Comparison of EMME Transit
Assignment Methods
Optimal Strategies
vs
Strategies with Variants (path assignment)
Karen Tsang
Bureau of Transport Statistics
Department of Transport
May 2011
Overview
A series of transit assignment experiments
◦ Compare standard and new methods
◦ To understand how flows are distributed
◦ Using simplified network (2 transit services)
◦ Variables : in-veh time, wait time, headway
choice between centroid connectors
Sydney Strategic Travel Model (STM) Network
◦ Rail Assignment Example
Standard vs New
Current
New Release
Module
5.31
5.32
Assignment
Preparation
5.11
5.32
Option
One option:
Two options:
1 = Optimal Strategies
1 = Optimal Strategies
* no path file
* path file saved
* assignment results same as current
if additional gc and variants inputs
not specified
2 = Strategies with Variants
* path file saved
* distribute flows based on frequency
and transit time (optional)
* multiple paths at centroid connectors
(optional)
Standard vs New
Current
New Release
Module
5.31
5.32
Assignment
Preparation
5.11
5.32
Option
One option:
Two options:
1 = Optimal Strategies
1 = Optimal Strategies
* no path file
* path file saved
* assignment results same as current
if additional gc and variants inputs
not specified
STANDARD
2 = Strategies with Variants
NEW
* path file saved
* distribute flows based on frequency
and transit time (optional)
* multiple paths at centroid connectors
(optional)
Experiment 1:
• Demand from origin to destination = 100 passengers
• No auxiliary (walk) link choice
• New service variables
Headway: 5, 10, 15 and 20 minutes
Travel time: 1 to 40 minutes in 1 min interval
Percentage of passengers using New service?
Experiment 1:
Assignment attributes and weighting factors:
In-vehicle time factor = 1.0
Auxiliary (walk) travel time factor = 2.0
Wait time factor = 2.0
Wait time = Headway/ 2
Boarding time = 5 min
Boarding time factor =1.0
Experiment 1:
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
Waiting time weight = 2.0
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
Waiting time weight = 2.0
100
100
New
Headway
= =5=5min
New
New
Headway
Headway
5min
min
New
Headway
= =10
min
New
New
Headway
Headway
=10
10
min
min
New Headway = 15 min
New
NewHeadway
Headway==15
15min
min
New Headway = 20 min
New
New
Headway
Headway
=
=
20
20
min
min
Base Headway = 10 min
Demand % on New Line
80
70
New Headway = 5 min
90
New Headway = 10 min
New
New Headway
Headway =
= 15
15 min
min
80
Demand % on New Line
90
60
50
40
30
New
New Headway
Headway =
= 20
20 min
min
70
60
50
40
30
20
20
10
10
0
0
0
5
10
15
20
25
Travel time on New Line
STANDARD
Distribution of flows
based on frequency
30
35
40
0
5
10
15
20
25
Travel time on New Line
30
35
40
NEW
Distribution of flows
based on frequency and transit time
Experiment 2: Reduced wait time weight
Wait time weight = 2.0 (Experiment 1)
Wait time weight = 1.0 (Experiment 2)
Experiment 2: Reduced wait time weight
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
Waiting time weight = 1.0
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
Waiting time weight = 2.0
100
100
New Headway = 5 min
New Headway = 5 min
New Headway = 10 min
New Headway = 15 min
New Headway = 20 min
Demand % on New Line
80
70
90
New Headway = 10 min
New Headway = 15 min
80
Demand % on New Line
90
60
50
40
30
60
50
40
30
20
20
10
10
0
New Headway = 20 min
70
0
0
5
10
15
20
25
Travel time on New Line
STANDARD
Wait time weight = 1.0
30
35
40
0
5
10
15
20
25
Travel time on New Line
30
STANDARD
Wait time weight = 2.0
Results from Experiment 1
35
40
Experiment 2: Reduced wait time weight
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
Waiting time weight = 1.0
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
Waiting time weight = 1.0
100
100
New Headway = 5 min
New Headway = 10 min
New Headway = 15 min
New Headway = 20 min
Demand % on New Line
80
70
New Headway = 5 min
New Headway = 10 min
New Headway = 15 min
New Headway = 20 min
90
80
Demand % on New Line
90
60
50
40
30
70
60
50
40
30
20
20
10
10
0
0
0
5
10
15
20
25
Travel time on New Line
30
STANDARD
Wait time weight = 1.0
Flow distribution based on
frequency
35
40
0
5
10
15
20
25
Travel time on New Line
30
NEW
Wait time weight = 1.0
Flow distribution based on
frequency and transit time
35
40
Experiment 3: Choice between connectors
• Demand from origin to destination = 100 passengers
• Choice between 2 centroid connectors
• New service variables
Headway: 5, 10, 15 and 20 minutes
Travel time: 1 to 40 minutes in 1 min interval
Percentage of passengers using New service?
Experiment 3: Choice between connectors
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
Waiting time weight = 2.0
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
100
100
90
New Headway = 10 min
80
Demand % on New Line
New Headway = 15 min
80
Demand % on New Line
New Headway = 5 min
New Headway = 10 min
New Headway = 15 min
New Headway = 20 min
90
New Headway = 5 min
New Headway = 20 min
70
60
50
40
30
70
60
50
40
30
20
20
10
10
0
0
0
5
10
15
20
25
Travel time on New Line
30
STANDARD
One service route is chosen
All or nothing
35
40
0
5
10
15
20
25
30
35
40
Travel time on New Line
NEW
Two centroid connector choices
Option 2 (logit)
Scale parameter = 0.2 (default)
where exp(-scale * transit time to destination)
Experiment 3: Choice between connectors
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
Waiting time weight = 2.0
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
100
100
90
New Headway = 10 min
80
Demand % on New Line
New Headway = 15 min
80
Demand % on New Line
New Headway = 5 min
New Headway = 10 min
New Headway = 15 min
New Headway = 20 min
90
New Headway = 5 min
New Headway = 20 min
70
60
50
40
30
70
60
50
40
30
20
20
10
10
0
0
0
5
10
15
20
25
Travel time on New Line
30
STANDARD
One service route is chosen
All or nothing
35
40
0
5
10
15
20
25
30
35
40
Travel time on New Line
NEW
Two centroid connector choices
Option 2 (logit)
Scale parameter = 0.5
where exp(-scale * transit time to destination)
Experiment 3: Choice between connectors
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
Waiting time weight = 2.0
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
100
100
90
New Headway = 10 min
80
Demand % on New Line
New Headway = 15 min
80
Demand % on New Line
New Headway = 5 min
New Headway = 10 min
New Headway = 15 min
New Headway = 20 min
90
New Headway = 5 min
New Headway = 20 min
70
60
50
40
30
70
60
50
40
30
20
20
10
10
0
0
0
5
10
15
20
25
Travel time on New Line
30
STANDARD
One service route is chosen
All or nothing
35
40
0
5
10
15
20
25
30
35
40
Travel time on New Line
NEW
Two centroid connector choices
Option 2 (logit)
Scale parameter = 0.8
where exp(-scale * transit time to destination)
Experiment 3: Choice between connectors
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
Waiting time weight = 2.0
Percentage of Demand taking the New Line
(Base line: Hwy=10, Time = 20)
100
100
90
New Headway = 10 min
80
Demand % on New Line
New Headway = 15 min
80
Demand % on New Line
New Headway = 5 min
New Headway = 10 min
New Headway = 15 min
New Headway = 20 min
90
New Headway = 5 min
New Headway = 20 min
70
60
50
40
30
70
60
50
40
30
20
20
10
10
0
0
0
5
10
15
20
25
Travel time on New Line
30
STANDARD
One service route is chosen
All or nothing
35
40
0
5
10
15
20
25
30
35
40
Travel time on New Line
NEW
Two centroid connector choices
Option 2 (logit)
Scale parameter = 1.0
where exp(-scale * transit time to destination)
Key Differences
1. Distribution of flows to attractive lines
Standard: based on frequency
New:
based on frequency and transit time (optional)
2. Distribution of flows between connectors at centroids
Standard: one path with best generalised time (AON)
New:
multiple paths at centroid connectors (optional)
Experiment Results
Flow Distribution on attractive lines
• Standard = Step function
• New = Step with transition logit curve
Flow Distribution between multiple centroid connectors
• Standard = AON
• New = AON or multiple preferred paths
Sydney Strategic Travel Model (STM)
Rail Network
Rail Stations:
Over 340
Rail Transit Lines:
80 (approx)
Rail Link Length:
Approx 2400 km
Sydney Strategic Travel Model (STM)
Rail Line Example
Sydney Strategic Travel Model (STM)
Rail Lines (Entire Network)
STM Network - Rail Assignments
Fixed Demand: 3.5-hr rail AM demand
Network: Rail Network with walk and bus access/egress
Assignment Methods:
Travel zone to Travel zone assignments
(1) Standard Assignment – Optimal Strategies
(2) Strategies with Variants
– Path Saved
– Distribution of flows based on frequency and transit time
– Scale Parameter = 0.5
Rail Assigned Volumes
Method 1: Optimal strategies (standard)
Rail Assigned Volumes
Method 2: Strategies with variants (new)
Rail Assigned Volume Differences
-6 %
+4 %
+1 %
+4 %
Green =
Less volumes in
new method
Red =
More volumes in
new method
Conclusions: Rail Assignment
• Rail assignment differences up to +/– 6%
• Minimise back-tracking
• Retain fast assignment run time
Standard = 1 minute, New = 2 minutes
• Increase attractiveness of express services