Virtual Work and the Unit-Load Method for Truss Deflections

What it is and where it is used

The unit-load method answers a very specific question: how far does one point on a structure move, in one chosen direction, under a given load set? It is the workhorse hand-calculation for the deflection of pin-jointed trusses — roof trusses, transmission towers, crane booms, bridge lattice girders, and aircraft wing ribs — but the same virtual-work framework also handles beams (bending), frames (bending + axial + shear), and torsion members.

• Design serviceability checks — codes such as Eurocode and the AISC Steel Construction Manual cap deflections (e.g. span/360 under live load); the unit-load method sizes members to meet them.
• Solving indeterminate structures — it supplies the flexibility coefficients used in the force (flexibility) method and in constructing influence lines via the Müller-Breslau principle.
• Camber and pre-set — computing the built-in upward camber a bridge truss needs so it sits flat under dead load.

Because it isolates a single degree of freedom, it is far quicker by hand than solving an entire displacement field, which is why it remains a core topic in every structural-analysis curriculum.

How it works — the virtual-work derivation

The method rests on the principle of virtual work: for a body in equilibrium, the external virtual work done by a virtual force set through the real displacements equals the internal virtual work done by the virtual internal forces through the real internal deformations.

Choose a virtual (dummy) system: a single unit load, 1 (dimensionless or 1 N/1 kN), applied at the point and in the direction of the desired displacement δ. External virtual work is simply 1 × δ. Internally, the unit load creates member forces n. The real loads create member forces N, so each bar of length L, area A, modulus E elongates by the real amount ΔL = NL/(AE). Equating external and internal virtual work over all members:

• 1 · δ = Σ n · (NL/AE) ⟹ δ = Σ (nNL)/(AE)

Every term is a scalar. Sign convention: tension positive, compression negative — so an n and N of the same sign contribute a positive (load-aligned) deflection. The method is exact for linear-elastic, small-displacement behaviour, which is why it obeys Betti's and Maxwell's reciprocal theorems.

Key quantities and a worked example

Symbols: δ = displacement sought; n = axial force in a member from the unit virtual load; N = axial force from the real loads; L = member length; A = cross-sectional area; E = Young's modulus (steel ≈ 200 GPa, aluminium ≈ 69 GPa). Units: with N,n in newtons, L in metres, A in m², E in Pa, δ comes out in metres.

Worked snapshot. Consider a single steel diagonal in a truss: N = +60 kN (tension), L = 4.0 m, A = 1,200 mm² = 1.2×10⁻³ m², E = 200 GPa. Its real elongation is ΔL = NL/(AE) = (60,000 × 4.0)/(1.2×10⁻³ × 200×10⁹) = 1.0×10⁻³ m = 1.0 mm. If the unit load produces n = 0.8 in that bar, its contribution to the joint deflection is n·ΔL = 0.8 × 1.0 mm = 0.80 mm.

• Sum such contributions over all members (a spreadsheet with columns n, N, L, A, and nNL/AE) to get the total joint deflection — typically 10–40 mm for a 20–30 m truss span, comfortably inside a span/360 ≈ 55–83 mm limit.

Using it in practice

In practice the calculation is organised as a table, one row per member:

• Analyse the truss for the real loads (method of joints/sections) to get every N.
• Remove the real loads, apply the unit load at the target joint/direction, re-analyse to get every n.
• Fill columns L, A, E, then compute nNL/AE for each member and sum.

Two powerful extensions cost almost nothing:

• Temperature change: replace NL/AE with the thermal elongation α·ΔT·L, giving δ = Σ n·α·ΔT·L (α_steel ≈ 12×10⁻⁶ /°C). A 30 °C swing on a 4 m bar elongates it 1.44 mm.
• Fabrication error / lack-of-fit: replace NL/AE with the misfit ΔL, so δ = Σ n·ΔL — this predicts the joint movement from members cut slightly long or short, and lets engineers deliberately pre-camber a truss.

To find horizontal and vertical movement at a joint, run two unit-load analyses (one per direction). A rotation is found by applying a unit moment instead of a unit force.

Comparison to related methods

The unit-load method is one member of the energy-method family, and its closest cousin is Castigliano's second theorem, δ = ∂U/∂P where the strain energy of a truss is U = Σ N²L/(2AE). Differentiating, ∂U/∂P = Σ (N/AE)(∂N/∂P)L; the partial derivative ∂N/∂P is numerically identical to n, so the two methods are algebraically the same. The unit-load approach is usually preferred because it avoids differentiation and works even when no real load acts at the point of interest — Castigliano then needs a fictitious variable P set to zero afterward.

• vs. double integration / moment-area: those give the full elastic curve of a beam but cannot handle a truss's discrete axial bars.
• vs. matrix stiffness / FEM: FEM returns every displacement at once and handles indeterminacy automatically, but for a single joint on a determinate truss it is enormous overkill — the unit-load table is faster and more transparent.

All these methods must agree, a fact guaranteed by Maxwell's reciprocal theorem: the deflection at A from a unit load at B equals that at B from a unit load at A.

Limits, pitfalls and significance

The method's power comes with sharp limits:

• Linear elastic only. It assumes Hooke's law and small displacements. Yielding, buckling of slender compression chords, or geometric nonlinearity (P-δ effects) invalidate the superposition it relies on.
• Axial-only for trusses. The classic δ = Σ nNL/AE ignores bending and shear; it is exact only for ideal pin-jointed members carrying pure axial force. Real bolted/welded joints carry secondary bending moments (typically 5–15% of stress), which the pure form omits.
• Sign-convention errors are the number-one student mistake — mixing tension-positive N with compression-positive n silently inverts the answer.
• Garbage-in on A and E: deflection scales inversely with A and E, so a mis-taken area or the wrong grade's modulus propagates directly into δ.

Its enduring significance is as the conceptual bridge between statics and the modern stiffness/FEM machinery: the flexibility coefficients it produces are exactly the off-diagonal terms of a compliance matrix, and virtual work is the variational principle underpinning every finite-element formulation used today.

Common hand methods for computing structural deflections, with their scope and effort.

Method	Governing expression	Best for	Effort / notes
Unit-load (virtual work)	δ = Σ nNL/AE	Deflection at ONE point/direction, any determinate truss	One virtual-load analysis per DOF wanted; handles temperature & fabrication error
Castigliano's 2nd theorem	δ = ∂U/∂P, U = Σ N²L/(2AE)	Deflection where a real load already acts	Needs a dummy variable P if no load at the point; algebraically equivalent to unit-load
Direct integration / double integration	EI d²y/dx² = M(x)	Beam deflection curves	Poor for trusses; gives whole elastic curve
Conjugate-beam / moment-area	Area of M/EI diagram	Beam slopes & deflections	Beams only, not axial-force trusses
Matrix stiffness (FEM)	{d} = [K]⁻¹{F}	All DOFs, indeterminate structures	Computer-based; overkill by hand for a single joint