Section 12-4 Design of Columns in Nonsway Frames • 595

EXAMPLE 12-2 Design of the Columns in a Braced Frame

Figure 12-28 shows part of a typical frame in an industrial building. The frames are
spaced 20 ft apart. The columns rest on 4-ft-square footings. The soil bearing capacity is
4000 psf. Design columns C–D and D–E. Use fcœ = 4000 psi and fy = 60,000 psi for
beams and columns. Use load combinations and strength-reduction factors from ACI Code
Sections 9.2 and 9.3.
1. Calculate the column loads from a frame analysis.
A first-order elastic analysis of the frame shown in Fig. 12-28 (loads acting on frame mem-
bers are not shown) gave the forces and moments in the table.

Column CD Column DE

Service loads, P Dead = 80 kips Dead = 50 kips

Live = 24 kips Live = 14 kips
Service moments at Dead = -60 kip-ft Dead = 42.4 kip-ft
tops of columns Live = -14 kip-ft Live = 11.0 kip-ft
Service moments at Dead = -21 kip-ft Dead = -32.0 kip-ft
bottoms of columns Live = -8 kip-ft Live = -8 kip-ft

A
68.5
9 in.
E A F

15 in.

16 in. 24 ft

A – A.

B A
D
A

20 ft

A C

4-ft-square
30 ft 25 ft

(c) Moments in columns

CD and DE, (kip-ft).

Fig. 12-28
Braced frame—Example 12-2.
596 • Chapter 12 Slender Columns

Clockwise moments on the ends of members are positive. All wind forces are assumed to
be resisted by the end walls of the building.
2. Determine the factored loads.
(a) Column CD:
Pu = 1.2 * 80 + 1.6 * 24 = 134 kips
Moment at top = 1.2 * -60 + 1.6 * -14 = -94.4 kip-ft
Moment at bottom = 1.2 * -21 + 1.6 * -8 = -38.0 kip-ft
The factored-moment diagram is shown in Fig. 12-28c. By definition (ACI
Code Section 2.1), M2 is always positive, and M1 is positive if the column is
bent in single curvature (Fig. 12-13c and d). Because column CD is bent in
double curvature (Fig. 12-28c), M1 is negative. Thus, for slender column
design, M2 = +94.4 kip-ft and M1 = -38.0 kip-ft.

(b) Column DE:

Pu = 82.4 kips
Moment at top = +68.5 kip-ft
Moment at bottom = -51.2 kip-ft
Thus M2 = +68.5 kip-ft and M1 = +51.2 kip-ft. M1 is positive because
the column is in single curvature.
3. Make a preliminary selection of the column size. From Eq. (11-19a) for
rg = 0.015,
Pu
Ag1trial2 Ú
0.401fcœ
+ fyrg2
134
=
0.4014 + 0.015 * 602
= 68.4 in.2 or 8.3 in. square
Because of the anticipated slenderness effects and because of the large moments, we will
take a larger column. Try 14 in. * 14 in. columns throughout.
4. Are the columns slender? From ACI Code Section 10.10.1, a column in a
braced frame is short if k/u>r is less than 34 - 12M1>M2 , which has an upper limit of 40.
(a) Column CD:
/u = 20 ft - 2 ft = 18 ft
= 216 in.
r = 0.3 * 14 in. = 4.2 in.
From Table 12-2, k ⬵ 0.77. Thus,
klu 0.77 * 216
⬵ = 39.6
r 4.2
From Eq. (12-20b),
M1 38.0
34 - 12a b = 34 - 12a - b = 38.8
M2 94.4
Because 39.6 7 38.8, column CD is just slender.
 Section 12-4 Design of Columns in Nonsway Frames • 597

(b) Column DE:

/u = 24 ft - 2 ft = 264 in.
k ⬵ 0.86
klu 0.86 * 264
⬵ = 54.1 7 40
r 4.2

Thus, column DE is also slender.

5. Check whether the moments are less than the minimum. ACI Code Section
[Link] requires that braced slender columns be designed for a minimum eccentricity of
10.6 + 0.03h2 in. For 14-in. columns, this is 1.02 in. Thus, column CD must be designed
for a moment M2 of at least

Pu emin = 134 kips * 1.02 in. = 11.4 kip-ft

and column DE for a moment of at least 7.0 kip-ft. Because the actual moments exceed
these values, the columns shall be designed for the actual moments.

6. Compute EI. Because the reinforcement is not known at this stage of the de-
sign, we can use Eq. (12-16) to compute EI. From Eq. (12-16),
0.40 Ec Ig
EI =
1 + b dns
where
Ec = 57,0002fcœ = 3.60 * 106 psi
Ig = 144>12 = 3200 in.4
0.40 Ec Ig = 4.61 * 109 lb-in2.

(a) Column CD:

1.2 * 80
b dns = = 0.716 (12-26a)
134
4.61 * 109
EI = = 2.69 * 109 lb-in.2
1 + 0.716
(b) Column DE:
1.2 * 50
b dns = = 0.728
82.4
4.61 * 109
EI = = 2.67 * 109 lb-in2.
1.728

7. Compute the effective-length factors. Two methods of estimating the effective-

length factors, k, have been presented. In this example, we calculate k by both methods, to
illustrate their use. In practice, only one of these procedures would be used in a given set of
calcultions. We begin with
©Ec Ic>/c
c = (12-29a)
©Eb Ib>/b
598 • Chapter 12 Slender Columns

where values for Ic and Ib can be taken as those given in ACI Code Section [Link].
Thus, Ic = 0.70 Ig(col), and Ib = 0.35 Ig(beam). For the assumed column section, use
Ic = 0.70 * 144/12 = 2240 in.4. For the beam section shown in Fig. 12-28b, the govern-
ing condition in ACI Code Section 8.12.2 gives the effective flange width as being equal to
one-fourth of the beam span, i.e., 90 in. Using that width gives Ig = 36,600 in.4, so
Ib = 0.35 * 36,600 = 12,800 in.4. As noted earlier in this text, Ig for the web portion of
the beam section is equal to approximately one-half of Ig for the full-beam section (web
and flange), and it often is used as an estimate of the cracking moment of inertia, Icr , for
the beam section. Thus, for the analysis used here, 0.35Ig(beam) is equal to approximately
0.70 Ig1web2 = 0.7 * 16 * 243/12 = 12,900 in.4. This approximation could be used at
this stage to save the time required to determine the effective flange width and then cal-
culate the moment of inertia of a flanged section.
In Eq. (12-29a), /c and /b are the spans of the column and beam, respectively, mea-
sured center-to-center from the joints in the frame.

(a) Column DE: The value of c at E is

Ec * 2240>124 * 122 Ec
cE = = 0.182
Eb * 12,800>300 Eb

where Ec = Eb. Thus, cE = 0.182. The value of c at D is (for Ec = Eb):

Ec * 2240>288 + Ec * 2240>240
cD = = 0.481
Eb * 12,800>360

The value of k from Fig. 12-26a is 0.63. The value of k from Table 12-2 was approxi-
mately 0.86.
As was pointed out in the discussion of Fig. 12-26, the effective-length nomographs
tend to underestimate the values of k for beam columns in practical frames [12-18]. Be-
cause Table 12-2 gives reasonable values without the need to calculate c, it has been used to
compute k in this example. Thus, we shall use k = 0.86 for column DE.

(b) Column CD: The value of c at D is

cD = 0.481
The column is restrained at C by the rotational resistance of the soil
under the footing and is continuous at D. From Eq. (12-34),
4Ec Ic>/c
c =
If ks
where If is the moment of inertia of the contact area between the footing and
the soil and ks is the coefficient of subgrade reaction, which is 160 psi/in.
from Fig. 12-27. Using , Ic = 1.0 Ig (not 0.7Ig) and assuming a footing depth
of 2 ft, so /c in Eq. (12-34) is 22 ft = 264 in., results in

484
If = = 442,000 in.4
12
4 * 3.60 * 106 lb>in.2 * 3200 in.4>264 in.
cc =
442,000 in.4 * 160 lb>in.3
= 2.47
 Section 12-4 Design of Columns in Nonsway Frames • 599

From Fig. 12-26: k = 0.77

From Table 12-2: k ⬵ 0.77
Use k = 0.77 for column CD.
8. Compute the magnified moments. From Eq. (12-23),
Mc = dns M2 (12-23)
where
Cm
dns = Ú 1.0 (12-24)
1 - 1Pu>0.75Pc2
(a) Column CD:
38.0
Cm = 0.6 + 0.4a - b (12-14)
94.4
= 0.438
p2EI
Pc = (12-25)
1k/u22
/u = 18 ft = 216 in
Using EI from step 6 and k = 0.77,

p2 * 2.69 * 109 lb-in.2

Pc =
10.77 * 216 in.22
= 960,000 = 960 kips
0.438
dns =
1 - 134>10.75 * 9602
= 0.538 Ú 1.0
Therefore, dns = 1.0. This means that the section of maximum moment
remains at the end of the column, so that
Mc = 1.0 * 94.4 = 94.4 kip-ft

Column CD is designed for Pu = 134 kips and Mu = Mc = 94.4 kip-ft.

(b) Column DE:
51.2
Cm = 0.6 + 0.4a b = 0.900
68.5
p2 * 2.67 * 109
Pc = = 511 kips
10.86 * 26422
0.900
dns = Ú 1.0
82.4
1 -
0.75 * 511
= 1.15
This column is affected by slenderness, so
Mc = 1.15 * 68.5 = 78.8 kip-ft
Column DE is designed for Pu = 82.4 kips and Mu = Mc = 78.8 kip-ft.
600 • Chapter 12 Slender Columns

4 No. 7 bars

14 in.

Fig. 12-29
Final column section for 14 in.
Example 12-2.

For both columns CD and DE the magnified moments are less than 1.4 times the first-
order moments, as required by ACI Code Section [Link].
9. Select the column reinforcement. Column CD is carrying both a higher axial load
and a higher moment, and thus, it will govern the design of the column section. We will use
the tied-column interaction diagrams in Appendix A, assuming an equal distribution of longi-
tudinal bars in two opposite faces of the column. The parameters required for entering the
interaction diagrams are

14 in. - 2 * 2.5 in.

g ⬵ = 0.64 ⬵ 0.60
14 in.

Assuming fPn = Pu = 134 kips,

fPn 134 kips

= = 0.68 ksi
bh 14 in. * 14 in.

Assuming fMn = Mc = 94.4 kip-ft = 1130 kip-in.,

fMn 1130 kip-in.

= = 0.41 ksi
2
bh 114 in.23

From Fig. A-6a 1g = 0.602, the required value for rg is approximately 0.012. Thus,

Ast1req¿d2 = 0.012 * Ag = 0.012 * 114 in.22 = 2.35 in.2

Use four No. 7 bars, Ast = 4 * 0.60 in.2 = 2.40 in.2 In summary, use a 14 in. * 14 in.
column section with four No. 7 bars, as shown in Fig. 12-29. ■

12-5 BEHAVIOR OF RESTRAINED COLUMNS IN SWAY FRAMES

Statics of Sway Frames

A sway (unbraced) frame is one that depends on moments in the columns to resist lateral
loads and lateral deflections. Such a frame is shown in Fig. 12-30a. The sum of the
moments at the tops and bottoms of all the columns must equilibrate the applied lateral-
load moment, V/, plus the moment due to the vertical loads, ©P¢. Thus,

© 1Mtop + Mbtm2 = V/ + ©P¢ (12-35)

 Section 12-5 Behavior of Restrained Columns in Sway Frames • 601

M top M top

M btm M btm

V−/ moments. P−¢ moments.

Fig. 12-30
Column moments in a sway
frame.

It should be noted that both columns have deflected laterally by the same amount ¢.
For this reason, it is not possible to consider columns independently in a sway frame.
If a sway frame includes some pin-ended columns, as might be the case in a precast
concrete building, the vertical loads in the pin-ended columns are included in ©P in Eqs.
(12-21) and (12-35). Such columns are referred to as leaning columns, because they depend
on the frame for stability.
The V–/ moment diagram due to the lateral loads is shown in Fig. 12-30b and that
due to the P– ¢ moments in Fig. 12-30c. Because the maximum lateral-load moments and
the maximum P– ¢ moments both occur at the ends of the columns, and hence can be
added directly, the equivalent moment factor, Cm, given by Eq. (12-14) does not apply to
sway columns. On the other hand, Eq. (12-11) becomes

Mo11 - 0.18P>PE2
Mc = (12-36)
1 - P>PE
602 • Chapter 12 Slender Columns

The term in brackets in Eqs. (12-11) and (12-36) has been left out of the Moment-Magnifier
equation in the ACI Code because the resulting change in the magnification does not vary
significantly.
It is also important to note that if hinges were to form at the ends of the beams in the
frame as shown in Fig. 12-30d, the frame would be unstable. Thus, the beams must resist
the full magnified end moment from the columns for the frame to remain stable (ACI Code
Section [Link]).
Loads causing sway are seldom sustained, although such cases can be visualized,
such as a frame that resists the horizontal reaction from an arch roof or a frame resist-
ing lateral earth loads. If a sustained load acts on an unbraced frame, the deflections
increase with time, leading directly to an increase in the P– ¢ moment. This process is
very sensitive to small variations in material properties and loadings. As a result,
structures subjected to sustained lateral loads should always be braced. Indeed, braced
frames should be used wherever possible, regardless of whether the lateral loads are
short time or sustained.

Mns and Ms Moments

Two different types of moments occur in frames:

1. Moments due to loads not causing appreciable sway, Mns

2. Moments due to loads causing appreciable sway, Ms

The slenderness effects of these two kinds of moments are considered separately in the
ACI Code design process because each is magnified differently as the individual columns
deflect and as the entire frame deflects [12-22]. Column moments that cause no apprecia-
ble sway are magnified when the column deflects by an amount d relative to its original
straight axis such that the moments at points along the length of the column exceed those
at the ends. In Section 12-2, this is referred to as the member stability effect or P–d effect,
where the lower case d refers to deflections relative to the chord joining the ends of the
column.
On the other hand, the column moments due to lateral loads can cause appreciable
sway. They are magnified by the P– ¢ moments resulting from the sway deflections, ¢, at
joints in the frame, as indicated by Eq. (12-35). This is referred to as the P– ¢ effect or
lateral drift effect.
ACI Code Section 2.1 defines the nonsway moment, Mns, as the factored end moment
on a column due to loads that cause no appreciable sidesway, as computed by a first-order
elastic frame analysis. These moments result from gravity loads. The 1977 through 1989
ACI Commentaries [12-13] defined “no appreciable sway” as being a lateral deflection of
¢>/ … 1>1500 of the story height at factored load levels. Gravity loads will cause small
lateral deflections, except in the case of symmetrically loaded, symmetrical frames. The
sway moment, Ms, is defined in ACI Code Section 2.1 as the factored end moment on a
column due to loads which cause appreciable sidesway, calculated by a first-order elastic
frame analysis. These moments result from either lateral loads or large unsymmetrical
gravity loads, or gravity loads on highly unsymmetrical frames.
Treating the P–d and P– ¢ moments separately simplifies design. The nonsway
moments frequently result from a series of pattern loads, as was discussed in Chapter 10.
The pattern loads can lead to a moment envelope for the nonsway moments. The maximum
end moments from the moment envelope are then combined with the magnified sway
moments from a second-order analysis or from a sway moment-magnifier analysis.
 Section 12-6 Calculation of Moments in Sway Frames • 603

12-6 CALCULATION OF MOMENTS IN SWAY FRAMES

USING SECOND-ORDER ANALYSES

First-Order and Second-Order Analysis

A first-order frame analysis is one in which the effect of lateral deflections on bending
moments, axial forces, and lateral deflections is ignored. The resulting moments and
deflections are linearly related to the loads. In a second-order frame analysis, the effect of
deflections on moments, and so on, is considered. The resulting moments and deflections
include the effects of slenderness and hence are nonlinear with respect to the load. Because
the moments are directly affected by the lateral deflections, as shown in Eq. (12-35), it is
important that the stiffnesses, EI, used in the analysis be representative of the stage
immediately prior to yielding of the flexural reinforcement.

Second-Order Analysis

Load Level for the Analysis

In a second-order analysis, column moments and lateral frame deflections increase more
rapidly than do loads. Thus, it is necessary to calculate the second-order effects at the
factored load level. Either the load combinations, load factors, and strength-reduction
factors from ACI Code Sections 9.2 and 9.3 or those from ACI Code Appendix C can be
used. We will continue to use the load combinations, load factors, and strength-reduction
factors from ACI Code Sections 9.2 and 9.3.

Lateral Stiffness-Reduction Factor

The term 0.75 in the denominator of Eq. (12-24) is the lateral stiffness-reduction factor,
fK, which accounts for variability in the critical load, Pc, and for variability introduced by
the assumptions in the moment-magnifier calculation. This factor leads to an increase in
the magnified moments. A similar but higher stiffness-reduction factor appears in a second-
order analysis. Two things combine to allow the use of a value of fK larger than 0.75 when
a second-order analysis is carried out. First, the modulus of elasticity, Ec, used in the frame
analysis is based on the specified strength, fcœ , even though the deflections are a function of
the Ec that applies to the mean concrete strength, that is, a concrete strength which is 600
to 1400 psi higher than fcœ . Second, the second-order analysis is a better representation of
the behavior of the frame than the sway magnifier given later by Eq. (12-41). The moments
of inertia given in ACI Code Section [Link] have been multiplied by 0.875, which,
when combined with the underestimate of Ec, lead to an overestimation of the second-
order deflections on the order of from 20 to 25 percent, corresponding to an implicit value
for fK of 0.80 to 0.85 in design based on second-order analyses.

Stiffnesses of the Members

Ultimate Limit State. The stiffnesses appropriate for strength calculations must estimate
the lateral deflections accurately at the factored load level. They must be simple to apply,
because a frame consists of many cross sections, with differing reinforcement ratios and
differing degrees of cracking. Furthermore, the reinforcement amounts and distributions
are not known at the time the analysis is carried out. Using studies of the flexural stiffness
of beams with cracked and uncracked regions, MacGregor and Hage [12-12] recommended
that the beam stiffnesses be taken as 0.4Ec Ig when carrying out a second-order analysis. In
604 • Chapter 12 Slender Columns

ACI Code Section [Link], this value has been multiplied by a stiffness-reduction factor
of 0.875, giving I = 0.35Ig.
Two levels of behavior must be distinguished in selecting the EI of columns. The lat-
eral deflections of the frame are influenced by the stiffness of all the members in the frame
and by the variable degree of cracking of these members. Thus, the EI used in the frame
analysis should be an average value. On the other hand, in designing an individual column
in a frame in accordance with Eq. (12-24), the EI used in calculating dns must be for that
column. This EI must reflect the greater chance that a particular column will be more
cracked, or weaker, than the overall average; hence, this EI will tend to be smaller than the
average EI for all the columns acting together. Reference [12-12] recommends the use of
EI = 0.8Ec Ig in carrying out second-order analyses of frames. ACI Code Section [Link]
gives this value multiplied by 0.875, or EI = 0.70Ec Ig for this purpose. On the other hand,
in calculating the moment magnifiers dns and ds from Eqs. (12-24) and (12-41), EI must be
taken as given by Eqs. (12-15) or (12-16).
The value of EI for shear walls may be taken equal to the value for beams in those
parts of the structure where the wall is cracked by flexure or shear and equal to the value
for columns where the wall is uncracked. If the factored moments and shears from an
analysis based on EI = 0.70Ec Ig for the walls indicate that a portion of the wall will
crack due to stresses reaching the modulus of rupture of the wall concrete, the analysis
should be repeated with EI = 0.35Ec Ig for the cracked parts of the wall.

Serviceability Limit State. The moments of inertia given in ACI Code Section [Link]
are for the ultimate limit state. At service loads, the members are cracked less than they are at
ultimate. In computing deflections or vibrations at service loads, the values of I should be
representative of the degree of cracking at service loads. ACI Code Section [Link] also
includes two other equations to calculate a refined value for the section moment of inertia,
I, when all of the member loads and longitudinal reinforcement are known. These equations
can be used to calculate deflections under either service loads or ultimate loads [12-23],
[12-24]. The equation for compression members (columns) is

25A st Mu 0.5Pu
I = ¢ 0.8 + ≤ ¢1 - - ≤ Ig … 0.875Ig (12-37)
Ag Puh Po
(ACI Eq. 10-8)

where Pu and Mu are either for the particular load combination under consideration or are
from the combination of loads that results in the smallest value of I in Eq. (12-37). For
service loads, the Pu and Mu values can be replaced by the acting loads. The value of I from
Eq. (12-37) does not need to be taken less than 0.35 Ig .
The equation for flexural members (beams and slabs) is

0.2bw
I = 10.10 + 25r2 ¢ 1.2 - ≤ Ig … 0.5Ig (12-38)
d
(ACI Eq. 10-9)

where r is the tension-reinforcement ratio at the section under consideration. For continu-
ous flexural members, I can be taken as the average of the values obtained from Eq. (12-38)
for the sections resisting the maximum positive and the maximum negative moments. The
value of I from Eq. (12-38) does not need to be taken less than 0.25 Ig .
 Section 12-6 Calculation of Moments in Sway Frames • 605

If the frame members are subjected to sustained lateral loads, the I-value in Eq. (12-37)
for compression members must be divided by 1 + b ds , where b ds is defined in Eq. (12-26b),
which is given in the following paragraphs.

Foundation Rotations

The rotations of foundations subjected to column end moments reduce the fixity at the
foundations and lead to larger sway deflections. These are particularly significant in the
case of shear walls or large columns, which resist a major portion of the lateral loads.
The effects of foundation rotations can be included in the analysis by modeling each
foundation as an equivalent beam having the stiffness given by Eq. (12-33).

Effect of Sustained Loads

Loads causing appreciable sidesway are generally short-duration loads, such as wind or
earthquake, and, as a result, do not cause creep deflections. In the unlikely event that sus-
tained lateral loads act on an unbraced structure, the EI values used in the frame analysis
should be reduced. ACI Code Section [Link] states that in such a case, I shall be divided
by 11 + b ds2, where, for this case, b ds is defined in ACI Code Section 2.1 as

maximum factored sustained shear within a story

b ds = (12-26b)
total factored shear in that story

Methods of Second-Order Analysis

Computer programs that carry out second-order analyses are widely available. The principles
of such an analysis are presented in the following paragraphs. Methods of second-order
analysis are reviewed in [12-12] and [12-25].

Iterative P– ¢ Analysis
When a frame is displaced sideways under the action of lateral and vertical loads, as shown
in Figs. 12-30 and 12-31, the column end moments must equilibrate the lateral loads and a
moment equal to 1©P2¢; that is,
© 1Mtop + Mbtm2 = V/c + ©P¢ (12-35)
where ¢ is the lateral deflection of the top of the story relative to the bottom of the story.
The moment ©P¢ in a given story can be represented by shear forces, 1©P2 ¢>/c, where
/c is the story height, as shown in Fig. 12-31b. These shears give an overturning moment
of 1©P¢>/c2 * 1/c2 = 1©P2¢. Figure 12-31c shows the story shears in two different
stories. The algebraic sum of the story shears from the columns above and below a given
floor gives rise to a sway force acting on that floor. At the jth floor, the sway force is
1©Pi2¢ i 1©Pj2¢ j
Sway force j = - (12-39)
/i /j
where a positive P¢>/ moment and a positive sway force both correspond to forces that
would overturn the structure in the same direction as the wind load would. The sway forces
are added to the applied lateral loads at each floor level, and the structure is reanalyzed,
giving new lateral deflections and larger column moments. If the deflections increase by
more than about 2.5 percent, new ©P¢>/ forces and sway forces are computed and the
606 • Chapter 12 Slender Columns

Fig. 12-31
Iterative P– ¢ analyses.

structure is reanalyzed for the sum of the applied lateral loads and the new sway forces.
This process is continued until convergence is obtained.
At discontinuities in the stiffness of the building or discontinuities in the applied
loads, the sway force may be negative. In such a case, it acts in the direction opposite that
shown in Fig. 12-31c.
Ideally, one correction should be made to this process. Although the P– ¢ moment
diagram for a given column is the same shape as the deflected column, as shown in Fig. 12-30c,
the moment diagram due to the P¢>/ shears is a straight-line diagram, similar to the V/
moments shown in Fig. 12-30b, compared to the curved-moment diagram. As a result, the
area of the real P– ¢ moment diagram is larger than that of the straight-line moment diagram.
 Section 12-6 Calculation of Moments in Sway Frames • 607

It can be shown by the moment–area theorems that the deflections due to the real diagram
will be larger than those due to the P¢>/ shears. The increase in deflection varies from zero
for a very stiff column with very flexible restraining beams, to 22 percent for a column that is
fully fixed against rotation at each end. A reasonable average value is about 15 percent. The
increased deflection can be accounted for by taking the story shears as g©P¢>/, where g is
a flexibility factor [12-25] that ranges from 1.0 to 1.22 and can be taken equal to 1.15 for
practical frames. Unfortunately, most commercially available second-order analysis programs
do not include this correction. For this reason, we shall omit g also.
Direct P– ¢ Analysis for Sway Frames
The iterative calculation procedure described in the preceding section can be described
mathematically as an infinite series. The sum of the terms in this series gives the second-
order deflection
¢o
¢ = (12-40)
1 - g1©Pu2¢ o>1Vus/c2
where
Vus = shear in the story due to factored lateral loads acting on the frame above the
story in question
/c = story height
©Pu = the total axial load in all the columns in the story
g M 1.15
¢ o = first-order deflection of the top of a story relative to the bottom of the story,
due to the story shear, Vus
¢ = second-order deflection
Both ¢ o and ¢ refer to the lateral deflection of the top of the story relative to the bottom of
the story.
Because the moments in the frame are directly proportional to the deflections, the
second-order moments are

Mo
M = ds Ms = (12-41)
1 - g1©Pu2¢ o>1Vus/c2

where Mo and M are the first- and second-order moments, respectively.

ACI Code Section [Link] defines the stability index for a story as

©Pu ¢ o
Q = (12-21)
Vus/c
(ACI Eq. 10-10)

Substituting this into Eq. (12-41) and omitting the flexibility factor g gives

Ms
ds Ms = Ú Ms (12-42)
1 - Q
Reference [12-25] recommends that the use of Eq. (12-42) be limited to cases where ds is
less than or equal to 1.5, because it becomes less accurate for higher values. This corre-
sponds to Q … 1>3. For this reason, ACI Code Section [Link] limits the use of Eq. (12-42)
to Q … 1>3. Example 12-3 illustrates the impact of Eq. (12-42) on magnified moments in
a frame.
608 • Chapter 12 Slender Columns

12-7 DESIGN OF COLUMNS IN SWAY FRAMES

Overview

For the 2008 edition of the ACI Building Code, sections dealing with the design of slender
columns were rewritten and rearranged. New equations were added, as discussed previously
for Eqs. (12-37) and (12-38), and other code sections were significantly modified. Because
of the general availability of structural analysis software that includes secondary bending
effects in sway frames, recommendations for the use of either nonlinear or elastic second-
order analyses are presented first in ACI Code Sections 10.10.3 and 10.10.4, respectively.
Moment magnification procedures, which are used in conjunction with first-order analysis
methods, are introduced in ACI Code Section 10.10.5, and then specific applications
for nonsway and sway frames are covered in ACI Code Sections 10.10.6 and 10.10.7,
respectively.
A separate stability check for sway frames, which was required in prior editions of
the ACI Code, is now covered by ACI Code Section [Link], which requires that the total
secondary moments in compression members shall not exceed 1.4 times the first-order
moments calculated for that member. Prior analytical studies [12-12] have shown that the
probability of a stability failure increases when the stability index, Q, calculated in
Eq. (12-21), exceeds 0.2. This is similar to the limit of 0.25 set by ASCE/SEI 7-10 [12-26]
for their definition of a stability coefficient denoted as u. Using a value of 0.25 results in
a secondary-to-primary moment ratio of 1.33, which is the basis for the ACI Code limit of
1.4 for that ratio. If that limit is satisfied, a separate stability check is not required.
ACI Code Section [Link] states that secondary bending effects shall be consid-
ered along the length of a compression member, including columns that are part of a sway
frame (sway story). Normally, structural analysis software that includes second-order
effects will give the maximum moments at the ends of a column. For slender columns
within a sway frame, it is possible that the maximum secondary moment will occur between
the ends of the column. This could be accounted for by either subdividing the column
(adding extra nodes) along its length or by applying the moment magnification procedure
for nonsway frames, which was covered earlier in Sections 12-2 and 12-4.

Computation of ds Ms
The following subsections discuss the ACI Code procedures for computing the magnified
sway moments, dsMs , in a sway frame.

1. Computation of ds Ms Using Second-Order Analyses

ACI Code Sections 10.10.3 and 10.10.4 allow the use of second-order analyses to compute
ds Ms. Elastic second-order analyses were discussed in Section 12-6. If torsional displace-
ments of the frame are significant, a three-dimensional second-order analysis should be used.

2. Computation of ds Ms Using Direct P– ¢ Analysis

As part of the moment magnification method for sway frames, ACI Code Section 10.10.7
permits the use of a direct calculation of P– ¢ moments via an equation similar to Eq. (12-42).
In ACI Code Section 10.10.7, this is written as
1
ds = Ú 1 (12-43)
1 - Q
(ACI Eq. 10-20)
 Section 12-7 Design of Columns in Sway Frames • 609

where
©Pu ¢ o
Q = (12-21)
Vus/c
(ACI Eq. 10-10)

Although there is no easy way to incorporate torsional effects into the calculation of ds Ms,
Section 12-9 discusses the extension of the Q method to second-order torsion.

3. Computation of ds Ms Using Sway-Frame Moment Magnifier

ACI Code Section 10.10.7 also allows the use of the traditional sway-frame moment mag-
nifier for computing the magnified sway moments

1
ds = Ú 1 (12-44)
1 - ©Pu>10.75 ©Pc2
(ACI Eq. 10-21)

Here, ©Pu and ©Pc refer to the sums of the axial loads and critical buckling loads, respec-
tively, for all the columns in the story being analyzed. In this case, the values of Pc are cal-
culated by using the effective lengths, k/u, evaluated for columns in a sway frame, with
b ds defined as

maximum factored sustained shear in the story

b ds = (12-26b)
total factored shear in the story

In most sway frames, the story shear is due to wind or seismic loads and hence is not
sustained, resulting in b ds = 0. The use of the summation terms in Eq. (12-44) accounts
for the fact that sway instability involves all the columns and bracing members in the
story. (See Fig. 12-30.) The format including the summations was first presented in
[12-1].
If 11 - ©Pu>0.75 ©Pc2 is negative, the load on the frame, ©Pu, exceeds the
buckling load for the story, ©Pc, indicating that the frame is unstable. A stiffer frame
is required.

Moments at the Ends of the Columns

The unmagnified nonsway moments, Mns, are added to the magnified sway moments,
ds Ms, at each end of each column to get the total design moments:
M1 = M1ns + ds M1s (12-45a)
(ACI Eq. 10-18)

M2 = M2ns + ds M2s (12-45b)

(ACI Eq. 10-19)

The addition is carried out for the moments at the top and bottom of each column. The
larger absolute sum of the resulting end moments for a given column is called M2, and the
smaller is called M1. By definition, M2 is always taken as positive, and M1 is taken as
negative if the column is bent in double curvature.
610 • Chapter 12 Slender Columns

Maximum Moment between the Ends of the Column

In most columns in sway frames, the maximum moment will occur at one end of the column
and will have the value given by Eqs. (12-45a) and (12-45b). Occasionally, for very slender,
highly loaded columns, the deflections of the column can cause the maximum column P–d
moment to exceed the P– ¢ moment at one or both ends of the column, in a fashion analo-
gous to the moments in the braced frame shown in Fig. 12-22c. As stated in the overview for
this section, ACI Code Section [Link] calls attention to this potential problem but does
not offer guidance. The ACI Commentary Section R.[Link] does suggest that this can be
accounted for in structural analysis by adding nodes along the length of the column.
This is a rare occurrence and prior editions of the ACI Code used Eq. (12-46) to
identify columns that may have P – d moments between the ends of the column that exceed
the P – ¢ moments at the ends.
/u 35
7 (12-46)
r Pu
œ
A fc A g

If /u>r exceeds the value given by Eq. (12-46), there is a chance that the maximum
moment on the column will exceed the larger end moment, M2 [12-27]. This would occur
if Mc, computed from Eq. (12-23), was larger than the end moments M1 and M2 from Eqs.
(12-45a) and (12-45b). If Mc 6 M2, the maximum design moment is at the end of the col-
umn and is equal to M2. If Mc Ú M2, the maximum design moment occurs between the
ends of the column and is equal to Mc.

Sidesway Buckling Under Gravity Loads

Another very rare event is the classical case of sidesway buckling under gravity loads
alone. As stated in the overview, ACI Code Section [Link] guards against this by requiring
that the secondary-to-primary moment ratio shall not exceed 1.4.

Minimum Moment

The ACI Code specifies a minimum moment M2,min (Eq. 12-22) to be considered in the
design of columns in nonsway frames, but not for columns in sway frames. This will be a
problem only for the gravity load combination U = 1.2D + 1.6L acting on a sway frame,
because this load combination does not involve ds Ms. For this load combination, we shall
design for the larger of M2 and M2,min.

EXAMPLE 12-3 Design of the Columns in a Sway Frame

Figure 12-32 shows the elevation and main floor plan for a five-story building. The
building is clad with nonstructural precast panels. There are no structural walls or other
bracing. The floor beams in the north–south direction are all 18 in. wide with an overall
depth of 30 in. The floor slabs are 6 in. thick. Assume all the floors are to be designed for
a superimposed dead load of 20 psf plus a live load of 80 psf, which includes a 20 psf par-
tition loading. The roof is assumed to have similar-sized structural members, carries a
superimposed dead load of 25 psf, and has a reduced live load, Lr , of 30 psf. Design an
interior and exterior column in the first-story level for dead load, live load, and wind