I NCORRECT U-TURNING OF VEHICLES AT INTERSECTIONS WITH TRAFFIC LIGHTS

: The article describes the problem of incorrect U-turns at intersections with traffic lights. Statistical data on road incidents related to U-turns are presented. Then, the international, Polish and foreign regulations concerning u-turning at intersections with traffic lights were analysed. The situations in which U-turns are allowed or prohibited are presented. The differences in design rules for junctions with U-turns in different countries have been taken into account. A literature review was also carried out that outlined various current U-turns around the world, including the design of turning places, the location of turning points, road safety when turning, and the impact of U-turns on traffic conditions. The further part of the article presents the results of field tests of the U-turn at 6 intersections located in Warsaw. The research was conducted by video observation. The results were broken down by age, gender, place of registration of the vehicle, type of vehicle, and the effect of incorrect turning. Data on road incidents at the examined intersections were also analysed. Data from the database kept by the Police were compared with the measurement data. A regression analysis was performed between the types of recorded incorrect manoeuvres and the number of accidents at the intersection. The results of statistical analysis carried out do not indicate the existence of a relationship between the number of identified incorrect U-turns and the number of road incidents at intersections. Based on the research, it was found that the phenomenon of incorrect U-turns at intersections with traffic lights is common, and the use of directional (protected) signals does not eliminate this phenomenon. The conclusions indicate practical solutions to reduce the number of illegally U-turning vehicles. The recommended actions are related to the stage of shaping the road network, designing the road geometry and organizing traffic and traffic lights, and auditing road safety, as well as the stage of road operation.


Introduction
Observations of road traffic on the streets of Warsaw show that despite improvements in traffic organization, modernization of traffic signals at intersections, and improvement in road safety, a large number of traffic accidents and collisions can still be observed, caused by the inappropriate, often illegal behaviour of traffic participants. The issue of road safety is related to many aspects of transport. The topic is important because road accidents not only have a direct impact on their victims but also on the travel times and operating costs of vehicle fleets (Rudyk et al., 2019). In the case of motor vehicle drivers, the problem of illegal U-turning can be observed. Data obtained from the Accident and Collision Recording System show that in 2010-2019, the number of accidents related to illegal turning in Warsaw oscillates around 100 events (SEWIK, 2020). Approximately 20-30% of these accidents occur at intersections with traffic lights operating in the three-colour mode. With non-functioning traffic lights, a maximum of one event per year is recorded. The statistics do not indicate significant changes in the number of accidents from month to month. However, variability within the week is noticeable -the number of accidents on Saturdays is about 1/4 less than on weekdays, and on Sundays, the number of accidents drops by half compared to weekdays. When analysing the problem of inappropriate vehicle U-turning in prohibited places, the high traffic volume in Warsaw, the lack of places where this manoeuvre can be performed safely, as well as the drivers' lack of respect (and often knowledge) of traffic regulations can be identified as the cause of this problem. The phenomenon of lack of knowledge of the rules or even deliberate violation of them is often overlooked during analyses related to the causes of drivers' behaviour (Muslim et al., 2018). Therefore, the need for empirical research related to drivers' incorrect U-turning was identified.

Literature review 2.1. U-Turning rules in legislation
The current road traffic regulations are contained in (MI, 2002), in (Sejm RP, 2012) and international regulations -(CoRT, 1968) and (CoRSaS, 1968). In turn, (MI, 2003) provides guidelines on the use of appropriate horizontal and vertical signage, types of traffic signals placed at intersections and road traffic safety devices.
The Vienna Convention on Road Traffic (CoRT, 1968) is an international treaty in force in the countries which are its signatories. It defines general rules for the movement of vehicles, their equipment, their interaction with pedestrians and cyclists. U-turning is covered only by one provision, indicating prohibition of this manoeuvre on a motorway. The Vienna Convention on Road Signs and Signals (CoRSaS, 1968) only defines the sign prohibiting a U-turn and does not refer to the admissibility of U-turns at intersections. In Poland, if there are no signs on the road or intersection, U-turning on the road or intersection is permitted. The manoeuvre of U-turning is not forbidden, also when signs are placed on the intersection: F-10, F-11, even if the possibility of U-turning is not indicated on them (a left turn permit is enough). Horizontal signs allowing to perform a U-turn at the intersection, with general signals at the intersection approach, are signs: P-8b, P-8c, P-8e, P-8g, P-8h, P-8i. U-turning is allowed only from the inner lane unless signs indicate that this manoeuvre may be executed from more than one lane. At intersections controlled by traffic lights, the U-turning manoeuvre is possible when the following signals are present: S-1, S-2, and S-3, but only with the arrow symbol for U-turning or U-turning and turning left (Fig. 1). The prohibition of U-turning may be introduced by vertical signs, horizontal signs and traffic lights (MI, 2002(MI, , 2003. Vertical signs prohibiting the manoeuvre of U-turning at an intersection are prohibition signs: B-21, B-23 and order signs: C-1, C-2, C-3, C-4, C-5, C-6, C-7, C-8. Road surface markings prohibiting U-turning on roads and intersections are: P-2a, P-2b, P-3a, P-3b, P-4. At intersections controlled by traffic lights, the U-turning manoeuvre is not possible when there are S-3 directional signals that do not directly indicate the possibility of U-turning, including signals intended for left-turning drivers (Fig. 2 Another direction is the use of an intersection where left-turn relations from the main road and the subroad are eliminated while leaving the possibility of straight traffic, which is referred to by the term "Median U-Turn Intersection Treatment" (MUTIT) (FHWA, 2007). This is equivalent to the junction known in Poland as a "cigar". When designing such an intersection, it is necessary to analyse geometric alternatives, the location of the intersection, rideability, MOEs, the possible use of traffic lights, road signs (including guide signs) and the impact on road safety. Research in this area focuses on defining criteria for the use of such a solution taking into account its efficiency and geometric parameters (Distefano et al., 2016). There are also studies showing traffic safety at such intersections, using simulation models and the traffic conflict analysis method (Kronprasert, 2020). They show an improvement in traffic safety after using a MUTIT-type intersection, but with an adequate distance between the U-turning places and the intersection with the cross road. Studies on the effectiveness of such intersections equipped with traffic lights (Bared et al., 2002) indicate an overall improvement in traffic conditions, but at the cost of increased delay for U-turning relationships. There has also been researched indicating the feasibility of using non-controlled intersections as U-turning locations and including an evaluation of traffic conditions at such intersections (Fan et al., 2013). The study also showed a solution to improve traffic conditions for turning vehicles by using double U-turning places, separate for light and heavy vehicles. Such a solution can be applied with narrower dividing lanes, but the literature only shows results of simulation studies and no examples of the application of such solutions in practice are known. A review of foreign literature indicates that there is a tendency to move U-turning places outside the intersection area. This improves traffic conditions for vehicles driving straight ahead (especially on the main road) and improves road safety by replacing dangerous left-turning manoeuvres outside the intersection. Also of importance is the fact that in such a case the driver has an easier task as he has to give way to a smaller number of traffic streams, which is a factor that improves safety (Szczuraszek et al., 2008). The disadvantages of such solutions include poorer legibility of the junction, which requires the use of appropriate guide signs, and increased delays by vehicles performing manoeuvres at the junction. This limits the use of such intersections to situations where roads of clearly different significance in the traffic system intersect.
In Poland, there are few roads with a wide dividing lane. Therefore, the possibility of using MUTIT and RCUT type crossings is limited. There are only a few turning places designated outside intersections. In most cases, these are used to improve the efficiency of traffic control and are placed before the approaches of controlled intersections. For this reason, most U-turning manoeuvres take place at intersections. Intersections in cities were often established as uncontrolled intersections, with the increase in traffic using traffic lights with general signals, and later using directional signals and dedicated phases for left turns.
Traffic safety analyses of U-turning movements are hampered by the fact that accident databases only include accidents at which the police were present. And even for these accidents, the information is often incomplete and does not allow the exact cause of the accident to be determined (Żukowska, 2015).

Research fields
Intersections located at various locations in the city of Warsaw, Poland, were taken into consideration to evaluate the U-turning manoeuvre. All intersections used S-3 left-turn signals during the study period, which prohibit turning according to regulations (MI, 2002). Intersections with different geometric layouts and fairly high traffic volumes were selected for the study. Seven intersections were evaluated. The research was conducted in September, October and November 2017. Intersection No. 1 -the intersection of Przyczółkowa Street with Wilanowska Avenue. The analyzed lane is located on the southern approach of the intersection. There are two lanes designated for left turns, each 3.5 m wide. During the measurements, a fixedtime control was in operation at the intersection. During the measurements, the conflicting turning traffic was moving from the lane marked with A-7 and P-13 signs. As a result of the non-adjustment of the intersection road signs to the regulations in force at the time (MI, 2003(MI, , 2015 and the lack of signalization in the entire area of the intersection, vehicles turning left and illegally U-turning from Przyczółkowa Street were moving simultaneously with vehicles turning right from Wilanowska Avenue. This is a dangerous situation that endangers the safety of road users.
In the period after the measurements were conducted, the traffic lights were modernized and the colliding right-turn relation was controlled with directional signalling devices. In the initial period of operation of this solution, a very high number of conflicts between vehicles illegally U-turning and those turning right were observed. This was also caused by the routing of routes with a turning relationship by the navigation systems. After notifications to the navigation system operator, made among others by one of the authors of the article, routes including turning at this intersection were no longer determined.

Research method
The research was carried out using a Panasonic Lumix camera. From the observation point, the drivers' behaviour at the observed intersection was recorded. The recordings from the camera allowed for a more in-depth analysis of the method and consequences of turning. The measurements were made in a way that was unnoticeable for the road users. This allowed eliminating the behaviour of drivers, whose actions could be changed during the observation.
The tests were carried out at different times and on different days of the week. The minimum sample size for which conclusions could be drawn was set as 25 turning vehicles for one measuring point. To obtain as many turning drivers as possible, measurements were made for 1.5 hours at each intersection. According to research (Hadi et al., 1995), 45% of traffic accidents at urban intersections were caused by heavy traffic. For this reason, the surveys were usually carried out during the morning or afternoon rush hours. The measurements were carried out during rain-free days in September, October and November 2017. Driving behaviour is influenced by many factors, related to the current traffic situation and directly to the driver, including the driver's gender, age, familiarity with the infrastructure they are on and the purpose of driving, the need to hurry, habits (Fuller et al., 2002). During the measurements taken, the turning drivers were classified into different categories such as the type of vehicle, the place of registration of the vehicle they were driving, the gender and age of the driver and the presence of a passenger.
In the category of the type of vehicle driven, a distinction was made between passenger cars, heavy vehicles (vans, trucks, buses) and single-track vehicles. Based on the number plates, vehicles were divided into three categories: registered in Warsaw, registered up to 30 km from Warsaw, and registered more than 30 km from Warsaw. The surveyed drivers were also divided into men and women. In the driver age category, three age ranges have been established, based on estimations only. These are 18-30 year-olds, 31-50 year-olds and 51+ year-olds.

Research findings 5.1. Results of fieldwork
Surveys were conducted and observation sheets completed for all locations. Overall, the number of incorrect U-turns during the 1.5 hours of observation at each location is shown in Table 1. The behaviour of drivers U-turning inappropriately at individual intersections was analysed in depth. The paper presents a comparative analysis of the results of measurements at individual intersections. Summary results of measurements at all intersections are presented in Table 2. Abbreviations used in Table 2 mean: RP -vehicle type (SO -a   The table shows that the most common vehicle was a passenger car. It was noted that those turning with a heavy vehicle would have difficulty manoeuvring due to the geometry of the intersection. This could be due to the large dimensions, large turning radius and low dynamic capabilities of the car. When analysing the percentage share of turning drivers in relation to the place of vehicle registration, the highest number of recorded vehicles was registered in Warsaw. At each studied intersection, the percentage share of women was significantly lower than that of men. At each intersection, the most numerous age group was between 31 and 50 years old. When measurements were made at different intersections, different effects of turning were detailed. To obtain a comparative analysis of the intersections, Table 3 shows the effects of turning at unauthorised places for each intersection. The effects of turning were defined as follows: Apassing safely without forcing priority on another traffic participant, B -forcing another driver to reduce speed slightly, C -forcing another driver to reduce speed significantly, D -forcing another driver to stop at the intersection, E -running into the kerb. Based on Table 3, graphs were created (Figure 3 and Figure 4), taking into account only the negative effects caused by illegal U-turning. Based on the data from the graphs, it is possible to identify the locations where drivers most frequently disrupted the flow of traffic. Intersection 3 recorded the highest proportion of negative impacts of illegal U-turning. Drivers were mainly entered on the kerb and forcing other road users to significantly reduce their speed.
The reason for such a bad result may be the prohibition of turning at 1.7 km before the intersection and the geometry of the intersection, mainly the radius and width of the island dividing the carriageways. Intersection No. 7 had the highest percentage of people who forced another traffic user to slightly reduce speed. Among the observed intersections, the largest number of people forced a significant change in speed at intersection No. 5. The main reason for this phenomenon was that they performed the turning manoeuvre too slowly.  Krukowicz, T., Firląg, K., Sterniczuk, E., Archives of Transport, 57 (1), [131][132][133][134][135][136][137][138][139][140][141][142][143][144][145]2021 Intersections No. 1 and No. 4 had the highest number of people who forced other drivers to stop their vehicle completely. In the first location, this was due to a lack of awareness of the priority at the given intersection. At the second one, it was due to waiting too long before carrying out the turning manoeuvre, which resulted in the signal being changed to prohibit entering the intersection. It can be deduced from the observations that there is a variation in the effects of the U-turning manoeuvre at different intersections.  (Table 4). Based on Table 4, graphs were created (Fig. 5, Fig.  6). depicting the percentage of female and male U-turners depending on the intersection. Based on the graphs in Figures 5 and Figure 6, it is possible to compare the effects of U-turning among women and men at different intersections. The greatest discrepancies in negative impacts can be observed at intersections #3 (a greater proportion of negative impacts among males), #5 (a greater proportion of negative impacts among males), #6 (a greater proportion of negative impacts among females), and intersection #7 (a greater proportion of negative impacts among males). At the remaining intersections, the differences are less than 10%. The persons who turned back more often at the selected measurement points were men. Based on the study, it can be concluded that they more often negatively affected the traffic flow while manoeuvring. On this basis, it can be concluded that the driver's gender has an influence on turning in a prohibited place.  B  A  B  1  91  9  89  11  100  0  2  95  5  100  0  87  13  3  80  20  10  0  50  50  4  72  28  85  15  73  27  5  87.5 12.5  75  25  60  40  6  73  27  65  35  86  14  7  56  44  67  33  86  14 Drivers between the ages of 31 and 50 were the most common group among those surveyed. At the two measurement points, 100% of the U-turning drivers passed safely, without obstructing other drivers. The overall share of drivers who more or less disrupted the traffic flow in this age category was 17%. This is the best result among the surveyed age groups. At intersection No. 3, the highest share of cases of performing a U-turning manoeuvre in a way that impeded the traffic flow of other vehicles was noted. Such behaviour constitutes as much as 50% of all recorded situations. Among all age groups, the overall percentage of negative consequences of turning is the highest for adults over 51 years old. Based on conducted measurements, it cannot be unambiguously determined which drivers in which age bracket cause most traffic obstructions at selected intersections. Different percentages of individual consequences of an improper manoeuvre were recorded at the measurement points. Based on the presented data, it may be concluded that the age bracket of drivers does not influence the decision to perform a U-turn.

Analysis of traffic accident data
The analysis of traffic accidents related to wrong U-turning in 2017 in Warsaw showed that 105 accidents took place in this period. Only one location has a significantly higher number of accidents of this type as many as 7. This is Górczewska Street, in the vicinity of the Wola Park Shopping Centre. In this place in that period was introduced temporary traffic organization associated with the construction of the second line of the underground. At the remaining intersections no more than 1 accident caused by an incorrect U-turn was observed. This gives rise to a suspicion that these accidents are qualified as other accidents including side events. Traffic accident statistics for 2017 were analysed for the intersections included in the analysis. The number of traffic accidents at each intersection is shown in Table 6. Table 6. Traffic accidents at analysed intersections Intersection

Number of accidents in 2017
Number of side crashes in 2017  1  9  1  2  18  3  3  2  0  4  7  2  5  0  0  6  7  3  7 1 0 There are numerous rear-end crashes at intersections that are unrelated to U-turning and lateral crashes related to lane changes, which were omitted from the analysis and are not shown in Table 6. The SEWiK database does not contain detailed information on accidents, so it is impossible to link a given accident to an intersection approach, and accidents classified as sideswipe accidents may occur in different situations and locations at the intersection. However, statistical analysis was carried out by determining Pearson and Spearman correlation coefficients between the measured numbers of misbehaviour, speeding violations, stopping violations, total

Summary and conclusions
Based on the conducted measurements, it may be stated that U-turning in prohibited situations is a phenomenon commonly occurring at controlled intersections in Warsaw. On average, 58 drivers U-turned at the analysed intersections during a 1.5hour observation period. Most cases of illegal turning take place safely, which is due to good visibility and legibility of priority rules and wide exits at most of the analysed intersections. The situation is different in the case of conflicts with vehicles moving on a signal allowing U-turning with the vehicles moving on the right turn on red signal (so-called "green arrow") from a perpendicular approach -in this case, the visibility is much worse and the priority rules are not clear, as drivers do not know what signal is displayed for the other driver. However, analyses of traffic accidents indicate that the number of recorded traffic accidents at the analysed intersections is not significant and is not statistically significantly related to the recorded irregularities while U-turning. The main conclusion that can be drawn from the research is that the use of an S-3 directional signal allowing only left turns is not an effective means of ensuring the elimination of illegal U-turning at the intersection. The reason for this situation is the widespread lack of knowledge of the provision prohibiting U-turning in such a situation.  It would be advisable to apply supplementary driver education aimed at maintaining and enriching knowledge of traffic regulations (Szczuraszek, 2008). Other conclusions concerning particular categories of traffic participants indicate that, although the problem concerns the whole cross-section of the driving population, educational activities should be directed in particular to older male drivers. The conclusions of the research should also be used at the various stages of the life cycle of road infrastructure. At the stage of traffic planning and forecasting, proper planning of the road network is essential, including the designation of U-turning places. Such places should result from traffic measurements and, in the case of a newly designed road, from traffic forecasts. The possibility of U-turn should not be eliminated at the initial stage of making traffic forecasts to identify places where it is necessary. This, therefore, requires an update of the macroscopic traffic models used to perform forecasts. At the stage of road design, it is necessary to design the possibility of U-turning at junctions where the traffic of U-turning vehicles is forecast. The surroundings of the road and the possibility of access to facilities located on the road should be analysed. The U-turning should provide adequate rideability. Solutions presented in foreign literature for the design of U-turning places allow the elimination of the phenomenon described in the article. However, these solutions can only be applied at selected intersections, and Polish roads, with narrow dividing lanes, do not allow their widespread use. Nevertheless, during the design of new roads and a thorough reconstruction of existing roads, such a solution may be considered. One of the methods of eliminating U-turning at unauthorised places is to provide an opportunity to safely perform this manoeuvre at designated places. Often U-turning at unauthorised places is caused by the lack of possibility to perform this manoeuvre on a long section before and after an intersection. During road operation, U-turning at forbidden places can be eliminated by proper drafting of maps used in car navigation systems -it is necessary to eliminate in these maps the relation of U-turning at a place, where it is forbidden in the current traffic organization, although in reality this activity is not performed. U-turns in prohibited locations may be caused by the navigation system's designation of such a route and the driver's reluctance to take a route with a longer travel time. This phenomenon increases especially during periods of congestion on the road network (Juhász et al., 2017). In cases where illegal turning is the cause of repeated traffic accidents at the operational stage, other traffic organisation measures should be applied, e.g. additional B-23 "No U-turning" signs. Although their use is not formally necessary when an S-3 traffic signal is used for vehicles turning left, it unambiguously prohibits drivers from U-turning at a given place. It should be noted that even the Road Safety Audit Manual (Podręcznik audytu BRD, 2019) does not provide guidance for evaluating a road for the placement of U-turning areas. In the opinion of the authors, this manual should be supplemented with issues related to U-turning, at each stage of the road safety audit. In particular, the need to pay attention to considering the location of U-turning places in conjunction with the development of the road environment should be emphasised. If on the section between intersections on a dual carriageway there are sources or destinations (including individual exits), there will be the phenomenon of U-turning connected with servicing those facilities, even not complying with the regulations. As such, provision must be made at the design stage for legal U-turning to and from these facilities. The importance of this issue increases as the road safety audit procedure is being extended to non-TEN-T roads in 2021 (Directive (EU) 2019/1936, 2019). The problem with implementing a U-turn is often the width of the dividing lane (e.g. intersection No. 3), which does not make it possible to ensure passage for all vehicles. In such cases, a solution applied by, among others, the Mobility and Transport Policy Office of the Municipal Office of the Capital City of Warsaw and the General Directorate for National Roads and Motorways (GDDKiA) in Warsaw is the use of the S-3 directional signal device permitting a U-turn and the placement of the B-23 "No U-turn" sign with a label reading "Not for passenger cars and single-track vehicles". It is also possible to additionally widen the road shoulder to ensure passability for long vehicles. These proposals apply to all types of traffic lights. The issues related to the proper handling of turning movements are dealt with in the earlier design stages, before the development of the traffic control algorithm. Therefore, once the presented conclusions are taken into account, any control algorithm can be used e.g. (Sathiyaraj et al., 2020), (Zhao et al., 2018), (Lin et al., 2020). During the operational stage of the road, more frequent traffic police checks are also recommended at intersections where the problem of improper U-turning has been diagnosed. Consistency and punishment of traffic violators should effectively improve road safety. In countries with strict penalties for noncompliance with traffic regulations, drivers are less likely to break them. The best examples of such countries are the USA and Sweden (Szczuraszek, 2008).