This study investigates the effect of rainfall on transit performance measures at the route level in the Puget Sound region of Washington State. Transit agencies are required to report certain performance metrics to the Federal Transit Administration (FTA), but performance measures can also be used to evaluate service and provide customers with information regarding the transit system. Using a three-year sample of archived automatic vehicle location (AVL) and hydrologic data the relationships between ridership, travel time, delay, and rainfall were investigated. The analysis of daily ridership and rainfall resulted in no statistically significant results, however, the results are supported by the existing research in this field. There was a generally negative trend in ridership with respect to rainfall. The analysis of travel time and rainfall did not result in the expected outcome. It was hypothesized that travel time would vary with rainfall, but that was not always the case. During many rainfall events the travel time remained average. The analysis of delay and rainfall shows that the impact of rainfall on delay is more complex than assumed. The delay during dry trips was different than the delay during light and moderate rain, but during heavy rain the statistical different disappeared. These results, implications for transit operators, and future research opportunities are discussed.

Walkability as a concept that captures the ability to walk from one place to another has multiple dimensions. Between traversability to being a proxy for better urban places, there are also numerous measurements of walkability that attempts to quantify certain or all aspects of walkability. It is, however, unclear, through a review of available literature, how these measurements of walkability relate to each other statistically. This methodology focuses on generating a framework for analysts to evaluate and prioritize pedestrian infrastructure. WalkScore™ (WS), HCM Pedestrian Level of Service (PLOS), Average Nodal Degree (AND), and Intersection Density are the four metrics selected for this analysis that focuses on distinctive aspects of walkability (proximity, amenity, network-connectivity, respectively). A sample of 51 street segments from the County of San Luis Obispo is selected according to their respective Average Daily Traffic (ADT) volumes. Pearson’s Correlations between the six combinations of relationships are measured, and the strongest correlation between the six relationships is between WalkScore™ and Intersection Density with an R2 of 0.44. A regression model that includes external factors such as population and adjacent land use is used to analyze and predict PLOS of the street segment. Although the model is not statistically significant, the goal of this research is to identify gaps in current and potential walkability of street segments in the sample. Therefore, this framework of using established walkability metrics to predict PLOS, and then distinguishing places for improvements is proposed as a result of this research to be used by government agencies to prioritize pedestrian infrastructure.

Urban transportation networks are vulnerable to various incidents. In order to combat the negative effects due to incident-related congestion, various mitigation strategies have been proposed and implemented. The effectiveness of these congestion mitigation strategies for incident conditions largely depends on the accuracy of information regarding network conditions. Therefore, an efficient and accurate procedure to determine the link states, reflected by flows and density over time, is essential to incident management. This thesis presents a user equilibrium Dynamic Traffic Assignment (DTA) model that incorporates the Cell Transmission Model (CTM) to evaluate the temporal variation of flow and density over links, which reflect the link states of a transportation network. Encapsulation of the CTM equips the model with the capability of accepting inputs of incidents like duration and capacity reduction. Moreover, the proposed model is capable of handling multiple origin-destination (OD) pairs. By using this model, the temporal variation of flows over links can be readily evaluated. The visualized prediction of link density variations is used to investigate the link state relationships. By isolating the effects of an incident, the parallel routes of a specific OD pair display the relationship of substituting for each other, which is consistent with the general expectation regarding such parallel routes. A closer examination of the density variations confirms the existence of a substitution relationship between the unshared links of the two parallel routes. This information regarding link state relationship can be used as general guidance for incident management purposes. / Master of Science

Data-driven intelligent transportation systems (ITS) are increasingly playing a critical role in improving the efficiency of the existing transportation network and addressing traffic challenges in large cities, such as safety and road congestion. This dissertation employs data dimensionality reduction, reinforcement learning, and discrete outcome models to improve traffic flow and transportation safety. First, we propose a novel data-driven technique based on Koopman operator theory and dynamic mode decomposition (DMD) to address the complex nonlinear dynamics of signalized intersections. This approach not only provides a better understanding of intersection behavior but also offers faster computation times, making it a valuable tool for system identification and controller design. It represents a significant step towards more efficient and effective traffic management solutions. Second, we propose an innovative phase-switching approach for traffic light control using deep reinforcement learning, enhancing the efficiency of signalized intersections. The novel reward function, based on speed, waiting time, deceleration, and time to collision (TTC) for each vehicle, maximizes traffic flow and safety through real-time optimization. Finally, we introduce a mixed spline indicator pooled model, an approach for multivariate crash severity prediction, addressing the limitations of previous models by capturing temporal instability. It carefully incorporates additional independent variables to measure parameter slope changes over time, enhancing data fit and predictive accuracy. The developed models are estimated and validated using data from the Central Florida region.

This thesis focuses on evaluating the appropriate speed at which vehicles should travel under different traffic conditions on freeways and its impact on crash frequency. The common belief is that the lower speed results in fewer crashes as reduced speed provides drivers with more time to react effectively and avoid collisions. However, this perspective overlooks the interplay among traffic speed, average spacing between consecutive vehicles, and the distance available for stopping a vehicle. Hence, we propose a safety parameter termed ‘Safety Correlate' (SCORE), which is defined as the proportion of average spacing relative to the stopping distance. To determine the relationship between SCORE and crash frequency, data from 366 4-lane urban freeway segments located in Virginia was analyzed and a Random-effects Poisson Lognormal model was developed. The obtained result indicated that the safety parameter SCORE is negatively associated with the annual hourly crash frequency, implying that the lesser the average spacing as a proportion of the stopping distance while traffic flow remains constant, the more frequent will be the crashes. Additionally, this research presents an application of SCORE in setting variable speed limits under various traffic flows. Overall, the study results provide valuable insights by investigating SCORE to improve traffic safety. Also, this research would help practitioners and policymakers to incorporate safety aspects while setting variable speed limits on freeways.

This dissertation has contributed to the pedestrian safety literature by assessing and comparing safety benefits and traffic efficiency among midblock Rectangular Rapid Flashing Beacon (RRFB) and Pedestrian Hybrid Beacon (PHB) sites. Video trajectory data were used to calculate pedestrian Surrogate Safety Measures (SSMs) and vehicles' delay. Regression models of SSMs and vehicles' delay revealed that PHB sites offer more safety benefits, at the expense of increased vehicles' delay, compared to RRFB sites. The presence of the PHB, weekday, signal activation, lane count, pedestrian speed, vehicle speed, land use mix, traffic flow, time of day, and pedestrian starting position from the sidewalk have been found to be significant determinants of the SSMs and vehicles' delay. Another avenue of pedestrian safety explored in this dissertation is the lag time. The study investigates survival likelihood and the lag time of non-instant pedestrian fatalities using random parameter Binary Logit and Ordered Logit models. The models were run on a dataset obtained from the Fatality Accident Reporting System (FARS) for the period of 2015-2019. The analysis revealed that weather, driver age groups, drunk/ distracted/ drowsy drivers, hit and run, involvement of large truck, VRU age group, gender, presence of sidewalk, presence of intersection, light condition, and speeding were common significant factors for both models. The factor found to be significant exclusively for the Binary Logit model includes Area type. Factors found to be significant exclusively for the Ordered Logit model include Presence of Crosswalk and Fire station nearby. The results validate the use of lag time as an alternative to crash count and crash severity analysis. The findings of this study pave the way for practitioners and policymakers to evaluate the effectiveness of midblock pedestrian crossing facilities, as well as to use lag time to investigate crashes and corroborate results from traditional crash-based investigations.

Owing to advancements in technology, substantial investments within the automotive industry, and the formulation of supportive state policies, the future landscape of the transportation sector is poised to witness a shift from traditional internal combustion engine vehicles (ICEVs) to electric vehicles (EVs). While EVs have made inroads in the market, they still face significant hurdles in the form of range anxiety and prolonged charging durations, inhibiting their widespread adoption. To tackle these challenges, a comprehensive approach to smart transportation electrification is proposed, emphasizing the pivotal roles of infrastructure development, particularly in the allocation of charging stations, and strategic operational decisions, including charging and platoon scheduling. This dissertation is structured around four essential components. The initial stage entails grasping the intricacies of charging demand, recognized as the foundational step before embarking on any transportation electrification initiative. Subsequently, the allocation of charging stations is addressed, with a specific focus on ride-sourcing vehicles, distinct from private EVs due to issues such as relocation time, waiting time, and dynamic pricing that affects spatiotemporal value of time (VOT) costs. This approach, which considers VOT costs, is essential in avoiding biased results in the planning of charging infrastructure for electrified ride-sourcing services. The third chapter centers on the optimization of charging and platoon scheduling, particularly within the context of long-haul freight vehicles. The objective here is to harness the flexibility of charging schedules to facilitate vehicle platooning, thereby reducing the demand for charging, and, consequently, energy consumption. This chapter involves the development of a mixed-integer programming model and explores various techniques, such as hyperparameter tuning and hybrid meta-heuristic methods, to optimize the model for large-scale applications. Lastly, the fourth chapter takes on the challenge of addressing uncertainty in scheduling problems. This is achieved by formulating a two-stage stochastic model and applying it within a hypothetical numerical example, providing a framework for optimizing charging station (CS) planning while accounting for uncertain operational parameters.

Recognizing the distinct non-motorist injury severity profiles by crash location (segment or intersection), we propose a joint modeling framework to study crash location type and non-motorist injury severity as two dimensions of the severity process. We employ a copula-based joint framework that ties the crash location type (represented as a binary logit model) and injury severity (represented as a generalized ordered logit model) through a closed form flexible dependency structure to study the injury severity process. The data for our analysis is drawn from the Central Florida region for the years of 2015 to 2021. The model system explicitly accounts for temporal heterogeneity across the two dimensions. A comprehensive set of independent variables including non-motorist user characteristics, driver and vehicle characteristics, roadway attributes, weather and environmental factors, temporal and sociodemographic factors are considered for the analysis. We also conducted an elasticity analysis to show the actual magnitude of the independent variables on non-motorist injury severity at the two locations. The results highlight the importance of examining the effect of various independent variables on non-motorist injury severity outcome by different crash locations.

Vehicle miles traveled (VMT) and its relationship with the built environment has been extensively studied. Most notably, five D variables of the built environment including density, diversity, design, destination accessibility, and distance to transit are the key variables included in this research to explain VMT generation from housing developments. This thesis uses prior research that developed robust statistical models and findings to create a framework to estimate VMT reduction affected by infill housing developed using incentives provided by the state compared to a regional comparator. The regional comparator is typically a suburban single-family housing development in the region. The models recommended for future application of the framework are based on ease of finding the data on variables included in the model and statistical robustness. The application of the framework in the Central Coast and San Francisco Bay Area regions of California shows that infill prototypes developed can generate an 11-27% reduction in VMT per capita. The findings are specific to a synthetic household defined for this study. The research provides ways to apply this framework to other regions of the state along with ideas to consider for future work. These ideas include exploring the VMT reduction potential based on households with different income levels appropriate for the regions, future modeling efforts, and selection of existing models. The findings of this thesis support that the combination of the five D variables can help attribute to a larger VMT reduction than the VMT reduction caused by the change of a single variable. When destinations are clustered, and jobs are available at a reasonable distance to the residence, a significant reduction in VMT is more achievable. The results inform public agencies on which locations are ripe for devoting further resources for incentivizing housing development to reach climate and housing goals.

In the United States, there are two schemes of operating traffic signal controls for permitted protected left turns (PPLT) namely the traditional five-section head system (known as Dog-House) and the flashing yellow arrow system (FYA). Past studies have agreed that these controls lead to decrease the average delay per left turn vehicle, decrease the protected green time, increase the left turn capacity, and enhance the intersection overall operation. The flashing yellow arrow (FYA) has been approved by the Federal Highway Administration as the national standard for the PPLT operations at signalized intersections. So, the Florida Department of Transportation also approved this new system and they are extensively replacing the traditional system with the new system on the area of Central Florida (Lin, et al, 2010). Both these systems have been used for a long time and there are some studies that evaluated these systems but there are limited number of projects that evaluated and/or compared between the two PPLT systems from the operational perspective. The main goal of this research is to study the characteristics of traffic operations and evaluate the effectiveness of the conversion from five-section head signal to the FYA treatments at 13 intersections located in Orlando, Florida. To reach this goal, detailed data collection efforts were conducted at 13 selected intersections in the central Florida area and appropriate statistical tests were conducted using the Minitab 17 Software. Statistical tests were attempted to fit different new regression models that correlate delay and left turn volumes as response variables against a set of independent variables that included permitted green time, opposing volume, percent of trucks, time gaps, speed, and land use type. In addition to fitting the data to regression models, these models were also analyzed for the purpose of detecting any significant differences between the five-section head treatment and FYA treatment. The statistical differences of converting the five-section head system to FYA system were discussed. The results in this thesis agreed with some of the previous studies and did not agree with others. In general, the flashing yellow arrow system was found to enhance the intersection operation, increase the number of left turn vehicles, and reduce the delay. Also, some suggestions and recommendations were made based on this study results.