1 AEROSPACE AE2307: Finite Element

Method
Lecture 2: Mathematical Formulation of FEM

2.1 Governing Equations (Partial Differential Equations)

2.1.1 Introduction to Governing Equations

In engineering, physical phenomena such as structural deformation, heat transfer, fluid flow, and
electromagnetic fields are modeled using partial differential equations (PDEs). These
equations describe how variables like displacement, temperature, and velocity change with
respect to time and space.

General Form of PDEs:

L(u)=f\mathcal{L}(u) = fL(u)=f

Where:

• L\mathcal{L}L is a differential operator (e.g., Laplacian, gradient, divergence),

• uuu is the dependent variable (e.g., displacement, temperature),
• fff is the source term (e.g., external force, heat source).

2.1.2 Examples of Governing Equations

1. Structural Mechanics:
o Stress-strain relationship: Describes the behavior of materials under load.
o Equilibrium equation: Balances internal stresses with external forces.

∇⋅σ+f=0\nabla \cdot \sigma + f = 0∇⋅σ+f=0

Where:

σ\sigmaσ is the stress tensor,

o
o
fff represents body forces (e.g., gravity).
2. Heat Transfer:
o Fourier’s law: Describes how heat flows through a material.

∇⋅(k∇T)+Q=ρc∂T∂t\nabla \cdot (k \nabla T) + Q = \rho c \frac{\partial T}{\partial

t}∇⋅(k∇T)+Q=ρc∂t∂T
 Where:

o TTT is temperature,
o kkk is thermal conductivity,
o QQQ is heat generation,
o ρ\rhoρ is density, ccc is specific heat.
3. Fluid Mechanics:
o Navier-Stokes equation: Governs fluid motion.

ρ(∂u∂t+u⋅∇u)=−∇p+μ∇2u+f\rho \left( \frac{\partial u}{\partial t} + u \cdot \nabla u \right)

= -\nabla p + \mu \nabla^2 u + fρ(∂t∂u+u⋅∇u)=−∇p+μ∇2u+f

Where:

o uuu is velocity,
o ppp is pressure,
o μ\muμ is dynamic viscosity.

2.2 Variational Principles and Weak Formulation

2.2.1 Variational Principles

Many governing equations can be derived from variational principles, where the physical
system seeks to minimize a certain functional (e.g., energy, work).

1. Principle of Minimum Potential Energy (Structural Mechanics):

o The actual displacement configuration of a structure minimizes the total potential
energy of the system.

Π(u)=∫V12σ⋅ϵ dV−∫Vf⋅u dV\Pi(u) = \int_V \frac{1}{2} \sigma \cdot \epsilon \, dV -

\int_V f \cdot u \, dVΠ(u)=∫V21σ⋅ϵdV−∫Vf⋅udV

Where:

o Π(u)\Pi(u)Π(u) is the potential energy functional,

o ϵ\epsilonϵ is the strain tensor.
2. Galerkin’s Method:
o Used to approximate solutions by projecting the governing equations onto a finite
set of basis functions.
o Key step in finite element formulation.

2.2.2 Weak Formulation

The weak form of a PDE is obtained by multiplying the governing equation by a test function
and integrating over the domain. It relaxes the continuity requirements for the solution, making it
suitable for numerical methods like FEM.

For a PDE: L(u)=f\mathcal{L}(u) = fL(u)=f

The weak form is: ∫VwL(u) dV=∫Vwf dV\int_V w \mathcal{L}(u) \, dV = \int_V w f \, dV∫V
wL(u)dV=∫VwfdV

Where www is the test function.

Advantages of Weak Form:

• Allows discontinuities at element boundaries.

• Reduces the order of the derivatives, making it easier to solve.

Example: Weak Form of Structural Equilibrium

The strong form of the equilibrium equation is: ∇⋅σ+f=0\nabla \cdot \sigma + f = 0∇⋅σ+f=0

The weak form is obtained by multiplying by a test function vvv and integrating:

∫Vv(∇⋅σ) dV=∫Vvf dV\int_V v (\nabla \cdot \sigma) \, dV = \int_V v f \, dV∫Vv(∇⋅σ)dV=∫VvfdV

Applying the divergence theorem:

∫Vσ:∇v dV=∫Vvf dV+∫Sv t dS\int_V \sigma : \nabla v \, dV = \int_V v f \, dV + \int_S v \, t \,

dS∫Vσ:∇vdV=∫VvfdV+∫SvtdS

Where ttt is the traction on the surface SSS.

2.3 Finite Element Discretization (1D, 2D, 3D Elements)

2.3.1 Discretization of the Domain

The finite element method discretizes the problem domain into smaller, simpler parts called
finite elements. These elements can be one-dimensional (1D), two-dimensional (2D), or three-
dimensional (3D).

1. 1D Elements:
o Used for problems like beams, trusses, and bars.
o Shape functions are simple linear or quadratic polynomials.

Example:
 o
A beam element with two nodes, each having one degree of freedom (DOF) for
displacement.
2. 2D Elements:
o Used for plane stress/strain problems, membranes, and shells.
o Typical elements include triangular (T3) and quadrilateral (Q4) elements.

Example:

o
A quadrilateral element with four nodes, each node having two DOFs
(displacement in xxx and yyy).
3. 3D Elements:
o Used for volumetric problems like solid mechanics.
o Elements can be tetrahedrons or hexahedrons.

Example:

o A tetrahedral element with four nodes, each node having three DOFs
(displacement in xxx, yyy, and zzz).

2.3.2 Element Shape Functions

Each finite element has associated shape functions that define how the solution varies within the
element.

• Linear Shape Functions: Provide a piecewise linear approximation, suitable for simple
problems.
• Quadratic or Higher-Order Shape Functions: Provide a more accurate representation
of the solution, especially for complex geometries.

2.3.3 Global Assembly

1. Element Matrices: Each element has its own stiffness matrix (for structural problems)
or conductivity matrix (for thermal problems), which relates nodal forces to nodal
displacements or fluxes to temperatures.
2. Global System: The local element matrices are assembled into a global system matrix
that represents the entire problem domain.

The finite element equation for a typical structural problem becomes: Ku=f\mathbf{K}
\mathbf{u} = \mathbf{f}Ku=f

Where:

• K\mathbf{K}K is the global stiffness matrix,

• u\mathbf{u}u is the vector of unknown nodal displacements,
• f\mathbf{f}f is the vector of applied forces.
2.4 Linear vs. Non-linear Analysis

2.4.1 Linear Analysis

In linear FEM, the material behavior, geometry, and boundary conditions are assumed to be
linear. This implies that:

• Material properties do not change with deformation (Hooke’s Law applies).

• Small deformations: Displacements are small, and the equilibrium equations remain
linear.

Linear FEM can be applied to simple, small-deformation problems such as:

• Elastic stress analysis of beams and plates.

• Static thermal problems without temperature-dependent material properties.

Characteristics:

• Simple to implement.
• Fast to compute.
• Suitable for small deformation problems.

2.4.2 Non-linear Analysis

In non-linear FEM, one or more assumptions of linearity are relaxed. This leads to more
accurate but computationally intensive solutions. Non-linearities can be classified into:

1. Geometric Non-linearity:
o Arises when large deformations occur, and changes in the geometry significantly
affect the system response.
o Examples: buckling of structures, large deflection of beams.
2. Material Non-linearity:
o Occurs when the material behavior is not linear (e.g., plastic deformation,
hyperelasticity).
o Examples: plasticity, viscoelasticity, creep.
3. Boundary Condition Non-linearity:
o Arises when boundary conditions change during the analysis (e.g., contact
problems).
o Examples: Contact between two bodies, frictional sliding.

2.4.3 Solving Non-linear Problems

Non-linear problems require iterative methods to solve. The most common techniques include:
 • Newton-Raphson Method: Iterative procedure to solve non-linear equations.
• Arc-Length Method: Used for solving problems involving snap-through or post-
buckling behavior.

The non-linear form of the finite element equation is:

K(u)u=f\mathbf{K(u)} \mathbf{u} = \mathbf{f}K(u)u=f

Where K(u)\mathbf{K(u)}K(u) is the stiffness matrix that depends on the displacement

u\mathbf{u}u.

2.5 Summary

In this lecture, we explored the mathematical formulation of the Finite Element Method,
focusing on:

• Governing Equations derived from physical principles that describe real-world

problems.
• The