𝑎|𝑝 =𝐵𝑙𝑎𝑎|𝑝 1 − 𝑃𝑛𝑎𝑎 1 − 0.48 𝑃𝑛𝑎𝑎

• Where: 𝐵𝑎𝑎|𝑝 = stopping potential capacity under MSB operation (bus/h) 𝐵𝑙𝑎𝑎|𝑝 = all-stopping potential capacity (bus/h) 𝑃𝑛𝑎𝑎 = proportion of non-stopping buses

Non-stopping Buses Potential Capacity from MSB Potential Capacity Model

𝐵𝑛𝑎𝑎|𝑝 =𝐵𝑙𝑎𝑎|𝑝𝑃𝑛𝑎𝑎

1 − 0.48 𝑃𝑛𝑎𝑎

• Where: 𝐵𝑛𝑎𝑎|𝑝 = non-stopping potential capacity under MSB operation (bus/h) 𝐵𝑙𝑎𝑎|𝑝 = all-stopping potential capacity (bus/h) 𝑃𝑛𝑎𝑎 = proportion of non-stopping buses

Stopping and Non-stopping Capacities under MSB (ASB Capacity 100 bus/h)

89 76

22

51

112 126

020406080

100120140

0 0.1 0.2 0.3 0.4

Pote

ntia

l cap

acity

(bus

/h)

Proportion of buses non-stopping

Stopping Non-stopping MSB

Conclusions for Application

• Microscopic simulation model can replicate deterministic BRT capacity estimation

• Bus-bus interference factor useful alternative to “effective loading areas” – relates dwell time variation

• Models developed can estimate BRT line capacity when non-stopping buses operating – Higher capacity than when all buses stop

Further Research

• More comprehensive validation using observational data

• Definition of “practical capacity” according to degree of saturation – Requires model of queuing / delay upstream of

station

With Thanks

• Queensland Department of Transport and Main Roads, TransLink Division – data and access

• Mr Hao Guo – smartphone data collection app

• Student survey team