In Europe, tens of thousands of people are killed in road traffic accidents, and more than 1 million are injured each year at a cost, which is estimated to exceed the total European Union budget by a factor of two. The majority of accidents involving injury occur within urban areas often at junctions, while the number of fatalities outside these areas is greater, largely as a result of higher speed. Traffic safety research has shown a biased interest in the problems associated with motorway and rural areas in the past.
There are many reasons, which advocate a greater interest in urban areas, in particular, those related to the safety of unprotected road users. In urban areas the traffic system context is more complex, where a mixed road user environment prevails and greater perceptual and cognitive demands are placed on road users. In the past, many of the more successful safety countermeasures have focused on designing the roadway to meet the needs and limitations of road users. These solutions have, however, proved to be very costly.
Today, new and relatively cheap technological solutions referred to as Intelligent Transport Systems (ITS) have been developed which have the capacity to reduce exposure, accident risk, and accident severity. While the long term effects of these systems are largely unknown, and problems associated with standardisation and legislation are in need of resolve, systems such as Intelligent Speed Adaptation and advanced traffic control systems have shown great potential with regard to the traffic safety problem in urban areas.
In order to effectuate this potential, a great deal of integrated multi-disciplinary research is required. 1. INTRODUCTION The last century of the second millennium has seen vast improvements in the living conditions and economic wealth of the industrialised nations of the world, and consequently a large growth in population, particularly in urban areas. The growth of cities and towns has also led to an increase in the need for mobility, and a consequent increase in the numbers and types of vehicles occupying the road infrastructure.
The exponential increase in the number of vehicles during the twentieth century has far outweighed the projected capacities and adaptive capabilities of the existing road infrastructure systems, this has resulted in a situation of congestion and frustration among road users of all types and has had significant detrimental effects for traffic safety in terms of the unacceptable numbers of road accidents involving fatality and injury. During 1997, there were approximately 45 000 fatalities, and 1. million injuries reported from road traffic accidents within the European OECD countries according to statistics taken from the International Road Traffic and Accident Database (IRTAD). The costs of such accidents within the European Union are estimated to be in the region of 160 billion ECU per year, thereby exceding the total annual budget for the EU in 1997 (89 billion ECU). Statistics indicate that while approximately two-thirds of all fatalities occur outside urban areas, two-thirds of the total number of injuries occur inside urban areas.
The outcomes of urban accidents are usually less severe in terms of the numbers of injuries and fatalities as a direct result of the greater limitations imposed on speed. Mainly for this reason, most of the international and national traffic safety research has focused on motorways and major roads that link towns and cities by traversing rural and suburban areas. 2 Why Focus on Traffic Problems in Urban Areas?
There are however, a number of very important reasons why the traffic system existing in urban areas should not be disregarded by prevailing research. Most importantly, there are a great number of people living in urban areas, and travel distances are generally shorter than in rural areas, which encourages people to use bicycles or to walk. According to the statistics for the European OECD countries, pedestrians amounted to 15 per cent, and cyclists 6 to per cent, of the total number of roadtraffic fatalities during 1997, respectively.
For other types of road users, such as vehicle drivers, there are significant differences in the types of accidents that occur in urban areas with a greater number of rear-end and turning collisions, and a larger percentage of collisions occurring at junctions. Generally, the urban environment can be regarded as more complex with many different types of road users with different needs, the physical and mental demands placed on road users are therefore much higher, and are reflected in the statistics by a comparatively greater number of accidents involving injury.
The Development of Traffic Safety in Urban Areas from a Historical Perspective In the past, European traffic safety in urban areas has most typically focused on speed management and traffic calming. Kjemstrup and Herrstedt (1992), identified three specific periods of development. During the first period prior to 1968, the major growth in the number of vehicles was initially countered by expanding the infrastructure where possible.
These measures soon proved insufficient, leading to the hazardous overuse of local roads that had previously catered for lighter traffic, cyclists and pedestrians. In the second period from 1969 to 1979, the traffic safety problem was approached by separating light and fast-moving heavy traffic, and by designing the traffic environment with an emphasis on uniformity and simplicity. “Traffic calming” initiatives became popular together with what became known as “Woonerf design”, which entailed physical speed-reducing measures (e. . humps and narrowings), and rules to govern speeds and priorities within urban areas. The third period from 1980 to 1990 saw increasing opposition to speed reducing measures by public and private interests. A new and less expensive solution was required, and eventually took the form of “environmentally adapted through-roads”, which took into consideration: flow-rates, composition of traffic, accident rates, pedestrian needs, and environmental sensitivity (in relation to noise levels air-pollution etc. ).
Research has shown considerable success in a number of European countries (e. g. Denmark, France, Germany), while others have neglected the potential offered by this concept. Kjemstrup and Herrstedt (1992) concluded at the time of writing, that automatic speed regulation devices were “a long way into the future” and that traffic calming through physical measures and the design of street space was the only currently available approach to achieving lower speeds and increased safety and security and an improved urban environment.
Despite safety countermeasures such as traffic calming techniques, various legislative measures, publicity campaigns, active police enforcement, improvements in vehicle safety standards, and local improvements in the infrastructure, the problems related to traffic safety in urban areas still prevail at an unacceptably high level. One area of research that has shown great promise is that concerning Intelligent Transport Systems (ITS) which exploit the latest information and communication technologies and integrate them into the traffic system.
This approach enables the already congested infrastructure to be utilised with much greater efficiency, and to some extent can resolve problems associated with human limitations. The future appears to have arrived for automatic speed regulation devices such as “Intelligent Speed Adaptation” (ISA), which are perceived to have great safety potential (see e. g. SNRA, 1996). There is however, an irrefutable need for establishing sound scientific knowledge related to how effective countermeasures, with or without the aid of ITS, can be designed, developed, and implemented to resolve urban area traffic safety problems.
The Need for Traffic Safety Strategies The World Health Organisation has referred to the existing traffic safety as a social and public health problem. In order to approach this problem, different European and world-wide traffic safety research programs have been initiated, which call for international co-operation between different institutions and establishments, and a conglomeration of different scientific disciplines in a major offensive to find ways to improve the present situation.
At a national level some countries have developed specific traffic safety strategies in order to approach the traffic safety problem. 3 In Sweden, the “vision zero” has been sanctioned by the government in order that individuals and organisations can develop an operational energy and innovative thinking towards a new and radical future situation where the desired ultimate goal is to have zero road-traffic fatalities (Tingvall, 1995). This is to be achieved by emphasising for individuals that loss of health is completely unacceptable, and that traffic safety is ultimately their responsibility.
The focus of attention is placed on active safety (i. e. the prevention of accidents), passive safety (i. e. reducing the severity of accidents upon occurrence), and improving the rescue services, health-care, and rehabilitation. Importantly, it is recognised that there are other interests in society in addition to the provision of safety (e. g. environmental issues and regional planning issues) which must be discussed in order to find a balance by authorities other than those in the road-transport sector in a social and political context.
Thus, the “vision zero” concept involves a wide spectrum of interest at many different levels in order to provide an integrated and far reaching approach to the traffic safety problem in Sweden. Delimiting and Defining Urban Areas for Traffic Safety Research In this report it is important to distinguish between those roads that can be regarded as part of an urban area road network, and those which are not. A reasonably straightforward definition of urban areas could be taken to include all types of roads provided that they fall within a town or city boundary.
This distinction is, however, too general to be of value for research purposes. The definition that is adopted by the Swedish Institute for Transport and Communications Analysis ( SIKA), and which will be used as a de facto definition in this report is: • roads which most often are directly adjacent to large numbers of buildings where people live and work • roads where there are many different types of road users (including cyclist and pedestrians) allowed to use the road • roads where there is a high density of road junctions, roundabouts, pedestrian crossings etc. n order to allow for a reasonable level of accessibility for all road users • roads where the maximum allowed speed is no greater than 50 km/h, or where a higher speed limit is posted, but the density of the surrounding buildings and the traffic conditions resemble those described above Roads that lead through built-up areas, but do not have intersections and are separated from the environment (by a fence) or restricted to motor vehicles are not considered to lie in an urban area (SIKA, 1999).
The Swedish National Road Authority makes a distinction between “central”, “intermediate”, and “suburban” roads (SNRA, 1989, p. 55). What are considered “urban” or “built-up” areas differs to some extent from country to country depending on the responsibilities and prevailing political climate of the local and central authorities with regard to different road categories, and environmental and traffic safety issues. The exact distinction remains therefore somewhat “fuzzy” at an international level. 2. SWEDISH AND INTERNATIONAL T RAFFIC S AFETY S TATISTICS : THE D IFFERENCES BETWEEN URBAN AND NON-URBAN AREAS The traffic safety situation in Sweden is described annually in an official publication entitled “Traffic Injuries” produced by the Swedish Institute for Transport and Communication Analysis (SIKA). This publication is based entirely on police reported accident data, and represents only those accidents, which involve personal injury.
The problem of under-reporting is acknowledged, and it is believed that the statistics presented might represent only as much as 30 per cent of the lighter injuries and approximately 60 per cent of the more serious injuries. Most injuries in relation to the different road user groups are believed to be represented to a level of approximately 50 per cent (e. g. caroccupants, motorcyclists, pedestrians). The most notable exception is found in the statistics for cyclists which are only believed to be represented to a level of 15 per cent (SIKA, 1999). Basic Swedish statistics and trends for the period 1960-1997 With 8. 5 million inhabitants, Sweden has one of the smallest populations among the European OECD nations. The infrastructure consists of a 210 000 km road network that is populated by around 4. 44 million motor vehicles. During the period between 1960 and 1997, the total number of accidents and the number of urban accidents has declined marginally (see Figure 1). The total number of fatalities and the number of fatalities in urban areas has, however, declined significantly (by more than 50 per cent since the mid-1960’s) during this time period (see Figure 2).
During the period from 1960 to 1997 the proportion of accidents involving fatality in urban areas has remained relatively constant, at an average of 32 per cent of the total number. Similarly, the proportion of accidents involving injury in urban areas has remained fairly constant during this time period at around 58 per cent of the total number. 20000 18000 1200 16000 14000 12000 1000 1400 No of Accid. 10000 8000 6000 4000 2000 0 1960 Accidents Urban Areas Accidents in total No of fatalities 800 600 400 Fatalities In total Fatalities urban areas 1970 1975 1980 1985 1990 1995 200 0 1965 1970 1975 1980 1985 990 1995 1960 1965 Year Year Figure 1: Absolute number of police reported accidents on Swedish roads that resulted in injuries or fatalities during the period 1960 to 1997, and absolute number for the same types of accidents in urban areas during the same period. Figure 2: Absolute number of police reported fatalities on Swedish roads during the period 1960 to 1997, and absolute number for the same type of accidents in urban areas during the same period. Comparing Swedish Urban and Non-Urban Statistics In 1997, 9 015 of the 15 752 reported road accidents involving fatality or injury occurred in urban areas.
These urban area accidents resulted in 154 fatalities, 1 765 serious injuries, and 11 593 less serious injuries. The remaining 6 737 accidents occurring outside urban areas resulted in 387 fatalities, 2 152 serious injuries, and 7 689 less serious injuries. The majority of the accidents that are reported to the police authorities involve one or more motor vehicles. The main classes of accidents are single vehicle accidents, vehicle-vehicle accidents, and accidents where vehicles and unprotected road users are involved (i. e. oped riders, cyclists, and pedestrians). 5 Single vehicle accidents usually occur in situations where drivers lose control over their vehicle, often as a result of driving too fast for the situation and prevailing road conditions. Many of these accidents occur at night-time and involve collisions with parked vehicles. With regard to vehiclevehicle accidents, a more in-depth look at the nature of the accident called for is in order to get a clearer picture of the different accident profiles occurring in Swedish urban and non-urban areas.
The main types of vehicle-vehicle accidents described by SIKA are: • Overtaking and Lane-Changing – Accidents where the two primarily involved vehicles are on the same road travelling in the same direction without any planned turn and where one of the vehicle drivers intended to either change lanes or overtake. • Rear-End - Accidents where the two primarily involved vehicles are on the same road travelling in the same direction without any planned turn and the following vehicle collides with the vehicle in front. No intended overtaking should be present in the situation. Meeting – Accidents in which both of the primarily involved vehicles collided while travelling in opposite directions with no planned turn. • Turning – Accidents where the two primarily involved vehicles are originally travelling on the same road in the same or opposite direction, and where one or both of the vehicles has planned to turn. • Crossing – Accidents where the two primarily involved vehicles are originally travelling on different roads, and where neither of the vehicles has planned to turn in a situation where the vehicles’ planned routes cross each other, or where one or both vehicles planned to turn. Others – Accidents that do not fit any of the above descriptions (e. g. reversing accidents and U-turn accidents). The differences in the number of accident occurrences between urban and non-urban areas are depicted in Figure 3. The figure indicates that almost all accident types are more frequent in urban areas, with the exception of single-vehicle accidents, accidents involving overtaking or lane changing, and meeting accidents.
Single vehicle accidents occur more often outside urban areas often as a result of loss of control at speeds, which are too high for the prevailing road conditions. Overtaking tends to occur less frequently within urban areas where a speed limit of 50 km/h or less is posted, and generally congested conditions restrict the opportunity for a manoeuvre of this type. Lane changing on the other hand occurs quite frequently within urban areas, but does not result in the same number of accidents most probably on account of lower speeds.
The number of rear-end accidents is greater within urban areas than in rural areas, there is a considerably larger number of situations per km of roadway which require vehicles to stop or yield. Similarly, there is a greater number of opportunities for turning and crossing accidents within urban areas due to higher levels of congestion and the higher number of traffic junctions. The most noticeable differences between urban and non-urban area accidents that are depicted in Figure 3, are those concerning motor vehicles and unprotected road users (i. . mopeds, cycles, and pedestrians). There is a comparatively greater number of unprotected road users in urban areas, but the absolute number of these accidents in comparison to single vehicle and vehicle-vehicle accidents suggest that this is one area that should be given great attention in the field of traffic safety research. The “others” category shown in Figure 3, which is not vehicle-vehicle related refers to accidents involving a vehicle and another type of obstacle than those already mentioned.
In rural areas and on motorways (and occasionally with urban areas) accidents occur that involve motor vehicles and wild animals that have strayed onto the road, larger animals such as moose can cause severe injury and sometimes fatality. Other types of motor vehicle accidents resulting in injury can also involve other forms of transportation that occupy or cross the roadway (e. g. trains, trams, tractors and other more unusual vehicles). Accidents that result in injury also occur between forms of transportation that do not involve the more common motor vehicles category, and in some cases pedestrians (e. . between cycles, mopeds, and/or pedestrians). 6 The fatality statistics reveal some interesting facts (see Figure 4). Most importantly, all types of vehicle-vehicle accidents result in a greater number of fatalities outside urban areas, despite the fact that some types occur more frequently within urban areas. This fact can be attributed largely to the higher speeds allowed on non-urban roads, and is exaggerated further by the number of fatalities in meeting accidents in non-urban areas where the collision speed is the sum of the two vehicles that are approaching in opposite directions.
The unprotected road users category is that which is of great concern to traffic researcher with regard to urban areas. The proportion of accidents involving motor vehicles and pedestrians or cyclists that result in fatality are considerably greater in non-urban areas as a result of the higher speed of the vehicles, but the absolute number of those accidents is considerably smaller outside urban areas. The traffic safety problems associated with unprotected road users are quite prominent.
The fatality risk independent of the type of area where the fatality occurred for pedestrians in relation to vehicle occupants given the relative number of travelled kilometres is 11 to 1, for cyclists 4. 5 to 1, and for moped riders 24 to 1. 3000 No of Accidents 2500 2000 1500 1000 500 Veh-Veh Crossing Veh-Veh Meeting Veh-Veh Turning Veh-Veh Rear-E Veh-Veh Others Veh-Pedestr Veh-Single Veh-Moped Veh-Veh LCh/Ov Veh-Cycle Others Veh-Pedestr Others Urban Accidents Non-urban Accidents 0 Accident Category
Figure 3: The number of urban and non-urban accidents for different road user accidents categories (LCh/Ov = Lane Changing and Overtaking, Rear-E = Rear-End, Pedestr = Pedestrian). 140 120 No of Fatalities 100 80 60 40 20 Veh-Single Veh-Veh Crossing Veh-Veh Meeting Veh-Veh Turning Veh-Veh Rear-E Veh-Veh Others Veh-Moped Veh-Veh LCh/Ov Veh-Cycle Urban fatalities Non-urban Fatalities 0 Accident Category Figure 4: The number of urban and non-urban fatalities for different road user accidents categories (LCh/Ov = Lane Changing and Overtaking, Rear-E = Rear-End, Pedestr = Pedestrian). During 1997 almost 200 more accidents occurred per month within urban areas rather than outside. In the urban environment, more accidents than average occurred from May to October, with a local minimum in July. July is also the month in which the majority of accidents outside urban areas occur. This can be explained by the fact that a large percentage of the Swedish population take their holidays in July, resulting in a lower number of road users within urban areas and a greater number on rural roads and motorways.
A look at the distribution of accidents occurring within urban areas according to the days of the week also shows a different pattern when compared to accidents occurring in nonurban areas. In urban areas, the number of accidents is above average on workdays, rising marginally towards Friday. A sharp decline in the number of accidents can be noticed during weekends. For times of day, the patterns are similar for accidents occurring within and outside urban areas. The main difference lies in the larger quantity of urban accidents. Most accidents occur in the afternoon between 16:00 and 18:00.
Few accidents occur during early morning hours before 6:00. Comparing Swedish and International Urban and Non-Urban Statistics In the 1999 summer edition of the European Transport Safety Council (ETSC) newsletter “Safety Monitor” it was stated that the fatality rates expressed in 100 million person/kilometres travelled were significantly greater for cyclists (6. 3), pedestrians (7. 5), and moped/motorcyclists (16) than all other modes of travel. This situation remained unchanged even when the fatality rates were expressed in terms of 100 million hours of relative exposure.
The ETSC also pointed out that there is a considerable difference (by a factor of 7) between the country with the most adverse fatality rate and the country with the least number of fatalities, and recommended that priority should be given to the traffic safety problems associated with unprotected road users. The European fatality statistics suggest that car travel is ten times safer than walking (although car travel is in itself ten times less safe than bus travel). The majority of international statistics in this report are extracted from the OECD IRTAD database.
This data is presented with due consideration to the problems of under-reporting and a general lack of exposure data. Currently twenty European nations (Austria, Belgium, Czech Republic, Denmark, Finland, France, Germany, Greece, Hungary, Iceland, Ireland, Italy, Luxembourg, Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, and United Kingdom) report data to IRTAD each year. Some of the more standard detailed data is, however, not available for some countries (most notably Greece, Luxembourg, and Portugal).
Besides these European countries there are data for Canada, USA, Australia, New Zealand, Japan, and South Korea. Table 1 indicates some of the basic demographic and traffic safety data for Europe (with Sweden shown separately), and other IRTAD countries for 1997. Table 1: Basic demographic and traffic safety data for the OECD countries who report their statistics to IRTAD. The OECD European countries are shown as a grand total (OECD Eur Total), and as an mean average (OECD Eur Average), the data for Sweden is also shown separately. Country/Continent Population (x 1000) Road Network (km) Land Area (sqkm)
Motor Vehicles (x 1000) Reported Accidents involving Injury 15 752 72 393 1 303 074 2 455 118 152 765 -9 482 780 399 246 452 Reported Accidents involving Injury in Urban Areas 9 015 50 015 850 258 1 564 411 107 019 -5 944 566 169 162 320 Reported Fatalities Reported Fatalities in Urban Areas Sweden OECD Eur Average OECD Eur Total USA Canada Australia New Zealand Japan South Korea 8 844 20 281 405 610 267 636 30 286 18 532 3 743 126 166 45 545 210 000 183 568 3 304 227 6 346 857 895 562 810 000 91 864 1 152 207 82 342 449 760 194 134 3 882 684 9 363 353 9 360 527 7 686 844 269 122 377 837 99 266 428 10 767 215 347 203 568 17 576 11 238 2 393 75 713 11 991 541 2 485 44 728 41 967 3 064 1 767 540 11 254 13 343 154 790 13 427 14 861 1 043 -146 5 512 5 378 The statistics presented in Table 1 show that Sweden is smaller in terms of population than the average OECD European country, yet larger in terms of the land area and the size of the road network. The number of accidents involving injury and the number of fatalities (regardless of whether they occurred within or outside urban areas) are also considerably smaller than the average. 8
An interesting comparison can be made between OECD Europe and the USA. The population of OECD Europe is approximately one-third larger than USA and the number of motor vehicles is only 5. 5 per cent higher. However, the road network in the USA is 48 per cent larger in terms of the number of kilometres, and the total number of reported accidents involving injuries is 47 per cent higher than in OECD Europe. The total number of fatalities for the USA is only 6 per cent smaller than for the OECD European countries despite the smaller population and larger road network.
The higher number of vehicles relative to the population could partly explain these differences. Before any more solid conclusions can be drawn important exposure data are required. The problem of under-reporting as previously mentioned, and the actual reporting method may also have confounding effects on the data. The rates for all accidents that resulted in injury per 10 000 population, and fatality rates per 100 000 population, for the different OECD countries and continents mentioned above are shown in Figure 5. The number of accidents that resulted in injury was not available for Australia.
The average rates for the OECD European countries were calculated individually on the data that were available (some data was not available for Greece, Luxembourg, and Portugal). The data presented in Figure 5 show that Sweden (and other Scandinavian countries) has a lower fatality rate and accident rate than other OECD European and non-European countries that reported data to the IRTAD database. 100 Fatalities per 100,000 population 90 80 Accidents involving injury per 10,000 population 70 No of Acc/Fat 60 50 40 30 20 10 0 Sweden OECD Europe (Av) USA Canada Australia New Zealand Japan Korea Country/Continent
Figure 5: Accident rates for all accidents that resulted in injury per 10 000 population, and fatality rates per 100 000 population, for the different OECD countries and continents. Looking at the international statistics in relation to urban and non-urban areas it can be noticed that all of the OECD countries, with the exception of Norway, had a higher proportion of accidents involving injury occurring within urban rather than outside. Sweden (57 per cent) came a close second to Norway (47 per cent), with the OECD European average matching very closely that of USA (63 and 64 per cent respectively).
Japan had the highest proportion of accidents involving injury occurring within urban areas (73 per cent). This might well be attributable to the high level of urbanisation in Japan and the resulting high proportion of urban roads. In Sweden, 28 per cent of all road-traffic fatalities occur in urban areas. This figure is lower than the OECD European country average (33 per cent) and lower than most of the countries and continents outside Europe with the exception of New Zealand (27 per cent). Japan has the highest urban area fatality rates (49 per cent) perhaps for the same reasons as mentioned above.
Within OECD Europe, Spain has the lowest recorded fatality rate in urban areas (20 per cent), closely followed by Norway (21 per cent), Austria (23 per cent), and Germany (24 per cent). Unfortunately, the IRTAD database is not so detailed concerning the exact types and involvement of different categories of road users concerning urban area accidents that involve injury and urban area fatalities. With regard to the fact that many accidents in urban areas involve unprotected road users it is worthwhile to take a look at the available IRTAD data concerning these categories despite the fact that 9 he proportions of accidents involving injury and fatalities in urban and non-urban areas are not available. The data indicates that Sweden has one of the lowest overall pedestrian fatality rates per 1 million population, and also a low cyclist fatality rate per 1 million population within (but not outside) the OECD European countries (see Figure 6). Obviously the proportional number of cyclists needs to be taken into account for the different countries in order to allow statements about their relative safety. 120 Fatality Rate (per 1 mill. pop. ) Pedestrian Fatality Rate per 1 million population 00 80 60 40 20 0 Cyclist Fatality Rate per 1 million population Sweden OECD Europe (Av) USA Canada Australia New Japan Zealand Korea Country/Continent Figure 6: Fatality rates for pedestrians and cyclists per 1 million inhabitants for the different OECD countries and continents. A Closer Look at Accidents that Occur at Junctions in Sweden Having identified the fact that the majority of accidents in urban areas occur at junctions it is interesting to look more closely at the nature of these accidents and the differences between existing types of junctions.
Figure 7 below indicates the main differences between road-links (i. e. the roads stretching between junctions), T-junctions, crossroad junctions, roundabouts, and other categories, in relation to the different types of road users involved. Unfortunately, no urban and non-urban breakdown of the statistics is available. 3000 Link 2500 No of Accidents 2000 1500 1000 500 0 Single Veh-Veh Veh-Cycle T-Junction Crossrd-Junc Roundabout Other Veh-Pedestr Accident Types Figure 7: Number of accidents on road-links and different junction types shown in accordance with the different types of road users involved.
Figure 7 shows that the majority of all single accidents occur on link roads. The accident statistics also show that accidents occurring on road-links have a higher frequency of fatalities and more serious injuries than those occurring at junctions. Of the 493 accidents involving fatality in Sweden during 1997, 329 occurred on link roads (i. e. 67 per cent), 69 at T-junctions (14 per cent), 68 at crossroadjunctions (14 per cent), and 4 at roundabouts (1 per cent). It is important to note that roundabouts are relatively uncommon in Sweden, between 3. and 10 times less in number than other types of crossing (Englund et. al, 1998). 10 Before discussing the vehicle-vehicle accidents, it is important to note that the majority of vehiclecycle and also the vehicle-moped accidents occur at T-junctions and crossroad-junctions. A far greater number of accidents occur at these types of junctions than on link roads. For pedestrians an almost equal number of accidents occur on road-links (523) and at T-junctions and crossroad-junctions taken together (498).
There is little difference between the accident and fatality rates for T-junctions and crossroad-junctions, and the data for roundabouts is not comparable on account of the relatively low number existing in the traffic infrastructure. The statistics show a larger number of vehicle-vehicle accidents occurring at T-junctions and crossroad-junctions taken together (3 436) in comparison with the number of accidents on road-links (2 216). It is interesting to examine the vehicle-vehicle accidents in relation to the different categories of accidents that occur (see Figure 8). 1000 No of Accidents 800 600 400 200 0
Veh-Veh Overt / LCh Veh-Veh Rear-End Veh-Veh Meeting Veh-Veh Turning Veh-Veh Crossing Veh-Veh Other Link T-Junction Crossrd-Junc Roundabou t Other Vehicle-Vehicle Accident Types Figure 8: Number of accidents on road-links and different junction types shown in accordance with the different types of vehicle-vehicle accidents (Overt/L Ch = Overtaking and Lane Changing). Figure 8 shows that very few ‘overtaking and lane changing’ accidents and very few ‘meeting’ accidents occur at junctions (these are more likely to fall into the ‘crossing’ and ‘turning’ categories when they occur at junctions).
It can also be noticed that a similar number of ‘rear-end’ accidents occur at T-junctions and crossroad-junctions, but these are far exceeded by the quantity occurring on road-links (this may reflect the way these accidents are coded with regard to distance in relation to a junction). Differences can be noticed at T-junctions and crossroad-junctions regarding the number of ‘turning’ and ‘crossing’ accidents. ‘Turning’ accidents occur with much greater frequency at Tjunctions (690) than at crossroad-junctions (381), whereas ‘crossing’ accidents are more likely to occur at crossroad-junctions (1030) than at T-junctions (481).
These statistics reflect differences in the frequencies of ‘turning’ and ‘crossing’ manoeuvres and possible conflict situations that occur at these two sites. These findings concur with those of the Swedish researcher Brude (1991) who points out that more than half of the accidents in urban areas occur at junctions. He also states that 4-legged junctions are 1. 5 to 2 times more accident-prone than 3-legged junctions, and consequently that redesigning 4-legged to three-legged junctions can enhance safety. 11 3.
THE M AIN CAUSES OF ACCIDENTS IN URBAN AREAS In the majority of traffic accidents, urban or otherwise, the cause cannot usually be traced to a single factor but rather to a combination of circumstances in a chain of events that are best described by the interactions between: a) the road user, b) the vehicle, and c) the roadway and prevailing environment. Any attempts to explain and improve traffic safety in urban areas must ideally adopt a “systems approach” in order to consider the contributions and interactions of all important factors, and to capture the dynamics and complexity of the traffic system (Hyden, 1994; Leveson, 1995).
The concept of traffic safety can ultimately be regarded as an emergent property of the actions and interactions of the main elements of the traffic system. This suggests that any systems analysis aimed at examining existing traffic safety problems must be performed at an appropriate contextual level in order to gain appropriate knowledge and insight into the existing complexities and dynamics of the system and it’s constituent elements at a given time. However, even if a systems approach is adopted, the subject of how and why accidents in urban areas actually occur presents a number of problems for researchers.
The main difficulties appear to lie in establishing the exact chain of events leading up to the accident. This information is of some relevance to the police authorities and insurance companies for the purposes of determining responsibility, but is often neglected with regard to important details that might provide a more in-depth systematic description of what actually occurred. Often, post crash interviews are susceptible to distortion as a result of trauma or fear of the consequences of being assigned responsibility.
Describing the conditions leading up to an accident is also made more difficult as they are often not monitored consciously by the driver who is often performing skills in what is best described as “automated” mode (Wickens, 1992). In the remainder of this section three important and relevant studies from different countries, which focus specifically on the reasons behind urban accidents, are described and discussed, and related to the traffic safety problem in urban areas in Sweden.
The English Study One of the most comprehensive and most interesting studies in relation to the causes of urban area accidents was carried out in the Leeds area in England at the end of the 1980’s ( Carsten, Tight, Southwell & Plows, 1989). During a one year period, Carsten and colleagues investigated 1 254 injury accidents reported to five different police sub-divisions that involved 2 454 participants (1 863 drivers, 463 pedestrians, and 128 cyclists) that occurred in the Leeds area on roads that had a speed limit of 40 mph (approx. 60 km/h) or less.
The accident data showed that 2 per cent of the accidents involved fatalities, and as many as 20 per cent involved serious injury. Nearly 70 per cent of all accidents occurred at junctions, and of these junctions 12 per cent were controlled by traffic-lights or stop signs, and 72 per cent by give-way signs, the remaining 16 per cent were uncontrolled. Besides the official police accident reports, questionnaires were administered by interview or post, and a visit was made to the accident site prior to a case conference that was convened to determine and classify the circumstances surrounding the accident.
An innovative method for describing accident causation was adopted in which four different levels could be combined in accordance with a multi-level coding scheme comprised of 150 different items in order to form what was described as a “chain of factors” approach. According to Carsten and colleagues (1989) a contributory factor behind an accident could be defined as “a road user or traffic systems failure without which the accident would not have happened”, the immediate failure that precipitated the accident.
The results of this study indicated that, of the first level factors (i. e. immediate failures that precipitated an accident), ‘unable to anticipate’ accounted for 29 per cent, ‘failure to yield ’ for 16 per cent, and ‘failure to anticipate ’ 10 per cent of the factors coded. ‘Unable to anticipate’ implies that the road user in question had the right of way, and a “reasonable road user” would also have been unable to anticipate the faulty behaviour of the other person, while ‘failure to yield’ implies that a “reasonable road user” would have perceived the upcoming danger earlier.
Results also show that as many as 44 per cent of drivers were “innocent victims of others’ mistakes”. Failure to yield was also a factor for adult and child pedestrians (66 and 78 per cent, respectively). 12 At the second level (i. e. failure that increased the likelihood of an accident), findings suggested that 62 per cent of the factors determined were situational problems. For the drivers the most important factors were ‘driving too fast for the situation’ (29 per cent) and ‘following too close’ (8 per cent). On the third level (i. e. oad user behaviour or lack of skill that led to failure) different road user skills are considered. It is found that especially pedestrians, and mostly children, “fail to look”, while the main problem of motor vehicle drivers usually is the misinterpretation of other road users’ intentions. Both groups also show “lack of judgement”, which includes a wrong estimation of the speed or the path of other road users. The most common problem on the fourth level (i. e. the explanation for the failure or behaviour) for drivers is their view being obstructed by something, either inside the car or outside. Impairment”, which mainly means the influence of alcohol, is much more common among adult pedestrians than among drivers. Some other points of interest to emerge from this study concerned the fact that 75 per cent of all accidents occurred within the first 5 kilometres of travel, and also that 93 per cent of the respondents had knowledge of the road where the accident occurred. Also, it is noteworthy that as many as 16 per cent of all pedestrians involved in accidents admitted consuming alcohol within a three hour period prior to the accident, compared to 4. percent of drivers. Some important differences were also found regarding different age groups, more specifically, the problems of younger drivers (most notably males) such as driving too fast for the situation differ from those of elderly drivers associated with deteriorating perceptual and cognitive abilities with increasing age. Perhaps one of the most important points concerns the fact that almost 50 per cent of the second most common level 1 factor, ‘ failure to yield ’, could be explained by lower level perceptual and cognitive factors (e. . ‘failure to look ’ and ‘failure to see’). On the basis of the important knowledge gained from their study, the authors were able to make a number of recommendations concerning road user issues for drivers (and riders), and pedestrians, and suggests ways in which traffic management and engineering issues might be brought to bear on traffic safety problems along with counter measures that involve improvements in publicity, education, and training.
The French Study Another major study which looked at some of the reasons behind accidents in urban areas, and the differences between these accidents and those occurring outside urban areas, was conducted by the French traffic researchers Malaterre and Fontaine (1993). This investigation was primarily aimed at identifying the safety needs of drivers, and the possibility of satisfying these needs by using the different Intelligent Transport Systems (ITS) functions suggested in the European PROMETHEUS programme.
The authors identified 17 different basic needs in relation to the driving task: “For each user the accident could have been avoided, if the need had been satisfied”. 3179 accidents involving 6049 road users were examined according to these needs. For approximately 20 % of the road users no needs were identified – they are considered the passive victims of the accident. The needs were grouped into the road user skills of :“control”, “prediction”, “estimates”, “detection”, and “status”. The results of this study show an overriding need for detection in urban areas (mainly at junctions), followed by the need for prediction (i. . predicting the manoeuvre of a road user who has already been detected). Detection problems were advocated in 61 per cent of the accidents analysed, and within this particular grouping of identified needs ‘transversal detection’ problems (i. e. detecting the approach of another road user on an intersecting lane at a junction, or similarly for a pedestrian, detecting the approach of a vehicle on the roadway he/she is preparing to cross) occurred most frequently (19 per cent of all accidents). Outside urban areas detection problems were still predominant over the other types of identified needs (45 per cent of all accidents).
However, the second most important need was identified as statusrelated diagnostic needs (i. e. mainly driver-related problems in relation to alcohol or fatigue, and space-time assessment needs) rather than prediction needs which were found to be somewhat less important in non-urban areas. The authors suggest that a large number of junction-related accidents in urban areas (i. e. meeting and turning accidents) could be avoided by the introduction of suitable Intelligent Transportation systems (ITS) which support driver needs such as ‘critical course determination’, ‘obstacle detection’, ‘vision enhancement’, and ‘safety margin determination’. 3 The American Study The North American researchers Retting, Williams, Preusser, and Weinstein (1995) have conducted another major study focusing specifically on traffic safety problems in urban areas. According to the authors, efforts to reduce the number and severity of accidents have been seriously inhibited by a general lack of important information with regard to the specific types of accidents that predominate in urban areas.
The primary purpose of their study was to develop a classification system based on pre-accident driver/vehicle behaviour in order that planners and policy makers could develop suitable countermeasures aimed at reducing the most prevalent types of urban accidents. The investigation conducted by Retting and colleagues was based on police reported accident data from three cities and one urban county, each from different states. The data were collected between August 1990 and July 1991 and included records from 4 526 accidents, many of these included property damage only in contrast to European police accident reporting.
The results of this study show that 56 per cent of all accidents occurred at intersections, and only 31 per cent of accidents resulted in injury. Pedestrian and cycle accidents were not included in the accident statistics, although it is reported that these would otherwise have accounted for 5 and 3 per cent of the total number of accidents, respectively. Five of the thirteen different types of accidents accounted for 76 per cent of the total number that occurred in the four urban areas, these included: ‘ran traffic control’ (i. . a situation where a vehicle that is required to stop, remain stopped, or yield disregards the requirement and collides with some other vehicle) with 22 per cent, ‘stopped or stopping’ (i. e. a situation where a vehicle that has stopped, or is stopping or just starting up in a travel lane is hit from the rear) with 18 per cent, ‘ran-off road’ (i. e. a situation where a vehicle leaves the travel lane(s) striking an object such as a parked cars) with 14 per cent, ‘lane-change’ (i. e. situation where a vehicle in a travel lane swerves or moves into another in the same direction that is already occupied) with 13 per cent, and ‘left-turn oncoming’ (i. e. a situation where a vehicle in the process of making a left turn in front of oncoming traffic is struck by or strikes a vehicle which is both coming from the opposite direction and which has a superior right of way) with 9 per cent. These five accident types also accounted for 83 per cent of the total number of accidents involving injury.
The rank order among these five accident types differed between areas. Also, the ‘ran traffic control’ and ‘left-turn oncoming’ accidents occurred most frequently at intersections. A closer look at the ‘ran traffic control’ category showed that 41 per cent of the intersections where the accidents occurred were controlled by stop signs, and that 31 per cent were controlled by traffic lights. It was also noticed that the ‘left-turn oncoming’ accidents generally occurred in the presence of traffic lights and were most often the result of a failure to yield to a right-of-way.
The ‘ stopped or stopping’ accidents, involving mostly rear-end collisions, were found to be the result of inattention. It was proposed that the most common types of accidents, described as ’ran traffic control’, could be resolved by better signal timing, the increased visibility of signals and signs, reduced speeds near intersections, red-light cameras, or redesigning the intersection in terms of traffic signals and signs. It is also recognised by the authors that different intersections are likely to have different accident type distributions depending on (e. g. oad geometry, population density, traffic density, methods of traffic control, and police enforcement strategies). Similarly, ‘stopped or stopping’ accidents might be reduced by better signal timing and reducing the number of stops. The ‘ ran-off road’ category of accidents might be reduced by better roadway design including the use of one-way systems, better lane delineation, and better lighting. Suggestions are also made regarding other possible solutions including changes in roadway design to reduce the possibility for conflict between different road users.
Comparison of the Swedish, American, English and French Data It is problematic to compare the four different countries directly because of the many differences in data collection methods, the grouping and inclusion of different road-users, and the many differences in the traffic systems. The most important differences that are noted include the fact that the French study does not consistently distinguish between accidents in rural and in urban areas, and the fact that the American data also includes property damage accidents and excludes other important accident types (e. g. accidents with unprotected road users).
Also, both the English and the French study give only sparse information as to the accident location. 14 One of the main conclusions that can be drawn concerns the fact that urban area junctions are accident-prone. In the USA 56 per cent of the accidents occurred at junctions, for England this number lies at 70 per cent, and in Sweden at least 46 per cent of the accidents that involve only motor vehicles occurred at junctions. This figure is probably larger, because it is likely that part of the rear-end accidents (21 per cent of all urban motor vehicle accidents in Sweden) also occurred at junctions.
In the English study it is stated that the majority of accidents for drivers occur as a result of an inability to anticipate, a failure to yield or a failure to anticipate (29, 16 and 10 per cent of all accidents respectively), the Swedish data suggests that 13 per cent of all urban area accidents are crossing accidents between two vehicles (i. e. suggesting a failure to yield or to anticipate), and also that 8 per cent of all urban area accidents are turning accidents (i. e. suggesting an inability or a failure to anticipate).
Also of importance in this comparison is the finding that 50 per cent of the accidents resulting from a failure to yield (which is the second most common form of accident) were attributed to perceptual factors for all road-user categories and a further 14 per cent to cognitive factors. The Swedish traffic researcher Brude (1993) arrives at nearly the same figure – he suggests that 61 per cent of all urban area accidents are attributable to detection related (i. e. perception and cognition) failures. A more detailed comparison can be made for certain types of accidents in Sweden and in the USA.
It has to be kept in mind, though, that the Swedish data does not include damage only accidents, and there is no information given if damage only accidents in the US show a different pattern of occurrence than injury accidents. Only accidents in urban areas that involve motor vehicles are considered. The percentages with which different accident types occur are quite similar in Swedish and American urban areas. The most common type in the US is ‘ran traffic control’ with 22 per cent, which might be matched against crossing accidents in the Swedish data (26 per cent).
Rear-end accidents make up 18 per cent in the US and 16 per cent in Sweden, single accidents in the US account for 14 per cent, but for 22 per cent in Sweden. The next common category in the US is lane-changing accidents with 13 per cent, which (including overtaking) account for only 3 per cent in Sweden. ‘Left turn oncoming’ accidents in the US make up 9 per cent of the urban accidents, while the figure for Sweden lies at 12 per cent, ‘left turn waiting’ accidents make up 3 per cent in both countries.
It can be seen that the distribution of accident types is quite similar for both countries, with the exception of single accidents and lane changing accidents. This last difference might be due to the fact that roads with more than one lane in each direction are much more common in American urban areas than in Swedish cities, which makes the occurrence of lane changes more likely in the US. 15 4. SAFETY ENHANCEMENT M EASURES IN THE TRAFFIC S YSTEM Road accidents are the result of a potentially large number of causal factors that exert their influence at approximately the same location and time.
In order to be able to propose, evaluate, and compare safety enhancement measures within the context of the traffic system, a suitable model is required. There are a number of models that can be applied for road safety management in order to describe the safety situation at a national or communal level. While many models tend to focus on the events surrounding the accident occurrence, or human error mechanisms, one of the more useful and more comprehensive models that is recognised internationally focuses on three dimensions related to: exposure in traffic; the risk of an accident given the exposure; and the consequences (i. . severity) of accidents (see e. g. Thulin & Nilsson, 1994; Rumar, 1996; and OECD, 1997). Other models also aim at predictability or effectiveness evaluation, perhaps using econometric modelling. There are also models that exist at the micro-level in order to describe safety problems at an individual level. These models are usually associated with the evaluation of subjective risk and are predisposed to problems associated with the fact that accidents are random and essentially unpredictable at the micro-level, requiring the use of other less valid proxy measures or safety indicators.
Many of these micro-models are not comprehensive enough to be of any great value in assessing risk within a systems context. Researchers in the field usually advocate a multi-disciplinary approach, although presently there are no known models that can bridge the macro-micro gap (see OECD, 1997 for a more complete description of different models). The three dimensions suggested in above model are adopted for a description of the more traditional approaches taken when dealing with the traffic safety problem.
Emphasis is given to urban area safety although many countermeasures may also have value in other areas. One of the advantages of this model is that it can be applied to the three primary elements of the traffic system (i. e. road users, vehicles, and the roadway environment), in order to form a 3 by 3 classification matrix for the many different traffic safety countermeasures (see Table 2 below). Table 2: Classification matrix for different types of traffic safety countermeasures related to the three main elements of the traffic system. Road-User Exposure E. g. mprovements in attractiveness of alternative (safer) modes of transport to relieve congestion and reduce travel time, etc. E. g. better use of other safer forms of (public) transport, ITS-functions which aid driver perception and decision-making, better driver training, etc. E. g. better use of in-vehicle injury prevention devices such as seat belts, better use of protective clothing by unprotected road users, etc. Vehicle E. g. more effective use of roadsystem through route-guidance, increases in parking costs, annual vehicle/road taxation, fuel taxation, etc. E. g. se of improved technology to provide better handling, ABS brakes, high-mounted brake lights, better man-machine, interface, warning systems, etc. E. g. better vehicle design and use of materials that can absorb energy on impact, provision of injury prevention devices such as air-bags, etc. Roadway E. g. better infrastructure design, differential toll-systems, improved traffic management strategies for better flow and capacity etc. E. g. improvements in roadway design, better visibility and lighting, separation of different types of road users to reduce possibility for conflicts etc.
E. g. Greater speed restrictions, Limited access to urban roads, Removal of dangerous roadside objects, etc. Accident Risk Accident Severity Safety Improvements Aimed at Individual Road Users For the different categories of road users, changing levels of exposure is very rarely stated as a primary goal for the purposes of improving the traffic safety situation. This may be related to the fact that accurate exposure data are very difficult to collect and are rarely presented in conjunction with accident statistics, thereby making exposure related improvements difficult to measure.
There are also problems associated with making comparisons between different modes of transport owing to the fact that exposure can be measured in terms of travel time or travel distance. In Sweden travel habit surveys that focus of establishing reliable exposure data have been performed on several occasions, and related to accident data (e. g. Thulin & Nilsson, 1994). 16 Factors, which tend to reduce road user exposure in general, are usually associated with the attractiveness of alternative modes of transportation.
For example, reductions in the use of private vehicles in favour of public transportation (e. g. buses and commuter trains) usually occur as a result of more competitive pricing combined with good levels of service, where the change of one transportation mode to another can also result in lower levels of congestion and a reduced exposure level in the traffic system in general. Careful regional planning is required to find a balance in the system that reflects the demands of the users in relation to the different forms of transportation and their perceived costs and benefits (i. . attractiveness), and the effects that such changes might have for the community at large. Another important factor regarding changes in modes of transportation is that many forms of public transport are safer by comparison than the use of private vehicles, and cycling and walking, thereby reducing not only exposure but also accident risk (ETSC, 1999a). Safety countermeasures that reduce the accident risk for road users are now commonly provided by the many different in-vehicle systems that enhance perceptual abilities and aid cognitive decisionmaking (see e. . Malaterre & Fontaine, 1993; ETSC, 1999b). Many of these types of systems that have to do with the way information from the environment is interpreted, manipulated and acted upon, are discussed in the following chapter concerning Intelligent Transportation Systems (ITS). The quality of driver education training is also an important factor, and has resulted in a number of new approaches such as ‘graduated licensing’, which is built upon the concept that driving tasks become gradually more difficult and unrestricted.
Generally, the majority of systems include three levels of tests, an initial provisional license, a preliminary licence, and a full licence (Englund et. al. , 1998). This form of graduated licensing has been shown to have a number of beneficial effects including a reduced risk for accident involvement (Smith, 1994). Inverted licensing, which encourages drivers to understand and accept responsibility for their actions has also been suggested (Klyve, 1998). Attempts to reduce accident risk through public safety campaigns and education programmes (e. . for school children) have often had a limited long-term effect on traffic safety (see e. g. OECD, 1986; Jarmark, 1992; Linderoth & Gregersen; 1994; Englund, Nyberg & Thiseus, 1997). Attempts to reduce accident severity for road users, particularly drivers and passengers of motor vehicles, has been the subject of a great deal of research by different vehicle manufacturers, particularly those who use the safety concept as an argument for marketing campaigns.
The actual road user injury protection that has been developed as a result of large investments in in-vehicle design and research is discussed below in relation to vehicle engineering. For the safety of drivers and their passengers, the most important factor often concerns the use of injury protection measures rather than their provision. This is especially important with regard to the fact that many drivers consider themselves better than the average (Svenson, 1981; Williams, Peak & Lund, 1995). In many European countries, seat-belt use is relatively low resulting in an unnecessary number of fatalities and serious injuries.
Many also fail to realise that the effectiveness of air bags is dependent on the use of seat belts (ETSC, 1999b). For unprotected road users the risk for (and consequences of) an accident can be reduced by ensuring easy detection by other road users through the use of suitably coloured clothing, reflexes, and cycle lights (OECD, 1998). Safety Improvements Through Vehicle Engineering Vehicle engineering has come a long way during the past century, a fact that is reflected by the relatively small number of accidents that are directly attributable to vehicle failure.
Modern vehicles undergo rigorous testing in order to ensure stability and handling, adequate steering charact