Engineering

Journal
First Quarter 2020 | Volume 57, No. 1

3 Dimensional Tolerances and Length

Determination of High-Strength Bolts
James A. Swanson, Gian Andrea Rassati,
and Chad M. Larson

33 A Reliability Study of Joints with Bolts

Designed with Threads Excluded but
Installed with Threads Not Excluded
James A. Swanson, Gian Andrea Rassati,
and Chad M. Larson

55 Reexamination of Shear Lag in HSS Tension

Members with Side Gusset Plate Connections
Akashdeep A. Bhat and Patrick J. Fortney

69 Continuous Bracing Requirements for

Constrained-Axis Torsional Buckling
Mark D. Denavit, William P. Jacobs V,
and Todd A. Helwig
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
Jack Klimp
Chairman
Stephen Knitter
Vice Chairman
Edward Seglias
Secretary/Legal Counsel
Charles J. Carter, SE, PE, PhD
President
Scott L. Melnick
Senior Vice President
Lawrence F. Kruth, PE
Vice President
Tabitha S. Stine, SE, PE
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
2000, Chicago, IL 60601.
Periodicals postage paid at Chicago, IL and additional mailing offices.
Postmaster: Send address changes to Engineering Journal in care of the
American Institute of Steel Construction, 130 E Randolph Street, Suite 2000, Subscriptions: [email protected], 312.670.2400
Chicago, IL 60601.
Copyright 2020 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.
Letter from the Editor
Dear Readers,

Hello and Happy New Year! As we head into the new year, I would like to take this opportunity
to recognize all of the hard work of our reviewers, last year and every year. Their contributions
are essential to the success of the Journal as we continue to strive to bring you the very best
articles and information in the steel construction industry. A list of our 2019 reviewers is posted
on the AISC website at www.aisc.org/ej.

Is there a steel design topic you would like to see in EJ? We are always looking for ideas for
papers. Authors interested in submitting papers should visit our website at www.aisc.org/ej for
author guidelines and submittal information.

Best wishes for a healthy and happy 2020!

Sincerely,

Margaret A. Matthew, P.E.

Editor

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 1

2 / ENGINEERING JOURNAL / FIRST QUARTER / 2020
Dimensional Tolerances and Length Determination of
High-Strength Bolts
JAMES A. SWANSON, GIAN ANDREA RASSATI, and CHAD M. LARSON

ABSTRACT
Structural engineers and detailers are often removed from the process of manufacturing bolts and, thus, the tolerances and variances that
go along with common manufacturing processes. While this does not represent a problem in most cases, being familiar with the manufac-
turing processes and tolerances associated with high-strength bolts can help prevent some problems from occurring before the design
process even begins, particularly when shorter bolt lengths are needed. This lack of familiarity, in some circumstances, might lead to mis-
taken assumptions regarding the location of the shear plane relative to the threads of the bolt, which may lead to incorrect designs. While an
engineer might presume that bolt strength would not control in such short grips, this paper will discuss the cases in which this can become
an issue. This paper summarizes the major variances between nominal and actual dimensions, evaluates some of the consequences that
those variances can have on design, presents solutions to those issues, and culminates with a proposed design procedure for proper length
determination of high-strength bolts with several illustrative examples.

Keywords:  Structural bolt, high-strength bolt, fastener, A325, A490, F1852, F2280, F3125, F3148, threads excluded, ASME B18.2.6.

INTRODUCTION increasing the design strength of the bolt. The paper focused
on the length tolerance of the bolts and the thickness tol-
B olts, like any other manufactured product, are speci-
fied by nominal values but have acceptable variances
or tolerances from those baseline values that are needed
erances of the washers and nuts that are used to complete
the bolting assembly, and culminated with a series of design
tables for commonly used bolt sizes that are useful tools for
during manufacturing. Because the governing standards for
structural engineers and detailers. Carter’s paper identified
high-strength bolts are maintained by ASTM International,
that the tolerance on bolt length is more critical than the
which references several American Society of Mechanical
washer and nut thickness tolerances and that while addi-
Engineers (ASME) standards, the details of bolt manufac-
tional tolerances on the shank length and thread transition
turing are often steps removed from the day-to-day atten-
region of bolts were considered, they were thought at the
tion of most structural engineers and connection detailers.
time to be small enough so as to be inconsequential.
The objective of this paper is to provide a summary of the
dimensional tolerances that are associated with the manu-
facture of bolts, evaluate some of the consequences that BOLT, NUT, AND WASHER GEOMETRY AND
those tolerances can have on structural steel design, and MANUFACTURING TOLERANCES
present solutions to those issues.
High-strength structural bolts are required by ASTM
Carter, in a 1996 Engineering Journal paper, presented
F3125/3125M-15a (2015) and F3148-17a (2017) to conform
an analysis of tolerances associated with high-strength bolt-
to the ASME Standard B18.2.6-19 (2019). Two types of
ing with an emphasis on developing expedient methods
dimensions, shown in Figures 1 and 2, are used in the lat-
of determining bolt length and when the threads of a bolt
ter standard: control dimensions and reference dimensions.
can be excluded from the shear plane of a bolted joint, thus
Control dimensions are those dimensions that are used dur-
ing manufacture to ensure quality control and conformance
with standards. Reference dimensions, on the other hand,
James A. Swanson, PhD, Associate Professor, Department of Civil and Archi- are dimensions that are typically provided for information
tectural Engineering and Construction Management, University of Cincinnati, only or for the purpose of calculating control dimensions,
Cincinnati, Ohio. Email: [email protected] (corresponding) and not for quality control or demonstration of conformance
Gian Andrea Rassati, PhD, Associate Professor, Department of Civil and to standards.
Architectural Engineering and Construction Management, University of Cin-
cinnati, Cincinnati, Ohio. Email: [email protected]
Chad M. Larson, President, LeJeune Bolt Co., Burnsville, Minn. Email:
Bolt Diameter
[email protected]
ASTM F3125 high-strength structural fasteners are gener-
ally available in diameters ranging from 2 in. to 12 in. in
Paper No. 2018-15 8‑in.-diameter increments, with the exception of 1a‑in.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 3

 Table 1.  Diameters and Tolerances for Standard Size High-Strength Bolts (ASME B18.2.6-19)
Nominal Bolt Maximum Body Minimum Body Permitted Swell Max Diameter plus
Diameter, E or db (in.) Diameter (in.) Diameter (in.) or Fin (in.) Max Swell (in.)
   2 0.515 0.482 0.030 0.545
   s 0.642 0.605 0.050 0.692
   ¾ 0.768 0.729 0.050 0.818
   d 0.895 0.852 0.060 0.955
1 1.022 0.976 0.060 1.082
18 1.149 1.098 0.060 1.209
1¼ 1.277 1.223 0.060 1.337
1a 1.404 1.345 0.090 1.494
12 1.531 1.470 0.090 1.621

bolts, which are not widely produced. ASTM F3148 high- a swell or fin under the head of the fastener that may occur
strength bolts are currently available in diameters ranging during manufacturing.1 As can be deduced from the table, a
from s in. to 18 in. in 8‑in. increments. The diameter of a bolt produced at the maximum body diameter with the max-
bolt, specified as E in Figures 1 and 2 and in ASME B18.2.6, imum permitted swell may have a final measured diameter
but more commonly referred to as db in structural engineer-
ing contexts, is a control dimension and has an over/under
tolerance ranging from approximately 0.015  in. for small
1 Swells and fins result primarily during hot forging of fastener heads. Hot
bolts to approximately 0.030 in. for large bolts. Actual val- forging is used in the manufacture of relatively few common bolt sizes,
ues of diameters permitted are shown in Table  1. In addi- however—mostly for larger diameter or longer bolts. Bolts up to approxi-
tion to the diameter tolerance, an allowance is included for mately 6 in. long are typically cold formed.

R R
E

LB Y LB Y
LG LT LG LT
H L H L

(a) Hex head with cut threads (b) Hex head with rolled threads

U U
R R
C

E
D

LB Y
LB Y
LG LT LS
LG LT LS H L
H L

(c) Twist-off spline drive with button head (d) Twist-off spline drive with hex head

Fig. 1.  Geometry of ASTM F3125 bolts (ASME B18.2.6-19).

4 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 Table 2.  Tolerances on Bolt Length
Nominal Bolt Diameter, E or db (in.) Nominal Bolt Length, L ≤ 6 (in.) Nominal Bolt Length, L > 6 (in.)
2 +0.00, −0.12 +0.00, −0.19
s +0.00, −0.12 +0.00, −0.25
w to 1 +0.00, −0.19 +0.00, −0.25
18 to 12 +0.00, −0.25 +0.00, −0.25

that is more than z in. larger than its nominal diameter and, from one manufacturer to another and even one machine to
in the case of a 12‑in. bolt, is nearly 8 in. larger than the another.
nominal diameter. As Shaw points out (2015), this, in part, ASME B18.2.6 provides specifications for bolts ranging
led to the increase in the size of standard holes for 1‑in.- in length from 12  in. to 10  in. in 4‑in.-long increments.
diameter and larger bolts in the 2016 AISC Specification However, the AISC Steel Construction Manual (2017) states
(AISC, 2016) and in the 8th edition of the AASHTO LRFD that high-strength bolts are generally furnished in length
Specification (AASHTO, 2017) and has been approved for increments of 4‑in. only up to a length of 5  in. and then
inclusion in the next edition of the RCSC Specification, in length increments of 2‑in. for longer bolts. The RCSC
expected to be published in 2020. Specification (2015), however, notes that the transition from
4‑in. to 2‑in. increments occurs at a length of 6 in. In prac-
Bolt Length, Shank Length, and Thread Length tice, availability of bolts of specific length is a function of
several variables, including manufacturing methods, tool-
The length of the bolt, specified as L in Figure 1, is a con-
ing, and market factors. A good rule of thumb is that bolt
trol dimension and for bolts without splines (ASTM F3125
lengths up to four diameters are generally stocked for all
Grades A325 and A490) is measured parallel to the axis of
diameters up to 12  in. and bolt lengths up to eight diam-
the bolt from the underside of the head of the bolt—the bear-
eters are commonly stocked for the more commonly used
ing surface of the head—to the end of the bolt. For bolts with
diameters of w in. to 18 in. Longer bolts or bolts in 4‑in.
twist-off splines (ASTM F3125 Grades F1852 and F2280),
length increments are available given sufficient lead time
the length, L, is measured from the bolt bearing surface to
and appropriate coordination with a supplier.
the center point of the groove between the threaded por-
The thread length, LT, is the distance from the last com-
tion of the bolt and the spline drive (ASME B18.2.6-19). For
plete thread near the shank to the extreme end of the bolt
F3148 spline drive bolts (ASTM F3148-17a), the length, L,
for Grades A325 and A490 bolts, to the center point of the
is measured from the bolt bearing surface to the first indi-
groove for Grades F1852 and F2280 bolts, or to the first indi-
cation of thread near the spline, as shown in Figure 2. The
cation of thread for F3148 bolts, as is shown in Figures 1 and
tolerances on the overall length of a bolt, shown in Table 2,
2. The thread length of structural bolts is generally shorter
range from +0 in. to approximately −8 in. or −4 in. depend-
than that of similar nonstructural bolts so as to more eas-
ing on the diameter and nominal length of the bolt. A looser
ily allow the threads of the bolt to be excluded from the
tolerance for bolts longer than 6 in. reflects once-common
shear plane, thus increasing the strength of the bolt when
manufacturing methods but is likely no longer required
it is subjected to shear. Although many resources, includ-
due to improved production practices, while methods vary
ing the AISC Manual and the RCSC Specification, include

R
E
D

LB Y
LG LT LS
H L

Fig. 2.  Geometry of ASTM F3148 bolts.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 5

 Table 3.  Bolt Lengths, Thread Lengths, and Shortest Lengths Routinely Produced
Nominal Bolt Diameter, Thread Length, Transition Length, Shortest Length
E or db (in.) LT (in.) Y (in.) Produced (in.)*
   2 1.00 0.19 14
   s 1.25 0.22 12
   w 1.38 0.25 1w
   d 1.50 0.28 2
1 1.75 0.31 24
18 2.00 0.34 2¾
14 2.00 0.38 2¾
1a 2.25 0.44 Special order
12 2.25 0.44 See text
* Bolts are occasionally produced in shorter lengths for special orders.

tables detailing the length of threads, LT , also shown here in LG,max, and a minimum body length, L B,min, which are cal-
Table 3, the thread length is a reference dimension in ASME culated as shown in Equations  1 and 2 using the nominal
B18.2.6, intended for calculation purposes only, and may overall length and the reference dimensions LT and Y. Values
actually vary from published nominal values. of LG,max and L B,min are tabulated in ASME B18.2.6 for each
Instead of controlling the thread length of a bolt, the diameter and length of fastener and are also shown herein
geometry of the bolt is controlled by the overall length of (in part) as Table A1. Because LT and Y are reference dimen-
the bolt, L, the grip gaging length, LG, and the body length, sions, no tolerances are provided for these dimensions and
L B. The grip gaging length, LG, is a control dimension mea- actual measured dimensions on finished product may vary
sured from the bearing surface of the head to the face of a from these published values.
thread ring gage (Figure  3) that is threaded onto the bolt
LG,max = Lnom − LT(1)
by hand until it stops at the thread runout. The bolt body
length, L B, which is also a control dimension, is basically
L B,min = LG,max − Y (2)
the length of the shank or body of the bolt. The body length
is more precisely defined as the distance measured from the
ASME B18.2.6 states that when the minimum body
bearing surface of the head to the last scratch of thread for
length, LB,min, is short enough, that the bolt shall be threaded
bolts with cut threads, as shown in Figure 1(a), or to the top
full length. Specifically, it says that when LB,min ≤ 2.5p for
of the extrusion angle for bolts with rolled threads, as shown
db ≤ 1 in. or when LB,min ≤ 3.5p for db > 1 in., the bolt shall
in Figures 1(b)–1(d). The transition length, Y, is a reference
be threaded full length, where p is the pitch of the threads on
dimension that represents the length of the transition region
the bolt. However, ASME B18.2.6 also says that bolts that
between the threads and body.
are threaded full length are permitted to have an unthreaded
The grip gage length and body length are used as con-
length under the head that is not longer than 2.5p for bolts
trol dimensions by specifying a maximum grip gage length,
1  in. in diameter or smaller and 3.5p for bolts larger than
1  in. in diameter. One implication of this is that there are
some lengths of bolts an engineer may expect to be fully
threaded that may, in fact, have a short unthreaded length
under the head. Another implication is that there are some
lengths of bolts an engineer may expect to definitely have
a shank but may, in fact, be fully (or mostly) threaded. The
former case is generally of little concern in most cases, but
the latter case may lead to a serious design issue.
With two exceptions, bolts that are fully threaded do not
carry a special designation identifying them as such. The
first exception is for bolts manufactured with nonstandard
dimensions, which are designated with an S—A490S, for
Fig. 3.  A thread ring gage. example. The second exception is that Grade A325 bolts up

6 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

to a length of four times their diameter may be manufactured Case 2
as fully threaded and are designated with a T—A325T, for
Next, consider a d in.-9×1w in. bolt with
example. Bolts with the “T” designation are for users who
may want a longer bolt that is fully threaded so they do not LG,max = 1w in. − 12 in.
need to order or inventory multiple shorter sizes. These bolts
= 0.25 in.
are common in markets such as the metal building indus-
try and are used in multiple grip ranges and in the threads
L B,min = 0.25 in. − 0.28 in.
included condition. The fully threaded T bolts are also per-
mitted to have an unthreaded length under the head that is = −0.03 in.
not longer than 2.5p for bolts 1 in. in diameter or smaller and
3.5p for bolts larger than 1 in. in diameter. Therefore use L B,min = 0.00 in.
Because L B,min is less than 2.5p, LG,max is taken as 2.5p =
Illustrative Cases 0.28  in. Because the value of L B,min is less than 2.5p, this
bolt, despite having a nominal thread length of LT = 12 in.
The next several cases consider d‑in.-diameter bolts of
and a nominal shank or body length of LB = 1w in. – 12 in. =
varying length. For all d‑in.-diameter structural bolts, there
4  in., would be considered fully threaded according to
are 9 threads per in. (TPI); thus the pitch of these bolts
ASME B18.2.6 and, like Case  1, may indeed be produced
is p = 9 in. and 2.5p = (2.5)/(9 in.) = 0.28 in. Further, all
as fully threaded, or it may have a small unthreaded and
d‑in.-diameter structural bolts have a threaded length, LT, a
transition length that could be as long as 0.28 in. These two
reference dimension, equal to 12 in. Finally, the transition
alternatives, both of which are in compliance with ASME
length, Y, also a reference dimension, has a value of 0.28 in.
B18.2.6, are shown as Figures 4(c) and 4(d), respectively.
for all d‑in.-diameter structural bolts.
Case 3
Case 1
Next, consider a d in.-9×2 in. bolt with
First, consider a d in.-9×12 in. bolt. The maximum grip gage
length, LG,max, is calculated using Equation 1 and the mini- LG,max = 2 in. − 12 in.
mum body length, L B,min, is calculated using Equation 2:
= 0.50 in.
LG,max = Lnom − LT(1)
L B,min = 0.50 in. − 0.28 in.
= 12 in. − 12 in.
= 0.22 in.
= 0.00 in.
Because LB,min is less than 2.5p, LG,max is taken as 2.5p =
L B,min = LG,max − Y (2)
0.28  in. Because this value of LB,min is less than 2.5p, this
= 0.00 in. − 0.28 in. bolt, despite having a nominal thread length of LT = 12 in.
and a nominal shank or body length of LB = 2 in. − 12 in. =
= −0.28 in.
2  in., would be considered fully threaded according to
ASME B18.2.6. Like Cases 1 and 2, the d‑in.-9×2‑in. bolt
Therefore, use L B,min = 0.00 in.
may be produced as fully threaded, or it may have a small
Because L B,min is less than 2.5p, however, LG,max is taken
unthreaded and transition length that could be as long as
as 2.5p  = 0.28  in. Because the value of LB,min is less than
0.28 in. These two alternatives, both of which are in com-
2.5p, the bolt is considered to be fully threaded. Because the
pliance with ASME B18.2.6, are shown in Figures 4(e) and
nominal thread length for a d‑in.-diameter bolt is 12  in.,
4(f), respectively.
which is equal to the nominal overall bolt length, the bolt
Three different variants of d‑in.-9×2‑in. bolts made by
would simply be considered as fully threaded.
different manufacturers are shown in Figure 5. The bolt on
Based on the ASME B18.2.6 standard, the d in.-9×12 in.
the left has a short unthreaded body with a diameter that is
diameter bolt may be produced without an unthreaded
less than the nominal diameter of the bolt, the bolt in the
body, or the bolt may have a small unthreaded and transi-
middle is basically all transition up to the body diameter,
tion length that could be as long as 2.5p. The diameter of
and the bolt on the right has a short body with a diameter
this unthreaded and transition length may be as large as the
equal to the nominal diameter of the bolt and a short transi-
full nominal diameter (over/under tolerances) but may be as
tion. All three bolts are in compliance with ASME B18.2.6.
small as the pitch diameter of the bolt. These two alterna-
tives, both of which are in compliance with ASME B18.2.6,
are shown as Figures 4(a) and 4(b), respectively. (The toler-
ance for overall bolt length is not illustrated in Figure 4.)
ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 7
Case 4 The case of the d‑in.-9×2  in. represents a potentially
serious design issue. A structural engineer or connection
Now, consider a d‑in.-9×24 in. bolt with
detailer would likely review tables in the RCSC Specifica-
LG,max = 24 in. − 12 in. tion (RCSC, 2015) or AISC Manual (AISC, 2017), see that
the thread length is listed as LT = 12 in. for a bolt that is 2 in.
= 0.75 in.
long, and, expecting the bolt to have a 2‑in.-long shank,
may design the bolt as if the threads are excluded from the
LB,min = 0.75 in. − 0.28 in.
shear plane, as is shown in Figure  6(a). The d‑in.-9×2  in.
= 0.47 in. bolt as supplied by the manufacturer in full compliance with
ASME B18.2.6  may however have a shank much shorter
Because this value of LB,min is greater than 2.5p, this bolt than 2 in. or have no shank at all, resulting in a significant
will have a shank that is at least 0.47 in. long with a diam- deviation from the engineer’s or detailer’s expectations, as is
eter equal to the nominal body diameter (±tolerances) and a shown in Figure 6(b).
transition region that may be as long as 0.28 in. This bolt, The shortest length bolts that are routinely produced
which is in compliance with ASME B18.2.6, is shown as for each diameter are shown in Table  3. With the excep-
Figure 4(g). tions of 1a‑in.- and 12‑in.-diameter bolts, bolts with the
It should be noted that the d‑in.-9×2‑in. is the shortest lengths shown in the table may or may not have a shank or
d‑in.-diameter high-strength bolt that is routinely produced. unthreaded body depending on the tooling and preferences
Because the d‑in. × 2-in. is considered to be fully threaded, of the manufacturer. Bolts with lengths greater than those
there is little demand for a d‑in.-diameter bolt shorter than shown in the table will have a shank.
2  in., although shorter ones are occasionally manufactured A direct application of the formulas for Lb and Lg in
upon request. The d‑in. × 24‑in. is the shortest d‑in.-diameter ASME B18.2.6 to 1a-in.- and 12-in.-diameter bolts would
high-strength bolt that is guaranteed to have a shank. show that 1a‑in.- and 12‑in.-diameter bolts with lengths

(a) (b) (c) (d) (e) (f) (g)

Fig. 4.  Conforming variants of d -in.-9×12-in., d -in.-9×1w-in., d-in.-9×2-in., and d -in.-9×24-in. bolts.

Fig. 5.  Three d-in.-9 × 2-in. bolts manufactured by three different manufacturers.

8 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

34  in. and shorter would be fully threaded. In Table  2 of its maximum permitted length, L = 4.00 in., and Figures 7(c)
ASME B18.2.6-10 (2010), however, 1a‑in.- and 12‑in.- and 7(d) show the bolt with its minimum permitted length
diameter bolts with lengths 3  in. and shorter were both L = 4.00 − 0.19 in. = 3.81 in. Additionally, Figures 7(a) and
shown as fully threaded. Going back to the 2006 edition, in 7(c) show the bolt with its minimum body length of L B  =
Table 2 of ASME B18.2.6-06 (2006), 1a‑in.-diameter bolts L B,min = 2.22 in., while Figures 7(b) and 7(d) show the bolt
with lengths 3 in. and shorter were shown as fully threaded with a slightly longer body length of LB = 2.38 in. that could
but 12‑in.-diameter bolts with lengths 3w  in. and shorter result with a shorter transition length of Y = 0.22 in. instead
were shown as fully threaded. This discrepancy has been of the reference value of Y = 0.28 in. All four variations are
resolved in ASME B18.2.6-19 (2019), and the corresponding within the ASME B18.2.6 tolerances for a d-in.-9×4-in.
table has been updated to be consistent with the equations. bolt, and consideration of these variations is essential when
As was stated earlier, 1a‑in.-diameter bolts are not routinely selecting a bolt with appropriate length for a joint.
produced and are available by special order only; they are
included in this discussion for the sake of completeness only. Nut and Washer Thickness Dimensions
ASME B18.2.6 also includes dimensions for heavy hex nuts
Case 5
and hardened washers. While dimensions and tolerances are
Finally consider a d-in.-9×4-in. bolt, where given for the width across the flats and the width across the
corners for nuts and for outer and inner diameters for wash-
LG,max = 4 in. − 12 in.
ers, the tolerances on thickness for both the nuts and wash-
= 2.50 in. ers are of primary concern for this paper and are shown in
Table 4. Nominal thicknesses are not provided for washers
L B,min = 2.50 in. − 0.28 in. in ASME B18.2.6, but E in. is often used as the nominal
thickness for all diameter washers (RCSC, 2015).
= 2.22 in.

Note that the transition length can vary from manufac- BOLT LENGTH DETERMINATION
turer to manufacturer; thus, depending on the actual values
Two design criteria that must be satisfied in selecting the
of Y and L (including tolerances for L), the thread length,
length of a bolt for a given joint are (1)  the bolt must be
LT, for this bolt could range anywhere from LT = (4.00 in. −
short enough that the nut can be either snug tightened or
0.19 in.) − 2.50 in. = 1.31 in. to LT = (4.00 in. − 0.00 in.) −
pretensioned without the threads of the nut running out onto
2.22 in. = 1.78 in., despite having a nominal reference value
the transition region of the bolt (“shanking out”), and (2) the
of LT = 12 in. In all cases, though, the body of the bolt will
bolt must be long enough that the nut can be threaded com-
be at least 2.22  in. long, and the grip gage length will not
pletely onto the bolt (zero or positive stick-out). These two
exceed 2.50 in.
cases are illustrated in Figures 8(a) and 8(b), which show a
Four different variations of the d-in.-9×4-in. bolt are
bolting assembly consisting of a 1-in.‑8×6-in. bolt, a nut, and
shown in Figure 7. Figures 7(a) and 7(b) show the bolt with

(a) (b)

Fig. 6.  Illustration of a joint with d -in.-9 ×2-in. bolts (a) as might be expected by the engineer of record and (b) as provided.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 9

 Table 4.  Nut and Washer Thicknesses
Nominal Bolt Heavy Hex Nut Thickness Washer Thickness
Diameter (in.) Maximum (in.) Nominal (in.) Minimum (in.) Minimum (in.) Maximum (in.)
 2 0.504 31 64 0.464 0.097 0.177
 s 0.631 39 64 0.587 0.122 0.177
 w 0.758 4764 0.710 0.122 0.177
 d 0.885 55 64 0.833 0.136 0.177
1 1.012 63 64 0.956 0.136 0.177
18 1.139 1764 1.079 0.136 0.177
14 1.251 1R 1.187 0.136 0.177
1a 1.378 1Y 1.310 0.136 0.177
12 1.505 1I 1.433 0.136 0.177

two F436 washers.2 Figure 8(a) illustrates the first criterion, Minimum grip = LG,max − Σt washers − δpretension (3)
which leads to the minimum grip for a given bolt length,
while Figure  8(b) illustrates the second criterion, which Maximum grip = L − Σt washers − tnut (4)
leads to the maximum grip for a given bolt length.3 These
Considering a 1-in.-8×6-in. bolt, the maximum grip gage
two criteria can be written mathematically as
length can be computed as
LG,max = 6 in. − 1w in.
2 The case of a bolting assembly with two washers—one under the head = 4.25 in.
and one under the nut—has been used in this paper because it represents a
situation that is useful for illustrating the calculations that are presented. For this bolt, with L < 8db, the change in length during
It should be noted, however, that while the use of two washers is neces- pretensioning, based on a half turn past snug being required,
sary in some situations, it is more common to use just a single washer. can be estimated as
3 Care should be taken to not confuse the “grip gage length” of a bolt with δpretension = 0.5/8
the “grip” of a joint. The former is a control dimension used in the manu-
facturing of bolts while the latter is the total thickness of a joint between = 0.0625 in.
the bearing surfaces of the bolt and nut, which, in this work, excludes the
thickness of F436 washers included with the bolting assembly. This is It should be noted that most of this elongation is expected
consistent with the definition of found in the RCSC Specification (2015)
to occur within the threaded region of the bolt. Thus,
but not with the definition found in the AISC Specification (2016).

Lmax = 4.00"

Lmin = 3.81"

LG.max = 2.50"

LB.min = 2.22"

(a) (b) (c) (d)

Fig. 7.  Different variations of d-in.-9×4-in. high-strength bolts per ASME B18.2.6-19.

10 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

assuming that two washers are used and using the minimum is that the bolt should be long enough so as to exclude the
washer thickness permitted in ASME B18.2.6, the minimum threads of the bolt from the shear plane(s) of the joint. Con-
grip can be determined using Equation 3: sidering the joint shown in Figure  9, where four plies are
joined, the bolt can be designed with the threads excluded
Minimum grip = LG,max − Σt washers − δpretension(3)
from the shear plane so long as none of the faying surfaces
= 4.25 in. − (2)(0.136 in.) − 0.0625 in. between plies intersects the threaded portion of the bolt.
This is true when
= 3.92 in.
Σt(n−1) + Σt washers under head ≤ L B,min + f (Y)(5)
Again, assuming that two washers are used, but using the
n 1
maximum washer thickness, maximum nut thickness, and
where Σt(n−1) is used as short hand for ∑ ti and f (Y) is some
minimum overall bolt length permitted in ASME B18.2.6, i=1
the maximum grip can be determined using Equation 4: fraction of thread transition length, Y. The threads can be
Maximum grip = L − Σt washers − tnut(4) excluded from the shear plane as long as
= (6.00 in.− 0.19 in.) Σt(n−1) ≤ L B,min − Σt washers under head + f (Y) (6)
– (2)(0.177 in.) − 1.012 in.
A fraction of the thread transition length, f (Y), is included
= 4.44 in. in Equations 5 and 6 because commentary to Section 2.3 of
the RCSC Specification (RCSC, 2015) can be interpreted as
Values for the minimum and maximum grip for all com-
permitting a bolt to be considered in the threads excluded or
mon diameter and length high-strength bolts are tabulated in
X condition when the shear plane passes through the tran-
Table A2. The values shown in Table A2 for the minimum
sition region of the bolt, though this section of the RCSC
grip include an elongation during pretensioning based on a
Specification is under review as of the writing of this paper.
half-turn of the nut for bolts where L ≤ 8db and two-thirds
Knowing that the actual value of Y may vary from one man-
turn of the nut for bolts where L > 8db. This is largely consis-
ufacturer to another, however, there is merit to simply and
tent with the RCSC Specification for joints where outer faces
conservatively taking f (Y) = 0.
of the gripped material are normal to the axis of the bolt but
If the Σt(n−1),max is defined as
is slightly conservative for bolts where L ≤ 4db where one-
third turn would be required. Σt(n−1),max = L B,min − Σt washers under head + f (Y)(7)
A third design criterion that is implemented in some cases

Minimum grip Stick out

LG,max

(a)

Maximum grip

(b)

Fig. 8.  Minimum and maximum grip for a given bolt length.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 11

then it can be stated that when Σt(n−1) for a joint is less than employed, an F436 washer or a c‑in.-thick common plate
Σt(n−1),max for a bolting assembly, then the bolt can be treated washer is again required to completely cover slotted holes, a
as if it is in the threads excluded or X condition. Again, con- beveled washer is required for sloping surfaces, and an F436
sidering a bolting assembly consisting of a 1-in.-8×6-in. bolt, washer is required under both the head and the nut when
a nut, and two F436 washers—one under the head and one material with a yield strength of less than 40 ksi is joined
under the nut—the Σt(n−1),max can be computed as using Grade A490 or F2280 bolts, except that a washer is
not needed under the head of an F2280 when it has a button
Σt(n−1),max = L B,min − Σt washers under head + f (Y)(7)
head like that shown in Figure 1(c). Further, when twist-off
= 3.94 in. − 0.177 in. + 0 in. bolts are used in a pretensioned or slip-critical joint, an F436
washer is required under the nut of the assembly, and when
= 3.76 in.
the calibrated wrench method of installation is used, an F436
where the maximum thickness of the washer was used and washer must be used under the turned element (either the
f (Y) was conservatively taken as zero. Values of Σt(n−1),max nut or head of the bolt), regardless of bolt grade or material
for common configurations of bolting assemblies com- strength. Finally, there are additional washer requirements
puted with f (Y) conservatively taken as zero are provided when direct-tension-indicators are used for pretensioning.
in Table A2. Given these requirements for washers, common bolting
In the past, other authors have related the location of the assembly configurations include no washers, a single washer
shear plane relative to the thread transition as a function of under either the nut or head, or two washers with one under
the bolt length and the thickness of the last ply in the joint, tn the nut and another under the head. Occasionally, multiple
(i.e., the ply closest to the nut). Because the thread length, LT, washers are used to allow more margin for nut rotation and
is a reference dimension, however, there is little assurance thread engagement during pretensioning. In those cases, the
that the transition will be where it is expected when its loca- washers are generally added under the nut. With few excep-
tion is based on LT. In this work, the location of the thread tions, there is little to be gained from using washers under
transition is determined as a function of the minimum body the head of the bolt, since doing so makes it more challeng-
length, L B,min, which is a control dimension, and is thus more ing to exclude the threads of the bolt from the shear plane.
reliable than the thread length. One exception is when the head of the bolt is the turned ele-
ment during tightening using the calibrated wrench method
Use of Washers of installation, in which case a single washer under the head
will suffice. Another exception is when Grade A490 or hex
The AISC Specification (AISC, 2016) refers to the RCSC
headed Grade F2280 bolts are used to connect material
Specification regarding the use of washers. According to the
with a yield stress less than 40 ksi. In other cases, washers
RCSC Specification (RCSC, 2015), washers are not required
other than F436 are used, including c‑in.-thick common
for snug-tightened joints except when a slotted hole is used,
washers, plate washers, direct-tension-indicating washers,
in which case an F436 washer or a c‑in.-thick common
or shims. In these cases, the thicknesses of those washers,
plate washer is required to completely cover the hole, or
shims, and/or washer plates must be included in the bolt
when sloping surfaces are joined, where a beveled washer
length calculations.
is required. When pretensioned or slip-critical joints are

t n- tn

t1 t2 ti tn

LB,min Y

Fig. 9.  Illustration of plies and threads in the grip of a bolt.

12 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

Procedure for Determination of Bolt Length that satisfies both criteria. In that case, try either
A procedure for the determination of bolt length for a joint adding washer(s) to the bolting assembly or
is summarized here. selecting a bolt that meets only the first criterion
(i.e., a bolt that has a minimum grip that is shorter
Step 1. Determine the type and location of washer(s) than the required grip).
required for the particular joint specified, taking care
Step 5. Unless bolts are not subject to shear or are
to consider washer requirements in the RCSC and
predetermined to be designed in the threads not
AISC Specifications related to hole type, location,
excluded or N condition, determine the maximum
material strength, and method of installation.
cumulative thickness of all but the last plate within
Step 2. Compute the required grip for the joint. The the grip of the bolt, Σt(n−1),max, and compare this
required grip should include all materials between value to Σt(n−1) that was determined in Step  3. If
the bearing surfaces of the bolt head and the nut Σt(n−1) is not greater than Σt(n−1),max then the bolt can
except for F436 hardened washers addressed in be designed in the threads excluded or X condition.
Step 1. Note that there is sometimes more than one bolt length
Step 3. Unless bolts are not subject to shear or are that will satisfy the design criteria for a given joint. It is
predetermined to be designed in the threads not up to the engineer to decide which one is desired. On one
excluded or N condition, determine the cumulative hand, choosing a longer bolt will result in more thread stick
thickness of all but the last plate within the grip of out past the exposed face of the nut and will make it more
the bolt, Σt(n−1) (i.e., the thickness of all of the plates likely that the threads will be excluded from the shear plane,
within the grip except for the plate closest to the resulting in higher strength from the fastener. On the other
nut). This quantity may depend on the way in which hand, though, choosing a shorter bolt will result in less
the bolt is installed in joints that are not symmetric thread stick out and more threads within the grip of the bolt,
about the mid-length of the joint grip and should which can improve the rotational capacity of the bolt dur-
include all materials between the bearing surface ing pretensioning and can improve the ductility of the bolt
of the nut and last faying surface of the joint except under tension. Note that the values shown for minimum grip
for F436 washers, which are included in Table A2. in Table  A2 include the minimum required turn of nut for
most cases. As such, when the required grip for a joint is
Step 4. Enter Table  A2 on the appropriate page for the close to the minimum grip for a bolt shown in Table A2, an
bolt diameter specified, and using the appropriate engineer may wish to select the next shorter bolt, if possible,
columns for the washer configuration determined to avoid potential problems with the nut “shanking out” on
in Step  1, select a bolt length that has both (1)  a the transition region due to over-rotation of the nut during
minimum grip that is shorter than the required grip pretensioning. Alternatively, an additional washer can often
and (2)  a maximum grip that is longer than the be added to the bolting assembly to reduce the likelihood of
required grip. Rarely, a bolt cannot be identified shanking out.

DESIGN EXAMPLES
Six examples are presented in this section to illustrate the use of the proposed procedure for determining the appropriate length
of bolts. Examples 1–4 are adapted from examples that were presented by Carter (1996). Examples 5 and 6 are intended to illus-
trate less commonly encountered design issues.

Example 1
Determine the bolt length for w‑in.-diameter ASTM F3125 Grade A325 snug-tight bolts in standard holes in a a‑in. single-plate
connection supporting a W21×50 beam (nominal t w = a in.)

Recommended Solution
The required grip for the joint is a in. + a in. = w in. Because the thickness of either of the plies joined is t = a in., Σt(n−1) = a in.
for the joint regardless of whether the bolt is installed through the single plate first or through the beam web first. No washers are
required for a snug-tightened joint using Grade A325 fasteners. Table A2 is used to select w‑in.-diameter bolts with a minimum
grip less than or equal to w in. and a maximum grip greater than or equal to w in. The following options are available:

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 13

 1. w-in.-10×1w-in. with no washers
Min grip = 0.20 in.
Max grip = 0.80 in.
Fully threaded

2. w-in.-10×2-in. with no washers

Min grip = 0.57 in.
Max grip = 1.05 in.
Σt(n−1),max = 0.37 in.

3. w-in.-10×2-in. with one washer

Min grip = 0.45 in.
Max grip = 0.88 in.
Σt(n−1),max = 0.37 in.

4. w-in.-10×24-in. with one washer

Min grip = 0.70 in.
Max grip = 1.13 in.
Σt(n−1),max = 0.62 in.

5. w-in.-10×24-in. with two washers4

Min grip = 0.58 in.
Max grip = 0.95 in.
Σt(n−1),max = 0.44 in.

Based on the options available for this joint, a w-in.-10×2-in. bolt is recommended because it would be of an acceptable length
to work either without washers or with a single washer. Because Σt(n−1),max = 0.37 in. in either case, which is smaller than Σt(n−1) =
a in., the bolt would need to be designed in the threads not excluded or N condition regardless of whether it is inserted through
the shear plate first or through the beam web first.
If it is required to exclude the threads from the shear plane, then a w-in.-10×24 in. bolt would work, though this assembly would
need at least one washer included to avoid shanking out the nut during installation (even for a snug-tight installation).

RCSC Method
Because washers are not required for this joint, the grip + washer thickness is w in. Per RCSC Specification Table C2.2, the
length of the bolt is determined by adding 1 in. to the grip + washer thickness. Thus, the length of the bolt can be determined as
Lreq = w in. + 1 in.
= 1w in.

Therefore, use Lreq = 1w in.

For the w-in.-diameter bolt, LT = 1a-in., thus the shank would be expected to be 1w-in. − 1a-in. = a in. long, and the engineer
may expect the bolt to be in the threads excluded or X condition. Note, however, that according to Table A1, the w-in.-10×1w-in.
bolt is considered to be fully threaded and would have an unthreaded and transition length that would be no longer than 0.25 in.

4 In the design examples presented in this work, when one washer is included, it will be under the nut, and when two washers are included, one will be under
the nut and one will be under the head.

14 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

Example 2
Determine the bolt length for w‑in.-diameter ASTM F3125 Grade A325 snug-tightened bolts in standard holes in a double-angle
connection with 2L5×3×c‑in. angles supporting a W27× 84 beam (nominal t w = v‑in.).5

Recommended Solution
The required grip for this joint is c in. + v in. + c in. = 1z in. Regardless of which direction the bolt is installed, Σt(n−1) for
the joint would be c in. + v in. = w in. Washers are not required for a snug tightened joint. Table A2 is used to select w‑in.-
diameter bolts with a minimum grip less than or equal to 1z in. and a maximum grip greater than or equal to 1z in. The fol-
lowing options are available:
1. w-in.-10×24-in. with no washers
Min grip = 0.82 in.
Max grip = 1.30 in.
Σt(n−1),max = 0.62 in.

2. w-in.-10×24-in. with one washer

Min grip = 0.70 in.
Max grip = 1.13 in.
Σt(n−1),max = 0.62 in.

3. w-in.-10×22-in. with one washer

Min grip = 0.95 in.
Max grip = 1.38 in.
Σt(n−1),max = 0.87 in.

4. w-in.-10×22-in. with two washers

Min grip = 0.83 in.
Max grip = 1.20 in.
Σt(n−1),max = 0.69 in.
Based on the options available for this joint, a w-in.‑10×24-in. bolt is recommended because it would be of an acceptable length
either without washers or with a single washer. Because Σt(n−1),max = 0.62 in. in either case, which is smaller than Σt(n−1) = ¾ in.,
the bolt would need to be designed in the threads not excluded or N condition.
If a threads excluded or X condition is required, a w-in.‑10×22-in. bolt would be acceptable, though this assembly would require
one washer under the nut to avoid shanking out the nut during installation (even for a snug-tight installation). Putting the washer
under the head, however, may cause the threads to be not excluded from the shear plane.

RCSC Method
When no washers are required, as would be the case for snug-tightened Grade A325 bolts, the grip + washer thickness is c in. +
v in. + c in. = 1z in. Thus, the length of the bolt can be determined using RCSC Specification Table C2.2 as
Lreq = 1z in. + 1 in.
= 2z in.
Therefore, use Lreq = 24 in.

5 It is noted that 2L4×32×c angles may be more commonly used than 2L5×3×c. However, the 2L5×3×c angles were selected for this example to maintain
consistency with the paper by Carter (1996), where it first appeared.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 15

For the w-in.-diameter bolt, LT = 1a in.; thus, the shank would be expected to be 24 in. − 1a in. = d in. long, and the engineer
would expect the bolt to be in the threads excluded or X condition. Note from Table A1, however, that the shank or body length
of a w-in.-10×24-in. bolt is guaranteed only to be 0.62 in. As a result, the 24‑in.-long bolt might actually be in the threads not
excluded or N condition. If the engineer considers a bolt with the shear plane passing through the thread transition to be treated
as if it is in the threads excluded or X condition, then from Table A1 it can be observed that the maximum grip gage length of
a 24‑in.-long bolt is 0.87 in, and the bolt could be treated as if the threads were excluded. This is a maximum length, however,
and the actual grip gage length may be anywhere between 0.62 in. and 0.87 in.

Example 3
Determine the bolt length for pretensioned w‑in.-diameter ASTM F3125 Grade A490 bolts connecting a 2‑in.-thick angle to a
W14×500 column flange (nominal t w = 32 in.). The calibrated wrench method will be used to pretension the bolts.

Recommended Solution
The required grip for this joint is 32 in. + 2 in. = 4 in. If the bolt is installed through the column flange first, then Σt(n−1) = 32 in.
but if the bolt is installed through the angle leg first, then Σt(n−1) = 2 in. Because the calibrated wrench method of installation is
to be used, at least one F436 washer is required under the turned element. A Grade A490 bolt was specified, so an F436 washer
is also required between the angle and the bearing surface of the head or nut since the angle would be made of 36-ksi material
unless otherwise designated.
Assuming that the bolt is installed through the column flange first and that the nut is the turned element, then a single washer
under the nut is required and from Table A2 a 52‑in.-long bolt can be selected. For a w‑in.-10×52 in. bolting assembly with
one washer under the nut, the minimum grip is 3.95 in. and the maximum grip is 4.38 in. In that case, Σt(n−1),max = 3.87 in. and
because this is larger than Σt(n−1) = 32 in., the bolt can be considered to be in the threads excluded or X condition. Note that a
w-in.-10×54-in. bolting assembly with one washer under the nut could be specified, but this would likely be available only by
special order.
Alternatively, assuming that the bolt is installed through the angle leg first and the nut is the turned element, then two washers
are required—one under the nut and one under the head of the bolting assembly. A 52‑in.-long bolt can again be selected, but in
this case, the minimum grip would be 3.83 in.; the maximum grip would be 4.20 in.; the Σt(n−1),max would be 3.69 in.; and because
this is larger than Σt(n−1) = 2 in., the bolt can be considered to be in the threads excluded or X condition.

RCSC Method
With a washer under the turned element, the grip + washer thickness is 32 in. + 2 in. + E in. = 4E in. Thus, the length of the
bolt can be determined as
Lreq = 4E in. + 1 in. 
= 5E in.

Therefore, use Lreq = 52 in.

For the w‑in.-diameter bolt, LT = 1a in.; thus, the shank would be expected to be 52 in. − 1a in. = 48 in. long, and the engineer
would expect the bolt to be in the threads excluded or X condition.

Example 4
Determine the bolt length for d‑in.-diameter ASTM F3125 Grade F2280 bolts in standard holes in an extended endplate moment
connection (1‑in.-thick plate) to a W14×132 column flange (nominal tf = 1 in.) The column and endplate are both made of mate-
rial with Fy = 50 ksi.

Recommended Solution
The required grip for this joint is 1 in. + 1 in. = 2 in. Because both plates joined are 1 in. thick, the Σt(n−1) is 1 in. regardless of
whether the bolt is installed through the column flange first or through the endplate first. Grade F2280 bolts are to be used, so
an F436 washer is required under the nut. Because 50‑ksi material is gripped by the F2280 bolts, a washer is not required under

16 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

the head. Table A2 is used to select d‑in.-diameter bolts with a minimum grip less than or equal to 2 in. and a maximum grip
greater than or equal to 2 in. The following options are available:
1. d-in.-9×34-in. with one washer
Min grip = 1.56 in.
Max grip = 2.00 in.
Σt(n−1),max = 1.47 in.

2. d-in.-9×32-in. with one washer

Min grip = 1.81 in.
Max grip = 2.25 in.
Σt(n−1),max = 1.72 in.

3. d-in.-9×32-in. with two washers

Min grip = 1.67 in.
Max grip = 2.07 in.
Σt(n−1),max = 1.54 in.

4. d-in.-9×3w-in. with two washers

Min grip = 1.92 in.
Max grip = 2.32 in.
Σt(n−1),max = 1.79 in.
Based on the options available for this joint, a d-in.-9×32-in. bolt is recommended because it would be of an acceptable length
either with one washer or with two washers. Because the smallest value of Σt(n−1),max for this bolt is 1.54 in., which is greater than
Σt(n−1) = 1 in., the bolt can be designed in the threads excluded or X condition, in either case.

RCSC Method
With a washer under the nut of the bolting assembly, the grip + washer thickness is 1 in. + 1 in. + E in. = 2E in. Thus, the
length of the bolt can be determined as
Lreq = 2E in. + 18 in.
= 3T in.
Therefore, use Lreq = 32 in.
For the d‑in.-diameter bolt, LT = 12 in.; thus the shank would be expected to be 32 in. − 12 in. = 2 in. long, and the engineer
would expect the bolt to be in the threads excluded or X condition.

Example 5
Determine the length for 1‑in.-diameter ASTM F3148 bolts in standard holes connecting a flange plate (1 in. thick) to the flange
of a W24×76 beam (nominal tf = n in.) in a moment connection. The beam and flange plate are both made of material with
Fy = 50 ksi.

Recommended Solution
F3148 bolts ship as matched bolting assemblies with one washer that is to be used under the nut. The required grip for this joint
is 1 in. + n in. = 1n in. To account for the tolerances in the depth of the beam, however, which are ±8 in. per ASTM A6
(AISC, 2017), the connection will be detailed allowing for the beam depth plus 8 in. Shims will be provided to accommodate
a gap of up to (2)(8 in.) between one beam flange and the adjacent flange plate. Thus the required grip for the bolts could be as
large as 1 in. + n in. + (2)(8 in.) = 1, in. If the bolts are installed though the beam flange first, the Σt(n−1) would be n in.
without shim plates and could be as large as , in. with shim plates. If the bolts are installed through the flange plate first, the

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 17

Σt(n−1) would be 1 in. without shim plates and could be as large as 14 in. with shim plates.
Table A2 is used to select 1‑in.-diameter bolts with a minimum grip less than or equal to 1n in. and a maximum grip greater
than or equal to 1, in. Based on these criteria, the following options are available:
1. 1-in.-8×32-in. with one washer
Min grip = 1.55 in.
Max grip = 2.12 in.
Σt(n−1),max = 1.44 in.

2. 1-in.-8×32-in. with two washers

Min grip = 1.42 in.
Max grip = 1.94 in.
Σt(n−1),max = 1.26 in.

3. 1-in.-8×3w-in. with two washers

Min grip = 1.67 in.
Max grip = 2.19 in.
Σt(n−1),max = 1.51 in.

Based on the options available for this joint, a 1-in.-8×32-in. bolting assembly with one washer is selected. This assembly will
accommodate a grip ranging from 1.55 in. to 2.12 in. with either one washer or two and will work in the joint either with or with-
out shims. Using just one washer would eliminate the need for and the added cost of the second washer. It can be further noted
that the minimum value of Σt(n−1),max for the options shown is 1.26 in. Thus, because this is greater than the maximum considered
Σt(n−1), the bolts can be designed in the threads excluded or X condition regardless of which of the options is selected, regardless
of the direction in which the bolts are installed, and regardless of whether or not shim plates are used. If for some reason the
1-in.-8×32-in. bolting assembly was unavailable, the 1 in.-8×3w in. bolting assembly with two washers would also work, though
this would require additional washers to be provided.

Example 6
Determine the appropriate length for s‑in.-diameter ASTM F3125 Grade F1852 bolts that are used in a double lap splice shear
joint where two 12‑in.-thick splice plates are used to connect a 22‑in. main member.

Recommended Solution
The required grip for this joint is 12 in. + 22 in. + 12 in. = 52 in. The Σt(n−1) is 12 in. + 22 in. = 4 in. regardless of the direc-
tion in which the bolts are installed. Because Grade F1852 bolts are specified, one hardened washer is required under the nut of
the bolting assembly.
Using a required grip of 52 in., Table A2 is used and it is noted that there is not a solution for a s‑in.-diameter assembly with
one washer tabulated. Despite this, there are several options for this joint.
Option 1.  From Table  A2, select a s-in.-11×7-in. bolting assembly with two washers. From the table, the minimum grip is
5.45 in. and the maximum grip is 5.77 in. If both washers are installed under the nut of the assembly, Σt(n−1),max would be 5.53 in.
Alternatively if one washer was installed under the head and one washer under the nut, then Σt(n−1),max would be 5.35 in. In either
case, Σt(n−1),max would be larger than Σt(n−1), indicating that the bolt could be designed in the threads excluded or X condition.
Option 2.  Investigate the use of an assembly with tighter overall length tolerances: Using a required grip of 52 in., Table A2 is
used to select a bolt with one washer that has a minimum grip that is smaller than the required grip. A s-in.-11×62-in. bolting
assembly is selected with a minimum grip of 5.07 in. and from Equation 4, the required length of the bolt can be determined as
L Rqd = Maximum grip + Σt washers + tnut(8)

L Rqd = 52 in. + (1)(0.177 in.) + 0.631 in.

= 6.31 in.

18 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

As long as the bolt has an actual length of at least 6.31 in., even with a washer and nut that are both as thick as they can be
while still within tolerance, the bolting assembly will work. The standard tolerance for overall length for this bolting assembly
is −0.25 in. To be acceptable, the tolerance would need to be 6.31 in. − 62 in. = −0.19 in. This is certainly achievable and can be
accommodated by discussing needs with a reputable bolt supplier. In this case, Σt(n−1),max would be 5.03 in., and because this is
larger than Σt(n−1) = 4.00 in., the bolt could again be designed in the threads excluded or X condition.
Option 3.  Investigate the use of a s-in.-11×6w-in. bolting assembly with one washer under the nut: Though this bolt length is
not included in Table A2, it would likely be available by special order given proper coordination with a bolt supplier. For the
s-in.-11×6w-in. bolt,
LG,max = Lnom − LT(1)
= 6w in. − 14 in.
= 5.50 in.

L B,min = LG,max − Y(2)

= 5.50 in. − 0.22 in.
= 5.28 in.

Minimum grip = LG,max − Σt washers − δpretension(3)

= 5.50 in. − (1)(0.177 in.) − (2 turn)/(11)
= 5.28 in.

Maximum grip = L − Σt washers – tnut(4)

= (6.75 in. − 0.25 in.) − (1)(0.177 in.) − 0.631 in.
= 5.69 in.

Σt(n−1) ≤ L B,min − Σt washers under head + f(Y)(6)

= 5.28 in. − (0 in.) + (0 in.)
= 5.28 in.

With these values, it can be seen that that the s-in.-11×6w-in. assembly with one washer under the nut would be acceptable and
could be designed in the threads excluded or X condition.

SUMMARY AND CONCLUSIONS least the 1981 edition of ASME B18.2.1. Despite the fact that
this misunderstanding has occurred in the design of numer-
The dimensional tolerances in ASME B18.2.6, to which ous structures since at least 1981, the authors are not aware of
F3125 and F3148 bolts must conform, consist of control a structural failure that has resulted from this specific issue.
dimensions and reference dimensions. Control dimensions While this misunderstanding can certainly lead to uncon-
and their tolerances are those that bolt manufacturers must servative designs, it might be considered only less conserva-
meet in order for their product to be in conformance with tive compared to actual demand. It should be noted that the
published ASME and, by extension, ASTM standards. application of resistance factors, documented overstrength
Resources that structural engineers and detailers commonly of fasteners (Moore et al., 2010), and other factors mitigate
have available, however, typically provide reference dimen- the risk associated with these incorrect design assumptions.
sions for bolts. This inconsistency can lead to variances Based on an analysis of available bolt diameters and
between the expectations of the engineer or detailer and what lengths, a series of tables was generated to aid in the length
is actually built, possibly in ways that are unconservative. determination of bolts for joints, considering the most puni-
It should be noted that this inconsistency has existed for tive combination of dimensional tolerances of high-strength
nearly four decades. Prior to the introduction of ASME bolts, hardened washers, and heavy hex nuts. The tables
B18.2.6, the dimensions of high-strength fasteners were present ranges of grip lengths that bolting assemblies can
maintained in ASME B18.2.1 (1981). The dimensional accommodate based on bolt diameter, bolt length, and
requirements in ASME B18.2.619 can be traced back to at washer configuration and also provide a tool for quickly

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 19

determining whether the threads in a bolting assembly can or X condition when they may actually be in the threads
be considered as “excluded” from the shear plane or whether not excluded or N condition. Examples were also presented
they should be considered as “not excluded.” Several design illustrating situations where an engineer or detailer may
examples are presented demonstrating the use of these expect bolts to have a small, but predictable, shank when, in
tables in determining the length of bolts and comparing the fact, they may—in full compliance with ASME B18.2.6—
results with traditional methods presented in engineering be manufactured fully threaded with no shank at all. It
and detailing resources. was shown that this second situation generally occurs for
The proposed design tables represent a departure from the the shortest length of each diameter bolt that is routinely
approach presented by Carter (1996), where acceptable bolt produced. While this represents a small percentage of the
lengths were tabulated for given grips. A similar approach bolt lengths available, it should be noted that approximately
was considered in the current work, but the resulting tables 90% of bolts sold have a length less than four times their
were substantially larger than those proposed, including diameter, and bolts that are considered to be fully threaded
some redundancies. Instead of centering on the required grip, per ASME B18.2.6 represent a disproportionately high per-
the tables proposed in this work focus on acceptable grips centage—possibly in excess of 35% for some diameters—of
for practically all available bolting assemblies. One notable bolts sold.6
difference between the tables presented by Carter and those The ambiguous case of the d-in.-9×2-in. bolt is encoun-
proposed herein is that Carter’s tables provide the minimum tered quite commonly. If d‑in.-diameter bolts were used to
ply thickness closest to the nut (tn,min) that is required to join two a‑in.-thick plates, a bolt with a length of 2 in. would
ensure that the threads are excluded from the shear plane. be appropriate. The engineer, based on tables in resources
The tables proposed herein instead provide the maximum readily available to him or her, may expect that the bolts
thickness of the plies closest to the head (Σt(n−1),max) that would be in the threads excluded condition, as is shown in
can be used to ensure that the threads are excluded from Figure 5(a). The bolts, however, would be considered as fully
the shear plane. This latter approach was chosen because threaded according to ASME B18.2.6 and would actually be
Σt(n−1),max is not dependent on the number of washers used in the threads not excluded condition, as is shown in Fig-
under the nut. For the example joints presented, bolt lengths ure 5(b). Note that selecting a d-in.-9×24-in. bolt instead of
determined using Carter’s method match those determined a d-in.-9×2-in. bolt would also be appropriate, but since this
using the method proposed herein, for the most part. In bolt is not considered to be fully threaded, it would have a
some cases, however, such as for w-in.-10×1w-in. and d-in.- minimum body length of L B,min = 0.47 in. and could reliably
9×2-in. bolts, Table 2 in Carter indicates that the bolts can be considered to be in the threads excluded condition.
be designed in the threads excluded or X condition for some Based on the analysis and examples included in this work,
ply thickness when, in fact, the bolts would be considered as the authors recommend that bolts that are considered to be
fully threaded according to ASME B18.2.6. fully threaded per ASME B18.2.6 be designed for shear
It was observed that when the most punitive combination strength only in the threads not excluded or N condition.
of tolerances for the bolt, washer, and nut were assumed to Bolts in this category are those that are the shortest pro-
occur simultaneously, the minimum grip for a given bolt duced for each diameter as is shown in Table 3 (in addition
length was actually larger than the maximum grip for the to 1a‑in.- and 12‑in.-diameter bolts up to and including
next shorter bolt of the same diameter and washer configu- 34  in. long), which are those bolts shown above the solid
ration. This generally occurred for bolting assemblies rang- lines in Tables A1 and A2.
ing from 2 in. to d in. in diameter that are longer than 6‑in.
nominal length, although cases were observed for bolting REFERENCES
assemblies of all diameters with multiple washers. Bolts lon-
AASHTO (2017), AASHTO LRFD Bridge Design Specifi-
ger than 6 in. have looser tolerances for nominal length than
cations, 8th Ed., American Association of State Highway
bolts 6 in. long and shorter. This, combined with the shorter
and Transportation Officials, Washington D.C.
threaded length of smaller-diameter bolts and the standard
practice of producing bolts longer than 5 in. in 2‑in.-long AISC (2016), Specification for Structural Steel Buildings,
increments, leads to a situation where it may be difficult to ANSI/AISC 360-16, American Institute of Steel Con-
determine an appropriate bolt length for a given joint. This struction, Chicago, Ill.
potential situation is not likely to lead to too many diffi- AISC (2017), Steel Construction Manual, American Insti-
culties in the field, but it could be eliminated if the rather tute of Steel Construction, Chicago, Ill.
permissive lower tolerance on bolt length was tightened up.
The design examples included illustrate potential situa-
tions where bolts selected using traditional methods of length
determination might be designed in the threads excluded
6 Based on an informal study of LeJeune Bolt Co. sales from 2015 to 2018.

20 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

ASME (1981), Square, Hex, Heavy Hex, and Askew Head ASTM (2017), Standard Specification for High Strength
Bolts and Hex, Heavy Hex, Hex Flange, Lobed Head, and Structural Bolt Assemblies, Steel and Alloy Steel, Heat
Lag Screws (In. Series), ASME B18.2.1, American Soci- Treated, 144 ksi Minimum Tensile Strength, In. Dimen-
ety of Mechanical Engineers, New York, N.Y. sions, ASTM F3148-17a ASTM International, West Con-
ASME (2006), Fasteners for Use in Structural Applica- shohocken, Pa.
tions, ASME B18.2.6-06, American Society of Mechani- Carter C.J. (1996), “Specifying Bolt Length for High-
cal Engineers, New York, N.Y. Strength Bolts,” Engineering Journal, AISC, Vol.  33,
ASME (2010), Fasteners for Use in Structural Applications, No. 2, pp. 43–54.
ASME B18.2.6-10, American Society of Mechanical Moore, A.M., Rassati, G.A., and Swanson, J.A. (2010), “An
Engineers, New York, N.Y. Experimental Analysis of Strength and Ductility of High-
ASME (2019), Fasteners for Use in Structural Applications, Strength Fasteners,” Engineering Journal, AISC, Vol. 47,
ASME B18.2.6-19, American Society of Mechanical No. 3, pp. 161–174.
Engineers, New York, N.Y., adopted and in press. RCSC (2015), 2014 Specification for Structural Joints Using
ASTM (2015), Standard Specification for High Strength High-Strength Bolts with Errata, Research Council for
Structural Bolts, Steel and Alloy Steel, Heat Treated, 120 Structural Connections, Chicago, Ill.
ksi (830 MPa) and 150 ksi (1040 MPa) Minimum Ten- Shaw Jr., R.E. (2015), “New Twists and Turns in Structural
sile Strength, In. and Metric Dimensions, ASTM F3125/ Bolting,” Structure Magazine, NCSEA, November.
F3125M-15a, ASTM International, West Conshohocken,
Pa.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 21

 Table A1.  Maximum Grip Gage Length and Minimum Body Length for Common Structural Bolt Sizes

Bolt Nominal Diameter and Threads per Inch

Length 2 in.-13 s in.-11 w in.-10 d in.-9 1 in.-8 18 in.-7 14 in.-7 1a in.-6 12 in.-6
L LG,max LB,min LG,max LB,min LG,max LB,min LG,max LB,min LG,max LB,min LG,max LB,min LG,max LB,min LG,max LB,min LG,max LB,min
(in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.)
14 0.19 0.06 0.23 0.00 0.25 0.00 0.28 0.00 0.31 0.00 0.50 0.00 0.50 0.00 0.58 0.00 0.58 0.00
12 0.50 0.31 0.23 0.03 0.25 0.00 0.28 0.00 0.31 0.00 0.50 0.00 0.50 0.00 0.58 0.00 0.58 0.00
1w 0.75 0.56 0.50 0.28 0.25 0.12 0.28 0.00 0.31 0.00 0.50 0.00 0.50 0.00 0.58 0.00 0.58 0.00
2 1.00 0.81 0.75 0.53 0.62 0.37 0.28 0.22 0.31 0.00 0.50 0.00 0.50 0.00 0.58 0.00 0.58 0.00
24 1.25 1.06 1.00 0.78 0.87 0.62 0.75 0.47 0.31 0.19 0.50 0.00 0.50 0.00 0.58 0.00 0.58 0.00
22 1.50 1.31 1.25 1.03 1.12 0.87 1.00 0.72 0.75 0.44 0.50 0.16 0.50 0.12 0.58 0.00 0.58 0.00
2w 1.75 1.56 1.50 1.28 1.37 1.12 1.25 0.97 1.00 0.69 0.50 0.41 0.50 0.37 0.58 0.06 0.58 0.06
3 2.00 1.81 1.75 1.53 1.62 1.37 1.50 1.22 1.25 0.94 1.00 0.66 1.00 0.62 0.58 0.31 0.58 0.31
34 2.25 2.06 2.00 1.78 1.87 1.62 1.75 1.47 1.50 1.19 1.25 0.91 1.25 0.87 0.58 0.56 0.58 0.56
32 2.50 2.31 2.25 2.03 2.12 1.87 2.00 1.72 1.75 1.44 1.50 1.16 1.50 1.12 1.25 0.81 1.25 0.81

22 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

3w 2.75 2.56 2.50 2.28 2.37 2.12 2.25 1.97 2.00 1.69 1.75 1.41 1.75 1.37 1.50 1.06 1.50 1.06
4 3.00 2.81 2.75 2.53 2.62 2.37 2.50 2.22 2.25 1.94 2.00 1.66 2.00 1.62 1.75 1.31 1.75 1.31
44 3.25 3.06 3.00 2.78 2.87 2.62 2.75 2.47 2.50 2.19 2.25 1.91 2.25 1.87 2.00 1.56 2.00 1.56
APPENDIX

42 3.50 3.31 3.25 3.03 3.12 2.87 3.00 2.72 2.75 2.44 2.50 2.16 2.50 2.12 2.25 1.81 2.25 1.81
4w 3.75 3.56 3.50 3.28 3.37 3.12 3.25 2.97 3.00 2.69 2.75 2.41 2.75 2.37 2.50 2.06 2.50 2.06
5 4.00 3.81 3.75 3.53 3.62 3.37 3.50 3.22 3.25 2.94 3.00 2.66 3.00 2.62 2.75 2.31 2.75 2.31
52 4.50 4.31 4.25 4.03 4.12 3.87 4.00 3.72 3.75 3.44 3.50 3.16 3.50 3.12 3.25 2.81 3.25 2.81
6 5.00 4.81 4.75 4.53 4.62 4.37 4.50 4.22 4.25 3.94 4.00 3.66 4.00 3.62 3.75 3.31 3.75 3.31
62 5.50 5.31 5.25 5.03 5.12 4.87 5.00 4.72 4.75 4.44 4.50 4.16 4.50 4.12 4.25 3.81 4.25 3.81
7 6.00 5.81 5.75 5.53 5.62 5.37 5.50 5.22 5.25 4.94 5.00 4.66 5.00 4.62 4.75 4.31 4.75 4.31
72 6.50 6.31 6.25 6.03 6.12 5.87 6.00 5.72 5.75 5.44 5.50 5.16 5.50 5.12 5.25 4.81 5.25 4.81
8 7.00 6.81 6.75 6.53 6.62 6.37 6.50 6.22 6.25 5.94 6.00 5.66 6.00 5.62 5.75 5.31 5.75 5.31
82 7.50 7.31 7.25 7.03 7.12 6.87 7.00 6.72 6.75 6.44 6.50 6.16 6.50 6.12 6.25 5.81 6.25 5.81
9 8.00 7.81 7.75 7.53 7.62 7.37 7.50 7.22 7.25 6.94 7.00 6.66 7.00 6.62 6.75 6.31 6.75 6.31
92 8.50 8.31 8.25 8.03 8.12 7.87 8.00 7.72 7.75 7.44 7.50 7.16 7.50 7.12 7.25 6.81 7.25 6.81
10 9.00 8.81 8.75 8.53 8.62 8.37 8.50 8.22 8.25 7.94 8.00 7.66 8.00 7.62 7.75 7.31 7.75 7.31
Notes: Cells above the solid line represent bolt lengths that are considered to be fully threaded per ASME B18.2.6. These bolts may or may not have a shank.
Shaded cells represent bolt lengths that are rarely produced. Italics values represent bolt lengths that are available by special order only.
 Table A2.  Minimum Grip, Maximum Grip, and Thread Condition for High-Strength Bolts
2-in.-Diameter Bolts
0 Washers 1 Washer 2 Washers Σt(n−1),max
Bolt Length Min Grip Max Grip Min Grip Max Grip Min Grip Max Grip 0 Washers 1 Washer
(in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.)
14 0.15 0.63 0.06 0.45 0.00 0.27 N N
12 0.46 0.88 0.36 0.70 0.27 0.52 0.31 0.13
1w 0.71 1.13 0.61 0.95 0.52 0.77 0.56 0.38
2 0.96 1.38 0.86 1.20 0.77 1.02 0.81 0.63
24 1.21 1.63 1.11 1.45 1.02 1.27 1.06 0.88
22 1.46 1.88 1.36 1.70 1.27 1.52 1.31 1.13
2w 1.71 2.13 1.61 1.95 1.52 1.77 1.56 1.38
3 1.96 2.38 1.86 2.20 1.77 2.02 1.81 1.63
34 2.21 2.63 2.11 2.45 2.02 2.27 2.06 1.88
32 2.46 2.88 2.36 2.70 2.27 2.52 2.31 2.13
3w 2.71 3.13 2.61 2.95 2.52 2.77 2.56 2.38
4 2.96 3.38 2.86 3.20 2.77 3.02 2.81 2.63
44 3.20 3.63 3.10 3.45 3.00 3.27 3.06 2.88
42 3.45 3.88 3.35 3.70 3.25 3.52 3.31 3.13
4w 3.70 4.13 3.60 3.95 3.50 3.77 3.56 3.38
5 3.95 4.38 3.85 4.20 3.75 4.02 3.81 3.63
52 4.45 4.88 4.35 4.70 4.25 4.52 4.31 4.13
6 4.95 5.38 4.85 5.20 4.75 5.02 4.81 4.63
62 5.45 5.81 5.35 5.63 5.25 5.45 5.31 5.13
7 5.95 6.31 5.85 6.13 5.75 5.95 5.81 5.63
72 6.45 6.81 6.35 6.63 6.25 6.45 6.31 6.13
8 6.95 7.31 6.85 7.13 6.75 6.95 6.81 6.63
82 7.45 7.81 7.35 7.63 7.25 7.45 7.31 7.13
9 7.95 8.31 7.85 8.13 7.75 7.95 7.81 7.63
92 8.45 8.81 8.35 8.63 8.25 8.45 8.31 8.13
10 8.95 9.31 8.85 9.13 8.75 8.95 8.81 8.63
Notes: For Min Grip and Max Grip, consider the total number of F436 washers used under the head and under the nut.
For the Σt(n−1),max, consider only the number of F436 washers under the head of the bolt.
“N” indicates bolts that should be designed in the “threads not excluded” or “N” condition only.
Cells above the heavy line represent bolt lengths that are considered to be fully threaded per ASME B18.2.6.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 23

 Table A2.  Minimum Grip, Maximum Grip, and Thread Condition for High-Strength Bolts (continued)
s-in.-Diameter Bolts
0 Washers 1 Washer 2 Washers Σt(n−1),max
Bolt Length Min Grip Max Grip Min Grip Max Grip Min Grip Max Grip 0 Washers 1 Washer
(in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.)
14 0.18 0.50 0.06 0.32 0.00 0.15 N N
12 0.18 0.75 0.06 0.57 0.00 0.40 N N
1w 0.45 1.00 0.33 0.82 0.21 0.65 0.28 0.10
2 0.70 1.25 0.58 1.07 0.46 0.90 0.53 0.35
24 0.95 1.50 0.83 1.32 0.71 1.15 0.78 0.60
22 1.20 1.75 1.08 1.57 0.96 1.40 1.03 0.85
2w 1.45 2.00 1.33 1.82 1.21 1.65 1.28 1.10
3 1.70 2.25 1.58 2.07 1.46 1.90 1.53 1.35
34 1.95 2.50 1.83 2.32 1.71 2.15 1.78 1.60
32 2.20 2.75 2.08 2.57 1.96 2.40 2.03 1.85
3w 2.45 3.00 2.33 2.82 2.21 2.65 2.28 2.10
4 2.70 3.25 2.58 3.07 2.46 2.90 2.53 2.35
44 2.95 3.50 2.83 3.32 2.71 3.15 2.78 2.60
42 3.20 3.75 3.08 3.57 2.96 3.40 3.03 2.85
4w 3.45 4.00 3.33 3.82 3.21 3.65 3.28 3.10
5 3.70 4.25 3.58 4.07 3.46 3.90 3.53 3.35
52 4.19 4.75 4.07 4.57 3.95 4.40 4.03 3.85
6 4.69 5.25 4.57 5.07 4.45 4.90 4.53 4.35
62 5.19 5.62 5.07 5.44 4.95 5.27 5.03 4.85
7 5.69 6.12 5.57 5.94 5.45 5.77 5.53 5.35
72 6.19 6.62 6.07 6.44 5.95 6.27 6.03 5.85
8 6.69 7.12 6.57 6.94 6.45 6.77 6.53 6.35
82 7.19 7.62 7.07 7.44 6.95 7.27 7.03 6.85
9 7.69 8.12 7.57 7.94 7.45 7.77 7.53 7.35
92 8.19 8.62 8.07 8.44 7.95 8.27 8.03 7.85
10 8.69 9.12 8.57 8.94 8.45 8.77 8.53 8.35
Notes: For Min Grip and Max Grip, consider the total number of F436 washers used under the head and under the nut.
For the Σt(n−1),max, consider only the number of F436 washers under the head of the bolt.
“N” indicates bolts that should be designed in the “threads not excluded” or “N” condition only.
Cells above the heavy line represent bolt lengths that are considered to be fully threaded per ASME B18.2.6.
Shaded cells represent bolt lengths that are rarely produced.

24 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 Table A2.  Minimum Grip, Maximum Grip, and Thread Condition for High-Strength Bolts (continued)
¾-in.-Diameter Bolts
0 Washers 1 Washer 2 Washers Σt(n−1),max
Bolt Length Min Grip Max Grip Min Grip Max Grip Min Grip Max Grip 0 Washers 1 Washer
(in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.)
14 0.20 0.30 0.08 0.13 — — N N
12 0.20 0.55 0.08 0.38 0.00 0.20 N N
1w 0.20 0.80 0.08 0.63 0.00 0.45 N N
2 0.57 1.05 0.45 0.88 0.33 0.70 0.37 0.19
24 0.82 1.30 0.70 1.13 0.58 0.95 0.62 0.44
22 1.07 1.55 0.95 1.38 0.83 1.20 0.87 0.69
2w 1.32 1.80 1.20 1.63 1.08 1.45 1.12 0.94
3 1.57 2.05 1.45 1.88 1.33 1.70 1.37 1.19
34 1.82 2.30 1.70 2.13 1.58 1.95 1.62 1.44
32 2.07 2.55 1.95 2.38 1.83 2.20 1.87 1.69
3w 2.32 2.80 2.20 2.63 2.08 2.45 2.12 1.94
4 2.57 3.05 2.45 2.88 2.33 2.70 2.37 2.19
44 2.82 3.30 2.70 3.13 2.58 2.95 2.62 2.44
42 3.07 3.55 2.95 3.38 2.83 3.20 2.87 2.69
4w 3.32 3.80 3.20 3.63 3.08 3.45 3.12 2.94
5 3.57 4.05 3.45 3.88 3.33 3.70 3.37 3.19
52 4.07 4.55 3.95 4.38 3.83 4.20 3.87 3.69
6 4.57 5.05 4.45 4.88 4.33 4.70 4.37 4.19
62 5.05 5.49 4.93 5.32 4.81 5.14 4.87 4.69
7 5.55 5.99 5.43 5.82 5.31 5.64 5.37 5.19
72 6.05 6.49 5.93 6.32 5.81 6.14 5.87 5.69
8 6.55 6.99 6.43 6.82 6.31 6.64 6.37 6.19
82 7.05 7.49 6.93 7.32 6.81 7.14 6.87 6.69
9 7.55 7.99 7.43 7.82 7.31 7.64 7.37 7.19
92 8.05 8.49 7.93 8.32 7.81 8.14 7.87 7.69
10 8.55 8.99 8.43 8.82 8.31 8.64 8.37 8.19
Notes: For Min Grip and Max Grip, consider the total number of F436 washers used under the head and under the nut.
For the Σt(n−1),max, consider only the number of F436 washers under the head of the bolt.
“N” indicates bolts that should be designed in the “threads not excluded” or “N” condition only.
Cells above the heavy line represent bolt lengths that are considered to be fully threaded per ASME B18.2.6.
Shaded cells represent bolt lengths that are rarely produced.
Italics values indicate bolt lengths that are available by special order only.
“—” indicates bolt assemblies where the nut may not be able to be threaded fully onto the bolt.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 25

 Table A2.  Minimum Grip, Maximum Grip, and Thread Condition for High-Strength Bolts (continued)
d-in.-Diameter Bolts
0 Washers 1 Washer 2 Washers Σt(n−1),max
Bolt Length Min Grip Max Grip Min Grip Max Grip Min Grip Max Grip 0 Washers 1 Washer
(in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.)
14 0.22 0.18 — — — — N N
12 0.22 0.43 0.09 0.25 0.00 0.07 N N
1w 0.22 0.68 0.09 0.50 0.00 0.32 N N
2 0.22 0.93 0.09 0.75 0.00 0.57 N N
24 0.69 1.18 0.56 1.00 0.42 0.82 0.47 0.29
22 0.94 1.43 0.81 1.25 0.67 1.07 0.72 0.54
2w 1.19 1.68 1.06 1.50 0.92 1.32 0.97 0.79
3 1.44 1.93 1.31 1.75 1.17 1.57 1.22 1.04
34 1.69 2.18 1.56 2.00 1.42 1.82 1.47 1.29
32 1.94 2.43 1.81 2.25 1.67 2.07 1.72 1.54
3w 2.19 2.68 2.06 2.50 1.92 2.32 1.97 1.79
4 2.44 2.93 2.31 2.75 2.17 2.57 2.22 2.04
44 2.69 3.18 2.56 3.00 2.42 2.82 2.47 2.29
42 2.94 3.43 2.81 3.25 2.67 3.07 2.72 2.54
4w 3.19 3.68 3.06 3.50 2.92 3.32 2.97 2.79
5 3.44 3.93 3.31 3.75 3.17 3.57 3.22 3.04
52 3.94 4.43 3.81 4.25 3.67 4.07 3.72 3.54
6 4.44 4.93 4.31 4.75 4.17 4.57 4.22 4.04
62 4.94 5.37 4.81 5.19 4.67 5.01 4.72 4.54
7 5.44 5.87 5.31 5.69 5.17 5.51 5.22 5.04
72 5.93 6.37 5.79 6.19 5.65 6.01 5.72 5.54
8 6.43 6.87 6.29 6.69 6.15 6.51 6.22 6.04
82 6.93 7.37 6.79 7.19 6.65 7.01 6.72 6.54
9 7.43 7.87 7.29 7.69 7.15 7.51 7.22 7.04
92 7.93 8.37 7.79 8.19 7.65 8.01 7.72 7.54
10 8.43 8.87 8.29 8.69 8.15 8.51 8.22 8.04
Notes: For Min Grip and Max Grip, consider the total number of F436 washers used under the head and under the nut.
For the Σt(n−1),max, consider only the number of F436 washers under the head of the bolt.
“N” indicates bolts that should be designed in the “threads not excluded” or “N” condition only.
Cells above the heavy line represent bolt lengths that are considered to be fully threaded per ASME B18.2.6.
Shaded cells represent bolt lengths that are rarely produced.
Italics values indicate bolt lengths that are available by special order only.
“—” indicates bolt assemblies where the nut may not be able to be threaded fully onto the bolt.

26 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 Table A2.  Minimum Grip, Maximum Grip, and Thread Condition for High-Strength Bolts (continued)
1-in.-Diameter Bolts
0 Washers 1 Washer 2 Washers Σt(n−1),max
Bolt Length Min Grip Max Grip Min Grip Max Grip Min Grip Max Grip 0 Washers 1 Washer
(in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.)
14 — — — — — — — —
12 — — — — — — — —
1w 0.25 0.55 0.11 0.37 0.00 0.19 N N
2 0.25 0.80 0.11 0.62 0.00 0.44 N N
24 0.25 1.05 0.11 0.87 0.00 0.69 N N
22 0.69 1.30 0.55 1.12 0.42 0.94 0.44 0.26
2w 0.94 1.55 0.80 1.37 0.67 1.19 0.69 0.51
3 1.19 1.80 1.05 1.62 0.92 1.44 0.94 0.76
34 1.44 2.05 1.30 1.87 1.17 1.69 1.19 1.01
32 1.69 2.30 1.55 2.12 1.42 1.94 1.44 1.26
3w 1.94 2.55 1.80 2.37 1.67 2.19 1.69 1.51
4 2.19 2.80 2.05 2.62 1.92 2.44 1.94 1.76
44 2.44 3.05 2.30 2.87 2.17 2.69 2.19 2.01
42 2.69 3.30 2.55 3.12 2.42 2.94 2.44 2.26
4w 2.94 3.55 2.80 3.37 2.67 3.19 2.69 2.51
5 3.19 3.80 3.05 3.62 2.92 3.44 2.94 2.76
52 3.69 4.30 3.55 4.12 3.42 3.94 3.44 3.26
6 4.19 4.80 4.05 4.62 3.92 4.44 3.94 3.76
62 4.69 5.24 4.55 5.06 4.42 4.88 4.44 4.26
7 5.19 5.74 5.05 5.56 4.92 5.38 4.94 4.76
72 5.69 6.24 5.55 6.06 5.42 5.88 5.44 5.26
8 6.19 6.74 6.05 6.56 5.92 6.38 5.94 5.76
82 6.67 7.24 6.53 7.06 6.39 6.88 6.44 6.26
9 7.17 7.74 7.03 7.56 6.89 7.38 6.94 6.76
92 7.67 8.24 7.53 8.06 7.39 7.88 7.44 7.26
10 8.17 8.74 8.03 8.56 7.89 8.38 7.94 7.76
Notes: For Min Grip and Max Grip, consider the total number of F436 washers used under the head and under the nut.
For the Σt(n−1),max, consider only the number of F436 washers under the head of the bolt.
“N” indicates bolts that should be designed in the “threads not excluded” or “N” condition only.
Cells above the heavy line represent bolt lengths that are considered to be fully threaded per ASME B18.2.6.
Shaded cells represent bolt lengths that are rarely produced.
Italics values indicate bolt lengths that are available by special order only.
“—” indicates bolt assemblies where the nut may not be able to be threaded fully onto the bolt.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 27

 Table A2.  Minimum Grip, Maximum Grip, and Thread Condition for High-Strength Bolts (continued)
18-in.-Diameter Bolts
0 Washers 1 Washer 2 Washers Σt(n−1),max
Bolt Length Min Grip Max Grip Min Grip Max Grip Min Grip Max Grip 0 Washers 1 Washer
(in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.)
14 — — — — — — — —
12 — — — — — — — —
1w — — — — — — — —
2 0.43 0.61 0.29 0.43 0.16 0.26 N N
24 0.43 0.86 0.29 0.68 0.16 0.51 N N
22 0.43 1.11 0.29 0.93 0.16 0.76 N N
2w 0.43 1.36 0.29 1.18 0.16 1.01 N N
3 0.93 1.61 0.79 1.43 0.66 1.26 0.66 0.48
34 1.18 1.86 1.04 1.68 0.91 1.51 0.91 0.73
32 1.43 2.11 1.29 1.93 1.16 1.76 1.16 0.98
3w 1.68 2.36 1.54 2.18 1.41 2.01 1.41 1.23
4 1.93 2.61 1.79 2.43 1.66 2.26 1.66 1.48
44 2.18 2.86 2.04 2.68 1.91 2.51 1.91 1.73
42 2.43 3.11 2.29 2.93 2.16 2.76 2.16 1.98
4w 2.68 3.36 2.54 3.18 2.41 3.01 2.41 2.23
5 2.93 3.61 2.79 3.43 2.66 3.26 2.66 2.48
52 3.43 4.11 3.29 3.93 3.16 3.76 3.16 2.98
6 3.93 4.61 3.79 4.43 3.66 4.26 3.66 3.48
62 4.43 5.11 4.29 4.93 4.16 4.76 4.16 3.98
7 4.93 5.61 4.79 5.43 4.66 5.26 4.66 4.48
72 5.43 6.11 5.29 5.93 5.16 5.76 5.16 4.98
8 5.93 6.61 5.79 6.43 5.66 6.26 5.66 5.48
82 6.43 7.11 6.29 6.93 6.16 6.76 6.16 5.98
9 6.93 7.61 6.79 7.43 6.66 7.26 6.66 6.48
92 7.40 8.11 7.27 7.93 7.13 7.76 7.16 6.98
10 7.90 8.61 7.77 8.43 7.63 8.26 7.66 7.48
Notes: For Min Grip and Max Grip, consider the total number of F436 washers used under the head and under the nut.
For the Σt(n−1),max, consider only the number of F436 washers under the head of the bolt.
“N” indicates bolts that should be designed in the “threads not excluded” or “N” condition only.
Cells above the heavy line represent bolt lengths that are considered to be fully threaded per ASME B18.2.6.
Shaded cells represent bolt lengths that are rarely produced.
Italics values indicate bolt lengths that are available by special order only.
“—” indicates bolt assemblies where the nut may not be able to be threaded fully onto the bolt.

28 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 Table A2.  Minimum Grip, Maximum Grip, and Thread Condition for High-Strength Bolts (continued)
14-in.-Diameter Bolts
0 Washers 1 Washer 2 Washers Σt(n−1),max
Bolt Length Min Grip Max Grip Min Grip Max Grip Min Grip Max Grip 0 Washers 1 Washer
(in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.)
14 — — — — — — — —
12 — — — — — — — —
1w — — — — — — — —
2 0.43 0.50 0.29 0.32 — — N N
24 0.43 0.75 0.29 0.57 0.16 0.40 N N
22 0.43 1.00 0.29 0.82 0.16 0.65 N N
2w 0.43 1.25 0.29 1.07 0.16 0.90 N N
3 0.93 1.50 0.79 1.32 0.66 1.15 0.62 0.44
34 1.18 1.75 1.04 1.57 0.91 1.40 0.87 0.69
32 1.43 2.00 1.29 1.82 1.16 1.65 1.12 0.94
3w 1.68 2.25 1.54 2.07 1.41 1.90 1.37 1.19
4 1.93 2.50 1.79 2.32 1.66 2.15 1.62 1.44
44 2.18 2.75 2.04 2.57 1.91 2.40 1.87 1.69
42 2.43 3.00 2.29 2.82 2.16 2.65 2.12 1.94
4w 2.68 3.25 2.54 3.07 2.41 2.90 2.37 2.19
5 2.93 3.50 2.79 3.32 2.66 3.15 2.62 2.44
52 3.43 4.00 3.29 3.82 3.16 3.65 3.12 2.94
6 3.93 4.50 3.79 4.32 3.66 4.15 3.62 3.44
62 4.43 5.00 4.29 4.82 4.16 4.65 4.12 3.94
7 4.93 5.50 4.79 5.32 4.66 5.15 4.62 4.44
72 5.43 6.00 5.29 5.82 5.16 5.65 5.12 4.94
8 5.93 6.50 5.79 6.32 5.66 6.15 5.62 5.44
82 6.43 7.00 6.29 6.82 6.16 6.65 6.12 5.94
9 6.93 7.50 6.79 7.32 6.66 7.15 6.62 6.44
92 7.43 8.00 7.29 7.82 7.16 7.65 7.12 6.94
10 7.93 8.50 7.79 8.32 7.66 8.15 7.62 7.44
Notes: For Min Grip and Max Grip, consider the total number of F436 washers used under the head and under the nut.
For the Σt(n−1),max, consider only the number of F436 washers under the head of the bolt.
“N” indicates bolts that should be designed in the “threads not excluded” or “N” condition only.
Cells above the heavy line represent bolt lengths that are considered to be fully threaded per ASME B18.2.6.
Shaded cells represent bolt lengths that are rarely produced.
Italics values indicate bolt lengths that are available by special order only.
“—” indicates bolt assemblies where the nut may not be able to be threaded fully onto the bolt.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 29

 Table A2.  Minimum Grip, Maximum Grip, and Thread Condition for High-Strength Bolts (continued)
1a-in.-Diameter Bolts
0 Washers 1 Washer 2 Washers Σt(n−1),max
Bolt Length Min Grip Max Grip Min Grip Max Grip Min Grip Max Grip 0 Washers 1 Washer
(in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.)
14 — — — — — — — —
12 — — — — — — — —
1w — — — — — — — —
2 — — — — — — — —
24 0.50 0.62 0.36 0.45 0.23 0.27 N N
22 0.50 0.87 0.36 0.70 0.23 0.52 N N
2w 0.50 1.12 0.36 0.95 0.23 0.77 N N
3 0.50 1.37 0.36 1.20 0.23 1.02 N N
34 0.50 1.62 0.36 1.45 0.23 1.27 N N
32 1.17 1.87 1.03 1.70 0.89 1.52 0.81 0.63
3w 1.42 2.12 1.28 1.95 1.14 1.77 1.06 0.88
4 1.67 2.37 1.53 2.20 1.39 2.02 1.31 1.13
44 1.92 2.62 1.78 2.45 1.64 2.27 1.56 1.38
42 2.17 2.87 2.03 2.70 1.89 2.52 1.81 1.63
4w 2.42 3.12 2.28 2.95 2.14 2.77 2.06 1.88
5 2.67 3.37 2.53 3.20 2.39 3.02 2.31 2.13
52 3.17 3.87 3.03 3.70 2.89 3.52 2.81 2.63
6 3.67 4.37 3.53 4.20 3.39 4.02 3.31 3.13
62 4.17 4.87 4.03 4.70 3.89 4.52 3.81 3.63
7 4.67 5.37 4.53 5.20 4.39 5.02 4.31 4.13
72 5.17 5.87 5.03 5.70 4.89 5.52 4.81 4.63
8 5.67 6.37 5.53 6.20 5.39 6.02 5.31 5.13
82 6.17 6.87 6.03 6.70 5.89 6.52 5.81 5.63
9 6.67 7.37 6.53 7.20 6.39 7.02 6.31 6.13
92 7.17 7.87 7.03 7.70 6.89 7.52 6.81 6.63
10 7.67 8.37 7.53 8.20 7.39 8.02 7.31 7.13
Notes: For Min Grip and Max Grip, consider the total number of F436 washers used under the head and under the nut.
For the Σt(n−1),max, consider only the number of F436 washers under the head of the bolt.
“N” indicates bolts that should be designed in the “threads not excluded” or “N” condition only.
Cells above the heavy line represent bolt lengths that are considered to be fully threaded per ASME B18.2.6.
Shaded cells represent bolt lengths that are rarely produced.
Italics values indicate bolt lengths that are available by special order only.
“—” indicates bolt assemblies where the nut may not be able to be threaded fully onto the bolt.

30 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 Table A2.  Minimum Grip, Maximum Grip, and Thread Condition for High-Strength Bolts (continued)
12-in.-Diameter Bolts
0 Washers 1 Washer 2 Washers Σt(n−1),max
Bolt Length Min Grip Max Grip Min Grip Max Grip Min Grip Max Grip 0 Washers 1 Washer
(in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.)
14 — — — — — — — —
12 — — — — — — — —
1w — — — — — — — —
2 — — — — — — — —
24 — — — — — — — —
22 0.50 0.75 0.36 0.57 0.23 0.39 N N
2w 0.50 1.00 0.36 0.82 0.23 0.64 N N
3 0.50 1.25 0.36 1.07 0.23 0.89 N N
34 0.50 1.50 0.36 1.32 0.23 1.14 N N
32 1.17 1.75 1.03 1.57 0.89 1.39 0.81 0.63
3w 1.42 2.00 1.28 1.82 1.14 1.64 1.06 0.88
4 1.67 2.25 1.53 2.07 1.39 1.89 1.31 1.13
44 1.92 2.50 1.78 2.32 1.64 2.14 1.56 1.38
42 2.17 2.75 2.03 2.57 1.89 2.39 1.81 1.63
4w 2.42 3.00 2.28 2.82 2.14 2.64 2.06 1.88
5 2.67 3.25 2.53 3.07 2.39 2.89 2.31 2.13
52 3.17 3.75 3.03 3.57 2.89 3.39 2.81 2.63
6 3.67 4.25 3.53 4.07 3.39 3.89 3.31 3.13
62 4.17 4.75 4.03 4.57 3.89 4.39 3.81 3.63
7 4.67 5.25 4.53 5.07 4.39 4.89 4.31 4.13
72 5.17 5.75 5.03 5.57 4.89 5.39 4.81 4.63
8 5.67 6.25 5.53 6.07 5.39 5.89 5.31 5.13
82 6.17 6.75 6.03 6.57 5.89 6.39 5.81 5.63
9 6.67 7.25 6.53 7.07 6.39 6.89 6.31 6.13
92 7.17 7.75 7.03 7.57 6.89 7.39 6.81 6.63
10 7.67 8.25 7.53 8.07 7.39 7.89 7.31 7.13
Notes: For Min Grip and Max Grip, consider the total number of F436 washers used under the head and under the nut.
For the Σt(n−1),max, consider only the number of F436 washers under the head of the bolt.
“N” indicates bolts that should be designed in the “threads not excluded” or “N” condition only.
Cells above the heavy line represent bolt lengths that are considered to be fully threaded per ASME B18.2.6.
Shaded cells represent bolt lengths that are rarely produced.
Italics values indicate bolt lengths that are available by special order only.
“—” indicates bolt assemblies where the nut may not be able to be threaded fully onto the bolt.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 31

32 / ENGINEERING JOURNAL / FIRST QUARTER / 2020
A Reliability Study of Joints With Bolts Designed
With Threads Excluded but Installed With
Threads Not Excluded
JAMES A. SWANSON, GIAN ANDREA RASSATI, and CHAD M. LARSON

ABSTRACT
This paper presents a reliability and probability study focusing on connections using relatively short bolts that have been shown in a com-
panion paper to have the potential to have been designed with threads excluded from the shear plane and then subsequently installed with
the threads not excluded from the shear plane. After an introduction outlining the background of the shear strength and associated design
of joints in various editions of the AISC Specification, the paper presents a structural reliability analysis as well as a probability study using
Monte Carlo simulations; finally, the paper discusses additional considerations and mitigating factors associated with this potential problem.
Calculated reliability coefficients and probabilities of failure are tabulated for joints using two diameter groups of 120-ksi bolts (from s in. to
1 in. and from 1s in. to 14 in.) and for joints using 150-ksi bolts. The paper provides an evaluation of the reliability of joints with bolts that have
been designed with the threads excluded from the shear plane but installed with the threads not excluded from the shear plane. Although it is
recommended that future designs involving short bolts be based on the assumption that the threads are not excluded from the shear plane,
this study provides a measure of the reliability of structures that have already been constructed with bolts designed assuming that the threads
were excluded but installed with the threads not excluded. The results show that the reliability of joints in this class is dependent on the grade
and size of the bolts used, on the length of the joint, and on which edition of the AISC Specification was used for design. It was found that
some joints in this class still meet the AISC target reliability for connections and that many joints meet the AISC target reliability for members.

Keywords:  structural bolt, high-strength bolt, fastener, A325, A490, F1852, F2280, F3125, threads excluded, threads not excluded, shear.

INTRODUCTION Specification (RCSC, 2015). In these tables, the thread

lengths of structural bolts are presented as constant for each
W hen designing a bolted joint for shear forces, one of the
first determinations that is made is whether the bolts
will be designed with the shear plane of the joint passing
given bolt diameter. Structural bolts, however, are manufac-
tured to conform to the American Society of Mechanical
Engineers (ASME) B18.2.6 Standard (ASME, 2019) where
through the shank or through the threads of the bolts. In the
the thread length is presented as a reference dimension and
former case where the threads are excluded from the shear
not a control dimension, which means that the thread length
plane (the X condition), the bolts have an available strength
may not be exactly equal to the values shown in tables.
that is higher than in the latter case where the threads are not
This situation often manifests itself in structural bolts
excluded from the shear plane (the N condition). To deter-
that are relatively short but still longer than the reference
mine whether or not threads will be excluded from the shear
thread length for their diameter, where the bolt may not have
plane, engineers and detailers often consult Table 7-14 of the
an unthreaded shank at all despite what is implied in some
AISC Steel Construction Manual (AISC, 2017), hereafter
design aids. Figure 1 shows three different d‑in. × 2‑in. bolts
referred to as the AISC Manual, or Tables  C-2.1 or C-2.2
that were made by three different producers, all in compli-
of the Research Council on Structural Connections (RCSC)
ance with ASME standards. Structural bolts that are d in.
in diameter have a published thread length of 12 in., which
leads to the assumption that a bolt that is 2 in. long should
have an unthreaded shank that is 2 in. long. Provisions in
James A. Swanson, PhD, Associate Professor, Department of Civil and Archi- the ASME standard, however, require this bolt to be man-
tectural Engineering and Construction Management, University of Cincinnati, ufactured as fully threaded, with a short unthreaded body
Cincinnati, Ohio. Email: [email protected] (corresponding)
with a diameter smaller than the nominal bolt diameter, or
Gian Andrea Rassati, PhD, Associate Professor, Department of Civil and
Architectural Engineering and Construction Management, University of Cin-
with an unthreaded shank with a diameter equal to the full
cinnati, Cincinnati, Ohio. Email: [email protected] nominal diameter of the bolt. This issue was addressed in
Chad M. Larson, President, LeJeune Bolt Co., Burnsville, Minn. Email: detail in a previous work (Swanson et al., 2019); Table 1 lists
[email protected] bolts that may be affected by this issue.
In addition to the issues noted for relatively short bolts,
it is possible for bolts of any length to have geometric
Paper No. 2019-15

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 33

 Table 1.  Bolt Lengths Considered to Be Fully Threaded Per ASME B18.2.6
Nominal Bolt Diameter, db Fully Threaded Bolt Lengths
2 in. L ≤ 14 in.
s in. L ≤ 12 in.
w in. L ≤ 1w in.
d in. L ≤ 2 in.
1 in. L ≤ 24 in.
18 in. L ≤ 2w in.
14 in. L ≤ 2w in.
1a in. L ≤ 34 in.
12 in. L ≤ 34 in.

deviations from nominal reference dimensions. A common This situation presents a potential problem for engineers
misunderstanding is that the length of the unthreaded shank and owners with structures already constructed with short
or body of a bolt can be computed simply by subtracting the bolts that have been designed in the threads excluded condi-
reference thread length from the overall length of the bolt. tion. Bolts that were designed to carry shear forces through
This is incorrect, though, because of tolerances on the length their shanks may actually be carrying shear forces through
and transition length of the bolt. A footnote to Table  7-14 their threads or transitions, meaning that they will likely
of the AISC Manual provides a reference to tolerances in have an available strength that is lower than expected.
ASME B18.2.6, where minimum body lengths (as well as Both the AISC Specification for Structural Steel Buildings
other dimensions) are found. The minimum body length, as (AISC, 2016) and the RCSC Specification include a factor
a control dimension, is not permitted to be smaller than pub- of 0.80 for bolts designed in the threads not excluded condi-
lished values. A d‑in. × 22‑in. bolt, for example, might be tion, thus bolts designed as X but installed as N may have
expected to have a shank length of 22 in. − 12 in. = 1 in. 20% less strength than expected. While this difference is
computed by subtracting the thread length from the overall less than the margin afforded by the resistance factor of 0.75
length. Consultation with Table 2 of ASME B18.2.6, how- that is used in the design of bolted joints,1 this may not be a
ever, shows a minimum body length, L B,min, of 0.72 in. for sufficient remedy in some circumstances.
that bolt. While it is sometimes argued that a bolt sheared It should be noted that this issue has existed for nearly
through its transition region demonstrates behavior compa- four decades; it can be traced back to at least the 1981 edi-
rable to a bolt sheared through its body; that may not actu- tion of ASME B18.2.1 (ASME, 1981). Bolts designed with
ally be the case. threads excluded from the shear plane have been discovered
during installation to be fully threaded on numerous occa-
sions, often resulting in replacement of the fasteners with
longer ones, often at great expense. While it is suspected
that this variation has likely gone undiscovered in the design
and construction of numerous other structures, the authors
are not aware of a structural failure that has resulted from
this specific issue. While this issue can certainly result in
designs that are unconservative, those designs might actu-
ally be only less conservative but still adequate when com-
pared to actual demand. Application of resistance factors,
documented overstrength of fasteners, conservatism built
into design equations, and other factors mitigate the risk

1 Note that there are several differences between the AASHTO LRFD
specification (AASHTO, 2017) relative to the AISC and RCSC specifi-
Fig. 1.  Three d -in.-9×2-in.-bolts cations, including the use of a resistance factor of 0.80 instead of 0.75.
made by three different manufacturers. These issues will be addressed briefly later in the paper.

34 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

associated with these incorrect design assumptions. The fol- is taken as 1.00 for joints with a length up to and including
lowing sections aim to quantify the reliability and risk asso- 38 in., but a value of Rlj = 0.80 is used for joints longer than
ciated with this issue. 38 in. The factor Rlj is similar to the factor Rj but accounts
for a more pronounced inequality in the distribution of shear
forces among the bolts in longer joints.
BACKGROUND OF BOLT SHEAR STRENGTH
According to the 2016 edition of the AISC Specification,
The most fundamental form of the design equation for a for bolts in a joint with a length not exceeding 38 in., where
connection, written in terms of load and resistance factor the threads are excluded from the shear plane (X), Rv  =
design (LRFD), is 0.625, Rnx = 1.00, Rj = 0.90, and Rlj = 1.00. For bolts in a joint
that has a length not exceeding 38  in., where the threads
ϕRn ≥ ΣγQ (1)
are not excluded from the shear plane (N), Rv = 0.625, Rnx =
where 0.80, Rj = 0.90, and Rlj = 1.00.
Q = applied load or load effect, kips Threads excluded, X:
Rn = nominal resistance, kips rnv = 0.563AbFu,bolt(3)
γ = load factor Threads not excluded, N:
ϕ = resistance factor rnv = 0.450AbFu,bolt(4)
On the left-hand side of the equation, we find the factored All bolts:
resistance, ϕRn, and this is a logical place to start. There are rnv = AbFu,bolt(5)
numerous limit states for a bolted joint where the bolts are
subjected to shear, including the shear strength of the bolt, where
the bearing and/or tear-out strength of the material around Fnv = the nominal strength per unit area of the bolt in
the bolt, block shear, a net section fracture, and possibly shear, ksi
joint slip depending on the connection type and the require- = RvRnxRjRljFu,bolt(6)
ments of the engineer of record.
The first two factors, Rv and Rnx, address the strength of indi-
vidual bolts and the variability of those strengths, whereas
Bolt Shear Strength
the third and fourth factors, Rj and Rlj, address the perfor-
The nominal shear strength of a bolt, rnv, in its fundamental mance of bolt groups in joints and the variability of those
form can be shown as (Kulak et al., 1987; Tide, 2010) joints.
rnv = AbFu,boltRvRnxRjRlj(2)
The Strength of Individual Bolts in Shear with
where: Threads Not Excluded
Ab = nominal unthreaded body area of the bolt, in.2
Three research studies have been conducted at the Univer-
Fu,bolt = ultimate tensile strength of the bolt material, ksi sity of Cincinnati since 2008 wherein the strength of indi-
Rj = length reduction factor for joints with L ≤ 38 in. vidual bolts was measured (Moore et al., 2008; Taylor et al.,
Rlj = length reduction factor for joints with L > 38 in. 2008; Roenker et al., 2017). These studies included ASTM
Rnx = thread condition factor (X vs N) A325, A325T, A490, F1852, and F2280 bolts of various
Rv = ratio of shear strength to tensile strength diameters manufactured by approximately a dozen different
producers.2 A summary of the grades and diameters of bolts
This form of the nominal strength equation includes the tested is shown in Table 2. All bolts considered herein were
area of the shank of the bolt, the ultimate tensile strength tested in single shear with the shear plane passing through
of the bolt material, and four reduction factors. A value of the threads. For the sake of clarity, A325, A325T, and F1852
0.625 is used for the first reduction factor, Rv, which is the bolts are treated as equivalent and are collectively referred
ratio of the shear strength of the bolt material to the ten- to as Group 120. Similarly, A490 (ASTM, 2014b) and F2280
sile strength of the bolt material (AISC, 2016). The second (ASTM, 2014d) bolts are also treated as equivalent and are
reduction factor, Rnx, accounts for the thread condition of the collectively referred to as Group 150.
bolts in the joint; a value of Rnx = 1.00 is used if the threads The results of the bolt tests are summarized in Fig-
are excluded (X) from the shear plane of the joint, and a ures 2 through 5 where the strength of bolts divided by the
value of Rnx = 0.80 is used when the threads are not excluded
(N) from the shear plane. A value of 0.90 is used for the third
factor, Rj, which represents a reduction in the strength of the
2 Test results that are referenced in this work were obtained for bolts that
bolts due to unequal distribution of shear force among the were produced before the ASTM F3125 (2018) standard was officially
bolts in typical joints having lengths up to 38 in. Finally, Rlj adopted.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 35

 Table 2.  Number of Each Grade and Diameter of Bolts Tested with Threads Not Excluded
Bolt Grade
Diameter, in. A325 A325T F1852 A490 F2280
2 0 0 0 0 0
s 20 15 5 30 0
w 86 50 10 64 15
d 100 15 5 104 15
1 79 20 5 92 15
18 21 10 5 25 5
14 20 25 0 30 0
1a 0 0 0 0 0
12 0 0 0 0 0

nominal area of their unthreaded shanks is shown as histo- A325 and A325TC bolts larger than 1 in. in diameter have
grams. Figure 2 shows data for Group 120 bolts sized s in. a minimum strength of Fu = 120 ksi. Prior to the adoption
through 14 in. It was observed, however, that the strength of ASTM F3125, ASTM A325 (2014a) and ASTM F1852
of Group 120 bolts exhibited a bimodal distribution where (2014c) required that bolts larger than 1 in. in diameter have
the strength of smaller diameter bolts sized s  in. through a minimum tensile strength of only Fu = 105 ksi. A bimodal
1 in., shown in Figure 3, was noted to be higher than that of distribution was not observed in the data for Group  150
larger diameter bolts sized 18 in. through 14 in., shown in bolts, shown in Figure 5.
Figure  4. When the larger-diameter Group 120 bolts were It was also noted that the mean shear strength of Group 150
considered separately as a subset of the complete dataset, bolts, 74.6 ksi, was slightly lower than the minimum required
a lower mean strength and lower standard deviation were value of Fu,bolt Rv Rnx  = (150  ksi)(0.625)(0.80)  = 75.0  ksi. In
observed for this class of fasteners. This difference may be fact, more than half of the Group 150 bolts that were tested
related to challenges associated with hardening the larger (211 of 395) had a shear strength less than 75.0 ksi. In con-
diameter bolts and may also be a reflection of the fact that trast, the average shear strength of the Group 120 bolts was
tests were performed on bolts manufactured prior to the greater than the minimum required value of (120 ksi)(0.625)
adoption of ASTM F3125 (2018), which requires that Grade (0.80) = 60.0 ksi, and only 10 of 491 Group 120 bolts tested

120

Group 120 Bolts n = 491 bolts

100 Threads N Ave = 68.5 ksi
d = s" to 14" St Dev = 4.60 ksi
80
Max = 94.3 ksi
Min = 57.4 ksi
Samples, n

60

40

20

0
40.00

42.50

45.00

47.50

50.00

52.50

55.00

57.50

60.00

62.50

65.00

67.50

70.00

72.50

75.00

77.50

80.00

82.50

85.00

87.50

90.00

Measured Ultimate Shear Strength per Unit Area, Fuv (ksi)

Fig. 2.  Shear strength per unit area of Group 120 bolts tested with threads not excluded.

36 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 120

Group 120 Bolts n = 410 bolts

100 Threads N Ave = 69.5 ksi
d = s" to 1" St Dev = 4.27 ksi
Max = 94.3 ksi
80
Min = 58.7 ksi
Samples, n

60

40

20

0
40.0

42.5

45.0

47.5

50.0

52.5

55.0

57.5

60.0

62.5

65.0

67.5

70.0

72.5

75.0

77.5

80.0

82.5

85.0

87.5

90.0
Measured Ultimate Shear Strength per Unit Area, Fuv (ksi)

Fig. 3.  Shear strength per unit area of smaller-diameter Group 120 bolts tested with threads not excluded.

30

Group 120 Bolts n = 81 bolts

25 Threads N Ave = 63.7 ksi
d = 18" to 14" St Dev = 2.89 ksi
Max = 70.2 ksi
20
Min = 57.4 ksi
Samples, n

15

10

0
40.0

42.5

45.0

47.5

50.0

52.5

55.0

57.5

60.0

62.5

65.0

67.5

70.0

72.5

75.0

77.5

80.0

82.5

85.0

87.5

90.0

Measured Ultimate Shear Strength per Unit Area, Fuv (ksi)

Fig. 4.  Shear strength per unit area of larger-diameter Group 120 bolts tested with threads not excluded.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 37

had a strength less than 60.0 ksi. This is thought to be a reflec- joint tend to experience loads that are lower than the average.
tion of the fact that, unlike Group 120 bolts, Group 150 bolts Kulak et al. further reported that the extent of the nonuni-
have a maximum specified tensile strength, which results in form load distribution was more severe in longer joints than
Group 150 bolts being manufactured with a strength that is in shorter joints. Kulak et al. recommended that a reduc-
closer to their minimum specified strength than Group 120 tion factor, referred to as Rj herein, of 0.80 be applied to the
bolts. This is also evidence that the value of Rnx = 0.80 might computed strength of bolts in all joints and that a second
possibly be too large. This reduction factor was proposed reduction factor, referred to as Rlj herein, of 0.80 be applied
as 0.75 by Fisher and Struik (1974), was then proposed as to the computed strength of bolts in joints that have a length
0.70 by Kulak et al. (1987), was reported as 0.833 by Frank greater than 50 in.
and Yura (1981), and was then reported as 0.762 by Moore The recommendations of Kulak et al. are reflected in
et al. (2008; 2010). Early versions of the RCSC Specifica- AISC Specifications up to and including the 2005 edition
tion incorporated a value of approximately 0.68 (RCSC, (AISC, 2005). The reduction factors were modified in
1964) but this was increased slightly to 0.70 in the 1976 edi- the 2010 edition of the AISC Specification (AISC, 2010),
tion (RCSC, 1976). The value of 0.80 was adopted with the though, based in part on work by Tide (2010), such that the
release of the LRFD version of the RCSC Specification in factor Rj of 0.90 is applied to joints of all lengths and the fac-
1988 (RCSC, 1988), though the value of 0.70 was retained in tor Rlj of 0.833 is applied to joints exceeding 38 in. in length,
the 1985 and 1994 editions of the ASD version of the RCSC as shown in Figure  6. The limit of L  > 38  in. was chosen
Specification (RCSC, 1985). This evolution was mirrored in based on the notion that few joints have a length equal to
the AISC Specification development with the exception that 38 in. using typical bolt spacings.
a value of 0.75 was adopted in the first edition of the AISC It should be noted, however, that Kulak et al. (1987)
LRFD Specification (AISC, 1986), but this was increased to reported that the distribution of load within a bolt group is
0.80 in the second edition (AISC, 1993). approximately uniform for short or compact bolt groups that
have a length up to approximately 10 in. The commentary
Consideration of the Effect of Joint Length to the 2016 AISC Specification states that, “[i]n connections
consisting of only a few fasteners and length not exceeding
In most designs, the strength of a group of bolts is taken as
approximately 16 in. the effect of differential strain on the
the sum of the individual strengths of each bolt in the group.
shear in bearing fasteners is negligible.” Further, the com-
It has been shown through testing of end-loaded bolted
mentary goes on to say that, “[f]or shear-type connections
shear joints, however, that the load transferred through the
used in beams and girders with lengths greater than 38 in.,
joint may not be equally shared among the bolts in the joint.
there is no need to make the second reduction,” and provides
Kulak et al. (1987) reported that the bolts near the beginning
examples shown in Figure 7 as references.
and end of a shear joint tend to experience loads that are
In this paper, compact joints will be defined as joints
higher than the average, while bolts near the middle of the
with a length, L, less than or equal to 16 in. Conversely, long

120

Group 150 Bolts n = 395 bolts

100 Threads N Ave = 74.6 ksi
d = s" to 14" St Dev = 3.81 ksi
80
Max = 87.5 ksi
Min = 60.1 ksi
Samples, n

60

40

20

0
100.0
50.0

52.5

55.0

57.5

60.0

62.5

65.0

67.5

70.0

72.5

75.0

77.5

80.0

82.5

85.0

87.5

90.0

92.5

95.0

97.5

Measured Ultimate Shear Strength per Unit Area, Fuv (ksi)

Fig. 5.  Shear strength per unit area of Group 150 bolts tested with threads not excluded.

38 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 Table 3.  Joint Efficiency, RjRlj, Based on Distribution of Bolt Forces in Shear Joints
(Tide, 2010)
AISC Specification Prior to 2010 AISC Specification 2010 and 2016
Standard Standard
Mean Deviation Mean Deviation
Compact 0.970 0.045 0.970 0.045
Intermediate 0.831 0.122 0.834 0.101
Long 0.820 0.160 0.821 0.163
All 0.881 0.598 0.881 0.598

joints will be defined herein as those joints with a length with lengths up to and including 14 in., comprising from two
greater than 50 in. when taken in the context of AISC Speci- to five rows of bolts arranged perpendicular to the applied
fications prior to 2010 and joints with a length greater than load. For those compact joints, the average value of RjRlj was
38 in. when taken in the context of the 2010 or 2016 AISC found to be 0.970 with a standard deviation of 0.045. These
Specifications. Joints that are neither compact nor long are values are shown in Table 3 along with means and standard
defined as intermediate length. deviations for compact, intermediate, and long joints as the
In 2010, Tide (2010) presented a discussion of the effects joint efficiency, which is defined within this context as the
of nonuniform loading of bolts in joints. Tide summarized ratio of the joint strength to the sum of the strengths of the
72 experiments wherein the value of RjRlj was measured for individual bolts in the joint.
joints ranging in length from 3.50 in. to 94.0 in., shown in Tide presented an equation for RjRlj, shown in Equation 7
Figure 8. Twenty-seven of those experiments were of joints and Figure 8, that is a linear function of the joint length,  L.

Fig. 6.  Comparison of design shear strength in the 2005 (Rj = 0.80) and 2010 (Rj = 0.90) AISC Specifications.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 39

 End loaded End loaded Non-end loaded

As shown

Fig. 7.  Examples of joints that are end-loaded and joints that are non-end-loaded (AISC, 2016).

1.20

1.00

0.80 Eq. 7
Experimental RjRlj

Eq. 8
0.60

0.40

0.20

0.00
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0
Joint Length, in.

Fig. 8.  Experimentally measured joint efficiency, RjRlj (Tide, 2010).

40 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

When evaluated against the 72 data points, the ratio of the For a smaller-diameter Group 120 bolt, this would be (0.625)
experimental data to the value predicted by the equation is (120 ksi) = 75.0 ksi. The average measured shear strength per
on average 1.05 with a standard deviation 0.146. The equa- unit area from Figure 3 is 69.5 ksi, which is approximately
tion proved to match the experimental data very well for 93% of the strength assumed during design. Of the 410 bolts
compact joints but demonstrated more deviation for longer in the category that were tested, 33 had a measured strength
joints, particularly for joints in excess of approximately greater than or equal to 75 ksi. It is also noted, however, that
40  in. long, where the ratio was as low as 0.80 for some there were two bolts with a measured strength below 60 ksi,
joints and as high as 1.40 for others. which is the nominal design strength of a single Group 120
bolt with the threads not excluded.
RlRlj = 1.075 − 0.0113L(7)
As a second example, based on the rationale for com-
where 0.75 ≤ RlRlj < 0.90 pact joints described earlier, the strength of bolts in com-
Recognizing the iterative nature and challenges associ- pact joints (joints with L  ≤ 16  in.), where the threads are
ated with designing a joint where the strength is a function not excluded from the shear plane (N), taking Rv  = 0.625,
of both the number of bolts and the length of the joint, Tide Rnx = 0.80, Rj = 1.00, and Rlj = 1.00, can be expected to be
ultimately recommended that RjRlj be taken as 0.90 for joints approximately
not longer than 38 in. and that RjRlj be taken as 0.75 for joints
rnv = AbFu,boltRvRnxRjRlj(2)
longer than 38 in.
A second equation proposed herein can be used to repre- = AbFu,bolt(0.625)(0.80)(1.00)(1.00)
sent the lower bound of the data in Tide (2010). This equa-
= 0.500AbFu,bolt(11)
tion, shown in Equation  8 and Figure  8, provides a lower
bound of the measured data within 1%. For joints designed using the AISC Specification prior to
2010, this is the same as the strength of bolts designed as
RlRlj = 1.02 − 0.0120L(8)
X. For joints designed using the 2010 or 2016 AISC Speci-
where 0.60 ≤ RlRlj < 0.90 fication, this is approximately 89% of the nominal strength
of bolts designed as X. In the former case, there would be
no deficiency in the design of a joint designed as X but
DISCUSSION
installed as N. In the latter case, the 11% difference in nomi-
For a bolt in a joint designed assuming that the threads would nal strength would be offset by the resistance factor to be
be excluded but installed with the threads not excluded, the applied to the joint.
ratio of the as-installed strength to the as-designed strength Finally, considering that the resistance factor assigned
would be to the bolt fracture limit state is 0.75, the ratio of nominal
strength in the N condition to design strength in the X condi-
Strength N Rv (0.80) R j Rlj Ab Fu,bolt
= = 0.80 (9) tion is 1.07, suggesting that the resistance factor may offset a
Strength X Rv (1.00) R j Rlj Ab Fu,bolt  strength deficiency due to inaccurate assumptions regarding
thread condition.
This represents a difference of 20%, and while at first glance
it might seem that bolted joints designed as X but installed Nominal strength N Rv (0.80) Rj Rlj Ab Fu,bolt
= (12)
as N must always be understrength, this is not necessarily Design strength X (0.75) Rv (1.00) Rj Rlj Ab Fu,bolt
the case. While the location of the shear plane is one of the
= 1.07 
more critical parameters affecting the strength of bolts in
shear joints, other factors may partially or completely off-
While these anecdotal examples show that a bolt designed
set the effect of shear plane location. These other factors
as X but installed as N may have sufficient strength to be
include inherent overstrength of bolt materials, conserva-
considered adequate without the need for remedial action,
tive approaches used to account for the distribution of forces
basing decisions on anecdotal evidence alone may not be
among the bolts in a joint, rounding up the number of bolts
appropriate in all cases. A more rational approach to quan-
required in a joint, and designing for connection forces that
tifying the safety of joints with bolts designed as X but
are conservative.
installed as N is to conduct a structural reliability analysis.
As an example, the ultimate shear strength per unit area
of a bolt sheared through its shank, not considering joint A Structural Reliability Analysis
length effects, would be
It can be shown that there is a high likelihood that joints
Fnv = (0.625)(1.00)Fu,bolt(10) with bolts designed as X but installed as N will actually have
= 0.625Fu,bolt a strength lower than expected during design. However, that
alone does not mean that the joint will fail. In order for a

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 41

failure to occur, the available strength of the joint has to be Thus, the resistance, referred to as Rm when taken as the
lower than the demand on the joint. mean of several instances, and the coefficient of variation
Equation 13 has been a generally accepted fundamental can be computed as
basis of the LRFD approach for steel structures for many
Rm = R BRp (16)
years (Fisher et al., 1978):
= (69.5 ksi)(0.970) 
⎛R ⎞
ln ⎜ m ⎟
⎝ Qm ⎠ = 67.4 ksi
β= (13)
VR2 + VQ2 
VR = VB2 + VP2 (17)
where 2 2
= (0.061) + (0.046)
Qm = mean value of load, kips
Rm = mean value of resistance, kips = 0.076 
VR = coefficient of variation of resistance
For this analysis, it will be assumed that the mean dead
VQ = coefficient of variation of load
load effect is 1.05 times the nominal value with a coefficient
β = reliability index of variation of 0.10 (Dm = 1.05Dnom with VD = 0.10) and that
In the following example, results of 410 tests of smaller- the mean live load effect is equal to the nominal value with
diameter (s in. to 1 in.) Group 120 bolts sheared in the N a coefficient of variation of 0.25 (Lm = 1.00Lnom with VL =
condition conducted at the University of Cincinnati are used 0.25). Using a ratio of live load to dead load of L/ D = 3.0,
to illustrate the procedure followed to calculate the reliabil- the mean load effect can be computed as
ity indices as part of this study. Based on the measured shear Qm = Dm + Lm(18)
strengths, the mean value, standard deviation, and coeffi-
cient of variation can be calculated as = 1.05Dnom + (1.00)(3.0Dnom)

⎛ V ⎞ = 4.05Dnom
RB = ⎜ ult ⎟ (14)
⎝ Ab ,nom ⎠ mean Similarly, the coefficient of variation for the load effect
= 69.5 ksi  with a ratio of L/ D = 3.0 can be computed as
σB = 4.63 ksi [ (1.05Dnom ) (0.10 )]2 + [ (3.00 Dnom ) (0.25 )]2
VQ = (19)
VB = 0.061 4.05 Dnom
= 0.187 
For example, during one experiment, a d‑in.-diameter
A325 bolt was tested in single shear with the threads not The basis for design is ϕRn ≥ ΣγQ, where the design resis-
excluded and failed at a load of 41.8 kips. For this bolt tance according to the 2010 and 2016 AISC Specifications,
π 7 2 assuming that the threads would be excluded from the shear
Ab,nom = ⎛ ⎞ ⎛ in.⎞ plane, is
⎝ 4⎠ ⎝ 8 ⎠
= 0.601 in.2 ϕRn = ϕRvRnx(RjRlj)Ab,nomFu,nom (20)
41.8 kips = (0.75)(0.625)(1.00)(0.90)Ab,nomFu,nom
Fuv =
0.601 in.2
For the load combination ΣγQ = 1.2D + 1.6L using a ratio
= 69.6 ksi of L/ D = 3.0
where Fuv is the measured shear strength per unit area, based ΣγQ = (1.2)(Dnom) + (1.6)(3.0Dnom)(21)
on the nominal area of the unthreaded body of the bolt.
Considering compact joints where L  ≤ 16  in., the joint = 6.0Dnom
efficiency, its standard deviation, and its coefficient of varia- Thus
tion are
(0.75)(0.625)(1.00)(0.90)Ab,nomFu,nom ≥ 6.0Dnom(22)
⎛ Vult ⎞
Rp = ⎜ (15)
⎝ Vexp ⎟⎠ mean Dnom ≤ 0.070Ab,nomFu,nom

= 0.970  Substituting this value into Equation 18,

σp = 0.045 Qm = 4.05Dnom(18)
Vp = 0.046
42 / ENGINEERING JOURNAL / FIRST QUARTER / 2020
 Table 4.  Reliability Indices for Bolted Shear Joints Designed as X
but Installed as N Based on Reliability Analyses for L/D = 3.0
AISC Specification Prior to 2010 AISC Specification 2010 and 2016
Grade 120 120 150 120 120 150
Diameter, in. s–1 18–14 w–18 s–1 18–14 w–18
Compact 3.94 3.59 3.24 3.36 2.99 2.65
Intermediate 2.61 2.29 2.01 2.28 1.94 1.64
Long 3.07 2.79 2.54 2.48 2.19 1.95

Table 5.  Approximate Probabilities of Failure for Bolted Shear Joints

Designed as X but Installed as N Based on Reliability Analyses for L/D = 3.0
AISC Specification Prior to 2010 AISC Specification 2010 and 2016
Grade 120 120 150 120 120 150
Diameter, in. s–1 18–14 w–18 s–1 18–14 w–18
Compact < 0.01% 0.02% 0.06% 0.04% 0.14% 0.41%
Intermediate 0.45% 1.10% 2.20% 1.12% 2.64% 5.01%
Long 0.11% 0.26% 0.55% 0.65% 1.41% 2.54%

= 4.05(0.070Ab,nomFu,nom) indices were computed for the ratio of live load to dead load
of 3.0 and are shown in Table 4. Approximate probabilities
= 0.284Ab,nomFu,nom(23) of failure that correspond to these reliability indices are
Finally, the reliability index can be computed as shown in Table  5. Reliability indices and probabilities of
failure for other ratios of live load to dead load are shown in
Rm
Reliability index = (24) Appendix A as Tables A1 and A2, respectively.
Qm
(67.4 ksi)(Ab,nom ) Overstrength Due to Discretization
=
0.284 Ab,nom Fu,nom
Most bolted joints are inherently overdesigned because of
67.4 ksi the practical necessity to select a discrete number of bolts
=
(0.284)(120 ksi) for a given joint. Suppose, for example, that a joint is sub-
67.4 ksi jected to a factored design load of Pu = 95 kips and d‑in.-
=
34.1 ksi  diameter Grade 120 bolts are being used. According to the
2010 or 2016 AISC Specification, the bolts would have a
⎛R ⎞ design strength of ϕrnv  = 30.7  kip/bolt in the X condition
ln ⎜ m ⎟ and ϕrnv  = 24.3  kip/bolt in the N condition. If the joint is
⎝ Qm ⎠
β= (13) designed as X, then 95  kips/ 30.7  kip/bolt = 3.07 bolts are
VR2 + VQ2  required, and the engineer would likely use four bolts.3
67.4 ksi ⎞ Thus, the bolts in this joint would have a design strength of
ln ⎛ (4 bolts)(30.7 kip/bolt) = 123 kips, which is 29% larger than
⎝ 34.1 ksi ⎠
= the required strength. Suppose that after the joint with four
(0.077)2 + (0.187)2 bolts is erected, it is discovered that the bolts were actually
= 3.36 installed as N. The design strength in that case would be
Considering the three classes of bolts (smaller Group 120,
larger Group 120, and Group 150), the categories of joints
based on joint length (compact, intermediate, and long), and 3 It is assumed in the analyses presented in this work that the theoretical
the two slightly different approaches employed in the design number of required bolts is always rounded up to the next largest whole
specifications (AISC Specifications prior to 2010 versus the number. It is recognized, however, that some engineers might be inclined
to round the number of required bolts down in some cases where it would
AISC 2010 and 2016 Specifications), a total of 18 reliability be only slightly unconservative to do so.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 43

(4 bolts)(24.3 kip/bolt) = 97.2 kips, which is still larger than for the simulation. The simulated live load, Lsim, was mod-
the required strength of 95 kips. This situation is referred to eled using a Gumbel distribution (Extreme Type I) with a
herein as discretization overstrength. mean of 1.00 times the nominal live load and a coefficient
Discretization overstrength is challenging to quantify of variation of 0.25 (Nowak and Collins, 2013). The required
deterministically, but there are a few factors that influence strength was calculated using the load combination ΣγQ =
when discretization overstrength is present and how large it 1.2Dn + 1.6Ln, and the total simulated load was calculated as
may be. The first factor is the number of rows of bolts that Qsim = Dsim + Lsim.
are in a joint parallel to the applied force. The incremental Bolt diameter was modeled as a uniform random variable
change in design strength is typically an integer multiple of over the range of s in. to 1 in. for smaller Group 120 bolts,
the number of parallel rows in the joint. For example, if there 18 in. to 14 in. for larger Group 120 bolts, or w in. to 18 in.
is a single parallel row of bolts, then bolts are added one at a for Group 150 bolts. The number of bolts was determined
time until the design strength exceeds the required strength by dividing the required strength by the design strength per
of the joint. On the other hand, if there are two parallel rows bolt in the X condition as ΣγQ/ ϕrnv and then rounding up to
of bolts, then two bolts are generally added at a time. A sec- the next highest integer, but using no fewer than two bolts.
ond factor is the size and strength of an individual bolt. If Spacing of the bolts in the direction of load, s, was taken as
small bolts are used in a joint, then incrementally increasing the larger of 3 in. or three times the bolt diameter, and the
the number of bolts in the joint increases the strength by a length of the joint was calculated as L  = (nbolts − 1)s. The
smaller amount than if larger bolts had been used. Addition- design strength and number of bolts was modified for joints
ally, if bolts are deployed in double shear, then the incre- whose length exceeded the transition values of 50 in. in the
mental strength increase would be double relative to bolts AISC Specifications prior to 2010 and 38 in. in the 2010 and
deployed in single shear. While this is interesting in a gen- 2016 AISC Specifications.
eral sense, it is not likely to have an influence on the fully The simulated ultimate shear strength per unit area of the
threaded short bolt problem because those bolts are likely bolts, Fuv,sim , was modeled as a lognormal random variable
not long enough to be deployed in double shear. Finally, the using the means and standard deviations shown in Figures 3,
size of the joint is also a consideration. When joints made 4, and 5. Figure  9 shows a histogram of simulated shear
of a larger number of bolts are designed, the incremental strength per unit area for smaller Group 120 bolts used in the
strength increase associated with adding one row of bolts is analysis with AISC Specifications prior to 2010. The simu-
smaller relative to the overall strength of the joint made up lated bolt strength, Rsim, was calculated as nboltsRjRljAbFuv,sim,
of a smaller number of bolts. This behavior, and the fact that where RjRlj was modeled as a uniformly distributed random
no fewer than two bolts can be used in any bolted joint, leads variable with a lower bound determined using Equation  8
to a higher discretization overstrength for smaller joints than and an upper bound of 1.00. Finally, the capacity-to-demand
for larger joints. ratio was computed as the ratio of Rsim to Qsim , and the prob-
ability of failure was taken as the number of simulations
Monte Carlo Simulations yielding a capacity-to-demand ratio less than unity divided
by the number of simulations. The simulated values of joint
Given the difficulty in deterministically quantifying the efficiency, RjRlj, for the analysis of smaller Group 120 bolts
discretization overstrength, an indirect approach was used designed using AISC Specifications prior to 2010 are shown
to assess the safety of bolted joints designed as “threads in Figure 10.
excluded” but erected as “threads not excluded.” A series of Figure 11 shows an overview of the results of the analysis
six analyses was conducted for end-loaded joints correspond- of smaller Group  120 bolts designed using AISC Specifi-
ing to each of the three classes of bolts and two versions of cations prior to 2010. Overall, the probability of failure for
the AISC Specification. Each of the analyses consisted of this analysis was 0.16%. The results were further subdivided
100,000  Monte Carlo simulations. Sample calculations for based on the length of the joints in each of the 100,000 simu-
one simulation are shown in Appendix B. lations and are shown in Table 6 along with the results from
The nominal dead load in each simulation, Dn, was the other analyses. Results for ratios of live load to dead load
modeled as a uniformly distributed random variable rang- other than 3.0 are shown in Appendix A as Table A3.
ing between approximately 0.50 times and 24 times the
design strength of a single bolt in the joint being consid-
ered. The simulated dead load, Dsim, was modeled as a SUMMARY
normally distributed random variable with a mean of 1.05 A question that will undoubtedly arise regarding structures
times the nominal dead load and a coefficient of variation that were designed with short bolts assuming that threads
of 0.10 (Nowak and Collins, 2013). The nominal live load, were excluded from the shear plane is whether remedial
Ln, was determined by multiplying the nominal dead load measures need to be taken to ensure the safety of the struc-
by the ratio of live load to dead load that was considered ture. The results of the reliability analysis and Monte Carlo

44 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 Table 6.  Approximate Probabilities of Failure for Bolts Designed as X
but Installed as N Based on Monte Carlo Analyses for L/D = 3.0
AISC Specification Prior to 2010 AISC Specification 2010 and 2016
Grade 120 120 150 120 120 150
Diameter, in. s–1 18–14 w–18 s–1 18–14 w–18
Compact < 0.01% <0.01% < 0.01% 0.01% 0.05% 0.18%
Intermediate 0.34% 0.70% 1.45% 0.72% 1.72% 2.91%
Long 0.02% 0.07% 0.26% 0.29% 0.73% 1.48%

25%

Group 120 Bolts n = 100,000 Simulations

Threads N Ave = 69.5 ksi
20% d = s" to 1" St Dev = 4.27 ksi
Max = 90.3 ksi
Min = 53.1 ksi
Frequency (%)

15%

10%

5%

0%
40.0

42.5
45.0

47.5

50.0

52.5

55.0
57.5
60.0

62.5

65.0
67.5

70.0

72.5
75.0

77.5

80.0
82.5
85.0
87.5

90.0

92.5
Simulated Shear Strength per Unit Area, Fuv,sim
Fig. 9.  Simulated shear strength per unit area of smaller Group 120 bolts with threads not excluded.

1.20

1.00
Simulated Joint Efficiency, RjRlj

0.80

0.60

0.40

0.20

0.00
0.00 10.00 20.00 30.00 40.00 50.00 60.00 70.00 80.00 90.00 100.00
Joint Length, L (in.)
Fig. 10.  Simulated joint efficiency used in the analysis of smaller Group 120 bolts per AISC Specifications prior to 2010.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 45

 Table 7.  Relationship between Probability of Failure and
Reliability Index for a Normally Distributed Dataset
Reliability Index Probability of Failure
1.0 15.9%
2.0 2.28%
2.5 0.621%
2.6 0.466%
3.0 0.135%
3.5 0.0233%
4.0 0.00317%
4.5 0.000340%

analyses described herein should help an engineer make that to 2010 have a higher reliability than those designed using
decision. The commentary to Chapter B of the 2016 AISC the 2010 and 2016 AISC Specifications.
Specification (AISC, 2016) states that target reliability
• Smaller Group 120 bolts demonstrated the highest level of
indices at L/ D  = 3 of approximately 2.6 for members and
reliability, and Group 150 bolts demonstrated the lowest
approximately 4.0 for connections were used in the develop-
reliability of the three classes of bolts considered in this
ment of the specification. The probabilities of failure associ-
study.
ated with these reliability indices are approximately 0.466%
and 0.00317%, respectively. Approximate probabilities of • Compact joints proved to be the most reliable, followed
failure for these and additional values of the reliability index by long joints, and then intermediate-length joints.
are presented in Table 7.
• Based on the Monte Carlo analyses, it can be concluded
It can be observed from the analyses described herein that
that compact joints designed using AISC Specifications
bolts designed as X but installed as N still have a substantial
prior to 2010, regardless of bolt grade and diameter, have
level of reliability, in some cases as high as target reliabili-
a reliability that approximately meets target reliabilities
ties cited in the commentary to the 2016 AISC Specifica-
for connections in the 2016 AISC Specification.
tion. The following conclusions can be made:
It is evident that bolts designed as X but installed as N
• Bolts in joints designed using AISC Specifications prior
may not meet the target reliability index for connections in

20%

18% AISC Specification prior to 2010

Designed as X, installed as N
16% n = 100,000 simulations
14% Rows of bolts = 2 to 30
Group 120 bolts d = s" to 1"
Frequency (%)

12% P (failure) = 0.16%

10%

8%

6%

4%

2%

0%
0.75

1.00

1.25

1.50

1.75

2.00

2.25
2.50

2.75
3.00

3.25

3.50

3.75

4.00

4.25

4.50

4.75
5.00

5.25

5.50

5.75

6.00

Ratio of Simulated Shear Strength to Simulated Load

Fig. 11.  Results of analysis of smaller Group 120 bolts per AISC Specifications prior to 2010.

46 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

many cases. However, a criterion that an engineer may con- the bolt diameter and the thickness and material strength of
sider imposing when deciding whether remedial measures the connected plies. This is explored in Appendix C, where
are necessary is whether the connection has a reliability that critical ply thicknesses are tabulated; joints with plies thin-
is at least equal to that of the members that are joined by it. ner than those tabulated will have bearing strengths that are
If that criterion is employed, then a maximum probability lower than the shear strength of the associated bolt in the N
of failure of approximately 0.466% would be acceptable. In condition.
that case, it could be stated that all compact joints designed The analyses presented herein focused on end-loaded
using any AISC Specification would satisfy that criterion. joints to the exclusion of eccentrically loaded joints. How-
Further, it could be stated that all joints, regardless of length, ever, because the strength of eccentrically loaded joints
designed in accordance with AISC Specifications prior to tends to depend mostly on the strength of the single most
2010 would also satisfy that criterion as long as Group 120 heavily loaded fastener in the bolt group, it could be postu-
bolts not larger than 1 in. in diameter were used. lated that the reliability of eccentrically loaded joints would
be similar to the reliability of compact end-loaded joints.
Additional Considerations This is supported by additional Monte Carlo simulations
that were conducted but are not described herein.
The analyses described herein were conducted under the
Finally, if after considering the analyses and discussion
assumption that the joints were designed for the loads to
presented herein, an engineer decides that remedial action
which they were actually subjected. If joints in this case
may be needed for a structure as a last resort, consider-
were actually designed, as is often done in practice, for
ation should be given to the idea of removing a representa-
artificially large loads, such as those corresponding to the
tive sample of the bolts in question to see if sufficient body
capacity of the member(s) being joined or a fraction of the
length may actually exist before prescribing a more exten-
load that would cause failure of the connected member(s)
sive and costly remediation solution.
[e.g., 2 uniformly distributed load (UDL)], then the reliabil-
It is noted that differences exist between the AISC and
ity of the joints would be increased, substantially in many
AASHTO-LRFD specifications—particularly differences
cases.
in the resistance factors, loads, and load combinations—
Shear joints designed as slip critical will likely be unaf-
that make it difficult to directly apply the reliabilities and
fected by the issue described herein. While it is noted that
probabilities of failure presented herein to bridge structures.
slip-critical joints must also be checked to ensure that the
Still, it is expected that the analyses presented herein will
strength of the bolts in bearing is satisfactory, a larger num-
be useful to bridge engineers nonetheless. It is noted that
ber of bolts is generally required in slip-critical joints than
the change in design strength between the 2005 and 2010
in bearing joints, and those additional bolts will likely pro-
AISC Specifications was approximately mirrored between
vide enough added strength in shear to provide a satisfactory
the 7th and 8th editions of the AASHTO-LRFD Specifica-
level of reliability.
tions. In the 7th and prior editions of the AASHTO-LRFD
Additionally, bolts deployed with more than one shear
Specification, the design strengths were rnv = 0.48AbFu,bolt Ns
plane (e.g., double shear, triple shear, etc.) are less likely to
and rnv  = 0.38Ab Fu,bolt Ns for threads excluded and not
be affected by the short bolt issue. Because bolts deployed
excluded, respectively, whereas in the 8th edition, the design
with more than one shear plane tend to be longer, they are
strengths were increased to rnv = 0.56AbFu,bolt Ns and rnv =
less likely to fall into the category of short fully threaded
0.45AbFu,bolt Ns, where Ns is the number of shear planes.
bolts. However, care should be taken with bolts of any length
The coefficients used in the 7th and prior editions of the
that are deployed in applications with more than one shear
AASHTO-LRFD Specifications are slightly lower than
plane to ensure that the threads of the bolts are excluded
those in the AISC Specifications prior to 2010; thus, the reli-
from all shear planes, if that is the assumption made during
ability of joints designed per those specifications would be
design.
favorably affected. It is further noted that the commentary to
One aspect of joint design that is absent from the analyses
the 8th edition of the AASHTO-LRFD Specification states
presented herein is the possibility that failure modes other
that it has been calibrated with a target reliability index of
than bolt shear may govern the strength of the joint. For
3.5 for main members under the Strength I load combina-
example, the strength of a joint with bolts designed as X but
tion over a 75-year design life. In a separate section, the
later discovered to actually be N may have been controlled
AASHTO commentary states that a reliability index of 4.5
by, for example, the bearing strength of the material around
was targeted for truss gusset plates under the Strength I load
the bolt instead of by the strength of the bolts. Revising the
combination at a dead load to live load ratio of 6.0, which
shear strength of the bolts in that case may have no effect
is outside of the range considered in this study (AASHTO,
on the strength of the joint or may result in a reduction in
2017).
strength less than the 20% implied by Equation 9. An analy-
sis of bearing strength is possible because it depends only on

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 47

 REFERENCES Fisher, J.W., Galambos, T.V., Kulak, G.L., and Ravindra,
M.K. (1978), “Load and Resistance Factor Design Crite-
AASHTO (2017), AASHTO LRFD Bridge Design Specifi-
ria for Connectors,” Journal of the Structural Division,
cations, 8th Ed., American Association of State Highway
ASCE, September.
and Transportation Officials, Washington D.C.
Fisher, J.W. and Struik, J.H.A. (1974), Guide to Design Cri-
AISC (1986), Specification for Structural Steel Buildings—
teria for Bolted and Riveted Joints, 1st Ed., John Wiley
Load and Resistance Factor Design, American Institute
and Sons, New York, N.Y.
of Steel Construction, Chicago, Ill.
Frank, K.H. and Yura, J.A. (1981), An Experimental Study of
AISC (1993), Load and Resistance Factor Design Specifi-
Bolted Shear Connections, FHWA/RD-81/148, Federal
cation for Structural Steel Buildings, 2nd Ed., American
Highway Administration, Washington, D.C., December.
Institute of Steel Construction, Chicago, Ill.
Kulak, G.L., Fisher, J.W., and Struik, J.H.A. (1987), Guide
AISC (2005), Specification for Structural Steel Buildings,
Design Criteria for Bolted and Riveted Joints, 2nd Ed.,
ANSI/AISC 360-05, American Institute of Steel Con-
John Wiley and Sons, New York, N.Y.
struction, Chicago, Ill.
Moore, A.M., Rassati, G.A., and Swanson, J.A. (2008),
AISC (2010), Specification for Structural Steel Buildings,
Evaluation of the Current Resistance Factors for High-
ANSI/AISC 360-10, American Institute of Steel Con-
Strength Bolts, Final Report Submitted to the Research
struction, Chicago, Ill.
Council on Structural Connections, January.
AISC (2016), Specification for Structural Steel Buildings,
Moore, A.M., Rassati, G.A., and Swanson, J.A. (2010), “An
ANSI/AISC 360-16, American Institute of Steel Con-
Experimental Analysis of Strength and Ductility of High-
struction, Chicago, Ill.
Strength Fasteners,” Engineering Journal, AISC, Vol. 47,
AISC (2017), Steel Construction Manual, American Insti- No. 3.
tute of Steel Construction, Chicago, Ill.
Nowak, A.S. and Collins, K.R. (2013), Reliability of Struc-
ASME (1981), Square and Hex Bolts and Screws (In. Series), tures, 2nd Ed., CRC Press, Boca Raton, Fla.
ASTM B18.2.1, American Society of Mechanical Engi-
Roenker, A.T., Rassati, G.A., and Swanson, J.A. (2017),
neers, New York, N.Y.
Testing of Torque-and-Angle High Strength Fasteners, A
ASME (2019), Fasteners for Use in Structural Applications, Technical Report Submitted to LeJeune Bolt Company,
ASTM B18.2.6, American Society of Mechanical Engi- March.
neers, New York, N.Y., Adopted and In Press.
RCSC (1964), Specifications for Structural Joints Using
ASTM (2014a), Standard Specification for Structural ASTM A325 or A490 Bolts, Research Council on Riv-
Bolts, Steel, Heat Treated, 120/105 ksi Minimum Tensile eted and Bolted Structural Joints of the Engineering
Strength, ASTM A325, ASTM International, West Con- Foundation.
shohocken, Pa. (withdrawn 2016).
RCSC (1976), Specification for Structural Joints Using
ASTM (2014b), Standard Specification for Structural Bolts, ASTM A325 or A490 Bolts, Research Council on Riv-
Alloy Steel, Heat Treated, 150  ksi Minimum Tensile eted and Bolted Structural Joints of the Engineering
Strength, ASTM A490, ASTM International, West Con- Foundation.
shohocken, Pa. (withdrawn 2016).
RCSC (1985), Allowable Stress Design Specification for
ASTM (2014c), Standard Specification for “Twist Off” Type Structural Joints Using ASTM A325 or A490 Bolts,
Tension Control Structural Bolt/Nut/Washer Assem- Research Council on Structural Connections of the Engi-
blies, Steel, Heat Treated, 120/105 ksi Minimum Tensile neering Foundation.
Strength, ASTM F1852, ASTM International, West Con-
RCSC (1988), Load and Resistance Factor Design Specifi-
shohocken, Pa. (withdrawn 2016).
cation for Structural Joints Using ASTM A325 or A490
ASTM (2014d), Standard Specification for “Twist Off” Type Bolts, Research Council on Structural Connections of the
Tension Control Structural Bolt/Nut/Washer Assemblies, Engineering Foundation.
Steel, Heat Treated, 150  ksi Minimum Tensile Strength,
RCSC (2015), 2014 Specification for Structural Joints Using
ASTM F2280, ASTM International, West Conshohocken,
High-Strength Bolts with Errata, Research Council for
Pa. (withdrawn 2016).
Structural Connections, Chicago, Ill.
ASTM (2018), Standard Specification for High Strength
Swanson, J.A., Rassati, G.A., and Larson, C.M. (2019),
Structural Bolts, Steel and Alloy Steel, Heat Treated, 120
“Dimensional Tolerances and Length Determination
ksi (830 MPa) and 150 ksi (1040 MPa) Minimum Ten-
of High-Strength Bolts,” Engineering Journal, AISC,
sile Strength, In. and Metric Dimensions, ASTM F3125,
Vol. 57, No. 1.
ASTM International, West Conshohocken, Pa.

48 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

Taylor, A.T., Rassati, G.A., and Swanson, J.A. (2008), Eval- Tide, R.H.R. (2010), “Bolt Shear Design Considerations,”
uation of the Resistance Factor for Fully Threaded High Engineering Journal, AISC, Vol. 47, No. 1.
Strength Fasteners, Final Report Submitted to the Metal
Building Manufacturers Association, September.

APPENDIX A

ADDITIONAL CALCULATED VALUES

Table A1.  Reliability Indices for Bolted Shear Joints Designed as X

but Installed as N Based on Reliability Analyses for Various Ratios of L/D
Ratio of Live Load to Dead Load
1 2 3 4 5
Smaller Gr 120 4.68 4.18 3.94 3.81 3.72
Compact Larger Gr 120 4.27 3.80 3.59 3.47 3.39
AISC Specification

Gr 150 3.78 3.42 3.24 3.14 3.07

Prior to 2010

Smaller Gr 120 2.71 2.66 2.61 2.58 2.55

Intermediate Larger Gr 120 2.33 2.32 2.29 2.27 2.25
Gr 150 2.00 2.03 2.01 2.00 1.99
Smaller Gr 120 3.16 3.12 3.07 3.04 3.01
Long Larger Gr 120 2.84 2.82 2.79 2.76 2.74
Gr 150 2.56 2.57 2.54 2.52 2.51
Smaller Gr 120 3.91 3.54 3.36 3.26 3.19
Compact Larger Gr 120 3.47 3.15 2.99 2.90 2.85
AISC Specification

Gr 150 2.99 2.76 2.65 2.58 2.53

2010 or 2016

Smaller Gr 120 2.35 2.32 2.28 2.25 2.23

Intermediate Larger Gr 120 1.94 1.95 1.94 1.92 1.91
Gr 150 1.58 1.64 1.64 1.64 1.64
Smaller Gr 120 2.49 2.50 2.48 2.46 2.45
Long Larger Gr 120 2.16 2.20 2.19 2.19 2.18
Gr 150 1.89 1.95 1.95 1.95 1.95

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 49

 Table A2. Probabilities of Failure for Bolted Shear Joints Designed as X
but Installed as N Based on Reliability Analyses for Various Ratios of L/D
Ratio of Live Load to Dead Load
1 2 3 4 5
Smaller Gr 120 < 0.01% < 0.01% < 0.01% < 0.01% < 0.01%
Compact Larger Gr 120 < 0.01% < 0.01% 0.02% 0.03% 0.04%
AISC Specification

Gr 150 < 0.01% 0.03% 0.06% 0.09% 0.11%

Prior to 2010

Smaller Gr 120 0.34% 0.39% 0.45% 0.50% 0.53%

Intermediate Larger Gr 120 0.98% 1.01% 1.10% 1.17% 1.22%
Gr 150 2.27% 2.13% 2.20% 2.27% 2.32%
Smaller Gr 120 0.08% 0.09% 0.11% 0.12% 0.13%
Long Larger Gr 120 0.22% 0.24% 0.26% 0.29% 0.31%
Gr 150 0.52% 0.51% 0.55% 0.58% 0.61%
Smaller Gr 120 < 0.01% 0.02% 0.04% 0.06% 0.07%
Compact Larger Gr 120 0.03% 0.08% 0.14% 0.19% 0.22%
AISC Specification

Gr 150 0.14% 0.29% 0.41% 0.50% 0.57%

2010 or 2016

Smaller Gr 120 0.93% 1.01% 1.12% 1.21% 1.28%

Intermediate Larger Gr 120 2.62% 2.53% 2.64% 2.74% 2.82%
Gr 150 5.72% 5.06% 5.01% 5.05% 5.09%
Smaller Gr 120 0.65% 0.62% 0.65% 0.69% 0.71%
Long Larger Gr 120 1.53% 1.39% 1.41% 1.44% 1.47%
Gr 150 2.95% 2.58% 2.54% 2.55% 2.57%

Table A3. Approximate Probabilities of Failure for Bolted Shear Joints Designed as X
but Installed as N Based on Monte Carlo Analyses for Various Ratios of L/D
Ratio of Live Load to Dead Load
1 2 3 4 5
Smaller Gr 120 < 0.01% < 0.01% < 0.01% < 0.01% < 0.01%
Compact Larger Gr 120 < 0.01% < 0.01% < 0.01% < 0.01% < 0.01%
AISC Specification

Gr 150 < 0.01% < 0.01% < 0.01% 0.04% 0.04%

Prior to 2010

Smaller Gr 120 0.21% 0.28% 0.34% 0.28% 0.31%

Intermediate Larger Gr 120 0.59% 0.70% 0.70% 0.76% 0.77%
Gr 150 1.43% 1.40% 1.45% 1.40% 1.51%
Smaller Gr 120 < 0.01% 0.02% 0.02% 0.03% 0.03%
Long Larger Gr 120 0.04% 0.07% 0.07% 0.11% 0.16%
Gr 150 0.14% 0.21% 0.26% 0.29% 0.30%
Smaller Gr 120 < 0.01% < 0.01% 0.01% 0.02% 0.01%
Compact Larger Gr 120 0.02% 0.02% 0.05% 0.05% 0.04%
AISC Specification

Gr 150 0.06% 0.11% 0.18% 0.22% 0.25%

2010 or 2016

Smaller Gr 120 0.56% 0.67% 0.72% 0.66% 0.74%

Intermediate Larger Gr 120 1.68% 1.66% 1.72% 1.74% 1.64%
Gr 150 3.37% 2.99% 2.91% 3.03% 3.12%
Smaller Gr 120 0.15% 0.27% 0.29% 0.35% 0.39%
Long Larger Gr 120 0.55% 0.58% 0.73% 0.79% 0.77%
Gr 150 1.38% 1.43% 1.48% 1.45% 1.64%

50 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 APPENDIX B

SAMPLE CALCULATIONS FROM MONTE CARLO SIMULATIONS

Case Group 150 Bolts Conforming to AISC 2010 and 2016 Specifications with L/ D = 3.0
Randomly select 1-in.-diameter bolts: Ab = 0.785 in.2
The design strength of a single bolt with threads excluded is:
Fnv = RvRnxRjRljFu,bolt(6)
= (0.625)(1.00)(0.90)(1.00)(150 ksi)
= 84.4 ksi

ϕrnv = ϕAbFnv (from Eq. 5)

2
= (0.75)(0.785 in. )(84.4 ksi) 
= 49.7 kip/bolt

Determine the design loads:

Set the minimum nominal dead load:
(0.50)(49.7 kip/bolt)
Dn,min =
[ (1.2)(1.0) + (1.6)(3.0) ]
= 4.14 kips

Set the maximum nominal dead load:

(24)(49.7 kip/bolt)
Dn,max =
[ (1.2)(1.0) + (1.6)(3.0) ]
= 199 kips

A nominal dead load of Dn = 136 kips is computed as a uniform random variable between 4.14 kips and 199 kips.
An “actual” dead load of Dsim = 134 kips is simulated as a normally distributed random variable with a mean of (1.05)(136 kips) =
143 kips and a standard deviation of (0.10)(136 kip) = 13.6 kips.
The nominal live load is computed as Ln = (3.0)(136 kips) = 408 kips.
An “actual” live load of Lsim = 529 kips is simulated as a random variable with Gumbel distribution using a mean of (1.00)
(408 kips) = 408 kips and a standard deviation of (0.25)(408 kips) = 102 kips.
The total nominal load is
Qn = Dn + Ln 
= 136 kips + 408 kips 
= 544 kips

The total design load is

ΣγQ = 1.2Dn + 1.6Ln 
= (1.2)(136 kips) + (1.6)(408 kips)
= 816 kips

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 51

The total “actual” load is
Qsim = Dsim + Lsim 
= 134 kips + 529 kips 
= 663 kips

With a design load of ΣγQ = 816 kips and a design strength of ϕrnv = 49.7 kip/bolt, the trial number of required bolts is
816 kips
= 16.4 bolts →   try 17 bolts
49.7 kip/bolt  

The spacing of the bolts is 3 in., thus the length of the joint would be L = (17 − 1)(3 in.) = 48.0 in.
Because the length is greater than 38 in., the factor Rlj is taken as 0.833 instead of 1.00, thus,
Fnv = (0.625)(1.00)(0.90)(0.833)(150 ksi)
= 70.3 ksi

ϕrnv = (0.75)(0.7854 in.2)(70.3 ksi)
= 41.4 kip/bolt

The number of required bolts is recomputed as

816 kips
= 19.7 bolts →   use 20 bolts
41.4 kip/bolt  
and the length of the connection is recomputed as L = (20 − 1)(3 in.) = 57.0 in.
The nominal shear strength per unit area of the bolt with the threads not excluded is simulated as a log-normally distributed
random variable with a mean of 74.6 ksi and a standard deviation of 3.81 ksi and is determined to be Fuv,sim = 78.0 ksi for this
simulation.
The factor RjRlj is simulated as a uniformly distributed variable with an upper bound of 1.00 and a lower bound of
RjRlj,min = 1.02-0.0120L(8)
= 1.02 − (0.0120)(57.0 in.)
= 0.336

Because RjRlj,min < 0.60, use RjRlj,min = 0.60.

For this simulation, a value of RjRlj = 0.619 was used.
The “actual” strength of this joint is simulated as
Rsim = (20 bolts)(0.785 in.2/bolt)(78.0 ksi)(0.619)
= 758 kips

Finally, the capacity-to-demand ratio for the joint is then

Rsim 758 kips
=
Qsim 663 kips
= 1.14

52 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 APPENDIX C

Table C1.  Ply Thickness Below which Bearing Strength Will Govern over Bolt Shear Strength, N
Fu,ply = 58 ksi Fu,ply = 65 ksi
Fu,bolt = 120 ksi Fu,bolt = 150 ksi Fu,bolt = 120 ksi Fu,bolt = 150 ksi
Bolt Dia. t crit t crit t crit t crit
(in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.) (in.)
 ½ 0.152 8 0.190 x 0.136 8 0.170 x
 s 0.190 x 0.238 4 0.170 x 0.212 x
 w 0.228 4 0.286 c 0.204 x 0.255 4
 d 0.267 4 0.333 c 0.238 4 0.297 c
1 0.305 c 0.381 a 0.272 4 0.340 c
18 0.343 c 0.428 v 0.306 c 0.382 a
14 0.381 a 0.476 ½ 0.340 c 0.425 v
1a 0.419 v 0.524 ½ 0.374 a 0.467 v
1½ 0.457 v 0.571 b 0.408 v 0.510 ½

Table C2.  W-Shapes (A992 Steel) for which the Bearing Strength of the Web Will Govern over Bolt Shear Strength, N
Group 120 Bolts Group 150 Bolts
Shape d = w in. d = d in. d = 1 in. d = w in. d = d in. d = 1 in.
W18 — — — — — ≤ W18×40
W16 — — ≤ W16×26 W16×26 ≤ W16×36 ≤ W16×40
W14 — ≤ W14×22 ≤ W14×30 ≤ W14×26 ≤ W14×34 ≤ W14×48
W12 W12×14 ≤ W12×19 ≤ W12×30 ≤ W12×19 ≤ W12×35 ≤ W12×45
W10 W10×12 ≤ W10×15 ≤ W10×26 ≤ W10×22 ≤ W10×26 ≤ W10×39
W8 W8×10 ≤ W8×13 ≤ W8×24 ≤ W8×24 ≤ W8×31 ≤ W8×35
W6 ≤ W6×9 ≤ W6×15 ≤ W6×20 ≤ W6×15 ≤ W6×20 All

BEARING STRENGTH COMPARISON the critical ply thickness can be found as a function of the
bolt diameter. Joints with ply thicknesses less than this will
For bolts in single shear, the bearing strength of a connected
have a bearing strength less than the bolt shear strength,
ply will govern when the bearing strength is less than or
calculated assuming that the threads are not excluded from
equal to the bolt shear strength, N:
the shear plane. These values are shown in Table C1, where
ϕbearing2.4tply dbFu,ply + ϕshear 0.450Ab Fu,bolt(C1) both decimal values and the nearest zth fractional values
are tabulated. Additionally, note that the tabulated values
Thus, joints that have a ply thinner than shown in the follow- are based on the case where deformation at the bolt hole
ing equation will be governed by bearing strength instead of at service load is a design consideration and that the bolt
the bolt shear strength. shear strength is calculated using the 2010 and 2016 AISC
d b Fu,bolt Specifications.
t ply ≤ (C2) When bolts are used to connect the webs of wide-flange
6.791Fu, ply 
shapes, shapes can be identified that will satisfy the mini-
mum ply thickness shown in Table C1. Those shapes are
Substituting common values of 58  ksi and 65  ksi for tabulated in Table C2 for Group 120 and Group 150 bolts for
Fu,ply, and 120 ksi and 150 ksi for Fu,bolt into this equation, the common bolt diameters of w in., d in., and 1 in.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 53

54 / ENGINEERING JOURNAL / FIRST QUARTER / 2020
Reexamination of Shear Lag in HSS Tension Members
with Side Gusset Plate Connections
AKASHDEEP A. BHAT and PATRICK J. FORTNEY
In memory of Patrick J. Fortney, who passed away in October, 2019.

ABSTRACT
This paper presents an evaluation of the shear lag factor for HSS tension members connected with two side plate gussets with longitudinal
welds as given in AISC Specification Table D3.1, Case 6b. The current AISC Specification for Case 6b does not permit weld lengths less than
the perpendicular distance between the welds and has the potential of producing negative shear lag factors. Similar issues previously existed
for members given in Case 4 of Table D3.1. However, the AISC Specification has adopted a mathematical model proposed by Fortney and
Thornton for Case 4 of Table D3.1. The work presented in this paper (1) offers a mathematical model for calculating the shear lag factor for
Case 6b derived by repurposing the model adopted by AISC for Case 4 of Table D3.1, (2) offers the results of a parametric study comparing
the results of the new mathematical model to the results using the current AISC method, and (3) discusses the protocols developed for use in
finite element analysis to evaluate the effectiveness of the proposed mathematical model. The proposed new mathematical model will permit
longitudinal weld lengths less than the perpendicular distance between the welds and removes the possibility of calculating a negative shear
lag factor, while better representing the redistribution of cross-sectional stress near the connection region.

Keywords:  Shear lag, HSS, longitudinal welds, gusset plates.

BACKGROUND away from the connected region is a region thought to be

subjected to the shear lag phenomenon. Along this critical
T he performance of a tension member is influenced by
factors like size, material, type of connection, and the
like. The manner in which the tension member is connected
length of the tension member, this shear lag effect can have
a detrimental effect on the tensile capacity of the tension
member.
has a direct influence on shear lag. Tension members are
Generally, the reduced capacity of a tension member is
generally connected to elements such as gusset plates using
addressed through the use of a shear lag factor, UA, applied
bolts or welds. When some, but not all the elements of the
to the gross area of the tension member. AISC Specifica-
cross section of a tension member (e.g., flange, web, leg,
tion for Structural Steel Buildings (AISC, 2016), Table D3.1,
etc.) are used to transfer the axial load to the connection, a
requires an accounting for shear lag for various types of end
nonuniform stress distribution occurs in the tension member
connections. For HSS members connected with two side
adjacent to the connection, this phenomenon is commonly
gusset plates with longitudinal welds (of length l), and con-
referred to as shear lag (Fortney and Thornton, 2012). Due
sidering the perpendicular distance between the welds as H,
to this effect, the entire cross section is not fully effective
the shear lag factor is calculated using Case 6b of Table D3.1
to carry the load at the critical section, resulting in reduced
as shown in Table 1.
design strength of the member.
In an HSS tension member connected with two side gus-
set plates on opposite faces, the longitudinal welds receive INTRODUCTION
the load from the gusset plate through shear. The shear in
the welds is then transferred to the HSS section as an axial The current AISC Specification procedure for determining
load. The transition from this localized stress to a uniformly the shear lag factor for Case 6b assumes that the welds will
distributed stress over the tension member cross section transfer the entire member load to the two walls parallel
with the gusset plates and assumes the walls that are per-
pendicular to the gusset plates are unconnected elements.
An additional assumption made for the current Case 6b is
that the perpendicular distance between the welds is equal to
Patrick J. Fortney, Associate Professor, University of Cincinnati, Cincinnati, the width of the H side (which is parallel to the gusset plate).
Ohio. In memory of Patrick J. Fortney, who passed away in October, 2019. However, the authors propose that all four walls are con-
Akashdeep Bhat, Graduate Student, University of Cincinnati, Cincinnati, Ohio. nected by the welds, and as such, the perpendicular distance
Email:[email protected] (corresponding)
between the welds could be taken as either H or B, depend-
ing on which sides of the HSS the gusset plates are attached
Paper No. 2019-06 to. With this, one could argue that the eccentricity, x, could

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 55

 Table 1.  Shear Lag Factor Case 6b
Case Description of Element Shear Lag Factor, UA Example
x
l ≥ H...UA = 1−
With two side l
6 Rectangular HSS
gusset plates B2
x=
4 (B + H )

Table 2.  Connection Eccentricity x

Perpendicular Distance between Welds x

B2
H x=
4 (B + H )

H2
B x=
4 (B + H )

be calculated in two different ways (as shown in Table  2) study was performed comparing the shear lag factor cal-
for a rectangular HSS, giving two possible shear lag factors. culated using the current AISC Specification to that calcu-
Furthermore, the current AISC Specification Case  6b lated using the proposed new mathematical model. A series
does not permit weld lengths less than the perpendicular dis- of finite element models were then developed in ABAQUS
tance between the welds. When the length of the connection, (2013) to evaluate the validity of the mathematical model.
l, is less than the connection eccentricity, x, and the gusset For the new model proposed, the welds are considered to
plates are attached to the B side, the equation produces a connect all four of the HSS walls, and the weld length need
negative shear lag factor. For example, for an HSS 20×4×2 not necessarily be equal to or greater than the perpendicular
with the gusset plate attached to the B side (4-in. side) with distance between the welds. The following presents the new
welds of length l  = 4  in., the connection eccentricity and proposed model.
the shear lag factor, using the current AISC Specification
Case 6b, can be calculated as:
MATHEMATICAL MODEL
H2
x=
4 (B + H ) Fortney and Thornton (2012) proposed a new method
for evaluating the shear lag factor for Case  4 as given in
=
( 20 in.)2 Table  D3.1 of the 2010 AISC Specification (AISC, 2010)
4 ( 20 in. + 4 in.) addressing the longitudinally welded end-connected mem-
= 4.17 in. bers in a more general way. The objective of that work was
x to develop a generalized procedure for calculating shear
UA = 1
l lag in plates, angles, channels, and tee members connected
4.17 in. with longitudinal welds. Consideration was given to connec-
=1 tions with weld lengths less than the perpendicular distance
4 in.
between the welds and also to address the condition where
= 0.04
connections have unequal weld lengths. Fortney and Thorn-
This is problematic because any connection, even one ton compared three models for evaluating the shear lag fac-
with a relatively short connection length, will provide some tor: (1)  the AISC model (prior to 2016), (2)  the Canadian
level of force transfer. Standards Association (CSA) model, and (3)  a fixed-fixed
To overcome these shortcomings, a mathematical model beam model.
was developed by repurposing the mathematical model The fixed-fixed beam model proposed by Fortney and
developed by Fortney and Thornton (2012) in AISC Speci- Thornton captures the biplanar shear lag effect due to
fication Table  D3.1, Case  4 (AISC, 2016). Subsequent to the connected and unconnected elements of a section. At
the development of the mathematical model, a parametric the time of that study, the current AISC Specifications

56 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

considered only uniplanar shear lag effects. The recommen- The generalized shear lag factor given for Case 4 is:
dation proposed by Fortney and Thornton was adopted by
3l 2 ⎛ x
the AISC Specification for Case 4 of Table D.3.1 with the U= 1− ⎞ (3)
by-product effect that certain shapes moved from Case 2 to 3l 2 + w 2 ⎝ l ⎠
Case 4. Based on the generalized biplanar model proposed,
The shear lag factor for the B walls is given by:
a similar approach may be taken that will address issues
with Case 6b of Table D3.1 for HSS tension members con- 1 ⎛ x⎞
UB = 2 ⎜1 − ⎟ (4)
nected with two side gusset plates. 1⎛ w⎞ ⎝ lw ⎠
It is postulated that the longitudinal welds connect all four 1+ ⎜ ⎟
3 ⎝ lw ⎠
walls of the HSS and each wall acts similarly to plates with
longitudinal welds (similar to plates as shown for Case 4 in 1 ⎛ 0⎞
= 2 ⎜1 − ⎟
Table D3.1). Thus, only in-plane shear lag exists, eliminat- 1⎛ B⎞ ⎝ lw ⎠
1+ ⎜ ⎟
ing the connection eccentricity (x = 0). Figure 1 shows the 3 ⎝ lw ⎠
migration of the load in the HSS walls to the four weld lines.
As can be seen in the Figure 1, each weld line attracts load 3lw2
=
from both walls connected by the weld. 3lw2+ B2 
The total tension load carried by each wall is proportional
The shear lag factor for the H walls is given by:
to the relative areas of each of the four HSS walls as given
by Equations 1 and 2. 1 ⎛ x⎞
UH = ⎜1 − ⎟ (5)
B ⎞ 1⎛ w⎞ ⎝ 2
l w⎠
PB = Pu ⎛ (1) 1+ ⎜ ⎟
⎝ B + H ⎠ 3 ⎝ lw ⎠
H ⎞ 1 ⎛ 0⎞
PH = Pu ⎛ (2) = ⎜1 − ⎟

⎝ B + H ⎠ 1 ⎛ H ⎞ ⎝ lw ⎠
2
1+ ⎜ ⎟
3 ⎝ lw ⎠
Note that the shear lag factor will be different for the B
and H walls since the l/ w ratio can be different for B and H. 3lw2
=
3lw2 + H 2 

Fig. 1.  In-plane shear lag in rectangular HSS connected with two side gusset plates with longitudinal welds.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 57

 Table 3.  Parametric Study Matrix
Perpendicular Distance
Configuration between Welds Length of Weld
Gusset plates on H side H Varied from 0.25H up to 2H with an increment of 0.25H
Gusset plates on B side B Varied from 0.25B up to 2B with an increment of 0.25B
H + B⎞ ⎛ H + B ⎞ with an increment
Varied from 0.25 ⎛⎜ ⎟ up to 2 ⎜ ⎟
⎝ 2 ⎠ ⎝ 2 ⎠
Gusset plates on H side H
H + B⎞
of 0.25 ⎛⎜ ⎟
⎝ 2 ⎠

H + B⎞ ⎛ H + B ⎞ with an increment
Varied from 0.25 ⎛⎜ ⎟ up to 2 ⎜ ⎟
⎝ 2 ⎠ ⎝ 2 ⎠
Gusset plate on B side B
H + B⎞
of 0.25 ⎛⎜ ⎟
⎝ 2 ⎠

Combining the effects of both walls gives a total shear lag 3l 2

factor as given in Equation 6: UM = (11)
3l + B 2 
2

BUB + HUH
UM = (6)
B+ H 
Parametric Study
Equation 6 can then be used for rectangular and square
A parametric study was performed based on the proposed
HSS as follows
equations for rectangular and square HSS sections; equa-
• Rectangular HSS tions 9 and 11, respectively, for AISC Specification Case 6b.
A total of 76 HSS sections were evaluated, where the thick-
3l 2 est walls were randomly selected from each HSS family. For
UH = (7)
3l + H 2 
2
each specimen, a configuration of gusset plates on the H
side and B side and weld lengths varying from 0.25H up to
3l 2
UB = (8) 2H, with an increment size of 0.25H, was considered. Vari-
3l 2 + B 2  ous such configurations were considered as illustrated in
BUB + HUH Table  3. A comparison of the shear lag factors calculated
UM = (9) based on current AISC Specification Case 6b and the new
B+ H 
equation is presented. Bar charts are plotted to compare the
shear lag factors calculated from the two methods. Compari-
• Square HSS
sons for HSS 12×10×2 (H× B×t) and HSS 20×4×2 (H× B×t)
H = B, UH = UB (10) are provided in Figures  2 and 3, respectively. Refer to the
trends observed in the following section for discussion of
the ratios shown above the bars in Figures 2 and 3. Note that

Fig. 2.  Comparison of shear lag factor for the proposed equation with the
current AISC Specification Case 6b for HSS 12×10×2 (gusset plate on H side).

58 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

the are no bars for the l/H and l/B less than 1, and no ratios is 1.25 and 1.5. When the ratio is 1.75 and 2.00, the
are provided because the current AISC Specification Case proposed model predicts a smaller factor relative to the
6b does not permit connections with l/H and l/B less than 1. current AISC model.
• When using the current AISC model, HSS sections with
Parametric Study
high H/B ratios can result in negative shear lag values (i.e.,
• For HSS sections that are almost square or square—for when l < x ). This can be observed in Figure 3 for the HSS
example, a HSS 12×10×2 —when the l/B ratio is greater 20×4×2 with l/B equal to 1.00. In this case, the shear lag
than 1, the difference in the shear lag factor calculated factor calculated using the proposed model is 0.27 and,
from the current AISC Specification Case  6b procedure from current Case  6b, is negative 0.04 (represented as
compared to the new proposed equation is negligible as zero in Figure 3). When l/B = 1.25, the proposed model
seen in Figure 2 for the currently permitted cases where predicts a value higher than the current Case 6b with the
the weld length is at least as long as the width of the H Ucurrent /Uproposed ratio being 0.63.
dimension. However, the shear lag factor calculated from
• For square HSS sections when l/B is 1.0, the ratio of
the proposed mathematical model is conservative when
Ucurrent /Uproposed is 1.17, and it continuously decreases to a
l/B ratio is equal to 1.
final value of 1.02 when the l/B ratio is increased to 1.25,
• Referring to Figure  2, another observation can be made 1.5, up to 2, as seen in Figure 4.
that as the weld length is increased, the ratio of Ucurrent /
• In general, it can be concluded that the shear lag factors
Uproposed continually decreases from 1.17 to 1.02
calculated using the new proposed equations are slightly
• For gusset plates on the B side of the rectangular HSS conservative as compared to current AISC Specification
member (refer to Figure 3), the shear lag factor calculated Case 6b except for the cases when H/B ratio is quite high.
using the proposed model is higher as compared to the The proposed model also provides a non-zero shear lag
current Case 6b. This was observed when the l/B ratio factor for all cases and will not produce a negative value.

/
*  For l B =1.00, U is negative using the current AISC Specification Case 6b.

Fig. 3.  Comparison of shear lag factor from the proposed equation with the
current AISC Specification Case 6b for HSS 20×4×2 (gusset plate on B side).

Fig. 4.  Comparison of shear lag factor from the proposed equation with the
current AISC Specification Case 6b for HSS 12×12×s (gusset plate on B).

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 59

• The new proposed model permits weld lengths less than the HSS material. The ductile damage criterion deals with
the perpendicular distance between the welds whereas the predicting the initiation of damage due to growth of voids
current procedure does not permit this. and nucleation and provides a relationship among fracture
strain, stress triaxiality, and strain rate. A reasonable value
of 0.1995 was used for the fracture strain based on the stress
FINITE ELEMENT ANALYSIS
strain curve used for the HSS material. When dynamic
The ABAQUS software package, using “explicit” analysis, explicit analysis is performed, it is required to provide
was selected to perform the finite element analysis (FEA). proper definition of density for all the materials. The density
A total of 31 members were modeled and evaluated in of the materials was taken as 0.1 kip/in3 as recommended
ABAQUS. The HSS members were modeled with square by Utsab Dhungana (2014). Table  4 shows the dimensions
corners in lieu of rounded corners, and fillet welds were used of the specimens as modeled in ABAQUS. At the connec-
in place of flare bevel groove welds, which would typically tions, the weld lengths considered were 0.5H, 1.0H, 2.0H,
be required for the rounded corners of an HSS. One model and 1.0B, where H is the longer side of the HSS member.
was developed using rounded corners and flare bevel groove The parameters shown in Table  4 are calculated using the
welds, and it was determined that the analysis is not sensi- following equations:
tive to this particular detail. As such, for simplicity, square
Length of HSS section,
corners with fillets welds were used for this study. The pre-
processing of the model was done in ABAQUS/CAE, and LH = l + 4H + 20 (12)
the ABAQUS dynamic explicit solver was selected to run all Length of gusset plate,
of the finite element analyses. The explicit procedure per- D
forms a large number of small time increments efficiently. It LG = l + 2 (13)
16 
uses a form of the central difference time integration method
to satisfy the dynamic equilibrium equations. Width of gusset plate,
W = H + 2tan(30)l(14)
Material Properties
Thickness of the gusset plate,
The modulus of elasticity was taken as 29,000 ksi, and Pois-
Applied load
son’s ratio was taken as 0.30  in accordance with industry TG = (15)
standards. The material for the gusset plate was defined as Fy W 
elastic perfectly plastic with a yield stress of 50 ksi. A simi-
Size of weld,
lar material type was defined for the weld material with an
ultimate stress of 70 ksi. The stress strain curve for ASTM 0.75 (Applied load )
D= (16)
A500 Grade B, as shown in Figure 5, was used for the HSS 1.392ln 
steel material and was scaled to RyFy and RtFu to account for
expected yield and tensile strengths, respectively. Expected ultimate tensile capacity of the HSS section
The “ductile damage initiation criterion” was assigned to = RtFu Ag(17)

Fig. 5.  Stress-strain curve for ASTM A500 Grade B.

60 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 Table 4.  Model Details
HSS Section Details Gusset Plate Details Weld Details
D
LH Rt Fu Ag LG W TG (16th of l
ID Member (in.) (kips) (in.) (in.) (in.) an inch) (in.)
1a* 38 6.50 11.3 0.50 36 2.0
1b 4×4×2 40 528 6.25 9.00 0.60 18 4.0
1c 44 9.13 13.5 0.40 9.0 8.0
2a* 47 8.75 17.5 0.60 46 3.0
2b 6×6×s 50 1015 8.88 13.0 0.80 23 6.0
2c 56 13.5 20.0 0.51 12 12
3a 56 9.88 19.8 0.75 47 4.0
3b 8×8×s 60 1390 11.0 17.0 0.80 24 8.0
3c 58 17.5 26.5 0.55 12 16
4a 74 10.5 21.0 0.80 36 6.0
4b 80 14.3 26.0 0.65 18 12
12×6×s 1580
4c 92 21.1 40.0 0.40 9.0 24
4d 50 10.5 15.0 1.10 36 6.0
5a 74 12.1 24.3 0.90 49 6.0
5b 12×12×s 80 2140 15.1 26.0 0.85 25 12
5c 92 25.6 40.0 0.55 13 24
6a 92 11.8 26.0 0.70 30 8.0
6b 100 17.9 35.0 0.51 15 16
16×4×s 1770
6c 116 33.0 53.0 0.35 8.0 32
6d 40 11.5 19.0 1.00 60 4.0
7a 92 14.1 28.3 1.10 49 8.0
7b 16×16×s 100 2900 19.1 34.5 0.85 25 16
7c 116 33.6 53.0 0.55 13 32
8a 110 13.0 32.0 0.55 24 10
8b 120 21.5 23.0 0.40 12 20
20×4×2 1730
8c 140 40.8 66.2 0.27 6.0 40
8d 40 9.88 15.8 1.20 47 4.0
9a 110 15.0 31.6 0.95 40 10
9b 120 22.5 43.1 1.00 20 20
20×12×s 2900
9c 140 41.3 66.2 0.45 10 40
9d 80 16.1 26 1.15 33 12
* Specimens were not used because the required weld sizes exceeded the width of the HSS wall.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 61

 For specimen  1a, the weld size required to transfer the capacity RtFu Ag of the HSS section. The FEA results for all
expected tensile capacity of the HSS section was calculated the specimens are provided in Table 5.
to be 2.25  in. Two such welds would be required on each The development of stresses in specimen 5b is shown in
wall for the desired geometry of the analytical specimen, Figures 7(a), (b), (c), and (d). Figure 7(a) shows the specimen
resulting in a dimension of 4.5 in., which is greater than the when there is zero load. Figure 7(b) shows the stress distri-
width of the HSS. A similar comment can be made for spec- bution when 85% of the applied load is transferred by the
imen 2a. Although a calculation can be made for the shear HSS section to the fixed end. At section b-b in Figure 7(b),
lag using the mathematical model, for the reasons previously it can be observed that there is a uniform distribution of
mentioned, specimens 1a and 2a were not able to be modeled stresses, whereas away from section b-b and closer to the
in ABAQUS. connection region, there is nonuniform stress distribution.
The interaction of welds with the gusset plates and the Similar stress distribution patterns were observed on each
HSS member was defined using tie constraints; tie con- wall because of symmetry. In Figure  7(c) at section c-c,
straints are found to be effective for welded connections necking was observed, and the load transferred by the HSS
(Ruffley, 2010). A fixed boundary condition was applied to section at that time was 86% of the applied load. Figure 7(d)
the gusset plates attached to one end of the HSS member, shows fracture initiation at section d-d when the load trans-
and load was applied on the gusset plates attached to the ferred by the HSS section is equal to 87% of the applied
other end of the HSS using kinematic coupling interactions. load. Figure 8 shows an axonometric view of specimen 5b
Figure 6 shows the assembly configuration, partitioning of after fracture.
the members, kinematic coupling interactions, and the welds Table 6 shows a comparison between shear lag factors cal-
attached to the gusset plate and HSS member via tie con- culated from the three methods: AISC Specification Case
straints for specimen HSS 12×12×s. 6b (UA), mathematical model (UM), and FEA (UF).
Using the values provided in Table  6, graphs were gen-
erated, as shown in Figure 9, to compare the values of the
RESULTS AND DISCUSSION
shear lag factor from the three methods. Figure 9 compares
All of the HSS specimens fractured at a section outside of the shear lag values for HSS 16×4×s (specimen 6). It can
the connection region. It was observed that the section frac- be observed that when l/H is less than 1.0, current AISC
tured at a load smaller than the calculated expected tensile Specification Case  6b does not permit such a connection,
capacity of the section. All of the specimens yielded, and but the mathematical model gives a value of UM  = 0.53,
necking was clearly visible, followed by complete rupture of which is conservative compared to the value obtained from
the HSS section. The reactions were taken at the gusset plate FEA (UF = 0.76). When l/H is 2.0, using the current AISC
at the fixed boundary end of the specimen. The maximum Specification Case 6b, the HSS section is 99% effective to
value of the reaction at the fixed end is taken as the fracture carry or transfer load thus giving a shear lag factor of UA =
load. The shear lag coefficient, UF, was calculated as the 0.99. However, the mathematical model and the FEA predict
ratio of this reaction force to the calculated expected tensile a shear lag value of UM = UF = 0.94.

Fig. 6.  Assembly configuration for HSS 12×12×s.

62 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 Table 5.  FEA Results
HSS Section Details Weld Configuration
Specimen Rt Fu Ag l Failure Load Shear Lag
Designation Member (kips) (in.) (kips) Factor, UF
1a 0.5H 2.0 NA NA
1b 4×4×2 528 1.0H 4.0 485 0.92
1c 2.0H 8.0 460 0.87
2a 0.5H 3.0 NA NA
2b 6×6×s 1010 1.0H 6.0 866 0.85
2c 2.0H 12 965 0.95
3a 0.5H 4.0 1160 0.83
3b 8×8×s 1390 1.0H 8.0 1190 0.86
3c 2.0H 16 1310 0.94
4a 0.5H 6.0 1350 0.85
4b 1.0H 12 1370 0.86
12×6×s 1580
4c 2.0H 24 1470 0.93
4d 1.0B 6.0 1480 0.94
5a 0.5H 6.0 1670 0.78
5b 12×12×s 2140 1.0H 12 1870 0.87
5c 2.0H 24 2010 0.94
6a 0.5H 8.0 1340 0.76
6b 1.0H 16 1510 0.85
16×4×s 1770
6c 2.0H 32 1660 0.94
6d 1.0B 4.0 1330 0.75
7a 0.5H 8.0 2080 0.72
7b 16×16×s 2900 1.0H 16 2480 0.86
7c 2.0H 32 2730 0.94
8a 0.5H 10 1160 0.67
8b 1.0H 20 1619 0.93
20×4×2 1730
8c 2.0H 40 1590 0.92
8d 1.0B 4.0 949 0.55
9a 0.5H 10 2290 0.79
9b 1.0H 20 2650 0.91
20×12×s 2900
9c 2.0H 40 2680 0.92
9d 1.0B 12 2440 0.84

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 63

 (a) Unloaded (b) 85% load (c) 86% load (d) 87% load
(necking) (fracture)

Fig. 7.  Development of stresses and fracture initiation in specimen 5b.

Fig. 8.  Axonometric view of specimen 5b after fracture.

64 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 Table 6.  Shear Lag Factors Comparison
AISC Mathematical
HSS Section Details Specification Model FEA
Specimen
Designation Section UA* UM UF
1a NA 0.43 NA
1b 4×4×2 0.88 0.75 0.92
1c 0.94 0.92 0.87
2a NA 0.16 NA
2b 6×6×s 0.88 0.75 0.85
2c 0.94 0.92 0.95
3a NA 0.43 0.83
3b 8×8×s 0.88 0.75 0.86
3c 0.94 0.92 0.94
4a NA 0.54 0.85
4b 0.96 0.81 0.86
12×6×s
4c 0.98 0.94 0.93
4d 0.67 0.54 0.94
5a NA 0.43 0.78
5b 12×12×s 0.875 0.75 0.87
5c 0.94 0.92 0.94
6a NA 0.53 0.76
6b 0.99 0.8 0.85
16×4×s
6c 0.99 0.94 0.94
6d 0.2 0.28 0.75
7a NA 0.43 0.72
7b 16×16×s 0.88 0.75 0.86
7c 0.94 0.92 0.94
8a NA 0.52 0.67
8b 0.99 0.79 0.93
20×4×2
8c 0.99 0.93 0.92
8d NA 0.21 0.55
9a NA 0.52 0.79
9b 0.94 0.8 0.91
20×12×s
9c 0.97 0.94 0.92
9d 0.74 0.61 0.84
* N/A denotes cases not currently permitted in AISC Specification for weld lengths less than H or B.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 65

 Fig. 9.  Shear lag factor comparison.

66 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 CONCLUSIONS TG Thickness of the gusset plate, in.
Based on the presented results, the following can be U Shear lag factor
concluded.
UA Shear lag factor calculated using the current AISC
1. From the parametric study, the proposed mathematical Case 6b
model is conservative as compared to the current AISC
UB Shear lag factor for B walls
Specification Table D3.1, Case 6b, except for some cases
when the gusset plate is attached on the B side of the HSS. UCurrent Shear lag factor calculated using the current AISC
Case 6b
2. The proposed model allows for weld lengths less than the
distance between the welds. UF Shear lag factor obtained from the finite element
analysis
3. The stress distribution pattern on which the mathematical
model is based has good correlation with the FEA results. UH Shear lag factor for H walls
4. The results from the finite element analyses are consistent UM Shear lag factor calculated using the mathematical
with the values of the shear lag factors calculated using model
the proposed mathematical model.
UProposed Shear lag factor calculated using the proposed
mathematical model
RECOMMENDATIONS
W Width of the gusset plate, in.
1. It is recommended that AISC adopt the mathematical
l Length of the weld, in.
model proposed here and revise Case 6b of Table D3.1 of
the AISC Specification. n Number of welds
2. Experimental testing should be performed to validate the x Connection eccentricity, in.
results of the finite element analysis and provide more
w Width of the HSS, in.
insights on some of the conclusions mentioned earlier.
3. The FEA models can be further refined by using actual REFERENCES
material properties, measured from coupon tests, for the
HSS ASTM A500 Grade B material. ABAQUS (2013), Abaqus Analysis User’s Manual, Ver-
sion 6.10, Simulia, Providence, R.I.
SYMBOLS AISC (2010), Specification for Structural Steel Buildings,
ANSI/AISC 360-10, American Institute of Steel Con-
Ag Gross area of the HSS section, in.2 struction, Chicago, Ill.
B Shorter side of the HSS section, in. AISC (2016), Specification for Structural Steel Buildings,
ANSI/AISC 360-16, American Institute of Steel Con-
D Weld size in sixteenths of an in. struction, Chicago, Ill.
Fy Yield stress, ksi Dhungana, U. (2014), “Shear Lag Factor for Longitudinally
Fu Ultimate tensile stress of the HSS section, ksi Welded Tension Members Using Finite Element Method,”
MS Thesis, School of Civil and Environmental Engineer-
H Longer side of the HSS section, in. ing, University of Cincinnati, Ohio.
LH Length of the HSS section, in. Fortney, P.J., and Thornton, W.A. (2012), “Recommenda-
tions for Shear Lag Factors for Longitudinally Welded
LG Length of the gusset plate, in. Tension Members,” Engineering Journal, AISC, Vol. 49,
PB Tension load carried by B wall, kips No. 1.
PH Tension load carried by H wall, kips Ruffley, D.J. (2010), “A Finite Element Approach for Model-
ing Bolted Top-and-Seat Angle Components and Moment
Pu Total tension load, kips Connections,” MS Thesis, School of Civil and Environ-
Rt Correction factor mental Engineering, University of Cincinnati, Ohio.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 67

68 / ENGINEERING JOURNAL / FIRST QUARTER / 2020
Continuous Bracing Requirements for
Constrained-Axis Torsional Buckling
MARK D. DENAVIT, WILLIAM P. JACOBS V, and TODD A. HELWIG

ABSTRACT
The design of floor and roof framing members is typically controlled by flexural demands; however, if a member serves as a chord or col-
lector, it can also be subjected to significant axial compression. Continuous restraint provided by the floor or roof diaphragm is commonly
assumed in design to provide adequate bracing of connected wide-flange members against minor-axis flexural buckling; however, these
members are still susceptible to major-axis flexural buckling and potentially to torsional buckling about a constrained axis located at the top
flange. In addition to the lateral restraint, floor and roof decking systems can also provide continuous torsional restraint through their flexural
stiffness and strength. This restraint can be used to increase the calculated constrained-axis torsional buckling strength or inhibit the mode
altogether. In this paper, the specific case of a wide-flange steel beam-column with both lateral and torsional restraint located at the top flange
is investigated, and torsional bracing requirements are derived. The focus of the study is on continuous torsional bracing and its effect on the
constrained-axis torsional buckling mode. The requirements are illustrated through a design example, and a parametric study is performed
examining typical floor and roof decking system configurations, identifying cases where improved design efficiency can be achieved.

Keywords:  Axial strength, constrained-axis torsional buckling, stability, wide-flange shapes.

INTRODUCTION Specification Section E4 (AISC, 2016). However, if the lat-

eral bracing is offset from the centroid, the member can fail
I n building structures, floor or roof framing members that
serve as chords or collectors can accumulate significant
axial load as they transfer loads from the diaphragm to
by constrained-axis torsional buckling as depicted in Fig-
ure 1(d) (for a case where the lateral bracing is located at the
top of the top flange of the member), and the expressions
the vertical elements of the lateral-force-resisting system.
provided in the Commentary on Section E4 are necessary as
The strength of these members in compression can be gov-
outlined in the next section of this paper.
erned by a number of different buckling modes. Figure  1
When wide-flange beams are laterally restrained by a
depicts the potential member buckling modes that might
decking system, the limit states of minor-axis flexural buck-
control the axial strength. Although designers are familiar
ling and torsional buckling [twisting about the shear cen-
with the flexural buckling modes for wide-flange columns
ter as shown in Figure 1(c)] are not applicable. As stated in
as shown in Figures 1(a) and 1(b), these members are also
AISC Seismic Design Manual (AISC, 2018), Table 8-1, this
susceptible to torsional buckling when the bracing does not
restraint occurs for bare steel deck (either steel roof deck
prevent twist and the unbraced length for torsional buckling
or composite steel deck prior to placement of the concrete)
is larger than the unbraced length for minor-axis flexural
when the ribs of the deck are perpendicular to the beam and
buckling. If the lateral bracing is located at the shear cen-
for composite slabs (i.e., composite steel deck with concrete
ter (which coincides with the centroid for wide-flange sec-
fill) in any orientation. The axial strength of the member for
tions), the torsional buckling mode as depicted in Figure 1(c)
these cases is calculated from the limit states of major-axis
is possible. The strength for torsional buckling about the
flexural buckling and constrained-axis torsional buckling.
shear center can be predicted using the expressions in AISC
Bare steel deck with ribs parallel to the beam is commonly
assumed to provide inadequate lateral restraint and thus is
not addressed in this paper.
The discussion thus far has focused on cases in which the
Mark D. Denavit, Assistant Professor, University of Tennessee, Knoxville, decking system only restrains lateral movement. However,
Tenn. Email: [email protected] (corresponding) decking systems also provide torsional restraint to the beam.
William P. Jacobs V, Principal, Stanley D. Lindsey and Associates, Ltd., While it is conservative to neglect this restraint, in many
Atlanta, Ga. Email: [email protected] cases the restraint is sufficient to brace the beam against
Todd A. Helwig, Professor, J. Neils Thompson Teaching Fellow in Civil Engi- constrained-axis torsional buckling or otherwise signifi-
neering, University of Texas at Austin, Austin, Tex. Email: [email protected].
edu
cantly increase the calculated strength. In this paper, brac-
ing requirements for constrained-axis torsional buckling are
developed. A detailed example is presented that illustrates
Paper No. 2019-07

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 69

the use of these requirements for common applications such accounting for slender elements is covered in more detail
as the design of collector beams or other members subjected later in this section as well as in the example problem that is
to large axial forces. Finally, the results of a broad paramet- presented in this paper.
ric study are presented with observations that highlight con- For constrained-axis torsional buckling with bracing off-
figurations where the benefits of the torsional restraint are set along the minor axis, as shown in Figure  2, the elas-
most effective. tic buckling stress is given by AISC Specification (2016)
Although the “beams” in the flooring or roofing system Commentary Equation C-E4-1, shown here as Equation 3.
that are discussed in this paper are actually “beam-columns” The fundamental equation for this mode was developed by
due to the combined axial force and bending moment, the Timoshenko and Gere (1961). The expression presented in
term “beams” is used throughout the paper for convenience. the Commentary is slightly modified as recommended by
The focus of the bracing requirements in the paper is the Errera and Apparao (1976) and Helwig and Yura (1999) to
stiffness and strength requirements necessary to control the account for practical limitations that often occur in practice.
constrained-axis torsional buckling mode from the axial
⎡ π 2 EI y ⎛ ho 2 ⎞ ⎤ 1
force component. Fe = ω ⎢ 2 ⎜ + a 2 ⎟ + GJ ⎥ 2 (3a)
⎣ Lcz ⎝ 4 ⎠ ⎦ Ag ro 
CONSTRAINED-AXIS TORSIONAL r 2o = r 2x + r 2y + a2(3b)
BUCKLING STRENGTH
where
The design of members for compression is governed by the E = modulus of elasticity of steel = 29,000 ksi
provisions of Chapter E in the AISC Specification (AISC, G = shear modulus of steel = 11,200 ksi
2016). The nominal axial compressive strength in the elas- Iy = minor-axis moment of inertia, in.4
tic and inelastic range is determined from the column curve
J = torsional constant, in.4
expressions given in Section E3:
Lcz = effective torsional length, in.
Pn = FcrAg(1) a = distance from centroid to brace point, in.
0.658(
Fy Fe ) ho = distance between flange centroids, in.
Fy when Fy Fe 2.25
Fcr = (2)
0.877Fe when Fy Fe > 2.25 rx = major-axis radius of gyration, in.
 r y = minor-axis radius of gyration, in.
where ω = finite brace stiffness factor = 0.9
Ag = gross cross-sectional area, in.2
Although wide-flange sections with slender elements are
Fcr = critical stress, ksi not typically used as columns, many beam-type sections are
Fe = elastic buckling stress, ksi slender for axial compression. As a result, a wide-flange
Fy = yield stress, ksi section that might be used in a flooring system may pos-
Pn = nominal compressive strength, kips sess slender elements for compression. In such a case, the
interaction of constrained-axis torsional buckling and local
If a section has slender elements, the provisions of Sec- buckling should be accounted for in accordance with AISC
tion E7 can be used to determine the effective area, Ae, that Specification Section E7. The effective area, Ae, is computed
is used in place of Ag in Equation 1. The methodology for as a function of the critical buckling stress, Fcr. For the case

Major-axis Minor-axis Torsional Constrained-axis

flexural buckling flexural buckling buckling torsional buckling
(a) (b) (c) (d)

Fig. 1.  Buckling modes.

70 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

of wide-flange steel members, Ae is computed as shown in control torsional buckling and constrained-axis torsional
Equation 4, where the effective widths of the flange and web buckling were developed by Helwig and Yura (1999), based
are computed using Equations 5 and 6, respectively. upon the results of a parametric finite element study. For the
case of discrete bracing and considering inelastic buckling,
Ae = Ag − 4(bf / 2 − be)tf − (h − he)tw(4)
the required torsional brace stiffness is given by Equation 7.
⎧
{P r }
2
bf 2 when λ f ≤ λ rf Fy Fcr
⎣( ho 4 ) + a ⎦
2 * ⎡ 2 2⎤
⎪⎪ u o − Pny
be = ⎨ ⎛ λ rf Fy ⎞ λ rf Fy b f βT =A (7a)
⎪ ⎜ 1 − 0.33
λf
⎟ 1.49
λf
when λ f > λ rf Fy Fcr ( nb 4 τEI y L ) ⎡⎣( ho2 4 ) + a 2 ⎤⎦
⎪⎩ ⎝ Fcr ⎠ Fcr 2


(5a) 2a
 A=4 ≥ 2.0 (7b)
E ho 
λ rf = 0.56 (5b)
Fy  *
2
EI y
Pny = 0.877 (7c)
bf L2 
λf = (5c)
2t f 
where
⎧ h when λ w ≤ λ rw Fy Fcr L = beam span, in.
⎪⎪
he = ⎨ ⎛ λ rw Fy ⎞ λ rw Fy Pu = required axial compressive strength, kips
⎪ ⎜ 1 − 0.24 ⎟ 1.31 h when λ w > λ rw Fy Fcr
nb = number of intermediate braces
⎝ λw Fcr ⎠ λw Fcr
⎪⎩ βT = required brace stiffness
(6a)

τ = stiffness reduction factor
E
λ rw = 1.49 (6b)
Fy  The stiffness reduction factor, τ, is applied to account for

the reduced stiffness and strength due to inelasticity and
h
λw = (6c) local buckling and has been modified from the original pre-
tw  sentation given in Helwig and Yura (1999) to reflect the pro-
visions in the current AISC Specification (2016), as shown
Further discussion of constrained-axis torsional buckling
in Equation 8.
strength including design tables and example calculations
are provided in Liu et al. (2013). AISC Design Guide  25 ⎧ − 2.724 ( P P ) ln ( P xP ) when P xP > 0.39
⎪ u y u y u y
(Kaehler et al., 2011) describes recommendations for com- τ=⎨
puting the constrained-axis torsional buckling strength for ⎪ x when Pu xPy ≤ 0.39
⎩
singly symmetric and tapered members. (8a)

Ae Fcr =Pu Ae
BRACING REQUIREMENTS x= (8b)
Ag 
Effective stability bracing must possess adequate stiffness
and strength. Brace stiffness and strength requirements to where
Py = axial yield strength, kips
= FyAg
x = ratio of the effective area (calculated iteratively for
a critical stress equal to the required axial strength
divided by the effective area) to the gross area

Alternatively, for previous versions of the AISC Specifica-

tion (e.g., 2010) where the strength of members with slender
elements was calculated using the net reduction factor, Q,
the stiffness reduction factor should be taken as Equation 9,
where Q is calculated at the required axial strength.
⎧ −2.724 ( P QP ) ln ( P QP ) when P QP > 0.39
⎪ u y u y u y
τ=⎨
⎪⎩ 1.0 when Pu QPy ≤ 0.39

(9)
Fig. 2.  Bracing offset along minor axis.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 71

 The required brace stiffness (Equation 7a) is modified for where βTb = required stiffness of the decking system.
the case of a member braced continuously at the top flange As noted by Helwig and Yura (1999), the required brace
by (1)  substituting in the location of the brace at the cen- stiffness limits the twist of the beam due to the applied load-
troid of the top flange (a  = ho/ 2) and (2)  converting from ing to a value equal to the initial twist imperfection. Thus,
discrete to continuous bracing by setting the term nb/ L equal the resulting brace strength requirement is the product of the
to unity (note that after this change, β T is expressed as the required brace stiffness, βT, and the initial twist imperfec-
torsional stiffness per unit length). Additionally, a resistance tion θ 0, as shown in Equation 13. The assumed initial twist
factor of ϕ = 0.75 is applied according to AISC Specifica- imperfection is based on a configuration where one flange
tion Appendix 6 (2016). The resulting expression is given by is straight and the other flange has a lateral initial out-of-
Equation 10. straightness with maximum lateral imperfection of L/ 500.
2 This results in an assumed initial twist imperfection given
1.5 ( Puro ho 2 2 )
2 *
Pny by Equation 14.
T = (10)
EI y ho 2  Mbr = βTθ0(13)
There are a number of factors that can affect the total L
0= (14)
brace stiffness, including the stiffness of the decking system 500ho 
as well as other stiffness components such as cross-sectional
distortion due to flexibility in the beam web and connection Often, the stiffness provided by the decking system is
stiffness between the decking system and beam. In the deri- more than sufficient to brace the beam, leaving strength as
vation presented herein, the total brace stiffness is based on the limiting brace requirement. In these cases, the required
the stiffness of the decking system acting in series with the moment strength can be reduced as a function of the ratio
distortional stiffness of the beam web, βsec, given in Equa- of required to provided stiffness as given by Equation  15
tion 11 (AISC Specification Equation A-6-13). In this paper, based on Equation  C-A-6-2 from the AISC Specification
the connection between the decking system and the beam Commentary (2016). The total provided stiffness, βprov, is
is assumed to be rigid. Accordingly, the required stiffness computed using an expression for springs in series given
of the decking system, βTb , is given by Equation 12 (AISC in Equation  16, where the stiffness provided by the deck,
Specification Equation  A-6-10). Note that if the web dis- βprov-b, is combined with the web distortional stiffness, βsec.
tortional stiffness is less than or equal to the required total βT θo
Mbr = (15)
brace stiffness (βsec ≤ βT), then Equation 12 is negative and ( 2 − βT β prov)

the required total brace stiffness cannot be achieved regard-
less of the stiffness provided by the decking system. 1 1 1
= + (16)
prov prov b sec 
3.3Et w3
sec = (11)
12h0  The example outlined in the following section illustrates
the use of these bracing provisions for details commonly
T
Tb = (12) found in practice.
T
1
sec 
DESIGN EXAMPLE

Given:
Consider a 24‑ft-long W18×35 ASTM A992 beam supporting a composite slab that has a total depth of 6 in., normal weight
concrete (ƒc′ = 3 ksi), and 3‑in. deep, 20-ga. composite steel deck. The deck spans perpendicular to the beams that are spaced at
10 ft. Steel headed stud anchors of ASTM A108 material with a diameter of w in. are provided at a spacing, s, of 1 ft. The beam
is assumed to be simply supported with twist restrained but warping deformations permitted at the ends.

Solution:
The problem is worked in multiple steps to illustrate the various modes as well as the contributions of the bracing. Baseline cal-
culations are first carried out to understand the flexural and torsional modes, neglecting the contributions of the composite slab.
It should be understood that although the lateral stiffness of the composite slab is neglected in the calculation of the minor-axis
flexural buckling mode, the lateral stiffness is required for the constrained-axis torsional buckling mode to occur. The shear
stiffness of most typical decking systems (bare composite steel or roof deck with ribs perpendicular to the braced member or
composite slabs in any orientation) is most often much larger than necessary to provide adequate lateral bracing.

72 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

Baseline Calculations
Determine the nominal strength for the limit state of constrained-axis torsional buckling neglecting the torsional stiffness pro-
vided by the decking system.
From AISC Manual Table 2-4, the beam yield stress is:
Fy = 50 ksi
From AISC Manual Table 1-1 (AISC, 2017a), the relevant section properties for the beam are:
Ag = 10.3 in.2 Ix = 510 in.4 Iy = 15.3 in.4 J = 0.506 in.4
bf = 6.00 in. d = 17.7 in. t w = 0.300 in. tf = 0.425 in.
rx = 7.04 in. r y = 1.22 in. bf/ 2tf = 7.06 h/ t w = 53.5
ho = 17.3 in. h = 16.1 in. k1 = 0.75 in.
The effective length of the beam is taken as the full span because twist is restrained but warping deformations are permitted at
the ends of the beam.
Lcz = L
= (24 ft)(12 in./ft)
= 288 in.

To determine the elastic buckling stress, Fe, using Equation 3a, first determine a and r 2o.
ho
a= (3b)
2
17.3 in.
=
2
= 8.65 in.

r 2o = r 2x + r 2y + a2
= (7.04 in.)2 + (1.22 in.)2 + (8.65 in.)2
= 126 in.2

⎡ π 2 EI y ⎛ ho 2 ⎞ ⎤ 1
Fe = ω ⎢ 2 ⎜ + a 2 ⎟ + GJ ⎥ 2 (3a)
⎣ L cz ⎝ 4 ⎠ ⎦ Ag ro
π 2 ( 29,000 ksi ) (15.3 in.4 ) ⎛ (17.3 in.)
2
2
⎞ ⎤ 1
= 0.90 ⎜ + ( 8.65 in. ) ⎟ + (11, 200 ksi ) ( 0.506 in. 4 ⎥
)
( 288 in.)
2 ⎜ 4 ⎟ ⎥ (10.3 in. )(126 in.2 )
2
⎝ ⎠ ⎦
= 9.42 ksi 

Compute the critical stress, Fcr , using Equation 2.

Fy 50 ksi
=
Fe 9.42 ksi
= 5.31 > 2.25

Because Fy /Fe > 2.25, the critical stress is determined as:

Fcr = 0.877Fe(2)
= 0.877(9.42 ksi)
= 8.26 ksi

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 73

Classify the flange for local buckling.
E
rf = 0.56 (5b)
Fy

( 29,000 ksi )
= 0.56
( 50 ksi )
= 13.5 

bf
f = (5c)
2t f
= 7.06 < rf 

Thus, the flanges are nonslender, and be = bf / 2.

Classify the web for local buckling.
E
rw = 1.49 (6b)
Fy

( 29,000 ksi )
= 1.49
( 50 ksi )
= 35.9 

h
w = (6c)
tw
= 53.5 > rw

Thus, the web is slender for axial compression.

Fy
To calculate the effective width, he, using Equation 6a, first determine the relationship between λw and rw .
Fcr

Fy ( 50 ksi )
rw = ( 35.9 ) (6)
Fcr (8.26 ksi )
= 88.3 > w = 53.5 

Thus, the critical stress is low enough that no reduction is necessary.

he = h  (6a)
Ae = Ag (from Eq. 4)

Compute the axial compressive strength, Pn, using Equation 1.

Pn = Fcr Ae(1)
= Fcr Ag
= (8.26 ksi)(10.3 in.2)
= 85.1 kips

The nominal strengths of the other modes of buckling assuming no bracing provided by the decking system are presented in
Table 1, where Pnx, Pny, Pnz, and Pnca are the computed strengths for the major-axis flexural, minor-axis flexural, torsional, and

74 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

 Table 1.  Axial Strength for Various Buckling Modes
Buckling Mode Axial Strength
Major-axis flexural buckling Pnx = 408 kips
Minor-axis flexural buckling Pny = 46.4 kips
Torsional buckling Pnz = 165 kips
Constrained-axis torsional buckling Pnca = 85.1 kips

constrained-axis torsional buckling modes, respectively. A comparison of the buckling capacities provides some interesting
insights into the behavior. The minor-axis flexural buckling strength is significantly smaller than the torsional buckling strength
because the unbraced length is the same for these two modes. Torsional buckling will always yield a larger strength than minor-
axis flexural buckling for wide-flange members of the same unbraced length. The table also demonstrates the significant reduc-
tion in the strength for the constrained-axis torsional buckling mode versus the torsional buckling mode, which will always be
the case when the location of bracing is offset along the minor-axis of the section. These values vary with the unbraced length
as shown in Figure 3.

Bracing Requirement Checks

For the configuration described previously, determine if the decking system is adequate to brace the beam against constrained-
axis torsional buckling at a required axial load of Pu = 250 kips.
To determine if the decking system is adequate to brace the beam, both stiffness and strength checks are necessary. First, the
required stiffness is calculated, followed by an evaluation of the available stiffness. The required strength is then calculated
(including a reduction based on the ratio of required stiffness to provided stiffness) and finally the available strength.

Fig. 3.  Variation of axial strength with length.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 75

Required Brace Stiffness
Recall that the value of x is the ratio of the effective area to the gross area at a critical stress equal to the required strength divided
by the effective area. The x value needs to be determined iteratively. For this case, x = 0.985, as confirmed in the following cal-
culations. A trial critical stress, Fcr, is calculated according to the definition of x given in Equation 8b.
Ae = xAg (from Eq. 8b)
2
= (0.985)(10.3 in. )
= 10.1 in.2

Pu
Fcr =
Ae
( 250 kips )
=
(10.1 in.2 )
= 24.8 ksi

As demonstrated in the baseline calculations, the flanges are nonslender (be = bf / 2) and the web is slender. Calculate the effective
width, he, using Equation 6.

Fy ( 50 ksi )
rw = ( 35.9 )
Fcr ( 24.8 ksi )
= 51.0 < w

⎛ λ rw Fy ⎞ λ rw Fy
he = ⎜ 1 − 0.24 ⎟ 1.31 h (6a)
⎝ λw Fcr ⎠ λw Fcr
⎛ ( 35.9 ) ( 50 ksi ) ⎞ ( 35.9 ) ( 50 ksi )
= ⎜ 1 − 0.24 ⎟ 1.31 (16.1 in.)
⎜ ⎟
( 53.5) ( 24.6 ksi ) ⎠ ( 53.5) ( 24.6 ksi )
⎝
= 15.5 in. 

Compute the effective area, Ae, using Equation 4.

Ae = Ag − ( h − he ) t w (4)
= (10.3 in. ) − ⎡⎣(16.1 in.) − (15.5 in.)⎤⎦ ( 0.300 in.)
2

= 10.1 in.2 

Compute the value x using Equation 8b.

Ae
x= (8b)
Ag
(10.1 in.2 )
=
(10.3 in.2 )
= 0.981 

The result is within the rounding error of the trial value; thus, x = 0.985 is confirmed.

76 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

Compute the stiffness reduction factor, τ, using Equation 8a.
Py = AgFy
= (10.3 in.2)(50 ksi)
= 515 kips

Pu ( 250 kips )
=
xPy ( 0.985) ( 515 kips )
= 0.493 > 0.39

Pu
Because > 0.39, τ is calculated as:
xPy

⎛P ⎞ ⎛ P ⎞
τ = −2.724 ⎜ u ⎟ ln ⎜ u ⎟ (8a)
⎝ Py ⎠ ⎝ xPy ⎠
⎡ ( 250 kips) ⎤ ⎡ ( 250 kips) ⎤
= −2.724 ⎢ ⎥ ln ⎢ ⎥
⎢⎣ ( 515 kips) ⎥⎦ ⎢⎣ ( 0.985) ( 515 kips) ⎥⎦
= 0.936 

*
Compute Pny , using Equation 7c.
2
* EI y
Pny = 0.877 (7c)
L2
2
( 29,000 ksi )(15.3 in.4 )
= 0.877 ( 0.936 ) 2
( 288 in.)
= 43.3 kips 

*
The value Pny represents the minor-axis flexural buckling strength considering the full length of the beam and with the level of
inelasticity expected at the required axial strength. The value varies as shown in Figure 4.
Compute the required total brace stiffness, βT, using Equation 10.

*
Fig. 4.  Variation of Pny with required axial strength.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 77

 2
1.5 ( Puro ho 2 2 )
2 *
Pny
T = (10)
EI y ho 2
2 2
1.5 ( 250 kips )(126 in.2 ) ( 43.3 kips )(17.3 in.) 2
=
( 0.75) ( 0.936 )( 29,000 ksi )(15.3 in.4 )(17.3 in.)2
= 10.1 kip-in./rad/in. 

Compute the distortional stiffness of the beam web, βsec, using Equation 11.
3
3.3Et w3 3.3 ( 29,000 ksi ) ( 0.300 in.)
sec = = (11)
12h0 12 (17.3 in.)
= 12.5 kip-in./rad/in. 

Compute the required stiffness of the decking system, βTb, using Equation 12.
βT
βTb = (12)
⎛ βT ⎞
⎜1 − ⎟
⎝ βsec ⎠
(10.1 kip-in./rad/in.)
=
⎡ (10.1 kip-in./rad/in.) ⎤
⎢1 − ⎥
⎢⎣ (12.5 kip-in./rad/in.) ⎥⎦
= 52.6 kip-in./rad/in. 

Fig. 5.  Variation of required torsional bracing.

78 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

The required stiffness of the decking system varies with the required axial strength as shown in Figure 5. As can be seen for
this case, the required stiffness is zero (or near zero due to simplifications in the derivation of the stiffness requirement) at the
constrained-axis torsional buckling strength. It is a fairly modest value at the torsional buckling strength. The required stiffness
of the decking system increases asymptotically to infinity for a required axial strength less than the major-axis flexural buckling
strength, indicating that the distortional stiffness of the web is insufficient to make major-axis buckling the controlling mode of
failure. Although it would be possible to reduce or eliminate cross-sectional distortion by providing transverse web stiffeners or
struts between the decking system and bottom flange at a few locations along the length of the beam, the fabrication costs are
likely to make such detailing impractical. Selecting a beam with a stockier web would be more prudent. It should be noted that
the W18×35 has a web slenderness of 53.5 and is among the most slender rolled W-shaped sections. Sections with stockier webs
will have fewer issues with cross-sectional distortion.

Available Brace Stiffness

To determine the available stiffness, the decking system is assumed to act in single curvature on each side of the span, as shown
in Figure 6, and the adjacent beams are assumed to have similar demands on the decking system as the example beam. Thus,
the stiffness contribution from each side is 2EI/L, where EI is the flexural rigidity of the decking system and L is the deck span
(Figure 7). Rigid connections are assumed between the beam and decking system.
On one side of the beam (left side in Figure 6), the concrete is in tension, and thus only the moment of inertia of the composite
steel deck is conservatively relied upon. The moment of inertia of 20‑ga. composite steel deck with a 3‑in. depth is 0.920 in.4/ft
(Sputo, 2014). On the other side of the beam (right side in Figure 6), the concrete is in compression, and the moment of inertia
of the composite slab is considered. The design moment of inertia of a 6‑in. total depth composite slab constructed with 20‑ga.
composite steel deck with a 3‑in. depth and normal weight concrete is 13.34 in.4/ft (Sputo, 2014). Perimeter beams or those adja-
cent to an opening should only consider the stiffness of the side where the decking system is present.

Fig. 6.  Assumed deck bending mode.

Fig. 7.  Deck bending stiffness model.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 79

Calculate the stiffness provided by the decking system, βprov-b, as the sum of the contributions from both sides.
2 EI ⎞ 2EI ⎞
β prov − b = ⎛ +⎛
⎝ L ⎠ left ⎝ L ⎠ right

2 ( 29, 000 ksi ) ( 0.920 in.4 ft ) 1 ft 12 in. 2 ( 29, 000 ksi ) (13.34 in.4 ft ) (1 ft 12 in.)
= +
(10 ft ) (12 in. 1 ft ) (10 ft ) (12 in. 1 ft )
= 574 kip-in./rad/in.

Because the stiffness provided by the decking system (βprov-b = 574 kip-in./rad/in.) is greater than the required stiffness of the
decking system (βTb = 52.6 kip-in./rad/in.), the decking system has sufficient stiffness to brace the beam against constrained-axis
torsional buckling at the required axial strength.

Required Brace Strength

The required brace strength is determined in accordance with Equation 15, which includes a reduction for surplus provided
stiffness. Although the available stiffness of the decking system is substantially larger than the required stiffness, the effects of
distortion need to be considered using Equation 16 to determine the provided total brace stiffness, βprov.
−1
⎛ 1 1 ⎞
β prov =⎜ + ⎟ (from Eq. 16)
⎝ β prov − b βsec ⎠
−1
⎡ 1 1 ⎤
=⎢ + ⎥
⎢⎣ ( 574 kip-in. rad in.) (12.5 kip-in./rad/in.) ⎥⎦
= 12.2 kip-in./rad/in. 

This calculation demonstrates that cross-sectional distortion associated with web flexibility dominates the stiffness of this brac-
ing system.
Compute the initial twist imperfection, θo, using Equation 14.
L
0 = (14)
500ho
( 288 in.)
=
500(17.3 in.)
= 0.033 rad 

Compute the required brace strength, Mbr , using Equation 15.

T o
M br = (15)
(2 T prov )
(10.1 kip-in./rad/in.) ( 0.033 rad )
=
2 (10.1 kip-in./rad/in.) (12.2 kip-in./rad/in.)
= 0.287 kip-in./in. 

Available Brace Strength

The calculation of the available strength will vary depending on the specific situation, but will generally include assessment of
the strength of the decking system, the strength of the connection between the decking system and beam, and the strength of
the beam web.

80 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

Deck Bending Strength
On one side of the beam (left side in Figure 6), the decking system is in negative bending, the concrete is assumed to be cracked,
and thus only the strength of the bare composite steel deck remains. The design strength of 20‑ga. deck that is 3 in. deep is
ϕMn,neg = 1.72 kip-in./in. (Sputo, 2014). On the other side of the beam (right side in Figure 6), the decking system is in positive
bending, and the composite slab can be relied upon. The design strength of 20‑ga. composite steel deck that is 3 in. deep and
supporting 6‑in. total depth normal weight concrete is ϕMn,pos = 5.10 kip-in./in. (Sputo, 2014). These strengths are parallel and
additive; thus, the strength of the decking system is calculated as:
ϕMn = ϕMn,neg + ϕMn,pos
= (1.72 kip-in./in.) + (5.10 kip-in./in.)
= 6.82 kip-in./in.

Connection Strength
The connection between the decking system and the beam is provided by bearing and through the steel headed stud anchor.
The interface between steel beams and composite slab has been studied extensively in the past, but predominantly under shear
loading. Little guidance is available in the literature for the calculation of the twisting moment strength of the connection. For
the purposes of this work, the moment is assumed to be taken by a force couple formed through tension in each steel headed
stud anchor and compression on the beam flange as shown in Figure 8(a). The moment strength per unit length along the beam
is then computed as the product of the strength of the controlling limit state of the force couple and the lever arm [taken as
bf / 3 for the triangular stress distribution shown in Figure 8(a)] divided by the stud spacing. The strength of the force couple is
computed from the limit states of steel headed stud tensile rupture, concrete pullout, concrete breakout, concrete crushing, and
beam flange yielding.

  
(a)  Beam cross-section view (b)  Beam side view

Fig. 8.  Beam-to-deck connection.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 81

Steel Tensile Strength
The strength of the steel headed stud anchor in tension is computed using the provisions of AISC Specification Section I8.3b
(2016). The diameter of the shank of the steel headed stud anchor is 0.75 in. and the ultimate strength, Fu, is 65 ksi.
Rn = Fu As
2
= ( 0.75) ( 65 ksi ) ( 0.75 in.)
4
= 21.5 kips

Concrete Pullout Strength

Pullout of the headed stud anchor is determined using the provisions of ACI 318, Section 17.4.3 (ACI, 2014). For a w‑in.-diameter
steel headed stud anchor, the diameter of the head is 1.25 in.
Compute the bearing area, Abrg, as
2 2
Abrg = d head d shank
4 4
2 2
= (1.25 in.) ( 0.75 in.)
4 4
= 0.785 in.2

Compute the pullout strength as

ϕRn = ϕ8Abrgƒc′
= (0.70)8(0.785 in.2)(3 ksi)
= 13.2 kips

Concrete Breakout Strength

Hawkins and Mitchell (1984) derived an equation for the breakout strength of steel headed stud anchors embedded in a compos-
ite slab with composite steel deck based on the height of the stud, Hs, and rib width at mid-height of the composite steel deck,
wr . The height of the stud is taken as the total depth of the composite slab (6 in.) minus the required cover (0.5 in.) minus the
height of the head of the stud (0.375 in.). Thus, the height of the stud, Hs, is taken as 5.125 in. The rib width at mid-height of the
composite steel deck, wr. is taken as 6 in.
Ac = 2 2Hs wr
= 2 2 ( 5.125 in.) ( 6.0 in.)
= 87.0 in.2
Rn = 4 fc Ac
= ( 0.75) 4 3,000 psi (87.0 in.2 )
⎛ 1 kip ⎞
= 14,300 lb ⎜ ⎟
⎝ 1,000 lb ⎠
= 14.3kips

As an alternative approach, Lawson and Hicks (2011) recommend taking the tensile strength of a headed stud anchor as 85%
of its shear strength. This method yields results in strengths somewhat higher than by evaluating each limit state individually.

Concrete Crushing Strength

The bearing strength of the concrete is computed using the provisions of AISC Specification Section J8 (AISC, 2016).

82 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

Compute the bearing area, A1, as the product of deck rib width (4.5 in.) and half the beam flange width [Figure 8(a)].
A1 = (4.5 in.)(bf / 2)
= (4.5 in.)(6.00 in./ 2)
= 13.5 in.2

Compute the bearing strength as

ϕPP = ϕ0.85ƒc′A1
= (0.65)(0.85)(3 ksi)(13.5 in.2)
= 22.4 kips

Beam Flange Bending Strength

The strength of the beam flange in bending is computed based on the plastic moment strength of a length of the flange equal
to the stud spacing at a distance k1 from the beam centerline and assuming the compressive couple force acts at a distance bf / 3
from the beam centerline.
t 2f s
Mn, flange = Fy
4
2
( 0.425 in.) (12 in.)
= ( 0.90 ) ( 50 ksi )
4
= 24.4 kip-in.
M n, flange
Rn =
( bf 3 k1 )
( 24.4 kip-in.)
=
( 6.00 in 3) ( 0.75 in.)
= 19.5 kips

From the preceding calculations, the controlling limit state for the force within the couple is concrete pullout of the steel headed
stud anchor. Combining that result with the lever arm of the couple and the spacing of the steel headed stud anchors, the control-
ling connection strength is calculated as:
ϕRn ⎛ bf ⎞
ϕMn = ⎜ ⎟
s ⎝ 3⎠
(13.2 kips) ⎛ 6.00 in. ⎞
=
(12 in.) ⎝ 3 ⎠
= 2.20 kip-in. in.

Beam Web Bending Strength

It is expected that the out-of-plane bending demand on the beam web is less than the required strength of the connection; how-
ever, it is unclear how much less. For these calculations, the bending demand of the web is conservatively taken as the required
brace strength. The strength of the beam web bending out of plane is computed as the plastic moment strength:
t w2
Mn = Fy
4
2
( 0.300 in.)
= ( 0.90 ) ( 50 ksi )
4
= 1.01 kip-in. in.
ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 83
 Table 2.  Parametric Study Properties
Property Case A Case B
Decking system stiffness, βprov-b (kip-in./rad/in.) 400 30
Decking system strength, ϕMn (kip-in./in.) 5 1
Connection force couple strength, ϕRn (kips) 10 0.4

Given that the strength of the beam web (ϕMn = 1.01 kip-in./in.) is less than that of either the decking system or the connection,
it is the controlling strength of the brace. Further, this strength is sufficient because it exceeds the required brace strength (Mbr =
0.287 kip‑in./in.).

Example Summary
Having met both the stiffness and strength requirements, the decking system is adequate to brace the beam against constrained-
axis torsional buckling at the required axial strength.
The calculations presented in this example are not intended to cover every possible situation. Engineering judgment is necessary
—especially when computing available braced stiffness and strength—to ensure that rational and reliable load paths exist and
that all relevant sources of flexibility have been accounted for. Cases such as perimeter beams, where a decking system is present
on only one side of the member, or bare composite steel or roof deck with a wide bottom flat, where the connection between the
decking system and beam may be flexible, should be approached with special care.

Parametric Study ⎛ 2ϕMn ⎞

βlimit = min ⎜ β prov , ⎟ (20)
The calculations to check the bracing requirements, such as ⎝ θo + ϕM n β prov ⎠ 
presented in the example, are quite involved and are gen-
erally impractical for typical designs. Noting that, in this The axial strength is calculated iteratively, not directly, as
section, a parametric study is presented with two goals: the maximum axial load for which the beam can be consid-
(1)  to identify conditions that are most beneficial for rely- ered braced against constrained-axis torsional buckling (i.e.,
ing upon the rotational bracing provided by the decking sys- satisfying Equation 19). This axial load, termed Pu,braced, is
tem and (2) to develop rules of thumb such that in specific analogous to the design constrained-axis torsional buckling
typical cases, the bracing provided by the decking system strength and comparable to design strengths for other buck-
can be taken advantage of without the need to perform full ling modes.
calculations. For the parametric study, two generic bracing cases are
Of interest in this study is the calculation of the axial defined. The first case, Case A, is representative of a com-
strength of the beam for a given decking system configura- posite slab similar to what was computed in the example
tion. The relevant parameters of the decking system con- presented in the previous section. The three defining factors
figuration are the provided stiffness, βprov-b, and the moment are the decking system stiffness, βprov-b, the decking system
strength, ϕMn (noting that the decking system, connection, strength, ϕMn, and the connection force couple strength, ϕRn.
or beam web may control strength). The two requirements Each of these values, given in Table 2, is marginally lower
can be stated as Equations 17 and 18. than what was calculated in the example so as to broaden
the range of applicability. The second case, Case B, is repre-
βprov ≥ βT(17) sentative of a steel roof deck (with ribs perpendicular to the
beam). The strengths and stiffnesses are accordingly lower
ϕMn ≥ Mbr(18) than that of the composite slab, including for the connection
Alternatively, noting the relationship between βT and Mbr force couple strength which is controlled, for example, by
in Equation  15, the requirements can be combined as in the uplift strength of a spot weld.
Equations  19 and 20, where the moment requirement has The variation of Pu,braced with length for a W18×35 beam
been converted into a stiffness requirement. is shown in Figure 9. The results in the figure show the two
different cases (Table  2), an additional case of βlimit  = 0,
βlimit ≥ βT(19) and the constrained-axis torsional buckling design strength
computed without accounting for any bracing. It is expected,

84 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

logically, that for βlimit  = 0, the maximum permitted axial In Figures 10(c) and 10(d), Pu,braced is plotted with respect
load will equal the constrained-axis torsional buckling to the design torsional buckling strength. Again, the greater
strength from the code equations (i.e., Pu,braced  = ϕPnca). stiffness and strength of Case  A are demonstrated in the
However, as can be seen in Figure  9, the two values dif- higher strength ratios. A key result shown in Figure 10(c) is
fer. The differences are due to simplifications made in the that for Case A and for all wide flange cross sections weigh-
derivation of the brace stiffness requirements and the fact ing 150 lb/ft or less, the constrained-axis torsional buckling
that the derivation was not intended to be applicable for zero strength is nearly or at least equal to the calculated torsional
brace stiffness. It is expected that for practical values of buckling strength [the ratio Pu,braced to ϕPnx has a minimum
βlimit, the method is accurate. value of 0.975  in Figure  10(c)]. For comparison, the ratio
The parametric study was performed by computing ϕPnca to ϕPnz has a minimum value of 0.371 over the same
Pu,braced for each wide flange shape in the AISC Shapes range (Figure 11). The lowest values in that ratio occur for
Database (AISC, 2017b) with a weight less than or equal to cross sections with lower web slenderness.
150 lb/ft, for both generic bracing configurations (Table 2) In Figures 10(e) and 10(f), Pu,braced is plotted with respect
and for a range of lengths from L/d = 5 to L/d = 50, where to the design major-axis flexural buckling strength. The
d is the section depth. The results of the parametric study second key result is shown in Figure 10(e). For Case A and
are presented in Figure 10. In each of the plots of Figure 10, for all wide flange cross sections weighing 150 lb/ft or less,
the ratio of Pu,braced to a design strength is plotted as a func- the constrained-axis torsional buckling strength is at least
tion of the unbraced length of the beam. Each line represents half of the calculated major-axis flexural buckling strength
one cross section. The lines are shaded based on the cross- [the ratio Pu,braced to ϕPnx has a minimum value of 0.577 in
section web slenderness, h/t w. The plots on the left-hand side Figure 10(e)]. For comparison, the ratio ϕPnca to ϕPnx has a
are for Case A, while the plots on the right hand side are for minimum value of 0.129 over the same range (Figure  12).
Case B. The lowest values in that ratio occur for cross sections with
In Figures 10(a) and 10(b), Pu,braced is plotted with respect higher web slenderness.
to the design constrained-axis torsional buckling strength. For the cases examined, the key results from Figure  10
These plots show most directly the increase in calculated are that the constrained-axis torsional buckling strength is
strength by accounting for the rotational stiffness of the nearly equal to or greater than either the torsional buckling
decking system. With the greater stiffness and strength of strength or half the major-axis flexural buckling strength;
Case  A, the calculated strength increases by as much as these are potentially useful rules of thumb. These results
a factor of 6. Whereas for Case  B, the calculated strength can be conservatively used for cases within the range of the
increases by as much as a factor of 3, but in some cases study—namely, wide flange shapes weighing 150  lb/ft or
decreases, indicating that steel roof deck is not effective in less, decking system properties meeting or exceeding those
providing bracing against constrained-axis torsional buck- listed in Table 2 for Case A, and span lengths between 5 and
ling. For both cases, the largest increases are for cross sec- 50 times the beam depth.
tions with higher web slenderness.

Fig. 9.  Variation of axial strength with length.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 85

 Fig. 10.  Parametric study results.

86 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

Fig. 11.  Comparison of axial strength, Pnca to Pnz .

Fig. 12.  Comparison of axial strength, Pnca to Pnx .

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 87

 CONCLUSIONS Mbr Required brace strength, kip-in.
Accounting only for the lateral restraint provided by steel Mn Nominal moment strength, kip-in.
roof deck or composite floor deck may lead to a situation
Mn,flange Nominal moment strength of flange, kip-in.
where the computed axial strength of a beam is controlled
by constrained-axis torsional buckling. The stiffness and Mn,neg Nominal negative moment strength of the decking
strength requirements developed in this paper allow for the system, kip-in.
inclusion of the rotational restraint provided by steel roof
Mn,pos Nominal positive moment strength of the decking
deck or composite floor deck. The inclusion of this restraint
system, kip-in.
can lead to a significant increase in the calculated axial
strength. The requirements are based on the results of prior Pn Nominal compressive strength, kips
studies, and an example was presented to illustrate their use.
Pnca Nominal compressive strength for the limit state of
A parametric study was performed to identify cases when
constrained-axis torsional buckling, kips
accounting for the rotational restraint was most beneficial
and to develop rules of thumb, with the key result being that Pnx Nominal compressive strength for the limit state of
for wide-flange beams that weigh 150 lb/ft or less, the rota- major-axis flexural buckling, kips
tional restraint provided by typical composite deck to inte-
Pny Nominal compressive strength for the limit state of
rior beams is sufficient to achieve strengths of at least 50%
minor-axis flexural buckling, kips
of the major-axis flexural buckling strength.
*
Pny Minor-axis flexural buckling strength with full
column length, kips
SYMBOLS
Pnz Nominal compressive strength for the limit state of
A Factor applied to ideal torsional brace stiffness to
torsional buckling, kips
control deformations and brace moments
Pp Nominal bearing strength, kips
A1 Loaded area of concrete, in.2
Pu Required axial compressive strength, kips
Abrg Bearing area, in.2
Pu,braced Maximum required axial compressive strength that
Ac Concrete breakout failure area, in.2
can be considered braced, kips
Ae Effective area, in.2
Py Axial yield strength, kips
Ag Gross cross-sectional area, in.2
Q Net reduction factor accounting for all slender
As Cross-sectional area of steel headed stud anchor, compression elements
in.2
Rn Nominal strength, kips
E Modulus of elasticity of steel = 29,000 ksi
a Distance from centroid to brace point, in.
EI Flexural rigidity of the decking system, kip-in.2
bf Width of flange, in.
Fcr Critical stress, ksi
be Reduced effective width of half-flange, in.
Fe Elastic buckling stress, ksi
d Depth of section, in.
Fy Yield stress, ksi
dhead Head diameter of steel headed stud anchor, in.
Fu Tensile strength, ksi
dshank Shank diameter of steel headed stud anchor, in.
G Shear modulus of steel = 11,200 ksi
ƒc′ Specified compressive strength of concrete, ksi
Hs Height of steel headed stud anchor, in.
h Web height, in.
Ix Major-axis moment of inertia, in.4
he Reduced effective web height, in.
Iy Minor-axis moment of inertia, in.4
ho Distance between flange centroids, in.
J Torsional constant, in.4
k1 Distance from web center line to flange toe of
L Beam or deck span, in. fillet, in.
Lcz Effective torsional length, in. nb Number of intermediate braces

88 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

ro Polar radius of gyration, in. REFERENCES
rx Major-axis radius of gyration, in. ACI (2014), Building Code Requirements for Structural
Concrete and Commentary, ACI 318-14, American Con-
ry Minor-axis radius of gyration, in.
crete Institute, Farmington Hills, Mich.
s Spacing of steel headed stud anchors, in. AISC (2010), Specification for Structural Steel Buildings,
t f Thickness of flange, in. ANSI/AISC 360-10, American Institute of Steel Con-
struction, Chicago, Ill.
tw Thickness of web, in.
AISC (2016), Specification for Structural Steel Buildings,
wr Rib width at mid-height of composite steel deck, ANSI/AISC 360-16, American Institute of Steel Con-
in. struction, Chicago, Ill.
x Ratio of the effective area to the gross area AISC (2017a), Steel Construction Manual, American Insti-
tute of Steel Construction, Chicago, Ill.
βlimit Limiting provided brace stiffness including
AISC (2017b), Shapes Database, v15.0, American Institute
consideration of brace strength
of Steel Construction, Chicago, Ill.
βprov Total provided brace stiffness AISC (2018), Seismic Design Manual, American Institute of
βprov-b Provided stiffness of the decking system Steel Construction, Chicago, Ill.
Errera, S.J., and Apparao, T.V.S.R. (1976), “Design of
βsec Web distortional stiffness
I-Shaped Columns with Diaphragm Bracing,” Jour-
βT Required brace stiffness nal of the Structural Division, ASCE, Vol.  102, No.  9,
pp. 1,685–1,701.
βTb Required stiffness of the decking system
Hawkins, N., and Mitchell, D. (1984), “Seismic Response
λf Width-to-thickness ratio for flange of Composite Shear Connections,” Journal of Structural
λrf Limiting width-to-thickness ratio for flange Engineering, ASCE, Vol. 110, No. 9, pp. 2,120–2,136.
Helwig, T., and Yura, J. (1999), “Torsional Bracing of Col-
λw Width-to-thickness ratio for web
umns,” Journal of Structural Engineering, Vol.  125,
λrw Limiting width-to-thickness ratio for web No. 5, pp. 547–555.
ϕ Resistance factor Kaehler, R.C., White, D.W., and Kim, Y.D. (2011), Frame
Design Using Web-Tapered Members, Design Guide 25,
θ0 Initial twist imperfection American Institute of Steel Construction, Chicago, Ill.
τ Stiffness reduction factor Lawson, R.M., and Hicks, S.J. (2011), Design of composite
Beams with Web Openings, SCI Publication P355, Steel
ω Finite brace stiffness factor = 0.90
Construction Institute, Berkshire, United Kingdom.
Liu, D., Davis, B., Arber, L., and Sabelli, R. (2013), “Tor-
sional and Constrained-Axis Flexural-Torsional Buckling
Tables for Steel W-Shapes in Compression,” Engineering
Journal, AISC, Vol. 50, No. 4, pp. 205–247.
Sputo, T. (2014), Floor Deck Design Manual, Steel Deck
Institute, Glenshaw, Pa.
Timoshenko, S.P., and Gere, J.M. (1961), Theory of Elastic
Stability, 2nd Ed., McGraw-Hill, New York, N.Y.

ENGINEERING JOURNAL / FIRST QUARTER / 2020 / 89

90 / ENGINEERING JOURNAL / FIRST QUARTER / 2020

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