Virtual prototyping pneumatic circuits for power transmission, for example braking circuits of trucks, is still a difficulty because of the complexity of the flow behavior in transonic conditions and of the coupling between local and macroscopic scales. These problems are met during system design, control synthesis and for static and dynamic performance analysis. Tuning accurate numerical models requires important costs and time when compared to other systems. The methodology proposed in this PhD thesis relies on numerically determining a data base that characterizes the local and macroscopic behavior of a circuit component according the variation from a reference point of some physical or geometrical parameters. The data bases are obtained from the extrapolation of the Mean Navier Stokes solution (RANS) for a given reference point with the help of a parametrization software dedicated to fluid mechanics (Turb'Flow). The main contribution of this thesis relies io the analysis of the solution obtained from the parametrization in two different cases: the De Laval nozzle and un "elbow" connecting element, which are elementary component in a circuit. We have shown that these two "simple" cases lead already to important difficulties in term of problem parametrization and calculation of the derivatives of the aerodynamic fields because of the problem dimension. In order to tackle this, we proposed to reduce the spatial discretization (mesh derefining) and we showed that this approach could sometimes lead to damp or move some phenomena (shocks). The second contribution of this work relies on evaluating the quality of the extrapolated solution and their validity domain, and on building links between local and macroscopic behavior. Finally, we proposed a method that allows the mass flow rate characteristic of a component to be determined from the calculation of the extrapolated solution issued from a limited number of reference points.
