Professional Documents
Culture Documents
<<
m^m UCRL-51928
October 8, 1975
LAWRENCE
LIVERMORE
LABORATORY
iliVV&sityofGaitotria/lJvemKtB
u r,---.;
; E LMTISUNUMITED,
)I3TSIBUTI0N OF T
NOTICE
"This report svas prepared as an account of work
sponsored by the United Stales Government. Neither
the United States nor the United Sores Energy
Research & Development Administration, nor any
of their employees, nor any of their contractors,-
subcontractors, or their employees, makes an;
warranty, express or Implied, or assumes any 'egai
liability or responsibiliiy for the accuracy,
completeness or usefulness of any information,
apparatus, product or process disclosed, or
represents that its use would not infringe
privately-owned rights."
* NTIS
"Pages Selling Price
1-50 $4.00
51-150 $5.45
151-325 $7.60
326-500 $10.60
501-1000 $13.60
Distribution Category
UC-80
m
LAWRENCE UVERMORE LABORATORY
University otCaSfomia Uvermofe.Calitcrria 94550
UCRL-51928
Abstract 1
Summary and Recommendations 1
Critical Structures and Equipment 4
Manufacturing Building 7
Receiving and Drum Storage . 8
Unloading Station and Scrap Recovery System 8
Bulk Storage and Mixed-Oxide Storage 8
Blending Column and Pressing Column 10
Transfer System 10
Ventilation System 11
Diesel Generators 11
Instrumentation 11
Methods of Structural Analysis 12
Manufacturing Building • 12
Equipment . . . . . 14
Definition of Terms 15
Available Methods 22
Comparison of Methods , • . . 28
Conclusions 34
Generation of Floor Spectra 36
Floor-Spectra Calculations 37
Ground Motion Used 37
Building Models . , . . . 37
Time-History Method 41
Approximate Methods 42
Comparison of Methods 52
Floor Spectra 52
Intangibles 53
Conclusions . , 57
Acknowledgments 57
References . . . . . . 58
Appendix. Description of the Kapur and Biggs Approximate Methods ... 60
-iii-
EVALUATION OF METHODS FOR SEISMIC ANALYSIS
OF MIXED-OXIDE FUEL FABRICATION PLANTS
Abstract
-2-
Recommended methods of analysis for manufacturing building
and critical equipment.
-3-
mass of equipment
m mass of supporting structure
natural frequency of equipment
f frequency of dominant support motion
-5-
PuO» bulk storage
Pressing column
Sintering furnaces
Gloved barrier
Blending column
-8-
Shatterproof
96 i n . window
42 i n .
oo oo
-l/2-in.-diam
leveling bolts
- 2 X 2 X 3 / 1 6 in.
structural tubing
(welded construction)
5/8-in.-diam
Front view
anchor bolts
(2 each leg)
-9-
BLENDING COLUMN AKD PRESSING Reinforced concrete -
COLUMN
are attached to the main building The pneumatic lines are asstuned to
-10-
powder oxides from the unloading sist of approximately ltf-in.-diameter,
station to the bulk storage bins and relatively lightweight (16-gauge)
also to transfer the blended powder pipe. The ducting will be anchored
from the blending column to the to the main building structure. The
mixed-oxide storage bins. Storage is box HEPA filters are lightweight and
assumed to be on an upper level to are approximately a 1-ft cube. The
take advantage of gravity feed for final HEPA filter bank consists of a
blending and pressing operations. filter frame that houses the box HEPA
filters.
VENTILATION SYSTEM Figure 6 shows a HEPA filter frame
designed in accordance with ORNL-
The ventilation system will employ 9
NSIC-65 guidelines. The frame is of
high-efficiency particulate air
welded stainless steel construction.
(HEPA) filters to minimize the release x
with dimensions of 1- 10- * 20-ft
of airborne oxides from the building.
and a total weight of 2000 lb.
The system will always direct the air
flow towards zones of greater radio
active contamination potential. DIESEL GENERATORS
-11-
Filter bank enclosure
Typical HEPA
filter, 2 4 X 2 4
X 1 2 i n . , 50 l b /
filter
/ - F r a m e , 1 ft d e e p ,
v
10 ft h i g h ,
20 f t l o n g . A l l
stainless stee!
welded construc
tion . Connected
to f l o o r , w a l l s ,
and roof of
enclosure.
-12-
Table 3. Kecommended models and used. The effects of site character
method of analysis.
istics (such as nard, intermediate,
and soft) on the response were
Foundation Method of
location Model analysis considered. Regulatory Guide 1.60
was used as the basis for ground-
Near or at Lumped mass or Response
3
motion input. compared the
grade finite element spectra
calculational results and the required
Deeply Finite element Response
embedded spectra 3
manpower and computer effort to form
the basis for our recommendations.
If time-history response quantities
are required, time-history analysis Also included in Ref. 4 were
by either modal or direct integra
discussions of the available methods
tion must be performed. If non
linear effects are important, a of analysis for both site response
time-history analysis by direct
integration must be performed. and structural response.
-13-
Table 4. Comparison of fundamental frequencies.
3
.'lanufacturing 10.4 8.6 2.9
4
Process 9.4 6.3 1.8
-15-
CH
Damping Factors
-M
The response of a Structure,
system, or component usually has
energy dissipation associated with it.
These energy losses may be caused by
internal energy dissipation within
the material or support or joint
slippage, and they are normally
accounted for by means of a damping
value.
Design damping values typically
range from 0.5 to 10% Cf critical
W^fc^S^ damping. Damping can ftave a signifi
cant effect on the dynamic response
Motion of the structure. Table 5 shows
-16-
In this study velocity-dependent single-degree-of-freedoni systems
damping was assumed, a standard (cantilever pendulums) with varying
modeling procedure in structural fundamental frequencies are fixed to
analysis. a moveable base. As the length of
Ground-Response Spectrum
(Free-Field)
-17
(b)
0.4
0.2
-18-
Figure 10 allows the fcree-field- resulting in a floor spectrum with
ground-nwtion spectrum currently used larger response values than ground-
to define seismic input for nuclear motion-spectrum values for a given
power reactors. This smooth spec frequency. Several methods available
trum is based on a statistical treat to generate floor spectra are
ment of approximately 47 earthquake discussed in the section Generation
records and is plotted on tripartite of Floor Spectra.
log paper for the convenience of
Rigid and Flexible System
having maximum acceleration, velocity,
and displacement values as a function In general, rigid systems are
1.0 g. Once the intensity of ground amplification over tne 1-g ground
-19-
Fig. 10. Design response spectra, scaled to 1-g horizontal ground
acceleration (from Regulatory Guide 1.60). (a) Horizontal,
(b) Vertical. Example: for 5% damping and £ = 0.25 cps,
D - 73.8 in., v = 115.93 in./s, a - 0,47 g.
-20-
multiplier of the acceleration of form of a response spectrum. Mode
gravity. The loads are distributed shapes and frequencies of the mathe
consistent with the mass or weight matical model must be calculated.
distribution of the system. Because of the way the modal
Generally, the coefficient is quantities are combined to get the
based on the seismicity and soil total response, this method yields
characteristics at the site as well only an approximation. The response
as the dynamic response character is calculated (such as for displace
istics, type, and function of the ment or loads) in each mode, and all
structural system under analysis. the calculated responses are
Another common way to determine the •superposed by the SRSS procedure.
coefficient is to use the response This procedure is used because the
spectrum defined for the structural maximum value in each mode may not
system's base motion and an assumed occur at the same time. The procedure
damping value. Often the peak of the is modified when closely spaced modes
acceleration spectrum is chosen to occur. Enough modes must be included
develop an equivalent static load to in the analysi J capture the
apply at the center of gravity of the complete structural response.
system. This, in effect, assumes The time-history method determines
tha:: the structure behaves as a the structural response to an
single-degree-of-freedom system in arbitrary force-time history or
resonance with the support or base acceleration-time history. The
motion. No natural-frequency cal method solves the equations of motion
culation is performed on the developed for the system at each
structural system. t .me step using one of two approaches:
mode superposition or direct
Dynamic Analysis integration.
In dynamic analysis the dynamic Using mode superposition Che
properties of the system (frequencies eigenvalue problem associated with
and mode shapes) are used to develop the model is solved to determine the
the seismic loads for design. One of mode shapes and frequencies of
two methods is used: the response- vibration. The response of each mode
spectrum method or the time-history is then determined and all modes are
method. combined to get the total response.
The response-spectrum method This approach is limited to the
requires the base motion to be in the linear response of structures.
-U-
The direct integration approach of the equipment will dictate the
is used to obtain a solution by step- suitable method of analysis.
by-step integration of the equations Figure 11 illustrates the
of motion. Solution of the eigenvalue standard approach used to develop
problem is net required, and non seismic loading to equipment located
linear effects may be included if at different levels in the building.
required. The input seismic loading to the
Either approach gives an "exact" building is usually expressed in the
time-history of response. A direct form of an accelerogram considered
integration analysis is equivalent to representative of the free-field
an analysis by the mode-superposition ground motion at the site (A_). The
approach in which all mode shapes ^nd free-field accelerogram is usually
frequencies are included and the (1) based on a single recorded earth
same time step, At, is used. Both quake accelerogram considered
approaches involve larger amounts of representative of the site or (2)
computer time than the response- developed from a free-field response
spectrum method to solve for the spectrum (S„) based on a statistical
dynamic response of a structure treatment of many past earthquake
because small time steps are required records.
to achieve meaningful results. When an acceleration time-
For more detailed discussion of history is required, a synthesized
the static and dynamic methods, accelerogram (A_) is derived from the
particularly the advantages and dis free-field response spectrum. An
advantages of each, see Ref. 4. example of this approach is Regulatory
Guide 1.60, which defines the seismic
Available Methods input for nuclear power reactors.
We have categorized the different The free-field ground motion (S
seismic analysis methods for or A_) is used as the exciting motion
equipment according to the dynamic to a lumped-mass or finite-element
response characteristics of both the model of the building. This model
building (or support structure) and usually includes soil springs, which
the equipment. The response permit soil-structure interaction
characteristics of the building will effects to be considered.
influence the choice of seismic
*
loading suitable for equipment anal Approximate approaches are discussed
in the section GenerctfyLon of Floor
ysis. The response characteristics Spectra.
-22-
The dynamic response of the model equipment. We then determine the
is calculated using a computer pro stiffness of the equipment. Finally,
gram. Response quantities include after these two stiffnesses are
member forces, displacements, known, the precision in the seismic
velocities, and accelerations through input load indicates the appropriate
out the building. Figure 11 analysis method.
illustrates calculated acceleration The two classes of stiffness, in
time-histories at different floor either buildings or equipment, are
levels (A-j). Depending on the rigid and flexible. Rigid systems
location of the equipment, the are those that exhibit no response
accelerograms can be used directly as amplification when subjected to
input to an equipment dynamic analysis. input motion. Flexible systems
An alternative technique is to develop exhibit response amplification. For
acceleration-response spectra example, for the design of power
corresponding to these floor accelero plants Regulatory Guide 1.60 suggests
grams, designated as S_, in Fig. 11. 33 Hz as the frequency cutoff for
These floor spectra are then used as rigid structures. Those structures
input to the dynamic analysis of with fundamental frequencies greater
equipment. than 33 Hz need not be designed for
An equivalent-static analysis of amplified acceleration levels.
equipment uses a single value from Flexible structures include those with
either the acceleration time-history frequencies less than 33 Hz. For our
include, for example, maximum ground between rigid and flexible buildings
ci -i *.J /* m a x
.flax, and equipment.
or floor acceleration (A G or Ap )
As shown in Fig. 12, methods 1
and ground- or floor-spectra values
through 4 pertain to the rigid-
that correspond to the frequency of building situation and i-^thods 3
^e ^B
the building or equipment (S„ , S_ ,
through 11 the flexible-building
S p ) . Appropriate damping values
situation. Because no dynamic re
must be used.
sponse characteristics or calculations
Figure 12 identifies 11 methods
are required for rigid buildings,
for evaluating the seismic integrity
methods 1 through 4 can determine
of equipment. To find the appropriate
seismic equipment leading directly
method we first determine the stiff
from the free-field ground motion.
ness of the building that houses the
-23-
cpeak
»—A," 1
A
A" s k
p|^r
F
t
s s A s
A * . -* A™ S
F[ A ~ F ° s e k
A/y^Vw^f
s r n
_L_ S°" P ' gs ^F
Static Dynamic
analysis analysis S t a t i c analysis . [ , Dynamic analysis_
of of o f equipment
Required of equipment
equipment
analysis
analysis must be p e r f o r m e d .
Fig. 12. Decision tree showing a n a l y s i s method to evaluate seismic I n t e g r i t y of equipment. Double under
scores indicate recommended methods. Definition of symbols: A„, A_ = acceleration time-history for
a BX a
ground, floor; A™ , Ap maximum a c c e l e r a t i o n for ground, floor; Sg, S„ = a c c e l e r a t i o n response
spectrum for f r e e - f i e l d ground, f l o o r ; S^ , S*J * value from a c c e l e r a t i o n response spectrum for
ground, floor; S , S , S„ = a c c e l e r a t i o n response spectrum value corresponding to frequency of
p
floor response spectrum (S„) or floor made regarding the selection process
value selected (S_ ) is less con* important) will fall into the
servative and consistent with the rigid class. As stated
calculated fundamental equipment earlier, the 33-hz value is
frequency. consistent with Ke&ula.cory
Method 10 (flexible building, Guide 1.60 for power plant
flexible equipment) also uses either buildings.
-2]
When enoosing between methods building, the methods most
2 and 3, methods 5 and t>, and frequently applied will be
methods o and 9 for most methods 1, 4, 7, or 10. The
buildings and equipment combi usual situation is to calcu
nations, uie latter method in late the response of many
each pair will be the most pieces of equipment located
practical because of the in one building. Method 10
undesirable conservatism is probably the best compro
associated with using the mise since floor spectra only
peak-response-spectrum value. need to be computed once.
Tne additional effort required 5* If differential support
to determine the fundamental motion need be considered for
frequency will be justified. the equipment (e.g., a pipe
When choosing between methods that extends over several
2, 3, and 4; methods 5» 6, floor levels), then a dynamic
and 7; and methods 9 and 10 analysis is required. The
because of tne single-degree- single analysis using a
of-freedom requirement, it coupled building-equipment
will generally be more model may be performed, or
practical to use methods 4, two separate analyses using
7, and 10. Most of the equip appropriate support motion
ment and buildings are may be conducted.
sufficiently complex that
they cannot be modeled as a Comparison of Methods
sinfle-degree-of-freedom
The dynamic response characteris
system. Further, once the
tics of the manufacturing building we
response characteristics
consider representative of future
nc'ossary to determine whether
fuel fabrication buildings is dis
or not the system responds as
cussed in detail in the section
a single-degree-of-freedom
Generation 03' Floor* Spectra. The
system are calculated, very
basis for the building design is
little additional effort is
discussed in the section Critical
required for the dynamic
Structures ayid Equipment.
response solution.
The fundamental frequencies of
Given only a single piece of
the building were calculated for two
equipment to evaluate in a
site conditions for both horizontal
-28-
and vertical ground motions. The given are 0.5 and UX free-field-
first site reflects an average site ground-motion spectra. These spectra
shear velocity of 3500 ft/s, and the were developed from the synthesized
second, a softer site, reflects a free-field accelerogram that was
shear velocity of 1500 ft/s. The used roi' the time-history response
calculational model included soil analysis of the building model. The
springs to account for soil-structure kX spectrum is consistent with that
interaction effects. The calculated used for the building response
fundamental frequencies for horizontal analysis. The 0.5% spectra are
motion were 7,7 and 9.7 Hz for the included to permit a comparison of
soft and hard sites, respectively. free-field motion with calculated
The calculated frequency for vertical building response motion for the
motion is 7.7 bz for both site same damping. The free-field ground
charac ter izat ions. motion reflects a 1-g maximum
Using the 33-Hz frequency value ground motion for both horizontal
for the division betveen rigid and and vertical directions. For other
flexible structure classes, the maximum horizontal free-field
manufacturing building must be motion the floor- and ground-
considered a flexible structural response spectra can simply be scaled.
system. The fundamental frequency of These figures clearly show
building horizontal motion is domi building influence on the seismic
nated by the end shear walls; for loads to equipment. The response
vertical motion the floor and roof characteristics of the building
slabs dominate the response character clearly amplify the ground motion.
istics. Even with substantial For example, Fig. 13 (horizontal
variations in wall and slab design motions, soft site) shows roof,
(e.g., decrease or increase of thick second-floor, and ground-floor peak,
ness by as much as 50%) the building spectral values of 114, 80, and 45 g,
would still fall within the flexible respectively., for 0.5% damping. The
rather than rigid class. effect of the building fundamental
Figures 13 and 14 show the frequency can also be observed to
acceleration floor spectra calculated force the largest amplification near
for different levels of the manufac that frequency. The ground-motion
turing building for both horizontal spectrum shows a lesser peak
and vertical motions. The floor amplification of 10 to 12 g at 0.5%
spectra are for 0.5% damping. Also over the 1-g maximum horizontal ground
-29-
Roof ( 0 . 5 % of critical damping)
Second floor
( 0 . 5 % of critical damping)
Ground floor
( 0 . 5 % of critical damping)
0 . 5 % of c r i t i c a l d a m p i n g , -
frt-e-field ground motion
4 % of c r i t i c a l damping,
f r e e - f i e l d ground m o t i o n "
8 10 12
Frequency — Hz
Fig. 13. Free-field ground motion and calculated floor response spectra,
horizontal motions, soft site.
-30-
Roof ( 0 . 5 % of critical damping)
Second
floor
Second floor
( 0 . 5 % of critical damping)
Ground floor
( 0 . 5 % of critical damping)
0 . 5 % of critical damping,
free-field ground motion
4% of critical damping,
free-field ground motion
8 10 12 14 16 18 20
Frequency — Hz
Fig. 14. Free-field ground motion and calculated floor response spectra,
vertical motions, soft site.
-31-
acceleration but over a. wider fre ground motion is based on a 1-g
quency range. Similar amplifications maximum horizontal ground motion.
are shown In Fig. 14.
Soil-structure interaction effects Table 6. Estimated fundamental
frequency of critical
influence the seismic loading to the equipment.
equipment, figure 1J shows a free-
tield-spectva amplification factor of Estimated
irequency
approximately A (45/12) over the Critical equipment (Hz)
ground floor 3t a frequency near the
Unloading station 10-20
building fundamental frequency. This
(gloved box)
amplification is directly attributed Bulk storage 25-50
to the soil springs. Mixed-oxide storage 25-50
Table 6 shows estimated funda Transfer system 5-20
mental frequencies of each of the (piping)
critical pieces of equipment. Varia Scrap recovery system 10-20
(gloved box)
tions in equipment design could
Ventilation system 5-20
substantially modify these frequency (ducting)
estimates. For example, the location Final HEPA frame 5-10
of support points plays a significant
role in the response characteristics
of the equipment. For this reason Table 7. Comparison of seismic
3
-32-
The equipment is assumed to be assumes the equipment frequency
located on the second flcor of the to coincide with the dominant
manufacturing building* The build free-field frequency on the 0.52
ing is assumed to have 4% damping damping curve.
and the equipment 0.5% damping.
Method 3 (rigid building, flexible
As previously indicated, ire
equipment). This method assumes
assume that future manufacturing
that the equipment frequency is
buildings will fall within the
determined and that the result
flexible-building class (methods 5
is a seismic equipment loading
to 11). However* for comparison we
that ranges from 1 to 12 g,
include seismic equipment loads
depending on the actual equipment
that reflect the rigid-building class
frequency.
(methods 1 to 4 ) . Methods 4, 10, and
Method 5 (flexible building, rigid
11 are not included in the comparison
because it would be necessary to equipment). Because the rigid
-33-
Method 7 (flexible building, rigid equipment analysis. It also
equipment). This method assumes requires the determination of the
the building has multiple-degree- fundamental frequency of the
of-freedom characteristics and equipment, resulting in equipment
yields a seismic equipment load loads that could vary from 4 to
of 4 g,. This value is based on 80 g depending on the actual
the maximum second-floor accelera equipment frequency.
tion, which is the same as the
Future fuel fabrication manufac
second-floor spectral acceleration
turing buildings and most critical
corresponding to a frequency of
equipment will fall into the flexible
33 Hz or greater at 0.5% damping.
building, flexible equipment class of
Method 8 (flexible building, methods 8 and 9.
flexible equipment). This method
requires a floor-response spectrum, Conclusions
-34-
the building must be modeled as a equipment will require a multiple-
multiple-degree-of-freedom system. degree-of-freedoin analysis. Further
For flexible equipment we more, the additional analysis effort
recommend method 10 over methods 8 for the frequency determination in
and 9. In most cases method 8 will method 10 is small beyond that already
yield loads that are too conservative. required by method 9.
Method 9 not only requires that the
For heavy equipment, which could
fundamental frequency be calculated,
affect the overall response of the
but also that the equipment responds
building, method 11 is necessary.
as a single-degree-of-freedom system.
With the exception of the simpler These recommendations are
equipment configurations, most summarized in Table 8.
Equipment Recommended
stiffness method Description
Almost all seismic analyses of There are tnree main reasons for
equipment involve tne use of floor- using one of the approximate methods
response spectra. In most instances in. lieu of the time-history approach:
a complete set of response spectra
1. A synthetic accelerogram
are calculated for the building prior
need not be generated
to equipment design. The floor
2. An approximate method is less
spectra will encompass motion in
complex
three orthogonal directions at aJl
3. Less analyst and computer
possible equipment locations and
time is required.
damping.
-36-
structural models of our MOFFP free-field ground motion and the 4%
building wera developed that encompass horizontal spectrum from Regulatory
the range of dynamic characteristics Guide 1.60. Xne figure illustrates
expected in the future. The four Liiat, wnile the agreement is not
consist of vertical and horizontal precise, the spectral content of the
response models founded on soft and synthetic eartnquake is adequate for
nard soil. Floor-response spectra our purposes- The response spectra
were generated witn tne time-history, of the artificial ground motion was
Kapur, and Biggs metnods for all four used as input to the approximate
models and tne results compared. methods.
-37-
10.0 1 I 1 , . Ml, 1 1 1 1 • 1 i | i 1 I 1 1 i 1 i_
-
- r- Spectrum of
" / synthetic
- RG 1.60 / accelerogram.
horizontal, i 4 % damping
4 % damping -v
- ^V
-= 1-0-
: :
^ RG 1.60
. vertical, -
4% damping
1 -
~\l/l
0.10 I . 1 i i i i i I i '1
0.10 1.0 10.0 100.0
Frequency — Hz
30 ft
Concrete beam elements
(mass less)
m = 39
2
4
Sheararea = 9X 10 in . 2
Second t
floor V. - 25.5 X 10° Bending inertia -
11
1.9X 1 0 " i n . "
<*
20 ft
Ground
floor
,1 T A
^
* * A (*>
^^-TN
m,= 40
w
I„*22.2X10°
* &
Soil springs
(kips/in.) ( i n . • kip/rad)
5 11
Soft soil 3.97X10 7.46X 1 0
5
Hard soil 21.6X 10 40.6 X 1 0 1 1
-39-
I , represent the mass and rotational and hard-soil models is the value of
inertia of the building roof, floors* che soil springs. In formulating
and tributary wall areas. The lateral this model it was important to include
stiffening effect of the interior; the vertical bending action of the
columns and partitions is negligible. upper-floor and roof slabs. Our
The beam-element shear area represents model allows for this effect and yet
that of the sidewalls. The bending provides a measure of economy in
inertia is that of the entire per computing and modeling time. The
imeter wall. The soil springs were model shown in Fig. 17 represents a
calculated by the procedures recom JO-ft-wide strip of the MOFFP.
mended by Wnitman and Richart. Figure 18 shows the origin of the
Figure 17 suows the analytical strip model. Because the building
model used for vertical MOFFP is symmetric, it was necessary to
response calculations, here again, model only one-naif the length of the
tne only difference between the soft- strip (105 ft) with the addition of
13// 14 15 16 17 18 19/201
— 30ft_
fyp.
8 9 10 U
U/tfi/f t fj\&
^h 2 ^ Z \ ^W Z "^
-Plane of symmetry
Fig. 17. Analytical model for MOFFP vertical response, elevation view. Soil
springs: soft soil, k - 1.00 * 10* kip/in.; hard soil, k - 5.42
z 2
x 10* kip/in.
-40-
-210 ft
D • D D
Region
modeled
.Typical 30 ft
.strip
210 ft
• • a
Fig. 18. Basis for the vertical MOFFP model, plan view.
symmetric boundary conditions at the overall soil spring for the building.
centerline as shown. The interior They were calculated as follows. The
vertical elements are the 24- x 24-in. vertical soil spring for the entire
columns. The horizontal elements and 210- x 210-ft foundation was computed
exterior vertical elements (at the by the method in Ref. 17. The k
left) ate beams IS in. thick by shown in Fig. 17 represents the
30 ft wide. The base mat has been fractional value of the overall
made artificially stiff to account spring for a 30- x 30-ft section of
for the supporting effect of the the slab (that is, k ,,M9).
' 2 overall '
underlying soil.
Time-History Method
The soil springs, k , represent
the appropriate fraction of the All the time-history analyses
-41-
were performed in the same manner. spectrum of the free-field ground
We u,,ed the 5APIV finite-element motion at this frequency. Tne 5.3-hz
18
code to make the calculations on a peak can be seen in the other floor
CDC-7600 computer. We used the spectra, but does not dominate when
direct-integration scheme with the the modal response is more evident.
free-field-ground-motion accelerogram The uneven character of tne spectrum
described above as input and 4% of of the input motion also accounts for
critical viscous damping. the split resonance peak occurring in
Because the equations of motion the soft-soil horizontal spectra.
are formulated in terms of relative Note in Fig. 19a that there appears to
displacements, the accelerations be two closely spaced resonances. In
output by S A P I V are relative to actuality there is a "valley" in the
free-field. The absolute acceleration spectrum of the input motion near the
time-history was obtained by a post fundamental frequency of the building.
processing operation that added the Figure 20 shows the vertical
ground acceleration to the structure MOFFP floor spectra calculated by
acceleration at each time step. The the time-history method. We
final step was to calculate the chose nodes 2, 8, and 16 of our
pseudo-acceleration response spectrum vertical model for floor-spectra
of the motion at each degree of calculations. They are representative
freedom of interest. This was done of the ground floor, second floor,
for 0.5% of critical (equipment) and roof, respectively. The split
damping. resonance phenomenon observed in the
models. For the most part, the The Kapur and Biggs approximate
horizontal spectra are dominated by methods for generating floor spectra
the fundamental frequency of the are well documented in Refs. 13 and
building. The one exception is the 14. However, for completeness the
ground-floor spectrum of the hard- two methods are outlined in the
soil model. In this case, the.15-g appendix to this report.
peak occurs at 5.3 Hz and results There are two essential differ
from a "spike" in the response ences between the two methods. The
120 1 1 1 I
>> -
•-Roof
100 1
- -
80
i
- ^s— Second floor -
in
60 \
- >— Grour d floor
-
40 \
- -
20
\
-
«
^ 2^ *-~.~—u
90 1 1 i
-(b) -
80
- /-fc> of -
70
- -
60 /
- S e :ond flo<ar "
50
/
-
/ 1 -
rj\V
40
- ound floor "
30
- / !
A, -
V
i
20
- -
10 *Jtm $ A A/
fjW f&Kfi — I—
—^
V-r- 1 r i '
8 10 12 14
Frequency — Hz
f i g . 19. Time-history-derived horizontal floor s p e c t r a , 1/2% damping.
(a) Soft s o i l , (b) Hard s o i l .
-43-
I
-8
4 6 8
Frequency — Hz
-44-
first is that Kapur's method is with the SAPIV code. The analytical
based on a hypothetical upper limit models used were identical to those
to the amplification of the building used in the time-history calculations.
and/or ground motion; Biggs' amplifi Figure 21 shows the results of
cation factors are based on the the modal analysis of the two
envelope of several time-history horizontal models. In both the soft-
analyses of a two-degree-of-freedom and hard-soil cases, the four lowest
system. Kapur used a time-history natural frequencies were input to the
analysis to estimate amplification approximate methods. None of the
factors only at equipment-building modes were closely spaced.
resonance. Table 10 gives the modal-analysis
The second difference is that data for the two vertical response
Kapur'a method involves artificially models. Here, the model was more
broadening the resonance peaks in the complex, and therefore the resulting
floor spectra by ±10%; Biggs' method mode shapes were more complex. Of
does not broaden the peaks. the 15 frequencies obtained for the
To our knowledge Biggs' method is soft- and hard-soil models, 10 were
published for only one set of chosen for input to the approximate
structure-equipment damping values: methods. The frequencies were chosen
4.0 and 0.5%. Kapur's method is for maximum response, based on the
published for a range of structure- participation factor, r, the eigen
equipment damping values that does vector, $>, and preliminary calcula
not include the two that Biggs reports. tions of approximate floor spectra.
We therefore extrapolated Kapur's It is of interest to note that
resonance amplification factor to 4.0 many of the modes of the vertical
and 0.5% structure-equipment damping model are closely spaced under the
to facilitate the comparison. The 19
criteria in Regulatory Guide 1.92.
extrapolated amplification value only This fact affects the approximate
affects Kapur's floor spectra near floor spectra because, as we point
building-equipment resonance. out in the appendix, the criteria
As noted previously, both methods and summing procedure in the Guide
require a modal analysis of the are used in the Biggs and Kapur
building and the response spectra of methods.
the free-field ground motion. The In performing the analysis we
modal analyses required by the discovered that in some instances the
approximate methods were accomplished approximate vertical spectra could
-45-
Soft soil Hard soil
f
4>sn T
sn
ln ln f
(Hz) r
ln s= 1 2 3 (Hz) r
in s= 1 2 3
7.30 9.94 0.128 0.0925 0.0526 9.69 8.71 0.147 0.0834 0.0150
14.8 2.59 0.0364 0.0395 0.0563 22.8 3.43 0.0797 0.122 0.0326
20.0 2.21 0.104 0.0549 0.0878 26.4 0.741 0.0475 0.0495 0.0094
28.9 1.29 0.0418 0.112 0.107 41.5 4.89 0.0055 0.0369 0.154
1? — *ln
2? T
2n
^fef*3n
Fig. 21. MOFFF horizontal response model, normal mode data input to approxi
mate floor spectra methods, lowest four normal modes. T, modal
participation factor; <j), eigenvector. (See Appendix for additional
definitions.)
-46-
Table 10. Normal-mode data for MOFFP v e r t i c a l response model.
Mode
r
No. £
ln i„
"n" S*2 8 16 (H > S=2 8 16
..b
<v
x
7.70 .987 .084 .081 1.16 7.72 .039 .0056 .192 .924
2 a,b 7.74 2.07 .124 .566 .333 8.29 .031 .0035 .556 .507
a , b ,c
3 !
7.98 .088 .102 .471 .744 8.42 .303 .020 .934 1.16
a , b , ,c .328 .072 8.82
4
a . 19 .453 .503 L.35 .0049 .270 .495
a , b , ,c
5
8.44 .484 .059 .803 .532 9.15 ..17 .035 .374 .235
6 « . b , ,c 9.63 .234 .029 .866 .315 9.89 .502 .018 .908 .385
7 a,b 11.43 .362 .080 .873 .577 11.77 .253 .022 .805 .440
a . b , ,c
8
12.67 .669 .062 .471 .349 13.08 .509 .015 .848 .468
9 a
13.23 .256 -0- .583 .418 16.88 .089 .047 .059 .105
b
io '= 16.04 .059 .151 .083 .150 19.67 .229 .057 .291 .517
11= 18.48 .106 .315 .382 .509 21.31 .974 .058 .037 .026
a.b, c
1 2
19.49 .787 .251 .400 .399 22.26 .752 .052 .258 .447
13 25.74 .482 .220 .019 .256 34.79 .047 .156 .018 .011
14= 27.04 .158 .680 .055 .174 35.47 .159 .248 .123 .074
15= 29.07 .051 .552 .009 .006 36.09 .709 .138 .124 .080
S o f t - s o i l modes
D
Hard-soil modes , nodes 8 and 16.
c
H a r d - s o i l modes , node 2.
-47-
I 1 1 i 1
1— 1 -T 1 1
(a)
120
.^Roof
- -
100
- -
• -
60
- /- Second floor -
40
20
" jX / - Ground f l Dor -
-jT 1 • i l l i i i i
90 i I 1 1 ' i i i i —i i i i' 1
"(b)
80
- y-Roof -
70
- / -
60
- -
50
- -
40
- ^^Seconc floor -
30
/ )f y^Groun 1 floor -
20
10
-
\>C -
t^\ _1 U -*"i 1 1 1 ,
10 20 30
Frequency — Hz
Fig. 22. MOFFP horizontal floor spectra by Kapur's method, 1/2% equipment
damping. (a) Soft-soil model, (b) Hard-soil model.
-48-
Frequency — Hz
Fig. 23. MOFFP horizontal floor spectra by Biggs' method, 1/2% equipment
damping, (a) Soft-soil model. (b) Hard-soil model.
-49-
200
20
Frequency — Hz
Fig. 24. MOFFP v e r t i c a l floor spectra by Kapur's method, 1/2X equipment
damping, (a) S o f t - s o i l model, (b) Hard-soil model.
-50-
240 - 1 "1 1 i ' 1 1 1 1 ""J- •! 1 1 1
- (a)
/-Roof
200 —
/
1/ -
160
- -
^ S e c o n d floor
V
120
/
-
80
40
" i\ (Wi J 1
/ - G r o u n d floor
-
.
S^V 1 1 _i i—i _i_.
Frequency — Hz
1
Fig. 25. MOFFP vertical floor spectra by Biggs method, 1/2% equipment
damping, (a) Soft-soil model, (b) Hard-soil model.
-51-
COMPARISON OF METHODS The artificial peak broadening in
Kapur's method can cause a great
Floor Spectra
deal of conservatism in the regions
Normally, the designer will draw on either side of the building
smooth response spectra that envelope natural frequencies. Second, as the
the calculated spectra. These ratio curves indicate, both the
smoothed curves are a much more Kapur and Biggs methods tend to
convenient and reasonable tool for produce very unconservative horizontal
design purposes. Also, the primary and vertical spectra at the ground
resonance peaks of the floor spectra floor when the soil is hard. The
are artificially broadened for design base mat moves, for th^. most part, as
purposes to allow for uncertainty in the soil moves. To model this
building frequencies. We felt these situation correctly, many more of the
measures would blur our comparison, higher building frequencies need to
particularly so since the spectrum be included. We did not include all
of the input motion was not smooth. the building frequencies in the
Therefore, we did not make these approximate-floor-spectra calculations
modifications on our floor spectra. and the resulting spectra at the
Kapur's method already had a ±10%- ground level were too low at some
peak-broadening feature built into frequencies.
it, but rather than tamper with the Excluding the cases above from
method, we left it in. consideration, the ratio plots
All floor spectra were calculated indicate that the Kapur floor spectra
at the same set of frequencies. The range from 0.6 to 4.5 times the time-
set contained 19el points between 0.10 history floor spectra; Biggs' range
and 33 Hz. Numerical comparison of from 0.8 to 3.7. In general, the
the floor spectra was somewhat methods tend to estimate the spectrum
complicated by the jagged nature of at the fundamental building frequency
the input; spectrum. This problem was very closely and to overestimate it
overcome by plotting the ratio of at higher frequencies.
each approximate floor spectrum to As currently formulated, the
the corresponding time-history- approximate-floor-spectra methods
spectrum vs. frequency. of Kapur and Biggs do not provide a
We feel that these ratio plots, reasonably constant degree of con
shown in Figs, 26 through 29, have servatism. Additional study and
two Important limitations. First, modification of these methods could
-52-
Kapur's method Biggs • met lod
Roof Roof
- - _ j •
\\
h
A /
'^K
f 1
A/ 7 __^
y
w
v^ /v
- -
• I
A Cs4/^—-
>TMu r
ir
\TvJ -v^- - \ _ / ' — ^ s : 1 JT
n /
*~T**AU.
/
V
1/ V
V \ /** JV \
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Frequency — Hz Frequency —- Hz
eliminate, for the most part, the method: (1) no synthetic accelerogram
regions where the spectra are uncon- is needed; (2) an approximate method
servative. However, the designer is less complex; and (3) less analyst
would still be faced with a degree of and computer time is required. How
method-introduced uncertainty and at ever, our experience indicates that
the same time a potential design- these reasons may not be \alid.
-53-
Kapur's method Biggs • met nod
.Roof .Roof
- •
\
""**». J
«w'
IA
^/
/^
/jv
•
. Second floor •
• -
- -
/
-
~,"^J
- - -
K / •
•
• -v J1
l l
/ \ /
-
I
-
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Frequency — Hz Frequency — Hz
-54-
Kapur's method Biggs' method
Roof Roof
- - -
- - - i -
- - - j
fvj ^ ^ "
J
1
J^. ^ J ~ -klV • ^ :
H^ ~~~~-
-JP^ -
- -
i i 1 ; i i-
Groi nd flc or Ground floor
- " 1 j
- - -
f\
1
\
-
^ A
J <\ J V,"»_, j ^ ^ ^ ^ ^
-
i 1
0 5 10 15 20 25 30 0 5 10 15 20 25 30
Frequency — Hz Frequency — Hz
-55-
Kapur's method Biggs' method
_ Roof
-
- -
^J"VJ
-
. Se< ond f oor
-
- -
\
:/\f \r\ -
. -
Gro"nrl f
-
- -
^ / \—
- -
0 5 10 15 20 25 30 10 15 20 25 30
Frequency — Hz Frequency — Hz
-56-
CONCLUSIONS 3-to-4 times the time-history
spectra at some frequencies.
Our study indicates that the time- Methods for generating the
history method of generating floor synthetic ground motion
spectra is the most suitable method required by the time-history
for fuel fabrication facilities and method are becoming widely
other installations of similar value available.
and Importance. The basis of this
The time-history method
conclusion can be summarized as
results in a minor increase
follows:
in complexity and a cost
1
. The approximate methods may increase tnat is minute
yield floor spectra that are compared to the project cost.
Acknowledgments
-57-
References
3. F.J. Tokars, R.C. Murray, and H.C. Sorensen, Seismic Response and Failure
Analyses of a Mixed-Oxide Fuel Fabrication Plant. Lawrence Livermore
Laboratory Rapt. UCRL-51755 (1975).
4. F.J. Tokarz, R.C. Murray, D.F. Arthur, W.W. Feng, L.H. Wight, and
M. Zaslawsky, Evaluation of Methods for Seismic Analysis of Nuclear Fuel
Reprocessing Plants, Lawrence Livermore Laboratory Rept. UCRL-51802,
Part 1 (1975).
5. Design Basis Tornado for Nuclear Power Plants. USAEC Regulatory Guide
1.76 (1974).
6. Building Code Requirements for Minimum Design Loads in Buildings and Other
Structures. American National Standards Institute Rept. ANSI A58.1-1972.
b. Standard Format and Content of Safety Analysis Reports for Nuclear Power
Plants (Revision I ) . prepared by the Regulatory Staff, USAEC (1972).
9. C.A. Burchsted and A.B. Fuller, Design, Construction, and Testing High-
Efficiency Air Filtration Systems for Nuclear Applications. ORNL-NSIC-65
(1970).
10. Reactor Site Criteria: Appendix A. Seismic and Geologic Siting Criteria
for Nuclear Power Plants, USAEC Rept. 100FR100 (1973).
11. Design Response Spectra for Nuclear Power Plants, USAEC Regulatory Guide
1.60 (1973).
12. Damping Values for Seismic Design of Nuclear Power Plants. USAEC
Regulatory Guide 1.61 (1973).
-58-
14. J.M. Biggs, "Seismic Response Spectra for Equipment Design in Nuclear
Power Plants," Paper #K4/7 First International Conference on Structural
Mechanics in Reactor Technology. Germany (1971).
15. J.M. Biggs and J.M. Roesset, "Seismic Analysis of Equipment Mounted on a
Massive Structure," Seismic design for Nuclear Power Plants. R.J. Hansen,
Ed. (MIT Press, Cambridge, MA, 1970), pp. 319-343.
17. R. V. Whitman and F.E. Richart, Jr., "Design Procedures for Dynamically
Loaded Foundations," Journal of the Soil Mechanics and Foundation Division
(November 1967).
18. K. Bathe, E.L. Wilson, and F.E. Peterson, SAPIV - A Structural Analysis
Program for Static and Dynamic Response of jLinear Structures. Earthquake
Engineering Research Center, College of Engineering, University of
California, Berkeley, CA, Rept. EERC 73-11 (1973).
20. R.H. Scanlon, "Earthquake Time Histories and Response Spectra," ASCE,
Journal of the Engineering Mechanics Division, 635 (August 1974).
21. C.A. Cornell and E.H. Vanmarcke, Artificial Earthquake Motions for Design,
Crystal River. Florida Site, report prepared for Weston Geophysical
Research, Inc. (February 1972).
•59-
Appendix
Description of the Kapur and Biggs Approximate Methods
:
at equipment frequency, f_.
nth natural frequency (Hz)
= acceleration value of the
of the building. (All
approximate floor spectrum
first subscripts " 1 " and
obtained by summing the
" 2 " refer to the building
in an appropriate
and equipment, 2n
manner.
respectively).
' equipment frequency.
1
KAPUR S APPROXIMATE METHOD
' modal participation factor
for mode n. For each equipment frequency of
• mode-shape value for mode interest, f , perform the following:
2
°2g
acceleration value from
response spectrum of free-
4-H
field ground motion at
If 0.9 < < 1.10, use che
equipment frequency, f~,
and damping, ^ '
-60-
If -r== < l a , Chen
in Ref. 3. The value of A t
2
depends on the structure and
.12
equipment damping and is not
listed in Ref. 3 for ^ - 0.04
[? 0
Kapur recommends that for closely
are provided so that it may spaced building modes, "the absolute
easily be extrapolated. We sum of the closely spaced modes
determined A. to be 23.0 for should be obtained before taking the
4 and 0.5% structure and square root of the sum of the squares
[5 if
step 2(A) should refer to Fig, 3
a
2g- and step 2(B) should refer to
Fig. 2 ] ,
-61-
In 2.2), C - 1.0
Fig. 3 and our Fig. A-l. One
1 25
enters the plot with the
0.50
0.40
S
"0 0.4 0.8 1.2 1.6 2.0 2.4 2.8 (? U
/» I'M
r
building frequency
lr/ 2\equi pment frequen c y / However, we followed the recom
mendation in Regulatory Guide 1,92
for closely spaced modes, just as
Fig. A-l. Biggs' amplification
factors. we did in using Kapur's method.
KC/gw
-62-