SIGNALIZED INTERSECTION DELAY MODELS--A PRIMER FOR THE UNINITIATED

Delay is being used increasingly as the primary indicator of level of service at signalized intersections, but for many traffic engineers delay estimation is a new task. Some of the currently available estimation techniques are introduced, and the assumptions on which they are based are examined. In general, these assumptions are unrealistic; so accurate delay estimates are not really possible. The difficulties are particularly acute when the arrival flows approach capacity. Some of the procedures avoid the worst of the problems at high flows by methods that, though they involve considerable mathematics, are based on modeling that is essentially qualitative rather than quantitative. Such methods abandon the quest for accuracy in favor of reasonableness: rather than attempting to provide right answers, they try to avoid answers that are terribly wrong. These models would seem to be useful as long as users do not expect too much from them, and the models can probably be somewhat improved. If accuracy is desired, however, a new generation of models that take more account of variations in travel demand over time is needed. The use of such models would require more information about traffic patterns than users are accustomed to providing.