Soil
behaviour in compression
Ronald B.J. Brinkgreve
Slide 1:
The learning objective of this lecture is:
To recognize and explain soil behaviour in compression
Slide 2:
The term compression is associated with a stress development that leads to a higher mean stress.
Typical modes of compression are:
Isotropic compression. This occurs, for example, during the application of the cell pressure in an
isotropically consolidated triaxial test.
>
One-dimensional compression. This type of compression is obtained, for example, in an
oedometer test or a constant rate of strain test. One-dimensional compression does not only
lead to an increase of the mean stress, but also to an increase in deviatoric stress, since the
difference between the vertical stress and the horizontal stress increases with further
compression. However, the main stress development is compression.
>
Typical phenomena observed in compression are:
Settlement: For example the settlement due to land reclamation or elevation
Consolidation: This is the time-dependent dissipation of excess pore pressures, causing
settlements. The settlement is caused by a transfer of stress from the pore water on to the soil
skeleton.
Secondary compression: This is the time-dependent settlement as a result of creep, after the
excess pore pressures have already dissipated.
De-structuration: This is the breakage of fabric or structure in natural soils upon loading beyond
the pre-consolidation stress level, which causes a significant increase in settlement.
Under conditions of one-dimensional compression one usually finds a constant ratio between
horizontal and vertical effective stress. This is the coefficient of lateral stress at rest, K0.
Note that soils do not really fail in compression. Although irreversible (plastic) strains might occur in
compression, soils do not really get in a state of failure under (isotropic or one-dimensional)
compression.
Slide 3:
Here is picture of a series of oedometer tests, with an example of a diagram that is obtained from
such a test. Here, the test is called consolidation test, which is a name often used for this test in the
United States. The diagram shows the vertical strain (in percent), plotted as a function of the applied
vertical stress, plotted on a logarithmic scale. Plotting stress versus strain gives an idea about
1
stiffness. Starting from zero, the line first goes into primary compression, then into unloading,
followed by reloading. You can observe the difference in stiffness between primary compression,
unloading and reloading. You can also see some hysteresis, which is the area between the latter two
lines.
Slide 4:
We can schematize an oedometer test apparatus as a stiff ring enclosing a soil sample. The soil
sample is subjected to vertical loading with lateral displacement constraints. The diagram on the
right shows the results of an oedometer test on Hostun sand of two different densities: Dense and
Loose. In this case, the axial strain is plotted along the horizontal axis and the axial stress is plotted
along the vertical axis. The steeper the curve, the stiffer the behaviour. It can be observed that the
dense sand is stiffer than the loose sand, and that, in general, unloading and reloading behaviour are
stiffer than primary loading behaviour.
Slide 5:
Here are the more curves for Dense and Loose Hostun sand, obtained from Jacques Desrues in
Grenoble. The stiffness that we measure from oedometer tests is the oedometer modulus, Eoed. This
stiffness is measured as the tangent stiffness along the curve at a particular vertical stress level. Note
that this modulus is different than the Youngs modulus E; we will get back to this in another lecture.
Slide 6:
In a standard oedometer test the lateral stress is not measured, but this can be done in a special
instrumented oedometer test. If we then plot the horizontal stress as a function of the lateral stress,
we can measure the K0-value, which is the ratio between horizontal stress and vertical stress. The
figure shows such a diagram for dense Monterey sand (after Sehn, 1990), with the horizontal stress
on the vertical axis and the vertical stress on the horizontal axis (I would personally plot this the
other way around). Anyway, you can see that during first loading, the ratio of horizontal over vertical
stress remains more or less constant, whereas during unloading the horizontal stress decreases much
less, such that the K0-value increases.
Slide 7:
In fact, K0 is defined in terms of effective stresses, and thus relates the horizontal effective stress to
the vertical effective stress. The lateral stress and K0-value also depend on the over-consolidation
ratio OCR. The diagram shows the K0-value as a function of the over-consolidation ratio for
Monterey sand of two different initial densities (after Wright, 1969). It can be seen that a higher OCR
gives a higher K0-value.
Slide 8:
Based on such measurements in the past, researchers have proposed empirical formulas to estimate
the ratio of horizontal over vertical effective stress. The most well-known formula is the one by Jaky
(1948), which says that K0 (for a normally-consolidated soil) is approximately equal to 1 minus the
sinus of the friction angle, phi. Later, this formula was extended by Mayne & Kulhawy to include the
effect of over-consolidation: K0 is approximately equal to (1 minus sinus phi) times OCR to the power
sinus phi. The power is usually simplified to a half (considering a friction angle of 30 degrees; sinus
phi = 0.5), which means that K0 is considered proportional to the square root of OCR.
Slide 9:
In some natural clays we can see another phenomenon, which is structure or bonding. The left
diagram shows the change of void ratio as a function of the vertical effective stress on a logarithmic
scale for a stiff clay. In fact, a change in void ratio is equivalent to a change in volumetric strain, and
2
since the lateral strain is zero in an oedometer test, the volumetric strain is equal to the vertical
strain. So this is just a different (but very common) way of plotting the results of an oedometer test.
The black dots show the behaviour of the natural clay, whereas the open dots show the behaviour of
a remoulded or reconstituted sample of the same clay. When plotting results in such a diagram, we
can obtain the so-called compression index, c sub c, for primary loading, and the swelling index, c sub
s, for swelling or unloading. The transition from the initial reloading part into primary loading is
indicated by the pre-consolidation stress. It can be seen that the natural clay has an extended
reloading part with an apparently higher pre-consolidation stress, but when this stress level is
reached, there is a high reduction in stiffness.
The right-hand diagram shows similar results for a soft structured clay, but now with the isotropic or
mean effective stress on a logarithmic scale on the horizontal axis. In this case the initial (reloading)
stiffness is very high, but as soon as the pre-consolidation stress is reached, the structure is
destroyed. This de-structuration is associated with a dramatic decrease in void ratio, so apparently
a very low stiffness, after which a more reasonable primary loading stiffness is obtained. The latter
coincides with the intrinsic compression line as obtained from reconstituted samples. The message
here is: be careful when loading structured natural clays: as soon as you pass the pre-consolidation
stress, a significant settlement can occur due to the effect of de-structuration.
Slide 10:
For clay type of soils, distinction can be made between primary compression and secondary
compression. The first is related with the process of undrained loading and consolidation; the second
is related to the time-dependent settlement after all excess pore pressures have dissipated. When
plotting vertical strain, volumetric strain or void ratio as a function of time on a logarithmic scale, a
typical S-shaped curve is obtained as shown in this diagram from Olsson, 2010. From the secondary
compression line we can obtain the creep index, alpha sub s or C sub alpha epsilon or alpha e,
depending on whether the vertical axis is strain or void ratio.
Slide 11:
In fact, the time-dependency in clays also introduces a rate-effect. With different loading rates, there
is an apparent shift of the pre-consolidation pressure, as shown in this diagram in which the vertical
strain is plotted against the vertical effective stress. A 10-times faster loading rate gives a 30%
increase in apparent pre-consolidation pressure.
Slide 12:
That brings us to the concept of Isotachs from Bjerrum, 1967. Consider a diagram in which void ratio
is plotted against vertical effective stress. In this diagram we can think of imaginary parallel curves
linking vertical stress to void ratio for different durations of sustained loading: instant compression, 3
years, 30 years, 300 years, 3000 years, etcetera. If an initial stress state at point A is kept constant for
3000 years, the soil will compress in time and the void ratio will increase accordingly until it reaches
the line for 3000 years in point B. If the stress is then increased instantaneously, the path will come
back on the instant compression line in point C, from which the void ratio will follow this line if the
stress is continuously increased.
Slide 13:
In order to summarize the previous slides: An oedometer or CRS test can tell us something about:
Stiffness: In case the stress is continuously increased, the oedometer stiffness, Eoed, can be
measured as the tangent line to the stress-strain curve. Alternatively, we can measure the
compression index, C sub c, when plotting the void ratio as a function of stress on a logarithmic
scale, and the swelling index, C sub s, if the test includes unloading.
3
The stiffness is a function of the stress level: this is stress-dependency of stiffness. Moreover,
the stiffness is different for primary loading and for unloading or reloading: this is stress-path
dependency of stiffness.
An oedometer test, with measurement of the lateral stress, can tell us something about the
ratio of horizontal over vertical stress, K0.
Moreover, we can obtain the pre-consolidation stress from an oedometer test or a CRS test.
Note that for clay-type soils, the observed pre-consolidation stress from a CRS test may be
affected by the loading rate.
An oedometer test can also indicate structure or bonding in natural soils.
Finally, we can observe phenomena related to time-dependent deformations, such as
consolidation and secondary compression or creep, if we plot the strain as function of time. In
this case, the stress that is initially applied remains constant.
Slide 14:
Considering the compression index and swelling index, there is a strong correlation with the plasticity
index. As you know, the plasticity index is the difference between the liquid limit and the plastic limit.
The diagram, from the book by Kulhawy & Mayne ( 1990) shows the compression index and the
swelling index (here indicated as the unload-reload index) for different values of plasticity index (in
percent). Based on this data, they came up with the correlation that the compression index is the
plasticity index divided by 74, and the swelling index is the plasticity index divided by 370. Note that
these correlations are really approximate; look at the large scatter in the diagram. Hence, they
should only be used as first approximations.
Slide 15:
Here is another correlation by Kulhawy & Mayne (1990) for normally-consolidated clays, relating the
creep index to the natural water content. Again, note that this correlation is very crude, especially
since it is based on data that is plotted on a double logarithmic scale, but it gives a first
approximation.
Slide 16:
It is interesting to plot the creep index against the compression index, as done by Whittle, 2010.
Different types of clays with different water content or plasticity index can have quite different creep
index and compression index values, but their ratio for different clays is in the same order.
Slide 17:
Considering time-dependent behaviour of clays, we have seen that creep follows consolidation after
undrained one-dimensional compression. In fact, undrained loading + consolidation is defined as
primary compression whereas creep is defined as secondary compression. There are different
hypotheses about primary and secondary compression:
In Hypothesis A, the End-Of-Primary (EOP) consolidation strain and the EOP pre-consolidation
stress are unique for a particular clay and independent from the consolidation time.
In Hypothesis B, the EOP strain increases and the EOP pre-consolidation stress decreases with
increasing duration of the consolidation time.
When comparing time-dependent behaviour in the field with lab test conditions, it is clear that the
consolidation process in the field takes much more time because of the larger layer thickness.
In Hypothesis A, the amount of strain obtained after full consolidation is supposed to be the same in
the field case as for the lab scale. What follows after the EOP is the creep.
4
In Hypothesis B, part of the creep is considered to occur already during the consolidation process and
since this period is longer in the field case, more total strains are generated at the EOP.
Considering effective stress principles, during the consolidation process there is a gradual transition
from excess pore pressure into effective stress, which triggers the creep process already before the
EOP. The longer the consolidation process takes, the more creep strains are accumulated, so this
would lead to larger total strain at the EOP. Moreover, a longer consolidation process would also
lead to a lower effective stress loading rate. As we have seen in Slide 11, the observed pre-
consolidation stress is affected by the loading rate, so it is not a constant. In fact, if we adopt
effective stress principles and Bjerrums concept of Isotachs, we would end up with Hypothesis B.
This seems to be supported by experimental evidence.
Slide 18:
We conclude the lecture Soil behaviour in compression with some references.
