An important application of advanced technology to transportation engineering involves the development of advanced traffic management systems (ATMS). Any proposed traffic-management action is essentially a forecast that the action will cause certain conditions, and uncertainty concerning traffic demand and/or the values of model parameters will introduce random error between what is expected and what actually occurs. This paper considers in some detail the problem of forecasting whether or not a given set of freeway on-ramp volumes are likely to cause overcapacity mainline demand. A simple model for the distribution of freeway traffic is considered, which leads to a straightforward interpretation of the impact of forecast uncertainty as a reduction in practial mainline capacity. The problem of efficient estimation-of-demand model parameters is then taken up; by embedding the demand model in a Markov-process model of freeway traffic flow, it becomes possible to compare three estimation approaches via Monte Carlo methods. These estimators show noticeable differences in statistical efficiency, which translate into practically significant differences in freeway mainline capacity. This suggests that improvements in traffic management may be possible through substitution of efficient for less-efficient statistical procedures.
|Original language||English (US)|
|Number of pages||15|
|Journal||Journal of Transportation Engineering|
|State||Published - Jul 1 1993|