Mode of Deposit Formation
Soil deposits that have been transported, particularly via water, tend to be made up of small
grain sizes and initially to be somewhat loose with large void ratios. They tend to be fairly
uniform in composition but may be stratified with alternating very fine material and thin
sand seams, the sand being transported and deposited during high-water periods when stream
velocity can support larger grain sizes. These deposits tend to stabilize and may become very
compact (dense) over geological periods from subsequent overburden pressure as well as
cementing and aging processes.
Soil deposits developed'where the transporting agent is a glacier tend to be more varied in
composition. These deposits may contain large sand or clay lenses. It is not unusual for glacial
deposits to contain considerable amounts of gravel and even suspended boulders. Glacial
deposits may have specific names as found in geology textbooks such as moraines, eskers,
etc.; however, for foundation work our principal interest is in the uniformity and quality of
the deposit. Dense, uniform deposits are usually not troublesome. Deposits with an erratic
composition may be satisfactory for use, but soil properties may be very difficult to obtain.
Boulders and lenses of widely varying characteristics may cause construction difficulties.
The principal consideration for residual soil deposits is the amount of rainfall that has
occurred. Large amounts of surface water tend to leach materials from the upper zones to
greater depths. A resulting stratum of fine particles at some depth can affect the strength and
settlement characteristics of the site.
Quality of the Clay
The term clay is commonly used to describe any cohesive soil deposit with sufficient clay
minerals present that drying produces shrinkage with the formation of cracks or fissures such
that block slippage can occur. Where drying has produced shrinkage cracks in the deposit
we have a fissured clay. This material can be troublesome for field sampling because the
material may be very hard, and fissures make sample recovery difficult. In laboratory strength
tests the fissures can define failure planes and produce fictitiously low strength predictions
(alternatively, testing intact pieces produces too high a prediction) compared to in situ tests
where size effects may either bridge or confine the discontinuity. A great potential for strength
reduction exists during construction where opening an excavation reduces the overburden
pressure so that expansion takes place along any fissures. Subsequent rainwater or even local
humidity can enter the fissure so that interior as well as surface softening occurs.
A clay without fissures is an intact clay and is usually normally consolidated or at least has
not been overconsolidated from shrinkage stresses. Although these clays may expand from
excavation of overburden, the subsequent access to free water is not so potentially disastrous
as for fissured clay because the water effect is more nearly confined to the surface.
Soil Water
Soil water may be a geological phenomenon; however, it can also be as recent as the latest
rainfall or broken water pipe. An increase in water content tends to decrease the shear strength
of cohesive soils. An increase in the pore pressure in any soil will reduce the shear strength. A
sufficient increase can reduce the shear strength to zero—for cohesionless soils the end result
is a viscous fluid. A saturated sand in a loose state can, from a sudden shock, also become
a viscous fluid. This phenomenon is termed liquefaction and is of considerable importance
when considering major structures (such as power plants) in earthquake-prone areas.
When soil water just dampens sand, the surface tension produced will allow shallow ex-
cavations with vertical sides. If the water evaporates, the sides will collapse; however, con-
struction vibrations can initiate a cave-in prior to complete drying. The sides of a vertical
excavation in a cohesive soil may collapse from a combination of rainfall softening the clay
together with excess water entering surface tension cracks to create hydrostatic water pres-
sure.
In any case, the shear strength of a cohesive soil can be markedly influenced by water. Even
without laboratory equipment, one has probably seen how cohesive soil strength can range
from a fluid to a brick-like material as a mudhole alongside a road fills during a rain and
subsequently dries. Ground cracks in the hole bottom after drying are shrinkage (or tension)
cracks.
Changes in the groundwater table (GWT) may produce undesirable effects—particularly
from its lowering. Since water has a buoyant effect on soil as for other materials, lowering the
GWT removes this effect and effectively increases the soil weight by that amount. This can
produce settlements, for all the underlying soil "sees" is a stress increase from this weight
increase. Very large settlements can be produced if the underlying soil has a large void ratio.
Pumping water from wells in Mexico City has produced areal settlements of several meters.
Pumping water (and oil) in the vicinity of Houston, Texas, has produced areal settlements
of more than 2 meters in places. Pumping to dewater a construction site can produce settle-
ments of 30 to 50 mm within short periods of time. If adjacent buildings cannot tolerate this
additional settlement, legal problems are certain to follow.
2-5 ROUTINE LABORATORY INDEX SOIL TESTS
Some or all of the following laboratory tests are routinely performed as part of the foundation
design process. They are listed in the descending order of likelihood of being performed for
a given project.
Water Content w
Water content determinations are made on the recovered soil samples to obtain the natural wa-
ter content w#. Liquid (W>L) and plastic (wp) tests are commonly made on cohesive soils both
for classification and for correlation studies. Water content determinations are also commonly
made in soil improvement studies (compaction, using admixtures, etc.).
Atterberg Limits
The liquid and plastic limits are routinely determined for cohesive soils. From these two limits
the plasticity index is computed as shown on Fig. 2-2a. The significance of these three terms
is indicated in Fig. 2-2a along with the qualitative effect on certain cohesive soil properties
of increasing either Ip or w/,. The plasticity index is commonly used in strength correlations;
the liquid limit is also used, primarily for consolidation estimates.
The liquid and plastic limit values, together with WM, are useful in predicting whether
a cohesive soil mass is preconsolidated. Since an overconsolidated soil is more dense, the void
ratio is smaller than in the soil remolded for the Atterberg limit tests. If the soil is located
below the groundwater table (GWT) where it is saturated, one would therefore expect that
smaller void ratios would have less water space and the WM value would be smaller. From
this we might deduce the following:
If WM is close to WL, soil is normally consolidated.
If WM is close to Wp, soil is some- to heavily overconsolidated.
If WM is intermediate, soil is somewhat overconsolidated.
If WM is greater than w/,, soil is on verge of being a viscous liquid.
Although the foregoing gives a qualitative indication of overconsolidation, other methods
must be used if a quantitative value of OCR is required.
We note that WM can be larger than H>L, which simply indicates the in situ water content
is above the liquid limit. Since the soil is existing in this state, it would seem that overbur-
den pressure and interparticle cementation are providing stability (unless visual inspection
indicates a liquid mass). It should be evident, however, that the slightest remolding distur-
bance has the potential to convert this type of deposit into a viscous fluid. Conversion may
be localized, as for pile driving, or involve a large area. The larger WM is with respect to WL,
the greater the potential for problems. The liquidity index has been proposed as a means of
quantifying this problem and is defined as
^ = WM-Wp = WM-Wp ( 2 1 4 )
WL ~ wp Ip
where, by inspection, values of Ii > 1 are indicative of a liquefaction or "quick" potential.
Another computed index that is sometimes used is the relative consistency,2 defined as
Ic = ^ f ^ (2-Ua)
IP
Here it is evident that if the natural water content WM ^ WL, the relative consistency is Ic ^
0; and if WM > WL, the relative consistency or consistency index IQ < 0.
Where site evidence indicates that the soil may be stable even where WM ^ WL, other
testing may be necessary. For example (and typical of highly conflicting site results reported
in geotechnical literature) Ladd and Foott (1974) and Koutsoftas (1980) both noted near-
surface marine deposits underlying marsh areas that exhibited large OCRs in the upper zones
with WM near or even exceeding Wi. This is, of course, contradictory to the previously given
general statements that if WM is close to Wi the soil is "normally consolidated" or is about to
become a "viscous liquid."
Grain Size
The grain size distribution test is used for soil classification and has value in designing soil
filters. A soil filter is used to allow drainage of pore water under a hydraulic gradient with
2
This is the definition given by ASTM D 653, but it is more commonly termed the consistency index, particularly
outside the United States.
Increases Decreases
Increasing 1P Dry strength
Permeability
Toughness at wP
Increasing wL Permeability Toughness at wP
Compressibility
Nonplastic Viscous fluid
range Plastic range
Water content w, percent
(a) Relative location of Atterberg limits on a water content scale. Note
that w s may be to the right of wP for some soils.
Idealized
Void ratio e
No further
reduction
in e Probable
Water content w, percent
Figure 2-2 The Atterberg limits and (b) Qualitative definition of the shrinkage limit.
some relationships to soil mass properties.
erosion of soil fines minimized. Frequently, the grain size test is used to determine the Dg5,
D 6 0 , Dio fractions (or percents). For example, on Fig. 2-3a, b the D 8 5 (size for which 85
percent of sample is smaller) is about 1.1 mm for the "well-graded" soil. The Dio size is
about 0.032 mm and was determined from the hydrometer test branch of the curve. The
percent clay (particles smaller than 0.002 mm) can be determined from a grain size curve
U.S. as of 1994 British (B.S.) German DIN French
Sieve no. mm Sieve no. mm Sieve no. mm Sieve no. mm
4 4.76
10* 2.00 8* 2.057 34* 2.000
20 0.841 16 1.003 31 1.000
30 0.595 30 0.500 500 0.500 28 0.500
36t 0.422 40Ot 0.400 27f 0.400
4Of 0.420
50 0.297 52 0.295
60 0.250 60 0.251 250 0.250 25 0.250
80 0.177 85 0.178 160 0.160 23 0.160
100 0.149 100 0.152 125 0.125 22 0.125
200 0.074 200 0.076 80 0.080 20 0.080
270 0.053 300 0.053 50 0.050 18 0.050
Breakpoint between sand and gravel.
Use for Atterberg limits.
(a)
Poorly graded
(uniform) soil
Percentfinerby weight
Sieve analysis
Hydrometer
Well-graded soil
Grain size, mm
Figure 2-3 (a) Various standard sieve numbers and screen openings; (b) grain size distribution curves.
such as this, which uses a combination of sieves and a hydrometer test. Typical sieve sizes
as used for sands and silts are shown in Fig. 2-3a.
Unit Weight y
Unit weight y is fairly easy to estimate for a cohesive soil by trimming a block (or length of a
recovered tube sample) to convenient size, weighing it, and then placing it in a volumetric jar
and measuring the quantity of water required to fill the container. The unit weight is simply
_ Weight of sample
Volume of jar - volume of water to fill jar
If the work is done rapidly so that the sample does not have time to absorb any of the added
water a very reliable value can be obtained. The average of several trials should be used if
possible.
The unit weight of cohesionless samples is very difficult (and costly) to determine. Esti-
mated values as outlined in Chap. 3 are often used. Where more accurate values are necessary,
freezing and injection methods are sometimes used; that is, a zone is frozen or injected with a
hardening agent so that a somewhat undisturbed block can be removed to be treated similarly
as for the cohesive sample above. Where only the unit weight is required, good results can be
obtained by recovering a sample with a piston sampler (described in Chap. 3). With a known
volume initially recovered, later disturbance is of no consequence, and we have
_ Weight of sample recovered
et
Initial volume of piston sample
The unit weight is necessary to compute the in situ overburden pressure po used to es-
timate OCR and is necessary in the computation of consolidation settlements of Chap. 5.
It is also used to compute lateral pressures against soil-retaining structures and to estimate
skin resistance for pile foundations. In cohesionless materials the angle of internal friction
<f> depends on the unit weight and a variation of only 1 or 2 kN/m 3 may have a substantial
influence on this parameter.
Relative Density Dr
Relative density is sometimes used to describe the state condition in cohesionless soils. Rel-
ative density is defined in terms of natural, maximum, and minimum void ratios e as
Dr = *max_~ €n (2-15)
^max ^min
It can also be defined in terms of natural (in situ), maximum, and minimum unit weight y as
7n
Dr = ( ~ 7min V ^ ) (2-16)
\7max - y m i n / V Y* /
The relative density test can be made on gravelly soils if the (—) No. 200 sieve (0.074 mm)
material is less than 8 percent and for sand/ soils if the fines are not more than about 12
percent according to Holtz (1973).
The relative density Dr is commonly used to identify potential liquefaction under earth-
quake or other shock-type loadings [Seed and Idriss (1971)]; however, at present a somewhat
more direct procedure is used [Seed et al. (1985)]. It may also be used to estimate strength
(Fig. 2-30).
It is the author's opinion that the Dr test is not of much value since it is difficult to obtain
maximum and minimum unit weight values within a range of about ±0.5 kN/m 3 . The average
maximum value is about this amount under (say 20.0 kN/m 3 - 0.5) and the minimum about
this over (say, 15.0 kN/m 3 + 0.5). The definition is for the maximum and minimum values,
but average values are usually used. This value range together with the uncertainty in obtain-
ing the in situ value can give a potential range in computed Dr of up to 30 to 40 percent (0.3
to 0.4). Chapter 3 gives the common methods of estimating the in situ value of Dr. A simple
