Institutional Repository
Technical University of Crete
EN  |  EL

Search

Browse

My Space

Multigrid cell-centered techniques for high-order incompressible flow numerical solutions

Mathioudakis Emmanouil, Mandikas Vasileios, Kozyrakis Georgios, Kampanis, Nikolaos A, Ekaterinaris, John A

Full record


URI: http://purl.tuc.gr/dl/dias/75F0BB2A-59FB-4165-A817-4E3294EC6EA5
Year 2017
Type of Item Peer-Reviewed Journal Publication
License
Details
Bibliographic Citation E. N. Mathioudakis, V. G. Mandikas, G. V. Kozyrakis, N. A. Kampanis and J. A. Ekaterinaris, "Multigrid cell-centered techniques for high-order incompressible flow numerical solutions," Aerosp. Sci. Technol., vol. 64, pp. 85-101, May 2017. doi: 10.1016/j.ast.2017.01.015 https://doi.org/10.1016/j.ast.2017.01.015
Appears in Collections

Summary

A multigrid pressure correction scheme suitable for high order discretizations of the incompressible Navier–Stokes equations is developed and demonstrated. The pressure correction equation is discretized with fourth-order compact finite-difference approximations. Iterative methods based on multigrid techniques accelerate the most demanding part of the overall solution algorithm, which is the numerical solution of the arised large and sparse linear system. Geometrical multigrid methods, using partial semicoarsenig strategy and zebra line Gauss–Seidel relaxation, are employed to efficiently approximate the solution of the resulting algebraic linear system. Effects of various multigrid components on the pressure correction procedure are evaluated and new high-order transfer operators are developed for the case of cell-centered grids. Their convergence rates are also compared with commonly used intergrid transfer operators. Furthermore, numerically comparisons between different multigrid cycle approaches, such as V-, W- and F-cycle, are presented. The performance tests demonstrate that the new pressure correction approach significantly reduces the computational effort compared to single-grid algorithms. Furthermore, it is shown that the overall high order accuracy of the numerical method is retained in space and time with increasing Reynolds number.

Services

Statistics