Event Type: PhD Defense

Optimization-based approaches are presented for the design of environmentally oriented road pricing and traffic rationing schemes, particularly with the objective of curbing human exposure to motor vehicle generated air pollutants. The focus on human exposure to pollutants advances previous road pricing and traffic rationing problems which primarily account for congestion minimization, emission minimization, or emissions constraints. Practical utilization of the proposed problems is hindered by their time-consuming nature, so surrogate-based algorithms are developed to accelerate the search for good problem solutions. Given that the algorithms are derivative-free, they can be applied to various types of computationally expensive transportation network design problems.

A toll design problem is proposed for selecting tolling locations and levels that minimize environmental inequality and human exposure to pollutants. A mixed-integer variant of the metric stochastic response surface algorithm and a hybrid genetic algorithm-metric stochastic heuristic are presented to solve the mixed integer toll design problem. Numerical tests suggest that the surrogate-based algorithms have superior performance relative to previous genetic algorithm-based methods.

In addition, an optimization problem is presented for the design of cordon and area-based road pricing schemes subject to environmental constraints. Flexible problem formulations are considered which can be easily utilized with state-of-the-practice transportation planning models. A surrogate-based solution algorithm that uses a geometric representation of the charging area boundary is proposed to solve cordon and area pricing problems.

Lastly, a bi-objective traffic rationing problem is considered where the planner attempts to maximize auto usage while minimizing pollutant exposure inequality, subject to constraints on the levels of greenhouse gas emissions and pollutant concentration levels. A personal pollutant exposure methodology is integrated with standard models used in transportation planning to simulate person-level pollutant intake. To solve this problem a surrogate-assisted differential evolution algorithm for multiobjective continuous optimization problems with constraints is proposed. A sample application illustrates a possible implementation of the traffic rationing problem and the ability of the proposed algorithm to find diverse feasible solutions

Since HFCVs are not yet in the market, there is not enough personal travel data with HFCVs to accurately estimate potential demand. Yet, for fuel companies, sufficient numbers of HFCVs are required before investment in more stations becomes profitable. Alternatively, for customers, sufficient numbers of stations are required before purchasing and operating HFCVs becomes a realistic alternative to ICEVs. So, the initial balancing between this supply and demand confliction is vital to the fate of HFCVs as a market force. This work investigates the effect of refueling availability on choosing HFCVs by finding saturation densities of refueling stations for these vehicles. Using a subsample of households in the NHTS 2009 dataset, we first use parameters of the utility of choosing Toyota Prius vs. Toyota Corolla. We then argue that the values of these coefficients can be transferred to AFVs in general, and used in a preference model for AFVs vs. ICEVs, provided that we also transfer the coefficients of the appropriate purchase and operating costs. Using these models as base, we express the operating costs of AFV with respect to the density of refueling stations and the mean value of time, which then are included in the logit model as variables. We then employ a dynamic normative model that accommodates both the “bandwagon” effect and the results of the estimation of the random utility model of choice to estimate proportions of AFVs in the market over time. Stabilized market proportions are then used for finding saturation densities of stations.

Then, using these results, a competition model is proposed to forecast supplies for HFCVs based on demands forecasted by the dynamic normative model. Feedback models are used connect results derived from the competition and dynamic normative models.

In this research, we propose a series of models designed to take advantage of availability of data—both structured and unstructured—from a variety of sources ranging from passive data, to questionnaires, to social media to analyze underlying patterns and trends of travel and activity behavior. The results support enhancements both in transportation planning and also in the application of programming to support such efforts.
First, a framework for automatically inferring the travel modes and trip purposes of human movement, when tracked by a GPS device, is introduced. We utilize a multiple changepoints algorithm to divide trajectories into segments using only speed data, with no use of referencing information or assumptions about the participants’ temporal or location contexts. Then, Random Forest is used to classify segments into moving and not—moving types. For moving segments, travel mode is predicted. Next, multiple machine learning algorithms are employed, validated, and tested to identify the most suitable model for inferring trip purposes. Estimation results indicate that Random Forest provides the best results. The overall prediction accuracy is over 80% on the testing set—both with and without data on socio-demographic variables—predicting “shop” trips with an accuracy of 92.1%, while its accuracy for “go home” and “studying” trips reaches 100%.
Additionally, we analyze data pertaining to responses to the introduction of light rail service taken in waves to complement and evaluate knowledge about how personal travel behavior varies over time of day, day of week, and between waves. Our results indicate that, although the average of activity duration varies significantly over days of week and waves, the random effect of these two factors on activity duration was minor; time of day contributed over one third of the total variance in the duration.
Finally, the dissertation demonstrates uses of Twitter data as a potentially important data source to understand comments, criticisms, and responses about light rail in Los Angles. This result can be useful for exploring trends among commuters and how their emotions varied according to the light rail line they used, the time of day, and the day of the week.

A large and growing body of literature associates proximity to major roadways with increased risk of many negative health outcomes and suggests that exposure to fine particulate matter may be a substantial factor. Directly emitted and non-reactive mobile source air pollutants such as directly emitted fine particulate matter can form large spatial concentration gradients along major roadways, in addition to causing significantly large temporal and seasonal variation in air pollutant concentrations within urban areas. Current modeling and regulatory approaches for minimizing exposure have limited spatial resolution and do not fully exploit the available data.

The objective is to establish a methodology for quantifying fine particulate matter concentration gradients due to mobile source pollutants and to estimate the resulting population exposure at a regional scale. A novel air dispersion modeling framework is proposed using EPA’s AERMOD with data from a regional travel demand model that can produce a high resolution concentration surface for a considerably large metropolitan area; in our case, Los Angeles County, California. We find that PM2.5 concentrations are highest and most widespread during the morning and evening commutes, particularly during the winter months. This is likely caused by a combination of stable atmospheric conditions during the early morning and after sunset in the evening and higher traffic volumes during the morning and evening commutes. During the midday hours concentrations are at their lowest even though traffic volumes are still much higher than during the evening. This is likely the result of heating during the day time which leads to unstable atmospheric conditions that cause more vertical mixing and lateral dispersion, reducing ground level PM2.5 concentrations by transport and dilution. With respect to roadway centerlines, PM2.5 concentrations drop off quickly, reaching relatively low concentrations between 150m to 200m from the center line of high volume roads. However, during stable atmospheric conditions (e.g., nighttime & winter season) concentrations remain elevated at distances up to 1,000m from roadway centerlines.

We will demonstrate the feasibility of our methodology and how integrating the dispersion modeling framework into the travel demand modeling process routinely performed when developing and analyzing regional transportation improvement initiatives can lead to more environmentally and financially sustainable transportation plans. Regional strategies that minimize exposure, rather than inventories, could be established, environmental justice concerns are easily identified, and projects likely to cause local pollution “hotspots” can be proactively screened out, saving time and money for the transportation agency.

A macroscopic relation between the network-level average flow-rate and density, which is known as the macroscopic fundamental diagram (MFD), has been shown to exist in urban networks in stationary states. In the literature, however, most existing studies have considered the MFD as a phenomenon of urban networks, and few have tried to derive it analytically from signal settings, route choice behaviors, or demand patterns. Furthermore, it is still not clear about the definition or existence of stationary traffic states in urban networks and their stability properties. This dissertation research aims to fill this gap.

I start to study the stationary traffic states in a signalized double-ring network. A kinematic wave approach is used to formulate the traffic dynamics, and periodic traffic patterns are found using simulations and defined as stationary states. Furthermore, traffic dynamics are aggregated at the link level using the link queue model, and a Poincar ́ map approach is introduced to analytically define and solve possible stationary states. Further results show that a stationary state can be Lyapunov stable, asymptotically stable, and unstable. Moreover, MFD is explicitly derived such that the network flow-rate is a function of the network density, signal settings, and route choice behaviors. Also the time for the network to be gridlocked is analytically derived.

Even with the link queue model, traffic dynamics are still difficult to solve due the discrete control at signalized junctions. Therefore, efforts are also devoted to deriving invariant continuous approximate models for a signalized road link and analyzing their properties under different capacity constraints, traffic conditions, traffic flow fundamental diagrams, signal settings, and traffic flow models. Analytical and simulation results show that the derived invariant continuous approximate model can fully capture the capacity constraints at the signalized junction and is a good approximation to the discrete signal control under different traffic conditions and traffic flow fundamental diagrams. Further analysis shows that non-invariant continuous approximate models cannot be used in the link transmission model since they can yield no solution to the traffic statics problem under certain traffic conditions.

For a signalized grid network, simulations with the link queue model confirm that important insights obtained for double-ring networks indeed apply to more general networks.

Transportation agencies tasked with forecasting freight movements, creating and evaluating policy to mitigate transportation impacts on infrastructure and air quality, and furnishing the data necessary for performance driven investment depend on quality, detailed, and ubiquitous vehicle data. Unfortunately, commercial vehicle data is either missing or expensive to obtain from current resources. To overcome the drawbacks of existing commercial vehicle data collection tools and leverage the already heavy investments into existing sensor systems, we present a novel approach of integrating two existing data collection devices to gather high resolution truck data – Weigh-in-motion (WIM) systems and advanced inductive loop detectors (ILD). Each source provides a unique data set that when combined produces a synergistic data source that is particularly useful for truck body class modeling. Since body configuration is closely linked to commodity carried, drive and duty cycle, and other operating characteristics, it is inherently useful for each of the above mentioned applications.

In this work we describe the physical integration including hardware and data collection procedures undertaken to develop a series of truck body class models. Approximately 33,000 samples consisting of photo, WIM, and ILD signature data were collected and processed representing a significant achievement over previous ILD signature models which were limited to around 1,000 commercial vehicle records.

Three families of models were developed, each depicting an increasing level of input data and output class resolution. The first uses WIM data to estimate body class volumes of five semi-trailer body types and individual predictions of two tractor body classes for vehicles with five axle tractor trailer configurations. The trailer model produces volume errors of less than 10% while the tractor model resulted in a correct classification rate (CCR) of 92.7%. The second model uses ILD signatures to predict 47 vehicle body classes using a multiple classifier system (MCS) approach coupled with the Synthetic Minority Oversampling Technique (SMOTE) for preprocessing the training data samples. Tests show the model achieved CCR higher than 70% for 34 of the body classes. The third and most complex model combines WIM and ILD signatures using to produce 63 body class designations, 52 with CCR greater than 70%. To highlight the contributions of this work, several applications using body class data derived from the third model are presented including a time of day analysis, average payload estimation, and gross vehicle weight distribution estimation.

This dissertation provides modifications and extensions to the Household Activity Pattern Problem (HAPP) to help move existing formulations from a laboratory prototype toward a more useable activity-based demand modeling product. Previous research on HAPP has been based on a pickup and delivery problem with time window constraints (PDPTW), which does not lend itself easily to application that is compatible with an activity-based forecasting model. Meanwhile, other research on activity based modeling lacks of the integration of household decisions regarding time-of-day arrival, activity duration and traffic congestion effects on travel. We borrow concepts from economic research and consider that each household member tries to obtain maximum utility by choosing arrival time of activities, choosing activity duration while minimizing travel times and travel costs throughout the course of the day. Chapter 1 provides the introduction and motivation of this research. Chapter 2 reviews pertinent literature relative to the activity-based approach, the HAPP model, and positions the dissertation research relative to the existing state-of-the-art. In Chapter 3 we propose extensions to HAPP (UHAPP) that incorporate time of day activity arrival utility and the utility of activity duration into HAPP as decision variables. In Chapter 4 we introduce the travel time-dependent household activity pattern problem model (TUHAPP), which extends the ability of HAPP to capture the time-of-day (TOD) difference in travel times and costs. In Chapter 5 we develop a framework using TUHAPP (UHAPP) as a regional activity-based demand model with a household travel survey. Chapter 6 provides conclusions and future research.

The present thesis is comprised of four essays that address important gaps in passenger- and cargo-airline research. Seminal studies in airline economics that rely on cross-section methods make critical homogeneity assumptions and preclude time-specific effects. The essays in this thesis use panel data, which allow for certain assumptions made by cross-sectional studies to be relaxed, while shedding light on the intertemporal features of air transport.

The first chapter investigates the cost structure of air cargo carriers by applying a total cost model used in passenger-airline studies. Using quarterly panel data (2003-2011) on the domestic operations and costs of FedEx Express and UPS Airlines, empirical results indicate that the air cargo industry exhibits increasing returns to traffic density and constant returns to scale. Accounting for carrier-specific differences in cost structure and network size, FedEx is found to be more cost efficient than UPS (a finding that is reversed when network size is not controlled). Individually, UPS exhibits substantial economies of density and constant returns to scale while FedEx’s cost structure is characterized by weak economies of density and constant returns to scale. Both carriers exhibit economies of size.

The next three chapters embody papers that use quarterly panel data of city-level air traffic, airline delay, and socioeconomic variables. Spanning 10 years (2003-2012), the panel structure of the data permits the use of fixed effects to control for city-specific heterogeneity.

The second chapter presents a paper prepared for the Airport Cooperative Research Program (ACRP). The study demonstrates the within-city traffic impacts of urban size, employment composition, and wages, providing new insights into the determinants of passenger and air cargo traffic. The essay also confirms that airport traffic is proportional to population, and that service-sector employment and higher wages induce passenger travel and goods movement. A city’s share of manufacturing employment, however, only impacts air cargo traffic. Passenger enplanements exhibit more sensitivity to the proportion of urban workers providing non-tradable services, compared to the share of workers in tradable service jobs.

The third chapter, co-authored with Andre Tok, examines the determinants of air cargo traffic in California. The study uses a shorter 7-year panel (2003-2009), and shows that service and manufacturing employment impact the volume of outbound air cargo. Total (domestic) air cargo traffic is found to grow faster than (proportionally to) population, while wages play a significant role in determining both total and domestic air cargo movement. Metro-level air cargo tonnage are also forecasted for the years 2010-2040, indicating that California’s total (domestic) air cargo traffic will increase at an average rate of 5.9 percent (4.4 percent) per year in that period.

The final chapter is co-authored with Volodymyr Bilotkach, and it provides the first evidence on the impact of airline delays on urban-sectoral employment. Controlling for unobserved city-specific differences, the empirical estimates of the effects of air traffic on total employment are comparable to previously reported measures. However, service-sector employment is found to be less sensitive to air traffic than other studies suggested. New evidence confirming that delays have a negative impact on employment is also provided, a finding that is robust to various model specifications.

The introduction of Senate Bill (SB 375) in 2008 stimulated more research linking travel behavior to the built environment. Smart growth tools mandated by this bill aim to reduce vehicle miles traveled (VMT), greenhouse gas (GhG) emissions and promote alternative modes to motorized travel. These tools encompass an array of land use improvements that are expected to influence active travel. Potential changes in the built environment may impact the frequency, amount and even the selection of routes for walking.

Data used in this dissertation was obtained from Phase I of the Expo Study, a three-phase travel survey of residents living near the Expo Light Rail Line in Los Angeles, CA. Respondents carried GPS devices and accelerometers to track locations and activity levels; and completed seven-day trip logs. Phase I of the survey was administered in Fall 2011, prior to the introduction of the Expo Line in April 2012.

This dissertation is comprised of three essays. The first essay uses a “place-oriented” approach to examine where active travel occurs in neighborhoods adjacent to the Expo Light Rail Line. This essay is based on the Behavioral Model of Environments, which emphasizes the influence of the physical environment on individuals’ travel behavior and route choices. Results indicate that the routes selected by pedestrians have higher densities of commercial and retail centers and better access to more transit stations.

The second essay uses an ecological modeling approach. Multilevel analysis of the effects of the built environment on active transport was performed in three geographic levels of aggregation near respondents’ homes. Examination of land uses at the half-mile extent yield the least number of significant results. In contrast, land uses examined at the 100-meter and half-mile distance from homes emphasize the importance of street connectivity and green space on increasing transport-related physical activity (TPA). This suggests the importance of analyzing the data at finer geographic levels.

The third essay proposes a practical methodology of pedestrian route analysis in which observed GPS-tracked routes were examined and compared to GIS-simulated shortest paths. The two route types were compared over deviations in trip-level travel indices, respondents’ socio-demographic traits, time of day variations and differences in objectively measured built environment features along both sets of routes. Results suggest that observed routes diverged more from shortest paths with increasing distance and were more circuitous beyond the 2.4-mile threshold. Most walks were completed after the AM Off Peak time. With the exception of the Evening time, observed routes were found to be much longer in all time periods especially in the AM Peak time. Moreover, higher densities of commercial centers, local businesses and green spaces were observed for GPS-tracked routes than shortest paths. These routes also had more street intersections and transit stops. Overall, results imply that pedestrians selected routes that were longer than the respective shortest paths and that may have been due to greater access to amenities and activity centers.

Understanding the choices that each individual in the population makes regarding daily plans and activity participation behavior is crucial to forecasting spatial-temporal travel demand in the region. The focus of this dissertation is developing a comprehensive mathematical/statistical framework to infer and replicate travel behavior of individuals in terms of their socio-demographic profiles. The framework comprises series of distinct modules that employ statistical segmentation, Bayesian econometrics, data mining, and optimization techniques to predict individuals’ activity types, activity frequencies, and the travel linkages that make them possible.

The key advantages of the model are: first; providing the likely content of activity agenda as part of the inference procedure, second; integrating transportation network topology within activity scheduling step, and third; integrating modal components. As part of the dissertation, a Graphical User Interface was developed for practical application of the model in transportation agencies. The data used for the analysis is the California Household Travel Survey data, 2000-2001. Testing the entire modeling system on an out-of-sample population—15% of the entire sample— shows that the model is able to predict on average 80.3% of daily activities of individuals correctly.