Introduction to partial differential equations

in fluid mechanics: their types and some analytical solutions

Numerical Fluid
analysis Mechanics

CFD

Computer
Science
 Typical types of partial differential equations

Classification of Partial Differential Equations: consider the case of

a second-order differential equation given in the general form as

∂2φ ∂2φ ∂2φ ∂φ ∂φ

A
2
+B
∂x∂y
+C
2
+D
∂x
+E
∂y
(
+ F φ = G x, y )
∂x ∂y
If the coefficients A, B, C, D, E, and F are either constants or functions of only (x, y) (do
not contain φ or its derivatives), the PDE is said to be a linear equation;

Otherwise it is a non-linear equation.

An important subclass of non-linear equations is quasilinear equations.

In this case, the coefficients may contain φ or its first derivative but not the second
(highest) derivative.
Classification of Partial Differential Equations:

∂2φ ∂2φ ∂2φ ∂φ ∂φ

A 2 +B +C 2 + D +E + Fφ = G ( x, y )
∂x ∂x∂y ∂y ∂x ∂y

if B 2 − 4AC = 0, the equation is Parabolic

if B 2 − 4AC < 0, the equation is Elliptic
if B 2 − 4AC > 0, the equation is Hyperbolic
Unsteady Navier-Strokes equations are elliptic in space and parabolic in time.

At steady-state, the Navier-Strokes equations are elliptic.

In Elliptic problems, the boundary conditions must be applied on all confining surfaces.
These are Boundary Value Problems .
Some examples
Heat transfer in a solid medium (no flow)
! " %
∂e ∂e ∂ $ ∂T '
ρ + ρU j = −P∇ ⋅ U + ρε + k
∂t ∂x j ∂x j $# ∂x j '&
⇒
" %
∂e ∂ $ ∂T '
ρ = k
∂t ∂x j # ∂x j '&
$

e = CT C is the specific heat,

Assume ρ ,C ,k are all constants
k
define thermal diffusivity α =
ρC
⇒
! !
( ) =α
∂T x,t 2
∂ T
or
( ) = α∇2T
∂T x,t
∂t ∂x j∂x j ∂t
 Special cases
1D transient heat transfer problem 2D steady-state heat transfer problem

! !
( ) = α∇2T
∂T x,t ( ) = α∇2T
∂T x,t
∂t ∂t
⇒ ⇒
( ) = α ∂2T
∂T x,t ∂2T ∂2T
0= + 2D Laplace equation
∂t 2 2 2
∂x ∂x ∂y

Transient problem: Equilibrium problem:

Solution depends on both the Solution is fully determined by the
initial condition and boundary boundary conditions
conditions
Parabolic PDE Elliptic PDE
An example of analytical solution for elliptic PDE
2D steady-state heat transfer problem
∂2T ∂2T
0= + 2D Laplace equation
2 2
∂x ∂y
0 ≤ x ≤ L, 0≤ y≤ L
T ( x,0) = T0
T ( 0, y ) = T ( L, y ) = T ( x, L ) = 0

Analytical solution by the method of separations of variables

+ L− y .
T ( x, y ) = ∑
4T0 $ x' (
sin & nπ ) sinh -nπ
) 0
( )
n=1,3,5,.... nπ sinh nπ % L( -, L 0/
Consistency and well-posedness
∂2T ∂2T
+ =0 in a 2D disk r = x2 + y2 ≤ R
2 2
∂x ∂y
∂T
Boundary condition:
∂r
(
r = R,θ ) = f θ ( )

We can solve in terms of r and angle 0 ≤ θ ≤ 2π

⇒ First write in cylindrical coordinates
1 ∂ $ ∂T ' 1 ∂2T
&r )+ =0
r ∂r % ∂r ( r 2 ∂θ 2
r d $ dA ' 1 ∂2 B
T =A r B θ () ( )
⇒ & r ) = − = n2 , n = 1,2,3,...
A dr % dr ( B ∂θ 2

d2A dA
r2 +r − n2 A = 0 ⇒ A = rn
2 dr
dr
∞
The solusion is then: T r,θ = a0 + ∑ r n +,an cos nθ + bn sin nθ -.
( ) ( ) ( )
n=1
θ =2 π ∂T θ =2 π ∂T
Divergence Theorem: 0 = ∫∫ ∇ 2T dA dz = ∫ R dθ dz ⇒ ∫ dθ = 0
V θ =0 ∂r θ =0 ∂r
θ =2 π
⇒ ∫
θ =0
( )
f θ dθ = 0 A necessary constraint for the boundary condition
 1D Transient heat transfer (Parabolic PDE)

𝜕𝑇 𝑥, 𝑡 𝜕 ! 𝑇 𝑥, 𝑡
=𝛼 , 0 ≤ 𝑥 ≤ 𝐿, 𝑡≥0
𝜕𝑡 𝜕𝑥 !

Initial condition: 𝑇 𝑥, 𝑡 = 0 = 𝑇"

Boundary conditions: 𝑇 𝑥 = 0, 𝑡 = 𝑇# , 𝑇 𝑥 = 𝐿, 𝑡 = 𝑇#

The analytical solution (by the method of separation of variables) is:

'
4 ! !
𝛼𝑡 𝑥
𝑇 𝑥, 𝑡 = 𝑇# + 𝑇" − 𝑇# . exp − 2𝑘 − 1 𝜋 ! sin 2𝑘 − 1 𝜋
𝜋 2𝑘 − 1 𝐿 𝐿
$%&

The normalized form

'
𝑇 𝑥, 𝑡 − 𝑇# 𝑥 𝛼𝑡 4 𝛼𝑡 𝑥
=𝐹 , ! =. exp − 2𝑘 − 1 ! 𝜋 ! ! sin 2𝑘 − 1 𝜋
𝑇" − 𝑇# 𝐿 𝐿 𝜋 2𝑘 − 1 𝐿 𝐿
$%&
A code to compute the analytical solution: temperature.f90

The normalized form

'
𝑇 𝑥, 𝑡 − 𝑇! 𝑥 𝛼𝑡 4 𝛼𝑡 𝑥
=𝐹 , # =* exp − 2𝑘 − 1 #𝜋 # # sin 2𝑘 − 1 𝜋
𝑇" − 𝑇! 𝐿 𝐿 𝜋 2𝑘 − 1 𝐿 𝐿
$%&
Tn1 = 0.0
Program HeatDiffusion Tn2 = 0.0
real pi,L,alpha,T0,TF Tn3 = 0.0
real t1,t2,t3,t4,t5,x Tn4 = 0.0
real Tn1,Tn2,Tn3,Tn4,Tn5 Tn5 = 0.0
integer k,ix do k=1,500
pi = 4.*atan(1.0) ! pi=3.14159.... c3 = (2.*k - 1.0)*pi
T0 = 100.0 c2 = c3*c3
TF = 20.0 c1 = 4./c3
alpha = 1.0 Tn1 = Tn1 + c1*exp(-c2*t1)*sin(c3*x)
L = 1.0 Tn2 = Tn2 + c1*exp(-c2*t2)*sin(c3*x)
t1 = 0.001*alpha/L**2 Tn3 = Tn3 + c1*exp(-c2*t3)*sin(c3*x)
t2 = 0.01*alpha/L**2 Tn4 = Tn4 + c1*exp(-c2*t4)*sin(c3*x)
t3 = 0.1*alpha/L**2 Tn5 = Tn5 + c1*exp(-c2*t5)*sin(c3*x)
t4 = 0.2*alpha/L**2 end do
t5 = 0.4*alpha/L**2
write(10,100) x, Tn1, Tn2, Tn3, Tn4, Tn5
do ix = 0,100 end do
x = 0.01*ix/L 100 format(2x,6F12.6)

End Program HeatDiffusion

9
The plot was done by NCL: demo4.ncl
10
1D linear horizontal acoustic wave equation
No viscous friction, isentropic flow [First assumption]
p 1 p
p = ρ RT , T = , dT = dp − dρ
ρR ρR ρ2R
γ
C dT R C C p ! ρ $ C
ds = v − d ρ = v dp − p d ρ ≈ 0 ⇒ = ## && γ= p
T ρ p ρ p0 " ρ0 % Cv
Small density and pressure fluctuations [Second assumption]
p − p0 ρ − ρ0 dp dρ
<< 1, << 1 ⇒ =γ
p0 ρ0 p0 ρ0 The speed of sound
Continuity and momentum equations cs = γ RT0
∂ρ ∂ ρ u
+
( )
= 0 ⇒ Linearize
∂ρ
+ ρ0
∂u
=0
∂t ∂x ∂t ∂x
∂u ∂u ∂p ∂u ∂p
ρ + ρu + = 0 ⇒ Linearize ρ0 + =0
∂t ∂x ∂x ∂t ∂x
Eliminate u from the above two
∂2 ρ ∂2 p ∂2 ρ γ p0 ∂2 ρ ∂2 ρ ∂2 ρ General solutions
= ⇒ = = γ RT0 = cs2
∂t 2 ∂x 2 ∂t 2 ρ0 ∂x 2 ∂x 2 ∂x 2 ρ = FLeft ( x + cst ) + FRight ( x − cst )
∂2 ρ ∂2 ρ
− cs2 =0
2
∂t ∂x 2
Hyperbolic PDF
1D acoustic wave

∂2 ρ ∂2 ρ
− cs2 =0 − ∞ ≤ x ≤ +∞, t ≥ 0
2
∂t ∂x 2
Initial conditiona: ρ x,0 − ρ0 = f x , ∂t !" ρ x,0 − ρ0 #$ = g x
( ) ( ) ( ) ( )
The solution is
x+cst
1 1
ρ x,0 − ρ0 = !" f x + cst + f x − cst #$ +
( ) ( ) ( ) ( )
∫ g τ dτ
2 2cs x−cst

At any given time, the solution at x p is determined by initial data within

x p − cs t ≤ x ≤ x p + cs t
Types of equations in CFD (Computational Fluid Dynamics)
Laplace Equation
∂2φ ∂2φ
+ =0
∂x 2 ∂y 2

Poisson Equation
∂2φ ∂2φ
+ = S(x, y)
∂x 2 ∂y 2

Laplace equations and Poisson equations are elliptic equations and generally
associated with the steady-state problems.

The velocity potential in steady, inviscid, incompressible, and irrotational flows satisfies
the Laplace equation.

The temperature distribution for steady-state, constant-property, two-dimensional

condition satisfies the Laplace equation if no volumetric heat source is present.
Fluid flow problems
generally have nonlinear terms due to the inertia or acceleration
component in the momentum equation. These terms are called advection terms. The
energy equation has nearly similar terms, usually called the convection terms, which
involve the motion of the flow field. For unsteady two-dimensional problems, the
appropriate equation can be represented as
∂φ ∂φ ∂φ " ∂2φ ∂2φ %
+u +v = B$ 2 + 2 ' + S
∂t ∂x ∂y # ∂x ∂y &
φ denotes velocity, temperature or some other transported property,
u and v are velocity components,
B is the diffusivity for momentum or heat, and
S is a source term.

The pressure gradients in the momentum or the volumetric heating in the energy
equation can be appropriately substituted in S.

The above PDE is parabolic in time and elliptic in space.

For very high-speed flows, the terms on the left side dominate, the second-order terms on
the right hand side become trivial, and the equation become hyperbolic in time and space.
The 2D Euler equation

∂u ∂u ∂u 1 ∂p
+u +v =− + fx
∂t ∂x ∂y ρ ∂x
∂v ∂v ∂v 1 ∂p
+u +v =− + fy
∂t ∂x ∂y ρ ∂y

∂φ ∂φ
u= , v=−
∂y ∂x
To eliminate pressure
∂ # ∂u ∂v & ∂u ∂u ∂2 u ∂u ∂v ∂2u ∂u ∂v ∂2 v ∂v ∂v ∂2 v ∂f x ∂f y
% − ( + + u + + v − − u − − v = −
∂t $ ∂y ∂x ' ∂y ∂x ∂x∂y ∂y ∂y 2 ∂x ∂x 2 ∂x ∂y ∂x∂y ∂y ∂x
∂y ∂x

∂2 u ∂2 u ∂f ∂f
→v +u + ....... = x − y B 2 − 4AC = u 2 > 0, the equation is Hyperbolic in space
∂y 2 ∂x∂y ∂y ∂x
 Comp. Math.
&
A more general & rigorous
Applications
definition
Hyperbolic of hyperbolic
conservation lawssystem with more unknowns
Jingmei Qiu

Outline Di↵erential form of conservation laws:

HCL
ut + rx · f(u) = 0, on ⌦
Summary
Comp. Math. (1)
& I .C . : u(x, t = 0) = u0 (x), on ⌦.
Applications
Hyperbolic: 1D system
Jingmei Qiu m = the number of components in 𝑢
• u(x, t): conserved quantities
Outline
The 1-D system (1) is hyperbolic if the m ⇥ m Jacobian matrix
HCL
Rd ⇥ R+ ! Rm ,
0 @f1 @f1 1
Summary
@u1 ··· @um
@fof the@problem;
• d: the 0dimension
A
f (u)
• m: the = of components
number = in· u~· .·
@u @fm @fm
• f(u): flux functions @u1 · · · @um

of the flux function has theRfollowing

m
! Rm property: For each value
of u,
• the eigenvalues of f 0 (u) are real,
8 / 53
• the matrix is diagonalizable, i.e. there is a complete set of
m linearly independent eigenvectors.
Examine the 1D linearized, isentropic acoustic wave eqns again

Recall the governing equations

𝜕𝜌 𝜕 𝜌" 𝑢 This is a 2-unknown system

+ =0 𝜌" 𝑢
𝜕𝑡 𝜕𝑥 𝜌 (
𝑈= , 𝐹 = 𝑝" 𝜌
𝑢
𝜕𝑢 1 𝜕𝑝 𝜌" 𝜌"
+ =0 Jacobian
𝜕𝑡 𝜌" 𝜕𝑥
0 𝜌"
𝜕𝐹 𝛾𝑝 𝛾𝑝
𝜌 ( = 0 → 𝜆&,! =±
𝑝 = 𝑝" 𝜕𝑈 𝜌" 𝜌 𝜌
𝜌"
= ± 𝛾𝑅𝑇 = ±𝑐*

Then same results! Two real eigenvalues à A hyperbolic system.

 Example: the 1D Euler system

Governing equations

Note:
!
3 − 𝛾 𝜌𝑢 !
𝜌𝑢 + 𝑝 = + 𝛾 − 1 𝜌𝐸
2 𝜌
𝜌𝑢! 𝜌𝑢 𝜌𝑢 # 1 𝛾 − 1 𝜌𝑢!
𝜌𝑢𝐻 = 𝑢 + 𝜌𝐶" 𝑇 + 𝑝 = 𝛾 𝜌𝐸 − 𝛾 − 1 , 𝜌𝐸 = 𝜌𝐻 +
2 𝜌 2𝜌! 𝛾 𝛾 2
Boundary conditions

The spatial boundary conditions in flow and heat transfer problems are of three general
types. They may stated as
on boundary: φ = φ1 ( x ) → Dirichlet BC

∂φ
on boundary: = φ 2 ( x ) → Neumann BC
∂n

! ! ∂φ
on boundary: a ( x ) φ + b ( x ) = φ3 ( x ) → mixed boundary conditions
∂n

∂φ ! ∂φ ∂φ
where the normal gradient is = n ⋅ ∇φ = nx + ny
∂n ∂x ∂y
Summary

• With respect to two independent variables, PDE can be classified into

Elliptic
Parabolic
Hyperbolic
Their solutions have different characteristics.

• Boundary conditions must be consistent with the governing PDE

• Linear PDE can often be solved analytically by the method of separation of variables

• The Navier-Stokes equations contain more than two independent variables as well as
more than one unknown fields, the nature of PDE depends on which two independent
variables are selected, or whether the flow is inviscid or not. The nature may also
change from one spatial region to another. For some special cases, the nature can be
shown explicitly.