Lecture # 02 Goverening Equations and Its Discritization

Lecture # 02 Goverening Equations and its Discritization
Getting started: CFD notation

∂u ∂u ∂u ∂2u ∂pu
PDE of p-th order f u, x, t, ∂x1 , . . . , , ,
∂xn ∂t ∂x1 ∂x2 , . . . , ∂tp =0
scalar unknowns u = u(x, t), x ∈ Rn , t ∈ R, n = 1, 2, 3

vector unknowns v = v(x, t), v ∈ Rm , m = 1, 2, . . .
∂ ∂ ∂
Nabla operator ∇ = i ∂x + j ∂y + k ∂z x = (x, y, z), v = (vx , vy , vz )
h iT
∇u = i ∂u
∂x + j ∂u
∂y + k ∂u
∂z = ∂u ∂u ∂u
∂x , ∂y , ∂z gradient
z
∂vx ∂vy ∂vz v
∇·v = ∂x + ∂y + ∂z divergence
   ∂vy

∂vz
i j k ∂y − ∂z k
    y
∇ × v = det  ∂ ∂ ∂ = ∂vx ∂vz  curl i j
 ∂x ∂y ∂z   ∂z − ∂x 
∂vy ∂vx
vx vy vz ∂x − ∂y x
∂2u ∂2u ∂2u
∆u = ∇ · (∇u) = ∇2 u = ∂x2 + ∂y 2 + ∂z 2 Laplacian
JetWings 1
Tensorial quantities in fluid dynamics
Velocity gradient
  1111111111111111
0000000000000000
0000000000000000
1111111111111111
∂vx ∂vy ∂vz
∂x ∂x ∂x
 ∂vy

∇v = [∇vx , ∇vy , ∇vz ] =  ∂vx ∂vz 
 ∂y ∂y ∂y 
∂vx ∂vy ∂vz v
∂z ∂z ∂z 1111111111111111
0000000000000000
0000000000000000
1111111111111111
Remark. The trace (sum of diagonal elements) of ∇v equals ∇ · v.
Deformation rate tensor (symmetric part of ∇v)

 
∂vx 1 ∂vy ∂vx 1 ∂vz ∂vx
 ∂x 2 ∂x + ∂y 2 ∂x + ∂z 
1 T  1

∂vx ∂vy

∂vy 1

∂vz ∂vy


D(v) = (∇v + ∇v ) =  ∂y + ∂x + 
2  2
∂y 2 ∂y ∂z 
1 ∂vx ∂vz 1 ∂vy ∂vz ∂vz
2 ∂z + ∂x 2 ∂z + ∂y ∂z
Spin tensor S(v) = ∇v − D(v) (skew-symmetric part of ∇v)
JetWings 2
Vector multiplication rules
Scalar product of two vectors

 
b1
 
a, b ∈ R3 , a · b = aT b = [a1 a2 a3 ] 
 b2  = a1 b1 + a2 b2 + a3 b3 ∈ R

b3
∂u ∂u ∂u
Example. v · ∇u = vx + vy + vz convective derivative
∂x ∂y ∂z
Dyadic product of two vectors

   
a1 a1 b1 a1 b2 a1 b3
   
3
a, b ∈ R , a ⊗ b = abT = 
 a2  [b1 b2 b3 ] =  a2 b1
  a2 b2 ∈R
a2 b3 
3×3
a3 a3 b1 a3 b2 a3 b3
JetWings 3
Elementary tensor calculus

1. αT = {αtij }, T = {tij } ∈ R3×3 , α ∈ R
2. T 1 + T 2 = {t1ij + t2ij }, T 1 , T 2 ∈ R3×3 , a ∈ R3

 
t t t
 11 12 13  P 3
3. a · T = [a1 , a2 , a3 ]  t21 t22 t23  = ai [ti1 , ti2 , ti3 ]
i=1 | {z }
t31 t32 t33 i-th row
    
t t t a t
 11 12 13   1  P 3
 1j 
4. T · a =  t21 t22 t23   a2  =  t2j  aj (j-th column)
j=1
t31 t32 t33 a3 t3j
  
1 1 1 2 2 2
t11 t12 t13 t11 t12 t13 3
P

  
5. T 1 · T 2 =  t121 t122 t123   t221 t222 t223  = t1ik t2kj
k=1
t131 t132 t133 t231 t232 t233
3 P
P 3
1 2 1 2 T
6. T : T = tr (T · (T ) ) = t1ik t2ik
i=1 k=1
JetWings 4
Divergence theorem of Gauß
Let Ω ∈ R3 and n be the outward unit normal to the boundary Γ = Ω̄\Ω.

Z Z
Then ∇ · f dx = f · n ds for any differentiable function f (x)
Ω Γ
Example. A sphere: Ω = {x ∈ R3 : ||x|| < 1}, Γ = {x ∈ R3 : ||x|| = 1}

√ p
where ||x|| = x·x= x2 + y 2 + z 2 is the Euclidean norm of x
Consider f (x) = x so that ∇ · f ≡ 3 in Ω and n = x

||x|| on Γ
Z Z
4 3
volume integral: ∇ · f dx = 3 dx = 3|Ω| = 3 π1 = 4π
Ω Ω 3
Z Z Z Z
x·x
surface integral: f · n ds = ds = ||x|| ds = ds = 4π
Γ Γ ||x|| Γ Γ
JetWings 5
Governing equations of fluid dynamics
Physical principles Mathematical equations

1. Mass is conserved • continuity equation
2. Newton’s second law • momentum equations
3. Energy is conserved • energy equation
It is important to understand the meaning and significance of each equation

in order to develop a good numerical method and properly interpret the results
Description of fluid motion z

v
Eulerian monitor the flow characteristics (x 1 ; y1 ; z1 )
in a fixed control volume k (x 0 ; y0 ; z0 )
i j y
Lagrangian track individual fluid particles as
they move through the flow field x
JetWings 6
Description of fluid motion

Trajectory of a fluid particle
z
x = x(x0 , t) dx
v = vx (x, y, z, t), x|t0 = x0
(x 1 ; y1 ; z1 )
dt
x = x(x0 , y0 , z0 , t) dy
(x 0 ; y0 ; z0 )
= vy (x, y, z, t), y|t0 = y0
k
y y = y(x0 , y0 , z0 , t) dt
i j
dz
z = z(x0 , y0 , z0 , t) = vz (x, y, z, t), z|t0 = z0
dt
x
Definition. A streamline is a curve which is tangent to the velocity vector

v = (vx , vy , vz ) at every point. It is given by the relation
y
dx dy dz v dy
=
vy
= = y(x) dx vx
vx vy vz
x
Streamlines can be visualized by injecting tracer particles into the flow field.
JetWings 7
Flow models and reference frames
Eulerian Lagrangian
S S
integral
V V
fixed CV of a finite size moving CV of a finite size
differential
dS dS
dV dV
fixed infinitesimal CV moving infinitesimal CV
Good news: all flow models lead to the same equations
JetWings 8
Eulerian vs. Lagrangian viewpoint
d
Definition. Substantial time derivative dt is the rate of change for a moving
∂
fluid particle. Local time derivative ∂t is the rate of change at a fixed point.
Let u = u(x, t), where x = x(x0 , t). The chain rule yields
du ∂u ∂u dx ∂u dy ∂u dz ∂u
= + + + = + v · ∇u
dt ∂t ∂x dt ∂y dt ∂z dt ∂t
substantial derivative = local derivative + convective derivative
Reynolds transport theorem

Z Z Z
d ∂u(x, t)
u(x, t) dV = dV + u(x, t)v · n dS
dt Vt V ≡Vt ∂t S≡St
rate of change in rate of change in convective transfer

= +
a moving volume a fixed volume through the surface
JetWings 9
Derivation of the governing equations
Modeling philosophy Generic conservation law

Z Z Z
1. Choose a physical principle ∂
u dV + f · n dS = q dV
∂t V S V
• conservation of mass
S n
f = vu − d∇u
• conservation of momentum V
dS f
flux function
• conservation of energy
Divergence theorem yields
2. Apply it to a suitable flow model Z Z Z
∂u
• Eulerian/Lagrangian approach dV + ∇ · f dV = q dV
V ∂t V V
• for a finite/infinitesimal CV
Partial differential equation
3. Extract integral relations or PDEs ∂u
+∇·f =q in V
which embody the physical principle ∂t
JetWings 10
Derivation of the continuity equation
Physical principle: conservation of mass

Z Z Z
dm d ∂ρ
= ρ dV = dV + ρv · n dS = 0
dt dt Vt V ≡Vt ∂t S≡St
accumulation of mass inside CV = net influx through the surface
Divergence theorem yields Continuity equation

Z
∂ρ ∂ρ
+ ∇ · (ρv) dV = 0 ⇒ + ∇ · (ρv) = 0
V ∂t ∂t
Lagrangian representation
dρ
∇ · (ρv) = v · ∇ρ + ρ∇ · v ⇒ + ρ∇ · v = 0
dt
dρ
Incompressible flows: dt =∇·v =0 (constant density)
JetWings 11
Conservation of momentum
Physical principle: f = ma (Newton’s second law)
total force f = ρg dV + h dS, where h=σ·n

dS
dV n body forces g gravitational, electromagnetic,. . .
h surface forces h pressure + viscous stress
g
Stress tensor σ = −pI + τ momentum flux
For a newtonian fluid viscous stress is proportional to velocity gradients:

1 2
τ = (λ∇ · v)I + 2µD(v), where D(v) = (∇v + ∇vT ), λ≈− µ
2 3
Normal stress: stretching Shear stress: deformation

“ ”
∂vy ∂vx
τxx = λ∇ · v + 2µ ∂v
∂x
x y
τxy = τyx = µ ∂x
+ ∂y
y
yx
∂v
= µ ∂v ∂vz
` x ´
τyy = λ∇ · v + 2µ ∂yy τxz = τzx +
xx
“ ∂z ∂x ”
∂v
τzz = λ∇ · v + 2µ ∂v
∂z
z
τyz = τzy = µ ∂v
∂y
z
+ ∂zy
x x
JetWings 12
Derivation of the momentum equations
Newton’s law for a moving volume

Z Z Z
d ∂(ρv)
ρv dV = dV + (ρv ⊗ v) · n dS
Z Z
= ρg dV + σ · n dS
V ≡Vt S≡St
Transformation of surface integrals

Z Z
∂(ρv)
+ ∇ · (ρv ⊗ v) dV = [∇ · σ + ρg] dV, σ = −pI + τ
V ∂t V
∂(ρv)
Momentum equations + ∇ · (ρv ⊗ v) = −∇p + ∇ · τ + ρg
∂t

∂(ρv) ∂v ∂ρ dv
+ ∇ · (ρv ⊗ v) = ρ + v · ∇v + v + ∇ · (ρv) = ρ
∂t ∂t ∂t dt
| {z } | {z }
substantial derivative continuity equation
JetWings 13
Conservation of energy
Physical principle: δe = s + w (first law of thermodynamics)
dS δe accumulation of internal energy

dV n
s heat transmitted to the fluid particle
h
g w rate of work done by external forces
Heating: s = ρq dV − fq dS Fourier’s law of heat conduction

q internal heat sources fq = −κ∇T
fq diffusive heat transfer
the heat flux is proportional to the
T absolute temperature
local temperature gradient
κ thermal conductivity
Work done per unit time = total force × velocity
w = f · v = ρg · v dV + v · (σ · n) dS, σ = −pI + τ
JetWings 14
Derivation of the energy equation

|v|2
Total energy per unit mass: E =e+ 2
e specific internal energy due to random molecular motion
2
|v|
2 specific kinetic energy due to translational motion
Integral conservation law for a moving volume

Z Z Z
d ∂(ρE)
ρE dV = dV + ρE v · n dS accumulation
Z Z
= ρq dV + κ∇T · n dS heating
V ≡Vt S≡St
Z Z
+ ρg · v dV + v · (σ · n) dS work done
V ≡Vt S≡St
Transformation of surface integrals

Z Z
∂(ρE)
+ ∇ · (ρEv) dV = [∇ · (κ∇T ) + ρq + ∇ · (σ · v) + ρg · v] dV,
V ∂t V
where ∇ · (σ · v) = −∇ · (pv) + ∇ · (τ · v) = −∇ · (pv) + v · (∇ · τ ) + ∇v : τ
JetWings 15
Different forms of the energy equation
Total energy equation
∂(ρE)
+ ∇ · (ρEv) = ∇ · (κ∇T ) + ρq − ∇ · (pv) + v · (∇ · τ ) + ∇v : τ + ρg · v
∂t

∂(ρE) ∂E ∂ρ dE
+ ∇ · (ρEv) = ρ + v · ∇E + E + ∇ · (ρv) = ρ
∂t ∂t ∂t dt
| {z } | {z }
substantial derivative continuity equation
dv
Momentum equations ρ = −∇p + ∇ · τ + ρg (Lagrangian form)
dt
dE de dv ∂(ρe)
ρ =ρ +v·ρ = + ∇ · (ρev) + v · [−∇p + ∇ · τ + ρg]
dt dt dt ∂t
Internal energy equation
∂(ρe)
+ ∇ · (ρev) = ∇ · (κ∇T ) + ρq − p∇ · v + ∇v : τ
∂t
JetWings 16
Summary of the governing equations
1. Continuity equation / conservation of mass

∂ρ
+ ∇ · (ρv) = 0
∂t
2. Momentum equations / Newton’s second law

∂(ρv)
+ ∇ · (ρv ⊗ v) = −∇p + ∇ · τ + ρg
∂t
3. Energy equation / first law of thermodynamics

∂(ρE)
+ ∇ · (ρEv) = ∇ · (κ∇T ) + ρq − ∇ · (pv) + v · (∇ · τ ) + ∇v : τ + ρg · v
∂t
|v|2 ∂(ρe)
E =e+ , + ∇ · (ρev) = ∇ · (κ∇T ) + ρq − p∇ · v + ∇v : τ
2 ∂t
This PDE system is referred to as the compressible Navier-Stokes equations
JetWings 17
Conservation form of the governing equations

Generic conservation law for a scalar quantity
∂u
+ ∇ · f = q, where f = f (u, x, t) is the flux function
∂t
Conservative variables, fluxes and sources

     
ρ ρv 0
     
U =  ρv  , F = 
  ρv ⊗ v + pI − τ ,
 Q=
 ρg 

ρE (ρE + p)v − κ∇T − τ · v ρ(q + g · v)
Navier-Stokes equations in divergence form
∂U
+∇·F=Q U ∈ R5 , F ∈ R3×5 , Q ∈ R5
∂t
• representing all equations in the same generic form simplifies the programming
• it suffices to develop discretization techniques for the generic conservation law
JetWings 18
Constitutive relations
Variables: ρ, v, e, p, τ , T Equations: continuity, momentum, energy
The number of unknowns exceeds the number of equations.
1. Newtonian stress tensor

1 2
τ = (λ∇ · v)I + 2µD(v), D(v) = (∇v + ∇vT ), λ≈− µ
2 3
2. Thermodynamic relations, e.g.

p = ρRT ideal gas law R specific gas constant
e = cv T caloric equation of state cv specific heat at constant volume
Now the system is closed: it contains five PDEs for five independent variables
ρ, v, e and algebraic formulae for the computation of p, τ and T . It remains to
specify appropriate initial and boundary conditions.
JetWings 19
Initial and boundary conditions
Initial conditions ρ|t=0 = ρ0 (x), v|t=0 = v0 (x), e|t=0 = e0 (x) in Ω
Boundary conditions Let Γ = Γin ∪ Γw ∪ Γout

w
Inlet Γin = {x ∈ Γ : v · n < 0}

ρ = ρin , v = vin , e = ein in out
prescribed density, energy and velocity

w
Solid wall Γw = {x ∈ Γ : v · n = 0} Outlet Γout = {x ∈ Γ : v · n > 0}

v=0 no-slip condition v · n = vn or −p + n · τ · n = 0
T = Tw given temperature or v · s = vs or s·τ ·n=0
∂T
fq
prescribed velocity vanishing stress
∂n = − κ prescribed heat flux
The problem is well-posed if the solution exists, is unique and depends continuously
on IC and BC. Insufficient or incorrect IC/BC may lead to wrong results (if any).
JetWings 20
Dimensionless form of equations
Motivation: sometimes equations are normalized in order to
• facilitate the scale-up of obtained results to real flow conditions

• avoid round-off due to manipulations with large/small numbers
• assess the relative importance of terms in the model equations
Dimensionless variables and numbers

t x v p T − T0
t∗ = , x∗ = , v∗ = , p∗ = , T∗ =
t0 L0 v0 ρv02 T1 − T0
ρv0 L0 inertia |v|

Reynolds number Re = µ viscosity Mach number M= c
Froude number Fr = √v0 inertia

Strouhal number St = L0
L0 g gravity v0 t0
v0 L0 convection µ
Peclet number Pe = κ diffusion Prandtl number Pr = ρκ
JetWings 21
Model simplifications
Objective: derive analytical solutions / reduce computational cost
Compressible Navier-Stokes equations
ρ = const µ=0
Incompressible Navier-Stokes equations Compressible Euler equations
Stokes flow boundary layer inviscid Euler equations potential flow
Derivation of a simplified model

1. determine the type of flow to be simulated
2. separate important and unimportant effects
3. leave irrelevant features out of consideration
4. omit redundant terms/equations from the model
5. prescribe suitable initial/boundary conditions
JetWings 22
Classification of partial differential equations
PDEs can be classified into hyperbolic, parabolic and elliptic ones

• each class of PDEs models a different kind of physical processes
• the number of initial/boundary conditions depends on the PDE type
• different solution methods are required for PDEs of different type
Hyperbolic equations Information propagates in certain directions at

finite speeds; the solution is a superposition of multiple simple waves
Parabolic equations Information travels downstream / forward in time;

the solution can be constructed using a marching / time-stepping method
Elliptic equations Information propagates in all directions at infinite speed;

describe equilibrium phenomena (unsteady problems are never elliptic)
JetWings 23
Classification of partial differential equations
∂u ∂u
First-order PDEs a0 + a1 ∂x1
+ . . . + aD ∂xD
=0 are always hyperbolic
D D
X ∂2u X ∂u
Second order PDEs − aij + bk + cu + d = 0
i,j=1
∂xi ∂xj ∂xk
k=1
coefficient matrix: A = {aij } ∈ RD×D , aij = aij (x1 , . . . , xD )

aij +aji
symmetry: aij = aji , otherwise set aij = aji := 2
PDE type n+ n− n0 n+ number of positive eigenvalues

n− number of negative eigenvalues
elliptic D 0 0
n0 number of zero eigenvalues
hyperbolic D−1 1 0
parabolic D−1 0 1 n+ ←→ n−
JetWings 24
Computational meshes
Degrees of freedom for the approximate solution are defined on a computational

mesh which represents a subdivision of the domain into cells/elements
structured block-structured unstructured
Structured (regular) meshes

• families of gridlines do not cross and only intersect with other families once
• topologically equivalent to Cartesian grid so that each gridpoint (or CV) is
uniquely defined by two indices in 2D or three indices in 3D, e.g., (i, j, k)
• can be of type H (nonperiodic), O (periodic) or C (periodic with cusp)
• limited to simple domains, local mesh refinement affects other regions
JetWings 25
Computational meshes
Block-structured meshes
• multilevel subdivision of the domain with structured grids within blocks
• can be non-matching, special treatment is necessary at block interfaces
• provide greater flexibility, local refinement can be performed blockwise
Unstructured meshes
• suitable for arbitrary domains and amenable to adaptive mesh refinement
• consist of triangles or quadrilaterals in 2D, tetrahedra or hexahedra in 3D
• complex data structures, irregular sparsity pattern, difficult to implement
JetWings 26
Discretization techniques
Finite differences / differential form

• approximation of nodal derivatives
• simple and effective, easy to derive
• limited to (block-)structured meshes
Finite volumes / integral form

• approximation of integrals
• conservative by construction
• suitable for arbitrary meshes
Finite elements / weak form

• weighted residual formulation
• remarkably flexible and general
• suitable for arbitrary meshes
JetWings 27
Finite difference method
Principle: derivatives in the partial differential equation are approximated

by linear combinations of function values at the grid points
1D: Ω = (0, X), ui ≈ u(xi ), i = 0, 1, . . . , N

X
grid points xi = i∆x mesh size ∆x = N
0 X
x0 x1 xi 1 xi xi+1 xN 1 xN
First-order derivatives
∂u u(x̄ + ∆x) − u(x̄) u(x̄) − u(x̄ − ∆x)
(x̄) = lim = lim
∂x ∆x→0 ∆x ∆x→0 ∆x
u(x̄ + ∆x) − u(x̄ − ∆x)
= lim (by definition)
∆x→0 2∆x
JetWings 28
Approximation of first-order derivatives
Geometric interpretation
u forward ba kward
entral ∂u
ui+1 −ui
∂x i ≈ ∆x forward difference
∂u
ui −ui−1
x x exa t ∂x i ≈ ∆x backward difference
∂u
ui+1 −ui−1
∂x i ≈ 2∆x central difference
xi 1 xi xi+1 x ∞
P (x−xi )n ∂nu
Taylor series expansion u(x) = n! ∂xn i , u ∈ C ∞ ([0, X])
n=0

∂u (∆x)2 ∂ 2 u (∆x)3 ∂ 3 u
T1 : ui+1 = ui + ∆x + 2
+ 3
+ ...
∂x
i 2 ∂x i 6 ∂x i

∂u (∆x)2 ∂ 2 u (∆x)3 ∂ 3 u
T2 : ui−1 = ui − ∆x + − + ...
∂x i 2 ∂x2 i 6 ∂x3 i
JetWings 29
Analysis of truncation errors
Accuracy of finite difference approximations

∂u ui+1 − ui ∆x ∂2u (∆x)2 ∂3u
T1 ⇒ = − − + ...
∂x i ∆x 2 ∂x2 i 6 ∂x3 i
forward difference truncation error O(∆x)

∂u ui − ui−1 ∆x ∂2u (∆x)2 ∂3u
T2 ⇒ = + − + ...
∂x i ∆x 2 ∂x2 i 6 ∂x3 i
backward difference truncation error O(∆x)

∂u ui+1 − ui−1 (∆x)2 ∂3u
T1 − T2 ⇒ = − + ...
∂x i 2∆x 6 ∂x3 i
central difference truncation error O(∆x)2
Leading truncation error
ǫτ = αm (∆x)m + αm+1 (∆x)m+1 + . . . ≈ αm (∆x)m
JetWings 30
Approximation of second-order derivatives

Central difference scheme

∂2u ui+1 − 2ui + ui−1 2
T1 + T2 ⇒ = + O(∆x)
∂x2 i (∆x)2
Alternative derivation
2 ∂u
∂u

∂ u ∂ ∂u ∂x i+1/2 − ∂x i−1/2
= = lim
∂x2 i ∂x ∂x i ∆x→0 ∆x
ui+1 −ui
∆x − ui −u
∆x
i−1
ui+1 − 2ui + ui−1
≈ =
∆x (∆x)2
Variable coefficients f (x) = d(x) ∂u

∂x diffusive flux
−ui
− di−1/2 ui −u

∂f fi+1/2 − fi−1/2 di+1/2 ui+1
∆x ∆x
i−1
≈ =
∂x i ∆x ∆x
di+1/2 ui+1 − (di+1/2 + di−1/2 )ui + di−1/2 ui−1
=
(∆x)2
JetWings 31
Approximation of mixed derivatives

∂2u ∂ ∂u ∂ ∂u y
2D: ∂x∂y = ∂x ∂y = ∂y ∂x
“ ” “ ”
∂u ∂u
−
∂2u
„ «
∂y ∂y
i+1,j i−1,j
= + O(∆x)2 yj +1
∂x∂y i,j 2∆x
yj
„ «
∂u ui+1,j+1 − ui+1,j−1
= + O(∆y)2 yj 1
∂y i+1,j 2∆y
„ «
∂u ui−1,j+1 − ui−1,j−1
= + O(∆y)2
∂y i−1,j 2∆y xi 1 xi xi+1 x
Second-order difference approximation

2
∂ u ui+1,j+1 − ui+1,j−1 − ui−1,j+1 + ui−1,j−1
= + O[(∆x)2 , (∆y)2 ]
∂x∂y i,j 4∆x∆y
JetWings 32
One-sided finite differences

∂u u1 − u0
= + O(∆x) forward difference
∂x 0 ∆x
? x0 x1 x2
backward/central difference approximations
would need u−1 which is not available
Polynomial fitting

∂u x2 ∂2u x3 ∂3u
u(x) = u0 + x + + + ...
∂x 0 2 ∂x2 0 6 ∂x3 0

∂u ∂u
u(x) ≈ a + bx + cx2 , ≈ b + 2cx, ≈b
∂x ∂x 0
approximate u by a polynomial and differentiate it to obtain the derivatives
u0 = a c∆x2 = u1 − u0 − b∆x
⇒
u1 = a + b∆x + c∆x2 −3u0 + 4u1 − u2
b=
u2 = a + 2b∆x + 4c∆x2 2∆x
JetWings 33
Analysis of the truncation error

∂u αui + βui+1 + γui+2
One-sided approximation ≈
∂x i ∆x

∂u (∆x)2 ∂ 2 u (∆x)3 ∂ 3 u
ui+1 = ui + ∆x + + + ...
∂x i 2 ∂x2 i 6 ∂x3 i
2
2 3
3
∂u (2∆x) ∂ u (2∆x) ∂ u
ui+2 = ui + 2∆x + + + ...
∂x i 2 ∂x2 i 6 ∂x3 i

αui +βui+1 +γui+2 α+β+γ ∂u ∆x ∂2u
∆x = ∆x ui + (β + 2γ) ∂x i + 2 (β + 4γ) ∂x2 + O(∆x2 )
i
Second-order accurate if α + β + γ = 0, β + 2γ = 1, β + 4γ = 0
∂u
−3ui +4ui+1 −ui+2
α = − 23 , β = 2, γ = − 21 ⇒ ∂x i = 2∆x + O(∆x2 )
JetWings 34
Application to second-order derivatives

∂2u αui + βui+1 + γui+2
One-sided approximation ≈
∂x2 i ∆x2

∂u (∆x)2 ∂ 2 u (∆x)3 ∂ 3 u
ui+1 = ui + ∆x + + + ...
∂x i 2 ∂x2 i 6 ∂x3 i
2
2 3
3
∂u (2∆x) ∂ u (2∆x) ∂ u
ui+2 = ui + 2∆x + + + ...
∂x i 2 ∂x2 i 6 ∂x3 i

αui +βui+1 +γui+2 α+β+γ β+2γ ∂u β+4γ ∂2u
∆x2 = ∆x2 ui + ∆x ∂x i + 2 ∂x2 + O(∆x)
i
First-order accurate if α + β + γ = 0, β + 2γ = 0, β + 4γ = 2
∂u
ui −2ui+1 +ui+2
α = 1, β = −2, γ=1 ⇒ ∂x i = ∆x2 + O(∆x)
JetWings 35
High-order approximations

∂u 2ui+1 + 3ui − 6ui−1 + ui−2
= + O(∆x)3 backward difference
∂x i 6∆x

∂u −ui+2 + 6ui+1 − 3ui − 2ui−1
= + O(∆x)3 forward difference
∂x i 6∆x

∂u −ui+2 + 8ui+1 − 8ui−1 + ui−2
= + O(∆x)4 central difference
∂x i 12∆x

∂2u −ui+2 + 16ui+1 − 30ui + 16ui−1 − ui−2
= + O(∆x)4 central difference
∂x2 i 12(∆x) 2
Pros and cons of high-order difference schemes

⊖ more grid points, fill-in, considerable overhead cost
⊕ high resolution, reasonable accuracy on coarse grids
Criterion: total computational cost to achieve a prescribed accuracy
JetWings 36
Grid transformations
Purpose: to provide a simple treatment of curvilinear boundaries
d nozzle dire t mapping d re tangle

P
P
inverse mapping
y a b a b
physi al domain omputational domain
body-tted grid artesian grid
x
The original PDE must be rewritten in terms of (ξ, η) instead of (x, y) and
discretized in the computational domain rather than the physical one.
∂u ∂u ∂u ∂u
Derivative transformations , ,... −→ , ,...
∂x ∂y ∂ξ ∂η
| {z } | {z }
difficult to compute easy to compute
JetWings 37
Finite volume method
The finite volume method is based on (I)

Integral conservation law (I)
rather than (D). The integral conservation Z Z Z
law is enforced for small control volumes ∂
u dV + f · n dS = q dV
∂t V
defined by the computational mesh: S V
S n
f = vu − d∇u
N
[ V
V̄ = V̄i , Vi ∩ Vj = ∅, ∀i 6= j
dS f
flux function
i=1 Z Z Z
Z ∂u
1 dV + ∇ · f dV = q dV
ui = u dV mean value V ∂t V V
|Vi | Vi
Partial differential equation (D)
To be specified ∂u
+∇ · f = q in V
• concrete choice of control volumes ∂t
∂u
• type of approximation inside them In steady state ∂t
= 0 so that
• numerical methods for evaluation ∇ · (uv) = ∇ · (d∇u) + q

of integrals and fluxes
JetWings 38
Definition of control volumes
Vertex-centered FVM Cell-centered FVM
1D ui 1D ui
ui+1 ui+1
ui 1 ui 1
Vi Vi
xi 1 xi xi+1 xi 1 x i 1=2 xi
x i 1=2 xi+1=2
2D 2D
Vi
Vi Vi
Vi
Different grids / control volumes can be used for different variables (v, p, . . .)
JetWings 39
Discretization of local subproblems
Integral equation for a single finite volume

Z Z Z
∂ui 1 X 1 1
+ f · nk dS = qi , ui = u dV, qi = q dV
∂t |Vi | Sk |V i | Vi |Vi | Vi
k
• the integral conservation law is satisfied for each CV and for the entire domain
• to obtain a linear system, integrals must be expressed in terms of mean values
Numerical integration pn (x )
Z n
X
f (x) dV ≈ wi f (xi )
V i=0
where wi ≥ 0 are the weights and xi are

the nodes of the quadrature rule x0 xi xn
Such formulae can be derived by exact integration of an interpolation polynomial
JetWings 40
Newton-Cotes quadrature rules for intervals
x1 +x2
1D x12 = 2 , V = (x1 , x2 ), |V | = x2 − x1
x1 x12 x2
Z
Midpoint rule f (x) dV ≈ |V |f12 exact for f ∈ P1 (V )
V
Z
Trapezoidal rule f1 + f2 exact for f ∈ P1 (V )
f (x) dV ≈ |V |
V 2
Z
Simpson’s rule f1 + 4f12 + f2 exact for f ∈ P3 (V )
f (x) dV ≈ |V |
V 6
Numerical integration for quadrilaterals/hexahedra

y
1 use a mapping onto a unit square

V
V^ and apply 1D quadrature rules in
each coordinate direction
x 0 1
JetWings 41
Newton-Cotes quadrature rules for triangles

2D x3
Midpoints
x1 + x2 x1 + x3 x2 + x3
x13 x123 x23 x12 =
2
, x13 =
2
, x23 =
2
x1 + x2 + x3
Center of gravity x123 =
3
x1 x12 x2
Z
f (x) dV ≈ |V |f123 exact for f ∈ P1 (V )
V
Z
f1 + f2 + f3
f (x) dV ≈ |V | exact for f ∈ P1 (V )
V 3
Z
f12 + f23 + f13
f (x) dV ≈ |V | exact for f ∈ P2 (V )
V 3
Z
3(f1 + f2 + f3 ) + 8(f12 + f23 + f13 ) + 27f123
f (x) dV ≈ |V |
V 60
exact for f ∈ P3 (V )
JetWings 42
Newton-Cotes quadrature rules for tetrahedra

3D x4
Center of gravity
x1234
x3 x1 + x2 + x3 + x4
x1234 =
4
x1
x2
Z
f (x) dV ≈ |V |f1234 exact for f ∈ P1 (V )
V
Z
f1 + f2 + f3 + f4 exact for f ∈ P1 (V )
f (x) dV ≈ |V |
V 4
Z
f1 + f2 + f3 + f4 + 16f1234 exact for f ∈ P2 (V )
f (x) dV ≈ |V |
V 20
JetWings 43
Interpolation techniques
Problem: the solution is available only at computational nodes (CV centers)

Interpolation is needed to obtain the function values at quadrature points
1
R
Volume integrals ui = |Vi | Vi
u dV ≈ u(x̄i ) midpoint rule
1
P R
Surface integrals f = vu − d∇u ⇒ |Vi | f · nk dS = Ic + Id
k Sk
Z Z
1 X 1 X
Ic = (v · nk )u dS, Id = d(nk · ∇u) dS
|Vi | Sk |Vi | Sk
k k
Approximation of convective fluxes vertex-centered FVM

ui
1 ui+1
1D: |Vi | ≡ ∆x = N, xi = i∆x ui 1
ui+1/2 − ui−1/2 Vi
Ic = v , v = const
∆x
xi 1 xi xi+1
How to define the interface values ui±1/2 ? x i 1=2 xi+1=2
JetWings 44
Upwind difference approximation (UDS)
Piecewise-constant solution Upwind-biased interface values
ui 1=2 ui+1=2
v>0 ui−1/2 ≈ ui−1 , ui+1/2 ≈ ui
ui+1
ui ui − ui−1
Ic ≈ v backward difference
ui 1 ∆x
x v<0 ui−1/2 ≈ ui , ui+1/2 ≈ ui+1

xi 1 xi xi+1
ui+1 − ui
xi 1=2 xi+1=2 Ic ≈ v forward difference
∆x
Taylor series expansion

v∆x ∂u (∆x)2 ∂2u
vui+1/2 = vui + +v + ...
2 ∂x i 8 ∂x2 i
a first-order accurate flux approximation, the leading truncation error resembles

a diffusive flux d ∂u v∆x
∂x with d = 2 being the numerical diffusion coefficient
JetWings 45
Central difference approximation (CDS)
Piecewise-linear solution Interpolation polynomial

ui 1=2 ui+1=2 xR − x x − xL
p1 (x) = uL + uR
ui+1 xR − xL xR − xL
ui
Averaged interface values
ui 1
ui−1 + ui ui + ui+1
ui−1/2 ≈ , ui+1/2 ≈
x 2 2
xi 1 xi xi+1
ui+1 − ui−1
Ic ≈ v central difference
xi 1=2 xi+1=2
2∆x
Taylor series expansions

∆x ∂u (∆x)2 ∂2u
ui+1 = ui+1/2 + 2 ∂x i+1/2 + 8 ∂x2 + ...
i+1/2

∆x ∂u (∆x)2 ∂2u
ui = ui+1/2 − 2 ∂x i+1/2 + 8 ∂x2 − ...
i+1/2

ui + ui+1 (∆x)2 ∂2u
Hence, ui+1/2 = − + ... (second-order accuracy)
2 8 ∂x2 i+1/2
JetWings 46
Linear upwind difference scheme (LUDS)
Piecewise-linear solution Upwind-biased extrapolation

ui 1=2 ui+1=2 3ui−1 −ui−2
ui−1/2 ≈ 2
v ui+1
v>0 3ui −ui−1
ui ui+1/2 ≈ 2
ui 1 3ui − 4ui−1 + ui−2

ui 2 Ic ≈ v
2∆x
x
3ui −ui+1
xi 2 xi 1 xi xi+1 ui−1/2 ≈ 2
v<0 3ui+1 −ui+2
xi 1=2 xi+1=2 ui+1/2 ≈ 2
LUDS is second-order accurate, equivalent 3ui − 4ui+1 + ui+2

Ic ≈ −v
to the one-sided 3-point finite difference 2∆x
The matrix is no longer tridiagonal (shifted, upper/lower triangular if Id = 0)

(m+1) (m+1) (m) (m)
Defect correction: ILU DS = IU DS + [ILU DS − IU DS ], m = 0, 1, 2, . . .
JetWings 47
Quadratic upwind difference scheme (QUICK)
Quadratic Upwind Interpolation for Convective Kinematics

xR −x xR −x
p2 (x) = uL xx−x
L
M
−xM xR −xL + uM xx−x
M
L
−x x
L R −xM
+ u R x
x−xL x−xM
R −xL xR −xM
ui 1=2 ui+1=2
Upwind-biased interface values
v ui+1
ui 3ui +6ui−1 −ui−2
ui−1/2 ≈ 8
ui 1 v>0
ui 2 ui+1/2 ≈ 3ui+1 +6ui −ui−1
8
x
3ui+1 + 3ui − 7ui−1 + ui−2
xi 2 xi 1 xi xi+1
Ic ≈ v
8∆x
xi 1=2 xi+1=2
3ui−1 +6ui −ui+1
ui−1/2 ≈ 8
• third-order flux approximation v<0 3ui +6ui+1 −ui+2
ui+1/2 ≈ 8
• second-order overall accuracy
3ui−1 + 3ui − 7ui+1 + ui+2
(because of the midpoint rule) Ic ≈ −v
8∆x
• marginally better than LUDS
JetWings 48
Evaluation of surface integrals

Approximation of convective fluxes
• any second-order finite volume scheme is a linear combination of CDS and
LU DS approximations (e.g. IQU ICK = 34 ICDS + 14 ILU DS )
• high-order schemes can be readily derived by polynomial fitting based on
pm (x), m > 2 but pay off only if the quadrature rule matches their accuracy
• a high-order scheme is guaranteed to produce better results than a low-order
one only asymptotically i.e. for sufficiently fine meshes
Approximation of diffusive fluxes

∂u ui − ui−1 ∂u ui+1 − ui slopes of the
≈ , ≈
∂x i−1/2 ∆x ∂x i+1/2 ∆x straight lines
Second-order accurate central difference

∂u
∂u

d ∂x i+1/2 − d ∂x i−1/2 ui+1 − 2ui + ui−1
Id = − ≈ −d
∆x (∆x)2
JetWings 49
Finite element method
Origins: structural mechanics, calculus of variations for elliptic BVPs
Boundary value problem Minimization problem







Lu = f in Ω ? Given a functional J : V → R
 u=g
0 on Γ0 ⇔ find u ∈ V such that


 n · ∇u = g1 on Γ1 J(u) ≤ J(w), ∀w ∈ V


 n · ∇u + αu = g on Γ
2 2 subject to the imposed BC
⊕ the functional contains derivatives of lower order

⊕ solutions from a broader class of functions are admissible
⊕ boundary conditions for complex domains can be handled easily
⊖ sometimes there is no functional associated with the original BVP
Modern FEM: weighted residuals formulation (weak form of the PDE)
JetWings 50
Fundamentals of the FEM
The Finite Element Method is a systematic approach to generating

piecewise-polynomial basis functions with favorable properties
• The computational domain Ω is subdivided into many
S
small subdomains K called elements: Ω̄ = K∈Th K.
• The triangulation Th is admissible if the intersection of
any two elements is either an empty set or a common
vertex / edge / face of the mesh.
Figure
• The finite element subspace Vh4:consists
Two nodal
of basis functions for the polygonal
piecewise-polynomial domain example
functions.
Typically, Vh = {v ∈ C m (Ω) : v|K ∈ Pk , ∀K ∈ Th }.
comprising the right hand side as well as the entries
• Any function v ∈ Vh is uniquely determined by Za finite number of degrees
a = a( ;at )certain
of freedom (function values or derivatives
i;j = rpoints
j r dx;
i i; jnodes).
called i= 1; : : : N;
j (12)

of the
• Each basis function ϕi nite element system
accommodates exactly A. degree
matrixone This is where the advantage
of freedom and of the nodal basis
function
has a small support becomes
so that apparent: the
the resulting above integrals
matrices reduce to integrals over supp( ) in the
are sparse. i
rst case and supp( ) \ supp( ) in the second. The supports of two nodal basis functions
i j
for linear triangles consists of all those triangles of which the corresponding node is a
vertex. The supports of two basis functions in our example are shown in Figure 5.
JetWings 51
Finite element approximation
The finite element is a triple (K, P, Σ), where

• K is a closed subset of Ω̄
• P is the polynomial space for the shape functions
• Σ is the set of local degrees of freedom
ϕi
Basis functions possess the property

 1 if i = j
ϕj (xi ) = δij =
 0 if i 6= j
x0 x i−1 xi x i+1 xN
N
X N
X N
X
uh (x) = uj ϕj (x) ⇒ uh (xi ) = uj ϕj (xi ) = uj δij = ui
j=1 j=1 j=1
Approximate solution: the nodal values u1 , . . . , uN can be computed by the

Ritz method provided that there exists an equivalent minimization problem
JetWings 52
Least-squares method
Idea: minimize the residual of the PDE
R(w) = Lw − f such that R(u) = 0 ⇒ Lu = f
R
Least-squares functional J(w) = Ω
(Lw − f )2 dx always exists
Necessary condition of an extremum

Z
d d
J(u + λv) = (L(u + λv) − f )2 dx =0
dλ λ=0 dλ Ω λ=0
R R R
Integration by parts: Ω
(Lu − f )Lv dx = Ω
L∗ (Lu − f )v dx − Γ
[. . .] ds = 0
Euler-Lagrange equation L∗ Lu = L∗ f where L∗ is the adjoint operator
• corresponds to a derivative of the original PDE

• requires additional boundary conditions and extra smoothness
• it makes sense to rewrite a high-order PDE as a first-order system
Advantage: the matrices for a least-squares discretization are symmetric
JetWings 53
Weighted residuals formulation
Idea: render the residual orthogonal to a space of test functions


∞
P  Lu = f in Ω
Let u = αj ϕj ∈ V0 be the solution of
j=1  u=0 on Γ
Residual is zero if its projection onto each basis function equals zero
R
Lu − f = 0 ⇔ Ω
(Lu − f )ϕi dx = 0 ∀i = 1, 2, . . .
∞
P R
Test functions v= βj ϕj ⇒ Ω
(Lu − f )v dx = 0, ∀v ∈ V0
j=1
Weak formulation: find u ∈ V0 such that a(u, v) = l(v) ∀v ∈ V0

R R
where a(u, v) = Ω
Lu v dx is a bilinear form and l(v) = Ω
f v dx
R
Integration by parts: Lu = ∇ · g(u) ⇒ a(u, v) = − Ω
g(u) · ∇v dx
JetWings 54
Finite element discretization

Continuous problem Discrete problem
Find u ∈ V0 such that Find uh ∈ Vh ⊂ V0 such that

a(u, v) = l(v), ∀v ∈ V0 a(uh , vh ) = l(vh ), ∀vh ∈ Vh′
N
X N
X
FEM approximations: uh = u j ϕj ∈ V h , vh = vj ψj ∈ Vh′
j=1 j=1
where Vh = span{ϕ1 , . . . , ϕN } and Vh′ = span{ψ1 , . . . , ψN } may differ
(Bubnov-)Galerkin method Vh′ = Vh → a(u, ϕi ) = l(ϕi ), ψi = ϕi

Petrov-Galerkin method Vh′ 6= Vh → a(u, ψi ) = l(ψi ), ψi 6= ϕi
N
X
Linear algebraic system a(ϕj , ψi )uj = l(ψi ), ∀i = 1, . . . , N
j=1
Matrix form Au = F with coefficients aij = a(ϕj , ψi ), Fi = l(ψi )
JetWings 55

Lecture # 02 Goverening Equations and Its Discritization

Uploaded by

Copyright:

Available Formats

Lecture # 02 Goverening Equations and Its Discritization

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Lecture # 02 Goverening Equations and Its Discritization

Uploaded by

Copyright:

Available Formats

Lecture # 02 Goverening Equations and its Discritization

Getting started: CFD notation

scalar unknowns u = u(x, t), x ∈ Rn , t ∈ R, n = 1, 2, 3

Tensorial quantities in fluid dynamics

Remark. The trace (sum of diagonal elements) of ∇v equals ∇ · v.

Deformation rate tensor (symmetric part of ∇v)

Spin tensor S(v) = ∇v − D(v) (skew-symmetric part of ∇v)

Vector multiplication rules

Scalar product of two vectors

Dyadic product of two vectors

Elementary tensor calculus

2. T 1 + T 2 = {t1ij + t2ij }, T 1 , T 2 ∈ R3×3 , a ∈ R3

Divergence theorem of Gauß

Let Ω ∈ R3 and n be the outward unit normal to the boundary Γ = Ω̄\Ω.

Example. A sphere: Ω = {x ∈ R3 : ||x|| < 1}, Γ = {x ∈ R3 : ||x|| = 1}

Consider f (x) = x so that ∇ · f ≡ 3 in Ω and n = x

Governing equations of fluid dynamics

Physical principles Mathematical equations

It is important to understand the meaning and significance of each equation

Description of fluid motion z

Eulerian monitor the flow characteristics (x 1 ; y1 ; z1 )

in a fixed control volume k (x 0 ; y0 ; z0 )

Description of fluid motion

Definition. A streamline is a curve which is tangent to the velocity vector

Flow models and reference frames

fixed CV of a finite size moving CV of a finite size

fixed infinitesimal CV moving infinitesimal CV

Good news: all flow models lead to the same equations

Eulerian vs. Lagrangian viewpoint

Reynolds transport theorem

rate of change in rate of change in convective transfer

Derivation of the governing equations

Modeling philosophy Generic conservation law

Derivation of the continuity equation

Physical principle: conservation of mass

accumulation of mass inside CV = net influx through the surface

Divergence theorem yields Continuity equation

Physical principle: f = ma (Newton’s second law)

total force f = ρg dV + h dS, where h=σ·n

For a newtonian fluid viscous stress is proportional to velocity gradients:

Normal stress: stretching Shear stress: deformation

Derivation of the momentum equations

Newton’s law for a moving volume

Transformation of surface integrals

dS δe accumulation of internal energy

Heating: s = ρq dV − fq dS Fourier’s law of heat conduction

Work done per unit time = total force × velocity

Derivation of the energy equation

Integral conservation law for a moving volume

Transformation of surface integrals

where ∇ · (σ · v) = −∇ · (pv) + ∇ · (τ · v) = −∇ · (pv) + v · (∇ · τ ) + ∇v : τ

Different forms of the energy equation

Total energy equation

Summary of the governing equations

1. Continuity equation / conservation of mass

2. Momentum equations / Newton’s second law

3. Energy equation / first law of thermodynamics

Conservation form of the governing equations

Conservative variables, fluxes and sources