Correlation analysis was performed to investigate the effects of drive cycle characteristics on distance-specific emissions (g/mile) and fuel economy (mpg) and consequently determine the most influential cycle metrics for modeling. A detailed analysis of linear and non-linear correlations was performed among cycle metrics to avoid collinearity and reduce the number of variables. The order of importance of the selected cycle metrics was determined. Results show that average speed with idle, number of stops per mile, percentage idle, and kinetic intensity were the most important cycle metrics affecting emissions and fuel economy. Preliminary regression analysis reinforced their importance for emissions modeling purposes.

Modeling transit bus emissions and fuel economy requires a large amount of experimental data over wide ranges of operational conditions. Chassis dynamometer tests are typically performed using representative driving cycles defined based on vehicle instantaneous speed as sequences of "microtrips", which are intervals between consecutive vehicle stops. Overall significant parameters of the driving cycle, such as average speed, stops per mile, kinetic intensity, and others, are used as independent variables in the modeling process. Performing tests at all the necessary combinations of parameters is expensive and time consuming. In this paper, a methodology is proposed for building driving cycles at prescribed independent variable values using experimental data through the concatenation of "microtrips" isolated from a limited number of standard chassis dynamometer test cycles. The selection of the adequate "microtrips" is achieved through a customized evolutionary algorithm. The genetic representation uses microtrip definitions as genes. Specific mutation, crossover, and karyotype alteration operators have been defined. The Roulette-Wheel selection technique with elitist strategy drives the optimization process, which consists of minimizing the errors to desired overall cycle parameters. This utility is part of the Integrated Bus Information System developed at West Virginia University.

West Virginia University, under contract to the Federal Transit Administration, has developed two tools to evaluate the pollutant emissions, greenhouse gases, and fuel economy of transit buses: a searchable database of transit vehicle emission data and a transit fleet emission inventory model. These tools, complemented by a transit vehicle life-cycle cost model developed by West Virginia University, Battelle, and Transit Resource Center for the Transportation Research Board, provide an interactive, approachable, and reliable method for users, primarily transit agencies, to evaluate overall fleet emissions and fuel consumption for optimization of fleet configuration and operation. These tools will be made available through an online website called the Integrated Bus Information System (IBIS). This paper describes development of the transit fleet inventory model and comparison with the U.S. Environmental Protection Agency (EPA) MOBILE6 and MOVES (motor vehicle emission simulator) emission inventory models. The IBIS model was developed from extensive chassis dynamometer data from several reference vehicles. Polynomial models were built through linear regression. These backbone models characterized the effects of driving activity on vehicle emissions and fuel consumption. Comparison of predicted emissions showed good agreement for hydrocarbon, carbon monoxide, and oxides of nitrogen emissions and acceptable agreement for particulate matter emissions. The EPA MOBILE6 model assumed constant values as a function of duty cycle for carbon dioxide emissions and fuel economy. The West Virginia University IBIS model and EPA MOVES model displayed similar trends for carbon dioxide emissions, but MOVES predicted substantially lower carbon dioxide levels.