Elsevier

Accident Analysis & Prevention

Volume 60, November 2013, Pages 277-288
Accident Analysis & Prevention

A tool for safety evaluations of road improvements

https://doi.org/10.1016/j.aap.2013.04.008Get rights and content

Highlights

  • Accident record alone is a misleading basis for targeting safety problems.

  • Proper EB tools result in reliable estimates on road safety improvements.

  • Such tools also help estimate the current road safety situation.

  • Tools with similar bases and principles improve the transferability of safety results.

Abstract

Road safety impact assessments are requested in general, and the directive on road infrastructure safety management makes them compulsory for Member States of the European Union. However, there is no widely used, science-based safety evaluation tool available. We demonstrate a safety evaluation tool called TARVA. It uses EB safety predictions as the basis for selecting locations for implementing road-safety improvements and provides estimates of safety benefits of selected improvements. Comparing different road accident prediction methods, we demonstrate that the most accurate estimates are produced by EB models, followed by simple accident prediction models, the same average number of accidents for every entity and accident record only. Consequently, advanced model-based estimates should be used. Furthermore, we demonstrate regional comparisons that benefit substantially from such tools. Comparisons between districts have revealed significant differences. However, comparisons like these produce useful improvement ideas only after taking into account the differences in road characteristics between areas. Estimates on crash modification factors can be transferred from other countries but their benefit is greatly limited if the number of target accidents is not properly predicted. Our experience suggests that making predictions and evaluations using the same principle and tools will remarkably improve the quality and comparability of safety estimations.

Introduction

Directive 2008/96/EC on road infrastructure safety management requires the establishment and implementation of procedures relating to road safety impact assessments by the Member States of the European Union (European Parliament, 2008). However, science-based safety evaluation tools are not widely used. Based on a comparison of definitions of hazardous road locations (also called black spots or hotspots) in eight countries, Elvik (2008a) argued that “currently applied operational definitions of hazardous road locations in most of the European countries included in this survey are not close to the state-of-the-art and are in need of considerable development in order to approach the state-of-the-art.” Furthermore, Elvik (2008b) and Montella (2010) have concluded that the empirical Bayesian (EB) method should be a standard in identification of hazardous location. Identification should be based on best estimate of accidents and it should be achieved by combining information from accident records and accident prediction models (Elvik, 2008b). Consequently, if the allocation of road safety improvements is not based on scientifically justified tools, one cannot expect safety work to be as effective as it should be.

The expected number of accidents after implementing a measure can be calculated as a product of the predicted number of accidents in the future without the measure (target accidents) and a crash modification factor (CMF, also called impact coefficient or accident modification factor) describing the effect of the planned measure. For example, a 0.9 CMF corresponds to a 10% reduction in the number of accidents.

The effects of measures are not always the same, but are highly dependent on the circumstances and implementation of the measure. Hence the term crash modification function has been proposed instead of crash modification factor (OECD/ITF, 2012, Elvik, 2009). CMFs are a proper way of transferring information on traffic safety effects if the quality of the safety studies and transferability of effects are properly taken into account (OECD/ITF, 2012).

CMF values should be based on reliable results from safety effect studies conducted in similar conditions to those where the CMFs are used. Reliable results can be obtained from well-controlled before-after studies, including a proper comparison group and controlling for e.g. regression-to-the-mean effect (Gross et al., 2010, Hauer, 1997).

Information about CMFs can be found online from the FHWA CMF clearinghouse (FHWA, 2012) and Austroads Road Safety Engineering Toolkit (ARRB, 2012). Information can also be obtained from publications such as the book by Elvik et al. (2009).

As mentioned above, the expected safety effect depends not only on the CMF estimate but also on the expected future number of target accidents, a quantity that is also insufficiently known. In some cases errors in target accidents can be substantial, even more so than errors in CMF. Probably no one uses a CMF suggesting a reduction of more than 100% of accidents, but that kind of error can occur when relying on accident records only. However, it is not infrequent for the estimation of future expected number of target accidents to be based on accident records with no modelling.

Consequently, in order to maximize efficient use of existing reliable safety knowledge, scientifically well-founded safety evaluation tools for road improvements are needed. The aim of this paper is to present specific software called TARVA that fits those needs. Tests conducted with TARVA, such as comparison of road safety between selected regions, are also described. Comparisons with other available methods are presented to demonstrate how important it is to use valid tools. Finally, the main conclusions are presented.

Section snippets

Tool for evaluating current safety and safety effects of road improvements

In this section, we present the principles of the TARVA programme along with specific calculations and coefficients calibrated using Finnish highway accident data from 2007 to 2011. Finally we explain the use and selected results of TARVA and its use for comparing the safety of selected regions.

Comparing regional road safety with TARVA

Road safety comparisons between regions are regularly performed in order to learn from the successes of the best performing regions, but without a proper approach the conclusions may be erroneous. Specifically, comparisons seldom take into account disparities between road categories, making it hard to identify where the differences originate. In the following, we demonstrate two comparisons that benefit from tools like TARVA.

The first case includes a scenario of investing funds for the safety

Comparison of alternative methods for accident prediction

This section aims to demonstrate how important it is to use valid tools such as TARVA. There are two comparisons that investigate to what degree relatively simple models, combined with accident records, predict future numbers of accidents on road and level crossings. In both comparisons the studied accident prediction methods were: (1) accident prediction model, (2) accident model combined with accident record using the EB method, (3) same average number of accidents for every entity, and (4)

Conclusions

We demonstrated a safety evaluation tool called TARVA. It uses EB safety predictions as the basis for selecting locations for implementing road-safety improvements. In addition, TARVA provides estimates of safety benefits of selected improvements. We compared the EB estimation method used by TARVA with predictions produced by the simple model alone and the accident record, which is currently quite a common estimation method.

The results showed that the most accurate estimates are produced by EB

Acknowledgements

The TARVA projects were funded by the Finnish Transport Agency, the Finnish Transport Safety Agency and the Lithuanian Road Administration in co-operation with Vilnius Gediminas Technical University. The authors wish to thank Panos Papaioannou, Risto Kulmala, Peter Hollo and the two reviewers for their valuable suggestions on an earlier draft of this paper.

References (15)

There are more references available in the full text version of this article.

Cited by (11)

  • Identifying crashes potentially affected by conditionally automated vehicles in Finland

    2023, Journal of Intelligent Transportation Systems: Technology, Planning, and Operations
  • Methodology for safety improvement programming using constrained network-level optimization

    2015, Transportation Research Part C: Emerging Technologies
    Citation Excerpt :

    In the paper, the researcher presents an approach to optimize the safety benefits in a given jurisdiction by maximizing the dollar value of the crashes saved at intersections each year over a multi-year planning horizon while predicting crashes simultaneously during the optimization procedure. Peltola et al. (2013) discussed a similar model, TARVA, a tool that estimates the traffic safety effects of road improvements in Finland and also predicts the expected number of road accidents if no measures are implemented. TARVA can be used for selecting locations for safety treatments and also to estimate the costs and safety benefits of road improvements, ultimately, evaluating the cost-effectiveness of combinations of safety improvements.

  • Safety assessment of Czech motorways and national roads

    2019, European Transport Research Review
View all citing articles on Scopus
View full text