Geographic Information Systems (GISs) provide a powerful framework for various tasks of transit management such as planning, performance evaluation, and marketing. GIS may be used to solve complex planning problems, assist in operations planning, and meet other management and operational needs. However, due to the changing nature of transit planning and operational data, transit planners and operators need to analyze the data over time, which requires a temporal GIS that is capable of storing, manipulating, and analyzing changes with respect to both time and space. Temporal GIS will allow planners and transit operators to analyze data within a certain time frame defined by the user and compare it to those in another time frame or to the current available data. This ability will improve prediction of future transit demand, evaluation of past operations, and analysis of day-to-day operations and management. This research entails the study of the applications of temporal GIS in transit planning and management. Transit management tasks are examined to identify those that may potentially benefit from temporal GIS. Their process, data needed, and possible applications are investigated. Examples of temporal GIS applications have been developed.

The estimation of pavement layer moduli through the use of an artificial neural network is a new concept which provides a less strenuous strategy for backcalculation procedures. Artificial Neural Networks are biologically inspired models of the human nervous system. They are specifically designed to carry out a mapping characteristic. This study demonstrates how an artificial neural network uses non-destructive pavement test data in determining flexible pavement layer moduli. The input parameters include plate loadings, corresponding sensor deflections, temperature of pavement surface, pavement layer thicknesses and independently deduced pavement layer moduli.

The detailed organic composition of atmospheric fine particles with an aerodynamic diameter smaller than or equal to 2.5 micrometers (PM 2.5) is an integral part of the knowledge needed in order to fully characterize its sources and transformation in the environment. For the study presented here, samples were collected at 3-hour intervals. This high time resolution allows gaining unique insights on the influence of short- and long-range transport phenomena, and dynamic atmospheric processes. A specially designed sequential sampler was deployed at the 2002-2003 Baltimore PM Supersite to collect PM2.5 samples at a 3-hourly resolution for extended periods of consecutive days, during both summer and winter seasons. Established solvent-extraction and GC-MS techniques were used to extract and analyze the organic compounds in 119 samples from each season. Over 100 individual compounds were quantified in each sample. For primary organics, averaging the diurnal ambient concentrations over the sampled periods revealed ambient patterns that relate to diurnal emission patterns of major source classes. Several short-term releases of pollutants from local sources were detected, and local meteorological data was used to pinpoint possible source regions. Biogenic secondary organic compounds were detected as well, and possible mechanisms of formation were evaluated. The relationships between the observed continuous variations of the concentrations of selected organic markers and both the on-site meteorological measurements conducted parallel to the PM2.5 sampling, and the synoptic patterns of weather and wind conditions were also examined. Several one-to-two days episodes were identified from the sequential variation of the concentration observed for specific marker compounds and markers ratios. The influence of the meteorological events on the concentrations of the organic compounds during selected episodes was discussed. It was observed that during the summer, under conditions of pervasive influence of air masses originated from the west/northwest, some organic species displayed characteristics consistent with the measured PM2.5 being strongly influenced by the aged nature of these long-traveling background parcels. During the winter, intrusions from more regional air masses originating from the south and the southwest were more important.

Self-consolidating concrete has been described as the most revolutionary development in concrete technology in several decades with the ability to flow freely through closely spaced reinforcements, expel entrapped air and self compact without vibration. Since it was first developed in Japan in the early 1980's, major development in the chemical admixture technology has made SCC more viable. An experimental study was conducted to identify the mechanical properties of SCC by optimizing the use of pozzolanic materials and local aggregates with some proposed statistical models. The research was focused to investigate compressive strength, splitting tensile strength, modulus of elasticity and drying shrinkage behavior of concrete. The results were established experimentally and compared with the available SCC research data based on extensive literature study. Besides the improved mechanical performance, results indicate that the use of pozzolanic materials and local aggregate in SCC is recommended in terms of its cost benefit value.

Traffic incidents are a major source of traffic congestion on freeways. Freeway traffic diversion using pre-planned alternate routes has been used as a strategy to reduce traffic delays due to major traffic incidents. However, it is not always beneficial to divert traffic when an incident occurs. Route diversion may adversely impact traffic on the alternate routes and may not result in an overall benefit. This dissertation research attempts to apply Artificial Neural Network (ANN) and Support Vector Regression (SVR) techniques to predict the percent of delay reduction from route diversion to help determine whether traffic should be diverted under given conditions. The DYNASMART-P mesoscopic traffic simulation model was applied to generate simulated data that were used to develop the ANN and SVR models. A sample network that comes with the DYNASMART-P package was used as the base simulation network. A combination of different levels of incident duration, capacity lost, percent of drivers diverted, VMS (variable message sign) messaging duration, and network congestion was simulated to represent different incident scenarios. The resulting percent of delay reduction, average speed, and queue length from each scenario were extracted from the simulation output. The ANN and SVR models were then calibrated for percent of delay reduction as a function of all of the simulated input and output variables. The results show that both the calibrated ANN and SVR models, when applied to the same location used to generate the calibration data, were able to predict delay reduction with a relatively high accuracy in terms of mean square error (MSE) and regression correlation. It was also found that the performance of the ANN model was superior to that of the SVR model. Likewise, when the models were applied to a new location, only the ANN model could produce comparatively good delay reduction predictions under high network congestion level.

A new method for measurement of bedload transport velocity using an acoustic Doppler current profiler (aDcp) is evaluated. Conventional bedload sampling involves physical samplers that are notoriously inaccurate and of limited use for characterizing the spatial and temporal distribution of bedload. The new technique utilizes the bias in aDcp bottom tracking due to movement on the river bed. This bias can be determined by comparing the boat velocity by differential global positioning system (DGPS) and by bottom tracking. The evaluation of the method had four components: field demonstration, laboratory calibration, development of an error model to separate the bedload velocity signal from the noise in the data, and use of the method in the field to characterize the spatial distribution of bedload transport velocity. The field demonstration involved concurrent aDcp and physical sampler measurements of bedload transport at stationary sampling stations in the gravel-bed reach of Fraser River. Mean bedload transport velocities measured using an aDcp were shown to correlate with mean bedload transport rates estimated with the physical samplers (r²=0.93, n=9). The laboratory calibration involved the creation of a synthetic bedload by dragging small cobbles over an artificial river-bed in a towing tank. It was shown that, despite high variability in the measurements that was due to instrument noise, the aDcp can separately estimate the mean magnitude and direction of the synthetic bedload velocity. However, due to excessive noise in individual beam velocities that did not appear to be present in the field data, the bedload velocity in the direction of transport was underpredicted by 79% on average. The error model is a new numerical method to probabilistically deconvolve the bedload velocity signal and the noise in the data. For data from Fraser River and from Norrish Creek, the probability density functions of the highly positively-skewed bedload velocity signal and the acoustic noise were resolved. The bedload velocity signal could be modelled as either a compound Poisson-gamma distribution or a gamma distribution. The acoustic noise was normally distributed and comparable to typical noise levels for aDcp water velocity measurements. Finally, field measurements from a moving boat in a sand-bed reach and a gravel-bed reach of Fraser River were used to characterize the spatial distribution of bedload transport velocity. The bedload velocity spatial distribution was shown to be significantly correlated with the spatial distributions of near-bed water velocity and depth averaged water velocity. Smoothing was achieved by both block averaging and kriging, which revealed coherent patterns in the bedload velocity spatial distribution. / Applied Science, Faculty of / Civil Engineering, Department of / Graduate

A mechanics-based model has been developed which predicts the timing of required ballast maintenance based on track settlement and roughness. With user-defined track material properties, loading characteristics, and maintenance technique to be applied, the model determines the rate of track roughness increase and applies the desired maintenance when such roughness exceeds a limit. This analysis is continued, usually over the life of the ballast, and the related life cycle cost is calculated. For another ballast material or maintenance technique, a lower or higher cost may be realized. In this way, the model user may observe the effect upon ballast maintenance and cost of using concrete rather than wood ties, a heavier axle load, or increasing the track stiffness. Because the life cycle cost is used, the frequent mistake of determining purchasing decisions based upon the least cost in the present is avoided. Instead, the model allows the user to optimize by considering the associated costs incurred during the component life.

The objective of this dissertation is to study the characteristics of subgrade soils under repeated loading and their influence on the design and performance of railway track. The roles of the subgrade as the track foundation are discussed. The mechanisms for different types of subgrade distress are examined. A method for the estimation of resilient modulus of subgrade soils is developed. This method takes into account the influence of soil physical state, stress state and soil type. The effect of soil physical state is quantified by combinations of two equations relating resilient modulus to moisture content. The effect of stress state is determined by equations relating resilient modulus at optimum moisture content to deviator stress. A method for the prediction of cumulative plastic strain of subgrade soils is developed which takes into account the influence of number of load cycles, stress state, soil physical state, and soil type. The stress state is represented by deviator stress. The influence of soil physical state is indirectly represented by the soil static strength. The parameters required for the prediction model are recommended for different soil classifications. Based on analysis with the GEOTRACK computer model, the effects of superstructure and substructure factors influencing track modulus are illustrated. The relationship between the track modulus and the track performance is analyzed. GEOTRACK is also used for analysis of the stress pulses generated in the subgrade under moving wheel loads. Two design methods are developed for the track granular layer thickness. The first one is intended to prevent the progressive shear failure of the subgrade and the other one is intended to prevent the excessive plastic deformation in the subgrade. The major considerations for these two methods include the magnitudes of the mix of dynamic wheel loads and the corresponding numbers of repeated loads, the soil type, the soil static strength, the subgrade thickness, the subgrade resilient modulus, and the resilient modulus for the granular layer. The granular layer thickness design method is evaluated by field test results. The field performance of the test track is reasonably interpreted by the design method developed in this dissertation.

At the National Geotechnical Experimentation Site (NGES) on the University of Massachusetts Amherst Campus, 44 pipe piles with diameters in the range of 60 to 168 mm and lengths of 1.5 to 10.7 m, were installed and load tested. The majority of piles were installed in a moderately plastic, varved clay crust with overconsolidation ratios in the range of 4 to 9. Open-ended and closed-ended (60$\sp\circ$ apex cone) piles were installed by driving and jacking. An extensive program of laboratory and in situ soil testing was also completed. Axial compressive load tests to failure were conducted at a predetermined time after installation and some piles were subjected to repeat load tests. Additionally, load tests were conducted on pile cone tips, independent of the pile shaft. Average unit skin friction was backcalculated from the interpreted pile failure load and estimated values of end-bearing. The investigation revealed that: (1) pile driving resulted in more intrusion of soil into the pile (less plugging), as compared to pile jacking; (2) plugging was a function of pile diameter and soil characteristics; (3) jacking resulted in higher skin friction capacity than driving; (4) skin friction capacity of piles depended on the degree of plugging; (5) pile skin friction capacity appeared to be a function of the lateral reconsolidation effective stress; (6) the operative lateral effective stress acting on the piles at failure appeared to vary over wide limits between values close to the at rest earth pressure (K$\sb{\rm o}$) and the reconsolidation lateral earth pressure K$\sb{\rm o}$), estimated from in situ test results; (7) the operative coefficient of friction at the soil-pile interface at failure was best estimated from the results of interface direct shear tests using remolded soil; and (8) repeat load tests on piles resulted in successively higher failure loads until the third load test, after which the capacity dropped slightly. The mechanism responsible appeared to be an increase in end-bearing. A new effective stress analysis for pile skin friction was proposed. The analysis allows for consideration of the installation method, degree of plugging during open-ended penetration and the estimated lateral stress conditions around displacement piles installed in overconsolidated clay.

State highway agencies in New England have recently adopted or are in the process of adopting performance specifications for new bituminous concrete pavements and overlays. These performance specifications cover issues such as asphalt content, in-place density, and initial ride quality. Associated with these specifications often are payment bonuses and penalties. Given the increasing trend to employ performance specifications, especially in initial ride quality, this dissertation focuses on the following questions. What are the factors that affect initial ride quality in new pavements and overlays, and what can be done to minimize the effects of those factors determined to be detrimental? What measurement devices, analytical methods, and measures are being employed to asses initial ride quality, and how suitable are these devices, analytical methods, and measures? Currently, the trend in the New England region is to measure initial ride quality of new pavements and overlays with the International Roughness Index (IRI). Although implementation of the IRI is gaining acceptance in the industry, there is research that questions its use for this purpose. Included in this dissertation are two field studies conducted in Massachusetts. The objective of these studies was twofold: (1) to focus on questions associated with paving operations and procedures, and their effect on roughness, and (2) to examine the suitability of an automatic road analyzer (ARAN) for the detection and measurement of initial roughness in new overlays. Based on these two field studies, it was determined that paver stops during the placement of a bituminous concrete overlay produced roughness that was easily detectable and measurable by a skilled inspector. It was also determined that the while the ARAN was capable of accurate and precise measurements of overall initial ride quality, it was not sufficiently sensitive to determine the exact location of this roughness caused by paver stops. For this reason, it was determined that the ARAN be used in conjunction with other measurement techniques.