[Intermediate → Advanced] — Two constants, one linear PDE, and most of structural engineering. The special case that carries the field.
The assumptions
Linear elasticity stacks four simplifications, and knowing which one you’re violating is the skill:
- Small strain — under ~1%, so the linearised strain tensor applies
- Small rotation — reference and deformed geometry are interchangeable
- Linear stress–strain — Hooke’s law, fully reversible
- Usually isotropic and homogeneous
What you get in return is enormous: superposition works. Solve one load case, solve another, add them. Every closed-form result in structural engineering depends on that.
Hooke’s law
The general anisotropic form:
has components, reduced by symmetries:
- and → 36
- Existence of a strain-energy function () → 21
- Isotropy → 2
That collapse from 21 to 2 is the payoff of material symmetry, and it’s why undergraduate mechanics is tractable.
For isotropic materials:
and are the Lamé parameters; is the shear modulus.
Inverted:
The engineering constants
Two independent, but five in common use, and converting between them is a constant chore:
| Meaning | Steel | Aluminium | Concrete | Rubber | |
|---|---|---|---|---|---|
| (Young’s) | axial stress / axial strain | 200 GPa | 70 GPa | 30 GPa | ~0.01 GPa |
| (Poisson) | −lateral / axial strain | 0.30 | 0.33 | 0.20 | ~0.4999 |
| (shear) | shear stress / shear strain | 80 GPa | 26 GPa | 12 GPa | ~0.003 GPa |
| (bulk) | pressure / volume change | 160 GPa | 70 GPa | 17 GPa | ~2 GPa |
The relations:
Two facts worth carrying:
for steel is ~200 GPa regardless of alloy or heat treatment. Stiffness comes from atomic bonding, which alloying barely changes. Strength varies by a factor of ten across steels; stiffness doesn’t. So if a part is too flexible, changing steel grade will not help — you must change geometry. This surprises people constantly.
means incompressible. Look at : as , . Rubber at is essentially incompressible, and that causes volumetric locking in finite elements — standard displacement elements go artificially stiff and give nonsense. The fix is mixed or reduced-integration elements, and it’s a routine practical problem. → FEM
Thermodynamic bounds: . Most materials sit at 0.25–0.35. Cork is ≈0, which is why a cork doesn’t bulge when you push it into a bottle. Auxetic materials have — they get fatter when stretched, which sounds impossible and is achieved with re-entrant microstructures.
Navier’s equations
Substitute Hooke’s law into the momentum balance and express everything in displacement:
Three coupled second-order PDEs in three unknowns. This is the complete statement of linear elastodynamics — everything else is a special case or a solution technique.
Statically, with no body force, it reduces to a form whose solutions are biharmonic (), which is why classical elasticity solutions involve Airy stress functions and complex-variable methods.
Elastic waves fall straight out, and the two speeds are physically distinct:
P-waves (pressure, longitudinal) are faster; S-waves (shear, transverse) are slower and cannot travel through fluids, since fluids have .
That last fact is how we know Earth has a liquid outer core: S-waves from earthquakes leave a shadow zone on the far side. A continuum-mechanics result that maps the planet’s interior. → hardware uses the same wave physics for ultrasonic sensing.
The two-dimensional simplifications
Full 3D is rarely necessary. Two reductions cover most practical work, and confusing them is a standard error.
Plane stress — a thin plate loaded in its plane. The out-of-plane surfaces are free, so:
The plate is free to thin. Use for: sheet metal, thin panels, membranes.
Plane strain — a long prismatic body, loaded uniformly along its length. The material is constrained axially:
Use for: dams, tunnels, retaining walls, a long pipe, a rolling contact line.
Thin ⟹ plane stress. Thick or long ⟹ plane strain. Pick the wrong one and stiffness is off by roughly — about 10% for metals, and more importantly the out-of-plane stress is either present or absent, which changes the failure prediction.
The convenient part: the governing equations are identical in form, so one solution serves both by substituting and to convert plane stress to plane strain.
Axisymmetry is the third reduction — a body of revolution under axisymmetric loading becomes 2D in . Pressure vessels, pipes, and shafts.
Saint-Venant’s principle
The result that makes engineering practical:
Far from the point of load application, the stress distribution depends only on the resultant force and moment, not on the details of how the load was applied.
Bolt, weld, or clamp a bar — beyond roughly one characteristic dimension from the attachment, the stress field is the same.
This is why beam theory works. You don’t model the bolt pattern; you apply a resultant and trust that the interesting region is far enough away. It’s also why you must not trust FE stresses at a point load or a fixed constraint — those are exactly the regions where the principle doesn’t apply, and the singular stresses there are artefacts of the idealisation, not predictions.
A point load in a continuum gives infinite stress. Real loads are distributed; the model isn’t.
Stress concentration
Geometry amplifies stress, and the classic result is worth memorising:
An elliptical hole in an infinite plate under uniaxial tension has a maximum stress at the hole edge:
For a circular hole (), . Three times the nominal stress, regardless of hole size.
Two consequences that matter:
Size doesn’t reduce concentration. A tiny hole concentrates stress as much as a large one — it’s the shape, not the scale. (Size matters for fatigue and fracture through other mechanisms, but not for .)
As the ellipse becomes a crack and . Continuum elasticity predicts infinite stress at a crack tip, which is the theory announcing its own limit. Fracture mechanics replaces the stress with the stress intensity factor , which is finite and measurable. → Failure and Yield
Design implication: sharp internal corners are where parts crack. Generous fillets are not cosmetic — going from a sharp corner to a radius can halve the peak stress, and it’s the cheapest fix available. Peterson’s charts tabulate for standard geometries and are still used daily.
Strain energy
Energy methods give you a second route to every problem, and often a faster one:
Castigliano’s theorem — deflection at a load point is . The quickest way to get a deflection in a frame or truss by hand.
The principle of minimum potential energy — the true displacement field minimises total potential energy among all compatible fields. This is the foundation of the finite element method, which searches a finite-dimensional space of trial functions for the minimiser. → FEM
Virtual work — the workhorse for deriving element formulations and for hand analysis of frames.
Uniqueness
For a well-posed problem — a positive-definite stiffness tensor and adequate boundary conditions — the linear elastic solution is unique (Kirchhoff’s theorem).
That’s a strong guarantee and it’s why linear elastic FE is reliable: there is one answer, and a converged solution has found it. Nonlinear problems lose this — buckling has multiple equilibria, plasticity is path-dependent, and contact problems can have several solutions. The certainty you get in linear analysis is genuinely special, and it’s worth noticing when you leave it behind.
“Adequate boundary conditions” means enough to eliminate rigid-body motion. Six constraints in 3D (three translations, three rotations). Fewer, and the stiffness matrix is singular — the most common beginner FE failure, reported as “insufficient boundary conditions” or a solver that simply won’t converge.
Related
- Constitutive Models — where Hooke’s law sits among alternatives
- Beams and Structures — the engineering specialisations
- FEM — solving this numerically
- Continuum mechanics map
Software engineering
Infrastructure & security
Computer science
Hardware & engineering