Mathematical and Physical Behavior of Equations

Dr. Zaib Ali

 Classification of Differential Equations

• ODE (Ordinary Differential Equation) - contains functions and derivatives of only

One Single Variable

• PDE (Partial Differential Equation) - contains functions and derivatives of Several

Independent Variables.

• Linear differential equations: - equation does not contain product of the dependent
variable and its derivative. Individual solutions to such equations can be
superimposed
• Non-linear differential equations: - equation does contain product of the dependent
variable and its derivative. Individual solutions to such equations cannot be
superimposed

• The order of a differential equation represents the highest order of a derivative (and
not that of the independent variables) appearing in it
 Classification of Differential Equations

• The mathematical behavior of a PDE can be either:

– Elliptic
– Parabolic
– Hyperbolic

• The knowledge of the mathematical behavior for a particular problem is of crucial

importance in selecting the appropriate numerical solution algorithm.

• Physical Behavior:
– Equilibrium problems
– Marching problems
– Eigenvalue Problems
 Determining the nature of PDE’s

• NS equations are system second order non-linear PDEs in four independent

variables.
• Many partial differential equations can be classified as either hyperbolic, parabolic,
or elliptic or mixed type.
• Consider a fairly simple system of quasi-linear equations:

• u and v represent a continuous velocity field throughout the xy space.

• the derivatives of u and v are finite values at this given point.
 Cramer’s Rule

• Consider any point in the xy plane, such as point P.

• Let us seek the lines (or directions) through this point (if any exist) along with the
derivatives of u and v are indeterminant and across which may be discontinuous.
• These special lines that we are seeking are called characteristic lines.
 Cramer’s Rule

• To find such lines, we recall that u and v are continuous functions of x and y and
write their total differentials as

• Above four Equations constitute a system of four linear equations with four
unknowns. These equations can be written in matrix form as:
 Cramer’s Rule

• Let [A] denote the coefficient matrix.

• Let us solve for the unknown ∂u/ ∂x, using Cramer's rule.
 Cramer’s Rule

• What happens if we choose to move in a direction away from point P such that [A]
in the above Eqn is zero?
• such characteristic lines indeed do exist, and we can find them by setting

• Expanding the determinant

• Divide by
 Cramer’s Rule

• This Equation is a quadratic equation in dy/dx. For any point in the xy plane, the
solution of Eq. will give the slopes of the lines along which the derivatives of u and
v are indeterminant.

• Direction of the characteristic lines through a given point in the xy plane

 Cramer’s Rule

• If D > 0 Two real and distinct characteristics exist through each point in the xy
plane. The original system of equations is called hyperbolic.

• If D = 0 Here the system of Eqs. is called parabolic.

• If D < 0 The characteristic lines are imaginary. The system of Eqs. is called
elliptic.

• Direct analogy with the general equation for a conic section from analytic geometry
 The Eigenvalue Method

• f1 and f2 are zero for simplicity

 The Eigenvalue Method

• The eigenvalues of [N] determine the classification of the system.

• If the eigenvalues are all real, the equations are hyperbolic.
• If the eigenvalues are all complex, the equations are elliptic.
• If λ = 0 or λ = ∞, the equations are parabolic.
• The eigenvalues may be a mix of both real and complex values. The mathematical
behavior of such equations then exhibits a mixed hyperbolic-elliptic nature.

• See Example 3.1 (J. D. Anderson, 1995)

 Physical Behavior of Different Types of Eqns.

• Each type of equation has a different mathematical behavior, and this reflects
different physical behavior of the flow fields as well.
• In tum, this implies that different computational methods should be used for solving
equations associated with the different classifications.

Hyperbolic Equations
• Two real characteristic
• Region of influence
• Domain of dependence
• Forward Marching
• Marching Solutions

Domain and boundaries for the solution of hyperbolic

equations. Two-dimensional steady flow.
Steady Inviscid Supersonic Flow

• Euler Eqns. Hyperbolic when M > 1

• Space Marching
Unsteady Inviscid Flow

• Euler Eqns. Subsonic or Subsonic

• Time Marching
Parabolic Equations

• One real characteristic

• Region of influence
• Forward Marching
• Marching Solutions
Steady Boundary-layer Flows

• Boundary Layer Eqns.

• Marching technique;
starting from initial
data at the nose of the
body.
 Parabolized Viscous Flows

• Boundary Layer Eqns. not valid

• If all the viscous terms in that involve derivatives in the streamwise direction are
neglected and the flow is assumed study, we can get what are called: Parabolized
Navier-Stokes Equations (PNS).
• PNS are simpler, i.e., contain less terms, than the full Navier-Stokes equations
and and they can be solved by means of a downstream marching procedure.
Elliptic Equations

• No real characteristic
• No Region of influence
• No domain of dependence
• Information propagated everywhere
• Simultaneous solutions
• Jury problems: boundary conditions
specified over whole boundary
Steady Subsonic Inviscid Flow

Similarly, Incompressible Inviscid

Flow also exhibits Elliptic behavior
 Boundary and initial conditions

• A PDE with spatial derivatives requires boundary conditions while a PDE with a
time derivative requires an initial condition.
• Boundary conditions mean defining the dependent variables and their derivatives
along the boundaries of the domain of the PDE.
• Initial condition means defining the time dependent variables at some initial state
(t = 0)

• Dirichlet boundary condition: - in which the dependent variables themselves are

prescribed along the domain boundary.
• Von Neumann boundary condition: - in which the normal gradient of the dependent
variables is prescribed along the boundary.
• Robin boundary condition: - in which the boundary conditions are a linear
combination of the Dirichlet and Von Neumann type.
• Mixed boundary conditions: - in which certain portions of the boundary are defined
as Dirichlet type, while others as Von Neumann type.
 Equations Classification

Classification of the main categories of fluid flow

 Flow Equations

• Flows involving both subsonic and supersonic flow regions – which is typical of
compressible high-speed flows – there is need to deal with both elliptic and hyperbolic
behavior at the same time.
• However, an elliptic solution algorithm cannot deal with the supersonic flow region, while a
steady hyperbolic solution algorithm is unsuitable for the subsonic one.
 Physical Behavior

Subsonic Flow ( M < 1)

Elliptic Behavior
(information propagates
both upstream and downstream)

Sonic Flow ( M = 1)
Parabolic Behavior
(information
propagates
downstream only,
behind the front)

Supersonic Flow( M < 1)

Hyperbolic Behavior
(information propagates
downstream only, and inside
the Mach cone)