[en] In this report we deal with the equation of the mean temperature field in rod bundles. Using homogenization techniques an equation is given (the simplest) in which an equivalent conductivity tensor appears. This conductivity has not the standard properties: the non symmetric part seems to be important

[en] This paper shows a comparison between 3D computation and some results of scale models experiments in the near field of a cooling tower. We compare the velocity and temperature fields and we found a rather good agreement with the measurements. The rough description of the shell in the computation gives rise to a pressure field which has not the same intensity as the measured field

[en] The discretization of advective terms in the numerical models is one of the main difficulty that we have to deal with in Fluid Mechanics. Different methods are proposed here using a characteristic approach which can be used in finite element discretization. Different examples are given which show the possibilities of the method

[en] This report deals with dispersion in a mean two dimensional flow such as the sea. A theory is presented leading to the definition of dispersion by a 2 x 2 tensor, the terms of which are expressed as a function of the velocity distribution of the flow in the vertical direction and of the turbulent viscosity coefficient. A pessimistic estimation (too small) of the coefficients of this tensor is obtained by calculating the velocity distribution in the vertical direction for the case of a flat infinite bed and no wind. The vertical turbulent velocity distribution used for this estimation was extracted from the literature. Both the permanent and tidal regime results are presented in the form of a nomograph. The dispersion tensor coefficients were also calculated from field measurements giving the velocity distribution in the vertical direction. These measurements were performed out at sea in the region of the Flammanville plant by the laboratory

[en] The principal difficulties of numerical order which appear when one hopes to realize a model in industrial hydraulics are the following: (i) Discrete representation of convection terms. (ii) Represention of the domain boundaries. In this report is described a mean to approach the problem, and examples are then given which allow to render an account of the possibilities offered by the numerical models

[en] An algorithm using the velocity components and stream function has been developped for the computation of non isothermal incompressible flows. It is applied with finite difference numerical methods, using rectilinear or arbitrary non orthogonal curvilinear grids. This programm was developped in order to allow the analysis of thermal-fluid problems which occur in the study of reactor components (especially in LMRBR), as many of the encountered geometries are simple enough for allowing a plane description of the flow. The comparison with experiments appeared to be fairly good, and the resulting model is found to be a very useful tool for the understanding and prediction of the flow and temperature interaction. (orig.)

[en] Finite difference numerical methods are available for the computation of unsteady nonisothermal flows with possibly strong buoyancy effects or head loss terms. The algorithm uses either velocity-pressure or velocity-stream-function formulations. The treatment of advective terms involves the method of characteristics. Arbitrary nonorthogonal curvilinear grids may be used, and turbulence is modeled by means of a k-epsilon eddy viscosity model. Two examples of application to liquid-metal fast breeder reactor thermal analysis are: hot plenum flow in a pool-type vessel during flow and thermal transients and unsteady flow in a pipe resulting from an inlet temperature change with a very low flow rate. For both cases, comparisons with experimental studies and applications to real reactors are shown

[en] An algorithm using the velocity components and stream function has been developped for the computation of non isothermal incompressible flows. It is applied with finite difference numerical methods, using rectilinear or arbitrary non orthogonal curvilinear grids. This program was developped in order to allow the analysis of thermal-fluid problems which occur in the study of reactor components (especially in LMFBR), as many of the encountered geometries are simple enough for allowing a plane description of the flow. The comparison with experiments appeared to be fairly good, and the resulting model is found to be a very useful tool for the understanding and prediction of the flow and temperature interaction. (orig.)

[en] Finite difference numerical methods are available for the computation of unsteady nonisothermal flows with possibly strong buoyancy effects or head loss terms. The algorithm uses either velocity-pressure or velocity-stream-function formulations. The treatment of advective terms involves the method of characteristics. Arbitrary nonorthogonal curvilinear grids may be used, and turbulence is modeled by means of a k-epsilon eddy viscosity model. Two examples of application to liquid-metal fast breeder reactor thermal analysis are: hot plenum flow in a pool-type vessel during flow and thermal transients, and unsteady flow in a pipe resulting from an inlet temperature change with a very low flow rate. For both cases, comparisons with experimental studies and applications to real reactors are shown

[en] The results of this document have to be distinguished in mathematical models applicable to small-area problems (horizontal scale comparable to depth) and models applicable to large-area phenomena (horizontal scales much greater than depth, quasi-hydrostatic approximation). In the case of the former, progress remains to be made in the simulation of turbulence and in the development of algorithms applicable under often very complex geometrical conditions. Excellent results are obtained by combining mathematical models with reduced-scale models, the former (on larger scales) providing the boundary conditions for the tank of the physical models. Large-area problems can be tackled only by means of mathematical models. These models are extremely efficient for the calculation of mesoscale circulation and transport of pollutants, but they all run into the same difficulty of simulating long-term problems and of determining drift currents. The principal difficulty faced by mesoscale or macroscale models is the determination of atmospheric conditions and of boundary conditions in the open sea. Mathematical models make it possible to determine the situation at every point of a given coastal zone and require only the conditions at the boundaries of the zone for this purpose. However, although these conditions at the boundary correspond to an experimental effort small in relation to total surveillance of the zone, they are essential to the predictions of the mathematical model, and efforts must be made to obtain the best possible boundary conditions. In addition to these experimental surveys at the boundaries, a certain number of observations within the zone are needed for the calibration of the model, i.e. for the determination of certain numerical coefficients appearing in the parametrization