Several dynamic planning tools use day-to-day adjustment processes that converge towards a "final" regime. The authors introduce and apply rigorous statistical stationarity tests to study the convergence of these day-to-day adjustment processes. These statistical tests are then used to evaluate when a stationary regime occurs after a policy change and what are the properties of the new stationary state. The procedure proposed is applied on a large network (Paris region). The authors conclude that stationarity tests provide a robust statistical component to be used for the evaluation of dynamic traffic simulation tools.
Abstract