Loading...

Employment of algebraic multigrid as a preconditioner to solve fully implicit mixed convection equations

Darbandi, M ; Sharif University of Technology | 2006

277 Viewed
  1. Type of Document: Article
  2. Publisher: 2006
  3. Abstract:
  4. The main purpose of the present work is to study the performance of an algebraic multigrid (AMG) algorithm as a preconditioner to the Krylov subspace methods, mainly GMRES methods. The method is used to solve the set of linear algebraic equations resulted from treating simultaneous simulation of fluid dynamics and heat transfer problems. The extended algorithm is fully implicit which results in a huge system of linear algebraic equations. Different parameters affecting the performance of the AMG are taken into account to enhance the performance of our extended algorithm. Because of high level of sparsity of the matrix of coefficients, the current results indicate that the AMG can be very effective if a proper storage method is chosen. The method is highly prone to be chosen as a high performance tool if robust coarsening schemes are selected and suitable coarsening threshold magnitudes are implemented. This latter implementation can remarkably minimize the solution time and the memory requirements. The method shows excellent scalability if the grid size increases. The study also shows that the achieved performance increases as the size of matrix increases
  5. Keywords:
  6. Algebraic multigrid (AMG) ; Krylov subspace methods ; Preconditioner ; Threshold magnitudes ; Algorithms ; Computational fluid dynamics ; Heat transfer ; Matrix algebra ; Problem solving ; Mixed convection
  7. Source: 44th AIAA Aerospace Sciences Meeting 2006, Reno, NV, 9 January 2006 through 12 January 2006 ; Volume 10 , 2006 , Pages 7055-7063 ; 1563478072 (ISBN); 9781563478079 (ISBN)
  8. URL: https://arc.aiaa.org/doi/abs/10.2514/6.2006-588