Engineering

Journal
Fourth Quarter 2021 | Volume 58, No. 4

223 Design for Local Web Shear at Brace

Connections: An Adaptation of the Uniform
Force Method
Rafael Sabelli, Brandt Saxey, Chao-Hsien Li,
and William A. Thornton

267 Interaction Strength of Steel-Concrete

Composite Beam-Columns Including the
Balance Point
Mark D. Denavit

279 Lateral-Torsional Buckling Research Needs

and Validation of an Experimental Setup in
the Elastic Range
Ryan Slein, Joshua S. Buth, Wajahat Latif,
Ajit M. Kamath, Ammar A. Alshannaq,
Ryan J. Sherman, David W. Scott,
and Donald W. White

Steel Structures Research Update

293 Structural Fire Engineering
Judy Liu
Engineering
Journal
American Institute of Steel Construction

Dedicated to the development and improvement of steel construction,

through the interchange of ideas, experiences, and data.

Editorial Staff
Editor Margaret A. Matthew, PE
Managing Editor Keith A. Grubb, SE, PE
Research Editor Judy Liu, PhD
Production Editor Erika Salisbury

Officers
Stephen H. Knitter
Chair
Hugh J. McCaffrey
Vice Chair
Edward Seglias
Secretary/Legal Counsel
Charles J. Carter, SE, PE, PhD
President
Scott L. Melnick
Senior Vice President
Carly Hurd, CAE
Vice President
Lawrence F. Kruth, PE
Vice President
Brian Raff
Vice President
Mark W. Trimble, PE
Vice President

The articles contained herein are not intended to represent official attitudes,
recommendations or policies of the Institute. The Institute is not responsible
for any statements made or opinions expressed by contributors to this Journal.
The opinions of the authors herein do not represent an official position of the
Institute, and in every case the officially adopted publications of the Institute
will control and supersede any suggestions or modifications contained in any
articles herein.
The information presented herein is based on recognized engineering
principles and is for general information only. While it is believed to be
accurate, this information should not be applied to any specific application
without competent professional examination and verification by a licensed
professional engineer. Anyone making use of this information assumes all
liability arising from such use.
Manuscripts are welcomed, but publication cannot be guaranteed. All
manuscripts should be submitted in duplicate. Authors do not receive a
remuneration. Guidelines for authors are printed on the inside back cover.
Engineering Journal (ISSN 0013-8029) is published quarterly. Subscriptions:
Members: one subscription, $40 per year, included in dues; Additional Member
Subscriptions: $40 per year. Non-Members U.S.: $160 per year. Foreign (Canada
and Mexico): Members $80 per year. Non-Members $160 per year. Published
by the American Institute of Steel Construction at 130 E Randolph Street, Suite Subscriptions: [email protected], 312.670.2400
2000, Chicago, IL 60601.
Copyright 2021 by the American Institute of Steel Construction. All rights Archives: Search at aisc.org/ej. Article downloads
reserved. No part of this publication may be reproduced without written are free for current members and are avaialable for a
permission. The AISC logo is a registered trademark of AISC. nominal fee for non-members.
Design for Local Web Shear at Brace Connections:
An Adaptation of the Uniform Force Method
RAFAEL SABELLI, BRANDT SAXEY, CHAO-HSIEN LI, and WILLIAM A. THORNTON

ABSTRACT
Recent literature has examined local shear forces in beams in chevron braced frames (Fortney and Thornton, 2015). Subsequently, design
methods based on optimal stress distributions to address these shears were developed (Sabelli and Arber, 2017; Sabelli and Saxey, 2021).
This paper extends those design methods to gusset connections at columns, utilizing the adaptability of the Uniform Force Method to facili-
tate design to reduce required member shear strength. The design model presented reduces the required member shear strength required,
as compared to the conventional application of the Uniform Force Method. The model allows for redistribution of force from the beam inter-
face to the column interface using a “bypass method,” as well as utilizing the gusset plate as part of a moment connection using the “haunch
method.” Finite element analyses are used to confirm the adequacy of a design employing these methods.

Keywords:  gusset plates, braced frames, truss connections.

INTRODUCTION specifically developed to allow the designer to proportion

and analyze connections to reduce required member shear
D esign of brace connections requires consideration
of local forces induced in the surrounding framing
members. Although the discussion of these forces initially
strength and thus to reduce the instances of connections
requiring web doublers. Similar to the UFM as originally
developed, the methods presented in this paper rely on
focused on beam midspan connections (Fortney and Thorn-
the lower bound theorem as presented by Thornton (1984)
ton, 2015), such forces also occur at beam-column-brace
for similar connections, demonstrating adequate strength
connections. These local forces are typically missed by
through investigation of an advantageous load path in a
analysis methods that neglect connection dimensions. Rich-
ductile connection and examining forces at gusset edges.
ards et al. (2018) studied midspan gussets; their finite ele-
This paper addresses the UFM and adds two options. To
ment analyses confirm the presence of these local forces,
further optimize designs, a “bypass” method is developed
as well as redistribution of stresses similar to those pos-
that permits assigning more of the brace force to the col-
ited by Sabelli and Arber (2017). [A design procedure for
umn than is possible using the UFM. This approach rec-
beam-to-column connections with full-height gussets based
ognizes that shear yielding of the beam does not constitute
on such a stress redistribution is presented in Sabelli and
formation of a complete mechanism and that inelastic shear
Saxey (2021).]
deformation of the beam web requires inelastic deformation
The Uniform Force Method (UFM) (Thornton, 1991;
of the column web or gusset. (Even inelastic deformation of
Muir and Thornton, 2014; AISC, 2017) is commonly used
both the beam web and column web does not constitute a
to analyze forces at gusset-plate connections. It is a power-
complete mechanism, and additional strength can be mobi-
ful tool that permits designers to proportion and optimize
lized as is discussed in the bypass section.)
connections. This study presents an adaptation of the UFM
Additionally, a “haunch” method is developed, permit-
ting the use of the gusset as part of a moment-resisting con-
nection. Equations are presented for UFM forces; for UFM
forces in combination with bypass forces; and for UFM,
Rafael Sabelli, Director of Seismic Design, Walter P Moore, San Francisco, bypass, and haunch forces combined.
Calif. Email: [email protected] (corresponding) This paper begins with the derivation of design equa-
Brandt Saxey, Technical Director, CoreBrace, West Jordan, Utah. Email: tions for the three methods, along with the combined force
[email protected]
equations. For clarity, a summary table presents the equa-
Chao-Hsien Li, PhD Candidate, University of California at San Diego, Calif. tion numbers that apply for the design quantities for each
Email: [email protected]
method. Design examples are presented for each method.
William A. Thornton, Vice President, Cives Engineering Corporation, Roswell,
Ga. Email: [email protected]
Finally, there is a brief presentation of validating finite ele-
ment analyses of one of the design examples.
Each of the methods presented (the adapted Uniform
Paper No. 2020-06R Force Method and the bypass method and haunch method

ISSN 0013-8029 ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 223

enhancements) depends on the lower-bound theorem to jus- at a maximum in either of two locations for each member
tify the use of a relatively simple design model. While the (considering the column above and below the beam as sepa-
design model is not intended to produce forces matching rate members). The maximum member shear may be due to
those of more sophisticated analytical models, the methods the total normal force, which is fully delivered by the gusset
are expected to result in designs with adequate strength. The at the section nearest the workpoint. For the column, one
finite element analyses performed confirm this strength. such section is just above the beam top flange in Figure 1
Additionally, the analyses produce force distributions simi- (section C1); for the beam, it is at the beam-to-column con-
lar to those determined using the design model. nection (section B1). Additionally, moments at the gusset-
This paper focuses exclusively on the design of braced- to-column or gusset-to-beam interface may be large enough
frame-connections to achieve the required strength of each such that the associated shear in the member at or near the
component for a defined set of forces. The design forces mid-length of the gusset (sections C2 and B2) may exceed
considered are brace axial forces, beam axial and shear the value at C1 and B1.
forces, and (for the haunch method) beam moment. Inelas- The member shear checks at C1 and B1 are straightfor-
tic deformation demands related to brace buckling are not ward. These are simply the normal forces transverse to the
considered; as such, the methods are appropriate for wind member axis delivered by the gussets (horizontal forces for
design and for the design of buckling-restrained braced the column; vertical forces for the beam). In the case of a
frames but are not sufficient for providing the required duc- beam such as shown in Figure 1, the transverse forces from
tility of special concentrically braced frames. the gusset above and below are additive. Similarly, column-
shear forces from gussets on opposite column flanges are
Member Shear Checks additive.
At sections C2 and B2, the member shear is a function of
The presence of local shears in beam and column webs is
both the portion of the normal force delivered between the
necessary for static equilibrium. All beam-column-brace
gusset end and the section in question, and the moment on
(and truss) connections must resist these forces, regardless
the gusset-flange interface. The combined effects of flexure
of the connection design and analysis techniques used. The
and normal force can be analyzed using a number of models,
shear is due to both the force normal to the member axis
three of which are shown in Figure 2. The elastic and (con-
and to the moment at the gusset-flange interface. While dif-
ventional) plastic distributions are adapted from the AISC
ferent analysis methods assign different force distributions,
Steel Construction Manual (2017), hereafter referred to as
there is no method that eliminates both normal force and
the AISC Manual. The third method, the “concentrated
moment in connections of members with non-zero depth.
stress method,” is described by Sabelli and Saxey (2021).
For gusset-plate connections, local member shear may be

Fig. 1.  Sections for shear check in beam and column.

224 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

[The results from the concentrated stress method match the where
“optimized plastic method” from the AISC Manual (2017) L = gusset length, in.
for cases in which the section is fully plasticized.] In each M = moment at gusset interface to beam or column
diagram the flexural stresses are sufficiently large so that flange, kip-in.
the shear at an interior section (B2) is greater than the shear N = normal force on beam or column flange, kips
at B1. Note that the precise location of the section B2 var-
e = 
eccentricity in force couple resisting moment M
ies among the stress-distribution models in Figure 2. For a
equal to half the gusset length L, in.
given combination of moment and normal force, the length
required for a limit on the maximum stress differs for each This shear loading from Equation 1 must be combined with
model. additional shear that may be present in the member.
In this paper the conventional plastic distribution is used.
In that distribution, the sections B2 and C2 are always at the
gusset mid-length, which simplifies the equations derived Sign Conventions
based on the moments at the gusset interfaces. However, Figure  3 shows a typical brace-connection diagram. Sub-
the concentrated stress method requires a lower force to scripts are employed in some equations to distinguish
transmit the same moment over the same length and thus actions and dimensions related to one gusset or one brace
can provide for a more efficient design (Sabelli and Saxey, from another. Gussets are designated 1 and 2. Dimensions
2021). and forces associated with each gusset are given the sub-
Using conventional plastic distribution, the shear at sec- script 1 or 2.
tion B2 or C2 is: Braces have two subscripts. The first pertains to which
M N side of the column the gusset connects to (1 or 2). The sec-
Vmid = + (1) ond pertains to which of the two braces connecting to a
e 2

Fig. 2.  Stress-distribution models.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 225

beam is indicated (above or below). Sign conventions match vertical force corresponding to the brace axial force, P, is
the figures such that positive brace axial forces F1,2 and F2,1 given by AISC Manual Equation 13-2:
correspond to compression and positive brace axial force β
F1,1 and F2,2 correspond to tension. Forces and angles per- Vc = P (2)
r 
taining to each brace carry the same designation subscript.
The term workpoint refers to the intersection of brace where
centerlines with the beam and column centerline. (Eccen- r = gusset centroid offset from workpoint per AISC Man-
tric workpoints are not addressed in this paper.) ual Equation 13-6:
Sabelli and Saxey (2021) derive local member shears
= (α + ec ) + (β + eb ) 
2 2
(3)
based on a full-height gusset that has continuous thickness.
Similar column shears occur in traditional gussets, and and
similar methods may be employed to utilize the gusset size P = brace axial force, kips
and an optimal stress distribution to eliminate the need for eb = eccentricity from beam flange to beam centerline,
local reinforcement. equal to half the beam depth, in.
The Uniform Force Method (UFM) (AISC, 2017) is a ec = eccentricity from column flange to column center-
commonly employed method of analyzing traditional gus- line, equal to half the column depth, in.
sets. The UFM utilizes the gusset dimensions 2α and 2β, as α = distance from column face to centroid of Uniform
well as the beam and column eccentricities eb and ec. Note Force Method force acting on beam flange, in.
that 2α and 2β are the dimensions of a virtual gusset, and
β = distance from beam flange to centroid of Uniform
the actual gusset dimensions (2α and 2β) may be different,
Force Method force acting on column face, in.
as shown in Figure 4 and discussed later. The method pre-
sented here is a supplement to the UFM, showing necessary The column horizontal force is given by AISC Manual
dimensions to eliminate the need for web reinforcement. Equation 13-3:
For additional guidance and background on the applica-
tion of the UFM readers should consult the AISC Manual ec
(AISC, 2017) and AISC Design Guide 29 (Muir and Thorn- Hc = P (4)
r 
ton, 2014).
The beam vertical force is given by AISC Manual
Force Equations Equation 13-4:

The UFM defines vertical and horizontal forces at the gus- eb

Vb = P (5)
set interfaces with the column and the beam. The column r 

Fig. 3.  Diagram with numbering and sign conventions.

226 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

The beam horizontal force is given by AISC Manual These relationships may also be expressed as:
Equation 13-5: α = r sin θ − ec (10)
α
Hb = P (6) β = r cos θ − eb (11)
r 

For both edges of the virtual gusset to be centered on the These virtual dimensions locate the centroids of the
forces acting on the beam and column flanges, the virtual forces acting on the beam flange and column flange. As
dimensions α and β must conform to the following relation- presented here, the UFM gusset forces are constrained
ship (AISC Manual Equation 13-1): to conform to the proportioning relationships defined by
Equations 7 through 11, regardless of the actual gusset pro-
α − β tanθ = eb tanθ − ec (7) portioning. Specifically, Equations 9, 10, and 11 are used to
convert the force equations into functions of r, which can
where
then be selected to optimize the connection. Thus, although
θ = brace angle from vertical, deg
three dimensions (r, α, and β) are used in the following
Equation 7 may also be expressed as: equations, they are constrained to each other and represent
a single variable in the design.
α + ec
β= − eb (8) In the UFM procedure [both as defined by Thornton
tan θ  (1991) and as applied here], these gusset forces act on the
beam and column at the points indicated on Figure 4 and
Equation 3 may be combined with Equation 7 and Equa-
are proportioned such that their force vectors pass through
tion 8 thus:
“control points” at the beam centerline at the column face
eb + β (for beam forces Hb and V b) and at the column centerline
r= (9)
cos θ at the beam top or bottom elevation (for column forces Hc
ec + α and Vc).
= The virtual dimensions α and β do not necessarily cor-
sin θ 
respond to the centroid of the gusset welds or bolted joints.

Fig. 4.  Uniform Force Method dimensions.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 227

They locate the centroids of the shear and normal forces at dimension, r, based on those forces in comparison to the
the beam or column flange, respectively. At those locations, member shear strength. From that dimension, the virtual
there is effectively no moment. However, if that location dimensions α and β are defined by Equations  10 and 11.
is not also the centroid of the joint (i.e., when the gusset Once those virtual dimensions are computed, the actual
dimension α ≠ α or β ≠ β), there is a resulting moment due gusset dimensions 2α and 2β are selected. There is not a
to the eccentricity. The AISC Manual provides methods unique relationship between the force distribution (repre-
of determining the relationship between the ideal gusset sented by the virtual dimensions r, α, and β) and the gusset
dimensions (2α and 2β) and actual dimensions (2α and 2β) geometry (i.e., 2α and 2β); a range of dimensions 2α and 2β
and for determining the resulting moments (AISC, 2017). may be used with the same virtual dimensions α and β, and
For efficient design, it is convenient to determine the mini- vice versa. The engineer may optimize the design by select-
mum virtual dimensions α and β to control the transverse ing values of α and β to produce forces that do not overload
loading on the column and beam so as to avoid overloading the beam (or connection) and column in shear at the B1 and
these members in shear and to subsequently select smaller C1 sections and subsequently select among a range of pos-
gusset dimensions 2α and 2β independently (i.e., not con- sible gusset dimensions 2α and 2β that are consistent with
strained to each other) and design the gusset and its inter- those forces and do not overload the beam and column in
faces for the resulting combination of normal force, shear shear at the B2 and C2 sections.
force, and moment.
Figure  4 shows the UFM forces acting on the gusset,
GUSSETS RESISTING UFM
along with the dimensions used in the UFM. (Forces act-
BRACE FORCES ONLY
ing on the member are of opposite sign.) The diagram on
the left shows a gusset with virtual dimensions 2α and 2β; This section derives closed-form solutions for minimum
forces on the column act at the centroids of the interfaces virtual and actual gusset dimensions based on UFM gus-
with the virtual gusset with no moment. The diagram on set forces if no other forces are required to be transmitted
the right shows the smaller gusset of dimensions 2α and by the gussets. The subsequent sections develop the (more
2β; forces acting at the centroids of the actual gusset edges complicated) equations that include other forces that the
include moments due to the difference between the virtual designer may elect to assign to the gusset: bypass forces to
and actual gusset dimensions. transfer more force to the column than would result from
In the following derivations, the relationship between α the application of the UFM and haunch forces for beam-end
and β defined by Equation 7 is maintained. This simplifies moments.
the method, while still permitting independent selection of
actual dimensions 2α and 2β that are not constrained by Determination of Column Effective Shear Strength
this relationship.
Column and beam shears are readily obtained using the
Using Equations 7 through 11 with Equations 2 through
UFM; they are Hc and V b, respectively, combined with any
6, the force equations can be presented in terms of the vir-
shear in the member from other sources. (If the moment at
tual dimension, r:
the interface is large, the resulting effect on local shear may
⎛ e ⎞ increase the member shear, as is discussed later.)
Vc = cos θ − b P (12)
⎝ r⎠ 
Hc ≤ VefCol (16)
e
Hc = c P (13) where
r  VefCol = effective column shear strength
eb
Vb = P (14) The column effective shear strength is reduced by addi-
r 
tional shear demands that must be simultaneously resisted
⎛ e ⎞ by the column. For rigid beam-to-column connections, the
H b = sin θ − c P (15) moment at the column face due to lateral loads entails a
⎝ r⎠ 
panel-zone shear in the column web and a corresponding
The larger the virtual dimension, r, the larger the portion shear in the column segments above and below the con-
of the brace force that is resisted in shear at the interfaces nection. Typically, the panel-zone shear is not critical as its
with the column flange and beam flange (Vc and Hb, Equa- direction is opposite that of the connection shear induced by
tions 12 and 15), and the less is resisted in the normal forces the braces. The column shear outside of the panel zone, how-
that cause member shear (Hc and V b, Equations 13 and 14, ever, needs to be considered in determining the total shear
which are used to evaluate sections C1 and B1, respec- demand:
tively). Subsequent equations solve for the minimum virtual
VefCol = ϕVn − Vcol (17)

228 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

where and right of the column (e.g., P1,1 and P2,1, as shown in Fig-
Vcol = column shear due to moment-frame behavior, kips ure  3). In this method, it is apportioned considering their
Vn = nominal member shear strength, kips horizontal components as follows:
ϕ = resistance factor P1,1 sinθ
VefC1,1 = VefCol (22)
P1,1 sin θ + P2,1 sinθ 
Determination of Beam and Beam-to-Column
Effective Shear Strength P2,1 sin θ
VefC 2,1 = VefCol (23)
P1,1 sin θ + P2,1 sin θ
The shear demand from the brace connection must not
exceed the beam effective shear capacity, considering this This apportionment is arbitrary and may be modified based
shear demand present in the beam due to lateral loads. Grav- on the demands Hc1,1 and Hc2,1. Apportionment as shown in
ity shear, Vg, and frame shear, VBMF, may further reduce the Equations 22 and 23, however, permits independent design
effective beam shear strength: of the two gussets.
VefBm = ϕVn − Vg − VBMF (18) The condition at the beam is evaluated similarly for con-

nections with both a brace above the beam and another
where below the beam; only a portion of the beam shear capacity
VBMF = beam shear due to moment-frame behavior, kips may be utilized to resist each UFM vertical beam force, V b.
Vg = beam shear due to gravity, kips The effective shear capacity of the beam, VefBm, must be
apportioned between the braces above and below the beam
When the vertical brace force is acting upward, such as in (see Figure 3):
the case of the braces on left-hand side of Figure 3 (or the
P1,1 cosθ
case with braces on the right-hand side during the reversal of VefB1,1 = VefBm (24)
the indicated force), the gravity shear, Vg, is in the opposite P1,1 cosθ + P1,2 cos θ 
direction of the brace-induced shear, and thus increases the P1,2 cosθ
effective shear strength of the beam; reduced-gravity load VefB1,2 = VefBm (25)
P1,1 cos θ + P1,2 cosθ 
combinations are appropriate for such cases.
The shear strength of the beam connection to the column, This apportionment applies to both the beam shear strength
VefConn, may be less than that of the beam: and the connection shear strength. As with the apportion-
VefConn = ϕRn − Vg − VBMF ≤ VefBm (19) ment of column shear strength (Equations 22 and 23), this

apportionment is arbitrary and may be modified based on
where the demands (V b1,1 and V b1,2 in this case).
Rn = nominal strength (of beam-to-column connection),
kips Determination of Minimum Virtual Dimensions
For convenience, a strength ratio, UC, is defined: The virtual dimension, r, determines the shear at the col-
VefConn umn section C1 and the beam section B1, and thus the mini-
UC = ≤1 (20) mum value for this dimension is established based on the
VefBm  effective shear strength at those sections.
Thus, The horizontal force must not exceed the column effec-
tive shear strength. Combining Equations 13 and 16 gives:
VefConn = UC VefBm (21)
ec P
r ≥ rminCol = (26)
Note that the factor Uc varies with the load combinations VefC 

that correspond to the shears Vg and VBMF.
where
Apportionment of Shear Strength for Multiple Gussets rminCol = minimum dimension, r, based on column shear
yielding, in.
For connections with gussets both above and below (e.g.,
P1,1 and P1,2), the column shear must be checked for both The condition at the beam is evaluated similarly. The shear
separately. For connections with gussets both left and right is set equal to the beam shear capacity (deducting shear
of the column (e.g., P1,1 and P2,1), the column forces Hc1,1 due to other sources), and the minimum dimension, r, is
and Hc2,1 of these gussets are additive, and only a portion derived based on the effective shear capacity of the beam-
of the column shear capacity may be utilized to resist each to-column connection (considering shear demands on the
of these forces. The effective shear capacity of the column, beam not due to the brace force).
VefCol, must be apportioned between the braces to the left The vertical force on the beam due to the brace force

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 229

from the UFM analysis is set equal to the effective beam- The larger the deviation of the actual dimension β from the
connection shear strength apportioned to that gusset: virtual dimension β, the larger the moment that must be
Vb ≤ UCVefB (27) resisted at the gusset interface.

As shown in Figure 5, the column shear at section C2 at
Per Equation  19, the connection effective shear strength, the gusset mid-height combines the shear required to trans-
VefConn, may be limited by the effective beam shear strength, mit the moment with half of the interface horizontal force:
VefBm. Hc M c
Combining Equation 14 and Equation 27 gives: Vmid = + (31)
2 β
eb P
r ≥ rminBm = (28) ⎛ β 1⎞
UCVefB  = Hc −
⎝ β 2⎠ 
where
This shear must be less than the column effective (appor-
rminBm = m inimum dimension, r, based on beam shear
tioned) shear capacity from Equation 17:
yielding, in.
Vmid ≤ VefC (32)
The maximum length, r, from the check of the column and
the beam controls the design of the gusset (or the need for The minimum dimension β may be determined by combin-
reinforcement): ing Equations 31 and 32:
β
r ≥ max (rminCol ,rminBm ) (29) β≥ for r ≥ rminCol (33)
VefC 1
+
The value of r selected is used in subsequent equations Hc 2 
for determination of UFM forces. The corresponding vir-
tual gusset dimensions α and β (determined using Eqs. 10 The minimum gusset dimension obtained using Equa-
and 11) are used in the determination of the minimum actual tion 33 should be compared to the gusset size required for
dimensions α and β, respectively. Note that by selecting one the brace-to-gusset connection.
of these two values (rminCol or rminBm), the design method If the virtual dimension, r, is equal to rminCol, Equation 33
constrains the corresponding evaluation of member shear can be combined with Equations 4 and 26 and simplified to:
(section C1 or B1) to indicate a demand-to-capacity ratio 2
of 1.0. While larger values of the virtual dimension, r, may β≥ β for r = rminCol (34)
3 
be used, selection of the lowest value tends to minimize
the member shears. It is possible that greater economy in If r is greater than rminCol, Equation  34 will result in a
the gusset can be achieved with a larger dimension, r, for larger value than Equation 33, which gives the true mini-
designs that can accommodate higher member shear. mum length; nevertheless, Equation  34 may be used for
Similar designs are performed for each gusset, deter- convenience.
mining the minimum dimension, r, for beam or connection Figure 5 shows column shear diagrams in the column due
shear yielding and column shear yielding and apportioning to the effects of the horizontal force, Hc (distributed over
the effective shear strength, considering the gusset on the the gusset height), and the moment, Mc. (The effects of the
opposite side of the beam or column for the respective gus- vertical force, Vc, are not shown because they do not con-
set design. tribute to column shear.) Two cases are shown. In the upper
set of diagrams, the virtual column dimension, r, is selected
Selection of Actual Gusset Dimensions: such that the column shear is exactly equal to the column-
Column Interface effective shear strength, and thus r = rminCol. In this case,
the column shear at section C1 at the bottom of the gusset,
As discussed earlier, the minimum virtual gusset dimen-
Hc, will equal the effective shear strength, VefC. As such, the
sions can be established based on the shear strengths at
shear at section C2 at the midheight, Vmid, must not exceed
sections C1 and B1, and the minimum actual gusset dimen-
the shear at the bottom, and the moment (and the eccentric-
sions can be established based on the virtual dimensions
ity causing the moment) must consequently be limited. In
and the shear strengths at sections C2 and B2.
the lower set of diagrams, the column shear at section C1 at
The required gusset dimension 2β at the column is deter-
the bottom of the gusset, Hc, is less than the effective shear
mined based on the combined effects of the force, Hc, corre-
strength, VefC. As such, the shear at section C2 at the gusset
sponding to the virtual gusset dimension, r, and, if β ≠ β (as
mid-height can be larger than the shear at section C1, and a
is typical), a moment, Mc, on the gusset-column interface:
greater moment and eccentricity can be tolerated; a larger
Mc = Hc (β − β)(30) eccentricity results from selecting a smaller dimension β.

230 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

If β > qβ, the maximum shear occurs at section C1. Con- Vb Mb
versely, if β < qβ, the maximum shear occurs at section C2. Vmid = + (37)
2 α
(If β = qβ, the maximum shear occurs for the entire bottom ⎛ α 1⎞
half of the gusset height between sections C1 and C2; this = Vb −
⎝ α 2⎠
case is not shown in the figure.)
Note that Equation  34 can be used to determine the This shear must be less than the beam shear capacity:
required shear strength of the column based on a selected
gusset size 2β such as might be determined based on the Vmid ≤ VefB (38)
brace-to-gusset connection. Assuming r minCol controls, Note that as this check is performed away from the connec-
Equations 9, 13, 16, and 34 can be combined to estimate the tion, the full (apportioned) beam shear effective strength,
required shear strength: VefB, is used without the reduction related to connection
ec cos θP effective shear strength, UC.
VefC ≥ (35)
3 The minimum dimension, α, may be determined by com-
β + eb
2  bining Equations 37 and 38:
α
This value can also be used to design web reinforcement for α≥ for r ≥ rminBm or VefBm >VefConn (39)
an already selected column and gusset dimension 2β. VefB 1
+
Vb 2 
Selection of Actual Gusset Dimensions:
Beam Interface The minimum gusset dimension obtained using Equa-
tion 39 should be compared to the gusset size required for
The required gusset dimension 2α at the beam is deter- the brace-to-gusset connection.
mined similarly, based on the force distribution corre- If the virtual dimension, r, is equal to rminBm, Equation 39
sponding to the virtual gusset dimension α. Similar to the can be combined with Equation 27 and simplified to:
condition at the column, if α ≠ α, there is a moment, Mb, on
α
the gusset-beam interface: α≥ for r = rminBm and VefBm >VefConn (40)
1 1
Mb = Vb (α − α ) (36) +
UC 2 
The shear at section B2 at the gusset mid-length com- If VefBm = VefConn, Equation 39 can be further simplified to:
bines the shear required to transmit the moment with half
2
of the interface vertical force: α ≥ α for r = rminBm and VefBm = VefConn (41)
3 

Fig. 5.  Shear in column due to horizontal force and moment.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 231

If r is greater than rminBm, Equations 40 and 41 will result in eb P sin θ
larger values than Equation 39, which gives the true mini- VefB ≥ (43)
U
mum length. α ⎛1+ C ⎞ + UC ec
⎝ 2⎠ 
Figure 6 shows beam shear diagrams and vertical load-
ing for two cases. In the upper set of diagrams, the connec- This value can also be used to design beam-web reinforce-
tion effective shear strength, VefConn, is equal to the beam ment for a given beam and desired gusset dimension 2α.
shear effective strength, VefBm. The limit on shear demand
at section B1 at the connection, V b, and the shear demand Selection of Gusset Thickness
on section B2 at the gusset mid-length are thus the same.
The gusset and its connections must be evaluated for the
(This occurs when the maximum eccentricity is defined
combined effects of moment, horizontal, and vertical forces
by Equation  41 for cases in which r  = rminBm and VefBm  =
at both the column interface (Mc, Hc, and Vc), and the beam
VefConn.) In the lower set of diagrams, a connection weaker
interface (Mb, Hb, and V b). The von Mises yield criterion
than the beam is shown (Uc < 1.0), and thus a higher shear
may be used to determine the minimum gusset thickness,
than V b can be resisted at the gusset mid-length. (A shear at
tg, considering forces at the interface with the column:
section B2 higher than V b results from selection of α < qα.
If α > qα, the shear at section B2 is less than Vb; this case ⎛ ⎞
2
⎡ Hc ⎤
2
Vc Mc
is not shown in Figure 6 but is similar to the upper diagrams tg ≥ ⎜ ⎟ +⎢ + ⎥ (44)
⎝ ϕ0.6Fy 2β ⎠ ⎢ ϕFy 2β ϕFy ( 2β ) 4 ⎥
2
in Figure 5.) ⎣ ⎦
Similar to Equation 34, Equation 41 can be used to deter- 2
mine the required shear strength of the connection based on 1 ⎛ Vc ⎞ 2 ⎛ 2 Mc ⎞
= ⎜ ⎟ + ⎜ Hc + ⎟
a selected gusset size 2α: ϕFy 2β ⎝ 0.6 ⎠ ⎝ β ⎠ 
e sin θP
UCVefB ≥ b (42) where
3
α + ec Fy = material specified minimum yield stress, ksi
2 
The minimum thickness at the beam interface is similarly
If the beam shear strength is equal to the connection determined:
shear strength (that is, if UC = 1.0), Equation 42 also defines
the required beam shear strength. If the connection shear 2 2
strength is the limiting factor (that is, if UC < 1.0), the ⎛ Hb ⎞ ⎡ Vb Mb ⎤
tg ≥ ⎜ ⎟ +⎢ + ⎥ (45)
y 2α )
( 2
required beam shear strength is determined from Equa- ⎝ ϕ 0.6Fy 2 α ⎠ ϕ
⎣ yF 2 α ϕF 4⎦
tions 5, 28, and 40: 2 2
1 ⎛ Hb ⎞ ⎛ 2 Mb ⎞
= + Vb +
ϕFy 2α ⎝ 0.6 ⎠ ⎝ α ⎠ 

Fig. 6.  Shear in beam due to vertical force and moment.

232 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

 The larger thickness from Equations  44 and 45 should and diagonal elongation of the gusset in the direction trans-
be used. If the required gusset thickness is excessive, a verse to the brace axis. Such a mechanism engages member
larger value of β or α (as appropriate) may be selected. [If shear and flexure outside of the connection region. When,
the length β or α required for a certain gusset thickness is as is typical, such forces are not considered in the mem-
desired, the required gusset length is the root of a fourth- ber design, the design of the connection, to be consistent,
power polynomial; various solution methods are available, should not depend on such forces. In such cases, the com-
including trial-and error and computer solvers. A closed- plete mechanism is limited by the capacity of the beam
form solution may also be derived using Ferrari’s formula and column to withstand flexural forces not considered in
(Euler, 1765).] the design in conjunction with the axial forces and other
actions that were considered. The mechanism may never-
Normal Forces on Column and Beam theless represent significant reserve strength over that of the
design.
The column and beam should each be evaluated for web
Often there is a great deal of economy to be achieved by
local yielding and web crippling for a required strength, Ru,
taking advantage of the additional capacity of the column to
due to the combined effects of normal force and moment
relieve the beam of some of the force. The UFM provides a
over a bearing length, N, of β or α, respectively:
method of transferring some of the vertical force: “Special
⎛ β 1⎞ Case #2” (AISC, 2017). An alternative method is presented
Ru = Vmid = Hc − (46)
⎝ β 2⎠  here in which the brace force is divided into two compo-
nents: a “UFM force” that is delivered to the column and
⎛ α 1⎞ beam per the method described above and a “bypass force”
Ru = Vmid = Vb − (47)
⎝ α 2⎠  that is delivered only to the column. The apportionment
is achieved by selecting a factor, λ, between 0.0 and 1.0.
The UFM portion of the force is λP and is addressed using
GUSSETS RESISTING ADDITIONAL FORCES the UFM as described previously (with modified effec-
tive column shear strength). The remaining bypass force
The preceding derivations are based on the gussets being is (1−λ)P and is assigned to the column, which must resist
designed using the UFM to transmit the brace force to the additional vertical and horizontal forces, as well as the
the beam and column. The methods developed (including corresponding moment. Note that for the design approach
selection of dimensions to preclude the need for member anticipated (in which β < β), this moment is in the oppo-
shear reinforcement) can be extended to gussets designed to site direction from the UFM column moment. The factor,
resist other forces as well. In the following section, forces λ, may be selected to reduce the required length of gusset
corresponding to two refinements are developed: gussets on the beam or to eliminate the overstress in the beam that
designed with a “bypass” method to transfer more of the would necessitate a web doubler.
brace force to the column (and less to the beam) than would The resulting column forces due to the bypass force are:
result from the application of the UFM, and gussets act-
ing as moment-transferring haunches. Equations for mini- HcBP = (1 − λ ) P sin θ (48)
mum virtual and actual gusset dimensions considering the
total gusset forces (UFM, bypass, and haunch forces) are VcBP = (1 − λ ) P cos θ (49)
derived. H
= cBP
tanθ 
Force Transfer from Beam to Column Using
Bypass Method McBP = HcBP (eb + β) − VcBPec(50)

Gussets proportioned in the manner described in the previ- where

ous section will have sufficient strength and will not cause λ = brace force apportionment factor
local yielding in the beam and column webs. Nevertheless,
Figure 7 shows these forces acting on the gussets.
such gussets may be overly large or otherwise undesirable,
If the beam strength is the limiting factor in sizing the
or such a design might entail reinforcing the web of one of
virtual gusset (r = rminBm and VefConn = VefBm), the factor, λ,
the members (typically the beam). Additionally, the shear
required to avoid beam web doublers may be obtained by
yielding of the web of one of the members does not consti-
modifying Equation 42:
tute a complete plastic mechanism; both webs must yield to
allow this to occur. VefB ⎡ 3 ⎤
λ≤ ⎢ α + ec ⎥ (51)
In fact, a complete mechanism requires extensive shear eb sin θP ⎣ 2 ⎦
yielding of both member webs over the entire gusset region

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 233

If the beam strength is not the limiting factor, the λ factor The bypass method can be used to design connections
required to avoid beam web doublers may be obtained by that satisfy this relationship to preclude the need for
modifying Equation 43: reinforcement.
VefB ⎡ ⎛ UC ⎞ In some circumstances, it may be convenient to transfer
⎤
λ≤ ⎢α 1+ + UC ec⎥ (52) force from the column to the beam (rather than from the
eb sin θP ⎣ ⎝ 2 ⎠ ⎦
beam to the column). In such circumstances the engineer
may either derive similar bypass equations for the beam or
If this bypass-force method is used, Equations  26 and 28
may use a value of λ greater than 1.0, increasing the beam
should be modified as follows:
UFM forces and reducing the total column forces.
λ Pec
rminCol = (53)
VefC − HcBP  Use of Gussets as Haunches for Moment Transfer
λ Peb It is possible to transfer the beam moment utilizing the gus-
rminBm = (54) set as a haunch as shown in Figure 8. The use of gussets as
UCVefB 
haunches will affect the force transfer at the gusset edges,
Similarly, the equations for minimum values of β and α and these forces must be combined with those related to
must be modified to address bypass forces; Equations  33 the brace axial force. The resulting equations for the mini-
and 39 are not valid for such cases. Modified equations mum virtual gusset dimension, r, to preclude shear yielding
(Equations 77 and 78) considering total forces and moments (determined in a subsequent section), and the correspond-
(including haunch forces) are presented in a subsequent ing virtual dimensions α and β, are necessarily more com-
section. plicated. For convenience, the method presented utilizes the
Equations 53 and 54 may be combined with Equation 48 beam moment at the column face, Mf, rather than at a loca-
to facilitate member selection. The result is an equation for tion based on the (as yet undetermined) gusset horizontal
λ: dimension. Thus, if the moment is based on beam flexural
ec strength (as might be the case for seismic design) projection
VefC − UCVefB of that moment based on an assumed gusset dimension and
eb
λ = 1− (55) corresponding beam shear is necessary.
P sin θ 
M f = MBM + VBMF ( 2α ) (57)
This can be combined with Equation  52 to derive the
required relationship between horizontal gusset length and where
column and beam connection strength: MBM = moment at a beam section aligned with the gusset
edge face, kip-in.
eb P sin θ − VefC eb
α≥ (56)
U
VefB ⎛1+ C ⎞ The designer has some discretion in selecting the height
⎝ 2⎠  of the force couple utilized. To permit independent design

Fig. 7.  Force proportioning.

234 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

of the gusset (considering the possibility of no gusset on the Thus, the total moment from beam “i,” Mfi, is divided into
opposite beam flange or a gusset of yet-to-be-determined four (potential) parts:
size), the gusset is assigned one-half of the beam moment M fi = MBCi ,1 + MBCi ,2 + MHi ,1 + MHi ,2 (60)
to be resisted by a force couple over a vertical distance from
the beam centerline to the virtual gusset centroid defined by It is expected, however, that in most cases, ρ will be selected
β. (Other values of this depth may be used, but this value as either 1.0 or 0.0 for each flange of each beam as is shown
provides for efficient gusset designs using simpler equations in Figure 8, and thus two of the components of Equation 60
in determining the appropriate dimension, β.) will be of zero value.
As shown in Figure  8, at conditions with a gusset, the For simplicity, the beam-connection moment is assigned
horizontal force that forms part of this moment-resisting to the beam flange as a force, Pf :
force couple may be resisted by the beam flange connection
M BCi , j
to the column, the gusset, or a combination of the two. Pfi , j = (61)
The value of the moment is 2Mf, assuming that the 2 hoi
moment is resisted by two force couples: one involving the ρi, j M fi
=
top flange or gusset and the web connection and the other hoi 

involving the bottom flange or gusset and the web connec-
tion. If the connection involves only the top, for example, where
and there is no bottom gusset and no bottom-flange con- ho = distance between beam flange centroids
nection to the column, the value of the moment assigned to
the top flange or gusset and beam web force couple is the The haunch moment results in a horizontal force, HcH
full Mf. (gusset-designation subscripts omitted):
The moment is apportioned between a beam-connection MH
HcH = (62)
moment, MBC, and a haunch moment, MH, using a distribu- db
+β
tion factor ρ (between 0.0 and 1.0); thus: 2
ρi, j M fi (1 − ρ ) M f
MBCi , j = (58) =
2  2 (eb + β ) 

where where
Mf db = beam depth, in.
MH i , j = (1 − ρi, j ) i (59)
2 
To satisfy equilibrium, there is an opposite horizontal force,
ρ =beam moment apportionment factor (between HbH, at the gusset interface with the beam:
beam flange and haunch force in gusset)

Fig. 8.  Gusset plates acting as haunches.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 235

 HbH = HcH(63) Equation 62, the horizontal forces at opposite flanges may
not be equal, and the difference must be resisted in the
The eccentricity between HcH and HbH is β, and there is
beam web-to-column connection to satisfy equilibrium.
thus a moment, βHcH, that must be resolved. A counteract-
This force, NH, is:
ing force couple is determined, with a vertical force, VcH,
at the column flange and an opposite vertical force, V bH, NH = Pf1,2 + H bH1,2 − ⎡⎣Pf1,1 + H bH1,1⎤⎦ (67)

which is assigned to the location α for simplicity:
VbH = VcH The haunch method may be used in conjunction with the
(64)
UFM with or without applying the bypass method.
β HcH
=
α  Total Gusset Forces: UFM, Bypass, and Haunch Forces
At the gusset centroid, HcH causes a moment, McH, due to Figure  9 shows these forces acting on the gusset edges.
the eccentricity between β and β: Forces shown are gusset forces; member forces are equal
and opposite. Note that the beam UFM forces are reduced
McH = (β − β ) HcH  (65) by the factor λ. In most cases, the haunch forces can have
The moment due to the vertical force at the beam interface, a beneficial effect; the designer should consider whether it
MbH, is: is appropriate to neglect that term or take a reduced value
to address uncertainty in the level of moment. For seismic
M bH = (α − α )VbH  (66) design, the potential for cyclic inelastic drift may result in
moments in the opposite direction of those shown, which
The horizontal force acting at the beam midheight may could be a more critical case. Determination of appropriate
not completely cancel with the horizontal force from the combinations of design forces is outside of the scope of this
remaining half of the moment addressed at the opposite study. At a minimum, connections should have sufficient
flange, potentially necessitating a small additional force, strength to resist maximum brace forces and maximum
NH, to be resisted at the beam-to-column connection. As beam forces both combined with each other (with consistent
shown in Figure 8, the horizontal force acting on the beam directions of forces as shown) and separately.
at one flange (Pf1,2 + HbH1,2) has a corresponding force at The direction of the UFM, haunch, and bypass column
the opposite flange resisted by a combination of flange and forces is shown in Figure  9. The resulting total column
gusset forces (Pf1,1 + HbH1,1). Due to potentially different forces (considering the direction of each component) are
values of ρ used in Equations  58 and 59, and potentially determined:
different moment arms used to determine these forces in

Fig. 9.  Total gusset forces.

236 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

 HcTot = λ Hc + HcBP − HcH  (68) defined by the virtual dimension, β, which is typically
greater than β, causing a moment, λMc, (when evaluated at
VcTot = λVc + VcBP + VcH  (69) the centroid of the gusset-column interface.). The haunch
McTot = λ Mc − McBP − McH  (70) force, HcH, (which acts in the direction opposite to λHc) is
applied at the same elevation, and the bypass force, HcBP,
The shear in the column panel zone is: (which acts in the same direction as λHc) is located on the
VPZ = HcTot − Pf  (71) brace centerline where it crosses the column face; these two
forces cause moments in the opposite direction of Mc.
The total beam forces (including the UFM, haunch, and
bypass forces) are: Minimum Virtual Gusset Dimensions Considering
Haunch and Bypass Forces
HbTot = λ Hb + HbH  (72)
In previous sections, Equations 26 and 28 provide the mini-
VbTot = λVb − VbH  (73) mum virtual dimensions, r, for cases with UFM forces only;
M bTot = λ Mb − MbH  (74) Equations  53 and 54 address combined UFM and bypass
forces. If haunch forces are included in the gusset design,
These equations are used to select the minimum virtual equations for minimum virtual dimensions, r, become more
dimension, r, considering the shear loading of the column, complicated (but nevertheless solvable). A satisfactory
HcTot, at the beam flange and the shear loading of the beam, value for this virtual dimension can be determined by trial
V bTot, at the column flange. Once this is selected, the actual and error, such that the beam and column are not overloaded
dimensions 2α and 2β may be selected considering the in shear by V bTot and HcTot, respectively. Alternatively, the
shear loading of the column and beam at the mid-length following methods can be used.
of the gusset as it is affected by the moments (McTot and Combining Equations  16 and 68 results in a quadratic
MbTot) that are functions of these dimensions. The interface equation for the minimum virtual dimension, r, considering
moments are then computed based on the selected dimen- column web shear, rminCol:
sions and are used in conjunction with the vertical and hori-
rminCol 2 ⎡⎣ (VefC − HcBP ) cos θ⎤⎦ + rminCol (75)
zontal forces to design the gusset and its connections to the
beam and column. ⎡ Mf ⎤
Figure 10 shows a free-body diagram of the gusset. The ⎢(VefC − HcBP )( ec − eb ) + (1 − ρ) − λ Pec cos θ⎥
⎣ 2 ⎦
centroid of the UFM column force, λHc, is at the location
+ [ λ Pec ( eb − ec )] ≥ 0


Fig. 10.  Free-body diagram of gusset.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 237

 Combining Equations  28 and 73 results in a quadratic ⎛ α HcBP ⎞ 1
equation for the minimum virtual dimension, r, considering β ≥ β− (77)
⎝ tan θ HcTot ⎠ VefC 1
beam web shear: +
HcTot 2 
rminBm2 ⎡⎣sin θUcVefBm⎤⎦ + rminBm ⎡⎣MH − ecUcVefBm − λ Peb sin θ⎤⎦
The minimum gusset dimension, α, is likewise determined
⎡ e ⎤ by combining Equations 27, 37, 73, and 74:
+ ⎢λ Pebec − MH b ⎥ ≥ 0 (76)
⎣ cos θ ⎦  α
α≥ (78)
VefB 1
The largest of the roots from Equations 75 and 76 is used. +
VbTot 2 
Excessively large values of the minimum dimension, r, may
indicate that more of the moment, Mf, should be transferred Note that this is based on the beam shear at the gusset mid-
by the beam flanges. Note also that both the effective col- length, and thus the connection factor, UC, is not used.
umn shear strength, VefCol, in Equation  75 and the effec- Designers may wish to begin trials using Equations  77
tive beam shear strength, VefB, in Equation 76 include terms and 78 with λ = 1.0 (i.e., with no bypass force). If the gusset
that, similar to Mf, are reflective of the moment-frame dimension 2α is excessive, the value of λ may be decreased
behavior, and the use of consistent loads is recommended. and the dimension β may be selected using Equation 77.
As discussed in the design examples, consideration of cases
without these flexural forces may be appropriate to address Beam-Web-to-Column Connection
the range of possible conditions.
Equation 76 presupposes a value of λ. If the design con- The sum of the horizontal forces at the beam web-to-
siders both the condition with moment and the condition column connection, HbW, is:
with moment equal to zero (for which haunch forces are H bW = NB + H bTot1,2 − H bTot1,1 + NH  (79)
likewise zero), Equation 52 may be used to select a value
of λ for the latter condition. As illustrated in the design where
example, the value of λ determined for the zero-moment NB = beam axial force delivered to the connection, kips
condition may be sufficient for all conditions; otherwise,
Note that the horizontal forces, Hb, from the gussets above
trial and error may be used.
and below counteract each other. For connections with only
The maximum virtual dimension, r, from the check of
one gusset, the required strength of this connection will be
the column and the beam from Equations 75 and 76 controls
significantly larger; this may be an important consideration
the design of the gusset (or the need for reinforcement), and
in determining how much force to divert using the bypass
therefore, the larger value should be used for the virtual
method, which increases Hb.
dimension, r, in subsequent equations per Equation 29. As
The sum of the vertical forces at the beam web-to-column
the haunch forces causing shear in the beam and column
connection, V bW, is:
are opposite those from the Uniform Force Method for the
typical direction of moment, member shear may not always VbW = Vg + VBMF + VbTot1,2 + VbTot1,1 (80)
be a governing consideration. Thus, selection of the min-
imum permissible value of the virtual dimension, r, may The forces in the beam flange are given in Equation  61.
not be the most economical choice in all cases; the gusset These web and flange forces are shared with the gusset
thickness could be optimized in some cases by selecting plate, and the designer should avoid connecting the beam
a higher value of r. The authors have not developed equa- to column in a manner that is not compatible with the con-
tions to solve for this optimum value; trial and error may be nection of the gusset to column such as a welded gusset
used. Note that if the design also considers a case without connection and bolted beam-web connection.
the moment, that latter case typically is governed by beam
shear and governs the required gusset thickness. DESIGN OF WELDS

Minimum Actual Gusset Dimensions Considering At a minimum, the welds connecting the gusset plates must
Haunch and Bypass Forces be capable of transferring the forces at the joint determined
for the design of the gusset. In some cases, such welds
The minimum gusset dimension, 2β, is governed by the may be insufficient to develop the full strength of the gus-
shear in the column at the gusset mid-height; β is defined set plate. As the design methods presented here utilize the
by combining Equations 16, 31, 68, and 70: lower-bound theorem, the ductility of such weak-weld/
strong-gusset joints must be demonstrated. It is important

238 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

 Table 1.  Summary of Equations
Design Quantity UFM Method Bypass Method Haunch Method
λ — (52) or (55) —
VefCol (17)
Required VefC (35) (35) [λP for P] or (55)
VefBm, VefConn (18), (19)
UC (20)
UCVefB [below; above] (24), (25)
Required UCVefB (42) (42) [λP for P]
Required VefB (43) (43) [λP for P]
rminCol (26) (53) (75)
rminBm (28) (54) (76)
r (29)
α, β (10), (11)
Vc, Hc, Vb, Hb (2), (4), (5), (6)
HcBP, VcBP — (48), (49) —
HcH, HbH — — (62), (63)
VcH, VbH — — (64)
HcTot, VcTot — (68), (69)
HbTot, VbTot — (72), (73)
Required β (33) (77)
Required α (40) (78) or (56) (78)
Mc (30)
McBP — (50) —
McH — — (65)
McTot — (70)
Mb (36)
MbH — — (66)
MbTot — (74)
Vmid (column, beam) (31), (37)
tg Maximum of (44) and (45)

to consider that the redistributions required to engage other SUMMARY OF EQUATIONS

elements necessitates that separate joints in the connection
The preceding discussion includes many intermediate
deform together. For example, the limitation of the gusset-
equations necessary for derivation of the design equations.
to-beam forces in the bypass method may assign more force
Additionally, some equations have multiple forms: simpler
to the beam-to-column connection, and the deformation
equations for the UFM and more terms with the introduc-
demands on the two weld groups are not independent. In the
tion of bypass and haunch forces. Table 1 presents the equa-
absence of a nonlinear or mechanism analysis demonstrat-
tion numbers to be used for the determination of design
ing that weld deformations are within their rupture limit,
quantities for the UFM, bypass, and haunch methods. (The
design of the weld for the gusset plate strength ensures the
haunch method equations include bypass forces.) Certain
ductility of the joint by favoring inelastic deformation of the
equations developed for the UFM are modified by substitut-
gusset or the web of the member over rupture of the weld.
ing λP for P; this is indicated in the table.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 239

 Table 2.  Brace Forces
P Pcosθ Psinθ
(kips) (kips) (kips)
P1,1 550 352 423
P1,2 460 295 353

DESIGN EXAMPLES
The connection shown in Figure 11 will be designed using the methods developed in this study.
The design is performed three times:
• Preliminary design (wind loads). The gusset is designed using UFM method without bypass forces.
• Redesign with bypass forces (wind loads). The gusset is redesigned using UFM and bypass to eliminate web reinforcement.
• Seismic design including haunch forces (and bypass forces).
The design forces are presented in Table 2. To facilitate subsequent calculations, the horizontal and vertical components of the
brace forces are determined and presented in the table. (P denotes the axial force in the brace.) The angle from vertical, θ, is
50.2°.
To facilitate comparison, the same brace axial forces are used for both the wind-load and seismic-load designs.
The beam and column forces delivered to the connection are shown in Table 3.
The seismic forces correspond to the formation of plastic hinges at each end of the beam, represented by the symbol Ecl. The
plastic-hinge moment is the maximum flexure that the beam can deliver and is thus not combined with the gravity moment,
although the corresponding shear is additive to the gravity shear. The plastic-hinge moment is 1.1 times the beam expected

Fig. 11.  Design example connection.

240 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

 Table 3.  Beam and Column Forces
Load Case or Combination Load Force
Vg 10 kips
Gravity 1.2D + L
Mf 100 kip-in.
VBMF 7.1 kips
Wind 1.0W Mf not used
Vcol 17.1 kips
VBMF 44.9 kips
Seismic Ecl Mf 7770 kip-in.
Vcol 108 kips
Vg + VBMF 17.1 kips
Gravity + wind 1.2D + L + 1.0W
Mf not used
1.2D + L + 1.0Ecl Vg + VBMF 54.9 kips
Gravity + seismic
0D + 0L+ Ecl Mf 7770 kip-in.

Table 4.  Member Sizes and Grades

Member Size Grade
Column W14×120 50 ksi
Beam W18×55 50 ksi
Gussets — 50 ksi

beam flexural strength, as required for the design of Special Concentrically Braced Frames and Buckling-Restrained Braced
Frames in the AISC Seismic Provisions (AISC, 2016) Sections F2.6b(b) and F4.6b(b), respectively.
The member sizes and materials are shown in Table 4. Members are ASTM A992 material, and plate is ASTM A572 material.
The connection strength is based on a 15-in.-deep portion of the beam complete-joint-penetration (CJP) welded to the column
flange:
ϕRn ≥ 176 kips
Based on the brace-to-gusset connection (not shown in this example), minimum dimensions for the gusset plate are 23  in.
wide and 17 in. high. The dimensions used for calculation allow for an extra inch for weld termination. The minimum gusset
interface centroids are:
β ≥ 8 in.
α ≥ 11 in.

1. Preliminary Design: Design of Top Gusset (with No Bypass or Haunch Force)

The following example shows the design of the top gusset to resist the force P1,2. The design considers the effect of the force
P1,1 on the beam, but otherwise the two gusset designs are independent of each other.
For preliminary design, the gusset will be proportioned to resist brace forces only, assuming the beam moment is transferred
to the column by the flanges (i.e., ρ = 1.0). Similarly, the bypass-force method will not be used. Both of these methods are
addressed in a gusset redesign.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 241

Column Effective Shear Strength
The effective column shear strength is reduced considering the column shear.
Vcol = 17.1 kips
VefCol = ϕVn − Vcol (17)
= 257 kips − 17.1 kips
= 240 kips 

Using Equation 35, the minimum shear strength is:

VefCol = VefC (35)
ec cos θP
≥
3−
β + eb
2
( 7.25 in.) 0.640 ( 460 kips )
=
3
(8.0 in.) + ( 9.05 in.)
2
= 101 kips 

Beam Effective Shear Strength

The effective beam shear strength is limited by the strength of the connection and is reduced considering the beam shear:
VefBm = ϕVn − Vg − VBMF (18)
= 212 kips − 10.0 kips − 7.1 kips
= 195 kips 
VefConn = ϕRn − Vg − VBMF (19)
= 176 kips − 10.0 kips − 7.1 kips
= 158 kips 
VefConn
UC = (20)
VefBm
158 kips
=
195 kips
= 0.814 

The apportioned effective beam shear strength that can be utilized for each gusset can be apportioned considering the vertical
components of the brace forces:
P1,1 cos θ
UCVefB1,1 = UC VefBm (from Eq. 24)
P1,1 cos θ + P1,2 cos θ
352 kips
= 0.814 (195 kips )
352 kips + 295 kips
= 86.2 kips 
P1,2 cos θ
UC VefB1,2 = UCVefBm (from Eq. 25)
P1,1 cos θ + P1,2 cos θ
295 kips
= 0.814 (195 kips )
352 kips + 295 kips
= 72.1 kips 

242 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

Using Equation 42, the minimum apportioned effective connection shear strength is:
eb sin θP (from Eq. 42)
UCVefB1,2 ≥
3
α + ec
2
( 9.05 in.)( 0.768 ) (460 kips )
=
3
(11.0 in.) + ( 7.25 in.)
2
= 135 kips > 72.1 kips 
Reinforcement is therefore required.
The beam shear strength is apportioned:
P1,2 cos
VefB1,2 = VefBm (25)
P1,1 cos θ + P1,2 cos θ
= (195 kips) ( 0.455)
= 88.7 kips 
The minimum apportioned effective beam shear strength is:
eb P sin θ
VefB ≥ (43)
U
⎛
α 1+ C ⎞ + UC ec
⎝ 2 ⎠
( 9.05 in.) (460 kips) ( 0.768 )
=
0.814 ⎞
(11.0 in.) ⎛⎝1+ + 0.814 ( 7.25 in.)
2 ⎠
= 150 kips > 88.7 kips 
Because VefB > VefB1,2, reinforcement is required.
If the proposed gusset length is to be used, both the connection strength and the beam strength are insufficient. Based on the
degree of this deficiency, it is anticipated that an unreinforced beam will require an excessively long gusset. (In fact, calcula-
tions were performed, and the required horizontal length exceeds 42 in.) Instead, the web will be reinforced with a a-in. dou-
bler, 15 in. deep (matching the available web depth of the connection) and extending horizontally to match the gusset length.
The reinforced strength for both the connection and the beam is determined and apportioned as before. The additional strength
is:
ϕRn ≥ ϕ0.6Fy dt
= (1.0 ) ( 0.6 ) ( 50 ksi ) (15 in.) ( a in.)
= 169 kips

The beam strength is:

ϕVn = 212 kips + 169 kips
= 381 kips

The connection strength is:

ϕVn = 176 kips + 169 kips
= 344 kips

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 243

The effective strengths are recalculated:
VefBm = ϕVn − Vg − VBMF (18)
= 381 kips − 10.0 kips − 7.1 kips
= 363 kips 
VefConn = ϕRn − Vg − VBMF (19)
= 344 kips − 10.0 kips − 7.1 kips
= 327 kips 
With this reinforcement, the value of Uc is:
VefConn
UC = = 0.900 (20)
VefBm 
UCVefB1,2 = 0.455 (327 kips )
= 149 kips > 135 kips o.k.

With this additional beam web reinforcement, the effective beam and effective connection capacities will be increased.
VefB1,2 = 0.455 (363 kips)
= 166 kips > 150 kips o.k.

Minimum Virtual Gusset Dimensions

The virtual gusset dimensions are determined based on the effective column and beam shear strengths calculated earlier.
ec P
rminCol = (26)
VefC
( 7.25 in ) ( 460 kips)
=
240 kips
= 13.9 in. 
eb P
rminBm = (from Eq. 28)
UC VefB1,2
( 9.05 in ) ( 460 kips)
=
0.900 (166 kips)
= 27.9 in. 
r ≥ max (rminCol ,rminBm ) (29)
≥ max (13.9 in.,27.9 in.)
= 27.9 in. 
This value will be used for subsequent calculations. The corresponding virtual dimensions are:
α = r sin θ − ec (10)
= (27.9 in.) ( 0.768 ) − 7.25 in.
= 14.2 in. 
β = r cos θ − eb (11)
= ( 27.9 in.) ( 0.640 ) − 9.05 in.
= 8.80 in. 

244 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

Gusset Forces and Member Shear Checks (at Connection)
The virtual gusset dimensions are used to determine the forces acting at the centroids of the virtual gusset interfaces with the
beam and the column:
β
Vc = P (2)
r
⎛ 8.80 in.⎞
= ( 460 kips )
⎝ 27.9 in.⎠
= 146 kips 
ec
Hc = P (4)
r
⎛ 7.25 in.⎞
= (460 kips )
⎝ 27.9 in.⎠
= 119 kips < VefC o.k.
eb
Vb = P (5)
r
⎛ 9.05 in.⎞
= (460 kips)
⎝ 27.9 in.⎠
= 149 kips < UCVefB1,2 o.k.
α
Hb = P (6)
r
⎛ 14.2 in.⎞
= ( 460 kips )
⎝ 27.9 in.⎠
= 234 kips 

Minimum Actual Gusset Dimensions and Member Shear Checks (at Mid-Length of Gusset)
The value of rminBm controls; therefore, r = rminBm and r > rminCol. The dimensions for α and β are calculated as follows:
β
β≥ (33)
VefC 1
+
Hc 2
8.80 in.
=
240 kips 1
+
119 kips 2
= 3.50 in. 
α
α≥ (40)
1 1
+
UC 2
14.2 in.
=
1 1
+
0.900 2
= 8.81 in. 

The following values will be used:

β = 8 in.
α = 11 in.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 245

Both of these values exceed two-thirds of the corresponding virtual dimensions:
β 8 in.
=
β 8.80 in.
= 0.909 > q
α 11 in.
=
α 14.2 in.
= 0.775 > q

As such, the maximum member shear will be at sections C1 and B1 rather than C2 and B2. For completeness, the evaluation
is shown next.
The moments due to the eccentricities between the actual dimensions and the virtual dimensions are:
Mc = H c ( β − β ) (30)
= 119 kips (8.80 in. − 8 in.)
= 100 kip-in. 
Mb = Vb ( α − α ) (36)
= 149 kips (14.2 in. − 11 in.)
= 479 kip-in. 
Check column shear:
Hc M c
Vmid = + (31)
2 β
119 kips 100 kip-in.
= +
2 8 in.
= 72.8 kips < VefCol o.k.

Check beam shear:

Vb M b
Vmid = + (37)
2 α
149 kips 479 kip-in.
= +
2 11 in.
= 118 kips < VefBm o.k.

Gusset Thickness
The gusset thickness is selected considering the combined forces on the gusset-to-column interface and the gusset-to-beam
interface. Considering the column side, the minimum thickness is:
2
1 ⎛ Vc ⎞ 2 ⎛ 2 Mc ⎞
tg ≥ ⎜ ⎟ + ⎜ Hc + ⎟ (44)
ϕFy 2β ⎝ 0.6 ⎠ ⎝ β ⎠
2 2
1 ⎛ 146 kips ⎞ ⎛ 2 100 kip-in. ⎞
= + 119 kips +
0.90 ( 50 ksi ) 2 (8 in.) ⎝ 0.6 ⎠ ⎝ 8 in. ⎠
= 0.390 in. 

246 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

Considering the beam side, the minimum thickness is:
2 2
1 ⎛ Hb ⎞ ⎛ 2 Mb ⎞
tg ≥ + V + (45)
ϕFy 2α ⎝ 0.6⎠ ⎝ b α ⎠
2 2
1 ⎛ 234 kips ⎞ ⎛ 2 479 kip-in. ⎞
= + 149 kips +
0.90 ( 50 ksi ) 2 (11 in.) ⎝ 0.6 ⎠ ⎝ 11 in. ⎠
= 0.460 in. 
A 2- in.-thick gusset is required, with overall dimensions of 23 in. wide by 17 in. high (providing extra length for weld termina-
tion). The gusset design is shown in Figure 12.

2. Redesign with Bypass Forces

The connection in this example is redesigned using the bypass method in order to avoid the need for the beam-web doubler.
Values for VefB and Uc correspond to the unreinforced condition.
The transfer factor λ is obtained from Equation 52:
VefB ⎡ ⎛ UC ⎞ ⎤
λ≤ ⎢α 1+ + UC ec⎥ (52)
eb sin θP ⎣ ⎝ 2 ⎠ ⎦
88.6 kips ⎡ ⎛ 0.814 ⎞ ⎤
= ⎢(11 in.) ⎝1+ + 0.814 ( 7.25 in.)⎥
( 9.05 in.) 0.768 ( 460 kips) ⎣ 2 ⎠ ⎦
= 0.592 
Bypass forces are calculated:
HcBP = (1 − λ ) P sin θ (48)
= (1 − 0.592 ) ( 460 kips ) 0.768
= 144 kips 

Fig. 12.  Preliminary gusset design.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 247

 HcBP
VcBP = (49)
tan θ
144 kips
=
1.20
= 120 kips 

Minimum Virtual Gusset Dimensions

The virtual gusset dimensions are determined:
λ Pec
rminCol = (53)
VefC − HcBP
( 0.592 ) ( 460 kips) ( 7.25 in.)
=
(240 kips ) − (144 kips )
= 20.7 in. 
λ Peb
rminBm = (54)
UCVefB
( 0.592 ) (460 kips )( 9.05 in.)
=
0.814 (88.6 kips )
= 34.2 in. 
r ≥ max (rminCol ,rminBm ) (29)
= max ( 20.7 in.,34.2 in.)
= 34.2 in. 
This value will be used for subsequent calculations. The corresponding virtual dimensions
α = r sin θ − ec (10)
= ( 34.2 in.) ( 0.768 ) − 7.25 in,
= 19.0 in. 
β = r cos θ − eb (11)
= ( 34.2 in.) ( 0.640 ) − 9.05 in.
= 12.8 in. 

Gusset Forces and Member Shear Checks (at Connection)

The virtual gusset dimensions are used to determine the forces acting at the centroids of the virtual gusset interfaces with the
beam and the column:
β
λVc = λP (from Eq. 2)
r
⎛ 12.8 in.⎞
= ( 0.592 ) ( 460 kips)
⎝ 34.2 in.⎠
= 102 kips 

248 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

 ec
λ Hc = λP (from Eq. 4)
r
⎛ 7.25 in.⎞
= ( 0.592 )(460 kips )
⎝ 34.2 in.⎠
= 57.8 kips 
eb
λVb = λP (from Eq. 5)
r
⎛ 9.05 in.⎞
= ( 0.592 )( 460 kips)
⎝ 34.2 in.⎠
= 72.1 kips 
α
λ Hb = λP (from Eq. 6)
r
⎛ 19.0 in. ⎞
= ( 0.592 ) (460 kips)
⎝ 34.2 in.⎠
= 151 kips 

Determination of Total Forces and Member Shear Checks (at Connection)

The total vertical and horizontal forces are used in the determination of minimum required gusset dimensions. (Moments can-
not be computed until the actual gusset dimensions are set.) The forces on the column edge of the gusset are:
HcTot = λ Hc + HcBP (from Eq. 68)
= 57.8 kips + 144 kips
= 202 kips < VefC o.k. 
VcTot = λVc + VcBP (from Eq. 69)
= 102 kips + 120 kips
= 222 kips 
The forces on the beam edge of the gusset are:
H bTot = λ Hb (from Eq. 72)
= 151 kips
VbTot = λVb (from Eq. 73)
= 72.1 kips < UCVefB o.k. 

Determination of Minimum Actual Gusset Dimensions

⎛ α HcBP ⎞ 1
β ≥ β− (77)
⎝ tan θ HcTot ⎠ VefC 1
+
HcTot 2
⎛ 19.0 in. ⎞ ⎛ 144 kips ⎞
12.8 in. −
⎝ 1.20 ⎠ ⎝ 202 kips⎠
=
240 kips 1
+
202 kips 2
= 0.900 in. 

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 249

 α
α≥ (78)
VefB 1
+
VbTot 2
19.0 in.
=
88.6 kips 1
+
72.1 kips 2
= 11.0 in. 
The following values will be used:
β = 8 in.
α = 11 in.

Determination of Moments at Interfaces and Member Shear Checks (at Mid-Length of Gusset)
The moments on the column edge of the gusset are:
λ Mc = λ Hc (β − β ) (from Eq. 30)
= ( 57.8 kips ) (12.8 in. − 8 in.)
= 279 kip-in. 
M cBP = HcBP ( eb + β ) − VcBP ec (50)
= (144 kips ) ( 9.05 in. + 8 in.) − (120 kips )( 7.25 in.)
= 1,590 kip-in. 
McTot = λ Mc − McBP (from Eq. 70)
= 279 kip-in. − 1,590 kip-in.
= −1,310 kip-in. 
The moments on the beam edge of the gusset are:
MbTot = λ Mb (from Eq. 74)
= λVb ( α − α ) (from Eq. 36)
= ( 72.1 kips ) (19.0 in. − 11 in.)
= 577 kip-in. 
Check column shear:
HcTot McTot
Vmid = + (from Eq. 31)
2 β
202 kips −1,310 kip-in.
= +
2 8 in.
= 62.6 kips < VefC o.k. 

Check beam shear:

VbTot MbTot
Vmid = + (from Eq. 37)
2 α
72.1 kips 577 kip-in.
= +
2 11 in.
= 88.5 kips < VefB o.k. 

250 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

Note that Equations 31 and 37 are modified to utilize total forces (HcTot, etc.).

Determination of Required Gusset Thickness

Considering the column side, the minimum thickness is:
2 2
1 ⎛ VcTot ⎞ ⎛ 2 M cTot ⎞
tg ≥ + HcTot + (from Eq. 44)
ϕFy 2β ⎝ 0.6 ⎠ ⎝ β ⎠
2 2
1 ⎛ 222 kips ⎞ ⎛ 2 −1,310 kip-in. ⎞
= + 202 kips +
0.90 ( 50 ksi ) 2 (8 in.) ⎝ 0.6 ⎠ ⎝ 8 in. ⎠
= 0.897 in. 
Considering the beam side, the minimum thickness is:
2 2
1 ⎛ H bTot ⎞ ⎛ 2 M bTot ⎞
tg ≥ + VbTot + (from Eq. 45)
ϕFy 2α ⎝ 0.6 ⎠ ⎝ α ⎠
2 2
1 ⎛151 kips ⎞ ⎛ 2 577 kip-in. ⎞
= + 72.11 kips +
0.90 ( 50 ksi ) 2 (11 in.) ⎝ 0.6 ⎠ ⎝ 11 in. ⎠
= 0.311 in. 
The minimum thickness is taken as the maximum of Equations 44 and 45. Thus, a plate 1 in. × 17 in. high × 23 in. wide can
be used. The web doubler in the beam is eliminated (as are the CJP welds of the doubler and beam web to the column) at the
cost of a thicker gusset (and larger gusset-to-column welds). Alternatively, if a thinner gusset is desired, the dimension β could
be increased beyond the original target value. Fillet welds are based on the strength of the gusset plate. The gusset design is
shown in Figure 13.

3. Design Considering Haunch Force (with Bypass Method)

In this example, a design of the connection will be performed considering requirements for buckling restrained braced frames.
(As discussed earlier, inelastic demands on the gusset plate due to brace buckling, as would be expected for the seismic
response of other braced-frame systems, are not addressed in this design method.) The example considers seismic forces, but
the method illustrated is applicable to design for wind loads as well.
For seismic design, the beam moments and shears considered correspond to the flexural yielding of the beam because the
frame is expected to undergo large displacements. While the beam end moment could be resisted by the beam-to-column
connection (as it was for the previous examples for wind load), there can be significant economy in utilizing the gussets as
haunches and eliminating beam-flange-to-column-flange welds. In this example, the haunch method is used in conjunction
with the bypass method utilized in the previous example.
The designer should consider the range of possible beam-end moments that could coincide with maximum brace forces (and
vice versa), as the drift ranges for the beam and for the braces are likely to be substantially different. For this example, the range
of moments is based on the beam expected moment strength in one direction (positive Mf) and zero in the opposite direction.
The authors do not intend to imply that this range is adequate for all (or even most) conditions. Determination of the appropriate
combinations of beam moment and brace axial forces for systems subject to inelastic drift requires further study and is outside
of the scope of this paper.
Values are presented in pairs in braces (“{}”), with the first value corresponding to the inclusion of forces (both shears and
moments) corresponding to moment-frame action (condition 1) and the second value without (condition 2). A separate final
check is performed using only the haunch forces from beam moment with brace forces taken as zero.
Given the difference in the drift that results in yield of the braces and that corresponding to yield of the beam, it is possible for
cyclic inelastic drift to result in a condition in which the value of beam moment is negative and the forces λHc and HcH (Eq. 68)
are additive. Consideration of such a condition is beyond the scope of this study.
The haunch force is based on an apportionment factor ρ = 0.0.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 251

 M f = {7770,0} kip-in.
M fi
MH = (1 − ρ) (59)
2
{7770, 0} kip-in.
= (1 − 0 )
2
= {3880,0} kip-in. 
Vcol = {108,0} kips

VBMF = {44.9, 0} kips

VefCol = ϕVn − Vcol (17)
= 257 kips − {108, 0} kips
= {149, 257} kips 
P1,2 cos θ
VefBm = ( ϕVn − Vg − VBMF) (from Eqs. 18, 25)
P1,1 cos θ + P1,2 cos θ
⎛ 295 kips ⎞
= ( 212 kips − 10.0 kips − {44.9, 0} kips )
⎝ 352 kips + 295 kips ⎠
= {71.4, 91.9} kips 
P1,2 cos θ
VefConn = (ϕRn − Vg − VBmF ) (from Eqs. 19, 25)
P1,1 cos θ + P1,2 cos θ
⎛ 295 kips ⎞
= (176 kips − 10.0 kips − {44.9, 0} kips )
⎝ 352 kips + 295 kips⎠
= {54.9, 91.9} kips 

Fig. 13.  Gusset design with bypass forces.

252 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

 VefConn
UC = (20)
VefBm
⎧ 57.5 kips 75.4 kips ⎫
=⎨ , ⎬
⎩ 74.0 kips 91.9 kips ⎭
= {0.769, 0.820} 
The apportionment factor λ is determined for condition 2 (Mf = 0) using the corresponding effective beam and connection
strength.
VefB ⎡ ⎛ UC ⎞ ⎤
λ≤ ⎢α ⎝1+ + ecUC⎥ (52)
eb sin θP ⎣ 2 ⎠ ⎦
( 91.9 kips ) ⎡ ⎛ 0.820 ⎞ ⎤
= (11 in.) 1+ + ( 0.820 ) ( 7.25 in.)⎥
( 9.05 in.)( 0.768 ) (460 kips ) ⎢⎣ ⎝ 2 ⎠ ⎦
= 0.617 
This apportionment factor based on condition 2 will be used for both conditions 1 and 2.
HcBP = (1 − λ ) P sin θ (48)
= (1 − 0.617 ) ( 460 kips) 0.768
= 135 kips 

Determination of Minimum Virtual Gusset Dimensions

The minimum required gusset dimension, r, is determined for each case:
⎡ Mf ⎤
rminCol 2 ⎡⎣(VefC − HcBP ) cos θ ⎤⎦ + rminCol ⎢(VefC − HcBP ) ( ec − eb ) + (1 − ρ) − λ Pec cos θ⎥ + [ λ Pec (eb − ec )] ≥ 0 (75)
⎣ 2 ⎦
rminCol ⎡⎣({163, 257} − 135) 0.640⎤⎦
2

⎡ {7765, 0} kip-in. ⎤
+ rminCol ⎢({163, 257} kips − 135 kips ) ( 7.25 in. − 9.05 in.) + − 0.617 ( 460 kips ) ( 7.25 in.) ( 0.640 )⎥
⎣ 2 ⎦
+ ⎣⎡0.617 (460 kips )( 7.25 in.) ( 9.05 in. − 7.25 in.)⎤⎦ ≥ 0

This is solved by the quadratic formula:
rminCol = {−1.50, 17.0} in.

⎡ e ⎤
rminBm 2 ⎡⎣ sin θUcVefBm ⎤⎦ + rminBm ⎡⎣ MH − ecUcVefBm − PFeb sin θ ⎤⎦ + ⎢λ Pebec − MH b ⎥ ≥ 0 (76)
⎣ cos θ ⎦
rminBm [ 0.768 {54.9, 91.9} kips ]
2

+ rminBm ⎡⎣{3880, 0} kip-in. − ( 7.25 in.) {54.9, 91.9} kips − 0.617 ( 460 kips ) ( 9.05 in.) 0.768⎤⎦
⎡ ⎛ 9.05 in. ⎞ ⎤
+ ⎢0.617 ( 460 kips ) ( 9.05 in.) ( 7.25 in.) − {3880, 0} kip-in. ≥0
⎣ ⎝ 0.640 ⎠ ⎥⎦

This is solved by the quadratic formula:
rminBm = {16.4, 34.1} in.

r ≥ max ( rminCol ,rminBm ) (29)

= {max ( −1.50, 16.4 ) ,max (17.0,34.1)} in.
= {16.4, 34.1} in. 

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 253

As discussed in the derivation of Equations 75 and 76, for cases in which the member shear is not a governing consideration,
selection of a higher value of the virtual dimension, r, could result in lower required gusset thickness. By trial and error, a value
of the virtual dimension r = 18.9 in. was found to optimize the required gusset thickness for condition 1. (The same gusset-edge
forces could also be obtained with λ = 1.0 and r = 35.2.) However, because this example also considers the zero-moment condi-
tion 2, no reduction results from such optimization, and such trial-and-error methods are generally not beneficial.
α = r sin θ − ec (10)
= {18.9, 34.1} in. ( 0.768 ) − 7.25 in.
= {7.20, 18.9} in. 
β = r cos θ − eb (11)
= {18.9, 34.1} in. ( 0.640 ) − 9.05 in.
= {3.00, 12.8} in. 

Determination of Forces at Interfaces

The minimum actual gusset dimensions β and α are functions of the UFM, bypass, and haunch forces:
β
λVc = λ P (from Eq. 2)
r
{3.00, 12.8} in.
= 0.617 ( 460 kips)
{18.9, 34.1} in.
= {45.5, 106} kips 
ec
λ Hc = λ P (from Eq. 4)
r
7.25 in.
= 0.617 (460 kips )
{18.9, 34.1} in.
= {109, 60.4} kips 
eb
λVb = λ P (from Eq. 5)
r
9.05 in.
= 0.617 (460 kips )
{18.9, 34.1} in.
= {136, 75.4} kips 
α
λ Hb = λ P (from Eq. 6)
r
{7.20, 18.9} in.
= 0.617 ( 460 kips )
{18.9, 34.1} in.
= {109, 158} kips 
HcBP
VcBP = (49)
tan θ
135 kips
=
1.20
= 113 kips 

254 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

 HcH = HbH (from Eqs. 62, 63)
MH
=
eb + β
{3880, 0} kip-in.
=
9.05 in. + {3.00, 12.8} in.
= {322, 0} kips 
VbH = VcH (64)
β
= HcH
α
{3.00, 12.8} in.
= {322, 0} kips
{7.20, 18.9} in.
= {134, 0} kips 

Determination of Total Forces and Member Shear Checks (at Connection)

The total vertical and horizontal forces are used in the determination of minimum required gusset dimensions. (Moments can-
not be computed until the actual gusset dimensions are set.) The forces on the column edge of the gusset are:
HcTot = λ Hc + HcBP − HcH (68)
= {109, 60.4} kips + 135 kips − {322, 0} kips
= {−77.1, 196} kips < VefC = VefCol o.k. 
VcTot = λVc + VcBP + VcH (69)
= {45.5, 106 } kips + 113 kips − {134, 0} kips
= {293, 219} kips 
The forces on the beam edge of the gusset are:
HbTot = λ Hb + HbH (72)
= {109, 158} kips + {322, 0} kips
= {431, 158} kips 
VbTot = λVb − VbH (73)
= {136, 75.4} kips − {134, 0} kips
= {1.90, 75.4} kips < UCVefB o.k.

Determination of Minimum Actual Gusset Dimensions

⎛ e ⎞
βHcTot − HcBP β + eb − c
⎝ tan θ ⎠
β≥ (77)
1
VefC + HcTot
2
⎛ ⎞⎛ ⎞ ⎛ 7.25 in.⎞
⎝{
3.00, 12.8} in. {−77.1, 196} kips − 135 kips {3.00, 12.8} in. + 9.05 in. −
⎠⎝ ⎠ ⎝ 1.20 ⎠
=
1
{149, 257} kips + {−77.1, 196} kips
2
= {5.60, 1.00} in. 

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 255

 α
α≥ (78)
VefB 1
+
VbTot 2
{7.20, 18.9} in.
=
{71.4, 91.9} kips 1
+
{1.90, 75.4} kips 2
= {0.200, 11.0} in. 
The following values will be used for the actual gusset:
β = 8 in.
α = 11 in.

It should be noted that condition 2 (which does not include the beam moment) is the governing condition. This is because the
beam shear is the governing consideration in selecting the gusset size, and beam shear induced by the haunch forces counter-
acts that due to the brace force. This highlights the need to consider the minimum value of the beam moment in design. It also
shows that in some cases the gussets can provide flexural resistance at essentially no cost.

Determination of Moments at Interfaces and Member Shear Checks (at Mid-Length of Gusset)
The moments on the column edge of the gusset are:
λ Mc = λ Hc ( β − β ) (from Eq. 30)
= {109, 60.4} kips ({3.00, 12.8} in. − 8 in.)
= {−543, 288} kip-in. 
M cBP = HcBP ( eb + β ) − VcBP ec (50)
=135 kips ( 7.25 in. + 8 in.) − 113 kips ( 7.25 in.)
= {1490, 1490} kip-in. 
McH = (β − β ) HcH (65)
= ({3.00, 12.8} in. − 8 in.) {32.6, 0} kips
= {−1600, 0} kip-in. 
McTot = λ Mc − McBP − McH (70)
= {−543, 288} kip-in. − {1490, 1490} kip-in. − {−1600, 0} kip-in.
= {− 433, − 1200} kip-in. 
The moments on the beam edge of the gusset are:
λ Mb = λVb (α − α ) (from Eq. 36)
= {136, 75.4} kips ({7.2, 18.9} in. − 11 in.)
= {−512, 597} kip-in. 
MbH = ( α − α )VbH (66)
= ({7.20, 18.9} in. − 11 in.) {134, 0} kips
= {− 505, 0} kip-in. 
MbTot = λ Mb − M bH (74)
= {−512, 597} kip-in. − {−505, 0} kip-in.
= { − 7.00, 597} kip-in. 

256 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

Check column shear:
HcTot McTot
Vmid = + (from Eq. 31)
2 β
{−77.1, 196} kips {− 433, − 1200} kip-in.
= +
2 8 in.
= {− 92.6, − 52.4} kips < VefC o.k. 
Check beam shear:
VbTot M bTot
Vmid = + (from Eq. 37)
2 α
{1.90, 75.4} kips {− 7.00, 597} kip-in.
= +
2 11 in.
= {0.300, 91.9} kips < VefB o.k. 

Determination of Required Gusset Thickness

Considering the column side, the minimum thickness is:
2 2
1 ⎛ VcTot ⎞ ⎛ 2 McTot ⎞
tg ≥ + H cTot + (from Eq. 44)
ϕFy 2β ⎝ 0.6 ⎠ ⎝ β ⎠
2
2 {− 433, − 1200} kip-in. ⎞
2
1 ⎛ {293, 219} kips ⎞ ⎛
= + {−77.1, 196} kips +
0.90 ( 50 ksi ) 2 (8 in.) ⎝ 0.6 ⎠ ⎝ 8 in. ⎠
= {0.725, 0.856} in. 
Considering the beam side, the minimum thickness is:
2 2
1 ⎛ H bTot ⎞ ⎛ 2 M bTot ⎞
tg ≥ + VbTot + (from Eq. 45)
ϕFy 2α ⎝ 0.6 ⎠ ⎝ α ⎠
2
1 ⎛ {431, 158} kips ⎞ 2 ⎛ 2 {− 7.00, 597} kip-in. ⎞
= ⎜ ⎟ + ⎜ {1.90, 75.4} kips + ⎟
0.90 ( 50 ksi ) 2 (11 in.) ⎝ 0.6 ⎠ ⎝ 11 in. ⎠
= {0.725, 0.324} in. 
The minimum thickness is taken as the maximum of Equations 44 and 45.

Evaluation of Connection for Moment Only

As discussed earlier, the connection should have sufficient strength to resist the required moment without the offsetting effects
of the brace forces.
The effective column shear strength in this case is increased due to the column shear (similar to column shear in a moment
frame being in the opposite direction from the beam flange force and thus reducing the panel-zone shear demand).
VefCol = ϕVn + Vcol
= 257 kips + 108 kips
= 365 kips

Similarly, the shear in the beam due to moment-frame behavior, VBMF, is in the opposite direction from the shear induced by
the haunch force, effectively increasing the effective beam shear:

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 257

 P1,2 cos θ
VefBm = ( ϕVn − Vg + VBMF ) (from Eqs. 18, 25)
P1,1 cos θ + P1,2 cos θ
= ( 212 kips − 10.0 kips + 44.9 kips) ( 0.455)
= 112 kips 
The forces on the column edge of the gusset are:
HcTot = HcH (from Eq. 68)
= 322 kips < VefCol o.k.
VcTot = VcH (from Eq. 69)
= 134 kips

The forces on the beam edge of the gusset are:

H bTot = H bH (from Eq. 72)
= 322 kips
VbTot = VbH (from Eq. 73)
= 134 kips

Because V bTot > UCVefB, reinforcement is required.

The moments on the column edge of the gusset are:
McTot = McH (from Eq. 70)
= 1,600 kip-in.

The moments on the beam edge of the gusset are:

MbTot = MbH (from Eq. 74)
= 505 kip-in.

Check column shear:

HcTot M cTot
Vmid = + (from Eq. 31)
2 β
322 kips 1,600 kip-in.
= +
2 8 in.
= 361 kips < VefC o.k. 

Check beam shear:

VbTot M bTot
Vmid = + (from Eq. 37)
2 α
134 kips 505 kip-in.
= +
2 11 in.
= 113 kips ≈ VefB o.k. 

Check gusset on column edge:

258 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

 2 2
1 ⎛ VcTot ⎞ ⎛ 2 M cTot ⎞
tg ≥ + H cTot + (from Eq. 44)
ϕFy 2β ⎝ 0.6 ⎠ ⎝ β ⎠
2 2
1 ⎛ 134 kips ⎞ ⎛ 2 1,600 kip-in. ⎞
= + 322 kips +
0.90 ( 50 ksi ) 2 (8 in.) ⎝ 0.6 ⎠ ⎝ 8 in. ⎠
= 1.05 in. 
Considering the beam side, the minimum thickness is:
2 2
1 ⎛ H bTot ⎞ ⎛ 2 M bTot ⎞
tg ≥ + VbTot + (from Eq. 45)
ϕFy 2α ⎝ 0.6 ⎠ ⎝ α ⎠
2 2
1 ⎛ 322 kips ⎞ ⎛ 2 505 kip-in. ⎞
= + 134 kips +
0.90 ( 50 ksi ) 2 (11 in.) ⎝ 0.6 ⎠ ⎝ 11 in. ⎠
= 0.588 in. 
The use of the gusset plates as haunches, in the absence of the offsetting brace forces, requires increasing the gusset thickness
and providing modest beam reinforcement. Alternatively, the gusset could be enlarged. For wind loading, it may be possible to
rely on the offsetting effects of the beam moment and brace forces to be in phase; for seismic loading consideration of out-of-
phase behavior is prudent.
The design is shown in Figure 14. Note that beam-flange-to-column-flange welds are not required due to the entire beam-end
moment being resisted by the gussets. As with the previous example, fillet welds are based on the strength of the gusset plate.

Fig. 14.  Gusset design considering haunch forces.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 259

 VALIDATION proportion of 2:1. Brace axial post-elastic stiffness in the
model is tuned such that the brace axial forces used in
In order to demonstrate that the design methods presented
design from Table 2 develop at a drift of 2.46%, which is the
in this study produce adequate designs, finite element anal-
drift that results in the design moment from beam plastic
yses were conducted on two models representing the design
hinging from Table 3. Thus, brace and beam forces trans-
in Example 3. The analyses corresponded to both the con-
mitted to the connection area closely match those used in
dition with the full expected beam moment (condition 1)
Example 3.
and the condition with zero moment (condition 2), which
For condition 2, no lateral drift is imposed, and the analy-
governed the design.
sis is force controlled. The design forces (brace axial forces
The two analysis models were constructed using Abaqus
from Table 2 and beam shear from Table 3) are imposed on
software. Flanges, webs, and gussets were meshed with
a constrained section of the beam at the connection bound-
shell elements in the connection region. Columns were
ary. Figure 15 shows diagrams of the models for conditions
meshed from the connection region to the column mid-
1 and 2.
height. Beams were meshed from the connection region to
Figure  16 shows the von Mises stresses and equivalent
the beam quarter-length for condition 1. Frame members
plastic strains in the connection region, with gusset inter-
were used for columns and beams outside these regions.
face forces determined by stress integration. The analysis
The models use an expected material strength of 55 ksi for
for condition 1 shows the formation of a beam plastic hinge
all elements, moderately higher than the design strength
close to the edge of the gusset. With the exception of the
employed in the design model (ϕFy  = 45 ksi). Consistent
plastic hinge region, the stress in the beam web is low, con-
with the design, beam flanges are not connected to the
sistent with the low forces indicated at section B2  in the
column flange and are separated by a small gap. A web-
design calculations (Equation 37, which indicates 1.2 kips
connection plate is joined to the beam web with a vertical
of beam shear from the top gusset).
line of tie elements and two 2-in. returns.
The analysis confirms that the connection can resist the
The analysis for condition 1 is displacement controlled,
design forces, including the beam plastic-hinge moment.
with the displacements at the two levels constrained to a

Fig. 15.  Diagrams of finite element models for condition 1 (a) and condition 2 (b).

260 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

This supports the validity of the haunch method, which uti- stresses so calculated also correspond roughly. Neverthe-
lizes the gussets to transmit the beam moment to the col- less, the difference is large enough in the case of the ver-
umn (rather than beam flange welds). As the plastic strain tical force to the beam to warrant caution in reliance on
is limited to the beam plastic hinge, the analysis confirms small values obtained from the difference of large forces,
that the design resistance can be achieved without excessive and the peak design stress is below the peak stress shown
ductility demand in any part of the gusset connection. in Figure  16, suggesting that providing for the full gusset
While an accurate design model is not required by the yield strength is prudent. Note that the edge forces from
lower-bound theorem, the forces at the interfaces are rea- the finite element analysis are computed in the deformed
sonably close to those in the design example: the difference condition, and thus are slightly out of alignment with the
between the design and analysis model values is small com- design calculations.
pared to the vertical or horizontal component of the brace It should be noted, however, that the values determined
force, as shown in Tables 5 and 6. by the haunch design method are moderately influenced
Table  5 shows that the apportionment of vertical and by some of the simplifying assumptions. First, the value of
horizontal forces between the beam and column is roughly the virtual dimension, r, was taken as the optimum value
consistent between the design and the analysis. Table  6 using trial and error; the method allows for other values
includes normal, shear, and flexural force at the gusset to be used. Second, the haunch forces are set to the same
edge at the column and at the gusset edge at the beam, for locations as the UFM forces, as shown in Figure 10. Third,
both the example calculation and from the finite element the apportionment factor, λ, employed for condition 1 is for
analysis. For both, stresses are calculated using the conven- convenience set equal to that for condition 2. Changes to
tional plastic method utilizing the plastic section modulus any of these assumptions would result in somewhat differ-
and computing a vector sum of normal and shear stress. ent design forces, and possibly greater difference between
(Stresses listed for the finite element analysis are computed calculated and analyzed values.
from total forces on the gusset edges; the stresses are not In contrast to condition 1, the finite element analysis of
those directly reported from the analysis. These stresses condition 2 (Figure 17) shows high beam shear in the con-
are only presented as a means of quantifying the com- nection region with substantial yielding both in the beam
bined effect of shear, normal, and flexural force at the gus- web (section B2) and in the connection plate (approxi-
set edge.) The comparison shows that design gusset-edge mately at section B1). In the design method (with the virtual

(a) Von Mises stress contour (b) Equivalent plastic stain contour

Fig. 16.  Finite element analysis results of Example 3, condition 1.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 261

 Table 5.  Comparison of Analysis and Design Vertical and Horizontal Forces for Condition 1
HcTot HbTot Sum
Horizontal Forces (kips) (kips) (kips) % Column % Beam
Example 3 calculation −77.1 431 353 −22% 122%
Finite element analysis −60.2 410 350 −17% 117%
VcTot VbTot Sum
Vertical Forces (kips) (kips) (kips) % Column % Beam
Example 3 calculation 292 2.70 295 99% 1%
Finite element analysis 268 29.1 297 90% 10%

Table 6.  Comparison of Analysis and Design Gusset-Edge Forces for Condition 1

Normal Shear Resultant
HcTot VcTot McTot Stress Stress Stress Angle
Forces at Column (kips) (kips) (kip-in.) (ksi) (ksi) (ksi) (deg)
Example 3 calculation −77.1 292 −459 13.7 20.8 24.9 33.3
Finite element analysis −60.2 268 −576
Stress calculated from finite element analysis forces 14.6 19.1 24.1 37.3
Normal Shear Resultant
VbTot HbTot Mbot Stress Stress Stress Angle
Forces at Beam (kips) (kips) (kip-in.) (ksi) (ksi) (ksi) (deg)
Example 3 calculation 2.70 431 0.0 0.100 22.4 22.4 0.40
Finite element analysis 29.1 410 −285
Stress calculated from finite element analysis forces 4.20 21.3 21.7 11.2

(a)  Von Mises stress contour (b)  Equivalent plastic stain conto

Fig. 17.  Finite element analysis results of Example 3, condition 2.

262 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

 Table 7.  Comparison of Analysis and Design Vertical and Horizontal Forces for Condition 2
HcTot HbTot Sum % %
Horizontal Forces (kips) (kips) (kips) Column Beam
Example 3 calculation 196 158 354 55% 45%
Calculation with ϕFy = 55 ksi 159 194 353 45% 55%
Finite element analysis 163 190 353 46% 54%
VcTot VbTot Sum % %
Vertical Forces (kips) (kips) (kips) Column Beam
Example 3 calculation 219 75.4 294 74% 26%
Calculation with ϕFy = 55 ksi 201 93.1 294 68% 32%
Finite element analysis 209 85.0 294 71% 29%

Table 8.  Comparison of Analysis and Design Gusset-Edge Forces for Condition 2

Normal Shear Resultant
HcTot VcTot McTot Stress Stress Stress Angle
Forces at Column (kips) (kips) (kip-in.) (ksi) (ksi) (ksi) (deg)
Example 3 calculation 196 219 −1200 35.4 15.7 38.7 66.2
Calculation with ϕFy = 55 ksi 159 201 −583 21.8 14.4 26.1 56.6
Finite element analysis 140 181 −403
Stress calculated from finite element analysis forces 17.2 13.0 21.5 53.0
Normal Shear Resultant
VbTot HbTot Mbot Stress Stress Stress Angle
Forces at Beam (kips) (kips) (kip-in.) (ksi) (ksi) (ksi) (deg)
Example 3 calculation 75.4 158 597 9.6 8.2 12.6 49.4
Calculation with ϕFy = 55 ksi 93.1 194 731 11.7 10.1 15.5 49.4
Finite element analysis 107 203 1030
Stress calculated from finite element analysis forces 15.3 10.5 18.6 55.4

dimension, r, set equal to rminBm), section B1 is constrained and stiffness to preclude significant ductility demands. The
to have a demand-to-capacity ratio of 1.0. Section B2 (using analysis also confirms that condition 2 is more critical than
Equation  37) has a calculated demand of 91.9 kips, equal condition 1 for this design.
to the effective capacity of the beam (Equation 18). Thus, The comparison of design forces from Example 3 to the
the analysis and design are consistent on a fundamental analysis forces for condition 2 shows that the design model
point: As the beam yields at these sections, the column can (which utilizes a specified minimum yield stress and a resis-
still continue to resist higher forces. Of note is that while tance factor) requires a greater portion of the brace force to
the beam web and its connection to the column reach yield bypass the beam. The more accurate finite element analysis
stress, the plastic strains remain low due to the redistribu- shows a distribution closer to the Uniform Force method,
tion of the forces in the connection to the column, consis- with somewhat less force bypassing the beam than in the
tent with the bypass design method. calculations, as shown in Tables 7 and 8. This difference is
The analysis confirms that the connection can support due in part to the disparity between the stresses permitted
the design loads for condition 2. This supports the use of the in the design model and the greater expected strength used
bypass method. It should be noted that the Uniform Force in the analysis. For comparison, design calculations were
Method without the bypass modification would indicate performed using ϕFy = 55 ksi (which required a bypass fac-
that this connection was inadequate. The plastic strains in tor, λ, of 0.76, as compared to the value of 0.617 from the
the beam web are small, confirming that after beam-web design example, corresponding to ϕFy = 50 ksi). Forces cor-
yielding, the bypass mechanism provides sufficient strength responding to that analysis are also in Tables 7 and 8.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 263

 CONCLUSIONS AND RECOMMENDATIONS MBC Moment at column face resisted by beam-to-column
connection, kip-in.
This study provides design equations that can be used in
the design of gussets in braced frames to minimize the Mb Moment at gusset–beam interface (from Uniform
required shear strength of the beams and columns in order Force Method analysis), kip-in.
to avoid the need for reinforcement. The design method
MbH Moment at gusset–beam interface from haunch
allows engineers to select the proportion of the load that
force, kip-in.
will contribute to beam or column shear and to redistribute
the load between beam and column. Additionally, equa- MbTot Total moment at gusset–beam interface, kip-in.
tions are provided that allow the gussets to be used to resist
MBM Moment at a beam section aligned with the gusset
beam-connection moment, which potentially allows for a
edge face, kip-in.
very economical design by eliminating the need for beam-
flange-to-column-flange welds. Mc Moment at gusset–column interface (from Uniform
The authors recommend this modified implementation Force Method analysis), kip-in.
of the Uniform Force Method, especially for conditions in
McBP Moment at gusset–column interface from bypass
which web doublers would otherwise be required. The use of
force, kip-in.
the bypass method provides additional economy, but engi-
neers should be cognizant of the limited study performed McH Moment at gusset–column interface from haunch
to date. Without additional study, the authors do not recom- force, kip-in.
mend bypass factors λ less than 0.6. The use of the haunch
McTot Total moment at gusset–column interface, kip-in.
method (with or without the bypass method) similarly pro-
vides for greater economy for connections with beam fixity. Mf Moment at column face (from beam), kip-in.
Based on the limited study performed to date, the authors
MH Moment at column face resisted by gusset, kip-in.
recommend that the haunch method be implemented con-
sidering both the upper and lower bounds of beam moment N Normal force on beam or column flange, kips
(conditions 1 and 2). Further study is required to establish
NB Beam axial force delivered to the connection, kips
the lower bound of this moment, especially for systems with
cyclic inelastic drift, for which the brace forces may reverse NH Horizontal force at beam-to-column web connection
within the elastic drift range of the beam. due to haunch forces, kips
Pf Beam flange force at connection to column, kips
SYMBOLS
Pi,j Brace axial force for brace j connecting to gusset i;
Fy Material specified minimum yield stress, ksi sign conventions are per the figures, kips
Hb Horizontal force transferred to beam by gusset Rn Nominal strength (of beam-to-column connection),
(from Uniform Force Method analysis), kips kips
HbH Horizontal force transferred to beam due to haunch Ru Required strength, kips
force, kips
UC Ratio of connection effective strength, VefConn, to
HbW Horizontal force at beam-web-to-column beam effective strength, VefBm
connection, kips
Vb Vertical force transferred to beam by gusset (from
HbTot Total horizontal force transferred to beam, kips Uniform Force Method analysis), kips
Hc Horizontal force transferred to column by gusset VbH Vertical force transferred to beam by gusset to resist
(from Uniform Force Method analysis), kips beam moment, kips
HcBP Horizontal force transferred to column by gusset VBMF Beam shear due to moment-frame behavior, kips
from bypass force, kips
VbW Vertical force at beam-web-to-column connection,
HcH Horizontal force transferred to column by gusset to kips
resist beam moment, kips
Vg Beam shear due to gravity, kips
HcTot Total horizontal force transferred to column by
Vc Vertical force transferred to column by gusset (from
gusset, kips
Uniform Force Method analysis), kips

264 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

VcH Vertical force transferred to column by gusset to β Distance from beam flange to centroid of Uniform
resist beam moment, kips Force Method force acting on column face, in.
VCol Column shear due to moment-frame behavior, kips β Half of gusset vertical dimension (centroid of
gusset–column interface), in.
VcBP Vertical force transferred to column by gusset from
bypass force, kips ϕ Resistance factor
VcTot Total vertical force transferred to column by gusset, λ Brace force apportionment factor (between Uniform
kips Force Method and “bypass force”)
VefBm Effective beam shear strength, kips ρ Beam moment apportionment factor (between beam
flange and haunch force in gusset)
VefBi,j Apportioned effective beam shear strength that can
be used to resist the force for brace j connecting to θ Brace angle from vertical, deg
gusset i; sign conventions are per the figures, kips
VefCol Effective column shear strength, kips REFERENCES
AISC (2016), Seismic Provisions for Structural Steel Build-
VefCi,j Apportioned effective column shear strength that
ings, ANSI/AISC 341-16, American Institute of Steel
can be used to resist the force for brace j connecting
Construction, Chicago, Ill.
to gusset i; sign conventions are per the figures, kips
AISC (2017), Manual of Steel Construction, 15th Ed.,
VefConn Effective beam-to-column connection shear American Institute of Steel Construction, Chicago, Ill.
strength, kips
Euler, L. (translated by Hewlett, J.) [published 1765; 1822
Vmid Shear in beam or column at mid-length or mid- translation], “Of a New Method of Resolving Equa-
height of gusset, kips tions of the Fourth Degree,” Elements of Algebra, Long-
man, Hurst, Rees, Orme, and Company, London.
Vn Nominal member shear strength, kips
Fortney, P.J. and Thornton, W.A. (2015), “The Chevron
db Beam depth, in. Effect—Not an Isolated Problem,” Engineering Journal,
eb Eccentricity from beam flange to beam centerline, AISC, Vol. 52, No. 2, pp. 125–164.
equal to half the beam depth, in. Muir, L.S. and Thornton, W.A. (2014), Vertical Bracing
Connections: Analysis and Design, Design Guide 29,
ec Eccentricity from column flange to column
AISC, Chicago, Ill.
centerline, equal to half the column depth, in.
Richards, P., Miller, B., and Linford, J. (2018), Finite Ele-
ho Distance between beam flange centroids, in. ment Evaluation of the Chevron Effect in Braced Frames.
r Gusset centroid offset (dimension from workpoint Brigham Young University Report No. SSRP-2018/02.
to brace control point in Uniform Force Method), Sabelli, R. and Arber, L. (2017), “Design of Chevron Gusset
in. Plates,” 2017 SEAOC Convention Proceedings.
rminBm Minimum dimension r based on beam shear Sabelli, R., and Saxey, B., (2021), “Design for Local Mem-
yielding, in. ber Shear at Brace Connections: Full-Height and Chev-
ron Gussets,” Engineering Journal, AISC, Vol. 58, No. 1,
rminCol Minimum dimension r based on column shear pp. 45–78.
yielding, in.
Thornton, W.A. (1984). “Bracing Connections for Heavy
tg Gusset thickness, in. Construction,” Engineering Journal, AISC, Vol.  21,
No. 3, pp. 139–148.
α Distance from column face to centroid of Uniform
Force Method force acting on beam flange, in. Thornton, W.A. (1991), “On the Analysis and Design of
Bracing Connections,” Proceedings of the AISC National
α Half of gusset horizontal dimension (centroid of Steel Construction Conference, Washington, D.C.
gusset–beam interface), in.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 265

266 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021
Interaction Strength of Steel-Concrete Composite
Beam-Columns Including the Balance Point
Mark D. Denavit
Note: A version of this paper appears in the proceedings of the 2020 Structural Stability
Research Council (SSRC) Annual Stability Conference

ABSTRACT
The maximum bending moment capacity of steel-concrete composite column cross sections occurs with concurrently applied axial com-
pression. This is seen in the shape of the interaction diagram, where the bending moment capacity increases with increasing axial com-
pression before reaching the balance point. The size of this bulged region of the interaction diagram can be significant, especially for
concrete-dominant sections. However, it is often neglected in design because of two stability-related concerns. First, the simple transfor-
mations that are recommended to convert cross-section strength to member strength produce illogical results near the balance point, with
member strength exceeding cross-section strength. Second, research has shown that the stiffness reductions used in elastic analyses are
not sufficient for highly slender concrete-dominant composite members subjected to high bending moments. This work seeks to address
these issues through the development of more advanced transformations and stiffness reductions. These new recommendations will more
accurately capture the strength of composite members and allow for more efficient designs.

Keywords:  composite construction, interaction strength, balance point, stiffness reduction.

INTRODUCTION to as the AISC Specification, in the 1986 edition (AISC,

1986). From that time until major revisions were made in the
S teel-concrete composite frames are an effective alterna-
tive to structural steel or reinforced concrete frames for
use as the primary lateral-force-resisting system of build-
2005 edition (AISC, 2005), the axial and flexural strengths
of composite beam-columns were based on calculations
that determined an equivalent steel section. This approach
ing structures. However, they have not yet been as widely
had limitations in that it was not applicable to columns
adopted in United States practice as they have in other parts
with steel ratios below 4%, and it often underestimated
of the world, notably East Asia. There are several barri-
the contribution of the concrete, particularly for concrete-
ers to the broader use of composite structures. Sequencing
dominant composite beam-columns with low steel ratios
issues in construction, which can lead to complications such
(Griffis, 2005). The current beam-column strength inter-
as difficult coordination of trades, can be a barrier. On the
action provisions (AISC, 2016) are based more directly on
other hand, innovative composite construction methods that
mechanics principles. The cross-section strength may now
resolve the sequencing issues can be highly efficient and
be determined using one of several methods; the two most
can reduce construction time (Griffis, 1992; Traut-Todaro,
commonly used are the plastic stress distribution method,
2019). Current design provisions are another barrier to the
which is applicable to most common composite column
wider adoption of composite construction. Despite recent
cross sections, and the more general strain-compatibility
advances (e.g., Lai et al., 2015; Denavit et al., 2016; Bruneau
method, which is comparable to approaches often taken
et al., 2018), design provisions for composite frames are
to compute reinforced concrete section strength. The plas-
not yet as comprehensive as those for the more traditional
tic stress distribution method is the primary method for
systems, nor do they consistently reflect the advantages of
assessing steel-concrete composite columns in the AISC
composite framing.
Specification (AISC, 2016) and other standards worldwide
Composite columns were introduced to the AISC Speci-
(CEN, 2004; SAC, 2014). It is accurate over a wide range
fication for Structural Steel Buildings, hereafter referred
of materials, cross-sectional geometries, and loading con-
ditions, but the method does result in significant unconser-
vative error for some cases. Cases of unconservative error
include encased composite members, also known as steel-
Mark D. Denavit, Assistant Professor, The University of Tennessee, Knoxville,
Knoxville, Tenn. Email: [email protected]
reinforced concrete (SRC) members, with high steel ratio,
high steel yield stress, or both (Behnam and Denavit, 2020).
The strain-compatibility method is conservative in nearly
Paper No. 2020-11 all cases but can be overly conservative in many cases. The

ISSN 0013-8029 ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 267

revisions in the 2005 edition also included an expansion of 1.81% (note that, for simplicity, the fillets between the web
the range of applicability of the provisions to members with and the flange are neglected in this work). The reinforcing
steel ratios as low as 1%. ratio (i.e., the ratio of area of reinforcing steel to gross area
Using the plastic stress distribution method, pairs of axial of the cross section) for this cross section is ρsr = Asr/Ag =
compression and bending moment strength are computed 0.40%. The concrete compressive strength is fc′ = 8 ksi, the
based on assumed plastic neutral axis locations. Selecting steel yield stress is Fy = 50 ksi, and reinforcing steel yield
many possible locations for the plastic neutral axis results strength is Fyr = 60 ksi. The longitudinal reinforcing has a
in an essentially continuous curve for the interaction dia- cover of 1d in. from the edge of the concrete to the edge of
gram. For example, the interaction diagram for the SRC the bar.
cross section shown in Figure 1 for bending about the major The example cross section shown in Figure 1 was selected
axis of the steel shape is shown in Figure 2(a). This cross to have low steel ratios, near the lower limits given in the
section has outside dimensions of 28 in. × 28 in., a W10×49 AISC Specification (AISC, 2016) since neglecting the
wide-flange steel shape, and four #8 reinforcing steel bars. balance point reduces the available strength for concrete-
The steel ratio (i.e., the ratio of area of steel to gross area dominant members more than it does for steel-dominant
of the cross section) for this cross section is ρs  = As /Ag  = members. This cross section is used in example analyses
described throughout this paper.
While the plastic stress distribution method can be used
to compute a continuous cross-section interaction diagram
by selecting many different plastic neutral axis locations,
doing so is burdensome by hand or spreadsheet. A set of
closed-form equations (AISC, 2017) has been developed to
compute key points on the curve which can then be used to
construct a multilinear interaction diagram. The points are
labeled A, C, D, and B as shown in Figure 2(a). Point A rep-
resents the pure axial strength; point B represents the pure
bending strength; point C has the same bending moment as
point B, but with axial compression; point D represents the
balance point, the point of maximum bending moment.
Computing the cross-section interaction strength is rela-
Fig. 1.  Example SRC cross section.
tively straightforward; however, it is not directly used in

6000 6000
A A

5000 C 5000 C
Axial Compression (kips)

Axial Compression (kips)

4000 4000

3000 3000
D Aλ D
Cλ
2000 2000

1000 1000 Dλ

B B
0 0
0 500 1000 1500 2000 0 500 1000 1500 2000
Bending Moment (kip-ft) Bending Moment (kip-ft)
(a)  Continuous vs. discrete (ACDB) (b)  Cross-section strength vs. beam-column strength

Fig. 2.  Interaction strength diagrams for the example SRC cross section.

268 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

design. As described in the AISC Specification Commen- conservative error introduced by neglecting the balance
tary (AISC, 2016), two reductions are applied to the nomi- point. The second goal is to reduce the unconservative error
nal cross-section interaction strength to obtain the available observed for highly slender, concrete-dominant composite
beam-column interaction strength. members, which may be exacerbated by the inclusion of the
The first is a stability reduction, where a factor equal balance point. To accomplish these goals, an alternative
to the ratio of the nominal axial compression strength stability reduction for interaction diagrams and an alterna-
with and without length effects (χ  = Pn /Pno) is applied to tive stiffness reduction to be used with the direct analysis
the ordinate (i.e., axial compression) of each point on the method are evaluated.
interaction diagram, leaving the abscissa (i.e., bending
moment) unchanged. This method is logical in that it yields
ALTERNATIVE STABILITY REDUCTION
the proper results for pure axial compression (point A)
FOR INTERACTION DIAGRAMS
and for pure bending moment (point B), but illogical and
potentially unconservative results arise in the intermediate As described in the previous section, the method for com-
points, particularly the balance point (point D). The balance puting the interaction strength of steel-concrete composite
point is the point of maximum moment and it occurs for a columns, which is referred to as Simplified Method 2 in the
nonzero axial compression. When the stability reduction is AISC Specification Commentary (AISC, 2016), neglects
applied in this simple manner, the resulting beam-column the balance point (point D). While any number of points
interaction point D lies outside of the cross-section interac- on the cross-section interaction diagram can be computed,
tion diagram, as shown in Figure 2(b). Note that the beam- only three points, A, C, and B are utilized for the avail-
column strength interaction diagram shown in Figure 2(b) able strength of composite beam-columns. Interaction dia-
was constructed with χ = 0.5, which corresponds to L c/H = grams computed following these recommendations for the
18.6 for the example cross section. example SRC cross section and for a variety of effective
The second reduction is to apply the resistance factors. lengths are shown in Figure 3(a). The conservativeness of
The resistance factors for composite columns are defined neglecting point D can be seen by comparing the interac-
as ϕc = 0.75 for axial compression and ϕb = 0.90 for flexure tion diagrams in Figure  3(a) to the cross-section interac-
in AISC Specification Chapter I (AISC, 2016). For com- tion diagram shown in Figure 2(a). The example SRC cross
bined bending and axial load, the AISC Specification Com- section is concrete-dominant, so the moment strength at
mentary recommends that axial compression of each point point D is significantly greater than that at point B.
be multiplied by ϕc and the bending moment of each point Interaction diagrams using an alternative method of
be multiplied by ϕb. This simple procedure may be uncon- applying the stability reduction are shown in Figure  3(b).
servative because it can lead to strength reductions which In this alternative method, points A, C, and B are computed
imply resistance factors greater than 0.90 for the intermedi- and reduced as before (i.e., factoring the ordinate by χ  =
ate points (Denavit, 2017). Pn/Pno). Noting that factoring just the ordinate for point D
Furthermore, when evaluated against advanced second- gives the illogical result of a point on the beam-column
order inelastic analyses, current design provisions can interaction strength diagram outside of the cross-section
result in unconservative errors for highly slender, concrete- interaction strength diagram, both the ordinate and the
dominant composite members subject to low axial loads abscissa of point D are reduced. The ordinate of point D is
and high bending moments (Denavit et al., 2016). Con- reduced by the same factor as the other points. The abscissa
cerns resulting from the simple reductions and potential is reduced such that the reduced point D remains on the line
unconservative error have led to the recommendation in the between point B and the original point D, thus ensuring that
AISC Specification Commentary to neglect point D in the the beam-column interaction strength does not exceed the
strength interaction diagram and to only consider points A, cross-section interaction strength. A summary of the reduc-
C, and B (AISC, 2016). tion applied to each point is presented in Table 1.
Neglecting the balance point can be highly conserva- The interaction diagram including point D and con-
tive, especially for stocky concrete-dominant columns. structed using the alternative stability reduction (denoted
Improved methods of determining interaction strength of as the ACDB interaction) provides a plausible alternative
steel-concrete composite beam-columns would have the to the interaction diagram currently recommended in the
potential of unlocking large amounts of strength and allow- AISC Specification Commentary (AISC, 2016) (denoted
ing composite columns to fulfil more of their potential. as the ACB interaction). However, the new interaction dia-
This work explores potential alternative approaches for gram must be rigorously evaluated to ensure that it results
including the balance point within the interaction strength in safe designs.
of steel-concrete composite beam-columns. This work When evaluating design provisions for beam-column
has two complimentary goals. The first goal is to reduce interaction strength, simply comparing available strengths

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 269

 Table 1.  Points on the Interaction Diagram
Cross-Section Strength Beam-Column Strength
Point M P M P
A 0 PA 0 χPA
C MC PC MC χPC
Da MD PD (1 − χ)MB + χMD χPD
B MB 0 MB 0
a
  Point D is not included with the ACB interaction diagram.

computed per design equations to the results of physical approaches. This work expands upon the results presented
experiments or advanced inelastic analyses can be mislead- by Denavit et al. (2016). The approach taken is to compare,
ing. In practice, available strengths are evaluated against for many different individual cases, the maximum applied
required strengths and required strengths are computed loads permitted by the design methodology to the applied
following particular rules (e.g., specific type of analysis, loads at which failure occurs according to second-order
defined stiffness). The provisions for an entire method inelastic analyses.
of design, encompassing both the available and required
strengths, must be considered in the evaluation. Benchmark Frames
Many notable studies have been conducted in this way,
The cases investigated are small frames that consist of a
including for structural steel columns and the development
single composite column as shown in Figure 4. The same
of the interaction equations in use today (Kanchanalai,
broad range of cross-section and frame parameters investi-
1977), for reinforced concrete columns (Hage and Mac-
gated by Denavit et al. (2016) were used in this work.
Gregor, 1974), for the development of the direct analysis
Four categories of cross section were investigated:
method (Surovek-Maleck and White, 2004), and for the
(1) circular concrete-filled steel tubes (CCFT), (2) rectan-
extension of the direct analysis method to composite frames
gular concrete-filled steel tubes (RCFT), (3) SRC subjected
(Denavit et al., 2016). Each of these studies duly consid-
to major-axis bending, and (4) SRC subjected to minor-axis
ered both the calculation of available strength and required
bending. Within these groups, sections were selected to
strength in their evaluations, albeit using somewhat different
Axial Compression (kips)

Axial Compression (kips)

Bending Moment (kip-ft) Bending Moment (kip-ft)

(a)  Simplified method 2 (ACB) (b)  Proposed method (ACDB)

Fig. 3.  Interaction strength diagrams for beam-columns with the example SRC cross section.

270 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

span practical ranges of concrete strength; steel ratio; and, L, leaning column load ratios, γ, and end restraints (rota-
for the SRC sections, reinforcing ratio. Steel yield strengths tional spring stiffnesses, k θ,top and k θ,bot) were investigated.
were selected as Fy = 50 ksi for wide-flange shapes, Fy = Each cross section was run with each frame resulting in
42 ksi for round HSS shapes, Fy  = 46 ksi for rectangular 1,200  individual cases for each of the CCFT and RCFT
HSS shapes, and Fysr = 60 ksi for reinforcing bars. Three groups and 2,880  individual cases for each of the SRC
concrete strengths were selected: fc′ = 4, 8, and 16 ksi. Note groups. Full details of the selected benchmark frames are
that the AISC Specification (AISC, 2016) limits concrete reported by Denavit et al. (2016).
strength to a maximum of 10 ksi. Concrete exceeding
this limit was included to investigate an extreme case and Second-Order Inelastic Analysis
because the limits may be revised in future editions.
Geometric and material nonlinear analyses using fiber-
With the selected CFT sections, the full range of permit-
based beam finite elements were used to obtain results
ted steel ratios is examined, including those associated with
against which the design methodologies are benchmarked.
noncompact and slender sections. However, local buckling
These analyses represent the “best guess” of the true behav-
is neglected in this study, both by not modeling it in the
ior of the frames. The uniaxial constitutive relations defined
inelastic analyses and by not including the strength reduc-
within the fiber representations of the cross sections were
tions in the design strength calculations. The effects of local
calibrated specifically for composite columns. As noted
buckling on the interaction strength of filled composite
previously, local buckling of the steel tube and other steel
members can be captured in design through the use of the
components was neglected. Initial system and member geo-
effective stress-strain method defined in AISC Specifica-
metric imperfections were directly modeled. Full details of
tion Section I1.2 (AISC, 2016) and in analysis through the
the analyses, including validation against the results of hun-
use of beam elements with specialized constitutive relations
dreds of physical experiments are reported by Denavit et al.
or shell elements (Lai and Varma, 2016). Nonetheless, local
(2016) and Denavit and Hajjar (2014).
buckling remains a complicated issue that was excluded
A sample of analysis results is presented in Figure 5 for
from this work for simplicity. Thus, the results of this study
various lengths of the sidesway inhibited frame with β = 1
are only strictly applicable to compact sections.
and with the example SRC cross section shown in Figure 1
As shown in Figure 4, both sidesway inhibited and side-
and described previously. A series of analyses was per-
sway uninhibited cases were investigated. The frames are
formed to obtain the results for each individual case shown
based on and expanded from those used in previous studies
in Figure  5. First, an analysis applying only vertical load
(Kanchanalai, 1977; Surovek-Maleck and White, 2004). For
(i.e., M = 0; see Figure 4) was performed to determine the
the sidesway inhibited frames, a range of column lengths,
peak load. In this analysis, load was applied and increased
L, and end moment ratios, β, were investigated. For the
in displacement control until a limit point was determined.
sidesway uninhibited frames, a range of column lengths,
The limit point was defined as when the lowest eigenvalue

γP P P
βM
H
kθ,top

Simply
supported Composite
Beam beam-column
L

Leaning
column
x
kθ,bot

Sidesway uninhibited Sidesway inhibited

frame frame

Fig. 4.  Benchmark frames.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 271

of the stiffness matrix was equal to zero. This coincides The nominal flexural stiffness of the composite columns is
with the maximum applied axial compression. Then, eight taken as EIeff as defined in the AISC Specification (AISC,
separate nonproportional analyses were performed with 2016). All stiffnesses are reduced by 0.8, and the flexural
different values of axial compression equally spaced from stiffness of the composite column is reduced by an addi-
zero to the maximum applied axial compression from the tional factor τb = 0.8. A notional lateral load of 0.002 times
axial-only analysis. In each of these nonproportional analy- the vertical load was included. The notional load was taken
ses, the specified level of axial compression is applied in as an additive load when the ratio of second-order drift to
load control then held constant. Subsequently, the lateral first-order drift was greater than or equal to 1.7. It was taken
load is applied and increased in displacement control until as a minimum lateral load otherwise. A sample of results is
a limit point was determined. The limit point was defined presented in Figure 6(a) for the example SRC cross section
as when the lowest eigenvalue of the stiffness matrix was and the same frames investigated in Figure 5.
equal to zero. This coincides with the maximum applied
moment. In each analysis, the applied loads and maximum Results
internal forces at the limit point are recorded. These are
The key result from these analyses is the error measured
the values shown in Figure 5. The same process was used
along a radial line from the origin between the interaction
for all other interaction diagrams developed using second-
diagrams constructed from the maximum applied loads per-
order inelastic analyses in this work.
mitted by the design methodology and the applied loads at
which failure occurs according to the second-order inelastic
Design Methodology
analyses. A sample comparison is shown in Figure 6(b) for
The maximum applied loads permitted by the design meth- the example SRC cross section and the sidesway inhibited
odology are obtained from an automated iterative process frame with L/H = 40, where H is the lateral dimension of
as the applied loads that produce maximum internal forces the cross section. For higher axial loads the interaction dia-
from an elastic analysis that lay directly on the design gram constructed from the inelastic analyses is outside the
interaction diagram [either the ACB such as shown in Fig- interaction diagram constructed from the design methodol-
ure  3(a) or the ACDB interaction such as shown in Fig- ogy, indicating conservative error of up to 70%. For higher
ure 3(b)]. The elastic analyses are performed by evaluating bending moments the opposite is true, albeit to a lesser
closed-form solutions to the governing differential equa- degree, with maximum unconservative error of up to 8%.
tion for the benchmark frames obtained from a computer In this range, the design methodology permits applied loads
algebra system. Only flexural deformations are considered. that the inelastic analysis indicates would result in failure.
Axial Compression (kips)

Axial Compression (kips)

Bending Moment (kip-ft) Bending Moment (kip-ft)

(a)  Applied loads (b)  Internal forces

Fig. 5.  Second-order inelastic analysis results for frames with the example SRC cross section.

272 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

 The error was evaluated for all benchmark frames, at The maximum unconservative error varies significantly
many angles within the M-P plane, and for both the ACB with section type, slenderness, and steel ratio. The greatest
and ACDB interaction diagrams. While the selected cross unconservative errors are seen for the slenderest and most
sections and frame parameters can be considered to span concrete-dominant cases. There is no specified limit on the
the practical range, the distribution of parameters within the level of unconservative error that can be tolerated within a
selection may not be representative of what is expected in design methodology. One reference identifies a 5% uncon-
practice. For instance, the selected set contains a far higher servative error as a reasonable maximum (ASCE, 1997).
proportion of very slender frames than would be expected Another study identified unconservative errors as large as
in typical construction. Accordingly, maximum and mini- 16% for structural steel columns designed according to the
mum error values are more meaningful than median or direct analysis method with direct modeling of member
average error values. Two of the most influential parame- imperfections (Wang and Ziemian, 2019). It is important to
ters within the set are the steel ratio, ρs, and the slenderness. note that large unconservative errors have been found using
Slenderness is defined by the parameter λoe (Equation  1) the ACB interaction as well (Denavit et al., 2016). Given
which is proportional to the effective length of the columns. that the ACDB interaction diagram is larger than the ACB
An effective length factor was computed and used for interaction diagram, use of the ACDB interaction diagram
determining λoe (note, however, that the available strength can only increase the maximum unconservative errors. The
was computed with an effective length factor of unity in increase in maximum unconservative error for each bin is
accordance with the direct analysis method). The frames presented in Table 4. Compared to the magnitude of error,
were separated into bins based on ranges of steel ratio and the increase due to the inclusion of point D is modest.
slenderness to better understand the error. The ranges used The primary reason to include point D is to reduce con-
to separate the frames based on slenderness are shown in servative error in the evaluation of strength. The decrease
Table 2. The maximum unconservative error for each of the in maximum conservative error by including point D for
bins for the ACDB interaction is shown in Table 3. each bin is presented in Table  5. As expected, the larg-
Pno est decreases in conservative error occur for stockier and
λ oe = (1) more concrete-dominant frames. This range is likely more
Pe  practical and common in construction than the highly
π 2EIeff slender members for which the high unconservative errors
Pe = (2) are seen, indicating that the addition of point D would
( KL )2  be highly beneficial. Nonetheless, given the increases in

2000
Second-order inelastic analysis
1800
Design methodology (ACDB)
1600
Axial Compression (kips)

Axial Compression (kips)

1400

1200

1000

800

600 8% maximum
unconservative error
400

200

0
0 200 400 600 800 1000
Bending Moment (kip-ft) Bending Moment (kip-ft)
(a)  Design methodology—ACDB interaction /
(b)  Comparison with OpenSees (L H = 40)

Fig. 6.  Maximum permitted applied loads for frames with the example SRC cross section.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 273

 Table 2.  Definition of Slenderness Ranges
Range Slenderness
I λoe ≤ 0.5
II 0.5 < λoe ≤ 1.0
III 1.0 < λoe ≤ 1.5
IV 1.5 < λoe ≤ 2.0
V 2.0 < λoe ≤ 3.0
VI 3.0 < λoe

Table 3.  Maximum Unconservative Error Based on Slenderness and Steel Ratio, ACDB Interaction
ρs I II III IV V VI
0.25 6.00% 14.6% 12.5% 13.7% 5.70% 5.90%
0.18 4.40% 12.4% 14.0% 15.9% 8.60% 9.10%
CCFT

0.11 5.20% 9.50% 14.4% 17.9% 11.4% 12.8%

0.06 6.40% 8.90% 11.7% 12.7% 19.3% 17.9%
0.02 5.40% 6.70% 7.00% 15.6% 24.8% 36.3%
0.28 1.70% 2.40% 1.90% 3.00% 0.00% 0.00%
0.19 4.30% 3.60% 5.20% 7.00% 0.90% 1.30%
RCFT

0.11 4.00% 4.60% 8.20% 11.4% 6.30% 7.10%

0.06 3.90% 4.90% 8.70% 6.20% 16.3% 15.7%
0.03 1.60% 0.50% 4.80% 10.5% 18.7% 21.8%
0.12 6.90% 5.90% 3.60% 4.70% 6.80% 2.10%
major-axis

0.09 4.70% 3.60% 3.80% 6.70% 8.90% 4.00%

SRC

0.04 2.00% 0.90% 2.40% 9.70% 14.3% 13.1%

0.01 2.10% 2.10% 5.00% 7.40% 14.7% 29.2%
0.12 17.4% 15.8% 14.9% 13.9% 14.1% 8.60%
minor-axis

0.09 13.8% 14.6% 10.3% 12.8% 13.0% 6.80%

SRC

0.04 5.50% 5.70% 8.00% 11.0% 13.8% 11.4%

0.01 2.10% 2.10% 4.30% 7.50% 11.2% 28.1%

maximum unconservative error, this approach cannot be flexural demands. Cases such as these are perhaps not
recommended for general use unless paired with additional often seen in practice, since most engineers wisely avoid
changes that reduce the maximum unconservative errors. this range. However, there is no slenderness limit within the
AISC Specification (AISC, 2016) and thus cases for which
large errors are recorded are permitted. One remedy to
ALTERNATIVE STIFFNESS REDUCTION
these high errors would be to further reduce the size of the
The previous section addressed the source of some of the interaction diagram. However, a different remedy related to
greatest conservative errors that exist in the provisions for the stiffness reduction may be more appropriate.
steel-concrete composite columns. The ACDB interaction The errors occur with low axial loads and high bending
diagram significantly reduced the level of conservative moments. High levels of concrete cracking are expected
error while only modestly increasing the unconservative in composite columns under this loading, which is more
error. However, the unconservative error was already high beam-like than column-like. The flexural rigidity used for
in some cases. The greatest unconservative errors occur composite columns when determining required strengths
for highly slender, concrete-dominant members with large within the direct analysis method is 0.8τb EIeff , where

274 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

 Table 4.  Percentage Point Increase in Maximum Unconservative Error Based on Slenderness and Steel Ratio
ρs I II III IV V VI
0.25 3.00% 0.00% 2.50% 1.50% 1.00% 0.70%
0.18 2.00% 0.00% 3.90% 2.20% 1.30% 1.00%
CCFT

0.11 2.40% 2.70% 5.10% 4.00% 3.10% 1.70%

0.06 0.40% 3.20% 3.20% 6.40% 6.30% 3.10%
0.02 0.00% 0.40% 7.00% 9.20% 7.70% 7.50%
0.28 0.00% 0.00% 1.90% 1.20% 0.00% 0.00%
0.19 0.00% 0.00% 2.40% 1.90% 0.90% 0.90%
RCFT

0.11 0.00% 1.70% 4.00% 3.70% 2.20% 1.70%

0.06 0.10% 2.60% 4.50% 5.50% 7.10% 3.50%
0.03 0.00% 0.00% 4.80% 6.20% 7.70% 4.80%
0.12 0.00% 0.00% 2.20% 2.30% 2.40% 0.70%
major-axis

0.09 0.20% 0.30% 2.30% 4.80% 2.60% 1.20%

SRC

0.04 1.00% 0.20% 1.70% 6.20% 4.70% 2.00%

0.01 0.00% 0.00% 2.90% 3.30% 3.00% 4.80%
0.12 0.00% 0.50% 0.20% 0.40% 0.20% 0.10%
minor-axis

0.09 0.30% 0.30% 0.70% 0.40% 0.20% 0.50%

SRC

0.04 0.00% 1.90% 1.80% 0.80% 2.70% 1.40%

0.01 0.00% 0.00% 2.20% 3.10% 3.40% 4.50%

Table 5.  Percentage Point Decrease in Maximum Conservative Error Based on Slenderness and Steel Ratio
ρs I II III IV V VI
0.25 0.00% 0.00% 0.00% 0.00% 0.00% 0.00%
0.18 0.00% 5.40% 0.00% 0.30% 0.00% 0.00%
CCFT

0.11 14.4% 16.5% 1.40% 0.60% 0.80% 0.10%

0.06 32.4% 28.8% 11.7% 3.20% 1.10% 0.70%
0.02 33.6% 32.3% 25.5% 9.20% 4.50% 2.10%
0.28 0.00% 0.00% 0.00% 0.00% 0.00% 0.00%
0.19 5.20% 0.00% 0.20% 0.40% 0.00% 0.00%
RCFT

0.11 16.9% 10.6% 1.80% 0.70% 0.80% 0.00%

0.06 31.9% 21.1% 4.70% 2.70% 0.00% 0.70%
0.03 34.4% 33.4% 18.2% 4.40% 1.60% 1.20%
0.12 13.7% 0.00% 0.80% 0.50% 0.30% 0.30%
major-axis

0.09 20.6% 6.20% 1.00% 0.80% 0.30% 0.30%

SRC

0.04 30.7% 12.6% 2.20% 0.70% 1.10% 0.30%

0.01 30.3% 29.0% 10.6% 2.40% 2.20% 0.00%
0.12 0.00% 0.00% 0.30% 0.40% 0.00% 0.20%
minor-axis

0.09 5.90% 4.10% 1.10% 0.50% 0.00% 0.10%

SRC

0.04 28.2% 9.50% 0.90% 0.40% 0.70% 0.20%

0.01 30.2% 28.5% 10.0% 2.20% 2.10% 0.00%

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 275

τb = 0.8 and EIeff is the flexural rigidity used within the col- (ACI, 2019) includes provisions for an effective flexural
umn curve for determination of axial compression strength. rigidity that varies with both axial compression and bend-
Further reductions to the stiffness would help eliminate the ing moment.
observed unconservative errors. An example alternative The effect of the alternative stiffness reduction on the
stiffness reduction factor, τb, is shown in Equation 3: maximum permitted applied loads for the example SRC
cross section is shown in Figure 8(a). The solid lines repre-
Mr ⎛ Pr ⎞
τ b = 1.25 − ⎜1 − 3 ⎟ ≤ 0.8 (3) sent the maximum permitted applied loads using a constant
Mn ⎝ Pno ⎠  τb = 0.8; the dashed lines represent the maximum permitted
applied loads using Equation 3. The percentage difference
This equation is based on prior work (Denavit and Hajjar,
between the two is shown in Figure  8(b). The reduction
2014). Data on the secant flexural rigidity was computed
is sufficient to eliminate the unconservative error [e.g., as
based on results from second-order inelastic analysis; an
shown in Figure 6(b).] There are also other attractive fea-
equation was then fit to the data. The variation of the τb
tures. The alternative stiffness reduction has no effect on
described by Equation  3 with internal forces is shown in
the pure bending strength, nor does it affect the strength
Figure 7. The reduction factor is a constant τb = 0.8 for much
when the axial compression is high. Also, as seen in Fig-
of the range. Only with high bending moment and low axial
ure 8(b), it has a greater effect on more slender members,
loads, where high levels of cracking are expected, does τb
for which additional conservatism is likely warranted. The
become less than 0.8 and vary with the axial compression
specific factors in Equation 3 should be refined and a wide
and bending moment.
ranging evaluation should be performed to ensure safety
Performing an elastic analysis with a stiffness reduc-
and accuracy, but these limited results show the promise of
tion that varies with internal forces can be cumbersome.
a moment-based stiffness reduction in efficiently eliminat-
However, there is precedent in U.S. practice. For structural
ing some of the largest unconservative errors observed in
steel members, the factor τb varies with axial compression.
the design provisions for steel-concrete composite framing
For reinforced concrete members, the ACI Building Code
systems.
Requirements for Structural Concrete and Commentary
Normalized Axial Compression (Pr /Pno)

Typ
i
inte cal A
rac CB
tion

Normalized Bending Moment (Mr /Mn)

Fig. 7.  Contour plot showing the variation of the alternative stiffness reduction factor.

276 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

 CONCLUSIONS REFERENCES
This work has highlighted some of the most pressing ACI (2019), Building Code Requirements for Structural
unresolved issues in stability and strength design of steel- Concrete and Commentary, American Concrete Institute,
concrete composite framing systems: (1) the large conser- Farmington Hills, Mich.
vative errors that result from neglecting the balance point AISC (1986), Load and Resistance Factor Design Specifi-
in the calculation of available strength and (2)  the large cation for Structural Steel Buildings, American Institute
unconservative errors that result from overestimation of the of Steel Construction, Chicago, Ill.
stiffness for very slender concrete-dominant members sub-
AISC (2005), Specification for Structural Steel Buildings,
jected to high bending moments. An alternative method of
ANSI/AISC 360-05, American Institute of Steel Con-
computing the available strength interaction diagram was
struction, Chicago, Ill.
proposed and evaluated against second-order inelastic anal-
yses for a broad range of cases. The results show that using AISC (2016), Specification for Structural Steel Buildings,
the proposed interaction diagram reduces the largest con- ANSI/AISC 360-16, American Institute of Steel Con-
servative errors but worsens existing unconservative errors. struction, Chicago, Ill.
To address this issue, an alternative stiffness reduction that AISC (2017), Steel Construction Manual, 15th Ed., Ameri-
varies with internal forces was proposed to better capture can Institute of Steel Construction, Chicago, Ill.
the occurrence of high levels of cracking and eliminate the ASCE (1997), Effective Length and Notional Load
unconservative errors. Initial studies with this alternative Approaches for Assessing Frame Stability: Implications
stiffness reduction showed promising results. Both alter- for American Steel Design, American Society of Civil
native approaches, once fully validated, have the potential Engineers, Reston, Va.
to improve the accuracy and safety of the stability design
Behnam, A. and Denavit, M.D. (2020), “Plastic Stress Dis-
provisions for steel-concrete composite framing. They can
tribution Method for Predicting Interaction Strength of
also set the stage for future developments such as design
Steel-Concrete Composite Cross Sections,” Journal of
provisions based on cross-section strength and the use of
Constructional Steel Research, Vol. 170, 106092.
high-strength materials.
Axial Compression (kips)
Axial Compression (kips)

Reduction in Maximum Permitted

Bending Moment (kip-ft) Applied Bending Moment
(a)  Maximum permitted applied loads (b)  Reduction in maximum permitted applied loads
with and without the alternative stiffness reduction from use of the alternative stiffness reduction

Fig. 8.  Results using the alternative stiffness reduction.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 277

Bruneau, M., Kenarangi, H., and Murphy, T.P. (2018), Con- Kanchanalai, T. (1977), The Design and Behavior of Beam-
tribution of Steel Casing to Single Shaft Foundation Columns in Unbraced Steel Frames, CESRL Report
Structural Resistance, NCHRP Research Report 872, No.  77-2, Structures Research Laboratory, Department
The National Academies Press. of Civil Engineering, The University of Texas at Austin,
CEN (2004), Eurocode 4: Design of Composite Steel and Austin, Tex.
Concrete Structures—Part 1-1: General Rules and Rules Lai, Z., and Varma, A.H. (2016). “Effective Stress-Strain
for Buildings, EN1994-1-1, European Committee for Relationships for Analysis of Noncompact and Slender
Standardization, Brussels, Belgium. Filled Composite (CFT) Members,” Engineering Struc-
Denavit, M.D. (2017), “Structural Reliability of Steel- tures, Vol. 124, pp. 457–472.
Concrete Composite Columns and Frames,” Proceedings Lai, Z., Varma, A., and Griffis, L. (2015), “Analysis
of the 8th International Conference on Composite Con- and Design of Noncompact and Slender CFT Beam-
struction in Steel and Concrete, Jackson, Wyo. Columns,” Journal of Structural Engineering, ASCE,
Denavit, M.D. and Hajjar, J.F. (2014), Characterization of Vol. 142, No. 1, 04015097.
Behavior of Steel-Concrete Composite Members and SAC (2014), Technical Code for Concrete Filled Steel Tubu-
Frames with Applications for Design, Newmark Struc- lar Structures, GB 50936-2014, Standardization Admin-
tural Laboratory Report Series, Newmark Structural istration of the People’s Republic of China, Beijing, China.
Laboratory Report NSEL-034, University of Illinois at Surovek-Maleck, A.E. and White, D.W. (2004), “Alterna-
Urbana-Champaign, Urbana, Ill. tive Approaches for Elastic Analysis and Design of Steel
Denavit, M.D., Hajjar, J.F., Perea, T., and Leon, R.T. (2016), Frames. II: Verification Studies,” Journal of Structural
“Stability Analysis and Design of Composite Structures,” Engineering, ASCE, Vol. 130, No. 8, pp. 1,197–1,205.
Journal of Structural Engineering, ASCE, Vol.  142, Traut-Todaro, J. (2019), “SpeedCore: Lateral System Inno-
No. 3, 04015157. vation for Today’s Construction Challenges,” Structure
Griffis, L.G. (1992), “Composite Frame Construction,” Con- Magazine, November, pp. 12–14.
structional Steel Design: An International Guide, Else- Wang, Y. and Ziemian, R.D. (2019), “Design by Advanced
vier Applied Science, London, UK, pp. 523–553. Elastic Analysis—An Investigation of Beam-Columns
Griffis, L.G. (2005), “Composite Design Provisions 2005 Resisting Minor-Axis Bending,” Proceedings of the
AISC Specification for Steel Buildings,” Proceedings of Annual Stability Conference, Structural Stability
the 2005 Structures Congress, ASCE, New York, N.Y. Research Council, St. Louis, Mo.
Hage, S.E. and MacGregor, J.G. (1974), The Second-Order
Analysis of Reinforced Concrete Frames, Structural Engi-
neering Report No. 49, Department of Civil Engineering,
University of Alberta, Edmonton, Alberta, Canada.

278 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

Lateral-Torsional Buckling Research Needs and
Validation of an Experimental Setup
in the Elastic Range
RYAN SLEIN, JOSHUA S. BUTH, WAJAHAT LATIF, AJIT M. KAMATH, AMMAR A.
ALSHANNAQ, RYAN J. SHERMAN, DAVID W. SCOTT, and DONALD W. WHITE

ABSTRACT
The AISC Specification Chapter F I-section member flexural resistance equations are a central part of structural steel design in the United
States. The provisions of Sections F4 and F5 address general singly and doubly symmetric I-section members. Analytical studies and
experimental tests subsequent to the implementation of these provisions within the 2005 AISC Specification suggest that the corresponding
inelastic lateral-torsional buckling (LTB) and tension flange yielding (TFY) resistance equations can be improved, resulting in significantly
larger predicted strengths in certain cases and somewhat smaller predicted strengths in other cases. Additional large-scale experimental
tests, specifically pushing into the inelastic LTB range, need to be conducted to further investigate these predictions. The broad objective
of the additional tests is to achieve a target reliability index of β = 2.6 for building design at a live-to-dead load ratio of 3.0 throughout the
design space involving all types of statically determinate I-section flexural members.

This paper discusses the need for these tests, specifically focusing on the details of how the test fixtures and bracing systems were config-
ured to minimize incidental restraint, which is a critical consideration when conducting flexural experimental testing. The paper discusses
the validation of the testing system by comparison of elastic buckling experimental results to analytical and numerical solutions.

Keywords:  lateral-torsional buckling, experimental testing, incidental restraint.

INTRODUCTION and members with compact, noncompact, or slender flanges

and/or webs, failing by plastic, inelastic, or elastic lateral-
T he Chapter F equations in the AISC Specification for
Structural Steel Buildings, hereafter referred to as the
AISC Specification (AISC, 2016b), provide a broad charac-
torsional buckling (LTB). Closely related equations exist
within the AASHTO LRFD Bridge Design Specifications
(AASHTO, 2020).
terization of the flexural resistance of all types of I-section
Relatively comprehensive assessments of analytical stud-
members, including rolled and welded members; members
ies and experimental test results conducted to date have
with doubly and singly symmetric cross-section profiles;
raised concerns that AISC Specification Sections F4 and
F5 flexural resistance provisions may not satisfy accepted
target reliability indices in certain cases pertaining to the
Ryan Slein, Graduate Research Assistant, Georgia Institute of Technology, LTB of I-section members (Subramanian et al., 2018; Sub-
Atlanta, Ga. Email: [email protected] (corresponding) ramanian and White, 2017). However, the experimental test
Joshua S. Buth, Civil Engineer II, Georgia Department of Transportation, data are quite sparse within a number of “regions” of the
Atlanta Ga. Email: [email protected]
corresponding design space. Additionally, AISC Specifica-
Wajahat Latif, Graduate Research Assistant, Georgia Institute of Technology, tion Sections F4 and F5 can be enhanced by eliminating
Atlanta, Ga. Email: [email protected]
the current tension flange yielding (TFY) limit state pro-
Ajit M. Kamath, Graduate Research Assistant, Georgia Institute of Technol-
ogy, Atlanta, Ga. Email: [email protected]
visions. Allowing for development of significant reserve
capacity involving yielding in flexural tension by incorpo-
Ammar A. Alshannaq, Graduate Research Assistant, Georgia Institute of
Technology, Atlanta, Ga. Email: [email protected] rating early tension yielding effects into the calculation of
Ryan J. Sherman, Assistant Professor, Georgia Institute of Technology, the cross-section yield moment to the compression flange,
Atlanta, Ga. Email: [email protected] Myc (Toğay and White, 2018).
David W. Scott, Professor and Chair, Department of Civil Engineering and Quality experimental data is critical for the validation
Construction., Georgia Southern University, Statesboro, Ga. Email: dscott@ of refined shell finite element analysis (FEA) procedures
georgiasouthern.edu
that can be employed to investigate the flexural resistances
Donald W. White, Professor, Georgia Institute of Technology, Atlanta, Ga.
within the design space more comprehensively.
Email: [email protected]
This paper focuses on the validation of the testing con-
figuration used to conduct inelastic tests to achieve the
Paper No. 2020-12 above objectives. Lateral-torsional buckling experimental

ISSN 0013-8029 ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 279

results can be sensitive to incidental restraint in the testing the blue members are the bracing reaction system, and the
configuration (Ziemian, 2010). The testing configuration gray members are the load frame. The loading and support
described in this paper utilizes polytetrafluoroethylene- fixtures and bracing details are discussed in the following
(PTFE-) coated spherical bearings for multi-rotational sections.
degree-of-freedom releases, mechanical bearings for the
single rotational degree-of-freedom releases (as part of a Load and Bearing Fixtures
Watt’s linkage bracing system), and lubricated roller packs
Roller boundary conditions were provided at the bearing
for single translational degree-of-freedom releases. The
locations. The overall boundary conditions were symmetric
effectiveness of these “releases” of rotational and transla-
about the mid-length of the test specimens. Longitudinal
tional constraints is evaluated directly by testing a speci-
translation was permitted via a lubricated roller pack com-
men in the elastic LTB range and comparing the measured
posed of four 2.5-in.-diameter solid steel rods. A 100-kip
responses to various analytical and numerical solutions.
load cell was located above each roller pack. Transverse dis-
placement was restrained at the bearing locations via Watt’s
TEST CONFIGURATION linkage braces discussed in the following section. The three
rotational degrees-of-freedom at the supports were released
The experimental test setup was designed to minimize inci-
via PTFE-lined spherical thrust bearings. The PTFE thrust
dental restraint and remove load-height effects. Inciden-
bearings allowed free in-plane rotation due to the major-
tal restraint can have a measurable impact on large-scale
axis bending of the specimens and free out-of-plane rota-
experimental LTB beam results by increasing the capacity
tion associated with flange warping and/or lateral bending.
beyond that based on the ideal boundary conditions (Zie-
The spherical bearing was seated in a counter-bored plate
mian, 2010). The test configuration was designed to fully
on top of the load cell. Weld beads were placed on the
release or fix selected degrees-of-freedom at the bearing,
flange of the specimen to securely seat the opposite side of
load, and bracing points. Load-height effects at the load
the spherical bearing on the specimen. Figure 3 provides a
points can make the calibration of design equations to
conceptual drawing of the bearing detail used for all tests
observed specimen behavior more complex. To eliminate
as well as a photograph of the final bearing detail prior to
load-height effects, the lateral and torsional displacements
placing the spherical thrust bearing at the top.
were restrained at the bearing and load points in all the
In addition, a PTFE spherical thrust bearing was located
experiments conducted in this research. The design of the
at the point of load application (e.g., at the midspan of the
test setup involved an extension of the concepts discussed in
three-point bending test specimens). Load-point bracing
the Structural Stability Research Council (SSRC) Technical
was provided via a Watt’s linkage system. Figure 4 shows
Memorandum No. 9 on flexural testing (Ziemian, 2010).
the conceptual and implemented boundary condition at the
Figure 1 provides an elevation view of a test specimen,
point of load application.
discussed throughout this paper, under three-point bend-
ing. The elevation view and measured section dimensions
Bracing
in Figure  1 are drawn to scale and show instrumentation
locations and the corresponding moment diagram. The bracing system was designed using a mechanical system
Figure  2 is a photograph of the test specimen within known as a Watt’s linkage. The Watt’s linkage restrains dis-
testing frame. The white member is the test specimen, placement perpendicular to the girder web, while allowing

8.5 ft 17 ft 17 ft 8.5 ft bf = 4.963 in.

SP-6 SP-7 SP-5
tf = 0.307 in.

SP-1 SP-8 SP-2 & SP-4 tw = 0.246 in.

SP-3 Brace point (typ.)
= horizontal string pot. h = 29.993 in.
= vertical string pot. Cb = 1.38, Ke = 0.82

Mmid Fyw = 61.9 ksi

Fyf = 58.2 ksi

Fig. 1.  South-facing elevation view of the elastic test specimen showing the instrumentation locations,
corresponding moment diagram, measured section dimensions and properties, and key design parameters Cb and Ke .

280 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

 Fig. 2.  Perspective view of the test configuration.

Fig. 3.  Bearing boundary condition detail: schematic (left) and implementation (right). The
spherical bearing was excluded in the photo to show the counter-bore. The roller pack was chocked
in the photo to prevent incidental movement during installation of the test specimen.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 281

free translation in the longitudinal and vertical directions. the linkage system to prevent deflections in the direction
Watt’s linkage bracing has been used previously by Yarimci perpendicular to the girder webs (i.e., the horizontal direc-
et al. (1967), Smith et al. (2013), and others. For the cur- tion in the figure) under large longitudinal displacements
rent study, the system was comprised of two 4-ft-long tie (i.e., displacements in the vertical direction in the figure).
rods with ball joint rod ends attached to the stiff reaction In addition, girder vertical deflections (i.e., deflections into
frames (painted blue in Figure 2) and a 6.5-in.-long center and out of the page in the figure) are accommodated by the
link. The center link is free to rotate about a vertical axis rotation of the center link about a vertical axis.
and transfers lateral forces to the girder through a cylindri- The Watt’s linkage system does an excellent job of releas-
cal mechanical bearing referred to as a flange block. The ing incidental constraint. Annotated photos describing the
center link is attached at its mid-length to a pillow block on system are provided in Figure  6 and 7. Several additional
each side of the 2.5-in.-thick plate containing the counter- unique features added to the design of the bracing system
bore at the load and support points and to a flange block include the following:
at the other brace points. Figure 5 illustrates the ability of

Fig. 4.  Load point boundary condition detail: schematic (left) and implementation (right).

Longitudinal axis of
the test specimen

Fig. 5.  Plan view illustration of a Watt’s linkage movement under a large deflection along the axis of the test specimen. The heavy
dashed blue lines represent the initial geometry of the linkage, the solid black lines represent a deformed geometry of the linkage, and
the red dashed line illustrates the path of the brace point (i.e., the middle of the center link) between the two geometries.

282 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

• Each end of the tie rods was threaded—left-hand threads NUMERICAL MODELING
on one end and right-hand threads on the other. The
Finite element methods can predict incipient buckling
opposing threads allowed for quick length adjustments
(bifurcation) by calculating the lowest eigenvalue using the
via rotation of the rod about its axis, without having to
equation
unbolt the ends, allowing for fine adjustments to ensure
the test specimen was plumb at each brace point. ⎡⎣K e. ff ⎤⎦ { Δ f } = λ ⎡⎣− K g. ff ⎤⎦ {Δ f } (1)

• Jam nuts were used to ensure the tie rod length did not
where Ke.ff is the elastic stiffness matrix only considering
change during the loading of the test specimens.
free degrees of freedom, Kg.ff is the geometric stiffness
• A rail system was implemented for rapid reconfiguration matrix calculated from element forces (or stresses) from a
between different unbraced lengths and bracing linear elastic analysis for a known reference load Pref, Δf is
configurations. The tie rods were bolted to vertical WT the displacement vector associated with the free degrees of
members containing a series of holes accommodating freedom, and λ is the lowest eigenvalue representing the
varying specimen heights. The bracing reaction frame ratio of the elastic critical load to Pref (McGuire et al., 2000).
was composed of wide flange rails that extended the Iterating Pref until λ goes to 1.0 results in a predicted
entire length of the test setup. Friction-based connections elastic linear buckling strength. For the test specimens,
by Lindapter (2019) were used to connect the vertical self-weight and the weight of the bracing components are
WT sections to the wide flange rails. Each WT slid along a constant value, so it is inappropriate to scale a constant
the rails to accommodate a range of specimen unbraced reference applied load, Pref, solely by λ. That is, the load-
lengths. ing on the system includes a constant load due to the ini-
tial self-weight of the specimen and bracing attachments
• Each brace point, including the Lindapter friction-based
to the specimen, plus the reference load multiplied by the
connection, was designed to accommodate a transverse
applied load parameter, λ. Once the initial load due to the
force of 20 kips. A flange block (a housed cylindrical
self-weight is established, then either Pref can be varied and
bearing) was used to release the rotation about the vertical
an eigenvalue solution sought such that λ = 1.0, or a con-
axis at the girder flanges for the Watt’s linkage system. At
stant applied reference load can be specified and an eigen-
the load application and bearing locations, pillow blocks,
value multiple of this reference load sought corresponding
with a different housing but the same internal cylindrical
to the buckling of the specimen. In either case, the internal
bearing, were selected. To prevent pull-out, the outside
forces are the sum of the constant forces due to the initial
diameter of the stem connecting the center link to the
pillow block was match-machined to the inside diameter
of the bearing for a press-fit connection. Additionally,
four set screws bear on flats on the center-link stem.

Fig. 6.  Components of the boundary conditions. Fig. 7.  Components of the bracing system.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 283

self-weight plus the forces due to the additional applied load the length of the members was selected such that the aspect
at buckling. In this research, these elastic buckling calcu- ratio of the web elements was close to 1.0. This mesh den-
lations were conducted using the finite element software sity has been determined to be sufficient for convergence of
system SABRE2 (White et al., 2020) based on thin-walled full nonlinear FEA solutions of various I-section members
open-section (TWOS) beam theory considering warping in prior research—for example, Prado and White (2015)—
torsion. as well as in these research studies.
The elastic effective length factor, Ke, is back-calculated The nonlinear shell FEA solutions in ABAQUS were
by equating the maximum internal moment within the criti- implemented through a modified RIKS arc length proce-
cal unbraced length, at the elastic buckling load level deter- dure. The modified RIKS algorithm is particularly useful
mined by the computational solution, to the equation for the in obtaining the post-buckling response for cases where the
elastic critical moment, Mcr, from recommended and AISC loading is proportional—that is, where the load magnitudes
Specification (2016b) provisions: are governed by a single parameter (Simulia, 2021). The
2 modified RIKS algorithm conducts a load-deflection analy-
Cb S xc π 2 E J ⎛ K e Lb ⎞ sis where the load is incremented by scaling the reference
Mcr = 1+ 0.078 (2)
( Ke Lb rt )2 S xcho ⎝ rt ⎠

load, or a set of applied reference loads, by the load parame-
ter. The definition of initial residual stresses, and the appli-
An expedient process to determine Ke is to employ the goal- cation of the applied loads to the test specimens starting
seek root-finding capability in Excel to iteratively solve for from their loaded state under their self-weight, was accom-
the value of Ke that satisfies the equality. plished by subdividing the analysis into multiple steps. Note
In Equation  2, Cb is the moment gradient factor, E is that the residual stress pattern follows half the magnitude
the elastic modulus, Sxc is the section modulus, Lb is the of the best-fit Prawel pattern based on recommendations
unbraced length of the critical section, rt is the radius of from Subramanian and White (2017), who state that this
gyration for LTB, J is the St. Venant torsion constant, and provides reasonable correlation with the mean results from
ho is the distance between the flange centroids. For the experimental tests. A first step was employed to solve for
slender-web members, J was taken equal to zero in Equa- the equilibration of the initial residual stresses on the geo-
tion  2, as well as in the calculation of the buckling load metrically imperfect model (the initial residual stress pat-
from SABRE2. The calculated value of Ke is used in all rec- tern generally does not satisfy equilibrium on the imperfect
ommended and AISC Specification strength calculations of structure geometry, nor at free-ends of the specimens). A
this research to directly account for the end restraint from second step was then employed to apply the constant self-
adjacent less-critical unbraced lengths due to continuity weight loads. Finally, the modified RIKS algorithm was
effects. An alternative means of calculating Ke is to use an applied in a third step to place the applied load incremen-
approximate method such as that proposed by Nethercot tally on the model.
and Trahair (1976). However, with the increased availabil-
ity of software capable of conducting basic elastic linear
VALIDATION OF THE TEST SETUP
buckling analysis (ELBA), the “exact” calculation of Ke is
THROUGH AN ELASTIC LTB TEST
the more appropriate solution for research evaluations.
ABAQUS Version 6.13 (Simulia, 2021) finite element A benchmark elastic LTB experiment was conducted to
analysis software was employed for developing GMNIA evaluate the effectiveness of the translation and rotational
(geometric nonlinear material linear-elastic analysis with releases in the test setup. Figure  1 provides an elevation
imperfections) finite element simulations in this research. view of the test specimen as well as the moment diagram.
The finite element mesh consisted of B31 beam elements The elevation view and measured section dimensions
for transverse stiffeners and S4R nonlinear shell elements in Figure  1 is drawn to scale and shows instrumentation
for all other components. The B31 is a two-node, three- locations. For the critical unbraced lengths adjacent to the
dimensional beam element that allows for transverse shear mid-span, Cb  = 1.38 from AISC Specification Commen-
deformation. The B31 element is based on linear-order tary Equation C-F1-2b (AISC, 2016b) and Ke = 0.82 for the
displacement interpolation and is compatible with the critical unbraced lengths adjacent to the mid-span (back-
S4R shell element. The S4R is a four-node, quadrilateral calculated from an elastic buckling analysis conducted
displacement-based shell element with reduced integration. using SABRE2). The resulting configuration slenderness
Both the B31 and the S4R elements are based on large- was well within the elastic LTB range.
strain formulations. The shell finite element mesh for the It is important to note that due to stable elastic post-
I-section specimens was generated using 12 elements across buckling response in the governing LTB mode, the speci-
the flange width and a minimum of 16 elements through men potentially can develop a maximum load capacity
the web depth. The dimension of the shell elements along larger than the elastic critical load, due to the large LTB

284 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

slenderness for the elastic LTB testing arrangements. Any The test specimen was fabricated by a prominent metal
actual post-buckling strength is directly dependent upon the building manufacturer and is representative of main frame
extent of early yielding due to the combined effects of the members in typical metal building frames. The web-to-
girder loads, the initial geometric imperfections, the initial flange welds are minimum size single-sided fillet welds.
residual stresses, and the amplified lateral bending of the Both flanges are fabricated from rolled bar stock, while
compression flange as the theoretical elastic critical load the web was cut from a coil. In addition, the specimen has
is approached. The peak load at which the elastic LTB test double-sided stiffener plates at all the brace points, includ-
reached is compared directly to the theoretical elastic LTB ing load and bearing locations, to control cross-sectional
resistance and to the capacity predicted from shell FEA test distortion.
simulation. The FEA simulation models included the evalu- Measured compression flange sweep and web out-of-
ation of the post-peak response. Additionally, plots of the flatness of the test specimen are shown in Figures 8 and 9,
horizontal displacement of the flanges versus the load were respectively. Geometric imperfection measurements were
employed to estimate the theoretical elastic LTB moments taken after the beam was installed and plumbed. Allowable
via Southwell, Meck, and Massey plots (Mandal and Cal- tolerances of Lb /480 for compression flange sweep and h / 72
ladine, 2002). for web out-of-flatness are specified in the Metal Building

Fig. 8.  Measured compression flange sweep.

Fig. 9.  Measured web out-of-flatness.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 285

 Table 1.  Peak Moment Contributions
Maximum Moment at Mid-Span Due to Dead Loads
Moment
Dead Load Calculation kip-ft (kip-in.)
Girder self-weight /
(0.036 klf) (51 ft)2 8 11.6 (139)
Mid-span attachments (0.38 kip) (51 ft)/ 4 4.95 (59.4)
Bracing attachments (0.07 kip) (8 ft) 0.540 (6.50)
Maximum Moment at Mid-Span Due to Applied Loads
Moment
Applied Load Calculation kip-ft (kip-in.)
Initial seating preload /
(0.50 kip) (51 ft) 4 6.38 (76.5)
Actuator load at failure (8.51 kips) (51 ft)/ 4 109 (1300)
Maximum moment at mid-span 133 (1600)

Systems Manual (MBMA, 2018). These tolerances are shows numerical strength predictions from an ABAQUS
approximately double the imperfection limits given by the nonlinear shell FEA test simulation and from a thin-walled
AISC Code of Standard Practice (AISC, 2016a). The other open-section (TWOS) beam theory inelastic buckling anal-
measured imperfections—that is, tension flange sweep, ysis using SABRE2, as well as estimates of the theoretical
combined flange warpage and tilt, and web off-center— elastic buckling load from a Southwell plot based on the
were documented but are not specified in this paper. A measured experimental displacements.
detailed force and moment tabulation is summarized in In addition, Figure 10 shows the normalized theoretical
Table  1, including the influence of all self-weights of the elastic buckling curve from Equation 2 for a range of mem-
testing specimen and bracing components, load applied bers having the same configuration as in Figure 1 but with
prior to zeroing the load cells at the start of the experiment, different effective unbraced lengths, using the calculated
and the measured peak load during the experiment. finite J for the specific specimen. This is the dashed black
The experimental strength of 1,600 kip-in. is normal- curve in the figure. Also shown in light gray is the LTB
ized by the moment corresponding to compression flange strength curve from the recommended provisions presented
yielding, Myc, of 4,760 kip-in. and plotted in Figure  10 at in Slein et al. (2021). This curve is based on J  = 0 since
the effective length, KeLb = 13.9 ft. Furthermore, Figure 10 the web for this cross section classifies as slender using

1.2 M/Myc Mtest/M

Mtest 0.332 -
1.0
Mn FEA 0.321 1.034
Mn SABRE2 0.307 1.081
0.8
Mn Southwell 0.348 0.954
Mn Elastic 0.307 1.081
M/Myc

0.6

0.4

0.2

0.0
0 2 4 6 8 10 12 14 16 18 20
KeLb (ft)

Fig. 10.  Comparison of experimental test results to numerical results from ABAQUS and SABRE2 to a
Southwell plot estimate of the elastic buckling load and to elastic LTB and design strength curves.

286 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

the recommended provisions. The “exact” elastic critical shows the measured moment versus section twist at the
moment for the specimen is determined using SABRE2, cross section where SP-7 and SP-8 are attached, overlaying
and the Ke value is determined using Equation 2 (using the the predicted moment-twist from the shell FEA.
solution based on J = 0 for this member because its web is The GMNIA solution from ABAQUS is capable of cap-
classified as slender in the recommended provisions). turing a capacity in the elastic test that is larger than the
Figure  11 shows the measured moment versus vertical theoretical elastic LTB strength due to the stable elastic
displacement for SP-2, overlaying the predicted moment- post-buckling response of the member. The contributions
maximum vertical displacement from the shell FEA. from elastic post-buckling strength are negligible for most
Figure 12 shows the measured moment versus lateral dis- practical LTB slenderness values; however, given the large
placements for SP-7 and SP-8, overlaying the predicted slenderness in this elastic LTB test, strengths larger than
moment-maximum horizontal displacement of the com- the theoretical elastic LTB resistance are possible. Fig-
pression flange and the corresponding horizontal displace- ure 14 shows the midspan moment versus the compression
ment of the tension flange from the shell FEA. Figure 13 flange lateral deflection from the GMNIA solution, a shell

Fig. 11.  Moment-vertical deflection.

Fig. 12.  Moment-horizontal deflection of both flanges.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 287

FEA geometric nonlinear (material linear-elastic) analysis flange lateral displacements increases abruptly, the mem-
with imperfections using the same model as in the GMNIA ber was very close to maximum capacity. The second-order
solution (referred to in the literature as a GNIA solution), amplifications of the girder lateral displacements and twists
and the ELBA solution obtained from SABRE2. in the experiment matched well with the theory.
As the theoretical elastic critical moment is approached,
the compression flange lateral displacement increases Experimental Estimation of Elastic Buckling Load by
rapidly, both in the full nonlinear (GMNIA) and the geo- Southwell, Meck, and Massey Plots
metrically nonlinear (GNIA) shell FEA solutions. The cor-
As a final evaluation of the test setup effectiveness—
responding rapid increase in the compression flange lateral
specifically the minimization of incidental restraint—
bending strains induces the onset of yielding within the
measured displacements and loads were used to generate
compression flange, resulting in a limit load in the GMNIA
Southwell, Meck, and Massey plots for the elastic tests.
solution. This behavior was observed during the experi-
These plots allow estimation of the elastic critical moment
ment, that when the rate of change of the compression

Fig. 13.  Moment-twist of the section.

Fig. 14.  Load-horizontal displacement curves from ABAQUS GMNIA

and GNIA solutions compared to the ELBA solution from SABRE2.

288 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

of the specimen, Mcr. Mandal and Calladine (2002) discuss φ/M versus u) and β (the inverse slope of u/M versus φ)—
the effectiveness and theoretical underpinnings of these that is, the geometric mean of α and β,
plots for LTB problems. Each of the methods plot a varia-
Mcr = αβ  (5)
tion of a displacement divided by a load term versus dis-
placement, then use the inverse of the slope of this curve to The Massey plot takes Mcr as the geometric mean of α
estimate Mcr. The authors interpreted the application of the (taken as the inverse slope of φ/M 2 versus φ), and β (taken
different estimation methods as described in the following as the inverse slope of u/M 2 versus u). The values of α and
discussion. β for the Meck and Massey plots are defined by the slope
The Southwell plot for LTB was generated using the of the corresponding plots, similar to Mcr on the Southwell
lateral displacement of the compression flange, u, as the plot. There is some complexity that is not captured by these
abscissa versus u/M as the ordinate. The lateral displace- methods since the critical unbraced lengths of the speci-
ment was measured using a linear string potentiometer, men are not flexurally and torsionally simply supported. As
located at the expected position of the maximum com- such, the second-order amplifications of the compression
pression flange lateral displacement. The estimate is not flange lateral deflection, u, and the twist, φ, do not have the
sensitive to the specific selected location as long as the same mathematical form as that of a simply supported col-
magnitude of the instrumentation noise is low compared to umn (for that matter, they do not have the same form when
the magnitude of the measurement. The maximum moment the LTB specimen is torsionally and simply supported,
at mid-span was calculated as which leads to the consideration of the alternative estima-
PL tion procedures other than the basic Southwell plot). How-
M= + Mo (4) ever, the methods agree reasonably well with each other
4 
for the elastic test. In the limit that the moment approaches
where P is the summation of load cell measurements at the the theoretical elastic buckling moment, it appears that the
end supports, L is the distance between the end supports, simple assumptions for the form of the second-order ampli-
and Mo is the mid-span dead load moment. Figure 15 shows fication of the displacements, embedded within the South-
the resulting Southwell plot for the elastic test. well, Meck, and Massey plots, apply reasonably well. The
The Meck plot was similar to the Southwell plot, but it calculated maximum moments are 1,660 kip-in. (0.348Myc),
also considers the twist of the cross section, φ. During the 1,640 kip-in. (0.344Myc), and 1,570 kip-in. (0.329Myc) from
elastic test the tension flange had negligible out-of-plane the Southwell, Meck, and Massey procedures, respectively.
motion, as shown in Figure  12; therefore, the twist of the These theoretical estimates show good agreement with
cross section is directly proportional to the lateral displace- the maximum moment of 1,600 kip-in. measured in the
ment of the compression flange. The Meck plot takes Mcr experiment.
as the square root of the product of α (the inverse slope of

Fig. 15.  Southwell plot.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 289

 QUANTIFICATION OF obtained in the FEA solution can be sensitive to varia-
INCIDENTAL RESTRAINT tions in the residual stresses and geometric imperfections
when the compression flange major-axis bending stress is
To quantify the ability of the testing system to minimize
relatively large at the LTB strength condition. The expected
incidental restraint, the buckling load solutions from South-
sensitivities of Mtest and MnFEA to the residual stresses
well, Meck, and Massey are compared with the result from
and geometric imperfections when KLb is close to Lr, and
an elastic linear buckling analysis (ELBA). The effective
the uncertainties in the precise values of these quantities,
length in the theoretical elastic buckling solution is back-
makes the use of Mtest/MnFEA less of a useful measure of
calculated directly from an ELBA in SABRE2, so the
incidental constraint within the testing system.
theoretical elastic buckling solution from Equation  2 and
the SABRE2 ELBA are the same. The reported SABRE2
solutions are from an inelastic nonlinear buckling analysis CONCLUSIONS
(INBA). Note that KeLb in the elastic test is much larger
Incidental restraint can have a measurable impact on exper-
than Lr; therefore, the major-axis bending moment is at a
imental LTB test results. Substantial attention was given
level where there is not a significant onset of yielding prior
to minimizing the constraint in the experimental setup
to reaching the maximum capacity, resulting in a similar
described in this paper. The setup design consists of ide-
INBA solution as ELBA. Structural steels such as ASTM
ally released degrees of freedom, while restraining speci-
A572 Grade 55 generally have a mean elastic modulus
fied degrees of freedom, using mechanical bearings in a
close to 29,500 ksi with a small coefficient of variation
Watt’s linkage bracing system. PTFE spherical thrust bear-
(Hartmann, 2005). Using this larger elastic modulus in the
ing allowed for free flange lateral bending and major-axis
ELBA solution gives a more accurate representation of the
and weak-axis rotation. Complications of load height were
physical response. The elastic critical moment scales by
bypassed with placing stiffeners at points of load application
29,500/ 29,000 given this change.
and bearing, along with a lateral brace of both flanges. For
For the elastic test, the Southwell, Meck, and Massey esti-
an accurate comparison to strength predictions, numerical
mates give Mcr/Myc ranging from 0.329 to 0.348. The ELBA
solutions were modeled to be representative of the physical
eigenvalue result from SABRE2, with E = 29,500 ksi, gives
specimen and manual predictions utilized a rigorously cal-
Mcr/Myc of 0.312. Therefore, in terms of elastic buckling
culated effective length factor.
load, the test shows an influence of incidental constraint of
To validate the effectiveness of the testing setup, an elas-
5 to 12%, depending on which estimates are used for deter-
tic LTB test was conducted and compared to analytical
mining the elastic critical moment experimentally.
and numerical predictions, as well as to theoretical elastic
To a lesser extent, the ability of the testing system to
critical moment estimates. The most direct prediction of
minimize incidental restraint can be quantified by compar-
the effectiveness of the setup is believed to be the compari-
ing the experimental load to the shell FEA load-deflection
son of the buckling load solutions from Southwell, Meck,
solution. The Mtest/MnFEA value was 1.03 for the elastic test.
and Massey to an elastic linear buckling analysis. Result-
However, the elastic test was halted prior to reaching the
ing data indicated an influence from incidental restraint on
limit load (when the compression flange lateral bending
the LTB strength ranging between 5 and 12%, depending
started to increase significantly) to ensure that no signifi-
on the methodology. The nominal difference can be attrib-
cant yielding occurred, since the specimen is subsequently
uted to the small amount of remaining restraint as well as
used for an inelastic LTB test, resulting in slightly low
other factors influencing the predictions, such as discrete
measured experimental strengths. This is more of a prob-
measurements of plate dimensions, material properties,
lem when KLb is in the proximity to Lr, due to the greater
and geometric imperfections. Ultimately, the results dem-
propensity for the onset of yielding due to the addition of
onstrate the designed test configuration minimizes inci-
amplified flange lateral bending stresses as the theoretical
dental constraint, resulting in accurate experimental LTB
elastic buckling load was approached. Additionally, poten-
test data.
tial overprediction of MnFEA can be a source of error. The
shell FEA test simulation potentially could have overpre-
dicted the strength, for example, by its use of one-half of ACKNOWLEDGMENTS
the best-fit Prawel residual stress profile (i.e., by assuming
The authors would like to acknowledge MBMA, AISC, and
residual stresses that are relatively small and symmetric
AISI for their generous support of this research; American
about the mid-width of the flanges), as well as by exclud-
Buildings Company, BlueScope Buildings North America,
ing the web off-center imperfections in the modeling of the
and Schulte Building Systems for their generous donations
specimen geometry, in this test. The maximum strength

290 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

of the test specimens; AISC and NUCOR Fastener Division Simulia (2021), Abaqus 6.13, Dassault Systems, https://
for material and fastener donations; and Thomas Murray www.3ds.com/products-services/simulia/services-sup-
of Virginia Tech for acting as the MBMA senior advisor to port/support/documentation/ (February 22).
the project. They would also like to thank the Georgia Tech Slein, R., Kamath, A.M., Latif, W., Phillips, M.L., Sherman,
Structural Engineering Laboratory staff and other GT per- R.J., Scott, D.W., and White, D.W. (2021), “Enhanced
sonnel for their many contributions to the project. Characterization of the Flexural Resistance of Built-up
I-Section Members,” SEMM Research Report 21-01,
REFERENCES School of Civil and Environmental Engineering, Georgia
AASHTO (2020), AASHTO LRFD Bridge Design Speci- Institute of Technology, Atlanta, Ga.
fications, 8th Ed., American Association of State and Smith, M.D., Turner, A.K., and Uang, C.-M. (2013), “Exper-
Highway Transportation Officials, Washington, D.C. imental Study of Cyclic Lateral-Torsional Buckling of
AISC (2005), Specification for Structural Steel Buildings, Web-Tapered I-Beams,” Department of Structural Engi-
ANSI/AISC 360-05, American Institute of Steel Con- neering University of California, San Diego, Calif.
struction, Chicago, Ill. Subramanian, L., Jeong, W.Y., Yellepeddi, R., and White,
AISC (2016a), Code of Standard Practice for Steel Build- D.W. (2018), “Assessment of I-Section Member LTB
ings and Bridges, ANSI/AISC 303-16, American Institute Resistances Considering Experimental Tests and Practi-
of Steel Construction, Chicago, Ill. cal Inelastic Buckling Design Calculations,” Engineering
Journal, AISC, Vol. 55, No. 1, pp. 15–44.
AISC (2016b), Specification for Structural Steel Buildings,
ANSI/AISC 360-16, American Institute of Steel Con- Subramanian, L. and White, D.W. (2017), “Resolving the
struction, Chicago, Ill. Disconnect between Lateral Torsional Buckling Experi-
mental Tests, Test Simulations and Design Strength
Hartmann, J.L. (2005), “An Experimental Investigation of
Equations,” Journal of Constructional Steel Research,
the Flexural Resistance of Horizontally Curved Steel
Vol. 128, pp. 321–334.
I-Girder Systems,” Doctoral Dissertation, University of
Maryland, College Park, Md. Toğay, O. and White, D.W. (2018), “Toward the Recogni-
tion of Unaccounted for Flange Local Buckling and Ten-
Lindapter (2019), Steel Connections Catalog, Lindapter
sion Flange Yielding Resistances in the ANSI/AISC 360
USA, http://www.lindapter.com/ (April 22).
Specification,” Proceedings of the SSRC Annual Stability
Mandal, P. and Calladine, C.R. (2002), “Lateral-Torsional Conference, Baltimore, Md.
Buckling of Beams and the Southwell Plot,” Interna-
White, D.W., Toğay, O., Slein, R., and Jeong, W.Y. (2020),
tional Journal of Mechanical Sciences, Vol. 44, No. 12,
“SABRE2-V2,” white.ce.gatech.edu/sabre (January 17).
pp. 2,557–2,571.
Yarimci, E., Yura, J.A., and Lu, L.W. (1967), “Techniques
MBMA (2018), Metal Building Systems Manual, Metal
for Testing Structures Permitted to Sway,” Experi-
Building Manufacturers Association, Cleveland, Ohio.
mental Mechanics, Vol.  7, No.  8, Reprint, Fritz Labo-
McGuire, W., Gallagher, R., and Ziemian, R.D. (2000), ratory Reports, Paper 112, http://preserve.lehigh.edu/
Matrix Structural Analysis, 2nd Ed., Wiley, New York, engr-civil-environmental-fritz-lab-reports/112.
N.Y.
Ziemian, R.D. (Ed.) (2010), Guide to Stability Design Cri-
Nethercot, D.A. and Trahair, N.S. (1976), “Lateral Buckling teria for Metal Structures, 6th Ed., John Wiley & Sons,
Approximations for Elastic Beams,” Structural Engineer- Inc., Hoboken, N.J.
ing, Vol. 54, No. 6, pp. 197–204.
Prado, E.P. and White, D.W. (2015), “Assessment of Basic
Steel I-Section Beam Bracing Requirements by Test Simu-
lation,” Structural Engineering, Mechanics and Materials
Report, School of Civil and Environmental Engineering,
Georgia Institute of Technology, Atlanta, Ga.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 291

292 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021
Steel Structures Research Update

Structural Fire Engineering

JUDY LIU

INTRODUCTION (associate project leader), Matthew Bundy, Brian Story,

Anthony Chakalis, Philip Deardorff, Selvarajah Ramesh,
A cross the United States, researchers are making exciting
discoveries and advances in structural fire engineer-
ing. Eleven of the leading scholars in the field are featured.
Xu Dai, and William Grosshandler. Team members Choe,
Ramesh, Dai, Hoehler, and Bundy won the Best Paper
Award at the 11th International Conference on Structures
Brief research highlights are organized by the topic areas
in Fire (SiF2020) for their paper, “Experimental Study on
of behavior and design of steel and composite structures
Fire Resistance of a Full-Scale Composite Floor Assembly
for fire, fire following earthquakes, and performance-based
in a Two-Story Steel Framed Building” (Choe et al., 2020b).
fire engineering. For each individual, related steel and fire
Dr. Choe’s honors also include the Department of Com-
research is also noted. Meanwhile, due to timing and cir-
merce Gold Medal in 2019, a group award for development
cumstances, there are structural fire engineering research-
of unique measurement capabilities for research at NIST’s
ers who do not appear in this article. You may find some of
National Fire Research Laboratory (NFRL).
their structural steel research in past articles or perhaps a
The research seeks to fill gaps in knowledge in the
future research update.
behavior of steel-concrete composite floors and beams with
simple shear connections subjected to fire. Previous studies
BEHAVIOR AND DESIGN OF STEEL AND were limited with regard to beam span, boundary condi-
COMPOSITE STRUCTURES FOR FIRE tions for the concrete on metal deck slabs, and other details
for the composite system (Choe et al. 2020a). Four 42-ft
Dr. Lisa Choe, Dr. Venkatesh Kodur, Dr. M.Z. Naser, Dr.
partially composite beam specimens were explored in the
Erica Fischer, and Dr. Amit Varma are contributing in
first phase of the work. Type of shear connection (welded-
various ways to the behavior and design of steel and com-
bolted double angle or shear tab) and slab continuity over
posite structures for fire. A two-story steel framed build-
the girder were the main parameters. The double angles
ing specimen is used for fire tests of full-scale composite
had greater rotational ductility than the shear tabs in fire
floor assemblies. System behavior and onset of collapse are
but could fail due to axial restraints in the cooling phase.
investigated in a numerical study of braced frame build-
The shear tabs were sensitive to the slab continuity in both
ings. Machine learning is used for autonomous evaluation
the heating and cooling phases (Choe et al. 2020a). In the
of fire resistance. Validated modeling techniques are used
second phase, 20-ft × 30-ft composite floor assemblies are
to study fire and fire following earthquake behavior of
exposed to natural gas-fueled compartment fires simulat-
steel moment-resisting frames. Comprehensive design pro-
ing standard fire environments [Figure 1(a)]. Gravity loads
cedures are developed for a new concrete-filled steel plate
are applied by frames connected to hydraulic actuators in
shear wall system.
the basement; adjacent bays are loaded with water-filled
drums [Figure 1(b)]. Test variables include area and type of
Fire Resistance of a Full-Scale Composite
steel reinforcement in composite floor slabs; fire protection
Floor Assembly
on exposed beams; and thermal restraints provided by the
Under way at the National Institute of Standards and adjacent bays, such as bay beam framing, beam-end connec-
Technology (NIST) is a series of four compartment fire tions, and slab continuity. In the first experiment, the fire
experiments on a full-scale two-story steel-framed build- test compartment represented an edge bay on the first floor
ing with composite floor slabs. Research structural engi- of the building. Prescriptive details of the floor specimen
neer Dr. Lisa Choe is leading the team: Matthew Hoehler complied with a 2-hr fire resistance rating. Connections
were shear tabs, and the minimum shrinkage reinforcement
was used in the floor slab. The reinforcement ruptured
before any membrane action could be developed, leading
to integrity failure of the floor slab prior to the specified
Judy Liu, PhD, Research Editor of the AISC Engineering Journal, Professor,
Oregon State University, School of Civil and Construction Engineering, Cor-
rating period (Choe et al., 2020b). Slab reinforcement and
vallis, Ore. Email: [email protected] fire performance of composite floor systems were explored
further with the second test in 2021 (Choe et al., 2021).

ISSN 0013-8029 ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 293

Design and Construction of Fire-Resistant Structures such as realistic loading, restraint and fire exposure scenar-
ios which can significantly influence the onset of instabil-
Dr. Venkatesh Kodur contributes practical structural fire
ity in the structure” (Venkatachari and Kodur, 2020). The
engineering tools and strategies founded on comprehensive
analysis of a 10-story braced frame building was conducted
numerical simulations and case studies. Dr. Kodur, Univer-
for different fire scenarios. The likelihood of collapse was
sity Distinguished Professor in Civil and Environmental
examined for parameters including fire location, tempera-
Engineering at Michigan State University, also serves as
ture and duration of the fire exposure, and number of com-
the Director of the Centre on Structural Fire Engineering
partments with fire exposure. Additional details can be
and Diagnostics. Dr. Kodur’s research includes mitigation
found in Venkatachari and Kodur (2020).
of fire-induced collapse of steel framed structures, utiliz-
ing comprehensive numerical models, and incorporating
Autonomous Evaluation of Concrete-Filled
the development of temperature-induced strength degrada-
Tube Columns
tion and instability in critical structural members with the
progression of horizontal and vertical fire spread in differ- Performance of structures under extreme conditions, arti-
ent compartments and stories. One example is a simplified ficial intelligence, and smart materials and structures are
approach for evaluating shear degradation in fire-exposed core interests for Dr. M.Z. Naser, an Assistant Professor in
steel and composite beams (Kodur and Naser, 2018). Fire Civil Engineering at Clemson University. Investigation into
safety of steel and composite bridges is another focus area; a temperature-induced moment-shear interaction (Naser and
specific aim is to develop practical tools for identifying vul- Kodur, 2020) and performance evaluation of concrete-filled
nerable bridges and developing mitigation strategies. Hon- tubes through machine learning (Naser et al., 2021) are just
ors include a 2018 Literati Award for Best Research Paper, some of Dr. Naser’s recent works. Honors include the 2018
“Effect of Local Instability on Fire Response of Steel Gird- Literati Award for Best Research Paper with Dr. Kodur.
ers” (Kodur and Naser, 2017) and a 2019 Outstanding Paper Dr. Naser has published extensively in structural fire engi-
Award at the 6th International Conference on Applications neering and machine learning, with a focus on “intelligent
in Structural Fire Engineering, Singapore, for “Evaluating tools that can accurately evaluate fire resistance and iden-
Fire Resistance of Composite Box Bridge Girders” (Zhang tify damage mechanisms in structures” and an eye toward
and Kodur, 2019). Dr. Kodur has also co-authored a text- “technologies to realize autonomous and self-diagnosing
book, Structural Fire Engineering. This comprehensive structures that can facilitate safe post-fire inspections and
book explains codes and standards; high-temperature mate- timely repairs” (Naser, 2020). His publications include the
rial properties; and behavior and strategies for enhancing Structural Fire Engineering textbook co-authored with Dr.
fire resistance of steel, concrete, composite, and timber Kodur (Kodur and Naser, 2020) and the Handbook of Cog-
structures (Kodur and Naser, 2020). nitive and Autonomous Systems for Fire Resilient Infra-
The work on steel and composite structures includes structures (Naser and Corbett, 2022).
system-level response of braced frame structures in fire Briefly highlighted here is Dr. Naser’s work on the struc-
scenarios (Venkatachari and Kodur, 2020). This work tural response evaluation of concrete-filled tube (CFT) col-
sought to fill knowledge gaps with respect to system behav- umns using machine learning (Naser et al., 2021). Dr. Naser
ior and the onset of collapse (Figure 2). The literature shows and collaborators noted the benefits of CFTs for building
that “limited attention is given to the effect of parameters design and construction, including strength, ductility, and

Fig. 1.  (a) Compartment fires with natural gas burners and (b) floor loading.

294 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

inherent fire resistance. They also noted that the adoption “These algorithms incorporate a supervised learning pro-
of high-strength materials might be inhibited by the range cess that mimics the natural selection process (i.e., Darwin-
of applicability of current code provisions and supporting ian evolution) to express hidden relations between a number
test data. The researchers used artificial intelligence (AI) of factors … The main advantage of these approaches over
to analyze a database of 3103 CFT columns tested under traditional soft computing techniques is their capability
different loading conditions and with a range of material to produce predictive expressions without relying on past
and geometric properties. Circular and square or rectan- formula or relationship” (Naser et al. 2021). As shown in
gular cross sections, with a maximum outside dimension the measured versus predicted capacity graphs in Figure 3,
of 40  in., had been tested under concentric and eccentric the predictions were generally better than those obtained
loading. The research team used genetic algorithms (GA) using AISC  360 (AISC,  2016), Eurocode  4 (CEN,  2009),
and gene expression programming (GEP) to develop mod- and AS 2327 (ASI, 2017). Additional details of this study
els capable of predicting the CFT column load capacities. can be found in Naser et al. (2021).

Fig. 2.  Deformed configuration of the building at the onset of fire-induced progressive collapse (time = 154 min).

  
(a)  AISC 360 (b)  AISC 360, Eurocode 4, and AS 2327

Fig. 3.  Measured versus predicted capacities for circular CFT columns under eccentric
loading; genetic algorithms (GA) and genetic expression programming (GEP) predictions.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 295

Development of Fire Modeling Capabilities structural fire engineering, and the modules are designed to
in OpenSees be integrated into courses across civil engineering under-
graduate curricula.
Simulating the behavior of steel and composite structures in
Simulation of fire performance of steel-frame build-
fire, educating the next generation of structural engineers,
ings is a theme area with multiple facets. Dr. Fischer’s
and designing simple connections for fire conditions are
research includes simulation of post-earthquake fire behav-
three primary themes for Dr. Erica Fischer. Dr. Fischer is
ior of industrial facilities, benchmarking of structural fire
an Assistant Professor in the School of Civil and Construc-
engineering modeling to large-scale experimental tests,
tion Engineering at Oregon State University. Recognized
and development of OpenSees capabilities to simulate
locally and nationally for research and teaching excellence,
steel gravity framing systems in fire. Some work to date
Dr. Fischer’s awards include AISC’s Terry Peshia Early
includes use of validated OpenSees modeling techniques
Career Faculty Award (2021). The American Society of
to investigate fire and fire following earthquake behavior
Civil Engineers (ASCE) honored Dr. Fischer with the 2019
of a three-story, three-bay moment-resisting frame (Mad-
Collingwood Prize for her paper, “Experimental Evalua-
dalozzo and Fischer, 2020). Five validation studies were
tion of Single-Bolted Lap Joints at Elevated Temperatures”
used to test component and system behavior, considering
(Fischer and Varma, 2018). Her more recent work on con-
temperature, forces, and displacements. Parameters for the
nections includes advanced connection modeling, leading a
fire and fire following earthquake investigations included
working group within one of AISC’s technical committees
locations of the compartment fire and the fuel load (Fig-
to synthesize numerical and experimental data on connec-
ure 4). Additional details can be found in Maddalozzo and
tion performance in fires and developing design provi-
Fischer (2020).
sions for connections exposed to various fire conditions.
Dr. Fischer also led a collaborative research review on the
SpeedCore—Concrete-Filled Composite Steel Plate
fire behavior of steel simple connections, identifying gaps
Shear Wall (CF-CPSW) Core
in knowledge and inconsistencies in testing methodologies
and reporting of results (Fischer et al., 2021). Education on Dr. Amit Varma is well known for research on the behavior
structural fire engineering extends past Dr. Fischer’s class- and analysis of steel and composite structures under fire
room and to development of teaching modules with the sup- loading and the development of design provisions for fire
port of AISC and the Structural Engineering Institute (SEI). and collapse resistance. Dr. Varma is the Karl H. Kettel-
The scope ranges from prescriptive fire protection design to hut Professor in the Lyles School of Civil Engineering at

Fig. 4.  Locations of compartment fires for moment-resisting frame study.

296 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

Purdue University and the Director of the Bowen Labora- plate shear wall system with cost, safety, and construction
tory for Large-Scale Civil Engineering Research. While schedule benefits, among others. An increased speed of
seeking to fill knowledge gaps in the fundamental behavior construction is partly because no fireproofing is required
of structural steel components and assemblies, Dr. Varma for SpeedCore walls at least 18  in. thick (AISC, 2021).
pioneered an innovative method for realistic structural fire Dr. Varma’s selections for 2017 and 2020 AISC Special
tests without the traditional gas furnace. Replication of this Achievement Awards and the 2021 AISC T.R. Higgins Lec-
unique test method at several structural engineering labora- tureship Award were based on his collection of work on the
tories across the world is just one indicator of Dr. Varma’s composite shear walls. More information about this hybrid
global impact in this field. His research group has produced core system, how it is designed and built, and how it works
numerous works advancing the state of the art in behavior, in fire can be found at AISC’s SpeedCore webpage (AISC,
analysis, and design for fire. Their achievements include 2021). Those looking for more information on the structural
publications on the fire behavior of steel and composite fire engineering and design of SpeedCore walls can view a
columns, composite floor assemblies, composite beams, series of recorded presentations on the “SpeedCore-Fire-
and steel connections (e.g., Hong and Varma, 2010; Well- Research” YouTube playlist (Figure 5; Varma, 2021). Com-
man et al., 2011; Selden et al., 2016; Fischer and Varma, puter programs to analyze and design SpeedCore walls for
2018). Dr. Varma’s group has also developed and bench- fire are available through an online repository (Varma and
marked detailed 3D nonlinear finite element models for Anvari, 2020).
predicting the observed experimental behavior of steel con-
nections and of steel and composite columns, floor assem-
FIRE FOLLOWING EARTHQUAKES
blies, and beams (e.g., Agarwal and Varma, 2011; Selden
and Varma, 2016; Choe et al., 2017; Fischer and Varma, Dr. Rachel Chicchi, Dr. Hussam Mahmoud, and Dr. Maria
2017). The benchmarked modeling approaches have been Garlock are advancing methods for fire following earth-
used to investigate the fire behavior, resistance, and overall quake (FFE) evaluations. Three-dimensional nonlinear
collapse of 3D steel and composite building structures (e.g., finite element building models explicitly include details of
Agarwal and Varma, 2014; Agarwal et al., 2014; Fischer et the gravity framing system and are used to compare dif-
al., 2019). The work has also been extended to evaluation ferent seismic damage–fire scenarios. A hybrid simulation
of multi-hazard behavior and design of steel building struc- framework incorporates thermal loads and seismic damage
tures under post-earthquake fire loading (e.g., Bhardwaj et in the physical substructure. The seismic performance of
al., 2019; Alasiri et al., 2020). nonstructural components (e.g., sprinklers) is incorporated
Dr. Varma’s most recent accolades recognize his achieve- into post-earthquake fire evaluations of steel frames.
ments in developing SpeedCore, a concrete-filled steel

Fig. 5.  Screenshot of video presentation on SpeedCore fire resistance experiments (Varma, 2021).

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 297

Post-Earthquake Fire Assessment model accounted for various limit states and failure modes
resulting from inelastic deformations, local and global
Post-earthquake fire assessment, critical temperature, sim-
instability, and connection damage/failure at elevated tem-
ple methods of analysis for fire-resistant design, and con-
peratures, while incorporating the effects of temperature
nections are among Dr. Rachel Chicchi’s topics of study.
on steel and concrete material properties” (Alasiri et al.,
Dr. Chicchi is an Assistant Professor in the University of
2020). Buildings with different levels of seismic damage
Cincinnati’s Civil and Architectural Engineering and Con-
were subjected to various fire scenarios. Shown in Figure 6
struction Management department. She has collaborated
is a fifth-story interior compartment fire and plots of the
with Dr. Fischer and Dr. Choe on a research review for fire
vertical deformation and temperature versus time for the
behavior of simple shear connections (Fischer et al., 2021).
failed interior gravity column. The researchers discovered
Analysis of single-angle connections by her research group
that the seismic and fire behaviors were largely decoupled;
will address one of the knowledge gaps identified in the
the gravity columns were the most critical components in
review. Work continues on simple methods of analysis to
the post-earthquake fire. Additional details and design rec-
determine demands on the structure at elevated tempera-
ommendations can be found in Alasiri et al. (2020).
tures (Chicchi and Varma, 2021). One extension of the work
is incorporation of local buckling for wide-flange columns
Multi-Hazard Simulation and Testing
into the simple equations. Another is an investigation into
the applicability of the column flexural buckling equa- A primary research thrust for Dr. Hussam Mahmoud is the
tion for high-strength steel. In a separate study, important assessment and development of resilient and sustainable
parameters affecting the critical temperatures of axially structural systems subjected to multiple hazards such as fire,
loaded wide-flange columns were identified and a closed- blasts, earthquakes, and/or wind loading. Dr. Mahmoud
form equation proposed (Sauca et al., 2021). Meanwhile, is the George T. Abell Professor of Infrastructure in the
post-earthquake fire assessment has evolved past an initial Department of Civil Engineering at Colorado State Univer-
research review (Chicchi and Varma, 2017) into detailed sity. He is also Director of the Structural Laboratory at Col-
studies and methods (e.g., Chicchi and Varma, 2018). orado State. Among his numerous recognitions and honors
Dr. Chicchi’s post-earthquake fire research includes is the 2017 Terry Peshia Early Career Faculty Award from
methods for post-earthquake fire assessment of steel AISC. Dr. Mahmoud’s contributions include new structural
moment frame buildings (Alasiri et al. 2020). A detailed systems as well as new performance-based design and life-
numerical methodology has been proposed and illustrated cycle analysis frameworks. Methods proposed include an
with a 10-story perimeter steel moment frame building approach for assessing steel column response under fire
case study. The nonlinear finite element model of the 3D loads considering with various boundary conditions and
building structure explicitly included the gravity framing temperature profiles (Memari and Mahmoud, 2018a). Also
system, including connections and composite slabs. “The considered are factors such as second-order effects and

0 600
0 15 30 45 60 75 90 105 120 135

500
-20
Vertical Deformation (in.)

400
Temperature (oC)

-40

300

-60
200

-80
100

-100 0
  Time (minutes)
(a) (b)

Fig. 6.  (a) Response of building to fifth-story interior compartment fire and
(b) vertical deformation and temperature versus time for failed interior gravity column.

298 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

residual stresses. A proposed framework for performance- She also serves as the program director of the Architecture
based analysis of fires following earthquakes incorporates and Engineering Program, is the Head of Forbes College at
uncertainties associated with fire hazard, gravity loads, Princeton University, and is Associated Faculty in both the
passive fire protection, and earthquake intensity (Memari School of Architecture and the Program in Latin Ameri-
and Mahmoud, 2018b). can Studies. Of her many accolades, perhaps the most
A hybrid simulation framework was developed and illus- notable in structural steel is the 2016 T.R. Higgins Lecture-
trated with a test of a small-scale specimen (Memari et al., ship Award for her steel structure-fire interaction research
2020). The four-story special concentrically braced frame (Garlock, 2016). Other awards include the Bronze Award
(SCBF) building was 5 ft × 5 ft in plan, with four bays in for the Advancement of Arc Welded Design, Engineering
each direction. A first story column was the physical test and Fabrication, given by The James Lincoln Arc Welding
specimen, the substructure model. The rest of the planar Foundation (2003) and a first-place poster in the category
frame was the numerical integration model (Figure  7). of Steel Structures, Composite Structures, and Connections
Leaning columns in the model represented the gravity in the 5th International Conference for Structures in Fire in
frames. The framework was first tested using a numerical East Lansing, Michigan, in 2010. The poster, co-authored
substructure-numerical integration model. For the physical with former student Spencer Quiel, was titled “Stress-Based
substructure model, a small fire furnace with four radiative Equations  for Predicting the Buckling Capacity of Steel
heaters and a self-reacting steel frame was used for thermal Plates Exposed to Fire” (Quiel and Garlock, 2010).
loads, and residual interstory drifts were used to represent Dr. Garlock notes that her research on fire has evolved
different seismic damage levels in the frame. The level of into fire following earthquake (FFE) community-level
residual interstory drift proved to be a significant factor in studies, specifically considering deep-basin seismic effects
the fire performance of the column. Additional details and such as those found in Seattle. A paper on FFE scenario
commentary on future studies can be found in Memari et studies in Seattle is forthcoming (Coar et al. 2021). Not
al. (2020). published, but in preparation, is a paper that examines the
effects of different earthquake types on passive and active
Community Level Studies and Nonstructural fire protection systems, and consequential structural-fire
Components interaction performance. Meanwhile, an archetype nine-
story steel moment-resisting frame (MRF) structure is
From steel connections for fire to community resilience
modeled using OpenSees finite element analysis software
to the Creative Art of Structural and Civil Engineering
and subjected to a suite of basin (Seattle) and non-basin
(CASCE, https://casce.princeton.edu/), Dr. Maria Garlock
(Los Angeles) earthquake records as developed by the Uni-
is advancing structural engineering design and the educa-
versity of Washington M9 Project (Frankel et al., 2018).
tion of future engineers. Dr. Garlock is a Professor of Civil
The resulting peak floor accelerations and interstory drift
and Environmental Engineering at Princeton University.

Fig. 7.  Physical and numerical substructure, residual damage state.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 299

ratios are used as intensity measures for existing component Performance-Based Design of Passive Fire Protection
fragility curves of the active (sprinkler) and passive (com- for Floor Systems
partment divider) fire protection systems (Pali et al., 2018;
Dr. Spencer Quiel is known for his research on buildings
Soroushian et al., 2014). The seismic performance of these
subjected to extreme loads such as fire, blast, and progres-
nonstructural components modifies the single-structure fire
sive collapse. Dr. Quiel is an Associate Professor in Civil
spread model based on the University of Delaware Disaster
and Environmental Engineering at Lehigh University. His
Research Center (Lee and Davidson 2010a, 2010b; Li and
experience in analysis and design of structures to resist
Davidson, 2013). The results of the fire spread model are
extreme loads includes four years at Hinman Consulting
used to apply time-dependent thermal loads to the seismi-
Engineers. As a Department of Homeland Security Grad-
cally damaged OpenSees finite element structural model
uate Fellow, Dr. Quiel also spent two months of his doc-
for structural performance evaluation of the steel frame
toral studies at the Building and Fire Research Laboratory
(Figure 8).
(BFRL) at the National Institute of Standards and Technol-
ogy (NIST). Dr. Quiel was the 2016 recipient of AISC’s
PERFORMANCE-BASED Milek Fellowship for his research on “Performance-Based
FIRE ENGINEERING (PBFE) Design of Passive Fire Protection for Floor Systems in
Steel-Framed Buildings.”
Dr. Spencer Quiel, Dr. Negar Elhami-Khorasani, and Dr.
For the Milek Fellowship, Dr. Quiel’s research group is
Thomas Gernay are improving simulations for perfor-
building a framework for performance-based design and
mance-based fire engineering (PBFE). A coordinated
analysis of steel-concrete composite floor systems to resist
computational-experimental investigation supports a
fire. The development of this framework for steel-framed
framework for realistic models of composite floor systems
buildings, motivated by the lack of design and analysis
with different fire ratings. A cost-effective hybrid simula-
guidance, is founded on a coordinated computational-
tion incorporates realistic boundary conditions and system
experimental investigation. Methods for quantifying real-
as well as element level response. Experimental character-
istic restraint of floor systems and modeling tools are being
ization of advanced high-strength steels tests the range of
produced. The laboratory experiments include five large-
applicability of current provisions.
scale structural-fire tests of partially restrained, partially

Fig. 8.  Fire spread model for structural performance evaluation of steel frame.

300 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

composite steel floor beam assemblies and a number of systems and fire fragility functions for steel frame build-
small-scale, spray-applied fire resistive material (SFRM) ings (Elhami-Khorasani et al., 2019; Gernay et al., 2018).
tests. Parameters for the large-scale tests included passive More recent research includes “Probabilistic Models for
fire protection (i.e., unprotected, coated with SFRM) and Temperature Dependent Strength of Steel and Concrete”
level of axial restraint (Kordosky et al. 2020). The experi- and “Developing Real-Time Hybrid Simulation to Capture
mental results have been used for validation of “thermal Column Buckling in a Steel Frame under Fire” (Qureshi
and structural analysis methods of varying complexity to et al., 2020a, 2020b). Qureshi et al. (2020b) received the
capture the flexural response and failure of the fire-exposed best student paper award at the International Structures in
composite floor beam specimens. The goal of this model- Fire Conference in 2020. Dr. Elhami-Khorasani was also
ing effort is to develop accurate, conservative predictions honored in 2020 with AISC’s Terry Peshia Early Career
of fire-induced mechanics and limit states at reduced com- Faculty Award, recognizing her structural steel research,
putational cost, thus increasing the accessibility of these teaching, technical service, and community outreach.
methods to practicing engineers” (Drury et al., 2020). A Dr. Elhami-Khorasani seeks to advance performance-
major outcome of the research will be tools for translat- based fire engineering (PBFE) through the development
ing fire resistance ratings into realistic models of composite of methods and models for explicit evaluation of reliabil-
floor systems for performance-based design of steel build- ity and fire safety. Probabilistic material models for steel
ings. Results have been promising, as demonstrated by com- and concrete incorporate uncertainties in material behav-
parisons of the final deflected shapes of the modeled and ior at elevated temperatures (Qureshi et al. 2020a). Qureshi
tested floor beam assemblies (Figure 9). Additional details et al. (2020a) also studied the influence of model choice
of the experiments and modeling efforts can be found in on structural failure assessment of steel and concrete col-
Kordosky et al. (2020) and Drury et al. (2020). umns exposed to fire. This work was in collaboration with
Dr. Thomas Gernay from Johns Hopkins University, Dr.
Methods and Models to Advance Performance-Based Ruben Van Coile from Ghent University in Belgium, and
Fire Engineering Dr. Danny Hopkin from Olsson Fire & Risk and the Uni-
versity of Sheffield in the UK. Meanwhile, Dr. Elhami-
Fire and elevated temperatures on steel structures is a
Khorasani’s work on real-time hybrid simulation provides
primary focus for Dr. Negar Elhami-Khorasani, whose
a cost-effective method for obtaining realistic element-level
research encompasses resilient communities, performance-
response of the physical substructure by properly incor-
based design, reliability, and multi-hazard analysis. Dr.
porating system-level response and interfacing boundary
Elhami-Khorasani is an Assistant Professor in the Depart-
conditions during the fire event through a numerical sub-
ment of Civil, Structural and Environmental Engineering
structure (Qureshi et al., 2020b). The proposed method was
at the University at Buffalo. In Dr. Elhami-Khorasani’s
successfully applied to a benchmark study of a large-scale
portfolio of work are topics such as parametric stud-
moment frame with buckling of one tube column subjected
ies for performance-based fire design of composite floor

0 0

-25 -1

-50 -2

-75 -3

-100
Displacement (mm)

Displacement (in.)

-4
-125 -5
(a)  Post-test photo Test Data
-150 -6
BP-TT
-175 BF-TT -7
-200 BP-LM
-8
BF-LM
-225 -9
S-TT
-250 -10
0 20 40 60 80 100 120 140 160 180
Time (min)
(c)  Vertical displacement time histories:
(b)  Model at failure: S-TT   Test vs. all model iterations

Fig. 9.  Results of testing and modeling for a protected W12×26 composite beam.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 301

to fire and subsequent frame collapse (Figure  10). “The of “next-generation” materials for economy, safety, and
real-time hybrid fire simulation is able to capture the construction efficiency. Dual-phase (DP) and martens-
dynamic behavior at the time of column failure, for which itic (MS) steels were tested at elevated temperatures and
the obtained result is also applicable to progressive collapse under steady-state (66 specimens) and transient-state con-
applications in general” (Qureshi, 2021). ditions (22 specimens). Steady-state tests began with heat-
ing to a target temperature and then loading until fracture.
Characterization of New Steels at Elevated Transient-state conditions involved loading to target stress
Temperatures levels and then heating at a constant rate. Several materi-
als and a half a dozen different grades of steel were evalu-
Risk-based methodologies, numerical modeling, and char-
ated for strength, ductility, and stress-strain behavior under
acterization of materials at elevated temperatures form the
temperatures ranging from ambient to 1290°F. Figure  11
core of Dr. Thomas Gernay’s contributions to performance-
shows the test set-up and a set of AHSS specimens after the
based fire engineering (PBFE). Dr. Gernay is an Assistant
steady-state tests. “With increasing temperature, the speci-
Professor in the Department of Civil and Systems Engineer-
mens became softer and showed more ductile failures, with
ing at Johns Hopkins University. The founder and principal
increasing fracture strain and clear necking phenomena”
investigator for Johns Hopkins’ Multi-Hazard Resilient
(Yan et al., 2020). Retention factors and predictive mod-
Structures research group, Dr. Gernay actively develops
els were developed for the different materials. At elevated
SAFIR, a nonlinear finite element software for modeling
temperatures, the tested AHSS showed larger reductions in
the behavior of structures in fire, and conducts research on
properties than their lower-grade counterparts. Revisions
topics ranging from properties of advanced high-strength
are needed to expand the applicability of current provi-
steels at elevated temperatures (Yan et al., 2020, 2021a,
sions to AHSS. Additional details can be found in Yan et
2021b) to behavior of cold-formed steel structures during
al. (2020).
and after fire exposure (Batista Abreu et al., 2020, 2021).
Dr. Gernay’s research portfolio also includes fire fragility
functions, probabilistic material models, and performance- SUMMARY
based fire design and tensile membrane action in composite
Structural fire engineering researchers are making
structures (Qureshi et al., 2020a; Gernay et al., 2018; Ger-
advances in the broad topic areas of behavior and design
nay and Khorasani, 2020).
of steel and composite structures for fire, fire following
Dr. Gernay’s experimental materials characterization at
earthquakes, and performance-based fire engineering. Spe-
elevated temperatures continues to strengthen computer
cific topics include behavior of full-scale composite floor
simulation capabilities, especially with respect to behavior
assemblies in fire, system behavior and onset of collapse of
of new steels. This includes steels more commonly found
braced frame buildings, and machine learning for autono-
in the automotive industry. Yan et al. (2020) investigated
mous evaluation of fire resistance. Comprehensive design
advanced high-strength steels (AHSS), taking the recent
procedures are developed for composite core wall systems.
demand for high-performance steels in the major projects
Hybrid simulation frameworks incorporate thermal loads,
such One World Trade Center as a harbinger of the adoption

Fig. 10.  Substructuring configuration for the benchmark study.

302 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

seismic damage, and realistic boundary conditions. Vali- Agarwal, A. and Varma, A.H. (2011), “Design of Steel Col-
dated modeling techniques are being used to study fire and umns for Fire Loading Including Effects of Rotational
fire following earthquake behavior of steel frames; realistic Restraints,” Engineering Journal, AISC, Vol. 48, No. 4,
models of composite floor systems are developed, and dif- pp. 297–314.
ferent seismic damage–fire scenarios are explored. Materi- Agarwal, A. and Varma, A.H. (2014), “Fire Induced Progres-
als characterization expands modeling capabilities to new sive Collapse of Steel Building Structures: The Role of
steels. The seismic performance of nonstructural compo- Interior Gravity Columns,” Engineering Structures, Spe-
nents (e.g., sprinklers) is incorporated into post-earthquake cial Issue on Fire Analysis of Structures, Vol. 58, pp. 129–
fire evaluations of steel frames, and fire following earth- 140. http://dx.doi.org/10.1016/j.engstruct.2013.09.020
quake is investigated at a community level.
Alasiri, M., Chicchi, R., and Varma, A.H. (2020), “Post-
Earthquake Fire Behavior and Structural Performance-
ACKNOWLEDGMENTS Based Fire Design of a Steel Moment Frame Building,”
Journal of Constructional Steel Research, Vol. 177.
Many thanks to the featured researchers for their contribu-
tions. A very special thank you to Dr. Erica Fischer whose ASI (2017), Australian Composite Structures Standard, AS
excellent suggestion and coordination made this article pos- 2327, Australian Steel Institute, Sydney, Australia.
sible. Any findings or recommendations are those of the Batista Abreu, J.C., Vieira Jr., L.C.M., Moreno Jr., A.L.,
researchers and do not necessarily reflect the views of the Gernay, T., and Schafer, B.W. (2020), “Experiments on
sponsors. Load-Bearing Cold-Formed Steel Sheathed Studs at Ele-
vated Temperatures,” Thin-Walled Structures, Vol.  156,
REFERENCES 106968.
AISC (2016), Specification for Structural Steel Buildings, Batista Abreu, J.C., Vieira Jr., L.C.M., Gernay, T., and Scha-
ANSI/AISC 360-16, American Institute of Steel Con- fer, B.W. (2021), “Cold-Formed Steel Sheathing Con-
struction, Chicago, Ill. nections at Elevated Temperature,” Fire Safety Journal,
Vol. 123, 103358.
AISC (2021), “Speedcore,” https://www.aisc.org/why-steel/
innovative-systems/SpeedCore/#55717. Bhardwaj, S.R., Sener, K.C., and Varma, A.H. (2019),
“Multi-Hazard Investigation and Testing of Composite
Agarwal, A., Selden, K., and Varma A.H. (2014), “Stability
(SC) Wall Piers: Seismic and Thermal Loads,” Nuclear
Behavior of Steel Building Structures in Fire Conditions:
Engineering and Design, Vol.  348, pp.  121–130. https://
Role of Composite Floor System with Shear-Tab Con-
doi.org/10.1016/j.nucengdes.2019.03.026
nections,” Journal of Structural Fire Engineering, Spe-
cial Issue on Structures in Fire-2012 Conference, Vol. 5,
No. 2, June. http://dx.doi.org/10.1260/2040-2317.5.2.77

 
(a) (b)

Fig. 11.  (a) Test set-up and (b) a set of AHSS specimens after testing.

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 303

CEN (2009), Eurocode 4: Design of Composite Steel and Elhami-Khorasani, N., Gernay T., and Fang, C. (2019),
Concrete Structures, Comite Européen de Normalisation, “Parametric Study for Performance-Based Fire Design
Brussels, Belgium. of US Prototype Composite Floor Systems,” Journal of
Chicchi, R. and Varma, A.H. (2017), “Research Review: Structural Engineering, ASCE, Vol. 145, No. 5.
Post-Earthquake Fire Assessment of Steel Moment Frame Fischer, E., Agarwal, A., and Varma, A.H. (2019), “Per-
Buildings,” Advances in Structural Engineering. https:// formance-Based Structural Fire Engineering of Steel
doi.org/10.1177/1369433217711617 Building Structures: Design-Basis Compartment Fires,”
Chicchi, R. and Varma, A.H. (2018), “Assessment of Post- Journal of Structural Engineering, ASCE, Vol. 145, No. 9.
Earthquake Fire Behavior of a Steel MRF Building in https://doi.org/10.1061/(ASCE)ST.1943-541X.0002370
a Low Seismic Region,” International Journal of Steel Fischer, E., Chicchi, R., and Choe, L. (2021), “Review of
Structures, Vol.  18, No.  4, pp.  1470–1481. https://doi. Research on the Fire Behavior of Simple Shear Connec-
org/10.1007/s13296-018-0183-y tions,” Fire Technology, April.
Chicchi, R. and Varma, A.H. (2021), “Comparison of Sim- Fischer, E. and Varma, A.H. (2017), “Fire Resilience of
ple and Advanced Methods of Analysis in AISC 360 for Composite Beams with Simple Connections: Paramet-
Fire Resistant Structural Design,” Engineering Journal, ric Studies and Design,” Journal of Constructional
AISC [accepted pending revisions]. Steel Research, Vol.  128, pp.  119–135. http://dx.doi
Choe, L., Ramesh, S., Dai, X., Hoehler, M., and Bundy, .org/10.1016/j.jcsr.2016.08.004
M. (2020b), “Experimental Study on Fire Resistance of Fischer, E. and Varma, A.H. (2018), “Experimental Evalua-
a Full-Scale Composite Floor Assembly in a Two-Story tion of Single-Bolted Lap Joint at Elevated Temperatures,”
Steel Framed Building,” SiF 2020—The 11th Interna- Journal of Structural Engineering, ASCE, Vol. 144, No. 1.
tional Conference on Structures in Fire, Brisbane, Aus- https://doi.org/10.1061/(ASCE)ST.1943-541X.0001911
tralia, November 30–December 2. Frankel, A., Wirth, E., Marafi, N., Vidale, J., and Stephen-
Choe, L., Ramesh, S., Grosshandler, W., Hoehler, M., son, W. (2018), “Broadband Synthetic Seismograms for
Seif, M., Gross, J., and Bundy, M. (2020a), “Behavior Magnitude 9 Earthquakes on the Cascadia Megath-
and Limit States of Long-Span Composite Floor Beams rust Based on 3D Simulations and Stochastic Synthet-
with Simple Shear Connections Subject to Compartment ics, Part 1: Methodology and Overall Results,” Bulletin
Fires: Experimental Evaluation,” Journal of Structural of the Seismological Society of America, Vol.  108.5A,
Engineering, ASCE. https://doi.org/10.1061/(ASCE) pp. 2,347–2,369.
ST.1943-541X.0002627 Garlock, M. (2016), “T.R. Higgins Lecture: Get Fired Up:
Choe, L., Ramesh, S., Zhang, C., and Clifton, C. (2021), What Structural Engineers Should Know about Fire
“Behaviour of Composite Floor Assemblies Subject to Design,” Proceedings of NASCC: The Steel Confer-
Fire: Influence of Slab Reinforcement.” Proceedings of ence, April 13–15, Orlando, Fla. https://www.aisc.org/
2021 Eurosteel Conference, September 1–3, University of education/continuingeducation/education-archives/t.r.-
Sheffield, United Kingdom (in production). higgins-lecture-get-fired-up-what-structural-engineers-
Choe, L., Zhang, C., Luecke, W., Gross, J., and Varma, should-know-about-fire-design-n79/
A.H. (2017), “Influence of Material Models on Predict- Gernay, T., Elhami-Khorasani, N., and Garlock, M.E.M.
ing the Fire Behavior of Steel Columns,” Fire Technology, (2018), “Fire Fragility Functions for Steel Frame Build-
Vol.  53, No.  1, pp.  375–400. http://dx.doi.org/10.1007/ ings: Sensitivity Analysis and Reliability Framework,”
s10694-016-0568-4 Fire Technology, Vol.  55, 1175. https://doi.org/10.1007/
Coar, M., Sarreshtehdari, A., Garlock, M.E.M., and Elhami- s10694-018-0764-5
Khorasani, N. (2021), “Methodology and Challenges of Gernay, T., and Khorasani, N. E. (2020), “Recommenda-
Fire Following Earthquake Analysis: An Urban Com- tions for Performance-Based Fire Design of Composite
munity Study Considering Water and Transportation Net- Steel Buildings Using Computational Analysis,” Journal
works,” Natural Hazards, in press. of Constructional Steel Research, Vol. 166, 105906.
Drury, M.M., Kordosky, A.N., and Quiel, S.E. (2020), Hong, S. and Varma, A.H. (2010), “Predicting Column
“Structural Fire Resistance of Partially Restrained, Par- Buckling under Fire Loading Using Fundamental Sec-
tially Composite Floor Beams, II: Modeling,” Journal tion Behavior,” Journal of ASTM International, ASTM,
of Constructional Steel Research, Vol.  167. https://doi Vol. 7, No. 1. http://dx.doi.org/10.1520/JAI102311
.org/10.1016/j.jcsr.2020.10594

304 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021

Kodur, V.K.R. and Naser, M. (2017), “Effect of Local Insta- Naser, M.Z. and Corbett, G. (2022), Handbook of Cognitive
bility on Fire Response of Steel Girders,” PSU Research and Autonomous Systems for Fire Resilient Infrastruc-
Review Journal, Vol. 1, No. 2, pp. 170–179. tures, Springer, to be released in 2022.
Kodur, V.K.R. and Naser, M.Z. (2018), “Approach for Shear Naser, M.Z. and Kodur, V.K.R. (2020), “Temperature-Induced
Capacity Evaluation of Fire Exposed Steel and Compos- Moment-Shear Interaction in Steel Beams.” International
ite Beams,” Journal of Constructional Steel Research, Journal of Steel Structures. Vol.  20, pp.  1,540–1,551.
Vol. 141, pp. 91–103. https://doi.org/10.1007/s13296-020-00388-4
Kodur, V.K.R. and Naser, M.Z. (2020), Structural Fire Naser, M.Z., Thai S., and Thai, T. (2021), “Evaluating Struc-
Engineering, McGraw-Hill Professional (Education), tural Response of Concrete-Filled Steel Tubular Columns
New York, N.Y. ISBN 126012858X, 9781260128581 through Machine Learning,” Journal of Building Engi-
Kordosky, A.N., Drury, M.M., and Quiel, S.E. (2020), neering. https://doi.org/10.1016/j.jobe.2020.101888.
“Structural Fire Resistance of Partially Restrained, Par- Pali, T., Macillo, V., Terracciano, M.T., Bucciero, B., Fiorino,
tially Composite Floor Beams, I: Experiments,” Journal L., and Landolfo, R. (2018), “In-Plane Quasi-Static Cyclic
of Constructional Steel Research, Vol.  167, https://doi Tests of Nonstructural Lightweight Steel Drywall Par-
.org/10.1016/j.jcsr.2020.105945 titions for Seismic Performance Evaluation,” Earth-
Lee, S.W., and Davidson, R.A. (2010a), “Application of quake Engineering and Structural Dynamics, Vol.  47,
a Physics-Based Simulation Model to Examine Post- pp. 1,566–1,588. DOI: 10.1002/eqe.3031
Earthquake Fire Spread,” Journal of Earthquake Engi- Quiel, S.E. and Garlock, M.E.M. (2010), “Stress-Based
neering, Vol 14.5, pp.  688–705. ISSN: 13632469. DOI: Equations for Predicting the Buckling Capacity of Steel
10.1080/13632460903336936 Plates Exposed to Fire,” Structures in Fire—Proceedings
Lee, S.W. and Davidson, R.A. (2010b), “Physics-Based of the Sixth International Conference, SiF’10, East Lan-
Simulation Model of Post-Earthquake Fire Spread,” Jour- sing, Mich., June 2–4.
nal of Earthquake Engineering, Vol. 14.5, pp. 670–687. Qureshi, R. (2021), “Evaluation of Steel Columns under
ISSN: 13632469. DOI: 10.1080/13632460903336928 Fire: Real-Time Hybrid Testing and Reliability Assess-
Li, S. and Davidson, R.A. (2013), “Parametric Study of ment,” PhD Dissertation, University at Buffalo, Buffalo,
Urban Fire Spread Using an Urban Fire Simulation Model N.Y.
with Fire Department Suppression,” Fire Safety Journal, Qureshi, R., Elhami-Khorasani, N., and Sivaselvan, M.
Vol.  61, pp.  217–225. ISSN: 03797112; DOI: 10.1016/j (2020b), “Developing Real-Time Hybrid Simulation to
.firesaf.2013. 09.017. URL: http://dx.doi.org/10.1016/j Capture Column Buckling in a Steel Frame under Fire,”
.firesaf.2013.09.017 Proceedings of 11th International Conference on Struc-
Maddalozzo, W. and Fischer, E.C. (2020), “Post-Earthquake tures in Fire, Queensland, Australia, November 30–
Fire Performance of Steel Buildings,” Proceedings of December 2.
the 17th World Conference on Earthquake Engineering Qureshi, R., Ni, S., Elhami-Khorasani, N., Van Coile, R.,
(17WCEE), Sendai, Japan, September 13–18. Hopkin, D., and Gernay, T. (2020a), “Probabilistic Mod-
Memari, M. and Mahmoud, H. (2018a), “Design Formula- els for Temperature Dependent Strength of Steel and
tion for Critical Buckling Stress of Steel Columns Sub- Concrete,” Journal of Structural Engineering, ASCE,
jected to Nonuniform Fire Loads,” Engineering Journal, Vol. 146, No. 6.
AISC, Vol. 55, No. 2, pp. 89–108. Sauca, A., Chicchi, R., Zhang, C., and Choe, L. (2021),
Memari, M. and Mahmoud, H. (2018b), “Framework for a “Critical Temperature of Axially Loaded Steel Members
Performance-Based Analysis of Fires Following Earth- with Wide-Flange Shapes Exposed to Fire,” Engineering
quakes,” Engineering Structures, Vol.  171. https://doi Journal, AISC, Vol. 58, No. 1.
.org/10.1016/j.engstruct.2018.05.099 Selden, K., Fischer, E., and Varma, A.H. (2016), “Experi-
Memari, M., Wang, X., Mahmoud, H., and Kwon, O. (2020), mental Investigation of Composite Beams with Shear
“Hybrid Simulation of Small-Scale Steel Braced Frame Connections Subjected to Fire Loading,” Journal of Struc-
Subjected to Fire and Fire Following Earthquake,” Jour- tural Engineering, ASCE, Vol.  142, No.  2. http://dx.doi
nal of Structural Engineering, ASCE, Vol.  146, No.  1. .org/10.1061/(ASCE)ST.1943-541X.0001381, 04015118
https://doi.org/10.1061/(ASCE)ST.1943-541X.0002466 Selden, K. and Varma, A.H. (2016), “Composite Beams
Naser, M.Z. (2020), “Autonomous Fire Resistance Evalua- under Fire Loading: Numerical Modeling of Behavior,”
tion,” Journal of Structural Engineering, ASCE, Vol. 146. Journal of Structural Fire Engineering, Vol.  7, No.  2,
https://doi.org/10.1061/(ASCE)ST.1943-541X.0002641 pp. 142–157. http://dx.doi.org/10.1108/JSFE-06-2016-011

ENGINEERING JOURNAL / FOURTH QUARTER / 2021 / 305

Soroushian, S., Maragakis, E., Zaghi, A.E., Echevarria, Yan, X., Xia, Y., Blum, H.B., and Gernay, T. (2020), “Ele-
A., Tian, Y. and Filiatrault, A. (2014), “Comprehensive vated Temperature Material Properties of Advanced High
Analytical Seismic Fragility of Fire Sprinkler Piping Sys- Strength Steel Alloys,” Journal of Constructional Steel
tems,” MCEER 14.0002. Research, Vol. 174, 106299.
Varma, A., (2021), “SpeedCore-Fire-Research.” [Video Yan, X., Xia, Y., Blum, H. B., and Gernay, T. (2021a), “Post-
playlist.] YouTube. https://www.youtube.com/playlist?list Fire Mechanical Properties of Advanced High-Strength
=PLv1o8rALuecH5N59niiCmrSmi3w_SZeci Cold-Formed Steel Alloys,” Thin-Walled Structures,
Varma, A.H. and Anvari, A.T. (2020), “Fire Design of Vol. 159, 107293.
SpeedCore Walls and CFT Columns,” Purdue University Wellman, E., Varma, A.H., Fike, R., and Kodur, V., (2011),
Research Repository. doi:10.4231/JQHG-1N93 “Experimental Evaluation of Thin Composite Floor
Venkatachari S. and Kodur, V.K.R (2020), “System Level Assemblies under Fire Loading,” Journal of Structural
Response of Braced Frame Structures under Fire Engineering, Special 9/11 Commemorative Issue, ASCE,
Exposure Scenarios,” Journal of Constructional Steel Vol.  37, No.  9, pp.  1002–1016. http://dx.doi.org/10.1061/
Research, Vol. 170, 106073, pp. 1–22. (ASCE)ST.1943-541X.0000451
Yan, X., Abreu, J.C.B., Glauz, R.S., Schafer, B.W., and Ger- Zhang, G. and Kodur, V.K.R. (2019), “Evaluating Fire Resis-
nay, T. (2021b), “Simple Three-Coefficient Equation  for tance of Composite Box Bridge Girders,” Proceedings:
Temperature-Dependent Mechanical Properties of Cold- 6th International Conference on Applications of Struc-
Formed Steels,” Journal of Structural Engineering, tural Fire Engineering (ASFE’19), pp. 1–8, Singapore.
Vol. 147, No. 4, 04021035.

306 / ENGINEERING JOURNAL / FOURTH QUARTER / 2021


Guide for Authors
Scope Engineering Journal is dedicated to the improvement and
advancement of steel construction. Its pages are open to all who
wish to report on new developments or techniques in steel design,
research, the design and/or construction of new projects, steel
fabrication methods, or new products of significance to the uses of
steel in construction. Only original papers should be submitted.
General Papers intended for publication should be submitted by email
Margaret Matthew, editor, at [email protected].
The articles published in the Engineering Journal undergo
peer review before publication for (1) originality of contribution;
(2) technical value to the steel construction community; (3) proper
credit to others working in the same area; (4) prior publication of the
material; and (5) justification of the conclusion based on the report.
All papers within the scope outlined above will be reviewed by
engineers selected from among AISC, industry, design firms, and
universities. The standard review process includes outside review by
an average of three reviewers, who are experts in their respective
technical area, and volunteers in the program. Papers not accepted
will not be returned to the author. Published papers become the
property of the American Institute of Steel Construction and are
protected by appropriate copyrights. No proofs will be sent to
authors. Each author receives three copies of the issue in which his
contribution appears.
Manuscripts Manuscripts must be provided in Microsoft Word format. Include a
PDF with your submittal so we may verify fonts, equations and figures.
View our complete author guidelines at aisc.org/ej.
Smarter. Stronger. Steel.
American Institute of Steel Construction
130 E Randolph St, Ste 2000, Chicago, IL 60601
312.670.2400 | aisc.org/ej