Queue evolutionsQueue evolution is one of the most important

factors in design of intersection signals. The evaluation compares the model-estimated and the field-observed queue evolutions. Lane 1 is scooter-prohibited, whereas lane 3 serves both scooters and cars.

With an approach of 190M long, it is observed that scooters arrive at the stop-line before cars with a leading time of 8 seconds, which reflects the real-world scooter-vehicle mixed traffic scenario.

Lane ChoiceThe field observation reflects the following

patterns:i. Cars tend to stay on the same travel lane, but

will perform lane changes to merge into the lane that serves their turning needs.

ii. Scooters, with high mobility, tend to choose the lane with fewer vehicles to advance to the downstream stop line.

Travel Time EstimationCars will decelerate and join the end-of-queue;

scooters may decide to filter through the stop queues and not reduce their speed to zero.

Field DataThe scooter-mixed traffic data was collected from the

northbound approaches of an 8-lane major arterial, Xinsheng South Road, in Taipei, Taiwan.

Traffic volume during data collection is about 9,000 vehicles per hour in the target direction; among all vehicle flows, 83% are scooters, 15% are cars, and 2% are buses.

Scooter Discharge RatesThe discharge rates for cars and scooters under non-

mixed traffic condition are set to be 1,800 and 14,000 vphpl, respectively.

An Empirical Study of the Scooter-Vehicle Mixed Traffic Propagation on Urban Arterialsby Chien-Lun Lan and Gang-Len Chang

University of Maryland, College Park

AbstractAbstract

Scooters are one of the primary transportation modes in many developing countries, but design guidelines and software for arterial signals accommodating heavy scooter-vehicle mixed flows are still in their infancy.

Traffic professionals often have no choice but to apply existing models that cannot address scooters’ complex maneuvers.

This study conducts field observations of the mixed traffic flow from their discharging to the formation of stop queues.

Based on the statistical analysis results, this study develops a series of formulations to describe the behavior of mixed traffic flows.

The developed models are evaluated and confirmed to be reliable to serve as the basis for designing arterial control plans.

Research BackgroundResearch Background

Field Observations To study scooter-vehicle mixed traffic, this

research selects two consecutive arterial approaches for field observation.

By using camcorders mounted on a high-rise building, the vehicle trajectories are retrieved on a frame-by-frame basis.

Research Objectives This research tries to understand the

fundamental properties of scooter-vehicle mixed traffic flows, focusing on empirical investigation of their macroscopic behavior.

This study further formulates all statistically validated empirical relations with a series of traffic models.

ConclusionsConclusions

This paper presents the results of empirical investigation with respect to the evolution of scooter-mixed traffic flows from discharging, propagation, to the formation of queues at the downstream intersection.

A series of concise equations are proposed to describe the complex propagation process over an arterial link.

The numerical evaluations of the proposed models with field observations have offered insights into the properties of mixed traffic flows, and can serve as the basis for development of an arterial signal progression model.

The Proposed ModelThe Proposed Model

Discharge ProcessIt is noticeable that scooters often form several

queue lines in a lane while cars can only form a single queue. The field survey reveals that the discharge rate for such mixed flows may vary with the number of scooters in the stop queue.

Acceleration/DecelerationThe average acceleration/deceleration

rates between cars and scooters from the field survey are different.

Comparing to cars• Scooters have higher initial acceleration

rate• Scooters brake later but harder

Merging into Stop QueueAs vehicles approach the stop line and observe

significant differences in queue length among the neighboring lanes, they may perform lane-changing to merge into the lane with shorter queue.

Scooter Filtering ProcessScooters, due to their small physical sizes, tend to

advance between vehicle queue-lines to reach available downstream spaces. It is observed from the field data that scooters’ filtering speed decreases as the mixed flow density increases.

EvaluationsEvaluations

0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%0

500

1000

1500

2000

2500

020004000600080001000012000140001600018000

Discharge Rates underDifferent Scooter-Mixed Ratios

Car ScooterSpace Occupied by Scooters in the Stop Queue (%)

Car D

isch

arge

Rat

e (c

ar/s

)

Scoo

ter D

isch

arge

Rat

e (s

coot

er/s

)

0

0.5

1

1.5

2

2.5

3

3.5

0 10 20 30 40 50 60 70 80

Acce

lera

tion

rate

(m/s

2 )

Distance from stop line

Acceleration Rate

Car

Scooter

0

0.5

1

1.5

2

2.5

0 10 20 30 40 50 60 70 80

Dec

. rat

e (m

/s2 )

Distance to the end-of-queue

Deceleration Rate

Car Scooter

60.0% 65.0% 70.0% 75.0% 80.0% 85.0% 90.0% 95.0% 100.0%1

1.5

2

2.5

3

3.5

Filtering Speed under Different Queue Conditions

Occupied space in the stop queue segment(%)

Filt

erin

g Sp

eed

(m/s

)

0%

20%

40%

60%

80%

100%

-4 -2 0 2 4 6 8 10 12

Lane

cha

ngin

g pr

obab

ility

(%)

Queue length difference (vehicles)

Lane Changing Probabilities under Different Queue Length Differences

Logit Model

Real Data

2 0.89

256

R

N samples

Time

Trav

el S

peed

Propagation distance l lL t

,ˆacc lv t

v

,ˆ ffs lv t ,ˆdec lv t

Time

Trav

el S

peed

v

FuPropagation

Filtering l lL t

l t

Speed trajectory of vehicles joining the end-of-queue with and without filtering behavior

0 10 20 30 40 50 60 70 800

10

20

30

40

50

60

Car Speed

Distance to the end-of-queue (m)

Spee

d (k

ph)

0 10 20 30 40 50 60 70 800

10

20

30

40

50

60

70

Scooter Speed

Distance to the end-of-queue (m)

Spee

d (k

ph)

Speed of vehicles during deceleration process

4% 19% 23% 47% 57% 67% 86%

REAL 626 2800 4168 6814 6267 8585 12462

ESTIMATED 596 2710 3231 6553 7913 9333 12000

Difference -5% -3% -22% -4% 26% 9% -4%

-50%

-30%

-10%

10%

30%

50%

0

2000

4000

6000

8000

10000

12000

14000

Scoo

ter d

isch

arge

rate

(vph

)

Scooter occupied space (%)

Scooter Discharge Rate

0

5

10

15

20

25

30

35

40

45

50

0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84

Que

ue L

engt

h (m

eter

s)

Time after red phase start (seconds)

Lane 1 (car-only)

Estimated Real

0

5

10

15

20

25

30

35

40

0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 57 60 63 66 69 72 75 78 81 84

Que

ue L

engt

h (m

eter

s)

Time after red phase start (seconds)

Lane 3 (scooter-mixed)

Estimated Real

0

0.5

1

1.5

2

2.5

3

0

1

2

3

4

5

6

7

8

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

Arriv

al R

ate

(veh

/sec

ond)

Que

ue le

ngth

(met

ers)

Time (seconds)

Queue Length & Arrival Rate of Lane 3

Arriving Scooters Arriving Cars

Estimated Real

Lane 1

Lane 2

Lane 3

2vI t

1vI t 1

,v mn t

2,v mn t

2,3,v mE t

2 2,v m vA t t

Sub-lane 1

Sub-lane 2

Sub-lane 3

2 2ˆm vt t

2 212 2

1 2 21

m

t tt t

u t t

2 2,v m vQ t t

1 1,v m vA t t

2v t

Lane ChoiceAcceleration

& DecelerationMerging into Stop Queues Stop Queues

Scooter Filteringthrough Stop Queues

Upstream Arrivals

2,v mX t

2,11,mx t

Acceleration/deceleration rates of vehicles

Sub-lane 1

Sub-lane 2

Sub-lane 3

ˆ l t l

ll

tt

u t

,lv mQ t

Filtering Speed lu t

Filtering Distance l t

Car Queue Line