See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/238790760

Newmark Seismic Deformation Analysis for Geosynthetic Covers

Article  in  Geosynthetics International · January 1998

DOI: 10.1680/gein.5.0120

CITATIONS READS
33 841

3 authors, including:

Neven Matasovic Jeanpierre Giroud

Geo-Logic Associates, Inc. 132 PUBLICATIONS   2,143 CITATIONS   
75 PUBLICATIONS   1,695 CITATIONS   
SEE PROFILE
SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Collaborative Research: The Integrity of Geosynthetic Elements of Waste Containment Barrier Systems Subject to large Settlements or Seismic Loading View project

NCHRP 12-114 - Guidance on Seismic Site Response Analysis with Pore Water Pressure Generation View project

All content following this page was uploaded by Neven Matasovic on 17 January 2014.

The user has requested enhancement of the downloaded file.

Technical Paper by N. Matasovic,
E. Kavazanjian, Jr. and J.P. Giroud

NEWMARK SEISMIC DEFORMATION ANALYSIS

FOR GEOSYNTHETIC COVERS

ABSTRACT: This paper investigates the impact of the following five assumptions on the
accuracy of Newmark seismic deformation analysis applied to geosynthetic cover systems:
(i) the potential failure mass is noncompliant; (ii) the dynamic response of the potential fail-
ure mass is uncoupled from displacement (slip); (iii) permanent displacements accumulate
in only one direction; (iv) vertical ground motions do not influence permanent displacement;
and (v) the yield acceleration is constant. Information presented in the literature indicates the
impact of the assumption of a noncompliant failure mass and the assumption of a seismic
response uncoupled from displacement is insignificant for typical geosynthetic cover sys-
tems. The results of computer analyses indicate that the effects of two-way sliding and verti-
cal ground motions can, in most practical cases, be neglected. However, the assumption of
a constant yield acceleration, when based on residual (or large displacement) shear strength,
may result in calculated displacements that are significantly larger than those calculated us-
ing a yield acceleration that degrades with accumulated displacement from a peak value to
a residual, or large displacement, value. Overall, results of this investigation indicate that
conventional Newmark analyses based upon residual shear strength yield conservative re-
sults when applied to geosynthetic cover systems.

KEYWORDS: Permanent displacements, Seismic deformation, Newmark method,

Composite cover, Residual strength, Two-Way sliding, Vertical acceleration.

AUTHORS: N. Matasovic, Project Engineer, E. Kavazanjian, Jr., Principal, GeoSyntec

Consultants, 2100 Main Street, Suite 150, Huntington Beach, California 92648, USA,
Telephone: 1/714-969-0800, Telefax: 1/714-969-0820; and J.P. Giroud, Senior Principal,
GeoSyntec Consultants, 621 N.W. 53rd Street, Suite 650, Boca Raton, Florida 33487, USA,
Telephone: 1/561-995-0900, Telefax: 1/561-995-0925, E-mail: nevenm@geosyntec.com,
edkavy@geosyntec.com, and jpgiroud@geosyntec.com, respectively.

PUBLICATION: Geosynthetics International is published by the Industrial Fabrics

Association International, 1801 County Road B West, Roseville, Minnesota 55113-4061,
USA, Telephone: 1/612-222-2508, Telefax: 1/612-631-9334. Geosynthetics International is
registered under ISSN 1072-6349.

DATES: Original manuscript received 6 July 1997, revised version received 21 December
1997 and accepted 9 January 1998. Discussion open until 1 September 1998.

REFERENCE: Matasovic, N., Kavazanjian, E., Jr. and Giroud, J.P., 1998, “Newmark
Seismic Deformation Analysis for Geosynthetic Covers”, Geosynthetics International, Vol.
5, Nos. 1-2, pp. 237-264.

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 237

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

1 INTRODUCTION

1.1 Classical Newmark Seismic Deformation Analysis

The most common approach to seismic deformation analysis is the rigid block sliding
on a plane approach described by Newmark (1965). In such an analysis, schematically
illustrated in Figure 1, it is assumed that a rigid block of weight W = mg, where m is
the mass of the block and g is the acceleration due to gravity, is resting on a horizontal
plane. When the plane is excited by a horizontal acceleration, a seismic inertia force,
[ah (t)]m, is induced in the block, where ah (t) is the horizontal acceleration time history
of the block. When the horizontal acceleration is expressed as a fraction of the accelera-
tion due to gravity, it is denoted by kh (t), where kh (t) = [ah (t)]/g, and is sometimes re-
ferred to as the seismic coefficient . Using the seismic coefficient, the seismic inertia
force on the block may be written as [kh (t)]mg.
At any time t, the forces at the block/plane interface are the interface normal force,
N, and the interface shear force, T. By force equilibrium, the interface shear force is
equal to the block inertia force, T = [kh (t)]mg, and the interface normal force is equal
to the weight of the block, N = W = mg. The magnitude of T is limited by the shear
strength of the interface, Tf . Using the Mohr-Coulomb failure criterion, the interface
shear strength may be expressed as Tf = aBL + N tanφ, where a is the interface adhesion,
φ is the interface friction angle, B is the width of the block/plane interface, and L is the
length of the block/plane interface.
The maximum possible block acceleration, ky , may be found by equating the block
inertia force at this acceleration to the interface shear strength: ky m =(ky /g) mg = aBL
+ N tanφ. Recognizing that mg = N, the maximum possible block acceleration (ex-
pressed as a fraction of the acceleration due to gravity) is found as ky /g = aBL / N + tanφ.
If the horizontal acceleration of the base plane remains less than or equal to ky , the block
and the plane move in unison. When the horizontal acceleration of the base plane ex-
ceeds ky , relative movement (yielding) between the block and the plane occurs. Be-
cause it represents the threshold above which relative movement (yielding) between the
block and plane occurs, ky is termed the yield acceleration.
Theoretically, when subjected to acceleration levels above ky , the block shown in
Figure 1 can move in both the positive and negative direction parallel to the plane (i.e.
x > 0 and x < 0, with the x-axis parallel to the acceleration), i.e. two-way block sliding
may occur. Because earthquake loading is a zero-mean process, if the plane is horizon-
tal and the block is allowed to move in both directions, the resulting permanent (or re-

W = mg

kh (t) W
Positive x direction
T
N
ah (t)
Figure 1. Basic elements of a classical Newmark analysis.

238 GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

sidual) displacement at the end of the earthquake shaking is generally close to zero.
Therefore, to apply the model shown in Figure 1 to the seismic deformation analysis
of a slope or an embankment, the classical Newmark analysis assumes the block can
move in only one direction, i.e. downslope. The classical Newmark analysis is then car-
ried out following the integration procedure schematically depicted in Figure 2 to cal-
culate the downslope movement. The direction of the horizontal inertia force which
causes downslope block movement is referred as the out-of-slope direction and is indi-
cated in Figure 1 as the positive x direction.
As illustrated in Figure 2, the classical Newmark procedure may be implemented by
numerical integration of the acceleration and velocity time histories of the block. First,
the velocity time history of the relative movement between the block and the plane is
calculated by integration of the acceleration time history of the plane modified by the
yield acceleration of the block, with relative block movement (sliding) beginning each
time the yield acceleration is exceeded in the out-of-slope direction and continuing un-
til zero velocity is calculated for the sliding block. The cumulative relative displace-
ment of the sliding block is then calculated by integrating the relative velocity time his-
Acceleration, ah

ky = constant
Relative velocity
Relative displacement

Time, t

Figure 2. Classical Newmark analysis integration scheme.

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 239

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

tory, as shown in Figure 2. Since relative displacement of the block only occurs between
the time the earthquake acceleration exceeds the yield acceleration in the out-of-slope
direction and the time when the relative velocity drops to zero, quiet intervals exist dur-
ing which there are no increments in the relative block displacement.
The value of relative displacement at the end of the base plane excitation is common-
ly called the (calculated) permanent seismic displacement or (calculated) seismic de-
formation. While the calculated displacements (and velocities) are properly referred to
as relative displacements (and velocities), for simplicity the attribute relative is com-
monly omitted.

1.2 Conventional Newmark Analysis for Geotechnical Problems

In conventional geotechnical practice, the potential failure mass established in a limit

equilibrium stability analysis is considered analogous to the sliding block on the plane
shown in Figure 1. The yield acceleration of the potential failure mass is established
as a function of the material and/or interface shear strength(s) and slope geometry using
a pseudo-static limit equilibrium analysis.
In a formal conventional Newmark analysis, the acceleration time history of the base
plane on which the block sits is generally evaluated from the results of seismic site re-
sponse analysis as the average acceleration time history of the potential failure mass
for which the pseudo-static yield acceleration has been calculated. Then, the seismical-
ly induced permanent displacement of the potential failure mass is calculated following
the procedure schematically depicted in Figure 2.
Less formal, classical (simplified conventional) analyses may also be performed us-
ing acceleration time histories selected from a catalog of recorded or synthesized earth-
quake ground motions and scaled to the peak horizontal ground acceleration, PHGA,
of the base plane motion. Alternatively, charts that summarize the results of previous
classical and conventional Newmark deformation analyses as a function of the yield
acceleration divided by the PHGA can be used.
The formal conventional Newmark seismic deformation analysis described above in-
volves the following five simplifying assumptions: (i) the potential failure mass is rigid
(noncompliant); (ii) the dynamic response of the failure mass is not influenced by
(coupled with) the permanent displacement (slip) that occurs along the failure surface;
(iii) permanent displacement accumulates in only one direction (the downslope direc-
tion); (iv) the vertical component of the ground motion does not influence the calcu-
lated permanent displacement; and (v) the yield acceleration of the potential failure
mass is constant.

1.3 Application to Geosynthetic Systems

Over the past six years, conventional Newmark seismic deformation analyses have
been used extensively for seismic design of geosynthetic liners and covers for landfills
and other waste containment systems. However, little direct field evidence of the accu-
racy of these conventional Newmark analyses exists. In fact, due to the many simplify-
ing assumptions inherent to the analysis, many engineers consider the permanent seis-
mic displacement calculated in a conventional Newmark analysis to be merely an index
of the seismic performance of the system rather than an engineering estimate of the an-

240 GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

ticipated seismic displacement of the geosynthetic liner or cover (Anderson and Kava-
zanjian 1995).
Experimental evidence on the accuracy of conventional Newmark analyses applied
to geosynthetic interfaces is summarized in Table 1. This evidence includes both geo-
technical centrifuge testing (Hushmand and Martin 1990; Zimmie et al. 1994; De 1996)
and shaking table testing (Kavazanjian et al. 1991; Yegian and Lahlaf 1992; Zimmie
et al. 1994; Yegian et al. 1995a,b; Yegian and Harb 1995; De 1996; Kramer and Smith
1997) under a variety of testing conditions. The results of these studies confirm that the
response of a geosynthetically lined rigid block to cyclical base excitation is in accor-
dance with the basic principles of the classical Newmark analysis. However, due to lim-
itations of the testing (adhesion can not be evaluated; testing is limited to the evaluation
of the peak friction angle), while previous studies indicate that the integration proce-
dure depicted in Figure 2 is accurate, these experiments provide little insight about the
overall impact of the simplifying assumptions made in a formal conventional Newmark
analysis on the accuracy of the computed deformations.
In the current paper, the impact of the five simplifying assumptions discussed in Sec-
tion 1.2 on the accuracy of the computed seismic deformations in a formal convention-
al Newmark analysis of geosynthetic cover systems is evaluated. The first two assump-
tions that the potential failure mass is noncompliant for displacement analysis
purposes and that the displacement is decoupled from the seismic response are dis-
cussed together (Section 2.1). The remaining three assumptions are discussed sepa-
rately (Sections 2.2 to 2.4).

2 PREVIOUS WORK

2.1 Effects of System Compliance and Decoupling

In a conventional Newmark seismic deformation analysis, the average acceleration

time history of the potential failure mass is assumed to represent the motion of the base
plane on which the sliding block in Figure 1 sits. The seismic response analysis typical-
ly used to calculate the average acceleration time history of the potential failure mass
generally assumes that there is no relative displacement (no sliding) at the base of the
potential failure mass. Therefore, it is implicitly assumed that the dynamic response of
the potential failure mass and the displacement (sliding) of this mass are two separate
process, i.e. they are decoupled. Furthermore, the use of the average acceleration time
history of the potential failure mass as the base plane motion implicitly treats the mass
as a noncompliant (i.e. rigid) mass for purposes of the displacement analysis (though
the compliance of the failure mass is considered in the seismic response analysis).
Lin and Whitman (1983) studied the seismic deformation response of earth dams to
gain insight into the decoupling and noncompliant assumptions of a conventional New-
mark analysis. Lin and Whitman (1983) modeled the seismic response of an earth dam
using a multi-degree-of-freedom system consisting of vertical stacks of lumped masses
connected by springs and viscous dashpots. A horizontal sliding element was
introduced within the model at varying locations to simulate shallow, intermediate, and
deep sliding wedges.

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 241

 Table 1. Summary of physical modeling involving geosynthetics.

242
Block Excitation
Numerical
Reference Test type Base Interface
Stiffness Type Variables simulation
inclination
Hushmand Centrifuge Stiff Horizontal Geomembrane/Geomembrane Harmonic Frequency 1.5 to 5 Hz Conventional
and Martin Geomembrane/Geotextile Transient Base acceleration 0.3 to 0.8g Newmark
(1990) Normal stress up to 8.4 kPa
Kavazanjian Shaking table Stiff Horizontal Geomembrane/Geomembrane Harmonic Frequency 1.5 to 5 Hz
et al. (1991) Geomembrane/Geotextile Base acceleration 0.3 to 0.8g ----
Normal stress up to 8.4 kPa
Yegian and Shaking table Stiff Horizontal Geomembrane/Geotextile Harmonic Frequency 2, 5, 10 Hz
Lahlaf Base acceleration up to 0.4g ----
(1992) Normal stress up to 13.6 kPa
Zimmie et Shaking table Stiff Horizontal Geomembrane/Geotextile Harmonic Frequency 5 to 40 Hz
al. (1994) Centrifuge Geonet/Geotextile Base acceleration 0.04 to 4g ----

GEOSYNTHETICS INTERNATIONAL
Normal stress 2.1 to 84 kPa
Yegian et al. Shaking table Stiff Horizontal Geomembrane/Geomembrane Harmonic Frequency 2, 5, 10 Hz
----
(1995a) Geomembrane/Geotextile Transient Base acceleration up to 0.6g
Yegian et al. Shaking table Stiff Horizontal Geosynthetics/Sand Harmonic Frequency 2, 5, 10 Hz
----
(1995b) Geomembrane/Geotextile Transient Base acceleration up to 0.6g
Yegian and Shaking table Stiff Horizontal Various geosynthetics Harmonic Frequency 1, 2, 5 Hz Yegian et al. (1991)
Harb (1995) 1V:3, 4, 6, 10H Geosynthetics Base acceleration up to 1g simplified approach

S 1998, VOL. 5, NOS. 1-2

De (1996) Shaking table Stiff Inclined Geomembrane/Geomembrane Harmonic Frequency 2.5 to 20 Hz Goodman and Seed
Centrifuge (2 and 4_) Base acceleration 0.15 to 0.22g, (1966) simplified
duration = 2 s Conventional
Normal stress up to 207 kPa Newmark
Kramer and Shaking table Compliant Inclined Geomembrane/Geotextile Harmonic Frequency is a function of Conventional
Smith Transient fundamental period of model Newmark,
(1997) Modified Newmark
MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers
MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

After subjecting the model to both simple harmonic and simulated earthquake mo-
tions, Lin and Whitman (1983) concluded that the decoupled approach generally pro-
vides a conservative estimate of permanent displacements (i.e. the permanent dis-
placements calculated assuming decoupling are generally larger than those that would
be obtained without assuming decoupling). This conservatism was the greatest at the
fundamental period of the system. Around the fundamental period of the system, Lin
and Whitman (1983) found that the magnitude of overprediction introduced by the de-
coupling assumption was negligible for shallow wedges (e.g. for cover veneer fail-
ures), approximately 20% for wedges of intermediate depth, and approached 100% for
deep wedges.
Gazetas and Uddin (1994) used two-dimensional finite element analyses with slip
elements to investigate the decoupling assumption. By comparing the permanent dis-
placements computed with slip elements with those calculated by a decoupled proce-
dure, based on the average acceleration time history (obtained by repeating the finite
element analysis without the slip elements) for hypothetical and actual rockfill dams,
Gazetas and Uddin (1994) concluded that the decoupling assumption generally does not
have a significant influence on the calculated permanent displacements. However, an
exception was noted for narrow band input motions that coincided with the fundamental
period of the dam, in which case the decoupled procedure overpredicted permanent dis-
placements by 100% or more.
Kramer and Smith (1997) developed a two degree-of-freedom analytical model to
take into account both slip at the base of the potential failure mass and mass compliance.
Kramer and Smith (1997) demonstrated the accuracy of their model in a limited number
of shaking table tests (see Table 1).
Analyses performed by Kramer and Smith (1997) indicated that, if the fundamental
period of the potential failure mass is very low with respect to the predominant period
of base excitation, as in the case of a cover soil veneer, the combined effect of neglecting
the slip at the base of the potential failure mass and mass compliance is not large and
is generally conservative. Consistent with the findings of Lin and Whitman (1983) and
Gazetas and Uddin (1994), when the fundamental period of the potential failure mass
was close to the predominant period of the base motion, displacements calculated in
a conventional Newmark analysis overpredicted the displacement calculated for the
two degree-of-freedom system by up to 100%.

2.2 Effect of Two-Way Sliding

By integrating only when the yield acceleration is exceeded on one side of the accel-
eration time history, conventional Newmark analysis ignores the cumulative effects of
block two-way sliding. The effect of ignoring two-way sliding has rarely been ad-
dressed in the technical literature. The reason for this omission lies in the fact that the
Newmark analysis is used primarily for seismic slope deformation analyses where the
horizontal acceleration needed to initiate upslope sliding (here referred to as the in-
slope yield acceleration) is so great that it is reasonable to assume that only downslope
sliding occurs. Consequently, the effect of two-way sliding has generally been ignored.
With the advent of geosynthetic cover systems with thin veneers of protective soil
cover, the effect of two-way sliding cannot be dismissed a priori. The relatively low
shear strength of many geosynthetic interfaces and, consequently, the relatively low in-

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 243

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

slope yield accelerations of a thin veneer cover suggest that the effect of two-way slid-
ing on the results of a Newmark deformation analysis warrants consideration. The ef-
fect of two-way sliding can easily be introduced into a Newmark analysis by integrating
the base plane acceleration time history modified by the out-of-slope yield acceleration
on one side and by the in-slope yield acceleration on the other side, keeping track of
the sign (direction) of the relative displacement. This approach is used in analyses de-
scribed in Section 3.4 to evaluate the impact of two-way sliding on the accuracy of con-
ventional Newmark analyses.

2.3 Effect of Vertical Acceleration Component

The effect of the vertical component of the earthquake motion on the sliding block
normal force and, consequently, on permanent displacement is ignored in a convention-
al Newmark analysis. While a vertical acceleration acting downwards on the block in
Figure 1 increases both the normal force, N, and the frictional component of the resist-
ing force, T, at the base of the block, a vertical acceleration acting upwards on the block
decreases both N and the frictional component of T.
Since the yield acceleration of the sliding mass is directly proportional to the limiting
value of T, and the limiting value of T depends upon the value of N, applying a vertical
acceleration to the block may increase or decrease the yield acceleration, depending on
the direction of the vertical acceleration. The magnitude of the change in yield accelera-
tion depends upon the relative contributions of interface adhesion and friction to the
yield acceleration and the magnitude and direction of the vertical acceleration.
The impact of the vertical acceleration on the yield acceleration was explored by Ling
and Leshchinsky (1997) in the context of the seismic stability of geosynthetic landfill
cover systems. Ling and Leshchinsky (1997) introduced a vertical pseudo-static force
into the limit equilibrium analysis of a landfill cover veneer. The vertical and horizontal
pseudo-static forces were assumed to act simultaneously in their analysis, with the hori-
zontal force acting out-of-slope and the vertical force acting upwards to produce the
maximum destabilizing effect. For vertical forces equal to 20 and 50% of the horizontal
force, Ling and Leshchinsky (1997) found the influence of the vertical force on calcu-
lated pseudo-static factors of safety practically insignificant.
The pseudo-static approach of Ling and Leshchinsky (1997) can be used to introduce
the vertical component of the ground motion into a Newmark deformation analysis. In
a conventional Newmark analysis, the yield acceleration is defined as the horizontal
acceleration which yields a pseudo-static factor of safety equal to unity. However, the
acceleration which results in a pseudo-static factor of safety equal to unity can also be
calculated either by assuming a direction for the resultant acceleration that is not hori-
zontal or by applying both horizontal and vertical accelerations simultaneously to the
potential failure mass.
Most of the computer programs commonly used in practice for limit equilibrium
slope stability analyses provide for one or both of these approaches. While this appears
to provide a straightforward manner in which to introduce the vertical component of
the ground motions into both limit equilibrium stability and Newmark deformation
analyses, it is not widely used. A major barrier to the adoption of this approach is the
lack of a rational means of determining the appropriate value of the vertical pseudo-
static force to use in the limit equilibrium analysis.

244 GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

Figure 3 shows horizontal and vertical components of a typical earthquake ground

motion. The frequency of these two ground motion components are significantly differ-
ent, and the peaks of the two components are typically out of phase. Consequently, it
is difficult, if not impossible, to establish a functional relationship between the two
components of ground motion in terms of the dynamic forces acting on the failure mass
and, therefore, between the horizontal and vertical pseudo-static forces used to calcu-
late the yield acceleration.
In fact, if the relationship between the vertical and horizontal ground motions is as-
sumed to be random, the vertical acceleration is as likely to decrease the yield accelera-
tion as it is to increase the yield acceleration, and the net impact of the vertical ground
motion on the permanent seismic deformation may be deduced to be negligible. Based
upon this logic, a vertical pseudo-static force should not be applied when calculating the
yield acceleration of a potential failure mass for use in a Newmark deformation analysis.
Yan et al. (1996a) developed a more formal approach of considering the vertical com-
ponent of the earthquake ground motions in a Newmark analysis. Yan et al. (1996a) ap-
plied the dynamic equation of motion in both the horizontal and vertical directions to
calculate the permanent seismic displacement of a rigid block sliding on a plane. In this
approach, the vertical and horizontal acceleration components of the ground motions
are applied simultaneously to the base plane. The Yan et al. (1996a) model also allows
two-way sliding, with out-of-slope and in-slope yield accelerations applied as ex-
plained in Section 2.2.
The results of a parametric study presented by Yan et al. (1996a) using actual accel-
eration time histories with the phase of the vertical and horizontal motions retained as-

(a) 0.6
Acceleration (g)

0.3

0.0

---0.3
Rio Dell - horizontal (270_)
---0.6
(b) 0.6
Acceleration (g)

0.3

0.0

---0.3
Rio Dell - vertical
---0.6
0 5 10 15 20 25 30
Time (s)
Figure 3. Rio Dell, California, accelerogram: (a) horizontal component; (b) vertical
component.

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 245

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

recorded indicate that including the vertical ground motion in the analysis can both in-
crease and decrease calculated permanent seismic displacement. An analysis of a
geosynthetic cover with an inclination of 1V:3H (1 vertical:3 horizontal) and an inter-
face friction angle of 8_ indicated that the influence of including the vertical ground
motion component in the analysis on the permanent seismic displacement was of the
order of 10%. The relative insensitivity of the calculated displacement to the vertical
motion was attributed to the random phasing and zero mean processes nature of the hor-
izontal and vertical components of the ground motion. Two-way sliding was also shown
to have little impact on the calculated displacements as the in-slope yield acceleration
was significantly higher than its out-of-slope counterpart for this example.

2.4 Degrading Yield Acceleration

In the classical paper by Newmark (1965), it was recognized that the assumption of
a constant shear strength on the block/plane interface, and therefore the implicit as-
sumption of a constant yield acceleration, might not always be appropriate. Newmark
suggested that modification of the method to include cyclic degradation of the shear
strength, i.e. degradation of the yield acceleration, might be warranted. However, this
suggestion has been largely ignored by the geotechnical and geosynthetics communi-
ties and, consequently, is not implemented in most computer programs for Newmark
analysis. Faced with the necessity of choosing a constant value of yield acceleration for
use in seismic displacement analysis of geotechnical systems, most engineers use a
yield acceleration evaluated from residual, or large deformation, shear strength param-
eters to provide a conservative basis for permanent seismic displacement assessment.
The influence of a degrading yield acceleration on the seismic deformation potential
of geosynthetic interfaces was studied by Matasovic et al. (1997). Matasovic et al.
(1997) developed a trilinear model for degradation of yield acceleration as a function
of displacement. In this model, shown in Figure 4, it was assumed that degradation of
the initial “peak” value of yield acceleration, ky1 , starts when the calculated permanent
displacement reaches an initial threshold displacement, δ1 . For a geosynthetic inter-
face, this initial threshold displacement value corresponds to the peak of the interface
shear force-displacement curve, as indicated in Figure 4a. After reaching the initial

(a) Peak parameters (b)

Large
Yield acceleration, ky

deformation or
residual
parameters ky1 δ2 − δ
Force

k y2 + (k y1 − k y2)
δ2 − δ1
ky2

δ1 δ2 δ1 δ2
Displacement, δ Displacement, δ
Figure 4. Composite landfill cover interfaces: (a) measured force-displacement
curve; (b) yield acceleration degradation model by Matasovic et al. 1997.

246 GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

threshold displacement, the yield acceleration degrades linearly with increasing inter-
face displacement until the ultimate “residual” yield acceleration, ky2 , is reached at a
second threshold displacement, δ2 . This second threshold displacement value corre-
sponds to the displacement at which the large deformation, or residual, shear strength
of the interface is reached.
Figure 5 illustrates how the degrading yield acceleration model impacts the results
of a Newmark deformation analysis. Figure 5 illustrates the difference between the re-
sults of a conventional Newmark analysis with a constant yield acceleration based upon
residual shear strength parameters and a Newmark analysis using the degrading yield
acceleration model. Analyses performed by Matasovic et al. (1997) support the logical
inference that cumulative displacement calculated with the degradation model illus-
trated in Figure 4 is consistently lower than the cumulative displacement calculated by

δ2 − δ
ky1
Acceleration, ah

Modified Newmark k y2 + (k y1 − k y2)

Conventional Newmark* δ2 − δ1
ky2
ky2

t
Relative velocity

Conventional Newmark*

Modified Newmark

t
Relative displacement

Conventional Newmark*

Modified Newmark
δ2

δ1

t1 t2 t
Figure 5. Comparison of classical and modified Newmark analysis integration
schemes.
Note: Yield acceleration evaluated using large deformation shear strength parameters (ky = ky2 ).

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 247

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

the conventional Newmark procedure with a constant yield acceleration based upon re-
sidual strength parameters.

3 NEWMARK SEISMIC DEFORMATION ANALYSIS FOR

GEOSYNTHETIC COVERS

3.1 General

A review of previous studies on the accuracy of conventional Newmark seismic de-

formation analysis indicates that the simplifying assumptions of decoupling the seismic
response of the potential failure mass from the permanent displacement of the mass and
of neglecting failure mass compliance in the deformation analysis introduce an accept-
able level of conservatism (no more than a factor of 2) into a conventional Newmark
analysis for a geosynthetic cover system. Therefore, the balance of this study focused
on the individual and combined effects of other parameters, i.e. two-way sliding, the
vertical component of the ground acceleration, and degradation of the yield accelera-
tion, on the results of Newmark seismic deformation analyses applied to geosynthetic
cover systems.
To study these effects, a computer program with provisions for these effects devel-
oped by Yan et al. (1996b) was used. This computer program incorporates the effects
of two-way sliding, formally takes into account the vertical acceleration component of
the ground motion, and employs the displacement-dependent yield acceleration model
proposed by Matasovic et al. (1997). A series of parametric studies was conducted using
this computer program to investigate the impact of these factors on the results of a New-
mark deformation analysis.

3.2 Model Parameter Development

In order to develop the parameters for the degrading yield acceleration model for the
parametric studies described herein, the results of a series of three direct shear interface
tests on a compacted soil-nonwoven geotextile-textured geomembrane-compacted soil
sandwich was used. The testing conditions were established to simulate the “typical”
geosynthetic cover system schematically shown in the inset of Figure 6.
This “typical” geosynthetic cover system consists of (top to bottom): vegetative cov-
er soil (silty sand); double-sided geocomposite drainage layer; textured, 1 mm thick
geomembrane; and compacted foundation soil. The testing was performed in accor-
dance with the ASTM D 5321 standard test method. Prior to the testing, the geomem-
brane and geotextile surfaces were wetted with water. The testing was carried out under
normal stresses, σn , of 6.9, 20.7, and 69.0 kPa, corresponding to approximate vegeta-
tive layer thicknesses of 0.4, 1.2, and 4.0 m, respectively. The displacement rate used
in the tests was 1 mm/minute. The testing results are presented in Figures 6 and 7.
Figure 6 shows the load-deformation curves from the three laboratory interface shear
tests. Figure 7 presents an interpretation of the test results of Figure 6 in terms of the
interface shear strength parameters a (adhesion) and φ (friction angle) for both peak and
large deformation (residual) conditions. The threshold displacements δ1 and δ2 , indi-

248 GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

4.5
δ1 = 20.1 mm 3.5
1
4.0
Range of δ2
3.5
Measured shear force (kN)

δ2 = 73 mm
3.0

2.5
σn = 6.9 kPa (cover thickness ≈ 0.4 m)
2.0 σn = 20.7 kPa (cover thickness ≈ 1.2 m)
δ1 = 9.7 mm σn = 69 kPa (cover thickness ≈ 4.0 m)
1.5 δ2 = 49 mm

1.0 δ1 = 5.3 mm
δ2 = 32 mm
0.5

0.0
0 20 40 60 80 100 120
Displacement, δ (mm)

Figure 6. Compacted soil-geotextile interface direct shear testing results.

Measured apparent shear stress (kPa)

80
Peak shear strength (a = 4.6 kPa, φ = 30.6 _)
70 Residual shear strength (a = 2.8 kPa, φ = 21.8 _)
60
50
40
30
20
10
0
0 10 20 30 40 50 60 70 80 90 100
Normal stress, σn (kPa)
Figure 7. Interpretation of compacted soil-geotextile interface direct shear testing
results.

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 249

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

cated in Figure 6 and used to develop the peak and large deformation failure envelopes
in Figure 7, were established by visual inspection.
Table 2 presents the static factors of safety, FSSTAT , and the corresponding peak and
large deformation yield accelerations for the three different normal stresses considered
in the analysis. Yield accelerations were calculated for a 1V:3.5H slope using the infi-
nite slope equations for veneer-type failures proposed by Matasovic (1991). Yield ac-
celerations ky1 and ky2 were calculated for the out-of-slope direction, while yield accel-
erations k iny1 and k iny2 were calculated for the in-slope direction. The data in Table 2 are
plotted in Figure 8 in the form of the yield acceleration degradation model of Matasovic
et al. (1997).

3.3 Representative Ground Motions

The parametric study presented herein was carried out using a set of five horizontal
acceleration time histories for earthquakes varying in moment magnitude, M, from 5.2
to 8.0. The five horizontal time histories used in the parametric analyses are summa-
rized in Table 3. This suite of horizontal time histories is shown in the form of accelera-
tion response spectra in Figure 9.

Table 2. Summary of the cover system parameters (1V:3.5H infinite slope).

Cover thickness Normal stress, σn Peak parameters Residual parameters

(m) (kPa) FSSTAT ky1 (g) in
k y1 (g) FSSTAT ky2 (g) k iny2 (g)
0.4 6.9 4.56 0.87 --1.91 2.94 0.50 --1.26
1.2 20.7 2.90 0.46 --1.34 1.91 0.23 --0.94
4.0 69.0 2.32 0.32 --1.14 1.55 0.14 --0.82

1.0
σn = 6.9 kPa (cover thickness ≈ 0.4 m)
Yield acceleration, ky (g)

0.8 σn = 20.7 kPa (cover thickness ≈ 1.2 m)

σn = 69 kPa (cover thickness ≈ 4.0 m)

0.6

0.4

0.2

0.0
0 20 40 60 80 100 120
Displacement, δ (mm)
Figure 8. Yield acceleration degradation model for three landfill cover thicknesses
analyzed by the modified Newmark analysis.

250 GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

Table 3. Summary of accelerograms.

Accelerogram Earthquake magnitude, M Accelerogram duration, Ds (s)

Lima, Peru 8.0 47.8
Rio Dell, California (horizontal component) 7.0 15.5
Rio Dell, California (vertical component) 7.0 19.0
Corralitos-Eureka Canyon Road, California 6.9 6.9
C4, Synthetic (Silva 1995) 6.7 6.7
Golden Gate Park, California 5.2 3.2

Notes: Magnitudes are based upon the Moment Magnitude scale. Durations are based upon the significant
duration as defined by Trifunac and Brady (1975).

2.4
Damping = 5%
Normalized acceleration (g)

2.0

1.6

1.2

0.8

Corralitos, California
0.4 Rio Dell, California (horizontal)
C4, Synthetic (Silva 1995)
Golden Gate Park, California
Lima, Peru
0.0
0.01 0.1 1.0
Period (s)
Figure 9. Acceleration response spectra of accelerograms considered in this study.

Figure 9 indicates that the selected suite covers a broad range of frequency contents
and predominant periods. Table 3 indicates that these time histories encompass a broad
range of duration, expressed in terms of significant duration, Ds , as defined by Trifunac
and Brady (1975). In addition to these horizontal time histories, the vertical component
of the Rio Dell time history from the M = 7.0 Petrolia earthquake was used to investigate
the influence of the vertical component of the ground motion on the results of the New-
mark deformation analyses.

3.4 Parametric Analyses

An initial set of Newmark deformation analyses was carried out using the two compo-
nents of the Rio Dell record, shown graphically in Figure 3, to investigate the effects
of two-way sliding and the vertical acceleration component on the computed deforma-
tion. The in- and out-of-slope yield accelerations listed in Table 2, corresponding to the
residual shear strength parameters evaluated at a normal stress of 69 kPa, were

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 251

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

employed in these analyses. The out-of-slope yield acceleration ky2 = 0.14g, corre-
sponding to 4 m of vegetative cover soil, and the horizontal component of the Rio Dell
accelerogram were used to establish benchmark seismic deformation values represent-
ing the results of a conventional Newmark analysis.
The calculations were performed for PHGA values of up to 1.1g, representing the
range of PHGA values considered likely to be encountered in practice. The effect of
two-way sliding was investigated using the out-of-slope yield acceleration of ky2 = 0.14g
and the in-slope yield acceleration of k iny2 = 0.82g presented in Table 2. The difference
between permanent displacements calculated with and without two-way sliding was,
even at the extreme PHGA level of 1.1g, negligible for all practical purposes.
Next, the effect of the vertical acceleration component on the calculated displacement
was investigated. The same Rio Dell accelerogram and the same set of yield accelera-
tions were used to establish the benchmark displacement values. To provide an upper
bound on the impact of the vertical component of the ground motions, the vertical com-
ponent of the Rio Dell accelerogram was scaled such that the peak vertical ground accel-
eration, PVGA, was equal to 1.7 times the PHGA. While the PVGA is typically assumed
to be 0.5 to 0.67 times the PHGA, the 1.7 value was chosen as the upper bound on the
ratio of PVGA to PHGA based upon examination of the most extreme strong ground mo-
tion records from the Loma Prieta, Northridge, and Petrolia California earthquakes.
In a conventional Newmark analysis, the accelerogram is integrated in two different
trials, once assuming sliding in the direction of the top half of the accelerogram and
once assuming sliding in the direction of the bottom half of the accelerogram. The larg-
est permanent displacement calculated in the two trials is considered the calculated dis-
placement. If the vertical component of the ground motion is included in the analysis,
there are four possible combinations of vertical and horizontal time histories and, there-
fore, four trials are required to exhaust all possibilities.
The results of the analysis in which the vertical component of the ground motion is
considered are summarized in Figure 10. Figure 10 indicates that the computed dis-
placement can be both increased and decreased by including the vertical motion in the
analysis. Consistent with conventional analysis, the results of the trial that produces the
largest calculated displacement is considered the calculated displacement. Even with
this assumption, Figure 10 indicates that the effect of the vertical acceleration compo-
nent on the permanent displacement calculated in a Newmark analysis is relatively
small and can be neglected for practical purposes, which is consistent with the previous
studies discussed in Section 2.3.
Based on the review of previous studies (Section 2.4), the effect of degrading yield
acceleration appears to be the most important effect for geosynthetic covers. Therefore,
the effect of a degrading yield acceleration was investigated using all five time histo-
ries. Conventional Newmark deformation analyses using a constant yield acceleration
were carried out for PHGA values of up to 1.1g using constant yield accelerations based
upon the peak and residual shear strength parameters and using the degrading yield ac-
celeration model. The analyses were carried out for the three cover configurations listed
in Table 2. Results of these analyses are presented in Figures 11 to 15.
Figures 11a, 12a, 13a, 14a, and 15a, presenting the results for the 0.4 m thick vegeta-
tive cover soil layer (σn = 6.9 kPa), indicate that the seismic displacement response of
the geosynthetic cover system calculated using the Newmark analysis with a degrading
yield acceleration (modified Newmark analysis) is the same as for a conventional New-

252 GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

0.30

Calculated permanent displacement (m)

Conventional Newmark analysis
Newmark analysis
0.25 with vertical
component
(PVGA = 1.7 PHGA)
0.20

0.15

0.10

0.05 σn = 69 kPa
Residual parameters
Rio Dell accelerogram
0.00
0.0 0.2 0.4 0.6 0.8 1.0
Peak horizontal ground acceleration (g)
Figure 10. Influence of the vertical acceleration component on the results of a
conventional Newmark analysis.

mark analysis based on peak shear strength parameters. This relative insensitivity of the
calculated displacement to degradation of yield acceleration is readily explained by the
fact that, due to the low normal stress, the initial (peak) value of yield acceleration is
relatively high (ky1 = 0.87g).
In Figures 11b, 12b, 13b, 14b, and 15b, presenting the results for the intermediate nor-
mal stress level (σn = 20.7 kPa) corresponding to 1.2 m of vegetative cover soil, the ef-
fect of a degrading yield acceleration on the calculated displacement becomes appa-
rent. Figures 11b, 12b, 13b, 14b, and 15b indicate that the bifurcation between the
results of a Newmark analysis based upon the peak strength and a Newmark analysis
using the degrading yield acceleration model starts when the permanent displacement
value calculated using the modified Newmark analysis reaches the initial threshold dis-
placement δ1 = 9.7 mm. Beyond the initial threshold displacement, the residual shear
stress parameters rapidly start to influence the displacement response in the degrading
yield acceleration case.
Figures 11b, 12b, 13b, 14b, and 15b also show that, for all five accelerograms, a rela-
tively high PHGA value (0.7g or more) is required to induce a displacement response
that converges on the displacement calculated using a constant yield acceleration based
upon the residual shear strength parameters. The same trend in calculated displacement
can be seen for the higher normal stress of 69 kPa, corresponding to 4 m of vegetative
cover soil, presented in Figures 11c, 12c, 13c, 14c, and 15c.
Figures 11 through 15 demonstrate that a conventional Newmark analysis based sole-
ly on the residual shear strength parameters always produces conservative results (i.e.
greater calculated displacements) compared to the degrading yield acceleration model
(modified Newmark analysis).

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 253

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

(a) 0.30 Classical Newmark analysis (peak parameters)

Classical Newmark analysis (residual parameters)
0.25 Modified Newmark analysis

0.20

0.15
Note: The classical (peak parameters)
0.10 and modified Newmark analysis give the
same results in this case.

0.05
δ1 = 5.3 mm
Calculated permanent displacement (m)

0.00
(b) 0.30

0.25

0.20

0.15

0.10

0.05
δ1 = 9.7 mm
0.00
(c) 0.30

0.25

0.20

0.15

0.10

0.05
δ1 = 20.1 mm

0.00
0.0 0.2 0.4 0.6 0.8 1.0
Peak horizontal ground acceleration (g)

Figure 11. Lima, Peru, accelerogram - comparison of classical and modified

Newmark analysis at different normal stress and peak horizontal ground acceleration
levels: (a) σn = 6.9 kPa; (b) σn = 20.7 kPa; (c) σn = 69.0 kPa.

254 GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

(a) 0.30 Classical Newmark analysis (peak parameters)

Classical Newmark analysis (residual parameters)
0.25 Modified Newmark analysis

0.20

0.15
Note: The classical (peak parameters)
0.10 and modified Newmark analysis give the
same results in this case.

0.05
δ1 = 5.3 mm
Calculated permanent displacement (m)

0.00
(b) 0.30

0.25

0.20

0.15

0.10

0.05
δ1 = 9.7 mm
0.00
(c) 0.30

0.25

0.20

0.15

0.10

0.05
δ1 = 20.1 mm

0.00
0.0 0.2 0.4 0.6 0.8 1.0
Peak horizontal ground acceleration (g)

Figure 12. Rio Dell, California, accelerogram - comparison of classical and modified
Newmark analysis at different normal stress and peak horizontal ground acceleration
levels: (a) σn = 6.9 kPa; (b) σn = 20.7 kPa; (c) σn = 69.0 kPa.

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 255

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

(a) 0.30 Classical Newmark analysis (peak parameters)

Classical Newmark analysis (residual parameters)
0.25 Modified Newmark analysis

0.20

0.15
Note: The classical (peak parameters)
0.10 and modified Newmark analysis give the
same results in this case.

0.05
δ1 = 5.3 mm
0.00
Calculated permanent displacement (m)

(b) 0.30

0.25

0.20

0.15

0.10

0.05
δ1 = 9.7 mm
0.00
(c) 0.30

0.25

0.20

0.15

0.10

0.05
δ1 = 20.1 mm

0.00
0.0 0.2 0.4 0.6 0.8 1.0
Peak horizontal ground acceleration (g)
Figure 13. Corralitos-Eureka Canyon Road, California, accelerogram - comparison of
classical and modified Newmark analysis at different normal stress and peak horizontal
ground acceleration levels: (a) σn = 6.9 kPa; (b) σn = 20.7 kPa; (c) σn = 69.0 kPa.

256 GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

(a) 0.30 Classical Newmark analysis (peak parameters)

Classical Newmark analysis (residual parameters)
0.25 Modified Newmark analysis

0.20

0.15
Note: The classical (peak parameters)
0.10 and modified Newmark analysis give the
same results in this case.

0.05
δ1 = 5.3 mm
0.00
Calculated permanent displacement (m)

(b) 0.30

0.25

0.20

0.15

0.10

0.05
δ1 = 9.7 mm
0.00
(c) 0.30

0.25

0.20

0.15

0.10

0.05
δ1 = 20.1 mm

0.00
0.0 0.2 0.4 0.6 0.8 1.0
Peak horizontal ground acceleration (g)

Figure 14. C4-synthetic accelerogram - comparison of classical and modified

Newmark analysis at different normal stress and peak horizontal ground acceleration
levels: (a) σn = 6.9 kPa; (b) σn = 20.7 kPa; (c) σn = 69.0 kPa.

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 257

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

(a) 0.30 Classical Newmark analysis (peak parameters)

Classical Newmark analysis (residual parameters)
0.25 Modified Newmark analysis

0.20

0.15
Note: The classical (peak parameters)
0.10 and modified Newmark analysis give the
same results in this case.

0.05
δ1 = 5.3 mm
0.00
Calculated permanent displacement (m)

(b) 0.30

0.25

0.20

0.15

0.10

0.05
δ1 = 9.7 mm
0.00
(c) 0.30

0.25

0.20

0.15

0.10

0.05
δ1 = 20.1 mm

0.00
0.0 0.2 0.4 0.6 0.8 1.0
Peak horizontal ground acceleration (g)
Figure 15. Golden Gate Park, California, accelerogram - comparison of classical and
modified Newmark analysis at different normal stress and peak horizontal ground
acceleration levels: (a) σn = 6.9 kPa; (b) σn = 20.7 kPa; (c) σn = 69.0 kPa.

258 GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

0.30

Calculated permanent displacement (m)

Conventional Newmark analysis
Modified Newmark analysis (MNA)
MNA with two-way sliding
0.25 MNA with vertical
component
(PVGA = 1.7 PHGA)

0.20

0.15

0.10

0.05
σn = 69 kPa
Residual parameters
Rio Dell accelerogram
0.00
0.0 0.2 0.4 0.6 0.8 1.0
Peak horizontal ground acceleration (g)
Figure 16. Influence of the second threshold displacement, δ2 , on the results of a
modified Newmark analysis.

The combined effects of a degrading yield acceleration, two-way sliding, and the ver-
tical acceleration component were also investigated using the Rio Dell accelerogram
and the yield accelerations corresponding to the higher normal stress of 69 kPa (repre-
senting 4 m of vegetative cover soil). The results of these analyses, presented in Fig-
ure 16, further substantiate the conclusion that the influences of the two-way sliding and
vertical acceleration component on calculated permanent displacements may be ne-
glected for practical purposes.
An additional set of parametric analyses was performed to investigate the sensitivity
of the degrading yield acceleration analysis to the threshold displacements δ1 and δ2 .
The selection of these parameters from the test results shown in Figure 6 was done sub-
jectively, as mentioned in Section 3.2. The selection of δ2 for the highest stress level
in Figure 6 is particularly subject to different interpretation. In order to investigate the
influence of δ2 on the results of the modified Newmark analysis, the analysis for the
normal stress level of 69 kPa (4 m of vegetative cover soil) was repeated for a reference
PHGA value of 0.61g with δ2 varied from 20.1 mm (equal to δ1 , corresponding to an
instantaneous drop from the peak strength to the residual strength) to 93 mm (the limit
of the testing apparatus). Results of this parametric study are shown in Figure 17.
Figure 17 indicates that, at the selected reference PHGA level of 0.61g, the results
of the displacement calculations are strongly dependent on the cumulative deformation
potential of the accelerogram employed in the analyses. Based upon the data in Fig-
ure 17 and in Table 3, the cumulative displacement appears to be related to the signifi-
cant duration and magnitude of the earthquake record. The discrepancy between the
displacements calculated for two extreme estimates of δ2 is large. For the motion with

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 259

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

Calculated permanent displacement (m)

0.25
Lima (M = 8, Ds = 47.8 s)
Rio Dell (M = 7, Ds = 15.5 s)
0.20 Corralitos (M = 6.9, Ds = 6.9 s)
C4, synthetic (M = 6.7, Ds = 6.7 s)
Golden Gate Park (M = 5.2, Ds = 3.2 s)
0.15
σn = 69 kPa
PHGA = constant = 0.61g
0.10

0.05

0.00
0 20 40 60 80 100 120
Second threshold displacement, δ2 (mm)
Figure 17. Influence of the vertical acceleration component on the results of a
modified Newmark analysis.

the largest significant duration and magnitude (the Lima, Peru, record), the calculated
displacement for the maximum value of δ2 is 0.04 m while the calculated displacement
of the minimum value of δ2 is over five times as great, 0.21 m. For the motion with the
smallest significant duration and magnitude (the Golden Gate Park, California record),
the discrepancy between the calculated displacement at the extreme values of δ2 is ne-
gligible. Figure 17 further indicates that, in the vicinity of the δ2 value of 73 mm that
was estimated subjectively, the calculated permanent displacement is relatively stable
for all five accelerograms.
Figure 17 also illustrates the importance of selecting accelerograms of appropriate
magnitude and/or duration for use in a Newmark deformation analysis. In several of the
chart solutions for Newmark displacement analyses used in current practice (e.g. Hynes
and Franklin 1984), the influence of earthquake magnitude and/or duration is ignored.
Ignoring earthquake magnitude and/or duration can lead to an unrealistic assessment
of seismic deformation potential. This is particularly true for small magnitude events
if the “upper bound” displacement curve from a deformation chart which ignores earth-
quake magnitude is employed. As indicated in Figure 17, in such a case the result may
be excessively conservative (i.e. the calculated displacement value may be too large).

4 CONCLUSIONS

Conventional Newmark seismic deformation analyses are based on the following five
simplifying assumptions: (i) the potential failure mass is rigid (noncompliant); (ii) the
dynamic response of the failure mass is not influenced by (coupled with) the permanent
displacement (slip) that occurs along the failure surface; (iii) permanent displacement
accumulates in only one direction (the downslope direction); (iv) the vertical compo-

260 GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

nent of the ground motion does not influence the calculated permanent displacement;
and (v) the yield acceleration of the potential failure mass is constant.
Review of relevant previous studies indicates that the assumptions of decoupled seis-
mic response and displacement and of a noncompliant failure mass are of relatively mi-
nor significance for assessing the deformation potential of geosynthetic covers for solid
waste landfills and other facilities, resulting in overprediction of the permanent dis-
placement by at most a factor of 2. The analyses presented herein also indicate that the
effects of two-way sliding and of the vertical component of the earthquake ground mo-
tions are, for most practical purposes, negligible for geosynthetic cover systems. How-
ever, for strain softening materials such as the interfaces of typical geosynthetic cover
systems, the effect of degradation of the yield acceleration from an initial peak value
to an ultimate residual value may be an important factor impacting the accuracy of con-
ventional Newmark analyses.
The analyses described herein for three representative geosynthetic cover configura-
tions using the yield acceleration degradation model proposed by Matasovic et al.
(1997) demonstrate that conventional Newmark deformation analyses which use a
constant yield acceleration based solely on residual shear strength parameters may be
excessively conservative, i.e. may lead to a calculated permanent displacement much
greater than the value calculated taking into account degrading yield acceleration. The
degree of conservatism depends to a large extent upon the value of the calculated seis-
mic deformation compared to the threshold deformations at which the peak and residual
strengths are mobilized. The analyses presented herein further demonstrate that selec-
tion of an acceleration time history of appropriate magnitude and/or duration is a key
factor in performing a seismic displacement analysis.
In deciding on whether to base a geosynthetic cover design on an analysis using a
constant yield acceleration or on an analysis using a degrading yield acceleration, a va-
riety of other factors in addition to the value of the calculated permanent seismic dis-
placement should be considered. Factors such as creep, the cyclic nature of earthquake
loading, and transient seismic displacements may accelerate the degradation of the in-
terface shear strength to the value corresponding to the residual shear strength parame-
ters. Therefore, evaluation of the appropriate interface shear strength values for use in
permanent seismic deformation analyses still requires more research.
Given the interface testing state of practice, two boundaries may be calculated for the
permanent displacement: (i) an upper boundary using a constant yield acceleration
based on residual strength parameters; and (ii) a lower boundary using a degrading
yield acceleration as indicated in the current paper.

ACKNOWLEDGMENTS

The authors wish to thank Mr. L. Yan of the University of Southern California, Los An-
geles, USA, who helped in the initial stages of this project and provided valuable sugges-
tions. The authors also wish to thank Dr. H. Sharma of GeoSyntec Consultants and an
anonymous reviewer of the manuscript who provided many valuable suggestions.

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 261

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

REFERENCES

Anderson, D.G. and Kavazanjian, E., Jr., 1995, “Performance of Landfills under Seis-
mic Loading”, Proceedings of the Third International Conference on Recent Ad-
vances in Geotechnical Earthquake Engineering and Soil Dynamics, University of
Missouri-Rolla, Vol. 3, St. Louis, Missouri, USA, April 1995, pp. 277-306.
ASTM D 5321, “Standard Test Method for Determining the Coefficient of Soil and Geo-
synthetic or Geosynthetic and Geosynthetic Friction by the Direct Shear Method”,
American Society for Testing and Materials, West Conshohocken, Pennsylvania,
USA.
De, A., 1996, “Study of Interface Friction of Landfill Geosynthetics: Static and Dynam-
ic”, Ph.D. Dissertation, Department of Civil Engineering, Rensselaer Polytechnic In-
stitute, Troy, New York, USA, 245 p.
Gazetas, G. and Uddin, N., 1994, “Permanent Deformation on Pre-Existing Sliding
Surfaces in Dams”, Journal of Geotechnical Engineering, Vol. 120, No. 11, pp.
2041-2061.
Goodman, R.E. and Seed, H.B., 1966, “Earthquake-Induced Displacements in Sand
Embankments”, Journal of Geotechnical Engineering, Vol. 90, No. SM6, pp.
125-146.
Hushmand, B. and Martin, G.R., 1990, “Layered Soil-Synthetic Linear Base Isolation
System”, SBIR Phase I Study Final Report, The Earth Technology Corporation, Long
Beach, California, USA, 86 p. (plus appendices)
Hynes, M.E. and Franklin, A.G., 1984, “Rationalizing the Seismic Coefficient Meth-
od”, Miscellaneous Paper GL-84-13, U.S. Army Engineer Waterways Experiment
Station, Vicksburg, Mississippi, USA, 34 p.
Kavazanjian, E., Jr., Hushmand, B. and Martin, G.R., 1991, “Frictional Base Isolation
Using a Layered Soil-Synthetic Liner System”, Proceedings of the Third U.S. Confer-
ence on Lifeline Earthquake Engineering, Cassaro, A.M., Editor, ASCE Technical
Council on Lifeline Earthquake Engineering Monograph No. 4, Los Angeles, Cali-
fornia, USA, August 1991, pp. 1139-1151.
Kramer, S.L. and Smith, M.W., 1997, “Modified Newmark Model for Seismic Dis-
placements of Compliant Slopes”, Journal of Geotechnical and Geoenvironmental
Engineering, Vol. 123, No. 7, pp. 635-644.
Lin, J.S. and Whitman, R.V., 1983, “Decoupling Approximation to the Evaluation of
Earthquake-Induced Plastic Slip in Earth Dams”, Earthquake Engineering and Struc-
tural Dynamics, Vol. 11, No. 5, pp. 667-678.
Ling, H.I. and Leshchinsky, D., 1997, “Seismic Stability and Permanent Displacement
of Landfill Cover Systems”, Journal of Geotechnical and Geoenvironmental Engi-
neering, Vol. 123, No. 2, pp. 113-122.
Matasovic, N., 1991, “Selection of Method for Seismic Slope Stability Analysis”, Pro-
ceedings of the Second International Conference on Recent Advances in Geotechni-
cal Earthquake Engineering and Soil Dynamics, University of Missouri-Rolla, Vol.
2, St. Louis, Missouri, USA, April 1991, pp. 1057-1062.

262 GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2

MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

Matasovic, N., Kavazanjian, E., Jr. and Yan, L., 1997, “Newmark Deformation Analy-
sis with Degrading Yield Acceleration”, Proceedings of Geosynthetics ’97, IFAI, Vol.
2, Long Beach, California, USA, March 1997, pp. 989-1000.
Newmark, N.M., 1965, “Effects of Earthquakes on Dams and Embankments”, Géo-
technique, Vol. 15, No. 2, pp. 139-160.
Silva, W., 1995, “Simulation of Strong Ground Motions Generated by Northridge
Earthquake at Southern California Landfills”, Synthetic Accelerogram C4, (digital
record).
Trifunac, M.D. and Brady, A.G., 1975, “A Study of the Duration of Strong Earthquake
Ground Motion”, Bulletin of the Seismological Society of America, Vol. 65, No. 1,
pp. 581-626.
Yan, L., Matasovic, N. and Kavazanjian, E., Jr., 1996a, “Seismic Response of Rigid
Block on Inclined Plane to Vertical and Horizontal Ground Motions Acting Simulta-
neously”, Proceedings of the Eleventh ASCE Engineering Mechanics Conference,
Vol. 2, Fort Lauderdale, Florida, USA, pp. 1110-1113.
Yan, L., Matasovic, N. and Kavazanjian, E., Jr., 1996b, “YSLIP_PM - A Computer Pro-
gram for Simulation of Dynamic Behavior of a Rigid Block on an Inclined Plane and
Calculation of Permanent Displacements of the Block”, User’s Manual, GeoSyntec
Consultants, Huntington Beach, California, USA, 21 p. (plus appendix)
Yegian, M.K. and Harb, J.N., 1995, “Slip Displacements of Geosynthetic Systems un-
der Dynamic Excitation”, Earthquake Design and Performance of Solid Waste Land-
fills, Yegian, M.K. and Finn, W.D.L., Editors, Geotechnical Special Publication No.
54, ASCE, proceedings of a specialty conference held in San Diego, California, USA,
October 1995, pp. 212-236.
Yegian, M.K. and Lahlaf, A.M., 1992, “Dynamic Interface Shear Strength Properties
of Geomembranes and Geotextiles”, Journal of Geotechnical Engineering, Vol. 118,
No. 5, pp. 760-779.
Yegian, M.K., Marciano, E.A. and Ghahraman, V.G., 1991, “Earthquake-Induced Per-
manent Deformations-Probabilistic Approach”, Journal of Geotechnical Engineer-
ing, Vol. 117, No. 1, pp. 35-50.
Yegian, M.K., Yee, Z.Y. and Harb, J.N., 1995a, “Response of Geosynthetics under
Earthquake Excitation”, Proceedings of Geosynthetics ’95, IFAI, Vol. 2, Nashville,
Tennessee, USA, February 1995, pp. 677-689.
Yegian, M.K., Yee, Z.Y. and Harb, J.N., 1995b, “Seismic Response of Geosynthetic/
Soil Systems”, Geoenvironment 2000, Acar, Y.B. and Daniel, D.E., Editors, Geotech-
nical Special Publication No. 46, ASCE, Vol. 2, proceedings of a specialty conference
held in New Orleans, Louisiana, USA, February 1995, pp. 1113-1125.
Zimmie, T.F., De, A. and Mahmud, M.B., 1994, “Centrifuge Modeling to Study Dy-
namic Friction at Geosynthetic Interfaces”, Proceedings of the Fifth International
Conference on Geotextiles, Geomembranes and Related Products, Vol. 1, Singapore,
September 1994, pp. 415-418.

GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2 263

 MATASOVIC, KAVAZANJIAN & GIROUD D Newmark Seismic Analysis for Geosynthetic Covers

NOTATIONS

Basic SI units are given in parentheses.

a = interface adhesion (Pa)

ah (t) = horizontal acceleration time history of the block (m/s2)
B = width of the block/plane interface (m)
Ds = significant duration of strong shaking (s)
FSSTAT = static factors of safety (dimensionless)
g = acceleration due to gravity (m/s2)
kh (t) = seismic coefficient (dimensionless)
ky = yield acceleration (m/s2)
ky1 = yield acceleration calculated on the basis of peak shear strength
parameters (m/s2)
k iny1 = ky1 needed to initiate upslope sliding (m/s2)
ky2 = yield acceleration calculated on the basis of residual shear strength
parameters (m/s2)
k iny2 = ky2 needed to initiate upslope sliding (m/s2)
L = length of the block/plane interface (m)
M = moment magnitude (dimensionless)
m = mass of the block (kg)
N = normal force (N)
PHGA = peak horizontal ground acceleration (m/s2)
PVGA = peak vertical ground acceleration (m/s2)
T = interface shear force (N)
t = time (s)
Tf = shear strength of the interface (N)
W = weight of the block (N)
δ = displacement (m)
δ1 = initial threshold displacement (m)
δ2 = second threshold displacement (m)
φ = interface friction angle (_)
σn = normal stress (Pa)

264 GEOSYNTHETICS INTERNATIONAL S 1998, VOL. 5, NOS. 1-2

View publication stats