Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Concrete Columns
Frédéric Légeron1 and Patrick Paultre, M.ASCE2
Downloaded from ascelibrary.org by Addis Ababa Science and Technology University on 08/26/19. Copyright ASCE. For personal use only; all rights reserved.
Abstract: This paper presents a new confinement model based on strain compatibility and transverse force equilibrium. This new
approach is capable of predicting the effectiveness of transverse reinforcement, which is key in modeling the behavior of high-strength
concrete confined with high-yield-strength steel. The model is validated on test results from more than 200 circular and square large-scale
columns tested under slow and fast concentric loading. In addition, results from about 50 square and circular large-scale columns tested
under constant axial load and reversed cyclic bending were also used in the assessment of the model adaptability to seismic-type loading.
All the predictions are in very good agreement with the experimental results. The model is especially effective in assessing the effec-
tiveness of high-yield-strength steel.
DOI: 10.1061/共ASCE兲0733-9445共2003兲129:2共241兲
CE Database keywords: High-strength concretes; Concrete columns; Confinement; Models.
Need for a General Confinement Model for Concrete The confinement models proposed in the past have limited
Columns validity in terms of concrete strength, transverse reinforcement
yield strength, column geometry, or load conditions. A unified
With the development of performance-based design methods, model capable of adequately reflecting these various conditions is
there is an increasing need for simplified but reliable analytical needed. This model should be based on a rational approach to the
tools capable of predicting the flexural behavior of reinforced confinement phenomenon rather than on multicriteria statistical
concrete members. In capacity design methods applied to multi- analyses. Cusson and Paultre 共1995兲 have proposed a rational
story buildings, strong columns/weak beams ensure that no story model that was limited to square high-strength concrete columns.
mechanisms develop. The ground floor columns must, however, The model was often considered too complex because it uses an
iterative procedure to predict the transverse reinforcement stress
be capable of developing sufficient ductility. Bridges often rely
at peak strength. This iterative procedure would not converge
solely on the capacity of piers to sustain large displacements with-
toward realistic values when initial guesses of the transverse re-
out collapsing. Design offices will be faced more and more with
inforcement strains were too far from the actual values 共Cusson
the need of predicting the deformation capacity of concrete mem-
et al. 1996, 1998; Razvi and Saatcioglu 1998兲. Experimental and
bers. While predicting the flexural behavior of concrete members analytical work carried out at the University of Sherbrooke in the
is a simple matter nowadays, a general approach to account for past decade on the behavior of concrete columns subjected to
confinement of concrete columns is still needed. By general, we concentric load as well as cyclic flexure and axial load 共Cusson
mean that the model should be applicable to the various cases and Paultre 1994, 1995; Paultre and Légeron 1999; Légeron and
engineers face in practice for seismic designs, i.e., different axial Paultre 2000; Paultre et al. 2001兲, combined with a number of
load levels 共from beam-type to essentially gravity-frame col- experimental data published in the past decade, allowed modifi-
umns兲, a wide range of concrete strength available 共20 to 140 cation of the Cusson and Paultre model to make it applicable to
MPa兲, normal- or high-strength confinement steel 共300 to 1,400 square and circular columns made with normal- and high-strength
MPa兲, and variable ductility demands 共from limited ductility to concrete and confined with normal- and high-strength steel. In the
fully ductile structures兲 to reach a given performance level. The process, the model has been simplified by removing the need to
main objective of this paper is to propose a rational uniaxial predict the transverse reinforcement stress at peak strength by an
stress-strain model to account for confinement of concrete col- iterative procedure. This paper presents the theoretical basis of the
umns with a wide range of concrete strength and transverse rein- proposed model which has been validated on 210 square and
forcement yield strength. circular columns tested under concentric compression and 50
square and circular columns tested under reversed cyclic bending
1
Senior Bridge Engineer, Jacobs Civil, Iselin, NJ 08830. and constant axial load.
2
Professor, Dept. of Civil Engineering, Univ. of Sherbrooke, Sher-
brooke PQ, Canada J1K 2R1. E-mail: patrick.paultre@courrier.usherb.ca
Note. Associate Editor: Joseph M. Bracci. Discussion open until July Confinement Model
1, 2003. Separate discussions must be submitted for individual papers. To
extend the closing date by one month, a written request must be filed with
the ASCE Managing Editor. The manuscript for this paper was submitted Circular Column Encased in a Continuous Envelope
for review and possible publication on August 22, 2001; approved on
January 17, 2002. This paper is part of the Journal of Structural Engi- It has long been known that confinement pressure increases the
neering, Vol. 129, No. 2, February 1, 2003. ©ASCE, ISSN 0733-9445/ capacity in compression 共Richart et al. 1928兲. In columns, con-
2003/2-241–252/$18.00. finement with hoops, spirals, or a steel casing is referred to as
f cc ⫽ f ⬘cc 冋 k 共 cc / ⬘cc 兲
k⫺1⫹ 共 cc / ⬘cc 兲 k 册 , cc ⭐ ⬘cc (11)
Fig. 4. Relation between ⬘h / ⬘c and I ⬘e Fig. 5. Relation between ⬘h / ⬘c , I ⬘e and stress-strain relationship of
transverse reinforcement steel expressed in terms of I e and h / ⬘c
tation is presented here, enabling a direct formulation in which
iteration is not necessary. Let us introduce the effective confine-
ment index at peak stress 共Cusson and Paultre 1995兲
Fig. 4 represents the strain compatibility condition in terms of
f ⬘ᐉe relation between I ⬘e and h⬘ / ⬘c . It is convenient to plot the force
I ⬘e ⫽ (18) equilibrium on the same graph for a graphical solution. Using Eq.
f ⬘c
共6兲, the effective confinement index is given by
Strength and ductility enhancements are expressed as a function
fh
of this nondimensional effective confinement index. Based on the I e ⫽ sey (24)
work by Cusson and Paultre 共1995兲, new relations are introduced f ⬘c
that cover a wide range of concrete strength where f h is a function of h / ⬘c . This expression shows that I e is
f ⬘cc a function of the stress-strain relationship of the confining steel.
⫽1⫹2.4共 I ⬘e 兲 0.7 (19) This relation is plotted in Fig. 5 for three column configurations
f ⬘c
confined with an elastic perfectly plastic steel. Eq. 共23兲 is also
⬘cc shown on the same graph. Since both Eqs. 共23兲 and 共24兲 should
⫽1⫹35共 I ⬘e 兲 1.2 (20) be satisfied simultaneously, the peak should occur at the intersect-
⬘c
ing point 关 (I ⬘e ) A for Column A, (I ⬘e ) B for Column B, and (I ⬘e ) C for
Dividing Eq. 共17兲 by ⬘c gives Column C兴. Indeed, the iterative procedure in the Cusson and
Paultre model 共1995兲 is intended to find this intersection point.
h⬘ ⬘cc ⬘ 兲 f ᐉe
共 1⫺ cc ⬘ The abscissa of this point gives the effective confinement index at
⫽ ⬘cc ⫺ (21)
⬘c ⬘c E ⬘cc ⬘c the peak, I e⬘ . An analytical solution can also be found. The slope
of the line representing the elastic range of steel in Fig. 5 is
where ⬘cc and E ⬘cc are unknown, since they are ‘‘equivalent’’
values. Assume that E ⬘cc ⫽␣ f ⬘cc / ⬘cc , where ␣⭓1 is a parameter f ⬘c
reflecting the ‘‘equivalent’’ column concept. Introducing Eqs. 共19兲 ⫽ (25)
sey E s ⬘c
and 共20兲 into Eq. 共21兲, yields
冋 册
When ⭐10, there is no intersection of the two curves in the
⬘h ⬘
1⫺ cc I ⬘e elastic range, and steel yields at peak 共Case A in Fig. 5兲. Hence,
⫽ 共 1⫹35共 I e⬘ 兲 兲 ⬘cc ⫺
1.2
⫻ (22)
⬘c ␣ 1⫹2.4共 I ⬘e 兲 0.7 I ⬘e ⫽ sey f hy / f ⬘c . If ⬎10, there might be an intersection point in
the elastic range 共Case C in Fig. 5兲 or transverse reinforcement
Fig. 4 presents experimentally determined h⬘ / c⬘ values ob-
may yield if steel yield strength is low enough 共Case B兲. This can
tained from about 80 columns tested under uniaxial compression
be summarized as
再
共Uzumeri 1980; Mander et al. 1984; Nagashima et al. 1992;
Sheikh and Toklucu 1993; Cusson and Paultre 1994兲. Fig. 4 also f hy if ⭐10
provides Eq. 共22兲 for ⬘cc ⫽0.43 and ␣⫽1.1. These values imply f ⬘h ⫽ 0.25f c⬘ (26)
that the column is in an average damaged state intermediate be- ⭓0.43 ⬘c E s ⬎ f hy if ⬎10
tween tie location and in between ties. Considering the high level sey 共 ⫺10兲
of uncertainties in the measurement of strain in ties, the predic- From this equation, it is evident that the more a column is
tions are in good agreement with experimental results. This pair confined, the more it is able to effectively use the yield strength of
of values represents a best fit of the data. The following simpler the transverse reinforcement. This fact is consistent with experi-
relation which provides good correlation with the test data is re- mental evidence 共Cusson and Paultre 1994兲.
tained 共Fig. 4兲: In the preceding, it was assumed that the transverse reinforce-
ment steel presented an elastic perfectly plastic stress-strain rela-
h⬘
⫽0.25⫹10I e⬘ ⭓0.43 (23) tionship. Even though this is rarely the case, strain hardening
⬘c effects can be neglected for normal-strength steel because the
Fig. 6. Relation between ⬘h / ⬘c , I ⬘e and stress-strain relationship of Fig. 7. Potential use of high-yield-strength steel
transverse reinforcement steel expressed in terms of I e and h / ⬘c and
accounting for strain hardening
It should be noted that for very well-confined concrete, the
onset of strain hardening is not usually reached at peak. This, axial strain could reach values approaching 0.05. At this strain
however, is not true for high-yield-strength steel presenting a level, strain hardening has generally begun and the confining
rounded stress-strain relationship with no marked yield stress. stress is underestimated when computed with the yield strength. It
The conventional yield strength defined for a 0.2% offset could is possible to use the strain compatibility to find the confined
underestimate the confining stress for very well-confined columns stress at cc50 . For the sake of simplicity, f hy is used in all the
where strain could be well in excess of the conventional yield cases, regardless of the confinement level.
strain. On the contrary, for lower confinement, the strain could be
well below the conventional yield strain but still in the curved
portion of the stress-strain curve, which means that the confining Use of High-Yield-Strength Steel
stress could be overestimated. A more refined representation of A case of particular importance is Column A in Fig. 5 which
the behavior of the steel in computing the effective confinement represents a configuration in which yielding of the transverse re-
index at peak can be used in this case 共Fig. 6兲. inforcement is always reached at peak load. For such columns,
the peak load in confined concrete is attained only when yielding
Postpeak Strain at 50% of Maximum Stress occurs in the ties. This is consistent with test results of concrete
columns confined with carbon-fiber casing. In these tests, there
The postpeak strain cc50 measured at 50% of the maximum stress are two possible failure modes: 共1兲 concrete failure for low con-
defines the postpeak shape of the stress-strain curve. The follow- finement and 共2兲 casing failure with axial stress increasing up to
ing equation is proposed: failure of the casing. The values of ⭐10 are then referred to as
cc50 the ‘‘unconditional yielding condition.’’ For practical applica-
⫽1⫹60I e50 (27) tions, it is of interest to know what effective sectional ratio of
c50
transverse reinforcement is necessary to reach this unconditional
where I e50⫽effective confinement index at cc50 共Fig. 3兲, and yielding condition. Setting ⫽10 in Eq. 共25兲, we can find the
c50⫽corresponding postpeak strain in the unconfined concrete critical sectional ratio of transverse reinforcement
measured at 0.5f ⬘c . The strain c50 is very difficult to measure
experimentally. It requires a very stiff testing machine, special f ⬘c
sey ⫽ (29)
control, and instrumentation. Few experimental values are avail- 10E s c⬘
able in the literature. In the absence of data, it is possible to use
c50⫽0.004 as suggested by Cusson and Paultre 共1995兲. It should where ⬘c ⫽strain corresponding to f ⬘c and can be taken as
be noted that for high-strength concrete, lower values have been 共Légeron 1998兲
reported 共Sheikh et al. 1994兲 depending on the type of aggregates ⬘c ⫽0.0005共 f ⬘c 兲 0.4 (30)
and mix proportioning, but in most cases, 0.004 is a reasonable
estimate for c50 . Introducing Eq. 共30兲 in Eq. 共29兲 yields
Experimental evidence from concentric compressive tests of 共 f c⬘ 兲 0.6
tied columns shows that transverse ties do not always reach yield sey ⬇ (31)
1,000
at peak loads, especially when the ties are made of high-yield-
strength steel. However, the confining steel always yields at When sey is larger than the value given by the right-hand side
cc50 . This is due to the large concrete expansion that takes place term of Eq. 共31兲, the confinement steel yields at peak load regard-
after the peak. Hence, I e50 is computed with f h ⫽ f hy and only the less of the steel’s yield strength. For such columns, increasing the
force equilibrium is necessary confinement stress can be achieved by increasing the yield
strength of the steel while keeping the same volumetric ratio of
f hy
I e50⫽ sey (28) transverse reinforcement. Consider, for example, a column made
f ⬘c of 100-MPa concrete 共Column A in Fig. 7兲 with sey ⫽0.025 and
Circular columns
Mander et al. 共1984兲 500 15 27–31 0.60–2.52 307–340 450
Li et al. 共1994兲 240 17 41–97 0.79–2.94 445–1318 300
Sheikh and Toklucu 共1993兲 203–356 27 35–36 0.59–2.44 452– 629 250– 400
Razvi and Saatcioglu 共1997兲 250 20 60–124 0.41–3.06 400–1000 300
Fig. 9. Prediction of stress in confinement steel at peak Fig. 11. Prediction of strain in confinement concrete at peak
Fig. 10. Prediction of maximum stress in confined concrete Fig. 12. Prediction of postpeak strain at 50% of peak
University of Canterbury
Watson and Park 共1994兲 400⫻400 9 39– 47 0.58 – 4.34 255–388
Downloaded from ascelibrary.org by Addis Ababa Science and Technology University on 08/26/19. Copyright ASCE. For personal use only; all rights reserved.
University of Sherbrooke
Légeron and Paultre 共2000兲 305⫻305 6 92–104 1.96 – 4.26 391– 418
Paultre et al. 共2001兲 305⫻305 6 79–110 2.27– 4.26 418 – 825
Robles et al. 共2001兲 300 6 97–109 1.10– 4.20 440–560
fectly plastic, which is not a good assumption for high-yield- Columns under Flexure and Constant Axial Load
strength steel with rounded stress-strain curves.
Columns are rarely loaded under pure axial compression only.
The predicted strength of the confined concrete are compared
The model is therefore used to predict the behavior of columns
to the experimental results in Fig. 10. The figure shows that the
tested under constant axial load and reverse cyclic flexure simu-
predictions are in good agreement with the experiments. Consid-
lating earthquake loading, which is a classical application of con-
ering only columns with numerical data given by the authors 共168
finement. For convenience, the experimental results are grouped
columns兲, the average ratio of experimental to predicted values is
by university 共Table 2兲. Seventeen columns were tested at the
1.03.
University of Toronto 共Canada兲 by Sheikh and Khoury 共1993兲,
The predictions of the strain at peak strength are compared to
Sheikh et al. 共1994兲, and Bayrak and Sheikh 共1998兲; 14 at the
the experimental results in Fig. 11. The predictions are very good,
University of Canterbury 共New Zealand兲 by Watson and Park
with a ratio of experimental-to-predicted values of 1.10, when
共1994兲, Li et al. 共1994兲; and 18 at the University of Sherbrooke
only the values reported by the authors are considered 共168 col-
共Canada兲 关12 square columns by Légeron and Paultre 共2000兲 and
umns兲.
Paultre et al. 共2001兲, and six circular columns by Robles, Bouaa-
Predictions of the postpeak strain cc50 when the stress in the
nani, and Paultre 共2001兲兴. All the columns are connected to a
confined concrete is 50% of its maximum strength are compared
massive stub representing a footing or beams and slab, typical of
to experimental results in Fig. 12. The predictions are very good,
columns in the first story of a building. The concrete strength of
with a ratio of experimental-to-predicted values of 1.13 共for 147
the 47 columns ranged from 30 to 120 MPa; the yield strength of
columns with numerical values reported by the authors兲.
transverse reinforcement ranged from 400 to 1,400 MPa. For each
specimen, the complete moment-curvature response was com-
Fig. 13. Prediction of maximum moment Fig. 14. Prediction of ultimate curvature
puted with the MNPHI computer program 共Paultre 2001兲 with a certain length, buckling of longitudinal bars, and spalling of con-
layer-by-layer analysis incorporating the proposed model. The crete cover. Finally, a small difference in the descending branch
factor ␣⫽0.85 was not used in flexure. Fig. 13 compares the of the experimental and the predicted moment-curvature diagram
experimental with the predicted maximum moments. The pre- leads to an important difference in ultimate curvature due to the
dicted values are in good agreement with the experimental results, definition in loads of the ultimate curvature. This means that ex-
even for highly axially loaded columns in which the confinement perimental error on moments 共or in the evaluation of the P⫺⌬
effect is the most sensitive. This is reflected by an overall mean effect兲 is translated into a large error in ultimate curvature, spe-
ratio between experimental and predicted maximum moment of cifically for well-confined columns.
1.03 with very small variability. The predictions are slightly con- Fig. 15 compares the predicted and experimental moment cur-
servative since the stub effects at the column bases, which pro- vature for six columns tested by Robles et al. 共2001兲. The col-
vide additional confinement to the tested columns, were not ac- umns are circular and constructed with high-strength concrete of
counted for. Fig. 14 compares the experimental with the predicted about 100 MPa. The columns whose responses are presented on
ultimate curvature. The ultimate curvature is defined as the cur- the right side of Fig. 15 共C100SH100N15, C100SH100N25, and
vature when bending capacity has dropped by 20% of the maxi- C100SH100N40兲 are confined with high-yield-strength-steel with
mum moment. The general trend is very well predicted. The mean f y ⫽560 MPa. It can be seen that the predictions of moment-
ratio between experimental and predicted ultimate curvature is curvature response for circular columns are very good.
0.96 with some scatter. However, the ultimate curvature is very Fig. 16 presents the predicted and experimental horizontal
difficult to measure due to the localization of damage along a force-tip displacement diagrams for six columns tested by
Légeron and Paultre 共2000兲 and by Paultre et al. 共2001兲. These reinforcement yield strength. A direct procedure is proposed to
diagrams were obtained by integrating the moment-curvature dia- determine the stress in the transverse reinforcement at pic strength
gram predicted with the model and using the plastic hinge length of confined concrete. The effectiveness of the model was gauged
recommended by Priestley et al. 共1996兲. The predicted responses against test results from more than 200 columns tested under
are in very good agreement with the experimental results in terms concentric compression and 50 columns tested under constant
of moment capacity, maximum horizontal forces, and ductility. axial load and reversed cyclic flexure. Predictions were found to
be in very good agreement with the experimental results.
Conclusions Acknowledgments
The writers acknowledge the financial support provided by the
This paper presents a uniaxial confinement model applicable to a
Natural Sciences and Engineering Research Council of Canada.
large range of concrete strength. In this model, the behavior of
confined concrete is related to a nondimensional parameter I e ,
which takes into account the amount of transverse confinement Appendix: Use of Model
reinforcement, the spatial distribution of the transverse and longi- Two columns in the test series presented in Fig. 16 are selected to
tudinal reinforcement, the concrete strength, and the transverse illustrate the use of the model.
f ⬘c ⫽78.7 MPa, ⬘c ⫽0.003, s ⫽4.26%, f hy ⫽438 MPa and K e The following symbols are used in this paper:
⫽0.688. A s ⫽ cross section of one leg of hoop;
1. Determine sey from Eq. 共5兲 A sh ⫽ area of transverse reinforcement within spacing s;
A shy ⫽ area of transverse reinforcement within spacing s
sey⫽ 12Kes⫽ 21⫻0.688⫻0.0426⫽0.0147
and perpendicular to direction y;
2. Calculate from Eq. 共25兲 c ⫽ diameter of circular column, diameter of circular
f ⬘c 78.7 confined concrete core;
⫽ ⫽ ⫽8.9 c y ⫽ dimension of rectangular column in y direction;
seyEsc⬘ 0.0147⫻200,000⫻0.003
Downloaded from ascelibrary.org by Addis Ababa Science and Technology University on 08/26/19. Copyright ASCE. For personal use only; all rights reserved.