Abstract
The aim of this research is to determine the effects and analysis the left turning accident rate in urban road between different intersections.留學生dissertation網
Previous studies found that there are particular parameters that had positive and negative effect to the accident rate.
Different accident rate analysis models discussed during researches.
The main methodology is according to the literature study about different left turning accident models summarize the various parameters. And then collection data from real situation in xxxxx urban road intersections to analysis it by SafeNET to determine which parameters for accident rate is positive and which one is negative. End up with parameters changing to define my own left turning accident rate analysis model.
The expectation of my own model is to use it to calculate the left turning accident rate for different kinds of intersections in different area (hopefully at the end I can achieve this purpose).
1 Introduction
1.1 Introductions and background
The important of road safety is becoming a hot topic during these years, especially in the developed countries. With the higher living standard more and more people starting concern the public safety problem. According to the data which is provided by European Road Statistics 2008[1], in 1990 the number of road accidents involving injury, for EU 15 is 1,342,900, then decrease to 1,13,7900 until 2006. From the figure maybe we will say this number is not big, but according to the statistic data from Word Bank in 2002, there are 1,170,000 people got killed each year by accident [2].
Based on the data from European Road Statistics 2008[1], in EU27 in 1991, the road fatalities are all most 2,000,000 and the road injuries are all nearly 1,900,000. But in 2006 the road fatalities are 42,000, while the road injuries are 28,000. As we can see, during this 15 years the road fatalities and injuries decreased remarkably, the reason the just because European unit put more and more attention on the problem of road safety.
The addition of a left-turn lane can improve the operations and safety at an intersection. Guidelines as to when to include a left-turn lane in intersection design are plentiful. Because of the quantity of methods, questions are asked regarding which method to use. This paper reviewed eight selected techniques and a number of criteria present in state manuals. Methods based on delay typically do not recommend a left-turn lane at lower left or through volumes when compared to methods based on conflict avoidance or safety. Because of the high benefits for crash reductions provided by left-turn lanes, a method that results in a recommendation at lower volumes would be preferred. The Harmelink model is a widely accepted approach that is based on conflict avoidance. The procedure was first proposed in 1967 and includes assumptions that may need to be revised. Findings from current research would suggest a critical gap of 5.5 sec (rather than 5.0 sec), a time to make left turn of 4.3 sec (rather than 3.0 sec), and a time to clear the lane of 3.2 sec (rather than 1.9 sec). A table was developed that lists suggested guidelines for installing left-turn lanes for operating speeds of 30, 50, and 70 mph (50, 80, and 110 km/h).#p#分頁標題#e#

And the general idea of left-turning guideline, which is the general useful method to determine the left turning lanes to achieve safety and effective purpose.
Left turning traffic situation for signalized and unsignalized intersections.
Left turning accidents between vehicles to vehicles in signalized intersection
Left turning accidents between vehicles to pedestrians in unsignalized intersection
Background of SafeNET what it can do and what we are going to do it with this research
An accident rate analysis tool which is can be used to determine which specific parameters and data affect the accidents rate by numeric level.
During our research we just focus on the left turning accident rate analysis try to base on real data to summarize a general model for vehicle to vehicle left turning accidents rate model by parameters changing in SafeNET anasysis.
1.2 Purpose
The main purpose of this research is to determine the main effects parameters of left turning accident rates and analysis it in urban road between signalized and unsignalized urban intersections. Analysis the characteristics for vehicle to pedestrians accidents model which include human factors and geographic factors. Then come up with a general model specific for vehicle to vehicle accident rate.
1.3 Methodology
The objectives of this research is to find out the effects parameters of left turning accident rate, analysis the general factors and end up with a model which can be used popularly to analysis vehicle to vehicle accident rate.
To achieve these objectives first of all a literature review which is study the current researches and the previous literature inside of this area to see which is the main parameters and factors for left turning accidents, then analysis characteristics of them, all these analysis research is focus on vehicle to pedestrians accidents model. This model is contains too many human factors which are vary from different road users and even same road users under different situation or condition all of these subjective and objective factors are all unpredictable problems. So it is impossible to analysis it by some particular mathematics models.
Compare with vehicle to pedestrians accidents situation, the vehicle to vehicle situation can be seen a more stable and fixed model. It can be simulated and analsized by different accident rate analysis tools. During this research SafeNET is used to analysis the parameters and then come out with a mathematics model utilized by vehicle to vehicle accident rate analysis.
Field data collection and modification is need for implementation in SafeNET then we need validation and calibration for the simulation and analysis to determine the final useful accident rate mathematics model.
1.4 Outline
During this research there are five main parts which are literature review, introduce SafeNET and the data collection & modification for different chosen scenario sites, implementation of SafeNET, validation and calibration result and then end up with conclusion & possibility of future research area. This outline describes the contents of each upcoming chapter.#p#分頁標題#e#
Chapter2: Literature Review
In the literature review chapter, the basic principles of left turning lane design, the general factors of left turning accident and characteristics of vehicle to pedestrians’ left turning accident compare with vehicle to vehicle accident are studied. Finally summarize the main factors and parameters which are effect and cause the left turning accidents.
Chapter3: Data Collection & Modification for Chosen Scenario Sites
Background, applications, advantages and disadvantages analysis of SafeNET compare with current other accident rate analysis tools are described in this chapter. What are the requirements data for analysis different type of intersection’s left turning accident and how these data can be collected are introduced as well. At the end, pick out the obviously incorrect data from the collected data for chosen scenario site.
Chapter4: Implementation of SafeNET
During implementation of SafeNET, the data collected from Norrkoping city two of signalized intersections and one of signalized roundabout, and two of unsignalized intersections will be input into SafeNET to analysis the different intersection accidents rates. After analysis by SafeNET, some of parameter will be modified by ideal principle which will be changed by manually. To check out will kind of parameter will affect the accidents rate significantly, and then give the feasible recommendations.
At the same time, the data will be used in the following chapter which is going to analysis all of the data and then come out with a mathematic model. For building the mathematic model plentiful data will be need. During this research the main data is   collected in Norrkiping intersections and Beijing intersections.
Chapter 5: Validation and Calibration Result
The most important thing in this part is by analyzing the plentiful data and selecting the proper mathematic method to come out the mathematic model. And then validate and calibrate the different parameters to determining which parameters are the positive affect and which are the negative parameters. The purpose of this analysis is to get the clear idea which kind of parameter will affect the accidents rate significantly. Then we can provide the accurate recommendations when building an intersection or provide the feasible suggestions when rebuild an unsafety intersection.
Chapter 6: Conclusion and Possibility of Future Research Area
In the last chapter, the results of analysed left turning accident rate in Norrkoping city’s intersection will be provided and the feasible commendations will also be provided. At the same time, the mathematic model will be summed up, which is a general model can be used in each kind of intersections, especially in the big cities where there are a lot of traffic volume. Because the data we built the model is based on Beijing city’s intersection.

WHAT IS THE PROBLEM
WHY IS LEFT TURNING IS A PROBLEM
HOW SERIOUS ABOUT THE PRO IN DIFFERENT COUNTRIES by stacticics#p#分頁標題#e#
HOW CAN RESERACHERS HAVE ANALYSIS THE PRO JOURSNERS PAPERS
WHAT WHERE THE PERIOUS SOLUTIONS FROM VEHICLES DRIVERS NEW ITS APPLICATIONS
THEN YOUR APPROUCH AND UNDERSTANDING IN SOVLING THIS PROBLEM
THEN MAKE A NEW MODEL
Check list for analysiing the left turning problem road safety audit

現在的dissertation要求比較簡單 也比較單一， 整體的思想是圍繞著 左轉交通事故發生率比較高來的，導師的要求也很單一，

定義什么是左轉交通事故 左轉交通事故的一些背景
為什么左轉交通事故發生率比較高，為什么左轉交通事故在全世界是一個大的難題
分析不同國家 左轉交通事故的嚴重性， 此章需要有大量的數據作為支持
根據閱讀的文獻，總結之前的研究者在此領域所做的研究，所對此問題的分析
從車輛的角度，行人的角度分析 此問題的之前的和現在的解決方案，同時分析 如何 應用智能交通的理論 解決此問題
根據前面的研究，給出本人對此問題的理解 以及解決方案，從而建立一個新的模型區解決此問題
中間可以生成一個CHECK LIST 根據交通安全審計的理論（此部分如果不明白我自己可以寫）

2Literature Review
2.1 Principles of left turning lane design
It is well know that for an intersection, if the left turning pass through frequency is high then the level of service of this intersection will be affected at the end the whole road’s level of service will be severely influenced. At the same time, accidents rates in the intersection also will increase with increased left turning pass through frequency compare with directly through traffic part. So the left turning traffic is become an significant part if require safety and effectively road environment. [1]
Left turning lane guidelines 
The addition of a left-turn lane can improve the operations and safety at an intersection. [2] Because of the high benefits for crash reductions provided by left-turn lanes, a method that results in a recommendation at lower volumes would be preferred. The Harmelink model is a widely accepted approach that is based on conflict avoidance. The procedure was first proposed in 1967 and includes.
A recent FHWA study found that the addition of a left-turn lane can result in reductions of crashes from 7 to 48 percent. Other studies have demonstrated the benefits of delay reductions with the installation of a left-turn lane on two-lane highways. Guidelines as to when to include a left turn lane in intersection design are plentiful. Some are based on minimizing conflicts in terms of the occurrence of a through vehicle arriving behind a turning vehicle; others are based on decreasing the amount of delay to through vehicles, and some are based on consideration of safety. Because of the quantity of methods, questions are asked regarding which method to use. For example, are certain techniques better for a rural versus an urban setting? Do the evaluations differ for number of lanes and for type of intersection? This paper reviewed eight selected techniques and a number of criteria present in state manuals. Some of the assumptions used in the techniques will be reviewed, and suggestions on changes to selected guidelines will be made.#p#分頁標題#e#

Common terms are used in several of the techniques. Figure 1 graphically shows the following movements that are used to determine the need for a left-turn lane in several of the guidelines:
Advancing Volume (VA) – the total peak hourly volume of traffic on the major road approaching the intersection from the same direction as the left-turn movement under consideration.
Left-Turn Volume (VL) – the portion of the advancing volume that turn left at the intersection.
Percent Left Turns (PL) – the percentage of the advancing volume that turn left; equal to the left-turn volume divided by the advancing volume (PL = VL ÷ VA).
Straight Through Volume (VS) – the portion of the advancing volume that travel straight through the intersection (VL + VS = VA).
Opposing Volume (VO) – the total peak hourly volume of vehicles opposing the advancin   Left Turning Volume (VL)   Opposing Volume (VO)
       Advancing Volume (VA)                     Straight Through Volume (VS)

 Figure1. Volumes for Use in Left-Turn Lane Warrant Methods

In 1985, the Transportation Research Board published NCHRP Report 279, Intersection Channelization Design Guide. In that report, data from Harmelink’s work were used to establish guidelines for determining the need for a left-turn lane. The following advice was provided for unsignalized intersections:

Left-turn lanes should be considered at all median cross-over on divided, high-speed highways.
Left-turn lanes should be provided at all unstopped (i.e., through) approaches of primary, high-speed rural highway intersections with other arterials or collectors.
Left-turn lanes are recommended at approaches to intersection for which the combination of straight through, left, and opposing volumes exceeds the warrants.
Left-turn lanes on stopped or secondary approaches should be provided based on analysis of the capacity and operations of the unsignalized intersection. Considerations include minimizing delays to right turning or through vehicles and total approach capacity.
2.2 DEVELOPMENT OF LEFT-TURN LANE GUIDELINES FOR SIGNALIZED AND UNSIGNALIZED INTERSECTIONS
It is generally accepted that the level of service (LOS) at intersections significantly affects the overall LOS of the road system. It is also known that the LOS at an intersection can be adversely affected by frequently allowing left-turning vehicles to block through traffic. In addition, crash rates tend to be higher at intersections than on through sections of a road. The separation of left-turning vehicles from through traffic is therefore an important condition for the safe and effective operation of intersections.

Existing guidelines for installing left-turn lanes have several limitations. They are mainly based on the traffic volumes at the intersection, and they use deterministic models with fixed gap acceptance and/or left-turn maneuver times. In addition, the guidelines for left-turn lanes for unsignalized intersections and signalized intersections must be specific for the type of intersection.#p#分頁標題#e#

In this study, new left-turn guidelines for both unsignalized and signalized intersections were developed on the basis of well-validated event-based simulation programs. Guidelines for unsignalized intersections were based on the percentage of left turns blocking through vehicles, whereas the guidelines for signalized intersections were developed using a minimum left-turn volume of either 85% left-turn capacity or LOS E delay (55 seconds/vehicle). In addition to the general guidelines, a prioritization tool that can be used to prioritize candidate intersections was developed. The prioritization tool accounts for both operational and safety aspects.

The purpose of this study was to develop left-turn lane guidelines for signalized and
unsignalized intersections. Specific objectives were as follows:
Critically evaluate existing guidelines for providing left-turn lanes at unsignalized and signalized intersections through a literature review.
Determine how engineers at the Virginia Department of Transportation (VDOT) use existing guidelines.
Develop, calibrate, and validate event-based simulation models to simulate traffic at
unsignalized and simple two-phase signalized intersections.
Develop new guidelines for left-turn lanes using the event-based simulation models that are applicable for all turning percentages.
Create a prioritization system for ranking the need for left-turn lane improvements among a number of candidate intersections.
 
This study was limited to developing guidelines for installing left-turn lanes at signalized and unsignalized intersections. Developing guidelines for left-turn signal phasing was not within the scope of this study. Although each event-based simulation program used for developing guidelines in this project could theoretically simulate any condition seen in the field, the scope of each program was limited to ensure that program coding could be completed in a timely manner.
Thus, the scope of the program was defined as follows:
A maximum of four approaches can be simulated using this program.
The program can handle all kinds of lane sharing and also accommodate up to four lanes on a given approach.
Unsignalized intersections on undivided highways are considered for only left-turn lane guidelines. VDOT.s Road Design Manual (2002) dictates that left-turn lanes are to be provided for traffic on non-access controlled divided highways.
The signalized intersection program can simulate actuated intersections but assumes simple two-phase signals.
The model is applicable only for left-turn lane analysis and cannot be used for rightturn lane analysis.
At signalized intersections, left-turns on the subject link must be permitted.
The effect of available sight distance and grades on the approaches is not examined explicitly.
The signalized program is a decision making tool for installing a left-turn lane and should not be used to determine left-turn phasing.#p#分頁標題#e#

The LTGAP program developed in this study does not fully implement multi-phase actuated signal controls. An installation of a left-turn lane at a signalized intersection could further improve its operation by implementing an exclusive left-turn phasing. It is recommended that further research be conducted for developing left-turn phasing guidelines.
If the proposed guidelines are accepted by VDOT, the interface of the prototype LTGAP program should be enhanced for professional use. Especially, the prioritization tool demonstrated in the project needs to be further incorporated into LTGAP.
The safety surrogate measure used in this study was solely based on conflict opportunities. Future research should quantify the extent to which conflict opportunities can predict crashes such that the impact of safety can be better incorporated into the prioritization tool.
2.3 Modeling left-turn crash occurrence at signalized intersections by conflicting patterns
Analyzing left-turning traffic is crucial for improving intersection operation and safety. This traffic may collide with many other traffic flows at signalized intersections, e.g., with leftturning traffic from the same approach or from the different approaches, and with through traffic from other approaches; therefore, left-turn crashes have many distinct conflicting patterns in vehicle maneuvers before collisions. Left-turn crashes are among the most frequently occurring collision types; based on the crash history of 1531 signalized intersections in the state of Florida, left-turn crashes rank third; following rear-end and angle crashes, and represent 16% of all intersection reported crashes (Abdel-Aty et al., 2005). At signalized intersections, differences exist in traffic volumes, site geometry, and signal operations, as well as safety performance on various approaches of intersections. Therefore, modeling the total number of left-turn crashes at intersections may obscure the real relationship between the crash causes (i.e., intersection characteristics, etc.) and their effects (i.e., leftturn crashes, etc.). However, at signalized intersections, traffic flows and signal operations of different approaches are interactive at intersections; therefore, disaggregating intersections into approaches will introduce correlation among observations from the same intersection. In this study, left-turn crashes were investigated at the approach level by conflicting patterns based on geometry and traffic-related explanatory variables using the appropriate statistical models, which are able to analyze correlated crash frequencies.
2.3.1 Factors affecting left-turn crash occurrence
Several studies have attempted to quantify the effects of traffic flow, intersection geometric design features, and traffic control and operational features on left-turn crash occurrence.
On signalized intersection approaches, left-turning traffic can be treated in one of three ways: “permissive”, “compound” (“protected/permissive” or “permissive/protected”), and “protected” and numerous studies have been conducted for evaluating their safety effect#p#分頁標題#e#
It has been found that as the width of a median increases, the sight distance for left-turning vehicles decreases significantly. For unprotected left-turn traffic at signalized intersections, vehicles turning left from opposing left-turn lanes often restrict each other’s sight distance pointed out that the minimum value of the offset can be zero, but cannot be a negative value in practical design, which would results in unsafe conditions.
2.3.2 Crash modeling strategy
Since at signalized intersections traffic flows and signal operations of different approaches are interactive, disaggregating an intersection into four approaches may produce correlation among the data. There are serious problems arising when basic count data models are used for correlated data, since basic count data models assume the dependent variables are independent.
For a model at the approach level, the data structures become a mixture of randomly selected intersections and highly correlated approaches within an intersection. Generalized Estimating Equations (GEE) provide an extension of generalized linear models (GLMs) to the analysis of correlated data, which can account for the correlation among the data. In addition, since crashes are rare events, disaggregating left-turn chttp://www.ttc-switch.com/dissertation_writing/rashes by collision pattern and by approach may lead to a high proportion of zeros and ones in some collision patterns. GEE with a binomial logit link function can be attempted for the patterns with fewer crashes.
2.3.3 Research objective
The primary purpose of this study is to investigate the relationship between left-turn crash occurrence and intersection features, i.e., geometry design features, traffic control and operational features, traffic flows, etc. The left-turn crashes were divided into different conflicting patterns based on the vehicle maneuvers before collisions, and then assigned to the approach with left-turning vehicles. GEEs, which provide an extension of generalized linear models (GLMs) to analyze correlated data, were applied for separate left-turn crash models to account for the correlation among the data.
2.3.4 Data collection
Information on intersection geometry design features, traffic control and operational features, traffic flows, and crashes from 2000 to 2005 were obtained for 197 four-legged signalized intersections from Orange and Hillsborough counties in the Central Florida area. The intersections with at least one state road were used since the data for these intersections are well-maintained. The intersections which had major changes in geometry, signal timing, and traffic flow during the study period were excluded; therefore, the traffic operations for the selected intersections remained fairly uniform during the study period. #p#分頁標題#e#
Geometry design features for the intersections were extracted by inspecting the aerial imagery contained in the software Google Earth (2009), which puts high-resolution aerial and satellite imagery and other geographic information on the desktop. The number of through lanes on each approach, the number of left turn lanes and whether they were exclusive, the presence of median on each approach, whether they had exclusive right-turn lanes on each approach, the types of left turn offset (negative, zero, or positive offset), the angle of the intersections, and the direction of each intersection roadway were identified.
2.3.5 Methodology
Negative Binomial regression is a common tool for modeling cross-sectional count data. Generalized Estimating Equations provide an extension of generalized linear models (GLMs) to the analysis of correlated data, which can account for the correlation among the repeated observations from the same intersections. The methods of modeling correlation in GEEs are described followed by the introduction of the assessment of GEE models using the Cumulative Residuals method and of how to compare variables’ main effects by using the type III analysis.
2.3.6 Summary and conclusions
Left-turn crashes account for a high percentage of total crashes at signalized intersections, and they are prone to be severe due to the relatively high conflicting speeds of involved vehicles and the angle of impact. This study investigated left-turn crash occurrences at signalized intersections.
A massive data collection effort has been done. A total of 197 four-legged signalized intersections were selected in the Central Florida area. Intersection characteristics and crashes were collected. When inspecting the crash reports, it was found that half of left-turn crashes were originally recorded as other crash types. All 3098 left-turn crashes were divided into nine patterns based on vehicle maneuvers. In the data, 41.0% of left-turn crashes involved injury, whereas the percentage of injury crashes was only 22.7% for all other crashes.
The estimation results showthat there are obvious differences in the factors that cause the occurrence of different collision patterns and that the factors, even when identical, influence each model in different ways.
First, the logarithm of the product of conflicting flows was usually the most significant factor in each model, which confirms the assumption that the frequency of collisions is related to the traffic flows to which the colliding vehicles belong and not to the sum of the entering flows. Second, the type of left-turn phasing is the most significant factor for Pattern 5 crashes. This study confirms the findings of the previous studies that “protected” phase is safer than “permissive” phase and that “compound” signal usually has more left-turn crashes than “permissive” phase; however, this study found different effects for Pattern 8. Third, the opposing crossing distance and the median on the entering approach have different safety effects for Patterns 5 and 8; a wide crossing distance is correlated with more Pattern 5 crashes but fewer Pattern 8 crashes. Fourth, a higher speed limit for the approach of the through movement is identified to be significant for Patterns 1, 3, 5, and 8. In addition, #p#分頁標題#e#
3 Data Collection & Modification for Chosen Scenar

4 Implementation

5 Validation and Calibration Res6 Conclusion and Possibilities of Future researche

[1]Development of left turn lane guidelines for signalized and unsignalized intersection
[2]http://www.ttc-switch.com/dissertation_writing/Left-Turn Lane Installation Guidelines by 2009-08-10