FEMA 1051 15seismically Isolated Structures
FEMA 1051 15seismically Isolated Structures
Contents
15.1 BACKGROUND .......................................................................................................................... 5
15.1.1 Concept of Seismic Isolation ................................................................................................ 5
15.1.2 Types of Isolation Systems ................................................................................................... 5
15.1.3 Design Process Summary...................................................................................................... 6
15.2 PROJECT INFORMATION ......................................................................................................... 7
15.2.1 Building Description ............................................................................................................. 7
15.2.2 Building Weights ................................................................................................................ 12
15.2.3 Seismic Design Parameters ................................................................................................. 13
15.2.4 Structural Design Criteria ................................................................................................... 14
15.3 PRELIMINARY DESIGN OF ISOLATION SYSTEM............................................................. 15
15.3.1 Elastomeric Isolation System .............................................................................................. 15
15.3.2 Sliding Isolation System ..................................................................................................... 20
15.4 ISOLATION SYSTEM PROPERTIES ...................................................................................... 24
15.4.1 Overview ............................................................................................................................. 24
15.4.2 Nominal Properties and Testing λ-Factors .......................................................................... 25
15.4.3 Aging and Environmental λ-Factors ................................................................................... 30
15.4.4 Specification λ-Factors........................................................................................................ 31
15.4.5 Upper- and Lower-Bound Force-Deflection Behavior ....................................................... 31
15.5 EQUIVALENT LATERAL FORCE PROCEDURE ................................................................. 33
15.5.1 Procedure ............................................................................................................................ 33
15.5.2 Structural Analysis .............................................................................................................. 34
15.5.3 Limitation Checks ............................................................................................................... 43
15.6 DYNAMIC ANALYSES ............................................................................................................ 44
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
15-2
Chapter 15: Seismically Isolated Structures
Chapter 17 of ASCE/SEI 7-16 (hereafter the Standard) addresses the design of buildings that incorporate
a seismic isolation system. It defines load, design and testing requirements specific to the isolation
system and interfaces with the other chapters of the Standard for design of the structure above the
isolation system and of the foundation and structural elements below.
The Standard has evolved over the years to reflect the state of art and practice in the field. The latest
evolution of the Standard incorporates a number of major revisions including:
Added a new section titled Isolation System Properties for the explicit development of property
modification (λ) factors which are used to account for the variation in a bearing’s nominal
mechanical properties.
Revised provisions to reflect a MCER-only basis for analysis and design.
Expanded the range of applicability of the equivalent lateral force (ELF) procedure.
Incorporated a more realistic distribution of shear force over the building height considering the
period of the superstructure and the effective damping of the isolation system.
Revised and clarified the seismic ground motion selection and scaling criteria.
Made a number of additions and changes to the testing requirements, such as dynamic testing
sequences, criteria for bearings to be classified as similar, and requirements for quality testing of
100% of production bearings.
Expanded commentary to compliment the provisions.
This example uses the Standard to analyze and design a three-story emergency operations center (EOC)
located in a region of high seismicity, classified as Seismic Design Category D. The building is base-
isolated to achieve a higher level of seismic performance in a major earthquake. The focus of this
example is given to the analysis and design of the isolation system with two common types considered: 1)
elastomeric bearings and 2) sliding bearings. Arbitrarily, the elastomeric isolation system is carried
through to final design, however generally the analysis is similar for sliding systems. Although the facility
is hypothetical, it is of comparable size and configuration to actual base-isolated EOCs and is generally
representative of base-isolated buildings.
In addition to the Standard, the following documents are either referenced directly, provide background,
or are useful aids for the analysis and design of isolated structures:
Constantinou Constantinou, M. C., Kalpakidis, I., Filiatrault, A. and Ecker Lay, R. A. (2011).
et al. LRFD-Based Analysis and Design Procedures for Bridge Bearings and Seismic
Bearings, Technical Report MCEER-11-0004, Multidisciplinary Center for Earthquake
Engineering Research, Buffalo, New York.
Constantinou Constantinou, M. C., Whittaker, A. S., Kalpakidis, Y., Fenz, D. M. and Warn, G. P.
et al. (2007). Performance of Seismic Isolation Hardware under Service and Seismic
Loading, Technical Report MCEER-07-0012, Multidisciplinary Center for Earthquake
Engineering Research, Buffalo, New York.
Fenz and Fenz, D. M., Constantinou, M. C. (2006). “Behavior of Double Concave Friction
Constantinou Pendulum Isolator”, Earthquake Engineering and Structural Dynamics 2006, 35:1403-
1421.
15-3
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
Kalpakidis and Kalpakidis, I. V., Constantinou, M. C., (2008). “Effects of Heating and Load History
Constantinou on the Behavior of Lead-Rubber Bearings”, Technical Report MCEER-08-0027,
Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.
McVitty and McVitty, W. J., Constantinou, M. C. (2015). Property Modification Factors for Seismic
Constantinou Bearings: Design Guidance for Buildings, Technical Report MCEER-15-0005,
Multidisciplinary Center for Earthquake Engineering Research, Buffalo, New York.
NEHRP NEHRP (2015). Recommended Seismic Provisions for New Buildings and Other
Structures, FEMA P-1050.
NIST NIST (2011). Selecting and Scaling Earthquake Ground Motions for Performing
Response-History Analyses, NIST GCR 11-917-15.
Sarlis and Sarlis A. A. S., Constantinou, M. C. (2010). Modeling Triple Friction Pendulum
Constantinou Bearings in Program SAP2000, Technical Report released to the engineering
community.
Sarkisian et al. Sarkisian et al. (2012). Property Verification of Triple PendulumTM Seismic Isolation
Bearings.20th Analysis and Computation Specialty Conference, ASCE.
Thompson et al. Thompson, A. C. T., Whittaker, A. S., Fenves, G. L., and Mahin, S. A. (2000).
“Property modification factors for elastomeric seismic isolation bearings.” Proc., 12th
World Conf. on Earthquake Engineering, Auckland, New Zealand, 1–8.
Warn and Warn G. P., Whittaker, A. S. (2007). Performance Estimates for Seismically Isolated
Whittaker Bridges, Technical Report MCEER-07-0024, Multidisciplinary Center for Earthquake
Engineering Research, Buffalo, New York.
Wolff et al. Wolff, E. D., Ipek, C., Constantinou, M. C., Morillas, L. (2014). Torsional Response of
Seismically Isolated Structures Revisited Engineering Structures 59, 462-468
York and Ryan York, K., Ryan, K. (2008). Distribution of Lateral Forces in Base-Isolated Buildings
Considering Isolation System Nonlinearity. Journal of Earthquake Engineering.
15-4
Chapter 15: Seismically Isolated Structures
15.1 BACKGROUND
15.1.1 Concept of Seismic Isolation
The concept of seismic isolation involves increasing the fundamental period of the structure away from
dominate frequencies of earthquake ground motion as well as providing damping to reduce response. It
can be likened to introducing an engineered soft-story in the structure, whereby a majority of the inelastic
action and displacement is concentrated at a single level (the isolation level) therefore protecting the
building above.
The potential advantages of seismic isolation and the advancements in isolation system products led to the
design and construction of a number of isolated buildings and bridges in the early 1980s. This activity, in
turn, identified a need to supplement existing seismic codes with design requirements developed
specifically for such structures. The application of the technology in the United States is now regulated by
building codes which invariably refer to ASCE 7 (the Standard) for analysis and design requirements.
These requirements assure the public that isolated buildings are safe, they provide engineers with a basis
for preparing designs and they provide building officials with minimum standards for regulating
construction.
Seismic isolation is a high-performance system that provides greatly improved seismic performance
compared to most other conventional seismic force resisting systems. That is, a seismically isolated
building and a conventional fixed-base building, designed to the respective minimum requirements of the
Standard will exhibit widely differing performance in a major earthquake. A seismically isolated building
is expected to have considerably better performance, with limited damage, whereas a conventional building
may be damaged to the point where it is uneconomical to rehabilitate. This is due to the inherent nature of
isolation, for example it reduces both displacements and accelerations in the superstructure. Although not
explicitly seeking damage control as an objective, indirectly the minimum requirements of the Standard
will give limited damage as a consequence of ensuring proper isolation achieved. For example, the
Standard limits the amount of inelastic action and drift that may occur in the superstructure to avoid
detrimental coupling with the isolation system. Moreover, the Standard requires that the isolation devices
be designed and tested to be functional under the effects of the Maximum Considered Earthquake (MCER).
The Standard recognizes that the properties of an isolation system, such as effective stiffness and
damping, can change during repeated cycles of earthquake response. This variability is acceptable
provided that the design is based on analyses that conservatively bound (limit) the range of possible
values of design parameters.
The first seismically isolated buildings in the United States were composed of either high-damping rubber
(HDR) or lead-rubber (LR) elastomeric bearings. Other types of isolation systems now include sliding
systems, such as the friction pendulum system, or some combination of elastomeric and sliding bearings.
Furthermore, at sites with very strong ground shaking there has been the application of supplementary
fluid viscous dampers in parallel with the bearings to control displacement. While generally applicable to
15-5
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
all types of systems, certain requirements of the Standard (in particular, prototype testing criteria) were
originally developed for isolation systems with elastomeric bearings tested at slow-speeds.
Isolation systems typically provide only horizontal isolation and are rigid or semi-rigid in the vertical
direction. A rare exception to this rule is the full (horizontal and vertical) isolation of a building in
southern California, isolated by large helical coil springs and viscous dampers. Three-dimensional
seismic isolation platforms are also currently being designed and implemented for earthquake
protection of sensitive computer equipment. While the basic concepts of the Standard can be extended
to full isolation systems, the requirements are only for horizontal isolation systems. The design of a full
isolation system requires special analyses that explicitly include vertical ground shaking and the potential
for rocking response.
1) Determine the site response spectrum using the USGS website and check if a site-specific ground
motion hazard analysis is required. Determine proximity to faults for use in the selection and
scaling of motions (i.e. near fault, high velocity pulse motions). Engage specialist expertise, as
appropriate, for the selection and scaling of motions for response history analysis.
2) Select bearing dimensions and properties and quickly assess adequacy using the equivalent lateral
force (ELF) procedure for estimating displacement and the stability criterion (see Section 15.3.1).
Be sure to include the effects of additional axial load due to vertical earthquake and overturning
moment effects, and additional displacement due to torsion. The design may be conservative at
this stage as the analysis is not yet refined. Use lower bound properties of the isolation system
and the MCER earthquake.
3) When a bearing design appears promising, proceed with a formal determination of the isolation
system properties (see Section 15.4) and conduct the ELF procedure for lower bound and upper
bound bearing properties on a three-dimensional mathematical model of the building (see Section
15.5). At this point the designer may want to think about distribution of bearing types (e.g. some
with and some without lead cores) so that torsion is minimized. Calculate bearing displacement
demands, story shear forces, story drifts and additional axial bearing loads. Note that for the
calculations of the axial loads the designer will have to perform a static analysis of the frame,
including torsion effects.
4) Perform an analysis to assess the vertical earthquake effects. This could be by simply adding
0.2SMSD or by more refined calculations. For example, constructing a representative vertical
response spectrum and performing a vertical response spectrum analysis (see Section15.6.4).
This could be done with an ETABS (or similar) comprehensive model or by simply treating the
structure as a rigid block with vertical stiffness derived from the vertical stiffness of the bearings.
This analysis will give the peak vertical load on each bearing due to the vertical ground motion.
This can be combined with the horizontal earthquake effects by combining 100% vertical + 30%
horizontal or 30% vertical +100% horizontal (this is consistent with the ASCE 7-10 philosophy
3
Adapted from a design steps list created by SUNY Dist. Prof. Michael Constantinou, University at Buffalo.
15-6
Chapter 15: Seismically Isolated Structures
for combining loads in orthogonal directions, but a statement to this effect is not clearly made in
ASCE 7-10).
5) Assess the adequacy of the bearings using the results of the ELF analysis. Determine if
uplift/tension is a problem. If so, modify the structure to reduce uplift/tension to acceptable
levels (for an elastomeric bearing, negative pressure not more than 3G, where G is the shear
modulus of the elastomer). The model for static analysis in step 3 above is useful in this step. If
not possible to modify the structure to achieve the desired effect, you may consider other options,
including modifying the isolation system to further reduce the shear force, replacing some
elastomeric bearings with another type of bearing or accepting damage to a limited number of
bearings provided the resulting structure response is acceptable.
6) Check requirements to see if a response spectrum or response history analysis is required. For a
response history analysis, model and analyze the structure in the lower and the upper bound
conditions for the MCER (see Section 15.6). The designer needs to decide on how to include
vertical earthquake effects. Options are to include nothing and simply add later the axial load
effect due to the vertical earthquake determined in Step 4, or add a +ag or –ag constant vertical
earthquake in the dynamic analysis where a is a portion of the acceleration determined in the
vertical earthquake response spectrum analysis (say if the peak value was 1.2g, then a = 0.3 1.2
= 0.36). Also, the designer needs to consider torsion. The best approach is to model only the
actual eccentricities (none in this example, because the building above the bearings and the
isolation system is symmetric) and not artificially shift the center of mass. Accidental torsion is
instead added using the same approach as in the ELF procedure or by using amplification factors.
7) Compare response history and ELF results on peak resultant bearing displacement, story shear
force and drifts (along the two principal building directions). Decide on the final values for
bearing displacements and forces to use in assessing adequacy of the bearings. These values
will be from the response history analysis values if its values are larger than the ELF values;
otherwise the values for evaluation cannot be less than a certain portion of the ELF values per the
Standard §17.6.4.
15-7
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
The structure (which is regular and symmetrical in configuration) has plan dimensions of 100 feet by
150 feet at all floors except for the penthouse, which is approximately 50 feet by 100 feet in plan.
Columns are spaced at 25 feet in both directions. Figures 15.2-2 and 15.2-3 are framing plans for the
typical floor levels (but with beam sizes for the first level) and the penthouse roof, respectively. The X-
direction is termed the longitudinal direction and the Y-direction is the transverse direction
15-8
Chapter 15: Seismically Isolated Structures
15-9
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
The vertical load-carrying system consists of concrete fill on ribbed steel deck floors, supported by steel
beams at 8.3 feet on center and steel girders at gridlines. Isolation devices (also referred to as isolators or
bearings) support each column below the first floor. The foundation consists of concrete spread footings.
The lateral force resisting system consists of a roughly symmetrical pattern of concentrically braced
frames. These frames are located on Gridlines B and D in the longitudinal direction and on Gridlines 2, 4
and 6 in the transverse direction. Figures 15.2-4, 15.2-5 and 15.2-6 show elevations of the longitudinal
and transverse framing. This framing is specifically configured to reduce the concentration of earthquake
overturning and uplift (tension) loads on the bearings. This is achieved by:
Increasing the number of bays with bracing at lower stories.
Locating braces at interior (rather than perimeter) gridlines since they have greater gravity loading.
Avoiding common end columns for bracing in the transverse and longitudinal directions.
15-10
Chapter 15: Seismically Isolated Structures
The isolation system has 35 identical lead rubber bearings located below each column. The first floor is
just above grade and the bearings are approximately 3 feet below grade to provide clearance below the
first floor for construction and maintenance personnel. A short retaining wall borders the perimeter of the
facility and provides approximately 2 feet of “moat” clearance for lateral displacement of the isolated
structure. Access to the EOC is provided at the entrances by segments of the first floor slab, which
cantilever over the moat.
Girders at the first-floor gridlines are much heavier than the girders at other floor levels and have
moment-resisting connections to columns. These girders stabilize the bearings by resisting moments due
to both vertical P-delta effects and horizontal shear loads. Column extensions from the first floor to the
top plates of the bearings are stiffened in both horizontal directions to resist these moments and to serve
as stabilizing haunches for the beam-column moment connections. The typical detailing of this
connection is shown in Figure 15.2-7.
15-11
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
The 2012 IBC Section 1607.10 permits an area-based live load reduction of not more than 60 percent for
elements with live loads from multiple stories. Therefore the axial component of live load on columns at
lower levels and on bearings have a total reduced live load (L) of L = 2,200 kips.
The effective seismic weight is taken as the dead load plus half the reduced life load, equal to 10,200 kips.
Alternatively, the seismic weight could be calculated in accordance with Standard §12.7.2, which gives
an effective seismic weight equal to the dead load plus a provision for partitions and the weight of
permanent fixtures. For this example the seismic weight is taken as the former case, D + 0.5L, to be
consistent with vertical load combination 1 specified in §17.2.7.1.
15-12
Chapter 15: Seismically Isolated Structures
Performance criteria. The performance criteria are determined according to Standard §1.5.1:
Occupancy Importance Factor (Standard Chapter 17): I = 1.0 (for an isolated structure)
Note: Standard Chapter 17 does not require use of the occupancy importance factor to determine the
design loads on the structural system of an isolated building (i.e., I = 1.0). However, the component
importance factor is still required by Chapter 13 to determine seismic forces on nonstructural components
of isolated structures (Ip = 1.5 for Occupancy Category IV facilities).
Design spectral accelerations. Chapters 11 and 21 of the Standard are used to determine the design
spectral accelerations. The Standard incorporates changes to the ground motions (new USGS spectral
accelerations and site coefficients) and new site-specific analysis requirements. Section 11.4.7 requires
that a ground hazard analysis be performed in accordance with Section 22.2 on sites with an S1 greater
than or equal to 0.6. For the purpose of this example, a generic site has been selected with details as
follows:
Short-period MCER spectral acceleration adjusted for site class (FaSS): SMS = 1.4
1-Second MCER spectral acceleration adjusted for site class (FvS1): SM1 = 0.9
15-13
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
Design basis.
Seismic force-resisting system (< 160 feet): Ordinary steel concentrically braced frames (OCBF)
Although Table 12.2-1 of the Standard limits the height of fixed-base OCBFs to 35 feet for this Seismic
Design Category D building, new provisions (§17.2.5.4) permit the use of OCBFs in seismic isolation
applications for heights up to 160 feet. This is permitted provided 1) the OCBF remains elastic for the
MCER event and 2) the displacement capacity to which the isolation system lock-up or impact the moat
wall shall be increased by a factor of 1.2.
Note: Standard §17.5.4.2 gives RI = 3/8R ≤ 2, RI = 1.2 however RI is limited to 1.0 in §17.2.5.4
since it is an OCBF.
Horizontal irregularity (of superstructure) (Standard Table 12.3-1: Type 1b): None
Vertical irregularity (of superstructure) (Standard Table 12.3-2: Type 1a, 1b, 5a, 5b): None
The Standard has been revised so that analysis and design need only be considered for the MCER event.
The horizontal, vertical and torsional earthquake effects are taken as follows:
Mass eccentricity - actual plus accidental: 0.05b = 5 ft (X direction); 0.05d = 7.5 ft (Y direction)
where QE are the effects of horizontal seismic forces and 0.2SMSD is the vertical seismic load effect. For
the ELF procedure, the combination of the horizontal earthquake effects in the X and Y directions are:
For the response history analysis, two perpendicular components of horizontal ground motion are
simultaneously applied in the model and therefore the combination of orthogonal effects is directly
accounted for.
Gravity loads (dead load and reduced live load): 1.4D and 1.2D + 1.6L
Maximum gravity and reduced earthquake loads (1.2D + 0.5L + 1.0E): 1.48D + 0.5L + QE/RI
Minimum gravity and reduced earthquake loads (0.9D - 1.0E): 0.62D - QE/RI
Note the load factor on L is taken as 0.5 for this example since the live load is not greater than 100 psf.
15-14
Chapter 15: Seismically Isolated Structures
Gravity loads (for example, long term load on bearings): 1.4D and 1.2D + 1.6L
Standard §17.2.7.1:
The preliminary design requires the determination of the isolation system properties (e.g., effective period
and damping), which depend on the type and size of the isolation bearings and the type and size of
supplementary dampers if they are also incorporated into the isolation system. The size of the bearings is
related to the amount of vertical load that they support and the maximum lateral displacement. This
displacement is a function of both the MCER ground motions at the building site and the effective period
and damping of the isolation system. Thus, preliminary design tends to be an iterative process.
A common approach is to perform parametric studies to get a theoretical understanding of the tradeoff
between forces and displacements. However the engineer must be mindful of specifying details (i.e. a
friction coefficient or a stiffness) of bearings not previously manufactured and tested as this will give
uncertainty in performance and potential for further iterations later in design. Furthermore, atypical bearings
may be more expensive and take longer to manufacture, test and deliver. A recommended approach is to
make contact with manufacturers early in the design process to view qualification test data (§17.8.1.1 of
the Standard) and use this data along with the details of the similar bearings as a basis for design.
The preliminary designs are based on the approach presented in Constantinou et al. (2011). This method
uses the ELF procedure and gives guidance for estimating the dynamic nominal properties of lead-rubber
and friction pendulum bearings based on the assumption that the manufacturer is unknown or
qualification test data is not available.
15-15
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
The characteristic strength Qd is the strength of the bearing at zero displacement and can be idealized as
being related to the diameter of the lead core DL and the effective yield stress of lead σYL:
DL2 YL
Qd (15.3-1)
4
Equation 15.3-1 implies that any contribution to the strength from rubber is included in σYL, which is a
reasonable simplification for low-damping rubbers used in LR bearings.
The post-elastic stiffness kd is related to the shear modulus of rubber G, the bonded rubber area (which is
a function of the diameter of the bearing and the lead core) and the thickness of rubber Tr (sum of the
individual rubber layers):
Gf L ( DB2 DL2 )
kd (15.3-2)
4Tr
The parameter fL accounts for the effect of the lead core on the kd and ranges in value from 1.0 to about
1.2. Only after repeated cycling is the value of fL close to unity.
Nominal properties and bounding. The two important properties to determine are the effective yield
stress of lead σYL and the shear modulus of rubber G. These properties are dependent on a variety of
parameters and are manufacturer specific.
There is uncertainty in σYL as it is dependent on the speed of motion, size and confinement of the lead core
(i.e. vertical load, manufacturer details with steel plates and installation of the lead core), and degrades from
cycle to cycle due to heating effects. Constantinou et al. (2011) provide guidance on a range of properties
based on high velocity, large amplitude testing of LR bearings. The dynamic nominal value (which is the
average of properties over 3-cycles of loading) of σYL may be in the range of 1.45 to 1.75 ksi. The upper bound
15-16
Chapter 15: Seismically Isolated Structures
is taken as the first cycle properties, recommended as 1.35 times the nominal properties. The lower bound
is assumed to be the nominal properties, which are close to the second cycle properties. Hence the values of
σYL taking into account uncertainty in the nominal value, scragging, speed of motion and heating effects is:
Upper bound σYL,max = 1.75 1.35 = 2.36 ksi
Lower bound σYL,min = 1.45 ksi
The shear modulus of rubber G depends on the rubber compound and manufacturing processes, which are
proprietary, as well as the on the frequency and conditions of loading. Constantinou et al. (2011) recommend
a nominal value in the range of 65 to 125 psi. Lower values of G are possible however few manufacturers
can reliably achieve this without experiencing significant scragging effects (temporary degradation of
properties with repeated cycling). For this example we seek the lowest shear modulus and consider a
range of 60-75psi for a competent bearing manufacturer. The associated scragging and aging effects are
taken as 1.1 and 1.1, respectively, times the nominal value. Hence the values of G are calculated as:
Upper bound Gmax = 75 1.1 1.1 = 91 psi
Lower bound Gmin = 60 psi
If the manufacturer is inexperienced then these values may not be conservative and it is recommended the
engineer use a wider range to account for uncertainties.
Preliminary design procedure. This procedure is based on examples in Constantinou et al. (2011) and
involves assessing the bearing stability, which is a critical check for preliminary sizing of elastomeric
bearings. Other adequacy checks are necessary but can be done later in design or by the bearing
manufacturer.
The bearing stability is critical at maximum displacements and large compression loads. Hence the lower
bound properties will be used for this analysis (lower bound on lateral stiffness) with a conservative
(high) estimate on the vertical load. An upper bound analysis (upper bound on lateral stiffness) may also
be conducted at the preliminary design stage as it typically results in the maximum forces on the structure
and maximum uplift demands on the bearings. This upper bound analysis can also be done after a
mathematical (e.g. ETABS) model is created, so as to better understand the distribution of inertial forces
over the height of the structure.
To enable iteration, the process is best executed using a spreadsheet and can be summarized as follows:
1. Approximate values for the variables DL, DB and/or Tr
2. Conduct the ELF procedure to calculate the maximum displacement DM
3. Increase DM for torsion and make a conservative estimate of the vertical compression load due to
dead, live, vertical earthquake and overturning effects.
4. Calculate the required individual rubber layer thickness to maintain stability of the bearing. An
acceptable design is typically one where the rubber layer thickness is in the range of 0.25 to 0.75
inches and where the shear strain (total displacement DM divided by Tr) is in the 200-250% range.
If this is not achieved then perform another iteration by altering the values in Step 1
Step 1
The first step involves sizing the lead core so that the strength of the isolation system is some desirable
proportion of the building weight W. In general Qd,total/W should be about 0.05 or greater in the lower
bound analysis. A ratio of 0.08 gives a system strength Qd,total of 816 kips and, by Equation 15.3-1, a lead
core diameter of DL = 4.5 inches for 35 LR bearings. As will be seen later, it is necessary to increase DL in
order to reduce displacements and maintain stability of the bearing. Guidance from Constantinou et al.
(2011) on the selection of DB and Tr is as follows:
15-17
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
Furthermore, the configuration of the isolation system may be altered, for example by arranging a mixture
of LR bearings and low-damping elastomeric bearings, or by installing supplementary dampers that act in
parallel with the isolation bearings. For simplicity, these options are not explored in this example.
Step 2
Using the LR bearing dimensions in Step 1 and properties in Section 15.3.1.2, construct a bilinear force-
displacement model of the isolation system. This is simply the sum of individual bearings strengths and
stiffnesses since the bearings are acting in parallel across the isolation plane. The ELF procedure (per
Section 15.5.1) is then conducted to calculate the maximum bearing displacement DM.
Step 3
This step involves estimating the critical combination of displacement (including torsion) and vertical
load for the stability calculation. Since this is preliminary design, conservative assumptions can be made
by calculating the displacement for a corner bearing and combining with the maximum vertical load which
occurs at any interior bearing. Since the Standard does not permit the total maximum displacement DTM to
be taken less than 1.15DM, and considering that the building is rectangular and has a uniform distribution
of bearing properties (i.e. not purposely configured to increase the torsion resistance), DTM of the corner
bearing is taken as 1.2 times DM calculated in Step 2. The vertical load from dead and live loads can be
roughly calculated based on tributary area, giving 380 kip and 90 kip, respectively, and the vertical load
from earthquake overturning can be approximated by taking the base shear KMDM, applying it as a vertical
triangular distribution over the height of the structure, distributing it to the braced frames and then
calculating the resistance to this overturning by the reaction from each bearing support. This crude and
conservative calculation gives a maximum overturning load of about 550 kip for the lower bound analysis.
Therefore from load combination Equation (15.2-2) the maximum compression load from preliminary
calculations is 1200 kips which is used for the stability calculation in Step 4.
Step 4
The empirical equations that follow, for assessing the required rubber layer thickness, are from
Constantinou et al. (2011). For a deformed hollow (assuming the lead core does not to contribute to
stability) circular bearing which is bolted to base plates top and bottom, the maximum thickness of rubber
layers, t, to maintain stability is calculated as follows:
1 DL 1 DL
2
DB DB2 ( sin )
GDB4
t 0.218 (15.3-3)
Tr DL2 Pu
1 2
DB
where G is the nominal shear modulus of rubber (ksi), DB is the diameter of the bearing (inch), DL is the
lead core diameter (inch), Tr is the total thickness of rubber (inch), Pu is the factored ultimate compression
load (kip), the factor of safety ϕ is taken as 1.1 and δ is calculated as follows:
2cos1 DTM D (15.3-4)
B
A summary of the preliminary design calculations is provided in Table 15.3-1. The first design iteration
required a rubber thickness of 0.072 of an inch and had a shear strain of over 300%, which is not practical
15-18
Chapter 15: Seismically Isolated Structures
to construct. Therefore the size of the lead core and diameter of the bearing were increased to get a rubber
layer that is practical to construct and to reduce the shear strain.
Uplift assessment. At this point in design it is also worthwhile to assess the potential for uplift at
bearings. Using preliminary estimates of axial loads along with the minimum vertical load combination,
Equation 15.2-3, the uplift demand is about -350 kip (in tension). Tension in elastomeric bearings should
be avoided, nevertheless, Constantinou et al. (2007) states that high quality manufacturers can sustain
tensile pressure of about 3G before cavitation occurs (where G is the shear modulus of the elastomer).
The issue however is that this cannot be known without testing of the production bearings and testing may
damage the bearings. Accordingly, it is recommended to avoid large tension demands that are close to the
capacity of 3GAr (~100 kip for this preliminary sized bearing using lower bound properties).
15-19
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
There are various solutions to consider for mitigating the effects of uplift:
1. Change the structural system to eliminate or reduce the bearing tension to an acceptable level, by:
a. Increasing the number of bays with braced frames.
b. Moving the braced frame location to where the columns have high dead loads or artificially
increasing the dead load (i.e. localized thickening of slabs or heavy facades).
c. Stiffening the beams above the bearings to distribute the tension over a larger number of bays.
2. Alter the bearing connection details to allow for limited and controlled uplift. This will be for the
small number of bearings experiencing uplift. Options include:
a. Have one end of the bearing bolted and the other end recessed/slotted (or with a “loose-nut”
condition) to allow limited and controlled uplift.
b. Use of an uplift restraining system.
These options require explicit modeling of the uplift behavior to quantify the effects of bearing
uplift, and testing under these conditions may also be required.
3. Perform more refined calculations for uplift, using a 3-dimensional mathematical model and by
using NLRHA.
Local uplift of individual elements is permitted (Standard §17.2.4.7), provided the resulting deflections
do not cause overstress or instability of the isolated structure. For this example, the braced frames are
well distributed and there is not scope to alter the structural system. Therefore a more refined calculation
of the uplift demands will be made using a three dimensional building model and using nonlinear
response history analysis. This is expected to reduce the calculated uplift demand.
Uplift on sliding bearings also needs careful consideration however there are examples where controlled
uplift of sliding bearings have been accepted on projects, for example the Mills-Peninsula Bay Hospital
and the new San Bernardino Court facility, Sarkisian et al. (2012).
The preliminary design involves determining the dynamic friction coefficient of the sliding interface μ
and the post-elastic stiffness kd. This stiffness is simply a function of the radius of curvature of the
concave plates. By determining these two variables, one can calculate the force-displacement behavior of
the bearing and in turn the effective stiffness and effective damping of the isolation system.
15-20
Chapter 15: Seismically Isolated Structures
Single, double and triple concave sliding bearings, with identical friction coefficients and identical radii
of curvature on the outer concave plates, can all be idealized by the rigid-linear model shown in Figure
15.3-3. The rigid-linear model will give a reasonable estimate of the global response of the structure for
triple concave sliding bearings. However, if in-structure accelerations and residual displacements, or
behavior beyond the MCER are of interest to the RDP then it may be appropriate to adopt a more
sophisticated force-displacement model. The formulation, implementation and validation of the triple
concave sliding force-deflection behavior, as well as for other configurations of multi-spherical bearings,
can be found in Fenz and Constantinou (2008).
The characteristic strength Qd of the sliding bearing is calculated as μ times the weight on the bearing:
Qd W (15.3-5)
The post-elastic stiffness kd for the double concave bearing with R1= R2 and h1=h2 is calculated as:
W W
kd (15.3-6)
2( R1 h1 ) 2 Reff
15-21
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
where Reff if the effective radius of curvature equal to (R1-h1). Furthermore the actual displacement
capacity d* for R1= R2, d1=d2 and h1=h2 is given by:
2( R1 h1 ) R
d* 2d1 eff 2d1 (15.3-7)
2 R1 R1
Nominal properties and bounding. It is recommended to make contact with bearing manufacturers in
order to help select a range of trial design friction coefficients and available radii of curvature and
diameters (i.e. displacement capacity) of concave plates.
As shown in Equation 15.3-6, kd is purely based on the geometry of the bearing. Since this can be constructed
with a high degree of tolerance by most manufacturers, the bounding of kd is not required for sliding bearings.
kd is largely dependent on the value of R1 and therefore this is the parameter the engineer can optimize.
Other bearing dimensions can be determined by the manufacturer. A partial list of previously manufactured
concave plates, from Constantinou et al. (2011), is R1 of 61, 88, 120, 156 and 238 inches. Increasing R1
gives a lower post-elastic stiffness. This typically results in a lower structural shear at the cost of a larger
displacement. Although there are greater P-delta moments and the bearing may require more material
(i.e. thicker top and bottom plates), this extra displacement can be readily accommodated by increasing
the displacement capacity, or by relation d1, with no change in the force-displacement behavior.
For this example, it is assumed that the manufacturer has already produced and tested bearings with an R1
of 88 inches, and that this will be used as a basis for design.
The coefficient of friction μ is affected by a number of factors, of which the sliding velocity, bearing
pressure (axial load divided by the contact area of the slider) and temperature are the most important.
Furthermore μ depends on the type and construction of the sliding interface, which is manufacturer
specific and proprietary. As such, a range of default μ values is not listed here for sliding bearings. Rather
it is recommended to view dynamic test data from the manufacturer without any effects of aging,
contamination and history of loading, that is, for a fresh bearing tested at normal temperature.
For example, Constantinou et al. (2011) approximate the nominal (three-cycle average) coefficient of
friction for concave sliding bearings of a particular manufacturer that uses a polytetrafluoroethylene
(PTFE) and stainless steel sliding interface, as follows:
where p is the contact pressure of the slider. Equation 15.3-8 is applicable for pressures of 2 to 8 ksi, with
a slider diameter of 11 inches, and tested amplitudes of 12 to 28 inches. Furthermore testing at velocities
of the order of 39 inch/s will have a lower μ than those predicted by Equation (15.3-8) by amounts of
about 0.01 to 0.02.
Hence for our preliminary design the average vertical load (D+0.5L) for all bearings is 290 kips, which
for a 12 inch diameter slider gives a bearing pressure of 2.6 ksi and μ of 0.096 less 0.015 (for high
velocities) = 0.081. This value is further adjusted for uncertainty in the nominal value by a factor of 0.80
to give a nominal coefficient of 0.065 rounded down to 6%. This value is also taken as the lower bound.
The upper bound value may be taken as the first cycle value, equal to 1.2 μ, multiplied by aging and
contamination effects taken as (1+0.75(1.1 1.1-1)) = 1.16 for an internal environment and a 1.2 factor
for uncertainty in the nominal value. Therefore the friction coefficient from preliminary design,
accounting for aging, contamination, velocity and heating effects, and uncertainty in the nominal value
are:
15-22
Chapter 15: Seismically Isolated Structures
Preliminary design procedure. The selection of the double concave sliding bearing dimensions, as
depicted in Figure 15.3-2, are explained in the preceding sections and are taken as:
Using Equations 15.3-5 and 15.3-6 to construct the force-displacement behavior, the ELF procedure (per
Section 15.5.1) is conducted to calculate the maximum bearing displacement DM and maximum base
shear using the upper and lower bound properties determined in 15.3.2.2.
Using Equation 15.3-7, the actual displacement capacity is 28.5 inches, which is greater than the lower
bound total maximum displacement of 23 inches. This displacement includes torsion taken as 1.2 times
DM. Therefore the displacement capacity of the bearings is adequate.
15-23
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
Given the importance of the isolation system, and uncertainty in the quality of different manufacturers,
the Standard has taken the approach of requiring the registered design professional (RDP) to determine
(in consultation with the manufacturer) the nominal design properties and to account for the likely
variation in those properties on a product- and project-specific basis. This is achieved through
incorporating new provisions (§17.2.8) for calculating the upper- and lower-bound force-deflection
behavior of the isolation system.
The following sections give guidance on interpreting test data and procedures for determining the nominal
mechanical properties of the bearings. To account for the variation in these properties, property
modification or λ factors are used to modify the nominal properties to an appropriate upper- or lower-
bound.
The concept of property modification factors was originally presented in Constantinou et al. (1999) and is
already implemented in bridge design codes. The approach is to assess the impact of a particular effect on
the bearing properties (e.g. heating, aging, velocity, etc), and if the effect is appreciable then assigning it a
λ-factor and accounting for the effect in analysis and design. Therefore the λ-factors encompass many
different effects and describe the deviation in properties their nominal value. For example, if an effect
causes a 10% increase in a nominal property than it is assigned a λ value of 1.10 and contributes to the
overall λmax factor. The λ-factors are determined through testing, rational analysis and engineering
judgment and are categorized in the Standard into three groups:
λae.max and λae.min which account for aging and environmental effects.
λtest,max and λtest.min which account for scragging, hysteretic heating and speed of loading.
The Standard then combines the λ-factors using Equations 17.2-1 and 17.2-2 as follows:
The λmax and λmin factors are applied to each nominal mechanical property of interest and therefore set the
upper- and lower- bound force-displacement behavior, respectively. For the LR bearing the shear
modulus of rubber G and the effective yield stress of lead σYL are the important properties. For the sliding
bearing, it is the friction coefficient μ. Each of these properties requires determination of property specific
λmax and λmin factors.
15-24
Chapter 15: Seismically Isolated Structures
In Equations 17.2-1 and 17.2-2 the Standard presumes a system property adjustment factor fa of 0.75, to
account for the conservative assumption of having full aging and environmental effects when the
governing earthquake occurs. However the RDP has discretion to increase fa based on the contributing
factors to λae and/or based on the significance of the structure.
The limits of Equations 17.2-1 and 17.2-2 do not apply when either:
a) qualification test data per §17.8.1.2 , (i.e. satisfying Items 2, 4, 7 and 8 of the similarity
requirements of §17.8.2.7), is approved by the RDP and is used to establish λ-factors.
b) project-specific dynamic prototype testing is conducted per §17.8.2.3, and that data is used
establish λ-factors.
Therefore the limits of Equations 17.2-1 and 17.2-2 are rarely expected to apply and are intentionally
wide. Caution is advised, however, as they may not be conservative for inexperienced manufacturers
with no, or limited test data. Regardless, it is implied by the Standard that dynamic testing is required in
such a case.
Furthermore, to supplement theses new provisions on isolation system properties, the Standard
incorporates a new clause: §17.8.1.1 Qualification Tests as well as new criteria in §17.8.2.7 Testing
Similar Units for when a bearing may be classified as similar. §17.8.2.3 has also been renamed Dynamic
Testing and implies that prototype testing shall be conducted dynamically.
The RDP has the flexibility and authority to determine what data is accepted as qualification test data.
However it is generally the responsibility of the manufacturer to conduct qualification testing. These tests
may be used to aid in the establishment of nominal properties and λ-factors, to characterize the longevity
of the bearing, and to develop models of the bearing for analysis. In practice it is assumed that
manufacturers have comprehensive databases of properties, based on research projects and past prototype
and production testing. This is already the case for two suppliers in the United States which have
compiled large databases.
The intent of the Standard is not that dynamic testing of bearings be conducted on every project. This can
be expensive and can only be performed at a limited number of facilities. The change in language in
§17.8.2.3 is identifying that most bearing types exhibit velocity dependence, and that the nominal
properties (average over three cycles) calculated from slow-speed testing will be different from the
nominal properties calculated from dynamic testing. The nominal properties are typically underestimated
by slow-speed testing, due to speed of loading effects, and furthermore the variation in those properties,
or bounds (λtest,max and λtest,min), are also underestimated by slow-speed testing due to less heating
effects. Therefore it is important to account for dynamic effects in the analysis and design.
Interpreting Test Data from Lead-Rubber Bearings. For this example, dynamic testing per
§17.8.2.2, Item 3 is conducted at a vertical load equal to D + 0.5L on two virgin (unscragged) bearings
tested at a normal temperature of 20°C. This test consists of three fully-reversed cycles at a displacement
amplitude of DM conducted dynamically at the effective period TM determined from the upper-bound
properties. Therefore the speed of loading effects and heating effects are directly accounted for by the
testing, since it is conducted at approximate peak earthquake velocities (i.e. 30-40 inch/sec).
Figure 15.4-1 illustrates the force-displacement behavior of a LR bearing tested at high-speed, with the
test data for two prototype bearings documented in Table 15.5-1. Table 15.5-1 gives the maximum
measured force and maximum measured displacement in the positive and negative directions, F+ and F-
and Δ+ and Δ- respectively, as well as the measured energy dissipated per cycle Eloop. Eloop is calculated by
15-25
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
numerical integration of the test loop. Although this test data is fictitious, it is generally representative of
high-speed test results from large LR bearings.
The test data in Table 15.4-1 is sufficient to determine the isolation system properties. Before calculating
the nominal properties, a few comments and recommendations are noted as follows:
The first cycle shows a distinctly higher strength and stiffness. This is due primarily to heating
effects on the lead core and also scragging effects of the rubber if the test is on a virgin/not
previously tested bearing (i.e. unscragged).
Sequential prototype testing with a small rest time for cooling may give an underestimate of the
strength of lead in the first cycle. Therefore it is recommended to determine properties for
analysis from only the initial test(s), of a regime of many tests.
The post-elastic stiffness kd is influenced by the lead core. Only after the first few cycles are its
effect negligible (i.e. in Equation 15.3-2 fL is equal to 1.0). Therefore it is recommended that the
first cycle is not used in the calculation in the shear modulus of rubber.
The shear modulus of rubber shall not be determined from a coupon test specimen that has a low
shape factor (bonded area divided by the area free to bulge). LR bearings used in buildings
typically have a shape factor greater than 10 and often larger than 20, and the rubber deforms
purely in shear. For coupon tests the rubber may be deforming in shear and bending, which
15-26
Chapter 15: Seismically Isolated Structures
results in an artificially low estimate of the shear modulus and consequently an unrealistically low
calculated effective stiffness of the actual bearing.
It is not possible to determine the scragging effects of rubber from lead rubber bearing test data as
the effects in the first cycle are completely masked by heating effect on the lead core. Data from
bearings tested from a virgin state without the lead core, or using data from plain natural rubber
bearings (if such bearings are also used in parallel in the isolation system) may be suitable.
Coupon tests may also be used to estimate scragging effects as the scragging effects are expected
to be similar on the bending response (coupon test) and shear stiffnesses (full size bearing test).
However size effects shall be taken into consideration since large bearings may exhibit variable
zones of curing through the volume of the bearing whereas small coupon samples are generally
uniformly cured throughout their thickness.
The stiffness of rubber is dependent on the shear strain; hence the post-elastic stiffness is
displacement-dependent. Although dynamic test data are only shown for the maximum displacement,
this is somewhat in conflict with the Standard which requires that properties envelope 0.5DM
up to and including DM. This implies that tests at other displacement amplitudes must also be
used in determining properties. Enveloping the post-elastic stiffness at different strains may result
in a multi-linear force-deflection loop (say, high post-elastic stiffness at smaller strains, which is
reduced at medium strains and increased again at large strains). It is the authors’ opinion that the
post-elastic stiffness should be a best-fit representation of the strains from 0.5DM to DM from
the dynamic test data in Figure15.4-1, using only a bilinear model, and that test cycles at other
displacement amplitudes (i.e. 0.5DM and 0.67DM) be viewed to verify consistency with this best-
fit. The reason for this is that, under the current Standards provisions, analyses are carried out
only at MCER conditions. Furthermore the bilinear representation is fitted to the test loop such
that it has the same values of effective stiffness at the maximum displacement.
The calculation of the mechanical properties for the two similar/prototype lead rubber bearings is given in
Table 15.4-2. The effective stiffness keff is calculated using Equation 17.8-1 of the Standard. The
characteristic strength Qd (fitted loop force at zero displacement) is determined using Equation 15.4-1,
using the energy dissipated Eloop, the displacement amplitude DM, and by assuming a yield displacement
Y, taken as 0.6 inches for all cycles.
Assuming that this strength is all from the lead core, per Equation 15.4-1, one can calculate the effective
yield stress of lead. It is noted that rubber may contribute small amount to Eloop (by relation, say 0.02 to 0.05
effective damping), and the interested reader is referred to Kalapakidis et al. (2008) for more information.
The post-elastic stiffness can be determined directly by a straight-line fit, or can be calculated based on
the Qd and keff by the following relationship:
Qd
keff kM kd (15.4-2)
DM
The shear modulus of rubber is then calculated by Equation 15.3-2, assuming the contribution to the post-
elastic stiffness from the lead core is negligible (fL=1) after the first cycle. As noted above, the first cycle
properties cannot be used, since the effects of heating on the lead core mask behavior of the rubber.
15-27
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
Fortuitously, the two prototype bearings have near identical properties. The nominal mechanical
properties are calculated as the average among the three cycles, and averaged for the two bearings.
Therefore the nominal properties are G = 58 psi and σYL = 1.69 ksi.
The next step is to determine the associated test λ-factors. The Standard requires λtest,max and λtest,min to
include variation from Item 2 of §17.8.2.2, which for the Item 2b dynamic testing consists of the
following sequence: continuous loading of one fully-reversed cycle at each of the following increments of
maximum displacement DM: 1.0, 0.67, 0.5, 0.25 followed by continuous loading of one fully-reversed
cycles at each of the following increments of DM: 0.25, 0.5, 0.67 and 1.0.
For the effective yield stress of lead σYL the upper bound is taken as the first cycle properties divided by
the nominal value, giving λtest,max=2.27/1.69=1.34. What to take for the lower bound is not so clear. In the
initial cycles of loading the lead loses strength due to hysteretic heating effects. This reduction in strength
is temporary and recoverable with adequate cooling time. The representative values in Table 15.5-2 show
a large difference between the σYL in the first and third cycles, which is not uncommon for large-scale
bearings tested at high-speed. It is the opinion of the authors that the lower bound should be based on
considerations of the seismic hazard and nominal isolation system properties (strength and stiffness).
Response history analysis studies by Warn and Whittaker (2007) demonstrate that about two (and less
than three) fully-reversed cycles at the maximum displacement are expected for isolation systems with a
yield strength to supported weight ratio (Qd/W) of 0.06 or larger and period based on a post-elastic
stiffness of 2.5 seconds or greater. Therefore the lower bound will be taken as the second cycle properties,
giving λtest,min=1.56/1.69=0.92.
A validated theory from Kalpakidis et al. (2008) can be used to quantify these heating effects, thus
reducing the need for extensive testing. For example, the tested bearing had a displacement amplitude of
15 inches whereas the preliminary design only required a 11.2 inch amplitude. Examples of using
simplified heating calculations are illustrated in McVitty and Constantinou (2015).
The process described above is believed to be consistent with the intent of the Standard, even though it
uses a different test sequence. Simplified heating calculations by Kalpakidis et al. (2008) show that the
reduction in σYL can be related to the total travel of the bearing. After two cycles of loading at DM the
tested bearing has experienced 2 4 DM = 120 inches of travel and σYL has reduced to a value of 1.56 ksi.
Had the Item 2b test been conducted, with a DM of 11.2 inches, the total travel for bounding would be
(1+0.67+0.5+0.25) 4 DM = 108 inches, which is less than 120 inches. Therefore the effects of heating
15-28
Chapter 15: Seismically Isolated Structures
are adequately captured by testing to two full cycles at a displacement of DM. It is also noted that the
effective damping βeff in Table 15.4-2 is under-predicted when compared to the preliminary design values,
as the bearings were tested to a larger displacement.
For the shear modulus of rubber G, the scragging effects need to be determined from other tests, and are
assumed in this example to be minor at λtest,max=1.1 and increased to 1.15 since the nominal value is based
on the last two cycles only. Data presented by Thompson et al. (2000) demonstrate that these scragging
effects are recoverable within a short period of time (see Constantinou et al. 2007 for more references).
Therefore the full effects of scragging should be incorporated in design, even though the bearing is
scragged in the production tests.
Sliding bearings. Representative high-speed dynamic force-displacement behavior for a sliding bearing
is illustrated in Figure 15.4-2. The test sequence is not too dissimilar from first half of the §17.8.2.2, Item
2b testing. It is noted that this data is for a triple concave sliding bearing, which is why there is a slope
(apparent yield displacement) upon reversal of the direction of displacement. For the double concave
sliding bearing described in this example the behavior would be more likened to the rigid-linear model of
Figure 15.3-3 (i.e. yield displacement Y=0 in Equation 15.4-1).
The process of determining the nominal mechanical properties and fitting a rigid-linear loop is a similar
process to that described in Section 15.4.2.1, with some important points to note:
The dynamic friction coefficient μ is the most important parameter to determine. It can be
calculated directly from the measured energy dissipated per cycle, using Equation 15.4-1 and
assuming a yield displacement Y = 0. The utility of this calculation is that it can be readily
15-29
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
obtained without determination/judgment of other parameters (say, for a triple concave sliding
bearing).
It is recommended to calculate the post-elastic stiffness based on the geometry of the bearing, per
Equation15.3-6. This is because the geometry can be constructed by most manufacturers with a
high degree of tolerance and theory predicts behavior very accurately. Any observed difference
between the fitted loop kd (based on Equation 15.3-6) and the test data kd may be due to (a)
fluctuations of the vertical load during the test, which affects the instantaneous value of the
friction coefficient, and (b) heating effects on the friction coefficient which are more pronounced
in high-velocity, large-amplitude cycles. Therefore the approach recommended is to accept that
the post-elastic stiffness does not vary and to assign any variability from cycle to cycle to the
frictional properties.
The hysteretic heating effects on sliding bearings are dependent on the friction coefficient (which
in turn is dependent on temperature), pressure and sliding velocity, as well as size of the bearing
components. Therefore the heating effects on sliding bearings are different from that explained
above for LR bearings, however the studies by Warn and Whittaker (2004), with regard to the
equivalent number of cycles of two at the maximum displacement, are still considered applicable.
The following sections list typical λae,max and λae,min expected of experienced manufacturers, and are
adopted for this example.
Elastomeric bearings. The aging and environmental factors for the shear modulus of rubber G are:
λae,min,G = 1.0
The aging and environmental factors for the effective yield stress of lead σYL of 99.99% purity are:
λae,max,σ = 1.0
λae,min,σ = 1.0
Sliding bearings. The aging and environmental factors for the friction coefficient μ for an unlubricated
PTFE-stainless steel sliding interface are:
15-30
Chapter 15: Seismically Isolated Structures
λae,min,μ = 1.0
This tolerance is required because the testing of a small number of prototype bearings may not necessarily
provide the best estimate of the nominal design properties. This potential discrepancy occurs because the
average of two prototype test results may be at the upper or lower end of the range of a larger population.
Alternatively, if past test data is used to establish nominal properties, there may be differences due to the
natural variability in properties and manufacturing variations.
For this example the specification tolerance on the average properties of all bearings, for each property of
interest (i.e. G, σYL or μ), is ±10%. That is:
λspec,max = 1.10
λspec,min = 0.90
Variations in individual bearing properties from the nominal design properties may be greater than the
tolerance on the average properties of all bearings, say ±15%. The wider specification tolerance for
individual bearings is not used for analysis of the isolation system but should be taken into account for
bearing connection design by amplifying the upper-bound analysis forces by the ratio of the λ-factors,
e.g., 1.15/1.10 for the example values here.
LR bearings force-displacement behavior. The maximum and minimum λ factors for the shear
modulus of rubber and effective yield stress of lead are calculated based on Equations 17.2-1 and 17.2-2,
as follows:
The upper- and lower- bound force-displacement behavior of the preliminary sized elastomeric isolation
system, based on the nominal values and λ-factors in Sections 15.4.2 through 15.4.4, is illustrated in
Figure 15.4-3.
15-31
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
0.4
0.35
0.3
Lateral Force/Weight
0.25
0.2
0.15
Displacement (inch)
Sliding bearings force-displacement behavior. The upper- and lower- bound force-displacement
behavior of the sliding isolation system, based on the preliminary design, is illustrated in Figure 15.4-4.
0.4
Lower Bound
0.35
Upper Bound
0.3
Lateral Force/Weight
0.25
0.2
0.15
0.1
0.05
0
0 2 4 6 8 10 12 14 16 18 20
Displacement (inch)
Figure 15.4-4 Preliminary Design Force-Deflection Behavior of Sliding Isolation System
A check of the minimum restoring force, per §17.2.4.4, gives a difference between the force at 0.5 DM and
1.0DM of 0.037W for the upper-bound properties and difference of 0.057W for the lower-bound
properties. Hence the Standards minimum limit of 0.025W is satisfied.
15-32
Chapter 15: Seismically Isolated Structures
The following calculations are illustrated, arbitrarily, for the elastomeric isolation system. The calculation
process for the sliding isolation system would be similar.
15.5.1 Procedure
The ELF procedure is an iterative process and is illustrated in the following equations for the preliminary
design bearing dimensions and lower bound properties. The terms below are defined in Section 15.3.
W
TM 2 (15.5-2)
kM g
10200
2 1.83 seconds
311 386
4. Calculate the effective damping βM. The yield displacement Y is assumed to be 0.6inch
4Qd ,total ( DM Y )
M (15.5-3)
2 kM DM 2
BM 1.44
6. Check the displacement matches what was initially assumed in Step 1 (Equation 17.5-1):
15-33
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
gSM 1TM
DM (15.5-4)
4 2 BM
386 0.9 1.83
11.18 inch 11.2 inch O.K
4 2 1.44
Modeling assumptions. To expedite calculation of loads on bearings and other elements of the seismic-
force-resisting system, a three-dimensional mathematical model of the building is developed and analyzed
using the computer program ETABS (CSI, 2013).
The vertical load on each bearing depends on their relative vertical compression stiffness Kv and soil
stiffness, the rigidity of the base level diaphragm/framing and the mass distribution. The compression
stiffness of a multi-layered elastomeric bearing can be approximated accordingly:
1
1 4
K v A ti (15.5-5)
i Eci 3K
where A is the bonded rubber area, ti is the individual rubber layers thickness, K is the bulk modulus of
rubber, assumed as 290 ksi and Eci is the compression modulus for incompressible material behavior
which is dependent on the rubber shear modulus, shape factor and the bearing geometry (see
Constantinou et al., 2007 for details). The compression stiffness for the LR bearings was calculated as
about 10,000 kip/in. The soil stiffness is taken as 150 pci and for a 5ft square footing beneath each
bearing column gives and vertical soil spring stiffness of 540 kip/in.
Since the ELF analysis is a linear-elastic method, the lateral effective stiffness of the bearings at the
maximum displacement is input into the model. For each bearing the effective stiffness is simply the
isolation systems effective stiffness, from Section 15.5.1 calculations, divided by the number of bearings.
A modal analysis of the structure shall be used to verify that the first two modes of vibration are
translational with a period similar to Section 15.5.1 calculations, with the third mode of vibration being
the torsional isolated mode.
Bearing dimensions and properties. The preliminary sizing of the bearing, per Section 15.3.1, used
quick but conservative calculations. Using the more refined calculations that are set out in the following
sections, it was decided to further optimize the sizing of the bearings to reduce structural shear. The
final dimensions of all the 35 LR bearings and their properties (as determined in Section 15.4) are as
follows:
15-34
Chapter 15: Seismically Isolated Structures
Using Equations 15.3-1 and 15.3-2 and details in Table 15.5-1, the ELF force-displacement behavior of
the isolation system can be constructed, as illustrated in Figure 15.5-1:
Lateral Force (kip)
4500
Lower
Bound 3500
Upper
2500
Bound
1500
500
-2500
-3500
-4500
Figure 15.5-1 Force-Deflection Behavior of Elastomeric Isolation System
Maximum displacement and effective period. The maximum displacement DM and effective period at
the maximum displacement TM is calculated using the ELF procedure in Section 15.5.1, which is
consistent with §17.5.3.1 and §17.5.3.2 of the Standard . The calculations for the upper- and lower-bound
properties are documented in Table 15.5-2.
15-35
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
Lateral seismic forces and vertical distribution. The lateral shear force required for the design of the
isolation system, foundation and other structural elements below the isolation system is given by Vb in
Equation 17.5-5. The overturning loads (i.e. axial loads) from the superstructure, which are used for the
design of the isolation system, foundation, and elements below the isolation system is given by the
unreduced lateral force Vst in Equation 17.5-7. Subject to the limits of §17.5.4.3, the base shear, Vs, for the
design of superstructure above the isolation level is taken as Vst reduced by the RI factor in accordance
with Equation 17.5-6. The results from these calculations for the upper- and lower- bound analysis are
given in Table 15.5-3.
The minimum lateral design force for structural elements above the isolation system Vs is initially
calculated as 3449 kip. However the value of Vs shall also be checked against §17.5.4.3 to ensure it
meets the minimum requirements. As shown in Table 15.5-4, the value of Vs is not governed by the
minimum requirements.
15-36
Chapter 15: Seismically Isolated Structures
The Standard has been revised to incorporate a more realistic distribution of lateral forces over the
buildings height (York and Ryan 2008) with details of the method explained in the Standard commentary.
Because the superstructure is much stiffer laterally than the isolation system, it tends to move as a rigid
body in the first mode, with a pattern of lateral seismic forces that is typically more uniformly distributed
over the height of the building. This is rather than an inverted triangular distribution, which is
representative of the first mode for a fixed-base building.
The method calculates the force of the base level, immediately above the isolation plane, then distributes
the remainder of the base shear among the other levels. The vertical distribution of the unreduced lateral
forces is given in Table 15.5-5. It is noted that §17.5.4.2 has an exception which recommends a more
conservative exponent term be used in Equation 17.5-7 when the hysteretic behavior of the isolation
system is characterized by an abrupt transition from pre-yield to post-yield. For an elastomeric isolation
system the transition is typically rounded and therefore the (1-2.5β) exponent is relevant.
15-37
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
Bearing vertical loads. The vertical/axial load on the bearings was calculated using the ETABS model
for both the upper- and lower-bound properties. In this case the upper-bound properties gave the critical
earthquake demands and are reported in Table 15.5-6 and 15.5-7. This table documents the loadings from
dead and reduced live loadings, as well as the envelope of the maximum and minimum demands from
horizontal earthquake and torsion actions. The X and Y directions referred to in the tables are illustrated
in Figure15.3-2. A negative sign denotes tension loading.
Since the isolation system and lateral-force resisting system have a symmetrical layout (in two
directions), only the critical demands are reproduced in the Table 15.5-6 through 15.5-9. That is, loads
and displacements at Gridlines 5, 6 and 7 (not shown) are similar to those at Gridlines 3, 2 and 1,
respectively; and loads and displacements at Gridlines D and E (not shown) are similar to those at
Gridlines B and A, respectively.
15-38
Chapter 15: Seismically Isolated Structures
Using the load combinations in Section 15.2.4, which incorporate vertical earthquake actions using
0.2SMSD, Table 15.5-7 shows the maximum and minimum downward forces for design of the bearings.
These forces result from the simultaneous application of gravity loads and unreduced earthquake story
forces (see Table 15.5-5) to the ETABS model. It is noted that tension stresses are developed in a number
of locations. This tension stress is acceptable for a high-quality elastomeric bearing if it is less than three
times the shear modulus of rubber. For the LR bearing in this example the capacity for the upper- and
lower bound properties is 3GA, equal to 80 and 120 kip, respectively. Hence there are four bearing
locations (Gridline A2, E2, A6 and E6) where there is a potential uplift issue. This issue is discussed
further in Section 15.7.1, after the vertical response spectrum and response history analyses.
15-39
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
Table 15.5-7 ELF Maximum and Minimum Vertical Loads on Bearings, Upper-Bound
Properties
Maximum Loads 1.48D + 0.5L + max(QEX+0.3QEY, 0.3QEX+QEY) + max(QE,Torsion) (kips)
Gridline 1 2 3 4
A 237 640 356 306
B 592 752 776 867
C 395 631 683 657
Minimum Loads 0.62D + min(QEX+0.3QEY, 0.3QEX+QEY) + min(QE,Torsion) (kips)
Gridline 1 2 3 4
A -4 -129 24 52
B -65 -8 57 12
C 35 92 225 255
Using the modal analysis from the ETABS model, the ratio of the 1st and 3rd modes, which are the
translational and torsional modes, respectively, for the lower-bound properties are 2.14sec/1.83sec or PT =
1.17. There could be a further reduction in the amplification factor (or increase in PT) by placing LR
bearings around the perimeter and plain elastomeric or sliding bearings on the interior of the building to
increase the torsional resistance. However the Standard places a minimum amplification factor of 1.15DM
for the corner bearings, regardless of how torsionally stiff the isolation system is. Part of the reason
behind increasing this minimum factor is that past experimental tests used to quantify the isolation
systems torsional response was on unrealistically torsionally stiff isolation systems (i.e. only four bearings
located on the perimeter).
The Standard minimum torsional amplification factor and total maximum displacements for various
bearing locations is given in Table 15.5-8. The minimum amplification factor is calculated using Equation
17.5-3 with the greater of the amplification for and X- or Y-direction reported in the table. It is noted that
the total maximum displacement calculated using the ETABS model (i.e. with the application of the
torsional moment at each floor level) gives similar values for this example.
15-40
Chapter 15: Seismically Isolated Structures
Table 15.5-8 Minimum Amplification Factors and ELF Total Maximum Displacement
Minimum Torsional Amplification Factor
Gridline 1 2 3 4
A 1.15 1.10 1.07 1.07
B 1.15 1.10 1.05 1.03
C 1.15 1.10 1.05 1.00
Upper-Bound Total Maximum Displacements, DTM (inches)
Gridline 1 2 3 4
A 9.6 9.1 8.9 8.9
B 9.6 9.1 8.7 8.6
C 9.6 9.1 8.7 8.3
Lower-Bound Total Maximum Displacements, DTM (inches)
Gridline 1 2 3 4
A 15.3 14.6 14.2 14.2
B 15.3 14.6 14.0 13.7
C 15.3 14.6 14.0 13.3
Bearing stability and shear strain assessment. A more refined calculation of the minimum required
individual rubber layer thickness for stability (per Equation 15.3-3), with compatible combinations of
maximum axial loads and total maximum displacements, for each bearing location are documented in
Table 15.5-9. The critical bearing location is Gridlines B4 and D4 which require a rubber thickness less
than 0.28 inches which is at the lower limit for what can be satisfactorily constructed by manufactures.
This would give a final bearing with 5.225/0.275 = 19 rubber layers each 0.275 inches (7 mm) thick. The
rubber shear strains due to lateral displacements are high at 294% in the corner locations, but are
achievable for a quality manufacturer.
The stability check is illustrated in this example since it is typically a governing criterion for the design of
the lead-rubber bearings. However, the design checks of the bearing stability, rubber shear strains, and
other components of the bearing (i.e. steel shim plates and end plate design) are typically the
responsibility of the manufacturer.
15-41
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
Table 15.5-9 Minimum Individual Rubber Thickness and Rubber Shear Strain
Upper-Bound Minimum Individual Rubber Layer Thickness, ti (inches)
Gridline 1 2 3 4
A 1.54 0.56 1.02 1.17
B 0.71 0.49 0.47 0.43
C 1.07 0.69 0.54 0.57
Lower-Bound Minimum Individual Rubber Layer Thickness, ti (inches)
Gridline 1 2 3 4
A 0.88 0.36 0.63 0.71
B 0.40 0.30 0.30 0.28
C 0.57 0.38 0.32 0.36
Lower-Bound Rubber Shear Strain, DTM/Tr (%)
Gridline 1 2 3 4
A 294 281 268 263
B 294 281 272 272
C 294 281 268 263
Story drifts. The Standard permits more liberal drift limits where the design of the superstructure is based
on a nonlinear response history analysis (NLRHA). The ELF procedure and response spectrum drift
limits are 0.015hsx for the reduced MCER level forces, which are increased to 0.020hsx for a NLRHA
(where hsx is the story height at level x). Usually a stiff system (e.g., braced frame) is selected for the
superstructure to limit damage to nonstructural components sensitive to drift and therefore the drift
demand is typically less than about 0.005hsx. Standard §17.6.4.4 requires an explicit check of
superstructure stability at the MCER displacement if the earthquake story drift ratio exceeds 0.010/RI.
The maximum story displacement of the structure above the isolation system is calculated by Equation
12.8-5 with Cd equal to RI (Cd = 1) and using Ie = 1.0. The upper-bound properties give the greater story
shear and therefore greater drift. The story drift in each direction, including accidental eccentricity, are
given in Table 15.5-10. Since the structure has braced frames, the calculated maximum story drift ratio is
well below the limit of 1.5%.
15-42
Chapter 15: Seismically Isolated Structures
If any of the items were not satisfied, then a dynamic analysis would be required. If items number 1, 2, 3,
4, 6 and 7(sic) are satisfied, then the response spectrum procedure is permitted. The ELF and response
spectrum procedures are both linear-elastic analyses, where the bearings behavior is represented by an
effective stiffness and effective damping at the maximum displacement DM. The response history
analysis, on the other hand, directly accounts for the nonlinear bearing behavior and is permitted for all
seismically isolated structures.
Table 15.5-11 Restrictive Requirements for ELF (and Response Spectrum) Analysis
Item Upper- Lower-
Requirement Check
No. bound bound
1 The structure is located on a Site Class A, B, C and
Site Class D O.K.
D.
2 The effective period of the isolated structure at the
maximum displacement, DM, is less than or equal to 1.5 sec 2.1 sec O.K.
5.0s.
3 The structure above the isolation interface is less than
or equal to 4 stories or 65 ft (19.8m) in structural
height measured from the base level. Exception: 4 stories with 50 ft height. O.K.
These limits are permitted to be exceeded if there is
no tension/uplift on the bearings.
4 The effective damping of the isolation system at the
maximum displacement, DM, is less than or equal to 24% 16% O.K.
30%.
5 The effective period of the isolated structure TM is
Tfb=0.43s
greater than three times the elastic, fixed-base period
TM = 1.5sec TM = 2.1 sec 3Tfb=1.3s
of the structure above the isolation system
O.K.
determined using a rational modal analysis.
6 The structure above the isolation system does not
No structural irregularity,
have a structural irregularity, as defined in Section O.K.
see Section 15.2.4.1
17.2.2.
7a The effective stiffness of the isolation system at the
maximum displacement, DM, is greater than one-third
464 > 409 239 > 164 O.K.
of the effective stiffness at 20 percent of the
maximum displacement.
7b The isolation system is capable of producing a
restoring force such that the lateral force at the
corresponding maximum displacement is at least 0.11W 0.12W O.K.
0.025W greater than the lateral force at 50 percent of
the corresponding maximum displacement.
7c The isolation system does not limit maximum
Assumed to have no
earthquake displacement to less than the total O.K.
restrictions less than DTM
maximum displacement, DTM.
15-43
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
The NLRHA gives the most realistic estimate of an isolated buildings response. Typically the
superstructure is modeled as elastic with nonlinear behavior confined to the isolation level. Two models
are required, one with the isolator units upper-bound force-deflection properties and one using the lower-
bound properties. Each model is subjected to at least seven different ground motion sets, where the values
used in design (for each response parameter of interest) are taken as the average of the seven ground
motion analyses maxima. The ELF analysis is still necessary to evaluate results of the dynamic analysis
and to obtain minima of response quantities.
In this example the NLRHA is used only to determine the final design displacements of the isolation
system, to assess the overturning loads and uplift, and to verify that ELF-based story forces used for
design of the superstructure are valid. If the NLRHA procedure were used as the primary basis for
superstructure design, then response results would be required for design of individual elements, rather
than for checking a limited number of global response parameters.
The nonlinear force-deflection characteristics of isolator units are modeled explicitly (rather than using
effective stiffness and damping, as in the ELF procedure). For most types of isolators, force-deflection
properties can be approximated by bilinear, hysteretic curves that can be modeled using commercially
available nonlinear structural analysis programs. The initial stiffness, yield strength and post yield
stiffness ratio of a single bearing for the upper- and lower-bound properties specified in ETABS is given
in Table 15.6-1. These properties give an isolation system response (for 35 bearings) matching that shown
in Figure 15.5-1.
More sophisticated nonlinear models may be necessary to accurately represent the response of isolators
with complex configurations or properties (e.g., triple pendulum sliding bearings), to capture stiffening
effects at very large displacements (e.g., of elastomeric bearings), to model rate-dependent effects
explicitly (Sarlis 2010, Kalapakidis 2008) or to model uplift behavior.
15-44
Chapter 15: Seismically Isolated Structures
Special modeling concerns for isolated structures include two important and related issues: uplift of
isolator units and P-delta effects on the isolated superstructure and the substructure. Typically, isolator
units have little or no ability to resist tension forces and can uplift when earthquake overturning (upward)
loads exceed factored gravity (downward) loads. To model uplift effects, gap elements may be used in
nonlinear models or tension may be released manually in linear models. For this example the uplift forces
from NLRHA do not exceed 3GA (see Section 15.7.1.1 for details) therefore uplift is not modeled
explicitly.
The vertical earthquake effects were not explicitly considered in the NLRHA model (i.e. only two
horizontal ground motion components are applied to the model) but are accounted for in the results later.
The effects of P-delta loads on the isolation system and adjacent elements of the structure can be quite
significant. The compression load, P, can be large due to earthquake overturning (and factored gravity
loads) at the same time that large displacements occur in the isolation system. Computer analysis
programs (most of which are based on small-displacement theory) may not correctly calculate P-delta
moments at the isolator level in the structure above or in the foundation below. Figure 15.6-1 illustrates
moments due to P-delta effects (and horizontal shear loads) for an elastomeric bearing and a flat sliding
bearing. The same concept applies for multi-spherical sliding bearings. For the elastomeric bearing, the
P-delta moment is split one-half up and one-half down. For the flat and single-concave sliding bearings,
the full P-delta moment is applied to the foundation below (due to the orientation of the sliding surface).
A reverse (upside down) orientation of the flat and single-sided sliding bearings would apply the full P-
delta moment on the structure above. For the double-concave sliding bearing, P-delta moments are split
one-half up and one-half down, in a manner similar to an elastomeric bearing, provided that the friction
(and curvature) properties of the top and bottom concave dishes are the same.
A V
P
C V
H1 P
H3
V B V D
E V G V
P P 15-45
H5 H7
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
Selection and scaling of ground motions. The Standard requires that ground motions be scaled to
match maximum spectral response in the horizontal plane. In concept, at a given period of interest, the
maximum spectral response of scaled records should, on average, be the same as that defined by the
MCER spectrum. The ground motion acceleration histories selection and scaling are illustrated in Chapter
3 of these NEHRP Design Examples.
For NLRHA, Standard §17.3 requires at least seven pairs of horizontal ground motion acceleration
histories be selected from actual earthquake records and scaled to match the MCER spectrum. Where the
required number of recorded pairs is not available, then the Standard permits the use of simulated
ground motion records. Selection and scaling of appropriate ground motions should be performed by a
ground motion expert experienced in earthquake hazard of the region, considering site conditions,
earthquake magnitudes, fault distances and source mechanisms that influence ground motion hazard at the
building site.
Standard §17.3.4 recognizes two types of scaling methods: amplitude scaling and spectrally matching,
and has different requirements for each. There are also different requirements if the site is within 3 miles
(5 km) of an active fault. For this example, the site is greater than 3 miles away from an active fault, and
amplitude scaling is the selected scaling method. In this case, the Standard requires that the earthquake
records are scaled to match a target spectrum over the period range of interest, defined as 0.75TM
determined using upper-bound isolator properties to 1.25TM using lower-bound isolator properties. This
gives a period range of interest of 1.1 to 2.6 seconds for the elastomeric isolation system. For each
period in this range, the average of seven square-root-of-the-sum-of-the-squares (SRSS) combinations
(of each pair of horizontal components of scaled ground motion) should be equal to or greater than 1.0
times the MCER spectrum.
The scaling factors are shown in Table 15.6-2 and reflect the total amount that each as-recorded ground
motion is scaled for NRLHA. Also shown are associated parameters which are useful for NLRHA.
Further illustration of selection and scaling of ground motion records is illustrated in Chapter 3 of these
NEHRP Design Examples.
15-46
Chapter 15: Seismically Isolated Structures
Orientation of ground motion components for analysis. Only for sites within 3 miles of an active
fault does the Standard specify how the two scaled components of each record should be applied to a
three-dimensional model (i.e., how the two components of each record should be oriented with respect to
the axes of the model). For other sites, the Standards commentary states that individual pairs of
horizontal ground motion components need not be applied in multiple orientations. Guidance on the
orientation of components of ground motions by NIST (2011) state that there is no systematic directional
dependence to ground motions at distant sites and that each pair of motions need only be applied to the
model in one orientation.
Since this example building is symmetrical in both directions and the site is further than 3 miles from an
active fault, the earthquake records were applied randomly to the model in one orientation only. The post-
processing of results per Section 15.6.4 accounts for whether the maximum response occurs in the
positive or negative direction.
On some projects, lack of guidance sometimes caused engineers to perform an unnecessarily large
number of response history analyses with each pair of ground motion records oriented in four or even
more different orientations. It is the author’s experience that these additional analyses have diminishing
returns. However, multiple orientations may still be necessary depending on project, and should be
assessed on a case-by-case basis (i.e. jurisdictional requirements or as requested for design verification).
Vertical Earthquake Spectrum. In the ELF procedure the vertical earthquake effects are accounted for
by simply adding or subtracting 0.2SMSD or 0.2 1.4 = 0.28g. The 0.2 factor is derived by multiplying the
horizontal spectrum at short periods by 2/3 and then using a 30% combination with horizontal earthquake
effects. This crude approach is also valid for NLRHA; however a more thorough method is illustrated
herein.
The Standard does not give explicit guidance of how to account for vertical motions, however
commentary has been added to address the matter. In §C17.3.3 one recommended approach to compute
the vertical design spectrum is to use the 2009 NEHRP Provisions in Chapter 23 where SDS is replaced
with SMS. The vertical spectra are a strong function of the natural period, source-to-site distance and local
site conditions, and relatively weak function of magnitude and faulting mechanisms. The 2009 Provisions
have simplified this so that the vertical spectrum can be calculated through the parameter SS (short-period
horizontal spectral acceleration for the site), as well as the site class classification. Using the 2009
Provisions the vertical coefficient Cv was calculated to be 1.38 for the D site class. The resulting vertical
response spectrum, along with horizontal spectra for comparison (both for the MCER event) is given in
Figure 15.6-2. There is a limit that the vertical response spectrum acceleration shall not be less than one-
half (1/2) of the corresponding horizontal spectral acceleration, which is why there is a “kink” in the
vertical spectra to be used for design.
For the design vertical spectrum the period refers to the vertical period and the spectral acceleration is in
the upwards or downwards direction.
15-47
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
1.8
Vertical Spectrum
1.6
Horizontal Spectrum
1.4
0.8
0.6
0.4
0.2
0
0 0.5 1 1.5 2
Period (seconds)
Figure 15.6-2 MCER Vertical Spectrum for Design
Analysis and Bearing Axial Loads. A vertical earthquake analysis requires careful modeling
considerations. These are outlined in the Standard commentary §C17.6.2, such as including all structural
elements in the model and adding more degrees of freedom (i.e. nodes along a beam or slab) so that the
mass is realistically distributed across the building footprint. Consideration of the soil-structure
interaction is also necessary and will require input from a geotechnical engineer. The modal analysis must
also capture the vertical excitation of the building, which may require hundreds of modes to obtain 90-
100% of the mass participation in the vertical direction.
A single degree of freedom analysis should be used initially to estimate the vertical period and resulting
vertical base reaction. The stiffness of the building in the vertical direction is calculated as the vertical
stiffness of the bearing (10,000 kip/in) and soil stiffness below (540 kip/in) acting in series, which gives a
total vertical stiffness of 35 bearings multiplied by 512 kip/in (1/512 = 1/10000 + 1/540), which is 17,920
kip/in. The corresponding vertical period would be:
10, 200
T 2 0.24 seconds
17920 386
The vertical spectral acceleration for a period of 0.24 seconds is 1.09g. However the multi-degree of
freedom ETABS model shows that the longest period for a vertical mode shape is around 0.34 seconds, so
we would expect a lower base reaction than 1.09g using the vertical response spectrum analysis.
The vertical response spectrum analysis resulted in a total reaction in the upwards/downwards direction of
7174 kip, which is equivalent to 7174/10200 = 0.7g. This corresponds to about 0.15SMS (i.e. 0.7
30%/1.4) which is less than the 0.2SMS used in Section 15.5 to account for vertical earthquake effects. The
response spectrum analysis also gives a more realistic distribution of vertical earthquake loads over the
building footprint. These bearing axial loads are given in Table15.6-3 and can be acting either in the
upward (tension) or downward (compression) directions. These maximum vertical earthquake loads can
be combined with the maximum vertical reactions due to horizontal earthquake loads using orthogonal
combinations corresponding to the 100%-30% rule per §C17.2.
15-48
Chapter 15: Seismically Isolated Structures
Introduction. The two independent ETABS models, which represent upper- or lower-bound bearing
properties, are analyzed for the set of seven pairs of horizontal ground motion records applied to the
model in the one orientation only (i.e. are not rotated). The post-processing in this section takes the
absolute maximum (i.e. maximum whether in the positive or negative directions) response for each
ground motion. The average of these absolute maxima responses over the seven ground motions is then
used for design.
Due to the many inputs required for NLRHA it is important to carry out verification checks. For example,
Figure 15.6-3 shows the NLRHA hysteretic response (base shear vs. displacement) of the isolation system
in the X and Y directions from ground motion 5, which compares well to the ELF procedure.
4000
X-direction
3000
Y-direction
2000
ELF procedure
1000
Base Shear (kip)
0
-15 -10 -5 0 5 10 15
-1000
-2000
-3000
-4000
Displacement (inch)
Figure 15.6-3 Base Shear Force-Displacement Behavior of Isolation System for Lower Bound
Properties, comparison of ELF and NLRHA for Ground Motion 5.
15-49
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
Torsion. The Standard §17.6.2.1 requires that the effect of torsion above the isolation interface,
considering the most disadvantageous position of eccentric mass, be considered. There are two
components of eccentric mass, the inherent eccentricity between the center of mass and center of rigidity
and the accidental eccentricity. This accidental eccentricity approach is used to indirectly account for
various effects, including: plan distributions of mass that differ from those assumed in design, variations
in the mechanical properties of structural components, non-uniform yielding of the lateral system, and
torsional and rotational ground motions.
Different models could be used to explicitly evaluate various locations of accidental mass eccentricity.
However, this approach would require multiple additional models to consider the most disadvantageous
location of accidental eccentric mass. To avoid doing an unnecessarily larger number of analyses, the
Standard §17.6.3.4.1 now permits the use of amplification factors to account for the effects of accidental
mass eccentricity.
In the NLRHA of this example, only the actual eccentricities (none in this case since the superstructure
and isolation system are symmetric) were modeled, with the calculated displacements and forces being
increased for accidental eccentricity effects during post-processing of the results. The resultant maximum
displacement, DM was multiplied by the amplification factor given in Standard Equation 17.5-3 (see
Table 15.5-8). This procedure of amplifying DM by a factor is the recommended method to account for
accidental torsion in NLRHA since it can be problematic to artificially alter the mass and/or center of
stiffness of the model because it changes the dynamic characteristics of the model and may
unintentionally improve performance. Furthermore it significantly reduces computational effort. For shear
and axial forces, the effects of accidental eccentricities were accounted for by using the ELF procedure.
That is, applying the torsion statically to the model and superimposing the results on the NLRHA results
to give the worst effect.
Peak isolation system displacement and base shear. The isolation system displaces simultaneously in
the X- and Y-directions at each increment in time with the resultant displacement being the vectorial
(SRSS) combination of these two components. Simply taking the maximum X or Y displacement over the
whole ground motion record and calculating their SRSS combination may be overly conservative. For
example, Figure 15.6-4 shows the displacement history of the isolation system in the X and Y directions
for ground motion 5, Loma Prieta, using lower-bound properties. The peak X, Y and SRSS displacements
were 11.9, 6.5 and 12.0 inches, respectively. The SRSS of the maximum X and Y displacements is overly
conservative at (11.92+6.52)1/2 = 13.6 inches.
15-50
Chapter 15: Seismically Isolated Structures
12
10 X-displacement
8
Y-displacement
6
SRSS
Displacement (inch)
4
2
0
-2
-4
-6
-8
-10
-12
0 5 10 15 20 25 30 35 40
Time (seconds)
Figure 15.6-4 Displacement Histories of Isolation System, Ground Motion 5, Lower Bound
Properties
Table 15.6-4 summarizes peak SRSS combinations of isolation system displacements and base shears for
each ground motion from the NLRHA. The ELF calculated displacement and base shear are also given
for comparison. The displacements are calculated at the center of mass of the isolation system and do not
include the effects of accidental mass eccentricity. This is accounted for by using the amplification factors
in Table15.5-8.
Table 15.6-4 NLRHA Peak SRSS Displacement and Base Shear of the Isolation System
Upper Bound Lower Bound
SRSS Max SRSS Max SRSS Max SRSS Max
Ground Motion Displacement Base Shear Displacement Base Shear
(inch) (kip) (inch) (kip)
GM 1 3.6 2300 7.5 2047
GM 2 6.7 3041 6.6 1905
GM 3 6.5 3162 9.1 2048
GM 4 4.4 2140 6.7 1991
GM 5 9.1 3846 12.0 2858
GM 6 8.0 2561 12.5 2947
GM 7 4.9 2789 7.4 2120
NLRHA Average 6.2 2834 (28%) 8.9 2273 (22%)
ELF 8.3 3853 (38%) 13.3 3183 (31%)
To avoid possible under-design, the Standard establishes lower-bound limits on results of dynamic
analysis to be used for design. These limits are established as a percentage of the corresponding parameter
calculated using the ELF procedure equations.
The total maximum displacement of the isolation system shall not be taken less than 80 percent of DTM
calculated in Section 15.5, where DM may be replaced with D’M per Standard Equation 17.6-1.
15-51
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
DTM,LB = 1.15 8.9 = 10.2 inches < 12.0 use 12.0 inches
DTM,UB = 1.15 6.2 = 7.1 inches > 6.7 use 7.1 inches
Therefore the displacement used to design the bearings shall be no less than DTM = 12.0 inches for lower
bound properties, as the minimum limit governs, and DTM = 7.1 inches for upper bound properties.
Since an OCBF structural system has been used the moat clearance and all elements crossing the isolation
interface shall designed for a displacement increased by a factor of 1.2 per §17.2.5.4:
DTM = 1.2 12.0 = 14.4 inches, rounded up to 15 inches around the building perimeter.
The isolation system, foundation and all structural elements below the isolation system shall be designed
for lateral force no less than 0.9Vb. Therefore the NLRHA forces below the isolation system would need
to be scaled up by a factor of:
Scale factor for forces below the isolation level = 0.9 3853/2834 = 1.2
Story forces. Table 15.6-5 summarizes average (of seven ground motions) absolute maximum
(maximum whether in the positive or negative direction) story shear force results at each level in the X
and Y directions from the NLRHA and compares these values with story shear forces calculated by ELF
formulas for unreduced design earthquake loads. The upper-bound isolation system properties gave the
greater shear forces. Figure 15.6-5 shows story shears calculated by ELF formulas and by the NLRHA.
Table 15.6-5 Summary of Peak Design Story Shear Forces and Comparison with Story Shear
Values Calculated Using ELF Methods, Upper Bound Properties
Method of Analysis
Response Parameter Peak NLRHA Result
ELF formulas
X-direction Y-direction
Penthouse 761 803 807
Third Story 2238 1787 1915
Second Story 3074 2392 2277
First Story 3449 2620 2230
Vb (Isolators) 3853 2834 2410
15-52
Chapter 15: Seismically Isolated Structures
50
45 NLRHA X-direction
40 NLRHA Y-direction
35 ELF
Elevation (feet)
30
25
20
15
10
5
0
-5
0 500 1000 1500 2000 2500 3000 3500 4000 4500
Shear force (kip)
Figure 15.6-5 Comparison of Story Shear Force based on NLRHA Average and ELF Procedures,
Upper-bound Properties
The minimum requirements of the Standard are that the base shear, Vb, shall not be taken less than 80
percent of that determined from the ELF procedure (for this regular structure) and that Vs shall not be
taken less than the limits in Table 15.5-4.
Therefore if the NLRHA demands are used for superstructure design, all demands on structural elements
above the isolation level need to be scaled up by the following factors:
Checking X-direction minimum structural shear Vs = 1.09 2620 kip >1940 kip (Table15.5-4) O.K.
Checking Y-direction minimum structural shear Vs = 1.28 2230 kip >1940 kip (Table15.5-4) O.K.
Bearing vertical loads. The combination of vertical/axial loads on the bearings to be used for design
consist of static dead and live loads, torsion (from the ELF procedure per Table 15.5-6), vertical
earthquake loads per the response spectrum analysis and given in Table 15.6-3, as well as the vertical
effects of horizontal earthquake loads from the NLRHA per this section.
The earthquake horizontal/overturning forces are calculated to be smaller in the NLRHA compared to the
ELF procedure. These force need to be increased by the ratio of the NLRHA base shear to the minimum
limit given by §17.6.4. This gives a scale factor of 1.2, per Section 15.6.5.3, on the earthquake
overturning load. The scaled and factored minimum axial loads on the bearings, which were critical for
upper bound properties, are given in Table 15.6-6.
15-53
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
Table 15.6-6 Maximum and Minimum Bearing Vertical Loads, Upper-Bound Properties
Maximum Loads 1.2D + 0.5L + max(QE)a + max(QE,Torsion)b + 0.3(EV)c (kips)
Gridline 1 2 3 4
A 215 568 325 287
B 535 702 736 829
C 369 651 698 665
Maximum Loads 1.2D + 0.5L + max(0.3QE) + max(0.3QE,Torsion) + EV (kips)
Gridline 1 2 3 4
A 223 522 371 351
B 531 736 850 918
C 419 753 986 903
Minimum Loads 0.9D + min(QE) + min(QE,Torsion) - 0.3EV (kips)
Gridline 1 2 3 4
A 14 -74 51 70
B -16 31 79 30
C 62 119 208 219
Minimum Loads 0.9D + min(0.3QE) + min(0.3QE,Torsion) - EV (kips)
Gridline 1 2 3 4
A 6 -27 5 6
B -4 -11 -29 -59
C 15 15 -74 -20
a. NLRHA average of maximum or minimum value from each of the seven ground motion records and scaled by a
factor of 1.2
b. Calculated per the ELF procedure
c. Vertical response spectrum analysis per Section15.6.4, Table 15.6-3
The vertical loads in Table 15.6-6 and displacements per Section 15.6.4.3 are slightly less critical than
that calculated from the ELF procedure (see Tables 15.5-7 and 15.5-8, respectively). Therefore the sizing
of the bearing from the ELF procedure is adequate, however and could be further optimized with another
design iteration.
15-54
Chapter 15: Seismically Isolated Structures
Although tension in elastomeric bearings is discouraged, high quality bearings can sustain significant
tensile deformation and this characteristic can be utilized if tension is unavoidable. Constantinou et al.
(2007) observes that the bearing have approximately the same stiffness in tension and compression up to
cavitation (where small cracks develop in the volume of rubber). Cavitation occurs at a negative pressure
over about 3G, where G is the shear modulus of rubber. Hence the bearing per Table 15.5-1, and using
nominal properties, can sustain a tension load of about 88 kip which is greater than 74 kip (which results
from upper bound properties). Therefore theoretically this is achievable; however the tension capacity of
the manufacturers’ isolator should be verified to the satisfaction of the RDP.
15-55
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
The Standard §17.8.2.2 requires that cyclic tests at different displacement amplitudes be performed for
typical vertical load (1.0D + 0.5L) and for maximum and minimum values of vertical load which are
based on the load combinations of Standard §17.2.7.1. For this example, maximum and minimum values
of vertical load are based on average of all the bearings peak earthquake response load (critical for upper
bound properties), as defined by the loads in Table 15.6-6. The test displacement is conservatively based
on the isolation system lower bound properties. This range of vertical load, from 30 kips to 590 kips,
addresses the intent of the cyclic testing requirements (i.e., to measure possible variation in effective
stiffness and damping properties).
Nominal effective stiffness (average over three cycles) for typical vertical load tested at high-speed
(or tested at low speed and adjusted for high speed response) and with a displacement amplitude
of 1.0DM:
Gnom ( DB2 DL2 ) 0.25 DL YL, Nom 0.058 (25.752 4.752 ) 0.25 4.752 1.7
2
kM
4Tr DM 4 5.225 10.4
kM = 8.5 kips/inch
Nominal energy dissipated per cycle (average over three cycles), Eloop or EDC, for typical vertical
load tested at high-speed (or tested at low speed and adjusted for high speed response) and with a
displacement amplitude of 1.0DM:
The range in this effective stiffness and EDC shall fall within the limits specified by the RDP, as defined
by the λ-factors determined in Section 15.4.
15-56
Chapter 15: Seismically Isolated Structures
15-57
FEMA P-1051, NEHRP Recommended Provisions: Design Examples
1) Elastomeric isolators for the Kunming Airport in China were installed in 2011 without
testing. As construction progressed and loads increased on the isolators, several showed
signs of delamination. Hundreds of isolators were removed and replaced.
2) Sliding isolators for a project in Italy in which production testing would have disclosed
defects in manufacturing.
Furthermore, production testing shall be used to confirm that the isolation system properties are
appropriate and within the specification tolerance assumed for analysis and design. As discussed in
Section 15.4.4, variations in individual isolator properties from the nominal design properties may be
greater than the tolerance (λspec) on the average properties of all isolators, provided the individual
tolerance is taken into account for isolator connection design and structure in vicinity of the bearing.
Quasi-static production testing is assumed to be the common practice for most manufacturers at present.
Therefore to evaluate the consistency of the nominal values measured from production testing, there must
be a relationship established between properties determined under quasi-static and to behavior under
dynamic conditions. Typically, this requires that the prototype isolators are tested under the same
conditions as the production testing to establish the criteria for acceptance of the production isolators, and
to be tested under dynamic conditions for obtaining the nominal values and related property modification
factors for analysis.
A recommended approach, which is now written into the Standard §17.8.2, is to perform the production
set of tests on the prototype bearings (these are the bearings which are used to calculate the dynamic
properties and bounds for analysis) prior to the sequence and cycles in §17.8.2.2. This testing is
recommended first as it represents the test conditions of productions bearings. That is, a virgin state and
not previously tested. Therefore the calculated nominal properties (average over three cycles), which are
multiplied by λspec to give the acceptable range, will include scragging/first-cycle effects and will not be
influenced by heating effects from previous testing. Hence, the acceptance criteria of the average of all
production bearings are:
where:
NomProto = Nominal property average over three cycles from testing specified in Table 15.7-1 from
prototype isolator units not previously tested (virgin state).
NomProd,Avg = Nominal property average over three cycles from testing specified in Table 15.7-2 from
production isolator units not previously tested (virgin state), averaged for all production bearings of a
common type.
15-58