ML20212L979
| ML20212L979 | |
| Person / Time | |
|---|---|
| Site: | 05000000, Diablo Canyon |
| Issue date: | 05/30/1978 |
| From: | Luco J CALIFORNIA, UNIV. OF, SAN DIEGO, CA, Advisory Committee on Reactor Safeguards |
| To: | Advisory Committee on Reactor Safeguards |
| Shared Package | |
| ML20150F500 | List:
|
| References | |
| FOIA-86-391 NUDOCS 8608250323 | |
| Download: ML20212L979 (19) | |
Text
.', !
^
{
{
~
4.
\\
REVIE110F Tile 'SEIS!!IC EVALUATION TOR POSTULATEI) 7.571 !!OSCRI EARTl! QUAKE, UNITS 1 AND 2, DIADLO CANYON SITE' by
~}
J. Enrique Luco M
A Report to the Advir.ory Cocaittcc on Reactor Safeguards U. S. Huc1 car Regylatory Commission.
30 Blay 19 78 86062503p PDR FOIA 0001 HOUCHS6-391 PDR
. +-i te!
4
{
REVIEtt Af1D REC 0ftlENDATIONS Af ter detailed review of the report 'Scismic Evaluation ic-Postulated 7.5;.1 ilosgri Earthquake' (Ref.1), I have the followi t comments and recommendations:
1.
Frec-Field Des i en Smetrum.
In my opinion, the f rc: -
field design spectrum used for re-evaluation of the Diablo Cam Nuc1 car Power Plant docs not reflect the strong motion at the r-for a 7.5 magnitude. carthquake at' an epicentral distance of 5 kiloneters, but rather the motion for a 6.511 earthquake at thni distance. The free-field design spectrum developed by Newnarh and ailopted by f1RC corresponds to a simplified version of the average of the tuo Pacoima Dam spectra recorded during the 6.5:'
San Fernando earthquake with the high-frequency portion reduced
~
by use of an ' effective' penh acceleration (Fig. 1).
The Blu:r design spectrum devcicped for the applicant closely folleur the M
Newmarh, spectrum.
The peak acceleration, velocity and displace I
ment controlling the high, intermediate and low frequency porti of the ficwmark design spectrum are in agreement with the averag (50t, percentile) peak valdes obtained by Trifunac (Ref. 2) for 6.51! carthquake while falling short by 40 to 60 percent from th-corresponding values for a 7.5}t carthquake (Tabic 1).
The peak values consistent with the IJeemark spectrun are also consideral' louer than those suggested in USGS circular 672 (Ref. 3) as r.hu in Tabic 1.
In addition, comparison of the Neumark and Blune J-sign spectra with estimates of the average response spectrum f: -
1.
A y
a g
e e
,--.--,,.m-.,
5
.h
(
(
a 7.St! carthquake as obtained by Trifunac (Ref. 4) also r. hows-differences of the order of 30 to 50 percent (Fig. 2).
The applicant has indicated that the thrust fault mechanir:-
and the location of the pacoima Dam instrument in the San Fernando carthquake may have increased the recorded peak accci-erstion. These possihic effect.s are negligibic in view of the fact th.at the standard deviatica for peak accelerations, which has not -been considered, corresponds to a factor of :.
- Also, t'
records for the Ms=-7.2 Ca:11, Russia carthquake of 1976 indicat peak horizontal acceleration of 0.Sg at an epicentral distance 10 kiJomuters. Correcting for attenuation using the Gutenber; '
e relation 1 cads to a peak acceleration of 1.03 at 5 kilomercre the Gazli carthquake in general agreement with the results of Trifunac sitd the USGS recommendation (Tabic 1).
In view of these facts,.I must conclude that the Neu:.iark.'
g Blume design spectra do not correspond to the ground motion for 7.5!! carthquake at an epicentrai distance of 5 kilometers.
I p-pore that the estimate of the arcrage respense spectrum fer M"7 S kilometers, epicentral distance and rock sites of Trifunac (r.
- 4) be used as design spectrum. This spectrum is consis tent wi t '
the only records availabic for large magnitude and short epicen-tral distances (San Fernando, Koyna and Ga:li) as wc11 ar. with -
USCS circular 672 recommendariens.
2.
'Ef fective' Peak Acceleration.
A judgmental factor h:-
been used to reduce the 1.15g peak acceleration recommended in USGS circular 672 to a value of 0.75g.
This ill-defined factor 2.
.i
~.
g%
.........,s...n~...
~
O c
(~
},,
s i
has been' used in the past to account for discrepancies on the icyc1 er damage observed as compared with the predictions of ordinary scismic analyses which do not account for the effects soil-structure interaction, are based on nominal values for damt ing and strength, assume linear behaviour and do not include tb energy dissipation in partitions and other non-structural cle acn ts. This catch-all reduction factor has no place in ti.c de-sign of carefully analyzed structures such as those in nucicar power pl ants.
Factors which may reduce the response or the J.
of damage should be identified and properly included in the s::
tural models.
In the case of Diablo Canyon, many of these fac*
have already bec'n incorporated in the analysis:
use of test strength rather than nominal values, use of higher than commen danping values, reduction by scattering of waves by large feu,J tions and possible inclusion of ductility.
The arbitrary redr-of the high-frequency components of motion affects the respene piping and equipment.
I recommend the climinati'on of this red:-
t' ion of the input motion.
On the Ef fect of Scattering of h'nves by Rigid Found.it 3.
The high-frequency components of the free-field motion have b:
reduced by the so-called tau filtering procedure to account by scattering of waves by the supposedly rigid foundations. Thi:
correction amounts to a reduction of the !!ewmark free-fic]d de.
spectrum by 20 to 30 percent for frequencies higher than 2 cps.
Slightly lower reductions have been used in the Blume's spectro-
. The correction for foundation scattering effects is based on th-e 3.
\\ e 6
.3.-
gy
--~~~~p is::
Ni
(
c e.-
assunption of a rigid foundation and hori:ontally propagating .:
waves. Although the assumption of a rigid foundation nar be-re.
sonabic, it must be recognized that deviations from the assumpt-Icad to localized higher stresses in the lower portions of the different structurcs. The assumptien of horizontally incident waves is highly questionabic considering that the epicentral di-tance is couparabic uith the focal depth.
Under these conditic:
the possibility of nearly vertically incident unves may not be ruled out.
For vertically incident waves the scattering by the foundations is practically nonexistent given the shallow enbcd ment.
Assuming for the sake of the argument that the scisnic exc tation at the site corresponds to horizontally incident Sil are I find that the reductions proposed by Newmark and Blume are.ic.
high when compared with analytical solutions.
For horizontallr incident 511 waves the reduction of the translational component-g of motion is coupled with the p.xistence of a marked torsional input to the structure (for' details refer to the attached paper-Th'c applicant has included ' accidental' eccentricitics of 5 and
,,g percent to represent these torsional effects. The us e o f an eccentricity of 5 percent corresponds to the use of a peak tor-sional acceleration at the base of the containment of the order of 0.025 rad /sec2 as may be inferred fron Table.1-5 of Ref. 1.
This torsional acceleration correspends to a tangential acceler ation at the base of the containment exterior.of 0.025 x 70/32=
0.0Sg.
The results of Itay and Jhaveri of IIRS/111ume presented in 4.
e g
.~~,y I.
E
o
(
(^
t Fig. 56 of Appendix D39A, but not used in the analysis, shee tha*
' corre a peak torsional acceleration of the order of 0.1 rad /sec-ponding to a peak tangential acceleration at the base of the con tainment exterior of 0.2g would be more appropriate.
It may be concluded that the use of a 5 percent cccentricity underestinatt the torsional input by a factor of four. 'This observation is c" sistent with the original work of Neumark (Ref. 5) which indicat that an eccentricity of the order of 25 percent would be neces r-the torsional effects induced by hori:ontally prep-to represent It must be mentioned that the increase in pen' goting 511 waves.
acceleration of 0.2g based on a more realistic estimate of the torsional input more than compensates for the reduction by tau-filtering from 0.75g to 0.67g for the containment exterior.
From the point of view of the analysis of thc. structural r.
sponse, it does not seem adequate to introduce the torsion.'1 2"
M through the use of ' accidental' cccentricitics.
Such procedur.'
which leads to the coupling of the torsional and transla tion il sponse in essentially symmetric structures distorts the rerrce and the natural frequencios of the sys tem.
The eficcts of the sional input may be significant for the turbine building in whi the possibility of portions of the structurc undergoing inelast deformatio.ns may increase the eccentricity.
Tf it is shown that the scire.ic excitation at the site cor the reductions responds mainly to horizontally incident waves, the translational and torsional response should be evaluated or To the basis of the more exact methods presently availabic.
5.
~
i
(
~
e include an exaggerated reduction of the translational motion..
out incorporating the full torsional effects is improper.
- 4.. Soil-Structure Interaction.
In Appendix D-LL3A of l'ei the applicant presents a conparison of the results obtained by fixed base analysis of the axisymmetric containment model with filtered spectra as input (F.S.Axisyn.) with those obtained r:
soil-structure interaction finite element model with the Ncte.:
free-field motion (without tau-filtering) used as surface cont.
mot ien (Pl.USil-SSI).
Based on the results shown in Fig. 3A-1 et Appendix D-LL3A, the applicant concluded that ' the use o f tau-filtered inputs with fixed base models as used for scismic ans:
of Diablo Canyon structures is censervative. '
This, comparison not valid, and the conclusion is not warranted by the anasysis For a valid ccmparison, we must requitt that the fixed base n '
- ymmet ric analysis and the fixed base PLUSil analysis give est -
M the same response everyuhcre except at high frequencies wherc :
fixed base PLUS!! results no ?' icluding the tau-filtering shoul '
slightly higher.
This is not jhc case as shown in Fig. 3 of t!-
report obtained from results. shown in Figs. 3A-1 and 3B-5 of Appendices D-LL3A and D-LL3B.
Since the fixed base PLUS!! modM is inconsistent with the fixed base axisymmetric model, no vali conclusion as to the effects of soil-structure interaction can obtained by comparisons of the type shown in Fig. 3A-1.
It nur be mentioned that it has been shown that two-dimensional models such as PLUSil may underestimate the rer.ponse at the top of the structure by 30 to 50 percent.
I l
6 c.
l l.
l n
l..
[
m
{-
f i
i In Appen lix D-LL3B, comparisens are presented of the respon:
for a fixed base and an SSI model both computed usinC PLU511 an'd the Newmark free-field spectrum (uithout t au-filte ring) as.contrs motion on the free-surface. Assunig that the results presented arc internally consistent, it is possibic to drau some tentative conclusions.
Fig. 3D-2 of Appendix D-LL3B indicates that the p accelerations on the containment exterior obtained including the SSI effects are approxic:ately 10 percent lower than those obtain on a rigid bene.
Since the SSI tesults automatically include th-cffects of scattering of waves by the foundation as well ::s the effects of radiation damping into the soil, it may be concluded that the reduction of 20 percent (0.75g to 0.6g) by tau-effect propcsed by Neumark and a 3,imilat reduction used by Blume are nc-conservative.
Figs. 3B-3 and 3B-4 of the same Appendix indicate that the s tory shear forces and overturning moments on the cent.'
ment exterior obtained including the SS1 arc equal or slightly higher than those obtained for the rigid base PLUS!! model.
In this case, any reduction of the fixed base results by tau-filter would underestimate the stresses in the structure.
Assuming that the PLUS!! results are correct and consistent, it may be concluded that the tau reduction proposed by Newnark a, 10ume everestimates the. reduction effects of wave scattering and soil-structure interaction for vertically incident shear waves.
In particular, the stresses conputed on the basis of spect ra re-duced by tau-filtering would underestimate the stresses that re-sult from the SSI P!.USil analysis by at 1 cast 20 percent.
7.
e L.
A t
=
=my-
=
a
,= q m 7
s*w._.m
=
(
(.
The applicant has indicated that -the shear wave veloci ti-the site execeds 3600 ft/sec. The lou-strain and iterated (or strain dependent) shcar uaves veloci tics used in the PLUSil SSI model are not reported.
I requ.:s t that this information be r.a.'
availabic.
In Appendix DLL-15-(Anendnent 53), a uni forn sheai wave velocity of 3500 ft/sec. is used.
I recon mend that, the tau-filtering approach be climina ted and that a complete three-dimensional soil-structure analysir vet tical and hori:ontally incident Si! waves be undertaken Th-npproach will have the advantage of producing rc.11is tic cs tin.:
of the wave scatterinC and torsional effects.
s The peak. spectral respenne' for the PLUS!! fixed base ansi-occurs at a frenuency of 5.3 cps chile the corresponding f rc :
for the axisynometric fixed base analysis is 1.5 cps, indicati wgN!ayy difference of 18 percent.
If this difference reficcts the a:c acy with which the fixed base fundamental frequency can be cer.,
ted, then it would seem that 'the peak widening of the fioer re-spons,c spectra of 5 pqrcent on the high frequency side may be sufficient.
The PLUSil SSk resonant frequency is IS peretut is
~
than the PLusil fixed base frequency. This again seems to in?ir that the.15 percent peak uidening of floor response spectra c:-
loe frequency side is not sufficient.
S.
Scisnic Itisk Analyses.
Several scisnic risk analyses the Diablo Canyon site have been performed.
The estimates obt:
for the probability of excedance of the Itos gri design spectrun
. fer by two orders of magnitude. The applicant (Appendix D-LL 1 8.
s 6
e i-
,h h
c:
(
1.
s estinates tha t the probability of exceeding an 'cffective' $c.
eration of 0.75g in 50 years is 0.1 percent.
Anderson and yt:
(Ref. 5) es timate that the probability of exceeding the high-frequency portion of the Ilosgri design spectrum in 50 years v.!
from 10 to 20 percent, depending on the scismicity model conr.
The difference co'rresponding to a factor of 100 to 200 can bt lyzed by considering the following factors:
(i)
The applicant considers the probability of c.xceda-of an 'cffective' acceleration of 0.75g while An.h son and Trifunac use as a basis of reference the 0.75g llosgri design spectrum.
The use by the ap-plica'nt of on ' c f fc e t iv e ' rather than 'instruacnt:.
acceleration of 0.75g reduces the probability cf e-
~
cedance by a factor of four.
(ii) The use of Blume's SAM-IV and SAM-V attenuatica U$ENEN r-lations as opposed to the use of the Trifunac's te lations leads to rc. duction of the probabilit;. of excedance by a factor of ten.
(iii)
The rest of th,c differences corresponding to a fac-tor of 2.5-4 can be attributed t.o the different scismicity models considered.
llaving isolated the causes of the discrepancies in rish cs:
nation, I will discuss them in detail.
I have indicated that t'
reduction of the peak acceleration to an 'cffective' icyc1 sheu' not be used in the analysis of nucicar power plants.
For the p-
, pose of es timating the risk of exceeding the !!os gri design spec:
9.
.m s
.........--------- --u-A
(
(
the anchor acceleration of 0,75 shettid be treated as actual 1-acceleration.
In this case, the probability of excedance in F-years as obtained by Blume's analysis would be of the order of percent (refer to Tabic II.S. D-LL 11) rather than 0.1 percen:
The main source of differences in scismic risk estinates be associated with the type of acccJeration-magnitude-distanc.-
relation used.
The applicant's risk analysis is based on the of the Blume's SA:1-IV'and SAM-V procedure.
In my opinion, t!.
procedure Icads to accelerations which do net reficct the s t" motion in the near soerce region of large magnitude carthqual.
If one considers the three largest carthquakes for which rece:
were obtained in the near source region, one finds that the e' scrred peak accelerations are three to ten times larger than :
predicted by the SAM IV-V procedure (Tabic 2).
Since the str:-
deviation for peak accelerations corresponds approxinately to factor cf two, it may be concluded that the SMt procedure is valid in the near source region'rf large carthquakes.
Ta b i c.'
indicates that Trifunac's reistions Jead to accurate estimates the observed peak acccierations (the average ratio of observed -
predicted peak acceleration is 1.07).
Fig. 41-I of Appendi.: D-41 shous that the use of th'c SMi procedurc '1 cads to probabili ti that are 10 times louer than these obtained on the basis of the Trifunac's relations for th: same scismici ty model.
Th e s ci s t.i -
city model described in Appendix D-LL 11 Jeads then to a proba!'
ity of exceeding a peak acceleration of 0. 75r. in 50 years of th-order of 4 percent.
6 33.
9
..-----s p
k.
4 2n.. m..
.w,.v m sm~ ~* ~ ~m m.a. ~~
--~~~r
,2
l s
The scismicity model used in Appendix D-LL 11 is bar.cd on : l seir.nic recurrence relation obtained by Smith for Central Coas -
California (Appendix D-LL 11A). These recurrence relatiens are based on the scismicity during the period 1930-1975 and do net clude the 7.271 1927 carthquake in the region.
The recurrence c-as shoun in Fig. 11A-2 of Appendix D-LL 11A underestinate the number of carthquakes uith nagnitudes larger than six, and are
- based on a neminal value for the pararacter b of 0.92.
Addition study by Smith (Appendix D-LL 45A) indicates that a more appre;-
ate value for b would bc 0.336.
The parameter b i.hich centrel:.
reistive contribution of the high magnitude carthquakes to the seismicity has a strong effect on the calculated risk.
The ute h=0.836 would increase the calculated probabilitics by a tactc:
tiro (refer to Tobic 45.3 of Appendi:. D-LL.15).
The scismicity model considered in Appendix D-LL 11 is c.'
g tent rith the scismicity ob tained in Appendix D-L!, 41 using the f.cologic record of fault disloca' tion (a= 3.12 in D-LL 11, a=2. SO 4
6 bar.cd on 10 years record and a= 3.20 based on 20 x 10 years record in D-LL 41).
The s.cismicity calculated on the basis of geologic record of lateral fault slip does not inc ' n.l e the seis city associated with vertical slip along the !!oscri fault.
!! a--
(Appendix D-LL 41A) quotes a report by Earth Science Associates dicating that the 'laterni slip uns probably subordinate t.c ver; movement.'
If this. is the case, the scirmicity should be incrt.
to account for vertical slip.
Considering all the factors mentioned, it seems that the 11.
O t
+
.x
.x
=. -..
= "
z x
~-"
'~~
~ ~ ~ -
f
{
\\
~
probability of 10 to 20 percent in 50 years ob tained by.\\nde r:
and Trifunac properly reflects the scismic risk of excedance o-t!Io licsgri design spectrum.
0 3
gg
-l 0
4,N$ 9
' t 's g
G D
e 12.
\\
l
.....--,c.--.....,,n a
.J i
~
REFERENCES 1.
Scismic Eraluation for Pos tulated 7.SM !!os gri darthquake, Units 1 and 2, Diablo Canyon Site, Pacific Cas and "Elcetric Company.
2.
Trifunac, !1.D., "Prelininary Analysis of the Peaks of Stren -
Earthquake tiotion-liependence of Peaks on liarthquake
- .:gni-tude, Epicentral Dir.tance and Recording Site Condititus:,"
Bull. Scism. Soc. of Ancr., Vol. 66, pp. 139-219 (1975).
3.
Page, R. A., D.M. Boore, II.P.. Joyncr, and li.W. Coulter.
Ground lotion Valties for Use in the Scismic Design of the Trans-Alaska Pipeline System, U.S. Geological Survey Circula r 672, 1972.
4.
"Irifunac, M.D., "Forecas ting the Spectral Amblitudes cf St r.
Earthquake Ground Motion," Sixth World Confere:1cc on Earth-quake Engineering, New Delhi, India, 1977.
5.
Neumark, N.M., " Torsion in Symmetrical Buildings," Fourt h World Conference on Earthquake Engineering, Vol. II, A-3, Santiago, Chile, 1969.
6.
Anderson, J.G., and M.D. Trifunac, Uniform Risk Absolute Acceleration Spectra for the Diablo Canyon Site, Cali forn i A Report to the Advisory Committee on Reactor Safeguirds,
g U.S. Nuc1 car Regulatory Commission, De ce r.b e r, 19 76.
,1 e
e 13.
que
+
b e
S
- [*%***
%g gp y w ere
-$ aphn, f.
fig M!!
f
(.
c TADLE1.
COM PARISON OF 1.'iAM! MUM CROUND.:OTIONS Peak valuca M = 6.5 M =
used by Newma r k*
T r iiuna c#
USGS' Trifu n.ic '
~
No. 672 max (C) 0.75
,0.69(1.29) 0.90 1.0 ~l.(2.0 0) a-max (in/sec) 24 23(40) 39 39(84) v d
(in) 8 8(19) 16 12(30)
- Ncvemark, N. M., "A Rationale for Development of Design Spect:-a fot Canyon Reactor F.icility," isppendix C Supplement No. 5, SER, i g
Canyon Nuclea r Poveer Station Units I and 2, NRC,1976.
Average (average + standard deviation) penh motion for rock at an epicen-distence R = 7.5 lan hascel on Trifunac, M. D., "Prelimina:, An +!.
of the Peaks of Strong Earthq uahn Ground Motion - Dependence of Pc
.on Earthquake Magnitude Epicentral Dintance and itecording Site C.-
tion s, " 13. S. S. A., 66, 169-219 (197 5).
Page, R. A., et al., "Cround Motion Values for U.cc in the Scismic De i:
, of the Trans-Alaska Pipeline System,"
CcoloC cal Survey Circular i
672, 1972 9
eq.
G 9
7
- I
VSM4
....L W.
? 2 ; c i s.s. e:
.,c cel e r a ; en+;
i l-SA:. IV - s A.: VC4)
Trifuna:(5) i Recorded Predicted Ratio Predicted Ratio i
Peak Peak Observed /
Peak Observed /
Acccl.
Accel.
Predicted Accel.
Predicted 1971 Pacoima(I) 1.25g 0.124g 10.08 0.839g 1,49 1967 Koyna(2) 0.63g 0.213g 2.96-0.765g 0.82 1976 Ga:li f) 0.80g 0.190:
4.21 0.900g 0.89 5.75' 1.07 t
(1) M-6.5, epicentrci distance 3 km, focal depth 15 km.
(2) M=6.5, epicentral distence 5 km, focal depth 5 km (essumed).
(3) M = 7. 2, epicent ral dis tance 10 km, focal depth 25 km.
(5)johs=12,000,5'=2.04,3=o (4) s=2, p=0.50.
u gg
-1 4
e d
g m e t
I 1
)*4*%i i
s e e. r-
~
~
y
'\\
j'\\;_
- STil, 1'
- e
/p
~ h<N,
'JN
+
\\
/
re 0
3N of p
t,? '
f.Q s
.t j-e f
,,f sue
'41
[
+
G gx g
/
E d
O j.
a' 4
p c, -
~
,,qW
~)
to e
~
a O
Y u
.P et fr
/
i i
e s
X i
L
[
g og
\\
- i J
o N
9
'c 2
s.
s%y
\\
\\
~~
- e
\\
t
.7,}
o O
g7 I
'*o,
,to'
~
s,
,.e e,r I
b l
I i
1 1
J't\\
f l
0.03.
e.05 c.t 0.2 07 l
5 to 3o so are Frequency, cps s
FIG.1 - DESICN SPECTRUM COMPARED WITH PACOIMA L
, DAM SPECTRA, 2 PERCENT DAMPING, 3
i i
7 i
5 4
O H
+
i I
I t.
. ~ _ _. _ -
{
(
i 9-m 13 In te:
11 in
'10 s
G
- 'S r-r o-tf an 5
iS T.
'_ n
~
o V
3 si <
T o
2
.o
_b v "O
E i,J I
c no e
f
,~.
18 a
w J.)
45 t0
. c
.r v
ow n
.g L
I In y
\\n l
N (u.*.
6o
/
J3 n
3 1
o n
.o s
u O
,o 9
e
/
i u
e"
/
6
\\
s g
a q
o
()
%I j
Zs, u
ci q.
~
W
.A o
j r
in j
y.z
(
3
\\~
\\
a N
d N
\\N N
\\
d g
+
i O
C.
to O
in O
lA
~
en d
n g
O (5, f> -). "S t
,. a,.
I e
- w 6
- "'"*'N*****
,%'P*****"**g****J*
~~
7
- m. ___
,} {s.( ['g$; jf,j '
d
.g,,
- Q g.
b,,
x...
.;s s
49 N
. a w'
.u.
.2'
.....u-n:.. ~....
...... ~
-a
.f.
. D:.
...n.
....m.#
- :+
s
.w mm.
[.
.4..
e
.,.g A a.co t.u r e Accetca n oy, 9 l
N m
O
'4'm 0
m I
P\\..
.= }
I s
.l
=..
}.
6,4 s
j..
e t,
..%........e s.
e
.. p.
, s' t
1 e.
+
- * * ~....*
..,.?.
l e
g.
.... w
... h.... _... :... -... lg g.;
t:q l
j g
-l g
- t..
n n.n.
1 9
..y i
x3-y 0 X.
r'}
O I)\\.
s.
.( %.;....
.. $. y__.._._.i..
s g.
o A
.3
..L f.
m
.I q
8
. S,'!
.... _... g ("
i y;
i i
0 tm.
t
~ *** ~. ~: :
~.
g
, -_ _., /,.
l.
=.
.r
%, l
...~.~DW*~--=='j.
y
.~/.
. =,
f
. ~ ~ ~ Y-.1 ~
...,..\\,......""~.."*..l..
o=*
== ~
~
I s.
- 6.,
o
%I
. /..
Ig.......e..
U I...
.~"'Il D
y
. j 4/)
s
_,..~......;-
3 o y
.-..../,.
o x.....,..
.S g (.
.p...
t
. "-i..... G1 l
i l'
O L.)..
b 6
v d
t s 00 l
)
l.
./
G
% 0 0 o :t -
c s
I-
'O
.'h.l &
,/..-
i, t'
to In y
i.9
.. -g
.4...
- ~..
.. ~.
s.-
N t+
I.
c.
1 a
g (h
[.
i s
4 6
.,e\\
~I
.)
O
~.
y,...
.g t
-