SAFE DESIGN OF RURAL ROADS BY NORMALIZED ROAD CHARACTERISTICS Thomas Richter, Prof. Dr.-Ing. Technische Universität Berlin, Department road planning and operation Benedikt Zierke, Dipl.-Ing. Technische Universität Berlin, Department road planning and operation ABSTRACT The standard designs of rural roads vary a lot and show several failings affecting the road safety negatively. The paper presents some results of a categorization of existing roads. This standardization of road design is believed to identify low volume roads and therefore harmonize the driving behaviour and increase the road safety. Based on empiric analysis road sections were dedicated and evaluated. The new single lane cross section for low volume roads showed positive results. Overall, the results of the project indicate that normalised road characteristics within design classes have positive effects on the driving behaviour and therefore the safety level on roads. 1 INTRODUCTION Within the project "safe design of rural roads by normalised road characteristics" (Richter & Zierke 2008) the TU Berlin was researching which road designs affect driving behaviour and road safety positively. The project aims are to provide self-explaining and recognizable roads and therefore increase road safety by creating design classes. Based on detailed accident analyses the design classes were derived. Besides the cross-sections including their road markings, the junction design and operation and the alignment were considered within the deliberations of safe design classes. The driving behaviour was examined in with/without and before/after situations on real roads. The empiric analyses on the test tracks were conducted by means of stationary radar measurements and the following-car technique on 30 different roads (over 10.000 km). Furthermore the detected self-explaining roads were analysed in the driving simulator of the TU Berlin. To ascertain the effect of these new kinds of roads they were compared with real roads implemented into the driving simulator. Besides driving speed, acceleration and lane keeping were measured along the stretch of road and locally on the real road as well as in the driving simulator. Current situation The standard designs of rural roads vary a lot. The question which road characteristics will lead to safe driving behaviour has yet to be answered. So far various studies have conducted research on specific road design elements, but have not been able to give an answer to the questions asked here. One approach to achieve a high quality road network is the standardisation of certain road types, which are supposed to have similar road Association for European Transport and contributors 2009 1
characteristics within their category, but show significant differences towards the other road types. In the course of the development of new guidelines for rural roads, the Richtlinien für die Anlage von Landstraßen (RAL) - Guidelines for the design of rural roads (FGSV 2008), the aim is to provide self-explaining and recognizable road types for motorists. According to the new RAL (FGSV 2008) rural roads are being classified by traffic relevance into one of four design classes. Thus, the designs of the cross-section and junction as well as the corresponding alignment are playing a decisive role. Expectations are that the general RAL (FGSV 2008) goals of normalizing rural roads, and through this positively influencing driver behaviour, will be enhanced by using characteristic designs. 2 SAMPLE OF TEST TRACKS The suitable road types were preselected based on the current state of discussion in task force 2.2.1 of the German Road and Transportation Research Association (FGSV) Designing new rural roads, which classifies four different categories. The main demands and goals of the separate road types were defined based on their role in the network (level of connection). On the one hand they are supposed to fulfil their regional development function with high road safety and a reasonable level of service and on the other hand they have to guarantee a low cost but sustainable protection of the environment and the surroundings. The consideration of the goals has to depend on the importance of the rural road for the network. This has to be the case because the often competing demands that the roads have to meet depend on their level of connection. In the target field of road safety the goals of safe horizontal alignment, safe turning and crossing, safe passing and overtaking and safe roadside could be deduced from the number and severance of the accidents in a conducted accident analysis. These goals can be met with a variation of measures. By using uniform road characteristics in cross-sections, alignment (radii sequences in particular), type of operation, overtaking concept and intersection layout the appropriate driving speed should become apparent to the driver. Considering these results the design classes now contain characteristic elements in the areas alignment, cross-section design, intersection design, operation type and roadside furniture. Based on the network function four design classes consisting of a total of seven road types were classified. Design class 1 has to manage the long distance traffic. It consists of a three to four lane cross section which is designed with a generous alignment to set the suitable speed to about 110 km/h. The connection of roads is processed by grade separated junctions. Agricultural traffic and bikes are not permitted. Association for European Transport and contributors 2009 2
Figure 1 typical cross sections of design class 1 Design class 2 consists of a two lane cross section with temporarily added passing lanes (about 15 % for each direction is intended). This 2+1 road still allows safe passing without using the lane of the oncoming traffic. Passing in the two lane cross sections is prohibited (see figure 2). The alignment is supposed to be semi generous to set the suitable speed to about 100 km/h. The connection of roads is processed by t-junctions with traffic lights. Bikes are not permitted. Figure 2 section typical cross section of design class 2 in the two lane cross Design class 3 consists of a two lane cross section which displays the regular rural road. The alignment is supposed to be semi adapted to set the suitable speed to about 90 km/h. The typical form of junction design is a roundabout. Association for European Transport and contributors 2009 3
Figure 3 typical cross section of design class 3 Design class 4 consists of a new approach of cross section design in Germany. For these low volume roads a new single lane road is believed to identify these roads and make them distinguishable to others. The suitable speed is at about 70 km/h. Figure 4 typical cross section of design class 4 3 DRIVING BEHAVIOUR ON THE TEST TRACKS 3.1 Methodology Driving behaviour as such cannot be described and must therefore be determined with the use of indicators. For this purpose three different parameter categories were used. The vehicle born parameters consist of parameters along the section and cross-section as well as the parameters on the driver s side of the vehicle. The vehicle born data can be split into longitudinal control (speed, acceleration and deceleration) and the transverse control (lateral acceleration, lane keeping). The empirical studies about driving behaviour have been conducted using two different methods. On the one hand local measurements were taken at certain locations along the test track. Relevant for the microscopic driving behaviour was data concerning the speed and lane keeping. On the other hand the driving behaviour of a single vehicle was monitored with the following-car technique along the entire test track. Association for European Transport and contributors 2009 4
Figure 5 inquiry of the driving behaviour The analysis of the empirical research was carried out using the aggregated data from speed, acceleration and lane keeping profiles. Furthermore the raw data were analysed statistically to verify the reliability of the conclusions or at least deduce observable trends. The results of the with/without and before/after study of the design classes show the positive effects of indicating normalised roads. For each design class one test track is shown below exemplarily. 3.2 Design class 1 Within design class 1 six test tracks with the characteristic design elements were analysed. Exemplary one road of about 5.5 km with three grade separated intersection is shown in figure 6. The profiles show that the driving speed within the open road and the fly over junctions are quite similar. Larger speed differences are recognized in between the cross section design. Due to the amount of lanes, the speed differs on these roads. An example of one test track is shown in figure 6. Figure 6 speed profile of a test track of design class 1 Association for European Transport and contributors 2009 5
3.3 Design class 2 Within design class 2 five test tracks with the characteristic design elements were analysed. Exemplary one road of about 9 km with two junctions with signal control is shown in figure 7. The speed profiles of the open road show that the driven speed is lower than for the test track of design class 1. As seen in figure 7 the speed limit within the intersections determine the speed within these sections. To achieve an appropriate speed within the intersections the length of the area with a speed limit before the intersections should not be undersized. Figure 7 speed profile of a test track of design class 2 3.4 Design class 3 Within design class 3 ten test tracks with the cross section design shown in figure 3 and roundabouts or regular junctions (mainly t-junctions) were analysed. Figure 4 shows a speed profile of an exemplary road of about 6 km with two roundabouts. Different to design class 1 and 2 the speed profile shows a considerably increased harmonization of the speed behaviour which is displayed by the lower vibration of the velocity. This is basically attributed to the lack of passing situations. Furthermore the top speeds a much lower. Association for European Transport and contributors 2009 6
Figure 8 speed profile of a test track of design class 3 3.5 Design class 4 Within design class 4 seven test tracks with the characteristic design elements were analysed. One of them was additionally equipped with a new road marking for a before/after study. After repaving this stretch of road it was the first test track in Germany on which the new road markings according to the new RAL (FGSV 2008) were applied. The road markings consist of two broken edge lines with an offset of 0.75 m from the edges of the road (2 m line length, S/L = 1/1) according to the RAL (FGSV 2008) (see figure 4). The old road marking existed of two solid edge lines which could hardly be seen and a spaced axis line (3 m length, S/L = 1/2). 5,70 2,85 2,85 5,70 4,20 0,75 Figure 9 0,75 Test track before (top) and after (bottom) the remarking Association for European Transport and contributors 2009 7
The speed profiles of the before/after study of the test track are shown in figure 4. The profiles reflect the expected influence of the alignment. An influence on speed in the area with radii of about 200 m can be noticed. Interesting is however, that the amount of speed reduction stays the same no matter which level of driving speed is measured. Figure 10 Speed profile, speed differences of the test track The average speed reduction for all percentile speed profiles was about 10 km/h for one direction. The other direction which already had a lower speed level in the before-situation showed a less noticeable speed reduction. For both directions though lower entering speeds in the villages could be ascertained. The local measurements confirm those results. They also show a reduction in speed for the after-situation by about 10 %. 4 DRIVING SIMULATOR STUDY 4.1 Methodology As a foundation for the tests in the simulator the existing knowledge about the composition of suitable road types was utilised. The defined road types were programmed for the tests in the simulator. They had to be driving behaviour optimised and include the requested road characteristics. For comparison real sections, which had been analysed during the empirical research were integrated into the tests. This allows an evaluation of the driving simulator comparing the real sections driven in the simulator and in situ on the one hand and a comparison between the real sections and the driving behaviour optimised sections on the other hand. 30 test runs per road type were conducted for the study. This means a total of approximately 6.000 km was driven on the test tracks. In the driving simulator similar measurements to the ones on the real roads were taken (speed, lane keeping). Additionally the visual behaviour of the test drivers was captured. Association for European Transport and contributors 2009 8
The results of the tests in the driving simulator were statistically analysed in the same way as the results from the in situ measurements to guarantee comparability. 4.2 Test sections In addition to the parameters defined in figure 1 a variation of road markings in design class 4 was analysed. Furthermore other vehicles with independent behaviour were added into the simulation in the course of this project. The traffic situation was hereby initiated location or event related. Figure 11 roads) characteristic cross-sections of design class 4 (low volume The driving optimised road sections were generated based on the suitable radii calculated. Therefore the circular curves with various radii and the associated clothoids and angles of aperture were strung together using a rational relation of radii. The average length of a driving optimised section is 5 kilometres. The sections can be connected indefinitely which allows driving without discontinuation. Each subject had to drive a total of 20 km on each test section which allowed identifying random events. The real test sections were also generated for the simulator comparing the GPS data from the empirical studies and the existing data from the site plan. The different cross section design is displayed in figure 6. 4.3 Driving behaviour The analysis of the average speed behaviour of all test subjects (n=45 rounds) without other traffic resulted in the profiles shown in figure 12. The average speed in the before-situation was about 95.2 km/h. The after-situation showed an average speed with a reduction of 5 % to about 91.3 km/h. An analysis by section for the open road and the junction area shows that the before-situation showed an average speed of 105.0 km/h and 83.4 km/h. For the after-situation with the new single-lane cross-section for the open road section a reduced speed of about 99.9 km/h was noticed. For the junction area the average speed went up to about 86.3 km/h. Association for European Transport and contributors 2009 9
Figure 12 speed profiles of the real driving simulator sections with the old (before) and the new (after) marking A precise consideration of the 85 % speed (V 85, the speed that 85 % of the drivers do not exceed) shows, equally to the results on the test track, a reduced speed in the open road section. Furthermore the speed reduction of high speed levels decrease stronger than on lower speed levels for the driving simulator section. The converse progress can be noticed at junctions. A high homogeneity of the speed profiles increases the safety level. The homogeneity itself can be displayed by several indicators. One of them is the frequency of the speed alterations over the road stretch. The total for alteration in the before situation was higher than in the after situation (46.5 1/h to 41.5 1/h). Therefore the speed along the stretch was more homogenous in the after-situation. On the other hand the homogeneity can be displayed by the variance of the drivers speed. Before the remarking the average variance was 8.2 km/h. The after-situation had a lower average variance of 7.8 km/h. The 70%-canal of the driving speed (definition of the normal driver) which displays the difference of the V 15 and the V 85 (the speed that 15% / 85% of the drivers did not exceed) did not show clear results. In contrast to the test sections with the real alignment (generous alignment) the driving behaviour adjusted alignment (adapted alignment) resulted in lower driving speeds. 5 CONCLUSION Overall the speed profiles of the test tracks which are designed with the characteristic elements of the design classes reflect the suitable speed as intended. This shows that normalised road characteristics within design classes have positive effects on the driving behaviour and therefore the safety level on roads. Association for European Transport and contributors 2009 10
In order to achieve the desired driving behaviour, it is not only important that roads can be distinguished from each other, which can also be enhanced by a uniform road layout within road types, but also which design elements achieve this distinction and uniformity. Therefore road markings are the essential elements to use. Especially on two lane cross-sections the different design standards were not indicated by the markings. By creating the new single lane cross-section the low volume roads are strictly separated from the standard roads. The implementation of the new design classes will take several years and will bring along problems in the conversion. Therefore recommendations have to be given to the planners to ensure the correct effects. Especially transition zones have to be dealt with cautiously. To ensure these needs the task force adjustment of existing rural roads was established last year. (Richter & Zierke 2009) REFERENCES FGSV (2008) RAL - Richtlinien für die Anlage von Landstraßen (Guidelines for the design of rural roads), Forschungsgesellschaft für Straßen- und Verkehrswesen, Köln, Entwurf (blueprint) Richter, Th., Zierke, B. (2008) Sichere Gestaltung von Landstraßen durch normierte Streckencharakteristik (Safe design of rural roads by normalised road characteristics), DFG-Zwischenbericht, Berlin, unveröffentlicht (unpublished) Richter, Th., Zierke, B. (2009) Ergänzungen der Richtlinien für die Anlage von Landstraßen (RAL) zur Anwendung beim am Bestand orientierten Um- und Ausbau (FE 02.293/2007/FRB) (Amendments of the RAL for the adjustment of existing rural roads), BASt-Zwischenbericht, Berlin, unveröffentlicht (unpublished) REFERENCES Since 2003 Thomas Richter is professor and head of the department road planning and operation at the Technische Universität Berlin. Since 2000 Prof. Richter is CIO of SHP- Ingenieure in Hannover. From 1993 to 2000 he was project manager of the SHP-Ingenieure for over 200 projects for traffic surveys, feasibility studies and the design of road infrastructure as well as further research projects in all fields. In between 1990 and 1993 Prof. Richter was after graduating from the University of Hannover research assistant at the University of Hannover. His research emphases were the design and the operation of junctions. Thomas Richter is a member of several task forces of the German Road and Transportation Research Association (FGSV) for new guidelines in Germany. Association for European Transport and contributors 2009 11
Technische Universität Berlin TIB 3 / 3-3, Gustav-Meyer-Allee 25 13355 Berlin, Germany richter@ils.tu-berlin.de, b.zierke@ils.tu-berlin.de Phone: + 49 (0) 30 314 72 553 Fax: + 49 (0) 30 314 72 884 Benedikt Zierke is scientific assistant at the department since 2006. Before he started his doctor (phd.) thesis on safe design of rural roads he was working during his course of study at the University of Hannover for the SHP- Ingenieure. His research activities at the Technische Universität Berlin contain the safe design of road infrastructure and traffic flow analysis. Benedikt is also part of the task force adjusting existing rural roads of the German Road and Transportation Research Association (FGSV). Association for European Transport and contributors 2009 12