16th World Conference on Earthquake, 16WCEE 2017

Santiago Chile, January 9th to 13th 2017

Paper N° 2715
Registration Code: S-L1462914560

DESIGNING NONSTRUCTURAL COMPONENT ANCHORAGE USING

APPROXIMATE FLOOR RESPONSE SPECTRA

B. Kehoe(1), M. Hachem(2)
(1)
Associate Principal, Wiss, Janney, Elstner Associates, Inc., [email protected]
(2)
Senior Associate, Degenkolb Engineers, [email protected]
…

Abstract
The seismic behavior of nonstructural components is affected by the interaction of the dynamic behavior of the
nonstructural component and the building. Building code requirements for the design of nonstructural components provide
approximate procedures for estimating the floor accelerations of a building based on assumptions regarding the dynamic
behavior of the building. Since the dynamic behavior of many nonstructural components is not usually well known, building
code design methodologies provide only limited consideration of the building’s dynamic characteristics and their effect on
the nonstructural component’s dynamic behavior.
The interaction between nonstructural components and the building could be determined by the development of floor
response spectra, but the effort required to produce floor response spectra are typically not cost effective. A procedure has
been developed and recently implemented into the U.S. building codes that allows for an improved consideration of the
interaction of nonstructural components and the building. This procedure produces approximate floor response spectra
based on an elastic modal analysis of the building. This approximate floor response spectra procedure is described and
results are compared to the building code design requirements and the results using floor response spectra generated from
response history analyses using several types of nonstructural components for example buildings. Recommendations are
made for conditions where the use of the floor response spectra procedures would be beneficial.
Keywords: nonstructural, floor response spectra
 16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017

1. Introduction
The design of nonstructural components for seismic loading is required for buildings in areas of moderate to
high seismicity. The calculated seismic design force prescribed in building codes provides a simplification of the
complex interaction of the behavior of nonstructural components attached to a building. The actual response of a
nonstructural component to earthquake shaking depends in large part on both the vibrational characteristics of
the building and the nonstructural component.

2. Background
The design seismic forces for nonstructural components prescribed by building codes has traditionally been
based on simplified procedures. There are a number of justifiable reasons for using simplified design procedures
for seismic design of nonstructural components, as described by Kehoe (2014). These simplified design
equations are generally considered to be conservative estimates of the seismic effects on nonstructural
components and failures of nonstructural components are typically related to lack of seismic design or
inadequate installation rather than deficiencies in the design procedures. Little effort however has been made to
understand how well these procedures predict the actual response of nonstructural components.
2.1 Building Code Design
The design procedure for nonstructural components in ASCE 7-10 Minimum Design Loads for Buildings and
Other Structures (2010) and its predecessor documents prescribes a horizontal seismic force be calculated and
applied to the component. Eq. (1) shows the equation used to calculate the horizontal seismic force for a
nonstructural component.

(1)

Where:
Fp is the design lateral seismic force on the nonstructural component,
SDS is the design spectral acceleration at short periods
Ip is the component importance factor
ap is the component response amplification factor
Rp is the component response factor
z is the height of attachment of the nonstructural component to the building, and
h is the height of the building.

This equation makes a number of simplifying assumptions. The period of vibration of the nonstructural
component is not explicitly required nor are the periods of vibration of the building structure. Thus, nonstructural
components are prescribed as being flexible or rigid, as indicated by the assigned value of the component
response amplification factor. The variation in the floor accelerations for the building are assumed to be linearly
increasing over the height of the building, which is based on the dynamic behavior of the building response
being predominantly due to the first mode. The nonstructural force also does not account for differences in
building behavior in different orthogonal directions.

2.2 Floor Response Spectra

A common simplification for the evaluation of the seismic effects on nonstructural components mounted within
a building is the characterization of the nonstructural components as secondary components that are decoupled
from the overall dynamic behavior of the structure. Based on this assumption, the response of the structure can
be determined and the resulting response used as input to determine the response of the nonstructural
components within the building. This process typically involves subjecting a model of the building to a response
history analysis and extracting from that analysis the acceleration responses at each floor level of the building,

2
 16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017

which can in turn be used to calculate response spectra at each floor. These acceleration response spectra are
referred to as floor response spectra.
Although floor response spectra can provide an estimate of a floor’s acceleration response, this type of analysis
is rarely used for the design of nonstructural components due to the effort required. The floor response spectra
are only valid for the specific ground motion response used as building input, thus a suite of time history
responses would need to be performed and aggregated to develop an appropriate design floor response spectrum
for each floor.
2.3 Approximate Floor Response Spectra Method
An alternative method for determining the seismic design force for nonstructural components has been
introduced in the 2016 Edition of ASCE 7 Minimum Design Loads for Buildings and Other Structures. This
procedure is based on the methodology described by Kehoe and Hachem (2003).
A modal analysis of the building is performed to determine the period and mode shapes for the significant
modes of the building. The number of modes that need to be considered is dependent on a number of factors
including the characteristics of the earthquake and the modal characteristics of the building. For short period
buildings, a minimum of three modes in each principal orthogonal direction of the building is recommended. For
buildings with longer fundamental periods, the fourth mode and possibly the fifth mode should be considered.
For each mode of the building, the maximum floor acceleration at each floor is calculated separately for
each mode in each orthogonal direction as the product of the spectral acceleration for the mode times the
participation factor for the mode at the floor, as described in Eq. (2).
PFin = ϕin Ln/Mn (2)
Where:
PFin is the Participation Factor at level i for mode n,
ϕin is the Eigenvector value at level i for mode n,
Ln is the dot product of the mass matrix and Eigenvector for mode n, and
Mn is the effective modal mass.
Some software programs, such as ETABS and SAP 2000 (CSI 2015), define the modal participation factor
as Ln/Mn and normalize the mode shapes such that the effective modal mass, Mn, is unity.
For nonstructural components that are flexible or flexibly supported, the maximum floor acceleration is
amplified by a magnification factor to account for the resonance between the nonstructural component and the
building. The peak modal response for a flexible component is approximated to be 5.0 when the period of the
nonstructural component equals one of the periods of the building based on an assumed damping ratio of 5
percent for the nonstructural component, which is consistent with the assumptions made in the ASCE 7 equation.
No amplification is assumed when the period of vibration of the nonstructural component is more than a factor
of 2 or less than 1/2 of one of the building periods. Because of the uncertainty in the periods of the building and
the nonstructural component, the generalized magnification factor in Fig. 1 includes a broad plateau of plus or
minus 20 percent where the period of the nonstructural component matches one of the building periods. This is
similar to the recommendations for evaluating nonstructural components for the U.S nuclear industry, which
suggests broadening the spectra by plus or minus 15 percent.

3
 16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017

Fig. 1 – Nonstructural Dynamic Amplification Factor

For each building mode considered, the maximum floor acceleration is multiplied by the magnification
factor to create a modal floor response spectrum, as shown in Eq. (3).
Ain = PFin San DAF (3)
Where:
Ain is the component acceleration at level i for mode n,
PFin is defined in Eq. (2),
San is the spectral acceleration for mode n, and
DAF is the dynamic Amplification Factor from Fig. 1.
Since the generalized spectrum in Fig. 1 is normalized by the ratio of the period of the nonstructural
component to the building period (TP/TX), the horizontal axis of each of these spectra is multiplied by the period
of the mode to produce a set of spectra relating the nonstructural component period to the acceleration. The
spectra are then combined and the design spectrum is taken as the envelope of the maximum acceleration value
at each period from each of the modal spectra. The minimum design acceleration is spectral acceleration at the
base of the building.
For the design of nonstructural components, the component acceleration, Ain, accounts for three of the
terms in Eq. 1: the ground motion term (0.4 SDS), the component response amplification factor (ap), and the
variation of response over the height of the building (1+2x/h). Thus, the nonstructural component design force
using the approximate floor response spectra method is shown in Eq. (4), where the acceleration used is the
maximum modal response acceleration for the level.
Fp = Max(Ain)Wp (Rp/Ip) (4)

3. Example Analyses
The approximate floor response spectra procedure has been performed for two sample buildings. Each building
model was subjected to a series of earthquake ground motions and the floor response spectra were calculated.
The calculated floor response spectra from the response history analysis is compared to the results of the
approximate procedure.
3.1 Example Building Responses

4
 16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017

Mathematical models of each of the two sample buildings were analyzed. One of the buildings was modeled
using SAP 2000 from Computers and Structures Inc. (CSI) (2015) and one building was modeled using ETABS
from CSI (2015).
3.1.1 25-Story Steel-Framed Building
An analysis was performed for a 25-story steel building. The building is roughly rectangular in plan and the plan
shape changes configuration at two levels over the height of the building. The analysis model is shown in Fig. 2.
The lateral force resisting system for the building includes steel moment-resisting frames and diagonal steel
braced frames. The floors are constructed with steel beams supporting concrete-filled metal deck slabs. The
building is supported on concrete piles. The exterior walls include precast concrete spandrel panels and insulated
glass panels.

Fig. 2 – Model of 25-Story Building

The dynamic characteristics of the building are summarized in Table 1.

Table 1 – Dynamic Characteristics for 25-Story Sample Building

Mode Period (sec) Direction
1 3.169 Longitudinal
2 2.784 Transverse
3 2.249 Torsion

5
 16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017

4 1.057 Longitudinal
5 0.837 Transverse
6 0.727 Longitudinal
7 0.594 Torsion
8 0.442 Transverse

The 25-story building was analyzed under eight ground motion acceleration time histories. These time history
records represent the two horizontal components of four different recorded earthquake motions. The earthquake
records used are from two California earthquakes. The earthquake records used and the peak ground acceleration
(PGA) for each record in each direction are listed in Table 2.

Table 2 – Ground Motion Records Used

Earthquake / Recording Station PGA - direction 1 PGA - direction 2
1994 Northridge /Corralitos - Eureka Canyon Rd 0.63 g 0.46 g
1989 Loma Prieta / Hollister 0.37 g 0.17 g
1994 Northridge / Newhall Fire Station 0.57 g 0.63 g
1994 Northridge / Sylmar 0.84 g 0.60 g

To apply the approximate floor response spectrum method, the response spectra for each of the ground
motion records was used. Fig. 3 shows a plot of the response spectra.

Fig. 3 – Ground Motion Response Spectra

3.1.2 4-Story Steel-Framed Building
An analysis was performed for a 4-story steel building. The structure consists of a full basement level, a one-
story podium level, and three story tower floors. The analysis model is shown in Fig. 4. The lateral force
resisting system for the building primarily consists of a space steel moment-resisting frame with moment

6
 16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017

connections to the column in two directions. The floors are constructed with steel beams supporting concrete-
filled metal deck slabs.

Fig. 4 – Model of 4-Story Building

The dynamic characteristics of the building are summarized in Table 3.

Table 3 – Dynamic Characteristics for 4-Story Sample Building

Mass Ratios Cumulative Mass Ratios
Mode Period Direction UX UY RZ Sum UX Sum UY Sum RZ
1 0.718 Y+X+RZ 8% 24% 17% 8% 24% 17%
2 0.638 X+Y 24% 13% 9% 32% 37% 26%
3 0.513 X 7% 3% 3% 39% 40% 30%
4 0.351 X+RZ 15% 2% 18% 54% 42% 47%
5 0.327 Y 3% 17% 5% 57% 58% 52%
…
13 0.029 Y 0% 27% 1% 59% 86% 65%
14 0.025 X 31% 1% 6% 90% 88% 71%
15 0.023 Y+X 8% 11% 3% 97% 99% 74%

The computed modes generally exhibit significant coupling between the modes. This is likely caused by the
asymmetric configuration of the building and the coupling between torsional and longitudinal modes.

7
 16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017

The 4-story building was analyzed under one ground motion acceleration time history that was recorded during
the 2014 Napa earthquake, at CGS-SMIP Station 68294 in Vallejo, California. The record had peak ground
accelerations (PGA) of 0.22g and 0.47g in direction 1 (90º component) and direction 2 (360 º component),
respectively. A linear time history analysis was performed using the linear modal combination method, with a
damping ratio of 2% in all modes.
To apply the approximate floor response spectrum method, the response spectra for each of the ground
motion components was used. Figure 5 shows a plot of the response spectra under 2% and 5% damping.

Fig. 5 – Ground Motion Response Spectra

3.3 Nonstructural Component Response

To evaluate the component responses for the sample ground motions, the peak ground accelerations are used in
lieu of the 0.4SDS term in Eq. (1). Table 4 summarizes the design accelerations for each of the sample
components for each sample ground motion at the roof level using the ASCE 7 equation (hx = hr). The values in
Table 4 account for the peak ground acceleration and the component amplification factor aP at the roof level,
which produces the maximum design acceleration. Note that ASCE 7-10 considers the wall cladding and the
interior partitions as rigid components with a component amplification factor of 1.0 and the water pipe and
appendage are considered flexible components with a component amplification factor of 2.5.

Table 4 –Component Design Accelerations (g) Using ASCE 7-10 for Sample Ground Motions

Ground Component
Motion/Direction Wall Cladding Interior Partition Water Pipe Appendage
Corralitas 1 1.43 1.43 3.59 3.59
Corralitas 2 1.89 1.89 4.72 4.72
Hollister 1 0.53 0.53 1.33 1.33
Hollister 2 1.11 1.11 2.77 2.77

8
 16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017

Newhall 1 1.74 1.74 4.37 4.37

Newhall 2 1.76 1.76 4.42 4.42
Sylmar 1 1.81 1.81 4.53 4.53
Sylmar 2 2.53 2.53 6.32 6.32

The fundamental periods of vibration have been calculated for several typical nonstructural components as
described below.
3.3.1 Exterior Wall Cladding
The period of vibration for a precast concrete wall panel was calculated assuming that the panel acts as a simply
supported beam spanning between connections, which occur at the floor framing. The cladding is assumed to be
reinforced concrete, 12.7 cm (5 in) thick, with a compressive strength of 27.6 MPa (4000 psi). The vertical span
between supports is assumed to be 4 m (13 ft. 1 inch). The effective moment of inertia is assumed to be 0.35
times the gross moment of inertia, as permitted by ACI 318. Based on these assumptions, the fundamental period
of the claddings is 0.077 seconds. The component is considered flexible since the period is slightly larger than
0.067 second, which is considered to be the upper bound limit for a component to be considered rigid.
3.3.2 Interior Partition
The period of vibration for an interior, non-loadbearing partition was calculated. The partition is assumed to
consist of light-gage steel studs at a spacing of 40.c cm (16 inches). The partition is assumed to span vertically
with anchorage to the floor system at the base and lateral bracing along the top of the partition. The height of the
partition from the floor to the lateral bracing at the top is assumed to be 2.75 m (9 ft). The steel studs have an
effective moment of inertia of 45.7 cm4 (1.1 in4). The gypsum board wall sheathing is assumed have a thickness
of 134 Pa (2.8 psf). Based on these assumptions, the fundamental period of vibration of the partition is 0.067
seconds. The component is considered rigid since the period is considered to be equal to the upper bound limit
for a component to be considered rigid.
3.3.3 Water Pipe
The period of vibration of a horizontally oriented steel pipe containing water was calculated assuming that the
pipe acts as a simply-supported beam between lateral supports. The pipe is assumed to have a diameter of 10.2
cm (4 in) and a thickness of 0.57 cm (0.226 in). The distance between supports is assumed to be 3.5 m (11.5 ft)
and the bracing is assumed to be rigid. The piping is assumed to be connected with threaded joints. Based on
these characteristics, the period of vibration for the pipe is assumed to be 0.057 seconds. The component is
considered rigid since the period is less than the upper bound limit for a component to be considered rigid.
3.3.4 Appendage
The period of vibration was calculated for a sign mounted on the side of a building with cantilever beams. The
length of the cantilever is 1.5 m (4.9 feet). The cantilever beam is a square structural tube about 76 mm wide
with a thickness of about 6.3 mm. The tributary weight of the sign on the cantilever is 678 kg (1500 pounds).
Based on these characteristics, the period of vibration for the sign is 0.35 seconds. The component is considered
flexible since the period is larger than 0.067 second and therefore should be considered flexible.

4. Comparison of Results
4.1 25-Story Sample Building
For each of the ground motions, the floor response spectra were computed for the roof level and the thirteenth
floor in each direction. The approximate floor spectra were also computed based on the procedure described
above. Fig. 6 shows a comparison of the spectra for the roof level in each orthogonal direction computed based
on the Corralitas direction 2 ground motion histories applied in the longitudinal and transverse building

9
 16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017

directions to the approximate roof response spectrum. In the figure, the spectra for each of the predominant
modes in the transverse direction are shown with dashed lines and the envelope of maximum of the responses is
shown in the dark solid line.

Fig. 6 – Roof Spectra for 25-Story Sample Building in Longitudinal and Transverse Directions
The results of the approximate floor response spectra procedure are compared with the results using the
floor response spectra for each ground motion at three floor levels for the example nonstructural components
described above. Table 5 summarizes the results for three of the ground motions applied in the longitudinal
direction of the building.

Table 5 – Comparison of Component Accelerations (g) for 25-Story Sample Building

Component Floor Corralitas2 Corralitas2 Hollister2 Hollister2 Newhall2 Newhall2
Level Computed Approx. Computed Approx. Computed Approx.
Accel (g) Accel (g) Accel (g) Accel (g) Accel (g) Accel (g)
Wall Roof 0.52 0.55 0.40 0.46 0.94 1.03
Cladding
16th 0.41 0.55 0.33 0.34 0.79 0.76
8th 0.40 0.55 0.42 0.44 0.97 0.99
Interior Roof 0.50 0.53 0.39 0.46 0.92 1.03
Partition
16th 0.38 0.53 0.32 0.34 0.75 0.76
8th 0.40 0.53 0.46 0.42 0.95 0.99
Water Pipe Roof 0.51 0.49 0.38 0.46 0.90 1.03
16th 0.39 0.49 0.31 0.34 0.73 0.76
8th 0.42 0.49 0.42 0.44 0.92 0.99
Appendage Roof 0.89 0.75 0.52 0.45 1.18 1.39
16th 0.87 0.75 0.46 0.34 1.31 0.92
8th 1.20 0.75 0.64 0.44 1.33 1.39

4.2 4-Story Sample Building

For each direction of the building and corresponding record, the floor response spectra were computed for the
fifth floor (roof) and the third floor. The approximate floor spectra were also computed based on the procedure
described above. Fig. 7 and Fig. 8 show a comparison of the spectra for the roof and third floors in each

10
 16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017

orthogonal direction computed based on the Napa ground motion histories applied in both building directions
simultaneously. In the figures, the spectra for each of the predominant modes are shown with dashed lines and
the envelope of maximum of the responses is shown in the dark solid line. All floor spectra were computed using
5% damping.

Fig. 7 – Roof Spectra for 4-Story Sample Building in Longitudinal and Transverse Directions

Fig. 8 – Third Floor Spectra for 4-Story Sample Building in Longitudinal and Transverse Directions
The proposed procedure appears to provide a good estimate of floor spectra over a wide range of periods, despite
the somewhat irregular configuration of the building. At short periods, the procedure tends to underestimate the
spectral accelerations, though the discrepancy is significantly reduced when using the ground level spectrum as a
minimum, especially at lower floors. It is noted that lateral component of the ground motion had a sharp peak in
the response spectrum at around 0.1 second, which was not observed in the computed spectra at the upper floors.

5. Conclusions
A floor response spectrum provides the expected design acceleration for non-structural components. The floor
response spectra vary depending on the ground motion, the dynamic properties of the building, and the location
within the building. A procedure for developing approximate floor response spectra has been introduced into the
latest version of ASCE 7 that uses the design response spectra in lieu of response history records to generate the
design spectra. The approximate floor response spectra has been shown to provide a reasonable estimate of a
floor response spectra by accounting for the amplification effects of each of the significant modes of vibration of
the building. The computed floor response spectra occasionally produces design accelerations that exceed the
approximate floor response spectra. This variation is due to peaks in the response spectra that are smoothed out
when developing response spectra for design. In this study, the base spectra tends to control the floor spectra at
short periods and the approximate floor response spectra may underestimate the component acceleration for
flexible components, but provides a good estimation of the component acceleration for rigid components.

11
 16th World Conference on Earthquake, 16WCEE 2017
Santiago Chile, January 9th to 13th 2017

The ASCE 7 component amplification factors may not accurately represent components that should be
categorized as rigid or flexible based on the actual dynamic properties of the component. The floor response
spectra computed using actual ground motions however, indicates that the upper bound period limit of 0.067
seconds for a component to be considered rigid may be increased since very little amplification of component
accelerations were found for components with periods up to about 0.10 seconds.
The results also showed that the ASCE 7 equation for designing nonstructural components overestimates
the design acceleration values across the range of component periods. These results also confirm previous
findings that for long period buildings there is less amplification of the horizontal acceleration over the height of
the building than prescribed by the ASCE 7 design equation.

6. References
[1] ASCE 4-98 (2000) Seismic Analysis of Safety-Related Nuclear Structures and Commentary,
American Society of Civil Engineers, Virginia, USA.
[2] ASCE/SEI. ASCE/SEI 7-16 (2016 anticipated) Minimum Design Loads for Buildings and Other
Structures, American Society of Civil Engineers, Virginia, USA.
[3] Computers and Structures, Inc. (2015), CSi Analysis Reference Manual, Computers and Structures,
Inc., Berkeley, California, USA
[4] Kehoe, BE (2014) “Defining Rigid vs. Flexible Nonstructural Components”, Proceedings of the 10
US National Conference on Earthquake Engineering, Anchorage, Alaska, USA.
[5] Kehoe BE, Hachem M, Procedures for Estimating Floor Accelerations (2003), ATC 29-2 Seminar
on Seismic Design, Performance, and Retrofit of Nonstructural Components in Critical Facilities,
Applied Technology Council, California, USA.

12