A posteriori error analysis of an augmented fully-mixed formulation for the stationary Boussinesq model. (1st February 2019)

• Record Type: Journal Article
• Authors: Colmenares, Eligio; Gatica, Gabriel N.; Oyarzúa, Ricardo
• Is Part Of: Computers & mathematics with applications. Volume 77:issue 3(2019)
• Page Start: 693
• Page End: 714
• Publication Date: 2019-02-01
• Subjects: Boussinesq model -- Finite element method -- Augmented fully-mixed formulation -- A posteriori analysis -- Reliability -- Efficiency; Electronic data processing -- Periodicals; Mathematics -- Data processing -- Periodicals; 510.28541
• DOI: 10.1016/j.camwa.2018.10.009
• Languages: English
• ISSNs: 0898-1221

Abstract

Abstract: In this paper we undertake an a posteriori error analysis along with its adaptive computation of a new augmented fully-mixed finite element method that we have recently proposed to numerically simulate heat driven flows in the Boussinesq approximation setting. Our approach incorporates as additional unknowns a modified pseudostress tensor field and an auxiliary vector field in the fluid and heat equations, respectively, which possibilitates the elimination of the pressure. This unknown, however, can be easily recovered by a postprocessing formula. In turn, redundant Galerkin terms are included into the weak formulation to ensure well-posedness. In this way, the resulting variational formulation is a four-field augmented scheme, whose Galerkin discretization allows a Raviart–Thomas approximation for the auxiliary unknowns and a Lagrange approximation for the velocity and the temperature. In the present work, we propose a reliable and efficient, fully-local and computable, residual-based a posteriori error estimator in two and three dimensions for the aforementioned method. Standard arguments based on duality techniques, stable Helmholtz decompositions, and well-known results from previous works, are the main underlying tools used in our methodology. Several numerical experiments illustrate the properties of the estimator and further validate the expected behavior of the associated adaptive algorithm.