Gabriel N. Gatica, Cristian Inzunza, Ricardo Ruiz-Baier, Felipe Sandoval:
A posteriori error analysis of Banach spaces-based fully-mixed finite element methods for Boussinesq-type models
In this paper we consider Banach spaces-based fully-mixed variational formulations that has been recently proposed for the Boussinesq and the Oberbeck-Boussinesq models, and develop reliable and efficient residual-based a posteriori error estimators for the 2D and 3D versions of the associated mixed finite element schemes. For the reliability analysis, we employ the global inf-sup condition for each equation defining the model, namely Navier-Stokes and heat equations in the case of Boussinesq, along with suitable Helmholtz decomposition in nonstandard Banach spaces, the approximation properties of the Raviart-Thomas and Clement interpolants, further regularity on the continuous solutions, and smallness assumptions on the data. In turn, the efficiency estimates follow from inverse inequalities and the localization technique through bubble functions in adequately defined local L^p spaces. Finally, several numerical results including natural convection in 3D differentially heated enclosures, are reported with the aim of confirming the theoretical properties of the estimators and illustrating the performance of the associated adaptive algorithm.
This preprint gave rise to the following definitive publication(s):
Gabriel N. GATICA, Cristian INZUNZA, Ricardo RUIZ-BAIER, Felipe SANDOVAL: A posteriori error analysis of Banach spaces-based fully-mixed finite element methods for Boussinesq-type models. Journal of Numerical Mathematics, vol. 30, 4, pp. 325-356, (2022).