2022 Presentation 2

You might also like

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 54

MODELLING URBAN TRANSPORTATION

(Ref: Orzutar & Willumsen, 2011; Rogers, 2003; Kanafani, 1983)

By: HA Quainoo
Date: September 2022
Contents
• The four-stage Transportation modelling
- Trip Generation models
-Trip Distribution models
- Modal Split Models
- Trip Assignment models
• The Generalised Cost of Travel
• Transportation Demand Analysis
- Elasticity of Demand
- Urban transportation Demand Models
• Transportation Policy & Demand Analysis in Urban
Transportation Planning
Recap: Trip Generation models (Ortuzar & Willumsen, 2011)
• Some Definitions:
(i) Trip / Journey: a one-way movement from a point of
origin to a point of destination
(ii) Home-based (HB) Trip: where the home of the trip-
maker is either the origin or the destination of the journey
(iii) Non-home-based (NHB) Trip: Where neither end of the
trip is the home of the traveller
(iv) Trip Generation: the total number of trips generated by
households in a zone (HB / NHB)
(v) Trip Production: the home end of an HB trip or as the
origin of an NHB trip
(vi) Trip Attraction: the non-home end of an HB trip of the
destination of an NHB trip
(vii) Activity: An endeavour / interest often associated
with a purpose but not necessarily linked to a fixed
location (e.g. shopping / going to cinema at different
locations; work, study, leisure, etc)
Classification of Journeys
• Purpose: In practice, a better understanding of travel / trip
generation models can be obtained if journeys by different
purposes are identified and modelled separately. In the case of
Home-based Trips, for example, different types are:
- Travel to work
- Travel to school / University (education trip)
- Shopping trips
- Social / recreational journeys
- Escort trips (to accompany / collect somebody else)
- Other journeys
(NB: Mandatory/ compulsory & Discretionary/ optional trips)
• Time of Day: the proportion of journeys by different purposes
varies with time of day
- Peak trips
- Off-peak trips
• Person Type: An important classification since
individual travel behaviour depends heavily on socio-
economic attributes. Categories usually used are:
- Income level (low, middle and high income)
- Car ownership (e.g. 0, 1, 2 or more)
- Household size and structure
Factors affecting Trip Generation
• Personal Trip Productions
- Income
- Car Ownership
- Family size
- Household structure
- Value of land
- Residential density
- Accessibility
• Personal Trip Attractions
- Roofed space available for industrial, commercial and other services
- Zonal employment
• Freight Trip Productions & Attractions (usually few trips)
- number of employees
- number of sales
- roofed area of firm
- total area of firm
Trip Generation models
• Linear/ Simple and multiple regression models
Example:
Ŷ = a + bX
Ŷ = a +b1X1 + b2X2 + b3X3 + ….bkXk
Growth-factor modelling
• Generally, if
Ti = Future Trips
ti = Current trips in zone i
Ei = Growth factor
Ti = t i * E i
• Estimation of the growth factor, Ei, is related to variables
such as population (P), income (I) and car-ownership (C) in a
function; e.g.
• Ei = ƒ (Pid , Iid, Cid)/ ƒ (Pic , Iic, Cic)
where subscript c = current year; d = design year
respectively
NB: Problem with growth-factor models: they either
underestimate / overestimate the total number of trips
• Example:
A zone with 250 households with cars and 250
households without a car. Assumptions:
(i)average trip generation rates of each group are as
follows:
- Car-owning households produce 6.0 trips/ day
- Non-car-owning households produce 2.5 trips /day
Estimate the number of future trips if in the future
all households will have a car (though population and
income remain constant)
• Solution:
- Current number of trips per day, ti
= (250)(2.5) + (250)(6.0)
= 2125 trips / day
- Future Trips = ti * Ei
where Ei = ƒ (Cid)/ ƒ (Cic) = 1/0.5 = 2

Hence Ti = 2125 *2 = 4250 trips / day


Method 2:
If it is assumed that the average trip rates will remain constant
(which the main assumption behind most popular forecasting
methods), then the future trips
Ti = 500 *6 = 3000 trips / day (thus, the growth factor
overestimates the total number of trips by almost 42%)
Trip Generations and Trip Attractions
• Assumption:
• Total number of trips originating (the origins
Oi) at all zones = Total number of trips
attracted (the destination Dj)

i.e. ∑ Oi = ∑ Dj (Not always the case)


i j
Trip Generation and Accessibility
• The classical four-stage transportation planning model
incorporates an iterative process between trip distribution
and assignment which leaves trip generation unchanged
• Main disadvantage: changes to the network are assumed to
have no effect on trip productions and attractions
• To counteract this assumption, modellers have incorporated
a measure of accessibility (ease / difficulty of making trips
to/from each zone) into trip generation equations
• Thus, instead of Oin = f (Hin), we have Oin = f (Hin, Ain),
• Where Hin = household characteristics; Ain = a measure of
accessibility by person type
• Generally, accessibility measures take the form:
Ain = ∑ f (Djn , Cij )
j

Where
Djn = a measure of attraction of zone j
Cij = the generalised cost of travel between zones i and
j
OR/
Ain = ∑ Djn exp (-Cij )
j

Where  = a calibration parameter from the gravity model


• Problem
Consider the following trip attraction models estimated using
a standard computing package (t-ratios are given in brackets);
Y = 123.2 + 0.89X1; R2 = 0.900
(5.2) (7.3)
Y = 40.1 + 0.14X2 + 0.61X3 + 0.25X4; R2 =0.925
(40.1) (1.9) (2.4) (1.8)
Y = -1.7 + 2.57X1 - 1.78X4; R2 = 0.996
(-0.6) (9.9) (-9.3)
Where Y = work trips attracted to the zone, X1 = total employment in the zone, X2 =
industrial employment in the zone; X3 = commercial employment in the zone; X4 =
service employment. Select the most appropriate model, explaining clearly why
(consider all its pros and cons)
The Generalised Cost of Travel
• Travel cost can be expressed in terms of distance, time
or money
• Travel impedance:
- Fare
- Travel time
- Tolls
- Fuel levy
- Level of service
- Safety / accident
- Comfort
- Affordability
- Accessibility
• Generalised Travel Cost
TRANSPORTATION DEMAND ANALYSIS (Kanafani, A., 1983)

• Transportation needs: due to socio-economic activities dispersed in


space
• What triggers mobility / travel?
- the quest for food and shelter
- production and consumption of commodities
- work; education; social functions; etc
• Objectives of transportation demand analysis
- To understand the complex relationships between spatial
distribution of socio-economic activities and transportation (an
essential component of transportation planning)
- To obtain a framework for estimating the need for
transportation and forecasting traffic volumes that will use
transportation facilities
- To enable the design of transportation facilities and evaluation
of their economic feasibility
Transportation Planning Process (Kanafani & Sperling, 1982)
SOURCE: TRB (NCHRP), 2012
Measuring Transportation needs (Kanafani, 1983)

• Traditionally measured in terms of traffic volume (flow of vehicles on


the road/ passengers on a flight/ tons of cargo on a truck, etc)
• However, traffic volume alone can be misleading:
- it represents a need influenced by the availability of transportation
services
- traffic volume in a congested transportation facility the true need
for transportation, since it excludes additional traffic that might flow
into the facility it additional capacity were available to carry it
- if additional capacity leads to an increase in traffic volume, then
the potential traffic is greater than the traffic volume originally
observed.
• A single measurement of traffic volume is thus insufficient to express
transportation needs
• Required: specification of different levels of traffic volumes that would
occur at different levels of service
Analogy: economics & demand for transportation
• The concept of demand in economic theory is adopted
to express the need for transportation
• Demand in economics, expressed by series of numbers
(kna. Demand schedule/ function) describe the levels
of consumption of a particular good at various price
levels
• Transportation demand is defined in the same manner
• Mobility (transporting people & goods) consumes time
& energy which translates into cost
• Traffic volumes that would occur at different levels of
cost represent the demand for transport
• Transportation demand is proportional to traffic volumes @
different levels of cost
• The development of specific volumes at specific
transportation facilities represents an interaction of the
demand for transportation with the service characteristics
of these facilities.
• Service characteristics  supply characteristics (hence, the
concept of demand and supply in interaction in
transportation demand analysis
• Thus:
- the demand for transportation results from the spatial
interaction of economic activities
- the flow of traffic at transportation facilities results from
the interaction of demand with the supply source
• Definition:
Transportation Demand Analysis: the process of relating the
demand for transportation to socio-economic activities that
generate it.
• Considerations:
- Type, level and location of human activities
- Demand for movement of people and goods b/n the
different space where the activities occur
-Accessibility
- Travel time and cost
- Safety, etc.
• Results of the analysis:
- Relationships, in the form of models b/n measures of
activity and measures of transportation demand
• Transportation demand, often expressed by a
relationship b/n traffic volumes & transportation cost
characteristics
- Transportation demand models
NB:
Demand analysis is distinct from traffic forecasting
- the main purpose is to achieve an understanding of
the determinants of the demand and of the manner by
which they interact and affect the development of traffic
volumes
- though the demand models provide a major input
into the forecasting process
Elasticity of Transportation Demand
• Elasticity: responsiveness / sensitivity of the demand for
transportation to changes in the supply characteristics
• Assumption:
- The demand function is a relationship between quantity,
xi, unit cost pi, and income I or budget, B
- All other factors are kept constant
-  = any variable (price, income, etc)
Thus, elasticity is a relation b/n differentials and is only valid for
infinitesimally small changes in variables
Then, e =  x/x = lim Δx/x
 / Δv→ 0 Δv/v is the
elasticity (at a point)

Note: the elasticity is defined w.r.t. the variable V.


When  is the cost, then it is referred to as
cost/price elasticity ei = xi/xi
pi/pi
When  is the consumer’s income I, it is called
the income elasticity eI = xi/xi
I/I
• Arc elasticity
- Elasticity values calculated by comparing discrete changes
in variables can be done either empirically / by calculating
such changes from a given demand function
• For observed / calculated changes Δx and Δv, the arc
elasticity is given by:
earc = Δx/x OR earc = Δ ln x
Δv/v Δ ln v

OR
earc = (x1 – x0)(v1+v0)
(x1 + x0)(v1 – v0)
Uses of demand elasticity
• Example:
Let the demand for air trips T between two cities be given by the function of the air
fare, p as T = Kp , where K is a constant, and  is the price (fare) elasticity of
demand
 = -2.0, suggests that the percentage change in traffic is twice the percentage
change in price

- Let the fare be doubled (i.e. increased by 100%).


- The elasticity of -2.0 would imply that this would cause traffic to decrease by 200%,
though a simple calculation shows that traffic will actually decrease by a factor of
4.
- Thus, elasticity values cannot be used to predict the impacts of such large
variations in variables
- The magnitude of changes to be interpreted using elasticity values depend on the
form of the demand function and on the elasticity values themselves
- When a demand function is known, then it can be used to predict the impacts of
changes
• NOTES:
- The use of elasticity value to interpret real changes are mere
approximations, and should be avoided when these changes are
large
- Elasticity as a relation b/n differentials is valid at the points at
which it is defined
- Hence its use to interpret the impacts of changes in demand
variables should be treated as approximations
- Elasticity values cannot be used to predict the impacts of large
variations in variables
- In addition to its use in predicting changes in demands,
elasticity, w.r.t. supply variables, is used to predict changes in the
resources consumed. For example:
- elasticity of trip demand w.r.t. price is useful in predicting changes in
total expenditures
- elasticity w.r.t. travel time is useful in predicting changes in total time
spent on travel
• Example
Suppose the demand for trips T is given by the following function of fare P:
T = 1000 p-2.0

Assuming an initial condition is given by P0 = 10 for which T0 = 10. A


change in fare to p1 = 15 results in traffic decreasing to T1 =4.4

earc = Δx/x = (4.4 – 10)/ (15 – 10) = -1.12


Δv/v 10 10

When the change is reversed so that p0 = 15 and p1 =10 for which T0 =4.4 and T1 = 10,
the elasticity becomes
earc = Δx/x = (10 – 4.4)/ (10 – 15) = -3.82
Δv/v 4.4 4.4
These values are different from the same point elasticity of -2.0. A closer
approximation is obtained when the arc elasticity is defined differently:
earc = Δ ln x = -2.05 which is closer to e= -2.0

Δ ln v

OR
earc = (x1 – x0)(v1+v0) = -1.94
(x1 + x0)(v1 – v0)
MODAL SPLIT MODELS (Ortuzar & Willumsen, 2011; Rogers, 2003)

• Factors affecting (travel) mode choice


• Utility Function (w.r.t. time, cost, etc)
• Modal Choice models:
e.g. The Logit Model
• Modal split & Transportation Planning
- Choice of transport mode is arguably the most important classic stage in
transportation planning
- The key role played by public transport in policy making
- Generally, travelling in public transport modes uses road space more
efficiently; produces fewer accidents and emissions than using private
cars
- Underground & other rail-based modes do not require additional road
space and therefore do not contribute to road congestion
- If some drivers could be persuaded to use public transport instead of
cars the rest of the car users would benefit from improved levels of service
- Thus, the issue of mode choice is probably the single most important
element in transport planning and policy making.
- It affects the general efficiency with which we can travel in urban areas,
the amount of urban space devoted to transport functions
• Factors affecting (travel) mode choice
Utility Function (w.r.t. time, cost, etc)

• What factors would make a traveller switch from plane


to train or road transport (public transport / private
car)?
• Modal split, therefore determines the proportion /
percentage of travellers who would use particular
modes of transport
• Models are based on Utility/ Disutility functions
• Utility depicts the level of satisfaction / dissatisfaction
(disutility) with a given mode of transport (e.g. journey
time, cost/fare)
• Probability that a given mode would be chosen is then
calculated once the utility function is determined.
• The function used to estimate the total utility provided
by a given mode is
• Generally, the utility function may be
expanded to take the form:
• Uk = ak + a1X1 +a2X2 + …….+anXn + Ꜫ0
Where:

Uk = Utility function for travel mode k


ak = Modal constant
Xn = modal attributes/ characteristics
variable (e.g. travel time, cost of travel)
an = coefficient associated with each
attribute
Ꜫ0 = error term
Trip Modal Split Models

Empirical Models:
(i) Diversion Curves
- These first models included only one/two attributes
/characteristics of the journey (generally in-vehicle travel time.
- A plot of the proportion of trips by mode I (TIij/Tij) against the
cost or time difference produces an S-curve
- These were empirical curves estimate what proportion of
travellers would be diverted to use a (longer but faster)
bypass route; hence the name diversion curves.
Empirical Modal-Split curve
(ii) Kirchoff’s Model:
- The proportion of trip makers between origin i and
destination j that chooses mode k as a function of the
respective generalised costs by mode k, Ckij is given by:

Where n = a parameter to be calibrated / transferred from


another location / time (usually b/n 4 and 9)
- Kirchoff’s model sometimes considered attractive
because the choice of mode depends on the ratio of costs
(to a power) and not their difference
• The empirical models have limited theoretical basis and
therefore their interpretation and forecasting ability are
doubtful

The Logit Model (synthetic model with strong theoretical


base)
- As part of travel demand forecasting, modal split analysis
seeks to investigate travellers’ decision in relation to mode
of travel
-Modes are preferable if they are relatively faster and
cheaper /safer (i.e. have competitive edge over other
modes (i.e. if the utility is higher than other modes of
travel)
• The probability of a commuter choosing mode i (train,
bus, car) is represented by the following multimodal
Logit (Choice) Model:

Pi = e U(i) / ∑n e U(r)
r=1

Where: U(i) = utility of mode i


U (r) = utility of mode r
n = number of modes available
For only two modes the Logit model (aka. Binary
Logit Model) takes the form of

Pi = 1 / (1 + e U2 – U1)
Where only a Bus and Car are considered

e.g. P (B) = 1 / (1+ e C – B)

P (C) = 1 / (1+ e B – C)
Example 1
Use a logit model to determine the probabilities of a group of 5000
commuters choosing between three modes of travel during the
morning peak hour: Private car, Bus and Light Rail. The utility functions
for the three modes are estimated using the following equations:
Uc = 2.4 – 0.2C – 0.03T
UB = 0.0 – 0.2C – 0.03T
ULR = 0.4 – 0.2C – 0.03T
Where:
C = cost (R); T = travel time (minutes)
For all workers:
The cost of driving is R4.00 with a travel time of 20 minutes
The bus fare is R0.50 with a travel time of 40 minutes
The rail fare is R0.80 with a travel time of 25 minutes.
Determine the number of workers using each mode of transport.
• Substitution of costs and travel times into the given
utility functions yield
• Example 2
The utility functions for the three modes are estimated
using the following equations:
Uc = 2.4 – 0.2C – 0.03T
UB = 0.0 – 0.2C – 0.03T
The local municipality constructs a bus lane in order to
alter the modal split in favour of bus usage. When in
operation, the bus lane will reduce the bus journey time
to 20 minutes but will increase the car travel time to 30
minutes. The cost of travel on both modes remains
unchanged.
Calculate the modal distributions for 1000 work
commuters using the route both before and after the
construction of the proposed new bus facility.
Solution:
• The baseline utilities for the two modes as in the first
example are:

You might also like