DISCUSSION
What is DG-FEM in general ??
How it is different from the Continuous Galerkin FEM?
Understanding DG-FEM through a scalar conservation
law
Application of DG-FEM to plane stress state/plane strain
state problems
PURELY DISCONTINUOUS
DG-FEM - CLASSIFICATION
DG
Discontinuo
FEM us Galerkin
Method
Interior Penalty
Method (for
elliptical and
parabolic
functions)
Dirchlet
Boundary
conditions are
used
THEIR SIGNIFICANCE
DG-FEM for
hyperbolic
functions
Approximation
solution is
constructed
element wise
NUMERICAL
FLUX is used for
defining at
element
boundaries
The weak form and the strong form of the above conservation law is
described here.
WEAK FORM : The residual/ weak form of the equation is written as shown below :-
Here, one thing to notice is that, RHS is the boundary domain, so at the
boundary, since we have multiple values of approximation solution, we need to
choose which solution yields the results. Hence we introduce the concept of
numerical flux here.
NUMERICAL FLUX : The numerical flux is incorporated in the described weak form as shown
below
Numerical Flux
Integrating the weak form once again, we obtain the strong form which is
represented as shown below
which is also termed as the classical local Discontinuous Galerkin formulation in weak
form
The mass and stiffness matrices are given as
NEW TERMINOLOGY : Numerical Flux functions Central Flux and Upwind flux
At the element boundaries, since there is discontinuity, the
solution at these locations are also not unique. Hence, flux functions
define a value for approximation function at the boundaries.
The choice of numerical flux impacts the stability of the solution.
(There are several ways of defining a numerical flux and it depends
on the type of problem we are dealing with)
Jump Operators
Average Operators
NUMERICAL FLUX IN DETAIL : For most of the problems, the numerical flux is generally chosen as
If
If
, then it corresponds to Central Flux, which is average of the
two solutions at the
interfaces
The role of numerical flux is to ensure stability of the mathematical
formulation.
Jump Operator is defined as follows:(First equation u being scalar, in the second equation u is a vector)
ADVANTAGES OF DG-FEM :Following are the important properties of DG-FEM : Solutions are piecewise smooth, often polynomial, but discontinuous at the
element interfaces
Boundary conditions and interface continuity and enforced only weakly.
All operators (jump, average) are local.
Schemes are well suited to variable order and element sizes, as all
information exchange is
across interfaces