Compressibility and Preconsolidation

Clay remembers the heaviest load it has ever carried.

Before this Lesson

Two warehouses built from the same plans stand a few miles apart, each on 20 ft of the same gray clay. Ten years after opening, one floor has settled about 11 inches. The other has settled less than 2.

The difference is history. Long ago the second site lay under another 20 ft or so of sand and gravel, which a river later carried away. Its clay was squeezed under that weight once, and it still behaves as if it remembers. This Lesson reads that memory and shows why it cuts the settlement so sharply.

Squeezing a slice of clay

A clay settles as water leaves its voids and its grains pack closer, as Changing Effective Stress showed. To predict how far a layer will go, the lab squeezes a slice of it.

In a consolidation test, or oedometer test, a disc of undisturbed clay about 2.5 in across sits in a stiff ring between two porous stones. The ring stops it from bulging sideways, so it can only get shorter, just like a layer under a fill that covers a whole site.

Load Dial gauge Loading cap Water drains out Porous stone Clay specimen Rigid ring Porous stone
A consolidation test, in cross section

Each load stays on until the water has drained and the height stops changing, often about a day, and then the load is doubled. With the excess pore pressure gone, every stress in the test is an effective stress, σ′\sigma', and the test reports the void ratio at each one.

Height follows 1 + e

How does a height become a void ratio? The grains and the water are, for practical purposes, incompressible, so when the specimen shortens, only the voids shrink, by exactly the water that left. In the phase box from Soil as Three Phases with one unit of solids, a saturated clay's water is ee tall and the whole box is 1+e1 + e. Squeeze it, and the solids stay at 1.

Water Solids e0 1 1 + e0 Water Solids ef 1 1 + ef Δe Squeeze Before After
The solids stay put; only the voids shrink
Drawn to scale for e0=1.2e_0 = 1.2 and ef=0.9e_f = 0.9: the box loses 0.3 of its original 2.2.

So a clay's height is proportional to 1+e1 + e. When the void ratio falls by Δe\Delta e from its initial e0e_0, the box loses Δe\Delta e of its original 1+e01 + e_0, and so does every slice of a layer of thickness HH:

ΔH=Δe1+e0 H\Delta H = \frac{\Delta e}{1 + e_0}\,H

The Handbook prints this under Ultimate consolidation settlement in soil layer as SULT=εvHSS_{ULT} = \varepsilon_v H_S, with the vertical strain εv=ΔeTOT/(1+e0)\varepsilon_v = \Delta e_{TOT}/(1 + e_0) and HSH_S the layer's thickness. "Ultimate" means after all the excess pore pressure has drained. Divide by 1+e01 + e_0, not 1+ef1 + e_f: a strain is a change over the original length. Run it backward, and a measured settlement gives Δe=(ΔH/H)(1+e0)\Delta e = (\Delta H / H)(1 + e_0).

Worked Example From a test's void ratios to a settlement
1 of 3

A 4.0 m clay layer has a void ratio of 1.15 at its middle. A consolidation test on a sample from there shows that under the effective stress a new building will bring, the void ratio falls to 1.02. Find the layer's ultimate settlement.

Equal ratios, equal squeezes

Plot a test's void ratios against σ′\sigma', and a pattern shows up. Once the clay carries more than it ever has, each doubling of the load squeezes out about the same void ratio. Going from 100 to 200 kPa takes out as much as going from 400 to 800, though the second step adds four times the stress.

One way to picture it: a soft clay's flat particles stack loosely, like a house of cards. As the stack tightens, each particle leans on more neighbors, so the clay stiffens roughly in step with the stress it already carries. An extra 100 kPa is a big deal to a clay at 100 kPa and a small one to a clay at 800.

A log scale gives every doubling the same length, just as it gives every factor of 10 one log cycle. So on a plot of ee against log⁡σ′\log \sigma', the e–log σ′ curve, those points fall on a straight line.

100 200 400 800 0.70 0.80 0.90 1.00 Effective stress, σ′ (kPa, log scale) Void ratio, e ×2 0.09 ×2 0.09 ×2 0.09
Each doubling of σ′ squeezes out the same void ratio
On the log axis the doublings are evenly spaced, and the line drops 0.09 across each one.

This is the virgin compression line, virgin because each stress on it is one the clay has never carried. The Handbook defines its slope, the drop in void ratio per log cycle, as the compression index:

CC=ΔeΔlog⁡σ′C_C = \frac{\Delta e}{\Delta \log \sigma'}

In the figure, CC=0.090/0.301=0.30C_C = 0.090/0.301 = 0.30. The log is base 10, and ln⁡\ln would divide the same Δe\Delta e by a number 2.3 times larger. Mind the symbol, too: CCC_C also names the coefficient of curvature from Grain Size and Plasticity. About sieves, it's curvature. About settlement, it's compression.

The clay remembers

Now run the test differently. Load a soft clay down the virgin line from A to B, at 800 kPa, then unload it to 200 kPa at C. It swells back only a little, because the particles packed at B don't spring back. Then reload it.

100 200 400 800 1600 0.60 0.70 0.80 0.90 1.00 Effective stress, σ′ (kPa, log scale) Void ratio, e Unload Reload CR CC A C D B Heaviest stress so far
Unload and reload: the clay remembers B
A to B and B to D lie on the virgin line. Unloading to C and reloading back to B follow a much flatter line.

From C back to B the clay is stiff, retracing a flat line and barely compressing. At B it bends onto the virgin line as if it had never left, and each doubling costs the full 0.09 again.

That's the memory. A clay keeps a record of the heaviest effective stress it has ever carried, its preconsolidation stress, σc′\sigma'_c, which the Handbook calls the past maximum consolidation stress. Below σc′\sigma'_c, the clay follows the flat recompression branch, and the Handbook defines its slope the same way, as the recompression index:

CR=ΔeΔlog⁡σ′C_R = \frac{\Delta e}{\Delta \log \sigma'}

A sample from the ground has already been through the first half of that experiment: geology loaded it, and erosion or the trip out of the borehole unloaded it. So its test curve starts flat and bends onto the virgin line at σc′\sigma'_c. A real test rounds the corner, and σc′\sigma'_c sits about where the flat and steep straight parts would meet if extended. The Handbook draws this as its Soil Consolidation Curve Over Consolidated Clay:

Range of recompression Range of virgin compression CR CC e0 σ′0 σ′c σ′f Effective stress, σ′ (log scale) Void ratio, e
The consolidation curve of an overconsolidated clay
σ0′\sigma'_0 is the effective stress in the ground today, and σf′\sigma'_f the stress after a new load. The slope changes at σc′\sigma'_c.

Adapted from the FE Reference Handbook, Soil Consolidation Curve Over Consolidated Clay

Here the two warehouses part ways. Load two layers of the same clay from the same σ0′\sigma'_0 to the same σf′\sigma'_f. If one layer's σc′\sigma'_c equals its σ0′\sigma'_0, its whole path runs down the steep virgin line. If the other's σc′\sigma'_c is above σf′\sigma'_f, its whole path stays on the flat branch, so its Δe\Delta e, and its settlement, is a fraction as large. A load that pushes past σc′\sigma'_c uses both slopes, which Consolidation Settlement turns into equations.

Checkpoint
1 of 3

Two points on the straight, steep part of a clay's test curve are e=0.95e = 0.95 at 200 kPa and e=0.78e = 0.78 at 800 kPa. What is the clay's compression index?

Normally consolidated or overconsolidated

To know which branch a layer starts on, compare its memory with what it carries today. The overconsolidation ratio is

OCR=σc′σ0′\text{OCR} = \frac{\sigma'_c}{\sigma'_0}

with σ0′\sigma'_0 the effective stress in the ground now. The Handbook uses the OCR in its at-rest earth pressure coefficient but never defines it, and this ratio is the standard meaning.

  • At OCR=1\text{OCR} = 1, the clay is normally consolidated. It carries the heaviest load it ever has, so it sits on the virgin line, and any new load follows CCC_C.
  • At OCR>1\text{OCR} > 1, the clay is overconsolidated. It has carried more before, so a new load travels the flat branch until it reaches σc′\sigma'_c.

Both stresses are effective. σc′\sigma'_c comes from a drained test, so take σ0′\sigma'_0 from the profile as in Effective Stress, with the water pressure out, at the middle of the layer, the point that stands for the whole layer.

Worked Example Is the clay overconsolidated?
1 of 3

At a site, 10 ft of sand lies on a 16 ft clay layer, with the water table 6 ft down in the sand. The sand weighs 110 lb/ft³ above the water table and 124 lb/ft³ below it, and the clay's γsat\gamma_{sat} is 118 lb/ft³. A consolidation test on a sample from the middle of the clay gives σc′=2,700\sigma'_c = 2{,}700 psf. Find the OCR and classify the clay.

Where the memory comes from

A clay becomes overconsolidated whenever something raises its effective stress for a while and then takes the rise away. The clay keeps the high-water mark.

  • Erosion strips off overlying soil, as at the second warehouse. σ\sigma at the clay drops while uu stays put, so σ′\sigma' falls below its old maximum. A glacier that melts away does the same.
  • Desiccation, drying near the surface, pulls the water left in the pores into tension, like water drawn up a straw. A negative uu makes σ′=σ−u\sigma' = \sigma - u climb, often far above what the overburden ever supplied, and the clay keeps that memory once it's wet again. That's why many soft clay deposits have a stiff crust a few feet thick.

Changing Effective Stress showed other ways to raise σ′\sigma' without adding anything on top. They count too.

Checkpoint

Which of these histories would leave a clay layer overconsolidated today?

A.
A river cut down through the valley and carried away 30 ft of the sand that covered the clay.
B.
Sediment has kept settling onto the clay from the lake above it ever since the clay was laid down.
C.
Pumping held the water table 25 ft lower for several decades, and it has since recovered to its old level.
D.
A reservoir flooded the site, which was already saturated to the surface, and the water over it has slowly deepened by 10 ft.
E.
A long drought dried and cracked the top few feet of the clay before the rains returned.

An estimate from the liquid limit

A consolidation test takes days and an undisturbed sample. When a question gives only index properties, the Handbook offers two correlations, under Compression index and Recompression index:

CC=0.009(LL−10)CR=CC6C_C = 0.009(LL - 10) \qquad C_R = \frac{C_C}{6}

The first says a clay that holds more water at its liquid limit has more open structure to collapse. Enter LLLL as a percentage: 45% goes in as 45. It's a rough fit to ordinary normally consolidated clays, so a test beats it when you have one. The second says the flat branch is about a sixth as steep as the virgin line, which is why the second warehouse settled about a sixth as much.

Checkpoint
1 of 2

A normally consolidated clay has a liquid limit of 52% and a plastic limit of 24%. Estimate its compression index.

Recap

  • Only the voids shrink, so height is proportional to 1+e1 + e, and ΔH=Δe H/(1+e0)\Delta H = \Delta e\,H/(1 + e_0).
  • Equal ratios of σ′\sigma' squeeze out equal Δe\Delta e, so the virgin line is straight on a log axis, with slope CC=Δe/Δlog⁡σ′C_C = \Delta e/\Delta \log \sigma' (base 10).
  • Below σc′\sigma'_c, the heaviest effective stress the clay has carried, it follows the flat branch, slope CRC_R; past it, the steep virgin line. That's why an overconsolidated clay settles less.
  • OCR=σc′/σ0′\text{OCR} = \sigma'_c/\sigma'_0 at mid-layer: 1 is normally consolidated, above 1 overconsolidated, from erosion, desiccation, or a water table lowered and later restored.
  • Without a test, CC=0.009(LL−10)C_C = 0.009(LL - 10) and CR=CC/6C_R = C_C/6. This CCC_C isn't the coefficient of curvature.