In the context of thermalhydraulics, Computational Fluid Dynamics (CFD) codes are widely used for design and safety analysis. CFD codes solve the Navier-Stokes equations in three dimensions. They mostly rely on the Reynolds-averaged formulation of the Navier-Stokes equations. This approach allows for a detailed representation of the flow while requiring a limited numbers of hypotheses (turbulence models, law of the wall). A fine spatial discretisation is needed in order to achieve good prediction capabilities. This implies a large number of control volumes. The computational resources necessary to carry out a calculation at the industrial scale, such as a two-phase flow transient on the entire primary side of a nuclear reactor, are often prohibitive by present-day standards.

In order to cut the computational cost, a coarser spatial discretisation can be retained. Depending on the case of interest, the best practise guidelines of the RANS approach might not all be respected. Further hypotheses need to be added in order to maintain the quality of the model’s predictions. Such models may include pressure drops, heat transfer correlations or mixing terms. This approach is often referred to as a porous media approach.
Regardless of the method, the system of interest is often restricted to an open-loop model, which requires boundary conditions for the equation system to be solved.
Multi-scale coupling methods aim at using each approach where it best suited. The rationale is to reduce the computational burden while capturing the relevant physical phenomena.
Multi-scale coupling can be either one-way or two-way. In a one-way coupling, boundary conditions obtained from a first calculation are used as boundary conditions for another calculation. There is no feedback from the second calculation on the first one. In a two-way coupling, the coupled codes exchange data in the form of boundary conditions, usually at each time step. There is feedback between the two codes. Two-way is the method that is selected in the following.
The boundary conditions used in the standard approach are developed for cases were only macroscopic data are available, flow rate and temperature at the inlet, pressure at the outlet. In the context of a multi-scale coupling, data that are more detailed can be available such as velocity and pressure fields. This thesis work aims at developing boundary conditions, which can take benefit of all the available data in order to make the coupling as seamless as possible.
As an example, in case of two code instances, each one solving a portion of a physical domain relying on the same discretisation and modelling options, the results obtained from these two instances should be identical to that of a single code instance relying on the same discretisation and modelling options solving the entire domain.