Master's Theses Completed With ITS Support, 1990-2004

2005

Bonstead, Tyler R. Red Cars Reborn: Transforming the Abandoned Pacific Electric Santa Ana Line into a 21st Century Transit System. M.S., Civil Engineering, 2005. Adviser: Michael G. NcNally

For the first half of the 20th Century, the Pacific Electric Santa Ana Line was an integral link in the transit network between Los Angeles and Orange County. But by the 1950's it was abandoned, and the car took over as the prime mover in the lives of Southern Californians. Now the corridor over which the Pacific Electric Red Cars once ran stands ready to move Southern Californians into the 21st century. This thesis examines the many different alternatives that could be constructed on the Pacific Electric right-of-way, and narrows the large field down to three. These three alternatives are described in more detail in the remainder of the paper. The first alternative examined in this paper is a general-purpose arterial, an extension of the current transit paradigm in the LA metro area. The second is bus rapid transit, in which large modern vehicles would run on their own busway, providing a level of service similar to rail. The final alternative is light rail, a fonn of rail transit that can run in city streets or its own exclusive right-of- way. All three alternatives give area residents a new choice for their commutes, with another major goal of drawing higher population densities to the surrounding area. With a new transit line along the Pacific Electric right-of-way, the area should be able to handle better the large numbers of new residents bound for the Southland.

Breiland, Christopher. The Effect of Allowing Single Occupant Hybrid Vehicles on High Occupancy Vehicle Lane Operations. M.S., Civil Engineering, 2005. Adviser: Will Recker

In an effort to reduce the environmental impacts of automobile travel, legislatures across the country are considering laws that would allow hybrid vehicles with a single occupant to use high-occupancy vehicle (HOV) lanes. Six states have already passed a “hybrid-HOV” law and Virginia has implemented its law over the objections of the Federal Highway Administration. While hybrid-HOV laws have been fairly well received by the public, the concern is that, over time, single occupant hybrid vehicles could increase congestion in HOV lanes to such a degree that the travel time incentive is lost. This thesis examines the potential impact of implementing California ’s hybrid-HOV law, AB 2628, on HOV lane operations in the Irvine , California area. The results of running a Paramics microscopic simulation on several demand scenarios reveal that AB 2628 will not have a significant impact on HOV lane operations. While AB 2628’s short-term impacts on HOV lane operations may be modest, the long-term implications are less favorable. The Virginia Department of Transportation has recommended that the legislature not extend the state’s current HOV law, and a high-demand scenario analyzed with Paramics suggests that HOV lane breakdown will eventually occur if hybrids are given unrestricted access to HOV lanes. Given that hybrids are being sold faster than they are being made, laws like AB 2628 currently seem out of place. Moreover, history has shown that once a group is given a privilege, like access to HOV lanes, revoking that privilege can be very difficult even if it is in the best interest of society.

2004

Tay, Larry Soo. Estimation of Trip Matrices for Traffic Simulation. M.S., Civil Engineering, 2004. 123 pp. Adviser: Michael G. NcNally

There is a rapidly increasing need to understand travel patterns in urban and metropolitan areas and to be able to capture their dynamics. The unavailability of ideal data measures has shaped and limited all of the attempted solutions to such a problem. To the researcher, understanding these changing travel patterns at finer levels can improve the effectiveness of infrastructure planning and the efficiency of control devices. Equally as important, but of greater interest to the practitioner, is the ability to be able to investigate changes in local and regional travel patterns at a moderately detailed level with time-efficient and computation-friendly tools; and to do so with growing amounts of confidence. These "tools", micro and macro simulators, have evolved tremendously over the last decade. The thesis documents the requirements for undertaking such a project, and attempts to investigate and evaluate some of those promising tools while generating results for a study area that has been burdened with traffic congestion.

Pages, Laia. Simulation Frameworks to Evaluate Modern Transit Designs. M.S., Civil Engineering, 2003. 79 pp. Adviser: R. Jayakrishnan

This thesis examines the simulation requirements to microscopically model newer transit systems such as Bus Rapid Transit Systems (BRT), and Demand Responsive Transit (DRT) systems, which could also be feeder systems to higher-capacity transit services. The study develops a general framework to model transit and fleet systems with specific attention on simulating BRT and DRT systems using the available network microscopic simulation software. The study then examines different simulation packages and discusses the differences that need to be considered if the proposed framework were implemented in each of them. Finally, a simulation scenario to model a combined system of BRT and High Coverage Point to Point Transit System (HCPPT) was performed. This simulation scenario is used to examine the variations in the performance of such a system when the supply of BRT buses varies.

Smolke, Brian Justin. An Evaluation of Freeway Bottlenecks using Microsimulation. M.S., Civil Engineering, 2003. 122 pp. Adviser: Michael G. McNally

There is an increasing reliance upon traffic simulation to assist in engineering and planning decisions. Many significant decisions regarding city planning are being made with the results of simulations as important input. Simulation is quite useful because it can be used to predict the effects on traffic conditions caused by demographic changes or modifications of roadway networks. Despite the extensive use of simulation, there seems to be a lack of documentation on the usage of models and the reliability of their results in terms of planning projects.
In this project, the traffic simulation program PARAMICS was employed to carry out a case study of a planning project. This case study is representative of the type of planning project that simulation can be used for in practice. The project involved modeling an area currently experiencing significant congestion during the peak rush hours. The area focused on the interchange of the I-5 and SR-55 freeways located in Orange County, California. The current conditions were first simulated. Several possible improvements to the existing roadway network were then modeled and their effects on traffic analyzed. This thesis will discuss methods used to develop a reasonable model of the actual roadway network within PARAMICS. The usefulness and the reliability of the results will also be examined.

2002

Alvarez, Veronica Maria. Safety Implications of Weaving Sections in Freeway Design: A Case Study in Orange County. M.S., Civil Engineering 2002. 93 pp. Adviser: Will Recker

Injury from motor vehicle collisions is the leading cause of death in the United States for persons aged 1 to 34. In California alone motor vehicle crashes annually kill nearly 4,000 people, injure approximately 280,000, and cost an estimated $15 billion. In increasingly crowded urban freeway systems, some of the worst congestion problems occur at weaving sections, where an entrance ramp is followed by an exit ramp. Only a few studies have been conducted that focus on obtaining a relation between the intrinsic features associated with weaving sections and accidents occurring in them. In this study a new and innovative methodology for studying the relation between traffic flow and accident conditions in weaving sections is proposed, based on the use of real time disaggregated data in order to distinguish the potential traffic conditions that could cause a incident. Accident characteristics provided by TASAS were matched with VDS loop real-time database for the case of Orange County freeways in 1998. Based on HCM methodology, new categories of types of weaving sections were added and several statistical tests were performed finding that sideswipe collisions have the highest likelihood of occurrence in weaving sections, with most of them happening in an interior lane. Additionally, all combinations including type C showed a higher probability of having accidents during periods of high-traffic volumes as well as during such extreme conditions as rainy weather. Finally, more than 70% of the observed time, weaving sections presented constrained operation during the period prior to the accident.

Janakiraman, Sankar. A Priori vs. Re-Optimization Methods for Time Constrained Routing and Scheduling of Local Vehicle Fleets. M.S., Civil Engineering, 2002. 53 pp. Adviser: Amelia C. Regan

The motivation for the research presented in this thesis is rooted in the operations of pickup and delivery fleets. The problem is formulated as a multiple traveling salesman problem with time windows (m-TSPTW) with asymmetric travel times.  The pick up and delivery problem models the situation in which a fleet of vehicles must satisfy a collection of transportation requests. Each request includes an origin and destination. Vehicles must be dispatched and routed so that each request is picked up at its origin and delivered to its destination. This simple problem structure can be used to model trucking operations, package delivery or public transit. In this thesis, an integer programming model is formulated to model the multiple vehicle pickup and delivery problem. The problem is solved as an m-TSPTW using two different methods. First, a network flow binary integer program formulation is proposed and solved using off the shelf optimization software (CPLEX). The second method uses a time window discretization method in which continuous variables (service times) are replaced by multiple discrete variables. That formulation is also solved using CPLEX.  A key issue addressed is the performance of static solutions for pickup and delivery operations in which the travel times are stochastic. This question is addressed by comparing the quality of a priori solutions generated using expected travel times versus solutions generated based on full knowledge of the true values of travel times.

Nandiraju, Srinivas. A New Cutting Plane Method for Integer Programming. M.S. Civil Engineering, 2002. 74 pp. Adviser: Amelia C. Regan

This thesis examines a new cutting plane (Twin Min Cut) method for solving integer programming problems. The goal of this research is to find better algorithms for solving integer programming problems which are practically very hard to solve and which emerge in many problems of interest to both academia and industry. In this research, the Twin Min Cut method is used to solve a variety of classical integer programming problems. The proposed cuts are compared to the classical Gomory Cuts, one of the classic cuts in literature, to gauge the relative performance of the method. We prove the analytical properties of this new cut and also look into the effectiveness of these cuts. Several techniques to improve the Twin Min Cut method are examined. These include finding deeper cuts and specific several implementation details, which are critical for the performance. We look into possibilities of embedding these cuts in a branch and cut framework, as cutting plane methods by themselves do not have good performance.

Wang, Huabing. Development of Estimation Procedures for Activity-Based Travel Demand Forecasting. M.S., Civil Engineering, 2001. 140 pp. Adviser: Will Recker

The typical but challenging problem in activity-based travel demand forecasting is focused on inferring the relative weights associated with potential components of the utility function. In this thesis, we implement an estimation procedure for the Household Activity Pattern Problem (HAPP) model in order to estimate the relative importance of factors associated with spatial and temporal interrelationships among the out-of-home activities that motivate a household's need or desire to travel. The method uses a genetic algorithm to optimize the utility function, a particular multidimensional sequence alignment method to deal with qualitative attributes of activity pattern (e.g., which household member to perform each activity, which vehicle to use...), and a time sequence alignment method to handle quantitative attributes of activity pattern ( e.g., starting time, duration...).

Jagannathan, Sreeram. Analysis of Dynamic Dispatching Heuristics for Carrier Fleet Operations in a GIS Framework. M.S., Civil Engineering, 1999. 77 pp. Adviser: Amelia C. Regan

This study examines some dispatching heuristics for carrier fleet operations, which are typically iynamic and inherently stochastic. The context of this research is truckload trucking operations where a single load is moved from its origin to its destination. A simulation framework is formulated to provide a testbed for the examination of these heuristics. A geographic information system (GIS) is interfaced with this framework to facilitate the use of a real travel network. A set of heuristic load assignment and reassignment strategies is investigated. These are referred to as the base case (BC), last load reassignment (LL), worst load reassignment 1 (WLED) and worst load reassignment 2 (WLER). All these strategies allow en-route diversion and are compared based on a set of performance measures. These strategies are evaluated relative to intermodal operations (truck, rail) of a trucking company operating in the Los Angeles area. This study is concerned only with the trucking portion of the operations. Three different congestion levels are considered -high, medium and low-levels and the performance of these heuristics under these conditions is evaluated.

Porwal, Himanshu. Dynamic Origin/Destination Demand Estimation Using True Section Densities. M.S., Civil Engiineering, 1999. 141 pp. Adviser: R. Jayakrishnan

This thesis presents a practical approach for dynamic origin/destination demand estimation. The proposed dynamic origin/destination estimation framework addresses many of the shortcomings of the existing formulations and presents a formulation for general networks and not just corridors. One unique feature of this framework is its use of section density as a variable instead of flow. The framework is built upon the foundation of static origin/destination matrix estimation by adding the temporal aspect. Two traffic assignment models, namely DYNASMART and DTA are used for assigning dynamic ODs onto the network and 1-Step Kalman Filter and Least Squares methods are used for optimizing the errors between the estimated and the true section counts. 1-Step Kalman Filter is considered as a special case of a Kalman Filter which is developed for future work with a rolling horizon estimation framework. In addition, this formulation also describes an infrastructure from which real-time traffic counts and other section data on various freeways could be collected and used in dynamic frameworks.

Yang, Xu. Parameter Optimization for a Microscopic Traffic Simulator using a Genetic Algorithm and an Application Programming Interface. M.S., Civil Engineering, 1999. 69 pp. Adviser: Will Recker

PARAMICS is an example of the latest generation of microscopic models for both freeway and surface street traffic simulations. To accomplish realistic traffic simulations, parameters that govern vehicle movements and behavior in simulation models need to be calibrated in accordance with the traffic pattern of the interested research area. This thesis presents the application of a Genetic Algorithm (GA) in finding optimal combinations of parameter values for PARAMICS. With the calibrated parameters, a mirror image of the real world traffic environment for the California ATMS Testbed is established in PARAMICS. An Application Programming Interface (API) is employed in PARAMICS in order to replicate real world traffic conditions in the calibration process. The API implementation in PARAMICS facilitates determination of the initial conditions of the PARAMICS model to match as close as possible the real transportation system. To validate the effectiveness of calibrated parameters, 30-second field loop detector data collected from the California ATMS Testbed are used. The calibration procedure was applied to midday off-peak traffic. The results indicate the promising performance of the Genetic Algorithm in the searching for optimal parameter combination, enabling PARAMICS to generate such traffic data as volume and occupancy that closely match the field data under different traffic conditions. Issues of how to use the Genetic Algorithm and Application Programming Interface in an area-wide network setup for parameter optimization in microscopic traffic simulation models are discussed. Conclusions and future research are also presented.

Parimi, Arun. Development of a Framework for Analyzing the Potential Impacts of Transportation Control Measures on Vehicle Emissions. M.S., Civil Engineering, 1998. 150 pp. Adviser: Will Recker

The 1990 Clean Air Act Amendments (CAAA) and the Intermodal Surface Transportation Efficiency Act of 1991 (ISTEA) have defined a set of transportation control measures to counter the increase in the vehicle emissions and energy consumption due to increased travel. The value of these TCM strategies is unknown as there is data available to measure the travel effects of individual TCM strategies and the models are inadequate in forecasting changes in travel behavior resulting from these strategies. The work described in this thesis begins to provide an operational methodology to overcome these difficulties so that the impacts of the policy mandates of both CAAA and ISTEA can be assessed. This research will demonstrate the benefits in vehicle emissions reduction based on optimal scheduling and linking of the activities performed by the individuals in a household. The potential of transportation policy options to alleviate vehicle emissions is determined in a comprehensive activity-based approach. The model formulated is tested under different policy scenarios, including an evaluation of potential benefits achieved by replacing all the vehicles in the fleet by vehicles conforming to present day emissions technology.

Fill, Ana Cristina. Transit Network Simulation and Line Selection Behavior. M.S., Civil Engineering, 1997. 146 pp. Adviser: R. Jayakrishnan

This thesis presents a transit simulation that can evaluate transit systems employing Advanced Public Transportation Systems (APTS). This simulation contains three major components to ensure its success. These components include: (a) real-time bus movement, (b) explicit passenger behavior modeling, and (c) network wide lines and passenger interaction. Although other implemented transit simulations often contain one or more of the above requirements, none incorporate all of them. Another very valuable feature included in the transit simulation described in this thesis regards its coordination with a traffic simulator that utilizes other Intelligent Transportation Systems (ITS) components. This integration allows the evaluation of strategies aimed at optimizing the transportation system as a whole rather than addressing different components individually. This thesis also demonstrates one of the several applications for this transit simulation. The chosen application examines the benefits of providing information to system users regarding the time of arrival of buses at their stops. Both the actual and psychological benefits to the passengers were considered. This application was selected because most cities implementing or planning to implement APTS technologies in their transit system have this application as one of their first priorities.

Compin, Nicholas Shawn. Rail Transit Station Development and the Municipal Land-Use Decision-Making Process. M.A., Social Ecology, 1996. 80 pp. Adviser: Marlon G. Boarnet

In recent years the idea of building medium to high-density residential development near rail transit stations has been gaining momentum with academics and policymakers alike. Such development, it is theorized, will increase rail transit ridership and reduce the total number of auto trips, thereby improving regional air quality and reducing roadway congestion. This paper specifically addresses the question of why, despite the attention given to the concept of transit-oriented development (TOD) by local, regional and state governmental units, there have been relatively few TODs constructed near rail stations in Southern California, and more specifically, near the stations of the San Diego Trolley, the oldest modern-day rail transit system in the region. Results from in-depth interviews with planning directors from the seven cities in which all Trolley stations are located indicate that although officials have been well- educated on the theoretical benefits of TOD, the implementation of station-area plans, including TOD, has been a priority in only one city. In addition, in those cities where station areas are seen as a potential focal point for development, most often it is due to economic benefits resulting from commercial land uses. Finally, the use of TOD for infill purposes has taken place in only one city. In the remaining municipalities where transit-based developments are scheduled, they are planned for construction on large portions of previously undeveloped property in outlying areas. The findings suggest that strategies which stress the regional benefits of rail transit, such as those involving TOD, may fall short when striving to influence local, land- use decisions. Instead, strategies which concentrate on increasing local economic benefits associated with rail transit-based development are viewed as being potentially the most effective way to influence municipalities into concentrating on station-area development.

Kim, Hye Jin. Application of Artificial Neural Networks in Discrete Choice Modeling. M.S., Civil Engineering, 1996. 99 pp. Adviser: Michael G. McNally

Traditional discrete choice analysis is performed via random utility models. They rely heavily on the statistical theory of model estimation and testing. A disadvantage of these models is the need to assume an underlying error term which in practice may vary case to case. The resulting model is valid only when the assumed error term is valid; the error term distribution, however cannot in general be validated. This type of parametric model restricts its applications and computing cost can be significant when the model takes on more than two choices. In this thesis, a non-parametric artificial neural network is proposed as an alternative discrete travel choice analysis. A non-parametric model negates the need for error specification. For the purpose of analysis, two sets of data were generated under three error term distributions with varying degree of error magnitudes. One set was use for design and the other for testing for the purpose of comparing the validity of the analysis. Comparison are made with two traditional discrete choice models (binary logit and probit) and the artificial neural network. The performance of the neural network is proven to be more robust than that of the traditional parametric formulation.

Kulkarni, Anup Arvind. The Influence of Land Use and Network Structure on Travel Behavior. M.S., Civil Engineering, 1996. 193 pp. Adviser: Michael G. McNally

There has always been considerable interest in the interaction between the land use-transportation system and travel behavior and as a result an ample body of work has emerged which examines various aspects of this complex relationship. Underlying this interest has been a fundamental assumption that the land use-transportation system has a considerable impact on travel behavior. Unfortunately, while there has been much work in this area, it has been difficult to develop and test the possible connection in a consistent statistical approach.
For this project, an integrated database of land use characteristics, network characteristics, and travel diary data was collected. This database was then analyzed by selecting sample neighborhoods to examine potential interactions. The objective being to develop conclusions about possible strategies involving changes to the land use-transportation system to influence travel behavior. This thesis aims to extend this body of work in new directions with the goal of exploring and understanding this complex relationship

Rindt, Craig. An Implementation and Evaluation Platform for Integrated Advanced Transportation Management and Information Systems. M.S., Civil Engineering, 1996. 125 pp. Adviser: Michael G. McNally

There is a growing body of new approaches, algorithms, and technologies for Advanced Traffic Management and Information Systems (ATMIS) which are designed to interface with other applications and with real-world instrumentation to improve transportation system performance. These applications range from detection and analysis techniques to management and control functions. This diversity yields the potential for many different solution configurations, the nature of which cannot be clear defined at this point. What is clear is that they will consist of multiple interdependent applications which require real-time information from the traffic system. This thesis develops a platform for the implementation and evaluation of integrated traffic solutions employing ATMIS technology. The platform consists of a hybrid mesoscopic/microscopic traffic simulation with scalable modeling detail and a communications library for implementing integrated ATMIS solutions as distributed algorithms. The design emphasizes a modular flexibility which simplifies the process of connecting new applications to a given integrated solution. Solutions implemented on the platform are capable of interfacing with the hybrid simulation, a real-world traffic system, or some combination thereof. This connection provides system data to the algorithms and allows them to manage and control the system interactively. The discussion includes the theoretical and practical concerns of both hybrid simulation and distributed transportation algorithm implementation, develops the implementation and evaluation platform based on this discussion, and provides an overview of an integrated traffic management system implemented on the platform.

Komerska, Richard J. Development of a Modeling Tool for the Preliminary Design of Personal Rapid Transit Networks. M.S., Engineering, 1995. 163 pp. Adviser: Michael G. McNally

A computer modeling tool was designed to be used in the preliminary design and analysis stage of a Personal Rapid Transit (PRT) planning effort. This program, called PRTnet, was created to help bridge the gap that currently exists between sophisticated microscopic simulation tools and simple uniform-grid prediction models. Sophisticated simulation tools, which include the effects of stochastic behavior, are necessary during the detailed design and analysis stage of a PRT planning exercise, but can be too time consuming for preliminary analysis. On the other hand, the simple uniform- grid models do not capture the unique characteristics of a real network, such as station location and origin-destination (O-D) demand. The PRTnet program allows the planner to input a candidate PRT network geometry, including station and link topology, and an O-D matrix. After utilizing an AII-Or-Nothing assignment strategy to load the network, the program outputs network-wide performance measures including required fleet size, average trip time and total vehicle-miles traveled. The PRTnet program permits rapid parametric analysis and "what if" studies to be performed, allowing the evaluation of many network variations within a relatively short timespan.

Logi, Filippo. A Software Architecture for the Integration of Advanced Transportation Management and Information Systems. M.S., Engineering, 1995. 181 pp. Adviser: Stephen G. Ritchie

This thesis describes a knowledge-based architecture that constitutes the design of a second generation expert system for Traffic Management Center operator support in incident management on freeway and arterial networks. The architecture, called TCM (Traffic Congestion Manager), is conceived as a structured collection of knowledge modules for the detection, verification and analysis of incidents and the selection of response strategies that integrate the various management and control means available to control center operators, such as signals, ramp meters and changeable message signs. The modular structure of the architecture provides different levels of abstraction making the knowledge embedded in the system easily accessible, understandable and modifiable. Furthermore, the system can provide a detailed analysis of problems, a deep explanation of the resolution process, and a set of alternative management plans.  After an introduction of the motivations that led to the choice of an artificial intelligence-based approach to the problem of incident management, and an outline of the most relevant works being carried on in this field, the various knowledge modules that compose the TCM architecture are described. First, the two high level units are introduced, one responsible for the detection and analysis of operational problems, and one for proposing adequate management plans. A detailed description of the lower level units, that perform basic tasks follows. The behavior of the system is later shown, through the definition of a complete example, that highlights the way the system approaches operational problems. Final considerations and some directions for further research conclude this work.

Vaideeswaran, Ganesh. Investigation and Implementation of Decomposition Algorithms for the Traffic Assignment Problem. M.S., Engineering, 1995. 143 pp. Adviser: R. Jayakrishnan

This thesis discusses the development and implementation of a new algorithm called the PD-GP (Physical Decomposition for Gradient Projection) algorithm for traffic assignment in urban networks. The algorithm is based on physical network decomposition and naturally renders itself for distributed processing. PD-GP could be used as part of on-line dynamic assignment frameworks and real-time network optimization strategies in ATMS (Advanced Traffic Management Systems). This algorithm is path-based and the solutions are in terms of path flows as opposed to link flows produced by conventional algorithms such as Frank-Wolfe. This feature combined with the excellent computational performance of the algorithm could be exploited in dynamic route guidance algorithms for ATIS (Advanced Route Guidance Strategies) also. A detailed discussion of the investigation and development of the PD-GP algorithm is given in this thesis. The algorithm is also implemented on a distributed platform and the details of the implementation are discussed. PD-GP is applied to hypothetical networks of various sizes under different physical decomposition schemes. The time savings and the solution quality produced are compared with that of the Gradient Projection (GP) algorithm. For the reasonably large networks studied, it is seen that time savings as high as 68% could be obtained by sequential implementation of the algorithm which increase to as high as 90% when the algorithm is implemented on the distributed platform. Since GP itself is several times faster than the commonly used Frank-Wolfe algorithm, such time savings are impressive. The solution quality produced is also good, though it depends on several factors such as the subnetwork selection and the network topology.

Yoon, Hankil. The DREAM Microkernel and a PC-Based Prototype Implementation. M.S., Electrical & Computer Engineering, 1995. 64 pp. Adviser: Kwang H. (Kane) Kim

Although demand for large-scale real-time (RT) applications is ever increasing, the-state-of-the-art remains inadequate for dealing with those applications, and the contemporary RT technologies still lack in flexibility, predictability and accuracy in both design and development of scalable RT safety-critical applications. In addition, extension of the object-oriented structuring approach to RT distributed computer system design has much less advanced than what was expected before. The new paradigms adopted by Kane Kim in the UCI DREAM Laboratory include general object-oriented structuring of RT distributed computing systems and design-time guarantee of timely responses. To support such design paradigms, the RTO.k object structuring scheme was developed principally by Kane Kim and Hermann Kopetz. To support RTO.k objects and conventional process-oriented RT computation, an RT distributed system kernel, called the DREAM kernel, was designed by Kane Kim with some assistance of his research staff in the UCI DREAM Laboratory. The DREAM kernel was aimed at possessing guaranteed timely service capabilities and strongly predictable behavior. The kernel architecture consists of five layers and each layer consists of multiple objects. The low-end four layers of this kernel are viewed as forming a microkernel named the DREAM microkernel, which essentially supports kernel threads. The DREAM kernel manages available machine time in a highly disciplined manner and involves both lower-level scheduling of kernel threads and high- level scheduling of application processes.  In this thesis research, a detailed design of the DREAM microkernel and its prototype implementation were carried out. Major issues faced during implementation and the implementation approaches adopted are discussed. The basic ideas and architectural principles adopted in the DREAM microkernel are considered to be of fundamental nature and useful in many practical operating system development environments.

Byrne, Timothy Edward. Regional Transportation Implications of Neotraditional Neighborhood Design. M.S., Engineering, 1994. 133 pp. Adviser: Michael G. McNally

The primary goal of this thesis is to test neotraditionalist beliefs that Neotraditional Neighborhood Design (NTND) leads to transportation benefits. The focus reflects on the regional aspects of NTND whereas most of the previous studies have focused on isolated developments or hypothetical subdivision designs. The City of Irvine's transportation forecasting model, the Irvine Transportation Analysis Program (ITAP), is utilized to present a more realistic and regional evaluation of NTND. A comparative assessment is presented based on the transportation impacts of NTND and conventional suburban design which began gaining popularity in the 1960s. Previous studies have shown greater transportation efficiency in neotraditional developments. From this analysis, it is evident that neotraditional design can lead to transportation benefits on the local level. These benefits accrue on a regional level but with diminishing returns to scale.

Marar, Arun Govind. An Improved Transit Assignment Algorithm. M.S., Engineering, 1994. 97 pp. Adviser: Michael G. McNally

This thesis presents exact analytical expressions for incorporation into transit network assignment frameworks. Multipath transit assignment models have been implemented in popular transportation planning software packages such as EMME/2 and QRS II. However, the use of approximate expressions for the expected waiting time at a node and the resulting ridership probabilities raise doubts about the reliability of the results. Recently proposed models are based on a fixed set of transit lines wherein the traveler chooses the strategy that allows the desired destination to be reached at minimum expected cost. One assumption made in existing multipath assignment models is Poisson arrival of transit lines. The reality however conforms more to fixed constant line headways and uniform random passenger arrivals which has been considered to be analytically more difficult than the Poisson arrival case. The exact expressions for waiting time and line probabilities are derived in this thesis based on a theoretical framework involving basic probability functions and properties of constant interarrival distributions. Recursive schemes are developed for computational implementation of these expressions which would facilitate their application to practical assignment. The applications can be either in the line enumeration phase or in the line ridership probability calculations of existing assignment models. Numerical examples are provided to illustrate the application of the recursive schemes.

Marca, James E. An Analysis of New Urban Rail Projects: Forecast Versus Actual Patronage and Impacts. M.S., Engineering, 1994. 205 pp. Adviser: R. Jayakrishnan

The actual ridership levels and economic impacts of new rail transit projects rarely come close to the levels predicted for these during the planning stages of the projects. This thesis reviews the recent experiences of new rail transit systems to examine reasons why this overprediction occurs. The forecasting and development of rail projects in Sacramento, Washington, D.C., Santa Clara County, San Francisco, and San Diego are reviewed in detail. It is concluded that the modeling processes used are highly susceptible to manipulation based on political considerations. A supplementary modeling scheme is proposed which takes a macroscopic rather than microscopic view of a proposed rail transit system. The macroscopic level of analysis clarifies the impact of the most important factors, and allows sensitivity analyses and “sanity checks” to be performed quickly and easily. It is hoped that this will reduce the incidence of overly optimistic rail transit impact predictions.

Rajadhyaksha, Sobodh Sudhir. Investigation and Implementation of a Gradient Projection Algorithm for the Traffic Assignment Problem. M.S., Engineering, 1994. 102 pp. Adviser: R. Jayakrishnan

This thesis takes a fresh look at the arguments against path-enumeration algorithms for the traffic assignment problem and provides the results of a gradient projection method. Traffic assignment problem refers to finding the flows that result in traffic networks under given origin-destination travel demands. The Frank-Wolfe algorithm is traditionally used for urban network assignment. This algorithm finds link-based flow solutions, and often shows slow convergence characteristics. However, it has been considered sufficient for planning purposes thanks to its applicability to large networks. The gradient projection algorithm finds path-based flow solutions and converges faster, but requires much larger storage for large networks.
The motivation behind the research is the orders of magnitude improvement in the availability of computer storage over the last decade. Faster assignment algorithms are necessary for real-time traffic assignment under several of the proposed Advanced Traffic Management Systems (ATMS) strategies. The results show that gradient projection converges in much fewer iterations than the conventional Frank-Wolfe algorithm. The computation time improvement is of the same order for small networks, but reduces as the network sizes increase. The computer implementation issues are discussed carefully, and potential schemes are provided to achieve a 10-fold speed-up for large networks also the algorithm has been applied to networks of up to 1000 nodes on typical computer work station. Certain data structures to save storage and solve the assignment problem for much larger networks are also discussed.

Stack, Robert William. A Multi-User Real-Time Knowledge-Based Expert System for Freeway Incident Response. M.S., Engineering, 1994. 92 pp. Adviser: Stephen G. Ritchie

The development of a prototype, multi-user, networked, real-time knowledge-based expert system to provide decision support to the California Department of Transportation and California Highway Patrol Transportation Management Center operators to more effectively accomplish their traffic management functions is described. The research focuses on developing an operational management and response system using knowledge obtained from experts in the traffic operations field. Knowledge specific to the California Department of Transportation (Caltrans) District 12 in Orange County, California was obtained from traffic management specialists and placed in the expert system. The prototype system, named "FIM" for Freeway Incident Manager, is implemented in G2, a real-time expert system shell. The prototype system is currently in the field at the new Caltrans District 12 Transportation Management Center (TMC) in Orange County .The development of incident detection, verification, and response knowledge bases for the system is described. The implementation of the system using G2 real-time expert system development software on a Sun workstation is described. Feedback from Caltrans District 12 TMC operators concerning the system was enthusiastic, and many enhancements were made to the system based on this feedback. Finally, directions for future research concerning the system are described. Such further research includes the integration of the expert system with other automated TMC functions such as integrated ramp metering and changeable message sign (CMS) control, so that the expert system can provide an integrated transportation management environment for the TMC.

Bilse, Douglas P. Superelevation Rates at Rural Section Curves. M.S., Engineering, 1993. 147 pp. Adviser: Will Recker

The current design practice in California dictates an arbitrary reduction in the superelevation rate of a curved section of rural highway when an intersection with a cross street is present. This thesis attempts to quantify the relative impact of these reduced superelevation rates to vehicles traveling along the rural highway and those traveling in and out of the cross street. The crashes most likely related to superelevation rate were analyzed along 98 Section Curves throughout California, and the prominent collision factor was excessive speed. Computer simulation results indicate that reducing the mainline superelevation rate tends to nominally reduce the limit speed above which loss-of -control crashes become more likely. Model results were applied to existing section curves. Results suggest that the practice of nominal reductions of superelevation rates on Section Curves accomplishes the desired goal of increasing the margins of safety on the cross stint while producing only minor reductions to the failure speeds of vehicles traversing the rural highway.

Deeter, Dean Lee. A Prototype Real-Time Expert System for Arterial Street Incident Management. 1993. M.S., Engineering, 1993. 86 pp. Adviser: Stephen G. Ritchie

An increasing number of cities are developing arterial street Traffic Management Centers (TMC's) capable of controlling city-wide traffic from one central location. One objective is to promote efficient progression and coordination along surface streets. Delays can arise when incidents or other operational problems occur along these arterial streets, possibly leading to gridlock. Presently, these TMC's are limited in their capabilities for response. In the foreseeable future, operators at these TMC's will need to make effective decisions within a short period of time concerning deployment strategies for changeable message signs, highway advisory radio and adaptive signal control for large areas of streets, potentially causing cognitive overload for the TMC operators. This thesis discusses the development of a prototype real-time knowledge based expert system (KBES) to demonstrate the feasibility of assisting the operators of surface street Traffic Management Centers in responding to traffic operational problems. The system, ARTIST (Arterial Real-time Traffic Incident reSponse Tool), replicates three stages of human response to operational problems: problem verification and classification, problem response, and monitoring of effects of responses. This paper describes the development of the verification, response and monitoring knowledge bases for the system, and their testing using G2 real-time KBES development software on a Sun SPARCstation.

van Hengel, Drusilla Ruth. External Variables and Environmental Attitude as Predictors of Personal Reduction of Drive Alone Behavior. M.A., Social Ecology, 1993. 95 pp. Adviser: Joseph F. Di Mento

Reduction of drive alone behavior is identified as an environmentally responsible behavior which can be described using Hardin's Commons Dilemma framework. Prior research in conservation behavior is reviewed. Because it is determined that socio-demographic variables are not reliable indicators of environmentally responsible behaviors, Ajzen and Fishbein's Reasoned Action Theory is utilized as a means by which the intention to reduce drive alone behavior can be better understood. The results suggest that although socio-demographic variables are useful for categorizing commuters by mode choice they do not make a significant contribution to the explanation of intention to use alternative modes above that made by attitudinal and belief variables pertaining to the environment and motor vehicles.

Cohen, Michael I. Analysis of the Potential Benefits of In-Vehicle Navigation Systems to Alleviate Special-Event Congestion. M.S., Engineering, 1992. 110 pp. Adviser: Michael G. McNally

This thesis presents a prototype tool for the simulation of traffic flow under a variety of congestion, information-supply and route guidance scenarios with emphasis on the congestion caused by traffic egress from special-events. The research extends the performance and application of DYNASMART, which performs real-time modeling of traffic flow in which a specified percentage of vehicles are equipped with In-Vehicle Navigation Systems (IVNS). The IVNS information is based on multiple paths updated at regular intervals. The driver response is currently assumed to be boundedly rational. The indifference bandwidth signifies the percentage: savings in travel time necessary to compel a driver to switch from the currently assigned route to the shorter route. The CONTRAM assignment program is used to provide DYNASMART with realistic dynamic equilibrium routes for the initial path assignment of unequipped vehicles.  The special-event traffic scenario is characterized by intense, short-term congestion in the presence of many alternative routes. The network and traffic demand of Anaheim, California are modeled with special-events traffic generated by Anaheim Stadium. A dynamic zonal O-D matrix was estimated using the COMEST program and observed link counts. Evaluations are made on possible benefits of IVHS in the case of no special-event, as well as special-events of various levels of attendance. Benefits of up to ten percent reduction in total travel time are feasible, with most benefits achieved by a market penetration of fifty percent. Higher benefits are possible in heavily congested parts of the network.

Neenan, Barbara. The Interjurisdictional Coordination of Katella Avenue Traffic Signals. M.S., Engineering, 1992. 144 pp. Adviser: Michael G. McNally

In recent years transportation planning has been experiencing an escalating emphasis towards increasing capacity and improving traffic management on urban streets and arterials to combat the effects of congestion. One measure which has proven to be particularly effective is the coordination of traffic signals. However, coordination efforts have traditionally been contained within city boundaries, with little or no communication between adjacent jurisdictions. This study details efforts to obtain regional coordination through the cooperative effort of five cities, the county, the state, and a regional funding agency. A traffic signal coordination project, incorporating a forty-intersection expanse of a major arterial, is documented from concept to implementation. Of particular interest are the financial, administrative, and political implications of coordination; traffic control hardware and software considerations and, coordinated timing plan designs and their impact on traffic flow.

Ramanathan, Balaji Venkatesh. Implementation of a Real-Time Information Processing Algorithm in TRANSYT-7F. M.S., Engineering, 1992. 103 pp. Adviser: Michael G. McNally

TRANSYT -7F is a deterministic and macroscopic traffic signal simulation and optimization program used throughout the world. Current limitations of TRANSYT - 7F include restrictions on the type of controllers in the network and lack of stochastic elements in the simulation. Moreover, the current simulation logic does not capture the dynamic information processing of traffic actuated signals. This research focuses on the development of a new information processing algorithm to complement the simulation logic currently used in TRANSYT -7F. This new logic allows the simulation of the dynamic information processing associated with modem traffic actuated signal controllers. Stochastic elements in traffic flow conditions may be introduced at each timestep. Queue lengths are monitored at every timestep, and associated effects on platoon dispersion are incorporated into the simulation. The resulting model is advanced as a tool for the evaluation of Intelligent Vehicle Highway Systems (IVHS) and Advanced Traffic Management Systems (ATMS) strategies.

Ryan, Sherry. An Assessment of the Potential Impacts of Neotraditional Neighborhood Design on Suburban Accessibility. M.S., Engineering, 1992. 97 pp. Adviser: Michael G. McNally

This thesis has two primary goals: the first is to establish a baseline evaluation of Neotraditiona1 Neighborhood Design (NTND), and the second is to test specific claims made by neotraditionalists as to the possible transportation benefits of employing this particular subdivision design. The first goal will be accomplished by orienting NTND in a historical context, which entails reviewing subdivision design trends of the past century and exploring how NTND has diverged or borrows from them. A literature review of the research and current thinking on the subject will also be provided. The second goal will be accomplished through simulation of hypothetical networks. This allows for the quantification of performance measures which can then be used to draw conclusions about the relative transportation impacts of neotraditional neighborhood design and conventional suburban design.

Zhang, Hongjun. A Real-Time Decision-Support System for Freeway Management and Control. M.S., Engineering, 1992. 110 pp. Adviser: Stephen G. Ritchie

Traffic incidents may cause about 60% of the congestion delay in the US. A number of US cities maintain Traffic Operations Centers (TOCs) for incident detection and response. Currently TOC operators are loaded with many low level tasks in incident management. This process could potentially be automated by an intelligent system. This thesis presents such a system: FRED (Freeway Real-time Expert system Demonstration, Version 2). FRED was designed to provide decision support for TOC operators in a simulated Intelligent Vehicle-Highway System (IVHS) operating environment. Currently FRED operates on an existing network of freeways near Anaheim, California. Based on the traffic information from detectors and an incident report database, FRED is able to detect and verify multiple incidents that occur on the freeway network, and give advice to TOC operators regarding incident response strategies. Such advice includes dispatching of major incident traffic management teams, posting messages on Changeable Message Signs that are installed on both freeways and arterial streets, and controlling traffic via ramp metering. FRED was designed in such a way that new developments in the area of incident management can be easily incorporated into the existing structure.

Dalal, Situ. Multifamily Residential Parking Demand Study: A Research Approach. 1991. M.S., Engineering, 1991. 81 pp. Adviser: Michael G. McNally

This thesis describes a practical research approach employed in conducting a parking demand study for multifamily residential apartment projects in the City of Anaheim. The primary objectives of this study were (a) to determine whether a relationship exists between project size and parking demand as well as between the rent of the apartment and parking demand, and b ) to determine the true parking demand of the existing market-rate apartments, affordable apartments, and senior citizens' apartments. A mail-back survey technique together with a field survey technique were utilized and are described here in. A detailed statistical analysis of the returned surveys did not indicate a relationship between the project size and the parking requirements, however, it was found that as the rent of an apartment increases, parking requirements also increase. Parking ratios were then developed incorporating the effect of apartment rent on parking demand. Field surveys indicated that the recommended parking ratios were more than sufficient to satisfy existing peak weekday and weekend parking demand including visitor parking demand.

Khan, Sarosh Islam. An Integrated Database Management-Expert System to Develop Highway Rehabilitation Projects. M.S., Engineering, 1991. 132 pp. Adviser: Stephen G. Ritchie

Highway resurfacing, restoration, rehabilitation and reconstruction can extend the life of an existing facility. At the same time, enhanced safety can be engineered into a project by making improvements to other features that relate to, for example, bridge improvements, minor widening of lanes and shoulders and minor alternations to horizontal and vertical alignment, in addition to pavement resurfacing and repair. Recent rehabilitation projects in California that were expanded to include safety enhancements resulted in original project-cost estimates underestimating costs, often by millions of dollars. This thesis presents an integrated database-management expert-system of computer program (4RSCOPE), which is a design tool to address this problem. The program has been developed to allow Caltrans engineers to gather data from both office records and field assessments, determine the design features to be included, and thus determine the project scope. The program is expected to become the standard system used by district engineers statewide in California to develop rehabilitation projects.

Landsman, Jaclyn K. Dynamic Travel Environment Change as an Outcome of Travel Related Stresses. M.S., Engineering, 1991. 87 pp. Adviser: Michael G. McNally

The past few decades have seen a tremendous increase in the mobilization of the general populace, especially through the use of the personal automobile. Along with the benefits associated with greater efficiency of movement afforded by car ownership there have been unforeseen drawbacks such as increased congestion, pollution, and a greater reliance on oil supplies which are currently being addressed. In acknowledgment of these efforts, this thesis attempt to shed some light on the motivation behind increases or decreases in car ownership by assessing the impact of travel related stresses on the household unit. Panel data from the Netherlands which incorporated individual travel diaries provided dynamic information on household travel patterns and socio-economic status, including car ownership status. A dynamic model was developed which associated deviations in household travel patterns from normal or mean travel patterns with changes in car ownership. Results indicated that external factors confounded prediction of car ownership change to the extent that deviation in travel patterns appear to be fleeting anomalies rather than harbingers to portentous events.

Prosser, Neil Anthony. A Real-Time Expert System for Freeway Incident Management. M.S., Engineering, 1991. Adviser: Stephen G. Ritchie

Deteriorating traffic conditions in urban areas of the United States have led to a reassessment of traffic management efforts, and in particular have led to interest in the employment of advanced technology in the vehicle-driver-highway environment to improve traffic flow without necessarily expanding the existing transportation infrastructure. Such "smart" car and "smart" road systems are collectively referred to as intelligent vehicle-highway systems (IVHS). In order to l effectively consider and implement future IVHS strategies, urban freeway Traffic Operations Centers (TOC's) will require some form of automated decision support. This thesis presents ~ prototype real-time expert system, FRED (Freeway Real-time' Expert system Demonstration), that is intended to demonstrate the provision of decision support to TOC personnel in the area of freeway incident management (incidents are events that disrupt the flow of traffic such as accidents, spilled loads and disabled vehicles). Incidents often lead to non-recurrent congestion, and the main aim of freeway incident management is to reduce the resulting delay to motorists. The case-study application of FRED to a six-mile section of the SR-91 Riverside Freeway Corridor is described in the thesis. The FRED system incorporates different problem-solving techniques and sophisticated data interfacing to provide what is believed to be a powerful and flexible computing environment for freeway incident management.

Catalina, Anthony James. A Systems Planning Study for the Bunker Hill Transit Tunnel. M.S., Engineering, 1990. 141 pp. Adviser: Stephen G. Ritchie

This thesis describes a methodology that can be followed to conduct a study to select a transit/transportation system to accommodate travel demand and enhance economic activities in the downtown of a major urban area,. The overall objective of the research is to investigate and apply this methodology to identify a series of promising transportation system alternatives and potentially applicable technologies, and to assess their impact on improving congested central business district mobility, environmental quality, and economic health. Current and future land use and transportation conditions, the socioeconomic structure of the city, travel-demand factors, cost factors, and local policy attributes all must be considered in the formulation of the alternatives. These research objectives are achieved through a case study conducted to identify and evaluate alternatives for utilizing the Bunker Hill Transit Tunnel (BH1T) and easements in downtown Los Angeles for mass transit purposes.

Guedes, Jose Eduardo L.S. The Simple Bound Improvement Procedure Applied to the Fixed Charge Transportation Problem. M.S., Engineering, 1990. 75 pp. Adviser: Bruce Lamar

This thesis presents the application of a new bounding technique, incorporated into a Branch-and-Bound (B&B) algorithm, used to solve the Fixed Charge Transportation Problem (FCTP).  The FCTP differs from the Transportation Problem in that the basic assumption of proportionality of cost and flow carried on a link is modified by the addition of a fixed charge incurred for any nonzero flow carried on a link. The bounding technique, known as the Simple Bound Improvement (SBI) procedure, focuses on the continuous variables, tightening bounds and reducing the feasible solution space. The performance of the new procedure was compared with the Standard B&B, and the Penalty Method. A computer code was developed and the comparison was made using a set of test problems obtained from the literature. The performance was compared based on two criteria: the computational time and the storage requirements. The first is measured by the CPU time, and the second by the Number of Candidate Problems solved and the Maximum Depth in the B&B enumeration tree. The results show that the Simple Bound Improvement procedure combined with the Penalty Method improves the CPU time by 52.2% on average, the number of candidate problems by 75.2 %, and the maximum depth by 40.3% when compared to the Standard Branch-and-Bound, and 29.6 %, 52.4% and 36.0%, respectively, when compared to the Penalty Method.  The results are very encouraging and suggest the use of this bounding technique as a tool for other problems that can be modeled as a Fixed Charge Transportation Problem.

Kamojjhala, Kishan Rao. Analysis of Truck Accidents on Freeway Connectors. M.S., Engineering, 1990. 99 pp. Adviser: Will Recker

This research involves examining the impact of freeway-to-freeway connector geometry on the truck loss-of-control accidents. In some previous studies it was established that the characteristics of freeway truck accidents differ significantly from freeway connectors to mainline freeway segments. It was also established that trucks are sensitive to threshold values of MSHTO geometric guidelines. In this particular study, 149 connectors with at least on truck accident were identified for the analysis from the Los Angeles and Orange counties. The truck accidents on these connectors during period 1979-87 consisted of 461 Type-A (overturn or hit object as collision type) and 1363 other accidents. Type-A or Loss-of-Control accidents were assumed to be the category of truck accidents which are related to connector geometry. Logit regression technique was used to explain the impact of connector geometric variables on the truck loss-of-control accidents. The results of the analysis indicated that the length of the critical curve, corresponding cross slope at maximum gradient type of ramp or shape of ramp of which significantly loop shaped curves were found to contribute to truck loss-of-control accidents. In the traffic factors speeding, improper turn found to be very significant reasons for truck loss-of-control accidents. When traffic factors were not included in analysis, it was found that steeper down gradients on connectors were found to be significantly related to loss-of-control or type-a accidents. The study indicates that there is a significant relationship between the connector geometry and it cannot be completely established unless a thorough analysis is conducted which enables to test the combinatorial effects of these geometric features on truck loss-of-control accidents. Further research is recommended in this direction.

Merheb, Omar Assad. Impact of Jersey Median Barriers on the Frequency and Severity of Freeway Accidents. M.S., Engineering, 1990. 93 pp. Adviser: Michael G. McNally

The design and operational advantages of freeway medians has long been a part of traffic engineering practice. Generally regarded as a substantial design safety factor, medians provide space to allow for sufficient recovery time for errant vehicles to correct their path and reenter the traffic stream. Design standards for medians are often difficult to meet, particularly in dense urban areas where the right-of-way cost are significant. In such cases, effective median barriers are warranted for high volume, high speed divided highways. The primary function of such barriers is to prevent out-of-control vehicles from crossing the median into opposing traffic and causing serious head-on accidents. Historically, these median barriers are thought to actually produce an increase in incident frequency, but a decrease in the number of serious accidents. An empirical study of this hypothesis is the subject of this thesis.

Nohalty, Paula D. A Stated Preference Analysis of Selected Freeway Demand Management Strategies. M.S., Engineering, 1990. 101 pp. Adviser: Michael G. McNally

Traditionally, behavioral research in transportation engineering has focused on revealed preference (RP) analyses. RP analyses attempt to explain past behavior or choices. RP analyses have several shortcomings, the foremost being its inability to predict demand for new or altered conditions. A relatively new analysis technique in transportation research is stated preference (SP) analysis, which can be used to predict demand for new or altered conditions.  This research presents a new application of SP analysis to assess a traveler’s preference for selected transportation demand management strategies. It also demonstrates a new approach for estimating SP models; the combined use of non-linear canonical correlation analysis and ordinary least squares (OLS) analysis to estimate models for SP ranking data. This analysis method explicitly recognizes that alternative rankings obtained in SP ranking experiments are ordinal and not continuous variables. This method is compared to ordered probit analysis and OLS regression.

Novarin, Alberto. Structural Equation Models of the Consequences of Freeway Accidents. M.S., Engineering, 1990. 92 pp. Adviser: Will Recker

Structural equation modeling is advanced as a procedure to analyze the causal relationships between variables representing various accident characteristics, such as level-of-service, collision location, number of injuries, total delay caused by the accident, etc. The method is centered on the estimation of three structural equation models of freeway accident consequences, namely probability of injuries, total delay, probability of "influence of alcohol" and probability of "non-driver's fault".  Using data drawn from more than 11,000 crashes that occurred over a 1- year period (November '87 -November '88) on 23 freeways in Los Angeles County, the method is demonstrated by analyzing the relationships between a set of explanatory variables and the endogenous indicators listed above," and between pairs of endogenous indicators.  The results show that the method is a useful tool in uncovering under- lying patterns in accident characteristics.

Phillips, Darryl Curtis. An Investigation of the Effect of Ramp Metering on Congested Freeways. M.S., Engineering, 1990. 380 pp. Adviser: Will Recker

Freeway congestion in many urban areas has become intolerable, and much effort is being put into researching ways to alleviate the problem. Ramp metering is in widespread usage, and should be evaluated to ensure that it is being used to best effect. Computer simulation was used for this study to avoid problems inherent in field data. The microscopic model INTRAS was used, and was modified to allow for a probabilistic diversion among alternate routes to a destination. Three cases were studied: a freeway with a bottleneck, a freeway with drivers able to avoid the freeway, and a freeway with drivers ultimately entering the freeway at some point. A sample network was developed for each case. Multiple simulation runs were made for each case, using varied metering strategies on the various ramps, with additional runs being made to evaluate the possible influence of other factors. Ramp metering was found to be quite effective in reducing delay caused by congestion, particularly when used in the vicinity of capacity bottlenecks and in areas where drivers are able to avoid the freeway completely. The implementation of ramp metering in all situations showed a reduction in delay over nonmetered freeways, perhaps due to more efficient merge operation. Recommendations are made for ways to maximize the beneficial impact of ramp metering. Further research is called for to determine the willingness of drivers to divert to alternate paths to minimize travel time.