Myr,Err, A. E. & ME!.C. C. (1983). Geotechnique 33, No. 3,
293-3Technische Hogeschool
Dell'
Earthquake-induced stresses in a poro-elastic foundation
supporting a rigid structure
A. E. MYNETT* and C. C. MEIt
A fluid-filled poro-elastic foundation supporting a long and rigid structure is subjected to a normally incident Rayleigh wave. For sufficiently low permeability and high frequency, the pore fluid and the solid matrix move together as a single phase, except in a thin boundary layer near the ground surface. Hence the dynamics of most of the porous medium is governed by the usual laws of elastodynamics with effective moduli which depend on both constituent phases. In this Paper, the dynamic stresses and pore pressure beneath the structure are studied. When the structure is small compared to the wave length, the neighbourhood of the structure is
approximately quasi-static with time being a parameter only. The analytical method of Muskhelishvili in elastostatics is then applied to obtain the effective stresses near the structure. By mass conservation, the corresponding pore pressure is found immediately. Boundary layer corrections are added to complete the solution. The so-called outer solution presented herein apply also to the pure elastodynamic case without pores, and the method can be easily modified for any large-scale stresses in the background. Scattered waves, being small, are not considered.
Une fondation poroelastique remplie de fluide et supportant une structure longue et rigide a ete soumise une onde de Rayleigh d'incidence normale. Pour une permeabilite suffisamment basse et une frequence suffisamment elevee le fluide interstitiel et la matrice solide se deplacent ensemble comme une seule phase, a l'exception d'une couche mince limite pres de la surface du sol. La dynamique de la plupart des milieux poreux est regie par les lois usuelles de l'elastodynamique avec des modules effectifs qui dependent des deux phases constitutives. Cet article etudie les contraintes dynamiques et la pression interstitielle au-dessous de la structure. Lorsque la structure est de grandeur reduite en comparaison de la longueur d'onde le voisinage de la structure est approximativement quasi-statique, le temps n'etant qu'un parametre. La methode analytique de Mushkelishvili dans l'elastostatique est alors appliquee pour obtenir les contraintes effectives pres de la structure. La pression interstitielle correspondante est trouvee tout de suite par la conservation des masses. Des corrections des couches limites sont ajoutees pour completer la solution. La solution dite exterieure presentee dans
Discussion on this Paper closes on 1 December 1983. For further details see inside back cover.
*Delft Hydraulics Laboratory.
t Massachusetts Institute of Technology.
Lab. v.
Scheepsbouw\kur:de'
293
l'article s'applique aussi aux cas purement elasto-dynamiques sans des pores, et la methode peut etre facilement modifiee pour des contraintes considerables a l'arriere-plan. Les ondes disperses sont negligees, puisqu'elles sont de valeur reduite.
INTRODUCTION
In this Paper the application of a recent boundary layer technique for obtaining approximate solutions of poro-elastic problems in geotechnical engineering is explored. The specific task is
concerned with a comparatively small rigid structure resting on the surface of a wetted porous ground during the passage of a seismic Rayleigh wave. Being confined to the vicinity of the ground
surface, the Rayleigh wave is attenuated weakly from the epicentre in contrast to P or SV waves,
and can therefore be the most destructive to
buildings. Knowledge of the stresses beneath a building when attacked by Rayleigh waves is
important to the estimation of foundation stability.
Mei & Foda (1981) showed that for frequencies
common to
seismic waves (5 10 rad/s), thepermeability of most of the ground materials (sand, rock, clay) is so low that relative motion between
the pore fluid and solid matrix is insignificant except near the unsealed ground surface. This
permits a boundary layer approximation whereby
a one-phase (solid-fluid composite) description
suffices away from the ground surface (i.e. in the
outer region) while a two-phase description is
necessary only within a thin boundary layer near
the unsealed ground surface. If Hooke's law is
assumed between the effective stresses and
displacements of the solid matrix, then the outer
composite medium moves according to the laws of
single-phase elasticity with appropriate material properties. In particular the following effective
density, Poisson ratio and Lame constant emerge = nP. + (I n) + 2mv ve= 2(1 + m) AC=(2v + 1/m) 1
where the parameter
ARCHla
a
=
MYNETT AND MEI NOTATION
amplitude of vertical ground acceleration
speeds of p and sv waves permeability
wave number = 27r/wave length
tiG 1
= stiffness parameter
13 1 - 2v
static void ratio
dynamic pore pressure in the
fluid amplitude of hydrodynamic pressure degree of saturation time velocity of fluid velocity of solid matrix spatial co-ordinates
dynamic effective stress in the
solid matrix
dynamic total stress in the composite medium
shear modulus of solid matrix half width of the structure 1 + m(3 - 4) = effective elasticity 1 +m parameter 2vG/(1 -2v) = Lame constant of solid matrix n G = 1 - 2v /3 (4)
essentially represents the stiffness ratio between the solid matrix and the pore fluid, G being the shear
modulus of the solid matrix and 13
the bulk modulus of the water-air mixture in the pores,while n is the porosity and v the Poisson ratio of the matrix.
If the total stress tensor Tr, is defined to be the sum of the effective stress cif and pore pressure p,
i.e.
where 1,1 = 1.2, then the boundary layer approach
yields the important result that the total normal component r and shear component T12 near the mud line y = 0 are essentially unaffected by the
boundary layer corrections in a and p, there
being negligible boundary layer correction for a12. Hence it follows that the traction conditions on themud line can be applied directly to the outer
problem, which can be solved first by existing methods of elastodynamics. Afterwards,
con-servation of mass implies that the outer pore
pressure is related to the normal stress components(2v
+ 1 Ini)G
= effective Lame 1 - 2v
constant of composite medium
InK,
14 = effective elasticity
para-27z
meter
v Poisson ratio of solid matrix ve (1 + 2mv)/2(1 +m) = effective
Poisson ratio of composite medium
# bulk modulus of fluid with air
bubbles
6 boundary layer thickness
Si) Kronecker delta Ps}
P.
density solid ty of(
fluid) p, np, + (1 - n)p, = effectivedensity of composite medium dimensionless independent vari-ables in the outer region
co wave frequency
;pi, dimensionless unknowns in the
outer region
ji dimensionless corrections in the boundary layer
( )° dimensional boundary layer cor-rection
()O dimensional outer approxi-mation
( )R Rayleigh wave component according to Po = -12m(ail° + 0-22°) for plane boundary 1 i104-T22°) (6a, b) 2(1 +m)
strain. Near the mud line y = 0 the
layer correction is of the format =
Ph =
I -i
exp (7)
where the boundary layer thickness is given by
H-
(Gkr2
rtIG I - 2v 1-1/2w fi 2(1 - v)] (8)
and .nt is chosen such that the outer pore pressure
and its boundary layer correction combine to
balance the externally applied pressure on the mud line.As a
consequence of Hooke's law, the,a
_
0 Tij = (5) a22b a Cp, ,n S x, y -L-
2mEARTHQUAKE-INDUCED STRESSES IN A PORO-ELAS1 IC FOUNDATION 295
boundary layer correction for the velocity or
displacement is negligible in contrast to the outer solution.
Although numerical schemes via integral equations or finite elements are available for many
geotechnical problems, it is preferable for both
economy and insight to apply analytical methods whenever possible. This is the case with small structures whose dimensions are much less than the elastic wave length. The region everywhere
outside the boundary layer shall be called the outer
region which can be further divided into the far and near fields from the structure. In the far field the wave length is the characteristic length. In the near field, i.e. the neighbourhood of the struc-ture, the characteristic length is the structural
dimension. Inertia effects are relatively small in this
near field so that both the local stresses of the primary wave and the induced stresses by the
presence of the structure are quasi-static. For plane
strain, the technique of complex functions can be applied as in Mei & Mynett (1981) and Mynett &
Mei (1982).
In this Paper we concentrate on the near field problem for a poro-elastic ground, the boundary layer corrections near the mud line being readily added afterwards. The results for the total outer stresses are directly applicable to a non-porous elastic ground for which the pore pressure is no longer significant, if the proper effective elastic constants are replaced by the actual constants for
the single phase medium.
The properties of the incident Rayleigh wave are needed first, in particular the approximate stresses
in the near field.
RAYLEIGH WAVES IN THE BACKGROUND
The space co-ordinates in the outer region are scaled by K-1 where K =2n11 is the Rayleigh wave number, and time by co-1, i.e.
X, Y = K(x, y), T= rat (9)
The dependent variables are scaled as follows: ?id = (u,°, vnl(alco)
(Air' "Fii)= (1)°, aij°1 T1j°)/P0 (10)
with
GKa
P0= (02 (11)
where a is the maximum vertical acceleration of
the ground surface and equation (13) is implied by
Hooke's law. With reference to Appendix I where
the outer problem is reduced to classical
elastodynamics as in Mei & Foda (1981), the
Rayleigh wave solution is (Graff, 1975)
(/) = Re A, egY E
çfr = Re 11,e'v E (12)
where A, and A2 are constants, and
E = ex p[iK(x CRr)] =- exp [i(X T)] = 1 CR2 IC p2 , s2 = I CR2/Cs2 (13)
and CR = (o/K denotes the phase speed of the
Rayleigh wave. The outer velocity components are ao
D,R = Re (iA egY + sA 2 er)E
ax aY 00
= = Re (qA egY iA2esY)E (14a-b)
ay ax
where the superscript or subscript R denotes
Rayleigh wave. The corresponding effective
stress components, related to the displacement of the solid matrix by Hooke's law, are given by
2v 2
7 71)
R = Re[(
1-2v 1_2v) x iA, 2sA2e2 E[(2
2vq 2 2v 2v 12v)
x iAi eqr + 2s A
elE
a 12R =Re[-2q,
, +(1 + s2) i A 2elE
(15a-c)
while the outer pore pressure follows immediately from equation (6)
=
Re[1/3
(q2 1)A
eglE
(16)nG
From equa ion (5), the corresponding total stress
components are 2v
2-2v
R = Ref[
q2l 2v
1-2v+(q2
iA,eqY2SA2esY}E nG .122R= Re12 2v 2 211-2v
1-2v
+(q2
nG 1)1 +2sA,ea}
E2R = Re [ 2qA, egr - (I + s2)iii,esr]E (17a-c)
By virtue of the boundary conditions on the ground surface i22R TI2R = 0 (18) "6-2!= Re q2 egr 2
where Y= 0; Al and A, are related by
2g
A, = iA1
1 s2
and the following expression for the wave velocity CR is obtained
_CR= (087+ 1.12ve)/(1 +v) (20)
C,
Once CR is found, q and s are known and the outer solution for the Rayleigh wave is complete. From equations (14) and (19) the horizontal and vertical accelerations of the ground surface are
a,R(x,o,T) = Re A , 1 1+ s2 2qs a2R(X,O,T) = Re i A ,
q(
s22)E 1 + s (21)Since the vertical ground acceleration a is used for normalization, we must have
A =
1+52 I (22)1 s2 q
A1 is defined to be real so that the outer vertical acceleration of the ground surface is
(a2o)I2
(19)
where Y = 0.
The boundary layer correction PR is obtained by specifiying that PR + = o on Y = 0 (Mei & Foda, 1981; Mynett & Mei, 1982). From equation (16) we then obtain i 13 22R = R = Re (q2 1)A, nG [ex YAE
P V2 6
(24)which must be added to
and PR for thesolution. Some numerical results have been given by Mei & Foda (1981).
Our aim is concerned with a rigid rectangular structure of width 2L and height H in welded
contact with the ground surface. We now add the
assumption that the typical structure dimension
(here the base width) L is much less than the
Rayleigh wave length, yet much greater than theboundary layer thickness 6, i.e. IC"' »L»6. The
Rayleigh wave solution can then be approximated
for Kx,Ky«1
inthe neighbourhood of the
structure.
NEAR FIELD APPROXIMATION OF THE BACKGROUND RAYLEIGH WAVE The near field dimensionless variables are
( (x, y)/L = (X, Y)/KL, T = tot (25)
v
= Re (q2 1) A ,
1v nG
.612K = 0
while keeping the other normalizations of
equation (10) unchanged. The Rayleigh wave solution (equation (15)) are then expanded for
small KL = E« 1. Keeping only the linear terms in the expansion and taking the real parts we get
tf(l) +f(3)
5-22R = E./(4) +1)6) & 12R =
(7) EPs) ./(9)
where the expansion coefficients are
= A,
( 2v
q2 2 2v4sq )
1 2v
1 2v
1 +52 COS Tfa)
2v q 22-2v
4s22v
1-2v
1 + s2 _To) Ai( 2v
1 _21,q1-2v
+1+52 sin r2-2v
4sq )
=Ai(2-2v
.72 2v 4sg1-2v
1-2v
1+52 sin r f(6) f(7) (26)Similarly, the normalized pore pressure and the
solid displacements can be expanded for small K L to give
pit Ep(I) Ep12) p(3)
CR a_ piR _oift Eu(I) +Eu(2)1+ ui3)
pit
7R_
-2Ra = EV 0/(2) + 1/(3)(28ac)where the coefficients for the pore pressure are po) A1 (g- I) cos t m ( I 2v (2) A
(q2 1)
.sint
P-
(I 2v)
p13)= A (q2 l)sin r (29ac) m (l 2v)The coefficients for the horizontal displacement
components are i)
2sg )
A,(
I+s, sin r
U2 = A I ± S2 1(q 2s2 q)COS T f (4) =A,(2
2va 2 2v4sq )
I 2v'
1-2v
1+52cost
f (5) A 2 2v 2 5 +2.52) =i(
q21-2v
I12.v 14 sin r1=0
= A , 2g(s q) cos t=0
(27ai) = a sin K(x t) (23))
-= &1R =f(4
(
sin -= + -qU3=
-
AI (1 2sci1+s2) cos
T (30a-c)
and the coefficients for the vertical displacement components are 1/(11= - A
i(q
2)
I" 1+S2v(2=
A1( 2s , sin 1+)
s-2qV(3) = A
i(q (31a-c)
sin r I +s2All coefficients have a simple harmonic time
dependence. Also, since both fiR and 6R are linear functions of ( ,ri) the same holds for o If the wave field in the half space is that of P-incidence or SV-incidence, as discussed in, e.g. Graff (1975), similar approximations result
in linear stress fields too. In other words, our
subsequent analysis can be readily modified for
any non-Rayleigh waves by changing the expansion coefficients f('), TO, etc.
These results give the background stress field.
We now examine the additional effect of the
structure.GROUND RESPONSE DUE TO THE MOTION OF THE SMALL STRUCTURE
In general the structure not only scatters waves
but also radiates waves by its induced motion. Around the structure whose dimension is L the variation of stresses must be comparable to that of
the Rayleigh waves, and is 0(KaKLGIco2). By
Hooke's law the associated skeleton displacement relative to the Rayleigh wave must be 0(K2aLlw2). Hence the associated skeleton inertia per unit
volume is 0(i) K2aL2). Now the stress force on a
unit volume of the skeleton is 0(K2aG1w2). The ratio of the skeleton inertia per unit volume to the stress force acting on it is then
0(p w2 L2/G,) 0 Psw2 2m
K (KL)2
G
= 0(KL)2 «1 which is very small since
0(p,co2IK2G)= 0(1)
(cf. equation (65), Appendix I). Consequently the induced stress field is also quasi-static. Below the
boundary layer near the unsealed mud line, the
outer boundary value problem for the near field is
to seek a biharmonic function f subject to the
boundary conditions
-.122 = il2 =0 (32) where I I > I, =
= u
0 (33)r= vo+Ewc, (34)
where I <1, = 0; Uo, Vo and Wo are the
horizontal, vertical and the angular displacements, respectively, of the rigid base, which are undetermined functions of T. The dimensionless
rotation Wo is 0(r,) times smaller than the dimensionless vertical displacement, owning to the
long wave length. The background stresses
function, due to
the free Rayleigh wave FR , depends on and rl linearly, so that the difference F PR must not have a linear growth as+112)1,2 CO. As in Mynett & Mei (1982), the
EARTHQUAKE-INDUCED STRESSES IN A PORO-ELASTIC FOUNDATION 297
problem can be separated into two parts
=
+ P
(35)where the superscript S designates the disturbance
due to the structure. Now Fs is also biharmonic
and must satisfy the following boundary conditions on r! = 0 1.22s = 12s (36) where I I >I Cs=
CR = v0u13)Euo)
(37) = vo +Ewo OR =V0
v(3) +E(wo v"1) (38) while should vanish away from the structure, i.e.0 as r (c2 ,12)2. (39)
The above boundary value problem can be
solved by the function-theoretic method of Muskhelishvili (1977). Specifically, we search in the complex r, plane= +M) for
an analyticfunction 0g)
in terms of which the stresscomponents are
ills+i225=
2[00+(C)]
(40)i22S ii12S = 46V;) (41)
while the displacements satisfy
0
-
_ 2 [Us+ i c c¢' (0 (42) The parameter,k` -
1 +in 1+ m(3- 4v) (43)varies monotonically from 1 to 3-4v as in varies
from 0 (incompressible fluid) to cc, (dry soil); see Mynett & Mel, 1982 for a plot. On the mud line, vanishing of surface traction z12 = i22 = 0 over
I I > 1 leads (from equation (41)) to the boundary
condition .0
P
= 0 + + I298
+ =0 (44)
where I I >1, = 0, (k+M (1)(+i0) and
(1)(xi0). Since the displacement is formally prescribed over I I <1 we have from (equation
(42))
a
()+ K (;) = (us + vs) 2g5'() (45)
a
where I I <1, = 0, II s'(*) = WO fitt'() and 0 gR'
= (//g
+il7R)'
2sq = { 1 1 + 52 sin .1- -1; 2qi q
)
/
1 +s2 COS TThe corresponding solution is
X(Of=2rci _ X +
C)
+ ( W, + iW2)X(C)The near
fieldstress components may be
obtained from equation (40) and (41), and the
amplitude of rotation Wo and the complex function of time (W, + iW2) are found. These coefficients
may be
obtained by invoking the dynamicequilibrium condition of the structure. DYNAMIC EQUILIBRIUM OF THE STRUCTURE
While the inertia of the ground matrix
isnegligible in the near field, inertia of the structure is
of the order 0(psal2) and is comparable to the
force acting on the structure by the ground
0(K2 aG121w2), since the ratio of the two forces is 0(p,, w2IGK 2) =A small structure in long Rayleigh waves is
expected to follow
the motion of the wave
displacement which is 0(aco- 2). In contrast, theadditional displacement due to the induced motion of the structure is only
(K2 LaG)L
(02 G
(KL)2
(.72«072.a a
Thus the acceleration of the structure can be
approximated by the acceleration of the Rayleigh wave at the centre of the base,= = 0). Using
this, the structural inertia in dimensionless form
becomes Ma,° 2sq horizontal
=,(1
cos r Po L 1 + s2 (50) and M a vertical 2q sin r (51).P L
cl A,(q
s2where M is
the mass of the
structure andM = Ma/P0 L. There are two contributions to the inertial moment about the origin. The first is due to the height of the centre of gravity 1112 above the ground surface. This inertial moment is of 0(p, 1.,3 a) for H 0(L). The other is due to the
angular acceleration of the structure 1w20. Now / = 0(p, L4), and 0 = 0fK(alca2), hence
1w20 = 0(p,L4K(a w2) co2) = 0((KL) ps a 12) which is much less than the first. Hence the inertial moment is dominated by IC1(H12L) A,(1 2sq
cost
+s2+s2-2sy
= 11-4-(HI2L) 1q(1 s)
cos r (52)These inertial forces must be balanced by the total forces acted on the base of the structure.
From equation (43)
(22
12) = (/)-()+ (PIO
(53) where I <I, since the Rayleigh wave part doesnot have any surface traction. Using equation (47) and Plemelj's formula we can show that
1i22()
al 2(0) + 1 =gs'()+
X + ()(WI +1W2) K, K 1 X +() 11 (54) 'C 2niJ-1 X+(0)(0)
where the last integral is the Cauchy principalvalue. By integrating equation (54) over the entire base, we obtain the resulting forces which are then
used to
write down the linear and angular
momentum equations for the structure.
Ci22()+iii2(0](14.
-=/i3A,(q
2g )sin
r 1 +s2 2sq ia A (1 COS I ±S2f
= Re f 22 -1 (55a) where X(C) = (t.", -1- 0- 1/2 +ig.( 1)- 1/2 with In lc Pc = " )7r-4)
I06)
= (47) (48) (49) 0(1). _1 = 2q,l(G)---= ICI(H/2L)'A (1
1 + s2
2sq )
CQS T (55b) Equations (55a and b),constitute three real
equa-tions to determine the three unknown coefficients Wo, W, and W2. In the Boussinesq theory for a,line
/load, the linear displacements U0 and 1/0 are
indeterminate in the two-dimensional context. As gs( *) is constant with respectto all the integrals in equation (55) can be evaluated by the techniques
explained in Muskhelishvili (1977). Details are
available in Mynett (1980). From equation (55a) we have 2 . 1-1.1 ±ii12):g .cl. Kc 1 ,
1
i
1 ' -+ (.11/,+1W2)i .V(0c1 1 K, + 1 +Kc+ 1 2_ ,fo
Xf
Kc"
f 2nt 's _1+(o),I(o 0
'I X+ Igo
2q5 'OW, + iW2)27r1 e 2gS4 Ice ' Kc -_--7: 2ni(W1 + iW2) = A71 . 1 1,(q -)sint
1 +s2 2q Ct A 1(11)
cost
i +s2 2sqTaking the reall and imaginary parts we obtain (1 =-2ir 1.+s2 COST 2sq Wv = _A - i - A , 1
2q )
2n I + .92 sin TAlso from equation; (55) we get
lC.
111 (--i22+7'El2gd gs'f
t (W! .i14(2)1 4".(0)rd-K+1
fa+gi
e 27ti j X + (o) +iw2)2ikc+h2.n+igs1:(1' + 4/42) (58)Equating the real part of the preceding ?result( to equation (52) yields 2n wo W2 zg,
(1 +.4p)
c,-F 1 (56). (57a, b) 2so (H/20 A I ' I +S2)cos T (59) which gives W0.With the outer total stresses thus determined
throughout the near
field,the pore pressure
follows from equation (6). Afterwards a boundary layer correction similar to equation (24) for pore pressure and effective stresses can be added. All ofFig. 11. Dynamic outer pore pressure f) = e/Pc, near a 'rigid'
structure subjected to a seismic Rayleigh wave. Dimen-sionless time r = cut is marked on the right of each figure EARTHQUAKE-INDUCED STRESSES IN A PORO-ELASTIC _FOUNDATION 299
K, 1 Ai) = K, -0
= (W,
-2 -3 -5 03 2 01 00 0-1 0 = =this should be superimposed to the background
Rayleigh wave solution (outer part plus boundary
layer correction) in order to obtain the complete
stress field.
NUMERICAL RESULTS
To demonstrate the essential features of the
complete solution, we present some sample results of computed dimensionless pore fluid pressure and effective solid stress components for the following inputs H/2L = 1; pjpw = p structure/p
water = 2-5; m= G/13 = 1 (with n = v =
a = vertical ground acceleration = 0-1g and
KL = E = 04.
Thus the structure is a solid of square
cross-section. The value of the
stiffness parameter m = 0/13 = 1 corresponds to a compressible pore fluid (for fully saturated pore fluid on =G/13 = 0-0044; see Mei & Foda, 11984First we present the dynamic outer solution in the neighbourhood of the structure, by including
the near field approximation for the background
Rayleigh wave field, but ignoring the correction in
the boundary layer which
istoo thin to be
incorporated in the contour plots. Quantitative
correction due to the boundary layer presence will
be considered later. The static response to the
weight of the structure is also excluded. In Figs* 1
4 contours of p and are plotted for different
;instants in a half period cot = r = 0, n/4, n/2,
3n/4, n. These results show the modification of the 'background pressures and stresses in the vicinity of the structure. After referring to equations (26)(29) for the behaviour of the Rayleigh wave
solution, we may observe for the particular instant- 11
of time cot = = 0 that the outer solutions io
and all are antisymmetric
in while the corresponding stress components of the Rayleigh'wave in the background are also antisymmetric.
The shear stress 5.12, however, is symmetric in
both in the background Rayleigh wave in the
resulting field.However at cat=t= n/2,
the solutions for 15, and a22 become symmetric, while Er,2 is antisymmetric in In general thesolutions behave quite differently after one quarter
wave period, but return to their original form
except for a sign reversal after one half wave
period.The outer pore pressure does not vanish at the
ground surface as the boundary condition requires, therefore a boundary layer correction is necessary. In Fig. 5 the value of j'7(, 0, "C) is plotted for t = 0, n/4, 7r/2, 3n/4, it. The required boundary layer correction is simply the real part of
*Throughout Figs 1-4, the abscissa is = x/L, while the, ordinate is = y/L..
77
77
77
Fig. 2. Dynamic outer effective normal stress component = (7110/P0 near a rigid structure in Rayleigh wave 1/3); = = 2 3 4 -77 2 3 4 -5 0 0.3 0-0 0-1 0.2 0.3 2 = -5 -3 3 5
-2 3 0 1 2 3 1 2 3 -4 2 0-0 -0-1
EARTHQUAKE-INDUCED STRESSES IN A PORO-ELASTIC FOUNDATION 301
0.3 0-2 1 0.1 - 0-1 - 0-2 - 0-2 -03 -0-3
404
5 "7 - 0.1 - 0.2\
-03 0.0 -0-1 0-00.0-
-4--5Fig. 3. Dynamic outer effective normal stress component
= o-22 Po near a rigid structure in Rayleigh wave
0.2---N-0!'t0 3r -4-4 0 '7 o -3 - 4 -5 0 -1 -2 -3 -4 -5 0 -1 -2 -3 -4 -5 -5 -3 -1 1 3 5 I Qz_CV I 0-1 --N?L° 076
-_---0 .1 0 1 ---°. 2 0 -3 0-5 -- 11---(07.0
- 0 .2 -- 0.6 0-1 - 0.2 3r -4- '7 o-
1 - 3 -4 -5 -1 -2 -3 -4 -5 - - '01 -0-01 -0-02 -0-05 0.01 CY05 0.02 oaFig. 4. Dynamic outer effective shear stress component a-12 = 120/P near a rigid structure in Rayleigh wave
0.1 0-2 0-01 0-10 0-02 0-2 0-1 -0-10 - 0-01 - 0-02 -5 -3 -2
302 1 -4 -2 0 2 4 08 0.4 -04 -08 0 8 of= 0 (of = IT 0.4
u
---
0 0 -0.4 -0.8 ^ 0.8 cot= IT 4 3rr (Dt = -4--0.4 p 00 -0.4 -0.8Fig. 5. Dynamic outer pore pressure ;5 = elP along the I
free surface (ti = 0) near a rigid structure in Rayleigh wave= -4 -2 0 2 4
EARTHQUAKE-INDUCED STRESSES IN A PORO-ELASTIC FOUNDATION 303
--):5(,0,r)exp (1
)E
The boundary layer feature for the same instants is shown in Fig. 6 where the cross-sectional profiles for the pore pressure are plotted for
= 2 and 2.
The solid line represents the complete solution including the boundary layer correction, which
vanishes at the free surface; and differs from the
outer solution (represented by the dotted curve)
only in the thin region near the free surface. In Fig. 7 we further plot-fr.,i22 and i 2 along the
base I
<I,
= o-. in is nearly the same as j) all the time so that the effective stress all is relatively small and the dynamic load is largely borne by the pore fluid. This sharing of load is expected to vary with the degree of saturation which influences the rigidity m of the composite medium.In conclusion, we have shown that the dynamic interaction of a rigid footing with a fluid-saturated
foundation can be solved by an approximate
analytical method if the porosity is low and the
footing is much smaller than the wave length but
much larger than the thickness of the boundary layer. The same idea can of course be extended
to three dimensions. ACKNOWLEDGEMENTS
We are grateful for the support of US National
Science Foundation through Grant (CME
792-1993). Computational support was also received
from Delft Hydraulics Laboratory during the
preparation of this Paper.APPENDIX I. ELASTIC WAVE EQUATIONS FOR A TWO PHASE MEDIUM
The linearized Biot (1956) equations for the momentum conservation of the solid and fluid phases are
avi Op
n-p,(1 n) =
(1 n)+(ui vi)
(60) Or ex; Ox, k au, Op n2pen
= - n---
(141- (61) at ex, kWhere Darcy's law of flow resistance is incorporated as the last term in each equation. Hooke's law is assumed between the effective stresses and strains in the solid skeleton. Subtracting the geostatic part, the dynamic Hooke's law may be written as
Oa, Gfar. au;)+ Aeu,
(62) et (3)e.,
3.0
ex,The equations for conservation of mass in the two phases can be combined to yield the storage equation
a ay, nOp
n(Ui
= (63)ex, ax, # et
where 13 is the bulk modulus of the pore fluid which depends on the degree of saturation S. the bulk modulus of thefullysaturated water /30 and the total pore
71
-
17-0 17-0 -5 -02-01 01 0.2 -0.2-0.1 0.10 2 3ffFig. 6. Cross-sectional profilesofpore pressure. Solid line: total (p°+ph)/ P and dotted line: outer approximation p. The scaleof the pore pressureisindicated inside the figures
5 3 -1 1 -5 --0.2-0.1 -02 0.1 02
ti
Fig. 7. Stress and pore pressure distribution along the
base of a rigid structure in Rayleigh wave. Curve 1:: curve,
2: 2 29 curve 3: t2 4s, 0 8 0 4 0 0 -0.8 08 0.4 0.0 0 4 0.8 1 -08 -04 00 0.4 08 0.8 0'4 0-0 0-4 -08
pressure po
1
I S
= + (64)
Pb Po
In terms of the dimensionless variables defined in equations (9)-(11), the dimensionless momentum equations (60) and (61) are then characterized by two ratios: ow2/G10 and tuolkGK2. The first ratio signifies the importance of inertia while the second interphase resistance. For a wide variety of soils and earthquake frequencies
EARTHQUAKEINDUCED STRESSES IN A PORO-ELASTIC FOUNDATION 305
ruolkGIC2» I (66)
From either momentum equation follows the result that
The storage equation (63) reduces to
. T
"G
13 OT
The momentum equations can be combined with equation (67) to give
p,c02 /2 a2
v2 +
V (V i;)- V (69)
G Pt 2 1-2v
where )2 denotes the effective density defined by equation (60). Hence, by eliminating 0j5/0-r
Nco2 12 02 172
(
1 /3= V2 + + V(V (70)
G at2 1 -2v nG
The velocity vector of plane strain elastodynamics can be described by two scalar potentials 4, and 0, i.e.
It = V0+V xiiez (71)
where ez = (0,0,1).0n substituting this into equation (70) and taking the divergence and curl alternatively, the elastic wave equations are obtained
zic+ 2G 2
i1r2 p,w2 12 V (/'
02 IP G v2
(3,2 p, a,' 12
where and correspond respectively to compressional and shear waves with propagation speeds Cr and C
REFERENCES
Biot, M. A. (1956). Theory of propagation of elastic waves in a fluid-saturated porous solid. J. Acoust. Soc. Am. 28, 168-191.
Graff, K. F. (1975). Wave motion in elastic solids. Ohio State University Press.
Mei, C. C. & Foda, M. A.(1981). Wave-induced responses in a fluid-filled poro-elastic solid with a free surface a boundary layer theory. Geophys. J. R. Asir. Soc. 66, 591-637.
Mei, C. C. & Mynett, A. E. (1983). Two-dimensional stresses in a saturated poro-elastic foundation beneath a rigid structure. I: A dam in a river. Numer. Analyt. Meth. Geomechan. 7, 57-74.
Muskhelishvili, V. I. (1977). Some basic problems of the mathematical theory of elasticity. Amsterdam: Nordhoff.
Mynett, A. E. (1980). Dynamic stresses and pore pressure in poro-elastic foundation beneath a rigid structure. DSc thesis, Massachusetts Institute of Technology. Mynett, A. E. & Mei, C. C. (1982). Wave-induced stresses
in a saturated poro-elastic sea bed beneath a rec-tangular caisson. Geotechnique 32, No. 3, 235-247.
pco21G10 = 0(1) (65) Cr2 = + 2G)//),; C, = Glp, (74)
and in physical dimensions.
(68)
-(721