Professional Documents
Culture Documents
North-Holland, Amsterdam
Ajaya K. G U P T A
Department of Civil Engineering, North Carolina State University, Raleigh, North Carolina 27695-7908, USA
Many nuclear power plant buildings, for example, the auxiliary building, have reinforced concrete shear walls as the primary
lateral load resisting system. Typically, these walls have low height to length ratio, often less than unity. Such walls exhibit
marked shear lag phenomenon which would affect their bending stiffness and the overall stress distribution in the building. The
deformation and the stress distribution in walls have been studied which is applicable to both the short and the tall buildings.
The behavior of the wall is divided into two parts: the symmetric flange action and the antisymmetric web action. The latter
has two parts: the web shear and the web bending. Appropriate stiffness equations have been derived for all the three actions.
These actions can be synthesized to solve any nonlineal cross-section. Two specific problems that of lateral and torsional
loadings of a rectangular box have been studied.
It is found that in short buildings shear lag plays a very important role. Any beam type formulation which either ignores
shear lag or includes it in an idealized form is likely to lead to erroneous results. On the other hand a rigidity type approach
with some modifications to the standard procedures would yield nearly accurate answers.
1. Introduction rate when the walls are indeed independent, but its
application to situations where various walls in a
Many nuclear power plant buildings, for example, nonlinear system are continuous at the vertical edges is
the auxiliary building, have reinforced concrete shear questionable. The beam approach avoids this problem
walls as the primary lateral load resisting system. Typi- by looking upon the building as a beam with nonlinear
cally, these walls have low height to length ratio, often cross-section. Chokshi and Lee [1] have modeled the
less than unity. It is well known that such walls exhibit buildings using the Timoshenko beam theory [3-5]. This
marked shear lag phenomenon, which would affect their approach allows for bending and shear deformations,
bending stiffness and the overall stress distribution in but does not address to the problem of shear lag.
the building. It is conceivable that these buildings could Hadjian and Atalik [6] have used Reissner's shear lag
be modeled using a finite element program in which formulation [7] to avoid the problem. Danay [8] uses a
each wall is discretized as several plate bending or plane "general" element with a parabolic stress distribution to
stress elements. Partly to avoid the cost of such a model the buildings. The problem with both the Reis-
numerical solution both in manpower and in comput- sner's and the Danay's formulations is that they presup-
ing, and also due to the fact that an elaborate finite pose the form o~f shear lag, which perhaps is adequate
element model may be an overkill considering all the for relatively taller buildings, but may not be so for
other uncertainties, resort is taken to simplified "stick" short buildings.
modeling techniques. These techniques typically use In the present paper the deformation and the stress
three degrees of freedom per floor - two horizontal distribution in wails have been studied which are appli-
translations and one rotation. Past investigators [1] have cable to both the short and the tall buildings. The
classified such techniques into two categories: the rigid- formulation for individual walls yields general stiffness
ity approach and the beam approach. In the rigidity relationships which can be integrated into a general
approach [2], the stiffness of each wall is calculated as purpose program. Specific behavioral aspects are also
an isolated wall. The stiffnesses of all the walls are studied which may lead to a simplified modeling similar
combined to form the story stiffness using simple trans- to that in ref. [2].
formation equations. This method is sufficiently accu-
20
i
% = S / 2 bt w ,
2b ~q
shown in fig. 3. The shear lag phenomenon is most
pronounced in a short wall, and is practically nonex- (b } WEB OR SHEARWALL
istent in a tall wall. Similar situation for the web wall is Fig. 2. Free body diagrams of walls in the box.
pictured in fig. 4. To study these stress states, a stress
function approach, in conjunction with a Fourier series
representation of the stress function if applied which is
discussed in detail in Appendix A.
T llll
Fig. 5 shows the flange wall with the forces F r on the
two edges which are given by
Ff = "rfhtf. (2)
F r = Kft). (3)
SHORT WALL
Fig. 1. Box structure subjected to lateral force. Fig. 3. Vertical stress distribution in the flange wall.
A.K. Gupta / Modeling of shear wall buiMings 71
6.0- /
,'/
,7
TALL WALL
5.0-
4.0-
± = 3_.0,~.~ /
o rf h ///
W
3.0-
111//
W
2D-
Y//
_'I
1.0.
SHORT WALL
o
Fig. 4. Vertical stress distribution in the web wall. o 0'.5 d.o ¢.s z'.o
ASPECT RATIO ( o/h )
Fig. 6, Variation of the flange stiffness ratio (1/rf) with the
This value changes for relatively short walls due to shear aspect ratio (a/h).
lag. The modified expression can be written as
Kf = rfKto. (5)
A plot between a/h and 1/rf is shown in fig. 6 which is based on the derivation in Appendix A. It is noted that
for small values of a/h (tall walls), 1/r r ~ 1 and for
short walls 1/rf ~- 3a/h which gives
-I
Kf = ~Et r for short walls. (6)
For practical application it is suggested that the value of
g f be taken as equal to smaller of the values given by
eqs. (4) and (6). Eq. (4) will govern for h >/3a, eq. (6)
I-
2b ~_]
l- c and the horizontal displacement t%, which are given
by
X~ ~2' = K ~ U , [wh = Kv, bUb" (7)
K,~.o = 2Etwb/3h,
Kwb = rwbKwb o,
Kwbo = Eb2tw/h 2.
(8)
(9)
The variations in the values of rw and r,~b in terms of
Fig. 7. The web wall.
b / h were obtained by a formulation similar to that
given for the flange wall in Appendix A, and are shown
in figs. 8 and 9. As expected, for tall walls, b/h --, O, r~
2.5~ and r,,,b --, 1. For short walls we have 1/r w -- b/h which
would give
Kw = ~£,w. (10)
As an approximation then, for practical application we
may use lower of the two values of K w given by eqs. (9)
and (10).
2.0-
The variation in l/rwb is more complex. As shown in
fig. 9, the value of l/rwb is zero at h / b = 0, it rises to
1.28 for h / b of 1.5, and then it starts going down until it
7 stabilizes at 1/rwb -- 1. A reasonable approximation ap-
T // pears to be l/rwb = 2 h / b for h / b < 0.5, and ]/rwb = l
beyond that. The substitution of this in eqs. (8) and (9)
1.5- leads to the following equation for short walls
0
(Z Kwh = Eb3tw/2h 3. (11)
/ /..,. ± = b_ Contrary to the observation in case of the stiffnesses K f
__.,,J__ll I - r~ h and K w which were lower for the short walls than would
be given by the corresponding expressions for the tall
1.0- wall, the value of K,,, b given by eq. (11) for short wall is
actually higher than that given by the tall wall expres-
sion in eq. (9). In fact, Kwh --, ~ as h ~ 0. In this case
therefore, for practical application, we should use higher
oi . Present Study
of the two values of Kwh obtained from eqs. (9) and
(11).
In addition to the edge shear bending the web wall is
subjected to a state of pure shear, as shown in fig. 2b,
which is accompanied with a horizontal shear displace-
-----Proposed Approximation
ment ush. We can write
1.4-
1.2-
I.O-
.0
2 I 2h
rwb b
0.8-
o_
(2:
¢q
(/)
tJJ
Z
t~.
0.6-
m
tJJ
0.4.
O.Z-
Present Study
- - - - - - Proposed Approximotion
O J
El = K w l v hs,
S 1 = KshlUshl, = KwblUbl , Fshl = 2a 1
h
F = F~hI +/'1, ul = UShl -- /'/bl' (22)
which give
Kwbl [ S1
°= x,~--;- t ~., - "')'
Fig. 11. Box structure subjected to torsional moment.
+ - - S 1 - K w b l u 1. (23)
"F = -~a Ksh 1
that the diaphragm is rigid in its own plane. The equi- Similarly for wall 2, we have
librium and the compatibility equations between the
walls and the diaphragm are given by
v = - ,rwb2/
K,~
- I 1'12 - K-2I.~)'
$2
T = 2 b S 1 + 2 a S 2, u 1 = b$, u z = aO, (21)
where T is the applied torsional moment and 0 the
corresponding rotation; S, and S 2 are the shear force
+-k-~ ) S2 - rwb2U~. (24)
and u~ and u 2 are the corresponding horizontal dis-
placements on the wails with length 2a and 2b respec- Using the same v and F for the two walls presumes
tively. The distribution of forces and the displacements compatibility and equilibrium at the vertical edges. Eqs.
on the two walls are shown in fig. 12. It is clear that all (23) and (24) with eq. (21) yield
the walls in this problem are behaving like the web wall
in the previous section. A1S1 + B l S 2 = CaO, A2S2 - B 2 S z = C20, (25)
I d l l l l c
2o
J I J l l
"1
"-- Fshl
!1i i i i i i i i i i i ii
Fshl
+ J J I I l I J J I l I J
I!F'
(O) FORCES AND DISPLACEMENTS FOR WALL I
!1............I!
~ sh2
J I J J a l l J l ~ ' I J q i i q t i r ! r i t i i i
r -
2b
(b) FORCES AND DISPLACEMENTS FOR WALL 2
Fig. 12. Free body diagrams of the walls.
in which For the purpose of comparison, we introduce here the
K stiffness of an open section in which the vertical edges
are assumed free with no compatibility and shear trans-
fer.
K,,,b Kwbza
C,=K+- 2ab(at, + bt,)E
WI K w2 ’
K open= (29)
h .
h Kwb2
B2=Zb+ Ksh2 ’ For the three idealizations, viz, the present formulation.
the closed section and the open section, the shear force
C, = K,,,b - K,,,a. distributions are given by.
Eqs. (21) and (25) give
T= K,O.
where
K
T
+ 2bKwb,
Kwb2 (KG +K:)
1
/
For the case of large h when shear lag can be ignored and (31)
eq. (26) gives
8a2b2t t E
K,,= h’2 (32)
Table 1
Comparison of shear force distribution and torsional stiffnesses, (S,/S,),,,,, = a/b, (S,/.S,),,, = 1, 1, = tZ
:,2b b,
rigorous and the resulting expressions, eqs. (26) and (30) The above observation is made on the basis of the
should give the most accurate results. The approximate studies performed here on the lateral load and torsional
expressions for Kwl and Kwh 1 give rise to three ranges behavior of a rectangular box section.
of h / a ratios, viz, when h / a <_0.5, 0.5 < h / a <_1.0
and when h / a > 1.0. Similarly, there are three ranges
for h / b ratios. Together, there are nine combinations of Appendix A. Analysis of the flange wall
these ranges. However, if we assume a > b, we can cut
down the number of combinations to six. For the pur- The coordinate system for the flange wall is shown in
pose of comparison, we assumed t 1 = t 2. Various values fig. 5. The boundary conditions are: at x = a, ox = 0 and
are summarized in table 1. %y = - ~-f; at y = 0, Oy = zxy = 0. The shear boundary
F r o m table 1, it is observed that the stiffness and the condition at x = a can be represented in terms of a
shear force distribution for the open section are rea- Fourier series:
sonably close to those calculated in the present study Txy = E Tm cos amy,
for h / a and h / b < 1.0. Therefore, as long as the lengths m~1,3 ....
of all the walls (2a or 2b) are greater than twice the
m~
height, the open section idealization is reasonably accu- r,, = ( - 1 ) °'+1)/2 47m a m= --. (A.1)
rate. Further, the closed section idealization essentially mlr ' 2h
governs the behavior for relatively tall buildings. In A stress function which satisfies the boundary condi-
between, however, none of the two idealizations give tions at y = 0 is
uniformly good results. An attempt was made to extend
the h / a / h / b _< 1 boundaries. Depending upon the = ~-~Fm sin amy. (g.2)
degree of accuracy desired, some relaxation may be
Substituting eq. (A.2) into the bihormonic equation of
possible, nothing very significant to report here.
the plane stress problem we get
F~iv - 2a,,F,i,
2 ii + 4
a m E n - O_, (A.3)
4, Conclusions
F m = A,~1 cosh amx + A,,EX sinh a,~x. (A.4)
The behavior of any wall can be studied in two
Applying the boundary conditions at x = a gives
parts: the symmetric part which constitutes the flange
action, and the anti-symmetric part constituting the web Am1 = - Cma sinh area, Am2 = Urn cosh otma,
action. The web action has two parts also - the web
shear and the web bending. Appropriate stiffness equa- Cm = ( - 1 ) t ' - 1 ) / 2 4~'f ( 2 a m a + s i n h 2 a , , a ) _ l . (A.5)
tions have been derived for all the three actions. These
actions can be synthesized to solve any nonlineal cross- Let us represent the vertical displacement anywhere in
section problem. The general formulation is presented the wall by o = E o m ( x ) cos amy. For zero Poisson's
in Appendix B. Two specific problems that of lateral ratio we can write vm = Fii/Ea. At the top corner of the
and torsional loadings of a box structure are presented vertical edge, x = a, y = 0,
in the preceding two sections.
In short buildings, shear lag plays a very important u = ~'~(1 + cosh 2ama)Cm/E. (A.6)
role. Any beam type formulation which either ignores
shear lag or includes it in an idealized form is likely to When h is large, eq. (A.6) gives
lead to erroneous results. Quite interestingly, on the
16~'fh2 ( - 1)t'~- ~)/2 ~'fh2 (A.7)
other hand, a rigidity type approach with some modifi-
v° = - - '3Ea
,7" E m3 2Ea "
cations to the standard procedures would yield nearly
accurate answers for short buildings, say when length of We denote the ratio of o0 and v by re which is given by
all the walls is at least twice the height. The suggested
modification is to concentrate effective areas of the 1 32 a ( - 1) t ' - 1 ~ / 2 1 + cosh 2area (A.8)
flange and the web walls at the ends of the web wall for rf ~r2 h ~ m2 2ama + sinh 2ot,,,a "
the purpose of calculating bending stresses in the flange
and the web walls, as explained in section 2. For calcu- The rf given above is the same rf used in eq. (5). Fig. 6 is
lating the building stiffness one only need include the based on eq. (A.8). The expressions for rW and rwb were
shear stiffness of the short walls. similarly derived and are not presented here.
78 A.K. Gupta / Modeling o["shear wall buildings
(B.2)
Based on the treatment in the text the following //////////r
force displacement relationships can be written o) FLANGE ACTION
Vwn Vwrl
Ffn = KfnVfn , Fwn = Kw.vw.,
I ¢
l
Fw n = g w b n U b n , an = KshnUshn. (B.3)
I I l l [ l / / l / l /
(b WEB BENDING ACTION
Sn
Uihn
Fy,Uy
F~, @ Ft.n
/ ///rz//~/, r
Fx ,U X (C) WEB SHEAR ACTION
Fig. 13. Cross-sectional plan of a shear wall building. Fig. 15. Various actions in a wall.
A.K. Gupta/ Modelingof shear wallbuildings 79
{F},~=[Kln(u)n, { F } . = [ S n F . a F n 2 ] T, { S ) = [Ks~o]( Uw },
where
h K + h Kshn h Kshn
[K]. = -- ~ shn (S.7)
h K
2a n shn 2 1 + 2 a . Kwbn 2 - - ~ + 1 + 2 a . Kwhn
The above 'stiffness matrix' is not symmetric because Note the [Ksso] matrix is symmetric. The { S } and the
the forces S., Fnl, Fn2 and the displacements u., V,1, v~2 (Uw } vectors can be rearranged, such that they first
are not exact pairs in the energy sense. have all the walls in X direction, then all the walls in Y
For the solution of the complete problem, each wall direction. Then eq. (B.12) can be rewritten as
' n ' can be viewed as a "finite element," which can be
assembled to form a global force displacement equation Xwxy U,~x
of the type
(!:/
(B.9a)
References [5] A,M. Ebner and D.P. Billington, Steady state vibrations of
damped Timoshenko beam, J. Struct. Div. ASCE 94 (ST3)
Ill N.C, Choksi and J.P. Lee, Shear coefficient and shear force (March 1968) 737-759.
distribution in nuclear power plant structures due to seismic [6] A,H. Hadjian and T.S. Atalik, Discrete modeling of sym-
loading, in: Proc. Internat, Symposium on Earthquake metric box-type structures, in: Proc. Internat. Symposium
Structural Engineering, St. Louis, Mo., August 1976, pp. on Earthquake Structural Engineering, St. Louis, Mo.,
629-642. August 1976, pp. 1151-1164.
[2] J.A. Blume, N.M. Newmark and L.H. Corning, Design of [7] E. Reissner, Analysis of shear lag in box beams by the
Muhistory Reinforced Concrete Buildings for Earthquake principle of minimum potential energy, Appl. Math. 1V (3)
Motions (PCA, 1961). (1946) 268-278.
[3] S. Timoshenko and J.N. Goodier, Theory of Elasticity [81 A. Danay, A general element for analysis of asymmetric
(McGraw Hill, New York, 1951). multistory buildings with varying cross-sections, Building
[4] G.R. Cowper, The shear coefficient in Timoshenko beam and Environment II (1976) 57-67.
theory, J. Appl. Mech. 33 (1966) 335-340.