Draft Manual of Road Safety

(Elective)

For
Master of Science in Transportation Engineering

Institute of Engineering
2016
1
Course Contents
1. Introduction: Safety Issues
 Safety scenario of different modes of transportation
Comparison of crash records of different modes of transportation
 Road deaths in national and international scenario
Trends of road crashes/fatalities in Nepal, zonal wise comparison, urban area wise
comparison, crashes in SRN and LRN, comparison with other countries
2. Human-Vehicle-Infrastructure in Road crashes
 Haddon matrix approach, safe system approach (safe speed, safe vehicles, safer roads
and roadsides)
 Factors involved in transportation crashes (Driver, vehicle, roadway, environment)
 Human factors in road design (driving task, inattentiveness, visibility, speed, P-R
time)
 Driver characteristics
 Human response process
 older driver characteristics
 pedestrian characteristics
 Bicyclists characteristics
 Road Users: Licensing system, Road rules
 Vehicles (roadworthiness, standard compliance, inspection and maintenance system)
 Vehicle characteristics (static, dynamics, kinematics
 Vehicles safety standards and design
 Roads: road design and maintenance
 Lane width, shoulder, horizontal and vertical curves, widening, road surface, sight
distance, drainage
 Environment: Traffic and general control measures
 Delineation, road markings, guide posts, warning signs including weather warning
3. Road Safety data and use
 Importance of Crash and Road Data, Crash records, crash types, recording system
 Relevance of different parameters to road safety: speed, volume, road/traffic control
inventory, asset management database
 Crash estimation methods
 Observed crash frequency and crash rate methods
 indirect safety measures
 empirical bayes method
 Network screening
 Introduction
 process
 Example of Crash rate screening, equivalent property damage only crash rate
frequency, relative severity index, relative rate

2
 4. Road Safety Audit/Inspection
 Definition and importance
Cost and benefits
 Stages, components, process
 Road safety audit manual (DoR, AusRoad)
 Road Safety Notes (Safer drains, Delineation; Safety Barriers; Identifying and treating
accident sites)
 Safety checks for new road projects
 Audit of road designs
 feasibility stage audit
 preliminary design stage audit
 detailed design stage audit
 preopening stage audit
 Safety audit for existing roads
 Audit of roadwork traffic schemes
 audit of land use developments
Audit of existing/newly designed roads will be linked to practical assignment

5. Crash Investigation, analysis and Treatment

 Crash data recording, coding, analysis
 Need of crash data
 coding criteria
 primary and secondary data
 technology available for data collection
 data accuracy
 Analysis of crash data
 Direct comparison of number of crashes
 direct comparison of crash rates, crash patterns
 statistical comparison
 Crash treatment process
 Relation between Road safety audit and crash location treatment
 Steps in crash location treatment
6. Engineering design of road safety measures
 Crash barriers (types)
 Delineation measures
 Traffic sign placement, Markings
7. Crash Costing and Analysis
 Human capital approach
 Willingness to pay approach
 Compensation payment approach

3
 8. Post-crash response
 Emergency preparedness
 Response and treatment
 Medical treatment and trauma management
9. Road safety education and public awareness
 Definition and importance
 Teachers guide, road users guide, children education
 Road safety awareness publicity campaign
10. Miscellaneous
 Act and guidelines related to road safety in Nepal
 Introduction to UN decade of Action on Road Safety

Practical:
Practical marks will be evaluated based on:
 Crash data collection and analysis: report submission/presentation
 Road safety audit of at least 500m section of road for a group of students (4-6) with report
preparation and presentation at the end

References:
 ITE Traffic Engineering Handbook 6th Ed. (2010)
 Principles of Traffic & Highway Engineering -2010; Garber-Hoel- Raju Sarkar
 Roess, RP., McShane, WR. and Prassas, ES. (1998), Traffic Engineering, Prentice Hall
 Austroads Road Safety Guidelines 2009
 US Highway Safety Manual 2009
 DoR Road Safety Note 4: Road Safety Audit Manual 1997, Safer drains, Delineation;
Safety Barriers; Identifying and treating accident sites
 Towards Safer Roads in Developing Countries: A Guide for Planners and Engineers; UK
ODA/TRL; 1994
 www.dotm.gov.np
 Road Traffic Injury Prevention Training Manual, WHO and IIT Delhi

4
Chapter 1: Introduction
1.1 Introduction to Safety Issues

Transportation is vital infrastructure for the overall economic development of a region or a nation.
Transportation is responsible for the development of civilizations from the very ancient times by
meeting travel requirement of people and transport requirement of goods. Transportation system
is often described as veins and arteries for the flow of economy of the nation. There is a strong
correlation between the quality of transport facilities and standard of living, because of which
society places a great expectation from transportation facilities. The adequacy of transportation
system of a country indicates its economic and social development.

Transportation systems analysis has now emerged as a recognized profession. More and more
government organizations around the world are becoming truly multi-modal in their orientation
and are opting a systematic approach to transportation problems. And the transportation system is
now being analyzed in terms of the various modes of transportation. One of the silent killers of
human life is transportation. Several statistics evaluates that more people are killed due to
transportation than great wars and natural disasters.

1.1.1 Safety scenario of different modes of transportation

Table1 below provides a breakdown of fatalities by mode of transportation in United States.
Overall, transportation fatalities have been declining with few exceptions such as motorcycle. For-
hire carriers and transportation professionals have better safety records than private operators. For
instance, commercial ships and their seasoned crews accounted for 41 vessel-related fatalities
while recreational boating experienced 672 fatalities. Transit fatalities have stayed about the same
while transit ridership has significantly increased.

Table 1: Transportation Fatalities by Mode: Select Years 1990-2010

Mode 1990 2000 2009 2010

Air (in aircraft and ground fatalities)
Large U.S. air carrier 39 92 52 2
Commuter air carrier 6 5 0 0
On-demand air taxi 51 71 17 17
General aviation 770 596 478 450

1
US Department of Transportation

5
 Highway (in vehicle and non-occupants)a 44,599 41,945 33,883 32,885
Pipeline, gas and hazardous liquid 9 38 13 25
Railroad (on train and non-occupants)b 729 631 544 601
Transitc 235 208 224 215
Waterborne
Vessel-related, commercial ship 85 53 49 41
Nonvessel-relatedd, commercial ship 101 69 58 43
Recreational boating 865 701 736 672
a Includes fatalities at railroad crossings. b Incidents and accidents; includes
commuter rail; excludes public highway-rail grade crossings involving motor
vehicles. c All reportable incident and accident fatalities. d For example, a person
on board stumbles falls overboard and drowns.
Table2 below shows injured persons by mode of transportation in the United States. The total
number of injured persons has been declining. For instance, recreational boating injuries decreased
by 6.1 percent in 2010. Transit-related injuries ticked upward by nearly 9.3 percent in 2010. This
reduction could be attributed to the fact that the U.S. Coast Guard and its partners have taken
efforts to educate boaters, especially since operator error accounts for 70 percent of boating
accidents.

2
US Department of Transportation

6
 Table 2: Injured Persons by Transportation Mode: Select Years 1990-2010

Mode 1990 2000 2009 2010

Air
Large U.S. air carriera 29 29 23 15
a
Commuter air carrier 11 7 1 2
a
On-demand air taxi 36 12 4 6
a
General aviation 409 309 273 256
b
Highway 3,230,666 3,188,750 2,217,000 2,243,000
Pipeline, gas and hazardous liquid 76 81 67 111
c
Railroad 22,736 10,424 7,227 7,376
d
Transit 54,556 56,697 21,420 23,414
Waterborne
Vessel-related, commercial ship 175 150 196 139
Nonvessel-related, commercial
shipe U 607 377 417
Recreational boating 3,822 4,355 3,358 3,153

a
Serious injuries only. bVehicle occupants, motorcyclists, pedestrians, pedal cyclists,
and other nonmotorists. cInjuries from train accidents, train and nontrain incidents,
and occupational illness. Includes Amtrak. Also includes train occupants and
nonoccupants except motor vehicle occupants at grade crossings. dAll accidents and
reportable incidents. Since 2008, only injuries requiring immediate medical treatment
away from the scene are reportable. Previously, any injury was reportable. Includes
commuter rail. eInjuries unrelated to vessel operations.

1.1.2 Road Crashes in International Scenario

There were 1.25 million road traffic deaths globally in 2013 – a figure that has plateaued since
2007 (see Figure below3).

3
Global Status Report on Road Safety, 2015, WHO

7
The population increase of 4% between 2010 and 2013 and an increase of 16% in registered
vehicles over the same period suggest that efforts to slow the increase in road traffic deaths may
have prevented deaths that would otherwise have occurred. But while the levelling out of road
traffic deaths in a context of rising motorization is encouraging, there are still no signs of an actual
decline, which is essential if the Decade of Action and Sustainable Development Goal targets are
to be realized. This suggests that while progress is possible, much more attention, political will
and resources are needed.

Ninety-percent of road traffic deaths occur in low- and middle-income countries, and while these
countries also account for 82% of the world’s population, they nevertheless bear a disproportionate
number of deaths relative to their level of motorization, as they account for only 54% of the world’s
registered vehicles (figure below).

8
While the global rate for road traffic deaths is 17.4 per 100 000, there is great disparity by income,
with rates more than twice as high in low- and middle-income countries than in the world’s high
income countries (figure below).

Road Traffic Deaths per 100000 population, by country

income statusa
Road Traffic Deaths per 100000 population

30
24.1
25
18.4 17.4
20
15
9.2
10
5
0
World
Medium income
Low income

High income

a Country income status was determined based on data from the World
Development Indicators Database WorldBank March 2015

Almost half of all deaths on the world’s roads are among those with the least protection –
motorcyclists, cyclists1 and pedestrians. However, the likelihood of dying on the road as a
motorcyclist, cyclist or pedestrian varies by region: the African Region has the highest proportion
9
of pedestrian and cyclist deaths at 43% of all road traffic deaths, while these rates are relatively
low in the South-East Asia Region (Figure below).

1.1.3 Status Quo of Nepal

Modes of Transportation in Nepal

Nepal’s transport sector accounts for a large part of domestic passenger and freight movements.
Being a landlocked country and due to high current and inadequate water depth in Nepalese river
the water transport has very limited potential. Air services were the main mode of transportation
in remote hills of Nepal till few years back and still contribute to passenger movements to key
commercial and tourist destinations, and to the transport of both passengers and goods into remote
hill areas. The single operating railway system covers only 52 km within the country which is in
pathetic condition. The road construction in Nepal is very difficult and costly due to the rugged
topography and poor geology. There are still 2 districts among 75 districts still not connected by
road. International trade movements are conducted through various land routes across Nepal India
border among which 7 routes are significant and there is only one land route across China Nepal
border. Limited high-value goods are carried by air. Therefore, Roads are the principal transport
mode in Nepal.

Road Network
The Department of Road (DoR), functioning under the Ministry of Physical Infrastructure and
Transport (MoPIT), is responsible for the construction and maintenance of the Strategic Road
Network (SRN). The Strategic Road Network (SRN), backbone of the National Road Network,
are the main national arteries providing inter-regional connections and links to regional and district
10
headquarters, international borders, key economic centers, touristic centers and the major urban
roads.

The Local Road Network (LRN), comprising of district roads, village or agricultural roads and
nonstrategic urban roads, are being constructed and maintained by the local governments and users
with the support from Department of Local Infrastructure Development and Agriculture Roads (
DoLIDAR), functioning under the Ministry of Federal Affairs and Local Development. Local
Governments consists of District Development Committees (DDCs), Village Development
Committees (VDCs) and Municipalities. DOLIDAR has recently conducted an inventory survey
and found that there is some 51000 km of Local Road Network in the country as of 2013.

Among the South Asian Countries, Nepal has a very low road density, not only in terms of serving
the population but also in providing accessibility to various parts of the country. Although the
strategic roads constitute about 20 percent of the National Road Network, it plays a very important
role in terms of the movement of people and freight. The strategic roads have high traffic volume
in comparison to district roads. At present, the Strategic Road Network consists of 21 National
Highways and 208 Feeder roads totaling 11636 km as of year 2011/12.

With the combined lengths of SRN and local road networks of 62579 km, the road density km per
100 sq. km is 42.51 and influenced population number per km is 425. If the serviceability is
considered, the road density drops down drastically.

Vehicle
At present more than 154599811 motorized vehicles are registered in the country with highest
share of motorized two wheelers is 77.9%. Light vehicles, car, Jeep and pickup share the second
highest as 9.5% and the public utility vehicle mainly Bus, Mini Bus and microbus share 2.9%. And
Crane, Dozer and Truck share (3.4%) and Tractors, three wheelers and other share 5.4%, 0.5% and
0.4% respectively.

11
Table 3: Total number of vehicles registered in Nepal

Road Crashes
More focus was in developing road length by constructing new roads and very low consideration
was given for maintenance and road safety till 80s. Most of the roads were under traffic to their
design capacity. Road construction followed standard geometrics with least concern for road
safety. As the traffic increased in these roads, accidents are also increasing in alarming rate.

Road crashes are increasing in Nepal due to increased vehicle fleet and speed. Most of the crashes
having any human injury are reported to the police. Crashes with minor injury and small damages
to the vehicles may be settled at the crash site with mutual understanding and may not be reported
to the police. A fairly large number of crashes are never reported to the police, mainly because the
involved parties want to settle the matter between them. Generally, only those crashes with high
injury or property damage or with disputes are reported and recorded in the police office. This
under-representation is assumed to be less pronounced for severe crashes. The figures do not give
the full crash picture.
Table 4: Annual Crash Description
Fiscal
S. No. No. of Crashes Fatality Serious Normal
Year
1 2049/50 1945 42 156 833
2 2050/51 1987 83 145 851
3 2051/52 2055 85 203 1028
4 2052/53 2272 104 237 1052

12
 5 2053/54 2396 72 230 1120
6 2054/55 2081 82 206 945
7 2055/56 2197 98 264 1448
8 2056/57 1875 75 132 1042
9 2057/58 2055 124 294 1294
10 2058/59 2180 75 190 1330
11 2059/60 2225 180 144 1908
12 2060/61 3652 86 471 1894
13 2061/62 3709 127 626 1697
14 2062/63 1989 83 459 1673
15 2063/64 2097 93 491 2179
16 2064/65 2211 120 611 2163
17 2065/66 2765 137 720 2248
18 2066/67 4104 146 748 3116
19 2067/68 4914 171 553 3632
20 2068/69 5096 148 396 3317
21 2069/70 4770 147 246 3431
22 2070/71 4672 143 229 3481
23 2071/72 4999 133 233 3643

Table 5: Yearly crash description by crash type and fatality rate

13
Total number of road fatalities in past decade was going decreasing in 2002, 2003 2004 and 2005
in comparison with base year 2001 but it became increasing since 2006 and seems very much
threatening in 2009. The number of death of passengers and pedestrians per 10000 vehicles is also
increasing steeply since 2006. The trend of minor injury, serious injury and fatality are also in
increasing trend and was steeply rising from 2006 to 2010.

From the table below, it is evident that the reasons of majority of crashes are negligence by the
driver and over speed which account for 43.7% and 18.7% respectively. Overtaking, high speed,
overload, drinking and driving are also the negligence of the driver. Therefore, enforcement and
awareness of the driver is lacking somewhere. Again, awareness is lacking regarding negligence
of pedestrian and cattle driving on the road. But a clear cut analysis cannot be made with the data.

Table 6: Crash causes

Table 7: Crash type comparison of South East Asian Countries

14
The table presented above shows that Nepal’s fatality rate per registered vehicles is around the
average rate within South Asia (17 per 10000 registered vehicles) but higher than the
corresponding rate in South East Asia and China.

The table below shows the extrapolated road crash rates by different regions of the country though
there is no official crash rates.

Table 8: Crash rates by region

15
16
The following summarizes the findings of road traffic crashes in Nepal based on past research and
monitoring4.
 About half of all the road traffic crashes nationwide occur in the Kathmandu Valley alone
where nearly half of the country’s fleet ply.
 The severity of the road traffic crash injuries in the Kathmandu Valley is less pronounced
than in the rural areas.
 Road traffic crash fatalities amongst the vehicle fleet are higher in the regions outside the
Kathmandu Valley.
 Pedestrians are the most vulnerable groups in road accidents because pedestrian-safety has
not been considered.
 People between 15 to 40 years of age are most affected in road-crash followed by those
above 50 years.
 In the urban areas, there is significant number of motor-cycle crashes.
 In the rural areas, there are significant number of trucks and bus crashes.
 Bus crashes along the long-distance routes are of serious concern accounting for 13% and
31%of all the fatalities and serious injuries, respectively.
 Single bus crash where the vehicle runs over the hill-roads represent the fatal road traffic
crash of catastrophic proportions.
 About 30 to 40% of the crashes happen after sunset when traffic is low.
 Driver negligence, drunk driving, random roadside parking, reckless pedestrian crossing,
poor road conditions, etc., were the major causes responsible for the road crashes.
 Crashes tend to cluster at the following locations.
o Urban areas: intersections
o Highways: bridge approaches, intersections, and roadside built-up areas.
 From a conservative estimate, the economic loss from road traffic crashes in Nepal was at
least NRs.22.7 billion (US$ 41.2 million) annually or 0.4 % of the GNP at 2007 price.
When the accident under-reporting are adjusted, the loss was 0.8% of the GNP annually.

4
Nepal Road Safety Action Plan 2013-20, Ministry of Physical Infrastructure and Transport, 2013

17
Chapter 2: Human, Vehicle, Infrastructure in Road Crashes

2.1 Safe System Approach

Safe System recognizes that there are limits to the forces the human body can withstand in a
collision, and seeks to ensure that no road user is subject to forces in a collision which will result
in death or an injury from which they cannot recover. It must be recognized that human error is a
feature of the road transport system and that while much can be done to reduce it, it cannot be
eliminated.

A Safe System should provide road users with:

 positive guidance in order to safely negotiate the road transport system
 clear direction regarding appropriate speed for the situation
 space to recover from mistakes, when mistakes do occur (in order to avoid collisions)
 protection from serious injury or death when crashes do occur.

The Safe System framework requires that:

 The road transport system is designed, built and maintained so that the forces road users
are subjected to in collisions will not result in death or serious injury.
 Roads and roadsides should be improved to reduce the risk of crashes and the severity of
impacts when crashes do occur; vulnerable road users must be considered in this process.
 Inclusion of effective active and passive safety systems in vehicles should be encouraged
to reduce the number of crashes and reduce the impact forces on occupants and road users
outside the vehicle.
 Speeds should be managed, taking account the risks on different parts of the road transport
system.

In addition, the realization of Safe System requires alert and compliant road users. This is achieved
through:
 limiting the admittance to the system to qualified drivers who have been trained
appropriately and do not suffer from disabilities which prevent them from driving safely,
and excluding drivers whose poor driving record makes them an unacceptable risk to the
community
 encouraging road users to understand crashes, the factors which contribute to them, and the
risks associated with them
 enforcement of the road rules, accompanied by publicity regarding the extent of
enforcement, the extent of offences detected, and the penalties imposed
 on-going advice, education and encouragement to promote safe road user behaviour.

Road safety is a core responsibility for road agencies. However, road safety goals cannot be
considered in isolation. A lot is expected of the road transport system and competing demands

18
must be acknowledged when considering road safety goals. In addition to its primary functions of
providing for the movement of people and goods to support economic and social activities, the
road transport system is expected to provide for all members of the community as road users in an
equitable manner. It is also expected not to intrude unreasonably on residential and recreational
areas, or on areas with high scenic or conservation values, and not to result in unacceptable levels
of pollution or resource depletion. It is expected to deliver the benefits of mobility while retaining
environmental quality, and to do so at reasonable cost whilst maintaining the highest levels of
safety for its users.

A Safe System aims to prevent road fatalities and serious injuries from occurring on the road
network, yet it seems these are almost a by-product of the competing demands on the road transport
system, the operation of which is essential for the efficient functioning of modern societies. As
such, ambitious road safety goals cannot be achieved by road safety agencies working alone;
instead cooperation and collaboration is required between a range of government and non-
government agencies as well as the general public.

The key pillars of safe system are

• Safer Roads
• Safer Speeds
• Safer Road Users
• Safer Vehicles
As stated earlier, multi stakeholder approach should be taken into account in safe system approach
of road safety as they have shared responsibility to make our safer.

2.2 Factors Involved in Transportation Crashes

While it is common to refer to the “cause” of a crash, in reality, most crashes cannot be related to
a singular causal event. Instead, crashes are the result of a convergence of a series of events that
are influenced by a number of contributing factors (time of day, driver attentiveness, speed, vehicle
condition, road design etc). These contributing factors influence the sequence of events (described
above) before, during and after a crash.

Before-crash events - reveal factors that contributed to the risk of a crash occurring, and how the
crash may have been prevented. For example whether the brakes of one or both of the vehicles
involved were worn;

During-crash events – reveal factors that contributed to the crash severity and how engineering
solutions or technological changes could reduce crash severity For example whether a car has
airbags and if the airbag deployed correctly;

After-crash events – reveal factors influencing the outcome of the crash and how damage and
injury may have been reduced by improvements in emergency response and medical treatment For
example the time and quality of emergency response to a crash.

Crashes have the following three general categories of contributing factors:

19
Human – including age, judgment, driver skill, attention, fatigue, experience and sobriety;
Vehicle – including design, manufacture and maintenance;
Roadway/Environment – including geometric alignment, cross-section, traffic control devices,
surface friction, grade, signage, weather, visibility.

By understanding these factors and how they might influence the sequence of events, crashes and
crash severities can be reduced by implementing specific measures to target specific contributing
factors. The relative contribution of these factors to crashes can assist with determining how to
best allocate resources to reduce crashes. Research by Treat into the relative proportion of
contributing factors was conducted in 1980 and the relative proportions are shown in the figure
below.

A framework for relating the series of events in a crash to the categories of crash contributing
factors is the Haddon Matrix. The chart below provides an example of this matrix. The Haddon
Matrix helps create order when determining which contributing factors influence a crash and which
period of the crash the factors influence. The factors listed are not intended to be comprehensive;
they are examples only.

2.2.1 Driver or Operator Action

The major contributing cause of many crash situations is the performance of the driver of one or
both (in multiple vehicle crashes) of the vehicles involved. Driver error can occur in many ways,
such as inattention to the roadway and surrounding traffic, failure to yield the right of way, and/or
traffic laws. These “failures” can occur as a result of unfamiliarity with roadway conditions,
traveling at high speeds, drowsiness, drinking, and using a cell phone or other distractions within
the vehicle.

20
2.2.2 The Vehicle Condition
The mechanical condition of a vehicle can be the cause of transportation crashes. Faulty brakes in
heavy trucks have caused crashes. Other reasons are failure of the electrical system, worn tires,
and the location of the vehicle’s center of gravity. Studies have indicated that 5 to 8.5% crashes
are directly caused by vehicle condition faults. Recent studies in UK indicate that 25% of goods
vehicles and 11% of large passenger carrying vehicles involved in crashes have contributory
defects. In developing countries it is likely that vehicle defects are more often a factor in crashes
as vehicle condition is generally much worse. The vehicle fleet is usually older with many vehicles
imported second hand. There may be difficulty in obtaining suitable spare parts.

Shortage of specialized equipment and tools, maintenance skills are often scarce and knowledge
of modern repair techniques is frequently poor. Short term thinking often dominates maintenance
and repair decisions. Lighting defects are common. Roadside survey conducted in Kathmandu in
1996 found only 40% of trucks and buses inspected had front lights and a nighttime survey of long
distance buses found two thirds with one or no rear lights. Vehicle lighting is even more important
in developing countries where road lighting and road marking and signs are inadequate and driving
conditions poor.

Overloading is a serious problem in many developing countries. Each vehicle has a designed
maximum weight which depends on the structural strength of the chassis, suspension, braking
system, tires and engine power. Overloading of freight transport vehicles has detrimental effects
on the economy of operation of national road networks. These effects are reflected in terms of
premature road surface deteriorations and adverse impacts on its structures which call for early
maintenances, overlaying and rehabilitations of roads. This is a case of economic loss to the
country. Overloading has another detrimental effect due to road traffic crashes thereby causing
loss of lives and properties. It may also cause premature aging of vehicles by wear and tear causing
financial loss to the truck owner. Another adverse effect occurs in terms of fumes of un-brunt fuels
causing environmental hazard for health of the public in general and emitting toxic gases.
Overloading may also cause slowing down the speeds of vehicles which may cause traffic
congestion on the roads. Vehicles react differently when the maximum weights which they are
designed to carry are exceeded. The vehicle will be less stable, difficult to steer and take longer
time to stop. The overloaded vehicle cannot accelerate as normal making it difficult to overtake.
At night, the headlights of an overloaded vehicle tilt up blinding oncoming drivers to possible
debris or obstructions on the roadway. Due to overloading brakes have to work harder and get
overheated and lose their effectiveness to stop the vehicle leading to accident. For example,
braking distance will increase, which causes drivers to misjudge stopping distances. In addition, a
raised center of gravity adds to the risk of a rollover.

Unofficial modification of goods vehicles by welding reinforcements or extensions to be chassis

or the addition of an extra axle can result in a vehicle with a seriously compromised safety
performance.

Buses are often constructed on truck chassis using materials and designs that offer little or no
protection to passengers in the event of a crash. Leg room between seats is often exceptionally

21
poor in order to squeeze the maximum number of seats in the vehicle and the seat frames
themselves are usually formed from angle steel, frequently causing amputation of limbs in a crash.
Emergency exits when provided are often blocked by a row of seats making their use difficult.

There is no recent data on accident costs in Nepal. But during the study of Road Connectivity
Sector I Project, vehicle accident costs for fatal, serious injuries and minor injuries as well as
average truck damage were estimated. The report shows 8% of total accidents in urban areas and
30% of total accidents in rural areas were those of truck accidents. The estimated costs were as
follows:

1. Fatal accident: NRs 609,236.0/person/time

2. Injury accident: NRs138,479.0/person/time

3. Net vehicle damage cost of a truck: NRs 180,308.45/vehicle/time

The costs show the extent of economic costs of accidents if occurred due to overloading.

2.2.3 The Roadway Condition

The condition and quality of the roadway, which includes the pavement, shoulders, intersections
and the traffic control system can be a factor in a crash. Highways must be designed to provide
adequate sight distance at the design speed or motorists will be unable to take remedial action to
avoid a crash. Traffic signals must provide adequate decision sight distance when the signal goes
from green to red. Railroad grade crossings must be designed to operate safely and thus minimize
crashes between highway traffic and rail cars. Highway curves must be carefully designed to
accommodate vehicles traveling at or below the design speed of the road.

2.2.4 The Environment

The physical and climatic environment surrounding a transportation vehicle can also be a factor in
the occurrence of transportation crashes with the most common being weather. All transportation
systems function at their best when the weather is sunny and mild and the skies are clear. Weather
on roads can contribute to highway crashes; for example, wet pavement reduces stopping friction
and can cause vehicles to hydroplane. Many severe crashes have been caused by fog because
vehicles traveling at high speeds are unable to see other vehicles ahead that may have stopped or
slowed down, creating a multivehicle pile-up. Geography is another environmental cause of
transportation crashes. Mountain ranges have been the site of air crashes. Flooded river plains,
swollen rivers, and mud slides on the pavement have caused railroad and highway crashes.

2.3 The Role of Human Factors in Road Safety

The interdisciplinary study of human factors applies knowledge from the human sciences such as
psychology, physiology, and kinesiology to the design of systems, tasks, and environments for
effective and safe use. The goal of human factors is to reduce the probability and consequences of
human error within systems, and associated injuries and fatalities, by designing with respect to
human characteristics and limitations.

22
Drivers make frequent mistakes because of human physical, perceptual, and cognitive limitations.
These errors may not result in crashes because drivers compensate for other drivers’ errors or
because the circumstances are forgiving (e.g., there is room to maneuver and avoid a crash). Near
misses, or conflicts, are vastly more frequent than crashes. One study found a conflict-to-crash
ratio of about 2,000 to 1 at urban intersections.

In transportation, driver error is a significant contributing factor in most crashes. For example,
drivers can make errors of judgment concerning closing speed, gap acceptance, curve negotiation,
and appropriate speeds to approach intersections. In-vehicle and roadway distractions, driver
inattentiveness, and driver weariness can lead to errors. A driver can also be overloaded by the
information processing required to carry out multiple tasks simultaneously, which may lead to
error. To reduce their information load, drivers rely on a-priori knowledge, based on learned
patterns of response; therefore, they are more likely to make mistakes when their expectations are
violated. In addition to unintentional errors, drivers sometimes deliberately violate traffic control
devices and laws.

2.3.1 Driving Task Model

Driving comprises many sub-tasks, some of which must be performed simultaneously. The three
major sub-tasks are:

Control: Keeping the vehicle at a desired speed and heading within the lane;

Guidance: Interacting with other vehicles (following, passing, merging, etc.) by maintaining a safe
following distance and by following markings, traffic control signs, and signals; and,

Navigation: Following a path from origin to destination by reading guide signs and using
landmarks.

Each of these major sub-tasks involves observing different information sources and various levels
of decision-making. The relationship between the sub-tasks can be illustrated in a hierarchical
form, as shown in figure below. The hierarchical relationship is based on the complexity and
primacy of each subtask to the overall driving task. The navigation task is the most complex of the
subtasks, while the control sub-task forms the basis for conducting the other driving tasks.

23
A successful driving experience requires smooth integration of the three tasks, with driver attention
being switched from one to another task as appropriate for the circumstances. This can be achieved
when high workload in the sub-tasks of control, guidance, and navigation does not happen
simultaneously.

2.3.2 Driver Characteristics

Attention and Information Processing
Driver attention and ability to process information is limited. These limitations can create
difficulties because driving requires the division of attention between control tasks, guidance tasks,
and navigational tasks. While attention can be switched rapidly from one information source to
another, drivers only attend well to one source at a time. For example, drivers can only extract a
small proportion of the available information from the road scene. It has been estimated that more
than one billion units of information, each equivalent to the answer to a single yes or no question,
are directed at the sensory system in one second. On average, humans are expected to consciously
recognize only 16 units of information in one second.

To account for limited information processing capacity while driving, drivers subconsciously
determine acceptable information loads they can manage. When drivers’ acceptable incoming
information load is exceeded, they tend to neglect other information based on level of importance.
As with decision making of any sort, error is possible during this process. A driver may neglect a
piece of information that turns out to be critical, while another less-important piece of information
was retained.

Scenarios illustrating circumstances in which drivers might be overloaded with information are
described in table below. Each may increase the probability of driver error given human
information processing limitations.

24
As shown in table above, traffic conditions and operational situations can overload the user in
many ways. Roadway design considerations for reducing driver workload are:
 Presenting information in a consistent manner to maintain appropriate workload;
 Presenting information sequentially, rather than all at once, for each of the control,
guidance, and navigation tasks; and,
 Providing clues to help drivers prioritize the most important information to assist them in
reducing their workload by shedding extraneous tasks.

In addition to information processing limitations, drivers’ attention is not fully within their
conscious control. For drivers with some degree of experience, driving is a highly automated task.
That is, driving can be, and often is, performed while the driver is engaged in thinking about other
matters. Most drivers, especially on a familiar route, have experienced the phenomenon of
becoming aware that they have not been paying attention during the last few miles of driving. The
less demanding the driving task, the more likely it is that the driver’s attention will wander, either
through internal preoccupation or through engaging in non-driving tasks. Factors such as increased
traffic congestion and increased societal pressure to be productive could also contribute to
distracted drivers and inattention. Inattention may result in inadvertent movements out of the lane,
or failure to detect a stop sign, a traffic signal, or a vehicle or pedestrian on a conflicting path at
an intersection.

Driver Expectation
One way to accommodate for human information processing limitations is to design roadway
environments in accordance with driver expectations. When drivers can rely on past experience to
assist with control, guidance, or navigation tasks there is less to process because they only need to
process new information. Drivers develop both long- and short-term expectancies. Examples of
long-term expectancies that an unfamiliar driver will bring to a new section of roadway include:
 Upcoming freeway exits will be on the right-hand side of the road;
 When a minor and a major road cross, the stop control will be on the road that appears to
be the minor road;
 When approaching an intersection, drivers must be in the left lane to make a left turn at the
cross street; and,
 A continuous through lane (on a freeway or arterial) will not end at an interchange or
intersection junction.

Examples of short-term expectancies include:

 After driving a few miles on a gently winding roadway, upcoming curves will continue to
be gentle;

25
  After traveling at a relatively high speed for some considerable distance, 123 drivers expect
the road ahead will be designed to accommodate the same speed; and,
 After driving at a consistent speed on well-timed, coordinated signalized arterial corridors
drivers may not anticipate a location that operates at a different cycle length.

Vision
Approximately 90 percent of the information that drivers use is visual. While visual acuity is the
most familiar aspect of vision related to driving, numerous other aspects are equally important.
 Visual Acuity – The ability to see details at a distance;
 Contrast Sensitivity – The ability to detect slight differences in luminance (brightness of
light) between an object and its background;
 Peripheral Vision – The ability to detect objects that are outside of the area of most accurate
vision within the eye;
 Movement in Depth – The ability to estimate the speed of another vehicle by the rate of
change of visual angle of the vehicle created at the eye ; and,
 Visual Search – The ability to search the rapidly changing road scene to collect road
information.

Visual Acuity
Visual acuity determines how well drivers can see details at a distance. It is important for guidance
and navigation tasks, which require reading signs and identifying potential objects ahead. Under
ideal conditions, in daylight, with high contrast text (black on white), and unlimited time, a person
with a visual acuity of 20/20, considered “normal vision,” can just read letters that subtend an
angle of 5 minutes of arc. A person with 20/40 vision needs letters that subtend twice this angle,
or 10 minutes of arc. With respect to traffic signs, a person with 20/20 vision can just barely read
letters that are 1 inch tall at 57 feet, and letters that are 2 inches tall at 114 feet and so on. A person
with 20/40 vision would need letters of twice this height to read them at the same distances. Given
that actual driving conditions often vary from the ideal conditions listed above and driver vision
varies with age, driver acuity is often assumed to be less than 57 feet per inch of letter height for
fonts used on highway guide signs.

Contrast Sensitivity
Contrast sensitivity is often recognized as having a greater impact on crash occurrence than visual
acuity. Contrast sensitivity is the ability to detect small differences in luminance (brightness of
light) between an object and the background. The lower the luminance of the targeted object, the
more contrast is required to see the object. The target object could be a curb, debris on the road,
or a pedestrian. Good visual acuity does not necessarily imply good contrast sensitivity. For people
with standard visual acuity of 20/20, the distance at which non-reflective objects are detected at
night can vary by a factor of 5 to 1. Drivers with normal vision but poor contrast sensitivity may
have to get very close to a low-contrast target before detecting it. Experimental studies show that
even alerted subjects can come as close as 30 feet before detecting a pedestrian in dark clothing
standing on the left side of the road. In general, pedestrians tend to overestimate their own visibility
to drivers at night. On average, drivers see pedestrians at half the distance at which pedestrians

26
think they can be seen. This may result in pedestrians stepping out to cross a street while assuming
that drivers have seen them, surprising drivers, and leading to a crash or near-miss event.

Peripheral Vision

The visual field of human eyes is large: approximately 55

degrees above the 175 horizontal, 70 degrees below the
horizontal, 90 degrees to the left and 90 degrees to the right.
However, only a small area of the visual field allows accurate
vision. This area of accurate vision includes a cone of about two
to four degrees from the focal point. The lower-resolution visual
field outside the area of accurate vision is referred to as
peripheral vision. Although acuity is reduced, targets of interest can be detected in the low-
resolution peripheral vision. Once detected, the eyes shift so that the target is seen using the area
of the eye with the most accurate vision.

Targets that drivers need to detect in their peripheral vision include vehicles on an intersecting
path, pedestrians, signs, and signals. In general, targets best detected by peripheral vision are
objects that are closest to the focal point; that differ greatly from their backgrounds in terms of
brightness, color, and texture; that are large; and that are moving. Studies show the majority of
targets are noticed when located less than 10 to 15 degrees from the focal point and that even when
targets are conspicuous, glances at angles over 30 degrees are rare. Target detection in peripheral
vision is also dependent on demands placed on the driver. The more demanding the task, the
narrower the “visual cone of awareness” or the “useful field of view,” and the less likely the driver
is to detect peripheral targets. Targets are seen in high resolution within the central 2-4 degrees of
the field of view. While carrying out the driving task, the driver is aware of information seen
peripherally, within the central 20 to 30 degrees. The driver can physically see information over a
180-degree area, but is not aware of it while driving, unless motivated to direct his or her attention
there.

Movement in Depth
Numerous driving situations require drivers to estimate movement of vehicles based on the rate of
change of visual angle created at the eye by the vehicle. These situations include safe following of
a vehicle in traffic, selecting a safe gap on a two way stop-controlled approach, and passing another
vehicle with oncoming traffic and no passing lane. The primary cue that drivers use to determine

27
their closing speed to another vehicle is the
rate of change of the image size. The figure
below illustrates the relative change of the
size of an image at different distances from a
viewer.

As shown in figure, the relationship between

viewing distance and image size is not a
linear relationship. The fact that it is a non-
linear relationship is likely the source of the
difficulty drivers have in making accurate
estimates of closing speed.

Drivers use the observed change in the size

of a distant vehicle, measured by the rate of change of the visual angle occupied by the vehicle, to
estimate the vehicle’s travel speed. Drivers have difficulty detecting changes in vehicle speed over
a long distance due to the relatively small amount of change in the size of the vehicle that occurs
per second. This is particularly important in overtaking situations on two-lane roadways where
drivers must be sensitive to the speed of oncoming vehicles. When the oncoming vehicle is at a
distance at which a driver might pull out to overtake the vehicle in front, the size of that oncoming
vehicle is changing gradually and the driver may not be able to distinguish whether the oncoming
vehicle is traveling at a speed above or below that of average vehicles. In overtaking situations
such as this, drivers have been shown to accept insufficient time gaps when passing in the face of
high-speed vehicles, and to reject sufficient time gaps when passing in the face of other low-speed
vehicles. Limitations in driver perception of closing speed may also lead to increased potential for
rear-end crashes when drivers traveling at highway speeds approach stopped or slowing vehicles
and misjudge the stopping distance available. This safety concern is compounded when drivers are
not expecting this situation. One example is on a two-lane rural roadway where a left-turning driver
must stop in the through lane to wait for an acceptable gap in opposing traffic. An approaching
driver may not detect the stopped vehicle. In this circumstance the use of turn signals or visibility
of brake lights may prove to be a crucial cue for determining that the vehicle is stopped and waiting
to turn.

Visual Search
The driving task requires active search of the rapidly changing road scene, which requires rapid
collection and absorption of road information. While the length of an eye fixation on a particular
subject can be as short as 1/10 of a second for a simple task such as checking lane position, fixation
on a complex subject can take up to 2 seconds. By understanding where drivers fix their eyes while
performing a particular driving task, information can be placed in the most effective location and
format.

Studies using specialized cameras that record driver-eye movements have revealed how drivers
distribute their attention amongst the various driving subtasks, and the very brief periods of time
(fixations) drivers can allocate to any one target while moving. On an open road, study drivers
fixated approximately 90 percent of the time within a 4-degree region vertically and horizontally
from a point directly ahead of the driver. Within this focused region, slightly more than 50- percent

28
of all eye fixations occurred to the right side of the road where traffic signs are found. This indicates
that driver visual search is fairly concentrated.

The visual search pattern changes when a driver is negotiating a horizontal curve as opposed to
driving on a tangent. On tangent sections, drivers can gather both path and lateral position
information by looking ahead. During curve negotiation, visual demand is essentially doubled, as
the location of street sign and roadside information is displaced (to the left or to the right) from
information about lane position. Eye movement studies show that drivers change their search
behavior several seconds prior to the start of the curve. These findings suggest that advisory curve
signs placed just prior to the beginning of the approach zone may reduce visual search challenges.
Other road users, such as pedestrians and cyclists, also have a visual search task. Pedestrians can
be observed to conduct a visual search if within three seconds of entering the vehicle path the head
is turned toward the direction in which the vehicle would be coming from. The visual search varies
with respect to the three types of threats: vehicles from behind, from the side, and ahead. Vehicles
coming from behind require the greatest head movement and are searched for the least. These
searches are conducted by only about 30 percent of pedestrians. Searches for vehicles coming from
the side and from ahead are more frequent, and are conducted by approximately 50 and 60 percent
of pedestrians, respectively. Interestingly between 8 and 25 percent of pedestrians at signalized
downtown intersections without auditory signals do not look for threats.

Perception-Reaction Time
Perception-reaction time (PRT) includes time to detect a target, process the information, decide on
a response, and initiate a reaction. Although higher values such as 1.5 or 2.5 seconds are commonly
used because it accommodates the vast percentage of drivers in most situations, it is important to
note that PRT is not fixed. PRT depends on human elements discussed in previous sections,
including information processing, driver alertness, driver expectations, and vision.
The following sections describe the components of perception-reaction time: detection, decision,
and response.

Detection
The initiation of PRT begins with detection of an object or obstacle that may have potential to
cause a crash. At this stage the driver does not know whether the observed object is truly something
to be concerned with, and if so, the level of concern.

Detection can take a fraction of a second for an expected object or a highly conspicuous object
placed where the driver is looking. However, at night an object which is located several degrees
from the line of sight, and which is of low contrast compared to the background, may not be seen
for many seconds. The object cannot be seen until the contrast of the object exceeds the threshold
contrast sensitivity of the driver viewing it.

Failures in detection are most likely for objects that are:

 More than a few degrees from the driver’s line of sight;
 Minimally contrasted with the background;
 Small in size;
 Seen in the presence of glare;
 Not moving; and,

29
  Unexpected and not being actively searched for by the driver.

Once an object or obstacle has been detected, the details of the object or obstacle must be
determined in order to have enough information to make a decision. As discussed in the next
section, identification will be delayed when the object being detected is unfamiliar and unexpected.
For example, a low-bed, disabled tractor trailer with inadequate reflectors blocking a highway at
night will be unexpected and hard to identify.

Decision
Once an object or obstacle has been detected and enough information has been collected to identify
it, a decision can be made as to what action to take. The decision does not involve any action, but
rather is a mental process that takes what is known about the situation and determines how the
driver will respond.

Decision time is highly dependent on circumstances that increase the complexity of a decision or
require it be made immediately. Many decisions are made quickly when the response is obvious.
For example, when the driver is a substantial distance from the intersection and the traffic light
turns red, minimal time is needed to make the decision. If, on the other hand, the driver is close to
the intersection and the traffic light turns yellow, there is a dilemma: is it possible to stop
comfortably without risking being rear-ended by a following vehicle, or is it better to proceed
through the intersection? The time to make this stop-or-go decision will be longer given that there
are two reasonable options and more information to process.

Decision-making also takes more time when there is an inadequate amount of information or an
excess amount. If the driver needs more information, they must search for it. On the other hand, if
there is too much information the driver must sort through it to find the essential elements, which
may result in unnecessary effort and time. Decision-making also takes more time when drivers
have to determine the nature of unclear information, such as bits of reflection on a road at night.
The bits of reflection may result from various sources, such as harmless debris or a stopped vehicle.

Response
When the information has been collected, processed, and a decision has been made, time is needed
to respond physically. Response time is primarily a function of physical ability to act upon the
decision and can vary with age, lifestyle (athletic, active, or sedentary), and alertness.

Speed Choice
A central aspect of traffic safety is driver speed choice. While speed limits influence driver speed
choice, these are not the only or the most important influences. Drivers select speed using
perceptual and “road message” cues. Understanding these cues can help establish self-regulating
speeds with minimal or no enforcement.

Perceptual Cues
A driver’s main cue for speed choice comes from peripheral vision. In experiments where drivers
are asked to estimate their travel speed with their peripheral vision blocked (only the central field
of view can be used), the ability to estimate speed is poor. This is because the view changes very
slowly in the center of a road scene. If, on the other hand, the central portion of the road scene is

30
blocked out, and drivers are asked to estimate speed based on the peripheral view, drivers do much
better.

Streaming (or “optical flow”) of information in peripheral vision is one of the greatest influences
on drivers’ estimates of speed. Consequently, if peripheral stimuli are close by, then drivers will
feel they are going faster than if they encounter a wide open situation. In one study, drivers were
asked to drive at 60 mph with the speedometer covered. In an open-road situation, the average
speed was 57 mph. After the same instructions, but along a tree-lined route, the average speed was
53 mph, The researchers believe that the trees near the road provided peripheral stimulation, giving
a sense of higher speed.

Noise level is also an important cue for speed choice. Several studies examined how removing
noise cues influenced travel speed. While drivers’ ears were covered (with ear muffs) they were
asked to travel at a particular speed. All drivers underestimated how fast they were going and drove
4 to 6 mph faster than when the usual sound cues were present. With respect to lowering speeds,
it has been counter-productive to progressively quiet the ride in cars and to provide smoother
pavements.

Another aspect of speed choice is speed adaptation. This is the experience of leaving a freeway
after a long period of driving and having difficulty conforming to the speed limit on an arterial
road. One study required subjects to drive for 20 miles on a freeway and then drop their speeds to
40 mph on an arterial road. The average speed on the arterial was 50 miles per hour. This speed
was higher than the requested speed despite the fact that these drivers were perfectly aware of the
adaptation effect, told the researchers they knew this effect was happening, and tried to bring their
speed down. The adaptation effect was shown to last up to five or six minutes after leaving a
freeway, and to occur even after very short periods of high speed. Various access management
techniques, sign placement, and traffic calming devices may help to reduce speed adaptation
effects.

Road Message Cues

Drivers may interpret the roadway environment as a whole to encourage fast or slow speeds
depending on the effects of the geometry, terrain, or other roadway elements. Even though drivers
may not have all the information for correctly assessing a safe speed, they respond to what they
can see. Drivers tend to drive faster on a straight road with several lanes, wide shoulders, and a
wide clear zone, than drivers on a narrow, winding road with no shoulders or a cliff on the side.
For example, speeds on rural highway tangents are related to cross-section and other variables,
such as the radius of the curve before and after the tangent, available sight distance, and general
terrain. The difficulty of the driving task due to road geometry (e.g., sharp curves, narrow
shoulders) strongly influences driver perception of risk and, in turn, driver speed.

Speed advisory plaques on curve warning signs appear to have little effect on curve approach
speed, probably because drivers feel they have enough information from the roadway itself and
select speed according to the appearance of the curve and its geometry. One study recorded the
speeds of 40 drivers, unfamiliar with the route, on curves with and without speed plaques.
Although driver eye movements were recorded and drivers were found to look at the warning sign,
the presence of a speed plaque had no effect on drivers’ selected speed. In contrast, a study of 36

31
arterial tangent sections found some influence of speed limit, but no influence of road design
variables. The sections studied had speed limits that ranged from 25 to 55 mph. Speed limit
accounted for 53 percent of the variance in speed, but factors such as alignment, cross-section,
median presence, and roadside variables were not found to be statistically significantly related to
operating speed.

2.4 Impacts of Road Design on the Driver

Intersections and Access Points

As discussed earlier, the driving task involves control, guidance, and navigation elements. At
intersections, each of these elements presents challenges:
 Control: The path through the intersection is typically unmarked and may involve turning;
 Guidance: There are numerous potential conflicts with other vehicles, pedestrians, and
cyclists on conflicting paths; and
 Navigation: Changes in direction are usually made at intersections, and road name signing
can be difficult to locate and read in time to accomplish any required lane changes.

In the process of negotiating any intersection, drivers are required to:

 Detect the intersection;
 Identify signalization and appropriate paths;
 Search for vehicles, pedestrians, and bicyclists on a conflicting path;
 Assess adequacy of gaps for turning movements;
 Rapidly make a stop/go decision on the approach to a signalized intersection when in the
decision zone; and,
 Successfully complete through or turning maneuvers.

Thus, intersections place high demands on drivers in terms of visual search, gap estimation, and
decision-making requirements that increase the potential for error. Road crash statistics show that
although intersections constitute a small portion of the highway network, about 50 percent of all
urban crashes and 25 percent of rural crashes are related to intersections. A study of the human
factors contributing causes to crashes found that the most frequent type of error was “improper
lookout,” and that 74 percent of these errors occurred at intersections. In about half of the cases,
drivers failed to look, and in about half of the cases, drivers “looked but did not see.

Errors Leading to Rear-End and Sideswipe Crashes

Errors leading to rear-end and sideswipe crashes include the following:
 Assuming that the lead driver, once moving forward, will continue through the stop sign,
but the lead driver stops due to late recognition that there is a vehicle or pedestrian on a
conflicting path.
 Assuming that the lead driver will go through a green or yellow light, but the lead driver
stops due to greater caution. Drivers following one another can make differing decisions
in this “dilemma zone”. As speed increases, the length of the dilemma zone increases.
Additionally, as speed increases, the deceleration required is greater and the probability of
a rear-end crash may also increase.

32
  Assuming that the lead driver will continue through a green or yellow light but the lead
driver slows or stops due to a vehicle entering or exiting an access point just prior to the
intersection, or a vehicle exiting an access point suddenly intruding into the lane, or a
pedestrian crossing against a red light.
 Changing lanes to avoid a slowing or stopped vehicle, with inadequate search.
 Distracting situations that may lead to failure to detect slowing or stopping vehicles ahead.
Distracting situations could include:
o Preoccupation with personal thoughts,
o Attention directed to non-driving tasks within the vehicle,
o Distraction from the road by an object on the roadside, or
o Anticipation of downstream traffic signal.

Errors Leading to Turning Crashes

Turning movements are often more demanding with respect to visual search, gap judgment, and
path control than are through movements. Turning movements can lead to crashes at intersections
or access points due to the following:
o Perceptual limitations,
o Visual blockage,
o Permissive left-turn trap, and
o Inadequate visual search.

A description of these common errors that can lead to turning crashes at intersections is provided
below.

Perceptual Limitations
Perceptual limitations in estimating closing vehicle speeds could lead to left turning drivers
selecting an inappropriate gap in oncoming traffic. Drivers turning left during a permissive green
light may not realize that an oncoming vehicle is moving at high speed.

Visual Blockage
A visual blockage may limit visibility of an oncoming vehicle when making a turn at an
intersection. About 40 percent of intersection crashes involve a view blockage. Windshield pillars
inside the vehicle, utility poles, commercial signs, and parked vehicles may block a driver’s view
of a pedestrian, bicyclist, or motorcycle on a conflicting path at a critical point during the brief
glance that a driver may make in that direction. Visual blockages also occur where the offset of
left-turn bays results in vehicles in the opposing left-turn lane blocking a left-turning driver’s view
of an oncoming through vehicle.

Permissive Left-turn Trap

On a high-volume road, drivers turning left on a permissive green light may be forced to wait for
a yellow light to make their turn, at which time they come into conflict with oncoming drivers who
continue through into a red light.

33
Inadequate Visual Search
Drivers turning right may concentrate their visual search only on vehicles coming from the left
and fail to detect a bicyclist or pedestrian crossing from the right. This is especially likely if drivers
do not stop before turning right on red, and as a result give themselves less time to search both to
the left and right.

Errors Leading to Angle Crashes

Angle crashes can occur due to:
o Delayed detection of an intersection (sign or signal) at which a stop is required;
o Delayed detection of crossing traffic by a driver who deliberately violates the sign or
signal; or
o Inadequate search for crossing traffic or appropriate gaps.

Drivers may miss seeing a signal or stop sign because of inattention, or a combination of
inattention and a lack of road message elements that would lead drivers to expect the need to stop.
For example, visibility of the intersection pavement or the crossing traffic may be poor, or drivers
may have had the right of way for some distance and the upcoming intersection does not look like
a major road requiring a stop. In an urban area where signals are closely spaced, drivers may
inadvertently attend to the signal beyond the signal they face. Drivers approaching at high speeds
may become caught in the dilemma zone and continue through a red light.

Errors Leading to Crashes with Vulnerable Road Users

Pedestrian and bicycle crashes often result from inadequate search and lack of conspicuity. The
inadequate search can be on the part of the driver, pedestrian, or 86 bicyclist. In right-turning
crashes, pedestrians and drivers have been found to be equally guilty of failure to search. In left-
turning crashes, drivers are more frequently found at fault, likely because the left-turn task is more
visually demanding than the right-turn task for the driver.

Examples of errors that may lead to pedestrian crashes include:

o Pedestrians crossing at traffic signals rely on the signal giving them the right of way, and
fail to search adequately for turning traffic.
o Pedestrians step into the path of a vehicle that is too close for the driver to have sufficient
time to stop.

When accounting for perception-response time, a driver needs over 100 feet to stop when traveling
at 30 mile per hour. Pedestrians are at risk because of the time required for drivers to respond and
because of the energy involved in collisions, even at low speeds. Relatively small changes in speed
can have a large impact on the severity of a pedestrian crash. A pedestrian hit at 40 mph has an
85-percent chance of being killed; at 30 mph the risk is reduced to 45 percent; at 20 mph the risk
is reduced to 5 percent.

Poor conspicuity, especially at night, greatly increases the risk of a pedestrian or bicyclist crash.
Clothing is often dark, providing little contrast to the background. Although street lighting helps
drivers see pedestrians, street lighting can create uneven patches of light and dark which makes
pedestrians difficult to see at any distance.

34
Interchanges
At interchanges drivers can be traveling at high speeds, and at the same time can be faced with
high demands in navigational, guidance, and control tasks. The following elements of design at
interchanges are often related to errors that can lead to crashes at interchanges that lead to crashes:
 On-ramp/merge length,
 Ramp spacing,
 Decision sight distance and guide signing, and
 Off-ramp radius.

On-Ramp/Merge Length
If drivers entering a freeway are unable to accelerate to the speed of the traffic stream (e.g., due to
acceleration lane length, the grade of the ramp, driver error, or heavy truck volumes), entering
drivers will merge with the mainline at too slow a speed and may risk accepting an inadequate gap.
Alternatively the freeway is congested or if mainline vehicles are tailgating, it may be difficult for
drivers to find an appropriate gap into which to merge.

Ramp Spacing
If the next off-ramp is close to the on-ramp, entering (accelerating) drivers will come into conflict
with exiting (decelerating) drivers along the weaving section and crashes may increase.(38,39)
Given the visual search required by both entering and exiting drivers, and the need to look away
from the traffic immediately ahead in order to check for gaps in the adjacent lane, it is easy to see
how sideswipe and rear end crashes can occur in weaving sections. Drivers may fail to detect
slowing vehicles ahead, or vehicles changing lanes in the opposing direction, in time to avoid
contact.

Decision Sight Distance and Guide Signing

Increased risk of error occurs in exit locations because drivers try to read signs, change lanes, and
decelerate comfortably and safely. Drivers may try to complete all three tasks simultaneously
thereby increasing their willingness to accept smaller gaps while changing lanes or to decelerate
at greater than normal rates.

Off-Ramp Radius
If the exit ramp radius is small and requires the exiting vehicle to decelerate more than expected,
the speed adaptation effect discussed in the previous section can lead to insufficient speed
reductions. Also, a tight off-ramp radius or an unusually long vehicle queue extending from the
ramp terminal can potentially surprise drivers, leading to run-off-road and rear-end crashes.

Divided, Controlled-Access Mainline

Compared to intersections and interchanges, the driving task on a divided, controlled-access
mainline is relatively undemanding with respect to control, guidance, and navigational tasks. This
assumes that the mainline has paved shoulders, wide clear zones, and is outside the influence area
of interchanges.

35
Driver Inattention and Sleepiness
Low mental demand can lead to driver inattention and sleepiness, resulting in inadvertent (drift-
over) lane departures. Sleepiness is strongly associated with time of day. It is particularly difficult
for drivers to resist falling asleep in the early-morning hours (2 to 6 a.m.) and in the mid-afternoon.
Sleepiness arises from the common practices of reduced sleep and working shifts. Sleepiness also
results from alcohol and other drug use. Shoulder-edge rumble strips are one example of a
countermeasure that can be used to potentially reduce run-off-road crashes. They provide strong
auditory and tactile feedback to drivers whose cars drift off the road because of inattention or
impairment.

Moving or Stopped Vehicles Ahead

Mainline crashes can also occur when drivers encounter slow-moving or stopped vehicles which,
except in congested traffic, are in a freeway through lane. Drivers’ limitations in perceiving closing
speed result in a short time to respond once the driver realizes the rapidity of the closure.
Alternatively, drivers may be visually attending to the vehicle directly ahead of them and may not
notice lane changes occurring beyond. If the lead driver is the first to encounter the stopped vehicle,
realizes the situation just in time, and moves rapidly out of the lane, the stopped vehicle is
uncovered at the last second, leaving the following driver with little time to respond.

Animals in the Road

Another common mainline crash type is with animals, particularly at night. Such crashes may
occur because an animal enters the road immediately in front of the driver leaving little or no time
for the driver to detect or avoid it. Low conspicuity of animals is also a problem. Given the
similarity in coloring and reflectance between pedestrians and animals, the same driver limitations
can be expected to apply to animals as to pedestrians in dark clothing. Based on data collected for
pedestrian targets, the majority of drivers traveling at speeds much greater than 30 mph and with
low-beam headlights would not be able to detect an animal in time to stop.

Undivided Roadways
Undivided roadways vary greatly in design and therefore in driver workload and perceived risk.
Some undivided roadways may have large-radius curves, mostly level grades, paved shoulders,
and wide clear zones. On such roads, and in low levels of traffic, the driving task can be very
undemanding, resulting in monotony and, in turn, possibly driver inattention and/or sleepiness. On
the other hand, undivided roadways may be very challenging in design, with tight curves, steep
grades, little or no shoulder, and no clear zone. In this case the driving task is considerably more
demanding.

Driver Inattention and Sleepiness

As described previously for the controlled-access mainline, inadvertent lane departures can result
when drivers are inattentive, impaired by alcohol or drugs, or sleepy. On an undivided highway,
these problems lead to run-off-road and head-on crashes. Rumble strips are effective in alerting
drivers about to leave the lane, and have been shown to be effective in reducing run-off-road and
cross-centerline crashes, respectively.

36
Inadvertent Movement into Oncoming Lane
The vast majority of head-on crashes occur due to inadvertent movement into the oncoming lane.
Contrary to some expectations, only about 4 percent of head-on crashes are related to overtaking.
Centerline rumble strips are very effective in reducing such crashes as they alert inattentive and
sleepy drivers. Although overtaking crashes are infrequent, they have a much higher risk of injury
and fatality than other crashes. As discussed previously, drivers are very limited in their ability to
perceive their closing speed to oncoming traffic. They tend to select gaps based more on distance
than on speed, leading to inadequate gaps when the oncoming vehicle is traveling substantially
faster than the speed limit. Passing lanes and four-lane passing sections greatly alleviate driver
workload and the risk of error involved in passing.

On roads with demanding geometry, driver speed choice when entering curves may be
inappropriate, leading to run-off-road crashes. Treatments which improve delineation are often
applied under the assumption that run-off-road crashes occur because the driver did not have
adequate information about the direction of the road path. However, studies have not supported
this assumption. Delineation that is purely auditory (i.e., rumble strips) does not have this negative
side- effect. Delineation that is purely auditory (i.e., rumble strips) does not have this negative
side-effect.

Slow-Moving or Stopped Vehicles Ahead

For the controlled-access mainline, rear-end and sideswipe crashes occur when drivers encounter
unexpected slowing or stopped vehicles and realize too late their closing speed.

Poor Visibility of Vulnerable Road Users or Animals

Vulnerable road user and animal crashes occur due to low contrast with the background and
drivers’ inability to detect pedestrians, cyclists, or animals in time to stop.

Crash Prevention
Crash preventive measures are classified into three categories: engineering, enforcement and
education.

1. Engineering measures
Road design: Geometric elements of the road (sight distance, width of the pavement, horizontal
and vertical alignment design details and intersections design elements) are checked and corrected
if necessary. Pavement surface characteristics (including skid resistance values) are checked and
suitable maintenance measures are taken to bring them up the design standards. Where necessary
construction of by-passes to separate through traffic from local traffic, design and construction of
grade separated intersections or flyovers to minimize delay and conflicts.
Maintenance of vehicles: Periodic maintenance of brakes, steering and lighting system of vehicles
plying on the road should be carried out and heavy penalties levied for defective vehicles.
Road lighting: Accident in night can be decreased by proper road lighting. Lighting is particularly
desirable at intersections, bridge sites and places where there are restrictions to traffic movements.

37
2. Enforcement measures
Speed control: To enable drivers of vehicles to develop correct speed habits. Also surprise check
on spot speed of vehicles should be done at selected locations and timings and legal action on those
who violates the speed limit should be taken.
Training and supervision: Strict test while issuing of driving license, testing of driving skill while
renewing.
Medical check: Test of vision and reaction time at prescribed intervals.
Observation of law and regulations: This is one of the most essential steps in enforcement for
prevention of accidents.

3. Education measures
Road users: It is very essential to educate road users for the various precautionary measures to use
the road way facilities with safety. The passengers and pedestrians should be taught the rules of
road, correct manner of crossings this may be possible by introducing necessary instructions in the
schools and poster exhibition and television showing serious results due to carelessness of road
users may also be useful.
Safety drive: Imposing traffic safety week when the road users are properly directed by the help
of traffic police and transport staff is the common means of training the public these days. Road
users should be impressed of what should be or what should not be done with the help of film ads
and documentaries. Proper driver training courses should be made available.

38
Chapter 3: Road Safety Data and Use
3.1 Importance of Crash and Road Data
The process of accurately investigating, analyzing and effectively treating crash locations relies
on the use of comprehensive and accurate crash data and data related to the road and traffic
characteristics at the crash locations. Comprehensive and accurate data enables the:
• Crash locations to be accurately determined
• Events associated with crashes to be identified
 identification of crash contribution and severity factors, thus providing the basis for
selecting targeted remedial treatment options
• identification of common factors across a number of crashes
• cost consequences of a single crash, all crashes at one location or several crashes with
common factors to be identified
• crash sites to be ranked so that treatment can be applied to those sites that will derive the
greatest safety benefits.

3.1.1 Minimum Reporting and Coding Criteria

There is a minimum set of data about each crash which is necessary as a basis for the sound and
satisfactory identification and investigation of a crash location. A reasonable knowledge of crash
data definitions and limitations is required to accurately interpret this information in any given
jurisdiction.
There is a high level of reporting of the most severe injury crashes and a lower level and variable
amount of reporting of lower cost crashes.

From a road safety perspective casualty data is the key indicator of the road safety problem. More
recently, and in close alignment to Safe System principles and objectives, fatal and serious injury
crashes and casualties are used as road safety key performance indicators. Lower severity
outcomes as well as property damage data (where this is available) can provide valuable additional
data that can support proposed countermeasure treatments.

The different crash severities are generally defined as follows:

• fatal crashes (one or more persons killed or died within 30 days)
• serious injury crashes (one or more persons admitted to hospital, although this is more
typically based on whether a person was injured and taken by ambulance to hospital)
• minor injury crashes (one or more persons injured who is not admitted to hospital, although
this is more typically based on a person not being taken by ambulance to hospital but
requiring medical treatment)
• non-injury crashes above threshold values which may vary across jurisdiction, plus those
where the property owner is not present.

3.1.2 Purposes of data collection

Road crash information is used by a wide variety of people for a wide variety of purposes
including:
• road safety practitioners, for developing remedial or pro-active road and traffic measures

39
 • police, who may be investigating whether they will charge a person with a criminal offence
in relation to a specific crash
• hospitals and health centres to monitor their health service requirements
• lawyers acting for clients in civil litigation, especially compensation for injuries and other
losses
• insurers, seeking facts before settling an insurance claim
• those with responsibility for road safety education and publicity, to ensure that their efforts
are well-targeted
• media
• police, in relation to enforcement activities, such as establishing the location for speed
cameras or breath testing stations
• safety administrators, exercising a duty to report statistical information on road crashes
• researchers, who need access to an accurate, reliable data base in order to conduct rigorous
research projects
• vehicle and component manufacturers and suppliers of highway materials, who wish to
assess the safety of their product, perhaps from a viewpoint of litigation, marketing or
product enhancement.

As the primary purposes for data collection within different organizations vary, the information
collected, the way in which data bases are established, the opportunities for data aggregation and
analysis, and interpretation of information will vary; the opportunities to supplement information
on one data base with information from another will be severely limited. For example, the data
base of a motor vehicle insurance company may not code crash location by a numerical geographic
system, as this information is peripheral to concerns of insurance claim assessment and financial
management. The total information on the data base is then not in a format which can be used by
someone seeking information about the safety performance of a particular location.

Primary data sources for crash reduction programs

The primary source of road crash information is the police crash report form. However these crash
outcomes are only known after the event, so police will sometimes attend relatively minor crashes
and (less often) will not attend some serious crashes.

 the police data base

 the police report form (which can provide a greater amount of detail regarding the crash
circumstances, for detailed analysis of individual sites; reference to police crash report
form details is usually necessary if a collision diagram is to be prepared).

Other data sources

Traffic data, such as traffic volumes (including turning volumes), pedestrian flows and vehicle
speeds may be helpful, depending upon the particular circumstances and problems at the site. In
some cases these will be available, but in other cases they may need to be collected.

Hospitals record the causes of injuries as well as their nature, extent and treatment. Crash data
sourced from police information may indicate whether someone was taken by ambulance or not,
but typically does not give information as to whether that person was hospitalised, for how long,
or whether the injuries sustained resulted in long-term impairment. With advances in technology,

40
and greater collaboration between agencies, it is possible to link this information together, allowing
for road safety treatments to be focused on those resulting in the most debilitating injuries. In
special circumstances, provided confidentiality of patient information is secure, hospital data may
be made available for research purposes. This has been done with good effect in the development
of countermeasures to reduce the severity of road crash injuries.

Insurance companies require claimants against policies to provide a description of the

circumstances in which the loss, damage or injury occurred. This information is not usually
released, although it has been released as consolidated data by some insurers. It provides a potential
for establishing the true extent of lower cost crashes which are not reported to the police.
Unfortunately it is generally not in a form which is compatible with the needs of crash location
analysis and treatment.

A further source of information on crash occurrences is tow truck operators’ records, although
there is no system in place for collecting this information.

From time to time in-depth crash studies are undertaken into the nature and causes of crashes in a
particular area. These studies are costly to undertake and involve specialist teams attending crash
scenes and taking measurements and recording crash features. The results are usually published in
special reports.

Local knowledge is an important source of information about safety problems on the road network.
Subjective information about crash problems must be regarded cautiously, but it can be a pointer
to problems or prompt further investigation. Information sources can include local residents,
businesses, safety groups, emergency service personnel, local medical practitioners, maintenance
contractors and local authority staff.

Interviews of road users, including people who have been involved in a crash at a site of interest,
in a structured format can be used to gain information for the development of crash
countermeasures.

Traffic conflict surveys may be used where the collection of crash data is not practical. These
involve field observations or video recording of conflicts (near misses).

Site investigations are a necessary component of any countermeasure development program and
will often yield insights into the crash history at a site.

Speed survey data also provides a source of information regarding speeding behaviour.

Technology Available for Data Collection

Computer based technology is being used to improve the accuracy of data collection in developed
countries. To improve the accuracy of location information Global positioning systems (GPS) or
satellite navigation systems are being used for accurate determination of a crash location. The
person attending the crash scene uses the system instead of, or as well as, documenting the location
in traditional terms (ABC Road, xx meters N/S/E/W of XYZ Street). This method has great
potential in rural areas where recording of the distance and direction to identifiable features can be
subject to significant error.

41
It is also now possible to use a geographical information system (GIS) or digital mapping to record
crash locations. This permits crash data to be incorporated within a relational data base, allowing
crash sites to be overlaid on plans showing other geographical information such as highway
features, traffic flows, intersection layouts and land uses.

New technology makes the initial collection and assessment of safety-related data easier and more
useful. As an example, in Main Roads Western Australia, the crash investigation team has recently
begun using video to assist in fatal and serious crash investigations. The video camera used records
GPS information and has a viewer that overlays all route information on the video display. This
assists in the completion of initial crash investigations. Furthermore, research is currently
underway on incorporating inventory data e.g. road geometry data and crash data with the video
information. This is to assist high risk crash route assessments for the identification of proposed
crash mitigation measures. This research is in its early stages of development.

To improve the accuracy and completeness of crash data

Menu-driven crash data capture programs can be used with laptop computers or tablets by police
attending a crash to ensure that all desired information is collected at the site. These programs can
include in-built logic and consistency checks on the data as it is entered.

Crash report forms can be arranged so that the information can be scanned into the data base, to
minimize costs and to reduce the opportunity for coding errors.

Accuracy of Crash Data

It is to know about the limitations of the data and take steps to resolve any anomalies which may
occur. The limitations include:

Under-reporting of crash data – although significant attempts are made to collect and record all
relevant crash data, not all non-fatal crashes make their way to the relevant crash data base. Surveys
conducted in New Zealand and in some Australian states have tried to compare data from crash
data bases with information from hospital admissions and other sources. Information from New
Zealand indicates only around 60% of serious crashes make it to the crash data base, with the
percentage significantly less for minor injury crashes (Alsop & Langley 2001). Australian based
research has also identified similar under-reporting of injury crashes (Cercarelli 1998). A review
conducted for Austroads (2005a) identified that reporting rates also varied by type of crash. For
example, reporting rates were lower for cyclists, pedestrians and motorcyclists.

Systematic reporting bias – this bias can result from the regulations or policies covering the
reporting of crashes. Reporting criteria vary between jurisdictions, resulting in crash experience
not being comparable. Numerically, property damage (non-injury) crashes constitute the bulk of
crashes: there may be as many as forty-one non-injury crashes for every casualty crash (James
1983). However, in some jurisdictions they are not reported, or are not recorded in the data base.
This results in an incomplete and systematically biased crash picture.

Random reporting bias – it is well established that crashes involving children, cyclists, pedestrians
and minor injury are substantially under-reported (James 1991). A similar situation applies to
crashes involving illegal activity, such as under-age driving, driving while intoxicated, riding a
42
motorcycle exceeding regulated capacity and carrying a pillion passenger when not permitted to
do so by regulation.

Further, it is common for some human factors (e.g. alcohol and drugs) and roadway factors (e.g.
the presence of a roadside culvert) not to be recorded. The absence of this information on the crash
report form may mean the absence of the factor or the failure to record it. Erroneous conclusions
can be made from the wrong interpretation of this absence of data.

Subjective bias – some crash forms require an assessment of possible contributing causes of the
crash. This in itself adds a subjective element, as the range of possible responses to the question
will be affected by the recorder’s experiences and the purposes (other than crash recording) to
which the information may be put. For example ‘failure to give way’ may be seen as a cause by
someone regularly involved in traffic law enforcement, whereas the same situation may be seen as
‘control device not visible’ by someone regularly involved with road environment safety. One
examination of the frequency of reported causes revealed distinct differences between police
districts (Vincent 1996). Similarly, speed and fatigue are not typically based on direct observation.

Reporting errors – it is important to recognise the circumstances under which a police officer
obtains information to complete a crash report. There will often be more pressing matters at a crash
scene. The officer may not have local knowledge or adequate training in incident investigation, so
some data items may be inadequately or wrongly recorded. Crashes do not always fit ‘standard’
formats and there may not be the motivation to fill in the form.

Coding errors – these can occur throughout the process from filling out the crash report form to
the data entry at the computer terminal. It is estimated that errors of this type are present in 5% of
crash files (Ogden 1996). They are unlikely to be revealed unless the data are used for detailed
investigation at individual sites. Typical problems include wrong direction for the north point,
wrong direction for one of two vehicle movements, selecting the wrong road user movement (DCA
code), for example ‘rear-end’ instead of ‘rear end into right turner’, and numerical coding errors.

Location errors – the location may be imprecise or wrong in the original police report form and
this will be carried through into the data base. If this continues through into coding, crashes at one
location may appear in two separate parts of the crash data base. The location reference system
may also be imprecise, so that a user of the data may not be able to accurately determine the
location (e.g. all mid-block crashes may be recorded as being half way between the adjacent
intersections).

Discontinuities over time – definitions or interpretations of field data may be changed over time
by those responsible for coding and reporting, so that data from one time period cannot be
compared with that of another. An abrupt change in recorded crash experience at a site should lead
an analyst to enquire whether there has been any discontinuity of this kind.

Delays – agencies responsible for data processing may not be sufficiently resourced: it may be
many months before information is available for analysis. Data may only be released annually.
This means that countermeasure development may be responding to historical crash patterns which
may be out of date.

43
Masked or hidden problem: it may be the case that a location is perceived as being so dangerous
that people avoid using it. In this situation the safety problem results in a reduction of amenity
(e.g. as pedestrians choose to cross the road somewhere regarded as safer) rather than resulting in
crashes.

Crash frequency is defined as the number of crashes occurring at a particular site, facility or
network in a one-year period. Crash frequency is calculated according to Equation below and is
measured in number of crashes per year.
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝐶𝑟𝑎𝑠ℎ𝑒𝑠
𝐶𝑟𝑎𝑠ℎ 𝐹𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 =
𝑃𝑒𝑟𝑖𝑜𝑑 𝑖𝑛 𝑌𝑒𝑎𝑟𝑠

Crash estimation refers to any methodology used to forecast or predict the crash frequency of:
 An existing roadway for existing conditions during a past or future period;
 An existing roadway for alternative conditions during a past or future period;
 A new roadway for given conditions for a future period.

Predictive method is to estimate the “expected average crash frequency” of a site, facility or
roadway under given geometric design, traffic volumes and for a specific period of time.

Expected average crash frequency is to describe the estimate of long-term average crash
frequency of a site, facility or network under a given set of geometric design and traffic volumes
in a given time period (in years).

As crashes are random events, the observed crash frequencies at a given site naturally fluctuate
over time. Therefore, the observed crash frequency over a short period is not a reliable indicator
of what average crash frequency is expected under the same conditions over a longer period of
time.

If all conditions on a roadway could be controlled (e.g. fixed traffic volume, unchanged geometric
design, etc), the long-term average crash frequency could be measured. However because it is
rarely possible to achieve these constant conditions, the true long-term average crash frequency is
unknown and must be estimated instead.

Crashes vary in the level of injury or property damage. The American National Standard ANSI
D16.1-1996 defines injury as “bodily harm to a person”. The level of injury or property damage
due to a crash is referred to in the HSM as “crash severity.” While a crash may cause a number of
injuries of varying severity, the term crash severity refers to the most severe injury caused by a
crash.

Crash severity is often divided into categories according to the KABCO scale, which provides
five levels of injury severity. Even if the KABCO scale is used, the definition of an injury may
vary between jurisdictions. The five KABCO crash severity levels are:

o K - Fatal injury: an injury that results in death;

44
 o A - Incapacitating injury: any injury, other than a fatal injury, which prevents the injured
person from walking, driving or normally continuing the activities the person was capable
of performing before the injury occurred;
o B – Non-incapacitating evident injury: any injury, other than a fatal injury or an
incapacitating injury, which is evident to observers at the scene of the accident in which
the injury occurred;
o C - Possible injury: any injury reported or claimed which is not a fatal injury, incapacitating
injury or non-incapacitating evident injury and includes claim of injuries not evident;
o – No Injury/Property Damage Only (PDO).

Crash evaluation refers to determining the effectiveness of a particular treatment or a treatment

program after its implementation. Where the term effectiveness is used, it refers to a change in the
expected average crash frequency (or severity) for a site or project. Evaluation is based on
comparing results obtained from crash estimation. Examples include:
o Evaluating a single application of a treatment to document its effectiveness;
o Evaluating a group of similar projects to document the effectiveness of those projects;
o Evaluating a group of similar projects for the specific purpose of quantifying the
effectiveness of a countermeasure;
o Assessing the overall effectiveness of specific projects or countermeasures in comparison
to their costs.

Data Needed for Crash Analysis

Accurate, detailed crash data, roadway or intersection inventory data, and traffic volume data are
essential to undertake meaningful and statistically sound safety analysis. This data may include:
o Crash Data: The data elements in a crash report describe the overall characteristics of the
crash. While the specifics and level of detail of this data vary from state to state, in general,
the most basic crash data consist of crash location, date and time, crash severity and
collision type, and basic information about the roadway, vehicles and people involved.
o Facility Data: The roadway or intersection inventory data provide information about the
physical characteristics of the accident site. The most basic roadway inventory data
typically include roadway classification, number of lanes, length, and presence of medians
and shoulder width. Intersection inventories typically include road names, area type, and
traffic control and lane configurations.
o Traffic Volume Data: In most cases, the traffic volume data required for the methods in the
HSM are annual average daily traffic (AADT). Some organizations may use ADT (average
daily traffic) as precise data may not be available to determine AADT. If AADT data are
unavailable, ADT should be used to estimate AADT. Other data that may be used for crash
analysis includes intersection total entering vehicles (TEV), and vehicle-miles traveled
(VMT) on a roadway segment, which is a measure of segment length and traffic volume.
In some cases, additional volume data, such as pedestrian crossing counts or turning
movement volumes, may be necessary.

45
3.3 Crash Estimation Methods

The development of new crash estimation methods is associated not only with increasing
sophistication of the statistical techniques, but is also due to changes in the thinking about road
safety. The following crash estimation methods are discussed:
o Crash estimation using observed crash frequency and crash rates over a short-term period;
o Crash estimation using observed crash frequency and crash rates over a long-term period
(e.g., more than 10 years);
o Indirect safety measures for identifying high crash locations. Indirect safety measures are
also known as surrogate measures;
o Statistical analysis techniques (specifically the development of statistical regression
models for estimation of crash frequency), and statistical methodologies to incorporate
observed crash data to improve the reliability of crash estimation models.

3.3.1 Observed Crash Frequency and Crash Rate Methods

Crash frequency and crash rates are often used for crash estimation and evaluation of treatment
effectiveness. The historic crash data on any facility (i.e., the number of recorded crashes in a
given period) is referred to as the “observed crash frequency”. “Crash rate” is the number of
crashes that occur at a given site during a certain time period in relation to a particular measure of
exposure (e.g., per million vehicle miles of travel for a roadway segment or per million entering
vehicles for an intersection). Crash rates may be interpreted as the probability (based on past
events) of being involved in an accident per instance of the exposure measure. For example, if the
crash rate on a roadway segment is one crash per one million vehicle miles per year, then a vehicle
has a one-in-a-million chance of being in an accident for every mile traveled on that roadway
segment. Crash rates are calculated according to equation below.

𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝐶𝑟𝑎𝑠ℎ 𝐹𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑖𝑛 𝑎 𝑃𝑒𝑟𝑖𝑜𝑑

𝐶𝑟𝑎𝑠ℎ 𝑅𝑎𝑡𝑒 =
𝐸𝑥𝑝𝑜𝑠𝑢𝑟𝑒 𝑖𝑛 𝑆𝑎𝑚𝑒 𝑃𝑒𝑟𝑖𝑜𝑑

Observed crash frequency and crash rates are often used as a tool to identify and prioritize sites in
need of modifications, and for evaluation of the effectiveness of treatments. Typically, those sites
with the highest crash rate or perhaps with rates higher than a certain threshold are analyzed in
more detail to identify potential modifications to reduce crashes. In addition, crash frequency and
crash rate are often used in conjunction with other analysis techniques, such as reviewing crash
records by year, collision type, crash severity, and/or environmental conditions to identify other
apparent trends or patterns over time. Advantages in the use of observed crash frequency and crash
rates include:
o Understandability –observed crash frequency and rates are intuitive to most members of
the public;
o Acceptance – it is intuitive for members of the public to assume that observed trends will
continue to occur;
o Limited alternatives – in the absence of any other available methodology, observed crash
frequency is the only available method of estimation.

46
Crash estimation methods based solely on historical crash data are subject to a number of
limitations. These include the limitations associated with the collection of data. Also, the use of
crash rate incorrectly assumes a linear relationship between crash frequency and the measure of
exposure. Research has confirmed that while there are often strong relationships between crashes
and many measures of exposure, these relationships are usually non-linear.

A (theoretical) example which illustrates how crash rates can be misleading is to consider a rural
two-lane two-way road with low traffic volumes with a very low observed crash frequency.
Additional development may substantially increase the traffic volumes and consequently the
number of crashes. However, it is likely that the crash rate may decline because the increased
traffic volumes. For example the traffic volumes may increase threefold, but the observed crash
frequency may only double, leading to a one third reduction in crash rate. If this change isn’t
accounted for, one might assume that the new development made the roadway safer.

Not accounting for the limitations described above may result in ineffective use of limited safety
funding. Further, estimating crash conditions based solely on observed crash data limits crash
estimation to the expected average crash frequency of an existing site where conditions (and traffic
volumes) are likely to remain constant for a long-term period, which is rarely the case. This
precludes the ability to estimate the expected average crash frequency for:
o The existing system under different geometric design or traffic volumes in the past
(considering if a treatment had not been implemented) or in the future (in considering
alternative treatment designs);
o Design alternatives of roadways that have not been constructed.

As the number of years of available crash data increases the risk of issues associated with
regression-to-the-mean bias decrease. Therefore, in situations where crashes are extremely rare
(e.g., at rail-grade crossings) observed crash frequency or crash rates may reliably estimate
expected average crash frequency and therefore can be used as a comparative value for ranking .

Even when there have been limited changes at a site (e.g., traffic volume, land use, weather, driver
demographics have remained constant) other limitations relating to changing contributing factors
remain. For example the use of motorcycles may have increased across the network during the
study period. An increase in observed motorcycle crashes at the site may be associated with the
overall change in levels of motorcycle use across the network rather than in increase in motorcycle
crashes at the specific site.

3.3.2 Crash Estimation using Statistical Methods

Statistical models using regression analysis have been developed which address some of the
limitations of other methods identified above. These models address RTM bias and also provide
the ability to reliably estimate expected average crash frequency for not only existing roadway
conditions, but also changes to existing conditions or a new roadway design prior to its
construction and use. As with all statistical methods used to make estimation, the reliability of the
model is partially a function of how well the model fits the original data and partially a function
of how well the model has been calibrated to local data. In addition to statistical models based on
crash data from a range of similar sites, the reliability of crash estimation is improved when historic
crash data for a specific site can be incorporated into the results of the model estimation.

47
A number of statistical methods exist for combining estimates of crashes from a statistical model
with the estimate using observed crash frequency at a site or facility. These include:
o Empirical Bayes method (EB Method)
o Hierarchical Bayes method
o Full Bayes method

The expected average crash frequency, Nexpected, is estimated using a predictive model estimate of
crash frequency, Npredicted (referred to as the predicted average crash frequency) and, where
available, observed crash frequency, Nobserved. The basic elements of the predictive method are:
 Predictive model estimate of the average crash frequency for a specific site type. This is
done using a statistical model developed from data for a number of similar sites. The model
is adjusted to account for specific site conditions and local conditions;
 The use of the EB Method to combine the estimation from the statistical model with
observed crash frequency at the specific site. A weighting factor is applied to the two
estimates to reflect the model’s statistical reliability. When observed crash data is not
available or applicable, the EB Method does not apply.

Basic Elements of the Predictive Models

The predictive models vary by facility and site type but all have the same basic elements:
o Safety Performance Functions (SPFs): statistical ”base” models are used to estimate the
average crash frequency for a facility type with specified base conditions.
o Accident Modification Factors (AMFs): AMFs are the ratio of the effectiveness of one
condition in comparison to another condition. AMFs are multiplied with the crash
frequency predicted by the SPF to account for the difference between site conditions and
specified base conditions;
o Calibration factor: multiplied with the crash frequency predicted by the SPF to account for
differences between the jurisdiction and time period for which the predictive models were
developed and the jurisdiction and time period to which they are applied.

While the functional form of the SPFs varies in the HSM, the predictive model to estimate the
expected average crash frequency Npredicted, is generally calculated using equation below.
𝑁𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑒𝑑 = 𝑁𝑆𝑃𝐹 𝑥 × (𝐴𝑀𝐹1𝑥 × 𝐴𝑀𝐹2𝑥 × … … × 𝐴𝑀𝐹𝑦𝑥 ) × 𝐶𝑥
Where
𝑁𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑒𝑑 = Predictive model estimation of crash frequency for a specific year on site type 𝑥
(crashes/year)
𝑁𝑆𝑃𝐹 𝑥 = Predicted average crash frequency determined for base conditions with the Safety
Performance Function representing site type 𝑥 (crashes/year)
𝐴𝑀𝐹𝑦𝑥 = Accident Modification Factors specific to site type 𝑥
𝐶𝑥 = Calibration factor to adjust for local conditions for site type 𝑥

Advantages of the Predictive Method

Advantages of the predictive method are that:

48
  Regression-to-the-mean bias is addressed as the method concentrates on long-term
expected average crash frequency rather than short-term observed crash frequency.
 Reliance on availability of limited crash data for any one site is reduced by incorporating
predictive relationships based on data from many similar sites.
 The method accounts for the fundamentally nonlinear relationship between crash
frequency and traffic volume.
 The SPFs in the HSM are based on the negative binomial distribution, which are better
suited to modeling the high natural variability of crash data than traditional modeling
techniques which are based on the normal distribution.

Safety Performance Functions

Safety Performance Functions (SPFs) are regression equations that estimate the average crash
frequency for a specific site type (with specified base conditions) as a function of annual average
daily traffic (AADT) and, in the case of roadway segments, the segment length (L). Base
conditions are specified for each SPF and may include conditions such as lane width, presence or
absence of lighting, presence of turn lanes etc. An example of a SPF (for roadway segments on
rural two-lane highways) is shown in below.
𝑁𝑆𝑃𝐹 𝑟𝑠 = (𝐴𝐴𝐷𝑇) × (𝐿) × (365) × 10(−6) × 𝑒 (−0.4865)
Where

𝑁𝑆𝑃𝐹 𝑟𝑠 = estimate of predicted average crash frequency for SPF base conditions for a rural two
lane two way roadway segment (crashes/year)
𝐴𝐴𝐷𝑇 = average annual daily traffic volume (vehicles per day) on roadway segment
𝐿 = length of roadway segment (miles)

While the SPFs estimate the average crash frequency for all crashes, the predictive method
provides procedures to separate the estimated crash frequency into components by crash severity
levels and collision types (such as run-off-road or rear-end crashes). In most instances, this is
accomplished with default distributions of crash severity level and/or collision type. As these
distributions will vary between jurisdictions, it is recommended that jurisdictions update these
defaults using their own data for crash severity levels and collision types. If sufficient experience
exists within an agency, some agencies have chosen to use advanced statistical approaches that
allow for prediction of changes by severity levels. The SPFs in the HSM have been developed for
three facility types (rural two-lane two-way roads, rural multilane highways, and urban and
suburban arterials), and for specific site types of each facility type (e.g. signalized intersections,
unsignalized intersections, divided roadway segments and undivided roadway segments).

Accident Modification Factors

Accident Modification Factors (AMFs) represent the relative change in crash frequency due to a
change in one specific condition (when all other conditions and site characteristics remain
constant). AMFs are the ratio of the crash frequency of a site under two different conditions
Therefore, an AMF may serve as an estimate of the effect of a particular geometric design or traffic
control feature or the effectiveness of a particular treatment or condition.

AMFs are generally presented for the implementation of a particular treatment, also known as a
countermeasure, intervention, action, or alternative design. Examples include illuminating an
unlighted road segment, paving gravel shoulders, signalizing a stop-controlled intersection, or

49
choosing a signal cycle time of 70 seconds instead of 80 seconds. AMFs have also been developed
for conditions that are not associated with the roadway, but represent geographic or demographic
conditions surrounding the site or with users of the site (e.g., the number of liquor outlets in
proximity to the site). Equation below shows the calculation of an AMF for the change in expected
average crash frequency from site condition ‘a’ to site condition ‘b’.
𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑐𝑟𝑎𝑠ℎ 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑤𝑖𝑡ℎ 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛 ′𝑏′
𝐴𝑀𝐹 =
𝐸𝑥𝑝𝑒𝑐𝑡𝑒𝑑 𝑎𝑣𝑒𝑟𝑎𝑔𝑒 𝑐𝑟𝑎𝑠ℎ 𝑓𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑤𝑖𝑡ℎ 𝑐𝑜𝑛𝑑𝑖𝑡𝑖𝑜𝑛 ′𝑎′

AMFs defined in this way for expected crashes can also be applied to comparison of predicted
crashes between site conditions ‘a’ and site condition ‘b’.

The values of AMFs are determined for a specified set of base conditions. These base conditions
serve the role of site condition ‘a’ in Equation above. This allows comparison of treatment options
against a specified reference condition. Under the base conditions (i.e., with no change in the
conditions), the value of an AMF is 1.00. AMF values less than 1.00 indicate the alternative
treatment reduces the estimated average crash frequency in comparison to the base condition. AMF
values greater than 1.00 indicate the alternative treatment increases the estimated average crash
frequency in comparison to the base condition. The relationship between an AMF and the expected
percent change in crash frequency is shown in equation below.
𝑃𝑒𝑟𝑐𝑒𝑛𝑡 𝑅𝑒𝑑𝑢𝑐𝑡𝑖𝑜𝑛 𝑖𝑛 𝐴𝑐𝑐𝑖𝑑𝑒𝑛𝑡𝑠 = 100 × (1.00 − 𝐴𝑀𝐹)
For example,
 If an AMF = 0.90 then the expected percent change is 100% × (1.00 − 0.9) = 10%
indicating a reduction in expected average crash frequency
 If an AMP = 1.2 then the expected percent change is 100% × (1.00 − 1.2) = −20%
indicating an increase in expected average crash frequency

50
Application of AMFs
Applications for AMFs include:
o Multiplying an AMF with a crash frequency for base conditions determined with a SPF to
estimate predicted average crash frequency for an individual site, which may consist of
existing conditions, alternative conditions or new site conditions. The AMFs are used to
account for the difference between the base conditions and actual site conditions;
o Multiplying an AMF with the expected average crash frequency of an existing site that is
being considered for treatment. This estimates expected average crash frequency of the
treated site. This approach is only recommended when a site-specific SPF applicable to the
treated site is not available. For example an AMF for a change in site type or conditions
such as the change from an unsignalized intersection to a roundabout can be used if no SPF
is available for the proposed site type or conditions;
o Multiplying an AMF with the observed crash frequency of an existing site that is being
considered for treatment to estimate the change in expected average crash frequency due
to application of a treatment. This approach is only recommended when site-specific SPFs
for both the existing and new roadway network are not available.

Applying Multiple AMFs

The predictive method assumes that AMFs can be multiplied together to estimate the combined
effects of the respective elements or treatments. This approach assumes that the individual
elements or treatments considered in the analysis are independent of one another. Limited research
exists regarding the independence of individual treatments from one another. AMFs are
multiplicative even when a treatment can be implemented to various degrees such that a treatment
is applied several times over. For example, a 4% grade can be decreased to 3%, 2%, and so on, or
a 6-foot shoulder can be widened by 1-ft, 2- ft, and so on. When consecutive increments have the
same degree of effect, Equation below can be applied to determine the treatment’s cumulative
effect.
𝐴𝑀𝐹(𝑓𝑜𝑟 𝑛 𝑖𝑛𝑐𝑟𝑒𝑚𝑒𝑛𝑡𝑠) = [𝐴𝑀𝐹(𝑓𝑜𝑟 𝑜𝑛𝑒 𝑖𝑛𝑐𝑟𝑒𝑚𝑒𝑛𝑡)]𝑛

51
Weighting using the Empirical Bayes Method
Estimation of expected average crash frequency using only observed crash frequency or only
estimation using a statistical model may result in a reasonable estimate of crash frequency.
However, as discussed, the statistical reliability (the probability that the estimate is correct) is
improved by combining observed crash frequency and the estimate of the average crash frequency
from a predictive model. While a number of statistical methods exist that can compensate for the
potential bias resulting from regression-to-the mean, the predictive method uses the empirical
Bayes method, herein referred to as the EB Method.

The EB Method uses a weight factor, which is a function of the SPF over dispersion parameter, to
combine the two estimates into a weighted average. The weighted adjustment is therefore
dependent only on the variance of the SPF, and is not dependent on the validity of the observed
crash data. The EB Method is only applicable when both predicted and observed crash frequencies
are available for the specific roadway network conditions for which the estimate is being made. It
can be used to estimate expected average crash frequency for both past and future periods. The EB
Method is applicable at either the site specific level (where crashes can be assigned to a particular
location) or the project specific level (where observed data may be known for a particular facility,
but cannot be assigned to the site specific level). Where only a predicted or only observed crash
data are available, the EB Method is not applicable (however the predictive method provides
alternative estimation methods in these cases). For an individual site the EB Method combines the
observed crash frequency with the statistical model estimate using equation below:
𝑁𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 = 𝑊 × 𝑁𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑒𝑑 + (1 − 𝑊) × 𝑁𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑

Where
𝑁𝑒𝑥𝑝𝑒𝑐𝑡𝑒𝑑 = expected average crashes frequency for the study period

52
𝑁𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑒𝑑 = predicted average crash frequency predicted using a SPF for the study period under
the given conditions
𝑊 = weighted adjustment to be placed on the SPF prediction
𝑁𝑜𝑏𝑠𝑒𝑟𝑣𝑒𝑑 = obverved crash frequency at the site over the study period

The weighted adjustment factor, w, is a function of the SPF’s overdispersion parameter, k, and is
calculated using equation below.
1
𝑊=
1 + 𝑘 × (∑𝑎𝑙𝑙 𝑠𝑡𝑢𝑑𝑦 𝑦𝑒𝑎𝑟𝑠 𝑁𝑝𝑟𝑒𝑑𝑖𝑐𝑡𝑒𝑑 )
Where
𝑘 = Overdispersion parameter from the associated SPF

As the value of the over dispersion parameter increases, the value of the weighted adjustment
factor decreases. Thus, more emphasis is placed on the observed rather than the predicted crash
frequency. When the data used to develop a model are greatly dispersed, the reliability of the
resulting predicted crash frequency is likely to be lower. In this case, it is reasonable to place less
weight on the predicted crash frequency and more weight on the observed crash frequency. On the
other hand, when the data used to develop a model have little over dispersion, the reliability of the
resulting SPF is likely to be higher. In this case, it is reasonable to place more weight on the
predicted crash frequency and less weight on the observed crash frequency.

3.4 Network Screening

3.4.1 Introduction
Network screening is a process for reviewing a transportation network to identify and rank sites
from most likely to least likely to realize a reduction in crash frequency with implementation of a
countermeasure. Those sites identified as most likely to realize a reduction in crash frequency are
studied in more detail to identify crash patterns, contributing factors, and appropriate
countermeasures. Network screening can also be used to formulate and implement a policy, such
as prioritizing the replacement of non-standard guardrail statewide at sites with a high number of
run-off-the-road crashes.

3.4.2 Network Screening Process

There are five major steps in network screening:
1. Establish Focus: Identify the purpose or intended outcome of the network screening
analysis. This decision will influence data needs, the selection of performance measures
and the screening methods which can be applied.
2. Identify Network and Establish Reference Populations: Specify the type of sites or facilities
being screened (i.e., segments, intersections, at-grade rail crossings) and identify groupings
of similar sites or facilities.
3. Select Performance Measures: There are a variety of performance measures available to
evaluate the potential to reduce crash frequency at a site. In this step the performance
measure is selected as a function of the screening focus and the data and analytical tools
available.

53
 4. Select Screening Method: There are three principle screening methods described in this
chapter (i.e., ranking, sliding window, and peak searching). The advantages and
disadvantages of each are described in order to help identify the most appropriate method
for a given situation.
5. Screen and Evaluate Results: The final step in the process is to conduct the screening
analysis and evaluate results.

STEP 1 - Establish the Focus of Network Screening

The first step in network screening is to establish the focus of the analysis. Network screening can
be conducted and focused on one or both of the following:
1. Identifying and ranking sites where improvements have potential to reduce the number of
crashes; and/or,
2. Evaluating a network to identify sites with a particular crash type or severity in order to
formulate and implement a policy (e.g., identify sites with a high number of run-off-the-
road crashes to prioritize the replacement of non standard guardrail statewide).

If network screening is being applied to identify sites where modifications could reduce the
number of crashes, the performance measures are applied to all sites. Based on the results of the
analysis, those sites that show potential for improvement are identified for additional analysis. This
analysis is similar to a typical “black spot” analysis conducted by a jurisdiction to identify the
“high crash locations.”

A transportation network can also be evaluated to identify sites which have potential to benefit
from a specific program (e.g., increased enforcement) or countermeasure (e.g., a guard-rail
implementation program). An analysis such as this might identify locations with a high proportion
or average frequency of a specific crash type or severity. In this case a subset of the sites is studied.

STEP 2 - Identify the Network and Establish Reference Populations

The focus of the network screening process established in Step 1 forms the basis for the second
step in the network screening process, which includes identifying the network elements to be
screened and organizing these elements into reference populations. Examples of roadway network
elements that can be screened include intersections, roadway segments, facilities, ramps, ramp
terminal intersections, and at-grade rail crossings.

54
A reference population is a grouping of sites with similar characteristics (e.g. four-legged
signalized intersections, two-lane rural highways). Ultimately prioritization of individual sites is
made within a reference population. In some cases, the performance measures allow comparisons
across reference populations. The characteristics used to establish reference populations for
intersections and roadway segments are identified in the following sections.

Intersection Reference Populations

Potential characteristics that can be used to establish reference populations for intersections
include:
 Traffic control (e.g., signalized, two-way or four-way stop control, yield control,
roundabout);
 Number of approaches (e.g., three-leg or four-leg intersections);
 Cross-section (e.g., number of through lanes and turning lanes);
 Functional classification (e.g., arterial, collector, local);
 Area type (e.g., urban, suburban, rural);
 Traffic volume ranges (e.g., total entering volume (TEV), peak hour volumes, average
annual daily traffic(AADT)); and/or,
 Terrain (e.g., flat, rolling, mountainous)

The characteristics that define a reference population may vary depending on the amount of detail
known about each intersection, the purpose of the network screening, the size of the network being
screened, and the performance measure selected. Similar groupings are also applied if ramp
terminal intersections and/or at grade rail crossings are being screened.

Segment Reference Populations

A roadway segment is a portion of a facility that has a consistent roadway cross section and is
defined by two endpoints. These endpoints can be two intersections, on- or off-ramps, a change in
roadway cross-section, mile markers or mile posts, or a change in any of the roadway
characteristics listed below.
Potential characteristics that can be used to define reference populations for roadway segments
include:
Number of lanes per direction;
 Access density (e.g., driveway and intersection spacing);
 Traffic volumes ranges (e.g., TEV, peak hour volumes, AADT);

55
  Median type and/or width;
 Operating speed or posted speed;
 Adjacent land use (e.g., urban, suburban, rural);
 Terrain (e.g., flat, rolling, mountainous); and,
 Functional classification (e.g., arterial, collector, local).
Other example roadway segment reference populations are: four-lane cross section with raised
concrete median; five-lane cross-section with a two-way, left-turn lane; or rural two-lane highway
in mountainous terrain. If ramps are being screened, groupings similar to these are also applied.

STEP 3 - Select Network Screening Performance Measures

The third step in the network screening process is to select one or several performance measures
to be used in evaluating the potential to reduce the number of crashes or crash severity at a site.
Just as intersection traffic operations analysis can be measured as a function of vehicle delay,
queue length, or a volume to-capacity ratio, intersection safety can be quantitatively measured in
terms of crash frequency, expected crash frequency, a critical crash rate, or several other
performance measures. In network screening using multiple performance measures to evaluate
each site may improve the level of confidence in the results.

Key Criteria for Selecting Performance Measures

The key considerations in selecting performance measures are: data availability, regression-to-the-
mean bias, and how the performance threshold is established.
Data and Input Availability

56
Typical data required for the screening analysis includes the facility information for establishing
reference populations, crash data, and traffic volume data and in some cases safety performance
functions. The amount of data and inputs that are available limits the number of performance
measures that can be used. If traffic volume data is not available or cost prohibitive to collect,
fewer performance measures are available for ranking sites. If traffic volumes are collected or
made available, but calibrated safety performance functions and over dispersion parameters are
not, the network could be prioritized using a different set of performance measures.

Regression-to-the-Mean Bias
Crash frequencies naturally fluctuate up and down over time at any given site. As a result, a short-
term average crash frequency may vary significantly from the long-term average crash frequency.
The randomness of accident occurrence indicates that short-term crash frequencies alone are not a
reliable estimator of long-term crash frequency. If a three-year period of crashes were to be used
as the sample to estimate crash frequency, it would be difficult to know if this three-year period
represents a high, average, or low crash frequency at the site compared to previous years.

When a period with a comparatively high crash frequency is observed, it is statistically probable
that a lower crash frequency will be observed in the following period. This tendency is known as
regression-to-the-mean (RTM), and also applies to the statistical probability that a comparatively
low crash frequency period will be followed by a higher crash frequency period.

Failure to account for the effects of RTM introduces the potential for “RTM bias”, also known as
“selection bias”. RTM bias occurs when sites are selected for treatment based on short-term trends
in observed crash frequency. For example, a site is selected for treatment based on a high observed
crash frequency during a very short period of time (e.g., two years). However, the site’s long-term
57
crash frequency may actually be substantially lower and therefore the treatment may have been
more cost effective at an alternate site.
Performance Threshold
A performance threshold value provides a reference point for comparison of performance measure
scores within a reference population. Sites can be grouped based on whether the estimated
performance measure score for each site is better or worse than the threshold value. Those sites
with a performance measure score worse than the threshold value can be studied in further detail
to determine if reduction in crash frequency or severity is possible.

The method for determining a threshold performance value is dependent on the performance
measure selected. The threshold performance value can be a subjectively assumed value, or
calculated as part of the performance measure methodology. For example, threshold values are
estimated based on: the average of the observed crash frequency for the reference population; an
appropriate safety performance function; or, empirical Bayes methods. Typically, the same
methods that require more data and can address RTM bias produce more reliable threshold
performance values.

Definition of Performance Measures

Crash Frequency

The site with the most total crashes or the most crashes of a particular crash severity or type, in a
given time period, is given the highest rank. The site with the second highest number of crashes in
total or of a particular crash severity or type, in the same time period, is ranked second, and so on.
The table below summarizes the strengths and limitations of the Crash Frequency performance
measure.

58
Equivalent Property Damage Only (EPDO) Crash Frequency
The Equivalent Property Damage Only (EPDO) Crash Frequency performance measure assigns
weighting factors to crashes by severity (fatal, injury, property damage only) to develop a
combined frequency and severity score per site. The weighting factors are often calculated relative
to Property Damage Only (PDO) crash costs. The crash costs by severity are summarized yielding
an EPDO value. Although some agencies have developed weighting methods based on measures
other than costs, crash costs are used consistently to demonstrate use of the performance measure.

Crash costs include direct and indirect costs. Direct costs could include: ambulance service, police
and fire services, property damage, or insurance. Indirect costs include the value society would
place on pain and suffering or loss of life associated with the crash.
The table below summarizes the strengths and limitations of the EPDO Crash Frequency
performance measure.

Probability of Specific Crash Types Exceeding Threshold Proportion

Sites are prioritized based on the probability that the true proportion, pi, of a particular crash type
or severity (e.g., long-term predicted proportion) is greater than the threshold proportion, p*i A
threshold proportion (p*i) is selected for each population, typically based on the proportion of the
target crash type or severity in the reference population. This method can also be applied as a
diagnostic tool to identify crash patterns at an intersection or on a roadway segment.

The table below summarizes the strengths and limitations of the Probability of Specific Crash
Types Exceeding Threshold Proportion performance measure.
Strengths Limitations
Can also be used as a Does not account for RTM bias
diagnostic tool
Considers variance in data Does not account for traffic volume
Some sites may be identified for further study because of unusually low frequency of
non-target crash types

Excess Proportions of Specific Crash Types

This performance measure is very similar to the Probability of Specific Crash Types Exceeding
Threshold Proportion performance measure except sites are prioritized based on the excess
proportion. The excess proportion is the difference between the observed proportion of a specific
collision type or severity and the threshold proportion from the reference population. A threshold
proportion (p*i) is selected for each population, typically based on the proportion of the target

59
crash type or severity in the reference population. The largest excess value represents the most
potential for reduction in average crash frequency. This method can also be applied as a diagnostic
tool to identify crash patterns at an intersection or on a roadway segment.

The table below summarizes the strengths and limitations of the Excess Proportions of Specific
Crash Types performance measure.

Relative Severity Index

Monetary crash costs are assigned to each crash type and the total cost of all crashes is calculated
for each site. An average crash cost per site is then compared to an overall average crash cost for
the site’s reference population. The overall average crash cost is an average of the total costs at all
sites in the reference population. The resulting relative severity index (RSI) performance measure
shows whether a site is experiencing higher crash costs than the average for other sites with similar
characteristics.

The table below summarizes the strengths and limitations of the Relative Severity Index
performance measure.

Crash Rate
The crash rate performance measure normalizes the frequency of crashes with the exposure,
measured by traffic volume. When calculating a crash rate traffic volumes are reported as million
entering vehicles (MEV) per intersection for the study period. Roadway segment traffic volumes
are measured as vehicle-miles traveled (VMT) for the study period. The exposure on roadway
segments is often measured per million VMT.

The table below summarizes the strengths and limitations of the Crash Rate performance measure.

Critical Rate

60
The observed crash rate at each site is compared to a calculated critical crash rate that is unique to
each site. The critical crash rate is a threshold value that allows for a relative comparison among
sites with similar characteristics. Sites that exceed their respective critical rate are flagged for
further review. The critical crash rate depends on the average crash rate at similar sites, traffic
volume, and a statistical constant that represents a desired level of significance.

The table below summarizes the strengths and limitations of the Critical Rate performance
measure.

Excess Predicted Average Crash Frequency Using Method of Moments

A site’s observed average crash frequency is adjusted based on the variance in the crash data and
average crash frequency for the site’s reference population. The adjusted observed average crash
frequency for the site is compared to the average crash frequency for the reference population.
This comparison yields the potential for improvement which can serve as a measure for ranking
sites.

The table below summarizes the strengths and limitations of the Excess Predicted Average Crash
Frequency Using Method of Moments performance measure.

Level of Service of Safety (LOSS)

Sites are ranked according to a qualitative assessment in which the observed crash count is
compared to a predicted average crash frequency for the reference population under consideration.
Each site is placed into one of four LOSS classifications, depending on the degree to which the
observed average crash frequency is different than predicted average crash frequency. The
predicted average crash frequency for sites with similar characteristics is predicted from an SPF
calibrated to local conditions.

The table below summarizes the strengths and limitations of the LOSS performance measure.

61
Excess Predicted Crash Frequency Using Safety Performance Functions (SPFs)
The site’s observed average crash frequency is compared to a predicted average crash frequency
from a SPF. The difference between the observed and predicted crash frequencies is the excess
predicted crash frequency using SPFs. When the excess predicted average crash frequency is
greater than zero, a site experiences more crashes than predicted. When the excess predicted
average crash frequency value is less than zero, a site experiences less crashes than predicted.

The table below summarizes the strengths and limitations of the Excess Predicted Crash Frequency
Using SPFs performance measure.

Expected Average Crash Frequency with Empirical Bayes (EB) Adjustment

The observed average crash frequency and the predicted average crash frequency from a SPF are
weighted together using the EB method to calculate an expected average crash frequency that
accounts for RTM bias. Sites are ranked from high to low based on the expected average crash
frequency.

The table below summarizes the strengths and limitations of the Expected Average Crash
Frequency with Empirical Bayes (EB) Adjustment performance measure.

Equivalent Property Damage Only (EPDO) Crash Frequency with EB Adjustment

Crashes by severity are predicted using the EB procedure. The expected crashes by severity are
converted to EPDO crashes using the EPDO procedure. The resulting EPDO values are ranked.
The EPDO Crash Frequency with EB Adjustments measure accounts for RTM bias and traffic
volume.

The table below summarizes the strengths and limitations of the EPDO Crash Frequency with EB
Adjustment performance measure.

Excess Expected Average Crash Frequency with Empirical Bayes (EB) Adjustment

62
The observed average crash frequency and the predicted crash frequency from a SPF are weighted
together using the EB method to calculate an expected average crash frequency. The resulting
expected average crash frequency is compared to the predicted average crash frequency from a
SPF. The difference between the EB adjusted average crash frequency and the predicted average
crash frequency from a SPF is the excess expected average crash frequency.

When the excess expected crash frequency value is greater than zero, a site experiences more
crashes than expected. When the excess expected crash frequency value is less than zero, a site
experiences less crashes than expected.

The table below summarizes the strengths and limitations of the Excess Expected Average Crash
Frequency with Empirical Bayes (EB) Adjustment performance measure.

STEP 4 - Select Screening Method

The fourth step in the network screening process is to select a network screening method. In a
network screening process, the selected performance measure would be applied to all sites under
consideration using a screening method.

In the HSM, there are three types of three categories of screening methods:
 Segments (e.g., roadway segment or ramp) are screened using either sliding window or
peak searching methods.
 Nodes (e.g., intersections or ramp terminal intersections) are screened using simple ranking
method.
 Facilities (combination of nodes and segments) are screened using a combination of
segment and node screening methods.

Segment Screening Methods

63
Screening roadway segments and ramps requires identifying the location within the roadway
segment or ramp that is most likely to benefit from a countermeasure intended to result in a
reduction in crash frequency or severity. The location (i.e., sub segment) within a segment that
shows the most potential for improvement is used to specify the critical crash frequency of the
entire segment and subsequently select segments for further investigation. Having an
understanding of what portion of the roadway segment controls the segment’s critical crash
frequency will make it easier and more efficient to identify effective countermeasures. Sliding
window and peak searching methods can be used to identify the location within the segment which
is likely to benefit from a countermeasure. The simple ranking method can also be applied to
segments, but unlike sliding window and peak searching methods, performance measures are
calculated for the entire length of the segment.
Sliding Window Method
In the sliding window method a window of a specified length is conceptually moved along the
road segment from beginning to end in increments of a specified size. The performance measure
chosen to screen the segment is applied to each position of the window, and the results of the
analysis are recorded for each window. A window pertains to a given segment if at least some
portion of the window is within the boundaries of the segment. From all the windows that pertain
to a given segment, the window that shows the most potential for reduction in crash frequency out
of the whole segment is identified and is used to represent the potential for reduction in crash
frequency of the whole segment. After all segments are ranked according to the respective highest
sub-segment value, those segments with the greatest potential for reduction in crash frequency or
severity are studied in detail to identify potential countermeasures.

Windows will bridge two or more contiguous roadway segments in the sliding window method.
Each window is moved forward incrementally until it reaches the end of a contiguous set of
roadway segments. Discontinuities in contiguous roadway segments may occur as a result of
discontinuities in route type, mileposts or routes, site characteristics, etc. When the window nears
the end of a contiguous set of roadway segments, the window length remains the same, while the
increment length is adjusted so that the last window is positioned at the end of the roadway
segment.

In some instances the lengths of roadway segments may be less than the typical window length,
and the roadway segments may not be part of a contiguous set of roadway segments. In these
instances, the window length equals the length of the roadway segment.

64
Peak Searching Method
In the peak searching method each individual roadway segment is subdivided into windows of
similar length, potentially growing incrementally in length until the length of the window equals
the length of the entire roadway segment. The windows do not span multiple roadway segments.
For each window, the chosen performance measure is calculated. Based upon the statistical
precision of the performance measure, the window with the maximum value of the performance
measure within a roadway segment is used to rank the potential for reduction in crashes of that site
(i.e., whole roadway segment) relative to the other sites being screened.
The first step in the peak searching method is to divide a given roadway segment (or ramp) into
0.1 mile windows. The windows do not overlap, with the possible exception that the last window
may overlap with the previous. If the segment is less than 0.1 mile in length, then the segment
length equals the window length. The performance measure is then calculated for each window,

65
and the results are subjected to precision testing. If the performance measure calculation for at
least one sub-segment satisfies the desired precision level, the segment is ranked based upon the
maximum performance measure from all of the windows that meet the desired precision level. If
none of the performance measures for the initial 0.1 mile windows are found to have the desired
precision, the length of each window is incrementally moved forward; growing the windows to a
length of 0.2 mile. The calculations are performed again to assess the precision of the performance
measures. The methodology continues in this fashion until a maximum performance measure with
the desired precision is found or the window length equals the site length.

The precision of the performance measure is assessed by calculating the coefficient of variation
(CV) of the performance measure.

A large CV indicates a low level of precision in the estimate, and a small CV indicates a high level
of precision in the estimate. The calculated CV is compared to a specified limiting CV. If the
calculated CV is less than or equal to the CV limiting value, the performance measure meets the
desired precision level, and the performance measure for a given window can potentially be
considered for use in ranking the segment. If the calculated CV is greater than the CV limiting
value, the window is automatically removed from further consideration in potentially ranking the
segment based upon the value of the performance measure.

There is no specific CV value that is appropriate for all network screening applications. However,
by adjusting the CV value the user can vary the number of sites identified by network screening
as candidates for further investigation. An appropriate initial or default value for the CV is 0.5.

Iteration #1
Table below shows the results of the first iteration. In the first iteration, the site is divided into 0.1
mi windows. For each window, the performance measure is calculated along with the CV.

The Coefficient of Variation for Segment B1 is calculated using the equation as shown below:

66
Because none of the calculated CVs are less than the CV limiting value, none of the windows meet
the screening criterion, so a second iteration of the calculations is required.

Iteration #2
Table below shows the results of the second iteration. In the second iteration, the site is divided
into 0.2 mi windows. For each window, the performance measure is calculated along with the CV.

Simple Ranking Method

A simple ranking method can be applied to nodes and segments. In this method, the performance
measures are calculated for all of the sites under consideration, and the results are ordered from
high to low. The simplicity of this method is the greatest strength. However, for segments, the
results are not as reliable as the other segment screening methods.
Node-Based Screening

67
Node-based screening focuses on intersections, ramp terminal intersections, and at-grade rail
crossings. A simple ranking method may be applied whereby the performance measures are
calculated for each site, and the results are ordered from high to low. The outcome is a list showing
each site and the value of the selected performance measure. All of the performance measures can
be used with simple ranking for node-based screening.

A variation of the peak searching method can be applied to intersections. In this variation, the
precision test is applied to determine which performance measure to rank upon. Only intersection-
related crashes are included in the node-based screening analyses.
Facility Screening
A facility is a length of highway composed of connected roadway segments and intersections.
When screening facilities, the connected roadway segments are recommended to be approximately
5 to 10 miles in length. This length provides for more stable results.

The table below summarizes the performance measures that are consistent with the screening
methods.

STEP 5 - Screen and Evaluate Results

The performance measure and the screening method are applied to the segments, nodes, and/or
facilities according to the methods outlined in Steps 3 and 4. Conceptually, for each segment or
node under consideration, the selected performance measure is calculated and recorded. Results
can be recorded in a table or on maps as appropriate or feasible.

68
The results of the screening analysis will be a list of sites ordered according to the selected
performance measure. Those sites higher on the list are considered most likely to benefit from
countermeasures intended to reduce crash frequency. Further study of these sites will indicate what
kinds of improvements are likely to be most effective.

In general it can be useful to apply multiple performance measures to the same data set. In doing
so, some sites will repeatedly be at the high or low end of the resulting list. Sites that repeatedly
appear at the higher end of the list could become the focus of more detailed site investigations,
while those that appear at the low end of the list could be ruled out for needing further
investigation. Differences in the rankings produced by the various performance measures will
become most evident at sites which are ranked in the middle of the list.

69
Chapter 4: Road Safety Audit/Inspection
4.1 Definition and Importance

A road safety audit is a formal examination of a future road or traffic project or an existing road in
which an independent, qualified team reports on the project’s crash potential and safety
performance.
The essential elements of the definition are that it is:
 a formal process and not an informal check
 carried out by people who are independent of the design or the road authority if an existing
road
 carried out by people with appropriate experience and training
 restricted to road safety issues. The essential elements of the definition are that it is:

A road safety audit is not:

 a way of assessing or rating a project as good or poor
 a means of ranking or justifying one project against others in a works program
 a way of rating one option against another
 a check of compliance with standards
 a substitute for design checks
 a crash investigation
 a redesign of a project
 something to be applied only to high-cost projects or only to projects involving safety
problems
 the name used to describe informal checks, inspections or consultation.
A road safety audit has the greatest potential for improving safety and is most cost effective when
it is applied to a road or traffic design before the project is built. It can be conducted on any design
proposal that involves changes to the ways road users will interact either with each or with the
physical environment. It is a formal process using a defined procedure. It must be conducted by
people who are independent and who have appropriate experience and training.
Road safety audits do not alter the need for safety first approach of a design. The road safety audit
process provides for an independent assessment to be made by a team specifically skilled in the
areas of crash prevention and road safety engineering.
Rather than checking for compliance, a road safety audit is checking fitness for purpose: will the
road or treatment work safety for its expected road users? Auditors should also fully understand
the safe system and be able to integrate this approach, as appropriate into the audit process.
Audit Cost and benefits
The cost of road safety audit varies depending upon the stage of audit and project size. This sis
typically less than 5% of road design cost. Design costs can be about 5% of total implementation
cost.

70
The benefit of road safety audit range from direct improvements in a design to incorporating things
to safety policies. They may be
 Safer new roads through crash prevention and crash severity reduction
 Safer road networks
 More uniform road environment that is easily understood by the road users and becomes
self-explaining to road users.
 Reduce whole of life costs of road schemes
 Providing one component of local and national crash reduction targets
 A reduced need to modify new schemes after they are built
 A better understanding and documentation of road safety engineering
 Eventual safety improvements to standards and procedures
 More explicit consideration of the safety needs of vulnerable road users
 The encouragement of other personnel in road safety.
The completion of design and existing road safety audits will also result in many qualitative
benefits. In addition to the safety related benefits of proactively identifying and treating specific
hazards the other benefits of the audit process include:
 Identification of improved design, construction and maintenance standards that influence
safety performance on an ongoing basis
 The role the audit plays in improving the general road safety awareness of operational staff
 The role the audit plays in providing the designer/asset owner with confidence in the safety
performance of the proposed project or road network.
An important element of the audit process is the recognition that an audit with no deficiencies
identified will still retain significant value in providing assurance of safety.

71
4.2 Stages, Components, Process

There are typically four

opportunities within the design
and development process for a
road or traffic project when a road
safety audit can be conducted,
regardless of the size or nature of
the project:
 at the feasibility stage
 once the preliminary
design stage has been
developed
 once the detailed design
stage is complete
 at the pre-opening stage
(or soon after the project
is complete).

A road safety audit may also be

conducted:
 for roadwork traffic
management required
during construction of
significant projects
 on the existing road
network.

A road safety audit can also be

undertaken at any stage of a
project’s life cycle or in stages throughout the design or construction of large projects.

The earlier a project is audited within the design and development process the better. If an
inappropriate concept or treatment (i.e. one with inherent safety problems in the particular context)
is chosen at the feasibility stage, it is very difficult and often impossible to remove the safety
problems at later design stages or once traffic is using it. Early auditing can also lead to the early
elimination of problems and, consequently, minimisation of wasted design time at later stages.

Road safety audits are applicable to all types of road projects, on all types of roads. It is not the
scale of the project that is important, but the scale of any potential hazard the design may
unwittingly hide. Developing a safe initial design is fundamental to the safety performance of a
road.

Road safety audits can be conducted on road projects as diverse as:

72
  new freeways
 major divided roads
 reconstruction and realignment projects
 intersection projects
 pedestrian and bicycle routes
 deviated local roads near major projects
 local area traffic management schemes and their component parts
 signal upgradings
 subdivision proposals
 crash reduction schemes
 safe routes to school projects
 maintenance with improvement activities, such as seal widening in conjunction with
rehabilitation.

Audits can also be conducted on projects that are ‘off-road’, but which affect nearby roads or create
off-road areas that effectively operate like roads. For example, a commercial development might
result in the following activities that may affect the safety of road users:
 vehicle/pedestrian conflicts in the new car park
 increased numbers of pedestrians crossing the adjacent road
 a spillover of parking onto an adjacent busy road
 restricted visibility or delays where vehicles access the development
 changed public transport circulation and access by users
 changed access/egress/unloading for delivery trucks
 changed traffic patterns on the adjacent road network.

Depending on the type of project and the stage of development a project has reached, each road
safety audit will consider different issues. However, the steps to be followed will generally be
similar. The steps are described in the flow chart in figure and can be thought of as a three phase
process:
The designer or client:
 selects an audit team
 provides all the documents
 holds a commencement meeting with the auditor.
The audit team:
 road safety audits all the documents and drawings
 inspects the site (including at night-time)
 repeats these two steps (as required)
 writes the audit report
 holds a completion meeting with the designer or client.
The designer or client:
 decides on the action required in response to the audit report, and its findings and
recommendations1
 documents these decisions
 implements the decisions by amending the design
 feeds the experience back into the organisation, to avoid similar design problems recurring
73
  where possible sends a copy of the documented decisions to the audit team leader.

The road safety audit process may use checklists to assist the auditor in considering the relevant
issues. Different checklists have been developed for the different stages of a project’s
development. The checklists are intended to be used as a prompt; they are not exhaustive lists
which cover every detail. The auditor should use individual judgement about the safety of any
feature. The checklists are not a substitute for knowledge and experience: they are an aid for the
application of that knowledge and experience and to make sure all factors are considered.

A road safety audit is a relatively straightforward process. The steps in the process are illustrated
in the flow chart in Figure below. In some organisational structures, and for some more minor
projects, some of the steps may be brief, but the sequence of steps will still apply. The steps apply
equally to design stage audits and other audits.

Selecting the Road Safety Audit Team

Objective: To select an audit team which is independent and has appropriate skills for the particular
project.

74
The most appropriate size of an audit team depends on the size and complexity of the audit task;
there is no optimum number of people, although teams of more than four people can be
unmanageable. Significant projects require at least two people.

The one essential ingredient in any road safety audit team is road safety engineering experience.
In addition, select people with relevant experience: is the project a freeway/a local
street/urban/rural, etc.

For small projects, an audit by one person can be effective; it depends on their skills and
experience. Avoid having a one-person team just because of cost; consider including someone
from within your organisation who is independent from the project. Every audit can be a training
exercise and it gives the audit team another pair of eyes.

Providing the Background Information

Objective: To provide the audit team with all the necessary information to allow an adequate
assessment of the project.

All relevant information must be provided to the audit team. The project designer should collect
all the necessary and relevant information in a usable form for the audit team. Information will
include scheme reports, data, drawings and relevant sections of contract documents. This step may
need to be initiated well before it is time to engage the road safety audit team. It may be necessary
to collect additional information, such as traffic volumes. This should be considered early enough
to avoid delays. The audit team also needs to be given a clear understanding of what tasks are to
be included in the audit.
Information to be provided to the audit team will typically include:

a) A clear statement of the expected outcome from the audit

This may require a written brief or a simple reference to the procedures and optional audit report
formats in these guidelines.

This concisely sets out the purpose of the project (i.e. the design project, not the audit), how it is
to be achieved, any deficiencies that need to be addressed, any design compromises that have been
made and the reasons, and any community input from prior discussion, correspondence and
consultations. For large projects, some of this information may be in reports used to support earlier
funding or programming decisions.

b) Site data
Site data includes:
 the design standards that have been used and any locations where they were not achieved
 traffic volumes including commercial and non-commercial components, cyclists and
pedestrians
 any previous road safety audit reports and the written responses to them; any known safety
issues which remain unresolved from earlier audits
 any environmental effects relevant to the location or the design, for example, weather
conditions (ice, fog, snow, etc.), animals, services, trees, historic buildings and topography.

75
c) Plans and drawings
These include:
 A set of drawings, at a scale suitable for the stage of design, showing the vertical and
horizontal alignment and other items relevant at the particular stage of audit. For example,
sign, linemarking and street lighting plans are essential at the detailed design and
preopening stages.
 Any other plans to cover adjacent roads or to describe adjacent land and its uses which
might be affected by the proposal or by the traffic changes it induces.

Holding a Commencement Meeting

Objective: To ensure the design team understands the audit process; to provide the audit team with
any additional information, identify key issues, constraints and potential issues requiring specific
consideration.

The most effective and efficient way to acquaint the road safety audit team with the background
to the project, and (if needed) to acquaint the designer/client with the audit process and the purpose
of the audit, is to convene a formal commencement meeting.

The commencement meeting provides the opportunity to explain to the audit team the project’s
purpose, any issues particular to this project, and any problems which have been experienced in
achieving planning, design or construction objectives. The designers may already have safety
concerns or queries about a particular aspect of their design. The audit team will not be able to
inspect the site under all traffic or weather conditions, so if particular conditions are important (for
example, traffic conditions at the end of each school day), the auditors should be advised. Plans
and background information are handed to the audit team, if this has not occurred before the
meeting.

If members of either party are unfamiliar with aspects of the audit process, this meeting is a good
time to explain the process and distinguish between the task of the audit team and the task of the
client. The audit team’s task is to identify and document any road safety concerns and
recommendations (if requested), while the client’s task is to respond to and act on those concerns
and recommendations (if requested).

Assessing the Documents

Objective: To safety audit the designs and background information and form conclusions about
the safety performance and crash potential of the road.

This phase takes place in parallel with site inspections; documents will need to be safety audited
before and after inspections. Before inspecting the site, initially peruse the documents (including
the plans) to record first impressions; list possible issues to be checked on site. Drawings, traffic
and crash data, field notes and other information should be assessed, using the checklists as
required. Identify any areas of the project which contain potential safety problems. If the
documents raise any questions, explanation should be sought from the designer or client before
the road safety audit report is written. Sometimes designers can be apprehensive about ‘outsiders’
coming in and commenting on their work; auditors can make use of the opportunity to ask
questions and allay designers’ fears.

76
The audit should confine itself to road safety aspects although a broad view of this should be taken.
For example, there can be road design elements which cause frustration or nuisance to road users,
where a direct relationship with crashes may be difficult to establish. Aspects like amenity or
aesthetics, unrelated to safety, should not be included in the auditor’s report. Likewise, traffic
capacity issues should not be included unless they have a bearing on safety.

Inspecting the Site

Objective: To see how the proposal interacts with its surroundings and nearby roads; to visualize
potential impediments and conflicts for road users.

It is essential for the road safety audit team to visit the site in daylight to appreciate any problems
relating to the present arrangements and, if appropriate, to visualise the future proposals and their
effects. Consider when is the most effective time to inspect the site as traffic conditions can vary
throughout the day or week.

A night-time inspection is also essential except where, in the experience of the client, there will be
nothing additional to observe. However, these circumstances should be rare. The visual
information available to road users can be markedly different at night-time and it can be surprising
what additional issues can be identified on a night-time inspection, even where work has not yet
commenced. The features on a road may be obvious in daylight, but deceptively hidden at night-
time. A night-time inspection permits a check on readability of the road and the effectiveness of
signs, markings and street lighting.

When on site, look beyond the limits of the design plans (or the limits of works at the pre-opening
stage); the inspection should include the adjacent sections of road. Transition or terminal zones,
where the new (and usually higher standard) road matches into the existing road system can often
be locations of greater hazard, as:
 road layouts and devices which previously operated safely can fail to do so once traffic
volumes, speeds or movements alter
 motorists may be unaware of the need to adjust their behaviour.

In addition, new roads or new traffic arrangements can often disrupt existing traffic and pedestrian
movement patterns.

The inspection should be undertaken from the point of view of all the likely road user groups and
not just motorists. Young and elderly pedestrians, truck and bus drivers, cyclists, motorcyclists,
elderly and disabled drivers have quite different safety needs:

Child pedestrians have a lower eye height to observe vehicles. Being small, they can be easily
hidden from a motorist’s view. They can act impulsively.
 Elderly pedestrians may be less agile, have poorer sight or hearing, or may have a poorer
ability in judging gaps and the speed of traffic.
 Truck and bus drivers have a higher eye height but this can lead to issues with delineation
and their visibility can be more easily affected by overhanging foliage. Their vehicles take
longer to stop and start moving, they are wider, and blind spots can be a problem.

77
  Cyclists are more seriously affected by surface conditions (for example, grates, potholes
and gravel) and gradients.
 Elderly drivers may be less able to recognise some traffic control features or judge gaps
due to cognitive difficulties.
 People with disabilities can be affected by poor eyesight and hearing or difficulties moving
around objects, near edges, between levels or at typical pedestrian speeds.
 Motorcyclists have rapid acceleration, but are susceptible to poor pavement conditions and
to ‘squeeze points’.

Consider how well the design caters for the different types of movements such as crossing the
road, entering the traffic stream or leaving it, as well as for travelling along the road. Consider
these for the different road user groups and the effects of different weather conditions.

Taking photographs or videotapes allows for later reference and possible inclusion in the report,
but they must not be used as a substitute for a site inspection; all audit team members should
inspect the site.

Writing the Road Safety Audit Report

Objective: To report on the audit’s findings, and if requested by the client, recommendations
regarding how identified safety deficiencies may be addressed.

The main task of the road safety audit report is to succinctly report on aspects of the project that
involve hazards and if requested by the client, recommendations about corrective actions. The
recommendations will usually indicate the nature or direction of a solution, rather than precise
details. Responsibility for that will rest with the client. The report provides the formal
documentation on which decisions about corrective action will be based.

A positive element of the design that improves safety can be mentioned in a road safety audit
report, but it is not necessary to mention them. The purpose of the report is not to rate the design,
but rather to address any road safety concerns.

In some cases, safety concerns may be identified but a recommendation, even a broadly directive
one, may not come to mind. In this case the safety issue should not be ignored; simply record the
finding (i.e. the safety concern) and under Recommendation write ‘Investigate treatment and
implement it’ or a similar comment.

In what order should items be listed? Sequentially along the length of a project? In the order in the
checklists? Grouped by common issues? There is no single best way of ordering findings and
recommendations, but the most important consideration is that the order be logical and helpful for
the report’s recipients when they consider the corrective actions. For example, where there are
three distinct intersections and ramps at a grade separated interchange and each one has identified
problems with the four elements of alignment, cross-section, delineation and visibility, it may be
better to discuss each site in turn, rather than discussing each design element about the different
sites in turn. In this way, any possible interaction between the problems at each site is more likely
to be recognised and addressed effectively. On long road projects, it may be more appropriate to

78
split the project into sections. In any event, if there is any concern that mutually dependent
recommendations are separated in the report, they should be cross-referenced.

Holding a Completion Meeting

Objective: To discuss the findings and recommendations4 for corrective action.

This meeting should involve the auditor and the client and/or the designer. It should not be viewed
as an opportunity to disagree with the findings or recommendations. Misunderstandings can be
resolved at this meeting, but it is preferable that this be done before the report is written. The
meeting can provide an opportunity for the client and/or the designer to ask for suggestions for
overcoming the identified problems.

In addition the completion meeting can be a useful opportunity to assist in training by familiarizing
participants with the full process and the nature of findings and recommendations.

Responding to the Audit Report

Objective: To deal with audit findings or recommendations in an effective manner; to judge
whether the recommendations of the road safety audit should be implemented and, where it is
decided otherwise, to give reasons in writing for the decision; to put agreed audit recommendations
into effect.

A. Procedures to deal with audit findings or recommendations

 This procedure needs to include well defined and documented procedures for dealing with
audit reports:
 who will respond to an audit report?
 who will sign off the corrective action report?
 who will ensure the agreed actions will be followed through?
 For each audit report: the action to be taken in response to each finding or recommendation,
by when, by whom and the current status of actions (has it been done yet?).

B. Responding to an audit report in writing

Road safety audit is a formal process. The audit report documents the audit team’s identified safety
concerns and recommendations (if requested by the client), to improve the safety of the design.
This must be responded to by the client (or the designer) with a written response to each audit
finding or recommendation. The response document must be signed by a representative of the
client. This response document, for example, may be a ‘corrective action report’ (CAR).

Audit recommendations are not mandatory. In the event of a crash, the audit documentation may
be sought by representatives of an injured person. It is important that audit recommendations are
given due consideration. If it is not possible to adopt a recommendation (for example, due to high
cost implications), is there another effective way of partly addressing the problem or can a solution
be staged over time? Reasons for not accepting findings and recommendations should be
adequately ocumented.

79
The client or designer may wish to call in an independent assessor for assistance with details of
how to respond to each audit findings or recommendation. It should be borne in mind that the audit
report will not include the design details of a solution to any problem.

Each finding or recommendation in the road safety audit report can be responded to by either:
 accepting it completely and designing a solution to overcome or reduce the problem, in
line with the audit findings or recommendation or in another equally effective way
 accepting it in part or in principle but, due to other constraints, implementing changes
which go only part of the way to resolving the safety problem
 � not accepting the finding or recommendation at all.

With the first point, the proposed action (for example, by whom and when) should be recorded. In
the case of the second or third points the reasons must be set out in writing. Also, with the third
point, if the finding is accepted, but the recommendation is rejected, this should be reflected in the
response.

Risk ranking of safety issues

The following tables may be useful to provide an indication of the level of risk and how to respond
to it. Determine into which category in Table below the issue best fits. From this select the risk
category in table below and its suggested treatment approach in Table below. This is not a scientific
system and professional judgement should be used.

80
D. Implementing the agreed changes
Once the corrective action report has been finalised, the agreed actions need to be implemented.
The designer has to develop design changes that address the safety problems. If one is at the pre-
opening stage, the actions need to be implemented as soon as possible on site. Temporary warning,
delineation or other treatment may be needed until the agreed solution is implemented.

Actions taken should be recorded (for example, description of work, by whom and when). This is
to fully close out the road safety audit finding as well as to factual record what works were
completed. Reasons for any variations from the proposed action must also be set out in writing.

E. The need for a subsequent audit

If it is decided to make significant changes to the design, a further audit of the revised design may
be appropriate, rather than waiting for the next design stage’s audit. This is a particularly important
consideration if the project has reached the detailed design stage and is to be built soon.

Closing the Loop – Feeding Back the Knowledge Gained

Objective: To disseminate the knowledge gained from an audit, for the wider benefit of road and
traffic designers.

Unless the knowledge gained from audits is fed back into the design process, there is a risk that
the same mistakes will be made again and again. Audits should be the catalyst for change, so that
the road safety engineering experience applied to one design can benefit future designs.

The opportunities for feedback include:

 Feedback into the current project.
 Feedback into other projects within the same organization. Ensure audit reports and
corrective action reports are widely circulated and discussed within the office. Deal with
any problem issues associated with peer review, standards vs safety, etc. Have designers
included in audit teams (on projects they are not associated with). Every couple of years,
review audit reports to see if there are common or repeated issues.
 Feedback generally to the profession. Include audit topics in professional development
seminars. Use road safety e-mail discussion groups and Internet sites.
 Feedback into revised standards. Contact your state road authority custodians of manuals
and standards about any examples of standard treatments which have compromised safety,
and request changes.

81
  Feedback to auditors. To improve future audits, advise your auditors about how you
responded to their audit (for example, send them a copy of the corrective action report).
This step is essential when audits occur at successive stages of a project.

As a way of gaining knowledge from audits, audited and un-audited design projects need to be
monitored for one to three years after they are built, to see whether crash problems are occurring
and, if so, whether the problems were anticipated in an audit. This can provide valuable feedback
into audit procedures: are sufficient designs being audited; do audit teams have the right mix of
people; are significant problems being identified; are the responses to audit reports appropriate?

4.2.1 The Audit of Road Design

Road safety audit has the greatest potential for improving safety when it is applied to a design
before a road or traffic project is built. It can be conducted on any proposal that is likely to alter
interactions between different road users, or between road users and their physical environment.

A road safety audit may be undertaken at one or more of the following stages as a design proceeds
from concept to implementation:
 the feasibility stage (including audit of the design brief)
 the preliminary design stage
 the detailed design stage
 the pre-opening stage (or post-opening if done just after the project opens).

Feasibility Stage Audits

By providing specific road safety engineering input at the feasibility stage of a project, road safety
audit can influence fundamental issues such as design standards, route choice, impact on and
continuity with the existing adjacent network, and intersection or interchange type, location,
number and layout.

For traffic management schemes or small-scale improvements this stage may be less significant,
but, until an audit is conducted, its value may be hard to judge. Where the basic choice of treatment
will affect safety performance, an audit, however brief, can be beneficial. The selection of an
inappropriate concept can be costly to rectify.

With larger projects, the selection of an inappropriate concept or design criteria at this stage may
be almost impossible to rectify later. Consider how difficult it would be to later try changing
designs if the following concepts are initially selected and designed, but are later judged to be
deficient in safety:
 a roundabout, rather than traffic signals
 a bypass to one side of a town, rather than the other
 an emergency breakdown lane on one side, rather than both sides of a bridge.

Why audit at the feasibility stage?

 to input safety engineering into the consideration of options
 to influence safety when there is the greatest scope for change

82
  to avoid obvious safety problems that can be locked in once designs commence or land is
acquired
 to ensure all likely road user groups have been considered in the design
 to check that the concept is compatible with the type of road and road user expectations
 to check that the design standards are compatible with the type of road and road user
expectations
 to look beyond the project and consider effects in transition areas and away from the project
 how does it fit into its environment?
 is it consistent?
 will staging involve compromises or be unsafe?
 is the scope of the project adequate, or are additional works needed elsewhere?
Preliminary Design Stage Audits
This audit occurs on completion of the preliminary road design or functional layout. If alternative
schemes have been developed for public consultation, each should be audited.

Drawings at approximately 1:1000 can be appropriate for the general alignment and 1:500 scale
for intersections and other specific locations of interest. Typical considerations will include
horizontal and vertical alignments, intersection layouts, the appropriateness of adopted standards
generally or at specific locations, access locations, whether all likely road user groups have been
considered and staging.

For larger projects, subsequent significant changes in road alignment become much harder to
achieve after this stage, as land acquisition and other associated legal matters commence.

The audit may identify unusual features. These may or may not be safety problems: engineering
judgement is required. Inconsistent or unexpected features can be a hazard where road users may
use them wrongly. Care is needed to ensure that an audit does not stifle an innovation that has a
good level of safety, simply because it is not a standard way of dealing with an issue.

Why audit at the preliminary design stage?

 an audit may not have been done previously
 to identify anything missed in a previous audit
 to avoid wasting costly design time if only a detailed design stage audit is done
 to check what standards have been used and what departures there have been from
standards
 to check that all likely road users have been considered, for example:
 can vehicles turn safely?
 can road users see each other?
 can road users see devices?
 is alignment and cross-section appropriate?
 is property access catered for?
 to check the adequacy of the road reservation width and its effect on batters
 to check intersection layouts and other conflict points
 to alert designers to areas where attention will be needed at a detailed design stage
 to check details at the connections to the existing road, for example:
83
  consistency
 fixed objects may be in a more vulnerable position.
Detailed Design Stage Audits
This audit occurs on completion of the detailed road design but before the preparation of
construction contract documents and finalisation of any land acquisition. This stage is the last
opportunity to change the design before construction commences. The audit reviews the plans that
will be used to build the project.

Drawings need to be at 1:500 or 1:200 scale (or larger), covering not only the general road layout
and alignment, but also intersection layout details, signing, linemarking, drainage, lighting,
fencing, landscaping, roadside objects, barriers, details of any signals, etc. The audit team
considers how these aspects of the design will affect the operation of the scheme for all the
different road user groups likely to use it.

This audit is very much concerned with the details especially of the road layout and the traffic
arrangements. Attention to detail at this design stage can do much to reduce the costs and
disturbance associated with last-minute changes that may otherwise occur if problems are not
identified until the pre-opening stage. Thus it is important that the auditors are provided with
drawings that contain the required details.

Why audit at the detailed design stage?

 audits may not have been done at previous stages
 to identify anything missed in previous stages and last chance to alter designs on paper
 to check what standards have been used and what departures there have been from
standards (if this has not been done during an earlier audit)
 to check the signing, linemarking and landscape plans
 to check that all likely road users have been considered, for example:
 can vehicles turn safely?
 can road users see each other?
 is alignment and cross-section appropriate and are fixed hazards present?
 to check the interaction of the detailed elements
 to check details at the connections to the existing road (especially consistency).

Pre-opening Stage Audits

The opportunities to rectify safety problems at this stage are limited, compared with audits at
earlier stages, but it is important to ensure detail is correctly implemented.

This audit involves a detailed inspection of a new scheme, its approaches and connections prior to
its opening. The new road or treatment is driven, ridden and walked (as appropriate) by the audit
team to ensure that the safety needs of all likely road users have been provided for.

The pre-opening audit is not simply an as-built check of the approved design, but also an
acceptance procedure on behalf of the travelling public. At this time, small modifications to some

84
aspects of the new work may be required to ensure that wrong messages will not be conveyed to
the road users in ways that compromise their safety.

A night-time inspection is essential. While it is needed for the obvious darkness-related issues like
signing, delineation and lighting, it is now recognised that a road layout that appears perfectly
acceptable during the daytime can give a totally different impression to its road users after dark,
causing specific safety problems.

If any major changes take place while the project is under construction, the client should seek road
safety engineering advice at the time, rather than relying on the pre-opening audit. A post-opening
audit can also be undertaken after a short period of operation, to assess how the road is actually
being used. Errors with the concept, or the detail implementation, are usually quick to surface and
these can be put right while contractual resources are still available.

Where a road carries traffic during the construction period and there is no ‘opening to traffic’, the
audit should be undertaken immediately following completion of construction activities,
preferably prior to removing any temporary traffic management or speed restrictions.

Why audit at the pre-opening stage?

 audits may not have been done at previous stages
 to identify anything missed in previous stages
 to check the inter-relationship of elements
 vertical and horizontal alignment
 things can look alright on plans, but not on site (in 3-D)
 to check that it’s built as designed
 designs and ‘incidentals’ can get changed on site
 spoil areas, services can get in the way
 landscaping gets added or expanded
 to check it at night-time
 confusion, visibility
 unplanned hazards can eventuate, like poles and pits not meant to be there
 signs can get lost in their background.

4.2.2 Road Safety Audit of Existing Roads

The road safety audit process can be applied to the existing road network, either in a route specific
manner (which yields detailed safety issues), or in a network wide manner (which yields more
general safety issues).

Throughout this guide the term road safety audit of existing roads is used but readers are advised
that other terms are equally applicable (e.g. road safety review and road safety inspection). There
is no single best way of carrying out a road safety audit of an existing road. Different ways are
discussed below: select one which suits your needs, making your primary aim to improve the level
of safety for the road’s users. The flow chart of activities in Figure 4.1 should still be followed,
making appropriate adjustments as required. Always include a daytime and night-time inspection.

85
The road safety auditing of existing roads is only one component of the assessment of existing
roads. The safety performance of existing roads should also include a crash investigation,
discussed below. When conducting audits on unsealed roads practitioners may refer to Chapter 11
of the Unsealed Roads Manual - Guidelines to Good Practice. This manual provides assistance in
the review of the safety requirements for unsealed roads

Identification and treatment of other potential hazards at a crash location, at the same time that the
crash causes are being treated. This may be done for little extra cost. Should crash records be
considered in the road safety audit of existing roads? A road safety audit of an existing road
involves informed judgments about the potential for future crash types along a road. Crash records
can be an important part of the information to be assessed in a road safety audit, but extreme care
should be taken to ensure that their consideration does not focus attention away from other
potential hazards. Also, consideration of past crashes without assessing the potential for other,
future crashes is not auditing or safety inspecting, but rather is crash investigation work. This is
covered in the Guide to Road Safety - Part 8: Treatment of crash locations. To avoid such
problems, some auditors wait until after their initial inspections before reviewing crash records.
One option might be to have only one team member look at the crash records before the inspection,
so that other members can focus on other issues. If considering crash records, check that they are
all still relevant and whether any recent changes to the road layout have removed the problem.

Do the road safety audit from the road users’ perspective

The road should be inspected from the point of view of each likely road user group and for the
different types of movement, such as crossing the road, entering the traffic stream or leaving it, as
well as for travelling along the road. This can best be achieved by using the road as road users do;
for example, by driving or riding at normal road user speeds, at night and day, in the dry and in
the wet, and by crossing the road where pedestrians would be expected to cross. Consider ways of
getting the comments of local road users.

What is a hazard?
During the road safety audit of an existing road, it can be easy to identify features that are not to
current standards or which are not ‘perfect’. Remember that standards do not necessarily equal
safety. A simple test to decide whether a feature is a potential hazard is to ask ‘What type of crash,
or what additional injury, could occur as a result of this feature?’ If you cannot think of a type of
crash resulting from the feature, it probably is not a hazard.

Road safety audit the whole network or parts of it?

Some road authorities seek to conduct a road safety audit of their whole road network on a regular
basis. Others select a random sample of roads, while others again limit their road safety inspections
of existing roads to sections of the network that are regarded as being of greatest potential hazard
to road users.

Types of road safety audits of existing roads

The aim of a road safety audit is to identify any existing safety deficiencies of design, layout and
road furniture. There should be a consistency of standards such that the road user’s perception of
local conditions assists safe behaviour.

86
Detailed single route or single site road safety audits
These road safety audits will result in comments about specific hazards along a route or at a single
site being road safety audited, with details about the nature of the hazard, its location and possible
options for remedial treatment. Unfortunately, there is almost no limit to the number of physical
deficiencies which can be identified on an existing road, compared with current guidelines on safe
road environments. These types of road safety audits, if undertaken over any significant length of
road, can result in a long ‘shopping list’ of identified deficiencies and a ‘wish list’ of
recommendations that is unlikely to be economically justified in the foreseeable future.

This raises the following issues:

 Typically, many of the identified deficiencies are maintenance issues. These could
be addressed by an appropriate maintenance program, rather than relying on
infrequent formal road safety audits.
 There is little benefit in conducting detailed road safety audits if resources do not
permit a majority of findings to be addressed within a foreseeable time.

Consequently, this detailed approach is best applied to specific, high-risk locations identified
through previous inspections or enquiries. It is also a useful complement to a crash location
investigation.

Broad network road safety audits

These audits will result in broad observations about consistency, adequacy of provision and extent
of maintenance. They will also identify any specific high-risk locations for later detailed
inspection. Typical issues can include:

Urban: street lighting: inconsistencies and pole placement; direction signs and street name signs:
provision and condition; road markings: level of maintenance; road surface condition (as a safety
issue); intersections: visibility and controls – provision and consistency.
Rural: guard fence: provision and end treatments; clearance to hazards intersections: visibility and
controls – provision and consistency; alignment: provision of warning at deficient locations;
delineation: adequacy and condition.

For any identified high-risk locations, a second detailed level of inspection can follow.
As a result of these road safety audits:
 inadequate road management practices can be identified
 new works programs can be initiated (e.g. a direction signing program and mass action
programs)
 changes in emphasis and priorities can be made to existing programs
 maintenance procedures can be changed to meet road users’ needs.

Following up the road safety audit

Once potential safety problems and corrective actions are identified, the next key task requires
determining the priority of corrective actions to be made. While some of the corrective actions
may be dealt with through road maintenance programs (e.g. faded or damaged signs, provision and
placement of advisory or regulatory signs, renewing of pavement markings, etc.), other corrective
actions may be more substantial and would require capital works funding (e.g. pole relocation,

87
provision of safety barrier, flattening of batters, etc.). Where capital works funding is a
requirement, and where the funding is not sufficient to correct all of the safety deficiencies deemed
as either intolerable or high-risk, then a road safety risk management approach should be
undertaken.

Application of crash risk models developed by ARRB Group in association with Austroads allows
road authorities to prioritize the treatment of potential safety problems so as to maximize the crash
risk reduction that may be achieved for the limited funds that are available. The evidence based
crash risk model is termed the Road Safety Risk Manager (RSRM). This may be applied to
prioritize the deficiencies for treatment so as to optimize the level of crash risk reduction for
available funding. This tool takes into consideration treatment costs, exposure and crash risk
reductions expected.

88
Chapter 5: Crash Investigation, Analysis and Treatment
To apply engineering treatments to a crash location, the causes of the crashes must first be
identified and then an effective treatment can be applied and this is known as crash investigation.
Definition of a Crash
A crash is defined as a set of events that result in injury or property damage, due to the collision
of at least one motorized vehicle and may involve collision with another motorized vehicle, a
bicyclist, a pedestrian or an object. The terms used in the HSM do not include crashes between
cyclists and pedestrians, or vehicles on rails.

What is a Crash Location?

A crash location (sometimes called a blackspot or hazardous road location) is:
• an individual site (e.g. an intersection or a bend in a road)
• a length of road (which could be e.g. urban or rural)
• an area of the road network (e.g. residential precinct, local traffic area or an entire suburb)
• locations across the road network which have a common hazardous feature (e.g.
substandard guard fence end treatments) and/or crash type (e.g. pedestrians).

The prevalence of crashes at only some locations, and the clustering of crash types at a single
location usually indicate that there are common causes for the crashes. It is the objective of crash
location treatment to identify these common causes and to counter them by applying appropriate
countermeasures.

Crash Risk
As risk is the product of three elements: probability, exposure and severity, a road safety strategy
must address all three elements. For a road agency these may include examples such as:

Influencing the probability of a crash

• applying sound traffic engineering and road safety engineering techniques through the
audit of new road designs and the treatment of known crash sites
• modifying road user behavior by appropriate design elements
• using well targeted education and enforcement programs
• applying appropriate speed management, including speed limits.

Influencing the exposure to a crash

• providing alternative, safer routes for vulnerable road users
• promoting safer forms of transport in preference to less safe forms.

Influencing the severity of a crash

• providing a more forgiving roadside environment (e.g. safety barriers)
• providing appropriate speed management
• providing good access for emergency services to reach crash sites.

89
It can be seen that the treatment of crash locations is just one element of a road safety strategy, but
it is an important and potentially very cost-effective part.

Treating Crash Locations

The treatment of crash locations involves a step-by-step process. Each of the steps needs to be
followed. Further, resources need to be applied, firstly to provide the crash information on which
all investigations are based, secondly to permit investigations and analysis to take place and thirdly
to permit the identified problems to be treated. For example:
 A data collection and verification system and a crash positioning protocol are needed, so
crash locations can be identified as accurately as possible.
 A comprehensive data base is needed, which includes details about a sufficient number of
crashes and crash features, so that problem locations and common crash features can be
identified.
 An appropriate criterion needs to be selected for defining ‘high’ crash locations. These
criteria may vary as the number of ‘high’ crash locations are effectively treated. The criteria
may also differ across programs funded by different levels of government (i.e. national,
state and local).
 A thorough diagnosis of the crash problems at a location is needed, so that the correct
conclusions may be drawn about contributing factors.
 Countermeasures need to be selected on the basis that they are known to be effective
against the particular problems identified, so that the problems are resolved.
 Safe design principles and road safety audit need to be applied to countermeasure design,
so that the countermeasure does not cause harm or result in new types of problems.
 An appropriate project ranking system is needed so that scarce resources can be applied
effectively to a program of potential countermeasures.
 Monitoring and evaluation of the effectiveness of countermeasures at site, route or network
level is needed to ensure that the targeted remedial treatments achieve their intended
purpose, while also continuing to improve knowledge associated with the treatment of
crash locations.

5.1 Crash data recording, coding, analysis

5.1.1 Importance of Crash and Road Data

The process of accurately investigating, analysing and effectively treating crash locations relies on
the use of comprehensive and accurate crash data and data related to the road and traffic
characteristics at the crash locations. Comprehensive and accurate data enables the:
• Crash locations to be accurately determined
• Events associated with crashes to be identified
 identification of crash contribution and severity factors, thus providing the basis for
selecting targeted remedial treatment options
• identification of common factors across a number of crashes
• cost consequences of a single crash, all crashes at one location or several crashes with
common factors to be identified
• crash sites to be ranked so that treatment can be applied to those sites that will derive the
greatest safety benefits.
90
5.1.2 Minimum Reporting and Coding Criteria
There is a minimum set of data about each crash which is necessary as a basis for the sound and
satisfactory identification and investigation of a crash location. A reasonable knowledge of crash
data definitions and limitations is required to accurately interpret this information in any given
jurisdiction.

There is a high level of reporting of the most severe injury crashes and a lower level and variable
amount of reporting of lower cost crashes.

From a road safety perspective casualty data is the key indicator of the road safety problem. More
recently, and in close alignment to Safe System principles and objectives, fatal and serious injury
crashes and casualties are used as road safety key performance indicators. Lower severity
outcomes as well as property damage data (where this is available) can provide valuable additional
data that can support proposed countermeasure treatments.

The different crash severities are generally defined as follows:

• fatal crashes (one or more persons killed or died within 30 days)
• serious injury crashes (one or more persons admitted to hospital, although this is more
typically based on whether a person was injured and taken by ambulance to hospital)
• minor injury crashes (one or more persons injured who is not admitted to hospital, although
this is more typically based on a person not being taken by ambulance to hospital but
requiring medical treatment)
• non-injury crashes above threshold values which may vary across jurisdiction, plus those
where the property owner is not present.

Purposes of data collection

Road crash information is used by a wide variety of people for a wide variety of purposes
including:
• road safety practitioners, for developing remedial or pro-active road and traffic measures
• police, who may be investigating whether they will charge a person with a criminal offence
in relation to a specific crash
• hospitals and health centres to monitor their health service requirements
• lawyers acting for clients in civil litigation, especially compensation for injuries and other
losses
• insurers, seeking facts before settling an insurance claim
• those with responsibility for road safety education and publicity, to ensure that their efforts
are well-targeted
• media
• police, in relation to enforcement activities, such as establishing the location for speed
cameras or breath testing stations
• safety administrators, exercising a duty to report statistical information on road crashes
• researchers, who need access to an accurate, reliable data base in order to conduct rigorous
research projects

91
 • vehicle and component manufacturers and suppliers of highway materials, who wish to
assess the safety of their product, perhaps from a viewpoint of litigation, marketing or
product enhancement.

As the primary purposes for data collection within different organizations vary, the information
collected, the way in which data bases are established, the opportunities for data aggregation and
analysis, and interpretation of information will vary; the opportunities to supplement information
on one data base with information from another will be severely limited. For example, the data
base of a motor vehicle insurance company may not code crash location by a numerical geographic
system, as this information is peripheral to concerns of insurance claim assessment and financial
management. The total information on the data base is then not in a format which can be used by
someone seeking information about the safety performance of a particular location.

Data relevant to road safety are collected every day in most countries, but for the data to be useful
for informing road safety practitioners, sufficient data must be collected, coded, validated and
stored safely.

5.1.3 Primary data sources for crash reduction programs

The primary source of road crash information is the police crash report form. However these crash
outcomes are only known after the event, so police will sometimes attend relatively minor crashes
and (less often) will not attend some serious crashes.

 the police data base

 the police report form (which can provide a greater amount of detail regarding the crash
circumstances, for detailed analysis of individual sites; reference to police crash report
form details is usually necessary if a collision diagram is to be prepared).

Other data sources

Traffic data, such as traffic volumes (including turning volumes), pedestrian flows and vehicle
speeds may be helpful, depending upon the particular circumstances and problems at the site. In
some cases these will be available, but in other cases they may need to be collected.

Hospitals record the causes of injuries as well as their nature, extent and treatment. Crash data
sourced from police information may indicate whether someone was taken by ambulance or not,
but typically does not give information as to whether that person was hospitalised, for how long,
or whether the injuries sustained resulted in long-term impairment. With advances in technology,
and greater collaboration between agencies, it is possible to link this information together, allowing
for road safety treatments to be focused on those resulting in the most debilitating injuries. In
special circumstances, provided confidentiality of patient information is secure, hospital data may
be made available for research purposes. This has been done with good effect in the development
of countermeasures to reduce the severity of road crash injuries.

Insurance companies require claimants against policies to provide a description of the

circumstances in which the loss, damage or injury occurred. This information is not usually
released, although it has been released as consolidated data by some insurers. It provides a potential
92
for establishing the true extent of lower cost crashes which are not reported to the police.
Unfortunately it is generally not in a form which is compatible with the needs of crash location
analysis and treatment.

A further source of information on crash occurrences is tow truck operators’ records, although
there is no system in place for collecting this information.

From time to time in-depth crash studies are undertaken into the nature and causes of crashes in a
particular area. These studies are costly to undertake and involve specialist teams attending crash
scenes and taking measurements and recording crash features. The results are usually published in
special reports.

Local knowledge is an important source of information about safety problems on the road network.
Subjective information about crash problems must be regarded cautiously, but it can be a pointer
to problems or prompt further investigation. Information sources can include local residents,
businesses, safety groups, emergency service personnel, local medical practitioners, maintenance
contractors and local authority staff.

Interviews of road users, including people who have been involved in a crash at a site of interest,
in a structured format can be used to gain information for the development of crash
countermeasures.

Traffic conflict surveys may be used where the collection of crash data is not practical. These
involve field observations or video recording of conflicts (near misses).

Site investigations are a necessary component of any countermeasure development program and
will often yield insights into the crash history at a site.

Speed survey data also provides a source of information regarding speeding behaviour.

5.1.4 Technology Available for Data Collection

Computer based technology is being used to improve the accuracy of data collection in developed
countries. To improve the accuracy of location information Global positioning systems (GPS) or
satellite navigation systems are being used for accurate determination of a crash location. The
person attending the crash scene uses the system instead of, or as well as, documenting the location
in traditional terms (ABC Road, xx metres N/S/E/W of XYZ Street). This method has great
potential in rural areas where recording of the distance and direction to identifiable features can be
subject to significant error.

It is also now possible to use a geographical information system (GIS) or digital mapping to record
crash locations. This permits crash data to be incorporated within a relational data base, allowing
crash sites to be overlaid on plans showing other geographical information such as highway
features, traffic flows, intersection layouts and land uses.

New technology makes the initial collection and assessment of safety-related data easier and more
useful. As an example, in Main Roads Western Australia, the crash investigation team has recently
begun using video to assist in fatal and serious crash investigations. The video camera used records
93
GPS information and has a viewer that overlays all route information on the video display. This
assists in the completion of initial crash investigations. Furthermore, research is currently
underway on incorporating inventory data e.g. road geometry data and crash data with the video
information. This is to assist high risk crash route assessments for the identification of proposed
crash mitigation measures. This research is in its early stages of development.

To improve the accuracy and completeness of crash data

Menu-driven crash data capture programs can be used with laptop computers or tablets by police
attending a crash to ensure that all desired information is collected at the site. These programs can
include in-built logic and consistency checks on the data as it is entered.

Crash report forms can be arranged so that the information can be scanned into the data base, to
minimise costs and to reduce the opportunity for coding errors.

5.1.5 Accuracy of Crash Data

It is to know about the limitations of the data and take steps to resolve any anomalies which may
occur. The limitations include:

Under-reporting of crash data – although significant attempts are made to collect and record all
relevant crash data, not all non-fatal crashes make their way to the relevant crash data base. Surveys
conducted in New Zealand and in some Australian states have tried to compare data from crash
data bases with information from hospital admissions and other sources. Information from New
Zealand indicates only around 60% of serious crashes make it to the crash data base, with the
percentage significantly less for minor injury crashes (Alsop & Langley 2001). Australian based
research has also identified similar under-reporting of injury crashes (Cercarelli 1998). A review
conducted for Austroads (2005a) identified that reporting rates also varied by type of crash. For
example, reporting rates were lower for cyclists, pedestrians and motorcyclists.

Systematic reporting bias – this bias can result from the regulations or policies covering the
reporting of crashes. Reporting criteria vary between jurisdictions, resulting in crash experience
not being comparable. Numerically, property damage (non-injury) crashes constitute the bulk of
crashes: there may be as many as forty-one non-injury crashes for every casualty crash (James
1983). However, in some jurisdictions they are not reported, or are not recorded in the data base.
This results in an incomplete and systematically biased crash picture.

Random reporting bias – it is well established that crashes involving children, cyclists, pedestrians
and minor injury are substantially under-reported (James 1991). A similar situation applies to
crashes involving illegal activity, such as under-age driving, driving while intoxicated, riding a
motorcycle exceeding regulated capacity and carrying a pillion passenger when not permitted to
do so by regulation.
Further, it is common for some human factors (e.g. alcohol and drugs) and roadway factors (e.g.
the presence of a roadside culvert) not to be recorded. The absence of this information on the crash
report form may mean the absence of the factor or the failure to record it. Erroneous conclusions
can be made from the wrong interpretation of this absence of data.

Subjective bias – some crash forms require an assessment of possible contributing causes of the
crash. This in itself adds a subjective element, as the range of possible responses to the question
94
will be affected by the recorder’s experiences and the purposes (other than crash recording) to
which the information may be put. For example ‘failure to give way’ may be seen as a cause by
someone regularly involved in traffic law enforcement, whereas the same situation may be seen as
‘control device not visible’ by someone regularly involved with road environment safety. One
examination of the frequency of reported causes revealed distinct differences between police
districts (Vincent 1996). Similarly, speed and fatigue are not typically based on direct observation.

Reporting errors – it is important to recognise the circumstances under which a police officer
obtains information to complete a crash report. There will often be more pressing matters at a crash
scene. The officer may not have local knowledge or adequate training in incident investigation, so
some data items may be inadequately or wrongly recorded. Crashes do not always fit ‘standard’
formats and there may not be the motivation to fill in the form.

Coding errors – these can occur throughout the process from filling out the crash report form to
the data entry at the computer terminal. It is estimated that errors of this type are present in 5% of
crash files (Ogden 1996). They are unlikely to be revealed unless the data are used for detailed
investigation at individual sites. Typical problems include wrong direction for the north point,
wrong direction for one of two vehicle movements, selecting the wrong road user movement (DCA
code), for example ‘rear-end’ instead of ‘rear end into right turner’, and numerical coding errors.

Location errors – the location may be imprecise or wrong in the original police report form and
this will be carried through into the data base. If this continues through into coding, crashes at one
location may appear in two separate parts of the crash data base. The location reference system
may also be imprecise, so that a user of the data may not be able to accurately determine the
location (e.g. all mid-block crashes may be recorded as being half way between the adjacent
intersections).

Discontinuities over time – definitions or interpretations of field data may be changed over time
by those responsible for coding and reporting, so that data from one time period cannot be
compared with that of another. An abrupt change in recorded crash experience at a site should lead
an analyst to enquire whether there has been any discontinuity of this kind.

Delays – agencies responsible for data processing may not be sufficiently resourced: it may be
many months before information is available for analysis. Data may only be released annually.
This means that countermeasure development may be responding to historical crash patterns which
may be out of date.

Masked or hidden problem: it may be the case that a location is perceived as being so dangerous
that people avoid using it. In this situation the safety problem results in a reduction of amenity
(e.g. as pedestrians choose to cross the road somewhere regarded as safer) rather than resulting in
crashes.

The raw data collected on the crash report needs to be coded into groups of relevant and useful
parcels of information. Crashes with similar characteristics are grouped under a code (DCC:
Descriptive Crash Codes); these groups along with location create clusters that practitioners will
be able to analyse. The data form has also collected contributing factors, which becomes part of

95
the practitioner’s information for analysis. All road crash data has patterns, patterns that indicate
behaviour, road condition, road design and vehicle issues, which lead to the crash. Most crash
types have tried and true remedial actions or solutions to reduce the number of crashes and also to
reduce the severity of crashes.
This Descriptive Crash Code (DCC) system identifies clusters of similar types and is important for
the treatment of crash locations. For example, cluster of rear end crashes at the approach to an
intersection, could provide clues to safety engineers of the deficiency of the road at the intersection
and lead to appropriate safety treatments. Depending on the primary and/or secondary contributing
factors (high approach speed, low amber time or poor/skid pavement surface etc.), it is possible to
suggest appropriate targeted road safety countermeasures to reduce these crashes in future.
Similarly, existence of many similar right-through crashes at an intersection may require replacing
permissive phase by protective phase of traffic signal. Too many head on crashes at a location may
be treated by median strip and /or by providing passing (overtaking) lanes or by banning
overtaking. Cluster of road departure crashes may call for the installation of roadside barriers
(guard rails). The proposed DCC system for Nepal is shown in the figure below.

96
5.2 Speed calculation from crash data

In order to draw conclusions of each road accident it is necessary to compute the original speeds
of the vehicles involved in various types of crashes.
Following are the typical modes of vehicular crashes:
1. Moving vehicle collides with a stationary vehicle.
2. Two approaching vehicles from different direction collide at an intersection.
3. Head on collision of vehicle approaching from opposite directions.
4. Moving vehicle collide with stationary object like tree, electric pole, parapet or any other
rigid structure.
Following assumptions are made while analyzing any particular crash.
1. When skid marks at the site of accidents are visible they are measured to find out actual
braking distance and assumed that 100% skid has taken place. When skid marks are not
visible it is assumed as a free collision, without brake being applied.
2. Impact of the colliding vehicles may be direct or oblique.
3. Friction coefficient of the pavement surface is either determined or suitably assumed
depending upon the conditions of the surface.
4. When two colliding vehicles of masses m1 and m2 with speed of v1 and v2 collide it is
assumed that both are perfectly plastic and both move together with common velocity of
v1 after collision.
m1v1 + m2v2 = (m1 + m2) v1
It means coefficient of restitution is zero.
If both vehicles are assumed perfectly elastic the coefficient of restitution will be unity.

v1  v 2  v12  v11
In case coefficient of restitution is e, then

e(v1  v 2 )  v12  v11.

The value of coefficient of restitution e is either known or suitably assumed.

To work out speed from skid distance

Skid distance ‘S’ for a vehicles of weight of ‘W’ to slow down from v1 to v2 m/sec speed is
obtained equating loss of kinetic energy to the work done against frictional force.

97
 W  (v 2  v 2 )  W  f  s
2g 1 2

Where
f - average friction factor or skid resistance
W- Weight of vehicle

v12  v22
S
Or, 2f g

Taking g = 9.8 m/sec2. If speed in kmph

V12  V22
S
254  f

If the skid distance ‘S’ is measured from skid marks, the initial speed of vehicle v1 can be worked
out

v1  v22  2  g  f  s
m/sec

V1  V22  254  f  s
kmph
If vehicle comes to stop after skid distance S, then v2 will be become zero.

v1  2  g  f  s
m/sec

V1  254  f  s kmph

Collision of moving vehicles with parked vehicles

Suppose vehicle 1 moving with speed v1 m/sec skid through a distance s1 after the effective
application of brakes, collides with a parked vehicle 2 and two vehicles skids together through a
distance s2 before coming to stop. The problem is to estimate initial speed v1 m/sec or V1 kmph
of a vehicle.
1. Before collision the vehicle 1 of weight W1 moving with initial speed v1 m/sec applies brakes,
skids through a distance s1 and attains speed v2 m/sec just before actual collision takes place.

v12  v22  2  g  f  s1
................ (i)
2. At collision with stationary vehicle 2 of weight W2 both start moving together at speed v3
m/sec. Here it is assumed that perfectly plastic impact has taken place. The relationship
between v2 and v3 is given by momentum equation.

98
 W1  W  W2 
  v2   1   v3
g  g 

 W  W2 
v2   1   v3
 W1 

Substituting the value of v2 in equation (i)

2
 w  w2 
v12  1   v3  2  g  f  s1
2
 w1  ................ (ii)
3. After collision vehicle 1 and 2 skid through distance s2 before coming to stop ( velocity v3
reducing to v4=0 )

 v32  2  g  f  s 2

v 32
Substituting this value of in equation (ii)
2
 W  W2 
v12   1   2  g  f  s 2  2  g  f  s1
 W1 

2
 W  W2 
 v1   1  2  g  f  s 2  2  g  f  s1
 W1 

In kmph unit

  W  W2 
2 
V1  254  f  s2   1   s1 
  W1  
 

Where s1 and s2 is skid distance in meter. When vehicle 1 does not apply brakes at all the
value s1 become zero.

Two vehicles approaching from right angles collide

Two vehicles A and B on approaching an intersection are assumed to skid on application of brakes;
they collide with each other and skid further in different direction as shown in figures.
The direction of the skidding vehicles after collision in this case depends upon the initial speeds
of two vehicles and their weight.

99
 S S VA3
If A2 and B2 are the skid distance of the vehicles after collision, the speed of vehicles and
VB3
just after collision may be found by the relation

VA3  254  f  s A 2
kmph (i)

v A3  2  g  f  s A 2
or m/sec

VB3  254  f  s B2
kmph (ii)

vB3  2  g  f  s B2
or m/sec
The momentum of the vehicles just after collision may be found using the speed values from
equations (i) and (ii) and these resolved in the original direction of motion of two vehicles. As per
assumption the momentum before collision is taken equal to the momentum after collision.

The three cases of right angled collision

Case (a)
A' B'

SA2 SB 2
B
A A

SA 1
SB1

WA W W
 VA2  B  SinB  VB3  A  CosA  VA3
g g g

WB
VA2   VB3  SinB  VA3  CosA
WA

WB W W
 VB2  A  VA3  SinA  B  VB3  CosB
g g g

WA
VB2   VA3  SinA  VB3  CosB
WB

Case (b)

100
 A'

S A2
SB2 B'
A A
B

SA1
S B1

WB
VA2   VB3  CosB  VA3  CosA
WA

WA
VB2   VA3  SinA  VB3  SinB
WB

Case (c)
A'

SA2

A A

B S
B2
SA1
SB1
B'

W
VA2  VA3  CosA  B  VB3  CosB
W A

WA
VB2   VA3  SinA  VB3  SinB
WB

VA1 VB1
The original speeds of vehicles and before skidding may be calculated by:

VA1  254  f  sA1  VA2 2

kmph

VB1  254  f  sB1  VB22

kmph

Numerical Example 1:

101
A vehicle of weight 2.5 tonnes skids through a distance equal to 38 m before colliding with another
parked vehicle 1.5 tonne. After collision both the vehicles skid through a distance of 15m before
stopping. Compute the initial speed of the moving vehicle. Assume coefficient of friction as 0.48.
Soln. Method (1) by steps
Let original speed of vehicle be v1 m/s, reduced to v2 m/s by applying brakes and skidding
through S1 = 38m. Just after collision let both the vehicles A and B start moving together
with speed v3 m/s and finally stops at v4=0, after skidding through a distance S2=15m and
take f = 0.48

After collision
Loss in kinetic energy of both vehicles together = work done against frictional force


WA  WB 2
2g  
v3  v24   WA  WB   f  S2

v32
 f  S2  0.48 15
2g

 v3  2  9.81 0.48 15  141.26 m/s

At collision
Momentum before impact = momentum after impact
WA v2 WA  WB
  v3
g g

WA  WB
v2   v3
WA

WA + WB 2.5 +1.5
= = 4
WA 2.5 2.5

WA  WB
 v2   v3  4 141.26 m/s
WA 2.5

Before collision
Loss of kinetic energy = work done against braking force in reducing the speed
WA 2
2g  
v1  v22  WA  f  S1

v12  2gfS1  v22  2  9.81 0.48  38  16  141.26  719.49

6.25

102
  v1 = 26.82 m/s

Original speed V1 = 26.82 × 3.6 = 96.55 kmph

Method (2)
The equation

  W W 
2 
v1  254f  s2  A B
  S1
  WA  
 

 2 
v 2  254  0.48  15  2.5  1.5   38 
  2.5  

v1 = 96.51 kmph

Numerical Example 2:
A vehicle skids through a distance of 40 m before colliding with another parked vehicle the weight
of which is 75 percent of the former. After collision both the vehicles skid through 14m before
stopping. Compute the initial speed of the moving vehicle. Assume coefficient of friction is 0.62.
Soln. Let original speed of vehicle v1 m/s
Reduced to v2 m/s by applying brakes and skidding through S1 = 40m
After collision let the vehicles A and B start moving together with speed v 3 m/s and finally
stops at v4= 0, after skidding through distance S2=14m take f = 0.62

After collision
Loss in kinetic energy of both vehicles together = work done against frictional force


Wa  Wb 2
2g  
v3  v 42   Wa  Wb   f  S2

v32
 0.62 14
2g

 v3  2  9.81 0.62 14  170.3  13.05 m/s

At collision
Momentum before impact = momentum after impact

103
 Wa v 2 Wa  Wb
  v3
g g

Wa  Wb
 v2   v3  1  0.75 170.3  22.84 m/s
Wa 1

Before collision
Loss of kinetic energy = work done against braking force in reducing the speed
Wa 2
 
v  v 22  Wa  f  S1
2g 1

Therefore initial speed of the vehicle before application of brakes

v12  2gfS1  v22  2  9.81 0.62  40  22.84  1008.24

 v1 = 31.75 m/s

Original speed V1 = 31.75 × 3.6 = 114.3 kmph

Numerical Example 3:
Two vehicles A and B approaching at right angles, A from West and B from South, collide with
each other. After the collision, vehicle A skids in a direction 47o North of West and vehicle B, 56o
East of North. The initial skid distances of the vehicles A and B are 32 and 18 m respectively before
collision. The skid distances after collision are 14m and 33 m respectively. If the weights of
vehicles A and B are 4 and 6 tonnes respectively, calculate the original speeds of vehicles. The
average skid resistance of the pavement is found to be 0.52.

Soln. Let the initial speeds of vehicles A and B before brake application be vA1 and vB1, the speeds
just before collision, after skidding through SA1 = 32 and SB1 = 18 be vA2 and vB2, the speeds
just after collision, be vA3 and vB3 and the final speed when the vehicles come to a stop is
zero; after skidding through further distance SA2 = 14 and SB2 = 33m;
f = 0.52

104
After collision
Loss in kinetic energy of each vehicle = work done against skid resistance
2
WA vA
For vehicle A, 3
 WAfSA 2
2g

 vA3  2gfSA 2  2  9.81 0.52 14 = 11.95 m/sec Similarly,

2
WBv B
For vehicle B, 3
 WBfSB2
2g

 VB3  2gfSB2  2  9.81 0.52  33 = 18.35 m/sec

At collision
Equating the moments before and after impacts after resolving the momentums along West
East direction
WA W W
 vA2  0  B sin B  VB3  A cos A  vA3
g g g

WB
 vA2  sin BvB3  vA3 cos A
WA

= 6 sin 56o × 18.35 – 11.95 × cos47o

4

= 14.67 m/sec
Resolving the momentums along South North direction,
WB W W
 vB2  A vA3 sin A  B vB3 cos B
g g g

WA
 vB2  v sin A  vB3 cos B
WB A3

= 4 × 11.95 × sin 47o + 18.35 × cos 56o

6

= 16.09 m/sec
Before collision
Loss in kinetic energy due to brake application = work done against brake application

i.e.
2g 
WA 2
vA1  vA
2
2 
= WAfSA3

 v2A1  2gfSA1  vA
2
2
= 2 × 9.81 × 0.52 × 32 + 14.672

105
 = 541.69
vA1  23.27m/ sec vA1  23.27 × 3.6 = 83.77 kmph

Similarly,
WB 2
2g 
vB1  vB
2
2 
 WBfSB1

v2B1  2gfSB1  vB
2
2
= 2 × 9.81 × 0.52 × 18 + 16.092 = 442.53

vB1  21.04 m/ sec vB1 = 21.04 × 3.6 = 75.74 kmph

Method (ii) by using the equations.

Using the equation for calculating speeds of vehicles just after collision;
VA3  254fSA 2  254  0.52 14 = 43 kmph

VB3  254fSB2  254  0.52  33 = 66 kmph

Using the equation for speeds of vehicles just before collision,

WB
VA2  sin BVB3  VA3 cos A  6 sin56o × 66 – 43 × cos 47o
WA 4

= 82.07 – 29.33 = 52.74 kmph

WA
VB2  V sin A  VB3 cos B  4 × 43× sin 47o + 66 × cos 56o
WB A3 6

= 20.97 + 36.91 = 57.88 kmph

Original speeds of vehicles before application of brakes are obtained as:

VA1  254fSA1  VA2 2  254  0.52  32  52.742 = 83.71 kmph

VB1  254fSB1  VB22  254  0.52 18  57.882 = 75.68 kmph

Numerical Example 4:
Two vehicles A and B approaching at right angles, vehicle A from West and vehicle B from South,
collide with each other. After the collision, vehicle A skids in a direction 64o North of East and
vehicle B, 35o North of East. The initial skid distances of the vehicles A and B before collision are
10 m and 12 m respectively. The skid distances of vehicles A and B after collision are 18 m and 21

106
m respectively. If the weights of vehicles A and B are 6 and 8 tonnes respectively, calculate the
original speeds of vehicles. The average skid resistance of the pavement is found to be 0.52.
Soln.
SA1 = 10 m SA2 = 18 m WA = 6 tonne
SB1 = 12 m SB2 = 21 m WB = 8 tonne
f = 0.52
Initial speeds of vehicles A and B are vA1 and vB1 respectively.
Speeds of vehicles A and B at collision are vA2 and vB2 respectively.
Final speeds of vehicles A and B after collision before coming to dead stop are v A3 and vB3
respectively.

WA v2A3
For vehicle A,  WAfSA 2
2g

 VA3  2gfSA2  2  9.81 0.52  18  13.55 m / sec

= 48.78 kmph
2
WBv B
For vehicle B, 3
 WBfSB2
2g

 VB2  2gfSB2  2  9.81 0.52  21  14.64 m / sec

= 52.70 kmph
At collision
Equating the moments before and after impacts after resolving the momentums along East
direction
For vehicle A
WB
VA2  VA3  cos A  cos B  VB3
WA

 VA2  13.55  cos 64o  8  cos 35o  14.6  21.93 m / sec

6

= 78.95 kmph

For vehicle B

107
 WB W W
 VB2  A VA3 sin A  B VB3 sin B
g g g

WA
 VB2  V sin A  VB3 sin B
WB A3

= 6 × 13.55 × sin 65o + 14.64 × sin 35o

8

= 17.53 m/sec = 63.11 kmph

Initial speeds
For vehicle A

VA1  2gfSA1  VA2 2  2  9.81 0.52 10  21.932

= 21.14 m/sec = 86.9 kmph

For vehicle B

VB1  2gfSB1  VB22  2  9.81 0.52 12  17.532

= 20.73 m/sec = 74.63 kmph

Numerical Example 5:
Vehicles A from west and B from south weigh 4 tonne and 5 tonne respectively approaching at
right angle collide with each other. After collision vehicle A skids 18 m in a direction 65° North
of East and vehicle B skids 8.0 m in a direction of 32° South of East. The initial skid distances of
vehicle A and B before collision are 3.5 m and 10 m respectively. Calculate the initial speeds of
the vehicles if the average skid resistance of the pavement is 0.61.
Soln.
SA1 = 3.5 m SA2 = 18 m WA = 4 tonne
SB1 = 10 m SB2 = 8 m WA = 5 tonne
f = 0.61
After collision
For vehicle A
VA3  2gfSA2  2  9.81 0.6118  14.68 m / sec

= 52.85 kmph
For vehicle B

108
 VB3  2gfSB2  2  9.81 0.61 8.0  9.78 m / sec

= 35.21 kmph
At collision
For vehicle A
W
VA2  VA3  cos A  B cos B  VB3
W A

VA2  14.68  cos50o  5  cos 32o  9.78  16.57 m / sec

4

= 59.65 kmph

For vehicle B
WA
 VB2  V sin A  VB3 sin B
WB A3

= 4 × 14.68 × sin 50o – 9.78 × sin 32o

5

= 5.46 m/sec = 19.66 kmph

Before collision (Initial speeds)
For vehicle A

VA1  2gfSA1  VA2 2  2  9.81 0.61 3.5  16.572

= 17.79m/sec = 64.04 kmph

For vehicle B

VB1  2gfSB1  VB22  2  9.81 0.6110  5.462

= 12.33 m/sec = 44.03 kmph

5.3 Crash Frequency, Crash Rates

Crash frequency is defined as the number of crashes occurring at a particular site, facility or
network in a one-year period. Crash frequency is calculated according to Equation below and is
measured in number of crashes per year.
𝑁𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝐶𝑟𝑎𝑠ℎ𝑒𝑠
𝐶𝑟𝑎𝑠ℎ 𝐹𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 =
𝑃𝑒𝑟𝑖𝑜𝑑 𝑖𝑛 𝑌𝑒𝑎𝑟𝑠

Crash estimation refers to any methodology used to forecast or predict the crash frequency of:

109
  An existing roadway for existing conditions during a past or future period;
 An existing roadway for alternative conditions during a past or future period;
 A new roadway for given conditions for a future period.

5.3.1 Observed Crash Frequency and Crash Rate Methods

Crash frequency and crash rates are often used for crash estimation and evaluation of treatment
effectiveness. The historic crash data on any facility (i.e., the number of recorded crashes in a
given period) is referred to as the “observed crash frequency”. “Crash rate” is the number of
crashes that occur at a given site during a certain time period in relation to a particular measure of
exposure (e.g., per million vehicle miles of travel for a roadway segment or per million entering
vehicles for an intersection). Crash rates may be interpreted as the probability (based on past
events) of being involved in an accident per instance of the exposure measure. For example, if the
crash rate on a roadway segment is one crash per one million vehicle miles per year, then a vehicle
has a one-in-a-million chance of being in an accident for every mile traveled on that roadway
segment. Crash rates are calculated according to Equation below.
𝐴𝑣𝑒𝑟𝑎𝑔𝑒 𝐶𝑟𝑎𝑠ℎ 𝐹𝑟𝑒𝑞𝑢𝑒𝑛𝑐𝑦 𝑖𝑛 𝑎 𝑃𝑒𝑟𝑖𝑜𝑑
𝐶𝑟𝑎𝑠ℎ 𝑅𝑎𝑡𝑒 =
𝐸𝑥𝑝𝑜𝑠𝑢𝑟𝑒 𝑖𝑛 𝑆𝑎𝑚𝑒 𝑃𝑒𝑟𝑖𝑜𝑑
Observed crash frequency and crash rates are often used as a tool to identify and prioritize sites in
need of modifications, and for evaluation of the effectiveness of treatments. Typically, those sites
with the highest crash rate or perhaps with rates higher than a certain threshold are analyzed in
more detail to identify potential modifications to reduce crashes. In addition, crash frequency and
crash rate are often used in conjunction with other analysis techniques, such as reviewing crash
records by year, collision type, crash severity, and/or environmental conditions to identify other
apparent trends or patterns over time. Advantages in the use of observed crash frequency and crash
rates include:
o Understandability –observed crash frequency and rates are intuitive to most members of
the public;
o Acceptance – it is intuitive for members of the public to assume that observed trends will
continue to occur;
o Limited alternatives – in the absence of any other available methodology, observed crash
frequency is the only available method of estimation.

Crash estimation methods based solely on historical crash data are subject to a number of
limitations. These include the limitations associated with the collection of data. Also, the use of
crash rate incorrectly assumes a linear relationship between crash frequency and the measure of
exposure. Research has confirmed that while there are often strong relationships between crashes
and many measures of exposure, these relationships are usually non-linear.

A (theoretical) example which illustrates how crash rates can be misleading is to consider a rural
two-lane two-way road with low traffic volumes with a very low observed crash frequency.
Additional development may substantially increase the traffic volumes and consequently the
number of crashes. However, it is likely that the crash rate may decline because the increased
traffic volumes. For example the traffic volumes may increase threefold, but the observed crash

110
frequency may only double, leading to a one third reduction in crash rate. If this change isn’t
accounted for, one might assume that the new development made the roadway safer.

Not accounting for the limitations described above may result in ineffective use of limited safety
funding. Further, estimating crash conditions based solely on observed crash data limits crash
estimation to the expected average crash frequency of an existing site where conditions (and traffic
volumes) are likely to remain constant for a long-term period, which is rarely the case. This
precludes the ability to estimate the expected average crash frequency for:
o The existing system under different geometric design or traffic volumes in the past
(considering if a treatment had not been implemented) or in the future (in considering
alternative treatment designs);
o Design alternatives of roadways that have not been constructed.

As the number of years of available crash data increases the risk of issues associated with
regression-to-the-mean bias decrease. Therefore, in situations where crashes are extremely rare
(e.g., at rail-grade crossings) observed crash frequency or crash rates may reliably estimate
expected average crash frequency and therefore can be used as a comparative value for ranking .

Even when there have been limited changes at a site (e.g., traffic volume, land use, weather, driver
demographics have remained constant) other limitations relating to changing contributing factors
remain. For example the use of motorcycles may have increased across the network during the
study period. An increase in observed motorcycle crashes at the site may be associated with the
overall change in levels of motorcycle use across the network rather than in increase in motorcycle
crashes at the specific site.

5.3.2 Crash Estimation using Statistical Methods

Statistical models using regression analysis have been developed which address some of the
limitations of other methods identified above. These models address RTM bias and also provide
the ability to reliably estimate expected average crash frequency for not only existing roadway
conditions, but also changes to existing conditions or a new roadway design prior to its
construction and use. As with all statistical methods used to make estimation, the reliability of the
model is partially a function of how well the model fits the original data and partially a function
of how well the model has been calibrated to local data. In addition to statistical models based on
crash data from a range of similar sites, the reliability of crash estimation is improved when historic
crash data for a specific site can be incorporated into the results of the model estimation.

A number of statistical methods exist for combining estimates of crashes from a statistical model
with the estimate using observed crash frequency at a site or facility. These include:
o Empirical Bayes method (EB Method)
o Hierarchical Bayes method
o Full Bayes method

111
5.4 Crash Location Treatment Process

The treatment of crash locations should be a methodical, step-by-step process.

1. Decide on the criteria for listing crash locations

Define the physical limits of individual locations, so that sections with similar characteristics are
considered together. Decide on the time period over which crash patterns are to be investigated.
All sites need to be compared using an agreed selection criterion. The preferred criterion is ‘cost
of crashes by crash type’ rather than a number of or rate of crashes. If necessary, select a crash
threshold, above which locations will be considered for inclusion as crash locations.

112
2. List all crash locations to investigate
Examine the information in the crash data base to identify locations which meet the definition of
crash location. Establish the cost of crashes at each location, over the agreed time period. Make a
list of all the locations which meet the minimum cost threshold selected. Ensure that locations are
sensibly defined, so that no location worthy of investigation is missed through being subdivided
in the data. Plan ahead for later monitoring.

Having identified all the sites worthy of investigation, each one should be examined in a step-by-
step fashion to identify the factors leading to crashes, develop solutions and organise having those
solutions implemented

3. Obtain all the relevant information

Obtain the crash data for the location. Be aware of the limitations on the availability and accuracy
of crash data. Obtain other information such as traffic volumes, recent changes in the road network
or traffic generating land uses, and any documented concerns about safety at the location.

4. Diagnose the problems

This is a three step process
a. analyse the crash data (including crash rates and densities) for any clustering by common crash
types or factors such as common approach legs, common weather or daylight, common age of
those involved, etc. Construct a factor matrix and draw a collision diagram. Is examination of the
original crash report forms warranted?
b. inspect the site from the perspective of the involved road users, as well as undertaking a close-
up examination of the site’s features and its users’ behaviour.
c. make any other investigations, then draw conclusions about the likely causes of crashes for
which there are common factors. There may be other types of contributing factors (e.g. speeding),
but focus on what it is about the road or traffic environment which is leading to crashes.

5. Select the countermeasures

Match the solutions to the problems. The key to the selection of countermeasures is to concentrate
on the particular crash types which have been identified in the diagnosis phase (Section 3) and
which are amenable to treatment with road or traffic engineering measures. Select the
countermeasure(s) and take account of the crash modification factors for each countermeasure.

6. Prepare a preliminary design

A preliminary design is required, so that its practicality can be confirmed and the cost of the
remedial treatment can be estimated. This design then needs to be road safety audited. Prior to
implementing the project, the design needs to be finalised, taking account of any audit
recommendations.

7. Establish the benefits and costs

113
Undertake an economic appraisal. Establish the costs (i.e. the initial design and construction costs
only) and the benefits (including reductions in crash costs by crash type). Decide whether to use
net present value (NPV) or benefit/cost ratio (BCR). Conduct sensitivity testing.

8. Document the findings

Draw together the documentation which has been undertaken through Steps 3 to 7 and set it out in
a format which allows this project to be assessed against other potentially worthy crash
countermeasure projects.

9. If there are several locations to treat - rank all treatments

Compare all projects’ NPV or BCR. An alternative ‘goals achievement approach’ can be used,
whereby projects are ranked but no attempt is made to assess their economic benefits against their
costs. These formalised forms of appraisal are simply an aid for decision making. They should not
be the only criterion for selecting safety improvement projects and their numerical answers should
not be a replacement for sound decision making.

10. Implement the treatment

Once the countermeasure treatment has obtained funding it can be installed. It is important that the
design which is being implemented accords with the results of the crash investigation. During the
implementation phase, traffic safety will continue to be important. Once the works have been
completed, the project should (where feasible) be the subject of a pre-opening road safety audit.

11. Monitor the treatment and evaluate its effectiveness

Monitoring is the systematic collection of data about the performance of road safety treatments
after their implementation. Evaluation is the statistical analysis of that data to assess the extent to
which the treatment (or a wider treatment program) has met crash reduction objectives. These tasks
are important to ascertain the positive and negative effects of a treatment and thus improve the
accuracy and confidence of predictions of that treatment’s effectiveness in subsequent
applications. It may take a number of years to collect sufficient data.

5.5 Countermeasure Selection and Design

Having identified the elements of the road and traffic environment which contributed to the crashes
and their severity, the next step involves consideration of countermeasures. For a solution to be
effective, it must be applied to a particular problem which it is known to affect. It must be an
effective countermeasure.

The aim of countermeasure development is to:

 select countermeasures which have been demonstrated to be effective in reducing the
incidence and/or severity of target crash types

114
  check that adopted countermeasures do not have undesirable consequences, either in safety
terms (e.g. lead to an increase in the number or severity of another crash type, or crash
migration) or in traffic efficiency or environmental terms
 be cost-effective, i.e. maximise the benefits from the whole program of expenditure over a
number of sites
 be efficient, i.e. produce benefits which outweigh the costs.
There are several criteria for countermeasure selection, including (Ogden 1996):
• Technical feasibility: can the countermeasure provide an answer to the safety problems
which have been diagnosed and does it have a technical basis for success?
• Economic efficiency: is the countermeasure likely to be cost-effective and will it
produce benefits to exceed its costs?
• Affordability: can it be accommodated within the program budget; if not, should it be
deferred, or should a cheaper, perhaps interim solution be adopted?
• Acceptability: does the countermeasure clearly target the identified problem and will it
be readily understandable by the community?
• Practicability: is there likely to be a problem of non-compliance, or can the measure
work without unreasonable enforcement effort?
• Political and institutional acceptability: is the countermeasure likely to attract political
support and will it be supported by the organisation responsible for its installation and
ongoing management?
• Legal conformity: is the countermeasure a legal device, or will users be breaking any
law by using it in the way intended?
• Compatibility: is the countermeasure compatible and consistent with other strategies,
either in the same locality or which have been applied in similar situations elsewhere?

It can be seen that the decision to adopt a particular countermeasure may involve more than a
simple matching of a solution to a problem.

Crash Modification Factors

Detailed information exists on the expected benefits that are likely to come from different
engineering treatments. The benefits are termed Crash Modification Factors (CMFs, expressed as
a fraction, indicating the expected remaining crashes) or Crash Reduction Factors (CRFs,
expressed as a percentage reduction). Ideally, the reduction in death and serious casualties would
be used, but unfortunately information that breaks crash reduction down by severity is rare.
Instead, the reduction in casualty crashes is often used as a proxy.

CMFs provide an indication of the expected remaining casualty crash rate following the
implementation of a typical countermeasure. This will help the practitioner choose the treatments
that are likely to bring about the greatest safety benefits depending on the crash types targeted.

The CMF serves as a simple multiplication factor applied to the existing crash rate. Therefore, a
CMF of less than one indicates a reduction in the crash rate, whilst a CMF greater than one
indicates an increase.

115
Reliable information on crash reduction has been produced by Austroads (2012b) for use in
Australian and New Zealand conditions. This information can also be found on the Austroads Road
Safety Engineering Toolkit. The table at the end of this section shows CMFs for various treatment
options. Extensive information on CMFs is also available from other sources too (e.g. Elvik et al.
2009; and the CMF Clearinghouse from the US, see www.cmfclearinghouse.org). All sources of
information on CMFs should be used with caution, as the environment where such treatments were
used may be different to local conditions. Evaluations of treatments also have variable quality, so
care should be taken when selecting these.

Crash countermeasure selection is not as simple as identifying the treatment with the lowest CMF.
The practitioner should identify the most applicable treatment for the crash type and road
environment considered, and then consider the relevant CMF.

Crash Reduction Factors (CRFs) are often also referred to. These provide information on the
expected percentage reduction (or increase) in casualty crashes (i.e. a CMF of 0.8 would be the
same as a CRF of 20%).

Care should be taken when applying CMFs to ensure that the treatment type being considered will
appropriately address the crashes of concern, and not result in any adverse impacts. Example below
presents practical examples of what to consider when applying CMFs.

Applying Crash Modification Factors

It is important to use expert judgement when applying CMFs. When implementing a change to a
traffic control to reduce one crash type, do not assume this will also reduce other crash types. In
some cases a measure that reduces one type of crash may result in more crashes of another type.

There is clustering of several crash types at a signalised crossroad. One type is the DCA
corresponding with a through-right collision. It is decided to fully control the right turn involved
in these crashes by installing a red arrow. The ‘remodel signals’ treatment a CMF of 0.6 for these
crashes. But what about the CMF for other crash types with this treatment?

The intersection also has a DCA corresponding with through-through collisions, but if the only
improvement is the new right turn red arrow, there will be no reduction in these crashes because
the right turn red arrow will not affect them. Only if other works are undertaken to target these
crashes, can the 0.7 CMF for crashes where signals have been remodelled be applied.

Consider the installation of a roundabout in a local street such as that shown in Figure D 9 as a
countermeasure to intersection crashes. The only crash type is through-through collisions. A
‘roundabout’ treatment has a CMF for these crashes of 0.3. But what about the changes to other
crash types for this treatment? Only apply the percentage increases to crashes of the relevant types
already reported. In this case there are none. But be realistic and use engineering judgment: are
there high cyclist numbers, which would make cyclist crashes likely with a roundabout? Might
rear-end crashes start to occur? How might they be avoided? To minimize the risk of new types of
crashes, have the new design road safety audited.

116
The development of CMFs is an ongoing process, with these values further refined as new
information comes to light. Most jurisdictions provide their own list of CMFs. Austroads also
conducts and publishes research in this field (see Austroads 2010b and Austroads 2012b).

Whilst each treatment acts to reduce crashes, the impact of each successive crash on reducing the
numbers of crashes is diminished.

In some cases there will be diminishing returns similar to those identified in the formula above. In
other cases, the addition of treatments may produce negligible additional benefit, or treatments
may act together to provide benefits greater than the sum of the individual treatments. Professional
judgement is required when making estimates of the combined benefits of more than one treatment.
Step 1: Select the most appropriate countermeasure/s
Using tables and road safety engineering experience and judgement, select either:
• one solution
• a combination of solutions

117
• two alternative solutions (e.g. lower and higher cost solutions with different crash
reduction effects), for the repeated crash types at the location.

118
119
120
Step 2: Apply CMFs
• Use CMF tables or alternative, robust sources of effectiveness values to determine the
effect of the proposed treatment. The cost information is to be used to stimate the economic
benefit of the proposal.
• Do not, as a first step, CMF table to find a solution which produces a good benefit/cost
ratio to justify a project. Step 1 must always be applied first, to match any solution to the
factors contributing to the crash types.
• For example, CMFs are available to estimate the crash reduction as a result of applying
audio-tactile edgelines as a countermeasure to combat run-off-road crashes.

121
Documentation
After these steps, document the selection of the countermeasure/s, in a format suitable as part of
the final crash location investigation and treatment report. It will include:
• a list of the proposed treatments which are designed to counter the identified crash causes
(with mention made about which treatment is aimed at which problem)
• a plan of the preliminary design of the countermeasures.

It may put forward two or more options to be considered, i.e. each with unique costs and effects
on crash reduction.

122
123
Chapter 6: Engineering Design of Road Safety Measures
6.1 Crash barriers

The purpose of road safety barrier systems is to shield vehicles from striking a hazard. But
impacting a road safety barrier is itself a hazard for vehicle occupants although usually less severe
than impacting a rigid object in the road reserve (e.g. pole or tree). Road safety barrier systems
may increase the likelihood of vehicle impacts because they are longer than the point hazards they
shield and are closer to the traffic.

Installation of a road safety barrier in front of a hazard requires space for dynamic deflection,
vehicle roll, system width, sight distance, sufficient length for terminals, and a run-out area for
terminals. On constrained road reserves it may be necessary to consider other options (such as
hazard relocation) because there is insufficient space to install a road safety barrier.

A possible outcome of a roadside safety assessment (using the hazard mitigation process in Figure
below) during the design of a new road or re-design of an existing road, is that roadside hazards
will need to be shielded by a road safety barrier system.

124
Road safety barriers are broadly described as flexible, semi-rigid or rigid. A barrier should only be
installed when the consequences of vehicle impact with the barrier are likely to be less severe than
the consequences of impact with the feature being shielded. Generally, the likelihood of striking a
barrier is greater than striking the hazard (e.g. a tree some distance further from the road).
However, the severity of an impact with the barrier is usually much less than that associated with
striking the hazard.

For hazards adjacent to existing roads, alternative options should be considered before a decision
is taken to install a barrier. These may include improvements to the road (e.g. alignment, cross
section, pavement surface, delineation) and/or the removal or treatment of hazards. Options for the
removal, treatment or shielding of roadside hazards should be considered during the planning and
design phases of projects.

6.1.1 General Requirements for Road Safety Barrier Systems

It is generally required that the use of road safety barrier systems should be:
Supported by technical literature and assembly instructions that clearly demonstrate the essential
mode of operation and prominently show the test level achieved in crash testing carried out
Selected and located in accordance with a recognised design procedure that is professionally

125
applied. The procedure is to take account of risk management techniques that address the
community of road users and neighbours that may be affected by the installation. The hazard
mitigation and design procedures in this guide meet these requirements.

 Erected in accordance with manufacturers’ instructions.

 Maintained in a manner that reflects specified requirements.
 Returned into service only after professional evaluation and execution of repairs.
 Fitted with end treatments and interface devices that are appropriate.

6.1.2 Factors Considered in Barrier Selection

A pre-requisite to the selection of a barrier is a risk assessment and an evaluation of the economic
viability of other possible measures compared to the provision of barriers. The various factors that
should be considered in selection of the type of barrier to be adopted are summarised in table
below.

The appearance of a road safety barrier system may also be an important issue at some sites where
compatibility with other architectural or geological features is essential.

Site Conditions
The key site factors that need to be assessed include road geometry, offset distances and cross
slopes.

Road geometry:
Some systems, such as wire rope barriers, have restrictions in regard to their use where the
horizontal and vertical alignment standards are less than that specified by the manufacturer.

Offset:

126
  The objective is to minimise both the probability of a barrier being impacted by an errant
vehicle and the severity of any collision with the barrier. In general, provided that the
roadside would enable an errant vehicle to recover, it is desirable that road safety barriers
be located as far as possible from the edge of the traffic lane as site conditions permit. This
will maximise the chance of the driver being able to regain control of the vehicle and also
minimise the length of barrier required and the hazard it presents. However, a greater offset
from the edge of the lane can result in larger impact angles, higher impact severity and a
higher probability of the barrier being penetrated. This aspect also requires consideration.
 It is essential that the most appropriate barrier is selected to suit the particular site. Rigid
barriers should generally be located between 1.0 m and 3.0 m (and no more than 4.0 m)
from the edge of the through lane as the angle of impact for errant vehicles may increase
with offset. At increasing impact angles the rigid barrier profile becomes ineffective and
injury severity increases.
 When located adjacent to horizontal curves, road safety barriers may need to be offset
further from the edge of the traffic lane so that they do not impede horizontal sight distance
 Sufficient width should be provided between the road safety barrier and the traffic lane to
enable stationary vehicles to park clear of through traffic. Also, the full width between the
pavement and a concrete barrier should be paved to ensure optimum barrier performance,
and consideration should be given to sealing the shoulder.
 When road safety barriers are used to shield embankments, consideration needs to be given
to the provision of adequate ground support as over time softening of the verge may occur.
For example, where the restraining mechanism is supported on posts, a clearance of not
less than 500 to 600 mm from the rear of the post to the top hinge point of a fill embankment
should be provided, although this may vary due to soil conditions, batter slope, post depth,
and other factors. In situations where post restraint is of concern, deeper post embedment,
closer post spacing or the use of soil plates may be considered. A soil plate is attached to
the bottom end of the post to increase the area of post available to resist moment forces
arising from vehicle impact. Reference should also be made to the manufacturer’s
specifications.

Cross slopes:
 Irrespective of the type of barrier being used it is preferable that the approach slope is
essentially flat because road safety barriers perform best when they are impacted by
vehicles with their centre of gravity at or near the normal position.
 In general, semi-rigid and flexible barriers should not be used where the slope in front of
the barrier is steeper than 10:1. An exception is where it is necessary to provide a drain in
a median and also to locate a barrier within the median in which case a 6:1 slope may be
acceptable, but only if it is not practicable to achieve 10:1.

Compatibility
As a general practice road authorities use a limited number of different, proven road safety barrier
systems on new construction and reconstruction. This practice has advantages in that maintenance
personnel need to be familiar with only a few systems and stocks of replacement parts are more
easily managed. Non-standard or special barrier designs need only be considered when site
characteristics or performance requirements cannot be met with standard systems.

127
Cost
The selection of a barrier should consider the life cycle cost of the systems and their safety
performance, including injury and property damage costs, and maintenance costs. Initial capital
cost of the barrier is only one component of economic evaluation; however, this is not to say that
the initial cost of the system is not an important budgetary and project management consideration.
The choice of a terminal treatment may also be a significant factor with respect to the cost of the
system.

Maintenance
A barrier can perform as intended only if it is properly installed and maintained. Maintenance
factors that need to be considered are:
 routine maintenance of the barrier itself
 impact repair
 effect of the barrier on adjacent road and roadside maintenance (pavement overlays, etc.)
 material and component storage requirements.

Aesthetics
Aesthetics are not normally an over-riding factor in the choice of barrier. However, greater
importance is now being placed on the aesthetics of road safety barriers, especially in recreational
and tourist areas. It may also be preferred for aesthetic reasons that a particular type of barrier is
used consistently along a road or section of road.

Field experience
There is no substitute for documented evidence of a barrier's performance in-service on the road.
This information provides feedback to designers and construction personnel on the performance
of various types of barrier in various situations. It is particularly important that road authorities
learn from both observing the results of impacts with barriers and from examining crash reports.

Environmental impact
Apart from the aesthetic appeal of the barrier, other environmental factors that may require
consideration are that:
 barriers that have a larger frontage area may contribute to a build-up of drifting snow or
sand, thereby affecting operation of the road and potentially the effectiveness of the barrier
 the use of certain preservatives in some wooden barriers or barriers that have wooden
components may be an issue
 some types of steel railing may deteriorate rapidly in highly corrosive environments
 solid barriers may block tourists’ views of scenic panoramas, or a driver’s sight distance
 fauna migration patterns.

6.1.3 Limitations on the use of barrier types

There are some key factors that must be considered in the selection of a barrier type to ensure that
it is suitable for the particular circumstances. However, it is important that designers also consult
the relevant national or state jurisdictional guidelines in choosing a barrier. It should be noted that
some national and state road authorities may have a product acceptance process for barriers
proposed for use on their roads.

128
129
6.1.4 Road Safety Barrier Design Process
A road safety barrier design can be undertaken in accordance with the figure below. In designing
a safety barrier system it is important that all system components meet the required test level over
the entire length of the barrier.

The figure shows that the road safety barrier design process includes:
1. Collection of information about the site such as geometry, speed zoning, AADT etc.
2. A clear understanding of the objectives of the proposed road safety barrier.
3. Choice of a trial lateral location for the barrier.
4. Confirmation that the clearance between the barrier and the hazard can accommodate the
required working width for the chosen barrier type.
5. The choice of a type of barrier for further consideration.
6. Where necessary, choice of an alternative barrier type and a repeat of points 3 and 4.
7. Development of detailed aspects of the barrier design such as the:
(a) transverse location of the barrier and any site modifications necessary to ensure that
impact height criteria are met
(b) points of need and length of need
(c) treatment of leading and trailing terminals
(d) details of interfaces where different types of road safety barrier systems meet (such
as a road safety barrier meeting a bridge barrier).

At all sites where hazards have been identified, it is important to realise that drivers expect a
consistency of treatment (in accordance with current design practice), along sections of road that
have similar features.

130
131
6.1.5 Types of barrier
Road safety barrier systems can generally be divided into three broad types comprising flexible,
semi-rigid and rigid barriers. As a general principle, if it is practicable to meet the requirements of
the following guidelines, the more flexible barrier should always be used as this minimises the
severity of any vehicle impacts with the barrier. However, flexible barriers have relatively large
deflections which may render them unsuitable in some situations. In some cases a barrier that will
contain large heavy vehicles may be required and this will be reflected in the containment level
specified.

Special barrier designs and modifications to existing designs have been developed for use where:
 there is a need to cater for vulnerable road users (i.e. motorcyclists and cyclists)
 the aesthetic appearance of the roadside is important.

In general, flexible barriers comprise tensioned wire ropes, semi-rigid barriers have horizontal
steel beams and rigid barriers are constructed with concrete.

132
For details of barrier profiles and materials (e.g. concrete strength and composition) designers
should refer to the relevant jurisdictional or manufacturers’ standard drawings and specifications.

The type of rigid barrier profile used varies between jurisdictions. The single slope and vertical
wall barriers have an advantage in that the road can be resurfaced without affecting the profile or
requiring expensive resetting of the base level of the barrier. This a key reason for their use in
some jurisdictions. There are two types of profile that have the same height but different slopes on
the wall.

To function correctly rigid barriers require some lateral restraint which may require a foundation.
Where drainage is required at the barrier location (e.g. in a narrow median) it is important the
drainage design is coordinated with the barrier design (i.e. location and depth of pipes, pits, etc.).

133
134
6.2 Delineation Measures
In a broad sense "delineation" stands for any device or treatment whose aim is to outline the road.
Delineating devices give the driver visual clues as to where the road is going. Many studies have
shown that they can have great safety benefits.

6.2.1 Importance of delineation

Delineation is always helpful, but it is particularly important at bends. Road crash records have
shown that more crashes happen on bends than on straight sections.

135
Good delineation of a road serves the following purposes:
 it shows the safe limits of the road,
 it helps drivers to control approach speed on bends,
 it improves lane discipline.
 it aids in identifying potentially hazardous situations such as sharp bends and obstacles,
and
 it helps to prevent crumbling of road edge and deterioration of shoulders.

Delineating devices will not make our roads completely safe. Nevertheless, studies elsewhere
show that an intelligent mix of some of the delineating devices can reduce accidents by as much
as 60 percent.

6.2.2 Types of delineating devices

In some respect delineators do warn of an impending hazard situation. In strict sense, however,
only road indicators can be considered as delineating devices. Hence, only the following are
discussed here:
 Centre and edge lines
 Post delineators
 Confidence blocks
 Earth-filled bitumen drums
 Chevron signs
 Reflective studs

6.2.3 Present practices

Some of the delineating devices, for example the centre line, edge line, post delineator, earth filled
bitumen drum and confidence blocks have already been in use in Nepal for a long time. However
due to the lack of a guideline, their use has not been consistent throughout the country.

Centre and edge lines

Except on single and intermediate lane roads as per NRS-2045, white centre and lane lines are
found to be widely used. Yellow edge lines have been introduced a few years ago on roads with
shoulders.

The usual practice has been to apply a single coat of paint, once or twice a year, or as permitted by
the allotted budget. The total cost of application at present is about Rs 200 per sq. m.

Post Delineator
These are prismatic concrete posts painted with black and white strips. However they are found to
be constructed in many shapes and sizes. Although relatively effective and economic elsewhere,
they have not got deserved consideration in Nepal. Post delineators have been found to be used
along curves, high embankments and floodways. They cost about Rs 600 per piece at present.

136
Confidence blocks (masonry parapet)
Standard designs for stone masonry confidence blocks and concrete (drum) blocks were introduced
in 1978 but different shapes and sizes are found in practice. They tend to be used as delineator-
cum-safety barrier. However, experience has shown that stone masonry confidence blocks are not
effective as safety barriers. They look solid, but they shear easily when hit, so they may be giving
drivers a false sense of security. Painting them white helps make them more visible, but, being
low, they get dirty quickly, and so they are not easily seen at night.

137
The cost of typical masonry and concrete (drum) confidence blocks without foundation is about
Rs 750 and Rs 300 per piece respectively. Confidence blocks of stone-filled gabion crates have
been used for the first time in Lamosangu-Jiri road, and on many other roads thereafter. They may
serve as a delineating device during day time in a limited way. However without reflectors, they
are not prominently visible during nights. Per cubic metre cost of stone-filled gabion crates is about
Rs 1000 at present.

Earth-filled bitumen drums

During the past, especially along roads constructed with Indian aid, earth-filled and white painted
bitumen drums have been used as delineators. They are too low to be clearly visible and the white
paint quickly gets dirty. However, they are better than nothing when the funds available are
insufficient. One drum will cost around RS 200.

138
6.2.4 Recommended delineation practices
Centre and Edge Lines
Centre and lane lines have long been considered a standard form of road delineation and should
be standard on all multi-lane roads, as defined by NRS-2045. They assist the driver to locate the
vehicle laterally on the roadway, and thus help in avoiding collision.

Edge lines help the drivers to be aware of the pavement edges and affect the position of vehicles
in a lane. By reducing the incidence of vehicles leaving the road, the edge lines also help to prevent
crumbling of road edge.

Post Delineator
As the posts stand out prominently, they provide very good delineation even when the road is
impounded with flood. The posts enable the driver to plan forward route, and thus need to be
consistent.

The addition of a small reflector would make the delineator much more visible at night when good
delineation is especially important. The additional cost would only be about Rs 100.

Chevron Signs
Delineation is critical on horizontal curves, especially on isolated curves with a radius less than
600m. As chevrons provide much greater visual impact, they are found to give a better long range
information on such curves.

Two type of chevron signs are recommended to be used in our context: signs with a set of three
chevrons and signs with a single chevron. The sign with three chevrons furnish more visual impact
and are suitable for very tight curves. The single chevron boards, however, provide 'dynamic'
visual information to the driver as they negotiate curves of larger radius.

At present the sign with three chevrons and the one with single chevron cost about Rs 6500 and
Rs 2500 respectively, if locally manufactured. Imported higher-quality signs will cost more than
double the amount.

139
Chevron signs provide a much better visual impact than post delineators. However, the chevrons
are many times costlier. Therefore for the time being, their use may have to be limited to sites
where loss-of-control accidents are common.

The sign with three chevrons are installed on sharp curves. The sign shall be installed at the crown
of the curve or at the location from where a sudden change in direction begins. The rest of the
curve shall be provided with post delineators. The sign can also be used to indicate temporary
diversions at roadwork sites - in which case it should have yellow chevrons on a black background.

140
Reflective Road Studs
Neither road paints nor thermoplastics produce a marking which is very visible at night, even when
glass beads (Ballotini) have been used. Markings are particularly difficult to see at night when it
is raining. This has led many countries to install retro-reflective road studs on the centre line, which
reflect the head light beam back to the driver.

Two types of retro-reflectors are presently available in the market: corner-cube and biconvex.
The first one is much more effective than the latter.

The studs may have one reflecting surface (uni-directional) or two (bi-directional), with each
reflecting surface fitted either with a corner-cube reflecting plate or with groups of bi-convex
reflecting beads.

Reflective studs provide better night time delineation than painted centre lines and edge lines,
especially under adverse weather condition. It is also useful in reducing encroachment across the
centre line at narrow bridges. A 33 percent reduction of accidents has been reported following the
installation of these delineating devices.

Raised reflective studs present a retro-reflective face to the oncoming traffic. These devices are
attached along the centre and edge lines to make the lines more visible especially during night
time. Recently some bi-convex type studs has been fitted on the road section in front of
Babarmahal. Due to the frequent driving of the vehicles over them some of the studs have already
been depressed into the road surface. However the remaining of the studs are functioning well.

141
6.2.5 Combination of delineation measures
For an effective delineation, the driver should be supplied with enough visual information. As each
of the delineating devices possess some unique advantages, a careful mix should be designed.

Consistency is very important for the installation of delineating devices. And any broken or
missing device must be, replaced promptly. For example, if post delineators are provided along a
50m radius curve on a road, all the similar curves on this road should also be equipped with post
delineators Absence of the delineation device on similar locations may cause accidents, and may
further lead to disbelief in the delineation system.

Any overuse of delineating devices, alone or in combination with other traffic signs, confuses
drivers. And the devices may lose their ability to gain due attention. It is not only uneconomic, but
may also lead to hazardous situations.

The delineating devices should be in harmony with other road furniture such as: safety barriers,
confidence blocks, traffic signs and signals. They should be according to the same standard and
augment each other’s' intended roles.

6.3 Traffic Sign Placement

Example: A truck traveling at 30 kmph is approaching a stop sign. At time to and at a distance of
15 m the truck begins to slow down by decelerating at 3 m/sec2. Will the truck be able to stop in
time?
Solution:
v  v0  at
Where v—final velocity = 0 m/sec2
v0 -- 30 kmph = 8.33 m/sec2
a—3 m/sec2
Therefore
0 = -3t + 8.33
Giving
142
 t = 2.78 sec
The distance covered by the truck in these 2.62 seconds is
1 1
x  v0 t  at 2  8.33 * 2.78  * (3) * 2.78 2  11.56  15
2 2
This indicates that the truck will stop just in time

Example: Timing of change (yellow) and clearance (all red) intervals of traffic signals. Consider
a case of intersection with approach speed of 50 kmph, a coefficient of friction 0.45 and assume
that the driver’s perception reaction time of 0.5 s.

Solution:

The safe stopping distance =

50 2
0.278 * 50 * 0.5   28.82
254 * 0.45
For a vehicle to safely clear the intersection from a point just at the safe stopping distance, it will
have to travel the stopping distance plus the width of the street plus one car length (to clear rear of
the vehicle)

If the street if 25 m wide and car is taken to be 5.8 m long, such a vehicle must traverse T =
28.82+5.8+25 = 59.62 m before vehicles from opposite approach are released. If the vehicle is
assumed to travel at its approaching speed of 50 kmph
Yellow time =
59.62
 4.3
50 / 3.6
The yellow and all red signal should be a total of 4.3 seconds long to accommodate the safe
clearance of vehicles unable to stop when the light changes.

Example: Sign placement

Toll plaza ahead—be prepared to stop
How far in advance of the toll plaza should such sign be placed?
Given that, it can be seen from the distance of 100m and that queued vehicles from the toll plaza
rarely extend more than 50 m from the gates. Approach speed is 60 kmph, coefficient of friction
is 0.35 and reaction time 2.5 s.

The sign must be seen in time to allow vehicles to stop safely before the end of the vehicle queue
at the toll plaza.
Solution:
SSD =
60 2
0.278 * 60 * 2.5 
254 * 0.35
= 82.2 m
The vehicle cue extends 50 m from the toll gates. Thus the driver must see the sign a minimum of
82.2+50 = 132.2 m from the gates. The sign itself may be read from 100m. Thus the sign must be
placed a minimum of (132.2 – 100) = 32.2m in advance of toll gates.

143
Chapter 7: Crash Costing and Analysis
Road traffic accidents (RTA) are commonly viewed in terms of individual personal losses or as
general statistics. Unlike aviation or rail crashes (where frequently, many persons may die in a
single crash), the large-scale economic and social impact of road crashes is rarely appreciated, as
road crash deaths and casualties normally only happen in ones and twos. Road crash costing
attempts to estimate the annual cumulative losses incurred by a country as a result of road crashes.

Several different methods exist for costing road crashes, including court awards and life insurance
contracts, but the two most common are the gross output method and the willingness to pay
method. The gross output method (also referred to as the human capital approach) is based on
assessing the economic consequences of road crashes, usually supplemented by a notional sum to
reflect pain, grief, and suffering for those involved and also for family and friends of those killed
and injured, as a proxy for crash costs. The willingness to pay method estimates the amount of
money people affected by a particular measure would pay to avoid an crash and produces a much
higher valuation of crash costs.

Since the late 1980s willingness to pay has increasingly been applied for crash costing in
industrialized countries. However, the gross output method has been the most commonly used
method in most countries over the past few decades. This method is recommended for developing
countries as its primary objective, increasing a country’s wealth, is thought more appropriate to
their needs. Many assumptions are required in crash costing and, whenever alternative values or
uncertainties present themselves, a conservative approach is recommended thus ensuring that a
indisputable minimum value is obtained of road crash costs in a country. If investment can be
justified on such a minimum value, it will certainly be justified on any other value.

WHY IS ROAD crash COSTING NEEDED?

Crash costing highlights the socioeconomic burden of road crashes. Developing countries face
many challenges and have many resource needs. Road safety tends not to receive due consideration
because not all road crashes and casualties are reported to the police and there is usually no other
system of estimating road crashes and the corresponding casualties nationwide.

There is also a problem with the perception that road crashes are random, unintentional, or
predestined; i.e., unavoidable. Road crashes are too often accepted as inevitable negative side
effects of motorization. However tragic the personal losses, road crashes are rarely perceived as a
serious drain on the economy and this leads to complacency towards road safety issues. This view
is totally wrong.

Road crashes have been shown to cost annually between 1 percent and 3 percent of GDP in
developing countries. The gross national product (GNP) is often more readily available than the
GDP figure and although usually slightly higher than the GDP, can be substituted for it for the
purposes of rough estimation. These are large sums that few countries especially developing
countries can afford to lose, year after year.

144
Knowledge of crash costs allows safety impacts to be economically justified. Road safety measures
have been frequently ignored or downplayed in cost benefit analyses on the grounds that the
associated costs and benefits are too intangible. Where road safety is included in cost benefit
analyses of road improvements, it is often only factored on a subjective basis and so does not get
applied in the consistent manner required for project comparisons. So road safety has generally
been severely underfunded as it is not possible to prove its cost-effectiveness without the use of
road crash cost values.

Key Components
The key components that need to be considered when examining this sector relate largely to the
various cost components. These can be classified into casualty-related costs (lost output, medical
costs, and the pain, grief, and suffering value), crash-related costs (property damage and
administration), and crash data. As shown in figure below, all are resource costs, except for the
pain, grief, and suffering component.

The cost of a crash is the sum of the casualty-related costs, plus the crash-related costs, while the
total cost of crashes in a country is the number of crashes by severity multiplied by their respective
crash cost.

Crash Data
To calculate total crash costs, the number of crashes and casualties by severity must be known.
While crash data is often taken for granted in industrialized countries, in developing countries with
little road safety awareness, crash data may be incomplete and inconsistent.

To permit cost estimation, casualty figures need to be provided separately by the traffic police for
each crash severity type, as shown in table below.

145
While the internationally accepted definition of a road crash death includes all related deaths within
30 days of the crash, many countries report only deaths occurring at the scene or within a few days.
Adjustment factors should, therefore, be applied when making international comparisons.
Published ministry of health figures, when available, should be compared with police reported
figures to ensure that an accurate assessment of total road accidents and casualties is achieved.

Serious injuries are defined as those that require hospitalization (at least one night) while slight
injuries require medical treatment but no overnight stay in a hospital. In many countries, the
number of police-reported road crash injuries is often unrealistically low and in such cases hospital
surveys should be conducted to identify the actual number of serious and slight injuries, based on
police or insurance company data.

Damage-only crashes are even less well documented than injury crashes, as few developing
countries require damage-only accidents to be reported. Early crash costings in the UK assumed a
damage-only crash ratio of six to every one injury crash. Subsequent insurance company surveys
increased this ratio and a recent postal survey has concluded that the damage-only ratio should be
more than 17 times that of injury crashes in urban areas of the UK. A nationwide average of 15
damage only crashes to each injury crash is now used in the UK for cost estimation. The
importance of estimating the number of damage- only crashes is shown by the fact that in the
United States (US), the cost of damage-only crashes is estimated to be higher than the cost of all
fatal crashes.

In order not to disregard the cost of damage- only crashes, their frequency will need to be estimated
and a conservative ratio (five or six damage-only accidents for every injury crash) can be used in
the interim until more accurate data is available. In Nepal, a ratio of 5:1 damage-only crashes to
injury crashes was recently used for urban areas and a lower ratio of 2:1 was accepted for rural
areas. The average cost of a personal injury crash is used in cost benefit analysis in the UK, but
this cost also includes expected damage-only crash costs. In developing countries, the average
injury crash cost may need to be factored to compensate for unreported injury crashes as well as
unreported damage-only crashes.

Lost Output
Lost output refers to the contribution RTA victims were expected to make to the economy with
future earnings weighted to present value (with an inflation rate currently in use in the country).
The contribution is usually measured by the average earnings plus any nonwage payments (e.g.,
national insurance contribution or rent subsidy).

Average wage rates are frequently used with assumptions required to reflect the amount of
agricultural labor and employment rates. Crash victim surveys have been conducted in some
countries but are generally on a small scale and often have had conflicting results. Urban pedestrian
deaths are commonly often assumed to be persons of low income yet a recent accident costing
exercise in Kerala, India, estimated victims’ income to be three times that of the local average per
capita income.

The “lost output” of RTA deaths is calculated as the average earnings multiplied by the number of
working years lost (i.e., average retirement age minus the average RTA fatality age) and then
weighted at the accepted government rate to a present day value. Unlike other major causes of

146
death in developing countries, road crashes strike down those in the prime of life (average RTA
fatality age is usually between 28 and 31 years), when they are arguably of the most productive
value to society.

Lost output for serious and slight injuries is the daily earning rate multiplied by the number of days
off work. This is usually derived from hospital and victim surveys. Cost is not normally assigned
to the amount of increased travel time caused by road crashes.

Vehicle Damage Costs

While all property-related damage costs should be valued (e.g., street furniture, guardrails, and
walls), vehicle damage costs are often the only property item valued in developing countries.
Vehicle damage costs include the repair or replacement cost (minus salvage value) insurance
claim, surveyor fees, and any business lost due to the vehicle being out of commission (although
this is rarely counted in developing countries). As the cost to society is being measured, vehicle
damage costs should be calculated net of import duties and sales tax. In the gross output method,
vehicle damage costs are often the largest cost component in analyses.

Insurance claims are the traditional source for vehicle damage costs, but the low rate of insurance
coverage in many developing countries raises questions as to how representative crash claims are.
Compensation is often less than damage costs, as policy coverage may be restricted, with the owner
usually paying the first portion (say US$100) or more of any damage. Insurance claim surveyor
reports, although still limited to insured vehicles, have the advantage of estimating vehicle damage
costs regardless of coverage limitations and often they will report the present value of the car, any
remaining salvage value, and will itemize the cost of labor and replacement parts separately. Where
they are available they provide a good source for estimating damage costs

Vehicle fleet operating companies can also be contacted for cost data but poor record keeping
practices and low levels of safety awareness frequently limit cost data availability. Vehicle fleet
companies should be contacted, however, to estimate the amount of business lost during post-crash
repair time.

Vehicle damage costs can also be collected from crash victim surveys, but this is a slow process.
Surveys were attempted in Nepal, with motorcyclists asked about accident damage costs, but the
results were disappointing. Repair workshops in Nepal were also contacted for accident repair
costs but little data was collected over three months.

Average vehicle damage costs should not be assumed to apply equally to all vehicles involved in
crashes. While pedestrian crashes are a serious problem in many developing countries and
contribute significantly to casualty figures and costs, pedestrian crashes often involve little or no
vehicle damage. Crash reporting systems such as the Transport Research Laboratory’s (TRL)
Microcomputer Accident Analysis Package (MAAP) can be configured to include data on vehicle
damage and can provide the number of reported crashes incurring no vehicle damage.

Medical Costs
Medical costs include emergency medical services, both inpatient and outpatient care, prescription
costs, service fees (X-rays and operations), and rehabilitation costs (including artificial limbs).

147
Medical costs are a difficult area in which to collect data. As they rarely account for more than 5
percent of crash costs, data collection can be kept to a minimum, as even gross errors have
relatively little effect on the overall accident costs.

Often, neither health ministries nor individual hospitals are able to estimate the cost of an inpatient
stay (per night) or an outpatient visit. Bed charges should not be used as they refer only to the
patients’ charge (and for partial service) and do not reflect the amount of subsidy received or the
true cost incurred. As a last resort, a hospital’s annual expenditure can be divided by the number
of inpatient days and outpatient visits. Outpatient visit cost can be assumed to be a quarter that of
an overnight stay, although this does not reflect capital costs. In developing countries, medical
costs do not reflect the reality of the situation, as scarce resources limit the hospital beds and
medical services available. The medical costs alone do not necessarily reflect the actual
opportunity costs.

Administration Costs
Administration costs from road crashes are incurred by the police and the insurance companies.
The situation is complicated, however, by the number of crashes not reported to the police and
those crashes not involving insured vehicles. The responsibility for crash reporting may also be
divided between several police forces or divisions, as in Bangladesh where traffic police report
damage- only crashes and general duty police report injury crashes.

Many developing countries for convenience adopt the ratios determined by research in the UK and
assume administration costs to represent 0.2 percent of the total resource costs in a fatal crash, 4
percent of serious crashes, 14 percent of slight crashes, and 10 percent of damage-only crashes.
An alternative method is to assume a token amount for those crashes involving police and/or
insurance involvement.

Pain, Grief, and Suffering Component

Early estimates of road crash costs focused exclusively on the direct economic costs and did not
attempt to consider pain, grief, and suffering. With the onset of social cost benefit analysis in the
1970s, a notional value for pain, grief, and suffering was included in industrialized countries to
reflect societies’ and the individual’s aversion to death. The original value for pain, grief, and
suffering of a RTA death was determined by the amount required to ensure all lives (no matter
what age and what remaining productivity) received a positive value under the net output method.

In the UK, pain, grief, and suffering values were increased several times throughout the 1970s and
1980s and ended at 38 percent of resource costs of a RTA death, 100 percent of a serious injury,
and 10 percent of a slight injury. While many developing countries have used these values, the
crash costing in India and Nepal have used 20 percent of lost output costs.

With the increasing introduction of the willingness to pay method, the human costs are almost
twice the resource costs in the UK. This method is inappropriate to apply in developing countries,
but does indicate that the calculation and use of a value based on gross output method perhaps still
underestimates the real cost of road crashes in a country.

148
Stages of Development
In order to produce accurate estimates of road crash costs and to incorporate these estimates into
cost benefit analyses, a developing country usually needs to proceed through a number of stages.
The major milestones of introducing and applying road crash costing are provided below, along
with the activity involved:

Accept 1 percent of annual national GDP as a minimum estimate of annual road crash costs.
This working estimate should be used to highlight the costs of road crashes incurred annually and
to emphasize the overall national cost of road crashes that are usually considered as individual
cases with separate financial and personal losses

Conduct crash costing exercise based on gross output method. Crash cost components can be
estimated with economic indicators available for lost output calculations and vehicle damage costs
derived from insurance company and surveyor data. Medical and administrative costs can be
assumed (small percent of total costs) and a conservative estimate initially used for estimating
pain, grief, and suffering (10 percent to 20 percent lost output). Pain, grief, and suffering estimate
should be agreed by a national road safety council or those involved in road accident consequences;
i.e., insurance company representatives, police, doctors, and legal experts. While some technical
assistance may be required for initial accident costing calculations, a local research institute could
be trained to conduct crash costing updates and revisions. TRL Road Note 105 provides guidance
on how to carry out costing in developing countries;

Assess extent of underreporting of injury and damage only crashes. If RTA injury is believed
to be underreported, hospital surveys should be conducted to produce a more accurate estimate of
crash injuries. Key hospitals along major highways should be surveyed using admission registers
(assuming RTA is listed as cause of admission). Table 2 shows the minimal data required by such
hospital surveys. Surveys of insurance and fleet companies should be undertaken to determine
frequency and average cost of vehicle damage in damage- only crashes;

Estimate total annual road crash costs and average crash costs. These estimates will be
determined locally and reflect both reported and unreported crashes, i.e., an average reported injury
crash cost would also include the costs of unreported injury and damage only crashes. These costs
would replace the GDP percentage estimate as the annual national cost of road crashes and should
be used in road safety programs and campaigns;

Conduct crash victim surveys. Crash victim surveys should be conducted to check the accuracy
of cost estimates. Further research on accident costs can be conducted by following up a random
sample (say 100 or so) road accidents by having researchers obtain from police records the names
of those involved in the selected accidents. These individuals or their families can be followed up
to identify how long the victim was in hospital, how many days’ work were missed, average salary,
and cost of vehicle repairs. Such surveys have been used in Sichuan, the People’s Republic of
China (PRC); and Karachi, Pakistan. Pain, grief, and suffering can be reconsidered and victim
stories considered for publicity and educational campaigns;

Introduce crash costs in project cost benefit analyses. Safety impacts should be evaluated based
on crash costs and engineering judgment on the predicted crash change; i.e., effect of road

149
widening on road crashes. Eventually, accident cost savings should be refined according to before
and after crash findings;

Target road safety funding to percentage of national road crash costs. The target of road safety
funding should be based on the national road crash cost estimate with anywhere between 10
percent and 50 percent requested for road safety (note that until recently, Japan allocated 0.6
percent of its annual GNP to road safety after calculating that crashes cost 1.3 percent of annual
GNP; i.e., safety funding allocation was about 50 percent of annual losses); and

Regularly revise estimates and consider willingness to pay method. Road crash costs should
be recalculated every three to four years and updated annually during the interim period by the per
capita GDP growth. Given the difficulty in adopting the willingness to pay method, countries
should first carry out a pilot project if they are considering moving towards such an approach, as
complex surveys are involved.

150
Chapter 8: Post Crash Response
The objectives of the post-crashes response is to improve capacity of the health-care systems to
provide emergency treatments and long-term rehabilitation for crash victims. To date, there is no
data on the response time to the post-crash victims in Nepal. There are a limited number of
ambulances run by the hospitals, private-sectors and social clubs. Prior to the launch of WHO 2004
initiatives to monitor road-safety injuries globally, there was no systematic reporting requirement
on the part of the Ministry of Health to make it accountable towards road-safety.

The role of emergency medical services in minimizing the consequences of road traffic crashes
lies immediately after the crash. The functions of an emergency medical service can be defined as:
 the provision of first aid and medical care to crash victims at the roadside;
 the transport of the victim to a hospital; and
 subsequent provision of a more definitive treatment.

About 50 percent of road traffic deaths happen within 15 minutes of the crash as a result of injuries
to the brain, heart, and large blood vessels. A further 35 percent die in the next 1-2 hours of head
and chest injuries, and 15 percent over the next 30 days from sepsis and organ failure. The time
between injury and initial stabilization is the single most important factor in patient survival, with
the first 30- 60 minutes being the most important.

The most serious injuries resulting from traffic crashes are head, spinal, and internal soft tissue
damage to vital organs. Early treatment and stabilization of these typical crash injuries can enhance
a patient’s timely and full recovery. Delay or well-intentioned but inappropriate first aid, can result
in death or permanent disability. Medical experience around the world has demonstrated that
stabilization of the injured person and hospitalization to a specialist center, within what they
describe as the “golden hour,” increases the patient’s potential for survival and full recovery.

Ambulance services are intended to meet the following needs:

 rapid response to life threatening or serious incidents;
 preservation of life at the scene;
 prehospital life support and patient stabilization; and
 reduction in death and serious injury for accident victims

One of the formal approaches of emergency medical care is to provide immediate first aid and
emergency care at the scene of the incident, then to transport the injured person to the emergency
department as fast as possible. It is there to supply skilled and definitive care, the so-called scoop
and run philosophy.

Another approach, an “informal system,” is found in many situations in countries in the Asian and
Pacific region. Since there is often no organized ambulance service, the injured are picked up by
bystanders or passing motorists and carried to the nearest emergency department by whatever
transport is available, usually without any treatment or first aid at the scene. This results in rapid
transport to the emergency department but without any resuscitation measures. The outcome of
this system depends to some extent on the capacity of the emergency department to deal with these
severe cases and to provide effective treatment on arrival.
151
Data management in hospitals and health care centres regarding RTAs is poor. However, there
have been some limited project-led initiatives in this area in the recent past to develop RTA
statistics within the participating hospitals. The MOHP is planning to introduce injury
surveillances to record RTA cases. The ministry is also contemplating developing three level of
care for crash victims (primary, secondary and tertiary). Nevertheless, priority for RTA cases
remain low compared to other injuries as the former is still not the most predominant form of
morbidity in the country.

The health centres throughout the country are linked to the centre through GPS. In addition, the
Kathmandu Metropolitan Traffic Police plans to install GPS devises on its patrol vehicles provide
opportunities to improve emergency response within the Kathmandu Valley and other areas if the
system is linked with the aim of enhancing post-crash response.

As per Nepal Road Safety Action plan, different activities along with their implementation
Main Activities Proposed Impact Following Implementation

Introduce toll-free telephone number for RTA

emergencies.  Reductions in the RTA fatalities.
 Facilities for users.

Develop ambulance policy for post-accident treatments

and emergency treatment training.  Reductions in post-accident fatalities.
 Improvement in emergency services.

Develop strategy and introduce revolving fund for RTA Relief for low income groups regarding
victims and disabled. rehabilitation, disability.

Open Trauma Care Centers and train for RTA injured.

 Increase facilities for trauma care.
 Develop skilled manpower in trauma-care.

Research and prioritize treatments for serious injuries Reduction of fatalities from predominant RTA
from RTAs. injuries.

Develop, introduce comprehensive injury surveillance

system in hospitals, health centers.  Improve the RTA database.
 Provide scientific basis for prioritizing RTA
injuries, their treatments and research.

Establish road-safety unit in the Ministry of health and

Population and institutionally enhance it.  Create awareness towards road-safety in the
health sector.
 Institutional development of the ministry.

Develop ambulance network along the major Improve ambulance services and their access.
highways, urban and rural roads.

objective are proposed in post-crash response.

152
Chapter 9: Road Safety Education and Public Awareness

9.1 Definition and Importance

9.2 Children Education

On average, 20 percent of all people killed in traffic crashes in developing countries are aged under
15. This is twice as high as in the developed world.
• Human error plays a large part in road accidents, being a contributory factor in about 95
percent of crashes.
• Teaching safety skills to children can provide lifelong benefits to society.

Roads in developing countries are often more unsafe than roads in industrial countries and the
traffic safety problems faced by children will often be greater in the developing world. Absence of
traffic education can leave children exposed to unnecessary risk. Since the traffic circumstances
and problems faced by such children are very different, it is inappropriate to simply use teaching
materials from developed countries. Local materials need to be developed. Although these may be
based on principles and materials from developed countries, they will need to be adapted to reflect
the needs, problems, and circumstances of relevance to local children.

In addition, an incremental approach is needed to improve road safety education. Road safety
provision should not rely on only occasional, isolated talks by visiting speakers but should include
regular practical training. Essential components in developing and improving this sector are as
follows:
• inclusion of road safety in the school curriculum appropriate to each age group;
• development and production of classroom materials;
• production of a teachers‘ guide and dissemination to all teachers;
• inclusion of road safety in teacher training courses; and
• coordination of activity and clearly defined responsibilities.

Children need to be made more aware of road safety and should be taught survival skills
appropriate to their age and needs. Teaching of road safety in schools is best done by teachers who
have themselves been trained on road safety issues and who can provide such instruction on a
regular basis to their students.

9.3 Road safety awareness publicity campaign

153
Publicity is the means through which the population is given information for a specific purpose,
and which contains a message that people can readily act upon. It differs from education in that it
does not involve a face-to-face interaction between the giver of the information and the recipient.
Thus it does not generally allow a dialogue. It involves the imposition of information and opinions
upon a whole population or a subset of that population. Its potential effectiveness depends on a
number of factors, including:

• credibility of the message;

• design of the message;
• implementation of the delivery; and
• the extent to which the country in question has a publicity culture.

Why needed?

As road user error is believed to be a factor in 95 percent of all road accidents, improving road
user behavior should always be a priority. With the ability to educate and influence the general
public, road safety publicity is needed in order to:
• create awareness of road accident threats and vulnerability of certain road users;
• educate road users as to what constitutes safe road user behavior;
• change attitudes and beliefs to a more positive road safety approach; and
• inform road users of changes in traffic regulations or operating conditions.

Key components
Most of the countries in the Asian and Pacific region employ road safety publicity techniques
based upon posters, leaflets, billboards, newspapers, and often radio and television. However, few
of these campaigns appear to be planned in the systematic manner necessary to ensure success.
The key components that need to be considered are:
• Problem Assessment
o Accident data
o Observation
o Police discussions
o Attitude testing
o Knowledge testing
• Campaign Design
• Campaign Message
• Target Audience Selection
• Pilot Testing
• Success Indicators and Methods of Evaluation
• Identify Any Other ‘Players’
• Sources of Funding
• Select Carriers of the Message
• Appeal
• Implementation
• Monitoring and Documenting

154
155
Chapter 10: Miscellaneous
10.1 Introduction to UN decade of action on Road Safety

For many years road traffic crashes have been acknowledged by the United Nations as a
considerable challenge to the achievement of health and development goals. In 2008, the
Commission for Global Road Safety called for a global road safety decade. This idea was
formalized in the Declaration which resulted from the First Global Ministerial Conference on Road
Safety, hosted by the Government of the Russian Federation in November 2009. The Decade of
Action for Road Safety 2011–2020 was proclaimed by the United Nations General Assembly in
March 2010.

The United Nations resolution A/RES/64/255 sets the goal for the Decade "to stabilize and then
reduce the forecast level of road traffic fatalities around the world by increasing activities
conducted at the national, regional and global levels.[7] The resolution calls on all Member States
to set road safety targets to be achieved during the Decade. While governments are expected to
lead on the implementation of activities, the resolution specifically calls for a multi-sectoral
approach that includes academia, the private sector, civil society, the media, victims and their
families.

Global Plan
The United Nations Road Safety Collaboration has developed a Global Plan for the Decade of
Action for Road Safety 2011-2020 as an overall framework for activities which may take place in
the context of the Decade. The categories or "pillars" of activities are: road safety management;
safer roads and mobility; safer vehicles; safer road users; and post-crash response.
Pillar 1 of the Global Plan focuses on the need to strengthen institutional capacity to further
national road safety efforts. It includes activities such as putting into practice major United Nations
road safety conventions; establishing a lead agency for road safety in the country involving
partners from a range of sectors; developing a national road safety strategy; and setting realistic
and long-term targets for related activities with sufficient funding for their implementation. It also
calls for development of data systems to monitor and evaluate activities.
Pillar 2 highlights the need to improve the safety of road networks for the benefit of all road users,
especially the most vulnerable: pedestrians, bicyclists and motorcyclists. Activities include
improving the safety-conscious planning, design, construction and operation of roads; making sure
that roads are regularly assessed for safety; and encouraging relevant authorities to consider all

156
forms of transport and types of safe infrastructure when they respond to the mobility needs of road
users.
Pillar 3 addresses the need for improved vehicle safety by encouraging harmonization of relevant
global standards and mechanisms to accelerate the uptake of new technologies which impact on
safety. It includes activities such as implementing new car assessment programmes so that
consumers are aware of the safety performance of vehicles, and trying to ensure that all new motor
vehicles are equipped with minimum safety features, such as seat belts. Other activities covered
include promoting more widespread use of crash avoidance technologies with proven
effectiveness, such as electronic stability control and anti-lock braking systems.
Pillar 4 focuses on developing comprehensive programmes to improve road user behaviour.
Activities include encouraging the development and adoption of model road safety legislation and
sustained or increased enforcement of road safety laws and standards. These efforts are combined
with public awareness and education to increase seat-belt and helmet wearing and to reduce
drinking and driving, speeding and other risks. It also calls for activities to reduce work-related
road traffic injuries and promotes the establishment of graduated driver licensing programmes for
novice drivers.
Pillar 5 Post Crash Response are to improve the post-crashes response, improve capacity of the
health-care systems to provide emergency treatments and long-term rehabilitation for crash
victims.

10.2 Nepal Road Safety Action Plan

The UN Global Action mandates member countries to develop their individual national plans for
the decade (2011 to 2020) incorporating interventions under the following five pillars to road-
safety.
 Road safety management
 Safer roads and mobility
 Safer vehicles
 Safer road users
 Post-crash response
In addition, the Global Action Plan recommends countries to develop their national action plans
for the decade in a manner that is consistent or can be carried forward to the regional plans.
In accordance with UN mandate, Nepal has also drafted national action plan on road-safety in line
to the Global Action Plan. All stakeholders are obliged to follow this action plan to improve and
manage road-safety in an integrated manner. It also sets out the activities that concerned agencies
need to implement to achieve the desired goals of reducing road traffic injuries and resultant
economic losses in Nepal. As the first national action plan, it is anticipated that GoN will revise
and update it in the future as necessary.

157
Safety Issues in Nepalese Roads
There are numerous safety issues on the Nepalese hill roads (which form a substantial portion of
the road network) such as poor visibility at blind corners; poor shoulders; unforgiving side-drains,
inadequate safety barriers at steep vertical drops; unscientific location of passing bays in single
lane roads; lack of climbing lanes; very steep gradients at numerous sections, narrow sections at
built-up areas, etc.. Along the roads in the plains (Terai), unforgiving drains, inadequate pedestrian
provisions, inadequate delineation at bridge/culvert crossings, narrow carriageway at build-up
areas, etc., are the predominant safety issues.

Road Safety Strategy

The detailed formulation of the road-safety strategy is proposed as one of the activity of this action
plan. However, the following is a brief outline of the strategy that will emerge.

Vision: Safe road-infrastructures and services backed with effective post-crash response and
conducive environment resulting in little or no casualties from the RTAs.

Mission:
(i) To mitigate the loss of life, properties and economic loss from RTAs.
(ii) To complement the broader mission of the National Strategy on the Prevention and Control of
Violence, Injuries and Disabilities
(iii) To meet the targets of the UN Decade of Action.
(iv) To provide a common framework for stakeholder agencies to implement the various
interventions required to mitigate RTAs outcomes.

A reflection of the background information of Nepal section shows the following facts:
 Both RTA fatalities and injuries are increasing.
 Road-users are most exposed to RTAs in the Kathmandu Valley but casualties are less
severe compared to other regions.
 RTAs are under-reported and analysis inadequate.
 Light vehicles, especially motor-cycles, dominate the vehicle fleet while mass
transportation is lacking.
 Pedestrian, population within 15- 40 years are most vulnerable users.
 There is a high number of motor-cycle accidents in the urban areas.
 Single bus accidents represent the RTAs with the most severity in the rural areas.
 RTAs are mainly caused by reckless driving and pedestrian recklessly crossing the streets.
 Safe pedestrian access is lacking aggravating pedestrian-safety.
 Safe-design (forgiving road) is lacking.
 Stakeholders’ interactions are ad-hoc and responsibilities are often duplicated.

158
Upon implementation of the various interventions proposed under the action plan, it is expected
that there will be positive impacts to various aspects concerning road-safety. There will be some
challenges as well as risks if proposed interventions are not implemented in the right manner. It is
necessary to take into account these risks when implementing the interventions.

159
160
161
162
163
164
165
166
167
The following outlines the national road- safety action plan for the period 2013 – 2020. These
plans are the product of one-on-one meetings, focused group discussions, correspondence with all
the stakeholders, discussion workshops, etc., and was finalized after reaching consensus during
the final workshop organized in December, 2011.

Activity for Pillar 1: Road Safety Management

Objectives: Set up a mechanism to improve capacity to manage road-safety through
 Adopt UN legal instruments
 Encourage creation of regional road-safety instruments
 Improve horizontal coordination amongst stakeholders
 Develop sustainable road-safety strategies and accident reduction targets
 improve accident data collection and research

Activity for Pillar 2: Safer Roads and Mobility

The objective of this pillar is to improve the inherent safety of the road networks for all road-users,
especially the most vulnerable groups (e.g. pedestrians, bicyclists and motor-cyclists). This will
be achieved through:
 Adoption of UN and international standards for the design of safe roads.
 Road safety audits and assessment
 Incorporating safe design practice during design, construction and operation of roads.

Activity for Pillar 3: Safer Vehicles

The objective of this pillar is to promote the universal adoption of both the active and passive
technologies that are available for safe vehicles through the harmonization with the global
standards, publicity and incentives for the consumers in their adoption.

Activity for Pillar 4: Safer Road Users

The objective of this pillar is to develop comprehensive programmes to improve road-user
behavior through the following activities:
 Sustained, stronger enforcement of traffic rules.
 Sustained road-safety awareness campaigns.
 Increased efforts to improve the use of seat-belts and helmets.
 Reduce drunk-driving and other risky behaviours
 Introduce better speed control
 Heavy penalty to undisciplined road-users including pedestrians.

Activity for Pillar 5: Post Crash Response

The objectives of this pillar are to improve the post-crashes response, improve capacity of the
health-care systems to provide emergency treatments and long-term rehabilitation for crash
victims.

168