iRAP Road Attribute Risk Factors

Curvature

This factsheet describes the road attribute risk factors used in the iRAP methodology for Curvature.
Curvature is a measure of the horizontal alignment of a road.

About risk factors

Risk factors, sometimes called crash modification factors (CMF), are used in the iRAP Star Rating methodology to
relate road attributes and crash rates. Risk factors (or CMF) are described by the Crash Modification Factor Clearing
House as follows:
A crash modification factor (CMF) is a multiplicative factor used to compute the expected number of crashes
after implementing a given countermeasure at a specific site.
For example, an intersection is experiencing 100 angle crashes and 500 rear-end crashes per year. If you
apply a countermeasure that has a CMF of 0.80 for angle crashes, then you can expect to see 80 angle
crashes per year following the implementation of the countermeasure (100 x 0.80 = 80). If the same
countermeasure also has a CMF of 1.10 for rear-end crashes, then you would also expect to also see 550
rear-end crashes per year following the countermeasure (500 x 1.10 = 550).

Related documents
This factsheet should be read in conjunction with:
 Star Rating Roads for Safety: The iRAP Methodology.
 Safer Roads Investment Plans: The iRAP Methodology.
 Star Rating and Investment Plan Coding Manual.

 Road Safety Toolkit (http://toolkit.irap.org).

Risk factors
Risk factors by road attribute category, road user type and crash type

Vehicle occupant Motorcyclist Pedestrian Bicyclist

Curvature Head-on Head-on
Run-off Run-off Along ** Along ** Run-off
LOC * LOC *
Straight or gently curving 1.0 1.0 1.0 1.0 1.0 1.0 1.0
Moderate curvature 1.8 1.8 2.0 2.0 1.8 1.8 1.8
Sharp curve 3.5 3.5 3.8 3.8 3.5 3.5 3.5
Very sharp 6.0 6.0 6.5 6.5 6.0 6.0 6.0

* LOC: loss of control. ** Risk of being struck by a vehicle or motorcycle.

Selection of risk factors
It is widely-recognised in the research literature that the curvature of a road is related to the risk of death or serious
injury. Recent reviews by ARRB Group and Austroads have been particularly influential for risk factors used by iRAP
(Turner et al., 2010). Taking the chart below as an example, the Curvature risk factors align with the following radius
ranges:
 straight or gently curving: greater than 900m
 moderate curvature: 500m to 900m

 sharp: 200m to 500m

 very sharp: 0m to 200m.
The risk factors for motorcyclists are inflated by around 10% to reflect the relatively lower stability of two-wheel
vehicles. Motorcyclists might also take bends at higher speeds than other drivers, thereby increasing their risk. The
pedestrian and bicyclist ‘along’ risk factors relate to the risk of these road users being struck by passing vehicles.
Crash rate by horizontal curvature radius (Austroads, 2010)

Background research and model development

Lynam (2012) explained the research background to the values used in earlier versions of the iRAP methodology. In
general, crash studies show that horizontal curves experience higher crash rates than straight road sections, with
rates varying from 1.5 to 4 times greater than the tangent sections in the range of bends they studied (Zegeer et al.,
1992). These additional crashes are mainly run-offs although some will also be head-on crashes where vehicles stray
over the centre-line of the bend. Wright and Robertson (quoted in Mak and Sicking, 2003) analysed 300 single-
vehicle, fixed-object fatal crashes in Georgia in an attempt to determine encroachment rates at bends and on
gradients by comparing the characteristics of the crash sites with controls 1 mile upstream of the crash sites. Bends
were significantly over-represented at the fatal crash sites, with the outside of the bend accounting for 70% of the fatal
crashes on bends. Downhill gradients of 2% or more were also found to have some effect.
British data (Hughes et al., 1997) suggest that on single carriageway roads an increase in bendiness of 1 degree per
km is associated with a 1% increase in crashes, while on dual carriageways, curves of 25-70 degrees and 70-90
degrees have about 20% and 250% more crashes than road sections with less than 25 degrees of curvature. Taylor
et al. (2002) suggest that the presence of severe bends (that is, those marked with a chevron - which are likely to be
greater than 60 degrees) on single carriageway rural roads each add 30% to the crash rate for one km of road. There
is additional supporting data from Walmsley and Summersgill (1998) for national roads in England, and from Lamm
(1999) for German roads.
Elvik and Vaa (2004) conclude that straightening curves reduces crash frequency if the curve radius is less than
2,000m, suggesting that replacing a curve of less than 200m radius by one of between 200m to 400m reduces
crashes by 50%; other examples are 200m to 400m radius increased to 600m giving an crash reduction of 33%;
400m/600m to 600m/1,000m giving a 23% reduction; and 600m/1,000m to 1,000m/2,000m giving an 18% reduction.
Although these data suggest a fairly consistent picture of the average effect of changing road curvature, other local
factors can have a major effect on crash frequency at a particular bend. Such factors generally reflect the extent to
which drivers fully anticipate the nature of the bend. A driver encountering a relatively sharp isolated bend on an
otherwise fairly high speed road is more likely to make a mistake in assessing the risk at the bend than a driver
negotiating one of a series of bends which have already caused a slower speed to be adopted. The driver might be
given warning of the nature of the bend in various ways; iRAP has attempted to take some account of this by including
a “quality of curve“ road attribute, but data on which to base a risk factors for this is very limited.
Turner et al. (2009) described the extensive data (described above) generated by the ARRB Group crash rates
database and other published literature appears to support these findings (see for example, McLean (1996) who
reviewed the TRB, Zegeer et al., Krammes et al. (1995) and New Zealand crash prediction models and compared
them in terms of the predicted crash risk on a curve of given radius relative to that on tangents).
In version 2.2 of the iRAP model, Curvature was described by four categories as shown below. It influenced the
vehicle occupant, motorcyclist and bicyclist models. The same values were used for urban, rural and semi-urban
roads.
Risk factors in earlier versions of the iRAP model

Curvature Vehicle occupants, motorcyclists and bicyclists

Straight or gently curving 1.0
Moderate curvature 2.2
Sharp curve 3.5
Very sharp 6.0

Primary references
The following publications are the primary references used in the selection of the iRAP road attribute risk factors. A
complete list of citations is available in: iRAP Road Attribute Risk Factors: Full Reference List.
Elvik, R, Hoye A, Vaa, T, and Sorensen, M. (2009). The Handbook of Road Safety Measures, Second Edition (2009)
Emerald Group Publishing Limited. ISBN 978-1-84855-250-0.
Lynam, D (2012). Development of Risk Models for the Road Assessment Programme. RAP504.12 and TRL Report
CPR1293, Published by iRAP and TRL and available at: http://www.trl.co.uk and at http://www.irap.org.
Mak, K. and Sicking, D. (2003). Roadside Safety Analysis Program – Engineer’s Manual. Transportation Research
Board (TRB) National Cooperative Highway Research Program (NCHRP) Report 492. ISBN 0-309-06812-6.
Turner, B. Steinmetz, L., Lim, A. and Walsh, K. (2012). Effectiveness of Road Safety Engineering Treatments. AP-
R422-12. Austroads Project No: ST1571.
Turner, B., Affum, J., Tziotis, M. and Jurewicz, C. (2009). Review of iRAP Risk Parameters. ARRB Group Contract
Report for iRAP.
Turner, B., Imberger, K., Roper, P., Pyta, V. and McLean, J. (2010). Road Safety Engineering Risk Assessment Part 6:
Crash Reduction Factors. Austroads AP-T151/10. ISBN 978-1-921709-11-1.
University of North Carolina Highway Safety Research Center and U.S. Department of Transportation Federal
Highway Administration (2013). Crash Modification Factors Clearing House: http://www.cmfclearinghouse.org/.
31 May 2013