An Adaptive Discontinuous Galerkin Technique with an Orthogonal Basis Applied to Compressible Flow Problems
Summary (1 min read)
2.1. Spatial discretization.
- Curved elements, which are essential for higher-order analysis on curved domains [1] , will require some modifications, e.g., the authors can use Gram-Schmidt orthogonalization with a different scalar product and induced norm.
- Shape functions would become element dependent and the matrix Ú Û BÜ qÝ would have to be computed and stored for every curved element of the mesh.
- This is not excessive because the total memory never exceeds that required for a global mass matrix and the number of curved elements is typically î ï.
3. Adaptive h-and p-Refinement.
- Adaptive analysis techniques have been shown to be highly effective for use in fluid mechanics problems (e.g., [14, 15] ).
- H-refinement consists of modifying element sizes while p-refinement consists of modifying polynomial orders.
- The authors describe a procedure to alter element sizes or polynomial orders using only local operations to change the approximation space.
3.2. H-Refinement.
- The DGM does not impose inter-element continuity of the approximated fields.
- For both refinement and coarsening operations, a 0 1 projection is performed to define the new solutions.
- With refinement, the identity projection is used.
- A loss of precision is associated with coarsening, as well as with the order reduction for p-refinement case.
Q dc A fQ (iv) with
- The increase is linear, which is optimal since the solution of this problem is self similar with the total length of the discontinuities (shocks and contact surface) growing linearly in time.
- The contact discontinuity, which turns to form the jet, is widened.
- After each time step (or sub-time step), the authors evaluate the solution at each integration point on element edges and faces.
- Rayleigh-Taylor problems are unstable; this asymmetry is, thus, not surprising.
- The mesh asymmetry induces a small perturbation that leads to significant modification to the flow.
4.2. Error Indicator.
- Parallel computation with dynamic load balancing will be essential for three-dimensional computations.
- The DGM software is being combined with a parallel data management system [8, 17] for this purpose.
Did you find this useful? Give us your feedback
Citations
404 citations
Cites background from "An Adaptive Discontinuous Galerkin ..."
...A basis for Pp(Ωj) is chosen to be orthogonal in L(2) [30, 14, 26] on Ωj and this leads to the Dubiner basis commonly used with spectral methods [24]....
[...]
262 citations
195 citations
182 citations
Cites methods from "An Adaptive Discontinuous Galerkin ..."
...Discontinuous Galerkin methods became popular following the pioneering work of Cockburn and Shu (1989, 1998) and have been widely adopted in computational fluid dynamics and other engineering applications (e.g., Bassi and Rebay 1997; Remacle et al. 2003)....
[...]
160 citations
References
8,174 citations
2,671 citations
2,319 citations
"An Adaptive Discontinuous Galerkin ..." refers methods in this paper
...Fryxell [9] used a piecewise parabolic method [21] with adaptive h-refinement to solve compressible RTI problems in two and three dimensions....
[...]
2,143 citations
"An Adaptive Discontinuous Galerkin ..." refers methods in this paper
...The Discontinuous Galerkin Method (DGM) was initially introduced by Reed and Hill in 1973 [16] as a technique to solve neutron transport problems....
[...]
2,119 citations
"An Adaptive Discontinuous Galerkin ..." refers background in this paper
...Cockburn and Shu [3, 5] describe a limiting procedure that prevents the approximate solution on an element from taking values outside of the range spanned by the neighboring solution averages....
[...]