0 Bewertungen0% fanden dieses Dokument nützlich (0 Abstimmungen)

43 Ansichten9 SeitenExisting and proposed models of the effective stiffness of reinforced
concrete columns subjected to lateral loads are assessed using the
experimental response of 329 concrete columns. Existing models
appropriate for design applications tend to overestimate the
measured effective stiffness and are unacceptably inaccurate,
because they generally neglect the influence of anchorage slip on
the effective stiffness of the column. A three-component model that
explicitly accounts for deformations due to flexure, shear, and
anchorage-slip is shown to provide a more accurate estimate of the
measured effective stiffness for the database columns. This model
is simplified by neglecting small terms and approximating the
results of moment-curvature analysis to obtain an accurate and
rational effective stiffness model appropriate for design applications.
For this model, the ratio of the measured stiffness to the calculated
stiffness had a mean and coefficient of variation of 1.02 and 22%
for circular columns and 0.95 and 25% for rectangular columns.

© © All Rights Reserved

PDF, TXT oder online auf Scribd lesen

Existing and proposed models of the effective stiffness of reinforced
concrete columns subjected to lateral loads are assessed using the
experimental response of 329 concrete columns. Existing models
appropriate for design applications tend to overestimate the
measured effective stiffness and are unacceptably inaccurate,
because they generally neglect the influence of anchorage slip on
the effective stiffness of the column. A three-component model that
explicitly accounts for deformations due to flexure, shear, and
anchorage-slip is shown to provide a more accurate estimate of the
measured effective stiffness for the database columns. This model
is simplified by neglecting small terms and approximating the
results of moment-curvature analysis to obtain an accurate and
rational effective stiffness model appropriate for design applications.
For this model, the ratio of the measured stiffness to the calculated
stiffness had a mean and coefficient of variation of 1.02 and 22%
for circular columns and 0.95 and 25% for rectangular columns.

© All Rights Reserved

Als PDF, TXT **herunterladen** oder online auf Scribd lesen

0 Bewertungen0% fanden dieses Dokument nützlich (0 Abstimmungen)

43 Ansichten9 SeitenExisting and proposed models of the effective stiffness of reinforced
concrete columns subjected to lateral loads are assessed using the
experimental response of 329 concrete columns. Existing models
appropriate for design applications tend to overestimate the
measured effective stiffness and are unacceptably inaccurate,
because they generally neglect the influence of anchorage slip on
the effective stiffness of the column. A three-component model that
explicitly accounts for deformations due to flexure, shear, and
anchorage-slip is shown to provide a more accurate estimate of the
measured effective stiffness for the database columns. This model
is simplified by neglecting small terms and approximating the
results of moment-curvature analysis to obtain an accurate and
rational effective stiffness model appropriate for design applications.
For this model, the ratio of the measured stiffness to the calculated
stiffness had a mean and coefficient of variation of 1.02 and 22%
for circular columns and 0.95 and 25% for rectangular columns.

© All Rights Reserved

Als PDF, TXT **herunterladen** oder online auf Scribd lesen

Sie sind auf Seite 1von 9

TECHNICAL PAPER

by Kenneth J. Elwood and Marc O. Eberhard

Existing and proposed models of the effective stiffness of reinforced

concrete columns subjected to lateral loads are assessed using the

experimental response of 329 concrete columns. Existing models

appropriate for design applications tend to overestimate the

measured effective stiffness and are unacceptably inaccurate,

because they generally neglect the influence of anchorage slip on

the effective stiffness of the column. A three-component model that

explicitly accounts for deformations due to flexure, shear, and

anchorage-slip is shown to provide a more accurate estimate of the

measured effective stiffness for the database columns. This model

is simplified by neglecting small terms and approximating the

results of moment-curvature analysis to obtain an accurate and

rational effective stiffness model appropriate for design applications.

For this model, the ratio of the measured stiffness to the calculated

stiffness had a mean and coefficient of variation of 1.02 and 22%

for circular columns and 0.95 and 25% for rectangular columns.

Keywords: bond slip; column; flexural stiffness; lateral loads; reinforced

concrete; yield.

INTRODUCTION

The assumptions made in estimating the stiffnesses of

structural members dominate the computed performance of

a building or bridge subjected to earthquake ground motions.

If these assumptions are used in a linear analysis, they control

predictions of the period of the structure, the distribution of

loads within the structure, and the deformation demands. The

member stiffnesses also control the yield displacement,

which in turn affects the displacement ductility demands

calculated as part of a nonlinear analysis.

The consequences of overestimating or underestimating

the actual stiffnesses of structural members depend on the

type of structural system and the response parameter of

interest. For example, a low estimate of the effective stiffnesses

of columns in a moment-resisting frame usually leads to a

conservative (high) estimate of the displacement demands.

In contrast, a low estimate of the effective stiffnesses for

columns in a shear-wall building would lead the designer to

unconservatively underestimate the elastic shear demands on

the columns. The need for an accurate estimate of effective

stiffnesses is even more crucial for time-history analyses, in

which the peaks and valleys of the ground-motion response

spectrum significantly influence the computed performance.

To assist engineers in developing numerical models for the

estimation of lateral deformation demands, most codes and

standards provide recommendations for member effective

stiffness. The Federal Emergency Management Agency

(FEMA) 356 seismic rehabilitation guidelines (ASCE 2000)

specify the most commonly used procedure for estimating

column stiffness in the U.S. This procedure has been adopted

into the Seismic Rehabilitation Standard, ASCE 41 (ASCE

2007a). It recently has been superseded by a new procedure

specified in ASCE 41 Supplement 1 (ASCE 2007b),

described in Elwood et al. (2007). A similar model is

included in the commentary to the New Zealand concrete

476

used recommendations by Paulay and Priestley (1992). ACI

318-08 (ACI Committee 318 2008) will be the first edition

of ACI 318 to provide stiffness recommendations specifically for lateral-load analysis. These code procedures

are convenient for preliminary analysis, because they

can be implemented without performing a moment-curvature

analysis and without knowing the details of the column

reinforcement. Simple effective stiffness models for

application in design have also been proposed by Mehanny et

al. (2001) and Khuntia and Ghosh (2004).

This paper uses data from the Pacific Earthquake Engineering

Research Center (PEER) Structural Performance Database,

developed by the second author and his students (Berry et al.

2004), to assess the accuracy of these practical methodologies and

to propose a new procedure.

RESEARCH SIGNIFICANCE

The assumed stiffness of a column dominates the results of

linear and nonlinear analyses of buildings and bridges

subjected to ground motions. Currently, most design

professionals assume that the column effective stiffness is a

fixed proportion (say 50 or 100%) of the gross-section stiffness.

Using a database of 329 columns with rectangular and

circular cross sections, this paper shows that existing procedures

for estimating column stiffness are inaccurate. Based on

simplifications of a three-component model, the paper

proposes a new procedure that is more rational, practical,

and accurate. The proposed procedure could be used immediately

by design professionals, and it could be incorporated into

design provisions, such as ASCE 41 (ASCE 2007a) or

ACI 318 (ACI Committee 318 2008).

DATABASE OF MEASURED

EFFECTIVE STIFFNESSES

The PEER Structural Performance Database (Berry et al.

2004) provided the data needed to evaluate the accuracy of

various models of column stiffness. The database contains the

cyclic force-deformation response, geometry, axial load, and

material properties for more than 400 tests of reinforced concrete

columns. A total of 366 of these columns were tested in

cantilever, double-curvature, and double-cantilever

configurations, which makes it possible to isolate the

columns stiffness/flexibility from other sources of flexibility,

such as a flexible supporting beam. To limit the analyses to

columns typical of practice, the axial load was limited to a

maximum of 0.66Ag fc, and the shear-span-to-depth ratio

MS No. S-2007-399.R1 received July 8, 2008, and reviewed under Institute publication

policies. Copyright 2009, American Concrete Institute. All rights reserved, including the

making of copies unless permission is obtained from the copyright proprietors. Pertinent

discussion including authors closure, if any, will be published in the May-June 2010 ACI

Structural Journal if the discussion is received by January 1, 2010.

Columbia, Vancouver, BC, Canada. He received his PhD from the University of

California, Berkeley, Berkeley, CA. His research interests include the behavior and

performance-based design of reinforced concrete structures under seismic loading. He

is Chair of ACI Committee 369, Seismic Repair and Rehabilitation, and a member of

ACI Committee 374, Performance-Based Seismic Design of Concrete Buildings, and

Joint ACI-ASCE Committee 441, Reinforced Concrete Columns.

Marc O. Eberhard, FACI, is a Professor at the University of Washington, Seattle,

WA. He received his BS in civil engineering, and materials science and engineering

from the University of California, Berkeley, and his MSCE and PhD from the University of

Illinois, Chicago, IL. He is a member of Joint ACI-ASCE Committee 445, Shear

and Torsion.

the Notation section).

The selected specimens include 221 columns with rectangular

cross sections with rectangular transverse reinforcement

and 108 columns with circular, octagonal, and square cross

sections with spiral transverse reinforcement (for brevity,

these database subsets will be referred to as rectangular

columns and circular columns, respectively). The

minimum, maximum, mean, and median properties of the

selected column databases are reported in Table 1. Both sets

of column data have wide ranges of column parameters; and

for both sets of data, the strength of over 90% of the included

columns was limited by their flexural strength. This statistic

is a consequence of the limit on the aspect ratio, which

eliminated many shear-critical columns. The distribution of

some parameters varies greatly between the two sets. For

example, compared with the rectangular column dataset, the

circular column set includes fewer columns with axial loads

above 0.3Ag fc or fc above 60 MPa (8700 psi). The circular

columns also tend to have lower values for db /D when

compared with the rectangular columns. These differences

circular columns were proportioned to represent typical

bridge practice, whereas most of the rectangular columns

were proportioned to reflect construction practice for buildings.

For each column, the envelope of the measured lateral

load-displacement relationship was corrected for P-delta

effects to give the effective lateral force envelope for each

column. The yield displacement and effective stiffness of

each column were determined as shown in Fig. 1. For

approximately 90% of the columns, the effective stiffness

was defined based on the point on the measured effective

force-displacement envelope that corresponded to the calculated

force at first yield, Ffirst yield (Fig. 1(a)). Adopting the same

definition of yield used by Benzoni et al. (1996) and others,

the yield force was defined as the first point at which the

tension reinforcement yielded or the maximum concrete

strain reached a value of 0.002, whichever came first. The

database does not include strain measurements, so the force

at first yield was determined with a moment-curvature

analysis, using a linear model for the steel and the Mander et

al. (1988) constitutive relationship for the concrete.

This definition could not be used for columns whose

strength did not substantially exceed the yield force (for

example, shear failures). For these columns, defined as those

whose maximum measured effective force Fmax was not at

least 7% larger than the calculated force at first yield, the

effective stiffness was defined based on the point on the

measured force-displacement envelope with an effective

force equal to 0.8Fmax (Fig. 1(b)).

Assuming the column is fixed against rotation at both ends and

has a linear variation in curvature over the height of the column,

stiffness from test data for: (a) yielding columns; and (b)

columns that did not yield.

Table 1Range of properties for database columns

Circular columns

(108 specimens)

Rectangular columns

(221 specimens)

Mini- MaxiMini- MaxiParameter mum mum Mean Median mum mum Mean Median

a/D

fc, MPa

(ksi)

18.9

(2.7)

1.5

fy, MPa

(ksi)

240

565

420

446

318

587

456

453

(34.8) (82.0) (61.0) (64.7) (46.2) (85.2) (66.2) (65.7)

D/db

0.005

v f c ,

MPa (psi)

P/Agfc

0.05

(0.6)

12

0.1

10.0

4.0

90.0 37.9

(13.1) (5.5)

4.0

1.5

34.4

(5.0)

21.0

(3.0)

48

27

0.99 0.27

(11.9) (3.3)

0.58

0.15

0.010

7.6

3.6

118.0 52.3

(17.1) (7.6)

3.2

36.5

(5.3)

28

12

32

18

16

0.19

(2.3)

0.09

(1.1)

0.71

(8.6)

0.32

(3.9)

0.30

(3.6)

0.10

0.00

0.63

0.23

0.20

stiffness.

477

referred to here as stiffness) can be defined as

gross-section stiffness (Table 2). The measured effective

stiffness can exceed EIg because the effect of the longitudinal

reinforcement on the transformed section is not accounted for

when determining EIg. Correlation coefficients for the rectangular

and circular columns (Rr and Rc , respectively) are also shown in

Fig. 2. The normalized effective stiffness increases most

consistently with increasing axial-load ratio (P/Ag fc) and aspect

ratio (a/D), with correlation coefficients ranging from 0.58 to

0.79. The normalized effective stiffness also increases with

increases in concrete compressive strength (fc). The normalized

stiffness decreases with an increase in the ratio of the steel

yield stress to concrete compressive strength (fy /fc). The

normalized stiffness correlates only weakly with the normalized

bar size and longitudinal reinforcement ratio.

F 0.004 a

EI effmeas = -----------------3 y

(1)

when the extreme concrete fiber reaches a maximum

compressive strain of 0.004, and y is the displacement at yield

according to Fig. 1 for an equivalent cantilever of length a.

Figure 2 demonstrates the influence of several key parameters

on the measured effective stiffness, expressed as a fraction of

the gross-section stiffness, EIg. Within the dataset, the

The models implemented in many of the structural codes

are similar in form to each other. Chapter 8 of ACI 318-08

(ACI Committee 318 2008) provides three options for

approximating member stiffnesses for the determination of

lateral deflection of building systems subjected to factored

lateral loads: (a) 0.35EIg for flexural members (P < 0.1Ag fc)

and 0.7EIg for compression members (P 0.1Ag fc ); (b)

0.5EIg for all members; or (c) as determined by a more

detailed analysis considering the reduced stiffness of all

members under the loading conditions. Figure 3 and Table 2

compare options (a) and (b) with the measured effective

stiffnesses from the column databases. Option (a) generally

overestimates the effective stiffness for axial loads below

0.4Ag fc. Option (b) overestimates the stiffness for columns

with low axial loads and underestimates the stiffness for

columns with high axial loads.

The figure and table also include evaluations of the effective

stiffness models from FEMA 356 (ASCE 2000), ASCE 41

Supplement 1 (ASCE 2007b), and Paulay and Priestley

(1992), all of which allow for interpolation between effective

stiffness values at low and high axial loads. The FEMA 356

(ASCE 2000), and the Paulay and Priestley (1992) recommendations also tend to overestimate the measured effective

stiffness for the columns with low axial loads, particularly

for the rectangular column dataset. Of these existing

procedures, ASCE 41 Supplement 1 (ASCE 2007b) provides

the best average estimate of the measured effective stiffness

(refer to Table 2). But none of these models are accurate. The

coefficient of variation for all of these models ranges from

35 to 58%, depending on the particular model and dataset.

As will be demonstrated in the following, this scatter can be

with existing code models (ASCE 41-S1 = ASCE Supplement 1

[ASCE 2007b]; PP92 = Paulay and Priestley [1992]).

for existing effective stiffness models

Circular columns (108 specimens)

Model

Minimum Maximum

Mean

Median

Coefficient of

variation, %

Minimum

Maximum

Mean

Median

Coefficient of

variation, %

Gross section

FEMA 356

0.13

0.25

1.21

1.96

0.39

0.74

0.34

0.64

55.1

48.1

0.10

0.19

1.22

1.95

0.37

0.68

0.33

0.64

58.1

48.8

ASCE 41 Supplement 1

ACI 318-08 (a)

0.42

0.24

2.11

1.81

1.02

0.76

0.91

0.69

39.1

46.7

0.27

0.14

1.95

1.74

0.82

0.59

0.83

0.53

36.0

49.5

Paulay and Priestley (1992)

0.25

0.26

2.42

1.62

0.78

0.68

0.67

0.61

55.1

41.3

0.19

0.17

2.43

1.56

0.75

0.58

0.65

0.56

58.1

44.4

Khuntai and Ghosh (2004)

0.26

0.21

1.24

1.58

0.61

0.68

0.56

0.62

37.7

41.8

0.16

0.16

1.29

2.03

0.49

0.66

0.49

0.62

41.2

54.3

478

on consideration of expected flexural deformations.

Using a computed moment-curvature relationship, as

shown in Fig. 4, the effective flexural stiffness of the column

EIflex can be determined based on the moment at first yield

of the column, Mfirst yield. The moment-curvature response

was determined for each column in the database based on

plane-section analysis and using the concrete constitutive

model by Mander et al. (1988) and a linear constitutive

model for steel. Figure 5 compares the code-based models

with the calculated flexural stiffnesses of the columns in the

database expressed as a fraction of the gross-section stiffness

(EIflex /EIg). For many of the columns considered, the models

provide an adequate estimate of the flexural stiffness.

Comparing the results in Fig. 3 and 5, however, it is apparent that

other sources of flexibility must be taken into account to

accurately estimate the total effective stiffness.

Other effective stiffness models incorporating the influence

of variables beyond axial load have been proposed in the

literature. For example, Mehanny et al. (2001) accounts for

the influence of the longitudinal reinforcement by introducing a model based on the transformed moment of inertia

and the balanced axial load

EI effcalc EI g, tr = ( 0.4 + P 2.4P b ) 0.9

(2)

THREE-COMPONENT MODEL

OF YIELD DISPLACEMENT

Several researchers (Sozen 1974; Priestley et al. 1996;

Lehman and Moehle 1998; Berry and Eberhard 2007) have

proposed estimating the yield displacement of an equivalent

cantilever column of length a as the sum of the displacement

components due to flexure, shear, and bar slip

y = flex + shear + slip

(4)

(modulus of rigidity).

Mehanny model consistently overestimates the measured

effective stiffnesses and has large coefficients of variation (38%

for circular columns and 41% for rectangular columns).

Khuntai and Ghosh (2004) recommend an effective stiffness

model for lateral-load analysis of reinforced concrete

frames, with and without slender columns, accounting for

the influence of longitudinal reinforcement and effective

eccentricity of the axial load. They propose the following

equation for compression members (P > 0.1Ag fc )

EI effcalc EI g = ( 0.80 + 25 ) ( 1 e D 0.5P P o ) 1.0

(3)

than the effective stiffness for flexural members determined

based on a similar model included in Khuntai and Ghosh

(2004). Compared with the Mehanny model, the Khuntai and

Ghosh model better predicts the average stiffness, but the

coefficient of variation for the ratio of the measured to calculated

effective stiffnesses exceeds 40% for the circular columns

and exceeds 50% for the rectangular column dataset (Table 2).

With the exception of ASCE 41 Supplement 1 (ASCE

2007b), all of the models considered were developed

primarily to provide an estimate to the flexural effective

stiffness (determined based on moment-curvature analyses)

and, hence, ignore additional flexibility due to bar slip and

shear deformations. Consequently, these models ignore the

important dependence of the effective stiffness on the aspect

ratio of the column, evident in Fig. 2(b). Rather than relying

on purely statistical models, it would be preferable to develop

a simple model, whose form is based on the theoretical

calculation of the yield displacement accounting for the

flexibility due to flexure, shear, and bar slip. As shown in the

following sections, this approach can provide a more

accurate estimate of the measured effective stiffnesses of the

database columns.

ACI Structural Journal/July-August 2009

(ASCE 41-S1 = ASCE Supplement 1 [ASCE 2007b]; PP92 =

Paulay and Priestley [1992]).

479

column datasets. Then, the three components of deformation

are combined into a single, nondimensional equation, which

can serve as the basis for the development of practical

effective-stiffness models based on gross-section properties.

Flexural deformations

The calculated flexural curvatures in a reinforced concrete

column can be integrated directly to estimate the column

deformations attributable to flexure. Alternately, assuming a

linear variation in curvature over the height of the column,

the contribution of flexural deformations to the displacement

at yield can be estimated as follows

2

a

a M 0.004

flex = ----- y = ----- ------------3

3 EI flex

(5)

compressive strain of 0.004, and y is the yield curvature, as

defined in Fig. 4.

Shear deformations

The column deformation due to shear within the elastic

range of response is small for most columns, but it can be

large (relative to others sources of deformation) for stocky

columns with high levels of shear demand. Before shear

cracking, this contribution can be estimated by assuming that

the effective shear modulus is equal to the gross-section,

isotropic elastic value (G = Ec /2.4). As the shear cracking

increases, the effective shear modulus reduces significantly.

For many applications, it is convenient to estimate the

shear displacement of an equivalent cantilever column by

idealizing the column as a homogeneous, isotropic material

with a constant, reduced shear modulus

M 0.004

shear = -------------A v G eff

(6)

section (5/6 of the gross-section area of a rectangular column

and 85% of the gross area of a circular column). The expected

effects of concrete cracking suggest that the effective shear

modulus should decrease as a function of the nominal

principle tensile stress. Nonetheless, for application in

engineering practice, the effective shear modulus Geff can be

approximated as one half the elastic value for all levels of

deformation. This value of the effective shear modulus was

selected to optimize the statistics for the effective stiffness

model developed below (Eq. (12)).

480

Slip of the reinforcing bars within the beam-column joints

or foundations further increases the lateral displacements.

This section derives an expression to estimate the lateral

displacement of a column due to bar slip prior to yielding of

the longitudinal reinforcement.

Moments at the ends of a reinforced concrete column tend

to cause tension in the longitudinal reinforcing bars, as

shown in Fig. 6. This tension force Ts must be resisted by the

bond stress u between the reinforcement and the footing or

joint concrete. If the bond stress is assumed to be constant,

equilibrium considerations lead to the following expression

for the length of bar required to resist Ts

db f s

l = -------4u

(7)

shown in Fig. 6, the slip of the reinforcing bar slip can be

expressed as

s db fs

slip = ------------8u

(8)

reinforcing bars slip is given by the ratio of slip to the

distance from the reinforcement to the neutral axis, c. Using

Eq. (8), and recognizing that (s/c) is equal to the curvature

at the section, the lateral displacement of an equivalent

cantilever column of length a due to slip of the reinforcement at

first yield can be expressed as follows

ad b f s first yield

slip first yield = a slip first yield = ---------------------------------8u

(9)

components) is defined as the displacement at an effective

force of F0.004; hence, slip from Eq. (4) can be determined by

multiplying Eq. (9) by the ratio F0.004 / Ffirst yield. Noting from

Fig. 4 that y = first yield (M0.004/Mfirst yield) = first yield(F0.004/

Ffirst yield), the following expression for slip is derived

ad b f s y

slip = ----------------8u

(10)

for elastic response range from u = 0.5 f c to 1.0 f c MPa

(u = 6 f c to 12 f c psi) (Otani and Sozen 1972; ACI

Committee 408 1979; Alsiwat and Saatcioglu 1992; Sozen

et al. 1992; Lehman and Moehle 1998). For the purpose of

this study, the average bond stress was taken as u = 0.8 f c MPa

(u = 9.6 f c psi). At first yield of the column, the stress in

the tension reinforcement, fs , used in Eq. (9) and (10) will

vary depending on the axial load on the column. For

columns with low axial loads, the tension reinforcement will

yield; hence, fs can be taken as equal to the yield stress, fy.

The stress in the tension reinforcement will decrease as the

axial load on the column increases, reaching zero when the

depth of the neutral axis is equal to the effective depth of the

column. Consequently, it is expected that the displacement

due to bar slip will increase with decreasing axial load.

ACI Structural Journal/July-August 2009

response, it is possible to include rotational springs at the

ends of the column elements to directly model the additional

flexibility from the slip of the longitudinal bars. The spring

stiffness can be determined as

8u M 0.004

8u

k slip = --------- ------------- = --------- EI flex

d b fs y

db fs

(11)

the effective stiffness of the column element, acting in series

with the bond element, can be taken as EIflex from a momentcurvature analysis (Fig. 4).

Contribution of components to total

yield displacement

As shown in Fig. 7, the contributions of flexure (Eq. (5)),

shear (Eq. (6)), and bar slip (Eq. (10)) varied consistently

with the axial-load ratio and aspect ratio. For both the rectangular

and circular columns, the flexural mode of deformation

contributed approximately 50 to 100% of the total deformation,

depending on the level of axial load and the aspect ratio. The

slip contribution ranged from 0% for columns with high

low axial loads. The results shown in Fig. 7 also indicate

that, except for stocky columns with high axial loads,

shear deformations contribute less than 15% of the yield

displacement for the columns in the database.

Effective stiffness

For engineering practice, it is convenient to use a single

effective stiffness for a column element. Expressing EIeff calc

as fraction of EIg and substituting Eq. (4) through (6) and

(10) for y, Eq. (1) can be expressed as a function of nondimensional parameters

EI eff calc

-------------------- = -----------------------------------------------------------------------------------------EI g

d

f

f

3 D

18 r 2 D 2 E c

1 + --- ----b- ---- ---s ---y + ------ ----v ---- -------8 D a f y u 5 D a G eff

(12)

column section in the direction of loading (r 2v = Ig/Av). For an

average bond stress value of u = 0.8 f c MPa (u = 9.6 f c psi),

an effective shear modulus, Geff , equal to one half the elastic

value, and using moment-curvature analysis to compute and

fs /fy, Fig. 8 and Table 3 show the ratio of the measured effective

stiffness to the effective stiffness determined using Eq. (12)

for the column databases.

EFFECTIVE STIFFNESS MODELS FOR PRACTICE

For many practical situations, particularly those in which

the column reinforcement has not yet been selected, it is

preferable to use a version of Eq. (12) that does not require

moment-curvature analysis. This section evaluates new

models for effective stiffness that include the influence of bar

slip. The models correspond to simplifications of Eq. (12), in

which the results of moment-curvature analysis (that is,

and fs /fy) are approximated and small terms are neglected.

According to Eq. (12), the ratio of the effective stiffness to

the gross-section stiffness is proportional to the normalized

flexural rigidity, . This normalized flexural rigidity varies

as function of axial load and aspect ratio.

effective stiffnesses.

Table 3Statistics for ratio of measured to calculated effective stiffness for proposed models

Circular columns (108 specimens)

Model

Minimum Maximum

Mean

Coefficient of

Median variation, % Minimum Maximum

Mean

Median

Coefficient of

variation, %

Equation (12)

Equation (16)

0.50

0.63

1.69

1.80

1.04

1.04

1.03

1.00

21.4

22.2

0.45

0.48

1.84

1.47

0.97

0.93

0.92

0.89

26.6

26.6

Equation (17)

Equation (18) with average bar size

0.64

0.57

1.76

1.59

1.04

1.02

0.99

1.00

23.5

22.0

0.48

0.46

1.68

1.63

0.92

0.95

0.91

0.94

26.9

25.5

481

primarily with the level of axial load, but also with the

amount of longitudinal reinforcement. Assuming a linear

stress-strain relationship for the concrete and steel, can be

expressed in terms of the normalized initial strain due to the

axial load, (P/AgEc)/o, and the relative stiffness of longitudinal

reinforcement, n. The normalized flexural rigidity can be

approximated as

P A g E c

approx = 0.2 + 1.3 ------------------+ n 1.0

o

(13)

the binomial theorem) an analytical solution for the moment

The coefficients in Eq. (13) were selected to provide the best

possible estimate of the flexural rigidity determined from

moment-curvature analysis (Fig. 4). Equation (13) provides a

reliable substitute for moment-curvature analysis for a wide

range of rectangular and circular columns. The ratio approx /

(where is determined from a moment-curvature analysis) has

a mean and coefficient of variation of 0.96 10.6% for

rectangular columns and 1.04 9.5% for circular columns.

For all of the 329 rectangular and circular columns, this ratio

ranged from a minimum of 0.69 to a maximum of 1.23.

If the reinforcement ratio has not yet been established,

can be approximated by substituting an average value of n

of 0.15 into Eq. (13), which results in the following relationship

P A g E c

approx = 0.35 + 1.3 ------------------ 1.0

o

(14)

the rectangular and circular column databases are 0.95 14.4%

and 1.04 20.7%, respectively. For the complete database,

the ratios ranged from 0.59 to 1.63. Equation (14), whereas

less accurate than Eq. (13), provides a good estimate of the

normalized flexural rigidity for cases where the longitudinal

reinforcement ratio is not yet known.

As with , the ratio of the steel stress at column yield to

the yield stress of the steel (fs /fy) can also be calculated with

moment-curvature analysis. Alternately, it is possible to

approximate this ratio by taking advantage of its dependence

on the level of axial load. Figure 9, which compares the steel

stress ratio determined based on moment-curvature analysis

with the normalized initial strain (P/AgEc)/o for the database

columns, indicates that the steel stress ratio can be approximated

for both circular and rectangular columns as follows

4 10 P A g E c

0.0 f s f y approx = --- ------ ------------------ 1.0

3 3 o

(15)

third term in the denominator, which is related to shear

deformations and tends to be small (refer to Fig. 7), by

approximating the flexural rigidity with Eq. (13), and by

approximating the steel stress ratio with Eq. (15). Taking

advantage of these relationships, the need for momentcurvature analysis can be eliminated, and the effective

stiffness can be approximated as

1.5 approx_Eq. 13

EI eff calc

--------------------- = --------------------------------------------------------------------------------------- 1.0 and 0.2

EI g

fs

d

D

b

1 + 110 ----- ---- -----------------------------------

D a f

y approx_Eq. 15

effective stiffnesses.

482

(16)

The two coefficients in Eq. (16) (1.5 and 110) were calibrated

to compensate for the elimination of the shear term and to

achieve a good match with the measured effective stiffness

for the rectangular and circular column databases. As shown

in Table 3, Eq. (16) provides similar levels of accuracy as

Eq. (12), without requiring moment-curvature analysis.

Using Eq. (14) (instead of Eq. (13)) to estimate , provides

a model that does not require knowledge of the longitudinal

reinforcement ratio without a significant decline in the

model accuracy (refer to Table 3).

ACI Structural Journal/July-August 2009

EI eff calc

1.5 approx_Eq. 14

--------------------- = --------------------------------------------------------------------------------------- 1.0 and 0.2

EI g

fs

d b D

D a f

y approx_Eq. 15

(17)

axial load as a fraction of Ag fc and recalibrating the

constants, which results in the following relationship

EI eff calc 0.45 + 2.5P A g fc

---------------------- = -------------------------------------------- 1.0 and 0.2

EI g

d

1 + 110 ----b- D

----

D a

(18)

bars, the db /D term in Eq. (18) can be estimated prior to

design for a given application. For bridge columns, the

longitudinal bars tend to be well distributed around the

column cross section, resulting in a smaller db/D term compared

with building columns, where higher axial loads and

architectural constraints tend to result in larger bar sizes relative

to the column dimension. Based on the db/D values for the

circular and rectangular column databases (Table 1), db /D can be

taken as 1/25 for bridge columns and 1/18 for building columns.

The form of Eq. (18) is consistent with the theoretical

formulation of effective stiffness based on Eq. (12) and the

summation of deformation components (Eq. (4) through (6) and

10). It provides a simple estimate of the effective stiffness

without requiring a moment-curvature analysis or selection of

the longitudinal reinforcement ratio. Figure 10 shows the ratio of

the measured effective stiffness for the columns in the

databases to the effective stiffness determined using Eq. (18)

and the recommended average values for db/D. The lack of

trends in Fig. 10 suggests that Eq. (18) properly accounts for the

dependence of the effective stiffness on column axial load and

aspect ratio. Furthermore, the data shows no bias with respect to

the longitudinal reinforcement ratio. As shown in Table 3,

Eq. (18) provides accuracy statistics that are consistent with

those found for the much more complex Eq. (12).

CONCLUSIONS

The effective stiffnesses of 108 spiral-reinforced columns

(with circular and octagonal cross sections) and 221 rectangular

columns were estimated from data in the PEER Structural

Performance Database (Berry et al. 2004). These data show that

the normalized effective stiffness of the columns increases with

increasing axial load, a trend that is reflected in many design

guidelines and codes. The data also show that EIeff /EIg

decreases with decreasing span-to-depth ratio, particularly for

low axial loads.

Seven existing models for column effective stiffness were

evaluated using the column databases. The existing models

were generally based on establishing an estimate for the

flexural rigidity and ignored the influence of bar slip and

shear deformations. All of the models tend to overestimate

the measured effective stiffness and resulted in coefficients

of variation ranging from 35 to 58%.

This paper presented a three-component model, which

explicitly combines the effects of flexure, bar slip, and shear

components of deformation. This model reproduces the

trends observed in the data and it provides an accurate estimate

of column stiffness. For this model, the ratio of the measured

effective stiffness to the calculated stiffness has a mean and

ACI Structural Journal/July-August 2009

columns and 1.04 21% for the circular columns.

For the purpose of design, this paper simplifies the threecomponent model, with little loss of accuracy, by introducing an

approximation to moment-curvature analysis, and by

accounting for the effects of bar slip in terms of span-todepth ratio and axial-load ratio. According to this procedure, the

effective column stiffness at yield can be estimated as

EI eff calc 0.45 + 2.5P A g fc

---------------------- = -------------------------------------------- 1.0 and 0.2

EI g

d D

1 + 110 ----b- ----

D a

(18)

columns and 1/18 for building columns. This model, appropriate

for implementation in design codes for concrete structures, is

more accurate than existing models. Implementation of the

proposed model resulted in a mean and coefficient of variation

for the ratio of the measured effective stiffness to the calculated

stiffness of 0.95 25.5% for rectangular columns and 1.02

22.0% for the circular columns.

ACKNOWLEDGMENTS

This work was supported in part by the Earthquake Engineering Research

Centers Program of the National Science Foundation, under Award Number

EEC-9701568 through the Pacific Earthquake Engineering Research Center

(PEER) and Natural Science and Engineering Research Council (NSERC)

of Canada. The views expressed are those of the authors and not necessarily

those of organizations cited here.

NOTATION

Ag

Asl

Av

a

aF0.004

aFfirst yield

b

c

D

=

=

=

=

=

=

=

=

=

db

Ec

Es

EIeff meas

EIflex

EIg

EIg,t

e/D

F0.004

=

=

=

=

=

=

=

=

=

Ffirst yield

Fmax

fc

=

=

=

fs

fy

Geff

l

M0.004

=

=

=

=

=

Mfirst yield

n

P

Pb

Po

Rc

Rr

rv

Ts

u

=

=

=

=

=

=

=

=

=

=

total area of longitudinal reinforcement

effective shear area of column cross section

shear span

moment at maximum concrete compressive strain of 0.004

moment at first yield (Fig. 4)

width of rectangular column section

distance from tension reinforcement to neutral axis

diameter (circular column) or column depth in direction

of loading (rectangular column)

nominal diameter of longitudinal bars

concrete modulus of elasticity

reinforcing steel modulus of elasticity

measured effective stiffness (Eq. (1))

effective flexural stiffness (Fig. 4)

gross bending stiffness

gross transformed bending stiffness

eccentricity ratio = M0.004 /PD

effective force at maximum compressive strain of 0.004

(Fig. 1)

effective lateral force at first yield (Fig. 1)

maximum measured effective force

concrete compressive strength (150 x 300 mm [6 x 12 in.]

cylinders)

longitudinal reinforcement steel stress at column fixed end

longitudinal reinforcement yield stress

effective shear modulus

length over which bond stress acts (Fig. 6)

aF0.004 = moment at maximum compressive strain at

0.004

aFfirst yield = moment at first yield (Fig. 4)

modular ratio (Es/Ec)

axial load (positive is compression)

axial compression at balanced failure condition

nominal axial load strength at zero eccentricity

correlation coefficient for circular columns

correlation coefficient for rectangular columns

radius of gyration using shear area (rv2 = Ig /Av)

tension force in one longitudinal bar (Fig. 6)

constant bond stress

483

slip

flex

shear

slip

=

=

=

=

=

=

y

o

=

=

y

first yield

slip

=

=

=

normalized effective flexural stiffness = EIflex/EIg

slip of longitudinal reinforcing bar (Fig. 6)

displacement at yield due to flexural deformations (Eq. (5))

displacement at yield due to shear deformations (Fig. (6))

displacement at yield due to bar-slip deformations

(Eq. (10))

displacement at first yield due to bar-slip deformations

(Eq. (9))

displacement at yield (refer to Fig. 1)

nominal strain at which concrete is assumed to yield

(0.002 in this study)

longitudinal reinforcement steel strain at column fixed

end

curvature at yield (Fig. 4)

curvature at first yield (Fig. 4)

rotation at end of column due to slip of reinforcing bars

(Fig. 6)

longitudinal reinforcement ratio (Asl /Ag)

REFERENCES

ACI Committee 318, 2008, Building Code Requirements for Structural

Concrete (ACI 318-08) and Commentary, American Concrete Institute,

Farmington Hills, MI, 465 pp.

ACI Committee 408, 1979, Suggested Development, Splice, and Standard

Hook Provisions for Deformed Bars, Concrete International, V. 1, No. 7,

July, pp. 44-46.

Alsiwat, J. M., and Saatcioglu, M., 1992, Reinforcement Anchorage

Slip under Monotonic Loading, Journal of Structural Engineering, ASCE,

V. 118, No. 9, Sept., pp. 2421-2438.

ASCE, 2000, Prestandard and Commentary for the Seismic Rehabilitation of Buildings, FEMA 356, Federal Emergency Management Agency,

Washington, DC, Nov.

ASCE, 2007a, Seismic Rehabilitation of Existing Buildings,

ASCE/SEI 41, American Society of Civil Engineers, Reston, VA.

ASCE, 2007b, Seismic Rehabilitation of Existing Buildings,

ASCE/SEI 41, Supplement 1, American Society of Civil Engineers,

Reston, VA.

Benzoni, G.; Ohtaki, T.; Priestley, M. J. N.; and Seible, F., 1996,

Seismic Performance of Circular Reinforced Concrete Columns under

Varying Axial Load, SSRP 96/04, Structural Systems Research, University

of California-San Diego, La Jolla, CA.

Berry, M. P., and Eberhard, M. O., 2007, Performance Modeling Strategies

484

Earthquake Engineering Research Center, University of CaliforniaBerkeley, Berkeley, CA, 213 pp.

Berry, M. P.; Parrish, M.; and Eberhard, M. O., 2004, PEER Structural

Performance Database Users Manual, Pacific Earthquake Engineering

Research Center, University of California-Berkeley, Berkeley, CA,

www.ce.washington.edu/~peera1.

Elwood, K. J.; Matamoros, A.; Wallace, J. W.; Lehman, D. E.; Heintz, J. A.;

Mitchell, A.; Moore, M. A.; Valley, M. T.; Lowes, L. N.; Comartin, C.; and

Moehle, J. P., 2007, Update of ASCE/SEI 41 Concrete Provisions,

Earthquake Spectra, Earthquake Engineering Research Institute, V. 23,

No. 3, Aug., pp. 493-523.

Khuntia, M., and Ghosh, S. K., 2004, Flexural Stiffness of Reinforced

Concrete Columns and Beams: Analytical Approach, ACI Structural

Journal, V. 101, No. 3, May-June, pp. 351-363.

Lehman, D. E., and Moehle, J. P., 1998, Seismic Performance of WellConfined Concrete Bridge Columns, PEER-1998/01, Pacific Earthquake

Engineering Research Center, University of California-Berkeley, Berkeley,

CA, 316 pp.

Mander, J. B.; Priestley, M. J. N.; and Park, R., 1988, Theoretical

Stress-Strain Model for Confined Concrete, Journal of Structural

Engineering, V. 114, No. 8, pp. 1804-1826.

Mehanny, S. S. F.; Kuramoto, H.; and Deierlein, G. G., 2001, Stiffness

Modeling of RC Beam-Columns for Frame Analysis, ACI Structural

Journal, V. 98, No. 2, Mar.-Apr., pp. 215-225.

NZS 3101:2006, 2006, Commentary on the Concrete Design Standard,

Part 2, Standards Association of New Zealand, Wellington, New Zealand.

Otani, S., and Sozen, M. A., 1972, Behavior of Multistory Reinforced

Concrete Frames during Earthquakes, Structural Research Series No. 392,

University of Illinois at Urbana-Champaign, Urbana, IL, 551 pp.

Paulay, T., and Priestley, M. J. N., 1992, Seismic Design of Reinforced

Concrete and Masonry Buildings, John Wiley & Sons Inc., New York.

Priestley, M. J. N.; Ranzo, G,; Benzoni, G.; and Kowalsky, M. J., 1996,

Yield Displacement of Circular Bridge Columns, Caltrans Seismic

Research Workshop, California Department of Transportation, Sacramento,

CA, 12 pp.

Sozen, M. A., 1974, Hysteresis in Structural Elements, Applied

Mechanics in Earthquake Engineering, V. 8, pp. 63-98.

Sozen, M. A.; Monteiro, P.; Moehle, J. P.; and Tang, H. T., 1992,

Effects of Cracking and Age on Stiffness of Reinforced Concrete Walls

Resisting In-Plane Shear, Proceedings of the 4th Symposium on Current

Issues Related to Nuclear Power Plant Structures, Equipment and Piping,

Orlando, FL.