Road Safety Manual
A manual for practitioners and decision makers
on implementing safe system infrastructure

You are here

8.2 Designing Infrastructure to Encourage Safe Behaviour

The Basics: Road user Capacities and Behaviours

Survivability of impacts

The Role of Speed Management in Safe System Elements and Application dealt with the relationship between traffic speed and crash survivability. The Safe System is designed around this basic concept; in particular, the speeds at which road users are unlikely to survive a collision. The speed at which impacts are survivable varies according to the type of collision (see The Safe System Approach). Some key design principles of a Safe System follow directly from this pattern of relationships.

As stated in Introduction, road users travelling on roads that have been designed with their characteristics and limitations in mind will avoid many of the mistakes that lead to collisions, but not all of them. Leaving the road and colliding with a barrier or a lighting pole should not result in death or serious injuries, so long as the vehicle’s speed was reasonably close to the speed for which the road was designed. This will not hold true if the speed is much higher than this design speed. Management of speeds is therefore of critical importance in achieving a Safe System. Since vehicle speed is usually under the direct and immediate control of the driver or rider, it is essential to provide the road user with clear signals about the type of road environment they are in, reinforced with specific information at appropriate points.

These requirements are largely addressed by the concept of self-explaining roads, which is discussed in Road Hierarchies and Self-explaining Roads in Designing Infrastructure to Encourage Safe Behavior.

Communication with road users

Messages relating to regulatory requirements, warnings of hazards, directions, and other useful information can be conveyed to drivers and other road users by a number of means. These include:

  • signs – static or changeable electronic signs, using words, pictograms or diagrams showing directions;
  • signals – at intersections, pedestrian crossings, and railway level crossings, and which cater for particular vehicle types e.g. buses, bicycles, emergency vehicles ;
  • road markings – these are a highly intuitive way of conveying messages, such as the path to follow, areas to keep off, and points at which to stop;
  • delineation – guide posts and retroreflective pavement markers have an important role to play in providing guidance in poor weather or at night. Pavement markers generally supplement the information provided by line marking.

The PIARC HFPSP guide discussed in Introduction includes actions to improve delineation or signage as possible remedial treatments for all of the three key requirements mentioned above, usually as a corrective measure to bring about a satisfactory resolution of the issue, or as a warning to indicate a potentially hazardous situation. Human factors considerations are critical in the design and provision of these treatments. The principal factors to be considered are:

  • Conspicuity – is the sign easily noticed?
  • Size and position of devices in relation to the required decisions or actions – e.g. Is the traffic control device sufficiently in advance of the required decision or driving action to allow the decision to be made or the driving action executed in an unhurried manner? These issues have to be considered at each site, initially by the road design team, and subsequently by the road manager during the road’s operational life.
  • Visibility – e.g. Does the design of the sign allow it to be easily read? Are signs or line markings faded or worn away so much that they are no longer effective? Are they still reflective at night? Checking and inspection to ensure visibility is a matter for the road manager. While physical measurements are the most reliable guide to the condition of traffic control devices, a well-managed programme of regular inspections by experienced personnel who are trained to make consistent assessments can be an acceptable basis for managing the condition of traffic control devices.
  • Comprehension – e.g. Can the signs (especially symbol signs) be readily understood by road users? The design of nearly all signs is specified in the standards or guidelines of individual jurisdictions; some of these designs comply with worldwide practice, some have proved effective over a long period of time, and newer signs may have been subject to rigorous testing before they were included in the standard. Difficulties can arise when new signs are required to cover unusual situations and the proposed signs are not adequately tested on road users.
  • Credibility – e.g. Are the messages they convey believable? This depends very much on how they are used in the local context. Overuse of signs can lead to them being widely disregarded. Consistency in the use of line marking, delineation and signing over lengths of road is essential in establishing the credibility of the system.

Note that a Safe System is not created solely by communication with road users. However, clear indications of the expected driving actions, especially speed choice and clear warning of hazards, can do much to reduce the number of collisions and to mitigate the severity of those that do occur. Therefore, human factors make important contributions to achieving a Safe System, and have a critical contribution to crash reduction on road systems that fall short of Safe System requirements.


Separation is required when operating speeds exceed the survivability limits for a particular class of road user. Once speeds exceed 30 km/h, it is essential to separate pedestrians from motorised traffic. This can be achieved by the provision of a raised footpath alongside the road, and the provision of pedestrian crossings at suitable intervals. These crossings may be unsignalised in situations where traffic speeds and/or volumes are low. Crossings should be signalised in busy environments to ensure pedestrians have ample opportunity to cross the road, and in high speed situations to reduce the chances of collisions with serious outcomes.

Safe System principles require that roads are designed to eliminate the possibility of road users being killed or seriously injured. At speeds of 50 km/h or above, side impacts are generally not survivable, so compliance with Safe System principles requires that cross-traffic movements should be controlled (e.g by roundabouts) at these speeds. While it may not always be practical to meet these requirements, consideration should always be given as to how best to avoid side impacts in environments where the operating speeds are 50 km/hr or above.

Head-on collisions are generally survivable up to speeds of 70 km/h. Compliance with Safe System principles therefore requires that where roads exceed 70 km/h, measures should be put in place to separate the opposing traffic streams, such as barriers or medians.

For the highest speed roads, complete separation is required by means of median separation, grade separation at intersections, exclusion of slow traffic, and fencing and barriers to exclude pedestrians. The maximum speed permitted on the highest standard roads varies from country to country, generally ranging between 100 km/h and 130 km/h.

It is well-understood that different standards of roads result in different levels of road casualties, although data on this point can prove elusive. Figure 8.3 is based on a comprehensive analysis of fatal and serious injury crashes and traffic data from the UK (Lynam & Lawson, 2005).

Figure 8.3 Fatal and serious collisions per billion vehicle kilometers traveled on a different class of road - Source: Lynam and Lawson (2005).

Note: VRUs (vulnerable road users) are excluded from the other categories.

The figure suggests the following:

  • Crash rates of all types are lowest for motorways and, with the exception of ‘other’ crashes, lower for dual carriageway roads than single carriageway roads.
  • Motorways have hardly any head-on crashes and very few crashes involving vulnerable road users or run-off-road crashes.
  • There is least difference between the three types of road in the ‘other’ category, which includes rear-end crashes.

The closer the road category approaches Safe System standards, the fewer opportunities drivers have to make mistakes, hence the lower the crash rate. Where the traffic lanes are separated, as in the case of motorways and dual carriageway roads, head-on crashes are almost eliminated, despite being the second most frequent type of crash on single carriageway roads. Crashes involving vulnerable road users are also very infrequent on motorways, which exclude pedestrians and cyclists. Crashes at junctions are the most frequent type of crash on both dual and single carriageways, and the second most frequent category on motorways despite the fact that the design of motorways ensures that the movement of traffic onto and off the motorway is by means of well-signposted and delineated ramps. The interaction between traffic on conflicting paths is a very demanding situation, where the many opportunities for driver error lead to a relatively high crash rate. Motorways and dual carriageways provide no specific protection from rear-end crashes, other than perhaps additional lanes to manoeuvre into.

The next three sections provide examples of where separation has resulted in the type of crash reductions on which Safe System thinking is based.

Dividing roads

Between 2002 and 2008, some 1800 km of ‘collision-free’ road was opened to traffic on the Swedish road network, created by providing a wire rope barrier to separate the traffic flows (Carlsson, 2009). Nearly all of this road length had a 2+1 configuration, i.e. lengths with two lanes in direction A and one lane in direction B, alternating with one lane in direction A and two lanes in direction B. These roads are not truly collision-free, as collisions with entering vehicles, vehicles in front, and roadside objects (including wire rope barriers) are still possible and do occur. Nevertheless, the results are impressive. Fatal and serious injuries were monitored over approximately 1300 km of the network and showed a reduction in fatalities and serious injuries of approximately 57%. Fatalities and serious injuries involving motorcyclists fell by 40–50%, and fatalities and serious injuries involving vulnerable road users (pedestrians and pedal cyclists, as well as motorcyclists) fell by 90%. The fatality rate for the 2+1 roads was equivalent to the fatality rate on roads built to full motorway standards. Further information on this approach is provided in the Case Study in Project-level and Network-level Approaches.

To date, claims for the effectiveness of flexible barriers are based on experience in HICs. There is every reason to expect, subject to correct installation, that they would be as effective in reducing vehicle to vehicle and run-off-road crashes in LMICS, and that they would have a role in both discouraging pedestrians from entering the carriageway and protecting pedestrians near the edge of the carriageway.

Although travel along divided roads is relatively safe, dividing a road can also divide the communities that live alongside it. Heavy traffic and high speeds, combined with a dividing barrier, makes it much more difficult for people or non-motorised vehicles to cross from one side of the road to the other. If a new road is driven through an existing settlement, or if an existing road through a settlement is divided, the social and business connections between the communities on either side of the road will be greatly reduced. Any individuals who continue to cross the road will do so at great risk from high speed traffic in an environment where drivers are not expecting to encounter pedestrians, unless satisfactory provision is made for them.

It is best to avoid this situation arising if possible by planning the route to circumvent affected communities rather than going through them. However, if this is not possible then provision should be made for footbridges or other forms of grade separation to allow pedestrians to cross. Other supporting treatments such as pedestrian railings may have to be used to encourage pedestrians to use the bridges.

Provision of footpaths

According to the US Federal Highway Administration (FHWA, 2010), the provision of footpaths (sidewalks) along both sides of a road reduces pedestrian ‘walk along the road’ crashes by 88%; even the provision of sealed shoulders wider than 1.2 metres (4 feet) reduces this type of crash by 71%. While footpaths are effective in reducing the target crash type, note that this type of crash is relatively unusual in developed HICs, with crashes involving crossing the road rather than walking along it. Providing a raised or otherwise separated footpath comes close to achieving a Safe System outcome; this is especially important in LMICs where drivers and riders frequently use the hard shoulder as an additional travel lane

Provision of motorcycle lanes

Malaysian experience, summarised in APEC (2011), has shown that dedicated lanes that separate motorcycles from other vehicles are highly effective in reducing motorcycle crashes. An early study (Radin Umar et al., 1995) found that the installation of a motorcycle lane along a section of major highway reduced crashes by 34%. A larger subsequent evaluation on the same route found a reduction of 39% in motorcycle crashes (Radin Umar et al., 2000). This study found that motorcycle lanes were of greatest benefit where the traffic volume was greater than 15,000 vehicles a day and the proportion of motorcycles in the traffic was between 20% and 30%.

The first Malaysian motorcycle lane designs were based on bicycle lane designs (Tung et al., 2008). This included the use of guardrails that were designed to protect low-speed cyclists from motorised vehicles that had left the roadway. However, the guardrails that were designed to protect cyclists from errant motorised vehicles were associated with almost 25% of motorcyclists’ fatal collisions with roadside objects and therefore increased the risk of serious injury for this road user group. This suggests that guardrail design for motorcycle lanes needs to be considered carefully.

Bicycle lane width standards were also found to be inadequate for motorcycle lanes. Hussain et al. (2005) observed the separation distances preferred by motorcyclists and concluded that the required operating width for a motorcycle was about 1.3 m, and that they were unlikely to overtake each other where the lane width was less than 1.7 m.

© ARRB Group

Speed Management

Management of traffic speeds by manipulating the road infrastructure has evolved considerably in recent years, but it has not yet been widely applied on the road network. For some time to come, speed limit enforcement is therefore likely to be vitally important for containing the amount of travel that occurs at unsafe speeds.

Road engineers currently have the following suite of techniques available for containing speeds to a desired maximum.

Traffic calming

Traffic calming is the general term given to engineering techniques for encouraging lower speeds, and now includes a variety of well-documented treatments. The Institute of Transportation Engineers has a website that provides a comprehensive overview of traffic calming measures (ITE, 2013). Fact sheets are available for some of the most widely used types of devices, i.e.:

  • devices that rely on vertical displacement – humps, flat-top platforms (speed tables), and raised platforms at intersections;
  • devices that rely on lateral displacement – chicanes and mini-roundabouts (neighbourhood traffic circles);
  • devices that rely on road narrowing – choke points and central islands;
  • street closures.

The website also has links to areas dealing with other types of treatments such as curb extensions, refuge islands, raised crosswalks and rumble strips; and links to topics such as speed reduction, accident frequency reduction, and reductions in cut-through traffic movements. A report from the UK’s Department for Transport provides a comprehensive summary of research on traffic calming (Department for Transport, 2007).

At its best, traffic calming is skilfully integrated with urban design through the use of street design, management of parking, landscaping and planting to create an environment where it is obviously a low-speed environment intended to accommodate pedestrians and cyclists and where the appropriate behaviour is obvious to all road users, including choice of the appropriate speed for drivers and riders.

Traffic calming principles have been widely applied in the establishment of low-speed zones in residential areas. This concept originally emerged in the Netherlands as the ‘woonerf’ or ‘living zone’; it has since been adopted in a number of countries under various forms. The key success factors are that the road network must carry low traffic volumes, the completed scheme must not be able to be traversed quickly, and the appearance of the streets should be changed.

In the UK, the first 20 mph zones produced substantial changes in speeds and crashes. A review found that the average speed reduction was approximately 14 km/h, the number of crashes fell by 60%, the number of crashes involving children fell by 70%, and the number of crashes involving cyclists fell by 29% (Webster & Mackie, 1996). Traffic flow within the zones fell by an average of 27%, and increased on the surrounding network by 12%. Despite this shift in traffic, there was little increase in crashes on these surrounding roads.

An extreme example of traffic calming is the shared space approach, referred to as ‘naked streets’ by sections of the press, pioneered by Hans Mordermann in the Netherlands. In streets and other public spaces where it is appropriate to give priority to pedestrians, removing all (or almost all) of the traffic signs and markings takes away the cues that this is an environment for motorised traffic and that drivers and riders should behave accordingly. Instead, it is made clear that this is predominantly an environment for pedestrians where drivers are expected to plan their vehicle movements to avoid inconveniencing pedestrians, and where a degree of communication between road users is necessary in deciding priority for vehicles or pedestrians. However, a comprehensive review of the available data was unable to conclude whether or not the shared space did in fact lead to crash reductions (Edquist & Corben, 2012). Although some of the sites studied did result in crash reductions, others showed increases, including sites where there had previously been no crashes. Poor design of the studies examined, and failure to take account of possible increased pedestrian activity, also made it difficult to arrive at a conclusion.

Thoughtful supporters of the shared space or ‘naked streets’ concept are careful to point out that it is only suited to those environments where it is appropriate to give priority to pedestrians, such as residential areas, open public spaces, and roads that cater for large numbers of pedestrian movements (e.g. shopping areas or cultural precincts). The functioning of society and the economy requires that traffic is given priority on most other roads, and consequently, a full range of signs, signals, markings and other devices are needed to communicate required behaviours to road users.

Combinations of treatments such as pavement markings, road narrowing and signage have been used effectively to reduce the speeds through settlements. A report from the UK’s DETR provides a comprehensive summary of research on traffic calming, including gateway treatments (Figure 8.4). Low-level gateway treatments were found to reduce speeds by less than 5 km/h; more substantial treatments by up to 11 km/h; and the most substantial treatments, which involved narrowing of the carriageway, by up to 16 km/h (DETR, 2007).

Makwasha and Turner (2013) found speed reductions associated with gateway treatments in New Zealand. Consistent with previous research, they found that speed reductions were greater at pinch point gateways where the roadway had been narrowed compared to gateways that consisted of signage alone. Consistent with speed reduction data, there was a 41% reduction in fatal and serious injuries at pinch point gateways, but small increases in crashes at the ‘sign only’ gateways. This agrees closely with earlier work by Taylor and Wheeler (2000) in the UK, which found a 43% reduction in fatal and serious injury crashes for gateway treatments alone, but reductions of 70% in these crashes when accompanied by downstream traffic calming treatments.

Figure 8.4 Threshold treatment with identity feature, median island, and vehicle activated speed feedback sign - Source: Dr Peter Cairney.

A detailed hypothetical example of how a gateway treatment might be provided is discussed in the PIARC HFPSP guide.

Gateways are also used at the entry to lower-speed zones within urban areas. However, as speeds are generally low at these points anyway, their effectiveness can be hard to evaluate (DETR, 2007).

Road Hierarchies and Self-explaining Roads

As mentioned in previous sections, road and traffic engineers have an effective array of techniques and devices available for communicating with the road user. Compliance on the road system is likely to be greatly improved if these techniques can be used in conjunction with basic road design to create road environments that give consistent messages to road users about the type of road they are, the function the road serves and, by inference, the type of driving behaviour and speed choice that is appropriate. This concept is generally referred to as a ‘road hierarchy’.

‘Self-explaining roads’ take this process further by designing the road system and its immediate surroundings to make the required driving actions obvious to the driver.

The European Commission website (EC, no date) describes self-explaining roads in the following terms:

The aim is that different classes of roads should be distinctive, and within in each class, features such as width of carriageway, road markings, signing, and use of street lighting would be consistent throughout the route. Drivers would perceive the type of road and "instinctively" know how to behave. The environment effectively provides a "label" for the particular type of road and there would thus be less need for separate traffic control devices such as additional traffic signs to regulate traffic behaviour.

Roads have different functions which require different traffic speeds and other behaviours, e.g. readiness to deal with cyclists and pedestrians (including young children). If these functions can be made explicit by the design and features of the road, it should be much easier to encourage drivers to behave appropriately. A road that is truly self-explaining would make other aspects of driver behaviour obvious, such as which traffic stream should give way to another, when the driver is approaching an intersection or a curve, where pedestrians are likely to cross the road, and where the driver should position the vehicle to make a turn across a traffic stream. A self explaining road would require few signs or line markings as the required driving actions would be conveyed intuitively by the way the road looks.

In the Netherlands, where the concept originated, four categories of road seem to be sufficient to cater for all needs (Theeuwes & Godthelp, 1995); these are: motorway, major inter-city roads, rural roads to connect residential areas to shopping and services, and woonerfs (or traffic-calmed residential zones). Other countries may find that they need more categories to cover their full range of road types (e.g. rural access roads, urban collector roads). The important point is that roads can be designed to create different expectations about how road users should act on them.

A recent application of self-explaining roads principles in a suburb of Auckland, New Zealand demonstrates how appropriate design – in this case retrofitting an area with planting and other low cost measures – can influence behaviour. After implementation, average speeds were lower on local streets but unchanged on collector roads. In both cases, the variability of speeds was reduced after implementation (Charlton et al., 2010). On local roads, vehicle numbers were reduced, vehicle lane keeping was less consistent, and signalling was less frequent. Also, pedestrian numbers increased, and pedestrians were less constrained in their movement patterns; however, these changes were not evident on the collector roads (Mackie et al., 2013). The authors interpreted these changes as indicating that a more relaxed, informal environment had been created on the local streets, consistent with the objectives of the project. These changed behaviours were accompanied by a 30% drop in crashes and an 86% reduction in crash costs.

The implications of self-explaining roads are especially profound for LMICs. The evidence is that drivers pick up powerful messages about the appropriate way to drive from the cues in the environment. Developments affecting parts of the road system that have customarily been used for social or commercial purposes should therefore be handled with particular care. If it is possible to retain the social or commercial function, then care should be taken to separate through traffic movements from the mixed activity area and ensure that a high-speed environment is not imposed on it. If it is not possible to retain the social and commercial functions, then a suitable alternative site for these activities should be found, and the new road facility which replaces the former mixed activity area should be clearly identifiable as primarily a traffic facility.

© ARRB Group

Road-based fatigue Countermeasures

There is a clear distinction between fatigue, which occurs with time spent on a task, and drowsiness which varies according to the time of day and how much sleep a person has had. The terms are often used interchangeably as they often occur together and have similar debilitating effects on driving. A recent review for the UK Department for the Environment, Transport and the Regions (Jackson et al., 2011) concluded that fatigue affects driving skills in three ways:

  • It increases the frequency of errors (e.g. the number of times a driver intrudes on a neighbouring lane).
  • It increases the size of errors (e.g. the distance that a driver intrudes on a neighbouring lane).
  • It increases the variability of errors (e.g. reduction in the amount of time a driver drives in the centre of the lane).

The most effective ways of managing fatigue for professional drivers appear to be through workplace fatigue management programmes, supported by programmes to ensure that drivers come to work well-rested by addressing lifestyle issues. However, there would also appear to be a role for infrastructure in countering the effects of fatigue.

Roberts and Turner (2008) identified specific areas where infrastructure-based countermeasures might be effective. These include:

Rest areas

Opportunities to rest are likely to be beneficial. It is well-established that short periods of sleep can restore the performance of fatigued drivers. However, there is uncertainty about the location of these facilities in relation to high-risk sections of road, and about the best type of facilities to be provided at different locations.

Monotony reduction treatments

Monotony reduction was thought to be worthwhile, but there was uncertainty about what type of monotony reduction would be effective. Note that the PIARC HFPSP guide suggests the creation of ‘sinuous, rhythmic road alignments’ (PIARC, 2015, p.37 – i.e. gently winding roads) to counter monotony by providing a constantly changing visual field, and suggests that monotonous vegetation and monotonous built environments be avoided.

Signage and road markings

Signs and road markings drawing attention to areas of high risk of fatigue crashes and advising of opportunities to rest at towns or rest areas were thought to have potential.

Audible line markings

Audible line markings are raised thermoplastic lines which create a whirring sound when driven over, alerting the driver that the vehicle is drifting onto the shoulder (when applied as an edge line) or across the centre of the road into the opposing lane (when applied as a centre line). These have proved to be highly effective in reducing crashes, but generally do not generate a sufficiently loud signal to be effective for trucks. In countries where there is widespread use of asphalt construction on rural roads, an equivalent treatment can be produced at lower cost by creating depressions in the asphalt by means of a special roller, or by milling grooves in the road surface. This is not possible when the road is of sprayed seal construction , which is typical of many roads in LMICs and HICs with low population densities. Audible line markings can also be applied to concrete roads, either by the application of thermoplastic markings or by milling grooves in the road surface.

Barriers and clear zones

If these other measures fail to prevent a fatigue-related incident, then barriers and/or clear zones at appropriate points have the potential to avoid serious injury.

Reference sources

No reference sources found.