Adjoint-based adaptive finite element method for the compressible Euler equations using finite calculus
2015
Kouhi Esfahani, Mohammad | Oñate Ibáñez de Navarra, Eugenio | Mavriplis, Dimitri | Universitat Politècnica de Catalunya. Departament d'Enginyeria Civil i Ambiental | Universitat Politècnica de Catalunya. GMNE - Grup de Mètodes Numèrics en Enginyeria
In this paper, an adjoint-based error estimation and mesh adaptation framework is developed for the compressible inviscid flows. The algorithm employs the Finite Calculus (FIC) scheme for the numerical solution of the flow and discrete adjoint equations in the context of the Galerkin finite element method (FEM) on triangular grids. The FIC scheme treats the instabilities normally generated in the numerical solution of the fluid equations through adding two stabilization terms, called streamline term and transverse term, to the original central-based discretized formulation. The non-linear system of equations resulting from the flow problem is solved implicitly using a damped Newton's method accompanied with the exact Jacobian matrix. A defect corrected scheme is implemented to iteratively solve the linear system of equations related to the adjoint problem benefiting from the transpose of the Jacobian matrix. At each iteration, the linear systems of equations resulting from the fluid and adjoint problems are solved using a preconditioned GMRES method. Having calculated the error of a specified output functional locally, an h-refinement methodology based on the element subdivision is performed to refine the candidate elements. The quality of the numerical results proves the capability of the presented approach for the adjoint-based error estimation and mesh adaptation problems in different flow regimes.
Показать больше [+] Меньше [-]Peer Reviewed
Показать больше [+] Меньше [-]Postprint (author's final draft)
Показать больше [+] Меньше [-]Библиографическая информация
Эту запись предоставил Universitat Politècnica de Catalunya
