Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

G - Articulo - Jibson - Predicting Earthquake-Induced Landslide Displacements Using Newmark's Sliding Block Analysis

Download as pdf or txt
Download as pdf or txt
You are on page 1of 23

PREDICTING EARTHQUAKE-INDUCED LANDSLIDE DISPLACEMENTS

USING NEWMARK'S SLIDING BLOCK ANALYSIS

Randall W. Jibson

ABSTRACT
Landslides typically cause a large proportion of earthquake damage, and the ability to
predict slope performance during earthquakes is important for many types of seismichazard analysis and for the design of engineered slopes. Newmark's method for modeling a
landslide as a rigid-plastic block sliding on an inclined plane provides a useful method for
predicting approximate landslide displacements; this method yields much more useful
information than pseudostatic analysis and is far more practical than finite-element
modeling. Applying Newmark's method requires knowing the yield or critical acceleration
of the landslide (above which permanent displacement occurs), which can be determined
from the static factor of safety and from the landslide geometry. Earthquake accelerationtime histories can be selected to represent the shaking conditions of interest, and those parts
of the record that lie above the critical acceleration are double integrated to determine the
permanent landslide displacement. For approximate results, a simplified Newmark method
can be used, which estimates Newmark displacement as a function of landslide critical
acceleration and earthquake shaking intensity.

INTRODUCTION
Most moderate and large earthquakes trigger landslides, and these landslides commonly
cause a significant proportion of total earthquake damage. Earthquakes having magnitudes
greater than 4.0 can trigger landslides on very susceptible slopes near the epicenter, and
earthquakes having magnitudes greater than 6.0 can generate widespread landsliding
(Keefer, 1984). Accurately predicting which slopes will move and the severity of that
movement, however, is difficult. This discussion, which is modified from that in Jibson
(1993), briefly reviews some published methods to predict earthquake-triggered slope
displacement and shows how these methods can be applied to practical problems. The
ability to predict approximate amounts of earthquake-induced landslide movement can be
used for regional seismic-hazard analysis and in designing slopes to withstand earthquake
shaking.
The seismic performance of a slope can be evaluated in several ways. The simplest and
most widely used approach is pseudostatic analysis, in which an earthquake acceleration
acting on the mass of a potential landslide is treated as a permanent static body force in a
limit-equilibrium (factor-of-safety) analysis. Different earthquake accelerations are applied
1

iteratively until the factor of safety is reduced to 1.0. The earthquake acceleration needed to
reduce the factor of safety to 1.0 is called the yield acceleration, the exceedance of which is
defined as failure. This procedure is simple and requires no more information than is
needed for a static factor-of-safety analysis. Pseudostatic analysis is useful for identifying
yield accelerations and, hence, peak ground accelerations (PGA) below which no slope
displacement will occur. In cases where the PGA does exceed the yield acceleration,
however, pseudostatic analysis has proved to be over-conservative in most cases: many
slopes experience transient earthquake accelerations well above their yield accelerations but
experience little or no permanent displacement (Newmark, 1965; Wilson and Keefer,
1983). The utility of pseudostatic analysis is thus limited because it provides only a single
numerical threshold below which no displacement is predicted and above which total, but
undefined, "failure" is predicted. In fact, pseudostatic analysis tells the user nothing about
what will occur when the yield acceleration is exceeded.
On the other end of the spectrum, advances in two-dimensional finite-element modeling
have facilitated very accurate evaluation of strain potentials and permanent slope
deformation (Seed and others, 1975; Taniguchi and others, 1983; Prevost and others, 1985;
Elgamal and others, 1987). These highly sophisticated methods require a broad spectrum of
data of extremely high quality and density, which, combined with the intensive computing
capacity required, make their general use prohibitively expensive (Chang and others, 1984).
Newmark (1965) proposed a method of analysis that bridges the gap between simplistic
pseudostatic analysis and sophisticated, but generally impractical, finite-element modeling.
Newmark's method models a landslide as a rigid-plastic friction block having a known
yield or critical acceleration, the acceleration required to overcome frictional resistance and
initiate sliding on an inclined plane (Fig. 1). The analysis calculates the cumulative
permanent displacement of the block as it is subjected to the effects of an earthquake
acceleration-time history, and the user judges the significance of the displacement.
Laboratory model tests (Goodman and Seed, 1966) and analyses of earthquake-induced
landslides in natural slopes (Wilson and Keefer, 1983) confirm that Newmark's method
fairly accurately predicts slope displacements if slope geometry, soil properties, and
earthquake ground accelerations are known. Newmark's method is relatively simple to
apply and provides a quantitative prediction of the inertial landslide displacement that will
result from a given level of earthquake shaking. Results from Newmark's method also are
useful in probabilistic analyses (Lin and Whitman, 1986; Yegian and others, 1991), which
further enhances their utility.

Figure 1. Sliding-block model used for Newmark analysis. The potential landslide is
modeled as a block resting on a plane inclined at an angle () from the horizontal. The
block has a known critical (yield) acceleration (ac), the base acceleration required to
overcome shear resistance and initiate sliding with respect to the base. The block is
subjected to a base acceleration (a) representing the earthquake shaking.

PAST APPLICATIONS OF NEWMARK'S METHOD


Newmark's method has been applied rigorously in a variety of ways to slope-stability
problems. Most applications have dealt with the seismic performance of dams and
embankments (Yegian and others, 1991; Makdisi and Seed, 1978), which was Newmark's
(1965) original intent. Newmark's method also has been successfully applied to landslides
in natural slopes (Wilson and Keefer, 1983). Several simplified approaches have been
proposed for applying Newmark's method; these involve developing empirical relationships
to predict slope displacement as a function of critical acceleration and one or more
measures of earthquake shaking. Many such studies plot displacement against critical
acceleration ratio-the ratio of critical acceleration to PGA (Franklin and Chang, 1977;
Makdisi and Seed, 1978; Ambraseys and Menu, 1988). Other studies have related critical
acceleration ratio to some normalized form of displacement: Yegian and others (1991)
calculated exceedance probabilities for displacement normalized by PGA, the equivalent
number of earthquake shaking cycles, and the square of the period of the base motion; Lin
and Whitman (1986) used simple, artificial ground-motion wave forms (triangular,
rectangular, or sinusoidal) to relate critical acceleration ratio to displacement normalized by
PGA. Jibson (1993) related Newmark displacement to critical acceleration and Arias
intensity, an earthquake-shaking measure described below. Miles and Ho (1999) and Miles
and Keefer (2000) compared results from these simplified methods with their own method
of rigorously integrating artificially generated strong-motion time histories.
Wilson and Keefer (1983) applied Newmark's method to a landslide triggered by the
1979 Coyote Creek, California, earthquake. The slide occurred near a strong-motion
instrument, and the landslide displacement predicted in the Newmark analysis using the
3

record from that instrument agreed well with the observed displacement. This method of
using real acceleration-time histories to predict displacements in natural slopes has been
applied to experimentally predict and map seismic slope stability in San Mateo County,
California (Wieczorek and others, 1985). It has been adapted to back-analyze shaking
conditions required to trigger landslides formed in the Mississippi Valley during the 181112 New Madrid earthquakes (Jibson and Keefer, 1993) and to reconstruct failure conditions
of other seismically triggered landslides (Jibson 1996; Jibson and Harp, 1996). It also forms
the basis for constructing digital seismic landslide hazard maps in southern California
(Jibson and others, 1998, 2001).

TYPES OF SLIDING-BLOCK ANALYSIS CURRENTLY IN USE


Since Newmark first introduced his analytical method in 1965, several variations have
been proposed that are designed to yield more accurate estimates of slope displacement by
modeling the dynamic slope response more rigorously. This, of course, involves a trade-off.
One great advantage of Newmark's method is it's theoretical and practical simplicity. This
simplicity, however, is the result of many assumptions that limit the accuracy of the results
in many cases. Chief among these limitations is the assumption of rigidity--that the
landslide block is perfectly rigid and experiences no internal deformation. This assumption
is reasonable for relatively thin landslides in stiff or brittle materials, but it introduces
significant errors as landslides become thicker and material becomes softer. More
sophisticated methods do a better job of modeling the dynamic elastic response of the
landslide material and thus yield more accurate displacement estimates, but again, there is a
trade-off in the complexity of the analysis and the difficulty in acquiring the needed input
parameters.
At present, analytical procedures for estimating permanent co-seismic landslide
displacements can be grouped into three types: rigid-block, decoupled, and coupled.
Rigid-block analysis. Rigid-block analysis is the analysis first developed by Newmark
(1965), which was described in detail below in the section "Conducting a Newmark
Analysis". To briefly summarize, it treats the potential landslide block as a rigid mass (no
internal deformation) that slides in a perfectly plastic manner on an inclined plane. Thus,
the mass experiences no permanent displacement until the base acceleration exceeds the
critical (yield) acceleration of the block; when the base acceleration exceeds the critical
acceleration, the block begins to move downslope. Displacements are estimated by doubleintegrating the parts of an acceleration-time history that lie above the critical acceleration.
Decoupled analysis. More sophisticated analyses were developed to account for the fact
that potential landslide masses are not rigid bodies but deform internally when subjected to
seismic shaking. The most commonly used of such analyses was developed by Makdisi and
Seed (1978) and estimates the effect of dynamic response on permanent slip in a two-step
procedure: (1) Perform a dynamic analysis of the slope (using programs such as QUAD4 or
SHAKE) assuming no failure surface; estimate acceleration time histories at several points
within the slope and develop an average acceleration time history for the slope mass above
4

the potential failure surface. (2) Use this average time history as the input in a rigid-block
analysis and estimate the permanent displacement. This approach is commonly referred to
as a decoupled analysis because the computation of the dynamic response and the plastic
slip are performed independently. Decoupled analysis thus does not take into account the
effects of slip on the ground motion.
Coupled analysis. In a coupled analysis, the dynamic response of the slide mass and the
permanent displacement are modeled together so that the effect of plastic slip on the ground
motions is accounted for. Lin and Whitman (1983) pointed out that the assumptions of the
decoupled analysis introduce errors in the estimation of total slip and compared results for
coupled and decoupled analyses. They showed that, in general, decoupled analysis yielded
conservative results that were within about 20 percent of the coupled results. More recently,
Rathje and Bray (2000) compared results from rigid-block analysis with linear and nonlinear coupled and decoupled analyses.

WHICH ANALYSIS SHOULD BE USED?


Selecting which of these types of analyses to conduct is critical in getting accurate
displacement estimates. The current state of knowledge suggests that the best basis for this
selection is the period ratio, Ts/Tm, the ratio of the fundamental site period (Ts) to the mean
period of the earthquake shaking (Tm) (Rathje and Bray, 1999; 2000). The fundamental site
period can be estimated as
Ts = 4h/Vs

(1)

where h is the maximum vertical distance between the ground surface and slip surface used
to estimate the yield acceleration and Vs is the shear-wave velocity of the materials above
the slip surface. Mean period of the earthquake shaking was defined by Rathje and others
(1998) as the inverse of the weighted average frequency over a frequency range of 0.25 to
20 Hz. Mean period can be estimated for rock site conditions as a function of earthquake
magnitude (M) and source distance (r, in km) as follows:
ln(Tm) = ln(0.411 + 0.0837(M - 6) + 0.00208r)
ln(Tm) = ln(0.516 + 0.00208r)

for M 7.25
for 7.25 M 8.0

(2a)
(2b)

As a general rule, coupled analysis gives good results for all conditions, but, of course, it
is the most complex to conduct. The following table provides general guidelines for
selecting between rigid-block and decoupled analysis in the terms of the period ratio.

Rigid-block analysis is thus appropriate for analyzing thin, stiff landslides having a period
ratio of 0.2 or less. Between 0.2 and 1, rigid-block analysis yields unconservative results
and should not be used. For period ratios between 1 and 2, rigid-block analysis yields
conservative results, but decoupled analysis give results closer to results from coupled
analysis, which is considered the most accurate result. For period ratios greater than 2,
rigid-block analysis yields highly over-conservative results that significantly overestimate
displacement.

CONDUCTING A NEWMARK ANALYSIS


Before describing how to apply Newmark's method, the limiting assumptions need to be
stated. Newmark's method treats a landslide as a rigid-plastic body, that is, the mass does
not deform internally, experiences no permanent displacement at accelerations below the
critical or yield level, and deforms plastically along a discrete basal shear surface when the
critical acceleration is exceeded. Thus, Newmark's method is best applied to coherent
landslides such as translational block slides and rotational slumps. Other limiting
assumptions commonly are imposed for simplicity but are not required by the analysis:
1. The static and dynamic shearing resistance of the soil are taken to be the same
(Newmark, 1965; Chang and others, 1984).
2. The effects of dynamic pore pressure are neglected. This assumption generally is
valid for compacted or overconsolidated clays and very dense or dry sands (Newmark,
1965; Makdisi and Seed, 1978).
3. The critical acceleration is not strain dependent and thus remains constant throughout
the analysis (Newmark, 1965; Makdisi and Seed, 1978; Chang and others, 1984;
Ambraseys and Menu, 1988). The accompanying programs do not require this and allow
for strain-dependant changes in critical acceleration.

4. The upslope resistance to sliding is taken to be infinitely large such that upslope
displacement is prohibited (Newmark, 1965; Chang and others, 1984; Ambraseys and
Menu, 1988).
The following sections outline the procedure for conducting a Newmark analysis and
provide simple examples of its application.

Critical (Yield) Acceleration


The first step in the analysis is to determine the critical or yield acceleration of the
potential landslide. One way to do this is to use pseudostatic analysis, where critical
acceleration is determined by iteratively employing different permanent horizontal
earthquake accelerations in a static limit-equilibrium analysis until a factor of safety of 1.0
is achieved.
Newmark (1965) simplified this approach by showing that the critical acceleration of a
potential landslide is a simple function of the static factor of safety and the landslide
geometry; it can be expressed as

(3)

ac = (FS - 1)g sin,

where ac is the critical acceleration in terms of g, the acceleration due to Earth's gravity; FS
is the static factor of safety; and is the angle (herein called the thrust angle) from the
horizontal that the center of mass of the potential landslide block first moves. Thus,
determining the critical acceleration by this method requires knowing the static factor of
safety and the thrust angle.

Factor of Safety
As noted by Newmark (1965), modeling dynamic slope response requires undrained or
total shear-strength parameters. During earthquakes, slope materials behave in an undrained
manner because excess pore pressures induced by dynamic deformation of the soil column
cannot dissipate during the brief duration of the shaking. Undrained strength also is called
total strength because the contributions of friction, cohesion, and pore pressure are not
differentiated, and the total strength is expressed as a single quantity.
The factor of safety can be determined using any appropriate method that uses undrained
or total shear strength. In materials whose drained and undrained behaviors are similar,
drained or effective shear strengths can be used if undrained strengths are unavailable or
difficult to measure. This allows great flexibility for users. For a rough estimate of
7

displacement, a simple factor-of-safety analysis, perhaps of an infinite slope using


estimated shear strength, could be used. On the other end of the spectrum, a highly detailed
site study could be conducted to determine the factor of safety very accurately. Clearly, the
accuracy of the safety factor, and the resulting predicted displacement, depends on the
quality of the data and analysis, but the user determines what is appropriate.

Thrust Angle
The thrust angle is the direction the center of gravity of the slide mass moves when
displacement first occurs. For a planar slip surface parallel to the slope (an infinite slope),
this angle is the slope angle. For simple planar block sliding, the thrust angle is the
inclination of the basal shear surface. For circular rotational movement, Newmark (1965)
showed that the thrust angle is the angle between the vertical and a line segment connecting
the center of gravity of the slide mass and the center of the slip circle. For irregular shear
surfaces, the thrust angle can be approximated visually, by estimating an "equivalent"
circular surface, or by averaging the inclinations of line segments approximating the
surface.

Calculation of Critical Acceleration


Figure 2 illustrates a simple hypothetical slope and the critical failure surface having the
lowest factor of safety (1.4) in undrained conditions. Newmark's (1965) geometric
construction indicates a thrust angle of 30o. According to Equation 3, a factor of safety of
1.4 and a thrust angle of 30o would yield a critical acceleration of 0.20 g.

Figure 2. Model of hypothetical slope: heavy line is basal shear surface; FS is factor of
safety; thrust angle is 30o.

Earthquake Acceleration-Time History


The most difficult aspect of conducting a Newmark analysis is selecting an input ground
motion, and many ways of doing so have been proposed. Most studies have used some
combination of the two approaches mentioned by Newmark: (1) scaling acceleration-time
histories from actual earthquakes to a desired level of PGA (Franklin and Chang, 1977;
Makdisi and Seed, 1978) and (2) using single or multiple cycles of artificial acceleration
pulses having simple rectangular, triangular, or sinusoidal shapes (Lin and Whitman, 1986;
Yegian and others, 1991). Both of these approaches yield useful results, but both also have
inherent weaknesses. Scaling an acceleration-time history by simply expanding or
contracting the acceleration scale does not accurately represent ground motion from
earthquakes of different magnitudes or proximities because magnitude and source distance
also affect the duration and predominant periods of shaking. And using simple artificial
pulses of ground shaking is an unnecessary oversimplification in light of the current
availability of digitized acceleration-time histories having a broad range of attributes.
Selecting a time history requires the user to know something of the shaking
characteristics or design requirements pertinent to the situation of interest. Common design
or hazard-assessment criteria include (1) a specified level of ground shaking, (2) a model
earthquake of specified magnitude and location, or (3) an acceptable design amount of
earthquake-triggered displacement.
9

Selecting a Time History for a Specified Level of Ground Shaking


Criterion (1) is by far the simplest; it requires only that the user locate a sampling of
digitized acceleration-time histories having the desired measure of earthquake shaking
intensity near the specified level. PGA is a common measure of ground-shaking intensity,
and digitized time histories having a wide variety of PGA's, even approaching 2 g,
currently are available.
A weakness of this approach is that PGA measures only a single point in an
acceleration-time history and is thus a rather crude measure of shaking intensity. A more
comprehensive and quantitative measure of total shaking intensity developed by Arias
(1970) is useful in seismic hazard analysis and correlates well with the distribution of
earthquake-induced landslides (Harp and Wilson, 1995; Jibson and others, 1998, 2001).
Arias intensity is the integral over time of the square of the acceleration, expressed as

(4)

where Ia is Arias intensity in units of velocity, g is the acceleration of Earth's gravity, a(t) is
the ground acceleration as a function of time, and T is the total duration of the strong
motion. An Arias intensity thus can be calculated for each directional component of a
strong-motion record. In cases where a given level of Arias intensity can be specified,
selecting a strong-motion record of similar intensity is quite simple, and currently available
records span a range of Arias intensities up to almost 25 m/s.
Selecting a Time History for a Specified Earthquake Magnitude and Location
Criterion (2) can be somewhat more difficult. If acceleration-time histories exist for
earthquakes of the desired magnitude that were recorded at appropriate distances, then they
can be used. Satisfying both magnitude and distance requirements is often impossible,
however, so it may be necessary to estimate shaking characteristics at the site of interest
using published empirical or theoretical relationships that predict PGA, duration, and Arias
intensity as a function of earthquake magnitude and source distance. Estimated shaking
characteristics can then be compared with those from existing time histories to provide a
basis for selecting appropriate records.
An example of this procedure is from the Mississippi Valley, where large earthquakes
occurred in 1811-12 but where no strong-motion records exist. The problem is to predict
the performance of a slope in a moment-magnitude (M) 6.2 earthquake centered at least 8

10

km away. If no time histories for that magnitude and distance existed, shaking
characteristics at the site would have to be estimated.
PGA in this example can be estimated using the attenuation relationship of Nuttli and
Herrmann (1984) for soil sites in the central United States:

log PGA = 0.57 + 0.50 mb - 0.83 log(R2 + h2)1/2 - 0.00069 R

(5)

where PGA is in centimeters per second squared, mb is the body-wave magnitude, R is the
epicentral distance in kilometers, and h is the focal depth in kilometers. A M-6.2
earthquake corresponds to mb=5.8 (Heaton and others, 1986). For mb=5.8, an epicentral
distance of 8 km, and a minimum focal depth of 3 km, Equation 5 predicts a PGA of 491
cm/s2 or 0.50 g. Methods for estimating strong-ground-shaking parameters in many
different tectonic environments were published by the Seismological Society of America
(1997).
Estimating the Arias intensity at the site can be done in more than one way. Wilson and
Keefer (1985) developed a relationship between Arias intensity, earthquake magnitude, and
source distance:

log Ia = M - 2logR - 4.1,

(6)

where Ia is in meters per second, M is moment magnitude, and R is earthquake source


distance in kilometers. For M=6.2 and R=8 km, Equation 6 predicts an Arias intensity at
the site of 1.97 m/s.
Arias intensity also correlates closely with the combination of PGA and duration. R.C.
Wilson (U.S. Geological Survey, unpublished data) developed an empirical equation using
43 strong-motion records to predict Arias intensity from PGA and a specific measure of
duration:

Ia = 0.9(PGA2)(D5-95%),

(7)

where Ia is in meters per second, PGA is in g's, and D5-95% is duration (hereafter called
Dobry duration) in seconds, defined as the time required to build up the central 90 percent
of the Arias intensity (Dobry and others, 1978). Estimating Arias intensities using this
method requires an estimate of the duration of strong shaking. Dobry and others (1978)
proposed an empirical relationship between duration and earthquake magnitude:

11

log D5-95% = 0.432M - 1.83,

(8)

where D5-95% is Dobry duration in seconds and M is unspecified earthquake magnitude


(probably local magnitude, ML). In the magnitude range of interest, ML values are generally
identical to M values (Heaton and others, 1986), so M=6.2 yields a Dobry duration of 7.1 s.
If this duration and the PGA of 0.50 g estimated above are used in Equation 7, an Arias
intensity of 1.59 m/s is predicted, which agrees fairly well with that estimated by Equation
6.
More recently, Abrahamson and Silva (1996) developed a much more rigorous equation
to estimate median values of D5-95%. For earthquake source distances greater than 10 km,
the following equation is used:

(9a)

where M is magnitude and r is source distance. For distances less than 10 km, the following
equation is used

(9b)

These three indices of shaking intensity-PGA, duration, and Arias intensity-form a


rational basis for selecting strong-motion records for analysis. Caution and judgment must
be used in making these estimates, however, because the process of combining values from
Equations 5-9, each of which has a range of possible error, compounds the uncertainty at
each step. For this example, three records are chosen whose shaking characteristics
reasonably match those estimated (Table 1). Selecting multiple records that span the range

12

of estimated shaking characteristics provides a range of displacements whose significance


the user can judge.
The frequency content of earthquake shaking also significantly affects slope response.
The mean period, defined above in equation 2, is the most useful measure of frequency
content and can be used to identify earthquake records that are appropriate for specific
analyses.

Selecting a Time History for a Specified Design Displacement


Criterion (3) differs from the first two in that a limiting damage level (landslide
displacement) is specified rather than the level of ground shaking. An example is to
estimate the maximum level of ground shaking a slope having a critical acceleration of 0.20
g could experience without exceeding 10 cm of displacement.
One approach to this problem is to simply iteratively analyze several strong-motion
records to find those that yield about 10 cm of displacement at ac=0.20 g. The magnitudes,
source distances, focal depths, PGA's, Arias intensities, and durations of these records
could then be examined to discern the approximate range of conditions the slope could
13

withstand. Obviously, this approach could be time consuming, but it would produce a
variety of possible threshold ground-shaking scenarios.
An easier approach to this type of problem is to apply the simplified Newmark method
discussed subsequently.

Calculating Newmark Displacement


Once the critical acceleration of the landslide has been determined and the accelerationtime histories have been selected, Newmark displacement can be calculated by double
integrating those parts of the strong-motion record that lie above the critical acceleration.
Several methods for doing this, some rigorous and others highly simplified, have been
proposed (Newmark, 1965; Makdisi and Seed, 1978; Chang and others, 1984; Ambraseys
and Menu, 1988). Perhaps the most useful rigorous method was developed by Wilson and
Keefer (1983). Figure 3A shows a strong-motion record having a hypothetical ac of 0.2 g
superimposed. To the left of point X, accelerations are less than ac, and no displacement
occurs. To the right of point X, those parts of the strong-motion record lying above ac are
integrated over time to derive a velocity profile of the block. Integration begins at point X
(Figure 3AB), and the velocity increases to point Y, the maximum velocity for this pulse.
Past point Y, the ground acceleration drops below ac, but the block continues to move
because of its inertia. Friction and ground motion in the opposite direction cause the block
to decelerate until it stops at point Z. All pulses of ground motion exceeding ac are
integrated to yield a velocity profile (Figure 3B), which, in turn, is integrated to yield a
cumulative displacement profile of the landslide block (Figure 3C).

14

Figure 3. Illustration of the Newmark algorithm, adapted from Wilson and Keefer (1983).
A, earthquake acceleration-time history with critical acceleration (dashed line) of 0.2 g
superimposed; B, velocity of the landslide versus time; C, displacement of landslide versus
time. Points X, Y, and Z are discussed in the text.
The algorithm of Wilson and Keefer (1983) permits both downslope and upslope
displacement by using the thrust angle to explicitly account for the asymmetrical resistance
to downslope and upslope sliding. If pseudostatic yield acceleration is used and the thrust
angle is not readily obtainable, the program can be simplified to prohibit upslope
displacement. This prohibition was justified by Newmark (1965), as well as others
(Franklin and Chang, 1977; Chang and others, 1984; Lin and Whitman, 1986; Ambraseys
and Menu, 1988,), because ac in the upslope direction is generally so much greater than ac
in the downslope direction that it can be assumed to be infinitely large. In most cases, the
upslope ac is greater than the PGA, and no error is introduced by prohibiting upslope
displacement. The Newmark algorithm we use in the accompanying program is based on
that of Wilson and Keefer but prohibits upslope displacement.
Integration programs for calculating Newmark displacement can be customized to
accept acceleration-time histories in either of two formats: successive pairs of time and
acceleration values (time files) or a single string of acceleration values sampled at a
constant time interval. The latter is the simpler approach and insures that the integration is
performed consistently throughout the time history. The accompanying programs are
designed for files of acceleration values at a constant time interval, but utility programs to
convert time files to acceleration files are included.

15

Digitized strong-motion records can be obtained in several ways. Analog strong-motion


records can be manually digitized to obtain a data file of time/acceleration pairs. Such a file
can be used in a Newmark integration program that accepts paired data, or it can be
resampled at a constant time interval by a simple linear interpolation program. Also, strongmotion records from many worldwide earthquakes are available in digital format from the
National Oceanic and Atmospheric Administration's National Geophysical Data Center in
Boulder, Colorado.

A SIMPLIFIED NEWMARK METHOD


The previous sections have described how to rigorously conduct a Newmark analysis.
Although this approach is straightforward, many of its aspects are difficult for many users,
and thus a simplified approach for estimating Newmark displacements would be helpful.
As discussed above, previous studies have proposed general relationships between
Newmark displacement and some normalized parameter(s) of critical acceleration (Franklin
and Chang, 1977; Makdisi and Seed, 1978; Lin and Whitman, 1986; Ambraseys and Menu,
1988; Yegian and others, 1991). Any of these that include parameters appropriate to a
problem of interest can be applied with relative ease. Most depend directly on PGA, which,
as noted, is a widely used but rather crude measure of shaking intensity. Therefore, a
simplified method based on Arias intensity, a better measure of shaking intensity, was
proposed by Jibson (1993).

Jibson's (1993) simplified Newmark method used an empirical regression equation to


estimate Newmark displacement as a function of shaking intensity and critical acceleration;
the model was calibrated by conducting Newmark analyses on 11 strong-motion records.
16

Jibson and others (1998, 2001) slightly modified the functional form of that equation to
make the critical-acceleration term logarithmic and used a much larger group of strongmotion records--280 recording stations in 13 earthquakes (Table 2)--to develop a new
regression equation. (With this larger data set, a logarithmic critical-acceleration term
yielded a much better fit than a linear term.) They analyzed both of the horizontal
components of acceleration from 275 of the recordings and a single component from the
remaining 5, which yielded 555 single-component records. For each record, they
determined the Arias (1970) intensity; then, for each record, they conducted a rigorous
Newmark analysis for several values of critical acceleration, ranging from 0.02 g to 0.40 g.
The resulting Newmark displacements were regressed on two predictor variables: critical
acceleration and Arias intensity. The resulting regression equation is
log Dn = 1.521 log Ia - 1.993 log ac -1.546 0.375,

(10)

where Dn is Newmark displacement in centimeters, Ia is Arias intensity in meters per


second, and ac is critical acceleration is g's (Jibson and others, 1998, 2001). The regression
equation is well constrained (R2 = 83%) with a very high level of statistical significance
(>99%); the model standard deviation is 0.375. This model yields the mean Newmark
displacement when the last term in the equation is ignored; the variation about this mean
results from the stochastic nature of earthquake ground shaking. Thus, two strong-motion
records having identical Arias intensities can produce different Newmark displacements for
slopes having the same critical acceleration. Therefore, Equation 10 yields a range of
displacements that must be interpreted with considerable judgment.
Thus, Newmark displacement, an index of seismic slope performance, can be estimated
as a function of critical acceleration (dynamic slope stability) and Arias intensity (groundshaking intensity). The accompanying program package includes a program to conduct a
simplified Newmark analysis using Equation 10.
Equation 10 can be applied to the example summarized in Table 1. For the lower
estimated Arias intensity of 1.59 m/s and a critical acceleration of 0.2 g, the mean value
from Equation 10 is 1.4 cm, and the range bracketing two standard deviations is 0.6-3.4 cm.
For the higher value of Arias intensity of 1.97 m/s, Equation 10 yields a mean value of 2.0
cm and a range of 0.8-4.7 cm. Displacements calculated from the three selected strongmotion records fall within this range; thus, the simplified Newmark method presented here
yields reasonable results.
Equation 10 can be applied to estimate the dynamic performance of any slope of known
critical acceleration because it is derived from generic values of critical acceleration that
are not site specific. Thus, several types of hazard analyses for earthquake-triggered
landslides can be developed:
1. If the Arias intensity at a site can be specified, and if the critical acceleration of the
slope can be determined, then the Newmark displacement can be estimated.

17

2. If critical displacement can be estimated and the critical acceleration of the slope is
known, then the threshold Arias intensity that will cause slope failure can be estimated.
3. If a critical displacement and Arias intensity can be estimated, then the threshold
critical acceleration below which slope failure will occur can be estimated.

INTERPRETING NEWMARK DISPLACEMENTS


The significance of Newmark displacements must be judged by their probable effect on
a potential landslide. Wieczorek and others (1985) used 5 cm as the critical displacement
leading to macroscopic ground cracking and general failure of landslides in San Mateo
County, Calif.; Keefer and Wilson (1989) used 10 cm as the critical displacement for
coherent landslides in southern California; and Jibson and Keefer (1993) used this 5-10 cm
range as the critical displacement for landslides in the Mississippi Valley. In most soils,
displacements in this range cause ground cracking, and previously undeformed soils can
lose some of their peak shear strength and end up in a weakened or residual-strength
condition. In such a case of strength loss, a static stability analysis in residual-strength
conditions can be performed to determine the slope stability after earthquake shaking
ceases.
Blake and others (2002) make the following recommendations for application of slidingblock analysis in southern California:

For slip surfaces intersecting stiff improvements (i.e., buildings, pools, etc.)
computed median displacements should be less than 5 cm.
For slip surfaces occurring in ductile (non-strain-softening) soil that do not intersect
engineered improvements (i.e, landscaped areas, patios), computed median
displacements should be less than 15 cm.
For slip surfaces in soils with significant strain softening (i.e., sensitivity > 2), if ac
was calculated from peak strengths, displacements as large as 15 cm could trigger
strength reductions, which in turn could destabilize the slope. For such cases, the
design should either be performed using residual strengths allowing median
displacements less than 15 cm, or using peak strengths allowing median
displacements less than 5 cm.

Any level of critical displacement can be used according to the parameters of the
problem under study and the characteristics of the landslide material. Highly ductile
materials may be able to accommodate more displacement without general failure; brittle
materials might accommodate less displacement. What constitutes "failure" may vary
according to the needs of the user. Results of laboratory shear-strength tests can be
interpreted to estimate the strain necessary to reach residual strength.
Predicted Newmark displacements do not necessarily correspond directly to measurable
slope movements in the field; rather, modeled displacements provide an index to correlate
with field performance. For the Newmark method to be useful in a predictive sense,
18

modeled displacements must be quantitatively correlated with field performance. In short,


do larger predicted displacements relate to greater incidence of slope failure? Analysis of
data from the Northridge earthquake facilitated answering this question; predicted
Newmark displacements were compared directly with an inventory of landslide actually
triggered. The results were then regressed using a Weibull model, which yielded the
following equation (Jibson and others, 1998, 2001):
P(f) = 0.335[1 - exp(-0.048 Dn1.565)],

(11)

where P(f) is the probability of failure and Dn is Newmark displacement in centimeters.


This equation can be used in any set of ground-shaking conditions to predict probability of
slope failure as a function of predicted Newmark displacement. Because this model was
calibrated using data from southern California, it may only be rigorously valid there. Jibson
and others (1998, 2001) discuss appropriate application of this probability model. The
accompanying program package has a program to estimate probability of failure based on
Equation 11.

DISCUSSION
Any idealized model is limited by its simplifying assumptions. The fundamental
assumption of Newmark's model is that landslides behave as rigid-plastic materials; i.e., no
displacement occurs below the critical acceleration, and displacement occurs at constant
shearing resistance when the critical acceleration is exceeded. This assumption is
reasonable for some types of landslides in some types of materials, but it certainly does not
apply universally. Many slope materials are at least slightly sensitive--they lose some of
their peak undrained shear strength as a function of strain. In such a case, Newmark's
method would underestimate the actual displacement, because the strength loss during
shear would reduce the critical acceleration as displacement occurs. For such materials, the
Newmark displacement might be considered a minimum displacement and so would be
unconservative.
Some highly plastic, fine-grained soils behave as visco-plastic rather than rigid-plastic
materials. The viscous response of these soils results in part from low permeability and
high cohesion, and the result can be a radically dampened seismic response. Some active,
slow moving landslides having factors of safety at or below 1.0 have experienced negligible
inertial displacement even during large earthquakes (Jibson and others, 1994) because of
viscous energy dissipation. In Newmark's method, displacement depends on the critical
acceleration, which, in turn, depends on the static factor of safety. Therefore, a landslide at
or very near static equilibrium should have a very low critical acceleration (theoretically,
ac=0 if FS=1) and thus should undergo large inertial displacements in virtually any
earthquake. Thus, Newmark's method probably overestimates landslide displacements in
visco-plastic materials.

19

Generally, Newmark's method has considered static and dynamic shear strength to be
the same and has ignored dynamic pore-pressure response; this has permitted use of static
shear strengths, which are much more easily determined than dynamic strengths. For many
soils, this assumption introduces little error, but static and dynamic strengths differ
significantly for some soils. In such cases, dynamic shear-strength testing may be required,
or static strengths can be adjusted by an empirical correction factor (Makdisi and Seed,
1978). Similarly, dynamic pore-pressure response, if considered significant, can be
measured in dynamic tests or accounted for empirically by reducing the static shear
strength.
The accompanying programs make conducting large numbers of analyses almost trivial,
and so the best approach for judging the likely performance of a slope is to select a large
number of earthquake records, perhaps 50-200, that have a reasonable range of properties
of interest and to then interpret the range of output displacements. Experience indicates that
the results tend to be log-normally distributed, with a few records yielding very high
displacements forming the right-hand tail of the distribution. Thus, mean displacements a
virtually always greater than median displacements, and standard deviations are fairly high.

CONCLUSION
Newmark's method is useful for characterizing seismic slope response. It presents a
viable compromise between simplistic pseudostatic analysis and sophisticated finiteelement modeling, and it can be applied to a variety of problems in seismic slope stability.
The simplified method presented here provides an easy way to estimate ranges of possible
displacement in cases where the seismic shaking intensity can be estimated. Probability of
failure can also be estimated in certain situations on the basis of a model calibrated using
data from the 1994 Northridge earthquake.

REFERENCES
Ambraseys, N.N., and Menu, J.M., 1988, Earthquake-induced ground displacements:
Earthquake Engineering and Structural Dynamics, v. 16, p. 985-1006.
Arias, A., 1970, A measure of earthquake intensity, in Hansen, R.J. (Editor), Seismic
design for nuclear power plants: Massachusetts Institute of Technology Press, Cambridge,
Mass., p. 438-483.
Blake, T.F., Hollingsworth, R.A., and Stewart, J.P., (eds.), 2002, Recommended Procedures
for Implementation of DMG Special Publication 117 Guidelines for Analyzing and
Mitigating Landslide Hazards in California: Los Angeles, Southern California Earthquake
Center, 127 p.

20

Bray, J.D., and Rathje, E.M., 1998, Earthquake-induced displacements of solid-waste fills:
Journal of Geotechnical and Geoenvironmental Engineering, v. 124, p. 242-253.
Chang, C.J., Chen, W.F., and Yao, J.T.P., 1984, Seismic displacements in slopes by limit
analysis: Journal of Geotechnical Engineering, v. 110, p. 860-874.
Dobry, R., Idriss, I.M., and Ng, E., 1978, Duration characteristics of horizontal components
of strong-motion earthquake records: Bulletin of the Seismological Society of America, v.
68, p. 1487-1520.
Elgamal, A.W., Abdel-Ghaffar, A.M., and Prevost, J.H., 1987, 2-D elastoplastic seismic
shear response of earth dams: Application: Journal of Geotechnical Engineering, v. 113, p.
702-719.
Franklin, A.G., and Chang, F.K., 1977, Earthquake resistance of earth and rock-fill dams:
permanent displacements of earth embankments by Newmark sliding block analysis: U.S.
Army Engineer Waterways Experiment Station, Miscellaneous Paper S-71-17, Report 5.
Goodman, R.E., and Seed, H.B., 1966, Earthquake-induced displacements in sand
embankments: Journal of the Soil Mechanics and Foundations Division, American Society
of Civil Engineers, v. 92, p. 125-146.
Harp, E.L., and Wilson, R.C., 1995, Shaking intensity thresholds for rock falls and slides:
Evidence from 1987 Whittier Narrows and Superstition Hills strong-motion records:
Seismological Society of America Bulletin, v. 85, p. 1739-1757.
Heaton, T.H., Tajima, F., and Mori, A.W., 1986, Estimating ground motions using recorded
accelerograms: Surveys in Geophysics, v. 8, p. 25-83.
Jibson, R.W., 1993, Predicting earthquake-induced landslide displacements using
Newmark's sliding block analysis: Transportation Research Board, Transportation Research
Record 1411, p. 9-17.
Jibson, R.W., 1996, Use of landslides for paleoseismic analysis: Engineering Geology, v.
43, p. 291-323.
Jibson, R.W., and Harp, E.L., 1996, The Springdale, Utah, landslide: An extraordinary
event: Environmental and Engineering Geoscience, v. 2, p. 137-150.
Jibson, R.W., Harp, E.L., and Michael, J.A., 1998, A method for producing digital
probabilistic seismic landslide hazard maps: An example from the Los Angeles, California,
area: U.S. Geological Survey Open-File Report 98-113, 17 p., 2 pl., scale 1:24,000.
Jibson, R.W., Harp, E.L., and Michael, J.A., 2001, A method for producing digital
probabilistic seismic landslide hazard maps: Engineering Geology, v. 58, nos. 3-4.

21

Jibson, R.W., and Keefer, D.K., 1993, Analysis of the seismic origin of landslides:
Examples from the New Madrid seismic zone: Geological Society of America Bulletin, v.
105, p. 521-536.
Jibson, R.W., Prentice, C.S., Langer, C.J., Rogozhin, E.A., and Borissoff, B.A., 1994,
Some observations of Landslides triggered by the 29 April 1991 Racha earthquake,
Republic of Georgia: Bulletin of the Seismological Society of America, v. 84, p. 963-973.
Keefer, D.K., 1984, Landslides caused by earthquakes: Geological Society of America
Bulletin, v. 95, p. 406-421.
Keefer, D.K., and Wilson, R.C., 1989, Predicting earthquake-induced landslides, with
emphasis on arid and semi-arid environments, in Landslides in a Semi-arid environment:
Inland Geological Society, Riverside, California, v. 2, 1989, p. 118-149.
Lin, J.S., and Whitman, R.V., 1983, Decoupling approximation to the evaluation of
earthquake-induced plastic slip in earth dams: Earthquake Engineering and Structural
Dynamics, v. 11, p. 667-678.
Lin, J.S., and Whitman, R.V., 1986, Earthquake induced displacements of sliding blocks:
Journal of Geotechnical Engineering, v. 112, p. 44-59.
Makdisi, F.I., and Seed, H.B.. 1978, Simplified procedure for estimating dam and
embankment earthquake-induced deformations: Journal of the Geotechnical Engineering
Division, American Society of Civil Engineers, v. 104, p. 849-867.
Miles, S.B., and Ho, C.L., 1999, Rigorous landslide hazard zonation using Newmark's
method and stochastic ground motion simulation: Soil Dynamics and Earthquake
Engineering, v. 18, p. 305-323.
Miles, S.B., and Keefer, D.K., 2000, Evaluation of seismic slope-performance models
using a regional case study: Environmental and Engineering Geoscience, v. 6, p. 25-39.
Newmark, N.M., 1965, Effects of earthquakes on dams and embankments: Geotechnique,
v. 15, p. 139-159.
Nuttli, O.W., and Herrmann, R.B., 1984, Ground motion of Mississippi Valley
earthquakes: Journal of Technical Topics in Civil Engineering, v. 110, p. 54-69.
Prevost, J.H., Abdel-Ghaffar, A.M., and Lacy, S.J., 1985, Nonlinear dynamic analysis of an
earth dam: Journal of Geotechnical Engineering, v. 111, p. 882-897.
Rathje, E.M., and Bray, J.D., 1999, An examination of simplified earthquake-induced
displacement procedures for earth structures: Canadian Geotechnical Journal, v. 36, p. 7287.

22

Rathje, E.M., and Bray, J.D., 2000, Nonlinear coupled seismic sliding analysis of earth
structures: Journal of Geotechnical and Geoenvironmental Engineering, v. 126, p. 10021014.
Rathje, E.M., Abrahamson, N.A., and Bray, J.D., 1998, Simplified frequency contenct
estimates of earthquake ground motions: Journal of Geotechnical Engineering, v. 124, p.
150-159.
Seed, H.B., Lee, K.L., Idriss, I.M., and Makdisi, F.I., 1975, The slides in the San Fernando
Dams during the earthquake of February 9, 1971: Journal of the Geotechnical Engineering
Division, American Society of Civil Engineers, v. 101, p. 651-688.
Seismological Society of America, 1997, Seismological Research Letters, v. 68, no. 1, 255
p.
Taniguchi, E., Whitman, R.V., and Marr, W.A., 1983, Prediction of earthquake-induced
deformation of earth dams: Soils and Foundations, v. 23, p. 126-132.
Wieczorek, G.F., Wilson, R.C., and Harp, E.L., 1985, Map Showing Slope Stability During
Earthquakes in San Mateo County, California: U.S. Geological Survey Miscellaneous
Investigations Map I-1257-E, scale 1:62,500.
Wilson, R.C., and Keefer, D.K., 1983, Dynamic analysis of a slope failure from the 6
August 1979 Coyote Lake, California, earthquake: Bulletin of the Seismological Society of
America, v. 73, p. 863-877.
Wilson, R.C., and Keefer, D.K., 1985, Predicting areal limits of earthquake-induced
landsliding, in Ziony, J.I. (Editor), Evaluating earthquake hazards in the Los Angeles
region?An earth-science perspective: U.S. Geological Survey Professional Paper 1360, p.
316-345.
Yegian, M.K., Marciano, E.A., and Ghahraman, V.G., 1991, Earthquake-induced
permanent deformations: Probabilistic approach: Journal of Geotechnical Engineering, v.
117, p. 35-50.

23

You might also like