Applicability Of Microscopic Traffic Simulation Models In Vehicle Emission Estimations

With the increasingly worsening of traffic emission pollution problems today, the effort that estimate emissions by an integration of traffic simulation models and emission models has gradually been becoming a fast-evolving research area. However, due to the lack of effective methods and indicators to characterize traffic behaviors, the estimation accuracy of emissions by such integration has not been verified and/or evaluated. In this context, this thesis presents a method to evaluate the estimation accuracy emissions based on the best explanatory parameter of energy consumption and emissions-vehicle specific power (VSP) distribution, and conducts the study on the applicability of microscopic traffic simulation models in vehicle emission estimations.First, the thesis synthesizes the state-of-the-art and state-of-the-practice on microscopic traffic simulation models, emission estimation models, and their integration, on the basis of which it discusses conventional methods and potential problems, establishing the context of the research in the thesis considering the trend of emission models.Second, after reviewing the VSP-distribution-based approach for emission estimations, the thesis conducts an analysis on the patterns of VSP distribution using both GPS measured real data and VISSIM simulated data, and then calculates their corresponding emissions. Subsequently, the errors between simulated and real-world results are measured by two indicators, the mean square error of VSP distributions and the relative error of emission factors. It is found from these analyses that the simulation results after the simulation models are calibrated by the traditional calibration methods cannot represent the microscopic driving behaviors of the real traffic operations from the perspective of emission estimations, leading to errors as high as 82.8%,53.6% and 29.6% for NOx, HC, and CO emission factors respectively. Consequently, it is necessary to conduct research on the calibration method of simulation models for emission estimations.Then, the sensitive analysis of parameters of simulation models on VSP distributions is conducted.7 parameters of simulation models, such as the desired speed distribution, are adjusted for a total of 14 times to determine their effect on simulation results using the mean square error of VSP distribution and its relative change rate as indicators. Further, the feasibility study of the parameter calibration of simulation models for the emission estimation is conducted. The study illustrates that, systematic errors exist in using microscopic traffic simulation models to represent second-by-second real driving behaviors. The errors cannot be reduced by the method of parameter calibration in simulation models. As a result, it is necessary to analyze reasons that cause errors from the internal mechanism of simulation models.Finally, the source of errors in using microscopic traffic simulation models to estimate emissions is explored by observing and analyzing the differences of speed and acceleration curves between simulated data and real-world data. It is found that the mechanisms of car-following and lane-changing models, especially in the description of accelerations, are the main reason that causes simulation errors. Based on the above analysis, several recommendations are proposed.