Imperial College London
Dr Matthew Eaton

DrMatthewEaton

Faculty of Engineering, Department of Mechanical Engineering

Reader
 
 
 
//

Contact

 

+44 (0)20 7594 7053m.eaton

 
 
//

Location

 

657City and Guilds BuildingSouth Kensington Campus

//

Summary

 

Publications

Citation

BibTex format

@article{O'Malley:2017:10.1016/j.pnucene.2017.03.014,
author = {O'Malley, B and Kophazi, J and Smedley-Stevenson, RP and Eaton, MD},
doi = {10.1016/j.pnucene.2017.03.014},
journal = {Progress in Nuclear Energy},
pages = {177--186},
title = {P-Multigrid expansion of hybrid multilevel solvers for discontinuous Galerkin finite element discrete ordinate (DG-FEM-SN) diffusion synthetic acceleration (DSA) of radiation transport algorithms},
url = {http://dx.doi.org/10.1016/j.pnucene.2017.03.014},
volume = {98},
year = {2017}
}
Download

RIS format (EndNote, RefMan)

TY  - JOUR
AB  - Effective preconditioning of neutron diffusion problems is necessary for the development of efficient DSA schemes for neutron transport problems. This paper uses P-multigrid techniques to expand two preconditioners designed to solve the MIP diffusion neutron diffusion equation with a discontinuous Galerkin (DG-FEM) framework using first-order elements. These preconditioners are based on projecting the first-order DG-FEM formulation to either a linear continuous or a constant discontinuous FEM system. The P-multigrid expansion allows the preconditioners to be applied to problems discretised with second and higher-order elements. The preconditioning algorithms are defined in the form of both a V-cycle and W-cycle and applied to solve challenging neutron diffusion problems. In addition a hybrid preconditioner using P-multigrid and AMG without a constant or continuous coarsening is used. Their performance is measured against a computationally efficient standard algebraic multigrid preconditioner. The results obtained demonstrate that all preconditioners studied in this paper provide good convergence with the continuous method generally being the most computationally efficient. In terms of memory requirements the preconditioners studied significantly outperform the AMG.
AU  - O'Malley,B
AU  - Kophazi,J
AU  - Smedley-Stevenson,RP
AU  - Eaton,MD
DO  - 10.1016/j.pnucene.2017.03.014
EP  - 186
PY  - 2017///
SN  - 0149-1970
SP  - 177
TI  - P-Multigrid expansion of hybrid multilevel solvers for discontinuous Galerkin finite element discrete ordinate (DG-FEM-SN) diffusion synthetic acceleration (DSA) of radiation transport algorithms
T2  - Progress in Nuclear Energy
UR  - http://dx.doi.org/10.1016/j.pnucene.2017.03.014
UR  - http://hdl.handle.net/10044/1/45566
VL  - 98
ER  -
Download