Incorporating Feedback in Travel Forecasting

The Clean Air Act Amendments of 1990 (CAAA) introduced new requirements for how transportation modeling for air quality analysis must be performed in nonattainment areas. Because of the degree to which vehicle emission rates (on a grams-per-mile basis) are affected by speed, specific attention has been given to how speeds are estimated and subsequently used in the travel forecasting and emissions estimation process. CAAA and guidelines issued in the years following introduction of the act require that speeds used in the process be realistic in comparison to what might be observed on the road and be reasonably consistent throughout the modeling process. In most traditional modeling processes that model trip generation, trip distribution, mode choice, route assignment, and emissions separately and sequentially, it has not been unusual to find different speeds (and travel times) used in different parts of the process. A description of two different research and development efforts that have produced new methods and guidelines for introducing feedback into the travel and emissions forecasting process to ensure consistent use of speeds is provided. COMSIS Corporation developed for FHWA a method for introducing feedback into the traditional four-step process by using an iterative process through all of the steps until the process converged to a stable set of link speeds. The methodology was used to test the effects of introducing feedback on model results under different levels of network congestion (feedback affects the results only when there is congestion in the network). The project resulted in a report documenting the methods, pitfalls, and common concerns for introducing feedback. A summary of the research conclusions from the project is provided.