Cet article a pour but la présentation d'un logiciel d'aide à la construction de modèles d'écoulement à partir de la méthode des distributions de temps de séjour (DTS). La théorie des distributions de temps de séjour ainsi que des méthodes de traçage couramment utilisées sont rappelées. Une description du logiciel est ensuite effectuée. Le modèle est basé sur l'association de modules élémentaires (réacteur parfaitement agité, réacteur piston. . . ), la détermination de la fonction de transfert étant effectuée en résolvant les bilans de matière dans l'espace de Laplace. Le logiciel peut donner la réponse à un signal d'entrée quelconque et plusieurs paramètres du modèle peuvent être optimisés. Différents couples de traceurs et de détecteurs sont donnés et les précautions à prendre lors des traçages sont décrites. Finalement, différentes études effectuées à l'aide du logiciel sont présentées afin de montrer les utilisations possibles. Improving the performance of an existing reactor or studying a new design requires modeling the flow of the different phases. Computational fluid dynamics can give an accurate a priori description of the velocity and concentration fields. However, this approach is too complicated in some cases (complicated geometry, random flow media, etc. ), and this results in an abundance of information. A simpler approach relies on the theory of Residence Time Distribution (RTD), which is a theoretical interpretation framework for tracer experiments. This paper describes a software package that simulates hydrodynamic models derived from RTD experiments. The concept of RTD, introduced by Danckwerts in 1953, is briefly reviewed. Then tracer experiments, commonly used tracers and detectors are dealt with. Finally, the care required in tracer experiments is described in detail. Visual inspection of the tracer response makes it is possible to guess the main characteristics of the flow pattern, such as dead zones, bypasses or recirculations (see Fig. 4), which must be represented by the reactor geometry. Based on these observations and also on the knowledge of the reactor studied, the flow pattern can be modeled with an association of elementary units such as a perfect mixing-cell, plug-flow reactor, perfect mixing-cell in series and perfect mixing-cell in series with exchange to a dead zone. Fig. 2 gives the typical outlet responses of these four elementary units to an ideal inlet pulse of a tracer. Complicated flow patterns can generally be modeled by a network of properly interconnecting elementary units. The software needs to be fed only with the description of this network. Then it derives and solves the corresponding mass-balance equations in the Laplace domain. Finally, it gives the impulse or step response at any node of the network. A convolution with an experimental inlet curve can also be made, and some model parameters can be fitted so as to recover an experimental outlet response at a given node. Various studies carried out using this software package are described in this article in order to point out the possible utilisations :1. An industrial crystalizer2. A new membrane reactor for large-scale mammalian cell cultures3. A waste-water treatment plant primary clarifier4. A ventilation system in environment engineering5. A concrete mixer in civil engineeringIn all these examples, good agreement between experimental and theoretical results was obtained. However, whatever the application, the agreement is meaningful only when :1. The geometry of the flow system is effectively reflected in the structure of the network2. Appropriate tracer experiments are performed in subparts of the flow system when numerous parameters have to be adjusted