Introduction
The problem of collisions at railway crossings is an on-going
one for rail operators, track providers and regulators and road
authorities in Australia. While the number of deaths and injuries
is small in comparison to other road casualties and has been
reduced considerably in recent years, they are the most serious
safety issue faced by the rail system in Australia.
The genesis of the present project was at a special meeting of
the Australian Transport Council (ATC) on 8 August 2002 that
considered the outcome of recent tests of locomotive auxiliary
lighting. It was agreed at that meeting that the SCOT Rail Group,
in consultation with the rail industry, develop a strategic
approach to managing the full range of level crossing issues. It
was further agreed that Austroads and Rail Group were to review
available research on train lighting and visibility and report back
to ATC at its meeting on 8 November 2002 on the need for any
further research. The Australian Transport Safety Bureau was
required to produce this review on behalf of Austroads and Rail
Group, and has commissioned ARRB Transport Research Ltd to
undertake the work.
The essential purpose of the present report is therefore to
advise on the need for, the feasibility of, and the potential
benefits of, further research into train lighting and conspicuity
that will deliver significant reduction in road safety trauma.
Number of collisions and their associated costs
In the period 1996-2000, it is estimated that approximately 36
crashes per year occurred at passive crossings throughout
Australia. These crashes resulted in an average of four deaths and
six serious injuries per year. The average annual cost of
collisions at railway level crossings was estimated to be at least
$24.8 million for all crossings, including $16.3 million for active
crossings and $8.3 million for passive crossings. As fatality data
for NSW were not available, these estimates were based on the
assumption that the distribution of injury outcomes for NSW is
similar to that in other states. The estimates are regarded as
minimums, since the data for South Australia, Western Australia and
the Northern Territory were incomplete. They are based on personal
injuries recorded in the road crash system and do not reflect the
major losses to the rail system that can occur as a result of a
collision with a road vehicle. Rail system loss data are not kept
in a systematic manner by all rail operators. Data that were
obtained indicated very high losses associated with some incidents.
The fatality rate at level crossings per 100,000 population is
considerably lower in Australia than in the United States and
Finland.
Case for improving train conspicuity
Since there are fewer locomotives (approximately 2300) than
passive crossings (approximately 6000), and since locomotive
lighting treatments are likely to cost less than even the
low-budget active warning systems currently being trialled,
treating locomotives appears to be an attractive option. Increasing
the conspicuity of locomotives would cost far less than providing
active treatments at all passive crossings. However, there is
presently insufficient research evidence to estimate the proportion
of collisions at passive crossings that would be prevented by such
treatments. While available data suggests that active warnings
would reduce crashes by more than 60 per cent (Schulte 1976), it is
not possible to say by how much increased conspicuity would reduce
collisions.
Vehicle/train collisions
Under Australian conditions, approximately 70% of collisions
occur during daylight and 30% occur at night. Daytime collisions
also predominate in the US, but the difference between daytime and
night-time crash occurrence is less marked than in Australia.
In Australia, approximately 65% of crashes involve trains
running into road vehicles and 35% involve road vehicles running
into the side of trains.
Seven contributing factors related to the driver of the road
vehicle have been identified:
- Not detecting the crossing
- Stalling
- Not detecting the train
- Being distracted
- Inaccurate expectancies
- Deliberate risk taking
- Misjudging train speed.
There are very few cases of stalled vehicles on the tracks, and
deliberate risk taking is not possible unless the driver has
already seen the train. The important contributing factors are, on
the one hand, not detecting the train, to which distraction and
inaccurate expectancies regarding the presence of a train may
contribute; and on the other hand, misjudging the speed of the
train. It is not known to what extent each of these factors
contributes to collisions at railway level crossings. It seems
inherently unlikely that adding more lights to the train would
result in more accurate (or at least more cautious) perceptions of
train speed. This leaves cases which involve not detecting the
train, distraction and expectations that a train will not be
present as events where adverse outcomes could be avoided by better
train conspicuity. Unfortunately, it is not possible to say what
proportion of cases this involves, or by how much increased
conspicuity (assuming effective increases in conspicuity were
possible) is likely to reduce this.
Lighting standards
The Australian Rail Operations Unit, in consultation with the
rail industry, is in the process of developing a draft Code of
Practice for the Defined Interstate Rail Network to provide uniform
guidance for the design and construction of rolling stock. The
relevant volume of the Code includes provisions for lighting and
reflectorised material on trains. The draft code has mandatory
provisions for headlights and 'road visibility lights', which are
similar to crossing lights. Reflectors are optional, but where
provided the reflective sheeting must be to Class 1A standard and
there are mandatory provisions for minimum dimension, placement and
colour.
The present report discusses characteristics of different
lighting systems and puts forward an outline for the development of
photometric models to predict the relative visibility of different
lighting treatments.
Empirical studies of the effectiveness of auxiliary lighting
treatments are reviewed. There is evidence to suggest that all
auxiliary lighting treatments are effective and increase
detectability or improve estimations of time to arrival compared to
headlights alone. A study for the US Federal Railroad
Administration showed that crossing lights were the most effective
treatment. Studies have also shown that strobe lights can improve
detection when added to locomotives previously equipped with
headlights alone. However, a recent study for Western Australian
Government Railways indicated that a single strobe light did not
improve detection when added to locomotives already fitted with
both headlights and crossing lights.
Possible means of improving train conspicuity
It is possible that daytime crashes might be reduced by adding
coloured strobe lights or by selection of colour schemes which
better contrast with the background against which locomotives on
particular lines are viewed.
Although reflectorised panels may be effective in improving
conspicuity, they may not be effective at crossings where the road
does not cross the rail track at right angles. Selfilluminated
devices, similar to those currently available as road delineators
(also known as cats eyes), may be a viable alternative.
Conduct of future research
If it is decided that the scale of the problem and the potential
benefits warrant further research, then it is recommended that the
following procedure be followed. Future research should proceed
first by careful modelling of the photometric properties of
proposed conspicuityenhancing treatments. Only once there is a
solid case established that a treatment has a high probability of
succeeding should any field work be undertaken. The photometric
modelling will probably have to be supplemented by photometric
measurements and some tests with subjects using real-life
visibility aids on a static locomotive to provide data for the
modelling process, resolve issues which cannot be resolved
theoretically and confirm the predictions of the models. The cost
of the photometric modelling exercise is estimated at $75,000 and
the supplementary program of measurements and tests at $15,000 to
$45,000. A study of the safety margins allowed by drivers when
crossing in front of a train would cost approximately $105,000,
including equipment and a pilot study to confirm the feasibility of
the method. The total program of recommended research would cost up
to $225,000.
Evaluation of the effectiveness of conspicuity treatments in
terms of crash reductions will not be practical, due to the small
number of crashes available for comparison, unless the proportion
of crashes prevented by the treatment is exceptionally high.