arnquake Engineering, Tenth Vir Conference 1992 Bakers, Roterdan. (SBN 9054100805
Seismic analysis of underground structures
C.Navaro
Escuela de Ingenieros de Caminos, Ciudad Universitaria, Madrid, Spain
ABSTRACT: This pay
Gonstantopoutos et a (19
extends to other types of strcture the simplified methodology proposed b
(1979) for the seismic design of tunnels. As a practical example, alarge structure of
reinforced concrete, of box shape and toally embedded in sil is analyzed. The dynaraic pressures acting on
walls, roof and floor, due to body and surface waves, are considered inthe analyses. A set of seismic load
Combination hy
i$ are proposed to account forthe differnt polarzation planes of the seismic’ waves, The
Influence of neighbouring buildings can be taken into account considering the new soil sress sates that they
produee.
LINTRODUCTION
Under seismic conditions, underground. structures
respond to diferent seismic waves propagating trough
soll media. One way t analyze this problem involves
three-dimensional finite element analyses “using
Appropriate transmitting boundaries, which allow
Sbeurate modelling ofthe sector andthe surrounding
Sci, These fll narmerical analyses should be cared out
in time domain to account properly for soi nonlinear
behaviour, However this methodology 1s expensive
‘wen utlized aya design tol ofthe stactre, because
Gezpands ing compute ine.
‘Since the pioneering paper of Yeh (1974), in which a
sapled shalyns fr bated pipes subjected to sete
‘was proposed, some atemps have been made 0
apply those theoretical concepts tothe svetaral design
Theavier and more complicated structures, sch a5
‘underground tunnels and galleries in fre-field ondion
‘Constantopoulos etal (1979 and 1980), Chrisano et
(1983) and Navazzo & Samaria (1988).
1h their methodology, Constantopouls et al. (1979)
assumed tht tunels move ina manner similar t that of
the surrounding sol, and ths sol-tnnel interaction
may be neglected, at least in its radional sense,
tecause the most tunnel vibration cnergy is radiated
Say by soil elastic waves. This method requires only
atic analyses and accounts very roughly for the
Gifference between tunnel and soll "gies. They
divided the seiamic tunnel analysis into two part:
transverse and longitudinal analyses. In the frst, 2
tel ose secton anand, ong te Sol
‘pressures against tunnel walls, oor ea
the different seismic waves. These pressures are
obtained from the sesses in sol in fee-eld condition,
{aking into account the changes in the sess istibuton
‘sound the tunnel originated by its physical presence on
Sell mass. The types of seismic waves considered in
this ‘analysis are" shear and. compression waves
Dropagating vetcally, and Rayleigh wave components
{GStosional and dlationa) traveling horizon and
1998
perpendiculary tothe longitudinal axis ofthe tunel.
Love waves are not considered in this part of the
‘alysis because they ae polrized in horizontal planes
Parallel wth ground suace
For longitudinal analysis, surface waves propagating
parallel tthe tnael axis are considered ang he tanael
Ean be modeled at an elastic beam connected by springs
tothe far ela (Winkler model, Nava and Sarasin
(0966) provided an analytical solution for this probe.
‘They considered the tunnel bending analyse, a3
suggested by Constanopouios etal (1979), analyzing
th effects on the tunnel caused by the dstorsional
component of Rayleigh and Love waves, but they
incorporated a "push-pull" analysis to account forthe
intivence ofthe dilatational component of Rayleigh
waves, improving tht suggested for Consaatopoulos
Teal (1970). To eheck the modified analysis, Nevaro
and Samar (1986) compazed the values 10 obtained
wth those recorded in actual earthquakes, observing a
High degee of accuracy.
a this paper, the methodology proposed by Cons-
tantopoulos eal (1979) extended wo other pes of
buried structures. As a practcal example, a eiforeed
oncrece large box (12.$+12.5°120 meters), protecting
Glesel tank, is analyzed seismically. Its founded on
fotact rock (shear wave velocity vp= 2,960 mls and
density p2= 2.71 tons/a’) andthe walls apd te roof
fein contact with a panulr backfill (v1 = 008 and
t= 206 tons/m), The rock material behaviour was
Sipposed elastic, and the variation of the dynamic
properties of the granular backfill versus shear sain
‘ete taken into account Hardin and Drevieh (1972).
‘Assructural wansverse secon is showa in Figure 1.
The seomic excitation cont of wo satiny
independent accelerograms with 4 broad frequenc
‘content. They ae defied at rockbackfi neta Ie
witha peak aceleraon of 0.2, in bots horizontal and
Yercal directions, and-match he response spect given
Inthe the Regulatory Guide 1.60.1050
Figure 1. Geometrical definition ofthe analyzed structure
ve Point A Point B
a Peetet
Bas a —— Poin B
a2 Pan
g 3
1 E
a.
‘ <
Frequency (Hz) Frequency (Hz)
Figure Horizon response spectra at points A andB (continous ine) and atts even fee-fed condition
(dete ie), Damping rao 5%.
2 SEISMIC ANALYSIS this figure he horizontal response speta at suerte
Points A’ and'B.Gee Figure 1 for situation),
Seeing to ronal sialon ee compres
2.1 Preliminary SSI sdies ce obbned in refi condone athe ame
Rages impor of actu ection
é in hese probems, finite element
falyses were cared out, nthe fequency domain,
{ing the vell-enown merical ool FLUSH (Lysmer
als (1975) whic taken into aocout dre dmeasional
effec in an approximate manne, The results conned
that SSt effec were not very portant forthe case of
fanels (Constanopoutos e¢ aL" (1979) and Naver
{@982), nor forthe probes understudy (Figure )-1n
levels, Maximum spectral ordinates appear for the
aural ‘ofthe soll layer. Differences of about
10% are observed: that for point A is greater than that of
free-fieid, whereas for poiat B the opposite occurs. For
verical excitation these differences are much les
2.2 Seismic environment
‘The determination of a seismic input for buried
1940suctres is very complex dve tothe randomness of
‘Etecion and the magnitude ofthe seismic modon, 0
Simplifying and conservative hypotheses should be
‘sed acount for the uncertainties of the seismic
‘vent The selamie waves considered are: verteally
propagiting body (shear and compression), and
Fortzntlly propagating Rayleigh waves, all polarized
in'vereal planes, and Love waves ‘polarized in
horizontal planes parallel to the ground surface. The last
two waves, normally known as surface waves, have &
Glpersive characte? when they propagate throogh &
layered half-space, iy wave phase velocity 33 &
{neton of the frequency of the wave component
onsidered. Forde case of Love waves, result ofthe
Aispersion equon are shown gaphcaly in Figure 3,
forthe simple layered halespace considered. For he
fundarmentl frequency range of an earthquake (1-5
Hay, Love wave components propagate 10 4 phase
well very near to the shear waves veloc ofthe
$Slpace Ail cneosin Ss acer when
{eating Rayleigh waves inthe sae sll sytem ving
Sal os))-Paptumennal nodes of so woven it
{he imple layered half-space, show that no changes in
Sell suesies re observed through layer depth This
‘means that infee-field condone, te maximum
stresses at layer points, due to surface waves, may be
computed a
panicle
o (or) = "Ble + Berd)
FREQUENCY (i)
LEGEND:
‘y= phase velocity of the Love wave component
(1) p1=2.060m3,v 1=3000m/s,92=2 53, v7=18600m/s
G) p1=2.06elm3,v1}=300m/s,p2=2 5u/m3,v2=2960m/5
Figure 3. Dispersion curves for a Love wave
propagating through a simple layered half-space.
1941
where Yparticle is the maximum soll panicle velocity,
wave is the wave phase velocity, E and G the elasticity
and shear moduli of soil respectively, and ¢ and = are
the normal and shear stresses inthe soil mass due 1 the
seismic wave component. The velocity of sol particles
‘may be obtained from the equation (Newmark etal
973):
e
here gay isthe si pea acceleration. Inthe ease of
Rayleigh waves, the sss sate induced in the soll
consists of normal stress, cng on vera! panes,
together wih a shear sues Gompotent Te fistcan be
Sorapie wing Eq (1) and considering & moles E
Sccordingly with the shear deformation. The send
Stress component may be assumed at 10% of he
Brevious one, for # layer Septh below 55 of he
Epieigh wavelength (Wait C588),
oll sess cased by vereally propagating
body waves may be eaicsiced by meant of the
computer program SHAKE (Schnabel etal 913) or
By using be simplified method suggested ty Seed and
Tass (1971) wo evalate sol shea susie caused by
the vera! propagation of sear waves The autor fag
checked tha he eppronimateexyesion ven by Seed
and Tdriss (1971) can also be used for verbally
‘ropegating compression waves, consienng the same
Enefcents bot ioducing the vercalacscration a
‘posnd evel
panicle (ms) = 1.2 ®
23 Seismic sol pressures
(Once he soi senses infeed, casted by each basic
weave typ, are known, sol pressure cisebutons can
te compat wing ast fit clement mesh in which
the sritore and a part of te surrounding sot ave
Todeled, as suggested by Constanopouion ct a
(1895) Ta hs mane, presture concentration ees
Structure comers, caused by the srture and scl
fulness ferences, an be fen int aoeount, Thus
the normal and shear swesedisbuton sting onthe
‘era wall se and athe le sous wld doe
to shear, compression and Rayleigh waves ar oined
igure.
‘This figure presents some pnts ofits. Fist of al,
the preaure disbutons donot follow the Mononoke:
Giabe theory (Seed and Whitman (1970) of seme
ath throt prediedons. This fs due tothe wall
fleibiliy the theory isnot sppcabe otis ye of
Sroctre, The shear sess values due to veteal
‘ropagnton of shear wave ae mach higher than hone
‘obtained from either Rayleigh or compression waves.
Note sit the fst reach a maximom of sbost 103
tons/m? whereas the second values between 0 and 2
tons/m2. Although the shear sess vales would be
Timid by soitwall fon phenoment, the sil forces
are walficor level oil be governed by sear wave
‘ropegncon The nomal sets tengo he wal doe
DYRavicigh and compression waves fre very smilar
thd in prtodee can be sasomed as constant long the
ral. Those exvsed by shear wave presen «minimum
Yiluc about the mide of he wall 84 maximum valsSubscripts P, § and R represent respectively Shear,
‘Compression and Rayleigh waves
Fie 4. Suess dssibution wound the wall caused by
ferent seismic waves, (Verical plane)
LEGEND:
= nommal sess
‘cm shear stress
Subscripts L and R represent respectively Love and.
Rayleigh waves
Figure 5. Swess distribution around walls comer caused,
by diferent seismic waves. (Horizontal plane)
1942
LEGEND:
+= Type of movement adquired by soil particles
S” Shear wave
Compression wave
R-Rayleigh wave, Directionof
L Love wave wave propagation
“igure 6. Seismic waves considered and their supposed
paeation planes,
at the wall comers, as consequence of the stress
Concentration effect mentioned. The shear force and
‘ending moment in the wall seem to be due 10 either
‘compression or Rayleigh waves. However axial force
tnd bending moment at wallfloor and walltoof rather
appear to be due to shear waves,
‘Figure 5 shows the suess distribution across a quarter
of ahorizontal cross-section, at the middle of structure
height. Only Rayleigh and Love wave effects are
‘considered in this figure. The normal stresses reach
‘imum values at the middle of the wall, and shear
stresses due to Love waves can be considered as
‘Accordingly to the polarized planes ofeach basic wave
type, shown in Figure 6, and to the terms of the
Regulatory Guide 1.92, the following seismic loads
‘combinations apply
9) S1-S2-P BW) S1-R
2 St-L1-P — &) $2-Ry
€)12-S2-P 9 RL
DUp-1y-P —b) Ra-Ly
where S, P, Rand L represent respectively any internal
load or bending moment due o 2 pressure disuibution
cconesponding t0 shear, compression, Rayleigh and
Love waves, andthe subscripts give the structural sides
where the seismic pressures are supposed acting. The
design loads should be computed as the envelope of
those calculated in esch combination hypothesis,
24 Influence of neighbouring buildings
‘When massive building ae founded near the considered
seuetre, this methodaiogy does nt access lead to
conservative design values. Recent works (Gémex-
‘Mass6 and Aralla (1984) snd Navarro (1992)) on this
Jase subject prove thatthe inluense of buildings ean
ieally alter the sess states generated in soll mass
‘garding tose obtained in free sotion.
So, forthe ease ofa Dullding (8,000 tons) founded atsurface (Figure 7(a)) on a soil system of the same
aynamic $ and seismic input as said before,
Figure 8) shows the sess sate, generated ito te
fol at diferent distances from the balding edge and in
the mile ofthe ayer depth, for horizontal excitation
For approximately 30 meters fom the bulaing edge,
the influence of the building hes disappeared. For
‘eal exciton the influence decreases much fase.
Figure 700) shows anoler eaze very common in
Nuclear Power Plants faites: a Reactor bulding of
120,000 tons weight and 60 mm diate, founded ona
talespace, wid a shear wave velocity of about 1100
‘ullding is surrounded by a granular backfill
Inver 10 m deep, whose dynamic properties ze those
proposed by. Hardin and Drocvieh (1972), The
Forlzonal seismie input isan accclerogram of peak
seeleration Gag, dened at backil see id
batching the response spectrum of the Regulatory
Guide 1:50, The dependence ofthe sess state oh
Gistance from the wall at an intesedate layer depth, i
Shown i Figure 8@). Ata distance of about 30 mec,
the influence of the building seems to dltappear
However, this de not exactly true Because Navarro
(4995) nas shown that an interference phenomena
ferweea Rayleigh waves, propagating Horizontally
through the Back layer tad aay froma the bullding
tall and verdeal maveling shear waves, occurs at ore
than 30 m fom the wallTo Mlustat tis, Figure 9
Shows the response spectra, fora damping ratio of 3%,
beeen
er ey cee
WS
@
obaned cdferent points ofthe layer. At abou 40 m,
al ordinates are greater han those obtained in
fecheid ‘Pie dies wc ey let
furetion of the wave Tength of the Raylelgh wave
(aver (1992),
3 CONCLUSIONS
‘This paper outlads a simplified methodology for the
design of lege suuctures other than tuntels and
lleres. Te is an extension of that proposed by
Sonstantopoutos et al (1979)
‘Some recommendation are made about the choice of
Rayleigh and Lave wave velocies to be used in the
‘nalts Ths geal smpliies the consideration of the
seas of the selmic waves
Figure 7. Different stuctural systems
Combination criteria, attending to the different
polarization planes of the seismic waves involved, are
rovided
Particular attention should be given when massive
buildings are near the buried stcucture because then
some ofthe ress of he splifed methodology may
‘ot be conservative.
as Toraonal es
a - vale
Bas 7 Sees
Bis hora ses
if versalanes
i aay
Bos i
Tien eae enaen “JD UU EDD Ee Ow
Disa rs ulin en) Disance tom bulliag val (n)
@
Figure 8. Stess stares
1943
®Ground surtace
Middle of back tayer
ACCELERATION (*8)
aurea sas
FREQUENCY (Hz)
@
FREQUENCY (Fiz)
o
Figure 9. Response spectra fora damping ratio of 5% at different distances
REFERENCES
Constantopoulos, LY. etal. 1979, Dynamic analysis of
vunnels. Proc. 3rd. Tat, Conf. Num. Meth.
Geomech..
Constantopoulos, LY. et al. 1980, Seismic analysis of
‘buried tunnels: Pros. 7th, World Con. Earthquake
Engng.
Christiano, P.P. et al. 1983. Response, of buried
Tunnels to earthquake excitation, Trans, 7th Int. Cont.
Sect Mech Reacon Tech Vol K: 287294,
Ewing, W. M. et al, 1957. Blastc waves in iayere
toed, Pentel
Gémez-Mass6, A. & I. Atala 1984. Finite element
‘versus simplified methods in the seismic analysis of
underground structures. Earthquake Engng. Stroct.
Dyn. 12: 347-367.
Hardin, BO. & VP. Dmevich 1972. Shear modulus
£6 dirping in sols design equations and caves. J
__ Soil Mech, Found, Div. ASCE 98: 603-624.
Lysmer, J. etal. 1975. FLUSH, a computer program
for approximate 3D analysis of soil-stucture
interaction problems. Rep. EERC 75-30.
Navarro, C. & Samartin, A. 1988, Simplified
Yongitadinal analysis of buried tunnels. Software for
ang. Workstations 4: 3-10
Navarro, C. 1992. Effect of adjoining structures on
seismic response of tunnels. Int. J. Num. Anel. Meth,
Geomech. Gn press).
"Newmark, NM eta. 1973, Seismic design specra for
‘Noclear Power Plans. Power Div. ASCE 98.
‘Schnabel, P.B, etal. 1975, SHAKE- A computer
‘gram for earthquake response analysis of horizontally
Iintered sites, Rep. EERC 72-12.
Seed, HB. & Idriss 1971. Simplified procedure for
evalua sll Lgutacton potential J Soll Mesh.
Found. Div., ASCE 97:1249-1273,
Sedd, H.B. & Whitman, R.V. 1970. Design of earth
‘taining structures for dynamic loads. Lateral stesses
in the. ground & design of earth retaining
structures: ASCE.
USNRC. Regulatory Guide 1.60. Design response
ta for Seismic design of Nuclear Power Plants
INRC. Regulatory Guide 1.92, Combining modal
1944
res-ponses and spatial components in seismic
response analysis.
Yeh, G. C.K. 1974, Seismic analysis of slender buried
‘beams. Bul, Seis. Soe, America 64: 1551-1562.
Wolf, LP. 1985. Dynamic soil-structure interaction.
Prentice-Hall.