Structural Beams in Torsion Trans. ASCE Vol. 101 (1936)

Lehigh University
Lehigh Preserve
Fritz Laboratory Reports
Civil and Environmental Engineering
1936
Structural beams in torsion, Trans. ASCE, Vol. 101

(1936) p. 857, Reprint No. 8 (36-1)
I. Lyse
B. G. Johnston
Follow this and additional works at: http://preserve.lehigh.edu/engr-civil-environmental-fritz-labreports

Recommended Citation
Lyse, I. and Johnston, B. G., "Structural beams in torsion, Trans. ASCE, Vol. 101 (1936) p. 857, Reprint No. 8 (36-1)" (1936). Fritz
Laboratory Reports. Paper 1154.
http://preserve.lehigh.edu/engr-civil-environmental-fritz-lab-reports/1154
This Technical Report is brought to you for free and open access by the Civil and Environmental Engineering at Lehigh Preserve. It has been accepted
for inclusion in Fritz Laboratory Reports by an authorized administrator of Lehigh Preserve. For more information, please contact
preserve@lehigh.edu.
Piea<;e.
Ve.:1uvv\ +0 ~ 16 1 4
fRITZ ENGrNEfRINC' lAnO .
L~HIGH UNIVERSity
.~
BETHLEHEM. PENNSyLVANIA
..
"LEHIGH UNIVERSITY PUBLICATION

VoI.IX
October, 1935
No. 10
;',:
.'
THE INSTITUTE OF RESEARCH

Circular No. 113
Science and Technology, No. 94
STRUCTURAL BEAMS IN TORSION

By
INGE LYSE, n.s. in C.E.
nRUCE G. JOHNSTON, M.S. in C.E.
LEHIGH UNIVERSITY
BETHLEHEM, PENNSYLVANIA
Price 20 cents
l
G1 0/
r
I~j
LEHIGH UNIVERSITY PUBLICATION
Pul~1is~~d
monthly duriug the calcndar year by Lehigh Univt\rl!iity, Ikt.b,lehem. Pennsylvania. Entered as sccond-class matter ltIarch 24; 1927,
at the Post Office at Bethlehem, PennsYlvania, under the
Act of August 24, 1912.
Circular
No.
Title
1. Organization of the Institute of Research of Lehigh University
2. Microscopical Studies of Anthracite.
3. Rate of Molecular Weight Increase in the Boiling of Linseed Oil.
4. A Preliminary Study of Magnesium-Base Alloys.
6. Absorption of Carbon Dioxide by Coal.
6. Studies in the Boiling of Linseed and China Wood Oils. 26 cents.
7. The Shakespeare Folios and the Forgeries of Shakespeare's Handwriting in the Lucy Packer Linderman Memorial Library at Lehigh
University. 26 cents.
8. Mazzini and Dante. 20 cents.
9. The Center, Function and Structure of Psychology. 16 cents.
10. Methods for Exciting and for Calibrating Tuning Forks. 16 cents.
11. Flat Luminous ~'lames. 10 cents.
12. The Pressure Vaccination Technic. 10 cents.
13. Studies in Drying Oils. 10 cents.
14. The Variant Issues of Shakespeare's Second Folio and Milton's First
Published English Poems. 26 cents.
16. The Hydrates of Calcium Nitrate. 10 cents.
16. Tower Absorption Coefficients. 10 cents.
17. The Testing of Audio-Frequency Transformer-Coupled Amplifiers.
26 cents.
18. Neurovaccine. 10 cents.
19. The Action of Bromine on Unsaturated Fatty Acids. 10 cents.
20. Equal-Slope Surfaces and Helices by Vectors. 10 cents.
21. Coal Conductivity Cell. 10 cents.
22. Volume Changes During Hydration of Gelatin, Cement and Plaster
of Paris. 10 cents.
24. Mononitro- and l>initrothlophenes. 10 cents.
26. Studies in Flame Propagation. 10 cents.
26. The Biochemistry of the Soaking of Hides. (Parts 1 and 2.) 20 cents.
27. Petroleum Wash-Oil Thickening in the Scrubbing of Coke-Oven
Gas. 10 cents.
28. Government. A Phase of Social Organization. 60 cents.
29. Aspects of Slavery and Expansion. 60 cents.
30. Lipid Distribution in Normal and Abnormal Liver Tissues. 10 cents.
32. Scientific Papers from Department of Physics. 10 cents.
33. Scientific Papers from Department of Mathematics. 26 cents.
34. Effect of Nitrate Oxygen upon Tannery Effluent. 10 cents.
36. Constitution and Nature of Pennsylvania Anthracite. 10 cents.
36. Scientific Papers from Department of Physics, 1929. 10 cents.
37. Studies in Drying Olls, 10 cents.
38. The Biochemistry of Soaking of Hides. 20 cents.
39. Papers from the Department of Metallurgy. 26 cents.
40. Scientific Papers from Department of Mathematics for 1929. 25 cents.
41. Floor Test in the George Mason Hotel, Alexandria, Va. 15 cents.
42. Investigation of Welded Connections between Beams and Columns.
16 cents.
43. Forms of Generalization and Their Causes. 10 cents.
44. Thermal Treatment of Natural Gas. 10 cents.
46. A Proof that the Induction Motor Circle Diagram Applies to the
Transmission Line. 10 cents.
46. Studies in the Drying Oils. 10 cents.
47. The Biochemistry of the Soaking and Liming of Hides. 10 cents.
48. The Hydration of Animal Skin by the Volume Change Method. 10
cents.
49. Studies of Some Properties of Gelatin. 10 cents.
60. Rate of Oxidation of Linseed Oil at 160 C. 10 cents.
61. Shakespeare Allusions and Parrallels. 60 cents.
62. Compressive Strength of Concrete in Flexure as Determined from
Tests of Reinforced Beams. 20 cents.
63. Energy of the Iron Arc. 10 cents.
54. The Relative Merits of Some Different Alloy Steels with Respect t.o
Certain Mechanical Properties, 10 cents.
66. The United States Indian Policy in Texas. 10 cents.
56. Logic as the Cross-Classification and Selection of Arbitrary Elements. 10 cents.
57. Studies in the Drying Olls. 10 cents.
58. Heats of Wetting and Adsorption on Zinc Oxide. 10 cents.
59. Some Experiments on the Soaking of Silks. 10 cents.
60. The Stab1Uzation of Blue Cupric Hydroxide. 10 cents.
(List continued on inside back cover)
AMERICAN SOCIETY OF CIVIL ENGINEERS

'Founded November 5. 1852
PAPERS
...
By INGE LYSE. 1 M. AM. Soc. C. E.,
AND BRUCE G. JOHNSTON,2 JUN. AM. SOC. C. E.
SYNOPSIS
Results of a study of the torsional properties of standard structural steel

beams are offered for discussion in this paper. The purpose of the investigation was to furnish a reliable basis for the design of structural members
subjectad to torsional loads. The relation between torque and stress on the
one hand, and between torque and twist on the other, for any piece subjected
to torsion involves a constant the value of which is a function of the
material and the shape of the cross-section. An accuratc method is given f~r
the evaluation of this torsion constant, K, for standard H-sections and I-sections, taking full account of all factors involved. This has been made possible by applying the "membran~' apalogy" to about sixty sections of widely
..varying flange, web, and fillet proportions.
The investigation included a study of the effect of end fixity in torsional
design, and shows how it may be obtained effectively. The proposed formulas
are applied to practical design problems, and are checked by torsional tests
on structural steel sections ranging in size from a 3-in. I-beam weighing 7.5
lb per ft, to a 12 by 12-in. beam weighing 190 Ib per ft.
HISTORICAL FOREWORD
The problem of pure torsion as applied to non-circular sections was first

treated correctly by Saint Venant (1)3 in 1855, and his general solution is
applicable to' any cross-section. In 1903, Prandtl (2) showed that if a thin
membrane were stretched across a hole having the shape of the cross-section
in question and distorted slightly, the equation of its surface had the same
form as the general differential equation involved in the torsion problem.
NOTE.-Discussion on this paper will be closed in August, 1935, Proceeding8.
Research Associate Prof. of Eng. Materials, Lehigh Vniv., Bethlehem, Pa.
Instr. in Civ. En!:., Columbia Vniv., Ncw York, N. Y.
(Formerly Lawrence
Calvin Brink Research Fellow in Civil Engineering, Lehigh Vniv. Bethlehem, Pa., in
Immediate Charge of. Torsion Investigation.)
The numbers in parentheses refer to references given in Appendix 1.
I
I . ._"
-470
Papers
Prandtl showed that by measuring the volume and slopes of the displaced
membrane a direct measurement of the torsional rigidity and stress was
obtainable. Prandtl's analogy, with a thin soap film as a membrane, was used
in several torsion investigations, first in England by Griffith and Taylor (3)
who studied the torsional strength of aeroplane sections in 1917, and, later,
in the United States by Trayer and March (4), who, in 1930, made similar
studies for the same purpose.
Important contributions to the torsion problem have been made by
Timoshenko (5). He has "hortened the pure torsional theory by slight
modifications of Saint Venant'a equations and by mathematical application
of the principles of the membrane analogy. He was also among the first to
consider the effect produced by preventing the warping of a cross-section.
This problem has had the attention of numerous investigators in connection
with problems of elastic stability and bu~kling during bending. Sonntag (7)
treatea the theoretical aspects of this problem in an article published in 1929.
INTRODUCTION
The investigation reported herein was undertaken as a study of all

available information on the subject, both theoretical and experimental,
supplemented by a considerable number of actual torsion tests of structural
steel beams and soap film experiments on various cross-sectional shapes.
The writers first considered testing beam sections 3 'ft in length, welded to
thick plates at the ends. A study of the problem, however, showed that
such beams would be sevEiral times stronger than if they were tested free- ,
ended, and that, unless the exact percentage of end fixity was known, it
would not be possible to draw definite conclusions from such tests.
In order to study the effect of end fixity directly, tests were first made on
eight sections of a 3-in. I-beam (7.5 lb per ft), varying from 3 in. to 4 ft
6 in. in length and cut from the same rolled section. The ends of each piece
were welded to plates 1 in. thick, and the specimens were tested in a standard
26000 in-lb torsion machine.
The results of these tests pointed the way to a revised genera] program, and
a torsion testing rig capable of applying torsional load up to 750000 in-lb was
designed. Provision was made for testing beams either fixed or free at the
ends, and with lengths of 1 ft 6 in., 3 ft, or 6 ft. Nineteen different tests
were made, twelve :on beams with the ends fixed by welding the side and
end plates to form a box section at the ends, and seven with ends free. The
beams ranged in sizes from the 3-in. I-beam, weighing 7.5 Ib per ft, to a
12 by 12-in. beam weighing 190 Ib per ft. Tensile and shearing properties
of the material in each type :of beam' were obtained by standard tensile tests,
round bar torsion tests, and slotted plate shear tests. Soap film' experiments
- on fifty-seven differently proportioned sections were made for the determination of the torsional rigidity.
This paper contains the final summary of all phases of the investigation.
Use has been freely made of the findings of previous investigators, for which
acknowledgment is made at appropriate points.
....
April, 1935
471
Notation.-The symbols in this paper are introduced in the text as they

occur,' and 'are summarized for reference in Appendix II.
THE TORSION THEORY
General Problem.-The solution of the torsional properties of a section

of any shape consists primarily in determining the distribution of lateral
shearing stresses over the cross-section. The shear components will be of
uneven distribution, except in the case of ,the circular section, and as a
result plane sections will be warped during twisting as shown in Fig. 1.
(Note that, in Fig. l(e), the web and each flange are warped as individual
rectangles, in addition to the warping of the section' as a whole.)
No Warping
I~'- - - 3 " - - - ' I - t .
- - - - - - - - 11;.,1"
r-I
la) CIRCULAR
SECTION
(b)
RECTAN'GULAR
'SECTION
(e) STRUCTURAL BEAM SECTION

FIG, I.-TWISTING
OF
BARS
VARIOUS CROSS-SECTIONS
OF
FIG. 2,-':'SECTIONS HAVING ApPROXIMATELY EQUAL

TORSIONAL RIGIDITY
It is assumed that the lateral displacements are proportional to the angular

twist and to the distance from the twisting axis (as is' the case in a circular
section). The longitudinal displacements cause the warping, and the resulting distribution of shearing stress is taken care of by introducing a "stressfunction," F, of x and y. This function must satisfy the diffe.rential equ~tion:
?:J2F
. , .. , ..........
....
]4
?:J2F
---1- ?:Jx2 .. ?:J'y2
== -,
G8
(1)
in which, G
the shearing modulus of elasticity, and ()
angle of twist, in
radians per inch. It may be show~ that the function, F, must be a constant. along the boundary of the section for solid ba~s, and, therefore, may
be chosen arbitrarily. as, equal to zero.
If the bound.ary conditions are such that Equation (1) may be solved
and the value of F determined, it is' possible to' eval~ate the torsion constant of the section and find the stress at any point in the cross-section.
Formulas for the torsion' constant and critical shearing stresses have been
derived in this manner for such sections as th'e -square, rectangle, eilipse;
equilateral triangle, and sector of a circle (1).
472
Papers
In the case of the circular shaft the shearing stress components have a
uniform distribution along each radius, and since the longitudinal shear is
likewise evenly distributed, there is no longitudinal warping of first-order
importance. The ",ell-known simple theory using the polar moment of
inertia is thus applicable to the case of the circular section.
If, in some way, warping which takes place in non-circular sections is
restrained or prevented, longitudinal fiber stresses will be introduced and" the
beam stiffened and strengthened.
The Membrane Analogy.-Equation (1) may be solved mechanically for
any cross-section by means of Prandtl's membrane analogy, thereby overcoming the mathematical limitations of the theoretical derivation.
In the application of this analogy, a soap film is stretched across an
opening having the same shape as the structural section under consideration.
The bubble is distended slightl~ by a variation in pressure. Prandtl showed
that the following relations obtain for this bubble: (1) The torsion constant,
K, is proportional to the total volume of the displaced bubble; (2) the shearing stress at any point is proportional to the maxirimm slope of .the
film at that point; and (3) the contour lines on the bubble give the direction
of maximum shearing stress.
The analogy is also useful as an aid in visualizing the rigidity and stress
distribution in various sections, and makes evident why the four sections
shown in Fig. 2 have approximately equal rigidities in pure torsion, since
the volumes of the. various soap bubbles are approximately the same in each
case. This would not be the case if the ends of the beams were restrained.
EVALUATION OF THE TORSION CONSTANT
Definition.-The torsion constant, K, is the measure of the torsional rigidity and twisting deflections. It is also a part of any formula for torsional
shearing stresses, and may be determined from test results by observing the
ratio of torsional moment to unit twist, in radians per inch, at any place
below the yield point of the beam, and dividing this ratio by the shearing
modulus of elasticity.
The Relation Between K and J.-When a torsional couple, T, is applied
to a circular shaft of radius, r, the maXImum shearing stress, 'r, at the surface, is given by:
Tr
T =
(2)
In terms of T Equation (2) may
in which, J
polar moment of inertia.
be te-arranged to read:
"~T
T= -
The torque, T, may also be expressed ip. terms of

T
JG
(3)
and G, thus:
(4)
A detailed description of the soap film studies is given in a thesis by Bruce G.

Johnston, Jun. Am. Soc. C. E., presented to Lehigh University in partial fulfillment of
the requirements for the degree of Master of Science..
Aprit 1935
473
For non-circular sections the torsional resisting moment may again be

expressed in terms of () and G, with the substitution of K, the torsion
constant, in place of J, thus:
'1' = KG 8
(5)
The torsion constant, K, is equal to the polar moment of inertia for circular
sections. Although for non-circular sections it is always less than the polar
moment of inertia, there is no direct relation between the two factors.
The Rectangle.-In dealing with structural shapes, two principal types
of section require, consideration, the rectangle, and the rectangle modified
by sloping sides, as in the flange of a standard I-beam. In the case of the
rectangle an accurate formula was derived originally by Saint Venant (1) :
l
K = -n b - 2 V n'
,
(6)
3
.
in which n
the breadth of a rectangular section; b
rectangular section; and V
.!!- >
praotically constant for
= the
length of a
a factor depending upon the ratiO,.!, but
3.
Fig. 3 shows the values of V for .! from

n
n'
1 to 3. For.! -ratios greater than 3, V

0.105, and for.! greater than 4,
n
n
V == 0.10504. Equation (6) finds a direct, qualitative interpretation in the
0.105
. 0.103
./
V-
r I
LW
~/
IOl
~ 0.099
0.097
0.095
1.00
J-r
~b----i-l
(b)
1.80
1.40
2.20
2.60
3.00
Side Ratio"*
FIG.
3. -
END
SECTIONS, WITH
CONSTANTS
<
3.
FOR
PART OF A SECTOR
,Wi
+-~
.-~
(e)
RECTANGULAR
(SEE EQDATION
TRAPEZOID
=cC
I
I~
~b---1
'C
(u)
SLOPING FLANGE SECTION
FIG. 4.
(6)).
soap film analogy. It is evident that for long rectangular sections the bubble
will be of constant cross-section along the central part, but at the two ends
it will be contracted and brought down to meet the small side. The
474
Papers
quantity, - 2 V n', then represents the "end loss," which for long sections
is evidently a function of n only.
It also follows that if the ends were made discontinuous, as if they were
parts of infinitely long rectangles, one might state withol}t error:
= l
= 1..
3 n
n3 b
3
and for any differential length, dx, along the section:
B
(7)
dx
..
(8)
The Section with Sloping Sides.-Equation (8) provides a basis for

evaluating K for the sloping flange section. Considering the section shown
in Fig. 4(a), let the thickness at any point be taken as r. Then, if the ends
are assumed discontinuous:
K
= -1
3
Jb
dx
(9)
Evaluating r in terms of m and n, and integrating:

K =
.!
12
(m
+ n) (m' + n')
(10)
which m = major flange thickness and n

minor flange thickness.
A deduction must be made for end effects, as in the case of the simple
rectangle, thus:
III
.!
K =
12
(m
+ n) (m' + n') -
V L m' - V s n'
(11)
in which V L and V s are the end constants, V, for the large end and the
small end, respectively, of the flange (see Fig. 5). The evaluation of these
two constants was the work of Professor J. B. Heynolds, through an analysis
of a section having the shape shown in Fig. 4(b) :5
V L = 0.10504 - 0.10000 S
0.08480 S' - 0.06746 S3
0.05153 S' . . (12)
and,
= 0.10504 + 0.10000 S + 0.084808' + 0.06746 sa + 0.05153 S' . . (13)

in which S = the total slope of the section; that is, m - n .
.
b
VB
Torsion Constant for H-Beams and I-Beams.-The foregoing supplies a

basis for evaluating the K-values of the component parts. Taking the section
shown in Fig. 4(c) as a basis for the sloping flange section, the sum of two
trapezoids and a small rectangular part is expressed by:
Kr
5
= b-
12
w (m
+ n) (m' + n') + l
.
w m3
2 V s n'
"Theory of Elasticity," by A. E. H. Love, Fourth Edition, p. 319.
(14)
I
- .. 1
April, 1935
.475
STRUOTURAL BEAMS IN TORSION
in which Kt
the K-value for the flange. The web IS considered as a
discontinuous section between the flanges, giving:
1
KID = ~ (d - 2 m) WI ., , , , (15)
3
in which KID
the K-value for the web; d
total depth of beam; and,
w
thickness of web (see Fig. 6). There still remains the evaluation of
the added rigidity due to the connection of the flange and web and also due
to the fillet at this point. It is evident that these will cause a considerable
"hump" in the soap bubble.
0.13
0.12
~
~a\\
:;'
~ 0.11
i~
0.10
<
'15''0
.......
...............
~
nd)
_.'
0.09
0.08
8
Pe
f.C
I---r--..
12
e n tag e
S I ci
--.
16
P
20
FIG. 5.-END CONSTANTS FOR K-VALUES OF FLANGES WITH

SLOPING SIDES.
f
'"
h
(b)
'"
H.BEAM SECTIONt
FIG. 6.
Trayer and March (4) in an investigation of aircraft strut sections

assumed this addition to the torsion constant to be proportional to the
fourth power of the diameter of the largest circle that can be inscribed at
the juncture of the web and flange (see Fig. 6), giving, as an additional
K-value:
...
in which D
= the
Gt
D'
diameter of an inscribed circle; and
,
Gt
,.,
=a
(16)
factor that
depends on two ratios, wand 2:...

m
Values of .(J. for sections with parallel-sided flanges, and for sections with
flange slopes of 1 on 6, are given in Fig. 7. These curves were determined
experimentally as the result of soap film tests which will be described subsequently. It will be noted in Fig, 'lea) that for parts of the curves to the
476
Papers
right of a-a, the lines are parallel and uniformly spaced. All standard
rolled beams are in this area, in which case, for parallel flange sides:
ex.
= 0.094
+ 0.070
.!.
(17)
O.IS
0.14
.JJ
o~
!'---
"-..
~ ~ V-- i'--.__
./
~
V
0.10
0.06
........
/V
/
/
/%
./
:/
"'"
...---
~
To Right of a-a o:0.094+0.070W
'vaIU"! of
~8
(a) PARALLEL,SIDED FLANGE
~
0
.:: 0.16 r---,----,----,..---,---,-----;r-----,r---,
'"
> 0.14
f---r--=-+---t--+--f=:::::=t--::::::=j::;;;;==1
0.02 ~"'-:=-:IJ1lj..c=-t---l----f-----
o <:----:-l---:"---1.:::--....L..--L---L--..L---,I
~
Values of
FIG.
7.-VALUES
OF
a.
FOR
i'i (Above) and iii: (Selow)
RIGHT-ANGLE
AND WEB.
JUNCTION
OF
I?LANGE
For flange sections with side slopes of 1 on 20 and 1 on 50 the following formulas represent an interpolation between the curves in Fig. 7:
For a slope of 1 on 20:
ex.
= 0.066 + 0.021.2. + 0.072.!.

m
(18)
and, for a slope of 1 on 50:

ex.
= 0.084 + 0.007 w + 0.0712:..

m
; .(19)
The various elements entering into the total K-values, can now be summarized as follows (refer to Fig. 6):
For sloping flange sections,
K
=b-
(m
+ n)
(m 2
+ n + ..!.
2
(d - 2 m) w
+ 2 ex. n' -
4 V s n' . . (20)
.-
April, 1935
477
Values of V L and VB for standard slopes are, as follows:
VL
VB
-1
6
0.09045
0.12441
1-
0.10026
,"
"
0.11026
20
..!-
0.10307
0.10707.
50
_1
"
0.10504
" .,
0.10504
00
and, for sections with parallel-sided flanges,
K = ~ bn3 + ..!- (d - 2 n)
3
3
SECTIONS
(a)
(b)
'0
(e)
(d)
(e)
RIGIDITY
STRENGTH
100.0
100.0
637.0
332.0
5.5
18.0
70.0
62.0
88.0
74.0
(j)
(g)
Nearly Exact
(h)
:E=w=~n
Nearly Exact
341.0
280.0
(Approx,)
(Approx.)
9.9
11.6
w+ 2
IX
D4
0.42016 n4
(21)
22.2
(Approx.)
22.8
(Approx)
6K'
.2
(,)
INFixed Ends
1'6"
78.1
38.3
(Approx.)
(Approx.)
FIG.
B.-ToRSIONAL RIGIDITY AND
'STRENGTH
OF
DIFFERENT
SEC'TIONS OF EQUAL CROSS-SECTION
AREAS.
FIG.
9.-lrORMULAS FOR STRESS CONCENTRATION IN FILLETS DUEl TO TORSION.

I
Diameter D may be determined by a large scale layout, or by the following

formulas:
For parallel-sided flange sections,
(n+r)2+ w (r
D=
2r
+n
+ :)
..'
(22)
478
STRUOTURAL BEAMS IN .TORSION
Eor sloping-sided flange sections,

.
(B
.D=
in which z
+ Z)'2 + w (r + ~)
B + r + z ....
,
.. ,
(23)
the maximum flange depth shown in Fig. 6, and,

B
r8
[~.!:.- + 1 - 1 - ~]
(24)
82
2r
Oomparative Efficienc'ies of Different 8ections.-Fig. 8 shows the comparative torsional efficiencies of certain different shapes, illustrating the
striking advantages of. the hollow box, or tubular, construction. These advantages would not obtain entirely if the section were built up by use of
bolts or rivets.
EXAMPLES OF. J!ESIQ..N.
General Statement
Structural members are_often required to carry torsionaL loads, generally
las ~ ~ec~nd~r~;' factor combined with bending. or direct stress~ _ Problems
involving the b'ending of unsymmetrical sections, and. problems of elastic
stability, such as the buckling of flanges dl!ring bending, also require a knowlr
edge of torsional properties.
.
,
In the design of short beams to carry tors1onal'"loads'considerable advantage may be obtained by fixing the ends, re'JUlti!1g: in~ iqc:reased str~ngth anc\
rigidity, with corresponding decrease, of angular deflection.. External fixity
is not needed-it is only necessary to box in the t~a"" fiang!l~ at -each~"D.d" and
thereby preyent, as nearly as possible, thei~ relative warping. 'The two
flanges then' act as two rectangular fixed-ended be~;ns ca~rYfng~;};te;~i.dis.
placement, mutually opposed in direction.
.
. 1 -. . In designing long beams,_the end effect tapers out rapidly toward ilie center;
and the formulas for pure torsion, free-ended, will be 'adequate_ and simpler
in their application. It would stil~ be of practical advantage and a' means of
additional safety, however, to "box" the ends of the beam.
Free-Ended Torsion
If a beam is to be designed as free-ended, only shearing stresses need be
'considered: - Regardless of the partial .restraint that does exist as an incidental feature of the details, such a design will be on the safe side. The
critic;al ,shearing stresses will occur along the outer surface of the beam
where the material is thickest, generally along the outside center lin~ of the
flange and along the inside re-entry fillets .
. Th~ she~ring stress is a function of the thickness of the material aiId the
following formula is proposed for the maximum shearing stress in :the flange
of an H-beam or an I-beam in free-ended torsion:
For, parallel-sided fiange sections,
,
:
TJ=
T (D
+ n)
2K
(25a)
....
April, 1935
479
for sloping flange sections,

T (D +m)
2K,
,
-
TI.= -'-'-'--~-
, '
:
. .,
. ,
(25b)
~'
and for shear stress in, the web,
Tw ., : . (26)
K
in which T
torsional moment. Although Equation (26) is in accord with
torsional theory it gave results which proved low by comparison with actual
tests. The following tentative formula was found to give better agreement:
Tw
T(w+O.3r)
(27)
2K
Equation (27) was adopted for use in the reported investigation. More
accurate stress formulas might be developed by further use of soap film tests,
taking slope measurements at critical points, and testing sections with
vario,!!s ratios of web, fillet, and flange as was 'done in determining the
torsion constant. An equation similar to Equation (27) will give practically
the same values for flange stress as Equation (25), thus:
Tw
_T(n+O.3r)
Tf-
(28)
Equation (25) was used in the reported investigation.

Ooncentration of Shearing Stress at Re-Entry Fillet
At the re-entry fillet of an I-beam torsional shearing stress concentrations
occur due to the sharp curvature of the fillet. These stresses are mostly of a
local nature and do not greatly influence the yielding of the beam as a whole.
Although they do not necessarily govern the design of the beam they are
important in ~he study of adequate fillet sizes and in the determination of
loads producing strain lines in the fillets. In a beam subjected to flexural
as well as torsional stresses the total stress concentration at the fillet is the
sum of the stresses due to these two causes.
The concentration of torsional stress at the fillet between flange and web
is illustrated in Fig. 9 for the formulas developed by Foppl(9), Trefftz(8),
Timoshenko(5), and Westergaard and Mindlin, and for soap film experiments
conducted by Griffith and Taylor(3), and by Oushman(13). The analysis
resulting in the Westergaard-Mindlin curve was communicated to the writers
by its originators. It is noted that all curves indicate a rapid increase in'
stress' concentration with' the decrease of the ratio, .!... , between the radius
.n
of the fillet and the thickness of the section next to it, particularly for
ratios of less than 1.0. Strain lines, therefore, will appear at relatively small
torsional tpoment for sections having small fillets.
As the fillets become increasingly large another factor is introduced since
the increased stress due to the greater thickness of material becomes of more.
480
-STRUOTURAL BEAMS IN TORSION
Papers
importance. The rigidity of the beam is increased in proportion to the third

power of the thickness, whereas the stress varies directly with the thickness.
Hence, for any given moment the stress would actually decrease. The curve
of soap film tests by Griffith and Taylor indicate this reverse effect, but
further experiments along this line should be conducted to establish these
relations more definitely.
Fixed-Ended Torsion
Assumptions.-If the ends of a nonccircular section are fixed in some

manner so as to prevent free warping of the end section, pure torsion no
longer exists.
Various investigators have studied this problem-Foppl(9), Timoshenko(5), Sonntag(7), and others. In 1930 an investigation on channel
sections was reported by Seely, Putnam, and Schwalbe (10). These investigators have generally been concerned with the problem of elastic equilibrium
involved in the side buckljng and twisting of a beam in bending without
hiteral support: Hence the torsion problem has been a secondary issue.
In the present tests both ends of structural beams have been fixed by
welding on heavy end plates and additional side stiffening plates between the
flanges at each end of the beam. The major stiffening effect produced is
that of fixing the flanges relative to each other. (Refer to Fig. 10.) The
FIG.
10.-BEAM
WITH
FIXED ENDS.
FIG.
1 1 : - TWISTING
OF
STRUCTURAL SHAPE,
prevention of the individual warping of the component rectangular parts

would taper out so rapidly that it would be of negligible importance; but,
by fixing the flanges with respect to each other, the effect of two opposed
fixed-ended beams is produced, as illustrated in Fig. 10.
The following assumptions have been made: (1) The flanges remain at
rjght angles to the web; (2) the angular deflection is small compared with
the length of the beam; (3) the bending of each flange about its weaker axis
is a negligible factor; (4) a point on the neutral axis of one flange can be
located with sufficient accuracy by co-ordinates (x, y) measured along the perpendicular to the original position of the axes; (5) the two ends of the beam
are held between mutually parallel planes as twisting takes place; (6) the
April,1935
481
displacement .of the flanges due to beam action is due to bending only (that
is, lateral shearing deflection is neglected; a correction is made afterward for
shearing deflection in very short beams); and (7) I u

moment of inertia
2
of one flange about the web axis.
Assumption (7) is quite accurate and of great convenience in the case of
standard I-beams and H-beams for which I u is given in all handbooks. The
following symbols in addition to those previously given will be used:
.!!:-~Elu
a =
KG
or, for steel,
~
K
0.806 h
and for sloping flange sections,
h=d-~(n+_Z_' )
z
in which h
+n
= distance between flange centroids; or, approximately,

h=d-
m+n
2
which is sufficiently accurate for practical purposes.
Derivatio;" of General Equations.-Consider the equilibrium of any cut
section as shown in Fig. 11. The outer torsional moment must be resisted by
the internal moment of the resisting forces, thus:
= T u + Qh
(29)
in which T u
torque required to twist a beam in a free-ended condition;
and Q = total lateral shearing force developed by pne flange. In terms of the
"
h
chosen co-ordinates, - dtJt
dy, approximately, and the twist, tJt, per unit
2
length
dl/l. :gence,
.
.
dx
_
dl/l
2 K G dy
T u - KG =-(30)
dx
h
dx
Furthermore, assuming that the larger moment of inertia of one flange is

equal to I u ,
2
E I u cPy _
dx!
(31)
Considering. Equations (29), (30), and (31):
E Iv cPy _ 2 K G dy _ - T
2
de
h'
dx
(32)
482
Differentiating with respect to x, and substituting a
Papers
!!...- ~ E Iv:
d'y
a' -
d'y _
- 0
dx'
dx'
As a general solution of this differential equation: y
A. sinh ~
a
==
KG
(33)
+ B cosh ~a + C + Dx
"r-
(34)
Torsion with Both Ends Restrained.-By proper evalution of the constants for the conditions obtaining in a beam fixed at both ends:
T h a (COSh ~ tanh _1_ _ sinh ~ + ~ _ tanh _1_) ... (35)
2KG
a
2a
a
a
2a
in which 1
length of the beam. For x
1, the total deflection equals,
Th (.!:... _
y =
_1_)
2 tanh
2KG
a
The moment in each flange equals,
in which u
E Tv d'y
2 dx'
2a
(J;
................ (36)
sinh u
cosh _1_.
2 a'
............... (37)
_1_ _ ~ ; and, furthermore, the shea; in each flange equals,

2 a
a
_ E 111 d3 Y
- T cosh u
= (38)
2
dx
h cosh _1_
2 a
The longitudinal stresses along the outer fibers of the flanges will. be
given by:
Q -
Me
u=-T
Mb
Ta b
sinh u
............. (39)
hIv cosh _1_

2 a
Critical stress will be either longitudinal stresses at the end of the beam,
or the sum of the lateral and torsional shearing stresses at the center of the
beam.
1
, and, consequently,
2 a
Tab
u = - - tanh
..................... (40)
h Iv
2 a
and at the center of the beam, u
0, giving,
At the end of the beam, u
Qc
T
h
= - sech
2
hcosh _1_
2 a
(41)
".
April,1935
483
derived for8 c
= dif;
dx
at the center; thus,
OC=KTG[O:'h(~~)
cosh
,]
(44)
This derivation (Equation (44)) has been based on the assumption (6) that
d~flection is due to bending only. As the beam is shortened, however, shearing deflection becomes of increasing' importance and should be considered.
Timoshenko(6) has indicated a strain energy method for calculating the
deflection due to shear when cross-sections are constrained from warping. By
combining his result for the simple cantilever beam' the correction,
1
2.95
~,
is obtained for a fixeJ-ended beam with point of inflection at

l'
.
the center; that is,
Oc= ~ fl,OOh (~~)

K G
cosh
IlJ (1 + 2.95~)
.......... (45)
l' .
and, denoting the "equivalent" torsion constant at the center by C c:
C = 0T
G
C
=K [ cos'Oh O(- ~r)

- 1
2a
If
+ 2.95
,.J.......
l'
(46)
484
Papers
Using Co as the measure of torsional shear developed at the center of

the beam instead of K and combining Equations (25) with Equations (42)
and (41), Equations (47) to (53) are derived for combined torsional and
lateral shearing stresses, as follows:
Total Maximum Shearing Stress at Ce.nter of Beam (Along Center Line
of Flange).-For parallel flange sections:
T
T [
b'
4 hlu cosh
ClJ
CD
+ n) 1
2 C.
for sloping flange sections:

T
= T [
b' (2 n
+ m)
J
1
+ D +m
12hmlucoSh(2la)
. 2C.
(47a)
(47b)
and the stress in the web may be computed by:
_ T(w+0.3r)
(48)
C.
_
Total Twisting Deflections of Fixed-Ended Beams.-Equation (46) provides an equivalent torsion constant based on the unit angular twist at the
center of the beam. A measure of the total twisting deflection of the beam
over the entire length is desired and c~m be obtained effectively by evaluating
an average equivalent torsion constant, which will be denoted as C....
The expression for total angular twist is, then:
TID
Tl
CAG
l/I=-
(49)
For very short beams most of the deflection is that producing shear and
Equation (36) is an expression for the total
deflection of the flanges due to bending only. Oonstant CA may be-evaluated
from Equation (36) in so far as bending deflections are concerned.
Co approaches CA in value.
The ratio of CA reduces to the following expression:

C.
_
C... ='
C.
et
(COSh 2 a -
l
- (COSh 2
a) -
1)
.............. (50)
sinh
a)
The graph of C is given on Fig. 12, permitting the quick calcula-
C.
tion of CA after Co is known. As the length approaches zero, CA _ should

approach Co in value and the curve on Fig. 13 gives the reduction to be made in
CA for various ib -ratios

in terms
,ofa
- factor to be multiplied by , (C A
As an example, let: C c
= 6.00 by Equation
(46); -
2a
= 2.00;
l
and, h
-Co).
= 5.00.
April, 1935
485
From Fig. 12, CA'
Co
= 1.42;
= 0.44; or, C = 6.00

A
and, from Fig. 13, the reduction
(1.42 - 0.044)
(0.42) (0.105)
= 8.28.
Torsion with One End Fixed and One End Free.-Often, in cases of combined bending and torsion, one end of the beam will be relatively unre1.6
>,.
1.5
--~
1.4
'"
lD
.~
~
1.3
G"lll"
1.1
FIQ..
00'
1.0
2.0
1.00
.E'"
0.80
'\
1\
~0.60
G"lll"
r--...
1.2
I'--- ........
3.0 4.0 5.0

Length Ratio
---
6.0
fa
---""0.40
'0
""
r---
7.0
8.0
0.20
00
c::
1.0
2.0
r--- r--
3.0
4.0
Length Ratio
l2.-POSITIVE FACTOIl TO OBTAI~

To BE USED IN CONJUNCTIO~
5.0
6.0
7.0
8.0
FIG. l3.-REDUCTION FACTOR FOR SHEAR

(FOR USE WITH I!'IG. 12 TO OBTAIN
FROM 0 ),
WITH NEGATIVE FACTOR FROM FIG. 13.
strained while the other end is fixed. At the free end there will be no lateral
shearing stresses in the flanges and the shearing stress formulas (Equations
(25 for free end torsion will apply. The evaluation of Equation (34) for
these end conditions gives the following: For maximum bending moinent in
the flange at .the unrestrained end:
T
al
M max = - tanh h
(51)
500....-----,,.----,-----..-----..---....,----....,------,25000
Maximum
DiLt Slress.,.
---..:.::...=-----
$.
.-
a.
g
~
Working Strength
~ of Beam
'5.'
.~
22000
~ 20000 .!i
-:g 400 \
3 0 0 1 - t - - \ - - I 1 - - , . . - - I - - - - + - - - - + - - - - + - - - - + - - - - ' - I 15000
\
\-\-
E 200
~-.~51
~a_____
-~(~S~~--
s\\e'~a(~\n;:?""~::=:..-+---+_---+_-------i 10000 ~
-~
~.
E
--"-
~ 100
Td
------ -- --- --7-Free

Working St'rgth
o Lo
12000
' -_ _--1
.~
5000
C:=:-:i==:====:=d
----l
----l
--l
--l
10
12
4
length
of
Beam
in
-.J 0
14
Feet
FIG. l4.-WORKING STRENGTH OF AN 8 BY 8-INCH BEAM WITH FIXED

ENDS (SHOWING LIMITING STRESSES FOR DIFFERENT LENGTHS).
::!O
486.
Papers
and for maximum direct fiber stress in the outer edges of the flange~ at the
restrained end: .
Tab
l
= - - tanh h Iv
a
The total angle of twist is given by:
(52)'
(J
if;
= -Ta
[ -l
tanh - I ] [ 1
+ 0.74 -b
(53)
KG
a
a
I'
Equation (53) will be accurate for all except very short beams.
Fig. 14 shows the application of the proposed formulas for maximum
longitudinal and shearing stresses to varying lengths of an 8 by 8-in. H-beam
fixed at both ends. It is noted that for lengths ranging between about 1 ft
and 9 ft the longitudinal stresses based on an allowable stress of 22 000 II? ,
per sq in. determine the design. For very short lengths and for long lengths
.the shearing stresses are critical.
.
.
Design of End Connection.-It is suggested that the end connections be
built as illustrated in Fig. 15. Oonnections of this type proved very satis-
.trn
~
(a)
. .'.
END SECTION
Numbers Indicate
Sequence of Tests
"
1__
(b)
FIG.
GENERAL VIEW
15.--':'-FIXED,E~DED
FIG. 16.-TYPES OF SECTIONS STUDIED BY SOAP
BEAM.
FILM.
factory
fixity.
flanges,
design.
in the actual torsion tests and provided comparatively complete end

The purpose of the end plates. is to prevent relative warping of the
and the following approximate analysis should serve as a guide in the
The moment in on~ flange at the end is obtained from Equation (37)
Ta tanh ..i.
'.' ,
(54)
h
2a
The value of l should be measured as the over-all length. Let Q.
the total
shear of the beam in the stiffening plate; and s
the distance between the
stiffening plates. Then, if the stiffening plates alone are assumed to fix the
ends of the beam:
]j[
Qs
(55)
April, 1935
Substituting
STRUOTURAL BEAMS IN. TOR~ION
III
,487
Equation (54):
Ta tanh J.(56)
h s
2a
The stiffener plate is welded to both flanges and to the end plate as well.
The stiffener and the adjoining part of the end plate act as a short xedended beam holding the flanges in place. No attempt was made t() analyze
the load distribution, and the design of the test beams was largely a matter
of judgment.
.
. The following tentative suggestions are made, as the result of the tests:
Q
(1) The length of the stiffener along the beam should be equal to abo~t
three-fourths the width of the flange for H-sections and to the full flange width
for I-beams;
(2) The thickness of the stiffener plate material should be gref,ltcr than
-that of the web thickness or greater than one-tenth the length of the stiffener plate along the beam;
(3) The stiffeners should be machined to a tightt oetween thc Jlanges
and should be welded to flange and end plate continuously on the outer part;
(4) The end plat~s shouid have' a .thickness equal to twice the l~aximu~
thickness of the beam m~terial; a~d
.
(5) The beam should be cut square and welded to the end p~atc with a
continuous flllet weld
about the entire
.
. beam end.
.-
The stiffener' plate and the weld, between it and the flangc should be
designed to. ;esist the she~r as computed by Equation (52).
, Design Examples
\
.-
General Remarlcs.-The most economic structural H-bcam or, I-beam' jor

torsional strength is one in which the material is most nearly of ~onstant
thickness throughout and is as thick and compact as obtainable.' Column se.ctions with parallel sided flanges and of the heaviest rolling in each series
,
most nearly satisfy these' requirements. '
The torsional design should be made with ends assumed to be free in the
case of riveted or bolted end connections; any'percentag~'-of end xity incidentally present wiII simply provide an additional factor o safety. Only
shearing stresses as computed by Equations (25) and (27) need be c~msider.ed
In free-end d e s i g n . '
,
Beams with boxed in and continuously welded end connectlons wiII bc
l
somewhat stiffer and stronger' dependin'g oh the length: Both ,longitudinal
stresses and shearing stresses must be consi.dered '(see ~ig. 14).' T~sts indicate that the shearing stresses generally determine the yield point of the
beam as a whole. The local direct stresses at the 'ends affect 'ini,tiillly .onl'ya
small part of the beam and are in the nature of secondary stresses. Shearing
yield o~ the other hand occurs along the entire beam length. The allowable
direct fiber stress wiII be made 22 000 lb per sq in. in the present discussion.
It is suggested that allowable ber stresses usual in secondary stress design
be applied in general to these stresses.
488
Papers
General Data.- The following data apply to all the examples: Allowable
working normal stress, a = 22 000 lb per sq in., for secondary stresses due to
fixed-end torsion; allowable working shear stress, '!
12000 lb per sq in.;
E
29 000 000 lb per sq in.; G
11 150 000 lb per sq in.; Poisson's ratio,
p.
0.30; l
over-all length of a beam, including stiffeners, along which a
. uniform torsional moment is assumed to act; 1 0 = 0.01745 radian; and 1
radian
57.30 degrees. Computations in these problems were made with a
10-in. slide-rule.
Design Example A.-A long beam with torsional deflection limited:
Assume a beam 20 ft long designed to resist a total torsional moment of
20 000 in-Ib with maximum total twist under the load limited to 1.2 degrees.
The procedure for designing the beam as free ended involves three steps:
(1) Determine the unit angle of twist, IJ, in radians per inch, thus, IJ
=
=
'= 1.2 X 0.01745

20 X 12
= 0.0000872 radian per inch;
(2) calculate the required K-
= G-T() =
20000
20.6
11 150 000 X 0.0000872
in.'; and (3) refer to standard tables of K-values and pick out the most economical section. In this manner, a Bethlehem section, 10 by 10 in., at 124 lb
20.37) will be satisfactory. The end connection will proper ft (with K
vide additional rigidity and will allow a small tolerance in picking sections.
Design Example B.-Analysis of the torsional strength of a short beam
(B8b, 8 by 8, at 67 lb per ft), with different end connections: The general
data applying to this case are: l
66 in., over-all; K
5.145 in.'; 111
88.6
in.'; n
0.933 in.; b
8.287 in.; D
1.206 in.; h
9.000 - 0.933
8.067
value from Equation (5), thus, K
..
in.; a
=
= 0.806 h . f4 = 27.0 in.;
.
~K
=
=
~=
2a
66
M
= 1.22;
=
=
cosh - 2a
1.8412 i and .
tanh - = 0.8397.
2a
The free-ended working strength is computed by Equations (25) thus:
2 K T (2) (5.145) (12000) = 57700 in-lb.
D + n 1.206 + 0.933
To determine the fixed-ended working strength, based on shear, compute
the equivalent torsion constant, Ce, for the center of the beam by Equation
(46), thus:
Ct
= 5.145(
1.8412 ) (
1.8412 - 1
1
1(
+ 2.95
)')
8~~9
= 10.77 info
Then, from Equation (47),

12000
T = - - - - - - - - - - - - - - - - = 108000 in-Ib
8.29"
1.206
0.933
4 X 8.067 X 88.6 X 1.841
2 X 10.86
See, for example. Bethlehem Manual of Steel Construction, Catalog S-47, 1934. p. 285.
. ..-.
April,1935 ,
489
To determine the fixed-ended working strength, based on longitudinal

fiber stresses at ends (tension or compression), apply Equation (40):
22000
T = - - - - - ' - ' - - ' - - - - - = 84000 in-lb
27 .0 X 8. 287 X O. 8397
8.067 X 88.6
The longitudinal stresses, therefore, determine the design of the beam, and
the allowable torsional moment is 84000 in-lb.
The shear to be resisted by the end plate is computed by Equation (56);
thus, if s is assumed as 6.5 in.,
Q = 84000 X 27.0 X 0.8397 = 36300 lb.

8.067 X 6.5
A i-in. plate fitted into the 7.13-in. space between the flanges and 6 in.
in length along the beam, will satisfy the requirements suggested. Assume a
.g.-in. fillet weld between the stiffener' and the flange. If s = 8.827 - 3 X i
= 6.41 in.,
Q
6.5 X 36 300
6.41
36 800 lb
The stress in the plate is:
r=
36 800
9 800 Ib per sq in.
X 6
. and in the weld:
= __3_6_8_0_0_ _ =
13900 Ib per sq in.

5
X 0.707 X 6
8
which is too high. Make the plate 71 in., long rather than 6 m. Then the
stress in the weld. is:
r
r = _6_ X 139 000
7.5
= 11100 Ib per sq in.
which is less than the limit of 11300 Ib per sq in. re<;lommended by the American Bureau of Welding.
TEST RESULTS
.'
.Soap Bubble' Tests: Purpose and Program

The purpose of the series of tests reported herein was toevalull,te, accurately, the torsion constant of structural H-beam and I-beam sections. 'Specifically, this problem narrowed down to determining the added torsional rigidity
introduced by the juncture of two rectangles, with fillets at the re-entry
corners, in excess of the torsional rigidity of these rectangles treated ,as
separate members. The problem, therefore, was to determine ex in Equation
490
Papers
Ci6). In order to establish the value of a for any shape of section it was
necessary to consider two-variables, w, the ratio of web to flange thickness,
n
and .!-, the ratio of fillet radius to flange thickness.

n
Furthermore, it was
essential to study sections with sloping flanges as well as those with parallel
sides.
A program of tests was outlined to cover a wide range of the two variables,
wand .!-. Although it would have been desirable to measure the slopes of
the bubbles and. thereby study the stresses,particularly in the fillets, such a
study would have greatly reduced the total number of tests possible. It was
thought better to establish the torsion constant definitely, in which case it
was only necessary to measure the volume of the displaced bubble. In each
series a basic web and flange thickness was adopted and after testing the
section with zero fillet radius, the various fillets were cut away in sequence as
shown in Fig. 16.
The curves obtained from the test results are shown in Fig. 7. It was
found that scattered points occurred along the lines of 0.0 and 0.2 fillet
radius. These variations may be due to the difficulty in machining the plates
with the small fillets, a slight inaccuracy causing considerable variation in
the diameter, D, of the inscribed circle. A further difficulty is encountered
in the plates of zero fillet radius due to the tendency of the soap film to
. jump across these sharp corners. Most of the structural beams actually rolled
have ratios of wand .!- both greater than 0.5 and in this ~rea the data were
n
n
.
quite consistent.
Tests of Steel Beams: Purpose and Apparatus
Torsion tests were made on steel beams, with several objects in view. The
sections themselves were chosen so as to give a range of shapes and sizes. as
great as possible, and tests of certain unu,sual shapes, which are not at present
standard, were nevertheless valuable in the investigation.
Tests of free-ended conditions were made on a number of beams in order
to check the results of the soap film experiments and the corresponding
method of calculating the torsion constant. In these tests the distribution
of shearing stress was studied, and a check was obtained on the proposed
approximate formulas for stress. Tests of fixed-end conditions were made
on different shaped beams and the effect of variations in length was studied.
A type. of end-connection design was developed to give a considerable degree
of fixity.
A standard torsion machine of 26000 in-Ib capacity was used to test
3-in., fixed-end I-beams, of various lengths. It was also used for torsion tests
of round bar samples of all material to determine the shearing modulus of
elasticity.
........
April, 1935
STRUCTURAL
BEA~l
11\ TORSlO:-I
491
The cable torsion rig, shown in Fig. 17, with an ultimate capacity of
750000 in-lb was used in the major tests: Most of the large beams were
tested in lengths of 6 ft, but two tests each were made on beams 1 ft 3 in.
and 3 ft long by means of the same sheaves and cables adapted for use with
.....-
FIG.
18.-LEVEL
BAR.
FIG.
Hl. -
CONNECTION
FOR
BEAMS.
FREEENDED
shorter top and bottom beams. During the tests of light beams, cables i in.
in diameter were used because of their flexibility and ease of handling, but in
the tests of the heavier and shorter sections the cable was changed to 1 in.
in diameter in order to develop the full capacity of the machine.
The sheaves were made of material 2 in. thick and were machined to a
minimum diameter of 17 in. A hole bored through one of the diameters
allowed continuous action and rever ing of the cable without fouling or introducing bending moment. This machine gave perfect sati faction in every
respect and was easily set up and dismantled. During tests the apparatus
was in such a state of balance that the heavy pulling beams could be easily
tilted either way by hand while maintaining a heavy torsional load of the
test specimen.
T "TorsioD TestiDg Machine of 750000lnchPounds Capacity," by Bruce G. Johnston.
Jun. Am. Soc. C. E., Engineering New8-Recal'd, Februar~' 28, 1935, p. 10.
492
Papers
Measuring Devices.-The level bar illustrated in Fig. 18 was built to

measure the change in relative altitude of two points 3 in. apart. It was
used to measure relative angle changes in all the beam tests. The micrometer
vernier permitted readings to 0.0001 in. and the level bubble was sensitive to
miClometer changes of. about. 0.0003 in. The total range of the instrument
was about 9 30' from the orginal level position.
The torsion meter was used in the torsion tests of round bars in the
26000-in-Ib torsion machine to measure the angular twist over a 3-in. section
of the bar. This device consisted of two steel collars attached by pointed
thumb screws to the bar and provided with two 1:1 000 Ames dials for
measuring the tangential displacements. The instrument was the same as
that used and described in a previous investigation at the Fritz Laboratory."
Twenty tensometers were used to obtain the strains at all critical points
during tests.
[ Test Procedure and Method.-In each torsion test, whether fixed or freeep-ded, there were three principal objectives: First, to learn as much as
possible about the strain and stress distribution; second, to measure the
torsional 'stiffnes~, or ratio of torsional load to angular twist; and third, to
l~arn the u$~ful torsional load-carrying limit of the beam as a whole.
: The strain and stress distribution was studied in two ways: First,. by
tensometers which were sensitive to changes in strain corresponding to from
150 to 300 1b. per sq in. in stress, depending on the model type; and second,
the beams were whitewashed with a mixture of water and hydrated lime
which showed the distribution and location of the first surface strain-slip
lines.
In computing the stresses from the observed strains the same values of
E, G, and /k, as those used in Design Example A, were adopted. The
torsional stiffness was gauged by measuring the relative angle changes bet-jveen two points along the beam by use of the level bar. In the free-ended
t~sts the angle changes were measured over a 36-in. length along the center
part of the beam. In the fixed-ended beam the unit angle change varied
along the length and a measure of the average stiffness was obtained by
measuring' the relative rotation of the end plates and of points a short distance from each end where the reinforcement ended.
The yield point of the beam as a whole was obtained by a study of the
torque-twist diagrams _and a knowledge of the load when the first strainslip lines appeared. In a few cases a definite drop of the beam was noted
and, in such cases, this was taken as the yield point. In .most of the tests
the yield point was taken as the torsional moment corresponding to the point
on the torque-twist diagram where the co-tangent of the slope was 1.5 times
the value of the co-tangent of the slope of the straight part.
Test Results.-Table 1, Appendix III, gives a general summary of all
the tests made, including dimensions of beams and computed X-values based
on actual dimensions. The dimersio'ns were obtained by means of micrometers
8 "Shearing
Properties and Poisson's Ratio of Structural and Alloy Steels," by
lnge Lyse, l\L Am. Soc. C. E., and H. J. Godfrey, Procee(!in.Q8, A.S.T.M., 1933, Vol. 33,
Pt. II.
April, 1935
493
and calipers. Readings were taken at a number of different places on the

beam, and averages from them were used to calculate the weight of the beam'
per foot of length. Furthermore, the beams were actually weighed and any
discrepancy between computed and actual weight was taken care of bY-'
adjusting the average measured dimension to give the actual weight.
, Table 2, Appendix III, presents the' physical properties based on tests
of samples taken from the test beams. The' tensile values are based on the
average of two tests of American Society of Testing Materials standard tension
test specimens (2-in. gauge length). The torsion tests and slotted-plate shear
tests were made in the same manner as described in a previous investigation
at the Fritz Engineering Laboratory.s In most cases, the tensile specimens
and round-bar torsion specimens were cut from the material where the
flange and the web join, as it is at this point that critical tor&ional !3tresses
develop. The material for the slotted-plate test specimens was. cut from the
webs of the beams.
Free-Ended Tests.-Free-ended tests were made on seven beams. Ea"h
specimen was held in the torsion rig by two bo}ts at each end which passed
with a loose fit through the web and through the two angles, the angles being
welded to end plates which, in turn, were bolted to the sheave plates of the
testing rig. Torsional moment was applied by means of lugs welded to
the 'end plates which engaged the flanges of the test specimens. The flanges
ill
.-
48
~\.
r---.----r---r--~..,.....----:::l
30 ,----i-----,
Material
Yield Point
401----\----+---1---7"--+-----1
.g
Stress Distribution
at Beam Yield Point
c:
&
..:
E 32 1---+----+--/"~-I----+------1
15
~
-g
241---+-,f-l--=+----::-::-_l--,---+---j
c:
iii
E
o
::E
~
'i!!
~
16
<= 12000 Ib per sq in. by Formula
<; 12000 Ib per sq in.by Tensometers

8 [--'-=,...--\----+--.........;1----+-----1
(b)SHEARING
STRESS
DISTRIBUTION
ON SURFACE
OF FLANGE AND
WEB
(a) f-DIAGRAM
o0~-"'"'":0-:.OO:::2:----:-0.0':!:0::-4-:--0'"'.0!-:0-:-6--'--=0.-!00=8:----J
0.01O
Unit Twist in Radians
FIG. 20.-FREE-ENDED TORSION TEST
T-2G,
A 12-INCH I-BEAM.
were thus free to warp and the beams were almost entirely unrestrained at'
the ends. Fig. 19 shows the details at one end of the largest beam tested
and is typical of all the free-end tests which were made. Figs. 20(a) and
21(a) show typical torque-twist diagrams of two of the tests. Figs. 20(b)
494
Papers
and 21(b) show the stress distribution based on tensometer readings taken
in these same tests. The tensometers were placed on the flange and web
surface at about the center cross-section of the beams and were set at an
angle'of 45 with the longitudinal axis of the beam, in order to measure
0.6
r----r---r----r-----,---:=.,
.~
"0
"
0.5 1__---+~~~+___+__7'--=-I__~-_+~--+~.~ 20 F=r==~;=.o'F~C=C====j
.e
.-5
~~
Drop -of Beam and Strain

Lines along Fillets
.g
is :9
I .
.to 10 1-</4-:""'~,.,.,.T'l'A+Iil+1'fl'm"TrrTn"r.:_--\-1
O.4I__---+-~_/_+_--~I__--_+--___j
E
'0
j
~
0.3
I__---+-"-j~--l-----I__---+-~___j
't'= 12000 Ib per sq.in.by Formula
"g
"
.1
< = 12000 Ib per sq "'.by Tensometers
::l;
:~ 0.2
~,
(blSHEARING
STRESS
DISTRIBUTION
ON SURFACE
OF FLANGE
. AND W~B
20
O.II--"I--+---I-~-+-----1'----__j
(a)
0.0004
0.0008
t=DIAG AM
0.0012
0.0016
12"X 12" 8eam

@ 190 Ib
0.0020
Unit Twist "in Radians
I~IG.
21. -
FREE-ENDED
TORSION
TEST
T-31, 'A 12-INCH
POUNDS PER FOOT.
BETHLEHEM
SECTION
AT
190
one of the principal strains which, in the' case of pure shear, will be equal
in both directions. In Figs. 20 (b) c,nd 21 (b) the shaded areas show the
stress distribution for '!m
12000 lb per sq. in.
The data of actual stresses from tensometer readings were' available at
only a limited number of points. In drawing the curves of stress distribution these data have been supplemented by known facts, deducible from the
general torsional theory, soap bubble tests, and from the actual beam tests;
that is: (1) The shearing stress equals z~ro -at outside corners; (2) there
is a "hump" in the stress curve at the outside center of the i flange; particularly
if the fillets and the web thickness are relatively large as compared with the.
flange thickn~s; and (3) theshearing--st~ess on the surface of the flange
and web is approximately proportional to the thickness of the ma.terial.
Table 3, Appendix. III,_ is a ,summary of the results obtained in the freeend tests. Good agreement" is shown between the torsion constant computed
from the measured dimensions and that obtained from the test results. The
test value for K was obtained from the slope of the tor.que-twist diagram
and from Equation (5); thus, T
= KG ();
or, K
= :!.....
GO
The maximum variation for' the K-value of test results was 6.7 and
the average variation of seven tests was 2.26 per cent. It is noted that
April,1935
~/
"
495
K for the heaviest beam tested 'was about two hundred times greater than
for the lightest beam and that the corresponding agreement for these tests
was expressed by a variation of 0.9 and 0.0 per cent.
The shearing stress computed by Equation (25) gave average stresses
7% less than those based on tensometer readings. The stresses in the web
by Equation (26) which, theoretically, should be correct, were much lower
than as computed from tensometer readings. Equation (27) was suggested
on the basis of these tests which, of course, are not complete enough to substantiate its adoption definitely. However, both Equations (25) and (27)
are believed to be usable for practical design purposes with ordinary values
of allowable shearing unit stress. The following special remarks apply to the
individual free-end tests:
In Test T-12 strain lines appeared along the fillets at 13500 in-Ib; and
along the outside center line of the flange at 15200 in-lb. The yiel9. point
of the beam as determined by the slope of the torque-twist diagram was
15900 in-lb.
In Test T-22 the freedom of the ends from restraint was checked by tensometers placed longitudinally near the ends. The strains were negligible.
The first shear strain line along the center line of the flange appeared at
21 210 in-lb. At 25610 in-Ib, strain lines progressed rapidly along the flange
and in the fillets, and a definite drop of the beam was noted. .
In Test T-25 strain lines appeared along the fillets at 50900 in-lb. Thereafter, the slope of the torque-twist diagram became nearly 50% greater
than for lower loads, maintaining nearly the same slope up to 100000 in-lb.
The yield point was taken as 50900 in-lb. Strain lines appeared along the
outside of the flange at 77 340 in-lb.
In Test T-26 (see Fig. 20) a slight checking in the fillets was noted at
23000 in-Ib, with a drop-of-beam yield point at 25000 in-Ib; strain lines
progressed along the outside of the flange at 26500 in-lb.
In Test T-30 the yield of the beam was noted by the 1.5 on 1.0 slope of
the t~rque-twist diagram at 64 000 in~lb. The first strain lines appeared
over ;the web at an indeterminate load due to the presence of heavy scale on
the section.
I~, Test T-31 (see Fig. 21) the beam and dropped strain lines occurred
along'Jhe fillets at 48 800 in-lb. The yield point was noted by the 1.5 on 1.0 slope
at 4~0 000 in-lb.
In.'. Test T-33 strain lines occurred along the fiilets at 65 600 in-lb. Yield
point'a nd strain lint7s appeared along the outside of the flange at 76400 in-lb.
Fixed-End Test;.-Twenty-two fixed-end tests were made on beams of

different sizes. Tests T-4 to T-13, inclusive, were made with different lengths
of a 3-in., 7.5-lb I-beam. These specimens were welded to end plates, 1 in.
thick, with a continuous i-in. fillet weld. The end plates were bolted to
1t-in. plates which were attached to the jaws of the standard 26000 in-Ib
torsion machine. The relative rotation of the end plates was measured by
level-bar readings taken at each end. The remainder of the beams were tested
in the cable-testing rig. All these beams were larger than the 3-in. I-beams
120,-----,------.-----,-
100
...,
--,
i---i-----r------j-----t-=====l
for
, 7Tensometers
Direct Stress
r'"
a
Shear Tensometers
rf. 80 t-----t-----+---=O'..,{-=----+-----+------1
en
Stiffeners
Yield Point
of Beam
'0
'1
1-------------6'0'~'
----------'--
>-3
::<l
5"
(c)
TOP VIEW OF FLANGE
5"
"0
C
1::t'
ro
60
t------j-----I'---II------+-----j------l
to
i."j
:>-
]j
I'::
en
:;;
0
'>-3.
401--+----
'40 ) - - - - - .
>-3
0
::<l
-g
SHEARING STRESS
AT SECTION
AT
201----<1'---+----+
THE YIELD POINT - - - - - - 1
STRESS IN THE BEAM
(b)
a-a
~ . 201---1-------''''''-=~.S
Tension in Top
>-",
Outer Fiber
.S ~
~~
~c.
en
0
Z
01-----------------===-""""'=----------+---1
~:9
~o 20 .
40
0,002
0.003
0,004
0.005
(d) DIRECT STRESS IN OUTER FIBERS OF FLANGE AT
YIELD POINT OF THE BEAM
Unit Twist in Radians
FIG, 22.-FIXED-ENDED TORSION TEST
.1
T-17,
A 6-INCH H-BEAM AT 41 POUNDS PER FOOT.
,
April,1985
497
and were welded to Ii-in. end plates with additional end stiffeners, fitted and
welded between the flanges, as was illustrated in Fig. 15.
Relative rotation of the end plates was observed on all beams by means
of level-bar observations, and the twist of the large beams having stiffener
plates was also measured at points just inside those plates. Strain readings
for longitudinal and shearing strains were observed wherever feasible.
Fig. 22(a) shows a typical fixed-end torque-twist diagram, and Fig.
22(b), 22(c), and 22(d) show the computed stresses from strain readings at
tlle yield point of the beam. These observations are typical of all the fixedend tests.
Table 4, Appendix III, gives the summary of the test results for fixedend beams. In computing the values of Oc and OA the question arose as to
the correct length to be used. If 100% end fixity were possible the correct
length would be slightly less than the over-all length and somewhat greater
t.han the length between the end stiffener plates. However, the over-all
length is the simplest approximation, and it gives the best results by comparison with the tests, except in the case of very short beams with end
stiffeners. In these two tests (T-19 and T-27, Table 4) the apparent percentage of end-fixity seems inconsistently high. Two tests have unusually
low percentages of end fixity (see Tests T-16 and T-24). The explanation
for this is given under the special remarks. The average percentage of end
fixity with Tests T-16 and T-24 omitted is 88.3 and all the 6-ft beams,
except T-24, have an end efficiency greater than 85 per cent.
In most cases the yield points of the beams were determined from the
slope of the torque-twist diagrams and the theoretical direct stresses computel on the basis of this yield-point torque are given in Column (14),
Table 4 (Appendix III). It is noted that, in spite of incomplete end fixity,
these stresses, in every case, are above the tensile yield-point strength of
the material as given in Table 2 (Appendix III). Hence, all the beams
.-
FIG. 23.-ILLUSTRATION OF STRAIN
PATTERN.
would have been designed afely on the basis of working, direct, fiber stresses.
The average of Column (14), Table 3 (Appendix III), is 55% more than the
average yield-point strength of the material in the test beams.
The computed and measured shearing stresses in the flange agree well
for all the 6-ft beams, with the exception of T-16 and T-24 in which the low

498
\
\
\
~\
c
a
\\
@
~
:E
~
~
RELATION BETWEEN LENGTH AND STIFFNESS
(a)
Papers
.4
{3.
~
2
24
--
Theoretical
K~:,
~L-
---
- -
20
- Com;t;dFree-End-;d stiffnes~
f------
---
\
\
RELATION BETWEEN LENGTH AND YIELDPOINT STRENGTH
(b)
1\
II\
W-,,"",
\
\
\
"
~
' -"-.
"'-...
--- -- ------:.i:------::f.--heoretlcal Torque for T m.. =21800 Ib per sq In
free-Ended Yield
........
------ ----
------
'" " ".
(Yield Point of Beams by Test
-~
\"
>0,,", '"' ,-, ,,",00 "
10
20
length
FIG. 24.-Tgs'r OF
------
30
of
f~r T
Beam
40
In
= 21 800 Ib per
s~ In
-----
-,
---
50
60
Inches
3-Ir-oCH I-BgAM AT 7.5 POUNDS pgR FOOT. FIX gO AT BOTH ENDS.
.
~
April, 1935
499
end fixity affects, directly, the shearing stress agreement. On the short 18-in.
beams with stiffener plates (T-19 and T-27), the discrepancy between the
computations and the test results is high, as might be expected. The computed
stresses in the web give an approximate check on the stresses indicated by
tensometers. Special remarks on fixed-end tests are, as follows:
..
,I
Tests T-4 to T-12 were 'of different lengths of the same 3-in. I-beams and
were well adapted to show the influence of end fixity on the strength and
stiffness with the length of beam the only variable. The flange of each beam
was whitewashed so that the appearance of first strain lines might be noted.
Fig. 23 shows the strain line pattern on the flange of one of these beams
after yield had taken place. Fig. 24(a) gives a graph of test results for
this series showing the influence of length upon the rigidity, and Fig. 24(b)
illustrates the influence of length upon the yield-point strength of the beams.
Special attention in this series of tests was given to Tests T-8 and T-10.
Stress measurements were taken along the extreme fiber of the flanges at
short intervals of length and the lateral bending moment in the flanges for
a definite torque load was . .computed from these. readings. The bend~ng
moments along the beam weteplo.tted.~!ln,cl.. the.' cilrves were differentiated to
give the lateral shear in the.flang~s. The.se results are compared in Fig. 25
with the theoretical variation.in shear by Equation (38).
Tests T-15 and T-16 should be compared with the free-end test, T-14, of
the same section. Test T-16 was a special run with additional stiffeners
placed midway along the section. These stiffeners were of the same type as
the end stiffeners and were parallel in a plane with the web. No outstanding
stiffener was provided. Although this additional stiffener gave the beam an
average stiffness 42% greater than Test T-15, it provided only 40.1% of
the theoretical stiffness of a beam 3 ft in length, rigidly fixed at each end.
The design of this beam would have been safe for strength, however, if
based on the 3-ft length and designed for the proper longitudinal working
stress.
Test T-17, in contrast to Test T-15, was of the heaviest 6-in. section
rather than the lightest. It should be noted that an effective fixity of 96.8%
was attained in this test. The end stiffeners were ~ in. thick: and 5 in. long.
The torque-twist diagram and data on stress distribution are given in Fig. 22
as typical of the results for fixed-end tests.
Test T-18, of a. subway column; provided an opportunity to observe a
section of extreme proportions.
Tests T-19 to T-21, together with free-end Test T-22 on the same sized
section, provided a series of different lengths of 8-in. H-section. It is noted
.that the shortest beam tested showed an end-fixity efficiency of 101%,
whereas the general trend for shorter beams should be less end fixity because
of the greater strains placed upon the end connection. This effect is explained by the fact that the over-all length was used in the computations.
Although this is good approximation for the longflr beams, the stiffness
changes rapidly in the short length range and the correct length is some
500
Papers
STRUCTURAT, BEAMS TN TORECION
10
~\
(a)
TEST T-l0
c
:.:J
1\
'//
~~
~0
10
15
j/
./Aetu.1
:x:
vi
"~ 020
)'
3-IN. I-BEAM 38.96 IN. LONG
Theoretical
"
"c~
\
\
\\
10
15
20
Distance from Center Line in Inches
14
.:10
o
I-
\
8
<b) TEST T-8
" "- \
"-
3-IN. I-BEAM 18..10 IN. LONG
0:.J
Ae!u.J
"
Theoretical
<3
,,~
10
......
-//
If
.6
l//
';(5","
~0
642
0/
-:;.;:'
024
10
Distance from Center line in Inches

FIG. 25.-SHEAR DISTRIBUTIONS 'IN FLANGES OF FIXED-ENDED BEAMS FOR A TORSIONAL
MOMENT OF 2 500 INCH-POUNDS.
Apn1,1935
501
.value less than that used. Fig. 26 shows the strain-line distribution near
the end of Beam T-21 after yielding had occurred.
Test T-23 is of the heaviest section of the 8-in. H-sections, whereas Tests
T-19 to T-22 were of the lightest weight section.
. '...
-\
FIG. 26.-STRAIN LINES AT THE FIXED END DB' BEAM .
Test T-24 is of interest because it had the lowest end-fixity efficiency of

the 6-ft beam sections, with an end fixity of only 45.5 per cent. The remaining 6-ft sections were more than 85% efficient in end fixity. The
explanation is found in the fact that the end stiffeners in this base were not
properly designed, and the beam not only pulled loose at the weld from the
end plate, but the end plate was badly warped during the test. The stiffeners
were -h in. thick and 6 in. long. Had they been 9 in. long and 1 in. thick,
as provided in the tentative Design Suggestions (1) and (2) (see heading,
"Design of End Oonnection"), it is believt2 tha~ much higher efficiency would
have resulted.
Tests T-27 to T-29 are of three different lengths of 12-in. I-beam. The
remarks concerning Tests T-19 to T-21 appear to apply in the case of these
beams also.
SUMMARY AND OONCLUSIONS
.'
Torsional Theory and the Torsion Constant.-The material presented in

this paper may be outlined undel" eight divisions, as follows:
(1) The essential features of the general torsion problem are outlined;
(2) The application of Prandtl's membrane analogy is presented;
(3) An accurate and detailed method of evaluating the torsion constant
of structural H-beams and I-beams is presented (this method is based on a
known theoretical evaluation combined with factors determined by experiment
from the membrane analogy) ;
502
Pape.r.s
(4) Formulas are proposed for the shearing stress in the flange and w"b'
due to pure torsion;
(5) The effect of 'shearing stress concentration in the fillets due to both
torsion and bending is discussed;
(6) The problem of torsion with either one or both ends of the beam
restrained is studied in detail and formulas for maximum shearing and
longitudinal fiber stresses are given;
(7) Specifications are suggested for the design of a welded end connection which will give a high degree of end-fixity efficiency in torsion; and,
(8) Detailed design examples are presented to illustrate the application
of the formulas to torsional design.
.Test Results.-The purp'ose of the soap film tests was to determine,

experimentally, the additive portion in the K-factor due to the added rigidity
-accruing from the juncture of the web and flange with the corresponding
fillets. To this end, soap film experiments were made on fifty-seven differently proportioned sections both with parallel and sloping flanges.. The
application of the membrane analogy as developed by this investigation was
restricted to volume measurement only. The method used was rapid, and,
.it is believed, gave results having an error considerably less than 1
per cent.
In computing the K-~alues of structural sections the experimentally determined part of K amounts, in the most extreme case, to 10%' of the total K.
Hence, an experimental error of 1% would give a possible error of only
0.1% in the total K-value.
_
Free-end torsion tests were made on 5even different beams ranging' in
size from a 6-in. H-beam @ 20 lb per ft to a 12-in. H-beam @ 190 lb per ft.
The heaviest beam had a torsion constant about two hundred times as great
as the lightest' beam.
The method of applying the torque to the ends of the beam provided a
high degree of end freedom. By measuring the unit twist of the frE;e-ended beams
and obtaining the slope of the torque-twist diagram, the free-ended
torsion tes'ts provided (through Equation (5 a definite check on the torsion
constant as computed by the proposed method. Using G
11150 000 lb per
sq in., which is theoretically correct for' E
29 000 000 lb per sq. in. and
JL
0.30, the test' results checked well with computed K-values. The maximum variation was 6.7% and the average for the seven tests was 2.26 per cent.
The yield point of the beams in both free-ended and fixed-ended tests was
determined by a study of the torque-twist diagram and in some cases by a
drop of the beam: In the free-ended tests the distribution of shearing stress
across the flange was studied by measuring, with tensometers, the principal
stresS at an angle of 45 around the center section.
The fixed-ended tests provided a means of studying the additional rigidity
and strength' over the free-ended tests, due to fixing the ends of the beams,
Twenty-two various lengths and sizes of beams were tested ranging from
.r.
...
April,1935
'.
STRUCTURAL BEAMs iN
Tofisi6~
a 3-in. I-beam at 7.5 Ib per ft, to a 10 by 12-in. beam at 62 Ib per ft. The
heaviest beam had a K-factor fifty-seven times greater than the least, and with
ends fixed the most rigid fixed-end beam hada rigidity three hundred and ten
times as great as the least rigid with ends fixed.
The distribution of longitudinal direct stresses in the extreme fibers of the
flanges'was studied by strain measurements taken along the outer edges of
the flanges by tensometers. The total shearing stress distribution at the
center section was studied by measurement of principal strains at an angle
of 45 degrees.
The fixed-ended tests furnished information as to the proper design of
welded end connections for high torsional rigidity.
The tensometers in both the free~ended and fixed-ended tests were of value
in studying the relative distribution of stress. Some of the results are
erratic. The warping of the section during twist made it difficult to obtain
a steady set-up for the tensometers, but in most cases the test results checked
fairly well with the formulas. In spite of variations in result and incomplete
end fixity it is noted that every beam tested would have been amply strong
if designed on a basis. of working longitudinal stresses at the ends.
The present investigation has covered, accurately, the question of torsional
rigidity and the evaluation of the torsion constant. The formulas for
stresses which are proposed are not exact but will be satisfactory for practical
design purposes.
.
,~.
ACKNOWLEDGMENT
~.
. ,
The torsional investigation was conducted as a co-operative project,the

:M:cClintic-Marshall Corporation, subsidiary to the Bethlehem Steel Company,
furnishing the steel beams and the material for constructing the test apparatus, and Lehigh University bearing all other expenses.
Through the courtesy of the Baldwin-Southwark Corporation, of Philadelphia, Pa., sixteen tensometers were loaned to the University (without
charge), during these tests.
Special acknowledgment is due Jonathan Jones, M. Am. Soc. C. E.,
Chief Engineer; C.H. Mercer, 1L Am. Soc. C. E., Consulting Engineer, and
Sterling Johnston, 11:. Am. Soc. C. E., Engineer, all of the McClinticMarshall CorpQration, and V. E. Ellstrom, Manager of Sales Engineering,
Bethlehem Steel Company, for their co-operation in the investigation.
The study was conducted as a regular research .project of the Fritz Engineering Laboratory, Lehigh University. C. C. Keyser, Laboratory Assistant,
and all others associated with the Laboratory contributed valuable help
throughout. Professor J. B. Reynolds, of the Department of Mathematics,
gave his assistance to the theoretical studies.
504
Papers
APPENDIX I
LIST OF REFERENCES
(1) Navier. Resume des Lecons. Third Edition, with Notes and Appendices
by Saint Venant. (Article 5 on "Torsion" presents Saint Venant's
studies in complete form.)
(2) Prandtl, L. Zur Torsion von Prismatischen Staben. Physikalische
Zeitschrift, IV, 1903, p. 758. (Brief presentation by Prandtl of the
membrani! analogy.)
(3) Griffith, A. A., and Taylor, G. I. Technical Reports of the Advisory
Committee for Aeronautics, No. 333, June, 1917. (Report covering
the experimental work with soap films by Messrs. Griffith and Taylor.)
(4) Trayer, George W., and March, H. W. The Torsion of Members Having Sections Oommon in Aircraft Oonstruction. Technical Report
of the Advisory Committee for Aeronautics, No. 334. (Reports on
the application of soap film tests applied to torsional rigidity of
wood airplane struts. Has an excellent bibliography on torsion and
summarizes formulas for numerous shapes.)
(5) Timoshenko. Theory of Elasticity. First Ediltion, 1934. (Chapter 9
presents a very complete study of all aspects pertaining to the
problem of pure torsion.)
(6) Timoshenko. Strength of Materials, Vol. 1.
(7) Sonntag, R. Zeitschrift fur Angewandte Mathematik und Mechanik,
1929, Vol. 9, p. 369. (A theoretical study of torsion with end
restraint at both ends.)
(8) Trefftz, E. Zeitschrift fur Angewandte Mathematik und Mechanik,
1922, Vol. 2, p. 263.
(9) Foppl, Aug., and Flippl, Ludwig. Drang und Zwang, Vol. 2, 1924.
(Complete study of all phases of torsion problem from ~ German
viewpoint.)
(10) Seely, Putnam, and Schwalbe. The Torsional Effeet of' Transverse
Bending Loads on Channel Beams. Bulletin No. 211, University
of Illinois. (Gives a theoretical discussion of the problem of channels as cantilever beams fixed at one end.)
(11) Campbell. Torsion Tests Made at Northwestern University. Engineering News-Record, Vol. 101, 1928, p. 154.
(12) Nadai. Plasticity. 1931. (Ohapters 19 and 20 present the problem
of torsion after Ithe material has yielded and during which plastic
flow occurs.)
_(13) Cushman, P. Allerton. Shearing Stresses in Torsion and Bending by
Membrane Analogy. (Presented before the American Society of
Mechanical Engineers in June, 1932.)
C
.", .
Aprit 1935
505
APPENDIX II
NOTATION
The following symbols, adopted for use in this paper, are presented as a
guide to' discussers:
a
b
d
e
h
i
m
n
p
r
s
a torsional factor in fixed-ended beams.

length of a rectangular section.
depth; total depth of beam.
eccentricity.
a subscript denoting "flange."
distance .between flange centroids.
= ~ (d - h).
major flange thickness.
= breadth
of a rectangular section; also, where defined, n = minor
flange thickness.
per unit area;
= pressure
radius of a fillet (authors' duplicate); also, where defined,
r = .variable thickness of a section.
between stiffening plates.
= distance
thickness; as a subscript, t, denotes "due to bending."
a substitution factor in Equation (37); as a subscript, u,

denotes "free ended."
'w
web thickness; as a subscript, w, denotes "web."
y
deflection.
area; cross-section area; also, where specifically' defined, A, B,
A
0, and D are coefficients in a general equation; as a subscript, A, denotes "average."
a substitution factor in Equation (23) (see, also, Symbol A).
B
constant; torsion constant equivalent to K for a fixed0
ended beam; 0 0
0 at the center; OA
average value
of 0 (see, also, Symbol A).
D
diameter; 'diameter of an inscribed circle (see, also, Symbol A).
elasticity; modulus of elasticity in tension and compression.
E
torsion stress function..
F
G
shearing modulus of elasticity.
I
rectangular moment of inertia; IIII and 111
I for a crosssection with respect to the X-axis and the Y-axis, respectively.
J.= polar moment of inertia.
J(
a torsion constant; K f and K w
values of K for flange and
web, respectively. .
L
length; as a subscript, L, denotes "large end."
bending moment.
M
total shear over a cross-section.
Q
u
=
=
....
total slope of a flange section
m-n
b
as a subscript,
S, denotes "small end."

T = torque; torsional moment; T a
torque required to twist 'a
beam in a free-ended condition.
506
Papers
= a factor depending on the .!!.- -ratio, but practically constant

n
for
.!!.n
greater than 3; V Land V s
= value of V for large end
and smail 'end, respectively, of the flange,

a factor that depends on two ratios, ~ and ~"
Cl
e
(J
po
p
cr
= unit elongation.
of twist, in radians per unit length;

at the center.
= angle
ratio.
= Poisson's
radius of curvature.
= unit
stress; normal stress; crt = longitudinal fiber stress in a
(Jc
(J
= flange due to bending.

unit shearing stress; "f = " for flange seCtions;
" = sections.
.
't"w
= " for web
total angle change between two cross-sections.
0/
APPENDIX III
TABLES
The tables presented herewith are introduced and adequately discussed

in the text of this paper.
TABLE l.-GENERAL SUMMARY OF ALL TESTS
..
Test
No.
WEIGHT, IN
M.EABURED DIMENSIONS, IN INCHES, ADJUSTED TO
POUNDS PER
FOOT
THE ACTUAL WEIGHT
~1l
I:en~th,
Naminal size
mm.. .
Nom- Actual
choo
inal
(1)
(2)
T-4'
T-5'
T-6'
T-7'
T-8'
T-9'
TIO
T-11'
T-12'
T-13'
T-14t
T-15'
T-16'
T-17'
T-18'
T-19'
T20
T-21'
T-22t
T:23'
T-24'
T25t
3-in. I-beam .......

3-in. [-beam .......
3-in. [-beam .......
3-in. [-beam .......
3-in. [-beam .......
3-in. [-beam .......
3-in. [-beam .......
3-in. [-beam, ......
3-in. [-beam .......
3-in. [-beam .......
6 by 6-in...........
6 by 6-in ..........
6 by 6-in ..........
6 by 6-in ..........
6 by IO-in .........
8 by 8-in ..........
8 by 8-in ..........
8 by 8-in ..........
8 by 8-in ..........
8 by 8-in ..........
10 by 12-in ... : ....
10 by 12-in ........
12-in. [-beam ......
12-in. I-beam ......
12-in. [-beam ......
12-in. [-beam ......
12-in. I-beam ......
12 by 12-in ........
12 by 12-in ........
T-26t
T-27'
T-28'
T-29'
T-3ot
T-31t
N3t
FIXed end.
Total'
depth, Length,
,b
d
Weh
thickness,
(4)
(5)
(6)
(7)
(8)
(9)
(10)
7.4
7.4
7.4
7.4
7.4
7.4
7.4
7.4
7.4
7.4
19.6
19.6
20.9
39.0
38.1
29.4
29.4
29.4
29.4
66.1
60.9
60.9
31.3
53.6
53.6
53.6
53.6
186.7
65.6
3.2
6.0
9.0
12.1
18.1
23.9
39.0
53.9
5.5
.8.8
72
72
72
72
72
18
36
72
72
72
72
72
72
18
36
72
72
72
2.98
2.98
2.98
2.98
.2.98
2.98
2.98
2.98
2.98
2.98
6.06
6.06
6.08
6.73
'6.20
8.05
8.05
8.05
8.05
9.05
'10.05
10.05
12.06
12.07
12.07
12.07
12.07 .
14.39
12.14
2.51
2.5[
2.5[
2.51
2.51
2.51
2.51
2.51
2.51
2.51
6.01
6.01
6.06
6.25
9.91
8.04
8.04
8.04
8.04
8.29
11.92
11.92
4.97
5.74
5.74
5.74
5.74'
12.56
12.03
0.342
0.342
0.342
0.342
0.342
0.342
0.342
0.342
0.342
0.342
0.257
0.257
0.303
0.501
0.431
0.290
0.290
0.290
0.290
.0.606
0.390
0.390
'0.350
0.730
0.730
0.730
0.730
1.069
0.409
0342
0.342
0.342
0.342
0.342
0.342
0.342
0.342
0.342
0.342
0:162
0:162
0.162
0.162
0.162
0.162
0'.162
0162
0.162
0.162
0.362
0:362
0.372
0.699
0.447
0.401
0.401
0.401
0.401
0.907
0.546
0.546
0.345
0'.449
0.449
0.449
0.449
1. 714
0.604
.....
.....
.....
.....
.....
.....
.....
.....
.....
.....
0.659
0.659
0.705
0.900
0.900
0.900
0.900
.....
.....
~ a.S
'i. 9.S
(11)
1rl
0
,,~
~"O..!f
.~
Eo<
gj~.~
(3)
72
i!J
~:9-5
Flange
thicknI'BS.
7.5
7.5
7.5.
7.5
7.5
7.5
7.5
7.5
7.5
7.5
20
20
20
40.5
40
31
31
31
31
67
62
62
31.8
65
55
55
55
190
65
-- --- -- ----- -
. t Free end.
Ii
(SEE FIG. 6)
(12)
(13)
-0.27- --0.532 0.0865

0.27
0.27
0:27.
0.27
0.27
0.27
0.27
0.27
0.27
0.28
0.28
0.24
0.35
0.35
0.41
0.41
0.41
0.41
0.42
0.45
0.45
0.42
0.54
0.54
0.54
0.54
0.62
0.58
0.532 0.0865
0.532 0.0865
0.532 0.0865
0.532 0.0865
0.532 0.0865
0.532 0.0865
0.532 0.0865
0.532 0.0865
0.532 0.0865
0.566 0.243
0.566 0.243
0.581 0.281
0.928 1. 738
0.735 0.802
0.676 0.463
0.676 0.463
0.676 0.463
0.676 0.463
1.208 4.912
0.919 2.101
0.919 2.101
0.882 0.833
1.244 3.318
1.244 3.318
1.244 3.318
1.244 3.318
2.186 48.437
0.953 2.227
.,.
....
April,1935
507
..
TABLE 2.-PHYSICAL TESTS OF MATERIAL IN TEST BEAMS
SHEARING PROPERTIES, IN POUNDS
PER SQUARE INCH

Samples;
taken
from
Test
Beam
No.
Stresses. in Pounds per

Square Inch
Shearing
modulua,
Modulus
of elasticity
G,in
thousands
of pounds
Appsrent per squars
shearing
Inch
ultimate
E,in
(1)
Ul'per
YIeld
point
Lower
yield
point
At
ultimate
(2)
(3)
(4)
38 900
thousands
of pounds
per square
inch
(5)
63550
29 100
32.0
41000
39 710
43 020
34 180
42 880
34890
40860
38310
32380
32910
37 860
35 790
37000
31 570
38290
31 580
35 530
33 900
30990
28 620
60450
58 130
60 220
56 570
61 280
61 180
59 600
56 330
59 850
60000
29200
29 150
28 500
28250
29 450
28 600
29 150
29 400
29400
29200
35.8
35.5
37.0
40.0
36.0
38.8
36 3
38.0
34.5
38.0
Average.. 38 090
34 110
59 740
29040
62.1
21800
52850
TABLE 3.-TESTS ON FREE-ENDED B:EAMS

TORSION
CONSTANT. K
SHEARING STRESSES: IIi POUNOS PER SQUARE INCH
.~"
Test No.
.. ~a;
;:::;0
Nominal
size
10'"
.Hl
""is
Z'"
(I)
(2)
T-I4.. .....
T-22 .. .....
T-25 . ...
T-26 .. .....
T-30 ..... ..
T-31 .... ...
T-33 .......
6 by 6-in .....
8 by 8-in .....
10 by 12-in ...
,12-in I-beam ..
12-in. I-beam..
.12 by 12-in.:.
12 by 12-in ...
Average ..
............
1l
U
8~
Ii:
1j,,,
,,0
l>.
P'I
(3)
(4)
20
31
62
31.8
55
190
65
0.243
0.463
2.101
0.833
3.318
48.440
2.227
....
......
. .....
:~r~
Yield
point
of
beam,
in inchpounds
!~
.'
Maximum, at the Yield Point
In the Flange
By
Equation
(25)
By
ten8ometera
In the Web
By
Equation
(27)
By
teu80m..
eters
--- - - - - - - - - - - - --- - - - - - - ~
(5)
(6)
(7)
(8)
(9)
(10)
0.243
0.0 15900
0.451 -2.6 25 610
1.960 -6.7 50 900
0.804 -3.5 25 000
3.380' +1.9 64000
48'870 ' +0.9 480 000
2.222 -{).2 76400
30 400
29 800
19 100
23 800
20 700
19300
26600
29 700
31 400
23 400
25400
20200
21 400
29300
22 400
22 600
12 600
14900
17 100
12400
19 700
22 200
17 900
17 700
18500
12 600
......
24240
25 830
17000
18 100
:S
~~
-'"
,Sm
8._
"",0
, (12)
--26000
23 900
24050
23 000
19000
20 750
......
. .....
._- - - - - - - - - --- - - - - - - ---'
- - - --2.26
22 780
508
Papers
TABLE 4.-TESTS OF FIXED-ENDED BEAMS
...,
Test No,
'"'~ib
Nominal
size
~:::l..e
c;
.S
8. fij
Q Q.,
S'~
(2)
(3)
T -4 ......
T -5 ......
T -6 ......
T -7 .....
T -8 ......
T -9 ......
T-10 ......
T-ll ......
T-12 ......
T-13 ......
T-15 .... ..
T-16 ......
T-17 ......
T-18 ......
T-19 ......
T-20 ......
T-21 ......
T-23 ......
T-24 ......
T-27 ......
T-28 ......
T-29 ......
3-in. I-beam ..
3-in. I-beam ..
3-in. I-beam ..
3-in. I-beam ..
3-in. I-beam ..
3-in. I-beam ..
3-in. I-beam ..
3-in. I-beam ..
3-in. I-beam ..
3-in. I-beam ..
6by6-in .....
6 by 6-in .....
6 by 6-in .....
6 by 10-in ....
8 by 8-in .....
8by 8-in.: ...
8 by 8-in .....
8by8-in .....
10 by 12-in ...
12-in. I-beam.
12-in. I-beam.
12-in. I-beam.
7.5
7.5
7.5
7.5
7.5
7.5
7.5
7.5
7.5
7.5
20
20
40.5
40
31
31
31
67
62
55
65
55
~~]g
Q)
O.~
gsg
S.
E"'Cf
.......
0
'51
S.9
<l-
H
(4)
] g.9
Q.l
::s
-3.2
-
E-<
(5)
0.57
6.0
0.57
9.0
0.57
12.1
0.57
18.1
0.57
23.9
0.57
39.0
0.57
53.9
0.57
5.5
0.57
8.8
0.57
72.0
13.1
36.0* 13.8
72.0
28.5
72.0
72.5
18.0
34.8
36.0
34.8
72.0
34.8
72.0
86.2
72.0 162.7
18.0
18.9
36.0
18.9
72.0
18.9
TORSION CONSTANTS
il~l:.::
a~~
~.~
~~ ::1l
0,.d
ls S~
~.9
blJ
<ll
Z
(1)
~
4"g
i~
~~
" '"'"
<ll
'S
:B
;:l
;;
:>
00
""
Values
of
ratio,
I
2a
CA
Ce,
computed
by
ComEqua- puted By test
tion (46)
(6)
(7)
(8)
(9)
(10)
(11)
0.0865
0.0865
0.0865
0.0865
0.0865
0.0865
0.0865
0.0865
0.0865
0.0865
0.2433
0.2810
1. 7375
0.8018
0.4631
0.4631
0.4631
4.9122
2.101
3.318
3.318
3.318
5.67
5.67
5.67
5.67
5.67
5.67
5.67
5.67
5.67
5.67
33.7
32:4
19.7
44.1
53.5
53.5
53.5
27.4
67.0
21.8
21.8
21.8
0.282
0.529
0.794
1.067
1.596
2.108
3.439
4.753
0.485
0.776
1.068
0.556
1.872
0.816
0.168
0.336
0.673
1.314
0.537
0.413
0.826
1.651
0.7998
0.4558
0.2823
0.1996
0.1341
0.1102
0.0913
0.0875
0.5004
0.2904
0.6186
1.896
2.431
2.917
20.927
7.496
2.346
9.479
15.104
32.059
11.636
5.1675
0.943
0.605
0.394
0.284
0.190
0.154
0.120
0.109
0.654
0.404
0.905
2.761
3.445
4.264
27.55
10.75
3.45
13.67
22.03
44.30
16.89
7.39
0.787
0.474
0.330
0.258
0.174
0.153
0.116
0.111
0.534
0.377
0.777
1.106
3.333
3.944
27.90
7.43
3.30
12.00
10.02
34.50
11.23
6.58
---- - --- ---------
UNIT STRESSES, IN POUNDS PER SQUARE INCH
Test No.
Yield
End
point
Ma::r"imum
fixity
of beam,
direct
(percent- by test in stress at
ages)
inchend, by
pounds Equation
(40)
Direct
stress
near end
by ten80metere
(15)
(1)
(12)
(13)
(14)
T -4 ......
T -5 ......
T .,6 . . . . . .
T -7 ......
T -8 ......
T -9 ......
T-10 ......
T-ll ......
1'-12 ......
1'-13 .....
T-15 ......
T-16 ......
T-17 ......
T-18 ......
T-19 ......
T-20 ......
T-21 ......
1'-23 ......
T-24 .... ..
T-27 ... . ,.
T-28 ......
1'-29 .......
83.5
78.3
83.8
90.8
91.6
99.4
96.8
101.8
84.2
93.4
85.8
40.1
96.8
92.5
101.0
69.1
95.7
87.8
45.5
77.9
66.6
89.0
26000
16000
10000
8500
6 500
6 160
5 500
5400
8000t
6000t
33 980
35 000
79 000
69000
151 600
115 000
65000
194 000
223 000
310 000
162 000
105 000
58700
70500
60 300
61 100
54 600
54500
50 100
49 300
32 900t
35 600t
72600
44 100
54 000
48700
40700
60000
61 600
53 300
57000
71 000
64300
57 100
Maximum Shearing Stress at the Yield Point

In the Flange
By
Equation
(47)
(16)
In the Web
By tenBometers
(17)
------ ----........ 30 500 . .. .. . . .

........ 24 500 ........
........ 20 360 . .......
........ 22 000
........
........ 22 800 ........
........ 25 400 . .......
. ....... 26 400 . . . ... ..
........ 27 000
........ 11 600t ........
........
'66' ioM
61 500t
40000
30300
49000
56700
56 300t
37 400
48 100
29400
68 500'
65800
12 OOOi
28000
11 100
27 800
17 000
13 000
14900
18 100
24000
16200
17 700
17900
22 700
'29'600'
23 200
29 000
16 700
25 800
19 100
19 500
24800
21 600
30 000
24 400
20200
36 in. = one-half total length on acconnt of additional stillener in middle.
By
Equation
(48)
(18)
By ten80mete1'8
(19)
........
........
. ... . ..
........
........
........
........
. .......
is'700'
'i9'700'
........
13 500
........
16 400
........
. .......
........
........ . .... .. .
........ . .......
........ ........
. ....... . .......
........
........ ........
........ ........
i5' 000'
7800
8600
12 400
18 100
14 500
13 450
5920
13 800
23 740
'f'
t Not yield point.
....
AMERICAN SOCIETY OF CIVIL ENGINEERS

Founded November 5. 1852
DISCUSSIONS
~"I

Discussion
By
MESSRS.
H.
M. WESTERGAARD AND R.
JOSEPH
B.
D.
MINDLIN, AND
REYNOLDS
H. lL WESTERGAARD,'" M. AM. Sao. C. K, and R. D. MINDLIN," JUN. AnI.

Sao. C. E. (by letter)."a-In Fig. 9 the authors show curves which represent
different formulas for the concentration of torsional shearing stresses at fillets.
The curve which is identified by the names of the writers was derived as a
result of correspondence with the authors, and for this reason its derivation
is given herein.
The process of approximate analysis is closely related to those used by
A. and L. FopplH and by S. Timoshenko,15 but differs in the following respects:
x
i-
PIG.
27.~I-BI:HI
IN TORSIO:'>.
FIG.
2S.-SIIEARING STRESSES
SECTION 1-2, IN FIG. 27.
AT
First, consideration of the influence of the straight edge of the cross-sectio~l

opposite the fillet led to a slightly greater concentratiqn factor near the beginning of the fillet, at Point 1 in Fig. 27, than that obtained by Timoshenko's
NOTE.-The paper by Inge Lyse, 1\1. Am. Soc. C. E .. and Bruce G. Johnston, Jun.
So~. C. ~., is published in this number of Proceedings.
This discussion Is printed In
Proceedtngs III order that the views expressed may be brought before all members for
further discussion.
12 Prof. of Theoretical and Applied Mechanics. Univ. of Illinois, Urbana, Illinois.
13 Research Asst. Dept. of Civ. Eng., Columbia. Univ., New York, N. Y.
3
1 a Received by the Secretan' March 1, Ill35.
14 "Drang und Z,vang," by A. and L. Foppl, Vol. 2, Second Edition, IV28, p. 73.
""Theory of Elasticity," by S. Timoshenko, 1934, p. 259.
Am.
510
WESTERGAARD AND MINDLIN ON STRUCTURAL BEAMS. IN TORSION
Discussions
formula; and, second, the writers had before them the detailed data obtained
by P. A. Oushman' in tests with soap films. These data show clearly that
the maximum shearing stress occurs at a point such as ''In in Fig. 27, and
that there is "a notable increase of stress from Point 1 to Point 'In. Special
consideration was given to this increase.
Let 'ro denote the shearing stress that would exist at Point 1 in Fig. 27
if the edge of the crosscsection at that point were straightened out by moving
the beginning of the fillet toward the left, when, at the same time, the thickness of the flange, t, the torsion factor, J{, and the total twisting moment, T,
are left unchanged. The stress, 'ro is defined by the formula:
'ro
= -Tt
Let 'r1 and 'rm denote the actual shearing stresses at Points 1 and
Then
..
(57)
the maximum stress at the fillet.
.....
'In,
'rm
being
.22' is the concentration factor at

To
Point 1 and
Tm
is the desired concentration factor for the fillet.
To
Prandtl's soap .film analogy, which the authors have used advantageously,
furnishes the key to the solution. It is noted that the soap film is stretched
over an opening shaped like the 'cross-section and is inflated a small amount
by an excess air pressure on one side. The shearing stresses on the original
cross-section follow the contour lines on the film, the edge being one of the
contour lines. Furthermore, the shearing stresses are proportional to
the slopes of the film, or proportional to the density of the contour lines.
The contour lines at Section 1-2 in Fig. 27 must be approximately perpendicular to that section. Accordingly, the shearing stresses, 'r, at that section
are approximately in the direction of y. Since the slope of the film is C'r,
c being a constant, the curvature of the film at Section 1-2 in the direction
dT
of x becomes c - .. The curvature of the film in the direction of y at Secdx
tion 1-2 is accounted for by the curving of the contour lines. If the radius of .
curvature of the contour lines at a particular point is R, then the curvature
of the film in the direction of y at the same point becomes c.:!.-.
The curva-
ture of the surface is the sum of the curvatures in the directions of x and y;
+ .:!.-).
that is, c (dT

dx
R
The equilibrium of tl,1e film requires that this combined curvature be
constant. Since R
r at Point 1 and R
00 at Point 2, it follows that,
dT ] +.22
[ dx
1
r
=[ ~J
dx
(58)
2
10 "Shearing- Stresses in Tor8ion and Bending' by l\Iembrane Annlogy," by P. A. Cushman, Doctoral Dissertation, Dniv. of Michigan, 1932.
; 1
April, 1935
WESTERGAARD AND MINDLIN ON STRUCTURAL BEAMS IN TORSION
511
Since Points 1 and 2 are on the same contour line, it is also required
that,
dx = 0
(59)
The diagram of the shearing stresses at Section 1-2 must be sh?ped about
as shown in Fig. 28. Equation (60) has been constructed so that it satisfies
this general requirement of shape as well as the specific requirements in
Equations (58) and (59), and, therefore, it may be assumed to represent
the shearing stress approximately:
T
Tl
[1 - (2 + .!..-) ~ + ~ ]
3r
2 tr
(60)
The constant curvature of the film may be computed as:

C
[dTJ =-~(1 _.!..-)
(~1)
dx 2
t
3r
If the fillet were some distance away, this curvature would remain the same,
"1 would be replaced by "0, and the term containing r would disappear.
Oonsequently,
To
= Tl
(1- tr)
3
(62)
which gives the concentration factor at Point 1,
......................... (63)
1-~
3r
The important indications of Equation (63) are preserved when the
following simpler formula is substituted:
~ =
To
+ .!. -
3r
(64)
As was mentioned, the maximum st.ress, "m, occurs at a point such as m in

Fig. 2'7. Equation (64) can be used to obtain an estimate of this stress by
making the following replacements: . t is replaced by a distance, tm, measured
along a curved section as drawn from m t~ n,' and "0 is replaced by a stress
corresponding to straight edges and a thickness, t m ; that is, according to
....
Equation (5'7), by
"0
t".. Thus, the concentration factor becomes,

t
:: =
t~ ( 1 + ~:)
(65)
A ~easonable estimate of tm is,
tm =t+0.3r
:(66)
512
REYNOLDS ON STRUOTURAL BEAMS
IN
TORSION'
Discussions
This value, substituted in Equation (65), leads directly to the formula represented graphically by the authors,
m
7
70
= 1.2+ ~ (..!..3
(67)
JOSEPH B. REYNOLDS,17 ESQ. (by letter).17a-The problem of the twisting

of I-beams has b~en studied thoroughly by the authors for special cases. It
is the purpose of this discussion to' compare the authors' results with a more
general theory for the twisting of I-beams under differing types of loading.
Only the effect of twisting is considered. The assumptions and the notation
are the same as those used in the paper.
Oonsider the requirements for equilibrium at a section of the beam at a
distance, x, from the fixed end as shown in Fig. 29. The inner moments at
...
.,
FIG. 29.
this section, together with the moment of the shears, must balance the outer
moment applied to the beam to the right of the section. This outer twisting moment, Ta;,will vary in general with the position of the. section; that
is, with x. For equilibrium at this section one must have:
Tt
+ Tw +
Q h = Ta;
(68)
in which Tt, T w , and Ta; indicate twisting moments of flanges, web, a~d
total of section, respectively.
In terms of the co-ordinates of points on the neutral axis, y =. I(x), the
following approximate relation may be written:
1
h de
dy
17 Prof. of Math. and Theoretical Mechanics, Lehigh Univ., Bethlehem. Pa.

N'Recelved by the Secretary March 16, 1935.
(69)
0.-
.....
April,1935
513
REYNOLDS ON STRUOTURAL BEAMS IN TORSION
and the twist per unit length is d8 , hence:
..
Tf
dx
Tw
= KG de =2KGdy
dx
(70)
hdx
Fr0m the relation between shear and bending moment:
~ E v I d"
y
dx'
=-
(71)
.
By means of Equations (70) and (71) and the value of a, Equation (68)
can be written :
3
a2
.
In
h X
w h'lC,
d y _
dx 3
3JL +
dx
=0
(72)
h Tz
= --.
2KG
The differential in Equation (72), has the general solution:

Y
=A
sinh~+ B cosh~ + C + Y,
a
a
(73)
in which y, is the particular solution satisfying Equation (72) and the

values of A, B, and C are determined by the boundary conditions for
the beam. Equation (73) is the same as the authors' Equation (34) for the
case considered. The value of y, is:
Yl=
1
:"j-~
1
-:"j:"
a eaXdx-zae
a ea Xdx .. (74)
Xdx-zae
For the usual types of loading, X can be written in the form:
= Xo +
+ X n xn
X, x + X. x' +
(75)
in which, X o , X" X.,

. . X n , are constants. When Equation (75) holds
successive integration by parts shows that Equation (74) may be written
in the form:
Yl=jxdx+a3(dX
dx
+ a2
3
d X
dx 3
+ a4
dsX
dx s
+ ....) .... (76)

n 1
If n is even in Equation (75) the last term in Equation (76) is d - X; if n

.
dx n - 1
n
is odd, the last term is d X. With the vallie of y, thus determined one may,
dx n
by Equation (73) and the relation,]{
'II
= ~2
EI d Y , write:
~2
= 2 K G[A sinh~ + B cosh~ + a2(dX + a3d" X3 + asds X,

s
h2
a.
dx
dx
...
)].(77)
514
Discussions
REYNOLDS ON STRUCTURAL BEAMS IN TORSION
and, since the shear is given)Jy:Q
= ~
EI
~~
. Q" = 2 KG (A cos h -x +. B'

d X +. a6 d X +. . .. ) . .(78)
smh -x +. a3 2
2
ah
dx
dx 4
The longitudinal stresses along the outer fibers of the flange will be given
by a
(j
Mb
-.-"
or by:
Iu
Eb (A sm
. h -x
= --.
2a
+. B cosh -x+. a2 d X+.

d' X+. a7 d- X +.
- a6 6
dx
dx'
and the lateral shearing stresses in the flanges by

2
Qb 2
't
. .) .. (79)
dx
= --, or by:
4Iu
4
Eb- ( A cosh -X +. B smh
dX +. a
6 d X
= -X +. a3 -+
. .. .). .. (80)
2
3
8a
a
a
dx
dx 4
.
The angle through which the beam twists comes from the relation,
(J
= 2 Y , and has the vahe:
B
x + . -2.C- + . 2O=2A sinh -x + . -2c
osh. h
a
h
a
h
h
2 a' X dX
Xdx+.-h
dx
2 a6
d' X
2 a7
d6 X
............. (81)
+'-h- X - 3+ ' - X - + '
dx
h
dx6
,;
In this manner, general values are derived for the principal variables of
interest in the case of twisted beams. The maximum values of these quantities will occur for differing values of x, depending upon the conditions surrounding the strained' beam. Example (a) demonstrates that Equations
(73) to (81) reduce to those given by the authors when the proper limitations
are applied.
.
Ex~mple (a).-Beam Twisted by Constant Torque, T, Applied at Its
Ends, with Both Ends Restrained.-In this case, T 1lJ. = T, a constant, and,
therefore, X
'r
hT
2KG
The constants A, B, and Care detE(fmined by
the requirements that for x

for x
= -
d2 y
2"' dx 2
tion (76) yields YI
o.
0, y
0, and
dy
dx
0.
:!f,urthermore,
Since the derivations of X are all zero, EquaIT
= 2KG
h T x and, py. Equation
.
(73):
T h a (COSh -=:.. tanh _1_ _ sinh -=:.. +. -=:.. _ tanh ._1_).. . (82)
2KG
a
2a
a
a
.
2a
which is the same as that presented by the authors in Equation (35). The
maximum displacement, Ym, occurs where x
1 and has the value given in
the authors' Equatiop (36).
..
.
1
April,1935
515
Similarly, Equation (77) reduces to the authors' E.quation (37) for the
......
moment in each flange, which has (where u
-., or x = 0), a maximum
value:
Nfm =
-Ta
h
l .
tanh 2 a
(83)
Equation (7S) reduces to Equation (3S) and Equation (7.9) becomes Equation (39).
By use of Equation (SO):
T
Tb'
cosh u
4Iv
cosh
2 a
(S4a)
and,
T m ""
= -T ~ .'"
4Iy
.
(S4b)
'" '"
which corresponds to the authors' Equation (42), and by Equation (Sl):

(J
"
Ta
l
l-)
= - ( cosh -x
tanh - -.
smh -x + -X
_ tanh
- ... (S5)
GK
a . 2a
a
a
2a
from which,
.(Jm""
= ;;(~ -
2 tanh 2
a)-
..
(S6)
Example (b).-Beam with One End Fixed, the Other Free, Under Con.
stant Torque, T.-In this case, X
hT
= ---
as in Example (a).
The
2KG
conditions determining A, B, and C are that for x = 0, y = 0, and
dy
: = 0,
dx
d'y_
_
and for x - l, - - - 0. The value of the deflection obtained is:
dx'
:. T-ha ( cosh -tan

x
h -l - s m
. h -x + -x- t a n1l l) .... (S7)
y=
2KG
From this all the other principal variables are readily found as in
Example (a).
Example (c).-Beam with Both Ends Fixed and a Uniform Eccentric
Load Along Its Length Producing an External Moment, T.-In this case,
- J
....
Tx
(1 - ~) ,
l
and, X
-.;,
Yl=
=~
2K G
2~TG (x
(1 _
~) giving,
_ ~') _ ~~~
(88)
516
. Discus~ions
The values of A, B, and 0 are determined by the requirements that for

x
dy
= 0, y = 0, and --,
= 0, and or x =
dx
-.!:JL = 0.
The deflection
dx
T ha ( cosh u _ coth _1_

2 J{ G sinh _1_
2a
+ ~ _ ~) ....... (89)
a
.. ,
,~
found is:
a1
2~
Example (d).-Beam with One End Fixed and the Other Free, Uniform
Eccentric Load Along Its Length.-Here, Tx

Yl = 2
h T (
J{
x - 21
(1 - ;)
Tha
and,
2 J{ G 1
The constants, A, B, and 0, are determined by the requirements that
= 0,
for x
= 0,
and dy
dx
= 0,
and for x
= 1,
d Y
dx 2
= 0.
The deflection
proves to be:
x)
T ha
2J{G [
'h(l
sm
--a 1a
cosha
_ tanh i.- + ~ sech i.- (COSh ~ a

1
a
a
+: -
2:
1)
~
(90)
It is thus shown that the formulas developed and substantiated by test

results in the authors' investigation, are identical with those obtained by a
general mathematical treatment of the subject of torsional resistance.
,.
'- .. -
~"
LEHI,GH UNIVERSITY PUBUCATION

......
;-"-5
,(,
(List continued from inside front cover)

Circular
roto. '
Title
61. A Study of Tannery Effiuent. 10 cents.
62. Hydration of Animal Skin by Volume-Change Method. 10 cents.
063. A Study of the Liming and Depilation of Animal Skin. 10 cents.
64. Tests of Reinforced Concrete Columns. 50 cents.
065. Studies in Colonial Connecticuf 'l'axation. 20 cents.
66. The Age of Roman Sacrificial Victims. 10 cents.
67. X-Ray Analysis of Slate. 10 cents.
68. A Critical Study of Pickling. 20 cents.
69. Gamma Ray Testing of 'Velds. 10 cents.
70. Silver__A Senate Racket. 10 cents.
071. Scientific Papers from the Department' of Mathematics. 20 cents.
072. X-Ray Studies of Animal Skin. I. Structure of Calf Skin as 1ndicatedby Molybdenum Radiation. 10 cents.
073. Scientific Papers from Department of Physics for the Year 1930. 10
cents.
074. Tests of Concrete Conveyed from a Central Mixing Plant. 10 cents.,
075. Arc Discharge Not Obtained ,in Pure Argon Gas. 10 cents.'
,
076. ,Tower-Absorption Coefficients-IV. 10 c e n t s . '
077. An X-ray Study of the Coagulation of Egg Albumin. 10 cettts.
078. A Proof of the Second' Law of '.!:'hermodynamics. 10 cents.
79. Studies In Philosophy. 10 cents. '
080: Some'Physical,Properties of Pen'risylvanla Anthracite and Related
Materials; 10 cents.
"
081. Third and Fourth Progress Reports on Column Tests at Lehigh Uni.
verslty. 20 cents.
'
82. Studies In the Drying Oils. 10 cents.
083. Tests of Riveted and Welded Steel Columns. 15 cents.
084. Correlation by Ordovician Bentonite. 10 cents.
085. Forces of Electric Origin In the Electric Arc. 10 cents.
086. Electric Welding and Electrk"Welds. 20 cents.
087. Synchronous-Mechanical Rectifier-Inverter. 20 cents.
088. Scientific Papers from Department of Physics for the Year 1931-33.
15 cents.
'
089. Simplifying Design and" Control of Concrete. 10 cents.
090. Scientific Papers from Department of Mathematics. 20 cents.
91. A Critical Study ()f Pickling, Parts II, III, and IV. 20 cents.
92. A New Surface, Tension Balance.: 10 cents.
'
093. X-Ray Study of the Transformation of Marcasite into Pyrite. 10
cents.
094. Studies In Hydrated Salts Systems. 10 cents.
95. Social Aspects of Commercial Banking Theory. 10 cents.
096. Relati()n Between Quality and Economy of C()ncrete. ,10 cents.
097. Studies in the Drying Oils. 10 cen t s.
098. Dielectric Losses 'in Rocksalt and the Raman Spectra of Three
Amlnes. 10 cents.
099. Tests of Reinforced Brick Columlls. 10 cents.
0100. An' Unsolved Problem In City Government: Political Leadership
and Expert Administration. 10 cents.
0101. Two Studies In Mathematics. 20 ce,nts.
0102~ Arcs in Inert Gases, II. 10 cents. "
0,103. "Propagation' of Elastic Waves In Ice, I, II. 10 cents.
0104. Geology Reprints. 1~ cents.
0105.!The Indian Trust Funds, 1797-1865. 10 cents.
0106. A ,Study of Physical Properties of Steel. 20 cents.
0107. Tower Absorption COefficients. 10 cen' s.
0108. Crime and Its Punishment In P.rovlnci"l Pennsylvania. 10 cents.
0109. A ,New Method of Investigathig Performance of Bearing Metals.
10 cen,ts.
.,'
, .
110. PapllTS In the Philosophy of Science. 10 cents.
0111. StudJes in Chrome Liquors. 20 cents.
112. Charles McLean Andrews and the Re-orlentation of the Study of
American Colonial Hi,story.10 cents.
'
.113. Structural Beams In Torsion. ,20 cents.
Limited number of coples'avallable for tree distribution.
..

Structural Beams in Torsion Trans. ASCE Vol. 101 (1936)

Uploaded by

Copyright:

Available Formats

Structural Beams in Torsion Trans. ASCE Vol. 101 (1936)

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Structural Beams in Torsion Trans. ASCE Vol. 101 (1936)

Uploaded by

Copyright:

Available Formats

Lehigh University

Civil and Environmental Engineering

Structural beams in torsion, Trans. ASCE, Vol. 101

Follow this and additional works at: http://preserve.lehigh.edu/engr-civil-environmental-fritz-labreports

"LEHIGH UNIVERSITY PUBLICATION

THE INSTITUTE OF RESEARCH

Science and Technology, No. 94

STRUCTURAL BEAMS IN TORSION

AMERICAN SOCIETY OF CIVIL ENGINEERS

Results of a study of the torsional properties of standard structural steel

The problem of pure torsion as applied to non-circular sections was first

STRUCTURAL BEAMS IN TORSION

The investigation reported herein was undertaken as a study of all

STRUCTURAL BEAMS IN TORSION

Notation.-The symbols in this paper are introduced in the text as they

General Problem.-The solution of the torsional properties of a section

I~'- - - 3 " - - - ' I - t .

(e) STRUCTURAL BEAM SECTION

FIG. 2,-':'SECTIONS HAVING ApPROXIMATELY EQUAL

It is assumed that the lateral displacements are proportional to the angular

---1- ?:Jx2 .. ?:J'y2

STRUCTURAL BEAMS IN TORSION

In terms of T Equation (2) may

The torque, T, may also be expressed ip. terms of

A detailed description of the soap film studies is given in a thesis by Bruce G.

STRUCTURAL BEAMS IN TORSION

For non-circular sections the torsional resisting moment may again be

the breadth of a rectangular section; b

rectangular section; and V

praotically constant for

a factor depending upon the ratiO,.!, but

Fig. 3 shows the values of V for .! from

1 to 3. For.! -ratios greater than 3, V

SLOPING FLANGE SECTION

STRUCTURAL BEAMS IN TORSION

The Section with Sloping Sides.-Equation (8) provides a basis for

Evaluating r in terms of m and n, and integrating:

which m = major flange thickness and n

0.08480 S' - 0.06746 S3

0.05153 S' . . (12)

= 0.10504 + 0.10000 S + 0.084808' + 0.06746 sa + 0.05153 S' . . (13)

Torsion Constant for H-Beams and I-Beams.-The foregoing supplies a

"Theory of Elasticity," by A. E. H. Love, Fourth Edition, p. 319.

STRUOTURAL BEAMS IN TORSION

FIG. 5.-END CONSTANTS FOR K-VALUES OF FLANGES WITH

Trayer and March (4) in an investigation of aircraft strut sections

diameter of an inscribed circle; and

depends on two ratios, wand 2:...

STRUOTURAL BEAMS IN TORSION

(a) PARALLEL,SIDED FLANGE

i'i (Above) and iii: (Selow)

= 0.066 + 0.021.2. + 0.072.!.

and, for a slope of 1 on 50:

= 0.084 + 0.007 w + 0.0712:..

STRUCTURAL BEAMS IN TORSION

Values of V L and VB for standard slopes are, as follows:

and, for sections with parallel-sided flanges,

9.-lrORMULAS FOR STRESS CONCENTRATION IN FILLETS DUEl TO TORSION.

Diameter D may be determined by a large scale layout, or by the following

STRUOTURAL BEAMS IN .TORSION