- : Ms Word Format
- : 80 Pages
- : ₦3000
- : 1-5 Chapters
- Click to DOWNLOAD Materials
RELATIONSHIP BETWEEN SPEED METRICS AND CRASH FREQUENCY AND SEVERITY: IMPLICAITONS FOR ROADWAY DESIGN
ABSTRACT
Reducing the number and severity of crashes on highways and streets is of high importance for government officials and transportation professionals in the United States. Substantial research has focused on various speed metrics, such as operating speeds and the posted speed limit, and their relationship to safety, such as crash frequency and crash severity. Crash severity is the safety measure most often linked to measures of speed and is based on dissipation of kinetic energy. However, many aspects of the relationships between speed metrics and crash frequency and risk have yet to be studied in depth, so a complete understanding of speeding-related crashes is unknown.
Design speeds are used to establish geometric design criteria, and operating speed results from the geometric design process. Posted speed limits may be established based on operating speeds or by statute. When posted speed limits are inconsistent with design or operating speeds, road safety performance may be affected. A more complete understanding of the relationship between safety performance and operating speeds, posted speed limits, and design speeds may produce rational speed limits and lead to improved safety performance on roadways.
This research combined real-time vehicle probe speed data, roadway inventory data, and crash data to assess crash risk and crash frequency.
This thesis first determined the risk of a crash on two-lane rural highways based on operating speed metrics, differences between speed metrics, and traffic volume data. Results from the crash risk analysis indicate that operating speeds in 1-minute and 5minute averages improve the statistical fit and prediction of binary logistic regression models. Higher traffic volumes and operating speeds higher than either the road average speed or road reference speed were associated with increased crash risk. Whereas, variations in travel speeds between vehicles were associated with decreased crash risk.
This thesis also analyzed the frequency of crashes on horizontal curve segments of two-lane rural roadways using operating speed data, differences among speed metrics, traffic volume data, roadway inventory data, and crash data. Negative binomial regression models improve the statistical fit and prediction of crash frequency models compared to random-effects negative binomial regression. Generally, increases in the differences between operating speed and road average speed and the differences between operating speed and inferred design were associated with an increase in crash frequency. Increases in the differences between inferred design speed and posted speed limit were also associated with an expected increase in crash frequency; however, increases in the operating speed variance and in the difference between operating speeds and posted speed limit were associated with an expected decrease in crash frequency.
TABLE OF CONTENTS
List of Tables…………………………………………………………………………………………………….. vii
List of Figures……………………………………………………………………………………………………… x
Acknowledgments……………………………………………………………………………………………….. xi
Chapter 1 Introduction………………………………………………………………………………………….. 1
Chapter 2 Literature Review………………………………………………………………………………….. 5
2.1 Crash Frequency – Speed Relationships…………………………………………………………. 5
2.1.1 Speed Limit – Safety Relationships…………………………………………………………. 5
2.1.2 Operating Speed and Speed Variance – Safety Relationships……………………… 6
2.2 Crash Severity – Speed Relationships…………………………………………………………… 12
2.3 Crash Risk – Speed Relationships………………………………………………………………… 16
2.4 Safety Performance Functions and Safety Statistical Modeling……………………….. 17
2.4.1 Safety Performance Functions……………………………………………………………….. 18
2.4.2 Crash Frequency Statistical Modeling……………………………………………………. 25
2.4.3 Crash Severity Statistical Modeling……………………………………………………….. 31
2.5 Summary………………………………………………………………………………………………….. 34
Chapter 3 Data…………………………………………………………………………………………………… 35
3.1 Sources of Data and Data Description………………………………………………………….. 35
3.1.1 Speed Data………………………………………………………………………………………….. 35
3.1.2 Roadway Inventory Data………………………………………………………………………. 38
3.1.3 Crash Data………………………………………………………………………………………….. 42
3.2 Crash Risk Database Development………………………………………………………………. 45
3.2.1 Merging Process………………………………………………………………………………….. 45
3.2.2 Supplementary Data…………………………………………………………………………….. 51
3.3 Crash Frequency Database Development……………………………………………………… 53
3.3.1 Merging Process………………………………………………………………………………….. 54
3.3.2 Supplementary Data…………………………………………………………………………….. 57
3.4 Descriptive Statistics………………………………………………………………………………….. 58
3.4.1 Crash Risk Database Descriptive Statistics……………………………………………… 58
3.4.2 Crash Frequency Database Descriptive Statistics…………………………………….. 63
Chapter 4 Statistical Analysis Framework……………………………………………………………… 66
4.1 Variables…………………………………………………………………………………………………… 66
4.2 Statistical Methods…………………………………………………………………………………….. 69
4.2.1 Binary Logistic Regression…………………………………………………………………… 69
4.2.2 Negative Binomial Regression………………………………………………………………. 71
4.2.3 Random Effects Negative Binomial Regression………………………………………. 73
4.2.4 Model Development and Variable Function Form Selection……………………… 75
4.2.5 Goodness of Fit Measures…………………………………………………………………….. 76
Chapter 5 Crash Risk Analysis Results…………………………………………………………………. 82
5.1 Operating Speed 1-Minute Averaged Data……………………………………………………. 83
5.1.1 Total crashes……………………………………………………………………………………….. 83
5.1.2 Fatal and Injury Crash Severity Levels…………………………………………………… 88
5.1.3 Fatal and Major Injury Crash Severity Levels…………………………………………. 93
5.2 Operating Speed 5-Minute Averaged Data……………………………………………………. 95
5.2.1 Total Crashes………………………………………………………………………………………. 95
5.2.2 Fatal and Injury Crash Severity Levels…………………………………………………. 101
5.2.3 Fatal and Major Injury Crash Severity Levels……………………………………….. 105
5.3 Operating Speed 15-Minute Averaged Data………………………………………………… 108
5.3.1 Total Crashes…………………………………………………………………………………….. 109
5.3.2 Fatal and Injury Crash Severity Levels…………………………………………………. 112
5.3.3 Fatal and Major Injury Crash Severity Levels……………………………………….. 116
5.4 Model Selection………………………………………………………………………………………. 119
5.5 Summary………………………………………………………………………………………………… 122
Chapter 6 Crash Frequency Analysis Results……………………………………………………….. 124
6.1 Time variable 1: Season and Year……………………………………………………………… 125
6.1.1 Total Crashes…………………………………………………………………………………….. 125
6.1.2 Fatal and Injury Severity Levels………………………………………………………….. 133
6.2 Time variable 2: Season……………………………………………………………………………. 139
6.2.1 Total Crashes…………………………………………………………………………………….. 139
6.2.2 Fatal and Injury Severity Levels………………………………………………………….. 146
6.3 Time variable 3: Year……………………………………………………………………………….. 149
6.3.1 Total crashes……………………………………………………………………………………… 149
6.3.2 Fatal and Injury Severity Levels………………………………………………………….. 155
6.4 Model Selection and Post-Estimation Testing……………………………………………… 155
6.5 Summary………………………………………………………………………………………………… 159
Chapter 7 Conclusions………………………………………………………………………………………. 161
7.1 Findings………………………………………………………………………………………………….. 161
7.1.1 Crash Risk Analysis Findings……………………………………………………………… 161
7.1.2 Crash Frequency Analysis Findings……………………………………………………… 162
7.2 Applications of this Research…………………………………………………………………….. 163
7.3 Recommendations for Future Research………………………………………………………. 164
References……………………………………………………………………………………………………….. 167
CHAPTER 1 INTRODUCTION
Reducing the number and severity of crashes on highways and streets is a highpriority for government officials and transportation professionals in the United States
(U.S.). Approximately 6.3 million police-reported crashes occurred in the U.S. in 2015
(NHTSA, 2016a). While the number of fatal crashes in the U.S. declined between 1994 (36,254 crashes) and 2014 (30,056 crashes), the most recent data indicate that this trend is not continuing (NHTSA, 2017a). There were 32,166 fatal crashes in 2015, but 34,439 fatal crashes in 2016 (NHTSA, 2017a). According to crash statistics, motor vehicle crashes were ranked as the 13th leading cause of death in the U.S. (NHTSA, 2016b). When separated into age categories, motor vehicle crashes were the number one leading cause of death for people between 16 and 24 years old and was among the top 10 leading causes of death for all people under 64 years old (NHTSA, 2016b).
The 32,166 fatal crashes in 2015 resulted in a total of 35,092 fatalities. Approximately 49 percent of the total fatalities in the U.S. (and 51 percent in Pennsylvania) occurred in rural areas (17,114 fatalities), yet the fatality rate in rural areas is 2.6 times higher than urban areas (NHTSA, 2017b). This suggests that crashes in rural areas are overrepresented, so traffic management efforts should be directed at roadways in these areas.
These crash statistics demonstrate the need for effective transportation safety improvement strategies to mitigate the frequency and severity of crashes on U.S. roadways. Because more than one-quarter of the fatal crashes in 2015 were attributed to speeding (NHTSA, 2017a), strategies to decrease the number of speeding-related crashes is of particular importance. According to the U.S. Department of Transportation (USDOT) Federal Highway Administration (FHWA), speeding is defined as “traveling too fast for conditions or in excess of the posted speed limit (FHWA, 2017).” However, speeding is considered to be a complex issue and mitigation strategies typically involve the interaction between road user behavior, engineering, education, and enforcement (FHWA, 2017).
In regards to engineering, substantial research has focused on various speed metrics, such as operating speeds and the posted speed limit, and their relationship to safety. Crash severity is the safety measure most often linked to measures of speed. However, many aspects of the relationships between speed metrics and safety measures have yet to be studied in depth, so a complete understanding of speeding-related crashes remains unknown. Further exacerbating this issue is that state transportation agencies have recently begun increasing posted speed limits on existing highways, particularly freeways and two-lane rural highways, producing inconsistencies between operating speeds, posted speed limits, and geometric design speeds.
When posted speed limits are increased to values that differ from those determined by original engineering and safety studies, inconsistencies between the speed metrics of those roadways might be produced. Current practices for setting posted speed limits include legislative/statutory laws, engineering studies, optimization, and the application of expert systems; however, the two most commonly used approaches are legislative/statutory and engineering studies. Using the legislative/statutory approach, laws dictate the speed limit based on the type of road and the surrounding area. Engineering studies involve collecting a sample of operating speed data in the field and posting the speed limit based on the 85th percentile operating speed of free-flow traffic, with consideration given to the roadway geometrics, roadside development, and accident rate. When speed limits are set without considering operating speeds, roadway features, and the surrounding area/environment, posted speed limits are not consistent with the actual and desired operating speeds. Therefore, raising speed limits on existing roads can result in regulatory limits that are either higher or lower than the design speed, engineering recommendations, or operating speeds, producing inconsistencies among speed metrics. The implications of such practices should be studied in order to determine the impacts on roadway safety.
Rational speed limits are needed in order for roadways to be consistent with driving expectations. A more complete understanding of the relationship between safety performance and operating speeds, posted speed limits, and design speeds may produce rational speed limits and lead to improved safety performance on roadways. This research focuses on quantifying the relationship between crash risk and crash frequency and various speed metrics using two-lane rural highway data from Pennsylvania.
Therefore, the research objectives include determining a method to quantify the relationships between operating speeds, posted speed limits, and design speeds on twolane rural highways, and to determine how the differences between these speed metrics affect safety performance of roadways. While all safety aspects are considered, the safety metric will primarily focus on crash frequency. This is because the relationship between severity and operating speed is well-documented (Elvik, 2009; Kockelman et al., 2006; Malyshkina and Mannering, 2008; SWOV, 2012), and generally concludes that more severe crash outcomes result from higher operating speeds.
The first objective of this research is to determine the risk of a crash based on different operating speed metrics using real-time vehicle probe speed data and traffic volume data. The speed metrics include operating speed (in 1, 5, and 15 minute averages), posted speed limit, design speed, and inferred design speed. These speed metrics, along with the differences among these speed metrics, are used to assess crash risk. The second objective of this research is to determine the frequency of crashes on curves and tangents using roadway data and the speed metrics previously described.
This research is significant to the field of engineering. While real-time speed data has been previously related to crash risk using detector data, real-time speed data has not been used before to assess crashes and the risk and frequency of crashes using probe data. This real-time vehicle probe speed data is linked to roadway data and crash data to assess the safety performance of two-lane rural roads. A direct relationship between crash frequency and operating speed metrics has not been previously determined.
This thesis is organized into five subsequent chapters. Chapter 2 provides a review of relevant literature and identifies limitations in the extant literature. Chapter 3 presents the data sources and descriptions of the data, including how the databases were developed. Subsequently, Chapter 4 provides the framework for the data analysis. This is followed by Chapter 5 that shows the results from the crash risk analysis. Chapter 6 shows the results from the crash frequency analysis. Finally, Chapter 7 describes the conclusions and recommendations from this research.
