Practical Performance Model For Bar Buckling
Practical Performance Model For Bar Buckling
Practical Performance Model For Bar Buckling
net/publication/228838083
CITATIONS READS
122 288
2 authors:
Some of the authors of this publication are also working on these related projects:
All content following this page was uploaded by Marc Olivier Eberhard on 28 March 2021.
Abstract: A practical model has been developed to predict, for a given level of lateral deformation, the likelihood that longitudinal bars
in a reinforced concrete column will have begun to buckle. Three relationships linking plastic rotation, drift ratio, and displacement
ductility with the onset of bar buckling were derived based on the results of plastic-hinge analysis, moment-curvature analysis, and the
expected influence of the confinement reinforcement. These relationships, which account for the effective confinement ratio, axial-load
ratio, aspect ratio, and longitudinal bar diameter, were calibrated using observations of bar buckling from cyclic tests of 62 rectangular-
reinforced and 42 spiral-reinforced concrete columns. A version of the drift ratio relationship is proposed for earthquake engineering
applications. The ratios of the measured displacements at bar buckling to the displacements calculated with the proposed model had a
mean of 1.01 and a coefficient of variation of 25% for rectangular-reinforced concrete columns. The corresponding mean and coefficient
of variation for spiral-reinforced columns were 0.97 and 24%, respectively.
DOI: 10.1061/共ASCE兲0733-9445共2005兲131:7共1060兲
CE Database subject headings: Bars; Buckling; Deformation; Concrete columns; Displacement; Ductility.
Fig. 3. Evaluation of approximation for plastic curvature = 1 / 20,  = 1 / 2, and = 0. In the Priestley et al. 共1996兲 model,
␣ = 0.08,  = 0, and = 0.022 共f y in MPa兲, with an upper limit on
L p of 0.044f ydb.
ment ratio, defined as eff = s f ys / f ⬘c , where s⫽volumetric The deformation relationships derived in this section adopt the
transverse reinforcement ratio; f ys⫽yield stress of transverse re- general form of Eq. 共8兲, in which ␣ , , and are unknown con-
inforcement; and f ⬘c ⫽concrete compressive strength. For ex- stants. The column length is included in Eq. 共8兲 to account for the
ample, Saatcioglu and Razvi 共1994兲 found that axial strain duc- moment gradient along the length of the cantilever, and the col-
tility capacity was approximately constant for a given effective umn depth is included to account for the influence of shear on the
confinement ratio. size of the plastic region. The longitudinal bar properties are in-
For development of the proposed deformation relationships, cluded to account for additional rotation at the plastic-hinge re-
the buckling strain is assumed to vary linearly as a function of the sulting from strain penetration of the longitudinal reinforcement
effective confinement ratio, as follows into the supporting element.
bb = 0共1 + 1eff兲 共6兲
Deformations at Bar Buckling
where 0 and 1⫽constants. The parameter 1 is expected to be
larger for spiral-reinforced columns because spiral reinforcement Three commonly used engineering demand parameters 共plastic
is more effective than rectangular reinforcement at confining the rotation, drift ratio, and displacement ductility兲 can be approxi-
core and longitudinal reinforcement. This relationship is similar mated by combining plastic-hinge analysis with the approxima-
to the relationship that Pantazopoulou 共1998兲 proposed for 20% tions of yield displacement, plastic curvature, and plastic-hinge
reduction in flexural strength. In this formulation the transverse length. By substituting the approximations for plastic curvature
confinement spacing, s, is taken into account by its effect on s. 共7兲 and plastic-hinge length 共8兲 into the definition of plastic rota-
By substituting Eq. 共6兲 into Eq. 共5兲 and combining constants, tion 共 p = pL p兲, the plastic rotation at the onset of bar buckling
the plastic curvature at the onset of bar buckling can be approxi- can be expressed as
mated with
冉 冊冉 冊
冢 冣
−1
0 1 + 1eff P L f ydb
p_bb ⬵ 共7兲 p_bb = C0共1 + C1eff兲 1 + C2 1 + C3 + C4
D P Ag f ⬘c D D
1 + 2
Ag f c⬘ 共9兲
where 0 , 1, and 2⫽constants.
The five constants in Eq. 共9兲 共C0 , … , C4兲 are combinations of the
constants included in Eqs. 共5兲 and 共7兲. By substituting Eqs. 共4兲
Plastic-Hinge Length and 共9兲 into Eq. 共2兲, and dividing by the column length, the drift
Numerous models have been proposed to estimate the plastic- ratio at the onset of bar buckling can be expressed as follows:
hinge length of structural members 共Sawyer 1964; Corley 1966;
Mattock 1967; Priestley and Park 1987; Priestley et al. 1996兲. In
many of these models, the expression for the plastic-hinge length
⌬bb
L
=
L
f y + C0共1 + C1eff兲 1 + C2
3Es D
P
冉
Ag f ⬘c
冊 −1
冉 冊
is proportional to the column length, L, column depth, D, and the
L f ydb
longitudinal reinforcement properties, as in the following equa- ⫻ 1 + C3 + C4 共10兲
tion: D D
L p = ␣L + D + f ydb 共8兲
If Eq. 共10兲 is divided by the yield displacement and multiplied by
where db⫽bar diameter of the tension reinforcement. For ex- the column length, the displacement ductility at the onset of bar
ample, in the equation developed by Mattock 共1967兲, ␣ buckling can be expressed as
⌬bb
⌬y
=1+
3Es
f y
冉
C0共1 + C1eff兲 1 + C2
P
Ag f ⬘c
冊 −1 umns兲 and Fig. 6 共spiral-reinforced columns兲 show the variation
of the drift ratio at the onset of bar buckling as a function of key
冉 冊
column properties. To isolate the effect of each property, the da-
1 1 f ydb tabase was organized into families, in which all columns in a
⫻ + C3 + C4 共11兲
L/D L/D D family had similar properties except for the property being stud-
ied. These families are connected with lines in Figs. 5 and 6. It
should be noted that the families do not take into consideration
The constants in Eqs. 共9兲, 共10兲, and 共11兲 can be evaluated from variations in the displacement history imposed on each column.
experimental observations of bar buckling. As expected from Eq. 共10兲, the drift ratio at the onset of longitu-
dinal bar buckling decreases with an increase in P / Ag f ⬘c , and in-
creases with an increase in eff , f ydb / D, and L / D.
Column Performance Database
To calibrate models of column behavior, the results of 467 cyclic, Calibration of Column Deformation Relationships
lateral-load tests of reinforced concrete columns were assembled
in a database, which is available on the World Wide Web at The column database was used to calibrate the column deforma-
www.ce.washington.edu/˜peera1 and http://peer.berkeley.edu/. tion relationships. Specifically, the values of the unknown con-
For each test, the database provides the column geometry, mate- stants 共C0 , … , C4兲 in Eqs. 共9兲, 共10兲, and 共11兲 were determined
rial properties, reinforcement details, loading configuration, a ref- such that 共1兲 the ratios of the measured damage displacements
erence, and test results. The test results provided include the digi- 共from the column database兲 to the calculated damage displace-
tal force-displacement history for the column 共or in a few cases,
ments had a mean value equal to 1.0; and 共2兲 the coefficient of
the force-displacement envelope兲, as well as the maximum col-
variation 共COV兲 of the ratios was minimized. The resulting values
umn deflection imposed before reaching various damage states,
of the constants for each measure of column deformation are
including the onset of bar buckling, ⌬bb. The definition of ⌬bb is
provided in Table 3, along with statistical measures of the accu-
illustrated in Fig. 4.
racy of the resulting equations. The objective functions 关共9兲–共11兲兴
To calibrate Eqs. 共9兲, 共10兲, and 共11兲, the tests were screened
according to the following criteria: 共1兲 the column needed to be used in the regression analyses contained a number of local
flexure-critical, as defined by Berry and Eberhard 共2004兲; 共2兲 the minima with similar magnitudes. As a result one cannot directly
aspect ratio had to exceed 1.9; 共3兲 the longitudinal reinforcement compare the values of C0 , … , C4 among the three equations and
had to be continuous 共unspliced兲; and 共4兲 the displacement pre- two column types. The COVs of the ratios of measured displace-
ceding bar buckling had to be documented. For the 62 ments to calculated displacements are similar for plastic rotation,
rectangular-reinforced and 42 spiral-reinforced column tests that drift ratio, and displacement ductility, ranging from 20 to 29%,
met the screening criteria, Tables 1 and 2 provide values of the depending on the particular measure of deformation, and depend-
maximum deformations preceding the onset of bar buckling, as ing on whether the columns were reinforced with spirals or rect-
well as key column properties. The mean values of the plastic angular ties.
rotations, drift ratios, and displacement ductilities for the Berry and Eberhard 共2003兲 showed that the accuracies of the
rectangular-reinforced columns were 0.043, 4.8%, and 6.0, re- estimates of ⌬bb calculated with Eqs. 共9兲, 共10兲, and 共11兲 can be
spectively. The corresponding mean values for the spiral- increased slightly by using more complex models of bar buckling.
reinforced columns were 0.060, 6.6%, and 6.8, respectively. However, the increases in accuracy did not justify the added com-
The experimental data support the general form of Eqs. 共9兲, plexity. Some of the scatter in the values of ⌬bb / ⌬bb_calc likely
共10兲, and 共11兲. For example, Fig. 5 共rectangular-reinforced col- arises from the influence of repeated deformation cycling
共Kunnath et al. 1997; Ranf et al. 2003兲. In addition, the identifi- displacement between 25 and 50 mm had been imposed on the
cation of the onset of bar buckling is subjective and may vary column.
among observers. The typical practice of imposing a series of
successively increasing cycles to discrete levels of deformation
leads to further scatter. For example, consider Fig. 4, in which bar Practical Implementation
buckling is identified to occur after an imposed displacement of
50 mm. The bars did not buckle at a displacement of 25 mm, but The accuracies of the deformation relationships based on plastic
it is impossible to know whether the bars would have buckled if a rotation 共9兲, drift ratio 共10兲, and displacement ductility 共11兲 were
similar. Since it is easiest to compute the drift ratio 共drift ratio can tion, the accuracy did not vary consistently with the ratio of the
be calculated without estimating the yield displacement, an esti- confinement spacing to the bar diameter 共s / db兲. Although the data
mate that introduces further error兲 practical recommendations available did not document the buckling mode, this slenderness
were developed based on the drift ratio relationship. For practical ratio would be expected to be significant for bar buckling between
implementation, the following relationship, a simpler version of two adjacent hoops or spirals. Since little data are available for
Eq. 共10兲, is proposed to approximate the drift ratio at the onset of large values of s / db , ke_bb should be taken as 0.0 for columns
bar buckling in reinforced concrete columns. with s / db exceeding 6. The accuracy of Eq. 共12兲 with the limit on
s / db is similar to that of Eq. 共10兲. The mean and coefficient of
⌬bb_calc
L
冉
共%兲 = 3.25 1 + ke_bbeff
db
D
冊冉 1−
P
Ag f ⬘c
冊冉 1+
L
10D
冊 variation for ⌬bb / ⌬bb_calc calculated using Eq. 共12兲 were 1.01 and
25% for rectangular-reinforced columns, and 0.97 and 24% for
spiral-reinforced columns.
共12兲
The form of Eq. 共12兲 is consistent with the expected behavior
where ke_bb = 40 for rectangular-reinforced columns and 150 for of reinforced concrete columns. The effective confinement ratio,
spiral-reinforced columns. The form of Eq. 共12兲 was obtained eff, accounts for the restraint against bar buckling that the trans-
from Eq. 共10兲 by combining the transverse and longitudinal rein- verse reinforcement provides. The values of ke_bb for spiral- and
forcement properties controlling the onset of bar buckling into a rectangular-reinforced columns differ because spiral reinforce-
new parameter, effdb / D, fixing the exponents of Eq. 共10兲 to 1.0, ment is more effective than rectangular ties at confining the core
and forcing the coefficients of Eq. 共10兲 to be the same for spiral and the longitudinal reinforcement. The normalized bar diameter,
and rectangular-reinforced columns, except for the coefficient db / D, reflects the increased stability of larger bars and the influ-
multiplying eff. ence of strain penetration into the column base. The axial-load
Berry 共2003兲 showed that the accuracy of Eq. 共12兲 does not ratio term is consistent with the results of moment-curvature
vary significantly with the axial-load ratio, aspect ratio, effective analysis, which indicate that, for the same maximum compression
confinement ratio, and longitudinal reinforcement ratio. In addi- strain, columns with higher axial loads have lower curvatures.
冉 冊
共%兲 Engineering Research Centers Program of the National Science
⌬demand
冉 冊 冉
L target Foundation, under Award Number EEC-9701568 through the Pa-
D ⌬bb_calc cific Earthquake Engineering Research Center 共PEER兲.
冊冉
eff = −1 共13兲
ke_bbdb P L
3.25 1 − 1+
Ag f ⬘c 10D
Fig. 9 shows the implications of using Eq. 共13兲 to proportion the
Notation
confinement reinforcement in a spiral-reinforced column. Fig. 9
The following symbols are used in this paper:
was developed for a buckling probability of 10%, which corre-
sponds to ⌬demand / ⌬bb_calc = 2 / 3 共Fig. 7兲. The confinement require- Ag ⫽ gross area of column cross section;
ments increase as drift ratio and axial load increase, and in some C0 , … , C4 ⫽ parameters used in deformation
cases, the required confinement to prevent bar buckling exceeds approximations;
the spiral requirements for earthquake-resistant construction 共Eq. D ⫽ column depth;
21-2兲 of the American Concrete Institute Building Code Require- db ⫽ bar diameter of longitudinal reinforcement;
ments for Structural Concrete 共ACI 2002兲. Similar figures can be Es ⫽ elastic modulus of longitudinal reinforcement;
generated for other column types and target probabilities of buck- f ⬘c ⫽ compressive strength of concrete;
ling. f y ⫽ yield stress of longitudinal reinforcement;
f ys ⫽ yield stress of transverse reinforcement;
G0 , G1 ⫽ parameters used in approximation of
normalized plastic curvature;
Conclusions
ke_bb ⫽ transverse reinforcement coefficient;
L ⫽ distance from the column base to the point of
The assembled database made it possible to develop a procedure
contraflexure;
to estimate the likelihood of longitudinal bar buckling in flexure-
L p ⫽ plastic hinge length;
dominant reinforced concrete columns based on imposed column
P ⫽ column axial load;
deformation.
␣ ,  , ⫽ parameters used in plastic hinge
The coefficients of variation of the ratio of the measured dis-
approximation;
placement to the calculated displacement at the onset of bar buck-
⌬ ⫽ total column displacement;
ling 共⌬bb / ⌬bb_calc兲 were similar for the relationships based on
⌬bb ⫽ column displacement at the onset of bar
plastic rotation, drift ratio, and displacement ductility. A modified
buckling;
version of the derived drift ratio relationship is proposed for
⌬bb_calc ⫽ calculated column displacement at the onset
earthquake-engineering applications.
of bar buckling;
冉 冊冉 冊冉 冊
⌬demand ⫽ expected column displacement demand;
⌬bb_calc db P L ⌬ p ⫽ plastic displacement of column;
共%兲 = 3.25 1 + ke_bbeff 1− 1+
L D Ag f c⬘ 10D ⌬y ⫽ yield displacement of column;
共14兲 bb ⫽ maximum compressive strain at the onset of
bar buckling;
where ke_bb = 40 for rectangular-reinforced columns and 150 for n ⫽ maximum compressive strain in concrete;
spiral-reinforced columns; eff = s f ys / f ⬘c , s⫽volumetric trans- y ⫽ yield strain of the tension reinforcement;
verse reinforcement ratio; f ys⫽yield stress of the transverse rein- 0 , … , 2 ⫽ parameters used in plastic curvature
forcement; db⫽diameter of the longitudinal reinforcing steel; approximation;
P⫽axial load, Ag⫽gross area of the cross section; f c⬘⫽concrete p ⫽ plastic rotation;
compressive strength; L⫽distance from the column base to the ⫽ parameter for yield curvature approximation;
point of contraflexure; and D⫽column depth. Because little data eff ⫽ effective confinement ratio;
were available for large values of s / db , ke_bb should be taken as s ⫽ volumetric transverse reinforcement ratio;
0.0 for columns in which s / db exceeds 6. p ⫽ plastic curvature;
The mean value of ⌬bb / ⌬bb_calc obtained with Eq. 共14兲 with the p_bb ⫽ plastic curvature at the onset of longitudinal
limit on s / db was 1.01 with a coefficient of variation of 25% for bar buckling;
rectangular columns and 0.97 with a coefficient of variation of p_n ⫽ plastic curvature at maximum compressive
24% for spiral-reinforced columns. The accuracy improves strain n;
slightly if a more complex expression is used, but the increase in p_norm ⫽ normalized plastic curvature; and
accuracy does not justify the increased complexity. By solving 1 , 1 ⫽ parameters used in buckling strain
Eq. 共14兲 for the effective confinement ratio, it is also possible to approximation.
proportion the confinement reinforcement for new columns based
on the column properties, the expected column deformation, and
the target probability of bar buckling. References