The Tension Test

A tension test pulls a bar apart and records every property on the way.

Before this Lesson

Clamp a piece of steel no thicker than a marker in a machine, pull until it snaps, and what you get is a sheet with two columns: load and elongation. Nothing on it says "modulus" or "yield strength". But the 36 ksi and 58 ksi the Handbook prints beside A36 steel started as a sheet like that one, and by the end of this Lesson you'll be able to pull those numbers out yourself.

What the machine records

A tension specimen is a bar with thick ends for the machine's grips and a slimmer straight middle. It's usually round, and sometimes a flat strip cut from plate. The machine pulls on the ends and reports the load, PP.

The stretch is measured only on the straight middle. The thick ends barely stretch, and they can slip in the grips, so they tell you little about the metal. Two marks on the middle set the gauge length, L0L_0, and a clip-on extensometer reads how far those marks move apart: the elongation, δ\delta.

round specimen grip extensometer reads δ P P d gauge length, L0 overall length d A0 = πd2/4 flat specimen w gauge length, L0 w t A0 = wt
Round and flat tension specimens

Divide by what the bar started with

Test a second bar of the same metal, and every row of the sheet changes. Give a bar twice the area, and it takes twice the load to do the same thing to the metal. Give it twice the gauge length, and the same stretch per inch shows up as twice the elongation. So divide both out. Load over the original area is engineering stress, and elongation over the original gauge length is engineering strain:

σ=PA0ε=δL0\sigma = \frac{P}{A_0} \qquad \varepsilon = \frac{\delta}{L_0}

Plot those, and every bar of that metal lands on one curve. The curve describes the material, not the bar. The Handbook prints both under Properties of Materials, in its Materials Science chapter, with FF for our PP and ΔL\Delta L for our δ\delta.

A0A_0 comes from dimensions measured before the test: πd2/4\pi d^2/4 for a round specimen, and width times thickness, wtwt, for a flat one. Take a round specimen 0.505 in. in diameter, with a 2 in. gauge length, and a reading of 9,000 lb at an elongation of 0.0031 in.:

A0=π(0.505)24=0.2003 in2σ=9,000 lb0.2003 in2=44,900 psi=44.9 ksiε=0.0031 in.2 in.=0.00155\begin{aligned} A_0 &= \frac{\pi (0.505)^2}{4} = 0.2003\ \text{in}^2 \\ \sigma &= \frac{9{,}000\ \text{lb}}{0.2003\ \text{in}^2} = 44{,}900\ \text{psi} = 44.9\ \text{ksi} \\ \varepsilon &= \frac{0.0031\ \text{in.}}{2\ \text{in.}} = 0.00155 \end{aligned}

Keep the units paired: pounds over square inches give psi, kips over square inches give ksi, and newtons over square millimeters give MPa.

The peak and the neck

Now keep pulling. Past the straight part, the metal hardens as it slips, so the load keeps climbing even though the whole bar is slowly thinning. At the top of the curve, hardening stops keeping up. One spot thins a little more than the rest, which makes it the weakest spot, which makes it thin faster. All further stretch goes into that waist, the neck, and the load falls until the bar breaks.

engineering strain, ε engineering stress, σ proportional limit ultimate tensile strength, σu fracture stretches evenly neck starts breaks in the neck
A whole test, from first load to fracture

The peak is the ultimate tensile strength, which the Handbook's Materials Science chapter lists as the maximum engineering stress:

σu=Pmax⁡A0\sigma_u = \frac{P_{\max}}{A_0}

The largest load on the sheet goes over the area the bar started with. Engineering stress stays on that original area to the end, neck or no neck, which is why the curve falls after the peak even though the metal in the neck hasn't gone soft. The load dropped, and the formula never noticed the neck.

The Handbook lists these properties but doesn't describe the test, so gauge length, the proportional limit, and necking are yours to carry in.

Checkpoint
1 of 3

A round specimen is 10 mm in diameter and 150 mm long overall, with a 50 mm gauge length. At one reading, it carries 20 kN. What is its engineering stress?

MPa

The slope of the straight part

EE is the slope of the straight part. The Handbook says so right under its Stress-Strain Curve for Mild Steel: "The slope of the linear portion of the curve equals the modulus of elasticity." When that line runs through the origin, one reading gives you the slope, and the Handbook prints that form under Uniaxial Loading and Deformation: E=σ/ε=(P/A)/(δ/L)E = \sigma/\varepsilon = (P/A)/(\delta/L). The 9,000 lb reading from earlier gives 44.9 ksi over 0.00155, or 29,000 ksi, which is what Table 1 - Typical Material Properties lists for steel.

The catch is that the reading has to be on the straight part, and a sheet doesn't announce where that ends. While the bar is on the line, equal steps of load give equal steps of elongation. The row where the elongation jumps by more than its share is past the proportional limit.

Worked Example E from one reading
1 of 2

A round aluminum specimen, 12.5 mm in diameter with a 50 mm gauge length, carries 15 kN in its linear range. The extensometer reads 0.089 mm. Find the modulus of elasticity.

When the line misses the origin

That one division assumes the line runs through the origin, and often it doesn't. A bar has to be seated in its grips, so the machine may already carry a small load when the extensometer is zeroed. The line is as straight as ever, but it starts partway up the load axis, and one row's stress over strain gives the slope of a line back to the origin instead.

So take two readings on the straight part and use the differences, which don't care where the line started. This is the Handbook's slope sentence as algebra:

E=ΔσΔε=ΔP/A0Δδ/L0E = \frac{\Delta\sigma}{\Delta\varepsilon} = \frac{\Delta P / A_0}{\Delta\delta / L_0}

Take a flat steel bar 1 in. wide and 0.25 in. thick, with a 2 in. extensometer. It reads 0.0002 in. at 2,000 lb and 0.00185 in. at 8,000 lb:

0 0.0005 0.0010 0.0015 0.0020 0.0025 0 2,000 4,000 6,000 8,000 10,000 elongation, δ (in.) load, P (lb) Δδ = 0.00165 in. ΔP = 6,000 lb one reading, back to the origin reading 1 reading 2 load on the bar when the extensometer was zeroed
Two readings on a line that misses the origin
A0=(1)(0.25)=0.25 in2Δσ=6,000 lb0.25 in2=24,000 psi=24 ksiΔε=0.00165 in.2 in.=0.000825E=240.000825=29,100 ksi\begin{aligned} A_0 &= (1)(0.25) = 0.25\ \text{in}^2 \\ \Delta\sigma &= \frac{6{,}000\ \text{lb}}{0.25\ \text{in}^2} = 24{,}000\ \text{psi} = 24\ \text{ksi} \\ \Delta\varepsilon &= \frac{0.00165\ \text{in.}}{2\ \text{in.}} = 0.000825 \\ E &= \frac{24}{0.000825} = 29{,}100\ \text{ksi} \end{aligned}

That's steel. The 8,000 lb reading alone gives 32 ksi over 0.000925, or 34,600 ksi, which isn't any steel.

Checkpoint
1 of 2

A flat aluminum bar, 1.5 in. wide and 0.125 in. thick, is tested with a 2 in. extensometer. In the linear range, it reads 0.0002 in. at 500 lb and 0.0018 in. at 2,000 lb. What is its modulus of elasticity?

ksi

The same straight part, measured sideways

While a bar stretches, it also thins. Measure across it as well as along it, and the straight part gives you Poisson's ratio too. The Handbook defines the ratio under Shear Stress-Strain as minus the lateral strain over the longitudinal strain. It prints no version with lengths and diameters, so you build each strain yourself, as a change over its own original size:

ε=δL0εlat=Δddν=−εlatε\varepsilon = \frac{\delta}{L_0} \qquad \varepsilon_{\text{lat}} = \frac{\Delta d}{d} \qquad \nu = -\frac{\varepsilon_{\text{lat}}}{\varepsilon}

The bar is getting thinner, so Δd\Delta d and the lateral strain are negative, and the minus sign makes ν\nu come out positive. It's always reported as a positive number. Take both readings on the straight part, since the ratio belongs to the elastic range.

Back to the 0.505 in. specimen. At its strain of 0.00155, its diameter has shrunk by 0.00024 in., so the lateral strain is −0.00024/0.505=−0.000475-0.00024/0.505 = -0.000475, and ν=0.000475/0.00155=0.31\nu = 0.000475/0.00155 = 0.31. Table 1 lists 0.30 for steel.

Checkpoint
1 of 2

In the elastic range of a tension test, a specimen shows an axial strain of 0.0011 and a lateral strain of −0.00034. What is Poisson's ratio?

A yield strength with no corner to point at

Many metals, aluminum alloys and high-strength steels among them, curve away from the straight line so gradually that no point on the bend stands out. Ask two people where yielding began, and you'll get two answers.

You've already met the way out. In Strain, Hooke's Law, and Poisson's Ratio, a bar unloaded past the proportional limit came down a line parallel to the straight part, because slipping doesn't change how stiff the atomic bonds are. The bar keeps whatever strain is left when that line reaches the axis, its permanent set:

εset=ε−σE\varepsilon_{\text{set}} = \varepsilon - \frac{\sigma}{E}

So every point on the bend has its own set: a little low on the bend, and more as you climb. Instead of hunting for where yielding begins, choose how much permanent strain counts as yielded. Engineers chose 0.2%, a strain of 0.002. No law of nature picks that number. It's small enough that a part is still essentially the shape you made it, and large enough to measure.

strain, ε stress, σ 0.002 σy slope E leaves 0.002 permanent strain 0.2% offset line unloading: same slope, E
Unloading lines from three points on the bend

Now you can see why the offset line runs parallel to the straight part. It isn't a drawing trick: it's the path the bar would take on the way down. Run it backward. Start at 0.002 on the strain axis and rise with slope EE. Where that line meets the curve is the one stress that would leave exactly 0.2% permanent strain, and that stress is the 0.2% offset yield strength, σy\sigma_y.

The Handbook's Materials Science chapter names the 0.2% offset method, and its Stress-Strain Curve for Mild Steel draws the dashed line from 0.002, but neither describes the construction.

With a table, there's nothing to draw, but the same idea works in numbers. Find εset\varepsilon_{\text{set}} for the rows on the bend. The yield strength is the stress where it reaches 0.002, interpolating if that falls between rows.

Worked Example Offset yield from a stress-strain table
1 of 3

A tension test on an aluminum alloy with E=70 GPaE = 70\ \text{GPa} gives these readings. Find its 0.2% offset yield strength.

The test's readings
StrainStress (MPa)
0.0020140
0.0040270
0.0050295
0.0060310
0.0070318
0.0100330

Starting from load and elongation

A raw sheet has no stress or strain on it yet, so convert first: σ=P/A0\sigma = P/A_0 and ε=δ/L0\varepsilon = \delta/L_0 for the rows you need. After that, it's the same job.

Watch the offset as you convert, because 0.002 is a strain, not a distance. On a 2 in. gauge, it's an elongation of (0.002)(2 in.)=0.004 in.(0.002)(2\ \text{in.}) = 0.004\ \text{in.}, and on a 100 mm gauge, it's 0.20 mm. Taking it as 0.002 in. on a 2 in. gauge finds the stress at only half the permanent strain you wanted.

A load-elongation plot is the same picture with relabeled axes, so you can leave the conversion for the end. Start the parallel line at 0.002L00.002L_0 on the elongation axis, read the load where it meets the curve, and divide by A0A_0.

Worked Example Offset yield from a load-elongation table
1 of 2

A round high-strength steel specimen, 0.505 in. in diameter with a 2 in. gauge length, has E=29,000 ksiE = 29{,}000\ \text{ksi}. Find its 0.2% offset yield strength from these readings.

The test's readings
Load (lb)Elongation (in.)
10,0000.0035
11,0000.0050
12,0000.0081
12,5000.0120

The finished plot

A question may hand you the finished plot, with the dashed line already drawn, and ask only what its marks mean.

Checkpoint
1 of 3
strain, ε stress, σ 0.002 A B C D
A tension-test curve with a dashed line

What does the slope of the dashed line represent?

A.
The 0.2% offset yield strength
B.
The ultimate tensile strength
C.
The modulus of elasticity
D.
Poisson's ratio

The yield and ultimate strengths are what design runs on. The Handbook's Table 2 - Average Mechanical Properties of Typical Engineering Materials prints them as σy\sigma_y and σu\sigma_u, 36 and 58 ksi for structural A36. In Steel Tension Members, you'll design with them as FyF_y and FuF_u. Comparing one curve with another waits for Reading Stress-Strain Curves.

Recap

  • Engineering stress and strain use the original area and the gauge length, σ=P/A0\sigma = P/A_0 and ε=δ/L0\varepsilon = \delta/L_0, so the curve describes the material, not the bar.
  • EE is the slope of the straight part: stress over strain from one reading when the line runs through the origin, and Δσ/Δε\Delta\sigma/\Delta\varepsilon from two readings when it doesn't.
  • The ultimate tensile strength is the peak, Pmax⁡/A0P_{\max}/A_0. Necking starts there, and the stress stays on the original area after it.
  • Poisson's ratio is −εlat/ε-\varepsilon_{\text{lat}}/\varepsilon in the elastic range, with εlat=Δd/d\varepsilon_{\text{lat}} = \Delta d/d, reported as a positive number.
  • A bar unloads parallel to its straight part, so a line from 0.002 with slope EE meets the curve at the stress that leaves 0.2% permanent strain. From load and elongation, the offset is 0.002L00.002L_0 of elongation, not 0.002 in.