a dynamic traffic assignment model for the gta...2014/12/11  · a dynamic traffic assignment model...

A Dynamic Traffic Assignment Model for the GTA

Hossam Abdelgawad, Ph.D., P.Eng

Islam Taha, PhD Student

Aya Aboudina, PhD Candidate

Baher Abdulhai, Ph.D., P.Eng.

STA and DTA

In a network with many OD zones and a time period of interest, for each OD pair and departure time, all used routes have equal and lowest experienced travel time (generalized cost). No user may lower his experienced travel time through unilateral action (deterministic DTA).

In a network with many OD zones, for each OD pair, all used routes have equal and lowest travel time (generalized cost). No user may lower his travel time through unilateral action (deterministic STA).

2

Experienced vs Instantaneous Travel Time

3

2

2

1

1

4

3

3

2

1

4

6

5

3

1

Link travel time

when entering the

link at each minute

Travel time for departing at min 1 = 3 min

Travel time for departing at min 2 = 6 min

4

(a) Instantaneous travel time calculation

3

2

2

1

1

4

3

3

2

1

4

6

5

3

1

Link travel time

when entering the

link at each minute

Travel time for departing at min 1 = 9 min

Travel time for departing at min 2 = 8 min

4

(b) Experienced travel time calculation

3

Why / When do you need Dynamic ?

Captures the onset and spread of congestion over space and time

Essential to design control strategies:

– Ramp Metering

– Dynamic Congestion Pricing

– Traffic Signal Control

– Evacuation Optimization

– ….etc.

4

14 16 18 20

Time of day (hour)

Speed (

km

/hr)

Base Case (Jameson Closed)

20

40

60

80

100

Applications – Mesoscopic Level Evacuation of City of Toronto

5

Optimized Evacuation of Toronto 2.25 M People

6

Avg Evacuation Time = 2 Hrs Network Clearance Time= 8 Hrs Avg Stop Time = 0.5 Hrs

Avg Evacuation Time = 7 Hrs Network Clearance Time= 30 Hrs Avg Stop Time = 6 Hrs

Op

tim

al

Pla

n (

OS

TE

)

Do

-No

thin

g

Calibration

Verification

Overall Modelling Approach

Input Data

….

….

……

DTA

Model

Parameters

Validation

Outputs Observed

Data

7

Verification of Input Data

Consistency, volume balance (multiple sources of data)

Volumes and Speeds (congested or uncongested)

8

A D

B

C

E

F

D = A+B+C D= E+F

Speed

Flow Capacity

Field Data

Model Output

High

Low

Notes on Demand Data

Capacity vs Demand

Time-dependent demand (demand profile)

9

Flo

w

Time

Capacity S

tart

poin

t o

f q

ue

ue

En

d p

oin

t o

f q

ue

ue

Analysis Period (say AM Peak from 7 am to 9 am)

Warm-Up

Period

Demand profile

Demand

Calibration: Iterative Process

10

Parameter Tweaking

Error Checking

Calibration

Behavior

Model

Demand

Observation and

Feedback

Model Capacity

Model Scope: GTA

11

Demand Shifting from GTHA to GTA

12

Without GTHA time shifting demand

Calibration Criteria

13

Example:

a) v= 10000, c = 9000 % Diff =10%

b) v= 1000, c=900 % Diff =10%

Example:

a) v= 10000, c = 9000 GEH=10.3

b) v= 1000, c=900 GEH=3.2

GEH < 5 (Perfect)

5 < GEH < 10 (acceptable)

10 < GEH < 15 (question data quality or model performance)

Calibration Results (Morning Peak)

14

0

1000

2000

3000

4000

5000

6000

7000

8000

Ho

url

y V

olu

me (

veh

/ h

r)

Volumes along HWY401-EB-Col Observed (lower limits) Observed (upper limits) Simulated

Calibration Results (Morning Peak)

15

0

20

40

60

80

100

120

140

Ho

url

y V

olu

me (

veh

/ h

r)

Speeds along HWY401-EB-Col Observed (lower limits) Observed (upper limits) Simulated

Calibration Results (Morning Peak)

16

within limits

Avg GEH = 12.8

HWY 401

QEW

DVP

HWY 404

HWY 400

HWY 410

HWY 403

Gardiner Exp

LakeShore Blvd

Thank You!

Hossam Abdelgawad Hossam.Abdelgawad@alumni.utoronto.ca

Baher Abdulhai baher.abdulhai@utoronto.ca