TRAFFIC ENGINEERING

TRANSPORTATION PLANNING & TRAFFIC ESTIMATION

OBJECTIVES
1. Identify highway system components
2. Define transportation planning
3. Recall the transportation planning process and its design purposes
4. Identify the four steps of transportation demand modeling and
describe modeling basics.
5. Explain how transportation planning and modeling process results are
used in highway design.
HIGHWAY SYSTEM COMPONENTS
1. Vehicle
2. Driver (and pedestrian/bikes)
3. Roadway
4. Consider characteristics, capabilities, and
interrelationships in design

Start with demand needs (number of lanes?)

TRANSPORTATION PLANNING
(ONE DEFINITION)
Activities that:
1. Collect information on performance
2. Identify existing and forecast future system
performance levels
3. Identify solutions

Focus: meet existing and forecast travel demand

WHERE DOES PLANNING FIT IN?
TRANSPORTATION PLANNING
IN HIGHWAY DESIGN
1. identify deficiencies in system
2. identify and evaluate alternative alignment impacts on system
3. predict volumes for alternatives
▪ in urban areas … model? … smaller cities may not need (few
options)
▪ in rural areas … use statewide model if available … else: see
lab 3-type approach (note Iowa is developing a statewide
model)
EXAMPLE
EXAMPLE
EXAMPLE
TRUCK TRAFFIC
PLANNING AT 3 LEVELS
▪ State … STIP Statewide Transportation Improvement
Program (list of projects)
▪ Regional … MPO Metropolitan Planning Organization
(>50,000 pop.), 25 year long range plan and TIP
(states now also do LRP)
▪ Local …project identification and prioritization
FOUR STEP MODEL
THE BASIC TRANSPORTATION MODEL…

1) Study Area Zones

Inputs 2) Attributes of Zones
 Socioeconomic Data
 Land Use Data
 “Cost” of Travel btw. Zones
3) The Road Network
Outputs

 Traffic Volume by Road Link

 Mode Splits
 Emissions
WHAT’S IN THE BLACK BOX?
The Four-Step Model
FOUR STEPS OF CONVENTIONAL
TRANSPORTATION MODELING
1. Trip Generation
2. Trip Distribution
3. Mode Split
4. Trip Assignment
OVERVIEW OF THE FOUR-STEP MODEL
P/A VERSUS O/D
Trip Production
Home end of the trip
(where the need to make a trip is generated)
Trip Attraction
Non-home end of the trip
(where the need to make a trip is satisfied)

Trip Origin
# of trips that begin in a zone
Trip Destination
# of trips the end in a zone
 P/A VERSUS O/D
Zone 1
Residential
Zone 1 = Production & Origin

Zone 1 =
Production & Destination

Zone 2 =
Attraction & Destination

Zone 2
Non-Residential

Zone 2 = Attraction & Origin

SOME GENERAL PROBLEMS WITH THE
CONVENTIONAL METHODOLOGY
▪ Focus on vehicular traffic
▪ Better models typically have a transit
component
▪ Typically forecasts huge increases in traffic
▪ Leads to engineers building bigger roads
to accommodate “forecast” traffic
▪ Which leads to induced traffic and congestion…
right back where we started when we
needed the bigger roads in the first place
SOME GENERAL PROBLEMS WITH THE
CONVENTIONAL METHODOLOGY
▪ Pedestrians and bicyclists
are rarely included
▪ Level of geography is difficult for non-motorized modes
▪ Network scale is insignificant
▪ Input variables are too limited
STUDY AREA
▪ Clearly define the area under consideration
▪ Where does one entity end?
▪ May be defined by county boundaries, jurisdiction, town centers
STUDY AREA
▪ May be regional
▪ Metropolitan area – Des Moines including suburbs,
Ankeny, etc.
▪ Overall impact to major street/highway network

▪ Local – e.g., impact of trips to new Ames mall

▪ Impact on local street/highway system
▪ Impact on intersections
▪ Need for turning lane or new signal – can a model do this
level of detail?
STUDY AREA
▪ Links and nodes
▪ Simple representation of the geometry of the
transportation systems (usually major roads or
transportation routes)
▪ Links: sections of roadway (or railway)
▪ Nodes: intersection of 2+ links
▪ Centroids: center of TAZs
▪ Centroid connectors: centroid to roadway
network where trips load onto the network
TRAVEL ANALYSIS ZONES (TAZS)
▪ Homogenous urban activities (generate same types of trips)
▪ Residential
▪ Commercial
▪ Industrial
▪ May be as small as one city block or as large as 10 sq. miles
▪ Natural boundaries --- major roads, rivers, airport boundaries
▪ Sized so only 10-15% of trips are intrazonal
FOUR STEPS OF CONVENTIONAL
TRANSPORTATION MODELING
▪ Divide study area into study zones
▪ 4 steps
▪ Trip Generation
▪ -- decision to travel for a specific purpose (eat lunch)
▪ Trip Distribution
▪ -- choice of destination (a particular restaurant? The
nearest restaurant?)
▪ Mode Choice
▪ -- choice of travel mode (by bike)
▪ Network Assignment
▪ -- choice of route or path (Elwood to Lincoln to US 69)
MODEL STEP 1
Trip Generation
TRIP GENERATION
▪ Calculate number of trips generated in each zone
▪ 500 Households each making 2 morning trips to work (avg.
trip ends ~ 10/day!)
▪ Worker leaving job for lunch

▪ Calculate number of trips attracted to each zone

▪ Industrial center attracting 500 workers
▪ McDonalds attracting 200 lunch trips
TRIP GENERATION
▪ Number of trips that begin from or end in each TAZ
▪ Trips for a “typical” day
▪ Trips are produced or attracted
▪ # of trips is a function of:
▪ TAZs land use activities
▪ Socioeconomic characteristics of TAZ population
TRIP GENERATION

ModelManager 2000™
Caliper Corp.
TRIP GENERATION
3 variables related to the factors that influence trip production and
attraction (measurable variables)
▪ Density of land use affects production & attraction
▪ Number of dwellings, employees, etc. per unit of land
▪ Higher density usually = more trips
▪ Social and socioeconomic characters of users influence production
▪ Average family income
▪ Education
▪ Car ownership
▪ Location
▪ Traffic congestion
▪ Environmental conditions
TRIP GENERATION
▪ Trip purpose
▪ Zonal trip making estimated separately by trip purpose
▪ School trips
▪ Work trips
▪ Shopping trips
▪ Recreational trips
▪ Travel behavior depends on trip purpose
▪ School & work trips are regular (time of day)
▪ Recreational trips highly irregular
TRIP GENERATION
▪ Forecast # of trips that produced or attracted by each TAZ for a
“typical” day
▪ Usually focuses on Monday - Friday
▪ # of trips is forecast as a function of other variables
▪ Attraction
▪ Number and types of retail facilities
▪ Number of employees
▪ Land use

▪ Production
▪ Car ownership
▪ Income
▪ Population (employment characteristics)
TRIP PURPOSE
▪ Trips are estimated by purpose (categories)

▪ Work

▪ School

▪ Shopping

▪ Social or recreational

▪ Others (medical)

▪ Travel behavior of trip-makers depends somewhat on trip purpose

▪ Work trips
▪ regular
▪ Often during peak periods
▪ Usually same origin/destination

▪ School trips
▪ Regular
▪ Same origin/destination

▪ Shopping recreational
▪ Highly variable by origin and destination, number, and time of day
HOUSEHOLD BASED
▪ Trips based on “households” rather than individual
▪ Individual too complex
▪ Theory assumes households with similar characteristics have similar
trip making characteristics
▪ However
▪ Concept of what constitutes a “household” (i.e. 2-parent family,
kids, hamster) has changed dramatically
▪ Domestic partnerships
▪ Extended family arrangements
▪ Single parents
▪ Singles
▪ roommates
TRIP GENERATION ANALYSIS
▪ 3 techniques
▪ Cross-classification
▪ Covered in 355
▪ Multiple regression analysis
▪ Mathematical equation that describes trips as a function
of another variable
▪ Similar in theory to trip rate
▪ Won’t go into
▪ Trip-rate analysis models
▪ Average trip-production or trip-attraction rates for
specific types of producers and attractors
▪ More suited to trip attractions
TRIP ATTRACTIONS
EXAMPLE:
TRIP-RATE ANALYSIS MODELS
For 100 employees in a retail shopping center, calculate the total
number of trips
 Home-based work (HBW)
= 100 employees x 1.7 trips/employee
= 170
 Home-based Other (HBO)
= 100 employees x 10 trips/employee
= 1,000
 Non-home-based (NHB)
= 100 employees x 5 trips/employee
= 500

Total = 170 + 1000 + 500 = 1,670 daily trips

MODEL STEP 2
Trip Distribution
TRIP DISTRIBUTION
▪ Predicts where trips go from each TAZ
▪ Determines trips between pairs of zones
▪ Tij: trips from TAZ i going to TAZ j
▪ Function of attractiveness of TAZ j
▪ Size of TAZ j
▪ Distance to TAZ j
▪ If 2 malls are similar (in the same trip purpose), travelers
will tend to go to closest

▪ Different methods but gravity model is most popular

TRIP DISTRIBUTION
▪ Determines trips between pairs of zones
▪ Tij: trips from TAZ i going to TAZ j

▪ Function of attractiveness of TAZ j

▪ Size of TAZ j
▪ Distance to TAZ j
▪ If 2 malls are similar, travelers will tend to go to
closest
▪ Different methods but gravity model is most
popular
TRIP DISTRIBUTION

Caliper Corp.
Maricopa County
GRAVITY MODEL
Tij = Pi AjFijKij
Σ AjFijKij

Qij = total trips from i to j

Pi = total number of trips produced in zone i, from trip generation
Aj = number of trips attracted to zone j, from trip generation
Fij = impedance (usually inverse of travel time), calculated
Kij = socioeconomic adjustment factor for pair ij
MODEL STEP 3
Mode Choice
MODE CHOICE
▪ In most situations, a traveler has a choice of modes
▪ Transit, walk, bike, carpool, motorcycle, drive alone

▪ Mode choice/mode split determines # of trips

between zones made by auto or other mode, usually
transit
 CHARACTERISTICS INFLUENCING
MODE CHOICE
▪ Availability of parking
▪ Income
▪ Availability of transit
▪ Auto ownership
▪ Type of trip
▪ Work trip more likely transit
▪ Special trip – trip to airport or baseball stadium served by
transit
▪ Shopping, recreational trips by auto
▪ Stage in life
▪ Old and young are more likely to be transit dependent

45
 CHARACTERISTICS INFLUENCING
MODE CHOICE
▪ Cost
▪ Parking costs, gas prices, maintenance?
▪ Transit fare

▪ Safety
▪ Time
▪ Transit usually more time consuming (not in NYC or DC …)

▪ Image
▪ In some areas perception is that only poor ride transit
▪ In others (NY) everyone rides transit

46
MODE CHOICE MODELING
▪ A numerical method to describe how people choose
among competing alternatives (don’t confuse model
and modal)
▪ Highly dependent on characteristics of region
▪ Model may be separated by trip purposes
UTILITY AND DISUTILITY FUNCTIONS
▪ Utility function: measures satisfaction derived from choices
▪ Disutility function: represents generalized costs of each choice
▪ Usually expressed as the linear weighted sum of the
independent variables of their transformation
U = a0 + a1X1 + a2X2 + ….. + arXr

U: utility derived from choice

Xr: attributes
ar: model parameters
LOGIT MODELS
▪ Calculates the probability of selecting a particular
mode

p(K) = ____eUk__
 eUk

p: probability of selecting mode k

LOGIT MODEL EXAMPLE 1
Utility functions for auto and transit
U = ak– 0.35t1 – 0.08t2 – 0.005c
ak = mode specific variable
t1 = total travel time (minutes)
Do you agree with the
relative magnitude of
t2 = waiting time (minutes) the time parameters? Is
there double
c = cost (cents) counting/colinearity?
LOGIT MODEL EXAMPLE 1 (CONT)
Travel characteristics between two zones
Variable Auto Transit
ak -0.46 -0.07
Do you agree with the
t1 20 30 relative magnitude of
the mode specific
t2 8 6 parameters? How much
effect does cost have?
c 320 100

Uauto = -0.46 – 0.35(20) – 0.08(8) – 0.005(320) = -9.70

Utransit = -0.07 – 0.35(30) – 0.08(6) – 0.005(100) = -11.55

LOGIT MODEL EXAMPLE 1 (CONT)
Uauto = -9.70

Utransit = -11.55

Logit Model:

p(auto) = ___eUa __ = _____e-9.70 ____ = 0.86

eUa + eUt e-9.70 + e-11.55

p(transit) = ___eUt __ = _____e-11.55 ____ = 0.14

eUa + eUt e-9.70 + e-11.55
LOGIT MODEL EXAMPLE 2
The city decides to spend money to create and improve bike trails so
that biking becomes a viable option, what percent of the trips will be
by bike?
Assume:
 A bike trip is similar to a transit trip
 A bike trip takes 5 minutes more than a transit trip but with no
waiting time
 After the initial purchase of the bike, the trip is “free”
LOGIT MODEL EXAMPLE 2 (CONT)
Travel characteristics between two zones
Variable Auto Transit Bike
ak -0.46 -0.07 -0.07
t1 20 30 35
t2 8 6 0
c 320 100 0

Uauto = -0.46 – 0.35(20) – 0.08(8) – 0.005(320) = -9.70

Utransit = -0.07 – 0.35(30) – 0.08(6) – 0.005(100) = -11.55

Ubike = -0.07 – 0.35(35) – 0.08(0) – 0.005(0) = -12.32

LOGIT MODEL EXAMPLE 2 (CONT)
Uauto = -9.70, Utransit = -11.55, Ubike = -12.32 Notice that auto lost
share even though its
“utility” stayed the
Logit Model: same

p(auto) = _____eUa ____ = _______e-9.70 ______ = 0.81

eUa + eUt +eUb e-9.70 + e-11.55 + e-12.32

p(transit) = _____eUt__ __ = ______e-11.55 ______ = 0.13

eUa + eUt +eUb e-9.70 + e-11.55 + e-12.32

p(bike) = _____eUt__ __ = ________e-11.55 ______ = 0.06

eUa + eUt +eUb e-9.70 + e-11.55 + e-12.32
MODEL STEP 4
Traffic Assignment (Route Choice)
TRIP ASSIGNMENT
▪ Trip makers choice of path between origin and
destination
▪ Path: streets selected
▪ Transit: usually set by route
▪ Results in estimate of traffic volumes on each
roadway in the network
PERSON TRIPS VS. VEHICLE TRIPS
▪ Trip generation step calculated total person
trips
▪ Trip assignment deals with volume not person
trips
▪ Need to adjust person trips to reflect vehicle
trips
▪ Understand units during trip generation phase
PERSON TRIPS VS. VEHICLE TRIPS
EXAMPLE
Usually adjust by average auto occupancy
Example:
If:
▪ average auto occupancy = 1.2
▪ number of person trips from zone 1 = 550
So:
Vehicle trips = 550 person trips/1.2 persons per vehicle
= 458.33 vehicle trips
TIME OF DAY PATTERNS
▪ Trip generation usually based on 24-hour period
▪ LOS calculations usually based on hourly time period
▪ Hour, particularly peak, is often of more interest than
daily
TIME OF DAY PATTERNS
▪ Common time periods
▪ Morning peak
▪ Afternoon peak
▪ Off-peak

▪ Calculation of trips by time of day

▪ Use of factors (e.g., morning peak may be 11% of daily
traffic)
▪ Estimate trip generation by hour
MINIMUM PATH
▪ Theory: users will select the quickest route
between any origin and destination
▪ Several route choice models (all based on some
“minimum” path)
▪ All or nothing
▪ Multipath
▪ Capacity restraint
MINIMUM TREE
▪ Starts at zone and selects minimum path to each
successive set of nodes
▪ Until it reaches destination node

2
(3) (2)
(7)
1 5
4
(4) (4)
3
Path from 1 to 5
MINIMUM TREE
2
(3) (2)
(7)
1 5
4
(4) (4)
3

1. Path from 1 to 5 first passes thru 4

2. First select minimum path from 1 to 4
3. Path 1-2-4 has impedance of 5
4. Path 1-3-4 has impedance of 8
5. Select 1-2-4
ALL OR NOTHING
▪ Allocates all volume between zones to minimum path
based on free-flow link impedances
▪ Does not update as the network loads
▪ Becomes unreliable as volumes and travel time
increases
MULTI-PATH
▪ Assumes that all traffic will not use shortest path
▪ Assumes that traffic will allocate itself to alternative
paths between a pair of nodes based on costs
▪ Uses some method to allocate percentage of trips
based on cost
▪ Utility functions (logit)
▪ Or some other relationship based on cost

▪ As cost increases, probability that the route will be

chosen decreases
CAPACITY RESTRAINT
▪ Once vehicles begin selecting the minimum path
between a set of nodes, volume increase and so do
travel times
▪ Original minimum paths may no longer be the
minimum path
▪ Capacity restraint assigns traffic iteratively, updating
impedance at each step
SIZING FACILITIES
SIZING FACILITIES
SIZING FACILITIES
Four-Step Model Problem
4-STEP MODEL PROBLEM
Zone Information
Zone 1 2 3 4

Households 250 500 300 100

Area (sq. miles) 1.0 0.5 1.0 2.0
Retail Employees 65 390 215 100
Non-Retail Employees 400 300 85 10
% Low Income 40% 20% 50% 20%
% Medium Income 60% 30% 50% 80%

Given
Kij = 1 (socioeconomic adjustment factor for use in gravity the
zone data above and the
model)

tables/charts on the following page,

Find the vehicle flows on the road network
to the left using “all-or-nothing” traffic
assignment.

This will require completing all four steps

of the transportation planning process…
 4-STEP MODEL PROBLEM
% of HH in Each Income Category Versus Auto Ownership Trip Attraction Rates
Autos/HH Attraction per
0 1 2+
per HH 1.0
Income
Retail Employee 5.0
Low 50 40 10
Non-Retail Employee 2.0
Medium 10 50 40
High 10 30 60

# of Trips per Household per Day

Travel Time Between Zones
Autos/HH Zone 1 2 3 4
0 1 2+
1 2 2 4 8
Income
Low 2 4 8 2 2 4 2 6
Medium 4 5 10 3 4 2 1 5
High 4 5 8
4 8 6 5 2

F Factor vs. Travel Time

Alternative Modes vs. Density of Attraction Zone
90
45
80
40
70 35

% Walking & Biking

60 30
F Factor

50 25
40 20

30 15

20 10

10 5

0 0
0 2 4 6 8 10 0 200 400 600 800 1000

Travel Time (min) Household Density (HH / sq. mi.) of Attraction Zone
BASIC STEP-BY-STEP PROCESS
▪ Trip Generation
▪ Using the socioeconomic & land use data along with the cross-
classification tables, come up with the number of trip
productions & attractions for each zone…
▪ Trip Distribution
▪ Create trip interchange table based upon “friction” between
zones using the travel time matrix, the F-Factor graph, and the
Gravity Model…
Gravity Model

Tij = Pi Aj Fij
ΣAjFij
(Productions)(Attractions)(Friction Factor)
= Sum of the (Attractions x Friction Factors) of the Zones
BASIC STEP-BY-STEP PROCESS
 Mode Choice
Create a separate trip interchange table for each mode based
upon, in this case, the relationship of land use household
density the percentage of people walking & biking…
 Trip Assignment
Distribute the vehicle trips to the street network using “all-or-
nothing” traffic assignment by assuming that all trips are
accommodated on the shortest possible path between
zones…
BASIC STEP-BY-STEP PROCESS
1 Travel Time = 2 2

Travel Time = 6
Travel Time = 5

3 Travel Time = 5 4
BASIC STEP-BY-STEP PROCESS
586 + 97 = 683
1 Travel Time = 2 2
787 + 159 + 40 = 986

Travel Time = 6
Travel Time = 5

97 + 215 = 312
40 + 66 = 106
95
0

32
3 Travel Time = 5 4
131
 QUESTIONS AND DISCUSSION

78
WASSALAM