Accuracy of Finite Volume/Staggered Grid Distributed Lagrange Multiplier/Fictitious Domain simulations of particulate flows. (22nd July 2015)

• Record Type: Journal Article
• Authors: Wachs, Anthony; Hammouti, Abdelkader; Vinay, Guillaume; Rahmani, Mona
• Is Part Of: Computers & fluids. Volume 115(2015)
• Page Start: 154
• Page End: 172
• Publication Date: 2015-07-22
• Subjects: Particulate flow -- Distributed Lagrange Multiplier/Fictitious Domain method -- Finite volume -- Solution accuracy -- Parallel computing; Fluid dynamics -- Data processing -- Periodicals; 532.050285
• DOI: 10.1016/j.compfluid.2015.04.006
• Languages: English
• ISSNs: 0045-7930

Abstract

Highlights: 3D particle-resolved simulations of suspension flow. Finite Volume/Staggered Grid-DLM/FD method. New multidimensional quadratic velocity reconstruction at particle boundary. Extended space accuracy study in challenging flow configurations. Error function of Reynolds number, solid volume fraction and particle angularity. Abstract: Our Distributed Lagrange Multiplier/Fictitious Domain method[1–3] has been recently implemented with a Finite Volume/Staggered Grid discretization scheme[4] and successfully applied to the settling of a single particle in an unbounded domain. Here we scrutinize the space and time accuracy of the computed solution in a broad range of flow configurations of growing complexity. We suggest a 2nd order interpolation operator to impose the rigid body motion constraint at the particle boundary based on Finite Element cubic quadratic basis functions. This enables us to benefit from the best of both discretization schemes: a fast Finite Volume/Staggered Grid Navier & Stokes solver and a high order Finite Element reconstruction at the particle boundary. The resulting space accuracy of the computed solution is improved without the need to employ any hydrodynamic radius calibration procedure for spherical particles. As a result, the method can also accurately be extended to non-spherical and angular particles. The order of space convergence is between linear and quadratic, depending on the flow configuration, and the magnitude of the error has a tendency to increase with the solid volume fraction, the Reynolds number and the angularity of the particles.