423

DIRECT ANALYSIS APPROACH FOR THE

ASSE MENT OF FRAME STABILITY:
VERIFICATION TUDIES

Andrea E. Malec'; & Donald W. White1

INTRODUCTION

In (Maleck and White 2003a and b), the authors have outlined an
alternative approach for analysis and design of steel framing systems,
lenned the Modified Elastic Approach. This method accounts for key
factors that affect system strength, specifically nominal residual stress
and geometric imperfection etTects, directly within a second-order
elastic analysis. In as such, a more rational analysis-design procedure
is obtained that eliminates the need for etTective length factors or
buckling solutions.

Various fonns of the Modified Elastic melhod have been discussed in

the recent literature, e.g., (Maleck and White 2001 and Deierlein et al.
1002). (Maleck and White 2oo3b) presents a summary of benchmark
validation studies from (Maleck 2001). (Maleck and White 2003a)
gives an overview of the Modified Elastic approach, and presents an
example design solution for a representative stability critical industrial-
Iype frame. This frame was originally studied in (Maleck 2001), and
example calculations have been discussed previously by Maleck and
While (200 I) and by Deierlein el al. (2002). However. in Ihese
references, resistance factors were applied only to the beam-column
strength tenns p. and M,. Subsequent studies have shown that in
general, resistance (~) factors must be applied to the nominal stiffnesses

I Senior Engineer. Engmeering Diagnostics. Inc .. Austm, TX

1 Associate Professor. Georgia Institute of Technology. Atlanta. GA
424 Maleck. While

in addition to the nominal strengths. The benchmark validation studies

in (Maleck 2001) and (Maleck and White 2oo3b). which are presented
in terms of nominal resistances. are unalTected by this issue. This is
beeause. if both the stilTness and strength terms are factored by an
approp riate uniform ~ prior to conducting the analysis and design
calcu lations, the results are identical to those obtained if the nominal
strengths and stifTnesses arc used in the calculations but the beam-
column interaction curves are subsequently factored along both axes by
the above ~ value.

In addition. the analysis-design solutions illustrated in (Maleck and

White 200 I) and (Deierlein et al. 2002) used a nominal rigidity of the
members of tEl, where T is a inelastic stiffness reduction factor
associated with the column strength. The original developments in
(Maleck 2001) were based on nominal rigidities of 0.9tEI for strong-
axis bending and O.StEI for weak-axis bending of I-shapes. with t
taken as the inelastic stilTness reduction factor associated with the
AISC-LRFD column curve (AISC 1999). Subsequent studies have
shown that these or similar minor reductions in the nominal section
rigidities. combined with the use of the ~ factor on the stilTness. are
essential to the accuracy of the method for structures that fail by elastic
or near elastic lateral sidesway buckling. Maleck and Wh ite (2003a)
gives an overall synthesis of the Modified Elastic procedure and
presents the results for the example frame using the base nominal
0.9tEI and O.StEI values from (Maleck 2001).

Recently. it has been suggested in discussions within AISC TCIO that

the use of the CRC t equation is desirable. and that a uniform factored
stilTness of O.StEI might be employed both for strong- and weak-axis
bending within the Modified Elastic procedure. Also. it has been
suggested that this procedure be referred to as the Direct Allalysis
method. This paper summarizes the key concepts associated with this
approach, and presents the results of fundamental benchmark
validation studies for this procedure.
 Direct AnalysIs Approach for the Assessment 425

THE OIRECf ANALYSIS APPROACH

OHrview. A general overview of the direct analysis approach is

presented here; for detailed discussions of the development of the
approach, the reader is referred to (Maleck & White 2003a) and
(Deierlein el. al 2002). The Direct Analysis approach is based on a
simple principle: if the parameters that afTect member nnd system
strength can be directly modeled (in a practical manner) within an
elastic analysis. the overall simplicity and rationality of the elastic
analysis-design calculations is improved. Based on SSRC Technical
Memorandum No. 5 (SSRC 1998), these parameters include (but are
not limited to) residual stresses, initial geometric imperfections and
boundary conditions. In the Direct Analysis approach, the efTects that
are not easily captured by simple modifications to elastic frame
analysis are addressed within the member resistance equations. The
modifications that can be easily made to an elastic analysis to better
estimate the strength limit stales behavior within the structural system
(e.g ., to more closely predict the internal forces obtained from a
rigorous distributed plasticity or plastic zone analysis) include:
I. Uniform reduction of the section flexural rigidity (EI,), based on
the level of axial force within the beam·coJumn members.
2. Specification of a nominal out-of-plumbncss, or lack-of-
verticality, of the structural framing.
By proper specification of nominal values for (I) the inelastic stifTness
reduction based on the behavior associated with the column strength
curve. and (2) column out-of-plumbness or a lack-of-verticality of the
structure based on erection tolerances such as specified in (AISC
2000), a simpler, more transparent and morc accurate analysis·design
approach is achieved.

Based on a wide range of studies. including the verification studies

presented in this paper, the following basic rules are suggested for
application of the Direct Analysis approach to tiered-type structural
framing:
426 Maleck, White

I) The nominal slifTnesses of all Ihe components wilhin Ihe

struclural syslem are faclored by a uniform value of 0.8. This
factor is applied to all the member rigidities, regardless of
orientation, and also to connection stifTncsses, column base
reslrainl slifTnesses, elc. (if a finite flexibility of these
components is considered in the design), There are two
contributors to the 0.8 factor on the stiffness. The first
contributor is a nominal reduction of 0.9. This reduction
factor accounts for the influence of distributed plasticity
efTects on the nominal stifTnesses as the strength limit stale
associated with the most critical component in the structural
system is approached . The second contributor is a resistance
+
factor of c 0.9. The product of these to factors is rounded
from 0.81 to 0.8.
2) For members in which the applied load p. exceeds O.SP" an
additional reduction factor
r=[P, 1 P, (I- P, I P, }] (I)

is applied, and therefore efTeclive member rigidities of

EI, = 0.8tEI (2)
are required for these types of members. In many types of
frames, Ihe O.SP, limit is not exceeded by Ihe required
strengths by any of the design load combinations. In cases
where this limit is exceeded, il is usually violated by only a
few columns within the structure. Furthennore, the value of
PII remains reasonably constant with design iterations in most
frames, and therefore an accurate bUI conservative value for t
can be selected rather easily.
3) A nominal frame nonverticality of WSOO is included in the
analysis by applying a notional load at each story level equal
to
N, - 0.002Y" (3)
 Direct Analysis Approach/or the Assessment... 427

where Y, is the factored design gravity load acting on the i~

story. Alternatively, the frame nonverticality may be directly
modeled by alteration of the perfect frame geometry.

If these minor modifications are performed in the context of a second-

order elastic analysis, the beam-column member strength checks can be
performed using the (AISC 1999) beam-column interaction equations
but with the P, term based on the actual member length. In short, the
need to calculate effective length iac/ors is eliminated, and thus the
member strength checks are greatly simplified.

While out-of-straightness can have an important influence on the

maximum strength of members in which the strength limit involves a
non-sway failure mode, the modeling of member out-of-straightness
within an analysis of the overall structural system is more cumbersome
than the modeling of a uniform frame nonverticality. In lieu of direct
modeling, the efTect of out-of-straightness on the strength is accounted
for in the axial strength term of the interaction equation.

One should note that the notional loads described by Eq. 3 are included
to model a physical attribute of the structure. They are not meant to be
a minimum horizontal load that can be neglected in the presence of a
larger applied lateral load; consequently they are additive to applied
lateral loads. For frames that have significant sidesway flexibility or
are subjected 10 large vertical loads. the influence of potential nominal
out-of-plumb imperfections on the internal second-order moments
within the structural system can be significant, even in the presence of
an applied lateral load (Maleck and White 2003a).

It should also be noted that the above reduction in the stifTness is

intended for use in the assessment of the strength only. Serviceability
limits should be checked using nominal stifTness values.

Special Considerations. For frames that are loaded near their vertical
load capacity and in which the structural system would tend to fail in
elastic sidesway buckling, the Direct Analysis approach behaves in a
428 Malec'" Wh,le

fashion that represents the true stability behavior more faithfully than
traditional buckling solution Or effective length based approaches. In
these type of structures. the strength of the system is reached due to
sIgnificant amplification of the sidesway deflections and the associated
internal moments as the limit of the structural resistance is approached.
This can be problematic if the Engmeer does not anticipate this
characteristic of the behavior. In traditional buckling solution or
effective length approaches, the internal forces and component
resistance ratios (i.e., the ratio of the required strengths to the design
resistances) tend to increase in only a mildly nonlinear fashion (due to
second-order elastic effects) as the design loads increase or if say some
of the components are reduced in size. However. when the Direct
Analysis Approach is applied to structures in which there are truly
large second-order amplifications of the lateral displacements based on
the reduced (or the actual inelastic) stiffnesses as the limit of resistance
is approached, the component resistance ratios can change in a highly
nonlinear fashion. For example, the engineer may find that the
interaction equation value for a beam·column member is 0.4, but if the
structure is checked for a slightly larger load, the interaction equation
value could rapidly increase beyond a value of 1.0 with only a small
increase in the required strength. These problems are significant only
for structures in which the sidesway amplification is excessive at the
required strength level. Therefore, they are likely to be important only
for certain special types of structures_

In general, the second-order analysis used in the direct analysis

approach must be rigorous; that is, the analysis should include both Pod
and P-O effects. Approximate Pod analysis methods are permitted only
if the applied axial loads on all columns satisfy the following limit:
(4)
where

(5)
and p. L is determined in the plane of bending. Equation 4 is a
conservative limit for which the influence of P-o moments on the
 Direct Analysis Approach for the Assessment .. 429

s,deSll'OY displacemenls can be neglected In lieu of a direct second-

order analysis. first-order analysis resulls may be modified by B I and
B2 amplification factors determined using the reduced stifTness.
However, the B2 amplification factor in (AISC 1999) is in efTect a P-d
analysis solution. If Eq . 4 is not satisfied, the form of this equation
given by LeMessurier (1977) including the CL term may be used to
account for the P-O effects on the sidesway deflections and internal
momenls. The B I amplification factor must be included in general
even when the axiallonds satisfy Eq. 4.

VERIFICATION ST DIES

Background. In (Maleck 2001), four small sensitive benchmark frame

configurations and two braced beam-column configurations were
studied to assess the accuracy of the proposed "Modified Elastic"
analysis-design approach. The complete set of ITames is similar to that
originally studied by Kanchanalai (1977) with the exception that the
efTects of initial geometric imperfections were included in the (Maleck
200 I) studies. Parameters considered in this study included
slenderness, member orientation (strong or weak-axis), beam-column
end restraint (G), and leaning column load (a). The results of these
studies are summarized in (Maleck and White 2003b).

In these studies, interaction curves were developed based on analysis

results from the "Modified Elastic" approach and the current LRFD
method for the studied frames . Both first-order (P versus M I) and
second-order (P versus M2) interaction curves were considered in these
studies. where
• M 1 is the maximum primary bending moment in the member
due to the applied loading. and
• M2 is the maximum internal second-order bendmg moment
The P versu~ M I interaction curves represented the maximum loadings
that can be applied to the benchmark structures; therefore. the P versus
M I interaction curves are referred to as both "applied loading" as well
430 Maleck, Wh,te

as "first-order" strength curves. These curves were compared to

interaction curves established by rigorous spread-or-plasticity analyses

As previously stated, these benchmark studies focused only on nominal

strengths. If only P, and M, are factored, IVltholit jactorlng the
st!IJness, then in the limit of structures that fail by elastic sidesway
buckling, factoring of P, and M, has a negligible innuence on the
calculated design resistance. The reason for this behavior is as follows.
In the limit of elastic sidesway buckling. the strength of the structure is
effectively controlled entirely by its elastic st!IJness. As the elastic
stability limit is approached, the internal moments tend to increase
dramatically such that large changes in internal moments are obtained
with only a small change 111 the externally applied loadings. As a
result. if the stiflhess is not reduced (e.g., by O.S) the design load
capacity in these types of structures is essentially predicted as the
nominal elastic sidesway buckling load, regardless of the fact that the
factored strengths f,P, (based on the actual unsupported length) and
",M, are used.

Design or Current Study. From the reasonably comprehensive set of

nominal strength studies performed in (Maleck 200 I), a subset of ten
strong-axis frame configurations and seven weak-axis frame
configurations are selected for additional study using factored strength
and stiffness values and the CRe tau equation (when required). The
subset of frames taken from (Maleck 200 I) that is used in this study are
shown in Fig. I. The chosen frames represent cases that exhibit the
largest unconservative and conservative errors in the initial nominal
study for either the direct analysis or current LRFD approach, or both.
Certain frames are also included because they exhibited significant
distributed plasticity efTects (SP_S60_GO, SP _SSO_GO) or failed by
elastic sides way buckling (UP_ S40_G I (12, SP_SO G3).
 Direct Analysis Approach/or the Assessment ... 431

~t
•"
j

1I<,_ up

••

'0
, '" I I
• .I "

tj
• ~ ·.L

~ .-~ ~
,

figu r. I. Benchmark Frames and Beam·Column (Maleck 200 I).

For Ihe plaslic zone Solulions. the Lehigh (Galambos and Keuer 1959)
residual stress pattern is used. and nominal geometric imperfections
equal 10 the fabricalion and ereclion lolerances in Ihe AISC Code of
Slandard Praclice (AISC 2000) are explicilly modeled. These
imperfections are:

60' tl500 (6)

and
00 - ti l 000 (7)
where
60 overall frame nonverticalily
432 Maleck, Wh,le

llo - maximum amplitude of a half sine wave member out-of-

straightness
I member length

When these parameters are used in a rigorous distributed plasticity

analysis. the resulting beam-column strength solutions closcly match
the current AISC LRFD beam-column strength procedures (AISC
1999) for the base case of W8x3 I members in strong-axis bending
(Maleck 2001. White and Clarke 1997. ASCE 1997). This should be
expected since the AISC LRFD bearn-column strength equations were
calibrated in part to the results from rigorous plastic zone solutions of
this type (ASCE 1997. Maleck and White 2oo3b). The analysis models
typically used for these distributed plasticity solutions have been based
on Euler-Bcrnoulli beam kinematics and in-plane response only.

LeMossurier's (1977) approach is used for the second-order elastic

design-analysis procedure in all of the sidesway-uninhibited
benchmarks considered wIthin this study. LeMessurier's method
accounts for both P-Iarge delta and P-small delta efTects and is. for all
practical purposes, exact for the sidesway problems studied in this
work. The "exact" closed-form analytical solutions are used to
determine the second-order elastic internal moments within the
sidesway-inhibited problem considered (beam-column BCS_. see Fig.
I). For the approximate (P-~) sidesway inhibited solutions, a B I
amplification factor is used in lieu of the "exact" closed-form solution.
For the direct analysis approach. the above solutions are based on the
efTective stifTness EI. (see Eq. 3). The approximate first-order
solutions are also detennined using LeMessurier's approach; in these
solutions the amplification term associated with P-O efTects is simply
omi"ed.

Result and Dis<ussion. A comparison of results for the studied

frames IS presented in Tables I (strong-axis) and Table 2 (weak-.xis).
The focus in this report is the error in the applied load curves (P vs.
M I); for a general discussion of the error associated with second-order
curves (P vs. M2), the reader is referred to (Maleck and White 2oo3b).
 Direct Analysis Approach f or the Assessment ... 433

Results of the factored solutions are presented for both the proposed
Direct Analysis approach and the current LRFD (1999) approach. For
both methods, errors from rigorous second-order analyses (P-li) and
approximate second-order analyses (P-t.) are presented. The error .s
defined as

(8)

where r" is the distance to the plastic zone strength along a radiallme
from the origin of the interaction curve plots. normalized in tcrm of
MIM, and PIP" and r is the corresponding distance along the same
radial line to the predicted design strength. Negative errors are
unconservativc.

In general, it can be seen from Tables I and 2 that the Direct Analysis
approach produces results with reasonable levels of unconservative
error in the applied load interaction curves, and improved accuracy of
these curves relative to that of the LRFD solutions, even for the cases
of maximum unconservative error. The error for the Direct Analysis
approach using a rigorous second·order analysis range from 6% to
+ 13 for strong axis bending and - 13 to +15 for weak-axis bending.
The corresponding LRFD errors range from - 8 to + 17 for strong-a.,is
and - 17 to + 17 for strong axis bending. The large unconservative
errors in the methods for the uBCS" problems in weak-axis bending
relate to the well known fact that the single AISC column curve
equation tends to give capacities that are somewhat liberal relative to
theoretical strengths for weak-axis buckling of I-shaped members
(Salmon and Johnson \996). The maximum and mmimum errors are
strongly dependent on the type of column end re traint, or more
specifically on the moment gradient within the members. The
conservative errors tend to be highest for the symmetric restrained-base
(SR.J frames, and these errors tend to be concentrated within the high
axial load regions for this case. Conversely. the highest uncollservative
errors are found in the braced, single curvature (BSC.) cases.
434 Maleck. White

Table I. Error in strong-axis benchma rk cases for rigorous ( P-O) and

approximate (P-6) second-order analyses

DIREC ANALYSIS LR

DHlgn.t1on C
P-.I
-
C c'
P-.<
C- c-
P-.I
C
. C
P- A
C
.
UP_S40_G1_cU 0 a 0 • 0 15 0 ,.
SP_52O_GO
SP_S<O_GO .
-2 3
3
·2
-,
3
3
-I
0
3
2
-1
0
3
2
SP_S60_GO ·3 0 -1' 0 •• • -. 3
SP_SIO_GO ·1
• -12 0 0 10 0 a
sp_SaO_G3 0
• 0
• 0 la 0 17
SR_S40_Gl 0 • 0
• 0 13 0 13
SR_SI O_GO
BCS_SIO .0
"
• .,.
0 12

• ....
0 3
2
0
-11
3
2
Bes suo -2 10 a • ·10 1
,
.....g. II -2
• I -. • II -2
, I ·3

Ta ble 2. Error in weak-axis benchma rk cases for rigorous (P-o) and

approximate (P-6) second-order analyses

o.RECT ANALYSIS LRFD

c-
P-.I
•
P-A
•
P-.I
. P-.<
.
O"~ nltl on

UP_W40_G1_0.2 -2
C

3
C

-3
C

1
C

0 I.
C C

0 I.
£

,. ."
-11 0 -23 0 -17 0 0
SP_W60_GO
,
SP_WIO_Gl
SR_W40_Gl
0
0 I. -2
0 I. ..
3 0
0 17
0

..
0
13

"3
SR_WlG_GO 0 a 0
• 3
De S_WIO •• 0 •• 0 · 10 0 ·13
.,.
0
BeS Wi 2D - 13 0
." 0 -13 0 0

....... II .• • ., • II ., 1 ., ,
 Direct Analysis Appraach/or the Assessment... 435

Applied load interaction curves are presented for the case of

SP_ SSO_GO in Fig. 2. This frame is representative of high error
associated with use of an approximate (P-L\) second-order analysis in
lieu of a rigorous (P-O) second-order analysis. In the absence of
applied moment, the maximum facto red strength predicted by the
plastic zone analysis is PIPy = 0.248; the conservative maximum
factored strength predicted by the Direct Analysis approach is PIP, =
0.236; and the unconservative prediction using an approximate second-
order analysis is PIP, = 0.27S . The maximum applied axial load
predicted by the plastic zone analysis exceeds the limit of PJP.L < 0. 15
(see Eq . 4) by approximately 25%, and the critical load predicted by
the Direct Analysis method exceeds this limit by approximately 21%.
There is an 18% difference in the resulting predictions of the critical
applied load between the rigorous and approximate second-order
analyses.

As can be seen in Fig. 2a, the Direct Analysis approach provides an

accurate estimate of the allowable applied loads when a rigorous
second-order analysis is used , Conservative error is present when the
axial load dominates interaction check ; however. up to the limit of Pu <
0. 15P. L , there is little to no error. For the SP_SSO_GO frame shown,
the maximum unconservative error is - 12% when an approximate
analysis is performed; this value is - S% at the limit where the axial
capacity predicted by the approximate analysis is equal to the
recommended limit of 0.15P.L . Similarly, for the SP_S60_GO the
unconservative error at P = O.l5P. L is - S%; this frame has the highest
increase in error when the P-O amplification effects are neglected (with
a maximum unconservative error of - 14% at the predicted axial
capacity in the absence of applied moment.) The unconservative error
associated with the approximate analysis for these critical cases at the
proposed applied axial force limit is somewhat high, but no larger the
maximum unconservative error predicted by the current LRFD
approach when a rigorous second-order analysis is used (see Table I).
436 Maleck, Wh,te

O'r----------------------------------,
-PZ ""1
- -...... AnIIy1i. RJgornuo"""'"
- . [)wed AnIItysI • . P·Oeb AnaItysts

02 I-__~...,,~:_-

0\

O~--------------------------~,

03
o

a.
02

Direct Analysis - " "

__~

~
-pz. ""1
--lRFO. RIgorous An8tysd
- -lRfD, P-OeHII aOIfyM

01 t-------------~~~".

O ~---- __- - - - - - - - - -__- - - -__~~~

o 02 04
MIM, " 00

b. LRFD (1999)

Figur.2. Frame SP_SSO_GO factored applied load curves.

 Direct Analysis Approach/or the Assessment 437

The LRFD results, presented in Fig. 2b, do not exhibit the sensitivity to
the second-order analysis method (P-~ vs. P-6) seen in the Direct
Analysis results (see Fig. 20). As can been seen in Table I, the
maximum error in the current LRFD approach is, in general, not
particularly sensitive to the accuracy of the second-order analysis
method used.

Figure 3 shows the applied load curves for SP W60_GO predicted by

the Direct Analysis approach. This frame exhibits the highest
unconservative error of the sway frames in weak·axis bending, and the
highe t increase in error when P-S effects are not directly included in
the analysis (see Table 2). At the limit that Pcr O.15PeL for the
approximate second-order analysis, the error is - 17%; this is equivalent
to the maximum unconservative error predicted by the LRFD approach
using a rigorous analysis .

• 5

•• ........
- o - d MeNd. RIgorous

- -Direct Melhod. P-o.tt •

• 3 An!Iy!!.
! .2
. 1'-

• o.
• 02 o. 01

Figur.3. Frame SP_W60_GO factored applied load curves, Direct

Analysis.
438 Maleclc, While

Results for the Direct Analysis method for frame SR_S80_GO are
shown in Fig. 4. The conservative error is typically highest in cases
where the member is subjected to reverse curvature bending. It can
also be seen from the plot that there is low sensitivity to the accuracy of
the second-order analysis; as would be expected since there is liule P-8
amplification of the moments in the case of reverse· curvature bending.
The above results suggest that the use of an appropriate equivalent
uniform rigidity (EI.) accounting for the effect of moment gradient
within the beam-column could produce improved accuracy over the
simpler approach selected in this research. However, making EI. a
function of the moment gradient would make its value more sensitive
to changes in the structure during design iterations and for different
design load combinations and add complexity to what is meant to be a
simple analytical procedure.

00

07

"
"
"o ~----

o 02 .. ..
__________--____________

MIllo
00
~~ __

Figure 4. Frame SR_S80_GO factored applied load curves, Direct

Analysis.

CONCLUSIONS

This paper summarizes the results of factored benchmark studies

designed to study the validity and accuracy of a variation on the
Modified Elastic method proposed by (Maleck and White 2002a),
 Direct Anal}~is Approachfor the Assessment ... 439

termed the Direct Analysis approach The Direct Analysis approach is

generally more accurate than the current AISC·LRFD (1999) method
in predicting allowable applied forces for stability critical benchmark
cases. Since the Direct Analysis method eliminates the need to
calculate column buckling loads or efTective length factors, it is simpler
to use than the current AISC·LRFD approach for many types of
problems. However, care must be taken when using approximate (p./!.)
second-order analysis algorithms in conjunction with the Direct
Analysis approach, as unconservative errors as high as ·14% for
strong-axis bending and -23% for weak-axis bending are achieved for
small. stability·critical benchmark frames. A limit on the applied axial
load of Pu < 0. 15 PeL for every column is suggested when an
approximate second-order analysis is used. A rigorous analysis that
includes P-O efTects is preferred for improved accuracy and reduced
unconservativc error.

REFERENCES

AISC (1999) Load and ReSIstance Factor Design Spectjicallon for

Steel Bllildings, Manual of Steel Construction· Load and Resist,,"ce
Factor Design, Third Edition. American Institute of Steel Construction,
Inc., Chicago, IL.

AISC (2000), Code of Standard Practice for Steel Blllidings and

Bridges, Manual of St..1 Construction - Load and Resistance Factor
Design, Second Edition, American Institute of Steel Construction, Inc.,
Chicago, IL.

ASCE (1997). EffectIve Length and NotIOnal Load Approaches for

AsseSSing Frame StabIlity: Implications for American Steel Design,
American Society of Civil Engin .. rs Structural Engineering Institute's
Task Commillee on EfTective Length under the Technical Commillee
on Load and Resistance Factor Design. 442 pp.

Deierlein, Gregory G., Haijar, Jerome F., Yura, Joseph A., White,
Donald W., and Baker, William F. (2002), "Proposed New Provisions
440 Maleclc, While

for Frame Stability Using Second-Order Analysis." Proceedings,

Annual Technical Session, Structural Stability Research Council,
Seattle, April

Galambos, T.V. and Ketter, R.L. (1959), "Columns Under Combined

Bending and Thrust," Journal o/Ihe Engineering Mechanics Division,
ASCE, 85(EM2), pp. \35- J52

Kanchanalai, T. (1977), "The Design and Behavior of Beam-Columns

in Un braced Steel Frames," AISI Project No. 189, Report No.2, Civil
Engineering/Structures Research Lab., University of Texas, Austin,
TX, 300 pp.

LeMessurier, W. J. (1977), "A Practical Method of Second Order

Analysis. Part 2: Rigid Frames," Engineering Journal, AISC, \3(4),
pp. 89-96.

Maleck, Andrea E. (2001), Second-Order Inelastic and Modified

Elastic Analysis and Design Evaluation of Planar Steel Frames, Ph.D.
Dissertation, Georgia Institute of Technology, 579 pp.

Maleck, A.E. and White. D.W (2001)., "A Modified Elastic Approach
for the Design of Steel Frames," Proceedings, Annual Technical
Session, Structural Stability Research Council, Fort Lauderdale, May.

Maleck, A. E. and White, D. W. (2003a), "Alternative Approaches for

Elastic Analysis and Design of Steel Frames. I: Overview," ASCE
Journal o/Slnlclllral Engineering, submllled/or review.

Maleck, A. E. and White, D. W. (2003b), "Alternative Approaches for

Elastic Analysis and Design of Steel Frames. II: Verification Studies,"
ASCE Journal a/Slroclural Engmeermg. submitted/or review

Salmon, Charles G. and Johnson, John E. (1996), Sleel SIn/clllres,

Design and BehaVior, Harper Collins, New York, 1024 pp.
 Direct Analysis Approach/or the Assessment 441

SSRC (1998), Guide to Stability Design Criteria/or Metal Stnlclllres,

5th Edition, T V. Galambos (ed.), Structural Stability research Council.
Wiley.

White. D.W. and Clarke, M.J. (1997), "Design of Beam·Columns in

teel Frames. I: Philosophies and Procedures, Journal 0/ Structural
Engineering, ASCE, 123( 12), 1556·1564.