Network Evaluation with Uncertain and Correlated Long-Term Demand

Predictions of future travel demands are inherently uncertain, implying that future network performance cannot be predicted with certainty. Assuming demand is deterministic, as is typically done in transportation planning, can result in inaccurate measures of network performance that may lead to incorrect policy decisions. This work presents a methodology for treating long-term origin-destination demands as random and possibly correlated in solving for user equilibrium. The effects of various types of correlations and distributions of travel demand on network performance are tested through numerical analysis. The results indicate that neglecting correlations between origin-destination demands can lead to inaccurate measures of network performance. Multiple sampling approaches are tested to determine the most efficient method for achieving accurate estimates of network performance.