View article

The traditional upper-level formulation in the bi-level OD matrix estimation process includes link flows and is mathematically an under-determinate problem. Various methods exist in literature to improve the quality of OD estimates by using additional traffic information including target OD, traffic speeds, travel times, turning proportions, trajectories, and partial OD flows. Irrespective of the type of traffic information, this study along with a few limited studies are of the opinion that structural information of OD needs to be accounted in the upperlevel formulation to improve the quality of final OD estimates. To this end, this study investigates if additional structural knowledge available through partial OD flows (referred as sub-OD flows) can improve the OD estimation quality. This study used two methods to capture the OD structure at two levels and refer to them as macro-OD and micro-OD structural information. Few studies in the past have applied the macro-OD structural information (captured using correlation coefficient) in OD estimation formulation. However, a method to account for micro-OD structure (captured using ratio of OD flows from the same origin and end into destinations with similar trip attractions) and integrate it into the upper-level objective function formulation was never developed and therefore is the major contribution of this study. The proposed methodology was demonstrated on Brisbane city network using synthetic link flows and partial OD flows from Bluetooth observations. The findings from this study revealed that exploiting the partial OD structural information through macro and micro sub-OD flows improved the quality of OD …