Final Publishable Report

CONTRACT N° : GRD2/2001/50088/RISER/S07.15369 PROJECT N° : GRD2/2001/50088 ACRONYM : RISER

TITLE : Roadside Infrastructure for Safer European Roads

PROJECT CO-ORDINATOR: Chalmers University of Technology

PARTNERS : Chalmers University of Technology (Chalmers) SE Vehicle Safety Research Centre (VSRC) UK Centre d’Études du Techniques de l’Équipement (CETE) FR European Union Road Federation (ERF) BE Fundación para la Investigación y Desarrollo en Transporte y Energía (CIDAUT) ES Technical University of Graz (TUG) AT Hierros y Aplananciones S.A (HIASA) ES Helsinki University of Technology (HUT) FI The Netherlands Organization for Applied Scientific Research (TNO) NL Volkmann & Rossbach GMBH & CO.KG (V&R) DE

PROJECT START DATE : 2003-01-01 DURATION : 36 Months

Date of issue of this report : 2006-07-14

Project funded by the European Community under the ‘Competitive and Sustainable Growth’ Programme (1998-2002)

Executive Summary

The Roadside Infrastructure for Safer European Roads (RISER) was a 5th Framework "Growth" project co-sponsored by the European Commission Directorate General for Transportation and Energy (DG-TREN). The project started 1 January 2003 and had a duration of 36 months. Ten contractors participated in the project, representing nine European countries.

The project had three specific objectives that addressed the issue of roadside safety, that is to say, the risks and consequences of vehicles leaving their travel lane and experiencing a violent event in the areas bordering the roadway. Often referred to as single vehicle accidents (SVA) or run-off-road crashes (ROR), these events represent about a third of road fatalities in the European Union member states. The main objectives of RISER were to:

1. develop a database(s) with information describing run-off-road crashes. 2. analyse the collected data to provide engineering and human factors links between the roadside infrastructure and safety and operational issues. 3. develop a set of best practice guidelines that will improve the state of roadside safety in Europe.

The RISER project has produced two separate accident databases that address the frequency and general circumstances of ROR crashes (approximately 265,000 cases) and specific driver, vehicle, and road infrastructure information (211 cases). These two databases are significant contributions to roadside safety that have not been previously collected under the framework of one project. From these databases, technical information was produced to indicate the risk of accidents with different types of roadside infrastructure, important information defining the configurations of ROR crashes, the response of drivers to roadside objects, and the response of infrastructure to different crash configurations in the real world or laboratory experiments.

A number of technical reports were generated by the RISER project to describe the different aspects of the project. The most important reports were deliverables D06: European Best Practice for Roadside Design: Guidelines for Roadside Infrastructure on New and Existing Roads and D08: European Best Practice for Roadside Design: Guidelines for Maintenance and Operations of Roadside Infrastructure. These two documents summarize the findings into documents that are suitable for road administrations, road safety researchers, road operators, and manufacturers of road equipment.

Table of Contents

EXECUTIVE SUMMARY...... II OBJECTIVES OF THE PROJECT...... 4 IDENTIFICATION AND COLLECTION OF DATA FOR THE RISER PROJECT ...... 6 DEVELOPMENT OF STATISTICAL DATABASE...... 6 DEVELOPMENT OF DETAILED DATABASE...... 7 COLLECTION OF ROAD OPERATOR DATA ...... 10 COLLECTION OF TEST DATA ...... 11 ROADSIDE INFRASTRUCTURE INFLUENCE ON DRIVER BEHAVIOUR ...... 13 Literature Review ...... 13 COLLECTION OF CURRENT DESIGN GUIDELINES...... 14 COLLECTION OF CURRENT MAINTENANCE GUIDELINES ...... 17 ANALYSIS OF ACCIDENT DATA/STUDY OF HUMAN PERFORMANCE...... 19 SUMMARY ...... 19 ANALYSIS OF STATISTICAL DATA ...... 19 ANALYSIS OF DETAILED DATA...... 21 Struck Infrastructure...... 22 SIMULATION STUDY...... 25 Accident reconstruction: PC-Crash...... 25 Accident reconstruction: Finite Elements...... 27 Accident reconstruction: MADYMO ...... 27 Modelling of vehicle infrastructure collisions:...... 29 Simulation matrix ...... 30 Driver Simulator Study ...... 30 SIMULATOR STUDY RESULTS ...... 34 DEVELOPMENT OF BEST PRACTICE GUIDELINES FOR ROADSIDE INFRASTRUCTURE. 36 DESIGN GUIDELINES ...... 36 Obstacle Identification...... 37 Safety Zone Criteria ...... 39 Recovery Zone Criteria ...... 40 Hazard Evaluation, Removal and Modification...... 42 Hazard Protection ...... 44 Monitoring ...... 45 MAINTENANCE AND OPERATIONS GUIDELINES...... 49 Routine Inspections ...... 49 Data Collection and Analysis ...... 50 Repair Plan ...... 50 Training ...... 52 LIST OF PROJECT DELIVERABLES...... 53 COMPARISON OF INITIALLY PLANNED ACTIVITIES AND WORK ACTUALLY ACCOMPLISHED...... 54 MANAGEMENT AND CO-ORDINATION ASPECTS...... 56 RESULTS AND CONCLUSIONS ...... 57 ACKNOWLEDGEMENTS ...... 58 REFERENCES ...... 60

iii

Objectives of the project

The number of fatal and serious injuries related to the occurrence of a single vehicle leaving the road is a significant portion of European road casualties. The range of casualty statistics varies from country to country, but the European average of single vehicle collision fatalities is one third of annual road fatalities[1]. The outcome of these accidents was highly dependent on the interactions between the vehicle and the roadside environment. The vision of the RISER project was to develop a knowledge base that can provide better roadside design tools and strategies as available resources were conspicuously incomplete. The objectives of the project were to provide a technical foundation upon which the implementation and operation requirements for European roadside areas and infrastructure could be based. Specific objectives were:

1. to identify all possible information sources that may be exploited to document run-of-road collisions. From this information databases for single vehicle collisions should be developed. In conjunction with database development, suggestions to streamline future data collection activities were a crucial output for the project. Two levels of data were envisioned: 1) a statistical database describing all types of single vehicle accidents to quantify the existing problems and 2) a database for detailed accident investigations to extract the important information (impact speed, impact angle, behavioural causation) necessary for the development of countermeasure strategies. 2. to develop a link between the standardisation activities, under CEN TC/226, and real world collision performance. This link was identified as the documentation of infrastructure behaviour in real world collisions for comparison to standardised testing required for road restraint systems. This connection was necessary for an end-user to interpret the specifications for new or existing systems. 3. to develop analysis procedures that can be applied in the design and operation of the roadside. These procedures are necessary tools for providing objective and technical support in decision making, as well as problem identification, procedures.

As a European Commission Research and Technical Development shared cost project, an important aspect of the project was to lift the level of European competence in roadside safety. Information systems and analysis procedures were developed to support the further improvements in the road system. Two critical innovations were anticipated 1) development of a detailed collision database for roadside collisions and 2) development of a catalogue of infrastructure performance. The first item required supplementary activities to ongoing EU activities in collision database due to the difficulty in connecting the currently available collision information with specific roadside features. The second item employed the developing simulation technology with crash test results so that a full picture of infrastructure performance could be achieved.

Through the development of information systems (like accident databases) and infrastructure cataloguing/evaluation through simulation, European safety experts could be trained for analysing roadside crashes.

The objectives would be met through the development of specific project outputs or deliverables. Interim deliverables were designed for the project progress, but the primary project outputs could be identified as:

1. a collision database containing information on single vehicle collisions exploiting existing and new data sources. 2. technical performance data for roadside infrastructure describing the physical interaction of vehicle and roadside in addition to the human factors influencing the accident events. 3. best practice guidelines for designing roadside environments including road safety audit approaches. 4. best practice maintenance guidelines identifying the operation and maintenance necessary to ensure adequate safety levels.

Details of the project activities that fulfilled the project objectives are described in the following chapters.

Identification and Collection of Data for the RISER Project

Summary Several sources of information were identified at the start of the project to help understand the issues of single vehicle collisions. These information sources included police, hospital, and insurance data, information collected by road operators and concessionaires, and previous reports or studies. This chapter discusses the different information sources investigated, how the information was modified to be exploited in the RISER project, and relevant results from the collection activities. Further analysis of this material and the spin-off studies conducted within the RISER project are described in the following chapter.

Development of Statistical Database The first task in the project was to identify sources of statistical level data for single vehicle collisions i.e. police data. The RISER statistical database consists of accident data from seven countries, Austria, Finland, France, Spain, Sweden, the Netherlands and the United Kingdom. The involved partners and the number of cases contributed from each country are presented in Table 1. In most countries, the accident data are reported by the police to the national road authorities. The national databases in the different countries are explained in Deliverable 1 [2].

Table 1. Number of cases from each contributor to RISER Statistical Database Number of Country Participant accidents Austria TUG 17 000 Finland HUT 5 000 France CETE 52 000 Spain CIDAUT 68 000 Sweden Chalmers 11 000 The Netherlands TNO 87 000 The United Kingdom VSRC 24 000 Total number of accidents 264 000

The primary focus in the RISER project was to reduce fatal and injury accidents in the roadside area. Therefore the accident selection criterion focused on single vehicle accidents. The criteria for the cases in the RISER statistical database include events where one vehicle run-off the road on rural roads.

The following selection criterion was used for all cases in the database:

1. Single vehicle accidents only (passenger cars, trucks and motorcycles) 2. No pedestrian involvement 3. Year 1999-2002 4. Fatal, Serious, Slight or Non-injury accidents 5. Minor rural roads excluded 6. Urban roads excluded

The variables included in each national database were review by Chalmers and a dataset for RISER statistical database was proposed. The proposal was set to 20 common variables, 14 variables related to the accident, vehicle and road environment and 6 variables related to the road user.

Crash Data Casualty Data ° Date ° Person Class ° Time ° Age ° Road Type ° Gender ° Carriageway Type ° Alcohol ° Road Condition ° Injury Severity ° Weather ° Seatbelt Usage ° Speed Limit ° Light ° No Of Vehicle Occupants ° Accident Type ° Road Alignment ° Hit Object ° Vehicle Type ° Deformation Location

The criterion to include a variable was that it should be of interest for RISER and more than one country could report it. All important variables connected to road infrastructure were chosen. Sub-variables and codes for each variable were selected and a database structure was developed (see Deliverable 1 [2] for further information). A simple database with two tables in Microsoft® Office Access was proposed and adopted by the involved partners. Chalmers coordinated the compilation of all the data from the partners.

Development of Detailed Database One goal of the RISER project was to review available databases for detailed accident information and to create a European database specifically for single vehicle accidents using existing sources. This database includes more specific collision information not available from the databases reporting national statistics.

The partners involved were initially asked to review the available databases in their respective countries and investigate the availability of using the data for the database. Table 2 shows the partners participating in this task, the source of accident data available to them and the number of cases they have entered onto the database.

Table 2. Contributors to RISER Detailed Database

Number of Participant Country Database source accidents Swedish National Road Administration Chalmers Sweden 68 (SNRA) Co-operative Crash Injury Study VSRC UK 30 (CCIS) CETE France CEESAR (EDA) 33 CIDAUT Spain CIDAUT (new cases) 11 TUG Austria TUG database 29 Traffic Safety Committee of Insurance HUT Finland 30 Companies (VALT) The TNO TNO (new cases) 10 Netherlands Total number of accidents 211

In order for the partners to be able to select the relevant cases for the database, a selection criteria document was devised. As this project was mainly involved with investigating single vehicle accidents on major arterial roads, the following criteria were used:

1. Single vehicle accidents where the significant damage and the most threatening impact in terms of injury was with the roadside infrastructure and not another vehicle. Single vehicle rollovers could be included if rollover initiated off the travel way. 2. No accidents involving pedestrians were to be included. 3. The vehicle must have left the travel way and made contact with or impacted objects (hazards) in the roadside environment (lateral or central reserve). Applicable objects include: ° safety barrier/guard rail/safety fence (including steel, concrete, wire rope safety fence and temporary barrier/fence), bridge parapet, crash cushions; ° sign post; ° a roadside embankment, ditch, slope, fill, etc; ° roadside pole (e.g. telegraph, lamppost); ° tree/hedge/wooden fence; ° bridge piers; ° arrester beds; ° rocks, boulders, etc.

4. Preferably accidents from the “last 3 years” (from 01/01/1999 onwards to match RISER statistical database). Good cases in 1998 were also to be considered.

5. Seat belt use was not required although it was preferable that the seat belt use was known for at least the driver. However, good cases where seatbelt use was not known were still to be considered. 6. All accident severities (Fatal, Serious, Slight and Non-injury accidents). 7. No minor rural roads or urban roads to be selected. 8. All production vehicles are applicable but focus was to be on passenger cars. 9. Must have detailed information on the roadside infrastructure (sketches and/or pictures) involved in the collision. Information of the vehicle damage and human injuries is necessary. The case data must support subsequent analysis/reconstruction of relevant infrastructure, vehicle and/or human factors (to be explained further in guidance notes to follow with the database forms).

The activity was lead by VSRC and supported by TUG for the software development. The RISER detailed database was built upon the existing ROLLOVER database, which includes relevant data fields for vehicle damage (e.g. deformation and intrusion) and occupant injuries. However, as this database did not include any information about the roadway, the struck infrastructure or accident causation details, new data forms were created and included in the database to cover these areas. As much as possible, compatibility with the PENDANT database was attempted.

The aim of the “Highway Details” data form was to gain detailed information on the roadside area(s) where the accident occurred. This included details for the cross section of the roadside where the accident occurred, such as the elevation, surface type and objects within the roadside (see Figure 1). One form was filled out for each area of roadway involved in the accident. For example, if a vehicle struck an object on the left and right hand-side of the road, a form should be filled out for each side of the road (see).

Carriageway Roadside / Median

A B C D

Width

G rad Elevation Carriageway ient edge line

Figure 1. Example of cross section of road showing definition of roadside/median sections

The “Struck Infrastructure” form was used to fill in as much information as possible about the specific type of infrastructure that caused the damage to the vehicle, such as a safety barrier, sign post or a steep embankment. One form was filled out for each struck infrastructure in that specific area of highway.

The "Causation and Human Factors" form was used to record as much information that was known about the cause of the accident, including key events that directly led to the accident and the risk factors which increased the likelihood of the accident occurring, and whether they were driver, vehicle or environment related. For example:

° Driver issues – distraction, mental state, physical state; ° Vehicle issues – design or the condition of vehicle; ° Environmental – road condition, poor visibility due to weather or the road geometry.

The database also includes a “Post Crash Data” form which allows data about the accident to be recorded for use in possible reconstructions.

In addition to the software, the database also allows the user to store further information about each accident, such as accident sketches and photographs, in files which can be linked through the database.

Collection of Road Operator Data HIASA investigated the assessment of the severity of a road restraint system based on accident data. The severity index used to assess a road section was defined as the number of fatalities per total reported accidents. The purpose of the investigation was to determine if this assessment is independent of the source of data utilized for the analysis (see Deliverable 1 [2]).

Two different sources of accident data were identified for use. One source was the official accident data source of the Spanish State – obtained from the General Direction of Traffic (DGT) and is a collection of reports from the Traffic Police. These data are published yearly. From DGT only injury accidents can be obtained no property damage accidents are reported.

The second accident data source was the records of the Road Operators (RO). The RO carries out the maintenance and repair of Road Restraint Systems as well as the care of the roadway in their regions of responsibility. The RO makes a report for each one of the accident repair works that is carried out. The reports are compiled into a database that is a representation of the accident history for the road section. Data obtained from the RO of includes data with both injury accidents and property damage accidents. The level of information will differ from police data, but can provide complementary information for establishing the difference between fatality and injury rates.

The following expression was used to calculate the probability of being killed when interacting with roadside infrastructure (Severity (S)).

S = num.fatalities / n Reported Severity S´= num.fatalities / N Real Severity where: n = Number of roadside collisions reported to DGT N = Number of roadside collisions reported by the Road Operator

The analysis of the Road Operator data from Spain shows that safety level assessments are biased by the source of the accident data. For the road network analyzed, the number of reported roadside collisions was 5 times higher in the RO data compared to the National Data (542 vs 99 cases respectively). Within the data collected for all accidents, the number of reported roadside collisions represents a higher proportion of the accident population. The RO data reported 81% of the accidents as being a roadside collision, whereas DGT data only reported 25% of the total accidents as being roadside collisions. These reporting differences leads to a different analysis result if one begins calculating the consequences of roadside collisions. When the severity is defined as the number of fatalities per single vehicle collision, the RO data has a much lower severity assessment (0.56) compared to the national statistics (DGT) data (3.03). These differences in severity may lead to the misjudgement of the effectiveness of a safety system or countermeasure.

Collection of test data The collection of test data was performed for comparison and analysis of the vehicle accelerations and occupant risk measurements of various products. Another intention was to represent several roadside infrastructure types and their mechanical behaviour.

Table 3 below presents the countries and partners that made different crash test data available and the actual source of the data. In total, data out of five individual countries was considered.

Table 3. Source of included crash test data RISER partner Source of used data Country Technical University of Graz Delta Bloc Europa GmbH Austria (TUG) Helsinki University of Helsinki University of Technology Finland Technology (HUT) (HUT) Volkmann und Rossbach Volkmann und Rossbach (V&R) Germany (V&R) Chalmers University of Swedish National Road and Sweden Technology (Chalmers) Transport Research Institute (VTI) The Netherlands Organization Ministry of Transport Netherlands The for Applied Scientific Research (Rijkswaterstaat) Netherlands (TNO)

Some of the mentioned partners made contacts with companies or test centres that could provide crash test data. Other partners contributed their own crash test results. In that way different crash test reports with similar content were retrieved. The most important values respectively test requirements and test data were chosen and tabulated:

° product, product name, material and profile ° performance level ° permanent and dynamic deflection ° ASI value and ASI class

° THIV ° PHD (Note that PHD has been deleted from further reporting of crash tests.) ° system length and height ° test length ° post spacing (if existent) ° anchorage ° test date and test house

Not all reports had all the information in this list.

The different road infrastructure types and the number of their collected test reports are presented in Table 4.

Table 4. Amount of crash test reports concerning road infrastructure types. Road infrastructure type Amount of crash test reports Concrete barriers 10 Concrete bridge barriers 1 Steel barriers 11 Steel bridge barriers 1 Steel barrier, computer simulated 1 Steel guardrails 17 Steel bridge guardrails 2 Steel restraint systems (gutter 1 construction) Crash cushions 12 Steel lighting columns 2 Wooden lighting columns 1 Traffic sign columns 1

Most of the reports mentioned above include several tests which mean that there is always more data behind one report than only one test.

Because the inequalities of the products and their test circumstances were really high, the collected crash test data was only partly comparable. Most and adequate test data was available about barriers and guardrails. Therefore these both groups have been focused in the evaluation. All together 68 crash tests were analysed (see RISER Deliverable 3 [3]).

A first look to the data (Figure 2) highlights that the data is fairly widely distributed. There are large differences even between the same or similar products. This is explainable by the different test conditions of the products like anchorage, post spacing, test length etc. Another reason for the wide data distribution is the special distribution of the crash test specification groups. The analysis shows that every crash test group has its own place within the data distributution (see Figure 2).

Figure 2. Approximate distribution of crash tests

The distribution of the crash tests depends on the different impact speeds and impact angles. The crash test group TB 21 for example, has the very low impact angle of 8 degree and impact speed of 80 km/h. Thus the severity results are quite low compared to other, more violent tests in Figure 2.

The results shown comparing ASI values with permanent deflection are similar to other results such as for THIV and PHD.

Another important and expected outcome of this analysis is the general trend that is the greater the permanent deflection, the lower the ASI or THIV value.

Roadside Infrastructure Influence on Driver Behaviour

Literature Review The role that roadside infrastructure has on driver behaviour was investigated in the project though literature review and driver simulator study. The collection of existing literature identifying the response of the driving public to changes in road design were conducted to serve as a starting point to include Human Factors principles within the roam of roadside infrastructure design by means of developing guidelines and analysis procedures necessary to select, implement, and operate a safe, efficient and affordable roadside infrastructure in the EU. The first step was to identify and review existing studies on what influence the roadside infrastructure has on driver behaviour.

Studies that were identified included driver aspects (e.g. sleep related research, driver vision) and road aspects (e.g. road lighting, changes in infrastructure), the influence of roadside changes on accident rates and on driver's behaviour. Some important studies that were identified include:

° study of the influence of paved shoulders on drivers' behaviour on departmental roads ° influence of paved shoulders on speed (case in Seine-Maritime, France) ° motorists' visual perception in rural areas ° evaluation of removal of trees on speeds ° paved shoulders and kinetics of light vehicles: effects of the pavement- shoulder drop-off ° evaluation of rumble strips, analysis of encroachments on motorway emergency lanes through a before-after observation ° signing and marking of substandard horizontal curves on rural roads ° summary of documents related to roadside infrastructure, such as rollover of vehicles on concrete barriers A total of 144 papers were found in and from, the available sources. Each paper was rated in terms of its relevance to the subject of the review. Ratings applied were ‘high’, ‘moderate’, and ‘low’ (where ‘low’ is to be distinguished from ‘none’). As a result 83 of the total of 144 papers was rated ‘none’, meaning when reviewed in detail the paper did not deal specifically with the driver-roadside-element interaction in some way (see Deliverable 2 [4]).

The main conclusions drawn from this activity were:

1. There are not that many studies that have actually looked into driver behaviour, or that – at least – present some ideas about the role of driving behaviour in the chain of events leading to a Run-Off-Road-accident (ROR- accidents)

2. The studies that are available, which are mainly field observational studies, do not really use the best ‘proxies’ for the risk with which drivers execute the driving task.

3. There is, on the other hand, a relative preponderance of accident-analytic studies, leading to predictive models or tools for benefit-cost analysis that can be applied to ROR-accidents. Useful as they are, for the reasons explained in the beginning this cannot be considered a sufficient basis for the development of adequate countermeasures.

4. Almost all countermeasures that are presently under discussion in the literature are directly infrastructure-based. It is surprising that only once or twice is there mention in the literature of the potential of advanced in-vehicle devices, such as Lane Departure Warning systems, as a remedy for ROR- accidents.

In the light of these results it was recommended that more systematic research on the response of the driving public to changes in a road design is to be conducted. The EU-RISER project will contribute explicitly to this objective.

Collection of Current Design Guidelines The purpose of this activity was to identify the current state-of-the-art in Europe with regards to roadside design guidelines (see Deliverable 5 [5]). The analysis was

limited to official national documents and does not include all the policies that are practiced at national and regional levels.

The roadside safety guidelines of seven European countries were reviewed. There was general agreement between the countries in terms of general practices. For example, all countries promote the safety zone and use the European Normatives to describe the performance of safety equipment. The general methodology to improve roadside safety is also largely the same among the countries. However the technical specification of each roadside safety element varies between the countries. For example, safety zones are much wider in the Netherlands than in the Great Britain and few countries treat the recovery zone as a separate entity.

The core information for designing a roadside design guideline were distilled from the information collected in Deliverable 5 [5]. The synthesis of the seven national policies into a common language document was useful for the further development of road infrastructure safety practices.

The Safety Zone The seven countries involved in this activity have adopted the “safety zone” concept as a zone adjacent to the road which is free of any obstructions which would be in the path of errant vehicles (see .

This philosophy of a “forgiving road” is the mere recognition that road users sometimes leave the running carriageway for explainable or unexplainable reasons.

Figure 3. Definition of Safety Zone

The Recovery Area The concept of the recovery area is not clearly identified as such among all contributing partners. This zone is used mostly for recovery manoeuvres, but it can provide extra space to avoid vehicles deriving from their normal paths, to avoid left- turning (or right- turning in the GB) vehicles, for emergency stopping of vehicles, for emergency rescue vehicles and also for the circulation of bikes and pedestrians off the driving lanes.

The main criteria for the recovery area dimensions are design speed, road type, traffic flow, and driving lane width.

As the recovery area is part of the safety zone in most of the guidelines, dimensions are not always provided. For the countries where the recovery area is clearly identified as a separate issue, the recovery area widens according to road type. On single carriageway roads for instance, dimensions vary one country from another (0.5m to 2m).

Roadside and Median Hazards Roadside hazards are continuous or punctual, natural or artificial, fixed objects or structures likely to deteriorate the consequences for an errant vehicle leaving its normal path (such as high decelerations to the vehicle occupants or rollovers). They can be located either on the verge or in the median reserve of the carriageway.

In the following analysis, hazards are divided in four categories:

1. Distributed hazards 2. Point hazards 3. Road restraint systems 4. Additional factors of risks

There are various factors of risks where the vehicle crashes against old roadside or median barriers, due to barrier components, improper dimensions of restraint systems, bad positioning or untreated terminations of safety barriers or fences.

Technical Treatment of Roadside Hazards for new Constructions For new constructions in all countries, there should not remain any obstruction in the safety zone otherwise shielded by a restraint system.

Technical Treatment of Roadside Hazards on Existing Roads The treatment of roadside hazards is the basic problem for the RISER research project.

Traffic flow, type of vehicles, distance of obstacle from the edge of carriageway and position in curve or in a straight road are the most frequent parameters used to treat the roadside obstacles on existing roads.

There are various methods of redesigning a roadside environment. The process listed below is more or less used by France, Finland, the Netherlands, Germany and Spain:

1. Remove the obstacle (all countries) 2. Redesign the obstacle so that it can be safely traversed (FI, FR, DE, GB, NL, SE) 3. Relocate away the obstacle (all countries) 4. Reduce impact severity by using an appropriate break-away device (FI, GB, NL, SE) (*) 5. Protect from the obstacle with a restraint system (all countries)

6. Delineate the obstacle (FI, FR, DE, GB, NL, ES) (**) (*) Reduce impact severity is an optional item in Germany and France, where shielding the obstruction is the preferred measure. It is not in use in Spain. (**) Delineation of the hazard is a transitory guideline in France which is being paced out. Delineation is not discussed in Sweden.

Restraint Systems Road restraint systems are used to protect the vehicle from the roadside environment. The underlying requirement is that the restraint system will result in a collision severity that is lower than a collision with the roadside environment.

The European requirements for Road restraint systems are covered in the test standard EN1317. The road restraints can be broadly divided into the following groups:

° Longitudinal barriers (contain and redirect vehicles) eg: guardrail ° Crash Cushions (energy absorbing structures for point obstacles) ° Transitions (connection between 2 different types of road equipment)

The test standard describes performance characteristics for different products following into these categories. However the standard does not describe the application of these products.

° Several restraint systems are treated in the national documents: ° Safety fences and safety barriers (all countries) ° Crash cushions and impact attenuators (all countries) ° Noses of island (the UK) ° Arrester beds (the UK) ° Wire rope safety fences (the UK, Sweden) ° Railings (the Netherlands) ° Low concrete walls (France)

Collection of Current Maintenance guidelines An inventory of existing guidelines in Europe related to the maintenance of roadside equipment and terrain was performed. Deliverable 7 [6] is a summary of seven questions about current practice which were answered by various means of documentation. Some questions lead to references in official documentation while other questions are only answered informally through interviews with maintenance officials. When documentation was not available, the information tended to reflect regional policies and may not be the common practice within the entire country. This makes it difficult to generalise the national maintenance practice for some countries.

Sufficient information about standards and guidelines is available from several countries and were used in the preparation of new common guidelines. The documentation collected indicates that in the various EU-countries we have, in general, a similar structure by dividing the road network into categories and prioritising the main roads regarding the maintenance efforts and intensity. The data

available from the referenced sources will be sufficient for preparation of common guidelines according to our assessment.

The current maintenance and operations practice for each country were assessed by identifying the answers to the following set of questions:

1. How often should inspections occur? 2. How often should repair works occur (plus response times)? 3. What are the criteria of component replacement? 4. Who reports the need for maintenance? 5. What are the maintenance procedures after a road accident? 6. Where is the information recorded and stored? 7. Are there training systems for those responsible for maintenance?

One goal was to identify a list of variables used in the maintenance activities which could be exploited for the later activity of developing maintenance guidelines related to safety.

The standards and guideline documents related to operation and maintenance of the roadside area cover an array of activities and physical features. With respect to roadside safety issues, infrastructure elements that were investigated should be relevant for single vehicle collisions.

Analysis of Accident Data/Study of Human Performance

Summary The information collected in the previous chapter provided an overview of the state- of-the-art in Europe in relation to simgle vehicle accidents. From this information, more detailed analyses were identified to take the understanding of the events causing the single vehicle accidents, as well as understanding the performance characteristics of different infrastructure elements. From the information collected and analysed in the previous chapter, the detailed analysis described in this chapter enabled the RISER consortium to take existing roadside safety practice and develop best practice guidelines for design and maintenance.

Analysis of statistical data An analysis of the statistical database was carried out to identify important characteristics of single vehicle accidents (SVA). There are great differences in the SVA severity, e.g. approximately 13 % of the French cases are fatal SVA, in comparison Great Britain has reported 3 % fatal SVA (see Figure 4). For the RISER database about 44 % of the fatal accidents are French cases and 36 % are Spanish SVA. Therefore the analysis for the complete dataset would be dominated by results from France and Spain, this fact points out, that the analysis must be done country specific, to avoid bias.

RISER SVA PER INJURY SEVERITY

100%

90%

80% 50.9% 58.1% 70% 60.0% 62.1% 66.3% 71.4% 75.3% 60%

50%

40%

30% 29.0% 42.1% 32.1% 29.6% 20% 28.0% 24.5% 21.4% 10% 12.9% 7.9% 8.3% 4.1% 5.7% 5.7% 6.9% 0% 3.3% 1 SWE 3 FIN 4 AUT 5 FRA 6 ESP 7 GB 8 NL Total DATABASE

1 Fatal 2 Serious 3 Slight

Figure 4. RISER single vehicle accidents by injury severity

Figure 5 shows the frequency of impacts versus the consequences of impacts based on the information in the RISER statistical database. The dashed lines are constant risk curves, the points in the upper right corner are high frequency, high fatality hazards. Trees are thus an important priority for improving roadside safety, but the

"No object" point indicates vehicles rolling over in the terrain. As expected, safety barriers have a high frequency of involvement in SVAs, but a lower fatality frequency.

40%

35%

30% Trees n o i 25% s i l l o C

f 20% o

y t i

r No object Ditch e

v 15% e

S Other Object Other Man Made 10% Object Safety Barrier

Post 5%

Other Natural Object 0% 0% 5% 10% 15% 20% 25% 30% Probability of Collision Figure 5. Risk information for various struck objects

For the accident type Austria, Finland and Great Brittan have comparable datasets. In Finland accidents on straight roads dominate whereas in Austria and Great Britain the SVAs seem to happen in the outer curve. Due to left hand traffic in GB the nearside (closest to roadside) in a right curve for GB can be compared to offside (closest to centreline) in a right curve in Finland (see Figure 6). This confirms other studies that have showed that single vehicle accidents occur more frequent in outer curves compared to inner curves.

Figure 6. Accident type distribution

As mentioned earlier the material is difficult to compare between the countries since the differences in the collection of the data in the national databases varies a lot. The main conclusions from the data collection activity were:

° there is a need to harmonize the data collection in the European countries ° external factors that affect road safety should not be neglected (regional parameters have influence on safety) ° the fatality rate of single vehicle accidents is high compared to other accident types ° there are great differences in the frequency, fatality and injury risk regarding road equipment ° that although motorcycle accidents are not a significant proportion of the accident statistics there is a high fatality rate within these kind of accidents ° crash severity peaks with bad weather conditions ° that longitudinal barriers can have a guiding function for drivers

The conclusions from the statistical data were an important guide for subsequent analysis in the RISER project.

Analysis of detailed data An in-depth analysis of the data in the detailed database was undertaken to identify weaknesses on the existing road network but also to identify when those weaknesses becomes hazardous. A number of areas were investigated, including the type of infrastructure impacted, set-backs of hazards and objects for the road edge, recovery area width, accidents involving safety barriers and an overview of accidents where there is evidence of an initial run-off with no impact.

Any statistical analysis of the detailed data was not relevant because the sample of 211 cases in the database is not a truly representative sample (e.g. there are a greater number of fatal cases in the database than would be expected in a 'representative' sample).

A "run-off" is the process of the vehicle entering the roadside from the main carriageway. Each accident in the database is either a single or a multiple run-off. A single run-off is when a vehicle enters the roadside and either does not return to the road or returns to the road and comes to a stop. A multiple run-off is when a vehicle enters the roadside (with or without impact), then returns to the main carriageway before entering the roadside at least one more time.

Run-offs are different to "events", which refers to the number of individual collisions in the roadside. By using the scene sketches and the "event" data, the type of run-off (i.e. single or multiple) could be determined for each accident.

In the database, there is a record for the number of lanes at an accident scene. However, these lanes often included slip road and turning lanes, so it was useful to be able to determine whether the road itself was generally a single or dual carriageway, not including the turning and slip lanes. The lane data, plus the scene

photographs, were used to determine whether each case in the database occurred on a single or dual carriageway road.

Struck Infrastructure The original 28 types of infrastructure elements have been grouped together into 6 main categories presented in Table 5. Table 6 displays a summary of the type of infrastructure which was impacted in the 211 database cases, including some of the general characteristics of the different type of accidents.

Table 5. Categorisation of infrastructure type Category Infrastructure type in database Safety barriers Bridge parapet; Safety barrier - steel, concrete, termination, wire rope safety barrier, other. Posts/poles Sign post; Lighting pole; Telegraph pole. Trees Tree. Other fences Non-safety fence; Hedge. Concrete & rock objects Wall; Bridge pier; Rock/boulder; Wall of pedestrian underpass. Sloping ground Embankment/slope; Ditch; Drainage gully. Other Other, Fog pole, Foot path, Unknown, N/A

Table 6. Summary of cases in the detailed database Tree Post/pole Rock/con- Sloping Non-safety Safety crete object ground fence barriers No. of Accidents 38 45 34 71 11 72 No of impacts/ 50 51 37 88 11 86 interactions Protected by safety 2 5 7 5 0 - barrier Fatal or serious 34 34 30 45 7 47 cases Slight or non-injury 4 11 4 26 4 25 cases Motorways 2 7 10 13 2 34 Non Motorways 36 38 24 58 9 38 Single 33 29 24 51 8 20 carriageways Dual carriageways 5 16 10 20 3 49 Single run-offs 34 39 31 54 8 52 Multiple run-offs 4 6 3 17 3 20

Only impact 30* 20 15 44 1 41 Fatal or serious 26 15 12 22 0 21 cases Slight or non-injury 4 5 3 22 1 20 cases Motorways 2 4 3 9 0 21 Non-motorways 28 16 12 35 1 18 Single 25 12 10 30 1 6 carriageways Dual carriageways 5 8 4 13 0 33 Single run-offs 26 18 13 38 1 29 Multiple run-offs 4 2 2 6 0 12 Greatest set-back 6.8m** 4.5m*** 6.4m**** No fatal No fatal 1.5m**** in fatal accident (6.8m (7.2m slight) slight) Smallest diameter 0.3m 0.2m - - - - in fatal accident** After impact 0m - 0m - 39m 0.6m - 32m 0m - 49m 1m 0m - distances 44m 208m * Only or 'main' impact in crash (21 cases where tree impact was only impact). ** Seatbelt worn ***Seatbelt not worn but would have made little difference as it was a side impact. ****Seatbelt not worn.

The results presented below focus on those accidents where only one infrastructure type was impacted, the “only impact” row in Table 6. The final conclusions from the detailed database analysis and the current guidelines recommendation can be found in Table 8 and Table 9.

Trees There where 19 fatal, 7 serious and 4 slight accident concerning trees. The set-back of trees from the carriageway edge line ranged from 2.0 m to 10.8 m. The diameter of the trees ranged from 0.15 m to 1.0 m.

From looking at the 30 cases where the tree was the main or only impact, there are no clear indications that trees become more dangerous when they are a particular diameter or a particular distance away from the road. The 'narrowest' tree which was

the main impact in a fatal accident was 0.2 m. In this accident, the driver who died was not wearing a seatbelt. The 'narrowest' tree impacted in a fatal accident where all occupants were belted was 0.3 m. The greatest set-back of a tree in a fatal accident where all occupants were belted was 6.8 m. The collision speed in the fatal accidents varied from 140 km/h to 75 km/h, the lowest impact speed for a serious accident was 45 km/h.

Posts and Poles There were 10 fatal, 5 serious and 5 slight accidents concerning posts and poles. 11 of the 20 collisions were side impacts and 9 were frontal impacts. 7 of the 11 side impacts led to a fatality, as did 3 of the 9 frontal impacts. The set-back of the posts/poles from the carriageway edge line ranged from 1.12 m to 4.5 m. The diameter of the posts/poles ranged from 0.08 m to 0.4 m. The impact speed in the fatal accidents varied from 100 km/h to 70 km/h, the lowest impact speed for a serious accident was 40 km/h.

Rock and Concrete Objects There where 11 fatal, 1 serious and 3 slight accidents with rock and concrete objects. There where only 6 set-back measurements, which ranged from 0.5 m to 6.4 m. In the fatal accident where the set-back is 6.4m, the vehicle had a side-impact (offs- side) with the rock wall. The driver was not wearing a seatbelt, but this may not have made any difference to the injury severity. There are very few measurements of the size of the object.

Sloping Ground There where 21 fatal, 1 serious, 19 slight and 3 non-injury accidents. In 25 cases, the first/only impact was with a cut slope or embankment, in 19 cases, it was with a ditch or a gully. Set-back of the start of the slope/ditch from the edge of carriageway ranged from 1.5 m to 6.8 m. Unfortunately, there are no measurements recorded for any of the fatal accidents. Impacted slope gradient ranged from -100% to +143%. In 29 of the 44 cases involving sloping ground only, a rollover occurred, of which 13 were fatal accidents. The impact speed in the fatal accidents varied from 90 km/h to 45 km/h.

Safety Barriers There where 8 fatal, 13 serious, 19 slight and 1 non-injury accident. In the dataset, barriers not fulfilling the standard EN1317 were included. The set-back of barriers from edge of carriageway ranged from 0.12 m to 7 m. As would be expected, barriers tended to be closer to the carriageway edge than other types of struck infrastructure. The height of the barriers did not vary too much, from 0.6 m to 0.83 m. In the database, accidents involving barriers occur more often on dual carriageway roads (motorway and non-motorways). There were more single than multiple run-offs, although the proportion of multiple run-offs was greater than with other types of struck infrastructure. Although there were fatal cases involving barriers, some were due to other factors in the accidents (e.g. non-car impacts, other subsequent impacts not recorded in the database). However, some were due to the failure of the barrier to protect the vehicle and its occupants.

Special attention in the analysis was made for:

° Accidents where vehicle was 'rebounded' by the barrier. ° Accidents where the barrier 'failed' to protect the vehicle & its occupants o Accidents where the vehicle was 'launched' into the air after an impact with a barrier termination. o Accidents where the vehicle travelled along the barrier. o Accidents where the vehicle broke through the barrier o Accidents where the vehicle went over the barrier.

The conclusion concerning safety barriers was that they should be considered as a roadside hazard when:

° Safety barriers which are not EN1317 compliant. ° Safety barriers which have been damaged or poorly installed. ° Safety barrier terminations which do not have a flared end.

In addition, although not directly hazardous, the following have potential to cause a hazardous situation:

° An inadequate length of safety barrier prior to a pre-defined hazard. ° Short gaps in between two safety barrier installations, in particular on central reservations.

Simulation study

Accident reconstruction: PC-Crash The detailed database analysis provides a thorough analysis of the 211 accidents collected for the RISER project. This analysis describes the critical patterns in single vehicle collisions that must be addressed in future guidelines for roadside safety.

Complementing the detailed database, reconstructions and simulations were conducted to identify some of the important impact parameters, such as speed and angle, that were not part of the original data collected. The whole analysis can be found in Deliverable 3 [3].

The PC CRASH analysis tool has been used for the reconstruction of accidents, allowing describing the impact conditions and also the vehicle kinematics during the event of the impact itself. It takes into account vehicle manoeuvres prior to the crash, and it is convenient for accidents which involved elements of the road environment with complex geometry such as embankments.

The accidents reconstructed for the detailed database were analysed to obtain impact conditions. Details from these reconstructions are stored in the RISER detailed database as well as internal reports maintained by the respective partner.

An example of the type of information collected by a detailed reconstruction of the accident is shown in Figure 7 below. It illustrates the exit speed of the vehicle compared to the posted speed for the road section.

180 Exit speed 160 Max road speed

140

120 ) h /

m 100 k (

e 80 e p S 60

Figure 7. Vehicles exit speed compared to the posted speed for the reconstructed cases.

Another important piece of information for the accident is the exit angle of the vehicle. The reconstructed cases included the initial departure, re-entry, and subsequent exit speeds and angles for the cases when applicable.

Figure 8 shows that 85% of all the reconstructed accidents had initial departure speeds from the roadway under 110 km/h. Figure 9 shows similar information for the exit angle for the vehicles. The 85%ile of vehicle initial departure angles are under 20 degrees. Both these distributions provide some justification for the standard test conditions for a heavy passenger vehicle.

100%

90%

80%

70% e g a

t 60% n e c r e P 50% e v i t a l u

m 40% u C

30%

20%

10%

0% 0 10 20 30 40 50 60 70 80 90 100 110 120 130 140 Speed [km/h] Figure 8. Cumulative Percentage Distribution of Exit Speeds

100%

90%

80%

70% t

n 60% e c r e P

v 50% i t a l u m u 40% C

30%

20%

10%

0% 0 5 10 15 20 25 30 35 Exit Angle [deg] Figure 9. Cumulative Percentage Distribution of Initial Exit Angle

Accident reconstruction: Finite Elements Simulation tools based on finite element method allow modelling with detail the behaviour of vehicles and structures under impact conditions. The purpose of this activity was to analyse the response of different roadside features such as metal barriers and poles when they deform under the impact loads that occur in crashes. Infrastructure types that were not well represented in the detailed accident database were also selected for more detailed reconstruction investigations.

The resulting vehicle and occupant behaviours during pole and post impacts were analysed in detail. For this purpose, a finite element vehicle model has been used that reproduces the crush behaviour of the frontal part of a small-size light, passenger car so that the most critical impact case is addressed. A control simulation of the vehicle model crashed into a rigid pole was performed and it was checked that its behaviour was consistent with the one observed in different regulations and test analyses

Results show that the potential for safety improvement is highly dependent on the design of the support structure. Poles of small or medium size, which are less exposed to weight and wind loads, can implement breakable connections to the ground and fulfil EN 40-3 at the same time. This improves the severity indices measured in the vehicle and safety for occupant.

Pole designs with larger sizes have demonstrated more difficulty for safety improvement, because of their geometries and masses. In these cases, stronger structural design requirements introduce that difficulty

As a conclusion, it is observed that a most critical interaction mode between poles and vehicles is the direct impact of vehicles against non-deformable or detachable posts. Breakaway mechanisms in the base are a convenient way to improve their safety in those locations where the detachment of a pole does not cause additional risks of eventual secondary impacts.

Accident reconstruction: MADYMO The MAYDMO simulation tool is able to analyse the phenomena that take place inside the vehicle during its interaction with the infrastructure. The above mentioned

tools (mainly finite element simulations) can provide a description of the overall vehicle kinematics, and then MADYMO assesses the interaction of the occupants with its envelope by using surrogates (crash test dummy models).

The mechanical interaction between the occupant and the vehicle parts involved, being a result of the behaviour of such different components as vehicle body and door structures, inner trimmings, and even restraint systems, were modelled by unified contact laws that reproduce a sample vehicle. The outputs of these simulations were an additional source of data that were analysed together with the standard injury parameters that were obtained from the vehicle motion.

Three representative cases were analysed. The first one involves a lighting pole, non-breakable and built in sheet metal. The second one is a rigid -cast iron- lamp post. The third one is a modified version of the second, in which a breakable fixing to the ground is implemented, using specific calibrated bolts. Figure 10 shows the comparison of the behaviour of the three cases.

Figure 10. Comparison of the occupant response in impacts against poles. Case 1 (left): sheet metal pole. Case 2 (centre): cast iron pole. Case 3 (right): cast iron pole with breakaway system.

Case 1 (sheet metal pole) was analysed via finite element simulations, with a severity index obtained of ASI=1.6. This would mean a considerable risk of injury for occupants, as is it observed in Figure 10, left column. Despite the action of occupant restraint systems (seat belt), the occupant motion implies high probability of direct impact of the head on the steering wheel. Case 2 (cast metal pole) is a dangerous impact. Previous analyses yielded an ASI value over 2, and the occupant simulations also predict high risk of occupant due to head impacting the steering wheel. Finally, case 3 demonstrates how a breakable –or eventually, deformable- pole would improve the safety performance. The predicted ASI value was 0.9, and Figure 10, right column, shows how the vehicle was slightly decelerated by the crash, but then continued its motion causing not so serious loads for passengers.

Modelling of vehicle infrastructure collisions: The modified vehicle model of Geo Metro was used for frontal, lateral and oblique impact. The vehicle reasonably replicates the major sequences of the collision event.

Frontal pole impact: The Geo Metro modified model was validated against the rigid pole for frontal impact scenarios. There has been found a good agreement between the test and the simulations with respect to the velocities histories and event timing.

Lateral pole impact: It has been noticed from the simulation results that the lateral pole impact is the most severe collision. The higher values of the injury parameters are caused by the extensive intrusion in the occupant compartment.

The modified model of the Geo Metro reasonably replicates the major scenes of the collisions.

Car to guardrail impact: The W-beam guardrail model replicates the basic phenomenological behaviour of the real interaction. It has been shown from the simulation results that the Geo Metro reasonably replicates the real sequences of the oblique impact.

Simulation matrix The results of the matrix show that the lateral pole impact is the most severe crash. In addition, the injury parameters increase with impact speed.

Stochastic simulation: The stochastic simulation showed that there is a strong correlation between injury criteria and impact speed. It has been noticed also that there is a strong correlation between the vehicles based criteria and the impact speed. However, the lateral position of the pole does not change the injury risks.

Driver Simulator Study The influence of driver behaviour due to different road and roadside elements was described in Deliverable 4 [7]. Three basic data sources are reported – the real world experience reported in the RISER detailed accident database, the findings of a detailed human factors based Road Scene analysis of ten selected accident sites, and a driving simulator experiment conducted at TNO Human Factors. The goal is to give specific insight into the question what the influence is of continuous roadside elements such as guardrails, barriers, and other roadside elements.

Twenty participants took part in the TNO driving simulator study. All participants (9 male and 11 female) had at least 5 years of driving experience. Their age varied between 29 en 53 years old, with an average of 46 years old. All drove more than 5000 kilometres per year and had done so for at least the past three years, with an average amount of 22 700 km/year.

All participants completed a total of 24 drives, in randomised order. The first drive was a practice drive of 5 minutes long. For all drives participants were asked to drive in the right lane as much as possible. They were not allowed to pass vehicles by overtaking in order to be able to measure a continuous influence of the guard-rail on the right side on lane position or speed.

The experiment was conducted in the high fidelity driving simulator of TNO Human Factors (Error! Reference source not found.). In this driving simulator the participant was seated in a BMW 318I mock-up that was placed on a motion base with six degrees of freedom. The participant watched a large radial screen on which the road and traffic environment was projected. The sounds of traffic in the environment and the car the participant was driving were also presented.

Figure 11. The TNO driving simulator driving on the motorway section without guard-rail on the right side, with emergency lane

Straight motorway sections On the (two-lane) motorway participants drove a total of 12 (mostly straight) sections, differing in road lay-out (with or without emergency lane), roadside elements (with or without a guard-rail or barrier), material of barrier (guard-rail or concrete step-barrier), and height of barrier (0.75m, 1.2 m and 1.6m). For details see Error! Reference source not found. and Error! Reference source not found.. The width of the lanes was 3.50m and the width of the vehicle is 1.70m.

Figure 12. Illustrations of guard-rails and barriers on straight sections on motorway with and without emergency lane. For all configurations of guard-rail and barrier a situation with and without emergency lane was presented to the participants.

Treatment of curves These conditions also consisted of driving on the motorway (two-lane). All rides had the same road layout with a narrow curve to the left with a radius of 700 meter and a length of 500 meter, starting at 2000 meter into the drive. The guard-rail in the curve on the right was treated in three different ways: 1) standard, 2) black and white stripes painted on the standard guardrail, and 3) black and white stripes painted on the guard-rail plus red and white stripes with arrows on top of the guard-rail (see Error! Reference source not found.).

Figure 13. Illustrations of the three different treatments in the curves

Rural road These conditions dealt with rural-road situations and obstacles alongside these roads. The rural roads chosen were two-lane two-directional roads (with very slight curves) with a maximum speed limit of 80 km/h. In all conditions, the only road marking present was an interrupted centre line marking. The choice to leave out the edge line markings was made, because, in this study we would like to make a distinction in behaviour (if present) as much as possible between different roadside conditions (see Error! Reference source not found.).

Figure 14. Illustrations of the five conditions on the rural road

Table 7. Overview of driving simulator conditions Condition 0 Practice drive Motorway Emergency lane Barrier/guard-rail Height 1 without No - 2 without concrete barrier 90 cm 3 without concrete barrier 120cm 4 without standard guard-rail 75cm 5 without guard-rail 120cm 6 without guard-rail 160cm 7 with no - 8 with concrete barrier 90 cm 9 with concrete barrier 120cm 10 with standard guard-rail 75cm 11 with guard-rail 120cm 12 with guard-rail 160cm Curves on motorway Emergency lane Treatment Total height 13 (driven twice) with standard 75 cm 14 (driven twice) with black/white stripes 75 cm 15 (driven twice) with stripes and strips 170cm Rural Road Trees Guard-rail Total height 16 30 meter At 4 ½ meter 75 cm 17 behind guard-rail At 4 ½ meter 75 cm 18 4,5 meter No - 19 2 meter No - 20 behind guard-rail Standard at 0.9 m 75 cm

Simulator Study Results Given the limited data in the database of accidents collected in the RISER project, a driving simulator study was developed to investigate two of the main factors leading to crashes – lateral position and speed. For the conditions on the motorway, where different types of guard-rails were introduced, it can be concluded that there was no effect on speed or on lateral position dependent on the type of guard-rail that was introduced. Overall no guidance of any type was found.

The presence or absence of an emergency lane did have an influence on driving behaviour. When an emergency lane was absent, drivers tended to choose a position on the road that is further away from the right side marking. How far away the driver chose to drive from the right side marking when an emergency lane was absent depended on the type of guard-rail. When the standard 75 cm high Dutch concrete barrier was introduced (this is the lowest guard-rail of all guard-rails introduced) and when the 90 cm high concrete barrier was introduced (the second lowest of all guardrails) in combination with an emergency lane drivers drive closer to the edge markings than in other conditions. However, when the emergency lane is absent they move further away from the right side edge markings in these conditions than in other conditions compared to the situation with an emergency lane. It could be the case that familiarity in the condition with an emergency lane plays a role.

Another conclusion based on these results and the results on the rural road is that when an obstacle is introduced drivers tend to temporarily move away from that obstacle and choose a position in the lane further away from the edge line. This effect was also found in earlier field studies when introducing a sound barrier or entering a tunnel entrance (Blaauw, & van der Horst, 1982; Bakker, & van der Horst, 1985; de Vos, Hoekstra, Pieterse, 1998; Martens & Kaptein, 1997, 1998). The effect was not found in the curve sections on the motorway.

The results found in the conditions with the curves on the motorway revealed that when drivers see a particular treatment for the second time they tend to be less impressed by it and, in this case, drive faster through the curve and closer towards the right edge marking than the first time. Also, when a new treatment (like the stripes) is introduced, speed goes down. This might, however based on the previous finding, be temporarily.

On the rural road drivers did change their speed when trees or guard-rails were introduced dependent on how close they were positioned to the road. When they were 4.5 m away or more, no effects on speed were found, whereas when they were positioned 2 m away or less, speeds were reduced.

For the lateral position a different picture occurred, when trees were introduced in combination with a guard-rail, drivers tend to choose a position away from the guard- rail and trees. However, when trees were introduced solely no effects on the lateral position chosen were found. In fact the position chosen in these two conditions was the same as in the condition with the trees at 30 m away from the roadside. One has to question therefore whether drivers understand the risks of trees alongside the road and whether that might be an underlying reason of the many incidents and accidents with trees. Trees along the road do not seem to influence driving behaviour that much and are not considered to be a serious hazard by road users. This is in line with the real life studies reported in France, (CETE 2000 & 2002) where the removal

of trees closer than 2 m from the lane edge had no influence on travel speeds. One important result from the French study was that after the trees have been cut down, the average number of accidents was halved, and the number of fatal accidents reduced by a factor of four. It is important to realize that the current guidelines recommend that no obstacles should be in the safety zone which is at least 4.5 m from the lane edge for 80 km/h roads in most European countries surveyed, and, if so, should therefore be removed or protected.

Development of Best Practice Guidelines for Roadside Infrastructure

The main objectives for the RISER project was the development of best practice guidelines for roadside infrastructure. Two documents were generated, one for the design and redesign of roadside areas and one for the maintenance and operations apspects related to different roadside infrastructure elements. The guidelines described in this structure are based on the existing guidelines obtained in the initial review of European policies. This information was then supplemented with the information gathered in the accident data, simulation studies, and human factors analyses conducted in the RISER project.

Design Guidelines The goal of any road authority, road operator and road designer is to provide the best service for the travelling public. The capacity of the system must allow the public to reach their destination in a timely manner without creating a safety risk for the vehicle occupants.

The physical layout of the road and roadside environment provides visual clues and signals to the vehicle operators. The type of road – a motorway or a small forest road – should be clear to the driver without explicit signs. The road width, types of lane markings, roadside geometry, etc. should provide clues to the driver that signal appropriate driving speeds and lane positioning for the type of road and indicate what type of other road users to expect.

The Self Explaining Road concept advocates a road and traffic environment that elicits safe driving behaviour simply by its design. By maintaining consistent and uniform road and roadside design procedures, the road meets drivers’ expectancy and drivers can anticipate changes in operating conditions (expecting at-grade intersections, pedestrians, etc.) even if they have not observed a sign stating that, for example, the motorway has ended. The physical layout of the road environment explains the driving context.

The Forgiving Roadside philosophy is simply the requirement that the roadside environment should not contain dangerous elements that will seriously injure or kill vehicle occupants that have unplanned trajectories off the carriageway. A fundamental component of this philosophy is the definition of an obstacle-free safety zone beside the carriageway. Since this is economically and functionally not always achievable, the introduction of passive safety equipment like road restraint systems (safety barriers), crash cushions, and energy absorbing (or break-away) posts to protect vehicles and minimize the consequences from dangerous impact hazards. It is important to recognize that all objects placed near a travel lane are potential impact hazards. The proper engineering design of passive safety infrastructure ensures that any subsequent impact with a safety device is much less severe than the resulting impact if the safety device was not in place.

The RISER project has produced a document [8] structured to follow the analysis procedures for best practice guidelines which may also be useful when conducting road safety audits for roadside safety (see Figure 15). The first step in evaluating

roadside infrastructure safety needs is to identify the types of fixed obstacles (or objects) adjacent to the road. These objects must be identified and inventoried in terms of type of obstacle and the lateral position to the road. From this information, the appropriate clear and recovery zone criteria should be defined for the road section to identify the critical areas in the roadside environment. From this information the first engineering judgements begin – identifying which obstacles are located in sensitive (affecting safety) areas and determining if these obstacles are a safety hazard.

Obstacle Chapter 1 Identification

Safety Zone Recovery Zone Chapter 2&3 Criteria Criteria

Hazard No Evaluation

Yes

Chapter 4 Hazard Yes Removal

Hazard Yes Final Monitoring Modification Solution

No Chapter 5 Chapter 6

Hazard Yes Protection

Figure 15. Procedure for Roadside Infrastructure Design. Chapter refer to [8]

Obstacle Identification Using the findings from the statistical and detailed database analyses and the review of current guidelines in Europe, definitions of roadside and median hazards have been produced. This includes minimum measures, impact speeds and set-backs that cause serious or fatal injuries for the accidents studied in RISER.

Table 8 and Table 9 outline each roadside object and the characteristics which define it as a roadside hazard (for example, size, location, frequency) according to the detail database analysis and the review of the current guidelines.

NOTE: These MINIMUM measures are conclusions from the RISER analysis. If guidelines within individual countries already include a greater margin of safety than those stated here (for example, smaller diameters, smaller heights, less severe gradients, slower speeds), then those national guidelines should also still apply.

Where possible, dangerous impact speeds have been identified from reconstructed cases in RISER detailed database. The dangerous impact speed is the minimum

speed at which a hazard can be impacted and still cause serious injuries to the occupants.

Given the limited cases (211) available in the detailed database, this data may not fully identify the range of impact conditions that cause serious or fatal injuries. Therefore the existing guidelines should be consulted to identify specific values for a country or region.

Table 8. Point Hazard Characteristics for Serious or Fatal Injuries in the RISER Detailed Database Hazard Diameter Dangerous Additional comments [m] impact speed [km/h] Trees and tree stumps >0.2 40 Typically >0.1 in many national guidelines The following poles/posts1 - Utility poles >0.2 40 - Standard lighting poles (wood, metal and concrete) - Posts of roadside signs >0.1 40 - Gantry/large traffic signs >0.1 40 - Supports/CCTV masts/High mast lighting columns - Supports/other high mast posts/poles. Rocks and boulders - - Bridge piers/pillars/abutments 50 Culvert ends/ headwalls/drainage - pipes Underpasses and other point hazards - Including those at the foot of (rivers, railway) an embankment Safety barrier terminations - Blunt barrier terminations and ramped ends which do not bend towards the roadside (see Chapter 4) 1 Does not include 'passively safe' posts and poles.

In Table 9, for slopes (cut and fill), ditches and drainage gullies minimum slope heights & gradients are given to identify when slopes becomes a hazard.

Table 9. Distributed Hazard Characteristics Identified in the RISER Detailed Database Hazard Height/ Gradient Dangerous Additional comments Depth [m] impact [m] speed [km/h] Cut (upward) slopes >1.0 >1:1 40 Fill (downward) >1.0 >1:1 40 In addition, ALL slopes/embankments embankments 6 m high or more (i.e. ALL set- backs). Ditches and drainage gullies >0.75 >1:3 40 (fore & back slope) Rock face cuttings/rock 50 Any exposed rock face fences cutting slopes <1.5 m above carriageway level. Retaining walls - Less than 1.5 m above carriageway level. Buildings/walls - Non-safety fences - Wire wildlife/boundary fences not considered as hazards. Old design safety barriers - Barriers not compliant with EN1317 and with poor performance records. Rows of trees/forests 40 Same measures as for individual trees. Adjacent roads, railways, - water hazards

Any of the previously defined point or distributed roadside hazards should also be treated as hazards if present within a central reserve.

A central reserve on any roads with a speed limit above 70 km/h which has a width of less than 10 m between opposing edges of the carriageway should itself be considered a hazard.

Although few conclusions can be made in this study regarding curves in roads and their potential as hazards, it is apparent from the detailed database analysis and from previous studies that curves in roads do impose an increased risk of a run-off-road accident. It is more difficult for a driver to recover from a run-off in a curve than on a straight section of road. Therefore, roadside hazards should be considered an even greater risk when located near curved sections of roads.

Safety Zone Criteria The dimensioning of a safety zone is a difficult process. A theoretical process using vehicle dynamics and human tolerance information provides results consistent with current practice if vehicle exit angles from the road are 5 degrees, which is the median value for the data collected in RISER. An alternative is to use the struck object set-back distance obtained from the accident data. In this latter approach, the data coming from RISER appear to support information from France, the US, and the Netherlands which shows that the risk of contact with an obstacle drops dramatically after the first few meters and most impacts with roadside obstacles occur in the first 10 m.

Most safety zones in Europe are specified to be between 6-10 m for travel speeds around 100 km/h. Safety zones are smaller for lower speeds and for 80 km/h roads, the same countries use 4.5-7 m as a safety zone width.

The RISER analysis provides two alternatives for designing the roadside safety zone.

1. Based on the risk of injury during an impact with a hazard, the safety zone can be dimensioned for allowable impacts with hazards. In this case the allowable impact speed for striking a hazard is given in Chapter 1 and the impact speeds are calculated from the information provided earlier in this chapter. 2. The safety zone can be dimensioned as the risk for a fatal impact with an object of a given set-back. Based on the RISER database, the set-back distances can be grouped into the categories based on the road characteristics.

The requirements for a well designed safety zone are that:

° the consequences of a run-off are reduced ° the width should be designed that most vehicles that leave the road do not leave the safety zone ° there should only be slopes that do not cause rollovers ° the surface should be homogenous and even to prevent rollover ° there should be no unprotected fixed objects located within the safety zone

Legislators and authorities should ensure that a safety zone only contains artificial structures that will collapse or break away on impact without significantly damaging an errant vehicle. Where allocation of the desired safety zone is not practicable, they should consider erecting an appropriate road restraint system.

Accident data collected in RISER indicates that most vehicle departures from the road were less than 20 degrees and 110 km/h. These run-off roads events involve a non-tracking (yawing) vehicle in about half of the cases. Impacts with roadside obstacles were observed up to 10 m from the road and 85 % of all roadside impacts happened within the first 7 m of the roadside. A roadside safety zone should be dimensioned to the local road conditions using the local accident data when possible.

Recovery Zone Criteria RISER’s analysis of different criteria for dimensioning the recovery zone has shown that the design of roadside environments is complex. For a road designer evaluating alternative designs and choosing among them is difficult because there are many levels of interaction between different road design components such as the road itself, speed, traffic volumes and terrain etc.

The information collected among RISER contributing partners’ national policies clearly show that the width of the recovery zone is different one country from another, some of the reviewed countries using the general roadside geometry to describe the safety risks for roadside environment.

From the research findings documented in the Design Guidelines [8], the recommended width of paved shoulders on non-motorway roads should be between

1 m and 1.5 m, value beyond which further widening does not seem to greatly improve safety (see Table 10 and Table 11). However these values can be smaller, above all in the outside of curves, and keep a significant efficiency. With paved shoulder widths greater than 1.5 m speeds and subsequent accident rates may increase. Further research should be conducted to confirm this assertion.

Table 10. Recovery Zone Width on Non-Motorway Roads Usage Recommended values Recovery of errant vehicles 1 m to 1.50 m Avoidance of overtaking and meeting vehicles 1 m to 1.50 m Avoidance of vehicles making a turn 1 m to 1.50 m Travel of vulnerable road users off the driving lane 0.5 m to 1.5 m

Table 11. Recovery Zone on Motorways Usage Recommended values Emergency lane 3 m to 4 m

Other findings documented in the Design Guidelines where:

° Paved shoulder condition and surface must have the same quality as the road. It must be constructed so as to bear the static load of heavy trucks. ° The material used to construct paved shoulders is very important. The recovery zone should have the same surface quality as the pavement and skid resistance should be identical to the carriageway. ° Loose materiel or grass shoulders degrade recovery and avoidance manoeuvres. In addition, bikers and cyclists are reluctant to travel on such uncomfortable material. ° Extra space should be installed opposite T-junctions and private accesses to enable avoidance of vehicles making a turn towards minor roads and private accesses. ° In combination with paved shoulders, audible road markings can be implemented alongside the main roads in order to alert the driver of an errant vehicle. ° Rumble strips or edge markings with jiggle bars should be provided to alert a motorist who is driving in a deteriorated driving situation. ° Paved shoulders should be implemented preferably on the outside of curves with radius greater than 200 m on single carriageway roads.

Road designers must recognize that the roadside environment and its design have a vital role to play in improving roadside safety. A great amount of research clearly show that a recovery zone has a positive effect on both accident rates and driver behaviour, provided that the abovementioned conditions of implementation are respected: a smooth and resistant surface made of asphalt or concrete, with no loose material, wide enough to allow vulnerable road users to make short trips off the carriageway but not too wide so that car drivers do not understand this roadway improvement as an extra driving lane.

For economical grounds, further research could be carried out to answer to the question: what is the optimal paved shoulder and lane widths?

Hazard Evaluation, Removal and Modification The development of a safe roadside environment depends on the identification of the hazards in the roadside environment. Once these hazards are identified and are known to be in the safety zone, suitable strategies (or countermeasures) are needed to protect traffic from these hazards. The Design Guidelines [8] provide guidance for the technical treatment of hazards on both new and existing roads.

As showed in Figure 5 trees are an important priority for improving roadside safety, but also vehicles rolling over in the terrain is a major problem in single vehicle accidents.

The hazards that are most common on European roads can be divided into 2 classifications: point hazards (see Figure 17) and distributed hazards (see Figure 20 and Figure 21). These two classifications create different procedures for selecting a countermeasure for the hazard.

Point Hazards Hazards that are restricted to a small area as can be subdivided into man made and natural features. From the RISER statistical database, the two primary point hazards overlooked in roadside areas are narrow objects like trees and poles. It is important that the safety zone is properly dimensioned for the road classification and that any trees or poles are removed from this area.

Figure 16. Passively Safe Lighting Column Figure 17. Hazardous Rigid Pole (picture courtesy Jan Wenäll, VTI Sweden)

Action 1 Remove the hazard. Man-made features in the roadside safety zone should only be there because of a functional requirement (lighting, signs, bridge support, etc.). If they are not required, then they should be removed. Trees and poles that are located in the safety zone but cannot be removed (aesthetic or functional requirements) need to be made less harmful to vehicle occupants.

Action 2 For man-made hazards, modify the hazard becomes the next step. For lighting and utility columns, energy absorbing and break-away structures are important structures to incorporate into the roadside area. The two different structures – deformable and rigid – are shown in Figure 16 and

Figure 17. The outcome from these two motorway accidents was quite different even though the impact speeds were quite similar.

Action 3 Natural features like trees are sometimes difficult to remove from the roadside area due to historic and aesthetic requirements. Therefore the third task is sometimes more applicable – protect the road user from the hazard (see Figure 18). Protecting the point hazard introduces added complexity to the roadside design. In addition to selecting a type of road restraint system, the designer must also consider the placement of the safety feature and the size of the area to be protected (see Figure 19). An important issue is the transition from single point hazards to distributed point hazards (such as a group of trees or lighting poles).

Figure 18. Rigid lighting columns protected Figure 19. The culvert is protected with a with guardrail short guardrail but the gantry pole and the old rigid lighting column are unprotected (Photo, HUT)

Distributed Hazards Distributed hazards by their nature encompass larger areas than point hazards (see Figure 20 and Figure 21). This can result in higher costs than for the point hazard. A similar process can be applied to distributed hazards as for the point hazards.

Figure 20. Unprotected rock cut close to the Figure 21. Untested guardrail with concrete carriageway posts and too low positioning

Action 1 Remove the hazard: As for point hazards, the nature of the hazard will determine what is possible. Man-made distributed hazards should be

designed so that they are not located within the safety zone. Similarly, natural hazards (rock faces, groups of trees, etc) should be set-backed from the road so that they are not located in the safety zone.

Action 2 Modify the hazard: Almost all distributed hazards of concern are related to the roadside geometry. RISER statistical accident data indicate that roadside geometry; including slopes, embankments and ditches (or no specific impacted object), contribute to almost half of all run-off-road accidents involving injury or fatality. These roadside features are believed to be the leading cause of rollover in single-vehicle, run-off-road accidents. The layout of the side slopes and ditches adjacent to the road are the main features that can be modified from a dangerous situation to a more gentle geometry.

Action 3 Hazard shielding: The application of road restraint systems (safety barriers) to protect distributed hazards is the best alternative when a hazard cannot be relocated outside the safety zone. The safety barrier must be selected to provide suitable protection for the exposed traffic and the dimensions of the hazard.

Hazard Protection An overview of passive safety road equipment was presented in the Design Guidelines [8] with a focus on road restraint systems, particularly safety barriers. The procedures for selecting other passive road equipment generally follow the same procedure. The first step is to identify the hazards that must be addressed. This will determine which type of passive safety road equipment is necessary. The hazard may be a lighting pole that can be replaced with an energy absorbing column or a rock cutting that needs to be shielded by a safety barrier.

The second step in selecting passive safety road equipment is to determine the containment level, or strength of the system. Safety barriers and crash cushions are classified by the size of the largest test vehicle used in the crash testing program. Energy absorbing poles are classified by the amount of impact energy they absorb in the crash test. Both of these ratings identify the structural capacity for the system.

The third step for selecting equipment is to identify the amount of space available for the systems dynamic performance. This is established by the proximity of the hazards being shielded. The location of the hazards is necessary to determine the working width of the system (for safety barriers), and the deflection classes for crash cushions.

The final step for determining the installation requirements of passive safety road equipment is to identify the length of the system. This is most relevant for safety barriers and is determined by the size and position of the hazard(s) and the expected accident configuration for the specific location.

Experience has shown that typical problems associated with road equipment are:

° Insufficient length of systems to shield hazards. ° Installations shielding hazards neglect neighbouring hazards. ° Insufficient free distances behind the system.

° Inappropriate end terminals for barriers.

Any passive safety structure used for protecting roadside hazards must be tested to European test requirements specified in EN 1317 (road restraint systems) and EN12767 (passive safety supports). Accident data and causation information analyzed in the RISER project (Deliverables 3 [3] and 4 [Error! Bookmark not defined.]) can be used to develop local policies for the selection and installation of roadside infrastructure.

The selection of road restraint systems should include maintenance and operation program of the road function (RISER Deliverable 8 [9]). Safety performance or roadside infrastructure can only occur if the equipment is maintained in good working order. This requires regular inspections and repairs when necessary. A reliable source of replacement parts and qualified service personnel is thus needed to keep all safety equipment within the manufacturers’ specifications.

Monitoring A road accident data collection system is required to monitor the performance of newly designed as well as existing roads. Through data collection, it is possible to learn more about the road transport system and help identify the need for safety countermeasures. For re-designed roads it is important to compare the safety levels of the road before and after the changes to find out if the changes were successful and quantify the benefit obtained with these modifications.

The current situation regarding accident data collection in European roads shows that, in order to find countermeasures to improve safety, more thorough information about the road infrastructure and the causation of the accident is required.

Statistic or base-level accident data are the most commonly available sources of information. These accident data, usually collected by the police, are often very general and the type of information differs considerably among European countries. All variables proposed to describe accidents in RISER statistical database are not fully available in national databases across Europe. Harmonised data elements describing European roadside infrastructure allow European-wide statistical studies. The information gathered in RISER provides a base to further develop this resource.

The data collected by the police is not sufficient for improving roadside infrastructure safety. Other information sources (i.e. road maintenance and detailed investigations, etc.) are needed to have a complete overview of the traffic environment. A valuable resource would be to have a system which collects data of all maintenance performed on road equipment and store it in a computerised database. When this data is compared to the police data, unreported accidents can be identified found and the frequency of accidents can be identified, independently of the injury severity. This makes it possible to monitor the actual accident rate, the performance of the road and roadside, and calculate the real costs raised by accidents on the road.

In-depth level accident data provide specific information that helps improve the knowledge of the accident event, and makes it possible to design or improve safety measures for identified problems. Detailed databases do not cover populations as large as the statistical ones, but the information they provide allows describing and analysing specific safety problems.

To reach a complete description of accident scenes and outcomes, it is important to not only have collection procedures covering a broad scope of information, but also a co-ordinated approach is desirable among the different collecting actors. The following procedures are a proposal to improve the type and quality of data collected for roadside safety issues.

Police Road accidents are reported differently in the European countries. Even if the police are often called to an accident scene, in some countries the police only report injury accidents, where in other countries they also report property damage accidents. It is preferable to register all of the accidents to get reliable statistics. By considering all types of accidents, it is possible to improve the usefulness of statistical indicators for accident risk rates and accident severity. These are used for the safety effectiveness evaluation of roads and road infrastructure.

When nationwide collection of accidents is not possible, an alternative is to collect data from all the accidents within a limited region. The geographical area selected should be representative of the whole country. Using statistical methods, the data can be scaled from the region to the whole country.

Maintenance and Operation Organisation In many countries the road maintainers report all repairs that are due to a road accident. Unfortunately not many countries have a system to store this data. Use of this data can contribute to finding:

° the safety performance of the road ° the actual cost for a road ° unreported (property damage) accidents

Hospital and Rescue Services Many injuries, especially to vulnerable road users, are never reported to the police. If a cyclist or a pedestrian falls on the road/pavement for example (without involvement of a motor vehicle) the injured person will probable be taken to hospital without being reported injured to the police. Only some of the countries (e.g. Sweden) have a system where both the police and the hospital report road accidents to the same database. Combining police and hospital data makes it possible to:

° improve the statistics on vulnerable road users ° compare injury outcome (police reported injury severity versus real injury severity) ° find unreported (injury) accidents

In order to improve the relevance of injury databases, it is important that unified procedures and criteria are set. Victims classified as “injured” should be followed up over fixed periods of time to report the final actual outcome of injuries, including eventually death. Currently, different methods are used throughout the European countries to report injuries.

In-depth Investigations When special studies are required, in-depth accident investigations must be performed. To find countermeasures for the road network it is important to include variables related to the road environment. Most of the existing variable lists are focusing on vehicles and the injuries to the road user e.g. the STAIRS protocol [10]. Within the RISER project an in-depth study on single vehicle accidents was performed. From the in-depth study a minimum set of road infrastructure variables was identified that is essential when investigating single vehicle accidents [2].

A variable matrix (see Table 12) was developed for a co-ordinated approach which is desirable among the different collecting actors presented above. The matrix is a proposal and should not be seen as a complete list. Fundamental variables are assumed to be collected by the police. Additional variables can be collected by road operators and the hospital and rescue services. In-depth accident investigations should collect all or part of the variables listed in Table 12, depending on the aim of the data collection.

All forms of accident data collections should include:

° date ° time ° type ° place (preferable GPS coordinates)

If the police for example could include more environmental variables to gain better statistics on road and roadside infrastructure it would enhance national databases and benefit the whole traffic safety area.

The colour coding for Table 12 is as follows:

Police

Road operators

Hospital and rescue services

Table 12. Variable Matrix for Accident Data Collection Background factors Pre-crash Crash Post-crash Driver under the influence of Age medicine, alcohol, drugs Injury severity Evacuation Time of accident within the drivers normal circadian Type of injuries (AIS code if Gender rhythm possible) Rescue

n Driver familiarity to the road a

m Type of road user system (surroundings) Injury cause Care u

H Vehicle speed compared to Placement in vehicle posted speed Impact points in the interior Died at scene Usage of passive safety Function of passive safety Post-crash times (alarm, Type of trip equipment systems to/from scene, to hospital) Driving licence (type, Death because of illness or possession time) suspicion of suicide Ejection, ejection path Make Technical failures Collision type Trapping Tyres (make, dimension, Rescue delayed due to the Model track depth) Collision speed vehicle Vehicle movement during Model year Tachograph data collision Possibility to open doors Colour Load (weight) Rollover movement Fire Kerb/gross weight Load contributed in accident Hit object Water submersion

e l Vehicle characteristics c i

h (driven axles, type of Vehicle movement on road Principal direction of force e

V gearbox, effect) before collision (PDOF) Type and position of Rest position (wheels, side, passive safety systems roof) Vehicle deformation (Collision Deformation Type of active safety Classification CDC if systems possible) Load influence on injury extent Light condition (daylight, Problem for rescue services Road location (urban/rural) darkness) Points of collision to get to the accident scene Traffic effected by the Type of road (classification) Weather condition Roadside type/outline accident Distance to medical rescue Traffic flow Road condition Hit object service Traffic flow at time of Distance to object in Posted speed accident roadside Temperature Road characteristics (geometry, surface type,

width, camber, Type of road- signs, signal Object protected by road Cost of the damage or t

n deformations) system, markings present restraint system repair e

m Road restraint system (type, n

o containment level, r i

v deformation, function, effect, n Roadside type Brake/skid marks failure) Repair time E Roadside characteristics (shoulder width and material, fore/back slope Road equipment influence gradients, ditch depth) Exit angle on injury extent Type of road equipment/ Preventive maintenance restraint system actions Road construction or maintenance zone Vehicle heading (north, south, west, east) The background factors are a description of the condition of the including components. Some of the background factors are the same for all accident types e.g. age, gender and type of road user etc. Other background factors are more specific depending on the accident type e.g. vehicle type, road geometry and roadside area etc. The background factors can also be placed under the different accident phases respectively if they are considered to be contributing factors in the accident. Example; if an accident occurred in a curve it might not be a contributing factor to the accident. However, if the curve camber was incorrect the curve can be considered as a pre-crash factor.

Maintenance and Operations Guidelines Maintenance and operations guidelines are an essential element of a road management concept, since the state of the art and the operational status of safety equipment and personnel are the key determinants for the achievable level of Infrastructure Road Safety (IRS). The main components of IRS for the roadside and median areas include the recovery and safety zones, road restraint systems, passive safety equipped support structures, road marking, and traffic signs.

Maintenance and operations of road safety equipment and infrastructure ensure that all safety related elements of the road system are operating as they were designed, tested, and approved. Maintenance of road equipment should not only be considered as the repair of broken or damaged equipment, but also as a potential monitoring system for the road network.

Maintenance and operations provide an important source of information for the road operator. It is crucial that an inventory of road infrastructure exists and the frequency of repairs, operational functions, and need for replacements can be identified. Without a maintenance programme, these critical issues cannot be addressed.

As essentials of a maintenance and operation management plan the RISER project have identified five areas of interest.

° Routine Inspection ° Data Collection ° Data Analysis ° Repair Plan ° Training

These topics cannot be considered separately, but as a total maintenance and operations programme. Information from one area is needed in several other areas.

Routine Inspections The results of the RISER study indicate that an inspection programme is necessary for identifying maintenance activities. It is important that three types of inspections are identified:

1. Safety inspections - are designed to identify defects likely to create a danger to the road users 2. Detailed inspections - are designed for routine maintenance tasks not requiring urgent execution 3. Safety patrols - are a supplement to safety inspections on the higher priority motorways and trunk roads

Damage or repair issues arising from these inspections should be prioritised for their repair urgency. At least three categories should be used:

° Level A – requiring prompt attention, as the defect presents an immediate or imminent hazard to road users

° Level B – defects which if not treated will get worse and cause major maintenance works at a later date ° Level C – defects that can be divided further with various levels of repair response times. Their response time can be adjusted to fit into a routine maintenance programme. ° Level D – defects requiring no repairs but should be monitored

The frequency of inspections must be determined for local conditions. It is crucial that high traffic roads (motorways and national roads) are inspected daily while minor roads have weekly inspections. Specific infrastructure inspections should be adjusted to suit the equipment performance requirements.

Reports from inspections should be incorporated into a database with basic information like date, location, and references to police reports when available. Pictures of the damage should be stored when possible.

Data Collection and Analysis The use of a digital database with photos obtained from inspection and maintenance activities will facilitate analyses of the road infrastructure performance and allow for better planning of investments of equipment and human resources. At the moment there is a lack of maintenance data in a suitable form suitable for these analyses as well as limiting the application of the Black Spot approach proposed in the RISER project. Therefore, further recommendations for the analysis and interpretations of maintenance data are not possible. However, the local use of Black Spot analyses can be adapted to incorporate maintenance data. It is crucial that all maintenance data should be stored in a suitable computer database that will allow processing.

Repair Plan There is a broad variation in the use of standards for repair plans between the participating countries in this study. We have to assume that new member states in the EU also will have different standards or no written standards or guidelines at all. A recommendation could be that the flowchart presented in Figure 22 could serve as a guideline for maintenance activities and that each country can define the different categories 1-6 in alternative C according to available road maintenance funds. A developed economy may then result in an upgraded category for a specific type of damage.

The alternatives A and B are applied for safety reasons and to avoid fast capital depreciation, when the damage is rapidly becoming worse.

The use of a digital database with photos and inspection and maintenance history will facilitate the choice of action as well as an analysis over time.

A. A hazardous situation? Site Attendance Permanent and Temporary Repair No Yes Repair

1. The damage has been reported X hours 30 days workdays at 8-16 hours? Yes

No 2. The damage has been reported Y hours 30 days outside normal working hours

B. The damage will become worse? Z hours 5 days Yes No

C. Other damage than A or B? Category 1-6

Yes Category 1 1 day 5 days

Category 2 5 days 20 days

Category 3 10 days 30 days No Category 4 20 days 60 days

Category 5 30 days next periodic maintenance

Category 6 N/A next periodic maintenance

D The damage requires no repair Monitor during regular inspections

Figure 22. Flowchart for choice of response time (temporary repair) and time until repair

Training A training programme for roadside infrastructure elements should be part of every national road safety policy. As identified previously, different categories of staff should participate in some level of training including inspectors, supervisors, road workers, office support staff to name a few. The level of training will be dependent on the role of the employee. Important topics to be covered include (but are not limited to):

° An overview of road and roadside infrastructure for hazard identification ° An understanding of the installation of road restraint systems and frangible posts/poles, plus an understanding of the use of road markings ° Principles of road safety auditing and risk assessment and their use as a method of inspecting roadsides for hazard identification ° Categorisations of defects and damage ° etc.

Training should be provided for new employees with refresher courses provided for individuals with training intervals suiting their job requirements. To date the UK has the best training system identified in the European Union and should be used as a reference.

List of Project Deliverables

The RISER project contract required the submission of twelve specific deliverables. Of these deliverables, eight are specific (public) reports and one dissemination website managed by one contractor. The relavant deliverables for public use have been distributed in a project CD available from the RISER consortium.

Table 13. Project Deliverables No. Submission Output Title of Deliverable Date from Tasks

D01 M 13 1.1,1.2, Accident Databases for Collisions with 1.4 Roadside Infrastructure

D02 M 12 1.3 Summary of Driver Behaviour and Driver Interactions with Roadside Infrastructure

D03 M 29 2.1,2.3, Critical Vehicle and Infrastructure 2.4 Interactions

D04 M 29 2.1,2.2 Identify the Envelope of Vehicle and Driver Response Prior to Collisions

D05 M 6 3.1 Summary of Current European Roadside Design Guidelines

D06 M 35 3.1,3.2, European Best Practice for Roadside 3.3,3,4, Design: Guidelines for Roadside 3.5 Infrastructure on New and Existing Roads.

D07 M 8 4.1 Summary of Maintenance and Operational Procedures for Roadside Infrastructure

D08 M 34 4.1,4.2, European Best Practice for Roadside 4.3 Design: Guidelines for Maintenance and Operations of Roadside Infrastructure

D09 M 6 5.4 Dissemination Plan.

D10 M 8 5.2 Website (www.riser-project.com)

D11 M 33 5.1,5.3 Dissemination material

D12 M 3 6.0 Detailed work plan on a task and partner level.

Comparison of initially planned activities and work actually accomplished.

The activities in the RISER project followed the original project plan that was proposed in the project proposal. The work package and task structure presented in Any major deviations from the work content of the Description of Work should be justified and their effects on the project discussed. Tables or charts should be used where appropriate.

Figure 23 Project Gannt Chart

Management and co-ordination aspects

The project was managed through a regular exchange of information between the partners and project coordinator. A critical factor in the success of the project was the use of frequent project meetings. These meetings usually were held over 2 days and were shared among the partners. This encouraged participation of all the partners.

An important activity that helped with the partner engagement in the project was the use of project workshops to encourage a dialogue between the project contractors and external participants, stakeholders in roadside safety problems.

All of the RISER partners were involved in the project deliverables, both in the supply of data as well as text contributions. The final deliverables list the main contributors to each deliverable and indicate the general level of involvement in the project.

There were some personnel changes during the duration of the project. The following list provides the contractors and the respective individuals involved wioht the RISER project,

Table 14. RISER Contractors and Key Researchers Organisation Acronym Country Chalmers University of Technology (Coordinator) CHALMERS Sweden Robert Thomson, Helen Fagerlind, Gunnar Lannér Centre d’Etudes Techniques de l’Equipement Normandie Centre CETE France Guy Dupré, Olivier Bisson Fundación para la Investigación y Desarrollo en Transporte y CIDAUT Spain Energía Juan M. García, Francisco López European Union Road Federation ERF Belgium Brendan Halleman, Jose Papí HIASA Grupo Gonvarri HIASA Spain Angel V. Martínez, Antonio Amengual Helsinki University of Technology HUT Finland Jarkko Valtonen, Marko Kelkka, Ute Gosse Netherlands Organisation for Applied Scientific Research TNO The Cees W. Klootwijk, Boudewijn Hoogvelt Netherlands Richard van der Horst, Selma de Ridder Graz University of Technology TUG Austria Heinz Hoschopf Vehicle Safety Research Centre, Loughborough University VSRC United Claire Naing, Julian Hill Kingdom Volkmann & Rossbach GMBH & CO.KG V&R Germany Wolfgang Wink

Results and Conclusions

The RISER project has made a significant contribution to the understanding of single vehicle accidents in Europe. During the duration of the project, two important data sources were developed to identify the characteristics of SVAs in Europe. This data became a foundation upon which further studies on the human factors, crash performance, and maintenance of roadside infrastructure elements could be developed.

The RISER documents provide a European reference that can be used to improve road safety levels through the improvement of roadside infrastructure. It is important to recognize that road infrastructure improvements benefit all road users and have no particular vehicle or driver requirements to be effective. Road infrastructure should be a democratic component of the road network, serving all road users.

The main output for RISER can be extracted from deliverables 6 and 8. These reports identify the application of the information gathered in the remaining deliverables. Of course more information can be obtained from review of all the reports, but the targeted user of the RISER data will benefit from a condensed summary of the RISER results in two concise reports.

Specific results from the RISER project include:

• Accident based analysis of roadside objects with criteria defining when these objects become hazardous to vehicle occupants

• Crash performance information for different roadside infrastructure elements identifying critical issues for selecting, installing, and maintaining roadside environments

• Information identifying the influence of roadside features on the neighbouring traffic flows

• Recommendations for monitoring safety levels of roadside features

Although not every roadside safety issue can be addressed in any single project, the RISER project has made a significant contribution to the understanding of single vehicle accidents in Europe. Continued research activities following the themes in RISER are recommended. Single vehicle accidents are a significant component of the annual fatal statistics for European roads and countemeasures addressing this accident type is a necessary component of any road safety program.

Acknowledgements

The RISER consortium is grateful for the cooperation and/or financial support of many organisations. First the support of DG-TREN and the “GROWTH” programme in the 5th Framework Research Programme must be acknowledged to facilitate this project to be conducted at a European level. The following list highlights some of the prominent national contributors to the RISER project.

From Sweden: • Kenneth Svensson and Kent Sjölinder at the Swedish Road Administration, Western Region office (Göteborg) for their assistance with the accident data collection • Anders Håkansson and Per Strömgren from The Swedish Road Administration, Department for Road and Street Design for information related to roadside design practice • Anders Ydenius and Anders Kullgren from Folksam Research for insurance industry information

From the United Kingdom

• Steve Proctor from TMS Consultancy in the UK, for his valuable contribution in the areas of highways design and design regulations. • The Highways Agency for their help and support throughout the project, in particular for the provision of valuable information resources. • The UK Department for Transport's Statistics Division, for the provision of UK STATS19 data for inclusion in the RISER Statistical Database. • The UK Cooperative Crash Injury Study for the provision of accident data for inclusion in the RISER Detailed Database. CCIS is funded by the Department for Transport (Vehicle Standards and Engineering Division), Autoliv, Daimler Chrysler, Ford Motor Company, LAB, Nissan Motor Europe, Toyota Motor Europe and Visteon. • CCIS is operated by teams from the Birmingham Automotive Safety Centre of the University of Birmingham; the Vehicle Safety Research Centre of Loughborough University; the Vehicle and Operator Services Agency of the DfT and TRL Limited. Further information on CCIS can be found at www.ukccis.org.

From the Netherlands • Fred Verweij from Ministry of Transport, The Netherlands for his expertise and support for the RISER project

From Finland: • The Finnish Road Administration • The Finnish Motor Insurers' Centre (VALT) for access to their accident database

From France: • The French Ministry of Transport for access to national accident statistics • Richard Driscoll at CEESAR for his processing of the indepth accident analysis

From Austria: • Accident data made available from ‘Kuratorium für Schutz und Sicherheit’

From Spain • General Directorate of Traffic (DGT) • Road operators in Asturias region for providing maintenance data

References

1 Collin, C. Statistics in Focus – Transport, Eurostat catalogue number CA-NZ-00- 003-EN-I, 2000 2 D01: Accident Databases For Collisions With Roadside Infrastructure, European Community R&TD Project, 5th Framework Programme “Growth”, Project "RISER" GRD2/2001/50088, 2004 3 D03: Critical Vehicle and Infrastructure Interactions, European Community R&TD Project, 5th Framework Programme “Growth”, Project "RISER" GRD2/2001/50088, 2005 4 D02: Summary of Driver Behaviour and Driver Interactions with Roadside Infrastructure, European Community R&TD Project, 5th Framework Programme “Growth”, Project "RISER" GRD2/2001/50088, 2004 5 D05: Summary of European Design Guidelines for Roadside Infrastructure, European Community R&TD Project, 5th Framework Programme “Growth”, Project "RISER" GRD2/2001/50088, 2003 6 D07: RISER D07 - Summary of Maintenance and Operational Procedures for Roadside Infrastructure, European Community R&TD Project, 5th Framework Programme “Growth”, Project "RISER" GRD2/2001/50088, 2006 7 D04: Identify envelope of vehicle and driver response prior to collisions, European Community R&TD Project, 5th Framework Programme “Growth”, Project "RISER" GRD2/2001/50088, 2005 8 D06: European Best Practice for Roadside Design: European Community R&TD Project, 5th Framework Programme “Growth”, Project "RISER" GRD2/2001/50088, 2006 9 D08: European Best Practice for Roadside Design: Guidelines for Maintenance and Operations of Roadside Infrastructure, European Community R&TD Project, 5th Framework Programme “Growth”, Project "RISER" GRD2/2001/50088, 2003 10 Standardisation of Accident and Injury Registration Systems, Final Report, Transport RTD Programme of the 4th Framework Programme, STAIRS Project RO-96-SC.204, 1999