Optimisation Topologique et Géométrique appliquées à la CFD en conduits (thèse 2016 - 2019)
Publications scientifiques au M2P2
2021
Isabelle Cheylan, Song Zhao, Pierre Boivin, Pierre Sagaut. Compressible pressure-based Lattice-Boltzmann applied to humid air with phase change. Applied Thermal Engineering, 2021, pp.116868. ⟨10.1016/j.applthermaleng.2021.116868⟩. ⟨hal-03180596⟩ Plus de détails...
A new compressible pressure-based Lattice Boltzmann Method is proposed to simulate humid air flows with phase change. The variable density and compressible effects are fully resolved, effectively lifting the Boussinesq approximation commonly used, e.g. for meteorological flows. Previous studies indicate that the Boussinesq assumption can lead to errors up to 25%, but the model remains common, for compressible models often suffer from a lack of stability. In order to overcome this issue, a new pressure-based solver is proposed, exhibiting excellent stability properties. Mass and momentum conservation equations are solved by a hybrid recursive regularized Lattice-Boltzmann approach, whereas the enthalpy and species conservation equations are solved using a finite volume method. The solver is based on a pressure-based method coupled with a predictor-corrector algorithm, and incorporates a humid equation of state, as well as a specific boundary condition treatment for phase change. In particular, boundary conditions that handle mass leakage are also proposed and validated. Three test cases are investigated in order to validate this new approach: the Rayleigh-Bénard instability applied to humid air, the atmospheric rising of a condensing moist bubble, and finally the evaporation of a thin liquid film in a vertical channel. Results indicate that the proposed pressure-based Lattice-Boltzmann model is stable and accurate on all cases.
Isabelle Cheylan, Song Zhao, Pierre Boivin, Pierre Sagaut. Compressible pressure-based Lattice-Boltzmann applied to humid air with phase change. Applied Thermal Engineering, 2021, pp.116868. ⟨10.1016/j.applthermaleng.2021.116868⟩. ⟨hal-03180596⟩
Isabelle Cheylan, Guillaume Fritz, Denis Ricot, Pierre Sagaut. Shape Optimization Using the Adjoint Lattice Boltzmann Method for Aerodynamic Applications. AIAA Journal, 2019, 57 (7), pp.2758-2773. ⟨10.2514/1.J057955⟩. ⟨hal-02468051⟩ Plus de détails...
The present work focuses on shape optimization using the lattice Boltzmann method applied to aerodynamic cases. The adjoint method is used to calculate the sensitivities of the drag force with respect to the shape of an object. The main advantage of the adjoint method is its cost, because it is independent from the number of optimization parameters. The approach used consists in developing a continuous adjoint of the primal problem discretized in space, time, and velocities. An adjoint lattice Boltzmann equation is thus found, which is solved using the same algorithms as in the primal problem. The test cases investigate new features compared to what exists in the literature, such as the derivation of the grid refinement models in the primal problem to obtain their adjoint counterparts, but also the derivation of a double-relaxation-time algorithm and the Ginzburg et al. interpolation at the wall ("Two-Relaxation-Time Lattice Boltzmann Scheme: About Parametrization, Velocity, Pressure and Mixed Boundary Conditions," Communications in Computational Physics, Vol. 3, No. 2, 2008, pp. 427-478). Regarding the unsteadiness of the primal problem, two methods differing in accuracy and computational effort are compared using a two-dimensional unsteady case. Finally, this first-of-a-kind adjoint solver is applied to a large-scale threedimensional turbulent case (the flow of air around a car at a speed of 130 km/h), which shows its usefulness in the industry.
Isabelle Cheylan, Guillaume Fritz, Denis Ricot, Pierre Sagaut. Shape Optimization Using the Adjoint Lattice Boltzmann Method for Aerodynamic Applications. AIAA Journal, 2019, 57 (7), pp.2758-2773. ⟨10.2514/1.J057955⟩. ⟨hal-02468051⟩
