Technical Manual for Strata

Albert R. Kottke
Xiaoyue Wang
Ellen M. Rathje

Geotechnical Engineering Center

Department of Civil, Architectural, and Environmental Engineering
University of Texas
October 2013
ii
 ABSTRACT

The computer program Strata performs equivalent linear site response analysis in the frequency domain
using time domain input motions or random vibration theory (RVT) methods, and allows for
randomization of the site properties. The following document explains the technical details of the
program, as well as provides a user's guide to the program.

Strata is distributed under the GNU General Public License which can be found
here:http://www.gnu.org/licenses/.

iii
iv
 ACKNOWLEDGEMENTS

This project was sponsored by the Pacific Earthquake Engineering Research Center’s Program of Applied
Earthquake Engineering Research of Lifelines Systems supported by the California Energy Commission,
California Department of Transportation, and the Pacific Gas and Electric Company.

This work made use of the Earthquake Engineering Research Centers Shared Facilities supported
by the National Science Foundation, under award number EEC-9701568 through the Pacific Earthquake
Engineering Research Center (PEER). Any opinions, findings, and conclusions or recommendations
expressed in this material are those of the author(s) and do not necessarily reflect those of the National
Science Foundation.

Additional support was provided by the Nuclear Regulatory Commission.

v
vi
 TABLE OF CONTENTS

Abstract ........................................................................................................................................................ iii

Acknowledgements....................................................................................................................................... v
Table of Contents ........................................................................................................................................ vii
List of Figures ............................................................................................................................................... ix
List of Tables ...............................................................................................................................................xiii
1 Introduction .......................................................................................................................................... 1
2 Site Response Analysis .......................................................................................................................... 3
2.1 Equivalent Linear Site Response Analysis ..................................................................................... 3
2.1.1 Linear Elastic Wave Propagation........................................................................................... 3
2.1.2 Equivalent-Linear Analysis .................................................................................................... 7
2.1.3 Dynamic Soil Properties ...................................................................................................... 10
2.2 Site Response Methods .............................................................................................................. 15
2.2.1 Time Series Method ............................................................................................................ 15
2.2.2 Random Vibration Theory Method ..................................................................................... 17
3 Variation of Site Properties ................................................................................................................. 27
3.1 Introduction ................................................................................................................................ 27
3.2 Random Variables ....................................................................................................................... 27
3.3 Statistical Models for Soil Properties .......................................................................................... 29
3.3.1 Layering and Velocity Model........................................................................................... 29
3.3.2 Depth to Bedrock Model ..................................................................................................... 40
3.3.3 Non-Linear Soil Properties Model ....................................................................................... 41
4 Using Strata ......................................................................................................................................... 43
4.1 Strata Particulars ......................................................................................................................... 43
4.1.1 Auto-Discretization of Layers .............................................................................................. 43
4.1.2 Location of Input Motions................................................................................................... 44
4.1.3 Interaction with Tables ....................................................................................................... 45
4.1.4 Non-Linear Curves ............................................................................................................... 46
4.1.5 Input Motions Dialog .......................................................................................................... 48
4.1.6 Interacting with Results ...................................................................................................... 50

vii
 4.2 Glossary of Fields ........................................................................................................................ 52
4.2.1 General Settings Page ......................................................................................................... 52
4.2.2 Soil Types Page .................................................................................................................... 55
4.2.3 Soil Profile Page................................................................................................................... 58
4.2.4 Motion(s) Page .................................................................................................................... 62
4.2.5 Output Specification Page................................................................................................... 69
4.3 Examples ..................................................................................................................................... 76
4.3.1 Example 1: Basic time domain ............................................................................................ 77
4.3.2 Example 2: Time Series with Multiple Input Motions ......................................................... 81
4.3.3 Example 3: RVT and Site Variation ...................................................................................... 83
5 References .......................................................................................................................................... 87

viii
 LIST OF FIGURES

Figure 2.1: The notation used in the wave equation. 4

Figure 2.2: Nomenclature for the theoretical wave propagation. 5

Figure 2.3: The input to surface transfer functions site in Table 2.1 considering different types of input. 6

Figure 2.4: The input to surface transfer functions site in Table 2.1 considering different types of input. 7

Figure 2.5: An example of the strain-time history and effective strain ( eff). 9

Figure 2.6: An example shear-wave velocity profile. 11

Figure 2.7: Examples of shear modulus reduction and material damping curves for soil. 12

Figure 2.8: The nonlinear soil properties predicted by the Darendeli (2001) model. 13

Figure 2.9: The mean and mean nonlinear soil properties predicted by Darendeli (2001). 14

Figure 2.10: Time domain method sequence: (a) input acceleration-time series, (b) input Fourier
amplitude spectrum, (c) transfer function from input to surface, (d) surface Fourier amplitude
spectrum, and (e) surface acceleration-time series (after Kramer 1996). 16

Figure 2.11: The comparison between the target response spectrum and the response spectrum
computed with RVT. 22

Figure 2.12: The relative error between the computed response spectra and the target response
spectrum. 22

Figure 2.13: The FAS computing through the inversion process. 23

Figure 2.14: RVT method sequence: (a) input Fourier amplitude spectrum, (b) transfer function from
input to surface, and (c) surface Fourier amplitude spectrum. 25

Figure 3.1: Two variables with a correlation coefficient of: (a) 0.0, (b), 0.99, and (c) -0.7. 28

Figure 3.2: A 10 layer profile modeled by a homogeneous Poisson process with = 1. 31

Figure 3.3: Transforming from a constant rate of = 1 to a constant rate of = 0.2. 31

Figure 3.4: A 10 layer profile modeled by a homogeneous Poisson process with = 0.2. 32

Figure 3.5: Toro (1995) layering model. (a) The occurrence rate (λ) as a function of depth (d), and (b) the
expected layer thickness (h) as a function of depth. 33

Figure 3.6: Transformation between a homogeneous Poisson process with rate 1 to the Toro (1995) non-
homogeneous Poisson process. 34
ix
Figure 3.7: A layering simulated with the non-homogeneous Poisson process defined by Toro (1995). 34

Figure 3.8: Ten generated shear-wave velocity (vs) profiles for a USGS C site class. (a) Using generic
layering and median vs, (b) using user defined layering and median vs. 40

Figure 3.9: Generated nonlinear properties assuming perfect negative correlation. 42

Figure 4.1: Layer Discretization group box. 44

Figure 4.2: Location selection: (a) top of bedrock, (b) switching to a fixed depth, and (c) fixed depth
specified as 15. 44

Figure 4.3: By clicking on the button circled in red all rows in the table are selected. 45

Figure 4.4: The nonlinear curve manager. 47

Figure 4.5: An initial view of the Input Motions Dialog. 49

Figure 4.6: An example a completed Input Motions Dialog. 49

Figure 4.7: Using the Output view to examine the results of a calculation. 51

Figure 4.8: Screenshot of the Project group box. 52

Figure 4.9: Screenshot of the Type of Analysis group box. 53

Figure 4.10: Screenshot of the Site Property Variation group box. 53

Figure 4.11: Screenshot of the Calculation Parameters group box. 54

Figure 4.12: Screenshot of the Layer Discretization group box. 54

Figure 4.13: Screenshot of the Soil Types group box. 55

Figure 4.14: Screenshot of the Bedrock Layer group box. 55

Figure 4.15: Screenshot of the Water Table Depth group box. 56

Figure 4.16: Screenshot of the Nonlinear Curve Variation Parameters group box. 56

Figure 4.17: Screenshot of the Darendeli and Stokoe Model Parameters group box. 57

Figure 4.18: Screenshot of the Nonlinear Property group box. 58

Figure 4.19: Screenshot of the Site Profile group box. 58

Figure 4.20: Screenshot of the Site Profile group box. 59

Figure 4.21: Screenshot of the Velocity Variation Parameters group box. 60

x
Figure 4.22: Screenshot of the Layer Thickness Variation Parameters group box. 61

Figure 4.23: Screenshot of the Bedrock Depth Variation Parameters group box. 62

Figure 4.24: Screenshot of the Motion Input Location group box. 62

Figure 4.25: Screenshot of the Input Motions group box (for Time Series Approach). 63

Figure 4.26: Screenshot of the RVT Motion Type drop-down list. 64

Figure 4.27: Screenshot of the User Defined FAS dialog. 64

Figure 4.28: Screenshot of the Response Spectrum Compatible dialog. 65

Figure 4.29: Screenshot of the Frequency Parameters… dialog. 66

Figure 4.30: Screenshot of the Seismological Source Theory dialog – Point Source Parameters tab. 67

Figure 4.31: Screenshot of the Seismological Source Theory dialog – Crustal amplification tab. 68

Figure 4.32: Screenshot of the Crustal Model dialog if Calculated is selected. 69

Figure 4.33: Screenshot of the Profiles tab. 70

Figure 4.34: Screenshot of the Time Series drop-down list. 71

Figure 4.35: Screenshot of the Time Series tab. 72

Figure 4.36: Screenshot of the Response and Fourier Spectra drop-down list. 72

Figure 4.37: Screenshot of the Response and Fourier Spectra tab. 72

Figure 4.38: Screenshot of the Ratios drop-down list. 73

Figure 4.39: Screenshot of the Ratios tab. 73

Figure 4.40: Screenshot of the Soil Type tab. 74

Figure 4.41: Screenshot of the Response Spectrum group box. 74

Figure 4.42: Screenshot of the Frequency Properties group box. 75

Figure 4.43: Screenshot of the Logging Properties group box. 75

Figure 4.44: The shear-wave velocity profile of the Sylmar County Hospital Parking Lot site (Chang,
1996). 77

Figure 4.45: Example of a plot with multiple responses. 83

xi
xii
 LIST OF TABLES

Table 2.1: The site properties of an example site. ........................................................................................ 6

Table 2.2: The values of the RVT calculation for the input motion. ........................................................... 24

Table 2.3: The values of the RVT calculation for the surface motion. ........................................................ 24

Table 3.1: The categories of the geotechnical subsurface conditions (third letter) in the Geomatrix site
classification Toro (1995). .................................................................................................................. 37

Table 3.2: Site categories based on (Toro (1995)). ............................................................................. 37

Table 3.3: Coefficients for the Toro (1995) model...................................................................................... 38

Table 3.4: Median shear-wave velocity (m/s) based on the generic site classification. ............................ 39

Table 4.1: Soil profile at the Sylmar County Hospital Parking Lot site Chang (1996). The mean effective
stress ( ) is computed assuming a of 1/2 and a water table depth of 46 meters. .................... 76

Table 4.2: Suite of input motions used in Example 2.................................................................................. 82

xiii
xiv
1 Introduction

The computer program Strata performs equivalent linear site response analysis in the frequency
domain using time domain input motions or random vibration theory (RVT) methods, and allows for
randomization of the site properties. Strata was developed with financial support provided by the
Lifelines Program of the Pacific Earthquake Engineering Research (PEER) Center under grant SA5405-
15811 and funding from the Nuclear Regulatory Commission. Strata is distributed under the GNU
General Public License which can be found here: http://www.gnu.org/licenses/.

The following document provides explanation of the technical details of the program. Chapter 2
provides an introduction to equivalent-linear elastic wave propagation using both time series and
random vibration theory methods. Using the time series method, a single motion is propagated through
the site to compute the strain compatible ground motion at the surface of the site or at any depth in the
soil column. Using random vibration theory, the expected maximum response is computed from a mean
Fourier amplitude spectrum (amplitude only) and duration. Chapter 3 introduces random variables and
the models that Strata uses to govern the variability of the site properties (nonlinear properties, layering
thickness, shear-wave velocity, and depth to bedrock). Chapter 4 discusses in the interaction with Strata
along with several tutorials that introduce the features found in Strata.

1
2
2 Site Response Analysis

Strata computes the dynamic site response of a one-dimensional soil column using linear wave
propagation with strain dependent dynamic soil properties. This is commonly referred to as the
equivalent linear analysis method, which was first used in the computer program SHAKE (Schnabel et al.
1972; Idriss & Sun 1992). Similar to SHAKE, Strata only computes the response for vertically propagating,
horizontally polarized shear waves propagated through a site with horizontal layers.

The following chapter introduces strain dependent soil properties, linear-elastic wave
propagation through a layered medium, and the equivalent linear approach to site response analysis.

2.1 Equivalent Linear Site Response Analysis

2.1.1 Linear Elastic Wave Propagation

For linear elastic, one-dimensional wave propagation, the soil is assumed to behave as a Kelvin-Voigt
solid, in which the dynamic response is described using a purely elastic spring and a purely viscous
dashpot (Kramer, 1996). The solution to the one-dimensional wave equation for a single wave frequency
(ω) provides displacement (u) as a function of depth (z) and time (t) (Kramer, 1996):

( ) [( )] [( )] (2.1)

In equation (1.1), A and B represent the amplitudes of the upward (-z) and downward (+z) waves,
respectively (Figure 2.1). The complex wave number (k*) in equation (2.1) is related to the shear
modulus (G), damping ratio (D), and mass density (ρ) of the soil using:

(2.2)

√ (2.3)

( √ ) ( ) (2.4)

3
G* and vs* are called the complex shear modulus and complex shear-wave velocity, respectively. If the
damping ratio (D) is small (<10-20%), then the approximation of the complex shear modulus in equation
(2.4) is appropriate. Strata uses the complete definition of the complex shear-modulus, not the
approximation, in the calculations.

Figure 2.1: The notation used in the wave equation.

Equation (2.1) applies only to a single layer with uniform soil properties and the wave
amplitudes (A and B) can be computed from the layer boundary conditions. For a layered system, shown
in Figure 2.1, the wave amplitudes are calculated using recursive formulas developed by maintaining
compatibility of displacement and shear stress at the layer boundaries. Using these assumptions, the
following recursive formulas are developed (Kramer, 1996):

( ) ( ) ( ) ( )
(2.5)
( ) ( ) ( ) ( )

where m is the layer number, hm is the layer height and αm* is the complex impedance ratio. The
complex impedance ratio is defined as:

(2.6)

At the surface of the soil column (m=1), the shear stress must equal zero and the amplitudes of the
upward and downward waves must be equal (A1=B1).

4
 Figure 2.2: Nomenclature for the theoretical wave propagation.

The wave amplitudes (A and B) within the soil profile are calculated at each frequency (assuming
known stiffness and damping within each layer) and used to compute the response at the surface of a
site. This calculation is performed by setting A1=B1=1.0 at the surface and recursively calculating the
wave amplitudes (Am+1,Bm+1) in successive layers until the input (base) layer is reached. The transfer
function between the motion in the layer of interest (m) and in the rock layer (n) at the base of the
deposit is defined as:

( )
( )( ) (2.7)
( )

where ω is the frequency of the harmonic wave. The transfer function is the ratio of the amplitude of
harmonic motion--either displacement, velocity, or acceleration--between two layers of interest and
varies with frequency. The transfer function (surface motion / within motion) for the site with the
properties presented in Table 2.1 is shown in Figure 2.3. The locations of the peaks in the transfer
function are controlled by the modes of vibration of the soil deposit. The peak at the lowest frequency
represents the fundamental (i.e. first) mode of vibration and results in the largest amplification. The
peaks at higher frequencies are the higher vibrational modes of the site.

For the example site (Table 2.1), the first mode natural frequency is 1.75 Hz (site period = 0.57
s). In the transfer function (Figure 2.3), the peak with the largest amplification occurs at this frequency.
The amplitudes of the peaks are controlled by the damping ratio of the soil. As the damping of the
system increases, the amplitudes of the peaks decrease which results in less amplification.

5
 Table 2.1: The site properties of an example site.

Property Rock Soil

Mass Density (ρ) 2.24 g/cm3 1.93 g/cm3
Height (h) Inf 50 m
Shear-wave Velocity (vs) 1500 m/s 350 m/s
Damping ratio (D) 1% 7%

Figure 2.3: The input to surface transfer functions site in Table 2.1 considering different types of
input.

The response at the layer of interest is computed by multiplying the Fourier amplitude spectrum
of the input rock motion by the transfer function:

( ) ( ) ( ) (2.8)

where Yn is the input Fourier amplitude spectrum at layer n and Ym is the Fourier amplitude spectrum at
the top of the layer of interest. The Fourier amplitude spectrum of the input motion can be defined
using a variety of methods and is discussed further in Sections 2.2.1 and 2.2.2.

One issue that must be considered is that the input Fourier spectrum typically represents a
motion recorded on rock at a free surface (i.e., the ground surface), where the upgoing and downgoing
wave amplitudes are equal ( ), rather than on rock at the base of a soil deposit, where the wave
6
amplitudes are not equal (Figure 2.4). The change in boundary conditions ( for a free surface,
at the base of a soil deposit) must be taken into account. The motions at any free surface are
referred to as outcrop motions and their amplitudes are described by twice the amplitude of the
upward wave ( ). A transfer function can be defined that converts an outcrop motion into a within
motion, and this transfer function can be combined with the transfer function in Equation (2.7) to create
a transfer function that can be applied to recorded outcrop motions on rock (Equation 2.9).

( )
⏟ ⏟ (2.9)

Motions recorded at depth (e.g. recorded in a borehole) are referred to as within motions and for these
motions the transfer function given in equation (2.7) can be used. Figure 2.3 shows the transfer function
(surface motion / outcrop motion) for the site profile presented in Table 2.1 using equation (2.9) where
the input motion is specified as outcrop. In comparison with the surface / within transfer function, the
surface / outcrop transfer function displays less amplification for all modes.

Figure 2.4: The input to surface transfer functions site in Table 2.1 considering different types of
input.

2.1.2 Equivalent-Linear Analysis

The previous section assumed that the soil was linear-elastic. However, soil is nonlinear, such that the
dynamic properties of soil (shear modulus, G, and damping ratio, D) vary with shear strain, and thus the

7
intensity of shaking. In equivalent-linear site response analysis, the nonlinear response of the soil is
approximated by modifying the linear elastic properties of the soil based on the induced strain level.
Because the induced strains depend on the soil properties, the strain compatible shear modulus and
damping ratio values are iteratively calculated based on the computed strain.

A transfer function is used to compute the shear strain in the layer based on the outcropping
input motion. In the calculation of the strain transfer function, the shear strain is computed at the
middle of the layer (z=hm/2) and used to select the strain compatible soil properties. Unlike the previous
transfer functions that merely amplified the Fourier amplitude spectrum, the strain transfer function
amplifies the motion and converts acceleration into strain. The strain transfer function based on an
outcropping input motion is defined by:

( )
( )
̈ ( )
(2.10)
( ( ) ( ))
( )

The strain Fourier amplitude spectrum within a layer is calculated by applying the strain transfer
function to the Fourier amplitude spectrum of the input motion. The maximum strain within the layer is
derived from this Fourier amplitude spectrum -- either through conversion to the time domain or
through RVT methods, further discussed in Section 2.2. However, it is not appropriate to use the
maximum strain within the layer to compute the strain-compatible soil properties, because the
maximum strain only occurs for an instant in time. Instead, an effective strain (γeff) is calculated from the
maximum strain. Typically, the effective strain is 65% of the maximum strain. An example of a strain
time-series and the effective strain is shown in Figure 2.5.

8
 Figure 2.5: An example of the strain-time history and effective strain ( eff ).

Equivalent-linear site response analysis requires that the strain dependent nonlinear properties
(i.e. G and D) be defined. The initial (small strain) shear modulus (Gmax) is calculated by:

(2.11)

where ρ is the mass density of the site, and vs is the measured shear-wave velocity. Characterizing the
nonlinear behavior of G and D is achieved through modulus reduction and damping curves that describe
the variation of G/Gmax and D with shear strain (discussed in the next section). Using the initial dynamic
properties of the soil, equivalent-linear site response analysis involves the following steps:

1. The wave amplitudes (A and B) are computed for each of the layers.
2. The strain transfer function is calculated for each of the layers.
3. The maximum strain within each layer is computed by applying the strain transfer function to
the input Fourier amplitude spectrum and finding the maximum response (see Section 2.2).
4. The effective strain (γeff) is calculated from the maximum strain within each layer.
5. The strain compatible shear modulus and damping ratio are recalculated based on the new
estimate of the effective strain within each layer.
6. The new nonlinear properties (G and D) are compared to the previous iteration and an error is
calculated. If the error for all layers is below a defined threshold the calculation stops.

After the iterative portion of the program finishes, the dynamic response of the soil deposit is computed
using the strain-compatible properties.
9
2.1.3 Dynamic Soil Properties

In a dynamic system, the properties that govern the response are the mass, stiffness, and damping. In
soil under seismic shear loading, the mass of the system is characterized by the mass density (ρ) and the
layer height (h), the stiffness is characterized by the shear modulus (G), and the damping is
characterized by the viscous damping ratio (D). The dynamic behavior of soil is challenging to model
because it is nonlinear, such that both the stiffness and damping of the system change with shear strain.
Section 2.1.2 introduced equivalent-linear site response analysis in which the nonlinear response of the
soil was simplified into a linear system that used strain-compatible dynamic properties (G and D). The
analysis requires that the strain dependence of the nonlinear properties within a layer be fully
characterized.

Defining the mass density of the system is a straight forward process because the density of soil
falls within a limited range for soil and a good estimate of the mass density can be made based on soil
type. Characterization of the stiffness and damping properties of soil is more complicated, the most
rigorous approach requiring testing in both the field and laboratory.

The shear modulus and material damping of the soil are characterized using the small strain
shear modulus (Gmax), modulus reduction curves that relate G/Gmax to shear strain, and damping ratio
curves that relate D to shear strain. The small strain shear modulus is best characterized by in situ
measurement of the shear-wave velocity as a function of depth. An example shear-wave velocity profile
is shown in Figure 2.6. The profile tends to be separated into discrete layers with a generally increasing
shear-wave velocity with increasing depth. Examples of modulus reduction and damping curves for soil
are shown in Figure 2.7. These curves show a decrease in the soil stiffness and an increase in the
damping ratio with an increase in shear strain.

10
 Figure 2.6: An example shear-wave velocity profile.

Modulus reduction and damping curves may be obtained from laboratory measurements on soil
samples or derived from empirical models based on soil type and other variables. One of the most
comprehensive empirical models was developed by Darendeli (2001) and is included with Strata. The
model expands on the hyperbolic model presented by Hardin and Drnevich (1972) and accounts for the
effects of confining pressure ( ), plasticity index (PI), over-consolidation ratio (OCR), frequency (f), and
number of cycles of loading (N) on the modulus reduction and damping curves.

11
 Figure 2.7: Examples of shear modulus reduction and material damping curves for soil.

In the Darendeli (2001) model, the shear modulus reduction curve is a hyperbola defined by:

(2.12)
( )

where a is 0.9190, γ is the shear strain and γref is the reference shear strain. The reference shear strain
(not in percent) is computed from:

( ) ( ) (2.13)

where is the mean effective stress and pa is the atmospheric pressure in the same units as . In the
model, the damping ratio is calculated from the minimum damping ratio at small strains (Dmin) and from
the damping ratio associated with hysteretic Masing behavior (DMasing). The minimum damping is
calculated from:

( ) ( ) ( )( ) (2.14)

where f is the excitation frequency (Hz). The computation of the Masing damping requires the
calculation of the area within the stress-strain curve predicted by the shear modulus reduction curve.
The integration can be approximated by:

( ) (2.15)

where:

12
 ( )
( ) { [ ] } (2.16)

(2.17)

The minimum damping ratio in equation (2.14) and the Masing damping in equation (2.16) are
combined to compute the total damping ratio (D) using:

( ) (2.18)

where b is defined as:

(2.19)

where N is the number of cycles of loading. In most site response applications, the number of cycles (N)
and the excitation frequency (f) in the model are defined as 10 and 1, respectively. Figure 2.8 shows the
predicted nonlinear curves for a sand (PI=0, OCR=1) at an effective confining pressure of 1 atm.

Figure 2.8: The nonlinear soil properties predicted by the Darendeli (2001) model.

A Bayesian approach was used in the Darendeli (2001) model to calculate the model
coefficients. One of the unique aspects of this model is that the scatter of the data about the mean
estimate is quantified. In the Darendeli (2001) model, the variability about the mean value is assumed to
be normally distributed. The normal distribution is described using a mean and standard deviation. The

13
mean values are calculated from equations (2.12) and (2.18). The standard deviation is a function of the
amplitude of the nonlinear property (i.e. and D). The standard deviation of the normalized
shear modulus (σNG) is computed by:

( )
√
( )
( ) ( ) (2.20)

√ ( ⁄ )

This model results in small σNG when is close to 1 or 0 and relatively large σNG when is
equal to 0.5. The standard deviation of the damping ratio (σD) is computed by:

( ) ( )√ ( )
(2.21)
√ ( )

In the damping ratio model, σD increases with increasing damping ratio. Using these definitions for the
standard deviation, the +σ modulus reduction and damping curve for sand at a confining pressure of 1
atm are shown in Figure 2.9.

Figure 2.9: The mean and mean nonlinear soil properties predicted by Darendeli (2001).

14
2.2 Site Response Methods

The previous section introduced transfer functions which transform the input Fourier amplitude
spectrum (FAS) into a FAS of strain or acceleration, and transfer functions can also be derived to
compute the response of a single-degree-of-freedom oscillator. In both the time domain and random
vibration theory methods, the same transfer functions are applied to the input FAS. The difference in
the methods is in how this FAS in the frequency domain is converted into time domain information.

2.2.1 Time Series Method

In the time series method, an input acceleration-time history is provided and the input FAS is computed
from that time series using the fast Fourier transform (FFT) to compute the discrete Fourier
transformation on the provided time series. The computed FAS is complex valued, and can be converted
into amplitude and phase information. Strata uses the free and open-source FFTW library
(http://www.fftw.org). The inverse discrete Fourier transform is used to compute a time series for a
given FAS. The details of the FFT process are not discussed here, but can be found on the FFTW
webpage.

In Strata, the time series is padded with zeros to obtain a number of points that is a power of
two. If a time series contains a power of two values, then it is padded with zeros until the next power of
two.

After the FAS of the motion has been computed it is possible to perform site response analysis
with the motion. The following is a summary of the steps to compute the surface acceleration time-
series for the site described in Table 2.1 (after Kramer 1996):

1. Read the acceleration-time series file (Figure 2.10a).

2. Compute the input FAS with the Fast Fourier transformation (FFT) (FIGURE 2.10b, only
amplitude is shown).
3. Compute the transfer function for the site properties (Figure 2.10c, only amplitude is
shown).
4. Compute the surface FAS by applying the transfer function to the input FAS (FIGURE 2.10d,
only amplitude is shown).
5. Compute the surface acceleration-time series through the inverse FFT of the surface FAS
(FIGURE 2.10e).

15
Figure 2.10: Time domain method sequence: (a) input acceleration-time series, (b) input Fourier
amplitude spectrum, (c) transfer function from input to surface, (d) surface Fourier amplitude
spectrum, and (e) surface acceleration-time series (after Kramer 1996).

16
2.2.2 Random Vibration Theory Method

The random vibration theory (RVT) approach to site response analysis was first proposed in the
engineering seismology literature (e.g. Schneider et al. (1991)) and has been applied to site response
analysis (Silva et al. 1997, Rathje and Ozbey 2006, Rathje and Kottke 2008). RVT does not utilize time
domain input motions, but rather initiates all computations with the input FAS (amplitude only, no
phase information). Because RVT does not have the accompanying phase angles to the Fourier
amplitudes, a time history of motion cannot be computed. Instead, extreme value statistics are used to
compute peak time domain parameters of motion (e.g. peak ground acceleration, spectral acceleration)
from the Fourier amplitude information. Due to RVT's stochastic nature one analysis can provide a
median estimate of the site response with a single analysis and without the need for time domain input
motions.

2.2.2.1 RVT Basics

Random vibration theory can be separated into two parts: (1) conversion between time and frequency
domain using Parseval's theorem, and (2) estimation of the peak factor using extreme value statistics.
Consider a time varying signal x(t) with its associated Fourier amplitude spectrum, X(f). The root-mean-
squared value of the signal (xrms) is a measure of its average value over a given time period, Trms, and is
computed from the integral of the times series over that time period:

√ ∫ [ ( )] (2.22)

Parseval's theorem relates the integral of the time series to the integral of its Fourier Transform, such
that equation (2.22) can be written in term of the FAS of the signal:

√ ∫ | ( )| √ (2.23)

where m0 is defined as the zero-th moment of the FAS. The N-th moment of the FAS is defined as:

∫ ( ) | ( )| (2.24)

The peak factor (PF) represents the ratio of the maximum value of the signal (xmax) to its rms value (xrms),
such that if xrms and the PF are known, then xmax can be computed using:

(2.25)

Cartwright and Longuet-Higgins (1956) studied the statistics of ocean wave amplitudes, and considered
the probability distribution of the maxima of a signal to develop expressions for the PF in terms of the
17
characteristics of the signal. Cartwright and Longuet-Higgins (1956) derived an integral expression for
the expected values of the peak factor in terms of the number of extrema (Ne) and the bandwidth (ξ) of
the time series (Boore 2003):

[ ] √ ∫ [ ] (2.26)

where the bandwidth is defined as:

√ (2.27)

and the number of extrema is defined as:

√ (2.28)

Boore (2003) illustrated the need to modify the duration used in the rms calculation when considering
requires modification for spectral acceleration to account for the enhanced duration due to the
oscillator response. Generally, adding the oscillator duration to the ground motion duration will suffice,
except in cases where the ground motion duration is short (Boore & Joyner, 1984). Boore and Joyner
(1984) recommend the following expressions to define Trms:

( ) (2.29)

(2.30)

(2.31)

where T0 is the oscillator duration, Tn is the oscillator natural period, and β is the damping ratio of the
oscillator. Based on numerical simulations, Boore and Joyner (1984) proposed n=3 and α=1/3 for the
coefficients in equation (2.29).

2.2.2.2 Defining the Input Motion

The input motion in an RVT analysis is defined by a Fourier amplitude spectrum (FAS) and ground
motion duration (Tgm). The FAS can be directly computed using seismological source theory (e.g Brune,
(1970, 1971)), or it can be back-calculated from an acceleration response spectrum (see Section
2.2.2.24). When the FAS is directly provided, the frequencies provided with the Fourier amplitude

18
spectrum represent the frequency range used by the program so it is critical that enough points be
provided.

Calculation of the duration for use in RVT analysis can be done using seismological theory or
empirical models. Atkinson and Boore (1995) recommended the following description of ground motion
duration ( ) for the Western United States using seismological theory:

⏟
⏟ (2.32)

where R is the distance in km, and the corner frequency (f0) in hertz is given by:

( ) (2.33)

where Δσ is the stress drop in bar, is the shear-wave velocity in units of km/s, and M0 is the seismic
moment in units of dyne-cm (Brune, 1970). The seismic moment (M0) is related to the moment
magnitude (Mw) by:

( ) (2.34)

For the Eastern United States, Atkinson and Boore (1995) proposed that the path duration effect be
distance dependent:

( )
{ (2.35)
( )
( )

Empirical ground motion duration models such as Abrahamson and Silva (1996) can also be used to
estimate the duration of the scenario event (Tgm). When such a model is applied, it is recommended that
Tgm be taken as time between the build up from 5% to 75% of the normalized arias intensity (D5-75).

2.2.2.3 Source Theory Model

Strata provides functionality for the calculation of a single-corner frequency point source model
originally proposed by Brune (1970) and more recently discussed in Boore (2003). The default values for
the Western United States and the Central and Eastern United States are taken from Campbell (1997).

19
2.2.2.4 Calculation of a FAS from an Acceleration Response Spectrum

The input rock FAS (Y(f)) can be derived from an acceleration response spectrum using an inverse
technique. The inversion technique follows the basic methodology proposed by Gasparini and
Vanmarcke (1976) and further described by Rathje et al. (2005). The inversion technique makes use of
the properties of the single-degree-of-freedom (SDOF) transfer function used to compute the response
spectral values. The square of the Fourier amplitude at the SDOF oscillator natural frequency fn (|Y(fn)|2)
can be written in terms of the spectral acceleration at fn ( ), the peak factor (PF), the rms duration of
the motion (Trms), the square of the Fourier amplitudes (| ( )| ) at frequencies less than the natural
frequency, and the integral of the SDOF transfer function (| ( )| ):

| ( )| ( ∫ | ( )| ) (2.36)
∫ | ( )|

Within equation (2.36), the integral of the transfer function is constant for a given natural frequency and
damping ratio ( ), allowing the equation to be simplified to (Gasparini & Vanmarcke, 1976):

| ( )| ( ∫ | ( )| ) (2.37)
( )

The peak factors in equation (2.37) depend on the moments of the FAS, which is currently undefined. So
the peak factors for all natural frequencies are initially assumed to be 2.5.

Equation (2.37) is applied first to the spectral acceleration of the lowest frequency (longest period)
provided by the user. At this frequency, the FAS integral term in equation (2.37) can be assumed to be
equal to zero. The equation is then applied at successively higher frequencies using the previously
computed values of | ( )| to assess the integral.

To improve the agreement between the RVT-derived response spectrum ( ( )) and the target
response spectrum ( ( )), the RVT-derived FAS is corrected by multiplying it by the ratio of the
two response spectra. This iterative process corrects the FAS from iteration i (| ( )|) using:

( )
| ( )( )| | ( )| (2.38)
( )

Additionally, the newly defined FAS is used to compute appropriate peak factors for each frequency. The
full procedure used to generate a corrected FAS is:

1. Initial FAS is computed using the Gasparini and Vanmarcke (1976) technique (Equation 2.37).
2. The acceleration response spectrum associated with this FAS is computed using RVT.
3. The FAS is corrected using equation (2.38).
4. The peak factors are updated.

20
 5. Using the corrected FAS and new peak factors, a new acceleration response spectrum is
calculated.

This process is repeated until one of three conditions is met:

1. maximum of 30 iterations,
2. a root-mean-square-error of 0.005 is achieved between the RVT response spectrum and the
target response spectrum, or
3. change in the root-mean-square-error is less than 0.001.

This ratio correction works very well in producing a FAS that agrees with the target response spectrum,
but the resulting FAS may have an inappropriate shape at some frequencies, as discussed below.

To demonstrate the inversion process, consider a scenario event of magnitude 7 at a distance of

20 km. The target response spectrum is computed using the Abrahamson and Silva (1997) attenuation
model (Figure 2.11). An initial estimate of the FAS is computed using the Gasparini and Vanmarcke
(1976) method and then the ratio correction algorithm is applied. This methodology (called “Ratio
Corrected”) results in good agreement with the target response spectrum (Figure 2.11), with less than
5% relative error as shown in Figure 2.12. However, the associated FAS slopes up at low and high
frequencies (Figure 2.13). The sloping up at low frequencies can be mitigated by extending the
frequency domain because the spectral acceleration at a given frequency is affected by a range of
frequencies in the FAS.

The frequency domain extension involves expanding frequencies to half of the minimum
frequency and twice the maximum frequency specified in the target response spectrum. For example, if
the target response spectrum is provided from 0.2 to 100 Hz (5 to 0.01 seconds), then the frequencies of
the FAS are defined at points equally spaced in log space from 0.1 to 200 Hz. The resulting response
spectrum essentially displays the same agreement with the target response spectrum (curve labeled
``Ratio and Extrapolated'' in Figures 2.11 and 2.12), but the FAS shows no sloping up at low frequencies
and less sloping up at high frequencies (Figure 2.13).

While the results in Figures 2.11-2.13 would appear to be adequate, it was observed that the
sloping up at high frequencies was affecting the RVT calculation. The peak factor depends on the 4th
moment of the FAS (Equations 2.27, 2.28), which is more sensitive to higher frequencies. Additionally,
seismological theory indicates that the slope of the FAS at high frequencies should be increasingly
negative due to a path-independent loss of the high-frequency motion (Boore 2003). To deal with these
issues, the slope of the FAS at high frequencies is forced down (curve labeled “Ratio, Extrap., & Slope
Forced” in Figure 2.13). The corrected portion of the FAS is computed through linear extrapolation in
log-log space from where the slope deviates from its steepest value by more than 5%. This solution
results in a slight under prediction (~3%) of the peak ground acceleration (Figure 2.11 and Figure 2.12).

21
Figure 2.11: The comparison between the target response spectrum and the response spectrum
computed with RVT.

Figure 2.12: The relative error between the computed response spectra and the target response
spectrum.

22
 Figure 2.13: The FAS computing through the inversion process.

2.2.2.5 Example of the RVT Procedure

The following is an example of random vibration theory applied to site response analysis to estimate the
peak acceleration at the top of the site described in Table 2.1: The earthquake scenario is a magnitude 7
event at a distance of 20 km, as described in the previous section.

1. Empirical relationships are used to specify the input rock response spectrum (Figure 2.11) and
ground motion duration ( 8.2 s).
2. Using the inversion technique, the FAS corresponding to the target response spectrum is
computed (Figure 2.14a). In this example, the peak acceleration of the input motion is
computed with RVT to allow for a comparison in the peak response between the surface and the
input. The RVT calculation results are shown in Table 2.2.

23
 Table 2.2: The values of the RVT calculation for the input motion.

Parameter Value Equation

Moments of FAS (m0, m2, m4) 0.0280, 93.84, 1.738x107 2.28
Bandwidth 0.1346 2.31
Number of extrema (Ne) 1123 2.32
Peak factor (PF) 3.325 2.30
Root-mean-square acceleration (arms) 0.0584 g 2.27
Expected peak acceleration from RVT (amax) 0.1942g 2.29
Target peak acceleration (PGA) 0.20 g ---

3. Compute the transfer function for the site properties (Figure 2.14b).
4. Compute the surface FAS by applying the absolute value of the transfer function to the input
FAS (Figure 2.14c). Using the surface FAS, the expected peak acceleration can be computed
using RVT, as presented in Table 2.3. The calculation shows that the site response increases the
peak ground acceleration by approximately 38%.

Table 2.3: The values of the RVT calculation for the surface motion.

Parameter Value Equation

Moments of FAS (m0, m2, and m4) 0.0635, 39.6356, and 1.6306x107 2.28
Bandwidth 0.3895 2.31
Number of extrema (Ne) 167.414 2.32
Peak factor (PF) 3.0588 2.30
Root-mean-square acceleration (arms) 0.0880 g 2.27
Expected maximum accleration (amax) 0.2692 g 2.29

24
Figure 2.14: RVT method sequence: (a) input Fourier amplitude spectrum, (b) transfer function from
input to surface, and (c) surface Fourier amplitude spectrum.

25
26
3 Variation of Site Properties

3.1 Introduction

A soil profile consists of discrete layers that vary in thickness based on the properties of the soil. The
layers are typically discretized based on the soil type, recorded from borehole samples or inferred from
a shear-wave velocity profile. In seismic site response analysis, each layer is characterized by a thickness,
mass density, shear-wave velocity, and nonlinear properties ( , and D). One of the challenges in
defining values for these properties is the natural variability across a site and the uncertainty in their
measurement. Because the dynamic response of a site is dependent on the soil properties, any variation
in the soil properties will change both the expected surface motion and its standard deviation.

In a simple system, the variability of the components can be analytically combined to quantify
the variability of the complete system, thus allowing for the expected value and variability of the system
response to be computed. In seismic site response analysis, the nonlinear response of the system does
not allow an exact analytic quantification of the variability of the site response. Instead, an estimate of
the expected surface response and its standard deviation due to variations in the soil properties can be
made through Monte Carlo simulations. Monte Carlo simulations estimate the response of a system by
generating parameters of the system based on defined statistical distributions and computing the
response for each set of input parameters. The following chapter introduces Monte Carlo simulations as
applied to site response analysis and presents the models that describe the variability of the layering,
shear-wave velocity, and nonlinear properties ( and D).

3.2 Random Variables

The goal of a Monte Carlo simulation is to estimate the statistical properties of the response of a
complex system. To achieve this goal, each of the properties of the system is selected from defined
statistical distributions and the response of the system is computed. The response is computed for many
realizations and the calculated response from each realization is then used to estimate statistical
properties of the system's response. While Monte Carlo simulations can be used on a wide variety of
problems, a major disadvantage is that a large number of simulations is required to achieve stable
results.

27
 Monte Carlo simulations require that each of the components in the system has a complete
statistical description. The description can be in the form of a variety of statistical distributions (i.e.
uniform, triangular, normal, log-normal, exponential, etc.), however the normal and log-normal
distributions typically are used because they can be easily described using a mean ( ) and standard
deviation ( ). For normally distributed variables, a random value (x) can be generated by:

(3.1)

where is the mean value, is the standard deviation, and is a random variable with zero mean and
unit standard deviation. Random values of  are generated and used to define the random values of x.

To generate multiple random variables that are independent, Equation 3.1 can be used for each
variable with different, random values of  generated for each variable. In the case of correlated random
variables, a more complicated procedure is required for the generation of values. The correlation
between variables is quantified through the correlation coefficient (). The correlation coefficient can
range from -1 to 1. Uncorrelated variables have =0 (Figure 3.1a). Positive correlation between variables
indicates that the two variables have a greater tendency to both differ from their respective mean
values in the same direction (Figure 3.1b). As  approaches 1.0, this correlation becomes stronger.
Negative correlation indicates that variables have a greater tendency to differ in the opposite direction
(Figure 3.1c).

Figure 3.1: Two variables with a correlation coefficient of: (a) 0.0, (b), 0.99, and (c) -0.7.

As discussed previously, independent random variables from a normal distribution are generated by
applying equation (3.1) independently to each random variable. By combining the multiple applications
of equation (3.1) into a system of equations, the generation of two independent variables is achieved by
multiplying a vector of random variables ( ⃗) by a matrix ([ ]) and adding a constant ( ⃗ ), defined as:
{ } [ ]{ } { } (3.2)

28
where and are random variables randomly selected from a standard normal distribution (
and ), and are the standard deviations of and , respectively, and and are the
mean values of and , respectively. Because the random variables and are independent
( ), the off-diagonal values in the matrix ([ ]) are zero.

Using the same framework, a linear system of equations is used to generate a pair of correlated
random variables. However, the off diagonal values in the matrix are no longer zero because of the
correlation between and . Instead, a pair of correlated random variables ( ⃗) is generated by (Kao,
1997):

{ } [ ]{ } { } (3.3)
√

Here, the first random variable ( ) is calculated based on the value of alone, while the second
random variable ( ) is a function of both and . Note that and still represent random and
independent variables generated from the standard normal distribution.

3.3 Statistical Models for Soil Properties

For the properties of the soil to be randomized and incorporated into Monte Carlo simulations, the
statistical distribution and properties of the soil need to be characterized. In this research, two separate
models are used. The first model, developed by Toro (1995), describes the statistical distribution and
correlation between layering and shear-wave velocity. The second model by Darendeli (2001) was
previously introduced in Section 2.1.3 and is used to describe the statistical distribution of the nonlinear
properties ( and D).

3.3.1 Layering and Velocity Model

In Strata, the randomization of the layering and the shear-wave velocity is done through the use of the
models proposed by Toro (1995). The Toro (1995) models provide a framework for generating layering
and then to vary the shear-wave velocity of these layers. The model for shear wave velocity variation
improves upon previous work by quantifying the correlation between the velocities in adjacent layers. In
previous models, one of two assumptions were made that simplified the problem: the velocities at all
depths are perfectly correlated and can be randomized by applying a constant random factor to all
velocities (McGuire et al. 1989; Toro et al. 1992), or the velocities within each of the layers are
independent of each other, and therefore can be randomized by applying an independent random
factor to each layer (Costantino el al. 1991). While these two assumptions simplify the problem, they
represent two extreme conditions. The Toro (1995) model makes neither of these assumptions; instead
the model incorporates correlation between layers.

29
3.3.1.1 Layering Model

The layering is modeled as a Poisson process, which is a stochastic process with events occurring at a
given rate ( ). For a homogeneous Poisson process this rate is constant, while for a non-homogeneous
Poisson process the rate varies. Generally, a Poisson process models the occurrence of events over time,
but for the layering problem the event is a layer interface and its rate is defined in terms of length (i.e.,
number of layer interfaces per meter).

In the Toro (1995) model, the layering thickness is modeled as a non-homogeneous Poisson
process where the rate changes with depth ( ( ), where d is depth from the ground surface). Before
considering the non-homogeneous Poisson process, first consider the simpler homogeneous Poisson
process with a constant rate. For a Poisson process with a constant occurrence rate ( ), the distance
between layer boundaries, also called the layer thickness (h), has an exponential distribution with rate .
The probability density function of an exponential distribution is defined as (Ang & Tang, 1975):

( )
( ) { (3.4)

The cumulative density function for the exponential distribution is given by:

( )
( ) { (3.5)

A random layer thickness with an exponential distribution is generated by solving equation (3.5) with
respect to thickness (h):

[ ( )]
( ) (3.6)

By randomly generating probabilities (F(h)) with a uniform distribution between 0 and 1 and computing
the associated thicknesses with equation (3.6), a layering profile was simulated for 10 layers with
(Figure 3.2). An exponential distribution with will be referred to as a unit exponential distribution.

30
 Figure 3.2: A 10 layer profile modeled by a homogeneous Poisson process with = 1.

Another way to think about generating exponential variables with a specific rate is to first
generate a series of random variables with a unit exponential distribution and then convert them to a
specific rate by dividing by the rate (see equation (3.6)). This process is shown graphically in Figure 3.3
and the associated layering is shown in Figure 3.4. In this example, the thicknesses (and depth) for =
1.0 (unit rate) are transformed to thicknesses (and depth) for = 0.2 (transformed rate). Here, each
thickness is increased by a factor of 5.0 (1/ ). A similar technique is used to transform random variables
generated with a unit exponential distribution into a non-homogenous Poisson process.

Figure 3.3: Transforming from a constant rate of = 1 to a constant rate of = 0.2.

31
 Figure 3.4: A 10 layer profile modeled by a homogeneous Poisson process with = 0.2.

For a non-homogeneous Poisson process with rate ( ) the cumulative rate ( ( )) is defined as
(Kao, 1997):

( ) ∫ ( ) (3.7)

( ) represents the expected number of layers up to a depth d. To understand the cumulative rate,
consider a homogeneous Poisson process with a constant rate (i.e., ( ) ). In this case, equation
(3.7) simplifies to ( ) . For (unit rate), ( ) , such that the expected number of
layers is simply equal to the depth. For (transformed rate), ( ) , such that the expected
number of layers is one-fifth the value of the unit rate because the layers are five times as thick. This
warping of the unit rate into a constant rate of 0.2 is represented by the straight line shown in Figure
3.3.

Transforming between the y-axis and x-axis in Figure 3.3 requires the inverse of the cumulative
rate function. For the homogenous case, ( ) , where u is the depth from an exponential
distribution with . For the non-homogenous case, the inverse cumulative rate function is used to
convert from a depth profile for (generated by a series of unit exponential random variables, )
to depth profile with a depth dependent rate. Before ( ) can be defined for the non-homogenous
process, ( ) and ( ) must be defined.

Toro (1995) proposed the following generic depth dependent rate model:

( ) ( ) (3.8)

32
The coefficients a, b, and c were estimated by Toro (1995) using the method of maximum likelihood
applied to the layering measured at 557 sites, mostly from California. The resulting values of a, b, and c
are 1.98, 10.86, and -0.89, respectively. The occurrence rate ( ( )) quickly decreases as the depth
increases (FIGURE 3.5a). This decrease in the occurrence rate increases the expected thickness of deeper
layers. The expected layer thickness (h) is equal to the inverse of the occurrence rate ( ( )) and
is shown in FIGURE 3.5b. The expected thickness ranges from 4.2 m at the surface to 59 m at a depth of
200 m.

Using equations (3.7) and (3.8), the cumulative rate for the Toro (1995) modeled is defined as:

( )
( ) ∫ ( ) [ ] (3.9)

The inverse cumulative rate function is then defined as:

( ) ( ) (3.10)

Using this equation a homogeneous Poisson process with (FIGURE 3.2) can be warped into a
non-homogeneous Poisson process as shown in FIGURE 3.6: TRANSFORMATION BETWEEN A
HOMOGENEOUS POISSON PROCESS WITH RATE 1 TO THE TORO NON-HOMOGENEOUS POISSON
PROCESS.. The resulting depth profile is shown in FIGURE 3.7.

Figure 3.5: Toro (1995) layering model. (a) The occurrence rate (λ) as a function of depth (d), and (b)
the expected layer thickness (h) as a function of depth.

33
Figure 3.6: Transformation between a homogeneous Poisson process with rate 1 to the Toro (1995)
non-homogeneous Poisson process.

Figure 3.7: A layering simulated with the non-homogeneous Poisson process defined by Toro (1995).

34
3.3.1.2 Velocity Model

After the layering of the profile has been established, the shear-wave velocity profile can be generated
by assigning velocities to each layer. In the Toro (1995) model, the shear-wave velocity at mid-depth of
the layer is described by a log-normal distribution. The standard normal variable (Z) of the ith layer is
calculated by:

[ ( )]
(3.11)

where is the shear-wave velocity in the ith layer, ( ) is the median shear-wave velocity at mid-
depth of the layer and is the standard deviation of the natural logarithm of the shear-wave
velocity. Equation (3.11) is then solved for the shear-wave velocity of the ith layer ( ):

{ [ ( )]} (3.12)

Equation (3.12) allows for the calculation of the velocity within a layer for a given median velocity at the
mid-depth of the layer, standard deviation, and standard normal variable. In the model proposed by
Toro (1995), values for median velocity versus depth ( ( )) and standard deviation ( ) are
provided based on site class. However, in the implementation of the Toro (1995) model in Strata the
median shear-wave velocity is defined by the user. Additionally, Strata includes the ability to truncate
the velocity probability density function by specifying minimum and maximum values. The standard
normal variable of the ith layer ( ) is correlated with the layer above it, and this inter-layer correlation is
also dependent on the site class. The standard normal variable ( ) of the shear-wave velocity in the top
layer (i=1) is independent of all other layers and is defined as:

(3.13)

where is an independent normal random variable with zero mean and a unit standard deviation. The
standard normal variables of the other layers in the profile are calculated by a recursive formula,
defined as:

√ (3.14)

where is the standard normal variable of the previous layer, is a new normal random variable
with zero mean and unit standard deviation, and is the inter-layer correlation.

Correlation is a measure of the strength and direction of a relationship between two random
variables. The inter-layer correlation between the shear-wave velocities proposed by Toro (1995) is a
function of both the depth of the layer (d) and the thickness of the layer (h):

( ) [ ( )] ( ) ( ) (3.15)

35
where is the thickness dependent correlation and is the depth dependent correlation. The
thickness dependent correlation is defined as:

( ) ( ) (3.16)

where is the initial correlation and is a model fitting parameter. As the thickness of the layer
increases, the thickness-dependent correlation decreases. The depth dependent correlation ( ) is
defined as a function of depth (d):

( )
( ) { [ ] (3.17)

where is the correlation coefficient at 200 m and is an initial depth parameter.

As the depth of the layer increases, the depth-dependent correlation increases. The final layer in
a site response model is assumed to be infinitely thick, therefore the correlation between the last soil
layer and the infinite half-space is only dependent on . Toro (1995) evaluated each of the parameters
in the correlation models ( , , , , ) for different generic site classes.

A site class is used to categorize a site based on the shear-wave velocity profile and/or local
geology. In the Toro (1995) model, the statistical properties of the soil profile (the median velocity,
standard deviation, and layer correlation) are provided for two different classifications schemes, the
Geomatrix and Vs30 classifications. The Geomatrix site classification classifies sites based on a general
description of the geotechnical subsurface conditions, distinguishing generally between rock, shallow
soil, deep soil, and soft soil (Table 3.1). In contrast, the Vs30 site classification is based on the time-
weighted average shear-wave velocity of the top 30 meters ( ) (Table 3.2), and requires site specific
measurements of shear-wave velocity.

Toro (1995) computed the statistical properties of the profiles for both the Geomatrix and
classifications using a maximum-likelihood procedure. The procedure used a total of 557 profiles,
with 541 profiles for the USGS classification and only 164 profiles for the Geomatrix classification.
The correlation parameters ( , , , , ) are presented in Table 3.3 and the median shear-wave
velocities in are presented in Table 3.4.

36
 Table 3.1: The categories of the geotechnical subsurface conditions (third letter) in the Geomatrix
site classification Toro (1995).

Designation Description
A Rock
Instrument is found on rock material ( > 600 m/s) or a very thin veneer (less
than 5 m) of soil overlying rock material.
B Shallow (Stiff) Soil
Instrument is founded in/on a soil profile up to 20 m thick overlying rock
material, typically a narrow canyon, near a valley edge, or on a hillside.
C Deep Narrow Soil
Instrument is found in/on a soil profile at least 20 m thick overlying rock
material in a narrow canyon or valley no more than several kilometers wide.
D Deep Broad Soil
Instrument is found in/on a soil profile at least 20 m thick overlaying rock
material in a broad canyon or valley.
E Soft Deep Soil
Instrument is found in/on a deep soil profile that exhibits low average shear-
wave velocity ( < 150 m/s).

Table 3.2: Site categories based on (Toro (1995)).

Average Shear-wave Velocity

Vs30 greater than 750 m/s
Vs30 =360 to 750 m/s
Vs30 =180 to 360 m/s
Vs30 less than 180 m/s

37
 Table 3.3: Coefficients for the Toro (1995) model.

GeoMatrix (m/s)
Property A &B C&D >750
360 to 180 to < 180
750 360
0.46 0.38 0.36 0.27 0.31 0.37
0.96 0.99 0.95 0.97 0.99 0.00
0.96 1.00 0.42 1.00 0.98 0.50
13.1 8.0 3.4 3.8 3.9 5.0
0.0 0.0 0.0 0.0 0.0 0.0
0.095 0.160 0.063 0.293 0.344 0.744

Profiles 45 109 35 169 226 27

38
 Table 3.4: Median shear-wave velocity (m/s) based on the generic site classification.

GeoMatrix (m/s)
Depth (m) A &B C&D >750 360 to 180 to < 180
750 360
0 192 144 314 159 145 176
1 209 159 346 200 163 165
2 230 178 384 241 179 154
3 253 193 430 275 191 142
4 278 204 485 308 200 129
5 303 211 550 337 208 117
6 329 217 624 361 215 109
7.2 357 228 703 382 226 106
8.64 395 240 789 404 237 109
10.37 443 253 880 433 250 117
12.44 502 270 973 467 269 130
14.93 575 291 1070 501 291 148
17.92 657 319 1160 535 314 170
21.5 748 357 1260 567 336 192
25.8 825 402 1330 605 372 210
30.96 886 444 1380 654 391 229
37.15 942 474 1420 687 401 246
44.58 998 495 1460 711 408 266
53.2 1060 516 1500 732 413 289
64.2 541 749 433 318
77.04 566 772 459 353
92.44 593 802 486 392
110.93 847 513 435
133.12 900 550
159.74 604
191.69 676
230.03 756

Ten generated shear-wave velocity profiles were created for a deep, stiff alluvium site using the
two previously discussed methods. In the first method, a generic site profile is generated by using the
layering model coefficients and median shear-wave velocity for a = 180 to 360 m/s site class, shown
in Figure 3.8(a). This approach essentially models the site as a generic stiff soil site. The second method
uses the layer correlation for the = 180 to 360 m/s site class, but the layering and the median shear-
wave velocity profile are defined from field measurements, shown in Figure 3.8(b). The site specific
layering tends to be much thicker than the generic layering as a result of the field measurements
indicating thick layers with the same shear wave velocity. In general both of the methods show an
increase in the shear-wave velocity with depth. However, the site-specific shear-wave velocity values are
significantly larger than the generic shear-wave velocity values. At the surface, the generic site has a
median shear-wave velocity of 150 m/s compared to the site specific shear-wave velocity of 200 m/s. At
a depth of 90 m, the difference is even greater, with the generic site having a median shear-wave
39
velocity of 470 m/s compared to the site specific median shear-wave velocity of 690 m/s. The difference
in shear-wave velocity is a result of the difference between the site specific information and the generic
shear-wave velocity profile.

Figure 3.8: Ten generated shear-wave velocity (vs) profiles for a USGS C site class. (a) Using generic
layering and median vs, (b) using user defined layering and median vs.

3.3.2 Depth to Bedrock Model

The depth to bedrock can be modeled using either a uniform, normal, or log-normally distributed
random variable. When using the normal or log-normal distribution, the median depth is based by the
soil profile. The variation in the depth to bedrock is accommodated by varying the height of the soil
layers. If the depth to bedrock is increased, then the thickness of the deepest soil layer is increased.
Conversely, if the depth to bedrock is decreased then the thickness of this deepest soil layer is
decreased. If the depth to bedrock is less than the depth to the top of a soil layer, then the soil layer is
removed from the profile.
40
3.3.3 Non-Linear Soil Properties Model

The Darendeli (2001) empirical model for nonlinear soil properties ( and D) was previously
discussed in Section 2.1.3. The Darendeli (2001) empirical model assumes the variation of the properties
follows a normal distribution. The standard deviation of and D varies with the magnitude of the
property and is calculated with equations (2.20) and (2.21), respectively. Because the variation of the
properties is modeled with a normal distribution that is continuous from to , the generated values
of or D may fall below zero. The most likely location for the negative values occurs when the
mean value is small, which occurs at large strains for and at low strains for D. Negative values
for either or D are not physically possible, therefore the normal distributions need to be
truncated. To correct for this problem, minimum values for and D are specified. The default
values in Strata are = 0.05 and D = 0.1%. Strata also includes the ability to specify maximum
values of and D.

and D curves are not independent of each other. Consider a soil that behaves more
linearly, that is to say the is higher than the mean . During a loading cycle, the area
inside the hysteresis loop would be smaller which is indicative of less damping within the system.
Therefore, as the linearity of the system increases, the damping decreases. To capture this effect, the
soil properties are assumed to have a negative correlation with the default value set at -0.5 (i.e. ρ=-0.5).

To generate correlated G/Gmax and D curves from baseline (mean) curves, the following
expressions are used for each shear strain value in the curves:

⁄ ( ) [ ⁄ ( )] (3.18)

( ) [ ( )] √ (3.19)

where and are uncorrelated random variables with zero mean and unit standard deviation,
[ ⁄ ( )] and [ ( )] are the baseline values evaluated at strain level , and are
the standard deviations computed from (2.20) and (2.21) at the baseline values of [ ⁄ ( )]
and [ ( )] , respectively, and is the correlation coefficient between G/Gmax and D. Equations
(3.18) and (3.19) must be applied at different strain levels, but the same values of and are used at
each strain level (i.e., perfect correlation between strain levels).

Using a correlation coefficient of -0.5, the nonlinear properties of sand (PI=0, OCR=0) at a
confining pressure of 1 atm were generated 10 times, shown in Figure 3.9. Three of the realizations
result in large shear modulus reduction curve relative to the mean. Because of the negative correlation,
the relative high shear modulus reduction corresponds to a relative low damping ratio.

41
Figure 3.9: Generated nonlinear properties assuming perfect negative correlation.

42
4 Using Strata

Strata is introduced through three examples that demonstrate the organization and most of the features
found in Strata. Before these three examples are discussed, some particular differences between Strata
and other site response programs are introduced.

With the exception of input motions, all of the input to Strata is entered via the keyboard, or
through copying and pasting from spreadsheets. The input file is not saved in the typical text format,
instead a binary format is used that is only readable by Strata. Furthermore, when the calculation is
complete no output text files are produced. Instead, the output can be directly viewed with Strata and
saved, once again to a binary format. There is the option for the data to be exported to text files that
can then be opened with a variety of applications, including Excel.

The screenshots and the description of all of the fields are based on Strata-rev-399.

4.1 Strata Particulars

4.1.1 Auto-Discretization of Layers

One of the biggest differences between Strata and other site response analysis programs is the fact that
the sublayers used in the calculation portion of the analysis are not defined by the user. Instead, the
user defines a velocity layer that is then subdivided into sublayers by Strata. This fundamental difference
exists because Strata allows for the layering and shear-wave velocity to vary across Monte Carlo
simulations (see Section 3.3.1.1) and therefore the required thickness of the sublayers changes.

The maximum thickness ( ) of the sublayers of ith velocity layer is taken as a fraction of the
minimum wavelength to be captured by the analysis:

(4.1)

where is the wavelength fraction which typically varies between and (anything greater than is
not recommended), is the maximum frequency of engineering interest which is typically around 20
Hz, and is the shear-wave velocity of the ith layer. The actual thickness of the sublayers is less than
the maximum thickness such that the velocity layer height divided by the sub-layer thickness is a whole

43
number. These parameters are defined in the Layer Discretization group box on the General Settings
page (Figure 4.1: Layer Discretization group box.). To prevent the layers from being auto-discretized,
click the Disable auto-discretization check box in the Layer Discretization group box (Figure 4.1: Layer
Discretization group box.). By clicking this box, the program will use the layers as defined on the Soil
Profile tab, which is the same approach used in most site response programs.

Figure 4.1: Layer Discretization group box.

4.1.2 Location of Input Motions

In other site response programs, the location of the input motion or the location of the requested
output (e.g., acceleration-time history) is generally referenced by a sublayer index. However, because
the sublayers are computed in Strata (and may change for each Monte Carlo realization), the location is
defined in terms of either the depth within the soil profile or at the top of the bedrock. When the
location is specified as Bedrock then the actual depth of the location may change if the depth of the
bedrock changes between Monte Carlo realizations. The location is specified by double-clicking the
appropriate cell, which will open a drop down list (Figure 4.2: Location selection: (a) top of bedrock, (b)
switching to a fixed depth, and (c) fixed depth specified as 15.), where the user can specify the depth as
Bedrock (Figure 4.2: Location selection: (a) top of bedrock, (b) switching to a fixed depth, and (c) fixed
depth specified as 15.a) or a fixed depth (Figure 4.2: Location selection: (a) top of bedrock, (b) switching
to a fixed depth, and (c) fixed depth specified as 15.b and Figure 4.2: Location selection: (a) top of
bedrock, (b) switching to a fixed depth, and (c) fixed depth specified as 15.c).

Figure 4.2: Location selection: (a) top of bedrock, (b) switching to a fixed depth, and (c) fixed depth
specified as 15.

44
4.1.3 Interaction with Tables

Table cells are selected with one click. After a cell is selected, the cell can be edited by typing. A cell's
edit mode can be directly entered by double clicking on the cell. In some cases, double clicking on a cell
will produce a widget to aid specifying input to the cell.

All tables used in Strata are dynamic, that is the number of rows can be changed. Rows are
added to the bottom of the list with the Add button. The Insert and Remove buttons are disabled until a
complete row has been selected, which is most easily achieved by clicking on the number next to the
row of interest. Multiple continuous rows can be selected by pressing the shift key while selecting the
rows. After rows have been selected: Add will add the same number of rows to the end of the table,
Insert will insert the same number of rows above the currently selected rows, and Remove will remove
the selected rows. All rows in the table can be selected by click on the button in the upper left portion of
the table as shown in Figure 4.3: By clicking on the button circled in red all rows in the table are
selected.. Some tables have cells that cannot be edited and have a light gray background (Figure 4.3: By
clicking on the button circled in red all rows in the table are selected.).

Figure 4.3: By clicking on the button circled in red all rows in the table are selected.

Data can be copied from spreadsheets and pasted into tables by first clicking on the table and:

1. Pressing Ctrl+v, or
2. Selecting Paste from the Edit menu in Strata, or
3. Right clicking on the table and selecting paste.

The table will automatically increase the numbers of rows to accommodate the size of the pasted data.

45
4.1.4 Non-Linear Curves

The nonlinear shear-modulus reduction and damping curves can be specified through three different
methods in Strata: (1) fixed models that are present by default and cannot be removed, (2) user defined
curves that can be used across projects, and (3) temporary models that only exist for the project.

Fixed Non-Linear Models

The following shear-modulus reduction and damping models are included by default:

 Darendeli (2001)
 EPRI (1993)
o Plasticity Based -- 10, 30, 50, and 70.
o Depth Based – 0-20, 20-50, 50-120, 120-250, 250-500, and 500-1000 ft.
 GEI (1993) – 0-50, 50-100, 100-250, 250-500, and >500 ft.
 GeoMatrix (Coppersmith, 1991) – 0-50, 50-150, and >150 ft.
 Idriss (1990) – clay and sand.
 Imperial Valley Soils (Turner and Stokoe, 1983) – 0-300 and >300 ft.
 Iwasaki (1976) – 0.25 and 1.0 atm.
 Peninsular Range (Silva et al., 1997) – 0-50 and 50-500 ft.
 Seed and Idriss (1970)
o For G/Gmax: sand mean and upper.
o For D: sand lower and mean.
 Vucetic and Dobry (1991) – Plasticity indices of 0, 15, 30, 50, 100, and 200.

User Defined Models

Non-linear curve models can be defined for use across multiple projects by adding models to the library.
The nonlinear property manager is opened by selecting Add/Remove Non-Linear Property Curves from
the Tools menu. Using the dialog (Figure 4.4: The nonlinear curve manager.), a new model can be
defined by following these steps:

1. Click the Add button to add a new curve to the normalized shear-modulus reduction or damping
models list (shown at the very bottom of the list).
2. Rename the model from “Custom” to something meaningful.
3. Use the Add button the right to add lines for specification of the data points for the curve.

Models defined in this manner will be added to the nonLinearCurves.strd file found in the Strata
installation folder.

46
 1 3

Figure 4.4: The nonlinear curve manager.

Temporary Models

If you want to define a curve without adding it to the library of models, simply select Custom from the
drop-down list in the Soil Types Table. Changing to the Custom model does not clear the previous
model’s data which allows for a model to be modified.

47
4.1.5 Input Motions Dialog

The Input Motions dialog is used to load a recorded motion into Strata and appears when the Add
button is clicked in the Input Motions dialog within the Motion(s) page. The dialog shown in Figure 4.5 is
what appears when Time Series approach is used. It allows the user to load a variety of time series of
ground motions in most formats using the following steps:

1. Click on the File button within the Input tab and select the acceleration time series text file. If
the file is from the NGA database, then the remainder of the form will be automatically
completed as shown in Figure 4.5. Regardless of the file format, the file is read and loaded into
the preview area.
2. The remaining fields need to be filled to reflect the information in the file. Information is
required for all fields except for the Description field. Fields can be completed by either typing
values in, or selecting from the file preview and dragging the selected text into the field. The
Start line and Stop line control which lines in the file contain the data. A zero value for the Stop
line will result in the data being read until the end of the file. The color of the file preview can
be refreshed by clicking on the Apply button.
The colors of the file preview have the following meanings:
 Green – headerlines: text found prior to the acceleration-time series data (these lines
are ignored).
 Blue – data lines: acceleration-time series data.
 Red – extraneous lines: text after the time series data (these lines are ignored).

An example of the colored data is shown in Figure 4.6.

3. The scale factor can be selected at this time or after the motion has been loaded. If the input
acceleration-time history is not in units of g (g=acceleration of gravity), then the scale factor
should be used to make this unit conversion. As the scale factor varies, the peak ground
acceleration and velocity change accordingly.
4. After the dialog has been completed, the acceleration time series, response spectrum and
Fourier amplitude spectrum can be viewed in the Plot tab.
5. Click OK to finish loading the file.

48
 Figure 4.5: An initial view of the Input Motions Dialog.

Figure 4.6: An example a completed Input Motions Dialog.

49
4.1.6 Interacting with Results

Prior to the calculation of the site response, the Results tab is disabled. After the site response
calculation has been completed, the Results tab will be enabled and selected. The remaining tabs are
placed in a read-only (or “locked”) mode where the fields can be reviewed but not edited. Therefore,
the input parameters used to generate the output are strictly linked for later reference by the user.

Strata can be unlocked by clicking on the Delete results icon in the upper left portion of the
screen, by selecting Delete results from the Tools menu, or by pressing F2. By unlocking Strata, all of the
results are deleted. If you wish, you may save the results and make changes to the input and re-run the
analysis. First, save the results using either the Save or Save As… command from the File Menu. Next
create a copy of the file by selecting Save As… from the File Menu and using a new name appropriate for
the new input parameters. Finally, unlock the newly created file.

Strata does not output any data files automatically, instead the results are saved to the project
file in a binary format. To work with the data outside of Strata, the data can be exported by selecting
Export... from the File menu. The exported data file is in a comma-separated values (.csv) format that
can be easily opened with Excel or another spreadsheet program. All data (even disabled results) are
included in the files. Data also can be copied from the table format and pasted into a spreadsheet.

Results can be viewed in both plot and table format. A screenshot of the Results tab is shown in
Figure 4.7. Each result represents a site realization subjected to an input motion. The output of interest
is selected from a drop-down box containing a list of all requested output. Individual values from this list
are selected by either clicking on the corresponding row in the Data Selection table, or by clicking on the
result in the plot. In both cases, the result is colored green if the result is enabled, or red if the result is
disabled. A result can be enabled or disabled by clicking on the checkbox next to the site realization that
produced the result. After the status of a result has been changed, the Recompute Statistics button will
become enabled indicating that the median and standard deviation (shown on the plot in solid and dash
blue lines, respectively) need to be updated. Click this button to recompute the statistics.

At the bottom of the table there are two buttons that allow all motions or sites related to the
currently selected result to be enabled or disabled. In the example (Figure 4.7), the motion
CHICHI06\TCU076-E.AT2 has been disabled. Whenever the status of a result is changed (e.g. from
disabled to enabled) the Recompute Statistics button will become enabled allowing the user to update
the median and standard deviation. The current plot can be printed by selecting Print… or Print to PDF…
from the File menu. The current plot can also be copied by right clicking on the plot and selecting Copy.

50
Figure 4.7: Using the Output view to examine the results of a calculation.

51
4.2 Glossary of Fields

The following section contains screenshots of the program and describes each of the fields. Assume that
information is required for each field, unless stated otherwise.

4.2.1 General Settings Page

Project Group Box

Project group box is used to provide some basic information about the newly-created project (Figure
4.8).

Figure 4.8: Screenshot of the Project group box.

 Title – name of the project (optional). The title is included in the first row of the exported data
and therefore can be extremely useful in identifying the output.
 Notes – a place to store information about the input (optional); only used to preserve your
sanity.
 Filename prefix – placed at the beginning of the name of all exported files (optional). The prefix
offers a means to distinguish a group of analyses.
 Units – defines gravity in the analysis. Changing the units does not perform any conversion; it is
recommended that the units be defined before any site information is provided.

52
  Save motion data checkbox – allows time series to be saved within the input file (optional). If
checked then the time series information is stored within the input file allowing one file to be
transferred between computers. However, larger input files will be generated.

Type of Analysis Group Box

Figure 4.9: Screenshot of the Type of Analysis group box.

 Method – either Linear Elastic or Equivalent Linear (EQL).

 Approach – either Time Series for a time series analysis, or Random Vibration Theory for RVT
analysis.
 Vary the properties checkbox – Controls if site properties will be varied in the calculation.

Site Property Variation Group Box

This group box is only enabled if the Vary the Properties check box in the Type of Analysis group box is
enabled. An example of what the group box looks like is shown in Figure 4.10.

Figure 4.10: Screenshot of the Site Property Variation group box.

 Number of realizations – number of sites to be generated. A site consists of nonlinear curves,

shear-wave velocity profile, layering thickness, and depth to bedrock – the variation of each of

53
 these properties is controlled elsewhere. Each of the input motions is propagated through a
generated site.
 Vary the nonlinear properties – controls if the shear modulus reduction and damping curves, as
well as the damping of the bedrock are varied.
 Vary the site profile – controls if the shear-wave velocity, layer thickness, and/or depth to
bedrock are varied.

Calculation Parameters Group Box

Figure 4.11: Screenshot of the Calculation Parameters group box.

 Error tolerance – the maximum error in the nonlinear properties between iterations.
 Maximum number of iterations – if the error tolerance is not achieved in this number of
iterations, the calculation ends.
 Effective strain ratio – the ratio between the effective shear strain (used for the strain
compatible nonlinear properties) and the maximum shear strain computed in the layer.

Layer Discretization Group Box

Figure 4.12: Screenshot of the Layer Discretization group box.

 Maximum frequency – maximum frequency of engineering interest

 Wavelength fraction – thickness of the layer relative to the wavelength at the maximum
frequency of engineering interest.
 Disable auto-discretization – by checking this check box, the auto-discretization is disabled, and
the program will use the layers as defined on the Soil Profile tab.
54
4.2.2 Soil Types Page

Soil Types Group Box

Figure 4.13: Screenshot of the Soil Types group box.

 Name – used for identifying soil layer in the program and output.
 Unit weight – total unit weight of the soil.
 G/G_max Model – model that describes the variation of normalized shear modulus with shear
strain.
 Damping Model – model that describes the variation of the damping ratio with shear strain.
 Notes – for your sanity only (optional).
 Varied – check box to identify if the nonlinear curves of the soil type will be varied. Only visible if
variation of the nonlinear properties is enabled.

Bedrock Layer Group Box

Figure 4.14: Screenshot of the Bedrock Layer group box.

 Unit weight – total unit weight of the infinite half space.

 Initial damping – damping of the infinite half space.
 Vary the damping of the bedrock - check box to identify if the damping in the bedrock is to be
varied

55
Water Table Depth Group Box

Figure 4.15: Screenshot of the Water Table Depth group box.

 Depth – depth of the water table. Only used to compute vertical effective stress for use in the
computation of cyclic stress ratio in the Results tab.

Nonlinear Curve Variation Parameters Group Box

This box described the standard deviation model for the nonlinear property curves for use in the
variation of these parameters (Figure 4.16). The standard deviation functions are described separately
for G/G_max and damping. The default functions are from Darendeli (2001).

Figure 4.16: Screenshot of the Nonlinear Curve Variation Parameters group box.

 Standard Deviation Model – model to describe the variability of the nonlinear parameters.
 Normalized shear modulus (G/G_max)
o Function – formula to describe the standard deviation of G/G_max. Enabled if standard
deviation model is set to custom.
o Max – maximum value of the normalized shear modulus.
o Min -- minimum value of the normalized shear modulus.
 Damping
o Function – formula for the standard deviation of the damping. Enabled if standard
deviation model is set to custom.
o Max – maximum value of the damping ratio.
o Min -- minimum value of the damping ratio.
 G/G_max, Damping Correlation Coefficient () – correlation coefficient between the varied
nonlinear curves.

56
Darendeli and Stokoe Model Parameters Group Box

This group box is only enabled when a soil type that uses the Darendeli and Stokoe (2001) model for
either of the nonlinear soil properties is selected. An example of what the group box looks like is shown
in Figure 4.17.

Figure 4.17: Screenshot of the Darendeli and Stokoe Model Parameters group box.

 Mean effective stress – mean effective stress of the soil in units of atmospheres.
 Plasticity index – plasticity index of the soil.
 Over-consolidation ratio – over-consolidation ratio of the soil.
 Excitation frequency – frequency of excitation.
 Number of cycles – number of loading cycles.

Nonlinear Property Group Box

For fixed models, this table will provide a read only view of the data points used in the models. The
values of the models can be edited by switching the model to custom. An example of what the group
box looks like is shown in Figure 4.18. The second column in the table changes between G/G_max and
damping depending on what type of model is selected in the Soil Type table.

57
 Figure 4.18: Screenshot of the Nonlinear Property group box.

 Strain (%) – shear strain in percent.

 G/G_max – normalized shear modulus.
 Damping (%) – damping ratio in percent.

4.2.3 Soil Profile Page

Site Profile Group Box

Figure 4.19: Screenshot of the Site Profile group box.

 Depth – depth of the top of the velocity layer – computed by Strata.

 Thickness – thickness of velocity layer.

58
  Soil Type – Soil Type associated with the velocity layer. This associates the velocity layer to a
specific unit weight and nonlinear properties for the analysis.
 Vs –shear-wave velocity of the layer.
 Minimum – minimum shear-wave velocity. Minimum value of shear wave velocity allowed
during Monte Carlo simulation
 Maximum – maximum shear-wave velocity. Maximum value of shear wave velocity allowed
during Monte Carlo simulation
 Varied – check box to identify if the shear-wave velocity of the layer is to be varied.

Variation of the Site Profile Group Box

The Variation of the Site Profile group is hidden unless the Site Property Variation checkbox is checked
on the General Settings page. For each of the functionalities (shear-wave velocity variation, layer
thickness variation, and depth to bedrock variation), there is a check box that controls if the
functionality should be activated (Figure 4.20). A group box for model parameters appears for each
functionality when the functionality is enabled.

Figure 4.20: Screenshot of the Site Profile group box.

59
Velocity Variation Parameters Group Box

Figure 4.21: Screenshot of the Velocity Variation Parameters group box.

 Layer specific standard deviation checkbox – Checkbox to identify if the standard deviation for
each velocity layer will be defined in the Site Profile table.
 Distribution – statistical distribution of the shear-wave velocity.
 Standard deviation – the site class for a generic standard deviation. If set to custom, a specific
value of standard deviation for the site can be defined.
 Correlation model – the site class for the correlation model. If set to custom, specific correlation
parameters must be defined in the Correlation Parameters group box.
 Correlation Parameters -- for information on these parameters see Section 3.3.1.2.
o Correl. coeff. at surface – correlation coefficient at the surface (ρ0).
o Correl. Coeff. at 200 m – correlation coefficient at a depth of 200 m (ρ200).
o Change in correl. with depth – change in correlation coefficient with depth (Δ).
o Depth intercept – depth intercept (d0).
o Exponent – exponent in the depth correlation term (b).

60
Layer Thickness Variation Parameters Group Box

Figure 4.22: Screenshot of the Layer Thickness Variation Parameters group box.

For more information on the parameters in this group box see Section 3.3.1.1.

 Coefficients – the parameters for the model used to vary layer thickness. If set to custom then
the model parameters a, b, and c are defined by the user.

Bedrock Depth Variation Parameters Group Box

61
 Figure 4.23: Screenshot of the Bedrock Depth Variation Parameters group box.

 Distribution – the statistical distribution of the bedrock depth, either uniform, normal, or log
normal.
 Standard deviation – standard deviation of the log normal or normal distribution. For a log
normal distribution this value is defined in log space.
 Minimum depth to bedrock – check box to identify if the distribution is to be truncated at a
minimum depth, and associated minimum depth.
 Maximum depth to bedrock – check box to identify if the distribution is to be truncated at a
maximum depth, and associated maximum depth.

4.2.4 Motion(s) Page

The Motion(s) Page is used to define the input motion(s) and where the input motion(s) is (are) input
into the soil profile. Depending on the type of analysis, the motion(s) are defined in different ways.

Motion Input Location Group Box

Figure 4.24: Screenshot of the Motion Input Location group box.

62
  Specify the location to input the motion(s) – The input location specifies where the input motion
should be specified. The adjacent box can be used to specify a specific depth or the bedrock. If
the input location is defined as bedrock then location always will be at the top of the half-space.

Input Motion(s) – for Time Series Approach

When the time series approach is selected, the Input Motions dialog looks like Figure 4.25.

Figure 4.25: Screenshot of the Input Motions group box (for Time Series Approach).

A recorded motion is added to the table by clicking on the Add button, and filling out the information in
the dialog (see Section 4.1.5 for more information). The following information is presented in the Input
Motion(s) dialog:

 Name – contains the directory and filename of the motion.

 Description – description of the time series (optional).
 Type – type of boundary conditions of the motion; either Outcrop or Within.
 PGA – the peak ground acceleration (PGA) of the scaled record in units of gravity.
 PGV – the peak ground velocity (PGV) of the scaled record in units of cm/s.
 Scale Factor – scale factor to convert the motion into units of gravity (if required, see Section
4.1.5) at the desired amplitude. Changing the scale factor changes the listed PGA value and PGV
value.

An entire suite of motions in the NGA (Next Generation Attenuation, http://peer.berkeley.edu/nga)

format can be added by clicking on the Load Suite push button which opens a file selection dialog. The
suite file is a simple text file with the extension ‘.csv’. Each line of the file contains the path name to the
motion and the scale factor separated by a comma. The path name can be either absolute or relative to
the path of the suite file. If examples were installed during the installation of Strata then an example
suite file named ‘suite-10-1.csv’ can be found in the examples directory.

Input Motion(s) – for Random Vibration Theory Approach

When the random vibration theory approach is selected there are a variety of methods that can be used
to specify the input motion. By clicking on the Add button a drop-down list can be used to choose from

63
these methods (Figure 4.26). There are three types of methods that you can use for defining RVT motion:
User Defined FAS, Response Spectrum Compatible, and Seismological Source Theory.

Figure 4.26: Screenshot of the RVT Motion Type drop-down list.

User Defined FAS Dialog

The User Defined FAS dialog is used to specify an acceleration Fourier amplitude spectrum (FAS) directly
(Figure 4.27). The following information must be provided:

 Name – specifies the name of the RVT motion.

 Description – provides further description of the motion.
 Type – type of boundary conditions of the motion; either Outcrop or Within,
 Duration – duration of shaking for use in RVT calculation.

Figure 4.27: Screenshot of the User Defined FAS dialog.

64
The simplest method of entering data into this table is pasting from a spread sheet. The columns
represent the following:

 Frequency – frequency in Hertz.

 Amplitude – Fourier amplitude at each frequency in units of gravity-seconds.

After entering FAS data and clicking on the Apply button, the response spectrum and Fourier Amplitude
Spectrum are plotted in the corresponding tab (RS Plot, FAS Plot). The calculated Response Spectrum
data also can be viewed in table format in the RS Data tab.

Response Spectrum Compatible Dialog

The Response Spectrum Compatible dialog (Figure 4.28) is used to specify an acceleration response
spectrum that is used to compute a Fourier amplitude spectrum through inversion (see Section 2.2.2.4).
The following information must be provided:

 Name – specifies the name of the RVT motion.

 Description – provides further description of the motion.
 Type – type of boundary conditions of the motion; either Outcrop or Within,
 Duration – duration of shaking.
 Damping of Target – damping ratio for the target response spectrum.

Figure 4.28: Screenshot of the Response Spectrum Compatible dialog.

65
The simplest method of entering data into this table is pasting from a spread sheet. The columns
represent the following:

 Period (s) – period of the oscillator in units of seconds.

 Spec. Accel. (g) – spectral acceleration of the oscillator in units of gravity.

After entering the response spectrum data and clicking on the Apply button, the response spectrum and
Fourier Amplitude Spectrum are plotted in the corresponding tabs (RS Plot, FAS plot). The response
spectrum and Fourier Amplitude Spectrum data points also can be viewed in table format on the RS
Data and FAS Data tabs. Note that the minimum and maximum values of the frequency, the number of
points, and whether the frequency interpolation is in log space or linear space can be specified by
clicking on the Frequency Parameters… button on the dialog (FIGURE 4.29).

Figure 4.29: Screenshot of the Frequency Parameters… dialog.

Seismological Source Theory Dialog

There is a Point Source Parameters tab and a Crustal Amplification tab in this dialog (Figure 4.30 and
Figure 4.31).

 Point Source Parameters tab (Figure 4.30)

o Name – specifies the name of the RVT motion.
o Description – provides further description of the motion.
o Moment Magnitude – moment magnitude of the event.
o Epicentral distance – distance to the epicenter in km.
o Depth – depth to the rupture in km.
o Parameter region – crustal region associated with the source and crustal properties.
Choose from Western US (Western NA), Central and Eastern US (Eastern NA), and
Custom. The Western US and Central and Eastern US values are from Campbell (2003).
o The parameters below are automatically specified if a Parameter region is selected.
They must be specified by the User if Custom is selected as the parameter region.

66
 o Stress drop – stress drop of the rupture (required for Custom location).
o Geometric attenuation coeff. – geometric attenuation coefficient. For example, if the
geometric attenuation relationship is 1/R and R=20, then this value would be 0.05.
o Path duration coefficient – factor used to define the portion of the duration that is
dependent on the path distance (Tp). See section 2.2.2.2 for details.
o Path attenuation – frequency dependent path attenuation Q defined by a coefficient (a)
and a power (b) using Q = a fb
o Shear velocity – shear-wave velocity at the rupture in km/s.
o Density – density of the rock at the rupture in g/cm3.
o Site attenuation – site attenuation o.
o Duration – duration of the event (D5-75).

Figure 4.30: Screenshot of the Seismological Source Theory dialog – Point Source Parameters tab.

67
 Crustal Amplification tab (Figure 4.31)
o Crustal Model – can choose from Western NA, Eastern NA, Custom and Calculated
(Figure 4.31 and 4.32). If Calculated is selected, user can define the thickness, shear-
wave velocity, and density of the crustal layers, as shown in Figure 4.32.
o Amplification – it allows for the crustal amplification to be computed for a site specific
crustal velocity profile using the quarter-wavelength method (Boore 2003).

Figure 4.31: Screenshot of the Seismological Source Theory dialog – Crustal amplification tab.

68
 Figure 4.32: Screenshot of the Crustal Model dialog if Calculated is selected.

4.2.5 Output Specification Page

The Output Specification page allows the user to specify which output they want provided by Strata.
The types of output are grouped in terms of Profiles, Time Series, Response and Fourier Spectra, Ratios,
and Soil Types.

Profiles Tab

The profiles tab allows you to select as output the variation of various parameters with depth (Figure
4.33). By checking a parameter, its variation with depth is shown in the results tab after computation.
The parameters that can be selected are listed below.

69
 Figure 4.33: Screenshot of the Profiles tab.

 Arias Intensity Profile – the Arias Intensity of the acceleration time history at a given depth, in
units of gravity.
 Damping Ratio – the damping ratio used in the final equivalent-linear iteration, in units of
percent.
 Dissipated Energy Profile – the dissipated energy per unit volume of soil ( at a given depth is
equal to the cumulative area bounded by the shear stress-shear strain hysteretic loops (Green
2001, Green and Teri 2005). The units of dissipated energy are stress. For this calculation, the
hysteresis loops in Strata are simulated by using the complex shear modulus to compute a
shear-stress time history. The dissipated energy is computed from the shear stress and shear
strain time histories using the trapezoid rule:

∑( )( )

 Final Shear-Wave Velocity Profile – the shear-wave velocity profile used in the final equivalent-
linear iteration.
 Initial Shear-Wave Velocity Profile – the initial small-strain shear-wave velocity profile.
 Peak Ground Acceleration Profile – the maximum acceleration at each depth, in units of gravity.
70
  Peak Ground Displacement Profile – the maximum displacement at each depth in units of cm/s
with depth.
 Maximum Error Profile – the maximum error at each depth in the equivalent-linear shear
modulus and damping ratio for the last iteration, in units of percent.
 Maximum Shear-strain Profile – the maximum shear-strain at each depth, in units of percent.
 Maximum Shear-stress Profile – the maximum shear-stress at each depth.
 Peak Ground Velocity Profile – the maximum velocity at each depth.
 Shear-Modulus Profile – the shear modulus at each depth used in the final equivalent-linear
iteration.
 Stress Ratio Profile – the ratio of the maximum shear stress to the vertical effective stress at
each depth.
 Stress Reduction Coefficient (r_d) Profile – the ratio of the shear stress at a given depth to the
equivalent shear stress for rigid conditions, defined as:

 Vertical Total Stress Profile – the vertical total stress ( ) at each depth.
 Vertical Effective Stress Profile – the vertical effective stress ( ) at each depth.

Time Series Tab

The time series of acceleration, displacement, velocity, shear-strain and shear-stress at any depth can be
selected for output. Click the Add button and select the appropriate time series from the drop-down list
(Figure 4.34).

Figure 4.34: Screenshot of the Time Series drop-down list.

After selecting the type of time series, it will be listed in a table on the Time Series tab (Figure 4.35). For
each time series, the following must be specified:

 Location – depth of the requested time series, either Bedrock or a specific depth.
 Type – boundary conditions of the response (Within or Outcrop).
 Baseline correct – if this box is checked then the acceleration-, velocity-, and displacement-time
series are baseline corrected using a polynomial.

71
 Figure 4.35: Screenshot of the Time Series tab.

Response and Fourier Spectra Tab

This tab is used to select the locations for which Acceleration Response Spectra and Fourier Amplitude
Spectra are requested. Click the Add button and select the appropriate type of spectra from the drop-
down list (Figure 4.36).

Figure 4.36: Screenshot of the Response and Fourier Spectra drop-down list.

After selecting the type of spectrum, it will be listed in a table on the tab (Figure 4.37). For each
spectrum, the following must be specified:

 Location – depth of interest, either Bedrock or a specific depth.

 Type – boundary conditions of the response (Within or Outcrop).

Figure 4.37: Screenshot of the Response and Fourier Spectra tab.

72
Ratios Tab

This tab is used to select the locations for which the ratios of Acceleration Response Spectra or Fourier
Amplitude Spectra are requested. Click the Add button and select the appropriate type of ratio from the
drop-down list (Figure 4.38). The following ratios may be requested:

 Acceleration Transfer Function (FAS_1/FAS_2) – the ratio of the Fourier Amplitude Spectrum of
the acceleration (i.e. transfer function) between locations 1 and 2.
 Spectral Ratio (Sa_1/Sa_2) – the ratio of the response spectra between locations 1 and 2.
 Strain Transfer Function – the ratio between the shear strain Fourier amplitude spectrum at
location 1 and the Fourier amplitude spectrum of the acceleration at location 2 (see Section
2.1.2).

Figure 4.38: Screenshot of the Ratios drop-down list.

After selecting the type of ratio, it will be listed in a table on the tab (Figure 4.39). For each ratio, the
following must be specified:

 Location 1 – depth of location 1, either Bedrock or a specific depth.

 Type 1 – boundary conditions at location 1 (Within or Outcrop).
 Location 2 – depth of location 2, either Bedrock or a specific depth.
 Type 2 – boundary conditions at location 2 (Within or Outcrop).

Figure 4.39: Screenshot of the Ratios tab.

73
Soil Types Tab

This tab is used to select the soil types for which the modulus reduction and damping curves are plotted.
This information is most useful when performing Monte Carlo simulations on the nonlinear properties.
Click the soil types from the list provided (Figure 4.40).

Figure 4.40: Screenshot of the Soil Type tab.

Response Spectrum Group Box

This box is used to specify for the response spectrum the damping ratio, the minimum and maximum
periods, the point count (i.e., number of periods), and whether the periods are equally spaced in log
scale or linear scale (Figure 4.41).

Figure 4.41: Screenshot of the Response Spectrum group box.

Frequency Properties Group Box

This box is used to specify for the Fourier Amplitude Spectra the minimum and maximum frequencies,
the point count (i.e., number of frequencies), and whether the frequencies are equally spaced in log
scale or linear scale (Figure 4.42).

74
 Figure 4.42: Screenshot of the Frequency Properties group box.

Logging Properties Group Box

This tab is used to specify the amount of information provided on the Compute page regarding the
equivalent-linear iterations. The logging level may be Low, Medium, or High (Figure 4.43).

Figure 4.43: Screenshot of the Logging Properties group box.

75
4.3 Examples

The following examples give a basic introduction to using Strata to perform equivalent linear site
response analysis. Each of the examples and required acceleration-time series can be installed by
including the examples option during installation. The examples files are found within the examples
folder in the installation path (e.g. C:\Program Files\Strata\examples), or by the shortcut in the Strata
start manual folder. The examples can be opened by either double clicking on the file, or by selecting
them from the ‘Open…’ item from the File menu.

All examples use the deep alluvium Sylmar County Hospital Parking Lot (SCH) site located in
Southern California for the site profile. The soil types and velocity layering of the site was proposed by
Chang (1996). The soil properties are listed in Table 4.1 with a water table at a depth of 46 meters. The
nonlinear properties for each of the layers were computed using the Darendeli (2001) empirical model
with PI=0, OCR=1, and the confining pressures listed in Table 4.1. The corresponding velocity profile is
shown in Figure 4.44 (Chang, 1996). The minimum and maximum values are those recommended in
Chang (1996). The time-averaged shear-wave velocity over the top 30 meters was computed as 273 m/s.

Table 4.1: Soil profile at the Sylmar County Hospital Parking Lot site Chang (1996). The mean
effective stress ( ) is computed assuming a of 1/2 and a water table depth of 46 meters.

Depth Vs
Range (m) Soil Type (m/s) (kN/m3) (kPa) (atm)
0 to 6 Alluvium (Sand) 200 (150 to 230) 18 54 0.36
6 to 31 Alluvium (Sand) 300 (240 to 350) 18 222 2.2
31 to 61 Alluvium (Sand) 460 (370 to 550) 19 562 5.6
Alluvium and Older
61 to 91 700 (580 to 750) 22 776 7.7
Alluvium (Sand)
91+ Bedrock 760 22

76
 Figure 4.44: The shear-wave velocity profile of the Sylmar County Hospital Parking Lot site (CHANG,
1996).

4.3.1 Example 1: Basic time domain

In the first example, the site response is computed for the Sylmar County Hospital Parking Lot site using
a single acceleration-time series. The input file for this example is named example-1-td.strata. In this
example, the steps required to compute the acceleration response spectrum at the surface of the site
are presented. The following steps assume that you are working from a new project.

4.3.1.1 General Settings Page

For the most part, the default values are acceptable on the General Settings page, but there are a few
options that one should be sure to define properly. The following steps are taken to define the proper
information:

1) A title should be created that accurately represents the project. For this example, this example
uses, “Sylmar County Hospital Site -- Time domain.”
2) The default unit system in Strata is the metric system. The units specified on the General
Settings Page must be in agreement with the units used throughout the input file.
3) The default calculation method is ‘Input Motions’ which is used for this example.

77
4.3.1.2 Soil Types Page

The information presented regarding the site information (Table 4.1 and Figure 4.44) is input into Strata
in two different steps. First, the Soil Types Page is used to define the soil types found within the profile.
These soil types are then spatially arranged in the Soil Profile Page. The following steps are required to
define the Soil Types for the Sylmar County Hospital Site:

1) In the Soil Types dialog, click the Add Button to create a new soil type and enter the information
found in Table 4.1. The nonlinear property model is selected from a drop down list. Select the
Darendeli & Stokoe (2001) nonlinear model for both the shear modulus reduction and damping
curves.

2) Additional information regarding the soil type is required for the Darendeli & Stokoe (2001)
nonlinear model and is found within the Darendli and Stokoe Model Parameters group box. The
soil is an alluvium with a plasticity index of 0 and an over-consolidation ratio of 1, which are the
default values for the PI and OCR. The excitation frequency and number of cycles will also be left
at the default values (1 Hz and 10, respectively). The Darendeli (2001) model requires the mean
effective stress for the soil, which is 0.36 atm for the center of the top alluvium layer.

3) Repeat steps (1) and (2) for the remaining three soil types.
4) The bedrock layer is defined by a unit weight of 22 kN/m3 and damping of 1%.

4.3.1.3 Soil Profile Page

The Soil Profile Page is used to spatially arrange the soil types and define the shear-wave velocity
associated with each layer. The following steps are used to input the soil profile information:

1) In the Site Profile table, click on the Add button to create a new soil profile layer.

78
 2) Define the thickness (6 m), soil type (Alluvium (0.36 atm)), and shear-wave velocity (200 m/s) of
the layer.
3) Repeat steps (1) and (2) until all of the layers are defined.

4.3.1.4 Motion (s) Page

For this example, the site response analysis is being computed for one recorded motion. This motion is
loaded using the following steps:

1) The motion will be input at the top of the bedrock, so the default location to input the motion is
correct.
2) The motion is loaded by clicking on the Add button in the Input Motion(s) dialog. This will open
a dialog box that assists in loading the file (see Section 4.1.5). Click on the File button in the
upper left portion of the dialog and select the VAS090.AT2 file from the NORTHR directory
(depending on where you installed Strata this might be “C:\Program
Files\Strata\example\motions\NORTHR”). Strata automatically parses the AT2 file and fills in the
required information. For motions not in the AT2 format, all of the boxes would need to be filled
in by hand.

79
 3) Click on the OK button to complete the loading of the file.

4.3.1.5 Output Specification Page

In this example, the goal is to compute the acceleration response spectrum at the surface of the site.
The requested output will also include the shear strain profile, the acceleration response spectrum at
the bedrock, and the spectral ratio between the surface and input rock motions.

1) For the acceleration-response spectrum at the surface, click on the Add button in the Response
and Fourier Spectra tab and choose Acceleration Response Spectrum from the drop-down list.
Assign the location of the layer to be at the surface (0 m) and select an outcrop boundary
condition.
2) To obtain the input acceleration response spectrum at the bedrock level, click on the add button
to generate another Acceleration Response Spectrum. For this row, use the drop down box in
the location column to set the depth to be bedrock and select the outcrop boundary condition.

3) For the spectral ratio, click on the Add button in the Ratios tab. The default values for the
locations are correct (Location 1 at 0 m, Location 2 at Bedrock).

4) The maximum shear-strain profile is enabled by clicking on the check box in the Profiles tab.

80
 5) Set the logging level to Medium in the Logging Properties. This logging level will report the
maximum error for each iteration.

4.3.1.6 Compute Page

The calculation is started by clicking on the Compute Button. If you wish you stop the calculation at any
time press the Cancel Button. After the calculation is complete, the Results Page will be selected
automatically.

4.3.1.7 Results Page

The Results Page allows the user to immediately view the results in a plot. To change the plotted
information, select the desired parameter from the Output dropdown list. The results can also be
viewed by clicking on the data tab in this page. The data can be exported to use in another program (like
Excel) by selecting Export from the File menu or by copy and paste to the spreadsheet.

4.3.2 Example 2: Time Series with Multiple Input Motions

In this example, the response at the surface of the site is computed for a suite of input motions. The site
properties and output are defined using the same procedure as in Example 1. In this example, the site
response is computed for a suite of 10 input motions, each with a different scale factor. This example
can be directly loaded from the ‘example-2-td.strata’ file in the examples directory.

The suite of input motions is presented in Table 4.2. Two different methods may be used to
enter these motions into Strata. The direct method is to add each motion individually using the
procedure described in Section 4.1.5. If the suite is composed only of AT2 files from the NGA database it
is possible to add them using a suite list file. This procedure can be convenient for suites that include a
large number of motions. The suite list for the suite in Table 4.2 can be found in the examples directory
and is named ‘suite-10-1.csv’. To load the suite, follow these steps:

1) Select the Motion(s) Page

2) Click on the Load Suite Button
81
 3) Select the appropriate suite file.

After the suite is loaded, switch to the Compute Page and start the calculation.

Table 4.2: Suite of input motions used in Example 2

File Scale Factor

CHICHI03\TCU138-W.AT2 1.17
NORTHR\H12180.AT2 1.15
CHICHI06\TCU076-E.AT2 1.19
LOMAP\GIL067.AT2 1
NORTHR\VAS090.AT2 1.17
NORTHR\LV3090.AT2 1.3
MORGAN\G01230.AT2 1.41
ITALY\B-CTR000.AT2 1.11
NORTHR\HOW330.AT2 1.2
ITALY\B-BAG270.AT2 1.48

After the calculation is completed the Results Page is selected. The plot shows the individual
response spectra (light gray), as well as the median (solid blue line) and plus and minus one standard
deviation (dashed blue line) response spectra (Figure 4.45). The currently selected record is shown in a
thick green line if the motion is enabled or in a thick red line if it is disabled. If the result is disabled then
it is not included in the statistics. Individual responses can be selected by either clicking on them in the
plot window, or selecting them from the table.

82
 Figure 4.45: Example of a plot with multiple responses.

4.3.3 Example 3: RVT and Site Variation

This example uses the same site properties as in the previous examples, but the input motion is defined
using a response spectrum and the site response calculation utilizes random vibration theory.
Additionally, the shear-wave velocity of the site is varied. This example can be directly loaded from the
‘example-3-rvt.strata’ file in the examples directory. After the soil types and site profile are defined
using the procedures presented in Section 4.3.1, the following changes need to be made:

1) General Settings Page

a) Select Random Vibration Theory from the Approach dropdown list in the Type of Analysis group
box.

b) Check the ‘Vary the properties’ check box to enable site property variation.

c) Set the number of realizations to be 30.

83
 d) Disable the variation of the nonlinear soil properties, and enable variation of the site profile.

2) Soil Profile Page -- now shown in the Soil Profile Page are the widgets that control the site profile
variation. This example uses a user specified standard deviation for the site, but uses the generic
correlation model.

a) Check the ‘Vary the shear-wave velocity of the layers’ to enable the setup of velocity variation
parameters.

b) The standard deviation (in natural log units) of the site is set in the box below the ‘Custom’
selection. For this example the standard deviation is specified as 0.15.

c) The Vs30 of site is used as a guide to select the 180 to 360 m/s correlation model.

d) Make sure the ‘Varied’ check box for every soil layer is checked, so that the variation of shear-
wave velocity of the layers is enabled.

84
3) Motion(s) Page

a) Click on the Add button and select ‘Response Spectrum Compatible’ from the dropdown list.

b) The response spectrum of the event is loaded into Strata by copying and pasting.

i) Using a spreadsheet program, open the file name ‘response-spectrum.csv’ from the
examples directory. If the .csv file is associated with Excel, this can be done by double
clicking on the file. The damping for this response spectrum is 5%.

ii) Select the columns of data from the spread sheet and select Copy from the Edit menu.

iii) Click on the Target Response Spectrum dialog and then right click and select Paste.

c) Define the duration of the event to be 6.68 seconds.

d) Click the Apply button, and the computed Fourier amplitude spectrum can be viewed by clicking
on the FAS Plot tab.

85
4) Compute Page

The necessary changes have been made to compute the response of the site using RVT with variation of
the shear-wave velocity. To see the results, start the calculation on the Compute page.

86
5 References

Abrahamson, N. A., & Silva, W. J. (1996). Empirical ground motion models. Upton, New York: Brookhaven
National Laboratory.

Abrahamson, N. A., & Silva, W. J. (1997). Empirical response spectral attenuation relations for shallow
crustal earthquakes. Seismological Research Letters, 94-127.

Ang, A. H.-S., & Tang, W. H. (1975). Probability Concepts in Engineering Planning and Design: Volume I--
Basic Principles. New York: John Wiley and Sons.

Ang, A. H.-S., & Tang, W. H. (1990). Probability Concepts in Engineering Planning and Design: Volume II--
Descision, Risk, and Reliability. Ang and Tang.

Atkinson, G. M., & Boore, D. M. (1995). Ground-Motion Relations for Eastern North America. Bulletin of
the Seismological Society of America, Vol. 85, No. 1, pp. 17-30.

Boore, D. M. (2003, March). Simulation of Ground Motion Using the Stochastic Method. Pure and
Applied Geophysics, 160(3-4), 635-676.

Boore, D. M., & Joyner, W. B. (1984). A note on the use of random vibration theory to predict peak
amplitudes of transient. Bulletin of the Seismological Society of America, 74(5), 2035-2039.

Brune, J. N. (1970). Tectonic stress and the spectra of seismic shear waves from earthquake. Journal of
Geophysics Research, 75(26), 4997-5009.

Brune, J. N. (1971). Correction. Journal of Geophysics Research, 76, 5002.

Campbell, K. W. (1997). Empirical near-source attenuation relationships for horizontal and veritcal
components of peak ground acceleraion, peak ground velocity, and pseudo-absolute accleration
response spectra. Seismological Research Letters, 154-179.

Campbell, K. W. (2003). Prediction of strong ground motion using the hybrid empirical method and its
use in the development of ground-motion (attenuation) relations in the Eastern North America.
Bulletin of the Seismological Society of America, 93(3), 1012-1033.

Cartwright, D. E., & Longuet-Higgins, M. S. (1956). The Statistical Distribution of the Maxima of a
Random Function. Proceedings of the Royal Soceity of London, Series A, Mathematical and
Physical, 237(1209), 212-232.
87
Chang, S. W.-Y. (1996). Seismic response of deep stiff soil deposits. Berkeley: University of California.

Coppersmith, K. (1991). Ground Motion Following Selection of SRS Design Basis Earthquake and
Associated Deterministic Approach. San Francisco, CA: GeoMatrix Consultants.

Costantino, C. J., Heymsfield, E., & Gu, Y. T. (1991). Site Specific Estimates of Surface Ground Motions for
the K-Reactor Site, Savannah River Plant. Upton, New York: Structural Analysis Division, Nuclear
Energy Department, Brookhaven National Laboratory.

Darendeli, M. B. (2001). Development of a new family of normalized modulus reduction and material
damping curves. Austin, Texas: The University of Texas.

EPRI. (1993). Guidelines for Determining Design Ground Motions. EPRI TR-102293.

Gasparini, D. A., & Vanmarcke, E. H. (1976). Simulated earthquake motions compatible with prescribed
response spectra. Cambridge, Massachusetts: Massachusetts Institute of Technology.

Geotechnical Engineers Inc. (GEI). (1983). Evaluation of Dynamic Soil Profiles for the F-Area Sand Filter
Structures. Boston, MA.

Green, R. (2001). Energy-Based Evaluation and Remediation of Liquefiable Soils. Blacksburg, VA: Ph.D.
Dissertation, Department of Civil and Environmental Engineering, Virginia Polytechnic Institute
and State University (Virginia Tech).

Green, R. a. (2005, 4). Number of Equivalent Cycles Concept for Liquefaction Evaluations - Revisited.
Journal of Geotechnical and Geoenvironmental Engineering, ASCE, pp. 477-488.

Hardin, B. O., & Drnevich, V. P. (1972, June). Shear modulus and damping in soils: design equations and
curves. Journal of soil mechanics and foundation engineering division, ASCE, 78, 667-692.

Idriss, I. M., & Sun, J. I. (1992). SHAKE91: a computer program for conducting equivalent linear seismic
response analyses of horizontally layered soil deposits. Davis, California: Center for Geotechnical
Modelling, Department of Civil and Environmentall Engineering, University of California.

Iwasaki, T., Tatsuoka, F., & Takagi, Y. (1976). Dynamic Shear Deformation Properties of Sand for Wide
Strain Range. Tokyo, Japan: Civil Engineering Institue, Ministry of Construction.

Kao, E. P. (1997). An Introduction to Stochastic Processes. Belmont, CA: Duxbury Press.

Kramer, S. L. (1996). Geotechnical Earthquake Engineering. Upper Saddle River, New Jersey: Prentice
Hall.

McGuire, R. K., Toro, G. R., O'Hara, T. P., Jacobson, J. P., & Silva, W. J. (1989). Probablistic seismic hazard
evaluations at nuclear plant sites in the Central and Eastern United States: Resolution of the
Charleston Earthquake Issue. Palo Alto, California: Electric Power Research Institute.

88
Rathje, E. M., & Kottke, A. R. (2008). Procedures for Random Vibration Theory Based Seismic Site
Response Analyses. Rockville, MD: Nuclear Regulatory Commission.

Rathje, E. M., Kottke, A. R., & Ozbey, C. M. (2005). Using inverse random vibration theory to develop
input Fourier amplitude spectra for use. 16th International Conference on Soil Mechanics and
Geotechnical Eningeering: TC4, (pp. 160-166). Osaka, Japan.

Schnabel, P. B., Lysmer, J., & Seed, H. B. (1972). SHAKE: A computer program for earthquake response
analysis of horizontally-layered. University of California, Berkeley, CA: Earthquake Engineering
Research Center.

Schneider, J. F., Silva, W. J., Chiou, S., & J.C., S. (1991). Estimation of ground motion at close distances
using the band-limited-white-noise model. Fourth Interntional Conference on Seismic Zonation
(pp. 187-194). Stanford, CA: EERI.

Seed, H. B., & Idriss, I. M. (1970). Moduli and Dynamic Factors for Dynamic Response Analyses.
University of California, Berkeley: Earthquake Engineering Research Center.

Silva, W. J., Abrahamson, N., Toro, G., & Costantino, C. (1997). Description and validation of the
stochastic ground motion model. Upton, New York: Brookhaven National Laboratory.

Toro, G. R. (1995). Probablistic models of site velocity profiles for generic and site-specific ground-motion
amplification studies. Upton, New york: Brookhaven National Laboratory.

Toro, G. R., Silva, W. J., McGuire, R. K., & Hermann, R. B. (1992). Probabilistic Seismic Hazard Mapping of
the Mississippi Embayment. Seismological Research Letters, 63(3), 449-475.

Turner, E., & Stokoe II, K. H. (1982). Static and Dynamic Properties of Clayey Soils Subjected to 1979
Imperial Valley Earthquake. Menlo Park, CA: USGS Geological Engineering Branch.

Vucetic, M., & R, D. (1991). Effect of Soil Plasticity on Cyclic Response. Journal of the Geoechnical
Engineering Division, ASCE, 89-107.

89