Section III:

Methodology for Small Cities

There are 73 cities as per 2011 census that have a population of more than 0.5 million. Most of these cities
require large investments in urban transport infrastructures (e.g., BRT, metros) to prevent unchecked growth
of private motorised transport and transform them into smart cities. Some of the decisions on transport
projects they take now would however have long-term consequences and therefore all these cities need to
go for a full CMP, which looks at accessibility for all socio economic groups and genders, studies impact of
transport system on safety, environment and CO2 emissions. The revised toolkit has provided a comprehensive
approach for these cities.

According to 2011 Census, 60% of urban population lives in towns and cities with population less than 0.5
million. Most of these cities have small size, short trip lengths, and high share of walking and will benefit
through improvement in operational effectiveness of para-transit and public transport systems. In the next
5-10 years, no major infrastructure changes in these cities are envisaged. In such a situation, undertaking a
full CMP is not required since it involves a reasonable time (minimum one year) and reasonable budget to
enable data collection, analysis and report write up. Such cities could still make use of CMP toolkit with the
following modifications which reduce the need for extensive modelling.

45
Preparing a Comprehensive Mobility Plan (CMP) – A Toolkit (Revised)

Table 1: Tasks to be followed for Small Cities

Task Modifications for Limited CMP

Task 2 Data Collection:
Table 5 provides an overview of the data needs for studying the existing transport system and
these can be reduced depending on data availability within the cities. Some reduction in the
data might be automatic, for example presence of cycle tracks, signalised crossing and data on
public transport may not be collected if these are not present in the city.
Task 3 Development of BAU Scenario
Refer to Task 3-4 Transport Demand Analysis
The Model Framework (Figure 6) recommends 4 Stage modelling for transport demand. 4 stage
modelling however involves setting up of computer based models and extensive data analysis,
therefore for smaller cities it is optional and they can instead go for the following approach.
They can estimate indicators for existing system (Table 10) based on data collection.
For future years based on stakeholder consultations target modes shares for horizon years can
be decided exogenously
The present trip rates and trip lengths for different modes for the present year can be computed
from a household survey.
A quick estimate of the future travel demand can then be done by using the cross classification
method (See Example Annex 12). Future projection of socio demographic as well as built form
variables can be used to predict the travel behaviour in the stated future year as shown in
Annexure 12.
Local Air Quality
Refer to Task 3-6: CO2 Emissions and Air Quality
Local air quality modelling requires dispersion modelling and this is optional for smaller cities
since they might not have air quality monitoring equipment necessary for model calibration
and also have low level of motorisation.
Task 4 Sustainable Urban Transport Scenarios
Once again, since 4 step modelling is optional, the cities can based on a stakeholder consultation
process decide the target modal shares for this scenario and then estimate the travel demand
from the nature of interventions proposed with regard to built environment.

46
Preparing a Comprehensive Mobility Plan (CMP) – A Toolkit (Revised)

Annexure 5. Four-Step Modelling

Model Framework (Four Step Modelling)

The four-step model approach for CMP needs to account for different social groups and gender (See the Figure
below) and for all modes of transport including NMT, para-transit and public transport and this is slightly different
from the conventional four-step modelling where there is no differentiation in terms of socio-economic groups
and gender, where the focus is mainly on motorised transport. Modeling software’s like QuantumGIS, ArcGIS,
TransCAD, CUBE, VISUM, EMME, OmniTrans, etc. can be used to create the travel demand model of the city.
However these softwares are designed primarily to model motorised modes like cars, two-wheelers and buses.
Visum is the only software among these that has specific modules on environment & emission modelling, and
modelling for NMT (PuTAux) as well as public transport modelling. Hence adequate care should be taken in
specifying the modelling parameters to suit the softwares for cycles. Various stages of the modelling procedure
have been explained in the following sections.

Four-Step Model Framework

94
 Annexures

The base year travel demand model is required to replicate the road network and travel patterns of the city
in modeling software and to test for various short-term measures that can be taken to improve the existing
transportation systems. The following table gives the input parameters and their data sources used for developing
the base year model.

Modeling components and input sources

Model Component Input Source
Traffic Analysis Zone Map Derived from Ward Map
Road Network Derived from Primary Data collected for road inventory &Link speeds and sec-
ondary data on road widths
(It can also be derived from open street maps, if VISUM software is used for
modeling. Open street maps helps in incorporating all the road characteristics).
Trip Production Patterns Household Interview Data and census
Trip Attraction Patterns Land Use Data from Master Plan andBuilding wise usage type from Property Tax
Database
Trip Distribution Trip length distribution patterns from Household Interview data to calibrate the
Gravity Model
Base Year Mode Shares Household Interview Data
Trip Assignment Traffic Volume Counts used for network calibration

Trip Generation
Trip generation involves estimating the total number of trips produced and attracted to each TAZ. Trip production
is dependent on socio-economic characteristics of households within the TAZ while trip attraction depends on
the land-use type of the TAZ as explained below.

Trip production
Household interview data is normally used to estimate the trip production trends for various types of households
using the following steps:
l Purpose-wise trips (eg., work, school/college, social, recreation, etc) produced in each household are
derived as a function of the socio-economic attributes of the household like household size, income and
vehicle ownership.
l Total number of households in each TAZ is derived from the census data or the property tax database and
its total households and number of trips produced are estimated.
l The socio-economic characteristics of each TAZ are derived from the HH Interview data.
l If detailed household level data is not available, TAZ level data and parameters like TAZ population,
employment opportunities etc. are used to derive the productions for each TAZ.

Trip Attraction
The number of trips attracted to each TAZ is estimated in this step. The attractiveness of a zone is a function of

95
Preparing a Comprehensive Mobility Plan (CMP) – A Toolkit (Revised)

the type of land-use of that zone. For example residential land uses produce trips while commercial, institutional
and industrial areas typically attract trips. Hence the existing land use mix is considered as the critical variable in
determining the trips attracted to each TAZ. Land use data at the city level is provided by the Master plan of the
city, but they are only indicative as the land use allocation in the master plan and the actual usage of land use is
observed to be varying widely in practice.

The Property tax data from the municipal corporations maintain building wise land use type and its plinth area.
Types of land use in the buildings include: Residential, Commercial, Educational, Industrial, Public Use, Shops,
Hospital, Cinema/Pub Entertainment, Others. Except residential, all other land use types attract trips. Hence, the
total plinth area of each type of attracting land uses can be calculated and used as a measure of attractiveness
of the TAZ.

Purpose-wise trips attracted to each zone from the household interviews is correlated with land use types in
each TAZ, using multiple linear regression technique to derive the relation between the trips attracted and
the land uses of the TAZ. Based on these equations, the number of trips attracted to each zone is re-calculated
using the equations. This however only gives the number of trips at the scale of the sample size of data, since the
sample trips are used for deriving the equation. Therefore these attractions are used as the relative attractiveness
of each zone. The attractions of each zone are then up scaled proportionally to the total attractions based on
the total trips produced for each purpose.

Trip Distribution
Trip distribution is used to derive the Origin-Destination (OD) matrix from theProduction Attraction (PA) table
prepared in trip generation. Gravity Method is generally adopted for trip distribution. In this method trips
between zone i and zone j (Tij) are distributed in proportion to the number of trips produced in i, number of trips
attracted in j and in the inverse proportion of the impedance between these zones i.e. travel time, travel cost,
relative safety etc.
Tij = Pi [Ai Fij ⁄ ∑Aj Fij]
Where,
Tij = trips produced at I and attracted at j,
Pi = total trip production at i,
Aj = total trip production at j,
Fij = (friction factor) or computed using the TLFD curves
i = origin zone,
j = destination zone

Trip Distribution can be carried out purpose wise or mode-wise based on city specific characteristics. (e.g.,) Trip
length distribution should be observed both purpose wise and mode wise, and whichever parameter has more
clearly defined trip length distributions should be selected for distribution. If the type of mode is affecting trip
length more, mode share split can be carried out before the trip distribution.The following is the step wise
procedure.

96
 Annexures

l The purpose wise peak hour trips are added up to get the total trips produced and attracted to each TAZ.
l The TAZ wise mode-share values can be derived from the HH Interview data and applied to the PA table
to get the mode-wise PA table for all zones.
l Current users: The mode share of public transport and cycles in each TAZ is derived from the household
interview data and is used to derive the PA table for current public transport and cycling trips. The PA table
can be for the peak hour or for the entire day based on the study requirements.
l Potential users: All the trips in the city form the potential public transport and cycle users in the city and it
is important to model these trips in parallel to estimate their potential shift to public transport and cycles
respectively.
l One of the features of the four-stage demand modeling process is that only the inter-zonal trips are
considered for assignment. Hence, the proportion of intra-zonal trips in each TAZ is calculated from the
HH Interview data and these trips are excluded from the demand modeling process.
l The PA table containing inter-zonal public transport trips is used as the input for trip distribution.
For public transport trips, the generalized cost is considered as impedance which is worked out based on time
taken for access, waiting, line haul, transfer, line haul and egress, and dis-utility of each of these in monetary terms.

Mode Choice
Mode choice models should be developed for all modes of transport including public transport and para-transit
modes. As discussed in Task 2-2 TAZ size for modelling thus needs to be small enough to cater to walk, bicycle
trips and account for impact of access/egress trips on public transport.

Mode choice equations

These are computed based on revealed and stated preference of individuals surveyed in the HH survey. A Multi-
nominal logit or Nested logit models or any other logit function were run to achieve the mode choice equations.
As stated mode choice is the dependent variable and socio-demographics of the individual, built form indicators
at the trip’s origin and end and travel cost are the independent variables in the equation.

Mode choice for walk and bicycle

One of the major differences in modelling NMT modes as compared to motorised modes is the impact of speed
on mode choice. Speed of NMT (walk and bicycle) is constant and there is negligible impact of congestion. While
other parameters like distance to be travelled, infrastructure quality, safety and security concerns have wider
impact over mode choice of walk and bicycle. Along with the mode-related parameters individual socio-economic
information needs to be accounted for modelling mode choice for NMT modes of transport.

Mode choice for public transport

Utility of public transport has minimum three inter-related segments i.e. access trip, haul trip and egress trip.
Studies have shown that access/egress trip has a significant impact over public transport as a mode choice. The

97
Preparing a Comprehensive Mobility Plan (CMP) – A Toolkit (Revised)

impact is not only in terms of public transport in vicinity to origin/destination but is also in terms of the discomfort
and disutility associated with the modes used for access/egress trips and mode interchanges. The utility function
for public transport thus involves waiting time and discomfort of changing modes other than mode related
parameters for access/egress trip and haul trip.

Trip Assignment
This step is performed to determine number of trips made by different modes on each of the existing transport
network link during peak and off-peak hour period. Trip assignment for NMT should account to land use and
density parameters in the vicinity of the infrastructure/facility. Trip assignment for bicycle also includes parameters
related to pavement quality, slope, traffic volume and speed. This involves using bicycle compatibility index (BCI)
and other such measures.

The person trip OD matrices for current and potential users are converted to vehicle trips based on the average
occupancy observed in each mode from the occupancy survey carried out in the city. However, the floating
populations coming into the city through the numerous entry points are captured from OD surveys at these
locations. These sample surveys are up scaled to total volume based on the traffic volume counts at those
locations. The OD matrices from these surveys are added to the OD from trip distribution to develop the overall
OD matrix of the city.

The mode-wise calibrated OD matrices derived from the above step are assigned on to the road network using
User-Equilibrium or Capacity Restraint methods based on Wardrops equilibrium49 for motorised modes. For
cyclists All or Nothing (AON) method is used in general by considering the minimum BCI or travel distance
between ODs of the cyclists as the determining factor for route choice. Since most links are assumed to have
enough capacity for cyclists and since cyclists are sensitive to safety and security issues more than the speed,
AON method is adopted.

Network Validation
The link flows observed from trip assignment are compared with the actual traffic flows observed from traffic
volume counts conducted at various locations across the city. If it is observed that the link flows from traffic
assignment vary from the traffic volume counts, the network needs to be re-checked for its accuracy. Some
missing links in the road network are identified through this procedure. However, the larger contributing factor
to this error can be the OD matrix derived from trip distribution. The OD matrix hasto be re-calibrated for it to
match the traffic volume counts. For this, an iterative process is available in modeling softwares called the OD
matrix estimation (TransCAD, CUBE)/ t-flow fuzzy (VISUM). Using this procedure, the network is calibrated to
match the actual volume counts observed on ground.

For details refer to Demand Assessment Module available at

https://www.dropbox.com/sh/99ngmessm2cgb76/IRv2lC9AwZ

49
De Dios Ortuzar, J and L. G. Willumsen (2001). Modelling transport, Wiley.

98