An immersed boundary method in OpenFOAM : Verification and validation. (3rd November 2017)

• Record Type: Journal Article
• Authors: Constant, E.; Favier, J.; Meldi, M.; Meliga, P.; Serre, E.
• Is Part Of: Computers & fluids. Volume 157(2017)
• Page Start: 55
• Page End: 72
• Publication Date: 2017-11-03
• Subjects: Immersed Boundary Method (IBM) -- OpenFOAM -- Bluff body -- Incompressible flows; Fluid dynamics -- Data processing -- Periodicals; 532.050285
• DOI: 10.1016/j.compfluid.2017.08.001
• Languages: English
• ISSNs: 0045-7930

Abstract

Highlights: A modified PISO algorithm integrating an efficient Immersed Boundary method is proposed, using an improved direct forcing approach. A rigorous characterization of the Immersed Boundary method is performed using an original verification technique. The numerical errors at various stages of the algorithm are precisely estimated. A thorough validation of the solver on relevant literature test-cases is provided. Abstract: The present work proposes a modified Pressure-Implicit Split-Operator (PISO) solver integrating the recent Immersed Boundary Method (IBM) proposed by [1] in order to perform reliable simulations of incompressible flows around bluff bodies using the open source toolbox OpenFOAM version 2.2 ([2] ). The (IBM) allows for a precise representation of fixed and moving solid obstacles embedded in the physical domain, using uniform or stretched Cartesian meshes. Owing to this feature, the maximum level of accuracy and scalability of the numerical solvers can be systematically achieved. An iterative scheme based on sub-iterations between (IBM) and pressure correction has been implemented in the native (PISO) solver of OpenFOAM. This allows one to use fast optimized Poisson solvers while satisfying simultaneously the divergence-free flow state and the no-slip condition at the body surface. To compute the divergence of the momentum equation (in the PISO loop) and the interpolation of the fluxes, we propose an hybrid calculation with an analytical resolution (using the kernel function equation) of the quantities involving the force term (singular quantities). A careful and original verification study has been carried out which allows to estimate three different errors related to the discretization and to the (IBM). Various 2D and 3D well-documented test cases of academic flows around fixed or moving solid bodies (cylinder and sphere) have been simulated and carefully validated against existing data from the literature in a large range of Reynolds numbers, R e = 30 − 3900 and in the frame of DNS and DDES OpenFOAM native models.