Numerical Methods for Hyperbolic

Conservation Laws

Introduction

Dr. Aamer Haque

http://math.iit.edu/~ahaque6
ahaque7@iit.edu

Illinois Institute of Technology

June 2, 2009
 Outline

Introduction
Mathematical/Numerical Modeling
Derivation of Conservation Laws
Examples of Conservation Laws
Hyperbolic Conservation Laws
 Modeling of Conservation Laws
Physical Model
Problem Conservation Laws
NO

YES
Mathematical Model Matches
PDEs Reality? Done!

YES Validation

Numerical Model Solves

Finite Volumes PDEs?
NO
Verification

Computational Model
Software/Hardware Solution
 V&V - Verification & Validation

Verification The process of determining that a model

implementation accurately represents the developer's
conceptual description of the model and the solution to
the model. (i.e. Solving the equations correctly)

Validation The process of determining the degree to

which the model is an accurate representation of the
real world from the perspective of the intended uses
of the model. (i.e. Solving the correct equations)

Source: AIAA Guide for the V&V of CFD Simulations

 Methods of V&V
Verification
Software engineering
Analytical solutions
Sod, Noh, Sedov
Benchmark numerical solutions
Collela-Woodward interacting blast waves
Conservation checks
Total energy
Validation
Experimental data

V&V requires quantitative metrics for comparisons

 What is a Conservation Law?
A conservation law is an equation that demonstrates
the global constancy of a particular physical quantity
for a closed system

V r ,t dV =constant

3
Position: r = x , y , z Time: t

3
Scalars: mass, energy :
3 3
Vectors: momentum :
Conservation Laws for Open Systems
An open system can exchange physical quantities
with its surroundings or with other systems (i.e. Flux)

We also allow the possibility of internal sources/sinks

V r , t 2 dV V r ,t 1 dV
t2 t2 t2
= t f in dt t f out dtt
1 1 1
V q dV dt

f in q f out
 Control Volume
A control volume V is a connected subset of space
of finite volume bounded by a surface S with finite area

S F
n
V

t2 t2
V r ,t 2 dV V r ,t 1 dV =t S Fn dSdtt V q dVdt
1 1

We shall assume that the control volume is fixed in space (Eulerian)

 Integral/Differential Equation Forms
t2 t t2
t V dVdt=t S Fn dSdtt V q dVdt
2

1 t 1 1

t2 t t2
t V dVdt=t V F dVdtt V q dVdt
2

1 t 1 1

[
]
t2
(I) t V
1 t
F q dVdt=0

(D) F = q (I) is true for all [t 1, t 2 ] ,V

t The term in brackets must be 0
We get (D)
 Continuity Equation

Many applications in continuum mechanics have

their flux defined by the following equation: F = v

When the source term is zero, we obtain the

Continuity Equation

v =0
t

e.g. Mass conservation: =

 Scalar Linear Advection Equations

Linear Advection Equation - Describes the transport of a

quantity u in 1D at constant velocity c

u u
c =0
t x

Advection-Diffusion Equation - Describes the transport of a

quantity u in 1D with constant velocity c and constant
diffusion coefficient D
2
u u u
c =D 2
t x x
 Traffic Flow

Describes the flow of traffic by tracking the density of

cars at any location x having velocity u

Continuity equation u=0
t x

Empirical relation u=u max

1

max

f =0
t x
 Berger's Equation

Simple 1D model of momentum transport in a fluid

ignoring: viscosity, density/pressure variations

u u
u =0
t x

u 1 2

t x 2
u =0
 Fokker-Plank Equation

Describes the evolution of the velocity distribution function f

v
The probability density of a particle having a velocity of v

f 2
v[ f v ]= D v f.
t

Applications to plasma physics and astrophysics

 Euler Equations
Physical model
Compressible fluid flow
Unsteady fluid motion (i.e. time-dependent)
Transport effects negligible (i.e. ignored)
No viscosity
No heat conduction
System of conservation laws
Mass
Momentum
Energy
Mathematical properties
Nonlinear
Strictly hyperbolic system
 Conservation of Mass

Mass cannot be created nor destroyed within a

control volume

We simply have the Continuity Equation

v =0
t
 Some Vector/Tensor Calculus

Gradient [ ]i = , i [ v ]ij =v i , j

Divergence v =v i , i [ A]i =a ij , j

Products vn =v i ni [ v v ]ij =v i v j

[ pI ]i = p ij , j = p

Index/Einstein summation specifies that:

comma denotes differentiation & repeated indicies are summed
 Some Vector/Tensor Theorems

Vectors Tensors

Gradient V dV =S n dS NA

Divergence V F dV =S Fn dS V A dV =S A n dS
T
Stokes S Fn dS=C Ft dc S A n dS=C A t dc
 Conservation of Momentum

Change in Momentum = Momentum Flux + External Forces

d
dt
V
v dV =S vn v dSS p n dS

v [ v v pI ]=0
t
 Conservation of Energy

Change in Energy = Energy Flux + Work Done

d
dt
V
E dV =S E vn dS S p vn dS

E [ E p v ]=0
t
 Equation of State

A closure condition is needed to relate pressure

to other fluid variables

We will mainly focus on the ideal gas EOS

p=1 e

Total Energy = Internal Energy + Kinetic Energy

1
E=e vv
2
 1D Planar Flow

U
F U =0
t x

[] [ ]
u
U= u F U = u 2 p
E E pu
1D Planar Isentropic & Isothermal Flows

U
F U =0
t x

U= [ ]

u
F U =
[ u
u 2 p ]
Isentropic Isothermal
2
p=C p=c
 1D Planar Linearized Flow

Describes small disturbances in a motionless gas

W W
A =0
t x

[ ]
0 0
W=

u[] A= c 2
0
0
 1D Systems of Conservation Laws
Conservation Form Conserved Flux
Variables Function
U
F U =0
t x

[] []
u1 f1
U x , t= F U =
Quasi-linear Form um fm
U U
AU =0
t x U :
m m
F :
m

Jacobian Matrix

fi
AU =
F
U [ ]
F
U ij
=
uj
m
A:
mm
 Eigenvalues & Eigenvectors
Eigenvalues i are solutions to the characteristic equation

A I=0

Left Eigenvectors Right Eigenvectors

i i i i
l A=i l A k =i k

[]
i
k1
l =[ l 1 lm ]
i i i i
k =
i
km
 Hyperbolic Conservation Laws

A system of conservation laws is hyperbolic if the Jacobian matrix A

has m real eigenvalues 1 , 2 , , m and a corresponding
set of linearly independent right eigenvectors k 1 , k 2 , , k m

The system is strictly hyperbolic if the eigenvalues are distinct

The system is linear if the Jacobian matrix A does not depend on U