[en] The R and D thermal hydraulic codes, notably the finite difference codes Melodie (2D) and ESTET (3D) or the 2D and 3D versions of the finite element code N3S were initially developed for incompressible, possibly dilatable, turbulent flows, i.e. those where density is not pressure-dependent. Subsequent minor modifications to these finite difference code algorithms enabled extension of their scope to subsonic compressible flows. The first applications in both single-phase and two flow contexts have now been completed. This paper presents the techniques used to adapt these algorithms for the processing of compressible flows in an N3S type finite element code, whereby complex geometries normally difficult to model in finite difference meshes could be successfully dealt with. The development of version 3.0 of he N3S code led to dilatable flow calculations at lower cost. On this basis, a 2-D prototype version of N3S was programmed, tested and validated, drawing maximum benefit from Cray vectorization possibilities and from physical, numerical or data processing experience with other fluid dynamics codes, such as Melodie, ESTET or TELEMAC. The algorithms are the same as those used in finite difference codes, but their formulation is variational. The first part of the paper deals with the fundamental equations involved, expressed in basic form, together with the associated digital method. The modifications to the k-epsilon turbulence model extended to compressible flows are also described. THe second part presents the algorithm used, indicating the additional terms required by the extension. The third part presents the equations in integral form and the associated matrix systems. The solutions adopted for calculation of the compressibility related terms are indicated. Finally, a few representative applications and test cases are discussed. These include subsonic, but also transsonic and supersonic cases, showing the shock responses of the digital method. The application of compressible flow pressure boundary conditions is presented in an appendix. (authors). 16 refs., 24 figs., 2 appends

[en] The capacity of turbogenerators in PWR is regulated with governing valves located at the admission of the high-pressure turbine. In this paper we present a comparison between measurements and a numerical simulation of the flow in a 2D mock up of this governing valve. To predict and simulate transonic flow at low Mach numbers, we present a new extension of two codes initially devoted to incompressible and unsteady flows (pressure based method). The codes use either FInite Difference Method or, for complex geometry, Finite Element Method. Predicting those kinds of flows is difficult due to strong coupling between physical phenomena like turbulence on one hand, and the complexity of industrial geometry on the other hand. The comparison of numerical results with pressure measurements and also with Schlieren photographs confirms the validation of this approach. The results show clearly how the method correctly captures the structure of the jet. (authors). 10 figs., 11 refs