Professional Documents
Culture Documents
net/publication/245305815
CITATIONS READS
2 507
5 authors, including:
Some of the authors of this publication are also working on these related projects:
Enhancement of the fire resistant behaviour of slim-floor composite beams with the use of advanced materials View project
All content following this page was uploaded by America Bendito on 30 September 2014.
Abstract: This paper proposes a new equation for the effective length factor 共k-factor兲 for reinforced concrete columns in braced frames.
The new formula is valid both for normal and high-strength concrete. The equation was obtained from a sensitivity analysis performed on
a two-dimensional nonlinear finite-element numerical model that takes into account the inelastic behavior of the concrete columns
共cracking, yielding, and second order effects兲. The numerical model was calibrated with 44 experimental tests performed by the writers’
research group. A comparative study was carried out between the numerical model and different national design codes, displaying
important differences with respect to all of them: the ACI code 共from 37 to ⫺3%兲, the Spanish code EHE 共from 26 to ⫺9.26%兲, and the
Eurocode 2 共from 14 to ⫺14%兲. It was decided to propose two additional simplified equations: one for checking and the second for
design.
DOI: 10.1061/共ASCE兲0733-9445共2009兲135:9共1034兲
CE Database subject headings: Buckling; Reinforced concrete; Concrete columns; Inelasticity; Effective length.
Introduction codes 共both for steel or concrete structures兲 through the use of
simplified equations of the effective length factor or the well-
The evaluation of the effective length factor 共k factor兲 in real known alignment charts.
concrete columns is not properly studied at the present time. This Typically the k-factor depends on the relative stiffness of the
is due to the fact that most of the equations to obtain such factor joints ⌿i, also called “end restrain factor” 共the sum of the column
are developed assuming a linear elastic material behavior or a stiffness divided by the sum of beam stiffness兲. This factor is used
reduced stiffness EI of the column, which is far from the real in the American code ACI_共American Concrete Institute 2005兲 or
behavior of reinforced concrete columns, where the strength of the Spanish EHE 共2001兲, which can vary from 0 to infinite
concrete, the reinforcement ratio, the slenderness, and the stiff-
ness of the joints have an important effect on the curvature of the 兺冉 冊 EI
L
兺冉 冊
columns
support. ⌿i = 共1兲
As is well known, the k factor transforms the buckling of a EI
column with different stiffness restraints at the ends in the buck- L beams
ling of another equivalent pinned-pinned column with an effective But Aristizábal-Ochoa 共1994兲 proposed a “fixity factor” 共i兲 that
buckling length 共Leff = kL兲. The differential equations of both varies from 0 to 1
problems have been widely solved for an elastic material 共Duan
冦 冧冦 冧
and Chen 1999兲 and they are implemented in the national design 1 K1
冉 冊
1 = R1 =
3 E cI g
1 1+
Assistant Professor, Dept. of Engineering, Núcleo Univ. Rafael R1 L
Rangel, Univ. de Los Andes, Venezuela 3101. E-mail: americab@ula.ve 共2兲
1 K2
冉 冊
2
Associate Professor, Instituto de Ciencia y Tecnologia del Hormigon, 2 = R2 =
Univ. Politécnica de Valencia, 46022 Valencia, Spain 共corresponding 3 E cI g
1+
author兲. E-mail: mromero@mes.upv.es R2 L
3
Associate Professor, Instituto de Ciencia y Tecnologia del Hormigon,
Univ. Politécnica de Valencia, 46022 Valencia, Spain. E-mail: jlbonet@ where Ec = elastic modulus of concrete; Ig = gross moment of iner-
cst.upv.es tia of the column section; L = unsupported length of the column;
4
Professor, Instituto de Ciencia y Tecnologia del Hormigon, Univ. K1 and K2 = stiffness of each spring 共end restraint condition兲.
Politécnica de Valencia, 46022 Valencia, Spain. E-mail: pmiguel@ These springs represent the stiffness of the two beams and the
cst.upv.es exterior column that arrive to the joint. Typically EI and K1 and
5
Professor, Instituto de Ciencia y Tecnologia del Hormigon, Univ. K2 are considered elastic.
Politécnica de Valencia, 46022 Valencia, Spain. E-mail: mafernan@ Both the end restraint factors and fixity factor result in the
cst.upv.es
same effective length factor because they are the solutions to the
Note. This manuscript was submitted on October 2, 2007; approved
on May 11, 2009; published online on August 14, 2009. Discussion pe- same differential equation in the elastic range, but use different
riod open until February 1, 2010; separate discussions must be submitted nomenclature. The elastic k-factor for nonsway columns varies
for individual papers. This paper is part of the Journal of Structural from 0.5 共clamped-clamped兲 to 1 共pinned-pinned兲.
Engineering, Vol. 135, No. 9, September 1, 2009. ©ASCE, ISSN 0733- For reinforced concrete structures there were a lot of studies in
9445/2009/9-1034–1039/$25.00. the elastic range regarding the design of slender columns. Cran-
Downloaded 21 Aug 2009 to 158.42.30.58. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
ston 共1972兲 proposed simplified equations for the effective length
factor. Also, a lot of work studying the influence of different end N
restraint conditions was completed by Hu et al. 共1993兲 who pro-
posed a new equation based on the partial fraction model.
Ka Elastic beam, equivalent
Inelastic Behavior to the spring Ka
Moreover, the theoretical problem of inelastic buckling was L
pointed out a long time ago with the tangent elastic modulus 2D reinforced
Hormigón armado 2D
theory and employing the reduced modulus theory or the Shan- Concrete column
ley’s theory 共Shanley 1947兲.
For steel structures, Yura 共1971兲 and Disque 共1973兲 presented Kb Elastic beam, equivalent
an inelastic k-factor method for steel structures using the concept to the spring Kb
of tangent elastic modulus, employing the same alignment charts
but modifying the end restrains factors. N
Besides, the problem for reinforced concrete structures has not
been deeply studied, although MacGregor et al. 共1970兲 and Breen
et al. 共1972兲 pointed out the necessity to get deeper in this subject.
A simple but good manner to include the inelastic behavior of Fig. 1. Geometry of the problem
beams and columns is the one implemented in the ACI code
共American Concrete Institute 2005兲, where the stiffness of beams
and columns for calculating the end restraint factor is reduced research group. The required numerical software had to be two-
using a fixed factor 共i.e., for columns EI = 0.7EcIg兲. dimensional 共2D兲 共membrane elements and trusses兲 to recreate:
But the influence of the k-factor on the behavior of reinforced 2D strain-softening constitutive relations, distributed cracking,
concrete structures is dependent on the strength of concrete bond-slip, plastic behavior of steel and concrete, the appearance
共Broms and Viest 1961兲, the slenderness, the fixity factors, and of plastic hinges, second order effects, etc. The software selected
the steel reinforcement ratio because they contribute to the non- was ATENA 共Cervenka 1998兲. The model included a biaxial frac-
linearity of the column. Moreover, if a column is cracked or ture criteria, tension stiffening, and quadratic isoparametric finite
yielded, its stiffness is lower than the elastic one. In this case the elements with 4 G integration points. So as not to extend the
k-factor will be lower than the elastic one. Conceptually this paper too much, the complex task of calibration can be read in
makes sense because it is as though the rotational springs are Bendito 共2006兲, where an error of 2.03% was achieved. Herein-
relatively more rigid, having a tendency toward the behavior of after, this virtual laboratory allowed performing more tests to pro-
the clamped-clamped column 共for which the elastic k is 0.5兲. pose a new equation for the buckling length.
With the actual sophistication of the numerical models, the The 2D finite-element software with membranes elements can-
concrete can be modeled closer each time to the real behavior. not simulate directly the classical rotational spring, because it has
Thereupon, Bazant and Xiang 共1997兲 studied the inelastic buck- only 2D degrees of freedom 共u and v兲. To do that, special purpose
ling of concrete columns in braced frames but focused the study geometry was created: a column with two elastic beams. These
to improve the method of analysis and not to obtain the k factor. beams represent the rotational springs 共Fig. 1兲.
They assumed a sine curve as the deflection curve of the column A preliminary study to test the geometrical 2D model was
and implemented all the nonlinearities of concrete. The improve- developed in three steps to verify that the buckling behavior is
ment consisted in considering the wavelength as unknown and acceptable.
variable during loading. Conceptually this is the same as the ef- 1. Column and springs 共that is, beams兲 were modeled initially
fective length factor. Later, Furlong 共1998兲 discusses about the using elastic materials to compare the numerical k factor
interest for practitioners to include a very complex method 共al- with the theoretical elastic solution. Both k factors were simi-
though more realistic兲 in the codes. lar to the second decimal.
Moreover, concrete technology has been improved consider- 2. The second test was accomplished supposing an inelastic be-
ably and now high-strength concrete 共HSC兲 can be easily ob- havior for columns but the springs were modeled with 0
tained, whose mechanical behavior cannot be extrapolated simply stiffness. Load-slenderness graphs were obtained to compare
from that of normal strength concrete 共NSC兲. The different sim- with the elastic Euler’s hyperbola values. Also normalized
plified methods that can be used for analysis in failure for slender load-slenderness curves were created to check coherence in
columns therefore need to be checked, so that their application
might be extended to HSC from NSC.
The objective of this paper is therefore to establish an im-
Table 1. Parameters of Study in the Sensitivity Analysis
proved k-factor equation which includes the complicated behavior
observed for reinforced concrete structures, in which inelastic de- Parameters of study Scope
formations are combined with tensile cracking and bond slip. The
Strength of concrete f c⬘ 30, 60, and 90 MPa
equation is limited to nonsway columns.
Longitudinal reinforcement ratio g 2, 3, and 4%
Geometric slenderness = L / h 20,30,35, 40, and 50
Yield stress of steel, f y 400 and 500 MPa
Numerical Simulation
Fixity factor of the rotational springs, 1 = 0.2 and 2 = 0.2
1 and 2 1 = 0.2 and 2 = 0.8
To simulate this behavior a nonlinear finite-element software was
1 = 0.8 and 2 = 0.8
selected and calibrated with 44 experiments performed by this
Downloaded 21 Aug 2009 to 158.42.30.58. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
4000 Inelastic k
Inelastic k
1000 λg = 0.80
h ρg=3%
0 0.70 ρg=2%
0 10 20 30 40
0.60
Slenderness
0.50
0 10 20 30 40 50 60
Fig. 2. Procedure to obtain the inelastic buckling coefficient
Slenderness
Sensitivity Numerical Study The inelastic k factor was compared with the codes ACI-318
共American Concrete Institute 2005兲, the Spanish code EHE
The variables that were studied are presented in Table 1. Both, 共1999兲, the Eurocode 2 共EC2兲 共European Committee for Stan-
column and beams had a square section of 30⫻ 30 cm2. The dardization 2004兲, and with a previous equation proposed by
reinforcement was four bars, located at each corner of the column Traver and Bonet 共2002兲. This last equation comes from a one-
in a symmetric distribution. The mechanical reinforcement cover dimensional finite-element analysis.
was fixed at 10% of the height and the width of the section. As it It was deduced that for all the cases and for any slenderness,
was said in the previous section, a real curve of the critical axial the inelastic k-factor was lower than that obtained using the equa-
load versus slenderness was obtained for each section configura- tion from the ACI and Spanish EHE code. Regarding the EC2 and
tion with the numerical model for a pinned-pinned column 共base concerning some of the slenderness, however, the inelastic effec-
curve兲. This curve improved the elastic Euler’s hyperbola, be- tive length was higher 共Figs. 5 and 6兲.
cause it includes the nonlinearities of the model. It was adjusted Higher differences were observed for the lower strength of
with a fifth degree polynomial. concrete 共f ⬘c 兲, lower reinforcement ratio 共g兲 and for the lower
The maximum load under compression was obtained for each stiffness rotational springs 共1 and 2兲. It is important to find out
parameter using the numerical model and also including the
k-factor
Inelastic k-factor ρ1=0.2, ρ2=0.2; fy=500 MPa; f’c=30 MPa; ρg= 2%
1.00
ρ1=0.2, ρ2=0.2; fy=500 MPa; ρg= 3% ACI, EHE
1.00 0.90
T-B
0.90 f´c=90 MPa 0.80
Inelastic k
k-factor
Downloaded 21 Aug 2009 to 158.42.30.58. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
k-factor Table 3. Values of a and b
ρ1=0.2, ρ2=0.8; fy=500 MPa; f’c=60 MPa;ρ
ρg= 2%
1.00
1 a b
ACI
0.90 0.2 ⫺0.25 0.86
EHE
0.8 ⫺0.16 0.69
0.80 T-B
k-factor
0.70 EC2
0.60 Inelastic k factor Replacing the values of a and b, the following equation is
0.50 obtained:
0 10 20 30 40 50 60
冉 冊
stiffness varies between 0 and infinity. Thereby, the k factor is i = Ri = 共6兲
3 E cI g
calculated in terms of the fixity factors “” initially with respect 1+ ␣
to a fixed slenderness 共 = 35兲 to include in the main part of the Ri L
equation only the variables of the strength of concrete and the where ␣ = slenderness correcting factor.
longitudinal reinforcement ratio. Later on, the slenderness will be The parameter ␣ can be calculated by performing the follow-
included using a correcting parameter. ing steps in sequence, for the other cases of slenderness 共 = 20,
For a particular case of f ⬘c = 60 MPa and a 2% of longitudinal 30, 40, and 50兲:
reinforcement ratio the procedure to obtain an equation of k is • The values of the fixity factors 1 and 2 are analytically re-
explained below. A graph of k − 2 is obtained, with fixed 1 共not placed in the corresponding equation of Table 4.
presented for simplicity兲. The first graph is for 1 = 0.2 and the
second one is for 1 = 0.8.
Only these two graphs are obtained to create initially a very Table 4. Equations of the Inelastic k Factor with respect to the Stiffness
simple equation of k, which is a linear interpolation of 2 Factors, with f y = 500 MPa
But what is very important is that the coefficients “a” and “b” are 2% k = 0.2012 − 0.28共1 + 2兲 + 0.90
not constant; they vary with 1. 3% k = 0.2012 − 0.28共1 + 2兲 + 0.92
If the values of a and b are presented in terms of 1, it can be 4% k = 0.2012 − 0.28共1 + 2兲 + 0.95
inferred that they are also linear functions. These coefficients,
shown in Table 3, are obtained from a trend line. g f c⬘ = 60 MPa
2% k = 0.1512 − 0.28共1 + 2兲 + 0.92
3% K = 0.1512 − 0.28共1 + 2兲 + 0.95
Table 2. Analysis of Errors 4% K = 0.1512 − 0.28共1 + 2兲 + 0.97
Values TB EHE ACI EC2
Maximum 共%兲 14.1 26.2 37.7 14.2 g f c⬘ = 90 MPa
Minimum 共%兲 ⫺7.7 ⫺9.2 ⫺3.6 ⫺13.8 2% K = 0.1512 − 0.28共1 + 2兲 + 0.95
Standard deviation 0.056 0.07 0.093 0.062 3% K = 0.1512 − 0.28共1 + 2兲 + 0.98
Average 共%兲 2.9 5.4 16.7 ⫺1.6 4% k = 0.1512 − 0.28共1 + 2兲 + 1
Downloaded 21 Aug 2009 to 158.42.30.58. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Table 5. Proposed Equation of k Factor for Checking Table 6. Proposed Equation for Design
Normal strength concrete k = 0.212 − 0.28共1 + 2兲 + A Normal strength concrete k = 0.212 − 0.28共1 + 2兲 + 0.95
共up to 50 MPa兲 where A = 0.025g + 0.85⬍ = 1 High-strength concrete k = 0.1512 − 0.28共1 + 2兲 + 1
High-strength concrete k = 0.1512 − 0.28共1 + 2兲 + B where g⫽longitudinal reinforcement ratio
共between 50 and 90 MPa兲 B = 0.03g + f c / 70⬍ = 1 Slenderness correcting factor ␣ = 0.04 – 0.4
共higher than 90 MPa兲 B = 0.025g + f c / 100⬍ = 1 where ⫽geometric slenderness
where g⫽longitudinal reinforcement ratio
Slenderness correcting factor ␣ = 0.04 – 0.4
where ⫽geometric slenderness
equation has a good accuracy for the calculation of the k factor
• The value of inelastic k factor is known because it was previ- for inelastic columns and elastic rotational springs.
ously obtained numerically. In this case, the value of k can be
related to only one unknown, the parameter ␣. Proposed Equation for Design
• As the equation of k depends on ␣ in a quadratic form; an
iterative procedure is performed. The first value of ␣ will be The proposed equation for design is simplified to not depend on
termed “alpha-trial.” the longitudinal reinforcement ratio. They depend on f ⬘c and 1
• The values of ␣ are adjusted until both values of k are and 2 共Table 6兲. Table 7 summarizes the errors for both equations
matched. Doing that, the effect of the last variable is in- 共checking and design兲. It can be noticed that the average error is
cluded in the procedure. 2.6% for checking and around 7.6% for design in the safe side,
• The values of ␣ are obtained in terms of the slenderness. A which improves greatly the existing methods in the codes.
relationship is obtained ␣ = 0.04 – 0.40.
Conclusions
Simplified Equations for Design and Checking
the Codes The calculation of the effective length factor in real concrete col-
umns is not properly addressed now. The reason is, most of the
In this section, simplified equations valid for implementation in research to obtain such length assumes a linear elastic material
national codes are presented. The equations are valid for normal behavior, which is not the case for reinforced concrete. There is
and HSC both for design and checking. no research study prior to the present one that uses 2D nonlinear
finite-element analysis to study the effective length factor.
• It was demonstrated that if the real behavior of the column is
Proposed Equations for Checking
modeled, the k-factor is lower than the elastic one.
The equations for checking are in terms of strength of concrete f ⬘c , • If a sensitivity study is performed, the strength of concrete and
the longitudinal reinforcement ratio g, and the fixity factors 1 the longitudinal reinforcement ratio have the same influence
and 2 共Table 5兲. on the inelastic k-factor coefficient, around 35 and 37%. How-
ever the yield stress of steel has not any influence.
• If a comparative study is performed between the numerical
Comparison between the Proposed Simplified
model and the different codes, it can be shown that there are
Equation for Checking and the Exact Inelastic k Factor
representative differences with respect to all of them: the ACI
Fig. 7 compares the proposed equation for checking with the code 共between 37 and ⫺3%兲, with the Spanish code EHE 共26
exact inelastic effective length factor to demonstrate that it has and ⫺9.26%兲, with the Eurocode 2 共between the 14 and
better accuracy than that existing in the codes. The errors have ⫺14%兲, and regarding Traver and Bonet 共14 and ⫺7%兲. It was
been diminished from the initial 14% until 5.7%. decided to propose a new equation for the effective length
If some random cases are computed for casual values of slen- factor for nonsway columns.
derness, rotational stiffness, and strength of concrete, the maxi- Three types of equations were proposed for the inelastic
mum error is as low as 1.8%. In conclusion, the proposed k-factor: one complete and two simplified 共checking and design兲.
It can be noticed that the medium error is 2.6% for checking and
around 7.6% for design in the safe side, which improves greatly
k-factor
ρ1=0.2,ρ2=0.8; fy=500MPa; 2%, 60MPa on the existing methods in the codes.
1,00
Inelastic k - factor
0,90
Simplified Equation
ACI
k-factor
0,80
Eurocode 2
Table 7. Error for Both Methods
0,70
Error for Error for
0,60 checking design
0,50 Values 共%兲 共%兲
10 20 30 40 50
Maximum 6.19 19.02
Slenderness
Minimum 0.00 0.00
Fig. 7. Comparison between the proposed equation for checking and Typical deviation 0.02 0.04
the inelastic k-factor Average 2.60 7.63
Downloaded 21 Aug 2009 to 158.42.30.58. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright
Acknowledgments de hormigón armado intraslacionales.” Ph.D. dissertation, Universitat
Jaume I, Castellon, Spain 共in Spanish兲.
The writers express their sincere gratitude to the Spanish “Minis- Breen, J. E., MacGregor, J. G., and Pfrang, E. O. 共1972兲. “Determination
terio de Fomento” for help provided through Project No. 13-12- of effective length factors for slender concrete columns.” J. Am.
2001 and “Ministerio de Educación” through Grant No BIA2005- Concr. Inst., 69共11兲, 669–672.
255. Broms, B., and Viest, I. M. 共1961兲. “Long reinforced concrete columns:
A symposium.” Trans. Am. Soc. Civ. Eng., 126共2兲, 308–400.
Cervenka, V. 共1998兲. “Applied brittle analysis of concrete structures.”
Proc., 5th Int. Conf. on Fracture Mechanics of Concrete Structures,
Notation
Freiburg, Germany, Vol. 4, 1107–1116.
Cranston, W. B. 共1972兲. “Analysis and design of reinforced concrete col-
The following symbols are used in this paper:
umns.” Research Rep. No. 20, Cement and Concrete Association,
Ec ⫽ elastic modulus of concrete; London.
f ⬘c ⫽ cylinder strength of concrete; Disque, R. O. 共1973兲. “Inelastic k-factor in design.” Eng. J., 10共2兲, 33–
f y ⫽ yield stress of steel; 35.
h ⫽ height of the column cross section; Duan, L., and Chen, W. F. 共1999兲. “Effective length factors of compres-
Ig ⫽ gross moment of inertia; sion members.” Structural engineering handbook, C. Wai-Fah, ed.,
K1 and K2 ⫽ stiffness of end springs 共end restrain CRC, Boca Raton, Fla.
condition兲; EHE. 共1999兲. Instrucción de hormigón estructural, Ministerio de Fo-
k factor ⫽ effective length factor; mento, Spain 共in Spanish兲.
L ⫽ unsupported length of the column; European Committee for Standardization. 共2004兲. “Eurocode 2: Design of
␣ ⫽ slenderness correcting factor; concrete structures: Part 1.” EC-2 ENV-1992-1-1.
⫽ geometric slenderness= L / h; and Furlong, R. W. 共1998兲.“Discussion: Inelastic buckling of concrete column
i ⫽ fixity factors. in braced frame.” J. Struct. Eng., 124共6兲, 721–722.
Hu, X. Y., Zhou, R. G., King, W. S., Duan, L., and Chen, W. F. 共1993兲.
“On effective length factor of framed columns in ACI code.” ACI
References Struct. J., 90共2兲, 135–143.
MacGregor, J. G., Breen, J. E., and Pfrang, E. O. 共1970兲. “Design of
American Concrete Institute. 共2005兲. “Building code requirements for slender concrete columns.” J. Am. Concr. Inst., 67共1兲, 6–28.
structural concrete and commentary.” ACI 318-05, ACI Committee Shanley, F. R. 共1947兲. “Inelastic column theory.” J. Aeronaut. Sci., 14共5兲,
318, Farmington Hills, Mich. 261–268.
Aristizábal-Ochoa, J. D. 共1994兲. “K-factor for columns in any type of Traver, J., and Bonet, J. L. 共2002兲, “Cálculo del factor de amplificación
construction: Nonparadoxical approach.” J. Struct. Eng., 120共4兲, de momentos en soportes esbeltos intraslacionales de hormigón ar-
1272–1290. mado en flexo-compresión esviada.” MS thesis, Technical University
Bazant, Z. P., and Xiang, Y. 共1997兲. “Inelastic buckling of concrete col- of Valencia, Spain 共in Spanish兲.
umn in braced frame.” J. Struct. Eng., 123共5兲, 634–642. Yura, J. A. 共1971兲. “The effective length of columns in unbraced frames.”
Bendito, A. 共2006兲. “Estudio numérico del pandeo inelástico de soportes Eng. J., 8共2兲, 37–42.
View publication stats Downloaded 21 Aug 2009 to 158.42.30.58. Redistribution subject to ASCE license or copyright; see http://pubs.asce.org/copyright