Abstract

We couple EMFAC with a dynamic mesoscopic traffic model to create an efficient tool for generating information about traffic dynamics and emissions of various pollutants (CO2, PM10, NOX, and TOG) on large scale networks. Our traffic flow model is the multi-commodity discrete kinematic wave (MCDKW) model, which is rooted in the cell transmission model but allows variable cell sizes for more efficient computations. This approach allows us to estimate traffic emissions and characteristics with a precision similar to microscopic simulation but much faster. To assess the performance of this tool, we analyze traffic and emissions on a large freeway network located between the ports of Los Angeles/Long Beach and downtown Los Angeles. Comparisons of our mesoscopic simulation results with microscopic simulations generated by TransModeler under both congested and free flow conditions show that hourly emission estimates of our mesoscopic model are within 4 to 15 percent of microscopic results with a computation time divided by a factor of 6 or more. Our approach provides policymakers with a tool more efficient than microsimulation for analyzing the effectiveness of regional policies designed to reduce air pollution from motor vehicles.

Abstract

Air quality in Southern California and elsewhere could be substantially improved if some gasoline powered personal vehicles were replaced by vehicles powered by electricity or alternative fuels, such as methanol, ethanol, propane, or compressed natural gas. Quantitative market research information about how consumers are likely to respond to alternative-fuel vehicles is critical to the development of policies aimed at encouraging such technological change. 

In 1991, a three-phase stated preference (SP) survey was implemented in the South Coast Air Basin of California to predict the effect on personal vehicle purchases of attributes that potentially differentiate clean-fuel vehicles from conventional gasoline (or diesel) vehicles. These attributes included: limited availability of refueling stations, limited range between refueling or recharging, vehicle prices, fuel operating costs, emissions levels, multiple-fuel capability, and performance. Respondents were asked to choose one vehicle from each of five sets of hypothetical clean-fuel and conventional gasoline vehicles, each vehicle defined in terms of attributes manipulated according to a specific experimental design. Discrete choice models, such as the multinomial logit model, are then used to estimate how the values of the attribute levels influence purchase decisions. The SP survey choice sets were customized to each respondent’s situation, as determined in the preceding Phase of the survey. The final Phase of the survey involved fuel-choice SP tasks for multi-fuel vehicles that can run on either clean fuels or gasoline. Preliminary results from a pilot sample indicate that the survey responses are plausible and will indeed be useful for forecasting.