Graduation Year

2024

Document Type

Dissertation

Degree

Ph.D.

Degree Name

Doctor of Philosophy (Ph.D.)

Degree Granting Department

Civil and Environmental Engineering

Major Professor

Qing Lu, Ph.D.

Committee Member

Fred F. Mannering, Ph.D.

Committee Member

Pei-Sung P. Lin, Ph.D.

Committee Member

Lu Lu, Ph.D.

Committee Member

Mingyang Li, Ph.D.

Keywords

Crash Frequency, Crash Severity, Continuous Pavement Friction, Macrotexture, Unobserved Heterogeneity

Abstract

Pavement shall accommodate current and predicted traffic needs in a safe, durable, and cost-effective manner. Among pavement surface characteristics, pavement friction is a critical factor contributing to roadway safety, as it directly impacts a driver's ability to execute steering, braking, and acceleration maneuvers safely. Ensuring an adequate level of pavement friction is essential to minimize the skidding risk for all road users. Continuous pavement friction measurement (CPFM) has recently become a proven approach, revolutionizing safety-related pavement management and skidding crash injury prevention. Meanwhile, pavement roughness has long been linked to vehicle fuel efficiency and pavement structural degradation. Perceptions of pavement roughness, however, may vary among users of different sociodemographic features and, therefore, potentially impact the allocation of resources for highway maintenance and rehabilitation when equity is considered. It is necessary to investigate the major factors influencing the user perception of pavement roughness. After an extensive and comprehensive literature review, this dissertation analyzed the impact of pavement characteristics on pavement safety and management via statistical heterogeneity models. It included four topics, summarized as follows.

Following a thorough literature review, the study first investigated the impacts of friction patterns and other pavement features on crash frequency. It leveraged district data collected via the Sideway-Force Coefficient Routine Investigation Machine (SCRIM) over 2,142 lane miles on state roads in Tampa Bay, Florida, in 2023. The CPFM data, including the friction (in terms of friction coefficient [SR40]) and macrotexture (in terms of mean profile depth [MPD]), were spatially matched to a 0.1-mile buffer around crash locations for roadway network analysis. Various statistics for the matched CPFM data were estimated for the development of safety performance functions (SPFs), using random parameter negative binomial (RPNB) and correlated RPNB models along with other input variables such as roadway, pavement, and traffic characteristics. Crash Modification Functions (CMFs) were then developed based on these analyses. The modeling results quantified the relationship between pavement characteristics and crash frequencies by region, pavement type, and crash type. The modeling results indicated that increasing friction or macrotexture was more likely to reduce crash frequencies in all the scenarios except the scenario of fatal and serious injury crashes on urban and suburban roads with dense-graded asphalt pavement or rigid pavement. In contrast, intensified variability of friction and macrotexture tended to increase crash frequencies in specific scenarios. Other pavement characteristics that impact crash frequencies include pavement roughness, rutting, and pavement cracking. Investigatory levels were developed to determine the safety-related friction demand and thresholds for safety review.

The SCRIM data, including the SCRIM Coefficient (SC) and macrotexture (MPD), was spatially matched to a 0.2-mile buffer around crash locations. The mixed logit models with heterogeneity in means and variances, alongside other factors such as roadway, traffic, and vehicle characteristics, were estimated based on the SCRIM data. The modeling results indicated that the highest pavement friction within the 0.2-mile buffer, marked by the maximum three-point moving average of the SC, most significantly impacted crash injury severity: higher pavement friction significantly reduced the likelihood of severe and fatal injuries. When the mean of MPD on dense-graded asphalt pavement surfaces fell within the range of 0.8-1.2 mm, there was a greater likelihood of reducing slight injuries. A mean MPD greater than 2 mm on open-graded asphalt pavement surfaces was associated with an increased likelihood of severe and fatal injuries. The analysis also found that a higher 5-minute harmonic speed was statistically more likely to result in severe injury or slight injuries and less likely to result in no injury. However, further investigation was needed to determine the impact of friction and macrotexture on crash injury severity on various roadway types and pavement conditions.

Previous studies have not created CPFM-based SPFs and CMFs for motorcycle crash frequency. This study utilized processed CPFM data to develop SPFs and CMFs for motorcycles, considering different facilities and pavement types across various population groups. The modeling results indicated that increased friction is more likely to reduce motorcycle crash frequencies, while intensified friction variability tended to increase crash frequencies in specific scenarios. Other pavement characteristics affecting motorcycle crash frequencies included macrotexture and pavement conditions (roughness, cracking, and rutting). Investigatory levels for friction and macrotexture were developed based on the SPFs to determine the friction demand for motorcycle safety.

This study also delved into the intricate details of how users perceive pavement roughness by conducting an in-depth analysis of data from previous research, which involved in-vehicle tests with individual users, various pavement conditions, and different types of vehicles. To ensure comprehensive analysis, this study utilized advanced machine learning and heterogeneity statistical methods, such as a classification tree, a random parameter random threshold order probability (RPRTOP), and a correlated RPRTOP model, to account for both group and individual unobserved heterogeneity and numerous variables in the dataset. The findings of the study uncovered a more extensive array of factors influencing roughness ranking compared to previous research. The results revealed that while physical measurements of pavement roughness (e.g., International Roughness Index), visible distress such as patches and faulting, and joints, were strong indicators of user roughness ranking, other factors including the specific route regularly used, participants’ age, income, gender, number of household infants, and interior vehicle noise levels also played a statistically significant role. Furthermore, the data indicated that two-way group random effects were statistically significant, highlighting the need to consider them in future studies. The outcomes of this study bridge the gap between making accurate predictions and understanding the underlying causality in research related to physical infrastructure measurement and user perceptions of infrastructure conditions. These findings may be integrated into pavement safety management.

Download

Share

COinS