Implementation of unstructured grid GMRES+LU-SGS method on shared-memory, cache-based parallel computers

Sharov, Dmitri; Luo, Hong; Baum, Joseph D.; Loehner, Rainald

doi:10.2514/6.2000-927

Cited by 45 publications

(27 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…During the last one and half decades, the unstructured grid methodology has clearly demonstrated its flexibility in tackling complex geometries, and in flow-based grid adaptations [5][6][7][8][9][10][11][12]. The success demonstrated by the unstructured grid method for steady flow problems has prompted considerable development for unsteady moving boundary problems.…”

Section: Discussionmentioning

confidence: 99%

Simulation of Moving Boundary Flow Using Overset Adaptive Cartesian/Prism Grids and DES

Wang¹

2003

View full text Add to dashboard Cite

The public reporting burden for this collection of information is estimated to average 1 hour per response, includ gathering and maintaining the data needed, and completing and reviewing the collection of information. Send comr of information, including suggestions for reducing the burden, to Department of Defense, Washington Head (0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302. Respondents should be a subject to any penalty for failing to comply with a collection of information if it does not display a currently valid OIV, Development of improved weapon systems requires better understanding of the complex aerodynamics created by moving boundaries. Carriage and release of conventional weapons from aircraft and aerial refueling fall in this category. The flow problems involving moving boundaries are very challenging to compute because the resultant flow field is intrinsically unsteady. Modeling such flow problems is made even more difficult due to the complex geometries and the highly turbulent namre of the unsteady flows. Under an exploratory grant from the AEOSR, a new simulation environment for moving boundary problems is being developed. This environment is designed to advance the state-of-the-art for moving boundary flow problems in the following areas: I. Improve the solution accuracy and efficiency through the use of overset adaptive Cartesian/prism grids. Key flow features will be automatically captured with solution based grid adaptations; 2. Improve the accuracy for highly turbulent unsteady separated flows encountered in moving boundary problems with a hybrid RANS/LES (Reynolds Averaged Navier Stokes/Large Eddy Simulation) approach named Detached Eddy Simulation (DES The report documents the research activities carried out since the grant award, and outlines future work. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS. AFRL-SR-AR-TR-04- SPONSOR/MONITOR'S ACRONYM(S) SPONSOR/MONITOR'S REPORT NUMBER(S) DISTRIBUTION/AVAILABILITY STATEMENT

show abstract

Section: Discussionmentioning

confidence: 99%

Simulation of Moving Boundary Flow Using Overset Adaptive Cartesian/Prism Grids and DES

Wang¹

2003

View full text Add to dashboard Cite

show abstract

“…For unstretched grids, LU-SGS works as in its original algorithm. Other techniques are investigated in this paper to improve the efficiency of LU-SGS by Message Passing Interface (MPI) (Barney 2014) parallelization (Wright et al 1995;Wissink et al 1997;Sharov et al 2000;Luo et al 2003) and node ordering techniques (George and Liu 1981;Borne 2000). First results have been presented (Otero and Eliasson 2013) and are improved in the present paper following a parameter study previously carried out (Otero and Eliasson 2014).…”

Section: Introductionmentioning

confidence: 99%

“…First developed for structured grids by Jameson and Yoon (1987), then successfully extended to hybrid structured-unstructured grids (Soetrisno et al 1996;Sharov and Nakahashi 1998), LU-SGS has been improved through different methods. The definition of a proper implicit operator (Chen and Wang 2000;Kim and Kwon 2005;Dwight 2006) and the combination with the GMRES solver as a preconditioning method (Sharov et al 2000;Nejat and Ollivier-Gooch 2005) are some of them. The method considers a linearization which is inexactly solved by a few steps of a symmetric Gauss-Seidel method through a forward and backward sweep.…”

Section: Introductionmentioning

confidence: 99%

Acceleration on stretched meshes with line-implicit LU-SGS in parallel implementation

Otero

Eliasson

2015

International Journal of Computational Fluid Dynamics

View full text Add to dashboard Cite

The implicit Lower-Upper Symmetric Gauss-Seidel solver is combined with the line-implicit technique to improve convergence on the very anisotropic grids necessary for resolving the boundary layers. The computational fluid dynamics code used is Edge, a Navier-Stokes flow solver for unstructured grids based on a dual grid and edge-based formulation. Multigrid acceleration is applied with the intention to accelerate the convergence to steady state. Lower-Upper Symmetric Gauss-Seidel works in parallel and gives better linear scaling with respect to the number of processors, than the explicit scheme. The ordering techniques investigated have shown that node numbering does influence the convergence and that the orderings from Delaunay and advancing front generation were among the best tested. 2D ReynoldsAveraged Navier-Stokes computations have clearly shown the strong efficiency of our novel approach line-implicit Lower-Upper Symmetric Gauss-Seidel which is four times faster than implicit Lower-Upper Symmetric Gauss-Seidel and line-implicit Runge-Kutta. Implicit Lower-Upper Symmetric Gauss-Seidel for Euler and line-implicit Lower-Upper Symmetric Gauss-Seidel for Reynolds-Averaged Navier-Stokes are at least twice faster than explicit and line-implicit Runge-Kutta, respectively, for 2D and 3D cases. For 3D Reynolds-Averaged Navier-Stokes, multigrid did not accelerate the convergence and therefore may not be needed.

show abstract

“…Over the last three years, FEFLO has been ported to both shared memory [26][27][28] and distributed memory [19, 29, and 30] (16) In order to simplify the algebra involved (and CPU), one may use, without noticeable deterioration of results:…”

Section: Comparison With Experimental Datamentioning

confidence: 99%

Numerical Simulation of H2/Air Detonation Using Detailed Reaction Models

Togashi

Löhner

Tsuboi

2006

44th AIAA Aerospace Sciences Meeting and Exhibit

View full text Add to dashboard Cite

show abstract

Implementation of unstructured grid GMRES+LU-SGS method on shared-memory, cache-based parallel computers

Cited by 45 publications

References 28 publications

Simulation of Moving Boundary Flow Using Overset Adaptive Cartesian/Prism Grids and DES

Simulation of Moving Boundary Flow Using Overset Adaptive Cartesian/Prism Grids and DES

Acceleration on stretched meshes with line-implicit LU-SGS in parallel implementation

Numerical Simulation of H2/Air Detonation Using Detailed Reaction Models

Contact Info

Product

Resources

About