Phd Dissertation

Simulation Study of Day-Night Variations in Emissions Impacts and Network Augmentation Schemes: An Application to PierPASS Policy for Port Trucks in California

Abstract

Freight operations are critical to our prosperity, but they also generate substantial external costs in the form of additional congestion, air pollution, and health impacts. Unfortunately these external costs are not well understood. In this dissertation, I focus on the drayage trucks that serve the San Pedro Bay Ports (or SPBP, i.e. the Ports of Los Angeles and Long Beach in Southern California), which is the largest port complex in the country. This research focuses on the PierPASS program, which shifts drayage trucks traffic from mid-day and peak hours to the evening and night hours. External costs from drayage trucks remain a major concern for communities adjacent to the ports because they bear a disproportionate fraction of the health impacts (respiratory and cardiovascular illnesses, cancers, and premature deaths) associated with the pollution generated by ports operations. In this context, the purpose of my dissertation is analyze the impacts of shifting freight traffic to off-peak periods with an emphasis on congestion, air pollution (NOx, and PM) and related health impacts. This impact analysis was conducted using a framework that integrates microscopic traffic simulation with emission estimation, air dispersion, and a health impact assessment. The research also developed a new approach for origin-destination demand estimation on large microscopic simulation network that is made by augmenting an existing simulation network. Thus the research makes both policy analysis and methodological contributions, and is expected to help enable policy makers to craft cleaner logistics policies. I found that PierPASS had little impact on traffic congestion and on overall emissions of various pollutants. However, PierPASS had a significant impact on the distribution of these emissions between day and night. During night-time, total port truck emissions increased by 71% for NOx and 72% for PM, while day-time emissions decreased by 9% for both NOx and PM. My dispersion analysis shows that PierPASS increased air pollutant concentrations during both day time and night time because of boundary layer effects. Finally, my health impact analyses using EPA’s BenMAP model show that the annual social costs due to PierPASS are $438 million.

MS Thesis

Feedback with Alternate Trip Assignment Approaches in the Four-Step Model

Publication Date

March 29, 2014

Author(s)

Abstract

The traditional Four-Step Model is widely used in the Transportation Planning and Forecasting Process. However, the model itself has a structural defect: it is only equilibrium in terms of Trip Assignment. Thus it is often viewed as an inadequate, partial equilibrium model. To achieve an overall equilibrium in the Transportation Planning and Forecasting Process, a feedback process can be introduced into the Four-Step Model. The objective of this research is to incorporate different feedback methods into the traditional Four-Step Modeling process to improve model performance. The specific approach herein is to examine the relative performance of direct and averaging feedback methods, and then to investigate the convergence of this two approaches by replacing User Equilibrium trip assignment with All-or-Nothing trip assignment during each feedback loop, but not in the original four step model application. An evaluation and comparison of these methods is presented. Two measures of effectiveness, the Root-Mean-Square Error (RMSE) and Total Vehicle Hours Traveled (VHT), are employed to test the convergence and evaluate the methods.

Phd Dissertation

ReMuLAA - A new algorithm for the route choice problem

Publication Date

March 14, 2014

Author(s)

Abstract

A new framework for analyzing the choice set formation for route choice models in transportation networks is presented and an algorithm is proposed. The algorithm is tested against a sample of GPS data for heavy trucks for the State of California. The results are presented in detail along with an analysis of both their qualitative and quantitative merits. A new algorithm for the route choice problem is also presented and its results analyzed against the state of the practice and state of the art. This new algorithm, ReMuLAA, is also the first known closed solution algorithm for the route choice problem using the Multinomial Logit Model (MNL) for an entire class of networks (Directed Acyclic Networks) without explicit route enumeration. A correction for the MNL model to account for route overlapping is also presented and the results are compared with other state-of-the-art route choice algorithms. The results of the application of ReMuLAA in a real world model are also presented and its advantages discussed.

MS Thesis

Left-turn elimination network analysis

Abstract

Left-turn movement volume takes small percentage of approach volume, however case delay to the majority of traffic flow at an intersection. Left-turn movement has longest averaged delay at an intersection itself. The idea of eliminating left-turn movement is to force a small number of left-turn trips to re-routing, as a results, all other trips would have less delay at an intersection. Elimination of some left-turn movements at a defined network may help improving network performance in term of network travel time, thus, a higher level of system optimization can be achieved. A network analysis is performed to compare network performance before and after left-turn elimination applied. Traffic assignment result is expected to be different because of trips re-routing after left-turn elimination.. Intersection control delay is expected to decrease at the intersection that where left-turn elimination is applied. As a result, network travel time is expected to drop because of the saving at intersection control delay. Signalized intersection control delay is a key of this study. Trips re-routing may happen because of prohibited left-turn movement, as well as the difference in turn penalty per movement if it is assigned, which would affect shortest path calculation. Trip re-routing may cause increase in total turn movement volume in network.

research report

Quantifying the Effect of Local Government Actions on VMT

Publication Date

February 13, 2014

Author(s)

Deborah Salon, Marlon Boarnet, Patricia (Pat) Mokhtarian

Abstract

This research uses empirical analysis of travel survey data to quantify how much Californians will change the amount that they drive in response to changes in land use and transport system variables. The study improves upon past research in three key ways. First, a dataset comprising merged information from five California-based household travel surveys was assembled. Second, a novel approach to control for residential self-selection was developed. Third, understanding heterogeneity in effects of variables on vehicle miles of travel (VMT) across two important dimensions – neighborhood type and trip type — was a focus. The effects of some land use and transport system characteristics do depend on neighborhood type, in ways that are intuitive but had not previously been empirically verified. Results of this research are embedded in the VMT Impact spreadsheet tool, which allows users to easily see the implications of this work for any census tract, city, or region in California.

Phd Dissertation

Interregional Commodity Flow Model Using Structural Equation Modeling: Application to the California Statewide Freight Forecasting Model

Publication Date

December 14, 2013

Author(s)

Abstract

Freight forecasting models are data intensive and may require many explanatory variables to achieve prediction accuracy. One problem, particularly in the United States, is that public data sources are usually available only at highly aggregate geographic levels, while models with more disaggregate geographic levels are required for regional freight transportation planning. A second problem is that supply chain effects are often ignored or modeled with economic input-output models which lack explanatory power. This study addresses these challenges by considering a Structural Equation Modeling approach, that is not confined to a specific spatial structure as spatial regression models would be, and allows for correlations between industries. The goal of the proposed methodology is to design a reliable and policy sensitive modeling framework for long term commodity flow forecasting that makes the best use of public available data sources. Practicality and improvement over previously available freight generation and distribution models are the highlights of this approach. There are two primary developed in this study. The first one is a structural commodity generation model. The second model is the Structural Equations for Multi-Commodity OD Distribution (SEMCOD) model. The proposed framework is implemented as a primary module in California Statewide Freight Forecasting Model (CSFFM) and will be used to update the California Transportation Plan (CTP 2015). The models are specified and estimated based on FAF3 data. It is shown that the proposed modeling framework provides a better fit to the data than independent regression models for each commodity. The three components of the models are: direct and indirect effects, supply chain elasticities at zone level and at origin-destination level, and intra-zonal supply-demand interactions. A validation of the geographic scalability of the model is conducted using a zoning system consisting of 97 county or sub-county zones in California.

Phd Dissertation

Improving On-Road Emissions Estimates with Traffic Detection Technologies

Abstract

Transportation has been a significant contributor to greenhouse gas and criteria air pollutant emissions. Emission mitigation strategies are essential in reducing transportation’s impacts on our environment. In order to effectively develop and evaluate on-road emissions reduction strategies, accurate quantification of emissions is the critical first step. The accuracy and resolution of the traffic measures needed by the emission models will directly affect the emission estimation results. This dissertation investigates the ability of traffic detection technologies to provide the traffic measures needed for accurate on-road emissions estimation. A review of traffic detection technologies is provided with insight into their capability and suitability for estimating emissions. The Inductive Vehicle Signature (IVS) system is identified as currently the most promising technology to couple with EPA’s latest MOVES emission model for estimating emissions. Models and algorithms based on the IVS detection system are developed to generate the two most important traffic measures for emission estimation: vehicle mix and average speed. The performances of the models are verified using real-world data. Assuming the IVS system and the models developed are deployed to generate vehicle mix and average speeds, the accuracy and reliability of the emissions estimation results based on these traffic measures are evaluated by simulating the operations of the models in the field using NGSIM data. Very promising results are obtained, which clearly demonstrates the capability of the IVS system for on-road emissions estimation. A Real-Time Emissions Estimation and Monitoring System based on the IVS technology is implemented on the I-405 freeway to estimate operational emissions on the road in real-time. Although average speed has been the most common input into emission models, the MOVES model is capable of using second-by-second vehicle speed trajectories to estimate emissions more accurately. Vehicle speed trajectories are becoming increasingly available thanks to the proliferation of GPS-enabled personal navigation devices and smartphones. Crowd sourced GPS data can also be used by emission models like MOVES to estimate emissions. This dissertation studies the use of a limited number of GPS speed trajectories to estimate emissions for all traffic on the road. Two fundamental questions are answered by this work: 1) how can GPS data be used for emissions estimation, and 2) how does the penetration rate of the GPS probes affect the emission results. With the methods proposed in this study, it is found that emissions can be estimated with high accuracy and reliability with even a very small penetration rate of GPS probes, when combined with the vehicle mix data generated from the IVS system. Discussions on the applications of the proposed systems and methods to various emissions analysis scenarios are also provided in this dissertation.

published journal article

The location selection problem for the household activity pattern problem

Abstract

In this paper, an integrated destination choice model based on routing and scheduling considerations of daily activities is proposed. Extending the Household Activity Pattern Problem (HAPP), the Location Selection Problem (LSP–HAPP) demonstrates how location choice is made as a simultaneous decision from interactions both with activities having predetermined locations and those with many candidate locations. A dynamic programming algorithm, developed for PDPTW, is adapted to handle a potentially sizable number of candidate locations. It is shown to be efficient for HAPP and LSP–HAPP applications. The algorithm is extended to keep arrival times as functions for mathematical programming formulations of activity-based travel models that often have time variables in the objective.

research report

Spatially Focused Travel Survey Data Collection and Analysis: Closing Data Gaps for Climate Change Policy

Abstract

This research explored the effect of small area land use policies on land use–travel behavior relationships. The authors pioneered methods to obtain travel data with sufficient spatial focus to shed light on how land use influences vehicle miles of travel. Travel diary surveys were obtained from four small neighborhoods in southern California. Results suggest differences in walking, transit, and passenger vehicle travel behavior associated with residing in areas with different built environment, land use, and transit access characteristics. Households in areas with higher employment accessibility tended to have more walking travel and lower vehicle miles of travel (VMT). Households within 1.5 miles of a rail transit station tended to have more transit ridership. Households within 0.5–1.0 miles of a rail transit station tended to have more walking travel, while households with higher levels of transit service were associated with lower household VMT. The methods developed advanced efforts toward low-cost, rapid travel data collection that can be used in before-and-after transportation program evaluations in the future.

working paper

Integration of Weigh-in-Motion and Inductive Signature Technology for Advanced Truck Monitoring

Abstract

Trucks have a significant impact on infrastructure, traffic congestion, energy consumption, pollution and quality of life. To better understand truck characteristics, comprehensive high resolution truck data is needed. Higher quality truck data can enable more accurate estimates of GHGs and emissions, allow for better management of infrastructure, provide insight to truck travel behavior, and enhance freight forecasting. Currently, truck traffic data is collected through limited means and with limited detail. Agencies can obtain or estimate truck travel statistics from surveys, inductive loop detectors (ILD) and weigh-in-motion (WIM) stations, or from manual counts, each of which have various limitations. Of these sources, WIM and ILD seem to be the most promising tools for capturing detailed truck information. Axle spacing and weight from existing WIM devices and unique inductive signatures indicative of body type from ILDs equipped with high sampling rate detector cards are complementary data sources that can be integrated to provide a synergistic resource that otherwise does not exist in practice, a resource that is able to overcome the drawbacks of the traditional truck data collection methods by providing data that is detailed, link specific, temporally continuous, up-to-date, and representative of the full truck population. This integrated data resource lends itself very readily toward detailed truck body classification which is presented as a case study. This body classification model is able to predict 35 different trailer body types for FHWA class 9 semi-tractors, achieving an 80 percent correct classification rate. In addition to the body classification model, the large data set resulting from the case study is itself a valuable and novel resource for truck studies.