7

Traffic Flow and Safety: Need for New

ModelsforHeterogeneousTraffic
Geetam Tiwari

INTRODUCTION

Knowledge of fundamental traffic flow characteristics—speed, volume and density—

and associated analytical techniques are essential requirements in planning, design,
and operation of transportation systems. The fundamental characteristics have been
studied at microscopic and macroscopic levels. Existing traffic flow models are based
on time headway, flow, time–space projectory, speed, distance headway and density.
These have lead to the development of a range of analytical techniques: demand–
supply analysis,capacity and level of service analysis, traffic stream modelling, shock
wave analysis, queuing analysis and simulation modelling. The focus of this research
has been homogenous traffic conditions—traffic which consists primarily of cars or
motorized vehicles of similar characteristics. We are now faced with a situation where
more than half of the world’s population is living in megacities in the developing
region. Large differences in income levels and social disparities have lead to the
development of ‘cities within cities’. Each level of the city, with its own level of
technology and land-use patterns, exists in close geographical proximity with other
cities of different patterns. This is reflected in the travel and traffic patterns existing in
this region. The same road space gets used by cars and buses, along with locally
developed vehicles for public transport such as three-wheeled scooter taxis, scooters
and motorcycles, bicycles, tricycle rickshaws, animal and human drawn carts.
Infrastructure which is designed on the basis of homogeneous traffic models has failed
to fulfill the mobility and safety needs of this traffic. Despite low levels of car ownership,
accidents and congestion continue to plague this region. This is the right time to
question the applicability of models which have been developed for a very different
situation.

Traffic Flow and Safety

Two basic parameters of traffic flow which influence the safety of road users are traffic
volume or flow measured in terms of number of vehicles per unit time and traffic
speed. The amount of traffic, often called traffic flow or more general exposure, is
often treated as a matter of routine. Relationship between accidents and flow is,
therefore, often assumed to be a simple (linear) relationship. One major reason could
72/ Injury Prevention and Control

be that it seems obvious that an increase in flow leads to an increase

in the number of accidents. If we have a system without road users
it is obvious that it cannot produce any traffic accidents. And if the
number of road users increases this will then obviously lead to
increased probability for an accident. Most modelling work has used
flow as one major explanatory variable. The most used model structure
is the exponential (Ekman, 1996).
E(A)= k* exposure y (Kulmala, 1995).
This approach fulfils some basic ideas about traffic and traffic
safety: it starts at the origin, zero flow creates no accident. Furthermore,
it is monotonically increasing with increasing flow, as long as y is
positive. However, this assumption is not completely supported either
by basic traffic theory or modern behaviour adaptation theories.
The results from estimations based on empirical data is often that y is
slightly less than 1. This means that accidents increase by increasing
exposure less than a linear approach would have given. This makes
intuitive sense also considering the relationship between flow, speed
and fatalities.

Flow, speed and fatalities

The standard traffic flow model assumes a parabolic curve between
speed and flow as shown in Figure 1.

depending on
the degree of
heterogeniety Figure 1Speed flow relationship (May, 1990)
in the traffic
mix, fatalities As number of vehicles increase on a network, speed remains constant
and injuries as long as the flow is below a certain value. This condition is defined
may decrease as level of service A. Speed reduces gradually until the flow reaches
even if the capacity. Consider this relationship with the relationship of speed
and fatalities and injuries shown in Figure 2. The estimates for
exposure of probability of pedestrian deaths at different impact velocities are: 5–
vulnerable 8 per cent at 30 kph, 25 per cent at 40 kph, 45–80 per cent at 50 kph,
road users and more than 85 per cent at 60 kph (European Transport Safety
increases Council, 1995; University of Zurich and Swiss Federal Institute of
 Tiwari /73

Technology, 1986.) For car occupants in crashes at 80 kph the likelihood of death is 20
times more than at 32 kph (IIHS, 1987). Clearly, increase in speed is associated with
disproportionate increase in number of fatalities. Also, the safe speed for car occupants is
much higher than for the pedestrian and bicyclists.
Speed influences energy consumption, pollution, noise, vehicle and road maintenance
costs, stress on road users and safety. In general, higher speeds have an adverse influence
on all these factors. The safety of road users is influenced both by the absolute speed of
vehicles and by the variation in speeds among vehicles on the road (Noguchi, 19**).
Other factors remaining constant, higher speeds increase the probability of a crash taking
place and the severity of injury in a crash, whereas a greater variations in speeds of
vehicles only increases the probability of the event. As illustrated in Figure 2, small
reductions in travelling speed result in large reductions in injuries and fatalities in both
urban and rural areas. This is because the stopping distance of a vehicle under braking is
proportional to the square of the original velocity and the damage to human beings is
related to the square of the impact velocity. Lower initial speeds means that the driver
has better control on the vehicle and the vehicle can stop much earlier and reduce the
probability of a crash. In the event of a crash the injuries are less severe at lower impact
velocities.
Probability of pedestrian fatality

Impact of speed km
Figure 2 Speed and pedestrian fatalities

The relationship between flow, speed and fatalities requires further consideration in
the case of mixed traffic and vulnerable road users. A heterogenous traffic mix has an
effect on traffic safety, specially traffic fatalities. Figure 3 shows the distribution of the
percentage of non-motorized vehicles (NMV) fatalities versus the percentage of motorized
vehicles (MV) trips comprising the location’s modal split (Fazio and Tiwari, 1995).
Theoretically, no NMV fatalities can result from a striking MV at the origin on the graph
because no MVs exist in the traffic stream at this point. When MVs account for 100 per
cent of the trips, no NMV fatalities occur because of the absence of NMVs in the traffic.
Presence of NMVs also has a calming effect on traffic speed. Data from Delhi also show
that as NMV flow increases, the average speed difference between MVs and NMVs
decrease specially on roads where mixing of MV and NMV takes place (high conflict
between MV and NMV). As the speed difference or initial speed reduces, number of
fatalities and injuries reduce. Therefore, depending on the degree of heterogeniety in
the traffic mix, fatalities and injuries may decrease even if the exposure of vulnerable
road users increases.
Models which predict number of fatalities, injuries and accidents based on a linear
relationship with motorization or flow are inappropriate if they do not include speed
implications. Rate of fatalities would depend on how the increased flow affects mean
speed of the traffic stream as well variation of speed of the traffic stream. The modal
74/ Injury Prevention and Control

diversity present in mixed traffic (primarily in less motorized

countries) makes the effect of speed even more critical because
the vulnerable road users (VRUs)— people outside cars and buses
constitute majority of the victims.

Figure 3 Degree of traffic homogeneity

Heterogeneous traffic Flow: present and future

Nearly 60 per cent of world population lives in developing r egions—

Africa, Asia and South America (Figure 4).

Western Europe
Southern Europe
Northern Europe
Eastern Europe
models which Europe

predict Western Asia

South-central Asia
number of Southeastern Asia

fatalities,
Eastern Asia
Regions

Asia

injuries and South America

accidents
Northern America
Central America

based on a Caribbean
American

linear Western Africa

relationship Southern Africa

Northern Africa

with Middle Africa

Eastern Africa

motorization 0 5
Africa
10 15 20 25 30 35 40 45 50

or flow are Percent

inappropriate Figure 4 Proportion of urban population in 1990 by region

if they do not These regions are characterized by dominance of large cities—more
include speed than 60 per cent of the urban population resides in million- plus
implications cities. Of the 100 largest cities 62 are in this region . A large
proportion of the population residing in these cities lives below
 Tiwari /75

the pover ty line, 29–60 per cent (Figure 5). Therefore, the demand for non-motorized
modes and pedestrians on highways and urban areas is inevitable.
Venezuela
Mexico
Guatemala
El Salvador
Colombia
Brazil
Latin America

Sri Lanka
Pakistan
Country

Nepal
Indonesia
India
Bangladesh
Asia (including China)

Zambia
Swaziland
Morocco
Gambia
Cote d’lvoire
Africa

0 10 20 30 40 50 60 70
Percent
Figure 5 Population below poverty line

Urban areas in developing countries experience such extremes of wealth and poverty
that they can be characterized as having dual economies. One serving the needs of the
affluent and featuring modern technologies, formal markets, and outward appearance
of developed countries. The other serves disadvantaged groups and is marked by
traditional technologies, informal markets and moderate to severe levels of economic
and political deprivation (Dimitrou, 1990).
Urban poverty, characterized by unemployment, dependence on the informal sector,
low wages and insecure jobs, has a direct bearing on travel and transport demand of a
large segment of the population residing in urban areas. Their dependence on transport
which enables them access to job markets becomes essential for survival. This need is
more critical for them than for those with high income and secure jobs. However, this
segment of population is also transport poor. Even a subsidized public transport remains
cost prohibitive for them.
Cities in developing countries are characterized by heterogenous traffic (mix of
non-motorized and motorized modes) and mixed land-use patterns. Non-motorized
vehicles are owned and used by a large section of the population (Figure 6).
Note: Animal carts & buses are negligible

Tokyo
George Town
Chiang Mai
Metro Manila
Surabaya
Shanghai
Kanpur
Dhaka
Hanoi
Phnom Penh
0 200 400 600 800 1000 1200
Number of vehicles per 1,000 population

Bicycle Cycle-Rickshaw Motorcycle

Other motor vehicle Bus Animal Cart

Figure 6 Comparison of vehicle composition

76/ Injury Prevention and Control

Car ownership rates in Asian countries are low compared to

North America and OECD countries. In 1993, it was 29 cars
per 1000 residents in East Asian countries compared to 561
cars per 1000 residents in North America, 366 in OECD
countries (AAMA, 1995) . Although the greatest growth rate in
the number of motor vehicles is expected in many Asian
countries, most of these increases in absolute numbers of vehicle
fleets will result from increases in the numbers of motorized
two-wheelers (MTWs) and three-wheelers (World Resources
Institute, 1996) . In Thailand, Malaysia, Indonesia and Taiwan,
two- and three-wheelers account for more than 50 per cent of all
motor vehicles.
Non-motorized transport (NMT) constitutes a significant share
of the total traffic in many Asian cities. Shanghai, Hanoi, Kanpur
and Tokyo all have a relatively high rate of bicycle ownership and
a high proportion of bicycle traffic (Figure 7). In Indian cities, the
share of NMT at peak hour varies fr om 30–70 per cent. The
proportion of trips undertaken by bicycles range between 15 and
35 per cent, the share tending to be higher in medium and small
sized cities. The patterns of NMT use changes with growth in city
size. In most NMT-dependent low-income cities, bicycles are used
for entire trips (e.g., commuting, shopping). In Kenya, despite
30 per cent several constraints, the NMT including walking are still the
of the world prevalent modes that provide more than 45 per cent of all the
population personal transport in urban centres. In a high-income city like
living in urban Tokyo, bicycles are increasingly used as a feeder mode to rail
poverty in stations as well as for shopping and other purposes (World Bank,
cities of 1995). Every motorized public transport trip involves access trips
by NMT at each end. Thus, NMT including walking continues to
developing play a very important role in meeting the travel demand in cities
countries is in developing countr ies.
transport Note: Animal carts & buses are negligible
poor … A Tokyo
sustainable George Town
transport Chiang Mai
system must Metro Manila
meet the Surabaya
demand of Shanghai
this captive Kanpur
ridership of Dhaka
non- Hanoi
motorized Phnom Penh
transport 0 20 40 60 80 100

existing in the Percent of vehicles

cities of the Bicycle Cycle-Rickshaw Motorcycle

South. Other motor vehicle Bus Animal Cart

Figure 7 Comparison of vehicle composition among cities

 Tiwari /77

Table 1 gives selected indicators for a few Indian cities. Regardless of city size, it
sho ws that nearly 40–60 per cent of households ha ve monthly incomes of
approximately US $50–60. In large cities like Mumbai, Delhi, Chennai, more than
60 per cent of people are employed in the informal sector. F or this population walking
and bicycling to work is the only mode of transport available. Assuming a minimum
of 4 trips per household per day at the cost of Rs 2 (US$ 0.05) per trip by public
transport, a household would need to spend Rs.320 (US$ 8) per month on transport.
For low-income people living in the outskirts of the city, the cost per trip may be two
or three times this amount depending on the number of transfers. On an average,
low-income households canno t spend more than 10 per cent of its income on
transpor t. This implies that a household’s income must be at least Rs.3200 (US$ 80)
to be able to use the public transport system at minimum rates. According to a
survey (ORG, 1994), approximately 28 per cent of households in Delhi ha ve a
monthly household income of less than Rs.2000(US$ 50).

Table 1
Indian city indicators

INDICATOR MUMBAI DELHI MADRAS BANGALORE LUCKNOW V ARANSI HUBLI / MYSORE GULBARGA TUMKUR
DHARWAD
Population(million)
10.26 8.96 5.65 4.47 1.8 1.08 0.68 0.7 0.33 0.19
Household Income Distribution (Quintile Boundaries US $)

I (poorest 20%)
374 290 347 385 291 268 284 373 258 287
V (richest 20%)
2497 3292 2781 2487 2181 2084 2009 2372 1951 1761
Informal Employment(%)
68 66 60 32 48 49 31 31 27 63
Motorized Vehicles(per 1000 pop.)
51 205 102 130 130 85 49 123 60 63
Data from: Society for Development Studies, Delhi

Sur vey results show that nearly 60 per cent of respondents find the minimum
cost of work trips by public transport (less than Rs.2 per trip) unacceptable (CRRI,
1988). Even at minimum costs, public transport trips account for 20 to 30 per cent
of family income for nearly 50 per cent of people living in unauthorized settlements.
This segment is very sensitive to the slightest variation in the cost of public transport
trips.
The data above show that an estimated 30 per cent of the world population living in
urban poverty in cities of developing countries is transport poor. In this segment it is
harder for individuals and households to save and build up assets, and reduce their
vulnerability to sudden changes/loss in income. Low incomes also make it difficult for
households to ‘invest’ in social assets such as education that can help reduce their
vulnerability in the future. Therefore, access to affordable transport is necessary for
survival. A sustainable transport system must meet the demand of this captive ridership
of non-motorized transport existing in the cities of the South.
78/ Injury Prevention and Control

URBAN TRANSPORT AND LAND-USE PATTERNS

Heterogenous traffic is likely to continue in less motorized

countries (LMCs) because of the socio-economic realities and
compulsions. Many cities in LMCs experience such extremes of
wealth and poverty that they can be characterized as having dual
economies—one for the affluent and the other for the
disadvantaged. Very often formal plans address the needs of one
set of city residents, failing to recognize the existence of the other
set who very often form the majority.
The planning framework as adopted in the preparation of
master plans in many Asian cities has been completely divorced
from resource assessment. The process also does not invoke any
procedures for involving community and bringing about
consensus on contentious issues. The net effect of the inadequacies
of the planning process has been that majority of urban growth
has long taken place outside the formal planning tools. Informal
residential and business premises and developments increasingly
this traffic is dominate new urban areas. Even in our megacities where half or
characterized more of a city’s population and many of its economic activities
are located in illegal or informal settlements, urban planners still
by lack of any rely on traditional master-planning approaches with their role
effective restricted to servicing the minority, high income residents. Few
channelization, weak attempts have been made to bring some coordination of
mode development and ‘services’ to informal areas through slum
segregation or improvement schemes.
control of Unlike traffic in cities in high income countries (HICs),
speeds. To bicycles, pedestrians and other non-motorized modes are present
in significant numbers on arterial roads and intercity highways.
the formally Their presence persists despite the fact that engineers designed
trained these highway facilities for fast moving uninterrupted flow of
planner it motorized vehicle.
looks like Increase in the level of congestion has been a major concern for
chaos moving planners and policy makers in metropolitan cities. In Delhi average
towards total speeds during peak hour range from 10–15 kph in central areas
gridlock. Yet and 25–40 kph on arterial streets (CRRI, 1992). Compared to Delhi,
average speeds in other mega cities are less. In 1993, Delhi’s traffic
people and fatalities were more than double that of all other major Indian cities
goods keep combined (Indian Express, 1994). Clearly, criteria for recommending
getting optimal speeds and congestion reduction does not include desired
through. And level of safety, pollution and land-use patterns.
may, by some There is ample evidence to illustrate the mismatch between
measures the careful planning and growing transportation problems. Unless
actually be we understand the basic nature of problems faced by our mega
doing better cities, the adverse impact of growing mobility on the environment
and safety would continue to multiply in future.
than in some The existence of an active informal sector introduces a high
controlled degree of heterogeneity in the socio-economic and land-use system.
conditions This is assumed to add to our problems of congestion and
 Tiwari /79

pollution. However, the informal sector is an integral part of the urban landscape
providing a variety of services at low costs, at locations with high demand for these
services. Many view hawkers, pavement shops, cycle and motor vehicle repair and
spareparts shops as unauthorized developments along the road that reduce the capacity
of the planned network. However, since the market demands these services, they
continue to exist and grow along arterial roads as well. It is quite clear that long term
land-use transport plans must address the needs of the informal sector.

Understanding Heterogeneous traffic flow in developing country cities

Most developing country cities have been classified as ‘low cost strategy’ cities
(Thomson, 19 77). In comparison with cities in the West, these cities consume less
transport energy. High densities, intensely mixed land-use, short trip distances, and
high share of walking and non-motorized transport characterize these urban centres
(Newman and Kenworthy,1989) .
Heterogenous traffic flow consists of modes of varying dynamic and static
characteristics sharing the same road space. Underlying concepts of traffic flow theory
in the US, Europe, and Australia are formed by motorized four-wheel road way traffic
dominating in those areas, i.e., homogenous traffic. All car following, lane changing
logic and system’s measure of effectiveness used in microscopic simulation programmes
ultimately use field data from these countries for caliberation.
In LMC cities, the road network is used by at least seven categories of motorized
and non-motorized vehicles. Vehicles ranging in width from 60 to 2.6 m, and capable
of maximum speeds ranging from 15 to 100 kph, share the same road space. All these
vehicles which have varied dynamic and static characteristics share the same carriageway.
This traffic is characterized by lack of any effective channelization, mode segregation
or control of speeds. To the formally trained planner it looks like chaos moving towards
total gridlock. Yet people and goods keep getting through. And may, by some measures
actually be doing better than in some controlled conditions.
In Delhi different traffic modes are not segregated and there is minimal enforcement
of speed limits. In this situation flow patterns result in a natural optimization of road
use due to self organization by road users. Though aggregate conflict data do not correlate
with fatalities, our data show that Delhi has high number of VRU fatalities. Therefore,
segregation and traffic calming techniques developed for Delhi conditions with special
reference to motorized two-wheelers are desirable. Techniques developed in HMCs do
not address the high volume of motorized two-wheelers and large variations in traffic
composition from site to site.
The peak hour motor vehicle flows in traffic with mix of MVs—cars, buses, two-
wheeler scooters, three-wheeler scooter— have been observed to be very high compared
to homogenous traffic sites of similar street width (Fazio et al., 1998). Having vehicles
of narrow widths in the traffic stream greatly increases the capacity of streets. Narrow
vehicles fill-in the lateral and longitudinal gaps between wide vehicles; heterogenous
traffic uses on-street space more efficiently than homogenous traffic (Figures 8, 9 and
10). Homogeneous traffic flow is modeled on the basis of lane discipline logic. An
ideal lane capacity is estimated of only passenger cars by using passenger car units or
equivalents.
For heterogenous traffic, having an ideal capacity by lane is mis-conceptual because
lane discipline is very loose. Vehicles have varying static and dynamic characteristics.
80/ Injury Prevention and Control

These share the same road space and move by sharing the lateral
as well as the linear gaps. For example, a motorcycle rider judges
whether the lateral distance (width) between a motor scooter and
bus is acceptable to progress on the roadway. Another motorcycle
rider in the same situation would have a different critical width
acceptance. If the width is unacceptable, then an entity is
constrained by preceding entities. Critical width acceptance
depends on three items. First, the travel speed of the vehicle/
entity itself. The next item is the physical width of the vehicle.
Distribution of the width acceptances of specific entity groups is
the third item, i.e., driver/rider/pedestrian behaviour. Each
vehicle/entity group has its own critical width acceptance.
Heterogeneous traffic can have many motorized two-wheelers,
motorized three- wheelers, bicycles, non-motorized three-wheelers,
cars, buses, trucks, animal-drawn carts, and human-powered push
and pull carts. Additionally, if sidewalk facilities are inadequate
or lacking, this diverse mixture contains significant on-road
pedestrian traffic. In homogeneous traffic, traffic entities form
one-dimensional queues (Figure 8); in heterogeneous traffic, mass
queues develop. These queues grow lengthwise as well as laterally.

Lane
discipline is
deficient in Homogeneous traffic has one dimensional queues Heterogeneoustraffichastwodimensionalormassqueues
heterogenous Figure 8Queuing theory
traffic not
because The ‘car following’ notion used in homogeneous traffic flow
driver models is not applicable in heterogenous traffic (Figure 9). Since
behaviour is cars do not comprise most of the traffic mixture, ‘car following’ is
significantly an incorrect term for heterogenous traffic. Secondly, since width
of entities vary greatly in heterogenous traffic, figuring out which
different, but leading entity/vehicle it is following is difficult. Leading entities
because may run parallel or staggered.
heterogeneous
traffic
consists of
entities of Homogeneous traffic has lane discipline
various
widths and
varying
dynamic Heterogeneous traffic has parallel/staggered entity-following
characteristics Figure 9 Car following
 Tiwari /81

Professionals have extensively derived models and algorithms from the ‘lane
changing’ notion of homogeneous traffic (Figure 10). Microscopic studies of this traffic
shows that the time headway between vehicles is an important flow characteristic that
affects safety, level of service, driver behaviour and capacity of a transportation system.
A minimum time headway must always be present to provide safety in the event that the
lead vehicle suddenly decelerates. The percentage of time that the following vehicle
must follow the vehicle ahead is one indication of level or quality of service. The
distribution of time headways determines the requirement and the oppurtunity for
passing, merging, and crossing. The capacity of the system is governed primarily by the
minimum time headway and the time headway distribution under capacity-flow
conditions.

Homogeneous traffic uses lane concept

Heterogeneous traffic uses width acceptance/entity envelope

Figure 10Lane changing

Clearly, underlying these concepts is the notion of lane discipline or lack of it.
Lane discipline is deficient in heterogenous traffic not because driver behaviour is
significantly different, but because heterogeneous traffic consists of entities of various
widths and varying dynamic characteristics. With homogeneous traffic, the width range
is approximately 2.1 m for cars to 2.6 m for trucks and buses. Homogeneous traffic
drivers find it optimal and advantageous to adopt lane discipline to traverse the roadway
space given the narrowness of the width range. For heterogeneous traffic, the width
range is approximately 6 m for pedestrians to 4.9 m for overburdened truck trailers.
Drivers, pedestrians, riders and animals find it optimal to advance by accepting lateral
gaps (widths) between preceding entities. Heterogeneous traffic uses road space more
efficiently than homogenous traffic. For this traffic, models based on width acceptance
can ultimately produce a good estimate of roadway capacity and assessments of
operations and safety of various facility designs.
This specific characteristic of heterogeneous traffic highlights the limitations of
conflict technique which has been developed for homogeneous traffic (Tiwari G e t
al.,1998). There are very few detailed studies on traffic patterns and their influence on
accidents in LMCs. We use here a study conducted in Delhi which involved conflict
analysis for prediction of fatal crash locations. Peak hour traffic at fourteen selected
locations were videotaped. Trained observers recorded traffic compositions at mid-
block, average space, mean speeds by mode and conflicts by type, reactor mode and
cause mode. The study showed a weak crash-conflict association. Definitions of conflict
developed in HMCs where motorized vehicles are the dominant modes are inadequate
in heterogeneous traffic situations like Delhi. In heterogeneous traffic streams, bicyclists,
82/ Injury Prevention and Control

motorized two-wheelers and motorized three-wheelers have the

freedom of moving laterally as well. These manoeuvers at low
speeds need not necessarily lead to fatal crashes. Kouabenan (1996)
and Salimen et al. (1992) have also reported that fatal accidents
generally differ in causation from non-fatal injuries. The 3.5 metre
wide lanes designed to fit four-wheeled motorized vehicles, such
as cars, trucks, and buses can fit at least two lanes of two wheelers.
These space sharing manoeuvers of heterogeneous traffic are also
recorded as conflicts. These manoeuvers at low speeds need not
necessarily lead to fatal crashes. For mixed traffic, we need to
develop better understanding of conflicts to differentiate between
conflicts that are a part of normal driving cycle and the ones that
are potential crashes.
These issues convince us that LMCs are experiencing a new
phenomenon in road traffic patterns and accidents for which there
is little precedence. These patterns are new and they need to be
understood through careful scientific research.

Safety in Heterogeneous traffic

The proportion of road users killed in various modes of transport

as a percentage of all fatalities in different countries is shown in
Figure 11. The data show that non-car occupants constitute almost
75 per cent or more of all fatalities in most developing countries.
This flows logically from the fact that this class constitutes the
majority of road users however, roads and other transport
infrastructure designs address the need of motorized vehicles only.
In addition, because VRUs are not protected by metallic or energy
absorbing materials, they sustain relatively more serious injuries
even at low velocity crashes.

U.S.A. (1995)

In many Australia (1990)

LMCs Norway (1990)

highways run The Netherlands (1990)

through rural Japan (1992)

areas with Malaysia (1994)

high density Colombo, Sri Lanka (1991)

populations Bandung, Indonesia (1990)

where most Thailand (1987)

people do
not have
Delhi, India (1994)

access to
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%

Pedestrians Bicyclists Motorised Motorised four wheelers others

motor Figure 11 Proportion of road users killed in different countries

vehicles.
 Tiwari /83

It is possible that wherever the proportion of VRUs is high as a proportion of all

road users similar crash rates will be experienced as in India. However, some less
motorized countries (LMCs) do not have bicycle use rates as high as those of countries
like India, China and Vietnam, and these countries have lower involvement of bicycle
fatalities (Tiwari and Saraf, 1997). In these countries where bicycle use rates are lower,
it appears that MTW and pedestrian fatality rates are proportionately higher. Data
show that in most developing countries, the urban poor who are heavily dependent
on NMVs, are the victims of road traffic crashes.
Almost 80 per cent of all cars are owned by 15 per cent of the world’s population
residing in North America, western Europe and Japan. On the other hand, in LMCs like
India and China, less than one in a hundred families owns a car. Car ownership levels in
LMCs are so low that even at a reasonable economic growth rate (say 7–10 per cent per
year) for the next few decades, most families in LMCs are not likely to own a car in the next
quarter of a century. A comparison of per capita levels of vehicle registration and proportions
of cars and motorcycles in a few countries is given in Table 2. These ownership patterns
show that total vehicles registered per capita, and car and motorcycle ownership levels
differ greatly in HMCs and LMCs. These different ownership levels influence traffic
behaviour patterns and composition of traffic fatalities.

Table 2
Vehicle Registration and Proportions

COUNTRY GNP TOTAL NUMBER MTW S AS % OF CARS AS %

PER CAPITA OF VEHICLES TOTAL VEHICLES TOTAL VEHICLES

US $ /1,000 PERSONS
Japan 34 630 640 20 58
USA 24 780 740 2 88
Germany 23 980 570 9 89
France 23 420 520 10 87
UK 18 340 410 3 86
Australia 18 000 610 3 76
Republic of Korea 8 260 206 24 33
Malaysia 3 140 340 56 34
Thailand 2 410 190 66 16
Philippines 950 32 26 28
Indonesia 810 58 69 15
Sri Lanka 600 50 60 13
China 530 21 40 24
India 320 30 67 14
Vietnam 210 27 91 9

These figures suggest two different phenomena that are relevant to road safety policies.
It appears that total vehicle registration levels remain below 100 per thousand persons in
countries that have per capita incomes of less than US $ 1,000 and that motorcycle
registrations decrease belo w 20 per cent of the total vehicle fleet only when per capita
incomes are much greater than US $ 8,000. The only exceptions are countries like China
84/ Injury Prevention and Control

where MTW production or availability has been controlled by

government policy. Even at per capita income levels of US $ 3,000
car ownership levels remain low and the proportion of MTWs can
be more than 50 per cent. Most LMCs, including India and
China, will not reach per capita income levels of US $ 3,000
in the next decade. As incomes increase, the poorest people in
countries like India and China will be able to own bicycles and
those who own bicycles today may opt to buy motorcycles when
they become richer. As the number of poor and lower middle
class people in these countries is larger than that belonging to
the upper class, we are likely to witness greater increases in
absolute numbers of bicycles and motorcycles than cars in the
next decade or so. Road safety policies and countermeasures
that are based on societies where car fleets constitute about 80
per cent of the vehicle fleet will not be adequate for most LMCs
where MT Ws comprise more than 40–50 per cent of the total
number of vehicles.
The prevalent high rates of pedestrian, bicycle and
motorcycle traffic in LMCs (proportions do differ from country
to country) result in VRU fatalities constituting 60–80 per cent
of all traffic fatalities (Mohan, 1992). These patterns of traffic
future traffic and accident in LMCs are not only different from those that
models must are prevalent in HMCs today, but are also different from the
account for experiences of HMCs in the past. The HMCs have never
the users of experienced road traffic that includes such a high proportion
different of motorcycles, buses and trucks sharing the same road space
transport with pedestrians and bicyclists. In addition, in the earlier part
modes having of this century when the present HMCs had low per capita
incomes, motor vehicles (including motorcycles) were relatively
conflicting more expensive and not capable of high velocities and
requirements… accelerations. Therefore, speeds were lower and number of
Highway vehicles using roads was less than that seen today. In a sense,
planning motor-vehicle technology, roadway quality and social systems
standards were more compatible. On the other hand, LMCs now have to
provide for plan for use of technologically advanced vehicles using relatively
services ‘less advanced’ roadways and enforcement systems.
needed by Because bicyclists and pedestrians continue to share road
space in the absence of infrastructure specifically designed for
motorized NMVs, they are exposed to higher risks of being involved in a
vehicle road traffic accident by sharing the road space with high speed
users.However, modes. Unlike cities in the West, pedestrians, bicyclists and
there are no MT Ws constitute 75 per cent of to tal fatalities in r oad tr affic
standards for cr ashes. Buses and trucks ar e involved in more than 60 per
providing cent of fatal crashes. Buses are often very crowded inside and
services significant proportion of passengers who die are those who
fall from footboards of buses. In addition, many indigenously
needed by designed vehicles (IDVs) such as tempos, jugar are present on
NMT roads of Indian cities because of the absence of efficient and
comfortable public transport services. These IDVs operate as
 Tiwari /85

paratransit modes thus serving a useful role in the context of existing social system
(Tiwari, 1994).
In HMCs a very large proportion of the population owns motorized vehicles.
In addition, these countries can afford to have roads parallel to expressways to be
used by local traffic and vehicles not allowed on expressways. In many LMCs
highways run through rural areas with high density populations where most people
do not have access to motor vehicles. Also, many expressways in LMCs do not
have parallel road links for slow and non-motorized traffic. This forces slow and
non-motorized traffic to use expressways and to cross them illegally where that
majority of the victims of road accidents on intercity highways are the vulnerable
road users.

FUTURE DIRECTIONS

Various road users have different and often, conflicting requirements. Motorized
vehicles need clear pavements and shoulders, while bicyclists and pedestrians need
shaded trees along the pavement to protect them from the summer sun. Owners
of private transport modes like MTW and automobiles prefer uninterrupted flow,
fewer stops and minimum delays at intersections, whereas public transport buses
require frequent stops for picking and discharging passengers. Motorized four-
wheeled vehicles like cars, buses, etc., perform better if they move in queues with
minimum braking and acceleration. Since our infrastructure design does not
account for the existing conflicting requirements of different modes, all modes
have to share the road space and operate in sub-optimal conditions.
Experience of past decades of long-term integrated land-use transport plan
exercise suggests that the existence of informal sector and their travel needs must
be recognized for preparing effective plans. This should encourage mixed land-use
patterns and transport infrastructure especially designed for bicycles and other
non-motorized modes.
Future traffic models must account for the users of different transport modes
having conflicting requirements. These models must account for the needs of
motorized vehicles for clear roads for uninterrupted traffic flow, at the same time
they must address the needs of bicyclists and pedestrians for shady trees, kiosks
for drinks, food and bicycle repair shops, etc., at shorter distances. Highway
planning standards provide for services needed by motorized vehicle users. However,
there are no standards for providing services needed by NMT. These services
mushroom along urban or inter-city highways to fulfill the demand of road users,
however their existence is viewed as ‘illegal encroachment’ on the designed road
space.
Motorized vehicles are designed to operate at much higher speeds for better fuel economy
and emission levels. Roads are also designed to increase throughput of motorised vehicles
only. These measures decrease safety of NMV occupants and pedestrians sharing the same
road space. Therefore, safe facilities—segregated lanes, convenient crossing opportunities
from the point of view of NMV users should form an integral part of the road designs. At
present these facilities are viewed as cost increasing measures which many developing
countries cannot afford due to resource crunch.
Urban streets passing through the commercial development and highways
passing through small towns serve multiple purposes. They carry through traffic.
86/ Injury Prevention and Control

However, the adjacent land-use generates cross-traffic and

demands multiple space usage, for example, space for parking
vehicles, space for hawkers and informal shopping, etc. The
existing design standards do not account for conflicting
demand between local traffic and through traffic resulting in
sub-optimal conditions, i.e., long delays for through traffic
and safety hazards for local traffic specially at off peak hours,
for both kinds of traffic.
We have to accept the fact that safety has to be promoted
in most LMCs within existing conditions. These include low
per-capita incomes, presence of mixed traffic, low capacity for
capital intensive infrastructure and different law enforcement
capabilities. This approach will be important for most LMCs
as they are not likely to experience economic growth rates which
puts them at par with HMCs within the next couple of decades.
This implies that pedestrians, bicyclists and motorized two--
wheeler (MTW) riders will remain dominant on LMC roads
for many decades. This group of road users will be called the
vulnerable road users (VRUs).
Such traffic systems are very complex and will need new
understanding. Therefore traffic flow and safety models need
to developed for this complex traffic to meet their specific
requirements not found in homogenous traffic conditions.
A major shift is required in design principles itself to
promote safety in LMCs. If a large number of users are
pedestrians, bicyclists and other slow moving vehicles then road
designs have to address their needs in addition to the needs of
motorized vehicles.Motorized vehicles can use a longer route
and over-bridges, however, a pedestrian or bicyclist would prefer
not to use an underpass or over-bridge just because it is safer
to do so. For this group of users convenience is an overriding
there is a priority.
need to There is a need to accommodate the conflicting
accommodate requirements of NMV occupants and pedestrians, and
the motorized traffic on our urban and inter-city highways. This
conflicting includes redesigning the road cross section setting more
requirements exclusive space for pedestrians and NMVs ,and giving
pedestrians and bicyclists priority over cars at certain places.
of NMV Speed control is perhaps the most important measure for
occupants reducing road traffic crashes in LMCs. Methods to control
and speed in urban and residential areas should be given the highest
pedestrians, priority. Currently we do not have a good understanding of
and how this can be done at low policing levels in the traffic mix
motorized seen in LMC urban areas. For rural roads there are no good
traffic on our designs for non-expressway safety with high VRU participation.
The quality of roads issue has to be addressed in terms of
urban and providing better facilities to non-motorized road users,
inter-city developing suitable designs for heterogenous traffic and those
highways for slowing traffic in residential areas. In many LMCs even
 Tiwari /87

on national highways, majority of people killed are pedestrians, bicyclists, those

using two-wheelers or involved in crashes with tractors/bullock carts. Therefore,
unless these issues are addressed and methods developed for area wide safety
improvements we will not be gaining much by concentrating on blackspot
treatment.

REFERENCES

AAMA, World Motor Vehicle Data, 1995 Edition. American Automobile

Manufactures’ Association: Detroit, USA. 1995.
Central Road Research Institute, Mobility Levels and Transport Problems of Various
Population Groups, CRRI, Mathura Road, New Delhi, India, 1988, p32.
CRRI, 1992, Development of Traffic and transport Flow Data Base for Road System
in Delhi Urban Area (New Delhi:Central Road Research Institute)
Dimitrou, H.T. Transport and Third World City Development. In Planning in Third
World Cities, ( H. Dimitrou and G.Banjo editors), Routledge, London, 1990.
Ekman, L., 1996, On the treatment of flow in traffic safety analysis: A non-parametric
approach applied on vulnerable road users. Bulletin 136 (Sweden: University of Lund,
Lund Institute of Technology, Department of Traffic Planning and Engineering). p.
22
European Transport Safety Council, 1995, Reducing Traffic Injuries Resulting from
Excess and Inappropriate Speed. (Brussels: European Transport Safety Council).
Fazio J. and Tiwari, G., 1995, Nonmotorized-motorized traffic accidents and conflicts
on Delhi Streets. Transport Research Record, 1487, 1995, pp. 68–74.
Fazio J., Hoque, MD. M., Tiwari, G., 1998, Fatalities of Heterogenous Traffic in
Large South Asian Cities. In Proceedings of the Third International Symposium on
Highway Capacity, vol.1, edited by Rikke Rysgaard.(Denmark: Copenhagen). pp. 423–
436.
Indian Express, 1994, Better Traffic policing urged, 26 February,
Insurance Institute for Highway Safety, 1987, IIHS Facts: 55 speed limit . (Arlington,
Va, USA: Insurance Institute for Highway).
Kulmala, R., 1995, Safety at rural three- and four-arm junctions: Development and
application of accident prediction models, (Espoo:VTT).
May, A. D., 1990, Fundamentals of Traffic Flow.
Mohan, Dinesh, 1992, Safety of vulnerable road users. Indian Highways, 20 (4),
pp. 29–36.
Newman, Peter and Kenworthy, Jeff, 1989, Cities and Automobile Dependence, An
International Source Book. (Brookfield, Vermont, USA: Grower Publising).
Noguchi K., 1990, In search of optimum speed: From the user’s viewpoint. Journal
of the International Association of Safety Sciences, Vol 14, No 1 .
Operations Research Group,1994, Household Travel Surveys in Delhi, Final Report ,
Sept., (New Delhi).
Tiwari, G., 1994, Safety aspect of public transort vehicles in developing countries.
Journal of Traffic Medicine .
Thomson, J. M., 1977, Great Cities and Their Traffic, (England: Penguin).
Tiwari, G, and R.Saraf, 1997, Bicycle Transport and Prevention of Theft – A Case
Study of Five Cities. (The Netherlands: Interface for cycling Expertise, I-CE).
88/ Injury Prevention and Control

Tiwari G., Mohan, Dinesh and Fazio, J, 1998, Conflict Analysis for prediction fatal
crash locations in mixed traffic streams, Accident Analysis and Prevention, Vol.30,
No.2, pp. 207–215.
University of Zurich and Swiss Federal Institute of Technology, 1986, The Car Pedestrian
Collision. Interdisciplinary Working Group for Accident Mechanics (Zurich: Switzerland).
World Bank, 1995, Non-Motorised Transport in Ten Asian Cities, Report TWU 20,
(Washington, DC, USA: The World Bank: Washington DC).
World Resource Institute, 1996, World Resource: A Guide to the Global Enviroment .
(Washington,DC, USA: The World Resources Institute).