ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 94-S39

A Performance-Based Approach for the Design of

Confining Steel in Tied Columns

by Shamim A. Sheikh and Shafik S. Khoury

A review of the development over the years of the ACI Code provisions for HISTORICAL REVIEW OF THE ACI CODE
confinement is presented. Based on the available experimental evidence, PROVISIONS FOR CONFINEMENT
the current Code requirements for the amount of confinement steel in tied The basic philosophy of the current ACI Code1 require-
columns are critically evaluated. It was concluded that the behavior of col- ments for confining steel is to maintain the axial load
umns designed according to the ACI Code may vary from unacceptably carrying capacity of the column after spalling of the cover
brittle to very ductile. While the amount of Code-required steel can be concrete. This philosophy is obviously based on strength
reduced in many cases, much larger amounts of lateral steel are needed in enhancement due to confinement. Ductility is not given due
other cases. A new design procedure is proposed in which the amount of importance although it is implied that the lateral steel would
lateral steel required is a function of the column ductility performance. The enhance section and member ductility. The ACI design
lateral steel content increases with an increase in the level of axial load, provisions for confining steel have changed over the past 30
and depends on steel distribution and the extent of lateral restraint pro- years through six editions of the Code from 1956 through
vided to the longitudinal bars. For any specific steel configuration, the pro- 1995. It is believed that some of the changes made from one
cedure lends itself to a simple design chart. The proposed method when edition to another are not suitable to provide sufficient
applied to realistically-sized specimens tested by different investigators ductility in reinforced concrete columns.
yielded excellent agreement with the experimental results. The 1956 and 1963 Codes—Neither of these Codes6
contained any equations to calculate the amount of confining
Keywords: columns (supports); confined concrete; standards; structural steel in tied columns. Both Codes required that at least #2
design; tied columns. bars (6.4 mm) be used for ties that are spaced apart not more
than 16 longitudinal bar diameter, 48 tie diameter, and the
INTRODUCTION least section dimension. No mention was made of ductility,
The need for ductile behavior of various structural confinement, or plastic hinge. However, the following equa-
components during a major earthquake has been demon- tion was suggested to calculate the volumetric ratio of
strated repeatedly during several seismic events. Although circular spiral reinforcement
it is preferable to dissipate seismic energy by post-elastic
deformations in beams, column hinging cannot be avoided
A f′
entirely in most buildings during severe earthquakes. To ρ s = 0.45 ⎛ -----g- – 1⎞ -----c- (1)
achieve sufficient ductility in columns, their potential ⎝A ⎠f
c yh
plastic hinge regions should be reinforced with appropri-
ately designed and detailed longitudinal and lateral where Ag = gross area of section, Ac = area of the concrete
confining steel. Design provisions for confinement steel in core measured to the outside diameter of spiral, fc′ =
various codes,1-3 however, do not contain adequate quanti- compressive strength of concrete, and fyh = yield strength of
tative relationships between the design parameters and lateral steel.
column performance. For columns subjected to axial load The only significant difference between the two editions
beyond a certain limit, heavy confining steel is required by of code was that related to steel detailing. While the 1956
the North American Codes irrespective of the level of axial Code required all the longitudinal bars to be laterally
load. In addition, there is no consideration given to the supported by tie bends, this requirement was considerably
distribution of longitudinal and lateral steel in a column relaxed in the 1963 edition (see Fig. 1) in which unsupported
which has been found to significantly affect the confine- middle bars were permitted as long as the clear distance
ment mechanism.4,5 As a result, the design may either be
very conservative for columns with well-distributed steel ACI Structural Journal, V. 94, No. 4, July-August 1997.
and subjected to low levels of axial load or unsafe for Received May 16, 1995, and reviewed under Institute publication policies. Copy-
right © 1997, American Concrete Institute. All rights reserved, including the making
columns in which only four corner bars are effectively of copies unless permission is obtained from the copyright proprietors. Pertinent
discussion will be published in the May-June 1998 ACI Structural Journal if
supported by tie bends and axial load is large. received by January 1, 1998.

ACI Structural Journal/July-August 1997 421

Shamim A. Sheikh is a Professor of civil engineering at the University of Toronto. He f′
is chairman of joint ACI-ASCE Committee 441, Reinforced Concrete Columns, a ρ s = 0.12 ---c (2)
member of joint ACI-ASCE Committee 442, Response of Concrete Buildings to Lat- fy
eral Forces, and of ACI Committee 368, Earthquake Resisting Concrete Structural
Elements and Systems. His research interests include confinement of concrete, earth-
quake resistance of reinforced concrete, and expansive cement and its application in The spiral pitch was limited to 3 in. (76 mm) and the
deep foundations.
maximum center-to-center spacing between ties was 4 in.
Shafik S. Khoury is an assistant professor in the Department of Structural Engineer- (102 mm). The minimum cross sectional area of tie (one leg)
ing at Alexandria University, Alexandria, Egypt. He received his PhD from the Uni- was specified as
versity of Houston in 1991. His research activities include concrete materials and
reinforced concrete columns.
lh ρs s
A tie = -----------
- (3)
2

where lh = the maximum unsupported length of the perimeter

tie, s = tie spacing, and ρs is the larger of the value calculated
from Eq. (1) and (2). In the format of the current Code
requirements,1 the amount of tie steel for square columns
with only perimeter ties can be written as

A f′ f′
Ash,c = 0.45shc ⎛⎝ -------g- – 1⎞⎠ -----c- , ≥ 0.12shc -----c- (4)
A ch f yh f yh

where Ach = Ac, Ash = the total cross sectional area of recti-
linear steel perpendicular to dimension hc (Ash = 2Atie), and hc
= lh. Eq. (4) assumes that for columns with square perimeter
hoops only, the efficiency of rectilinear confining steel is 50
percent of that of spirals. For the case of square hoop with one
supplementary crosstie in each direction, the implied effi-
Fig. 1—Reduction of interior ties in 1963 ACI Code. ciency of the rectilinear ties is increased to 66 percent of that
of spirals. Application of lateral pressure on the concrete core
at larger number of points results in better confinement;4
between an unsupported bar and a supported bar did not therefore the Code’s assumption was quite rational. The
exceed 6 in. (152 mm). This change was primarily based on minimum bar size allowed for ties was also increased from #2
experiments in which the ultimate column strength was the (6.4 mm) as specified in the 1963 Code, to #3 (9.5 mm) for
only criterion used. Since no attempt was made to evaluate longitudinal bars #10 (31.8 mm) or smaller and at least #4
ductility, this relaxation in interior ties appeared plausible.
(12.7 mm) for #11 (34.9 mm) or larger longitudinal bars.
Whereas the change was technically sound for most steel
arrangements, allowing perimeter ties only for all situations The 1983 Code—The maximum tie spacing was changed
is not appropriate as it is now well-known that columns with from 4 in. (102 mm) to the smaller of 4 in. (102 mm) and
only four corner bars supported by tie bends may fail in a one-quarter of the minimum section dimension. While the
brittle manner.7-9 The single perimeter tie is not able to requirements for the amount of spiral reinforcement were
support the middle longitudinal bars effectively after cover similar to those specified in the previous edition [Eq. (1) and
spalling; these bars would buckle and push the ties outward, (2)], the total cross sectional area of rectilinear lateral steel
thus releasing a considerable amount of confinement. The (including crossties) was given by
1956 Code provided very efficient steel detailing which
would have provided excellent confinement with small tie ′
A f f′
spacing. Since 1963, this provision of the Code has not Ash,c = 0.3shc ⎛⎝ -------g- – 1⎞⎠ ---c , ≥ 0.12shc ---c (5)
changed. A ch fy fy
The 1971 and 1977 Codes—The special provisions for
seismic design were introduced in the 1971 edition of the It can be seen that this requirement is similar to that given
Code in Appendix A and were retained without any substan- by Eq. (4) except for the numerical coefficient 0.45 which has
been reduced to 0.3. No clear explanation for this change from
tial changes in the 1977 Code. The importance of ductility
the 1977 Code was provided. From Eq. (1), (2), and (5), it can
was outlined, and related significant terms were defined. The be shown that for columns with square perimeter ties, the effi-
plastic hinge was defined as the region where ultimate ciency of ties as confining steel varied from 50 to 75 percent
section moment capacity may be developed and maintained of that of spirals (Fig. 2). As for spirals, the lower limit in Eq.
while the inelastic deformation is increased significantly. (5), which is applicable to columns in which Ag /Ach ≤1.4 sets
The concept of “Strong Column-Weak Beam” was intro- the minimum confinement for the purpose of ductility.
duced in an attempt to prevent column hinging. The volu- The Code also required that this lateral steel be distributed
metric ratio of spiral ρs was given as in Eq. (1) with the lower over regions where inelastic action is considered to be likely.
limit provided by Eq. (2) that will be applicable to large The length of this region was defined to be above and below
columns in which Ag/Ac is less than 1.27. each connection and on both sides of any section where flex-

422 ACI Structural Journal/July-August 1997

ural yielding is likely to occur, or in other words, where plastic
hinges are expected. The use of crossties with a 180 deg hook
at one end and a 90 deg hook at the other end was allowed for
the first time to provide ease of construction. Efficiency of the
90 deg hooks in confining the concrete core and preventing the
premature buckling of longitudinal bars has proven to be some-
what doubtful, especially under high axial load levels.7-9
The 1989 and 1995 Codes—In view of the importance of
providing reinforced concrete structures with adequate
toughness to respond inelastically under severe seismic
attacks, Appendix A was moved to form Chapter 21 in the
main body of the 1989 Code.1 Eq. (1), (2), and (5) remained
unchanged except that the factor 0.12 in Eq. (5) was changed
to 0.09. This change was based on the observed behavior of
tied columns, which had properly detailed hoops and cross- Fig. 2—Comparison of effectiveness of spirals and rectilinear
ties, and made the relative efficiency of rectilinear lateral ties.
reinforcement reasonably uniform for all sizes of columns
(Fig. 2). Figure 3 shows a comparison of the behavior of
columns with different steel arrangements tested under
concentric compression.10,11,14 It is clear that the relationship
between the efficiencies of circular and rectilinear lateral
steel is not as simple as assumed in the code. The following
points can be made in this regard: 1. The current code equa-
tions assume that all steel configurations in tied columns
result in similar column behavior. Extensive experimental
evidence indicates that section ductility and strength varied
significantly from one configuration to another7-12 under
axial load only as well as under combined axial load and
bending moment; 2. The Code philosophy of maintaining
axial load capacity of a section after spalling of the cover
concrete ignores the most important parameter, ductility,
when a column is subjected to axial load and flexure. A
measure of ductility should be included in the design equa-
tions; 3. The axial load level in the column has a great effect
on the column behavior.7-9,12,13 Since considerably high
levels of axial load are permitted by many codes for seismic Fig. 3—Comparison of circular and rectilinear confinement.
design of columns, the detrimental effects of higher axial
loads must be compensated for by using larger amounts and mental program. This procedure includes the effects of two
more efficient configuration of lateral steel. The confine- additional variables which are not considered in the current
ment equations in the 1982 NZS Code2 were almost similar Code equations and have been proven4,5,7-13 to significantly
to Eq. (5) except for the additional multiplication factor of affect the confinement effectiveness and consequently the
0.5 + 1.25(Pe /φfc′Ag) where Pe is the column axial load, and column behavior. These variables are the level of axial load and
φ is the strength reduction factor. At values of Pe /φfc′Ag less the steel configuration. In the proposed method the required
amount of confining steel is increased with an increase in
than 0.4, the ACI Code was more conservative than the 1982
ductility demand. The proposed procedure lends itself to a
NZS Code and for higher axial load, the NZS Code requires design chart which should be of interest to researchers, and
more lateral steel than the ACI Code, up to 50 percent more practitioners engaged in the design of ductile moment-
for Pe/φfc′Ag equal to 0.75. Neither code attempted to quan- resisting reinforced concrete frames.
titatively relate the required amount of lateral steel to column
performance. In the 1995 version of the New Zealand Code,2 PROPOSED APPROACH
the amount of lateral steel is related to the level of axial load The proposed approach was developed using the experi-
and is aimed at producing highly ductile columns. mental results of twenty-nine large-sized specimens reported
elsewhere.7-9,14 Initial development of this design procedure
RESEARCH SIGNIFICANCE is given in Ref. 15. Some background data and the procedure
is briefly explained here.
In the light of the previous research and ensuing conclusions
Column performance—In evaluating the column perfor-
about the inability of the existing ACI Code provisions for
confinement to provide columns with an adequate level of mance and studying the effects of different variables,
ductility in many circumstances, the need for a procedure for the ductility and toughness parameters defined in Fig. 4 were
design of confining steel becomes clear. Such a procedure that used. These include curvature ductility factor μφ, cumulative
relates confinement parameters to the column performance is ductility ratio Nφ, and energy-damage indicator E. Wherever
proposed here on the basis of results from an extensive experi- used, subscripts t and 80 indicate, respectively, the value of

ACI Structural Journal/July-August 1997 423

Fig. 4—Ductility parameters.

Fig. 5—Relationships between curvature and displacement

ductility factors.

the parameter until the end of the test (total value) and the
value until the end of the cycle in which the moment is
dropped to 80 percent of the maximum value. Energy param-
eter ei represents the area enclosed in cycle i by the M-φ loop.
All other terms are defined in Fig. 4 except Lf and t which
represent the length of the most damaged region and section
depth of the specimen, respectively. The energy-damage indi-
cator E is similar to the one proposed by Ehsani and Wight16
for force-deflection curves. Table 1 lists the available ductility
parameters for all the specimens considered in this analysis.
To relate various ductility parameters, energy index E80
and cumulative ductility ratio Nφ80 are plotted against curva-
ture ductility factor μφ in Fig. 5. Data from nine similar spec-
imens that were tested under similar conditions with constant
axial load and cyclic lateral loads were used in the construc- Fig. 6—Effects of axial load and concrete strength.
tion of this figure. A reasonable correlation exists between
the parameters in the figure. For μφ of 16, the values for Nφ80
and E80 are 64 and 575, respectively. A column section with found to depend on several factors such as length L, section
this level of deformability is defined as highly ductile. With size, and longitudinal bar diameter, but it is unaffected by
a μφ value of 8 to 16, the section is defined as moderately parameters that comprise confining steel.7,8,13,17 Since
ductile and the low ductility column has μφ < 8. With this confinement of concrete in columns will only affect μφ
correlation between ductility parameters, the specimens directly, curvature ductility rather than displacement
tested under monotonic flexure (last 15 specimens in Table 1) ductility is therefore used as a parameter in the proposed
could also be considered in the analysis. In typical columns procedure. For drift-based design story/column drift can then
of framed structures curvature ductility factor μφ and be easily calculated using μφ and plastic hinge length for
displacement ductility factor μΔ are directly related. specific geometric and loading conditions.
Assuming an elasto-plastic section response and constant Axial load level—Increased axial load reduces ductility
curvature over an equivalent plastic hinge length (Lp), the significantly.7-9 The level of axial load is generally
main variables that affect the relationship between μφ and μΔ measured by indices P/fc′Ag and P/Po. For columns with
are Lp, the column length L between the point of maximum similar fc′, both these indices provide similar comparison.
moment and the point of contraflexure and the type of lateral However, for different fc′ values in columns the comparison
load applied. The equivalent plastic hinge length has been using P/fc′ Ag may not remain valid. Fig. 6 shows moment-

424 ACI Structural Journal/July-August 1997

Table 1—Details of specimens
Lateral steel Axial load level Ductility ratio Energy indicator
Researchers Specimen fc′ , ksi Spacing, in. ρs, % fyh, ksi Ash/Ash,c P/fc′ P/Po μφ Nφ80 Nφt E80 Et
AS-3 4.81 4.25 1.68 73.6 1.43 0.599 0.498 19.0* 63 74 610 753
FS-9 4.70 3.75 1.68 73.6 1.46 0.761 0.628 8.0 37 44 154 163
ES-13 4.72 4.50 1.69 67.3 1.34 0.758 0.626 6.0 15 26 53 110
Khoury AS-17 4.54 4.25 1.68 73.6 1.52 0.765 0.626 12.0 52 58 402 443
and
Sheikh10 AS-18 4.75 4.25 3.06 67.3 2.41 0.768 0.636 17.5 80 92 897 1156
AS-19 4.68 4.25 1.30 72.2 1.12 0.467 0.386 19.0 85 129 631 1230
F-9H 8.45 3.75 1.68 73.6 0.81 0.641 0.615 5.0 — — — —
E-13H 8.36 4.50 1.69 67.3 0.76 0.631 0.605 2.5 — — — —
AS-3H 7.85 4.25 1.68 73.6 0.88 0.619 0.585 10.5 31 35 178 204
Sheikh, Shah,
AS-18H 7.93 4.25 3.06 67.3 1.44 0.639 0.605 14.0 43 59 384 458
and Khoury11
AS-20H 7.78 3.00 4.30 67.3 2.10 0.643 0.607 16.5 80 98 935 1170
F-9L* 6.81 3.75 1.68 73.6 1.01 0.637 0.583 7.0
Patel and
E-13L 6.95 4.50 1.69 67.3 0.91 0.649 0.597 5.0
Sheikh16
A-17L 7.12 4.25 1.68 73.6 0.97 0.658 0.609 10.3
E-2 4.55 4.50 1.69 70.0 1.45 0.611 0.443 10.0
A-3 4.61 4.25 1.68 71.0 1.44 0.603 0.492 28.5
F-4 4.67 3.75 1.68 71.0 1.41 0.595 0.491 21.3
D-5 4.53 4.50 1.68 71.0 1.39 0.460 0.387 20.0
F-6 3.95 6.81 1.68 70.0 1.65 0.747 0.578 10.3
D-7 3.80 2.13 1.62 68.0 1.52 0.777 0.618 16.0
E-8 3.76 5.00 0.84 70.0 0.87 0.776 0.600 3.5
Yeh and
F-9 3.84 3.75 1.68 71.0 1.72 0.769 0.589 6.2
Sheikh12
E-10 3.81 2.50 1.68 71.0 1.73 0.766 0.585 5.2
A-11 4.05 4.25 0.77 68.0 0.72 0.737 0.576 8.3
F-12 4.85 3.50 0.82 67.0 0.63 0.601 0.499 9.4
E-13 3.95 4.50 1.69 70.0 1.67 0.738 0.571 12.3
D-14 3.90 4.25 0.81 67.0 0.73 0.748 0.600 7.3
D-15 3.90 4.50 1.68 71.0 1.61 0.748 0.600 12.3
A-16 4.92 4.25 0.77 81.0 0.71 0.600 0.500 13.2
*Lightweight aggregate concrete specimens

curvature responses of four columns. Effect of a change in

axial load on the column behavior can be evaluated from
Specimens AS-3 and AS-17, which are almost identical in
every other regard. Increase in load from 0.66fc′Ag to
0.77fc′Ag resulted in a significantly less ductile behavior. Curva-
ture ductility factor μφ was reduced by about 45 percent.
Specimens AS-3 and AS-3H contained the same amount
of tie steel and were tested under similar axial loads as repre-
sented by index P/fc′Ag. Specimen AS-3H made with higher
concrete strength, fc′, displayed much lower ductility. Spec-
imens AS-3 and AS-18H contained about 45 percent more
tie steel than that required by the ACI Code1 and both were
tested under P/fc′Ag approximately equal to 0.6. Ductility Fig. 7—Concept of effectively confined concrete area.
and energy dissipation capacity of the higher strength
concrete specimen is considerably lower. Specimen AS-17 is
comparable to Specimen AS-18H from a point of view that concrete and the distribution of confining pressure which are
both contained about 150 percent of the code-required tie in turn highly affected by the distribution of longitudinal and
steel contents and both were subjected to axial loads that lateral steel and the extent of lateral restraint provided to the
were approximately 60 percent of the ultimate load capacity bars.4,5,9 Figure 7 explains the concept of effectively
Po. Moment-curvature responses and ductility parameters of confined concrete area within a column core in two different
these two specimens are reasonably similar. It can be steel configurations.5 With larger number of longitudinal
concluded that the amount of tie steel required for a certain bars laterally supported by tie bends, the area of effectively
flexural response of columns for a given P/Po is approxi- confined concrete is increased and the efficiency of confine-
mately proportional to the concrete strength. ment improves considerably. From a four-bar configuration
Steel configuration—The effectiveness of confining steel to an eight-bar configuration the efficiency improvement is very
primarily depends on the area of the effectively confined large. Beyond the eight-bar configuration, the confinement

ACI Structural Journal/July-August 1997 425

 Limiting conditions for steel configurations—For earth-
quake design, it is believed that only the two top categories
of ductility, high and moderate, are needed to be discussed
here. Among the specimens tested during this research
program none of the specimens with Configuration E
(Category I) resulted in high section ductility factor (μφ).
The axial load level in these specimens was high but in other
studies (e.g., Ref. 12) columns with E sections tested under
low axial load level (P < 0.3Po) also showed unsatisfactory
behavior. Based on the observed experimental performance
and the analytical evidence (Fig. 8), Category I configuration
is not recommended for high ductility columns.
Specimen ES-13, which contained 34 percent more steel
than required by the Code, showed very poor behavior with
μφ = 2.5. Also under medium level of axial load (about 0.4-
0.45Po), Specimen E-2 with 1.45 times the Code required
steel exhibited μφ of only about 10. Therefore, the use of
Configuration E in moderately ductile columns should be
limited to lower range of axial load (P < 0.40Po). For conser-
vative design, the Category I configurations are recom-
mended for moderate ductility columns only if the applied
axial load is less than the balanced load Pb.
With regard to Category II configurations, the effective-
ness of hooks not anchored in the core has been a controver-
sial issue. Although some researchers concluded that the
supplementary crossties allowed columns to perform in a
Fig. 8—Effect of steel configuration. ductile manner,12,18,19 the axial load levels in these tests were
low (about 0.1Po to 0.3Po). Recent research7-9 has shown that
the use of 90 deg hooks in Section F (Fig. 9) may provide
sufficient restraint to the middle bars up to a certain stage of
loading, but at large deformations the 90 deg hooks tend to
open, and the restraint provided to the bars becomes ineffec-
tive resulting in a loss of confinement. None of the reported
specimens showed satisfactory performance under high
levels of axial load even when lateral steel content was in
excess of the Code requirements (see Table 1). Although
Specimen F-4 indicates very ductile performance, the same
section in Specimen F-9 under higher axial load level shows
undesirable behavior for seismic resistance. It should be
noted that the 90 deg hooks were not always in the zone of
maximum deformation due to the monotonic nature of flex-
Fig. 9—Categories of steel configurations. ural loading in this set of specimens. Therefore, high apparent
ductility in some specimens may not be repeatable. Accord-
efficiency does not increase as significantly with the increase in ingly, it is recommended that the use of Category II configu-
the number of laterally-supported longitudinal bars. rations to produce high-ductility columns be limited to cases
Figure 8 shows moment-curvature responses of two with low levels of axial load. These columns can be used for
column specimens ES-13 and FS-9. These specimens and moderate ductility if axial load does not exceed 0.4Po.
Specimen AS-17 in Fig. 6 are almost identical in all regards The limiting conditions under which the three categories
except steel configuration. Specimen AS-17 displayed more of steel configurations may be reliably used for moderate
ductile behavior (also see Table 1) than Specimen FS-9 and high ductility columns are outlined in Fig. 10. It should
which in turn is tougher than Specimen ES-13. be emphasized here that some configurations under such
Based on this concept and extensive experimental conditions may require a higher amount of lateral steel than
data4,7,8,9,12,14 steel configurations may be divided into the other configurations.
following three main categories (see Fig. 9):
• Category I: where only single-perimeter hoops are used AMOUNT OF CONFINING STEEL
as confining steel. General form of proposed equation
• Category II: in addition to the perimeter hoops supporting The relationship between the amount of lateral steel as
four corner bars, at least one middle longitudinal bar at recommended by the current Code Ash,c and the suggested
each face is supported at alternate points by hooks that are amount of lateral steel Ash is taken as:
not anchored in the core. At other points the supporting
hooks are anchored in the core.
Ash = (Ash,c)Y (6)
• Category III: in which a minimum of three longitudinal bars
are effectively supported by tie corners on each column face
and hooks are anchored into the core concrete. where Y is a factor expressed as

426 ACI Structural Journal/July-August 1997

 Y = αYpYφ (7)

where α is a parameter that accounts for the confinement

efficiency including configuration and the lateral restraint
provided to the longitudinal bars. Parameters Yp and Yφ take
into account the effect of axial load level and the section
ductility demand, respectively.
Parameter α—Parameter α is assumed to be equal to
unity for Category III configurations. This factor is expected
to be greater than unity for Category I configurations even
for their use under limiting conditions prescribed earlier. For
such a case, the value for α is estimated in a later section. Use
of Category II configurations is subjected to imposed limita-
tions because some of the hooks are not anchored in the core
as previously discussed. It is reasonable to assume a value of
α equal to unity for these configurations in situations where
opening of these hooks does not take place until after suffi-
Fig. 10—Limiting conditions for steel configurations.
cient ductility is exhibited.7-9 In the event of high axial load
levels, the value of α would be much greater than unity;
however such an application should be avoided.
Development of expressions for parameters Yp and Yφ—
Eq. (6) and (7) for sections with at least three longitudinal
bars effectively restrained on each face (α = 1) reduce to

A sh
-----------
- = Yp Yφ (8)
A sh, c

After investigating several possible forms of expressions

for Yp and Yφ, the following simple forms were selected,
a3
Y p = a 1 + a 2 ⎛ ------⎞
P
⎝P ⎠
, and (9)
o

b2
Yφ = b1 ( μφ ) (10) Fig. 11—Required amount of tie steel as affected by axial load.

where a1, a2, a3, b1, and b2 are constants to be determined

empirically.
As a starting point, since the two parameters Yp and Yφ are
independent of each other, the value of Yφ is assumed to be
unity for highly ductile sections with μφ equal to or greater
than 16. Specimens meeting this requirement are AS-3, AS-
18, AS-19, AS-20H, A-3 and F-4. Using the results from
these specimens, a least squares analysis was performed to
find constants a1 and a2 for selected values of a3 that ranged
from 1 to 6. Corresponding to each chosen value of a3, and
consequently obtained values for a1 and a2, the constants b1
and b2 in the expression for Yφ (Eq. 10) were then deter-
mined using the test results for those 16 specimens in which
α = 1.0. These included all the specimens with A and D
configurations and Specimens F-4 and F-12 from Table 1.
Specimen A-3 was not included in the analysis since its μφ Fig. 12—Required amount of tie steel as affected by curvature
was unusually large compared with other similar specimens. ductility factor
Minimization of the total cumulative error for all the 16
specimens was the only criterion used to select the final where Yexp = Ash/Ash,c , and Ypred = Yp Yφ. The best fit curves
values of the empirical constants. The cumulative error e2 from this analysis are shown in Fig. 11 and 12. The expressions
was calculated as
for parameters Yp and Yφ are given below
16

∑ ( Yexp – Ypred )
2 2 P 5
e = (11) Y p = 1 + 13 ⎛ ------⎞ (12)
⎝P ⎠
1 o

ACI Structural Journal/July-August 1997 427

 Table 2—Calculated α values for specimens with
Configuration E
Specimen ES-13 E-13H E-13L E-2 E-8 E-10 E-13
Yp 2.25 2.05 1.99 1.22 2.01 1.89 1.79
Yφ 0.27 0.10 0.22 0.49 0.15 0.23 0.62
α 2.20 3.73 2.09 2.43 2.95 3.99 1.51

μ
Y φ = -----φ- (16)
18

Equation (15) provides a conservative estimate for Eq. (12)

for most of the axial load range up to P/Po equal to 0.65. It
should be noted that the allowable axial load for tied
columns is 0.56Po (1). Eq. (16) gives a slightly conservative
alternative to Eq. (13). Considering Eq. (15) and (16), Eq. (14)
can be rewritten as:
Fig. 13—Comparison of experimental and predicted curvature
ductility factors. P μ μ
A sh = α 6 ------ – 1.4 -----φ- A sh, c ≥ α -----φ- A sh, c (17)
Po 18 18

and Design for Category I configurations—The parameter α

may be estimated in this case by using the experimental
( μφ )
1.15 results. Values for α were calculated using Eq. (14) for all
Y φ = ------------------ (13) the specimens with Configuration E from Table 1 and are
29 listed in Table 2. The average value of α is about 2.70 which
implies that the amount of lateral steel needed in sections
The correlation coefficients for Eq. (12) and (13) are 0.99 with Configuration E to attain a specific ductility demand
and 0.93, respectively. The high coefficients indicate excel- may be two to three times that required for sections with
lent agreement between the analytical and the experimental Configuration A. The experimental M-φ relationships
values. The cumulative error e2 over the 16 specimens used reported previously8 confirm this finding.
in the final analysis was 0.519 yielding as average error of The factor α for Category I configurations may also be
0.032 per specimen. The parameter Yφ, when checked for estimated by adopting the concept of “effectively confined
highly ductile column sections, was found to be almost unity concrete core area”5 as shown in Fig. 7. The ratio between
for the average value of μφ equal to 18.5. the area of effectively confined concrete and the total
Final form of design equation—Based on the above, the concrete area λ at tie level is given by
amount of lateral steel in tied columns may be calculated
using the following expression n

∑ Ci
2
1.15
⎧ ( μφ )
P-⎞ 5 ⎫------------------A
A sh = α ⎨ 1 + 13 ⎛ ----- i=1
⎝ P ⎠ ⎬ 29 sh, c (14) λ = 1 – ---------------- (18)
⎩ o ⎭ 5.5A co

Factor α is unity for Category III configurations and for where Aco = the core area enclosed by the center line of
Category II configurations as long as the prescribed limiting perimeter hoop; Ci is the base of the curve representing the
conditions are met. However, for Category I configurations,
area which is not effectively confined; and n = the number of
the α value is greater than unity.
The above procedure is applied to all those specimens these curves.
tested during this program in which longitudinal bars were It may be reasonably assumed here that the configuration
effectively supported laterally. Comparison between the parameter α is proportional to 1/λ. Since α = 1 for Category
analytical and the experimental curvature ductility factors is III configurations, α for Category I configurations (αI) may
shown in Fig. 13. The correlation coefficient is 0.94 with an be written as αI = λIII/λI where λIII and λI can be calculated
average difference between the test and the predicted values using Eq. (18). For the specimens in which the longitudinal
less than 10 percent. As mentioned above, minimization of bars are uniformly distributed around the core perimeter, the
the total cumulative error was the only criterion used in the λ values for Configurations A and O (Fig. 9) are 0.636 and
development of Eq. (14). No attempts were made to mini- 0.273, respectively. Hence, αI = 2.33. The ductility of
mize the error in individual columns. Section E with eight longitudinal bars, four corner bars, and
Fig. 11 and 12 also show simplified versions of equations four unsupported middle bars was sometimes observed to be
for Yp and Yφ as given below even worse than that of Section O with only four corner
bars.20 It may be reasonable, therefore, to conclude that the
factor α for Category I configurations may range from 2.3 to
P
Y p = 6 ------ – 1.4 ≥ 1.0 (15) 2.7. An average value of 2.5 is thus assumed for all
Po configuration types in this category.

428 ACI Structural Journal/July-August 1997

 As suggested before, the use of Category I configurations
may be reliable only for moderate ductility columns under
axial load level below the balance point. For this low axial
load, 1 + 13(P/Po)5 ≈ 1.0. Taking α equal to 2.5, Eq. (14) and
(17) reduce to Eq. (19) and (20), respectively.

1.15
( μφ )
A sh = ------------------A (19)
11.5 sh, c

μ
A sh = ------φ- A sh, c (20)
7.2

Based on Eq. (19) and (20), it can be stated that for axial
load below the balance point, the current ACI Code steel
may be sufficient to provide μφ of about 7 to 8 for sections
with Configuration E. However, for a moderately ductile Fig. 14—Lateral steel requirements.
column with μφ = 12, the required amount of lateral steel
should be 50 percent higher than that required by the Code.

Design chart
On the basis of the proposed equations [Eq. (14) and (17)],
a design chart is constructed in Fig. 14 for columns in which
a minimum of three longitudinal bars are effectively supported
laterally in each face (Category III configurations). The
amount of required lateral steel increases with an increase in
the axial load level and an increase in the ductility demand.
Three ductility zones as discussed earlier are indicated in the
figure which shows that the Code prescribed amount of tie
steel may be adequate to provide high ductility columns only
if the applied axial load is less than 0.4Po , and moderate
ductility columns under higher axial loads as long as at least
three longitudinal bars are effectively supported laterally on
each column face. Under high levels of axial load, the Code
required amount of lateral steel may not be sufficient to meet
high ductility demand. The same figure can be used for Cate-
gory II sections as long as the limiting conditions shown in
Fig. 10 are satisfied. For columns with Category I configura- Fig. 15—Application of the design procedures.
tions (α = 2.5), the code-provided amount of lateral steel will
be insufficient to provide even moderately ductile columns suggested to be the smallest of B/3, 6db and 200 mm (8
under axial loads exceeding balanced load (P/Po ≈ 0.3). in.).
Effect of hoop/tie spacing APPLICATION OF THE
Experimental and theoretical evidences show that hoop PROPOSED DESIGN APPROACH
spacing plays a significant role in the mechanism of confine- The proposed equation is applied to a 700 mm (27.6 in.)
ment.4,5 Larger ratio of hoop spacing s to core width B will square column previously reported by Park et al.21 The
result in smaller area of effectively confined concrete in the results are presented graphically in Fig. 15 and are compared
core (Fig. 7). The procedure presented here for the sake of with the ACI and NZS Code requirements as well as the
simplicity does not include tie spacing as an active param- lateral steel required according to Park.22 For comparative
eter. However, it should be noted that the test data on which purposes, the analytical equation proposed here was
the equations are based were obtained from specimens in converted to be a function of (P/fc′Ag) instead of (P/Po).
which tie spacing varied from 0.20B to 0.43B. In this prac- Under high axial loads, the steel required for highly ductile
tical range of spacing, the confinement mechanism has columns according to Park is significantly less than that
reasonably high efficiency. Another important reason to based on the proposed equation. However, for curvature
limit spacing is to avoid premature buckling of longitudinal ductility factor μφ equal to 10, the requirements according to
bars when a column is subjected to seismic excursions in the Park are somewhat conservative compared with the
inelastic range. In the specimens considered here, tie spacings proposed curve. The amount of steel required according to
varied between 3.4db and 7.2db where db is the bar diameter. the ACI Code is inadequate to achieve μφ = 10 under high
In this range of s/db, premature buckling of the longitudinal axial loads even for well-configured columns. The 1982 NZ
bars can generally be avoided.2,13 The proposed procedure code requirements for lateral steel produced highly ductile to
can be used to design the confining steel for a given column moderately ductile columns for most of the axial load range.
performance as long as the tie spacing is less than 0.43B and In the 1995 version of the NZ code, the effect of axial load is
7db. For a conservative design the limit to the tie spacing is more severe compared with the 1982 version. The required

ACI Structural Journal/July-August 1997 429

Table 3—Application of proposed equations
Lateral steel Longitudinal steel Axial load level A sh Results
----------
-
Specimen fc′ , MPa Size @ Spacing, mm fyh, MPa No. Size fy1, MPa P/fc′ Ag P/Po A sh, c μφ, exp. μφ, pred.
Specimens* reported by Mugurauma and Watanabe26
BH-2 115.8 6 mm @ 35 792.3 12 φ13 mm 399.6 0.423 0.446 1.395 20.3 20.9
AH-2 85.7 6 mm @ 35 792.3 12 φ13 mm 399.6 0.629 0.632 1.885 15.6 15.6
BL-2 115.8 6 mm @ 35 328.4 12 φ13 mm 399.6 0.423 0.446 0.578 9.0 9.7
AL-2 85.7 6 mm @ 35 328.4 12 φ13 mm 399.6 0.629 0.632 0.781 4.1‡ 7.3
Specimens† reported by Azizinamini et al.27
NC-2 41.4 #4 @ 102 414 8 #8 414 0.20 0.20 0.888 16.8 16.7
NC-5 41.4 #4 @ 102 414 8 #8 414 0.30 0.30 0.888 16.4 16.3
*
Column section = 200 x 200 mm; core section = 176 x 176 mm; configuration = Type D
†
Column section = 18 x 18 in.; core section = 14.5 x 14.5 in.; configuration = Type A
‡
In Ref. 26, the M-φ envelope curve for this specimen was reported only up to the maximum moment. The post peak descending part of the curve was not given.

amount of lateral steel, according to the 1995 NZ code, is even in well-configured columns to meet the high ductility
similar to that proposed by Park22 for μφ = 20. demand. However, at low axial load levels (P ≤ 0.4Po), the
Equation (14) is also applied to six specimens reported by code requirements may be relaxed. For steel configurations
Muguruma and Watanabe23 and by Azizinamini et al.24 in which only four corner bars are adequately restrained
Details of these specimens are given in Table 3 which also laterally, the ACI Code design will produce columns with
compares the analytical values of curvature ductility factors inadequate ductility for most of the axial load range. The
with the experimental values. The comparison shows excel- proposed equation when applied to realistically-sized speci-
lent agreement. The only difference may be noticed in Spec- mens tested by different investigators yielded excellent
imen AL-2. It is believed that the actual value of μφ for this agreement with the test data.
specimen is greater than the experimental value listed in
Table 3 since the M-φ envelope curve for this specimen was ACKNOWLEDGEMENTS
reported only up to the maximum moment. The post-peak Research reported here was supported by grants from the Natural
Sciences and Engineering Research Council of Canada and the U.S.
descending part of the curve was not provided. National Science Foundation.

SUMMARY AND CONCLUSIONS REFERENCES

The development over the years of the ACI Code provi- 1. ACI Committee 318, “Building Code Requirements for Reinforced
sions for the design of confining steel in tied columns is Concrete (ACI 318-95) and Commentary (318R-95),” American Concrete
presented. It appears that some of the changes made from Institute, Farmington Hills, MI, 1995, 369 pp.
one edition of the Code to the next are not suitable to provide 2. Code of Practice for the Design of Concrete Structures (NZS 3101:1982
ductile columns. The current Code requirements are evalu- and NZS 3101:1995), Standards Association of New Zealand, Wellington.
3. Code for the Design of Concrete Structures for Building (CAN 3-
ated based on experimental evidences, and the areas in which
A23.3M84), Canadian Standards Association, Rexdale, Ontario, 1984, 281 pp.
the requirements need modification are explored. 4. Sheikh, S. A., and Uzumeri, S. M., “Strength and Ductility of Tied
Although extensive experimental evidence indicates that Concrete Columns,” Journal of the Structural Division, ASCE, V. 106,
ductile behavior of confined concrete columns depends signif- ST5, May 1980, pp. 1079-1102.
icantly on the level of axial load as well as the distribution of 5. Sheikh, S. A., and Uzumeri, S. M., “Analytical Model for Concrete
steel and the lateral support provided to the longitudinal bars, Confinement in Tied Columns,” Journal of the Structural Division, ASCE,
these factors are not considered in the current Code equations V. 108, ST12, Dec. 1982, pp. 2703-2722.
for confining steel. Ductility demand is also not given due 6. ACI Committee 318, “Building Code Requirements for Reinforced
Concrete,” American Concrete Institute, Detroit, 1956, 1963, 1971, 1977,
importance in the Code design procedure. Therefore, columns 1983, 1989.
designed according to the current Code may display brittle 7. Sheikh, S. A., and Khoury, S., “Confined Concrete Columns with
behavior under certain circumstances and may be overly Stubs,” ACI Structural Journal, V. 90, No. 4, July-Aug. 1993, pp. 414-431.
conservative in terms of ductility in other cases. The required 8. Sheikh, S. A.; Shah, D. V.; and Khoury, S. S., “Confinement of High-
amount of lateral steel should be a function of axial load level, Strength Concrete Columns,” ACI Structural Journal, V. 91, No. 1, Jan.-
steel configuration, and the expected ductile performance. Feb. 1994, pp. 100-111.
To account for the factors discussed above, a procedure for 9. Sheikh, S. A., and Yeh, C. C., “Tied Concrete Columns Under Axial
Load and Flexure,” Journal of the Structural Division, ASCE, V. 116, No. 10,
the design of confining steel in tied columns is proposed.
Oct. 1990, pp. 2780-2801.
Three categories of steel configurations are outlined. 10. Bertero, V. V., and Felippa, C., Discussion of “Ductility of Concrete”
Limiting conditions under which these categories of steel by H. E. H. Roy and M. A. Sozen, Proceedings of the International Sympo-
configurations may reliably be used are suggested. A sample sium on Flexural Mechanics of Reinforced Concrete, ASCE-ACI, Miami,
design chart is also presented which can be used to determine 1964, pp. 227-234.
the amount of lateral steel required for a specific ductility 11. Iyengar, K. T. R. J.; Desayi, P.; and Reddy, K. N., “Stress-Strain
level in a column that contains a minimum of three laterally- Characteristics of Concrete Confined in Steel Binders,” Magazine of
Concrete Research, V. 22, No. 72, 1970, pp. 173-184.
supported longitudinal bars in each face and the steel is
12. Ozcebe, G., and Saatcioglu, M., “Confinement of Concrete Columns
uniformly distributed. A comparison between the amount of for Seismic Loading,” ACI Structural Journal, V. 84, No. 4, July-Aug.
lateral steel as suggested by the new approach and the Code 1987, pp. 308-315.
requirements indicates that under moderate-to-high levels of 13. Priestly, M. J. N, and Park, R., “Strength and Ductility of Concrete
axial loads, the Code requirements may not be sufficient Bridge Columns Under Seismic Loading,” ACI Structural Journal, V. 84,

430 ACI Structural Journal/July-August 1997

No. 1, Jan.-Feb. 1987, pp. 69-76. Also, Discussion by K. Sakai, and S. A. 19. Rabbat; B. G., Daniel, J. L.; Weinmann, T. L.; and Hanson, N.W.,
Sheikh, ACI Structural Journal, V. 84, No. 6, Nov-Dec. 1987, pp. 543-547. “Seismic Behavior of Lightweight and Normal Weight Concrete Columns,”
14. Patel, S., and Sheikh, S., “Behavior of Light-Weight High-Strength ACI JOURNAL, Proceedings, V. 83, No. 1, Jan.-Feb. 1986, pp. 69-79.
Concrete Columns Under Axial Load and Cyclic Flexure,” Report No. 20. Sheikh, S. A.; Khoury, S.; and Sakai, K., “Ductility of Confined
UHCEE 91-5, Department of Civil and Environmental Engineering, Concrete Columns,” Symposium on Ductility, Japan Concrete Institute,
University of Houston, Dec. 1992. Tokyo, Japan, 1990, 8 pp.
15. Khoury, S., and Sheikh, S. A., “Behavior of Normal and High 21. Park, R.; Priestly, M. J. N.; and Gill, W. D., “Ductility of Square-
Strength Confined Concrete Columns with and without Stubs,” Report Confined Concrete Columns,” Journal of the Structural Division,
No. UHCEE 91-4, Department of Civil and Environmental Engineering, ASCE, V. 108, ST4, Apr. 1982, pp. 929-950.
University of Houston, Dec. 1991, 345 pp.
16. Ehsani, M. R., and Wight, J. K., “Confinement Steel Requirements 22. ACI Committee 318, “Proposed Revisions to: Building Code
for Connections in Ductile Frames,” Journal of the Structural Division, Requirements for Reinforced Concrete,” ACI Structural Journal, V. 86,
ASCE, V. 116, ST 3, Mar. 1990, pp. 751-767. No. 3, May-June 1989, pp. 342-347.
17. Sakai, K., and Sheikh, S. A., “What Do We Know about Confinement 23. Muguruma, H., and Watanabe, F., “Ductility Improvement of High-
in RC Columns (A Critical Review of Previous Work and Code Provisions),” Strength Concrete Columns with Lateral Confinement,” High Strength
ACI Structural Journal, V. 86, No. 2, Mar.-Apr. 1989, pp. 192-207. Concrete, SP-121, American Concrete Institute, Farmington Hills, MI,
18. Johal, L. S.; Azizinamini, A.; Musser, D. W.; and Corley, W. G., 1990, pp. 47-60.
“Seismic Evaluation of Columns to Improve Design Criteria for Transverse 24. Azizinamini, A.; Corley, W. G.; and Johal, L. S.; “Effects of Trans-
Reinforcement,” Proceedings, 5th Canadian Conference on Earthquake verse Reinforcement on Seismic Performance of Columns,” ACI Structural
Engineering, Ottawa, July 1987, pp. 799-806. Journal, V. 89, No. 4, July-Aug. 1992, pp. 442-450.

ACI Structural Journal/July-August 1997 431