SlideShare a Scribd company logo
1 of 19
Download to read offline
Foundations of CFD
(AM5630)
Outline
 What is CFD ?
 Steps in CFD
 Steps in CFD
 Governing Equations
 Tensor notation
CFD books
Richtmyer R D and Morton K W, Difference methods for
initial value problems, Inter science publishers, 1967.
Chung T J, Computational Fluid Dynamics, CUP, 2002
Anderson J, Essential Computational Fluid Dynamics
Joel H. Ferziger and Milovan Peric Computational Methods
Joel H. Ferziger and Milovan Peric Computational Methods
for Fluid Dynamics
John C. Tannehill, Dale Anderson Richard Pletcher,
Computational Fluid Mechanics and Heat Transfer, 2nd Ed.
Steps in CFD
 Identify the design/ optimization/ failure analysis
problem
 Identify a suitable Mathematical model
 Choose the appropriate flow domain of interest.
 Suitable PDE’s, IC’s and BC’s.
 Suitable PDE’s, IC’s and BC’s.
 Discretize the region of interest.
 Develop the system of algebraic equations.
 Solve the system of equations.
 Perform post-processing.
 Develop an engineering solution.
Continuum Hypothesis
V
m
V
V δ
δ
ρ
δ
δ *
lim
→
=
 Air at STP: δV*=10-9 mm3, which contains 3x107
molecules which is sufficient to define a nearly constant
density.
Forces on a blob of fluid
force
pressure
Net
on
Accelerati
Mass .
* =
v
δ
V
r
force
pressure
Net
on
Accelerati
Mass .
* =
g
x
p
Dt
D
j
ji r
r
ρ
τ
ρ +
∂
∂
+
∇
−
= )
(
V
g
v
forces
viscous
dS
p
Dt
D
v
S
r
r
)
(
)
(
V
)
( δ
ρ
δ
ρ +
+
−
= ∫
forces
viscous
+
g
v
forces
viscous
v
p
Dt
D
v
r
r
)
(
)
(
V
)
( δ
ρ
δ
δ
ρ +
+
∇
−
=
forces
Body
+
Equations of motion








∂
∂
+
∂
∂
+
∂
∂
+
∂
∂
−
=
∂
∂
+
∂
∂
+
∂
∂
+
∂
∂
2
2
2
2
2
2
1
z
u
y
u
x
u
x
p
z
u
w
y
u
v
x
u
u
t
u
ν
ρ








∂
∂
+
∂
∂
+
∂
∂
+
∂
∂
−
=
∂
∂
+
∂
∂
+
∂
∂
+
∂
∂
2
2
2
2
2
2
1
z
v
y
v
x
v
y
p
z
v
w
y
v
v
x
v
u
t
v
ν
ρ





 ∂
+
∂
+
∂
+
∂
∂
−
=
∂
∂
+
∂
∂
+
∂
∂
+
∂
∂ 2
2
2
1 w
w
w
p
w
w
w
v
w
u
w
ν
ρ
j
j
i
i
j
i
j
i
x
x
u
x
p
x
u
u
t
u
∂
∂
∂
+
∂
∂
−
=
∂
∂
+
∂
∂ 2
1
ν
ρ
0
=
∂
∂
i
i
x
u





 ∂
+
∂
+
∂
+
∂
−
=
∂
+
∂
+
∂
+
∂ 2
2
2
z
y
x
z
z
w
y
v
x
u
t
ν
ρ
0
=
∂
∂
+
∂
∂
+
∂
∂
z
w
y
v
x
u
 Some times called suffix notation. Much of the
notation can be introduced in the context of vectors
with so called Summation convention
 The rule is, if a suffix is repeated in any term, there is
an implied summation
Tensor Notation
j
i
j
x
u
u
∂
∂
3
2
1
x
u
u
x
u
u
x
u
u i
i
i
∂
∂
+
∂
∂
+
∂
∂
 Range convention: If a suffix occurs just once in a
term, then it is assumed to take all of the values 1,2,3
in turn.
 j is called dummy index – indicating an implied
summation.
 i is called free index – indicating the eqn. above
represents 3 components of a vector eqn.
3
3
2
2
1
1
x
u
x
u
x
u
∂
+
∂
+
∂
Tensor Notation (ctd…)
)
/
(
)
.
( ρ
p
t
−∇
=
∇
+
∂
∂
u
u
u








∂
∂
−
=
∂
∂
+
∂
∂
ρ
p
x
x
u
u
t
u
i
j
i
j
i
0
u =
∇.
i
i
x
u
∂
∂








∂
∂
∂
+








∂
∂
−
=
∂
∂
+
∂
∂
j
j
i
i
j
i
j
i
x
x
u
p
x
x
u
u
t
u 2
ν
ρ
u
u
u
u 2
)
/
(
)
.
( ∇
+
−∇
=
∇
+
∂
∂
ν
ρ
p
t
Disturbance propagation in a flow
Subsonic Sonic Supersonic
0
)
1
( 2
2
2
2
2
=
∂
∂
+
∂
∂
−
y
x
M
φ
φ
1

M 1
=
M 1

M
PDE - classification
0
2
2
2
2
2
=
+
+
∂
∂
+
∂
∂
+
∂
∂
+
∂
∂
∂
+
∂
∂
G
F
y
E
x
D
y
C
y
x
B
x
A φ
φ
φ
φ
φ
φ
0
4
2

− AC
B
0
4
2
=
− AC
B
Elliptic
Parabolic
0
4 =
− AC
B
0
4
2

− AC
B
Parabolic
Hyperbolic
Elliptic PDE
 Can be solved by specifying BC’s on a complete
contour enclosing the region.
 It is a BVP
 At all points in the domain Ω
Γ
0
4
2

− AC
B
 An elliptic PDE has no real characteristic curves.
 An elliptic PDE has no real characteristic curves.
 A disturbance is propagated instantly in all
directions in the region.
0
:
2
2
2
2
=
∂
∂
+
∂
∂
y
x
Examples
φ
φ
)
,
(
2
2
2
2
y
x
f
y
x
=
∂
∂
+
∂
∂ φ
φ
Parabolic PDE
 BC’s may be closed on one direction. But,
remain open at one end of the other direction.
 It is a mixed BVP
 At all points in the domain Ω
Γ
0
4
2
=
− AC
B
 For a parabolic PDE there exists one char. line.
 For a parabolic PDE there exists one char. line.
 Typically, an initial distribution of dependent
variable and two sets of BC’s are required for a
complete description of the problem.
2
2
:
x
T
t
T
Examples
∂
∂
=
∂
∂
α 2
2
x
u
t
u
∂
∂
=
∂
∂
ν
Hyperbolic PDE
 The eqns. can be solved by specifying the conditions
only at a portion of the boundary, the other
boundaries remain open.
 It is an IVP
 At all points in the domain
Ω
0
4
2

− AC
B
 For a Hyperbolic PDE, two
P
A
Γ
 For a Hyperbolic PDE, two
real char. exists per piont.
 Typically, we need to specify conditions at one
part of the boundary in order to determine the
solution in a given region.
2
2
2
2
2
:
x
a
t
Examples
∂
∂
=
∂
∂ φ
φ
A
B C
Tensor Notation (ctd…)
( )
j
k
ijk
i
i
x
u
∂
∂
=
×
∇
= ε
ω u
u
×
∇
( )
k
j
ikj
i
i
x
u
∂
∂
=
×
∇
= ε
ω u
( )








∂
∂
−
∂
∂
=
×
∇
=
k
j
j
k
ijk
i
i
x
u
x
u
ε
ω
2
1
u
?
3
2
1 ω
ω
ω
is
what
k
i
ik
x
u
EXPAND
∂
∂
τ
j
i
i
j
j
i
x
u
x
u
x
u
EXPAND
∂
∂








∂
∂
+
∂
∂
Outline
 PDE and their classification
 Understanding characteristics

When do you call a PDE is well-posed ?
 The problem in fact has a solution

 The solution is unique
 The solution is unique

 The solution depends continuously on
the data given in the problem.

PDE - classification
0
2
2
2
2
2
=
+
+
∂
∂
+
∂
∂
+
∂
∂
+
∂
∂
∂
+
∂
∂
G
F
y
E
x
D
y
C
y
x
B
x
A φ
φ
φ
φ
φ
φ
Assume ϕ = ϕ(x,y) is a solution of the differential equation.
By definition, the 2nd order derivatives along the
characteristic curves are indeterminate and indeed, they
characteristic curves are indeterminate and indeed, they
may be discontinuous across the characteristics.
Thus differentials of ϕx and ϕy is which
represent changes from location (x,y) to (x+dx, y+dy)
across the characteristics may be expressed as,
However, no discontinuity of the first derivatives is
allowed i.e., they are continuous functions of x and y.
Assume ϕ = ϕ(x,y) is a solution of the differential
equation.
PDE - classification
∂
∂
∂ 2
2
2
φ
φ
φ
dy
y
dx
x
d x
x
x
∂
∂
+
∂
∂
=
φ
φ
φ dy
y
x
dx
x ∂
∂
∂
+
∂
∂
=
φ
φ 2
2
2
dy
y
dx
x
d
y
y
y
∂
∂
+
∂
∂
=
φ
φ
φ dy
y
dx
y
x 2
2
2
∂
∂
+
∂
∂
∂
=
φ
φ
H
y
C
y
x
B
x
A =
∂
∂
+
∂
∂
∂
+
∂
∂
2
2
2
2
2
φ
φ
φ








+
+
∂
∂
+
∂
∂
−
= G
F
y
E
x
D
H
Where
φ
φ
φ
The three eqns. above can be used to solve for the 2nd order
derivatives of ϕ
Using Cramer’s rule,
2nd order PDE’s
dy
dx
dy
dx
C
B
A
dy
d
d
dx
C
H
A
y
x
y
x
0
0
0
0
2 φ
φ
φ
=
∂
∂
∂
Since it is possible to have
discontinuities in the 2nd order
derivatives of the dependent variable
across the characteristics, these
derivatives are indeterminate. Thus,
setting the denominator equal to zero,
0
0
0 =
dy
dx
dy
dx
C
B
A
Solving this quadratic yields the equations of the
characteristics in physical space.
This yields a quadratic equation.
0
2
=
+






−






C
dx
dy
B
dx
dy
A
?
,
=



β
α
dx
dy
Do it now !
0
)
1
(
2
2
2
=
∂
+
∂
− M
φ
φ
Classify the following equations :
0
1
;
0 2
2
≈
∂
∂
∂
∂
+
∂
∂
−
=
∂
∂
+
∂
∂
=
∂
∂
+
∂
∂
y
p
y
u
x
p
y
u
v
x
u
u
y
v
x
u
ν
ρ
0
2
2
2
2
2
=
+
+
∂
∂
+
∂
∂
+
∂
∂
+
∂
∂
∂
+
∂
∂
G
F
y
E
x
D
y
C
y
x
B
x
A φ
φ
φ
φ
φ
φ
0
)
1
( 2
2
=
∂
+
∂
−
y
x
M
Overview
Boundary conditions are a required component of the
mathematical model.
Boundaries direct motion of flow.
Specify fluxes into the computational domain, e.g.
mass, momentum, and energy.
mass, momentum, and energy.
Fluid and solid regions are represented by cell zones.
Material and source terms are assigned to cell zones.
Boundaries and internal surfaces are represented by
face zones.
Boundary data are assigned to face zones.
Boundary conditions
When solving the Navier-
Stokes equation and continuity
equation, appropriate initial
conditions and boundary
conditions need to be applied.
conditions need to be applied.
In the example here, a no-slip
boundary condition is applied at
the solid wall.
Neumann and Dirichlet boundary
conditions
When using a Dirichlet boundary condition, one prescribes
the value of a variable at the boundary, e.g. u(x) =
constant.
When using a Neumann boundary condition, one
When using a Neumann boundary condition, one
prescribes the gradient normal to the boundary of a
variable at the boundary, e.g. ∂nu(x) = constant.
When using a mixed boundary condition a function of the
form au(x)+b∂nu(x) = constant is applied.
Note that at a given boundary, different types of boundary
conditions can be used for different variables.
Outline
 Discretization
 Finite Differences

0
2
2
2
2
2
=
+
+
∂
∂
+
∂
∂
+
∂
∂
+
∂
∂
∂
+
∂
∂
G
F
y
E
x
D
y
C
y
x
B
x
A φ
φ
φ
φ
φ
φ
Discretization
1 i
i-1 i+1 N
i
i-1 i+1
j+1
i,j
j
j -1
Grid generation : Structured (vs) Unstructured Taylor Series
)
(
)
( i
x
x φ
φ =
i
i
x
x
x








∂
∂
−
+ 3
3
3
!
3
)
( φ
H
x
n
x
x
i
n
n
n
i
+








∂
∂
−
+
φ
!
)
(
K
i
i
x
x
x 





∂
∂
−
+
φ
)
(
i
i
x
x
x








∂
∂
−
+ 2
2
2
!
2
)
( φ
Any continuous differentiable
function (ϕ), in the vicinity of xi ,
can be expressed as a Taylor
series:
i
i
i
i
i
x
x
x
x
x 





∂
∂
−
+
≈ +
+
φ
φ
φ )
(
)
(
)
( 1
1
i
i
i
i
i
i
i
i
x
x
x
x
x
x
x
x 







∂
∂
−
+






∂
∂
−
+
≈ +
+
+ 2
2
2
1
1
1
!
2
)
(
)
(
)
(
)
(
φ
φ
φ
φ
i
n
n
n
i
i
n x
n
x
x
TE 







∂
∂
−
+
∞
=
∑
φ
!
)
(
: 1
3
i
n
n
n
i
i
n x
n
x
x
TE 







∂
∂
−
+
∞
=
∑
φ
!
)
(
: 1
2
Do it now !
i
i
i
i
i
x
x
x
x
x 





∂
∂
−
+
≈ +
+
φ
φ
φ )
(
)
(
)
( 1
1
i
i
i
i
i
x
x
x
x
x 





∂
∂
−
+
≈ −
−
φ
φ
φ )
(
)
(
)
( 1
1
)
(
)
(
)
(
)
(
1
1
x
O
x
x
x
x
x i
i
i
i
i
∆
+
−
−
=






∂
∂
+
+ φ
φ
φ
)
(
)
(
)
(
)
(
1
1
x
O
x
x
x
x
x i
i
i
i
i
∆
+
−
−
=






∂
∂
−
− φ
φ
φ
)
(
)
(
)
(
)
( 1
x
O
x
x
x
x
i
i
i
∆
+
∆
−
=






∂
∂ −
φ
φ
φ
2
)
(
? x
O
x i
∆
+
=






∂
∂φ
i
i
i
i
i
x
x
x
x
x
x
x
x 







∂
∂
−
+






∂
∂
−
+
≈ 2
2
2
!
2
)
(
)
(
)
(
)
(
φ
φ
φ
φ
Derivative and its approximation Validation and Verification
Are we solving the right equations ?
(vs)
Are we solving the equations right ?
Consistency
Consistency
0
→
TE
FDE
PDE
TE −
=
Do it now !
i
i
i
i
x
x
x
x
x
x
x 







∂
∂
∆
+






∂
∂
∆
+
≈
∆
+ 2
2
2
!
2
)
(
)
(
)
(
)
(
φ
φ
φ
φ
)
(
?
2
2
x
O
x i
∆
+
=








∂
∂ φ )
(
)
(
2
2
1
2
2
2
x
O
x
x
i
i
i
i
∆
+
∆
+
−
=








∂
∂ +
+ φ
φ
φ
φ
2
2
2
)
(
? x
O
x i
∆
+
=








∂
∂ φ
i
i
i
i
x
x
x
x
x
x
x 







∂
∂
∆
+






∂
∂
∆
+
≈
∆
+ 2
2
2
!
2
)
2
(
)
2
(
)
(
)
2
(
φ
φ
φ
φ
Outline
 Discretization
 Finite Differences
 Explicit (vs) Implicit
Do it now !
i
i
i
i
x
x
x
x
x
x
x 







∂
∂
∆
+






∂
∂
∆
+
≈
∆
+ 2
2
2
!
2
)
(
)
(
)
(
)
(
φ
φ
φ
φ
)
(
?
2
2
x
O
x i
∆
+
=








∂
∂ φ
i
i
i
i
x
x
x
x
x
x
x 







∂
∂
∆
+






∂
∂
∆
+
≈
∆
+ 2
2
2
!
2
)
2
(
)
2
(
)
(
)
2
(
φ
φ
φ
φ
)
(
)
(
2
)
1
(
*
2
)
2
( 2
1
2
2
2
x
O
x
x
Eq
Eq i
i
i
i
∆
+
∆
+
−
=








∂
∂
⇒
− +
+ φ
φ
φ
φ
Do it now !
2
2
2
)
(
? x
O
x i
∆
+
=








∂
∂ φ
i
i
x
x
x






∂
∂
∂
∂
=








∂
∂
)
(
2
2
φ
φ
x
x
x i
i
∆






∂
∂
−






∂
∂
≈ +
φ
φ
1
i
i

 x
∆
x
x
x
i
i
i
i
∆






∆
−
−






∆
−
≈
−
+ 1
1 φ
φ
φ
φ
( )
( )2
2
1
1 2
x
O
x
i
i
i
∆
+
∆
+
−
= −
+ φ
φ
φ
[ ]
2
2
2
)
(
,
)
(
? y
x
O
y
x i
∆
∆
+
=








∂
∂
∂ φ
Do it now !
[ ]
2
2
2
)
(
,
)
(
? y
x
O
y
x i
∆
∆
+
=








∂
∂
∂ φ
y
y j
i
j
i








∂
∂
−








∂
∂
≈
−
+ ,
1
,
1
φ
φ
( )( )
( ) ( )
[ ]
2
2
1
,
1
1
,
1
1
,
1
1
,
1
,
4
y
x
O
y
x
j
i
j
i
j
i
j
i
∆
∆
+
∆
∆
+
−
−
=
−
−
+
−
−
+
+
+ φ
φ
φ
φ
x
y
y j
i
j
i
∆

 ∂

 ∂
≈
−
+
2
,
1
,
1
Explicit (vs) Implicit
2
2
x
T
t
T
∂
∂
=
∂
∂
α
( )
2
2 T
T
T
T +
−
∂
( )
( )
t
O
t
T
T
t
T n
i
n
i
∆
+
∆
−
=
∂
∂ +1
Explicit
Implicit
( )
( )2
2
1
1
2
2
2
x
O
x
T
T
T
x
T i
i
i
∆
+
∆
+
−
=
∂
∂ −
+
( )
( )2
2
1
1 2
x
O
x
T
T
T
n
i
i
i
∆
+






∆
+
− −
+
( )
( )2
1
2
1
1 2
x
O
x
T
T
T
n
i
i
i
∆
+






∆
+
−
+
−
+
Outline
 Explicit (vs) Implicit
 Stability analysis
Explicit (vs) Implicit
2
2
x
T
t
T
∂
∂
=
∂
∂
α
( )
( )
t
O
t
T
T
t
T n
i
n
i
∆
+
∆
−
=
∂
∂ +1
1 i
i-1 i+1 N
n
n
n
T
T
T 
 ∂
−
+ 2
1 n
n
n
x
T
t
O
t
T
T








∂
∂
=
∆
+
∆
−
+
2
2
1
)
( α
1
2
2
1
)
(
+
+








∂
∂
=
∆
+
∆
−
n
n
n
x
T
t
O
t
T
T
α
Explicit
Implicit
Explicit Scheme
( )
( )
n
i
n
i
n
i
n
i
n
i T
T
T
x
t
T
T 1
1
2
1
2 −
+
+
+
−
∆
∆
+
=
α
( )
( ) )
(
2 2
2
1
1
1
2
2
t
O
x
O
x
T
T
T
t
T
T
x
T
t
T n
i
n
i
n
i
n
n
∆
+
∆
=








∆
+
−
−
∆
−
−








∂
∂
−
∂
∂ −
+
+
α
α
Explicit
1 i
i-1 i+1
N
Time level n
i
Time level n+1
i
Time level n+1
i-1 i+1
Implicit
( )
( )
1
1
1
1
1
2
1
2 +
−
+
+
+
+
+
−
∆
∆
+
= n
i
n
i
n
i
n
i
n
i T
T
T
x
t
T
T
α
Implicit Scheme
1 i
i-1 i+1
N
Time level n
n
i
n
i
i
n
i
i
n
i
i d
T
c
T
b
T
a =
+
+ +
−
+
+
−
1
1
1
1
1 TDMA
i
Time level n+1
i-1 i+1
Implicit
( )
( )
1
1
1
1
1
2
1
2 +
−
+
+
+
+
+
−
∆
∆
+
= n
i
n
i
n
i
n
i
n
i T
T
T
x
t
T
T
α
Semi-Implicit Scheme
1 i
i-1 i+1
N
Time level n
( )
( )
( )( )







+
−
−
+
+
−
∆
∆
+
=
−
+
+
−
+
+
+
+
n
i
n
i
n
i
n
i
n
i
n
i
n
i
n
i
T
T
T
T
T
T
x
t
T
T
1
1
1
1
1
1
1
2
1
2
1
2
β
β
α
β formulation
( ) 







∆
+
−
=
∆
− −
+
−
+
2
1
1
1
1
2
2 x
T
T
T
t
T
T n
i
n
i
n
i
n
n
α Richardson’s
( ) 







∆
+







 +
−
=
∆
−
−
−
+
+
−
+
2
1
1
1
1
1
1
2
2
2 x
T
T
T
T
t
T
T
n
i
n
i
n
i
n
i
n
n
α Dufort-
Frankel
Different Schemes
( )








∆
=
∆ 2
2 x
t
α
Frankel
( )
( )
( )
( ) 







∆
+
−
+
∆
+
−
=
∆
− −
+
+
−
+
+
+
+
2
1
1
2
1
1
1
1
1
1
2
2
2
1
x
T
T
T
x
T
T
T
t
T
T n
i
n
i
n
i
n
i
n
i
n
i
n
i
n
i
α
Crank-
Nicolson
Explicit
 The solution algorithm is simple to set up.
 For a given ∆x, ∆t must be less than a specific limit
imposed by stability constraints. This requires many
time steps to carry out the calculations over a given
interval of time.
Implicit
Implicit
 Stability can be maintained over much larger values of
∆t. Which implies, fewer time steps are needed to
carry out the calculations over a given interval.
 Since matrix manipulations are usually required at
each time step, the computer time per time step is
larger than that of the explicit approach.
Discrete perturbation analysis
After m steps
i-1 i+1
m
ε
i
After m steps
1 i-1 i+1
N
m
ε
m
ε
− m
ε
−
i
ε
1 i
i-1 i+1
N
Time level n
Time level n+1
1 i
i-1 i+1 N
i
After m steps
1 i-1 i+1
N
Stability of Explicit schemes
2
2
x
T
t
T
∂
∂
=
∂
∂
α
1 i
i-1 i+1 N
Explicit : FTCS
( )2
1
1
1
2
x
T
T
T
t
T
T n
i
n
i
n
i
n
i
n
i
∆
+
−
=
∆
− −
+
+
α
( )
x
t ∆
∆
Let us introduce a disturbance ε at
and find the influence on the grid points at higher time levels
n
i
T
( ) ( )
( )2
1
1
1
2
x
T
T
T
t
T
T n
i
n
i
n
i
n
i
n
i
∆
+
+
−
=
∆
+
− −
+
+
ε
α
ε
Stability of Explicit schemes
( ) ( )
( )2
1
1
1
2
x
T
T
T
t
T
T n
i
n
i
n
i
n
i
n
i
∆
+
+
−
=
∆
+
− −
+
+
ε
α
ε
i
T
assume n
i ∀
( )
( )2
1
0
2
0
x
t
T n
i
∆
+
−
=
∆
−
+
ε
α
ε
( ) 















∆
∆
−
=
+
2
1
2
1
x
t
T n
i α
ε
( ) 







∆
∆
= 2
x
t
d
let α
d
T n
i
2
1
1
−
=
+
ε
1
2
1
)
(
1
2
1 −
≥
−
≤
− d
or
d
1
≤
⇒ d
Stability of Explicit schemes
When the error reaches all the grid points, after many time
steps, approximately with the same magnitude,
2 possibilities may be considered :
(i) Error at time m, have the same sign
(ii) Error at time m, have alternate signs.
( )
( )2
1
2
x
t
T m
m
m
m
m
i
∆
+
−
=
∆
−
+
ε
ε
ε
α
ε
m
m
i
T ε
=
+1
No stability constraint
;
;
;
,
1
1
m
m
i
m
m
i
m
m
i
T
T
T
Let
ε
ε
ε
=
=
=
−
+
;
;
;
, 1
1
m
m
i
m
m
i
m
m
i T
T
T
Let ε
ε
ε −
=
=
−
= −
+
Under what conditions will the solution be stable, if the sign
of the error alternates ?
Stability of Explicit schemes
;
;
;
, 1
1
m
m
i
m
m
i
m
m
i T
T
T
Let ε
ε
ε −
=
=
−
= −
+
Under what conditions will the solution be stable, if the sign
of the error alternates ?
( )
m
m
m
m
m
i d
T ε
ε
ε
ε −
−
−
+
=
+
2
1
d
T
m
n
i
4
1
1
−
=
+
ε
( ) m
m
i d
T ε
4
1
1
−
=
+
Solution will be stable if, 1
4
1
)
(
1
1
≤
−
≤
+
d
or
T
m
n
i
ε
This requirement leads to,
( )2
2
1
x
t ∆
≤
∆
α
0

∂
∂
−
=
∂
∂
a
x
u
a
t
u
1 i
i-1 i+1 N
The quantity u is convected along these lines with a constant a.
A number of FD approximations can be constructed as follows :
u
u
u
u n
n
n
n
−
−
+1
Schemes for Hyperbolic Equations
( )
x
u
u
a
t
u
u n
i
n
i
n
i
n
i
∆
−
−
=
∆
− +
+
1
1
Eulers’ FTFS
Stability analysis indicates that, the method is unconditionally
unstable.
( )
x
u
u
a
t
u
u n
i
n
i
n
i
n
i
∆
−
−
=
∆
− −
+
+
2
1
1
1
Eulers’ FTCS
This explicit formulation is also unconditionally unstable.
Schemes for Hyperbolic Equations
0

∂
∂
−
=
∂
∂
a
x
u
a
t
u
1 i
i-1 i+1 N
( )
x
u
u
a
t
u
u n
i
n
i
n
i
n
i
∆
−
−
=
∆
− −
+
1
1
First order upwind
differencing
Stability analysis indicates that, this method is stable, when c ≤ 1
Stability analysis indicates that, this method is stable, when c ≤ 1
Use Forward differencing for the spatial derivative if a  0
( )
( )
x
u
u
a
t
u
u
u n
i
n
i
n
i
n
i
n
i
∆
−
−
=
∆
+
−
−
+
−
+
+
2
2
1
1
1
1
1
1
The Lax method
Stability analysis indicates that, this method is stable, when c ≤ 1.
( )
x
u
u
a
t
u
u n
i
n
i
n
i
n
i
∆
−
−
=
∆
− +
+
1
1
Sources of Error
2
2
x
T
t
T
∂
∂
=
∂
∂
α
1 i
i-1 i+1 N
Explicit : FTCS
( )2
1
1
1
2
x
T
T
T
t
T
T n
i
n
i
n
i
n
i
n
i
∆
+
−
=
∆
− −
+
+
α
( )
x
t ∆
∆
A : Analytical Solution
D : Exact solution of the difference equation
Discretization error = A - D
N : Numerical solution on a digital computer
Round-off error = N - D
D
N
Error −
=
ε
:
Stability of Explicit schemes
( ) ( )
( )2
1
1
1
1
1
1
2
x
D
D
D
t
D
D n
i
n
i
n
i
n
i
n
i
n
i
n
i
n
i
n
i
n
i
∆
+
+
+
−
+
=
∆
+
−
+ −
−
+
+
+
+
ε
ε
ε
α
ε
ε
( ) ( ) 1
1
1
2 D
D
D
D
D n
i
n
i
n
i
n
i
n
i +
−
=
− −
+
+
α
D : Exact solution of the difference equation; Hence, it exactly
satisfies the difference equation.
( ) ( )
( )2
1
1 2
x
D
D
D
t
D
D i
i
i
i
i
∆
+
−
=
∆
− −
+
α
( )
( )2
1
1
1
2
)
2
(
)
1
(
x
t
n
i
n
i
n
i
n
i
n
i
∆
+
−
=
∆
−
⇒
− −
+
+
ε
ε
ε
α
ε
ε
Error (ε) also satisfies the
difference equation; Solution is
stable, if
1
1
≤
+
n
i
n
i
ε
ε
Round-off error variation
Random variation of ε with x can be
analytically expressed as a Fourier series as
follows : x
ik
m
m
m
e
A
x ∑
=
)
(
ε
Round-off error variation
3
,
2
,
1
2
=






= m
m
L
km
π
∑
=
=
2
/
1
)
(
)
,
(
N
m
x
ik
m
m
e
t
A
t
x
ε
∑
=
=
2
/
1
)
(
N
m
x
ik
m
m
e
A
x
ε
However, we are interested in the variation of ε with time.
Stability of Explicit schemes
( )2
)
(
)
(
)
(
2
x
e
e
e
e
e
e
t
e
e
e
e x
x
ik
at
x
ik
at
x
x
ik
at
x
ik
at
x
ik
t
t
a m
m
m
m
m
∆
+
−
=
∆
− ∆
−
∆
+
∆
+
α
Simplify now !
x
ik
at m
e
e
by
divide ,
Use following identities :
2
)
cos(
x
ik
x
ik
m
m
m
e
e
x
k
∆
−
∆
+
=
∆
2
)
cos(
1
)
2
(
sin2 x
k
x
k m
m ∆
−
=
∆
Use following identities :
Outline
 Elliptic PDE
 Parabolic PDE (2-d)
 Iterative Methods
1-D parabolic PDE
2
2
x
T
t
T
∂
∂
=
∂
∂
α
1 i
i-1 i+1 N
Explicit : FTCS
( )2
1
1
1
2
x
T
T
T
t
T
T n
i
n
i
n
i
n
i
n
i
∆
+
−
=
∆
− −
+
+
α
( )
x
t ∆
∆
( ) 2
1
2
≤
∆
∆
x
t
α
Parabolic PDE Elliptic PDE
0
2
2
2
2
=
∂
∂
+
∂
∂
y
T
x
T
2
2 +
−
+
− T
T
T
T
T
T
)
,
(
2
2
2
2
y
x
f
y
T
x
T
=
∂
∂
+
∂
∂
( ) ( )
0
2
2
2
1
,
,
1
,
2
,
1
,
,
1
=
∆
+
−
+
∆
+
− −
+
−
+
y
T
T
T
x
T
T
T j
i
j
i
j
i
j
i
j
i
j
i
( )
( )
( ) 0
2
2 1
,
,
1
,
2
2
,
1
,
,
1 =
+
−
∆
∆
+
+
− −
+
−
+ j
i
j
i
j
i
j
i
j
i
j
i T
T
T
y
x
T
T
T
0
)
1
(
2 ,
2
1
,
2
1
,
2
,
1
,
1 =
+
−
+
+
+ −
+
−
+ j
i
j
i
j
i
j
i
j
i T
T
T
T
T β
β
β
Parabolic PDE : FTCS








∂
∂
+
∂
∂
=
∂
∂
2
2
2
2
y
T
x
T
t
T
α
( ) ( ) 







∆
+
−
+
∆
+
−
=
∆
− −
+
−
+
+
2
1
,
,
1
,
2
,
1
,
,
1
,
1
, 2
2
y
T
T
T
x
T
T
T
t
T
T n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
α
( ) ( ) 2
1
2
2
≤








∆
∆
+
∆
∆
y
t
x
t α
α
,
,
1 y
x
if ∆
=
∆
=
α
( ) 4
1
2
≤








∆
∆
⇒
x
t
( )
( )
( )
( )
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i T
T
T
y
t
T
T
T
x
t
T
T ,
1
,
,
1
2
,
1
,
,
1
2
,
1
, 2
2 −
+
−
+
+
+
−
∆
∆
+
+
−
∆
∆
+
=
α
α
Stability analysis indicates, the
method is stable if,
Elliptic PDE
0
2
2
2
2
=
∂
∂
+
∂
∂
y
T
x
T
( ) ( )
0
2
2
2
1
,
,
1
,
2
,
1
,
,
1
=
∆
+
−
+
∆
+
− −
+
−
+
y
T
T
T
x
T
T
T j
i
j
i
j
i
j
i
j
i
j
i
)
,
(
2
2
2
2
y
x
f
y
T
x
T
=
∂
∂
+
∂
∂
( ) ( )
∆
∆ y
x
( )
( )
( ) 0
2
2 1
,
,
1
,
2
2
,
1
,
,
1 =
+
−
∆
∆
+
+
− −
+
−
+ j
i
j
i
j
i
j
i
j
i
j
i T
T
T
y
x
T
T
T
0
)
1
(
2 ,
2
1
,
2
1
,
2
,
1
,
1 =
+
−
+
+
+ −
+
−
+ j
i
j
i
j
i
j
i
j
i T
T
T
T
T β
β
β
( )
( )
)
1
(
2 2
2
2
2
β
α
β +
−
=
∆
∆
=
y
x
Let
0
,
1
,
2
1
,
2
,
1
,
1 =
+
+
+
+ −
+
−
+ j
i
j
i
j
i
j
i
j
i T
T
T
T
T α
β
β
( )
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T 1
,
2
1
,
2
,
1
,
1
2
1
,
)
1
(
2
1
−
+
−
+
+
+
+
+
+
= β
β
β
3
T
Jacobi iterative method
1
T
2
T
4
T
The Analogy between iterative methods
( )
( )
( )
( )
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i T
T
T
y
t
T
T
T
x
t
T
T ,
1
,
,
1
2
,
1
,
,
1
2
,
1
, 2
2 −
+
−
+
+
+
−
∆
∆
+
+
−
∆
∆
+
=
α
α
( )
⇒
=








∆
∆
=
4
1
,
1 2
x
t
α ( )
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i T
T
T
T
T
T
T ,
1
,
1
,
,
1
,
1
,
1
, 4
4
1
−
+
−
+
+
+
+
−
+
+
=
( )
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T 1
,
2
1
,
2
,
1
,
1
2
1
,
)
1
(
2
1
−
+
−
+
+
+
+
+
+
= β
β
β
( )
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i T
T
T
T
T ,
1
,
1
,
1
,
1
1
,
4
1
−
+
−
+
+
+
+
+
=
From the discretization of an Elliptic PDE,
( )
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T 1
,
1
,
,
1
,
1
1
,
4
1
1 −
+
−
+
+
+
+
+
=
⇒
=
β
Parabolic PDE : Implicit








∂
∂
+
∂
∂
=
∂
∂
2
2
2
2
y
T
x
T
t
T
α
( ) ( ) 







∆
+
−
+
∆
+
−
=
∆
− +
−
+
+
+
+
−
+
+
+
+
2
1
1
,
1
,
1
1
,
2
1
,
1
1
,
1
,
1
,
1
, 2
2
y
T
T
T
x
T
T
T
t
T
T n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
α
N
n
j
i
n
j
i
y
n
j
i
y
n
j
i
y
x
n
j
i
x
n
j
i
x T
T
d
T
d
T
d
d
T
d
T
d ,
1
1
,
1
1
,
1
,
1
,
1
1
,
1 )
1
2
2
( −
=
+
+
+
+
−
+ +
−
+
+
+
+
−
+
+
n
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i f
T
e
T
d
T
c
T
b
T
a ,
1
1
,
,
1
1
,
,
1
,
,
1
,
1
,
1
,
1
, =
+
+
+
+ +
−
+
+
+
+
−
+
+
1
1
,
2
2
,
2
1
2
,
1
2
,
2
2
,
2
1
3
,
2
2
,
2
1
2
,
2
2
,
2
1
2
,
3
2
,
2
+
+
+
+
+
−
−
=
+
+ n
n
n
n
n
n
T
e
T
b
f
T
d
T
c
T
a
In a matrix form Solving a system of Equations
n
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i f
T
e
T
d
T
c
T
b
T
a ,
1
1
,
,
1
1
,
,
1
,
,
1
,
1
,
1
,
1
, =
+
+
+
+ +
−
+
+
+
+
−
+
+
 Direct Methods
Cramers’ rule
Cramers’ rule
 Gauss Elimination
 Iterative Methods
 Point iterative methods
Line iterative methods
Outline
 Elliptic PDE
 Parabolic PDE (2-d)
 Approximate factorization
TDMA (Thomas Algorithm)
2
2
x
T
t
T
∂
∂
=
∂
∂
α
1 i
i-1 i+1 N
Implicit scheme
( )2
1
1
1
1
1
1
2
x
T
T
T
t
T
T n
i
n
i
n
i
n
i
n
i
∆
+
−
=
∆
− +
−
+
+
+
+
α
( )
x
t ∆
∆
( )
( )
1
1
1
1
1
2
1
2 +
−
+
+
+
+
+
−
∆
∆
+
= n
i
n
i
n
i
n
i
n
i T
T
T
x
t
T
T
α
n
i
n
i
i
n
i
i
n
i
i d
T
c
T
b
T
a =
+
+ +
−
+
+
−
1
1
1
1
1 TDMA
In a matrix form Gauss Elimination
Thomas Algorithm
Thomas Algorithm
This slide is only notional !
Elliptic PDE
0
2
2
2
2
=
∂
∂
+
∂
∂
y
T
x
T
( ) ( )
0
2
2
2
1
,
,
1
,
2
,
1
,
,
1
=
∆
+
−
+
∆
+
− −
+
−
+
y
T
T
T
x
T
T
T j
i
j
i
j
i
j
i
j
i
j
i
)
,
(
2
2
2
2
y
x
f
y
T
x
T
=
∂
∂
+
∂
∂
( ) ( )
∆
∆ y
x
( )
( )
( ) 0
2
2 1
,
,
1
,
2
2
,
1
,
,
1 =
+
−
∆
∆
+
+
− −
+
−
+ j
i
j
i
j
i
j
i
j
i
j
i T
T
T
y
x
T
T
T
0
)
1
(
2 ,
2
1
,
2
1
,
2
,
1
,
1 =
+
−
+
+
+ −
+
−
+ j
i
j
i
j
i
j
i
j
i T
T
T
T
T β
β
β
( )
( )
)
1
(
2 2
2
2
2
β
α
β +
−
=
∆
∆
=
y
x
Let
In a matrix form
Parabolic PDE : Implicit








∂
∂
+
∂
∂
=
∂
∂
2
2
2
2
y
T
x
T
t
T
α
( ) ( ) 







∆
+
−
+
∆
+
−
=
∆
− +
−
+
+
+
+
−
+
+
+
+
2
1
1
,
1
,
1
1
,
2
1
,
1
1
,
1
,
1
,
1
, 2
2
y
T
T
T
x
T
T
T
t
T
T n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
n
j
i
α
N
n
j
i
n
j
i
y
n
j
i
y
n
j
i
y
x
n
j
i
x
n
j
i
x T
T
d
T
d
T
d
d
T
d
T
d ,
1
1
,
1
1
,
1
,
1
,
1
1
,
1 )
1
2
2
( −
=
+
+
+
+
−
+ +
−
+
+
+
+
−
+
+
n
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i f
T
e
T
d
T
c
T
b
T
a ,
1
1
,
,
1
1
,
,
1
,
,
1
,
1
,
1
,
1
, =
+
+
+
+ +
−
+
+
+
+
−
+
+
1
1
,
2
2
,
2
1
2
,
1
2
,
2
2
,
2
1
3
,
2
2
,
2
1
2
,
2
2
,
2
1
2
,
3
2
,
2
+
+
+
+
+
−
−
=
+
+ n
n
n
n
n
n
T
e
T
b
f
T
d
T
c
T
a
In a matrix form
 The coefficient matrix is Pentadiagonal .
 Solution procedure is very time consuming.
 How to over come this inefficiency ?
ADI : Alternating Direction Implicit method Fractional Step Methods
Approximate Factorization methods
 What is the order of accuracy of this scheme ?
Fractional Step Methods
Solving a system of Equations
n
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i f
T
e
T
d
T
c
T
b
T
a ,
1
1
,
,
1
1
,
,
1
,
,
1
,
1
,
1
,
1
, =
+
+
+
+ +
−
+
+
+
+
−
+
+
 Direct Methods
Cramers’ rule
Cramers’ rule
 Gauss Elimination
 Iterative Methods
 Point iterative methods
Line iterative methods
Outline
 ADI
 Fractional step methods
 Approximate factorization

ADI : Alternating Direction Implicit method Fractional Step Methods
The Navier-Stokes equations
u
u
u
u 2
p
1
∇
+
∇
−
=
∇
+
∂
∂
ν
ρ
t
2
2
x
T
t
T
∂
∂
=
∂
∂
α
( )
( )
1
1
1
1
1
2
1
2 +
−
+
+
+
+
+
−
∆
∆
+
= n
i
n
i
n
i
n
i
n
i T
T
T
x
t
T
T
α
0
=
⋅
∇ u
n
2
n
n
n
n
1
n
p
1
u
u
u
u
-
u
∇
+
∇
−
∇
⋅
−
=
∆
+
ν
ρ
t
Anything wrong ?
Governing equations
n
2
n
n
n
n
1
n
p
1
u
u
u
u
-
u
∇
+
∇
−
∇
⋅
−
=
∆
+
ν
ρ
t
0
1
n
≠
⋅
∇ +
u
n
2
n
n
n
n
1
n
p u
u
u
u
u ∇
∆
+
∇
⋅
∆
−
=
∇
∆
+
+
t
t
t
ν
ρ
( )
n
2
n
n
n
1
n
2
p u
u
u
u ∇
∆
+
∇
⋅
∆
−
⋅
∇
∆
=
∇ +
t
t
t
ν
ρ
ρ
0
1
n
=
⋅
∇ +
u
Is there any problem now ?








∇
+
∇
−
∇
⋅
−
∆
+
= n
2
n
n
n
n
p
1
~ u
u
u
u
u ν
ρ
t
Operator Splitting Methods








∇
+
∇
−
∇
⋅
−
∆
+
= n
2
n
n
n
n
p
~ u
u
u
u
u ν
ρ
β
t






∇
+
∇
−
∇
⋅
−
∆
+
= +
+ n
2
1
n
n
n
n
1
n
p
1
u
u
u
u
u ν
ρ
t 





∇
+
∇
−
∇
⋅
−
∆
+
= p u
u
u
u
u ν
ρ
t
( )
n
1
n
1
n
p
p
~ β
ρ
−
∇
∆
−
=
− +
+ t
u
u
φ
ρ
∇
∆
−
=
−
+ t
u
u ~
1
n
( )
n
1
n
p
p β
φ −
= +
u
~
2
⋅
∇
∆
−
=
∇
t
ρ
φ
Operator Splitting Methods
φ
∇
∆
−
=
+ t
u
u ~
1
n
u
~
2
⋅
∇
∆
−
=
∇
t
ρ
φ
φ
β +
=
+ n
1
n
p
p
( )
Γ
+
Γ
−
∆
=
∇ u
u ~
1
n
ρ
φ
t
What is a suitable BC for φ ?
φ
ρ
∇
∆
−
=
+ t
u
u ~
1
n
(1) Predict
(2) Compute
(3) Compute the new velocity and pressure field
u
~
φ
1
n
1
n
p +
+
u
Fractional step Methods
The above eqn. can be split as,
1
n
2
1
n
n
n
n
1
n
p
1 +
+
+
∇
+
∇
−
∇
⋅
−
=
∆
u
u
u
u
-
u
ν
ρ
t
0
1
n
=
⋅
∇ +
u
0
~ n
n
n
=
∇
⋅
∆
+ u
u
u
-
u t
1
n
1
n ~
~ +
+
∇
∆
−
= p
t
ρ
u
u
u
~
~
1
2
⋅
∇
∆
−
=
∇ +
t
pn ρ
0
=
∇
⋅
∆
+ u
u
u
-
u t
0
~
~
~
~
~ 2
=
∇
∆
+
= u
u
u ν
t
0
1
n
=
⋅
∇ +
u
Operator Splitting Methods








∇
+
∇
−
∇
⋅
−
∆
+
= n
2
n
n
n
n
p
~ u
u
u
u
u ν
ρ
β
t
u
~
2
⋅
∇
∆
=
∇
t
ρ
φ
φ
β +
=
+ n
1
n
p
p φ
ρ
∇
∆
−
=
+ t
u
u ~
1
n
φ
β +
= p
p
ρ
(1) Predict
(2) Compute
(3) Compute the new velocity and pressure field
u
~
φ
1
n
1
n
p +
+
u
x
v
y
u
∂
∂
−
=
∂
∂
=
ψ
ψ








∂
∂
−
∂
∂
−
=
∂
∂
+
∂
∂
y
u
x
v
y
x 2
2
2
2
ψ
ψ
Outline
 Assignment 3  Q(3).
 Iterative methods

Assignment 3 – Q(3)
0
2
2
2
2
=
∂
∂
+
∂
∂
y
T
x
T
( ) ( )
0
2
2
2
1
,
,
1
,
2
,
1
,
,
1
=
∆
+
−
+
∆
+
− −
+
−
+
y
T
T
T
x
T
T
T j
i
j
i
j
i
j
i
j
i
j
i
( )
( )
)
1
(
2 2
2
2
2
β
α
β +
−
=
∆
∆
=
y
x
If
0
)
1
(
2 ,
2
1
,
2
1
,
2
,
1
,
1 =
+
−
+
+
+ −
+
−
+ j
i
j
i
j
i
j
i
j
i T
T
T
T
T β
β
β
( )
1
Jacobi Iteration
Gauss-Seidel iteration
 Line Gauss-Seidel iteration
Iterative Techniques
( )
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T 1
,
2
1
,
2
,
1
,
1
2
1
,
)
1
(
2
1
−
+
−
+
+
+
+
+
+
= β
β
β
( )
1
1
,
2
1
,
2
1
,
1
,
1
2
1
,
)
1
(
2
1 +
−
+
+
−
+
+
+
+
+
+
= k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T β
β
β
( )
1
1
,
1
,
2
1
,
1
1
,
2
1
,
1 )
1
(
2 +
−
+
+
+
+
+
− +
−
=
+
+
− k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T β
β
Setting up the Eqns.
( )
1
1
,
2
1
,
2
1
,
1
,
1
2
1
,
)
1
(
2
1 +
−
+
+
−
+
+
+
+
+
+
= k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T β
β
β
Gauss-Seidel iteration
( )
1
,
1
,
,
1
,
1
2
,
)
1
(
2
−
+
−
+ +
+
+
+
= j
i
j
i
j
i
j
i
j
i T
T
T
T
T β
β
β
( )
1
1
,
1
2
3
,
1
2
1
2
,
0
2
,
2
2
1
2
,
1
)
1
(
2
1 +
+
+
+
+
+
+
= k
k
k
k
k
T
T
T
T
T β
β
β
Can we accelerate the convergence ?.
( )
1
1
,
2
1
,
2
1
,
1
,
1
2
1
,
)
1
(
2
1 +
−
+
+
−
+
+
+
+
+
+
= k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T β
β
β
Point Successive Over Relaxation (PSOR)
RHS
on
T
subtract
add k
j
i,
( )
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T
T
T ,
2
1
1
,
2
1
,
2
1
,
1
,
1
2
,
1
, )
1
(
2
)
1
(
2
1
β
β
β
β
+
−
+
+
+
+
+
= +
−
+
+
−
+
+
( )
( )
1
1
,
1
,
2
1
,
1
,
1
2
,
1
,
)
1
(
2
)
1
( +
−
+
+
−
+
+
+
+
+
+
+
−
= k
j
i
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T
T β
β
ω
ω
As solution progresses, must approach
k
j
i
T,
1
,
+
k
j
i
T
Do we have an optimum, ?
ω
Optimum ω
 No general guidelines.
 Optimum is calculated for limited applications
 Elliptic + Dirchlet BC’s
Can we accelerate the convergence ?.
Line Successive Over Relaxation (LSOR)
( )
1
1
,
1
,
2
1
,
1
1
,
2
1
,
1 )
1
(
2 +
−
+
+
+
+
+
− +
−
=
+
+
− k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T β
β
1
1
2
1
)
1
(
2 +
+
+
+
+
− k
k
k
T
T
T ω
β
ω
Do we have an optimum, ?
ω
( )
1
1
,
1
,
2
,
2
1
,
1
1
,
2
1
,
1
)
1
)(
1
(
2
)
1
(
2
+
−
+
+
+
+
+
−
+
−
−
+
=
+
+
−
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i
T
T
T
T
T
T
ωβ
ω
β
ω
β
ω
LSOR can be introduced to ADI as well !
Outline
 Mid Term Exam
1st March 2012 (Thursday), 8-8:50 am
(Venue : CRC 103)
 Iterative Methods (ctd…)
 Review for Mid-Term Exam

Solving a system of Equations
n
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i
n
j
i
j
i f
T
e
T
d
T
c
T
b
T
a ,
1
1
,
,
1
1
,
,
1
,
,
1
,
1
,
1
,
1
, =
+
+
+
+ +
−
+
+
+
+
−
+
+
 Direct Methods
 Cramers’ rule
 Gauss Elimination
 Iterative Methods
 Iterative Methods
 Point iterative methods
 Jacobi (simplest)
 Gauss-Seidel
 PSOR
 Line iterative methods
 Line Gauss-Seidel
 LSOR
The End
k 1
+
k
Simple Iterative solvers at a glance
0
2
2
2
2
=
∂
∂
+
∂
∂
y
T
x
T
( ) ( )
0
2
2
2
1
,
,
1
,
2
,
1
,
,
1
=
∆
+
−
+
∆
+
− −
+
−
+
y
T
T
T
x
T
T
T j
i
j
i
j
i
j
i
j
i
j
i
( )
( )2
2
2
y
x
If
∆
∆
=
β
0
)
1
(
2 ,
2
1
,
2
1
,
2
,
1
,
1 =
+
−
+
+
+ −
+
−
+ j
i
j
i
j
i
j
i
j
i T
T
T
T
T β
β
β 0
)
1
(
2 ,
1
,
1
,
,
1
,
1 =
+
−
+
+
+ −
+
−
+ j
i
j
i
j
i
j
i
j
i T
T
T
T
T β
β
β
( )
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T 1
,
2
1
,
2
,
1
,
1
2
1
,
)
1
(
2
1
−
+
−
+
+
+
+
+
+
= β
β
β
( )
1
1
,
2
1
,
2
1
,
1
,
1
2
1
,
)
1
(
2
1 +
−
+
+
−
+
+
+
+
+
+
= k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T β
β
β
( )
1
1
,
1
,
2
1
,
1
1
,
2
1
,
1 )
1
(
2 +
−
+
+
+
+
+
− +
−
=
+
+
− k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T β
β
Can we accelerate the convergence ?.
( )
1
1
,
2
1
,
2
1
,
1
,
1
2
1
,
)
1
(
2
1 +
−
+
+
−
+
+
+
+
+
+
= k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T β
β
β
Point Successive Over Relaxation (PSOR)
RHS
on
T
subtract
add k
j
i,
( )
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T
T
T ,
2
1
1
,
2
1
,
2
1
,
1
,
1
2
,
1
, )
1
(
2
)
1
(
2
1
β
β
β
β
+
−
+
+
+
+
+
= +
−
+
+
−
+
+
( )
( )
1
1
,
1
,
2
1
,
1
,
1
2
,
1
,
)
1
(
2
)
1
( +
−
+
+
−
+
+
+
+
+
+
+
−
= k
j
i
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T
T β
β
ω
ω
As solution progresses, must approach
k
j
i
T,
1
,
+
k
j
i
T
Do we have an optimum, ?
ω
Optimum ω
 No general guidelines are available.
 Optimum is calculated for limited applications
 Elliptic + Dirchlet BC’s
Can we accelerate the convergence ?.
Line Successive Over Relaxation (LSOR)
( )
1
1
,
1
,
2
1
,
1
1
,
2
1
,
1 )
1
(
2 +
−
+
+
+
+
+
− +
−
=
+
+
− k
j
i
k
j
i
k
j
i
k
j
i
k
j
i T
T
T
T
T β
β
Can you modify this by introducing ω
Do we have an optimum, ?
ω
( )
1
1
,
1
,
2
,
2
1
,
1
1
,
2
1
,
1
)
1
)(
1
(
2
)
1
(
2
+
−
+
+
+
+
+
−
+
−
−
+
=
+
+
−
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i
k
j
i
T
T
T
T
T
T
ωβ
ω
β
ω
β
ω
LSOR can be introduced to ADI as well !
ADI  for Parabolic PDE
ADI  for Elliptic PDE
The solution procedure can be accelerated by
introducing relaxation parameter )
(ω
Outline
 Streamfunction-vorticity formulations
 Incorporation of upwind for
Governing equation for vorticity transport
u
u
u
u 2
1
∇
+
∇
−
=
∇
⋅
+
∂
∂
ν
ρ
p
t






∇
+
∇
−
=
∇
⋅
+
∂
∂
×
∇ u
u
u
u 2
1
ν
ρ
p
t
u
ω ×
∇
=
u)
u
2
1
u
u)
u
u ⋅
∇
+
×
×
∇
=
∇
⋅ (
(
u
ω
-
ω
u
u)
ω )
(
)
(
( ∇
⋅
∇
⋅
=
×
×
∇
0
ω =
⋅
∇
ω
u)
ω
ω 2
( ∇
=
×
×
∇
+
∂
∂
ν
t
u)
u ×
∇
∇
=
∇
×
∇ (
2
2
u
ω
-
ω
u
u)
ω )
(
)
(
( ∇
⋅
∇
⋅
=
×
×
∇
ω
u
ω
ω 2
)
( ∇
+
∇
⋅
= ν
Dt
D
Tracking vorticity distributions
Vorticity

 Vorticity ω is governed by an evolution eqn. which is
much simpler than N-S.
 Unlike u, ω can neither be created nor destroyed in
the fluid interior.
u
ω ×
∇
=
the fluid interior.
 It is transported throughout the flow field by familiar
processes such as, advection and diffusion.
 Localized distributions of ω remain localized, which
is not the case with velocity field.
 So, an Eddy in a turbulent flow blob of vorticity
and its associated rotational and irrotational motions.

Governing Equations
Stability is governed by,
Operator Splitting Methods








∇
+
∇
−
∇
⋅
−
∆
+
= n
2
n
n
n
n
p
~ u
u
u
u
u ν
ρ
β
t
u
~
2
⋅
∇
∆
=
∇
t
ρ
φ
φ
β +
=
+ n
1
n
p
p φ
ρ
∇
∆
−
=
+ t
u
u ~
1
n
φ
β +
= p
p
ρ
(1) Predict
(2) Compute
(3) Compute the new velocity and pressure field
u
~
φ
1
n
1
n
p +
+
u
x
v
y
u
∂
∂
−
=
∂
∂
=
ψ
ψ








∂
∂
−
∂
∂
−
=
∂
∂
+
∂
∂
y
u
x
v
y
x 2
2
2
2
ψ
ψ
Governing Equations
Central difference form for the above equation,
The above difference equation numerically allows
checkerboard velocity distribution.
Discrete Checkerboard velocity
Discrete Checkerboard for pressure Governing Equations
u
u
u
u 2
p
1
∇
+
∇
−
=
∇
+
∂
∂
ν
ρ
t
0
=
⋅
∇ u
For viscous, incompressible flows :
ω
∂
In non-primitive (ψ-ω) form :
ω
u)
ω
ω 2
( ∇
=
×
×
∇
+
∂
∂
ν
t
u
ω ×
∇
=
x
v
y
u
∂
∂
−
=
∂
∂
=
ψ
ψ
;








∂
∂
+
∂
∂
=
∂
∂
+
∂
∂
+
∂
∂
2
2
2
2
y
x
y
v
x
u
t
ω
ω
ν
ω
ω
ω








∂
∂
−
∂
∂
−
=
∂
∂
+
∂
∂
y
u
x
v
y
x 2
2
2
2
ψ
ψ
MAC algorithm
 Harlow and Welch (1965)
 One of the earliest methods of solving N-S in
primitive variables.
 Hirt and Cook (1972)
 A pressure Poisson equation is formulated
 A pressure Poisson equation is formulated
 The momentum equations are used for the
computation of velocities.
Temporal Evolution MAC algorithm
Define pressure correction terms as :
MAC algorithm
Define pressure correction terms as :
Continuity equation transforms into :
Governing Equations
Central difference form for the above equation,
The above difference equation numerically allows
checkerboard velocity distribution.
Discrete Checkerboard velocity Discrete Checkerboard for pressure
Outline
 Checker board pressure patterns
 Staggered grids

One-Dimensional N-S
)
2
(
1
)
1
(
0
2
2
x
u
x
p
x
u
u
t
u
x
u
∂
∂
+
∂
∂
−
=
∂
∂
+
∂
∂
=
∂
∂
ν
ρ
1 i
i-1 i+1 N
1
+
n
n
)
3
(
1
1
2
2
1 +
+








∂
∂
−








∂
∂
+
∂
∂
−
=
∆
−
n
n
n
i
n
i
x
p
x
u
x
u
u
t
u
u
ρ
ν
)
4
(
~
2
2 n
n
i
i
x
u
x
u
u
t
u
u








∂
∂
+
∂
∂
−
=
∆
−
ν








∂
∂
−
=
∆
−
⇒
−
+
x
p
t
u
u i
n
i
ρ
1
~
)
4
(
)
3
(
1
One-Dimensional N-S
1 i
i-1 i+1 N
0
2
1
1
1
1
=
∆
−
=
∂
∂ +
−
+
+
x
u
u
x
u n
i
n
i
)
4
(
1
~
1








∂
∂
−
=
∆
−
−
+
x
p
t
u
u i
n
i
ρ








∆
−
−
=
∆
− −
+
+
x
p
p
t
u
u i
i
i
n
i
2
1
~
1
1
1
ρ
x
p
p
t
u
u i
i
i
n
i
∆
−
∆
−
= −
+
+
2
~ 1
1
1
ρ
?
1
1 =
+
+
n
i
u ?
1
1 =
+
−
n
i
u
1
1
2
2 ~
~
2
2
−
+
−
+
−
=
∆
+
−
∆
i
i
i
i
i
u
u
x
p
p
p
t
ρ
Checkerboard pressures
 This delinking between velocity and pressure
at a grid point is an impediment and results in
zig-zag type checkerboard pressure values.
 To avoid this, MAC algorithm introduces
Staggered grids in place of “collocated grids”.
Staggered grids in place of “collocated grids”.

One-Dimensional N-S
1 i
i-1 i+1
N
( )
0
2
/
2
1
2
/
1
1
2
/
1
=
∆
−
=
∂
∂ +
−
+
+
x
u
u
x
u n
i
n
i
i-1/2 i+1/2 i+3/2
( )
2
/
1
2
/
1
1
1 ~
~
2
−
+
−
+
−
∆
=
∆
+
−
i
i
i
i
i
u
u
t
x
p
p
p ρ
( )








∆
−
−
=
∆
− +
+
+
+
x
p
p
t
u
u i
i
i
n
i 1
2
/
1
1
2
/
1 1
~
ρ
2D - Governing Equations
Central difference form for the above equation,
The above difference equation numerically allows the
checkerboard velocity distribution.
Such a distribution is not representative of a physical
flow field.
This is NOT an issue for compressible flows, as
inclusion of density variations would wipe out
checkerboard pressure pattern.
Discrete Checkerboard velocity Discrete Checkerboard for pressure Pressure Correction schemes
The 2-D Navier-Stokes A Staggered grid Staggered grid
)
1
(
0
=
∂
∂
+
∂
∂
y
v
x
u
Write a central
Write a central
differencing expression
for the above equation,
around the grid point (i,j).
Staggered grid
Difference equation for
x-momentum equation
about (i+1/2,j).
Outline
 Governing Equations in FM
 Differential form : FDM
 Integral form : FVM, FEM


Finite Volume Method
 FDM  regular grids
 Industrial problems  Complex domains
 FDM  coordinate transformation
 Loss of computational efficiency and accuracy


 FVM  Integral form of eqns. (greater flexibility in
handling complex domains)
 In FVM  conservation laws are applied on the
elementary volumes.

1-D steady diffusion equation
)
1
(
)
(
)
(
)
(
φ
φ
φ
ρ
ρφ
S
V
t
+
∇
Γ
⋅
∇
=
⋅
∇
+
∂
∂ r
)
2
(
0
)
( =
+
∇
Γ
⋅
∇ φ
φ S
Generic transport equation for the property ϕ
)
2
(
0
)
( =
+
∇
Γ
⋅
∇ φ
φ S
)
3
(
0
)
( =
+
∇
Γ
⋅
∇ ∫
∫ CV
CV
dv
S
dv φ
φ
)
4
(
0
)
( =
+
∇
Γ ∫
∫ CV
A
dv
S
dA φ
φ
1-D steady diffusion equation
)
5
(
0
=
+






Γ S
dx
d
dx
d φ
1-D diffusion equation for the property ϕ

 dT
d
)
6
(
0
=
+






S
dx
dT
k
dx
d
Discretization in FV
Physical boundary that coincides with CV boundary.
Discretization
)
5
(
0
=
+






Γ S
dx
d
dx
d φ
)
6
(
0
=
+






Γ ∫
∫ CV
CV
dv
S
dv
dx
d
dx
d φ
)
7
(
0
=
+






Γ
−






Γ dv
S
dx
d
A
dx
d
A
w
e
φ
φ

 −

 d φ
φ
φ =


Γ
dφ
)
8
(







 −
Γ
=






Γ
PE
P
E
e
e
e x
A
dx
d
A
δ
φ
φ
φ
)
10
(
P
P
u S
S
dv
S φ
+
=
)
9
(
?
=






Γ
w
dx
d
A
φ
Substitute (8), (9) and (10) in Eqn.(7) and rearrange NOW!
Discretization
Discretized equations of the above form must be setup at
each nodal point.
Modify the discretized equation to incorporate BC.
Solve the resulting system of linear algebraic equations.
1-D Heat conduction example
At grid points 2, 3, 4 :
1-D Heat conduction example
At the boundary
point 1:
?
How
dx
d
A
dx
d
A
w
e






Γ
−






Γ
φ
φ
Rearrange the above eqn. in the following form :
dx
dx w
e 



1-D Heat conduction example
At the boundary
point 5:
?
How
dx
d
A
dx
d
A
w
e






Γ
−






Γ
φ
φ
Rearrange the above eqn. in the following form :
dx
dx w
e 



Outline
 FVM
 1-D heat conduction equation
 2-D diffusion equation
 1-D convection-diffusion equation
 1-D convection-diffusion equation

Finite Volume Method
 FVM  Integral form of the conservation laws are
discretized directly in the physical space.
 Use a mesh (where the cell centre refers to the grid
points, while the cell faces coincide with the domain
boundaries).
 FVM has great advantage that the conservative
 FVM has great advantage that the conservative
discretization is automatically satisfied (by directly
using the integral form of the conservation laws).

1-D Heat conduction example
At grid points 2, 3, 4 :
E
E
W
W
P
P T
a
T
a
T
a +
=
At the boundary
point 1:
?
How
dx
d
A
dx
d
A
w
e






Γ
−






Γ
φ
φ
BC at A : Applying TA
Rearrange the above eqn. in the following form :
dx
dx w
e 



BC at B : Applying TB
At the boundary
point 5:
?
How
dx
d
A
dx
d
A
w
e






Γ
−






Γ
φ
φ
Rearrange the above eqn. in the following form :
dx
dx w
e 



1-D Heat conduction example
At the boundary
point 5:
?
How
dx
d
A
dx
d
A
w
e






Γ
−






Γ
φ
φ
Rearrange the above eqn. in the following form :
dx
dx w
e 



Resulting algebraic equations
Comparison of the numerical solution FVM for 2-d Diffusion problems
When the above governing
equation is integrated over the
CV, we obtain
Flux through the CV faces
Substitute above expressions and rearrange NOW!
Formulation of algebraic expressions Four Basic Rules in FVM
 Rule 1  Flux consistency across the control volume faces
in
exit q
q =
0
,
, 
W
E
P a
a
a
 Rule 2  Positive coefficients
0
,
, 
W
E
P a
a
a
P
P
u S
S
S φ
+
=
v
S
a
a P
neighbors
P ∆
−
= ∑
 Rule 4  Sum of neighboring coefficients :
 Rule 3  Negative slope linearization of the source term.
Outline
 FVM
1-D convection-diffusion equation

Steady 1-D convection-diffusion equation
Integrating the transport and continuity equation,
Steady 1-D convection-diffusion equation
Introduce the following variables,
1-d CDE : central differencing
F/D ratio : 1.25
1-d CDE : central differencing
F/D ratio : 5
1-d CDE : upwind differencing
F/D ratio : 5
QUICK scheme  2 u/s GP + 1 d/s GP
A 2-D test case 2-D test case : Upwind differencing 2-D test case : QUICK
TVD schemes
 TVD : Total Variation Diminishing
 TVD is specially formulated to achieve oscillation-
free solutions.
 Upwind  most stable. But, introduces high level of
false diffusion.
 CDE, QUICK  Spurious oscillations or wiggles,
 CDE, QUICK  Spurious oscillations or wiggles,
when Peclet number is high
 In turbulent flows, wiggles can give rise to physically
unrealistic negative values and instability.
 In TVD, the tendency towards oscillation is
counteracted by adding an artificial diffusion
fragment or weighting towards upstream
contribution.
The END

More Related Content

Similar to lec_slides.pdf

On Approach to Increase Integration Rate of Elements of a Switched-capacitor ...
On Approach to Increase Integration Rate of Elements of a Switched-capacitor ...On Approach to Increase Integration Rate of Elements of a Switched-capacitor ...
On Approach to Increase Integration Rate of Elements of a Switched-capacitor ...BRNSS Publication Hub
 
Week 8 [compatibility mode]
Week 8 [compatibility mode]Week 8 [compatibility mode]
Week 8 [compatibility mode]Hazrul156
 
formulation of first order linear and nonlinear 2nd order differential equation
formulation of first order linear and nonlinear 2nd order differential equationformulation of first order linear and nonlinear 2nd order differential equation
formulation of first order linear and nonlinear 2nd order differential equationMahaswari Jogia
 
Electromagnetic Scattering from Objects with Thin Coatings.2016.05.04.02
Electromagnetic Scattering from Objects with Thin Coatings.2016.05.04.02Electromagnetic Scattering from Objects with Thin Coatings.2016.05.04.02
Electromagnetic Scattering from Objects with Thin Coatings.2016.05.04.02Luke Underwood
 
differentiol equation.pptx
differentiol equation.pptxdifferentiol equation.pptx
differentiol equation.pptxPlanningHCEGC
 
On Optimization of Manufacturing of Field-effect Transistors to Increase Thei...
On Optimization of Manufacturing of Field-effect Transistors to Increase Thei...On Optimization of Manufacturing of Field-effect Transistors to Increase Thei...
On Optimization of Manufacturing of Field-effect Transistors to Increase Thei...BRNSSPublicationHubI
 
An Approach to Analyze Non-linear Dynamics of Mass Transport during Manufactu...
An Approach to Analyze Non-linear Dynamics of Mass Transport during Manufactu...An Approach to Analyze Non-linear Dynamics of Mass Transport during Manufactu...
An Approach to Analyze Non-linear Dynamics of Mass Transport during Manufactu...BRNSS Publication Hub
 
Numerical Solution of Diffusion Equation by Finite Difference Method
Numerical Solution of Diffusion Equation by Finite Difference MethodNumerical Solution of Diffusion Equation by Finite Difference Method
Numerical Solution of Diffusion Equation by Finite Difference Methodiosrjce
 
Mit2 092 f09_lec20
Mit2 092 f09_lec20Mit2 092 f09_lec20
Mit2 092 f09_lec20Rahman Hakim
 
First order linear differential equation
First order linear differential equationFirst order linear differential equation
First order linear differential equationNofal Umair
 
Existence results for fractional q-differential equations with integral and m...
Existence results for fractional q-differential equations with integral and m...Existence results for fractional q-differential equations with integral and m...
Existence results for fractional q-differential equations with integral and m...IJRTEMJOURNAL
 
Existence results for fractional q-differential equations with integral and m...
Existence results for fractional q-differential equations with integral and m...Existence results for fractional q-differential equations with integral and m...
Existence results for fractional q-differential equations with integral and m...journal ijrtem
 
Quantum assignment
Quantum assignmentQuantum assignment
Quantum assignmentViraj Dande
 
Calculus of variations
Calculus of variationsCalculus of variations
Calculus of variationsSolo Hermelin
 
APPLICATION OF HIGHER ORDER DIFFERENTIAL EQUATIONS
APPLICATION OF HIGHER ORDER DIFFERENTIAL EQUATIONSAPPLICATION OF HIGHER ORDER DIFFERENTIAL EQUATIONS
APPLICATION OF HIGHER ORDER DIFFERENTIAL EQUATIONSAYESHA JAVED
 
article_imen_ridha_2016_version_finale
article_imen_ridha_2016_version_finalearticle_imen_ridha_2016_version_finale
article_imen_ridha_2016_version_finaleMdimagh Ridha
 

Similar to lec_slides.pdf (20)

On Approach to Increase Integration Rate of Elements of a Switched-capacitor ...
On Approach to Increase Integration Rate of Elements of a Switched-capacitor ...On Approach to Increase Integration Rate of Elements of a Switched-capacitor ...
On Approach to Increase Integration Rate of Elements of a Switched-capacitor ...
 
presentation
presentationpresentation
presentation
 
Week 8 [compatibility mode]
Week 8 [compatibility mode]Week 8 [compatibility mode]
Week 8 [compatibility mode]
 
formulation of first order linear and nonlinear 2nd order differential equation
formulation of first order linear and nonlinear 2nd order differential equationformulation of first order linear and nonlinear 2nd order differential equation
formulation of first order linear and nonlinear 2nd order differential equation
 
Electromagnetic Scattering from Objects with Thin Coatings.2016.05.04.02
Electromagnetic Scattering from Objects with Thin Coatings.2016.05.04.02Electromagnetic Scattering from Objects with Thin Coatings.2016.05.04.02
Electromagnetic Scattering from Objects with Thin Coatings.2016.05.04.02
 
differentiol equation.pptx
differentiol equation.pptxdifferentiol equation.pptx
differentiol equation.pptx
 
On Optimization of Manufacturing of Field-effect Transistors to Increase Thei...
On Optimization of Manufacturing of Field-effect Transistors to Increase Thei...On Optimization of Manufacturing of Field-effect Transistors to Increase Thei...
On Optimization of Manufacturing of Field-effect Transistors to Increase Thei...
 
UNIT-III.pdf
UNIT-III.pdfUNIT-III.pdf
UNIT-III.pdf
 
An Approach to Analyze Non-linear Dynamics of Mass Transport during Manufactu...
An Approach to Analyze Non-linear Dynamics of Mass Transport during Manufactu...An Approach to Analyze Non-linear Dynamics of Mass Transport during Manufactu...
An Approach to Analyze Non-linear Dynamics of Mass Transport during Manufactu...
 
Numerical Solution of Diffusion Equation by Finite Difference Method
Numerical Solution of Diffusion Equation by Finite Difference MethodNumerical Solution of Diffusion Equation by Finite Difference Method
Numerical Solution of Diffusion Equation by Finite Difference Method
 
01_AJMS_195_19_RA.pdf
01_AJMS_195_19_RA.pdf01_AJMS_195_19_RA.pdf
01_AJMS_195_19_RA.pdf
 
01_AJMS_195_19_RA.pdf
01_AJMS_195_19_RA.pdf01_AJMS_195_19_RA.pdf
01_AJMS_195_19_RA.pdf
 
Mit2 092 f09_lec20
Mit2 092 f09_lec20Mit2 092 f09_lec20
Mit2 092 f09_lec20
 
First order linear differential equation
First order linear differential equationFirst order linear differential equation
First order linear differential equation
 
Existence results for fractional q-differential equations with integral and m...
Existence results for fractional q-differential equations with integral and m...Existence results for fractional q-differential equations with integral and m...
Existence results for fractional q-differential equations with integral and m...
 
Existence results for fractional q-differential equations with integral and m...
Existence results for fractional q-differential equations with integral and m...Existence results for fractional q-differential equations with integral and m...
Existence results for fractional q-differential equations with integral and m...
 
Quantum assignment
Quantum assignmentQuantum assignment
Quantum assignment
 
Calculus of variations
Calculus of variationsCalculus of variations
Calculus of variations
 
APPLICATION OF HIGHER ORDER DIFFERENTIAL EQUATIONS
APPLICATION OF HIGHER ORDER DIFFERENTIAL EQUATIONSAPPLICATION OF HIGHER ORDER DIFFERENTIAL EQUATIONS
APPLICATION OF HIGHER ORDER DIFFERENTIAL EQUATIONS
 
article_imen_ridha_2016_version_finale
article_imen_ridha_2016_version_finalearticle_imen_ridha_2016_version_finale
article_imen_ridha_2016_version_finale
 

More from MalluKomar

Fluid Mechanics.pptx study of fluids is very important
Fluid Mechanics.pptx study of fluids is very importantFluid Mechanics.pptx study of fluids is very important
Fluid Mechanics.pptx study of fluids is very importantMalluKomar
 
GEOTHERMAL ENERGY.pptx type renewable energy source
GEOTHERMAL ENERGY.pptx type renewable energy sourceGEOTHERMAL ENERGY.pptx type renewable energy source
GEOTHERMAL ENERGY.pptx type renewable energy sourceMalluKomar
 
Hydrodynamic cavitaion on system for agr
Hydrodynamic cavitaion on system for agrHydrodynamic cavitaion on system for agr
Hydrodynamic cavitaion on system for agrMalluKomar
 
Dimensional-Analysis-Mr-Raman-Gahlaut.pdf
Dimensional-Analysis-Mr-Raman-Gahlaut.pdfDimensional-Analysis-Mr-Raman-Gahlaut.pdf
Dimensional-Analysis-Mr-Raman-Gahlaut.pdfMalluKomar
 
material selection.pptx
material selection.pptxmaterial selection.pptx
material selection.pptxMalluKomar
 
material selection.pptx
material selection.pptxmaterial selection.pptx
material selection.pptxMalluKomar
 
Surface Coating.pptx
Surface Coating.pptxSurface Coating.pptx
Surface Coating.pptxMalluKomar
 
5-powdermetallurgy-130304041022-phpapp02.pptx
5-powdermetallurgy-130304041022-phpapp02.pptx5-powdermetallurgy-130304041022-phpapp02.pptx
5-powdermetallurgy-130304041022-phpapp02.pptxMalluKomar
 
powder metallurgy.pptx
powder metallurgy.pptxpowder metallurgy.pptx
powder metallurgy.pptxMalluKomar
 
powder metallurgy.pptx
powder metallurgy.pptxpowder metallurgy.pptx
powder metallurgy.pptxMalluKomar
 
Surface Coating.pptx
Surface Coating.pptxSurface Coating.pptx
Surface Coating.pptxMalluKomar
 
Surface Coating.pptx
Surface Coating.pptxSurface Coating.pptx
Surface Coating.pptxMalluKomar
 
vdocument.in_heat-treatment-process.pptx
vdocument.in_heat-treatment-process.pptxvdocument.in_heat-treatment-process.pptx
vdocument.in_heat-treatment-process.pptxMalluKomar
 
What is Python.pptx
What is Python.pptxWhat is Python.pptx
What is Python.pptxMalluKomar
 
Calculus class-notes
Calculus class-notesCalculus class-notes
Calculus class-notesMalluKomar
 
Gate 2022 brochure
Gate 2022 brochureGate 2022 brochure
Gate 2022 brochureMalluKomar
 

More from MalluKomar (17)

Fluid Mechanics.pptx study of fluids is very important
Fluid Mechanics.pptx study of fluids is very importantFluid Mechanics.pptx study of fluids is very important
Fluid Mechanics.pptx study of fluids is very important
 
GEOTHERMAL ENERGY.pptx type renewable energy source
GEOTHERMAL ENERGY.pptx type renewable energy sourceGEOTHERMAL ENERGY.pptx type renewable energy source
GEOTHERMAL ENERGY.pptx type renewable energy source
 
Hydrodynamic cavitaion on system for agr
Hydrodynamic cavitaion on system for agrHydrodynamic cavitaion on system for agr
Hydrodynamic cavitaion on system for agr
 
Dimensional-Analysis-Mr-Raman-Gahlaut.pdf
Dimensional-Analysis-Mr-Raman-Gahlaut.pdfDimensional-Analysis-Mr-Raman-Gahlaut.pdf
Dimensional-Analysis-Mr-Raman-Gahlaut.pdf
 
MODULE-3.pptx
MODULE-3.pptxMODULE-3.pptx
MODULE-3.pptx
 
material selection.pptx
material selection.pptxmaterial selection.pptx
material selection.pptx
 
material selection.pptx
material selection.pptxmaterial selection.pptx
material selection.pptx
 
Surface Coating.pptx
Surface Coating.pptxSurface Coating.pptx
Surface Coating.pptx
 
5-powdermetallurgy-130304041022-phpapp02.pptx
5-powdermetallurgy-130304041022-phpapp02.pptx5-powdermetallurgy-130304041022-phpapp02.pptx
5-powdermetallurgy-130304041022-phpapp02.pptx
 
powder metallurgy.pptx
powder metallurgy.pptxpowder metallurgy.pptx
powder metallurgy.pptx
 
powder metallurgy.pptx
powder metallurgy.pptxpowder metallurgy.pptx
powder metallurgy.pptx
 
Surface Coating.pptx
Surface Coating.pptxSurface Coating.pptx
Surface Coating.pptx
 
Surface Coating.pptx
Surface Coating.pptxSurface Coating.pptx
Surface Coating.pptx
 
vdocument.in_heat-treatment-process.pptx
vdocument.in_heat-treatment-process.pptxvdocument.in_heat-treatment-process.pptx
vdocument.in_heat-treatment-process.pptx
 
What is Python.pptx
What is Python.pptxWhat is Python.pptx
What is Python.pptx
 
Calculus class-notes
Calculus class-notesCalculus class-notes
Calculus class-notes
 
Gate 2022 brochure
Gate 2022 brochureGate 2022 brochure
Gate 2022 brochure
 

Recently uploaded

CCS355 Neural Network & Deep Learning UNIT III notes and Question bank .pdf
CCS355 Neural Network & Deep Learning UNIT III notes and Question bank .pdfCCS355 Neural Network & Deep Learning UNIT III notes and Question bank .pdf
CCS355 Neural Network & Deep Learning UNIT III notes and Question bank .pdfAsst.prof M.Gokilavani
 
HARMONY IN THE HUMAN BEING - Unit-II UHV-2
HARMONY IN THE HUMAN BEING - Unit-II UHV-2HARMONY IN THE HUMAN BEING - Unit-II UHV-2
HARMONY IN THE HUMAN BEING - Unit-II UHV-2RajaP95
 
HARMONY IN THE NATURE AND EXISTENCE - Unit-IV
HARMONY IN THE NATURE AND EXISTENCE - Unit-IVHARMONY IN THE NATURE AND EXISTENCE - Unit-IV
HARMONY IN THE NATURE AND EXISTENCE - Unit-IVRajaP95
 
ZXCTN 5804 / ZTE PTN / ZTE POTN / ZTE 5804 PTN / ZTE POTN 5804 ( 100/200 GE Z...
ZXCTN 5804 / ZTE PTN / ZTE POTN / ZTE 5804 PTN / ZTE POTN 5804 ( 100/200 GE Z...ZXCTN 5804 / ZTE PTN / ZTE POTN / ZTE 5804 PTN / ZTE POTN 5804 ( 100/200 GE Z...
ZXCTN 5804 / ZTE PTN / ZTE POTN / ZTE 5804 PTN / ZTE POTN 5804 ( 100/200 GE Z...ZTE
 
College Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service NashikCollege Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service NashikCall Girls in Nagpur High Profile
 
IVE Industry Focused Event - Defence Sector 2024
IVE Industry Focused Event - Defence Sector 2024IVE Industry Focused Event - Defence Sector 2024
IVE Industry Focused Event - Defence Sector 2024Mark Billinghurst
 
Sachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
Sachpazis Costas: Geotechnical Engineering: A student's Perspective IntroductionSachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
Sachpazis Costas: Geotechnical Engineering: A student's Perspective IntroductionDr.Costas Sachpazis
 
Model Call Girl in Narela Delhi reach out to us at 🔝8264348440🔝
Model Call Girl in Narela Delhi reach out to us at 🔝8264348440🔝Model Call Girl in Narela Delhi reach out to us at 🔝8264348440🔝
Model Call Girl in Narela Delhi reach out to us at 🔝8264348440🔝soniya singh
 
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...VICTOR MAESTRE RAMIREZ
 
Internship report on mechanical engineering
Internship report on mechanical engineeringInternship report on mechanical engineering
Internship report on mechanical engineeringmalavadedarshan25
 
Past, Present and Future of Generative AI
Past, Present and Future of Generative AIPast, Present and Future of Generative AI
Past, Present and Future of Generative AIabhishek36461
 
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130Suhani Kapoor
 
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur EscortsCall Girls in Nagpur High Profile
 
SPICE PARK APR2024 ( 6,793 SPICE Models )
SPICE PARK APR2024 ( 6,793 SPICE Models )SPICE PARK APR2024 ( 6,793 SPICE Models )
SPICE PARK APR2024 ( 6,793 SPICE Models )Tsuyoshi Horigome
 
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Dr.Costas Sachpazis
 
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escortsranjana rawat
 
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...ranjana rawat
 
(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts
(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts
(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escortsranjana rawat
 

Recently uploaded (20)

★ CALL US 9953330565 ( HOT Young Call Girls In Badarpur delhi NCR
★ CALL US 9953330565 ( HOT Young Call Girls In Badarpur delhi NCR★ CALL US 9953330565 ( HOT Young Call Girls In Badarpur delhi NCR
★ CALL US 9953330565 ( HOT Young Call Girls In Badarpur delhi NCR
 
CCS355 Neural Network & Deep Learning UNIT III notes and Question bank .pdf
CCS355 Neural Network & Deep Learning UNIT III notes and Question bank .pdfCCS355 Neural Network & Deep Learning UNIT III notes and Question bank .pdf
CCS355 Neural Network & Deep Learning UNIT III notes and Question bank .pdf
 
HARMONY IN THE HUMAN BEING - Unit-II UHV-2
HARMONY IN THE HUMAN BEING - Unit-II UHV-2HARMONY IN THE HUMAN BEING - Unit-II UHV-2
HARMONY IN THE HUMAN BEING - Unit-II UHV-2
 
HARMONY IN THE NATURE AND EXISTENCE - Unit-IV
HARMONY IN THE NATURE AND EXISTENCE - Unit-IVHARMONY IN THE NATURE AND EXISTENCE - Unit-IV
HARMONY IN THE NATURE AND EXISTENCE - Unit-IV
 
ZXCTN 5804 / ZTE PTN / ZTE POTN / ZTE 5804 PTN / ZTE POTN 5804 ( 100/200 GE Z...
ZXCTN 5804 / ZTE PTN / ZTE POTN / ZTE 5804 PTN / ZTE POTN 5804 ( 100/200 GE Z...ZXCTN 5804 / ZTE PTN / ZTE POTN / ZTE 5804 PTN / ZTE POTN 5804 ( 100/200 GE Z...
ZXCTN 5804 / ZTE PTN / ZTE POTN / ZTE 5804 PTN / ZTE POTN 5804 ( 100/200 GE Z...
 
College Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service NashikCollege Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
College Call Girls Nashik Nehal 7001305949 Independent Escort Service Nashik
 
Exploring_Network_Security_with_JA3_by_Rakesh Seal.pptx
Exploring_Network_Security_with_JA3_by_Rakesh Seal.pptxExploring_Network_Security_with_JA3_by_Rakesh Seal.pptx
Exploring_Network_Security_with_JA3_by_Rakesh Seal.pptx
 
IVE Industry Focused Event - Defence Sector 2024
IVE Industry Focused Event - Defence Sector 2024IVE Industry Focused Event - Defence Sector 2024
IVE Industry Focused Event - Defence Sector 2024
 
Sachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
Sachpazis Costas: Geotechnical Engineering: A student's Perspective IntroductionSachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
Sachpazis Costas: Geotechnical Engineering: A student's Perspective Introduction
 
Model Call Girl in Narela Delhi reach out to us at 🔝8264348440🔝
Model Call Girl in Narela Delhi reach out to us at 🔝8264348440🔝Model Call Girl in Narela Delhi reach out to us at 🔝8264348440🔝
Model Call Girl in Narela Delhi reach out to us at 🔝8264348440🔝
 
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
VICTOR MAESTRE RAMIREZ - Planetary Defender on NASA's Double Asteroid Redirec...
 
Internship report on mechanical engineering
Internship report on mechanical engineeringInternship report on mechanical engineering
Internship report on mechanical engineering
 
Past, Present and Future of Generative AI
Past, Present and Future of Generative AIPast, Present and Future of Generative AI
Past, Present and Future of Generative AI
 
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
VIP Call Girls Service Hitech City Hyderabad Call +91-8250192130
 
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Meera Call 7001035870 Meet With Nagpur Escorts
 
SPICE PARK APR2024 ( 6,793 SPICE Models )
SPICE PARK APR2024 ( 6,793 SPICE Models )SPICE PARK APR2024 ( 6,793 SPICE Models )
SPICE PARK APR2024 ( 6,793 SPICE Models )
 
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
Sheet Pile Wall Design and Construction: A Practical Guide for Civil Engineer...
 
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur EscortsHigh Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
High Profile Call Girls Nagpur Isha Call 7001035870 Meet With Nagpur Escorts
 
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
(ANVI) Koregaon Park Call Girls Just Call 7001035870 [ Cash on Delivery ] Pun...
 
(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts
(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts
(MEERA) Dapodi Call Girls Just Call 7001035870 [ Cash on Delivery ] Pune Escorts
 

lec_slides.pdf

  • 1. Foundations of CFD (AM5630) Outline What is CFD ? Steps in CFD Steps in CFD Governing Equations Tensor notation CFD books Richtmyer R D and Morton K W, Difference methods for initial value problems, Inter science publishers, 1967. Chung T J, Computational Fluid Dynamics, CUP, 2002 Anderson J, Essential Computational Fluid Dynamics Joel H. Ferziger and Milovan Peric Computational Methods Joel H. Ferziger and Milovan Peric Computational Methods for Fluid Dynamics John C. Tannehill, Dale Anderson Richard Pletcher, Computational Fluid Mechanics and Heat Transfer, 2nd Ed. Steps in CFD Identify the design/ optimization/ failure analysis problem Identify a suitable Mathematical model Choose the appropriate flow domain of interest. Suitable PDE’s, IC’s and BC’s. Suitable PDE’s, IC’s and BC’s. Discretize the region of interest. Develop the system of algebraic equations. Solve the system of equations. Perform post-processing. Develop an engineering solution. Continuum Hypothesis V m V V δ δ ρ δ δ * lim → = Air at STP: δV*=10-9 mm3, which contains 3x107 molecules which is sufficient to define a nearly constant density. Forces on a blob of fluid force pressure Net on Accelerati Mass . * = v δ V r force pressure Net on Accelerati Mass . * = g x p Dt D j ji r r ρ τ ρ + ∂ ∂ + ∇ − = ) ( V g v forces viscous dS p Dt D v S r r ) ( ) ( V ) ( δ ρ δ ρ + + − = ∫ forces viscous + g v forces viscous v p Dt D v r r ) ( ) ( V ) ( δ ρ δ δ ρ + + ∇ − = forces Body + Equations of motion         ∂ ∂ + ∂ ∂ + ∂ ∂ + ∂ ∂ − = ∂ ∂ + ∂ ∂ + ∂ ∂ + ∂ ∂ 2 2 2 2 2 2 1 z u y u x u x p z u w y u v x u u t u ν ρ         ∂ ∂ + ∂ ∂ + ∂ ∂ + ∂ ∂ − = ∂ ∂ + ∂ ∂ + ∂ ∂ + ∂ ∂ 2 2 2 2 2 2 1 z v y v x v y p z v w y v v x v u t v ν ρ       ∂ + ∂ + ∂ + ∂ ∂ − = ∂ ∂ + ∂ ∂ + ∂ ∂ + ∂ ∂ 2 2 2 1 w w w p w w w v w u w ν ρ j j i i j i j i x x u x p x u u t u ∂ ∂ ∂ + ∂ ∂ − = ∂ ∂ + ∂ ∂ 2 1 ν ρ 0 = ∂ ∂ i i x u       ∂ + ∂ + ∂ + ∂ − = ∂ + ∂ + ∂ + ∂ 2 2 2 z y x z z w y v x u t ν ρ 0 = ∂ ∂ + ∂ ∂ + ∂ ∂ z w y v x u Some times called suffix notation. Much of the notation can be introduced in the context of vectors with so called Summation convention The rule is, if a suffix is repeated in any term, there is an implied summation Tensor Notation j i j x u u ∂ ∂ 3 2 1 x u u x u u x u u i i i ∂ ∂ + ∂ ∂ + ∂ ∂ Range convention: If a suffix occurs just once in a term, then it is assumed to take all of the values 1,2,3 in turn. j is called dummy index – indicating an implied summation. i is called free index – indicating the eqn. above represents 3 components of a vector eqn. 3 3 2 2 1 1 x u x u x u ∂ + ∂ + ∂ Tensor Notation (ctd…) ) / ( ) . ( ρ p t −∇ = ∇ + ∂ ∂ u u u         ∂ ∂ − = ∂ ∂ + ∂ ∂ ρ p x x u u t u i j i j i 0 u = ∇. i i x u ∂ ∂         ∂ ∂ ∂ +         ∂ ∂ − = ∂ ∂ + ∂ ∂ j j i i j i j i x x u p x x u u t u 2 ν ρ u u u u 2 ) / ( ) . ( ∇ + −∇ = ∇ + ∂ ∂ ν ρ p t Disturbance propagation in a flow Subsonic Sonic Supersonic 0 ) 1 ( 2 2 2 2 2 = ∂ ∂ + ∂ ∂ − y x M φ φ 1 M 1 = M 1 M
  • 2. PDE - classification 0 2 2 2 2 2 = + + ∂ ∂ + ∂ ∂ + ∂ ∂ + ∂ ∂ ∂ + ∂ ∂ G F y E x D y C y x B x A φ φ φ φ φ φ 0 4 2 − AC B 0 4 2 = − AC B Elliptic Parabolic 0 4 = − AC B 0 4 2 − AC B Parabolic Hyperbolic Elliptic PDE Can be solved by specifying BC’s on a complete contour enclosing the region. It is a BVP At all points in the domain Ω Γ 0 4 2 − AC B An elliptic PDE has no real characteristic curves. An elliptic PDE has no real characteristic curves. A disturbance is propagated instantly in all directions in the region. 0 : 2 2 2 2 = ∂ ∂ + ∂ ∂ y x Examples φ φ ) , ( 2 2 2 2 y x f y x = ∂ ∂ + ∂ ∂ φ φ Parabolic PDE BC’s may be closed on one direction. But, remain open at one end of the other direction. It is a mixed BVP At all points in the domain Ω Γ 0 4 2 = − AC B For a parabolic PDE there exists one char. line. For a parabolic PDE there exists one char. line. Typically, an initial distribution of dependent variable and two sets of BC’s are required for a complete description of the problem. 2 2 : x T t T Examples ∂ ∂ = ∂ ∂ α 2 2 x u t u ∂ ∂ = ∂ ∂ ν Hyperbolic PDE The eqns. can be solved by specifying the conditions only at a portion of the boundary, the other boundaries remain open. It is an IVP At all points in the domain Ω 0 4 2 − AC B For a Hyperbolic PDE, two P A Γ For a Hyperbolic PDE, two real char. exists per piont. Typically, we need to specify conditions at one part of the boundary in order to determine the solution in a given region. 2 2 2 2 2 : x a t Examples ∂ ∂ = ∂ ∂ φ φ A B C Tensor Notation (ctd…) ( ) j k ijk i i x u ∂ ∂ = × ∇ = ε ω u u × ∇ ( ) k j ikj i i x u ∂ ∂ = × ∇ = ε ω u ( )         ∂ ∂ − ∂ ∂ = × ∇ = k j j k ijk i i x u x u ε ω 2 1 u ? 3 2 1 ω ω ω is what k i ik x u EXPAND ∂ ∂ τ j i i j j i x u x u x u EXPAND ∂ ∂         ∂ ∂ + ∂ ∂ Outline PDE and their classification Understanding characteristics When do you call a PDE is well-posed ? The problem in fact has a solution The solution is unique The solution is unique The solution depends continuously on the data given in the problem. PDE - classification 0 2 2 2 2 2 = + + ∂ ∂ + ∂ ∂ + ∂ ∂ + ∂ ∂ ∂ + ∂ ∂ G F y E x D y C y x B x A φ φ φ φ φ φ Assume ϕ = ϕ(x,y) is a solution of the differential equation. By definition, the 2nd order derivatives along the characteristic curves are indeterminate and indeed, they characteristic curves are indeterminate and indeed, they may be discontinuous across the characteristics. Thus differentials of ϕx and ϕy is which represent changes from location (x,y) to (x+dx, y+dy) across the characteristics may be expressed as, However, no discontinuity of the first derivatives is allowed i.e., they are continuous functions of x and y. Assume ϕ = ϕ(x,y) is a solution of the differential equation. PDE - classification ∂ ∂ ∂ 2 2 2 φ φ φ dy y dx x d x x x ∂ ∂ + ∂ ∂ = φ φ φ dy y x dx x ∂ ∂ ∂ + ∂ ∂ = φ φ 2 2 2 dy y dx x d y y y ∂ ∂ + ∂ ∂ = φ φ φ dy y dx y x 2 2 2 ∂ ∂ + ∂ ∂ ∂ = φ φ H y C y x B x A = ∂ ∂ + ∂ ∂ ∂ + ∂ ∂ 2 2 2 2 2 φ φ φ         + + ∂ ∂ + ∂ ∂ − = G F y E x D H Where φ φ φ The three eqns. above can be used to solve for the 2nd order derivatives of ϕ Using Cramer’s rule,
  • 3. 2nd order PDE’s dy dx dy dx C B A dy d d dx C H A y x y x 0 0 0 0 2 φ φ φ = ∂ ∂ ∂ Since it is possible to have discontinuities in the 2nd order derivatives of the dependent variable across the characteristics, these derivatives are indeterminate. Thus, setting the denominator equal to zero, 0 0 0 = dy dx dy dx C B A Solving this quadratic yields the equations of the characteristics in physical space. This yields a quadratic equation. 0 2 = +       −       C dx dy B dx dy A ? , =    β α dx dy Do it now ! 0 ) 1 ( 2 2 2 = ∂ + ∂ − M φ φ Classify the following equations : 0 1 ; 0 2 2 ≈ ∂ ∂ ∂ ∂ + ∂ ∂ − = ∂ ∂ + ∂ ∂ = ∂ ∂ + ∂ ∂ y p y u x p y u v x u u y v x u ν ρ 0 2 2 2 2 2 = + + ∂ ∂ + ∂ ∂ + ∂ ∂ + ∂ ∂ ∂ + ∂ ∂ G F y E x D y C y x B x A φ φ φ φ φ φ 0 ) 1 ( 2 2 = ∂ + ∂ − y x M Overview Boundary conditions are a required component of the mathematical model. Boundaries direct motion of flow. Specify fluxes into the computational domain, e.g. mass, momentum, and energy. mass, momentum, and energy. Fluid and solid regions are represented by cell zones. Material and source terms are assigned to cell zones. Boundaries and internal surfaces are represented by face zones. Boundary data are assigned to face zones. Boundary conditions When solving the Navier- Stokes equation and continuity equation, appropriate initial conditions and boundary conditions need to be applied. conditions need to be applied. In the example here, a no-slip boundary condition is applied at the solid wall. Neumann and Dirichlet boundary conditions When using a Dirichlet boundary condition, one prescribes the value of a variable at the boundary, e.g. u(x) = constant. When using a Neumann boundary condition, one When using a Neumann boundary condition, one prescribes the gradient normal to the boundary of a variable at the boundary, e.g. ∂nu(x) = constant. When using a mixed boundary condition a function of the form au(x)+b∂nu(x) = constant is applied. Note that at a given boundary, different types of boundary conditions can be used for different variables. Outline Discretization Finite Differences 0 2 2 2 2 2 = + + ∂ ∂ + ∂ ∂ + ∂ ∂ + ∂ ∂ ∂ + ∂ ∂ G F y E x D y C y x B x A φ φ φ φ φ φ Discretization 1 i i-1 i+1 N i i-1 i+1 j+1 i,j j j -1 Grid generation : Structured (vs) Unstructured Taylor Series ) ( ) ( i x x φ φ = i i x x x         ∂ ∂ − + 3 3 3 ! 3 ) ( φ H x n x x i n n n i +         ∂ ∂ − + φ ! ) ( K i i x x x       ∂ ∂ − + φ ) ( i i x x x         ∂ ∂ − + 2 2 2 ! 2 ) ( φ Any continuous differentiable function (ϕ), in the vicinity of xi , can be expressed as a Taylor series: i i i i i x x x x x       ∂ ∂ − + ≈ + + φ φ φ ) ( ) ( ) ( 1 1 i i i i i i i i x x x x x x x x         ∂ ∂ − +       ∂ ∂ − + ≈ + + + 2 2 2 1 1 1 ! 2 ) ( ) ( ) ( ) ( φ φ φ φ i n n n i i n x n x x TE         ∂ ∂ − + ∞ = ∑ φ ! ) ( : 1 3 i n n n i i n x n x x TE         ∂ ∂ − + ∞ = ∑ φ ! ) ( : 1 2
  • 4. Do it now ! i i i i i x x x x x       ∂ ∂ − + ≈ + + φ φ φ ) ( ) ( ) ( 1 1 i i i i i x x x x x       ∂ ∂ − + ≈ − − φ φ φ ) ( ) ( ) ( 1 1 ) ( ) ( ) ( ) ( 1 1 x O x x x x x i i i i i ∆ + − − =       ∂ ∂ + + φ φ φ ) ( ) ( ) ( ) ( 1 1 x O x x x x x i i i i i ∆ + − − =       ∂ ∂ − − φ φ φ ) ( ) ( ) ( ) ( 1 x O x x x x i i i ∆ + ∆ − =       ∂ ∂ − φ φ φ 2 ) ( ? x O x i ∆ + =       ∂ ∂φ i i i i i x x x x x x x x         ∂ ∂ − +       ∂ ∂ − + ≈ 2 2 2 ! 2 ) ( ) ( ) ( ) ( φ φ φ φ Derivative and its approximation Validation and Verification Are we solving the right equations ? (vs) Are we solving the equations right ? Consistency Consistency 0 → TE FDE PDE TE − = Do it now ! i i i i x x x x x x x         ∂ ∂ ∆ +       ∂ ∂ ∆ + ≈ ∆ + 2 2 2 ! 2 ) ( ) ( ) ( ) ( φ φ φ φ ) ( ? 2 2 x O x i ∆ + =         ∂ ∂ φ ) ( ) ( 2 2 1 2 2 2 x O x x i i i i ∆ + ∆ + − =         ∂ ∂ + + φ φ φ φ 2 2 2 ) ( ? x O x i ∆ + =         ∂ ∂ φ i i i i x x x x x x x         ∂ ∂ ∆ +       ∂ ∂ ∆ + ≈ ∆ + 2 2 2 ! 2 ) 2 ( ) 2 ( ) ( ) 2 ( φ φ φ φ Outline Discretization Finite Differences Explicit (vs) Implicit Do it now ! i i i i x x x x x x x         ∂ ∂ ∆ +       ∂ ∂ ∆ + ≈ ∆ + 2 2 2 ! 2 ) ( ) ( ) ( ) ( φ φ φ φ ) ( ? 2 2 x O x i ∆ + =         ∂ ∂ φ i i i i x x x x x x x         ∂ ∂ ∆ +       ∂ ∂ ∆ + ≈ ∆ + 2 2 2 ! 2 ) 2 ( ) 2 ( ) ( ) 2 ( φ φ φ φ ) ( ) ( 2 ) 1 ( * 2 ) 2 ( 2 1 2 2 2 x O x x Eq Eq i i i i ∆ + ∆ + − =         ∂ ∂ ⇒ − + + φ φ φ φ Do it now ! 2 2 2 ) ( ? x O x i ∆ + =         ∂ ∂ φ i i x x x       ∂ ∂ ∂ ∂ =         ∂ ∂ ) ( 2 2 φ φ x x x i i ∆       ∂ ∂ −       ∂ ∂ ≈ + φ φ 1 i i   x ∆ x x x i i i i ∆       ∆ − −       ∆ − ≈ − + 1 1 φ φ φ φ ( ) ( )2 2 1 1 2 x O x i i i ∆ + ∆ + − = − + φ φ φ [ ] 2 2 2 ) ( , ) ( ? y x O y x i ∆ ∆ + =         ∂ ∂ ∂ φ Do it now ! [ ] 2 2 2 ) ( , ) ( ? y x O y x i ∆ ∆ + =         ∂ ∂ ∂ φ y y j i j i         ∂ ∂ −         ∂ ∂ ≈ − + , 1 , 1 φ φ ( )( ) ( ) ( ) [ ] 2 2 1 , 1 1 , 1 1 , 1 1 , 1 , 4 y x O y x j i j i j i j i ∆ ∆ + ∆ ∆ + − − = − − + − − + + + φ φ φ φ x y y j i j i ∆   ∂   ∂ ≈ − + 2 , 1 , 1 Explicit (vs) Implicit 2 2 x T t T ∂ ∂ = ∂ ∂ α ( ) 2 2 T T T T + − ∂ ( ) ( ) t O t T T t T n i n i ∆ + ∆ − = ∂ ∂ +1 Explicit Implicit ( ) ( )2 2 1 1 2 2 2 x O x T T T x T i i i ∆ + ∆ + − = ∂ ∂ − + ( ) ( )2 2 1 1 2 x O x T T T n i i i ∆ +       ∆ + − − + ( ) ( )2 1 2 1 1 2 x O x T T T n i i i ∆ +       ∆ + − + − +
  • 5. Outline Explicit (vs) Implicit Stability analysis Explicit (vs) Implicit 2 2 x T t T ∂ ∂ = ∂ ∂ α ( ) ( ) t O t T T t T n i n i ∆ + ∆ − = ∂ ∂ +1 1 i i-1 i+1 N n n n T T T   ∂ − + 2 1 n n n x T t O t T T         ∂ ∂ = ∆ + ∆ − + 2 2 1 ) ( α 1 2 2 1 ) ( + +         ∂ ∂ = ∆ + ∆ − n n n x T t O t T T α Explicit Implicit Explicit Scheme ( ) ( ) n i n i n i n i n i T T T x t T T 1 1 2 1 2 − + + + − ∆ ∆ + = α ( ) ( ) ) ( 2 2 2 1 1 1 2 2 t O x O x T T T t T T x T t T n i n i n i n n ∆ + ∆ =         ∆ + − − ∆ − −         ∂ ∂ − ∂ ∂ − + + α α Explicit 1 i i-1 i+1 N Time level n i Time level n+1 i Time level n+1 i-1 i+1 Implicit ( ) ( ) 1 1 1 1 1 2 1 2 + − + + + + + − ∆ ∆ + = n i n i n i n i n i T T T x t T T α Implicit Scheme 1 i i-1 i+1 N Time level n n i n i i n i i n i i d T c T b T a = + + + − + + − 1 1 1 1 1 TDMA i Time level n+1 i-1 i+1 Implicit ( ) ( ) 1 1 1 1 1 2 1 2 + − + + + + + − ∆ ∆ + = n i n i n i n i n i T T T x t T T α Semi-Implicit Scheme 1 i i-1 i+1 N Time level n ( ) ( ) ( )( )        + − − + + − ∆ ∆ + = − + + − + + + + n i n i n i n i n i n i n i n i T T T T T T x t T T 1 1 1 1 1 1 1 2 1 2 1 2 β β α β formulation ( )         ∆ + − = ∆ − − + − + 2 1 1 1 1 2 2 x T T T t T T n i n i n i n n α Richardson’s ( )         ∆ +         + − = ∆ − − − + + − + 2 1 1 1 1 1 1 2 2 2 x T T T T t T T n i n i n i n i n n α Dufort- Frankel Different Schemes ( )         ∆ = ∆ 2 2 x t α Frankel ( ) ( ) ( ) ( )         ∆ + − + ∆ + − = ∆ − − + + − + + + + 2 1 1 2 1 1 1 1 1 1 2 2 2 1 x T T T x T T T t T T n i n i n i n i n i n i n i n i α Crank- Nicolson Explicit The solution algorithm is simple to set up. For a given ∆x, ∆t must be less than a specific limit imposed by stability constraints. This requires many time steps to carry out the calculations over a given interval of time. Implicit Implicit Stability can be maintained over much larger values of ∆t. Which implies, fewer time steps are needed to carry out the calculations over a given interval. Since matrix manipulations are usually required at each time step, the computer time per time step is larger than that of the explicit approach. Discrete perturbation analysis After m steps i-1 i+1 m ε i After m steps 1 i-1 i+1 N m ε m ε − m ε − i ε 1 i i-1 i+1 N Time level n Time level n+1 1 i i-1 i+1 N i After m steps 1 i-1 i+1 N Stability of Explicit schemes 2 2 x T t T ∂ ∂ = ∂ ∂ α 1 i i-1 i+1 N Explicit : FTCS ( )2 1 1 1 2 x T T T t T T n i n i n i n i n i ∆ + − = ∆ − − + + α ( ) x t ∆ ∆ Let us introduce a disturbance ε at and find the influence on the grid points at higher time levels n i T ( ) ( ) ( )2 1 1 1 2 x T T T t T T n i n i n i n i n i ∆ + + − = ∆ + − − + + ε α ε
  • 6. Stability of Explicit schemes ( ) ( ) ( )2 1 1 1 2 x T T T t T T n i n i n i n i n i ∆ + + − = ∆ + − − + + ε α ε i T assume n i ∀ ( ) ( )2 1 0 2 0 x t T n i ∆ + − = ∆ − + ε α ε ( )                 ∆ ∆ − = + 2 1 2 1 x t T n i α ε ( )         ∆ ∆ = 2 x t d let α d T n i 2 1 1 − = + ε 1 2 1 ) ( 1 2 1 − ≥ − ≤ − d or d 1 ≤ ⇒ d Stability of Explicit schemes When the error reaches all the grid points, after many time steps, approximately with the same magnitude, 2 possibilities may be considered : (i) Error at time m, have the same sign (ii) Error at time m, have alternate signs. ( ) ( )2 1 2 x t T m m m m m i ∆ + − = ∆ − + ε ε ε α ε m m i T ε = +1 No stability constraint ; ; ; , 1 1 m m i m m i m m i T T T Let ε ε ε = = = − + ; ; ; , 1 1 m m i m m i m m i T T T Let ε ε ε − = = − = − + Under what conditions will the solution be stable, if the sign of the error alternates ? Stability of Explicit schemes ; ; ; , 1 1 m m i m m i m m i T T T Let ε ε ε − = = − = − + Under what conditions will the solution be stable, if the sign of the error alternates ? ( ) m m m m m i d T ε ε ε ε − − − + = + 2 1 d T m n i 4 1 1 − = + ε ( ) m m i d T ε 4 1 1 − = + Solution will be stable if, 1 4 1 ) ( 1 1 ≤ − ≤ + d or T m n i ε This requirement leads to, ( )2 2 1 x t ∆ ≤ ∆ α 0 ∂ ∂ − = ∂ ∂ a x u a t u 1 i i-1 i+1 N The quantity u is convected along these lines with a constant a. A number of FD approximations can be constructed as follows : u u u u n n n n − − +1 Schemes for Hyperbolic Equations ( ) x u u a t u u n i n i n i n i ∆ − − = ∆ − + + 1 1 Eulers’ FTFS Stability analysis indicates that, the method is unconditionally unstable. ( ) x u u a t u u n i n i n i n i ∆ − − = ∆ − − + + 2 1 1 1 Eulers’ FTCS This explicit formulation is also unconditionally unstable. Schemes for Hyperbolic Equations 0 ∂ ∂ − = ∂ ∂ a x u a t u 1 i i-1 i+1 N ( ) x u u a t u u n i n i n i n i ∆ − − = ∆ − − + 1 1 First order upwind differencing Stability analysis indicates that, this method is stable, when c ≤ 1 Stability analysis indicates that, this method is stable, when c ≤ 1 Use Forward differencing for the spatial derivative if a 0 ( ) ( ) x u u a t u u u n i n i n i n i n i ∆ − − = ∆ + − − + − + + 2 2 1 1 1 1 1 1 The Lax method Stability analysis indicates that, this method is stable, when c ≤ 1. ( ) x u u a t u u n i n i n i n i ∆ − − = ∆ − + + 1 1 Sources of Error 2 2 x T t T ∂ ∂ = ∂ ∂ α 1 i i-1 i+1 N Explicit : FTCS ( )2 1 1 1 2 x T T T t T T n i n i n i n i n i ∆ + − = ∆ − − + + α ( ) x t ∆ ∆ A : Analytical Solution D : Exact solution of the difference equation Discretization error = A - D N : Numerical solution on a digital computer Round-off error = N - D D N Error − = ε : Stability of Explicit schemes ( ) ( ) ( )2 1 1 1 1 1 1 2 x D D D t D D n i n i n i n i n i n i n i n i n i n i ∆ + + + − + = ∆ + − + − − + + + + ε ε ε α ε ε ( ) ( ) 1 1 1 2 D D D D D n i n i n i n i n i + − = − − + + α D : Exact solution of the difference equation; Hence, it exactly satisfies the difference equation. ( ) ( ) ( )2 1 1 2 x D D D t D D i i i i i ∆ + − = ∆ − − + α ( ) ( )2 1 1 1 2 ) 2 ( ) 1 ( x t n i n i n i n i n i ∆ + − = ∆ − ⇒ − − + + ε ε ε α ε ε Error (ε) also satisfies the difference equation; Solution is stable, if 1 1 ≤ + n i n i ε ε Round-off error variation Random variation of ε with x can be analytically expressed as a Fourier series as follows : x ik m m m e A x ∑ = ) ( ε Round-off error variation 3 , 2 , 1 2 =       = m m L km π ∑ = = 2 / 1 ) ( ) , ( N m x ik m m e t A t x ε ∑ = = 2 / 1 ) ( N m x ik m m e A x ε However, we are interested in the variation of ε with time.
  • 7. Stability of Explicit schemes ( )2 ) ( ) ( ) ( 2 x e e e e e e t e e e e x x ik at x ik at x x ik at x ik at x ik t t a m m m m m ∆ + − = ∆ − ∆ − ∆ + ∆ + α Simplify now ! x ik at m e e by divide , Use following identities : 2 ) cos( x ik x ik m m m e e x k ∆ − ∆ + = ∆ 2 ) cos( 1 ) 2 ( sin2 x k x k m m ∆ − = ∆ Use following identities : Outline Elliptic PDE Parabolic PDE (2-d) Iterative Methods 1-D parabolic PDE 2 2 x T t T ∂ ∂ = ∂ ∂ α 1 i i-1 i+1 N Explicit : FTCS ( )2 1 1 1 2 x T T T t T T n i n i n i n i n i ∆ + − = ∆ − − + + α ( ) x t ∆ ∆ ( ) 2 1 2 ≤ ∆ ∆ x t α Parabolic PDE Elliptic PDE 0 2 2 2 2 = ∂ ∂ + ∂ ∂ y T x T 2 2 + − + − T T T T T T ) , ( 2 2 2 2 y x f y T x T = ∂ ∂ + ∂ ∂ ( ) ( ) 0 2 2 2 1 , , 1 , 2 , 1 , , 1 = ∆ + − + ∆ + − − + − + y T T T x T T T j i j i j i j i j i j i ( ) ( ) ( ) 0 2 2 1 , , 1 , 2 2 , 1 , , 1 = + − ∆ ∆ + + − − + − + j i j i j i j i j i j i T T T y x T T T 0 ) 1 ( 2 , 2 1 , 2 1 , 2 , 1 , 1 = + − + + + − + − + j i j i j i j i j i T T T T T β β β Parabolic PDE : FTCS         ∂ ∂ + ∂ ∂ = ∂ ∂ 2 2 2 2 y T x T t T α ( ) ( )         ∆ + − + ∆ + − = ∆ − − + − + + 2 1 , , 1 , 2 , 1 , , 1 , 1 , 2 2 y T T T x T T T t T T n j i n j i n j i n j i n j i n j i n j i n j i α ( ) ( ) 2 1 2 2 ≤         ∆ ∆ + ∆ ∆ y t x t α α , , 1 y x if ∆ = ∆ = α ( ) 4 1 2 ≤         ∆ ∆ ⇒ x t ( ) ( ) ( ) ( ) n j i n j i n j i n j i n j i n j i n j i n j i T T T y t T T T x t T T , 1 , , 1 2 , 1 , , 1 2 , 1 , 2 2 − + − + + + − ∆ ∆ + + − ∆ ∆ + = α α Stability analysis indicates, the method is stable if, Elliptic PDE 0 2 2 2 2 = ∂ ∂ + ∂ ∂ y T x T ( ) ( ) 0 2 2 2 1 , , 1 , 2 , 1 , , 1 = ∆ + − + ∆ + − − + − + y T T T x T T T j i j i j i j i j i j i ) , ( 2 2 2 2 y x f y T x T = ∂ ∂ + ∂ ∂ ( ) ( ) ∆ ∆ y x ( ) ( ) ( ) 0 2 2 1 , , 1 , 2 2 , 1 , , 1 = + − ∆ ∆ + + − − + − + j i j i j i j i j i j i T T T y x T T T 0 ) 1 ( 2 , 2 1 , 2 1 , 2 , 1 , 1 = + − + + + − + − + j i j i j i j i j i T T T T T β β β ( ) ( ) ) 1 ( 2 2 2 2 2 β α β + − = ∆ ∆ = y x Let
  • 8. 0 , 1 , 2 1 , 2 , 1 , 1 = + + + + − + − + j i j i j i j i j i T T T T T α β β ( ) k j i k j i k j i k j i k j i T T T T T 1 , 2 1 , 2 , 1 , 1 2 1 , ) 1 ( 2 1 − + − + + + + + + = β β β 3 T Jacobi iterative method 1 T 2 T 4 T The Analogy between iterative methods ( ) ( ) ( ) ( ) n j i n j i n j i n j i n j i n j i n j i n j i T T T y t T T T x t T T , 1 , , 1 2 , 1 , , 1 2 , 1 , 2 2 − + − + + + − ∆ ∆ + + − ∆ ∆ + = α α ( ) ⇒ =         ∆ ∆ = 4 1 , 1 2 x t α ( ) n j i n j i n j i n j i n j i n j i n j i T T T T T T T , 1 , 1 , , 1 , 1 , 1 , 4 4 1 − + − + + + + − + + = ( ) k j i k j i k j i k j i k j i T T T T T 1 , 2 1 , 2 , 1 , 1 2 1 , ) 1 ( 2 1 − + − + + + + + + = β β β ( ) n j i n j i n j i n j i n j i T T T T T , 1 , 1 , 1 , 1 1 , 4 1 − + − + + + + + = From the discretization of an Elliptic PDE, ( ) k j i k j i k j i k j i k j i T T T T T 1 , 1 , , 1 , 1 1 , 4 1 1 − + − + + + + + = ⇒ = β Parabolic PDE : Implicit         ∂ ∂ + ∂ ∂ = ∂ ∂ 2 2 2 2 y T x T t T α ( ) ( )         ∆ + − + ∆ + − = ∆ − + − + + + + − + + + + 2 1 1 , 1 , 1 1 , 2 1 , 1 1 , 1 , 1 , 1 , 2 2 y T T T x T T T t T T n j i n j i n j i n j i n j i n j i n j i n j i α N n j i n j i y n j i y n j i y x n j i x n j i x T T d T d T d d T d T d , 1 1 , 1 1 , 1 , 1 , 1 1 , 1 ) 1 2 2 ( − = + + + + − + + − + + + + − + + n j i n j i j i n j i j i n j i j i n j i j i n j i j i f T e T d T c T b T a , 1 1 , , 1 1 , , 1 , , 1 , 1 , 1 , 1 , = + + + + + − + + + + − + + 1 1 , 2 2 , 2 1 2 , 1 2 , 2 2 , 2 1 3 , 2 2 , 2 1 2 , 2 2 , 2 1 2 , 3 2 , 2 + + + + + − − = + + n n n n n n T e T b f T d T c T a In a matrix form Solving a system of Equations n j i n j i j i n j i j i n j i j i n j i j i n j i j i f T e T d T c T b T a , 1 1 , , 1 1 , , 1 , , 1 , 1 , 1 , 1 , = + + + + + − + + + + − + + Direct Methods Cramers’ rule Cramers’ rule Gauss Elimination Iterative Methods Point iterative methods Line iterative methods Outline Elliptic PDE Parabolic PDE (2-d) Approximate factorization TDMA (Thomas Algorithm) 2 2 x T t T ∂ ∂ = ∂ ∂ α 1 i i-1 i+1 N Implicit scheme ( )2 1 1 1 1 1 1 2 x T T T t T T n i n i n i n i n i ∆ + − = ∆ − + − + + + + α ( ) x t ∆ ∆ ( ) ( ) 1 1 1 1 1 2 1 2 + − + + + + + − ∆ ∆ + = n i n i n i n i n i T T T x t T T α n i n i i n i i n i i d T c T b T a = + + + − + + − 1 1 1 1 1 TDMA In a matrix form Gauss Elimination Thomas Algorithm Thomas Algorithm This slide is only notional !
  • 9. Elliptic PDE 0 2 2 2 2 = ∂ ∂ + ∂ ∂ y T x T ( ) ( ) 0 2 2 2 1 , , 1 , 2 , 1 , , 1 = ∆ + − + ∆ + − − + − + y T T T x T T T j i j i j i j i j i j i ) , ( 2 2 2 2 y x f y T x T = ∂ ∂ + ∂ ∂ ( ) ( ) ∆ ∆ y x ( ) ( ) ( ) 0 2 2 1 , , 1 , 2 2 , 1 , , 1 = + − ∆ ∆ + + − − + − + j i j i j i j i j i j i T T T y x T T T 0 ) 1 ( 2 , 2 1 , 2 1 , 2 , 1 , 1 = + − + + + − + − + j i j i j i j i j i T T T T T β β β ( ) ( ) ) 1 ( 2 2 2 2 2 β α β + − = ∆ ∆ = y x Let In a matrix form Parabolic PDE : Implicit         ∂ ∂ + ∂ ∂ = ∂ ∂ 2 2 2 2 y T x T t T α ( ) ( )         ∆ + − + ∆ + − = ∆ − + − + + + + − + + + + 2 1 1 , 1 , 1 1 , 2 1 , 1 1 , 1 , 1 , 1 , 2 2 y T T T x T T T t T T n j i n j i n j i n j i n j i n j i n j i n j i α N n j i n j i y n j i y n j i y x n j i x n j i x T T d T d T d d T d T d , 1 1 , 1 1 , 1 , 1 , 1 1 , 1 ) 1 2 2 ( − = + + + + − + + − + + + + − + + n j i n j i j i n j i j i n j i j i n j i j i n j i j i f T e T d T c T b T a , 1 1 , , 1 1 , , 1 , , 1 , 1 , 1 , 1 , = + + + + + − + + + + − + + 1 1 , 2 2 , 2 1 2 , 1 2 , 2 2 , 2 1 3 , 2 2 , 2 1 2 , 2 2 , 2 1 2 , 3 2 , 2 + + + + + − − = + + n n n n n n T e T b f T d T c T a In a matrix form The coefficient matrix is Pentadiagonal . Solution procedure is very time consuming. How to over come this inefficiency ? ADI : Alternating Direction Implicit method Fractional Step Methods Approximate Factorization methods What is the order of accuracy of this scheme ? Fractional Step Methods Solving a system of Equations n j i n j i j i n j i j i n j i j i n j i j i n j i j i f T e T d T c T b T a , 1 1 , , 1 1 , , 1 , , 1 , 1 , 1 , 1 , = + + + + + − + + + + − + + Direct Methods Cramers’ rule Cramers’ rule Gauss Elimination Iterative Methods Point iterative methods Line iterative methods
  • 10. Outline ADI Fractional step methods Approximate factorization ADI : Alternating Direction Implicit method Fractional Step Methods The Navier-Stokes equations u u u u 2 p 1 ∇ + ∇ − = ∇ + ∂ ∂ ν ρ t 2 2 x T t T ∂ ∂ = ∂ ∂ α ( ) ( ) 1 1 1 1 1 2 1 2 + − + + + + + − ∆ ∆ + = n i n i n i n i n i T T T x t T T α 0 = ⋅ ∇ u n 2 n n n n 1 n p 1 u u u u - u ∇ + ∇ − ∇ ⋅ − = ∆ + ν ρ t Anything wrong ? Governing equations n 2 n n n n 1 n p 1 u u u u - u ∇ + ∇ − ∇ ⋅ − = ∆ + ν ρ t 0 1 n ≠ ⋅ ∇ + u n 2 n n n n 1 n p u u u u u ∇ ∆ + ∇ ⋅ ∆ − = ∇ ∆ + + t t t ν ρ ( ) n 2 n n n 1 n 2 p u u u u ∇ ∆ + ∇ ⋅ ∆ − ⋅ ∇ ∆ = ∇ + t t t ν ρ ρ 0 1 n = ⋅ ∇ + u Is there any problem now ?         ∇ + ∇ − ∇ ⋅ − ∆ + = n 2 n n n n p 1 ~ u u u u u ν ρ t Operator Splitting Methods         ∇ + ∇ − ∇ ⋅ − ∆ + = n 2 n n n n p ~ u u u u u ν ρ β t       ∇ + ∇ − ∇ ⋅ − ∆ + = + + n 2 1 n n n n 1 n p 1 u u u u u ν ρ t       ∇ + ∇ − ∇ ⋅ − ∆ + = p u u u u u ν ρ t ( ) n 1 n 1 n p p ~ β ρ − ∇ ∆ − = − + + t u u φ ρ ∇ ∆ − = − + t u u ~ 1 n ( ) n 1 n p p β φ − = + u ~ 2 ⋅ ∇ ∆ − = ∇ t ρ φ Operator Splitting Methods φ ∇ ∆ − = + t u u ~ 1 n u ~ 2 ⋅ ∇ ∆ − = ∇ t ρ φ φ β + = + n 1 n p p ( ) Γ + Γ − ∆ = ∇ u u ~ 1 n ρ φ t What is a suitable BC for φ ? φ ρ ∇ ∆ − = + t u u ~ 1 n (1) Predict (2) Compute (3) Compute the new velocity and pressure field u ~ φ 1 n 1 n p + + u Fractional step Methods The above eqn. can be split as, 1 n 2 1 n n n n 1 n p 1 + + + ∇ + ∇ − ∇ ⋅ − = ∆ u u u u - u ν ρ t 0 1 n = ⋅ ∇ + u 0 ~ n n n = ∇ ⋅ ∆ + u u u - u t 1 n 1 n ~ ~ + + ∇ ∆ − = p t ρ u u u ~ ~ 1 2 ⋅ ∇ ∆ − = ∇ + t pn ρ 0 = ∇ ⋅ ∆ + u u u - u t 0 ~ ~ ~ ~ ~ 2 = ∇ ∆ + = u u u ν t 0 1 n = ⋅ ∇ + u Operator Splitting Methods         ∇ + ∇ − ∇ ⋅ − ∆ + = n 2 n n n n p ~ u u u u u ν ρ β t u ~ 2 ⋅ ∇ ∆ = ∇ t ρ φ φ β + = + n 1 n p p φ ρ ∇ ∆ − = + t u u ~ 1 n φ β + = p p ρ (1) Predict (2) Compute (3) Compute the new velocity and pressure field u ~ φ 1 n 1 n p + + u x v y u ∂ ∂ − = ∂ ∂ = ψ ψ         ∂ ∂ − ∂ ∂ − = ∂ ∂ + ∂ ∂ y u x v y x 2 2 2 2 ψ ψ
  • 11. Outline Assignment 3 Q(3). Iterative methods Assignment 3 – Q(3) 0 2 2 2 2 = ∂ ∂ + ∂ ∂ y T x T ( ) ( ) 0 2 2 2 1 , , 1 , 2 , 1 , , 1 = ∆ + − + ∆ + − − + − + y T T T x T T T j i j i j i j i j i j i ( ) ( ) ) 1 ( 2 2 2 2 2 β α β + − = ∆ ∆ = y x If 0 ) 1 ( 2 , 2 1 , 2 1 , 2 , 1 , 1 = + − + + + − + − + j i j i j i j i j i T T T T T β β β ( ) 1 Jacobi Iteration Gauss-Seidel iteration Line Gauss-Seidel iteration Iterative Techniques ( ) k j i k j i k j i k j i k j i T T T T T 1 , 2 1 , 2 , 1 , 1 2 1 , ) 1 ( 2 1 − + − + + + + + + = β β β ( ) 1 1 , 2 1 , 2 1 , 1 , 1 2 1 , ) 1 ( 2 1 + − + + − + + + + + + = k j i k j i k j i k j i k j i T T T T T β β β ( ) 1 1 , 1 , 2 1 , 1 1 , 2 1 , 1 ) 1 ( 2 + − + + + + + − + − = + + − k j i k j i k j i k j i k j i T T T T T β β Setting up the Eqns. ( ) 1 1 , 2 1 , 2 1 , 1 , 1 2 1 , ) 1 ( 2 1 + − + + − + + + + + + = k j i k j i k j i k j i k j i T T T T T β β β Gauss-Seidel iteration ( ) 1 , 1 , , 1 , 1 2 , ) 1 ( 2 − + − + + + + + = j i j i j i j i j i T T T T T β β β ( ) 1 1 , 1 2 3 , 1 2 1 2 , 0 2 , 2 2 1 2 , 1 ) 1 ( 2 1 + + + + + + + = k k k k k T T T T T β β β Can we accelerate the convergence ?. ( ) 1 1 , 2 1 , 2 1 , 1 , 1 2 1 , ) 1 ( 2 1 + − + + − + + + + + + = k j i k j i k j i k j i k j i T T T T T β β β Point Successive Over Relaxation (PSOR) RHS on T subtract add k j i, ( ) k j i k j i k j i k j i k j i k j i k j i T T T T T T T , 2 1 1 , 2 1 , 2 1 , 1 , 1 2 , 1 , ) 1 ( 2 ) 1 ( 2 1 β β β β + − + + + + + = + − + + − + + ( ) ( ) 1 1 , 1 , 2 1 , 1 , 1 2 , 1 , ) 1 ( 2 ) 1 ( + − + + − + + + + + + + − = k j i k j i k j i k j i k j i k j i T T T T T T β β ω ω As solution progresses, must approach k j i T, 1 , + k j i T Do we have an optimum, ? ω Optimum ω No general guidelines. Optimum is calculated for limited applications Elliptic + Dirchlet BC’s Can we accelerate the convergence ?. Line Successive Over Relaxation (LSOR) ( ) 1 1 , 1 , 2 1 , 1 1 , 2 1 , 1 ) 1 ( 2 + − + + + + + − + − = + + − k j i k j i k j i k j i k j i T T T T T β β 1 1 2 1 ) 1 ( 2 + + + + + − k k k T T T ω β ω Do we have an optimum, ? ω ( ) 1 1 , 1 , 2 , 2 1 , 1 1 , 2 1 , 1 ) 1 )( 1 ( 2 ) 1 ( 2 + − + + + + + − + − − + = + + − k j i k j i k j i k j i k j i k j i T T T T T T ωβ ω β ω β ω LSOR can be introduced to ADI as well ! Outline Mid Term Exam 1st March 2012 (Thursday), 8-8:50 am (Venue : CRC 103) Iterative Methods (ctd…) Review for Mid-Term Exam Solving a system of Equations n j i n j i j i n j i j i n j i j i n j i j i n j i j i f T e T d T c T b T a , 1 1 , , 1 1 , , 1 , , 1 , 1 , 1 , 1 , = + + + + + − + + + + − + + Direct Methods Cramers’ rule Gauss Elimination Iterative Methods Iterative Methods Point iterative methods Jacobi (simplest) Gauss-Seidel PSOR Line iterative methods Line Gauss-Seidel LSOR
  • 12. The End k 1 + k Simple Iterative solvers at a glance 0 2 2 2 2 = ∂ ∂ + ∂ ∂ y T x T ( ) ( ) 0 2 2 2 1 , , 1 , 2 , 1 , , 1 = ∆ + − + ∆ + − − + − + y T T T x T T T j i j i j i j i j i j i ( ) ( )2 2 2 y x If ∆ ∆ = β 0 ) 1 ( 2 , 2 1 , 2 1 , 2 , 1 , 1 = + − + + + − + − + j i j i j i j i j i T T T T T β β β 0 ) 1 ( 2 , 1 , 1 , , 1 , 1 = + − + + + − + − + j i j i j i j i j i T T T T T β β β ( ) k j i k j i k j i k j i k j i T T T T T 1 , 2 1 , 2 , 1 , 1 2 1 , ) 1 ( 2 1 − + − + + + + + + = β β β ( ) 1 1 , 2 1 , 2 1 , 1 , 1 2 1 , ) 1 ( 2 1 + − + + − + + + + + + = k j i k j i k j i k j i k j i T T T T T β β β ( ) 1 1 , 1 , 2 1 , 1 1 , 2 1 , 1 ) 1 ( 2 + − + + + + + − + − = + + − k j i k j i k j i k j i k j i T T T T T β β Can we accelerate the convergence ?. ( ) 1 1 , 2 1 , 2 1 , 1 , 1 2 1 , ) 1 ( 2 1 + − + + − + + + + + + = k j i k j i k j i k j i k j i T T T T T β β β Point Successive Over Relaxation (PSOR) RHS on T subtract add k j i, ( ) k j i k j i k j i k j i k j i k j i k j i T T T T T T T , 2 1 1 , 2 1 , 2 1 , 1 , 1 2 , 1 , ) 1 ( 2 ) 1 ( 2 1 β β β β + − + + + + + = + − + + − + + ( ) ( ) 1 1 , 1 , 2 1 , 1 , 1 2 , 1 , ) 1 ( 2 ) 1 ( + − + + − + + + + + + + − = k j i k j i k j i k j i k j i k j i T T T T T T β β ω ω As solution progresses, must approach k j i T, 1 , + k j i T Do we have an optimum, ? ω Optimum ω No general guidelines are available. Optimum is calculated for limited applications Elliptic + Dirchlet BC’s Can we accelerate the convergence ?. Line Successive Over Relaxation (LSOR) ( ) 1 1 , 1 , 2 1 , 1 1 , 2 1 , 1 ) 1 ( 2 + − + + + + + − + − = + + − k j i k j i k j i k j i k j i T T T T T β β Can you modify this by introducing ω Do we have an optimum, ? ω ( ) 1 1 , 1 , 2 , 2 1 , 1 1 , 2 1 , 1 ) 1 )( 1 ( 2 ) 1 ( 2 + − + + + + + − + − − + = + + − k j i k j i k j i k j i k j i k j i T T T T T T ωβ ω β ω β ω LSOR can be introduced to ADI as well ! ADI for Parabolic PDE ADI for Elliptic PDE The solution procedure can be accelerated by introducing relaxation parameter ) (ω Outline Streamfunction-vorticity formulations Incorporation of upwind for Governing equation for vorticity transport u u u u 2 1 ∇ + ∇ − = ∇ ⋅ + ∂ ∂ ν ρ p t       ∇ + ∇ − = ∇ ⋅ + ∂ ∂ × ∇ u u u u 2 1 ν ρ p t u ω × ∇ = u) u 2 1 u u) u u ⋅ ∇ + × × ∇ = ∇ ⋅ ( ( u ω - ω u u) ω ) ( ) ( ( ∇ ⋅ ∇ ⋅ = × × ∇ 0 ω = ⋅ ∇ ω u) ω ω 2 ( ∇ = × × ∇ + ∂ ∂ ν t u) u × ∇ ∇ = ∇ × ∇ ( 2 2 u ω - ω u u) ω ) ( ) ( ( ∇ ⋅ ∇ ⋅ = × × ∇ ω u ω ω 2 ) ( ∇ + ∇ ⋅ = ν Dt D
  • 13. Tracking vorticity distributions Vorticity Vorticity ω is governed by an evolution eqn. which is much simpler than N-S. Unlike u, ω can neither be created nor destroyed in the fluid interior. u ω × ∇ = the fluid interior. It is transported throughout the flow field by familiar processes such as, advection and diffusion. Localized distributions of ω remain localized, which is not the case with velocity field. So, an Eddy in a turbulent flow blob of vorticity and its associated rotational and irrotational motions. Governing Equations Stability is governed by, Operator Splitting Methods         ∇ + ∇ − ∇ ⋅ − ∆ + = n 2 n n n n p ~ u u u u u ν ρ β t u ~ 2 ⋅ ∇ ∆ = ∇ t ρ φ φ β + = + n 1 n p p φ ρ ∇ ∆ − = + t u u ~ 1 n φ β + = p p ρ (1) Predict (2) Compute (3) Compute the new velocity and pressure field u ~ φ 1 n 1 n p + + u x v y u ∂ ∂ − = ∂ ∂ = ψ ψ         ∂ ∂ − ∂ ∂ − = ∂ ∂ + ∂ ∂ y u x v y x 2 2 2 2 ψ ψ Governing Equations Central difference form for the above equation, The above difference equation numerically allows checkerboard velocity distribution. Discrete Checkerboard velocity Discrete Checkerboard for pressure Governing Equations u u u u 2 p 1 ∇ + ∇ − = ∇ + ∂ ∂ ν ρ t 0 = ⋅ ∇ u For viscous, incompressible flows : ω ∂ In non-primitive (ψ-ω) form : ω u) ω ω 2 ( ∇ = × × ∇ + ∂ ∂ ν t u ω × ∇ = x v y u ∂ ∂ − = ∂ ∂ = ψ ψ ;         ∂ ∂ + ∂ ∂ = ∂ ∂ + ∂ ∂ + ∂ ∂ 2 2 2 2 y x y v x u t ω ω ν ω ω ω         ∂ ∂ − ∂ ∂ − = ∂ ∂ + ∂ ∂ y u x v y x 2 2 2 2 ψ ψ MAC algorithm Harlow and Welch (1965) One of the earliest methods of solving N-S in primitive variables. Hirt and Cook (1972) A pressure Poisson equation is formulated A pressure Poisson equation is formulated The momentum equations are used for the computation of velocities.
  • 14. Temporal Evolution MAC algorithm Define pressure correction terms as : MAC algorithm Define pressure correction terms as : Continuity equation transforms into : Governing Equations Central difference form for the above equation, The above difference equation numerically allows checkerboard velocity distribution. Discrete Checkerboard velocity Discrete Checkerboard for pressure Outline Checker board pressure patterns Staggered grids One-Dimensional N-S ) 2 ( 1 ) 1 ( 0 2 2 x u x p x u u t u x u ∂ ∂ + ∂ ∂ − = ∂ ∂ + ∂ ∂ = ∂ ∂ ν ρ 1 i i-1 i+1 N 1 + n n ) 3 ( 1 1 2 2 1 + +         ∂ ∂ −         ∂ ∂ + ∂ ∂ − = ∆ − n n n i n i x p x u x u u t u u ρ ν ) 4 ( ~ 2 2 n n i i x u x u u t u u         ∂ ∂ + ∂ ∂ − = ∆ − ν         ∂ ∂ − = ∆ − ⇒ − + x p t u u i n i ρ 1 ~ ) 4 ( ) 3 ( 1 One-Dimensional N-S 1 i i-1 i+1 N 0 2 1 1 1 1 = ∆ − = ∂ ∂ + − + + x u u x u n i n i ) 4 ( 1 ~ 1         ∂ ∂ − = ∆ − − + x p t u u i n i ρ         ∆ − − = ∆ − − + + x p p t u u i i i n i 2 1 ~ 1 1 1 ρ x p p t u u i i i n i ∆ − ∆ − = − + + 2 ~ 1 1 1 ρ ? 1 1 = + + n i u ? 1 1 = + − n i u 1 1 2 2 ~ ~ 2 2 − + − + − = ∆ + − ∆ i i i i i u u x p p p t ρ
  • 15. Checkerboard pressures This delinking between velocity and pressure at a grid point is an impediment and results in zig-zag type checkerboard pressure values. To avoid this, MAC algorithm introduces Staggered grids in place of “collocated grids”. Staggered grids in place of “collocated grids”. One-Dimensional N-S 1 i i-1 i+1 N ( ) 0 2 / 2 1 2 / 1 1 2 / 1 = ∆ − = ∂ ∂ + − + + x u u x u n i n i i-1/2 i+1/2 i+3/2 ( ) 2 / 1 2 / 1 1 1 ~ ~ 2 − + − + − ∆ = ∆ + − i i i i i u u t x p p p ρ ( )         ∆ − − = ∆ − + + + + x p p t u u i i i n i 1 2 / 1 1 2 / 1 1 ~ ρ 2D - Governing Equations Central difference form for the above equation, The above difference equation numerically allows the checkerboard velocity distribution. Such a distribution is not representative of a physical flow field. This is NOT an issue for compressible flows, as inclusion of density variations would wipe out checkerboard pressure pattern. Discrete Checkerboard velocity Discrete Checkerboard for pressure Pressure Correction schemes The 2-D Navier-Stokes A Staggered grid Staggered grid ) 1 ( 0 = ∂ ∂ + ∂ ∂ y v x u Write a central Write a central differencing expression for the above equation, around the grid point (i,j).
  • 16. Staggered grid Difference equation for x-momentum equation about (i+1/2,j). Outline Governing Equations in FM Differential form : FDM Integral form : FVM, FEM Finite Volume Method FDM regular grids Industrial problems Complex domains FDM coordinate transformation Loss of computational efficiency and accuracy FVM Integral form of eqns. (greater flexibility in handling complex domains) In FVM conservation laws are applied on the elementary volumes. 1-D steady diffusion equation ) 1 ( ) ( ) ( ) ( φ φ φ ρ ρφ S V t + ∇ Γ ⋅ ∇ = ⋅ ∇ + ∂ ∂ r ) 2 ( 0 ) ( = + ∇ Γ ⋅ ∇ φ φ S Generic transport equation for the property ϕ ) 2 ( 0 ) ( = + ∇ Γ ⋅ ∇ φ φ S ) 3 ( 0 ) ( = + ∇ Γ ⋅ ∇ ∫ ∫ CV CV dv S dv φ φ ) 4 ( 0 ) ( = + ∇ Γ ∫ ∫ CV A dv S dA φ φ 1-D steady diffusion equation ) 5 ( 0 = +       Γ S dx d dx d φ 1-D diffusion equation for the property ϕ   dT d ) 6 ( 0 = +       S dx dT k dx d Discretization in FV Physical boundary that coincides with CV boundary. Discretization ) 5 ( 0 = +       Γ S dx d dx d φ ) 6 ( 0 = +       Γ ∫ ∫ CV CV dv S dv dx d dx d φ ) 7 ( 0 = +       Γ −       Γ dv S dx d A dx d A w e φ φ   −   d φ φ φ =   Γ dφ ) 8 (         − Γ =       Γ PE P E e e e x A dx d A δ φ φ φ ) 10 ( P P u S S dv S φ + = ) 9 ( ? =       Γ w dx d A φ Substitute (8), (9) and (10) in Eqn.(7) and rearrange NOW! Discretization Discretized equations of the above form must be setup at each nodal point. Modify the discretized equation to incorporate BC. Solve the resulting system of linear algebraic equations. 1-D Heat conduction example At grid points 2, 3, 4 :
  • 17. 1-D Heat conduction example At the boundary point 1: ? How dx d A dx d A w e       Γ −       Γ φ φ Rearrange the above eqn. in the following form : dx dx w e     1-D Heat conduction example At the boundary point 5: ? How dx d A dx d A w e       Γ −       Γ φ φ Rearrange the above eqn. in the following form : dx dx w e     Outline FVM 1-D heat conduction equation 2-D diffusion equation 1-D convection-diffusion equation 1-D convection-diffusion equation Finite Volume Method FVM Integral form of the conservation laws are discretized directly in the physical space. Use a mesh (where the cell centre refers to the grid points, while the cell faces coincide with the domain boundaries). FVM has great advantage that the conservative FVM has great advantage that the conservative discretization is automatically satisfied (by directly using the integral form of the conservation laws). 1-D Heat conduction example At grid points 2, 3, 4 : E E W W P P T a T a T a + = At the boundary point 1: ? How dx d A dx d A w e       Γ −       Γ φ φ BC at A : Applying TA Rearrange the above eqn. in the following form : dx dx w e     BC at B : Applying TB At the boundary point 5: ? How dx d A dx d A w e       Γ −       Γ φ φ Rearrange the above eqn. in the following form : dx dx w e     1-D Heat conduction example At the boundary point 5: ? How dx d A dx d A w e       Γ −       Γ φ φ Rearrange the above eqn. in the following form : dx dx w e     Resulting algebraic equations
  • 18. Comparison of the numerical solution FVM for 2-d Diffusion problems When the above governing equation is integrated over the CV, we obtain Flux through the CV faces Substitute above expressions and rearrange NOW! Formulation of algebraic expressions Four Basic Rules in FVM Rule 1 Flux consistency across the control volume faces in exit q q = 0 , , W E P a a a Rule 2 Positive coefficients 0 , , W E P a a a P P u S S S φ + = v S a a P neighbors P ∆ − = ∑ Rule 4 Sum of neighboring coefficients : Rule 3 Negative slope linearization of the source term. Outline FVM 1-D convection-diffusion equation Steady 1-D convection-diffusion equation Integrating the transport and continuity equation, Steady 1-D convection-diffusion equation Introduce the following variables, 1-d CDE : central differencing F/D ratio : 1.25
  • 19. 1-d CDE : central differencing F/D ratio : 5 1-d CDE : upwind differencing F/D ratio : 5 QUICK scheme 2 u/s GP + 1 d/s GP A 2-D test case 2-D test case : Upwind differencing 2-D test case : QUICK TVD schemes TVD : Total Variation Diminishing TVD is specially formulated to achieve oscillation- free solutions. Upwind most stable. But, introduces high level of false diffusion. CDE, QUICK Spurious oscillations or wiggles, CDE, QUICK Spurious oscillations or wiggles, when Peclet number is high In turbulent flows, wiggles can give rise to physically unrealistic negative values and instability. In TVD, the tendency towards oscillation is counteracted by adding an artificial diffusion fragment or weighting towards upstream contribution. The END