You are on page 1of 14

AE-618A FINITE ELEMENT METHODS FOR FLUID FLOWS

Basic Principles of Numerical Methods


for Shock Capturing

Ashish Shukla (12172)


Rishav Choudhary (12575)

Outline
A brief introduction to shock capturing methods, its
necessity, types, and advantages.
A brief introduction to Shock Fitting Methods
Comparison of Shock Capturing & Shock Fitting Methods
A short discussion on SUPG method.

References
1. A.N.Brooks, T.J.Hughes, Streamline upwind/Petrov-Galerkin
formulations for convection dominated flows with particular emphasis
on the incompressible Navier-stokes equations, COMPUTER METHODS
IN APPLIED MECHANICS AND ENGINEERING, 32 (1982) 199-259
2. T.J.Hughes, G.Scovazzi,T.E. Tezduyar, Stabilized methods for
compressible flows, REVIEW PAPER (2008)
3. T.J.Hughes and M.Mallet, A new finite element formulation for
computational Fluid dynamics: IV. A discontinuity-capturing operator
for multidimensional advective-diffusive systems, COMPUTER METHODS
IN APPLIED MECHANICS AND ENGINEERING, 58 (1986) 329-336
4. T.E. Tezduyar, M.Senga, Stabilization and shock-capturing parameters
in SUPG formulation of compressible ows, COMPUTER METHODS IN
APPLIED MECHANICS AND ENGINEERING, 195 (2006) 16211632
5. V.M.K.R.Kotteda, Viscous flow in a mixed-compression intake:
Instabilities, M.TECH THESIS, AEROSPACE ENGINNERING, IIT KANPUR
(2009)
6. The Finite Element Method for Fluid Dynamics, Sixth Edition, O.C.
Zienkiewicz, R.L. Taylor, P. Nithiarasu

Why shock capturing methods?


High speed flows are convection dominant. Compressibility
effects become important. The simple numerical methods like
Finite Element Method using Galerkin approach, and Finite
Difference method fail to give the right solution in this case.
Computation of flow through shock waves is an extremely
difficult task because of discontinuous variation in flow variables,
i.e. pressure, temperature, density, and velocity, across the
shock.
The computational solution for such flow problem without shock
capturing parameters shows regions of undershoots and
overshoots across sharp gradients.
The next slide shows a physical situation which we wish to
simulate numerically. The results obtained with and without
shock capturing methods are compared.

Shock capturing method

With the finite element approximation in which all the variables


are interpolated usingC0 continuity the exact reproduction of
shocks is not possible. In all finite element solutions we therefore
represent the shocks simply as regions of very high gradient.
In order to avoid the spurious oscillations, additional viscosities
are introduced which will help in damping out any oscillations
caused by shocks and, yet, deriving sharp solutions.
Such procedures using artificial viscosities are known as shock
capture methods. The concept of adding additional viscosity or
diffusion to capture shocks was first suggested by von Neumann
and Richtmyer47 as early as 1950.

Types of Shock Capturing methods

Classical Methods
Modern Methods

Central discretization

Upwind based discretization

Do not consider the direction of


wave propagation

Consider the direction of wave propagation

Artificial diffusion terms are usually


linear and same amount is applied
uniformly to all grid points.

Non-linear diffusion term are used with an


automatic feedback mechanism which
adjust the amount of dissipation in any
cell.

Provides accurate results in the case


of smooth and weak shock solutions.

MacCormack method, Lax-Wendro


method, Beam-Warming method etc.

proven to be stable and accurate even for


problems containing strong shock waves.
Total Variation Diminishing (TVD)
schemes, Flux-Corrected Transport
scheme(FCT), Streamline Upwind
Petrov/Galerkin (SUPG) method etc.

Shock Fitting Method


In this method ,shock waves are explicitly introduced in
the solution using appropriate shock relations (Rankine
Hugoniot relations).
The flow is first solved without taking shocks into
consideration. Then, an assumption is made regarding
the position of the shock. The Rankine-Hugoniot relations
give the downstream values of physical variables. The
streamlines are found out using these. The process is
repeated iteratively to get the true location of the shock
which satisfies all the boundary conditions.
Shock fitting consists of locating and tracking the motion
of the discontinuities, which are treated as boundaries
between regions where a smooth solution to the
governing PDEs exists.

Comparison of Shock capturing and shock


fitting method
S.no

Parameters

Shock Capturing

Shock Fitting

Location of
shock

We do not require the location


of the shock .This has obvious
consequences in terms of
coding simplicity.

An assumption has to be made


regarding the initial location of
shock. This followed by an
iterative process is used to find
the correct location. RankineHugoniot relation are used to
approximate the space-time
evolution of the discontinuity.

Mesh
refinement

In order to bring the


numerical thickness of the
captured wave towards its
physical value, mesh
refinement in the shock
normal direction is needed,
which causes an increase in
computational cost

We require more no. of nodal


points where shock is present,
So in this case no grid points
are wasted as in shock
capturing method.

Shock
Strucuture

Shock is numerically smeared


rather than discontinuous

This does not happen in this


case.

SUPG Method
(Streamline Upwind/Petrov-Galerkin)
Upwind Differencing on the convective term can give solutions
that are devoid of unphysical oscillations.
The upwinded convective term can be constructed simply by
adding artificial diffusion to a central difference treatment.
To arrive at this formulation through finite element methods,
there are two paths.
The first method involves addition of artificial viscosity to the
original diffusion term, which is a function of the convective
velocity. The second method involves use of modified basis for
the weight functions (different from the trial functions), which
give more weight to the upstream nodes.
The latter is called the Petrov-Galerkin method.

Problems with upwind schemes: Order of accuracy is lower than central difference scheme.
Over-diffusivity of solutions in presence of source and transient
terms.
Crosswind diffusion problem when the flow is skew to the mesh.
SUPG Approach: The artificial diffusion operator is constructed to act only in the flow
direction, eliminating the possibility of any crosswind diffusion.
Instead of adding the artificial diffusive term only to the diffusive
part of the initial difference scheme, the operator now acts on all
the terms of the original convection-diffusion equation (with source
terms as well as transient terms).
The weight functions across the inter-element boundaries are
discontinuous. They are chosen in the same way as described for
the normal Petrov-Galerkin method.
Addition of a shock-capturing parameter, which is found out using
entropy variables, to the above developed SUPG form enables the
tracking of the discontinuities or shock in the flow.

SPUG formulation of Navier-Stokes


Equation
In terms of conservation variables, the compressible Navier-Stokes
equations can be written in the vector form as
(U/t) + (Fi/xi) (Ei/xi)= 0 on t (0, T)
Here U =

Here Fi & Ei
are Euler & Viscous flux vectors respectively. Using these, the
quasi-linear form of Navier-Stokes equation can be constructed as:-

We define the function spaces as follows:-

The SUPG form can be stated as follows:Find:- Uh Sh : Wh Vh , the following is satisfied:-

Here, the first element level series represents the SUPG stabilization
term with stabilization coefficient , while the second series represents
the shock capturing term with as the shock capturing parameter.

THANK YOU!!!!

You might also like