Design and Detailing of High Strength Concrete Columns

by

Stephen J. Foster

School of Civil and Environmental Engineering

The University of New South Wales, Sydney, Australia

 2

THE UNIVERSITY OF NEW SOUTH WALES

SCHOOL OF CIVIL AND ENVIRONMENTAL ENGINEERING

BIBLIOGRAPHIC DATA SHEET ISSN 0077-880X

1. UNICIV Report No. R375 2. ISBN: 85841 342 6 3. Date: March, 1999

4. Title and Subtitle :

Design and Detailing of High Strength Concrete Columns

5. Author (s)
Stephen J. Foster BE, MEngSc, PhD, Senior Lecturer, School of Civil and
Environmental Engineering, The University of New South Wales, Australia.
6. Keywords
High strength concrete, columns, cover spalling, strength, ductility, tie
spacing, design.

7. Abstract

In this report the behaviour, strength and ductility of high strength concrete columns is
reviewed. Particular emphasis is given to the issue of cover spalling and its influence on
strength and ductility. Cover spalling in columns occurs after cracking along the interface of
the cover and the core is initiated by a triaxial stress condition. This cracking occurs before
the concrete reaches its uniaxial compressive strength and leads to the condition of early
cover spalling. If sufficient ties are provided, and detailed to provide efficient and effective
confinement in the localised region, the column can maintain a sufficient level of ductility. It
is shown that while using high strength ties does not increase the strength of high strength
concrete columns, they can improve the ductility. Design guidelines are suggested for
strength and for maximum tie spacings and tie arrangements.

8. Number of pages 36
 3

Table of Contents

1. Introduction ...................................................................................................................... 4

2. The Behaviour of Concentrically Loaded HSC Columns ............................................ 5

3. The Effect of Load Eccentricity Early Cover Spalling ............................................... 13

4. The Rectangular Stress Block - Theory ....................................................................... 15

5. A Modified RSB Model For AS3600. ........................................................................... 19

6. Comparison of RSB Models .......................................................................................... 22

7. Design For Ductility ....................................................................................................... 23

8. Conclusions ..................................................................................................................... 31

9. References ....................................................................................................................... 34
 4

1. Introduction

The increase in strength and ductility of normal strength concrete columns afforded by well-

detailed lateral confinement reinforcement is well documented (Sheikh and Uzumeri (1),

Mander et al. (2), Saatcioglu and Razvi, (3)). Questions have been posed, however, as to

whether or not similar detailing is suitable for high strength concrete (HSC) columns and, if

not, what amount of confinement reinforcement is necessary to obtain a satisfactory level of

ductility in high strength concrete columns (Razvi and Saatcioglu (4), Foster and Attard (5),

Pessiki and Pieroni (6)). Existing code provisions for minimum amounts of lateral

reinforcement are based on experiences with normal strength concrete, however, the

influence of concrete strength on effectiveness of confinement needs to be taken into

consideration.

Experimental studies on high strength concrete columns under concentric axial loading have

shown that the strength is affected by spalling of the cover and the inability of the concrete

core to carry increased loads after the cover is shed. Various theories have been postulated

for this observed behaviour, including buckling of the cover shell (Paultre et al. (7)) and

restrained shrinkage in the cover shell combined with shrinkage of the high strength concrete

around the reinforcing steel (Sundaraj and Sheikh (8), Collins et al. (9)). In a recent paper by

Foster et al. (10) it was shown that cracking occurs at the cover-core interface as a result of

the triaxial stress condition induced by confinement of the core. It was also shown that this

interface cracking occurs irrespective of the concrete’s strength.

Experimental studies (Cusson and Paultre (11), Foster and Attard (12), Razvi and

Saatcioglu (13)) have shown that at the point of cover spalling, the tie steel is not at yield.

This leads to the question as to what advantages can high strength ties provide to the
 5

behaviour of a column section given that at the point of cover spalling the confinement is

limited by the tie strain. This issue is addressed in this paper together with design models for

the strength and ductility requirements of HSC columns.

2. The Behaviour of Concentrically Loaded HSC Columns

While all columns designed to AS3600 (14) are required to incorporate a minimum

eccentricity of 0.05D, where D is the overall depth of the section, much insight is gained by

reviewing the behaviour of concentrically loaded columns. The load versus deflection

behaviour of a HSC column under load is shown in Figure 1. As the column is loaded the

displacement increases in a near linear manner. The proportional limit, as shown in Figure 1,

is dependent on the strength of the concrete and can be as high as 80 percent of the spalling

load for very high strength concrete. Beyond the proportional limit, micro-cracking within

the core gives the increased lateral expansion necessary to activate the confining pressure

provided by the tie reinforcement. High strength concrete has a higher proportional limit than

normal strength concrete leading to a lower expansion of the core at the peak load and, thus,

lower tie strains at the spalling load.

cover spalling load

second peak load

proportional limit
Load

residual strength

Displacement

Figure 1 – Load-deflection behaviour of HSC columns

 6

While for HSC columns the strains in the ties at the peak load are small, they are not

insignificant. In a recent paper by Foster et al. (10), it was shown that the initiation of cover

spalling is an inevitable consequence of the triaxial stress state in the cover. Further, the

mechanism behind the initiation of the separation exists equally for conventional strength as

for high strength concretes and within the constraint σ 3 < f cp ; where σ 3 is the axial

compression stress in the concrete and f cp is the uniaxial strength of the in-place concrete. In

providing ties to constrain the core tension stresses are set up at the cover-core interface, as

shown in Figure 2. The greater the amount of confinement provided to the core, the greater

the tension stress at the cover-core interface due to the higher restraint provided to the core

against Poisson’s growth. Figure 3 shows the load path for the stresses at a point at the

cover-core interface for a circular column (although the same applies for the case of

rectangular sections). The point is in a triaxial tension-compression-compression stress state,

with the tension in the radial direction. The circumferential compression is small, relative to

the axial compression, and has only a small influence on the behaviour. Cracking occurs at

the interface of the cover and the core once the internal stresses reach the failure surface. The

experimental and numerical evidence (Liu et al. (15)) suggests that this is at an axial stress of

the order of 0.85 f cp and at a tensile stress of the order of 0.26 f t , where f t is the uniaxial

tension strength. Once cracking occurs at the cover-core interface the cover is free to buckle

or spall away from the section. This process is possibly accelerated by dilation of the

longitudinal steel after yielding.

When the cover spalls the core is required to take the additional load shed by the cover. In

tests on HSC columns by Cusson and Paultre (11) it was observed that the cover spalled

suddenly causing a drop in load of 10-15 percent. In the finite element studies of Liu et

al. (15) it was shown that the drop in load is a function of the ratio of the area of the cover to
 7

Figure 2 - The mechanics of cover spalling (after Foster et al., 10).

-1.4
σ 3/fcp

load path -1.2

-1 σ2 = 0.02σ3
cover-core
interface
cracking -0.8

-0.6
triaxial
failure
surface -0.4

-0.2

0.08 0.06 0.04 0.02 0 -0.02 -0.04 -0.06 -0.08

σ 1/fcp

Figure 3 – Load path for points along the cover-core interface.

the gross cross sectional area. In poorly confined columns, or columns with a high cover area

to gross area ratio, the dynamic effects set-up by the spalling of the cover leads to a sudden

and catastrophic failure. In well-detailed sections expansion of the core activates the

confinement provided by the ties and failure is arrested. At this point the column may again

carry increased load to a second peak. The second peak load may be higher or lower than
 8

spalling load and depends on the volumetric ratio of the ties, the yield strength of the ties, the

strength of the concrete and the efficiency of the tie arrangements. The second peak load

corresponds to yielding of the ties and the axial strain at the peak load is a function of the tie

yield strength.

Sheikh and Uzumeri (16) proposed a procedure to determine the capacity of the confined

core based on a conceptual model in which it was postulated that the area of the effectively

confined concrete in a column is less than the core area (refer Figure 4). The effectively

confined concrete is assumed to arch between the points where the lateral steel exerts a

confining pressure on the concrete. In the case of rectilinear lateral reinforcement, the area of

effectively confined core is less than the core area even at the tie level and is further reduced

away from the tie level. In the case of circular or spiral ties, the reduction of the core area to

an effectively confined area takes place only along the longitudinal axis of the column.

Mander et al. (2), Saatcioglu and Razvi (3) and Cusson and Paultre (17) further refined the

concept advanced by Sheikh and Uzumeri. In these studies it was hypothesised that the

confining pressure on the core dissipates between ties and away from longitudinal bars with

the full confining pressure acting over a reduced or effective core area, Aeff . The core area,

Acore , is normally defined by the enclosed area inside the perimeter of the center lines of the

outer ring of spirals or ties. The effective core area is given by

Aeff =ke Acore (1)

where ke is the confinement effectiveness coefficient ( k e ≤ 1) . The confining pressure is

calculated by assuming the ties to have yielded and the equilibrating stresses on the core to be

uniformly distributed. Examples of calculating the confining stress ( fr ) for some common

sections are shown in Figure 5.

 9

(b)

Effectively
confined
core Cover

Poorly (c)
confined
core

(a)

Figure 4 - Effectively confined area in tied concrete columns; a) square and b) circular
sections; c) 3D view of a square column.

fr =
2 Atie f yt
fr =
4 Atie f yt
fr =
(
2 Atie f yt 1 + 2 )
ds s bc s bc s
(a) (b) (c)

Figure 5 - Calculation of confining pressures for some common section types.

 10

For circular columns with tie or spiral reinforcement, assuming a parabolic arch between the

ties with a 45 degree tangent slope following the concept advanced by Sheikh and

Uzumeri (16), the confinement effectiveness factor is

2
⎛ s* ⎞
ke = ⎜⎜ 1 − ⎟
⎟ (2)
⎝ 2d s ⎠

where s* is the clear spacing between the ties or spirals as used by Mander et al. (2) and d s

is the diameter of the tie or spiral reinforcement. For square or rectangular sections a

modified form of the Sheikh and Uzumeri model is used with the confinement effectiveness

parameter given by

⎛ 1 n ⎞ ⎛ s* ⎞ ⎛ s* ⎞
ke = ⎜ 1 −
⎝ α Acore
∑ wi2 ⎟⎜
× 1−
⎠ ⎝ 2bc ⎠
⎟ ⎜1 −
⎝ 2dc ⎠
⎟ (3)
i =1

where wi is the ith clear distance between adjacent tied longitudinal bars, bc and d c are the

core dimensions to the centreline of the ties across the width and depth of the section,

Acore = bcd c , n is the number of spaces between tied longitudinal bars and α is constant

which is equal to 6 if the arches of the effectively confined concrete are assumed to be

parabolic and have an initial tangent of 45 degrees.

The core strength ( fo ) can be obtained using a modified form of the Richart et al. (18)

equation

f o = f cp + ke C f r (4)

where f cp is the unconfined in-situ concrete strength and C is a confinement parameter. For

the analyses that follow, the confinement parameter was taken as C = 4 for f c′ ≤ 80 MPa and
 11

C = 3 for f c′ > 80 MPa , as recommended by the FIP/CEB (19). The capacity of the confined

core including the longitudinal reinforcement is then given by

Pcore = f o ( Acore − Ast ) + Ast f sy (5)

where Ast is the area of longitudinal tension reinforcement and f sy is the yield strength of

the longitudinal reinforcement.

In Figure 6a, the data of Sheikh and Uzumeri (1), Toklucu and Sheikh (20), Cusson and

Paultre (11) and Razvi and Saatcioglu (13) has been normalised against the squash load and

plotted against the cylinder strength. In Figure 6a, the squash load Po is calculated as

Po = k3 f c′ Ac + Ast f sy (6)

where f c′ is the measured cylinder strength, Ac is the gross area of concrete in the section,

and k3 is a factor to account for the difference in the cylinder strength to that of the in-situ

( )
concrete k3 = f cp f c′ . Razvi and Saatcioglu (13) measured k3 ≈ 0.9 and this value was

taken for all specimens compared in Figure 6. Figure 6a shows that for cylinder strengths

greater than 80 MPa the peak load is approximately 0.85 times the squash load calculated by

Eq. (6). From the evidence presented, the spalling load can be taken as

Pspall = 0.85k3 f c′ Ac + Ast f sy (7)

and the maximum capacity of the section is then obtained from

(
Pmax = max Pspall , Pcore ) (8)
 12

2 Razvi and Saatcioglu (1996)

Cusson and Paultre (1994)
1.8
Sheikh and Uzumeri (1980)
1.6 Toklucu and Sheikh (1992)

1.4

1.2
Ptest /No

0.8

0.6

0.4

0.2

0
20 30 40 50 60 70 80 90 100 110 120 130 140
Cylinder Strength (MPa)

(a)

2 Razvi and Saatcioglu (1996)

Cusson and Paultre (1994)
1.8
Sheikh and Uzumeri (1980)
1.6 Toklucu and Sheikh (1992)

1.4
Ptest / Pmax

1.2

0.8

0.6

0.4

0.2

0
20 30 40 50 60 70 80 90 100 110 120 130 140
Cylinder Strength (MPa)

(b)
Figure 6 - Comparisons between experimental data normalised against the squash load
for conventional and HSC columns loaded in concentric compression.
 13

In Figure 6b, the experimentally obtained peak loads are normalised against the theoretical

capacity including the longitudinal steel, as calculated using Eqs. 5, 7 and 8. Figure 6b

indicates that for all strength concrete the cover is ineffective after the initiation of the

cover-core cracking. The peak load then becomes the greater of the spalling load (taken as

the load corresponding to the commencement of cover-core cracking) and the capacity of the

effectively confined core (as given by Eq. 8). Comparison of Figures 6a and 6b, however,

indicate that the spalling loads are only of concern for columns with f c′ ≥ 60 MPa .

3. The Effect of Load Eccentricity Early Cover Spalling

Ibrhim and MacGregor (21) analysed data from a number of experimental studies for

concentrically, and eccentrically, loaded columns. Two of their figures are reproduced in

Figure 7. Comparing Figures 7a and 7b suggests that the early cover spalling effect is not

significant for eccentrically loaded sections. Once sufficient bending is introduced such that

d n ≤ D , where d n is the depth of the neutral axis measured from the extreme compression

fibre, no reduction in load occurs due to early spalling of the cover. The reasons for this

remain unclear.

Figure 7a shows that k3 = 0.85 is a reasonable lower bound for eccentrically loaded columns.

For concentrically loaded columns, Figure 7b shows a lower bound of k3k4 = 0.72 , where k 4

is introduced to account for the effect of cover spalling on the squash load.
 14

1.2

1.1

0.9
k3

0.8
Schade
Kaar et al.
0.7 Ibrahim et al.
Swartz et al.

0.6
0 20 40 60 80 100 120 140
f'c (MPa)

(a)

1.2 Sheikh et al.

Richardt et al.
Martinez et al.
1.1 Yong et al.
Cusson et al.

1
k3 k4

0.9
k3k4 = 0.85

0.8
k3k4 = 0.72
0.7

0.6
0 20 40 60 80 100 120 140
f'c (MPa)

(b)

Figure 7 - k3 versus concrete strength for (a) eccentrically and (b) concentrically loaded

specimens (after Ibrahim and MacGregor, 21)

 15

4. The Rectangular Stress Block - Theory

For any given stress distribution through a concrete beam-column section, an equivalent

rectangular distribution of stresses can be determined such that

1. the total volume under the true and equivalent rectangular stress blocks are equal, and

2. the centroid of the equivalent rectangular stress block lies at the centroid of the true

stress block.

The relationship between the true and rectangular stress blocks is shown in Figure 8. Point 1

is necessary to ensure equilibrium of forces and point 2 for moment equilibrium. Three

parameters are required to define the equivalent rectangular stress block (refer Figure 8),

namely k1 , k3 and γ (where γ = 2k 2 ). For a given stress-strain relationship, and assuming

that plane sections remain plane, k2 (and therefore γ ) is determined such that the location of

the centroids of the true and rectangular stress blocks coincide. Equilibrium of forces is then

applied to determine k1 such that the volume under the true and rectangular stress blocks are

equal. The k3 parameter represents the ratio of the strength of the in-situ concrete to that of

the standard cylinder and is unvaried for both the true and equivalent rectangular stress

blocks. The k3 factor can be directly determined from tests on concentrically loaded

specimens.

Figure 8 - Equivalent Rectangular Stress Block.

 16

It is generally agreed that as the concrete strength increases the generalised stress-strain curve

becomes increasingly triangular. For the case of a triangular stress-strain relationship, again

assuming that plane sections remain plane, equilibrium gives γ = 2 3 and k1 = 3 4 . These

limits are absolute if a perfectly rational model is to be established and, thus, for all strengths

of concrete

γ ≥2 3 k1 ≥ 3 4 (9)

The squash load of a short concrete column is given by

( )
Po = k3k 4 f c′ Ag − As + As f sy (10)

where Ag is the gross area of the section, Ast is the area of steel through the cross section,

f c′ the cylinder strength and f sy the yield strength of the longitudinal reinforcement. As

discussed above, the parameter k4 is introduced into Eq. 10 to account for premature spalling

of the cover shell.

A number of stress blocks have been proposed for the design of HSC beam-columns,

including those that appear in the design codes CSA-94 (22) and NZS-95 (23). The

parameters developed for these and other models such as those presented by Ibrahim and

MacGregor (21), and Attard and Stewart (24) were developed from a pool of experimental

data which included concentrically loaded HSC column specimens. In these tests it was

observed that the cover shell failed “prematurely” (as measured against expectations

developed from theories developed for conventional strength concrete columns). The

parameters for the different stress block models developed are compared in Table 1,

including the parameters for a modified AS3600 model presented in this report.
 17

Table 1 - Rectangular stress block parameters for HSC ( f c′ in MPa).

Reference α 1 = k1k 3 γ

CSA-A23.3-94 (22) 0.85 − 0.0015 f c′ ≥ 0.67 0.97 − 0.0025 f c′ ≥ 0.67

NZS3101-1995 (23) 1.07 − 0.004 f c′ . − 0.008 f c′

109
0.75 ≤ α1 ≤ 0.85 0.65 ≤ γ ≤ 0.85

Ibrahim and Macgregor (21) 0.85 − 0.00125 f c′ ≥ 0.725 0.95 − 0.0025 f c′

. ( f c′) . ( f c′)
Attard and Stewart (24) −0.1 −0.091
129 ≥ 0.71 γ = 109 ≥ 0.67

Modified AS3600 . − 0.0032 f c′

102 . − 0.008 f c′
109
(refer Section 5) 0.64 ≤ α1 ≤ 0.85 0.67 ≤ γ ≤ 0.85

The Canadian code model (22) was calibrated against a range of experimental data including

data for concentrically loaded HSC columns failing in concentric compression. The squash

load given by CSA-94 is

( )
Po = α1 f c′ Ag − As + As f sy (11)

with α1 = 0.85 − 0.0015 f c′ ≥ 0.67 (12)

The depth of the stress block , for sections in combined bending and compression, is given by

λ = 0.97 − 0.0025 f c′ ≥ 0.67 (13)

 18

Comparing Eqs. 10 and 11 gives α1 = k3k4 . With the rectangular stress block (RSB) shown

in Figure 8 it can then be concluded that Eq. 11 and the RSB are consistent provided k1 ≈ k4 .

Further, noting that in CSA-94 k3 = 0.9 , then the CSA-94 formulation leads to

k1 = k4 = 0.94 − 0.00167 f c′ ≥ 0.75 (14)

which satisfies the limits given by Eq. 9. A similar analogy can be made for the models given

by NZS-95 (23) and by Ibrahim and MacGregor (21).

Attard and Stewart (24) proposed a set of RSB parameters based on a probabilistic model.

The model varies from previous models in that the interaction curve gives the mean section

strength. Thus, if all else was equal, 50 percent of the experimental data would fall each side

of the interaction curve. This differs from other models where the characteristic cylinder

strength in the field is substituted for the mean cylinder strength used in the laboratory.

Attard and Stewart use the squash load given by Eq. 10, with

k3 = 1.05 − 0.0009 f cm (15)

k3k 4 = 0.92 − 0.0009 f cm (16)

and where f cm is the mean cylinder strength in MPa. The mean cylinder strength is taken as

f cm = f c′ + 7.5 MPa (17)

Like previous reviews of the data, the calibration of the Attard and Stewart spalling factor

does not give any insight into the mechanics of cover spalling. The generalisation of the

Attard and Stewart spalling factor to other than rectangular sections and for sections with

various cover/core ratios is yet to be demonstrated. This is discussed further in Section 6

below.
 19

The height of the Attard and Stewart stress block (refer Figure 8) for eccentrically loaded

sections is given by

. ( f c′)
−0.1
k1k 3 = 129 ≥ 0.71 (18)

with a depth of

γ = 1.09( f c′ )−0.091 ≥ 0.67 (19)

Reviewing the data Attard and Stewart used in developing Eq. 15, a constant value of

k 3 = 0.95 represents the data reasonably well. Substituting k 3 = 0.95 into Eq 18 gives

. ( f c′)
−0.1
k1 = 136 ≥ 0.75 (20)

The limit on Eq. 20, however, is not reached until f c′ = 380 MPa .

5. A Modified RSB Model for AS3600.

Much of the work on the development of a rational RSB model for inclusion in the Australian

building code for concrete structures, AS3600 (14), has started with the premise that a new

relationship needs to be determined for all strengths of concrete. This requires calibration of

the models against existing conventional strength concrete data. In this paper a modified

AS3600 RSB model is developed starting with the premise that the existing AS3600 model

performs well over the range of concrete strengths for which it is calibrated; and,

presupposing that a minimalist approach is desirable. It is also noted that the current AS3600

model meets all the boundary constraints for a rational model with k 3 = 0.85 and k 4 = 10
. .

The requirements then are for a rational model that meets the known boundary constraints
 20

(implied by Eq. 9) and should be valid for concrete strengths beyond that currently produced

commercially.

In Figure 6 data is plotted for a number of reinforced columns tested in concentric

compression for a wide range of concrete strengths. The data indicates that inclusion of a

cover spalling mechanism into a design model becomes more important as the concrete

cylinder strength is increased beyond 60 MPa. For reinforced columns cast with concrete

strengths of less than 60 MPa other mechanisms, such as the increase in strength due to

confinement, compensate for the reduction in section capacity due to the onset of spalling.

The effects on capacity may be further disguised by the better ductility of conventional

strength concrete compared to that of HSC. The squash load is taken as the maximum

capacity of the section under concentric loading and is the greater of the capacity of the

section at the point of cover spalling or the capacity of the confined core (see Section 2).

Figure 6 indicates that k 4 can be given by

k 4 = 1.2 − 0.0038 f c′
(21)
0.85 ≤ k 4 ≤ 1.0

In the squash load equation in AS3600 (14) k 3 = 0.85 and k 4 = 10

. . Therefore with the RSB

parameters used in AS3600-1994 it can be interpreted that k1 = 10

. and

2k 2 = 1.09 − 0.008 f c′
(22)
0.67 ≤ 2k 2 ≤ 0.85

The triangular stress block limit is reached at f c′ = 54.2 MPa . The axial load carried by a

section is given by
 21

n
C = k1 2k 2 (k3 f c′ )d nb + ∑ ( As f s )i (23)
i =1

where d n is the distance from the extreme compression fibre to the neutral axis, As is the

area of reinforcing steel at level i, f s is the stress in the steel at level i and n is the number of

layers of reinforcing steel. The moment carried by a symmetric section is

n
M = k12k 2 (1 − k 2 )(k3 f c′ ) d n2 b + ∑ ( As f s )i di (24)
i =1

where di is the distance from the plastic centroid of the section to the reinforcing steel at

level i. The extension to HSC can be obtained by manipulating k1 such that in the limit the

stress block area factor k12 k2 = 0.5 . The k12 k2 factor is the multiplication factor necessary to

convert the area under the true stress block to an equivalent rectangle. The lower limit of

k12 k2 = 0.5 represents the conversion of a triangle to an equivalent rectangle. Of equal

importance is the factor k1 2 k 2 (1− k 2 ) which represents the moment carried by the concrete.

Taking 2 k 2 given by Eq. 14, and setting the triangular stress block limits at

f c′ = 120 MPa (which appears reasonable when reviewing the stress strain curves obtained on

standard cylinders) leads to

k1 = 12
. − 0.0038 f c′
(25)
0.75 ≤ k1 ≤ 10
.

The stress block can now be defined for the case where the neutral axis lies within the section

where γ = 2k 2 is given by Eq. 22, and

α1 = k1k 3 = 102
. − 0.0032 f c′
(26)
0.64 ≤ α1 ≤ 0.85
 22

The squash load is given by Eq. 10 with

α 2 = k3k 4 = 1.02 − 0.0032 f c′

(27)
0.72 ≤ α 2 ≤ 0.85

A straight line is used to join the squash load given by Eqs. 10 and 27 to the point on the

interaction curve defined by Eqs. 22 and 26 with c = D , where D is the depth of section. The

axial compression-bending interaction diagram defined by Eqs. 10, 22, 26 and 27 give

identical curves to the current AS3600 for f c′ ≤ 54.2 MPa and obey the triangular stress

block boundary limits set out in Eq. 9 for the extension to high, and very high, strength

concretes.

6. Comparison of RSB Models

In Figure 9a the factor defining the area under the rectangular stress block ( k1 2 k 2 ) is plotted

for a number of RSB models. The change in slope in the NZS-95 and in the proposed

modified AS3600 model, for increasing concrete strengths, is due to reaching the limits in k 2

and k1 , respectively. Thus while the limits of Eq. 9 are satisfied in that each parameter

approaches that for the triangular stress block, they do so at different concrete strengths. In

the Canadian code (CSA-94) model, the k1 and k 2 limits are reached simultaneously at

f c′ = 120 MPa . In the model proposed by Attard and Stewart, the triangular stress block limit

is not reached until f c′ = 230 MPa , although the limits given by Eqs. 18 and 19 are not

reached until f c′ = 390 MPa and f c′ = 210 MPa , respectively. Figure 9b compares the

moment factor k1 2 k 2 (1− k 2 ) for the CSA-94 (22), NZS-95 (23), Attard and Stewart (24) and

modified AS3600 model. The proposed modified AS3600 model compares favourably with

the CSA-94 and NZS-95 models upto f c′ = 80 MPa . At f c′ = 80 MPa the NZS-95 model hits

its limit, a limit which is significantly higher than that defined by a triangular stress block.
 23

While the CSA-94 model reaches its limit at f c′ = 120 MPa , the calibration of the CSA-94

model was limited to f c′ ≤ 80 MPa (22).

In Figures 10, 11 and 12 comparisons are made for the various stress blocks for a series of

87 MPa 250 mm by 150 mm rectangular columns and 74 MPa and 92 MPa 150 mm by 150

mm square columns, respectively, tested by Foster and Attard (5, 12). Comparisons are made

with k 3 = 0.85 and k 3 = 10

. where it is shown that all the models discussed lead to similar

results. The reasons for the closeness of the RSB models compared in Figures 10 to 12 is

explained by a convergence of the area and moment factors shown in Figure 9 for concrete

strengths in the range f c′ = 70 ~ 100 MPa . Larger differences can be expected between the

CSA-94, NZ-95 and modified AS3600 models for concrete strengths greater than 100 MPa,

and for the Attard and Stewart model for concrete strengths of less than 50 MPa and greater

than 100 MPa.

In Figure 13, the modified AS3600 model is compared with 36 eccentrically loaded columns

tested Lloyd and Rangan (25). A good correlation is shown between the test data and the

model predictions.

7. Design for Ductility

Ductility is an important issue when it comes to the detailing of all concrete members, but is

of particular importance for of HSC columns due to the brittle nature of the concrete. Again it

is worth reviewing the behaviour of columns under concentric loading as this represents the

extreme condition. Ductility in columns is derived from confinement provided to the core

and, thus, is a function of the yield strength of the ties, the concrete strength, the volumetric

ratio of tie reinforcement and the arrangement of the ties. Liu et. al. (15) showed that while
 24

increasing the strength of the ties does not improve the load capacity of a HSC column, it

does give an improvement in the ductility.

One measure of ductility is given by the I10 index, where I10 is calculated similar to that set

out in ASTM C1018 (26) for the measurement of toughness. The I10 parameter is the area

under the load versus strain curve at a strain of 5.5 times the yield strain, relative to the area

under the curve for a strain equal to the yield strain. The yield strain is taken as 1.33 times the

strain corresponding to a load on the ascending curve of 0.75Pu (see Foster and Attard (5)).

The area under the load versus strain curve up to 5.5 times the yield strain is chosen such that

for a perfectly elasto-plastic material I10 = 10 while for a perfectly elastic-brittle material

I10 = 1 .
 25

1
CSA-94
NZS-95
0.9 Attard and Stewart (1998)
Modified ACI-318 model

0.8
k1.2k2

0.7

0.6

triangular stress
0.5
block limit

0.4
0 20 40 60 80 100 120 140 160
Concrete Strength (MPa)

(a)

0.5 CSA-94
NZS-95
Attard and Stewart (1998)
Modified ACI-318 model
0.45
k1.2k2(1-k2)

0.4

0.35
triangular stress
block limit

0.3
0 20 40 60 80 100 120 140 160
Concrete Strength (MPa)
(b)

Figure 9 - Comparison of (a) the area factor; and (b) the moment factor for various
RSB models
 26

fcm = 87 MPa Experimental Data

k 3 = 0.85 CSA-94
NZS-1995
3500
Attard and Stewart (1998)
Modified AS3600 model
3000

2500
Axial Load (kN)

2000
8Y12

1500
W4@60
250

1000

500
150

0
0 20 40 60 80 100 120
Moment (kNm)

(a)

fcm = 87 MPa Experimental Data

k 3 = 1.0 CSA-94
3500 NZS-95
Attard and Stewart (1998)
3000 Modified AS3600 model

2500
Axial Load (kN)

2000 8Y12

1500 W4@60
250

1000

500 150

0
0 20 40 60 80 100 120
Moment (kNm)

(b)

Figure 10 - Comparisons of various RSB models with 250 mm x 150 mm rectangular

columns tested by Foster and Attard (2) (a) k 3 = 0.85 ; and (b) k 3 = 1.0 .
 27

Exp. - 74 MPa
150 CSA-94
NZS-95
2000 Y12 Attard and Stewart (1998)
Modified AS3600 model

150
W6

1500
Axial Load (kN)

1000

fcm = 74 MPa
500
k 3 = 0.85

0
0 10 20 30 40 50
Moment (kNm)

(a)

150 Exp. - 74 MPa

CSA-94
Y12 NZS-95
2000 Attard and Stewart (1998)
150

W6 Modified AS3600 model

1500
Axial Load (kN)

1000

fcm = 74 MPa
500
k 3 = 1.0

0
0 10 20 30 40 50
Moment (kNm)

(b)

Figure 11 - Comparisons of various RSB models with 74 MPa 150 mm x 150 mm

columns tested by Foster and Attard (10) (a) k3 = 0.85 ; and (b) k3 = 1.0 .
 28

150
Experimental
2500 Y12 CSA-94
NZS-95

150
W6 Attard and Stewart (1998)
2000 Modified AS3600 model
Axial Load (kN)

1500

1000

fcm = 92 MPa
500 k 3 = 0.85

0
0 10 20 30 40 50 60
Moment (kNm)

(a)
150
Experimental
2500 Y12 CSA-94
150

W6 NZS-95
Attard and Stewart (1998)
Modified AS3600 model
2000
Axial Load (kN)

1500

1000

fcm = 92 MPa
500 k 3 = 1.0

0
0 10 20 30 40 50 60
Moment (kNm)

(b)

Figure 12 - Comparisons of various RSB models with 92 MPa 150 mm x 150 mm

columns tested by Foster and Attard (10) (a) k3 = 0.85 ; and (b) k3 = 1.0 .
 29

2500

k3 = 0.85
2000
Predicted Axial Load (kN)

1500

1000

Test/Predicted
500 Mean = 1.08
StdDev = 0.11

0
0 500 1000 1500 2000 2500
Peak Axial Load Test (kN)

Figure 13 - Comparison of peak axial loads versus predicted axial load using the
modified ACI-318 model for 36 columns tested by Lloyd and Rangan (25).

A number of authors (4, 27, 28, 29) have indicated that ductility is a function of the

confinement parameter ρ s f yt fc' where ρs is the lateral reinforcement volumetric ratio, fyt is

the yield strength of the tie reinforcement and f c′ is the concrete strength. Razvi and

Saatcioglu (13) suggested that this parameter be multiplied by a second parameter

representing the efficiency of the tie reinforcement arrangement. In Figure 14, the I10

ductility index for 40 columns tested by Razvi and Saatcioglu (13) is plotted against

ke ρ s f yt f c′ . The hollow data markers represent the results of columns where large changes in

load were recorded. For these columns an exact value for I10 can not be obtained, but upper

and lower bounds can be calculated. The data point is plotted at the average of the upper and
 30

10 Circular columns
Square columns

7
I10

6
CC-17
CC-18
5

4 CS-5

3
1 2 3 4 5 6 7 8 910 20 30
ke ρ s f yt f c′ (%)

Figure 14 – I10 ductility parameter versus the effective confinement ratio

lower bound and have a possible error in I10 of approximately ± 0.8 . From Figure 14, a

relationship between the ductility index and the effective confinement parameter is obtained,

that is

(
I10 = 1.9 ln 1000 ke ρ s f yt f c′ ) (28)
or
1.7 I10
ke ρ s f yt f c′ = (29)
1000

While more data is needed for columns with low amounts of confining steel, some important

design conclusions can be determined from the data available. Specimens CC-17 and CC-18

(refer Figure 14) had no longitudinal reinforcement. Thus, it is concluded that the

arrangement of the longitudinal reinforcement is an important parameter in obtaining

ductility, even in circular columns. In column CS-5, 1000 MPa ties were used to increase the

confinement ratio, however, the tie spacing was 120 mm (D 2) . For most of the columns with
 31

I10 < 8 sudden changes in the load-strain data were recorded. In no case where the tie

spacing was greater than (D 2.5) , or the distance between tied longitudinal bars was greater

than (D 2.5) , did the columns perform adequately with respect to ductility, regardless of the

strength of the ties. For regions of low to moderate seismicity satisfactory ductility may be

obtained with I10 ≥ 8 . Therefore, from Eq. 29, adequate ductility is achieved provided that

the effective confinement parameter is ke ρ s f yt fc′ ≥ 0.07 and that the tie spacing, and the

spacing between tied longitudinal bars, is not greater than (D 2.5) .

Applying the above principles for the hinge region of a 600 mm square column cast with

80 MPa concrete, having a uniform arrangement of 12 Y36 bars longitudinally, 40 mm of

cover and 4 legged 16 mm diameter ties in each direction, requires 500 MPa ties at 100 mm,

700 MPa ties at 140 mm or 1000 MPa ties at 180 mm centres to achieve a ductility level of

I10 = 8 . Outside the localised zones the tie spacing may be increased but should not exceed

the lesser of D/2 and 300 mm.

8. Conclusions

In this paper, research on the behaviour of high strength concrete columns under combined

axial load and bending has been reviewed towards the synthesis of a rational rectangular

stress block model for the design of conventional and high strength concrete columns. Tests

have shown that concentrically loaded columns fabricated with HSC fail at loads lower than

that their theoretical squash loads, where the squash load is calculated from the uniaxial

strength of plain concrete and the total cross-sectional area. Whilst the early cover spalling of

concentrically loaded HSC concrete columns has long been recognised, the mechanics of

early cover spalling are not well understood. Finite element modelling of concentrically
 32

loaded circular columns shows that a cracking plane develops at the cover core interface

when the axial stress in the concrete is less than the concretes uniaxial compressive strength.

This cracking occurs when the tensile stress at the cover-core interface reaches its triaxial

strength limit, which is considerably below the uniaxial tension strength of the concrete.

Once cover-core interface cracks have developed, the cover concrete is free to spall or buckle

away.

Analysis of experimental data on concentrically loaded square and circular columns shows

that the strength of the section is limited by either the spalling load or by the capacity of the

confined core. For columns with moderate to high quantities of tie reinforcement, typical of

HSC columns, the spalling load can be taken as 0.85 times the squash load where the squash

load is calculated on the gross cross sectional area. In this paper, a semi-rational model is

developed where the early cover spalling is included in calculating the squash load, and is

separated from the RSB parameters developed for eccentrically loaded sections. A modified

version of the AS3600 compression-bending interaction model has been developed starting

with the premise that the current model works well for conventional strength columns. Also

discussed in this paper are the boundary conditions that a rational RSB model must satisfy.

The modified AS3600 model developed is calibrated such that for concrete with

f c′ ≥ 120 MPa the equivalent RSB corresponds to a triangular stress block. The proposed

modified AS3600 model shows a good correlation against other stress block models and

against experimental data for f c′ ≤ 100 MPa .

The ductility of reinforced concrete columns is dependent on the confinement provided by

the transverse reinforcement. The confinement effectiveness is related to the concrete

strength, the tie spacing and configuration, the yield strength of the ties, the tie diameter, the

cover and the arrangement of the longitudinal reinforcement. A parameter that reflects the
 33

level of confinement is the effective confinement parameter ke ρ s f yt f c′ . To obtain a ductility

level of I10 ≥ 8 an effective confinement parameter of ke ρ s f yt f c′ ≥ 0.07 is required but with

the additional condition that the spacing of the ties, and the spacing between tied longitudinal

bars, should not exceed the lesser of (D 2.5) and 300 mm. Outside of the hinge region the

spacing of the ties can be relaxed but should not exceed the minimum requirements of

(D 2.5) and 300 mm. For columns with high axial loads and low moments, the hinge region

may be difficult to identify. In these columns the requirement that ke ρ s f yt f c′ ≥ 0.07 should

extend for the length of the column.

 34

9. References

1. Sheikh, S.A., and Uzumeri, S.M., “Strength and ductility of tied concrete columns”, J. of
Struct. Engrg., ASCE, Vol. 106, No. 5, 1980, pp. 1079-1102.

2. Mander J.B., Priestley M.J.N., and Park R., “Theoretical stress-strain model for confined
concrete”, ASCE, Journal of Structural Engineering, Vol. 114, No. 8, 1988, pp.
1804-1825.

3. Saatcioglu, M., and Razvi, S.R., “Strength and ductility of confined concrete”, J. Struct.
Engrg., ASCE, Vol. 118, No. 6, 1992, pp. 1590-1607.

4. Razvi, S.R., and Saatcioglu, M., “Strength and Deformability of Confined High-Strength
Concrete Columns”, ACI Structural Journal, Vol. 91, No. 6, 1994, pp. 678-687.

5. Foster, S.J. and Attard, M.M., “Experimental tests on eccentrically loaded high strength
concrete columns”, ACI Structural Journal, Vol. 94, No. 3, 1997, pp. 2295-2303.

6. Pessiki, S. and Pieroni, A., “Axial load behavior of large-scale spirally-reinforced high-
strength concrete columns”, ACI Struct. Journal, Vol. 94, No. 3, 1997, pp. 304-314.

7. Paultre, P., Khayat, K.H, Langlois, Y., Trundel, A., and Cusson, D., “Structural
Performance of Some Special Concretes”, 4th International Symposium on Utilization of
High-Strength/High-Performance Concrete, Paris, 1996, pp.787-796.

8. Sundararaj, P., and Sheikh, S.A., “High Strength Concrete Columns Under Eccentric
Loads”, Department of Civil Engineering, University of Toronto, Canada, Research
Report, 1992.

9. Collins, M.P., Mitchell, D., and MacGregor, J.G., “Structural Design Considerations for
High-Strength Concrete”, Concrete International, Vol. 15, No. 5, 1993, pp. 27-34.

10. Foster, S.J., Liu, J. and Sheikh, S.A., “Cover spalling in HSC columns loaded in
concentric compression”, Journal of Structural Engineering, ASCE, Vol. 124, No. 12,
1998.

11. Cusson, D. and Paultre, P. “High strength concrete columns confined by rectangular ties”,
J. of Struct. Engrg., ASCE, Vol. 120, No. 3, 1994, pp. 783-804.
 35

12. Foster, S.J. and Attard, M.M., “Ductility and strength in HSC columns”, First
Engineering Foundation conference on high strength concrete, Kona, Hawaii, July 13-18,
1998.

13. Razvi, S.R and Saatcioglu, M. (1996). “Tests of high strength concrete columns under
concentric loading”, Dept. Of Civil Eng., University of Ottawa, Report OCEERC 96-03,
1996, 147 pp.

14. AS3600, “Concrete Structures Code”, Standards Association of Australia, 1994.

15. Liu, J., Foster, S.J., and Attard, M.M., “Behaviour of tied high strength concrete columns
loaded in concentric compression”, UNICIV Report R-372, School of Civil and
Environmental Engineering, The University of New South Wales, Sydney, Australia,
1998, 143 pp.

16. Sheikh, S.A. and Uzumeri, S.M., “Analytical model for concrete confinement in tied
columns”, J. of Struct. Engrg., ASCE, Vol. 108, No. 12, 1982, pp. 2703-2722.

17. Cusson, D. and Paultre, P., “Stress-strain model for confined high-strength concrete”, J.
of Struct. Engrg., ASCE, Vol. 121, No. 3, 1995, pp. 468-477.

18. Richart, F.E., Brandtzaeg, A. and Brown, R.L., “The failure of plain and spirally
reinforced concrete in compression”, Bulletin 190, University of Illinois, Engineering
Experimental Station, Champaign, Illinois, 1929, 74 pp.

19. FIP/CEB, “High strength concrete - state of the art report”, Federation Internationale de
la Precontrainte, London, 1990, 61 pp.

20. Toklucu, M.T., and Sheikh, S.A., “Behaviour of reinforced concrete columns with
circular spirals and hoops”, Department of Civil Engineering, University of Toronto,
Canada, Research Report, 1992, 330 pp.

21. Ibrahim, H.H.H., and MacGregor, J.G., “Modification of the ACI rectangular stress block
for high-strength concrete”, ACI Structural Journal, Vol. 94, No. 1, 1997, pp. 40-48.

22. CSA-A23.3-94, “Design of Concrete Structures”, Canadian Standards Association, 1994.

23. NZS-3101, “Design of Concrete Structures”, Standards New Zealand, Wellington,1995.

 36

24. Attard, M.M., and Stewart, M.G., “An improved stress block model for high strength
concrete”, Research Report No. 154.10.1997, Department of Civil, Surveying and
Environmental Engineering, University of Newcastle, Newcastle, Australia, 1998, 42 pp.

25. Lloyd, N.A., and Rangan, B.V., “High strength concrete columns under concentric
compression”, Research Report No. 1/95, School of Civil Engineering, Curtin University
of Technology, Perth, Western Australia, 1995, 199 pp.

26. ASTM C1018, “Standard Test for Flexural Toughness and First-Crack Strength of Fibre-
Reinforced Concrete (Using Beam with Third-Point Loading)”, 1992, pp. 514-520.

27. Martinez, S., Nilson A.H., and Slate, F.O., “Spirally Reinforced High-Strength Concrete
Columns”, ACI Journal, Proceedings Vol. 81, No. 5, 1984, pp. 431-442.

28. Bjerkeli, I., Tomaszewicz, A. and Jensen, J.J., “Deformation Properties and Ductility of
Very High Strength Concrete”, Utilization of High Strength Concrete - Second
International Symposium, SP-121, American Concrete Institute, Detroit, 1990,
pp. 215-238.

29. Sugano, S., Nagashima, T., Kimura, H., Tamura, A. and Ichikawa, A., “Experimental
Studies on Seismic Behaviour of Reinforced Concrete Members of High Strength
Concrete,” Utilization of High Strength Concrete - Second International Symposium,
SP-121, American Concrete Institute, Detroit, 1990, pp. 61-87.