Dynamic Bearing Capacity
Dynamic Bearing Capacity
Dynamic Bearing Capacity
CHAPTER 2
2.1
at either the ground surface or at a certain depth below the ground surface that is
necessary to mobilize the full shear strength of the soil along some continuous failure
surface. The critical load depends not only on the mechanical properties of the soil but
on the size and shape of the footing. The problem of evaluating the critical static load
has been treated by number of researchers, for examples, Terzaghi,
1943; Meyerhof,
1951; Vesic, 1973, but a mathematically rigorous solution has been attained only under
greatly simplified assumptions.
In comparison with the extensive studies on the static bearing capacity of shallow
foundations,
dynamic bearing capacity of soils. Where the latter information is available, the analyses
and experimental investigations deal essentially with vertical vibrations as in machine
foundations. Very little is available in literature describing studies of the bearing capacity
of soils under earthquake type loads.
It was reported in literature that many buildings failed during earthquakes by soil
bearing
capacity failure.
to the liquefaction
28
phenomenon of the soil- a condition where the mean effective stress in a saturated soil,
essentially loose saturated cohesion less soil, is reduced to zero. For example, during the
Niigata, Japan, earthquake of June, 1964, apartment buildings at Kawagishicho were
tilted, some by as much as 60, yet there was almost complete absence of other
(structural) damage. This indicates that the bearing capacity of the soil was reduced to
a very low value.
Relatively few cases of foundation failures in partially saturated and/or dense sand
have been reported in the literature during seismic events. An example of such a failure
occurred during the Miyagiken-Oki earthquake of magnitude 7.8 on June 12, 1978,
northeast of Sendai, Japan, where the foundations of several oil storage tanks suffered
from bearing capacity failure and excessive settlement (after Okamoto,
1984). The
subsoil for the oil storage tanks consisted of a layer of fine sand 65 m thick, which had
been densified by vibrio-flotation prior to the construction of the tanks. Ishihara et. al.,
(1980) reported little or no damage to similar nearby tank facilities where the sand was
compacted using compaction piles. Other failures of oil tank foundations on partially
saturated, compacted sands during the Niigata earthquake in 1964 were reported by
Watanabe (1966).
During a series of shaking table tests on flexible retaining walls, Richards et. al.,
(1990) observed that at certain levels of horizontal acceleration the dry, dense C-109
Ottawa,
- defined as
phenomenon whereby the sand behaves as a viscous fluid. The fluidization phenomenon
29
was noticed to be initiated at moderate horizontal acceleration, less than 0.3 g, where
g is the acceleration of earth's gravity. Model footings of various sizes, shapes and
weights were placed on a layer of dry, dense, C-109 Ottawa sand in a test box (1 m x
1 m x 5 m long) and was subjected to horizontal excitation on a shake table at the State
University of New York (SUNY) to determine movements of the model footings at the
fluidization stage. It was found, that, at various acceleration levels below 0.5 g, all these
footings moved vertically. For example, a cylinder (178 mm in diameter) was found to
settle about 6 mm at a peak horizontal acceleration of 0.3 g as shown in Fig 2.1a. At
higher level of acceleration 0.5
<
kh
<
and the footing settled more than 25 mm as if it was in a viscous fluid (Fig. 2.1b). A
wooden box which was partially submerged in the sand was found to rise about 6 mm
as shown in Fig. 2.1 c. This study by Richards et. al., (1990) shows that the bearing
capacity of the dry loose sand can be reduced at even moderate horizontal acceleration
level. Richards, et. al., (1990) used Mohr's circle representation to show the effects of
the induced inertia forces on soil behavior.
2.1.1
loads. Triandafilids (1965) presented a solution for the dynamic response of continuous
footings supported on saturated cohesive soil
(et>
30
(a)
23'
Time {sec}
1.0 rO.7Bg .
Cl
.:t
0.5
.~ as
1:
u 0.5
~
k'11
1.0
1.0
Slope ,
"'.
(b)
Fig. 2.1.
I \}
12.Smm/.sec
"l'"
2
... '26mm
Time (secJ
(C)
rime fsec)
31
loads. By assuming that the soil behaves like a rigid plastic material and the slip surface
IS
circular, he computed the maximum angular rotation the footing might undergo.
Prakash and Chummar (1967) presented an analysis for determining the response
of continuous foundations resting on a c-<p soil and subjected to transient horizontal loads
applied above the level of the footing base. The geometry of the failure surface is
assumed to be a logarithmic spiral with its center at the corner of the footing base. This
study shows that the maximum horizontal load that can be tolerated before failure occurs
is equal to 0.38 of the vertical static load. The maximum angular rotation that the footing
might undergo is variable and depends on the soil properties
(c,<p)
width (B).
Sarma and Iossifelis (1990) presented a study on the seismic bearing capacity of
shallow footings due to horizontal accelerations.
and Iossifelis (1990) analysis is similar to that presented by Sokolovski (1962) for spread
footings under an applied static inclined load. There are two differences between the
analysis of Sarma and Iossifelis (1990) and the analysis of Sokolovski (1962). Firstly,
Sarma and Iossifelis (1990) added the inertia force of the soil mass to the forces used in
32
calculating the soil bearing capacity factor due to the soil weight (N,,). Secondly, they
used the limit equilibrium technique of slope stability with inclined slices (Sarma, 1979)
to evaluate the bearing capacity factors. The results of Sarma and Iossifelis (1990)
analysis for
NqE
was found to agree with the values determined by Meyerhof (1953), and
N-yE
NeE
the soil mass was accounted for by Sarma and Iossifelis (1990). The changes in the
active and passive earth pressures due to the applied acceleration by the seismic loadings
could reduce the seismic bearing capacity factors to lower values than those due to the
load inclination. Sarma and Iossifelis (1990) did not consider such an effect. Also, they
did not consider the effect of the vertical acceleration on the bearing capacity of their
granular soil.
Richards, et. al., (1991, 1993) developed charts for the variation of the ratio of
seismic bearing capacity factors to their static counterparts with horizontal accelerations.
Richards, et. aI., (1993) used a Coulomb type mechanism and approximated the failure
surface by wedges similar to Rankine wedges for static case (Fig. 2.2). The effect of the
inertia forces of the vertical load and the overburden pressure as well as the soil cohesion
was not considered in these analysis.
33
\\\\\\\\
\\\\\\\\
(1973)
Richards et. al.. (1993)
Fig.2.2.
Comparison between the failure surface of Vesic (1973) and Richards, et.
aI., (1993).
34
2.1.2
of shallow foundations. Selig and Mckee (1961) investigated the behavior of a number
of small footings resting on dense uniform Ottawa sand and subjected to vertical
impulsive dynamic loads as well as static loads. In their conclusion, they stated that"
for the static tests in dense sand the classical general shear failure occurred. fJowever,
for equal or greater total settlement under the dynamic load an entirely different mode
of sand distortion occurred. The mode was very similar to that of the local shear failure
that occurs for the same footing loaded statically in loose sand. Only after the third
repeated impact on the same footing was there any evidence of formation of a well
defined shear plane for the dynamic tests. And in these cases the total settlement was
quite large (about 75 % of the footing width)."
Carroll (1963) conducted unconsolidated-undrained
the effect of strain rate on undrained cohesion of clay. Carroll suggested that the
dynamic undrained cohesion (Cu(dyn)
at high strain rate (
> 50%/sec)
35
initial stress is zero, the strength of both soils under a single dynamic stress pulse (N = 1,
where N is the number of cycles) was seen to be considerably greater than the static
strength by factor of 1.4 for the Bay mud and 1.52 for the compacted silty clay. This
results agree with the results of Carroll (1963). Under the effect of cyclic loading, the
strength for the San Francisco Bay mud and the compacted siIty clay were found to be
in the range of 0.8 to 1.0 and 1.0 to 1.2 greater than the static strength, respectively.
Taylor (1968) stated that" when a soil is repeatedly loaded, it may fail after a
number of cycles at a combined (static and dynamic) stress considerably less than the
failure stress observed in a conventional test." Based on the strength characteristics
of
cohesive soil under dynamic loading, Taylor (1968) outlined an experimental method for
checking that the cohesive soil will not fail under the effect of repeated cyclic loadings
at the chosen static and dynamic safety factors.
1.
2.
3.
(CT/q)
(CT/q),
(.:lCT/q)
CT
where
= l/fe-1/f"
(.:lCT/q),
where
.:lCT
Check the stability of the design due to the chosen load factors (fs and fe)
by a proof test in which the soil is subjected to cyclic loads equal to the
calculated loads above with more than 50 cycles.
36
5.
if failure does not occur after 50 cycles of dynamic loading then the
dynamic load factor (fd) must be at least 2. If failure occurs in the test,
the desired value of fd has not been attained, and the preliminary design
must be revised.
This method is based on laboratory testing and therefore the testing setup must
be available for the designer to run the required tests. Also, it is costly and time
consummg.
Vesic et. al., (1965) studied experimentally the dynamic bearing capacity of a
sandy soil under strip footings loaded vertically with different rate of transient loads, All
tests were performed using a circular, rigid, rough plate, having a diameter of 101.6 mm
and resting on the surface of homogeneous dense sands. The sand bed was prepared in
a watertight steel box 127 cm square and 178 cm deep. No shaking was applied to the
sand bed except that caused by the transient load. They showed that the bearing capacity
was about 30 percent lower than the static bearing capacity when the footing was
subjected to dynamic vertical loads. Vesic et. al., (1965) suggested that the reduction in
strength is equivalent to a decrease in the angle of internal friction of about 2 degrees.
The effects of overburden pressures and soil cohesion were not considered in this study.
Okamoto (1956) used a box, 15 cm width x 60 cm length x 30 cm depth,
containing sandy soil which he shook horizontally on a shake table to determine the
seismic bearing capacity of shallow footings on a dense sand with angles of friction
ranging from 38 to 50. The footing was simulated by a rectangular test plate, 6.1 cm
37
x 15.0 cm, and loaded vertically. As the box was shaken horizontally at a maximum
acceleration of 0.38 g for 1.3 seconds, the test plate settled into the sand. When the test
plate was loaded obliquely, the test plate slid horizontally due to shear failure of the soil
beneath the plate before settlement begins. The bearing capacity was found to decrease
linearly with increasing horizontal accelerations for both dry and saturated sand. These
test results were compared with a theoretical solution which consider the resultant of the
weight and the inertia forces of the footing as inclined load applied to the soil at an angle
equal to tan-1(kh) from the vertical axis. Good agreement between the theoretical and
experimental results were found. The theoretical solution does not consider the inertia
force of the soil, the surcharge loading, and the cohesion of the soil.
2.2
SOIL SETTLEMENT
For cohesionless soil, most of the settlement occurs by the time the construction
IS
complete.
densification and the development of soil fluidization. The available studies on each type
of these settlements are presented below.
2.2.1
IMMEDIATE SETTLEMENT
The immediate settlement for cohesionless soils is assumed to be completed by
the end of the construction period. In many investigations, different methods have been
reported for calculating the immediate settlement. Some of these methods were derived
38
using the theory of elasticity, while others are based on correlation between settlement
and Standard Penetration Tests or Cone Penetrometer Tests. The most common methods
used in practice are listed in Table 2.1.
2.2.2.
conducted using vibratory compactors, (Bernhard (1952), D' Appolonia, et. al. (1969),
Moorhouse and Baker (1969), Converse (1953. Because cohesion less soils are known
to densify due to accelerations, many investigators tried to simulate field conditions to
study this problem by mounting small containers filled with sand on vibrating tables.
D' Appolonia and D' Appolonia (1967) conducted tests on air-dry sand using a
vibrating table apparatus. The sample is contained in a mold 10.2 cm in diameter and
8.9 cm high. Tests were conducted by placing air-dry sand in the mold, applying the
surcharge and amplitude and frequency of vertical vibration until the volume of the
sample in the mold remained constant over the last 20 minutes of the test. From this
study it was found that an increase in the surcharge load reduces t~e maximum density
obtainable by vibration and an increase in acceleration higher than the peak acceleration
of vibration can cause a decrease in compacted density. At accelerations less than 1.0
g very little densification was shown.
Whitman and Ortigosa (1968) conducted a similar investigation
D' Appolonia and D' Appolonia (1967) and concluded that:
to that of
)l+B
Table 2.1.
Based on relation
the
theory
of elasticity,
where E,
isof
A correlation
between
allowable
Empirical
between
the
Spbearing
Calculation
formula
Remarks
penetration
tests
(N), settlement
and footing
width
(B).
a functionwhere
of SPT
blowbecounts
and I is
Elastic
E, can
estimated
pressure (P), from
blow N
count
value
from
usingstandard
empirical
formula
suchtaken
as asequation,
S=-W[1-J.L2]
of width
(ft)+ for
identical
load foot).
the=influence
E,
10B[N
15]
factor.
ksf (kips applied
per square
S=8P footing
pressure.
s= qBI[1_J.L2]
the plate and the settlement of a square
Author
N
for B$;4
B>4 B+l
It
for
It
S= 12P[~r
S=S~~r
1-
B+l
i=l 0.1
Es w.
F. f1q d
"n
210g(-t-)
C "1 -o.s(-('~1
Author
L
1
(1 +0.4 ~)
:-r
Author
qF
2BLxB D
footing
5]
{~r
applyingbased
correction
for footing
size, for different type
2
developed
by Terzaghi
This method
on table
loadfactors
displacement
square plate on the surface of a sand layer and
B Remarks
BL
q=K)that can be obtained from
I
subgrade reaction
sand,and
anddepth.
L is the
Iof
Author
shape,
Where
I
KsCalculation
islength
the modules
relationships
obtained
by Iload
testing
1 formula
ft of
-footing
(atD+-)
(at-) B
where m=-ay 2
BxB
0 ay 1.5m
K =K
43
1.
2.
when the dynamic stresses are small compared to the initial overburden
pressure, there is still no noticeable densification.
3.
vertical
accelerations
during
an
earthquake
produce
very
little
densification.
Barkan (1962) studied the effect of horizontal vibration on soil densification by
using a direct shear apparatus mounted on vibrating table. He developed a functional
relationship between void ratio and horizontal acceleration coefficient (~,) for all soils
densified by vibration. Barkan approximate this relationship between void ratio and ~
by,
(2.1)
In which
emin
and
em3X
on soil
densification by mounting the base of a direct shear test on a horizontal vibrating table
with a variable frequency (0 cps to 60 cps) and amplitude (0 cm to 0.191 cm). The study
was conducted on Ottawa C-109 sand and steel balls. The results of this study confirm
the existence of the relationship in equation (2.1) found by Barkan (1962).
44
Youd (1970), uses the same testing configuration as Youd (1968), studied the
influence of the normal pressure on the parameter ((3) in equation 2.1 at acceleration
levels greater than 0.5 g. The parameters
emin, emm
of the normal pressure and the material properties. The variation of these parameters
with the applied normal pressure was given in tables by Youd (1970). In agreement with
the observation of Barkan (1962), horizontal acceleration was found to cause considerable
soil densification.
Timmerman and Wu (1969) used a triaxial equipment for dynamic testing to
study the relationship between sand deformation and stress condition. The tests were
conducted on Ottawa Sand, a fine Ottawa Sand, and
normal and radial stresses were applied at frequencies of 2.5 Hz, 5 Hz, 10 Hz, and 25
Hz. The acceleration was less than 0.5 g. Most of the specimens were subjected to
100,000 load cycles. The effect of cyclic stresses on the sand densification was found to
be more important than the effect of the acceleration. D' Appolonia (1970) also indicated
that the repeated cyclic stress for acceleration lower than 1.0 g is the major cause for soil
densification.
Silver and Seed (1971a) investigated the soil densification due to cyclic shear
strain in the range of 0.01 % to 0.5 %. Static direct simple shear test equipment developed
by the Norwegian Geotechnical Institute (NGI) was modified so that cyclic horizontal
shear stress could be applied. This investigation showed that the settlement due to soil
compaction is not significantly affected by values of vertical static stress and depends
45
only on the shear strain amplitude induced in the sample with cyclic shear strains
exceeding
0.05
%.
Seed and Silver (1972) proposed that the settlement of sand during earthquakes
is produced mainly by the horizontal components of ground shaking where in most of
cases the acceleration levels involved will be substantially less than 0.5 g. Seed and
Silver (1972) outlined a procedure for estimating the possible amount of this settlement
based on the results of previous studies by Silver and Seed (1971a, 1971b). These
procedures as outlined by Das (1983) are:
1.
2.
Divide the soil layers into n layers. They need not be of equal thickness.
3.
4.
CTz
5.
6.
Convert the irregular strain histories obtained for each layer (Step 5) into average
shear strain and equivalent number of uniform cycles.
7.
soil
specimens from each layer to obtain the vertical strains for the equivalent number
of strain cycles calculated in Step 6. This has to be done for the average effective
vertical stress levels
8.
CTz
calculated in Step 6.
46
n
tlH=
where
Ez(i)
L
i=l
(2.2)
EZ(i)Hi
is the average vertical strain determined in Step 7 for layer i and Hi is the
layer thickness. The calculated settlement by this procedure for Jensen Filtration Plant
of the Metropolitan Water District of Southern California located in the Granda Hills area
of the San Fernando Valley was found to be 33 mm. When this filtration plant was
shaken by the San Fernando Earthquake of Feb. 9,1971 with maximum acceleration of
0.6 g, the observed settlement was found to be between 89 mm to 102 mm. The
computed settlement is about one third the observed one. It has been suggested by Pyke,
et. al., (1975) that the major cause of the discrepancy between the computed and actual
settlements is that soils are subjected to multidirectional rather than the unidirectional
shaking under actual earthquake conditions.
Pyke, et. al., (1975) studied the effect of multidirectional
settlement. They reported from studies using a shake table that the effect of the second
component of motion has the same effect as the first component, and the computed
settlement caused by one component might be approximately double to account for the
effect of shaking in different horizontal directions.
Further,
47
Seed (1971) method was found to be 33 mm. It was assumed that the second component
of motion was similar to the one used and then the computed settlement was double (66
mm) to account for the effect of shaking in varying horizontal directions. The total
settlement was equal to the settlement due to the horizontal components (66 mm) plus
50% of that due to the vertical component (33 mm) to become 99 mm, which compares
much more favorably with 88 mm to 102 mm settlement that has been attributed to
compaction caused. by the earthquake.
Timmerman and Leelanitkul (1980) investigated experimentally the progressive
axial strain of Ottawa sand samples of various relative densities caused by vibrating
foundations.
shearing strain amplitudes between 10-5 and 10-2 . The data show a linear log-log
correlation
between the measured progressive axial strain and the calculated cyclic
shearing strain amplitude. The predicted deformations using the triaxial data was found
to agree well with data from other research using cyclic simple shear data like that of
Seed and Silver (1972). Leelanitkul (1980) outlined the procedure for predicting the
progressive settlement in cohesionless layer subjected to vertical vibrations using triaxial
results.
2.2.3
FLUIDIZATION SETTLEMENT
Richards, et. al., (1990) showed that significant settlement of shallow footings can
occur due to soil fluidization independent of the initial void ratio. The amount of this
48
settlement is based on the peak acceleration as well as its duration. Richards, et. al.,
(1993) have suggested that the pattern of foundation settlement can be predicted by the
sliding-block model proposed for gravity retaining walls by Richards and Elms (1979).
Because of the differences between the movements of gravity retaining walls and shallow
footings, this technique may not give a reasonable prediction
shallow footings . For example, a gravity retaining wall will move horizontally along the
ground surface when driving forces due to the horizontal inertia force of the wall and
the active force of the soil exceeds the shear force at the base of the wall. However, the
foundation will move vertically inside the soil due to soil bearing capacity failure.