VIRTUAL BASE ISOLATION BY BUILDING SOFTENING WITH DRIFT CONTROL

PROVIDED BY FLUID VISCOUS DAMPERS

Michael Gemmill1, Kurt Lindorfer1, H. Kit Miyamoto2

ABSTRACT

Large mass and high story heights are common requirements for many
data storage and collocation facilities. These building characteristics, which are
typically considered design obstacles, actually provide a unique opportunity for
reducing seismic response through behavior similar to that of a base isolated
building.

In accordance with the 1999 SEAOC Blue Book (SEAOC, 1999)

recommendations for passive energy dissipation, the building’s Lateral Force
Resisting System (LFRS) is designed for strength requirements only, resulting in
a relatively flexible LFRS, while Fluid Viscous Dampers (FVD) are incorporated
to limit story drifts to acceptable levels. Due to the high building mass, large story
heights of 18’-0” (5.5m), and long period LFRS, the building exhibits a
fundamental period of 1.45 seconds, compared to 0.5 seconds for a typical two-
story moment frame building. The long period LFRS emulates the response of a
traditional base isolated system by reducing the acceleration on the building and
its contents, while story drifts are controlled by FVD.

There are many benefits to this “virtual isolation” system and

incorporation of the SEAOC Blue Book recommendations. With the elimination
of the maximum drift requirements, the moment frames are substantially lighter
than a traditionally framed building, thus lowering the structural steel cost of the
LFRS. The long period structure also produces significantly reduced forces in the
foundation elements. Velocity and displacement are reduced significantly through
the use of the FVDs, which protects the sensitive contents of the building. These
benefits lead to a reduced response resulting in an enhanced performance level
during a major seismic event.

Introduction

Several common structural design requirements for a facility housing high tech
equipment are low floor displacements, low floor accelerations and immediate occupancy status
after a major seismic event. When subjected to high ground accelerations, traditional LFRS are
unable to achieve both low floor displacements and accelerations. A true enhanced performance
level, such as an immediate occupancy state following a Maximum Capable Earthquake (MCE),
is also very difficult to obtain with a traditional structural system since it relies highly on
nonlinear behavior of the LFRS. The incorporation of FVD into the LFRS provide a drastic
reduction in nonlinear behavior while maintaining low floor displacements and accelerations
during a major seismic event.

__________________
1
PARADIGM Structural Engineers, Inc., 251 Kearny Street, 7th Floor, San Francisco, CA 94108,
2
Marr Shaffer & Miyamoto Structural Engineers, Inc., 1450 Haylard Dr., Suite One, West Sacramento, CA 95691
 Many high-tech facilities house a large amount of heavy equipment, which result in a
very high building seismic mass. Large story heights are also often required to allow for the
installation of mechanical and electrical support systems. A large mass building with large story
heights can be a major design obstacle for a traditional LFRS in a region of high seismicity. The
use of FVD can be a very efficient way of countering these design obstacles. The 1999 SEAOC
Blue Book states that when incorporating FVD, the LFRS shall be designed for strength
requirements only, with no limitation on drift demands. The FVD are in turn added to control
drifts to acceptable levels. This approach is intended to allow for a higher-period building, with
corresponding lower force demands, while maintaining acceptable drift limits by the addition of
dynamic stiffness through the use of FVD. This case study showed that this design philosophy is
very effective in producing buildings that are capable of achieving an immediate occupancy
status after a MCE event.

Building Description

The building is a two-story, 165,000 square foot (15,325 m2) data storage facility in Santa
Clara, California. The building is intended to house sensitive computer equipment, large
numbers of batteries, typically near the perimeter, weighing in excess of 400psf., a large number
of mechanical units resting atop a roof mechanical platform, and extensive electrical conduits.
The seismic mass of the building is approximately three times that of a typical office building.
The building has relatively high story heights of 18’-0”. The developer required a seismic
performance level of an immediate occupancy state following a Design Basis Earthquake (DBE),
which exceeds the requirements of the 1997 Uniform Building Code (UBC) (ICBO, 1997), for
an “essential services” facility. In addition, low floor displacements and accelerations were
required to protect the sensitive equipment housed on the second floor and roof mechanical
platform.

The LFRS consists of 56 total bays of two-story special moment resisting frames
(SMRF), utilizing Reduced Beam Sections (RBS) at the connections. A total of 104 nonlinear
FVD with a 400 kip. capacity were incorporated along the perimeter of the building, within two-
story “X” braced frames. The fundamental period of the building is 1.45 seconds, resulting in a
code prescribed base shear of 0.052g.

More traditional LFRS were not chosen for various reasons. A concrete shear wall
system was not chosen because high floor acceleration and associated potential equipment
damage was a concern of the client. A traditional braced frame system was eliminated as an
option due to the inherent high floor accelerations, as well as the need for deep foundation
elements to resist overturning forces. In addition, it is extremely difficult, if not impossible, to
achieve the enhanced performance levels desired by the client, with these traditional systems
which rely on inelastic behavior. Base isolation, though theoretically a viable option, was not
selected due to the high cost associated with designing the system to the stringent requirements
of the UBC.
 Input Time Histories

The building is located 16km from the Hayward fault and 17km from the San Andreas
fault, within a region of very high seismic activity. A site-specific probabilistic seismic hazard
analysis (PSHA) was performed to estimate the magnitude of ground acceleration at the site.
The PSHA modeled the faults in the Bay Area as linear sources and assigned earthquake
activities to the faults. Site-specific spectra at the ground surface were estimated using stiff soil
attenuation relationships consistent with the subsurface conditions encountered at the site.
(Gouchon, 2000) DBE is defined as a 500 year return event, and MCE is defined as a 1000 year
return event. Spectral matching was performed to provide appropriate time histories for both
DBE and MCE levels. Site specific response spectra for a 5% damping are shown in Figure 1,
along with corresponding UBC response spectra graphs. Time histories were chosen based on
similarities in magnitude and distance to the target spectra. Three earthquakes were incorporated
for each level of seismic hazard (6 total). The worst case results for acceleration, velocity and
displacement were used in design of the LFRS. The time history values are shown in Table 1.

1.6
Spectral Acceleration (g)

1.4
1.2 1 99 7 U B C D B E

1 1 99 7 U B C M C E
0.8
D e s ig n B as is
0.6 E a rthq ua k e
M a x im u m C ap ab le
0.4 E a rthq ua k e
0.2
0
0 0 .5 1 1 .5 2 2.5 3 3.5 4 4.5

P e rio d (s e c o n d s )

Figure 1. Response spectra graphs for 5% damped building

Table 1. Time history values

Epicentral Peak Peak Peak
Earthquake Time History Magnitude Hazard Distance Acceleration Velocity Displacement
Level (km) [g.] [in./sec.] [in.]
Landers Joshua Tree 7.4 DBE 15 0.504 28.20 9.12
Loma Prieta Los Gatos 6.9 DBE 23 0.464 29.64 10.08
Imperial Valley (1979) Differential Array 6.6 DBE 26 0.492 34.56 14.76
Kocaeli Duzce 7.4 MCE 90 0.606 36.36 14.28
Loma Prieta Los Gatos 6.9 MCE 23 0.575 37.08 12.96
Landers Yermo 7.4 MCE 84 0.556 29.64 13.32
 Virtual Isolation

The relatively flexible moment frames emulate a traditional base isolation system. In a
traditional base-isolated structure,

The system decouples the building or structure from the horizontal components of the
ground motion by interposing structural elements with low horizontal stiffness between
the structure and the foundation. This gives the structure a fundamental frequency that is
much lower than both its fixed-base frequency and the predominant frequencies of the
ground motion.” (Naeim and Kelly, 1999),

The two fundamental advantages of a base isolated system are low acceleration and drift for the
superstructure. Both of these qualities are captured in the low frequency moment frame with
FVD system. In addition, the performance of a base-isolated system rises with an increase in
mass of the ground floor. The large mass of the 2nd floor of the subject project produces an
analogous effect on the performance of the LFRS.

Design and Analysis Procedure

A stick model, based on story stiffness and mass, was first developed in order to obtain
an initial estimate of the required critical damper properties, including maximum force, damping
coefficient and nonlinear damping exponent. Time History Analyses (THA) were used to
determine the seismic demand on the structure. The analyses were performed on a trial and error
basis, with a final result commensurate with our performance requirements of maximum
allowable building drift.

The stick model was then transformed into a two-dimensional model and analyzed in
ETABS (CSI, 1999) using the damper properties from the previous model. The number of
dampers was selected based on the damping coefficient required to produce an acceptable drift.
Additionally, the quantity of dampers was chosen such that a maximum damper force of 400kips
was produced under a DBE level event. The 400kip level was deemed an acceptable and
economical maximum force level for this structure. The model was analyzed using three DBE
time histories [Table 1.], which produced a slight variation of damping properties of the FVDs
from the preliminary stick model. The beams and columns were modeled as linear elements, and
their demand-to-capacity ratios (DCR) were determined. A maximum DCR of approximately
0.9 was observed. A DCR of less than two for ductile elements is generally considered an
acceptable level for immediate occupancy, per FEMA 273 (FEMA, 1997) when using an
unreduced ground acceleration on a linear structure.

A true nonlinear model was built using RAM Perform (RAM International, 2000)
software to verify the results in the ETABS model. This software has advanced nonlinear
modeling capabilities for the nonlinear beam and column elements. The beam elements
consisted of rigid end zones from the centerline of the column to the column face, an elastic
beam segment from the column face to the centerline of the RBS, zero length plastic moment
hinges at the centerline of the RBS and an elastic beam segment between the hinges. The
column elements consisted of a rigid end zone at the base, a zero length moment hinge above the
rigid end zone, and an elastic element above the hinge. The beam-column joint was modeled as
an elastic panel zone element, comprised of four pin-connected rigid links with a rotational
spring. The dampers were modeled as a nonlinear viscous element with an elastic bar
representing the steel driver brace. LRFD was used for design using a load combination of 1.2D
+ 1.0L + 1.0E. The model was subjected to both DBE and MCE level time history events. Beam
rotations and demands, column rotations and demands, panel zone DCR, damper DCR, and
interstory drifts were recorded for worst case DBE and MCE events.

Analysis Results

All elements that had a possibility of experiencing inelastic response, including RBS and
column bases, were modeled as deformation controlled elements using nonlinear components.
Inelastic limits were checked for these elements based on FEMA 273 requirements. All
elements that were expected to remain elastic, including panel zones and portions of beams and
columns, were modeled as force controlled elements using linear components. Force levels were
checked for these elements based on standard steel design equations with a stress reduction
factor of 1.0.

All deformation and force level results corresponded to an immediate occupancy level for
both the DBE and MCE level events, surpassing the client’s performance requirements. The
LFRS remained fully elastic throughout the MCE event, except for onset of yielding that was
experienced in several panel zones. Based on FEMA 273 requirements for this LFRS, an
immediate occupancy level is achieved if the maximum beam rotation is less than 1.7% and the
maximum column rotation is less than 1.6%. A maximum rotation of approximately 0.9% for
both the RBS and column bases, occurred during the MCE level event, which indicates an
immediate occupancy level was achieved. The dampers were designed to possess a nonlinear
exponent of 0.4. This nonlinearity limits the increase in axial force above the design value
resulting from the MCE level event. Thus, a maximum DCR of 0.82 occurred in the damper
elements. Several of the panel zones had a DCR of approximately 1.1 under the MCE level
event. This slight overstress was considered acceptable, since no overstrength was considered
when determining the capacity of the panel zones. P-delta effects were checked for gravity
columns and proved negligible due to the displacement control provided by the FVD. Refer to
Table 2 for a summary of the analysis results.

Comparison to UBC-Designed Building

A three-dimensional model without dampers was built using ETABS and designed solely
based on the 1997 UBC, including drift requirements. Due to the drift requirements, the columns
required by the UBC were substantially larger than those of the building with FVD. Table 3
shows a comparison of the weight of the LFRS for the building without dampers to the weight of
the LFRS for the building with dampers. In addition to the increase in structural steel, the elastic
forces on the foundation elements are higher on the UBC building due to the decrease in period
of 30%.
 Assuming a cost of structural steel to be approximately $1/lb. ($2,000/ton), the total cost
of the LFRS for the UBC building is $1,765,000. The dampers for the project cost $929,000.
The total cost of the SEAOC building w/ FVD is $2,078,000. Therefore, the increase in the cost
of the SEAOC building with dampers is $313,000 or about $1.90/s.f.. This is approximately a
1% increase of total construction cost.

Table 2. Summary of nonlinear analysis results

Deformation
Controlled Force Rotation (rad) Performance
Components Level Level
RBS "hinge" DBE 0.0071 Immediate
Occupancy
MCE 0.0088 Immediate
Occupancy
Column base DBE 0.0077 Immediate
Occupancy
MCE 0.0089 Immediate
Occupancy
Force
Controlled Force Demand/Capacity Performance
Components Level (DCR) Level
Beam (not at RBS) DBE 0.65 Immediate
Occupancy
MCE 0.74 Immediate
Occupancy
Column above base DBE 0.74 Immediate
Occupancy
MCE 0.8 Immediate
Occupancy
Panel zones DBE 0.87 Immediate
Occupancy
MCE 1.09 Immediate
Occupancy
Viscous dampers DBE 0.67 Immediate
Occupancy
MCE 0.82 Immediate
Occupancy

Conclusions

For a cost of approximately $1.90/s.f., an immediate occupancy level at an MCE level event was
achieved, as opposed to a collapse prevention state for the UBC building. This is a relatively
small cost increase for the drastic increase in building performance. Life cycle analyses would
show that over the life of the building, the structure with dampers would be less expensive.
Therefore, the incorporation of FVD and the associated increase in seismic performance can be
very cost effective. This design philosophy could be incorporated into more traditional buildings
such as office buildings, commercial buildings, schools and hospitals. For these buildings, the
enhanced seismic performance would be very feasible and cost effective.

Table 3. Comparison of UBC building without Energy Dissipation System (EDS) to SEAOC
Building with EDS
UBC Building SEAOC Building
w/out EDS w/ EDS
Lateral columns W14x550 W14x211
Total weight (kips) 1452 557
Lateral floor beams W33x116 W33x116
Total weight (kips) 204 204
Lateral roof beams W21x62 W21x62
Total weight (kips) 109 109
Tube braces - TS10x10x5/8
Total weight (kips) - 175
Damper gussets - 1-1/2" Plates
Total weight (kips) - 104
Total weight of struct.
steel for LFRS (kips)/(tons) 1765/883 1149/575

References

CSI, 1999, ETABS – Integrated Design and Analysis Software for Building Systems,
Computers and Structures, Inc., Berkeley, CA.

ICBO, 1997, Uniform Building Code, International Conference of Building Officials,

Whittier, CA.

FEMA , 1997, NEHRP Guidelines for the Seismic Rehabilitation of Buildings, Federal
Emergency Management Agency, Washington, D.C.

Gouchon, J., 2000, Geotechnical Investigation- Martin Avenue, Treadwell and

Rollo, San Francisco, CA.

Naeim, F. and Kelly, J., 1999, Design of Seismic Isolated Structures, John Wiley & Sons,
Inc., New York, NY.

RAM International, 2000, RAM Perform 2D, RAM International, Carlsbad, CA.

SEAOC ,1999, Recommended Lateral Force Requirements and Commentary, Seismology

Committee, Structural Engineers Association of California, Sacramento, CA.

Singh, J.P., 1998, Site Specific Ground Motion for High-Tech Seismic Design, U.S.-Japan
Workshop on the Improvement of Structural Design and Construction Practices, Honolulu, HI