Nuclear Engineering and Design 374 (2021) 111046

Contents lists available at ScienceDirect

Nuclear Engineering and Design

journal homepage: www.elsevier.com/locate/nucengdes

Seismic response of electrical equipment subjected to high–frequency

ground motions
Sugandha Singh, Abhinav Gupta *
Center for Nuclear Energy Facilities and Structures, North Carolina State University, USA

A R T I C L E I N F O A B S T R A C T

Keywords: Recent ground motion studies conducted at nuclear facilities indicate that in Central and Eastern United States,
High frequency ground motions the ground motion response spectra exceed the safe–shutdown earthquake spectra at high frequencies. Even
Seismic response of electrical equipment though high–frequency ground motions do not cause structural damage, high–frequency accelerations may
Cabinets and control panels
propagate through the structure and interfere with the output of the safety–related equipment such as relays
Nonlinear mounting of cabinets
Effect of localized nonlinearities
required to ensure safe shutdown of the plant during a seismic event. Hence, it is essential to seismically qualify
electrical equipment subjected to high–frequency accelerations. The seismic qualification is conducted by
comparing in–cabinet response spectra evaluated at the equipment locations with the capacities of equipment
obtained from shake table tests. The in–cabinet response spectrum is obtained from analysis of nuclear power
plant building and electrical cabinet in which the equipment are mounted. This study is based on the hypothesis
that the high–frequency motions do not reach the equipment as the small displacements induced by such motions
are filtered out by the geometric nonlinearities. The effect of two different types of nonlinearities are studied: (1)
a gap in the connection between electrical cabinet and floor; (2) sliding friction between electrical cabinet’s base
and the floor. The results from an analyses of various different cases show that the high–frequency motions do
not reach the relays if the maximum displacement of building floor is less than the gap. Even if the displacement
is larger than the gap, the in–cabinet spectral accelerations are not excessively high. On the contrary, results from
a conventional linear analysis give excessively high unrealistic spectral accelerations.

1. Introduction ground motions have high-frequency content due to the presence of hard
rock soil profile unlike the motions recorded in Western United States.
1.1. Purpose and motivation The frequency content of ground motion response spectra in CEUS
ranges from 15 to 45 Hz as compared to the frequency content of
A nuclear plant should be able to safely shutdown and maintain USNRC’s Reg. Guide 1.60 design safe shutdown earthquake (SSE)
necessary operations to avoid core melt during or after an earthquake. spectra which ranges from 1 to 10 Hz (USNRC, 2014). An example of the
The safety-related electrical equipment are critical for safe shutdown of difference in response spectra at a site in CEUS is shown in Fig. 1. The
plant and thus, are required to maintain their functionality. Following seismic hazard studies in accordance with NTTF recommendation
the damages caused by the earthquake and subsequent tsunami at required the nuclear plants to evaluate the vulnerability of their SSCs to
Fukushima-Daiichi nuclear power station in Japan, USNRC appointed a high-frequency ground motions.
Near-Term Task Force (NTTF) to review the insights from Fukushima EPRI (2007) shows that the high-frequency ground motions do not
Daiichi accident and provide recommendations for enhancing the safety cause structural damage to the power plant or heavy industrial struc­
of nuclear power plants in the United States (USNRC, 2014). NTTF’s tures and equipment because the frequency ratio (the ratio of frequency
Recommendation 2.1 requires the licensees to reevaluate the seismic of excitation to the natural frequency of the system) is much greater than
and flooding hazards at their sites and update the structures, systems, unity resulting in a low value of dynamic amplification factor and hence,
and components (SSCs) to safeguard against the updated hazards if lower response. On the other hand, structures with higher natural fre­
necessary. Recent studies (SSHAC, 1997; EPRI, 2013; PEER, 2015) quencies do not undergo large relative displacements or stresses due to
indicate that in Central & Eastern United States (CEUS), earthquake small displacement amplitude of high-frequency ground motions.

* Corresponding author.
E-mail address: agupta1@ncsu.edu (A. Gupta).

https://doi.org/10.1016/j.nucengdes.2020.111046
Received 31 August 2020; Received in revised form 17 December 2020; Accepted 28 December 2020
Available online 29 January 2021
0029-5493/© 2021 Elsevier B.V. All rights reserved.
S. Singh and A. Gupta Nuclear Engineering and Design 374 (2021) 111046

Fig. 1. Ground motion response spectrum and safe–shutdown earthquake

Fig. 2. Ground motion response spectrum used in the analysis.
spectrum in CEUS.

However, the high-frequency ground motions may propagate through Table 1

the structure and the electrical cabinets to the safety-related electrical Maximum ground acceleration and displacement.
devices such as relays. The output signal of relays is important for safe Acceleration (g) Displacement (mm)
shutdown of the plant and may be influenced by the high-frequency
Low-Frequency High-Frequency Low-Frequency High-Frequency
content in ground motions. Usually, relays are tested on shake tables
before being used in nuclear facilities (EPRI, 2014; EPRI, 2015; IEEE, 0.159 0.415 5.986 11.007

2013). Historically, the relays have not been tested for excitations with
frequencies higher than 16 Hz. The acceleration sensitivity of relays to earthquake (1952). TAFT earthquake has frequency content between 1
high frequency accelerations is not well understood (EPRI, 2015). There and 6 Hz and peak spectral acceleration occurs at 2.8 Hz. The high-
has been limited observations that certain electrical devices that were frequency ground motion used in the analysis is obtained from the
seismically qualified for low frequency ground motions did not perform design spectrum of a nuclear power station on the Eastern US. It has a
their intended function during high frequency shaking (EPRI, 2007). frequency content in the range of 12–55 Hz and peak spectral acceler­
Some of the more recent guidelines particularly in the non-nuclear ap­ ation occurs at 35 Hz. The ground motion response spectra for both
plications require testing of nonstructural systems up to 33 Hz but these motions are shown in Fig. 2 in which both the grounds motions are
guidelines are applicable to only newer facilities and are not necessarily normalized to have peak ground acceleration of 1 g. The peak ground
applicable to nuclear equipment. acceleration and displacement are given in table 1.
Engineers at a nuclear power plant qualify the electrical devices by
evaluating the seismic demands inside the electrical cabinets and
comparing it with the capacities. The method to evaluate seismic de­ 1.2. Summary of existing studies
mands amplifies the floor response spectrum or in–structure response
spectrum (ISRS). Amplification of ISRS due to the electrical cabinet is Many studies (Stafford, 1975; Djordjevic and O’Sullivan, 1990;
based on the type of the cabinet. The amplified spectrum is called Djordjevic, 1992; Rustogi and Gupta, 1998; Gupta et al., 1999; Yang
in–cabinet response spectrum (ICRS) which gives the seismic demands et al., 2002; Rustogi and Gupta, 2004; Sankaranarayanan, 2007; Cho
on the devices like relays. The method suggested in EPRI (2015) uses et al., 2011; Vlaski, 2013; Herve, 2014; Gupta et al., 2019) have focused
empirical data to calculate the amplification factors. The approach ig­ on understanding the seismic behavior of electrical cabinets to deter­
nores many factors such as cabinet mounting arrangement, location of mine the factors that influence the in-cabinet response spectra. A sum­
equipment on the cabinet, geometric nonlinearities, etc. In general, the mary of the main observations made in these studies is discussed below
use of empirical amplification factors have been found to be inaccurate in order to understand the dynamic behavior of cabinets.
which can often render some devices seismically unacceptable for use in Gupta et al. (1999) proposed a method for evaluating in-cabinet
a plant. The extensive seismic probabilistic risk assessment studies response spectra in which the response of a cabinet is represented by
(SPRA) conducted by the plants in United States as an outcome of the only a few significant modes. A significant mode which governs ICRS
NTTF’s Recommendation 2.1, experienced that several relays in the can either be a global cantilever mode of the cabinet or a local mode of
plants could not be seismically qualified because the ICRS generated the door, internal frame or back wall depending on where the relay is
using simple amplification of ISRS with high frequency content had very mounted. These mode shapes are calculated by Rayleigh-Ritz method to
large spectral accelerations. represent the cabinet behavior. However, for some cabinets, the signif­
This paper presents a preliminary assessment of the effect of local­ icant global mode can also occur due to rigid body rocking of cabinet.
ized nonlinearities that exist in the cabinet mounting arrangements at The rigid body rocking of cabinet is explored in more detail by Yang
the base. It is intended to evaluate whether localized nonlinearities at et al. (2002) and Makris and Zhang (1999). These studies propose for­
the base due to gaps that exist around anchor bolts would filter out the mulations to evaluate rocking stiffness for different mounting arrange­
small displacements associated with high frequency ground motions. ments in cabinets which facilitates incorporation of cabinet rocking
The study is based on observations from the analysis of simple repre­ mode in Ritz vector approach. Rustogi and Gupta (2004) use modal data
sentative systems and compares the ICRS obtained from a linear analysis obtained from shake table and in-situ tests of cabinets to validate finite
to that from a nonlinear analysis. Two different ground motions are used element analysis and develop a basis for the Ritz vector approach. Gupta
in the analysis. Since low-frequency ground motions are also likely to and Yang (2002) modify the originally proposed Ritz vector approach by
occur in various regions, the systems are subjected to a low-frequency incorporating cabinet rocking mode and a partial rotational constraints
ground motion in addition to a high-frequency ground motion. The provided by structural members in a cabinets.
low-frequency ground motion used in the analysis is the TAFT The studies mentioned above do not consider two important aspects
associated with rocking behavior. First, when a cabinet is anchored and

2
S. Singh and A. Gupta Nuclear Engineering and Design 374 (2021) 111046

Fig. 3. Local mode shapes of cabinet door. Fig. 4. Local mode shape of cabinet internal frame.

undergoes rocking, there is likely to be a gap between cabinet base and

anchor bolt head. It is, thus, hypothesized that since the displacements
caused by high-frequency ground motions are relatively small, these
displacements can be filtered out due to the gap and consequently the
high-frequency ground motion do not reach the relays. Herve (2014)
and Vlaski (2013) discuss that high-frequency vibrations due to an
aircraft impact filter out before reaching equipment due to the presence
of small gaps. Secondly, the hole for the bolt at cabinet base plate is
usually larger than the bolt diameter. When a seismic motion acts at the
cabinet base, the cabinet initially acts as fixed-base (linear-elastic) sys­
tem until the friction force resists the seismic force. When seismic force
exceeds the maximum friction force, the base may slip from its position
causing rigid body displacement of the cabinet. This results in a
nonlinearity.
Herve (2014) studies the propagation of high-frequency vibration in
the structure, mainly caused due to an aircraft impact. The work by Fig. 5. Cup–like deformation of cabinet base plate around the anchor bolt.
Herve (2014) builds on the experimental observations by Vlaski (2013)
that high-frequency vibrations tend to filter out before reaching the cabinet. Figs. 3 and 4 show the mode shapes of local modes for a cabinet
equipment. Herve (2014) proposes a method to model and analyze a door and internal frame, respectively.
nonlinear elastic system in which the system has a 1 mm gap in the Llambias et al. (1989) illustrate the significance of modeling the
spring-mass-damper system. The model proposed by Herve (2014) forms structural details of a cabinet base for evaluating accurate ICRS. It
the very basis of the research presented in this paper. It is extended to studied two electrical cabinets experimentally as well as analytically to
study the effects of different gap lengths in bolted connections (1 mm, 5 understand their seismic behavior and calculate in-structure amplifica­
mm) subjected to both low-frequency as well as high-frequency ground tions. At low amplitudes of excitation, both experimental and analytical
motions. The true nature of interaction between the anchor bolt and the results predicted same natural frequencies and mode shapes. At higher
cabinet is quite complex in nature and involves multiple different as­ amplitudes of excitation, however, the experiments show that the nat­
pects. An attempt has been made in this paper to address one of these ural frequency of cabinet decreases with increase in the level of exci­
aspects that can induce impact and thereby lead to impact induced high tation. The cabinets behave in a nonlinear manner since the first mode is
frequency vibrations which can then be magnified by earthquake mo­ observed to be a rigid body rocking mode. Similarly, Lee et al. (1990)
tions containing high frequencies. address the effects of in situ shimming at the cabinet base on the seismic
response. Their study shows that shim plates cause a discontinuity at the
2. Seismic behavior of electrical cabinets base of the cabinet. Thus, the stiffness of cabinets is less than that in
shake table tests where the boundary condition is predominantly fixed.
Gupta et al. (1999) show that only one or a few significant modes are Moreover, Lee and Abou-Jaoude (1992) study the effects of base uplift
sufficient to evaluate accurate ICRS. This observation is based on com­ on cabinet response by modeling an SDOF system supported on a rigid
parison of ICRS obtained by using all the modes and by using only one beam which in turn is supported by two vertical compression springs.
significant mode. It is shown that the in-cabinet response spectra ob­ Their study shows the importance of modeling base flexibility in
tained from both the analyses are almost identical. This observation is determining dynamic characteristics of the cabinet.
then used to propose mathematical functions and Rayleigh-Ritz method In Ritz vector approach (Gupta et al., 1999), mathematical functions
to represent cabinet behavior by a simple two degree of freedom system. are used to represent the global bending mode. The global rocking mode,
The significant mode depends on various factors such as the location of however, cannot be considered without a knowledge of the rocking
relays, dynamic characteristics of a cabinet, etc. The location of relays stiffness at the base of cabinet. Thus, Gupta and Yang (2002) studied
inside the cabinet plays an important role in evaluating the in-cabinet three types of widely used mounting arrangements for cabinets and
response spectra. Relays can be mounted on any component of the developed formulations for calculating rocking stiffness imparted by
cabinet such as cabinet door, internal frame, back wall, etc. The fre­ these arrangements. The formulations are developed based on the re­
quency at which maximum amplification of in-cabinet response spectra sults of experimentally validated finite element models. The mounting
occurs may or may not coincide with the fundamental frequency of the arrangements of cabinets typically have a base plate fixed to the floor by

3
S. Singh and A. Gupta Nuclear Engineering and Design 374 (2021) 111046

discussed earlier, the displacements due to high-frequency ground mo­

tions are relatively very small. Therefore, this study is conducted as a
preliminary exploration to evaluate the hypothesis that the high-
frequency ground motions would filter out due to the existing gaps be­
tween the cabinet base plate and the anchor bolts.
Nonlinearity may also occur due to friction that may exist at the
cabinet base. This nonlinearity occurs because the anchor bolt may
exhibit some gap to allow cabinet sliding. Either the hole for the bolt at
cabinet base plate is a little larger than the bolt diameter or the base
plate may experience a tear around anchor bolt as has been observed
experimentally. Hence, the cabinet acts as a fixed-base structure (line­
arly elastic) as long as the seismic force is less than the friction force.
When the seismic force exceeds the friction force, the cabinet base can
slip from its position causing a rigid body motion of the cabinet resulting
in a nonlinearity due to friction. These nonlinearities can influence the
propagation of the high-frequency seismic motions and the in-cabinet
Fig. 6. Cabinet mounting arrangement.
response spectra.

anchor bolts. The base plate undergoes localized cup-like deformation as

4. Nonlinear models at cabinet base
shown in Fig. 5 which is a region of high base plate curvature. The base
plate curvature is negligible in the regions far from the anchor bolts and
In this study, we propose two models to represent the nonlinear
thus do not contribute to the rotational stiffness. Rustogi and Gupta
seismic behavior of cabinets and understand their influence on the fre­
(2004) compare results of in-situ and shake table tests on cabinets with
quency content of motions that reach relays. The first model represents a
those obtained from the Ritz vector approach as originally proposed.
gap that may exist between the cabinet base plate and the bolt head. The
The main differences between the tests and the analysis occur due to the
second model represents the effect of sliding friction between the bolt
presence of global rocking mode in experimental studies. Rocking occurs
and the base plate. A detailed discussion on these models is presented
when the overturning moment due to inertial forces is more than the
below.
restoring forces of gravity about an edge.

3. Proposed Premise/Hypothesis 4.1. Model-I: Gap between cabinet base and anchor bolt

As discussed above in detail, the seismic behavior of a cabinet de­ This model is used to represent the gap as shown in Fig. 6 in between
pends on the dynamic properties of the cabinet itself as well as the the bolt head and the cabinet base plate. Herve (2014) considered a
structural configuration of its mounting arrangement. Cabinets mounted spring-mass-damper system with a gap of 1 mm which is based on the
with anchor bolts undergoes rigid body rocking which usually affects the recommendations by EPRI (2007) and International Atomic Energy
fundamental mode of the cabinet. The rigid body rocking mode may be Association (IAEA, 2012) as the cut-off displacement for the case of
the significant mode by itself or in combination with a local mode. These high-frequency motions due to an airplane impact. The force–displace­
factors affect the resulting in-cabinet response spectra and hence the ment relationship used in their study to represent this model considers a
frequency content of cabinet acceleration at the equipment level. This gap of 1 mm in both the tension and the compression.
paper focuses solely on the effects of rigid body rocking and sliding of Contrary to the recommendation by Herve (2014), the gap associated
cabinets. Hence, the cabinets models studied in this paper are idealized with sliding or rocking behavior in a cabinet would exhibit zero force
cases to understand the equipment response due to such factors. When a only in tension. The behavior in compression between the cabinet base
cabinet base is anchored to the floor and the cabinet undergoes rocking, plate and the anchor bolt is linear elastic with no gap. Furthermore, we
there can be a gap between the cabinet base plate and the bolt head. As consider two different magnitudes of gaps, 1 mm and 5 mm. Fig. 7 shows
the force–displacement relationship of this model. Fig. 6 shows that for

Fig. 7. Force–displacement relationship for cabinet model–I.

4
S. Singh and A. Gupta Nuclear Engineering and Design 374 (2021) 111046

Fig. 8. Force–displacement relationship for cabinet model–II.

displacement between 0 mm and a predetermined gap, there is zero friction force, t is the time instant, Δt is the time step used in the analysis
resisting force in case of tension whereas for displacements more than and f(t) is the spring force at time ‘t’.
the gap, the resisting force follows linear elastic relationship. Hence, it is
a nonlinear elastic model that exhibits geometric nonlinearity. The 5. Application case studies
model follows a simple set of equations of motion which are described
by Eqs. (1) and (2). In the equations below, m denotes the mass, c is the In order to achieve the objective of this study, analysis of various
damping constant, k is the stiffness, ugap is the gap length (1 mm or 5 cases are conducted by creating representative building-cabinet systems
mm), u is the displacement, u̇ is the velocity, ü is the acceleration of the and considering two different earthquake ground motions. For
cabinet, respectively. Also, üf is the floor acceleration which is an input simplicity and for facilitating appropriate interpretation of the phe­
at the base of cabinet. nomenon, both the building and the cabinet are represented by a
respective SDOF system. The two ground motions considered in this
mü + cu̇ = − müf , 0 ≤ u ≤ ugap (1)
study are same as those given in Fig. 2. Even though Fig. 2 shows
spectrum curves normalized to a PGA of 1 g, the actual motions used in
mü + cu̇ + ku = − müf , u < 0 (2)
the numerical study correspond to the actual PGA values for these mo­
tions. The low frequency ground motion represented by the Taft earth­
mü + cu̇ + k(u − ugap ) = − müf , u > ugap (3)
quake record has a PGA of 0.159 g and that for the high frequency
ground motion, represented by the actual design spectrum of a nuclear
power plant on Eastern US, is equal to 0.415 g. At first, it may appear
4.2. Model-II: Friction effect
inappropriate to consider different PGA values for the two motions.
However, the problem being studied is a nonlinear problem in which the
Following USNRC’s Near Term Task Force (NTTF) recommendations
behavior is highly dependent upon the amplitudes of displacements.
2.1 and 2.2, many plants in CEUS have completed their Seismic Proba­
Therefore, the actual motions are used without any modifications. As
bilistic Risk Assessment (SPRA) and found that the relays in many cab­
noted in Fig. 2, the peak of the spectrum curve for the low frequency
inets do not qualify based on the acceptable risk goals. One of the
earthquake record occurs around 3 Hz and that for the high frequency
solutions suggested and pursued by the industry is based on utilizing the
motion close to 35 Hz. Consequently, two different variations of the
ductility (elastic-perfectly plastic) considerations. The model-II repre­
building and the cabinet are considered. Typical buildings in nuclear
sents the effect of friction on seismic behavior of cabinets (Kon­
power plants have low frequency modes for which the frequencies lie in
stantinidis and Nikfar, 2015). As discussed earlier, the diameter of bolt
the range of 1 to 5 Hz. Therefore, a 3 Hz SDOF system is selected to
hole in the base plate is often a little larger than the bolt diameter.
represent a low–frequency building (or a low–frequency cabinet).
Consequently, the friction forces would act at the interface of the base
Further, a 35 Hz SDOF system is considered to represent a high­
plate and the floor before sliding begins. The cabinet would act as a
–frequency building (or a high–frequency cabinet). One may argue that
linear elastic material when the seismic force is less than the friction
the typical building frequencies are much less than 35 Hz. However, it
force. When the seismic force becomes greater than the friction force,
must be noted that every building will have higher order modes in the
the cabinet would exhibit a slip in the direction of motion. As the di­
frequency range of interest in this study. These modes will exhibit
rection of motion changes, the cabinet would again exhibit a linear
appreciable amplifications when subjected to a high frequency ground
elastic behavior until the time seismic force exceeds the friction force.
motion. The high frequency building modes contribute primarily to the
Thus, the behavior is primarily hysteretic in nature and identical to that
floor motions at lower elevations in a building. Similarly, there are
of an elastic-perfectly plastic material. Fig. 8 shows the force–displace­
hundreds of cabinets in a nuclear plant that are located at various
ment relationship for this model. The equations of motion for this model
different elevations throughout the building. Typically, a global rocking
can be written as follows:
cabinet mode is a low frequency mode. A local mode of the cabinet panel
mü + cu̇ + ku = − müf , |f (t)| ≤ fs max (4) or frame can be either a low frequency or a high frequency mode.
Therefore, the high frequency systems are meant to be representative of
mü + cu̇ + (ku − fs max ) = − müf , |f (t)| ≥ fs max (5) higher order modes for both the buildings and the cabinets. Typical
damping ratios for concrete buildings such as auxiliary building and
When the direction of velocity changes,
service building in nuclear plants are on the order of 5%. Therefore, the
f (t + Δt) = f (t) + (k(u(t + Δt) − u(t) ) ) (6) damping ratio of building is taken as 5%. Similarly, the damping ratio
for cabinets is typically taken as 2% in almost all plants and their SPRA
mü(t + Δt) + cu̇(t + Δt) + f (t + Δt) = − müf (t + Δt) (7) studies. Therefore, the damping ratio for cabinets is taken as 2%.
While the building model is linear-elastic in all cases, different
where k is the stiffness before cabinet slips, fs_max is the maximum static

5
S. Singh and A. Gupta Nuclear Engineering and Design 374 (2021) 111046

Fig. 9. Steps for analyzing in–cabinet response spectrum for each case.

nonlinear models are used to represent the cabinet. A total of four

Table 2
cabinet models are considered to represent different scenarios. One of
Maximum Total Floor Acceleration in ‘g’ units.
these four cabinet cases correspond to the traditional/conventional case
of linear-elastic or fixed-base model. In addition, two cases correspond Ground Motion Low-Frequency Building High-Frequency Building
to a gap nonlinearity. One considers a small gap of 1 mm whereas the Low-Frequency 0.483 0.161
other a relatively larger gap of 5 mm. The fourth case corresponds to the High-Frequency 0.248 1.125
elastic-perfectly plastic model representing the effect of friction between
the cabinet base plate and the floor. The floor motion, evaluated from an
• Case-5 (L–H–L): Low–frequency earthquake, High–frequency (35
analysis of SDOF building model, is used as an input at the base of a
Hz) building, and low–frequency (3 Hz) cabinet.
cabinet (Katona et al., 1995). Relatively, a very small time-step of
• Case-6 (H–H–L): Same as Case-5 for high-frequency earthquake
integration equal to 0.0001s is used in this study in order to appropri­
input.
ately consider the effects of high frequency ground motions. Fig. 9 il­
• Case-7 (L–H–H): Low–frequency earthquake, High–frequency (35
lustrates the various steps in the analysis of a single system
Hz) building, high–frequency (35 Hz) cabinet.
where M denotes the lumped mass of building or cabinet in kg; f
• Case-8 (H–H–H): Same as Case-7 for high-frequency earthquake.
denotes the frequency of building or cabinet in Hz; and ξ denotes the
percent damping ratio of building or cabinet. The nonlinear analysis for
For each of these cases, both the linear and nonlinear models are
all the cases is conducted in OpenSEES (McKenna and Fenves, 2001).
considered for comparison purposes.
Overall, the various cases considered in this study can be enumerated as:

6. Discussion of results
• Case-1 (L–L–L): Low–frequency earthquake, Low–frequency (3 Hz)
building, and Low–frequency (3 Hz) cabinet.
The results from the analyses for various application cases enumer­
• Case-2 (H–L–L): Same as Case-1 except that the ground motion is
ated above are discussed in this section. The observations from these
high–frequency earthquake.
results are used to evaluate the validity of the hypothesis presented
• Case-3 (L–L–H): Low–frequency earthquake, Low–frequency (3 Hz)
earlier. To begin with, the two types of building models, low–frequency
building, and high–frequency (35 Hz) cabinet.
building with natural frequency of 3 Hz and high–frequency building
• Case-4 (H–L–H): Same as Case-3 except that the ground motion is
with natural frequency of 35 Hz are analyzed for both the low-frequency
high-frequency earthquake.
and the high-frequency ground motions. The peak floor acceleration for
each building induced by both ground motions are shown in Table 2.

6
S. Singh and A. Gupta Nuclear Engineering and Design 374 (2021) 111046

Table 3
Maximum Total and Relative Floor Displacement in ‘mm’ units.
Relative Displacement Total Displacement

Ground Motion Low-Frequency Building High-Frequency Building Ground Motion Low-Frequency Building High-Frequency Building

Low-Frequency 1.352 0.0033 Low-Frequency 6.134 5.99

High-Frequency 0.694 0.0232 High-Frequency 11.08 11.024

Fig. 10. ISRS of low–frequency building subjected to both ground motions. Fig. 12. Comparison of fixed-base cabinet ICRS with ISRS for case 1 (L–L–L)
and case 2 (H–L–L).

Fig. 11. ISRS of high–frequency building subjected to both ground motions.

Fig. 13. Comparison of fixed-base cabinet ICRS with ISRS for case 3 (L–L–H)
Further, maximum total and relative floor displacements are shown in and case 4 (H–L–H).
Table 3. The in-structure response spectra (ISRS) for a low–frequency
building subjected to both the low and high frequency ground motions high–frequency ground motions.
are shown in Fig. 10. As anticipated, the amplifications are much greater Fixed-Base Case-1 (L–L–L) and Case-2 (H–L–L): Next, a fixed-base
for the case of low frequency ground motion even though the PGA of low linear-elastic cabinet is subjected to floor motions from different
frequency ground motion is only 0.159 g compared to 0.415 g for high buildings. The resulting cabinet acceleration time histories are then used
frequency ground motion. Moreover, the ISRS amplitude at the peak (3 to generate the in-cabinet response spectra (ICRS). Fig. 12 compares the
Hz) is less for high frequency ground motion because the input spectra ICRS for Case-1 (L–L–L) and Case-2 (H–L–L). It also illustrates the rela­
for high–frequency ground motion shown in Fig. 2 has smaller spectral tively high spectral accelerations in ICRS compared to the corresponding
accelerations at 3 Hz as compared to that of low–frequency ground values in the in-structure response spectra (ISRS). The primary reason
motions. for high spectral accelerations is the tuning of the building frequency
The floor response spectra for the case of a high–frequency building with the cabinet frequency.
subjected to both the ground motions are compared in Fig. 11. As seen in Fixed-Base Case-3 (L–L–H) and Case-4 (H–L–H): Fig. 13 compares the
this figure, the building truly behaves rigidly for the low frequency ICRS for these two cases of fixed base cabinets. As seen in this figure, the
ground motion and there is no amplification as such. The floor spectrum spectral accelerations in ICRS are relatively much less than those
is almost identical to the ground motion response spectrum (GMRS). observed in Fig. 12. This observation is due to detuning of cabinet (rigid)
However, the amplifications are significantly larger in the case of high and building frequencies. It can be seen that the spectral acceleration for
frequency ground motion. The 35 Hz building which is conventionally the case of high frequency motions are relatively less than those for low
considered to be rigid actually resonates with the input motion. The ISRS frequency motion which is anticipated given that the building frequency
amplitude at the peak (35 Hz) is less for low-frequency ground motion is in resonance with the peak of low frequency GMRS.
because the input spectra for low–frequency ground motion shown in Fixed Base Case-5 (L–H–L) and Case-6 (H–H–L): The various curves
Fig. 2 has smaller spectral accelerations at 35 Hz as compared to that of for ICRS and ISRS for these two cases of fixed base cabinets are

7
S. Singh and A. Gupta Nuclear Engineering and Design 374 (2021) 111046

Table 4
Maximum Total Cabinet Acceleration in ‘g’ units.
Case Number Linear 1 mm Gap 5 mm Gap Elastic Perfectly Plastic

1 (L-L-L) 4.747 5.583 1.592 2.116

2 (H-L-L) 3.349 1.118 0.471 0.216
3 (L-L-H) 0.485 4.993 2.192 0.419
4 (H-L-H) 0.26 3.121 0.317 0.214
5 (L-H-L) 0.715 0.475 0.722 0.638
6 (H-H-L) 0.426 0.286 0.229 0.325
7 (L-H-H) 0.168 2.088 0.762 0.169
8 (H-H-H) 15.83 1.715 0.819 0.324

Table 5
Maximum Total Cabinet Displacement in ‘mm’
Case Number Linear 1 mm Gap 5 mm Gap Elastic Perfectly Plastic
Fig. 14. Comparison of fixed-base cabinet ICRS with ISRS for case 5 (L–H–L)
and case 6 (H–H–L). 1 (L-L-L) 14.85 18.085 10.891 9.456
2 (H-L-L) 11.107 5.134 9.178 6.502
3 (L-L-H) 6.135 7.028 8.864 5.947
4 (H-L-H) 5.407 5.261 5.933 5.407
5 (L-H-L) 6.163 7.095 11.244 6.388
6 (H-H-L) 11.107 11.396 12.315 11.107
7 (L-H-H) 5.99 6.875 7.027 5.991
8 (H-H-H) 11.21 10.562 9.492 10.203

pulses resonate with the low–frequency cabinet and therefore we

observe amplifications in the ICRS in the vicinity of 3 Hz.
Fixed Base Case-7 (L–H–H) and Case-8 (H–H–H): These cases form the
very basis of conducting this study. The high–frequency building –
high–frequency cabinet cases are representative of a higher order
building mode and higher frequency cabinet mode. The complete system
is rigid, and no amplification is observed between GMRS, ISRS, and ICRS
when subjected to low frequency ground motion. This is shown in
Fig. 15. However, when this system is subjected to a high frequency
Fig. 15. Comparison of fixed-base cabinet ICRS with ISRS for case 7 (L–H–H). motion, the amplifications observed in ICRS are excessively large.
Fig. 16 shows the corresponding spectra. As seen in this figure, the
spectral accelerations at the peak of ICRS are almost unimaginable in
real-life. Granted that this is only a simple system and that the ampli­
fications illustrated in this example are much greater than what might be
observed in a real-life system. Even if the corresponding amplifications
in the real-life cases are less than those observed in Fig. 16, the ampli­
fications will be fairly high which can be explained as follows. For a
SDOF system, the ratio of maximum acceleration response amplitude to
1
the amplitude of a harmonic base excitation is ≈ 2ξ where ξ is the
damping of the system. For a harmonic ground motion that is tuned with
the building frequency and also with the cabinet frequency, the total
theoretical amplification possible for ICRS can be shown to be (2ξ)− 3.
For a damping ratio of 5% in the building and 2% in the cabinet, a 2%
ICRS would correspond to an amplification factor of 6250. A small
amplitude ground acceleration of 0.1 g would lead to an in-cabinet
spectral amplitude of 625 g. This shows that it is theoretically possible
Fig. 16. Comparison of fixed-base cabinet ICRS with ISRS for case 8 (H–H–H). to have excessively large values of ICRS if a linear elastic analysis is
considered. For a ground motion with relatively small PGA of 0.1 g, such
compared in Fig. 14. As seen in this figure, the high–frequency building large spectral accelerations are unrealistic in real-life applications.
resonates with the high frequency ground motion leading to large values Given the argument that arises based on the observations from
of spectral accelerations around 35 Hz in the ISRS. However, the spectral Fig. 16, it is important to explore the role of nonlinearity at cabinet
accelerations at 35 Hz in the corresponding ICRS are much less because mounting arrangement. Next, the ICRS are evaluated for the same sys­
the cabinet is quite flexible with respect to the high frequency floor tems but with different types of nonlinearities at cabinet base. These
motion and acts almost like an isolation system. It is interesting to note ICRS are compared with each other as well as to the corresponding ICRS
that spectral accelerations in ICRS are relatively high around 3 Hz which for fixed base cases. First, the peak cabinet accelerations as well as
is the frequency of the low–frequency cabinets for both low and high maximum total cabinet displacements obtained by analyzing all models
frequency ground motions. While this observation is anticipated for the for each case are compared in Tables 4 and 5.
low frequency ground motion, the corresponding high spectral accel­ Nonlinear Base Case-1 (L–L–L): For this system, the building and
erations at 3 Hz for high frequency ground motion occur because the cabinet are in resonance with the ground motion. Hence, there is a high
ground motion does not contain only high frequency content. It also amplitude of spectral acceleration at resonating frequency 3 Hz for the
contains some low amplitude low frequency input. These low amplitude fixed-base case. On the other hand, the amplitude of peak spectral

8
S. Singh and A. Gupta Nuclear Engineering and Design 374 (2021) 111046

Fig. 17. ICRS for nonlinear base case 1 (L–L–L).

Fig. 18. ICRS for 5 mm gap cabinet- nonlinear base case 1 (L–L–L). Fig. 19. ICRS for nonlinear base case 2 (H–L–L).

acceleration for 1 mm gap model is even higher than the fixed-base case gap. Second, an overall amplification can be seen in the peak spectral
because the maximum displacement of the floor is more than 1 mm gap. accelerations compared to GMRS. This observation is attributed to the
The maximum floor displacement is 6.14 mm. When the cabinet base, in maximum floor displacement being slightly greater than the gap.
this case, hits the anchor bolt, an impulse is generated due to the high Nonlinear Base Case − 2 (H–L–L): In this case, the ICRS are evaluated
velocity of the cabinet which acts as an additional load. Repeated im­ for same systems as in Figs. 17 and 18 but for a high frequency motion.
pulses thus resonate with natural frequency of cabinet and further adds The new ICRS curves are compared in Fig. 19. As discussed earlier, the
to the already large response from linear analysis at the 3 Hz value. In high frequency ground motion also contains some low frequency con­
contrast, the peak spectral acceleration is much less for the case of 5 mm tent which resonates with the low–frequency building as well as low­
gap as compared to the fixed-base case. Essentially, a 5 mm gap filters –frequency cabinet. Consequently, an amplification can be observed in
out most of the displacement of the cabinet. Even though the maximum the ICRS around 3 Hz for the case of fixed-base analysis. The response is
floor displacement (6.14 mm) is slightly larger than 5 mm, the floor comparatively less for the elastic-perfectly plastic case due to energy
motion for almost all the duration of this motion remains below 5 mm.
Therefore, no amplification is observed. Furthermore, this large floor
displacement occurs only for a very short duration during the entire
floor motion and the response due to the corresponding impulse is
almost nonexistent. In addition to the gap boundary condition, Fig. 17
also gives ICRS for the case of elastic-perfectly plastic boundary condi­
tion (friction). As seen in the figure, the hysteresis loop dissipates some
energy and therefore the peak of ICRS is less than that of fixed base case.
The relatively low values of spectral accelerations in the ICRS for 5 mm
gap case make it difficult to observe the true variation in Fig. 17.
Therefore, Fig. 18 gives the ICRS for only the 5 mm gap case. This figure
can be used to make two observations. First, the peak spectral acceler­
ation occurs at 1.7 Hz instead of 3 Hz. The frequency shifts because of
reduction in stiffness of the cabinet, i.e., essentially zero stiffness when
relative displacement of floor and cabinet is between zero and 5 mm
Fig. 20. ICRS for 5 mm gap cabinet-nonlinear base case 2 (H–L–L).

9
S. Singh and A. Gupta Nuclear Engineering and Design 374 (2021) 111046

generated for 5 mm gap case has smaller amplitude at low frequency

followed by a smooth curve.
This is a very peculiar observation and in order to understand it
better, a fast fourier transform (FFT) is evaluated and compared for
acceleration time histories of various cases. Fig. 22(a) – (c) give the FFT
plots for the case of fixed-base system, 1 mm gap case, and 5 mm gap
case, respectively. FFT for cabinet acceleration time history in the fixed-
base case shows that the primary frequencies contained in this motion
occur between 1 and 5 Hz with maximum input at 3 Hz. On the contrary,
the FFT of 1 mm gap shows that there are many frequencies from 3 to
100 Hz contained in the in-cabinet acceleration time history which
would resonate with different oscillator frequencies while generating
the response spectrum curve. This is because of the repeated impulses
generated during this motion as the floor displacement is more than the
gap. This impulse causes the cabinet to vibrate in its natural frequency
Fig. 21. ICRS for nonlinear base case 3 (L–L–H). under free vibration (35 Hz) which then keeps repeating because of
repeated impulses. FFT plot for 5 mm gap cabinet shows a very periodic
dissipation by the hysteresis. ICRS generated by analyzing 1 mm gap pattern of frequency content of acceleration time history. Again, as
model also has a lower amplitude of peak spectral acceleration than the explained for 1 mm gap cabinet, the impulse is generated when cabinet
fixed-base cabinet but higher than elastic-perfectly plastic case. The base hits the bolt head and the cabinet then oscillates in its natural
response for 5 mm gap case is relatively very less. For 1 mm gap case, the frequency until the next impulse. The maximum contribution to the
peak spectral acceleration is still quite high and around 10 g. It is so response is from 3 Hz frequency (as expected, since the cabinet is rigid as
because the maximum floor displacement is significantly large and equal
to 11.25 mm due to flexible nature of building and a much higher PGA of
the high frequency ground motion. On the other hand, the response of 5
mm gap models is relatively less because the gap filters out quite a bit of
the floor displacement. For completeness, Fig. 20 shows the ICRS for
only the 5 mm gap case.
Nonlinear Base Case-3 (L–L–H): Fig. 21 shows the ICRS of high­
–frequency cabinet mounted on low–frequency building subjected to
low–frequency ground motion. In this case, since the cabinet is rigid
with respect to the building, the fixed-base model has same ICRS as the
ISRS. For elastic-perfectly plastic model, the ICRS is same as that of the
fixed base case but continues to have slightly higher amplitude before
merging at peak cabinet acceleration which is same for both models.
ICRS of models with gaps, on the other hand, show a very different
trend. The trend is similar to linear elastic model until about 5 Hz fre­
quency and after that for 1 mm gap cabinet, relatively higher amplitude
of spectral acceleration can be observed. On the other hand, ICRS Fig. 23. ICRS for nonlinear base case 4 (H–L–H).

Fig. 22. Fast Fourier transform plot for case 3 (L–L–H).

10
S. Singh and A. Gupta Nuclear Engineering and Design 374 (2021) 111046

acceleration for ICRS of 1 mm gap cabinet occurs at around 2.10 Hz

instead of 3 Hz (which is the frequency content ISRS) and that of 5 mm
gap cabinet occurs at even lower frequency of around 1.7 Hz as observed
earlier in Fig. 18 as well. The elastic-perfectly plastic model initially
behaves as a linear-elastic model until seismic forces are more than
friction forces which leads to hysteretic behavior. Thus, the trend of
ICRS is similar but the peak spectral acceleration is slightly smaller than
the fixed-base model.
Nonlinear Base Case-6 (H–H–L): A high-frequency ground motion is
applied at the base of the system discussed above and Fig. 25 shows the
ICRS for this case. The building’s natural frequency, when subjected to
high-frequency ground motion, resonates with ground motion’s fre­
quency at 35 Hz and thus peak spectral acceleration occurs at 35 Hz. But
the cabinet is flexible with the natural frequency of 3 Hz. Thus, the
response of the fixed-base cabinet is less as compared to ISRS with peak
cabinet acceleration of only 0.4 g. Similar to all other cases, elastic-
Fig. 24. ICRS for nonlinear base case 5 (L–H–L).
perfectly plastic model has the same trend as that of the fixed-base
model with somewhat smaller amplitudes of peak spectral accelera­
compared to the building, the the ICRS should retrace floor response
tion and peak cabinet acceleration. Moreover, both the gap models are
spectra in fixed-base analysis and thus ICRS has main frequency content
filtering out the displacement and high-frequency of ground motion.
of 3 Hz) and then impulse reults in oscillations at high frequencies. From
Thus, ICRS of both gap models have very less peak spectral accelerations
FFT plot, we can observe that there is a smooth curve-like envelop on
and even lesser peak cabinet acceleration at around 0.23 g for 5 mm gap
higher frequencies, thus ICRS generated by using 5 mm gap cabinet is a
cabinet and 0.29 g for 1 mm gap cabinet.
smooth curve.
Nonlinear Base Case-7 (L–H–H): Fig. 26 shows ICRS of a system with
Nonlinear Base Case-4 (H–L–H): Examining ICRS shown in Fig. 23 for
high–frequency cabinet mounted on high–frequency building subjected
a high–frequency cabinet mounted on low–frequency building subjected
to low-frequency ground motion. Since ISRS is same as GMRS, the fre­
to high-frequency ground motion, it can be observed that fixed-base and
quency content of total floor acceleration remains 3 Hz, and hence the
elastic-perfectly plastic cases have similar nature of response as that of
cabinet is too rigid for this case. Thus, ICRS for fixed-base as well as
previous case but with smaller amplitude. The 1 mm gap model has
similar nature of ICRS (since maximum floor displacement in this case is
11.25 mm) with slightly lesser amplitude than that of previous case. On
the contrary, 5 mm gap filters out most of the displacement and hence
high-frequency ground motion. But the nature of ICRS seems to be
similar to the previous case but with smaller amplitudes.
Nonlinear Base Case-5 (L–H–L): Fig. 24 shows ICRS of a low­
–frequency cabinet mounted on a high–frequency building subjected to
low-frequency ground motion. In this case, the building is too rigid to be
affected by low-frequency ground motion and hence, the relative
displacement is negligible thereby resulting in an ISRS that is same as
GMRS. Due to this, the main frequency of total floor acceleration is 3 Hz
(input ground motion) which resonates with cabinet frequency resulting
in the high amplitude of peak spectral acceleration of ICRS. Since the
maximum floor displacement (6.01 mm), in this case, is higher than that
of gaps 1 mm and 5 mm, peak spectral acceleration and peak cabinet
acceleration in ICRS for gap models is also more than that of ISRS.
However, it can be noticed that the frequency of peak spectral Fig. 26. ICRS for nonlinear base case 7 (L–H–H).

Fig. 25. ICRS for nonlinear base case 6 (H–H–L).

11
S. Singh and A. Gupta Nuclear Engineering and Design 374 (2021) 111046

7. Conclusions

In summary, the primary observations and conclusions drawn from

the results are:

• A linear-elastic analysis can result in excessively high (unrealistic)

spectral accelerations in the case of a high–frequency building and
high–frequency cabinet subjected to high frequency ground motion.
Currently, the plants are faced with precisely this situation as every
building and every cabinet has a high frequency mode that are
perfectly or nearly tuned with each other and with the frequency of
input motion. However, as illustrated in this study, it is a theoretical
condition and even a very small degree of nonlinearity reduces such
high unrealistic accelerations to more realistic values. If the results of
Fig. 27. ICRS for nonlinear base case 8 (H–H–H). a linear-elastic analysis are used in seismic qualification of relays or
other electrical instruments, then a large number of the instruments
elastic-perfectly plastic models are similar to ISRS with slightly higher are likely to fail the qualification and the cost of operating and
amplitudes. On the other hand, the maximum displacement of the floor upgrading the plant will be unrealistic too.
(6.01 mm) is more than that of the gap and hence ICRS of gap models • The peak spectral acceleration and peak cabinet acceleration in ICRS
have higher amplitudes. The nature of ICRS for gap cabinets is same as for nonlinear cases of gap models depend on the maximum floor
that of the ICRS shown in Fig. 21. displacement. If the maximum floor displacement is less than the
Nonlinear Base Case-8 (H–H–H): Finally, the high-frequency ground gap, the ICRS is observed to be significantly less, i.e., they filter out
motion is applied at the base of the system discussed above, and Fig. 27 the motion as hypothesized at the beginning of this study.
shows ICRS. The natural frequency of fixed-base model resonates with • If floor displacement is greater than the gap, then ICRS has a rela­
the frequency content of input floor acceleration and hence, the peak tively larger amplitude in the high-frequency range in the case of
spectral acceleration of ICRS of fixed-base cabinet is very high as high–frequency cabinets and higher amplitudes in the low-frequency
compared to all other models and ISRS. It should be noted here that only range for low–frequency cabinets. However, the amplitude values
because the system is perfectly in resonance with the ground motion, the are of similar order as observed in existing designs. In such cases,
peak spectral acceleration is very high, otherwise it is not practical for there exist possibility of repeated impulses which can result in high
this high amplitude to occur. This is precisely the problem being faced frequency oscillations. The relays and other instruments must be
by the plants in Central and Eastern US currently. Theoretically, a linear- qualified for such motions which are a reality in actual plants due to
elastic analysis results in acceleration levels that are unrealistic and not the existence of such nonlinearities in real-life.
observed in typical earthquakes even when the earthquakes have had • ICRS calculated by analyzing cabinet with elastic-perfectly plastic
high frequency content. This case is the primary motivation behind this model follows a trend similar to that of the fixed-base cabinet but
study. with somewhat smaller amplitudes mainly due to energy dissipation
Fig. 28 shows the ICRS of all the other cabinet models as their true in the hysteretic loop. This observation also illustrates that cabinet
variation is difficult to observe in Fig. 27. The nature of ICRS in elastic- model–I considered in this study is very different from cabinet
perfectly plastic cabinet is similar to that of fixed-base cabinet but with model–II and even though model–II is also nonlinear, it is not closer
significantly smaller peak spectral acceleration as well as peak cabinet to reality.
acceleration. On the other hand, the maximum floor displacement
(11.18 mm) is more than the gap. Therefore, the ICRS for both gap
models is observed to be similar to that in previous cases. The amplitude Declaration of Competing Interest
is smaller for 5 mm gap cabinet as compared to the previous case.
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence

Fig. 28. ICRS for nonlinear base case 8 (H–H–H; without fixed-base cabinet).

12
S. Singh and A. Gupta Nuclear Engineering and Design 374 (2021) 111046

the work reported in this paper. IEEE, 2013. IEEE Standard for Seismic Qualification of Equipment for Nuclear Power
Generating Stations. IEEE Std 344TM-2013.
Katona, T., Kennerknecht, H., Henkel, F.O., 1995. Earthquake design of switchgear
Acknowledgement cabinets of the VVER-440/213 at Paks. In: Transactions of the Thirteenth
International Conference on Structural Mechanics in Reactor Technology, 1995; III,
This research was supported by Center for Nuclear Energy Facilities Porto Alegre, Brazil, August 13–18.
Konstantinidis, D., Nikfar, F., 2015. Seismic response of sliding equipment and contents
and Structures at North Carolina State University. Resources for the in base-isolated buildings subjected to broadband ground motions. J. Int. Assoc.
Center come from the dues paid by member organizations and from the Earthquake Eng. 44 (6), 865–887. https://doi.org/10.1002/eqe.2490.
Civil, Construction, and Environmental Engineering Department and Lee, B.J., Abou-Jaoude, C., 1992. Effect of base uplift on dynamic response of electrical
and mechanical equipment. Seismic Eng. PVP 237-2, 145–150 (ASME).
College of Engineering in the University. Lee, B.J., Berak, E.G., Passalugo, P.N., 1990. Effect of base shim plates on seismic
qualification of electrical control panels. In: Proceedings of the 1990 Pressure Vessels
References and Piping Conference, PVP, vol. 197, ASME, pp. 301–306.
Llambias, J.M., Sevant, C.J., Shepherd, D.J., 1989. Non-linear response of electrical
cubicles for fragility estimation. Transactions of 10th International Conference on
Cho, S.G., Kim, D., Chaudhary, S., 2011. A simplified model for nonlinear seismic
Structural Mechanics in Reactor Technology, Anaheim, USA, August 22–27.
response analysis of equipment cabinets in nuclear power plants. Nuclear Eng. Des.
Makris, N., Zhang, J., 1999. Rocking Response and Overturning of Anchored Equipment
241(8), 2750–2757. https://doi.org/10.1016.
under Seismic Excitations. PEER 1999/06.
Djordjevic, W., 1992. Amplified response spectra for devices in electrical cabinets. Proc.
McKenna, F., Fenves, G., 2001. The OpenSEES Command Language Manual: version 1.2.
4th Symp. on Current Issues Related to Nuclear Power Plant Structures, Equipment
Pacific Earthquake Engineering Center, University of California, Berkeley.
and Piping, Orlando, Florida, USA.
PEER, 2015. NGA-East: Median Ground-Motion Models for the Central and Eastern North
Djordjevic, W., O’Sullivan, J.J., 1990. Guidelines for development of incabinet amplified
America Region. PEER Report No. 2015/04, April 2015.
response spectra for electrical bench boards and panels. 1990. In: Rep. No., EPRI-NP-
Rustogi, S.K., Gupta, A., 1998. Incabinet response spectra. Tech. Rep. No. C-NPP-SEP 21/
5223, Stevenson and Associates, Inc, Woburn, Mass.
98, 1998, Center for Nuclear Power Plant Structures, Equipment and Piping, N.C.
EPRI, 2007. Program on technology innovation: The effects of high-frequency ground
State Univ., Raleigh, N.C.
motion on structures, components, and equipment in nuclear power plants. Palo
Rustogi, S.K., Gupta, A., 2004. Modeling the dynamic behavior of electrical cabinets and
Alto, CA: 2007.1015108.
control panels: Experimental and analytical results. J. Struct. Eng 130 (3), 511–519.
EPRI, 2013. Seismic Evaluation Guidance: Screening, Prioritization and Implementation
SSHAC, 1997. Recommendations for Probabilistic Seismic Hazard Analysis: Guidance on
Details (SPID) for the Resolution of Fukushima Near-Term Task Force
Uncertainty and Use of Experts. NUREG/CR-6372, UCRL-ID- 122160, Vol. 1.
Recommendation 2.1: Seismic. Palo Alto, CA: 2013.1025287.
Sankaranarayanan, R., 2007. Seismic Response of Acceleration-Sensitive Nonstructural
EPRI, 2014. High frequency program: High frequency testing summary. Palo Alto, CA,
Components Mounted on Moment-Resisting Frame Structures. Ph.D. Dissertation.
2014.3002002997.
University of Maryland, College Park.
EPRI, 2015. High frequency program: Application guidance for functional confirmation
Stafford, J. R., 1975. Finite element predictions of the dynamic response of power plant
and fragility evaluation. Palo Alto, CA, 2015.3002004396.
control cabinets. In: Proc. 2nd ASCE Specialty Conf. on Structural Design of Nuclear
Gupta, A., Cho, S-.G., Hong, K-.J., Han, M., 2019. Current State of in-cabinet response
Plant Facilities, New York.
spectra for seismic qualification of equipment in nuclear power plants. Nuclear Eng.
USNRC, 2014. Regulatory Guide 1.60: Design Response Spectra for Seismic Design of
Des. 343, 269–275.
Nuclear Power Plant. Rev. 2, July 2014.
Gupta, A., Rustogi, S.K., Gupta, A.K., 1999. Ritz vector approach for evaluating incabinet
USNRC, Feb 2014. Recommendation for enhancing reactor safety in the 21st century.
response spectra. Nuclear Eng. Des. 190 (3), 255–272. https://doi.org/10.1016/
USNRC, Tech. Rep. 46.
S0029-5493(99)00076-X.
Vlaski, V., 2013. Reduction of External Hazard (Fast Impact) Induced Vibrations. F/AB-
Gupta, A., Yang, J., 2002. Modified Ritz vector approach for dynamic properties of
58. 1st Conference on Technical Innovation in Nuclear Civil Engineering TINCE
electrical cabinets and control panels. Nuclear Eng. Des. 217(1), 49–62. https://doi.
2013, Paris, October 28–31.
org/10.1016/S0029-5493(02)00133-4.
Yang, J., Rustogi, S.K., Gupta, A., 2002. Rocking stiffness of mounting arrangements in
Herve, G., 2014. Improvement of the evaluation of high frequency content in the
electrical cabinets and control panels. Nuclear Eng. Des. 219 (2), 127–141. https://
calculation of impact floor response spectra. 2nd Conference on Technical
doi.org/10.1016/S0029-5493(02)00279-0.
Innovation in Nuclear Civil Engineering TINCE 2014, Paris.
IAEA, 2012. Safety Assessment of NPP Structures against Human Induced External
Events. IAEA SAFETY REPORT SERIES, DD1086 Draft, Rev.: R1- 1, 6 November.

13