Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

2001 - Interaction Surfaces of Reinforced-Concrete Sections in Biaxial Bending

Download as pdf or txt
Download as pdf or txt
You are on page 1of 7

INTERACTION SURFACES OF REINFORCED-CONCRETE SECTIONS

IN BIAXIAL BENDING

By A. Fafitis,1 Fellow, ASCE

ABSTRACT: A method for the computation of the interaction surface of reinforced-concrete sections subjected
to axial load and biaxial bending is developed. The method is analytically exact and it is based on using Green’s
theorem to transform the double equilibrium integrals into line integrals along the compressive perimeter of the
concrete section. For up to third-degree polynomial stress-strain relations for concrete, Gauss integration with
only three sample integration points yields exact results. This is the reason the method is computationally
efficient. Note that the concrete stress-strain relations recommended by the American Concrete Institute and by
Downloaded from ascelibrary.org by HARVARD UNIVERSITY on 06/23/14. Copyright ASCE. For personal use only; all rights reserved.

Eurocode 2 of the European Union both fall within the above limitation. The method is demonstrated by a
numerical example.

INTRODUCTION can be achieved by fictitious cuts, as indicated by sides 4-5


and 10-11 in Fig. 1.
Reinforced-concrete columns are subjected to biaxial bend- The main assumptions in the following analysis are that
ing as a result of their positions in a structure, their unsym- plane sections remain plane (Bernoulli’s assumption) and that
metrical geometry, and the nature of the external loading. Col- the tensile stress of the concrete is negligible.
umns at corners or with their principal directions different The stress-strain relationship of concrete is of the form
from those of the floor plan or columns resisting earthquake
or wind loads are usually subjected to biaxial bending in ad- ␴c = f¯ (ε ␺) (1)
dition to axial load.
The problem of biaxial bending has received some attention where ε ␺ = normal strain.
in the past (Furlong 1961; Brondum-Nielsen 1985; Hsu 1989; Because of Bernoulli’s assumption, the strain ε␺ can be ex-
Dundar 1990). However, in the majority of buildings, the bi- pressed in terms of ␺, the distance from the neutral axis
axial effect on columns is either ignored (e.g., by analyzing ε ␺ = εcu␺/c (2)
the lateral force resisting frames as plane frames) or accounted
for by simplifying procedures {e.g., the design of corner col- where εcu = maximum compressive strain in the concrete; and
umns described in the Uniform building code [International c = depth from the maximum compressive fiber to the neutral
Conference of Building Officials (ICBO) 1997]}. These sim- axis. From Fig. 1 it is seen that
plifications are usually conservative and they are, in most
cases, justified. The interaction of frames, for example, de- ␺ = y ⫹ c ⫺ max Y (3)
pends on the ability of reinforced-concrete elements (usually therefore the stress can be expressed
beams) to transfer torsion, which is limited because of the
great reduction of torsional stiffness after cracking. ␴c = f (␺) = f ( y ⫹ c ⫺ max Y ) = g( y) (4)
An additional reason for avoiding the biaxial bending in
design is the complexity and the computational intensity of The axial resultant Nc and the moment components Mxc and
the problem. The construction of the interaction diagram for Myc carried by concrete are

冕冕
uniaxial bending is much simpler than the construction of the
interaction surface for biaxial bending. The objective of this Nc = g( y) dA (5)
study is to develop an algorithm for the construction of an A

冕冕
interaction surface of an arbitrary section. The proposed
method is exact and computationally efficient. With a modern
personal computer, the construction of the surface takes sec- Mxc = yg( y) dA (6)
A
onds, thus allowing the designer to interact in real time by
making the appropriate changes and monitoring their effect on
the behavior and strength of the section. Myc = ⫺ 冕冕 A
xg( y) dA (7)

BIAXIAL BENDING PROBLEM where A = compressive area of the section enclosed by the
perimeter a, 2, 3, 4, 5, 6, 7, b, c, 10, 11, d, as shown in Fig.
Consider the reinforced-concrete section of Fig. 1, subjected 1. The negative sign indicates compression. The coordinate
to an axial force PZ and a bending moment with components system Oxy may be a centroidal or a plastic-centroidal system
MX and MY. To use Green’s theorem, which involves line in- with x parallel to the neutral axis, as shown in Fig. 1. The
tegration, the section must be defined by a line that is closed choice of the coordinate system is discussed later.
and reducible. For sections with voids (hollow sections), this The strain and stress distributions shown in Fig. 1, are the
1 result of the external axial force PZ and the bending moments
Civ. and Envir. Engrg. Dept., Arizona State Univ., Tempe, AZ 85287-
5306. MX and MY . The equilibrium of those external forces with the
Note. Associate Editor: Joseph Bracci. Discussion open until Decem- internal force resultants Nc, Mxc, and Myc can be written as
ber 1, 2001. To extend the closing date one month, a written request must follows:
be filed with the ASCE Manager of Journals. The manuscript for this
technical note was submitted for review and possible publication on Sep- PZ = Nc ⫹ Ns (8)
tember 5, 2000; revised February 12, 2001. This technical note is part of
the Journal of Structural Engineering, Vol. 127, No. 7, July, 2001. MX = (Mxc ⫹ Mxs)cos ␪ ⫹ (Myc ⫹ Mys)sin ␪ (9)
䉷ASCE, ISSN 0733-9445/01/0007-0840–0846/$8.00 ⫹ $.50 per page.
Technical Note No. 21285. MY = ⫺(Mxc ⫹ Mxs)sin ␪ ⫹ (Myc ⫹ Mys)cos ␪ (10)

840 / JOURNAL OF STRUCTURAL ENGINEERING / JULY 2001

J. Struct. Eng. 2001.127:840-846.


Downloaded from ascelibrary.org by HARVARD UNIVERSITY on 06/23/14. Copyright ASCE. For personal use only; all rights reserved.

FIG. 1. Arbitrary Reinforced-Concrete Section

where Ns, Mxs, and Mys = resultants of the steel stress; and ␪ (e.g., multicell boxes, bridge piers, shear walls with end col-
= angle from the positive direction of the neutral axis to the umns), and corner columns. The problem of integration be-
positive X-axis, anticlockwise. Positive direction of the neutral comes more complex in biaxial bending because the integra-
axis is defined as the direction that leaves the compressive tion area (compressive zone) is defined by the neutral axis,
fibers of the section to the left (Figs. 1 and 3). The case of which rotates and migrates continuously as Nz, Mx, and My
total tension is trivial (because only steel can take tension), vary.
and it is not considered in this study. The integration problem has been addressed by dividing the
The biaxial moment problem is thus reduced to determining section either in strips parallel to the current direction of the
the concrete strain εcu, depth to the neutral axis c, and direction neutral axis or in small rectangular elements. This brute-force
of the neutral axis ␪ that will satisfy the equilibrium equations solution has been employed successfully (Brondum-Nielsen
[(8)–(10)]. The problem is nonlinear because the concrete con- 1985; Hsu 1989; Tsao and Hsu 1993). However the division
stitutive law function g( y) is nonlinear and its tensile stress is of the section into small rectangular elements may require
neglected and also because the steel is a nonlinear material. A some complex computer code. Some elements at the borders
solution can be obtained by a trial-and-error procedure or some of the section will not be rectangular, and the calculation of
other approximate method [e.g., Newton-Raphson method their area and centroid introduces additional computational dif-
(Dundar and Sahin 1993)]. The usual procedure is to assume ficulties. There is also the question of the relationship between
a concrete strain εcu and an angle ␪, and by varying the depth required accuracy, size of the discretizing grid, and computa-
to neutral axis c, to satisfy (within some acceptable tolerance) tional effort.
the equilibrium equations. If for a given ␪ the equilibrium The way to overcome the computational difficulties of dou-
equations are not satisfied, a new ␪ is tried. ble integration, without compromising accuracy, is the trans-
Another approach is to construct the ‘‘interaction surface.’’ formation of the double integrals by the techniques of ele-
For a given geometry of the section and material constitutive mentary calculus. The method presented in this paper is an
laws and a fixed ultimate compressive strain εcu, the angle ␪ ‘‘exact integration method’’ in which the double integrals are
is incremented, and for each ␪, the depth to neutral axis c is transformed by Green’s theorem to single line integrals along
incremented. The points (Pz, Mx, and My) produced in this way those sides (or segments thereof) that are in compression. This
create a surface in the Pz-Mx-My space, which is the interaction method is general and can be used to calculate other double
surface corresponding to a given maximum concrete compres- integrals for section properties such as area moment and mo-
sive strain. All points inside this surface correspond to com- ment of inertia.
binations of axial force and bending moments that would not
produce compressive strain in the concrete exceeding the ul- GREEN’S THEOREM AND ITS APPLICATION TO
timate compressive strain. In this sense then, the points inside BIAXIAL BENDING
the interaction surface are ‘‘safe’’ points. The importance of
the interaction surfaces in design stems from the fact that they Green’s theorem transforms the double integration over an
define a safe space. The construction of interaction surfaces is area A into a line integration along the closed line L that en-
the objective of this study. closes the area. The derivation can be found in calculus text-
From (5)–(7), it is apparent that the calculation of the stress books (Kaplan 1959). It is formally stated as follows:

冕冕 冖 冖
resultants Nc, Mxc, and Myc involves the evaluation of double
integrals over the compressive zone of the cross section. The
(⭸Q /⭸x ⫺ ⭸P/⭸y) dx dy = P dx ⫹ Q dy (11)
evaluation of double integrals presents computational difficul- A L L
ties, especially when the integration area is of a complicated
shape. In concrete structures such sections are not uncommon where P and Q = two functions of x and y; A = area of inte-
JOURNAL OF STRUCTURAL ENGINEERING / JULY 2001 / 841

J. Struct. Eng. 2001.127:840-846.


gration; and L = closed curve that encloses the area A. If L is stitutive function g( y) is required. In the ACI code, the func-
polygonal, which is the case in most concrete sections, then tion g( y) is a zero-degree polynomial (constant stress block),
the line integrals consist of the summation of the integrals whereas in Eurocode 2, g( y) is a second-degree polynomial
along the straight sides of the polygon. The theorem can be and a constant. In both these most important concrete models,
applied to the biaxial problem, as it is formulated in (5)–(7), the highest degree of the integrant of (17) is 4; therefore, the
if one lets third-order Gauss quadrature employed here will yield the ex-
act value of the integration. This is true for any model that
P=0 (12) can be expressed in terms of polynomials of no higher than

Q=
1
r⫹1 冕 x r⫹1y sg( y) dx (13)
third degree [e.g., the models suggested by Hognestad et al.
(1955) and Kent and Park (1971)]. For models of higher de-
gree or for nonpolynomial models (Sargin 1971; Fafitis and
where r and s = nonnegative integers. With this definition of Shah 1985), the third-order quadrature will yield an approxi-
P and Q, (11) becomes mation of the integral of (17). The approximation can be im-

冕冕 冖
proved by employing higher-order quadratures.
Downloaded from ascelibrary.org by HARVARD UNIVERSITY on 06/23/14. Copyright ASCE. For personal use only; all rights reserved.

1
xry sg( y) dx dy = x r⫹1y sg( y) dy (14)
A r⫹1 L LINE INTEGRATION
The stress resultant R, of the concrete compressive zone (Nc The line integral of (17) is evaluated along the sides of the
or Mxc or Myc), is thus reduced to the line integral compressive zone. The sides are defined by the x,y-coordinates

R=
1
r⫹1 冖L
x r⫹1y sg( y) dy (15)
of the endpoints, as shown in Fig. 2.
The equation of side l is
x = ␣l ⫹ ␤l y (18)
This is the basic equation of the present formulation of the
biaxial problem. Depending on the values of r and s, the left- ␣l = xl ⫺ ␤l yl (19)
hand side of (15) represents
␤l = ( xl⫹1 ⫺ xl)/(yl⫹1 ⫺ yl) (20)
• The axial force Nc [(5)] for r = 0 and s = 0 where (xl, yl) and (xl⫹1, yl⫹1) are the starting and ending points,
• The bending moment Mxc [(6)] for r = 0 and s = 1 respectively, of side l.
• The bending moment ⫺Myc [(7)] for r = 1 and s = 0 [note With (18) the line integral in (17) becomes
the negative sign in (7)]

If the function g( y) = 1, then the left-hand side of (15)


represents
Sl = 冖
l
(␣l ⫹ ␤l y)r⫹1y sg( y) dy = 冖 l
Gl ( y) dy (21)

Gl ( y) = (␣l ⫹ ␤l y)r⫹1y sg( y) (22)


• The area A, for r = 0 and s = 0
• The area moment Sx, for r = 0 and s = 1 Note that, if side l is vertical, x is constant (x = xl = xl⫹1, ␣l
• The area moment Sy, for r = 1 and s = 0 = xl, and ␤l = 0), and (18)–(20) are not needed. If side l is
• The moment of inertia Ix, for r = 2 and s = 0 horizontal, then dy = 0 and the contribution of that side to the
• The moment of inertia Iy, for r = 0 and s = 2 integral is zero. This observation is used in the computer pro-
• The product of inertia Ixy, for r = 1 and s = 1 gram in this study. For a given direction ␪ of the neutral axis,
the section is rotated by ⫺␪ and the neutral axis becomes
The right-hand side of (15) is a line integration along the horizontal. The integration is carried over the segment of the
sides of the integration area A and can be written perimeter of the section that lies on the compressive zone,

冖 冘
which is above the horizontal neutral axis.
1 1 The numerical evaluation of (21) takes the form
x r⫹1y sg( y) dy = Sl (16)

冕 冘
r⫹1 L r⫹1 l Ng

where Sl = Gl ( y) dy ⬵ ( yl⫹1 ⫺ yl) wi Gl ( yi) (23)


l i=1

Sl = 冖l
x r⫹1y sg( y) dy (17) where Ng = order of the Gauss quadrature (in this study Ng =

is the integral along the l side of the closed polygon that en-
closes the compressive zone of the section.
If the constitutive law, which is expressed by the function
g(y), is integrable, then the integrals Sl could be explicit in
closed form. If the function g( y) is not integrable, then the
integrals Sl will be evaluated by a numerical scheme (e.g.,
Gauss quadrature). The common constitutive laws such as the
‘‘stress block’’ recommended by the American Concrete Insti-
tute (ACI) code (1995) or the ‘‘parabolic-rectangular’’ of Eu-
rocode 2 [Comité Euro-International du Béton-Fédération In-
ternational de la Précontrainte (CEB-FIP) 1993; Eibl 1995;
Becket and Alexandrou 1997] are integrable functions, and the
calculations based on (17) are ‘‘exact.’’ However, in writing
computer code, it is often more convenient to integrate nu-
merically avoiding integration formulas. The code written for
this investigation employed the third-order Gauss quadrature
(Alkhafaji and Tooley 1986). Only the definition of the con- FIG. 2. Definition of Side l

842 / JOURNAL OF STRUCTURAL ENGINEERING / JULY 2001

J. Struct. Eng. 2001.127:840-846.


3); wi = ith weight; and yi = sample point at which the function NUMERICAL EXAMPLE
G( y) is evaluated. Numerical values for wi and yi can be found
in numerical analysis textbooks (Alkhafaji and Tooley 1986). Calculate the axial force PZ and the bending moments MX
and MY of Fig. 3(a) for a neutral axis at ⫺45⬚ and a distance
from the plastic centroid equal to 9.972 in., as shown. The
STEEL CONTRIBUTION constitutive law of the concrete is assumed parabolic rectan-
The contribution of the steel does not present computational gular [Fig. 3(b)], and the steel is assumed elastoplastic with
difficulties. The steel bars are assumed discrete points with yielding stress fy = 60 ksi and elastic modulus Es = 30,000 ksi.
area Asj, coordinates xsj, ysj, and stress fsj. The total steel axial The section is rotated about its plastic centroid by 45⬚, as
force and bending moment resultants are shown in Fig. 4. The compressive zone is a, 1, 2, 3, b. The
line integration will be along the sides a-1, 1-2, 2-3, and 3-b.

冘 The coefficients ␣l and ␤l of (18) for each side are given in


Nb

Ns = Asj fsj (24) Table 1, together with the integration limits yl and yl⫹1.
j =1
Downloaded from ascelibrary.org by HARVARD UNIVERSITY on 06/23/14. Copyright ASCE. For personal use only; all rights reserved.

冘 Axial Force Nz (r = 0, s = 0)
Nb

Mxs = ysj Asj fsj (25)


j =1 The axial force NZ is the sum of Nc, the force carried by


Nb
concrete, and Ns, the force carried by steel. The axial force Nc,
carried by the compressive zone, is given by (15) for r = 0, s
Mys = ⫺ xsj Asj fsj (26) = 0, and g( y), as defined by (4) and depicted in Fig. 4(b). The
j =1
numerical evaluation of the integral of (15) requires the eval-
To avoid double counting of the concrete area that is dis- uation of Sl [(16) and (17)]. The Gauss points yi for each side
placed by the steel bars, the force Asj fcj is subtracted from the of the corresponding values of g( yi) are given in Table 2. With
bar force Asj fsj ( fcj being the concrete compressive stress at the these values, one calculates Gl (yi) by (22) and Sl by (23).
centroid of the bar). The detailed calculations for S2 for side 1-2 (Fig. 4) are (1)
At a strain larger than the breaking strain of the steel, the from Fig. 4(a) and Table 1, one has Yi = Y1 = ⫺3.725, Yi⫹1 =
bar is assumed broken with stress fsj = 0. Y2 = 17.488, ␣l = 14.142, and ␤l = ⫺1; and (2) from Table 2,
The origin O of the coordinate system of Fig. 1 coincides the first Gauss index is ␭l = 0.1127 and the first point on side
with the plastic centroid. The plastic centroid is the centroid 1-2 [Fig. 4(a)] is at a distance 0.0338L1-2 from the starting
of the section in its ‘‘plastic’’ state; that is, when all the steel point of side 1-2 (L1-2 is the length of the side). Therefore the
has yielded and the whole concrete section is under the max- y-coordinate of the first Gauss point is y1 = Yi ⫹ ␭1 (Yi⫹1 ⫺ Yi)
imum compressive stress (Wang and Salmon 1998). The ad- = ⫺3.725 ⫹ 0.1127(17.488 ⫹ 3.725) = ⫺1.334 in. The other
vantage of using the plastic centroid as the origin of the co- Gauss points, y2 and y3, are calculated in a similar way, and
ordinate system is that at the maximum axial compressive they are given in Table 2. By (23) one has
force the corresponding moment is zero. However if the stress-
strain law of the steel does not have a distinctive yielding S2 = (17.488 ⫹ 3.725)[0.278(14.142 ⫹ 1.334)(⫺3.39)
plateau (e.g., strain-hardening law), the plastic centroid may ⫹ 0.444(14.142 ⫺ 6.881)(⫺4.25)
not offer any advantage and the geometric centroid may be
used as the origin. ⫹ 0.278(14.142 ⫺ 15.097)(⫺4.25)] = ⫺575.93 kip

FIG. 3. Rectangular Reinforced-Concrete Section (Numerical Example)

JOURNAL OF STRUCTURAL ENGINEERING / JULY 2001 / 843

J. Struct. Eng. 2001.127:840-846.


Downloaded from ascelibrary.org by HARVARD UNIVERSITY on 06/23/14. Copyright ASCE. For personal use only; all rights reserved.

FIG. 4. Rectangular Section Rotated 45⬚ (Numerical Example)

TABLE 1. Side Coefficients and Integration Limit law for concrete, and elastoplastic law for steel (as defined in
Parameter a-1 1-2 2-3 3-b Fig. 3)—is
Yi ⫺9.952 ⫺3.725 17.488 3.346 PZ = Nc ⫹ Ns = ⫺1,780.25 ⫺ 217.04 = ⫺1,997.29 kip
Yi ⫹1 ⫺3.725 17.488 3.346 ⫺9.952
␣1 21.592 14.142 ⫺20.834 ⫺14.142 Bending Moments Mx (r = 0, s = 1) and My (r = 1, s = 0)
␤1 1 ⫺1 1 ⫺1
The bending moment carried by the concrete is given in
(14): the x-component Mxc, for r = 0 and s = 1, and the y-
Similarly for sides a-1, 2-3, and 3-b, one has S1 = ⫺144.03 component Myc, for r = 1 and s = 0. The values of g( y) at the
kip, S3 = ⫺625.82 kip, and S4 = ⫺434.47 kip. Gauss points of each side are given in Table 2, and they are
The axial force carried by the concrete is the same used for the axial force Nc.
For Mxc the calculations of Sl for each side are analogous to
Nc = (S1 ⫹ S2 ⫹ S3 ⫹ S4)/(r ⫹ 1) = ⫺144.03 ⫺ 575.93 the calculations for Nc . For side 1-2 (Fig. 4), one has, by (22)
⫺ 625.82 ⫺ 434.47 = ⫺1,780.25 kip and (23) and with r = 0 and s = 1

The calculation of the stress and force carried by the steel S2 = (17.488 ⫹ 3.725)[0.278(14.142 ⫹ 1.334)(⫺1.334)(⫺3.39)
is straightforward. The strain εsi at each bar is given in Table ⫹ 0.444(14.142 ⫺ 6.881)(6.881)(⫺4.25) ⫹ 0.278(14.142
3 and shown in Fig. 4(c). From the strain, one calculates the
corresponding stress and finally the force Fsi. Note that, from ⫺ 15.097)(15.097)(⫺4.25)] = ⫺1,255.89 kip-in.
the stress of each compressive bar, one subtracts the stress of
Similarly for sides a-1, 2-3, and 3-b, one has S1 = 827.55 kip-
the concrete at the location of that bar to avoid double count-
in., S3 = ⫺5,519.41 kip-in., and S4 = 305.81 kip-in.
ing the compressive concrete area displaced by the bar.
The bending moment Mxc carried by concrete is, by (14)
The total force Ns carried by the steel is
and (16)
Ns = 10.86 ⫺ 26.90 ⫺ 70.80 ⫺ 70.80 ⫺ 59.40 = ⫺217.04 kip Mxc = (S1 ⫹ S2 ⫹ S3 ⫹ S4)/(r ⫹ 1) = 827.55 ⫺ 1,225.89
The total force carried by the section for the given location ⫺ 5,519.41 ⫹ 305.81 = ⫺5,611.94 kip-in.
of the neutral axis (Fig. 4)—at a maximum compressive con-
crete strain εcu = ⫺0.0035, parabolic-rectangular stress-strain The moment Mxs carried by steel is calculated based on the

TABLE 2. Concrete Stress at Gauss Points

Gauss index Side a-1 Side 1-2 Side 2-3 Side 3-b
Gauss weight
i ␭i wi yi g( yi) yi g( yi) yi g( yi) yi g( yi)
1 0.1127 0.278 ⫺9.269 ⫺0.37 ⫺1.334 ⫺3.39 15.894 ⫺4.25 1.844 ⫺3.99
2 0.5000 0.444 ⫺6.849 ⫺1.52 6.881 ⫺4.25 10.417 ⫺4.25 ⫺3.314 ⫺2.84
3 0.8873 0.278 ⫺4.429 ⫺2.47 15.097 ⫺4.25 4.940 ⫺4.24 ⫺8.473 ⫺0.77

844 / JOURNAL OF STRUCTURAL ENGINEERING / JULY 2001

J. Struct. Eng. 2001.127:840-846.


TABLE 3. Reinforcing Steel The moment Mys carried by steel is calculated based on the
Area x y εs fs Fs force at each bar and its distance from the x-axis given in Table
i (in.2) (in.) (in.) (in./in.) (ksi) (kip) 4
1 1.27 3.725 ⫺12.210 0.00029 8.641 10.974 Mys = ⫺[10.86(3.725) ⫺ 26.90(12.210) ⫺ 70.80(⫺3.346)
2 1.27 12.210 ⫺3.725 ⫺0.00079 ⫺21.122 ⫺26.825
3 1.27 ⫺3.346 11.831 ⫺0.00278 ⫺55.750 ⫺70.802 ⫺ 70.80(⫺7.588) ⫺ 59.40(⫺11.831)] = ⫺1,188.89 kip-in.
4 1.27 ⫺7.588 7.588 ⫺0.00224 ⫺55.750 ⫺70.802
5 1.27 ⫺11.831 3.346 ⫺0.00170 ⫺46.732 ⫺59.350 The total y-component of the bending moment is
My = ⫺2,074.14 ⫺ 1,188.89 = ⫺3,263.03 kip-in.
force at each bar and its distance from the x-axis ( y-coordi-
nate), as shown in Table 3 Bending Moments in Original Coordinate System
Mxs = 10.86(⫺12.210) ⫺ 26.90(⫺3.725) ⫺ 70.80(11.831) (MX and MY)
Downloaded from ascelibrary.org by HARVARD UNIVERSITY on 06/23/14. Copyright ASCE. For personal use only; all rights reserved.

The moments MX and MY are calculated using the transfor-


⫺ 70.80(7.588) ⫺ 59.40(3.346) = ⫺1,606.01 kip-in.
mation formulas [(9) and (10)]
The total x-component of the bending moment is
MX = (⫺7,217.94)cos(45) ⫹ (⫺3,263.03)sin(45)
Mx = ⫺5,611.94 ⫺ 1,606.01 = ⫺7,217.95 kip-in.
For Myc, the calculations for Sl for each side are analogous. = ⫺7,411.72 kip-in.
For side 1-2 (Fig. 4), one has, by (22) and (23) and with r = MY = ⫺(⫺7,217.94)sin(45) ⫹ (⫺3,263.03)cos(45)
1 and s = 0
S2 = (17.488 ⫹ 3.725)[0.278(14.142 ⫹ 1.334)2(⫺3.39) = 2,796.55 kip-in.
The axial capacity PZ is the same in the two coordinate sys-
⫹ 0.444(14.142 ⫺ 6.881)2(⫺4.25)
tems.
⫹ 0.278(14.142 ⫺ 15.097)2(⫺4.25)] = 6,921.38 kip-in.
Similarly for sides a-1, 2-3, and 3-b, one has S1 = ⫺2,282.38 CONCLUSIONS
kip-in., S3 = 7,513.42 kip-in., and S4 = 5,838.62 kip-in. Note The presented method for the construction of interaction
that the negative sign in the calculations for Sl above is the surfaces of concrete sections is computationally efficient be-
result of the adopted sign convention for My, as indicated in cause the double integration is reduced to the numerical eval-
(7). uation of a few line integrals. The number of line integrations
The bending moment Myc carried by concrete is, by (14) is equal to the number of the sides of the section in the com-
and (16) pressive zone. Each of these integrals involves typically three
Myc = ⫺(S1 ⫹ S2 ⫹ S3 ⫹ S4)/(r ⫹ 1) sample points; thus, the number of computer operations is very
small.
= ⫺(⫺2,282.38 ⫺ 6,921.38 ⫹ 7,513.42 ⫹ 5,838.60)/2 The construction, by a Pentium 400 MHz processor, of the
isobars (i.e., Mx-My curves corresponding to constant axial
= ⫺2,074.14 kip-in. forces) of the corner shear wall shown in Fig. 5 took about 3

FIG. 5. Corner Shear Wall

JOURNAL OF STRUCTURAL ENGINEERING / JULY 2001 / 845

J. Struct. Eng. 2001.127:840-846.


Downloaded from ascelibrary.org by HARVARD UNIVERSITY on 06/23/14. Copyright ASCE. For personal use only; all rights reserved.

FIG. 6. Mx-My Isobar Diagrams for Corner Shear Wall Using ACI and Eurocode 2 Models for Concrete

s each. Three of these isobars are shown in Fig. 6. Each one members subjected to biaxial bending and axial load.’’ Comp. and
was constructed according to the ACI and Eurocode 2 models. Struct., 49(4), 643–662.
Eibl, J. (1995). Concrete structures Euro-design handbook, Ernst & Sohn,
It is clear that the computation time would allow the interac- Frankfurt, Germany.
tive use of the method as a design tool. Fafitis, A., and Shah, S. P. (1985). ‘‘Predictions of ultimate behavior of con-
The method can be extended to prestressed concrete sec- fined columns subjected to large deformations.’’ ACI J., 82(4), 423–433.
tions by including the stress-strain law of the cable in addition Furlong, R. (1961). ‘‘Ultimate strength of square columns under biaxially
to that of the conventional reinforcement. eccentric loads.’’ ACI J., 57, 1129–1140.
Hognestad, E., Hanson, N. W., and McHenry, D. (1955). ‘‘Concrete stress
distribution in ultimate stress design.’’ ACI J., 27(4), 455–479.
REFERENCES Hsu, T. C. (1989). ‘‘T-shaped reinforced concrete members under biaxial
bending and axial compression.’’ ACI J., 86(4), 460–468.
Alkhafaji, A. W., and Tooley, J. R. (1986). Numerical methods in engi- International Conference of Building Officials (ICBO). (1997). ‘‘Struc-
neering practice, Holt, Rinehart and Winston, New York. tural engineering design provisions.’’ Uniform building code, Vol. 2,
American Concrete Institute (ACI). (1995). ‘‘Building code requirements Whittier, Calif.
for structural concrete.’’ ACI 318-95, Detroit. Kaplan, W. (1959). Advanced calculus, Addison-Wesley, Reading, Mass.
Becket, D., and Alexandrou, A. (1997). Introduction to Eurocode 2, E & Kent, D. C., and Park, R. (1971). ‘‘Flexural members with confined con-
FN Spon, London. crete.’’ J. Struct. Div., ASCE, 97(7), 1969–1990.
Brondum-Nielsen, T. (1985). ‘‘Ultimate flexural capacity of cracked po- Sargin, M. (1971). ‘‘Stress-strain relationships for concrete and the anal-
lygonal concrete sections under biaxial bending.’’ ACI J., 82(7), 863– ysis of structural concrete sections.’’ Study No. 4, Solids Mech. Div.,
869. University of Waterloo, Waterloo, Ont., Canada.
Comité Euro-International du Béton-Fédération International de la Pré- Tsao, W. H., and Hsu, T. C. (1993). ‘‘A nonlinear computer analysis of
contrainte (CEB-FIP). (1993). Model code, Thomas Telfold, London. biaxially loaded L-shaped slender reinforced concrete columns.’’ Comp.
Dundar, C. (1990). ‘‘Concrete box sections under biaxial bending and and Struct., 49, 579–588.
axial load.’’ J. Struct. Engrg., ASCE, 116(3), 860–865. Wang, C., and Salmon, C. G. (1998). Reinforced concrete design, Addi-
Dundar, C., and Sahin, B. (1993). ‘‘Arbitrarily shaped reinforced concrete son-Wesley, Reading, Mass.

846 / JOURNAL OF STRUCTURAL ENGINEERING / JULY 2001

J. Struct. Eng. 2001.127:840-846.

You might also like