Strain, Hooke's Law, and Poisson's Ratio

Pull on a bar and it stretches, thins, and springs back, until it doesn't.

Before this Lesson

Bend a paperclip a little and let go, and it springs back as if nothing happened. Bend it hard and it stays bent. You knew that. But watch closely as you let go of the hard bend: even then, it springs back a little. Part of the bend returns and part is gone for good, and one curve tells you how much of each.

Stretch per length

Hang the same load from two rods of the same steel and the same thickness, one 10 ft long and the other 20 ft. Every inch of either rod carries the same stress, so every inch stretches by the same tiny amount. The 20 ft rod has twice as many inches, so it stretches twice as far.

So the total stretch tells you as much about the rod as about the steel. The part that belongs to the steel is the stretch per unit of length, the normal strain. The Handbook prints it under Uniaxial Loading and Deformation:

ε=δL\varepsilon = \frac{\delta}{L}

Here δ\delta is the change in length and LL is the original length. The units cancel, so strain is a pure number: 0.002 means 0.002 in. per inch, or 0.002 mm per mm, or 0.2%. Like stress, it's positive in tension and negative in compression.

Dividing force by area did the same job for stress. Do both, and the member's size drops out. Test a thin bar, a thick bar, and a long bar of one steel. Load against stretch gives three different lines. Stress against strain gives one, and that line belongs to the steel, not to any bar.

0 0.05 0.10 0.15 0.20 0.25 0 10 20 30 stretch, δ (in.) load, P (kip) Load against stretch 0 0.0004 0.0008 0.0012 0 10 20 30 strain, ε stress, σ (ksi) Stress against strain thick bar thin bar long bar thick bar thin bar and long bar thick bar: A = 2 in2, L = 10 ft thin bar: A = 1 in2, L = 10 ft long bar: A = 1 in2, L = 20 ft dots: each bar at P = 30 kip
Three bars of one steel, plotted two ways

One curve for the material

Now keep pulling until the bar breaks, and plot the whole story. For a ductile metal such as mild steel, it runs like this:

stress, σ strain, ε strain hardening necking linear region brittle material: fracture slope = E proportional limit yield ultimate strength fracture ductile metal
The stress-strain diagram of a ductile metal, with a brittle material for contrast
Not to scale. On a true scale, the straight part would nearly hug the stress axis.
  • Linear region, up to the proportional limit. Stress and strain rise in step. The atoms are being pulled slightly apart, and they act like very stiff springs.
  • Yield. The curve bends over. Layers of atoms start to slip past one another, and that stretch stays. This is the yield strength, σy\sigma_y, from the Handbook's Table 2.
  • Strain hardening. The slipping gets harder as it goes, so the stress has to keep climbing to get more stretch.
  • Ultimate strength, σu\sigma_u. The peak of the curve. From here on, the bar necks down at one spot.
  • Fracture. The bar breaks. The curve dips before this only because it keeps dividing by the original area while the neck shrinks.

A brittle material, such as cast iron or glass, skips most of that: close to a straight line, and then it snaps.

The Handbook's Stress-Strain Curve for Mild Steel labels only the yield strength, so the other names are yours to carry in.

The straight part is the most useful piece. Its slope is the modulus of elasticity, EE, which measures the material's stiffness. The Handbook's Hooke's Law entry gives the uniaxial case:

σ=Eε\sigma = E\varepsilon

Strain has no units, so EE has the units of stress. Steel's is 29,000 ksi, far above any stress it will ever see, because strains are tiny: at 29 ksi, steel's strain is only 0.001.

To read EE off a curve, pick a point on the straight part and divide its stress by its strain. The proportional limit is where the straight part ends. Up to there the spring picture holds all the way: take the load off and the member comes back down the same line to zero strain, however many times you do it. Permanent stretch really starts a little above that limit, but for metals the two are close enough to treat as one.

Checkpoint
1 of 3
0 0.008 0.016 0.024 0 100 200 300 400 500 600 700 strain, ε stress, σ (MPa) A B C D
A metal's stress-strain curve with four lettered points

Which point is the proportional limit?

A.
Point A
B.
Point B
C.
Point C
D.
Point D

Past the limit, only part of it comes back

Now load a bar past the proportional limit and take the load off. It doesn't retrace the curve, and it doesn't get back to zero. It comes down a straight line, parallel to the original straight part.

Picture a stiff spring hooked to a block on a rough table. Pull gently, and only the spring stretches. Pull hard enough, and the block slides too. Now let go. The spring relaxes, at the same stiffness as ever, and the block stays where it slid to.

The metal's bonds are the spring, and its slipping layers are the block. Slip doesn't change how stiff the bonds are, so the way down has slope EE. At any point on the curve, the strain is those two parts added together:

εtotal=σE⏟comes back+εset⏟stays\varepsilon_{\text{total}} = \underbrace{\frac{\sigma}{E}}_{\text{comes back}} + \underbrace{\varepsilon_{\text{set}}}_{\text{stays}}
stress, σ strain, ε σ proportional limit loaded to here E unloading: same slope, E permanent set σ/E comes back εtotal
Unloading from a point past the proportional limit

The part that stays is the permanent set, εset=εtotal−σ/E\varepsilon_{\text{set}} = \varepsilon_{\text{total}} - \sigma/E. On a graph you can draw it: from the marked point, run a line parallel to the straight part down to the strain axis, and read where it lands.

The part that comes back sets how much the bar shortens when the load is removed: that recovered strain times the length, (σ/E)L(\sigma/E)L. Use the stress at the marked point, not the stress at the proportional limit, because the harder you've pulled, the farther the spring has stretched. That's your paperclip's small spring-back after a hard bend. The Handbook says nothing about unloading, so this is yours to carry in too.

Worked Example A bar loaded past its limit
1 of 3
0 0.005 0.010 0.015 0.020 0.025 0 10 20 30 40 50 60 strain, ε stress, σ (ksi) 0.003, 30 ksi D (0.020, 50 ksi)
A stress-strain curve with a marked point D

An 8 ft bar is loaded to point D, then unloaded. Find its permanent set and how much it shortens when the load comes off.

Several materials on one plot

On a shared plot, each question looks at a different feature of the curves:

  • Stiffest: the steepest straight part, right at the origin. Stiffness is EE, the initial slope. How high a curve climbs is a separate question.
  • Most plastic deformation: the longest run of strain past the end of the straight part. A material with a long run is ductile.
  • Brittle: it breaks at or near the end of its straight part, with almost no plastic run. Height doesn't settle this either.
Checkpoint
1 of 3
stress, σ strain, ε A B C D
Stress-strain curves of four materials

Which material is the stiffest?

A.
Material A
B.
Material B
C.
Material C
D.
Material D

A bar that stretches also thins

Pull on a rubber band and watch it get skinny. Steel does the same thing, far too slightly to see. Squeeze a bar instead, and it bulges.

So there are two strains: the longitudinal strain along the bar, ε=δ/L\varepsilon = \delta/L, and the lateral strain across it, εlat=Δd/d\varepsilon_{\text{lat}} = \Delta d/d. For a given material, they move in a fixed ratio, which is Poisson's ratio. The Handbook defines it in words under Shear Stress-Strain:

ν=−lateral strainlongitudinal strain\nu = -\frac{\text{lateral strain}}{\text{longitudinal strain}}

The two strains always have opposite signs, so the minus sign makes ν\nu come out positive. Solved for what you usually want, it's εlat=−νε\varepsilon_{\text{lat}} = -\nu\varepsilon.

unloaded L d in tension P P d + Δd Δd is negative δ longitudinal strain: ε = δ/L, positive lateral strain: εlat = Δd/d, negative
A rod before and under tension

For ordinary materials, ν\nu lies between 0 and 0.5. At 0, a bar doesn't thin at all, and cork comes close, which is why a cork slides into a bottle without bulging out. At 0.5, it thins just enough to keep its volume, and rubber comes close. Steel sits at about 0.3. The Handbook doesn't state that range, but the values in its Table 1 - Typical Material Properties sit inside it:

Elastic constants of common materials
MaterialEE (Mpsi)EE (GPa)GG (Mpsi)GG (GPa)ν\nu
Steel29.0200.011.580.00.30
Aluminum10.069.03.826.00.33
Cast iron14.5100.06.041.40.21
Wood (fir)1.611.00.64.10.33
Brass14.8–18.1102–1255.8400.33
Copper171176.5450.36

Source: FE Reference Handbook, Table 1 - Typical Material Properties (adapted)

GG comes in the next section. The Handbook's full table also lists α\alpha, which waits for Thermal Strain and Indeterminate Bars.

Keep the units paired, with stress and EE in the same unit. 29.0 Mpsi is 29×10629 \times 10^6 psi, or 29,000 ksi, and 200 GPa is 200,000 MPa.

Table 1 is for when no specific alloy is named. For an alloy in Table 2 - Average Mechanical Properties of Typical Engineering Materials, such as A36, use that table instead. It prints EE in 10310^3 ksi, so its 29.0 is also 29,000 ksi.

A rod's diameter change is then a chain of four short steps:

σ=PA  ⟶  ε=σE  ⟶  εlat=−νε  ⟶  Δd=εlat d\sigma = \frac{P}{A} \;\longrightarrow\; \varepsilon = \frac{\sigma}{E} \;\longrightarrow\; \varepsilon_{\text{lat}} = -\nu\varepsilon \;\longrightarrow\; \Delta d = \varepsilon_{\text{lat}}\,d
Worked Example How much a steel rod thins
1 of 3

A steel rod, 1 in. in diameter and 8 ft long, carries 20 kip in tension. Using Table 1, find its axial strain and the change in its diameter.

A question can also skip the numbers and ask only what's true of the ratio.

Checkpoint

Which statements about Poisson's ratio are true?

A.
A round rod in compression gets thinner.
B.
In a rod under tension, the axial strain is positive and the lateral strain is negative.
C.
Values for ordinary engineering materials lie between 0 and 0.5.
D.
It equals the change in diameter divided by the change in length.
E.
For steel, it is about 0.3.

Sliding instead of stretching

Lay a thick book flat and push its top cover sideways. The bottom stays put, each page slides a little on the one below it, and the square spine leans over into a parallelogram. No page got longer, and yet the book deformed.

Two ideas carry over from stretching. Force per area is still stress, called shear stress, τ\tau, when the force lies in the face instead of pointing through it. And the deformation is a right angle closing up, so the shear strain, γ\gamma, is the change in that angle, in radians. The Handbook sketches it under Stress and Strain as the reduction in a 90° angle.

In the elastic range these two are proportional as well, which is the shear form of Hooke's law. The Handbook prints it under Shear Stress-Strain, solved for γ\gamma:

γ=τG\gamma = \frac{\tau}{G}

GG is the shear modulus, which Table 1 calls the modulus of rigidity. Sliding and stretching are different deformations: EE resists one, and GG the other. For steel, GG is 11.5 Mpsi against an EE of 29.0, so using EE where GG belongs makes the answer about 2.5 times too small.

You'll meet this in a bearing pad, a block of rubbery elastomer that sits under a bridge girder and lets it creep sideways as it grows and shrinks with the temperature. A block bonded between two plates behaves the same way.

F A = ab a b h F Δ h γ
A bearing pad under a horizontal force

Stress, then strain, then displacement:

τ=FAγ=τGΔ=γh\tau = \frac{F}{A} \qquad \gamma = \frac{\tau}{G} \qquad \Delta = \gamma h

The area is the face the force slides along, the pad's top, A=abA = ab, not its side. The last step is really tan⁡γ=Δ/h\tan\gamma = \Delta/h, but for a small angle in radians, tan⁡γ≈γ\tan\gamma \approx \gamma.

The Handbook has no pad formula and lists no elastomers, so a question will state GG.

Worked Example A bridge bearing pad
1 of 3

An elastomeric bearing pad is 12 in. long, 8 in. wide, and 2 in. tall, with G=150 psiG = 150\ \text{psi}. A horizontal force of 2.88 kip acts on its top. Find the shear stress, the shear strain, and the displacement of the top face.

Three constants, only two of them free

EE, GG, and ν\nu look like three separate facts about a material, but they aren't. Slide a square element into a diamond, and one diagonal has stretched while the other has shortened. So shear is tension and compression in disguise, and how a material resists sliding is already fixed by how it resists stretching and how much it thins. The Handbook gives the link under Shear Stress-Strain:

G=E2(1+ν)G = \frac{E}{2(1 + \nu)}

Since ν\nu lies between 0 and 0.5, GG is always between E/3E/3 and E/2E/2, which is a quick check on any answer. To go the other way, rearrange to ν=E/(2G)−1\nu = E/(2G) - 1.

Try it on steel, and you get 29.0/2.6 = 11.2 Mpsi, while Table 1 prints 11.5. The table lists typical values, and they don't all fit the formula exactly. When a question gives you EE and ν\nu, use the formula.

Checkpoint
1 of 3

A titanium alloy has E=110 GPaE = 110\ \text{GPa} and ν=0.34\nu = 0.34. What is its shear modulus?

GPa

Recap

  • Strain is stretch per length, ε=δ/L\varepsilon = \delta/L. With stress, it strips out the member's size, so one curve describes the material, and its initial slope is EE, in σ=Eε\sigma = E\varepsilon.
  • Up to the proportional limit, a member returns to zero strain. Past it, unloading runs parallel to the initial slope: σ/E\sigma/E comes back, so the bar shortens by (σ/E)L(\sigma/E)L, and the rest stays as permanent set.
  • On a shared plot, the stiffest curve starts steepest, the most plastic has the longest run past its straight part, and the brittle one has almost no run.
  • A rod also thins: εlat=−νε\varepsilon_{\text{lat}} = -\nu\varepsilon, applied to the diameter. The strains have opposite signs, ν\nu lies between 0 and 0.5, and steel's is about 0.3.
  • For a bearing pad, τ=F/A\tau = F/A over the face the force slides along, γ=τ/G\gamma = \tau/G, and Δ=γh\Delta = \gamma h. And G=E/[2(1+ν)]G = E/[2(1 + \nu)], so any two of the constants give the third.