Seismic Protection of Structures with Base Isolation
1. Concrete Solutions 09 Paper 7a-3
Seismic Protection of Structures with Modern Base Isolation Technologies
Luis Andrade1 and John Tuxworth2
1
Senior Structural Engineer, Green Leaf Engineers
2
Director, Green Leaf Engineers
Synopsis: Increased resistance to earthquake forces is not always a desirable solution for buildings
which house contents that are irreplaceable or simply more valuable than the actual primary structure (eg
museums, data storage centres, etc). Base isolation can be employed to minimize inter-story drifts and
floor accelerations via specially designed interfaces at the structural base, or at higher levels of the
superstructure.
This paper presents the design comparison of two isolation systems (lead-rubber bearings, and friction
pendulum bearings) for a five-story reinforced concrete framed building. The response of the base-case,
fixed-structure, and isolated systems is compared for dynamic analysis to actual historical records for five
significant seismic events.
Keywords: bearing, concrete, damping, dissipation, drift, isolation, inter-storey, lead-rubber, pendulum,
seismic.
1. Introduction
Conventionally, seismic design of building structures is based on the concept of increasing resistance
against earthquake forces by employing the use of shear walls, braced frames, or moment-resistant
frames. For stiff buildings these traditional methods often result in high floor accelerations, and large inter-
story drifts for flexible buildings. With both scenarios building contents and nonstructural components may
suffer significant damage during a major event, even if the structure itself remains basically intact.
Obviously this is an undesirable outcome for buildings which house contents that are irreplaceable, or
simply more costly and valuable than the actual primary structure (eg museums, data storage centers,
etc).
The concept of base isolation is increasingly being adopted in order to minimize inter-story drift and floor
accelerations. In this instance the control of structural forces and motion is exercised through specially
designed interfaces at the structural base — or potentially at a higher level of the superstructure — thus
filtering out the actions transmitted from the ground. The effect of base isolation is to essentially uncouple
the building from the ground.
This paper presents the design comparison of two isolation systems — Friction Pendulum System (FPS)
and Lead-Plug Bearings (LPB) — for a five-story reinforced concrete framed building. The response of the
fixed-base structure is compared to base-isolated cases for five different historical time-history records for
significant earthquake events.
2. Base Isolation Systems
There are two common categories of large-displacement base (or seismic) isolation hardware: Sliding
Bearings and Elastomeric Bearings. This paper considers Friction Pendulum Systems (FPS) and Lead-
Plug-Bearings (LPB), which belong to the first and second categories respectively.
2.1 Friction Pendulum System (FPS)
A FPS is comprised of a stainless steel concave surface, an articulated sliding element, and cover plate.
The slider is finished with a self-lubricating composite liner (e.g. Teflon). During an earthquake, the
articulated slider within the bearing, travels along the concave surface, causing the supported structure to
move with gentle pendulum motions as illustrated in Figure 1(a) and 1(b). Movement of the slider
1
2. Concrete Solutions 09 Paper 7a-3
generates a dynamic frictional force that provides the required damping to absorb the earthquake energy.
Friction at the interface is dependent on the contact between the Teflon-coated slider and the stainless
steel surface, which increases with pressure. Values of the friction coefficient ranging between 3% to 10%
are considered reasonable for a FPS to be effective, Wang (1).
The isolator period is a function of the radius of curvature (R) of the concave surface. The natural period is
independent of the mass of the supported structure, and is determined from the pendulum equation:
T = 2π R / g (1)
where g is the acceleration due to gravity.
The horizontal stiffness (KH) of the system, which provides the restoring capability, is provided by:
kH = W / R (2)
where W is the weight of the structure.
The movement of the slider generates a dynamic friction force that provides the required damping for
absorbing earthquake energy. The base shear V, transmitted to the structure as the bearing slides to a
distance (D), away from the neutral position, includes the restoring forces and the friction forces as can be
seen on the following equation, where μ is the friction coefficient:
W
V = μW + D (3)
R
The characterised constant (Q) of the isolation system is the maximum frictional force, which is defined as:
Q = μW (4)
The effective stiffness (keff) of the isolation system is a function of the estimated largest bearing
displacement (D), for a given value of μ and R, and is determined by:
μW W
k eff = V / D = + (5)
D R
A typical hysteresis loop of a FPS can be idealized as shown in Figure 1(c).
Force
Vmax
1
Q kH
keff
(a) 1 Dmax
Displacement
(b)
(c)
Figure 1. Motion in a FPS (a) initial condition, (b) displaced condition at maximum displacement,
(c) Idealized Hysteresis Loop of a FPS
The dissipated energy (area inside the hysteretic loop) for one cycle of sliding, with amplitude (D), can be
estimated as:
E D = 4 μWD (6)
Thus the damping of the system can be estimated as:
2
3. Concrete Solutions 09 Paper 7a-3
ED 2 μ
β= = (7)
4πk eff D 2
π D/R+μ
2.2 Lead-Plug Bearings (LPB)
The elastomeric LPB which are generally used for base isolation of structures consist of two steel fixing
plates located at the top and bottom of the bearing, several alternating layers of rubber and steel shims,
and a central lead core as shown in Figure 2(a). The elastomeric material provides the isolation
component with lateral flexibility; the lead core provides energy dissipation (or damping), while the internal
steel shims enhance the vertical load capacity whilst minimizing bulging. All elements contribute to the
lateral stiffness. The steel shims, together with the top and bottom steel fixing plates, also confine plastic
deformation of the central lead core. The rubber layers deform laterally during seismic excitation of the
structure, allowing the structure to translate horizontally, and the bearing to absorb energy when the lead
core yields.
The nonlinear behavior of a LPB isolator can be effectively idealized in terms of a bilinear force-deflection
curve, with constant values throughout multiple cycles of loading as shown on Figure 2(b).
Force
Vmax
1
kd
Q
keff
ki 1
Dmax
Dy
Displacement
(a) (b)
Figure 2. LPB isolator (a) components, (b) Idealized Hysteresis Loop of a LPB
The natural period of the isolated LPB system is provided by:
W
T = 2π (8)
k eff g
The characterised strength (Q) is effectively equal to the yield force (Fy,) of the lead plug. The yield stress
of the lead plug is usually taken as being around 10MPa. The effective stiffness (keff ) of the LPB, at a
horizontal displacement (D) being larger than the yield displacement (Dy) may be defined in terms of the
post-elastic stiffness (kd,) and characteristic strength (Q), with the following equation:
k eff = k d + Q / D (9)
As a rule of thumb for LPB isolators, the initial stiffness (ki) is usually taken as 10 x kd , Naeim et al (2).
The energy dissipated for one cycle of sliding, with amplitude (D) can be estimated as:
E D = 4Q ( D − D y ) (10)
Following on from this assumption, it has been shown by Naeim et al (2) that the effective percentage of
critical damping provided by the isolator can be obtained from:
3
4. Concrete Solutions 09 Paper 7a-3
ED 2 Q ( D − Q / 9k I
β= = (11)
4πk eff D 2
π (k i D + Q) D
3. Model-building Configuration
A reinforced concrete moment-resisting frame was adopted as the structural system for the analysis
building. Figure 3 (a) and 3(b) show the structural configuration of the building in plan.
(a) (b)
Figure 3. Structural configuration plans (a) 1st to 3rd floors. (b) 4th and 5th floors.
Self weight of the structure was based on a concrete density (γ ) = 24 kN/m3. Super-dead loads of 1 kN/m2
was also applied to represent floor finishes, and 140 mm thick, 2.5-m high hollow masonry partitions with
a density of (γ ) = 15 kN/m3 were considered to contribute as a line-load along beams of 4.9 kN/m. The
imposed (live) load applied in each floor was taken as 2 kN/m2. Story heights were taken as 3 m.
The Universal Building Code was considered in relation to seismic classification and variables, so as to
enable consistency of symbols and nomenclature throughout the paper. Most international standards
including AS 1170.4:2007 are either based on, or align significantly with, UBC 1997(3). It was assumed
that the building ‘model’ was located in a Seismic Zone 4 of source Type A, and rests on a soil profile
Type C.
4. Design Parameters
According to Mayes et al (4), an effective seismic isolation system should have the following
characteristics:
• sufficient horizontal flexibility to increase the structural period and accommodate spectral demands of
the installation (except for very soft soil sites),
• sufficient energy dissipation capacity to limit displacement to a practical level,
• adequate rigidity to enable the building structure to behave similarly to a fixed base building under
general service loadings.
As recommended by both Naeim et al (2) and Mayes et al (4), a target period (T) of 2.2 seconds was
adopted for the isolated structure — approximately 3 times the fixed-base fundamental period (TF ) of 0.7
seconds.
Following UBC 1997, the target design displacement can be calculated as:
4
5. Concrete Solutions 09 Paper 7a-3
( g / 4π 2 )C VD T
DD = (12)
BD
where CVD is a seismic coefficient, and BD is a damping coefficient which is a function of the effective
damping β.
From UBC 1997 Table 16-R, CVD = 0.56. An affective damping of 15% was assumed for both LPB and
FPS — to be confirmed at the end of the design. From Equation 12, the design displacement = 220 mm.
The effective stiffness for both bearing types was calculated following the formulas presented previously.
Properties including damping, hardness, modulus of rigidity, modulus of elasticity and poisons ratio (for
LPB), and friction coefficient (for FPS) were adopted from manufacturer’s data.
As the performance of LPB isolators is weight dependant, three different sizes were incorporated in the
model. The positions nominated in Figure 5 were adopted to promote an economical design. Final design
parameters and details for each isolator type are provided following.
Detailed design calculations have been omitted for clarity, however iterative calculation is required to
ascertain effective stiffness and effective damping as both are typically displacement dependent.
Figures 4(a) & 4(b) display cross-sectional details for isolator characteristics summarised in Tables 1 and
2 respectively.
R=1200mm
(a) (b)
Figure 4. Geometrical characteristics of Base Isolators (a) FPS. (b) LPB Type A
Table 1. Design Parameters of FPS isolators.
Symbol Value Nomenclature
T (sec) 2.2 (Design Period)
β (%) 15 (Effective damping)
BD 1.38 (Damping factor)
DD (mm) 220 (Design displacement Eq. 12)
R (mm) 1200 (radius of curvature, calculated from Eq. 1)
μ 0.057 (friction coefficient)
(Force reduction factor, UBC 1997 Table A-16-E, Concrete special moment
RI 2.0
resisting frame)
W (kN) 7318 (Total weight of the building)
Keff (kN/m) 7961 (Total effective stiffness Eq. 5)
kH (kN/m) 6085 (Non-linear stiffness Eq. 2 )
31033
ki (kN/m) (Elastic stiffness, taken as 51kH)
0
Q (kN) 416 (Frictional force Eq. 4)
Dy (m) 1.4 (Yield displacement calculated as Q / ( ki- kH )
β (%) 14.9 (Check of assumed effective damping Eq. 7)
5
6. Concrete Solutions 09 Paper 7a-3
Table 2. Design Parameters of LPB isolators.
Parameter Value Nomenclature
T (sec) 2.2 (Design period)
β (%) 15 (Effective damping)
BD 1.38 (Damping factor)
DD (mm) 220 (Design displacement Eq. 12)
G (MPa) 0.45 (Shear modulus)
T (sec) 2.2 (Design period)
Isolator Nomenclature
Parameter
Type A Type B Type C
Wi (kN) 1030 740 510 (Axial load on isolator)
Keff (kN/m) 840 604 416 (Effective stiffness calculated from Eq. 8)
ED (kN-m) 38.9 28.0 19.3 (Global energy dissipated per cycle, calculated from Eq. 11)
Q (kN) 43.9 31.5 21.7 (Short term yield force, calculated form Eq. 10)
Kd (kN/m) 642 461 318 (Inelastic stiffness, calculated form Eq.9)
Ki (kN/m) 6422 4614 3180 (Elastic stiffness, taken as 10kd)
Kd / Ki 0.10 0.10 0.10 (Stiffness ratio)
Dy (mm) 7.6 7.6 7.6 (Yield displacement, calculated as Q/9 kd)
Fy (kN) 48.8 35.0 24.1 (Yield Force calculated as kiDy)
Figure 5. Location of LPB isolators Type A, B and C.
5. Modal Analysis
SAP2000 structural analysis software is capable of Time History Analysis, including Multiple Base
Excitiation. SAP2000 facilitates the dynamic modeling of base isolators as link elements, which can be
assigned various stiffness properties. This stiffness values for both FPS and LPB isolators were calculated
as detailed in previous sections of this paper. Calculations associated with the following summary and
totaling some one-hundred pages have been excluded from the paper.
Table 3 provides the fundamental period for the three cases studied: structure with fixed base; with FPS
isolators; and with LPB isolators, as derived from an SAP2000 modal analysis. It can be seen that the
periods obtained for both types of isolator are close to the target period (T = 2.2 sec) recommended by
Naeim et al (2) and Mayes et al (4). Figure 6 shows the shape of the first mode of vibration for the 3
models. In addition to influencing fundamental period Figure 6 shows the isolators’ influence on modal
shape.
Table 3. Fundamental Periods
Model Fundamental Period, T (sec)
Fixed Base 0.73
LPB 2.23
FPS 2.05
6
7. Concrete Solutions 09 Paper 7a-3
(a) (b) (c)
Figure 6. First mode of vibration for (a) fixed base building, (b) FPS isolated building and (c) LPB
isolated building.
6. Time History Analysis
A nonlinear analysis was carried out in SAP2000 in order to test the response of the structural systems,
and to validate isolator functionality. The models were subjected to the following historical seismic time-
history records:
• 1940 Imperial Valley Earthquake, El Centro Record (Richter Scale 7.1),
• 1979 Imperial Valley Earthquake, El Centro Record, Array #5 (Richter Scale 6.4),
• 1989 Loma Prieta Earthquake, Los Gatos Record (Richter Scale 7.1),
• 1994 Northridge Earthquake, Newhall Record (Richter Scale 6.6),
• 1995 Aigion Earthquake, Greece (Richter Scale approx. 5)
A seismologist is of invaluable assistance when selecting applicable time-histories, however guidance for
selecting scaling records can be gleaned from codes, Kelly (5). The events chosen for consideration in
this paper represent several of the major earthquakes in recorded history, with the 1995 Aigion
Earthquake in Greece being of similar magnitude to the Newcastle earthquake of 1989 (Richter Scale 5.6)
Figure 7 shows maximum response values for each of the earthquake records for roof acceleration,
elastic base shear, inter-storey drift, and isolator displacements.
Maximum roof acceleration is dominated by the 1989 Loma Prieta earthquake record which yields a value
of about 36 m/sec2 for the fixed base structure, while for the isolated structures is in the order of 8.5
m/sec2 (76% reduction) (see Figure 7(a)).
Maximum elastic base shears are dominated also by the 1989 Loma Prieta earthquake. An elastic base
shear of approximately 120%W (where W is the building’s weight) for the fixed base building is reduced to
35%W (68% reduction) and 45%W (63% reduction) for LPB and FPS isolators respectively (see Figure
7(b)).
Maximum Inter-storey drifts for fixed base and isolator cases are again generated by the 1989 Loma
Prieta Earthquake, with values of about 129mm for the fixed base structure and 25mm (81% reduction)
and 35mm (73% reduction) for LPB and FPS respectively (see Figure (c)). The drift ratio derived for Level-
1 of the fixed base structure is 4.3%, about twice the maximum limit of 2% imposed by the UBC 1997. The
FPS isolated structure displays a value of 1.15% which is well under the limit.
Figure 7(d) shows maximum isolator displacements in the order of 473mm and 469mm. It can be seen in
Figure 7(e) that these values are round 215% of the isolator design displacement of 220 mm, indicating
that both isolator systems would fail during the 1989 Loma Prieta Earthquake and 1994 Northridge
Earthquake.
7
8. Concrete Solutions 09 Paper 7a-3
Force-Displacement hysteresis loops for the FPS and LPB isolator (Type A), as subjected to the 1989
Loma Prieta earthquake record, are provided in Figures 8(a) and 8(b). These curves follow the
mathematical models presented in section 2 of this paper. Elastic and post-elastic stiffness can be
obtained as the slopes of the first two initial segments.
Roof Acceleration Elastic Base Shear
40 140
35
LBS LRB
120
Acceleration (m/sec/sec)
FPS FPS
30
V / W (%)
Fixed Base 100
25 Fixed Base
80
20
60
15
40
10
20
5
0 0
1940 El 1979 El 1989 Loma 1994 1995 Aigion 1940 El 1979 El 1989 Loma 1994 1995 Aigion
Centro Centro Prieta Northridge Centro Centro Prieta Northridge
Earthquake Record Earthquake Record
(a) (b)
1st Floor Inter - Story Drift Isolator Displacement
140 500
LRB LRB
450
Isolator Displacement (mm)
120 FPS
FPS 400
100 350
Fixed Base
300
Drift (mm)
80
250
60 200
40 150
100
20
50
0 0
1940 El 1979 El 1989 Loma 1994 1995 Aigion 1940 El 1979 El 1989 Loma 1994 1995 Aigion
Centro Centro Prieta Northridge Centro Centro Prieta Northridge
Earthquake Record Earthquake Record
(c) (d)
Time History Displacement / Design Value
250%
LRB
FPS
200%
150%
100%
50%
0%
1940 El 1979 El 1989 Loma 1994 1995 Aigion
Centro Centro Prieta Northridge
Earthquake Record
(e)
Figure 7. Comparison of Response to the 5 earthquake records (a) roof acceleration, (b) elastic
base shear (c) 1st floor inter-story drift, (d) isolator displacement, (e) time history displacement /
design value utilization ratio.
8
9. Concrete Solutions 09 Paper 7a-3
The energy dissipated by each isolator is provided by the area inside each loop cycle. Effective damping
can be calculated using Equations 7 or 11 and compared with the assumed design value. Note that there
is seemingly an anomaly present in Figure 8 (a), as maximum ‘-ve’ deflection for the FPS isolator
corresponds to a reduction in elastic base shear. This anomaly was evident only for the Loma Prieta
earthquake, and further study is required to ascertain why this issue occurred.
(a) (b)
Figure 8. 1989 Loma Prieta Earthquake Record. Force-displacement hysteresis loops for (a) FPS
isolator (b) LPB isolator Type A.
Lead Plug Bearing Friction Pendulum System Fixed Base
40.0
Acceleration (m/sec/sec)
20.0
0.0
-20.0
-40.0
0 5 10 15 20 25 30
Time (sec)
Lead Plug Bearing Friction Pendulum System Fixed Base
9000
Base Shear (kN)
4500
0
-4500
-9000
0 5 10 15 20 25 30
Time (sec)
Lead Plug Bearing Friction Pendulum System
500
Isolator Displacement
250
(mm)
0
-250
-500
0 5 10 15 20 25 30
Time (sec)
Figure 9. Time-history results for 1989 Loma Prieta earthquake record. (a) Roof acceleration, (b)
elastic base shear, (c) isolator displacement.
9
10. Concrete Solutions 09 Paper 7a-3
Finally, time-history results for the Loma Prieta earthquake record are shown in Figure 9. It can be noticed
from Figures 9(a) and 9(b) how the response in time of the isolated system is significantly less than the
fixed base structure, specially between the first 10 to 15 seconds of the seismic excitation. Figure 9(c)
compares the two types of isolators’ lateral displacements, which appears to be less for the FPS.
7. Conclusions & Recommendations
It can be seen that resultant accelerations, elastic base shears and inter-storey drifts were all effectively
reduced by the adoption of Lead-Plug and Friction-Pendulum isolator systems, resulting in significant
improvement in modeled building performance, and a very likely minimisation of post-event losses. For
the ground conditions and sway-frame structural system adopted, LPB & FPS base isolation would be
excellent options to reduce structural and non-structural damage, and to protect building contents. Both
the LPB and FP systems provided a comparative reduction in roof level accelerations (up to 76%);
however the LPB provided the best reduction in elastic base shear, and inter-storey drift (at first floor). For
the adopted bearing characteristics, the FPS provided greatest control of isolator displacement — a
significant serviceability constraint with respect to boundary conditions.
Response of the isolated structural framing systems was dominated by the time-history record of the 1989
Loma Prieta Earthquake. The second highest intensity experienced by the test structure was due to 1994
Northbridge earthquake. The isolator design displacement (being a function of the nominated isolator
characteristics) of both systems was exceeded by these earthquakes, indicating alternate properties/sizes
would be required to accommodate higher intensity events.
Further work is recommended to establish applicability of these base-isolation systems for the common
braced-frame structural framing paradigm, and also to confirm suitability (or lack thereof) for high-rise
construction, and or use on deep alluvial soil strata as evident in Australian centers such as Newcastle.
8. References
1. Wang, Yen-Po, “Fundamentals of Seismic Base Isolation”, International Training programs for
Seismic Design of Building Structures.
2. Naeim, F. & Kelly, J. M., “Design of Seismic Isolated Structures: From Theory to Practice”, John
Wiley & Sons, Inc. 1999.
3. International Conference of Building Officials, ICBO (1997), “Earthquake Regulations for Seismic-
Isolated Structures”, Uniform Building Code, Appendix Chapter 16, Whittier, CA.
4. Mayes, R. & Naeim, F., “Design of Structures with Seismic Isolation”, Earthquake Engineering
Handbook, University of Hawaii, CRC Press, 2003.
5. Kelly, T. E., “Base Isolation of Structures Design Guidelines”, Holmes Consulting Group Ltd, July
2001.
10