Finite Element Method Lecture Notes: Learn FEM Through Clear Explanations and Worked Procedures

Introduction

Learning the finite element method requires much more than reading equations or memorizing matrix formulas. Students must understand how each mathematical expression is developed, where it belongs in the analysis, what physical behavior it represents, and how it contributes to the final displacement, reaction, strain, or stress.

For many beginners, the most difficult part of FEM is not the mathematics itself. The real challenge is understanding how the individual stages fit together. A stiffness matrix may appear simple when studied alone, but its purpose becomes much clearer when it is connected to element behavior, coordinate systems, boundary conditions, and the complete structural model.

Worked procedures reveal the full path from a physical problem to a numerical solution. They show how a structure is idealized, how it is divided into finite elements, how the element equations are developed, and how the separate contributions are assembled into one global system. The lecture-based presentation in this collection supports this structured approach by explaining both the theory and the reasoning behind each computational step.

From a Physical Structure to a Numerical Model

Every finite element analysis begins with a real or idealized physical system. This system may be an axial bar, a truss, a beam, a frame, a plate, a solid body, or a heat-conduction domain.

Before writing equations, the analyst must first understand the expected physical behavior. Important questions include:

· What type of deformation is expected?

· Which loads act on the structure?

· Where is the structure supported?

· Which material properties are required?

· Is the behavior one-dimensional, two-dimensional, or three-dimensional?

· Are the deformations assumed to be small?

· Is the material considered linear and elastic?

These questions determine the type of finite element that should be selected and the assumptions that will be used.

The continuous structure is then replaced by a finite number of smaller regions called elements. The points connecting the elements are called nodes. Instead of solving for the displacement at every point in the body, FEM solves for a limited number of nodal values and uses interpolation functions to estimate the behavior between the nodes.

This is the central idea of the finite element method: a complicated continuous problem is transformed into a finite system of algebraic equations.

A Reliable Solution Sequence

A typical finite element problem follows a clear and repeatable sequence. Although the details vary between bars, beams, trusses, frames, plates, and solids, the general procedure remains nearly the same.

1. Define the Physical Problem

The first step is to identify the geometry, material properties, loads, supports, and expected behavior of the structure.

For an axial bar, the required material and geometric properties commonly include:

· Young’s modulus (E)

· Cross-sectional area (A)

· Element length (L)

· Applied axial forces

· Prescribed nodal displacements

A careful physical description prevents later confusion about signs, directions, units, and boundary conditions.

2. Idealize the Structure

The analyst selects a mathematical model that represents the important features of the real structure.

A slender bar subjected mainly to axial loading may be modeled using one-dimensional bar elements. A roof framework may be represented using truss elements. A bending member may require beam elements, while a building frame may require frame elements capable of representing both axial deformation and bending.

The model should be simple enough to solve efficiently but detailed enough to represent the actual behavior accurately.

3. Discretize the Domain

The structure is divided into finite elements connected at nodes. This process is called discretization or meshing.

A coarse mesh contains relatively few elements and may provide only an approximate solution. A finer mesh usually gives better accuracy but requires more calculations.

Mesh design should account for:

· Changes in geometry

· Changes in material properties

· Load application points

· Support locations

· Regions of high stress gradients

· Holes, corners, notches, and other discontinuities

A good mesh is not necessarily uniform. Smaller elements are often used in critical regions, while larger elements may be sufficient elsewhere.

4. Number the Nodes and Degrees of Freedom

Each node is assigned a number, and each independent nodal displacement is assigned a degree-of-freedom number.

For a one-dimensional axial bar, each node normally has one displacement degree of freedom. For a two-dimensional truss, each node usually has two translational degrees of freedom. Beam and frame elements may also include rotational degrees of freedom.

Consistent numbering is essential because the degree-of-freedom numbers determine where each element contribution is placed in the global matrix.

5. Develop the Element Equations

Each element is described by an equation of the form

[ {(e)}{(e)} = ^{(e)} ]

where

· (^{(e)}) is the element stiffness matrix,

· (^{(e)}) is the element nodal displacement vector,

· (^{(e)}) is the element nodal force vector.

For a uniform axial bar element,

[ ^{(e)} =

]

This matrix relates the nodal displacements of the element to the nodal forces required to produce those displacements.

6. Assemble the Global Equations

The element matrices are combined to form the global stiffness matrix:

[ = ]

where

· () is the global stiffness matrix,

· () is the global nodal displacement vector,

· () is the global nodal force vector.

Assembly is performed by placing each term of the element stiffness matrix into the global matrix according to the element’s degree-of-freedom mapping.

When two or more elements share a node, their stiffness contributions are added at the corresponding global matrix location.

7. Apply Loads and Boundary Conditions

The load vector is constructed from applied nodal loads, distributed loads, body forces, thermal effects, or other actions.

Boundary conditions are then introduced. These may prescribe zero displacement at a fixed support or a known nonzero displacement caused by settlement, thermal expansion, or controlled movement.

Boundary conditions are necessary not only for representing supports but also for eliminating rigid-body motion and making the global system solvable.

8. Solve for Nodal Displacements

After the constraints are applied, the reduced system is solved for the unknown nodal displacements.

[ = ^{-1} ]

In practical computations, the inverse of the stiffness matrix is usually not calculated directly. Instead, efficient numerical methods such as Gaussian elimination, matrix factorization, or iterative solvers are used.

The nodal displacement vector is the primary solution of most displacement-based finite element analyses.

9. Calculate Reactions, Strains, and Stresses

Once the displacements are known, support reactions can be obtained from the original global equations.

Element strains and stresses are then calculated from the element displacement vectors.

For a uniform axial bar element, the strain is

[ = ]

and the stress is

[ = E = E ]

The element axial force is

[ N = A = (u_2-u_1) ]

These secondary quantities often have the greatest engineering importance because they are used to assess strength, safety, deformation limits, and possible failure.

Following this same organized sequence repeatedly helps students recognize the common framework behind many different finite element problems.

Understanding Stiffness Physically

For a uniform axial bar, the stiffness factor is

[ ]

This expression has a direct physical meaning.

Increasing the cross-sectional area (A) makes the bar more resistant to axial deformation. A thicker bar therefore has greater stiffness.

Increasing Young’s modulus (E) also makes the bar stiffer. Materials with a large elastic modulus require greater force to produce the same strain.

Increasing the length (L), however, makes the bar more flexible. A long bar stretches more than a short bar when both are subjected to the same axial force.

The relationship can also be understood from the familiar axial-deformation equation:

[ = ]

Rearranging gives

[ F = ]

Thus, (AE/L) is the force required to produce one unit of relative axial displacement.

Connecting the matrix to this physical meaning is much more valuable than memorizing it in isolation. The finite element stiffness matrix is not merely a collection of numbers. It is a mathematical representation of how the element resists deformation.

Why the Element Matrix Has Positive and Negative Terms

The axial-bar stiffness matrix contains both positive and negative entries:

[ ^{(e)} =

]

The negative terms indicate that the internal force depends on the relative displacement of the two nodes.

If both nodes move by the same amount in the same direction, then

[ u_1=u_2 ]

and the bar experiences no extension:

[ u_2-u_1=0 ]

Therefore, no strain or internal axial force develops. The element undergoes rigid-body translation.

If the second node moves farther than the first, the element elongates. If the first moves farther than the second, the element shortens. The matrix automatically accounts for these possibilities through its pattern of positive and negative terms.

This simple example shows how a stiffness matrix contains physical information about compatibility, equilibrium, deformation, and rigid-body motion.

Why Assembly Deserves Special Attention

An individual element matrix describes only a small part of the complete model. Assembly combines all element equations into one system representing the entire structure.

Suppose two bar elements are connected in a line:

· Element 1 connects nodes 1 and 2.

· Element 2 connects nodes 2 and 3.

Each element has a (2 ) stiffness matrix. However, the complete structure has three nodal displacement degrees of freedom. Its global stiffness matrix is therefore (3 ).

If both elements have stiffness values (k_1) and (k_2), the global stiffness matrix becomes

[ =

]

The middle diagonal term contains

[ k_1+k_2 ]

because node 2 is connected to both elements. Both elements resist movement of that node, so their stiffness contributions are added.

This is the physical basis of assembly.

Errors in node numbering or degree-of-freedom mapping can place correct element terms into incorrect positions. The element matrices may be perfectly derived, but the global model will still be wrong.

Worked procedures make the local-to-global relationship visible by showing:

· The connectivity of each element

· Its local degree-of-freedom vector

· The corresponding global degree-of-freedom numbers

· The positions where each matrix entry is placed

· The locations where shared contributions are added

Assembly deserves careful study because it is one of the central ideas of FEM.

Local and Global Coordinate Systems

Some elements are most easily described in a local coordinate system aligned with the element itself.

For example, the local axis of a truss element follows the direction of the member. In local coordinates, the member carries only axial force, and its stiffness matrix is relatively simple.

However, the complete structure is usually described using global horizontal and vertical axes. The local stiffness matrix must therefore be transformed into global coordinates before assembly.

This transformation ensures that the element’s resistance is represented correctly in the structural coordinate system.

For a two-dimensional truss element inclined at an angle (), the direction cosines are

[ c= ]

and

[ s= ]

The global stiffness matrix contains combinations such as

[ c2,s2,cs ]

These terms distribute the element’s axial stiffness into the global horizontal and vertical directions.

Understanding coordinate transformation helps students see how one-dimensional member behavior contributes to multidirectional structural response.

Boundary Conditions and Structural Stability

The global stiffness matrix is often singular before supports are applied. This is not necessarily an error. It usually indicates that the unconstrained structure can undergo rigid-body motion.

For example, a free bar can translate without deforming. A free two-dimensional body can translate horizontally, translate vertically, and rotate.

Supports remove these rigid-body motions by prescribing selected nodal displacements.

A fixed support may impose

[ u=0 ]

while a roller support may prevent displacement in one direction but allow motion in another.

Boundary conditions must be sufficient to stabilize the structure without introducing unnecessary constraints. Too few constraints leave the system unstable. Too many constraints may change the physical problem and create artificial reactions.

Worked procedures should therefore explain not only how boundary conditions are entered, but why each constraint is needed.

A Simple Worked Procedure

Consider a uniform axial bar of length (L), area (A), and Young’s modulus (E). The left end is fixed, and a force (P) is applied at the right end.

The element stiffness matrix is

[ =

]

The displacement vector is

[ =

]

and the nodal force vector is

[ =

]

where (R_1) is the unknown support reaction.

The finite element equation is

[

=

]

Because the left end is fixed,

[ u_1=0 ]

The second equation becomes

[ u_2=P ]

Therefore,

[ u_2 = ]

The strain is

[ = = ]

The stress is

[ = E = ]

The first equilibrium equation gives the reaction:

[ R_1=-P ]

The negative sign indicates that the reaction acts in the direction opposite to the applied force.

This simple example demonstrates the full FEM sequence:

1. Define the element.

2. Form the stiffness matrix.

3. Write the nodal equations.

4. Apply the boundary condition.

5. Solve for displacement.

6. Calculate strain and stress.

7. Recover the reaction.

8. Check equilibrium.

The Role of Shape Functions

Finite element analysis solves directly for nodal values, but the displacement inside an element must also be approximated.

For a two-node axial bar, the displacement may be written as

[ u(x) = N_1(x)u_1+N_2(x)u_2 ]

where (N_1) and (N_2) are shape functions.

For an element extending from (x=0) to (x=L),

[ N_1(x) = 1- ]

and

[ N_2(x) = ]

Therefore,

[ u(x) = (1-)u_1

· u_2 ]

These functions have important properties:

[ N_1(0)=1, N_1(L)=0 ]

and

[ N_2(0)=0, N_2(L)=1 ]

Thus, the interpolation reproduces the correct nodal displacement at each end.

Differentiating the displacement gives the strain:

[ = = -u_1

· u_2 ]

or

[ = ]

Shape functions therefore connect the nodal unknowns to the displacement and strain fields inside the element.

Why Worked Procedures Are Essential

Finite element theory can appear fragmented when students study stiffness matrices, shape functions, coordinate transformations, assembly, boundary conditions, and stress recovery as separate topics.

A complete worked procedure brings these ideas together.

It demonstrates:

· How the physical model determines the element type

· How the mesh determines the number of unknowns

· How interpolation produces strain

· How material behavior produces stress

· How energy or equilibrium produces stiffness

· How assembly produces the structural equations

· How supports modify the global system

· How nodal displacements lead to stresses and reactions

Repeated exposure to complete procedures helps students develop a reliable problem-solving routine. Eventually, they begin to recognize the same structure even when the element type or engineering application changes.

Common Errors in FEM Solutions

Students often make mistakes that are procedural rather than theoretical.

Inconsistent Units

Young’s modulus, dimensions, forces, and areas must use a compatible system of units. Mixing meters with millimeters or pascals with megapascals can produce errors of several orders of magnitude.

Incorrect Connectivity

If an element is assigned to the wrong nodes, its stiffness terms will be assembled incorrectly.

Sign Errors

Forces and displacements must follow the selected positive directions. A consistent sign convention should be established before calculations begin.

Missing Constraints

An insufficiently constrained model may undergo rigid-body motion, producing a singular stiffness matrix.

Excessive Constraints

Unnecessary supports may prevent realistic deformation and create artificial reaction forces.

Incorrect Coordinate Transformation

For inclined elements, an error in the angle, direction cosines, or transformation matrix can change the entire structural response.

Premature Rounding

Rounding intermediate values too aggressively can reduce accuracy, especially in systems involving several elements.

Misinterpretation of Stress

Large stresses near sharp corners, point loads, or idealized supports may represent stress concentrations or mathematical singularities rather than uniform physical stress fields.

Worked examples help students identify where these errors occur and how to avoid them.

Verify the Answer

A finite element result should never be accepted only because a computer or calculator produced it. Verification is an essential part of the analysis.

Check Global Equilibrium

The total support reactions should balance the applied loads.

For a statically loaded structure,

[ F_x=0 ]

[ F_y=0 ]

and, where applicable,

[ M=0 ]

Failure to satisfy equilibrium may indicate incorrect loading, assembly, signs, or reaction recovery.

Check Prescribed Displacements

Every constrained degree of freedom should satisfy its prescribed value. A fixed degree of freedom should have zero displacement unless a nonzero value was intentionally assigned.

Inspect the Deformed Shape

The deformation pattern should be physically reasonable.

A member subjected to tension should generally elongate. A downward transverse load should usually produce downward bending. Unexpected deformation directions may indicate sign errors, incorrect supports, or coordinate-transformation mistakes.

Check Symmetry

If the geometry, loading, and supports are symmetric, the displacement and reaction results should normally reflect that symmetry.

Compare with a Simple Analytical Solution

Whenever possible, compare the FEM result with a known solution for a bar, beam, truss, or other simplified system.

Even an approximate hand calculation can reveal a major modeling error.

Perform a Mesh-Refinement Study

The model should be solved using progressively finer meshes. If the quantity of interest approaches a stable value, confidence in the numerical result increases.

For example, displacement values may be compared for meshes containing:

[ 2,,, ]

elements.

A result that changes significantly with every refinement has not yet demonstrated convergence.

Examine Stress Carefully

Displacements often converge more smoothly than stresses. Stress values may be especially sensitive near concentrated loads, supports, sharp corners, cracks, and material interfaces.

The analyst should distinguish between meaningful high stress and numerical behavior caused by an idealized singularity.

Building Confidence Before Advanced Applications

The best way to learn FEM is to begin with simple elements and gradually move toward more complicated models.

A useful progression is:

· Axial bars

· Springs

· One-dimensional heat conduction

· Plane trusses

· Beams

· Two-dimensional frames

· Triangular and quadrilateral elements

· Axisymmetric solids

· Plates and shells

· Three-dimensional solids

· Dynamic and nonlinear analysis

Simple problems are not merely elementary exercises. They reveal the structure of the method clearly and allow each step to be checked by hand.

Once students understand the complete solution sequence for a bar or truss, they are better prepared to study larger matrices, numerical integration, isoparametric elements, nonlinear materials, contact, dynamics, and commercial finite element software.

The Difference Between Using Software and Understanding FEM

Modern software can generate a mesh, assemble matrices, solve thousands or millions of equations, and display colorful stress contours. However, software cannot replace engineering judgment.

A student who understands the FEM procedure can evaluate whether:

· The selected element type is appropriate

· The mesh is sufficiently refined

· The supports represent the real structure

· Loads have been applied correctly

· Material properties are consistent

· Stress concentrations are physically meaningful

· The solution has converged

· The results satisfy equilibrium

· The deformation is realistic

Without this understanding, a visually impressive result may still be incorrect.

Lecture notes and worked procedures provide the theoretical foundation needed to use software responsibly.

What This Learning Approach Provides

A structured lecture-based approach helps students:

· Follow the major stages of an FEM solution

· Understand the physical meaning of stiffness matrices

· Connect element equations to complete structural models

· Learn local-to-global assembly systematically

· Apply loads and boundary conditions correctly

· Calculate displacements, reactions, strains, and stresses

· Identify common modeling and computational errors

· Verify results using equilibrium and physical reasoning

· Understand mesh refinement and convergence

· Build confidence before studying advanced applications

The goal is not simply to obtain a numerical answer. The goal is to understand why the procedure works, what the result means, and how to judge whether it is trustworthy.

Finite element analysis becomes much easier when it is learned as an organized sequence rather than as a collection of disconnected equations. A step-by-step presentation allows students to see how physical behavior, mathematical modeling, matrix equations, and engineering interpretation work together.

Would a clear, systematic, and worked-procedure approach make finite element analysis easier for you to master?

🛒 [Buy the book on Amazon]

🛒 [Buy the book on Gumroad]

🛒 [Buy the book on Payship]

📬 Subscribe to our newsletter for free and be the first to receive exciting updates, exclusive content, special offers, delivered straight to your inbox. Whether you’re a parent, teacher, or student enthusiast, there’s something for everyone — don’t miss out!