The general strategy adopted in the development of a computational tool performing the identification of parametric models based on the residence time distribution (RTD) theory is exposed Two main aspects of the modelling procedure are presented : the structural discrimination of the various solution schemes, and the parameter estimation step. The master problem is solved by a stochastic procedure based on a Simulated Annealing algorithm while the slave problem is solved by a nonlinear deterministic procedure. An industrial application is proposed for the description and the quantification of the air now patterns and the associated pollutant transfers observed in ventilated indoor spaces, and particularly in nuclear enclosures.