Robust transportation network analysis with uncertain and correlated long-term origin-destination demands

A methodology for treating long-term origin-destination demand as random and possibly correlated in solving for deterministic user equilibrium is presented in this thesis. The effects of various types of correlations, demand variances, and congestion levels are tested through numerical analysis. Network performance is measured using the expected value and variance of total system travel time. This thesis shows that, if correlations do exist between origin-destination demands, assuming that the demands are deterministic or stochastic and independent may lead to inaccurate measures of network performance.