Dynamic Soil Properties 2015 PDF
Dynamic Soil Properties 2015 PDF
Dynamic Soil Properties 2015 PDF
References:
1. Kramer, S.L., Geotechnical Earthquake Engineering, Prentice Hall,
1196. Chapter 6.
2. Bozorgnia, Y., and Bertero, V.V., Earthquake Engineering, from
Engineering Seismology to Performance Based Engineering, CRC
Press, 2004, Chapter 4.
3. Desai, C.S., and Christian, J.T. Numerical Methods in Geotechnical
Engineering, McGraw Hill, 1976, Chapters 2 and 3.
4. Das, B.M., Principles of Soil Dynamics, 2nd Ed., CENGAGE Learning,
2011
5. Luna, R., and Jadi, H. “Determination of Dynamic Soil Properties
Using Geophysical Methods,: Proc. 1st Int. Conf. on the Application
of Geophysical Methods and NDT Methodologies to Transportation
Facilities and Infrastructure, St . Louis, MO, 2000.
Important Dynamic Properties of Soils
• Cyclic Shear Modulus (G), Damping (, ), and
Poisson’s ratio ()
• Cyclic Strength, undrained shear strength of
sands and silts; cyclic shear strength of clays
(softening)
• Liquefaction properties
• Seismic compression of compacted soils
Shear Modulus, G
• Shear strain () dependent, highly non linear
• Early models are equivalent elastic.
• Recent models are non linear, viscous plastic.
Methods to measure Shear Modulus
• In Situ
‐ Seismic Reflection Test
‐ Seismic Refraction Test
‐ SASW Test
• Laboratory Figure 6.8 Different methods for creation of impulsive disturbances for
seismic geophysical tests : (a) shallow explosive; (b) vertical impact; (c)
‐ Resonant Column Test horizontal impact
‐ Cyclic Triaxial Test
‐ Cyclic Simple Shear Test
‐ Shaking Table Test
• Empirical Correlations
‐ SPT
‐ CPT
‐ Dilatometer
‐ Pressuremeter
Important to note for Linear Models
• Equivalent linear models are only approximate
• Can not be used for permanent deformation
or failure analysis
• Still used as it provides efficient computation.
Geophysical Seismic Tests
a. SV‐waves, in which the motion of the particles is in the plane of propagation as shown by the
arrows in Figure 4.37b
b. SH‐waves, in which the motion of the particles is perpendicular to the plane of the
propagation, as shown by a dark dot in Figure 4.37c
Since incident P
and SV waves
involve particle
motion
perpendicular to
the plane of
interface, there
will be both
reflected and
refracted P and
SV waves
Geophysical Survey Test
• Measures G (and ) by measuring wave
velocities and attenuations.
• P waves travel at 1700 m/sec in water,
whereas medium‐soft/loose soils generally
have a vp of less than 1500 m/sec. Thus is soft
saturated soils, the P‐waves basically travels
through the water.
• S‐wave does not have this problem.
Seimic Reflection Test
• Useful for large scale &/or deep stratigraphy,
not useful for shallow layers
By measuring t1 and knowing x and vp1 from the direct wave calculation, the thickness of the
upper layer can be calculated as
(6.7)
Inclined Layering
The characteristics of deeper layers may be evaluated using reflections from deeper interfaces
(Griffiths and King, 1965; Ewing et al., 1957; Kleyn, 1983). The method is limited by the difficulty
associated with determining the arrival time of the reflected waves, particularly for cases in
which reflected waves arrive while the receivers are still responding to direct waves.
Interpretation of results for profiles with low‐velocity layers may also be difficult.
for equal distance geophones
x x
A B From Snellius:
Note: xA has a ‐ sign z = z’
Geometry
for small :
CA + CB = 2 z
C
Seismic Refraction Test
Snell’s Law
ii ic
90o
ir
Figure 6.12 Wavefronts for first‐arriving waves in a seismic refraction survey. Note that
first arrivals near the source are from direct waves, but at distances greater than the
critical distance,xt, the first arrivals are from head waves. (After Corps of Engineers,
1979.)
Figure 6.13 Travel path for first arrival
when xn > xc*
(6.12)
Substituting the result of Snell’s law for critical incidence, sin ic = v1/v2, and the trigonometric
identity cos2ic =1 ‐ sin2ic , and rearranging yields
(6.13)
If a receiver was placed exactly at the critical distance, xc, the direct wave and the head wave
would reach it at exactly the same time (i.e., tdn=thn). Consequently, from equations (6.11) and
(6.13)
(6.14)
From which
(6.15)
For the case of multiple horizontal layers, the travel time‐distance diagram will exhibit more than
one break in slope, as illustrated in Figure 6.14. The distances corresponding to these slope
breaks can be used, along with the slopes themselves, to determine the thickness of deeper
layers. The thickness of the kth layer, for example, would be given by (Corps of Engineers, 1979)
(k≥2) (6.16)
The velocity of each layer is smaller than the layer immediately below it. For many geologic
conditions this is a good assumption, but the results of a seismic refraction test can also be
misleading. A low‐velocity layer underlying a higher‐velocity layer (i.e., a velocity reversal) will not
appear as an individual segment on the travel time‐distance diagram. Instead, it will cause the
computed depths of the layer boundaries to be greater than the actual depths (Redpath, 1973).
Also, blind zones, where a subsurface layer exists but is not indicated by the travel time‐distance
diagram, can be caused by insufficient layer thickness or insufficient velocity
Figure 6.14 Travel time‐distance diagram for multiple
horizontal layers. (After Redpath, 1973.)
In such cases the head wave from a deeper layer can overtake the head wave of an intermediate
layer before it reaches the ground surface. The undetected existence of a blind zone will cause
the computed depth of the deeper layer to be less than the actual depth (Redpath, 1973). In
cases where the velocity increases continuously with depthm ray paths will be curved rather than
straight. Where velocity is proportional to depth, the ray path will become circular arcs.
Inclined Layering
Figure 6.16 Reverse profiling used to identify irregular or
sloping interfaces. Concept of apparent velocity in
underlying layer is also illustrated. (After Redpath, 1973.)
Irregular Layering
Note:
1. TD1 and TD2 need not be on a
curve intersection
2. Each curve’s slope must be past
v1 (already receiving head
waves).
(6.29)
This technique does not require boreholes and is another in‐situ method used to measure the
shear modulus (G) of all types of soils. In this test, an electromagnetic oscillator at high frequency
930 to 1000 cycles/second, cps) or a rotating mass type oscillator to produce low frequency
vibrations (less than 30 cps) are used. The shear wave velocity is computed from the Rayleigh
wave‐length measured with receivers placed along the ground surface, and the frequency of
vibration at the source using the following equation (SW‐AJA, 1972; Gazetas, 1991):
Vs –VR = fλR (4)
Where,
f = Frequency of Vibration and λR = Rayleigh wave length
The effective depth of R‐wave has been empirically related to the soil layer at a depth equal to
one‐half the wave‐length, λR (Heukelom and Foster, 1960, Fry, 1963 and 1965, Ballard 1964).
Spectral Analysis of Surface Wave Test
(SASW)
• Uses the shape of a dispersion curve [plot of
Rayleigh wave velocity vs. frequency or wave length]
• Performed using 2 vertical receivers on the ground in
line with an impulsive or random noise source.
Figure 6.19 Typical configuration of source and receivers in a SASW test. Receiver spacing is changed in such a way that d1 + d2
remains constant.
• Receiver records are transformed into the frequency domain using FFT. [Matching the
records with a series (with different frequencies) of Rayleigh waves propagation between
receivers].
• From FFT the phase difference, (f) for each frequency, f, is obtained.
The corresponding travel time between receivers can be calculated for each frequency from
(6.27)
Since the distance between receivers, Δd= d2‐d1, is known, the Rayleigh wave phase velocity and
wavelength can be calculated as functions of frequency :
(6.28)
(6.29)
Dispersion Curve
vr and thickness of layers are
determined by matching theoretical
curves with measurements.
Inversion method after Haskel [1950]
and Thomson [1953] solution is an
example theoretical solution for
horizontal elastic layers.
For profiles with vr varying with
depth, higher modes Rayleigh waves
should be considered [Gucunski &
Woods, 1991; Tokimatsu, 1992]
Figure 6.20 Experimental dispersion curve from
SASW test. (After Gucunski and Woods, 1991; used
by permission of the University of Missouri‐Rolla.)
SASW Procedure
The two signals in time domain (y1(t) and y2(t)) from Figure 4.7 are translated in frequency
domain using a Fast Fourier Transform, obtaining the related linear spectra (Y1 (ω) and Y2 (ω)).
Using spectral analysis technique, it is then possible to get information about the quality of the
records and eventually the phase velocity as a function of frequency.
Figure 4.7 Example of SASW signals : (a) whole signals; (b) wave‐train arrivals. (Site : ENEA; source:
6kg hammer; d=2m)
The assessment of signal quality is made using the “coherence function”, namely a spectral
quantity obtained comparing different registrations, that is a measure of the degree by which
input and output signals are linearly correlated. A value close to unity is an index of good
correlation and hence the recorded signals can be considered genuine and unaffected by ambient
noise.
Phase velocity as a function of frequency can be obtained from the phase of the average Cross‐
Power Spectrum. The quantities involved in the evaluation of the dispersion curve, in the
succession according to which they are evaluated, are the following
• Auto‐power spectra (Figure 4.8c,d)
(4.3)
(4.4)
• Cross Power Spectrum
(4.5)
• Phase of Cross Power Spectrum (4.8a)
(4.6)
Coherence function (Figure 4.8b)
(4.7)
Figure 4.8 Spectral quantities evaluated from the signals of Figure 4.7 : (a) Phase of cross‐power spectrum; (b) Coherence
function; (c) Auto‐power spectrum (first receiver); (d) Auto‐power spectrum (second receiver)
• Time delay between the receivers :
(4.8)
• Phase velocity of Surface waves :
(4.9)
where D is the distance between the two receivers.
• Wave length :
(4.10)
The use of the cross‐power spectrum phase for the evaluation of the frequency dependent time
delay is based on the hypothesis that the wave group under consideration is composed of a single
mode of propagation and hence the phase velocity is function only of frequency (Aki and
Richards 1980). Under such assumption the signal that is associated to a wave propagating along
the x direction can be expressed as a superposition of harmonic waves of the same mathematical
form of that propagating in a homogeneous half space (Equation 3.9)
(4.11)
Where y(x,ω) is the amplitude spectral density, k(ω) is the wave number and φ(ω) is a constant
phase term due to effects other than propagation. The Fourier transform of such signal is :
Hence considering the cross‐power spectrum relative to two different detection of the same
wave along the x‐axis at location x1 and x2 :
(4.13)
And recalling that
(4.14)
Equation (4.8) can be obtained with few manipulations.
SASW Advantages/Disadvantages
• Quick test, no borehole needed.
• Can detect low velocity layers
• Can reach large depths, > 100 m.
• Measured vs compares well with cross hole
results.
• Applicable up to gravels and debris flow deposits.
• Require special equipments and highly
experienced operators
Seismic Cross Hole Methods
Figure 6.21 Seismic cross‐hole test : (a) direct measurement using two‐hole
• Source and receivers placed within the same depth of each layer. Source and receivers
need to be fixed on borehole walls. Shear energy rich source is preferred for easy
detection.
• Test at various depths. Receivers shall be oriented in the particle motion direction.
• Observe arrival times of common phase, i.e. first arrival, first peak, first trough etc., is
easily noticeable
Impulse sources
• Explosives create P and S waves
complicated.
• Mechanical Impulse:
‐ vertical (e.g. SPT device)
‐ torsional (torsional impact on rod)
‐ best if polarity of source is reversible
Figure E6.5
Particulars
• Can test individual layers (most layers are nearly
horizontal).
• Reliable up to 30 to 60 m depths for mechanical impact
sources and more for explosives
• May require borehole deviation survey (layers may not be
accurately estimated)
• May not measure accurate velocity if high velocity layers
exist nearby. In such case, need more advanced
interpretation method to account refraction.
• To estimate damping, 3 or more bore hole attenuation
records can be used, if geometry is properly defined,
radiation damping can be separated from material
damping.
Seismic Down Hole (Up Hole) Test
• Can be performed in a single borehole easier & cheaper
• Receivers can be placed at various depths.
• S wave is generated easier in down hole test
• Reverse polarity easily achieved , by applying the impulse in opposite directions
easy to separate P from S wave
Particulars
• Up hole test measures wave velocities in a similar way
the seismic wave is carried to the ground surface.
• Down hole test allows detection of layers that can be
hidden in refraction surveys.
• Low velocity layers can be detected by close spacing
the receivers in a down hole test.
• Potential interpretation difficulties due to hole boring
disturbance, casing and fluid effects, insufficient or
excessive impulse source, back ground noise and
ground water effects.
• Material and radiation damping makes it difficult to
measure depths of more than 30 to 60 m.
Figure 6.23 Travel‐time curve from down‐hole test in San Francisco Bay area. (After Schwarz and
Musser, 1972.)
Seismic Cone Penetration Test (S‐CPT)
• Re. Down hole Test
Cross hole S‐CPT test can
be done using 2 sets of
seismic cones [Baldi,
1988].
Field Methods
• Use empirical correlations from field
measurements with G and (details later)
• Includes: SPT, CPT, Dilatometer, Pressure
meter
SPT
• (N1)60 is needed:
Nm measured
Em actual hammer energy
CN overburden correction
Eff = hammer weight x
height of fall Figure 6.25 SPT overburden correction factr.
(After Liao and Whitman, 1986.)
Dilatometer Test
• Similar to this is LLT (lateral load test).
• Test intervals typically 10 – 20 cm penetration
• Lift off pressure (p0 ,contact) @ 0.05 mm
membrane movements
• Pressure @ 1.1 mm (p1) membrane movement
is recorded
• Hydrostatic pore pressure, uo, is recorded
• Corrected for p0, and vo’.
• Dilatometer Test Results:
‐ Material Index:
‐ Horizontal Stress Index:
‐ Dilatometer Modulus:
= 34.7 for 2.4” membrane diameter &
1.1 mm deflection
• Correlated with G, and liquefaction
resistance
Pressure meter test
• Measures stress‐strain and strength
• Using Cavity Expansion theory, the stress‐
strain and shear strength parameters.
• Disturbance is less for self boring pressure
meters or push in pressure meters.
Laboratory Tests
• To analyze behavior of soil parameters to
major influencing factors, e.g. strain
amplitude, frequency, confining pressure, etc.
Resonant Columns Test
• For Low Strain Dynamic Properties
• Specimen, hollow or solid cylindrical (same
size as in Triaxial)
• Electro magnetically applied harmonic
(frequency and amplitude controlled)
torsional or axial loading. Random and
impulse loading can also be applied.
• Loading starts at lower frequency, then
increased to the natural frequency of the
system (sample and apparatus).
• Within the specimen, the torques resistance, T, is:
J is polar moment inertia of specimen
I is mass polar moment inertia of specimen
• The loading apparatus produces a torque, T, according to:
Io is mass polar moment inertia of apparatus
• Assuming harmonic rotations
• Rotation at base (z = 0) is zero, and torque force equality, we
get:
Where fn is the fundamental frequency in hertz. In this case the rotations would follow a quarter‐
sine‐wave pattern over the height of the specimen at the fundamental frequency. Adding the
mass of the loading system results in more linear variation of rotation and, consequently, more
uniform strain conditions over the height of the specimen.
For longitudinal loading, the analogous equation is
(6.38)
Where W is the weight of the specimen, W0 the weight of the loading system, and
the longitudinal wave propagation velocity.
If the loading system was massless (I0 = 0), equation (6.37) would degrade to
Cyclic Triaxial Test
Figure 6.33 Time histories of deviator stress and stress paths for (a) isotropically
consolidated conditions, (b) anisotropically consolidated conditions with cyclic
deviator stress amplitude greater than deviator stress during consolidation
(producing stress reversals), and (c) anisotropically consolidated conditions with cyclic
deviator stress amplitude less than deviator stress during consolidation (no stress
reversals)
Anisotropically consolidated triaxial tests are performed to model conditions in and beneath
slopes where initial static shear stresses exist. Figure 6.33b refers to an anisotropically
consolidated specimen for which the cylic deviator stress amplitude is greater than the deviator
stress during consolidation. Stress reversals also exist in this situation, even though the cyclic
deviator stress is no longer symmetric about the ρ‐axis. If the cyclic deviator stress.
Figure 6.34 NGI cyclic simple shear apparatus. Soil specimen is contained within wire‐reinforced
rubber membrane. (After Airey and Wood, 1987.)
The simple shear apparatus, however, applies shear stresses only on the top and bottom surfaces
of the specimen. Since no complementary shear stresses are imposed on the vertical sides, the
moment caused by the horizontal shear stresses must be balanced by non‐uniformly distributed
shear and normal stresses. The effects of non uniformity of stresses can be reduced by increasing
the diameter/height ratio greater than about 8:1 (Kovacs and Leo, 1981). Conventional simple
shear apparatuses are limited by their inability to impose initial stresses other than those
corresponding to Ko conditions. Pyke (1973), used a large shaking table to produce a two
directional large‐scale simple shear apparatus with a diameter/height ratio of about 9 in one
direction and 20 in the other.
Cyclic Torsional Shear Test. Many of the difficulties associated with the cyclic triaxial and cyclic
direct simple shear tests can be avoided by loading cylindrical soil specimens in torsion. Cyclic
torsional shear tests allow isotropic or anisotropic initial stress conditions and can impose cyclic
shear stresses on horizontal planes with continuous rotation of principal stress axes. They are
most commonly used to measure stiffness and damping characteristics over a wide range of
strain levels.
Ishihara and Li (1972) developed a torsional triaxial test that used solid specimens. Dobry et al.
(1985) used strain‐controlled cyclic torsional loading along with stress‐controlled axial loading of
solid specimens to develop a CyT‐CAU test has proven effective for measurement of liquefaction
behavior. Torsional testing of solid specimens produces shear strains that range from zero along
the axis of the specimen to a maximum value at the outer edge. To increase the radial uniformity
of shear strains, others developed hollow cylinder cyclic torsional shear apparatuses while hollow
cylinder tests offer perhaps the best uniformity and control over stresses and drainage, specimen
preparation can be difficult and equipment is not widely available.
Figure 6.35 hollow cylinder apparatus. The specimen is enclosed within internal and external
membranes on which internal and external pressures can be applied independently. Application of
cyclic torque induces cyclic shear stresses on horizontal planes
Shaking Table Tests. In the early years of geotechnical earthquake engineering, virtually all
physical model testing was performed on shaking tables. Shaking table research has provided
valuable insight into liquefaction, post earthquake settlement, foundation response, and lateral
earth pressure problems. Most shaking tables utilize a single horizontal translation degree of
freedom, but shaking tables with multiple degrees of freedom have also been developed. Shaking
tables are usually driven by servo hydraulic actuators (Figure 6.36); their dynamic loading
capacities are controlled by the capacity of the hydraulic pumps that serve the actuators. Large
pumps and large actuators are required to produce large displacements of heavy models at
moderate or high frequencies.
High gravitational stresses cannot be produced in a shaking table test. Though the contractive
behavior associated with high normal stresses at significant
We can simulate depths by placing soil very loosely during model preparation, the process of
preparing such models is quite difficult. Because of the low normal stress levels, the contribution
of factors that produce a cohesive component of strength will be greater in the model than in the
prototype. Correction procedures (e.g., Hettler and Gudehus, 1985; Iai, 1989) have been
developed to aid in the interpretation of shaking table test results.
Centrifuge Tests. In a centrifuge test, a I/N‐scale model located at a distance, r, from the axis of a
centrifuge (Figure 6.37) is rotated at a rotational speed, , which is sufficient to raise the
acceleration field at the location of the model to N times the acceleration of gravity. In principle,
the stress conditions at any point in the model should then be identical to those at the
corresponding point in the full‐scale prototype. The overall behavior (e.g., displacements, failure
mechanism, etc.) should also be identical.
Centrifuge tests are restricted to much smaller models than even moderate sized shaking tables.
Figure 6.37 Cross section through a geotechnical centrifuge. (After O’Reilly, 1991.)
Similitude considerations are very important in the planning and interpretation of centrifuge
tests. Scaling factors for a number of parameters are shown in Table 6‐2. the scaling factors show
how dynamic events are speeded up in the centrifuge. For example, the stresses and strains in a
30‐m (100‐ft)‐ high prototype earth dam could be modeled with a 30‐cm (1‐ft)‐high centrifuge
model accelerated to 100g. A harmonic 1‐Hz base motion lasting 10 sec at the prototype scale
would be modeled by a 100‐Hz motion lasting 0.1 sec in the model. The dissipation of any
generated pore pressures, however, would occur 10,000 times as fast in the model as in the field.
For this reason, viscous fluids such as glycerin or silicon oil are often used as pore fluids in
centrifuge models.
High speed transducers and data acquisition systems are required to obtain useful results in
dynamic centrifuge tests. Because the scaling laws apply to all parts of the model, miniaturized
transducers and cables are required to minimize their influence of the model.
Table 6‐2 Scaling Factors for Centrifuge Modelinga
prototype and that the soil properties are not rate dependent.
Mechanics of Particulate Matter
Hertz (1881) studied the behavior of identical spheres of radius, R, compressed against each
other by a normal, force, N (Timoshenko and Goodier, 1951), and showed that
(6.40)
Where G and v are the elastic constants of the spheres and δN is the change in distance between
the centers of the spheres. For a cubically packed array of spheres loaded along one of the
packing axes (Figure 6.38), the average normal stress is obtained by dividing the normal force by
its tributary area, that is
(6.41)
Figure 6.38 Cubically packed assemblage of
spheres subjected to normal stress, σ, and shear
stress, τ, that produce interparticle contact forces
N and T. (After Dobry et al., 1982.)
Then the tangent modulus for uniaxial loading is given by
(6.42)
Which suggests that the stiffness should, theoretically, vary with the cube root of the axial stress.
When a tangential force, T, is applied, elastic distortion causes the centers of the spheres to be
displaced perpendicular to their original axis (Mindlin and Deresiewicz, 1953; Dobry et al., 1982)
by an amount
(6.43)
Where f is the coefficient of friction between the spheres. When T becomes equal to fN, gross
sliding of the particle contacts occurs. This gross sliding is required for permanent particle
reorientation; consequently, volume changes cannot occur and excess pore pressures cannot be
generated when gross sliding does not occur.
(6.44)
The shear strength corresponding to the initiation of gross sliding is called the threshold shear
strain. When the properties of quartz (E = 11 x 106 psi (7.6 x 107 kPa), v = 0.31, f = 0.50) are
substituted into equation (6.44), the threshold shear strain is given by
(6.45)
Where σ is in psf. For confining pressures of practical interest (500 to 4000 psf (25 to 200 kPa)),
equation (6.44) would predict a threshold shear strain between 0.01 and 0.04%. Real soils, of
course, do not consist of regular arrays of spherical particles, but the existence of a threshold shear
strain very close to that predicted by equation (6.44) has been observed experimentally for sands
under both drained (Drnevich and Richart, 1970; Youd, 1972; Pyke, 1973) and undrained (Park and
Silver, 1975; Dorby and Ladd, 1980; Dobry et al., 1982) loading conditions. The simple idealized
analysis of a regular array of spheres helps illustrate the reason for its existence. Experimental
evidence suggests that the volumetric threshold shear strain increases with plasticity index (PI); the
volumetric threshold shear strain of a clay with PI = 50 is approximately one order of magnitude
greater than that of a sand, with PI = 0 (Vucetic, 1994). Experimental evidence also indicates
(Vucetic, 1994) that soils exhibit linear elastic behavior below a linear cyclic threshold shear strain,
γtl, that is approximately 30 times smaller than γtv.
(Equivalent) Linear G and models
G = c/ c
Actually use Gsec
NOT Gtan
Width of hysteresis
Dissipated
energy loop is a measure
of damping
2 x area of
Strain energy
See wave propagation notes, Slides # 35 ‐ # 52
Hysteresis loop. Referring to Figure B.16, the damping ratio is taken as the ratio of the area of the
hysteresis loop to the area of the shaded triangle, all divided by 4π. The damping characteristics
of a linear system can also be evaluated from its frequency response characteristics. Setting the
magnification factor expression [equation (B.44)] equal to half‐power tuning ratios, shown in
Figure B.17, can be approximated as
from Damped Forced Vibrations
The general solution to the eq. of motion for damped forced vibration can be
obtained by combining the complementary and particular solution:
Superscript
]
• By varying (=/n), we get the magnification
curve as shown.
• For M = Mmax/2,
and
• We get:
1.0
• Approximately:
or
Other Measures of Energy Dissipation
In addition to the damping ratio, ξ, a number of other parameters have been used to describe
energy dissipation characteristics. Seismologists, for example, often work with the quality factor.
(B.63)
In vibration analysis, the loss factor Compare with Kelvin Voigt’s
viscosity: = 2 (G/)
(B.64)
And specific damping capacity
(B.65)
Are often used (Goodman, 1988).
It is important to remember that the damping ratio, and any of these other parameters, are
simply parameters used to describe the effects of phenomena that are often poorly understood.
They allow the effects of energy dissipation to be represented in mathematically convenient
manner. For most soils and structures, however, energy is dissipated hysteretically (i.e., by
yielding or plastic straining of the material). In such cases the behavior is more accurately
characterized by evaluating the nonlinear response of the system.
Factors influencing G
• G = F (, c, e, ’m, PI, OCR, Ncycles, …..)
Strain Dependency of G
Figure 6.40 Backbone curve showing typical variation of Gsec with shear strain
Gmax
• When geophysical tests are performed
@ < 3 x 10‐4 %:
Gmax = vs2
(6.49)
• Where F(e) is a function of the void ratio, OCR the overconsolidation ratio, k and
overconsolidation ratio exponent (Table 6‐3), σ’m the mean principal effective stress [σ’m = (σ’1
+ σ’2 + σ’3 )/3], n a stress exponent, and pa is atmospheric pressure in the same units as σ’m
and Gmax. Hardin (1978) proposed that F(e)= 1/(0.3+0.7e2), while Jamiolkowski et al. (1991)
suggested that F(e)=1/e1.3. the stress exponent is often taken as n=0.5 but can be computed
• The stress exponent is often taken as n = 0.5 but can be computed for individual soils from
the results of laboratory tests at different effective confining pressures. It should be apparent
that Gmax, pa, and σ’m must be expressed in the same units. Equation (6.49) can also be used
to adjust measured Gmax values to represent conditions that are different (e.g., increased
effective stresses) from those at which the measurements were made.
Table 6‐3 Overconsolidation ratio Exponent, k
Plasticity Index k
0 0.00
20 0.18
40 0.30
60 0.41
80 0.48
≥100 0.50
Other empirical relationships have been proposed for specific soil types. The maximum shear
modulus of sand, for example, is often estimated as
(6.50)
Table 6‐4 Estimation of K2,max
e K2,max Dr(%) K2,max
0.4 70 30 34
0.5 60 40 40
0.6 51 45 43
0.7 44 60 52
0.8 39 75 59
0.9 34 90 70
Source : Adapted from Seed and Idriss
From Su data for clays
Table 6‐5 Values of Gmax/sua
Overconsolidation Ratio, OCR
Plasticity index 1 2 5
15‐20 1100 900 600
20‐25 700 600 500
35‐45 450 380 300
Source : After Weiler (1988).
a Undrained strength measured in CU triaxial compression.
Effects of Strain Rate & Degree of Consolidation
Evaluation of shear modulus can be complicated by rate and time effects (Anderson and Woods, 1975,
1976; Anderson and Stokoe, 1978; Isenhower and Stokoe, 1981). Rate effects can cause Gmax to increase
with increasing strain rate. The influence of strain rate on Gmax increases about 4% per tenfold increase
in strain rate. Rate effects can be significant when comparing Gmax values obtained from field shear
wave velocity measurements (usually made with the use of impulsive disturbances which produce
relatively high frequencies) with values obtained from laboratory tests.
The shear wave velocity, and hence Gmax, increases approximately linearly with the logarithm of time
past the end of primary consolidation to an extent that cannot be attributed solely to the effects of
secondary compression. The change of stiffness with time can be described by
(6.51)
where ΔGmax is the increase in Gmax over one log cycle of time and (Gmax)1000 is the value of Gmax at a
time of 1000 min past the end of primary consolidation. NG increases with increasing plasticity index, PI,
and decreases with increasing OCR (Kokushu et al. 1982). For normally consolidated clays, NG can be
estimated from the relationship.
Anderson and Woods (1975) showed that some of the discrepancy between Gmax values from field and
laboratory tests could be explained by time effects, and that NG could be used to correct the Gmax values
from laboratory tests to better represent actual in situ conditions.
Update on Strain Rate Effect to Measured G
Matesic, L., Vucetic, M. “Strain Rate Effect on Soil Secant Modulus at Small
Cyclis Strains,” ASCE, JGGE, June 2003.
Clay
Sand
Summary
• Measured Go,secant increases with strain rate
• Effect more pronounced in clay than in sands
• As the strain rates increases, the effects of
strain rate to Go,secant reduces. For sands in the
test, increasing strain rate beyond (0.003 to
0.006%)/sec. gives less than 1% change in
Go,secant, while in the tested clay, a change of
about > 1.5 % is still observed when the strain
rates increased from 0.005 to 0.009 %/sec.
In Situ
Relationship Soil Type References Comments
Test
Increasing Factor Gmax
Cementation, c Increase with c
Plasticity index Increases with PI if OCR >1;
Stays about constant if OCR = 1
Strain rate, ϒ No effect for non‐plastic soils; increases with ϒ
for plastic soils (up to ~ 10% increase per log
cycle increase in ϒ)
Number of loading cycles, N Decreases after N cycles of large ϒc , but
recovers later with time in clays: increases with
N for sand
Source : Modified from Dobry and Vucetic (1987)
Shear Modulus Reduction
Figure 6.42 Modulus reduction curves for fine grained soils of different plasticity.
(After Vucetic and Dobry (1991). Effect of soil plasticity on cyclic response. Journal
of Geotechincal Engineering, Vol. 117, No. 1. Reprinted by permission of ASCE.)
• Shape of curves governed by PI
• Threshold strain, th, increases with PI
• Curves applicable for both fine‐coarse grained
soils, PI = 0 curve is curve for sand.
• See next slide for generic formula.
• Effect of ’ significant for sands only [see the
2nd next slide]
Modulus reduction behavior were combined by Ishibashi and Zhang (1993) in the form
(6.52)
where
Influence of ’ to reduction curve
PI = 50
Figure 6.43 Influence of mean effective confining pressure on modulus reduction curves for (a)
nonplastic (PI=0) soil, and (b) plastic (PI=50) soil. (After Ishibashi (1992). Discussion, Journal of
Geotechnical Engineering, Vol. 118, No. 5 . Reprinted by permission of ASCE.)
• Effect of number of cycles for clays
Pore pressure generation and structural changes can cause the shear strain amplitude of a soil
specimen to increase with increasing number of cycles, under stress‐controlled harmonic loading
conditions. The shear stress amplitude would be observed to decrease with increasing number
of cycles, if clay or saturated sand specimens are loaded harmonically under strain‐controlled
undrained conditions. For cohesive soil, the value of the shear modulus after N cycles, GN, can be
related to its value in the first cycle, G1, by:
GN = δG1 (6.53)
Where the degradation index δ is given by δ = N‐t and t is the degradation parameter (Idriss et al.,
1978). The degradation parameter has been shown to decrease with increasing PI and increasing
overconsolidation ratio, and to increase with increasing cyclic strain amplitude (Idriss et al., 1980;
Vucetic and Dobry, 1989; Tan and Vucetic, 1989). The effects of stiffness degradation on modulus
reduction behavior is shown in Figure 6.44.
Effect of number of cycles for clays
Figure 6.44 Effect of cyclic degradation on shear modulus.
(After Vucetic and Dobry (1991). Effect of soil plasticity on
cyclic response, Journal of Geotechnical Engineering, Vol
117, No. 1. Reprinted by permission of ASCE.)
Table 6‐9. Effect of Environmental and Loading Conditions on Modulus Ratio (at a Given
Strain Level) of Normally Consolidated and Moderately Overconsolidated Soils
Increasing Factor G/Gmax
Cementation, c May increase with c
Plasticity index Increases with PI
Cyclic strain, ϒc Decreases with ϒc
(6.54)
Table 6‐9. Effect of Environmental and Loading Conditions on
Damping Ratio of Normally Consolidated and Moderately
Overconsolidated Soils.
Increasing Factor Damping ration, ξ
Confining pressure, σ’m Decreases with σ’m ; effect decreases
with increasing PI
Void ratio, e Decreases with e
Geologic age, tg Decreases with tg
Cementation, c May decrease with c
Overconsolidation ratio, OCR Not affected
Plasticity index Decreases with PI
Cyclic strain, ϒc Increases with ϒc
Strain rate, ϒ Stays constant or may increase with
ϒ
Number of loading cycles, N Not significant for moderate ϒc and
N.
Source : Modified from Dobry and Vucetic (1987)
Cyclic Non Linear Models
• Follows actual stress‐strain path.
• Variety of cyclic non linear models:
a. Backbone curves
b. Series of rules governing unload‐reload
behavior, stiffness degradation etc.
• More complex models account for pore
pressure generation and dissipation
Characteristics of Non Linear Models
• Have non linear backbone curves
• Have loading and unloading curves
MORE ADVANCED
• Include cyclic stiffness degradation
SIMPLER
behavior
• Include pore pressure generation and
dissipation
• Account for irregular loading effects
• Include densification effects
max
Gmax
2
2
2
(r2 , r2 )
Masing rules
Back Bone Curve
Figure 6.47. Extended Masing rules: (a) variation of shear stress with time; (b) resulting stress‐
strain behavior (backbone curve indicated by dashed line).
Additional rules
(extended Masing Rules)
• If unloading or reloading curve exceeds the
maximum past strain and intersects the
backbone curve, it should follow the backbone
curve until next stress reversal.
• If unloading or reloading curve crosses an
unloading or reloading curve from previous
cycle, it should follow that of the previous
cycle.
Features:
• Simple model for numerical computation.
• Hysteric damping is included.
• To avoid spurious response at small , can be
increased at small .
• Model does not account for v due to cyclic
shear stresses.
Advanced Constitutive Models
• Yield surface
• With hardening rules, change in yield surface
shape and values when plastic deformation
occur.
• With flow rules, plastic strains due to
increasing shear strain.
• Mostly containing difficult to obtain
parameters.
Cyclic Strength of Soils
• Exceeding limiting strength or deformations
• Cohesionless tied to liquefaction
• Cohesive, related to undrained shear strength
‐ Cyclic stress produces cyclic and increases
average as well.
‐ Monotonic strength (at the end of cyclic
loading)
Figure 6.48. Definitions of average and
cyclic shear stress and shear strain. (After
Goulois et al., 1985. Used by permission of
ASTM)
Cyclic Strength
• Depends on both cycl and ave
• When ave is small, unidirectional is small,
however, cyc can become large for large cycl
• When ave is large, unidirectional is large, cyc
will be large even for lowcycl
• When ave = 0, failure is defined in terms of cyc
(usually 3 %). [See next slide]
• For low stress ratios (below critical level of
repeated loading, CLRL), failure strain will never
be reached. CLRL increases with PI (0.05 for NP
silt to 0.55 for SF Bay Mud).
f > 5 %
f O 3 %
Figure 6.49 Variation of cyclic strength ratio with number of cycles for different soils (After Lee and Focht, 1976.)
For ave 0
• cyc and ave will depend on cyc and ave
• cyc predominantly depends on cyc and N
• ave predominantly depends on ave and N
• See Figure 6.50
Figure 6.50. Variation of average shear strain with average
shear stress, cyclic shear stress, and number of cycles in
cyclic direct simple shear tests on plastic Drammen clay.
τstrength of a reference strength measured under slow loading
conditions with τave = 0. cyclic loading applied at a period of
10 sec. (after Goulois et al., 1985. Used by permission of
ASTM)
Monotonic Strength
• Post earthquake strength must reflect effects
of cyclic loading by the earthquake.
• For saturated soils, controlled by e and
structure.
• When cyclic loading increases pore pressures,
’ drops, stiffness will also drop.
• Cyclic load may substantially effect soil
structure. [See next slide, samples with similar
initial e].
Figure 6.51. Effect of cyclic loading on subsequent monotonic undrained loading
behavior of triaxial specimens of a slightly plastic silt. (1) stress=‐strain behavior; (b)
prior cyclic loading. Specimens 2 to 6 were subjected to different levels of cyclic
loading prior to monotonic loading. Note the dilative nature of the stress paths of
specimens 2 to 6 compared to specimen 1. (After Castro and Christian, 1976.)
Less than 10 % strength
reduction for cyc < 50 %
f,static
Figure 6.52. Effect of peak cyclic strain on
monotonic strength after cyclic loading. (After
Thiers and Seed, 1969. Used by permission of
ASTM)
Questions?
• Why is the effect of “structure disturbance”
change small for post cyclic Su at large strain?
• Why are the cyclic deformations at similar
deviator stress larger than the monotonic
ones?
• Do you think that Thiers and Seed’s
observation, that at cyc < 50 % f,static , will also
produce small reduction in for high OCR
clays?