Development of a hybrid RANS-LES model for the prediction of unsteady loads at the wall

This work is devoted to turbulence modelling using hybrid RANS-LES approaches. These approaches offer the best potential to reach the compromise between the capability of LES to capture the large-scale structures in regions of interest, and the low-computational cost of RANS calculations in regions where LES resolution is not required or too CPU-demanding, notably in near-wall regions. The industrial objective for EDF is to use these methods to predict unsteady loads on large components (reactor coolant pumps) of pressurized water reactors. In this regard, a new formulation of HTLES (Hybrid Temporal Large-Eddy Simulation) is developed, in order to improve the theoretical foundation of the model, applying the Hybrid- Equivalence criterion. The interest of this model lies in the use of temporal filtering, to control the energy partition between resolved and modeled scales, ensuring a consistent bridging between RANS and LES models. The HTLES approach is applied to several RANS models, and the model is calibrated in decaying isotropic turbulence. In addition, a similar analysis shows for the first time that it is possible, from theoretical considerations, to derive a hybrid RANS-LES model based on scale modifications in a RANS model, which explicitly tends towards a standard LES model in LES regions. The control of the RANS-to-LES transition in near-wall regions is a major concern in this work. In order to improve the behavior of the HTLES model, a two-fold shielding function (depending on physical parameters and mesh criteria) and an internal consistency constraint are introduced, aiming at enforcing the RANS mode when the activation of the LES mode is not suitable. The validation process of the upgraded version of the model is carried out on channel and periodic-hill flows, reaching the expected outcomes: accuracy of the predictions and robustness of HTLES to grid coarsening. Moreover, it is shown that resolved vortices coming from the LES zone penetrates into the near-wall RANS region down to the wall, enabling the hybrid model to fairly reproduce the low frequencies. The capacity of HTLES to provide information on energy and pressure spectra at the wall is assessed, which is decisive to predict unsteady loads. Hence, HTLES offers a cost-saving alternative to a wall-resolved LES. Finally, the work focuses on a zonal control of HTLES (enforcing the RANS mode in large portions of the domain where the LES mode is not required), applying the ALF (Anisotropic Linear Forcing) at the RANS-to-HTLES interfaces to promote turbulent fluctuations. This feasibility study offers promising results. (author)