DESIGN OF FOUNDATIONS FOR LIGHT STRUCTURES

ON EXPANSIVE SOILS

By John D. Nelson1, Daniel D. Overton2, and K.C. Geoff Chao2

INTRODUCTION

The design of foundations for light structures on expansive soils is more complex
and more critical than that for non-expansive soils sites. It will be more expensive to
develop a site having expansive soils than a site with non-expansive soils, and the
cost of poor design will have more costly effects. Foundations will cost more, and
most likely the site investigation and foundation design will cost more as well.
Although soil sampling methods are generally similar to those for other sites, the
laboratory testing, analysis, and design are more complex. Soils must be
characterized to deeper depths, and the movement of water even in unsaturated zones
must be considered.

Attention was focused on expansive soils approximately 30 to 40 years ago, when

construction on such soils was observed to experience damage different from what
engineers were accustomed to seeing. In many instances, damage due to heave was
attributed to settlement. During construction of the Friant-Kern canal much research
was initiated, and a significant jump in the “learning curve” occurred. Much of the
early work in this area was done in California. The soils there were of a clay nature
and experienced moderate to low expansion potential. Construction and design
procedures were developed to deal with their problems. Portions of Texas have
black cotton soils which may have very high expansion potential. Procedures were
developed for construction on these soils commonly using stiffened mat designs.

In the Front Range area of Colorado, a different type of expansive soil was
encountered. There the soils were highly over-consolidated claystone and clayshale
of the Pierre and Denver formations. These soils exhibit very high expansion
potential with very high swelling pressure. These soils exist over a wide area and
extend into Canada. In the colder climates, it is common for structures to have
basements to keep foundations below frost depth. The pier and grade beam system
was developed for these areas, and the basement walls typically serve as the grade
beam. Other foundation systems, such as over-excavation and replacement with
non-expansive soil are used. Stiffened mat foundations are used less frequently.

The geotechnical engineer typically does not perform the structural design, but is
called upon to provide the geotechnical input for the design. This includes
determination of the soil properties, and computation of soil, slab, and pier
____________________________________________________________________
1. Professor of Civil Engineering, Colorado State University, Fort Collins, Colorado and Corporate
Consultant, MFG, Inc., Fort Collins, Colorado.
2. Senior Geotechnical Engineer, MFG, Inc. 3801 Automation Way, Suite 100, Fort Collins,
Colorado 80525, Telephone 970-223-9600.
movement. In this paper, the important factors to be included in such computations
are presented. The movement of water in the subsoils, particularly in the unsaturated
zones are discussed and field observations are presented. Methods of predicting
free-field heave are reviewed and comparisons are made between predicted values
and observations. Pier design including methods for both rigid and elastic piers is
presented. The force in the piers is computed for use in designing the reinforcing
steel. Input parameters for stiffened mat foundations are discussed. Example
calculations are presented and discussed.

THE NATURE OF EXPANSIVE SOIL

Both residual and transported soils may be expansive. The main factor is the
presence of minerals that exhibit expansive characteristics. The mineral itself does
not expand, but the introduction of water into the spaces between solid particles
produce large pressures, which if not counteracted by applied forces cause the
particles to move apart, causing heave. The propensity for the water to be pulled into
the soil is known as the soil suction. The expansive potential of a soil is highly
dependent upon the soil suction. The following sub-sections will discuss soil suction
and the potential for water contents to increase in a soil profile.

Soil Suction

The driving force for introduction of water into the space between soil particles is
“soil suction”. Soil suction describes the pressure exerted on water tending to suck it
into the soil. It consists of two parts. One component is the matric suction, and the
other is the osmotic suction.

The matric suction, hc, is similar to capillary rise in soils. This is depicted in Figure
1 in which a column of granular soil is placed in a pan of water. The situation shown
is for the case where the column of soil was saturated and allowed to drain. The soil
will remain saturated to some height depicted by the point hd. Below this point,
surface tension of the air-water interface prevents the water from draining further.
The air-water interface in the pore water causes the water pressure to be lower than
the air pressure, causing the water to be in tension. The tension in the water is the
matric suction. The difference between the air and water pressure, (ua – uw)
quantifies the matric suction, and is a function of the radius of the spherical surface
depicted in Figure 1c. This in turn is a function of the pore size, and hence, the grain
size of the soil. The matric suction at a point in the soil in terms of head, hc, is equal
to the height of that point above the water table as shown in Figure 1a. The smaller
the grain size of the soil, the greater the tension in the water, and the higher will be
the capillary rise in the soil.

The capillary head depicted by hd in Figure 1 is called the displacement head. This
represents the air pressure that would be necessary to displace the water from the
pore spaces. This is sometimes called the air entry pressure, if it is expressed in
terms of pressure.

2
MATRIC SUCTION, hc

AIR

hd
SOLIDS

Ua = AIR
PRESSURE
0 100%
WATER
DEGREE OF SATURATION Ts
(a)
(b)
R

Uw = WATER
(c) PRESSURE

Figure 1 Matric Suction in Soil

Osmotic suction, ho, is influenced by the chemistry of the system. Figure 2 shows a
chamber of water divided by a semi-permeable membrane. On one side of the
membrane is pure water, and on the other side is a salt solution. The semipermeable
membrane will allow water molecules to pass through it but not salt molecules.
Because the salt in the solution on the left of the membrane in Figure 2 has an
affinity for water, it will pull water molecules into that space from the space on the
right side of the membrane. The pressure that the salt solution can exert on the pure
water depends on the concentration of the salt in the solution. This pressure is called
the osmotic pressure. In terms of head, this pressure is represented by ho in Figure 2.

Figure 3 shows two idealized clay particles in close proximity to each other. The
electrical charges on the faces of the clay particles hold salt cations in close
proximity to balance the charges. As a result, the concentration of salt in the water
in the space between the particles is higher than in the water outside of that space.
Thus, the electrical field around the clay particle serves as a “pseudo-semi-
permeable” membrane. As water is introduced into the system, the high
concentration of salt between the particles causes pressure to be exerted on the water
molecules tending to force them into the space between the particles. This pressure
is the osmotic suction of the soil.

3
 ho

SEMIPERMEABLE
MEMBRANE

SALT PURE
SOLUTION WATER

Figure 2 Osmotic Pressure Across a Semipermeable Membrane

WATER MOLECULES
CLAY MINERAL

SALT CATIONS

PSEUDO-SEMIPERMEABLE MEMBRANE

Figure 3 Pseudo-Semipermeable Membrane Effect

Causing Osmotic Suction in Clay

4
The total suction is the sum of the matric and the osmotic suction. In terms of their
effect on the behavior of soil, they must be considered independently (Miller and
Nelson, 1993). However, for changes in water content within the ranges normally
encountered in geotechnical engineering applications, changes in osmotic suction are
small. Thus, the change in total suction may be considered to be equal to the change
in matric suction.

The most common expansive mineral is montmorillonite. Because of the nature of

the parent crystal, the montmorillonite minerals are very small and thin, and have a
large surface area. This leads to high water contents, and large values of soil suction.

Depth of Heaving

The depth of soil that is contributing to heave at any instant of time depends on two
factors. These are the depth to which water contents in the soil have increased since
the time of construction, and the expansion potential of the various soil strata. As
water migrates through a soil profile different strata become wetted, some of which
may have more swell potential than others. Consequently, the zone of soil that is
contributing to heave varies with time.

In order to determine the amount of heave that will occur at a particular time it is
necessary to know what zone of soil is being wetted and the expansive nature of that
soil. This in turn depends on the manner in which the groundwater migrates in the
soil. Furthermore, the effect of soil heave on a structure will depend on the
particular member of the foundation that is being affected by the heave, and where
that member is located in the soil profile. Movement of the soil surface will begin
almost immediately after construction, whereas some time will be required for the
soil at deeper depths to become wetted. Thus, a slab-on-grade will begin to heave
almost immediately, but movement of piers will be delayed sometimes for several
years.

Although several years may be required for heave to occur, conservative design will
consider the ultimate amount of heave that can occur in the lifetime of the structure.
Prediction of the ultimate heave requires that the largest zone of expansive soil that
can contribute to heave must be defined, properties of expansive soil in that zone
must be determined, and this zone must be assumed to become wetted at some time
during the life of the structure.

The term “Active Zone” has been in common usage in the field of expansive soils.
However, the usage of that term has taken different meaning at different times and in
different places. Therefore, for purposes of clarity and consistency, the following
four definitions have been put forth (Nelson, Overton, and Durkee, 2001).

1. Active Zone is that zone of soil that is contributing to heave due to soil
expansion at a particular point in time. The depth of the active zone will vary
as heave progresses, and therefore varies with time.

5
 2. Zone of Seasonal Moisture Fluctuation is that zone of soil in which water
contents change due to climatic changes.

3. Depth of Wetting is the depth to which water contents have increased due to
external factors. Such factors could include capillary rise after the
elimination of evapo-transpiration from the surface, infiltration due to
irrigation or precipitation, or introduction of water from off-site.
Underground sources may include broken water lines, perched water tables,
flow through more permeable strata that are recharged at distant locations, or
a number of other factors.

4. Depth of Potential Heave is the depth to which the overburden vertical

stress equals or exceeds the swelling pressure of the soil. This represents the
maximum depth of Active Zone that could occur.

Subsurface Movement of Water

An idealized water content profile in a uniform soil at an undeveloped site in a dry

climate is shown by profile A in Figure 4. It is seen that below some depth,
designated by zs, an equilibrium water content exists. Above that depth, the water
content decreases due to water losses from the ground surface, usually evapo-
transpiration. If a large cover is placed on the ground surface such that the water
content profile shown in Figure 4 is isolated from any edge effects, surface water
losses are eliminated, and the water content profile will come into equilibrium with
the environment as shown, for example, by profile B. If the ground surface is
subjected to temperature fluctuations such as due to summer and winter climates, the
water contents in the zone affected by temperature changes will fluctuate about
Profile B. Profile C, for example would be typical for winter conditions and Profile
D would be typical for summer conditions. The zone in which temperature effects
occur, and depths below that in which climatic effects can change the water content
define the zone of seasonal moisture fluctuation. The depth of this zone would be
less than or equal to the depth zs.

One obvious example of a surface cover is the placement of a slab-on-grade.

Another example of a cover is an irrigated lawn, where water is supplied at a rate
equal to the evapo-transpiration needs of the grass. Although some water content
fluctuations may occur in the upper 6 to 12 inches of the soil profile depending on
the type of grass that is planted, the effect below this surficial zone is to eliminate all
water flux across the upper boundary.

6
 0

B ZONE OF
D SEASONAL
FLUCTUATION
C

DEPTH, z

Zs

0 WATER CONTENT, w

Figure 4 Idealized Water Content Profile

However, if the water introduced by irrigation exceeds the evapo-transpiration needs

of the grass, as is normally the case, a net excess of water exists at the soil surface.
In that case, water will migrate downward. The leading edge of the zone of water
content increase is known as the “wetting front”.

McWhorter and Nelson (1979) developed a methodology to analyze seepage from a

surface water source into an unsaturated soil. Figure 5 shows a water content profile
beneath a surface water source at times t1 and t2 (t2 >t1). The migration of a wetting
front that moves downward with time is evident. Below the wetting front, the water
content is the same as that which existed prior to introduction of the water source.
The depth to the wetting front is the depth of wetting.

This wetting front will continue to move downward as long as the total potential of
the soil above the wetting front is higher than that below the wetting front. For a
uniform soil profile the soil suction in the unsaturated zone below the wetting front
will be greater than that in the soil above the wetting front where water contents are
higher. The difference in soil suction between the wetter and drier zones, along with
gravity, will result in downward flow of water. The wetting front will continue to
migrate downward until a truly impermeable boundary or a water table is reached.
For considerations of expansive soils, even very plastic clays do not represent a truly
impermeable boundary. The movement of the wetting front in those soils may be
very slow, but even small increases in water content may cause significant amounts
of heave.

7
 GROUND
SURFACE

t1
DEPTH, z
t2

WETTING
FRONT
WATER CONTENT
INITIAL WATER ABOVE WETTING
CONTENT FRONT

0
WATER CONTENT, w

Figure 5 Schematic Distribution of Water Content

(After McWhorter and Nelson 1979)

Figure 6 shows actual measured data from the Colorado State University field test
site (Nelson et al., 1994). The similarity between Figures 4 and 6 are evident. The
slight variation of the two water content profiles near the ground surface in Figure 6
show the effects of temperature.

The above discussion relates to a site in which the soil is fairly uniform, or in which
the soil exists as a series of uniform strata. It assumes that there are few or no
discontinuities or zones in which flow patterns will be concentrated. In areas where
the beds exhibit some dip, water movement may be different than described above.
Figure 7 depicts water movement at a site in Loveland, Colorado. At that site the
beds dip to the east at about 20o. Water sources at the site included irrigation and a
small pond to the west of the building. Continuous core was taken during drilling
and between sampling. In the core, the bedding planes were evident, and along
many of them, iron staining could be seen, indicating oxidation due to water
traveling along the beds. The clayshale was soft and iron stained for a distance of 1
or 2 inches on either side of the bedding planes. It was postulated that joints and
cracks also existed in near vertical directions. These discontinuities provided flow
paths between bedding planes. Based on those observations, a conceptual model of
water movement as depicted in Figure 7 was developed.

8
 0

-2

-4

-6
0 DAYS
(JANUARY 1993)
365 DAYS
-8 (JANUARY 1994)

DEPTH (FT)
596 DAYS
(SEPTEMBER 1994)
-10

-12

-14

-16

-18

-20
0 5 10 15 20 25
WATER CONTENT (%)

Figure 6 Water Content Profile from CSU Site (Nelson, et. al., 1994)

OFFICE
IRRIGATION AND PRECIPITATION
SPACE

SOIL
POND
BACKFILL
MANUFACTURING
FLOOR

BEDROCK

CRAWLSPACE

BEDDING PLANES

JOINTS, FRACTURES

PIERS

Figure 7 Water Movement in Dipping Bedrock in Loveland, Colorado

9
Figure 8 shows the water content profile for pre-construction conditions and during
the investigation of the distress at the Loveland site. Observations of the core as
described above show that the water contents are not as uniform as indicated in
Figure 8. In fact, adjacent to the bedding planes the soil is quite wet and water
migrates outward from there into the zones between the bedding planes. The water
movement had progressed to a depth of about 35 feet at the time of the investigation.
As water continues to migrate into deeper bedding planes along the joints and cracks,
the apparent “wetting front” will continue to move downward, and the clayshale at
deeper depths will increase in water content. Water will continue to migrate outward
from the cracks and fissures into the spaces between.

0 0

3 10
DEPTH BELOW GROUND SURFACE (METERS)

DEPTH BELOW GROUND SURFACE (FT.)

6 TOP OF PIERS 20
BEFORE
CONSTRUCTION
(1990)
AFTER
9 CONSTRUCTION 30
(1998)

12 40

BOTTOM OF PIERS

15
50

18
60

21
70
0 5 10 15 20 25

WATER CONTENT (%)

Figure 8 Water Content Profile in Dipping Bedrock in Loveland, Colorado

The general soil profile for a site in Denver is shown in Figure 9. The soil profile
consists of relatively flat-lying beds. The coal seam is a water-bearing stratum that
appears to be recharged at some location off-site. Water content profiles at the site
are shown in Figure 10. Irrigation at the surface has been discontinued for about 2
years, and the entire area has been in a severe drought for the past 3 years. From
Figure 10 it is seen that the main source of water is the migration of water upward
and downward from the coal seam due to soil suction. This illustrates the
introduction of soil water from off-site sources even though infiltration from the
surface is almost non-existent.

10
 DEPTH LITHO- SOIL DESCRIPTION
(FT) LOGY
GRASS ON SURFACE GROUND SURFACE
0
FILL, SANDY SILTY CLAY (0-5.5')

5
SILTY CLAY (5.5'-10')

10
WEATHERED CLAYSTONE (10'-20')

15

20
CLAYSTONE (20'-34.5')

25

30

35
COAL (34.5'-42.5')

40

CLAYSTONE (42.5'-E.O.B.)
45 E.O.B. @ 45'

Figure 9 Soil Profile for Denver Site Showing Coal Seam

0

Fill
5

Silty Clay

10

Weathered Claystone
15

20

June, 2000
depth (ft)

August, 2000
November, 2000
25
Claystone December, 2000
July, 2001
August, 2002
30

35

Coal

40

45 Interbedded Claystone
& Sandstone

50
0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 75 80

volumetric water content (%)

Figure 10 Water Content Profile for Denver Site

11
LABORATORY TESTING – OEDOMETER TESTS

Oedometer tests for expansive soils use the same apparatus as for standard
consolidation tests on soft soils. Tests on expansive soils differ in the loading
procedures that are used. Two basic oedometer swell tests are described below.

Consolidation-Swell Test

In the consolidation-swell test a soil sample is placed in a consolidation ring and

subjected to a prescribed vertical stress, termed the “seating pressure”. After loading
under the seating pressure for a period of time, usually about 24 hrs, the sample is
inundated and allowed to swell while still being loaded at the seating pressure. The
seating pressure may represent the overburden pressure, overburden pressure plus the
applied load from the structure, or some other arbitrary surcharge. After swelling,
the sample is subjected to additional load in increments, and may be unloaded, also
in increments. The swelling pressure is defined as the vertical stress required to
compress the sample back to the volume it had at the time of inundation.

An idealized plot of consolidation-swell test data is shown in Figure 11a. In that

figure, σ′i is the stress at which the sample is inundated and σ′s represents the
swelling pressure according to the above definition. It should be noted that because
the excess pore fluid pressures are allowed to dissipate to zero after each load
increment, it is not necessary to differentiate between effective and total stress in
Figure 11a.

The consolidation-swell test is similar to the Expansion Index test that is widely used
in southern California. In the Expansion Index test the seating pressure is prescribed
to be 144 psf, and the sample is typically not loaded back to determine the swelling
pressure.

Constant-Volume Test

In the constant-volume test the sample is placed in the oedometer and subjected to a
prescribed seating pressure as in the consolidation-swell test. However, when the
sample is inundated the sample is prevented from swelling. The swelling pressure is
the maximum applied stress required to maintain no swell. When the swelling
pressure stops increasing after inundation, the sample may be rebounded by
complete load removal or incremental load removal. Alternatively, it may be loaded
beyond the swell pressure and unloaded following the conventional consolidation
test procedure. An idealized plot of constant-volume test data is shown in Figure
11b.

The swelling pressure measured in the constant-volume test is significantly smaller

than that determined in the consolidation-swell test. This is because the sample
imbibes water in the consolidation-swell test, and the swelling pressure includes the
work necessary to expel the water. In the constant-volume test, the swelling pressure

12
represents only that load necessary to avoid imbibition of water. Correlations have
been developed between the swelling pressures measured in the two different tests.

% SWELL
Cc

eo

Cr

INITIAL SWELL
SURCHARGE PRESSURE

( 'sc )
cs
APPLIED STRESS (LOG SCALE)

Figure 11a Consolidation-Swell Test Data

CS
eo

SWELL
PRESSURE

( 'sc )
cv

APPLIED STRESS (LOG SCALE)

Figure 11b Constant Volume Test Data

13
LABORATORY TESTING - SOIL SUCTION

The most common method of measuring matric suction is the pressure-plate method.
In this method, a pat of soil is placed on a porous plate that has a high air entry
pressure as shown in Figure 12. The bottom of the plate is in contact with water. A
prescribed air pressure is applied to the chamber above the plate. The air pressure
forces water out of the soil, and when equilibrium is reached, the soil is removed,
and its water content is measured. The difference between the air pressure and the
water pressure is the matric suction, (ua – uw) = hc. The matric suction can be plotted
versus the water content to obtain the soil water characteristic curve as shown in
Figure 1a.
Ua

SOLID

WATER

AIR

HIGH AIR ENTRY

Uw PRESSURE STONE
WATER

Figure 12 Axis Translation Technique

The easiest and most common method of measuring total suction is the filter paper
method. A standard filter paper is calibrated by placing it in several different sealed
containers over different salt solutions of known concentrations. The total potential
of the air in the container is in equilibrium with that of the salt solution, i.e., the total
suction. The filter paper imbibes water vapor from the air until it is also in
equilibrium with the air. A calibration curve in the form of suction versus water
content of the filter paper is plotted.

To measure the suction of the soil, a sample is placed in a sealed jar along with a
piece of filter paper that is not in contact with the soil. The jar is placed in a constant
temperature area, such as a standard ice chest, and allowed to sit for approximately 7
days. In this time period the filter paper comes into equilibrium with the soil water
and the air. The filter paper is removed, its water content is measured, and the total
suction is obtained from the calibration curve of the filter paper.

The only special piece of equipment that is needed for this procedure is a balance
that can read to a resolution of 0.0001 gram.

The osmotic suction can be computed as the difference between the total suction and
the matric suction.

14
HEAVE PREDICTION

Heave prediction is generally accomplished using one of two methods. The two
methods are generally described as the “soil suction method” and the “oedometer
method.” In the “soil suction method” the two stress state variables, soil suction and
effective stress, are both used, and indices relating volumetric change to each stress
state variable must be defined. In the oedometer method, only the effective stress is
defined and the soil suction is considered indirectly, by assuming that the final value
of matric soil suction is zero at the end of inundation of the soil.

Methods of predicting heave using the “soil suction method” have been described by
Miller et al. (1995) and McKeen (1992). Although the testing methods required for
this method are not complex, nor difficult to perform, they are not normally
conducted in ordinary geotechnical engineering laboratories. On the other hand,
most geotechnical engineering laboratories routinely conduct oedometer tests, or
some variation of that test. Therefore, only the oedometer method will be discussed
below.

Prediction of Free-Field Heave by the Oedometer Method

Free-field heave is the heave that will occur at the surface of a soil profile if no
surcharge or stress is applied. Because the stress imposed by the weight of an
unloaded slab-on-grade is small, slab heave is approximately equal to the free-field
heave.

A method for prediction of free-field heave using oedometer test data is outlined in
Nelson and Miller (1992). A refinement of that method is presented below.
Comparison of heave predictions using this method with measurements at a number
of sites shows that heave can be predicted with reasonable accuracy using data from
commercial soil laboratories.

To predict the heave of a soil profile, the soil is divided into a number of layers, n, of
thickness z. The general equation for heave is (Nelson and Miller, 1992):

n ⎡ C z
⎛ σ′ ⎞ ⎤
ρ = ∑⎢ log⎜⎜ f ⎟⎟ ⎥
ρ i
Equation 1
⎣ (1 + e o ) i
1 ⎢ ⎝ σ′s ⎠ i ⎥⎦

where: ρ = free-field heave

Cρ = heave index
σ′f = final effective stress state
σ′s = swelling pressure from the constant-volume oedometer test
eo = initial void ratio
zi = layer thickness

15
Determination of Heave Index, Cρ

The heave index, Cρ, can be determined from consolidation-swell test data along
with data from constant-volume tests. However, the constant-volume test data can
be approximated from consolidation-swell test data, as will be described below.
Thus, all of the needed data can be obtained from the consolidation-swell test.

Figure 13a shows the state of stress at four points in a soil profile. Figure 13b shows
idealized curves for consolidation-swell tests performed on several samples at
different seating pressures corresponding to the overburden pressures at points 1
through 4 shown in Figure 13a. At depth z1 the change in void ratio or percent swell
that will take place during inundation will correspond to point A. At depth, z2, the
percent swell will correspond to point B. The sample in consolidation-swell test B
(depth z2) is not able to imbibe as much water as that in test A because of the higher
seating pressure. A similar case is true for sample C.

At the depth of potential heave the effective overburden stress is equal to the
swelling pressure. This point is represented by point D, which corresponds to the
constant-volume oedometer test.

Thus, it is seen that at any point in the soil within the zone of potential heave, the
amount of swell that will occur corresponds to a point on line ABCD in Figure 13b.
Thus, the value of the heave index, Cρ, in Equation 1 is the slope of the line ABCD.

An accurate method to determine Cρ would be to perform several consolidation-swell

tests at different inundation pressures and a constant-volume test. However, a good
approximation for Cρ can be obtained as shown in Figure 14, which utilizes the
results of a consolidation-swell test and a constant-volume test. It is assumed that
line ABCD can be approximated by a straight line such as line C in Figure 14. The
value of Cρ is then given as

Δe s
Cρ = Equation 2
⎛ σ′ ⎞
log⎜⎜ s ⎟⎟
⎝ σ′i ⎠

where: Δes = the increase in void ratio due to swelling during inundation
σ′s = swell pressure from a constant-volume test
σ′i = applied stress at the time of inundation.

16
 Z1 ( vo')A
Z2
( vo')B Z3
DEPTH OF
POTENTIAL
( vo')C HEAVE

vo'= 's

Figure 13a Location of Samples for Figure 13b

Ch C
VOID RATIO

( i')A ( i')B ( i')C S'

APPLIED STRESS (LOG SCALE)

Figure 13b Consolidation-Swell Tests on Samples
with Different Initial Surcharge

17
 3
CONSOLIDATE SWELL TEST DATA

es
Ch
2 C

ecv
Cc
VOID RATIO

eo 1 4
0

CONSTANT VOLUME TEST DATA

σi' σs' (σs')cs

APPLIED STRESS (LOG SCALE)

Figure 14 Comparison of Oedometer Test Approaches for Obtaining
Heave Prediction Parameters

Most soils reports only present the results of consolidation-swell tests. Constant-
volume tests are seldom performed. If only consolidation-swell test data is available,
the value of σ′s can be estimated using the swelling pressure obtained in the
consolidation-swell test. Experience in the Front Range area of Colorado has shown
that the value of σ′s lies at approximately 60% of the distance between σ′i and the
swelling pressure from the consolidation-swell test, (σ′s)cs. Thus, the swelling
pressure from the constant volume oedometer test can be approximated by the
equation:

σ′s = σ i′ + 0.6 [(σ′s )cs - σ i′] Equation 3

Accuracy of Heave Prediction

Near the foothills in the western part of the Denver area, the beds of the clayshale
have been tipped upward during the formation of the Rocky Mountains. To the east
of this area, the beds are more flat lying. In the steeply dipping bedrock areas
expansive soils problems are more severe than in the areas where the beds are more

18
flat lying. In those areas, it is not uncommon to encounter heave in amounts of 1 to
2 feet or more. The differential heave is more severe in these areas, and time rate of
heave is greater due to the higher permeability of clayshale along the steeply dipping
beds.

Data from 13 houses in the Denver and Fort Collins, Colorado area were analyzed to
evaluate the accuracy of the heave prediction method. In the Denver area, 2 of the
houses are located on steeply dipping beds in the Pierre Formation clayshale, and 10
of the houses are located on flat lying beds in the Denver Formation
claystone/clayshale. Only one house in the Fort Collins area was included in this
study. This house was located on the Pierre Formation clayshale. However, the dip
of the beds in Fort Collins is not as steep as those in the steeply dipping area in
Denver.

Data from only consolidation-swell tests was available. This data was obtained from
soil reports prepared prior to construction by commercial geotechnical engineering
companies. Computations of free-field heave were conducted using the data from
the soil borings located most closely to the lots being considered.

At the time that these computations were made it was assumed that the swelling
pressure from the constant-volume oedometer test would be equal to half of that
measured in consolidation-swell tests. More recent experience has shown that
Equation 3 is a better representation. However, the results presented below, indicate
the general accuracy of the heave prediction method.

The observed values of heave for the houses in Denver were estimated from
measurements of differential movement between basement slabs-on-grade and first
floor joists supported on basement wall grade beams. Elevations at the end of
construction were not available, and therefore, it was assumed that the spacing
between the basement slabs and the first floor joists conformed to the design values
shown on framing plans. The observed heave reported for the house in Fort Collins is
the differential heave of the spread footing that occurred after 2 years. Thus, the
observed values have a substantial degree of uncertainty in them.

The measurements were made only 2 to 7 years after construction. Preliminary

analyses of wetting front movement in these slabs showed that as much as 30 years
or more could be required for the entire zone of potential heave to become fully
wetted. Taking into account the projected future heave, and the fact that some
movement of the basement wall grade beams likely has occurred due to pier heave,
the estimated actual ultimate heave was assumed to be two times the measured
heave. Although the accuracy of that factor may be questioned, it can be stated with
some certainty that the actual amount of heave will be higher than the observed
value.

Figure 15 shows the predicted heave plotted against the corrected measured heave, or
the “actual” heave. It is seen that the values of predicted heave are greater than the

19
 “actual” heave. A considerable amount of scatter exists in the data. Conclusions
drawn on the basis of that data must consider the nature of the data that is presented
and the manner in which they were obtained. One purpose of the analysis was to
evaluate how well heave can be predicted based on data presented in standard, state-
of-the-practice soils reports.

18.0

16.0
ρp=2ρa
Flat Lying Beds
- Denver Shale
14.0 ρp=ρa
Fort Collins
- Pierre Shale
12.0
ρp=0.5ρa
Steeply Dipping Beds
Predicted Heave (in.)

10.0 - Pierre Shale

8.0 Solid Points Were Computed Using Ch

Open Points Were Computed Using C s

6.0

4.0

2.0

0.0
0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 22.0 24.0 26.0

Estimated Actual Heave (in.)

Figure 15 Actual Heave Versus Predicted Heave

Lines showing predicted values within a factor of 2 of the “actual” values are shown
in Figure 15. It is seen that almost all points fall within these limits.

Significant inaccuracy is inherent in this data. However, when it is considered that

no correction was made for sample disturbance, actual constant-volume tests were
not performed, and the actual value of heave is based on uncertain data, the results
look good. They show that reasonable heave predictions can be made and some
reliance can be placed on the predictions.

This data shows that with more careful control on the data that is gathered, and with
the continuation of research, it will be possible, in the future, to predict heave of
expansive soils with a fairly high degree of accuracy using the oedometer method.
Recently an investigation has been conducted on a larger structure with carefully
controlled sampling, testing, and surveying. Heave has been predicted within about
20% for that case.

20
PIER DESIGN

Figure 16 shows a typical detail of a type of drilled pier and grade beam foundation
system that is used in the Front Range area of Colorado. A void space must be
maintained between the grade beam and the soil to prevent uplift forces from being
transmitted directly to the grade beam as the soil heaves. Typically a structural floor
is supported on the grade beam. Slab-on-grade floor systems were used in the past,
but did not function well because of intolerable movements due to heave of the
subsoils.

10% SLOPE
DO NOT BACKFILL
2-#4 CONT. TOP & BOTTOM
UNTIL LOWER FLOOR
IS IN PLACE

#4 @ 20" VERTICAL

-#4 @20" HORIZONTAL

BRACE WALL PRIOR

TO BACKFILL
10'

FLOOR SYSTEM

6" VOID BETWEEN CAISSONS

1.5'
GRADE AT AIR SPACE

CONCRETE CAISSON

2-#8 DOWELS EACH CAISSON,

EXTEND FULL DEPTH OF
CAISSON & MIN. 2' INTO WALL

Figure 16 Typical Pier and Grade Beam Detail

The piers are typically uncased, cast-in-place drilled piers, usually without belled
bottoms. The principle on which the piers are designed is to found them in a sound
stratum at sufficient depth so as to provide sufficient anchorage to minimize
movement under the uplift forces exerted by the expansive soil.

If a stable non-expansive stratum exists sufficiently near the surface, the pier may be
designed as a rigid member anchored in that stratum so as to prevent movement.

21
However, if the depth of potential heave is large, the required length of pier may be
too long to be practical. The pier may then be designed as an elastic member in an
elastic medium. The predicted heave of the pier is computed, and the structure must
be designed to accommodate that amount of movement.

Rigid Pier Design

The forces acting on a pier in expansive soil are shown in Figure 17. The principle
of the design is that the negative skin friction below the depth of potential heave plus
the dead load, P, must resist the uplift pressures produced by the swelling pressures
exerted on the pier above that point.

Pdl

UPLIFT SKIN
FRICTION

Zp DEPTH
OF
POTENTIAL
HEAVE

L
fu

(L - Zp)
NEGATIVE SKIN
FRICTION PROVIDING
ANCHORAGE

fs d

Figure 17 Forces on Piers

22
The equation for required length of a rigid straight shaft pier is,

1 ⎡ Pdl ⎤
L = zp +
fs ⎢α 1σ ′s z p − πd ⎥ Equation 4
⎣ ⎦

where, zp = the depth of potential heave

fs = the negative skin friction below the depth of potential heave
α1 = a coefficient of uplift between the pier and the soil
σs ′ = the swelling pressure
Pdl = minimum dead load on the pier
d = diameter of the pier

The maximum tensile force generated in the pier, Pmax, would occur at the bottom of
the depth of potential heave and would be equal to,

Pmax = Pdl - fuzpπd Equation 5

The terms in Equation 5 have been defined above with reference to Equation 4.

A more complete treatment of rigid pier design is presented in Nelson and Miller
(1992).

Rigid pier design works well if a non-expansive stratum exists sufficiently near the
surface, such that anchorage for the pier can be achieved. This is not uncommon
where the formation consists of strata of sandstones and claystones, and a sandstone
stratum underlies an expansive claystone.

However, in a deep deposit of expansive soil, the material below the depth of
potential heave is basically the same material as that in the expansive zone. In this
case if the negative skin friction below the depth of potential heave is approximately
the same as the uplift skin friction, the required pier length is approximately twice
the depth of potential heave. If the depth of potential heave is, for example, 30 ft, the
required length of the pier would be about 60 ft. Such a pier length is generally not
practical for a residential structure. It would be appropriate to use a shorter pier if
the movement of the shorter pier was within tolerable limits. In that case an elastic
pier design would be appropriate.

Elastic Pier Design

In the space available in this paper, a detailed discussion of elastic pier design is not
possible. Therefore, the discussion will be limited to prediction of pier movement
and internal forces for straight shaft piers, i.e., a pier with no belled bottom. For a
more complete discussion of this subject, the reader is referred to Nelson and Miller
(1992).

23
The uplift skin friction along the side of the pier may be considered to be uniform
along the length of the pier or may increase with depth. A case for the former would
be the situation where the soil within the depth of potential heave has generally the
same swelling pressure throughout. A case for the latter would be where the
soundness of a claystone stratum increases with depth as a result of weathering of the
upper material, and hence, the swelling pressure increases with depth.

Figure 18 shows pier normalized heave plotted as a function of the ratio of pier
length to depth of potential heave. The data in this figure is based on design charts
presented in Nelson and Miller (1992). Two curves are presented depending on
whether the soil-pier skin friction is constant with depth or whether it varies linearly
with depth.

Figure 19 shows the normalized maximum tensile force in the pier as a function of
the ratio of pier length to depth of potential heave. The maximum force is
normalized to the frictional force that would exist if the uplift friction were applied
to the entire length of the pier.

For the case where the skin friction varies linearly with depth, this force, PFS, is equal
to,

PFS = − 0.5 fumLπd Equation 6a

For the case where the skin friction is uniform with depth, this force, PFS, is equal to,

PFS = −f um Lπd Equation 6b

From Figure 19 it can be seen that for all but relatively long piers, this ratio is
approximately equal to 0.5.

24
 UNIFORM LINEARLY
UPLIFT INCREASING
FRICTION UPLIFT FRICTION

fu fum

A B
1.0

0.8
A
p 0.6

o
0.4
B
0.2

0
0 0.5 1.0 1.5 2.0

L/ Zp
Figure 18 Normalized Pier Heave Vs L/zp

1.0

0.8

A
Pmax 0.6

PFS 0.4
B
0.2

0
0 0.5 1.0 1.5 2.0

L/Z
Figure 19 Normalized Force in Piers Vs L/zp

25
Accuracy of Pier Movement Prediction

There is not much data available with which to verify the pier movement predictions.
One reason is that pier movement is smaller and occurs more slowly than free-field
heave (slab movement). Also, in most cases, there is not a stable benchmark against
which to measure pier movement.

The authors have had relatively recent experience with two cases in which fairly
reliable data has been available. One is a commercial, relatively heavy masonry
building, located in Loveland, Colorado. This facility was discussed above. It is
located on the Pierre Shale that is dipping to the east at about 20o at that location.
The building has a deep basement with a crawl space below. The piers are 23 feet
long with the top being 21 feet below finished grade.

Piers are 18 to 42 inches in diameter with dead loads varying from 60 to 250 kips.
The depth of wetting at the time of our investigation was about 35 feet. Piers in one
corner of the building had heaved 11 inches. Calculations indicated that maximum
heave would be about 15 inches. Considering that additional heave will occur after
the time of investigation, it is reasonable to conclude that the predicted heave will be
within about 25% of the actual heave.

The other building is located on the flat-lying Denver formation claystones. This
also is a relatively heavy building supported on piers. The ground floor is a slab-on-
grade and has heaved as much as 5 inches in places. The piers are 18, 24, or 30
inches in diameter and are 25 or 28 feet deep. The grade beams supported on the
piers are relatively shallow, so that the tops of the piers are only 2-1/2 to 3 feet below
finished grade. Deep benchmarks have been established that are founded well below
the depth of potential heave to allow for accurate monitoring of pier movement. The
maximum pier movement to date (12 years after construction) is 4 inches. Analyses
have been conducted to predict pier heave as a function of time. Comparison of the
predicted pier movement to date and extrapolation to the ultimate heave has
indicated that the difference between the predicted pier movement and the actual
measured heave is approximately 15 to 20%.

These two cases indicate that with careful sampling and testing, and with accurate
definition of the subsoil profile, pier heave can be predicted with good accuracy.

STIFFENED SLAB DESIGN – INPUT PARAMETERS

Although the actual design of a stiffened slab-on-grade falls within the responsibility
of the structural engineer, the geotechnical engineer is generally asked to provide the
input parameters related to the subgrade soils. The following discussion presents the
general pattern of displacement of a stiffened slab-on-grade, and discusses the
important input parameters.

26
The general pattern of displacement beneath a stiffened slab can be considered to be
what is termed either center heave or edge heave. These are depicted in Figure 20.
The center heave is the most common to occur. It results as water contents beneath
the center of the slab increase faster than at the edges. This is usually the result of
elimination of evapo-transpiration and capillary rise from a deeper water source such
as a perched water table, or even just a zone in which water contents are not
influenced by seasonal changes.

ORIGINAL GROUND SURAFCE

(a)

ORIGINAL GROUND SURAFCE

(b)

Figure 20 Two Principal Mound Shape Conditions: (a) Center Heave and (b)
Edge Heave (Walsh, 1978, Courtesy of CSIRO Australia)

Edge heave is normally observed only in very dry climates and in soils with low
capillary rise. The edge heave is frequently the result of changes in surface drainage
patterns or irrigation. Because of the dry climate, and lack of subsurface moisture,
the center shows little heave and the edges undergo the most movement.

The discussion below will focus primarily on considerations of center heave. Figure
21 shows the mound profiles that will exist for a stiffened slab for conditions of
center heave. Fig 21a shows the profile that will result beneath a weightless slab
with no stiffness, i.e., a membrane placed on the soil surface. The mound is
characterized by the maximum heave at the center with no stress applied at the
surface, ymax. This value of ymax would correspond to the free-field heave, ρmax.

Figure 21b shows the mound shape of a slab of infinite stiffness with both center and
edge loading. If the swelling pressure of the soil exceeds the applied load the slab
will heave. In the center of the slab, because the slab is of infinite stiffness, the
mound will have a flat shape. The distance from the edge to where the slab contacts
the soil mound is termed the edge distance, E. The maximum heave occurs under the
central flat portion of the slab and is equal to ym, which is smaller than ymax in Figure
21a.

27
The situations depicted in Figures 21a and 21b represent the two extremes of
possible situations. In reality, a situation somewhere between these two will exist
such as shown in Figure 21c.

The most commonly used slab design procedure currently used in the United States
is the Post Tensioning Institute (PTI) procedure. The parameters needed for the PTI
procedure are as follows.

GROUND PROFILE AFTER

SLAB CONSTRUCTION
ymax ya

ORIGINAL GROUND
LEVEL
(a)
P P

ym

(b)
P P

ym

(c)

Figure 21 Mound Profile for Various Cases of Slab-on-Grade:

(a) Mound Profile After Slab Placement if no Load is Applied,
(b) Mound Profile for Infinitely Stiff Slab with Load, and
(c) Mound Profile for Flexible/Semirigid Slab with Load

Maximum differential heave, ym, is a percentage of the maximum free-field heave,

ρmax. Often the percentage is taken as a value between 33% and 100%. However, it
is more reliable to calculate the value of ym from Equation 1, taking into account the
loading at the surface produced by the weight of the slab and the applied dead load.

Edge moisture variation distance, e, can be estimated on the basis of climate using
the chart by Wray (1978) presented in Figure 22. That figure indicates values of
edge moisture variation distance up to 6 feet. However, research has indicated that

28
the value of e may be as large as the depth of seasonal moisture change (McKeen
and Johnson, 1990; Durkee, 2000). In many cases this may approach the distance
from the edge of the slab to the center of the slab.

EDGE MOISTURE VARIATION DISTANCE (FT)

NOTE: USE UPPER RANGE FOR
FISSURED SOIL
6
CENTER LIFT
5

3
EDGE LIFT
2

0
ARID DRY MOIST WET

CLIMATE

Figure 22 Approximate Relationship Between Thornthwaite Moisture Index

and Edge Moisture Variation Distance (Wray, 1978)

Allowable soil bearing pressure, qa, can be taken as the unconfined compressive
strength of the soil.

Slab-subgrade friction coefficient, f, can be assumed as a percentage of the angle of

internal friction of the soil. Generally this percentage is assumed to be 50 to 100%.

OVEREXCAVATION AND REPLACEMENT

This system had been used for several years and its use gradually declined. Until the
past 10 years or so, the maximum practical depth of overexcavation was considered
to be about 4 feet. More recently, problems have been observed with pier and grade
beam foundation systems in the areas west of Denver, where the bedrock beds are
steeply dipping. Large shear forces are exerted on the piers because the primary
direction of swell is perpendicular to the beds. Cast-in-place concrete piers typically
have little if any shear reinforcement, and in some cases, piers were observed to have
suffered shear failure.

Thompson (1992) presented observations of damage to residences, and concluded

that the frequency of damage to structures was less if the depth of non-swelling
overburden over the expansive claystone was less than 10 feet. In response to this
conclusion, Jefferson County, Colorado designated an area called the Designated

29
Dipping Bedrock Area (DDBA), in which the beds of the Pierre Shale dip at angles
greater than 30o. The Land Development Regulations require that in this area
foundations for residences be located on 10 feet or more of non-expansive
overburden. This has led to a renewed interest in the overexcavation and
replacement method.

In many cases the expansive soil is overexcavated to depths of 10 below basement

walls or footings on a subdivision wide basis. In other cases it is done on an
individual house basis. The intent is to replace this soil with non-expansive soil, but
more usually it is replaced with the same soil, on the assumption that reworking of
the soil and compaction to proper specifications will minimize the expansion
potential of the remolded soil.

Generally, spread footing foundation systems are then placed on the overexcavated
soil. There have been a limited number of situations in which pier and grade beam
foundations were used.

The soil used to replace the overexcavated soil should be a non-expansive soil with
low permeability. Although granular soils are non-expansive, they are highly
permeable, and if water infiltrates into them, they serve as a small underground
reservoir that provides a constant source of water to the underlying expansive soil.

Typical specifications for the replacement soil is that it should have at least 50%
passing the No. 200 mesh sieve, it should have a liquid limit less than 30%, and it
should be non-expansive. It is important that the soil should be tested to meet these
specifications during the quality control of the project. In particular, consolidation-
swell tests should be performed to assure that the soil is, in fact, non-expansive.

The use of this method in the DDBA has met with mixed results. Some houses have
exhibited signs of intolerable differential heave, but in many of these cases, builders
did not follow all recommendations in the soils report. A number of houses have
performed satisfactorily. The regulations have been in effect since 1995, and more
time is needed to evaluate the success of this method.

EXAMPLE FOUNDATION DESIGN

For purposes of illustrating the above design procedures, two hypothetical subsoil
profiles have been defined. One is typical of a site with relatively highly expansive
claystone representative of a site along the Front Range of Colorado. The other is a
site with a clay of moderate expansion potential typical of a site in San Diego that
was described to me by a local engineer.

The calculations will first be presented without comment, and then will be discussed
in the next section.

30
High Expansion Site (Colorado)

This site is characterized by residual soils and weathered claystone, overlying a deep
deposit of sound claystone that begins at a depth of 6 ft. The structure will have a
basement with a structural floor, and the grade beam foundation walls will be
founded 8 feet below original grade. The minimum dead load on the piers will be 15
kips. The claystone was tested in a consolidation-swell test and exhibited 9.6% swell
when inundated at a seating pressure of 1,000 psf. The swelling pressure was
measured to be 16,000 psf. The total density of the claystone is 136 pcf, with a
water content of 17%. The example will calculate the following.

1. Free-field heave at the bottom of the grade beam

2. Required length of a rigid pier for no movement
3. Required reinforcement steel for the rigid pier
4. Predicted pier heave of a 14 in. diameter pier, 40 feet long
5. Required reinforcement steel for the 40 feet long pier
6. Free-field heave of the surface of the non-expansive fill for 10 feet of
overexcavation
7. Maximum differential heave, ym, of a stiffened mat placed on the original
soil at a depth of 8 feet.
8. Maximum differential heave, ym, of a stiffened mat placed on the
overexcavated and replaced soil at a depth of 8 feet.

Calculations

1. The swelling pressure for a constant volume oedometer test is estimated

using Equation 3. Thus,

σ′s = 1,000 + 0.6(16,000 − 1,000) = 10,000psf = 10ksf Equation 7

The term, Cρ/(1 + eo) in Equation 1 can be determined from Equation 2 by

considering that,

Δe
Cρ 1 + eo
= Equation 8
1 + eo log( σ′s )
σ′i

and noting that Δe / (1 + eo) is the percent swell which, in this case, is 9.6%.

Therefore,

Cρ 0.096
= = 0.096 Equation 9
1 + eo 10,000
log( )
1,000

31
The depth of potential heave, zp, is computed by equating the overburden
pressure to the swelling pressure. Thus,

136z p = 10,000 psf Equation 10

z p = 73.5ft Equation 11

In computing heave, the depth of potential heave is divided into several

layers and the heave of each layer is computed. The total heave is the sum of
heave over all layers. For this example the soil was divided into 30 layers.
The depth to be divided includes the depth of excavation and the depth of
potential heave below the bottom of the excavation. Each layer is then (73.5
+ 8.0) / 30 or 2.72 feet thick. The upper three layers are mostly in the zone of
excavation, and contribute little of any to the heave. The fourth layer extends
from a depth of 8.17 ft. to 10.89 ft. The midpoint of the layer is 1.53 feet
below the bottom of the excavation. The effective stress at that depth is ,

σ′f = 1.53 × 136 = 208 psf Equation 12

From Equation 1, the heave of that layer is

208
ρ 4 = 0.096 × 2.72 × log( ) = −5.27 inches Equation 13
10,000

The negative sign indicates that movement is upward and can be ignored
recognizing that heave is upward. This computation is then repeated for all
layers and the increments of heave are summed. These computations lend
themselves well to computation by simple spreadsheets. The spreadsheet
showing all computations for this case are shown in the Appendix. The total
heave is computed to be 36.5 inches.

2. The required length of a rigid pier is calculated by equating the uplift

forces shown in Figure 17 to the sum of the negative (anchorage) skin
friction forces and the dead load.

The uplift skin friction is equal to,

f u = α1σ′s Equation 14

Where α1 is a coefficient of uplift between the pier and the soil. The value of
α1 can reasonably be assumed to be between 0.10 and 0.25 (Nelson and
Miller, 1992). A value of 0.15 will be assumed here.

32
The uplift skin friction forces are,

14
Fu = f u z p πd = (0.15 × 10) × 73.5 × π × ( ) = 404 kips Equation 15
12

The negative (anchorage) skin friction can be calculated by

f u = α s σ ′h Equation 16

Where:
αs is a coefficient of negative friction between the pier and the
soil, and
σh′ is the lateral stress acting on the pier in the anchorage zone.

The value of αs should be similar to that of α1. Whereas that is most likely
true in most soils, tests at Colorado State University have shown that it may
be somewhat higher in the stiff and, sometimes sandy, claystone in Colorado.
Therefore, a value of 0.25 will be used here. In the next example, for a site in
clays in California a value of 0.15 will be used for αs.

The lateral pressure will be taken as being equal to the swelling pressure of
the claystone.

Thus,

14
Fs = (0.25 × 10)(L − z p )π ( ) = 9.16 × (L − z p ) Equation 17
12

Summing all forces including the dead load,

9.16 × (L − z p ) + 15 = 404 kips Equation 18

(L − z p ) = 42.5 ft Equation 19

L reqd = 42.5 + 73.5 = 116 feet Equation 20

3. The maximum tensile force in the pier is,

Pmax = 15 − 404 = −389 kips Equation 21

The negative sign indicates that the force is tensile. If conventional Grade 60
steel reinforcement, with an allowable design stress of 40 ksi is used, the
required area of steel is,

33
 389
(A s ) reqd = = 9.72 in 2 Equation 22
40

This is probably too much steel to fit in a 14 in. diameter pier. An alternative
is to use high tensile strength DWYIDAG with an allowable design stress of
105 ksi. In that case the required amount of steel is,

389
(A s ) reqd = = 3.7 in 2 Equation 23
105

4. The heave of a 40 feet long pier is computed using Figure 19. Because
the claystone is stiff and has a high swelling pressure, the pier-soil
interaction is considered to be uniform with depth. This corresponds to
case A in Figure 18.

L 40
= = 0.54 Equation 24
z p 73.5

From Figure 18,

ρp
= 0.73 Equation 25
ρo

The free-field heave was previously calculated to be 36.5 inches. Thus, the
predicted pier heave is,

ρ p = 0.73 × 36.5 = 27 inches Equation 26

5. Figure 19 will be used to compute the maximum force in the pier, for use
in computing the required amount of reinforcing steel. From Equation
6b, the value of PFS is calculated to be,

14
PFS = −f u Lπd = −(0.15 × 10) × 40 × π × ( ) = −220 kips Equation 27
12

Again, the minus sign indicates that the force is tensile.

From Figure 19,

Pmax
= 0.51 Equation 28
PFS

Pmax = −0.51 × 220 = −112 kips Equation 29

34
 For Grade 60 steel,

112
(A s ) reqd = = 2.8 in 2 Equation 30
40

6. The top 10 feet below the excavation is assumed to be replaced with non-
expansive compacted fill having a dry density of 115 pcf. Heave
calculations are the same as for the calculations of free-field heave above
except that the upper 18 feet (8 feet of excavation for the basement and
10 ft of overexcavation) of the soil profile consists of non-expansive fill.
The heave calculations are shown in the Appendix. The computed free-
field heave for this case is 23.5 inches.

7. To compute the maximum differential heave for a stiffened mat

foundation, ym, the calculations are the same as for free-field heave
except that the weight of the stiffened mat and the dead load of the
structure are included in the computation of the stresses at the midheight
of the layer. To facilitate the computations by the same computer
spreadsheet as was used previously, the mat weight and the dead load are
represented by an equivalent load placed by 4 ft of the same fill. Thus,
for this case, the spreadsheet shows 8 ft of fill and 4 ft of excavation. The
calculations are shown in the Appendix. The computed heave, ym, is 29.3
inches. Thus, ym, is 80% of ρo.

8. In this case it is desired to compute the value of ym if the soil below the
mat is overexcavated and replaced with non-expansive fill for a depth of
10 ft. As in the previous computation, the weight of the mat and the dead
load are represented by an equivalent load imposed by 4 ft of fill. Thus,
the spreadsheet shows 18 feet of fill and excavation of only 4 ft. The
predicted heave, ym, is 20.2 inches. This is 86% of the heave for the case
when 10 ft of soil were overexcavated and replaced.

Moderately Expansive Site (California)

This site is characterized by a deep deposit of residual clay. A consolidation-swell

test was conducted on this sample at a seating pressure of 144 psf. The sample
swelled by 8% and had a swelling pressure of 6,000 psf. The total density of the soil
is 125 pcf and the water content is 19%. Two alternative foundation systems are
being considered for the structure. One is a pier and grade beam foundation with the
grade beams founded 4 feet below the original grade. The minimum dead load on
the piers will be 15 kips. The length of the piers is not specified but the tolerable
total heave of any pier is 2 inches. The second foundation system is a stiffened mat
placed at a depth of 4 feet either on the original soil or on 4 feet of overexcavated
and replaced soil. The example will perform the same calculations as in the above
example, except instead of computing pier heave of a pier of specified length the
required length of pier to avoid intolerable movement will be calculated.

35
1. In this case it is assumed that a crawl space is excavated under the floor
so that access is provided to facilitate maintaining the void below the
grade beam. The swelling pressure for a constant volume oedometer test
is estimated to be,

σ′s = 144 + 0.6 × (6,000 − 144) = 3,658 psf Equation 31

Also,

Cρ 0.08
= = 0.057 Equation 32
(1 + eo ) 3,658
log( )
144

To compute the depth of potential heave,

125z p = 3,658 psf Equation 33

z p = 29.3 feet Equation 34

The computer spreadsheet showing the computations for this case is included
in the Appendix. The computed heave is 8.6 inches.

2. In the computation of the required length of a rigid pier, a value of 0.15 is

used for α1 as was discussed above for the rigid pier calculations for the
previous example. The uplift skin friction forces are,

14
Fu = f u z p πd = (0.15 × 3.658) × 29.3 × π × ( ) = 59 kips Equation 35
12

As discussed above, a value of 0.15 will be used here for αs.

The anchorage force, Fs, is,

14
Fs = (0.15 × 3.658) × (L − z p ) × π × ( ) = 2.01 × (L − z p ) Equation 36
12

Summing all forces including the dead load,

2.01 × (L − z p ) + 15 = 59 Equation 37

(L − z p ) = 21.9 feet Equation 38

36
 L reqd = 21.9 + 29.3 = 51.2 feet Equation 39

3. The maximum tensile force in the pier is,

Pmax = 15 − 59 = −44 kips Equation 40

The negative sign indicates that the force is tensile. If conventional Grade 60
steel reinforcement, with an allowable design stress of 40 ksi is used, the
required area of steel is,

44
(A s ) reqd = = 1.1 in 2 Equation 41
40

4. In this case the length of the pier has not been specified, but the tolerable
ultimate heave of the pier is 2.0 in2. Figure 19 will be used to find the
required length of a pier to limit movement to this amount.

ρ p 2.0
= = 0.23 Equation 42
ρ o 8.6

In this case because the soil is clay and not as stiff as the claystone in the
previous example, the soil-pier interaction stress will be assumed to increase
linearly with depth. This corresponds to case B in Figure 18. From Figure
18,

L
= 1.1 Equation 43
zp

L = 29.3 × 1.1 = 32 feet Equation 44

5. Figure 19 will be used to compute the maximum force in the pier. For a
32 ft long pier,

14
PFS = −0.5f u Lππ = −0.5 × (0.15 × 3,568) × 32 × π × ( )
12
= −31.4 kips Equation 45

From Figure 20 for Case B,

Pmax
= 0.43 Equation 46
PFS

Pmax = −0.43 × 31.4 = −13.5 kips Equation 47

37
 For Grade 60 steel,

13.5
(A s ) reqd = = 0.34 in 2 Equation 48
40

6. The free-field heave that would occur if the upper 4 ft of expansive clay
below the depth of excavation is replaced with non-expansive fill was
computed using the spreadsheet in the Appendix. The fill was assumed to
have the same properties as in the previous example. The computed
heave is 5.1 inches.

7. The maximum differential heave, ym, for a stiffened mat foundation

placed at a depth of 4 ft on the original soil was calculated in the same
way as in the previous example. As in the previous computation, the
weight of the mat and the dead load are represented by an equivalent load
imposed by 4 ft of fill. The computed heave, ym, is 5.0 inches. Thus, ym,
is 58% of ρo.

8. In this case it is desired to compute the value of ym if the soil below the
mat is overexcavated and replaced with non-expansive fill for a depth of
4 ft. As in the previous computation, the weight of the mat and the dead
load are represented by an equivalent load imposed by 4 ft of fill. The
predicted heave, ym, is 3.1 inches. This is 61% of the heave for the case
when 4 ft of soil were overexcavated and replaced.

DISCUSSION AND CONCLUSIONS

Both sites that were hypothesized for the examples represent conditions that have
higher expansion potential than the average, but are not overly extreme. Soils have
been observed in each of the areas with expansion similar to those used on the
examples. However, the depths of the deposits are usually not uniform for such
great depths. Generally there exist some interbedded strata with lower expansion
potential as well. Nevertheless, the results of the calculations are consistent with
values that have resulted in actual cases.

The predicted free-field heave for the hypothetical sites in Colorado and California
are 36.5 and 8.6 inches respectively. It must be kept in mind that this represents the
ultimate amount of heave that will eventually occur. If the depth of potential heave
is very deep it would require long times for that heave to occur. Heave in the
amount of over 2 ft has been observed within periods of 5 or 6 years, particularly in
some areas of the steeply dipping bedrock in west of Denver. For the California site,
it is very reasonable to expect that in a few years heave in the amount of over 4
inches would occur.

38
For both cases, very long rigid piers are required. For a light structure these lengths
are not practical. It is not practical to design the pier for no movement, and
therefore, either the pier heave of a given length pier is calculated, or the length of
pier needed to limit movement to a prescribed tolerable amount is calculated. In the
case of the Colorado site, the movement of the 40 feet long pier is still beyond
practical limits, and an alternative foundation scheme should be considered. If none
can be found that are within economic constraints, the site is not “buildable”. For the
California site, a 32 feet long pier is well within practical limits. Piers of 35 feet
length are not unusual even in hard claystone.

A large amount of steel is required for the rigid pier in the Colorado site. If that pier
was actually going to be used it would be necessary to use high strength steel. The
amounts of steel required for the California site and for the shorter piers at each site
are reasonable.

It is important to check the bond between the steel and concrete. It is better to use
several smaller bars are than a few large bars. Of course, it is not desirable to use too
many bars that would hamper placement of the concrete. The steel must be tied and
spacers must be used to assure proper bar spacing and appropriate cover between the
steel and the soil. The use of small size aggregate, such as pea gravel, and plasticizer
in the concrete, helps to prevent segregation of aggregate, and facilitates flow into
the drilled pier.

Overexcavation and replacement with non-expansive fill reduced the predicted free-
field heave at both sites to about 60 to 65% of the predicted value for the case
without overexcavation and replacement. This is a significant reduction.

It is important that the fill was assumed to be non-expansive. It is common practice

to remold the original soil and use it as the fill. Although the fill is not as expansive
as it was in its original state, it still is expansive, and in many cases, has been
observed not to function properly. Specifications should specifically state that the
replacement fill must be non-expansive, and the fill soil must be tested as part of the
quality control phase of the project.

The input parameter, ym, for a stiffened mat foundation was calculated to range from
80 to 86% of the free-field heave at the Colorado site, and 58 to 61% at the
California site. It is evident that the load on the stiffened mat is more effective in
reducing heave at the site with the lower expansion potential. However, the
combined use of overexcavation and stiffened mat foundation reduced the heave at
both sites by 35 to 45%. This can be an effective design feature.

39
 APPENDIX
SPREADSHEETS FOR CALCULATIONS
REFERENCES

Durkee, D. B. (2000), “Active Zone and Edge Moisture Variation Distance in

Expansive Soils”, Dissertation submitted in Partial Requirement for the PhD Degree,
Colorado State University, Fort Collins Colorado.

McKeen, G. (1992). “A Model for Predictive Expansive Soil Behavior”, Proceedings

of the 7th Intl. Conf. on Expansive Soils. Dallas, Texas (USA), 1-6.

McKeen, R.G. and Johnson, L.D., (1990), “Climate-Controlled Soil Design

Parameters for Mat Foundations”, J. Geotech. Eng. ASCE 116(7):1073-1094, July.

McWhorter, D. B. and Nelson, J. D. (1979), "Unsaturated Flow Beneath Tailings

Impoundments," J. GeotechEngDiv, ASCE, Vol. 105, No. GT11, November.

Miller, D.J. and Nelson, J.D. (1993). “Osmotic Suction as a Valid Stress State
Variable in Unsaturated Soil Mechanics”, Unsaturated Soils, Geotechnical Special
Publication No.39, ASCE.

Miller, D.J., Durkee, D.B., Chao, K.C., and Nelson, J.D., (1995), “Simplified Heave
Prediction for Expansive Soils,” 1st Intl. Conf. On Unsat. Soils, Paris, 1995.

Nelson, J.D., Durkee, D.B., Reichler, J.D., and Miller, D.J. (1994), “Moisture
Movement and Heave Beneath Simulated Foundation Slabs on Expansive Soils”,
prepared for US Army Corps of Engrs. by Civil Engr. Dept., Colorado State
University.

Nelson, J.D. and Miller, D.J. (1992) Expansive Soils. Problems and Practice in
Foundation and Pavement Engineering. John Wiley and Sons, New York.

Nelson, J.D., Overton, D.D., and Durkee,D.B. (2001). “Depth of Wetting and the
Active Zone.” Expansive Clay Soils and Vegetative Influence on Shallow
Foundations, ASCE, Houston, Texas 95-109.

Thompson, R. W. (1992), “Performance of Foundations in Steeply Dipping

Claystone,” Proc. 7th Intl. Conf. Expansive Soils, Dallas, Texas (USA), 84-88.