INCORPORATING CYCLING IN
OTTAWA-GATINEAU TRAVEL MODEL
Surabhi Gupta & Peter Vovsha, Parsons Brinckerhoff Inc.
Ahmad Subhani, City of Ottawa
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Ottawa-Gatineau Region
National Capital Region
Fourth Largest Urban area in Canada (1.2 million population; 670K jobs)
Ottawa region is 2,796 square kms (1,800 square miles)
On the banks of the Ottawa, Rideau, and Gatineau rivers
Transportation System
Good Network of Highways
Transit Options – Local Buses, Express Buses, O-Train
Transitway (BRT)
LRT (under development)
Other Modes –
Spring/Summer/Fall – Bicycling
Winter – Skating on the Canal
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Why Cycling?
Targets for Cycling in Ottawa
Strong growth in cycling observed between 2005 and 2011
Regional bicycling share 2005 2011
Daily 1.25% 1.75%
Peak Periods 1.5% 2.5%
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Promoting Bike-and-Ride
4
Focusing on Cycling Networks around Transit Oriented Development Zones
Providing ample parking at Transit Stations
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Existing Model
Java based population synthesizer
Daily tour-based structure for travel generation and spatial
distribution
EMME based
Limited segmentation (HH variables)
Conventional trip-based mode choice and traffic/transit
simulations for AM and PM
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Objective
Bike as an additional mode in mode choice model
Detailed bike routing/assignment based on LOS measures
with cross-impacts of auto traffic and bike movements
Develop LOS measures for bicycles that include link and
node-level factors
Consider effect of perceived bike LOS measures on mode
choice
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Model
Traffic Stress (Mineta)
BLOS (Landis et al)
BCI (FHWA-RD-98-095)
BLOS (NCHRP-
616) SFCTA Model
BLOS (Dixon)
League of IL Bicyclists
Bike Route Preferences
(Stinson and Bhat)
NY State BCI Variables
Link-Level Type
Vehicle Flow Rate Y Y Y EN Number of Through Lanes Y Y Y EX Speed Y Y Y Y Y Y EX % Heavy Vehicles Y Y Y Y EN Pavement Condition Y Y Y Y EX Lane Width Y Y Y Y EX Shoulder (yes/no) Y Y Y Y EX Parking (Yes/No) Y Y Y Y Y EX Parking Width Y Y Y EX Bike Lane Blockage (Yes/no) Y Y EX
Bike Facility /Lane/Shoulder Y Y Y Y Y EX
Bike Lane Width Y Y EX Shoulder Width Y Y Y EX Type of Road: Residential, Arterial etc Y Y Y EX Trip Generation Intensity Y EN Driveways (Yes/No) Y EX Unrestricted Sight Distance Y EX Terrain (Flat, Hilly, Mountainous) Y EX AADT Y Y EX Median (Yes/No) Y EX Curb Lane Width Y Y Y EX
Node/Turn Level Type Signal (Yes/No) Y EX Right Turn Lane (Yes/No) Y Y Y EX Left Turn Lane (Yes/No) EX Right Turn Pocket Lane (Yes/No) Y EX Cross Street Width Y EX Turn Movements EX Right Turn Lane Length Y EX Left Turn Lane Length EX
Variables affecting Bike LOS & route choice
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Summary of State of Practice
Large number of qualitative studies
First attempts to incorporate bike in mode choice (San
Francisco, Portland, LA)
No examples yet of bicycle network assignment
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Bicycle LOS and Volume-Delay Function
LOS can be segmented based on User class and Bike facility
type
Types of Users
Classification A (Mineta Report, Dill et al., Geller) – Strong, Enthusiastic and
Interested
Example splits in Portland (Dill et al.): 4% Strong, 9% Enthusiastic, 56%
Interested, 31% Never
Classification B (Bhat et al.) – Experienced and Inexperienced
Types of Bike Facilities - Grade Separated, Exclusive Lane,
Mixed-Traffic
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Bicycle travel time affected by:
Auto volume – high V/C ratio for autos implies a steeper bicycle
VDF as they have to navigate through high congestion for mixed-
traffic
Bicyclist type – Stronger/experienced bicyclists have higher free
flow time and lower sensitivity to congestion and auto traffic
Bike lane type – Easier to navigate through dedicated bike lane than
mixed traffic
Total effective capacity – effective capacity available to bikes
conditional on the modeled traffic volumes
Bicycle LOS and Volume-Delay Function
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Bicycling Facility Types in Ottawa Region
Multi-use Stone pathway and Asphalt pathways (shared by
pedestrians and bicyclists)
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Bicycling Facility Types in Ottawa Region
Exclusive Bike Lanes (physically separated or marked)
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Bicycling Facility Types in Ottawa Region
Sharrow Lanes
Bikes allowed in mixed traffic
Paved Shoulders
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Bike Travel Time Function (VDF)
Bike link
travel time
(VDF)
= Link Delay
Factor
(LDF)
× Bike Free
Flow Time
Cycling
Conditions
Auto
Congestion
Factor
(ACF)
×
bt
Reduced Speed
due to Auto
Congestion
)( bVt
Congestion
effect due
to other
Bicycles
×
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Link Delay Factor (LDF)
Delay (travel time) experienced by the bicyclist:
𝐋𝐢𝐧𝐤 𝐃𝐞𝐥𝐚𝐲𝐢𝐣𝐦 = 𝐋𝐃𝐅𝐢𝐣𝐦 × 𝐁𝐢𝐤𝐞 𝐅𝐫𝐞𝐞 𝐅𝐥𝐨𝐰 𝐓𝐢𝐦𝐞 ∀ 𝐢, 𝐣 ∈ 𝐀 ∀𝐦 ∈ 𝐌
Where, the link delay factor (LDF) is defined as:
𝐋𝐃𝐅𝐢𝐣𝐦 = 𝟏 + 𝐋𝐎𝐒𝐢𝐣𝐦 ∀ 𝐢, 𝐣 ∈ 𝐀 ∀𝐦 ∈ 𝐌
In turn, 𝐋𝐎𝐒𝐢𝐣𝐦 is defined as:
𝐋𝐎𝐒𝐢𝐣𝐦 = 𝐌𝐚𝐱 𝐟 𝐀𝐢𝐣, 𝐏𝐢, 𝐏𝐣, 𝐌𝐦 × 𝐀𝐢𝐣 , 𝟎
Where:
𝐀𝐢𝐣 = Link specific variables,
𝐏𝐢 = Downstream node-specific variables,
𝐏𝐣 = Upstream node-specific variables,
𝐌𝐦 × 𝐀𝐢𝐣 = Link-user specific interaction variables.
Link Delay
Factor (LDF)
1+LOS
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Link Delay Factor (LDF)- Example
Variables
Network Attribute Bicyclist
Units Value [A] Effect Multiplier
[B] [A] x [B]
Link-Level
Bicycle Lane (yes/no) N/A 0 Decrease -1.12 0
Sharrow Lanes N/A 1 Decrease -0.5 -0.5
Bike Lane Width Feet 5 Decrease -0.4 -2
Curb Lane Width Feet 10 Decrease -0.0498 -0.498
Traffic Speed kph 35 Increase 0.01375 0.481
Curb Lane Volume Vph 600 Not Good 0.002 1.2
Other Lane Volume Vph 1200 Not Good 0.0004 0.48
Parking Lane (yes/no) N/A 1 Increase 0.506 0.506
% Heavy Vehicle Volume Ratio 15 Increase 0.034 0.51
Frequency of driveways N/A 3 Increase 0.019 0.057
Pavement Condition (good/bad) 0-4 0 Increase 0.05 0
Node-Level
Signal N/A 1 Increase 0.011 0.011
LOS 0.23625
Free Flow Travel Time (FFTIME) mins 6
Delayed Travel Time (FFTIME x
(1+LOS)
mins 7.48
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Auto Congestion Factor (ACF)
Auto
Congestion
Factor
(ACF)
Reduced Speed
due to Auto
Congestion
Reduction in Bike Speed due to Auto congestion:
Only affects the bicycles in mixed traffic
For bicycle lanes and multi-use pathways, ACF = 1.0
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Link Volume Delay Function (Bicycle)
0
5
10
15
20
25
30
35
0 1000 2000 3000 4000 5000 6000
Tra
vel T
ime (
Min
s)
Bike Volume
High Auto V/C
Medium Auto V/C
Low Auto V/C
Parameters
α β γ θ μ ν δ LDF
5000 6 0.1 4 1 1 0.5 4 0.1 1.2
𝑡 𝑉𝑏 = 𝑡𝑏 × 1 + 𝛿𝑏 𝑉𝑏𝐶𝑒𝑓𝑓
𝐸𝑥𝑝 𝑒𝑓𝑓
Bike VDF is given by:
Congestion effect due
to other Bicycles
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Link Volume Delay Function (Auto)
Impact of bicycles on auto travel times
Presence of bicycles increase auto travel
times
Bicycles take up capacity, and since they move
slower than autos, they take up more capacity
than their physical dimensions
The reduction in capacity due to bicycles
accounts for these impacts
3
3.5
4
4.5
5
5.5
0 1000 2000 3000 4000 5000
Tra
vel T
ime (
min
s)
Auto Volume
High Bike
Volume
Medium Bike
Volume
Low Bike
Volume
High Bike
Volume
Medium Bike
Volume
Low Bike
Volume
Bikes in Mixed-
Traffic
Dedicated Bike Lane
Parameters
ζ β
5000 3 0.1 3 1 1.33
PCE 0.8
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Iterative Auto-Bicycle Assignment
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Modeled Bike Flow Map: AM Peak Hour
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Model Validation
0
200
400
Westbound Eastbound
Count Volume
Laurier at Metcalfe
0
100
200
300
Southbound Northbound
Count Volume
Protage Bridge
0
100
200
Southbound Northbound
Count Volume
Alexandre Bridge
0
50
100
150
Southbound Northbound
Count Volume
Colonel By Dr
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
Future Improvements: Turn Penalties
j
i
k
k
Turn
i—j—k
k
𝒌 k
𝒌
j
i
𝑓𝑘 𝑉𝑎 , 𝑉𝑏, …
𝑓𝑘 𝑉𝑎 , 𝑉𝑏, …
𝑓𝑘 𝑉𝑎 , 𝑉𝑏, …
For turn i-j-k:
𝑡 𝑉𝑎 = 𝑇𝑃𝑎𝑖𝑗𝑘 + 𝑡0 × 𝐿𝐹 × 1 + 𝜁𝑎 𝑉𝑎 + 𝑝𝑏𝑙𝑏 𝑉𝑏
𝐶
𝛽𝑎
𝑡 𝑉𝑏 = 𝑇𝑃𝑏𝑖𝑗𝑘 + 𝐴𝐶𝐹 × 1 + 𝜁𝑏 𝑉𝑏𝐶𝑒𝑓𝑓
𝐸𝑥𝑝𝑒𝑓𝑓
Innovations in Travel Modeling, April 27-30, 2014, Baltimore, MD
QUESTIONS!