Highways Agency

Chapter 2 - UK Experience

2.1 This chapter describes the rationale behind the development of safety barrier advice and the basis of the UK standard. It also briefly describes some recent UK research and its relevance to the development of a new standard.

2.2 This chapter also reports an analysis of accident data where vehicles have left the carriageway on the nearside of English major roads and compares this with accident levels elsewhere in the UK and those on minor roads. The Group was also interested in understanding the extent to which safety barriers are installed on English major roads. Within the allowable timescales it was possible to examine data for sample road lengths and to look specifically at motorway bridges over rail lines.

UK Standards

2.3 The current UK safety barrier standard ("Safety Fences and Barriers" TD 19/85, see Annex 3) is part of the Highways Agency Design Manual for Roads and Bridges (DMRB) and was first issued in 1985. It has been supplemented over time by publication of detailed drawings and advice on the use of new types of barrier, for example Interim Advice Notes 24, 26 and 34 on the use of temporary safety barriers on road works and clearance requirements. Further advice on the use of safety barriers adjacent to bridges is contained in the standard for bridge parapets ("The Design of Highway Bridge Parapets", BD 52/93). Safety barrier requirements also form parts of other standards in the manual. An updated safety barrier standard bringing together all the advice for the UK is being prepared.

2.4 The Highways Agency produces a range of standards and advice contained in its Design Manual for Roads and Bridges. The Manual sets a standard of good practice for the design, assessment and operation of trunk roads (including motorways). It is also applicable, in part, to other roads with similar characteristics. Scotland, Wales and Northern Ireland are co-signatories to the Manual and have in general adopted the Manual for their road networks (the manual contains specific National variations of some of the standards). The DMRB is commended for use by local highway authorities, as the advice is directly applicable to local roads where vehicle speeds are 50mph and over. Where the DMRB is used for minor, lower speed local roads, it is for the local highway authority to decide on the extent to which the documents in the Manual are appropriate.

2.5 The DMRB embodies the collective experience of the Highways Agency, its agents and design organisations over many years, and the other major highway authorities, and as such, represents a guide to best practice. It provides technical requirements and guidance resulting from research and practical experience in the management of the trunk road system in the UK and from overseas. Many of the standards are based on safety needs and they are reviewed and updated as necessary.

2.6 The various parts of the DMRB are the responsibility of the relevant professional or technical specialists within the Highways Agency. They have the responsibility to monitor continuously the performance of the standards by obtaining feedback and keeping abreast of relevant developments in the appropriate technical fields. Standards, Advice Notes and Specifications are reviewed and monitored by commissioning research or other studies as required. When amendments or additions to the DMRB are thought to be appropriate, the Highways Agency convenes a Technical Project Board. This comprises of representatives from Scotland, Wales and Northern Ireland, the Highways Agency and appropriate representatives from industry or local government, who review the proposed amendment or addition, or steer its development. Only if the Technical Project Board is in agreement does the new or amended standard go forward to the National Road Administration Chief Engineers for their approval.

2.7 The DMRB has been prepared for use by appropriately qualified and experienced professionals. It is not a statutory or regulatory document, neither does it cover every requirement in exhaustive detail. Many matters are left to the professional expertise and judgement of users, whilst other matters are covered elsewhere, for example in British and European standards. The British Standards Institution has published a companion document to TD 19/85 as a Published Document PD 6634 Parts 1 to 4 (1999), which describes the different types of vehicle containment systems and their technical development. The technical requirements in the DMRB are to be considered as a desirable minimum and must be adhered to for motorways and trunk roads unless a departure from the requirements is approved (known as a Departure from Standards). The procedure for considering and granting a Departure from Standards involves the designer formally submitting a proposal for departing from requirements with full particulars and a justification to the technical specialists responsible for overseeing the standard. Departures from Standards are only granted in exceptional circumstances where there is little practical alternative except at disproportionately high cost. Designers use their expertise in deciding what additional provision is appropriate at particular locations.

Safety Barrier Development

2.8 The early safety barriers were developed in the USA in the late 1950s and refined through full-scale impact testing, engineering judgement and experience. Many countries began to develop their own safety barriers based on these early systems and applied them to meet their own circumstances. Prime factors influencing the provision of safety barriers included road geometry such as width, profile and horizontal and vertical curvature. Other factors covered traffic volume, speeds and composition.

2.9 The prime requirements for safety barriers are to prevent vehicles from crossing from one carriageway to the other and to prevent errant vehicles (that is, vehicles leaving the carriageway) from impacting with or entering roadside hazards. A safety barrier is intended to absorb some of the energy from the impact caused by an errant vehicle striking it and to redirect the vehicle along the line of the barrier so that it does not turn around, turn over or re-enter the stream of traffic. This is called containment. Safety barriers can be either flexible, such as a steel wire rope or a steel beam, which absorb much of the energy from an impact, or relatively rigid such as concrete barriers, with less absorption of energy. Both types can have the same level of containment but may have very different stiffnesses. Use of standards for safety barriers in the UK result in barriers being provided to protect road users where there is a risk as identified by the standard and where their use can be justified, both in cost and safety benefit terms. Safety barriers are only used where the consequences of a vehicle striking a barrier are considered likely to be less serious than if the vehicle were to continue unrestrained. They cannot offer complete protection from injury to the road user.

2.10 International developments have led to a very wide range of safety barrier designs, which are applied to meet a range of impact considerations by differing vehicles, on a variety of roads. In all instances in the UK barriers have been accepted following validation using full-scale, standardised impact tests, the parameters of which have been determined to meet different national requirements. Full-scale repeatable and reproducible impact tests carried out under strictly controlled conditions continue to be used to validate barrier systems. Two separate impact tests are generally required for this validation. The first test uses a vehicle ranging from a 1500kg saloon car (such as a Ford Granada) to a 30t rigid tanker. This test is mainly used to assess the containment (or 'strength') of the barrier. The second test uses a 900kg vehicle (such as a Ford Fiesta) to assess the reaction of a lighter vehicle impacting the barrier and the severity of any injuries to occupants in the vehicle when it impacts such a barrier.

2.11 In the UK, a study of the types of vehicle in general use led to the adoption of a 1500kg saloon car and an impact speed of 70mph as the most appropriate standard parameters that safety barriers would be expected to control. It was recognised that there was a very wide variety of vehicles in use, but at the same time it was accepted that it would be impractical and uneconomical to use a range of vehicles to validate designs of safety barriers. Following analysis of the dynamics of vehicle trajectory and of accident data, an impact angle of 20 degrees was adopted as being a representative angle of approach. Testing to these parameters (ie a 1500kg vehicle impacting a road restraint system at 70mph, and at 20 degrees) led to the designation of 'normal' containment for a barrier dealing with these parameters. Studies considering heavy goods vehicles led to the adoption of 16t (such as a 2-axle rigid flat bed) and 30t (such as a 4-axle rigid tanker) vehicles for barrier testing.

2.12 The implementation of the European Directive on Construction Products (89/106/EEC, December 1988 as amended by Council Directive 93/68/EEC, July 1993) resulted in the need to harmonize national safety barrier standards across Europe to a common European Standard (EN 1317). This Standard establishes general criteria for testing road restraint systems, with specific performance and test criteria for safety barriers and crash cushions. Four of the six parts of the Standard have been published.

2.13 While it is international practice to validate safety barrier systems by a limited number of successful full-scale impact tests, it is also appreciated that many vehicle impacts with barriers on the live road network lie outside the range of the validation tests. Vehicles will impact with barriers at a variety of speeds, some in excess of the 70mph performance parameter, at a variety of angles of approach and under a variety of levels of driver control. Whilst testing can provide some indication of the likely performance of a safety barrier on the road network, these tests should be regarded as a procedure to select only those systems that meet the minimum performance criteria. Monitoring of in service performance and feedback from accident data, where impacts occur at different angles and speeds, is a recognised means of updating and developing products. As products and circumstances change in the future, so will the validation requirements.

2.14 Having a range of approved safety barrier systems of different performance characteristics leads to the requirement for describing where and to what extent they should be deployed. Taking the most widely used performance parameter as an example (that is to contain a 1500kg saloon car, with minimum effect on the occupants), this indicates the selection of a flexible steel safety barrier with 'normal' containment. These barrier systems are installed on the majority of central reserves on major roads in England, primarily to inhibit vehicles crossing the central reserve. Flexible barriers that meet this level of containment are also installed on the nearside of major roads to prevent contact with defined hazards. An alternative, where there is limited space to allow for the deflection of a flexible barrier, would be to select a rigid concrete barrier. However, there is anecdotal evidence to support the belief that an impact with a rigid barrier will result in more serious consequences for the vehicle occupants than a flexible barrier.

2.15 The objective of installing nearside safety barriers is to reduce the consequences of vehicles leaving the carriageway and encountering a hazard. The UK standard requires normal containment barriers to:

• provide protection to road users from engaging with roadside hazards (high embankments, water, rail lines, tight curves, solid objects, sign and gantry legs, lighting columns and trees);
• be positioned at least 1.2m from the edge of the carriageway;
• be 0.61m high (to the centre of the barrier beam);
• be a minimum length of 30m in advance of a hazard.

The minimum length of 30m was related to the minimum length at which a safety barrier can be expected to perform most effectively, based on the impact tests.

2.16 It is part of the continuing responsibility of the Highways Agency to monitor the overall need for safety barriers and the degree of containment required taking into account a wide variety of parameters, one of which could be the possibility of an errant heavy goods vehicle. Accident data for vehicles over 3.5 Tonnes in weight indicates that, while they constitute 10% of the total vehicles on major roads, they are involved in only 7% of nearside single vehicle accidents (see Annex 4 Tables A4.3.5 and A4.4.8). Monitoring and accident reports suggest that although the normal level containment safety barrier is designed and tested for 1500kg vehicles, they may also help to contain some heavy goods vehicles. However, if such a vehicle hit the supporting section of a bridge over a major road, then there could be the possibility of a catastrophic structural failure. There are therefore instances where a containment level higher than that for a 1500kg car could be justified for either the nearside or the central reserve. There are a number of barrier systems offering high and very high containment available for use at these particularly vulnerable locations. As the consequences to the occupants of a car striking these barriers is thought likely to be more serious than for impacting normal containment barriers, their use has to be carefully considered and justified.

2.17 Details of the high containment Double Rail Open Box Beam (DROBB) are available in the Highways Agency's Manual of Contract Documents for Highway Works. The very high containment Higher Vertical Concrete Barrier (HVCB) and requirements for the positioning of these barriers are contained in drawings currently issued to designers on a site specific basis by the Highways Agency. The standard requires high containment DROBB barriers, which have been validated for use with vehicles up to 10 tonnes to be used where there is an exceptional hazard to the occupants of vehicles. The very high containment HVCB, which has been validated with vehicles of up to 30 tonnes, can be used at sites where there is an abnormally high risk, such as the risk of an errant HGV causing a catastrophic structural failure.

2.18 The Group discussed the content of the current UK standard and considered that it covered the relevant areas where advice is needed. However, it had been difficult to identify all of the background and reasoning behind the standard, as the document is written for use by highway design practitioners. The Group considered that it would not always be clear to others with an interest in safety what lay behind the current advice. The Group would wish to see more background to the advice to aid the dialogue with third parties and suggested that a more formal framework for regular reviewing and updating standards may be more transparent.

UK Research

2.19 The development of advice on safety barriers to contain cars has taken place over many years, primarily by TRL on behalf of the Highways Agency. TRL, under a similar arrangement, have developed a range of barrier systems to contain heavy goods vehicles.

2.20 The Group was made aware of three topics of UK research ("Accident Study into Safety Fence Provision", TRL 1998) that suggested that changes to the current nearside safety barrier standard should be considered. These topics are described below:

1. One item defined as a hazard is a lighting column. The frequency of their installation on the nearside affects the likelihood of an impact between a vehicle and the column. Spacing and distance from the edge of the carriageway are factors affecting safety. More specific advice on the protection of lighting columns is required following an analysis of accident data.
2. At present, safety barriers eventually end by being anchored to the ground at each end, where the barrier angles down from its standard height to ground level. This results in a 'ramp' effect at the end of barriers. Fatal accidents have been reported where cars have hit these ramped ends and then been directed into the hazard that the safety barrier was protecting. Full-scale impact tests have been carried out by TRL to study this phenomenon and the research has led to the development of performance criteria for energy absorbing crash cushions that could be used at the ends of safety barriers. These criteria are now included in the European Standard. A study by TRL of the frequency of barrier endings reported the potential benefit of reducing the number of open gaps between successive safety barriers, where the gap was less than 50m. Accidents where vehicles struck the ends of safety barriers have led to the need for advice on the removal of short gaps in safety barriers.
3. TRL has studied the application of safety barriers on the nearside of roads. One particular area of interest has been the geometry of embankments at the roadside. This has identified the need to define not only the height but also the gradient of any embankment when justifying the provision of a safety barrier. The ability of an errant vehicle to cross the slope of an embankment is an issue to be taken into account.

The Group considered that the current programme of UK research will lead to improvements to the standard and other advice. The results of such research need to be examined to see how appropriate they are before they are incorporated into new advice. The Group discussed the UK research that had been made available to them and considered that where appropriate it should be included in a revised standard.

Analysis of Accident Data

2.21 The Group wanted initially to be aware of the safety record on UK roads in relation to other countries. UK roads are among some of the safest in the world and compare favourably with other European and developed countries. In terms of fatalities per head of population, England has the lowest number; other parts of the UK reflect well (see Annex 6 Table A6.9). An alternative way of looking at the safety record on UK roads is to look at the number of fatalities per distance travelled (see Annex 6 Table A6.10). Here the UK falls into the middle band with other European countries falling either side.

2.22 The Group asked for a detailed analysis of accident data where single vehicles had left the carriageway on the nearside. These data are only available from accident records (known as STATS 19 forms) completed by the police at the scene of a reported accident where there have been injuries. TRL maintains a database of these records (some 240,000 reports being added each year) and were asked to carry out a variety of analyses to assist the Group in its work. The analyses presented (see Annex 4) are mainly based on English data with summary tables showing broad comparisons with other parts of the UK (see Tables 2.2 and 2.3 below). The STATS 19 reporting system is only applicable for injury accidents and there is currently no mechanism for collecting information relating to damage only accidents.

2.23 The STATS 19 records hold a variety of information about each accident and the resulting injuries. The data are verified by a lengthy process of checking and is made available on an annual basis for Great Britain in the form of statistics published by DTLR. Highway authorities use the data to identify particular accident problems and places on the road network that require a more detailed safety study. The time available to the Group for analysis only allowed for a statistical approach to the data to be taken. Of particular interest to the Group was the breakdown of the data by road type (major, minor, etc) and roadside hazards that caused injuries. Details of individual accidents, for example driving conditions, were not analysed due to time constraints.

2.24 There is evidence that an appreciable proportion of non-fatal injury accidents are not reported to the police (a number of studies have investigated under-reporting by comparing police and hospital statistics). Therefore the accident data collected using STATS 19 forms will under-report the level of injury accidents. The DTLR has analysed this under-reporting and makes comments on it in the introduction to their annual statistics reports (for how this under-reporting may be allowed for, see Annex 5). This development is not relevant for the purposes of comparing numbers of accidents, as it applies to all the accident data, and so no adjustment has been applied to the data for this chapter. An adjustment has been used in Chapter 3 and is essential when estimating risks and the severity of injuries associated with accidents.

2.25 The Group wanted to look at the breakdown of accidents in order to take a view on the adequacy of the safety barrier standard. Before examining the data in detail they asked to look at some broader road statistics about road types and traffic levels that placed their task of looking at major roads into a wider context.

2.26 The Group were made aware of the proportion of the English road network to which their work could be applied. Despite major roads only being 5% of the total road network they carry around 45% of the traffic (see Table 2.1).

Table 2.1 Percentage of Traffic on English Roads (based on Annex 4 Table A4.3.1 and Annex 4 Table A4.3.3)

2.27 The Group wanted to be satisfied that their considerations could be applied to the rest of the UK, even though the Group was considering data covering only English accidents. Table 2.2 shows the lengths of motorway and trunk road and other roads in the UK for comparative purposes while Table 2.3 shows broad levels for nearside accidents on major roads in England, Scotland and Wales.

Click to view Table 2.2

Table 2.3 - Comparison of nearside accident figures on major roads across parts of the UK (from STATS 19 data)
Notes: 1 - road types are not classified exactly the same in all countries and figures are based on the closest comparisons
2 - details for Northern Ireland are not available

2.28 The Group were satisfied from their considerations of the comparison data and on the advice of representatives from Scotland, Wales and Northern Ireland, that recommendations made by the Group based on English accident data would be relevant to the rest of the UK.

2.29 The Group wanted to understand the relationship between the type of road and the quantity of nearside single vehicle accidents. The data indicated that:

• such accidents are only 5% of accidents on all English roads (see Annex 4 Section 4 Table A4.4.3);
• they account for 11% of the annual road fatalities in England (see Annex 4 Section 4 Table A4.4.14);
• the occurrence of these accidents is significantly greater in absolute terms on single carriageway and minor roads (see Figure 2.2); and
• about one quarter of these accidents in England occur on the major road network, which represents less than 5% of the total road network, whilst carrying over half of all traffic.

Click to enlarge chart

The Group discussed these figures and agreed that there are a significant number of single vehicle accidents on the major road network but that they present a more serious problem on minor roads.

2.30 TRL undertook a fuller analysis for the Group and examined the number of injury accidents where particular roadside hazards are struck or entered. The results are in Figure 2.3. As these accidents have all involved injuries they show that hitting objects at the roadside results in injuries to the road user and that there is a range of roadside hazards involved in injury accidents. They further show that hitting a safety barrier or even not hitting a roadside object at all ('hitting nothing') can also result in injury accidents. Hitting nothing could include, for example, leaving the carriageway and overturning down an embankment.

Click to enlarge chart

2.31 TRL have separately investigated the ability of safety fences to contain and redirect vehicles by sampling some typical lengths of English motorway at sites with characteristics that are similar over a length of at least 400m ("Accidents at hard shoulder discontinuities on dual 2-lane and 3-lane motorways", TRL 1998). They classified sites according to the major characteristic of each site, as:

• level, in cutting or on embankment, all substantially without safety barriers;
• level, in cutting or on embankment, substantially with safety barriers; and
• on a bridge (with or without safety barriers)

and considered the rates of personal injury accidents at the sample sites. Accident figures for lengths of motorway with and without safety barriers suggested that providing safety barriers does reduce the risk of accident injuries where there are embankments, for example, but raises the risk on a level section. It was found that where sites were level with no safety barrier, there was a lower risk of injury in accidents than sites which were level and where a safety barrier was provided (see Annex 4 Section 5). The figures also show that for single vehicle accidents, accident rates for motorway lengths with safety barrier, level run off, cutting or bridge parapet are fairly similar.

2.32 The results from this TRL work are generally supported by the examination of the trends in reported nearside injury accidents on English major roads over a period of nine years (1990-1998) as shown in Annex 4 Section 4 Table A4.4.18. Injury accidents where vehicles have struck roadside hazards show a general downward trend whilst accidents where safety barriers are struck are increasing. This should not be surprising; over time more safety barrier has been provided on major roads and so the probability of a nearside accident involving a barrier should be expected to increase.

2.33 The Group discussed the results from all the TRL work and considered that increased use of safety barriers could account for the reduction in hazards struck. The Group recognised that barriers themselves can cause injury accidents, as can a vehicle running off the road and not striking a roadside hazard. This tended to suggest that as safety barriers were installed to protect the road user, there should be a clear relationship between the provision of a barrier and the risk associated with the hazard that is being protected (see also paragraph 3.6).