Changing Effective Stress

Lower the water table, and the ground presses down harder without adding any soil.

Before this Lesson

In 1977 the hydrologist Joseph Poland was photographed beside a utility pole in California's San Joaquin Valley, southwest of Mendota. Signs up the pole mark where the ground had been in 1955 and, near the top, in 1925. In 52 years the ground he stood on had sunk about 9 m, nearly 30 ft.

Nobody dug that ground away or piled anything on it. Farmers pumped groundwater to irrigate, the water pressure deep down fell, and the clay layers there squeezed together under the same weight they'd always carried. This Lesson tracks what moves effective stress, and what happens when it reaches zero.

Two quantities, one subtraction

Effective stress is still σ′=σ−u\sigma' = \sigma - u, at every moment, so any change in it is

Δσ′=Δσ−Δu\Delta\sigma' = \Delta\sigma - \Delta u

That's the whole Lesson in one line. σ′\sigma' moves only when σ\sigma and uu move by different amounts, so for each event that follows, the question is which of the two it moves, and how far. In the San Joaquin Valley, pumping cut uu and left σ\sigma nearly alone, so σ′\sigma' went up.

Weight added on top

Spread a new layer of fill across a whole site, or anything else that wide, such as a broad stockpile. Picture the column of ground on one square foot, as in Effective Stress, anywhere on the site. Every column gets the same new weight on top, so every point below gains the same total stress:

Δσ=γfillHfill\Delta\sigma = \gamma_{fill} H_{fill}

A surcharge stated as a pressure is simply Δσ\Delta\sigma. Under a single footing, by contrast, the added stress spreads and fades with depth, a case the Handbook gives no method for. The water table stays at the same elevation, so in the long run uu is unchanged and σ′\sigma' gains the full Δσ\Delta\sigma. The Handbook's Vertical Stress Profiles with Surcharge diagram draws exactly this: a surcharge it calls SS, added to σ\sigma and σ′\sigma' at every depth, with uu left alone.

Worked Example A fill on a familiar site
1 of 2

At the site from Effective Stress, sand lies on a 12 ft clay layer, with the water table 6 ft down. At point A, in the middle of the clay, σ=2,368\sigma = 2{,}368 psf, u=874u = 874 psf, and σ′=1,494\sigma' = 1{,}494 psf. A 5 ft fill with γ=120\gamma = 120 lb/ft³ is spread over the whole site. Long after it's placed, what are σ\sigma and σ′\sigma' at A?

Fill, γ = 120 lb/ft³ Sand, γ = 110 lb/ft³ Sand, γsat = 125 lb/ft³ Clay, γsat = 118 lb/ft³ A New ground surface Original ground surface Water table 5 ft 6 ft 8 ft 12 ft 6 ft
The site with a new fill on top

The first day on clay

The example said "long after" for a reason: A is in clay. For σ′\sigma' to rise, the grains must press together harder, so the skeleton they form must squeeze a little tighter, and it can do that only if some water leaves the voids. Water barely compresses, and it creeps through a clay's fine pores. On the day the fill goes on, almost none has left.

Karl Terzaghi, who founded soil mechanics, pictured this as a spring in a cylinder of water, under a piston with a small hole in it.

Just loaded 600 psf Water carries: 600 psf Spring carries: 0 psf Hole = drainage Spring = soil skeleton (Δσ′) Water = pore water (Δu) Draining 600 psf Water carries: 300 psf Spring carries: 300 psf Water escapes Drained 600 psf Water carries: 0 psf Spring carries: 600 psf
The spring and the water share the load as water leaks out
The load is the same 600 psf in all three states. Only how it's shared changes.

At the instant of loading, the spring can't compress until water gets out of the way, so the water carries the whole load. As water leaks out, the piston sinks and the spring takes over, until it carries everything and the water is back at its starting pressure.

A saturated clay under a new fill does the same. Right after the fill goes on, Δu=Δσ\Delta u = \Delta\sigma and Δσ′=0\Delta\sigma' = 0. That rise in uu is the excess pore pressure. It pushes water out toward the sand above or below, and as it bleeds away, σ′\sigma' climbs to its full gain and the clay settles. How long that takes is the subject of Time Rate of Consolidation. Sand drains so fast that it reaches the long-term state at once.

Meanwhile the clay's strength hasn't changed, though its load has. That's why the end of construction is often the riskiest moment for an embankment on soft clay.

Moving the water table

Pumping, a drought, or wells around an excavation can lower the water table. For a point below the new water table, two things change in the band of soil the water leaves:

  • The band drains, so it weighs its moist γ\gamma instead of γsat\gamma_{sat}, and σ\sigma drops a little.
  • The free surface is lower, so huh_u shrinks by the drop, and uu falls by γW\gamma_W per foot of it.
Sand, γ = 110 Sand, γsat = 125 Clay, γsat = 118 A hu = 14 ft Before pumping Sand, γ = 110 Drained, γ = 110 Clay, γsat = 118 A hu = 6 ft After pumping Ground surface Old water table 6 ft 8 ft 12 ft Unit weights in lb/ft³
The same site before and after pumping
The band from 6 to 14 ft drains, and A ends up 6 ft below the water table instead of 14.

The second effect wins easily. For the site's sand, each foot of drop takes 125−110=15125 - 110 = 15 psf off σ\sigma but 62.4 psf off uu, so σ′\sigma' gains 47.4 psf. In general, with Δz\Delta z the drop,

Δσ′=[γW−(γsat−γ)]Δz=(γ−γ′) Δz\Delta\sigma' = \left[\gamma_W - (\gamma_{sat} - \gamma)\right]\Delta z = (\gamma - \gamma')\,\Delta z

It's the trade from the water-table problems in Effective Stress: each foot of the band swaps γ′\gamma' for γ\gamma. Every point below the new water table gains the same amount. In a clay the gain arrives only as its water drains, which is why the valley took decades to sink, and questions ask for the long-term value.

Use the moist unit weight the question gives for the drained band. If it says the soil stays saturated, σ\sigma doesn't change and the gain is γW\gamma_W per foot. Raising the water table runs the trade backward, until the water reaches the ground.

Worked Example Pumping the site down
1 of 2

Wells lower the water table at the site in the figure from 6 ft down to the top of the clay, 14 ft down. The drained sand weighs 110 lb/ft³, like the sand that was always above the water table. With no fill this time, find the long-term change in σ′\sigma' at A, which started at 1,494 psf, and its new value.

Fills, pumps, and floods each move σ\sigma, uu, or both. Before you calculate, it pays to know which way each one goes.

Checkpoint
1 of 3

A long drought lowers the water table in a deep sand by 2 m, and the sand it leaves dries to its moist unit weight. At a point deeper in the same sand, how do σ\sigma, uu, and σ′\sigma' change?

A.
σ\sigma unchanged, uu falls, σ′\sigma' rises
B.
σ\sigma falls slightly, uu falls more, σ′\sigma' rises
C.
σ\sigma and uu fall by the same amount, σ′\sigma' unchanged
D.
σ\sigma rises, uu falls, σ′\sigma' rises

Pressure from below

Sometimes a clay caps a sand fed with water from higher ground, such as distant hills where the sand comes to the surface. The clay keeps that water from rising, so it's under more pressure than the water table above would give: an artesian, or confined, aquifer.

The sand's water pushes up on the base of the clay with the pressure that holds up a piezometer's water, so there, huh_u is the height of the piezometric level above the base, not the depth below the water table. The Handbook calls huh_u the "uplift or pressure head," and here it earns the first name.

Clay, γsat = 120 lb/ft³ Sand (confined aquifer) Ground surface and water table Piezometric level 8 ft hu = 28 ft 20 ft B
A clay layer over an artesian sand
The water table is at the ground surface, and the piezometer fills to 8 ft above it.

At B, the base of the clay:

σ=120(20)=2,400 psfu=(20+8)(62.4)=1,747 psfσ′=2,400−1,747=653 psf\begin{aligned} \sigma &= 120(20) = 2{,}400\ \text{psf} \\ u &= (20 + 8)(62.4) = 1{,}747\ \text{psf} \\ \sigma' &= 2{,}400 - 1{,}747 = 653\ \text{psf} \end{aligned}

Ordinary groundwater would give u=1,248u = 1{,}248 psf and σ′=1,152\sigma' = 1{,}152 psf, so the extra 8 ft of head takes 499 psf off the grains. Raise the level to about 18.5 ft above the ground, and σ′\sigma' at B reaches zero: the water would lift the clay off the sand.

Questions may give the piezometric level as a height, a depth, or an elevation. Turn it into a height above the point. Inside the clay, uu changes from one end to the other as water seeps up, so questions ask at the base.

Checkpoint

The ground at a site is at El. 40.0 m. Its 2 m of sand, with γ=17.5\gamma = 17.5 kN/m³, lies on a 5 m clay layer with γsat=19.0\gamma_{sat} = 19.0 kN/m³, and the water table is at the top of the clay. Below the clay is a confined sand, and a piezometer screened in it fills to El. 41.5 m. What is the effective stress at the base of the clay?

kPa

When effective stress reaches zero

Push water up through sand instead of clay, and you can turn the effect up as far as you like. Picture a tank with a layer of sand LL thick on a screen, under standing water. A tube feeds water in below the screen from a reservoir whose surface stands Δh\Delta h above the tank's, so water flows up through the sand and spills over the top.

Reservoir Overflow Sand Flow L Δh
Water forced up through a layer of sand

At the base of the sand, σ\sigma hasn't changed: it's still the water above plus the sand. But uu there now holds up a column reaching the reservoir's level, so it has gained γWΔh\gamma_W \Delta h, which comes straight off the still-water σ′=γ′L\sigma' = \gamma' L:

σ′=γ′L−γWΔh\sigma' = \gamma' L - \gamma_W \Delta h

The water flowing up drags on the grains and holds up part of their weight. σ′\sigma' reaches zero when Δh/L\Delta h / L, the gradient through the sand, reaches the critical hydraulic gradient:

ic=γ′γW=γsat−γWγWi_c = \frac{\gamma'}{\gamma_W} = \frac{\gamma_{sat} - \gamma_W}{\gamma_W}

The Handbook prints the second form, unnamed, under Factor of safety against seepage liquefaction. From GsG_s and ee, the Archimedes form γ′=(Gs−1)γW/(1+e)\gamma' = (G_s - 1)\gamma_W/(1 + e) gives

ic=Gs−11+ei_c = \frac{G_s - 1}{1 + e}

A typical sand, with Gs=2.65G_s = 2.65 and e=0.65e = 0.65, has ic=1.00i_c = 1.00, and most sands land near 1, because submerged sand weighs about as much as water. Keep γ′\gamma' on top: γsat\gamma_{sat} gives a number near 2.

What's left of the sand? Its grains touch, but nothing presses them together, and a clean sand's strength is pure friction, which needs that pressing. So with σ′=0\sigma' = 0 its strength is zero, whatever its friction angle. Dense or loose, it behaves like a thick liquid, and a stone or a footing sinks into it. That's quicksand. It isn't a kind of sand but any sand with enough water rising through it, and when the flow stops, the same sand is firm again. At an excavation the floor seems to boil, so the failure is also called boiling.

Checkpoint
1 of 2

A sand has Gs=2.68G_s = 2.68 and e=0.72e = 0.72. What is its critical hydraulic gradient?

Checkpoint

Part of an excavation floor is a dense, gravelly sand with a friction angle of 40°, and part is a loose, fine sand with a friction angle of 30°. Both have γsat=124\gamma_{sat} = 124 lb/ft³, and water seeps up through both. Which statements are true?

A.
The dense sand needs a steeper exit gradient to go quick, because of its higher friction angle.
B.
Both go quick at the same exit gradient.
C.
Either one becomes quicksand at that gradient, and firm again when the flow stops.
D.
Once both are quick, the dense sand still has more strength than the loose one.

Boiling at an excavation

In the field, the tank is an excavation. Steel sheet piles driven into sand hold back a river or high groundwater while the excavation goes in beside them, and pumps keep the water inside down at the floor. Outside it stands Δh\Delta h higher, so water flows down along the sheets, under their tips, and up through the floor.

Sheet pile River Excavation Water kept at floor level Sand Δh Flow path, length L Exit gradient, ie
Seepage under a sheet-pile wall

Where it comes out, the water is flowing up through sand, as in the tank, and the gradient there is the exit gradient, iei_e. If the head is lost evenly along a flow path of length LL, ie=Δh/Li_e = \Delta h / L. Questions give either iei_e or a head loss and a flow length. The Handbook's Factor of safety against seepage liquefaction compares the two gradients:

FSs=icieFS_s = \frac{i_c}{i_e}

Above 1, the floor holds. Keep ici_c on top, because the inverse reads a safe floor as a failing one. For a required factor of safety, turn it around: the largest allowable exit gradient is ie,max=ic/FSsi_{e,max} = i_c / FS_s, so the flow path must be at least

Lmin=Δhie,max=FSs ΔhicL_{min} = \frac{\Delta h}{i_{e,max}} = \frac{FS_s\,\Delta h}{i_c}

That's why one standard fix for a boiling excavation is longer sheets: driving them deeper spreads the same head loss over more sand.

Worked Example Is the cofferdam floor safe?
1 of 4

A sheet-pile cofferdam holds back a river whose surface stands 4.5 m above the excavation floor, where pumps keep the water inside. The shortest flow path, down the outside of the sheets, under the tips, and up into the floor, is 12.0 m long. The sand has γsat=19.6\gamma_{sat} = 19.6 kN/m³. Find the factor of safety against boiling.

Recap

  • Δσ′=Δσ−Δu\Delta\sigma' = \Delta\sigma - \Delta u, so σ′\sigma' changes only when σ\sigma and uu change by different amounts.
  • A widespread fill adds Δσ\Delta\sigma at every depth, and in the long run σ′\sigma' gains all of it. On saturated clay, at first Δu=Δσ\Delta u = \Delta\sigma.
  • Lowering the water table raises σ′\sigma' by (γ−γ′)Δz(\gamma - \gamma')\Delta z, with the moist γ\gamma in the drained band. Flooding saturated ground leaves σ′\sigma' alone.
  • Under a clay over an artesian sand, huh_u at its base is the height of the piezometric level above it.
  • At ic=(γsat−γW)/γWi_c = (\gamma_{sat} - \gamma_W)/\gamma_W, near 1, upward seepage drives σ′\sigma' to zero, and a cohesionless soil has no strength, whatever its friction angle. Against boiling, FSs=ic/ieFS_s = i_c / i_e.