You are on page 1of 30

ENCN353: Geotechnical Engineering

Module 4: Elastic deformation and


settlements of shallow foundations
1 Introduction
The next set of lectures deal with the use and applicability of elastic
methods to determine stress changes and corresponding
settlements due to shallow foundation loads for flexible foundations,
before the onset of failure at higher loads. In essence, this is looking
at “Serviceability Limit State” behaviour rather than the Ultimate
Limit State (which we covered in Module 3).

We need to have a good understanding about how a 3-dimensional


element of material will behave when loaded.

You are now aware that in load-deformation behaviour:

1.

2.

However, we can start with the idea of linear elasticity and work from
there to build up a picture of how soil deforms in 3D when loaded.
That is, we assume the soil can be described using concepts of
elasticity with appropriate elastic constants. Then we can add in the
non-linear behaviour where needed.

By the end of this module you should be able to:


1. Develop and manipulate the 3D elasticity equations for soil,

2. Select, justify and calculate the appropriate soil parameters


(e.g. G, E, K, E’, Ku etc.) for different loading scenarios,

3. Illustrate the difference between tangent and secant modulus


and explain when each might be used,

4. Calculate soil stresses due of surface loads using


Boussinesq’s solution, Newmark’s chart and Fadum’s charts,

5. Analyse soil settlement problems and explain and justify the


assumptions and idealisations in your analysis.

ETB/JJMH Page 1
ENCN353: Geotechnical Engineering

2 The selection of elastic parameters


2.1 Fundamentals of elasticity
It is important to understand how different elastic constants (modulii
G, E and K and Poisson’s ratio ν) relate to each other. We can then
use these to estimate short term (immediate or “undrained”) soil
deformations and long term (“drained”) deformations.

2.1.1 Normal stresses and strains


Consider a cube of dimensions x = y = z, subjected to forces Px, Py
and Pz normal to the sides.

The normal stresses are:

Let us assume that under these forces the cube is compressed by


∆x, ∆y and ∆z, in their respective directions. The strains (assumed
small) are:

The volumetric strain is:

ETB/JJMH Page 2
ENCN353: Geotechnical Engineering

2.1.2 Shear stresses and strains


Consider the XZ plane and apply a shear force F that causes the
square to distort as shown below.

The shear stress is:

The shear strain (assuming small) is:

We can apply the same principles to the other two directions.

2.1.3 Poisson’s Ratio


If we apply a compressive load in one direction to a deformable
cube we find the cube does not just deform in that direction but also
laterally. I.e.:

The ratio of the lateral strain


to the vertical strain is called
Poisson’s ratio:

Note the negative sign!

ETB/JJMH Page 3
ENCN353: Geotechnical Engineering

2.1.4 Hooke’s Law in 3D


Hooke’s law for an isotropic elastic material:

∆ε x = 1 ( E )(∆σ x − ν ∆σ y − ν ∆σ z )

∆γ xy = 1 ( E ) [2 (1 + ν )∆τ ]xy

We can add together direct strains (assuming small) to obtain


volumetric strain:

And similarly isotropic stress:

2.1.5 Elastic material constants


A reminder of the elastic material “constants” from solid mechanics
(NB: they are true constants for many materials, but not for soil,
however, they are still useful to us!):

Young’s modulus:

Shear modulus:

ETB/JJMH Page 4
ENCN353: Geotechnical Engineering

Bulk (volumetric) modulus:

Poisson’s ratio:

(NB: these are secant definitions, tangent definitions would be ratios


of change in stress to change in strain, e.g. ∆σ/∆ε).
Gives bulk modulus, K as:

And we find shear modulus G:

Proofs and special cases:


We put the subscript “u” for undrained behaviour or prime (‘) for
drained behaviour – i.e. EU or E’, KU or K’ and GU or G’.

Hence:

However, it is assumed that soil can only compress by virtue of a


loss in void ratio. In a saturated soil this can only happen during
drained loading as water is expelled from the pores. Therefore, for
undrained loading (i.e. short term behaviour), as ∆εvol = 0, KU = ∞.
This means that the undrained Poisson’s ratio, νU = 0.5.
Terzaghi’s principle says the soil grain contact network (skeleton) is
responsible for taking all the shear stress (because water cannot
take shear). Hence the effective and total shear stresses are
identical and G (undrained) equals G (drained): GU = G’.

ETB/JJMH Page 5
ENCN353: Geotechnical Engineering

Finally, the drained (effective stress) and undrained (total stress)


values of Young’s Modulus, E’ and EU, and Poisson’s ratio, ν’ and
νU, are linked by the shear modulus G = G’ = GU.
We can measure G relatively easily in a triaxial undrained
compression test. If the cell pressure is held constant, ∆σ2 = ∆σ3 = 0,
so:

From the Mohr’s circle plot above we can relate stress and strain
increments so that:

ETB/JJMH Page 6
ENCN353: Geotechnical Engineering

And (remembering G’ = GU):

2.2 Appropriate soil moduli


It is extremely important that we have an idea of the soil stiffness for
our particular working load, in order to obtain a reasonable
settlement estimate…
The figure below shows typical stiffness response of a soil with
increasing strain in terms of stiffness modulus E versus axial strain,
in say, a triaxial test (your first lab!). This curve is sometimes called
the “stiffness degradation curve”.

ETB/JJMH Page 7
ENCN353: Geotechnical Engineering

Soil stiffness may be altered by changing the loading direction. This


“recent stress history” effect results in the soil behaving more stiffly
upon unloading and then reloading – for example, this can happen
when excavation is followed by construction of a foundation.
Stiffness will also be different in the case of a pre-loaded or over-
consolidated soil (i.e. on an unload-reload line) compared with a soil
undergoing first-time compression (on a normal compression line).

These two factors lead to the different initial stiffnesses as seen in


the above figure.
Hence it is important that an elastic stiffness is selected that is
appropriate for the particular design situation. In other words, the
design loads need to be compared with the initial conditions of the
soil to determine the correct stress-strain response during
construction and design life.

ETB/JJMH Page 8
ENCN353: Geotechnical Engineering

2.3 Which stiffness to use? Tangent or secant?


2.3.1 Definitions
The definition of tangent and secant modulus is given below:

Secant modulus
The secant modulus is found by drawing a straight line from a datum
point to a point on the non-linear stress-strain curve. It is defined as
the ratio of change in stress to strain as measured from the datum
point:

Tangent modulus
The tangent modulus is found by drawing a tangent to the stress-
strain curve at a particular point (i.e. it is the slope of the stress-
strain curve). It is defined as the rate of change of stress with strain:

Reference strain
Both types of modulus, whether Esecant, or Etangent are usually
referenced at a particular strain level – e.g.

ETB/JJMH Page 9
ENCN353: Geotechnical Engineering

2.3.2 Which stiffness?


In an elastic calculation the strain depends on the stress and
modulus – hence an iterative procedure might be needed to find a
value of modulus that is consistent with the strain. Secant modulus
is easier to use for this purpose.
In general the “average” soil deformation response across a stress
change is the important parameter for settlement – hence, the use of
secant stiffness is often desired.
Care needs to be taken however, to judge each case on its merits!

2.3.3 Variable stiffness with depth


Gibson soil
In many cases soil stiffness tends to increase with increasing mean
effective stress – this means that the soil will become stiffer with
depth. This type of soil is sometimes referred to as a “Gibson” soil.
There are some standard solutions (see later) which deal directly
with this type of soil. However, it is also possible to subdivide layers
of soil with different (increasing) characteristic stiffness to obtain
settlement estimates beneath an arbitrarily (non standard) shaped
foundation.

Layered soils
Finally, with different geological formations, there may also be layers
of different stiffness overlying each other. It is important to treat each
with a different characteristic stiffness to obtain reasonable
estimates of settlement.

ETB/JJMH Page 10
ENCN353: Geotechnical Engineering

3 Soil stresses due to surface loads


3.1 Boussinesq’s solution
In 1885, Boussinesq published solutions for the determination of the
stresses and strains at any point for an isotropic, homogeneous,
elastic half-space, acted upon by a point load at its surface.
• Isotropic:

• Homogeneous:

• Half-space:

ETB/JJMH Page 11
ENCN353: Geotechnical Engineering

The result for a point load can be made useful to us by integrating


over any given area so that a finite foundation size can be applied to
a soil surface and corresponding stresses and strains can be
determined at any point in the ground.
One such useful case is that of a uniform circular surcharge acting
over a radius R with magnitude q. The resulting change in vertical
stress ∆σz, at a depth z below the centreline, determined from
integration of Boussinesq’s solution is:

Equation (i)

A similar expression may be obtained for stresses due to a strip


loading and for points away from the centreline.
The figure below shows the resulting contours of vertical stress
increase for both a circular footing and a strip footing under a
uniform surcharge. Note how the stresses decay much more rapidly
for the circular footing case due to the load being able to spread in
two orthogonal directions compared to just one for the strip footing.

We can see from the form of the equations above that stiffness (a
source of some uncertainty, as discussed earlier) does not come
into the calculation of stress. This means that, even if the soil is not
behaving truly elastically, or homogeneously the Boussinesq
solution can work rather well. In particular it has been shown that
where the soil stiffness increases with depth (sometimes referred to
as a “Gibson” soil) or where there are layers of different soil present,
there is often little difference between the “true” stress distribution
and that predicted by Boussineq.

ETB/JJMH Page 12
ENCN353: Geotechnical Engineering

Contours of increase in vertical stress (a) below a circular footing of diameter B,


and (b) below a strip footing of width B, subjected to a uniform surcharge q.
A few cautionary cases, leading to significant deviation from ideal
Boussineq results:

ETB/JJMH Page 13
ENCN353: Geotechnical Engineering

The following two methods use Boussineq’s solution integrated over


a circular area to estimate changes in vertical foundation stresses
and thereby, settlements.

3.2 Newmark’s method for arbitrarily shaped foundations


In 1942 Newmark devised a graphical method which directly uses
equation (i) to determine vertical stresses at a point at any depth
due to an arbitrarily shaped surcharge. This can then be used to
determine settlements, as we shall see later.
The chart is shown below and is given as a handout for you to work
on.

The Newmark chart is essentially a plan view of the surface of an


infinite elastic half-space, divided into n elements. Each element,

ETB/JJMH Page 14
ENCN353: Geotechnical Engineering

when loaded with a uniform stress q, will result in an increase in


vertical stress of q/n at depth z below the centre of the chart. The
depth z is linked to the scale of the chart (i.e. “scale line” = depth z)
and must be shown on it.
The chart has 200 elements in total, so that if any element is loaded
by a uniform stress of q, this will result in an increase in vertical
stress of q/200 or 0.005q at a depth z below the centre, x. Hence, it
can be used to calculate the increase in vertical stress at any
location within the soil mass due to the application of a surface load
of arbitrary size and shape.
How to use:
1. Draw a plan sketch of the foundation outline such the length
for the scale line equals the depth of interest z and so that the
point of interest x is at the centre of the chart.
2. Count the number of blocks Nq covered by the foundation
loading with surcharge q (note, q could vary across the
foundation, so care needs to be taken here to apportion N with
each value q). Group together parts of blocks.
3. The vertical stress ∆σz at depth z beneath point x is given by:

Equation (ii)

3.2.1 Example 1 – embankment over a pipeline


It is wished to construct a 5m high embankment to carry a road over
some low-lying land. The embankment is 5m wide at the crest and
6m at the base. It may be treated as a uniform surcharge of base
width 6m with soil unit weight γ=20KN/m3. Within the ground at the
site, a gas pipeline already exists at a depth of 8m. The
embankment will run approximately perpendicular to the pipe. The
engineers are concerned about the loading (and deformation) of the
pipe.
Using Newmark’s chart, calculate the additional vertical loads acting
along the pipe on the embankment centreline, at the edge and at a
midpoint between on either side to obtain a loading profile. Plot the
profile.

Problem geometry:

ETB/JJMH Page 15
ENCN353: Geotechnical Engineering

Solution using Newmark’s chart:

ETB/JJMH Page 16
ENCN353: Geotechnical Engineering

We need to determine the stress at a number of points along the


pipe. Using symmetry, draw the embankment on the Newmark
charts at various positions along the pipe, ensuring the scale line
equals 8m (the depth of interest).
Counting the squares (equal to N), q is surface load (equal to
20kN/m3 * 5m = 100kPa), so using the previous equation, we can
draw a table of data:
A B C D E F

Centreline (m) 0 3 6 9 12 18

∆σz (kPa)

ETB/JJMH Page 17
ENCN353: Geotechnical Engineering

Distribution of stress:

3.2.2 Example 2a – A 3.0m × 4.5m concrete foundation


A rectangular concrete slab, 3m × 4.5m, rests on the surface of a
soil mass. The load from the slab is 2025kN. Use Newmark’s
analysis to determine the vertical stress in the soil at 3m:
(a) under the centre of the slab,
(b) under the edge of the slab,
(c) at a distance of 1.5m from a corner,

ETB/JJMH Page 18
ENCN353: Geotechnical Engineering

Strategy: Scale according to depth of interest and plot the


foundation on a Newmark’s chart, with the point of interest over the
centre of the chart. The scale on the chart is set equal to the depth.
The uniformly distributed stress is:

For A: Count the squares for one quarter and multiply by 4 (bi-
directional symmetry):

For B: Count the squares for one half and multiply by 2 (symmetry)

For C: There is no symmetry, so all squares need to be counted.

NB: There is a more complex problem in the Powrie textbook.

3.3 Fadum’s charts for rectangular footings


In 1948, Fadum presented a chart-based solution to the Boussinesq
case of the loading of a rectangular area by a uniform surcharge.
Given that many foundations are essentially rectangular in nature,
this method provides for faster analysis compared with the otherwise
more versatile Newmark method. It is particularly useful for
combinations of rectangular loads on multilayered soils, as results
can be superposed.
Given a rectangularly loaded area of length L and width B, the
increase in vertical stress ∆σv at a corner at depth Z can be
determined using Fadum’s chart. This gives values of an influence
factor Iσ, in terms of dimensionless parameters m=L/Z and n=B/Z
(note that L and B are interchangeable).

ETB/JJMH Page 19
ENCN353: Geotechnical Engineering

Such that for an applied surcharge q, the change in vertical stress is:

Because the analysis is elastic, one can superpose results to obtain


the stress change at a point due to complex foundation shapes and
variable surcharges. This is shown schematically in the figure below:

ETB/JJMH Page 20
ENCN353: Geotechnical Engineering

The use of superposition to enable analysis of a complex foundation


can best be illustrated by an example.

3.3.1 Example 2b – A 3.0m × 4.5m concrete foundation (again!)


For the foundation loading previously analysed (repeated below)
using Newmark’s analysis, use Fadum’s chart to determine the
vertical stresses in the soil at 3m depth for points A, B and C.

Strategy: The slab is rectangular and the equations for a uniformly


loaded rectangular area are for the corner of the area. Divide the
area so the point of interest is at a corner. You may need to extend
the effective loaded area if a point of interest is outside the true
loaded area (e.g. Point C). The extension is fictitious, so will then
need to subtract another area in order to obtain the true result (as
shown above).

ETB/JJMH Page 21
ENCN353: Geotechnical Engineering

Point A: A is at the corner of 4 rectangles. Find stress increase for


one rectangle of size B=1.5m, L=2.25m, and multiply the result by 4.

Point B: B is at the corner of two rectangles of dimensions


3m×2.25m. Find stress increase for one such rectangle and multiply
by 2.

Point C: C is outside the rectangular slab. Extend the rectangle to C


– resulting in the stress increase due to a rectangle of dimensions
4.5m×4.5m, then subtract the stress increase for the smaller
rectangle of dimensions 1.5m×4.5m.

(Note there is a small difference in the value for C from the


Newmark’s chart analysis. This could be due to small errors in
reading off either chart. Neither method is “more accurate” than the
other.)

ETB/JJMH Page 22
ENCN353: Geotechnical Engineering

4 Soil settlements due to surface loads


Two elastic approaches are available for the calculation of surface
settlements. One uses the results of the above type of analysis
(Newmark or Fadum) and vertical stress changes assuming the soil
can be divided into distinct horizons. The other uses charts and
tables to directly give settlements based on particular regular
foundation shapes.
4.1 Settlements from vertical stress and strain behaviour
Both Newmark and Fadum’s chart methods give vertical stresse
increases in the soil due to a surcharge at arbitrarily chosen points.
These results can be combined with a knowledge of the stress-strain
response of the soil to estimate settlements in each layer.
Terzaghi (1943) suggested that, because settlements tend to be
predominantly vertical in nature, the one-dimensional soil modulus,
E0’ could be used in combination with vertical stress changes to
produce an estimate of settlement, δρ, over a soil depth h. That is:

So,

For settlement at the surface of the soil ρ, the contribution of each


soil layer to settlement is then summed.

4.1.1 Example 2c – A 3.0m × 4.5m concrete foundation (yet again!)


Taking either the results for vertical stress change from Newmark’s
or Fadum’s methods, we can determine the settlement of the soil
layers beneath if we know the one dimensional soil modulus E0’.

ETB/JJMH Page 23
ENCN353: Geotechnical Engineering

Assume that the soil has an increasing stiffness with depth, such
that E’ = 10MPa at the surface increasing to 20MPa at 12m with
v’=0.25 throughout, and we choose to split the calculation into two
layers as shown below:

Calculating the change in E0’ with depth for v’ = 0.25:


Recall:
E (1 − ν )
E0 =
(1 + ν )(1 − 2ν )

Average E0’ =

It just so happens we have the values for the change in stress at A,


B and C at 3m. We can also work out values for the stresses at 9m.
These results for 3m and 9m are given below. Calculate the surface
settlements at A, B and C.
Point A B C
Change in stress at 3m (kPa) 63 45 13.5

Change in Stress at 9m (kPa)

The settlement of each layer is given by:

Taking the layers as 0-6m below ground with E0’=15MPa and 6-12m
below ground with E0’=21MPa, the overall settlement at A is:

ETB/JJMH Page 24
ENCN353: Geotechnical Engineering

ρ=

And at B is:

And at C is:

(So we see we can get a settlement trough outside the loaded zone,
which might affect adjacent structures!)

Potential shortcomings:
• Assumes that most deformation is vertical compression so
short-term settlements due to shear at constant volume cannot
be calculated.
• Division into only two layers of soil is crude – more layers with
different E0’ and ∆σ'v would be better.

4.2 Settlements from standard solutions


There are formulae and charts available, derived from Boussineq’s
solution for a uniform, isotropic elastic material, for standard patterns
of surface load which directly yield surface settlements via Young’s
modulus E, (rather than the one-dimensional modulus E0’). These
should be used with care as they are sensitive to the elastic
parameters assigned to the soil, soil anisotropy etc.
One advantage they have over use of one-dimensional analysis is
that it is possible to obtain estimates of undrained settlements due to
shearing as well as those eventual settlements due to consolidation
(i.e. by using different E’ and EU values for the drained and
undrained cases, respectively).

ETB/JJMH Page 25
ENCN353: Geotechnical Engineering

4.2.1 Circular foundation


The following formulae have been derived from Boussinesq’s
solution for a circular foundation of diameter B, loaded with a
uniform vertical stress of q:
Flexible plate, settlement at centre:

Flexible plate, settlement at edge:

Rigid plate, (settlement at edge or centre):

4.2.2 Rectangular foundation


Also from Boussinesq’s solution, the following formula should be
used with the tabulated values of influence factor, Iρ, below for a
rectangular foundation of length L and width B (where L>=B).

L/B Iρ L/B Iρ L/B Iρ L/B Iρ


1.0 0.561 1.6 0.698 2.4 0.822 5.0 1.052
1.1 0.588 1.7 0.716 2.5 0.835 6.0 1.110
1.2 0.613 1.8 0.734 3.0 0.892 7.0 1.159
1.3 0.636 1.9 0.750 3.5 0.940 8.0 1.201
1.4 0.658 2.0 0.766 4.0 0.982 9.0 1.239
1.5 0.679 2.2 0.795 4.5 1.019 10.0 1.272
Influence factors for calculating the settlement below the corner of a uniformly
loaded rectangular area at the surface of a homogeneous isotropic elastic half-
space.

ETB/JJMH Page 26
ENCN353: Geotechnical Engineering

4.3 A cautionary note – influence lines


Some textbooks on foundation analysis and some design codes
suggest that you can draw a 45º line from the edge of a foundation
downwards through the soil to assess the influence of one
foundation on another. This can be useful but caution is needed not
to over-simplify design problems (see the pressure “bulbs” on page
18)…
Furthermore, design codes sometimes suggest that foundation
loads spread at 1:2 (more conservative on loading) or 1:1 (less
conservative) through the soil for design purposes. These rules of
thumb can be useful for design but should be used with care.
Methods directly derived from Boussinesq result in more accurate
stress distributions and these should be used in cases where using
the true loading distribution would alter / influence the design.

Example (i): Empirical rule from Liu & Evett, 2005 “Soils and Foundations” for
minimum footing spacing.

ETB/JJMH Page 27
ENCN353: Geotechnical Engineering

Example (ii): From Tomlinson (2004) “Foundation Design and Construction”, the
influence of overlapping pressure distributions (using Boussinesq) on minimum
column spacing for stiff soil over a soft soil.

5 Estimation of settlements from SPT and CPT


correlations
It is possible to use field derived correlations between in-situ test
data (e.g. Standard Penetration Test (SPT) blow-count N-values or
Cone Penetration Test (CPT) cone resistance, qc values for different
types of soil) and foundation settlement. These empirical relations
can be found in the likes of Bowles “Foundation Analysis and
Design” (USA) and Tomlinson “Foundation Design and
Construction” (UK). As with all empirical correlations, care is needed
to ensure that the conditions match that for which the correlations
were derived. This may require no little engineering judgement.

ETB/JJMH Page 28
ENCN353: Geotechnical Engineering

Useful approximate stiffness and Poisson’s ratio for typical soils.

ETB/JJMH Page 29
ENCN353: Geotechnical Engineering

(
Spare Newmark’s chart: recall: ∆σ z = 0.005 ∑ N q q ) where z, the
depth of interest, is set as the length of scale line.

ETB/JJMH Page 30

You might also like