Engineering Structures 48 (2013) 3742

Contents lists available at SciVerse ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Nonlinear analysis of slender high-strength R/C columns under

combined biaxial bending and axial compression
M. Bouchaboub , M.L. Samai
Department of Civil Engineering, Faculty of Sciences, University Mentouri of Constantine, Algeria

a r t i c l e i n f o a b s t r a c t

Article history: A numerical technique for the analysis of high-strength reinforced concrete slender columns subjected to
Received 2 March 2012 end loadings is presented. The nite difference method is applied to calculate the load corresponding to a
Revised 30 August 2012 specied deection, considering both material and geometric nonlinearities. In this way, the complete
Accepted 31 August 2012
loaddeection curve was computed for X columns that were also tested experimentally. The analysis
Available online 21 November 2012
procedure is applicable for columns with hinged extremities, although only isolated columns are evalu-
ated herein.
Keywords:
2012 Elsevier Ltd. All rights reserved.
Slender
Columns
High strength
Concrete
Biaxial

1. Introduction 2. General description

Reinforced concrete columns are generally subjected to eccen- The column under consideration is a rectangular cross-section
tric compression as a result of their location in the structure, their reinforced concrete column made using high-strength concrete.
cross-section or the type of forces they bear. Many columns are The system is subjected to an axial compressive force and biaxial
subjected to this kind of loads, for example the corner columns bending (Fig. 1). The column is divided into a number of segments.
of a building or the piles of a bridge. Fig. 1 schematically represents In this analysis, the following assumptions are made:
the axial force and moments applied to a column bearing this type
of load.  Plane sections before bending remain plane after bending.
The fact that concrete has become more resistant during the last  Lateral deections of the column are small compared to the
few years as a result of the development made in the quality of length.
materials should also be taken into consideration. For instance,  The cross-section properties are preserved in subdividing the
when HSC is used to reduce the dimensions of the cross-section section into nite elements.
of columns, their slenderness is increased by comparison with col-  Creep and shrinkage deformations are excluded.
umns in normal-strength concrete of similar length and supposed  Shear and torsional deformations are neglected.
carrying capacity which increases the possibility that the ultimate
load of HSC columns may sometimes be governed by instability 3. Stressstrain relationships of materials
rather than by strength.
This paper concentrates on the carrying capacity of slender, 3.1. Concrete
high-strength reinforced concrete columns under shod-term load-
ing, with hinged extremities. Concrete is assumed to carry no tensile stress in tension (Fig. 2).
Columns made with conventional concrete were also studied For concrete in compression, the stressstrain law for HSC given in
for comparison. the FIB textbook on Structural Concrete [1] has been used.
The rcec relationship may be approximated by the following
function: for |ec| < |ec,lim|
 2
Eci ec ec
Ec1 ec1  ec1
Corresponding author. Address: Cite 100 logts no. 59, Ain mlila 04300, Algeria. rc 
Eci

ec
fcm 1:a
1 Ec1
2 ec1
E-mail address: [email protected] (M. Bouchaboub).

0141-0296/$ - see front matter 2012 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.engstruct.2012.08.030
38 M. Bouchaboub, M.L. Samai / Engineering Structures 48 (2013) 3742

My

Mx

My Fig. 3. Generalized stresses and generalized strains and discretization of the cross-
Mx section.

Fig. 1. Axial compression and biaxial bending moments.

and generalized strain vector in an incremental form. For practical
applications and for nite changes of F and D:
DF Q  DD 3
The elements of the matrix Q are formulated by discretizing the
cross-section into nite elements, calculating the axial force and
the resisting moments of the individual elements and summing
them up to form the elements of the matrix, as explained later.
In Fig. 3, the shape of the used mesh is shown. The optimal
mesh size is obtained after a study of the differences observed
for 112 simulations.
The centroidal strain of the cross-section and the two curva-
Fig. 2. Stressstrain relationships of concrete.
tures are rst assumed. The values obtained from the previous cy-
cle could be reasonably used as initial proposed values. With the
aid of the assumption of linear strain distribution, the centroidal
where Eci is the tangent modulus, rc the compression stress, ec the strain of each nite area can be calculated as follows:
compression strain, ec1 = 0.0022, Ec1 = fcm/0.0022 secant modulus
e e0 y/x x/y 4
from the origin to the peak. fcm is the mean value of compressive
strength, it may be estimated from: where the origin of the coordinates is taken at the center of the
cross-section, Fig. 3. The stresses of different nite areas corre-
fcm fck Df 1:b
sponding to the strains are calculated from the materials stress
where Df = 8 MPa. fck is the characteristic compressive strength of strain curves. The tangent stiffness moduli are calculated for each
cylinders tested at the age of 28 days. nite area. The generalized stresses of the section are then calcu-
lated as follows:
Z
3.2. Steel
Ni r  dA
The stressstrain relationship for steel is assumed to be elastic Z
perfectly plastic in both compression and tension with yield point M xi r  y  dA 5
fy. The value of fy should correspond to a 0.2% offset in the charac- Z
teristic re diagram. M yi r  x  dA

Comparing the internal forces, Ni, Mxi, and Myi with the external
4. Momentcurvaturethrust relationships
forces, N, Mx and My, the difference force vector between the two
sets of forces, can be calculated:
The procedure applied to develop the momentcurvature
8 98 9
thrust relationship is the cross-section tangent stiffness matrix ap-
< DN >
> < N  Ni >
=> =
proach for biaxial bending [2,3]. For a biaxially loaded section, we dF DMx Mx  Mxi 6
can dene the generalized stress vector, F, and the generalized >
: >>
;: >
;
DM y My  Myi
strain vector, D, as follows:
8 9 8 9 If the calculated difference force vector is within the allowable tol-
< N >
> = < e0 >
> = erance, (jdFj 6 103 F), a solution has been obtained. Otherwise, an
F Mx D /x 2 iterative procedure would be required, in which the tangent stiff-
>
: >
; >
: > ;
My /y ness matrix of the cross-section, Q, has to be developed.
0 1
where N is the axial force on the section, Mx, and My are the mo- Q 11 Q 12 Q 13
ments about the X and Y coordinate axes, respectively, e0, is the B C
Q @ Q 21 Q 22 Q 23 A 7
centroidal strain and /x and /y are the curvatures about the X and Q 31 Q 32 Q 33
Y coordinate axes, respectively. Due to material nonlinearity, it is
useful to express the relationship between generalized stress vector The elements of the matrix Q are formulated as follows:
 M. Bouchaboub, M.L. Samai / Engineering Structures 48 (2013) 3742 39

8 9 8 9
X
m X
< De >
n
Q 11 aij  Etij 8
> = < DN >
> =
i1 j1
dF D/x Q 1  DM x 17
>
: > > >
X
m X
n D/y ; :
DM y
;
Q 12 aij  yij  Et ij 9
i1 j1 Then,
X
m X
n 8 9
>
< e De >
=
Q 13 aij  xij  Etij 10
i1 j1 D D DD /x D/x 18
>
: >
;
Q 21 Q 12 11 /y D/y
Xm X n
The tangent stiffness matrix is updated for the current strains and
Q 22 aij  y2ij  Et ij 12
i1 j1
the previous procedure is repeated until the difference in force vec-
X
m X
n tor is within the allowable specied tolerance. In this way, the mo-
Q 23 aij  xij  yij  Etij 13 mentcurvaturethrust relations can be developed.
i1 j1 In Fig. 4, the obtained momentcurvaturethrust relations are
Q 31 Q 13 14 plotted for constant axial load with different values of moment
about the y-direction.
Q 32 Q 23 15
Xm X n
Q 33 aij  x2ij  Etij 16 5. Method of analysis
i1 j1
One of methods used for the design of columns is the nite dif-
where Et is the tangent modulus, m and n refer to the number of -
ference method [4].
nite areas in the X and Y directions, respectively, and i and j refer to
This method is illustrated for the case of the two hinged col-
the location of the nite area in the X and Y directions, respectively.
umns (Fig. 5). The column length, is divided into a nite number
The correction values of the generalized strains are calculated
of elements, each with a length Dx. In an arbitrary section the
and then added to the original generalized strain vector to give up-
deection is yi. the curvature, /i, at this section may be approxi-
dated values of strains which in turn are used to calculate the cor-
mately calculated as follows:
responding stresses. The generalized strain correction vector can
be calculated as follows: yi1  2yi yi1
y00i  19
Dx2

Fig. 4. Examples of momentcurvaturethrust relationships.

40 M. Bouchaboub, M.L. Samai / Engineering Structures 48 (2013) 3742

with y00i /i .
This equation denes the geometrical aspect of the problem. To
nd out if an equilibrium state is possible for a given value of the
axial load, N, and a given eccentricity, e, an iteration process can be
started by assuming a reasonable value for the deection, y0, at
midspan. In addition to the axial force, N, the total moment in this
section:
M 0 Ne y0 20

The curvature, /i, can be obtained from the momentcurvature

thrust relationships. So it is now possible to calculate the deection,
y1, in the next section. The total moment there becomes:
M 1 Ne y1 21

And from the momentcurvaturethrust relationships, the magni-

tude of /1 can be obtained. In general it follows from Eq. (19) that:

yi1 2y1  yi1  /i Dx2 22

Using this equation it is possible to calculate the deection in each

section from the results of earlier calculated deection. Thus the
deection at the ends of the column is determined.
However, this has to be zero; if it is not, the calculation is re-
peated, introducing new, more suitable values of the deection,
y0, at midspan until the boundary condition is fullled. In this
way only one equilibrium state can be found, this representing
one point on the load deection curve (Fig. 6).
To estimate the ultimate load using this method it is clear that
several equilibrium states have to be calculated, for increasingly
larger values of N.

6. Method verication

Based on the numerical procedure discussed above, a computer Fig. 6. Flow chart for the present computer program.
program was developed to provide numerical results. The compu-
tation time is not signicant. In fact, the completion of the entire
program took almost 03 min. Comparisons between the results obtained from the computer
program with available test data are made here.
Numerous papers are available in the literature related to the
behavior of columns made of conventional concrete. Some of them
are the tests performed by Drysdale and Huggins [5], Ramamurthy
[6], Wu and Huggins [7], Wang and Hsu [8], Tsao and Hsu [9], Hsu
et al. [10], and Kim and Lee [11]. Nevertheless, there are few
authors who have proposed tests on high strength concrete col-
umns [1216]. Most authors have studied the behavior of high
strength concrete subjected to uniaxial bending, whilst tests on
this type of columns subjected to biaxial bending are very scarce.

Table 1
comparison between analytical results and columns tested by Germain and Espion
[15].

Column Length Eccentricity Stirrup Cylinder Failure load (KN)

designation (m) (mm) diameter strength
Exp Present Pana/
(mm) (MPa)
analysis Pexp (%)
A-1/36-R 3.78 5 8 86.9 1750 1770 101
A-1/36-O 3.78 5 8 87.3 1640 1620 99
A-1/18-R1 3.78 10 6 92.0 1922 1900 99
A-1/18-Q 3.78 10 8 89.3 1524 1450 95
A-1/18-O 3.78 10 6 91.4 1489 1430 96
A-1/18-R2 3.78 10 8 85.9 1479 1530 103
A-1/12-O 3.78 15 6 89.4 1256 1310 104
A-1/9-R 3.78 20 6 91.8 1140 1150 101
A-1/9-O 3.78 20 6 94.3 1100 1090 99
B-1/90-O 4.38 2 6 92.0 1388 1420 102
B-1/36-O 4.38 5 6 93.1 1315 1340 102
B-1/18-O 4.38 10 6 91.3 1153 1150 99

Fig. 5. Finite difference method: hinged column. fy = 542 MPa.

 M. Bouchaboub, M.L. Samai / Engineering Structures 48 (2013) 3742 41

Fig. 7. Loaddeection curves.

Olivier Germain and Espion [15] have tested 12 slender, two- Table 2
hinged, HSC columns. The cross-section was square 180 mm by Comparison between analytical results and columns tested by Drysdale and Huggins
180 mm and reinforced by four bars, each with a diameter of [5].
l2 mm (with yield strength at 542 MPa). The distance from the cen- Column ex ey Cylinder Failure load (KN)
ter of the reinforcing bars to the concrete surface was about (mm) (mm) strength
Exp Present Pana/Pexp
37 mm, Stirrups with a diameter of 6 or 8 mm. (MPa)
analysis (%)
In Table 1, the details of the columns tested by Olivier Germain
A-1-C 17.96 17.96 26.83 167.31 165 98
and Espion are described, and a comparison between the analytical A-1-C 17.96 17.96 26.83 167.31 165 98
results and the experimental data is given in Fig. 7. A-3-C 17.96 17.96 27.73 171.32 167 97
The last values of strains in test curves has never been reached C-2-A 23.47 9.72 26.69 175.77 171 97
C-3-A 23.47 9.72 28.42 185.11 178 96
in the numerical modeling of the structural behavior of these col-
D-1-A 25.4 0.0 30.35 173.10 190 109
umns. It is due to the difculty of the convergence of the analysis in D-2-C 25.4 0.0 29.18 176.66 188 106
the post-peak region. E-1-C 35.2 14.58 26.80 145.06 130 90
The range of difference in the load carrying capacity of the col- E-2-A 35.2 14.58 30.35 149.96 141 94
umn between the two sets of results is +4% to 5%, which is very F-1-A 11.73 4.86 28.49 261.20 253 97
F-2-A 11.73 4.86 29.04 253.64 255 99
good.
A comparisons have been made with experimental results ob- fy = 387 MPa.
tained by Drysdale and Huggins [5]. Descriptions of the columns
made of conventional concrete tested are shown in Fig. 8. The com-
parison with Drysdale and Huggins is given in Table 2. The analyt- Table 3
ical results have shown very good correlation with experiments. Comparison between analytical results and columns tested by Pallars et al. [12].
In order to ascertain the validity of the method, comparisons Test ex ey Cylinder Failure load (KN)
have been made with experimental results obtained by Pallars (mm) (mm) strength
Exp Present Pana/Pexp
et al. [13] who have tested slender HSC columns in bi-axial bend- (MPa)
analysis (%)
ing combined with axial compression.
k30_a0.5_n1 17.88 4.47 107 552.77 520 94
The cross-section was rectangular 200 mm by 100 mm. Table 3 k30_a1_n1 14.14 7.07 104 496.71 490 98
presents the tested columns. The terminology used to refer to each k30_a1_n2 28.28 14.14 108 342.02 320 94
test is shown as kx_an_nN, where kx: x indicates the slenderness k30_a2_n1 8.94 8.94 100 456.97 570 124
of the specimen. an: n is the ratio between relative eccentricities fy = 558 Mpa, L = 3000 mm.
(from 0 to 1). nN: N is a non-dimensional applied eccentricity.

The analytical results have shown a good correlation with

experiments (Fig. 9).

7. Conclusions

The nite difference method is utilized here to analyze slender

columns under biaxial bending combined with axial compression.
Comparison between the test data and the theoretical failure load
shows the advantage of using this method in predicting the
strength and the behavior of biaxially loaded high-strength rein-
Fig. 8. Cross-section of columns tested experimentally by Drysdale and Huggins. forced concrete slender columns.
42 M. Bouchaboub, M.L. Samai / Engineering Structures 48 (2013) 3742

Fig. 9. Loaddeection curves for column k30_a0_n2 and k20_a0_n4.

The method was also applied for columns made of conventional [5] Drysdale RG, Huggins MW. Sustained biaxial load on slender concrete
columns. J Struct Div, Proc ASCE 1971;97(5):142343.
concrete and the analytical results have shown very good correla-
[6] Ramamurthy LN. Investigation of the ultimate strength of square and
tion with experiments. rectangular columns under biaxially eccentric loads. In: Proceedings:
The general trend of the method is to be conservative. Only few reinforced concrete columns; 1966. p. 26398.
points were on the other side (unconservative). [7] Wu H, Huggins MW. Size and sustained load effects in concrete columns. J
Struct Div 1977;103(ST3):493506.
HSC concrete may be brittle in pure compression, but the failure [8] Wang GG, Hsu CT. Complete biaxial loaddeformation behavior of RC columns.
of slender columns (whether HSC or not) is strongly affected by J Struct Eng 1992;118(9):2590609.
second-order bending effects that induce a kind of structural duc- [9] Tsao WH, Hsu CTT. Behaviour of biaxially loaded square and L-shaped slender
reinforced concrete column. Mag Concrete Res 1994;46(169):25767.
tility before material failure, even with the smallest eccentricities. [10] Hsu CTT, Hsu LSM, Tsao WH. Biaxially loaded slender high-strength reinforced
The maximum carrying capacity always corresponded to a sta- concrete columns with and without steel bre. Mag Concrete Res
bility failure, rather far from the material failure. 1995;47(173):299310.
[11] Kim JK, Lee SS. The behaviour of reinforced concrete columns subjected to
axial force and biaxial bending. Eng Struct 2000;23:151828.
[12] Pallars L, Bonet JL, Miguel PF, Fernandez Prada MA. Experimental research on
References high strength concrete slender columns subjected to compression and biaxial
bending forces. Eng Struct 2008;30:187994.
[1] Federation international du bton Structural Concrete. Textbook on behaviour [13] Claeson C, Gylltoft K. Slender concrete columns subjected to sustained and
design and performance. Updated knowledge of the CEB-FIP Model Code 1990, short-term eccentric loading. ACI Struct J 2000;97(1):4552.
vol. 1. Lausanne; 1999. p. 2930. [14] Sarker PK, Adolphus S, Patterson S, Rangan BV. High-strength concrete
[2] Chen WF, Shoraka MT. Tangent stiffness method for biaxial bending of columns under biaxial bending. Recent advances in concrete technology;
reinforced concrete columns, vol. 35-I. IABSE Publications; 1975. 2000. p. 21734 [Special Publication 200-14].
[3] El-Metwally SE, El Shahhat AM, chen wf. 3-D nonlinear analysis of R/C slender [15] Germain O, Espion B. Slender high-strength RC columns under eccentric
columns. Comput Struct 1990;37(5):86372. compression. Mag Concrete Res 2005;57(6):36170.
[4] COMITE Euro-international du Bton. CEB-FIP manual of buckling and [16] Lee JH, Son HS. Failure and strength of high-strength concrete columns
instability. Lancaster: The Construction Press; 1978. subjected to eccentric loads. ACI Struct J 2000;97(1):7585.