Civil & Environmental Engineering Theses and Dissertations
Permanent URI for this collectionhttp://hdl.handle.net/1903/2753
Browse
8 results
Search Results
Item NON-DESTRUCTIVE TESTING FOR QUALITY ASSURANCE OF CONCRETE & PERFORMANCE PREDICTION OF BRIDGE DECKS WITH MACHINE LEARNING(2022) Ghahri Saremi, Setare; Goulias, Dimitrios DG; Civil Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Non-destructive testing (NDT) methods are particularly valuable in the quality assurance (QA) process since they do not interfere with production of concrete and reduce testing time and cost. NDTs can provide early warnings in meeting strength requirements at early ages of concrete as well as long term strength. NDTs are also valuable in providing evaluation of health of in-service infrastructures such as bridge and pavement. The results of this study can be used for potential adoption of an NDT-based QA plan. Their adoption in QA will provide the opportunity to test a larger portion of concrete during assessment without a significant increase in QA cost and testing time. To achieve that purpose, the selected NDTs should be fast, accurate, reliable and simple to run. The NDT methods explored in this study included infrared thermography, ultrasonic pulse velocity (UPV), fundamental resonance frequency, rebound hammer, ground penetrating radar (GPR), and ultrasonic pulse echo (UPE). Different sets of NDTs were selected in each experimental study undertaken in this dissertation appropriate to the research objectives and goals in each case. For strength gain monitoring, (i.e., maturity modeling during early ages of hydration), the suggested NDTs need to provide an assessment of the mechanical properties of concrete. To assess the concrete quality during production and/or construction the selected NDTs should rapidly identify potential issues concerning uniformity and/or the presence of production and placement defects. For evaluating the condition of concrete bridge decks with asphalt overlays, GPR response was used to detect layer thickness and concrete quality and to evaluate reinforcement condition. For addressing the transition from lab to field results, machine learning modeling was used to predict the structure condition. Therefore, two artificial neural network (ANN) models were proposed and assessed in this study to predict the condition of bridge decks in Maryland and Massachusetts. Thus, the objectives of this research were to identify and assess alternative NDT methods that can be used in: i) monitoring and/or estimating strength gain (i.e., maturity modeling) in concrete; ii) evaluating concrete uniformity and production quality; iii) detecting and measuring the extent of delamination in concrete slab representing small scale field conditions; iv) evaluating GPR in assessing the condition of pavement layers, concrete quality and reinforcement in bridge decks; and v) employing machine learning modeling to predict the condition of bridge decks.Item EVENT-DRIVEN OPERATION OF DISTRIBUTED SYSTEMS WITH ARTIFICIAL INTELLIGENCE TECHNOLOGIES AND BEHAVIOR MODELING(2022) Montezzo Coelho, Maria Eduarda; Austin, Mark A; Civil Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)This dissertation aims to enhance decision making in urban settings by integrating artificial intelligence technologies with distributed behavior modeling. Today’s civil engineering systems are far more heterogeneous than their predecessors and may be connected to other types of systems in completely new ways, making the task of system design, analysis and integration of multi-disciplinary concerns much more difficult than in the past. These challenges can be addressed by combining machine learning formalisms and semantic model representations of urban systems, that work side-by-side in collecting data, identifying events, and managing city operations in real-time. We exercise the proposed approach on a problem involving anomaly detection in an urbanwater distribution system and a metrorail system.Item National-Level Origin-Destination Estimation Based on Passively Collected Location Data and Machine Learning Methods(2021) Pan, Yixuan; Zhang, Lei; Civil Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Along with the development of information and positioning technologies, there emerges passively collected location data that contain location observations with time information from various types of mobile devices. Passive location data are known for their large sample size and continuous behavior observations. However, passive location data also require careful and comprehensive data processing and modeling algorithms for privacy protection and practical applications.In the meantime, the travel demand estimation of origin-destination tables is fundamental in transportation planning. There lacks a national origin-destination estimation that provides time-dependent travel behaviors for all travel modes. Passive collected location data appeal to researchers with the potential of serving as the data source for large-scale multimodal travel demand estimation and monitoring. This research proposes a comprehensive set of methods for passive location data processing including data cleaning, activity location and purpose identification, trip-level information identification, social demographic imputation, sample weighting and expansion, and demand validation. For each task, the thesis evaluates the state-of-the-practice and state-of-the-art algorithms, and develops an applicable method jointly considering the different features of various passive location data sources and the imputation accuracy. The thesis further examines the viability of the method kit in a national-level case study and successfully derives the national-level origin-destination estimates with additional data products, such as trip rate and vehicle miles traveled, at different geographic levels and temporal resolutions.Item INTRODUCING A GRAPH-BASED NEURAL NETWORK FOR NETWORKWIDE TRAFFIC VOLUME ESTIMATION(2021) Zahedian, Sara; Haghani, Ali; Civil Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Traffic volumes are an essential input to many highway planning and design models; however, collecting this data for all the roads in a network is not practical nor cost-effective. Accordingly, transportation agencies must find ways to leverage limited ground truth count data to obtain reasonable estimates at scale on all the network segments. One of the challenges that complicate this estimation is the complex spatial dependency of the links’ traffic state in a transportation network. A graph-based model is proposed to estimate networkwide traffic volumes to address this challenge. This model aims to consider the graph structure of the network to extract its spatial correlations while estimating link volumes. In the first step, a proof-of-concept methodology is presented to indicate how adding the simple spatial correlation between the links in the Euclidian space improves the performance of a state-of-the-art volume estimation model. This methodology is applied to the New Hampshire road network to estimate statewide hourly traffic volumes. In the next step, a Graph Neural Network model is introduced to consider the complex interdependency of the road network in a non-Euclidean domain. This model is called Fine-tuned Spatio-Temporal Graph Neural Network (FSTGCN) and applied to various Maryland State networks to estimate 15-minute traffic volumes. The results illustrate significant improvement over the existing state-of-the-art models used for networkwide traffic volume estimation, namely ANN and XGBoost.Item PARAMETRIC AND NON-PARAMETRIC APPROACHES FOR THE PREDICTION OF THE DIFFUSION OF THE ELECTRIC VEHICLE(2020) Bas Vicente, Javier; Cirillo, Cinzia; Zofío Prieto, José Luis; Civil Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Driven by environmental awareness and new regulations for fuel efficiency, electric vehicles (EVs) have significantly evolved in the last decade, yet their market share is still much lower than expected. In addition to understanding the reasons for this slow market penetration, it is crucial to have appropriate tools to correctly predict the diffusion of this innovative product. Recent works in forecasting the EV market combine substitution and diffusion models, where discrete choice specifications are used to address the former, and Bass-type to account for the latter. However, these methodologies are not dynamic and do not consider the fact that innovation occurs through social channels among members of a social system. This research presents two advanced methodologies that make use of real data to evaluate the adoption of the EVs in the State of Maryland. The first consists of a disaggregated substitution model that considers social influence and social conformity, which is then embedded in a diffusion model to predict electric vehicle sales. The second, in contrast, relies on non-parametric machine learning techniques for the classification of potential EV purchasers. Both make use of data collected through a stated choice experiment specifically designed to capture the inclination of users towards EVs.Item Uncertainty quantification of a radiative transfer model and a machine learning technique for use as observation operators in the assimilation of microwave observations into a land surface model to improve soil moisture and terrestrial snow(2020) Park, Jongmin; Forman, Barton A; Civil Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Soil moisture and terrestrial snow mass are two important hydrological states needed to accurately quantify terrestrial water storage and streamflow. Soil moisture and terrestrial snow mass can be measured using ground-based instrument networks, estimated using advanced land surface models, and retrieved via satellite imagery. However, each method has its own inherent sources of error and uncertainty. This leads to the application of data assimilation to obtain optimal estimates of soil moisture and snow mass. Before conducting data assimilation (DA) experiments, this dissertation explored the use of two different observation operators within a DA framework: a L-band radiative transfer model (RTM) for soil moisture and support vector machine (SVM) regression for soil terrestrial snow mass. First, L-band brightness temperature (Tb) estimated from the RTM after being calibrated against multi-angular SMOS Tb's showed good performance in both ascending and descending overpasses across North America except in regions with sub-grid scale lakes and dense forest. Detailed analysis of RTM-derived L-band Tb in terms of soil hydraulic parameters and vegetation types suggests the need for further improvement of RTM-derived Tb in regions with relatively large porosity, large wilting point, or grassland type vegetation. Secondly, a SVM regression technique was developed with explicit consideration of the first-order physics of photon scattering as a function of different training target sets, training window lengths, and delineation of snow wetness over snow-covered terrain. The overall results revealed that prediction accuracy of the SVM was strongly linked with the first-order physics of electromagnetic responses of different snow conditions. After careful evaluation of the observation operators, C-band backscatter observations over Western Colorado collected by Sentinel-1 were merged into an advanced land surface model using a SVM and a one-dimensional ensemble Kalman filter. In general, updated snow mass estimates using the Sentinel-1 DA framework showed modest improvements in comparison to ground-based measurements of snow water equivalent (SWE) and snow depth. These results motivate further application of the outlined assimilation schemes over larger regions in order to improve the characterization of the terrestrial hydrological cycle.Item Extreme Precipitation Projections in a Changing Climate(2019) Hu, Huiling; Ayyub, Bilal M.; Civil Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Global climate is changing at an alarming rate, with an increase in heat waves, wildfires, extreme weather events, and rising sea levels, which could cost the United States billions of dollars in lost labor, reduced crop yields, flooding, health problems, and crumbling infrastructure. Reports by hundreds of US climate scientists from 13 federal agencies in the Fourth National Climate Assessment (2018) predict that the US economy will shrink by as much as 10% by the end of the century if global warming continues with current trends. Extreme precipitation, in particular, has led to significant damage through flooding, bridge scouring, land-slides, etc.; therefore, it is critical to develop accurate and reliable methods for future extreme precipitation projection. This dissertation proposes new methods of improved projections of such extremes by appropriately accounting for a changing climate. First, this dissertation studies how to model extreme precipitation using Markov Chains and dynamic optimization. By incorporating day-to-day serial dependency and dynamic optimization, the model improves the accuracy of extreme precipitation analysis significantly. The dissertation also examines future projections of extreme precipitation. State-of-the-art methods for future precipitation projections are based on downscaled Global Climate Models (GCMs), which are not always accurate for extreme precipitation projection. This work studies accuracy when using downscaled GCMs for extreme precipitation and designed new methods based on copulas to improve the accuracy. Finally, the above methods are applied to the analysis of future trends of intensity-duration-frequency (IDF) curves, which, in turn, have extensive applications in designing drainage systems. To incorporate geographic influence on local areas, a machine-learning-based solution is proposed and validated. The results show that the gradient boosting tree can be used to accurately project future IDF curves for short durations. It is also projected that short-duration intensity will increase up to 23% for the selected representative stations in this century. In summary, this dissertation systemically studies different aspects of improvements and applications of extreme precipitation projection. By using mathematical models, such as copula and Markov Chains as well as various machine-learning models (i.e., gradient boosting tree), extreme precipitation projection can be made significantly more reliable for use.Item SENSITIVITY ANALYSIS OF SUPPORT VECTOR MACHINE PREDICTIONS OF PASSIVE MICROWAVE BRIGHTNESS TEMPERATURES OVER SNOW-COVERED TERRAIN IN HIGH MOUNTAIN ASIA(2018) Ahmad, Jawairia; Forman, Barton A; Civil Engineering; Digital Repository at the University of Maryland; University of Maryland (College Park, Md.)Spatial and temporal variation of snow in High Mountain Asia is very critical as it determines contribution of snowmelt to the freshwater supply of over 136 million people. Support vector machine (SVM) prediction of passive microwave brightness temperature spectral difference (ΔTb) as a function of NASA Land Information System (LIS) modeled geophysical states is investigated through a sensitivity analysis. AMSRE ΔTb measurements over snow-covered areas in the Indus basin are used for training the SVMs. Sensitivity analysis results conform with the known first-order physics. LIS input states that are directly linked to physical temperature demonstrate relatively higher sensitivity. Accuracy of LIS modeled states is further assessed through a comparative analysis between LIS derived and Advanced Scatterometer based Freeze/Melt/Thaw categorical datasets. Highest agreement of 22%, between the two datasets, is observed for freeze state. Analyses results provide insight into LIS’s land surface modeling ability over the Indus Basin.