Issues about calibration and validation in traffic simulations models are presented in this paper. Among such issues are misconceptions about calibration and validation, a simulation methodology. There are two main phases, conceptual calibration concerns mode nucleus and operation calibration is associated with the context study. The methodology issue with specific context is also addressed. A statistical approach is also proposed to build an experiment ranging from sensitivity analysis to adequate statistical analysis, as the outputs are non-stationary. The work was focused on micro-simulation models and experiment has been carried out using Paramics simulator. The main objective of a more broad work is to start building simulation efforts though more methodological rather than ad-hoc studies and give confidence on the stability of any developed model at the operational level for a specific context study. For the covering abstract see ITRD E114174.
Abstract