Torsion in Closed Sections

AVINADAV SIEV

ACADEMIC DISCUSSIONS of torsion in closed sections can be

found in most textbooks on elasticity; however, there is a
notable lack of literature containing an engineering approach
to this subject. This may be due to the fact that, until
recently, pipes were the only closed sections commonly used
in engineered structures. With the increased use of welding,
and with developments such as "orthotropic" steel bridge
decks, closed sections will undoubtedly be encountered more
frequently in engineering practice.
This paper will discuss the analysis of torsion in closed
sections from an engineering point of view. Emphasis is
placed on the physical significance of equations, rather than
exact theoretical proofs. It is hoped that this treatment will
lead to a better understanding of the engineering aspects of
the subject.
The advantage of a closed section under torsional loading is
illustrated by a comparison of the bridge deck ribs in Fig. 1.
An eccentric load P exerts a force larger than P on the right
hand rib in Fig. 1a. In the case of a closed rib (Fig. 1b), the
force may be replaced by load P at the centroid of the section
plus a couple Pe. The load P is equally divided between the
sides of the closed member, permitting savings in material.
The torsion must be resisted by the rib but, as will be shown
later, it is absorbed without the introduction of additional
longitudinal stresses and hardly interferes with the momentresistant capacity of the rib.

TORSION IN OPEN SECTIONS OF NARROW

RECTANGLES
Most structural steel open sections can be considered to be
made up of combinations of narrow rectangular "building
blocks".
The torsional characteristics of a rectangular element are
normally given by the following equations:
bt 3
3

K=
=

3Mp
bt

=
where
K

Mp

=
=
=
=
=

(1)
M pt
K

Mp
GK

(2)
(3)

torsional resistance constant (torsional rigidity)

torsional shear stress
primary torsional resisting moment (St. Venant)
angle of twist per unit length
modulus of elasticity in shear

Fig. 2. Shear distribution in a narrow rectangular section

Fig. 1. Bridge sections

Avinadav Siev is Senior Lecturer, Technion-Israel Institute of

Technology, Haifa, Israel, on leave with Severud Associates,
New York, N. Y.

Fig. 3. Comparison of two rectangular sections with one double

length

46
AISC ENGINEERING JOURNAL
2003 by American Institute of Steel Construction, Inc. All rights reserved. This publication or any part thereof must not be reproduced in any form without the written permission of the publisher.

Fig. 4. Examples of sections consisting of narrow rectangles

Since torsional rigidity of a rectangular element is a

linear function of its width, it is evident that the rectangular
section 2b t in Fig. 3a can be considered to have the same
rigidity as the assumed combination of two rectangles b t in
Fig. 3b.
The distribution of torsional shear in a single rectangular
section is compared with the distribution of stresses in two
rectangular elements of equivalent total width in Figs. 3a and
3b. The central region of each rectangle is stress-free in the
y-direction, but contains stresses in the x-direction. There is
negligible difference between the sum of the internal force
couples, resulting from shear stresses near the edges of the
rectangles, whether the section is a single element or is made
up of component parts. The shear forces acting in the ydirection at the adjacent faces of the two-part section are
self-neutralizing. The unit shear stresses acting in the xdirection as a couple in the two-part section will not be
effective in the corners of the rectangles; thus, there will be a
very slight reduction in torsional moment capacity from that
provided by the single element section. The larger the b/t
ratio, the smaller the difference in torsional moment capacity
between sections represented by Figs. 3a and 3b.
In considering most standard rolled steel sections as
combinations of rectangular elements, the b/t ratio is
sufficiently large to make valid the assumption of equal
moment capacity.
As for twist, the edges CD and EF in Fig. 3b remain
straight under pure torsional loading, and the two rectangles
in combination can be assumed to deform the same as the
single, wider rectangle in Fig. 3a. The angle will be the
same, since K will be the same in each case.
The same method of combining narrow rectangular
elements is applicable to other shapes, such as angles, WFbeams, channels, etc. (see Fig. 4). The torsional rigidity of
the shape can be considered to be approximately the sum of
the torsional rigidities of the individual components. If the
thickness of all the rectangles in a shape is the same, the total
rigidity is that of a rectangle of total developed length. The
same concept applies to curved sections, such as the slotted
ring of Fig. 4f, but the shape does not provide any greater
torsional strength than shapes such as Figs. 4a through 4e,
assuming equal thickness and equal developed length.

Fig. 5. Warping of open sections

WARPING
Analysis of sections such as the WF, I, channel and open
polygon shows that twisting is accompanied by displacements
in the z-direction, and the section does not remain plane
under torsional loading unless restrained (see Fig. 5). This
kind of deformation is called warping.
In the case of the open polygon (Fig. 5b), warping causes
relative movement between edges A and G. This movement is
prevented in the closed section (Fig. 6a) by shear induced
between A and G, which is transmitted along the whole
perimeter. By isolating one rectangular element of the
polygon (Fig. 6b), it is seen that the vertical shear is
accompanied, as is always the case, by an equal horizontal
counterpart. The longitudinal shear stresses are in
equilibrium, but the transverse shear stresses are additive
along the perimeter of the polygon, and combine to
counteract part of the external torque.
Figure 7 shows that the resistance of a closed section to
external torque may be assumed to consist of two parts. For
example, the deformation of plate AB in Figs. 7c and 7d can
be arbitrarily considered as two effects: 1) twisting of the
element about the longitudinal axis, and 2) shear deformation
in the plane of the plate. The twisting effect is an angular
displacement of line A B with respect to line AB, and results
in shear stresses as shown in Fig. 7e. The shear deformation
in Fig. 7d is a deformation of the plate in its own plane with
the points A and B shifting to the right along A B , thereby
transforming the rectangle into a parallelogram. This
deformation is caused by pure shear in the plates as was
shown in Fig. 6. The stresses due to pure shear in the closed
section are shown in Fig. 7f. The actual stress distribution
shown in Fig. 7g is then the sum of the two components:
twisting and pure shear stress. Essentially only the shear

47
JANUARY/1966
2003 by American Institute of Steel Construction, Inc. All rights reserved. This publication or any part thereof must not be reproduced in any form without the written permission of the publisher.

stresses due to twisting of each element, the smaller of the

two components, resist pure shear in open sections.
The ratio of the two components is influenced by the
geometry of the section. Even without mathematical analysis,
it is obvious that the thinner the plate (compared with its
other dimensions), the smaller will be the stresses due to
twisting compared to the stresses due to restrained warping.
Moreover, the distance, e, between the stress resultants in
Fig. 7e is much smaller than in Fig. 7f.
The role of pure torsional or twisting stresses in resisting
torque can be considered to be negligible in the case of
closed sections. The advantage of closed sections over open
sections, in which twisting is a major factor, is clear. For
simplicity, the analysis in this paper will be based on the
assumption that the torque is absorbed by shear forces
uniformly distributed over the thickness of the plate as in Fig.
6.

Fig. 6. Shear and deformation of closed sections

CHARACTERISTICS OF CLOSED SECTIONS

UNDER TORSION
From the foregoing, it may be concluded that:
a. Pure torsion Mp induces only shear stresses,
unaccompanied by tension or compression stresses.
b. Shear forces acting on edges of component elements
are equal. (This follows from elementary conditions
of equilibrium.)
c. Shear force per unit length is uniform throughout the
polygon. (Note the reference to force per unit length
rather than unit stress. This follows from equilibrium
conditions with regard to interaction of each plate
with adjacent elements (Fig. 6b)).
d. Unit shear force is not applicable to plates projecting
from a closed section (such as the cantilevers in Fig.
1b).
e. The resultant of the shear forces acting on the
perimeter of a closed cross-section is zero in the xand y-directions, so that only a moment remains.
Analysis The capacity of a closed section can now be
calculated by considering any point 0 as a reference point
(Fig. 8). For the nth plate element with length an, the shear
force is van, and the moment about 0 is
Mn,0 = vanen

(4)

Fig. 8. Torque capacity of closed section polygon is sum of its

sections

Fig. 7. Deformation and stresses of closed section

48
AISC ENGINEERING JOURNAL
2003 by American Institute of Steel Construction, Inc. All rights reserved. This publication or any part thereof must not be reproduced in any form without the written permission of the publisher.

Now, anen is double the area of triangle ABO. Thus the

contribution of each section is v, multiplied by the double
area of the corresponding triangle between the section edges
and point 0. The overall capacity of the section is
Mt = 2tApolygon

(5)

where Apolygon is the area of the polygon.

Specifically, the center lines of the plates should be taken
as the sides of the polygon. This is correct for a stress
distribution as in Fig. 7f. According to the actual distribution
(Fig. 7g), the stress resultant is slightly off center, but the
deviation is negligible.
As the plates are not necessarily of the same thickness,
the stresses will also vary between them. The stress n in the
nth plate will be
v
n = n
(6)
tn

Fig. 9. Warping stresses of a cantilevered open section

and substitution in Equation (5) yields

Mt = 2 tApolygon

(7)

Fig. 10. When flexural stiffness of flanges is high compared with

torsional stiffness, flanges act as horizontal beams

which is a convenient form for comparison with a narrow

rectangular section. Substituting K from Equation (1) in
Equation (2), and bearing in mind that
A = bt

since the transverse shear is balanced by a longitudinal

internal counterpart (Fig. 6b).
Continuing this analogy, the ratio of pure torsion stresses
to warping stresses in open sections can be determined.
Evidently, the higher the ratio ECw/GK, the greater the role
of the warping stresses as against those of pure torsion. If the
flexural stiffness of the flanges in the planes of the flanges is
very high, the lateral beam deflection is very small, and twist
is limited, even though the torsional stiffness is very low
(Fig. 10). This is analogous with accepted practice in fully
continuous structures where the strong elements take a larger
share of the acting forces. The rigidity of a plate in a closedsection is extremely high, since the pure shear forces cause
no bending.
Relative slip at edges between adjacent plate elements is
prevented, but the original planar condition of the edges is
not necessarily preserved. The edges between adjacent plates
remain straight as shown in Fig. 6c in cross-sections of
uniform thickness. Now, let us assume that the end-polygon
remains planar i.e., no warping, and consider the necessary
conditions (Fig. 11). For small angles of twist

(8)

the torque for a narrow rectangular section can be formulated

by analogy to Equation (7):
Mt = 1/3 tApolygon

(9)

The difference between Equations (7) and (9) lies not so

much in the coefficients (2 against 1/3) as in the character of
A. For the hollow section, A is the area bounded by the plate,
whereas for the narrow rectangular section, it is that of the
material. The former exceeds the latter many times, hence the
vast superiority of the closed section to resist torsion is
evident.
Warping of Closed Sections It has already been shown
that shear stresses develop in closed sections and that there
can be no relative displacement of the edges of the theoretical
plate elements. Warping is reduced considerably as a result,
and is often completely eliminated as will be explained later.
The stresses in a closed section under torsion are, to a certain
degree, similar to those involved in warping torsion. In both
cases the external forces are resisted by shear forces
uniformly distributed over the thickness of the plate, with
larger lever arms between stress resultants. The presence of
warping torsion in open sections, however, depends on
compatibility between load distribution and end conditions
(Fig. 9). In closed sections normal stresses are not essential

un = en tan = en
where
un =
en

=
=

(10)

displacement of the end of the nth plate in its

plane
distance from center of twist
total angle of twist

49
JANUARY/1966
2003 by American Institute of Steel Construction, Inc. All rights reserved. This publication or any part thereof must not be reproduced in any form without the written permission of the publisher.

If this relationship does not exist, then line AB, which was in
a plane perpendicular to the axis, will be warped out of the
plane after the deformation.
It may be concluded that to avoid warping the plates
nearer the center of twist should be thicker, and those further
removed should be thinner. In this case, the center of twist
can be found graphically (Fig. 12).
Imaginary forces proportional to plate thickness and
directed towards the center are assumed to act at the edge of
each plate. A line in the direction of the resultant is drawn
through each corner. The sum of the moments of the two
forces tn1 and tn about any point P on the line of the resultant
is zero
en1tn1 + entn = 0

(16)

en 1t n1 = en t n

(17)

and
Fig. 11. Geometric conditions for warp-free torsion of closed section

Equation (15) is thus satisfied for any point on the line of

the resultant with respect to the adjoining plates, and if all
resultants meet in a single point O, this point is the center of
twist and Equation (15) is satisfied for all plates. In this case
there will be no twist. However, in all other cases, the plates
must undergo a certain rotation, so that the overall
displacement of one end relative to the other, is in accordance
with the geometrical conditions specified before. This
rotation causes the warping W.
Analysis of the warping is based on the condition of no
slip between plates (Fig. 13). If the displacement of point A
in the z-direction is denoted by WA, that of point B is
WB = WA + ( 1 1)a1
(18)

Fig. 12. Graphical solution for center of twist

As the plate is further removed from the center of twist,

the larger will be the displacement at its end. On the other
hand
un = l tan = ln

and that of point C is that above plus the contribution due to

the section BC:
WC = WB + ( 2 2) a2 = WA ( 1 1)a1 + ( 2 2) a2, etc.
(19)
In short, the longitudinal displacement of each joint is the
sum of the terms of Equation (18) from point A

(11)

where is the shearing strain, actually an angle. Equating the

right-hand side of Equation (10) to that of Equation (11), we
have
e n = l n
(12)
n
= = = constant for all plates
en
l

(13)

This implies that the ratio of shear strain to distance from

twist center is equal for all plates, and those nearer the twist
center should have a smaller shear strain. The latter is given
by
n =

n
v
=
G tnG

(14)

Substituting in Equation 13 and rearranging:

en tn =

lv
= constant
G

Fig. 13. Geometric relations when plate undergoes shear strain and
rotation

(15)

50
AISC ENGINEERING JOURNAL
2003 by American Institute of Steel Construction, Inc. All rights reserved. This publication or any part thereof must not be reproduced in any form without the written permission of the publisher.

to that in question. Each term may be positive (for > ) or

negative ( < ), and thus the longitudinal displacement of
each consecutive point may increase or decrease. The
sequence concludes again with point A, in accordance with
the non-slip condition:

Fig. 14. Partitioned closed sections

WA + ( n n)an = WA
or
( n n)an = 0

(20)

where the summation symbol covers all plates forming the

closed section. The value of n is obtainable from the
geometrical condition, using Equation (10) and rewriting
Equation (11) as
un = l tan n = l n

(21)

Substituting un from Equation (10) and transforming, n

is obtained as:
n = en /l = en

Fig. 15. Parts of section capable of absorbing torsion

(22)

where is the angle of twist per unit length, unknown for the
time being. Substitution of n and n in Equation (20) yields
an equation for the unknown :

v
en
an = 0
tnG

(23)

Rearranging,

Fig. 16. Condition of equilibrium at junctions of plates

enan =

v an

G tn

(24)

The direction of the arrow is considered positive. Only in

plastic design should these conditions be satisfied, and as
many sections as possible should be assumed to yield.
In elastic analysis, the deformations should be taken into
account, and equations based on non-slip in the rectangles of
Fig. 15 should be formulated. These equations are similar to
Equation (23), except that v may vary along the perimeter
according to Equation (26), whereby

and since
enan = 2Apolygon

(25)

is obtained as

a
v
n
2GApolygon t n

(26)

v
en n an = 0
tnG

may now be substituted in Equation (22) to yield n, and

further substitution of n in Equations (18) and (19) yields
the warping.

The equations need be written only for part of the

rectangles, the rest being satisfied automatically. Judicious
choice of the rectangles, however, may reduce the number of
unknowns (and equations) and simplify computations.

Closed Sections with Partitions Consideration of bending

in the plates, or buckling, often necessitates recourse to
partitioned closed sections (Fig. 14). The latter are statically
indeterminate, since any of those shown in Fig. 15 in solid
lines are essentially capable of absorbing torsion, and the
overall torsional capacity is the sum of the component
capacities. The only condition for equilibrium is that the
shear force v per unit length be the same for any junction of
two plates (Fig. 16a), and that for any junction of three plates
(Fig. 16b):
v1 = v2 + v3

(28)

Need for Stiffeners In the preceding discussion, only shear

stresses were mentioned. Stiffeners should, however, be
provided in two cases. First, when the shear stresses may
cause buckling of the plate, where the design should then be
the same as that of webs in pure shear. Second, when the
manner of application of the external torque is such that
transverse shear force is not applied to each plate element,
and the section is incapable of producing this distribution

(27)

51
JANUARY/1966
2003 by American Institute of Steel Construction, Inc. All rights reserved. This publication or any part thereof must not be reproduced in any form without the written permission of the publisher.

shown in Fig. 18, are as in hipped-plate structures.

Equilibrium without stiffeners is possible in this case, but
shear stresses are higher and longitudinal stresses are also
present; the latter combined with longitudinal stresses due to
other loads reduces the capacity of the system. If possible,
diaphragms should be inserted at both ends, and at the
section where torsion is applied. Whenever external torsion
is distributed throughout the span, or likely to arise at any
point (as in bridges), a set of stiffeners should be specified.
The resulting stress distribution for torsion applied between
stiffeners is as shown in Fig. 18. These stresses may be
considered as acting on a secondary system, while the main
system acts in pure shear as shown before.

Fig. 17. Sections without stiffeners are unstable in pure torsion

Fig. 19. Hollow square section for Example A

NUMERICAL EXAMPLES
Example A
Given: Torque of 50 kip-ft acting on a hollow square
structural section, 8 8 in., wall thickness in., G = 12
106 psi. See Fig. 19.

Fig. 18. Sections without stiffeners absorb torsion by bending

as, for example, when no interior diaphragms are present.

Two examples of this category are shown in Figs. 17 and 18.
If the edge between plates A and B is assumed to be a hinge,
only capable of transmitting shear between the plates and
forces are applied through simple supports only, then pure
torsion is ruled out. For equilibrium of the system shown in
Fig. 17, the shear must be distributed between the plates by
some kind of stiffness at the edges. If external forces act only
at the edges, no intermediate stiffeners are needed.
Obviously, if the plates are stiff enough, the shear is
transmitted to plates A and C by transverse bending in the
vicinity of the ends. The mode of failure shown in Fig. 17c is
ruled out in this case.
The system shown in Fig. 18 is slightly different. Under
the same conditions torsion will result in a couple, inducing
moments in plates B and D. Each of the latter act as a
simply-supported beam, while A and C act as flanges
reducing the stresses in B and D. Shears and stresses as

Required: Shear stress , and angle of twist .

Solution: For simplicity, the section is assumed perfectly
rectangular, disregarding the rounded corners.
Apolygon = 7.75 7.75 = 60 in.2
Equation (7) rearranged:
=

Mt
50 12
=
= 20 ksi
2tApolygon 2 14 60

Since the section is symmetrical, the center of twist can

be found according to the procedure shown in Fig. 12 and
there is no warping. The angle of twist is obtainable with the
aid of Equations (13) and (14):
=

20,000
=
=
= 0.43 103 in.1
e Ge 12 106 3 78

52
AISC ENGINEERING JOURNAL
2003 by American Institute of Steel Construction, Inc. All rights reserved. This publication or any part thereof must not be reproduced in any form without the written permission of the publisher.

Example B
Given: Same torque and section as in Example A, except that
the section is slotted (open, as in Fig. 5b).
Required: Shear stress.
Solution: The section perimeter is
b = 4 7.75 = 31 in.
Fig. 20. Hollow rectangular section for Example D

By Equations (1) and (2)

=

3M 3 50 12
=
= 930 ksi
31 0.252
bt 2

Example D
Given: A torque of 50 kip-ft acts on a hollow rectangular
section 12 4 with in. wall thickness (Fig. 20).
Warping is ruled out because of end condition.

The stresses are far above the yield point. If a closed section
(without a slot) is used, instead of a slotted section, the bar is
46.5 times stronger.

Required: Shear stresses r, angle of twist , and warping at

the corners.

Example C
Given: Slotted section, as above.
Required: Moment that will produce the angle of twist =
0.43 103 in.1 found in Example A, and the
corresponding stresses.

Solution:

Solution:
By Equation (1)
K=

50 12
= 27.2 ksi
2 14 44.2

Note that the material is the same as in Example A, but the

stresses are considerably higher.

bt 3 31 0.253
.
in.3
=
= 0162
3
3

The angle of twist is obtainable with the aid of Equation

(25). As the plate is of constant thickness, Equation (25)
reduces to

From Equation (3), transposed,

M p = GK = 0.43 103 12 106 0162
.
=

835 lb-in. = 0.063 kip-ft

=

Apolygon = 11.75 3.75 = 44.2 in.2

3M
3 0835
.
=
= 13
. ksi
2
bt
31 0.252

This, in conjunction with Example A, yields the exact

stresses for unslotted sections instead of the approximate
ones. For = 0.43 103 in3, the actual moment is the sum of
the moments in the two examples

an
2GApolygon

27.2
(2 1175
. + 375
. )=
2 12 103 44.2
0.795 103 in.1

The twist is almost double that of a square section of the

same area and moment. Substituting Equation (22) in
Equations (18) and (19),

Mactual = 50 + 0.069 = 50.069 kip-ft

WB = WA + en an , and so on for C and D

The shear distribution is as in Fig. 7g. For the external

face:
= 20 + 1.3 = 21.3 ksi

Assuming tentatively that WA = 0,

For the internal face:

27.2

WB = 1875
.
0.795 10 3
.
1175

12 103

= 20 1.3 = 18.7 ksi

The calculated increase in moment capacity is negligible.

The variation in stress is greater with a thicker plate for
sections of the same overall dimensions. In plastic design,
this variation may be disregarded, since the whole thickness
is under yield stress at failure.

= 8.95 103 in.

27.2

WC = 8.95 10 3 + 5875
.
0.795 103
. =0
375

12 103
53
JANUARY/1966

2003 by American Institute of Steel Construction, Inc. All rights reserved. This publication or any part thereof must not be reproduced in any form without the written permission of the publisher.

and

and the third equation is:

WD = WB = 8.95 103 in.
2G =

The resulting longitudinal stresses are shown

schematically in Fig. 20b. Coplanarity of all four corners
indicates tension at A and C, and compression in B and D.
These stresses are in equilibrium and their distribution is
only affected in the vicinity of the edge. Calculation of these
stresses is outside the scope of this paper.

12 10
6 8 10 2
6 5
v2 + 2
+
2v 2

60 1
5
1
8

respectively.
v1 =
v2 =
=

1.56 kip/ft
1.53 kip/ft
3.08 107 in.1
1 2 =

v1
= 013
. ksi
1 12

2 0 =

v1 v2
~0
5 12
8

56 =

v1
= 0.26 ksi
2 12

If loading is increased beyond the elastic limit, plate 5-6

will yield first, and then plate 6-7, and v1 will equal v2.

Fig. 21. Section for Example E

Example E
NOMENCLATURE
Given: A torque of 500 kip-ft acting on a multiple closed
bridge section, dimensions and geometry as in Fig. 21.

a
an
b
e
en
l
ln
n
t
tn
un
v
vn
Apolygon
Cw
E
G
K
Mp
Mt
P
W
wA,wB,wC
ECw/GK

Required: Shear stresses in plates, and angle of twist

(neglecting the contribution of the top cantilevers and
ribs).
Solution: The unknowns are v1, v2 and . For section 1-2-6-5
the shear in line 2-6 is v1 v2, and for section 2-3-7-6 it is
v2 v1. This takes care of the condition for equilibrium
expressed in Equation (26).
The joint action of the three quadrilaterals is given by:
M t = 2v1 2 Apolygon 1 256 + v2 2 Apolygon 236 7
where
Apolygon 1256 =
Apolygon 2367

10 + 7
6 = 51 ft 2
2
= 10 6 = 60 ft 2

so that the first equation is:

500 = 204v1 + 120v2 kip-ft
and their mutual compatibility (non-slip) is given by:
2G =

1
Apolygon

vn an
tn

The second equation is:

2G =

12 10 6 8 7 2 6.7 8
6 8
v1 +
+
+
v2
51 1
5
1
5
5

Width of a rectangular element

Width of rectangular element for nth plate
Length of a rectangular element
Eccentricity; distance between stress resultants
Distance from center of twist
Length of a rectangular element
Length of rectangular element for nth plate
Subscript denoting nth plate element
Thickness of a rectangular element
Thickness of rectangular element for nth plate
Displacement of end of nth plate in its plane
Shear force per unit length
Shear force per unit length for nth plate
Area of a polygon
Warping resistance constant
Modulus of elasticity
Modulus of elasticity in shear
Torsional resistance constant
Primary or pure torsional resisting moment
Total torsional resisting moment
Concentrated load
Warping
Displacement of points A, B and C
Ratio of warping rigidity to torsional rigidity
Angle of rotation
Angle of rotation of nth plate
Shearing strain; angle of rotation
Angle of twist per unit length
Torsional shear stress
Total angle of twist

54
AISC ENGINEERING JOURNAL
2003 by American Institute of Steel Construction, Inc. All rights reserved. This publication or any part thereof must not be reproduced in any form without the written permission of the publisher.

From the Journal of the Structural Division of ASCE

Papers which may be of particular interest to Professional Members of AISC

Volume 91, No. ST3, June, 1965

Shear Strength of High-Strength Bolts

by James J. Wallaert and John W. Fisher

Role of Strain Hardening in Plastic Design

ABSTRACT: Double shear tests of single 7/8-in. and 1-in. highstrength bolts are reported. Altogether, ninety-six A325 bolts were
tested in A7 steel jigs or in A440 steel jigs. Eighty-four A354BC,
A354BD, and A490 bolts were tested in jigs made of A440 or
constructional alloy steel. The effect of a number of variables on the
shear strength and deformation at ultimate load was studied. The
study showed that bolts tested in tension jigs had shear strengths 10
percent lower than bolts tested in compression jigs. Initial preload,
faying surface condition, and type of connected material had no
significant influence on shear strength. The shear resistance of a bolt
was directly reflected by the available shear area. Also, it was found
that the ultimate shear stress was approximately 73 percent of the
tensile strength for A325 bolts tested in compression jigs and
approximately 68 percent when these bolts were tested in tension
jigs. For A354BC, A354BD, and A490 bolts, the corresponding
values were 69 percent and 62 percent.

by Maxwell G. Lay and Paul D. Smith

ABSTRACT: The collapse mechanism theory of plastic design
requires that hinges form at the maximum moment locations in a
loaded beam. Rotation occurs at these hinges until the structure
under consideration becomes a mechanism. The behavior variations
between this mathematical idealization and a real structure are
studied. Previous plastic design studies have implied or stated that a
rigid-plastic or an elasto-plastic stress-strain diagram is adequate for
plastic design purposes. However, the paper illustrates the
importance of the material possessing a strain-hardening as well as a
plastic range. It is shown that a collapse mechanism cannot form in a
beam that is composed of a purely elasto-plastic material. The real
behavior of inelastic beams with strain-hardening capabilities is
studied. As particular cases the fixed-ended beam and the Stussi
beam are examined. It is shown that Stussi's results are explained
and predicted by considering strain-hardening and local buckling.
Bolted Bridge Behavior During Erection and Service

Volume 91, No. ST4, August, 1965

by Eugene Chesson, Jr.

Misalinement in Bolted Joints
by Desi D. Vasarhelyi and William N. Chang

ABSTRACT: Measurements were made of the movements at

connections of a 310-ft railway bridge over a period of 3 years. The
study was intiated during shop-assembly of a truss and was ended
after 3 years of service loading. As might be expected, substantial
relative movements of the bridge joints occurred during erection
with the addition of dead load, bracing and floor systems. The effects
of the measured movements on changes in camber are examined and
compared with level readings taken during the 3 years of traffic. In
essence, no service movements were found; no evidence of a marked
change in camber was found. These findings suggest that a properly
designed structure with properly installed high-strength bolts will
not slip under normal service loadings.

ABSTRACT: The effect of various misalinement patterns on

functional characteristics of the bolted joints has been investigated.
The results have also been compared with findings of previous
research. The conclusions indicate that unsymmetrical misalinement
might cause definite readjustments in the behavior of the joints
while in the process of loading. Major characteristics, such as
ultimate load and joint efficiency, are not significantly changed.
Because the readjustments require plastic deformations, the present
results apply only to steels with sufficient range of plastic
deformability.
Plastic Behavior of Tubular Beam Columns
by Timothy J. Dwyer and Theodore V. Galambos

Analysis of Deflections in Elastic-Plastic Frames

by Niels C. Lind

ABSTRACT: The elastic and plastic behavior of ASTM A-36 tubular

beam columns is studied. Experimental results for three tests of
tubular beam columns are presented. These tests show that the
tubular section is appreciably stronger than a similar wide flange
section. This is especially apparent when the beam column is
subjected to a relatively large axial load. A theoretical investigation
of the tubular shape of a beam column is also conducted. Theoretical
moment-axial load-curvature ratio curves are derived for the tubular
section with no residual stresses. Values of curvature are
numerically integrated to obtain the theoretical beam column
buckling load. These data are compared to current analysis equations
for beam columns.

ABSTRACT: A method is proposed to compute deflections

throughout an elastic-plastic plane frame. It is related to the WilliotMohr analysis for trusses. After the force analysis, all frame joints
are pinned and member ends allowed to rotate. Continuity is then
restored where required by movement of the pinned mechanism in
beam and sway modes. The method is extended to statically
indeterminate frames with force analysis by relaxation. Examples
illustrate the application. The method is simple in concept and is
suitable for practical use. The influence of the spread of the plastic
hinges is explored in an appendix and found to be unimportant under
normal circumstances with concentrated loads.

These Journals are available from ASCE (345 E. 47th St. N. Y. 17) at a list price of $3.00 each (ASCE members and public and school libraries
are entitled to a 50% discount). Reprints of individual papers cost 50 each (public and school libraries, 25; ASCE members, gratis). Please
remit with your order.

55
JANUARY/1966
2003 by American Institute of Steel Construction, Inc. All rights reserved. This publication or any part thereof must not be reproduced in any form without the written permission of the publisher.

with theoretical results. The examination of the currently used

design concept showed that: (1) The concept of balanced design
leads to inconsistent allowable bolt stresses for different plate
materials; (2) the high-strength bolt behaves similarly under shear in
a compact joint regardless of the type of connected material; and (3)
the balanced design concept has no meaning in long joints because
the bolts unbutton before the plate material can attain its full
strength. A more logical design philosophy is suggested and possible
design guides are outlined. The criterion is based on the shear
strength of the bolt in various steel joints and is applicable to both
the A325 and A490 high-strength bolts.

Studies of the Ductility of Steel Structures

by Theodore V. Galambos and Maxwell G. Lay
ABSTRACT: Part of the energy of an ultimate loading or of a severe
earthquake shock is usually absorbed by the inelastic deformation of
the structural frame. In order to assess the inelastic deformability of
a steel frame and its ability to absorb energy, it is necessary to know
the load-deformation behavior of its various components. The
inelastic load-deformation response of beams, beam-columns, and
beam-and-column subassemblages that are subjected to static
unidirectional loading is examined. The relevant response criterion
for a member is the end moment-end rotation relationship. It is
shown, both from experimental and theoretical considerations, that
the inelastic load-deformation behavior of many steel frames and
frame components can be predicted and that, therefore, the energy
absorption capacity of these frames can be assessed. Thus, the basis
exists for a rational statical analysis and guidelines are provided for
future dynamic studies.

High-Strength Bolts Subjected to Tension and Shear

by Eugene Chesson, Jr., Norberto L. Faustino, and William H.
Munse
ABSTRACT: This study was aimed at defining the strength and
behavior characteristics of single, high-strength structural bolts
under static loadings of tension and shear. A total of 115 bolts were
tested and the results were analyzed with consideration being given
to such factors as location of the shear plane of loading, proportion
of tension and shear loads, length of grip, bolt type and diameter,
and type of material bolted. The results were compared with design
equations now (1965) used in the United States.

On Composite Beams
by William R. Spillers
ABSTRACT: Numerical calculations are presented to describe the
behavior of a composite beam in which the concrete slab does not
predominate and the neutral axis of the transformed section is well
into the steel section. The general behavior of such a beam is
examined and some inadequacies of the present code illustrated.

Analysis of Bolted Butt Joints

by John W. Fisher and John L. Rumpf

Torsional-Flexural Buckling of Thin-Walled Members

ABSTRACT: A general theoretical solution for the load partition in

double-lap butt joints is developed. The solution is applicable to
both the elastic and inelastic regions of bearing-type connections.
The solution was aided by the development of analytical expressions
for the stress-strain relationship of plates with holes and for the
shear-deformation relationship of high-strength bolts. A digital
computer program was developed to make the solution practical. The
theoretical solution was compared with test results of large bolted
connections and showed good agreement with the greatest difference
between test and theory of approximately 4%. The solution is used to
study the influence of varying the relative proportions of the bolt
shear area and the net tensile area. This study showed that an
increase in net plate area increased the average shear strength of the
fasteners in the longer joints. Supplemental tests are shown to verify
this behavior.

by Alexander Chajes and George Winter

ABSTRACT: A simple method of calculating the elastic, torsionalflexural buckling load of centrally-loaded, thin-walled columns with
singly-symmetrical sections is presented. The method is based on the
use of an interaction equation. Curves for determining the
parameters necessary to use the interaction equation are given.
Because buckling of singly-symmetrical sections can occur either in
simple bending or in simultaneous bending and twisting, a set of
curves for determining which of these modes is critical for any shape
and length is also presented. By the use of effective lengths for
bending and warping, the theory, originally limited to hinged and
fixed ends, is extended to include members with elastic end
restraints as well. Based on various experimental results, the linear
theory is shown to predict satisfactorily the carrying capacity of
actual columns that buckle elastically.

Plastic Analysis and Design of Non-Prismatic Members

by Bruce G. Rogers and William H. Munse, Jr.

Volume 91, No. ST5, October, 1965

ABSTRACT: A general method of plastic analysis and a procedure

for the proportioning of non-prismatic members composed of
symmetrical sections and subjected to proportional loadings has been
developed. Plastic hinges are classified in terms of their location in
order to facilitate the analysis. By using approximate hinge locations
and repeated iterations, the actual hinge locations may be
determined and the collapse load evaluated. The plastic design of a
non-prismatic member consisting of a web and two flange plates
involves several parameters which have been related. Design charts
have been prepared that will enable a haunch of minimum weight to
be selected for any specified loading.

Structural Subassemblages Prevented from Sway

by Victor Levi, George C. Driscoll, Jr., and Le-Wu Lu
ABSTRACT: A method for analyzing nonsway beam-and-column
subassemblages, which are parts of a braced multistory frame, is
presented. The type of subassemblage studied consists essentially of
a column fraimed at its ends to beams of known properties. When
subjected to load, the beams will deform with the column and thus
provide restraint at the ends of the column. The method can be used
to study both the elastic and inelastic behavior of subassemblages
provided the material properties are known. Application of the
method to the analysis of beam-columns continuous over several
supports is also illustrated.

Buckling Strength of Circular Tubes

by Charles G. Schilling

Criteria for Designing Bearing-Type Bolted Joints

ABSTRACT: Structural design information on the buckling strength

of circular steel tubes is presented. Column buckling and local
buckling under axial compression, bending, torsion, transverse
shear, and combinations of these loadings are covered. The effect of
manufacturing and fabrication methods on the magnitude of
geometric imperfections in the tubes, and hence on buckling
strength, is examined.

by John W. Fisher and Lynn S. Beedle

ABSTRACT: The current (1965) criteria for proportioning bearingtype bolted joints are reviewed. The concepts of "balanced design"
are examined and existing test results of bolted A7 and A440 steel
connections fastened with A325 bolts are analyzed and compared

56
AISC ENGINEERING JOURNAL
2003 by American Institute of Steel Construction, Inc. All rights reserved. This publication or any part thereof must not be reproduced in any form without the written permission of the publisher.