Soil Mechanics 2020
Soil Mechanics 2020
Soil Mechanics 2020
These notes are intended to accompany lectures in the course on Soil Mechanics.
1. These notes provide the essential information for the study of Soil Mechanics.
The lectures and associated powerpoint slides are based on these notes.
2. There are some examples in the notes. Additional examples will be discussed
in the lecture and tutorial sessions.
3. The final examination will be based entirely on the material contained in the
course notes. There is no need to purchase any text book. There are many
books that cover the course material, provide additional examples and may
assist your understanding. Some of these are included in the references in the
course outline.
CIVL2410 SOIL MECHANICS
INTRODUCTION
As the name Soil Mechanics implies, the subject is concerned with the deformation and
strength of bodies of soil. It deals with the mechanical properties of the soil materials and
with the application of the knowledge of these properties to engineering problems. In
particular, it is concerned with the interaction of structures with their foundation material.
This includes both conventional structures and also structures such as earth dams,
embankments and roads which are themselves made of soil.
As for other branches of engineering, the major issues are stability and serviceability.
When a structure is built it will apply a load to the underlying soil; if the load is too great
the strength of the soil will be exceeded and failure may ensue. It is important to realise
that not only buildings are of concern, the failure of an earth dam can have catastrophic
consequences, as can failures of natural and man-made slopes and excavations. Buildings
or earth structures may be rendered unserviceable by excessive deformation of the
ground, although it is usually differential settlement—where one side of a building settles
more than the other—that is most damaging. Criteria for allowable settlement vary from
case to case; for example, the settlement allowed in a factory that contains sensitive
equipment is likely to be far more stringent than that for a warehouse. Another important
aspect to be considered during design is the effect of any construction on adjacent
structures, for example the excavation of a basement and then the construction of a large
building will cause deformations in the surrounding ground and may have a detrimental
effect on adjacent buildings or other structures such as railway tunnels.
Many of the problems arising in Geotechnical Engineering stem from the interaction of
soil and water. For example, when a basement is excavated water will tend to flow into
the excavation. The question of how much water flows in needs to be answered so that
suitable pumps can be obtained to keep the excavation dry. The flow of water can have
detrimental effects on the stability of the excavation and is often the initiator of landslides
in natural and man-made slopes. Some of the effects associated with the interaction of
soil and water are quite subtle, for example, if an earthquake occurs, then a loose soil
deposit will tend to compress causing the water pressures to rise. If the water pressures
should increase so that they become greater than the stress due to the weight of the
overlying soil, then a quicksand condition will develop and buildings founded on this soil
may fail.
Soil mechanics differs from other branches of engineering in that generally there is little
control over the material properties; we have to make do with the soil at the site and this
is often highly variable. By taking samples at a few scattered locations we have to
determine the soil properties and their variability. At this stage in a project knowledge of
the site geology and geological processes is essential to successful geotechnical
engineering.
1
Soil mechanics is a relatively new branch of engineering science, the first major
conference occurred in 1936 and the mechanical properties of soils are still incompletely
understood. The first complete mechanical model for soil was published as recently as
1968. Over the last 40 years there has been rapid development in our understanding of
soil behaviour and the application of this knowledge in engineering practice. The subject
has now reached a phase of development similar to that of structural mechanics a century
ago and the words of William Anderson in 1893 about structural engineers are relevant
today for geotechnical engineering, "There is a tendency among the young and
inexperienced to put blind faith in formulas (computer programs), forgetting that most of
them are based upon premises which are not accurately reproduced in practice and which,
in any case, are frequently unable to take into account collateral disturbances which only
observation and experience can foresee and common sense provide against."
2
1 SOILS AND THEIR CLASSIFICATION
1.1 Introduction
Soil mechanics is concerned with particulate materials (soils) found in the ground that
are not cemented and not greatly compressed. These soils usually have a sedimentary
origin, however, they can also occur as the result of rock weathering without any
transport of the particles. The soil particles can have varying sizes, shapes and
mineralogies, although these properties are usually interrelated. For instance, the larger
sized particles are generally composed of quartz and feldspars, minerals that have high
strengths, and the particles are fairly round. The smaller sized particles are generally
composed of the clay minerals kaolin, illite and montmorillonite, minerals that have
low strengths, and form plate like particles. One of the most important aspects of
particulate materials is that there are gaps or voids between the particles. The amount
of voids is also influenced by the size, shape and mineralogy of the particles.
Because of the wide range of particle sizes, shapes and mineralogies in a typical soil, a
detailed classification of each soil would be very expensive and inappropriate for most
geotechnical engineering purposes. However, some form of simple classification
system giving information about the engineering properties is required on all sites. Why
is this necessary?
• Usually the soil on site has to be used. Soils differ from other engineering
materials in that one has very little, if any, control over their properties.
• The extent and properties of the soil at the site have to be determined.
• Cheap and simple tests are required to give an indication of the engineering
properties, such as stiffness and strength for preliminary design.
To achieve this continuous samples are recovered from boreholes, drilled to a depth
that will depend on the scale of the project. Observation of the core enables the different
soil layers to be determined and then classification tests are performed for these
different strata. The extent of the different soil layers can be determined by correlating
the results from different boreholes and this information is used to build a picture of the
sub-surface profile.
3
1.2 Particle Size Definitions
The precise boundaries between different soil types are somewhat arbitrary, but the
following scale is now in use worldwide.
where C, M, F stand for coarse, medium and fine respectively and the particle sizes are
in millimetres.
Note:
• the logarithmic scale. Most soils contain mixtures of sand, silt and clay particles,
so the range of particle sizes can be very large.
• not all particles less than 2 µm are comprised of clay minerals and some clay
mineral particles can be greater than 2 µm (and a micron, µm, is 10-6 m).
These include sands, gravels and larger particles. For these soils the grains are well
defined and may be seen by the naked eye. The individual particles may vary from
perfectly round to highly angular reflecting their geological origins.
These include the silts and clays and have particles smaller than 60 µm.
• Silts
These can be visually differentiated from clays because they exhibit the property
of dilatancy. If a moist sample is shaken in the hand water will appear on the
surface. If the sample is then squeezed in the fingers the water will disappear.
Their gritty feel can also identify silts.
• Clays
Clays exhibit plasticity, they may be readily remoulded when moist, and if left
to dry can attain high strengths
• Organic
These may be of either clay or silt sized particles. They contain significant
amounts of vegetable matter. The soils as a result are usually dark grey or black
and have a noticeable odour from decaying matter. Generally, only a surface
4
phenomenon but layers of peat may be found at depth. These are very poor soils
for most engineering purposes.
Different procedures are required for fine and coarse-grained material. Detailed
procedures are described in the Australian Standard AS 1289.A1 Methods of testing
soil for engineering purposes. These will be demonstrated in a laboratory session.
• Coarse
Sieve analysis is used to determine the distribution of the larger grain sizes. The
soil is passed through a series of sieves with the mesh size reducing
progressively and the proportions by weight of the soil retained on each sieve
are measured. There are a range of sieve sizes that can be used and the finest is
usually a 75 µm sieve. Sieving can be performed either wet or dry. Because of
the tendency for fine particles to clump together, wet sieving is often required
with fine-grained soils.
• Fine
To determine the grain size distribution of material passing the 75 µm sieve the
hydrometer method is commonly used. The soil is mixed with water and a
dispersing agent, stirred vigorously and then allowed to settle to the bottom of
a measuring cylinder. As the soil particles settle out of suspension, the specific
gravity of the mixture reduces. A hydrometer is used to record the variation of
specific gravity with time. By making use of Stoke’s Law, which relates the
velocity of a free-falling sphere to its diameter, the test data is reduced to
provide particle diameters and the % by weight of the sample finer than a
particular particle size.
5
1.5 Grading curves
The results from the particle size determination tests are plotted as grading curves.
These show the particle size plotted against the percentage of the sample by weight that
is finer than that size. The results are presented on a semi-logarithmic plot as shown in
Figure 2 below. The shape and position of the grading curve are used to identify some
characteristics of the soil.
100
80
60
% Finer
40
W
20
C P U
0
0.0001 0.001 0.01 0.1 1 10 100
Particle Size (mm)
Some typical grading curves are shown on the figure. The following descriptions are
applied to these curves
The use of names to describe typical grading curve shapes and positions has developed
as the suitability of different gradings for different purposes has become apparent. For
example, well graded sands and gravels can be easily compacted to relatively high
densities which result in higher strengths and stiffnesses. For this reason, soils of this
type are preferred for road bases. The suitability of different gradings is discussed in
some detail by Terzaghi and Peck (1967).
From the typical grading curves, it can be seen that soils are rarely all sand or all clay
and, in general, will contain particles with a wide range of sizes. Many organisations
have produced charts to classify soils giving names for the various combinations of
particle sizes. One such example is given in Figure 3 below.
6
0
100
10
90
20
80
30
70
40 Clay
Cla
(% 60
y
50
es
Siz
Siz
50
e s (%
60
d
Sa n
)
70
30
80 Clay-Sand Clay-Silt
20
90
Silty Sand Sandy Silt 10
100 Sand
0
0 10 20 30 40 50 60 70 80 90 100
Silt Sizes (%)
LOWER MISSISSIPPI VALLEY DIVISION,
U. S. ENGINEER DEPT.
Important observations from Figure 3 are that any soil containing more than 50% of
clay sized particles would be classified as a clay, whereas sand and silt require 80% of
the particles to be in that size range. Also, any soil having more than 20% clay would
have some clay like properties.
The hydrometer test is usually terminated when the percentage of clay sized particles
has been determined. However, there are significant differences between the behaviour
of the different clay minerals. To provide additional information on the soil behaviour
further classification tests are performed. One such set of tests, the Atterberg Limit
Tests, involve measuring the moisture contents of the soil at which changes in the soil
properties occur.
These tests are only used for the fine-grained, silt and clay, fraction of a soil (actually
the % passing a 425 µm sieve). If we take a very soft (high moisture content) clay
specimen and allow it to dry, we would obtain a relation similar to that shown in Figure
4.
As the soil dries, its strength and stiffness will increase. Three limits are indicated, the
definitions of which are given below. The liquid and plastic limits appear to be fairly
arbitrary, but recent research has suggested they are related to the strength of the soil.
7
Volume SL = Shrinkage Limit
PL = Plastic Limit
LL = Liquid Limit
SL PL LL
Moisture Content (%)
This is the moisture content the soil would have had if it were fully saturated at the
point at which no further shrinkage occurs on drying.
In the shrinkage test the soil is left to dry and the soil is therefore not saturated when
the shrinkage limit is reached. To estimate SL, it is necessary to measure the total
volume, 𝑉𝑉𝑇𝑇 , and the weight of the solids, 𝑤𝑤𝑠𝑠 . Then:
𝛾𝛾𝑤𝑤 𝑉𝑉𝑇𝑇 1
SL = 𝑚𝑚𝑐𝑐 = − (2)
𝑤𝑤𝑠𝑠 𝐺𝐺𝑠𝑠
This is the minimum water content at which the soil will deform plastically
This is the minimum water content at which the soil will flow under a small disturbing
force.
8
1.6.4 The Plasticity Index (PI or Ip)
I𝑝𝑝 = LL − PL (3)
𝑚𝑚𝑐𝑐 − PL 𝑚𝑚𝑐𝑐 − PL
LI = = (4)
LL − PL I𝑝𝑝
The Atterberg Limits and relationships derived from them are simple measures of the
water absorbing ability of soils containing clay minerals. For example, if a clay has a
very high LI and LL, it is capable of absorbing large amounts of water and, for instance,
would be unsuitable for the base of a pavement. The LL and PL are also related to the
soil strength.
Remember that only the fraction finer than 425 µm is tested in the Atterberg Tests. If
this fraction is only small (that is, the soil contains significant amounts of sand or
gravel) it might be expected that the soil would have better properties. While this is true
to some extent, it is important to realise that the soil behaviour is controlled by the finest
10–25% of the particles.
Several systems are used for classifying soil. This is because these systems have two
main purposes:
1. To determine the suitability of different soils for various purposes (p. 8, Data
Sheets).
The reason for the large number of such systems is the use of particular systems for
certain types of construction and the development of localised systems.
9
5. High compressibility silts
6. Low to medium compressibility clays
7. High compressibility clays
8. Peat, organic soils
The standard system used worldwide for most major construction projects is known as
the Unified Soil Classification System (USCS). This is based on an original system
devised by Casagrande. Soils are identified by symbols determined from sieve analysis
and Atterberg Limit tests.
Coarse-Grained Materials
If more than half of the material is coarser than the 75 µm sieve, the soil is classified as
coarse. The following steps are then followed to determine the appropriate 2 letter
symbol:
• If more than half of the coarse fraction is sand, then use prefix S
• If more than half of the coarse fraction is gravel, then use prefix G
First determine the percentage of fines, that is, the % of material passing the 75 µm
sieve.
If W or P are required for the suffix, then 𝐶𝐶𝑢𝑢 and 𝐶𝐶𝑐𝑐 must be evaluated:
𝐷𝐷60
𝐶𝐶𝑢𝑢 =
𝐷𝐷10
2
𝐷𝐷30
𝐶𝐶𝑐𝑐 =
(𝐷𝐷60 × 𝐷𝐷10 )
10
If M or C are required they have to be determined from the procedure used for fine
grained materials discussed below. Note that M stands for Silt and C for Clay. This is
determined from whether the soil lies above or below the A-line in the plasticity chart
shown in Figure 5.
For a coarse-grained soil which is predominantly sand, the following symbols are
possible:
Fine-Grained Materials
These are classified solely according to the results from the Atterberg Limit Tests.
Values of the Plasticity Index and Liquid Limit are used to determine a point in the
plasticity chart shown in Figure 5. The classification symbol is determined from the
region of the chart in which the point lies.
11
The final stage of the classification is to give a description of the soil to go with the 2-
symbol class. For a coarse-grained soil, this should include:
• stratification
• degree of compactness
• cementation
• moisture conditions
• drainage characteristics
The information required, along with all the details of the Unified Classification
Procedure is given in Figure 6. Note that slightly different information is required for
fine-grained soils.
12
Figure 6: Unified Soil Classification Chart
13
Example 1: Classification using USCS
Classification tests have been performed on a soil sample and the following grading
curve and Atterberg limits obtained. Determine the USCS classification.
100
80
60
% Finer
40
20
0
0.0001 0.001 0.01 0.1 1 10 100
Particle Size (mm)
% fines (% finer than 75 µm) = 11% ∴ coarse grained, dual symbols required
2. Determine % of different particle size fractions (to determine G or S) and 𝐷𝐷10 , 𝐷𝐷30 ,
𝐷𝐷60 from grading curve (to determine W or P)
3. From the Atterberg Test results determine its Plasticity Chart location
From Plasticity Chart, the point lies below the A-line; hence second suffix = M
Soil is a three-phase material that consists of solid particles which make up the soil
skeleton and voids which may be full of water if the soil is saturated, may be full of air
if the soil is dry, or may be partially saturated as shown in Figure 1.
solid
water
air
2.1 Units
Length metres
Mass tonnes (1 tonne = 103 kg)
Density (mass/unit volume) t/m3
Weight kilonewtons (kN)
Stress kilopascals (kPa) 1 kPa = 1 kN/m2
Unit Weight kN/m3
𝜌𝜌𝑤𝑤 = 1 tonne/m3
= 1 g/cm3
In most applications, it is not the mass that is important, but the force due to the mass
and the weight, 𝑤𝑤, is related to the mass, 𝑚𝑚, by the relation
𝑤𝑤 = 𝑚𝑚𝑚𝑚
15
where 𝑔𝑔 is the acceleration due to gravity. If 𝑚𝑚 is measured in tonnes and 𝑤𝑤 in kN, 𝑔𝑔 =
9.8 m/s2.
Because the force is usually required, it is often convenient in calculations to use the
unit weight, 𝛾𝛾 (weight per unit volume).
𝑤𝑤
𝛾𝛾 =
𝑉𝑉
𝑚𝑚𝑚𝑚
𝛾𝛾 =
𝑉𝑉
= 𝜌𝜌𝜌𝜌
Another frequently used quantity is the Specific Gravity, 𝐺𝐺𝑠𝑠 , which is defined by
density of material 𝜌𝜌
𝐺𝐺𝑠𝑠 = =
density of water 𝜌𝜌𝑤𝑤
It is often found that the specific gravity of the materials making up the soil particles
are close to the value for quartz, that is:
𝐺𝐺𝑠𝑠 ≈ 2.65
For all the common soil forming minerals, 2.5 < 𝐺𝐺𝑠𝑠 < 2.8.
We can use 𝐺𝐺𝑠𝑠 to calculate the density or unit weight of the solid particles:
and hence the volume of the solid particles if the mass or weight is known:
𝑚𝑚𝑠𝑠 𝑤𝑤𝑠𝑠
𝑉𝑉𝑠𝑠 = =
𝐺𝐺𝑠𝑠 𝜌𝜌𝑤𝑤 𝐺𝐺𝑠𝑠 𝛾𝛾𝑤𝑤
16
2.3 Void Ratio and Porosity
Using volumes is not very convenient in most calculations. An alternative measure that
is used is the void ratio, 𝑒𝑒. This is defined as the ratio of the volume of voids, 𝑉𝑉𝑣𝑣 , to the
volume of solids, 𝑉𝑉𝑠𝑠 , that is:
𝑉𝑉𝑣𝑣
𝑒𝑒 =
𝑉𝑉𝑠𝑠
A related quantity is the porosity, 𝑛𝑛, which is defined as ratio of the volume of voids to
the total volume.
𝑉𝑉𝑣𝑣
𝑛𝑛 =
𝑉𝑉𝑇𝑇
Now,
𝑉𝑉𝑣𝑣 𝑉𝑉𝑣𝑣 𝑛𝑛
𝑒𝑒 = = =
𝑉𝑉𝑠𝑠 (1 − 𝑛𝑛)𝑉𝑉𝑇𝑇 1 − 𝑛𝑛
and hence:
𝑒𝑒
𝑛𝑛 =
1 + 𝑒𝑒
The degree of saturation, 𝑆𝑆, has an important influence on the soil behaviour. It is
defined as the ratio of the volume of water to the volume of voids
𝑉𝑉𝑤𝑤 𝑉𝑉𝑤𝑤
𝑆𝑆 = =
𝑉𝑉𝑎𝑎 + 𝑉𝑉𝑤𝑤 𝑉𝑉𝑣𝑣
The distribution of the volume phases may be expressed in terms of 𝑒𝑒 and 𝑆𝑆; by knowing
the unit weight of water and the specific gravity of the particles, the distributions by
weight may also be determined as indicated in Table 2.
𝑉𝑉𝑤𝑤
𝑆𝑆 =
𝑒𝑒𝑒𝑒𝑠𝑠
17
Table 2: Distribution by volume, mass and weight in soil
Note that Table 2 assumes a solid volume 𝑉𝑉𝑠𝑠 = 1 m3. All terms in the table should be
multiplied by 𝑉𝑉𝑠𝑠 if this is not the case.
Several unit weights are used in Soil Mechanics. These are the bulk, saturated, dry and
submerged unit weights.
The bulk unit weight is simply defined as the weight per unit volume:
𝑤𝑤𝑇𝑇
𝛾𝛾𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 =
𝑉𝑉𝑇𝑇
When all the voids are filled with water, the bulk unit weight is identical to the saturated
unit weight, 𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠 , and when all the voids are filled with air, the bulk unit weight is
identical with the dry unit weight, 𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 . From Table 2 it follows that:
When S = 1,
𝛾𝛾𝑤𝑤 (𝐺𝐺𝑠𝑠 + 𝑒𝑒)
𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠 =
1 + 𝑒𝑒
When S = 0,
𝛾𝛾𝑤𝑤 𝐺𝐺𝑠𝑠
𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 =
1 + 𝑒𝑒
Note that in discussing soils that are saturated, it is common to discuss their dry unit
weight. This is done because the dry unit weight is simply related to the void ratio; it is
a way of describing the amount of voids.
The submerged unit weight, 𝛾𝛾 ′ , is sometimes useful when the soil is saturated and is
given by:
𝛾𝛾 ′ = 𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠 − 𝛾𝛾𝑤𝑤
18
2.6 Moisture Content
The moisture content, 𝑚𝑚𝑐𝑐 , is a very useful quantity because it is simple to measure. It
is defined as the ratio of the weight of water to the weight of solid material
𝑤𝑤𝑤𝑤
𝑚𝑚𝑐𝑐 =
𝑤𝑤𝑠𝑠
If we express the weights in terms of 𝑒𝑒, 𝑆𝑆, 𝐺𝐺𝑠𝑠 and 𝛾𝛾𝑤𝑤 as before, we obtain
and hence,
𝑒𝑒𝑒𝑒
𝑚𝑚𝑐𝑐 =
𝐺𝐺𝑠𝑠
Note that if the soil is saturated (𝑆𝑆 = 1), the void ratio can be simply determined from
the moisture content.
A sample of soil is taken using a thin walled sampling tube into a soil deposit. After the
soil is extruded from the sampling tube, a sample of diameter 50 mm and length 80 mm
is cut and is found to have a mass of 290 g. Soil trimmings created during the cutting
process are weighed and found to have a mass of 55 g. These trimmings are then oven
dried and found to have a mass of 45 g. Determine the phase distributions, void ratio,
degree of saturation and relevant unit weights.
2. Distribution by Volume
19
Air volume, 𝑉𝑉𝑎𝑎 = VT − 𝑉𝑉𝑠𝑠 − 𝑉𝑉𝑤𝑤 = 14.9 × 10−6 m3
3. Moisture Content
4. Void Ratio
5. Degree of Saturation
6. Unit Weights
If the sample were saturated, there would need to be an additional 14.9 × 10-6 m3 of
water. This would weigh 146.2 × 10-6 kN and thus the saturated unit weight of the soil
would be:
20
Example 2: Calculation of Unit Weights
A soil has a void ratio of 0.7. Calculate the dry and saturated unit weight of the material.
Assume that the solid material occupies 1 m3, then assuming 𝐺𝐺𝑠𝑠 = 2.65 the distribution
by volume and weight is as follows.
26.0
Dry unit weight, 𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 = = 15.3 kN/m3
1.7
(26.0 + 6.87)
Saturated unit weight, 𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠 = = 19.3 kN/m3
1.7
If the soil were fully saturated the moisture content would be:
6.87
𝑚𝑚𝑐𝑐 = = 0.264 = 26.4%
26.0
Alternatively, the unit weights may be calculated from the expressions given earlier
(which are on p. 5 of the Data Sheets):
(𝐺𝐺𝑠𝑠 + 𝑒𝑒)𝛾𝛾𝑤𝑤
𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠 =
1 + 𝑒𝑒
𝐺𝐺𝑠𝑠 𝛾𝛾𝑤𝑤
𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 =
1 + 𝑒𝑒
See also pp. 4–5 of the Data Sheets for most of the definitions and equations given in
this section.
21
3 COMPACTION
• The principal reason for compacting soil is to reduce subsequent settlement under
working loads.
• Compaction reduces the voids ratio making it more difficult for water to flow
through soil. This is important if the soil is being used to retain water, such as
would be required for an earth dam.
• Compaction can prevent the build-up of large water pressures that cause soil to
liquefy during earthquakes.
There are several types of test which can be used to study the compactive properties of
soils. Because of the importance of compaction in most earth works standard
procedures have been developed. These generally involve compacting soil into a mould
at various moisture contents.
Soil is compacted into a mould in 3–5 equal layers, each layer receiving 25 blows of a
hammer of standard weight. The apparatus is shown in Figure 1 below. The energy
(compactive effort) supplied in this test is 595 kJ/m3. The important dimensions are:
Because of the benefits from compaction, contractors have built larger and heavier
machines to increase the amount of compaction of the soil. It was found that the
Standard Compaction test could not reproduce the densities measured in the field and
this led to the development of the Modified Compaction test.
22
3.3.2 Modified Compaction Test AS 1289-E2.1
The procedure and equipment is essentially the same as that used for the Standard test
except that 5 layers of soil must be used. To provide the increased compactive effort
(energy supplied = 2072 kJ/m3) a heavier hammer and a greater drop height for the
hammer are used. The key dimensions for the Modified test are:
handle
collar (mould
extension) metal guide to control
drop of hammer
cylindrical
base plate
soil mould hammer for
compacting soil
To assess the degree of compaction it is important to use the dry unit weight, 𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 ,
because we are interested in the weight of solid soil particles in a given volume, not the
amount of solid, air and water in a given volume (which is the bulk unit weight). From
the relationships derived previously, we have:
𝐺𝐺𝑠𝑠 𝛾𝛾𝑤𝑤
𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 =
1 + 𝑒𝑒
𝐺𝐺𝑠𝑠 𝛾𝛾𝑤𝑤
𝑒𝑒 = −1 (1)
𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑
Because 𝐺𝐺𝑠𝑠 and 𝛾𝛾𝑤𝑤 are constants, it can be seen that increasing dry density means
decreasing voids ratio and a more compact soil.
23
In the test, the dry density cannot be measured directly; what are measured are the bulk
density and the moisture content. From the definitions, we have:
Thus, we have:
(1 + 𝑚𝑚𝑐𝑐 )𝑤𝑤𝑠𝑠
𝛾𝛾𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 = = (1 + 𝑚𝑚𝑐𝑐 )𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 (2)
𝑉𝑉𝑇𝑇
This allows us to plot the variation of dry unit weight with moisture content, giving the
typical response shown in Figure 2 below. From this graph, we can determine the
optimum moisture content, 𝑚𝑚𝑐𝑐,𝑜𝑜𝑜𝑜𝑜𝑜 , for the maximum dry unit weight, �𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 �𝑚𝑚𝑚𝑚𝑚𝑚
Dry Unit Weight, 𝜸𝜸𝒅𝒅𝒅𝒅𝒅𝒅
�𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 �𝑚𝑚𝑚𝑚𝑚𝑚
𝑚𝑚𝑜𝑜𝑜𝑜𝑜𝑜
If the soil were to contain a constant percentage, 𝐴𝐴, of voids containing air where:
𝑉𝑉𝑎𝑎
𝐴𝐴 (%) = × 100
𝑉𝑉𝑇𝑇
𝐴𝐴 𝑉𝑉𝑤𝑤 + 𝑉𝑉𝑠𝑠
1− =
100 𝑉𝑉𝑇𝑇
24
Then, a theoretical relationship between 𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 and m for a given value of 𝐴𝐴 can be
derived as follows:
𝐴𝐴
𝛾𝛾𝑏𝑏𝑏𝑏𝑏𝑏𝑏𝑏 𝑤𝑤𝑤𝑤 + 𝑤𝑤𝑠𝑠 (𝑤𝑤𝑤𝑤 + 𝑤𝑤𝑠𝑠 ) �1 − �
𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 = = = 100
1 + 𝑚𝑚𝑐𝑐 𝑉𝑉𝑇𝑇 (1 + 𝑚𝑚𝑐𝑐 ) (𝑉𝑉𝑤𝑤 + 𝑉𝑉𝑠𝑠 )(1 + 𝑚𝑚𝑐𝑐 )
Now,
𝑤𝑤𝑠𝑠
𝑉𝑉𝑠𝑠 =
𝐺𝐺𝑠𝑠 𝛾𝛾𝑤𝑤
𝑚𝑚𝑤𝑤 𝑤𝑤𝑤𝑤 𝑚𝑚𝑐𝑐 𝑤𝑤𝑠𝑠
𝑉𝑉𝑤𝑤 = = =
𝜌𝜌𝑤𝑤 𝛾𝛾𝑤𝑤 𝛾𝛾𝑤𝑤
Hence,
𝐴𝐴 𝐺𝐺𝑠𝑠 𝛾𝛾𝑤𝑤
𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 = �1 − �� � (3)
100 𝐺𝐺𝑠𝑠 𝑚𝑚𝑐𝑐 + 1
If the percentage of air voids is zero, that is, the soil is totally saturated, then this
equation becomes:
𝐺𝐺𝑠𝑠 𝛾𝛾𝑤𝑤
𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 = � � (4)
𝐺𝐺𝑠𝑠 𝑚𝑚𝑐𝑐 + 1
From this equation, we see that there is a limiting dry unit weight for any moisture
content and this occurs when the voids are full of water. Increasing the water content
for a saturated soil results in a reduction in dry unit weight. The relation between the
moisture content and dry unit weight for saturated soil is shown on the graph in Figure
3. This line is known as the no-air-voids (NAV) line.
impossible
Dry Unit Weight, 𝜸𝜸𝒅𝒅𝒅𝒅𝒅𝒅
25
3.5 Effect of Water Content during Compaction
As water is added to a soil (at low moisture content), it becomes easier for the particles
to move past one another during the application of the compacting forces. As the soil
compacts, the voids are reduced and this causes the dry unit weight (or dry density) to
increase. Initially then, as the moisture content increases so does the dry unit weight.
However, the increase cannot occur indefinitely because the soil state approaches the
no-air-voids line which gives the maximum dry unit weight for a given moisture
content. Thus, as the state approaches the no-air-voids line, further moisture content
increases must result in a reduction in dry unit weight. As the state approaches the no-
air-voids line a maximum dry unit weight is reached and the moisture content at this
maximum is called the optimum moisture content.
Increased compactive effort enables greater dry unit weights to be achieved which,
because of the shape of the no-air-voids line, must occur at lower optimum moisture
contents. The effect of increasing compactive energy can be seen in Figure 4. It should
be noted that for moisture contents greater than the optimum, the use of heavier
compaction machinery will have only a small effect on increasing dry unit weights. For
this reason, it is important to have good control over moisture content during
compaction of soil layers in the field.
increasing compactive
Dry Unit Weight, 𝜸𝜸𝒅𝒅𝒅𝒅𝒅𝒅
energy
It can be seen from this figure that the compaction curve is not a unique soil
characteristic. It depends on the compaction energy. For this reason, it is important
when giving values of �𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 � and 𝑚𝑚𝑐𝑐,𝑜𝑜𝑜𝑜𝑜𝑜 to also specify the compaction procedure
𝑚𝑚𝑚𝑚𝑚𝑚
(for example, Standard or Modified).
The table below contains typical values for the different soil types obtained from the
Standard Compaction Test.
26
Table 1: Typical values for a standard compaction test
Typical Values
�𝜸𝜸𝒅𝒅𝒅𝒅𝒅𝒅 � (kN/m3) 𝒎𝒎𝒄𝒄,𝒐𝒐𝒐𝒐𝒐𝒐 (%)
𝒎𝒎𝒎𝒎𝒎𝒎
Well graded sand, SW 22 7
Sandy clay, SC 19 12
Poorly graded sand, SP 18 15
Low plasticity clay, CL 18 15
Non plastic silt, ML 17 17
High plasticity clay, CH 15 25
Note that these are typical values. Because of the variability of soils, it is not appropriate
to use typical values in design and tests are always required.
To control the soil properties of earth constructions (e.g. dams, roads) it is usual to
specify that the soil must be compacted to some pre-determined dry unit weight. This
specification is usually that a certain percentage of the maximum dry density, as found
from a laboratory test (Standard or Modified), must be achieved.
For example, we could specify that field densities must be greater than 98% of the
maximum dry unit weight as determined from the Standard Compaction Test. It is then
up to the contractor to select machinery, the thickness of each lift (layer of soil added)
and to control moisture contents in order to achieve the specified amount of compaction.
a) accept b) accept
Dry Unit Weight, 𝜸𝜸𝒅𝒅𝒅𝒅𝒅𝒅
reject reject
27
There is a wide range of compaction equipment. For pavements, some kind of wheeled
roller or vibrating plate is usually used. These only affect a small depth of soil and, to
achieve larger depths, vibrating piles and drop weights can be used. The applicability
of the equipment depends on the soil type as indicated in the table below
For soils without any fines (sometimes referred to as cohesionless), the standard
compaction test is difficult to perform. For these soil types it is normal to specify a
relative density, 𝐼𝐼𝑑𝑑 , that must be achieved. The relative density is defined by:
𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑒𝑒
𝐼𝐼𝑑𝑑 = (5)
𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚
Note that if: 𝑒𝑒 = 𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 , 𝐼𝐼𝑑𝑑 = 1 and the soil is in its densest state
𝑒𝑒 = 𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 , 𝐼𝐼𝑑𝑑 = 0 and the soil is in its loosest state
The expression for relative density can also be written in terms of the dry unit weights
associated with the various voids ratios. From the definitions, we have:
28
𝐺𝐺𝑠𝑠 𝛾𝛾𝑤𝑤
𝑒𝑒 = −1
𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑
and hence,
1 1
− 𝛾𝛾
�𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 �𝑚𝑚𝑚𝑚𝑚𝑚 𝑑𝑑𝑑𝑑𝑑𝑑 �𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 �𝑚𝑚𝑚𝑚𝑚𝑚 �𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 − �𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 �𝑚𝑚𝑚𝑚𝑚𝑚 �
𝐼𝐼𝑑𝑑 = = (6)
1 1 𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 ��𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 �𝑚𝑚𝑚𝑚𝑚𝑚 − �𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 �𝑚𝑚𝑚𝑚𝑚𝑚 �
−
�𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 �𝑚𝑚𝑚𝑚𝑚𝑚 �𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 �𝑚𝑚𝑚𝑚𝑚𝑚
The description of the soil will include a description of the relative density. Generally,
the terms loose, medium and dense are used where:
Note that you cannot determine the unit weight from knowing 𝐼𝐼𝑑𝑑 . This is because the
values of the maximum and minimum dry unit weights (void ratios) can vary
significantly. They depend on soil type (mineralogy), the particle grading and the
angularity.
29
4 EFFECTIVE STRESS
A saturated soil is a two-phase material consisting of a soil skeleton and voids which
are saturated with water. It is reasonable to expect that the behaviour of an element of
such a material will be influenced not only by the forces applied to its surface but also
by the water pressure of the fluid in the pores.
Soil Soil
Soil
Figure 1: (a) Original soil sample, (b) Case 1: soil loaded by an applied weight,
W and (c) Case 2: soil loaded by water weighing W
In examining the reasons for this observed behaviour, it is helpful to use the following
quantities:
vertical force
vertical stress, 𝜎𝜎𝑣𝑣 = (1)
cross– sectional area
and to define an additional quantity the vertical effective stress, by the relation:
30
Let us examine the changes the vertical stress, pore water pressure and vertical effective
stress for the two load cases considered above:
𝑊𝑊 𝑊𝑊
Case 1 0
𝐴𝐴 𝐴𝐴
𝑊𝑊 𝑊𝑊
Case 2 0
𝐴𝐴 𝐴𝐴
Another situation in which effective stresses are important is the case of two rough
blocks sliding over one another, with water pressure in between them as shown in Fig
2.
The effective normal thrust transmitted through the points of contact will be
𝑁𝑁 ′ = 𝑁𝑁 − 𝑈𝑈 (3a)
The frictional force will then be given by 𝑇𝑇 = 𝜇𝜇𝑁𝑁 ′ , where µ is the coefficient of
friction. For soils and rocks, the actual contact area is very small compared to the cross-
sectional area, so that 𝑈𝑈/𝐴𝐴 is approximately equal to 𝑢𝑢𝑤𝑤 , the pore water pressure. Hence
dividing through by the cross sectional area, 𝐴𝐴, this becomes:
𝜏𝜏 = 𝜇𝜇𝜎𝜎𝑣𝑣′ (3b)
where 𝜏𝜏 is the average shear stress and 𝜎𝜎𝑣𝑣′ is the vertical effective stress.
31
both deformation and strength of soils depend upon the effective stress. This was
originally suggested by Terzaghi in the 1920’s—and Equation 2 and similar relations
are referred to as the Principle of Effective Stress.
It is clear from the definition of effective stress that, in order to calculate its value, it is
necessary to know both the total stress and the pore water pressure. The values of these
quantities are not always easy to calculate but there are certain simple situations in
which the calculation is quite straightforward. The most important is when the ground
surface is flat as is often the case with sedimentary (soil) deposits.
surcharge, q
If we consider the equilibrium of a column of soil of cross sectional area, 𝐴𝐴, it is found
that:
32
4.2.2 Calculation of Pore Water Pressure
water table
𝐻𝐻
𝑃𝑃
Suppose the soil deposit shown in Figure 4 has a static water table, as indicated. The
water table is the water level in a borehole and, at the water table, 𝑢𝑢𝑤𝑤 = 0. The water
pressure at a point, 𝑃𝑃, is given by:
A uniform layer of sand 10 m deep overlays bedrock. The water table is located 2 m
below the surface of the sand which is found to have a void ratio 𝑒𝑒 = 0.7. Assuming
that the soil particles have a specific gravity 𝐺𝐺𝑠𝑠 = 2.7, calculate the effective stress at a
depth 5 m below the surface.
33
2. Calculation of dry and saturated unit weights
a) b) c)
𝑤𝑤𝑠𝑠 26.46
𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 = = = 15.56 kN/m3
𝑉𝑉𝑇𝑇 1.70
26.46 + 6.86
𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠 = = 19.60 kN/m3
1.70
Note that, in practice, if the void ratio is known, the unit weights are not normally
calculated from first principles considering the volume fractions of the different phases.
This is often the case for saturated soils because the void ratio can be simply determined
from:
𝑒𝑒 = 𝑚𝑚𝑐𝑐 𝐺𝐺𝑠𝑠
34
The unit weights are calculated directly from the formulae in the Data Sheets, that is:
In general, the state of stress in a soil cannot be described by a single quantity, the
vertical stress. To fully describe the state of stress, the nine stress components (6 of
which are independent), as illustrated in Figure 7 need to be determined. Note that, in
soil mechanics, a compression positive sign convention is used.
𝜎𝜎𝑧𝑧𝑧𝑧
𝜎𝜎𝑦𝑦𝑦𝑦 z
𝜎𝜎𝑧𝑧𝑧𝑧
𝜎𝜎𝑥𝑥𝑥𝑥
𝜎𝜎𝑦𝑦𝑦𝑦
𝜎𝜎𝑧𝑧𝑧𝑧
y
𝜎𝜎𝑥𝑥𝑥𝑥
𝜎𝜎𝑦𝑦𝑦𝑦
𝜎𝜎𝑥𝑥𝑥𝑥 x
35
Example 2: Effects of Groundwater Level Changes
Initially a 50 m thick deposit of a clayey soil has a groundwater level 1 m below the
surface. Due to groundwater extraction from an underlying aquifer, the regional
groundwater level is lowered by 2 m. By considering the changes in effective stress at
a depth, 𝑧𝑧, in the clay, investigate what will happen to the ground surface.
Due to decreasing demands for water the groundwater rises (possible reasons include
de-industrialisation and greenhouse effects) back to the initial level. What problems
may arise?
Assume:
The vertical total and effective stresses at depth z are given in the table below.
At all depths, the effective stress increases and, as a result, the soil compresses. The
cumulative effect throughout the clay layer can produce a significant settlement of the
soil surface.
When the groundwater rises, the effective stress will return to its initial value and the
soil will swell and the ground surface will heave (up). However, due to the inelastic
nature of soil, the ground surface will not in general return to its initial position. This
may result in:
• surface flooding
• uplift of buildings
36
5 STEADY STATE FLOW
5.1 Introduction
It is important to know the amount of water that will enter a pit during construction or
the amount of stored water that may be lost by percolation through or beneath a dam.
The behaviour of soil is governed by the effective stress, which is the difference
between total stress and pore water pressure. When water flows, the pore water
pressures in the ground change. A knowledge of how the pore water pressure changes
can be important in considering the stability of earth dams, retaining walls, etc.
Because the pores in soils are so small, the flow through most soils is laminar. This
laminar flow is governed by Darcy's Law, which will be discussed below.
𝑃𝑃
Note: 𝑧𝑧 is
𝑧𝑧(𝑃𝑃) measured UP
from the datum
datum
𝑢𝑢𝑤𝑤 (𝑃𝑃)
ℎ(𝑃𝑃) = + 𝑧𝑧(𝑃𝑃) (1)
𝛾𝛾𝑤𝑤
In this equation, 𝛾𝛾𝑤𝑤 is the unit weight of water (9.8 kN/m3) and 𝑢𝑢𝑤𝑤 (𝑃𝑃) is the pore water
pressure.
37
Note:
𝑢𝑢𝑤𝑤 (𝑃𝑃)
The quantity is usually called the pressure head.
𝛾𝛾𝑤𝑤
The quantity 𝑧𝑧(𝑃𝑃) is called the elevation head (its value depends upon the choice of a
datum).
The velocity head (not shown in Equation 1) is generally neglected. The only
circumstances where it may be significant is in flow through rock-fill but, in such a
situation, the flow will generally be turbulent and so Darcy's Law is not valid.
2m
static water table
𝑋𝑋
1m
5 𝑚𝑚
1m
𝑃𝑃
impermeable stratum
𝑧𝑧(𝑃𝑃) = 1 m
4𝛾𝛾𝑤𝑤
ℎ(𝑃𝑃) = +1= 5m
𝛾𝛾𝑤𝑤
𝑧𝑧(𝑋𝑋) = 4 m
𝛾𝛾𝑤𝑤
ℎ(𝑋𝑋) = +4 =5m
𝛾𝛾𝑤𝑤
It appears that, when there is a static water table, the head is constant throughout the
saturated zone.
38
3. Calculation of Head at P (datum at the water table)
𝑧𝑧(𝑃𝑃) = −4 m
4𝛾𝛾𝑤𝑤
ℎ(𝑃𝑃) = −4= 0m
𝛾𝛾𝑤𝑤
𝑧𝑧(𝑋𝑋) = −1 m
𝛾𝛾𝑤𝑤
ℎ(𝑋𝑋) = −1 =0m
𝛾𝛾𝑤𝑤
When there is a static water table, the head is constant throughout the saturated zone,
but its numerical value depends on the choice of datum.
It is very important to carefully define the datum. The use of imaginary standpipes can
be helpful in visualising head. The head is then given by the height of the water in the
standpipe above the datum.
Note also that it is differences in head (not pressure) that cause flow.
∆ℎ
Soil Sample
∆𝑙𝑙
During his fundamental studies of the flow of water in soil, Darcy found that the flow
𝑄𝑄 was:
39
• Proportional to the cross-sectional area 𝐴𝐴.
∆ℎ
𝑄𝑄 = 𝑘𝑘𝑘𝑘 (2a)
∆𝑙𝑙
𝑄𝑄 = 𝑘𝑘𝑘𝑘𝑖𝑖
(2b)
𝑣𝑣 = 𝑘𝑘𝑘𝑘
Note that the actual average velocity of the water in the pores (the groundwater velocity)
is 𝑣𝑣⁄𝑛𝑛 where 𝑛𝑛 is the porosity. The groundwater velocity is always greater than the
Darcy velocity.
inlet
constant head
device load
H
manometer
outlet
sample
L
device for flow
measurement
porous disk
40
This is similar to Darcy's experiment. The sample of soil is placed in a graduated
cylinder of cross sectional area, 𝐴𝐴, and water is allowed to flow through. The discharge,
𝑋𝑋, during a suitable time interval, 𝑇𝑇, is collected. The difference in head, 𝐻𝐻, over a
length, 𝐿𝐿, is measured by means of manometers.
𝑋𝑋 𝐻𝐻
= 𝑘𝑘𝑘𝑘
𝑇𝑇 𝐿𝐿
𝑋𝑋𝑋𝑋 (3)
𝑘𝑘 =
𝐴𝐴𝐴𝐴𝐴𝐴
A piston is used to compact the soil because the permeability depends upon the void
ratio.
standpipe with
cross-sectional
area of 𝑎𝑎
porous
disc 𝐻𝐻1
𝐻𝐻
𝐻𝐻2
𝐿𝐿 sample of
area, 𝐴𝐴
𝛿𝛿𝛿𝛿
The flow in the standpipe is −𝑎𝑎 𝛿𝛿𝛿𝛿
41
𝐻𝐻
The flow in the sample is 𝑘𝑘𝑘𝑘 𝐿𝐿
and thus
𝛿𝛿𝛿𝛿 𝐻𝐻
−𝑎𝑎 = 𝑘𝑘𝑘𝑘 (4a)
𝛿𝛿𝛿𝛿 𝐿𝐿
𝑘𝑘𝑘𝑘
−𝑎𝑎 ln 𝐻𝐻 = 𝑡𝑡 + 𝐶𝐶 (4b)
𝐿𝐿
Now initially at time 𝑡𝑡 = 𝑡𝑡1, the height of water in the permeameter is 𝐻𝐻 = 𝐻𝐻1 while,
at the end of the test, 𝑡𝑡 = 𝑡𝑡2 and 𝐻𝐻 = 𝐻𝐻2 and thus:
𝐻𝐻1
𝑎𝑎𝑎𝑎 ln �
𝐻𝐻2 � (4c)
𝑘𝑘 =
𝐴𝐴(𝑡𝑡2 − 𝑡𝑡1 )
Soils exhibit a very wide range of permeabilities and while particle size may vary by
about 3–4 orders of magnitude, permeability may vary by about 10 orders of magnitude.
10-1 10-2 10-3 10-4 10-5 10-6 10-7 10-8 10-9 10-10 10-11 10-12
Silts Homogeneous
Gravels Sands
Fissured & Weathered Clays
Clays
Permeability is often estimated from correlations with particle size. For example:
𝑘𝑘 = (𝑑𝑑10 )2
This expression was first proposed by Hazen in 1893. It is satisfactory for sandy soils
but is less reliable for well graded soils and soils with a large fines fraction.
42
5.5 Mathematical Form of Darcy's Law
Because of their geological history, soils tend to be deposited in layers and hence have
different flow properties along the layering and transverse to the layering.
z (vertical)
∆𝑧𝑧
B C
∆𝑥𝑥
O x (horizontal)
Suppose that the permeability in the horizontal plane is 𝑘𝑘ℎ , then the velocity in the 𝑥𝑥
direction, 𝑣𝑣𝑥𝑥 , is approximately given by:
ℎ(𝐶𝐶) − ℎ(𝐵𝐵)
𝑖𝑖𝑥𝑥 ≈ (5a)
∆𝑥𝑥
𝜕𝜕ℎ
𝑣𝑣𝑥𝑥 = −𝑘𝑘ℎ
𝜕𝜕𝜕𝜕
The negative signs in these equations have been introduced because flow occurs in the
direction of decreasing head.
Similarly, if the permeability in the vertical direction is kv then the velocity 𝑣𝑣𝑧𝑧 is given
by:
ℎ(𝐴𝐴) − ℎ(𝐵𝐵)
𝑖𝑖𝑧𝑧 ≈ (5b)
∆𝑧𝑧
𝜕𝜕ℎ
𝑣𝑣𝑧𝑧 = −𝑘𝑘𝑣𝑣
𝜕𝜕𝜕𝜕
Should there also be flow in the y direction, this is similarly governed by:
𝜕𝜕ℎ
𝑣𝑣𝑦𝑦 = −𝑘𝑘ℎ
𝜕𝜕𝜕𝜕
43
5.5.1 Plane Flow
In many situations, as in the dam shown below, there will be no flow in one direction
(usually taken as the y direction).
cross-section of a
long dam
flow
soil
impermeable bedrock
In order to be able to analyse the complex flows that occur in practice, it is necessary
to examine the water entering and leaving an element of soil. Consider plane flow into
the small rectangular box of soil shown below:
𝑣𝑣𝑧𝑧
𝑣𝑣𝑥𝑥 D soil
B ∆𝑧𝑧
element
∆𝑥𝑥
44
The net flow into the element is:
net flow = [𝑣𝑣𝑥𝑥 (𝐵𝐵) − 𝑣𝑣𝑥𝑥 (𝐷𝐷)]∆𝑦𝑦∆𝑧𝑧 + [𝑣𝑣𝑧𝑧 (𝐶𝐶) − 𝑣𝑣𝑧𝑧 (𝐴𝐴)]∆𝑥𝑥∆𝑦𝑦 (6a)
For steady state seepage the flow into the box will just equal the flow out, so the net
flow in will be zero, thus dividing by ∆𝑥𝑥∆𝑦𝑦∆𝑧𝑧 and taking the limit for an infinitesimal
element, it is found:
𝜕𝜕𝑣𝑣𝑥𝑥 𝜕𝜕𝑣𝑣𝑧𝑧
+ =0 (6b)
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
𝜕𝜕 𝜕𝜕ℎ 𝜕𝜕 𝜕𝜕ℎ
�𝑘𝑘ℎ � + �𝑘𝑘𝑣𝑣 � = 0 (7a)
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
For a homogeneous material in which the permeability does not vary with position, this
becomes:
𝜕𝜕 2 ℎ 𝜕𝜕 2 ℎ
𝑘𝑘ℎ 2 + 𝑘𝑘𝑣𝑣 2 = 0 (7b)
𝜕𝜕𝑥𝑥 𝜕𝜕𝑧𝑧
and for an isotropic material in which the permeability is the same in all directions
(𝑘𝑘ℎ = 𝑘𝑘𝑣𝑣 ):
𝜕𝜕 2 ℎ 𝜕𝜕 2 ℎ
+ =0 (7c)
𝜕𝜕𝑥𝑥 2 𝜕𝜕𝑧𝑧 2
For the more general situation, in which there is flow in three dimensions, the continuity
equation becomes:
For a homogeneous material in which the permeability does not vary with position (x,
y, z), this becomes:
𝜕𝜕 2 ℎ 𝜕𝜕 2 ℎ 𝜕𝜕 2 ℎ
𝑘𝑘ℎ 2 + 𝑘𝑘ℎ 2 + 𝑘𝑘𝑣𝑣 2 = 0 (9b)
𝜕𝜕𝑥𝑥 𝜕𝜕𝑦𝑦 𝜕𝜕𝑧𝑧
and for an isotropic material in which the permeability is the same in all directions:
𝜕𝜕 2 ℎ 𝜕𝜕 2 ℎ 𝜕𝜕 2 ℎ
+ + =0 (9c)
𝜕𝜕𝑥𝑥 2 𝜕𝜕𝑦𝑦 2 𝜕𝜕𝑧𝑧 2
45
6 FLOW NETS
6.1 Introduction
Let us consider a state of plane seepage, as for example in the dam shown in Figure 1.
phreatic line
unsaturated
soil
drainage
blanket
𝑧𝑧
flow
𝑥𝑥
For an isotropic material, the head satisfies Laplace's equations, thus analysis involves
the solution of:
𝜕𝜕 2 ℎ 𝜕𝜕 2 ℎ
+ =0
𝜕𝜕𝑥𝑥 2 𝜕𝜕𝑧𝑧 2
46
6.2 Representation of Solution
At every point (𝑥𝑥, 𝑧𝑧) where there is flow there will be a value of head, ℎ(𝑥𝑥, 𝑧𝑧). In order
to represent these values, we draw contours of equal head as shown on Figure 2.
equipotential (EP)
It thus follows:
𝜕𝜕ℎ 𝜕𝜕ℎ
𝑑𝑑𝑑𝑑 + 𝑑𝑑𝑑𝑑 = 0 (1b)
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
𝑑𝑑𝑑𝑑 𝜕𝜕ℎ⁄𝜕𝜕𝜕𝜕
� � =− (1c)
𝑑𝑑𝑑𝑑 EP 𝜕𝜕ℎ⁄𝜕𝜕𝜕𝜕
It is also useful, in visualising the flow in a soil, to plot the flow lines (FL), these are
lines that are tangential to the flow at a given point and are illustrated in Figure 2.
It can be seen from Figure 2 that the flow lines and equipotentials are orthogonal. To
show this, notice that, on a flow line, the tangent at any point is parallel to the flow at
that point so that:
47
It follows immediately that:
𝑑𝑑𝑑𝑑 𝑣𝑣𝑥𝑥
� � =
𝑑𝑑𝑑𝑑 FL 𝑣𝑣𝑧𝑧
Now, from Darcy’s Law,
𝜕𝜕ℎ
𝑣𝑣𝑥𝑥 = −𝑘𝑘
𝜕𝜕𝜕𝜕 (2b)
𝜕𝜕ℎ
𝑣𝑣𝑧𝑧 = −𝑘𝑘
𝜕𝜕𝜕𝜕
Thus,
𝑑𝑑𝑑𝑑 𝜕𝜕ℎ⁄𝜕𝜕𝜕𝜕
� � =
𝑑𝑑𝑑𝑑 FL 𝜕𝜕ℎ⁄𝜕𝜕𝜕𝜕
and so,
𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
� � ∙ � � = −1 (3)
𝑑𝑑𝑑𝑑 EP 𝑑𝑑𝑑𝑑 FL
and thus the flow lines and equipotentials are orthogonal in an isotropic material.
Consider a pair of flow lines; clearly the flow through this flow tube must be constant
and so, as the tube narrows, the velocity must increase. Suppose now, we have a pair of
flow lines as shown in Figure 3.
Y
ℎ + ∆ℎ
ℎ + 2∆ℎ
Z FL
T
t EP X ∆𝑄𝑄
y EP
FL
x
z
∆𝑄𝑄
48
Suppose that the flow per unit width (in the y direction) is, ∆𝑄𝑄, then the velocity, 𝑣𝑣, in
the tube is given by:
∆𝑄𝑄
𝑣𝑣 = (4a)
|𝑦𝑦𝑦𝑦|
Also, let us assume that the potential drop between any adjacent pair of equipotentials
is ∆ℎ, then it follows from Darcy’s Law:
∆ℎ
𝑣𝑣 = 𝑘𝑘 (4b)
|𝑧𝑧𝑧𝑧|
∆𝑄𝑄 |𝑦𝑦𝑦𝑦|
= (4c)
𝑘𝑘∆ℎ |𝑧𝑧𝑧𝑧|
Using an identical argument to that used in developing Equation 4c, it can be shown
that:
∆𝑄𝑄 |𝑌𝑌𝑌𝑌|
= (4d)
𝑘𝑘∆ℎ |𝑍𝑍𝑍𝑍|
|𝑦𝑦𝑦𝑦| |𝑌𝑌𝑌𝑌|
= (5)
|𝑧𝑧𝑧𝑧| |𝑍𝑍𝑍𝑍|
Thus, each of the elemental rectangles bounded by the given pair of flow lines and a
pair of equipotentials (having an equal head drop) have the same length to breadth ratio.
Next, consider a pair of equipotentials cut by flow tubes each carrying the same flow,
∆𝑄𝑄, as shown in Figure 4.
49
∆𝑄𝑄 B
∆𝑄𝑄
C EP ℎ
D
d FL A
b FL
EP ℎ + ∆ℎ
a
c
Then we see that, if it is assumed that each of the tubes is of unit width (in the y
direction) then the velocity in the tube is:
∆𝑄𝑄
𝑣𝑣 = (6a)
|𝑐𝑐𝑐𝑐|
∆ℎ
𝑣𝑣 = 𝑘𝑘 (6b)
|𝑎𝑎𝑎𝑎|
∆𝑄𝑄 |𝑐𝑐𝑐𝑐|
= (6c)
𝑘𝑘∆ℎ |𝑎𝑎𝑎𝑎|
∆𝑄𝑄 |𝐶𝐶𝐶𝐶|
= (6d)
𝑘𝑘∆ℎ |𝐴𝐴𝐴𝐴|
Hence, again, a pair of flow tubes carrying equal flows will intersect a given pair of
equipotentials in elemental rectangles which have the same length to breadth ratio.
In drawing flow nets by hand, it is usual to draw them so that each flow tube carries the
same flow and so that the head drop between adjacent equipotentials is equal. In such
cases, all elemental rectangles will be similar. It is usually most convenient to draw the
net so that these rectangles are 'square' (it is possible to draw an inscribed circle). This
is illustrated in Figure 5.
50
Figure 5: Inscribing circles in a flow net
To calculate quantities of interest, that is the flow and pore pressures, a flow net must
be drawn. The flow net must consist of two families of orthogonal lines that ideally
define a square mesh and that also satisfy the boundary conditions. The three most
common boundary conditions are discussed below.
𝐻𝐻 − 𝑧𝑧
𝐻𝐻
equipotential
𝑧𝑧
datum
51
Now,
So,
(𝐻𝐻 − 𝑧𝑧)𝛾𝛾𝑤𝑤
ℎ= + 𝑧𝑧 = 𝐻𝐻
𝛾𝛾𝑤𝑤 (7)
That is, the head is constant for any value of 𝑧𝑧, which is, by definition, an equipotential.
Alternatively, this could have been determined by considering imaginary standpipes
placed at the soil boundary; for every point, the water level in the standpipe would be
the same as the water level. The upstream face of the dam shown in Figures 1 and 2 is
an example of this situation.
permeable
soil
𝑣𝑣𝑛𝑛 = 0
𝑣𝑣𝑡𝑡
flow line
impermeable material
The phreatic surface shown in Figures 2 and 8 is also a flow line marking the boundary
of the flow net. A phreatic surface is also a line of constant (zero) pore pressure as
discussed below.
Sometimes a portion of saturated soil is in contact with air and so the pore pressure of
the water just beneath that surface is atmospheric. The phreatic surface shown in Figure
8 below is an example of such a condition. We can show from the expression for head
52
in terms of pore pressure that equipotentials intersecting a line of constant pore pressure
do so at equal vertical intervals as follows:
𝑢𝑢𝑤𝑤
ℎ= + 𝑧𝑧
𝛾𝛾𝑤𝑤
Thus,
∆𝑢𝑢𝑤𝑤
∆ℎ = + ∆𝑧𝑧
𝛾𝛾𝑤𝑤
Now,
∆𝑢𝑢𝑤𝑤 = 0
and so,
∆ℎ = ∆𝑧𝑧 (8)
2. Draw a coarse net which is consistent with the boundary conditions and which
has orthogonal equipotential and flow lines. (It is usually easier to visualise the
pattern of flow so start by drawing the flow lines.)
3. Modify the mesh so that it meets the conditions outlined above and so that the
rectangles between adjacent flow lines and equipotentials are square.
53
6.6 Calculation of Quantities of Interest from Flow Nets
In most problems, we know the head difference (𝐻𝐻) between inlet and outlet and thus:
𝐻𝐻
∆ℎ = (9)
number of potential drops
15 m
ℎ =15 m
5m ℎ =0
𝑃𝑃
ℎ =3m
ℎ =12 m ℎ =6m
ℎ =9m
For example, let us assume that the depth of water retained by the dam is 15 m and that
downstream of the dam, the water table is level with the ground surface. For this case
it can be seen that the total head drop is 15 m. Inspection of Figure 2 or Figure 9 shows
that there are 5 potential drops and hence the head drop between each pair of potentials
is ∆ℎ = 15/5 = 3 m.
The flow net has been drawn so that the elemental rectangles are approximately square;
thus, referring to Figure 3 and Equation 4, it can be seen that between each pair of flow
tubes, the flow is:
It should be noted that, in the development of this formula, it was assumed that each
flow tube was of unit width and so Equation 10a gives the flow per unit width (into the
page).
54
Suppose that the permeability of the underlying soil is 𝑘𝑘 = 10-5 m/sec (typical of a fine
sand or silt), then the flow between each pair of flow tubes is:
There are 5 flow tubes and so the total flow per unit width of dam is:
𝑄𝑄 = 5 × 10−5 × 3 m3 /s/m𝑤𝑤𝑤𝑤𝑤𝑤𝑤𝑤ℎ
and, if the dam is 25 m wide, the total flow under the dam:
𝑄𝑄 = 25 × 5 × 10−5 × 3 m3 /s
The flow per unit width can alternatively be calculated from the formula:
𝑁𝑁𝑓𝑓
𝑄𝑄 = 𝑘𝑘𝑘𝑘 (10b)
𝑁𝑁ℎ
Equation 10b is given in the Data Sheets to calculate the flow per unit width. In this
equation, 𝑁𝑁𝑓𝑓 is the number of flowtubes (the number of flowlines – 1) and 𝑁𝑁ℎ is the
number of equipotential drops (the number of equipotential lines – 1).
Note that there are occasions where Equation 10b cannot simply be applied, but
Equation 10a will always be applicable for individual flow tubes. It is often necessary
to determine ∆ℎ from consideration of a single flow tube. If a square flow net has been
constructed, that value of ∆ℎ will apply to all flow tubes.
The pore pressure at any point can be found using the expression:
𝑢𝑢𝑤𝑤
ℎ= + 𝑧𝑧
𝛾𝛾𝑤𝑤
Now referring to Figure 9, suppose that we wish to calculate the pore pressure at the
point 𝑃𝑃. Taking the datum to be at the base of the dam, it can be seen that 𝑧𝑧 = −5 m
and so:
Figure 10 below shows a long vessel, 20 metres wide, stranded on a sand bank. It is
proposed to pump water into a well point, 10 metres down, under the centre of the vessel
to assist in towing the vessel off. The water depth is 1 metre.
The sand has a permeability of 3 × 10-4 m/sec. Assuming that a head of 50 m can be
applied at the well point calculate:
55
• The pore pressure distribution across the base of the vessel
• The total upthrust due to this increase in pore pressure
• The rate at which water must be pumped into the well point
well point
56
5m 2.5 m 1.8 m
A B C D E
Figure 11: Enlarged view of flow net in the vicinity of the base of the vessel
For convenience, these are chosen to be where the EPs meet the vessel (B to E) and at
the vessel centreline (A). Hence calculate the pore water pressures.
A B C D E
Head (1) ℎ1 − 4.5∆ℎ ℎ1 − 5∆ℎ ℎ1 − 6∆ℎ ℎ1 − 7∆ℎ ℎ1 − 8∆ℎ
Head (2) ℎ2 + 4.5∆ℎ ℎ2 + 4∆ℎ ℎ2 + 3∆ℎ ℎ2 + 2∆ℎ ℎ2 + ∆ℎ
Head (1 and 2) (m) 20.5 18.33 14.0 9.67 5.3
Pressure (kPa) 201.1 179.8 137.3 94.9 52.3
𝒖𝒖𝒘𝒘 = 𝜸𝜸𝒘𝒘 (𝒉𝒉 − 𝒛𝒛)
3. Measure the lengths off the flow net (note that diagram must be drawn to scale) and
hence calculate force from pressure distribution
For simplicity, assume linear variation in pressure between points. Then the TOTAL
UPTHRUST (per unit length of the vessel) is:
57
7 FLOW NETS FOR ANISOTROPIC MATERIALS
7.1 Introduction
Many soils are formed in horizontal layers as a result of sedimentation through water.
Because of seasonal variations, such deposits tend to be horizontally layered and this
results in different permeabilities in the horizontal and vertical directions.
Consider the horizontally layered deposit, shown in Figure 1, which consists of pairs of
layers, the first of which has a permeability of 𝑘𝑘1 and a thickness of 𝑑𝑑1 overlaying a
second which has permeability 𝑘𝑘2 and thickness 𝑑𝑑2 .
𝑘𝑘 = 𝑘𝑘1 𝑑𝑑1
𝑘𝑘 = 𝑘𝑘2 𝑑𝑑2
First, consider horizontal flow in the system and suppose that a head difference of ∆h
exists between the left and right-hand sides as indicated in Figure 2. It then follows,
from Darcy’s Law, that:
∆ℎ ∆ℎ
𝑣𝑣1 = 𝑘𝑘1 ; 𝑄𝑄1 = 𝑘𝑘1 𝑑𝑑 (1a)
𝐿𝐿 𝐿𝐿 1
and,
∆ℎ ∆ℎ
𝑣𝑣2 = 𝑘𝑘2 ; 𝑄𝑄2 = 𝑘𝑘2 𝑑𝑑 (1b)
𝐿𝐿 𝐿𝐿 2
58
ℎ = ℎ0 ℎ = ℎ0 − ∆ℎ
𝑣𝑣 = 𝑣𝑣1 𝑑𝑑1
𝑣𝑣 = 𝑣𝑣2 𝑑𝑑2
𝐿𝐿
It therefore follows:
𝑄𝑄1 + 𝑄𝑄2 ∆ℎ
𝑣𝑣 = = 𝑘𝑘ℎ (2a)
𝑑𝑑1 + 𝑑𝑑2 𝐿𝐿
where:
Next, consider vertical flow through the system, shown in Figure 3. Suppose that the
superficial velocity in each of the layers is 𝑣𝑣 and that the head loss in layer 1 is ∆ℎ1 ,
while the head loss in layer 2 is ∆ℎ2
ℎ = ℎ0
𝑣𝑣 𝑑𝑑1
ℎ = ℎ0 − ∆ℎ1
𝑣𝑣 𝑑𝑑2
ℎ = ℎ0 − ∆ℎ1 − ∆ℎ2
𝐿𝐿
59
In layer 1,
∆ℎ1
𝑣𝑣 = 𝑘𝑘1
𝑑𝑑1
so,
𝑣𝑣𝑑𝑑1
∆ℎ1 = (3a)
𝑘𝑘1
Similarly, in layer 2:
∆ℎ2
𝑣𝑣 = 𝑘𝑘2
𝑑𝑑2
𝑣𝑣𝑑𝑑2 (3b)
∆ℎ2 =
𝑘𝑘2
The total head loss across the system will be ∆ℎ = ∆ℎ1 + ∆ℎ2 and the hydraulic
gradient will be given by:
𝑣𝑣𝑑𝑑1 𝑣𝑣𝑑𝑑2
∆ℎ ∆ℎ1 + ∆ℎ2 +
𝑘𝑘 𝑘𝑘2 (3c)
𝑖𝑖 = = = 1
𝑑𝑑 𝑑𝑑1 + 𝑑𝑑2 𝑑𝑑1 + 𝑑𝑑2
∆ℎ
𝑣𝑣 = 𝑘𝑘𝑣𝑣 (3d)
𝑑𝑑
and hence,
𝑑𝑑 𝑑𝑑1 𝑑𝑑2
= + (3e)
𝑘𝑘𝑣𝑣 𝑘𝑘1 𝑘𝑘2
Example 1
Suppose that that the layers are of equal thickness 𝑑𝑑1 = 𝑑𝑑2 = 𝑑𝑑0 and that 𝑘𝑘1 =
10−8 m/s and that 𝑘𝑘2 = 10−10 m/s, then:
and,
𝑑𝑑0 + 𝑑𝑑0
𝑘𝑘𝑣𝑣 = = 1.98 × 10−10 m/s
𝑑𝑑0 /10−8 + 𝑑𝑑0 /10−10
showing that, as is generally the case, the vertical permeability is much less than the
horizontal.
60
7.3 Flow Nets for Soil with Anisotropic Permeability
Plane flow in an anisotropic material having a horizontal permeability 𝑘𝑘ℎ and a vertical
permeability 𝑘𝑘𝑣𝑣 is governed by the equation:
𝜕𝜕 2 ℎ 𝜕𝜕 2 ℎ
𝑘𝑘ℎ + 𝑘𝑘𝑣𝑣 =0 (4)
𝜕𝜕𝑥𝑥 2 𝜕𝜕𝑧𝑧 2
The solution of this equation can be reduced to that of flow in an isotropic material by
the following simple device: introduce new variables defined as follows:
𝑥𝑥 = 𝛼𝛼𝑥𝑥̅
(5a)
𝑧𝑧 = 𝑧𝑧̅
𝑘𝑘ℎ 𝜕𝜕 2 ℎ 𝜕𝜕 2 ℎ
+ =0 (5b)
𝛼𝛼 2 𝑘𝑘𝑣𝑣 𝜕𝜕𝑥𝑥̅ 2 𝜕𝜕𝑧𝑧̅2
Thus, by choosing:
𝑘𝑘ℎ
𝛼𝛼 = � (5c)
𝑘𝑘𝑣𝑣
It is found that the equation governing flow in an anisotropic soil reduces to that for an
isotropic soil, viz.:
𝜕𝜕 2 ℎ 𝜕𝜕 2 ℎ
+ =0 (5d)
𝜕𝜕𝑥𝑥̅ 2 𝜕𝜕𝑧𝑧̅ 2
and, so, the flow in anisotropic soil can be analysed using the same methods (including
sketching flow nets) that are used for analysing isotropic soils.
61
Example 2: Seepage in an Anisotropic Soil
Suppose we wish to calculate the flow under the dam shown in Figure 4:
2.5 m
𝐻𝐻1
impermeable dam
𝐻𝐻2
𝑧𝑧 𝐿𝐿
𝑧𝑧
soil layer
𝑥𝑥
impermeable bedrock
Figure 4: Dam on a permeable soil layer over impermeable rock (natural scale)
For the soil shown in Figure 4, it is found that 𝑘𝑘ℎ = 4𝑘𝑘𝑣𝑣 and therefore:
4𝑘𝑘𝑣𝑣
𝛼𝛼 = � =2
𝑘𝑘𝑣𝑣
so,
𝑧𝑧 = 𝑧𝑧̅
62
5m
𝐻𝐻1
𝐻𝐻2
𝑧𝑧 𝑥𝑥 = 𝐿𝐿
𝑥𝑥̅ = 𝐿𝐿/2
𝑧𝑧
𝑥𝑥𝑥𝑥
impermeable bedrock
The flow net can now be drawn in the transformed co-ordinates and this is shown in
Figure 6.
5m
It is possible to use the flow net in the transformed space to calculate the flow
underneath the dam by introducing an equivalent permeability:
63
A rigorous proof of this result will not be given here, but it can be demonstrated to work
for purely horizontal flow as follows:
ℎ = ℎ0 ℎ = ℎ0 − ∆ℎ ℎ = ℎ0 ℎ = ℎ0 − ∆ℎ
𝑋𝑋 𝑋𝑋�
∆ℎ
∆𝑄𝑄 = 𝑘𝑘ℎ 𝑡𝑡 (7a)
𝑥𝑥
∆ℎ ∆ℎ 𝑘𝑘ℎ
∆𝑄𝑄 = 𝑘𝑘𝑒𝑒𝑒𝑒 𝑡𝑡 = 𝑘𝑘𝑒𝑒𝑒𝑒 𝑡𝑡 � (7b)
𝑥𝑥̅ 𝑥𝑥 𝑘𝑘𝑣𝑣
Example 3
Suppose that in Figure 6, 𝐻𝐻1 = 13 m and 𝐻𝐻1 = 2.5 m, and that 𝑘𝑘𝑣𝑣 = 10−6 m/s and
𝑘𝑘ℎ = 4 × 10−6 m/s. The equivalent permeability is:
The total head drop is 10.5 m and there are 14 head drops and thus:
(13 − 2.5)
∆ℎ = = 0.75 m
14
64
For a dam with a width of 50 m,
7.4 Piping
Many dams on soil foundations have failed because of the sudden formation of a piped
shaped discharge channel. As the store water rushes out, the channel widens and
catastrophic failure results. This results from erosion of fine particles due to water flow.
Another situation where flow can cause failure is in producing ‘quicksand’ conditions.
This is also often referred to as piping failure.
In order to analyse this situation, consider water flowing upwards through the element
shown in Figure 8.
Elevation:
𝑢𝑢2
𝑧𝑧 = 𝑧𝑧2 ; ℎ = ℎ2 ; 𝑢𝑢 = 𝑢𝑢2
𝑧𝑧 = 𝑧𝑧1 ; ℎ = ℎ1 ; 𝑢𝑢 = 𝑢𝑢1
𝑢𝑢1
Plan:
area = 𝐴𝐴
The pore pressure can be calculated from the head and so:
65
For piping to occur, the uplift must be greater than the self-weight of the soil:
𝛾𝛾𝑤𝑤 (ℎ1 − ℎ2 ) − 𝛾𝛾𝑤𝑤 (𝑧𝑧1 − 𝑧𝑧2 ) > 𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠 (𝑧𝑧2 − 𝑧𝑧1 ) (8c)
or, alternatively:
𝑖𝑖 > 𝑖𝑖𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
where:
(ℎ1 − ℎ2 )
𝑖𝑖 = hydraulic gradient =
(𝑧𝑧2 − 𝑧𝑧1 )
and,
(𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠 − 𝛾𝛾𝑤𝑤 )
𝑖𝑖𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 = critical hydraulic gradient =
𝛾𝛾𝑤𝑤
Example 4
Suppose the dam shown in Figure 6 is 39 metres wide (this may be determined from
the scale drawing), the water levels are the same as in the previous example (𝐻𝐻1 = 13 m
and 𝐻𝐻1 = 2.5 m) and the saturated unit weight of the soil is 18 kN/m3. Piping is most
likely to occur at the toe of the dam, the hydraulic gradient there can be obtained from
the flow net:
Thus,
0.75
𝑖𝑖 = = 0.67
1.125
Now,
18 − 9.81
𝑖𝑖𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 = = 0.83
9.81
The safety factor against piping failure is thus 𝑖𝑖𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 /𝑖𝑖 = 0.83/0.67 = 1.25, which is
probably not adequate given the potentially disastrous consequences of a piping failure.
66
8 ONE-DIMENSIONAL SETTLEMENT BEHAVIOUR
Soils are often subjected to uniform loading over large areas, such as shown in Figure
1, from an embankment. Under such conditions, soil which is remote from the edges of
the loaded area undergoes vertical strain, but no horizontal strain. That is strains, and
hence surface settlement, only occur in one dimension.
embankment
𝑥𝑥
soil layer 1
𝑧𝑧
soil layer 2
rock
The accuracy of this assumption depends on the relative dimensions of the loaded area
and thickness of the soil layer. If the area is relatively large and the thickness of the soil
layer relatively small, then the assumption of 1D conditions will be reasonable.
67
8.2 The Oedometer
The behaviour of soil during one-dimensional loading can be tested using a device
called an oedometer, which is shown schematically in Figure 2. The one-dimensional
condition in which the vertical strain 𝜀𝜀𝑧𝑧𝑧𝑧 ≠ 0 and the lateral strains 𝜀𝜀𝑥𝑥𝑥𝑥 = 𝜀𝜀𝑦𝑦𝑦𝑦 = 0 is
also referred to as confined compression.
load displacement-measuring
device
loading cap
cell
water
porous disks
The soil is loaded under conditions of no lateral strain (expansion), as the soil fits tightly
into a relatively rigid ring.
Uncontrolled drainage is provided at the top and bottom of the specimen by porous
discs (two-way drainage). In more sophisticated oedometer apparatus, control of
drainage is possible.
A vertical load is applied to the specimen and a record of the settlement versus time is
made. The load is left on until all settlement ceases (usually 24 hours, although this
depends on the soil type; impermeable clays may take longer).
The load is then increased (usually by a factor of 2, so the vertical stresses might be e.g.
20, 40, 80, 160 kPa). When the maximum load is reached, the soil is unloaded in several
increments. If desired, reloading can be carried out. At each step, time–settlement
records are made.
The relationships between void ratio and effective stress, and settlement and time are
found from the test. The methods by which these are obtained will be explained in the
laboratory classes.
It is conventional to plot the void ratio versus the logarithm of the effective stress in
examining the behaviour of soil, rather than plotting the relationship between effective
stress and strain as is often done in materials testing. The reason for this is that the
relationship between effective stress and void ratio is fundamental to an understanding
68
of soil behaviour. The relationship obtained is similar to that between effective stress
and strain because changes in void ratio and strain are simply related as shown later.
The volume strain, 𝜀𝜀𝑣𝑣 , of an element of material is defined to be the change in volume,
∆𝑉𝑉, divided by initial volume, 𝑉𝑉0.
∆𝑉𝑉
𝜀𝜀𝑣𝑣 = − (1)
𝑉𝑉0
The volume strain is related to the vertical (axial) strain. To show this, consider Figure
3.
a) b)
∆𝑥𝑥
∆𝑥𝑥(1 − 𝜀𝜀𝑥𝑥𝑥𝑥 )
Figure 3: Deformation of a soil element (a) before deformation and (b) after
deformation
𝑉𝑉 − 𝑉𝑉0
𝜀𝜀𝑣𝑣 = − (2b)
𝑉𝑉0
69
8.4 Relation between Volume Strain and Void Ratio
For most soils, the skeletal material is far stiffer than the soil composite and thus,
referring to Figure 4, it can be seen that the relationship between volume strain and void
ratio is:
∆𝑒𝑒
𝜀𝜀𝑧𝑧𝑧𝑧 = − (3b)
1 + 𝑒𝑒0
a) b)
Figure 4: Deformation of soil element (a) before deformation and (b) after
deformation
70
8.5 Behaviour of Soil under One-Dimensional Loading
𝐴𝐴
Void Ratio, e 𝐶𝐶
𝐵𝐵
𝐷𝐷
Upon reloading the soil beyond 𝐵𝐵, the soil continues along the path that it would have
followed if loaded from 𝐴𝐴 to 𝐷𝐷.
If the current effective stress, 𝜎𝜎 ′ , is equal (note that it cannot be greater than) to the pre-
′
consolidation stress, 𝜎𝜎𝑝𝑝𝑝𝑝 , then the deposit is said to be normally consolidated (NC).
𝜎𝜎 ′ = 𝜎𝜎𝑝𝑝𝑝𝑝
′
(normally consolidated) (4a)
During deposition of a soil (which usually takes place through sedimentation), the
weight of the soil (which increases with depth below the surface) causes a decrease in
void ratio. Suppose that at a particular depth below the surface the soil is represented
by point 𝑃𝑃 in Figure 6. If the soil is now subjected to an effective stress increase under
71
1D conditions, the path that will be followed in the 𝑒𝑒 − log10 𝜎𝜎 ′ plot will be along the
extension of the deposition line as shown in Figure 6. A soil which lies at any point on
this line is called normally consolidated and the line is called the normal consolidation
line (NCL).
Normally consolidated soils are usually found as recent alluvial deposits and are mainly
composed of silt and clay sized particles. It is extremely rare to find normally
consolidated soils inland, away from the rivers or lakes in which they were deposited.
𝒆𝒆
impossible
states
𝑃𝑃
normal
consolidation
line
over-consolidated
states
𝜎𝜎 ′ < 𝜎𝜎𝑝𝑝𝑝𝑝
′
(over-consolidated) (4b)
Note:
𝜎𝜎 ′ > 𝜎𝜎𝑝𝑝𝑝𝑝
′
(not possible) (4c)
It may be seen that, for the same increase in effective stress, the change in void ratio
will be much less for an over-consolidated soil (from 𝑒𝑒0 to 𝑒𝑒𝑓𝑓 ) than it would have been
for a normally consolidated soil. Hence, settlements will generally be much smaller for
structures built on over-consolidated soils.
Most soils are over-consolidated to some degree. This can be due to the effects of:
shrinking and swelling of the soil on drying and rewetting; changes in ground water
levels; and unloading due to erosion of overlying strata.
72
𝑂𝑂
𝑄𝑄
𝑒𝑒 = 𝑒𝑒0 𝐹𝐹
𝑒𝑒 = 𝑒𝑒𝑓𝑓 𝑃𝑃
𝑅𝑅
(logarithmic scale)
𝜎𝜎0′ 𝜎𝜎𝑓𝑓′ ′
𝜎𝜎𝑝𝑝𝑝𝑝
The distance from the normal consolidation line has an important influence on the soil
behaviour. This is described numerically by the over-consolidation ratio (OCR). The
OCR is defined as the ratio of the pre-consolidation stress to the current effective stress.
′
𝜎𝜎𝑝𝑝𝑝𝑝
OCR = (5)
𝜎𝜎 ′
𝒆𝒆 𝐷𝐷
𝐶𝐶
𝐴𝐴
𝐹𝐹
𝐵𝐵
′
𝜎𝜎𝑝𝑝𝑝𝑝 𝐥𝐥𝐥𝐥𝐥𝐥 𝟏𝟏𝟏𝟏 𝝈𝝈′
73
Steps in the construction are given below:
1. Determine the point of maximum curvature 𝐴𝐴. (It’s important to draw the graph
to a sensible scale)
4. Draw the extension of the straight line (normally consolidated) portion of the
curve 𝐷𝐷𝐷𝐷.
5. Where the line 𝐷𝐷𝐷𝐷 cuts the bisector (𝐴𝐴𝐴𝐴) of angle 𝐶𝐶𝐶𝐶𝐶𝐶, is the pre-consolidation
stress.
For a normally consolidated soil the pre-consolidation stress will be the same as the
vertical overburden stress (due to weight of overlying soil) existing at the depth from
which the sample was taken. Some unloading of the sample will take place during
sampling so that a pre-consolidation stress may be detected upon reloading in the
oedometer at the point where the soil is loaded back to the stress state existing in the
ground.
74
𝒆𝒆
Figure 10 shows a portion of the 𝑒𝑒 − log (𝜎𝜎 ′ ) plot for a normally consolidated soil.
𝐼𝐼
𝑒𝑒𝐼𝐼
𝐹𝐹
𝑒𝑒𝐹𝐹
Suppose that a soil is in an initial state 𝐼𝐼 and, after loading, moves to the final state, 𝐹𝐹,
as shown in Figure 10.
𝑒𝑒𝐹𝐹 −𝑒𝑒𝐼𝐼 ∆𝑒𝑒
slope of IF = log ′ ′ = log ′ ′ (6a)
10 �𝜎𝜎𝐹𝐹 �−log10 (𝜎𝜎𝐼𝐼 ) 10 (𝜎𝜎𝐹𝐹 /𝜎𝜎𝐼𝐼 )
Because the relationship between effective stress and void ratio can be closely
approximated by a straight line, the slope is a constant. The slope constant, 𝐶𝐶𝑐𝑐 , is called
the compression index.
75
∆𝑒𝑒
= −𝐶𝐶𝑐𝑐 (6b)
log10 (𝜎𝜎𝐹𝐹′ /𝜎𝜎𝐼𝐼′ )
The above equation can be used to calculate the final void ratio from the known final
effective stress and initial conditions as follows:
A similar approach is possible if the soil is over-consolidated and the final stress is less
than the pre-consolidation stress; this is shown in Figure 11.
Again suppose that a soil is at an initial state 𝐼𝐼 and after loading moves to a final state,
𝐹𝐹, as shown in Figure 11. As before, we have:
𝑒𝑒𝐹𝐹 −𝑒𝑒𝐼𝐼 ∆𝑒𝑒
slope of IF = log ′ ′ = log ′ ′ (7a)
10 �𝜎𝜎𝐹𝐹 �−log10 (𝜎𝜎𝐼𝐼 ) 10 (𝜎𝜎𝐹𝐹 /𝜎𝜎𝐼𝐼 )
𝑒𝑒𝐼𝐼 𝐼𝐼
𝑒𝑒𝐹𝐹
𝐹𝐹
As the relationship between effective stress and void ratio is approximately linear, thus:
∆𝑒𝑒
log10 (𝜎𝜎𝐹𝐹′ /𝜎𝜎𝐼𝐼′ )
= −𝐶𝐶𝑟𝑟 (7b)
The constant 𝐶𝐶𝑟𝑟 is called the recompression or swelling index. Again this equation can
be used to determine the final void ratio provided the final effective stress and initial
conditions are known, as follows:
Suppose the initial state of the soil is given by point 1 in Figure 12, the point at which
it reaches the pre-consolidation stress is denoted by 2 and the final state is denoted by
76
3. The resulting change in void ratio as the soil moves from the initial state 1 to the final
state 3 can be considered to occur in two distinct stages: Stage 1 in which the soil is
over-consolidated and Stage 2 in which the soil is normally consolidated.
𝑒𝑒1 1
2
𝑒𝑒2
𝑒𝑒3 3
Stage 1 (1 → 2)
Stage 2 (2 → 3)
Since the soil is normally consolidated, the current state of effective stress will be the
′
pre-consolidation stress and thus the final value of the pre-consolidation stress (𝜎𝜎𝑝𝑝𝑝𝑝 )
′
will be 𝜎𝜎3 .
If the soil is at point 3, where it is normally consolidated, and is unloaded so that the
effective stress drops, the change in void ratio should be determined from Equation 7c
for over-consolidated soil.
77
9 CALCULATION OF SETTLEMENT
𝑥𝑥
𝐻𝐻
Δ𝑆𝑆
𝑧𝑧
For confined compression the horizontal strains are negligible i.e. 𝜀𝜀𝑥𝑥𝑥𝑥 = 0 and 𝜀𝜀𝑦𝑦𝑦𝑦 = 0
and thus:
Δ𝑆𝑆 ∆𝑒𝑒
𝜀𝜀𝑧𝑧𝑧𝑧 = = 𝜀𝜀𝑣𝑣 = −
𝐻𝐻 1 + 𝑒𝑒0
Thus,
∆𝑒𝑒𝑒𝑒
Δ𝑆𝑆 = − (1)
1 + 𝑒𝑒0
the settlement of a thicker layer can be calculated by dividing the layer into a number
of sub layers as shown in Figure 2. This is necessary because both the initial and final
effective stress vary with depth as do the void ratio and the OCR.
The settlement of the soil layer is calculated by calculating the settlement of the
individual sublayers and adding them; in doing this, it is assumed that the void ratio and
78
the effective stress are constant throughout the sublayer and equal to their values at the
centre of the sublayer.
Thus,
Δ𝑒𝑒𝑖𝑖 𝐻𝐻𝑖𝑖
for sublayer 𝑖𝑖, Δ𝑆𝑆𝑖𝑖 = −
1 + 𝑒𝑒𝑖𝑖
so that,
𝑛𝑛 𝑛𝑛
Δ𝑒𝑒𝑖𝑖 𝐻𝐻𝑖𝑖
total settlement, 𝑆𝑆 = � Δ𝑆𝑆𝑖𝑖 = � �− � (2)
1 + 𝑒𝑒𝑖𝑖
i=1 i=1
In solving this problem, it will be assumed that the gravel is far less compressible than
the clay and thus that the settlement of the gravel can be neglected. The settlement of
the clay layer will be calculated by dividing it into two sublayers
2m
GRAVEL
5m 𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑,𝑔𝑔 = 18 kN/m3
𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠,𝑔𝑔 = 22 kN/m3
A 4m
CLAY
𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠,𝑐𝑐 =?
B 4m
1. Unit weights
In order to commence the calculations, it is first necessary to calculate the unit weight
of the clay; this is shown schematically in Figure 4.
79
a) b)
or,
(𝐺𝐺𝑠𝑠 + 𝑒𝑒)𝛾𝛾𝑤𝑤
𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠 = = 19.06 kN/m3
1 + 𝑒𝑒
2. Initial state at 𝐴𝐴
3. Final state at 𝐴𝐴
Notice that the final effective stress exceeds the initial pre-consolidation stress and thus
the clay moves from being initially over-consolidated to finally normally consolidated.
The soil in the first sub layer moves from being over-consolidated to normally
consolidated and so the calculation of the change in void ratio must be made in two
stages.
80
′
Stage 1: Soil over-consolidated (𝜎𝜎 ′ < 𝜎𝜎𝑝𝑝𝑝𝑝,𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 )
′
Δ𝑒𝑒1 = −𝐶𝐶𝑟𝑟 × log10 (𝜎𝜎𝑝𝑝𝑝𝑝,𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖𝑖 /𝜎𝜎𝑖𝑖′ )
′
Stage 2: Soil normally consolidated (𝜎𝜎 ′ = 𝜎𝜎𝑝𝑝𝑝𝑝 )
Now,
𝐻𝐻Δ𝑒𝑒
Δ𝑆𝑆 = −
1 + 𝑒𝑒
4 120.00 179.52
Δ𝑆𝑆 = [0.05 × log10 � � + 0.2 × log10 � �]
1.8 91.12 120.00
Δ𝑆𝑆 = 0.0911 m
5. Initial state at 𝐵𝐵
6. Final state at 𝐵𝐵
now,
𝐻𝐻Δ𝑒𝑒
Δ𝑆𝑆 = −
1 + 𝑒𝑒
4 176.56
Δ𝑆𝑆 = [0.2 × log10 � �]
1.8 128.16
Δ𝑆𝑆 = 0.0620 m
8. Total settlement
81
9.2 Calculation of Stress Changes
The calculation of settlement depends upon knowledge of the initial and final effective
stress within each sub layer of the deposit. The initial effective stress state can be
determined from knowledge of the bulk unit and the position of the water table. The
increase in total stress can be estimated using the theory of elasticity. (Note the soil is
in general not really elastic; however in the working stress range this assumption
provides reasonably accurate estimates of the stress increases due to the applied loads)
𝑥𝑥
𝑧𝑧
𝐻𝐻 → ∞
3𝑃𝑃𝑧𝑧 3
Δσzz =
2𝜋𝜋𝑅𝑅 5
(1 + 𝜈𝜈)𝑃𝑃𝑃𝑃
Δσxx + Δσyy + Δσzz = (3)
𝜋𝜋𝑅𝑅 3
(1 + 𝜈𝜈)𝑃𝑃 𝑧𝑧 2
Δ𝑢𝑢𝑧𝑧 = �2(1 − 𝜈𝜈) + 2 �
2𝜋𝜋𝜋𝜋𝜋𝜋 𝑅𝑅
where:
𝑅𝑅 = �𝑥𝑥 2 + 𝑦𝑦 2 + 𝑧𝑧 2
and,
𝐸𝐸 = Young ′ s modulus
𝜈𝜈 = Poisson′ s ratio
Δ𝑢𝑢𝑧𝑧 = vertical displacement due to load
82
The symbol Δ is used to indicate that each of the quantities in Equation 3 represents the
increase in the particular quantity due to the applied load.
The solution for a point load is important because it can be used to develop solutions
for distributed loads by integration. Some of these solutions are presented in the Soil
Mechanics Data Sheets.
9.3.1 Stresses due to Circular Foundation Loads applied at the Ground Surface
5m
𝑝𝑝 = 100 kPa
𝑟𝑟
2m
𝑧𝑧
5m
A B
There is a simple analytic expression (given in the Data Sheets) for points on the centre
line under a circular load:
3
−
𝑎𝑎2 2
Δσzz = 𝑝𝑝 �1 − �1 + 2 � �
𝑧𝑧 (4)
where:
83
𝑧𝑧 = the depth of interest = 2 m
3
Δσzz = 100 × �1 − [1 + (1.25)2 ]−2 � = 75.6 kPa
In this case, there is no simple analytic expression and the solution must be found by
using the influence charts given in the Data Sheets, reproduced in part in Figure 7. Note
that this chart can also be used for points on the centre line for which 𝑟𝑟 = 0.
Now,
𝑧𝑧 2
= = 0.8
𝑎𝑎 2.5
𝑟𝑟 5
= =2
𝑎𝑎 2.5
and, using the Data Sheets, Δ𝜎𝜎𝑧𝑧𝑧𝑧 /𝑝𝑝 = 0.03 and, so, Δ𝜎𝜎𝑧𝑧𝑧𝑧 = 3.0 kPa
84
Figure 7: Influence factors for a uniformly loaded circular area of radius 𝒂𝒂
85
9.3.2 Stresses due to Rectangular Foundation Loads applied at the Ground Surface
(plan view) 𝐵𝐵
𝐿𝐿
uniformly-distributed surface stress, 𝑝𝑝
𝑧𝑧
point immediately
beneath one of the
rectangle’s corners
(elevation view)
Many loads which occur in practice are applied to foundations that may be considered
to consist of a number of rectangular regions. It is thus of interest to be able to calculate
the vertical stress increases due to a uniformly distributed load acting on a rectangular
loaded area. This is shown schematically in Figure 8.
The vertical stress change at a distance 𝑧𝑧 below one of the corners of the rectangular
load may be determined from a chart which is given in the Data Sheets and is
reproduced in Figure 9.
86
Figure 9: Influence factors for uniformly loaded rectangular areas
This chart can be used to determine the value of stress increase at any point in an elastic
layer, the method for doing this is illustrated below.
This situation is shown schematically in Figure 10. The stress change is required at a
depth 𝑧𝑧 below point 𝑂𝑂.
The first step in using the influence charts is to break the rectangular loading up into a
number of components each having a corner at 𝑂𝑂, this is relatively simple as can be
seen in Figure 10.
87
It thus follows that at the point of interest, the stress increase Δ𝜎𝜎𝑧𝑧𝑧𝑧 (𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴) is given by:
𝐷𝐷 𝑇𝑇 𝐶𝐶
(plan view) 𝑋𝑋 𝑂𝑂 𝑍𝑍
𝐴𝐴 𝑌𝑌 𝐵𝐵
𝑧𝑧
O (point of interest)
(elevation view)
Example 2
Suppose we wish to evaluate the increase in stress at a depth of 2 m below the point O
due to the rectangular loading shown in shown in Fig. 11, when the applied stress over
𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 is 100 kPa.
𝐷𝐷 𝑇𝑇 𝐶𝐶
2m
(plan view) 𝑋𝑋 𝑂𝑂 𝑍𝑍
3m
𝐴𝐴 𝑌𝑌 𝐵𝐵
3m 2m
𝑚𝑚 = 𝐿𝐿/𝑧𝑧 = 1
𝑛𝑛 = 𝐵𝐵/𝑧𝑧 = 1
88
Thus,
∆𝐼𝐼σ = 0.175
and so,
𝑚𝑚 = 𝐿𝐿/𝑧𝑧 = 1.5
𝑛𝑛 = 𝐵𝐵/𝑧𝑧 = 1
Thus,
∆𝐼𝐼σ = 0.194
and so,
𝑚𝑚 = 𝐿𝐿/𝑧𝑧 = 1.5
𝑛𝑛 = 𝐵𝐵/𝑧𝑧 = 1.5
Thus,
∆𝐼𝐼σ = 0.216
and so,
𝑚𝑚 = 𝐿𝐿/𝑧𝑧 = 1.5
𝑛𝑛 = 𝐵𝐵/𝑧𝑧 = 1
Thus,
𝐼𝐼σ = 0.194
and so,
5. Increase in stress
The stress increase at a point vertically below a point 𝑂𝑂 which is outside the loaded area
can also be found using the influence charts shown in Figure 9.
𝑋𝑋 𝑍𝑍 𝑂𝑂
𝐷𝐷 𝐶𝐶 𝑇𝑇
(plan view)
𝐴𝐴 𝐵𝐵 𝑌𝑌
This is achieved by considering the stress 𝑞𝑞 acting on 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴 to consist of the following:
It thus follows that at the point 𝑂𝑂, the stress increase, Δ𝜎𝜎𝑧𝑧𝑧𝑧 (𝐴𝐴𝐴𝐴𝐴𝐴𝐷𝐷), is given by:
90
𝑋𝑋 𝑍𝑍 𝑂𝑂 𝑋𝑋 𝑍𝑍 𝑂𝑂
𝑞𝑞 𝑞𝑞 𝑞𝑞
𝐷𝐷 𝐶𝐶 𝑇𝑇 𝐷𝐷 𝐶𝐶 𝑇𝑇
𝑞𝑞 𝑞𝑞 𝑞𝑞
𝐴𝐴 𝐵𝐵 𝑌𝑌 𝐴𝐴 𝐵𝐵 𝑌𝑌
Stage 1 Stage 2
𝑋𝑋 𝑍𝑍 𝑂𝑂 𝑋𝑋 𝑍𝑍 𝑂𝑂
𝑞𝑞 𝑞𝑞
𝐷𝐷 𝐶𝐶 𝑇𝑇 𝐷𝐷 𝐶𝐶 𝑇𝑇
𝑞𝑞 𝑞𝑞
𝐴𝐴 𝐵𝐵 𝑌𝑌 𝐴𝐴 𝐵𝐵 𝑌𝑌
Stage 3 Stage 4
Example 3
𝑋𝑋 𝑍𝑍 𝑂𝑂
1m
𝐷𝐷 𝐶𝐶 𝑇𝑇
(plan view) 2m
𝐴𝐴 𝐵𝐵 𝑌𝑌
10 m 1m
Suppose the rectangular area 𝐴𝐴𝐴𝐴𝐴𝐴𝐴𝐴, shown in Figure 14, is subjected to a surface stress
of 100 kPa AND it is required to calculate the vertical stress increase at a point 1.5 m
below the point 𝑂𝑂.
91
1. For rectangular loading 𝑂𝑂𝑂𝑂𝑂𝑂𝑂𝑂
𝑚𝑚 = 𝐿𝐿/𝑧𝑧 = 0.67
𝑛𝑛 = 𝐵𝐵/𝑧𝑧 = 0.67
Thus,
∆𝐼𝐼σ = 0.121
and so,
𝑚𝑚 = 𝐿𝐿/𝑧𝑧 = 7.67
𝑛𝑛 = 𝐵𝐵/𝑧𝑧 = 0.67
Thus,
∆𝐼𝐼σ = 0.167
and so,
𝑚𝑚 = 𝐿𝐿/𝑧𝑧 = 7.67
𝑛𝑛 = 𝐵𝐵/𝑧𝑧 = 2.00
Thus,
∆𝐼𝐼σ = 0.240
and so,
𝑚𝑚 = 𝐿𝐿/𝑧𝑧 = 2
𝑛𝑛 = 𝐵𝐵/𝑧𝑧 = 0.67
Thus,
𝐼𝐼σ = 0.164
and so,
92
∆σ𝑧𝑧 = 𝑝𝑝 𝐼𝐼σ = −100 × 0.164 = −16.4 kPa
5. Increase in stress
9.3.3 Stresses due to Foundation Loads of Arbitrary Shape applied at the Ground
Surface
Newmark’s chart provides a graphical method for calculating the stress increase due to
a uniformly loaded region of arbitrary shape resting on a deep homogeneous isotropic
elastic region.
Newmark’s chart is given in the data sheets and is reproduced in part in Figure 10. The
procedure for its use is outlined below
The scale for this procedure is determined by the depth 𝑧𝑧 at which the stress is to be
evaluated, thus 𝑧𝑧 is equal to the distance 𝑂𝑂𝑂𝑂 shown on the chart.
Draw the loaded area to scale so that the point of interest (more correctly its vertical
projection on the surface) is at the origin of the chart, the orientation of the drawing
does not matter
Count the number of squares (𝑁𝑁) within the loaded area; if more than half the square is
in count the square otherwise neglect it.
∆σ𝑧𝑧𝑧𝑧 = 𝑁𝑁 × 0.001 × 𝑝𝑝
Example 4
Suppose a uniformly loaded circle of radius 2 m carries a uniform stress of 100 kPa. It
is required to calculate the vertical stress at a depth of 4 m below the edge of the circle.
The loaded area is drawn on Newmark’s chart to the appropriate scale (i.e. the length
𝑂𝑂𝑂𝑂 is set to represent 4 m) as shown in Figure 10.
It is found that the number of squares, 𝑁𝑁 = 206 and so the stress increase is found to be:
This result can also be checked using the influence charts for circular loading and it is
then found that:
93
𝑧𝑧 𝑟𝑟
= 2 ; = 1 ; ∆σ𝑧𝑧𝑧𝑧 /𝑝𝑝 = 0.2
𝑎𝑎 𝑎𝑎
and so,
∆σ𝑧𝑧𝑧𝑧 = 20 kPa
94
10 ANALYSIS OF CONSOLIDATION
From the response of soils under one-dimensional conditions, it is apparent that when
the effective stress increases, there will be a tendency for the soil to compress. However,
when a load is applied to a saturated soil specimen, this compression does not occur
immediately. This behaviour is a consequence of the soil constituents, the skeletal
material and pore water, being almost incompressible compared to the soil element;
deformation can only take place by water being squeezed out of the voids. This can
only occur at a finite rate and so, initially, when the soil is loaded it undergoes no
volume change.
Under one dimensional conditions, this implies that there can be no vertical strain and
thus no change in vertical effective stress. For 1D conditions, we have:
When the load is first applied, the total stress increases but, as shown above for 1D
conditions, there can be no instantaneous change in vertical effective stress: this implies
that the pore pressure must increase by exactly the same amount as the total stress as:
Subsequently there will be flow from regions of higher excess pore pressure to regions
of lower excess pore pressure, the excess pore pressures will dissipate, the effective
stress will change and the soil will deform (consolidate) with time. This is shown
schematically in Figure 1.
Time Time
Effective Settlement
Stress
Time Time
95
10.2 Derivation of the Equation of Consolidation for 1D Conditions
If we assume that the pore fluid and soil skeleton are incompressible, then:
volume decrease of the soil = volume of pore fluid which flows out
and thus,
rate of volume decrease of the soil = rate at which pore fluid which flows out
𝑣𝑣(𝑧𝑧, 𝑡𝑡)
soil ∆𝑧𝑧
element
𝜕𝜕𝜕𝜕
𝑣𝑣(𝑧𝑧 + Δ𝑧𝑧, 𝑡𝑡) = 𝑣𝑣 + Δz
𝜕𝜕𝜕𝜕
𝜕𝜕𝜕𝜕
[𝑣𝑣(𝑧𝑧 + Δ𝑧𝑧, 𝑡𝑡) − 𝑣𝑣(𝑧𝑧, 𝑡𝑡)]𝐴𝐴 = Δ𝑧𝑧𝑧𝑧
𝜕𝜕𝜕𝜕
𝜕𝜕𝜀𝜀𝑣𝑣
Δ𝑧𝑧𝑧𝑧
𝜕𝜕𝜕𝜕
and thus,
𝜕𝜕𝜕𝜕 𝜕𝜕𝜀𝜀𝑣𝑣
= (3)
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
96
where:
It will also be assumed that Darcy’s Law holds and thus that
𝜕𝜕ℎ
𝑣𝑣 = −k v (4)
𝜕𝜕𝜕𝜕
In applying Darcy’s Law, it is only the velocity due to the consolidation process that is
of interest and consequently the head in Equation 4 is the excess head due to the
consolidation process (not the total head). The excess head is related to the excess pore
water pressure by:
Note that the elevation is not involved in Equation 5 because it only relates the excess
heads and water pressures. From Equations 3–5 it follows that:
∂ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜀𝜀𝑣𝑣
�𝑘𝑘𝑣𝑣 � = − (6)
∂z 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
If it is also assumed that the soil element responds elastically to a change in effective
stress, then:
where:
with,
97
The value of 𝑚𝑚𝑣𝑣 must be determined over the appropriate effective stress range because
it depends on the mean effective stress. This can be seen by considering the relation
between voids ratio and effective stress:
𝑒𝑒 = 𝐴𝐴 − 𝐶𝐶 log10 (𝜎𝜎 ′ )
and hence,
𝐶𝐶𝐶𝐶𝜎𝜎 ′
𝑑𝑑𝑑𝑑 = −
2.3𝜎𝜎 ′
Now,
Δ𝑒𝑒 𝐶𝐶𝐶𝐶𝜎𝜎 ′
𝜀𝜀𝑣𝑣 = − = = 𝑚𝑚𝑣𝑣 𝑑𝑑𝜎𝜎 ′ (9)
1 + 𝑒𝑒 2.3(1 + 𝑒𝑒)𝜎𝜎 ′
Thus, 𝑚𝑚𝑣𝑣 depends on both voids ratio, 𝑒𝑒, and effective stress, 𝜎𝜎 ′ .
At a boundary where the soil is free to drain, the pore water pressure will be constant
and will not change during consolidation. For such a boundary, the excess pore water
pressure will be zero.
𝜕𝜕𝜕𝜕
= 0 at an impermeable boundary (11b)
𝜕𝜕𝜕𝜕
At the instant of loading, there is no volume strain and thus no change in vertical
effective stress. At this instant, the excess pore water pressure will be equal to the initial
increase in total stress.
98
10.5 The Equation of Consolidation for a Homogeneous Soil
𝜕𝜕 2 𝑢𝑢 𝜕𝜕𝜕𝜕 𝜕𝜕𝜎𝜎𝑒𝑒
𝑐𝑐𝑣𝑣 = − (13)
𝜕𝜕𝑧𝑧 2 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
where:
𝑘𝑘𝑣𝑣
the coefficient of consolidation, 𝑐𝑐𝑣𝑣 =
𝑚𝑚𝑣𝑣 𝛾𝛾𝑤𝑤
The coefficient of consolidation (𝑐𝑐𝑣𝑣 ) can be estimated using the oedometer apparatus
as can the coefficient of volume decrease (𝑚𝑚𝑣𝑣 ). The procedure to do this will be
discussed in the laboratory classes. It is difficult (time consuming) to measure the
permeability of clays (𝑘𝑘𝑣𝑣 ) and so the value of permeability is usually inferred from the
values of 𝑐𝑐𝑣𝑣 and 𝑚𝑚𝑣𝑣 .
Figure 3 represents a layer of clay of thickness 2𝐻𝐻 subjected to a uniform surface stress
𝑞𝑞 applied at time 𝑡𝑡 = 0 and held constant thereafter. The clay layer is free to drain at
both its top and bottom boundaries. This is called two-way drainage.
2𝐻𝐻 𝑧𝑧
Figure 3: Homogeneous clay layer free to drain from both upper and lower
boundaries
The increase in stress throughout the layer and does not vary with time and so,
𝜎𝜎𝑒𝑒 = 𝑞𝑞
99
Equation 10 therefore becomes:
𝜕𝜕 2 𝑢𝑢 𝜕𝜕𝜕𝜕
𝑐𝑐𝑣𝑣 = (14a)
𝜕𝜕𝑧𝑧 2 𝜕𝜕𝜕𝜕
The clay layer is free to drain at its upper and lower boundary and so,
Initially, the excess pore pressure will just match the increase in total stress so that there
will be no instantaneous volume change and thus:
where:
1
𝛼𝛼𝑛𝑛 = �𝑛𝑛 + � 𝜋𝜋
2
𝑧𝑧
𝑍𝑍 = (a dimensionless distance)
𝐻𝐻
𝑐𝑐𝑣𝑣 𝑡𝑡
𝑇𝑇𝑣𝑣 = (a dimensionless time)
𝐻𝐻 2
Note that 𝐻𝐻, which occurs in both dimensionless quantities, is the maximum drainage
path length.
The settlement of the soil layer can be determined by summing the vertical (= volume)
strains, giving:
2𝐻𝐻 2𝐻𝐻
and the variation of settlement with time can be obtained by substituting in Equation 15
which gives the variation of u with time and depth.
2𝐻𝐻 ∞
1 2
𝑆𝑆 = � 𝑚𝑚𝑣𝑣 𝑞𝑞 �1 − 2 � sin(𝛼𝛼𝑛𝑛 𝑍𝑍) 𝑒𝑒 −𝛼𝛼𝑛𝑛𝑇𝑇𝑣𝑣 � 𝑑𝑑𝑑𝑑
𝛼𝛼𝑛𝑛
0 𝑛𝑛=0
100
giving,
∞ 2
𝑒𝑒 −𝛼𝛼𝑛𝑛𝑇𝑇𝑣𝑣
𝑆𝑆 = 𝑚𝑚𝑣𝑣 𝑞𝑞2𝐻𝐻 �1 − 2 � � (16b)
𝛼𝛼𝑛𝑛2
𝑛𝑛=0
Noting that the final settlement of the layer, 𝑆𝑆∞ = 𝑚𝑚𝑣𝑣 𝑞𝑞2𝐻𝐻, the settlement may be
written:
∞ 2
𝑆𝑆 𝑒𝑒 −𝛼𝛼𝑛𝑛𝑇𝑇𝑣𝑣
= 𝑈𝑈 = �1 − 2 � �
𝑆𝑆∞ 𝛼𝛼𝑛𝑛2 (16c)
𝑛𝑛=0
The variation of excess pore pressure within the layer is shown in Figure 4 (p. 25, Data
Sheets).
The lines on Figure 4 represent the variation of pore pressure with depth at different
non-dimensionalised times (𝑇𝑇). These lines are known as isochrones. It can be seen that,
initially, the excess pore pressure is constant (𝑢𝑢/𝑞𝑞 = 1) throughout the layer. With time
the pore water flows from the interior of the layer to the drainage boundaries and the
excess pore pressures dissipate until, after a very long time, there are no excess pore
pressures.
101
The variation of settlement with time is most conveniently plotted in the form of the
degree of settlement (𝑈𝑈) versus dimensionless time (𝑇𝑇𝑣𝑣 ) and this is illustrated in Figure
5 (also in the Data Sheets).
There are several useful approximations for the degree of settlement, viz.:
4𝑇𝑇𝑣𝑣
𝑈𝑈 = � (𝑇𝑇𝑣𝑣 ≤ 0.2)
𝜋𝜋
(17)
8 −𝜋𝜋2𝑇𝑇 /4
𝑈𝑈 = 1 − 𝑒𝑒 𝑣𝑣 (𝑇𝑇𝑣𝑣 > 0.2)
𝜋𝜋 2
Alternatively, Figure 5 may be used. It is worth remembering that 𝑈𝑈 = 0.5 when 𝑇𝑇𝑣𝑣 =
0.197.
102
10.6.2 One-Way Drainage
𝐻𝐻 𝑧𝑧
impermeable base
The increase in stress throughout the layer and does not vary with time and so,
𝜎𝜎𝑒𝑒 = 𝑞𝑞
𝜕𝜕 2 𝑢𝑢 𝜕𝜕𝜕𝜕
𝑐𝑐𝑣𝑣 2 = (18a)
𝜕𝜕𝑧𝑧 𝜕𝜕𝜕𝜕
The clay layer is free to drain at its upper boundary and, as before,
𝜕𝜕𝜕𝜕
= 0 when 𝑧𝑧 = 𝐻𝐻 for 𝑡𝑡 > 0 (18c)
𝜕𝜕𝜕𝜕
Initially, the excess pore pressure will just match the increase in total stress so that there
will be no instantaneous volume change and thus:
103
Reference to Figure 4 reveals that Equation 15 also satisfies the condition
𝜕𝜕𝜕𝜕
= 0 when 𝑧𝑧 = 𝐻𝐻 for 𝑡𝑡 > 0
𝜕𝜕𝜕𝜕
and is thus also the solution for one-way drainage (the two-way drainage problem can
be viewed as two one-way drainage problems ‘back to back’). Further examination
reveals that although the expression for final settlement differs for the two cases the
expression for degree of settlement is precisely the same.
Figure 7 shows a soil profile, it can be assumed that the sand and gravel are far more
permeable than the clay and so consolidation in them will have occurred
instantaneously.
gravel
Final settlement = 40 mm
clay 5m 𝑐𝑐𝑣𝑣 = 0.5 m2 /yr
impermeable base
It is assumed that the final settlement has for each of the clay layers has been determined
by the methods described in the previous sections and that their values are as indicated
on Figure 7. It is required to find the settlement after 1 year.
𝑐𝑐𝑣𝑣 𝑡𝑡 0.4 × 1
𝑇𝑇𝑣𝑣 = = = 0.1
𝐻𝐻 2 22
104
2. Settlement of the lower layer
𝑐𝑐𝑣𝑣 𝑡𝑡 0.5 × 1
𝑇𝑇𝑣𝑣 = = = 0.02
𝐻𝐻 2 52
3. Total settlement
𝑐𝑐𝑣𝑣 𝑡𝑡 𝑐𝑐𝑣𝑣 × 2
𝑇𝑇𝑣𝑣 = = = 80 000𝑐𝑐𝑣𝑣
𝐻𝐻 2 0.0052
Since the degree of settlement for the two case is the same, the two values of the
dimensionless time, 𝑇𝑇𝑣𝑣 , are equal and so:
𝑐𝑐𝑣𝑣 × 𝑡𝑡
80 000𝑐𝑐𝑣𝑣 =
100
The data in the previous example can be used to calculate 𝑐𝑐𝑣𝑣 . The dimensionless time
for 50% consolidation is 𝑇𝑇𝑣𝑣 = 0.197 (from Figure 5, 𝑇𝑇𝑣𝑣 ≈ 0.2), thus:
0.197 = 80 000𝑐𝑐𝑣𝑣
105
11 NUMERICAL SOLUTION OF THE 1D CONSOLIDATION EQUATION
11.1 Introduction
The 1D equation of consolidation cannot be solved analytically except for some very
simple situations. For more difficult cases, it is necessary to use approximate numerical
techniques. One numerical technique that can be used for consolidation problems is the
finite difference approach. In this method, the solution is evaluated at a number of
points at different times as indicated on Figure 1 below.
𝑡𝑡 = 0 𝑡𝑡 = 𝑡𝑡1 𝑡𝑡 = 𝑡𝑡2
𝛥𝛥𝛥𝛥 𝒕𝒕
1
𝛥𝛥𝛥𝛥
2
4
𝒛𝒛
The 1D consolidation equation and the boundary conditions are approximated by finite
difference formulae. These can be derived by referring to Figure 2 below and taking
local axes at 𝐵𝐵:
𝐴𝐴 𝑢𝑢 = 𝑢𝑢𝐴𝐴 𝑧𝑧 = −Δ𝑧𝑧
𝐵𝐵 𝑢𝑢 = 𝑢𝑢𝐵𝐵 𝑧𝑧 = 0
u
z
𝐶𝐶 𝑢𝑢 = 𝑢𝑢𝐶𝐶 𝑧𝑧 = +Δ𝑧𝑧
106
Suppose that the excess pore pressure at any time 𝑡𝑡 can be approximated by a parabola:
The constants in this equation can be related to the values of the excess pore pressures
at points 𝐴𝐴, 𝐵𝐵, 𝐶𝐶. Taking 𝐵𝐵 as the origin for 𝑧𝑧 gives:
so that:
𝑎𝑎1 = 𝑢𝑢𝐵𝐵
𝑢𝑢𝐶𝐶 − 𝑢𝑢𝐴𝐴
𝑎𝑎2 =
2Δ𝑧𝑧 (1c)
𝑢𝑢𝐴𝐴 + 𝑢𝑢𝐶𝐶 − 2𝑢𝑢𝐵𝐵
𝑎𝑎3 =
2Δ𝑧𝑧 2
Thus, evaluating the slope and curvature of u at the point 𝐵𝐵 (𝑧𝑧 = 0), it is found:
𝜕𝜕 2 𝑢𝑢 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
𝑐𝑐𝑣𝑣 2
= − (2a)
𝜕𝜕𝑧𝑧 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
where 𝑞𝑞 is the change in total stress, due to applied loads, from the initial equilibrium
situation when the excess pore pressures were zero.
When this equation is evaluated at any point in the soil, it is equivalent to evaluating
the equation at point 𝐵𝐵 and hence the finite difference formulae developed above can
be introduced so the equation becomes:
107
If the above equation is now integrated from times 𝑡𝑡 to 𝑡𝑡 + Δ𝑡𝑡, it is found that:
𝑡𝑡+Δ𝑡𝑡
𝑐𝑐𝑣𝑣
Δ𝑢𝑢𝐵𝐵 = Δ𝑞𝑞𝐵𝐵 + 2 � [𝑢𝑢𝐴𝐴 + 𝑢𝑢𝐶𝐶 − 2𝑢𝑢𝐵𝐵 ] 𝑑𝑑𝑑𝑑 (2c)
Δ𝑧𝑧
𝑡𝑡
where Δ𝑢𝑢𝐵𝐵 = 𝑢𝑢𝐵𝐵 (𝑡𝑡 + Δ𝑡𝑡) − 𝑢𝑢𝐵𝐵 (𝑡𝑡) and Δ𝑞𝑞𝐵𝐵 = 𝑞𝑞𝐵𝐵 (𝑡𝑡 + Δ𝑡𝑡) − 𝑞𝑞𝐵𝐵 (𝑡𝑡).
𝑡𝑡+Δ𝑡𝑡
� 𝐹𝐹(𝑡𝑡) 𝑑𝑑𝑑𝑑 ≈ 𝐹𝐹(𝑡𝑡)Δ𝑡𝑡
𝑡𝑡
𝑡𝑡 𝑡𝑡 + Δ𝑡𝑡 𝑡𝑡
Figure 3: Approximate integral evaluation
where:
𝑐𝑐𝑣𝑣 Δ𝑡𝑡
𝛽𝛽 =
Δ𝑧𝑧 2
or,
𝑢𝑢𝐵𝐵 (𝑡𝑡 + Δ𝑡𝑡) = Δ𝑞𝑞𝐵𝐵 + 𝑢𝑢𝐵𝐵 (𝑡𝑡) + 𝛽𝛽[𝑢𝑢𝐴𝐴 (𝑡𝑡) + 𝑢𝑢𝐶𝐶 (𝑡𝑡) − 2𝑢𝑢𝐵𝐵 (𝑡𝑡)] (2e)
Suppose the solution for 𝑢𝑢 has been found up to time 𝑡𝑡. The applied load will be known
at time 𝑡𝑡 + Δ𝑡𝑡 and so the quantity Δ𝑞𝑞 is known. This means all the quantities on the
right hand side of Equation 2e are known and thus that 𝑢𝑢 at time 𝑡𝑡 + Δ𝑡𝑡 can be
calculated. Thus, a knowledge of the distribution of 𝑢𝑢 at time 𝑡𝑡 means that the
distribution of 𝑢𝑢 at time 𝑡𝑡 + Δ𝑡𝑡 can be inferred. Now, the initial distribution of 𝑢𝑢 can
always be determined and thus the solution can be found by ‘marching’ forward in time.
11.4 Stability
108
𝑐𝑐𝑣𝑣 Δ𝑡𝑡 1
𝛽𝛽 = ≤
Δ𝑧𝑧 2 2
At a free draining boundary, there is no impediment to flow and so the pore pressure
remains constant and thus the excess pore water pressure is zero; this is illustrated in
Figure 4a.
drainage boundary, 𝑢𝑢 = 0
𝐵𝐵
saturated soil
saturated soil
𝐴𝐴
Δ𝑧𝑧
𝐵𝐵
At an impermeable boundary, such as that illustrated in Figure 4b, there can be no flow
in a direction perpendicular to the boundary. As outlined earlier, this implies:
𝜕𝜕𝜕𝜕
=0 (3a)
𝜕𝜕𝜕𝜕
109
The finite difference analogue of this equation is:
𝑢𝑢𝐶𝐶 − 𝑢𝑢𝐴𝐴
=0 (3b)
2Δ𝑧𝑧
and hence,
Suppose that a 4 m layer of clay, shown in Figure 5, which is free to drain at its upper
boundary and rests on an impermeable base, is subjected to a surface loading of 64 kPa.
𝑞𝑞 = 64 kPa
4 sublayers
4m 𝑐𝑐𝑣𝑣 = 2 m2 /yr
𝑚𝑚𝑣𝑣 = 0.0003 m2 /kN
impermeable bedrock
110
1. Divide the deposit into layers—this fixes the value of Δ𝑧𝑧
In this case, the deposit is divided into 4 sublayers (all with the same thickness) and
thus Δ𝑧𝑧 = 1 m.
Thus,
5. March the solution forward using the finite difference equation and introducing the
boundary conditions
111
6. Calculate settlement
S = � 𝜀𝜀𝑣𝑣 𝑑𝑑𝑑𝑑
0
𝐻𝐻
In the above equation, the integral of the excess pore pressure cannot be evaluated
exactly because the excess pore pressures are only calculated at the grid points.
However, the integral can be evaluated approximately using numerical techniques. The
simplest approach, and that used here, is to use the trapezoidal method.
Thus,
𝐻𝐻
1 1
� 𝑢𝑢 𝑑𝑑𝑑𝑑 ≈ (𝑢𝑢0 + 𝑢𝑢1 )Δ𝑧𝑧 + ⋯ + (𝑢𝑢𝑛𝑛−1 + 𝑢𝑢𝑛𝑛 ) Δ𝑧𝑧
2 2
0 (6)
𝑢𝑢0 + 𝑢𝑢𝑛𝑛
= Δ𝑧𝑧 �� � + 𝑢𝑢1 + ⋯ + 𝑢𝑢𝑛𝑛−1 �
2
where:
𝑢𝑢𝑖𝑖 = 𝑢𝑢𝑖𝑖 (Δ𝑧𝑧, 𝑡𝑡)
112
The settlement at other times can be similarly calculated from the excess pore pressures
hence the values can be determined as shown in the table below.
Suppose now that the previous example is solved using a step size of 2 months but
keeping the number of layers the same. If this is the case, β = 1/3. The numerical
solution proceeds as above, but now using the more complex form of the finite
difference equation, viz. Equation 2e; the solution is shown in the table below:
𝑡𝑡 (mths) 0 2 4 6 8 10 12 14 16 18
𝑞𝑞 (kPa) 64.00 64.00 64.00 64.00 64.00 64.00 64.00 64.00 64.00 64.00
𝑧𝑧 = 0 m 64.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
𝑧𝑧 = 1 m 64.00 64.00 42.67 35.56 30.81 27.65 25.28 23.44 21.92 20.61
𝑧𝑧 = 2 m 64.00 64.00 64.00 56.89 52.15 48.20 45.04 42.32 39.92 37.74
𝑧𝑧 = 3 m 64.00 64.00 64.00 64.00 61.63 59.26 56.63 53.99 51.39 48.86
𝑧𝑧 = 4 m 64.00 64.00 64.00 64.00 64.00 62.42 60.31 57.85 55.28 52.68
dummy 64.00 64.00 64.00 64.00 61.63 59.26 56.63 53.99 51.39 48.86
settlement (mm) 0.00 9.60 16.00 20.27 23.82 26.90 29.67 32.20 34.54 36.73
𝑢𝑢𝐵𝐵 (𝑡𝑡 + Δ𝑡𝑡) = Δ𝑞𝑞𝐵𝐵 + 𝑢𝑢𝐵𝐵 (𝑡𝑡) + 𝛽𝛽[𝑢𝑢𝐴𝐴 (𝑡𝑡) + 𝑢𝑢𝐶𝐶 (𝑡𝑡) − 2𝑢𝑢𝐵𝐵 (𝑡𝑡)]
= 0 + 59.26 + 0.3333 × [48.20 + 62.42 − 2 × 59.26] = 56.63 kPa
The results for the two analyses are quite close. After 18 months, the settlement
predicted in Example 1 is 36 mm, which compares well with the settlement calculated
in Example 2, viz. 36.7 mm.
Suppose fill having unit weight 20 kN/m3 is placed at a rate of 0.5 m/month for 12
months after which no more load is applied.
The analysis only differs from that in the previous examples in that the value of Δ𝑞𝑞
needs to be included in the finite difference equation. Choosing β = 0.5 with 4 layers
gives a time step of 0.25 years as before. The results are shown in the table below:
113
𝑡𝑡 (yrs) 0 0.25 0.5 0.75 1 1.25 1.5
𝑞𝑞 (kPa) 0 30 60 90 120 120 120
𝑧𝑧 = 0 m 0 0 0 0 0 0 0
𝑧𝑧 = 1 m 0 30 45 60 71.25 52.5 46.88
𝑧𝑧 = 2 m 0 30 60 82.5 105 93.75 82.5
𝑧𝑧 = 3 m 0 30 60 90 116.25 112.5 105
𝑧𝑧 = 4 m 0 30 60 90 120 116.25 112.5
dummy 0 30 60 90 116.25 112.5 105
settlement (mm) 0 4.5 13.5 24.75 38.25 48.94 56.81
Note that when the load is applied gradually, the excess pore pressure at the permeable
upper boundary remains at zero. This is because there is no instantaneous change in
load.
If the calculation is repeated for the case in which there are 5 sublayers and a time step
of 0.1 years is adopted, this gives 𝛽𝛽 = 0.3125 and the results are shown below:
𝑡𝑡 (yrs) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
𝑞𝑞 (kPa) 0.00 12.00 24.00 36.00 48.00 60.00 72.00 84.00 96.00
𝑧𝑧 = 0 m 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
𝑧𝑧 = 0.8 m 0.00 12.00 20.25 27.09 33.04 38.38 43.27 47.80 52.04
𝑧𝑧 = 1.6 m 0.00 12.00 24.00 34.83 44.78 54.00 62.64 70.78 78.50
𝑧𝑧 = 2.4 m 0.00 12.00 24.00 36.00 47.63 58.86 69.66 80.07 90.11
𝑧𝑧 = 3.2 m 0.00 12.00 24.00 36.00 48.00 59.89 71.60 83.10 94.35
𝑧𝑧 = 4 m 0.00 12.00 24.00 36.00 48.00 60.00 71.93 83.72 95.33
dummy 0.00 12.00 24.00 36.00 48.00 59.89 71.60 83.10 94.35
settlement (mm) 0.00 1.44 3.78 6.74 10.21 14.13 18.45 23.13 28.16
Again the settlements at 1.5 years are quite similar. Thus, although greater refinement
of the grid leads to more accurate excess pore pressures and settlements, there is, in
practice, little advantage to using Β ≠ 0.5.
114
Example 4: Abrupt Change of Load
Suppose that, in the case detailed in Example 1, a further surcharge of 32 kPa is added
after 12 months. The solution in this case is best handled in two stages.
Stage 1 follows exactly the path outlined in Example 1 and is detailed in the table below.
Just after 1 year, the load is abruptly increased and, since there can be no instantaneous
volume strain, there can be no increase in effective stress and no change in settlement.
This means that the increase or decrease in applied stress must be matched by a
corresponding increase or decrease in pore water pressure. This enables the excess pore
water pressure to be calculated.
Stage 2 of the calculation then proceeds in the same way as in Stage 1 or in Example 1.
This is shown in the table given below:
115
12 SETTLEMENTS OF STRUCTURES
maximum
settlement
soil layer
As discussed earlier, the skeletal soil material and the pore water are relatively
incompressible and any change in volume can only occur due to change in the volume
of the voids. For the volume of the voids to change, pore water must flow into or out
of a soil element. Because this cannot happen instantaneously, when a load is first
applied to a soil, there cannot be any immediate change in its volume. For one-
dimensional conditions with no lateral strain, this implies that there is no immediate
vertical strain and hence that the excess pore pressure is equal to the change in vertical
stress. However, under more general conditions, both lateral (or horizontal) and vertical
strains can occur. Immediately after load is applied, there will be no change in volume
but the soil deformations will result in an initial settlement. This is said to occur under
undrained conditions because no pore water has been able to drain from the soil. With
time, the excess pore pressures generated during the undrained loading will dissipate
and further lateral and vertical strains will occur. Ultimately, the settlement will reach
its long term or drained value.
When the load is first applied to the soil, there will be a tendency for the more highly
stressed parts of the soil to compress and thus for there to be a reduction in the volume
of the voids. The pore water will respond to this tendency towards a decrease in volume
by undergoing an increase in pore water pressure and so initial excess pore water
pressures will develop. Subsequently, there will be a flow of water from regions of
high excess pore water pressure to regions of low excess pore water pressure and the
load induced excess pore water pressures will dissipate. This is the process of
consolidation and, during this process, the soil will undergo a settlement which varies
with time. Ultimately, after a long period of time, all the excess pore water pressures
will have dissipated and the settlement of the soil will cease and it will reach its long
term or drained settlement (the term drained is used because all excess pore water
pressures have dissipated and there will be no further drainage of water from the voids
although the voids will still remain saturated). The process of consolidation is shown
schematically in Figure 2.
116
It should be stated that the process described above represents a simplification because
some soils tend to creep. For such soils, there will be additional creep settlements even
though the effective stress does not change.
Total
Stress
Time
Excess Pore
Pressure
Time
Effective
Stress
Time
Figure 2a: Variation of stress and pore pressure at a typical point under a
footing
Settlement
consolidation final
settlement settlement
initial settlement
Time
117
12.2 Analysis of Settlement under Three-Dimensional Conditions
Previously the settlement under foundations has been estimated assuming purely one-
dimensional conditions. However, it is clear from consideration of the stress changes
(predicted by the theory of elasticity) under the centre and edges of various loaded areas
that, in general, the stress changes may differ significantly from those deduced using
the purely one-dimensional assumption.
If it is hypothesised that the soil can be treated as a linear isotropic elastic material, then
solutions for the settlement can be obtained using the theory of elasticity. This
assumption involves a considerable level of approximation which is necessary because:
In an isotropic elastic solid it is found that Hooke’s Law relates the changes in stress to
the changes in strain as described in Equation 1:
where Δ𝜀𝜀𝑥𝑥𝑥𝑥 , Δ𝜀𝜀𝑦𝑦𝑦𝑦 , Δ𝜀𝜀𝑧𝑧𝑧𝑧 denote the strains which arise from the changes in stress Δ𝜎𝜎𝑥𝑥𝑥𝑥 ,
Δ𝜎𝜎𝑦𝑦𝑦𝑦 , Δ𝜎𝜎𝑧𝑧𝑧𝑧 and where 𝐸𝐸 is Young’s modulus and 𝜈𝜈 is Poisson’s ratio.
Hooke’s Law in this form does not apply to soil except for undrained conditions, which
will be discussed later. For soil, the correct relationship is one between effective stress
and strain as shown below:
118
′
Δ𝜎𝜎𝑥𝑥𝑥𝑥 − 𝜈𝜈 ′ �Δ𝜎𝜎𝑦𝑦𝑦𝑦
′ ′
+ Δ𝜎𝜎𝑧𝑧𝑧𝑧 �
Δ𝜀𝜀𝑥𝑥𝑥𝑥 = ′
𝐸𝐸
′
Δ𝜎𝜎𝑦𝑦𝑦𝑦 ′
− 𝜈𝜈 ′ (Δ𝜎𝜎𝑧𝑧𝑧𝑧 ′ )
+ Δ𝜎𝜎𝑥𝑥𝑥𝑥
Δ𝜀𝜀𝑦𝑦𝑦𝑦 = (1b)
𝐸𝐸 ′
′
Δ𝜎𝜎𝑧𝑧𝑧𝑧 ′
− 𝜈𝜈 ′ �Δ𝜎𝜎𝑥𝑥𝑥𝑥 ′
+ Δ𝜎𝜎𝑦𝑦𝑦𝑦 �
Δ𝜀𝜀𝑧𝑧𝑧𝑧 = ′
𝐸𝐸
where 𝐸𝐸 ′ is called the effective stress (drained) Young’s modulus and 𝜈𝜈 ′ is called the
effective stress (drained) Poisson’s ratio and where the increments of effective stress
are related to the increments of total stress and the increment of pore water pressure by:
′
Δ𝜎𝜎𝑥𝑥𝑥𝑥 = Δ𝜎𝜎𝑥𝑥𝑥𝑥 − Δ𝑢𝑢
Δσ′𝑦𝑦𝑦𝑦 = Δ𝜎𝜎𝑦𝑦𝑦𝑦 − Δ𝑢𝑢 (1c)
′
Δ𝜎𝜎𝑧𝑧𝑧𝑧 = Δ𝜎𝜎𝑥𝑥𝑥𝑥 − Δ𝑢𝑢
The relationship between effective stress and strain can always be used to calculate the
deformation of soils. However, to do so it is necessary to know both the change in total
stress and the change in pore water pressure. The change in total stress can usually be
estimated using elastic solutions but the change in pore pressure is, in general, very
difficult to determine.
One important case where the effective stresses are known is in the long term. In this
situation all excess pore water pressures have dissipated and thus the change in effective
stress is equal to the change in total stress. The settlement can then be calculated using
the effective stress–strain relations.
′
Δ𝜎𝜎𝑥𝑥𝑥𝑥 − 𝜈𝜈 ′ �Δ𝜎𝜎𝑦𝑦𝑦𝑦
′ ′
+ Δ𝜎𝜎𝑧𝑧𝑧𝑧 �
Δ𝜀𝜀𝑥𝑥𝑥𝑥 = ′
𝐸𝐸
′
Δ𝜎𝜎𝑥𝑥𝑥𝑥 ′
(1 + 𝜈𝜈 ′ ) − 𝜈𝜈 ′ �Δ𝜎𝜎𝑥𝑥𝑥𝑥 ′
+ Δ𝜎𝜎𝑦𝑦𝑦𝑦 ′
+ Δ𝜎𝜎𝑧𝑧𝑧𝑧 �
= (2)
𝐸𝐸 ′
′
Δ𝜎𝜎𝑥𝑥𝑥𝑥 (1 + 𝜈𝜈 ′ ) − 3𝜈𝜈 ′ (Δ𝜎𝜎𝑚𝑚 ′ )
=
𝐸𝐸 ′
where:
′ ′ ′
′
�Δ𝜎𝜎𝑥𝑥𝑥𝑥 + Δ𝜎𝜎𝑦𝑦𝑦𝑦 + Δ𝜎𝜎𝑧𝑧𝑧𝑧 �
mean effective stress, Δ𝜎𝜎𝑚𝑚 =
3
119
12.4 Behaviour of an Elastic Soil under Undrained Conditions
It was shown above that the long term behaviour of soil can be analysed using Hooke’s
Law since all excess pore pressures have dissipated and so the effective stress equals
the total stress. Another important case which can be analysed using Hooke’s Law is
immediately after loading when no water has drained out of the soil pores and no excess
pore pressures have dissipated, i.e. undrained behaviour. To establish this, note that,
under such conditions, there can be no volume change and thus:
The volume strain can be calculated using Equations 2 and 3a, giving:
If the volume strain is zero, the change in mean effective stress is zero and thus:
′
3(1 − 2𝜈𝜈 ′ )Δ𝜎𝜎𝑚𝑚
Δ𝜀𝜀𝑣𝑣 = =0
𝐸𝐸 ′
Then,
′
Δ𝜎𝜎𝑚𝑚 = Δ𝜎𝜎𝑚𝑚 − Δ𝑢𝑢 = 0
Thus,
This enables the increment in excess pore water pressure to be expressed in terms of
the total stress. Using this relation and substitution into Equation 2 leads to the
following relation between total stress and strain:
This and similar expressions for Δ𝜀𝜀𝑦𝑦𝑦𝑦 and Δ𝜀𝜀𝑧𝑧𝑧𝑧 are equivalent to Hooke’s Law for
undrained loading, which may be written as:
120
Δσ𝑥𝑥𝑥𝑥 − 𝜈𝜈𝑢𝑢 �Δσ𝑦𝑦𝑦𝑦 + Δσ𝑧𝑧𝑧𝑧 �
Δ𝜀𝜀𝑥𝑥𝑥𝑥 =
𝐸𝐸𝑢𝑢
Δσ𝑦𝑦𝑦𝑦 − 𝜈𝜈𝑢𝑢 (Δσ𝑧𝑧𝑧𝑧 + Δσ𝑥𝑥𝑥𝑥 )
Δ𝜀𝜀𝑦𝑦𝑦𝑦 = (5)
𝐸𝐸𝑢𝑢
Δσ𝑧𝑧𝑧𝑧 − 𝜈𝜈𝑢𝑢 �Δσ𝑥𝑥𝑥𝑥 + Δσ𝑦𝑦𝑦𝑦 �
Δ𝜀𝜀𝑧𝑧𝑧𝑧 =
𝐸𝐸𝑢𝑢
The quantities 𝐸𝐸𝑢𝑢 and 𝜈𝜈𝑢𝑢 are called the undrained Young’s modulus and Poisson’s ratio
respectively. By comparing Equations 4 and 5, it can be seen that these quantities are
related to the drained or effective stress relations as follows:
3E ′
Eu =
2(1 + 𝜈𝜈 ′ ) (6)
𝜈𝜈𝑢𝑢 = 0.5
It is interesting to note that, so far, there has been no mention of shear behaviour; for
shear stresses and strains, Hooke’s Law may be written as:
Δσyz
Δ𝛾𝛾𝑦𝑦𝑦𝑦 =
𝐺𝐺 ′
Δσ𝑧𝑧𝑧𝑧
Δ𝛾𝛾𝑧𝑧𝑧𝑧 = ′ (7)
𝐺𝐺
Δσ𝑥𝑥𝑥𝑥
Δ𝛾𝛾𝑥𝑥𝑥𝑥 =
𝐺𝐺 ′
where 𝐺𝐺 ′ is a material property called the shear modulus which is related to the effective
stress parameters as follows.
′
𝐸𝐸 ′
𝐺𝐺 = (8)
2(1 + 𝜈𝜈 ′ )
𝐸𝐸 ′ 𝐸𝐸𝑢𝑢 𝐸𝐸𝑢𝑢
𝐺𝐺 ′ = = = = 𝐺𝐺𝑢𝑢 (9)
2(1 + 𝜈𝜈 )
′ 3 2(1 + 𝜈𝜈𝑢𝑢 )
This shows that the shear modulus (and shear strain) is unaffected by the state of
drainage in the soil.
It is important to emphasise that the relation between effective stress parameters and
undrained parameters is based on many approximations (soil assumed elastic) and
should not be expected to be exact. Thus, although the undrained value of Poisson’s
121
ratio will be precisely 1/2 for a saturated soil because of incompressibility, the
undrained Young’s modulus should be measured directly rather than determined from
the effective E´ value.
• The values of 𝐸𝐸 ’ and 𝜈𝜈 ′ are both dependent on the mean effective stress,
1 ′ ′ ′
3
�Δ𝜎𝜎𝑥𝑥𝑥𝑥 + Δ𝜎𝜎𝑦𝑦𝑦𝑦 + Δ𝜎𝜎𝑧𝑧𝑧𝑧 �, with the moduli increasing with stress level.
• It is advisable to use an appropriate secant modulus for the expected strain level
under the footing.
Typical values may be selected from the following values given in the p. 65 of the Data
Sheets.
122
These typical values should be used with caution. Soils are extremely variable materials
and considerable expertise is needed to determine accurate parameters.
A cuboidal soil specimen is in equilibrium with a uniform stress acting on all faces of
100 kPa and no pore pressure, that is, 𝑢𝑢 = 0. The vertical stress is then increased by
90 kPa with the stresses on the other faces remaining constant and with the sample
prevented from draining. Calculate the vertical and lateral strains if 𝐸𝐸 ′ = 10 MPa and
1
𝜈𝜈 ′ = 4.
𝜈𝜈𝑢𝑢 = 0.5
3E ′
Eu = = 12 MPa
2(1 + 𝜈𝜈 ′ )
Now, the total stress changes are Δ𝜎𝜎𝑥𝑥𝑥𝑥 = 0 kPa, Δ𝜎𝜎𝑦𝑦𝑦𝑦 = 0 kPa, Δ𝜎𝜎𝑧𝑧𝑧𝑧 = 90 kPa.
Hence,
90
Δ𝜀𝜀𝑧𝑧𝑧𝑧 = = 0.0075
12000
Changes in effective stress are needed to evaluate the effective Hooke’s Law relations.
123
�Δ𝜎𝜎𝑥𝑥𝑥𝑥 + Δ𝜎𝜎𝑦𝑦𝑦𝑦 + Δ𝜎𝜎𝑧𝑧𝑧𝑧 �
Δ𝑢𝑢 = Δ𝜎𝜎𝑚𝑚 =
3
90
= = 30 kPa
3
′ ′ ′
Hence Δ𝜎𝜎𝑥𝑥𝑥𝑥 = −30 kPa, Δ𝜎𝜎𝑦𝑦𝑦𝑦 = −30 kPa, Δ𝜎𝜎𝑧𝑧𝑧𝑧 = 60 kPa.
′
Δ𝜎𝜎𝑥𝑥𝑥𝑥 − 𝜈𝜈 ′ �Δ𝜎𝜎𝑦𝑦𝑦𝑦 ′ ′
+ Δ𝜎𝜎𝑧𝑧𝑧𝑧 � −30 − 0.25(−30 + 60)
Δ𝜀𝜀𝑥𝑥𝑥𝑥 = = = −0.00375
𝐸𝐸 ′ 10000
′ ′ (Δ𝜎𝜎 ′ ′ )
Δ𝜎𝜎𝑦𝑦𝑦𝑦 − 𝜈𝜈 𝑧𝑧𝑧𝑧 + Δ𝜎𝜎𝑥𝑥𝑥𝑥
Δ𝜀𝜀𝑦𝑦𝑦𝑦 = = −0.00375
𝐸𝐸 ′
′
Δ𝜎𝜎𝑧𝑧𝑧𝑧 − 𝜈𝜈 ′ �Δ𝜎𝜎𝑥𝑥𝑥𝑥′ ′
+ Δ𝜎𝜎𝑦𝑦𝑦𝑦 � 60 − 0.25(−30 × 2)
Δ𝜀𝜀𝑧𝑧𝑧𝑧 = ′
= = 0.0075
𝐸𝐸 10000
If the same sample from Example 1 is now allowed to drain and consolidate, without
any change to the applied stresses, what strains will develop?
Only an effective stress analysis is relevant. Total stress analysis cannot be used because
the total stress parameters (𝐸𝐸𝑢𝑢 ,𝜈𝜈𝑢𝑢 ) are only relevant to undrained loading, that is, when
deformation occurs at constant volume.
In this example, during consolidation the total stresses remain constant. The effective
′ ′ ′
stress changes are thus: Δ𝜎𝜎𝑥𝑥𝑥𝑥 = +30 kPa, Δ𝜎𝜎𝑦𝑦𝑦𝑦 = +30 kPa, Δ𝜎𝜎𝑥𝑥𝑥𝑥 = +30 kPa (they
are all equal to the reduction in pore water pressure). Then, from Hooke’s Law:
Note that the total strains due to the undrained loading followed by consolidation are:
The same total strains are obtained if the load is applied slowly so that no pore pressures
are obtained. In this case, the pore pressure change is zero and hence the change in total
′ ′
stress is the same as the change in effective stress (Δ𝜎𝜎𝑥𝑥𝑥𝑥 = 0 kPa, Δ𝜎𝜎𝑦𝑦𝑦𝑦 = 0 kPa,
′
Δ𝜎𝜎𝑥𝑥𝑥𝑥 = +90 kPa). The strains are then given by:
′
Δ𝜎𝜎𝑧𝑧𝑧𝑧 ′
− 𝜈𝜈 ′ �Δ𝜎𝜎𝑥𝑥𝑥𝑥 ′
+ Δ𝜎𝜎𝑦𝑦𝑦𝑦 ′
� Δ𝜎𝜎𝑧𝑧𝑧𝑧
Δ𝜀𝜀𝑧𝑧𝑧𝑧 = = ′ = 0.009
𝐸𝐸 ′ 𝐸𝐸
124
′
Δ𝜎𝜎𝑥𝑥𝑥𝑥 ′
− 𝜈𝜈 ′ �Δ𝜎𝜎𝑦𝑦𝑦𝑦 ′
+ Δ𝜎𝜎𝑧𝑧𝑧𝑧 ′ )
� −𝜈𝜈 ′ (Δ𝜎𝜎𝑧𝑧𝑧𝑧
Δ𝜀𝜀𝑥𝑥𝑥𝑥 = = = Δ𝜀𝜀𝑦𝑦𝑦𝑦 = −0.00225
𝐸𝐸 ′ 𝐸𝐸 ′
Note that the strains are identical to those determined as a result of undrained loading
followed by consolidation. This result is not surprising when it is remembered that this
is an elastic analysis.
125
13 SETTLEMENT OF STRUCTURES
𝑃𝑃
𝐻𝐻 soil layer
rigid bedrock
(1 + ν′ )Δ𝜎𝜎𝑧𝑧𝑧𝑧
′
− 𝜈𝜈 ′ �Δ𝜎𝜎𝑥𝑥𝑥𝑥
′ ′
+ Δ𝜎𝜎𝑦𝑦𝑦𝑦 ′
+ Δ𝜎𝜎𝑧𝑧𝑧𝑧 �
Δ𝜀𝜀𝑧𝑧𝑧𝑧 = (1b)
𝐸𝐸 ′
It has been shown previously how the Boussinesq solution for the stresses in an elastic
half space due to a point load acting on the surface can be used to determine the stress
distributions under a variety of shapes of loaded areas (circles, rectangles and arbitrary
126
shapes). The same solution can be used to determine the surface settlements, 𝑆𝑆𝑟𝑟 as a
function of the distance, 𝑟𝑟, from a point load 𝑄𝑄, as:
𝑄𝑄(1 − 𝜈𝜈 2 )
𝑆𝑆𝑟𝑟 = (2)
𝜋𝜋𝜋𝜋𝜋𝜋
𝑄𝑄
𝑟𝑟
𝑆𝑆𝑟𝑟
𝐻𝐻 → ∞
𝑄𝑄(1 − 𝜈𝜈 2 )
𝑆𝑆𝑟𝑟 =
𝜋𝜋𝜋𝜋𝜋𝜋
𝑑𝑑𝑑𝑑
𝑎𝑎
𝑑𝑑𝑑𝑑
𝑟𝑟
127
The settlement is then given by:
𝑎𝑎 2𝜋𝜋
1 − 𝜈𝜈 2
𝑆𝑆𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 = � � 𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞𝑞
𝜋𝜋𝜋𝜋𝜋𝜋
0 0 (3)
2𝑞𝑞(1 − 𝜈𝜈 2 )𝑎𝑎
=
𝐸𝐸
For other positions under the circular load and for other shapes the integration is not so
straightforward and, in many cases, analytical solutions will not be possible.
Also, a limitation of this (Boussinesq) solution is that it assumes the soil layer is
infinitely deep. This rarely occurs in practice as more generally a relatively shallow soil
layer usually overlies rock.
The procedure adopted in practice is to make use of charted solutions that are available
for a number of commonly encountered situations. Some of these are given in the data
sheets and are discussed below. For other solutions, the book "Elastic solutions for Soil
and Rock Mechanics" by Poulos and Davis should be referred to.
𝑃𝑃 = 𝜋𝜋𝑎𝑎2 𝑝𝑝𝑎𝑎𝑎𝑎
rigid
2𝑎𝑎
ℎ soil layer
rigid bedrock
The configuration being considered is shown in Figure 4a and the solution is presented
in terms of a settlement factor, 𝐼𝐼𝑝𝑝 . The settlement, 𝑆𝑆, is given by the expression:
𝑝𝑝𝑎𝑎𝑎𝑎 𝑎𝑎
𝑆𝑆 = 𝐼𝐼 (4a)
𝐸𝐸 𝑝𝑝
where:
128
𝑎𝑎 = radius of the loaded area
𝐸𝐸 = soil modulus
𝐼𝐼𝑝𝑝 = settlement factor read from Figure 4b (p. 45 of the Data Sheets).
Ip = Note that Ip depends on the value of Poisson’s ratio ν.
Figure 4b: Settlement factors for a rigid circular footing on a soil layer
129
Example 1
𝑎𝑎 1.5
= = 0.167
ℎ 9
500
𝑝𝑝𝑎𝑎𝑎𝑎 = = 70.7 kPa
𝜋𝜋(1.5)2
𝑃𝑃
𝑝𝑝𝑎𝑎𝑎𝑎 =
𝑏𝑏 2
(4b)
𝑏𝑏
𝑎𝑎 =
√𝜋𝜋
Soils often have a modulus that increases with depth. The soil does not necessarily
change its nature with depth, the reason for the increase in modulus is that the mean
effective stress increases with depth and so, because the modulus increases with the
mean effective stress, the modulus varies with depth. Often the variation with depth is
approximately linear and so can be approximated by the relation:
The modulus increases linearly from 𝐸𝐸0 at the surface as shown schematically in Figure
5.
130
𝑃𝑃 = 𝜋𝜋𝑎𝑎2 𝑝𝑝
2𝑎𝑎
soil layer
𝐸𝐸0 + 𝑚𝑚𝑚𝑚
A charted solution is available for this modulus variation for the case of a flexible
circular footing (𝑝𝑝 constant) resting on an infinitely deep soil layer. The settlement may
be expressed in the form:
𝑝𝑝𝑝𝑝
𝑆𝑆 = 𝐼𝐼 (6)
𝐸𝐸0 𝑝𝑝
where:
131
Figure 6: Influence chart for flexible circular load on non-homogeneous soil
132
Example 2
An oil tank applies a uniform stress of 75 kPa over a circular area with diameter 20 m.
Calculate the immediate settlement if the undrained modulus increases linearly from 2
MPa at the surface to 5 MPa at a depth of 10 m.
𝐸𝐸 = 𝐸𝐸0 + 𝑚𝑚𝑚𝑚
5 = 2 + 10𝑚𝑚
𝑚𝑚 = 0.3 MPa/m
𝐸𝐸0
𝜂𝜂 =
𝑚𝑚𝑚𝑚
2
= = 0.67
0.3 × 10
Now,
Thus,
𝑝𝑝𝑝𝑝
𝑆𝑆 = 𝐼𝐼
𝐸𝐸0 𝑝𝑝
75 × 10 × 0.6
= = 0.225 m
2000
13.5 Settlement under the Edge of a Flexible Strip Load on a Finite Soil Layer
The configuration is shown in Figure 7a. The settlement at the edge takes the form:
𝑝𝑝ℎ
𝑆𝑆 = 𝐼𝐼 (7)
𝜋𝜋𝜋𝜋 𝑝𝑝
where:
For a rigid strip footing, the settlement can be estimated by averaging the centre and
edge settlements of an equivalent flexible footing.
133
𝑝𝑝
𝐵𝐵
ℎ soil layer
rigid bedrock
134
Figure 7b: Settlement factor for edge of flexible strip on a soil layer
135
Example 3
8m 10 m 18 m
embankment (+)
(−)
2m
15 m 15 m
2m
The embankment loading consists of a strip loading of intensity +50 kPa and width of
18 m for which:
ℎ 15
= = 0.83
𝐵𝐵 18
and a strip loading of intensity -50 kPa and width of 2 m for which:
𝐵𝐵 2
= = 0.13
ℎ 15
𝑝𝑝ℎ +50 × 15
𝑆𝑆1 = 𝐼𝐼𝑝𝑝 = × 1.10
𝜋𝜋𝜋𝜋 𝜋𝜋 × 9000
𝑝𝑝ℎ −50 × 15
𝑆𝑆2 = 𝐼𝐼𝑝𝑝 = × 0.58
𝜋𝜋𝜋𝜋 𝜋𝜋 × 9000
136
−50 × 15
𝑆𝑆 = 𝑆𝑆1 + 𝑆𝑆2 = × (1.10 − 0.58) = 0.0138 m
𝜋𝜋 × 9000
If a footing is embedded, the settlement will be reduced. Two cases are shown in Figure
9a for which some solutions are available, both for a very deep elastic layer. The
settlement reduction factors are given in Figure 9b (p. 48, Data Sheets). To use these
solutions, the settlement must be found using the previously derived solutions for the
load resting on the surface.
𝑧𝑧 𝑧𝑧
𝐷𝐷 𝐷𝐷
Figure 9a: Loads applied below the surface in a deep elastic layer
137
13.7 Selection of Elastic Parameters
This component is due to deformations in the soil immediately after loading. As has
been discussed previously, immediately after a load is applied, water has no time to
drain out of the voids and so there is no volume change. Hence, any deformation must
occur at constant volume.
To estimate the initial settlement, 𝑆𝑆𝑖𝑖 , due to the constant volume deformation the
undrained (total stress) parameters, 𝐸𝐸𝑢𝑢 and 𝜈𝜈𝑢𝑢 = 1/2, are used in the analyses described
above.
As observed earlier when the load is applied over a very large area, the situation
approaches one-dimensional conditions, for which the initial undrained settlement is
zero.
In principle effective stress parameters could be used to determine the settlement but,
because the excess pore pressures generated by the load vary throughout the soil, the
analysis is not straightforward and the simple elastic formulae cannot be used.
This is due to deformations arising from volume changes which occur as a consequence
of the excess pore water pressures (which have been generated immediately after
loading) dissipating and allowing the effective stresses to come into equilibrium with
the applied loads. Finally, all excess pore water pressures will have dissipated and the
final settlement, 𝑆𝑆𝑡𝑡𝑡𝑡 , can be determined by using 𝐸𝐸 ′ and 𝜈𝜈 ′ in the settlement formulae
developed previously.
The settlement due to consolidation, 𝑆𝑆𝑐𝑐 , can be determined indirectly from the final
settlement 𝑆𝑆𝑡𝑡𝑡𝑡 ,, and the immediate settlement, 𝑆𝑆𝑖𝑖 , by:
Settlements due to creep cannot be predicted using the simple elastic formulae and are
usually only significant for soft soil sites.
138
13.8 Calculation of the Settlement at any Time
For relatively impermeable clayey soils in the short term, undrained deformations
occur. It is normally assumed that construction occurs sufficiently quickly so that no
drainage occurs and the settlement at the end of construction is then the immediate
settlement, 𝑆𝑆𝑖𝑖 . For sandy soils, the total final settlement is reached in the short term and
there is no time dependent response; thus it is assumed that consolidation is
instantaneous. Note that there will be soils that have intermediate properties and the
initial settlement will be partly drained. The extent of the drainage (consolidation) will
depend on the boundary conditions and the coefficient of consolidation.
For clayey soils, the time settlement behaviour can be visualised as shown in Figure 10.
construction
time
load
time
settlement
consolidation
settlement, 𝑆𝑆𝑐𝑐
total final
settlement, 𝑆𝑆𝑡𝑡𝑡𝑡
initial
settlement, 𝑆𝑆𝑖𝑖
time
The settlement at any time 𝑡𝑡 can then be calculated from the three components described
above and it is found that:
𝑆𝑆𝑡𝑡 − 𝑆𝑆𝑖𝑖
𝑈𝑈 = (9b)
𝑆𝑆𝑡𝑡𝑡𝑡 − 𝑆𝑆𝑖𝑖
139
Clearly,
𝑈𝑈 = 0 when 𝑡𝑡 = 0
and,
𝑈𝑈 = 1 when 𝑡𝑡 = ∞
Solutions for 𝑈𝑈 versus 𝑇𝑇 for a variety of boundary conditions are given on pp. 50–58 of
the Data Sheets. In general, these charts use the non-dimensionalised time factor, 𝑇𝑇,
given by 𝑐𝑐𝑣𝑣 𝑡𝑡/ℎ2 , where ℎ is the thickness of the soil layer irrespective of the boundary
conditions. (Note that this is different from the definition used for 1-D consolidation).
Example 4
Determine the immediate settlement, the final settlement and the settlement 1 year after
the end of construction of a rigid circular footing 5 m in diameter which supports a load
of 1.5 MN and is founded on a 5 m thick clay layer overlying gravel. The clay layer has
the following uniform properties: 𝐸𝐸 ′ = 5 MPa, 𝜈𝜈 ′ = 0.2, 𝑐𝑐𝑣𝑣 = 0.5 m2 /yr and 𝐸𝐸𝑢𝑢 =
6.25 MPa.
𝑎𝑎 2.5
= = 0.5
ℎ 5
Thus,
𝐼𝐼𝑝𝑝 = 0.63
140
with:
1500
𝑝𝑝𝑎𝑎𝑎𝑎 = = 76.39 kPa
𝜋𝜋 × 2.52
Thus,
Thus,
𝐼𝐼𝑝𝑝 = 0.95
141
3. Calculation of settlement after 1 year
Thus, for:
𝑐𝑐𝑣𝑣 𝑡𝑡 0.5 × 1
𝑇𝑇 = = = 0.02
ℎ2 52
ℎ
=2
𝑎𝑎
142
(b) For the case of a Permeable Footing (PTPB)
The degree of consolidation can be determined from Figure 12 and it is found that 𝑈𝑈 =
0.5 and so the settlement after 1 year is:
143
14 SOIL STRENGTH
14.1 Introduction
Soils are essentially frictional materials. They are comprised of individual particles that
can slide and roll relative to one another. In the discipline of soil mechanics, it is
generally assumed that the particles are not cemented.
One consequence of the frictional nature is that the strength depends on the effective
stresses in the soil. As the effective stresses increase with depth, so will the strength, in
general.
The strength will also depend on whether the soil deformation occurs under fully
drained conditions, constant volume (undrained) conditions or with some intermediate
state of drainage. In each case, different excess pore pressures will occur, resulting in
different effective stresses and hence different strengths. In assessing the stability of
soil constructions, analyses are usually performed to check the short term (undrained)
and long term (fully drained) conditions.
The limiting shear stress that may be applied to any plane in the soil mass is found to
be given by an equation of the form:
where:
𝑐𝑐 = cohesion (apparent)
𝜙𝜙 = friction angle
This is known as the Mohr-Coulomb failure criterion. The parameters 𝑐𝑐 and 𝜙𝜙 are not
generally soil constants. The Mohr-Coulomb criterion is an empirical criterion and the
failure locus is only locally linear. Extrapolation outside the range of normal stresses
for which it has been determined is likely to be unreliable.
144
• the use of total or effective stresses.
𝑐𝑐𝑢𝑢 and 𝜙𝜙𝑢𝑢 are referred to as the undrained (total) strength parameters. These
parameters can only be determined from undrained tests.
The undrained strength parameters are not soil constants, they depend strongly
on the moisture content of the soil. The total stress criterion has limited
applicability as it is only valid if soil deformation occurs without drainage.
The undrained strengths measured in the laboratory are only relevant in practice
to clayey (low permeability) soils that initially deform without drainage and that
have the same moisture content in-situ.
The engineering strength of soil materials is often determined from tests in either the
shear box apparatus or the triaxial apparatus.
The soil is sheared along a predetermined plane by placing it in a box and then moving
the top half of the box relative to the bottom half. The box may be square or circular in
plan and of any size; however, the most common shear boxes are square, 60 mm x 60
mm and test specimens are typically 20 mm thick. Larger boxes of 300 mm × 300 mm
are used to test soil with larger (gravel) sized particles. The shear box is constructed in
two separate halves, which may be held together by locating screws so that the box can
be filled with the soil to be tested.
145
top platen normal load
A load normal to the plane of shearing may be applied to a soil specimen through the
lid of the box. Provision is made for porous plates to be placed above and below the
soil specimen. These enable drainage to occur which is necessary if a specimen is to be
consolidated under a normal load and if a specimen is to be tested in a fully drained
state. The soil specimen may be submerged, by filling the containing vessel with water,
to prevent the specimens from drying out. Undrained tests may be carried out but, in
this case, solid spacer blocks rather than the porous disks must be used.
Notation:
𝑁𝑁 = normal force
𝐹𝐹 = tangential (shear) force
Usually only relatively slow drained tests are performed in shear box apparatus. For
clays, the rate of shearing must be chosen to prevent excess pore pressures building up.
For freely draining sands and gravels, tests can be performed quickly. Tests on sands
and gravels are usually performed dry as it is found that water does not significantly
affect the (drained) strength.
Provided there are no excess pore pressures, the pore pressure in the soil will be
approximately zero and the total and effective stresses will be identical. That is, 𝜎𝜎𝑛𝑛 =
𝜎𝜎 ′𝑛𝑛 .
The failure stresses thus define an effective stress failure envelope from which the
effective (drained) strength parameters, 𝑐𝑐 ´ and 𝜙𝜙 ′ , can be determined.
146
Typical Test Results:
shear load
(𝐹𝐹)
shear stress
(𝜏𝜏)
At this stage, we are primarily interested in the stresses at failure. It is observed that for
a set of initially similar soil samples, there is a linear failure criterion that may be
expressed as:
𝜏𝜏 = 𝑐𝑐 ′ + 𝜎𝜎𝑛𝑛′ tan 𝜙𝜙 ′
From this the effective (drained) strength parameters, 𝑐𝑐 ′ and 𝜙𝜙 ′ , can be determined.
A peak and an ultimate failure locus can be obtained from the results each with different
𝑐𝑐 ′ and 𝜙𝜙 ′ values. All soils are essentially frictional materials and continued shearing
results in them approaching a purely frictional state where 𝑐𝑐 ′ ≈ 0. Normally
consolidated clays (OCR = 1) and loose sands do not usually show peak strengths and
have 𝑐𝑐 ′ = 0, whereas, over-consolidated clays and dense sands have 𝑐𝑐 ′ > 0. Note that
dense sands (OC clays) do not possess any true cohesion (bonds) and the apparent
cohesion results from the tendency of soil to expand when sheared.
147
As a soil test the shear box is far from ideal. Disadvantages of the test include:
In practice, the shear box is used to get quick and crude estimates of the failure
parameters. It is sometimes used to obtain undrained strengths but this use should be
discouraged.
The triaxial test is carried out in a cell and is so named because three principal stresses
are applied to the soil sample. Two of the principal stresses are applied to the sample
by a water pressure inside the confining cell and are equal. The third principal stress is
applied by a loading ram through the top of the cell and therefore may be different to
the other two principal stresses. A diagram of a typical triaxial cell is shown below.
148
A cylindrical soil specimen as shown is placed inside a latex rubber sheath which is sealed
to a top cap and bottom pedestal by rubber O-rings. For drained tests, or undrained tests
with pore pressure measurement, porous disks are placed at the bottom and sometimes at
the top of the specimen. For tests where consolidation of the specimen is to be carried out,
filter paper drains may be provided around the outside of the specimen in order to speed
up the consolidation process.
Pore pressure generated inside the specimen during testing may be measured by means of
pressure transducers. These transducers must operate with a very small volume change,
since fluid flowing out of the specimen would cause the pore water pressure that was being
measured to drop.
14.3.2.1 Stresses
𝐹𝐹 = deviator load
𝜎𝜎𝑟𝑟
The term 𝐹𝐹/𝐴𝐴 is known as the deviator stress and is usually given the symbol, 𝑞𝑞.
Hence we can write 𝑞𝑞 = 𝜎𝜎𝑎𝑎 − 𝜎𝜎𝑟𝑟 = 𝜎𝜎1 − 𝜎𝜎3 (the axial and radial stresses are principal
stresses).
• volumetric compression if the soil is free to drain. The effective stresses will
increase and so will the strength.
149
• increasing pore water pressure if soil volume is constant (that is, undrained).
As the effective stresses cannot change it follows that Δ𝑢𝑢 = Δ𝜎𝜎𝑟𝑟 .
14.3.2.2 Strains
From the measurements of change in height, 𝑑𝑑ℎ, and change in volume, 𝑑𝑑𝑑𝑑, we can
determine:
where ℎ0 is the initial height and 𝑉𝑉0 is the initial volume. The conventional small strain
assumption is generally used.
It is assumed that the sample deforms as a right circular cylinder. The cross-sectional area,
𝐴𝐴, can then be determined from:
𝑑𝑑𝑑𝑑
1 + 𝑉𝑉 1 − 𝜀𝜀𝑣𝑣
0
𝐴𝐴 = 𝐴𝐴0 � � = 𝐴𝐴0 � �
𝑑𝑑ℎ 1 − 𝜀𝜀𝑎𝑎
1+
ℎ0
It is important to make allowance for the changing area when calculating the deviator
stress,
There are many test variations. Those used most in practice are
Cell pressure applied without allowing drainage. Then, keeping cell pressure
constant, increase deviator load to failure without drainage.
150
As a test for investigating the behaviour of soils, the triaxial test has many advantages over
the shear box test:
Typical results from a series of drained tests consolidated to different cell pressures would
be as follows.
q
increasing
cell pressure
𝜀𝜀𝑎𝑎
The triaxial test gives the strength in terms of the principal stresses, whereas the shear box
gives the stresses on the failure plane directly. To relate the strengths from the two tests,
we need to use some results from the Mohr circle transformation of stress.
𝜏𝜏
𝜎𝜎3 𝜎𝜎1 𝜎𝜎
151
14.4 Mohr Circles
The Mohr circle construction enables the stresses acting in different directions at a point
on a plane to be determined, provided that the stress acting normal to the plane is a
principal stress. The Mohr circle construction is very useful in Soil Mechanics as many
practical situations can be approximated as plane strain problems.
The sign convention is different to that used in Structural Analysis because, for Soil
Mechanics, it is conventional to take the compressive stresses as positive.
Sign convention:
Let us consider the stresses acting on different planes for an element of soil:
a) 𝜎𝜎1 b)
1sin 𝛼𝛼
𝛼𝛼 𝜏𝜏𝛼𝛼
𝜎𝜎3 1cos 𝛼𝛼
1
𝛼𝛼 𝜎𝜎𝛼𝛼 𝛼𝛼
Figure 7: An element of soil showing (a) the stresses on a plane at angle 𝜶𝜶 to the
minor principal stress and (b) the relevant lengths.
𝜎𝜎1 𝜎𝜎3
𝜎𝜎𝛼𝛼 = (1 − cos 2𝛼𝛼) + (1 + cos 2𝛼𝛼)
2 2
𝜎𝜎1 + 𝜎𝜎3 𝜎𝜎1 − 𝜎𝜎3
𝜎𝜎𝛼𝛼 = − cos 2𝛼𝛼
2 2
and, similarly,
(𝜎𝜎1 − 𝜎𝜎3 )
𝜏𝜏𝛼𝛼 = sin 2𝛼𝛼
2
152
τ
(𝜎𝜎𝛼𝛼 , 𝜏𝜏𝛼𝛼 )
𝑅𝑅
𝜙𝜙 2𝛼𝛼 𝜎𝜎
𝜎𝜎3 𝜎𝜎1
𝑝𝑝
where:
and, failure occurs on a plane at an angle 𝛼𝛼 from the plane on which 𝜎𝜎3 acts, so:
𝜋𝜋 𝜙𝜙
𝛼𝛼 = � − �
4 2
Failure will occur when we can find any direction such that:
153
τ
𝑅𝑅
𝜙𝜙 𝑐𝑐 𝜎𝜎
𝜎𝜎3 𝜎𝜎1
𝑐𝑐 cot 𝜙𝜙 𝑝𝑝
As mentioned above it is the effective strength parameters, 𝑐𝑐 ′ and 𝜙𝜙 ′ , that are the
fundamental soil strength parameters. To use these parameters, the Mohr-Coulomb
criterion must be expressed in terms of effective stresses, that is:
𝜏𝜏 = 𝑐𝑐 ′ + 𝜎𝜎𝑛𝑛′ tan 𝜙𝜙 ′
with:
1 + sin ϕ′
𝑁𝑁𝜙𝜙 =
1 − sin ϕ′
𝜎𝜎𝑛𝑛′ = 𝜎𝜎𝑛𝑛 − 𝑢𝑢
𝜎𝜎1′ = 𝜎𝜎1 − 𝑢𝑢
𝜎𝜎3′ = 𝜎𝜎3 − 𝑢𝑢
154
Note that the difference between the total and effective stresses is simply the pore pressure,
𝑢𝑢. Thus, the total and effective stress Mohr circles have the same diameter and are
displaced along the 𝜎𝜎 axis by the value of the pore pressure.
In drained laboratory tests, the loading rate is sufficiently slow so that all excess pore water
pressures will have dissipated. From the known pore water pressures the effective stresses
can be determined.
The behaviour of drained tests must be interpreted in terms of the effective strength
parameters, 𝑐𝑐 ′ and 𝜙𝜙 ′ , using the effective stresses. It is possible to construct a series of
total stress Mohr Circles but the inferred total strength parameters have no relevance to
the soil behaviour.
The effective strength parameters are generally used to check the long term (that is, when
all the excess pore pressures have dissipated) stability of soil constructions. However, for
sands and gravels, pore pressures dissipate rapidly, and, for these permeable soils, the
effective strength parameters can also be used for assessing the short term stability. In
principle, the effective strength parameters can be used to check the stability at any time
for any soil type but, to do this, the pore pressures in the ground must be known and in
general they are not.
The behaviour of undrained tests may be interpreted in terms of the effective strength
parameters, 𝑐𝑐 ′ and 𝜙𝜙 ′ , using the effective stresses. In a triaxial test with pore pressure
measurement, this is possible. The behaviour may also be interpreted in terms of the total
strength parameters, 𝑐𝑐𝑢𝑢 and 𝜙𝜙𝑢𝑢 . However, if the total stress parameters are being used,
they must be determined from Unconsolidated Undrained tests if they are to be relevant
to the soil in the ground.
Let us consider the behaviour of three identical saturated soil samples in undrained triaxial
tests. No water is allowed to drain and three different confining pressures are applied
(samples are unconsolidated). The Mohr circles at failure will be as follows:
155
𝜏𝜏
𝜎𝜎
𝜎𝜎3′ 𝜎𝜎1′ 𝜎𝜎3 𝜎𝜎1
Figure 10: Mohr circles for undrained failure at different confining pressures
Because all samples are at failure, the effective stress failure condition must also be
satisfied and, because all the circles have the same radius, there must be a single effective
stress Mohr circle. The different total stress Mohr circles indicate that the samples must
have different pore water pressures.
The explanation for the independence of the undrained strength on the confining stress is
that increasing the cell pressure without allowing drainage has the effect of increasing
the pore pressure by the same amount (Δ𝑢𝑢 = Δ𝜎𝜎𝑟𝑟 ). There is therefore no change in
effective stress. As it is the effective stresses that control the soil behaviour, the
subsequent strength is unaffected. The change in pore pressure during shearing is a
function of the initial effective stress and the moisture content. As these are identical
for the three samples, an identical strength is obtained. As will be shown later, the fact
that the moisture content remains constant is the most important factor in having a
constant strength.
In some series of unconsolidated undrained tests, it is found that for different soil
samples from a particular site 𝜙𝜙𝑢𝑢 is not zero, or 𝑐𝑐𝑢𝑢 is not constant. If this occurs, then
either:
The total stress strength parameters, 𝑐𝑐𝑢𝑢 and 𝜙𝜙𝑢𝑢 , are often used to assess the short term
(undrained) stability of soil constructions. It is important that no drainage should occur
otherwise this approach is not valid. Therefore, for sands and gravels which drain rapidly
a total stress analysis would not be appropriate.
For soils that do not drain freely, this approach is the only simple way of assessing the
short term stability because in general the pore water pressures are unknown.
156
Note, however, that it is possible to measure an undrained strength for any type of soil in
the triaxial apparatus.
Example 1
Analytical Solution
Hence,
and,
Graphical Solution
𝜏𝜏
26°
17.5
𝜎𝜎
157
15 STRESS-STRAIN BEHAVIOUR OF SOILS
In practice, sands are usually sheared under drained conditions because their relatively
high permeability ensures that excess pore pressures are not generated. This behaviour can
be investigated in a variety of laboratory apparatus. We will consider the behaviour in
simple shear tests. The simple shear test is similar to the shear box test but it has the
advantage that the strain and stress states are more uniform, enabling us to investigate the
stress-strain behaviour. The name simple shear refers to the plane strain mode of
deformation shown below:
𝜎𝜎
𝑑𝑑𝑑𝑑
𝜏𝜏
𝑑𝑑𝑑𝑑
𝐻𝐻
𝛾𝛾𝑥𝑥𝑥𝑥
𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
𝛾𝛾𝑥𝑥𝑥𝑥 = 𝜀𝜀𝑧𝑧 = − = 𝜀𝜀𝑣𝑣
𝐻𝐻 𝐻𝐻
For this deformation, there are only two non-zero strain components; these are the shear
strain, 𝛾𝛾𝑥𝑥𝑥𝑥 = 𝑑𝑑𝑑𝑑/𝐻𝐻 and the normal strain, 𝜀𝜀𝑧𝑧 = 𝑑𝑑𝑑𝑑/𝐻𝐻. The volume strain, 𝜀𝜀𝑣𝑣 = 𝜀𝜀𝑧𝑧 .
For sands, the two most important parameters governing their behaviour are the relative
density, 𝐼𝐼𝑑𝑑 , and the effective stress level, 𝜎𝜎 ′ . The relative density is defined by:
𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑒𝑒
𝐼𝐼𝑑𝑑 =
𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚
where 𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 and 𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 are the maximum and minimum void ratios that can be measured in
standard tests in the laboratory and 𝑒𝑒 is the current void ratio. This expression can be re-
written in terms of dry density as:
𝐺𝐺𝑠𝑠 𝛾𝛾𝑤𝑤
𝛾𝛾𝑑𝑑 = (1a)
1 + 𝑒𝑒
and hence,
1 1
� �
𝛾𝛾𝑑𝑑,𝑚𝑚𝑚𝑚𝑚𝑚 − 𝛾𝛾𝑑𝑑
𝐼𝐼𝑑𝑑 = (1b)
1 1
� �
𝛾𝛾𝑑𝑑,𝑚𝑚𝑚𝑚𝑚𝑚 − 𝛾𝛾𝑑𝑑,𝑚𝑚𝑚𝑚𝑚𝑚
158
Sand is generally referred to as dense if 𝐼𝐼𝑑𝑑 > 0.6 and loose if 𝐼𝐼𝑑𝑑 < 0.3.
The influence of relative density on the behaviour can be seen in the plots below for tests
all performed at the same normal stress.
𝜏𝜏 dense (𝐷𝐷)
medium (𝑀𝑀)
loose (𝐿𝐿)
𝛾𝛾
𝑒𝑒 loose (𝐿𝐿)
medium (𝑀𝑀)
dense (𝐷𝐷)
𝛾𝛾
medium (𝑀𝑀)
𝛾𝛾
loose (𝐿𝐿)
159
The following observations can be made:
• All samples approach the same ultimate conditions of shear stress and void ratio,
irrespective of the initial density.
• Initially dense samples attain higher peak angles of friction (𝜙𝜙 ′ = tan−1(𝜏𝜏/
𝜎𝜎 ′ )).
• Initially dense soils expand (dilate) when sheared and initially loose soils
compress.
The influence of stress level can be seen in the plots below where the two dense samples
have the same initial void ratio, 𝑒𝑒1 , and similarly the loose samples both have the same
initial void ratio, 𝑒𝑒2 .
𝜏𝜏 𝐷𝐷2 𝑒𝑒 𝐿𝐿2
𝐿𝐿1
𝐷𝐷1
𝐷𝐷1
𝐿𝐿2 𝐿𝐿1
𝐷𝐷2
𝛾𝛾 𝛾𝛾
𝜀𝜀𝑣𝑣 𝜏𝜏 𝐷𝐷1
𝐷𝐷1 𝜎𝜎 ′ 𝐷𝐷2
𝐷𝐷2
𝛾𝛾 𝐿𝐿2
𝐿𝐿1 𝐿𝐿1
𝐿𝐿2
𝛾𝛾
160
The following observations can be made:
• The ultimate values of shear stress and void ratio depend on the stress level, but
′
the ultimate angle of friction (𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢 = tan−1(𝜏𝜏/𝜎𝜎 ′ )𝑢𝑢𝑢𝑢𝑢𝑢 ) is independent of both
density and stress level.
• Initially dense samples attain higher peak angles of friction (𝜙𝜙 ′ = tan−1(𝜏𝜏/
𝜎𝜎 ′ )), but the peak friction angle reduces as the stress level increases.
• Initially dense soils expand (dilate) when sheared and initially loose soils
compress. Increasing stress level causes less dilation (greater compression).
All soil, when sheared, will eventually attain a unique stress ratio given by 𝜏𝜏/𝜎𝜎 ′ =
′
tan 𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢 and reach a critical void ratio which is uniquely related to the normal stress. This
ultimate state is referred to as a Critical State, defined by:
The locus of these critical states defines a line known as the Critical State Line (CSL).
This may be represented by:
′
𝜏𝜏 = 𝜎𝜎 ′ tan 𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢
𝜏𝜏 CSL
𝜏𝜏
CSL
𝜎𝜎 ′
′
𝜏𝜏 = 𝜎𝜎 ′ tan 𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢
𝑒𝑒 = 𝑒𝑒0 − 𝜆𝜆 ln 𝜎𝜎 ′
′
𝜎𝜎
𝑒𝑒
𝑒𝑒 𝑒𝑒
CSL
CSL
𝑒𝑒 = 𝑒𝑒0 − 𝜆𝜆 ln 𝜎𝜎 ′ 𝑒𝑒 = 𝑒𝑒0 − 𝜆𝜆 ln 𝜎𝜎 ′
𝜎𝜎 ′ ln 𝜎𝜎 ′
161
At critical states, soil behaves as a purely frictional material:
′
𝜙𝜙 ′ = 𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢 ′
= 𝜙𝜙𝑐𝑐𝑐𝑐 = constant = 𝑓𝑓(mineralogy, grading, angularity)
During a simple shear test on dense sand, the top platen is forced up against the applied
normal stress. Work must be done against this external force, in addition to the work done
in overcoming friction between the particles. Thus, the frictional resistance of the soil may
′
appear to be greater than 𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢 . Another way to demonstrate this is to consider a “saw-
tooth” analogy.
𝑃𝑃
𝑄𝑄
𝛼𝛼 𝐹𝐹
𝑁𝑁
𝑄𝑄 = 𝐹𝐹 cos 𝛼𝛼 + 𝑁𝑁 sin 𝛼𝛼
𝑄𝑄 𝐹𝐹 cos 𝛼𝛼 + 𝑁𝑁 sin 𝛼𝛼
=
𝑃𝑃 −𝐹𝐹 sin 𝛼𝛼 + 𝑁𝑁 cos 𝛼𝛼
𝐹𝐹
𝑄𝑄 �𝑁𝑁� + tan 𝛼𝛼
=
𝑃𝑃 1 − � 𝐹𝐹 � tan 𝛼𝛼
𝑁𝑁
Now,
𝑄𝑄 𝐹𝐹 ′
= tan 𝜙𝜙 ′ ; = tan 𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢
𝑃𝑃 𝑁𝑁
′
tan 𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢 + tan 𝛼𝛼
tan 𝜙𝜙 ′ = ′
1 − tan 𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢 tan 𝛼𝛼
′
𝜙𝜙 ′ = 𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢 + 𝛼𝛼 (3)
𝑑𝑑𝑑𝑑 𝑑𝑑𝜀𝜀𝑣𝑣
tan 𝛼𝛼 = ≈
𝑑𝑑𝑑𝑑 𝑑𝑑𝑑𝑑
162
15.1.5 Peak Conditions
𝜏𝜏 peak
ultimate
𝜙𝜙 ′
𝜎𝜎 ′
However, this approach obscures the fact that 𝑐𝑐 ′ is only an apparent cohesion. An
alternative method of presenting the results is to determine the maximum friction angle
′
𝜙𝜙𝑝𝑝𝑝𝑝 which in shear box type tests is simply given by tan−1(𝜏𝜏/𝜎𝜎 ′ ). The relation between
′
𝜙𝜙𝑝𝑝𝑝𝑝 and effective stress is then as shown below.
′
𝜙𝜙𝑝𝑝𝑝𝑝
′
𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢
𝜎𝜎 ′
The position of the lines in this plot is a function of the mineralogy and angularity of the
soil.
′ ′
Note that even loose sand can have 𝜙𝜙𝑝𝑝𝑝𝑝 > 𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢 if the stress is low enough. This means
that loose sands may expand when sheared.
163
15.1.6 Implications for Stability Analysis
′
If you choose to use 𝜙𝜙𝑝𝑝𝑝𝑝 (or 𝑐𝑐 ′ , 𝜙𝜙 ′ with 𝑐𝑐 ′ ≠ 0) in stability calculations, then you are
saying that, everywhere on the critical failure surface, the soil will be dilating at failure. In
most practical cases, this is unlikely to be realistic. For instance, consider the case of a
retaining wall.
𝜏𝜏
wall failure C
surface B
A
𝛾𝛾
𝜏𝜏
𝜏𝜏 B
A C
A
1B C
𝛾𝛾
𝛾𝛾
The behaviour of clays is essentially identical to that of sands. The data however is usually
presented in terms of the soils stress history (OCR) rather than relative density.
To predict the behaviour of soil, we need to combine the CSL with our previous
knowledge concerning the consolidation behaviour. Experience has shown that the CSL
is parallel to the normal consolidation line and lies below it in a void ratio–effective stress
plot.
164
𝑒𝑒
Normal Consolidation
Line (NCL)
swelling
line
CSL
ln 𝜎𝜎 ′
We find that normally consolidated clays behave similarly to loose sands and heavily over-
consolidated clays behave similarly to dense sands. As the OCR increases, there is a
gradual trend between these extremes. The response in drained simple shear tests with 𝜎𝜎 ′
constant is as follows:
𝜏𝜏 CSL 𝜏𝜏
OCR = 1
OCR = 8
𝜎𝜎 ′ 𝛾𝛾
𝑒𝑒 𝜀𝜀𝑣𝑣
OCR = 8
NCL 𝛾𝛾
CSL OCR = 1
𝜎𝜎 ′
165
15.2.1 Undrained Response
In an undrained test, volume change is prevented and therefore the void ratio must remain
constant. Because the soil always heads towards a critical state when sheared, it is possible
to show the path that will be followed in an 𝑒𝑒– 𝜎𝜎 ′ plot. This is shown below for normally
consolidated (OCR = 1) and heavily over-consolidated (OCR > 8) samples having the
same initial void ratio. Once the final states in this plot are known, so too are the final
states in the 𝜏𝜏– 𝜎𝜎 ′ plot. Also, if the final total stresses are known, then the excess pore
pressures can be determined.
𝜏𝜏 CSL 𝜏𝜏
OCR = 1
OCR = 8
𝜎𝜎 ′ 𝛾𝛾
𝑒𝑒 𝑢𝑢 (+ve)
OCR = 1
NCL 𝛾𝛾
CSL OCR = 8
(−ve)
𝜎𝜎 ′
• Knowledge of the Critical State Line enables an explanation for the existence of
apparent cohesion (undrained strength) in frictional materials.
• It is also clear that, if the moisture content changes, then so will the undrained
strength, because failure will occur at a different point on the CSL.
166
15.3 Differences between Sand and Clay
When considering the behaviour of sands and clays, we generally use different parameters.
For sands, stress level and relative density are considered to be the important parameters,
whereas, for clays, the parameters are stress level and stress history (OCR).
However, the broad patterns of behaviour observed for sands and clays are very similar.
To understand why different “engineering” parameters are used, it is useful to consider the
positions of the consolidation and CSL lines in the void ratio–effective stress plot.
𝒆𝒆
clay
loose
dense sand
NCL NCL
𝐥𝐥𝐥𝐥𝐥𝐥 𝝈𝝈 (𝐌𝐌𝐌𝐌𝐌𝐌)
0.1 1 10 100
Figure 12: Comparison of typical void ratio–effective stress behaviour for a sand
and a clay
167
16 EARTH PRESSURES (RANKINE’S METHOD)
𝐹𝐹
Some force is required to support the soil. This force may be provided by
If the force is too small, the soil behind the wall will reach a state of failure with the wall
moving away from the soil (active failure). If the force is too large, the soil will reach
another state of failure with the wall moving into the soil (passive failure).
In Rankine’s method, it is assumed that the wall is frictionless. The normal stress acting
on the wall will therefore be a principal stress. If the wall is vertical and the soil surface
horizontal, the vertical and horizontal stresses throughout the retained soil mass will be
principal stresses. In this situation, the vertical stress at any depth can be simply
determined, as follows:
𝛾𝛾1 𝑑𝑑1
168
The horizontal stress can then be calculated from the Mohr-Coulomb failure criterion. If
short term stability is being considered, this can be achieved using undrained (total stress)
parameters while, if long term stability, is being considered, drained (effective stress)
parameters must be used.
The implications of this expression are most easily investigated by considering the
response of soil adjacent to a frictionless retaining wall. Then, we can identify two
limiting conditions:
There is insufficient force to support the soil. Assuming that the vertical stress is given
simply by the weight of the overlying soil and does not change during deformation, the
minimum horizontal stress may be determined from:
The force on the wall is greater than the resistance provided by the soil. The horizontal
stress reaches a maximum value given by:
In the Rankine method, a stress state is found that is in equilibrium with the applied
loads and has the soil at failure. In plasticity theory, this approach is referred to as a
lower bound method—a method which can be shown to produce safe, conservative
solutions.
The relation between active and passive states can be seen by considering the Mohr
circles as shown below.
169
𝝉𝝉 𝜏𝜏 = 𝑐𝑐 + 𝜎𝜎𝑛𝑛 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡
Figure 3: Mohr circles showing the relation between active and passive states
For a given 𝜎𝜎𝑣𝑣 , it is impossible for the horizontal stresses to drop below 𝜎𝜎ℎ,𝑚𝑚𝑚𝑚𝑚𝑚 or rise
above 𝜎𝜎ℎ,𝑚𝑚𝑚𝑚𝑚𝑚 .
A total stress analysis is only appropriate if the soil remains undrained. In practice, this
implies that total stress analysis can only be used to investigate the stability of clayey
soils with low permeabilities.
Use undrained parameters, 𝑐𝑐𝑢𝑢 and 𝜙𝜙𝑢𝑢 , and total stresses, 𝜎𝜎1 , 𝜎𝜎3 , 𝜎𝜎𝑣𝑣 , 𝜎𝜎ℎ with:
1 + sin 𝜙𝜙𝑢𝑢
𝑁𝑁𝜙𝜙 = ; 𝑐𝑐 = 𝑐𝑐𝑢𝑢 (3)
1 − sin 𝜙𝜙𝑢𝑢
2𝑐𝑐𝑢𝑢
−
�𝑁𝑁𝜙𝜙
𝑧𝑧
𝐻𝐻
170
• 𝜙𝜙𝑢𝑢 ≠ 0. This implies that the undrained strength, 𝑠𝑠𝑢𝑢 , increases with depth. It does
not imply that the soil is unsaturated; if this were true, an undrained analysis
would be inappropriate.
• Values of 𝑐𝑐𝑢𝑢 and 𝜙𝜙𝑢𝑢 can be measured for sands from undrained triaxial tests.
However, these are almost never relevant because of drainage
• Tension cracks. The analysis indicates negative, tensile stresses at the surface.
However, soil particles cannot provide tension. The negative stresses have to
come from suctions in the pore water. It is difficult to rely on the tensile forces
and they are usually ignored. The tensile stresses reduce the force required for
stability of the wall. Ignoring the tensile stresses therefore gives a more
conservative solution. The pressure distribution on the wall becomes:
𝑧𝑧0
2𝑐𝑐𝑢𝑢 �𝑁𝑁𝜙𝜙
𝑧𝑧0 = (4)
𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠
If water is available, it can fill up the tension crack and provide additional
pressures on the wall. In this situation, the pressure diagram becomes
𝑧𝑧0 water
𝛾𝛾𝑤𝑤 𝑧𝑧0
soil
171
• The position of the water table is only important in as far as it affects the total
stresses.
Use effective soil parameters, 𝑐𝑐 ′ and 𝜙𝜙 ′ , and effective stresses, 𝜎𝜎1′ , 𝜎𝜎3′ , 𝜎𝜎𝑣𝑣′ , 𝜎𝜎ℎ′ with:
1 + sin 𝜙𝜙 ′
𝑁𝑁𝜙𝜙 = ; 𝑐𝑐 = 𝑐𝑐 ′ (5)
1 − sin 𝜙𝜙 ′
and 𝜎𝜎 ′ = 𝜎𝜎 − 𝑢𝑢.
2𝑐𝑐 ′
−
�𝑁𝑁𝜙𝜙
𝑧𝑧
𝐻𝐻
𝑐𝑐 ′ , 𝜙𝜙 ′ , 𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑
• 𝑐𝑐 ′ and 𝜙𝜙 ′ are peak strength values. It is generally more appropriate and safer to
′
use the ultimate or critical state parameters, 𝑐𝑐 ′ = 0 and 𝜙𝜙 ′ = 𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢 .
• For passive failure, the critical state parameters give a smaller force on the wall
than the peak strength parameters. Again, this gives a safe, conservative estimate.
In dry sand, the limiting passive pressure is given by:
172
• It is important to remember to use effective stresses, 𝜎𝜎𝑣𝑣′ = 𝜎𝜎𝑣𝑣 − 𝑢𝑢, when
calculating the horizontal effective stresses, 𝜎𝜎ℎ′ . Then, to calculate the total
horizontal stress on the wall, the pore water pressure must be added to obtain
𝜎𝜎ℎ = 𝜎𝜎ℎ′ + 𝑢𝑢.
• If the water level is not the same on each side of the wall, water will flow. The
pore water pressures must then be determined from a flow net before calculating
𝜎𝜎𝑣𝑣′ .
Example 1
Layer 1:
A clay layer; so will be undrained in the short term, so it will require a total stress
(undrained) analysis:
1 + sin 𝜙𝜙𝑢𝑢
𝑐𝑐 = 𝑐𝑐𝑢𝑢 = 20 kPa ; 𝑁𝑁𝜙𝜙 = = 1.19
1 − sin 𝜙𝜙𝑢𝑢
173
This gives the pressure distribution on the wall shown in Figure 9a. The analysis
predicts tensile stresses between the soil and the wall. These are not likely, and a tension
crack may develop. Because the water table is at the surface, the crack will fill with
water, and a more pessimistic pressure distribution is shown in Figure 9b.
𝑧𝑧0 = 2.91 m
𝛾𝛾𝑤𝑤 × 2.91
26.4 26.4
Figure 9: Pressure distribution (a) assuming tensile stresses can develop and (b)
assuming that the tension cracks fill with water
Layer 2:
A sand layer, so excess pore pressures will dissipate rapidly. Therefore, a total stress
analysis cannot be used. For sand in the short term, assume that it is fully drained and
therefore we must use an effective stress analysis.
1 + sin 𝜙𝜙 ′
𝑐𝑐 = 𝑐𝑐 ′ = 0 kPa ; 𝑁𝑁𝜙𝜙 = = 3.69
1 − sin 𝜙𝜙 ′
Active failure; thus 𝜎𝜎1′ = 𝜎𝜎𝑣𝑣′ and 𝜎𝜎3′ = 𝜎𝜎ℎ′ . From the Mohr-Coulomb criterion,
Layer 3:
A clay layer; therefore total stress (undrained) analysis for short term.
1 + sin ϕu
𝑐𝑐 = 𝑐𝑐𝑢𝑢 = 50 kPa ; 𝑁𝑁𝜙𝜙 = =1
1 − sin ϕu
174
𝜎𝜎1 = 𝜎𝜎3 + 2𝑐𝑐𝑢𝑢
𝒛𝒛 𝝈𝝈𝒗𝒗 𝝈𝝈𝒉𝒉
8 135 35
10 165 65
2.91 m
28.5
2.09 m
26.4
56
3m
93.7
35
2m
65
The force required to prevent active failure can be determined from the pressure
diagram:
= 393.7 kN/m
175
Example 2
A 5 m high retaining wall retains a clayey soil, which overlies a highly permeable
sandstone. If the water level remains at the surface of the clay in the retained soil and is
level with the top of the sandstone, determine the minimum force required to maintain
the stability of the wall for short and long term. The soil parameters are:
′
𝑐𝑐𝑢𝑢 = 37 kPa ; 𝜙𝜙𝑢𝑢 = 5° ; 𝑐𝑐 ′ = 0 kPa ; 𝜙𝜙𝑢𝑢𝑢𝑢𝑢𝑢 = 25° ; 𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠 = 19 kN/m3
5 m clayey soil
sandstone
Figure 11: Analysis of a retaining wall in the short and long term
Short term:
Allowing for tension crack filling with water, pressures acting on the wall will be:
𝑧𝑧0 = 4.25 m
4.25 × 9.81
11.9
176
Long term:
Effective stress analysis; pore pressures required – have to be determined from a flow
net.
𝑋𝑋
5m
At 𝑋𝑋,
𝑧𝑧 = 2/3 × 5 = 10/3 m
𝑢𝑢 = 𝛾𝛾𝑤𝑤 (ℎ − 𝑧𝑧) = 0
Hence,
177
17 EARTH PRESSURES (COULOMB'S METHOD)
17.1 Introduction
soil
movement
wall
movement
assumed
failure plane
This approach, known as the limit equilibrium method, is widely used in geotechnical
engineering. Experience has shown that it gives solutions that agree reasonably well
with observations of collapse of real soil structures.
Its main disadvantage is that the common layered soil profile cannot be simply accounted
for.
𝜏𝜏 = 𝑐𝑐 + 𝜎𝜎𝑛𝑛 𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡
Whenever the soil is at failure, a Mohr-Coulomb locus of this general form can be used;
however, the appropriate values for c and φ will depend on the type of analysis. In a total
stress (undrained) analysis, 𝑐𝑐 = 𝑐𝑐𝑢𝑢 and 𝜙𝜙 = 𝜙𝜙𝑢𝑢 , whereas, in an effective stress analysis,
𝑐𝑐 = 𝑐𝑐 ′ and 𝜙𝜙 = 𝜙𝜙 ′ .
178
We next need to consider the forces acting on the failure plane:
Shear Force:
soil
movement 𝜏𝜏 𝑇𝑇 = ∫ 𝜏𝜏𝜏𝜏𝜏𝜏
Normal Force:
𝜎𝜎
𝑁𝑁 = ∫ 𝜎𝜎𝜎𝜎𝜎𝜎
assumed Cohesive Force:
failure plane
𝐶𝐶 = ∫ 𝑐𝑐𝑐𝑐𝑐𝑐
If the soil properties are constant, we can write the failure criterion in terms of forces as:
𝑇𝑇 = 𝐶𝐶 + 𝑁𝑁 tan 𝜙𝜙 (1)
To facilitate the analysis, we split the unknown forces, 𝑇𝑇 and 𝑁𝑁, into two forces, 𝐶𝐶, which
is generally known and acts parallel to the failure surface, and a resultant, 𝑅𝑅, acting at a
known angle, 𝜙𝜙, to the normal to the failure surface.
𝑇𝑇 𝐶𝐶
𝜙𝜙
𝑁𝑁
𝑅𝑅
𝑅𝑅 cos 𝜙𝜙 = 𝑁𝑁
(2)
𝑅𝑅 sin 𝜙𝜙 = 𝑇𝑇 − 𝐶𝐶 = 𝑁𝑁 tan 𝜙𝜙
Failure does not always occur within the soil mass. For the failure of the soil structure,
a mechanism is required and, for the case of a retaining wall, this means slip must also
occur between the wall and the soil. We assume that the failure conditions can be
described by a Mohr-Coulomb criterion, that is:
𝜏𝜏 = 𝑐𝑐 + 𝜎𝜎 tan 𝜙𝜙
179
This can also be expressed in terms of forces as:
As with Rankine's method, a total stress analysis is only appropriate if the soil remains
undrained and, in practice, this is only true if the stability of clayey soils is being
investigated.
Consider the undrained active failure of a wall of height, 𝐻𝐻, in a saturated clayey soil with
undrained parameters, 𝑐𝑐𝑢𝑢 and 𝜙𝜙𝑢𝑢 , and between the wall and the soil undrained parameters,
𝑐𝑐𝑤𝑤 and 𝜙𝜙𝑤𝑤 .
𝐻𝐻 tan 𝜃𝜃
soil 𝐶𝐶2
𝐻𝐻
𝑊𝑊
movement 𝐶𝐶1
where:
𝐶𝐶2 = 𝑐𝑐𝑤𝑤 𝐻𝐻
1 2
𝑊𝑊 = 𝐻𝐻 tan 𝜃𝜃 𝛾𝛾
2
180
Now we can draw the polygon of forces as the only unknowns are the magnitudes of 𝑅𝑅1
and 𝑅𝑅2 .
𝜙𝜙𝑤𝑤
𝑅𝑅2
𝑅𝑅1
𝑊𝑊 (90 − 𝜃𝜃 − 𝜙𝜙𝑢𝑢 )
𝐶𝐶2
𝜃𝜃 𝐶𝐶1
By eliminating 𝑅𝑅1 we can obtain an expression for the unknown force on the wall 𝑅𝑅2 in
terms of 𝜃𝜃.
In drawing a polygon of forces, the forces must be drawn so that the arrows indicating
their directions all point the same way as you move around the polygon. It is possible that
a polygon of forces cannot be constructed and, if this happens, it indicates that the assumed
failure mechanism is incorrect.
181
The force 𝑅𝑅2 determined from the polygon may not be the only force acting on the wall.
In the example above, the forces acting on the wall are:
𝜙𝜙𝑤𝑤 𝐻𝐻
𝑅𝑅2
𝑉𝑉
𝐶𝐶2 𝐹𝐹𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡 𝐹𝐹𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡
We are often interested in the horizontal force, that is, 𝑅𝑅2 cos 𝜙𝜙𝑤𝑤 . For the active case, we
need to find the maximum value of this quantity. This requires different failure
mechanisms to be investigated to find the value of 𝜃𝜃 that gives 𝐻𝐻𝑚𝑚𝑚𝑚𝑚𝑚 . This can be easily
determined by graphical methods.
In the design of retaining walls, it is necessary to also check that the vertical component
of the load can be resisted by the soil.
During total stress analyses, it is necessary to allow for the presence of tension cracks,
and, if water is present, the possibility that these cracks will fill with water.
2𝑐𝑐𝑢𝑢 �𝑁𝑁𝜙𝜙
The tension crack depth is determined from Rankine's method giving 𝑧𝑧 = .
𝛾𝛾
The problem geometry and the forces acting on the soil wedge then become:
𝑧𝑧
𝑊𝑊1
𝐻𝐻 𝐶𝐶2∗
𝑊𝑊2 𝐶𝐶1∗
𝜃𝜃 𝜃𝜃
𝑅𝑅2∗ 𝜙𝜙𝑤𝑤 𝜙𝜙𝑢𝑢
𝑅𝑅1∗
182
If the tension cracks fill with water, this has no influence on the polygon of forces, but the
water provides an additional horizontal force 𝑈𝑈 on the wall as shown below.
𝑈𝑈
𝑅𝑅2∗
𝐶𝐶2∗ 𝐹𝐹𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡𝑡
Consider the retaining wall shown below with the soil properties as indicated. Determine
the minimum force required for stability assuming that the failure involves a plane at an
angle 30° to the vertical passing through the toe of the wall. Ignore the possibility of
tension cracks.
𝐷𝐷
𝐵𝐵 10° Soil Properties:
soil 𝑐𝑐𝑢𝑢 = 10 kPa
movement
𝜙𝜙𝑢𝑢 = 10°
5m
𝑐𝑐𝑤𝑤 = 2 kPa
30°
𝜙𝜙𝑤𝑤 = 20°
𝛾𝛾 = 20 kN/m3
𝐴𝐴
183
𝐷𝐷
𝐵𝐵
𝑊𝑊
𝐶𝐶𝑎𝑎𝑎𝑎 𝐶𝐶𝑎𝑎𝑎𝑎
30° 10°
𝑅𝑅𝑎𝑎𝑎𝑎
20° 𝑅𝑅𝑎𝑎𝑎𝑎
𝐴𝐴
𝐴𝐴𝐴𝐴 = 6.4 m
𝐴𝐴𝐴𝐴 = 5 m
wedge area = 8 m2
Hence,
20°
𝑅𝑅𝑎𝑎𝑎𝑎
𝑅𝑅𝑎𝑎𝑎𝑎
50°
160 10
30°
64
184
17.3 Effective Stress Analysis
Effective stress analysis is always appropriate but pore water pressures must be known.
The failure criterion must now be written in terms of effective stress so that:
𝜏𝜏 = 𝑐𝑐 ′ + 𝜎𝜎 ′ tan 𝜙𝜙 ′
𝑇𝑇 = 𝐶𝐶 ′ + 𝑁𝑁 ′ tan 𝜙𝜙 ′ (4)
where 𝑁𝑁 ′ = 𝑁𝑁 − 𝑈𝑈 and 𝑈𝑈 is the force due to the pore water pressure acting on the failure
plane.
𝑇𝑇 𝐶𝐶 ′
𝜙𝜙 ′
′
𝑁𝑁
𝑅𝑅 ′
𝑈𝑈
𝑈𝑈
• When the soil is dry the pore pressures everywhere will be zero and the effective
stresses will equal the total stresses. However, only an effective stress analysis is
appropriate.
• If sliding occurs between the soil and a wall, appropriate effective stress failure
parameters must be used. The effective parameters between any interface (e.g. a
′
wall) and the soil should be based on the ultimate conditions so that 𝑐𝑐𝑤𝑤 = 0 and
′ ′
𝜙𝜙𝑤𝑤 = 𝜙𝜙𝑤𝑤,𝑢𝑢𝑢𝑢𝑢𝑢 .
185
For active failure the mechanism giving the greatest force is needed
For passive failure the mechanism giving the least force is needed
The most critical mechanism is unlikely to give an accurate estimate of the failure
load because observation of real soil shows failure rarely occurs on planar surfaces.
• To assist in selecting the appropriate values of 𝜃𝜃 to use for the assumed failure
planes in the soil, it is helpful to remember that the failure plane is inclined at an
angle (𝜋𝜋/4 − 𝜙𝜙/2) to the direction of the minor principal stress 𝜎𝜎3 .
𝜏𝜏
𝐹𝐹
𝜙𝜙 2𝛼𝛼
𝜎𝜎3 𝜎𝜎
If the wall is vertical and frictionless then 𝜎𝜎3 will be horizontal, and the angle of
the failure plane will be at (𝜋𝜋/4 − 𝜙𝜙/2) to the vertical. If the wall is rough, then
𝜎𝜎3 will not be horizontal and the angle of the failure plane will change.
• In the presence of steady state seepage, it may be necessary to draw a flow net to
determine the pore water forces 𝑈𝑈 acting on the soil wedge.
186
All the possible forces acting on a failing soil wedge have now been determined for an
effective stress analysis. These are:
𝐵𝐵 D
𝐶𝐶𝑎𝑎𝑎𝑎 direction of
𝑊𝑊 𝐶𝐶𝑎𝑎𝑎𝑎 movement of soil
wedge
𝑈𝑈𝑎𝑎𝑎𝑎
𝑅𝑅𝑎𝑎𝑎𝑎 𝜙𝜙 ′
′
𝜙𝜙𝑤𝑤 𝜃𝜃
𝑅𝑅𝑎𝑎𝑎𝑎
𝑈𝑈𝑎𝑎𝑎𝑎
𝐴𝐴
𝐵𝐵 D
direction of
𝐶𝐶𝑎𝑎𝑎𝑎 movement of soil
𝑊𝑊 𝐶𝐶𝑎𝑎𝑎𝑎 wedge
𝑅𝑅𝑎𝑎𝑎𝑎
𝑈𝑈𝑎𝑎𝑎𝑎 𝑅𝑅𝑎𝑎𝑎𝑎
′
𝜙𝜙𝑤𝑤 𝜃𝜃 𝜙𝜙 ′
𝑈𝑈𝑎𝑎𝑎𝑎
𝐴𝐴
187
Example 2: Effective Stress Analysis
Considering the same wall and geometry as in Example 1, but now using the effective
strength parameters for the soil as given below.
Soil Properties:
𝐷𝐷
𝐵𝐵 10° 𝑐𝑐 ′ = 5 kPa
soil 𝑋𝑋 𝜙𝜙 ′ = 10°
movement
′
𝑐𝑐𝑤𝑤 = 2 kPa
5m
′
𝜙𝜙𝑤𝑤 = 20°
30°
𝛾𝛾𝑑𝑑𝑑𝑑𝑑𝑑 = 20 kN/m3
𝛾𝛾𝑠𝑠𝑠𝑠𝑠𝑠 = 22 kN/m3
𝐴𝐴
𝐴𝐴𝐴𝐴 = 6.4 m
𝐴𝐴𝐴𝐴 = 5 m
𝐴𝐴𝐴𝐴 = 5.77 m
𝐵𝐵𝐵𝐵 = 2.89 m
Hence,
′
𝐶𝐶𝑎𝑎𝑎𝑎 = 2 × 5 = 10 kN/m
′
𝐶𝐶𝑎𝑎𝑎𝑎 = 5 × 6.4 = 32 kN/m
𝑊𝑊 = 0.5 × 5 × 2.89 × 22 + (8 − 0.5 × 5 × 2.89) × 20 = 174.5 kN/m
188
Water pressures on the soil wedge due to groundwater table are as follows:
𝑈𝑈𝑎𝑎𝑎𝑎
𝑈𝑈𝑎𝑎𝑎𝑎
20°
𝑅𝑅𝑎𝑎𝑎𝑎
𝑈𝑈𝑎𝑎𝑎𝑎 60°
𝑅𝑅𝑎𝑎𝑎𝑎
𝑊𝑊
𝑈𝑈𝑎𝑎𝑎𝑎 60°
′
𝐶𝐶𝑎𝑎𝑎𝑎
30°
′
𝐶𝐶𝑎𝑎𝑎𝑎
189
′
From the polygon of forces, we obtain 𝑅𝑅𝑎𝑎𝑎𝑎 = 17 kN/m. The forces acting on the wall
are:
𝑅𝑅𝑎𝑎𝑎𝑎 𝑈𝑈𝑎𝑎𝑎𝑎
′
𝐶𝐶𝑎𝑎𝑎𝑎
Note that 𝑁𝑁𝑎𝑎𝑎𝑎 is largely due to water pressure. However, due to water on the other side of
the wall, the net resistance required for stability is only 15.97 kN/m.
Example 3
The figure below shows an example where there is steady state seepage towards a wall
which has a vertical drain behind it and the water level to the right of the wall is maintained
at a height 𝐻𝐻𝑤𝑤 . The wall supports soil which has a saturated unit weight of 𝛾𝛾, an ultimate
′
friction angle of 𝜙𝜙𝑐𝑐𝑐𝑐 and a sloping soil surface. The wall is rough and has an angle of
′
friction, 𝜙𝜙𝑤𝑤 .
Calculate the limiting horizontal force which is required to prevent failure by the wall
moving away from the soil.
The only difference between this and the previous example is that the flow net must be
used to determine the force due to the pore water along the assumed failure plane, 𝐴𝐴𝐴𝐴.
The pore water pressures can be determined in the normal way. However, in this situation,
the pore pressure can be determined simply by noting that the top flow line is the phreatic
190
surface where 𝑢𝑢 = 0. As the head is constant on any equipotential, the head at 𝐷𝐷 is equal
to the head at 𝐸𝐸 and the difference in pore pressure is simply 𝛾𝛾𝑤𝑤 times the vertical distance
between 𝐸𝐸 and 𝐷𝐷.
′
𝜙𝜙𝑐𝑐𝑐𝑐 ′
𝑈𝑈𝑤𝑤 = 0.5𝛾𝛾𝑤𝑤 𝐻𝐻𝑤𝑤2
𝑈𝑈 from 𝜙𝜙𝑤𝑤
flow net ′
𝑅𝑅 ′ 𝑅𝑅𝑤𝑤
191