Energy and Variational Methods

Introduction and Context

Energy methods provide a powerful alternative to solving structural equilibrium equations directly. Instead of determining every internal force from local equilibrium, they relate external work, strain energy, displacement, and stability. Variational methods generalize this idea by expressing structural behavior as the stationary value of a functional. They form the theoretical basis of the unit-load method, Castigliano’s theorems, the principle of virtual work, Rayleigh–Ritz methods, finite-element methods, and modern structural optimization.

These methods are especially valuable for indeterminate structures, frames, plates, shells, nonlinear systems, and structures with complicated loading or geometry. At postgraduate level, their importance lies not only in obtaining numerical answers but also in understanding assumptions, admissible fields, convergence, stability, and the relationship between continuous mechanics and discrete computational models.

Learning Objectives

After studying this chapter, the reader should be able to:

1. distinguish external work, complementary energy, strain energy, and potential energy;
2. derive and apply virtual-work and energy theorems;
3. calculate displacements and redundants using Castigliano’s theorems and the unit-load method;
4. formulate structural equilibrium as a variational problem;
5. explain admissibility, stationarity, compatibility, and convergence;
6. apply the Rayleigh–Ritz approach to approximate structural response;
7. identify limitations associated with linearity, material behavior, boundary conditions, and numerical approximation.

1. Fundamental Energy Concepts

1.1 Work and strain energy

For a gradually applied load, the incremental external work is

where P is an applied force and Δ is its displacement in the force direction. For a linearly elastic member loaded from zero to a final force P,

The work is stored as strain energy U. In a general three-dimensional elastic body,

where σ_ij is the stress tensor, ε_ij is the strain tensor, V is the body volume, and repeated indices imply summation. The factor one-half applies to linear elastic loading from an unstressed state.

For common structural actions, the total strain energy can be written as

Here N, M, V, and T are axial force, bending moment, shear force, and torsional moment; E is Young’s modulus; G is shear modulus; A is cross-sectional area; I is second moment of area; J is torsional constant; κ is the shear correction factor; and L is member length. Terms may be omitted when their deformation contribution is negligible. For slender beams, bending energy commonly dominates; for deep beams, shear energy may be important.

Dimensional consistency is evident: M²/(EI) has units of force, and multiplication by dx gives force-length, the unit of energy.

1.2 Complementary energy

Complementary energy is naturally expressed in terms of stresses or generalized forces. For a linearly elastic member it is numerically equal to strain energy, but the variables used are different. In nonlinear elasticity, the two quantities need not be interpreted identically.

For a uniaxial material,

where P' is a dummy load variable. If the load–displacement relation is linear, U = U^* = PΔ/2. Complementary-energy formulations are particularly useful when assumed stress fields satisfy equilibrium.

1.3 Total potential energy

For a conservative structural system, the total potential energy is

where W_e is the potential of external loads. For prescribed concentrated forces P_i and displacements Δ_i,

with signs governed by the chosen coordinate convention. The equilibrium configuration makes Π stationary:

For a stable linear elastic system, the stationary value is a minimum. Stationarity alone does not guarantee stability; the second variation must also be considered. A sufficient local stability condition is

for every nonzero admissible variation.

2. Principle of Virtual Work

A virtual displacement is an imagined, infinitesimal, kinematically admissible displacement imposed at a fixed load state. It need not occur physically. The principle of virtual work states that a body is in equilibrium if and only if the virtual work of external forces equals the virtual work of internal stresses for every admissible virtual displacement:

In continuum form,

where b_i are body-force components, t_i are prescribed surface tractions, F_ki are concentrated forces, u_i are displacements, and S is the traction boundary. Virtual displacements vanish on displacement-prescribed boundaries.

The principle is valid for small or finite deformation theories when the appropriate strain and equilibrium descriptions are used. The simple linear form assumes compatible kinematics, equilibrium, and constitutive behavior that permits the stress–strain work expression.

3. Unit-Load and Virtual-Work Methods

For a linearly elastic structure, the displacement at a point in a specified direction is obtained by applying a unit load in that direction to the actual structure. The general expression is

The lower-case quantities n, m, v, and t are internal actions caused by the unit load. The upper-case quantities arise from the actual loading. The expression assumes linear elasticity, small displacements, unchanged geometry, and compatible boundary conditions. It is often called the Maxwell–Mohr integral.

For a beam in which bending dominates,

The unit load must be applied at the desired displacement location and in the desired direction. For a rotation, the virtual action is a unit moment. For a relative displacement, the unit load is applied as an equal and opposite pair.

Worked Example 1: Cantilever-tip displacement

A prismatic cantilever of length L carries a downward point load P at its free end. Neglect shear and axial deformation. Determine the vertical tip displacement.

The bending moment due to the actual load at a section measured from the fixed end is M(x)=P(L-x). A unit downward load at the tip gives m(x)=L-x. Therefore,

Since ∫₀^L(L-x)²dx=L³/3,

The result has units force × length³ divided by force × length², hence length. The displacement is downward in the direction of the applied load.

4. Castigliano’s Theorems

For a linearly elastic structure with independent generalized forces Q_i and corresponding displacements q_i, the strain energy is a function of the forces. Castigliano’s second theorem states

Thus, displacement is obtained by differentiating strain energy with respect to the corresponding force. If Q_i is a moment, q_i is the associated rotation.

When strain energy is expressed as a function of generalized displacements, the first theorem gives

These relations require conservative loading and differentiability of the energy function. In practical use, all independent loads must be retained as variables, including a dummy load if a displacement not associated with an existing force is required.

For a beam whose energy is dominated by bending,

If M depends on Q, then

Since ∂M/∂Q is the bending moment diagram produced by a unit value of Q, this is identical to the unit-load method.

5. Indeterminate Structures and Redundant Forces

Energy methods are particularly effective for statically indeterminate structures. Select redundant forces R_j and express internal actions as functions of the applied loads and redundants. Compatibility requires the displacement in each released redundant direction to equal its prescribed value, often zero.

For a redundant with zero displacement,

For a prescribed displacement Δ_j⁰,

For a linear system, this produces simultaneous equations of the form

where f_ij is the flexibility coefficient representing displacement in direction i caused by a unit redundant in direction j, and d_i is the compatibility displacement after removal of redundants. Maxwell’s reciprocal theorem gives f_ij=f_ji for linear elastic structures with compatible boundary conditions.

6. Variational Formulation of Structural Problems

A variational problem seeks a function that makes a functional stationary. For an Euler–Bernoulli beam with transverse displacement w(x), flexural rigidity EI, and transverse load q(x), the total potential energy may be written as

where primes denote derivatives with respect to x, and W_boundary represents work of applied boundary forces or moments. The admissible functions satisfy the essential, or geometric, boundary conditions such as prescribed displacement and rotation.

Taking the first variation,

Integrating the first term by parts twice gives the governing differential equation

under the stated sign convention. The boundary terms produce natural boundary conditions involving bending moment and shear force. This derivation demonstrates an important principle: the strong differential equation and boundary conditions are consequences of a single energy statement.

Essential and natural boundary conditions

Essential conditions prescribe the primary variable, such as displacement or rotation. They must be imposed directly on the trial functions. Natural conditions prescribe force-like quantities, such as moment, shear, traction, or axial force. They arise automatically from the variational statement and need not be imposed strongly in a Ritz formulation.

Confusing these two types is a common source of incorrect solutions. A trial function that violates an essential boundary condition is inadmissible, even if it appears to satisfy the differential equation approximately.

7. Rayleigh–Ritz Method

The Rayleigh–Ritz method approximates the unknown field by a finite series:

where a_r are unknown coefficients and φ_r(x) are admissible trial functions satisfying all essential boundary conditions. Substitution into Π gives a function of the coefficients, Π(a₁,a₂,...,a_n). Stationarity requires

For a linear elastic problem this becomes

where [K] is the stiffness matrix, {a} is the coefficient vector, and {f} is the generalized load vector. The matrix is symmetric for conservative linear elastic systems.

Approximation and convergence

A Ritz solution is an admissible displacement field, so it satisfies essential constraints exactly but may not satisfy equilibrium pointwise. As the trial space is enriched, the potential energy generally approaches the exact minimum from above for a stable linear elastic problem. Convergence depends on the suitability and completeness of the trial functions, not simply on their number.

Trial functions should be independent, sufficiently smooth for the energy integral, and capable of representing the dominant deformation modes. Poorly selected functions can cause slow convergence or ill-conditioning. Polynomial, trigonometric, orthogonal, and problem-specific functions are common choices.

Worked Example 2: One-term Ritz approximation

Consider a simply supported beam of length L, constant EI, and uniform load q₀. Choose

This satisfies w(0)=w(L)=0, the essential displacement conditions. The curvature is

The potential energy is

Using ∫₀^Lsin²(πx/L)dx=L/2 and ∫₀^Lsin(πx/L)dx=2L/π,

Stationarity gives

and therefore

The maximum displacement occurs at midspan because the sine function is largest at x=L/2. This is an approximation to the exact beam response, but the selected function captures the correct symmetry and boundary displacements.

8. Stability and Second Variation

The first variation identifies equilibrium; the second variation assesses whether that equilibrium is stable. In a stable elastic system, a small admissible displacement away from equilibrium increases potential energy. At buckling, the second variation becomes zero in a critical mode.

For an ideal pin-ended column under axial force P, a simplified total potential energy is

The first term is bending energy and the second represents the destabilizing potential of the compressive load. Choosing w=a sin(πx/L) gives the critical condition

This is Euler’s critical load for a pin-ended, prismatic, straight column under ideal conditions. Other end conditions replace L by an effective length KL, giving P_cr=π²EI/(KL)². The formula does not account for imperfections, residual stress, inelasticity, or lateral-torsional interaction.

9. Complementary-Energy and Mixed Methods

Displacement-based formulations use kinematically admissible fields and are closely associated with total potential energy. Force-based formulations use statically admissible stress or internal-force fields and are associated with complementary energy. Mixed methods treat both displacement and stress variables as unknowns.

A statically admissible field satisfies equilibrium and force boundary conditions but may not satisfy compatibility exactly. A kinematically admissible field satisfies compatibility and displacement boundary conditions but may not satisfy equilibrium pointwise. These distinctions support upper- and lower-bound reasoning in appropriate linear elastic problems and are important in finite-element error assessment.

Mixed and hybrid finite elements can be advantageous when stress accuracy, incompressibility, constraints, or interface conditions are important. However, they require careful choice of interpolation spaces to avoid spurious modes and loss of stability.

10. Application to Finite-Element Analysis

The finite-element method can be derived directly from virtual work. For an element, write the displacement field as

where [N] contains shape functions and {d} contains nodal degrees of freedom. The strain is

where [B] is the strain–displacement matrix. For linear elasticity, {σ}=[D]{ε}, with [D] the constitutive matrix. The element stiffness matrix is

and the assembled equilibrium equations are

where [K] is the global stiffness matrix and {F} is the global nodal-load vector. The matrix is symmetric positive definite after sufficient displacement constraints are applied to remove rigid-body motion, provided the material and model are stable.

Energy concepts assist in mesh refinement and verification. A decreasing displacement or energy error under systematic refinement is evidence of convergence, but it is not by itself proof that the model represents the real structure. Boundary conditions, material properties, load idealization, and element formulation must also be validated.

11. Illustrative Case: Energy Analysis of a Portal Frame

Background

Consider an illustrative, two-dimensional steel portal frame used to support a service platform. The columns and beam are prismatic, joints are rigid, and lateral wind loading acts at the beam level. The frame is statically indeterminate, and the design question is the horizontal roof displacement.

Problem

A direct equilibrium solution requires member stiffness relations and compatibility equations. The engineering team also needs to assess whether axial deformation and panel-zone flexibility can be neglected.

Analysis and approach

The frame is first idealized using small-displacement linear elasticity. A unit horizontal load is applied at the roof point where displacement is required. Internal actions from the actual wind load are denoted by N,M,V; those from the unit load are n,m,v. The predicted displacement is

The analysis is repeated with and without axial terms. If the difference is small relative to the project serviceability limit, bending-only modeling may be acceptable for preliminary work. If the frame is tall, heavily braced, or has flexible connections, axial and connection deformation must be retained.

A finite-element model is then constructed from the same variational formulation. Mesh refinement is performed near joints and load introduction regions. Reactions are checked against global equilibrium, and the computed strain energy is compared with external work. For a linear load case, the total external work under gradual loading should equal the strain energy, while work under final-load notation satisfies U=PΔ/2 for a single load–displacement pair.

Outcome

The energy calculation identifies the dominant deformation mechanism and provides an independent check on the finite-element displacement. Disagreement would prompt examination of sign conventions, releases, boundary conditions, member axes, shear deformation, or load application.

Learning points

1. Unit-load methods are efficient for selected displacements.
2. Energy decomposition reveals whether bending, axial, shear, or torsional deformation governs.
3. Variational and finite-element results should be checked by equilibrium and energy balance.
4. A numerically converged model can still be physically inappropriate if the idealization is poor.

12. Limitations, Professional Practice, and Contemporary Developments

Classical energy methods commonly assume linear elasticity, conservative loading, small deformation, stable equilibrium, and sufficiently smooth fields. They require modification for plasticity, cracking, hysteresis, contact, large rotations, follower forces, dynamic loading, and path-dependent materials. In nonlinear systems, the work depends on loading history, and a single scalar strain-energy function may not exist.

For dynamic systems, kinetic energy must be included. A Lagrangian may be written as

where T is kinetic energy. Hamilton’s principle then yields the equations of motion under suitable variations and endpoint conditions. Damping and nonconservative forces require additional treatment.

In professional practice, energy methods should be used with documented assumptions, unit checks, independent verification, and sensitivity studies. Design codes may prescribe specific load combinations, material models, imperfection factors, and acceptance criteria; energy calculations do not replace those requirements.

Computational developments include nonlinear finite elements, isogeometric analysis, topology optimization, and machine-learning-assisted surrogate models. AI can help identify reduced trial spaces, detect anomalous simulation results, or accelerate parametric studies, but it cannot independently validate physical assumptions. Training data, extrapolation, explainability, and engineering responsibility remain important governance issues.

Chapter Summary

Energy methods express structural response through work and stored energy rather than relying only on local equilibrium. Strain energy is obtained from stress–strain action, while total potential energy combines internal energy with the potential of external loads. The principle of virtual work equates internal and external virtual work and leads directly to displacement formulas, finite-element equations, and variational formulations.

The unit-load method and Castigliano’s theorems provide efficient displacement calculations in linear elastic structures. Redundant forces can be determined by differentiating strain energy and imposing compatibility. The Rayleigh–Ritz method approximates displacement fields using admissible functions, and its stationarity equations produce a finite system of algebraic equations. The second variation is essential for distinguishing stable equilibrium from instability, including buckling.

These methods are powerful but depend on appropriate assumptions and modeling judgment. Verification must include dimensional checks, equilibrium checks, energy balance, convergence studies, and assessment of whether neglected deformation mechanisms are significant.

Key Terms

Strain energy: Energy stored in a deforming elastic body.

Complementary energy: Energy-like functional expressed naturally in terms of stress or generalized force variables.

Total potential energy: Internal strain energy minus the potential of external conservative loads.

Virtual displacement: An imagined infinitesimal, kinematically admissible displacement.

Virtual work: Work associated with a virtual displacement or virtual strain.

Unit-load method: Calculation of a displacement by applying a unit force or moment in the desired direction.

Castigliano’s first theorem: Relation giving generalized force by differentiating energy with respect to displacement.

Castigliano’s second theorem: Relation giving displacement by differentiating strain energy with respect to force.

Flexibility coefficient: Displacement in one direction caused by a unit force in another direction.

Redundant: An unknown force or moment beyond those obtainable from equilibrium alone.

Compatibility: Requirement that deformations form a continuous and kinematically possible field.

Admissible displacement: A displacement field satisfying essential boundary conditions and required regularity.

Essential boundary condition: A prescribed displacement or rotation condition.

Natural boundary condition: A prescribed traction, force, shear, or moment condition.

Variational principle: Statement that a functional is stationary for the physical solution.

Rayleigh–Ritz method: Approximation method using a finite set of admissible trial functions.

Second variation: The second-order change in a functional used to assess stability.

Complementary admissibility: Satisfaction of equilibrium and force boundary conditions by an assumed force field.

Flexural rigidity: Product EI, representing bending stiffness.

Energy convergence: Approach of an approximate energy or response toward the exact solution as the approximation space is enriched.

Review Questions

Short Answer Questions

1. Define strain energy and explain why the factor one-half appears for linear elastic loading.
2. Distinguish essential and natural boundary conditions.
3. State Castigliano’s second theorem.
4. What is meant by a kinematically admissible displacement field?
5. Why is the second variation relevant to structural stability?

Long Answer Questions

1. Derive the unit-load expression for displacement of a linearly elastic beam, including bending and axial contributions.
2. Explain the principle of virtual work and show how it leads to the weak form of a beam equation.
3. Discuss the similarities and differences between total potential-energy and complementary-energy formulations.
4. Derive the Rayleigh–Ritz equations for a beam using a finite series of admissible functions.
5. Explain how energy methods can be used to analyze statically indeterminate structures and verify finite-element models.

Analytical / Application Questions

1. A cantilever carries a tip force and a tip moment. Formulate its total bending strain energy and use Castigliano’s second theorem to obtain both the tip displacement and tip rotation.
2. For a simply supported beam under a nonuniform load, select two admissible sine trial functions and formulate the two Ritz equations. State how convergence could be assessed.
3. A portal-frame finite-element model gives a roof displacement that differs substantially from a unit-load calculation. Develop a systematic diagnostic procedure using equilibrium, boundary-condition, stiffness, energy, and mesh checks.

Multiple Choice Questions with Answers

Question 1

For a linearly elastic member loaded gradually from zero to a final force P, the stored strain energy is:

A. PΔ

B. PΔ/2

C. 2PΔ

D. P/2Δ

Correct Answer: B. PΔ/2

Question 2

In the unit-load method, the unit load is applied:

A. At the centroid of the structure

B. At the support with the largest reaction

C. At the point and in the direction of the desired displacement

D. Only at the location of maximum moment

Correct Answer: C. At the point and in the direction of the desired displacement

Question 3

Castigliano’s second theorem gives a displacement by differentiating strain energy with respect to:

A. Young’s modulus

B. The corresponding generalized force

C. Member length

D. Cross-sectional area

Correct Answer: B. The corresponding generalized force

Question 4

Which condition must a Ritz trial function satisfy directly?

A. All equilibrium equations pointwise

B. All natural force boundary conditions

C. Essential displacement boundary conditions

D. The exact bending-moment diagram

Correct Answer: C. Essential displacement boundary conditions

Question 5

For a stable equilibrium configuration, the second variation of total potential energy is generally:

A. Negative for every admissible variation

B. Zero for every admissible variation

C. Positive for every nonzero admissible variation

D. Independent of the material stiffness

Correct Answer: C. Positive for every nonzero admissible variation

Question 6

For bending-dominated response, the unit-load displacement integral is:

A. ∫Mm/EI dx

B. ∫EI Mm dx

C. ∫M+m/EI dx

D. ∫M/mEI dx

Correct Answer: A. ∫Mm/EI dx

Question 7

The Euler critical load for an ideal pin-ended prismatic column is:

A. πEI/L

B. π²EI/L²

C. EI/(πL²)

D. π²EA/L²

Correct Answer: B. π²EI/L²

Question 8

In a linear elastic finite-element formulation, the element stiffness matrix is commonly written as:

A. ∫[B][D][B]dV

B. ∫[B]ᵀ[D][B]dV

C. ∫[D]ᵀ[B][B]dV

D. ∫[N]ᵀ[N]dV only

Correct Answer: B. ∫[B]ᵀ[D][B]dV

Question 9

A redundant force in an indeterminate structure is determined through energy methods primarily by imposing:

A. A compatibility condition

B. A temperature condition only

C. A density condition

D. A change in coordinate origin

Correct Answer: A. A compatibility condition

Question 10

For a cantilever of length L with a tip load P, constant EI, and bending deformation only, the tip displacement is:

A. PL/EI

B. PL²/(2EI)

C. PL³/(3EI)

D. 3PL³/EI

Correct Answer: C. PL³/(3EI)