ML20147H474
| ML20147H474 | |
| Person / Time | |
|---|---|
| Site: | Oyster Creek |
| Issue date: | 02/26/1988 |
| From: | Nagai Y GENERAL PUBLIC UTILITIES CORP. |
| To: | Dromerick A NRC |
| Shared Package | |
| ML20147H481 | List: |
| References | |
| 6385F, C320-88-0270, C320-88-270, NUDOCS 8803080495 | |
| Download: ML20147H474 (41) | |
Text
{{#Wiki_filter:GPU Nuclear Corporation UCIeSr Ore Upper Pond Road Parsippany, New Jersey o7o54 L[ p[ g 201 316 7000 [g//////M/ / TELEX 136-482 Writer's Direct Dial Nurnber: hMEf7 50 ~ N? P'Og February 26, 1988 C320-88-0270 i Mr. A.W. Dromerick U.S. Nuclear Regulatory Comission Washington, DC 20555
Dear Mr. Dromerick:
1
SUBJECT:
Information Requested by A.W. Dromerick's Letter : l Dated December 16,1987 re: Oyster Creek Seismic Floor Response Spectra Methodology This letter transmits the enclosed infomation as outlined below. The infomation was requested by your letter dated December 16, 1987 concerning Oyster Creek seismic floor response spectra. As per our telephone conversation on February 10, 1988, the information is being forwarded directly to you in order to expedite your review. Our formal submittals will be made < in the near future. Although the enclosed documents are advanced copies, it is expected that there will be very few (if any) changes. l 1 Listed below are the materials being transmitted by this letter, j . Y3riatiors in soil properties - This material provides rt.sponses to Item 1 mentioned in your letter of December 16, l 1987. Part A of this material provides reasonings behind l choosing the soil property bounds used in the Oyster Creek SSI analyses and Part B presents a study of the effects of strain - , dependent soil properties on reduction of motion in the SSI analyses. Enclosure 2. Modeling uncertainties - This material provides responses to Item 2 mentioned in your letter of December 16, 1987. Under this item, NRC staff wanted us to provide published reports of comparisons between SUPERFLUSH analytical results and measured responses of structures under actual seismic events. This is ' to obtain assurance that SSI modeling and computations, as implemented in the SUPERFLUSH code, properly capture the structural responses in the field. 8803000495 080226 PDR ADOCK 05000219 P PDR GPU Nuclear Corporaton is a substd ary of General Pubic UtAtes Corporation
Mr. A.W. Dromerick February 26, 1988 Page 2 Enclosure 3. Computational parameters and their limitations - This infonnation is provided in response to Item 3 of your December 16, 1987 letter. It describes the enveloping procedures that have been followed to obtain final floor response spectra. Enclosure 4. Verification and validation of computer code with the measured results - This information is provided in response to Item 4 of your December 16, 1987 letter. Enclosure 5. Effect of saturated soil - This information is provided in response to Item 6 of your December 16, 1987 letter. It describes the soil saturation condition at the Oyster Creek reactor building area and how the saturation effects were considered in the reactor building SSI analyses. Enclosure 6. Torsional effects on structures - This information addresses Item 7 of your December 16, 1987 letter. It describes the study done to assess the torsional effects of Oyster Creek reactor building mathematical model on the computed horizontal floor response spectra. Please note that information requested by Item 5 of your December 16, 1987 letter, "Power spectral density for the time history being used" is still in the process of being gathered. It will be forwarded to you as soon as the information becomes available. If you have any questions concerning the materials being transmitted by this letter, please contact me. 1 Very truly yours i
./kl fosIiNagai Ulicensing E gineer /jbw j Enclosure cc: Civil / Structural Manager - L. Garibian (w/o attach.)
Director - Engineering & Design - G.R. Capodanno Director - Engineering Projects - D.X. Croneberger " i Lic. & Reg. Affairs Director - J.R. Thorpe " i Manager BWR Licensing - M.W. Laggart !
- Manager Engineering Mechanics - A.P. Rochino "
Manager Special Projects - E.F. O'Connor " ' 1 l 6385f ;
ENCLOSURE 1 VARIATIONS IN SOIL PROPERTIES The attached packages have been developed in response to information requested by USNRC audit staff (Ref. memo f rom Mr. A. Droserick (NRC) to Mr. P. Fiedler (CPU), dated December 16, 1987). The attached material provides responses to item 1 - Variations in Soil Properties - mentioned in the NRC meno. Part A provides reasonings behind choosing the soil property bounds used in the Oyster Creek SSI analyses. Part B presents a study of the effects of strain-dependent soil properties on reduction of motion in the SSI analyses. I l i j LTG151ume i
PART A SOIL PROPERTY BOUNDS USED IN OYSTER CREEK REACTOR BUILDING SSI ANALYSES Introduction s The following paragraphs describe the study done by URS/Blume to obtain the soil property bounds to be used for SSI analyses. These soil property bounds are considered to account for realistic soil property variations that may occur at the Oyster Creek site. Measured Soil Properties at Ovster Creek Site Area Shear (S) and compression (P) vave velocities were measured in the site area. These are presented in Figure 1. The S- and P-wave velocities are given for four layers - a 15-f t top layer overlying 25-f t, 65-ft, and 35-ft layers. Shear wave velocities increase with dapth f rom 600 f t/see for the top layer to 1.400 ft/see for the fourth layer. Compression wave velocities are also measured for these layers. Poisson's ratio for each of the layers is derived from its shear and compression wave velocities. The water table varies between elevation +3 ft 6 in. to +10 ft, as monitored in the well network at the site since 1984 The reactor building is embedded 53 ft below grade. This places the baseast of the reactor building in the third layer (dense Cohansey sand). The third layer extends below the reactor building basemat and is the most important layer for consideration of SSI ef fects on the reactor building. The third layer has a measured shear wave velocity of 1,200 ft/sec. Relative densities were also obtained from relative density tests performed on undisturbed tube samples. The average relative density of the sands in j the third layer was determined to be 94%, with the maximum and minimum values being 101% and 78%. 1 4 (g
r Best Estimate Soil Properties For the purposes of computing the best estimate soil shear modulus to be used in SSI analyses, the field-seasured seismic shear wave velocities have been used. For the third layer, this yields the following best estimate soil shear modulus: 2 0 = o V,
= 0 1200
= 5,366 kst f
where G = Shear modulus in ksf a = Soil mass density, K-sec 2/gg V, = Soil shear wave velocity, ft/see We have also calculated the shear modulus for the third layer from tested relative density (D,) values as follows: ! Average value of rD for the third layer is 94%. This is assumed to be best estimate value. Shear modulus of sand is also given by the following 1 formula { l, G = K2 /O' where og is the effective mean principal stress and K 18 2 8 Parameter which is a function of relative density and shear strain. K = 2 73 for D, = 94% and low shear strain (10""%) from Figure 2 (Ref. Fig. 5 of reference 1) The effective stress of is calculated at the middle of the third layer (72.5 ft below grade) as follows: j i LEG /Blume
P 4 Assuming water table at eleistion +10 f t (saximum level): a; - 13.5 x 120 + 59.0 x (120 - 62.4)
= 5.018.4 psf l
9 Assuming water table at elevation +3.5 ft (minimum level): a; - 20 x 120 + 52.5 x (120 - 62.4) .
= 5,424 psf '
The average e; - (5,018.4 + 5,424)/2 = 5,221.2 psf The best estimate G = 73 ((5,221.2)
= 5,274.8 ksf
= 5,275 kaf i
The best estimate shear modulus value of 5,275 ksf. calculated from the relative density test r6sults, compares very well with 5,366 ksf computed f rom shear wave velocity measurements. So, 5,366 ksf was chosen as the best estimate shear modulus for the third layer. Determination of Upper and Lower Bound Soil Properties
- To determine realistic upper and lower bound soil properties, we have exas-ined the actual soil property variations reported for the site soil tests.
For the third layer, the relative density variations were reported to be j 75% (minimus) and 101% (maximus). s i The Kg values associated with these densities are obtained f rom Figure 2 as 1 follows: K 2
=
61 for D; = 78% and low shear strain (10'"!) l
=
78 for D r= 101% and low shear strain (10 #%) 1 i 1 tr@Blume ; i i 1
~ . . . -- -. .
4 Upper value of G = K 2( !
= 78 0 Z21.2
= 5,636 ksf i
= 105% of best estigate C of 5,366 ksf 1.ower value of G = K 2(
= 61 /5z21.2
= 4,408 ksf
=
0.82% of best estimate C of 5,366 ksf ' Tros the above calculations, it is judged that conservative estimates of upper bound and lower bound shear moduli say be taken as 1.25 G be and 0.67 Gbe, where Gbe is best astimate shear modulus. The K2 values corre-sponding to upper bound and lower bound soil properties are 93 and 50. Seed and Idriss (Ref. 1) examined K2 values determined from several in-situ shear wave velocity measurement tests reported in the literature. The range of values of K2 for dense to extremely dense sands were found to be i 44 to 86. Thus, a range of 50 to 93, chosen for K2 values for the dense Cohansey sand layer at the Oyster Creek site, covers the possible variation in K2 of such dense sands as reported in the literature. Figure 3 presents the best estimate, upper bound, and lower bound soil pro-files used in Oyster Creek reactor building SSI analyses. Conclusion From the above study, it is concluded that the soil properties used in SSI analyses of Oyster Creek reactor building are based on actual soil property
- measurements in the site area. The soil property variations considered are realistic and conservative bounds of actual soil property variations ob-tained from site soil test results.
Reference
- 1. Seed, H. B., and Idriss, I. M., "Soil Moduli and Damping Factors for Dynamic Response Analyses," Report No EERC-70-10, University of l 1
California, Berkeley, December 1970. l l l I l l (LN
LAYER NO. MEASURED P-WAVE MEASURED SHEAR POISSON' S LAYER DEPTH VELOCITY WAVE VELOCITY RATIO (FT) VP (FT/SEC) Vs (FT/SEC) - _______ _____=- - - - - - q_ _ _. s-El 23 [ ,,.r..-- 1 1400 600 0.39 15 T"G.' Medium denss,ty ."..:.r.3 -
? -
- .1
> ,- ' .9'a 1. san.d . . .. *n**I.-..~.:'3 7 yg g.
. . , . . ' . ~~ = ..\
2 5200 1000 0.48 25 , y .,7. .T* ; ,Ag. .,7g. ,, . ;. ...wvel O' 72.S'
- $. ".h. .*'..Q.*
.%. *.'.**,y}$.y*
. *.*.!..*=.
*,hy
- 9. 5' h/////,,
bhny;hFi .Y:.Yi..!*?$\h)]el.$ i - EL. ; l Dennesend gj6W l:
% sM p$$)h../. ,,lq;;::y.jjj:.gg; .ig.BUILDING I
, 3 5600 1200 9.48 65 - '-- i gw , $p BASEMAT OF REACTOR e
*..;.....,...~
. . .~- . . . .
. RIDDLE . ..~, : Sand + clay 4 s'it.~.;",*[8*
' ~ * ~ '
OF 3RD l*- - ^- * '* ~1 4 5900 1400 0.47 35 LAYER , l {.j-}-{; .
.-c.. ..,'0*;[4 .:s , fij; ,j.{,. .y:;.lJg. .[.}
v- - :., 4 L 6 .-:. w: : : :: t L.J. . ,-s.. .: :: t 1;' :. a
*h:"1fi .
jfn-[Mi- ' send *
.;.i?N *N.:'kl.
;."l ~.*j M
-l-^' " ~ N- : sij S.N:;vYN:2.;:
- .. - . q. : _.g . .:..
i l'g h Bedrock NOTE: WATER TABLE VARIES BETWEEN EL. +3.5' TO +10.0' [/)
/ FIGURE 1: SOIL SEISMIC WAVE VELOCITY PROFILE AT OYSTER CREEK PLANT
~
i 1 1 I ; i
.1L $ ,
e - 8 i e $ ' _ W E E W 5" 0
/ E $
12 9 e
/ =
8
$ gW b
n iW I
?
9 E@.
/ ,
/ i I
l '- ' of g gg g . l 9 s O @@ 8 g f l m a e aa a 4i g g a gg g g g i I ' ' I f 1 I l l I o' g R 8 8 9 2 2 2 o-1 2 , 7 LR&Blume , A
l l LOW STRAIN POISSON' S LAYER DEPTH SHEAR MOOLLUS (KSF) RAT 10 (,T) BEST UPPER LOWER EST, MATE BOUND BOUND -
,,.r.. . .
El23 1342 1678 895 8.39 15 ,F . ...-. .. ,. .... ,y3*5: .' Medium deneety ..,:.t
* ~ '# !" I. I~
- . .. ~
~ -C .G- s . ~sand ., . - . .W ~?:-:, yy g. -
',- . .4 . . . . . . .
3727 4659 2485 9.48 25 ,
., Q 7 E
. . . .. ... .. . ... .. . .,_g . . .. . . e 7 I O'
~
. El-11*
. : ... -:?rx. .: g. :. '
t:: s: :Oc.t. a .< . SSRd ; '7
' "... ...'. [
[. I51 Y e
}(:'.itCohenery sendl . i!65* ",,0!
. 5366 6708 3577 8.48 65 '
s:' v :.,v,- e. . z: :p n .;f>.ii: e; :*-
-- c.* 4. _ ;.;l$ '
- Q_. ;s .
8 . . . . . . . . . . . . .. . . ..
. .*s . . Sand + CISy h Nk.***.~ S*
> ; ._ _ =.,y.; - . , . .
- _ .1 9134 4869 9.47 35 p' .! *I 7344 i'; ::e' ? .I - .. . . ,-a. ,. YE-f.?".,.J
^ *'-: .L .i ':s *.Ti.. ".' .... '
L 4+:z. .
- t. '.:. *,
.s:d.o.:..
g.;. . :
- Dense. .Nc
.-, :: . ; .- : n-sand . .4-.1700' .k.,...
.Q. I
..C
.; .1_.:#;;= -7 :
-c . - . . : .;..- .
%%/h / Sedrock NOTE: WATER TABLE VARIES BETWEEN EL. +3.5' TO +19.0'
,,_3. ,<..___,.,,,_E,,,_,__
g
PART B EFFECT OF STRAIN-DEPENDENT S0IL PROPERTIES ON REDUCTION OF MOTION OYSTER CREEK REACTOR BUILDING SSI ANALYSES In,troduction In implementing the SSI analyses for Oyster Creek reactor building, URS/Blume has used s t rain-dependent soil properties in the computer code SuperFLUSH. The strain dependence of soil shaar modulus and damping with shear strain are given in Figures 1 and 2. These are the well-known Seed and Idriss curves for sand (Ref.1). The soil shear modulus reduces with ! increased strain, and this is given as a ratio of low-strain (10~4%) shear ! sodulus, G
- m. Usually, the shear moduli calculated from shsar wave veloc-ities measured in tha field from downhole and crosshole data are assumed to be low-strain shear moduli, i.e., G ,x.
The best estimate, upper bound, and lower bound low-strain shear moduli for the Oyster Creek site are shown in Figure 3. The best estimate low-strain ' shear noduli are calculated f rom the shear wave velocities seasured in the Oyster Creek site area. The upper bound and lower bound moduli are 1.25 and 0.67 times thn best estimate values. Three SSI analyses were conducted with SuperFLUSH for the three low-strain soil profiles, and the soil layer , i properties (shear modulus and damping) were iterated to correspond to the shear strain state in the soil as per Figures 1 and 2. The floor response spectra, computed f rom these three SSI analyses, were enveloped to obtain ! the final spectra for the Oyster Creek project. l Parametric Study 1 To investigate the effect of strain-dependent properties on the computed seismic responses, we have used deconvolution procedures to compute the motion at the basemat level of the reactor building (el. -29 f t 6 in.) for strain-independent soil profile (i.e., soil properties were not varied with shear strain). The shear moduli of the soil layers were assumed to remain constant at the low-strain best estimate shear modulus values as given in l Figure 3. The corresponding soil damping was assumed to be 1%. The best
-,- LF2/Blume 1
3 estimate soil profile was chosen because these properties correspond to the field 9sessured shear wave velocities. The notion at the ground surface ' I (11. +2 3 f t 6 in.) was assumed to be the design time history (0.165g ZPA) - I for the deconvolution computation. , The response spectrum for the deconvolved action at al. -29 ft 6 in. for the above-sentioned strain-independent best estimate soil profile was com-puted. This was compared with the response spectra at the same level, com-puted for tha project using the envelope of spectra f rom strain-dependent i i best estimate, upper bound, and lower bound soil profiles. The ratio of the spectral amplitudes of the project spectrum to the strain-independent response spectrum is plotted as a function of frequency in Figure 4. A ratio value of less than 1.0 implies that consideration of strain-dependent soil properties and enveloping results from the three soil cases give lower response spectral amplitudes at that frequency than a strain-independent analysis with best estimate soil case would yield. As shown in Figure 4 (shaded area), in most f requency ranges, the spectral amplitudes are in-creased (by as much as 50% to 70%) due to consideration of strain-dependent
- soil properties and enveloping the three soil profiles. In the rest of the frequency ranges, there is a reduction of less than 101, except in a very ,
narrow range, where it goes up to abuut 20% reduction in a f requency range of 2.5 ha to 4 ha. ) l i To further investigate the effects of soil strain-dependent properties on j , soil motion, we considered a case where soil shear atrain is assumed te . I consistent with Seed and Idriss curves at a strain of 10~ 32. This is a very conservative estimate of the possible soil strain during an SSE event. The best estimate soil modulus at 10-3% strain is 93.4% of the low-strain (10 4t) value. Corresonding damping is 1.6%. Deconvolved motion at this soit condition was computed, and spectral ratios similar to Figure 4 are plotted in Figure 5. As shown in Figure 5 (shaded area), in most f requency j ranges, the spectral amplitudes are increased due to consideration of s t rain-de pendent soil properties. In the rest of the frequency ranges, )l there is less than 10% reduction in spectrtl amplitudes. 4 ESANume i i l
Conclusions It is concluded from the above study that the consideration of strain- ' dependent soil properties in Oyster Creek SSI analyses does not lead to ' significant reduction in basemat socion. In most frequency ranges, spec- , tral amplitudes are increased due to the strain-dependent soil cases ehen compared to strain-independent, best-estimate soil case . In other f re-quency ranges, the dif ferences are less than 10% when compared to responses obtained assuming very conservative soil strains (10~3%) during seismic ' action and only 7% reduction in soil shear modulus and 1.6% damptng in soil. Even for the extreme case of no reduction in soil shear modulus and i 1% soil damping, the reductions are so'stly less than 10% in all frequency ranges of interest. References
! 1. Seed and Idriss (1970), "Soil Moduli and Damping Factors for Dynamic Response Analyses," EERC Report 70-10 University of California, Berkeley.
i j I I
- l i
I i , i I k e f 4 1 l 4 I j LES/Blume 4 i l
9 d g t I
- C3 L g
.5 s -
;: .L 2
m 1 v1 u 2.
? -
a Q M k e 0 0 8
' 4
.5
. =
' 7 a s
Q . a 8 a i a w e l ? C bw
.2 u
i S I' E a C t C i i i I i i i i 1 & 9 9 h 9 9 * . 9 N 7 0 o o o o o o o o o suoned,,ot .,4 to sninPON .tesyg
- h A 'uisangassysassntnpmasey[ (N
! 30 g i t
28 - ,
- 2 6 -- _
a
' 24 - /
/
22 - ' ~~
~'
E 20 -. ~ o _ ~. e 18 - - a 9 IS - "! y i i 3 o 14 -
) ~ , 3 12 -
8 n. 2
< 10 - -
o - 8- . 6-j. i Lf - g, . - i 4- .
- s 2- , s' g; O "
1 x e ~_ O jg 4 /0 ~' lb'* /O*' /O" l0 s i SHEAR STRAIN , percent l.
l
- 1 l
l l L
. Si as g I h_
w h_ _m ww $ G h"'""! 5 =ba* --*~ -- Ik P.:.~.u7.(.h, s:;m%:Phy..*'
;r '. t , * , .Q.!,d..e.:.:-"'
.a ..:; : c:n ..:x.::- ze
". .' *th'! < .3.:'- .u .- *..! .;; . - Y. *+ l
.**.;.'.., , xy l8;::. .;-ll o .*
o
?. . .
. ,'-. F i .'
. ' V.*.i [.* ;. . '8 ;*$' '
=g:;.f).:( .:'/;
l i'fpv> R$O:+l I{ ,....:M..- i l5,. h
$.$6-4..
5
. 'n. s.,
.. c
. .+.m,%.. . . . . .J . R ,.. z.
. fL4..t"1.r.t ;woll!.
e - 5 I. y *.e
. . 1
.; s:: / .
** ' . -: 'fff U.h.%.'
3 W W
.*,..'a'!
!.,e.s g.
d.'*.]"dE2 *:
!3!!F
' . . it.. ;. # y. .
.v
- i. lf::7'?. .
.f, U
N:'*'5$p
.d.i$3.... . . .
g
.'1 .,bE b;e..pu: 6. ..i?h * .*khahs
- Y.s x
' l .$.'.
. . N.bh5 y s1 _
i m s 4 I w a I a. l l g .$ g
** g g a 5 l El I I E : ;
i si B s a a s E R en m e s a g a * *
- IaE m Is! !-
-l g _= s a
- a a
c
> 4 I 4===
4 % SMS
-a E
. l l
#- l
[ Mlg- ' 1 l
l . 2
. 1.9 -
3
- " 1.8 -
2 3. 1.7 - ( 2.j 1.6 - , 1.5 -
- a. - .
} 1.4 -
,- '/
!! g 1.3 - [ S' ,
{
$ 1.2 - . -
R},
}i } ~
S,'7.
/ //-M::&i 1
A =; 0.9 - w v ij O.8 - N5 ' 5 0.7 - 55 0.6 - 2 :
- : 0.5 -
1 L O.4 - [ .[ 0.3 - a 3 0.2 -
- e s 0.1 -
5 0 , i . : : i C 10 20 30 40 FREQUENCY (HZ.) 4 Figure 4. Effect of Strain-Dependent Soil Properties on Reduction of Motion at ( Basemat Level (El. -29'-6"), Oyster Creek Reactor Building i l
$ 2
/
f 1.9 - j)
~2 1.8 - '
- 2 1.7 -
- u 1.6 - /
- 0 /
u a 1.5 - / 3 j 1.4 - / 7 7 '/'7 '
/
1 / / /
// ~
$$ 'UC hsu /
<// /$'E/b, '
'~
i* i L - 1 0.9 - M v 3 g 0.8 -
*5 5 0.7 -
- j. j 0.6 -
- : 0.5 -
55 O.4 -
$ 3 0.3 -
t ai
=
0.2 - (( 0.1 - - 2 2 O i i i i i i : 0 10 20 30 40 FREQUENCY (HZ.) Figure 5. Effect of Strain-Dependent Soil .'roperties on Reduction of Motion at Basemat I.evel (El. -29'-6"), Oyster Creek Reactor Building 2
- ~ . - . .
,g
;7 Ssismic soil-structura interaction ef fects at Fukushima Nuclear Power Plant in the Miyagiken-oki earthquake Masafumi Narikawa The Tokyo Electric Power Company, Ir.c.
Takekazu Udaka Eachquake Engineering Technology, Inc. Mitsuo Okumura Kozo Keikaku Engineering Inc. l l 1 INTRODUCTION ' " The finite element method provides a powerful tool for the. solution of seismic soil-structure interaction problems especially with complicated properties, geometries, and boundary conditions (EET 1983). However, only limited field observations of strong excitation at nuclear power i plants are available at present. This paper describes soil-structure I interaction analyses of the.Fukushima Daiichi Nuclear Power Plant complex using motions recorded at the plant during the Miyagiken-oki earthquake and evaluates the adequacy of the analytical methods used in current design practice. 2 DESCRIPTION OF THE REACTOR BUILDING L The reactor building, which houses a 1100 MWe BWR Mark II reactor is constructed of reinforced concrete and is structually isolated.from the adjacent turbine building and radwaste building. The reactor building is about 73m in height from the base of the foundation mat and is 68.5m x 68.3m in plan at the basement level and 45.5m x 42.5m in plan at the roof level. The reactor building is partially embedded and is I founded on a mudstone at a depth of 17m below ground surface. { l 3 HIYACIKEN-OKI EARTHQUAKE OF 1978 AND OBSERVATION SYSTEM On June 12, 1978, the Fukushima Daiichi Nuclear Power Plant experienced { a strong earthquake. The earthquake was assigned a Richter magnitude of l 7.4 with the epicenter being approximately 140km from the plant and the i focal depth being 40km. Four accelerometers of the moving coil type were installed at the time of the earthquake and all four recorded motions. The accelerometers are located under the roof (P01), on the refueling floor (P02), at the top of the base mat (P03) and in the mudstone at a depth of 31m below ground surface (PO4). The duration of strong motion was more than 30 seconds. The observed peak accelerations were 145 to 208 Cal at the plant roof level and 60 to 84 Cal inside the mudstone. Following the Miyagiken-oki earthquake, the total number of observation points has been increased. Cross-sectional view of the locations of the seismographs are shown in Fig. I and more explanations are available in reference (Narikawa 1987). ia . . .: .e. 4.. . . . . .. L . . . L . . . , . . . n .1 /. L . . e A . -
+
l S.wja, ijnecin
, og,i*
. n. . . ..unn 4m,
... N ,
I < *
,1,
< &M,
-..n.w A A L0 1 ""J l " -
1 4 I
??. .
\ v;p ;o n
, - !!! C"
]
U6 3"
~
l F i o C-) g :
-1
- a _ C" = ,, , L :8
. -i
~I y
- n3 I 22 2 ::!: 2:::<-
., n ,
Fis. 1 Stismoca u n Locations Fis. 2 Soit Pnoptaties Fis 3 Moett 4 SOIL-STRUCTURE INTERACTION ANALYSES i l 4.1 Outline of analytical approach ' The computer program "SuperFLUSH" (EET 1983, Lysmer 1975) has been used in order to analyze the response of the Unit 6 reactor building during the earthquake. This program uses the complex response method of computing the response of a finite element model. In this study, seismic soil-structure interaction analyses have been carried out using a special deconvolutional procedure. Commonly, in soil structure interaction analyses the control motion is specified at the surface of the free field away f: om the structure. However, the locations P01 to PO4 are too close to the reactor building to avoid soil-structure interaction effects, and may not be considered to be free field motions. In this series of studies, the control motions were considered to be at two of the ob-served locations in the finite element model and a deconvolutional procedure has been used to estimate the free field ground surface motion. ne deconvolutional procedure was applied to the recorded accelerations at both P03 and P04. HowcVer, since the ratio of vertical excitation to horizontal excitation was negligible, only the horiz:ntal component was considered at this location. The results obtained from a previous study (Narikawa,1987) based on a multi-point observation system were also utilized to obtain better understanding of the soil-structure behavior at Unit 6 reactor building site and the motion recorded at Unit 1 site was also used as a control motion for simulation of the response of Unit 6. 1 4.2 Material properties The soil properties used in the analyses are shown in Fig. 2. The same shear wave velocities as used by Narikawa (1987) were used for the soil-structure interaction studies. The dataping ratios used were higher than those used in the previous study because of the higher level of excita- , 1 kTM 2 Su sea.:A dm iAs edw ced: e R h W eim'< W
s tion. n o reactor building itself wcs.reprasentsd by three. flexible " - barms cod tho fcundction was mod 31cd by a sories of rigid bacms. Tha - structure properties are also identica1' to those used by Narikawa (1987) and were determined by computing effective shear areas, mass moments, moments of inertia and masses based on the original blue prints of the reactor building. A damping ratio-of'2% was assigned to all structural components to be compatible with the previous. analyses. The finite element model used in the analyses is shown .in Fig. 3. A semi-infinite half. space (EET 1983. Joyner 1975) is assumed at a depth of 65m below ground surface and energy transmitting boundaries-(EET 1983, Waas 1972) are attached at the vertical boundaries.in order to transmit energy beyond the finite element mesh. 4.3 Analyses based on the motions recorded at P03 and PO4 Comparisons of the recorded response spectra and. computed response spectra which assume P03 as a control point at the foundation mat are shown in Figs. 4(a) through 4(c). The. solid lines and dotted lines in the figures are recorded and computed response spectra, respectively, i The comparisons show excellent agreement with each other. Similar com - ! parisons of recorded and computed response spectra are shown in Figs. 5(a) through 5(c), taking the control point to be P04. The. discrepancy between the observed and computed motions at PO4 in the .ange of 0.10 to 0.30 seconds that is shown in Fig. 4(c) significantly af fected the. 1 structural ruponse, nis discrepancy may be due to the irregularity'. 1 of the geometry and material properties of the real soil and rock layers. l l
!)
. -l l
! . :ss,unisas .l l.
-'- . g- . n s ownia u ,c
! 1"",' "'
u.
~
=m.n. ;,g
. c
~ 3
" . .. . s il - ;;
. l .3 - .
5 h I '.
. y .
- i. j .
_ ~
~ ,
a_ . amnim u = [ -
. . g* 'u i
. ... ,e n ,, , ,
=c . <,i l
i
- - . wi. .
vi- lg. - . I ! I
,t - - .t's
' l l i_ . imni= u .--
. .. . ..' nl , m u ,,, ] i
'O
! 1""' 3 - =c
..c,,,,,. co g!" -
t
?
l.l-r j , l- . . . M.% ,.- - ,
~ -
fn t (seasiteot W % seosu $*ttsea em Case.et %'im as Al f n. % (somines er hseou Saicisa se (a. ins %ne si 4 kW7 3
..w s . onw,.
r.. ..Se.da . ..c . 't,cunu c fu e rt.. .A.G4 o c. t i s e.H 8 (f:. a2 L -sW w . . a .. . . , a ns . . e A.
4.4 Anclyses cssu2ing c loccl irragularity ct PO4 - - In order to compensate for the disert.pancy observed at location PO4, local irregularity of the material properties has been introduced to the analyses. n e properties in the vicinity of location P04 were assumed to be slightly sof ter for the upper layers and slightly stiffer below 30m as compared to those used in the free field. The incident wave component of the recorded motions was then computed at a depth of 143m below ground surface and used as the control motion in further analyses. n is procedure is illustrated schematically in Fig. 6. Comparisons of recorded and computed response spectra and transfer functions in N-S direction are shown in Figs. 7(a) through 7(d), and Figs. 8(a) through 8(c), respectively. The transfer functions obtained are relative to the motion at P04. Figs. 9(a) through 9(d) show comparisons of the recorded and computed response spectra in E-W direction. The computed values are in good agreement with the recorded values. PP M M _
~4,~ o ,,- m,,, a_ . si nani u m u, .
E'M Z
@~= = !. ZD E
E., t
! I -. , !
! ! ll s..J.<
%# Pett Dit sum
..... u .... ... ....,,. . . ~ ,,,, i- . .i i.ni= u = .
1.'eoewsies Lets i.e snasi , j - es cess e (33
. ---- Coev,i s gg _ .
I
- g. .
e- ~._ -.-.,~_.
. B-l llEMI a .3
,,, . hi m, u i'i = = r,; - 3
- = - .
,o in l
_ --. c 1 g, , # = E , , 7' ,. - I
.a. , , . . I ,,, .
i [ .. . i'.j n .,, = = r,: - }..
-,_ . ent. u .
, as e
, , ; k. (**e 'i o g
4 sg
- ae -
bga" = t'!.""il , g j., . - I-5 + , ~ -
.. ~. % 1 **i. 3. .. em am na 7 ACMPdCT ll'l2) fis. 6 ( .sius a lesewta f.mc.iwa in. 7 (s==sisms e 8tiemu letteseas wies a tat 6 i. .sies si Rn in-la MS M O Y Pio 2 -
- H.y.pa .c4 wt6) J s. g
4.5 Anclyses b:s:d on ths motion rccord:d et Unit 1 The motion recorded at Unit 1 at a depth of 14m below the ground surface was also used as an input motion and the corresponding incident wave component was obtained at a depth of 143m using the mudstone properties given by Tanaka (1980). Figs. 10(a) through 10(d) show comparisons of the recorded .snd computed response spectra at Unit 6. Good agreement is again obtained*.
~ =
{ . (.s $ltl'fl3 A1 PCl , g. , 8 l lit ((11311 PCI j -na casm s IA>
]* -== hteen s IA 5 *. (se ette . . . . Campett e g!- -
j ,.g - 5 ei 5
- g. ,y l. g.
A
~
i r .. - r i _ . <- >ieni= ar m . i_ . ..i i.cn. .i .
; - k= = *
**~ (8*v't.
'8' 3
au . . m
- . 5 ... ( ,,,t 3 ll = -
j j -. - t 4 l= . .
...-=*a g
.= .
i i - i
- g. , E 8 3lECfl311 P03 . g. 5181ECfl3 Af PC3 3 - ===. i(>
3 - = c. . ici
- . * * = (avvite =., -~. Ceestas
{j = - . l= . . t y g- . .
- g. .
"~ , . _
-m -
- g. , ( V llEClla Af P0e ,
- g. , 8 5 $]E(Il3 Af 80s ,
3
=<.a. m 3
===. m 5 =.
gg . == Carvits . lg _ . ~*= (aref t3 . I t
- g. .
g-
*=,. ww .. ..
~
.. ' = , - a. .w s . ..
Ma!OD tttCM) MaICO 16t!MI f at. 9 (awasinses w kneerse $nties Asseim fis, la Cavasises w tapou Lettiaa laws e a loca lostuste at P0m (L-81 saa % tion kceses s' We 1 ( Ar'91 4 cet,,,u .,.n. e m e.u.v.d + .ese a % u u- w
5 CONCLUSION In order to'obtain a better understanding of the behavior of the reactor building, a number of dynamic soil-structure interaction analyses have i been performed using the motions recorded during the Miyagiken-oki l earthquake of 1978. The results of the analyses show good agreement ' with the recorded responses. The results presented herein along with ; those presented by Narikawa (1987) indicate that this class of soil- ; structure interaction analyses can be very useful in evaluating the ' observed responses of reactor buildings and should give increased confidence to their use in design analyses, i l ACKNOWLEDGEMENT Many valuable suggestions from Dr. H. Tanaka are deeply appreciated. l 1 REFERENCE ' Earthquake Engineering Technology, Inc. (EET) 1983. SuperFLUSH Manual Vol. I - Vol. III. , Joyner, W.B. & Chen, A.T.G. 1975. Calculation of Nonlinear Ground I Response in Earthquake. Bull. Seis. Soc. Am. 65:1315-1336. Lysmer, J. , Udaka, T. , Tsai, C.F. & Seed, H.B. 1975. FLUSH - A Computer Program for Approximate 3-D Analysis of Soil-Structure Interaction Problems. EERC 75-30, U.C. Berkeley. Narikawa, M., Udaka, T. & Okumura, M. 1987. Seismic Soil-Structure Interaction Behavior at Fukushima Nuclear Power Plant Based on Multi-Point Observation. 9th International Conference on SMIRT K5/8. Tanaka H. & Nakahara, M. 1980. Investigation of soil-Building Interaction Behavior of a BWR Plant During Miyagiken-oki Earthquake of 1978. Proc. of 7th World Conference on Earthquake Engineering Vol. 6. 3 Waas, G. 1972. Linear Two-Dimensional Analysis of Soil-Dynamics Problems in Semi-Infinite Layered Media. Ph.D. Thesis, U.C. Berkeley. I I i
~~
K5/7 h
%er sen %%s. rut:n ettntt, er W.s m, $er c v :. ,6. ......~,..-.a e
g Ssismic soil-structure interaction behavior at Fukushima Nuclear Power Plant based on multi-point observations Masafumi Narikawa The Tokyo Electric Power Company, Inc. Takekazu Udaka Earthquake Engineering Technology Inc. Mitsuo Okumura Kozo Keikaku Engineering Inc. 1 INTRODUCTION Evaluation of soil-structure interaction effects is one of the most important processes in the seismic design of nuclear power plants. In order to better understand soil-structure interaction effects and to accomplish more safely designed structures it is therefore very desirable to obtain observations of field performance and to perform analytical simulations for both large and small earthquakes. On September 14, 1982, a minor earthquake occurred off the coast of the Tohoku district of the Main Island of Japan. The earthquake was assigned a Richter magnitude of 5.0 with the epicenter being approximately 37km from the Fukushima Daiichi Nuclear Power Station of The Tokyo Electric Power Company and the focal depth being approximately 60km. Acceleration records were obtained at thirteen locations both within and outside the Unit 6 reactor building. Analytical simulations have subsequently performed to ) i study the dynamic response of the reactor building and to examine the ! adequacy of the analytical methods used in current design practice. ; 1 2 DESCRIPTION OF THE REACTOR BUILDING The reactor building of Unit 6 of the Fukushi.aa Daiichi Nuclear Power Plant (a BWR Mark II type, 1100 MWe) is approximately 73m high from the bottom of base mat (OP-4.0m) to the top of the structure (OP+68.6m). The building is partially embedded and is founded on a mudstone at an ; elevation of 17m below ground sur'iace. The plan dimension of the I reactor building is 68.5m x 68.3m at the bottom and 45.5m x 42.5m at the top. The reactor building is constructed of reinforced concrete and is structually isolated from the adjacent turbine and radwaste buildings. 3 MULTI-POINT OBSERVATION SYSTEM Cross-sectional view of the approximate locations of the seismographs are shown in Fig. 1. Two seismographs are installed at the roof level (Op+65.5m, P01 and P11), two at the refueling floor level (OP+51.5m, P02 and P10), one at OP+19m level -(P08), and two in the basement floor (OP+1m. P03 and POS), resulting in total of seven seismographs inside the reactor building. Five seismographs are installed outside the reactor building: two (PO4 and P13) in the mudstone at a depth of kMP I
. . ;. 4. . /.. . . . . . r a . a . l. s~ i . 4. . . n .1 7- . . .a . ././... ..
-31c below ground surface near tha reactor building, ona (P14) et a dspth 1
of -143m below ground surface, one at the surface (P07) and one at depth j of -17m (P12) . 'the seismographs P07 and P12 are located 130m north of the reactor building, and are considered to be located in a nearly i perfect free field environment. The observed peak acceleration during the earthquake on September 14, 1982 were 26 to 28.6 Cal at the plant , roof level and 20.5 to 26.5 Cal at the free field ground surface. The j approximate duration of motions was 40 seconds. J l 4 SOIL-STRUCTURE INTERACTION ANALYSIS l l 4.1 Outline of analytical approach i Dynamic soil-structure interaction analyses have been performed utilizing i the computer program "SuperFLUSH" (EET 1983), which uses the complex ! response method of computing the response of a finite element model l (Lysmer 1975). The recorded motion at the ground surface in the free j field was taken as control motion for the analysis. A semi-infinite half space was assumed at depth of 65m at the bottom of the finite ele- < ment model. A technique for separating incident and reflected components I of the incoming motion was used and only incident component is applied I at the bottom of the finite elenent model. The reflected component, I including the disturbance due to the reactor building, is absorbed by ! the half space (Joyner 1975, Udaka 1981). This approach eliminates any l uncertainty caused by the assumption of any arbitrary depth for the finite element model. Energy transmitting boundaries (EET 1983, tysmer 1975, Waas 1972) were attached at the vertical boundaries of the finite element model to simulate the existence of semi-infinite soil layers beyond the finite element model. Soil-structure interaction analyses were performed in both N-S and E-W directions and rocking motion of the reactor building along an axis approximately 45 degrees f rom the N-S direction was computed by taking the resultant accelerations of the N-S and E-W excitations. Torsional behavior of the structure was also simulated by use of a traveling SH wave concept. 1 Sees avg ( LO(i f, O leg
. . . .ie >.i A 1
. . . N ,
A A% ;
-,,. n .
" AA l L0
. 3, M -
- ~
g
' M 3a 2
n-
-n j E n, p
C C 4-fa ._m:- g ; L e n - , a -!
..n - - . .
o --
. tia --- e. ne.
33 H!!!! 2 2 2!!!f* Flo.1 Stisacsam Locations Flo. 2 Soit Paestarits Fio. 3 Moctt I K C/8 2
&2 sri.s' m - ; w.ca.-*r u r % ! Km s. % \'.. m h v.
G e.-t em.J m ni . W .* s sw.%
l . 4.2 Materici prcperties
'Ihe soil properties used in the analyses are shown in Fig. 2. The reactor building was modeled in the N-S direction by three shear beams whose properties were obtained by computing effective shear areas, mass moments, moments of inertia, and masses based on the original blue prints for- the reactor building. The base mat was modeled as a rigid beam.
A damping ratio of 2% was assumed for all structural components. Fig. 3 shows the finite element model used for soil-structure interaction analyses. 4.3 Comparison of recorded and computed responses 4.3.1 Horizontal motions Fig. 4 shows comparisons of the recorded and computed responses in the N-S direction at P10, located on the refueling level, at P08, located on 0?+19m level, and at POS, located in the basement. Similarly, Fig. 5 shows comparisons of the recorded and computed responses in the E-W direction. All the response spectra are computed for a damping ratio of 5%. The transfer functions shown in Fig. 6 for the N-S direction and in Fig. 7 for the E-W direction are relative to the free field ground surface motions (P07). A comparison of the recorded and computed maximum accelerations is tabulated in Table 1. The computed values are in extremely good agreement with the recorded values. m - I . ' 6
- i. . ..i nmna u ,i.
. ,_ . ... .i m u , , ,,,
i*. =c==>
* ** ~* (er etv e iu :
=c .. us
$ .. ...... (e,y, g g il - .
lj- . I I l+# . g... d 1 d
'1 I I I
- i. . si nmu u u . , . i. uimi= u .
i au... m =<==. m
=
l,-
.. . .. ( . ,s s
-{ .~~~ Cow ves e g .
gg. . I l- . g ... .
- -W . .
-m ,.
, . ..s oun= u m , c4 mmu= u m I
= c. .
-i_ an==s ici I, (e itis
,(i
(, ~-~~ Cee.it s l gg- . gg. . I t g... . i w s. M*!OC (MCDC) i .. ,
>. , m.
Mal 0C fMCact Iss. e (ee ene, y hymsg 1,gge 4 eg.gi p ,,, g g,,,, , , g, ,,y g, , ,g q , K r/r a has e.
.. u.1 . .. p
. ne r. h t.e~ be m a s , % w.c= pe m
. 1
" i . . .
..! ,. . .......u.. - ,a;- ,; , , , - ! " p, i.....un- --;gn; ,,, -
su - vs - j,. - ..su . - g,.
-a,.
s
' '5 " - -
. .a..
s
, _ - x% , -
.. i i . . i . i j
- g. ".
i
... .. .u n. -- ;g ,; , , , - ! .'. . .. .. .u , i
__ .g. , i
. . . - %. - . . . - %,.u - I
.. - e.iu . - K,. - >an -
t ,
% ._- _ C *l % .___
l i i . . . . i i
!.. . . . . .u n .
---. .gg; , , , - }..
.n
. . .. . . .u n. ---- .gn; ws l
- 1. .
l 9 ,.. - 2. n . - g,. - *.a . - i
! " ;an ~ l i ..
- a. ! ' D.= w= ..
.. a an !
rata.t<v bzl ratAt<r inti l Iss. 6 (s=. twas c. f.asus. F.,.< tic.s 1413 fis. 7 (cp=..a,$ t .s f..awt. Twicisms 6( si
)
i Tabt. 1 C.ersarlson of Maat.un Accelerat tene Pot M2 rtn t.u s rn) Poi M s p.u .O n th.u J J. ) i n'.It'd. a.7 7.m tismrit u 24.s. 2o.4 2't . 4 9.2 n.t N.I g,g as umiu 18. t. Jt.8 20.2 i t .u ""~it'.~n-* '.v 7 cmrmn 32.7 io.) is.) 1o.2 s.5 a.s uait can 4.0.2 P.ocking ne seiscographs P03 and POS are located on a line which is nearly 45 degrees from the N-S axis. Therefore. the accelerations obtained by the subtraction of the observed vertical components at P03 from POS can be assumed to be twice the rocking component of the foundation along this axis. Likewise. the computed vertical responses due to both N-S and E-V horizontal excitations at the foundation level are simply com-bined to obtain the rocking response on the axis 45 degrees from the N-S-direction. A comparison of the recorded and computed rocking response spectra is shown in Fig. 8. D e computed response is again in excellent agreement with the recorded response. The minor differences that can be seen may be due to the assumption in the analysis that the structure is perfectly symmetric when this is not actually true.
.-I ,u.. , .I ..i ,m i -
- c~~ -
,y .
145 Ps:s gl . . e.s g
, , ... .. ..; ..m _ y.u.
g .. . 3
.o,. .. . .. i 8 PLA23D IMCael "A I .
- 0 <
s.ag , .' { tt.f tB .t . 61 2 i Ice. 8 (a'. 160. .F .t .pngt set t t..
. . . . . . , s. w ., u m ii . t t=u% *na ...r n h-. . u. s sn.a A A. .;.1 w d.*o k-.h,** R.h sM . = u le W
'4.3.3 Torsional behavior In order to evaluate the torsional response of Fukushima Unit 6, the base foundation was assumed to be perfectly rigid as shown in Fig. 9.
The torsional components with respect to the N-S and E-W directions can then be expressed as follows: Ens = (P03NS-P05NS)/Rg,
,, Eq. (1)
%, = (P05EW-P03EW)/Rus If I. is equal to 5,,, because of the assumption of rigidity, the foundation has a torsional movement with angular acceleration of Bus -
The observed torsional motions were then computed using Eq. (1) and the response spectra for U : and *ds w multiplied by a factor Run as shown in Fig. 10. It is apparent that the reactor building has a rigid body torsional component in the order of 10% of the translational component. The corsional components En Re, and Es,Rus can also be calculated using the recorded responses at P03 and traveling SH wave concepts (Udaka 1979) with phase velocities of 2000m/see in N-S and 3000m/see in E-W directions, respectively. A phase velocity of 2000m/sec corresponds to an incident angle of approximately 15 degrees to the vertical while a phase velocity of 3000m/see corresponds to an incident angle of 10 degrees. Comparisons of response spectra based on the observed motion computed using Eq. (1) and the traveling SH wave concept are shown in Fig. 11. The results shown suggest that torsional behavior at the site may be well explained by the traveling SH wave concept. Finally U : and Es, were computed using the responses computed in the soil-structure interaction analysis and traveling wave concepts with phase velocities of 2000m/sec for the N-S and 3000m/sec for the E-W directions, respectively. Comparison of response spectra for recorded and computed torsional components are shown in Fig. 12. The good agree-ment that can be seen between recorded and computed results should in-crease confidence in the concept of traveling SH waves for evaluating the torsional response of structures. The computed torsional component 8=s Re, was then added to the computed translational components. The computed response spectra with the tor-sional component included are shown in Fig. 13. It may be seen that the torsional behavior has no significant effect on the total translational behavior in this case. 5 CONCLUSION The results of the analyses of horizontal and rocking motions show excellent agreement with the recorded responses. The torsional move-ment of the reactor building was also studied and was found to have no significant effect on the horizontal responses. These results indicate that the methods used herein are both useful and suitable for evaluation of soil-structure interaction effects and also that analysis based on motions generated by small magnitude earthquakes can be very informative in evaluating the adequacy of analysis and design procedures. w k.r/s c s.. e s e % e. % v s pi x
- l
.*, 1 a
1 l l ACKNOWLEDGEMENT l l Many valuable suggestions from Dr. H. Tanaka are deeply appreciated. l l REFERENCES l Earthquake Engineering Technology, Inc. (EET) 1983. SuperFLUSH Manual Vol. I - Vol. III. Joyner, W.B. & Chen, A.T.C.1975. Calculation of Nonlinear Cround' Response in Earthquake. Bull. Seis. Soc. Am. 65:1315-1336. l Lysmer, J. , Udaka, T. , Tsai, C.F. & Seed, H.B. 1975. FLUSH - A Computer Program for Approximate 3-D Analysis of Soil-Structure Interaction Problems. EERC 75-30, U.C. Berkeley. Udaka , T. , Lysme r, J. & Seed , H.B. 1979. Dynamic Response of Horizontally Layered Systems Subjected to Traveling Seismic Waves, . l Proc. 2nd U.S. National Conference on Earthquake Engineering, Stanford. ; Udaka, T., Okumura, M. & Tada, K. 1981. Soil-Structure Interaction j Analyses for Verying Seismic Environments and Boundary Conditions, j 6th International Conference on SMIRT. k(a):k3/1. , Waas, C.1972. Linear TVo-Dimensional Analysis of Soil-Dynamics Problems in Semi-Infinite Layered Media. Ph.D. Thesis, U.C. Berkeley. us a. l l l l l j I.+ in . . t .s. ,. i. . !*"" - Nc =st ii. - g , ,,,
-. - - u a. wi. .. c., i;. -- - - ce*,'in <w an % . l 1
, rou n6.. rom /in gj VI !
s~~ -
.d l--
'W
.- - w . -
Ma330 gggpos ,,,,,, n .is w iu. = u s,m u ,.
! *" i
% = <== . s.un., u .n i. = =-' . .. c.s a...,,,, . . .
l
--- kc.no tu g Wie, 1
natsosuceet
.. l
,.=- P u n6.'Mah/sl .
jj Fis.12 kc. ness a.s Coverse es se lectu
. en v.iin . i.u ., u,n I ...
l ... .
._ JK .
ii, i., l
]
l l n- -....i.,s. a~ Bu.ese tW s.4 Ca.cier ii,
- a. .
-I, . .-
Bat ten t P se M6. 20rsus) . b *~-*- -*
- Coev't e Il...
11 - l l... . l--
?.. N ,- .. .. .
Je. .. ..
*tesco suceen pseloo estcoei Els. Il 8E 64 S.f t f aa 8. testie.
. Osianas a som W s.. hee gig,g gi m Winatla m t M
,,,,g,g,g,,,,,,,
r .. <........a. ......+. .... ... . . . , , . . . . . . , . . . Y# C
ENCLOSURE 2 MODELLING UNCERTAINTIES The following paragraphs describe our response to the modelling uncertain-ties item mentioned by USNRC audit staf f (Ref. Memo f rom Mr. A. Dromerick (NRC) to Mr. P. B. Fiedler (GPU), dated December 16, 1987, iten 2, page 1). Under this item, NRC staf f wanted us to provide published reports of comparisons between SuperFLU5H analytical results and seasured responses of structures under actual seismic events. This is to obtain assurance that SSI modelling and computations as implemented in the SuperFLUSH code prop-erly capture the structural responses in the field. The following documents are submitted herewith in response to the above-mentioned USNRC request:
- 1. Validation of SuperFLUSH code with Measured Earthquake Results : The ;
following papers present comparisons of SuperFLUSH analytical results and actual measured earthquake responses at Fukushima Unit 6 plant (BWR MK.II, 1100 MWe) in Japan. The earthquakes were Miyagiken-Oki (June 12, 1978, 7.4 magnitude , 140 km epicentral distance) and Tohoku (September 14, 1982, magnitude 5, 37 km epicentral distance). i la. Narikawa, Udaka, and Okumura, "Seismic Soil-structure Interaction Effects at Fukushima Nuclear Power Plant in Miyagiken-Oki Earth- j quake," Paper K5/7, SMiRT-9, Lausanne, 1987. Ib. Narikawa, Udaka, and Okumura, "Seismic Soil-structure Interaction Behavior at Fukushima Nuclear Plant Based on Multi-point Observa-tions," Paper K5/8, SMiRT-9, Lausanne, 1987. ! l
- 2. Comparison of Analysis Results from FLUSH Group of Programs with Hea_-_
suced Earthquake Results: The following papers compara results from SSI analyses using FLUSH group of programs with measured results from actual earthquakes in nuclear plant structures or scaled models. LF2/Blume
~ . - - _ _ . - __ _. _ _. ._ __ _ _ _
2a. Valera, Seed, Tsai, and Lysmer, "Soil-structure Interaction Effecta at the Humboldt Bay Power Plant in the Ferndale Earth-quake of June 7, 197 5," J. Geot. Engrg. Div., ASCE, Vol. 103, No. GT10, October 1977. 2b. Berger, Fierz, and Kluge, "Predictive Response Computations for Vibration Tests and Earthquake of May 20, 1986, Using an Axisym-metric Finite Element Formulation Based on the Complex Response Method and Comparison with Measurements--A Swiss Contribution," EPRI/NRC/TPC Lotung SSI Workshop, 1987. 2c. Sato, Nakai, Yamazaki, Mita, and Ishii, "Soil-structure Interac- I tion Analysis of Quarter-Scale Model Using AXERA Code," EPRI/NRC/TPC Lotung SSI Workshop, 1987. I 2d. Tseng, et al., "Soil-structure Interaction Analyses of Quarter- l Scale Containment Model Experiment in Lotung, Taiwan; Part 3: FLUSH Method," EPRI/NRC/TPC Lotung SSI Workshop, 1987. i l l l l LM/Blume
1142 OCTOBER 1977 GT10 N OCTOBER 1977 GTM Board. Vol. 37,1958. pp. 576-581.
- 36. Washins. R. K.. "Influence of Soil Characteristics on the Deforniation of Embedded Fiendde Pipe Culverts." Bulictm 223. Highway Research Board 1959. pp.14-24.
g7. waams. . x -ra.iu,e Cond.t.or of no.bie Culvuts Entdded in Sod." Nccedsags. Highway Research Board. Vol. 39,1960. pp. 36t-371. UMR N T '-C
- 88. Wessergaard. H. M.. "Computation of Stresses in Concrete Roads." Nccedings. G WTL-c Hashway Research Board. Vol. 5. Pt. I.1925, pp. 90-I12. r^L' a bM rg.yg,y,p undrotighe.te <OnEwiicityand no a.tr>
ENGINEERING DIVISION
- 90. Zhesnochkin.B. N.,"Analysisof Elastically Embedded Bars,"(in Russian)Stroutdat.
194s.
- 93. Zamnictmann. H., the Bernhamsqr des Eisenbahn Ober6 eases. (The Calculatwn o' Radway Super Structures) Bc Isn. Germany.1888. SEISMIC SOIL-STRUCTURE INTERACTION EFFECTS AT HowsoleT BAv PowEn PLANT i By Julio E. Valera,' M. ASCE H. Beisen Seed,8F. ASCE, C. F. Taal,*
, and Johns Lysener,* M. ASCE d One of the many controversial aspects of nuclear power plant design in the i past several years has been that of evaluating the seismic soil-structure interaction effects during design levels of earthquake shaking. Basically, two methods of approach are available for determining these effects: (1) Complete interaction analyses that attempt to make some evaluation of the variations in carthquake [ motions, both in the structure and in the soil in which it is embedded, and ! (2) inertial interaction analyses in which the motions in the soil surrounding ' the structure are considered to be some representative average motion having V the same characteristics at all points (10). The forener approach has usually been applied through the use of finite element methods of analysis while the latter, although it can be performed using finite element techniques, has usually I been associated with half space analyses of clastic or viscoelastic layered systems. It appears to be the prevailing opinion "that for near surface structures, good results can be obtained by a well performed analysis of either type. However, for embedded structures, the compicte interaction analysis approach comes closest to representing in a rational way all the important aspects of the problem" (1). The principal limitation of this approach at the present time is usually i considered to be the cost of the analysis and, in some cases, the less expensive inertial interaction approach may provide resuks of sufficient accuracy for practical purposes. However, as increasingly efficient and versatile computer programs are developed for finite element analyses, and as progressively more ! Nuec.-lhscussion open until March I.1978. To calend the closing dase one nionth. i a written request snuss be filed with the Edesor of Technical Putd.casions. ASCE. This p.eper is part of the copyrigheed Journal of the Geosechnical Engmeering thvision. ~ Proceedengs of the American Society of Civd Engineers. Vol.103. No. GTIO. October. 1977. Manuscript was submissed for review for possabic pubhcation on February 8.1977.
' Partner. Dames & Moore. San Francisco. Cahf.
i 8 Prof. of Civ. Engrg . Univ. of Cahforma Berkeley. Cahf.
' Grad. Research Asst.. Dept. of Civ. Engra. Univ. of Cahfornia. Berkeley. Cahf.
- Prof. of Civ. Engrg.. Umv. of Cahfornia. Berkeley. Cahl.
1143 1
'E 1146 OCTOBER 19n GT10 GTt3 SEISMIC EFFECTS 1:7 fackl(Storage Building) were 0.35 g and 0.26 g in the transverse and longitudinal a number of soil profiles involving clay strengths varying considerably in the directions respectively, making these the strongest earthquake motions to which upper 20 ft (6 m) as shown in soil profiles A, B, and C in Fig. 5. Resuks a nuclear power plant has so far been subjected. However there was no observable for all profiles investigated fell within the range represented by profiles A and damage to the facihty resulting from these mo* ions. B.
I A fortuitous aspect of the records obtained from the Humboldt Bay Plant I was the fact that the soil conditions at the plant site had been determined o . zs- :rs-rur- ,,,,,,, , , , ,
! by a comprehensive ficM investigation only approx 12 months before the [ ** yg",,
as ,,,,,,," carthquake occurred. In fact, extensive liquifaction and soil-structure interaction ,,,, , analyses using finite element procedures, with accompanying determinations ""*""*~'M
. , , , , , , , , %t% .
of soil characteristics at the site, had been carried out prior to the carthquake " wwas maarr of June 7,1975.ncsc studies were performed by Dames & Moore using analytical
"~
techmques developed at the University of California at Berkeley (5). In this ,, respect, it is interesting to note that these analyses had predicted a peak '"""* h tN
- acceleration at the base of the Refueling BuiWing of 0.13 g for a free-field }
ground surface acceleration of 0.25 g, while the subsequent earthquake produced an average peak acceleration at the base of the Refueling Buildcs, of 0.14 j,, , i s for an average free-field ground surface acceleration of 0.30 g. This resuk ,, alone, predicted in advance of the event and published in design reports, is of considerable interest. ,n _'.'".
* '""** *
- m,. = c an
- Though these facts are of major importance, perhaps the most significant feature of the June 7 event is the opportunity it provided to check the adequacy suo-of seismic design procedures against the known performance of a prototype structure under known field conditions of considerable intensity. De results ans -
of such an evaluation are presented in the following pages. 88*** w a=+=,'. aa'aa oon me in, ,.,.,. Snu Comemone ase Sot Paorasmas FML 3.-Seil8 Preslee at HuaiheMt Boy Power Plant Site (1 ft - SJeg g 1 peg
= 47.9 N/m )
A general description of the tubsurface soil conditions at the plant site has been presented by Valera and Brady. A crosssection through Unit 3 in the - g = north-south direction is shown in Fig. 2. Basically, the soils around the Refueling -
- r.cr: -
. 7.~
Building consist of approx 25 ft (8 m) of medium to stiff clay [ increasing to i approx 30 ft (9 m) at the Storage Building], underlain successively by approx 30 ft (9 m) of medium-dense to dense sand,10 ft (3 m) of very stiff clay l, l, and then a deep bed of dense sand containing some clay lenses extending to *
,, ,,, , ,,, , 9 ,, y j ,
a depth of approx 400 ft (I20 m). All of the soils surrounding the Refueling .-
-- ==---
~ '
Building are overconsolidated, with an average overconsglidation ratio of at least 6-8, indicating,that the coefficient of carth pressure at rest in the sands = wouW be on the order of one or more. %c soil profile and soil properties I used in the pre-earthquake soil-structure interaction studies are shown in Figs. 3(a) and 4. respectively. The soit profiles and soil properties used in the present f h" ' .~~ f'.. *L 4 . . study are shown in Figs. 3,4 and 5. He profile for the conditions adjacent 7 '. ./.' 4
~ " * * * * " " * *
- to the Refueling Building was identical to ahat used in the pre-earthquake analyses.
At the site of tim Storage Building itself, where the free-field records were RG. 4.-Averne* Dynemale SeN Properties
< obtained, there is suoc uncertainty about the actual strength of the top 30 l ft (9 m) of clay, because the closest boring is at least 100 feet (30 m) away and there is considerable scatter in the measured values of shear strength for De dynamic shear moduli and damping characteristics of the soils were undisturbed samples of clay taken from three borings surrounding the buildmg. determined by standard soil testing procedures using resonant column tests and his uncertainty is reflected by the ranges of strength values for these soils cyclic-triaxial tests on undisturbed samples reconsolidated under the M-situ 4 indicated in Fig. 3(b). To allow for tisis uncertainty, analyses were made for confining pressures. Dese are shown in Fig. 4. It is pertinent to note that 1
i g OCTOBER 1977 GTto GTIO GEISMIC EFFECTS 1149 these dynamic properties were determined and filed with the Nuclear Regulatory Commission before the carthquake of June 7.1975. At the time the studies developed in the free-fickt are analyzed by a deconvolution procedure for the were initiated (early 1973)it was not considered necessary to make determinations soil deposit alone to determine the motions which would have to be developed of fickt shear wave velocities since it was clear from preliminary studies that at a considerable depth below the ground surface [c.g.,150 ft-200 ft (46 m-61 4 m)] m order to produce the actual ground surface motions by transsnission of body waves (vertical shear waves) through the soil depossa. His can be "g $]M MMQ accomplished through the use of a computer program such as SHAKE (6). nese same base motions are then used to analyze the response of a finite
- _.-.___ g g Q - g"" g* - E _ [
- 7
.~,,_~'..,7
" ;7 .. -. - ~. element model of the soil-structure system and the results of this latter analysis harm N ' "" "" are checked by ensuring that the required freefield motions are indeed developed m
in the free field. He basic requirements of a suitable analysis and computer
-. .-- =* a - -- - ' *
- program (10) are that it should be capable of considering: (1) ne variation of ground motions with depth;(2) the three-dimensional nature of the problem; i
I b O)Ihe cffccts of adjacent structures on each ather,if appropriate;(4) the variation _ of soil characteristics with depth; and (5) the nonlinear stress-strain and energy-absorbing characteristics of the soil.
.. g .. g w. .
Raoutre or Pus-Easrn-aam Asantvan l _, ,,, . . De pre-earthquake studies performed by Dames and Moore were made using the computer programs SHAKE and LUSH (4). Analyses were carned out g pyg geit pressee et storage Building Used for Decanvahstion Studlee for cross sections in the north-south and cast-west directions (Fig.1) and for It ft - eJes m. j per - 47.9 N/m'l various levels of peak ground surface acceleration. ne soil properties shown in Figs. 3(a) and 4 together with the structure c ea""' - ; characteristics shown in Tables I. 2 and 3 were assisped to the finite element
. _ _ . , '._~ ~ ~ -,.- model. Damping values of 4% and 7% were used for the structures for analyses j I conduc:ed using peak ground surface accelerations of 0.25 g and 0.4 3.respec-tively.
f] l
' O "~ i O From the resuks of the initial studies it was found that the effects of the 7fi '
f j 7 i adjxent structures on the response of the bursed reactor caisson were relatively minor. Hus, the adjacent structures were not wated in the finite element j Ane -.-; l i ! ? nulet used for the later studies. Since transsritting boundernes are not inchsded l l in the computer program LUSH it was necessary to use an extensive mesh
-y- w- l in the horizontal direction to ensure that the computed response of the Reactor
.. .. . .. m..-- r Caisson and Refueling Building was not influenced by the Boundary conditions
- - ~~':: j I-
- only necessary to consider the response of the soil deposit to a depth of about h . one-half the structure width below the base of the structure; consequently,
_],,,,,,,,,,,,,,_ the base of the analytical model was taken at a depth of 150 ft (46 m) below
~ * ~ ~ ~ ~ ~ ~ the ground surface. '
FIG. g --Scenenestic Sm - tetien of Sotstrucsureinterm W W % Decesvolution Studies.-In performing a deconvolution analysis of a ground Element Reedel yudace MM&ctn6ne a cmeWn@ moMor use b a Wructure ,
'"'##**" * ** * *" ""#"*'I ' "I
- l N'9"*"CY shear moduli at moderate to large strains, such as can be determined by components of the ground surface motion in order to obtain meaningful resuhs.
Dere are two reasons for this requirement: str:.iacontrolled cyclic loading triaxial tests, were required for the analysis. Consmis hnunnenom Amatves Pesou" 1. He specified ground surface motion may contain hesh frequency coenpo. ne general procedure for snaking a complete interaction analysis (7)is shown nents which would not, in reahty, be developed for the site conditions under schematically in I^ig. 6. First, in which the known ground surface motions consideration.This is particularly true for sites consistir.g of deep [250-ft (76-m)] Iulics of soil or mcluding layers of soft to medium-staff clay and sand (7).
I 33.,o OCTOBER 1977 GT13 GT10 SEISMIC EFFECTS 1151
- 2. Deconvolution by a wave propagation analysis using equivalent-line:tr The spectra for this time history closely match the Nuclear Regulatory Comngission properties to represent the nonlinear stress-strain characte..stics of tne soil (NRC) design spectra stipulated in Regulatory Guide 1.60. In these studies, et was necessary to use a cutoff frequency of 15 Hz-20 Hz in order to ensure TABLE 1.-Structurni n:' _.2 of Reacter Caiseen t ik accebbs at M M M hme excesshe.
Acceleration time histories computed at various depths withm. the free-field soil profile are also shown in Fig. 7. It may be seen that there is both a decrease Depth Shear modulus, in below ground pounds per square Density in pounds ,. surface. In feet foot x 10' per cabic foot Poisson,a ratio I g1) (2) (3) I4I **a** **'= a8 l 3 l 78-78 33 0 a2 -= 78-87 4IN ' "
. .m = nm m. m. . . , ,.
Note: I ft = 0.305 m; I psf - 47.9 N/m'; I pcf = 16.0 kg/m'. "*'"**** _ - .t _ . _ m a sta urust.3 Depth below ground surface, in feet Weight of mass. in kips ?
'i' 25 i
76 k-* W My $ 37 44 .,
** * * * " " ' ' ' * " " * * "" aa 31 43 .g*= =a . . .
57 47 78 54 Nose: I ft = 0.305 m; I kap = 453.6 kg. as ILE Mr8D't 'se TABLE 3.-Structural ? , A of Refueling Bulkling ** l m hlh I 5 k ARL AM taltA. , m Depth Sheer modulus,in ! above ground pounds per aquero Density. in pounds .. surface, in feet foot x 10* per cubic foot Poisson's ratio (1) 12) (3) (4) < n as, n .m , , , , , , , , , , , , 0-17.5 5 23 0.2 10 0.2 17.5-35 5 , FR 7 -liertserreal a%etion h w g % g % Nose: I ft = 0.305 m; I psf = 47.9 N/ma ; I pcf = 36.0 kg/m'. h for Free Field Candlesene (s. = g.25 g)in n-A-" 4 in the amplitude of the motion and an increase in the frequency content, with inevitably leads to an excessive amplification, with depth, of high frequency an increase in depth within the profile. motions. SeN-Serveture Interacties Analyses.-Using the base motions computed at a depth of 150 ft (46 m)in the deconvolution studies, analyses were then made, in the pre-earthquake deconvolution analyses, the acceleration time history using the program LUSH and a suitably fine but extensive mesh, to compute shown at the top of Fig. 7 was used as the free-ficId ground surface moten. the response of the soil-structure system. Computations were made for a variety
s 1152 OCTOBE21977 GT10 GT10 SEISMIC EFFECTS 1153 of soit properties, and envelope spectra for motions at various levels within and recorded peak accelerations at instrument locations in the structure. Such the structures were finally selected for design, based on the range of computed results supplemented by engineering F.ulgment. a comparison is shown in Table 4. It may be seen that the values show a remarkably high degree of agreement although there is some indication that TABLE 4.-e-- ' - of flocorded and Computed Acceleratione the actual stiffness of the structure was somewhat less than that used in the
"--==a= 1 an
- - == o
, *' *" o* .
, , *l ,** 9 or o,s o.
Maximum Maximum Accelerations for Recorded Motione acceleratione 85- - e,- [ _ go, ,,,g Location Elevation Longitudinat "- Transverse motione , !< x- ); '""**""' - (1) (2) (3) (4) (5) ****'"** a n- - zn-
- Free.feeld IStorage I tw IJing) +12 0 35 g 0.26 g 04g 0.25 g *- fI u- f -
Refueling twilJing + t2 0.25 g 0.20 g 0.21 g 015 g Reactor caiswa -66 0 86 g 0.12 g 0 22 g 0. 8 ) g "- f g eZ - el- ,g,- no-
*--m _._,,,,
9 n - - g ~es= -.s.,, I p- a n ,,. . . I re FIG. 9.-A,,.a.,etion Distributions Cm _12 by r _ ; . ^' _ _.. of Hecorded Surface
. -r heetione (1 ft = 0.306 ml ,
I
-o s .
f '2"""".^, .i ?.. ., 7 ~ , ,,' . I - . . _ e , _ , a- I s m - j - n .- m-I %,
*o f
F-i" - P-- F ! a e- ' i i _.o
}=-O l--@
I, as- -
-m*
m- - __g nn-I
. -aco a. . * *5" i*"*
l R(L 10.-Ellect of Seit Preste on Verteesen of t .h a n.,seien with Depese I Crg_12 by Deconvehstion of Ground Surface Motion flooerde 11 ft = 0.385 ml analysis. Nevertheless, the good agreement in these values is an encouraging 1 aspect of the analytical procedure uwd in the studies. RG. g.-Finite Elemem Meeh Used for Analysis (1 ft = t.385 ml " in the course of these s'udies, analyses were made for ground surface motions Pbst-carthquake studies of soil-structure interaction effects were performed having peak accelerations of OA e and 0.25 g. Since these are in the range following the same basic procedure as that described in the aforementioned, of peak accelerations developed in the transverse and longitudinal directions but using the computer programs SHAKE and FLUSH H). since the latter during the June 7 carthquake, it is of interest to compare the values of computed provides a more versatile capabdity than LUSH and is also more economical. Advantage was taken of the results obtained in the earlier studies, and the
, I l j 11 *#. OCTOBEft 1977 GTit, GT10
- SEISMIC EFFECTS 1156 I
effects of the adjacent structures were therefor 2 neglected in the analyses. high frequencies ts might be expected for a deep soil condition such as that Because the program FLUSH uses transmitting boundaries. it was only necessary at the Humboldt Bay Plant site. to use the finite element mesh shown in Fig. 8 for the sod-structure interaction 2. The results of soil-structure interaction analyses made with a cut-off analyses. frequency of 12.5 Hz will be comparable to those made using lusher cut-off I Decemvekstion Studlee.-As stated previously, there are valid reasons why frequencies. Since there is a marked reduction in computer costs associated ! some fihering of a given ground surface motion is required in performing a with the use of a lower cut-off frequency, the soil-structure interaction studies deconvolution analysis to determine motions at various depths. To determine described in the following section were made for these conditions.
' the sigmficance of such effects for the recorded motions at the Humboldt flay site, deconvolution analyses were made for Soil Profile A in Fig. 5 at the Storage SeN-Structure Imeeraction Sandles.-Having determined the base motions re-Building site and for the recorded surface motions, using filtering or cut-off quired in the soil profile at a depth of 150 ft (46 m) to produce the recorded frequencies of 20 Hz.15 Hz, and 12.5 Hz. The resuks of these studies, in motions at the ground surface under free-field conditions, the same motions j actms of the computed variation of maximum acceleration with depth in the i soil profile, are shown in Fig. 9. It may be seen that the cut-off frequency. , , ,
- c..
.[ .... kme=. .
- =
.....=;- 1..
\
l l g n
, : I. ; . _ . , . . ..
l ; l.w.. -
, .1 ai s
.f i ; :. ,,,e a . -
,% i i . ee. - .
p.~-Ec
,,,___, 1. . . m - P-- -- -
. y -
I.. .
.. y _
4.i, , l-= . - __ 4, . ., .
' 3..
L.... . m.y*.3 1 4. .
; t_4 2.. .: s:.i==- T = --
jj=-..._.p i sss-
+ - -
j _. A... ( m
- t. ..m.s _.
~
m 1..~ . m . .. N e<f
. w ,,, ,, ,,A 1 s. . - . .
- . . - .- _ .- - i..
s.
~'; . . . 1..
FIG.11. en; ' : of Recorded and e- ; _nf Specsra in RefueNng Su5 ding-E*Y FIG.12. *- ri of Spectra for #- ,if Motions in Free-Fised and at Bene of Stnscture-Mealmem Groun.4 Seeface Accelerosien - e.3 g (1 ft - eJes m) within the ranse investigated, had little m. fluence on the results of the analysis, all of the studies for both the longitudinal and transverse recorded motions showing a marked decrease in magnitu& of the peak acceleration from the were uscJ as excitation at the base of the soil-structure model shown in Fig.
) ground surface to a depth of approx 30 ft (9 m) and below. In fact, the peak 3 to compute the motions developed: (I) At the base of the structure; and I
accelerations computed to develop in the free-field at the level of the base (2) in the structure at the level of the ground surface, where motions were 3 of the Refueling Building [ approx 85 ft (26 m)] is in the range of 0.10 3-0.14 recorded during the urthquake of Jun( 7. Separate analyses were made for i g, or less than 60% of the maximum acceleration at the ground surface. the longitudinal and transverse records of free. field motion and for the various It may also be seen, from Fig.10. that generdy similar results are obtained soit profiles. The ranges of analytical resuks are shown in Fig.11 in the form t whether Soil Profile A or B is used for the analysis. Akhough they are not of response spectra, and are also compared with the spectra for the recorded shown, resuks for Soil Frofile C fell within the "age shown for Soil Profiles motions. A and B. Thus, it would seem reasonable to condude from these results that: It may be seen that for both lungitudinal and transverse motions, the recordeo motions at the base of the structur : are in reasonably good agreement with I. The recorded ground surface motions have no significant content of very those computed using the finite element procedure for implementation of an
r s GT10 q GT13 SEtSMIC EFFECTS 1157 ng OCTO8ER 1977 practice. At the present time, regidatory requirennents for determining soil-struc-ideahzed" complete interaction vaalysis. For both compos ents of motion. the 8" . interaction effects for embedded sanactures such as the Refueling Besidens analysis procedure indicates a higher peak in the response spectrum at a frequency require the specification of a design or control motion at the ground surface of appros 3 Hz than actually developed but considered overall, the agreement having a designated maximum acceleration and a tinue history whose spectruen between computed and recorded base motion spectra is both patifying and closely matches a standard design spectrum shape specified by the NRC. Gare encourasses. the avermee peak acceleration recorded in the free-field at the Hanboldt Bay Similarly. the recorded motions in the sanacture at sround level fall essentially _ plant was 0.3 3. at would seem reasonable to compare the motaons recorded within the ranse cosaputed by the interaction analysis procedure, providing further at the base of the Refueling Building with those computed following an approved confirmation of the ability of a complete interaction analysas to compute the *le5isa procedure consistent with a peak free-field around surface acc.a ration t strucs.aral response with an adequate e"egree of accuracy in this case. of 0.3 g and the standard design spectrum shape. 'Ihis is, in fact, the motsoet It is recognized, of course. that one such test of the C=MIity of ar.y whose spectral shape as shown in the upper left corner of Fig.12. An acceleration i analytical procedure does not necessasily provide proof tlut it wiB always lead r - - .. -.
.s lh:n .. -
1 . l u i
,.. n!
I .) [- x.h~ i & N a f l A _.- ,,_, % l. i :-
.n ,
! / '* d .' I-
*?(i;-a i a ., ,,,,,,,,,,,,
4l l-(C E "' m s a-2 e _.p u r_ I* : i..~ l 7 marw . . .. 2:27. '~ '." / t a ..
, , ~ ~ '
!.=. .:.7.*:=l" m e - ,
,l - = ~
. i ,: . . ;
ee ~e s
/ es- [**
"::gr.:.za I T.
]=. *
/ N .. ^
,~ g ea g r8 e , , }. .
- 2. '-~~. =w="
.e . . .
e, a > _=
.= 5 . - . .==+ .= ..
f . .. N M- ~ - : , of spesera for e--e a sneesen se yees Fluid and at anse J pts.13. c_ ' _ et spesere for Design and flesseded AAmelone (1 ft - eJe5 *f semeeum tseine senc os len pteeedume 11 ft - e.3e5 nd d sul r time history having this spectrum and having a duration of appcom 16 see was so good evahastions of field performance. Nevertleeless, in the current absence f' of any other opportunity to check analytical methods for computing response used en the following analyses. isador strong shaking of protosype structures, the results obtaGaed in evi ,a this Regulatory practice permits the deconvolution of this motion and the analysis ) of soil-structure interaction effects using finite element methods as previously
- siegle case can give designers increased confidence in the usefulness of the described but it also requires that
- (I) Analyses be made for the most likely j analytical tools at their disposal.
values of soil moduli and for values of soil modub which are increased and A"*8caen8'v e' IdRC Desas Wa== reduced by a factor of I.5 to allow for possible uncertainties in soil property deseminations;(2) the envelope of the resuhing spectra for motions cosnputed I . In addition to their function as a check of the adequacy of procedures for for a point in the free-field at the level of the base of the structiare be not
! enelysing soil-structure interaction, the records obtained at the Humboldt Bay less than 609E, of the spectral accelerations for the ground surface control snotion; Power Plant are also useful in investigating the adequacy of required design l;
W ins OCTOBER 1:77 GT13 GT10 SEISM.C EFFECTS 1159 and O} the structural response be evaluated for motions having a spectral shape the spectrum for the recorded motions come close to that for the computed enveloping those computed at the base of the structure for free.ficid notions base motion envelope spectrum. meeting she requirements of (1) and (2). One means of increasing the free.ficId spectra to meet the 60% of surface A typecal set of calculations for the same ground surface control motion, control motion requirement is to increase the ground surface acceleration for but for the three different values of soil moduli, are shown in Fig.12. In the control motion for one or more of the analyses, so that after deconvolution this figure, the control motion is shown in the upper left-hand corner, the spectra it meets the free. field requirements. In the present case, this could be achieved for the computed motions in the free. field at the level of the base of the structure by increasing the control motion for the analysis performed using values of are shown in the lower left-hand corner and the spectra for the computed motions soil moduli reduced by 30% With a satisfactory degree of accuracy, this leads at the base of the structure are shown on the lower right.bnd corner. For to corresponding increases of 30% in both the free-field spectrum at a depth the anrJysis conducted with the most likely values of soil moduli and the reduced of 85 ft (26 m) and the spectrum for motions at the base of the structure. De superimposed spectra for the three analyses with this modification are is - shown in Fig.14 and the envelope of the spectra for computed motions at
,~ a, n QOM the base of the structure is compared with the spectra for the rnotions recorded at the base of the structure in Fig.15. It may be seen from Fig.14 shot the
? '",,,,,
* -f '"""
envelope of free-field spectra now comes very close to meeting the design om ip spectral requ:rements at this location; thus the envelope of spectra for motsons f'4 developed at the base of the structure as shown in Fig.15 would be esacatially l Dcm acceptable foe design purposes. His envelope provides a comfortable margin of safety above the spectra for the recorded base motions and would seem J w
/f ;,_- - --
o g/ . . . . to indicate that, at least for these strong motion records, the current design e' 2, requirements provide an adequate Lut not excessively conservative margin of f 7 i y l h-- safety for analyses conducted in the manner described in the aforementioned. m.--==a== Similar studies for other methods of evaluating soil-structure interaction effects
* %** am we--. would presumably throw some light on the degree of conservatism they introduce 3
U"O into the design procedure. m e. -a t I am-
"* He preceding pages present the results of a study of the distribution of
__-_6 tea * ";" " *" *"" ground motions and structnral sesponse in the Humboldt Bay Nuclear Power
. M --ll=-- Plant during the Ferndale carthquake of June 7,1975. Based on a knowledge E - of the motions recorded at ue ground surface in the free-field, computations
*"~~ are made using an idealized complete interaction procedure based on finite element analysis, to determine the characteristics of the motions likely to develop M~A{3 .-.r.
fu. at the base of the buried reactor caisson at a depth of 85 ft (26 m) below u * ! * * " *
- the ground surface and within the Refueling Building at the ground surface level. De computed motions are shown to be in reasonably good agre+ ment
- Fio.1s.-.t;t et spectre for Design and Reeerded testione et Sese of with those recorded at these locations during the same earthquake. In addition, the recorded motions are compared with sluse computed by an analysis procedure i
Structure H ft - 3J35 mi which generally meets existing regulatory requirements, and it is shown that the regulatory reqidrements lead to an adequate but not excessively conservative soil moduli, the control modion was filtered at 10 Hz, while for the analysis margin of safety based on the motions recorded in this event. with increased soil moduli, the control motion was filtered at 20 Hz. He envelope It is of interest to note that lambe has recently made a study of the accuracy of the computed spectra for the motions at the base of the structure is compared of engineering predictions of soil behavior under static-loading conditions (3). ' with the motions recorded at the base of the structure m Fig.13. For this purpose he classified predictions into five groups as follows: i la may be seen that although the free. field motions fail to meet the NRC design spectral acceleration requirements in the frequency range from approx 1. Type A: Prediction made before the event. ' 2 Hz-5 Hz, the envelope spectrum for the computed motions at the base of 2. Type B: Prediction made during the event but before the results are known. i the structure is nevertheless higher than the spectra for the recorded base motions 3. Type B1: Prediction made during the event but with results krown at at all frequencies. In fact, only at frequencies of about 4.5 Hz-5.5 Hz does the time.
i l 1100 OCTOBER 1977 GT13 GT13 SEISMIC EFFECTS 13;t
- 4. Type C Prediction made after the event but before the resuks are known. Embedded Structures During Earthquakes" sponsored by the Nasiraal Science
- 5. Type CI: Prediction made after the event but with results known at the Foundation. 'Ihe support of the foundation and of the Pacific Gas and Electric
- time. Company in providing the besic data required for the study and in encouraging i the investigation is gratefully acknowledged
! He concluded that "Type C predictions are autopsees. Our professional literature l contains the resules of more Type Cl predictions than any other type. Autopsies Apaanseg -Marusmune j can of course be very helpful in contributing to our knowledge. However, one , snust be suspicious when an author uses a Type Cl prediction to prove that i. An.lyses go, Sois-Seructwe Interaction Effects for Nuclear Power plants." Proyess i any prediction 9chnique is correct". Lambe also concluded that predacted resuks Report of the Ad Hoc Comniitsee on Sod-Structwe Inseraction. Nuclear Seructwes within a factor of two of observed field performance constitute very good and Maserials Comaussee of the Structwal Division of ASCE. haat M. Idriss. th. . . predictions. It would seem optamistic to expect any better success in predicting y'ya'c a8 Nc 8 y i S Sec ASC onference on Saractistal Desagn dynamic behavior of soel or soil-structure systems. , 2. Hwang. R.. "Seisneic Response of Embedded Strucawes." thesis presented so stie However. the prediction of the base motion peak accelerations shown in Univeisity of Calefornse at Berheicy. Berkeley. Cahf.. in IW4. in partial fuitelinient Table 4. based on the assuanpaion that the ground surface niotions with peak of the regewesents for the dryee of Doctor of ph.a phy. j accelerations of 0.25 g and 0.40 g in the free field was clearly a Class A prediction J. Iambe. T. W.. "Ptedictions in Soil Enesseering." Thirteeneh Rankaug 14ctuse.
,,",,," V No 3M g4 l essing Lambe's tw - "_ J in that the study describing these results was carned 4, g C l
oest before the event of June 7.1975 had occurred; nevertheless, the degree of Sosi-Structwe Synsems." Report N. EENC 74 4. Earthquake F - genearch
, of samdarity between peak acceleration values assusned and developed m the Center. Universsey of Cahiernia at Berkeley. Berkeley.Cahf.. Apr!. iW4 l free ficht and those predicted and developed at the base of the Reactor Caisson 5. Lysmer. J.. et ad.. "FLUSH-A Compasser Propane for Approminiese 3.D Analysis
' o Semcture latenctson Problems." Report %. EERC 75-30. Earehead* Response would scene to sinow that the prediction was highly satisfactory. Similarly, although the more detailed analyses described in the preceding pages N
- er eley. er e ey. Cahl.
using the amate general procedure were neede after the event. it snight reasonably 6. Schnabel. P. B.. Lynnier. J., and Seed. H. B. "SH AKE: A Comp. ster Proyens be clainied that they represent a Class A predoction since liney permeisted virtually for Earthquake Response Analysis of Horiaanta5y Layered Sises." Jtsport N. EERC i neo latitude for .
* -ion of the resuks. in that they were based on: Amarch Center. Uneverssey of Cahternia at Berkeley. -
} 7. Seed. H. B.. Lysmer. J.. and Hwaag. R., "Sail-Sensesme Inseraction Analyses for i !. A methnd of analyse.s developed price to the event. Evalumsing Seismic Response." Report %. EERC 744. Earthquake Engsq
- 2. Soil properties established and filed with the Nuclear Regulatory Commission Research Censer. Universesy of Cahfornia as Berkeley. Berkeley. Calif., Apr.. IM4.
' g price to the event. 8 S**d .H. B.. Lynnier. J.. and Hwang. R., "So.3 Seructure inseraction Analyses for
- 3. Fined aspsface anot*was established by the event. **."No.
101 GT3. Proc. Paper iIJit May IM5 pp 4ML4<7"' """'* " A"'"*f of
- 9. Seed. H. B.. Uses. C. and Lysaner. J., "Siec Dependem Spectra for ET ' Resis-Nevertheless, alte ithers would be the first to agree that the good agreenient tant Desipi." Report No. EERC 74-!I. Earthquake Enemeering Research Cemeer*
i in t!ais one case between preactaf and developed mossoses at the base of the University of Cahfornia se Berkeley. Parkeley. Caist., Nov.. IM4. I structure does not necessarily prove the =dap==cy. for au cases, of the nicthod 10. Seed. H. B.. Wheenian. R. V.. and Lysmer. J.. "Sost-Senacture Interaction Effects ) of analysis used. Clearly, cosnpensating errors neight be involved whose effects have not been fidy appreciated. On the other hand, it is an encennag:nig start [w all.. Oct., g,73. het delr Structwal Geos af } and the results obtasned obviously give sonne degree of justifLstion for the i ancthod uced 'Ibey neight also induc course.given an equaldegree of justification i for other methods which neight be used for analyzing soil-structure interaction l effects. 'these are nipiificant facts in a field in which no other data esists I by wenich the adequacy of analytical procedures can be checked. At the same , j tense, it is clear that any method of analysis wisich provides r. poor prediction of the resuks obtassied. baand on the known values of soil and structural properties s j and the neotions recorded at the groissed surface, mu t be considered of dubeous i vahdity for future predictions of probable buildeng response d I .
! h-i
! The study described in the preceding pages was conducted as part of an _ e
~
) investigation on "Analysis of Soil-Sanscture lateraction Effects for Massive l _ _ _ _ _ _ _ _ _ _ _ _ _. _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ - - - - - - _ _ - - . _ _ _ _ - - _ _ _ _ - _ _ -}}