Mathematical models to predict vehicle fuel consumption and emission metrics are presented in this thesis. Vehicle fuel consumption and emissions are complex functions to be approximated in practice due to numerous variables affecting their outcome. Vehicle energy and emissions are particularly sensitive to changes in vehicle state variables such as speed and acceleration, ambient conditions such as temperature, and driver control inputs such as acceleration pedal position and gear shift speeds, among others. Recent empirical studies have produced large amounts of data concerning vehicle fuel consumption and emissions rates and offer a wealth of information to transportation planners. Unfortunately, unless simple relationships are found between fuel consumption and vehicle emission metrics, their application in microscopic traffic and macroscopic planning models becomes prohibitive computationally. This thesis describes the development of microscopic energy and emission models using nonlinear multiple regression and neural network techniques to approximate vehicle fuel consumption and emissions field data. The energy and emission models described in this thesis utilized data collected by the Oak Ridge National Laboratory. The data include microscopic fuel consumption and emission measurements (CO, HC, and NOx) for eight light duty vehicles as a function of vehicle speed and acceleration. The thesis describes modeling processes and the tradeoffs between model accuracy and computational efficiency. Model verification results are included for two vehicle driving cycles. The models presented estimate vehicle fuel consumption within 2.5% of their actual measured values. Vehicle emissions errors fall in the range of 3-33% with correlation coefficients ranging between 0.94 and 0.99. Future transportation planning studies could also make use of the modeling approaches presented in the thesis. The models developed in this study have been incorporated into a microscopic traffic simulation tool called INTEGRATION to further demonstrate their application and relevance to traffic engineering studies. Two sample Intelligent Transportation Systems (ITS) application results are included. In the case studies, it was found that vehicle fuel consumption and emissions are more sensitive to the level of vehicle acceleration than to the vehicle speed. Also, the study shows signalization techniques can reduce fuel consumption and emissions significantly, while incident management techniques do not affect the energy and emissions rates notably. / Master of Science

The transportation sector is a major contributor to U.S. fuel consumption and emissions. Consequently, assessing the environmental impacts of transportation activities is essential for air-quality improvement programs. Current state-of-the-art models estimate vehicle emissions based on typical urban driving cycles. Most of these models offer simplified mathematical expressions to compute fuel consumption and emission rates based on average link speeds while ignoring transient changes in a vehicle's speed and acceleration level as it travels on a highway network. Alternatively, microscopic models capture these transient effects; however, the application of microscopic models may be costly and time consuming. Also, these tools may require a level of input data resolution that is not available. Consequently, this dissertation attempts to fill the void in energy and emission modeling by a framework for modeling vehicle fuel consumption and emissions mesoscopically. This framework is utilized to develop the VT-Meso model using a number of data sources. The model estimates average light-duty vehicle fuel consumption and emission rates on a link-by-link basis using up to three independent variables, namely: average travel speed, average number of stops per unit distance, and average stop duration. The mesoscopic model utilizes a microscopic vehicle fuel consumption and emission model that was developed at Virginia Tech to compute mode-specific fuel consumption and emission rates. This model, known as VT-Micro, predicts the instantaneous fuel consumption and emission rates of HC, CO and NOx of individual vehicles based on their instantaneous speed and acceleration levels. The mesoscopic model utilizes these link-by-link input parameters to construct a synthetic drive cycle and compute average link fuel consumption and emission rates. After constructing the drive cycle, the model estimates the proportion of time that a vehicle typically spends cruising, decelerating, idling and accelerating while traveling on a link. A series of fuel consumption and emission models are then used to estimate the amount of fuel consumed and emissions of HC, CO, CO2, and NOX emissions for each mode of operation. Subsequently, the total fuel consumed and pollutants emitted by a vehicle while traveling along a segment are estimated by summing across the different modes of operation and dividing by the distance traveled to obtain distance-based average vehicle fuel consumption and emission rates. The models are developed for normal and high emitting vehicles. The study quantifies the typical driver deceleration behavior for incorporation within the model. Since this model constructs a drive cycle which includes a deceleration mode, an accurate characterization of typical vehicle deceleration behavior is critical to the accurate modeling of vehicle emissions. The study demonstrates that while the deceleration rate typically increases as the vehicle approaches its desired final speed, the use of a constant deceleration rate over the entire deceleration maneuver is adequate for environmental modeling purposes. Finally, the study validates the model on a freeway and urban arterial network. The results demonstrate that the model provides accurate estimates of vehicle fuel consumption and emission rates and is adequate for the evaluation of transportation operational projects. / Ph. D.

The transportation sector has a significant impact on the environment both nationally and globally since it is a major vehicle fuel consumption and emissions contributor. These emissions are considered a major environmental threat. Consequently, decision makers desperately need tools that can estimate vehicle emissions accurately to quantify the impact of transportation operational projects on the environment. Microscopic fuel consumption and emission models should be capable of computing vehicle emissions reliably to assist decision makers in developing emission mitigation strategies. However, the majority of current state-of-the-art models suffer from two major shortcomings, namely; they either produce a bang-bang control system because they use a linear fuel consumption versus power model or they cannot be calibrated using publicly available data and thus require expensive laboratory or field data collection. Consequently, this dissertation attempts to fill this gap in state-of-the-art emission modeling through a framework based on the Virginia Tech Comprehensive Power-Based Fuel consumption Model (VT-CPFM), which overcomes the above mentioned drawbacks. Specifically, VT-CPFM does not result in a bang-bang control and can be calibrated using publicly available vehicle and road pavement parameters. The main emphasis of this dissertation is to develop a simple and reliable emission model that is able to compute instantaneous emission rates of carbon monoxide (CO), hydrocarbons (HC) and nitrogen oxides (NOx) for the light-duty vehicles (LDVs) and heavy-duty diesel trucks (HDDTs). The proposed extension is entitled Virginia Tech Comprehensive Power-Based Fuel consumption and Emission Model (VT-CPFEM). The study proposes two square root models where the first model structure is a cubic polynomial function that depends on fuel estimates derived solely from VT-CPFM fuel estimates, which enhances the simplicity of the model. The second modeling framework combines the cubic function of the VT-CPFM fuel estimates with a linear speed term. The additional speed term improves the accuracy of the model and can be used as a reference for the driving condition of the vehicle. Moreover, the model is tested and compared with existing models to demonstrate the robustness of the model. Furthermore, the performance of the model was further investigated by applying the model on driving cycles based on real-world driving conditions. The results demonstrate the efficacy of the model in replicating empirical observations reliably and simply with only two parameters. / Ph. D.