Truss Modeling of Concrete Columns in Compression

Panos D. Kiousis, A.M.ASCE1; P. G. Papadopoulos2; and H. Xenidis3

Abstract: A truss-based simulation of reinforced concrete columns under compression is presented in this study. The concrete truss
elements are modeled based on advanced constitutive equations that take into consideration confinement dependent hardening followed by
Downloaded from ascelibrary.org by Guru Nanak Dev Engineering College on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.

softening. The computational advantages, as well as the challenges of this approach, are discussed. The effects of confining steel content
and spacing on the failure mechanisms of concrete columns under compression are examined using a parametric analysis. This analysis
verifies the well-established trends that strength and ductility increase with increasing transverse steel content and decreasing tie spacing.
It is also found that improvements on strength and ductility have limits. Finally, it is found that the amount and spacing of transverse
reinforcement influences the type of failure, which can be one of shear or concrete crushing.
DOI: 10.1061/共ASCE兲EM.1943-7889.0000142
CE Database subject headings: Concrete columns; Trusses; Plasticity; Ductility; Strength; Compression.
Author keywords: Concrete; Truss; Plasticity; Column; Ductility; Strength.

Introduction simulation of concrete behavior is certainly not novel 共Bažant

1997, Papadopoulos and Xenidis 1999; Fraternali et al. 2002;
The advent of the finite element theory and computer technology Salem 2004; Li and Tran 2008兲. Truss-based analyses, known as
has enabled sophisticated and accurate analyzes of many difficult “strut-and-ties,” are main-stream computational tools for concrete
structural problems. Especially important are the solution meth- design, especially in zones of stress-concentrations. However, the
ods that have been developed for nonlinear problems where ana- strut-and-tie approach aims for simplicity and examines ultimate
lytical solutions are difficult to obtain 共Stevens et al. 1991; Fafitis equilibrium of imposed failure configurations of concrete struc-
and Won 1994; Kwon and Spacone 2002; Zhang et al. 1994; Liu tural elements. The approach presented here is more extensive
and Foster 2000; Zhao et al. 2004兲. Material nonlinearity in these and aims to provide a more accurate research tool to recreate or
predict the path to failure, including a complete load-deformation
problems is often modeled using the incremental theory of plas-
record as well as the development of damage mechanism that
ticity 共Chen 1982兲 combined with special integration techniques
leads to failure. The main contribution of this study is the presen-
over the complex loading paths that are being developed within
tation of a theoretical and computational tool to examine the ef-
the mass of the structure 共Ortiz and Popov 1985; Ortiz and Simo
fects of discrete passive confinement to concrete columns
1986兲. The solution of concrete problems is often attempted using
subjected to compressive forces. The analysis accounts for the
this approach with the added difficulties associated with damage nonlinear response of concrete, including the effects of confine-
and softening stress-strain relations, which can result in math- ment to the hardening, softening, and failure characteristics of the
ematical and numerical instabilities 共Bazant 1976, 1984兲. Tech- material. This approach leads to solutions that are computation-
niques to resolve these issues have been presented with various ally stable, and captures localized failures due to crashing or
levels of success 共De Borst et al. 1993; Vardoulakis and Aifantis shearing of the material. The simulation is two-dimensional and
1991; Liu et al. 2001兲. its original intent was to observe qualitatively the response of
In this study, the behavior of concrete columns is examined concrete columns to loads. However, it is shown here that the
based on a truss-model simulation. Significant practical advan- quantitative performance of the simulation is also acceptable.
tages of this approach include: 共1兲 one-dimensional elements are
used, which are simpler conceptually and computationally and 共2兲
one-dimensional nonlinearity can be modeled using the secant Simulation of the Concrete Column
modulus of the total stress-strain relation, which, even in the case
of softening behavior, maintains a positive stiffness. Truss-based The method of analysis that is presented here addresses the prob-
lem of a short column under pure compression 关Fig. 1共a兲兴. The
1
Associate Professor, Div. of Engineering, Colorado School of Mines, symmetry of the problem allows the examination of one module
Golden, CO 80401 共corresponding author兲. of width equal to b / 2 and height equal to s, where b = column
2
Assistant Professor, Dept. of Civil Engineering, Aristotle Univ. of width and s = tie spacing. The boundary conditions for the module
Thessaloniki, Greece. are presented in Fig. 1共b兲. It is noted here that for short columns
3
Associate Professor, Dept. of Civil Engineering, Aristotle Univ. of subjected to uniform compression, the transverse reinforcement
Thessaloniki, Greece.
impacts directly the response of concrete within the column core.
Note. This manuscript was submitted on June 15, 2009; approved on
January 12, 2010; published online on January 25, 2010. Discussion pe- On the other hand, the axial reinforcement impacts the column
riod open until January 1, 2011; separate discussions must be submitted capacity and ductility mainly by direct superposition. Thus, in the
for individual papers. This paper is part of the Journal of Engineering analysis presented here, the axial reinforcement is omitted to fa-
Mechanics, Vol. 136, No. 8, August 1, 2010. ©ASCE, ISSN 0733-9399/ cilitate a clear demonstration of the effects of confinement on the
2010/8-1006–1014/$25.00. response of concrete. The effects of the axial reinforcement are

1006 / JOURNAL OF ENGINEERING MECHANICS © ASCE / AUGUST 2010

J. Eng. Mech., 2010, 136(8): 1006-1014

 再冎 冤 冥再 冎
1
0
␧x a ux
=2 共2兲
␧y 1 uy
0
b

where the element dimensions a and b are defined in Fig. 2.

The stress-nodal force relations are

再冎 冤 冥再 冎 再 冎 冋 册再 冎
1
0
␴x 2 b Fx Fx d b 0 ␴x
共3兲
Downloaded from ascelibrary.org by Guru Nanak Dev Engineering College on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.

= or =
␴y d 1 Fy Fy 2 0 a ␴y
0
a

where d = element thickness 共Fig. 2兲.

Substituting Eqs. 共2兲 and 共3兲 into Eq. 共1兲 results in

再 冎 冋 册 冋 册 冤 冥再 冎 再 冎
1
0
Fx d b 0 E 1 ␯ b ux Fx
= 2 ⇒
Fy 2 0 a 1 − ␯2 ␯ 1 1 uy Fy
0
a

冤 冥再 冎
Fig. 1. Simplification of the column to a module between two ties b
␯
Ed a ux
= 共4兲
considered in two examples where the predictive capacity of the 1 − ␯2 a uy
model is demonstrated in comparisons with actual experiments. ␯
b

The nodal force-displacement relation for the truss of Fig. 2 can

Development of the Truss Geometry be easily developed using the stiffness method as

再冎 冤 冥再 冎
In the present analysis, only two-dimensional simulations are con- A1 c3 sc2
sidered. A concrete element, subjected to concentric compression, + A3 A3
Fx a a a ux
is modeled here as a special truss-assembly which is geometri- = 2E 2 共5兲
cally defined and sized to provide the same elastic stiffness as the Fy sc A2 s3 uy
A3 + A3
concrete continuum under plane-stress conditions 共Fig. 2兲. Simu- b b b
lations based on plane stress are more appropriate at the early
stages of loading where the reactive stressing of ties is minimal. where s = sin ␪ and c = cos ␪, where A1, A2, and A3
Following crack development, the ties are activated introducing a = cross-sectional areas of the horizontal, vertical, and diagonal
complex triaxial state within the concrete core, while the concrete elements, respectively, as shown in Fig. 2.
cover remains in a mostly biaxial state of stress. Corrections to Equating first row, second column terms in Eqs. 共4兲 and 共5兲
account for the three-dimensional state of stress in the core are results in
discussed later in this paper.
Note that the plane-stress relations of elasticity are expressed d␯ sc2 ␯ ad ␯ ld
2 = 2A3 ⇒ A3 = 2 = ⇒ A3
as follows: 1−␯ a 1 − ␯ 2sc
2
1 − ␯2 2sc

再冎
␴x
␴y
=
E
冋 册再 冎
1 ␯
1−␯ ␯ 1
2
␧x
␧y
共1兲 =
␯ ld
1 − ␯ sin 2␪
2 共6兲

The strain-nodal displacement relations are where l = a / c = b / s = length of the diagonal truss member.
The same result is verified by equating the second row, first
column terms in Eqs. 共4兲 and 共5兲.
Similarly equating the first row, first column terms in Eqs. 共4兲
and 共5兲 results in

2 冉 A1
a
+ A3
c3
a
= 冊d b
1 − ␯2 a
⇒ A1 =
1 db
2 1 − ␯2
− A 3c 3 ⇒ A 1

= 冉
1 db
2 1−␯ 2 −
␯ ldc2
1 − ␯2 s
冊 共7兲

Finally, equating the second row, second column terms in Eqs. 共4兲
Fig. 2. Truss model of the plane-stress element and 共5兲 gives

JOURNAL OF ENGINEERING MECHANICS © ASCE / AUGUST 2010 / 1007

J. Eng. Mech., 2010, 136(8): 1006-1014

Downloaded from ascelibrary.org by Guru Nanak Dev Engineering College on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 3. Geometric configurations of the truss simulations

2 冉 A2
b
+ A3
s3
b
冊=
d a
1 − ␯2 b
⇒ A2 =
1 da
2 1 − ␯2
− A 3s 3 ⇒ A 2
Fig. 4. Uniaxial stress-strain of concrete

= 冉
1 da
−
␯ lds2
2 1 − ␯2 1 − ␯2 c
冊 共8兲
the products A1 f r and A2 f ⬘c must remain approximately constant as
To ensure that cross-sectional areas A1 and A2 are positive it is A1 and A2 change with the ratio of a to b. It is noted here that A3
concluded that does not vary significantly with the ratio of a to b.
1
cot2␪ ⬍ and tan2␪ ⬍ ␯ 共9兲
␯
Constitutive Equations
or
Concrete is a pressure sensitive material with both strength and
冑␯ ⬍ tan ␪ ⬍ 1
or 冑␯ ⬍ b ⬍ 1
共10兲 ductility increasing with confining pressure. A one-dimensional
冑␯ a 冑␯ elastoplastic model that describes the strength and deformation
A typical value of Poisson’s ratio of concrete is 0.20. Thus, for characteristics of the concrete truss elements is presented here.
concrete, Eq. 共10兲 can be written as
Concrete Compression Strength
⬍ 冑5
1 b
⬍ 共11兲
冑5 a The compression strength of concrete increases linearly with con-
fining pressure as follows:

冉 冊
The approach outlined above resolves the issue of creating a truss
with equivalent initial stiffness to that of a solid concrete element. ␲ ␾
f ⬘cc = f ⬘c − f l · tan2 + 共13兲
However, the issue of strength must also be addressed. When the 4 2
concrete truss 关Fig. 2共b兲兴 is loaded in axial compression, the hori-
zontal truss elements fracture early in the loading process and do where f ⬘cc = confined concrete strength; f l = confining stress; and
not participate in the load resistance. For the concrete truss that ␾ = angle of internal friction of concrete.
carries no horizontal restriction of deformation, such as transverse
reinforcement, the vertical load is carried almost exclusively by Concrete Axial Stress-Strain in Compression
the vertical elements. Thus, the truss element strength must be
proportioned to the ratio of the actual cross section of the concrete Following Fig. 4, the uniaxial stress-strain diagram of concrete is
element 关Fig. 2共a兲兴 to that of the sum of the vertical elements 关Eq. modeled by a two-parts equation. The prepeak response of con-
共8兲兴 crete is modeled as 共Kent and Park 1971兲

⬘
f ctr
f ⬘c
=
da
da
␯ lds 2 = 2
共1 − ␯2兲a2
a 共1 + ␯兲 − ␯l 2 =
1 − ␯2
b2
共12兲 冋
f c = − f ⬘cc sign共⑀兲 · 冉 冊 冉 冊册
⑀
⑀cco
2
−2·
⑀
⑀cco
共14兲
− 1 − ␯
1 − ␯2 1 − ␯2 c a2 where ⑀ = concrete strain 共positive in tension兲 and ⑀cco = strain at
confined peak stress. In the case of zero confinement, the peak
where f ⬘c = compression strength of concrete and f ctr ⬘
= compression strength of the concrete truss element. Based on stress f ⬘cc and the corresponding strain ⑀cco are equal to their un-
Eqs. 共7兲 and 共8兲, it is noted that when a is smaller than b 关Fig. confined values f ⬘c and ⑀co, respectively.
3共a兲—tall truss configuration兴 A1 tends to be large while A2 be- The postpeak response of concrete, up to the confined crushing
comes small. When a and b are equal 关Fig. 3共b兲—balanced truss strain ⑀ccu, is modeled as
configuration兴, A1 and A2 become equal. Finally, when a is larger
f c = − f ⬘cc关1 + Zmc · 共⑀ + ⑀cc0兲兴 共15兲
than b 关Fig. 3共c兲—short truss configuration兴, A2 tends to be large
while A1 becomes small. These observations become important, where Zmc = softening coefficient. In the case of zero confinement,
because while the relations 共6兲–共8兲 provide the cross sections for Zmc and ⑀ccu acquire their unconfined values Zm and ⑀cu, respec-
the required stiffness compatibility between continuum and truss tively. Note that Zm, and Zmc are positive when expressing soft-
simulation, they do not address the issue of strength. In general, ening.

1008 / JOURNAL OF ENGINEERING MECHANICS © ASCE / AUGUST 2010

J. Eng. Mech., 2010, 136(8): 1006-1014

Downloaded from ascelibrary.org by Guru Nanak Dev Engineering College on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 5. Modeling of softening parameter Zmc

The confined peak strain ⑀cc0 is a function of confinement as Fig. 6. Evaluation of ultimate strain ⑀ccu of confined concrete
follows:

fl

冉 冊
⑀cc0 = ⑀c0 + 0.0277 · 共16兲
f ⬘c ⑀ccu f ⬘c fl
= 1 + 35.3 − 6.25 · · 共19兲
Eq. 共16兲 was derived based on experimental data reported by Mei ⑀cu f ⬘co f ⬘c
et al. 共2001兲 and matched very well a larger set of experimental
where ⑀cu = ultimate compressive 共crushing兲 strain of the uncon-
data presented by Attard and Setunge 共1996兲, as well as Imran
fined concrete and ⑀ccu = ultimate compressive strain of the con-
and Pantazopoulou 共1996兲.
fined concrete.
The softening coefficient Zmc is modeled based on a relation
The reasonable success of Eq. 共19兲 to predict the ultimate
presented by Samra 共1990兲
strain ⑀ccu of confined concrete is demonstrated in Fig. 6, for
0.5 experimentally observed ultimate strains up to approximately

冑
Zmc = 共17兲 0.05
3 + 0.002f ⬘c 3 bc ⬘ is assumed for strains larger than ⑀ccu
A residual strength f ccr
+ · ␳s · − 0.002
f ⬘c − 1000 4 s to be equal to a fraction of the confining stress f l
where ␳s = volumetric content of transverse reinforcement based
on the confined core volume; bc = core width; and s = transverse ⬘ = ␣r · f l
f ccr 共20兲
reinforcement spacing. It is interesting to note in Eq. 共17兲 that the
For the purposes of this study, it is assumed that ␣r = 1, which
yield strength f y of steel is absent. This is based on Cohn and
simply reduces a failed confined element to a hydrostatic state of
Ghosh 共1972兲, who found that column ductility is influenced
stress.
much more by the amount and spacing of transverse reinforce-
ment than it is by its grade.
Eq. 共17兲 relates the softening coefficient Zmc to the element Concrete Axial Stress-Strain in Tension
confinement in an indirect and design oriented approach through
The axial stress-strain relation of concrete in tension is described
the transverse steel content and spacing. Such relation cannot be
by Eq. 共14兲 up to a failure strain ⑀ct as shown in Fig. 4.
used directly in the material constitutive model approach that is
adapted in this study. Instead, Eq. 共17兲 was evaluated for concrete
compression strengths of 28, 42, and 56 MPa for a variety of core Steel Axial Stress-Strain Relation
widths, and transverse reinforcement spacing and diameters, and
The axial stress-strain relation for the reinforcing steel bars is
translated to confining stress f l. The results of this analysis were
expressed by Eq. 共21兲 and shown in Fig. 7
curve-fitted to the dimensionless Eq. 共18兲

冉 冊
f ⬘c 0.5 Elastic region − ⑀y ⱕ ⑀ ⱕ ⑀y f s = Es · ⑀s

冉 冊
Zmc · = 共18兲
f ⬘co fl 2.2
+ 0.05
f ⬘c Flow region ⑀ ⱕ − ⑀y or ⑀ ⱖ ⑀y fs = ⫾ fy
where f ⬘co = reference compression strength, selected here as 28
MPa. Hardening region ⑀ ⱕ − ⑀h or ⑀ ⱖ ⑀h
Comparisons of Eq. 共17兲 for different concrete compression
strengths to Eq. 共18兲 are presented in Fig. 5. ⑀ − ⑀y
fs = ⫾ fy + 共21兲
Modeling the confined ultimate strain ⑀ccu is a challenging 1 ⑀ − ⑀y
problem which has not been discussed sufficiently in the pub- +
Esp ⫾共f u − f y兲
lished literature. A statistical correlation of the experimental data
published by Imran and Pantazopoulou 共1996兲 and Mei et al. where the ⫾ symbol indicates a positive sign in tension and a
共2001兲, produced the following equation: negative sign in compression.

JOURNAL OF ENGINEERING MECHANICS © ASCE / AUGUST 2010 / 1009

J. Eng. Mech., 2010, 136(8): 1006-1014

Downloaded from ascelibrary.org by Guru Nanak Dev Engineering College on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 9. Theoretical failure plane definition

Fig. 7. Stress-strain relation of reinforcing steel

sections: 共1兲 an unconfined region bounded by the free surface

Two-Dimensional Compression Simulation
and a second order parabola with peak ordinate of s / 4, and 共2兲 the
remaining region, which is confined. This is consistent with ear-
General lier studies 共Mander et al. 1988; Karabinis and Kiousis 1996兲. For
This study presents a two-dimensional simulation of a three- the purposes of this study, a truss element is considered confined
dimensional problem. The plane-stress simulation is more repre- if at least 50% of its length is in the confined region.
sentative of the early stages of loading, where the reactive To account for the three-dimensional nature of the problem
confinement of the core is minimal. To account for the more and improve upon the assumption of plane-stress behavior, the
complex triaxial state of stress that develops when the effects of calculated transverse stresses are applied in both horizontal direc-
confinement become pronounced the following corrective steps tions when calculating elastoplastic stiffness and strength. Con-
are introduced: finement is developed due to the action of the ties and is
1. Identify the confined and unconfined regions within a col- calculated as
umn.
2. Consider the three-dimensional state of stresses in the con- Fl
stitutive equations. fl = 共22兲
s·d
3. Select the proper truss-shape for this problem: short, bal-
anced, or tall. where Fl = average force of the ties; s = tie spacing; and d = depth
The column simulation is confined to a module of length s, of the column cross section.
bounded by two ties 共Fig. 1兲. The element is loaded under “de- The selection of truss configuration 共short, balanced, or tall—
formation controlled” conditions, where the top boundary dis- Fig. 3兲 influences the development of failure tendencies of the
places uniformly downwards, forcing compression 共Fig. 8兲. Since structure. To achieve reasonably accurate results and avoid at the
this is a discrete simulation of the continuum, confining stresses same time using a very high mesh refinement, the proper truss
on the truss elements do not result naturally from the analysis and configuration must be selected when possible. The average stress
must be imposed separately. The column mass is divided into two development of the tied column element is presented in Fig. 9共a兲.
The compression strength expression 共13兲 is consistent with the
Mohr-Coulomb criterion. The Mohr circle of stresses at failure is
calculated in Fig. 9共b兲 and shown in Fig. 9共a兲. Thus, for a friction
angle of ␸ = 30°, which is common for concrete, the failure plane
is oriented at 60°, which has a 1.73:1 slope. Thus, a tall configu-
ration with approximately 1.73:1 height to width ratio is preferred
for this problem because it results naturally to shear failure planes
that are commonly observed experimentally.

Aim of the Parametric Investigation

A parametric investigation is conducted here to examine the ef-
fects of confinement characteristics on the strength and ductility
of the concrete columns. The confinement characteristics that are
examined here are:
1. The confining steel content ␳s which is defined as the volu-
metric content of transverse reinforcement based on the con-
fined core.
2. The relative spacing s / b of the transverse reinforcement.
Fig. 8. Discretization To achieve the goals of the parametric investigation, columns

1010 / JOURNAL OF ENGINEERING MECHANICS © ASCE / AUGUST 2010

J. Eng. Mech., 2010, 136(8): 1006-1014

 Table 1. Summary of Numerical Experiments
s/b
␳ 0.0811 0.162 0.27 0.325 0.379 0.433 0.487 0.595 0.6495 0.974
0.005 1.72 1.26 1.39 1.22 1.18 1.03 0.96 1.05 1.05 0.88
1.25 1.19 1.19 1.09 1.02 1.01 1.03 0.92 0.93 0.82
0.0075 2.09 1.53 1.72 1.49 1.40 1.22 1.15 1.02 1.12 0.96
1.50 1.34 1.27 1.21 1.15 1.12 1.10 1.01 1.00 0.79
0.01 2.57 1.78 1.94 1.76 1.52 1.39 1.31 1.16 1.03 1.13
1.54 1.48 1.35 1.33 1.24 1.22 1.21 1.10 1.11 0.87
0.015 3.03 2.24 2.49 2.26 1.89 1.59 1.65 1.47 1.05 1.48
Downloaded from ascelibrary.org by Guru Nanak Dev Engineering College on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.

1.83 1.74 1.63 1.58 1.49 1.45 1.42 1.29 1.11 0.95
0.02 3.64 2.58 2.77 2.89 2.05 1.88 1.86 1.64 1.01
2.05 1.93 1.76 1.74 1.59 1.55 1.50 1.31 1.16
0.03 3.65 2.61 3.23 3.23 2.17 2.08 1.83 1.68
2.05 1.93 1.77 1.75 1.59 1.55 1.53 1.33
Note: Each cell contains the strain ratio ␧ccu / ␧cu 共top兲 and the stress ratio f cc / f c⬘ 共bottom兲.

with square cross sections 共400⫻ 400 mm兲 with ties that vary in ductility with steel content are clearly demonstrated. Similarly, a
spacing and volumetric content, as summarized in Table 1, are series of computational experiments to investigate the effects of
loaded to failure under axial compression. steel-tie spacing while maintaining a constant content of 1.00% is
presented in Fig. 12, where the increase of strength and ductility
with decreasing tie spacing are demonstrated. The cumulative ef-
Results
fects of the tie content and spacing to the column ductility and
The discrete simulation of loading rectangular columns to failure strength are presented in Figs. 13 and 14.
results in certain identifiable features. These are briefly described A number of useful conclusions can be drawn:
in Fig. 10. Deformations are initially elastic. The first recogniz- • In all cases examined here, both confined strength f cc and con-
able feature is due to tensile vertical cracking which results in a fined failure strain ␧ccu increase with increasing tie steel con-
minor loss of stress and a small reduction in stiffness while the tent and decreasing tie steel spacing.
material behavior continues to be elastic. This feature is amplified • Both f cc and ␧ccu gains have an upper limit as the tie steel
by the discrete nature of the truss modeling and is typically not
observed in actual experiments. The second feature is due to spal-
ling of an approximately parabolic-shape concrete mass which is
mostly unconfined. Stress loss is observed at this point, which
increases with the tie spacing, while the stress-strain relation is
inelastic. Continued loading leads to further local failure, attrib-
uted to additional spalling, and is identifiable by a stress drop.
Further loading results is a highly nonlinear response character-
ized by softening and eventual failure. Failure is mainly associ-
ated to a shear band development in cases of low to moderate
transverse reinforcement. However, in cases of closely spaced and
large diameter transverse reinforcement 共␳s ⬎ 2%兲, failure is
mostly due to concrete crushing.
A typical series of computational experiments to investigate
the effects of steel-tie content while maintaining constant tie spac-
ing is presented in Fig. 11, where the increase of strength and Fig. 11. Effect of steel-tie content for constant spacing 共s = 0.43b兲

Fig. 10. Typical column stress-strain relation describing the progres-

sive failure mechanism Fig. 12. Effect of steel-tie spacing for constant content 共1.0%兲

JOURNAL OF ENGINEERING MECHANICS © ASCE / AUGUST 2010 / 1011

J. Eng. Mech., 2010, 136(8): 1006-1014

Downloaded from ascelibrary.org by Guru Nanak Dev Engineering College on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 15. Simulation of Column #6 共Scott et al. 1982兲

content increases. For large tie spacing 共s ⬇ b兲, this limit was
reached at steel content of 1.5%. For smaller tie spacing 共s
⬍ 0.6b兲, this limit was reached at steel content of 2%.
• Within the ranges examined in this study, the magnitude of
␧ccu approaches an upper limit for small tie spacing 共s / b
⬇ 0.2兲 and a lower limit for large tie spacing 共s / b ⬇ 0.6兲. This
is clearly demonstrated in Fig. 14. Similar limits for the con-
fined strength f cc were outside the range examined here.
• In cases of large tie spacing and low tie content, it was ob-
served that the strength and ductility of the “confined” con-
crete were slightly inferior to those of the unconfined concrete.
Fig. 13. Increase in ductility and strength with tie content and spac- This effect, which is the result of crack initiation around the
ing ties without adequate “compensation” by sufficient confine-
ment, has also been observed experimentally, although it has
not been adequately discussed. Yong et al. 共1988兲 observed
this behavior experimentally for columns with rectangular and
diamond shaped ties at spacing s = b, but reported this behavior
as “not affected” by the ties. Hong et al. 共2006兲, observed a
similar behavior for columns confined only with rectangular
ties and s = 0.6b. However, they offered no discussion.
• The loss of strength and ductility that is observed in cases of
large tie spacing may be overcome in some cases with a suf-
ficient increase of confinement volume, i.e., large diameter
ties, which can increase the resistance of the core sufficiently
to overcome stress-concentration losses. However, the re-
quired steel content becomes rather excessive 共⬇2%兲, which
leads to the conclusion that reducing the tie spacing is
preferable.

Comparisons with Experimental Data

To evaluate the quantitative capability of this model a number of
simulations of actual experiments were performed. Only two typi-
cal simulations of actual experiments are presented here for
economy of space. The simulation of Column #6 by Scott et al.
共1982兲 is presented in Fig. 15, while the simulation of Column #3
by Razvi and Saatcioglu 共1989兲 is presented in Fig. 16. In general
it was found that simulations using this model resulted in earlier
failure predictions than observed by the actual experiments. This
is attributed to the two-dimensional nature of the simulation
which lacks space diagonal elements. The plane diagonal ele-
ments are subjected to larger strains and thus earlier failure. It is
also observed that the simulation typically overestimates the
stresses for columns confined as described in Fig. 16. The two-
dimensional nature of this simulation assumes a more uniform
Fig. 14. Increase in ductility and strength with tie spacing and con- confinement than the actual confinement provided by the rein-
tent forcement configuration of Fig. 16, thus predicting higher

1012 / JOURNAL OF ENGINEERING MECHANICS © ASCE / AUGUST 2010

J. Eng. Mech., 2010, 136(8): 1006-1014

 eds., ASCE, Reston, Va., 1377–1379.
Chen, W. F. 共1982兲. Plasticity in reinforced concrete, McGraw-Hill, New
York.
Cohn, M. Z., and Ghosh, S. K. 共1972兲. “The flexural ductility of rein-
forced concrete sections.” Struct. Eng. Int. (IABSE, Zurich, Switzer-
land), 32共11兲, 53–73.
De Borst, R., Sluys, L. J., Mühlhaus, H.-B., and Pamin, J. 共1993兲. “Fun-
damental issues in finite element analyses of localization of deforma-
tion.” Eng. Comput., 10共2兲, 99–121.
Fafitis, A., and Won, Y. H. 共1994兲. “Nonlinear finite element analysis of
concrete deep beams.” J. Struct. Eng., 120共4兲, 1202–1220.
Fraternali, F., Angellilo, M., and Fortunato, A. A. 共2002兲. “Lumped stress
Downloaded from ascelibrary.org by Guru Nanak Dev Engineering College on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.

Fig. 16. Simulation of Column #3 共Razvi and Saatcioglu 1989兲 method for plane elastic problems and the discrete-continuum ap-
proximation.” Int. J. Solids Struct., 39共25兲, 6211–6240.
Hong, K.-N., Han, S.-H., and Yi, S.-T. 共2006兲. “High-strength concrete
column confined by low-volumetric-ratio lateral ties.” Eng. Struct.,
strength than actually observed. It is concluded that the overall 28共9兲, 1346–1353.
predictive performance of this model is acceptable, despite the Imran, I., and Pantazopoulou, S. J. 共1996兲. “Experimental study of plain
simplifications introduced by the two-dimensional modeling. concrete under triaxial stress.” ACI Mater. J., 93共6兲, 589–601.
Karabinis, A., and Kiousis, P. D. 共1996兲. “Strength and ductility of rect-
angular concrete columns: A plasticity approach.” J. Struct. Eng.,
Summary and Conclusions 122共3兲, 267–274.
Kent, D. C., and Park, R. 共1971兲. “Flexural members with confined con-
A truss-based approach aiming to clarify the confined response of crete.” J. Struct. Div., 97共ST7兲, 1969–1990.
concrete columns subjected to axial loading is presented in this Kwon, M., and Spacone, E. 共2002兲. “Three-dimensional finite element
study. Advanced nonlinear modeling is introduced to account for analyses of reinforced concrete columns.” Comput. Struct., 80共2兲,
199–212.
the effects of confinement on the ductility as well as hardening
Li, B., and Tran, C. T. N. 共2008兲. “Reinforced concrete beam analysis
and softening characteristics of concrete. The truss approach pro- supplementing concrete contribution in truss models.” Eng. Struct.,
vides certain computational advantages, as the constitutive model 30共11兲, 3285–3294.
becomes one-dimensional with confinement, which allows for Liu, J., and Foster, S. J. 共2000兲. “A three-dimensional Finite element
easy implementation of the secant modulus of the total stress- model for confined concrete structures.” Comput. Struct., 77共5兲, 441–
strain relation, which, in turn eliminates problems associated with 451.
negative stiffness matrices due to softening. Unfortunately, Liu N., Tang W. H., and Ng, C. W. W. 共2001兲. “Probabilistic FEM for
trusses do not produce very well conditioned stiffness matrices reliability of strain softening media.” Finite Elem. Anal. Design,
and local failures may result in the process of the equilibrium 37共8兲, 603–619.
iterations leading to premature computational instabilities. Tem- Mander, J. B., Priestley, M. J. N., and Park, R. 共1988兲. “Theoretical
porary additions of very compliant elastic springs connecting stress-strain model for confined concrete.” J. Struct. Eng., 114共8兲,
each node to a virtual rigid wall behind the truss resolved this 1804–1826.
problem. At each increment, once equilibrium is achieved, the Mei, H., Kiousis, P. D., Ehsani, M. R., and Saadatmanesh, H. 共2001兲.
“Confinement effects on high strength concrete.” ACI Struct. J.,
springs are removed.
98共4兲, 548–553.
The two-dimensional simulation of the three-dimensional Ortiz, M., and Popov, E. P. 共1985兲. “Accuracy and stability of integration
problem examined in this study limits the accuracy of the analy- algorithms for elastoplastic constitutive relations.” Int. J. Numer.
sis. For example, it is not easy to separate the effects of applying Methods Eng., 21共9兲, 1561–1576.
transverse reinforcement as a set of square ties 共Fig. 16兲, versus a Ortiz, M., and Simo, J. C. 共1986兲. “An analysis of a new class of inte-
combination of square and octagonal, or diamond shape ties 共Fig. gration algorithms for elastoplastic constitutive equations.” Int. J.
15兲. Efforts were made, however, to account for the three- Numer. Methods Eng., 23共3兲, 353–366.
dimensional nature of the problem in the constitutive model by Papadopoulos, P. G., and Xenidis, H. 共1999兲. “A truss model with struc-
allowing equal development of confinement in both transverse tural instability for the confinement of concrete columns.” J. Euro-
directions. The effects of confinement, expressed by tie spacing pean Earthquake Engineering, 2共1兲, 57–80.
and volumetric content, on the stress-strain characteristics of the Razvi, S. R., and Saatcioglu, M. 共1989兲. “Confinement of Reinforced
column, including the failure modes, are quite useful and experi- concrete columns with welded wire fabric.” ACI Struct. J., 86共5兲,
mentally verifiable. A future three-dimensional truss simulation is 615–623.
Salem, H. M. 共2004兲. “The micro truss model: An innovative rational
expected to address most concerns that were raised here.
design approach for reinforced concrete.” J. Adv. Concr. Technol.,
2共1兲, 77–87.
Samra, R. M. 共1990兲. “Ductility analysis of confined columns.” J. Struct.
References Eng., 116共11兲, 3148–3161.
Scott, B. D., Park, R., and Priestley, M. J. N. 共1982兲. “Stress-strain be-
Attard, M. M., and Setunge, S. 共1996兲. “Stress-strain relationship of con- havior of concrete confined by overlapping hoops at low and high
fined and unconfined concrete.” ACI Mater. J., 93共5兲, 1–11. strain rates.” ACI J., 79共1兲, 13–27.
Bažant, Z. 共1997兲. “Fracturing truss model: Size effect in shear failure of Stevens, N. J., Uzumeri, S. M., Collins, M. P., and Will, G. T. 共1991兲.
reinforced concrete.” J. Eng. Mech., 123共12兲, 1276–1288. “Constitutive model for reinforced concrete finite element analysis.”
Bazant, Z. P. 共1976兲. “Instability, ductility, and size effect in strain- ACI Struct. J., 88共1兲, 49–59.
softening concrete.” J. Eng. Mech. Div., 102共2兲, 331–344. Vardoulakis, I., and Aifantis, E. C. 共1991兲. “A gradient flow theory of
Bazant, Z. P. 共1984兲. “Is strain-softening mathematically admissible?” plasticity for granular materials.” Acta Mech., 87共3–4兲, 197–217.
Proc., 5th ASCE-EMD Specialty Conf., A. P. Boresi and K. P. Chong, Yong, Y.-K., Nour, M. G., and Nawy, E. G. 共1988兲. “Behavior of laterally

JOURNAL OF ENGINEERING MECHANICS © ASCE / AUGUST 2010 / 1013

J. Eng. Mech., 2010, 136(8): 1006-1014

 confined high-strength concrete under axial loads.” J. Struct. Eng., sign, 18共1–3兲, 51–58.
114共2兲, 332–351. Zhao, Z. Z., Kwan, A. K. H., and He, X. G. 共2004兲. “Nonlinear finite
Zhang, Y.-G., Lu, M.-W., and Hwang, K.-C. 共1994兲. “Finite element element analysis of deep reinforced concrete coupling beams.” Eng.
modeling of reinforced concrete structures.” Finite Elem. Anal. De- Struct., 26共1兲, 13–25.
Downloaded from ascelibrary.org by Guru Nanak Dev Engineering College on 11/07/22. Copyright ASCE. For personal use only; all rights reserved.

1014 / JOURNAL OF ENGINEERING MECHANICS © ASCE / AUGUST 2010

J. Eng. Mech., 2010, 136(8): 1006-1014