Semisequicentennial Transportation Conference Proceedings
May 1996, Iowa State University, Ames, Iowa

Traffic Forecasting as if Intersection Control Matters (And Vice Versa)

Sam Granato

Linn County Regional Planning Commission,
City Hall, 6th Floor,
Cedar Rapids, Iowa 52401.

Many improvements have been made in the past few years in measuring vehicle capacity and delay on arterial streets and intersections. These techniques are well known within both the transportation planning and traffic engineering communities. However, most planners continue to use traffic forecasting techniques concerning capacity and travel time that they know to be inaccurate, due to their being wedded to functionally obsolete software programs. Several new traffic forecasting models have been developed recently that take into account up-to-date research in highway capacity, and the planning staff in the Cedar Rapids metro area has implemented one of these new modeling methods for the latest update to the metropolitan transportation plan. Traffic forecasts are now more realistic given traffic control options and limitations and more useful in developing a financially feasible transportation plan. WARNING: The practicing transportation planner or traffic engineer may find the contents of this paper to be unusual or unexpected. While the author assumes full responsibility for the statements and opinions presented, he assumes no liability for any mental anguish or suffering on the part of the reader.


Traffic engineers know that arterial street capacity and the level of congestion depend on intersection traffic control and management strategies, as well as pavement width (number of through and turning lanes) and access control. However, planners have historically been unable to take these strategies into account in developing traffic forecasts. They do understand that these strategies help determine the travel times on streets and therefore the choice of travel routes used by the motoring public. However, the analytical procedures traditionally used to make traffic forecasts assume that arterial street capacities are fixed values (based on pavement width) regardless of intersection turning movements or the traffic control utilized. Traffic engineers, in turn, have had difficulty in determining how traffic control strategies affect travel route choices and historically have calculated benefits for these strategies as if traffic counts and patterns were not affected by them.

Automated traffic forecasting procedures have been developed recently that partially bridge the gap that has always existed between traditional planning and engineering modeling methodologies. The Cedar Rapids metro area planning staff now utilizes these procedures, in which traffic forecasts are constrained by and adapt to intersection congestion as calculated by Highway Capacity Manual (HCM)-related techniques. Instead of a presumption that arterial streets have fixed capacities, the new model process calculates capacities and travel time (delay) by the HCM techniques, and routes traffic based on the shortest-time paths as calculated. (The process is iterative, as delay and travel time on city streets is due in part to traffic volumes and intersection turn movements, but these volumes and turn movements are based on assessments of the fastest route between all travel origins and destinations including delay at intersections.)

Another important change locally is that the modeled distribution of travel adapts to the (forecasted) level of congestion in order to hold fairly constant (within constraints imposed by land use) the amount of time people spend traveling. While research on travel behavior has shown that this is how the motoring public has been adapting to increasing levels of congestion in many cities (1), the traffic forecasting process in many urban areas does not take this into account. (The motoring public also adapts to reduced congestion by changing where and how far they are willing to travel. Case in point: Access to the city of Seattle, Washington, from suburban communities to the east is via several bridges over Lake Washington and connecting waterways. The widening of the heavily-congested Interstate 90 "Floating Bridge" in 1990 led to a 66 percent increase in traffic volume, from 64,000 to 106,000 vehicles per day, with no reduction in traffic volumes on the alternative bridge routes (2).)

These procedures are a radical departure from the traffic forecasting process used in the 1970s and 1980s and still in use in most other metropolitan areas. Modeling error (the difference between current traffic counts and base-year forecasts) was cut roughly in half from results obtained previously from more traditional planning methods (3), and the error for high-volume streets (AADT > 15,000), on average, is now statistically indistinguishable from sample traffic count error.

This new model process has been utilized to prepare future-year traffic forecasts that are both more credible and more useful in preparing financially feasible transportation plans. However, it is a long-ingrained habit in the transportation profession (both planning and engineering) to treat future traffic forecasts as independent of roadway network development and traffic control—as something to be designed for. Also, a traffic impact study or a signal timing/offset study is typically conducted with the assumption that current intersection turning movements are fixed, and the benefits of signal coordination or impacts of land development along an arterial street are calculated on that basis. While it is recognized within the engineering community that such projects can influence background traffic volumes by changing travel time and delay on the street of interest (4,5), there are few if any analytical methods for calculating such changes outside the traffic forecast model process. Without consideration of this process, a before-and-after field study, if confined to one street corridor, may understate the benefits of a project in terms of congestion or mobility.

The HCM methods, despite challenges raised to them from individual field studies, do a much better job in estimating travel times on arterial streets than traditional traffic forecasting models. Locally, the modeled travel times on eight major arterial corridors were found to be within 10 percent of field-measured travel times, on average. Most planning agencies do not make any determinations regarding accuracy of their modeled travel times, and the only studies located on this subject have found that, at best, the accuracy of the traditional model process is on the order of 15 percent error for freeways and 40–50 percent error for arterial/collector streets (6,7). (It should be remembered that the traditional traffic forecasting model was developed in an era when freeways, not arterial streets, were the primary focus of inquiry.)

It is often argued that planning (traffic forecasting) methods are best suited for planning new roads, not for making traffic engineering decisions. The point being made here is not to use forecasting models exclusively to make engineering decisions, but to assist in determining whether a road-building strategy or a traffic management strategy (or both) is appropriate for specific street corridors or sections of a city.

References

  1. P. Gordon et al. The Commuting Paradox: Evidence from the Top Twenty. Journal of the American Planning Association, Autumn 1991, pp. 416–420.
  2. Biennial Traffic Count Program. Washington State Department of Transportation.
  3. S. Granato. A Case Study in Calibration: Refining Traffic Forecasting Models for Small Urban Areas. PC-TRANS, Spring 1994, pp. 12–15.
  4. W. Recker, et al. Interjurisdictional Coordination of Katella Avenue Traffic Signals. U.S. Department of Transportation Publication No. DOT-T-93-24, August 1992, p. 10.
  5. Institute of Traffic Engineers. Traffic Access and Impact Studies for Site Development: A Recommended Practice. ITE Publication No. RP-020B, 1991, p. 16.
  6. P. Stopher. Predicting TCM Responses with Urban Travel Demand Models. Transportation Planning and Air Quality II, American Society of Civil Engineers, New York, New York, 1993.
  7. R. Dowling and A. Skabardonis. Improving Average Travel Speeds Estimated by Planning Models. In Transportation Research Record No. 1366, TRB, National Research Council, Washington, D.C., 1992, pp. 68–74.

CTRE is an Iowa State University center, administered by the Institute for Transportation.

Address: 2711 S. Loop Drive, Suite 4700, Ames, IA 50010-8664

Phone: 515-294-8103
FAX: 515-294-0467

Website: www.ctre.iastate.edu/

Iowa State University--Becoming the Best