Trip Generation CE 451/551

1
Trip GenerationCE 451/551
Grad students need to discuss projects at end
of class
Source NHI course on Travel Demand Forecasting
(152054A)
2
Terminology

• Trip generation
• Person trip
• Vehicle trip
• Trip end
• Trip production
• Trip attraction
• Trip purposes
• Home-based work (HBW) trip
• Non-home based (NHB) trip others
• Special generator
• Socioeconomic data
• Demographic data

Image http//www.angryspec.com/scrounge.htm
3
Trip purposes

• Practice has shown that better travel forecasting
  models are obtained if trips by different
  purposes are identified and modeled separately.
  The most common trip purposes are
• HBW
• HBO
• NHB
• In TDF, trip productions and attractions are used
  to represent the ends of a trip. A production is
  the home end of an HB trip and the beginning of a
  NHB trip.
• HB trips (urban) constitute 70 of all trips

Others?
4
Trips, by purpose (the objective)
PA Table
5
Typical Trip Generation Process
Demographic and Socioeconomic inputs
Cross Classification Model
Trip Productions by zone, by purpose
Employment, attraction landuse data
Balance (system-wide)
Regression model
PA Tables, by purpose
Trip Attractions by zone, by purpose
6
Balancing attractions to productions
Rule of thumb original estimates of total
production and attractions should be within 10
of each other.
7
What is trip generation a function of?

• Land use
• Intensity
• Location/accessibility
• Time
• Type (person, transit, auto, walking )

Photo by enUserAude, taken on March 7, 2006
Graphic source http//www4.uwm.edu/cuts/utp/route
loc.pdf
8
Trip Generation

• Determine number of trip ends
• Methods
• Regression
• Cross Classification (tables)
• Rates based on activity units (ITE)

Image www.caliper.com
9
Regression
aggregation hides variability
Estimating a model

• Aggregate (zonal) or disaggregate (household)
• Linear or nonlinear
• Dependent (Y) variable is trips
• Independent (Xi) variables are
• Household attributes
• E.g., population, auto ownership, income level
• Employment attributes
• E.g., number of employees or size of
  establishments
• Could include network attributes?
• Be careful of co-linearity, power
• Can use your own data (best?) or borrow
  parameters

10
http//xkcd.com/503/ This work is licensed under
a Creative Commons Attribution-NonCommercial 2.5
License. This means you're free to copy and share
these comics (but not to sell them). More details
11
Cross classification models
See wiki on Contingency tables

• Breaks the trip generation process into steps
• Relies on aggregate data collected from surveys
  (like Census), like average income by
• income categories
• auto ownership
• Trip rate/auto
• Trip purpose
• Resembles regression, but non-parametric (like
  regression with dummy variables)
• Groups households in different strata
• 1-4 submodels (table based)
• Improved by adding info

• Advantages
• No prior info on shape of curves must be assumed
• Simple, easy to understand
• Can be used to account for time, space
• Disadvantages
• Does not permit extrapolation
• No goodness of fit measures
• Requires large sample size

From Amarillo 1990 model docs, ITE
12
One step Cross classification model (productions)
HBW
2007 eq.
0-8000
8K-16K
16K-32K
32K-56K
56K plus
Note US avg. median HH income 30K in 1990
is now 50,000 (2007)
From Amarillo 1990 model
13
One step Cross classification model (productions)
NHB
2007 eq.
0-8000
8K-16K
16K-32K
32K-56K
56K plus
From Amarillo 1990 model
14
Multi-step Cross Classification ExampleSource
ITE (Univ. of Idaho)
15
First Develop the family of cross class curves
and find number of households in each income group
00
Given (from survey)
Note orange lines show how to develop the curves
M
H
L
L
16
Now find percent of households in each auto
ownership/income group class
17
A
Given (from survey)
15K 25K 55K
18
Now find trips per households in each auto
ownership/income group class
19
Given (from survey)
B
20
Now find trips by purpose in each income group
class
21
Given (from survey)
C
22
Recall the problem
For the zone multiply the number of households
in each income group (00) by the percent of
households owning certain number of cars by
income group (A) to get the total number of
households by auto ownership in each income
group (00 x A) see next slide series Multiply
the result (00xA) by the number of trips
generated by each income group/auto ownership
category (B) to get trips by income group/auto
ownership category (00xAxB). Sum to get trips by
income level (?(00xAxB)). Multiply this sum by
the percent of trips by purpose (C) to get trips
by purpose by income group (Cx?(00xAxB)). Sum
over all income groups to get (total trips by
purpose from the zone). ANS
23
00
A
Low
x
Med
High

00xA
B

x
24
00xAxB
C
x

Cx?(00xAxB)
25
Cross classification model (attractions)
Note Less data than for productions, can use
cross-class or regression, most common
classification is by type of employment
1998 Austin, TX household travel survey
26
See also Wisconsin Trip Rate Files(Madison has
annotation)
Experience Based Analysis
Click in slideshow mode
27
Typical trip gen application

• Traffic engineers use rates (e.g. ITE), why?
  (data, peak)
• Planners use cross class and regression, why?
  (purpose, forecasting)
• Can we use rates in the TDF? How?
• http//www.ite.org/tripgen/Trip_Generation_Data_Fo
  rm.pdf

28
Special generators

• Shopping malls (large)
• Hospitals (different)
• Military institutions
• Airports (large)
• Colleges and universities (large, different)
• Stadiums (off peak)
• Elderly housing (small)

Click in slideshow mode