Sensitivity analysis for transit equilibrium assignment and applications to uncertainty analysis

• Develop an equilibrating mechanism for the hyperpath-based transit equilibrium assignment problem (TEAP). • Propose a hyperpath-based gradient projection solution algorithm for solving the TEAP. • Develop the analytical sensitivity analysis approach for the hyperpath-based TEAP. • Develop the sensitivity-based uncertainty analysis approach for estimating the output variations with different sources of uncertainties. • Avoid burdensome computations of simulating the uncertainties from model input/ parameter in uncertainty analysis. Systematic uncertainty analysis can be used to quantitatively evaluate variation in model outputs and identify the critical sources of uncertainty to improve the reliability and stability of a system. To analyze the effects of uncertainties in transit networks that may be caused by probabilistic travel demand, congestion, or vehicle frequencies, this study develops a sensitivity-based uncertainty analysis approach as a post-analysis tool for equilibrium transit systems. The congestion effect is considered in the waiting time and in-vehicle travel time of a passengers’ route-choice model. The hyperpath concept is used to manage passengers’ riding strategies due to the common-line problem at transit stops. A hyperpath-based gradient projection (GP) solution algorithm is developed for the solution of the variational inequality formulation of the transit equilibrium assignment problem (TEAP). A restricted sensitivity analysis approach originally developed for road networks is re-developed for the TEAP in transit networks. An analytical sensitivity-based approach is derived to conduct uncertainty analysis for the TEAP, which enables the simultaneous propagation of uncertainties from different input sources to the model outputs. Numerical examples are provided for the following purposes. (1) To demonstrate three applications of the sensitivity analysis of the TEAP, namely the perturbed solution estimation problem, the critical parameter identification problem, and the paradox analysis problem. (2) To illustrate the use of uncertainty analysis of the TEAP, such as estimating the variance and confidence level of model outputs with respect to various model inputs/parameters as random variables, and ranking the importance of arcs using sensitivity-based uncertainty analysis. (3) To demonstrate the applicability of the proposed approach to real transit networks. The findings demonstrate not only the importance of the analytical sensitivity analysis development for the TEAP, but also for the practical applications of sensitivity and uncertainty analyses.