Shear Strength

Soil strength is friction between grains, and friction needs the grains pressed together.

Before this Lesson

In 1964, an earthquake shook Niigata, Japan. In one apartment complex, several four-story concrete buildings came through it barely cracked, yet ended up tilted at wild angles, one nearly on its side. The buildings hadn't failed. The ground under them had.

That ground was loose sand full of water. The shaking tried to pack the grains tighter, which meant squeezing water out, and water can't leave in the few seconds of shaking. So its pressure rose until it held up the grains, and the buildings, by itself. With nothing pressing them together, the grains couldn't grip each other, and the sand flowed like a thick liquid. This Lesson measures that grip and shows when water takes it away.

Friction between grains

Fill a shoebox with dry sand, cut it in two at mid-height, and set a weight on the lid. Push the top half sideways, and at some point the sand gives way along the cut. Double the weight, and it takes about twice the push. That's the crate on the floor from Static Friction, with sand on both sides of the sliding surface: friction rises to μsN\mu_s N and no further, with tan⁡ϕ=μs\tan\phi = \mu_s. Here the sliding surface runs through the soil, so ϕ\phi belongs to the soil, and it's called the angle of internal friction.

The laboratory version is the direct shear test. A normal force PP presses the specimen, and a shearing force TT pushes the upper half until the soil fails along the plane between the halves. Dividing both by that plane's area AA gives stresses, which don't depend on the box's size. The Handbook's Geotechnical section gives both:

σN=PAτ=TA\sigma_N = \frac{P}{A} \qquad \tau = \frac{T}{A}
Shear plane, area A P T
A direct shear box
The lower half is held fixed. The soil can fail only along the plane between the halves, of area AA.

In stresses, the friction ceiling is σNtan⁡ϕ\sigma_N\tan\phi. Many soils, clays especially, also hold together a little with no normal stress at all, from bonds between particles. That's the cohesion, cc, and it simply adds on. The Handbook prints the result under Shear stress at failure. It's the Mohr-Coulomb criterion:

τF=c+σNtan⁡ϕ\tau_F = c + \sigma_N\tan\phi

τF\tau_F is the shear strength, the most shear stress a plane can carry with σN\sigma_N pressing on it. Plot it against σN\sigma_N and you get the failure envelope, a line that crosses the shear axis at cc and rises at the angle ϕ\phi. The stresses on any plane are a point on this plot. Below the line, the plane holds. On it, the plane fails. Above it is out of reach, because the soil gives way on the way there. So checking a plane takes one comparison: the shear stress acting on it against τF\tau_F at the same normal stress.

Normal stress, σN Shear stress, τ φ c Holds Fails Mohr-Coulomb failure: τF = c + σN tan φ Can't be reached
The Mohr-Coulomb failure envelope
The Handbook's diagram also draws a Mohr circle touching the line, which belongs to triaxial tests.

Adapted from the FE Reference Handbook, Geotechnical section, Mohr-Coulomb failure diagram

Only the grains grip

Now fill the voids with water. On any plane through the soil, the water pushes with pressure uu over practically the whole area, as it did on the horizontal slice in Effective Stress. That push holds the grains apart and carries part of the normal stress. And water has no shear strength at all: you can't build a pile of it. So the share of σN\sigma_N the water carries buys no friction. Only the share the grain contacts carry, the effective normal stress, presses grains together:

σN′=σN−uτF=c′+σN′tan⁡ϕ′\sigma'_N = \sigma_N - u \qquad \tau_F = c' + \sigma'_N\tan\phi'

The primes on c′c' and ϕ′\phi' mark parameters measured in terms of effective stress. The Handbook prints the criterion without primes and calls σN=P/A\sigma_N = P/A the total normal stress, which is what the box applies. In a test slow enough for the water to drain, the pore pressure stays near zero, so in the box the two are equal. In the ground they aren't, and σN′\sigma'_N is the one that goes in.

On a horizontal plane at some depth, the normal stress is the vertical stress, so σN′\sigma'_N is the σ′\sigma' from the effective-stress profile. That's why this topic keeps saying it: find σ′\sigma' first.

Worked Example Strength on a plane 16 ft down
1 of 3

A silty sand has c′=100c' = 100 psf and ϕ′=30∘\phi' = 30^\circ. The water table is 6 ft below the ground, and the sand weighs 112 lb/ft³ above it and 124 lb/ft³ below it. A nearby load puts a shear stress of 700 psf on a horizontal plane 16 ft down. Does the soil fail on that plane?

When the pore pressure rises

Hold the total stress on a plane fixed and raise the pore pressure, the way rain soaking into a hillside or a rising water table does. Each psf the water picks up is a psf the grains no longer carry, so σN′\sigma'_N falls by Δu\Delta u, and the strength falls with it:

ΔτF=Δutan⁡ϕ′\Delta\tau_F = \Delta u\tan\phi'

The cohesion doesn't change. Only the friction term does. And the shear stress the plane has to carry, set by gravity and the loads, doesn't change either. On the envelope plot, the plane's stress point slides left by Δu\Delta u at the same height, toward the line. Raise uu far enough and the point reaches the line, and the soil fails without a pound of load being added. That's how many landslides start, as Slope Stability shows, and it's what took the grip from the sand at Niigata.

Effective normal stress, σ′N Shear stress, τ c′ φ′ Δu Δu tan φ′ Before After τF = c′ + σ′N tan φ′
A rise in pore pressure moves the stress point toward the envelope
The shear stress stays the same while σN′\sigma'_N drops by Δu\Delta u. Along the envelope, that costs Δutan⁡ϕ′\Delta u\tan\phi' of strength.
Checkpoint
1 of 2

On a plane through a sand, the total normal stress is 2,400 psf and the pore pressure is 750 psf. The sand has c′=0c' = 0 and ϕ′=34∘\phi' = 34^\circ. What is the shear strength on the plane?

psf

Two tests fix the envelope

Each direct shear test holds one normal force on the specimen and records the shearing force at failure. Divided by the area, that pair is one point on the envelope. One point can't fix a line that has an intercept, so you run at least two tests at different normal forces. The slope between the two points is tan⁡ϕ\tan\phi, and the cohesion is whatever strength is left at either point once its friction is taken off:

tan⁡ϕ=τ2−τ1σ2−σ1c=τ1−σ1tan⁡ϕ\tan\phi = \frac{\tau_2 - \tau_1}{\sigma_2 - \sigma_1} \qquad c = \tau_1 - \sigma_1\tan\phi

To get ϕ\phi itself, take tan⁡−1\tan^{-1} of the slope, with the calculator in degrees. And watch the area's units. A specimen measured in inches gives stresses in psi, and 1 psi is 144 psf. Skipping the area altogether is sneakier: ϕ\phi still comes out right, because both stresses in the slope are divided by the same AA, but cc comes out as a force. Tests are usually sheared slowly enough for the water to drain, so the results are c′c' and ϕ′\phi'.

Worked Example Two tests on one soil
1 of 2

Two direct shear tests on a 60 mm × 60 mm specimen of the same soil fail at these forces: P=360P = 360 N and T=252T = 252 N in the first, P=720P = 720 N and T=432T = 432 N in the second. Find cc and ϕ\phi.

A clean sand needs one test

A clean sand has no cohesion. Its grains don't stick to each other, which is why you can't carve a vertical wall out of dry sand. All of its strength is friction, so c=0c = 0, and its envelope passes through the origin. That's the second point for free, so one test fixes the line:

tan⁡ϕ=τFσNϕ=tan⁡−1τFσN\tan\phi = \frac{\tau_F}{\sigma_N} \qquad \phi = \tan^{-1}\frac{\tau_F}{\sigma_N}

With ϕ\phi in hand, the strength at any other normal stress is σNtan⁡ϕ\sigma_N\tan\phi. A line through the origin also means strength is proportional to normal stress: double σN′\sigma'_N and you double the strength. So you can even scale straight from the test without finding the angle.

Checkpoint
1 of 2

A direct shear test on a dry, clean sand in a 60 mm × 60 mm box fails at a shearing force of 315 N under a normal force of 450 N. What is the sand's friction angle?

°

Fast loads and slow ones

Everything so far took uu as the steady groundwater pressure. Right after a load goes on, it may not be.

Load a saturated soil, and at first the water takes the new load, because water barely compresses and the grains can't move closer until some of it leaves. That extra pressure, the excess pore pressure, lasts until the water escapes. A clean sand's open voids let it out in minutes, long before a building is finished, so a sand is analyzed as drained, with c′=0c' = 0 and ϕ′\phi'.

A saturated clay's tiny voids take months or years. Put a fill on one in a few weeks, and at the end of construction the water still carries nearly all of it. σ′\sigma' hasn't grown, so the strength hasn't either, and more fill wouldn't help: the water takes that too. So right after quick loading, the clay's strength is one fixed number, its undrained shear strength sus_u. Plotted against total normal stress, the envelope is horizontal:

ϕ=0τF=c=su\phi = 0 \qquad \tau_F = c = s_u

That's an undrained analysis: sus_u takes the place of cc, and the friction term drops out. The grains haven't lost their friction. ϕ=0\phi = 0 only says that while the water can't escape, extra load buys no extra strength. The Handbook says nothing about drained or undrained strength and has no symbol for sus_u, so a question gives it to you, sometimes calling it the cohesion.

Over the years, the water seeps out and the fill's weight passes to the grains. Then the clay is drained too: uu is back to the steady groundwater pressure, with no excess, and the strength is c′+σN′tan⁡ϕ′c' + \sigma'_N\tan\phi'. That's the long-term analysis. So the real question is how fast the load comes on against how fast the soil drains. An earthquake is fast even for a sand, and that was Niigata.

σ′N τF Clean sand φ′ c′ = 0 σ′N τF Clay, drained φ′ c′ σN τF Saturated clay, undrained su φ = 0
Envelopes of a clean sand, a drained clay, and an undrained saturated clay
The undrained envelope is plotted against total normal stress, the other two against effective normal stress.
Checkpoint
1 of 3

Three specimens of one soil are sheared in direct shear under normal stresses of 1, 2, and 3 ksf. Every one fails at a shear stress of 0.9 ksf. What are cc and ϕ\phi for these tests?

A.
c=0c = 0, ϕ=42.0∘\phi = 42.0^\circ
B.
c=0c = 0, ϕ=24.2∘\phi = 24.2^\circ
C.
c=0c = 0, ϕ=16.7∘\phi = 16.7^\circ
D.
c=0.9c = 0.9 ksf, ϕ=0\phi = 0

Recap

  • Soil strength is friction plus a little cohesion, τF=c+σNtan⁡ϕ\tau_F = c + \sigma_N\tan\phi, with tan⁡ϕ\tan\phi playing the friction coefficient. A plane whose shear stress falls below the envelope holds, and one on it fails.
  • Only the grains grip, so strength runs on effective stress: τF=c′+σN′tan⁡ϕ′\tau_F = c' + \sigma'_N\tan\phi'. Find σ′\sigma' first. A rise in pore pressure at constant total stress costs Δutan⁡ϕ′\Delta u\tan\phi' of strength.
  • A direct shear test gives σN=P/A\sigma_N = P/A and τ=T/A\tau = T/A. Two tests give tan⁡ϕ\tan\phi as the slope and cc as the intercept. Keep the area's units straight.
  • A clean sand has c=0c = 0, so one test gives ϕ=tan⁡−1(τF/σN)\phi = \tan^{-1}(\tau_F/\sigma_N), and its strength is proportional to σN′\sigma'_N.
  • Drained, for a sand or for a clay in the long term: c′c' and ϕ′\phi', with no excess pore pressure. Undrained, for a saturated clay loaded quickly: ϕ=0\phi = 0 and τF=su\tau_F = s_u, a horizontal envelope.