Scholarly article on topic 'Safety effectiveness of pavement design treatment at intersections: Left turning vehicles and pedestrians on crosswalks'

Safety effectiveness of pavement design treatment at intersections: Left turning vehicles and pedestrians on crosswalks Academic research paper on "Civil engineering"

CC BY-NC-ND
0
0
Share paper
Academic journal
IATSS Research
OECD Field of science
Keywords
{"Left-turning vehicle" / "Pedestrian safety" / "Gap acceptance" / "Swedish traffic conflict method" / "Pavement design treatment"}

Abstract of research paper on Civil engineering, author of scientific article — Hasina Iasmin, Aya Kojima, Hisashi Kubota

Abstract Pedestrians are the most vulnerable road users as they are more exposed than other road users. Pedestrian safety at road intersections still remains the most vital and yet unsolved issue. One of the critical points in pedestrian safety is the occurrence of accidents between left-turning vehicle and pedestrians on crosswalks at signalized intersections. A crosswalk is a place designated for pedestrians and cyclists to cross vehicular roads safely. Drivers are expected to give priority to pedestrians or cyclists during interactions between them on the crosswalk. If a driver exhibits non-yielding behavior, the interaction will turn into a collision. This study examined the safety effect of three crosswalks designed with different materials such as red-colored material or brick pavement based on a safety performance study. The safety performance study considered left-turning driver's gap acceptance behavior and the severity of traffic conflict events between left-turning vehicles and pedestrians. The results of the study indicates that using brick pavement on a crosswalk increases the safety level of the crosswalk. Drivers at such crosswalks are more acquiescent to the priority rule.

Academic research paper on topic "Safety effectiveness of pavement design treatment at intersections: Left turning vehicles and pedestrians on crosswalks"

ARTICLE IN PRESS

[ATSSR-00120; No of Pages 9

IATSS Research xxx (2016) xxx-xxx

Contents lists available at ScienceDirect

IATSS Research

Safety effectiveness of pavement design treatment at intersections: Left turning vehicles and pedestrians on crosswalks

Hasina Iasmin *, Aya Kojima, Hisashi Kubota

Department of Civil and Environmental Engineering, Graduate School of Science and Engineering Saitama University, 255 shimo okubo, Sakura-ku, Saitama, 338-8570,Japan

ARTICLE INFO ABSTRACT

Pedestrians are the most vulnerable road users as they are more exposed than other road users. Pedestrian safety at road intersections still remains the most vital and yet unsolved issue. One of the critical points in pedestrian safety is the occurrence of accidents between left-turning vehicle and pedestrians on crosswalks at signalized intersections. A crosswalk is a place designated for pedestrians and cyclists to cross vehicular roads safely. Drivers are expected to give priority to pedestrians or cyclists during interactions between them on the crosswalk. If a driver exhibits non-yielding behavior, the interaction will turn into a collision. This study examined the safety effect of three crosswalks designed with different materials such as red-colored material or brick pavement based on a safety performance study. The safety performance study considered left-turning driver's gap acceptance behavior and the severity of traffic conflict events between left-turning vehicles and pedestrians. The results of the study indicates that using brick pavement on a crosswalk increases the safety level of the crosswalk. Drivers at such crosswalks are more acquiescent to the priority rule.

© 2016 International Association of Traffic and Safety Sciences. Production and hosting by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Available online xxxx

Keywords: Left-turning vehicle Pedestrian safety Gap acceptance

Swedish traffic conflict method Pavement design treatment

1. Introduction

Road intersections are very important and crucial locations where conflicts between different road users travelling from different directions are easily generated. To control traffic from different directions, traffic signals are very commonly used at intersections. They are operated for controlling the movement of conflicting road users. At a signalized intersection, usually, left-turning vehicle drivers (in a left-hand traffic system) are permitted to use the same signal phase that is allocated for the through vehicle. For operational efficiency of traffic, pedestrians are also allowed to use the same signal phase along with through vehicles from the same direction. The through vehicles and pedestrians move parallel to each other, while left-turning vehicles turn through the crosswalk by cutting the walking line of pedestrians (Fig. 1). Since a crosswalk is the place designated for pedestrians to cross the road safely, the maneuvering of turning vehicles at crosswalks is characterized by their compliance with the priority rule: in interactions between left-turning vehicles and pedestrians on a crosswalk, the pedestrians should be allowed to pass first.

* Corresponding author. E-mail addresses: hasina@dp.civil.saitama-u.ac.jp (H. Iasmin), kojima@dp.civil.saitama-u.ac.jp (A. Kojima), hisashi@dp.civil.saitama-u.ac.jp (H. Kubota).

Peer review under responsibility of International Association of Traffic and Safety Sciences.

Interactions between pedestrians and left-turning vehicles at crosswalks are critical situations in which the driver has to show modest behavior by changing his speed to avoid collision with the pedestrian. If the driver fails to do this, the interaction will turn into a collision. Accident data reveal that numerous accidents involved left-turning vehicles and pedestrians on crosswalks. In Japan, in the period 2008 to 2012, 49% of all pedestrian accidents occurred at signalized intersections. Of these accidents, 7.8% fatalities involved left-turning vehicles and pedestrians [1]. This indicates that left-turning vehicles do not give priority to pedestrians properly on crosswalks. Many situations, including invisibility problem, traffic volume, road geometry, road user behavior, and traffic signaling policy, may influence such accidents. Some researchers identified several factors influencing the yielding behavior of drivers, including speed limits and number of lanes [2], pedestrians' distance from the curb [3], pedestrians' clothes [4], and the number of pedestrians waiting to cross [5]. The number and positions of pedestrians were also studied in some other works [6,7].

Among all road users, pedestrians are the most vulnerable road user as they are the least unprotected users. In a collision between a car and a pedestrian, the severity is high for the pedestrian. In such a collision, usually, the car does not face any danger except for the risk of hitting other objects or cars because of the swerving motion for avoiding collision with the pedestrian. Therefore, it is important to ensure pedestrian safety on crosswalks. Some countermeasures such as roundabouts, raised crosswalks, curb extensions, raised intersections, and right-turn

http://dx.doi.org/10.1016/j.iatssr.2016.04.001

0386-1112/© 2016 International Association of Traffic and Safety Sciences. Production and hosting by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

ARTICLE IN PRESS

H. Iasmin et al. / IATSS Research xxx (2016) xxx-xxx

Fig. 1. Pedestrian and car directions when the signal is green.

Some researchers found that there is a close relationship between road geometry and road user behavior [8-10]. Considering this, the main purpose of this study is to investigate the effect of brick pavement or red color at intersections on the behavior of left-turning vehicle and evaluate their contribution to pedestrian safety on crosswalks. In this study, a comparative safety analysis with three pavement design scenarios was conducted on the basis of the gap acceptance of left-turning cars and the traffic conflict analysis method. The three scenarios are listed below:

1. No pavement design

2. Red colored pavement design

3. Brick pavement design

Each of these three signalized intersections were installed in a segment of an urban road one by one (Fig. 2). Except for the pavement design, all other traffic characteristics were almost the same (Tables 1,2).

This remaining parts of this paper are arranged as follows: After an introduction, the background related to gap acceptance and traffic conflict analysis study is described. Next, the paper describes the safety performance method, linking gap acceptance and Swedish traffic conflict analysis. Thereafter, the result and conclusions and future works related to this study are discussed.

channels, were implemented in order to improve pedestrian safety. However, all of these countermeasures have some limitations depending on the different road characteristics. It is more desirable to explore and execute new low-cost engineering solutions to improve pedestrian safety on crosswalks.

2. Study Background

In order to study safety performance, two topics are reviewed and summarized: the gap acceptance study and the traffic conflict analysis.

Fig. 2. Three sites with different pavement designs (Source: Google Map).

Table 1

Geometric and traffic characteristics at observational sites.

Intersection Average left turning car Average Intersection corner angle Width of Major road (m) Width of Minor road (m)

pedestrian /cyclist

(veh/h) Ped. Cyc.

No pavement design 6 12 17 90° 6 6

Red colored pavement 8 9 13 90° 6 6

Brick pavement 5 7 13 90° 6 6

ARTICLE IN PRESS

H. Iasmin et al./ IATSS Research xxx (2016) xxx-xxx

Table 2

Traffic signaling time at observational sites.

Intersection

Survey time

Green time (s)

Yellow time

All red (s)

Red time (s)

All red (s)

Total cycle (s)

No pavement design Red colored pavement Brick pavement

9.00 am to 4.00 pm 9.00 am to 4.00 pm 9.00 am to 4.00 pm

26 31 30

80 80 80

2.1. Gap acceptance studies

Drivers always seek the right opportunity to cross the intersections. This opportunity is termed "gap," and the behavior is called "gap acceptance" [11]. A driver has to wait for an acceptable gap or lag larger than his critical gap. Otherwise incorrect gap acceptance may cause accidents between road users [12,13]. If drivers tend to accept small gaps, the probability of collision between road users may increase. Gap acceptance is a well-known method to study the manner in which drivers move into a priority area where they must give way to other road users. Many researchers have modeled drivers' gap acceptance behaviors toward pedestrians or cyclists in the priority area. Sun. et al. applied logit and probit models to analyze drivers' yielding patterns at an unsignalized pedestrian crosswalk [5]. A logistic-regression model was developed to predict the yielding or gap acceptance behavior considering different factors, including the presence of pedestrians crossing [14].Asano et al. modeled the gap acceptance behavior of left-turning drivers to predict how a driver considers the position of the pedestrian [7]. A gap acceptance study was conducted to interpret drivers' overlooked behavior towards cyclists at roundabouts [15]. Different population and places had different critical gaps [16].

The main objective of this study is to evaluate how a driver moves into crosswalk with different pavement designs in the presence of pedestrians or cyclists on or near the crosswalk. Gap acceptance study is suitable for interpreting the changes in the behavior due to different crosswalk pavement designs.

workshop was held in Oslo, Norway, in 1977. An internationally recognized definition of traffic conflict is as follows:

"An observable situation in which two or more road users approach each other in space and time to such an extent that there is a risk of collision if their movements remain unchanged." [20]

Various conflict indicators (time to collision, post-encroachment time, deceleration rate, gap time, etc.) were developed to measure the severity of an interaction. The interaction between road users can be described as a continuum of safety-related events (Fig. 3). Based on the concept of the continuum of traffic events (Fig. 3), the Swedish traffic conflict technique was proposed by Hyden [21 ].

Svensson proposed a traffic safety process as an extension of the Swedish traffic conflict technique to study the relationship between traffic events in terms of safety and safety-related events [22]. By using this approach, it is possible to compare locations with different geometrical designs based on the distribution of safety-related events on the time to accident (TA)/conflicting speed (CS) graph (Fig. 4). Events of different levels of severity provided relevant feedback to the involved road users.

For understanding how it is likely to have conflicts on crosswalks, the Swedish traffic conflict technique was used in the present study to determine if the pavement design can influence the diversion of vehicles from a collision course to a normal course.

3. Methodology

2.2. Traffic conflict study

3.1. Study Area

For safety assessment, it is important to know how it is likely to have conflicts at conflict area, where road users travelling from different directions meet. The conventional method of traffic safety assessment mainly considered the occurrence of traffic crashes. Crashes are rare events, and not all crashes are reported [17,18]. The behavioral or situational aspects of events are not included in police crash data. For traffic safety measure, it is very important to understand the connection between behavior and safety. The traffic conflict analysis method is a surrogate safety-measuring tool for crash data analysis. The concept of traffic conflicts was first proposed by Perkins and Harris at General Motors Laboratory in the USA [19]. The first international traffic conflicts

The main purpose of this study is to analyze the effect of pavement design on the conflicts and unsafe behavior of left-turning vehicles when they interact with pedestrians. Potential locations were selected using Google Street View information. Thereafter, through actual site visits, the real features of the locations were observed. Three favorable locations for the study were chosen. The study area is located near Nishikawaguchi station in Japan. All the intersections are situated in a residential area. At each of these intersections, one urban road is

Fig. 3. The pyramid showing the interactions between road users as a continuum of events [21].

Fig. 4. TA/CS graph defining the different severity levels [22].

ARTICLE IN PRESS

4 H. Iasmin et al. / IATSS Research xxx (2016) xxx-xxx

Obseved view Geometric sketch

Fig. 5. Illustrations of intersections: (1) No pavement design, (2) red pavement design, and (3) brick pavement design.

intersected by three local residential roads (Fig. 5). All of these intersections are signalized. However, there is no separate signal phase for pedestrians. Pedestrians follow the same signal time as through vehicles. Left- and right-turning vehicle also share the same signal phase for completing their maneuver. Geometric and traffic characteristics and traffic signaling information is presented in Tables 1 and 2. The difference in pavement design with other geometric characteristics is shown in Fig. 5.

3.2. Safety performance analysis

According to the priority rule, drivers of left-turning cars should control their behavior when crossing the crosswalk if pedestrians are present. Hence, a safety performance study considering left-turning vehicle behavior was conducted. The analysis was based on gap acceptance study and conflict analysis of behavioral observations, based on observation of the movement of left-turning cars and pedestrians using video cameras. For conflict analysis, the Swedish traffic conflict technique was used. When driver reaches a crosswalk, he/she has to search opportunity to cross the crosswalk safely when a pedestrian is present. This opportunity is called the gap. If the gap is large, the driver can accept the gap easily. However, if the gap is small, the decision procedure becomes complicated. In this complicated situation, a wrong decision will lead to a collision. If driver does not take evasive action, he may collide with the pedestrian. Such a situation is called a conflict. To compare the three locations, it is important to measure the severity of these conflict events and their proper distribution. Fig. 6 shows the flowchart of the safety evaluation procedure adopted in this study.Details of the safety evaluation procedure used in this study are given in the following subsections.

3.2.1. Gap acceptance studies

Gap acceptance is commonly used to study the manner in which drivers move into a priority area where they must give way to other road users. In this study to understand the driver's tendency to give priority to pedestrian or cyclist on crosswalk gap acceptance study is used.

In this study, a gap is considered as an opportunity for a left-turning car to cross the pedestrian crosswalk during an interaction with a pedestrian. The terms used are defined as follows:

"A lag is the required time for a single pedestrian to reach the conflict

area."

"A gap is the time difference between two successive pedestrians

reaching the conflict area."

"The conflict area is defined as the area that is covered by a car on the crosswalk of an outflow road." This is so because all potential conflicts with pedestrians or cyclists occur within this area.

The gap is recorded at the point in time where the left-turning car driver decides whether he accepts or rejects the gap. Since precise determination of this point is very difficult, in this study, the decision point is assumed that when a driver reaches near the crosswalk of an inflow road. According to the definition of lag and gap, the time in seconds available to the left-turning car before a pedestrian arrives in the conflict area was measured (Fig. 7).

To estimate the gap acceptance probability distribution, empirical data were collected. Thereafter, the gaps and lags were

ARTICLE IN PRESS

H. Iasmin et al. / IATSS Research xxx (2016) xxx-xxx

Fig. 6. Flowchart of the safety evaluation procedure in this study.

divided into several bins of size 1.0 s owing to the limited sample size. The acceptance probability for each bin can be calculated by using Eq. 1.

P(x) =

No. of observed accepted gaps/lags No. of observed accepted and rejected gaps/lags

To analyze the left turn gap, the acceptance logistic regression method was used [23]. Logistic regression was found to be appropriate for modeling a situation in which drivers have numerous opportunities where they have to take the yes/no decision. In this method, the critical gap is defined as the median of the accepted gaps that are accepted by 50% of the drivers. The logistic regression model is represented as follows:

P(x) =

1 e-bc -b1x

Fig. 7. Definition of gap.

Where P(x) is the probability of accepting a gap or lag, x; b0 and b-i are intercept parameter and slope parameter, respectively.

3.2.2. The Swedish traffic conflict technique

The Swedish traffic conflict technique by following the methods proposed by Svensson was used [22]. A conflict presumes a collision course.

ARTICLE IN PRESS

Fig. 8. Camera calibration procedure by Kinovea.

Fig. 9. An interaction between a left-turning vehicle and pedestrian, and the detected vehicle motion parameters a: Tracking an object b: Video screenshot with trajectory c: Moving trajectory of a left-turning vehicle and pedestrian d: Speed of a left-turning vehicle and pedestrian.

ARTICLE IN PRESS

H. Iasmin et al. / IATSS Research xxx (2016) xxx-xxx

Table 3

Observation samples used to analyze left turn gap acceptances.

Measurement No pavement design Red colored Brick pavement

Accepted gap/lag 19 26 14

Rejected gap/lag 17 18 16

"Conflict: A conflict is either an event that would have led to a collision if both road-users had continued with unchanged speeds and directions or a near-miss situation where at least one of the road-users acts as if there were a collision course."[21]

"Collision course: Unless the speed and/or the direction of the road-users changes, they will collide." [22]

The severity distribution of the conflicts was analyzed by using the TA/CS graph (Fig. 4). In this graph, different events with different severity levels indicate different intensities of severity (see Table 5) [24]. It is presumed that all events with a TA/CS value are parts of the same traffic safety process as the crashes and serious conflicts. The definition of the severity of a conflict is based the two variables: TA (Time to accident) and CS (conflicting speed), implying that each event related to the collision course are represented using these two variables.

TA is the time that remains from one of the road users have started an evasive action, until a collision would have occurred if the road users had continued with unchanged speeds and directions.

CS is the speed of the involved road user at the moment when the evasive action (i.e., braking) starts. TA is estimated for this CS.

If road users are not in a collision course, there is no possibility of an accident. Hence, the pre-condition of collision course was assumed when an evasive action was taken by a road user. As pedestrians change their speeds continuously, it is difficult to know the point where they start taking evasive action. Hence, only the events in which evasive action was taken by the driver were considered in the analysis. Basically, five different types of evasive action can be taken for avoiding collisions:

i. Braking

ii. Swerving

iii. Accelerating

iv. Braking + Swerving

v. Accelerating + Swerving

If a driver took any of these evasive actions, the events were recorded. By using the speed and distance data obtained using the video analyzing software, chances for collision if drivers did not change their current speed were determined (Fig. 8). If the road-users were in a collision course, the event was counted as an interaction.

3.3. Data collection

Data of gap acceptance and conflicts and unsafe behavior were acquired by using video cameras installed on high-rise buildings. In Japan, it is very difficult to put camera anywhere because of strict rules and regulations. After selecting the potential site, the most important task was to search for a location for setting up the video cameras.

Table 4

Statistical parameters of the fitted lag/gap acceptance probability distributions.

intersection Regression coefficient Estimate Standard error R2

No pavement b0 -4.909 1.658 0.734

b1 1.828 0.633

Red colored b0 -8.494 3.336 0.887

b1 2.543 1.004

Brick pavement b0 -9.745 4.271 0.841

b1 2.232 0.980

100% 90%

■S 50%

30% 20%

/ /. /

/ . ► /

/ ; /< ►

-1 # t ♦ no pavement design ♦ red coloured ♦ brick pavement — — — - no pavement design -brick pavement

__j ✓

10 11 12

Gap size [sec)

Fig. 10. Predicted probability that drivers accept gaps to take a left turn through a crosswalk.

Permission from three building owners near the location was obtained to use their upper floor stair cases to install video cameras. The video observation view of each intersection with the studied crosswalk (red circle in Fig. 5) can be seen in a screenshot of the recorded video (see Fig. 5). The camera continuously recorded traffic scenarios from each observation site from 9.00 am to 4.00 pm for five consecutive weekdays in December 2014 and January 2015. It was winter season, and the weather was sunny and clear.

3.4. Data processing

In this study, owing to small sample size, only cars are considered. Following left-turning cars in a platoon are not considered in the data processing. To study gap acceptance decision and severity of evasive action, these events are selected for data extraction when a car reaches near the crosswalk of an inflow road where pedestrians are present (Fig. 7). In each event, the behaviors of drivers when pedestrians were present near the crosswalk were observed to determine if they accepted or rejected the gap or took any evasive action.

All data, including the positions and timings of left-turning cars and pedestrians and cyclists were extracted from video data using video analyzing software Kinovea. Kinovea is a free and open source (GPL2) French software created in 2009 as a tool for movement analysis [25]. This software is mainly used for sport analysis. Using parameter calibration by using the geometric data of the marking lines, Kinovea can calculate motion parameters such as position, speed, and acceleration of sports cars, athletes, and players. Hence, it is possible to use Kinovea for traffic study. The Butterworth filter was used for filtering data in Kinovea. Fig. 8 shows an outline of the detailed description of the video-analyzing procedure.

100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

7 B-Ï B—s l-l 1 f i

A esign oloured

■ red

i / * brick pavement

f r» f

0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 Speed of left turn vehicle at conflict area (km/hr)

Fig. 11. Cumulative distribution of observed vehicle speed in the conflict area for different pavement designs.

ARTICLE IN PRESS

H. Iasmin et al. / IATSS Research xxx (2016) xxx-xxx

0123456789 10 11 12 TA -value

Fig. 12. Severity of events when a car brakes [22].

Fig. 9a and b shows a screenshot of real interaction between the left-turning vehicle and pedestrian. In this interaction, the driver rejects the gap with an evasive action (braking). If he had not taken this action, a collision may have followed. Fig. 9c and d shows the motion parameters of this interaction. From these motion parameters gap time, speed at conflict area, TA, and time to collision are extracted.

4. Results and discussion

4.1. Left-turn gap acceptance behavior

Owing to low traffic volume, only cars were considered for analyzing the gap acceptance of left-turning vehicle. From the dataset, a total 110 individual gap decisions were recorded, where 36 observations corresponded to no pavement design; 44 observations, to red pavement, and 30 observations, to brick pavement design (Table 3). As stated in the methodology, the observed lags and gaps were divided into bins of 1.0 s (0-1 s, 1.1-2 s, 2.1-3 s, 3.1-4 s, 4.1-5 s, 5.1-6 s, 6.1-7 s, 7.1-8 s, 8.1-9 s). The acceptance probability can be calculated by using Eq. 1. Logistic regression was used to fit these plots. As an indicator of model fit, the mean values of Nagelkerke's R2 for each individual regression model was reported for each analysis [26].

Table 4 lists the estimated values of statistical parameter from logistic regression analysis. The estimated values of slope parameter b1 for without any pavement design are smaller than those for red colored pavement design and brick pavement design. This indicates that a driver at an intersection with red color and brick pavement has significantly lower probability of accepting a smaller gap than at a typical intersection without any design. This can be attributed to the visual effect of red color and brick pavement on the driver. When a driver sees a different design, he/she may think more about crossing and become careful. Therefore, the driver may notice the pedestrian at an early stage of his approaching the crosswalk. Fig. 10 shows the predicted probability of gap acceptance by a left-turning driver based on regression results. It shows that a driver at a brick pavement crosswalk has significantly lower probability of accepting a smaller gap. Comparatively probability of accepting smaller gaps is lower for a driver at a red pavement crosswalk than at a typical intersection.

4.1.1. Vehicle speeds at conflict area

A crosswalk is a shared space for pedestrians and left-turning vehicles. In a shared space, the vehicle speed is one of the most important parameters for safety evaluation. The risk of the causality of crashes increases with increasing free travelling speed [27,28]. It is rational to use the vehicle speed at the conflict area as an indicator of the severity of the conflict. For three intersections, the cumulative distributions of observed vehicle speed at conflict area after accepting a specific lag or gap are shown in Fig. 11. The cumulative distributions are developed by dividing the speeds into bins of 2.0 km/h size (0-2 km/h, 2.1-4 km/h, 4.1-6 km/h, 6.1-8 km/h, 8.1-10 km/h, 10.1-12 km/h, 12.1-14 km/h, 14.1-16 km/h, 16.1-18 km/h, 18.1-20 km/h, 20.1-22 km/h, 22.1-24 km/h, 24.1-26 km/h). A t-test was performed to check the significance of the difference between speed distributions at 95% confidence level. The corresponding t-values are shown in Fig. 11.

For the gap acceptance study, drivers at crosswalks of red and brick pavements tend to accept larger lags and gaps than the drivers at crosswalk without any pavement design. Fig. 11 shows clearly that the speed of a left-turning car at a conflict area of a crosswalk with any pavement design is significantly higher than the two treatments when accepting gaps and lags. The 85th percentile speed of left-turning cars on crosswalks with red and brick pavements is lower than the speed at crosswalks without any pavement design. Under such conditions, it is clear that presence of red or brick pavements on a crosswalk decreases the severity of interaction. The results of the t-test showed that the difference between speeds was not significant for the red and brick pavements. However, Fig. 11 shows that 7.6% drivers at crosswalks with red pavements and 21.4% drivers at crosswalks with brick pavements travelled through the conflict areas with speed less than 10 km/h when they found pedestrians near the crosswalk. For decreasing the speed at when the gap, the brick pavement is more efficient than the red colored pavement.

4.2. Conflict study result by TA-CS graph

A total of 44 observations (no pavement design =17; red pavement = 13; brick pavement = 14) was selected for the analysis. Left-turning drivers took evasive action to avoid collisions with pedestrians. For each event, TA and CS were measured.

Fig. 12 shows the plots of the conflicts with measurable TA and speeds of left-turning vehicles at the time of evasive action at three studied sites. Table 5 summarizes the frequency of interactions between left-turning vehicles and pedestrians for different severity levels. With respect to the TA-CS graph or Swedish traffic conflict technique, events at level 26 or higher are more dangerous. They consist of serious conflicts and accidents. Table 5 shows that there are no serious conflicts in any intersections. As the traffic volume is low and road size is small, the seriousness of conflicts is very low. The severity of the events in the severity level 20-25 is fairly high. They are very close to accidents. At these levels, the safety margin is very small, which may change the situation on the road to a critical event. If drivers are not careful, these types of situation may change to an accident easily. The typical intersection without any pavement design has 15 events (88.23%) in this level, indicating a fair chance of accidents. In this typical intersection, when drivers and pedestrians move in a collision course, they always surprise

Table 5

Observation of conflicts at three sites and the distribution of conflicts with regard to the type of conflict and severity level.

Severity level

Intensity

Remark

No pavement design

Red-colored pavement

Brick pavement

26-30 20-25

19 or less

Highest severity Consists of serious conflicts and injury crash

Fairly high severity Strong closeness to an unsafe dimension

Lower severity Danger is not very imminent. Risk is low

(88.23%) 2

(11.77%)

(57.14%) 5

(42.86%)

(28.57%) 10

(71.43%)

ARTICLE IN PRESS

H. ¡asmin et al. / ¡ATSS Research xxx (2016) xxx-xxx

each other. They are very close to serious events. For intersection with red pavement, the number of fairly high severity events is relatively high (57.14%). It seems that the drivers at this intersection took the decision for evasive action very late. When a driver reaches the crosswalk, he/she may not stop without confirming the pedestrian position. Hence, the evasive action was delayed, and the remaining distance became less. Crosswalks with brick pavements had very few events (28.57%) with fairly high severity, and most of the interactions (71.43%) were in the lower severity level. In other words, using brick pavements on crosswalks is a very good way of converting collision courses to normal road situations at early stage. When a driver reaches the crosswalk, before confirming the presence of a pedestrian, he/she slowed because of the road design. Because bricks on a crosswalk may give a driver a different visual effect that make him/her to understand that it is a pedestrian area. This makes a driver careful, and take evasive action a very early stage of interaction. Theoretically, left-turning vehicle and pedestrian will collide if they do not change their speeds, but the intensity in very low.

5. Conclusion

This work investigated the safety performance of pavement design on crosswalks when left-turning vehicles interact with pedestrians on crosswalks. Crosswalks are very crucial areas for road users, especially pedestrians who are the most vulnerable road users. Hence, the safety of pedestrians at intersections should be ensured. This was the main research question of this study was whether a pavement design can enhance pedestrian safety. Two intersections with different pavement designs and one typical intersection without any design on the crosswalk were selected. Gap acceptance study and conflict study were performed for safety evaluation. The key conclusions of this study are summarized here:

The result of the gap acceptance study confirmed the positive effect of using red color or brick pavements. The results show that at a crosswalk without any design, drivers accept smaller gaps than in the case of the other two crosswalks. When drivers clear the conflict area of a typical crosswalk, they do so at relatively high speed. In the case of red design and brick pavements, drivers accept small gaps less frequently. The visual effect of design on crosswalks may give drivers more time to think before turning. Hence, drivers reject the small gaps.

The conflict study endorses the positive effect of brick pavement design. At typical crosswalks, the frequency of fairly high severe conflicts is more. This means that drivers take evasive action at the last moment. This tendency may increase the chance of accidents. If the timing of taking evasive action is wrong, these events cause very serious accidents. However, conflict data for red pavements also show that the frequency of fairly high severities is not very low. Drivers take evasive action at a later stage. Conflict study of the brick pavement design confirmed that it is safer to use brick on crosswalks. In the case of this intersection, when left-turning cars and pedestrians move in collision course, the event has very low severity. Brick design on crosswalks make driver more alert at an early stage of entering the crosswalk. Hence, the driver takes evasive action very fast. The remaining distance to collide becomes small and TA also increases.

Thus, brick design on crosswalks increases the alertness of left-turning vehicles. Red pavement also has a positive effect on increasing the safety of pedestrians. However, the lowering capacity of conflict severity is more in the case of brick pavement design. Thus, it can be concluded that brick pavements on crosswalks is suitable for pedestrian safety at intersections. Pavement design effect on right-turning vehicle (left-hand traffic system), safety performance of other type of pavement design, and visual effect of pavement design on driver can be explored in future studies.

Acknowledgement

This study was conducted under the project "ÎS^EÂÎTÎil & № t M

(Principal Investigator: Ayako Taniguchi, Tsukuba University), supported by The General Insurance Association of Japan.

References

Institute for Traffic Accident Research and Data Analysis, ITARDA information, Japan, http://www.itarda.or.jp/itardainfomation/english/info100_e.pdf2013. S. Turner, K. Fitzpatrick, M. Brewer, E. Park, Motorist yielding to pedestrians at unsignalized intersections: findings from a national study on improving pedestrian safety, Transp. Res. Board 1982 (2007) 1-12.

V. Himanen, R. Kulmala, An application of logit models in analysing the behaviour of pedestrians and car drivers on pedestrian crossings, Accid. Anal. Prev. 20 (3) (1988) 187-197.

W.A. Harrell, The impact of pedestrian visibility and assertiveness on motorist yielding, J. Soc. Psychol. 133 (3) (1993) 353-360.

D. Sun, S.V.S.K. Ukkusuri, R.F. Benekohal, S.T. Waller, Modeling of motorist-pedestrian interaction at uncontrolled mid-block crosswalks, Proceedings of the 82nd TRB Annual Meeting, Washington, 2003.

L. Leden, Pedestrian Risk Decrease with Pedestrian Flow: A Case Study Based on Data from Signalized Intersections in Hamilton, Ontario, Accid. Anal. Prev. 34 (2002) 457-464.

W.K.M. Alhajyaseen, M. Asano, H. Nakamura, Left-turn gap acceptance models considering pedestrian movement characteristics, Accid. Anal. Prev. 50 (2013) 175-185. H. Iasmin, A. Kojima, H. Kubota, Yielding behavior of left turning driver towards pedestrian/cyclist: Impact of intersection angle, J. East. Asia Soc. Transp. Stud., December 2015 (In Press, Corrected Proof, Available online 27).

V.G. Stover, Issues relating to the geometric design of Intersections, Proceedings of the 8th international Conference in ACCESS Management, Baltimore, MD, USA, 2008.

V.G. Stover, F.J. Koepke, Transportation and Land Development, second ed. Institute of Transportation Engineers, Washington DC, USA, 2002.

R.J. Troutbeck, W. Brilon, Unsignalized intersection theory, Turner-Fairbank Highway Research Center, Virginia, 1992.

L. Tijerina, J.D. Chovan, J. Pierowicz, D. Hendricks, Examination of signalized intersection, straight crossing path crashes and potential IVHS countermeasures (Final report DOT-VNTSC-NHTSA-94-1), Columbus, OH: Battelle, July 1994. H. Preston, R. Storm, M. Donath, C. Shankwitz, Review of Minnesota's rural crash data: Methodology for identifying intersections for intersection decision support (IDS), Rep No. MN/RC-2004-31, Minnesota Department of Transportation, St. Paul, MN, 2004.

N.M. Rouphail, E.I. Schroeder, J. Bastian, Event-Based Driver Yielding and Pedestrian Gap Utilization Models for Micro-Simulation Applications, North Carolina State University, 2008.

M.-B. Herslund, N.O. Jorgensen, Looked-but-failed-to-see errors in traffic, Accid. Anal. Prev. 35 ( 2003 ) 885-891.

A. Moshe, Pollatschek P. Abishai, L. Moshe, A decision model for gap acceptance and

capacity at intersections, Transp. Res. B 36 (2002) 649-663.

A. Englund, N.P. Gregersen, C. Hydén, P. Lovsund, L. Aberg, Trafiksakerhet - En

kunskapsoversikt, KFB & Studentlitteratur, Lund, Sweden, 1998.

G.B. Grayson, A.S. Hakkert, Accident Analysis and Conflict Behaviour, in: J.

Rotherngatter, R.A. de Bruine (Eds.), Road-user and Traffic Safety, Van Gorcum

1987, pp. 27-59.

S.R Perkins, J.I. Harris, Traffic conflict characteristics: Accident potential at intersections, Highw. Res. Rec 225 (1968) 45-143.

F. Amundson, C. Hydén, Proceedings of the First Workshop on Traffic Conflicts, Institute of Economics, Oslo, 1977.

C. Hydén, The development of a method for traffic safety evaluation: The Swedish traffic conflicts technique, Department of Technology and Society, Lund University, 1987.

A. Svensson, A method for analysing the traffic process in a safety perspectiveDoctoral thesis University of Lund, Lund Institute of Technology, Department of Traffic Planning and Engineering, 1998.

A. Agresti, An Introduction to Categorical Data Analysis, Wiley, New York, 1996. A. Svensson, C. Hydén, 2006. Estimating the severity of safety related behavior, Accid. Anal. Prev. 38 (2006) 379-385.

Kinovea, 0.8.23 Experimental version; Joan Charmant and Contrib, http://www. kinovea.org.

N. Nagelkerke, A note on a general definition of the coefficient of determination, Biometrika 78 (1991) 691-692.

C.N. Kloeden, G. Ponte, A.J. McLean, Travelling speed and the risk of crash involvement on rural roads, Report CR 204, Australian Transport Safety Bureau ATSB, Civic Square, ACT, 2001.

G. Nilsson, The effects of speed limits on traffic accidents in Sweden, Proceedings of the international symposium on the effects of speed limits on traffic accidents and transport energy use, Dublin, OECD, Paris, 1982 (1981).