In the Netherlands, cycling and walking are two very important modes of transportation, but also vulnerable modes because of their unprotected state. By constructing bicycle tracks and sidewalks, vulnerable road users on road sections are separated from other traffic. Still, many crashes happen on these types of infrastructure. The most common causes for crashes on bicycle tracks are steering off the road (against the kerb or in the verge), cycling into a bollard, and a slippery road surface. In the goal of creating a safer environment for bicyclists and pedestrians, there is an on-going debate about the usefulness of kerbs as a separation between these two modes. This resulted in the project: Prevention of accidents with kerbs and other types of edges by Rijkswaterstaat. Architects and urban designers have applied many different forms of kerbs as there is no official guideline on how to design them in urban space. Kerbs can thus be found in many different designs, using different colours, slopes and heights (see Figure 1). Although the different kerb types are commonplace in practice, their safety effects are barely examined. This research fills in parts of the knowledge gap between the identified problem (accidents with kerbs) and the lack of research done on (the safety of) different types of kerbs. The main research question that is answered is: How do different types of kerbs, varying in slope, relate to the behaviour of bicyclists and pedestrians? The scope lies with situations within the built-up area for bicycle-pedestrian infrastructure, physically separated from car traffic, excluding cars from this research. Only straight road sections are examined. Within this research, kerbs are categorised into three different types: right, sloped, and levelled. The analysis on the behaviour of bicyclists and pedestrians on these different kerb types is done by the means of camera surveys. Video recordings are made of 12 locations (4 per kerb type), in which the traffic is analysed for 120 minutes per location. The video analysis consists of counting the number of cyclists, pedestrians, scooters and other traffic user types. Also the infrastructure characteristics such as kerb type, width of the bicycle track and sidewalk, and colour contrast are measured. The behaviour of the traffic users is operationalised as a traffic user moving onto another traffic user?s infrastructure. For a cyclist, this is described as a cyclist moving onto the sidewalk, abbreviated by COS (Cyclist On Sidewalk). For scooters, this behaviour translates to SOS (Scooter On Sidewalk), and for pedestrians, this translates to a Pedestrian moving Onto the Bicycle track (POB). The focus of this research is on finding a relationship between these three kerb types and the COS behaviour. In finding this relationship, many other infrastructural variables are of influence to measuring the behaviour of cyclists and pedestrians in relation to the kerb. The kerb safety variables that are included in the analysis are: Kerb Type, Kerb Height, Kerb Slope, Width of Sidewalk, Width of Bike Track, Width of Shoulder, One Way or Two Way Bicycle Track, Traffic Intensities, Traffic user characteristics (age and gender), Presence of Colour Contrast, and Obstructions & Installations. There are more variables possibly influencing the behaviour of traffic users around kerbs (e.g. weather effects, influence of cars, curve situations). For making this research practicable, these variables are excluded (their possible effect is minimised) by taking similar study locations or by applying a specific scope. Within the camera survey, a total of 14,502 bicyclists and 3578 pedestrians were observed. The number of cyclists that moved onto the sidewalk (COS) was 42 (0.21%). Figure 2 shows these 42 COS cases divided into the three different kerb types. It shows that in the attempt to separate bicyclists and pedestrians, the right kerb functions best, followed by the sloped variant. The following additional conclusions can be drawn per kerb type: - Right kerb: With only 1 case of COS (0.02%), the right kerb shows that cyclists do not move onto the sidewalk often, in absolute sense, but also compared to the other kerbs. Since a vertical kerb (though dimensions vary) is not designed to allow cyclists to move onto the sidewalk, this effect was as hypothesised. - Sloped kerb: Having 5 cyclists on the sidewalk (0.09%) within the observed data set, the number of COS cases for sloped kerb types still is very low. Taken into account that this type allows cyclists to move onto the sidewalk more easily than the right kerb, the degree to which they actually do it, is still relatively low. - Levelled kerb: With 36 COS cases, the levelled kerb shows a significant difference between the COS numbers of the right and levelled kerb. Still, with 0.76%, the extent to which this happens compared to the total amount of cyclists for levelled kerbs is relatively low. The main conclusion that can be drawn from these observations is that COS behaviour is a rare event, and that all kerb types function well in terms of separating bicyclists from pedestrians. Given the 42 cases of cyclists moving onto the sidewalk, an in-depth analysis is executed to see what are the most common reasons for moving onto the sidewalk. The specific motives for moving onto the sidewalk are mostly unclear (22 out of 42 cases). Other multiple observed reasons are for passing other road user(s) (6) and to stop and check their phone (4). Within the locations observed, COS did not lead to any unsafe (accident-related) situations within the occurrences. COS movements happen for both genders and within all age groups. The data further shows that a COS movement in this research is often related to a bicyclist in close proximity (in 57% of the cases a cyclist was within 1 meter). This could indicate that cyclists nearby have a relationship to cycling on the sidewalk. Related to the intensity, the presence of cyclists within 30 meters of the COS movement is also relatively high (occurred in 32 out of the 42 COS cases). Binomial logistic regression is performed to ascertain the effects of all included variables (Width of Sidewalk, Obstructions & Installations, etc.) on the likelihood that cyclists move onto the sidewalk. The kerb type is found as the only variable statistically significant affecting the COS in the logistic regression model. A similar analysis is executed for the other traffic users. A significant relation was not found for the pedestrians on the bicycle track (POB) compared to the kerb type. Therefore no further conclusions can be made upon the pedestrians, and it is assumed that other infrastructural variables are of a higher influence than the kerb type. For the scooters, only two cases in which they went onto the sidewalk (SOS) are observed, both for the levelled kerb type. As numbers are small, no statistical relationship to the kerb type was found. Attempts are made on relating the cyclist?s trajectory to the different kerb types. The program „Image Tracker? and „Trajectory Viewer? are used for researching the video data. Due to a high distortion of the images in the orthorectification phase, it was unable to do accurate measurements of the trajectories. For researching the trajectories with these software tools, it is highly recommended to place the cameras such that a top view is created. Relating these research findings to literature and safety, additional conclusions for the design and traffic safety policies on kerbs are made. Literature shows that right kerbs are not the most preferred option since they do not allow cyclists and pedestrians to make mistakes, in other words, they are not forgiving. The low amount of COS indicates that all kerb types function well in terms of separating bicyclists from pedestrians. In the ambition of separating bicyclists from pedestrians because of their different speeds and masses, as well as designing a forgiving kerb to accept people making mistakes, both the sloped and levelled kerb come out as best practice. However, no research was found on how the different types of sloped and levelled kerb function in terms of forgivingness. In choosing between kerb types, a trade-off should be made in terms of forgivingness and level of separation. If separation of traffic users is assumed more important, a sloped kerb is preferred. If higher level of forgivingness is assumed more important, a levelled kerb is preferred. More research is needed in measuring what the safest dimensions and designs of kerb types are, and to what extent they are actually forgiving. The behaviour of bicyclists and pedestrians should also be researched around curves and intersections. COS behaviour might be more common in these situations. (Author/publisher)
Abstract