what is the model??? a primer on transportation demand forecasting models shawn turner theo...
TRANSCRIPT
What is the Model??? A Primer on Transportation
Demand Forecasting ModelsShawn TurnerTheo Petritsch
Keith LovanLisa Aultman-Hall
Objectives
• Provide overview of motor vehicle travel demand forecasting process
• Discuss options and alternatives for including biking and walking into vehicle forecasting models
Origin
• Transportation Research Board Committee on Bicycle Transportation
• Subcommittee recommendation January 2004
What’s the Problem?
• The “model” forecasts “tons” of motorized traffic and so engineers must build bigger roads
• The “model” rarely includes peds and bikes• In fact, the “model” structure and assumptions
do not work for bikes and peds• But decision makers like numbers, so we need
to use other techniques to quantify the activity in non-motorized transportation
The Model1. Road Network2. Zones in Study Area3. Attributes of Zones
Traffic Volume by Road Link (section)
Defining the Study Area
1. Define boundaries
2. Establish the network
3. Create the zones
Defining the Study Area
1. Define Boundaries• Numerous states have statewide models• Most models are used at the regional or
metropolitan planning organization level
-study area may be larger than appropriate-statewide may be appropriate for tourism
PROBLEM
Defining the Study Area
2. Establish the network• main arterial roads are represented as a series of links
and nodes• links are defined by speed and capacity• turns are allowed at nodes
Defining the Study Area
2. Establish the network
- links are defined by speed and capacity
- turns are allowed at nodes
-only main roads and intersections are included-even collector roads are excluded-off-road facilities are not included
PROBLEM
Defining the Study Area
3. Create traffic analysis zones
-uniform land use
-bounded by major roads
-neighborhood size (a few blocks or more)
Defining the Study Area
3. Create the zones
-State of CT model – 2000 zones (5500 square miles and 3.4 million people)
-Lexington-Fayette County Ky - *** zones (population 250,000 and 293 square miles)
Defining the Study Area
3. Create the zones
-all modeled trips begin in a zone and are destined for a zone
-zone size is so large that most bike and pedestriantrips start and end in the same zone and do not use the network being modeled-the typical zone attributes of population and employment are not necessarily enough to predictlevels of biking or walking
PROBLEM
The Model1. Road Network2. Zones in Study Area3. Attributes of Zones
Traffic Volume by Road Link (section)
What’s in the Black Box?
• The four-step model– Originally developed in the 1950s and 1960s
for interstate highway planning– Many advanced newer modeling techniques
have been developed by researchers– For the most part agencies still use the “four
step” model
The Four Step Model
1. Trip Generation
2. Trip Distribution
3. Mode Choice
4. Trip or Route Assignment
The Four Step Model
• Trip Generation – use of linear regression to predict the number of trips beginning from and destined for each zone based on population and employment
-there are relatively few bicycle and pedestrian tripsbut note this process considers total trips at this point, non-motorized trips have not necessarily been excluded-trip generation can be conducted by trip purpose (work or shopping or total daily trips are common) but note that recreational and discretionary tripsare hard to include
PROBLEM
The Four Step Model
• Trip Distribution – use of gravity models to link trip origin zones
to trip destination zones– a trip matrix is produced
The Four Step Model
• Trip Distribution – Trip interchanges increase with decrease
distance between zones– Trip interchanges increase with increased
zone “attractiveness” (square footage of retail or population for example)
The Four Step Model
• Trip Distribution
-all trips are still lumped together
-but bike and ped trips are affected differently
by distance
-many bike and ped trips are intra-zonal and this
distance is hard to representPROBLEM
The Four Step Model
• Mode Choice– utility choice models predict the probability
that a trip between a pair of origin and destination zones will be made by each mode
– drive alone, carpool, transit are commonly included
– some agencies do little mode choice as most of the trips they are modeling occur by automobile
The Four Step Model
• Mode Choice-Many agencies exclude bikes and peds completely
-The small percentage of bike/ped trips makes it
difficult to calibrate models and include them
-Typical survey methods result in under-reporting of
discretionary trips, short trips and bike/ped tripsPROBLEM
The Four Step Model
• Traffic or Route Assignment– for a trip between an origin zone and a
destination zone the sequence of links (roads) in the network for routing are selected
-often based on just minimum distance or time
-some models do more than one route for each OD pair
The Four Step Model
• Traffic or Route Assignment
-The routes used by bikes and walkers are not
in the network
-Trips starting and ending within one zone are
not assigned
-Factors beyond time and distance are also
important to non-motorized routing
PROBLEM
The Four Step Model
• Conclusions for Bike and Ped Modes– The geographic scale could be wrong– The network is wrong– The input variables are too limited– Trip purpose is usually either too specific or
too general
Options and Alternatives
– 2 options1. Tweak existing model or “post process”
intermediate or final model output
2. Use sketch planning tools outside of formal model
– Burden of proof should be on transportation professional
– Remind modelers how results will be used:• Relative demand for prioritization• Mode shift potential