The ERRABL project, funded in the context of the Maths-Vives PEPR program (https://www.maths-vives.fr/) and in collaboration with INRIA Paris and University of Caen, aims to improve the reliability and computational efficiency of hydrodynamic models used to assess tidal-stream energy resource, focusing on the Alderney Race (Raz Blanchard), in Normandy, one of Europe’s most energetic tidal sites. High-fidelity computational models such as Telemac3D (used and developed at University of Caen, https://www.opentelemac.org/index.php/presentation?id=18) provide accurate simulations of tidal flows but remain computationally expensive and uncertain due to complex physics (e.g. turbulence, bottom friction) and uncertain inputs (bathymetry, tidal forcing). The computational cost becomes prohibitive when attempting to model the flow over several tidal cycles within a tidal turbine farm, where the turbines are represented using actuator disk models.
To overcome these limitations, ERRABL develops hybrid model–data approaches and metamodels (surrogate models) that (i) correct physical models using data assimilation and machine learning, (ii) quantify uncertainties on key physical quantities such as the Annual Energy Production (AEP), and (iii) enable optimization of tidal array layout design under uncertainty (i.e., determining the optimal number and placement of turbines within a given area while accounting for hydrodynamic interactions between the turbines).
In this framework, a postdoc hired by INRIA and detached at Institut Jean Le Rond D’Alembert, the Mechanics Institute of Sorbonne University, will be in charge of the development of data-driven improvements to hydrodynamic models, the development of surrogate models for fast wake field modeling, and his use for the optimization of tidal array layout design under highly uncertain conditions.
