Sei sulla pagina 1di 14

EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS, VOL.

7, 555-568 (1979)

BASE ISOLATION SYSTEMS FOR EARTHQUAKE PROTECTION OF


MULTI-STOREY SHEAR STRUCTURES
D. M. LEE*

Department of Civil Engineering, California Institute of Technology, Pasadena, California, US.A.


AND
I. C . MEDLANDt

Department of Theoretical and Applied Mechanics, University of Auckland, Auckland, New Zealand

SUMMARY
This paper is a study of the effectiveness of a wide range of bilinear hysteretic isolation systems in shielding multistorey 2-D shear structures from earthquake excitations. Important parameters of the isolation system are identified and their effect on structure response noted. It is shown that isolation systems can be constructed which
allow the structure proper to remain purely elastic even during very strong ground motions. It is further shown
that the shear responses and base displacements of structures on these isolation systems can be accurately estimated from elastic response spectra of the forcing earthquakes.
The philosophy of structure isolation is discussed and a n introduction given to the physical devices currently
available to provide it.
INTRODUCTION
Many of the multi-storey buildings being constructed today have natural periods of vibration that lie
within the region in which earthquakes have their dominant energies. Furthermore, the elastic internal
damping of such structures is typically rather small and therefore, when shaken by an earthquake, the
vibrations generated in these structures can resonantly increase, stressing it beyond its elastic limit. Current
design practices account for this possibility by allowing the structure to develop inelastic hinges at beam
ends during strong ground motions. These hinges provide not only increased flexibility but also energyabsorbing capacity, both of which help to limit the earthquake generated forces. Such inelastic deformations
require large displacements, however, and cause progressive breakdown of the structural components as
well as severe and expensive secondary damage (e.g. fracture of partitions, windows, etc.). In many cases
the force limiting inelastic action of the structural components cannot occur until after this secondary
damage has occurred. Moreover, the ability of the structure to form the number and style of plastic hinges
required is by no means guaranteed.
Since flexibility and energy-absorbing capacity are desirable attributes but the large deformations and
structural damage that these entail are not, it would seem logical to mount a structure at its base on flexible,
energy-absorbing devices which, by limiting the forces transmitted into the structure proper (i.e. the structure
above the base-foundation interface), confine inelastic behaviour to the specially designed isolation system.
The techniques of so mounting a structure and the responses of these structures to earthquake forcing
. Kelly
~
have been studied under a variety of titles by many workers: Fintel and Kahn,l Chopra el ~ 1and
et aL3 (the soft first storey); Delfosse4 (the Gapec system); Plichon and Jolivet5 (an aseismic foundation
system) and Sinner et aZ.,63' Priestly et aZ.* and Lee and Medlands (the base isolation system). A more
complete review of the literature is contained in Lee and Medland.lo

Research Fellow. Formerly Graduate Student, Department of Theoretical and Applied Mechanics, University of Auckland, Auckland, New Zealand.
t Senior Lecturer.

Received 16 February 1979


Revised 14 June 1979

0098-8847/79/0607-0555$01.OO
1979 by John Wiley & Sons, Ltd.
39

555

5 56

D. M. LEE A N D I. C. MEDLAND

The majority of the above studies have suggested using laminated rubber pads to support the structure
weight and to introduce the required structure base-foundation flexibility. These pads, which consist of
thin sheets of rubber bonded to interleaving steel plates, have been used for many years to support bridge
decks but have more recently been the subject of extensive testing11-13 to determine their suitability for
structure isolation. These tests have demonstrated that shear strains of 100 per cent can be sustained for
many hundreds of cycles without damage. Examples of alternative support systems include roller bearings
and sliding supports.14
Much work has been done to devise cheap, reliable and effective energy-absorption devices for use with
the essentially elastic load-supporting system (rubber pads, rollers, etc.). Among those which have been
devised, tested and found to be satisfactory are absorbers which utilize the plastic torsion and bending of
mild steeP5-17 and the extrusion of lead.l* One-piece devices which combine both the load-bearing flexible
support and the energy-absorbing capacity in a single unit have also been constructed. One such device
utilizes a laminated rubber pad with a central, vertically mounted core of lead, constrained to deform in
pure shear between the pads steel plates.lg Another device utilizes a laminated rubber pad whose upper
surface is free to slide on the structure base when the shear across the pad reaches a critical level5 At the
time of writing the former device was being used to support a four-storey office block in Wellington, New
Zealand,2O while the latter device has already been employed to support the Koeberg Nuclear Power Plant
in South A f r i ~ a . ~
Figure 1 illustrates a possible foundation arrangement of a base isolation system in which the rubber
pads are mounted on removable wedges which, with local jacking, can be moved to adjust the pad elevation
or to allow it to be removed for testing or examination. (It is expected, however, that the rubber pads currently being used to support structures will outlast the economic life of the building and ought never to

**I.:

;:u

-~..w..~.A*u*;

.,-=a

a,:

.*.El

m.p:i--d*a.-A

Building Base

3~

&steretic

, ,-:.,

Dampers1

.:1 -

m-0

-d*a:

Fixed Foundation

Figure 1. Schematic representation of one unit of a possible foundation arrangement with base isolation

@
Force

Hysteretic Domper

Displ.

Mb

Rubber PodsY

(a)

(b)

Figure 2. (a) An N-storey lumped parameter, base isolated, 2-D shear structure whose rubber pad supports and hysteretic
dampers have together the bilinear hysteretic characteristics shown in (b), (for horizontal motions)

EARTHQUAKE PROTECTION OF MULTI-STOREY SHEAR STRUCTURES

557

need replacement.) The hysteretic dampers are located in a crawl space from where they can be conveniently accessed to correct any residual plastic offset following an earthquake. This space also provides an
area in which flexible plumbing and electrical connections can move. Use of the previously mentioned
one-piece devices would allow the arrangement of Figure 1 to be simplified considerably.
The aim of this paper, however, is not to investigate the physical properties and installation details of
any particular isolation system but rather to investigate the effect of a bilinear hysteretic isolation system
on the responses of multi-storey 2-D shear structures of the type shown in Figure 2. (This type of isolation
system was chosen as combinations of the devices examined11-19 do result in an essentially bilinear hysteretic
support whose properties remain essentially constant over many yield cycles.) Specifically, this study is
a detailed variation of those parameters considered to be of importance to the response of such isolated
structures, namely the structure period, the forcing earthquake and the defining parameters of the bilinear
hysteretic base isolation system, k,, k , and F.
STRUCTURE MODELS, RESPONSE PARAMETERS AND EARTHQUAKES USED

The N-storey shear structures considered in this investigation have floor masses and column stiffness uniform over the height, a base mass Mb of one-sixth the total mass of the structure proper ( N M ) , and fundamental periods when isolated (i.e. T, (UI)) of one-tenth the number of storeys. Orthogonal modal damping
of 5 per cent of critical was applied in each mode of vibration, with the dynamic solution routine adjusting
this damping as the stiffness of the isolation system changed from its elastic value, k,, to its post-yield
value, k,, and vice versa. In a larger
of which this paper forms a part, the amount of viscous damping, the base mass and reductions in the interstorey column stiffnesses (at constant structure fundamental
period) were shown to be only minor parameters in controlling the responses of isolated structures.
Four quantities are used to describe the essential features of the structure response. The first three of
these quantities are respectively the maximum interstorey shear force, S (which normally occurs at or near
the structure base), the maximum structure-base foundation displacement, d,, and the residual plastic
offset of the base from the foundation following an earthquake, aFL. The fourth quantity, B,, the 'bulge
defining parameter', is used to quantify the relative magnitudes of the other interstorey shear forces with
respect to their maximum value, S, by means of a set of normalized shear distributions as shown in Figure
3(b). These distributions are uniquely defined by the set of static horizontal floor forces, Fi, in Figure 3(a).
Fi is the positive horizontal distance from ordinate i / N on the vertical axis to the swinging arm OC, pivoted
about point 0 at angle 0 to the vertical. The load set for a given 0 produces a shear distribution of bulge
parameter.
BD = i-cos

(1)

As the angle 0 increases from zero, the intersection point, C, moves from E along ED then up DA, varying
the force distribution from a uniform one (0 = 0, BD = 0) to the common code provision case of an inverted
triangle of loads (0 = 45 degrees, B D z0.3), to the obviously overconservative situation of all load being
applied at the top floor (0 = 90 degrees, BD = 1.0). Thus increasing values of BD indicate shear distributions
which have increasingly more significant shear forces in higher levels of the structure, relative to that at
the base. The advantage of this method of describing shear distributions, over say a least squares deviation
from a linear distribution, is that these distributions not only closely approximate those observed in this
study but also encompass common code distributions and furthermore, given only BD and the number
of storeys, are easily reproduced. The dynamic solution routine takes the value of B,, defining the shear
distribution which is equal to or just larger than the actual distribution at every level of the structure, as
being the bulge of that actual shear distribution. This was done in order to have a single parameter with
which to display the effect of different isolation system parameters in promoting or inhibiting higher mode
responses in the structure.
A wide range of strengths and types of earthquakes were used to excite the isolated structures. The set
of earthquakes selected for this purpose included the Niigata (N.S., E.W., 16 June 1964) records?, the
artificial records A1 to D2 of Jennings et aZ.22and the Caltech digitized and corrected records AOOl (N.S.),

558

D. M. LEE AND I. C. MEDLAND

Storey
Number

'INID
0.0

0.5

I.o

F, (=Horiz. Force/Max

Horiz. Force)

'0.0 0.2 0.4 0.6 0.0 1.0


Normalised Interstorey Shear Force

(0)
(b)
Figure 3. (a) Normalized floor force distributions, (b) Normalized interstorey shear distributions produced from (a) for a
specific N

A008 (N79"E), B032 (S86"W), B033 (N65"E) and B034 (N85"E), C041 (S14"W, N76"W) and DO56 (N21"E,
N69"W).23This final set includes data from the earthquakes at, respectively, E l Centro (18 May 1940),
Eureka (21 December 1954), Puget Sound (29 April 1965), Parkfield (27 June 1966), Pacoima Dam and
the Castaic Old Ridge Route, San Fernando Valley (9 February 1971).
RESULTS
As the aim of this work was a parameter investigation and as there were an infinity of possible combinations of these parameters, a condensed procedure was used whereby only one parameter was varied at
one time, all others being held constant. The results of this study are therefore presented in three sections,
the first of which deals with the effect of varying the structure period and earthquake forcing of structures
mounted on a given isolation system. The second section then deals with the effect of varying the properties
of the isolation system being used to shield a given structure from a given earthquake. In the third and
final section these results are combined, with four different isolation systems (having parameters chosen
with the aid of the preceding analyses) being used in turn to isolate structures of different periods from a
variety of earthquakes.
Section I: structure period and earthquake variation
Figure 4 shows the maximum shear response, S , of one- to eighteen-storey structures (T, (UI)= 0.11.8 s) to the El Centro (N.S., May 1940) earthquake, both when unisolated and when mounted on an isolation system having bilinear hysteretic parameters of elastic stiffness k , = 5.0 W m-l, post-yield stiffness
k , = 1-0W rn-l and yield force F= 0.05 W, W being the weight of the structure above, but not including
the base (i.e. the structure proper). Structure shears are also expressed as a fraction of W. Two very important benefits of base isolation are immediately apparent, namely that not only is the maximum structure
shear force dramatically lowered but also that this lowered ,value is essentially independent of the original
fundamental period of the unisolated structure. The first of these effects is due to the much greater flexibility and damping of the isolated structure compared to the unisolated structure while the second is a
consequence of the effective period of the isolated structure being governed (for strong forcing) primarily
by the post-yield stiffness of the isolation system. The maximum base displacement was approximately
8 cm for all the structures while the residual plastic offset was less than 0.6 cm.
To examine the effect of different earthquakes on the responses of isolated, as opposed to unisolated,
structures, structures of T, (UI) = 0.3, 0.6 and 0-9 s were excited by each of the previously listed earthquakes. These responses were then linearly regressed against a variety of response spectrum parameters of
the forcing earthquakes and the degree of correlation noted for each parameter. As expected, the maximum

EARTHQUAKE PROTECTION OF MULTI-STOREY SHEAR STRUCTURES

0.5 -

4,lJnisolated

k-w-5

0.4 -

0.3

0.2

0.1

559

'\

'\
'\

\\*---*---x
s
e

Isolated

Figure 4. Maximum interstorey shear forces in multi-storey structures of unisolated fundamental period, TI (UI). El Centro
(N.S., May 1940) earthquake excitation

shear force in the unisolated (and 5 per cent damped) structures was most closely correlated with the
value of the 5 per cent damped acceleration response spectrum of the forcing earthquake, at an oscillator
period equal to TI(UI). The responses of the isolated structures were, however, also found to be very
closely correlated with response spectrum data of the forcing earthquakes but with the value of the 10
per cent damped velocity response spectrum at an oscillator period of TI (PY), the fundamental period
of the structure when mounted on a linear spring of stiffness equal to the post-yield stiffness of the isolation
system. Figure 5 superimposes plots of the maximum shear force for the isolated and unisolated cases,
each against its best describing parameter. The solid lines are the least squares fitted regression lines while
the dashed lines represent 95 per cent confidence intervals. The close correlation observed for the isolated
structures is particularly important as it allows one to estimate accurately the maximum value of the very
5% Damped Spectral Acceleration at T , (UI) ( r n ~ - ~ )
1

15.0

10.0

5.0

20.0
I

10%Damped Spectrol Velocity at TI (PY) (ms-I)


Figure 5. Linear least squares regression fits of the maximum shear force in: -unisolated structures, against the 5 per cent
damped spectral acceleration of the forcing earthquake at TI (UI) and in -isolated structures against the 10 per cent
damped spectral velocity of the forcing earthquake at TI (PY)

560

D. M. LEE AND I. C. MEDLAND

much inelastic response of the isolated structures, using only elastic response spectrum data. Thus Code
of Practice design of base isolated structures could be undertaken as easily as that currently done for unisolated structures. No inelastic forced vibration analysis would be necessary.
Section II: isolation system parameter variation
From the results presented in Section I it was concluded that the basic effects of varying the isolation
system parameters could be adequately investigated using a single earthquake (El Centro, N.S., May
1940) and a single structure ( N = 6, TI(UI) = 0.6 s). This is the approach taken in this section.
Now an assumption to date in the analysis has been that the isolation system has infinite vertical stiffness.
Generally this is not the case and, hence, the structure will be free not only to translate horizontally on
its flexible supports but to move vertically and to rock on them as well. Although straight vertical motions
are not expected to increase structure shears significantly, rocking of the structure does indeed increase
them as shown in Figure 6, where TR is the rigid body rocking period of the structure on its isolation system (of k , = 5.0Wm-l, k , = 1.OWm-l and F = 0.05 W ) . It can be shownz4that for structures having equal
floor masses, a rectangular floor plan and the isolation system elastic support distributed uniformly over
the base area, the rocking period is approximately given by

where T, is the period of the total structure mass on the total vertical stiffness of the isolation system and
h/b is the ratio of structure height to structure width in the direction of the rocking motions. From equation (2) and Figure 6 it can be seen that to keep rocking effects from becoming significant, the vertical
stiffness of the isolation system ought to be as large as possible but certainly large enough so that the rigid
Storey
Number
6

(Max. Interstorey Shear Force) /W

Figure 6. Maximum interstorey shear forces in six-storey, base-isolated structures with rigid body rocking periods of TI.
El Centro (N.S., May 1940) excitation

body rocking period of the structure is below the region where the forcing earthquake has its dominant
energies. Now, because of the virtual incompressibility of rubber, the vertical stiffness of a laminated
rubber pad of a given height is directly proportional to the square of the number of interleaving steel plates
and this can be arbitrarily increased to provide any desired vertical stiffness, without altering the pads
horizontal shear stiffness. Thus rocking effects can be eliminated to any degree required whether rubber
pads or roller bearings (infinite vertical stiffness) are used to isolate the structure. The remainder of this
work therefore assumes that rocking motions have been eliminated from the isolation system, i.e. TR is
set equal to zero.
Figure 7 shows the responses of an identical set of six-storey structures when mounted on a variety of
isolation systems and excited by the El Centro earthquake motions. The isolation systems post-yield stiffnesses, k,, were held constant while the elastic stiffnesses, k,, and yield force, F, were varied. The symbol
0 on the plots of Figure 7 (and those of Figure 8) locates the set of isolation system parameters used in

56 1

EARTHQUAKE PROTECTION OF MULTI-STOREY SHEAR STRUCTURES

SA+

.zc
.I5
.I0

.o:

(cm)

20 15 -

1.25-

10 -

2.01

*&
L
g
<-

5.0-

1.01

5I

.01 ,025

.05

.I0

.I5

.01 .025

.05

25 .I0

.I5

F/W
(C)

(d)

Figure 7. The effect of the isolation system yield force and elastic stiffness (at constant post-yield stiffness) on the responses
of an isolated structure (N = 6 , TI (UI) =0.6 s) being excited by the El Centro (N.S., May 1940) earthquake motions.
(a) Maximum interstorey shear force, (b) Shear distribution bulge-defining parameter, (c) Maximum base-foundation
displacement and (d) Residual plastic offset of the structure base from the foundation

Section 1 and serves as a reference point for a visual comparison of the benefits or otherwise to be obtained
from varying the isolation system parameters from these previously considered values.
Now it is intuitively obvious that to limit shear responses, the isolation system must have good energyabsorption properties for the strength of forcing expected. This, however, will not be the case if the elastic
stiffness of the isolation system is close to its post-yield stiffness or if its yield force is very small. In either
case the isolation system will behave essentially as a linear spring whose stiffness is close to that of the
isolation systems post-yield stiffness. The effect of such inappropriate isolation systems can be seen in
the increased structure shear forces and base displacements, for low elastic stiffness and yield force, in
Figures 7(a) and 7(c). A similar effect occurs when the yield force is too large. The isolation system behaves
essentially as a linear spring whose stiffness is close to that of the fully elastic isolation system. It is worthy
of note that of all the isolation systems considered in Figure 7, the reference system (0)
considered in
Section I provides the lowest maximum structure shear force. However, although the maximum structure
shear force cannot be reduced below its Section I value by adjusting k, and/or F, it can be seen from
Figure 7(b) that it is possible to reduce greatly the bulge of the structures shear distribution by such adjustments. Comparison of the shear distributions in Figure 9 for the reference (k, = 5.0 W m-l, k , = l.OWm-l,
F = 0.05W) and the A isolation system ( k , = 2.5Wm-l, k , = I*OW, F = 0.05W) illustrates the degree
to which the shear forces in the upper levels of the structure can be reduced without greatly increasing
the base shear force. B , is reduced from 0.38 to 0.06.
The residual plastic offsets of the structure were determined by halting the earthquake excitation after
significant ground motion had finished (30s) and by then allowing the structure vibrations to decay to
rest. These offsets, as shown in Figure 7(d), were less than 2 cm for all the isolation systems investigated.

562

D. M. LEE A N D I. C. MEDLAND

Figure 8 shows how the structure responses change as the yield force and post-yield stiffness of the isolation system are varied but its elastic stiffness is held constant. It can be seen that reducing the post-yield
stiffness, k , can greatly reduce the maximum shear force but at the expense of increasing the bulge of the
shear distribution, the maximum displacement of the base and the expected residual plastic offset.
The general observation can be made from Figure 8(a) that by reducing both the yield force and the postyield stiffness together, the structure shear forces can be reduced to any desired level. There are, however,
practical problems in constructing an isolation system of very low post-yield stiffness and it can be seen
from Figure 8(c) and (d) that greatly increased base displacements and residual offsets would be expected
for such systems. If the yield force is too low, the isolation system will yield during strong winds or minor
earthquakes, prematurely fatiguing its components. Also if the elastic stiffness of the isolation system is
too low, then the structure could have its fundamental period lying in the region where wind loading could
cause unpleasant resonant oscillations and even eventual yielding of the isolation system. The authors
consider that for most applications k, = 2.5 W m-l, k , = 0.5 W m-l and F = 0.025 W would be the lower
limits imposed on the isolation system parameters by the above considerations.

s/ w
.20-

.I5

.I0

.05 -

I I

.01 .025

.05

.I0

.I5

0.1 .025

.05

.I5

F/W
(a)

dB
(cm)

20

-0.05

15

10
5

.01 .025

.05

.I0

.I5

F/W
(C)

Figure 8. The effect of the isolation system yield force and post-yield stiffness (at constant elastic stiffness) on the response
of an isolated structure ( N = 6 , TI (UI) =0.6s) being excited by the El Centro (N.S., May 1940) earthquake motions.
(a) Maximum interstorey shear force, (b) Shear distribution bulge-defining parameter, (c) Maximum base-foundation
displacement and (d) Residual plastic offset of the structure base from the foundation

With the above constraints in mind, four sets of isolation system parameters (as identified and listed
in Table 1) were selected as defining isolation systems which gave increasing shear protection to a structure
while keeping the base displacements within reasonable limits. The shear distributions for the six-storey
structure mounted on each of these isolation systems and excited by the El Centro motions are given in
Figure 9. The numbers on each distribution denote the maximum base displacements in centimetres.
Examining Figure 9 it can be seen that the shear forces arising in a structure excited by a large earthquake

563

EARTHQUAKE PROTECTION OF MULTI-STOREY SHEAR STRUCTURES

Table I. Parameters of selected, bilinear hysteretic base isolation systems


Isolation system
identification
Reference
(cf. Section I)
A
B
C
D

(m-I)

(Post-yield stiffness, k 2 ) /W
(m-9

5.0

1 .o

0.05

2.5
2.5
2.5
2.5

1 .o

0.05

0.5
0.05

0.05
0.05
0.025

(Elastic stiffness, kl)/ W

0.05

(Yield force, F)/ W

such as El Centro can be reduced to very low levels ( 0.05 W ) for what are easily accommodated levels
of base displacement ( 15 cm or 6 in). It ought to be remembered that the ground accelerations employed
here have a peak value of 0.35 g and could induce peak accelerations of the order of 0.8 g in an unisolated
elastic structure, i.e. shear forces of the order of 0.8W; a factor of 16 times greater than those for the isolated structure.
N

(Mox. Interstorey Sheor Force)/W

Figure 9. Shear distributions of six-storey structures when mounted on selected isolation systems and excited by the El
Centro (N.S., May 1940) earthquake motions

Section III: a close examination of four isolation systems


The results of the preceding section have indicated how the responses of isolated structures are affected
by the parameters of their isolation systems. These results were, however, derived for a single structure
being excited by a single earthquake and consequently provide only a first basis on which to debate the
efficacy of different isolation systems. The isolation systems examined in Section I1 might therefore produce
different structure responses if different earthquake forcing or if structures of different periods were used.
Certainly one would expect the scale of the shear response, base displacement and residual offset to increase
if a larger earthquake were employed. Specifically one might ask how effective the isolation systems A to
D of Table I would be in shielding structures of different fundamental periods from a variety of earthquakes. Can correlations similar to that of Figure 5 be found between the responses of structures mounted
on these isolation systems and simple parameters of the forcing earthquakes? How, in more generality, do
the displacements and residual plastic offsets of the structure base increase as the isolation systems become
more 'flexible' (i.e. as the shear forces transmitted through the isolation system decrease) ? To answer these
questions six-storey structures of unisolated fundamental period, TI (UI) = 0-3, 0.6 and 0-9 s were mounted
on each of the four isolation systems A to D and excited by eight of the previously specified set of earthquake records. These eight records were the artificial B1, Cl and D1 records and the Caltech digitized
records of the El Centro (N.S.-AOOl), Parkfield (N65OE-BO33, N85"E-B034), and Pacoima Dam
(S 14"W and N76"W-C04 1) earthquakes.

564

D. M. LEE AND I. C. MEDLAND

For the structures mounted on a given isolation system the maximum shear response, S, generated during
each of the earthquakes was linearly regressed against selected values of the response spectra of the forcing
earthquakes. These response spectra values were selected at the fundamental T, (E), and second mode,
T,(E), periods of the structure when mounted on the elastic isolation system and when mounted on the
post-yield isolation system, TI (PY) and T , (PY). (The former periods are those of the structure when
mounted on an elastic spring of stiffness, k,, while the latter are those of the structure when mounted on
an elastic spring of stiffness k,.) Linear least squares regression fits were made between S and response
spectra values at Tl (E) and T , (E), both individually and together, and between S and response spectra
values at Tl (PY) and T , (PY), again both individually and together. In each case the 5, 10 and 20 per cent
damped velocity and acceleration response spectra were used, making a total of 36 regression fits attempted
for each isolation system. It was found that for all four isolation systems, S was most closely correlated
with data from the 10 per cent damped velocity response spectrum of the forcing earthquake, although at
oscillator periods of Tl (PY) for systems A and B but at oscillator periods of T , (PY) for systems C and
D. This close correlation with data at different oscillator periods for different isolation systems can be
understood in the light of Figure 9 where it is apparent that isolation systems A and B produce first mode
dominated shear distributions whereas isolation systems C and D produce shear distributions where
second mode effects are sufficiently dominant to control the maximum structure shear force (i.e. the base
shear is not the maximum shear). Plots of S versus the best earthquake parameter for each of isolation
systems A, B, C and D are superimposed in Figure 10 with their abscissa values scaled so that each plot
has the same horizontal extent. (The numbers in brackets above each line denote the factor by which the
abscissa values have been multiplied before being plotted.) In order to avoid confusion on the graph, the
extents of the 95 per cent confidence intervals are indicated by the vertical bars at the right-hand end of
each regression line. No distinction has been made between plot points from structures of different periods,
only between structures on different isolation systems. Examining Figure 10 it can be seen that isolation
systems A to D reduce the maximum structure shear forces much as expected from the results of Section 11,
so justifying the technique of picking isolation systems on the basis of their effect with one structure and
a single earthquake.

sl

Isoln.

S vrs

sp.vel

(I0 %d ), (It;

0.3

T; (PY)

0.2

T2(PY)

0.I

T2(PY)

0.2

0.4

0.6 0.0

Spectrol Velocity

10
.

1.2

1.4

(loo/,domped) (m/s)

1.6

Figure 10. Plots of the best correlations between the maximum interstorey.shear force in selectively isolated structures
and response spectrum data of the forcing earthquakes

Values of the bulge-defining parameter, B,, for the three structures on each of the four isolation systems
were regressed against response spectrum data in the same fashion as for S but no strong correlations were
discovered. B, did, however, depend strongly on both the isolation system type and the unisolated fundamental period of the structure, as can be seen in Table 11. In general the bulge of the structure shear distribution decreased as Tl (UI) decreased and increased as the flexibility of the isolation systems increased.
(Isolation system D is, for example, considered more flexible than A because it transmits lower shear
forces into the structure proper.)

565

EARTHQUAKE PROTECTION OF MULTI-STOREY SHEAR STRUCTURES

Table 11. Means (BD)and standard deviations (u) of the 'bulge defining parameter', BD, for structures of unisolated fundamental period TI (UI), when mounted on isolation systems A to D of Table I
Tl (UI) = 0.3

Isolation system
A

BD
0.062
0.046
0.253
0.344

B
C
D

0.044
0.026
0.094
0.103

Tl (UI) = 0.6
&I

0-132 0.118
0.249 0.150
0.428 0.127
0.489 0.115

Tl (UI) = 0.9

BD

0.298
0.372
0.507
0.597

0.096
0.132
0.130
0.077

Thus base isolation is of particular benefit to low period (i.e. rigid) structures as not only is the maximum
shear force reduced greatly from its unisolated value but the fraction of this shear distributed over the upper
storeys of the structure is also reduced (i.e. for low period structures, B, is generally smaller in isolated
than unisolated structures). Rigidity of the structure proper also ensures that little internal structural movement and hence little secondary damage occurs. Masonry structures, presently considered unsafe in seismic
regions, could become economic alternatives if mounted on a base isolation system.
For each isolation system the maximum structure base displacements, d,, were found to be most closely
correlated with the value of the 10 per cent damped velocity response spectrum of the forcing earthquake
at a period of Tl (PY). Such a result is not surprising as movement of the base is determined primarily by
the fundamental mode response of the structure, even if the maximum shear force, occurring above the
structure base (cf. C and D in Figure 9), is determined largely by second mode effects. Figure 11 shows,
for each of the isolation systems, regression plots of d, against values of the 10 per cent damped velocity
response spectra of the forcing earthquakes at oscillator periods of TI (PY). The dashed lines represent
approximate 95 per cent confidence intervals on the data. There was no significant variation in dB with Tl
(UI), for structures mounted on a given isolation system. An interesting feature of Figure 11 is that the
base displacements do not increase very significantly as the flexibility of the isolation systems increases
dB 6 0 -

(cm) 5o

dg 60 IIsoln. Syst. C.
(cm)

Isoln. Syst. 8.

40
30
20
10

/
.2

.4

.6 .8

1.0 1.2

1.4 1.6

Sp. Vel. (10%d), at TI(PY), (rns-l)

Sp.Vel.(IO%d), at TI(PY);(ms-')

(C)
(d 1
Figure 1 1 . Plots of the best correlation between the maximum base displacements in selectively isolated structures and
values of the 10 per cent damped velocity response spectrum (at Tl (PY) of the forcing earthquakes

566

D. M. LEE AND I. C. MEDLAND

(i.e. from A to B to C to D here). The average maximum isolation system displacements were, in fact,
16.2 cm, 17.3 cm, 19.5 cm and 19.4 cm for structures mounted on isolation systems A, B, C and D respectively, an increase of approximately 20 per cent. This increase is rather small when one considers that the
corresponding average maximum structure shear forces of 0.24 W , 0.17 W , 0.09 W and 0.06 W, respectively,
have been reduced by a factor of approximately 75 per cent. Thus the more flexible isolation systems do
not necessarily generate increases in base displacement in direct proportion to the decreases in shear force
which they provide. The accuracy with which the base displacements can be estimated from response
spectrum data does, however, decrease as the flexibility of the isolation system increases (cf. the widths
of the 95 per cent confidence intervals in Figure 11). Greater care would therefore be needed in designing
the more flexible isolation systems to ensure that excessive base displacement did not occur. Low postyield stiffness isolation systems could also allow greater than expected base displacements for earthquakes
with a prominent drift. Limiting stops or overrider restraining springs may be prudent for such installations.
The residual plastic offset, IS,,, of the structures on each of the isolation systems, displayed no correlation with any of the response spectrum parameters previously considered. There was, however, a weak
correlation between the residual plastic offset and the 20 per cent damped Spectral Intensity (S.I.) of the
forcing earthquake. (The 20 per cent damped S.I. is the area, between 0.1 and 2.5 s, under the 20 per cent
damped velocity response spectrum of the earthquake.) Linear least squares regression plots of ISpL versus
S.I. are given in Figure 12 where it can be seen that residual offsets tend to increase as the post-yield stiffness, k,, of the isolation system decreases (i.e. from A to B to C) and, for small k,, to decrease as the isolation system yield force decreases (i.e. from C to D). (This effect can be observed in Figure 8(d) for the
El Centro excited structure.) For a sufficiently high post-yield stiffness (e.g. isolation system A with k , =
1.0 W m-l) the residual plastic offset is very small (< 1 cm) and is effectively independent of the earthquake strength.

.4

.8

1.2

1.6

2.0

2.4

S.I. (209d); (m)


(a 1
SPL
20

Isoln. Syst.

D.

16 12

84I)

.8

1.2

1.6

2.0

2.4

Figure 12. Residual plastic offsets of selectively isolated structures plotted against the 20 per cent damped spectral intensity
of the forcing earthquake

EARTHQUAKE PROTECTION OF MULTI-STOREY SHEAR STRUCTURES

567

In Figures 10-12, no distinction has been made between data points arising from structures of different
unisolated fundamental period (0.3, 0.6 or 0.9 s). It is only in determining the shape of the isolated structures shear distribution that this pre-isolation period has any effect.
DISCUSSION

The analyses in this paper have been based on the El Centro earthquake, with larger and smaller earthquakes being used to indicate how the isolated structures responses vary with earthquake strength. If,
however, one were concerned with designing specifically for very large earthquakes such as the Pacoima
Dam record, a new data set like that of Section I1 ought to be computed using such a record. One would
still, however, be faced with the choice of trading off shear protection against the expected base displacements and residual offsets. If very large base displacements are to be accommodated, then roller bearing
supports or Teflon slip pads might be preferred over laminated rubber pads which can lose effective supporting area for large displacements. If rubber pads are used for the structure support, then their pIan
dimensions ought to be of the order of 10 times greater than their expected displacements.
An important feature of structures mounted on the more flexible isolation systems is the decreased
sensitivity of their maximum interstorey shear response to the strength of the earthquake. This effect can
be observed in Figure 10 where it can be seen that a given increase in the strength of some design earthquake will cause a smaller increase in shear for a structure mounted on isolation system D rather than
isolation system A. The more flexible isolation systems therefore provide a better shear force buffer against
unknown ground motions although the designer then faces the problem of allowing for the possibility of
very large base displacements. Obviously, a balance must be struck in each situation between the shear
reducing characteristics and the base displacement increasing characteristics of a given isolation system.
The plots of Section 111 (and those of Sections I and 11) are intended as a guide to the designer in striking
this balance.
CONCLUSIONS
(i) The maximum shear responses of multi-storey shear structures are greatly reduced by base isolation
and these reduced responses are effectively independent of the pre-isolation structure period.
(ii) The rocking stiffness of the isolation system should be sufficiently high in order to prevent structure
rocking motions from increasing the upper storey shear forces.
(iii) A detailed examination of the parameters defining the bilinear hysteretic isolation system revealed
that its post-yield stiffness is the most important parameter in determining the maximum shear response
of an isolated structure. In contrast, the elastic stiffness of the isolation system had little effect on the base
shear although both the elastic and the post-yield stiffness were important in determining the upper storey
shear forces. Low values of the isolation system yield force produced high base displacements but generally
low residual plastic offsets. Low yield forces also produced almost linear structure shear distributions.
(iv) Close relationships were established between the forced responses of isolated structures and selected
values of elastic response spectra of the forcing earthquakes. These relationships were quantified and
confidence limits derived for estimates of structure response using these relationships. Therefore, if statistical information is available on the earthquake motions expected at a given site, then statistical limits can
be placed on the expected response of a given structure on a given isolation system, without having to
perform an ensemble of forced vibration analyses.
(v) Very low shear responses can be achieved in structures subjected to extremely strong earthquake
forcing, if the designer is prepared to allow for the expected base displacements and the possibility of
having to reset residual plastic offsets following such an earthquake.
(vi) Because a base isolation system can be designed to allow a given structure to remain completely
elastic under the expected earthquake forcing, no reconstitution or repair of the structure proper should be
necessary after an earthquake. This expected reduction in damage could be used to obtain lower earthquake damage insurance premiums. Furthermore, because the structure proper is not expected to undergo

568

D. M. LEE AND I. C. MEDLAND

inelastic deformation, there will be little need for ductility in the beam-column connections, which should
allow a more simple detailing there and a general lowering of structural strength. In reinforced concrete
structures fewer stirrups would be needed. All of these features add up to a considerable saving in design
and construction costs which would help to offset, or even neutralize, the cost of installing a base isolation
system.
REFERENCES
1. M. Fintel and F. R. Kahn, Shock absorbing soft story concept for multistory earthquake structures, Proc. Am. Concr.
Inst. 66, 381-390 (1969).
2. A. K. Chopra, D. P. Clough and R. W. Clough, Earthquake resistance of buildings with a soft first storey, Eurthqu.
Eng Struct. Dyn. 1, 347-355 (1973).
3. J. M. Kelly, J. M. Edinger and C. J. Derham, A practical soft story earthquake isolation system, Rep. No. UCBIEERC
77/27, Univ. of California, Berkeley (1977).
4. G. C. Delfosse, The Gapec system: a new highly effective aseismic system, Proc. 6th W d Conf. Earthq. Engng, New
Delhi, India 3, 163 (1977).
5 . C. Plichon and F. Jolivet, Aseismic foundation systems for nuclear power plants, Proc. S.M.I.R.T. Conf., London,
England, Paper No. C190/1978 (1978).
6. R. I. Skinner, J. L. Beck and G. N. Bycroft, A practical system for isolating structures from earthquake attack, Earthqu.
Eng Struct. Dyn. 3, 297-309 (1975).
7. R. I. Skinner and G. H. McVerry, A practical system for isolating nuclear power plants from earthquake attack,
Nucl. Engng and Design 36, 287-297 (1976).
8. M. J. N. Priestley, R. L. Crosbie and A. J. Carr, Seismic forces in base isolated masonry structures, Bull. N.Z. Nat.
SOC.Earthq. Engng, 10, 55-68 (1977).
9. D. M. Lee and I. C. Medland, Estimation of base isolated structure responses, Bull. N . Z . Nut. SOC.Eurthq. Engng,
11, 234-244 (1978).
10. D. M. Lee and I. C. Medland, Base isolation-An historical development and the influence of higher mode responses,
Bull. N . Z . Nut. SOC.Earthq. Engng, 11, 219-233 (1978).
11. J. L. Chameau and H. C. Shah, Dynamic testing of Gapec isolators, John A. Blume Earthq. Engng Center, Stanford
University, California (1978).
12. J. M. Eidinger and J. M. Kelly, Experimental results of an earthquake isolation system using natural rubber bearings,
Rep. No. CJCBIEERC 78/03, Univ. of California, Berkeley, California (1978).
13. R. G. Tyler, Dynamic tests on laminated rubber bearings, Bull. N . Z . Nut. SOC.Eurthq. Engng, 10, 143-150 (1977).
14. R. G. Tyler, Dynamic tests on P.T.F.E. sliding layers under earthquake conditions, Bull. N.Z. Nat. SOC.Eurthq.
Engng, 10, 129-138 (1977).
15. R. I. Skinner, J. M. Kelly and A. J. Heine, Energy absorption devices for earthquake resistant structures, Eurthqu.
Eng Struct. Dyn. 3, 287-295 (1975).
16. R. G. Tyler, A tenacious base isolation system using round steel bars, Bull. N.Z. Nat. SOC.Eurthq. Engng, 11, 273281 (1978).
17. R. G. Tyler, Tapered steel energy dissipators for earthquake resistant structures, Bull. N.Z. Nut. SOC.Eurthq. Engng,
11, 282-294 (1978).
18. W. H. Robinson and L. R. Greenbank, An extension energy absorber suitable for the protection of structures during
an earthquake, Eurthqu. Eng Struct. Dyn. 4, 251-259 (1976).
19. W. H. Robinson and A. G. Tucker, A lead rubber shear damper, Bull. N.Z. Nat. SOC.Earthq. Engng, 10, No. 3 (1977).
20. L. M. Megget, Analysis and design of a base isolated reinforced concrete frame building, Bull. N.Z. Nut. SOC.Eurthq.
Engng, 11, 245-254 (1978).
21. Digitized strong motion earthquake accelerograms in Japan, Pub. Assoc. for Sci. Info. in Japan (1972).
22. P. C. Jennings, G. W. Housner and N. C. Tsai, Simulated earthquake motions, Report, Earthq. Engng Res. Lab,
Calif. Inst. of Tech., Pasadena, Calif. (1968).
23. Strong motion earthquake accelerograms-digitized and plotted data, Vol. 11, Corrected accelerograms and integrated ground velocity and displacement curves, Part A-EERL 71-50 (1971), Part E-EERL 72-50 (1973), Part CEERL 72-51 (1973), Part D-EERL 72-52 (1973), Calif. Inst. of Tech., Pasadena, Calif.
24. D. M. Lee, The effect of base isolation on multi-story shear structures, Ph.D. Thesis, Dept. of Theor. and Appl. Mech.,
University of Auckland, Auckland, New Zealand, 1978.

Potrebbero piacerti anche