RECONCILING USER AND SYSTEM OBJECTIVES IN TRAFFIC CORRIDOR MANAGEMENT

This dissertation addresses the problem of improving the efficiency of a traffic corridor through the development of a computerized procedure for determining means of reconciling the individual corridor user's objectives with that of the system's. An LP based model specifically structured for traffic corridor assignment and incorporating dual objective functions designed to obtain either a User or a System Based assignment (accurately as well as efficiently) is formulated. The model also contains the capability of simulating the effect of ramp metering (which is singled out for analysis as the most feasible form of corridor control) through an adjustment to the right hand sides of a subset of the LP constraints. To facilitate the determination of optimal ramp metering rates (i.e., those which force the User Based assignment to maximum conformance with the System Based assignment), an algorithm is developed and presented. This algorithm makes use of the dual simplex method to systematically cut the region of feasible assignments so as to veer the User Based assignment toward that of the System Based optimum solution.; The above reconciliation procedure is applied to a case study consisting of a 15-mile stretch of the Long Island Expressway and the parallel Northern State Parkway. The application of the procedure reveals that the imposition of ramp metering at the direct access points only (i.e, key ramp control) results in a reduction of 80% in the Excess Travel Time attributable to the User Based assignment. Extension of control to all transfer facilities results in an additional 11% convergence. From another perspective, the results indicate that partial 'key ramps only' control will produce a reduction in the total corridor travel time of 2.9%, while full control will bring about a 3.3% reduction in corridor travel time.; The sensitivity of the procedure to inevitable errors in the O-D data was investigated. It was determined that ramp metering rates derived through the use of O-D tables containing normal errors with a coefficient of variation as high as 0.20, still produced a significant improvement in the User Based assignment for the case study at hand.; Other uses for the LP model were explored. The relative improvement in the performance of the corridor to be gained through various alternatives such as (1) the reduction in the overall vehicle miles of travel (VMT) or (2) the addition of a lane to a critical stretch of roadway, were evaluated with the use of the model.