DYNAMICS OF STRUCTURES ON TWO-SPRING

FOUNDATION ALLOWED TO UPLIFT

By Solomon C.-S. Yim, 1 A. M. ASCE and Anil K. Chopra, 2 M. ASCE

ABSTRACT: Investigated in this paper is the dynamics of structures with their

foundation mat supported only through gravity and thus permitted to uplift
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

from the supporting system. In its fixed base condition the structure is idealized
as a single-degree-of-freedom system attached to a rigid foundation mat, which
is supported at each edge by a spring-damper element. Analytical expressions
are presented for the free vibration response of the system and the effects of
foundation uplift are examined. An effective numerical procedure, based on
expressions of the Rayleigh-Ritz concept, to evaluate the structural response to
earthquakes is presented. Based on the response spectra presented it is shown
that foundation mat uplift has the effect of reducing the base shear for short-
period structures, with these reductions being especially significant for the more
slender structures.

INTRODUCTION

Standard earthquake analysis of buildings under the assumption that

the foundation and soil are firmly bonded, often produce a base over-
turning moment that exceeds the available overturning resistance due
to gravity loads, which implies that a portion of the foundation would
intermittently uplift during an earthquake. By investigating the dynam-
ics of rocking rigid blocks, Housner was the first to recognize the cor-
relation between foundation uplift and the good performance of seem-
ingly unstable structures during earthquakes (3). Idealizing the structure
as a one-degree-of-freedom oscillator, Meek analytically investigated the
effects of foundation uplift on the earthquake response of a flexible
structure (5). Meek concluded that foundation uplift leads to reduction
in the structural deformation. Later, Meek extended his study to mul-
tistory, braced-core buildings (6), and found that significant reduction
in structural deformation can again be obtained by permitting uplift of
the rigid base of the core. Experiments on steel building frames with
columns permitted to lift-off during vibration, conducted on the Berke-
ley shaking table (2,4), demonstrated that allowing column uplift under
an extreme earthquake would enhance the chances of a structure sur-
viving in a functional condition, with damage held to a minimum. The
flexibility of the supporting soil was ignored in these analytical studies
(5,6); it was not incorporated in the shaking table tests either (2,4).
Recently, Psycharis (7) studied the effects of foundation uplift on
structural response using two simple models to represent foundation
flexibility and energy dissipation: (1) Two spring-damper elements sym-
metrically placed under the base of the structure; and (2) Winkler foun-
dation with spring-damper elements distributed over the base of the
2
'Grad. Student, Dept. of Civ. Engrg., Univ. of California, Berkeley, Calif.
Prof., Dept of Civ. Engrg., Univ. of California, Berkeley, Calif.
Note.—Discussion open until December 1, 1984. To extend the closing date
one month, a written request must be filed with the ASCE Manager of Technical
and Professional Publications. The manuscript for this paper was submitted for
review and possible publication on August 23, 1983. This paper is part of the
Journal of Engineering Mechanics, Vol. 110, No. 7, July, 1984. ©ASCE, ISSN 0733-
9399/84/0007-1124/$01.00. Paper No. 18986.
1124

J. Eng. Mech. 1984.110:1124-1146.

 structure. The analytical parts of his study included several aspects of
the dynamics of rigid blocks as well as flexible structures. Also he nu-
merically examined the effects of base uplift on the dynamic response
of one example single-story structure and one multistory superstructure.
Based on these results, he concluded that "for flexible superstructures,
it seems that uplift always increases the angle of rotation of the foun-
dation mat but the effects on the structural deflection and the resulting
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

stresses are not clear. In general, it cannot be concluded whether uplift

is beneficial to the structure or not, since this depends on the parameters
of the system and the characteristics of the excitation."
The objective of this paper is to develop a better understanding of the
effects of transient foundation uplift on building response, considering
a wide range of the important system parameters. For this purpose it is
appropriate to choose structural idealizations that are relatively simple
but realistic in the sense that they incorporate only the most important
features of foundation uplift. In its fixed base condition, the structure is
idealized as a single-degree-of-freedom system attached to a rigid foun-
dation mat which is flexibly supported. The flexibility and damping of
the supporting soil is represented by two spring-damper (in parallel)
elements, one at each edge of the foundation mat.

SYSTEM CONSIDERED

It is desirable to begin with the simplest possible structural system,

so for this purpose we consider the idealized representation of a one-
story structure shown in Fig. 1(a). It is a linear structure of mass m,
lateral stiffness k and lateral damping c, which is supported through the
foundation mat of mass m0 resting on two spring-damper elements, one
at each edge of the foundation mat, connected to the base which is as-

— — BONDED CONTACT,

U P U F T PERMITTED

M /
Mc = pb
//
/]2kfbe
-0< / ' 1
»c.p/2k,b 8 < < i

/ -pb

//
l a ) SYSTEM CONSIDERED

FIG. 1.—Flexible Structure on Two Spring-Damper Element Foundation and Its

Properties
1125

J. Eng. Mech. 1984.110:1124-1146.

 sumed to be rigid. The columns are assumed to be massless and axially
inextensible, the foundation mat is idealized as a rigid rectangular plate
of negligible thickness with uniformly distributed mass, and it is pre-
sumed that horizontal slippage between the mat and supporting ele-
ments is not possible. The stiffness and damping coefficients of the
foundation model are assumed constant, independent of displacement
amplitude or excitation frequency. Thus, the frequency dependence of
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

these coefficients, as for a viscoelastic half space (9), is not recognized;

nor is the strain dependence of these coefficients for soils (8) considered
in this study.
In previous studies, the impact between the foundation mat of a flex-
ible structure (5), or the base of a rocking rigid block (3), and the rigid
supporting soil was idealized as a perfectly inelastic collision with the
vertical velocity and associated kinetic energy of the structural mass dis-
sipated instantaneously. However, in the system idealization selected
here, the energy is gradually dissipated through the viscous damping
mechanism of the foundation. But if the foundation springs are rela-
tively stiff, the typical case for buildings and soils, the energy dissipation
occurs within a very short duration. Thus, the two types of energy dis-
sipating mechanisms would be expected to similarly affect structural re-
sponse. Psycharis also concluded that the impact effect can be modeled
well by the simpler viscous damping mechanism (7).
Prior to the dynamic excitation, the foundation mat rests on the spring-
damper elements only through gravity and is not bonded to these sup-
porting elements. Thus, a supporting element can provide an upward
reaction force to the foundation mat but not a downward pull. During
vibration of the system this upward reaction force will vary with time.
At any instant of time the total reaction force—which for a supporting
element consisting of a spring and damper in parallel is the sum of the
elastic and damping forces—reaches zero, one edge of foundation mat
is in the condition of incipient uplift from the supporting element.
In particular, if the damper is absent, the relation between the reaction
force and displacement of an edge i (left or right) of the foundation mat
is shown in Fig. 1(b). The upward reaction force is related linearly to the
downward displacement of the foundation mat through the spring stiff-
ness kf, but no reaction force is developed if the displacement is up-
ward; the displacements are measured from the unstressed position of
the springs. If the foundation mat were bonded to the supporting ele-
ments the force-displacement relation will be linear, as shown in Fig.
1(b), valid for downward as well as upward displacements.
Consider the foundation mat and its supporting elements without the
superstructure with a static force p acting in the downward direction at
its center of gravity (eg). The relation between the static moment M
applied at the e g and the resulting foundation-mat rotation 6, limited
to angles much smaller than the slenderness ratio b/h, is shown in Fig.
1(c) for unbonded as well as bonded conditions. If the mat is not bonded
to the supporting elements the M-0 relation is linear until the foundation
mat uplifts from one of the supporting elements; thereafter no additional
moment can be developed. Uplift occurs when the rotation reaches 0C =
p/2kfb with the corresponding moment Mc - pb. The downward force
is p = (m + m0)g, the combined weight of the superstructure and foun-
1126

J. Eng. Mech. 1984.110:1124-1146.

 dation mat, prior to any dynamic excitation, but would vary with time
during vibration.
Next consider the entire structural system with a gradually increasing
force fs applied in the lateral direction. If the foundation mat is bonded
to the supporting elements, which along with the structure have linear
properties, the lateral force can increase beyond limit if the overturning
effects of gravity forces are neglected. However, if the mat is not bonded
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

to the supporting elements, one edge of the foundation mat is at incip-

ient uplift when the lateral force reaches fsc = (m + m0)gb/h. Thus, the
maximum base shear that can be developed under the action of static
forces is
b
Vc = (m + m0)g - (la)
n
The structural deformation associated with this base shear is
(m + m0)g b
u = {lb)
< ~ir—h
and at incipient uplift, the foundation-mat rotation
(m + m0)g
9c = — — — lc
2kfb
which is consistent with the preceding paragraph and Fig. 1(c).
The base excitation is specified by the horizontal ground motion with
displacement ug(t) and acceleration ug(t). The vertical component of
ground motion is not considered in this study. Under the influence of
this excitation the foundation mat would rotate through an angle 6(f)
and undergo a vertical displacement v(t), defined at its e g relative to
the unstressed position. Prior to dynamic excitation, the vertical dis-
placement is vs = (m + m0)g/2kf, the static displacement due to the total
weight of the superstructure and foundation mat. During the dynamic
excitation, the vertical displacement v will remain constant at the initial
static value if the foundation mat is bonded to the supporting elements,
but it will vary with time in the unbonded case. The displaced config-
uration of the structure at any instant of time can be defined by the
deformation u(t), foundation-mat rotation 9(f)/ and vertical displace-
ment v(t) at the center of gravity of the foundation mat.

EQUATIONS OF MOTION

The differential equations governing the small-amplitude motion of the

system of Fig. 1(a) can be derived by considering the lateral equilibrium
of forces acting on the structural mass m, and moment and vertical equi-
librium of forces acting on the entire system (see Appendix B, Ref. 10).
Assuming that the structural-foundation system and excitation are such
that the amplitudes of the resulting displacement and rotation responses
are small so that sin 9 and cos 9 may be approximated by 9 and 1, re-
spectively, these equations may be expressed as
mil + m(h%) + cu + ku = -mii„(t) (2a)
1127

J. Eng. Mech. 1984.110:1124-1146.

 m0b2 .. b2 . b
-—£ (hQ) -cu + elCf—2 (h%) + €2cf-v - ku
b2 b
+ ^kf-2 (M) + e2kf-v = 0. (2b)

b • b
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

(m + m0)v + ^CfV + e2cf~ (h%) + ^kfv + e2fy- (hQ) = -(m + m0)g (2c)

in which €! and e2 depend on whether one or both edges of the foun-

dation mat are in contact with the supporting elements; ex = 2 and e2 =
0 for contact at both edges, €i = 1 and e2 = ±1 if right or left edge is
uplifted. A supporting spring-damper element provides an upward re-
action force to the foundation mat
Rt(t) = -kfVt(f) - wit), i = l,r (3a)
in which v{(t) = v(t) ± bQ(t), i = I, r (3b)
As mentioned earlier, a downward reaction force cannot be developed
because the foundation mat is not bonded to the supporting elements,
i.e. Rj(t) a 0, i = I, r. If, at some instant of time, this condition is not
satisfied at the left or right edge of the foundation mat, i.e. JR,-(f) < 0, /
= I, r that particular edge uplifts from the supporting element which no
longer provides any reaction to the foundation mat. Because the spring-
damper element cannot extend above its initial unstressed position, i.e.
Vj(t) < 0, ; = I, r, an edge of the foundation mat would uplift at the
instant of time this condition is not satisfied. These two uplift criteria
are equivalent if each supporting element included only a spring.
The earthquake response of the system depends on the following di-
mensionless parameters: (1) co = Vk/m, the natural frequency of the rig-
idly supported structure; (2) £ = c/2ma>, the damping ratio of the rigidly
supported structure; (3) p = a)„/a>, where a);, = \/2kf/(m + m0) is the ver-
tical vibration frequency of the system of Fig. 1(a) with its foundation
mat bonded to the supporting elements; (4) £B = 2cf/2(m + m0)cov, the
damping ratio in vertical vibration of the system of Fig. 1(A) with its
foundation mat bonded to the supporting elements; (5) a = h/b, a slen-
derness-ratio parameter; and (6) 7 = m0/m, the ratio of foundation mass
to superstructure mass.
When the equations of motion and uplift criterion are expressed in
terms of these dimensionless parameters, for a given excitation, it is ob-
served that, under the small-amplitude assumption, the foundation-mat
rotation 0 depends on the width of the foundation mat; however, the
structural deformation u, the product of the foundation-mat rotation and
half base width Qb, and the vertical displacement v, do not depend on
the size of the structure. If we were to consider the large-amplitude mo-
tion of the structure including the possibility of its overturning, the size
parameter h would play an important role, just as in the dynamics of
rigid blocks (3). However, the foundations of most buildings are ex-
pected to undergo only small rotations and uplift displacements. At these
small-amplitude motions the deformation response of the system de-
pends on the h/b ratio but not separately on h.
1128

J. Eng. Mech. 1984.110:1124-1146.

 The equations of motion for the system of Fig. 1(a) are nonlinear as
indicated by the dependence of the coefficients ex and e2 on whether
both edges of the foundation mat are in contact with the supporting
elements, the left edge is uplifted, or the right edge is uplifted. How-
ever, for each of the three contact conditions the governing equations
are linear, but they depend on the contact condition. Thus, during the
time duration that one particular contact condition is valid, the system
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

behaves as a linear system, but the vibration properties of the linear

system are different for the three contact conditions. Thus the dynamic
response of the nonlinear system of Fig. 1(a) can be viewed as the se-
quential response of three linear systems.
The equations of motion can be specialized for the undamped system
with massless foundation mat by substituting m0 = 0, and c = cf = 0. In
particular, the inertia and damping terms in Eq. 2b are zero and, follow-
ing the usual approach to static condensation, 6 can be expressed in
terms of u and v from
b2 b
-ku + ^kf— (h%) + z2kf- v = 0 (4)

and substituted into Eqs. 2a and 2c. The reduced system consists of the
resulting two differential equations in the two unknown "dynamic" de-
grees-of-freedom. As presented later, this reduced system of equations
provides a basis for approximate analysis of systems with damping and
foundation-mat mass, i.e. c ¥> 0, Cf ¥> 0, and m0 ¥< 0.

NATURAL VIBRATION FREQUENCIES AND MODES

For the special case of y = 0, the natural frequencies and mode shapes
of the system of Fig. 1(a) can be determined by solving the eigenvalue
problems associated with the reduced version of Eq. 2. The resulting
two-degree-of-freedom eigenvalue problem is solved for the three linear
systems corresponding to the three sets of values for ex and e 2 . The nat-
ural frequencies of the linear system with both edges of the foundation
mat in contact with the supporting elements are
<x>i = couj; 0)2 = co0 =wp (5a)

p2
in which %= A/ 2+ (5b)

and the corresponding mode shapes are

4>[ = <d>? (1 - 6?) 0> (6a)
$1 = (0 0 1) (6b)
where the three terms in each mode shape are ordered to correspond
to r T = {u h% v). As shown in Fig. 2(a), the system vibrating in the sec-
ond mode undergoes uncoupled vertical motion, without lateral defor-
mation of the structure or rotation of the foundation mat, at the natural
frequency a>v defined earlier. The motion of the system vibrating in the
first mode consists of lateral deformation of the structure and rotation
1129

J. Eng. Mech. 1984.110:1124-1146.

 (a) CONTACT AT BOTH EDGES OF FOUNDATION MAT
?

f="=l
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

b) UPLIFT AT ONE EDGE OF FOUNDATION MAT

FIG. 2.—Free Vibration Mode Shapes for Two Cases: (a) Foundation Mat in Con-
tact with Supporting Elements at Both Edges; and (b) One Edge of Foundation
Mat Uplifted

of the foundation mat without any vertical displacement of its e g (rel-

ative to its static equilibrium position).
The natural frequencies of the linear system with uplift at one edge
(left or right edge) of the foundation mat are
Xj = 0; X2 = o)X2 (7a)

1(1 + a 2 )p 2
in which X2 = W ^ + 2 (7b)

and the corresponding mode shapes are

itf = (o 1 ± i \ (8a)
«|»J = (X2 (1 - X2) f a ) (8b)
Eqs. 7 and 8 are applicable to the system with uplift at left or right edge
of the foundation mat. Wherever both algebraic signs appear simulta-
neously the upper sign applies to uplift at the left edge and the lower
sign to uplift at the right edge. The mode shapes described by Eq. 8 are
plotted in Fig. 2(b) relative to an initial reference position (shown in dashed
lines) which is somewhat arbitrary. The system vibrating in the first mode
undergoes rigid body rotation about one edge of the foundation mat,
whereas in the second mode motion occurs in all three degrees of free-
dom: lateral deformation of the structure, vertical displacement and ro-
tation of the foundation mat. If the frequency ratio p exceeds the slen-
derness-ratio parameter a, which in turn is much larger than unity, it
can be shown that X2 is close to a, the value determined directly for the
structure of Fig. 1(a) but supported directly on the rigid base without
the spring-damper elements (5).
For the general case with some foundation-mat mass (7 ¥= 0), the nat-

1130

J. Eng. Mech. 1984.110:1124-1146.

 ural frequencies and mode shapes can be determined by solving the ei-
genvalue problem associated with Eq. 2. The three-degree-of-freedom
eigenvalue problem was solved for the three linear systems correspond-
ing to the three sets of values for €1 and e2 (see Appendix C, Ref. 10).
For the linear system with both edges of the foundation-mat in contact
with the supporting elements, the first two natural frequencies and modes
are described by Eq. 5a and Fig. 2(a) even when the foundation mat has
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

some mass. However, now a>i depends also on 7 and is no longer de-
scribed by Eq. 5b. Because the natural frequency and the shape of the
first mode involves wj (Eqs. 6a and 7a), they also depend on 7. The in-
fluence of 7 on «! decreases monotonically as 7 becomes smaller. For a
fixed value of the frequency ratio (3, the natural frequency and shape of
the second mode is independent of the mass ratio 7. A third vibration
mode appears (Fig. 2(a)) when the foundation mat has some mass. Just
like the first mode, the third mode includes lateral deformation of the
structure and rotation of the foundation mat without any vertical dis-
placement of its e g (relative to its static equilibrium position). It can be
shown through numerical results and a mathematical limiting process
that the natural frequency of the third mode is at least an order of mag-
nitude higher than the other two frequencies, and it tends to infinity as
the mass of the foundation mat approaches zero (see Appendix C, Ref.
10).
For the linear system with uplift at one edge of the foundation mat,
the first two natural frequencies and modes are described by Eqs. 7 and
8 and Fig. 2(b), even when the foundation mat has some mass. How-
ever, k2 is no longer described by Eq, 7b and depends on 7. The effect
of mass ratio 7 on the natural frequency and shape of the second mode
decreases monotonically with 7. The natural frequency and shape of the
first mode, which is a rigid body mode, is independent of the mass ratio
7. The third vibration mode, which appears when the foundation has
some mass, includes displacements in all the three degrees of freedom
just as in the second mode. Just as in the earlier case, the natural fre-
quency of the third mode is at least an order of magnitude higher than
the second mode frequency and it tends to infinity as the foundation-
mat mass approaches zero (see Appendix C, Ref. 10).

FREE VIBRATION RESPONSE

Undamped System.—Starting with the initial state of the structural

system of Fig. 1(a), wherein both edges of the foundation mat are in
contact with the supporting elements, the response of the system with
massless foundation mat to any initial displacements and velocities can
be determined as the superposition of the free vibration responses in
the two modes of vibrations §1, and <$>2 • In particular, if an initial ve-
locity x(0) is imparted in the lateral direction to the mass m of the su-
perstructure, the second vibration mode will not be excited and the re-
sponse will be entirely due to the first mode. The structural displacement
vector tT = (u h% v) is then described by

r(t) = §1 sin (Hit - vs<$>2 (9)

1131

J. Eng. Mech. 1984.110:1124-1146.

 provided the motion is small enough that both edges of the foundation
mat remain in contact with the supporting elements. From Eq. 9, the
structural displacements at the point of incipient uplift of one edge of
the foundation mat are
i (10)
uc = h% =
au> pV
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

These displacements are same as those presented in Eq. 1, specialized

for massless foundation mat, which were obtained from static con-
siderations.
The critical velocity x(0)c is defined as the initial velocity that results
in the maximum displacements r(f), from Eq. 9, equal to the displace-
ments given by Eq. 10. It can be shown that

Va 2 + p2 g
x(0)c = (11)
ap
and it depends on the slenderness-ratio parameter a and frequency ratio
p. For purposes of the subsequent discussion it is useful to introduce
the normalized initial velocity x(0) = x(0)/x(0)c. If the initial velocity is
less than the critical velocity, i.e., x(0) < 1, both edges of the foundation
mat will remain in contact with the supporting elements and no uplift
will occur; the free vibration response will be described by Eq. 9.
If the initial velocity exceeds the critical velocity, i.e., x(0) > 1, Eq. 9
will be valid only until the first uplift occurs. One edge of the foundation
mat will uplift at the time instant u (t) given by Eq. 9 reaches the uc value
of Eq. 10. During the time duration that uplift of this edge continues,
the displacement response can be expressed as

'(*) = gP + Cit + Qi *1
2 1 + a"

C 2 sin(M + 8)+i U^ i|) 2 (12)

wherein t is the time measured from the onset of uplift and, where two
algebraic signs appear, the upper sign is to be used if the left edge of
the foundation mat uplifts and the lower sign applies to uplift of the
right edge. The four constants Q , C 2 , C3 and 8 can be determined from
the displacements and velocities of the system at the onset of uplift, by
transforming them to modal coordinates.
After vibration of the structure for some time with one edge of the
foundation mat uplifted, as described by Eq. 12, the foundation mat will
re-establish contact with supporting elements at both its edges. Mea-
suring time t' from this instant, the response of the structure during the
time span it continues to vibrate without uplift is given by
x(t) = Djcfo sin (&>!*' + 8j) + D2<i>2 sin (<o2f' + 82) - vs$i (13)
Both vibration modes may now contribute to the response, in contrast
to Eq. 9 wherein the second vibration mode did not appear. The four
constants D j , D 2 , 8j and 82 can be determined by standard procedures
1132

J. Eng. Mech. 1984.110:1124-1146.

 BONDED CONTACT, UPLIFT PHEVENTED
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

UNDAMPED STRUCTURE DAMPED STRUCTURE { £ = 0 . 0 5 , £ v =0.4)

FIG. 3.—Response of Structure (a = 10, (3 = 8, 7 = 0) to Initial Velocity *(0) = 2

for Two Conditions of Contact between Foundation Mat and Supporting Elements:
(a) Bonded Contact Preventing Uplift; and (b) Unbonded Contact Only through
Gravity with Uplift Permitted

from the displacements and velocities of the structure at the time when
contact at both edges of the foundation mat is re-established.
Numerical results for the response of a structure to normalized initial
velocity x(0) = 2 are presented in Fig. 3(a) for two conditions of contact
between the foundation mat and the supporting spring-damper ele-
ments: (1) Bonded contact preventing uplift; and (2) unbonded contact
only through gravity with uplift permitted. The response quantities pre-
sented as a function of time are deformation u, foundation-mat rotation
0, total lateral displacement x = u + hd, vertical displacements vt and vr
of the left and right edges of the foundation mat, and uplift index which
is zero if, there is no uplift, +1 if the left edge uplifts and - 1 if the right
edge uplifts.
If the foundation mat is bonded to the supporting elements, the struc-
tural response is described by Eq. 9 for all time. The frequency of this
simple harmonic motion is coi, defined in Eq. 5a, and the amplitudes of
the response quantities are given by the critical values of Eq. 10 multi-
plied by the normalized initial velocity, which is 2 in this case.
The structural response is initially described by Eq. 9 even when uplift
is permitted, resulting in response identical to the bonded case. When
the deformation u and foundation-mat rotation 0 reach their critical val-
ues (Eq. 10), the left edge of the mat uplifts at time a (Fig. 3(a)) and the
subsequent response is described by Eq. 12. The deformation u oscillates
with frequency X2 about a value = uc(h2/h2 + b2) which for a wide range
of a is close to uc. This higher frequency response appears also in Vi(t)
and vr(t) and slightly in 0(f), but the total lateral displacement x(t) varies
smoothly with time without any noticeable contribution of the higher
frequency mode. Contact at both edges of the foundation mat is re-es-
tablished at time b and the response is described by Eq. 13 until the right
edge uplifts at time c; at which time Eq. 12 takes over again until contact

1133

J. Eng. Mech. 1984.110:1124-1146.

 at both edges is re-established at time d and Eq. 13 is back in the picture.
This response behavior is repeated during every cycle of free vibration
as seen in Fig. 3(a). However, the response is almost, but not precisely,
periodic because the initial conditions are not exactly repeated at the
beginning of each vibration cycle. Whereas no vertical velocity is im-
parted to the structural mass when free vibration is initiated at t = 0, a
small vertical velocity may be present at the beginning of subsequent
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

cycles of free vibration. In addition to the slight differences in the re-

sponse details from one cycle to the next, this perturbation can produce
additional very short duration uplift episodes, as seen in the lowest por-
tion of Fig. 3(a).
Because of uplift of the foundation mat, the maximum structural de-
formation u is reduced but the maximum values of all the other response
quantities—0, x, v, and vr—are increased. Noting that the time scale in
Fig. 3(a) is normalized with respect to the natural vibration period T of
the rigidly supported structure, it is apparent that the flexibility of the
supporting elements has the effect of lengthening the vibration period
to Tc, and uplift of the foundation mat results in even a longer period
T
Effects of Damping.—Expressed as a combination of modal contri-
butions, the response of the undamped system to an initial velocity was
given by Eq. 9 during the time that both edges of the foundation mat
remain in contact with the supporting elements, by Eq. 12 during the
time one edge is uplifted, and by Eq. 13 after contact has been re-es-
tablished at both edges. When damping in the structure and the sup-
porting elements is included, the responses of a system with massless
foundation mat will be described by Eqs. 9,12, and 13 with appropriated
modifications to account for modal damping. In Eqs. 9 and 13, these
include exponential decay terms e^""1"' and e~in"nt', respectively; and sub-
stitution of the damped frequency co^, instead of co„ in the harmonic terms,
where ®'n = o>„ V I - £2 and the modal damping ratios are

P3 / a2 \
= i+ v (14fl)
^ VTwA ^ )
&= fc (14k)
In Eq. 12 only the harmonic term is modified to include the exponential
decay term e~fcXzf and the damped frequency \ 2 is substituted instead of
the undamped frequency X2/ where X2 = X2 V I - 11 and the damping
ratio is

p 3 (l + a 2 ) 1 / 2 / 2a2 \
(15)
^W^rK^yH
For the structure with both edges of the foundation mat in contact
with the supporting elements, the damping ratio for the second vibra-
tion mode—which involves uncoupled vertical motion without lateral
deformation or base rotation—is simply £„, the damping ratio of the
structure in vertical vibration (Eq. 14b). In the first mode which includes
lateral deformation of the structure coupled with rotation of the foun-
1134

J. Eng. Mech. 1984.110:1124-1146.

 dation mat about its e g , the damping ratio is a linear combination of £„
and (j, where the latter is the damping ratio of the rigidly supported
structure in lateral vibration. For the linear system with uplift at one
edge, the first vibration mode is a rigid-body mode without any damp-
ing. The damping ratio for the second vibration mode, which includes
lateral structural deformation coupled with foundation-mat rotation, is
a linear combination of £„ and (,.
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

The response of the structure considered earlier, but now including

damping in the structure as well as supporting elements, to normalized
initial velocity *(0) = 2 is presented in Fig. 3(b). From the selected values
€ = 0.05, and &, = 0.4, Eqs. 14 and 15 lead to & = 0.03, & = 0.4 and £2
= 0.25. Because of damping the tendency of the foundation mat to uplift,
is reduced, resulting in the uplift duration decreasing with each vibra-
tion cycle; after a few cycles the foundation mat remains in a continuous
contact with the supporting elements and the responses u, 8 and x decay
exponentially at a rate controlled by £i. The vibratory term in the re-
sponse during the time one edge is uplifted (Eq. 12) decays exponen-
tially at a rate defined by £2 / which in this case is large and results in a
quick decay of the high-frequency oscillations in u about the uc value.
The corresponding high-frequency effects in the other responses also
disappear.
Based on the aforementioned considerations and numerical results of
Fig. 3(b), we can examine the dependence of damping effects on the
normalized initial velocity x(0). If ±(0) exceeds 1 only slightly, the foun-
dation mat will uplift from its supporting elements only for relatively
short durations of time; for most of the time, neither edge of the foun-
dation mat will uplift. The response of the structure will therefore be
dominated by Eq. 13. The second mode does not contribute to the struc-
tural deformation, which will therefore decay exponentially at a rate con-
trolled by the damping ratio & for the first mode. As discussed earlier,
this damping ratio is a linear combination of £ and £„.
If the normalized initial velocity x(0) is much larger than 1, the foun-
dation mat will remain in contact with its supporting elements only for
a small fraction of a vibration cycle, too short to cause any significant
decay of response. The left or right edge of the foundation mat will be
uplifted during a major part of a vibration cycle. In this condition, the
rotation and vertical displacement of the foundation mat are primarily
due to the undamped rigid-body mode. The high-frequency (= X2) con-
tributions to the structural deformation and vertical displacements at the
edges of the foundation mat (Fig. 3(a)) essentially disappear due to
damping (Fig. 3(b)), because they are due to the second mode (Eq. 12)
which has a modal damping £2 • For typical parameter values X2 is several
times higher than the overall frequency, thus several vibration cycles of
the second mode occur during each uplift phase (Fig. 3(a)). The number
of cycles is sufficient to essentially eliminate this high frequency re-
sponse for a wide range of values of damping and other structural pa-
rameters. Considering that the rigid body mode is undamped and the
high frequency response is eliminated by damping, the structural de-
formation response is insensitive to the damping values ij and £„ if
i(0) is significantly larger than 1. This conclusion is supported by nu-
merical results not presented here.

1135

J. Eng. Mech. 1984.110:1124-1146.

 ANALYSIS PROCEDURE

The response of the system of Fig. 1(a) to specified ground motion can
be analyzed by numerical solution of the equations of motion (Eq. 2).
These equations are nonlinear as indicated by the dependence of the
coefficients ej and e2 on whether one or both edges of the foundation
mat are in contact with the supporting elements. For each of the three
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

contact conditions—contact at both edges, left edge uplifted, and right

edge uplifted—the governing equations are linear, but they depend on
the contact condition. Thus, during the time duration that one particular
contact condition is valid, the system behaves as a linear system, but
vibration properties of the linear system are different for the three con-
tact conditions. Thus, at any time, the numerical analysis of response
over the next time-increment consists of two steps: (1) Determining the
contact condition valid for that time increment; and (2) solution of the
equations of motion with the corresponding values for ^ and e 2 .
As mentioned earlier, for each contact condition the linear system has
three degrees-of-freedom and the natural frequency of the third mode
is at least an order of magnitude higher than the second mode fre-
quency, and it tends to infinity as the mass of the foundation-mat ap-
proaches zero (see Appendix C, Ref. 10). Because the time-increment
employed in numerical analysis of response is controlled by the vibra-
tion frequency of the highest mode, very short time increments must be
used to accurately represent the contributions of the third mode. Be-
cause its frequency is at least an order of magnitude higher than that of
the second mode, the time-increment would be so short to be almost
impractical. Fortunately, the contribution of the third mode to the re-
sponse of each of the three linear systems (corresponding to the three
contact conditions) is negligible (see Appendix C, Ref. 10). By eliminat-
ing the effects of this high frequency mode in the numerical analysis,
the time increment will be controlled by the natural frequencies of the
first two vibration modes.
The obvious procedure to eliminate the high frequency effects is to
express the structural displacements as a linear combination of the first
two vibration modes of the system for the appropriate contact condition.
However, this approach has the disadvantage that the choice of the time-
varying modal coordinates depends on the contact condition. The more
convenient approach is to transform the equations of motion to a set of
displacements that can be used for all contact conditions. To this end
the rotation 9 is expressed in terms of u and v from Eq. 4:
2
1
7 7 U (16)
ei L«/
The displacement vector can therefore be expressed as
1 0
u
(17a)
^kfXb) \7Jb
.v J
. 0 1 J
or r = Tf. (17b)
1136

J. Eng. Mech. 1984.110:1124-1146.

 Substituting the transformation of Eq. 17 into Eq. 2 and premultiplying
both sides by the transpose of the transformation matrix T leads to a
reduced system of two differential equations in the two unknowns u
and v.
The eigenvalue problem associated with the reduced system of two
differential equations was solved. The resulting natural frequencies and
shapes of the two vibration modes were compared with the 'exact' val-
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

ues for the lower two modes, obtained from analysis of the original three-
degree-of-freedom eigenvalue problem. The agreement was found to be
excellent over a wide range of the system parameters 7, a and 0 (see
Appendix C, Ref. 10). Thus, the reduced system of equations accurately
retains the vibration properties of the lower two vibration modes, while
eliminating the high-frequency third mode.
The reduced system of equations is integrated numerically using an
implicit method with linear variation of acceleration in each time-step.
Appropriate governing equations are used, consistent with the contact
condition at the beginning of the time-step. The necessary modifications
are incorporated if the contact condition changes during a time-step (see
Appendix D, Ref. 10). By eliminating the high-frequency, third vibration
mode it was possible to employ a much larger integration time-step than
would otherwise be appropriate. The integration time-step At = 0.01 sec
used in this investigation is ten times longer than that necessary to ac-
curately solve the equations of motion in their original form (7), resulting
in considerable saving in computational effort.
The procedure described to eliminate the contributions of the high-
frequency, third vibration mode may be viewed as an application of the
classical Rayleigh-Ritz method, wherein the two Ritz vectors are de-
scribed by the columns of the transformation matrix T. If the foundation
mat is massless, this procedure is equivalent to the static condensation
approach outlined in a preceding section.

EARTHQUAKE RESPONSES

The response of a structural system to the north-south component of

the El Centro, 1940, ground motion computed by the numerical proce-
dures described in the preceding section is presented in Fig. 4. Re-
sponses are shown for two conditions of contact between the foundation
mat and the supporting spring-damper elements: (1) Bonded contact
preventing uplift; and (2) unbonded contact only through gravity with
uplift permitted. In the first case, the structural response is entirely due
to the first natural vibration mode of the system with both edges of the
foundation mat in contact with the supporting elements. Thus, the re-
sponse behavior is similar to a single-degree-of-freedom system. When
uplift of the foundation mat is permitted, the response behavior is much
more complicated. During the initial phase of the ground shaking, both
edges of the foundation mat remain in contact with the supporting ele-
ments. As the ground motion intensity builds up, the two edges of the
foundation mat alternately uplift in a vibration cycle. In this example
uplift occurs every vibration cycle during the strong phase of ground
snaking, with the duration of uplift depending on the amplitudes of
foundation mat rotation. As the intensity of ground motion decays to-

1137

J. Eng. Mech. 1984.110:1124-1146.

 • BONDED CONTACT, UPLIFT PREVENTED
• UNBONDED CONTACT, UPLIFT PERMITTED
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

-W^>^V^v^-^VA/V

10 15 20
TIME I , SEC.

FIG. 4—Response of Structure (a = 10, B = 8, 7 = 0, T = 1 sec, £ = 0.05, £„ =

0.4) to El Centra Ground Motion for Two Conditions of Contact between Foun-
dation Mat and Supporting Elements: (a) Bonded Contact Preventing Uplift; and
(b) Unbonded Contact Only through Gravity with Uplift Permitted

ward the later phase of the earthquake, the foundation mat no longer
uplifts, and both edges remain in contact with the supporting elements.
The effects of foundation-mat uplift on the maximum response of the
structure due to earthquake ground motion are similar to those observed
in a preceding section during free vibration. The maximum deformation
of the structure is close to uc when foundation-mat uplift is permitted,
which is a significant reduction compared to the response when foun-
dation-mat uplift is prevented. Because of damping, the small-amplitude
oscillations at the higher frequency X2 damp out and are not present in
the earthquake response results of Fig. 4. The rotation of the foundation
mat and vertical displacements of the two edges of the foundation mat
are significantly increased due to the rigid body rocking mode of the
system, permitted by uplift. This mode provides the dominant contri-
bution to these responses during uplift but does not affect the structural
deformation.
In order to study the effects of foundation-mat uplift on the maximum
response of structures, response spectra are presented. The base shear
coefficient
. , 2
* max
ku„ 2ir\ U„
(18)
W mg
in which Vmax = the maximum base shear, and w = the weight of the
1138

J. Eng. Mech. 1984.110:1124-1146.

 superstructure, is plotted as a function of the natural vibration period
of the corresponding rigidly supported structure. For each set of system
parameters a, (3, 7, £ and (•„, such a response spectrum plot is presented
for two conditions of contact between the foundation mat and the sup-
porting spring-damper elements: (1) Bonded contact preventing uplift;
and (2) unbonded contact only through gravity with uplift permitted.
Also presented are the results for the corresponding rigidly supported
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

structure, which is simply the standard pseudoacceleration response

spectrum, normalized with respect to gravitational acceleration. In-
cluded in the response spectra plots is Vc, the critical base shear coef-
ficient associated with the maximum value of base shear Vc obtained
from static considerations (Eq. la):
Vc I + 7
Vc = — = - (19)
w a
This critical base shear coefficient depends on the mass ratio 7 and slen-
derness-ratio parameter a, but is independent of the vibration period T.
The response spectra presented in Fig. 5 are for systems with massless
foundations (7 = 0) and a fixed set of system parameters a, p, £, and £„
subjected to the El Centro ground motion. The differences between the
response spectra for the two linear systems, the structure with foun-
dation mat bonded to the supporting elements and the corresponding
rigidly supported structure, are due to the change in period and damp-

- ,-v-"\ -

- 1 \
1 \

» -1 V
- X
"* \
*"N^^v_»rf—^t \ CRITICAL BASE SHEAR
0.10 COEFF. Vz
0.08
0.06 -
0.04

0.010
0.008
0.006 -
RIGID SUPPORTING SYSTEM
0.004
F L E X I B L E SUPPORTING SYSTEM
BONDED CONTACT, UPLIFT PREVENTED
__ UNBONDED CONTACT .UPLIFT PERMITTED
~
1 1 1 11 1 1 1 1 1
0.4 0.6 0.8 I 2 4
VIBRATION PERIOD T, SEC.

FIG. 5.—Response Spectra for Structures (a = 10, p = 8, 7 = 0, £ = 0.05, iv

0.4) Subjected to El Centro Ground Motion for Three Support Conditions
1139

J. Eng. Mech. 1984.110:1124-1146.

 ing resulting from support flexibility (Eqs. 5 and 14). The base shear
developed in structures with relatively long vibration periods is below
the critical value and the foundation mat does not uplift from its sup-
porting elements. If foundation mat uplift is prevented, the maximum
base shear at some vibration periods may exceed the critical values; for
the selected system parameters and ground motion Fig. 5 indicates that
the critical value is exceeded for all vibration periods shorter than the
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

period where the linear spectrum first attains the critical value. If the
foundation mat of such a structure rests on the spring-damper elements
only through gravity and is not bonded to these elements, uplift occurs
and this has the effect of reducing the base shear. However, the base
shear is not reduced to as low as the critical value based on static con-
siderations (Eq. la). Although foundation-mat uplift is initiated at the
time instant the base shear in a vibrating structure reaches the instan-
taneous dynamic critical value, depending on the state—displacement,
velocity and acceleration—of the system the deformation and base shear
may continue to exceed the critical values. Furthermore the dynamic,
critical base shear may exceed the static critical value because the time-
dependent vertical force p may exceed the gravity loads. Because the
base shear developed in linear structures (foundation-mat uplift pre-
vented) tends to exceed the critical value by increasing margins as the
vibration period decreases, the foundation mat of a shorter-period struc-

2.0

1.0
0.8
0.6

0.4

CRITICAL BASE SHEAR

£ 0.08

i 0.06

Ul 0.02
CO
<
m
0.010
0.008
0.006
RIGID SUPPORTING SYSTEM

" F L E X I B L E SUPPORTING SYSTEM

0.004 p = A, BQNOEDCONTACT.UFLIFTPREVENTEO
UNBONDED CONTACT, UPLIFT PERMITTED \
BONDED CONTACT, UPLIFT PREVENTED
0.002 UNBONDEDCONTACT.UPLIFT PERMITTED

0.1 0.2 0.4 0.6 0.8 I 2 4 6 6 10 20

VIBRATION PERIOD T, SEC.

FIG. 6.—Response Spectra for Structures (a = 10, y = 0, £ = 0.05, £„ = 0.4) Sub-

jected to El Centre Ground Motion for Two Values of Frequency Ratio p = 4 and
12

1140

J. Eng. Mech. 1984.110:1124-1146.

 ture has a greater tendency to uplift, which in turn causes the critical
base shear to be exceeded by a greater margin, although it remains well
below the linear response.
The response spectra presented in Fig. 6 are for two values of the
frequency ratio p with all other system parameters kept constant. As
mentioned earlier, the differences in the response spectra for the two
linear systems, the structure with foundation mat bonded to the sup-
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

porting elements and the corresponding rigidly supported structure, are

due to the change in period and damping resulting from support flex-
ibility (Eqs. 5 and 14). The period change appears as a shift in the re-
sponse spectrum to the left, with larger shift for smaller values of p, i.e.,
for the more flexible supporting elements. The nonlinear response spec-
trum for the structure with foundation mat permitted to uplift shifts sim-
ilarly to the left, with uplift initiated at longer periods as (3 increases.
But for the period shift, the shape of the linear as well as nonlinear
response are affected little by the frequency ratio p.
As presented in Eq. 19, the critical base shear coefficient is inversely
proportional to the slenderness-ratio parameter a. This would suggest
that maximum base shear will be smaller for relatively slender struc-
tures, which is confirmed by response spectra presented in Fig. 7. The
foundation mat of a slender structure has a greater tendency to uplift
resulting in greater reductions in the base shear. Uplift of the foundation

2.0

1.0"
o.e

0.6

0.4
V
^6 0.2 ~W\\ s-V\
CRITICAL BASE SHEAR
COEFF, V^FOR a =5
H
Z \
UJ
\ \\
5 0.08
C
°'° \\
u.
LU
o
0.06 *-- __\ \ \ CRITICAL BASE SHEAR
\ \ c O E F F . Vc FOR a = 20
o
cc 0.04 \
< V \ ™X\A
UJ
X
<n -
LI 0.02 VA \\r
<
CO
-
0.010
0.008 RIGID SUPPORTING SYSTEM

0.006 - F L E X I B L E SUPPORTING SYSTEM

0.004 a = 2 0 , BONDEDCONTACT,uPUFTPREVENTED \
a = 2 0 , UNBONDED CONTACT UPLIFT PERMITTED!

0.002 ..— a = 5 , UNBONDED CONTACT UPLIFT PERMITTED

1 1 1 1 1 1 1 I I I 1 \
0.4 0.6 0.8 I
VIBRATION PERIOD T, SEC.

FIG. 7.—Response Spectra for Structures (p = 8, 7 = 0, £ = 0.05, £„ = 0.4) Sub-

jected to El Centro Ground Motion for Two Values of Slenderness Ratio Parameter
a = 5 and 20
1141

J. Eng. Mech. 1984.110:1124-1146.

 mat occurs at all vibration periods shorter than the period where the
linear response spectrum attains the critical value. This period depends
on the slenderness ratio a in a complicated manner, because the critical
base shear coefficient as well as the period shift of the linear response '
spectrum (relative to the standard pseudo-acceleration spectrum) both
depend on a. For relatively slender structures the critical base shear coef-
ficient is smaller (Eq. 19), but the period shift is larger (Eq. 5).
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

SUMMARY AND CONCLUSIONS

The equations of motion for the structure with its foundation mat per- i
mitted to uplift are nonlinear with foundation stiffness and damping
parameters dependent on whether the foundation mat is in contact with
the supporting systems at one or both edges. However a linear system !
corresponding to any instantaneous contact condition of the foundation
mat can be defined. This linear system has three degrees-of-freedom and |
the natural frequency of the third mode is at least an order of magnitude ':
higher than the second mode frequency and it tends to infinity as the
foundation-mat mass approaches zero. Because the contribution of this
high frequency mode to the response is negligible, the Rayleigh-Ritz
concept was employed to eliminate this mode from the analysis. It was
then possible to employ a much larger integration time-step than would ,
otherwise be practical, resulting in considerable savings in computa- I
tional effort. i
Analytical examination of the governing equations and system prop- \
erties and the numerical results, presented here and elsewhere (10), in-
dicate that the earthquake response of uplifting structures is controlled I
by the following system parameters, listed in more or less descending
order of importance: (1) Natural vibration frequency u of the rigidly sup- i
ported structure; (2) slenderness-ratio parameter a; (3) ratio 7 of foun-
dation mass to superstructure mass; (4) (3 = coB/w where a>v is the vertical
vibration frequency of the system with its foundation mat bonded to \
supporting elements; (5) damping ratio £ of the rigidly supported struc- I
ture; and (6) damping ratio £,v in vertical vibration of the system with its j
foundation mat bonded to the supporting elements.
If we had considered the large-amplitude motion of the structure in-
cluding the possibility of its overturning the response of a structure would
be influenced also by its size. Similarly the size influences even the small-
amplitude response if p-8 effects are considered, but the results pre- j
sented indicate that these effects are insignificant for most buildings (10).
A study of the response spectra plots presented leads to the following
conclusions:

1. The base shear developed in structures with relatively long vibra-

tion periods is below the critical value, the maximum base shear that
can be developed with the structure supported only through gravity, f
and the foundation mat does not uplift from its supporting elements. i
2. For short period structures, the base shear exceeds the critical value j
if foundation-mat uplift is prevented; for such structures, permitting uplift |
has the effect of reducing the base shear—to values somewhat above
the critical value. 1
1142

J. Eng. Mech. 1984.110:1124-1146.

 3. The foundation mat of a slender structure has a greater tendency
to uplift resulting in greater reductions in base shear.
4. The shape of the response spectrum is affected little by the fre-
quency ratio p, but it does have the effect of shifting the response spec-
trum to the left with larger shift for smaller (3, i.e., for the more flexible
supporting elements.
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

The possibility of transient uplift of a portion of the foundation mat

or of a few individual footings, as the case may be, should be considered
in the analysis of response of structures subjected to intense earthquake
ground motion. Because foundation uplift generally has the effect of re-
ducing the structural deformations and forces, there is no need to pre-
vent it but, on the contrary, it is desirable to permit it. The foundation,
underlying soil and the structural columns should be properly designed
to accommodate the transient uplift and the effects of subsequent impact
on contact.
Although the maximum base shear is reduced because of foundation-
mat uplift for slender structures over a wide range of fixed-base vibra-
tion period, very short period structures may be exceptions to this be-
havior. Their response, as seen in results presented elsewhere (1), may
increase because of foundation mat uplift. However these vibration pe-
riods are unrealistically short for slender structures representative of me-
dium-rise to high-rise structures with longer vibration periods. Thus, the
increased computed response of very short period structures in spite of
foundation mat uplift is of little practical consequence.
Because they are based on the response results presented here and in
the more complete report (10), all for one selected ground motion, strictly
speaking the aforementioned conclusions may not be valid for other ex-
citations or range of structural parameters. However, as observed else-
where (1), the response behavior of structures, excluding those with very
short fixed-base vibration periods, is simple when their foundation mat
is permitted to uplift even though the system is nonlinear. In a sense
the behavior seems to be even simpler than that of the corresponding
linear system—the structure with its foundation mat bonded to the sup-
porting soil. In particular, the maximum base shear of uplifting struc-
tures does not depend significantly on the details or intensity of the
ground acceleration ug(t), but is controlled by the critical shear Vc (Eq.
la) for the structure (10). Thus it seems that the aforementioned conclu-
sions should be also valid for a variety of ground motions.
Although the results of this work have provided an understanding of
the basic effects of transient foundation uplift on the earthquake re-
sponse of structures, there is an important limitation in the application
of these results to actual buildings. In reality the foundation stiffness
and damping parameters depend on the displacement amplitude and,
excitation frequency, in contrast to the constant parameters employed
in this work. In either case, the foundation parameters are very difficult
to evaluate because they would depend on the details of the foundation
design, including the degree of foundation embedment and the base
deformability. Reliable methods to evaluate foundation parameters for
actual buildings need to be developed, so that the beneficial effects of
foundation uplift can be considered in the design of buildings.
1143

J. Eng. Mech. 1984.110:1124-1146.

 ACKNOWLEDGMENT

This research was conducted u n d e r Grants PFR79-08261 a n d CEE81-

05790 from the National Science Foundation for which the authors are
grateful.

APPENDIX I.—REFERENCES
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

1. Chopra, A. K., and Yim, S. C.-S., "Simplified Earthquake Response Analysis

of Structures with Foundation Uplift," Journal of Structural Engineering, ASCE,
submitted for publication.
2. Clough, R. W., and Huckelbridge, A. A., "Earthquake Simulation Tests of
a Three-Story Steel Frame with Columns Allowed to Uplift," Report No.
UCB/EERC 77-22, Earthquake Engineering Research Center, University of
California, Berkeley, Calif., Aug., 1977.
3. Housner, G. W., "The Behavior of Inverted Pendulum Structures During
Earthquakes," Bulletin of the Seismological Society of America, Vol. 53, No. 2,
Feb., 1963, pp. 403-417.
4. Huckelbridge, A. A., and Clough, R. W., "Seismic Response of Uplifting
Building Frame," Journal of the Structural Division, ASCE, Vol. 104, No. ST8,
Aug., 1978, pp. 1211-1229.
5. Meek, J. W., "Effects of Foundation Tipping on Dynamic Response," Journal
of the Structural Division, ASCE, Vol. 101, No. ST7, July, 1975, pp. 1297-1311.
6. Meek, J. W., "Dynamic Response of Tipping Core Building," Earthquake En-
gineering and Structural Dynamics, Vol. 6, No. 5, Sept.-Oct., 1978, pp. 437-
454.
7. Psycharis, I. N., "Dynamic Behavior of Rocking Structures Allowed to Uplift,"
Report No. EERL 81-02, Earthquake Engineering Research Laboratory, Cali-
fornia Institute of Technology, Pasadena, Calif., Aug., 1981; also see Earth-
quake Engineering and Structural Dynamics, Vol. 11, pp. 57-76 and 501-
521, 1983.
8. Seed, H. Bolton, and Idriss, I. M., "Soil Moduli and Damping Factors for
Dynamic Response Analyses," Report No. UCB/EERC 70-10, Earthquake En-
gineering Research Center, University of California, Berkeley, Calif., Dec,
1970.
9. Veletsos, A. S., and Verbic, B., "Vibration of Viscoelastic Foundations," In-
ternational Journal of Earthquake Engineering and Structural Dynamics, Vol. 2,
No. 1, July-Sept., 1973, pp. 87-102.
10. Yim, C. S., and Chopra, A. K., "Effects of Transient Foundation Uplift on
Earthquake Response of Structures," Report No. UCB/EERC 83-09, Earth-
quake Engineering Research Center, University of California, Berkeley, Calif.,
June, 1983.

APPENDIX II.—NOTATION

The following symbols are used in this paper:

b = half width of foundation mat;

c = lateral damping coefficient of superstructure;
Cf = damping coefficient of supporting element of two-element
foundation;
g = acceleration of gravity;
h = height of superstructure;
k = lateral stiffness of superstructure;
kf = stiffness of supporting element of two-element foundation;
M = static base m o m e n t applied at e g of foundation mat;

1144

J. Eng. Mech. 1984.110:1124-1146.

 Mc = maximum static base moment at full uplift of foundation mat
with contact reduced to one edge;
m = mass of superstructure;
m0 = mass of foundation mat;
Ri = upward reaction of left supporting element of two-element
foundation;
Rr = upward reaction of right supporting element of two-element
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

foundation;
T = natural vibration period of the rigidly supported structure;
Tc = rocking vibration period of flexibly supported structure with
uplift prevented;
Tu = rocking vibration period of flexibly supported structure with
uplift permitted;
u = structural deformation;
uc = maximum structural deformation under the action of static
lateral force;
iig(t) = horizontal ground acceleration;
"max = maximum structural deformation during an earthquake;
^max = a base shear coefficient = Vmax/w;
Vc = a base shear coefficient = Vc/w;
Vc = maximum base shear that can be developed under the action
of static lateral force;
=
Vmax maximum base shear;
v = vertical displacement of e g of foundation mat;
vi = vertical displacement of left edge of foundation mat;
vr = vertical displacement of right edge of foundation mat;
vs = vertical displacement of foundation mat under the action of
gravity forces;
w = weight of superstructure;
x = lateral displacement of structure relative to supporting foun-
dation = u + hQ;
x(0) = initial velocity of structure;
x(Q)c = minimum initial velocity of structure inducing foundation-mat
uplift of structure supported on two-element foundation;
i(0) = normalized initial velocity of structure;
a = slenderness ratio parameter;
P = w^/w;
7 = ratio of foundation mass to superstructure mass;
li = damping ratio of the z'th vibration mode after uplift;
8 = rotation of foundation mat;
0C = static rotation of foundation mat at full uplift with contact re-
duced to one edge;
\; = undamped natural vibration frequency of the z'th mode after
uplift;
i = damping ratio of the rigidly supported structure;
<•„ = damping ratio in vertical vibration of the system of Fig. 1 with
its foundation mat bonded to the two-element foundation;
£,• = damping ratio of the z'th vibration mode of system with both
edges of foundation mat in contact with foundation;
<!>,• = z'th vibration mode of system with foundation mat in contact
with both edges of foundation;
1145

J. Eng. Mech. 1984.110:1124-1146.

 >|/, = ith vibration mode of system after uplift;
co = natural vibration frequency of the rigidly supported structure;
a), = natural vibration frequency of the ith mode of system with
both edges of foundation mat in contact with supporting ele-
ments; and
co„ = vertical vibration frequency of the system with its foundation
mat bonded to the supporting elements.
Downloaded from ascelibrary.org by Auburn University on 09/11/13. Copyright ASCE. For personal use only; all rights reserved.

1146

J. Eng. Mech. 1984.110:1124-1146.