Professional Documents
Culture Documents
Normalized Modulus Reduction and Material Damping Ratio Curves for Bay
of Campeche Sand
Victor M. Taboada, and Vishal Dantal, NGI Inc.; Diego Cruz Roque, Francisco Flores Lopez,
and Procoro Barrera Nabor, Instituto Mexicano del Petroleo
This paper was prepared for presentation at the Offshore Technology Conference held in Houston, Texas, USA, 2–5 May 2016.
This paper was selected for presentation by an OTC program committee following review of information contained in an abstract submitted by the author(s). Contents
of the paper have not been reviewed by the Offshore Technology Conference and are subject to correction by the author(s). The material does not necessarily reflect
any position of the Offshore Technology Conference, its officers, or members. Electronic reproduction, distribution, or storage of any part of this paper without the
written consent of the Offshore Technology Conference is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words;
illustrations may not be copied. The abstract must contain conspicuous acknowledgment of OTC copyright.
Abstract
Predictive equations for estimating normalized shear modulus and material damping ratio of sand are
presented in this paper. The equations are based on a modified hyperbolic model and a statistical analysis
of existing isotropically consolidated resonant column and strain-controlled cyclic direct simple shear test
results for 252 specimens obtained from the Bay of Campeche. Two independent modified hyperbolic
relationships are fitted to model stiffness (G/Gmax)-strain using two parameters and material damping
ratio-strain curves using four parameters. Variables used in the equation for normalized shear modulus
are: confining pressure; shear-strain amplitude; a reference strain, defined as the shear strain at which the
shear modulus has reduced to 0.5Gmax, and a curvature parameter which controls the rate of modulus
reduction, such as the model suggested by Darendeli (2001). The equation for damping ratio D, is
expressed in terms of the reference strain, defined as the shear strain for a 50% increase in material
damping ratio (i.e. D/Dmax ⫽ 0.5), a curvature parameter which controls the rate of material damping ratio
increase, the minimum material damping ratio Dmin, and the maximum material damping ratio Dmax,
similar to the equation suggested by Gonzalez and Romo (2011). It is found that the Bay of Campeche
sand exhibit more linear response and lower damping ratio than other sands reported in the literature. The
uncertainties associated with the predictive equations are quantified. A case study is provided to illustrate
an application of the predictive equations to seismic response analysis and the importance of considering
confining stress. The predictive equations of normalized shear modulus reduction G/Gmax and Damping
ratio curves are easy to apply in practice, and are useful in the analysis of granular strata and offshore
structures subjected to earthquake loading when site specific laboratory testing is not available.
Introduction
Background
The Bay of Campeche is located in the large bay comprising the southern portion of the Gulf of Mexico
between the Yucatan Peninsula to the east, the Isthmus of Tehuantepec to the south, and the coast of
Mexico at Veracruz to the west. The Bay of Campeche is limited to that section enclosed approximately
by 91° W. Longitude on the east to 95° W. Longitude to the west and 20° N. Latitude on the north to the
Mexican coast on the south, as indicated by the rectangular area in Figure 1.
2 OTC-26878-MS
The Bay of Campeche covers an area of about 15,540 square kilometers, and is the largest oil field in
Mexico which has prompted an extensive installation of shallow water fixed offshore platforms type
jacket. The oil platforms installed in the Bay of Campeche must be designed against earthquake loading
since the piles can be subjected to large seismic lateral loadings.
The dynamic structural analyses of the oil platforms require acceleration time histories or acceleration
spectra that already include the soil amplification of the earthquake motions. Two of the most important
dynamic soil properties required to conduct an equivalent-linear seismic site response analysis to evaluate
the soil amplification are: i) a curve of G/Gmax versus cyclic shear strain ␥, also called modulus reduction
curve, where G is the shear modulus and Gmax is the maximum shear modulus at very low shear strains
of the order of 10⫺4%, and ii) a curve of equivalent hysteretic or material damping ratio D versus ␥, where
D is defined from the measured area WD inside a complete hysteretic loop, which corresponds to the
energy dissipated in one cycle, and the maximum strain energy stored during one cycle WS, through the
basic expression (see Figure 2):
(1)
Figure 2—Hysteresis loop for one cycle of loading showing Gmax, G and D (Zhang et al., 2005)
OTC-26878-MS 3
Typically, G/Gmax decreases and D increases as ␥ becomes larger, and in fact it has been observed that
a fast decrease in G/Gmax with ␥, corresponding to a strongly nonlinear soil, is associated with a strong
increase of D with ␥ in the same soil, and vice versa.
In general, G/Gmax increase when the confining pressure m increases. A simple way of explaining this
trend is by focusing on the dependence of the shear modulus G on m. The maximum shear modulus
, where m ⫽ 0.4 to 0.5 and A is a constant. For an isotropically consolidated sand,
Coulomb’s strength law indicates that the shear strength max increses linearly with m (max ⫽ m sin
for pure shear loading, where ⫽ angle of internal friction of the soil). As G ⫽ /␥ (Figure 2), at large
strains, G ⫽ max ␥ ⫽ m sin /␥ that is, G increases linearly with m at large strains, while G ⫽ Gmax
increases approximately with m0.5 at small strains. Therefore, G/Gmax is proporsional to m1-m ⬇ m0.5
and as m increases G/Gmax increases. Conversely, the damping ratio D decreases.
It is generally difficult by any means to perform in situ tests in which large strains are imposed
uniformly in the soil as it exists in the field. In realization of this, most of the recent efforts for pursuing
the strain dependency of modulus and damping of in situ soils have been directed towards the conduct of
laboratory tests on undisturbed samples which are regarded as representing intact conditions in the field.
Unfortunately, it is not always possible to perform dynamic laboratory testing to obtain the shear
modulus reduction curve and material damping ratio curve of all the soil strata found in the offshore soil
deposit, and the geotechnical engineer is left to use curves developed mainly for onshore soils with
different geological settings than those of the Bay of Campeche sand. Thus, a characterization of the cyclic
stress-strain response of the sand in shear that facilitate prediction of the shear modulus reduction curve
and material damping ratio curve of sands of the Bay of Campeche are needed.
To cover this need, a database has been established and tailored for an equivalent linear character-
ization of the cyclic response of sand soil units in the Bay of Campeche. The data have been collected
from offshore soil investigations performed between 2012 and 2015 that include abundant results of
resonant column tests and cyclic direct simple shear tests.
of uniform fine sand. Offshore flow behaves like laminated and low speed, allowing the deposition of
sediments from the size of the clay.
In general, at the seafloor there is a layer of very soft to soft olive gray clay, with shell fragments with
a thickness ranging from 5 to 25 m. However, there are areas with presence of granular soil; these areas
usually correspond to the mouths of the rivers of the area and surfaces where coral formations emerge or
areas close to the coast where the waters are shallow. An example of this is the coast located in front of
Ciudad del Carmen, where the seafloor sediments consist of a dense layer of sand about 10 m thick
overlaying a firm to hard clay layer. The surficial clay layer reaches its largest thickness in the northeast
part where it reaches 25 m. The thickness of the clay layer reduces to the southwest, allowing the
underlying sand to appear superficially to the edge of the continental shelf.
Another important feature of the stratigraphy of the Bay of Campeche is that below the sand layer that
underlies the surface clay layer is an alternating sequence of clays and sands. The former vary from stiff
to hard clays, and the latter are medium dense to very dense sands.
Stiffness and damping ratio database of sandy soils
The existing laboratory G/Gmax and D data used in this study are compiled from offshore geotechnical
investigations performed in the Bay of Campeche between 2012 and 2015 that include isotropically consoli-
dated resonant column and strain-controlled cyclic direct simple shear test results for 252 specimens of sand.
Histograms of the 252 specimens with respect to the water content, void ratio, relative density, fines
content, carbonate content, confining pressure, coefficient of earth pressure at rest and submerged unit
weight of the specimen are presented in Figure 3(a-h). As shown in Figure 3(a) and 3(b), most of the water
content and void ratio of the specimens vary in the ranges of 20%-30%, and 0.6-0.8, respectively.
The relative density of the specimens presented in Figure 3(c), is mostly dense in the range of
50%-80%, with only four loose specimens with relative density below 35%, and twenty-four very dense
specimens with relative density higher than 85%. About 55% of the specimens had a fines content in the
range of 5%-10% (Fig. 3d), classified as sand with silt, and about 45% of the specimens with fines content
higher than 12 % but less than 40%, classified as silty sand.
According to the carbonate content shown in Figure 3(e), about 70% of the soil specimens are siliceous
sands with carbonate content less than 10%, and the rest of the specimens are calcareous sands with
carbonate content between 10% and 40%. The effective confining pressures presented in Figure 3(f) are
between 100 kPa and 700 kPa, with an average of 400 kPa.
The coefficient of earth pressure at rest is mostly between 0.45 and 0.5, as observed in Figure 3(g)
while the submerged unit weight presented in Figure 3(h) is mostly between 20 kN/m3 and 21 kN/m3.
The sand samples were obtained offshore through the drill pipe by percussion (hammer) with 57.2-
mm-OD, 54.0-mm-ID thin-walled steel tubes. In very dense sand, percussion samples are obtained either
with a 63.5-mm-OD, 54.0-mm-ID tapered, thick-walled tube sampler, or a 76.2-mm-OD, 63.5-mm-ID
split-spoon sampler.
Wireline percussion sampling consists of lowering a hammer and slide assembly through the drill pipe
to the desired sampling depth. The hammer is a 79-kg sliding weight that is raised and dropped
approximately 1.5 m to achieve a maximum penetration of 0.6 m, or a maximum of 30 blows. In very
dense sands, blow counts in excess of 30 blows are sometimes required to drive the sampler into the
formation in order to obtain an adequate amount of sample for laboratory soil testing.
The sand samples were extruded from the sample tubes, examined and visually classified. Soil samples
were tested with diluted hydrochloric acid solution (10% concentration) to obtain a qualitative assessment
of the carbonate content. Natural water content and unit weight measurements were performed. After
testing, representative portions of each soil sample were sealed in airtight containers for shipment to the
laboratory for further testing. The test specimens were taken from extruded samples that were sealed in
the airtight containers and trimmed to the corresponding diameter.
Resonant column tests were performed on solid, cylindrical-shaped, soil specimens approximately 38
mm in diameter and 76 mm in length. Each test specimen was back-pressure saturated to about 140 kPa
and then isotropically consolidated to 0.5, 1.0 and 2.0 times the in-situ effective confining pressure.
Strain-controlled cyclic direct simple shear (DSS) tests were conducted on soil specimens of approx-
imately 18 to 19 mm in height, trimmed to approximately 50 mm in diameter. Each granular test specimen
was consolidated to an effective vertical consolidation stress equal to the estimated in situ effective
vertical stress. Each test specimen was allowed to consolidate to about one log cycle of time or 24 hours,
whichever was less, past the end of primary consolidation (T100) before applying 20 cycles of sinusoidal
cyclic horizontal loads at a frequency of 1.0 Hz (for earthquake loading conditions). The test was
conducted while maintaining the specimen at constant volume, with the pore pressures estimated by
measuring the changes in the vertical stress during cycling. Each specimen was subjected to a specified
nominal average cyclic shear strain of 0.25%, 0.5% or 1.0%.
where G is the shear modulus at any strain, Gmax is the maximum shear modulus at ␥ ⫽ 0.0001%, and
␥r is the reference shear strain, which is defined by max/Gmax. The disadvantage of this approach is the
6 OTC-26878-MS
difficulty in finding max. The authors also indicated that this true hyperbolic relationship did not generally
fit their data (Harding and Drnevich, 1972)
Darendeli (2001) proposed a modified hyperbolic model based on testing of intact sand-gravel samples:
(3)
where a is called the curvature parameter, and ␥r is the reference strain value at which G/Gmax ⫽ 0.5.
This model uses only two parameters, and the reference strain provides an efficient normalization of shear
strain.
In order to comprehend the non-linear elastic behavior of marine sands, and to produce a best-fit
functional relationship, a new database has been constructed incorporating shear modulus degradation
curves from the Bay of Campeche sand. This curve-fitting process facilitates prediction of shear modulus
degradation curves. The relationship between normalized shear modulus (G/Gmax) and shear strain (␥) for
the 252 selected tests is plotted in Figure 4.
Figure 4 —Shear modulus degradation data from database for Bay of Campeche sands
Equation (3) is adopted in this study to model the variation of G/Gmax with ␥. For the sake of simplicity,
the curvature parameter a, is assumed independent of the effective confining pressure, , and equal to
a ⫽ 1.08. The effective confining pressure is calculated by
(4)
where is the effective vertical stress, is the effective horizontal stress and Ko is the coefficient
of earthpressure at rest.
The best-fit functional relationship for the shear modulus degradation data of Figure 4 is shown in
Figure 5 as a modified hyperbolic equation in the form proposed by Darendeli (2001) and presented in
equation (3). The mean, lower bound and upper bound curve-fitting parameters are:
OTC-26878-MS 7
where F(e) is a function of the void ratio, A and m are material constants and , is the effective
confining pressure defined by equation (4). Considering the void ratio function 1/(1⫹e)3 proposed by
Harding and Black (1966), equation (5) can be formulated as a dimensionless equation for strain-
dependent shear modulus as (Oztoprak and Bolton, 2013):
(6)
where A(␥) and m(␥) are strain-dependent parameters, and Pa is a reference pressure of 100 kPa
(effectively, the atmospheric pressure). The use of equations (3) and (6) in sequence can provide a method
for refining optimum functional relationship, as proposed by Oztoprak and Bolton (2013).
With the available information on , and e from the database, log-log graphs of G(1⫹e)3/Pa
against are shown in Figure 6, where the strain-dependent parameter A(␥) and m(␥) are
deduced for various strains.
8 OTC-26878-MS
Figure 6 —Shear modulus variation with effective confining stress: (a) ␥ ⴝ 0.001%; (b) ␥ ⴝ 0.015%; (c) ␥ ⴝ 0.25%; (d) ␥ ⴝ 0.5%; (e) ␥ ⴝ 1%.
To plot the results of equation (6) on the mean curve of G/Gmax, against ␥, in equation (6) can be
assigned as 400 kPa, which is the average value of the medium stress range in Figure 3. Since G/Gmax is
to be calculated, the void ratio function cancels, and is not required. Using , A(␥) and m(␥) from Figure
6, the G/Gmax values can readily be calculated and placed on Figure 7, where they lie close to the mean
curve, but not exactly on it. Since the normalized shear modulus gives a more reliable relation than the
scatter relationship shown in Figure 6, it is preferred to modify the A(␥) and m(␥) values from the values
listed in Figure 6 so as to fit exactly the mean hyperbolic curve presented with solid red line in Figure 7.
The refined values of A(␥) and m(␥) appear in Figure 8.
Figure 7—Calibration of m(␥) and A(␥) using equation (6) and mean G/Gmax curve of database ( )
OTC-26878-MS 9
Figure 8 —Modified relations of A(␥) and m(␥) to predict shear modulus using equation (6)
Once the calculated G/Gmax values of equation (3) have been fitted to the mean hyperbolic curve in
Figure 7, for the mean effective confining pressure , ⫽ 400 kPa, other values can be inserted in
equation (6), and appropriate values of ␥r can be derived from equation (3) to fit the new curves. The
resulting family of degradation curves for varying is shown in Figure 7. If the resulting m(␥) relation
is compared with the published data of others researchers, as shown in Figure 9, it now creates a trend line
appropriate to the Bay of Campeche sands, hence verifying the modification explained above.
Figure 9 —Comparison of strain-dependent values of m(␥) with those from other work (After Oztoprak and Bolton, 2013).
According to the database curve in Figure 5, G/Gmax is 0.5 at a shear strain between 0.044% and 0.14%,
with a mean of 0.08%. Using a constant value, a ⫽ 1.08 in equation (3), and changing the confining
pressure progressively from 100 kPa to 700 kPa, leads to the interesting offsetting of the modulus
degradation curves towards higher values of strain, as seen in Figure 7. The trend of ␥r increasing with
effective confining stress is shown in normalized form in Figure 10. For the most common confining
stresses (70 kPa to 700 kPa), the Bay of Campeche sands of this database produce a regression for ␥r as
follows:
(7)
10 OTC-26878-MS
Figure 10 —Trend of ␥r with normalized effective confining pressure, derived from Figure 7
Validation of G/Gmax
In the case of unavailability of the shear modulus degradation curve for the Bay of Campeche sand, it is
possible to calculate it with equations (3) and (7) with a ⫽ 1.08 proposed in this paper. Comparison
between measured and predicted values can be validated against the database. G/Gmax between predictions
and measurements can best be assessed by plotting predicted against measured values for the 936 data
points accumulated from all the tests. This is presented in Figure 11, where it can be seen that 85% of the
predictions lie within a factor of 1.3 and 0.7 from the measurements.
Figure 11—Comparison of measured and calculated G/Gmax values (936 data points)
The best-fit functional relationship for the material damping ratio data of Figure 12 is shown in
Figure 13 as a modified hyperbolic equation similar to that proposed by Gonzalez and Romo (2011)
in the form:
(8)
OTC-26878-MS 11
where, Dmax and Dmin are the maximum and minimum material damping ratio, respectively; ␥rD is the
deformation corresponding to 50% of increase in material damping ratio D (i.e. D/Dmax ⫽ 0.5), and aD
is a curvature parameter characteristic of the curve D-␥.
Figure 12—Material damping ratio data from database for Bay of Campeche sands
Figure 13—Fitting a hyperbola to the damping ratio data and curve-fitting parameters
According to the database curve in Figure 12, D/Dmax is 0.5 at a shear strain between 0.08% and 0.35%,
with a mean of 0.165%. Using a constant value of aD ⫽ 1.85 (i.e. independent of confining pressure) in
equation (8), and changing the effective confining pressure progressively from 100 kPa to 700 kPa in
equation (6), leads to the offsetting of the material damping ratio curves towards higher values of strain
as seen in Figure 14. The trend of ␥rD increasing with effective confining stress is shown in normalized
form in Figure 15. The Bay of Campeche sands of this database produce regression for ␥rD as follows:
12 OTC-26878-MS
(9)
Figure 14 —Prediction of D for different using equation (8) and mean D curve of database ( )
Figure 15—Trend for evolution of ␥rD with normalized confining pressure, derived from Figure 14
The trend of minimum material damping ratio Dmin, with normalized effective confining pressure
plotted in Figure 16 produces a regression for Dmin as follows:
(10)
OTC-26878-MS 13
Figure 16 —Trend of minimum material damping ratio with normalized effective confining pressure
Figure 17 demostrates that a simple linear relation can be derived between (Dmax – Dmin) and
normalized effective confining pressure using all the available data, as:
(11)
In the case of unavailability of the material damping ratio D and its increase by straining for the Bay
of Campeche sand, it is possible to calculate them with equations (8), (9), (10), (11) and a constant
curvature parameter of the material damping curve of aD ⫽ 1.85 independent of , proposed in this
14 OTC-26878-MS
paper. Comparison between measured and predicted values of material damping ratio D can be validated
against the database for all the 936 data points accumulated from all the tests. This is presented in Figure
18, where it can be seen that 80% of predicted values lie within a factor of 1.3 and 0.7 from the
measurements.
Figure 18 —Comparison of measured and calculated material damping ratio (D) values (936 data points)
Figure 19 —Comparison of recommended G/Gmax and D curves for m’ⴝ100 kPa with Seed and Idriss (1970)
Figure 20 —Comparison of recommended G/Gmax and D curves for m’ⴝ100 kPa for the Bay of Campeche sands with Seed and Idriss
(1970), Vucetic and Dobry (1991), and Darendeli (2001) curves
For shear strains lower than 0.09%, the recommended G/Gmax curve for Bay of Campeche sand for m’
of 100 kPa presented in Fig. 20(a) is more linear than Darendeli (2001) G/Gmax curve at the same
confining pressure and Vucetic and Dobry (1991) for PI ⫽ 0. For shear strains higher than 0.09%, the
recommended G/Gmax curve of Bay of Campeche sand is slightly lower than mean of Seed and Idriss
(1970) and Vucetic and Dobry (1991) curves. For shear strains higher than 0.001%, the recommended
damping ratio curve for the Bay of Campeche sand for m’ of 100 kPa presented in Fig. 20(b) is lower
than both Darendeli (2001), and Vucetic and Dobry (1991) damping curves, and Seed and Idriss (1970)
mean damping curve.
Model uncertainty
There are discrepancies between the measured data and the predicted G/Gmax and material damping ratio
(D) values, as can be seen in Figure 11 and Figure 18. The recommended curves of shear modulus
degradation and material damping ratio obtained using equations (3) and (8) are considered as mean
curves and are function of multiple variables (i.e. a, aD, ␥r, ␥rD, Dmin, and Dmax), with each variable
exhibiting some degree of uncertainty.
The standard deviation (SD) of the predicted and measured G/Gmax and material damping ratio for
three effective confining pressures 200, 400 and 700 kPa was estimated and presented in Figure 21. Also
16 OTC-26878-MS
presented in this figure are the recommended mean curves and the measured data. It can be seen that the
⫾1 SD ranges for the three effective confining pressures are similar. This results agree with observed
scatter in the compiled data (see Figure 5 and Figure 13), and are generally consistent with uncertainties
exhibited in the data sets compiled by Seed and Idriss (1970), Vucetic and Dobry (1991), and Darendeli
(2001).
Figure 21—Recommended mean and 1 SD range G/Gmax-␥ and D-␥ curves for m’ of 200, 400 and 700 kPa
Application
To illustrate the effect that G/Gmax and D curves have on seismic response analysis, the soil deposit
at a platform location in the Bay of Campeche is considered, consisting of six sand strata, one stratum
of very stiff lean clay and two strata of hard clay, as illustrated in the borehole log presented in Figure
22.
OTC-26878-MS 17
Figure 22—Borehole log of the soil deposit considered in the application example
18 OTC-26878-MS
In-situ measurements of shear wave velocity with P-S Logging system were available at the site and
are plotted with dotted line in Figure 23. This figure also presents with solid line the median shear wave
velocity profile as discretized for input into the widely used equivalent linear method that is solved in the
frequency domain and implemented in the computer program SHAKE (Schnabel et al., 1972), as modified
by Idriss and Sun (1992) with the pre- and postprocessor routines of SHAKE 2000 (Ordoñez, 2000).
Figure 23—Shear wave velocity (Vs) measured in-situ and Vs profile as input into SHAKE and DEEPSOIL
The median shear wave velocity profile presented with solid line in Figure 23 was also used as input
in the widely used non-linear time domain site response analysis code DEEPSOIL (Hashash et al., 2015)
to compare the results of this program with the linear-equivalent program SHAKE2000.
Input time history
The acceleration time histories used to generate the ELE and ALE input motions of the site response
analysis was the recorded Loma Prieta 1989 earthquake, shown in Figure 24. Characteristics of the
earthquake and recording station for this recorded motion are presented in Table 1.
OTC-26878-MS 19
Figure 24 —Recorded acceleration time history (HDA 165) used in the site response analysis
HDA USGS 1656 Hollister Diff. Array / 1989-10-18 HDA165 6.93 17.5 45.1 0.27
Loma Prieta 1989
This acceleration time history was spectrally matched to a normalized target bedrock spectrum obtained
from a probabilistic seismic hazard analysis (PSHA) performed for the Bay of Campeche and then scaled to
two levels of out cropping peak ground acceleration (PGA), 0.06g and 0.12 g that are representative of the
extreme level earthquake (ELE) and abnormal level earthquake (ALE) events, respectively.
Extrapolation of soil properties beyond bottom of borehole
The spectrally matched acceleration time history was defined for a base stratum with a shear wave velocity
of approximately 620 meters per second (m/sec). At the site, the shear wave velocity at the borehole
termination depth was estimated to be less than the required 620 m/sec. Below the bottom of the borehole,
it was assumed that shear wave velocity (Vs) increased as a function of the effective vertical stress (’v)
in accordance with the relationship:
(12)
The stress exponent (n) generally ranges from 0.25 to 0.5 for sandy and clayey soils, respectively. For
the Bay of Campeche sand, n was estimated to be 0.27 (Taboada et al., 2015). A site-specific constant,
k, was back-calculated using the interpreted shear wave velocity and effective vertical stress near the
bottom of the borehole. Using the above relationship, the shear wave velocity profile was extrapolated to
a depth at which the shear wave velocity was calculated to be 620 m/sec. Below the bottom of the
borehole, it was assumed that the unit weight increased linearly with depth. For this study, the unit weight
profile was linearly extrapolated from the estimated value at the bottom of the borehole to a value of 22.0
kN/m3 at the depth at which the shear wave velocity was calculated to be 620 m/sec.
Input to SHAKE program and results
The effect of using different types of shear modulus degradation and material damping ratio curves on the
acceleration spectra at the surface of the soil deposit presented in Figure 22 was investigated with the
computer program SHAKE2000 and four different sets of G/Gmax and Damping ratio curves, including
the model formulated in this study for the Bay of Campeche sands. The four sets of curves used for
20 OTC-26878-MS
comparison are Seed and Idriss (1970) mean estimate for sand, Vucetic and Dobry (1991) for plasticity
index (PI) equals to zero, Darendeli (2001) for PI equals to zero and for different confining stresses, and
finally the model formulated in this study for Bay of Campeche sand.
The soil profile used for the SHAKE analysis was divided in 42 layers with varing thicknesses. The
layers at the top of soil column were thinner than the layers at the bottom of the soil column. The
submerged unit weight used to obtain the total unit weight for the SHAKE analysis are presented in Figure
22 and the shear wave velocity for the layers found upto the end of the soil boring at a depth of 131.1 m
below seafloor is presented in Figure 23.
Since the selected input motion was processed and modified to be compatible with a target spectrum
that was developed for bedrock sites, this motion was defined as ⬙outcrop⬙ bedrock motion and was input
at a depth where the shear wave velocity is estimated to be about 620 m/sec. At the selected platform soil
boring site, the borehole performed was terminated at a depth of about 131.1 meters where the shear wave
velocity is estimated at about 465.0 m/s. The available data were extrapolated below the termination
depths to estimate the depths at which a bedrock–like medium would be present. For the platform soil
boring site, the depth at which input motions were applied was estimated to be 217.6 meters below the
seafloor. Details regarding the procedures used to estimate the input depth (i.e., the base of the site
response model) and the extrapolation of the shear wave velocity and unit weight beyond the bottom of
the borehole are provided in the previous section.
The modulus degradation and damping models proposed by Seed and Idriss (1970), and Vucetic and
Dobry (1991) are independent of confinig stress therefore a single degradation curve and damping curve
were implemented throughout the soil profile all the way to the bottom of the borehole. The model
proposed by Darendeli (2001) and the current proposed model for the Bay of Campeche sand presented
in this study are dependent on confining stresses therefore six different shear modulus degradation curves
and damping ratio curves were used along the soil profile. The following six effective confining stresses
were used to evaluate the shear modulus degradation and material damping curves: 40, 100, 200, 400, 685
and 1000 kPa.
The soils below the bottom of the borehole were assumed to be essentially linear. Consequently, no modulus
degradation (i.e. G/Gmax ⫽ 1) and a damping ratio of 2 percent were used to characterize those soil layers.
Figure 25 shows for 200 – and 500 –year return period events, corresponding to the ELE and ALE
events, respectively, the estimated 5–percent damped response spectra at seafloor calculated from the
SHAKE analyses using the shear modulus degradation and damping ratio curves based on Seed and Idriss
(1970) mean curve, Vucetic and Dobry (1991) for PI ⫽ 0, Darendeli (2001) for PI ⫽ 0 and the model
formulated in this study for Bay of Campeche sand. This figure also shows the response spectra at seafloor
computed with the non-linear site resonse analysis program DEEPSOIL.
Figure 25—Acceleration response spectra at seafloor for the ELE (PGA ⴝ 0.06g) and ALE (PGAⴝ0.13g) events
OTC-26878-MS 21
Profiles of normalized shear modulus, damping ratio, peak shear strain, and peak acceleration with depth
were calculated for each of the four sets of shear modulus degradation and damping ratio curves considered in
the SHAKE analysis. Figure 26 shows the results for 500 –year return period, corresponding to the ALE event
from each of those cases from the SHAKE analysis, as well as the profiles of peak shear strain and peak
acceleration from the DEEPSOIL analysis from the ground surface down to the depth of 180.0 m.
Figure 26 —Profiles of dynamic parameters and peak ground response for ALE conditions
It can be observed in Figure 25 that the acceleration response spectrum at seafloor with highest
amplitude computed from the SHAKE analysis is obtained when the set of G/Gmax and damping ratio
curves recommended in this study for the Bay of Campeche sands are used since they present the most
linear response in the G/Gmax curve (see the comparison of curves presented in Figure 20) which produced
the highest computed profile of G/Gmax presented in Figure 26, and the lowest damping ratio curve of the
four sets of curves considered in the analysis (see Figure 20) which generated the lowest computed profile
of damping ratio presented in Figure 26.
The acceleration response spectrum at seafloor with the lowest amplitude computed from the SHAKE
analysis is obtained when the G/Gmax and damping ratio curves presented by Vucetic and Dobry (1991)
are used since they have the largest non-linear response in terms of G/Gmax curve (see Figure 20) which
produced the lowest computed profile of G/Gmax presented in Figure 26, and the lowest damping ratio
curve from the four sets of curves considered in the analysis (see Figure 20) which produced one of the
highest computed profile of damping ratio below 85.0 m presented in Figure 26, and slightly lower from
85.0 m and above than the highest computed damping ratio profile obtained using Seed and Idriss (1970)
mean curves.
22 OTC-26878-MS
The amplitude of the acceleration response spectrum computed at seafloor with SHAKE using the
curves proposed by Darendeli (2001) is lower than the amplitude of the response spectrum computed
when the curves G/Gmax and damping ratio recommended in this study for the Bay of Campeche sand are
used. This difference in amplitude is more pronounced for the ALE event, as observed in Figure 25.
A comparison between non-linear and linear-equivalent methods using the shear modulus degradation
and damping ratio curves developed in this study for the Bay of Campeche sands was performed using
the site response analysis programs DEEPSOIL and SHAKE2000, respectively. The results of both
programs in terms of acceleration spectra computed at seafloor are presented in Figure 25, and the
computed profiles of peak shear strain and peak acceleration with depth are presented in Figure 26.
As expected, the equivalent linear and non-linear analyses provide different results. The difference
between the two programs, DEEPSOIL and SHAKE2000, in the amplitude of the acceleration response
spectrum at seafloor (see Figure 25) is not significant for the ELE event of about 15% at periods of about
0.34 and 0.72 seconds since the developed shear strains are small of the order of 0.03% in the top 60 m
and the non-linear response is not very important at this very low shear strain level. However, the
difference in the amplitude of the computed acceleration spectrum at seafloor between both programs is
important of about 32% for the ALE event at periods of about 0.34 and 0.72 seconds, as observed in
Figure 25, since the developed shear strains in the top 60 m are rather large in some instances of about
0.35% (one order of magnitude compare to the ELE event), as observed in Figure 26, and at this shear
strain level the non-linearities in the response become more important.
Comparisons between the predictive equations developed in this study and the earlier previously
published curves show that the predictive equation of this study of G/Gmax are more linear and the
predictive equation of damping ratio provide less damping than the previously published curves for sands.
The predictive equations for G/Gmax and damping ratio are shown to make predictions that are accurate
within one standard deviation of random error, as determined from the 1872 data points.
As illustrated in the application presented at the end of the paper, the highest amplitude in the acceleration
spectra at seafloor is computed when the predictive equations presented in this study are used and the lowest
when the Vucetic and Dobry (1991) curves are used, illustrating the importance of considering the confining
stress in the shear modulus degradation curves and the damping ratio curves. As expected, it is found that when
using the predictive equations presented in this study, the differences between linear-equivalent and non-linear
site response analyses are found to be more important at high shear strain levels of the order of 0.35% and
almost insignificant when the shear strains are very low of the order of 0.03%.
The predictive equations developed in this paper are not applicable for carbonate sands or for sands that
exhibit a reduction of material damping ratio at large shear strains where a dilative response is observed,
and the hysteresis loops acquire a characteristic ⬙banana⬙ shape.
References
Darendeli, B.M. (2001). Development of a new family of normalized modulus reduction and material damping curves.
PhD dissertation, University of Texas at Austin, Texas, USA.
Gonzalez Blandon, C.M., and Romo Organista, M.P. (2011). Estimation of dynamic properties of clays. Revista de
Ingenieria Sismica, No. 84, pp 1–23. Mexico City, D.F., Mexico (in Spanish).
GulfBase (2012). http://www.gulfbase.org/facts.php
Harding, B.O., and Drnevich, V.P. (1972). Shear modulus and damping in soils. Journal of the Soil Mechanics Foundation
Division, ASCE 98(7), pp 667–692.
Harding, B.O., and Black, W.L. (1966). Sand stiffness under various triaxial stresses. Journal of the Soil Mechanics
Foundation Division. ASCE 92, No. SM2, pp 667–692.
Hashash, Y.M.A., Musgrove, M.I., Harmon, J.A., Groholski, D.R., Phillips, C.A., and Park, D. (2015). DEEPSOIL 6.1,
User Manual. Urbana, Illinois, Board of Trustees of University of Illinois at Urbana-Champaign.
Idriss, I.M., and Sun, J.I. (1992). User’s manual for SHAKE91, Center for Geotechnical Modeling, Department of Civil
and Environmental Engineering, University of California, Davis, 13 pp. (plus Appendices).
Ordoñez, G.A. (2000). SHAKE2000: a computer program for the 1-D analysis of geotechnical earthquake engineering
problems ( http://www.shake2000.com/index.htm).
Oztoprak, S., and Bolton, M.D. (2013). Stiffness of sands through a laboratory database. Geotechnique, Vol. 63, No. 1,
pp 54 –70.
Roblee, C., and Chiou, B. (2004). A proposed geoindex model for design selection of non-linear properties for site
response analysis. Proc. NSF/PEER International Workshop on Uncertanties in Nonlinear Soil Properties and their
Impact on Modeling Dynamic Soil Response, University of California at Berkeley, Berkeley, California, (
http://peer.berkeley.edu/lifelines/Workshop304/).
Schnabel, P.B., Lysmer, J., and Seed, H.B. (1972). SHAKE: A Computer program for earthquake response analysis of
horizontally layered sites,⬙ Report No. EERC 72-12, University of California, Berkeley.
Seed, H.B., and Idriss, I.M. (1970). Soil moduli and damping factors for dynamic response analysis. Report No. EERC
70-10, Earthquake Engineering Research Center, Berkeley, California.
Taboada, V.M., Cruz, D., Barrera, P., Renovato, S.D., Hernández, J.M., and Gan, K.C., 2015. Predictive equations of shear
wave velocity for Bay of Campeche Sand, Proceedings of the third International Symposium on Frontiers in Offshore
Geotechnics III (ISFOG 2015), Oslo, Norway, 10-12 June 2015, Taylor & Francis Group, London, pp 1115–1120.
Vucetic, M., and Dobry, R. (1991). Effect of soil plasticity on cyclic response. Journal of Geotechnical Engineering, Vol.
117, No. 1, pp 89 –107, January, 1991.
Worzel, J.L., Leyden, R., and Ewing, M. (1968). Newly discovered diapirs in Gulf of Mexico, American Association
Petroleum Geologists, Bulletin 52, pp 1194 –1203.
Zhang, J., Andrus, R.D., and Hsein Juang, C. (2005). Normalized shear modulus and material damping ratio relationships.
Journal of Geotechnical and Geoenvironmental Engineering, ASCE, Vol. 131, No. 4, April 1, 2005.