Elements af

STRENGTH OF MATERIALS
Elements of

STRENGTH OF

S. TIMOSHENKO
Professor of Engineering Mechanics^ Emeritus
Stanford University
MATERIALS
FOURTH EDITION
( An East- West Edition

Affiliated East-West Press Private Ltd.

New Delhi
 D. VAN NOSTRANI3 COMPANY ,
INC.
120 Ak^xandt'r 8t., Princeton, New Jerst'y {Principal Office)*
24 West 40 Stret't, New York 18, New York

T> Vax Nostranu Company, 1/iu.

358, K*‘nsin^ton Hi^k Street, lA>ndon, \V 14, Kuultind

P). Van Nostrani) C\ympany' (Camida). Pt[>

25 HoliinRcr Road, Tc^ronto 10, Canada

Fourth Edition^ January 19SO

Published by \V. n. Ten Brorck for AFFIPIA r ED EAS'I -W ES 1'

PRE:SS private ltd., C 57 Defence C:olon^ New Delhi 3
, ,

India, and primed by S* M. Balsaver at CSHA PR IN FERS,

National Douse, T-'ulloch Road, Bomba\ i, India.
 PREFACE TO THE FOURTH EDITION

This textbook is an outgrowth of Timoshenko’s two-volume Strength of

Materials,first published in 1930. Whereas the two-\'olume edition presents

both elementary and advanced topics, the present volume is considerably

abridged and is designed primarily for undergraduate courses in elementary
strength of materials in American colleges and engineering schools

This fourth edition of Elements of Strength of Materials has beeiiVom-

pletely rewritten but in so doing an attempt has been made to retain the
same general approacli to the subje(*t that characterizes the original work.
This consists primarily in proceeding gradually from the simplest cases to
the m(?re complex ones and relying on physical and geometrical considera-
|,ions of deformation to estaV>bsh the patterns of stress distribution under

various types of loading. This, of course, characterizes the ‘’strength of

”
materials approach” as contrasted witli that of the ‘‘theory of elasticity
Such an approach may seem old-fashioned to .some, but the authors firmly
believe that, for the beginner, it repre^sents a sounder pedagogy. We must
ail learn to walk before we attempt to run.

In this edition, the subject with some rearrangement and

matter,
additions, remain.s essentially the same as
previous editions. Significant
in
changes and new features may be listed as follows: In the first ehapter,
the ideas of stress and strain within the elastic range of behavior are treated
thoroughly before introducing the complica lions associated with nonlinear
stress strain behavior. The second chapter begins with a discussion of the
stress conditions on an oblique section of a bar in tension in order that the
complete stress-strain diagram with proportional limit, yield point, ultimate
strength, etc,, may be better appreciated. This chapter also contains a new
section on Plastic Analysis or Limit Dcusign. ,

Throughout the book, the material on Strain Energy, pre^'iously confined

to one chapter has been distributed among several chapters in
at the end,
order to better integrate this material with the .st of the subject. Thus
Chapter II contains one section on strain energy of tension or compression;
Chapter IV, one on strain energy of torsion; and Chapter VIII, one on
strain energy of bending. This permits discussion of stress under dynamic
loads in parallel with that under static loads and gives more life and variety
to the* problems. Anyone preferring to postpone omit these sections on
V
VI PREFACE

strain energy can easily do so without loss of continuity in the rest of the

material.
Chapter III beginvS with a discussion of stresses in thin-walled pressure
vessels which serves to introduce the problem of biaxial stress. Analysis of
biaxial stress is then developed in detail and Mohr’s circle is introduced.
This leads logically to a discussion of pure shear which is essential to a
proper treatment of torsion as taken up in Chapter IV. To round out the
chapter on biaxial stress, the final section deals with the problem of riveted
and welded joints for pressure vessels. This material is greatly abridged in
the present edition, since it is felt to belong more properly to a design course.
In Chapters V and VI, the question of bending stresses and shearing
stresses in beams is taken up. This material, although somewhat rear-
ranged, is not greatly different from that in the third edition. Chapter
VII treats the problem of plane stress and the notion of principal stresses.
Applications to principal stresses in beams and vstresses due to combined
bending and torsion are fully treated. The chapter ends with sections on
the analysis of plane strain and the use of strain rosettes.
Chapter VIII is devoted to methods of calculating deflections of beams.
These include the differential equation of the elastic line, the moment-area
theorems, and the method of superposition. Statically indeterminate
beams are discussed in Chapter IX. Since the concept of strain energy
has been developed in earlier chapters, it is natural at this point to discuss
Castigliano’s theorem and its application to statically indeterminate
problems. This chapter ends with a new section on limit analysis of
statically indeterminate beams, using the concept of the plastic hinge.
Chapter X, dealing with the theory of columns, has been rewTitten so as to
emphasize the rational approach and minimize the attention given'' to
empirical column formulas. The text proper ends with a chapter on the
mechanical properties of materials. This treatment has been greatly
expanded from that in the third edition and many important new develop-
ments are included. It is hoped that the inclusion of such material in an
undergraduate textbook will give the student a better appreciation of the
importance of the experimental side of the subject of Strength of Materials.
Notations and sign conventions used in the third edition hav*^ been
changed in a number of instances in order to bring them into closer agree-
ment with accepted usage at the present time. This is most notable in the
use of (j and r for normal and shearing stress. Most of the problems are new,
at least to the extent of given numerical data, and answers are given to all
problems.

S. Timoshenko
 CONTENTS

PAGE

PREFACE TO THE FOURTH EDITION V

NOTATIONS IX
T. TENSION, COMPRESSION, AND SHEAR: I 1

Introduction Internal Force; Stress

. . . Elasticity; Strain . . . . . . Stati-
cally Indeterminate Problems in Tension and I"* mpression . . , Thin
Rings

11. TENSION, COMPRESSION, AND SHEAR: II 26

Variation of Stress with Aspect of Cross-Section . . . Stress-Strain
Diagrams; Working Stress . . . Limit Design Strain Energy in . . .

Tension and Compression . . . Stress Concentration in Tension or

Compression Members

III. BIAXIAL TENSION AND COMPRESSION 50

Stresses in Thin- Walled Pressure Vessels . , . Further Analysis of
Biaxial Stress Mohr’s Circle for Biaxial Stress
. . . Pure Shear . . . . . .

Riveted and Welded Joints in Pressure Vessels

IV. TORSION 70
• Torsion of a Circular Shaft Close-Coiled Helical Spring Strain . , . . . .

Ent-rgv" inShear and Torsion Torsion of Thin-Walled Tubes . . . . . .

Shaft of Rectangular or Profile Section

/ V. STRESSES IN BEAMS: I 95
Shearing Force and Bending Moment Shear and Bending Moment . . .

Diagrams Bending Stresses in Beams

. . . Various Shapes of Cross- . . .

Sections of Beams Shear Stresses in Bending . Stresses in Built-

. . . . .

up Beams
VI. STRESSES IN BEAMS. II 139
Plastic Bending of Beams . . , Beams of Two Materials . . . Reinforced
Concrete Beams . . Bending of Beams of Arbitrary Cross-Section . . ,

Shearing Stresses in Beams of Thin-Walied Pr*ofile Section Bending . . .

Stresses in Curved Beams

VII. ANALYSIS OF PLANE STRESS AND PLANE STRAIN 173
General Case of Plane Stress . , . Principal Stresses in Beams . . .

Stresses Due to Combined liending and Torsion . . . Plane Strain . . .

The Strain RosetU'

VII
vni CONTENTS
vju, i)eflf:ctk)X of bp:ams •
197
Equation tho Elastic- Lino
of .'"Tho Moment-Area .

Mofliod . . . f )ofl(H‘ti(>ns l)V Sup<Tposition Stnim Eneri?y of HendinE .

. j. I)rfl(‘ctjons l>uo Shearing Strain

IX STATIC VLLY IN DKTEUMIXATE KEAMb 231

Al(‘tliod of Sup(Tf)o.sitiori . . Theon m of Three Moments . . . Tlie
“
Thoonnii of Casti‘i;liano . .^Applications of Castigliano’s Theorem to
Stnticalh I ndid'o-Tinnatf Proldems . . Limit Analysis of Statioalh'
I ndeteriiuiiat(‘

X. lUlF.OJlV OF COLl'M\>. 264

Va c(‘iitri(' Loading Columns; EuIit's Column
of a Short Strut . . Lon.^
I'orniula Idirtlier I)iscussi(»n of Euler’s Column Formula The . . .

Sc'-ant I'ormiila Ht'pn'sentatiori of mp(‘rfections by Equivalent I

Ef cciitj’u iC . , E,mpiri(‘al Column Formula*^

XT Ml.Cll WICAI. ]M(OPEimivS OF MATERIALS 294
'iA'hMi* Test .Stretching of Steel H(>yond the
Yield Point.
\'i- Id Point 'fvpv's of Fractures in Tension
. . Compression Tests . . .

I'ests of MatiTials Ender Combined Stresses Striuigth Theories . . .

Impart Te«ts Fatigue' of Metals Fatujue I nder Combined

. . . . .

. Fatiiiiiio and Str(‘>s Concentrations Physical Properties . , .

o( Metals at Hiijli 'rempciatures

AeiM:Nuix \ 341
A\'( nij^e I’liy'icai Prf>pertit*s <tf C«.mmon Metals Avcracie Phvsn^al .

Pfopertc's and AN'»»rkin{^ Stresses tor Stnietural ‘rimber Averagf' . . .

PnqxTties of
i’h'>s]cal ibnldiu^i; Stone. Pnek and CVauTet(‘ . . .

Mei'hameai Proiierties of Steels

VPPKNDIX n. MOMENTS OF INERTIA OF PLANE AREAS 346

APPEXiilX 11-1 : TVDLES 357
Formulas for Eh Elements of Wide
inents of <''ommon Sections . , .

Fhomi’ Sect oils Eleinent.s of American Standard I-beam Sections

. .

I'h mimt'^ oi dvineiican Standard Channel Sections Elements of . . .

I.(jual \nt;l(’ Sections Idcments of Unequal An^le Sections. . .

INDEX
 NOTATIONS

A area
a, b, c diinensions
c distance from neutral axis to extreme fiber
d diaxaeter
E modulus of elasticity
c eccentricity
F force
G shear modulus
g gravitational acceleration constant
h height; depth of a beam
hp horsepower
/ moment of inertia of area
I radius of gyration
j polar moment of inertia of area
K stress concentration factor
k symbol for \F/EI; spring constant ;
factoi
I length
AI betiding moment
N normal force
n factor c»f safety; r.p.ni ;
number
P force; load
V pressure per unit area; pitch
Q force; staticalmoment of area
Q load per unit length
R reaction, radius
r radius; radius of gyration
S stress resultant
s arc length
T torque; temperature
t thickness
V strain energy
u strain energy pier unit volume
V shearing force; volume
V velocity
W weight total load
;

IX
 NOTATIONS
load per unit length : weight per unit volume
forces
coordinates
section modulus
temperature coefficient of expansion angle ;

angle
shearing strain ;
weight density
deflection; total elongation
tensile or compressive strain
hlope of elastic line: angle of twist per unit length
Poisson^s ratio
radius of curvature; radial c<jordinate
normal stress
shearing stress
angle of twist; angular coordinate
angular velocity
 CHAPTER I

TENSION, COMPRESSION, AND SHEAR: I

1.1 Introduction
Various struruires and machines —
bridges, cranes, airplanes, ships,
etc. —
will be found, upon examination, to consist of numerous parns or
members connected together in such a veay as to perform a us('ful function
and to withstand externally applied loads. C'onsider, for example, the
simple press shown in Fig. 1.1a. The function of this press is to tc^.st speci-
mens of various materials in compression. To accomplish this, the speci-
men is placed on the floor of the base A and the end of the screw is forced
down against it by turning the handwheel at the top. This action subjects
the specimen as well as the lower portion of the screw to axial compresmm
(Fig. l.ld) and the side members N to axial tension (Fig. 1.1b). It will be

Fto. 1.1
2 TENSION, COMPRESSION. AND SHEAR: I

observed also that the crosshead M is subjected to bending (Fig. 1.1c) and
the upper part of the screw to twist or torsion (Fig. Lie). Detailed study of
these four basic types of loading of structural members essentially con-
stitutes the subject matter of Strength of Materials. In this and sub-
sequent chapters, they will be taken up in order.
Analysis and design of any structure or machine like the press in Fig. 1.1
involve two major questions: (a) Is the ".structure strong enough to with-
stand the loads applied to it and (b) is it stiff enough to avoid excessive
deformations and deflections? In Statics, the members of a structure were
treated as rigid bodies; but actually all materials are deformable and this
property will henceforth be taken into account. Thus Strength of Materials
may be regarded as the statics of deformable or elastic bodies. For example,
it is clear that compression of the specimen in Fig. 1.1a can he increased
only by advancing the screw of the press downwards through the crosshead
M. This relative displacement between two parts of the machine is partly
accounted for by shortening of the sp)ecimen and the lower part of the screw
and partly by extension of the side bars N as well as some liending de-
flection of the crosshead M. Thus, the amount of compressive force on the
specimen that will correspond to one turn of the handwheel will depend
upon the relative stiffness of the various members of the machine.
Both the strength and stiffness of a structural member are functions of its
size and shape and also of certain physical properties of the material from
which it is made. These physical properties of materials are largely
detennined from experimental studies of their behavior in a testing ma-
chine. The study of Strength of Materials is aimed at predicting just how
these geometric and physical properties of a structure will influence its
behavior under service conditions. The applications of the subject are
broad in scop? and will be found in all branches of engineering. We begin
with a study of the simplest type of loading, namely, axial tension or com-
pression of a straight prismatic bar.

1,2 Internal Force; Stress

In Fig. 1.2, a prismatic bar AB is subjected to axial tension by Ihe

action of a vertical load P applied at B and acting along the axis AB of the
bar, the proper weight of which is neglected. This action on the bar
stretches it slightly and also tends to pull it apart, i.e., to produce rupture.
This tendency to rupture is resisted by internal forces within the bar, i.e.,

by actions and reactions between its various particles. To visualize these

internal forces, imagine that the bar is cut by a section mn perpendicular to
its and that the lov/er portion is isolated as a free body (Fig. 1.2b) At
axis .

the lower end of this portion of the bar, the external force P is applied. On
 INTERNAL FORCE; STRESS 3

the upper end are the internal forces represent- ^

ing the actions of the particles of the upper part
of the bar on those of the lower part. These
forces are continuously distributed over the
cross-section mn. In dealing with such distri-
buted forces, the intensity of force, i.e., the force
per unit area, is of great importance. Visualiz-
ing the bar as made up of a bundle of longi-
tudinal fibers, each of which carries its fair
share of the load, it apf)ears reasonable to
assume, in this case, that the distribution of
*
forces over the cross-section will be uniform
From the condition of equilibrium of the free
body (Fig. 1.2b), it is seen that the resultant of
this uniform distribution of internal forces must
lx* equal to the external load P. Thus, if A de-
notes the cross-se(!tional area of the bar and
(T, the force per unit area, we have S = aA
= P, from which
P
" = I'
This force per unit area is called the stress in the bar; the total tension
S — sometimes called the stress resultant. Force is usually measured
<rA is

in pounds and area in square inches so that stress has the dimension of
pounds per square inch, denoted by ‘^psi."'

non-un^onn; but this effect is very localued and will be ij^ored for the present. For
further discussion see Art. 2.5, p. 46.
4 TENSION, COMPRESSION, AND SHEAR: I

In order that the applied load P in Fig. 1.2 will actually induce a uniform
stre.ss <T ov’^er each cross-sc^ction of the bar as assumed above, its line of

action must pass through the centroid of each cross-section, i.e., P must
act along the (‘entroidal axis of the bar. To prove this, consider an arbitrary
shape of (Toss-section as.shown in Fig. 1.3 and let dA be any element of area

therein. Then for the assumed uniform stress distribution, a is constant

over the cross-section and the element of force acting on dA is cdA, normal
to the plane of the section. The resultant of these parallel forces is

N = fad A = afdA = aA , (a)

also normal to the section.

The S can be found from the
point of application of the stress resultant
theorem moments: namely, the moment of the resultant about either of
of
the coordinate axes x or y must equal the algebraic sum of moments of the
elemental forces <jaA about the same axis. Thus, denoting by x and y the
coordinates of the point of application of the resultant, we have

(tAx = jx<rdA = aJxdA = (tAxc,1

aAy - jyadA = crjydA =

where Xc and yo are the coordinates of the centroid C of the cross-section.
From e(^s. (b), it is seen that x = Xc and y = ijc. Thus for a uniform stress
distribution, the stress resultant S acts through the eeniroid of the cross-
section. Furthermore, it can be seen from Fig. 1.2b that the force S must
be (‘ollinear with the applied force P. Therefore, P can produce a uniform
stress distribution over each cross-section only if it acts through their

centroids.*
All of the foregoing discussion applies also to the case of a short post or
stnd subje(‘ted to a compressive load P as shown in Fig. 1,4. Here also the

*A tensile load P that does not act along the centroidal axis of a bar will produce bend-
ing as well as tension of the bar. This case is discussed in Art. 10.1, p. 264.
 INTERNAL FORCE; STRESS 5

load P must act along the centroidal axis of the post to produce the uniform
compressive stress cr indicated in Fig. 1.4b. In the case of compression
members, this condition is sometimes difficult to fulfill, so that the com-
pression of long slender struts or columns requires special consideration
which will be taken up later in Chapter X.
Direct Shear, Referring again to Fig. 1.2, let us consider now" in some
detail the connection between the tension member AH and the ceiling at its

upper end. Clearly, in the interests of good design, this connection should
be strong enough to develop the full load-carrying capacity of the bar AB
itself. An enlarged detail of this connection is shown in Fig. 1.5a, where it is

seen that the load P on the tension member must be transmitted to the fork
by the horizontal pin connecting the two parts. A free-body diagram of this
pin is shown in Fig. 1.5b and it is seen that the pm is primarily in a condition
of shear which tends to cut it across the sections mn* Assume now" that the
internal shearing forces resisting this tendency are uniformly distributed
over each of the cross-sections mn. Then denoting by Tqv the shear force
per unit area, i.e., the average shear stress,we see that equilibrium conditions
of the middle portion of the pin require that TrvAh = P, from which

r«v = £, (1.2)

where A, is the total area in shear — in this ease, twice the cross-sectional
area A of the pin.
Since shearing conditions are never as simple as assumed above, it must
be realized that the average shear stress as calculated from eq. (1.2) may be
o»ly a rough approximation to the actual stresses that exist in the material.
Nevertheless, lacking any more exact knowledge of the true stress distribu-
tion, the designer is often forced to use this simple concept of average shear
stress as a basis for design.
In dealing with various kinds of machines and structures, the engineer
frequently encounters members subjected to simple direct tension, com-
pression, or shear as discussed above. The general problem of design of
such members consists in proportioning them so that they can safely and
economically withstand the loads that the}’' have to carry. As a basis of
doing this, many materials have been tested in the laboratory to establish
their strength or resistance to rupture under various types of loading and
thereby establish allowable or safe working siressest to be used in design.

*There is also some bending of the pin, but if the clearances are small this will be of
secondary importance. Only the shearing action will be considered in the present dis-
cussion.
tThe establishment of working stresses is a very complex question which will not be
discuss^ in any detail at this point. For furtlier discussion, see Art. 2.2, p. 32.
G TP^NSION, COMPRESSION, AND SHEAR: I

An allowable working stress is usually taken as 1/n times the value of the
stress at which failure of the test specimen took place. Thus in using such
a AVorking stress, the designer has a so-called factor of safety n to allow for
overloading or other unforeseen adverse effects. Losing these somewhat
working stresses together with eqs. (1.1) and (1.2), the
arbitrarily assigned
designer can determine the proper dimensions for the various members of a
machine or structure subjecled to the action of given loads. Or, if the
structure has already been biuii, he can establish safe values for the allow-
able loads in a similar maimer.

KXAMCLf: A \'(‘rtical load /"

1 = 5{X)0 U> isMjpported by two inclined stctd wires
and lU' as shown in I'^ig. LG Determine the required (Toss-sectional area of 1

Fig, 1.6

each wire if the allowable working stress in tcuision is o-„, - 10,0(X) psi and the
angle 6 = 30®.
solution, a free-body diagram of the ring C is shown in Fig. 1.6a, and tjie cor-
responding triangle of forces in Fig. 1.6b. This triangle being equilatfTal, wA‘ have
5 = P = 5000 lb. Then from eq. (LI), the necessary cross-sectional area is

S 5000
A - 0.5 sq in.
O’ «; 10,000

EXAMPLE 2. a deep- well pump is operated by a vertical prismatic

The piston of
steel rod of length I
- 320
attached to a crankat its upperend as ahowm in Fig. 1.7.
ft

Determim‘ the extreme values of tinisile and compressive stress a in the rod if the
resistance on the piston during the dowmstroke is 200 lb and during the upstroke is
 INTERNAL FORCE; STRESS
2000 lb. The cross-sectional area of the rod is A *= 0.338 sq in. and its density is

490 Ib/cu ft.

SOLUTION. The weight of the rod is

wlA = ^ X 320(12) X 0.338 « 368 lb.

The maximum tensile stress will occur at the top of the rod during the upstroke.
Under these conditions, the total tension is

Sm.K = 368 -h 2000 = 2368 lb.

The corresponding maximum tensile stress is

S 2368 .

The greatest compressive stress will occur at the lower end of the rod during the
downstroke. For these conditions S = 200 lb compression and

5 200
^ 592 psi.
A “ 0.338

EXAMPLE 3. A tension rod made up of two parts as shown in Fig. 1.8 is designed to
carry a total load P = 20,000 lb. What is the proper diameter d for the connecting
bolt if the allowable working stress in shear is ^ 10,000 psi?
SOLUTION. Assume that the bolt fits snugly in the holes through the prongs of the
fork and that the clearances are small so that bending action on the bolt will be
minimized. Then the bolt is esseixtially in a condition of direct shear across the
sections aa and eq. fl.2) may be used. This gives

F ^ 20,000
A 2 sq in.
Tw 10, (KX)

Thus the required cross-sectional area A of the bolt is 1 sq in. and the corresponding
diameter d « 1.13 in.
 TENSION, COMPRESSION, AND SHEAR: I

PROBLEMS
1* A short hollow cast-iron cylinder with a
w’all 'thickness of 1 in. is to carry a

compressive load P = 150 kips (1 kip = 1000 lb). Compute the required outside
diameter do if the working stress in compression is Cre = 12,000 psi. Am. do =
4.98 in.

2. Solve the preceding problem if the wall thickness is to be one-tenth of the

outside diameter do. Ans. do = 6.65 in.
3. A steel wire hangs vertically under its owm weight. What is the greatest
length can have if the allowable tensile stress is tr* = 30,000 psi? The specific
it

weight of steel is 490 Ib/cu ft. /ln,9. I = 8820 ft.

4. The upper portion of the tension rod in Fig. A is a l-in.-square aluminum bar

for which <Ty, = 8000 psi. The lower portion is a square steel bar for W'hich (Tu. =
20,000 psi. Calculate the safe load F based on the aluminum and the proper cross-
sectional dimension a for the steel bar to develop the full strength of the as8embl3^
Ans. a ^ 0.4 in.

Fig. \

5. If th(‘ aWow^able w^orking stress lor the pump rod in f'ig. 1.7 is reduced to
<Tu, = 6000 psi,w'hat is the required cross-sectional area .4*^ Use all other data Us
given in Example 2. Ans. A == 0,407 sq in.

Three pieces of wood having

6. in. X IJ-in.-square sections are glued together
as shown in Fig. B. If the assembly carries a total load P * 4000 lb as showm, what
IS the average shearing stress in the glued joints? Ans. r^v = 444 psi,
‘*7. For the wood truss shown in Fig. C, allowable stresses in direct compression

and shear are - 400 psi and “ 125 psi, respectively. Compute the proper
dimensions a and h for the notched joints at the ends of the horizontal memlier.
Atw. a « 4 in.; 5 = 1.25 in.
8. In Fig. D a lever is attached to a spindle by means of a i-in.-square key 1 in.
long. If the average shear stress in the key is not to exceed 8000 psi, what is the safe
value of the' load P applied to the end of the lever? Ans. P ^ lb.
9. Two lengths of 2i-in .-diameter shafting are in line and are coupled together by a
sleeve that fits over one end of each shaft. To prevent relative rotation between the
tw^o portions of the shaft, f-in. -diameter pins pass through the sleeve and shafting
at right angles to the axis of the shaft. These pins are purposely made of a weak
material so ac to prevent overloading of the machine. If the maximum torque to be
transmitted bj the .ihaft is 10,000 in.-ib, what should be the ultimate shearing
strength of the rnaU^rial chosen for the pins? Ans, 16,300 psi.
lOr The hub of a pulley may be fastened to a l-in.-diameter sikaft either by a
 INTERNAL FORCE; STRESS

square key or by a pm as shown m Fig. F. (\mipuU‘ tfu* iorque tliat (‘ac*h connection
can safely transmit if the a\^orage shearing stn^ss in the k('v or pin is not to exceed
10,000 psi. Ans. Tk =- 1250 in. -Ih, for key; = 1105 in -lb, fs?' pm

1.3 Elasticity; Strain

Let LIS consider again the ease of a prismatn' bar in

simple tension as shown in Fig, 1.9. As the tensile load P
is gradually increased, the bar will be. oh.served to stndch
slightly so that for each value of P there will be a (•.in-

responding small elongation 6 of the bar as shoAMi

Subsequently, if the load is gradually diminished to ^eru.
the elongation 5 will either eomplel('ly or partially dis-
appear, i.e., the bar tends to ^eassume its initial length /.

This property of a material to return partially or com-

pletely to its initial shape after unloading is called
elasticity. If the bar completely recovers its initial length
the material is said to be perfectly elastic; if not, it is .said

U\be only partially elastic. Experiments show hat many t

structural materials like steel, aluminum, wood, and even

concrete may
he considered as perfectly elastic within
limits, not excessively loaded. For the present dis-
i.e., if

cussion. we confine our attention to this clastic behavior of a materia)

Referring to Fig. 1.9, it may be assumed that, under tension, all longitu-
dinal fibers of the bar are stretched uniformly. Thus, we define the elonga-
tion per unit length of the bar by

which is called the tensile strain or simply the. strain. Similarly, for a bar
in axial compression, the quantity c will define the contraction per unit
length of the bar, or the compressive strain. Tensile strain will be considered
positive; compressive strain, negative. It should also be noted that strain is a
dimensionless quantity, l>eing a length divided by a length.
Hooke's Law, Experiment with prismatic bars of various materials in
tension have showm that, within the range of elastic behavior of the
10 TENSION, COMPRESSION, AND SHEAR: I

material, the elongation 6 is P and the

proportional to both the tensile force
length I of the bar. It is also observed to be inversely proportional to the
cross-sectional area A of the bar. Expressed algebraically, 6 Pl/A^ or

= 1 iL
E A ^ AE'
' (1.4)

where E is a constant for any given material and is called the modulus of
elasticity of the material in tension. The relationship expressed by eq. (1.4)
is known as Hookers law^ after Robert Hooke who first established it by
experiment in 1678.
Using the notations
P 6

from eqs. (1.1) and (1.3), Hooke’s law can also be written in the form

e 7; or <7 = Ee. 0.5)

E
Expressed in words, this leads to the familiar concise statement of Hooke’s
law: Stress is proportional to strain. The modulus of elasticity E is seen to
represent the factor of proportionality between stress and strain and
may be directly defined by
_ stress
? (a)
c strain

Since the strain < is dimensionless, it may be concluded that the modulus of
elasticity E has the dimension of stress, i.e., pounds per square inch, psi.
From its definition (a), the modulus of elasticity E is seen to represent
that stress a which would produce a tensile strain c = 1 ;
in other wor^s,
that tensile stress under which a bar would be stretched to twice its original

length if the material could remain perfectly elastic throughout such

excessive strain. As this observation would lead us to expect, the modulus
of elasticity is a very large quantity for most materials. In the case of steel,
for example, a value E = 30(10)® psi is usu-
ally used,* ^
If we plot tensile stress n against rain e

from eq. (1.5), we obtain the straight line

0*4 in Fig. 1.10. This plot represents the
so-called stress-strain diagram for a material
in simple tension within the range of its
elastic behavior. The modulus of elasticity
E is seen to represent the slope of the
stress-strain line OA,
'“For value.j of F for other commonly used structural materials, see Tables I, II, III,
Appendix A, p. 341.
 ELASTICITY; STRAIN 11

above discussion applies also to the case of axial compression

All of the
of a bar. For most materials, the modulus of elasticity in compression is
found to have the same value as for tension. In calculation, tensile stress
and strain are to be considered as positive: compressive stress and strain,
as negative.
Poisson's Ratio. Referring again lo the tension member uj Fig. 1.9,
experiments show further that its axial ek>ngatio!i isalwa 3’'s accompanied
by a lateral contraction and that the ratio

un it lat eral c ont raction

^ unit axial elongation

is constant for a given material within its range ol elastic behavior. This
constant y. is called Pu)sson's raho^ after the French mai hcinatician who
predicted its existenee and value, using the molecular theory of structure
of the material. For so-called isotropic materials whifdi have the same
properties in all\»isson concluded that m -
directions, I Careful measure-
ments on sti’uctural metals in ten.sic.n usually confirm this v:ilue quite
closely. For vtriK'tural steel, experiments indicate that approximately
fi
^ o.:h

Knowing (he modulus of elasticity E arid Poisson's ratio /i for a given ma-
terial, the change in dimensions and voiurrte of a. prismatic bar in tension can
easily be calcula1(‘d Before deformation, the volume of the bar is F = AL
After deformation, the new length q — /(I + t) and the new cross-sectional
area Ai = A(i “ Thus the new smlume is = Ail\ — Al{l + c)
(1 — fiey or, since a is a small r|uanlity compared with unity, Fi — AZ

(1 + e — 2/icb ITus the change in \oiume is AV" -- - F == Ah

(I -- 2ii) and the anil vohimr change beeome.s

AF Ah(\ - 2m) ^ ~
Y 2m). (1.6)

Since it is unlikely that any material would diminish

volume under in

tension, we may expect that m will alw'ays be less than i. For such a
material as rubber, m approaches the above limit and the volume of rubber
rcmakis practically constant during extension. Other materials like con-
crete and cork have smaller values of m and increase slightly in volume
when subjected to tension The value of m for concrete can be taken as 0,1,
while for cork, m = 0.
The above discussion of axial tension can be applied also to the cc.3e of

axial compression of a prismatic bar. Jn such case, the longitudinal con-

traction of the bar will be accompanied by lateral expansion and for cal-
culating this expansion, the same value for m as in the case of tension can
be used.
12 TENSION, COMPRESSION, AND SHEAR: I

EXAMPLE 1. Referring to Fig. 1.1 calculate the

total elongation 6 of a prismatic bar of length I and
cross-sectional area A which hangs vertically under
its own weight.
SOLUTION. Due to the uniformly distributed weight
of the bar, the total tension at the cross-section mn
will

.S, = yAx, (b)

where y is the weight per unit volume of the material.

As a result of this tension, an element of length dx sit-
uated just above the cross-section mn will elongate by
the small amount

obtained from e(i. (1.4) by replacing P by and /

by dx. Summing up all such elemental elongations
Fio. 1.11 along the full length of the bar givt's

rs.d2
‘•J.Aif
Substituting S, = yAx from eq. (b), this becomes
becor

It may be noted that since the total weight of the bar is W = yAl, eq. (e) can also be
expressed in the form

C'omparing the result with eq, (1.4), we see that in this case of uniformly varying
tension, the total elongation 5 is just half what it would be if the tension were ecpial
to W throughout the length of the bar.
EXAMPLE 2. A homogeneous slender prismatic bar of total length 21 rotates with
constant angular velocity w in a horizontal plane about a fixed axis through its mid-
point as shown in Fig, 1.12. The cross-^sectional area of the bar is A and its weight
 ELASTICITY; STRAIN 13

per unit volume is 7. Find the maximum stress o and the total elongation h of either

half of the bar due to centrifugal tension.

SOLUTION. Consider an element of the bar of length (if at the distance f from the
axis of rotation. The centrifugal force on this element due to its motion in a circular
path of radius f is

dF - 7^
9

where yAd^/g is the mass of the element and is its radial acceleration. This force
acts radially outwards along the axis of the bar as shown. The total tension 5, at
any cross-section mn defined by the distance x from the axis of rotation is the sum of
all these centrifugal forces between f = x and f = Thus /.

It
-to*^ = -T— /)»
(/- - i\
a:*), (f)
iff

This ta'nsile force has its maximum value at the mid-point of the bar (i = 0) where

yAf^H'^
mn* •“

The corresponding maximum stress is

_ 7^
/I 2g

It will be noted that this maximum stress increases as the square of both the angular
velocity co and the length /, so that a long bar rotating at high speed will be very

severely stressed.
The total elongation of either half of the bar will be

'S.dar
'

AE
Substituting the value of from eq. (f) and integrating

f^yAiii^dx yu}H^
b = (F - x^) (g)
0 ~2gAE

This may also be written in the form

(s')
ig AE 3<7 AE
where 11" = yAlia the weight of one half of the bar. This is the same elongation that
would take place in a massless string of length I having a mass particle of weight W/S
at its outer end.

PROBLEMS
1. What tensile force P will be required to produce a unit strain « = 0.0008 in a
steel rod that has a circular cross-section of diameter d ~ § in., if S = 30(10)* psi?
Ans. P = 4710 lb.
 14 TENSION, COMPRESSION, AND SHEAR: I

2 An aluminum bar 6 ft long has a l-in.-square cross-section over 2 ft of its length

.

and a l-in.-diameter circular cross-section over the other 4 ft. How much will the
bar elongate under a tensile load P = 3500 lb, if ^ = 10.5(10)® psi? Ans. 5 =
0.0284 in.
3. A steel bar 10 ft long has a circular cross-section of diameter dj = 1 in. over
half its length and diameter ^2 = i in. over the other half, (a) How much will the
bar elongate under a tensile load P = 8000 lb? (b) If the same volume of material is
rolled into a bar 10 ft long and of uniform diameter d, what will be the total elonga-
tion 5 of this bar under the same tension P ~ 8000 lb? Ans. (a) S = 0.102 in.;
(b) d = 0.0652 in.
^ 4. A steel bar having a J in. X J in. cross-section is loaded as shown in Fig. A.
Neglecting localized irregularities in stress distribution in the vicinity of the points

6
'

4 '

^ ->

1
^

'"
4^ 2

Ftq. A

of application of the loads, compute the net increase in the length of the bar. .4 ns.
8 = 0.0370 in.
A twisU'd wire cable
1200 ft long is fastened at its upper end and hangs vertical-
ly in adeep shaft. Assuming E — 15(10)® psi, and y = 4W lb per cu ft, calculate the
total elongation of the cable under the influon(*e of its own weight. An$. 5 = .96 in. 1

6. A uniformly tapering sU*el rod 10 ft long has diamet^T di = 2 in. at its upper
end and diameter d = 1 in. at its longer end. This rod hangs vertically and carries a
^

Umsile load P = 10,000 lb at the lower end. Calculate the total elongation 5 of th(‘
rod, neglecting its own weight. Am. 3 = 0.0255 in.
7. 2, p, 6, assume that the vertical rod of the mine pump
Referring to Example
is made of steel (E =
30(10)® psi) and that all other data are as given in the exami)l(‘.
With these data, find the propc^r radius r of the crank at the upper end in order to
attain an 8-in. stroke of the piston at the lower end. Am. r = 4.42 in.
*

8. For the simple structure shown in Fig. B, member BC is a steel wire having
diameter d = ^ in. and member .IP is a wood strut of l-m.-square cross-section.
Calculate thij horizontal and vortical components of the displacement of point B due

a
 ELASTICITY; STRAIN 15

to a vortical load P = 400 lb acting a? shown. For stool, = 30(10)® psi; for
wood, /i’u. - 1.5(10)® psi. ,lns. 5/. = 0.0290.045 in.
in.; 5, =
9. If the rotating stool bar in Fig. ] .12 has a total length 21 — 10 ft, what angular
speed of rotation may it have with.out exceeding a maximum allowable ttaisilo stress
of 40,000 psi? Ans. 1660 r.p.m.
10. The vertical stec*] bar in Fig. C has cros^-st'otional area .1 over th(’ length 1 2 -i ,

cross-sectional area .li over the length /i, and modulus of elasticity E. Find the
ratio P1/P2 such that the vertical displacement of point .1 will In* Z('ro. -Uw.

Pi 2

11 . In testing a concrete cylinder in compression, the original diarnetiT d = 0 in.

was increased by 0.0005 in. and the original hmgth / = 12 in. was decreased by 0 01
in. under a total compr(*ssive load P = 52,000 lb. Compub' the values of tin*
modulus of elasticity E and Poi.^^aon’s ratio ju. Uis. E ^ 2(10)® psi, y. = 1/11.
.

12. A prismatic bar of cross-sectional area .1 and lengtli /. fastened at its upper
end, bangs vertically under the action of its own weight and a tensih' force P at its
lower end. ( -aiculate the increase m volume of the bar if E, y, and 7 are given, Ans.

1.4 Statically Indeterminate Problems in Tension and Com-

pression

Consider the simple structure made up of three tension rnemliers, each of

cross-sectional area A and modulus of elasticity arranged as show'ij in
Fig. 1.13a and subjected to a vertical load P Sit A. Under the action of this
load^each bar will be subjected to some tension and a free-body diagram of
the connecting pin A will be as shown in Fig. 1.18b, where Y denotes the
tensile force in each inclined bar (equal because of symnietry) and X, the

Fm. 1.13
16 TENSION, COMPRESSION, AND SHEAR: I

tensile force in the vertical bar. Then from equilibrium considerations of

this pin, we must have
X -f 2Fcosa = P. (a)

It will be seen at once that w^hile this equation of statics dehnes a

relationship between the tensile forces X and Y, it is insufficient to deter-
mine their values uniquely. For this reason, the structure is said to be
statically indeterminate. Looked at in another way, it will be observed that
any tw^o of the three bars are sufficient to constrain the pin A in the vertical
plane of the figure and therefore a third bar is said to be redundant. In
general, any structure which contains more constraints than are necessary
for geometric rigidity of the structure, i.e., one which contains redundant
constraints, will prove to be statically indeterminate. In fact, if the bars of
the structure in Fig. 1.13a were absolutely rigid and undeforinable, as
assumed in statics, there would be no way to ascertain how the load P was*
divided among the three bars. However, the bars are actually elastic and
stretch slightly under tension. Furthermore, since they are connected
together by the pin A, it is evident that the amounts they stretch must be
related in some way by the geometry of the structure. This is shown in
Fig. 1.13c, Denoting by d t-he elongation of the vertical bar and by 5i the
elongation of each inclined bar and keeping in mind that these elongations
are extremely small compared with the dimensions of the structure, it will
be seen from Fig. 1.13c that
= 5 cos a. (b)

This equation of geometry or consiste?U deformation is the key to the problem.

Assuming that the tensions in the bars do not exceed their elastic iimiU and
using Hooke's law, we have

XI Yl, Yl
0 — ~7^ and, ,
5i =_ — ~T~r
’

AE AE AEco^^a

where U = i/cos a is the length of each inclined bar. Substitution of these

values into eq. (b) gives
Yl XI
Tin
AE cos a
==
Tn cos a,
AE
from which F = X cos^ a. (c)

Substitution of this value of F into eq. ta) gives

X + 2X cos^ a = P,
from w'hich

^ 1+2 cos® a
«“ and F
1+2^ cos® a ,
 STATICALLY INDETERMINATE PROBLEMS 17

Then the total vertical deflection of point A becomes

s =
AE
^ (—L\.
+2
\1 a) cos^

This example illustrates one case of a statically indeterminate system in-

volving axially loaded members. Others will be shown in the examples
which follow. In all cases of systems involving redundant elements,
equations of statics will be insufficient to determine all of the unknown!
forcesand must be supplemented by equations of consistent deformation
based on the geometry of the system.

KXAMPLK 1. A hollow st(^el / = 12 in., inside diamefcT d = 6 in.,

cylinder of length
and uniform wall thickness < = with concrete and compressed l>etween
J in. is filled
= 100,000 lb as shown in Fig. 1 .14a. Calculate the
rigid parallel plat<‘s by a load F
compressive stress in each material and the total shortening of the cylinder if
F, = 30(10)® psi for tlie steel and — 2(10)® psi for the concrete. Assume that
both materials obey Hooke’s law .

SOLUTION. A free-body diagram for the upper portion of the cylinder is shown in
Fig. 1.14b and from statics it follows that

OsA, -h (TcAc = P. (d)

Also since both materials must short(*n tlic saTue amount, because of the rigid (md
]jlates,and are of the same height I, their umt I’ompn'ssiv^e strains must be equal,
so that

(e)
Es

Phis shows that the two materials arc stressed in the ratio of their moduli of elastic-
ity, For the given vaiu(!s <if these moduli, E»/Ec = 15, so that a, = ISCf, and
eq. (d) gives
\hP
1 5. U Ac 15.L-f A,
18 TENSION, COMPRESSION, AND SHEAR: I

Using the given numerical data, A, = 2.41 in.^ and Ac = 28.3 in.^ so that
(Tc = 1560 psi and cr, = 23,400 psi. The total shortening of the cylinder is 5 = f/ =
(ac/Ec)l = 0.000775 X 12 = 0.00936 in.
EXAMPLE 2. A rigid bar AB is hinged at A and supported a horizontal position
in
by two identical vertical sh^el wires as shown in Fig. 1.15a. Find the tensile forces
iSi and »So induced in these wires by a vertical load P applied at B as showm.

soLUi'ioN. (‘onsidenng the (‘quiiibrium of the })ar AB, the proper weigiit of which
is neglech^d, and taking moments of all forces with respect to the hinge J ,
w’ehave

+ - PL (f)

'Mso, since u\B remains straight and simply rotab's slightly around point A, the
elongations and 52 of the w'ires must be in th(‘ ratio a/b, i.e.,

Then from Hooke’s law, Si/S>: ~ a/b and cq. (f) gives

Pal Pbl
a» + “ a= 4-
6"'

EXAMPLE 3. A cold-rolled steel bolt of length / = 12 in. passes through a hard-

drawm copper tube of the same length, as shown in Fig. 1 .16a, and the nut at the left
end is turned up just snug at room temperature T = 70° F. Subsequently the nut is
tightened up n ~ J turn and the entire assembly is raised to a temperature T =*
140° F. Wliat stresses will exist in the bolt and the tube under these conditions?
The cross-sectional area of the steel bolt is A* « J sq in., its modulus of elasticity
F, - 30(10)" psi, the coefficient of thermal expansion is a, 6.5(10)"" in./in./P^
and the thread pitch p ~ | in. For the copper tube, Ac = i sq in., Ec = 16(10)" psi,
and a. = 9.3(10)-" in./ in./F°.
SOLUTION. Consider that portion of the assembly to the right of a section vm as a
free body (Fig 1.16b). I'hen from static equilibrium it is seen that the compressive
force Sc in the copper tube must balance the tensile force in the steel bolt, i.e.,

= Sc. (h)
 STATICALLY INDETERMINATE PROBLEMS 19

Furthermore, since the final length of the bolt and the tube must be the same, it

follows that the shortening of the tube plus the extension of the bolt must be equal to
the thread displacement of the nut along the bolt. Expressed algebraically,

Sg + 6c « pti (i)

wherein the total extension of the steel bolt is

and the total shortening of the copper tube is

Sri
Or = XrlAT
ArEr

AT being the net increase in Uunperature. Substituting these expressions for 5, and
6r into eq. fi) and noting from (vp ph) that S, ~ Sc = S. we obtain

pn
I

For the gjven nurnericjil data this gi\c<

--
N 18.t)701b

The corresponding stresse-^ are - d7,340 psi, U,‘rision and (Tc = 24,900 psi,

(‘ornpression.

Fig. 1.16 Fio. A

PROBLEMS
The bottom chord member of a bridge consists of three steel eye-bars placed
t.

side by side as shown in Fig. A. Each bar m.a^' be assumed to have a uniform cros-s-
section, 4 in. X I in. throughout its "‘cf^-nter to center’^ length of 16 ft. The total ten-
sion in the compositci member is to be 200,000 lb. If the middle bar is accidentally
0.03 in. shorter than the other two Ixdore assembly, what will be the final unit
tensile ^ress in tliLs bar? Am, a — 19,800 psi.
20 TENSION, COMPRESSION, AND SHEAR: I

£ A concrete pedestal 8 in. high and having a 4- in. X 4-in.

square cross-section is reinforced by four J-in. -square steel
bars as shown in Fig. B. Calculate the safe value of the
compressive load P if the allowable working stresses for steel
and concrete are <t, = 20,000 psi and Cc ®= 1,000 psi. Moduli
of elasticity for steel and concrete are =* 30(10)* psi and

Ec = 3(10)* psi, respectively. Aub, P * 25,0001b.

Prestressed concrete beams are made in the following
manner: Steel wires are stretched between rigid end plates to
a tensile stress ctq as shown in Fig. Ci. Concrete is then
poured around them to form the beam as shown in Fig. Cj.
After the concrete sets, the external forces Q are removed and
the beam is left in a prestressed condition. If the moduli of
elasticity of steel and concrete arc in the ratio 12:1 and their
cross-sectional areas are in the ratio 1 :15, what are the final
residual stresses in the two materials? Ans. = 5<ro/9; <Tc
Fig. B = — <ro/27.
A trussed wood beam of length 2^ = 10 ft consists of a
4.

4-in. X square timber and two steel rods each of diameter d = J in. as shown
4-in.
in Fig. D. The angle 2ot - 20®, E^ = 1.2(10)* psi, and E. = 30(10)* psi. If the

Fig. C Fig. D
double-acting turnbuckles in the steel rods have 10 threads per inch, how many
turns should be given to each turnbuckle after they are snug in order to prestress the
wood to 3(f00 pai in compression? Neglect deformation of struts. Ans. 11.6 turns.

Fig. E Fig. F Fig. G Fig. H

 STATICALLY INDETERMINATE PROBLEMS 21

.1 A vertical prismatic bar is fastened at its upper end and supported at its lower

end by an unyielding floor as shown in Fig. E. Determine the reaction R exerted by

the floor on the lower end of the bar if external loads Pi and P 2 are applied at inter-
mediate points as shown. The following numerical data are given: Pi = 3000 lb,
P 2 6000 lb, a = 4 in., 6 = 8 in., c = 12 in. Ans. R = 3500 lb.
^6; A square post is made up of two pieces, one aluminum and one steel, side by
side as shown in Fig. F. If P, = 30(10)® psi and Ea = 10(10)® psi, find the proper
eccentricity e of the applied compressive load P so that each material will be under
uniform compressive stress. Ans. 6=- 1.00 in.
A rigid bar AB that is 6 ft long is suspended in a horizontal position by two
vertical wires attached to its ends. The wire at the left end A is 6 ft long, made of
copper [Ec = 15(10)® psi], and has a cross-sectional area A^ = 0.12 sq in. The wire
at the right end P is 8 ft long, made of steel [E, = 30(10)® psi] and has a cross-
sectional area A, = 0.08 sq in. At what distance x from the left end A of the bar
.IP should a vertical load P be applied to produce equal vertical deflections of A
and P? Neglect the weight of the bar AB. Ans. x = 3.00 ft.
8. A steel bar 5 ft long has a cross-sectional area of 2 sq in. for 2 ft of its length and
3 sq in. for 3 ft of its length. At room temperature, the bar fits exactly between un-
yielding walls at its two ends. If P = 30(10)® psi and the coefficient of thermal
expansion is a = 0.0000065 in./in. /F®, what compressive force will be induced in
the bar by a temperature rise of 50 F®? Assume no bending of the bar. Ans.
S = 24,400 lb.
9. A is slipped over a steel bolt and held in place by a nut that is
bronze sleeve
turned just snug. Compute the temperature rise required to stress the bronze to
4000 psi compression. Use the following data: A, = 0.50 in.®, At * 0.75 in.®,
P, « 30(10)® psi, Et =* 12(10)® psi, a. =* 0.0000065 in./in./F®, « 0.0000100 in./
in./F®, Ans. T = 152 F®.
To, A vertical steel rod fastened at its upper end has cross-sectional area A 1 =
1.00 sq in. over the length l\ = 12 in. and A 2 = 0.50 sq in. over the length h = 18 in.
as shown in Fig. G. The rod is also supported at C by a loose-fitting steel sleeve
having cross-sectional area A 2 * 0.50 sq in. and length /a = 5 in. Calculate the
deflection ^6 of the lower end of the rod due to a load P = 10,000 lb, assuming that
the sleeve is just snug when P * 0. Ans. Bb = 0.0138 in.

11. For the system shown in Fig. 1.13, the two side members are steel wires i in.
in diameter and the middle member is an aluminum wire } in. in diameter. The
length I of this wire is 20 in. and the angle a = 30®. At room temperature T « 70® F
and with no external load P, the assembly is free from stresses. Calculate the stress
in each of the wires due to an external load P = 1000 lb and a rise in temperature to
200® F. Ans. = 29,400 psi, tension; <Ta * 7920 psi, tension.
12. The side members of the square frame shown in Fig. H are aluminum bars of
1 sq in., cross-section and the diagonal members are steel wires i in. in diameter. If

the assembly is free from stresses at room temperature T = 70® F, what tensile
forces will exist in the diagonal wires if the entire frame is increased in temperature
to P = 170® F? Use the following numerical data: E» * 30(10)® psi, Ea « 10(10)®
psi, a, « 70(10)“’ in./in./F®, and a„ = 128(10)”’ m./in./F®. Ans. S * 772 lb.

1.5 Thin Rings

Consider the case of a thin circular ring subjected to the action of
shown in Fig. 1.17. If the cross-
unifonlily distributed radial loading as
 TENSION, COMPRESSION, AND SHEAR: I

sectional area A of the ring is constant along the circumference and the
thickness t is small compared with the radius r, such loading will produce
uniform circumferential strcwss and strain in the ring and the problem can
be treated as one of simple tension or compression. The distributed load
may be due to either internal or external pressure or centrifugal forces in

loo, 1.17 Fjg. 1.18

;b(' rase of a rotating ring. In any case, its intensity q may be defined in
pounds per incli of circimifercncc of the 'mean center line of rrxdius r.
To examine the internal forces induced by such loading, consider *the
tMpiiIi])rimn of an cjoment of length fis = ni$ as shown in Fig. I.17b, where
aS'denotes the cinMiniferential tension or hoop tension and qrdO is the element
of exteriial radial load. I'^.quating the algebraic stun of projections of
Iliese forces in the radial direction and noting that sin (d6/2) — dd/2 gives

qr de - 2^ = 0.

from \vhich N = qr, (1.7)

If the disinhuted loading q is directed radially inwards, eq. (1.7) will give
the iniiform circuniferential compression in the ring.
Since the thickness i of the ring is assumed small compared with its

mean radius r, the hoop tension S can be taken as uniformly distributed*

over the cro.ss-seci ion .1 and the unit stress becomes
aS qr
(1.8)
.4

•For thick nngs, hcc Timoshenko, Strength of Materiah, Vol. 2, p. 205.

 THIN RINGS 23

Likewise, the circumferential strain, uniform around the circumference,

becomes
qr
= <T
(1.9)
E IT
It may be noted that, because the circumference and diameter of the ring
are in the constant ratio tt, the unit diametral extension will be the same as
the circumferential strain, i.e., cd = t.. This notion of diametral strain €d
willbe useful in discussing problems of shrink Jits as will be shown in some
of the examples to follow.
The uniform radial loading shown in Fig. 1.17a can be produced by
rotating the ring about its geometric axis normal to lht‘ plane of the ring
with some constant angular velocity o) usually specified in radians per
.second (secO- Fnder such conditions, each element of the ring of mass dm
moves with constant speed in a circular path of radius r and has a normal
acceleration ~ coV, directed towards the center of the ring. The cor-
responding centrifugal force is toVdm directed radially outward For an
element of length ds ~ nid and weight w per unit length of circumference
the mass dm ~ 'W)d6 g. Such centrifugal forces on all elements of the ring
are seen to represent a uniform distribution of radial loading of intensity

w
== -co-r-
u (a)
g

Substituting this into eq. (1.7), the corresponding hoop tension S becomes

o = or = — (b)
g

Likewise from eq. (1.8), the hoop stress is

S w
cr (")
A Ag
where 7 = w/A is the weight per unit volume of the material from which
the ring is made. It will be noted that this hoop stress due to centrifugal
forces increases as the square of both the angular velocity w and the radius
r of the ring. Thus a ring of large radius that rotates at high speed can be
subjected to very high stresses,
EXAMPLE 1. A long steel water pipe having a mean diameter d ~ 24: in. and wall
thickness t
— i in. carries an internal pressure p =
150 psi, Fig. 1.18. Calculate the
magnitude of the hoop stress in the pipe wail and.' the increase in its diameter.
SOLUTION. Consider a section of the pipe having a length of 1 in. as a ring, the
cross-sectional area of w^hich will be A = 1 in. X i in. = 0.125 in.^ Then the
internal pressure p represents the intensity ol load per unit length of circumference
of this ring and eq. (1.8) gives

pr 150 X 12
^ “ 14,400
A “
psi.
0.125
24 TENSION, COMPRESSION, AND SHEAR: I

From eq. (1.9), the circumferential strain becomes

14,400
e 4.8 (10)-^
30(10)

Since this also represents the diametral strain, the increase in diameter is

Ad = 4.8(10)-^ X 24 = 0.0115 in.

EXAMPLE 2. A steel sleeve of small thickness t is to be shrunk onto a solid shaft of

diameter d as shown in Fig. 1.19. Neglecting deformation of the shaft, find the
proper initial inside diameter of the sleeve if the hoop stress induced in the sleeve by
the shrink fit is not to exceed a prescribed working stress <Tw

Fig. 1,19

SOLUTION. Using eq. (1.9), the allowable circumferential strain in the sleeve,
equal to its diametral strain, must not exceed the value

where Ad is the difference between the diameter d of the shaft and the initial inside
diametcT of the sleeve. From this expression

Taking, for example, (Tu- = 30,000 psi, E= 30(10)® psi, and d = 4 in., this gives

30,000
X 4 0.004 in.

Thus tlie initial inside diameter of the sleeve should be not less than 4.000 —
0.004 - 3.996 in.
EXAMPLE 3. At a temperature of 420® F, a brass hoop of 20 in. inside diameter
and ^ in. thick, fits snugly over a steel hoop of 20 in. outside diameter and 1 in, thick
as shown in Fig, 1.20. Both hoops are 1 in. wide in the direction normal to the plane
of the figure. Subsequently, the system cools down to a temperature of 70®F. Cal-
culate the radirj pressure q set up between the tw^o hoops and the hoop stress at in
the brass due to the shrink fit. The following data are given; E, = 30(10)® psi,
E, = 13(10)® psi, a, = 6.5(10)“® in./in./F®, = 10.4(10)“® in./in./F®. *
 THIN RINGS 25

SOLUTION. Neglecting the small difiFerence in the mean radii of the two hoops, we
assume that each hoop has a mean radius r * 10 in. Then it may be concluded that
the circumferential strains due to combined cooling and stressing are approximately
equal for the two hoops. This permits us to write

9r
aiAT - a^T-f .

AhEb
(ai, - ot,)AT A.E,
from which 9
A.E, r

A,E,
Substitution of the given numerical data in this expression gives q 730 lb per in.
between the two hoops. Then for the hoop stress in the brass,
for the radial pressure
we have
7300
»
” -2! * 14,000 psi.
Ab * 0.50

PROBLEMS
1. At what rpin will a thin steel ring of mean diameter d « 6 ft reach its ulti-

mate strength (o-u == 76.000 psi) in hoop tension? Am. n = 2680 rpm.
2. Determine the limiting peripheral speed of a copper ring if the working stress
i;

is <r„. = 3000 psi and the weight per unit volume is 550 lb per cu ft. Am. v ^ 159 fps.

3. A thin steel ring having an inside diameter of 19.96 in. is heated and shrunk
over a thin cast-iron ring having an outside diameter of 20.00 in. If each ring has
cross-sectional area A — 1.00 sq in., what are the final hoop stresses induced in the
two rings? Use Et ~ 30(10)® psi, Ee = 15(10)® psi. Ans. o’, = =* 20,000 psi.

4. The steel strap of a brake band is 2 in. wide and ^

in. thick. The brake lining
is J in, thick and is riveted to the strap by pairs of rivets which reduce the
effective width of the strap to IJ in. on account of the rivet holes. If the tensile
stress the steel strap is not to exceed 16,000 psi, what normal pressure per inch of
’in

circumference may the 10-in. drum exert on the brake lining? Am. q = 325 Ib/in.
5. A leather belt 0,35 in. thick overruns a 15-in. diameter pulley turning at 1200
rpm. What tensile stress is set up in the belt due to centrifugal action? Leather
weighs Ib/cu in. Am. or = 80 psi.
6. What angular velocity oj of the shrunk-fit ring in Fig. 1 .20 will be required to
relieve the shrink-fit pressure existing under the conditions given in Example 3?
What will be the hoop stress in the brass ring at this angular velocity? Am. co =
605 sec“*; at — 29,400 psi.
 CHAPTEK II
TENSION, COMPRESSION, AND SHEAR: II

2.1 Varialioii oT with Aspe<*t of f>o^>-Srt*tioii

III -lu- 1 ‘ase* oi‘ {iNiai ton-uai of a priMoatic har, l i^ 2 la. the ou a
normal cro.ss-sertion nin iniil(»rni aiul ha» rh(‘ niagnituclo a F^A as
disciEsscd in Art. 1 2. I.ot u^' rini.^idcr now ihc t'l’ ^tross on an oblique-
cross-section pej cutting the b-ar at an angle </> wuli the normal cros^-^eriion

P\ |/77

— jir T” ^

Fir,. IM

ran. First, \\a i-nlaU' that portion of the bar the left of the oblitpie
section pq as a free body and repn-^em the action of the renaoed portion
on this free body by the st r^^'.^ le-ultant A as .shown in 'jg. 2.1i>. fVoin the I

equilibrium condilion, tliis julernal force N must be t'qiial, oppositf, and

collinear with the external ftn‘ce J\ as .-hown. Resolving the force -S into
compimc-nt" .V and Q, normal and tangential, re'-ipoctivciy, to the plane pg,
we find
A' - /"‘-oso: Q - FAn<t>. (a)

Then since tlie area ot the oblique seetion pq is

A' — A /eos (/».

the corresponding stresse.- are

A’ F
 VARIATION WITH CROSS-SECTION 27

These are called, respectively, the normal stress and the shear stress on the
oblique section pq^ the aspect of which is defined by <>. We see that when
= 0 and the section pq coincides with the normal section mn, eqs.
(2.1) give
p
(o'n)mnx ~ (b)

and T = 0, as they should. However, as </> is increased, the normal stress

(7n diminishes, until when 0 = t/ 2. = 0. Thus there is seen to be no
normal lateral stress between the longitudinal fibers of a prismatic bar in
tension. On
the other hand, w'ith increase in the angle 0, the shear stress t
increases to a maximum value

r'max = - — (c)

when <t>
= 7r/4 and then diminishes to r = 0 w^hen <j> = 7r/2.

These observations lead us to consider more carefully the question of the

strength of a bar in simple tension. If the bar is made of a material that is

much weaker in shear than if is in cohesion,

that failure willit may happen
take place due to relative slipping between twT) parts of the bar along a 45°-
plane where the shear stress is a maximum, rather than due to direct rupture
across a normal section where the normal stress is a maximum. For example,
a short wood post loaded in axial compression, as shown in Fig. 2.2a, may

Fig. 2.2
28 TENSION, COMPRESSION, AND SHEAR: II

actually fail by shearing along a jagged plane inclined roughly by 45° to the
axis of the post. In such case, we may still specify the value of Pj at which
this failure occurs as the ultimate strength of the wood in compression, even
though the failure is not a true compression failure of the material.
Similarly, during a tensile test of a flat bar of low-carbon steel with
polished surfaces, it is possible to observe a very interesting phenomenon.
At a certain value of the tensile stress a = PjA visible slip bands approxi-
mately inclined by 45° to the axis of the bar will appear on the flat sides of
the specimen as shown in Fig. 2.2b. These lines, called Lueders' lines, indi-
cate that the material is failing in shear, even though the bar is being loaded

in simple tension. This relative sliding along 45° -planes causes the speci-
men to elongate axially, and after unloading it will not return to its original
length. Such apparent stretching of the bar due to this slip phenomenon is

called plastic yielding. Again, the axial tensile stress cry.p. = P/A at which
this occurs may be designated as the yield stress in tension, even though the
failure is not a true tension failure of the material. These matters will be
discussed further in the next article.
Formulas (2.1), derived for the case of axial tension, can be used also for
axial compression, simply by changing the sign of P/A. We then obtaiji
negative values for both the normal stress and the shear stress t. The
o-n

complete state of stress on a thin element l)etween two parallel obilqucj

sections for axial tension and axial compression are compared in Fig. 2.3.
The directions of these stresses associated with axial tension (Fig. 2.3a) will
be considered as positive; those associated with axial compression (Fig.
2.3b), as negative. Thus Cn is positive when it is a tensile stress and negative
when it is a compressive stress. By reference to Fig. 2.3, the rule for sign of

The shear stress r on any face of the element

shear stress r will be as follows;
willbe considered positive when it has a clockwise moment with respect
to a center inside the element (Fig. 2.3a). If the moment is counter-
 VARIATION WITH CROSS-SECTION 29

clockwise with respect to a center inside the element, the shear stress
is Stated in a different way, the shear stress on any surface of a
negative.
body will be considered to be of positive sign if it points in a direction cor-
responding to clockwise rotation about a center inside the body, otherwise

of negative sign. Several examples of both positive and negative shear

stress are shown in Fig. 2.4. These sign conventions, while arbitrary, must
nonetheless be carefull}" observed to avoid confusion.
Returning to the case of a bar in axial tension, let us consider now the
stresses on an oblique section p'q' at right angles to the section pq, as
shown in Fig. 2.5. To obtain the stresses (t\ and r' on this section, we need
m

\
^
\
/o'""
(o)
n

only to replace by 90® + in eqs. (2.1). Then remembering that sin

(90° + = cos 0, w^hile cos(90° + <>)
= - sin this gives

P
= "j = P ^
ffn cos^(90° + <t>) -T sin“* 0,

(21')
fp . IP
^ 2 A ~ ~2 A
These stresses on the plane p'g' act as shown in Fig. 2.5b.
80 TENSION, COMPRESSION, AND SHEAR: II

The complete sot of stresses given by eqs. (2.1) and (2.T) are called
c<mj)lementary stresses because th(‘y occur on mutually perpendicular
planes. Comparing the two sets of formulas, we observe that
O'n + cr'n — F^/A,

r' = — r.
Thus the sum of normal and a\ on any two mutually perpen-
stresses an
dicular sections of a bar in axial tension is constant and ecpial to P/A, the

normal stress on the normal section mn. Also, complcmeniarii shear stress es
arc always ecfual in maqm^de hut opposite in sign.
TTie~THiuallry of Tom^ :Mitary shear stresses
such as r and r' on the faces of a rectangular
clement
(iMg. 2.0) also can bo established from the equilib-
^ iiuin conditions of the element itself, as follows:*
Lot dz (haioto the thi(‘kness of the element normal
^ to the plane of th(‘ paper and ds, ds', the lengths of
its edges. Then the areas on which r and t' act wall
be, respectively, dsdz and ds'dz. Multiplying the
Fig. 2.6 shear stresses by the areas on which they act, we
obtain two counteracting couples, the moments of
which must balance each other. Thus

ridsdz) X ds' = r{ds'dz) X ds,

from which r = r', where t has already been represented as negative in

Fig. 2.6.

EXAMPLE 1. A short steel bar having a 1-in. X Tin. square cross-section is sub-
jected to compressive forces P = 25,000 lb axially applied as shown in Fig. 2.7.

Fig. 2,7

Compute the complete set of complementary stresses on the sides of the rectangular
clement A oriented as shown.

SOLUTION Taking P/A = —25,000 psi and = 30® in eqs. <t> (2.1)

an = -25,000 X (0.866)2 = -18,750 psi,

r = -12,500 X 0.866 = -10,820 psi.

 VARIATION WITH CROSS-SECTION 31

Similarly, from eqs. (2.1')

tr', = -25, (XK) X (0.50)‘ = -1)250 p.si,

r' - -(-12, .500) X 0 800 - + 10,820 psi.

As a chock on these results, it can he note<l tliat thi'v satisfy ('qs. (d).

PROBLEMS
K A high-strength specimen having a circular cross-section of
tensile test
diameter d - 0.505 a total tension 1* — 23,000 Ih. Determine the
in. carri(‘s
maximum shear stress t,uhx and the normal stress an on the plane of this maximum
shear. Am. r.nax = 57,500 psi; a„ ~ 57,500 psi.
A steel rod of diameter d = j in. and length I ~ 20 in. fits snugly between un-
yielding supports .1 and B at room timiperature (70° F) as shovvn in Fig. A. ('^()m-
pute tlie normal stress <r„ and the shear stress r on the oblique section pg if the
temperature drops to —30° F. The coefficient of thermal expaiision for the bar is
a ~ 7.0(10)“® in.. in./F° and the modulus of elasticity is K - 30(10)® psi. Am.
<Tn = 10.500 psi; 7 = —10,500 psi.

Fig. a

prismatic bar carrying an axial t(*nsile strt‘ss a^ Ls cut by an oblique vSr'ction

pq as shown in Fig. B. If the normal and shear stress, resp('ctiv('iv, on this section
are an = 12,000 psi and r = 4000 psi, find the value of and the angle </> deffining
the aspect of the section pq. Ans. a^ -
13,333 psi; (f) = 18° 20'
4)r^cfcrring tx) the case of axial tension of a prismatic bar as shown Fig. 2.5, the m
following data are given; A = 1 sq in., P = .5(X)0 lb, </> = 30°, Calculate the
(r'n, r, r', for sections pq and p'q'.
stresses Cn, Ans. an = 3750 psi; a\ — 1250 psi;
T = — r' = 2165 psi.
5. A 6-in. diameter concrete test cylinder is subjected to a compressive load
P = 113,200 lb axially applied. Calculate the normal and shear stress on an
oblique cross-section defined by the angle 0 = 45°. Ans. an = —2000 psi; r =
—2000 psi.
Referring to Fig. B, find the value of 0 for which an = 2o''n.
6. What is the cor-
responding shear stress t? Ans. 0 = 35° 16'; r = 0 472 a^.

P'

Fig. C
34 TENSION, COMPRESSION, AND SHEAR: II

taken into consideration that at stresses below the proportional limit the
material may be considered as perfectly elastic, and beyond this limit a
part of the strain usually remains after unloading the bar, i.e., permanent set

occurs. In order to have the structure in an elastic condition and to remove

the possibility of a permanent set, it is usual practice to keep the working
stress well below the proportional limit. In the experimental determination
of this limit, sensitive measuring instruments (extensometers) are necessary
and the position of the limit depends to some extent upon the accuracy
with which the measurements are made. In order to eliminate this difficulty,

one usually takes the yield point or the ultimate strength of the material
as a basis for determining the magnitude of the working stress. Denoting
by ayy, (Ty
p , o-^it respectively the working stress, the yield point, and the
ultimate sti’ength of the material, the magnitude of the working stress
will be determined by one of the two following equations:

n }ii

Here n and /o are us\ially called factors of safety, which determine the
magnitude of th(' working stress. In the case of structural steel, it is

logical to take the yield point as the basis for cal(‘ulating the working
stress, because hoiv a considerable permanent set may occur wdiich is

not permissible in engineering structures. In such a case a factor of safety

n — 2 will give a conservative value for the working stress provided that
only constant loads are acting upon the stru(‘ture. In the cases of suddenly
applied loads or variable loads (and these occur very often in machine
parts), larger factors of safety may be necessary. For brittle materials
such as cast iron, concrete, and various kinds of stone and for such material
as wood, the ultimale strength is usually taken as a basis for determining
the working v^tresses

The magnitude of the factor of safety depends very much upon the
accuracy with which tlie external forces acting upon a structure are known,
upon the accuracy with which the stresses in the members of a structure
may be calculated, and also upon the homogeneity of the materials used.
The common practice of speaking of working stresses and factors of
safety based on some characteristic stress such as the yield point of steel
or the ultimat;e strength of cast-iron is somewhat dangerous and mis-
leading. If Pi, Pi, ,F*k are a set of external loads for a structure, what

we really mean by a factor of safety n is that if all these loads are in-
creased to nPi, nPi,. nPk, the structure will be just on the verge of
failure, wffiere “failure” of course must be clearly defined. It may mean,
in the case of a steel structure, that collapse due to yielding of some mem-
l>ers can occur, or, in the case of a concrete structure, that some member

is on the verge of fracture.

 DIAGRAMS, WORKING STRESS 35

In the case of a statically determinate system, the stresses in its members

will all increase in proportion to the applied loads even though the pro-
portional limit is exceeded. Thus, in such systems, the use of a working
stress as defined by one of ecjs. (2.2) gives the desired objective as stated
above. However, in the case of a statically indeterminate system like
those discussed in Art. 1.4, the stresses in the members depend on their
deformations. Thus, the methods of analysis used in Art. 1.4 and based on
Hookers law will not be applicable beyond the proportional limit. In
such cases, the use of working stresses as defined by ecjs. (2.2) will be more
or less meaningless.

2.3 Limit Design

We shall here discuss briefly a method of analysis of statically in-
determinate structures which will enable us to predict the loading under
which the structure will collapse due to simultaneous yielding of some or
all of itsmembers. Such an analysis, of course, is automatically limited to
steel structures where there is a pronounced yield point of the material.
The discussion will also be confined to structures made up of simple tension
or compression members under uniform stress. When we have the magni-
tudes of such collapse loads and specify the working loads as l/n times these
values, we have realized a true factor of safety n against complete failure
of the structure. This philosophy of limit design or 'plastic analysis is
gaining wide favor among structural engineers because its use results
generally in more efficient and economical designs.’*'
To begin with, it is customary to idealize the portion OABC of the stress-
strain diagram for steel as shown in Fig. 2.8a to that in Fig. 2.10, where it is
assumed that proportionality holds up to the yield stress and that therer
after the material yields indefinitely. Such an idealized material is said to

lo-

Fig. 2.10 Fig, 2.11

*Fo»’ those especially interested 4n this subject the following references may be con-
sulted: J. A. Van den Brock, Limit Design, John Wiley & Sons, N. Y., 1935. American
Institute of Steel Construction, Plastic Design in ^teel, 1959.
36 TENSION, COMPRESSION, AND SHEAR: II

be either perfectly elastic or perfectly plastic depending on whether the

stress below or at the yield point. It will be further assumed here that
is

^he yield point has the same value in both tension and compression. For
ordinary structural steel, this value may be taken as 40,000 psi.
To illustrate now the idea of limit design, let us consider the statically
indeterminate system shown in and already discussed in Art. 1.4,
F'ig. 2.11
assuming now that all three members are made of steel and have the same
cross-sectional area A. As the external load P is gradually increased, the
axial forces X and V in the members increase and are in the constant
ratio shown by eq. (c), p. 16. This continues until at a certain value Pi
of the external load, the force in the vertical bar reaches the magnitude
X = (Ty i.e., the stress in this bar reaches the yield point. At this
time, the vertical bar becomes plastic and will stretch thereafter without
further increase in tension, but the inclined bars having the tensions
Y = X cos- a are still elastic and capable of carrying further load. Thus
as the l(jad P is increased beyond the value Pi, the tensile force A' in the
vertical bar remains constant (<ry.p A) and the tensile forces V in the
inclined bars continue to increase. However during this phase of the in-
crease in external load the relationship (c) on p. 16 is no longer valid.
We have only to satisfy the equilibrium condition

X+ 2K cos a - P. (a)

h'inally, the inclined bars also reach the yield point and w^e have A= F=
(Ty.pA. Substituting these values into eq. (a), we obtain

Pyp = (Ty p A(1 + 2 cos a). (b)

This repiesents the so-called collapse load or limit load of the structure.*
A working load P = Py.p./n will now have a true factor of safety n against
complete collapse of the system. We see from this discussion that a plastic
analysis of a statically indeterminate system is actually simpler than an
elastic analysis as given in Art. 1 4.

As another example of plastic analysis of a structure, consider the case of

a square frame with pin joints loaded as shown in Fig. 2.12a. This structure
is seen to be statically indeterminate, having one redundant member, but
we make no attempt here to calculate the axial forces induced in the bars
during elastic behavior. As the external load P is gradually increased, the
axial forces in the bars all increase elastically until finally a load Pi is
reached at which some bar begins to yield and thereafter will take no further
load. With further increase in P beyond the value Pi, the truss behaves as

*Throughout the above discussion it is assumed that the deformed configuration of

the structure is not essentially different from the undeformed configuration, i.e., small
change in the angle a due to deformation of the bars is neglected.
 LIMIT DESIGN 37

Fig. 2.12

a statically determinate system. During this phase of loading, the axial

forces in the bars that are still elastic continue to increase until finally a
•second bar reaches the yield stress. Wheh happens, the truss becomes
this
non-rigid and can collapse more or less freely due to steady yielding of the
two bars that have reached the plastic condition. Thus the limit load has
been reached.
To discover which two bars will first reach the yield stress, we consider
the equilibrium conditions of joints A and B and conclude that at all times
we must have
S: = .Si = % and S, = ~ (c)
V2
Thus if each bar has the same cross-sectional area A, it is clear that the
diagonals 2 and 4 will be the two bars to reach the yield stress. When
first

this happens, the system can collapse as shown in Fig. 2.12b and from
equilibrium considerations of joint A, we may write

c = .Si + %• (d)
V2
Expressing Si in terms of S 4 from eqs. (c), this becomes

P —
V2
Finally, setting 84 = 82 — (Ty.p A, we obtain for the collapse load

j-v 20" y p.A —

“^2 \^o’y.p.4. (e)

The above result is based on the assumption that each bar has the same
cross-sectional area A. Thus even in the limit condition, the bars 1, 3, 5
are not working to full capacity, their unit stresses being only see
eqs. (c).Clearly, we may now reduce the cross-sectional areas Ai, As, A# to
A/^J2 without reducing the value of the limit load Py.p. given by eq. (e).
:i8 TExNSION, COMPRESSION, AND SHEAR: li

Therefore, taking jU = Aa = A and Ai = Az == At = A/^J2, we obtain

the optimum limit design of the truss where all bars will reach the yield
condition simultaneously when the applied load P = ^|2<Ty,^,A• Using a
factor of safety n, the corresponding safe working load will be =
^j2(Ty p A/n.
EXAMPLE L A tension member consists of
two pieces of flat steel plate 12 in. wide by |
in. thick connected together by a riveted joint
as shown in Fig. 2.13. The rivets are | in. in
diameter and the material has a yield stress in
shear, Ty p = 20,000 psi. Calculate the safe
tensile load Pu, for the member based on a factor
of safety n = 2 against complete failure due to
shear yielding of the rivets.
SOLUTION. This system is highly statically
indet(UTiunate, and during (dastic behavior of the
n.aterial a calculation of the true distribution of
load among the various rows of rivets represents
an impossible problem. However, as the load P
on the member is gradually increased, one row
of rivets aft<T another reaches the yield condition
in shear and thereafter transmits a constant
amount of load until finally all rivets will have
reached the yield condition. At such time, the
total load that the splice can transmit is

Py, = 10X2(rypA),
where A — 7r(0.87^»)Vd = 0.601 sq in. is the cross-sectional area of one rivet. Thus
the limit load for the spliced member is Py p. = 10 X 2(20,000 0.601) = 240,400 X
lb, and the safe working load {n = 2) becomes Pu — 120,200 lb. This example
shows the advantage of plastic analysis in riveted joints.

PROBLEMS
1. A v(*rtical load P supported by five steel wires symmetrical!}^ arranged as
is

shown in Fig. A and having cross-sectional area A = 0.10 sq in. Calculatti the
i‘ach
limit load for the system if <Ty p. = 40,000 psi and a = 45®, 0 = 30®. Ans. Py p =
16,600 lb.
 LIMIT DESIGN 3U

2. A rigid bar AB
A and supported in a horizontal position by two
is hinged at
steel wires as each wire has cross-sectional area A = 0.125 sq in.
shown in Fig. IL If

and a yield point of 36,000 psi, what is the safe w orking load Pu based on a factor ol
safety n = 2 against collapse due to yielding of the wires? Ans. Pu, — 2250 lb
3. A rigid bar AB of total weight \V — 2000 lb is suspended in a horizontal posi-
tion by three steel wires arranged as showui in Fig. C. If eacli wire has a cross-
sectional area of 0.125 sq in. and a yield point of 36.000 psi, w'hat additional safe load
P can be carried at D if a factor of safety n = 2 is desired against collapse of the
system due to yielding? Ans. = 3833 lb.

\ 4.
^ ^

‘
i

D c A
Qb f <

j
^ '
In

Fig. C

4. A horizontal load P is supjiorted by a bracket consisting of three short st(H‘l rods

iirrangt'd in a vertical plane as showmdn
D. f each rod has a crgs.s-si'ctional area
Fig. 1

o^ 1 sq in., ciilculato the .safe working load based on collajise due to yiidding of t!u'
jnaterial. 'Fake c^^ - 40.000 psi and n ~ 2. \m. Pu - 27,320 lb
5. If the cross-sectional areas of the inclinf'd bars in Lig. I) n'lnain at 1 sq in ,

what IS the least cross-sectional area the vertical bar w’lthout rciliiciug
/li for tlie

safe working load Pw calculated in Problem 4‘^ Ans. .1 0.366 sq in. i

2.4 Strain Energy in Tension and Compression

Let us consider again the case of a prismatic bar in simple tension, Fig.
2,14a.. Assuming elastic behavior of the material, the load-defleotioii
diagram will be a straight line OA as showm in Fig. 2.14b, and for any value
of the tensile load P' the corresponding elongation of the bar is denoted by
5'. Now if an increment of load dP' is added to P\ the elongation 5' wdll

increaseby the amount and the load P' does positive w^ork, P'd5'. This
work, represented by the area of the shaded strip in Fig. 2.14b, is stored in
the bar in the form of potential energy or strain energy as it is more com-
monly called. Subsequently, if the increment of load dP' is removed, the
lower end of the bar moves up through the distance dh' and the stored
energy P'd6' is transformed back into the w'ork of raising the external load
P' through the distance d5'. Thus, we may regard the elastic bar as a spring
in which energy can be stored or released accordingly as the load P' is
increased or decreased. This property of an elastic bar to absorb and release
energy with changes in loading is very important in dealing with time-
varying -or dynamic loads in structures and machines.
40 TENSION, COMPRESSION, AND SHEAR: II

The total energy stored in the bar of Fig. 2.14a under a tensile load P is
equal to the sum between 0 and B and is
of all such elemental strips as P'dd'
represented by the area OAB in Fig. 2.14b. Denoting this strain energy by
U, we have

Since expression (2.3) is valid only if Hookers law applies, we also have,
between P and 6, the relationship:
PI
6 =
AE
Using this, expression (2.3) can be written in either of the following two
forms:
P^l
U -
2AE
or U= 21
(2.4)

XT P'

(a

In the first form, the strain energy i.s represented as a function of the
tension P in the bar; in the second form, it is represented as a function of the
elongation 5 of the bar. Both forms will be found useful in further discus-
sions.
In some applications, the strain energy of tension per unit volume is of
importance. In the case of uniform tension of a prismatic bar as in Fig.
2,14a, this may be obtained simply by dividing expressions (2.4) by the
volume Al of the bar. Thus, using the notation u = Uf Al,

w = 2^ or u = — , (2.5)

where a = P/ is the tensile stress and e = S/tis the tensile strain in the bar.
The greatest amount of strain energy per unit volume that a material can
 8TRAIN ENERGY 41

aljsorb without exceeding its proportional limit can be found by substituting

this proportional limit for a in the first of eqs. (2.5). This quantity is then
called the modulys of resilience of the material in tension. Taking the case of

steel, for example, with a proportional limit cTp.i. = 30,000 psi, we obtain
for the modulus of resiliej^ce ur = 15 in.-lb per iri^

The foregoing discussion of strain energy of a bar in tension can be used

also in the case of axial compression. In such case both a and € are negative,
but since they appear to the square in the expressions for strain energy,
this has no effect on the final result. In short, strain energy is always a
positive scalar quantity.
As an example of an application of dynamic
eqs. (2.4) to a problem of
loading, consider the simple arrangement for producing tension by impact,
Fig. 2.15. A weight W, starting from rcvst, falls through a height h and

strikes a flange at the lower end B of a prismatic bar of length Z, the upper
end A of the bar being fixed. It is desired to find the maximum elongation
and stress that will be induced in the bar by such an impact. In discussing
the problem for elastic deformation, wc assume that the mass of the bar is
negligible compared with that of the the weight IT, and that there is no loss
of energy due to impact between the w^eight and the flajige at B.
After striking the flange, the w^eight W continues to
move dowmwards, stretching the bar as it goes. How-
(^ver, due by the stretched
to the resisting force offered
bar, it decelerates rapidly and soon comes to rest hav-
ing stretched the bar an amount 5. At this instant the
tension in the bar is a maximum and the strain energy
.stored is given by the se(‘ond of eqs. (2.4). Neglecting
minor losses, ihis stored stTain energy in the bar must
be equal to the work done by IT in falling through the
total vertical distance h + d. Thus, we write

W{h +&) = (a)

Introducing the notation Sat = Wl/AE for the static

elongation of the bar under the action of the dead
weight IT, and solving equation (a), we obtain

6 = s.t + = 6., + (b)

where v — yl2gh is the velocity of the falling body at the moment of striking
the flange at B.
If the height h is large in comparison with Sat, eq. (b) reduces to
42 TENSION, COMPRESSION, AND SHEAR: II

The corresponding tensile stress * in the bar is then

bJK Wir
a = (c)
I \Al '

2g

Th(i expression under the radical is directly proportional to the kinetic-

energy oi modulus of elasticity of the material
the falling Ixxiy and to the
of the bar, and inversely proportional to the volume A1 of the bar. Hem^e
the stress can be diminished not only by an increase in the cross-sectional
area but also by an increase in the length of the bar or by a decrease in the
modulus E. This is quite different from static tension of a bar, where the
stress i& independetit of the length I and the modulus IL
Consider now another extreme ease, in which h is ecpial to zero, that is,
the body IE is suddenly put on the flange B without an initial velocity.
Although in this case there is no kinetic ene rgy at the beginning of extension
of the bar, the problem is quite different from that of a static loading of
the bar. In the case of a static tension, we assume a gradual application of
the load and conseciuently there is always ectuilibrium between the acting
lr)ad and the resisting forces of elasticity in the bar. The (tuestion of th('

kinetic energy of the load does not enter into the problem at all under such
condition.s. In the ca>o of a sudden application of the load, the elongation
of the bar aiul the stress in the bar are zero at the beginning, and the
suddenly applied load begins to fall under the action of its own weight.
During this motion the resist ing force of (he bar gradually increases until it
just ei^uals ir when the vertical displacement of the w^eight is But at
this moment th(' load has a certain kinetic energy acquired during the dis-
placement Snt, hence it. continues to move downward until brought to rest
by the resisting force in the bar. Tlie maximum elongation for this condi-
tion is obtained from ec} (b) by setting v ~ 0. Then

5 = 25st, (d)

that is, a suddenly applied load, due to dynamic conditions, produces a

deflection which is twice as great as that obtained when the load is applied
gradually.
The above discussion of impact is based on the assumption that the
stress in the bar remains within the proportional limit. Beyond this limit
the problem becomes more involved because the elongation of the bar is no
longer proportional to the tensile' force. Assuming that the tensile test
diagram does not depend upon the speed of straining the bar, elongation
beyond the elastic limit during impact can be determined from an ordinary
tensile test diagram such as is shown in Fig. 2.16. For any assumed maxi-

*lt is here a«<*umed that the general srhotne of stress distribution under impact is the
same as that under static load.
 strain energy 43

mum elongation 6, the corresponding area OADF gives the work necessary
to produce suchan elongation; this must equal the work W’(/i + 5) pro-
duced by the weight W. When WQi + B) is equal to or larger than the

total area OA BC of the tensile test diagram, the falling body will fracture
the bar.
From this it follows that any change
form of the bar which results
in the
in diminishing the total area OARC
diagram diminishes also the
of the
resisting power of the bar to impact. In the grooved specimens shown in
Fig. 2.18, for instance, the plastic flow of the metal will be concentrated at
the groove and the total elongation and the work necessary to produce
fracture will be much smaller than in the case of the prismatic bar shown in
the same figure. Such grooved specimens are very weak in impact; a slight
shock may produce fracture although the material itself is ductile. Mem-
bers having rivet holes or any sharp variation in cross-section are similarly
weak against impact.
The resistance of a bar to impact also depends upon the ductility of the
material. This may be seen from the tensile test diagrams of Fig. 2.17.
The curve OAC represents the load-deformation curve for a material of
higli strength but low ductility, while the curve OAB is for a material of
lower strength but high ductility. The more ductile material is capable of
absorbing much more energy before rupture than the less ductile material,
as is evider>t by comparing the shaded areas under OAB and OAC. For
this reason, a ductile material has a greater resistance to fracture under
shock loading.

EXAMPLE 1. Three tension members having the dimensions shown in Fig. 2.18
each carry the same t(msile load P. Compute the amounts of strain energy stored in
the three cases, assuming that the stn^ss is uniformly dLstributed over each crOi:is-
section.
SOLUTION. From the first of eqs. (2.4), the strain energy in the prismatic bar,
case (a), is
44 TENSION, COMPRESSION, AND SHEAR: II

wliere A = vd^/4. For case (b), the strain energy is

2(iA)E '
2AE 162AE 16

For case (c), thf‘ strain Riit'rgy is

P\0M) Pn

Ftg. 2,18

Comparison of these three expressions shows that the? strain energy biicomes
smaller and smaller as the cross-sectional area of the bar is increased over more and
more of its length. Thus, as already noted above, a grooved bar is very ineffective
in absorbing energy under dynamic loadmg. It has l)ecn found, for example, that
the life of engine-head stud bolts as shown in Fig. 2.19 can be increased by turning

Fig. 2.19 Fig. 2.20

down the skuk of the bolt to the root diameter of the threaded portion, thus in-
creasing the amount of strain energy that the bolt can absorb under a given tensile
load.
 STRAIN ENERGY 45

EXAMPLE 2. A weight W = 10,000 lb attached to the end of a steel wire rope

(Fig. 2.20) moves downward with a constant velocity = 3 ft per second. WTiat i;

stresses are produced in the rope when its upper end is suddenly stopped? The free
length of the rope at the moment of impact is / = 60 ft, its net cross-sectional area
is A « 2.5 sq in., ]5 X 10‘‘ psi.
and E =
SOLUTION. Neglecting the mass of the rope, the total energy in the system at any
instant is the sum of the elastic strain energy in the rope, the kinetic energy of the
falling weight, and its potential energy relative to some datum plane (here assumed
to be a horizontal plane through the lowest position of the weight). Let 5*1 represent
the static elongation of the rope caused by W, and 5, the total elongation of the rope
when the weight is at its lowest position. If it is assumed that there arc no losses in
the over-all energy of the system, the energy just prior to stopping (when at an
elevation 5 — above the datum plane) may be equated to the energy in the lowest
position. Hence

AE6^
+ J «» + »'(« - fi.t)
21 21

Introducing W obtained from the known relation = Wl/AE into the al)ove
equation, one gets

8 8m +

Lpon sudden stopping of point .1, the tensile strcvss in tin* roi)e increases in the ratio

A-i + jl = .
^
a,. ^S.t\AEg yjgS.,

Thus using the given numerical data, we have

a = 4000(1 + 4.17) = 20,700 psi.

We see that the dynamic stress in this case is more than five times the corresponding
statir stress W/ .

PROBLEMS

1. A prismatic steel bar, 10 in. long, is subjected to axial compressive forces

E = 4000 lb. Compute the amount of strain energy (a) if the cross-sectional area
:

A = 4 sq in., (b) if A = 2 sq in.; E = 30(10)® psi. Arts, (a) V = \ in.-lb.; (b)

U= % in.-lb.
2. A prismatic steel rod of Iqngth I and cross-sectional area A hangs vertically
under its own weight. How much strain energy w stored in the bar if its weight per
unit volume is 7? Ar^s. U = y^AE/^E.
3. Compute the strain energy per unit volume and per pound which can be stored
in th6 following materials in tension without exceeding their proportional limits:
(1) structural steel; 7 = 0.284 Ib/in.^ E— 30(10)* psi, <Tp.\ = 30,000 psi. (2) tool
steel; 7 = 0.284 Ib/in.^, E — 30(10)* psi, o-p.i. = 120,000 psi. (3) rubber; 7 —
0.0336 lb7in.’, E = 300 psi, (Tp.i. = 300 psi.
46 TENSION, COMPRESSION, AND SHEAR: II

4. A vertical load P is supported by three steel

wires each of the same cross-sectional area A (Fig.
A). Compute the* vertical deflection b by equating
the strain energy JJ of the system to the external
work Pb/2 of the load P. Compare the solution with
the one made in Art. 1.4, p. 15. Assume a = 30°.
A ns. b = 0.435 Pl/AE,
5. A 1-lb weight is attached to the end of a nylon
string of length I and cross-sectional area A = 0.01
sq in. The other end of the string is tied to a fixed
support at A, The weight is now dropped from rest
at the level A and falls the full length I of the string
in the gravity field. What maximum tensile stress a
is induced in the string if it stretches elastically
whil(' bringing the falling weight to rest*^ Neglect
the weight of the string and explain why the maximum stress a is independent
of /. The modulus of elasticity for nylon in tension is /? = 4(10)'' psi. Ann. <7,nax'
= 9000 psi.
6. The lOOO-lb hammer of a pile driver freedy falling through a height /? = 3 ft
strikes the top of a wood jiilo 20 ft long and 12 in. in diameter, assumed fixed at its
lower end. Determine the maximum c-omiiressive stress <t induced in the pile if
E ~ 1.5(10)® psi. Neglect the weight of the pile and energy loss(‘s due to impact.
A ns. arnax = 20(X) psi.
7. For the system shown determine the height h through which the
in Fig. 2.15,
weight W should Ik* order to pnidiu^^ a maximum stress a = 30,000
allowed to fall in

psi in the rod. The following numerical data are given: — 30 lb, I =• 6 ft, W
A = 0.5 in E ~ 30(10)® psi. An/?, h = 18 in.
.

8. Referring t,o the system shown in Fig. 2.20, and using all data as in Ex. 2, find
the maximum stress a if there is a short spring bcAwecai tlu' end of the cable and the
load W. Statically, this .spring elongates ] in. per HXX) lb of tension, i.e., it has a
spring constant k = 2000 Ib/in. An,s\ a = 7220 psi.

2.5 Strc.ss ("oneeiilralion in Ten^^oll or Conipres.sion IVlenibers

So far in the diseiission of simple tension and eoinprossion, it has been as-
sumed that the bar has ii prismatical form. Then for centrally applied
forces, the stress at some distance from the ends is uniformly distributed
over the cross-section. Abrupt changes in cross-section give rise to great
irregularities in stress distribution. These irregularities are of particular
importance in the design of machine parts subjected to variable external
forces and to reversal of stresse.s. Irregularity of stress distribution at such
places means that at certain points the stress is far above the average value
and, under the action of reversal of stresses, progressive cracks are likely
to start gradually from such {Kjints. The majority of fractures of machine
part.s in service can be attributed to such progressive cracks.
Stress concentration is a matter which is frequently overlooked by
designers. That the neglect of this factor in much of our engineering design
has not led to mort', frequent disaster is due to the employment of a large
 STRESS CONCENTRATION 47

factor of safety in stress analysis and to the beneficial effect of local yielding
upon stress distribution. This does not justify, however, our disregarding it
in those cases where it does enter. In this article, the importance of this
factor will be illustrated by treating briefly stress concentration as it occurs
in tension or compression members. This abbreviated treatment*** will be
sufficient, it is hoped, to acquaint the student with the occurrence and
effects of stress concentration and to help him recognize those cases where it

must be given consideration.

A simple example of stress concentration occurs in a rectangular plate
with a small ctrcAilar hole at tlio center w'hen subjected to tension (Fig. 2.21a).

111]11 _L_

i
A L
1

TT Tt n
h 1 t* 1 111
(b) (c /

Fig. 2.21

Investigation shows f that in this case there is a high stress concentration

at points m and n at the edge of the hole. The distribution of stresses over
the cross-section through the center of the hole is shown by the shaded area.
If the size of the hole is small compared with the width of the plate, the
magnitude of the normal stress at any point of this cross-section, a distance
r from the center of the hole, is given by the equation!

^
2V 4r* 16 r*) ^

*For a rather complete discussion of this subject, together with references to current
literature, seeTimoshenko’s Strength of Materiala, 3d ed., Part II, Chap. VIIl, and the
article by M. M. Frocht, “PhotoeJastic Studies in Stress Concentration,” Mechanical
Engineering, August 1936. Also see papers in the Proc. Soc. Exp. Stress Analysis by the
following: Durelli and Murray, Vol. I, No, I, p. 19; Berkey, Vol. I, No. II, p. 66;
Hetenyi, Vol. I, No. I, p. 147. For numerous excellent illustrations of failures originating
at pojnts of stress concentration, see Battelle Memorial Institute, Prevention of Fatigue
of Metals, John Wiley and Sons, New York, 1941.
tSee Timoshenko, Strength of Materials, 3d ed.. Part II, Art. 56, p. 301.
tFor derivation, see Timoshenko’s Theory of Elasticity, McGraw-Hill Book Co., 1934,
p. 75.
48 TENSION, COMPRESSION, AND SHEAR: II

in which ao is the uniform stress at the ends of the plate. It can be seen that
the stress concentration is highly localized in this case. At m and w, where
a = 3<ro, the stresses decrease rapidly with increase in the distance from
these overstressed points; at a distance from the edge of the hole equal to
the radius of the hole (r = d), cr = ao only. Due to the bending action
around the hole of Fig. 2.21a, a compressive stress of magnitude o-o is set up
at the top and the bottom of the hole.
In the case of a ^all elliptical hole (Fig. 2.21b), the maximum stress is
likewise at the ends of the horizontal axis of the hole and is given by the
equation

O-rrnx = <^0^1 + 2^^ (b)

Tliis stress increases with the ratio h/c so that a very narrow^ hole perpen-
dicular to the direction of tension produces a very high stress concentration.
This explains why cracks perpendicular to the direction of forces tend to
spread. This spreading can be stopped by drilling holes at the ends of the
crack to eliminate the small radii at the ends of the crack which prodiic(‘
high stress concentration.
Small semicircular grooves in a plate subjected to tension (Fig. 2.21c)
also produce high stress concentration. Experiments show that at points

m and n the stresses are about three times the average stress applied at the
ends of the plate. The standard tensile test specimen for cement {Yig. 2.22)
is another example of a tension member with sharp variation in cross-

section. Experiments show that the maximum stress occurs at points m

 STRESS CONCENTRATION 49

and n and that this stress is about 1J times the average stress over the cross-
section mn.
In the case of a plate having two portions of different widths, the maxi-
mum stress occurs at points m and ri in Fig. 2,23. 'Fhis stress may be
represented by the equation:
O’ n»ax ~ O'

in which <7o is the uniform tensile stress applied at the end of the narrow
portion and K is a numerical factor which depends on the radius r of fillets

and on the ratio b/a. This factor

usually called the /actor of stress concen-
is

tration. Values of this factor are given by the curves in Fig. 2.23.* It will
lie seen that the stress con(‘entration factor increases with reduction of the
radius of the fillet. A more
satisfactory stress distribution at the weak
section can be obtained by using a larger radius of hllet. This fact is now
well recognized in machine design and in shop practice. It is the custom to
specify a minimum radius of fillet to prevent the development, of progressive
cracks at reentrant corners.
All the conclusions reached above concerning stress distribution assume
that the maximum stresses are within the proportional bruit of the material.
Beyond the proportional limit, stress distribution depends on the ductility
of the materia). A ductile material can be subjected to considerable stretch-
ing beyond the yield point without great increase in stress. Due to this
fact, the stress distribution beyond the yield point becomes more and
more uniform This explains why, with ductile
as the material stretches.
materials, holes and notches do not lower the ultimate strength when the
notched piece is tested statically. However, in the case of a brittle material
such as glass, the high stress concentration remains up to the point of break-
ing. This causes a substantial weakening effect, as demonstrated by the
decrease in ultimate strength of any notched bar of brittle material.
The above discussion shows, therefore, that the use of notches and re-
entrant corners in design is a matter of judgment. In the case of ductile
structural steel, high stress concentration is not dangerous provided there

is no alternating stress. In the case of brittle material, points of stress con-

centration may have a great weakening effect and such places should be
eliminated or the stress concentration reduced by using generous fillets.
In members subjected to reversal of stress, the effect of stress concentration
must always be considered, as progressive cracks. are likely to start at such
points even if the material is ductile.

*The8e curves are taken from article by M. M. Frocht, loc. cit.^ p. 47.
 CHAPTER III

BIAXIAL TENSION AND COMPRESSION

3.1 Stresses in Thin- Walled Pressure Vessels

In preceding chapters we have discussed simple

tensiiui or compression of a material in one direction.
Very often we may encounter the case of a plate or
sheet of malerial subjected to tension or compression in
two perpendicular directions at the same time. To see
how such biaxial stress may arise, let us consider, in
Fig. 3.1, a closed thin-walled container having the form
of a surface of revolution and subjected to internal
pressure of intensity p. Such containers as this are often
Fig. 3.1 encountered in stress analysis problems and are called
pressure vessels. If the wall thickness i of such a vessel is
small compared with its principal radii of curvature, the wall will have very
little bending resistance and acts primarily as a membrane in which the
stresses are tangential to the middle surface of the wall and uniformly dis-
tributed across its thickness. Such stresses are called membrane stresses
and are easil^^ calculated from equations of statics. Isolating an element A
of the wall cut out by two and two meridians, we conclude
parallel circles
from symmetry conditions that only normal stresses cri and a2 act on its
edges and we obtain the case of biaxial stress.
To calculate these stresses, we refer to Fig. 3.2 and introduce the follo^v-
ing notations.

<Ti = tensile stress in meridional direction or meridional stress.

(T2 = hoop stress.
tensile stress in circumferential direction or
I = uniform thickness of shell wall.
Ti — radius of curvature of meridian at A .

r2 ~ radius of curvature perpendicular to meridian at A.

dB\ = angle subtended by meridian arc of element.
dd2 = angle subtended by arc normal to meridian at A .

dsi = riddi = dimension of element in meridional direction.

ds2 = r2d02
— dimension of element in circumferential direction.*
50
 THIN. WALLED TRESSURE VESSELS 51

Then the stress resultants acting on the edges of the element are (Tids 2 t and
cr^dsit as shown in Fig. 3.2b. The two .stress resultants in the meridional

Fig. 3.2

direction have a resultant, in the direction of the normal to the element

equal to

(T\ds2t dOi — .
(y,)

In the same manner, the stress resultants in the (urcumfcrential direction

have a normal resultant equal to

(T^dsidS’it
(J2dsit dd^, (b)
n
The sum of these normal forces is in equilibrium with the normal pressure
force on the inside surface of the element ;
thus

aidSidSit aTdSids^t
q
,

— ,
ptt,SioS2>
^
,

'
(C)
ri r2

from which
ffi
_l_
_ p (3.1)
ri rj t

Applications of this formula for thin-walled pressure vessels of various

shapes will now be illustrated by several exampies.

EXAMPLE 1 . Calculate the membrane stressesa i and cr? for a thin- walled spherical
veasel of radius r and v/all thickness t if it is subjected to uniform internal pressure of
intensity p.
SOLUTION. In this case cri = at = tr and fi = ra = r. Then eq. (3.1) at once
reduces to

(d)
52 BIAXIAL TENSION AND COMPRESSION

EXAMPLE 2. Calculate the membrane stresses <ti and cto for a cylindrical tank of
radius r, length I, and wall thickness t if it carries a uniform internal pressure
p (see
Fig. 3.3).
SOLUTION. In this case - co and r2 = r so that eq. (3.1) reduces at once to

This is the same as the result obtained for hoop stress in the case of a thin ring as
discussed in Art. 1.5, p. 22.
To find the princi])al stress ai in the longitudinal direction, we cut the cylinder
in two by a section normal to its axis of revolution and consider the equilibrium of
that portion to one side of this section as shown in Fig. 3.3b. In this case the re-

(Tj 2vrt

(b)

3.3

sultant thrust on the end f)f the tank is pinr^ and this force must loe balanced by the

uniform longitudinal stress <Ti around the circumfenmce of the cylinder. Thus

<Ti • 27rr/ = irr^p,

from which (0

Comparing eqs. (e) and (f), we see that in the case of a circular cylindrical tank sub-
jected to uniform internal pressure, the longitudinal stress cry is just half as large as
the hoop stress o’ 2 .

EXAMPLE 3. An open conical tank uniformly suspended around its upper rim is
filledwith water to a depth h as shown in Fig. 3.4. Calculate the membrane stresses
at the level mn and find the value of y for which each of these stresses will be a
maximum.
.SOLUTION. Since rj = 00 in the case of a conical vessel, eq. (3.1) gives

0-2 = (g)
t
 THIN-WALLED PRESSURE VESSELS 53

where ro is the radius of circumferential curvature at the level mn as shown. Sub-

Ptitutinp, in eq. (g),

7* o = y tan a
COS a
and

p ^ w{k - ?y),

where w denotes the weight per unit volume of

water, we obtain

_ w(h — y) y tan a
(h)
t cos a

for the hoop stress at the level mn. To find the

value of y for w'hich this stress is a maximum, we
set the d^'rivative of expression (h) equal to zero
and obtain

doi w tan a
{h - 2y) = 0 ,
Fig. 3.4
dy i cos a

from whicli y = h/2. Then with this value of y, eq. (h) gives

^ ^
wh’^ tan a
(hO
4^ cos a
The str(‘ss (Ji in the meridional direction at the level inn is found from th(^ condi-
tion that the total weight of th(‘ shaded volume of water in Fig. 3.4 is supiiorted by
the vertical com})on(‘nt of the meridional tension on the circumference mn. Thus

(Ti •
2tij tan a t cos a — ivQt — »y)7r/y^ tan‘‘^ a -f- tan'-* a
3
This gives

(Ti - Winn a/ hy - ~y^\

2 \
I ii)
21 cos a\ ^ )
Again setting dai/dy = 0, we
^’e obtain

dui
Ti wt^na/ 4 \
dy

from which y = 3V4- With this value of //, cq. (i) becomes
tan a
(tr i)inax — (iO
cos a

PROBLEMS
1. What maximum allow^able internal pressure p for a standard 2-in.-
is the
diameter water pipe if the allowable tensile stress is 16,000 psi? The actual
steel
inside and outside diameters of such a pifie an' 2.067 in. and 2.375 in., respectively.
Ans. p — 2380 psi.
2. Calculate the largest safe diameb'r I) for a spherical pressure vessel made of
 54 BIAXIAL TENSION AND COMPRESSION

thin magnesium plate 0.10 in. thick if the allowable tensile stress is 12,500 psi and
the internal pressure p = 20 psi. A ns. — 250 in. D
3. A vertical steel standpipe is 16 ft in diameter and stands 100 ft high. If the
allowable working stress in tension is = 16,000 psi, what is the required wall
thickness Assume that the pipe is filled with water for which w = 62.4 Ib/cu ft.
Ana. f = 0.26 in.
For the conical vessel shown in Fig. 3.4, the following data are given:
4. = 10
ft, a = 22° 30'. Determine the required wall thickness t if the tank is filled with salt

water (iv = 64 Ib/cu ft) and the working stress in tension is Cw == 18,000 psi.
dns. t = 0.0033 in.
5. A truncated conical tank having the dimensions shown in Fig. A is filled with
water (w; = 62.4 Ib/ft^). Calculate the membrane stresses a] and an for an element
A of the wall situated as shown in the figure if t = 0.012 in. Ans. aj - 1620 psi;
0-2 = 7760 psi.

6- A cylindric.al water tank having a hemispherical bottom as shown in Fig. H

hangs from its upper rim and is filled with water. Calculate the membrane stresses
.

a I and a 2 for a small element of the wall at A if the thickness ist = 0.012 in. Ans.
a I = 5720 psi a 2 = 5460 psi.
;

7. A thin hemispherical shell of uniform thickness t and radius R is supported on a

smooth horizontal floor as shown in Fig. C and subjected to its own weight of
intensity g per unit of surface area. Compute the membrane stresses o and a 2 for an ’

element at A. Ans.

Fig. C Fig. D
 THIN-WALLED PRESSURE VESSELS 55

8, A rubber torus inflated to a pressure p = 10 psi has the dimensions shown in

Fig. D. Calculate the membrane stresses on an element at A if a = 12 in., b — 15
in., c — 9 in., t = 0.1 in. Am. ai = 350 psi; co = 150 psi.

3.2 Further Analysis of Biaxial Stress

In the preceding article, it has been shown how an element of material
may be subjected to tensile or compressive stresses in two perpendicular
directions simultaneously. Considering again such an element (Fig.3.5a),
we shall now study this state of biaxial stress in tnore detail. Here, we
denote the two given stresses by cr* in the x direction and jy in the p-
direction and assume that is algebraically the larger of the two.
We may now inquire what stresses exist on a plane whose normal n makes
the angle <#> with the .r-axis, as shown. To answer this question, we isolate,
as a free body, a triangular portion abc of the element as shown in Fig.3.5b.
Let dAx and dAy denote the areas of the faces ab and 5c, respectively, and
dA the area of the face ac. Then the stress resultants on the faces ab and

(a) (b) (c)

Fig. 3 5

be c^dAx and cydAy, as shown. To balance these forces,

ac, respectively, will

we need both a normal stress resultant OndAn and a shear stress

shall
resultant rdA „ on the inclined face ac of the element as shown. Noting that
dAx — dAn cos and dAy = dAn sin <t> and projecting all forces on the n-
axis, we must have, for equilibrium of the element,

OndAn = o-x(dAn cos (ff) COS 0 + (Ty{dAn sin </)) sin <#>,

from which
cTn = o-x cos^ A- <Ty sin* + o-y) + J((rx — ay) cos 2<l}. (3.2a)

In the same way, projecting £^1 forces on the element on the direction ac,
we obtain
rdAn = (Tx(dAn cos 4>) sin ^ — ay{dAn sin </») cos <t>.

from which
T = (<r, — ay) sin </> cos = ^(cr* — ay) sin 2 (p. (3.2b)
56 BIAXIAL TENSION AND COMPRESSION

Thus expressions (3.2a) and (3.2b) give the magnitudes of normal stress (r„

and shear stress t on any plane, the orientation of which is defined by <l>.

If we take an angle <(>' = 90° + <l> and remember that sin (90° +<<>) —
cos (p while cos (90° +<?>)== —sin eqs. (3.2a) and (3.2b) become

crj = (Tx sin^ <t> + ay cos^ (p = ^{<Tx + ay) — i(<^r — ay) cos 20, (3.3a)

r' = —{ax — ay) eos 0 sin 0 = — ^{ax — ay) sin 20. (3.3b)

These expressions represent the normal and shear stresses on a plane per-
pendicular to the plane ac (see Fig. 3.5c). As discussed in Art. 2.1, the
stresses given by eqs. (3.2) and by eqs. (3.3) are called complementary
stresses. In this more general case of biaxial stress, the sum of comple-
mentary normal stresses an and an is again constant and equal to a* + ay,
the sum of the two given stresses. Likewise the complementary shear
stresses r and t' are seen to be equal in magnitude but of opposite sign.*
We continue to use, for shear stress r, the sign convention discussed in
Art. 2.1: that is, a shear stress on any face of an element is positive when
its sense of rotation about a center inside the element is clockwise. Thus
the negative shear stress r has been shown in its proper direction ni
Fig. 3.5c.
Further examination of eq. (3.2a) will show that as 0 varies from zero to
ir/2, the normal stress cr„ varies from ((rn)max = ax, when 0 = 0, to ((rn)m»n =
Oy, when 0 = 7r/2. Thus the stresses ax and <Ty represent, respectively, the
maximum and the minimum values of normal stress and are called the
principal stresses. Similarly, it will be seen from eq. (3.2b) that the shear
stress T is a maximum when 0 = 7r/4 and that the magnitude of this maxi-
mum shear stress is

^luHx ~ iKax ^i/)j (3 *1)

i.e., half the difference between the two principal stresses. If the two
principal stresses are equal {ax == ay), there will oe no shear stress on any

plane such as ac in Fig. 3.5b.

With proper changes in sign, eqs. (3.2) and (3.3) can be used also if one
or both of the principal stresses ax and ay should be compression, i.e.,

negative. If one principal stress is tension and the other is compression,

then the tensile stress will be taken as a* and the compressive stress as ay.

If both principal stresses are compression, the smaller one should be taken
as ax, since it has already been assumed that ax is algebraically greater
than ay.

So far, we have considered only the state of stress for an element sub-
jected to biaxial tension or compression. We shall now discuss briefly
the strain or deformation of such an element. When there are principal
stresses ax and ay in both principal directions x and y of the element (Fig.
3.5a), the strain in either of these directions will depend not only upon
 ANALYSIS OF BIAXIAL STRESS 57

the stress in that direction but also upon the stress in the orthogonal
direction because of the Poisson ratio effect discussed in Art, 1.3. In the
X direction, within the elastic limit, the plate will have positive strain
= ffj/E and at the same time it will have negative strain €* =
due to lateral contraction associated with the tensile stress o-y. The same
reasoning holds for total strain in the y direction. Thus in the case of
biaxial tension of a thin plate, the total or net strains will be

MtTj/ ]

E~~E' i

(3.5a)
I

E E ‘

In the ^-direction normal to the plane of the plate there will be lateral
contraction due to the Poisson ratio effect associated with each of the princi-
pal stresses cr* and ay so that the net strain in this direction becomes

-h (Ty). (3.5b)

Very often in experimental stress analysis the principal strains ti and ey

will be measured directly by strain gages. Then the corresponding stresses
<T* and ay can be calculated from ec4S. (3.5a), which, for this purpose, may be

put in the form

O’x *“ *
^
1 - #1
2
*

(3.6)
_ («» + f

If the plate carries compression in either principal direction, eqs. (3.5) or

(3 ()) can still be used by taking the corresponding tr or e with negative sign.
EXAMPLE A thin circular steel plate of radius r and thickness t is subjected to
1.

radial stress a unifornxly distributed around its circumference as shown in Fig. 3.6.
Determine the state of stress on any element such as A and also the unit volume
change of the entire plate.
SOLUTION. Since both the plate and the
external loading are symmetrical around the
center 0, it follows that the deformed plate
will remain perfectly circular in form. This
means that both the radial and circumferential
strains must be uniform and equal throughout
the plate. Then it fallows from eqs. (3.5a) and
reference to the elements A and B in Fig. 3.6
that (Tx = cTy = (T. Thus we have the case of
equal principal stresses.
Before deformation, the volume of the plate
is Fo *= ttH. After deformation, the new
radius is r(l -f €,) and the new thickness is
FI +€,). Thus the new volume is F = Fio. 3.6
58 BIAXIAL TENSION AND COMPRESSION

wrHil -h -f (z)- Exiianding this and keeping only terms of first order in
the small quantity e, it reduces to E = TrrH(l +
€ 14* ^x), and the total volume

change becomes AV = E — Vo = TrrH(€t +

2c*). The corresponding unit volume
change is AV/Vo = 6^ 2€r. Now for <r* =
+ = c. eqs. (3.5a) give

Cr = €y = - (1 ~
and eq. (3.5b) gives

With these values for strains, the unit volume change becomes

AE 2g
;(1 - 2m ). (a)
Eo

Taking E = 30(10)'’ psi. a - 30(10)^ psi, and m = i* this gives

AE a
^ 0.001
E” 30(U))«

I'RO HLKMS
1. Determine the stresses Cr,, c'n, t, and r' for the element in Fig. 3.5 if cTt =
10.000 p.si, (7y = —5000 psi, and 0 — 30°. A ns, cf„ = 6250 psi; ej'n = —1250 psi;
r =: — r' = 6500 psi.
2. D(dt‘rniine tli(‘ stresses cr„, o-'„, r, and r' for tl\(‘ olennuit in Fig. 3.5 if a* —
-5000 psi, (Ty - - 10,000 psi, and 0 - 30°. Ans. = -6250 psi; a'n = -8750
psi; T — — t' = 2165 psi.
3. For the <‘lement in Fig. 3.5, the following values an* given for the |)rincipal
stresses: a* = 4000 psi, Uy = 3000 psi. Calculate the value of the angle’0 for which
the shear stress r will be a maximum and evaluate the corresponding stresses (r„, a'n,
Tmax- Ans. (7n = (t' n = 3500 psi Tmax = 500 psi. ;

4. Using the data given in Problem 1, calculate the principal strains c* and ty if
E = 30(10)« fKsi andM = 0.3. Ans. e* = 383(10)-«; c^ = -267(10)'®.
5. A rectangular plate subjected to biaxial stress is found to have principal
strains c, = 0.00100 and €y = — 0.0(X)70 as determined by strain gages. What are
the corresponding principal 20,000
stresses if = 30(10)® psi and m = 0.3? Am. o’* =
26.000 psi; Cy = —13,160 psi,
6, A cube of concrete is compressed in two perpendic-
ular directions by the arrangement shown in Fig. A.
Calculate the unit volume change for the cube if it is
4 in. on a side. Assume that the compressive stresses
are uniformly distributed over its faces and take P =
lb, M = 0.1, and E = 4(10)® psi. Ans. AV/V ==
-0.000707.
7. A thin circular brass membrane is suppdi;ted around
its rim by a rigid circular brass ring and at rtoifi tem-

perature (70°F) the membrane is free from stres^ What

principal stresses o-* and cry will be induced in the mem-
brane if its temperature falls to zero degrees? Assume
 ANALYSIS OF BIAXIAL STRESS 50

that E - 14(10)® psi, m = 0.3 and a = 10.4(10)“® in./in./F°. Arts. =

i 4,550 psi.

8. The thin-walled in Fig. 3.3, p. 52, has radius r = 20 in.,

cylindrical tank shown
wall thickness t = isubjected to internal gas pressure p = 110 psi.
in., and is

Calculate the magnitude of shear stress t along a 45°-helix. A/w. r = 4400 psi.
9. Derive a formula for the maximum shear stress T,„ax in the plane of the wall of
the conical tank filled with water as shown in Fig. 3.4, p. 53. Ans.

_ Swh^ tan a
rmnx —
64f cos a

3.3 Mohr’s Circle for Biaxial Stress

Contiider again the rase of a thin plate subjected to biaxial tension as

shown in 3.7a. Then as discussed in Art, 3.2, the normal and shear stresses
on any plane whose normal n makes an angle 0 with the .r-axis will be given
by eqs. (3 2) repeated below for convenience of reference.

+ (Tt,) + ^(<rx — (Ty) COS 20,

(a)

r = - ay) sin 20.

Ir can easily be shown that these are the equation.s of a circle in a a-7 plane,
with the angle 0 as a parameter. Introducing the notations

“b ~ rniaxi (b)

we can write ocjs. (a) in the more compact form

(T n “ O'fxv H” T'n'.ax COS 20,
(c)
T = r„,ax sin 20.

Then to eliminate the parameter 0, we note from the second of eqs. (c) that

cos 20 = .^1 — sin^ 20 = -^1

—

Substitution of this in the first of eqs. (e) gives

^11 ^BV — (d^

or (<rn *“ (T^y)^ + = Tmax^.

This is the standard form for the equation of a circle having a radius
Tmax = —
and Centered on the cr-axis at <r„v =
ffy) + (^v) as shown
in Fi^ is called MohEs arch for biaxial stress and it is very
3.7b. This
useful as a graphical means of solving eqs. (a).
Consider, for example, any point D on the circle and denote the angle
(W BIAXIAL TENSION AND COMPRESSION

ACD by 2<t> as shown. Then from geometry it can be seen that the co-
ordinates of point ]) are

()E ~ OC + Cl) cos '2<t>

== |(<7x 4" 4' — (^y) cos 2<^,

1)E - CD sin 20 = |(crx ~ <ry) sin 20.

Thus the coordinates of point J) in Fig. 8.7b represent the values of <Tn

and T for th(‘ defined by 0 in Fig. 3.7a.

plane whose aspect is

For each different aspect of plane in Fig. 3.7a as defined by 0, there is a

corresponding point on the circle, the coordinates of which represent the
normal arid shear stress For example, when 0 = 0, the
on that plane.
normal and point D on the circle coincides with
n coincides with the j-axis
point A, giving cr„ = cr^ and r = 0. Whfui 0 = 45"", 20 = 90° and point I)
falls at F giving = ^((Tx + (^y) and
a-,, ). When
0 = 90°, the normal n coincides wi4i the //-axis and the corresponding
position of point 1) is at B on the circle indicating On = <^y and r = 0. It
will be noted also that the coordinates of point D' diametrically opposite
point D on the circle give the complementary stresses On and r' as defined
by eqs. (8.8). Thus all possible information about the stresses on various
planes can be found from Mohr\«5 circle.
If either principal stress is compression, it must be taken with negative
sign so that in general the center C of Mohr's circle may lie to either side of
the origin in Fig. 3.7b, but always on the <T-axis. Several particular cases
are illustrated in the following examples.

^:xAMPLE 1 Construct Mohr's circle for the case of biaxial stress of a thin plate
.

where Cy = —ax as shown in Fig. 3.Sa.

SOLUTION. In this case a^y — i(ax -f ay) = 4((t, — (Tx) = 0 and the center C of
the circle falls at the origin, while its radius Xmax == — <Ty) = J(2o’x) = a^.
Thus Mohr's circle will be as shown in Fig. 3.8b.
 MOHR'S CIRCLE FOR BIAXIAL STRESS 61

(a) (b)

Fig. 3.8

EXAMPLE 2. Construot Mohr’s circle for the case of simple tension where cr^ = 0,
as shown in Fig. 3.9a.

(a) (b)

Fig. 3.9

SOLUTION. In this case, - i (<^r + cr^) - and tjie circle is tangent to the r-
axis asshown in Fig. 3.9b.
EXAMPLE 3. Construct Mohr’s circle for the case of biaxial stress shown in Fig.
3.1 where o-j is tension and cry is compression. Assume \<Ty\ ~ 2 |cr,| .

Fig. 3.10

Thus Mohr’s
+
SOLUTION. In this case <r»y = §((t, a,) = -iffxandrm.x
circle will be as shown in Fig. 3.10b.
= i(ffi - O=
G2 BIAXIAL TENSION AND COMPRESSION

PROBLEMS
1. Construct Mohr’s circle for the particular case of biaxial tension where
(J X “
2. A having radius r = 5 ft and wall
circular cylindrical steel boiler (see Fig. 3.3)
thickne.vs t = subjected to internal pressure p = 100 psi. Using Mohr’s
i in. is

circle, find the normal stress (r„ and the shear stress r on the edge of a helix which
makes an angle <t> = 60° with a generator. Arw. <Tn = 10,500 psi; r = 2600 psi.
3. Construct Mohr's circle for the case of biaxial stress shown in Fig. 3.7, if
(Xi ~ 10,000 psi o-y = —5000 psi. From this circle, find the stresses (r'n, t, r', for

an clement so onentaU'd that (p - 30°. Ans. (t„ = 6250 psi; (r\ = —1250 psi;
T - — t' = 6500 psi
4. Find the normal and shear stresses on the element of the preceding problem if </>

IS so chosen that the shear stress is a maximum. Ans. ffr - — = 2500 psi;
T - ~ r' = 7500 psi.
5. Referring to the conical water tank shown in Fig, 3.4, assume that a = 30°,
= 12 ft, t = 0.05 in. and w = 64 Ib/cu ft. Construct IMolir’s circle in such case
for an element of the w’all, the location of which is defined by y = ih, and find the
magnitude of normal stress (Xn on the plane of maximum shear stress at this point.
Ans. (Tn — 2130 psi.
6. Referring to Fig. 3.10 and assuming cr^ = 10,000 psi, cry = —20,000 psi, find
the value of defining the plane on which an vanislies. What is the magnitude of
shear stress r on this plane? Ans. 4> = 35° 15': r ~ 14,100 psi.

3.4 Pure Shear

Let us consider now the pariicular case of biaxial stress where a^ —ay
as shown in Fig. 3.11a. In such ease, Mohr's circle wdll be as shown in
Fig. 3.11b. From points F and Fi on the circle, it is seen that the maximum
shear stress on 45 '‘-planes in r^ax — Azaj and also that the normal stresses
On and o'n on these planes vanish. This means that the. square element
Gcbd oriented at 45° to the directions of principal stress is subjected to shear
stresses only on its edges and is said to be in a state of pure shear. This
parti(‘ular state of stress is of special interest and it will now be examined in
further detail,
Consider first, the deformation of the element acbd shown to a larger

(b) (c)

Fia. 3.11
 PURE SHEAR 63

scale in Fig. 3.11c. Since there are no normal stresses on its edges, these
edges will remain unchanged in length during deformation of the element.
The horizontal diagonal ab will simply elongate and the vertical diagonal
will contract so that, after deformation, the element will have the form of a
rhombus as shown by dotted lines. Thus, the original right angles at a
and b become t/2 — 7 and those at c and d become 7r /2 + 7, wher 7 defines
the amount of angular deformation of the element. This quantity is a pure
number and is called the shearing strain, analogous to the tensile or com-
pressive strain e as discussed in Art. 1.3. To visualize this stress condition
more readily, we rotate the clement acbd by 45^ and place the edges b'd'
and bd in juxtaposition as shown in Fig. 3 12. Then the angle rbc' seen to i-i

represent the previously defined 7 and its value is

cc' b
~ be
~T
where 5 is the lateral displaccMiU'Ut ot the
upper face of the element relative to the lower
faceand is tlie distance between the.se faces.
/

Thus the analogy between shearing strain 7

and extensional strain e is complete, except
that in the case of shear the displacement b
takes place at right angles to the length / so
that we see the ratio b'l as an angle.
Within the elastic limit of a material, it is
reasonable to assume that the shearing strain
7 is proportional to th(‘ shearing stress r that
Fig. 3.12
produces it. In fact, experiments with
materials in pure shear show^ this to be the case, so that

where the constant of proportionality, denoted by G, is called the modulus

of elasticity in shear or simply the shear modulus. Like the tension or com-
pression modulus it has the dimension of stress, Ib/in.^
In the case of pure shear, there will be no change in volume of an element
during deformation. To show this, wt must return to eqs. (3.5) defining
ihe strains € 1 , ty, e-j, in the case of biaxial stress. "Then for <ji
= —ctu — r,

these expressions become

ex = ^ (1 + m), ty = - (i + m), €* = 0. (3.8)

g
Then the unit volume change AVfV = + ey + e, — 0.

It is seen from Fig. 3.11a that the linear strains €, and in the directions
04 BIAXIAL TENSION AND COMPRESSION

of the diagonals ab element <ichd must be geometrically related

and cd of the
to the shearing strain 7. Out of this, comes an important relationship
between the shear modulus G and the tension modulus E, which will now be
established. Fsing the first two of eqs. (3.8) for linear strains and referring
to Fig. 3.11c, we see that

oa' — (>a(\ + ex) = oa I + (I +m) ,

I

(a)

oc/ = OC(l -f €.,) = 1 (1 + /x)

Alsc; from the geometry of the right triangle c'oa'y we have

1 - (1 + m)
or'
tan oa'r/ ~ tan (b)
oa'
f
^
(f + m)
since oa = oc in eqs. (a). Finally, from trigorjometry,

tan 7 - tan ^
4 2 1 - 7/^
tan .

(c)
1 +y/2
1 + tan j tan ^
4 2

.since tan and (an y/'2 ~ 7/2, the angle 7 being very small.
r/4 - 1

Comparing expressions (b) and (c). we conclude that

’2
- g (1 + .1 (d)

Substituting for y its vnliK' from e((. (3.7), thi.s becomes

i; S
E
fioiu which G ^ ^TT—;
-2(1
—
+ m)r (3.9)

This theoretical relationship between the shear modulus G and the

tension modulus E is in good agreement with that found by experiment.
Taking, for mild steei, E = 30 (10)® psi and m = 0.3, we find G = 1.5 (10)® 1

psi. If Poisson’s value of the ratio ^ = 0.25 is used, eq.

theoretical
(3.9) gives G = 12 (10)®psi. It will be shown in the next chapter that the
modulus of elasticity in shear can bo determined experimentally by twisting
a circular shaft. Such experiments with structural steel specimens show
the value of G to liewithin the above limits, '
 PURE SHEAR i\5

PROBLKMS
1. Jri Fig. 3.11a, (Tx =
—o-y = 20,000 psi. Calculate the magnitude of the shear-
ing strain y element acbd, if G = 12(10)® psi. Am. y = 0.00167.
for the
2. A cIos(h1 right circular cylindrical tank 10 in. in diainetci is made of J-m*
steel plate and is subjected to an inti^rnal pressure of intensity p = 5(K) psi. Cal-
culate the maximum shearing strain y in the wall if K ~ 30(10)® psi and Poisson's
ratio M = Am. y - 0.433(10)”^:
0.3.
3. Referring to Fig. 2.21a on p. 47, eah-ulate the magnitude of maximum shearing
strain y at the circumlercnce of the small circular hoh^ if the axial tensile stress is
(Tj
- 10,000 psi and p = 0.25. Aris. y = 0.00125.

3.5 Riveted and Welded Joints in Pressure Vessels

In the actual fabrication of vari-

ous kinds of pressure vessels as dis-
cussed in Art. 3.1, it is usually neces-
sary to have one or more joints or
seams as shown in Fig. 3.13. Such
joints may be either riveted or welded
depending upon the material and the
service conditions to which the vessel
will be subjected. The general ob-
jective is to obtain a joint that will be
tight and as strong as possible.
Riveted joints for boilers and other
pressure vessels are of two kinds: lap joints and butt joints as illusti'ated
in Fig. 3.14a and 3.14b. The rivet.s in Mich joints are driven red hot so that,
after cooling, they scpuMv.e the plates tiglilly together. Thus when intt*rnal

pressure is applied to the shell, ndativi* motion between the plates is pre-
ented by friction as well as })y the shearing strength of the rivets. In fact,
ouK after the friction is overcome do the rivets begin to work in shear. We
see that the behavior of a riveted joint under load is extrmnely complex. To
simplify the problem, it is usual practice to neglect the friction completely
and to assume that the riv(‘ts carry the load in shear with the shearing stress

Fig. 3.14
66 BIAXIAL TENSION AND COMPRESSION

uniformly distributed over the cross-section of each rivet as discussed on

p. 5.
Besides shearing of the rivets, there are other ways by which a riveted
boiler joint like those in Fig. 3.14 can fail after friction has been overcome.
These possible modes of failure are illustrated in Fig. 3.15. As already dis-

Fig. 3.15

cussed, the rivet may fail in shear across the section aa, Fig. 3.15a, al-
though it may undergo considerable bending before this happens. The
resistance of the rivet lo shear failure can be increased by increasing its
diameter. A second possible mode of failure is shown in Fig. 3.15b, where
cru.shing of the plate in compression just behind the rivet allows the joint
to open up. Resistance to this type of faih4re can also be increased by using
a larger diameter for the rivet or by increasing the thickness of the plate.
If the rivets are too -‘losely spaced, the net section of the plate between
rivet holes will be so much nKluced that failure of the joint may take place
due to tearing of the plate in tension between rivets as shown in Fig. 3.1 5o.
Resistance to such failure can be increa.sed by increasing the spacing or 'pitch
p of tlie rivets along the .seam (see Fig. 3.13).
The problem of design of a riveted boiler joint consists of adjusting the
plate thickness, the rivet diameter, and the riv^et pitch so that the joint is

equally strong against each mode of failure. For such design of steel pres-
sure vi^ssel joints, the ASME Boiler Code recommends the following ulti-
mate stresses to be used with a suitable factor of safety, say n = 5.

tension: at = 55,000 psi,

shear: t = 44,000 psi,

crushing: Jc = 95,000 psi.

Obviously, the strength of a riveted boiler joint can never be as great as

that of the shell itself. The ratio of the strength of the joint to that of the
shell proper is called the efficiency of the joint.

EXAMPLE 1. The cylindrical container .shown in Fig. 3.13 is made of steel plate
of thickness ^ rivets are fj in. in diameter and the pitch p = If in.
} in. The
Calculate the ultimate strength of the joint and its efficiency, using A.S.M.E.
Boiler Code specifications.
SOLUTION. Referring to Fig. 3.14a, let us consider a length of joint equal to the
 JOINTS IN PRESSURE VESSELS 07

riv(4 pitch p = 1.625 in. 'fhcHi for this length f)f iiiiflisturl)e(l shell the strength is
55,000 X 0.25 X =
22,300 lb. The croM.s-sectioiial area of one rivet is 0.371
1.625
in.’^ and the com\sporiding shearing strength is 44, (KK) X 0.371 = 16,300 lb. Tlie
projected area behind one rivet is 0.25 X ji - 0.172 in.'-', and the corresponding
crushing stnnigth is 05,0(K) X 0.172 - lti,3(K) lb. The net cross-s('ctJonal area of
plate between rivet holes is 0.25(1.625 — 0.68S) — 0.234 in.'^ and the corresponding
U'aring strength of the joint i.s 55,000 X 0.234 - 12,000 lb. This is the smallest
value and therefore reuresents the strength of th(‘ joint. I'he corresponding efficiency
is12,900 22,300 = 0.58 or 58 per cent.
The efficiency of the joint can be improved by increasing the rivet pitcli so that
the tearing strength comes up to the strengths in shear and crushing. The equation
for determining the optimum pitch is

(p
- 0.688)0.25 X 55,000 = 16,300,

from which p = 1.875 in. The strength of this length of undisturbed shell is

55,000 X 0.2.5 X 1.875 ~ 25,800 lb. Hence the new efficieney becomes 16,300
25,800 = 0.63 or 63 per cent.

Welxied Joints. With the pre.seiit day advaruH'S in welding techniques,

welded joints for pressure vessels and structural connections are rapidly re-
placing riveted joints. Typical lap lotdds and butt undds for the longitudinal
seam of a pressure vessel are shown in Tig. 3,10. Tu the (‘.ase of a lap weld,

( a ) Lap Weld ( b ) Butt Weld

Fig. 3.16

Fig. d.lOa, the thinnest section through the weld bead is called the throat
of the w^eld. The lengthbead is called the length of the weld. The
I of the
product of the throat dimension and the length of the bead gives the cross-
sectional area of the weld to be used in calculating its strength. In the case
of a butt weld, Fig. 3.16b, the thickness t of the plate would be taken as the
throat dimension.
The calculation of the strength of a welded joint is largely empirical
because of the difficulties invDlved in making any rational analysis owing
to the presence of stress concentrations. It is usually assumed that the
strength of a weld the cross-sectional area of the throat multiplied by an
is

arbitrary working stress. The Code for Fusion Welding specifies a working
stress of 16,000 psi in tension or compression for shielded arc welds. If the
weld is subjected to shear, the working stress is 13,600 psi.
 BIAXIAI. TEXSIOxN AND COMPRESSION
EXAMPLE 2. A 6 in. X 6 in. X i in.
iinglp-section t»‘nsionmem})er is welded
to a gusset plaU' by two side fillet welds
‘2*
T- which act in direct shear as shown in Fig.
°2
3.17. The total tensile force in the mem-
i
Q
( her IS F —
103,500 lb and each weld bead
has a throat dimension t = 0.353 in. If
1

“'Xi- — '^filler
the working stress for the weld metal in
shear is = 13,600 psi, what lengths h
and h should the two weld beads have?
Fig. 3.1< SOLUTION. The line of action of the ten-
sile force P acts through the centroid of
the angle section defim'd by the distances a, 1.6S in and = 4.32 in. In
order to balance the load /^ the resultant of the two WTld resistances must have
the same line of action as P. Then if these two forces are denoted by Q\ and Q 2 it ,

iollows that the following condition^: of e()uilibrium must be fulfilled:

Q. + - P,

QiOi

from which, with tin' given numerical data.

Qi - 74,50011). Q 2
=- 20,00011)

Since th(‘ .^h(‘aring resist auee ol eaeh beail jx'r inch of length is

«/
== 0 353 > 13,000 - 4.S()0JI>/in..

the eorresjionding lengths h and 1 2 should be

74,500 29,m)0
/, 15.5 m ,

4806”
6.05 in.
"isoo

P HO R L I’ M^

,
1. A 4 ill X I m. xtei'l stra)) is spliced as shown h’lg The rivets are in. in m 1

diameter. '['he allowable \\f»rking stresse\s are Tu — 10,000 psi in shear and
jTu = 16,000 psi m tension. C aleulah* the safe load P for the spliecal strap and tin'
efficiency (4 tiie joint, l/i.v. P ™ 47,100 lb; eff. == 73.6%.
2. Referring to Fig, 2.13, p. 3S find the yield limit load P for the spliced ine-nber
if the rivet row I is omitted. What is the efficiency of the joint under these condi-
tions? Assume (7, p
- 40,000 i)si. r, ,>
-- 20,000 psi. Ans. Py p
= 216,000 lb,
('flf. - 60';h
Referring to Pig, 3.17, find the number of |-in.-diameter rivets required for
3.
this connection if shear is assumed to govern and the working stress in shear is
Tu, — 10,000 psi. Assunu' that the rivc'ts are so arranged that the centroid of the
group lies on the line of action of the tensile load A 71 S. 17 rivets.
4. A water tank made (4 J-in. steel plate has a riveted lap joint. The
dianieter rivi'ts a-e arranged in three rows. The pitch in the two outer rows is 5 in.-
that m
the inner row is 21 m. Using working stresses Gu, ^ 16,000 psi in tension
and Tw — 12,000 ])si in sliear, find the safe internal pressure p if the tank is 4 ft in
diamettT. An.v. p = 207 psi.
 5, JOINTS IN PRESSURE VESSELS 69

A spiral-riveted penstock 40 in. in diameter is made of steel plate ^ in. thick.

The spiral seam is a single riveted lap joint with rivets, each rivet having a
safe resistance to shear of 5000 lb. The pitch of the spiral (helix) is 7 ft. What is

the required rivet spacing if the internal water pressure is = 160 psi? Assume
that there is no longitudinal load on the pensi-ock. Atis. 2.81 in.

6. A double-riveted butt joint for a 6-ft-diameter steam boiler has the following
dimensions: rivet diameter ^ in,, outer row pitch = 5j in.^ inner row pitch
= 2J in., thickness of main plate = J in., thickness of cover plates = A
in. Using
ASME Boih^r Code recommendations, find the mode of failure, the iiltiniate internal
pressure p, and the efficiency of the joint. Ans. p — 640 psi; efT. — 83.7%.
7. A steel plate 4 in. wide by | in. thick is connected to a gusset plate by means
of two .side filUd welds as shown in Fig. B. Calculate the required bead length I
of th(' welds if P — 48,000 lb and the working stre.ss for the wxJd metal in shear is
r.. =-' 13,G(K) psi. Ans. I - 6.67 in.
8. A st(‘el bar of width b and thickness t is to be \v(3ld(*d to a gusset plate by means
of two side fillet wedds as shown in Fig. B. Using the same working stress in shear
for the w'cld miTal as in the Problem 7, calculate the bead length I of each fillet to
develop a working stress of (r„, = 1 6,000 psi for tension in the bar. / =• 0.8336.

9. A 5 in. X 3^ in. X i in. angle section is to be welded to a plate by side fillet

welds as shown in Fig. 3.17, The 5-in. log of the angle ii-on lies against the plate.
If 7^ ~ 50,000 Ib and acts along the centroidal axis of the angle section, what are
the required lengths h and I 2 of the beads? Ans. li — 9.42 in I 2 — 4.43 in. ;

10 If, in the pn'ceding problem, the weld bead at the ht^ol of the angle iron has a
.

Uin. leg dimension and that at the toe has a f-in. leg dimension, calculate the re-
(juired lengths h and I 2 of the tw^o welds. Ans. li = 7.06 in.; I 2 = 4.47 in.
 CHAPTER IV

TORSION

4.1 Torsion of a Circular Shaft

CorMcier a circular shaft built Jii at the upper end and twisted by a couple
applied to the lower end (Fig 4.1a). ft (;au be shown by measurements at
the surface that circular sectioris <)f the sliaft remam circulaT durjrig twist,
and that their diameters and the distan(‘es between them do not change
provided the angle of twist is small.
AdisC'Iike element of the shaft, such as that adjacent to the section mn
;tnd shown as a free body in Fig. 4.1b, will be in the following state of strain
There will be a rotation of its bottom cross-section with reference to its top
through an angle d<t>. A thin element abed of the surface of the disc whose
sides were vertical before strain takes the form shown in Fig. 4.1b. The

lengths of the sides remain essentially the same and only the angles at the
comers change. Thus we may conclude that the element is in a state of
P'ure shear (see An. 3.4) and the magnitude of the she^aring strain, measured
by the angle cac'. is given very closely by
e’e

70
 CIRCULAR SHAFT 71

Since c'c is the small arc of radius r subtended by the angle c'c r d<t>.

Thus
__ c'c _ r d4>
(a)
^ Ojc' dx
For a shaft twisted by a torque at the end, the angle of twist 4> is projX)rtion-
al to the distancex of the cross-section from the fixed end and hence d4>/dx
is a constant. This constant represents the angle of twist per unit length of the

shaft and will be called 6, Then, from (a),

y = rO. (b)

The shearing stresses which act on the sides of the element and produce the
above shearing strain have the directions shown. The magnitude of each,
from eq. (3.7), is
T = Gy = Grd. (4.1)

So much for the state of stress on an element at the surface of the shaft.
As for the state of stress within the shaft, the assumption will now be
made that not only the circular boundaries of the cross-sections of the shaft
remain undistorted but also that the cro,ss-sections themselves remain plane
and rotate as if absolutely rigid that is, every diameter of the cross-section
;

remains straight and rotates through the same angle. Tests of circular shafts
show that the theory developed on this assumption is in very good agree-
ment with experimental results. Such being the case, the discussion for the
element abed at the surface of the shaft (Fig. 4.1b) will hold also for a
similar element within the shaft, whose radius p replaces r (Fig. 4.1c). The
thickness of the element in the radial direction is considered as very small.
Such elements are then also in pure shear, and the shearing stress on their
side is

r = GpS. (4.2)

This states that the shearing stress varies directly as the distance p from
the axis of the shaft. Fig. 4.2 pictures this stress distribution in the plane of

Fig. 4.2 Fig. 4.3

 TORSION

the cros.s-sectif)!! .'tnd ulso the (‘ornpIcrnoDtary shearing stresses in an axial

plane (see p. ‘JO). The inaxiinum stress occurs in the outer surface of the
shaft, where p --- r.

For a ductile niatei lal, plastic flow begins first in (his outer surface. For a
material which is weaker in shear longitudinally than transversely for -

ins(,an(‘(\ a wooden shaft with the fibers parallel to the axis the first -

('racks will be prodiaa^d by shearing stresses acting in 1 he axial sections and

flaw will appear on the surface; of the shaft in the longitudinal direction. In
the case of a inaterial which is weaker in tension than in shear ' for
instance, a circular shaft of cast iron or a cylindrical piece of chalk —a
crack along a helix inclined at 45° to the axis of the shaft often occurs
(Fi^r 'ITe explanation is simple. The state of pure shear is etpiivalent
to one of tension m one direction and ecpial compression in the perpendicu-
lar direction (see Art. 5.1) A rectangular element cut from the outer
layer of a twisted sliaft with sides at 45° to the axis will be subjected to such
stresses, as showm in lug. 4 5. The tensile stresses shown produce the heli(*al

crack mentioned.
The relationship betw'een the appli(‘d tonpie T and the stresses which it

produces will now be found From the e(|uilii>rium of that portion of the
shaft iHd.ween the ixittom and the section mn, it can be concludial that the
shearing sti*esses distributed over the cross-section 7nn are statically
opuivalent to a couple equal and opposite to the external torque T. iuir
each element of area dA (lug. l.lcj. the shearing force is rdA. The mo-
ment of thi.> force about the axis of the shaft is (rdA)p = Gdp^dA, from o(\
(4.2). The Tobil resisting tor(]ue T about the axis of the shaft is the summa-

tion, takim over the entire cross-sectional area, of these moments of th('

individual clement-’; that is,

7’
=I oep- dA = ad p- dA = G 0J. (()
^

where

is defined as the polar moment of inertia of the circular cross-section.* For a

circle of diameter d, J - 7rd‘*/32, so that

T =
and
r 52
(4.3)
G Td^'

Thus 0, the ; igle of twist per unit of length of the shaft, varies directly avS

th(' applied tonpic and inversely as the modulus of shear G and the fourth

Appendix H, p. 540.
 CIRCULAR SHAFT 73

power of the diameter, If the shaft is of length /, the total angle of twist
will be

</> = /?/
II (4.4)
(fj

This equation is useful in the experimental verification of the theory, and is

checked by numerous experiments which justify the assumptions made in

its derivation. It should be noted that experiments in torsion are commonly

used for determining the modulus of materials in shear. If the angle of tw'ist
produced in a given shaft by a given torque be measured, the magnitude of

G can be easily calculated from e(i. (4.4).

Substituting d from ecj. (4. .4) in eq. (4.1), we obtain an eciuation for
calculating the maximum sh(‘aring stress in iwist of a solid circular shaft

ibr
~ Tr “ (4.5)
mP
We see that this stress is proportional to the applied lonpie T and inversely
proportional to the cube of the diameter of the shaft.
If we substitute 6 from eep { in eip ( 1 2), w(' olitain an ex})ression for
th(’ shi^anng stri’ss r at niu/ point in a solid circular shaft;

Tp
r ~- (4.6)
,1

In practical applications, the diameter of the shaft must soiinMirnes be

calculated from the horsepower wdiich it. is reipiired to transmit t li\ eii the
horsepower hp, the speed n in rpm, and the torque T in inch pounds, a
formula coimci'ting these quantities is derived as follows; Sin(*e the work
done by the applied tortpie 7’ piT revolution of the shaft is T Oir, the work
dune per minute at n rpm will be T 'IirrL Then with 1 Iqi = 35,000 ft-lb

per minute, we havi^

2TnT
12 X 33,000
or
,,
7 = 12 X 33,000 X hp —
63,000 (hp) .

in.-lb
,,
(4.0
2Trn n

When the horsepower and rpm are given, t he corresponding torque T can
be computed from this formula and then used in any of the foregoing
equations to compute the shear stress or angle, of twisi in the shaft.
In this way it may Ix'. shfiwii that the required diameter d of a S'^did

circular shaft, to transmit hp liorsepower at n rpm with a maximum allo\v-

able working stress Tw i’u shear, will be

(d)
 TORSION

Hollow Shaft. From the preceding discussion of torsion of a solid shaft of

circular cross-section,it is seen that only the material at the outer surface

of the shaft can be stressed ro the limit assigned as an allowable working

stress. All of the material within the shaft will work at a lower stress and is

not being used to full capacity. Thus in tUose cases where weight reduction
is important, it is advantageous to use hollow shafts. In discussing the
torsion of a hollow shaft, the same assumptions will be made as in the case
of a solid shaft. The general expression for shearing stress, eq. (4.2), will
apply. However, in calculating the internal resisting torque T in this case,
we sum the moments of the elemental forces rdA only over the region from
Pi = Jdt to po = as shown in Fig. 4.4. In this way, we again obtain, foi
the internal resisting torque,

T = Ge / dA = GdJ,
I
where now
<

is the polar moment of inertia of the hollow' cross-section. Thus, with the
proper value of/, the basic equations (4.2), (4.4), and (4.6) apply also to a
circular shaft of hollow' cross-section.
Taking, for example, the case w’^here di = \do, the angle of twist </>, eq.
(4.4), and the maximiim shear stress Xmax, eq. (4.6), will be found to be about
6 per cent larger than in the case of a solid shaft having the same outside
diameter do. But the reduction in w'eight will be 25 per cent.

50 hp on

EXAMPLE 1 . The in Fig. 4.5 is made of steel, has diameter

solid line shaft shown
d = 1 .5 in., and runs at 525 rpm.
supported in bearings so placed that bendmg
It is
of the shaft will be negligible. A driving belt feeds 50 hp to the left hand pulley
while 30 hp and 20 hp, respectively, are taken off by belts overrunning the middle
and right hand pulleys. Compute the maximum shear stress r induced in the shaft
and the total angle of twist <t>. Assume G= 12(10)® psi.
 CIRCULAR SHAFT 75

SOLUTION. The greater torque is in the left-hand portion of the shaft. Its value
from eq. (4.7) is

T = 63,000 X ^-
OaO
6000 in.-lb.

The corresponding maximuiri shear stress from eq. (4.5) is

16 X 6000
r 9060 psi.
TT X (1.6)3
Similarly, for the right-hand portion of the shaft which transmits 20 hp, T = 2400
in.-lb and r = 3620 psi.
The total angle of twist is the sum of the angles of twist and <^2 in the two
portions of the shaft. Using eq. (4.4), this becomes

=
6000 X 120 2400 X 240 = = 12®27'.
<t>\ <t>2 0.217 rad
12(1 0)« X 0.497 1200) » X 0.497
The value of J used in the above ealeiilation is

^ 3^(1.
»•- 0.497 i„..

'

' EXAMPLE 2, A stepped solid circular shaft is built-in at its ends and subjected
to an externally applied torque To at the Determine
should(;r as shown in Fig. 4.6.
the angle of rotation where To is applied.
of the shoulder section
SOLUTION, l^his is a statically indeterminate system because the shaft is built-in
at both ends. All that we can find from statics is that the sum of the two reactive
torques Ta and Tjt at the built-in ends of the shaft must be equal to the applied
torque To. Thus
T^4-Tb^ To. (f)

From consideration of consistent deformation, we see that the angle of twist in

each {xirtion of the shaft must be the same, i.e.,

(t>a = 4>b = 4>0- (g)

Using eq. ‘(4.4) for angle of twist, expression (g) becomes

Tacl _ Tsb __
~
GJa " GJb
(h)

This defines the ratio between Ta and Tb as

(i)
Tb Jb(^

-M
•^4
Ca
T X
Fig. 4.6
76 TORSION

From this expression, we see that in the particular case where J ~ J ih

for a shaft of uniform cross-section, the reactive t()r(|ues Ta and Th are simply in
the inverse ratio f^»f the lengths a and b. From cqs. (f) and (i;, we obtain, for the more
general case,
To
Ta = T„ = (j)

J B(i

Using either of these values in eejs. (Ii), we have for t]i(‘ angle of rotation 0i) of tin'
junction
Tpoh
(k)
{•f Ab + J rt<i)(r

PROBLEMS
1 The shaft of a "J'-socket wrench
. is J in. in diameter and 18 in. long. If the
allowable working stress in shear is r^. =
what is the maximum twisting 10,000 psi,
moment that it is safe to exert with the wrench, ami through what angle will the
shaft twist under the a(‘tion of this torque^ A ns. Ty, = 215 in.-lb; = 3.44°. '
<t>

2. A ste(‘l shaft of diamet(‘r d — 2 in. and hmgth I

— 20 in. is twisted in a testing
ma(4une until one end rotates through an angle 0 -- 0.6° with n'spind to the other
end. For this angle of twist, the m(‘asiired torque load is T — 9860 in.-lb. What is

themaximum shear stress r in tiu' shaft and wdiat is the valui' of the shc'ar modulus
G? Ans. T = 6280 psi; G = 12(10)« psi
3. A solid st(‘el shaft of diameter d = 3 in lias a working stress in shear, == 3000
psi and the allow abh' angle of twist ])(‘r foot of length is (’alculate the safe
hirque T that the shaft mav transmit. Ans Ty, - 11,600 in.-lb.
i. A steel shaft diameter turns at 10,000 rpiii. What U’ th(‘ maximum
1 in. in
powa^r that such a shaft ma\ develof) without exce(‘ding an assigned working stress
in shear of r„, = 4500 psh'’ .hes*. 0.274 hj).
5. Determine the proiK*r diam<‘t(*r d for a solid stet4 shaft to transmit 200 hp
at 105 rpni if the working stress m shear is 6000 psi. -h/,9. d ~ 4.67 in.
6. Determuu' tiu* propiT diarnetiT d for a solid steel shaft to transmit 300 hp
at a speed of 3()00 rpm if the w orking stress in shear is 7-„, — 6000 psi. Ari-s. d ~ 1.65
in.

7. A hollow' .<teel .shaft is to have outside diametcT d arul inside dianud-er d/2.
Calculate’ the pn^jier value of d for the shaft if it is to transmit 2(X) hp at 105 rpm
with a w'orking stri's^ in shear of 6000 psi. Jws. d = 4.78 in.

^Geor Gear
N
Gotc
house
d,
i

T
0
/ d,
L_ 4
4*-- 3
'-
- ’
5 *1

Fig. a

3. Figure A represents a plan vimv of a sluice gate installation for liontrolling

water flow. To raise (*ach of the three vc'rtically sliding gates requires a force of
 CIRCULAR SHAFT 77

20,000 lb to bo oxort/od by th(‘ corresponding; g;car (7.5-iii. pitch diameter) on a

vortical rack attached to each g;ote stem What are the n*(iuired diametfTS du d.*. rfa,
for the threo portions of the shaft if -= 12,000 psi‘^ Assum(> a steel shaft, G =
12(10)® psi. An^. rf, - 5.75 m., ~ 5.00 in., cU = 4.00 m.
9. For the s^^stem in Fig;. H, power is transmitted from the gear .1 to the gear D.
If the pitch dianieti'rs of the gears B and C are m tlu* ratio J 3, \\hat ls the proper
ratio of shaft diamc’ters diido for both shafts to hav(‘ th(‘ same maximum shear
stress r? .Ins. di.do = 0.093.
10. For the stepped shaft shown in Fig. 4.6, following data are given:
the
(I
='-
30 in., b = 30 in., di = 1 .00 in., d-i
— 1.20 in.,To = (>IK)0 in.-ll> ('aleulate the
maximum shear stress in the shaft and the angl(‘ through vvhieh tin' shoulder s(‘ction
will rotate. .Assume G = 12(10)® psi. .bKs\ r == 11,900 psi; ~ 2.85°.
11. A prismatic shaft with built-in ends is subjected to th(‘ action of externally
applied twisting moments Ti and T-. as shown in Fig. What are, the internal
torques Ta, Ti,,Tr, in th(' thrta* portions a, 5, c, of the sliaft? The following data
are given: a = 30 in., 5-50 in:, c = 40 in., 7b — 12,000 in. -lb, and To - 24,(K)0
in.-ll). Ins. T„ - 17.000 in.-lh; 7\ - 5000 in.-lb; T, = 19.000 in.-lb

12.

Fig. C Fig. D

A solid circular aluminum shaft of diameter d — 2 in. is twisted in a testing

rnar'hine as ^hown in Fig. 1). strain gag(* plac(‘d along a 45°-heli\ on thi‘ surface
of the shaft .shows a positi' c strain e =
955(10) b wluui T = 12,000 in. -lb. W^'hat
IS the shear modulus C for the iiiatenar* A ns. G ~ 4(10}® [)si

4.2 Close-(^oiled Helical Spring;

An interesting appliealion of the theory of torsion arises in t he ease of a

close-eoiled helical spring.* Assume that su(‘h a spring, wuiund from a wire
of solid circular cross-section on a eireiilar core, is subjected to the action of
axial forces P (Fig. 4.7a), and that any one coil li(\s nearly in a plane per-
pendicular to the axis of the helix. Considering the equilibrium of the upper
portion of the spring hounded by an axial section such as mn (Fig. 4.7b), it

can be concluded from equations of statics that the stress resultant on the
cross-section mn of the coil reduces to a shearing force P through the
center of the cross-section and a couple acting in a counterclockwise direc-
tion in the plane of the cross-section of magnitude PR, where R is the radius
of the cylindrical surface containing the center line of the spring. The
*For a complete treatise on springs, .sih* Mechanical Spnmjs by A, M Wahl, Penton
Publishing Co., Cleveland, 1944.
78 TORSION

couple PR twists the coil and causes a maximum shearing stress given by
eq. (4.5), which becomes here

(a)

where d is the diameter of the cross-section mn of the wire. Upon this stress
due to twist, that due to the shearing force P is superposed. For a rough

Ftg. 4.7

approximation, this shearing force is assumed to be uniformly distributed

over the cross-section ; the corresponding shearing stress will be

rr
IP
T (b)
irfP

At the point m the directions of t and r" coincide so that the maximum
shearing stress occurs here and has the magnitude

= r + r = — (^1 +_
j
(4,8)

It can be seen that the second term in the. parentheses, which represents the
effect of the shearing force, increases with the ratio d/P. It becomes of
practical importance in heavy helical springs, such as are used on railway
cars. Due to this term, points such as m on the inner side of a coil are in a
less favorable condition than points such as n, for at point n the shearing
stresses t' and r'' act in opp>osite directions. Experience with heavy springs
shows that cracks usually start on the inner side of the coil
 HELICAL SPRING 79

There is another reason for expecting higher

stresses at the inner side of the coil. In calculating
the stresses due to twist, we use eq. (a), which was
derived for cylindrical bars. In reality each element
of the spring will be in the condition shown in Fig.
4.8. It is seen that if the cross-section hf rotates
with respect to oc, due to twist, the displacement of
the point b with respect to a will be the same as that
of the point / with respect lo c. Due to the fact that the distance ab is

smaller than the distance #/, the shearing strain at the inner side ab will be
larger than that at the outer side c/, and therefore the shearing stresses
produced by the couple PR will be larger at b than at /. Taking this into
consideration, together with the effect of the shearing force, we replace eq.
(4.8) by the following equation* for calculating the rnaximuin shearing
stress:
WPR (4.9)
7r(P \4m ~ 4 m /

in which m = 2R/d. It can be seen that the correction factor in the paren-
theses increases with a decrease of m; for instance, in ease m == 4 this
factor is about 1.40 and for m= 10 it is equal to 1.14.

In calculating the deflection of the spring, usually only the effect of the
twist of the coils is taken into consideration. For the angle of twist of one
element between the two adjacent cross-sections mn and m'n' (Fig. 4.7c),
using eq. (4.4), in which Rda replaces i, we obtain

, . P R Rda
’ ’

d4>

Due to this twist, the lower portion of the spring rotates with respect to the
point 0 (Fig. 4.7a), and the point of application B of the force P describes
the small arc Bfi' — a dip. This movement of B is easily pictured by
imagining all of the spring as rigid except the element between mn and
m'n'. The vertical component of this displacement is

PR} da
dd B'B" = BB'- = Rd<l>^ (c)
a JG
The complete deflection of the spring is obtained by summation of the de-
B'B" due to each element mnm'n', over the length of the spring.
flections
Then
, l^^PR}^ 64nPB»
4 10 )
.
(

•For the derivation of this formula, see A. M. Wahl, "'Stresses in Heavy Closely Coiled
Helical Springs, “Trons. A.S.M.E,t 1929, VoL 61, Paper No. APM-61-17
80 TORSION

inwhich n denotes the number of coils. The net horizontal displacement in

one complete turn is zero.
The ratio jP/ 5 for a given spring is called the spring constant^ denoted by /c.

Thus, from e(\. (4.10), we have

P ^
( 4 11 )
.

5 MRhi
Two springs are said to be of the same stiffness if their spring constants arc
(•(jual. The spring constant k can be varied by changing the material, the
wire diameter, the core radius, or the number of coils.

hXAMPLK 1 . Two closo-coilcd helical springs wound from the same wdre but with
different core radii are assembled as shown in Fig. 4.9 and compressed between
rigid plates at their ends. C alculate the maximum shear stress induced in each
spring if the wire diameter d \ in. and P - 100 lb. 7'he (‘ore radii are a.s shown
in the figure.

SOLUTION, two springs must have the same over-all shortening 5, it

Sinc(‘ the
follows from F is divided between them in the inverse ratio
eq. (4.10) that the load
of the cubits of thidr core radii. Thus denoting by Fi the load carried by the outside
spring and by Fa that carried by the inside spring, we have

Li -?L
~
Pi 64
Also
Pi -f Po = F == 1001b.

From those two equations, we obtain P\ - 29.7 lb and Pa = 70.3 lb. Substituting
these values, together with the other given data, into eq. (4.9), the corresponding
 HELICAL SPRING 81

iiiaxiniuin shear stresses in the two springs are found to be r 2860 psi and
T‘i
- 5380 psi.
KXAMPLio 2. A e(jnical spring as shown in Fig. 4.10 is siiV)jeeted to a compressive
load F. Each coil is assumed to lie essentially in a horizontal plane and the sliape
of the s})iral in plan vie\A' is defined by the equation

ft = ft. + (d)
27rn

where R is the radius at any point .4 on the spiral and a is the angle measured as
shown. It is required to develop a formula for the spring <'onstant k,
SOLUTION, 'j'he conditions of an element of the spring at .1 of huigth Rd^x will he
the same as for the element in Fig. 4.7c. Hence, using eq. (c),

Substituting now the value of R from eq. (d), we obtain

, ~
= 32? P"’T„
ird*G ,
,
(ft 2 -ft.)a|>
rf«
Ififtrt
= ^(ft.^+ft./)(ft. + ftj),

where J has been replac’ed by TTr/"*/ 32. Thus the spring constant is

^ ^ ^ 0^ ‘

5 16n(/?> + /?.')(/?! +^ 2)

PHOBLEMS
1. For th(' helical spring shown in Fig. A, the following data are given: P ^ 400
lb, R ~ 3 in., d — -J
in., n ~ 12, and G = 12(10)® jisi. (-aleulab', the maximum
sliear stre.<=^s and the (‘xtension 5, Aris. Tm ~ 17,200 psi; 5 == 2.18 in.
2. If the limiting shear stress for the spring shown in Fig. A is Tm, = 20,000 psi,
what is the maximum safe load? Ans. Pm = 465 lb.
3. The
h(‘lical spring shown in Fig. H has rii + 712 ~
n coils, is built-in at both
ends,and c.arries a load P applied to an intermediate plate. C’alculah^ the reactions
Ri and R 2 at the ends of the spring. A /is. R /R 2 = uzlriu Ri R 2 — P. 1

Fig. a Fig. B Fig. C

82 TORSION

4. Using formula (4.9), find the safe load P for the conical spring in Fig. 4.10 if
the working stress in shear is r,,. = 45,000 psi. The spring is wound from a 1-in.-
diameter steel rod with Ri — 2 in. and R 2 — S in. Ans. — 1010 lb.
5. A rigid bar AB weighing 20 lb and carrying a load P=
80 lb is supported by
three springs having spring constants A: 1 = 1(X) Ib/in., 60 lb/ in., kz = '10 Ib/in,
=
(Fig. C). If the unloaded springs were all of the same length, find the distance x
such tliat AB will be horizontal. Ans. x = 7.5 in.

4.3 Strain Energy in Shear and Torsion

Consider, in Fig. 4.11, an element of elastic material in a state of pure

shear. The strain energy stored in such an element may be calculated in
the same way as for the case of simple tension (see Art. 2.4). During de-
formation of the element, the top face cd moves horizontally through the
distance ydy relative to the bottom face ah as the shear stress is gradually

i
^

'V

Fig 4.11 Fig. 4.12

increasing from zero to the final value r. Then the work done by the shear
force rd.<; dz (ui the top face is \Tdx dz • ydy. Since the shear forces cn the
sides acand bd do no work, this represents the total strain energy stored in
the element. Dividing by the volume dx dy dz of the element, we obtain

^ = fa)
y
for the strain energy per unit volume.
Noting that, within the elastic limit of the material, y = t/(j, expression
(a) may be written in either of the following two forms:

u = or w = (4.12)
 STRAIN ENERGY 83

The first of these equations expresses the strain energy as a function of the
shear stress t; the second, as a function of the shear strain 7. The limit of
elastic strain energy per unit volume will be obtained by setting r in the
first equation equal to the elastic limit in shear.
Having expressions (4.12) for strain energy of shear per unit volume, the
total strain energy in a solid circular shaft of radius r and length I subjected
to twisting moments T at its ends (Fig. 4.12) can easily be found. Denoting
by Tmnx the maximum shear stress at the surface of the shaft, the shear
stress at any intermediate radius p will be Tmax(p/r). Then, from the first of
eqs. (4.12), the strain energy per unit volume at the radius p will be

Tmax^
u (b)
2Gr2
The energy in the elemental tube of length Z, radius p, and thickness dp
will he
^ ^
dU = udV = Z^-L.i.2rpd.p. (c)

Summation of expression (c), from p = 0 to p = r, gives for the total strain

energy in the twisted shaft,

This is seen to be just half the value that would be obtained if all the
material were stressed to the maximum value r^ax.Noting that rn^x =
Tr/J, where ./ = TrrV2, expression (4.13) can be written also in the form

U- ~TH
(4.13')

This strain energy U for a shaft in torsion

may be obtained in another way by using ex-
pression (4.4) for the angle of twist in the shaft.
This shows that the relation between torque T
and angle of twist <t> is linear within the elastic
limit of the material as shown by the torque-
twist diagram OAB in Fig.
4.13. For any small
increment d<t> of the angle of twist, the work
done by the acting torque is represented by the
area of the shaded strip in this diagram. Thus
as the torque is gradually increased from zero
to any final value T, the total work, equal to the
energy stored, is represented by the area OAB

TT T* /-1 \
84 TORSION

where 4> = TlfGJj from eq. (4.4). With this relationship between T and
expression (d) may be written in either of the following two forms:

( 4 14 )
.

The first of these equations expresses the strain energy in terms of the
torque T; the second, in terms of the angle of twist Expressions (4.14)
may be used either for a solid circular shaft with J = 7rd^/32 or for a hollow
circular shaft with J = T(do^ — d»^)/32.
If the internal torque varies along the length of a shaft as in the case of a
conical spring (Fig. 4.10), we consider one elemental disc of length dx and
under twisting moment Tx Then the angle of twist in this elemental disc is

d<l} = ^dx, (e)

dx
where
d^ _
dx ~ GJ (f)

is the angle of twist per unit thickness of the elemental disc. Substitution
of d<t) for 4> ‘^md dx for I in the second of eqs. (4.14), gives, for the strain
energy in the disc.

d[> -

Then a summation over tfie full length I of the shaft gives, for the total
strain energy

= ( 4 15 )
.

KXAMPiiE1. Determine the deflection 6 for the close-coileci helical spring in

Fig. 4.7a,by using the expression for strain energy of torsion.

SOLUTION. The twisting moment on each and every element of the coil like that
shown in Fig. 4.7c is PR. lienee, by the first of eqs. (4.14), the total strain energy
in the spring of length 2TrRn is

(P/2) 2. 2TrRn
_
2GJ

Equating this to the w^ork Fd/2 of tlie load P during deflection, we obtain

2TrnFR^ e>4nPR\
~ GJ ~ Gd*

which agrees with expression (4.10) on p. 79.

EXAMPLE 2. Verify expression
on p. 81, for the deflection 5 of the conical
(e)
spring loaded ns showm in Fig. 4.10, by using oq. (4.15) for strain energy of twist.
SOLUTION. At any point A on the coil (Fig. 4.10), the torque = P/2, where the
 STRAIN ENERGY 85

radius R varies with a. according to eq. (d) on p. 81. Then from expression (f)

above, the angle of twist per unit length of coil at A is

PR

Substituting this into eq. (4.15), taking dx = R da and changing the limits of
integration accordingly, we obtain

Equating this to the work Pd/ J. of the load P, we. obtain

P"" PR\
~ Jo GJ
which agrees with eq. (e), p. 81.

KXAMPLK 3. A solid steel shaft with a flywheel at one end rotates at constant
speed n = 120 rj^m, Fig. 4.14. If the bearing A suddenly freezes, what maximum
shear stress Tnmx will be produced in the shaft due to dynamic effects? Assume
/ = 5 ft, d == 2 in., the weight of the flywhend \\ — 100 lb, and its radius of gyration
1 = 10 in.

ROLUi'ioN, l^he maximum shear stress in the shaft will occur when the total
kinetic energy of the flywheel has been absorbed by the shaft. This energy is

WW ^ 100 X 10^ X {At Y

*
2050 in.-lb.
2g 2 X 386

2^
Fig. 4.14

Setting tills equal to the strain energy U as given by eq. (4,13) and solving for
Tmax, we find
11.5(10)« X 2050
'Tmax 22,400 psi.
TT X 1 X 60

PROBLEMS
1. Two shafts of solid circular cross-section are identical except for their diam-
eters di and d 2 Under the $aihe torque T, what will be the ratio of the amounts of
W
.

strain energy stored in each shaft? Am. U = {di/diY.

1 2

2. A solid circular shaft and a thin-walled circular tube made of the same material
and having the same weight are stressed in torsion to the same maximum shear
What is the ratio of the amounts of strain energv sUired in the two shafts?
stress T.
Am. f/i/f/2 - i
86 3. TORSION

The shown in Fig. A is wound from a steel wire having diameter

conical spring
d and has n = 10 coils. The plan view of the spring represents an Archi-
== 0.1 in.

medean spiral for which R = oa, where a = 0.1 in. and a is an angle measured as

Fw. A Fig. F Fig. C

shown It the shear modulus G = 12('10)® psi, what is the value of the
in Fig. 4.10.
spring constant k? Avs. k = 0.0302 lb /in.
4. A flywheel of weight W
= 3S.6 lb and radius of gyration i = 10 in. is riKumtcd
at the middle of a solid steel shaft of diameter d = 2 in. and length I as <hown in
Fig. H. The shaft rotates in hearings A and B at an angular speed n = 120 rpiii.
If both bearings suddenly freeze so that the ends of the shaft become ifu ked, the
shaft will have to absoib the kinetic energy of the flywheel. Calculate the shortest
length / of tht‘ shaft for which this can be done without exceeding a maximum shear
stress Tu, = 12,000 jisi in the shaft, dris. /,nin = 7.00 ft.

5, A solid steel shaft of diameter d - { in, fits loosely inside a hollow steel shaft
of inside diameter d ^ in. and outside diameter di — | in. as shown in Fig. C.
=
A pin .‘1.1 prevents relative rotation between the ends of the shafts at tin' left.
Pinholes at the right arc initially at right angles to each other as shown. Th(‘ two
shafts are now twist'd in opposite directions until the pinholes at B line up and a
pin BB IS then inserted. How much strain energy will be locked in th(' system if
I = 100 in.? Ans. U - 45.6 in.-lb.

4.4 TorHion of Thin- Walled Tubes

In the case of a hollow shaft of circular cross-section for which the inside
diameter is very nearly equal to the outside diameter, we speak of the shaft

as a thin^walled tube. For such a tube in torsion (Fig. 4.15), the polar
moment of inertia of the cross-section cannot be calculated with good
accuracy from the formula

and it is preferable to use the approximate expression

J = P^dA j^dA = 2irrH, (a)

 THIN-WALLED TUBES 87

where r is the radius of the mean center line

and t is the wall thickness. Then assuming for
such a thin-walled tube that the shear stress r
isuniform across the wall and equal to the
value at the mean radius r as shown in Fig.
4.15b, we obtain from eq. (4.6)

T \
(b)
(b) \
2irrH \
\

Likewise, from eq. (4.4), the angle of twist of

the tube becomes

Tl Tl \
^ (c) \
GJ 2TrHa

Tn Art. )L4, it was shown that the -^tate of

pure shear such as exists on the element A in

Fig. 4.15a is equivalent to biaxial teuMon and Fig. 4.15
compression on an element oriented at 45° to
the axis of the tube, like the element B in Fig. 4.15a. From this, we see that
a long narrow strip of the wall coinciding with the 45°-helix shown in Fig.

4.15a is subjected to axial compression, and if the wall of the tube is very
thin, such a helical strip may buckle. This phenomenon can be demon-
strated by rolling a sheet of paper into a tube and then subjecting it to
torsion. Analysis of this problem* shows that for a long steel tube under
torsion, we should have the ratio t/r > %o to avoid the danger of buckling
at normal working stresses.
Introducing, in eqs. (b) and (c), the notations

Ao = TTf^ = area enclosed by mean center line,

8 = 2Tr = length of mean center line,

we obtain
T
(d)
2Ao/
and
^ _ TSl
~ 2AS (e)

The above formulas can be used for calculating the shear stress and angl^
of twist in a thin-walled tube of arbitrary cross-section as shown in Fig.
*See L. H. Donnell, Stability of Thin-Walled Tubes under Torsion^ Nat Adv, Comm.
Aeronautics, Tech. Kept. 479, 1933.
88 TORSION

4.16. When such a thin tube of noncirciilar cross-section is twisted, the

cross-sections rotate slightly, one with respect to another, but they
do not
remain plane. After twist, each cross-section becomes slightly warped, and

if no restraint against this warping exists at the ends of the tube, it takes
place in such a waythat the shear strain 7 of every element of the wall
such as A has the same magnitude regardless of the radial distanee p of that
element from the centroid of the cross-section. Thus the shear stress r,

proportional to the shear strain 7 ,

is uniform throughout the wall of the
tube.
To relate this shear stress r to the external torque T, we consider any
cross-section of the tube as shown For an element ds of the
in Fig. 4.16b.
mean center line, the shear force is rids as shown, where t is the thickness of
the wall, assumed constant. The moment of this force about point 0 is

dT — rtds • r,

where r is the distance from 0 to the tangent to the mean center line.
Summing such elemental moments over the entire length s of the mean
center line, we obtain
T =H rds.
j
We see that the quantity rds under the integral sign is just double the area
of the small shaded triangle of base length ds and altitude r in Fig. 4.16b.
 THIN-WALLED TUBES 89

Thus the integral of this quantity over the full length « represents double
the area enelosed by the mean center line of the wall. Denoting this area
by A 0 we have
,

r = Tf . 2Ao,
from which
T
T ^
(4.1G)
2A,i

which coincides with eq. (d) above.

To calculate the angle of twist 0 of the tube, we use the method of strain
energy as discussed in Art. 4.2. From the first of eqs. (4.12) the total strain
energy in the tube is

U= X sit,

where sit is the volume of material in the tube and t is the shear stress,
uniform tliroughout the volume. Equating this strain energy to the work
T <t>/2 of the applied torque during twist, we obtain

T~W
T0

from which

^ Yg
Since T 2AotT from eq. (4.16), this becomes
TSl
0 = (4.17)
2AoC/
which agrees with eq. (e) above.
In practi(;ul problems we often eU(‘ountcr thin-walled tubular members
of other than circular cross-section and ecjs. (4.10) and (4.17) are very
useful in the analysis of their behavior under torsion. The following
examples will serve to illustrate several such problems.
EXAMPLE 1. Two thin-walled tubular members made of the same material have
the same length, the same wall thickness, and the same total weight and are sub-
jected to the same torque T. If their cross-sections are circular and square, respec-

1i !1 11 t1 “1
r 1

1 1

1
1

1
1

1
1

4 ;'
1
1

1
1

1
1

1
1

1
1

1
1

Fig. 4.17
 90 TORSION

lively, as shown in Fig. 4.17, what are the ratios of their shear stresses r and of their
. angles of twist <^?
SOLUTION. From eq. (4.16), the ratio of shear stresses is

T
(0
7 ‘

where A'o and are the areas enclosed by the mean center lines of the same
length s (Fig. 4.17). For the circle A'o = sV4ir while for the square, A"o = sV16.
Substituting these values into eq. (f), w^e find

r' ^47r_ir
“ l6 ” 4 0.785.

From eq. (4.17), the ratio of angles of twdst is

0.616.
<t>" r'M'o \4/
Wc conclude from this that the circular cross-section represents a more efficient
use of the material.
EXAMPLE 2. A stainless steel tube having an
elliptical cross-section as shown in Fig. 4.18
is subjected to torsion. If the allowable stress
in shear psi,
is what is the
== 10,000
corresponding allow^able angle of twist per
unit length of the tube? The following data
are given: G = 12(10)® psi, a = 3.00 in.,

b = 2.00 in., ^= J in.

S 01.UTION. From eq. (4.17), the angle of
twist per unit length of tube will be

0 _ Tg
T “ 2A^
The area enclosed by the mean center line is Ao = Tab = 67r sq in. The length s of
the mean center line may be calculated with good accuracy from the approximate
formula

\{a + h) Mab

which gives s = 15.86 in. Then the above expression for 6 becomes

10,000 X 15.86
0.000350 radian/in.
2 X 6ir X 12(10)®
PROBLEMS
1. A
tubular aluminum shaft having a square cross-section with outside dimen-
sion a =
1 in. must safely carry a twisting moment T = 636 in.-lb. Calculate the
proper wall thickness t if the working stress is = 6000 psi. An«. t « 0.060 in.
2. For the aluminum tube described in the preceding problem, calculate the angle
of twist per unit length of tube under the applied torque T = 636 in.-lb. The shear
modulus G * 4(10)® psi. Atw. 6 « 0.00319 radian/in.
3. A thin-walled stainless steel tube has the rectangular cross-section shown in
Fig. A. How does the angle of twist per unit length of tube due to torsion vary
 THIN-WALLED TUBES 91
4.
with the ratio a « a/b if the total length of mean centerline 5 = 2 (a -f ft) and the
applied torque T
remains constant? Ans. 0 varies as (1 4- cx)^/a^.
The cross-section of a thin-walled steel tube has the form of an equilateral
triangle with mean center line s = 3n = 3 in. and wall thickness / = in. If
Tw — 8000 psi, what is the safe twisting moment T for the tube? Am. T — 433 in.-lb.
-T fr

5..

Fig. a Fig. B

A long thin-walled tube of circular cross-section has a conical taper, the

diameter of the mean center line at the small end being do and that at the large end
being 2do (Fig, B). The wall thickness is t, the length is L and the applied torque, T.
Calculate the angle of twist of one end of the taf)e with respect to the other.

Am.
m
2TdoHG
6 What IS the ratio between the shear stress<'s rn and n near the two ends of the
conical tube in the above problem'^ 4ns. to/t/ = 4.

4.5 Shaft of Rectangular or Profile Section

The prolilem of twist of a shaft of rectangulai’ eross-
sectioii is complicated, due to the warping of cross-
scctioiis during twist. This warping can be shown
experimentally with a rectangular bar of rubber on
whose faces a system of small squares has been traced.
It is seen from Fig, 4.19* that during twist the lines

originally perpendicular to the axis of the bar become

curved. This indicates that the distortion of the small
squai-es, mentioned above, varies along the sides of this
cross-section, reaches a maximum value at the middle,
and disappears at the corners. We thiuefore expect that
the shearing stress will vary as this distortion: namely,
it is a maximum at the middle of the sides and zero at
the corners of the cross-section. Investigation of the
problem t indicates that the maximum shearing stn^ss
occairs at the middle of the longer sides of the rectangu-
lar cross-section and is given by the equation

“ Jl
-
ocbc^’
(4.18)

*This figure is reproduced from Bach’s Elastizital und FesUgkeit.

tThe complete solution is due to de Saint Venant, MHn des Savanis HrangerSj t. 14
(1855).
92 TORSION

inwhich b is the longer and c the shorter side of the rectangular cross-section
and a is a numerical factor depending upon the ratio b/c. Several values of
a are given in Table 4.1.
T he angle of twist per unit length in the case of a rectangular cross-section
is giveii by the eciuation
r
( 4 19 )
.

ffbc^G

The values of the numerical factor are given in the third column of
Table 4.1.

Table 4.1 DATA FOR THE TWIST OF A

SHAFf OF RECTANGI:LAR CROSS-SECTION
b/c a
1.00 .208 .141
1.50 .231 .196
1.75 .239 .214

2.00 .246 .229

2.50 .258 .249
;eoo ,267 .263

4,00 .282 .281

(i.CK) .299 .299
8.00 .307 .307

lO.(K) .313 .313

oo .333 .333

It is seen that in the case of a very narrow rectangular cross-section, such

as that for a thin strip of sheet metal, a and P equal \ and the ec^uations for
the maximum shearing stress and the angle of twist per unit of length be-
come
:yj\
~ bc^’
(4.20)

3T
(4.21)

These equations are of practical importance because they can be used not
only for a narrow rectangle but also for approximate solutions in other cases
in which the width of the cross-section is small. For instance, in the case of
the cross-sections of unifcjrrn thickness showui in P'igs. 4.20a and b, the angle

of twist is obtained Irom eq. (4.21) by putting in this ciiuation for b the
developed length of the center line, namely, b = 0r in the case of the section

represented in Fig. 4.20a, and h = 2a ~ r in the case represented in Fig.

4.20b. The inaximum stress for the first of these two sections will be ob-
tained from eq. (4.20). For the angle section (Fig. 4.20b) the maximum
 RECTANGULAR SHAFT 93

stress is at the reentrant ofn-nor.* This luaximnrn stress is obtained by

multiplying the stn^ss given by eq. (4.20) by a stress eoneentration factor.
This factor has b(?en found to vary with the ratio of lillet radius r to thick-
ness c: values of this fact(»r are given in Table 4.2.

Table 4.2 STRESS CT)NCE\TRAT10N FAC14)R

r/r Factor
1
6 2i
1
4 2i
I
2
i IJ

roxAMPLK 1. Two thin-walled tubes of (‘ircular cross-section are identic.al e.xcept

that one seamless and the other is split as shown in Fig. 4.21. If r is the radius
is

of the mean centiT line and t is the wall thickness, tind th(‘ ratio of th(‘ir angles of
twist 0 per unit length of tube when they are subjeehnl to the same torque T,
SOLUTION. For the seamless tube, we obtain from eejs. (4 10) and (4.17) of Art. 4.4,

Th T
""
2^?Gt

Using eq, (4.21) for the split tube when* b ~ 27rr and c - I, we obtain

e
*
=-HL (h)
2TrGt^

Thus
(c)

Taking, for example, t/r = 1/10, this gives 6^/62 ^ 1/300, i.e,, t}u- seamless tube
is 300 times as stiff in b>rsion as the eorresponding split tulx?.

EXAMPLK 2. For the structural angle s^'ction shown in Fig. 4.20b, tie* iollowiug
dimensions are given: n = 4 in., c — ^ in. A sU‘d bar having this cros.s-section is
4 ft long and is subjccUd to twisting moments T = 2500 iii,-lb at its ends. Compute*
the maximum shear stress r and the angle of twist between the two ends ol the
bar if the fillet radius r = J in.

*The above methods apply where the cross-sections of the structural shapes are free
to warp.
94 TORSION
SOLUTION. Taking 6 == 2a — c = 7.5 in. in eq. (4.20),

3 X 2o00

Then with r/c = \ 1 the stress concentration factor is 2 from Table 4.2 ‘

above and the maximum shear strtvss at the reentrant corner becomes r = 2 X 4000
= 8000 psi
The
1, total angle of twist from eq. (4.21) becomes
3 X 2500 X 48
0 = 0/ 0.032 radian.
7.5 X ihV'^ X 12(10)<'

PROBLEMS
Calculate the ratio between the maximum shear stresses ri and T 2 induced in

the seamless and split tubes, respectively, in Fig. 4.21. Aiis. tj/t 2

2. An aluminum channel
the cross-section
lias

1/2 — 1/2
shown in Fig. A. The
radius at the two
fillet

reentrant corners is J in. Calculate the safe

'

V2 f twisting moment T for the channel if the working

1 1 3"
3 ; ;

1
stress in shear is = 8000 psi. Ans. T 5000
1 : :

in. -lb.
t
. 10': 3. Calculate the angle of twist 0 per unit
1

length of the channel in Fig. A under torque

Fig. a T = 5000 in. -lb if C = 4(10)® psi. Ans.d = 0.002
rad/ in.
4. A steel bar having a 1 in. X
i in. rectangular (Toss-section is subjected to
twisting moments T = 200 in.-lb applied to its ends. Compute the maximum
shear stress. Ans.Tmax = 11,350 psi.
5. If the bar described in the preceding problem is 10 ft long, w'hat is the angle
of twist 0 between its two ends under the action of the torque T = 200 in.-lb?
Assume G ~ 11.5(10)® psi. Ans. 0 0.475 rad.
 CHAPTER \

STRESSES IN BEAMS: I

5.1 Shearing Force and Bending Moment

A structural member that is reasonably long compared with its lateral
dimensions when suitably supported, and subjected to transverse forces so
applied as to induce bending of the member in an axial plane, is called a
beam. Several examples are shown in Fig.
5.1. The beam in Fig. 5.1a, supported by
a pin a t A and a roller at B is called a simple
beam. The one in Fig. o.lb, built into a
wall at B and free at A, is called a cantilever (a)
beam. Since, in both of these eases, the ZEITIIL
conditions of support are such that the re-
actions can be found from eciuations of (b)
statics, these beams are said to be statically
determinate On the other hand, the beam
J e.-
in Fig. 5.1c, supported by a pin at A and
rollers at both C and Z?, is statically inde- (c)

terminatCj since the reactions cannot be Fro. 5.1

found from equations of statics alone. For
the present, we confine our attention to statically determinate beams.*
Consider now a cantilever beam AB subjected to external loads and P2
as shown in Fig. 5.2a. If we imagine this beam to be cut by a section mn,
we see that the applied forces tend to displace the left-hand portion of the
beam relative to the right-hand portion, which is anchored in the wall.
This tendency is resisted by internal forces between the two parts of the
beam. Thus if we isolate that portion of the beam to the left of the section
mn as a free body, we represent the action of the built-in portion theron b>
distributed forces as shown in Fig. 5.2b. The true distribution of these
internal forces on the section mn is complicated, but to maintain equilib-
rium of the free body, they must be statically equivalent to the equilibrant

*Method8 of dealing with statically indeterminate beams will be discussed in Chapter

IX.

95
9ti ST1«ESBE.S IN BEAMS: 1

of the applied extenial forces P^and P-i. This stress resultant on the section
mn can always he represented by a force applied at the centroid of the
cross-section and a conple, both in the axial plane of the applied loads.
Furthermore, the force can, in com-
turn, be resolved into rectangular
ponents iVi, normal to the plane of the section, and W, lying in the plane of
the section. Thus, in Fig. 5.2(^ we represent the stress resultant on any

Fig. 5.2

MV‘lion mn by the three quantities Nxj Vxy and M*, called, respectively, the
normal force, the shear force, and the bending moment, at that section.
These (luantities will lie considered positive when they have the directions
shown in Fig. o.‘Jc. vSuch Mgn conventions, although arbitrary, must be
carefully obs(>rv^ed to avoid confusion.
To pursue an element of the beam
this question further, let us consider
between two adjacent shown in Fig. 5.Ba. Then normal
cros.s-sections as
forces N, shear forces V, and bending moments M
on the two faces of this
element will be considered as positive when directed as shown. We see that
a positive normal force is directed away from that face of the free body on
which it acts. Vlso a positive shear force isone that has a clockwise sense of
rotaLion about a point inside the free body. Finally the bending moments
 BENDINd MOMENT 97

M are positive when they tend to bend the element concave upwards.
Negative directions of A", F, and M
are shown in Fig. 5,:ih
Using the three equations of equilibrium

SX. = 0, = SA/. = () (a)

for the free body in Fig. o.2(*, the normal force A"x, the shear force Fx, and
the bending moment Mx at any cross-section mn can readily be calculated.
Using the centroid of the section mn as a moment center, these equations
give

Nx = Pi sin a,
]

= P 2 - Pi cos a,
Vx [
(b)

Mr = P 2 (.r — a) ~ Pi (‘OS a x. j

These expressions, of course, hold only for a < x < L For 0 < x < a, we
consider the equilibrium conditions of that portion of the beam to the left of
a section pq (Fig. 5. 2a) and obtain

Nx = Pi sin «,

Fx= —Pi cos a, ^ (c)

Mx — —Pi cos a • X. )

It should be noted from expressions (l>) and (c) that, in general, AT^, 1%, and
Mx vary with the distance .r defining the location of the cross-section at
which they occur.

w lb per unit length

(0) (b) (c)

Fig. 5.3 Fig. 5.4

98 stressp:s in beams: t

As another example, consider the simple beam AB carrying a uniformly

distributed transverse load of intensity w as shown in Fig. 5.4a. Con-
sidering the entire beam as a free body the reactions at A and B are found to
be wl/2 each, as shown.
Then to evaluate the shearing force and bending moment at a chosen
section mn, we consider the equilibrium of that portion of the beam to the
left of this section (Fig. 5.4b). Acting on this free body, we have the re-
action v)l/2 at A and that part of the distributed load between A and the
section mn. The resultant of this portion of the distributed load will be a
vertical force wx ac'ting at the distance x/2 from A as shown. On the face of
the section boundary, we have a .shear force \\ and a bending moment Mx,
but in this case, since none of the (external forces has any horizontal com-
ponent, there will be no normal force Nx. Now using e(^s. (a), we obtain

wx.
(d)
, , xol wx* I

= T''
~ T' ''

If we consider the e((uilibrium of that portion of the beam to the right of

the section ma(Fig. 5.4c), we obtain

wl
1 w{l
..
- .

x) - \

y, I

(e)

= ^(/ - ,r) - ^(/ - x)- !

These expressions can readily be reduced to coincide with expressions (d)

above. Since the entire beam is in equilibrium, we must otitain the same

values for shear forr-e and bending moment at a chosen section, by using the
equilibrium conditions for either portion of the beam.
A between shearing force and bending moment at
significant relationship
any beam will now be shown. In
cross-section of a transversely loaded
Fig. 5.5a, consider an element of the beam cut out by t'wo adjacent cross-
sections mn and pq, distance dx apart. On the left-hand face of this element,
we represent the shear force and })euding moment by Vx and Mx (assumed
positive). Then if no external load is applied between the cross-sections mn
and and bending moment on the right-hand face of the
pq, the shear force
element will Mx + dMx as shown, where dMx is the change in
be Vx and
fiending moment between mn and pq. Since the element is in equilibrium,
the algebraic sum of moments of these forces about point 0 must be zero
and we find
— Mx 4" {M “b dMx) — Vj/dx = 0,
 BENDING MOMENT 99

from which = Vx (5.1)

ax

Thus at any cross-section of the beam between points of application of

concentrated loads, the rate of change of bending moment with respect to x
is equal to the shear force.

Fio. 5.5

If there is some distributed load of intensity xc between mn and pq^ the

free-body diagram of the element will be as shown in Fig. 5.5b. Here again,
we equate to zero the algebraic vsum of moments of all forces with respect
to point 0 and obtain

-Mx + {M, + dMr) - V.dx + wdx = 0.

Neglecting the last term as a small quantity of second order in this ex-
pression, it reduces to
dMx — Vx dx = 0,

which yields the same relationship (5.1) obtained above.

Equating to zero the algebraic sum of vertical forces on the element in
Fig. 5.5b, we find

V, - (Fx 4 dFx) -wdx^ 0,

dV
from which = — w;. (5.2)
dx r

Thus when there is distributed load of intensity a; between the cross-

sections mn and pg, the shear force also changes along the beam and its rate
1(X) STRESSES IN BEAMS: I

of change with respect to .r is equal to the intensity of load but with opposite
sign.
If there is a concentrated load P on the beam between the cross-sections
mn and pq, the free-body diagram of the element will be as shown in
Fig. 5.5c. In this case, we denote the shear force on the left-hand face of the
element by Fx and that on the right-hand face, by F'*. Then equating to
zero the algebraic sum of vertical forces on the element, we obtain
F'x = Fx - P.

Thus in this case there is an abrupt change in the shear force over the length
cix. Accordingly, it may be concluded from eq. (5.1) that there will be a
corresponding discontinuity in the derivative dMJdx at the point of
application of a concentrated load P on the beam.
EXAMPLE 1. A simple beam AB carries a triangular distribution of transv^erse
load as shown in Fig. 5.6, .the maximum intensity of load at B being wo. At what
cross-section (defined by x) does the maximum bending moment occur and what is
its magnitude?
SOLUTION. The resultant load on the beam is wqI/2 acting at a point C distance
1/3 to the left of B. Thus, by XM « = 0, for the entire b(‘am as a free body,

RJ ~ (wol/2)(l/3) = 0,
from which Ra -tt'
()

The free-body diagram for a portion of the beam to the left of a section mn is shown
in Fig. o.bb. The resultant of that part of the distributed load which acts on this
free body is woX^/2l applied at point D, distance x/3 to the left of the section mn.
Positive shear force F^ and bending moment A/x act on the section boundary as
shown. From the equilibrium conditions for this free body, we find

U’o/ WoX^ )
Fx
1 zT’ I

(f)
wolx Wot’
M X
I

Fro. 5.6
 BENDING MOMENT 101

It will be noted that these expressions satisfy the relationship expressed by eq.
(5.1); i.e.,

dMj:_ wd ivcx'^
_
~d7 ~ ~ T 21
''

Setting this expression for dMr/djr — 0, we find

(r)

defining the location of that section for w'hich has its maximum value. Finally,
substituting this value of j: into the second of eqs. (f), we obtain

(h)

EXAMPLE 2. Two horizontal beams AC and BC are connecUHl by a hinge C and

reston three supports at points D, E, F, as shown in Fig. 5.7a. Both beams carry
uniformly distributed load of intensity w = 1000 lb p(‘r ft. Calculatf' the reactions
Rd, Re, Rf, and evaluate the maximum positive bending moment induced in the
beam AC.
SOLUTION. Considering CB as a fre(‘ body (Fig. 5.7b) and using the equilibrium
equations XMc = 0 and 21% = 0, we find Rf = 8333 lb and Rc = 1667 lb. Con-

•^ = 1000 lb per ft

(c) (b)

Fig. 5.7

sidering -4C as a free body (Fig. 5.7c) and using the equilibrium equations ZMd = 0
and SF, = 0, we find Re =
8333 lb and Rd - 7333 lb.

The shear force at any section mn between E and D at the distance x from A is

V X — —wx + Rs = —lOOOx 4- 8333 = dMr

—i
102 STRESSES IN BEAMS: 1

Settinj^ tliis expression equal to zero, we nonelude that Mr has a maximum value
at X - S.;i3 ft. In this ro^rion

Mr = Re (X - 4) -
vSubstitutinp; herein x = S.33 ft, Ri = s333 lb, and w = 1000 lb per foot, we obtain
- 1390 ft-lb.

PHOBLKMS
For th(' sinipl<‘ beam m
A, evaluate the shear force and bending moment
at a section just to the left of the point of application of the 4000-lb load. Arts.
r = 1025 lb; M = 27,750 ft-lb .

4000 lb

B
1T; I
! !
ri
^

6-
I

Fig. a
Sfl Calculate the bending moments at cross-sections C and D of the beam shown
in Fig. B. A ns. Mr = 0; Md - -15,000 ft-lb.

p
1200 6/n 1 w per ft

i
jgi
TTT mrni
fi*— -5'—" — 5'--'— 5'
—v/-
5.
Fig. B Fig. C
3. A simply supported beam with overhanging ends carries transverse loads as
shown in Fig. C. moment
If wl = P, what is the ratio a/l for which the bending
Mr beam will be zero? A?i^. a/l - i.
at the middle of the
4. A simply supported beam carries a linearly varying transverse loading as
sliovvn in Fig, D. The maximum intensity is Wo at each end of the beam. Calculate
the maximum positive bending moment. Ans. 3/max = 0.016 wol^.

TrnTrT>^
4 C
i-
Fig. D
A simple beam carries a linearly varying transverse loading as shown in Fig. E.

Calculate the maximum positive bending moment. Ans. 3/max = Wol^/SQyjS.

X
Fig. E
 BENDING MOMENT 103

6. Derive general expressions for normal force A",, shear force and bending
moment at the cross-section mn of the beam shown in Fig. F. .4ns. Nj = -|-F;
V, = -0.2P; M, = -0.2Pt.

C
r 30ClO lb

2 '

000
I— 4^.
.

500 lb ft
Id ft
1

n:i 1 n 1 1 ! ! ! T ijj ! ii : i

r
-4-- A L. . _ J

Fig. F Fig. G

T, Calculate the sliear force V/f and the bending moment M at the section just
to the left of the wall at P, for the cantilever beam in Fig. G. .4ns. T/# - —8200 lb;
Mb = -30,400 ft-lb.
The simple beam .4P in Fig. H carries a varying transverse load the intensity
^
of which at any point x is defined by the expression Wx — Wu sin tx/I, where Wo is
the maximum intensity at the center of the beam. Evaluate the maximum bending
moment in the beam. Ans, ~ icoP/tt*.

9. For the beam supported and loaded as shown in Fig. I, calculate the shear
force Vj) and the bending moment M
at the cross-section D. Ans, V d - —250 lb:
Md = -f 500 ft-lb.

1000 !b

1
1000 lb
nil 1 1 10 '
rT
1 1
1000
nTTnrr
lb

1
• 1 y i U — 2'

1000 lb
1

Fig. I

5.2 Shear and Bending Moment Diagrams

We have seen in the preceding article that the shear forceF* and bending
moment Mx in a transversely loaded vary with the beam will, in general,
distance x defining the location of the cross-section on which they occur.
For this reason, it is often advantageous to use a graphical representation of
104 STRESSES IN BEAMS: I

the variation of these quantities along the axis of the beam. To do this, we
let the abscissa indicate the position of the section, and the ordinate the
corresponding value of shear force or bending moment. Such graphical
representations are called shearing force and bending moment diagrams.

Fig. 5.8

To illu.strate. let us consider as an example the case of a simple beam A B

carrying a .single concentrated transverse load P (Fig. 5.8a). From equilib-
rium corisid^^ratioris of the entire beam, the reaction.s are found to be

Pa
Ra
T I

Then for any cro.s.s-section mn to the left of P, i.e., for 0 < jt < a, it can be
concluded from the equilibrium of that portion of the beam between A and
mn that

Vx = H j-j Mx = H— (a-)

From these expressions, we see that the shear force is constant between A
and the point of application of the load P, while the bending moment
varies linearly with x. For x = 0, the bending moment is zero and for

X = a it is Pabll. The corresponding diagrams are shown by the straight

lines ac and a'c/ in Figs. 5.8b and c.
 BENDING MOMENT DIAGRAMS 105

For a cross-section pq to the right of the load P, i.e., for a < x < U we
obtain in the same way

T-P^ = —
Pb P(h - 1} Pa
v, =
]

'
=-T’ i
(b)

M. = - P{x -a)= +Pa^l -

^ j

Thus again, the shear force is constant but of negative sign, while the
bending moment varies linearly with x. At x = I, its value is zero while at
X = a, it becomes Pah /I. The corresponding diagrams of shear force and
bending moment for this part of the beam are shown by the straight
lines cband c'b' in Figs. 5.8b and c.
Viewing the (‘(uuplete diagrams of shear force and bending moment, we
see that the point of appli(‘ation of the load P there is an abrupt change in
the shear force from ~{-Pb/l to —Pafl. Correspondingly, there is a sharp
discontinuity in the slope of the bending moment curve a' c'b'.
Ifseveral transverse forces act on a simple beam as shown in Fig. o.Ua,
the same reasoning as above shows tliat between transverse loads, the shear
force remains constant and the bending moment varies linearly. Thus the
siiear force* diagram will have the general torm shown in Fig. 5,9b, whil(» the
106 STRESSES IN BEAMS; I

bending moment diagram will be of the form shown in Fig. 5.9e. It should
be noted that corresponding to ea(;h abrupt change in the magnitude of
shear force there is a discontinuity in the slope dMxJdx of the bending
moment diagram.
The actual construction of the diagrams in Fig. 5.9, for given numerical
values of the loads Pi, P2, P3 and distances between their points of applica-
tion, is quite simple. First the reactions Ri and R 2 will be calculated. Then

on (‘omputing values of shear force and bending moment at only three

cross-sections — say just to the left of each applied load — the diagrams
can be constructed simply by connecting these key points by straight lines
as shown, l^'inally it should be noted that the maximum ordinate of the
bending moment diagram (under the load P2) occurs at that section where
the shear force changes sign, since here also the sign of dMjdx changes ac-
cordingly. If, going along the x-axis, the shear force changes from a
positive to a negative value at a certain cross-section (as in Fig. 5.9b),
we have a maximum bending moment at this section. A change in the
shear force from a negative to a positive value, conversely, will indicate a
minimum bending moment. In general, several changes in sign of shear
force may occ\ir along the length of the beam. If so, there will be several
maxima or minima in the bending moment and each one must be in-
vestigated to ascertain the numerically largest bending moment in the
beam.
Let us consider now a simple beam AB carrying a uniform distribution of
transverse load of intensity w shown in Fig. 5.10a. In this case the re-
as
actions are each equal to wl/2. Then at any section distance x from the
left end A we have
,

wl
Vx -- w,
.
(c)
wlx wx^
M. 2 T
(see eqs. (d), p. 98). From the first of these expressions, we see that the
shear force Vx has the positive value wl/2 at x =
0 and the negative value
— wl/2 at X = 1; furthermore, Vx varies linearly with x. Hence we may
construct the shear force diagram in Fig. 5.10b simply by drawing the
straight line ab as shown.
From the second of expressions (c), we see that the bending moment Mx
varies quadratically with x so that the corresponding curve a parabola. is

Its ordinate is zero for x =

0 and also for x = 1. It has its maximum
ordinate at x — 1/2, which is Mmax = wP/S. Thus drawing a parabola
through the points a', c\ in Fig. 5.10c, we obtain the complete bending
moment diagram as shown. Again, it will be noted that the maximum
 BENDING MOMENT DIAGRAMS 107

w per unit length

(C)

Fig. 5.10

bending moment occurs at the section for which Vx — dMz/dx 0, in ^

accordance with eq. (5.1).
In Fig. 5.11a, we consider the case of a cantilever beam carrying a
uniformly distributed load between A and C as shown. Following the sign
conventions of Fig. 5.3, we see that in this case both the shear force and the
bending moment are negative throughout the beam. Specifically, for a
section mn betw^een A and C,

From the first of expressions (d), we construct the straight line ac in Fig.
5.11b and from the first of expressions (e), the straight line ch. This
completes the shear force diagram as shown. From the second of expres-
sions (d), we construct the parabola aV in Fig. 5.11c and from the second
of expressions (e), the straight line c'5'. This completes the bending mo-
ment diagram as shown. Since there is no abrupt change in the shear force
at section C, there is correspondingly no discontinuity in slope dM jdx of
the bending moment curve at this section. Thus in PTg. 5,11c, the straight
line c'6' is tangent to the parabola a'c' at c\
108 STRESSES IN BEAMS: I

Fiii. 5.11

Wo noto in Fig. 5.11c, that the numerically largest bending moment

Mu — — occurs at the built-in end of the beam, but that there
appears to be no corresponding change of sign in the shear force at this
section. To clarify this situation, we must consider that portion of the
beam which is encased in the wall (see Fig. 5.12). Assuming that the beam
bears on the wall only at points B and D, we will have reactions

V)l 3wl\
Rb = ”2 ~r “jT" — and Ru (f)
oa 8a

directed aii shown in Fig. 5.12a. Then for any cross-section mn between D
and we have

3ieP
\ X — '\-Rd = + ^

8a
(g)
Mx ^ —RijX^ —
Sa

where x measured to the left from D, as shown. From these expressions,

is

we may now construct diagrams showing shear force and bending moment
variation along the encased portion BD of the beam (Figs. 5.12b and c).
These diagrams show^ that there is a change in the sign of the shear force
from negative to positive at B which confirms the condition of a minimum,
i.e., a negative maximum of M* at B.
 BENDING MOMENT DIAGRAMS 109

It will be shown in the next article that we may expect the highest
bending stresses beam at the section where the bending moment
in a is

numerically a maximum; therefore, the location of such sections is of

practical importance. They are most easily identified by a change in sign of

(c) (c)

F»ci. 5.12 Fig. 5.13

the shear force which can he easily detected from a study of the shear force
diagram. However, there are many other useful and significant applications
of both shear force and bending moment diagrams, as will be seen later.
EXAMPLK 1. Construct shear force and bending moment diagrams for the
cantilever beam loaded as shown in Fig. 5.13a.
SOLUTION. In the region 0 < a; < a, we have

y, = -P; M. = -Px. (h)

In the region a < x < I, we have

F.= -P+P = 0,

M, = -Px + P(x - a) = -Pa.j

The corresponding diagrams are shown in Figs. 5.13b and c.
EXAMPLE 2. Construct shear force and bending moment diagrams for the simple
supported beam with overhang, loaded as shown in Fig. 5.14a.
SOLUTION. The reactions are
Ra = 0, Rb ^ wi
Between A and C, we have
y. -0. ii/, -0, (j)
no STRESSES IN BEAMS: I

where x is measured to the right from A .

Between C and B, we have

= -wx, Mx = — (k)

where x is measured to the right from C.

lietween B and D, we have

F. = - xy = -^(1 - X^' (1)

where x is measured to the right from B. I'he corresponding shear and moment
diagrams are showm in Figs. 5.14b and c.
EXAMPLE 3. Construct shear force and bending moment diagrams for the simply
supported beam with overhanging ends loaded as shown in Fig. 5.15a.
SOLUTION. From equilibrium of the entire beam, w^e find Ri = = P(1 + 2a/l)
directed as shown. The shear force in each overhang is constant and equal to — P.
The shear force in the middle portion is constant and equal to — P P(1 + 2a/ L) +
= +2Pa//. The corresponding shear force diagram is shown in Fig. 5.1 5h.

*y'/unit length

P P^

o-

• I (o)
I I

ZPa
I I lEs.
X

(b)

c\Pa

bPa (c)

Fig. 5.15

Since the beam issubjected only to concentrated forces, the bending moment
must vary linearly between points of application of transverse loads. The bending
moment over the left support is —Pa, while that over the right support is +Pa.
Plotting these points in Fig. 5.15c and drawing the straight lines a5, bcy and cd, we
obtain the complete bending moment diagram for the beam as shown.
 BENDING MOMENT DIAGRAMS 111

PROBLEMS
1. Construct, to scale, the complete shear force and bending moment diagrams
for the beam in Fig. A, p. 102.
For the beam in Fig. C, p. 102, the following numerical data are given: P - 50
2.
“ 200 Ib/ft, / = 10 ft, a = 3 ft. Construct shear force and bending moment
lb, te

diagrams to scale.
3. Construct general shear force and bending moment diagrams for the simple
beam loaded as shown in Fig. D, p. 102.
4. Construct general shear force and bending moment diagrams for the simple
beam loaded as shown in Fig. E, p. 102.
5. Construct to scale, the complete shear force and bending moment diagrams
for the cantilever beam loaded as chown in Fig. G, p. 103.
6. Construct, to scale, the complete shear force and bending moment diagrams
for the beam in Fig. I, p. 103,

24.000 lb

+ fOOOlb
.
2'-—
LUuuuLiLiuLiLiLiLi. 1 f B
L I%rTiTT>-.r
, - n 1 /

Ml
1

1
11
n
L
n
4' -4,
J 111 111111 11 u LI ij 11Ll
S' -k
D
4 — 4'

-1000 lb

Fig. a Fig. B

7. The beam shown in Fig. A carries a total distributed load W = 24,000 lb,
between D and E and is supported by a uniformly distributed reaction between A
and B. Construct shear force and bending moment diagrams to scale.
8. The shear force diagram for a simple beam supported at its ends is shown in
Fig. B. Construct the loading and bending moment diagrams.

5.3 Bending Stresses in Beams

Let us consider the beam AB transversely loaded as shown in Fig. 5.16
together with its shear force and bending moment diagrams. We note that
the middle portion CD of the beam from shear force and that its
is free
bending moment Mx = Pa is uniform between C and Z>. This condition is
called pure bending.
To investigate the state oi internal stress produced by pure bending, we
must examine the deformation which takes place within the material. In so
doing, we shall assume that the beam is prismatic and that it has an axial
plane of symmetry which we take as the xy-plane." W’^hen the applied loads
also act in such a plane of symmetry, bending will take place only in that
plane. We assume further that the material is homogeneous and that it
obeys^ Hooke’s law, the modulus of elasticity in tension being the same as
that in compression. Then since the bending moment is uniform between C
and Dj it is reasonable to assume that the bending deformation will also be
irj STRESSES IN BEAMS: I

p p uniform; i.e., the portion CD of the

beam will take the form of a circular
arc as shown in Fig. 5.17. In this de-
formed configuration, each cross-
section, originally plane, is assumed
to remain plane and normal to the
longitudinal fibers of the beam.*
As a result of the deformation
shown in Fig. 5.17, fibers on the con-
vex side of the beam are elongated
slightly while thoseon the concave
side are shortened slightly. Some-
where in between the top and bot-
tom of the beam, there is a layer of
fibers which remain unchanged in
length. This is called the neutral
surface.The intersection of this
neutral surface with the axial plane of symmetry is called the neutral axis
of the beam. Its intersection with the plane of any cross-section is called
lh(* neutral axis of that section. After deformation, the planes of two adjacent
cross-sections inn and jpq intersect at 0. We denote the angle between these
planes by dd and note that dd = dx/p where 1/p is the curvature of the
iKMitral axis of the beam.
In Fig. 5.17, we now draw through point b on the neutral axis, a line p'q'
parallel to mn and indicating the original orientation of the cross-section pq

•Careful strain measurements in the laboratory confirm this assumption for the case
of pure bending.
 VARIOUS SHAPES OF CROSS-SECTIONS J13

before bending. From we see that the segment cd of any

this construction,
fiber at the distance ij from the neutral surface elongates by the amount
(I'd = yds. Since its original length was cd' = dx, the corresponding
strain is

ydB
= -^ = n/p
.

(a)
.

If a fiber on the concave side of the neutral surface is considered, the

distance /y will be negative and the strain is also negative. ITius all fibers

on the convex side of the neutral surface are in tension while those on the
concave side are in compression. Experiments indicate that the lateral
deformation of fibers is tlie same as in simple tension and compression.
Thus the stress in each fiber will be proportional to its longitudinal strain,

E^ (5.3)

Thi.sshows that the fiber str(‘sses ax due to pun* bending vary linearly with
distance from the neutral surface, so long as the material follows Hooke’s
.7

law. Such stress distribution over the depth of the beam is shown in Fig.
5.18. The position of the neutral axes Oz of tlu*. cross-se(‘tion may nt)w be
found from the condition that these stresses distributed ovr*r the section
must give rise to a resisting eouple M.

Let dA denote an element of area of the cross-section at the distance y

from the neutral axis (Fig. 5.18). Then the eleiVieiit ol force on this area
is axdA, Using eq. (5.3), this becomes

cr.dA =-ydA.
P

Now resultant normal fonje Nx on the section (puic

since there must be no
114 STRESSES IN BEAMS: I

bending), the integral of over the entire area of the section must
vanish, i.e.,

y dA =0. (c)
-f
p Ja

Since E/p ^ 0, we conclude from this that

dA = Aye = 0, (d)
Ly
where A is the total cross-sectional area and yc is the distance from the
neutral axis to its centroid. Finally, since A 0, we conclude that Vc — 0.

Thus tke neutral axis of the cross-section passes through its centroid.
The moment of the elemental force (XxdA about the neutral axis of the
section is dM = ya^dA. The sum of these elemental moments over the
total area of the section must produce the bending moment on that M
section. Thus

M I
ya.dA =-
P JA
/
i/dA. (e)
JA

The integral in this expression in which a summation is made of each ele-

ment of area dA* multiplied by the square of its distance from the neutral
axis ( 2 -axis), is called the moment of inertia of the cross-sectional area with
respect to that axis.* Introducing the tiotation

(f)

for this quantity, cq. (e) may be written in the form

I ^ K
El
(5.4)
p

This shows us that the curvature l/p of the axis of the beam is proportional
to the bending moment M
and inversely proportional to the quantity E/,
called the flexural rigidity of the beam. We see that this flexural rigidity
reflects both the stiffness of the material as measured by E and the pr(‘por-
tions of the cross-sectional area as measured by 7.
Substituting the value of 1/p from eq. (5.4) into eq. (5.3) above, we obtain

My
O-x (5.5)
I

It is seen that this bending stress will be a maximum in those fibers furthest
removed from the neutral surface, tension on the convex lower face of the
beam and compression on the concave upper face. Denoting the distances
•See Appendix B, p. 351.
 VARIOUS SHAPES OF CROSS-SECTIONS 116

to extreme fibers in tension and compression, respectively, by ci and as

shown in Fig. 5.18, we obtain from eq. (5.5)

Mci, il/Co
^max '
0’'inin (5.5a)
I I

If the cross-section is S3’'mmctrical with respect to its centro^dal axis,

Cl = C2 = c and the extreme fiber stresses in tension and compression
are equal.
Introducing the notations
/
2i, (g)
Cl

called the section moduli^ eqs. (5.5a) can also he expressed in the form

^^max
M _ M (5.5b)
1 ^2

In the case of a rectangular cross-section of width b and depth h (Fig. 5.19a),

Cl = C2 = h/2,
bh\ bh^

For a circular section of diameter d (Fig. 5.19b), ci = C2 c//2,

icd?
/ = 32*

In the case of the trapezoidal section shown in Fig. 5.19c, Ci < C2 which
makes Zi > Za. Thus if the beam is bent concave upwards, the maximum
compressive stress in the fibers of the top face will be greater than the maxi-
mum tensile stress in the fibers of the bottom fa'ce. In the case of a beam
made of cast-iron this may be advantageous, since cast-iron is stronger in
compression than it is in tension. Various shapes of cross-sections for beams
will be discussed in more detail in the next article.
All of the foregoing theory has been developed for the case of pure
bending, i.e., constant bending moment along the length of the beam. In
lie STRESSES IN BEAMS: I

such case, the shear force at each cross-section

is zero and the normal

due to bonding are the only ones produced. In the case of non-
stresses
uniform bending of a beam where the bending moment varies from one
cross-section to another, there is a shear force at each cross-section and
shearing stresses are also induced in the material.* The deformation as-
sociated with these shearing stresses causes warping of the various cross-
sections so that plane cross-sections before bending do not remain plane
after bending. This complicates the problem, but more elaborate analysis
shows that the normal stresses due to bending, as calculated from e(j. (5.5),
are not greatly altered by the presence of these shearing stresses. Thus it is

justifiable to use the theory of pure bending for (calculating the normal
stresses in the case of non-uniform bending and it is accepted practice to
do so.

EXAMPLE 1. A
locomotive axle with outboard bearings A and B will be subjected
to transverse loads as shown in Fig. 5.20. Determine the maximum bending stress a
induced in the axle if a = 13.5 in., / — 59 in., diameter d ~ 10 in., and P — 26,000
lb. Find, also, the deflection 6 at the mid-point of the axle.

P P

SOLUTION. The bending moment between wheel loads at Z) and E is M= Pa —

26,000 X 13 5 in. -lb. The section modulus is Z - 7rdV32. Thus eq. (5.5) gives

_ M 26,000 X 13.5 X 32
cr... - ^ TT X 10’
3580 psi.

From cq. (5.4), the radius of curvature of the circular arc DCE, in pure bending, is

_ £/ _ 30(10)’ X^X (10)« _ .

M 26,000 X 13.5 X 64
’

Now, referring to Fig. 5,20, we see that the deflection CoC = 5 is

5 = p(l — cos B), (h)

where B = l/2p = 0.000705 radian. For such a small angle, we may take cos B =
1 — ^V2. Then cq. (h) becomes
59 X 59
0,0104
2 8p* " 8p 8 X 41,900
in.

‘These shearing stresses are discussed in Art. 5.5.

 VARIOUS SHAPES OF CROSS-SECTIONS ii;

LXAMPLE Calculate the rnaximuTn bending stress that will bo induced in a

2.
steel wire of diameter d — }-i 2 m. if it is wound on a drum of diameter D — 20 in.
SOLUTION. For this ealeulation, wc use eq. (5.3). Thus

A- .3000)^ 1
, _ _ 4f),ll)0 psi.
" p' 10.016 ^64
It must b(‘ assumed that this does not exceed the yield point of the steel, so that the
material behaves elastically.
EXAMPLE 3. A simple beam 20 ft long is to carry a uniform load of 480 lb per ft,
including the weight of the beam itself, (a) Select a suitable Douglas fir beam with
rcf-tangular si'ction of depth-width ratio h/b ~ 1.2 and an allowable^ working stress,
<Ju. = 1200 psi, (b) Select a suitable st^‘el beam of standard 1-section to carry the
same* load, if the allowabh^ stress is Gu - 18.000 psi.

SOLUTION. Referring to Fig. 5.10, we see that, for a umformls' loaded simple
l)(‘am. the maximum bimding moment at the middle is

„
“
480 X 20 X 20
.17 111 fix 8* " 24,000 ft-lb. ( 1
)
K

Also, for a rectangular cros.s-section of d(‘pth h and width b ~ /pT.2, the section
modulus is

5/r h'^
' ""
7.2

'Fhen with - Gu ~ 1200 psi, eq. (5.5b) becomes

24,000 X 12 X 7.2^
1200

from w'hich

Tlius the wood beam should have a 10-m. X l2-in. s(‘ction.

For the steel I-beam, eq. (5.5b) becomes

18,000Z = 24,000 X 12,

from w^hich tiie required section modulus Z — 16 in.? Now turning to Table B.3 of
Appendix B, p. 364 we find that an 8T-23 standard I-beam has a sc'ction modulus
Z — 16 in.^ and will be satisfactory.

PROBLEMS

1. A thin steel rule having a cross-section 0.025 in. X 1.00 in. is bent by couples
applied at its end.s = 10 in. of the circular arc subtends a central
so that a length I

angle 6 = 60°. Calculate the maximum stress induced in the rule and the magnitude
of the bending moment M. Ans. a = 39,300 psi; M = 4.09 in -lb.
118 STRESSES IN BEAMS: I

iT A girder simply supported at A and B is uniformly loaded on the overhangs as

shown in Fig. A. The intensity of load is w - 10 kip /ft. Th(‘ girder is of standard

I-section of depth d = 30 in. and having a moment of in<Ttia / = 9150 in.'* De-
termine the ma.ximum fiber stress in the middle p(>rtion of th(' beam and the

upward deflection 8 of the mid-point C. Am. = 9S40 psi; 8 = 0.157 in.

3A 4A

Fjg a Fig. B

3. Rej^eat the .solution of the })rereding problem if the beam is of wood and has
a squan’ cross-s('ction 12 in. X 12 in. and a modulus F= 1.5(10)® psi. Assume that
w == 1000 lb/ ft (Truax - 20S0 psi; 8 - 1.67 in.
4. A simple beam of span / = 14 ft, (‘arries a uniformly distributed load of
~
intensity w 250 Ib Tt. Caiculate the maximum bending stress if the beam has a
rectangular cross-section of width b ~ 5| in, and depth b = 7f m. Ans. =
13.50 psi.
5. A wood beam having a r('etangula,r eross-se('tif>n H in. wid(‘ X 12 in. deep is

supported and load(‘d as shown in Fig. H, Locate the ‘‘dangerous section” and
calculate the maximum bending stress. Arts, (Tma* = 1050 psi.
6. A temporary dam in a water channel i.s formed by setting vertical 3 X 12-in.
planks betwetm guide rails A A and BB as shown in Fig. C. Assuming no support
at D, cakailate th(‘ maximum bending stress induced in eaidi plank whim the water
depth on the left side of the dam is 6 ft as shown. Am. crn,Bx = 2000 psi.

Fig. C Fig. D

7. During a certain manufacturing operation each spoke of a flywheel (considered

as a cantilever beam) is subjected to the loading shown in Fig. D. Each spoke has
elliptical cross-section whose major and minor axes vary from X 2J in. at A
to 4 X 2\ in. at B. Calculate the maximum bonding stress induced in the spoke.
Ans. (Tmax = 1650 psi.
8. Calculate ^he center-to-center spacing of 2 X 8-in, floor joists to carry a floor
load of 50 lb per sq ft if the clear span is 16 ft and the allowable working stress
(Tu. = 1000 psi. Am. 13.3 in.
 VARIOUS SHAPES OF CROSS-SECTIONS 119

9. Calculate the nominal depth h of floor joists spaced at 16 in. center to center if
the floor load is 140 lb per sq ft. The clear span is 10.5 ft and the joists are 2 in.
thick; the working stress == 1000 psi. .Irw. 10 in.
10. Astandard 1-boain 20 ft long is to be supported and uniformly loaded as
shown in Fig. E.The supports A and B will be so placed that the negative bending
moment over either support will be numerically equal to the positive bending
moment at mid-span. If the allowable bending stress otu- = 18,000 psi, select a
suitable standard I-beam. Ans. 71-20.

12.
Fig. E Fig. F

M. A cast-iron beam 12 ft long is supported and loaded as shown in Fig. F. The

cross-section of the beam is an inverted T, 6 in. de(*p by 4 in. wide, stem and flang(‘
each being 1 in. tlnek. Find th(‘ safe magnitude of P, if working strcs.se.s in k‘nsion
and compression are (Ti = 6(X)0 psi, (7^. = 10,000 psi, respeetiveiy. Ans. P = 1670 lb.
A standard channel beam OOL.120) is bUi)ported and loaded as shown in
Fig. G. The triangular distribution of load is defined by its maximum intensity Wo
at B. If the allowable working stress (tension or compression) is ir,,, = 20,000 psi,
what is the safe value of ico? .4 ns. wo = 780 Ib/ft.

Fig. G Fig. H
13. Aheavy wheel of weight W
= 4000 lb is rolled very slowly across a simple
beam AB of span Z = 20 ft (Fig. H). Calculate the maximum bending stress in-
duced in the beam, if its section modulus Z - 12 in.^ Ans. (Tmai = 20,000 psi.
14. A simple beam AB carries two wheel loads Wi - 1000 lb and \V j = 2000 lb
as shown in Fig. I. The wheels are 5 ft apart but can have any position on the beam
as defined by the distance x. Find the value of x to make the bending moment under
the wheel W
2 a maximum and calculate the corresponding maximum bending stress.

The section modulus of the beam is Z = 7.3 in.^ Ans. x = 5.83 ft; = 20,700
psi.

Fig. I Fig. J
120 8TRESSES IX BEAMS: I

15. Prove that the depth h nf cantilever beam of rectangular eros-^-section

should vary linearly with x to attain a beam of constant strength under uniformly
distribuk^rl load (Fig- *1) \s.sum(' that the flexure formula for prismatic beams
applies.

5.4 Various Shapes of (!ross-Sec*licms ol' fleauis

From the discussion in Art 5 d it follows that the niaxinmrn tensile or

compressive s1res.> in a IxnU bar is proportional to the disTanei' of th(‘ most
remote fibers from the neutral axis. Heiua* if tlu* material has the same
strength m tensi(»n and compressifin, it will be logical to choose shape's oi
cross-section ni whi(*h the centroid is at the middle of tlie depth of the Ix'ani
In this maniif'r the saiiu' t.aclor ol safety for libers in tension and for those
in cornpressKm \^ ill be obtaiiuxi This is th<‘ underlying idea in the choice of

sections symnK'trical with respe'ct to the neutral axis for materials su(‘h as
structural steel, which have about the same yield point in tension and com-
pression. Snch shown in Fig. 5.2 la, b, and e. If the
cross-scci lolls aie
seetion is not symmetrical vvuth respect to the neutral axis, for example a
rail section, tlu* material is freciueiitly distributed between the head and the

base so as to have the centroid near the middle ol its height.

For a material of low strength in tension an<l high strength in compression
for example, cast-iron or concrete - the best cross-sect ii/ii for a beam is

not symmetrical with respect to the neutral axis but is such that the dis-

(c) (d)

Fio. 0.21

taiices.ri and co from the neutral axis to the most remote fibers in tensicn

and comprc'ssion are in the same proportion as the strengths of the mat/Crial
in tension and in compression. In this manner equal strength in tension and
compres.sion is obtained. For example, with a 7’-section (Fig. 5.21d), the
position of the centroiil of the section may lie adjusted along the height of
the section by properly proportioning its flange and web.

In designing a beam to undergo bending, not only the conditions of

strength should be satisfied but also the condition of economy in the weight
of the beam. Of two cross-sect ions having the same section modulus, that
is, satisfying the condition of strength with the same factor of safety, the
one with the smaller area is more economical.
 VARIOUS SHAPES OF CROSS-SECTloXS 12J

In comparing various .sha])os of cross-sc^'tions, consid('i‘ ^rst the rec-

tangular section of depth // aj)d width h ( Uig. r).2\a'i The section luodvilns is

Z (a)
6

where A denotes the cross-sectional area. It is seen that the re(*t angular
cross-section becomes mon^ and more economical with increase in its
depth h. However there is a certain limit to this increase and the question
of lateral instability of the beam arises as the section becomes too narrowv
The collapse of a beam of very narrow^ rectangular section may be due not to
overcoming the strength of tlv material but to sidewise buckling.*
In the case of a circular cross-section of diameter r/ (Fig. 5. 2 lb),

ird^
(b)
32

Comparing circular and scpiare cross-sections of the same area, the side k
of the square will be h = d\^/2, for which eq. (a) gives

Z = 0.147 .4d.

Comparison of this wuth lb) shows a sciuare cross-section to be more

economical than a circular one.
Consideration of the stress divStribution along the depth of the cross-
section (Fig. 5.18) leads to the conclusion that for economical divsigu most of
the material of the beam should be put as far as possible from the neutral
axis. The theoretically ideal case for a given cross-sectional area A and
depth h would be to distribute each half of the area at a distance h/2 from
the neutral axis as shown in Fig. 5.22. Then

and the section modulus becomes

Fig. 5.22 Fig. 5.23

*For a discussion of lateral buckling of beams, soe Timoshenko, Elastic Stability^

McGraw-Hill Book Co., Inc., New York, 1960.
122 STRESSES IN BEAMS: I

This ideal limit, although unattainable, may be approached in practice by

using an I-section or wide-flange section with most of the material in the
flanges (Fig. 5.21e). Due to the necessity of putting part of the material in
the web, the limiting condition (c) can never be realized; but for standard
wide-flange sections^ we have

(d)

Comparison of (d) with (a) shows that an I-section us more economical than
a rectangular section of the same depth. Furthermore, due to its wide
flanges, an I-beam will be more stable with respect to sidewise buckling
than a rectangular section having the same depth and section modulus.
In the design and selection ‘of beam s(‘ctions, reference is usually made to
a handbook* giving various propc'ities such as moment of inertia, section
modulus, etc,, of standard structural shapes.These shapes such as I-
sections (Fig. 5.21e) and U-sections (Fig; 5.21f) are usually specified by
the nominal depth in inches and the weight in pounds per lineal foot of
beam. An 8-in., 18.4-lb I-beam, for instance, is 8 in. deep, weighs 18.4 lb
per ft, and is designated simply by 81-18.4. Such beams having extra wide
flanges are specified in a separate class. For example, the notation 12WF65
denotes a wide-flange section having a nominal depth of 12 in. and weighing
65 lb per lineal foot.

EXAMPLE 1 A prumatio beam in pure bending has a traf)ezoidal cross-section as

.

shown in Fig. 5.23, the top fibers being in compression. If allowable working
stresses in Umsioii and compression are at = 5000 psi and ac = 8000 psi, calculate
the ratio of bases hi/b^ for maximum economy.
SOLUTION. Let Cl and denote distances to extreme fibers from the neutral
(centroidal) axis as shown. Then from eq. (5.5a) we have

8000 = yci,

from w'hich ci/c2 = 8/5. Also ci = h. Hence

Cl = (c)

The distance yc to the centroid of the trapezoid is

h( b2^2b, \
(0
3\ 62 -h /

•See, for example, Sled Construction by the American Institute of Steel Construction,
New York, 1959; also Appendix B-1, p. 357,
 VARIOUS SHAPES OF CHOSS^SKCT IONS 123

Fxjuating this to the valiii* of e-j from (o), wo find

hi 11

KXAMPLK 2. A heam of square eross-sectioii

a X a is bent in the sertieal plain' of one of its

diagonals as shown in Fig. 5.24. Show that, fo^ a

given moment, the maximum fiber stress(‘^ can
be reduced by cutting off the shaded corners to a
depth aa/ \^ and calculate the optimum \alue of a.
SOLUTION. The inomciit of inertia of the com-
plete cross-section is / — a*/ 12 and the cor-

responding section modulus is Z — 12 (see

Ap})endix B). The mom(‘nt of inertia of the
reduced section will be obtained by adding to the
moment of inertia of the square baie about its
diagonal hd that of the two parallelogram.s rfdh
and g(dh about their common base dh. Thus

1 — a - ^^y2 aa <7(1 ~ a) <d( l

— a)'^
4-2 (1 --f 3.^)
l2~ i2

and the corrc'sponding s(‘cti(>n modulus becoim'S

z = -
<7(1
p——- a),
= 7^7
12
«H1 - «)' n + 3«). (g)

'Phis sc'ction modulus is a maximum for that value of a which makes dZ/dcc -- 0.
Making this diffiTcntiation. we find a - }(> ami the corresponding niaxirnuin section
modulus becomes, from (g),

V2a-'
- 1.053 ’
(h)
12

Thus, by cutting ofT th<' corners, is increased by about 5 per

the section modulus
(‘(all and the maximum bending be reduced by this percentage.
stre.sses will
To understand how this removal of material actually incre'ascs the stnmgth of the
s(H*tion, v\c must observ(‘ tliat removal of the small corner areas simply reduces the
distance to extreme fib('rs m a. gu'ater proportion than it reduca's the moment of
inertia of th(‘ section. Thus tla^ section modulus Z — I/c is actually increased.

PROBLEM.S

1. A simply supported cast-iron beam is to ha\e the inverted T-section shown in

Fig. A. the allov'able stresses for cast-iron in Uaisioii and compression are
If
—
(Ti 4000 psi and a, 8000 psi, calculate the propiT stem thickness t of the section.
Am, i - 2 in.
2. For the channel lieain shown in Fig B, it is desired to have the ratio of extreme
fiber bonding strt*sses o', : cr, =3:7. What is the profx^r wail thickness t to realize
this condition? Am. i = 2 in.
124 STRESSES IN BEAMS; I

-H*j r K-

Fig. a Fig. B

3. A prismatic bar having a semicircular cross-section is subjeijted to pure bending

as shown in Fig. C. Caiculab^ the ratio at/a, of extreme fiber Jesses in bending.
Ans. ai/a, = 0.736,
i.

Fig. 0 hiG. D
E A rertanguiar wood beam i.s to be cut from a circular log of cliarnerei d (Fig. !>,).

C'alculute the ratio 6 'h to attain a beam of rna.vimum strength in })ure bending.
Ariy, h/ h ~ \ \2.
3. show that the section modulus of a circular cioss-soction
H(‘fcrriTig to Fig. E,
will be iiKToaseti by cutting off the shaded segments to a depth d and find the value
of d to make the section modulus a rnaxinium, At?,?, 6 O.Oild.
6. A 12 X
square pine timber to be used as a beam is given somew^hat the
J2-iri.

form cjf serriii ircuiar grooves of (iiameter d ~ 1(1 in, along the
an {-beam by cutting
sides of the beam (Fig, F) If the allowable working stress for the material in tension
.

or compression is a,, = 1400 p.si, what is the maximum beading moment that such
a beam can safely resist? Ans. — 24,0(X) ft-lb.
7, A makeshift steel I-beam is formed by welding 1 -in. -diameter steel pipes to the

edges of a i-iri. steel plate 10 in. deep as shown in Fig. G. The wall thickness of the
pipe is i in. Calculate the section modulus Z. Ans. Z - 6.95 in,*’

Ftg. K Fig. F Fig G Fig. H

 VARIOUS SHAPES OF CROSS-SECTIONS 125

8 . A beam having originally a cross-section in the form of an equilateral triangle

is to be cut off slightly at the top as shown in Fig. H. Calculate the value of a,
defining the amount cut off, in order to make the section modulus Z2 = I/C 2 for the
remaining trapezoid a maximum. Ans. a ~ 0.13.

5.5 Shear Stresses in Bending

In Art. 5.1 we have seen that in the case of a beam bent by transverse
loads there is, in general, both a bending moment and a shear force Vx at
each cross-section. In Art. 5.3 we have seen how the bending moment
represents the resultant of a certain linear distribution of normal stresses ctx

over the cross-section. Similarly, the shear force on any cross-section

must be the resultant of a certain distribution of shear stresses r over that
section. The question to be considered now is just what that distribution
must be to satisfy equilibrium conditions of the various elements of the
hesLm.
We begin with the simplest case of a beam of rectangular cross-section
(Fig. 5.25) and assume that Vx is the resultant shear force on a chosen
cross-section. Dividing the cross-section into infinite.simal strips parallel to
the 2 -axis, it is reasonable to assume that on any
one' such strip, the shear

stress r will beuniform across the width b of the beam and parallel to the
^-axis as shown. We also know from the discussion of Art 2.1 that such
shear stress on one side of the prismatic element mn must be accompanied
by equal shear stresses on each of the other three sides of the element (see
Fig. 5.25). Thus, we must observe at the outset that there will be horizontal
shear stresses between layers of the beam parallel to the neutral plane as w^ell
as transverse shear stresses between cross-sections and that at any point in
126 STRESSES IN BEAMS: I

the beam these complementary shear stresses are equal in magnitude. This
at once leads us to conclude that the shear stress t must vary as we go from
top to bottom of the beam. For if we consider the prismatic element m'n'
whoso upper face coincides with the free top surface of the beam, we see
that the shear stress t must vanish. The same conclusion holds for an ele-
ment one face of which coincides with the bottom surface of the beam.
Thus the shear stress r must vary with y and must vanish at y = ±zhl2.
The key to the true law of variation of r with y comes from examining the
equilibrium of an elemental block pnpiUi of the beam between two adjacent
cross-sc^ctions mn and mini distance dx apart as shown in Fig. 5.26a. The
bottom face bd:v of this block coincides with the bottom surface of the beam
and is free from stress. Its upper face is parallel to the neutral plane and at
the arbitrary distance yi therefrom and is acted upon by the horizontal
shear stress t existing at this level. The end faces of the block are repre-
sented by the shaded portion of the cross-section of the beam as shown in
Fig. 5.26b. These end faces are acted upon by the normal bending stresses

(c)

Fig. 5.26

<r. The complete elemental block, isolated as a free body, is shown in

Fig. 5.26c. There are. of course, also shear stresses on the end faces pn and
piUi, but we will be interested only in the equilibrium of this block in the x-
direction; hence, these shear stresses will not enter in the equation of
equilibrium and are not shown on the free body.
Before going further, we may observe at once that if the bending moment
in the beam has the same magnitude at sections mn and mini, the nomal
stress distributions on the ends pn and pi/q of the elemental block (Fig.
5.26c) will be identical. Then for equilibrium of the block in the x direction,
we conclude that r == 0. This simply verifies the fact that pure bending
can induce no shear stresses in the beam.
Consider now the more general case of varying bending moment, de-
noting by M
and M
dM the moments at the cross-sections mn and
mini, resjjectively. Then the normal force acting to the left on an elemental
area dA of the end face pn of the block will be, from eq. (5.5),
 SHEAR STRESSES IN BENDING 127

My
(rdA dA.
I

The sum of these forces over the face pn of the block will be

dA. (a)
Jy. I

In the same manner, the sum of the normal forces over the face piUi of
the block becomes

f
(M -b dM)y
>

dA. (b)
iv. I

The shear force acting on the upper face of the block is

rh di, (c)

where r is the shear stress at the arbitrary level /yi. The forces (a), (b),
(c), must be in equilibrium; hence

rb dx

from which

r ydA,
dx Ibjy^

or. by using eq. (5. 1),

r i/dA. (5.0)

The integral in this expression is seen to represent the statical momxni^

about the neutral axis of the cross-section, of tht^ shaded portion ppnn
of the cross-section, i.e,, of that portion of the cross-section below (or
above) the arbitrary level y\ at which the shear stress r is re«][uired. De-
noting this statical moment by Q, eq. (5.6) takes the form

VQ
" = 7^- (5.7)

To how this shear stress varies with

see the distance \h from the neutral
axis,we must now examine the variation of Q with t/i-

Referring to Fig. 5.26b, we see that for a rectangular cross-section

dA - bdy and Ci = /i/2, so that

ydA — b (d)
128 STRESSES IN BEAMS: I

This same statical moment will be obtained as the product of the shaded
area

and the distance

of its centroid from the neutral axis of the cross-section

vSubstiUiting (d) into eq. (57), we obtain

T = (5.8)

This shows tliat the shear stress r vanes parabolically with fji as shown in

Fig. 5.27b. When = dihf2^ r = 0, and w^hen //i

== 0, = Vh-/HI.
Noting that / = bh^/ 12, we have

w^here A hh is the total area of tlie cross-.se<*tion Thus the maxinmin

shear stress (horizontal or vertical) oc(*urs at the neiurat axis (/yi - 0 )

and is 50 per cent larger than the average shear stress VIA,

Since the shear stress r varies from top to bottom of the beam, it f* Hows
that the shear strain 7 = r/G must vary in a similar way. Thus originally
plane cross-sections of the beam become warped after bending. This
warping can be demonstrated by oending a rubber beam as shown in
Fig. 5.28. If straight lines mn and 'pq are scribed on the side of the beam
before bending, they will become curved lines m'n' and after bending
as shown! in the figure. At the points m\ p\ n', q\ the shearing strain is
zero so that the curves m'n^ and p'g' remain normal to the upper and
lower surface of the beam after bending. At the neutral surface, the angles
between the tangents to these curves and the normal sections mn and pq
 SHEAR STRESSES IN BENDING 129

are equal to the shear strain r^nax/G. As long as the shear force W remains
constant along the beam, the warping of all cross-sections is the same;
that is mn' = pp\ nn' — qq\ etc. Thus the sh(*ar stresses r do not con-
tribute to the longitudinal strains in the fibers and the distribution of
normal stresses cr is the same as in the case of pure bending.
A more elaborate theoretical investigation of this problem shows that
warping due to shear strain does not substantially" affect the longitudinal
fiber strains, even when W
varies along the beam^ provided this variation
is continuous as in the case of a beam carrying a uniformly distributed
load. In the case of concentrated loads on the beam, however, the stress
distribution near points of application of external loads becomes more
complicated, but these irregularities are very localized and do not appreci-
ably affect the over-all stress distribution in the beam. Thus it is usually
justifiable to use flexure formulas derived for pure bending in the case*, of
non-uniform bending, as wo have already done in the preceding articles.
In discussing the distribution of shear stress in the web of an I-beam,
\ce r*an proceed in about the .viirie manner as fora beam of rectangular
cross-section. Referring to Fig. 5 29a, lei //. denote the level at w^hich
wc wish to evaluate the shear stn Then with the dimensions of the
section as shown, tlu‘ statical moinent. of the shaded portion of the cross-
section about th<‘ neutral axis is
130 STRESSES IN BEAMS: 1

Taking yi = 0, this gives for the maximum shear stress at the neutral
axis,

Taking //i = /ii/2, the minimum shear stress in the web at the junction
with the flange becomes

From eq. (5,7'), it is seen that between tji = 0 and //i — hi/2f the shear
stress varies parabolically as represented by the horizontal ordinates of
the diagram in Fig. 5.29b. The area of this diagram multiplied by the
thickness t of the web represents the total shear force V\. carried by the
web. Calculations show that for the usual proportions of standard I-

sections and WF-sections, V,v- F, i.e., the flanges contribute very little
to the total shear force V carried by the cross-section, and need not be
considered.
In this regard it will only be pointed out that eq. (5.7) is not applicable
in studying shear stresses in the flanges anyway, since it is not justifiable
to assume that the shear stress will be uniform across the width 6 of a
flange. For example, we see at once that for iji hi/2, the shear stress
over the free surfaces ah and cd of the flange must be zero, while across
the junction be, it has the value given by eq. (f). This non-uniformity
across the width b of the flange prevails to a considerable extent throughout
the flange, and is too complex to be analyzed by elementary methods.

6"
i
t „
10

(o) (b)

Fig. 5.30

EXAMPLE 1. A simply supported wood beam of rectangular cross-section carries

a concentrated load P at its mid-section as shown in Fig. 5.30. Allowable working
stresses in tension or compression and in sliear parallel to the grain are given as
follows; = 1000 psi, ru, = 160 psi. \\'hat is the safe value of the load P?
SOLUTION. The cross-sectional area A and section modulus Z are

..1 = 6A = 60 in.‘; Z = —o
hh}
== 100 in.=
 SHEAR STRESSES IN BENDING 131

The maximum shear force V" and maximum bending moment M are
P PI
r = = 12Pm.-lb.

Then from cqs. (5.9) and (5.5), wo have

dr P_ M \2P
~ 2 80’ Z “ lOO'

Setting th(‘se maximum stresses equal to tlie given working stresses and solving for
the load P in each ease gi\'es

Pi = 80 tu- = 12,800 lb, based on shear stress,

Pa = 8.33(ru; = 8333 lb, based on bending stress.

rhus the bending stress go\'erns and the safe load is P = 8333 lb.
KX\Mp].E 2. A beam AP supported at its (Mids has a span / — 4 ft and carries a
uniform load of intensitv w ~ 2(XX) lb ft over the full span. The cross-section of the
beam is a T-seetion having the dimensions shown in Fig. 5.31. Calculate the
maximum shear stress r, induced in th(‘ beam.

Fia. 5.31 Fig. 5.32

soLuaioN. The maximum shear force occurs on a section just inside one of the
supports and has the value equal to the reaction, i.e.,

r,„., = = 4000 lb.

The centroid of the section is located by the distance

Vc
4(1) +8X4 ,
2.83 in.

The moment of inertia of the cross-section is

= ^*-1- - 12(1.83)^ = 75.4 in.‘

The maximum shear stress occurs at the neutral axis of the cross-section; the
statical moment of ilic shaded area of the stem, with respect to this axis is

Q = 5.17 X 1 X J(5.17) = 13.34 in.^

132 .STRESSES IN BEAMS: I

Suhstitutinj; thes(^ f|uantitios into eq. (5.7), we obtain

Y9 4000 X 13.34
=
r 708 psi.
It 75.4 X 1

example: 3. Culrulate the maxinuim shear stress induced at the neutral axis of
a beam of cir^ailar cross-section (Fig. 5.32} if the total shear force at the section is V.
soLU'rioN. The moment of inertia of the cross-section is

The statical moment about the neutral axis of the shaded .semicircular area is

TT^ 4r
Q
T
With these values, ecj (5.7) becomes

VQ 4V' X 2r» 4 V
(g)
Ib Trr* X 3 X 2r 3 .4

where A ~ xr^ is the area of the circular cross-section.

PROBLEMS
d. Referring to Fig. 5.30, calculate the inaxiinum shear stress induced in the
beam if th(' load P - 10,000 lb is placed one foot to the rigiit of support A. /Ins.
= 187.5 psi.
2. A wood beam simply supported and uniformly loaded has a rectangular cross-
section 4 iu. wide X 6 in. deep. The allowable working stresses in flexure and shear
are <t^ — 1200 psi and = 160 psi. Calculate the span length I below whicli the
shear stress will govern and above which the bending stress will govern the safe
lof^. A ns. ( = 45 in.
3* A simply supported beam of span ( = 10 ft carries a uniform load of intensity
w. The cross-section of the beam is a rectangle 8 in. wide X 12 in. deep. The
maximum flexure stress due to bending is 1200 psi. Calculate the maximum shear
stress. Ans. r^ax = 120 psi.
4, A laminated wood beam is made up of three 2 X 4-in.planks glued together to
form a solid cross-section 4X6 in. as shown in Fig. A. The
allowable shear stress
in the glued joints is Tu; = 50 psi. If the beam is 6 ft long and simply supported at

ir
c v/’^z/A

-z' H
m
Fig. a Fig. B Fig. C
 SHEAR STRESSES IN BENDINC; m
its ends, what is the safe load P that can carried at the middle of the span?
What is the corresfxjndinK maximum flexure stress? Ans, P == 1800 Ib; (Tnux = 1350
psi.

5. A bridge consists of tw^o parallel girders covered wuth cross planks.

sinTiple

Each a redwood log 1 ft in diameter and the clear span / = 8 ft. The
girder is

allow^able stresses in bending and shear arc cr^, = 1000 psi, and = 120 psi.
What is the maximum weight IT of a steam roller, considered as a concentrated
load, that can safely (;ross the bridge* at very slow' speed? Assume that the allow-
able stresses are over and above those existing in the girders due to the weight of the
bridge itself. Ans. \V ~ 14,140 lb.
6, In a building structure, the overhanging end of a large wide-flange girder
supports a vertical load P of such magnitude that the flexure stress at A (Fig. B) is
9000 psi. Calculate the maximum and minimum values of the shear stress in the
web if the beam has a 24WF130 section (see Appendix B). .tns. = 9930 psi;
Tm ^.^
1
7 980 psi
7: A .short cantilever beam (extruded magnesium) has the cross-section shown in
Fig. C. Calculate the shear stress across ah if V ~ r>0.(K)0 lb. yln,s*. r - 2500 psi.

5.6 Stresses in Built-up Beams

Fabricated or built-up beams and girders are frecpiently found in

engineering practice. S(‘veral examples of such iieains are shown in Fig.

5.33. Figure 5.33a represents the cross-section of a simple box beam

liuilt up of four wood planks held together by nails or screws spaced at
intervals along the length of thebeam, ]*^igure 5.33b rcpn;senls the cross-
section of a wood girder formed simply by bolting two rectangular limbers

(o ) { b) {^)

Fig. 5.33

together and Fig. 5.33c represents a typical cross-section of a steel girder

formed by riveting angle sections and cover plates to the edges of a web-
platje.

The stresses in such built-up beams are usually calculated on the as-
sumption that the parts are rigidly connected so that the beam behaves
134 STRESSES IN BEAMS: 1

like one of solid section. The computations will then involve: (a) designing
the beam as a solid beam and (b) designing and spacing the connections
so as to fulfill the requirements of (a). In the first step, the formulas for
solid beams are used, making allowance for the effect of rivet holes, slots,
etc., by using reduced cross-sectional areas. The calculations required
to design Uie connei^ting ekunents can Ix'st )>e discussed in connection with
particular cases.
Let us consider first the case of a box beam having the cross-sect ional
dimensions shown in Fig. ff.lUa. We assume that the beam is simply
supported at its ends with a span / — b ft and that it is to carry a vertical

(q) (b)

load P at mid-span. The allowable working stress m bonding is 1000

psi. It is required to find tiie propi^r spacing of lag scnwvs, each of which
can transmit a sliear force F - 000 II >.

The moment cf inertia of the cross-section with respect to its neutral

axis zz is

4 X (8F 8 X (IF
+8X (4.5F == 495 in.

from which F = 5500 lb. Thus the shear force at each cross-section of the
beam becomes

K = = 2750 lb.
^
 STRESSES IN BEAMS ( BUIT/r-tJ P '

135

The? shear .stress across the junctions ah ami rd of the section, from e(). (5.7),
is

VQ 2750 X ‘M\ .

The coriTspondinji; horizontal shear i(>rce (o be transmit b^d bv e^acti screw is

/'
- ]{\rT)
from which
F (U)0
() in,
'
2r X 50

Built-up T-beams are more fre([uently used in practice than built-up

wood beams. Such a beam (Fig. 5 M5) is first designed as one of solid
section, making suitable deductions in cros>- sectional area to allow for the
weakening effect of the rivet holes, and using a suitable working stress a-„,.
The size and spacing of rivets will then Ik* selected so as to develop the
nominal flexure strength of the beam. To illustrate, k‘t us consider the
required spacing of the ri^'ets whidi (*onnect the flange angles to the W(d).

(q)

Fig. 5.35

Referring to Fig. 5.35, let e denote the .spacing of these rivets and consider
corresponding cross-sections mn and m\ni at which t he bending moments arc
M and M + AAf ,
respectively. Due to the difference Ail/ in these bending
moments, the normal stresses as on the sections mn and mini will be un-
balanced and there is a tendency for thi^ flange represented by the shaded
portion of the cross-section in Fig, 5. 35a, to slide with respect to the web.
This sliding is prevented and by the shearing strength of the
V)y friction
rivet. Neglecting friction, the force on the rivet becomes equal to the differ-

ence in normal stress resultants in the flange at sections mn and mini.

This' difference is

AM ydA, (a)
I
136 STRESSES IN BEAMS: I

where jy dA is the wStatical moment of the shaded area of the flange about

the neutral axis of the section. Now using the relationship dM/dx = F, eq.
(5.1), and replacing dM by AAf and dx by e, we have
Ail/ = Fe. (b)

Substituting this in eq. (a) and also denoting

j y dA by Qp, we obtain

F = (5.10)

This represents the total shear force that must be carried by each rivet.
Observing that each rivet is in double shear and assuming an allowable
average shear stress for the rivet, the force F is readily computed for any
given rivet size. The corresponding spacing e is then calculated from
eq. (5.10).

EXAMPLE 1. A box beam is made up of two 10Li20 channel sections connected

tc» 16-in, X i-in. side plates with }-in.-diameter rivets spaced at 5 in, as shown in
Fig, 5.36. and painting, acce.ss holes are cut in the side
Also, to allow for riveting
plates as shown. The beam has a span Z = 20 ft and is to carry a concentrated load
P at the middle. To allow for the weakening effect of rivet holes, the working stress
for bending is chosen as 0"^ = 16,000 psi. Calculate the safe load P, the average
shear stress r in the side plates along be between access holes, and the shear stress
in the rivets

Fiu. 5.36

SOLUTION, Referring to Table B.4 of Appendix B, we find that the smaller

moment of inertia of one channel about its own centroidal axis is 2.8 in.^ and fhe
cross-sectional area of one channel is 5.86 in.^ Then the moment of inertia of the
entire cross-section, ignoring rivet holes, is

L - 2(2.8 -f- 5.86 (7.39)*] -h = 944.4 iii.^

 STRESSES IN BEAMS (BUILT-UP) 137

Now using eq. (5.5a), and noting that the maximum bending moment is M « Pi/4,
we obtain
Me P X 240 X 8
=
T 4 X 944.4
16 (XX)
, psi,

from which the safe load P == 31,480 lb. The corresponding shear force at any
cross-section in 7 = JP = 15,740 lb.
To calculate the average shear stress in the side plates between h and c, we must
first calculate the statical moment Q of half the cross-sectional area in I^g. 5.36a,
with respect to the neutral axis of the cross-section. This becomes

Q = 5.86 X 7.39 H- 1 X 4 X 6 - 67.3 in.*

The total shear force P to be carried between cross-sections through a and d can
now be found from eq. (5.10) by taking e — 15 in., the distance bf'tween centers of
access holes. Thus

15,740 X 15
P = X 67.3 = 16,800 lb.
944.4

The web area available to transmit this force is 1 X 3 -- 3 in.® Hence the average
shear stress in the side plates along he is

P 16.800
r iiv 5600 psi.
A 3

To calculate the shear stress in each rivet, we must first hnd the statical moment
of one channel section about the neutral axis of the crosR-section. This becomes

Qr - 5.86 X 7.39 - 43.3 in.®

Then, again using eq. (5.10) and taking e -= 5 in., we find

1 5,740 X 5
P - X 43.3 - 3610 lb.
944.4

There are two rivets each of cross-sectional area A ~ 0.442 in.'*' to transmit this
force. Hence the average shear stress in one rivet bc^comes

3610
4090 psi.
2 X 0.442

PROBLEMS
1. A built-up beam having the cross-seetjon shown in Fig. A consists of two

standard 41-7.7 steel beams connected by f-in.-diamett'r rivets spaced on 4-in.

centers along the length of the beam. The l)eAm is 5 ft long, simply supported,
and uniformly loaded so as to develop a maxinmiA f>ending stress of 16,0(X) psi.
Calculate the average shear stress induced in the rivets nearest the ends of the
beam. Ans, r = 5350 psi.
2. A cantilever beam is composed of two 6 x 6-in. timbcjrs held together by bolts
and connector rings as shown in Fig. B 'i’he bait holes are J in. in diameter and
each connector ring can safely transmit a force of 6000 lb in shear. If the load
P = 5000 lb, what is the required spacing of the bolts? Ans. e - 9.6 in.
138 STRESSES IN BEAMS: I

p
U - -- 3 I/J— — 6
"

rp —
j rp c-i r— ~
"“T'
“
‘p- 12
1 1 «
'
mT^
i

L
1

-gu til tij tb d^r [4^

! e- *1

Fig. a Fig. Fig. (

3. C’aF'ulato the inaxiTiiuiii bending stn‘ss<r in the beam o( th(' pna-edirig j)rf)bj(‘m,
assuming that the first 2 in. from the wall. 1 ns. a — 1590 psi.
bolt on the hdt i.s

I. \ box beam like that slu’wn in Fig 5.33a is made of four (i X 1-in. uootl
planks (‘onnect«‘d by sen'vvs, f‘aeh of which can sah'ly transmit a shikar force of
250 lb. Cailculati' tlie mmimnm s par mg of screws along the length of the b(*am if
tlie maximum shixir force 1" =- 1000 lb. .<!/?.<? e = 4.3S in.
5. A built-up gin’tT is made bv rnading four 0 X 4 X ^-m. aiigh's to the (“dges
oi a 10 X 1-in. jilati using 'in.-diam(*ter rivets, as slumn in Fig. C\
,
1
riu* girdi'r
i.s 20 ft long, simply support(‘d at th(‘ nds, and carries a lamcentrated load I* at
<

the middle. Falcailati' tln’ rivet spacing e to di'velop tfie full fie.vural strength of
the girder if (r„. ~ 10,000 psi and the aliowabk; average .sfiear stress in tiu’ rivi'ts
is Tu, = 00(X) psi. [ns. V S.5 in.
 CHAPTER VI

STRESSES IN BEAMS: 11

6.1 Plastic Bending of Beams

The preceding theory bending ha> hevn (h‘veloped on the 'assumptions
of
that plane a beam befori' Inuiding remain plane after
(‘ross-se(‘tions of
bending and that th(‘ material follows Hooke’s law. These assumptions
lead to the condition of a linear distribution of stri'sses ovtT the depth of the
beam. This theory applies with good aeiairacy for many materials, so long
as the def(»rmations remain withm the elastii* limit. However, for the
bending of a steel beam btres.sed beyond thi* yield point, it is no longer
permissible to assume proportionality between stress and strain m the
longitudinal fibers. Sueh bending of a steel beam beyond the elastic range
of strain is eallod plastic beading.
To develop a theory of plastie bending
of steel beams, we begin with a considera-
tion of the stress-strain diagram for
structural steel as shown in Fig. 2.8a,
p. TF Since the elastic limit stress and the
yield stress are close together and since
the plastie strain during yielding may be
many times greater than the clastic strain
customary to idealize^
before yielding, it is

diagram by three straight

this stress-strain
lines as shown in Fig. 6.1. This diagram
assumes that proportionality between
stress and strain holds up to the yield stress ry and that for any strain
j,

beyond this, the stress remains constant and^ equal to ay.p It is also .

assumed that the mateiral has the same yield point in tension and
compression.
Now let us consider a prismatic beam of arbitrary symmetrical cross-
section subjected to pure bending in the plane of symmetry as .shown in
Fig. 6.2. If the bending moment M
is not too large, the beam will be elastic,

the distribution of bending stresses over the depth of the beam will be linear
139
140 STRESSES IN BEAMS; II

(Fig, 6.2a), and the maximum stress in the most remote fiber from the
neutral axis will be less than the yield stress. This represents the case of
elastic bending as discussed in the preceding chapter.
If the bending of the beam is increased beyond the elastic range, experi-
ments show that plane cross-sections before bending continue to remain
plane after bending. Thus we always have the strain in each fiber propor-
tional to its distancefrom the neutral axis. Some of the fibers more remote
from the neutral axis will acquire strains beyond the elastic limit. However,
we see from the stress-strain diagram, Fig. 6.1, that the stresses in these
overstrained fibers remain constant and equal to the yield stress <Ty p.. This
condition is represented in Fig. 6.2b. The central unshaded portion of the
beam is still elastic while the shaded outer regions have become plastic.

<r < ^yp

A«ui rol Axis.

I
5
a‘-<iTyp -

(a) (c)

Fig. 6.2

With further increase in bending, more and more of the inner fibers reach
the yield condition until finally the entire beam, with the exception of a very
thin layer at the neutral axis, becomes plastic. This condition is represented
in Fig. 6.2c. The corresponding bending moment is called the plastic
moment for the section and is denoted by Mp. Neglecting strain hardening
in the outer fibers, no further increase in the bending moment can be at-
tained. Thus the plastic moment Mp represents the limiting strength of the
beam in bending.

To evaluate the plastic moment Mp for a given cross-section, Fig. 6.3a, we

assume that in the fully plastic condition every fiber of the beam is stressed
 PLASTIC BENDING OF BEAMS 141

to the yield point ay p., those above the neutral axis being in compression
and those below the neutral axis being in tension, as shown in Fig. 6.3b.
Then since the sum of the internal forces over tlie entire cross-section must
reduce to a couple equal to the bending moment Mp, we conclude that

a.vp A, - ay,p Aj -0, (a)

and
(Ty p All/, + (Ty p A22/2 = (b)

where Ai, A2 are the areas below and above the neutral axis, respectively,
and 2/1 and 2/2 are the distances to their centroids. For locating the neutral
axis, we obtain from eq. (a)

( 6 . 1)

where A = Ai -|- A2 is the total area of the cross-section. Thus the neutral
axis divides the total area of the cross-section into equal parts. be It will
noted that the location of the neutral axis for plastic bending is, in general,
different from that for elastic bending (see p. 1 14). Replacing A 1 and A2 by
A/2 in eq. (b), we obtain
A
^ (yi 4 Hz). (6.2)

For the particular case' of a rectangular cross-section, A = 6/i, = 2/2

=
/i/4and eq. (0.2) becomes

The maximum elastic moment for the same section is

Thus, for the rectangular cross-section, we have the ratio

My ^ 3
M, 2

and conclude that the plastic moment is 50 per cent greater than the maxi-
mum elastic moment for the same rectangular section.
For a circular cross-section of radius r, A = rr*, = 4r/3ir and
eq. (6 A) becomes
142 STRESSES IN BEAMS: II

The maximum elastic moment for the same section is

Z = (f)

Thus, in this case, the ratio

M,
from which the plastic moment is s(‘en to he 70 per cent |z;reater than the
maximum elastic moment.
If M'o write eq. (ti.2) in the form

P (g)

^U//1 -h

we S(^e then the quantity + ^2 ) may he consirlered as a s(H*tion

modulus for th(^ case of plastic hendina- Denoting this ipiantity hy Z,„
we have

^/» ~ (//i + Jh)’ (O.S)

Applyitif^ this formula to the particular case of a [‘iWi oO wide flange

section, we have, from Tiihle H.2 of Appendix H, A ~ 14.71 in.*’, aiid

?/i
== ih
8.08 X 0.641 X 5.78 + 5.46 X 0.:i71 X 2.78
= ,
1-82
.

in.

Thus
= i X 14.71 X 9.64 - 70.9 in.\

while the (dastic section modulus is Z = 64 7 in.’’ Eor such a section the
ratio

Me Z G4.7

and the plastic moment is only JO per cent greater than the maximum
moment.
elastic
In the practical design of beams, the allow^able loads are sometimes
selected on the basis of the plastic moment that can be dev'eloped by the
seciion rather than the maximum elastic moment. This procedure,
analogous to that of Art. 2.3, often results in a more efficient use of the
material and is called limit design. To illustrate, let us consider the case
of a simply supported l>eam loaded at the middle as shown in Fig. 6 4.
 PLASTIC BENDING OF BEAMS 143

If the load F is not too large, the

entire beam will bo elastio and wo
have the deflertion cuiwe AFB a.s

shown in Vig ().4a Increasing the

load gradually, a value will sor.n b<‘

reached for which th(‘ maximum

extrerm^ fiber stress at (
’ will be e(|-ial

to although the beam is

(Tv ,). still

completely elastic. This load, d(‘-

noted by Pe, represents the elastic

limit load for the beam. As P is

increased beyond this valiuu tlu'

('
fibers in tin* neighborhood of point
anaximum bending moment) begin
to yield and this yielding will progn^ss until, when tlu' bending moment at C
reaches the value 3/;., a pUisiir fiin<}( will beformi'd at f\ Most of the beam
will "till be elasti(‘ but with the phi'-tic hinge at C, it continues to deflect
without further increase in th(' load, Th<‘ value of P
which produces this
condition is called the limit load and is denoted by Pj 'To calculate this
limit load, wo hav(‘

"2 -
I
from which Pi = 4Mp/L A safe working load taken as P,„ - Pe/h
ill have a faetor of safely n based on tin* elasiie limit load A safe working
load taken as 7A- - P/./^h will have a factor of safety n\ based on the
plastic limit load,l^sing this second procedure for .selecting the allowabU*
working load represents the idea of limit design.

F.x>^MPLE 1. A stepjied steel shaft of circular <Toss-sectiou is built-in at C and

louded at th(‘ free end A as shown in Fig. 0 ( aiculatr' th(‘ ratio of dianaUers

d, d-: in*order that pluv^tie hirig(‘ conditions will devoioi) siinultaneously at

’
and C.

A 1^2
1
2

1 -
5 r
; t

'

i
^ 2 ^
;

Fig. (1.5

soLGTiciN. 'rhe bf'nding moment at B is —Pl/2 while that at C is —Pl\ thus

they an; in thf' ratio I 2, and the plastic sedion moduli at B and C must he in
:

the same ratio. From this r(‘(juirement, we have (see eq. e)

144 STRESSES IN BEAMS: II

Fig. 6.6

^ „
^ ” ^
1

2
,

from which di/d^ - 1/^2 - 0.794.

EXAMPLE 2. Calculate the plastic moment J/^. for the stool T-section shown in
Fig. 6.6 if the yield stress 40,000 psi.
<7-y
p
—
SOLUTION. For fully developed plastic bending, the neutral axis divides the total
area into two equal parts. Hence in this case the neutral axis coincides with NN
the junction between sti'm and flange as shown. Then from cq. (6.2), we hav('

/4fp = 40,000 X 3(1.5 -I- 0.5) - 240,000 in.-lb.

PROP LEMS

J. Locate the neutral axis NN for fully plastic bejiding of a steel beam of triangular
cross-section as shown A. Am. hi — 0.707 h.
in Fig.
2. Calculate ihc plastic section modulus for the triangular cross-scction in Fig. A.
A ns. Zp = 0,0977 bh*.
3. A simply supported beam has a span / ^ 5 ft and a 3 X 6-in. rectangular
steel
cross-section. If the uniformly loaded and ctj ^ - 40,000 psi, what is the
beam is

maximum intensity of the load that the beam can support? Am. w ~ 28,800 Ib/ft.

Fig. B

4. A has a square cross-section with two semicircular grooves cut in

steel beahi
its sides as shown Calculate the plastic moment Mp that this beam can
in Fig. B.
develop if (r, p. = 36,000 psi. Ans, Mp = 740,000 in.-lb.
5. A cantilever beam consists of a steel rod of diameter d that has a centered
bore of diamett:/- di for one-third its length as shown in Fig. C. What should be
the diameter di of the bore for plastic hinges to form simultaneously at A and C
under the action of a load P at the free end? Ans. di « 0.87 ^d.
 BEAMS OF TWO MATERIALS 145

6.2 Beams of Two Materials

Beams made of two or more encoun-

different materials are frequently
tered in engineering practice. Common
examples are wood beams rein-
forced by steel plates and concrete beams reinforced by imbedded steel
rods. The theory of bending of such beams within the elastic range of the
materials is quite simple. For a l>eam of two materials, the procedure is
to transform the composite beam into an equivalent beam of one material.

Fig, 6.7

Consider, for example, the case* of a wood l>\'im of rectangular cross-

section reiniorced on the bottom by shown in Fig. 6.7.
a steel plate as
Assuming that no slip takes place between the wood and steel during
bending, the theory of solid l)eams will still apply. That is, plane cross-

sections before bending remain plane after bending and the strains in the
longitudinal fibers are proportional to their distances from the neutral
axis. Thus for any bending cur\atiire 1/p within the elastic range of the
material, the normal fiber stress in the wood, at the distance y from the
neutral axis, is

<r = E^y/p

and the corresponding element of normal force on the area dA = bdy is

dy, (a)
P

where is the modulus of elasticit}^ of wood. Similarly, for an elemental

area of the steel,

dF, ^ — by dy, ^'b)

p

where E, is the modulus of elasticity of steel. In the case of pure bending

these elemental forces, summed over the total areas of wood and steel,
respectively, must have a net resultant force equal to zero, and the sum of
their moments about the neutral axis must be equal to the resisting moment
14 (> STRESSES IN BEAMS: II

cli'voloped by the sertioii. Without actually making these summations,

we note that the results will be iiiifhanged if we write eq. (a) in the equiv-

alent form

dF„ = hj y dy. (a')

This shows that we may regard the wood portion (4 the eross-seetion of
\\idth b as equivalent to a steel stem of redueeti width

as shown in J-'ig (l.71>. Under a givcui Icmding, iIk* composite section in

Fig. () 7a and the trufisformnl scctutn in Fig t).7b will both develop the
same resisiing moment
As soon as we hav(' the traiL'^forined section, the problem of the bending
of the beam of two miUerials is nnluced to that of the bending of a steel
beam of 7\s(M*tion which can lu‘ treated on the basi^ of tht‘ theory previ(nisly
developed in Art for a given Ixmding moment M on the section,
the extreme tcmsih' stress at. the bottom will bo

Mci

and at the top th( compressive stress will be

(T2 — Mo,
~j~’

where I is the moment of mmaia of the transformed section, Fig. 6.7b.

These bending stresses are, of course, for the equivalent steel section
To obtain the true maximum compressive stress in the wood, the stress
(Tj must be reduced in the ratio £'«/£'«.
In the above diseiission, the less stiff material (wood) was transformed

Fro. 6.8 Fig. 6.9

 BEAMS OF TWO MATERIALS 147

into an equivalent width bi of the stiffer material (steel). The reverse

procedure will obviously be equally satisfactory.
If the compo.«ite beam consists of more that two materials, the trans-

formed section may, in general, be obtained in terms of any one of them.

For example, Fig. (i.Sa represents the rectangular eros.s-seetion of a lam-
inated beam of three materials having moduli of elasticity Ei > K2 > E^.
Then a suitable transformed section of the stiffest material will be that
shown in Fig. G.8b, where 61 = b, In = E^b^'Ex, and Ih = Ezb/Ex.
Jf the composite beam does not have a rectangular cross-section, the
problem of finding an equivalent, transformed se(‘.tion may be more com-
plicated. Consider, for example, a beam of two materials having a circular
cross-section as shown in lug. ().9a and as.sume Ey > Then in the
transformed section, Fig. G.Ob, each elemental strip of width b in the lower
half of the cross-section must I)e reduced to width by = (E2/Ei)b. This
will be accomplished by making the lower half of the transformed section
elliptical as shown.

r.XAMPLE A timber beam having an 8 X 12-in. rectangular cross-section is

1.

reinforced topand bottom by steel plates 8 in. wide X i in. thi(’k as shown in Fig.
6.10a. The moduli of elasticity are E, = 30(10)* psi and E,o ^ 1.5(10)* psi. The
allowable working stresses are a, = 16, OCX) psi and cr^. = 1200 psi. Find the maxi-
mum allowable bonding moment for the section.
SOLUTION. The ratioF, = ; 1 : 20. Hence the transformed section will be
an I-section with w^eb thickness by — 5/20 ~ 0.4 in, as showm in Fig. 6.10b.

Fig. 6.10 Fig. 6.11

To determine the safe bending moment, wc must first decide which allowable
working stress wdll govern. The allowable stress in the web of the equivalent
beam will be (t', = 20 X 1200 = 24,000 psi. The allowable stress in the flanges is
a, = 16,000 psi, hence the flange stress governs.
The moment of inertia of the transformed section is
0.4(6)*'
7 + 4(6.25)* + 3
370 in.^
148 STRESSES IN BEAMS: II

Then, from eq. (5.4)

= - 0-.
c
^X
6.5
16.000 = 911,000 in.-lb.

EXAMPLE 2. The cross-section of a small beam cut from a sheet of 7-ply ply-
wood has the dimensions shown in Fij^. 6.11a. Alternate layers of the plywood
have the grain parallel to the length of the beam. The beam is 4 ft long, simply
supported, and loaded at the middle. The modulus of elasticity parallel to the
grain is Fj = 1.6(10)* pel, that perpendicular to the grain is Fj = 0.4(10)® psi.
The corresponding working stresses are ~ 1200 psi and cr? = 300 psi. Calculate
the safe value of the load P.
SOLUTION. The ratio of the smaller modulus to the larger is F? = 1 4. :

Hence the width of layers having the grain across the axis of the beam should be
reduced in the ratio I 4 and the transformed section will be as shown in Fig. 6.11b.
:

The moment of inertia of this transformed section is

_ 1(0.875 )^ 07J)^.125V
^
^
~ 2[0.09375(0.25)^J - 0.0439 in.^
12 12

The corresponding bending moment for th(t section is

/ 0 0430
X 12(X) - 120 in.-lb.
c 0.437

and the safe load P at the middle of the beam is

4d/,„ax 480
........ .
U) lb

If a laminated beam of two materials has many layers, it will be justifiable to

express the moment f»f inertia of the transformed section b}’' the approximate
formula
hh^
"] 2~2 iT"'
where h) = is the reduced width of tJie softer layers. v^iiLstjtuf,i;ig this
value of b\ into eq. (d), we obtnm

1 -

PROBLEMS
1. A simply supporUxi beam 10 ft long and carrying a concentrated load F at the

middle has the cross-section shown in Fig. 6.7a. Taking P - 2000 ib, EJE, =
1/20, b
- 4 m., h = 6 in., and t - 1/2 in., calculate the maximum tensile stress in
the sU'cl and tht' maximum compressive stress in the wood. .4ns. a, ~ 7640 psi;
= 1300 psi.
2. A wood l>eam of rectangular cross-section 6 in. wide by 8 in. deep is reinforced
at the top by a steel plate 2 in. wide and I in. thick and at the bc'ttom by a steel
plate 6 in. wide and 1/2 in. thick. The moduli of elasticity are F, = 30(10)® psi
and Ew = LfiClO)® psi and the allowable stress in the steel is tr, = 16,000 psi.
 BEAMS OF TWO MATERIALS 149

Calculate the safe bending moment and the corresponding maximum

for the section
fiber stress cr„ in the wood. Arts, M =
308,000 in.-lb; (r«, « 655 psi.
3. A
laminated wood beam has a rectangular cross-section 8 in. wide X 12 in.
deep. Thelayers arc each ^
in. thick, and alternate layers have their grains crossed
at right angles. The ratio E 2 E\ = 1 4 and the corresponding working stresses
: :

are cri = 1200 psi parallel to the gram and erz = 300 psi across the grain. Calculate
the safe bending moment for the section. Am. == 12,000 ft-lbl M
4. A bimetallic strip such as is used in thermostats is made of a strip of copptir
(Er = 17(10)® psi) and a strip of aluminum {F.., - 10(10)® psi), each in. wide 1

and iV thick, solidly fused together along the junction plane as sliown in Fig. A.
Calculate the radius of curvature p due to pure b^'iidmg of the strip if = 10 in.-lb- M
What is the corresponding maximum stress in the copper? Aws. p = 208 in.;
cTr = 4450 psi.

5.

A steel shaft with circular crf)s.s-so(‘tion of (lianu'ter d — 10 in. hnsan alumi-

num core of diameter d/2 ^'=
5 in. as shown m Fig. H. (?al'‘ulat(' the safe bending
moment M if the allow’able stress in the steel is a, ^ 20,000 psi. Atih. 47 ~ 157,000
ft-ib.

6.3 Reinforced Cone re te Beams

As mentioned in the pre^'eding article, onende beams r(‘inforeed with c

steel represent an important example of beams of two difh'rfuit materials

A complete treatment (;f the theory of reinforced eonen l(' is beyond tiie
scope of this book: we givt' here only a brief diMaissioii ot this important,
problem.*
It is well known that the strength of eonerete is mu(‘h greater in eom-
pressioii than in tension, lienee a rectangular beam of eonereto will fail
from the tensile stre.sst's on the convex side. 'Fhe beam can be greatly
strengthened by imbedding steel burs <ni the (‘f)nv(^x side as shown in

Fig. 6.12. Since concrete grips the steed strongly, there will he no sliding
of the steel bars with re.spect to the eonende during bending and tiie

methods developed in the previous artade can' also he used here for cal-
culating bending srre.sse.s. In practice, the ero.ss-seiditinal an^a of the steel
bars is usually such that the tensile strength of the; eoncrele on the convex

*For a more complete t,r(‘atment ot reinforced concrote, see C. \V. Dunham, The Thcon/
and Prartice of Reinforced Corurete. McGraw-Hill Hook Co Inc., New Vork, 1953 ,
150 STRESSES IN BEAMS: II

side is overcome before yielding of the and at larger loads

steel begins,
the steel alone takes practically Hence it is established
all the tension.
practice in calculating bending stresses in reinforced-concrete beams of the
type shown in Fig. 0.12 to assume that all the tension is taken by the steel
and all the compression by the concrete.

(a) (b) (c)

Fig. 6.12

Concrete does not follow Hooke\s law, and a compression test diagram
for this material has a shape similar to that for cast-iron (see Fig. 2.8b).
As the compressive stress increases, the slope of the tangent to the diagram
decreases, that is, the modulus of concrete decreases with increase in stress.
In calculating stresses in reinforced-concrete beams, it is usual practice
to assume that Hooke’s law holds for concrete and to compensate for the
variable modulus by taking a lower value for this modulus than that
obtained from compression tests at small compressive stresses. In specifi-
cations for reinforced concrete it is usually assumed that n, the ratio of
the modulus of elasticity of the steel to that of the concrete, is

n 15.

Transforming the area of the steel A, into an area of concrete nA.,

equivalent as far as elastic properties are concerned, the transformed
cross-section is that shown in Fig. 6.12c. The stress distribution follows
a linear law, vsince plane cross-sections remain plane during bending and
since Hooke’s law^ is assumed to hold for the concrete. Under these
conditions, the neutral axis will lie at the centroid of the shaded cross-
section (Fig. 6.12c). This requires that the first moment of the shaded
area above the neutral axis mm with respect to mm must equal that of
the shaded area below; that is,

led
(bkd)
^ = nA. (d - M).

This equation is quadratic in terms of fc, the value of which defines the
position of the neutral axis.
 REINFORCED CONCRETE BEAMS 151

If the ratio of the area of the steel .4, to that of the area of. the concretf
above the center of the steel be denoted by

the solution of the above equation gives

K = \2pn + (pa)’ — pn. (6.4)

Instead of using the moment of inertia of the transformed section as in

the preceding article, more direct to proceed as follows. The average
it is

stress acting over the area above the neutral axis is (Tc/2 where is the

maximum compressive stress in the concrete. The total compressive force

C is (<rc/2) bkd. For a linear stress distribution in bending, the center
of pressure of C is at the centroid of the triangle which represents the stress
distribution; that is, at z =
With no axial forces acting on the beam,
^kd.
the tension T must equal C and these two forces constitute a c.ouple whose
moment is the moment of resistancV and is ecjual to the prodiu^t of C and
the moment arm jd, where = — k/'A, Hence the bending moment, I

Me based on a maximum stress a,, in the coniTete is

M. = bl.-d(jd) = a^d-. (6.5)

I ^

As the maximum allowable fiber stress in the steel will probably not
be produced by the same value of bending moment as produced Ccj it
is necessary to compute also the bending moment M„ based on a,. The
couple used for this purpose is T(jd) where T = is the force in the

steel. Then
Mg = (JaAgjd. ( 6 6)
.

In determining the safe bending moment that may be applied, the smaller
of the tw^o values Me and A/, should be u.sed.
If the bending moment be given rather than the maximum allowable
fiber stresses, the value of the maximum fiber stresses caused by the
given moment can be computed in the same manner as above and then
examined to see if either or both exceed permissible values.
To protect the steel from damage by fire, the reinforcement in beams,
girders, and columns should not l)e placed nearer the exposed surface
than in. This specification fixes a lower limit for the depth of concrete
below^ the reinforcing bars designated as a in Tig, 6.12b.

EXAMPLE 1. If n = 15 and c, = 650 psi, determine the safe load at the middle
of a reinforced-concrete beam 10 ft long, freely supported at the ends, and having
6 10 in., d = 12 in., A, = 1.17 sq in. What is the stress in the steel? Consider

two cases: (a) if the weight of the concrete is neglected; (h) if the concrete weighs
150 lb per cu ft.
152 STRESSES IN BEAMS; II

SOLUTION, (a) area of the reinforcing rods is nA, = 15(1.17) =

The transformed
17.55 sq in. The neutral axis 1-1 must be
so located that it is a centroidal axis for
the shaded areas shown in Fig. 6.13a. This requires that = 0, i.e., XM
aoy) 17.55 (12 - y),
^
from which y 4.%5 in.
The compressive force C
is the product of the compression area and the average

compressive stress acting thereon; namely, C == (650/2)(10 X 4.965) = 16,100 lb.

Then since T - C, and the moment arm of the couple is (12 — 4.965/3) == 10.35
in., the moment based on a safe stress of 650 psi in the concrete is

J/e - 16|100(10,35) = 166,500 in.-lb.

The maximum bending moment is P//4; equating this to 166,500 in.-lb, one finds
P » 5570 lb.
Since T -- C ^ 16,100 lb, the stress in the steel is P/A = 16,100/1.17 * 13,760
psi.
(b) The concrete weighs 150(10)(14)/144 = 146 lb per ft of length of the 10 X
14-in. cross-section, allowing 2 in. of concretebelow the reinforcing rods. The
maximum bending moment cau.se(l by this uniform load is == [146(10)*/ 8] 12 =

21,900 in.-lb. Hence the moment available fur carrying the concentrated load P is
166,500 - 21,900 = 144,600 in.-lb. Then P//4 = 144,000 and P - 4820 lb.

From the foregoing disciussion it us seen that the bending moment which
produces the maximum permissible stress in the ct>ncrete of a reinforced-
concrete beam tvill not necessarily stress the steel to its safe limit. Or it

may, on the other aand, overstress the steel, and if the bending moment is

reduced so as to lower the steel stress to a safe value, then the concrete will
be understressed. Both of these (conditions are unfavorable from the stand-
point of economy, since the full strength of either the steel or the concrete
is not being utili 2.ed. When a reinforced concrete beam is being designed,
however, it is possible to so proportion the ratio of the steel to the concrete
in the cross-section that the maximum permissible stresses in the concrete
and the steel are realized simultaneously. The beam then is said to have
balancfd reinforcement. In most practical cases, balanced reinforcement can

Fig. 6.13 Fig. 6,14

 REINFORCED CONCRETE BEAMS 163

only be approximated on account of desi^^n conditions, such as size of re-

inforcing rods available, etc.
In ideal balanced reinforcement, the tensile stress in the equivalent
concrete, which replaces the steel and whose area is n times as great as that
of the steel, is to be (rjn at. the same time as the compressive stress in the
outside fiber of the actual concrete is Since these stresses are proportional
to their distances from the neutral axis,

Oe _ g, /n 1 — A: Ct
(67)
d —kd k ruxc

Once k is determined, the remaining procedure is exactly the same as

before and is illustrated in the following numerical example.
EXAMPLE 2. Design a concrete beam »vith balanced reinforcement to sustain a
bending moment of 2,000,000 in. -lb. Width of beam to be 20 in., cr, = 16,000 psi;
cr, - 600 psi; n - 15.

bC»Lr'riON. The' transformed area and the stress distribution an* shown in Fig.
The teasilo stress in thetransformed area of concrete equivalent to the
ongiiiai steel is one-fifteenth of that in the steel since its area lias been multiplied
by 15. The distribution of stress intensities is assunied to be a straight line as shown
m Fig. 6.14b. The neutral axis is located by using eq. (6.7), i.e.,

1
j-J Jij,m
/t
' ^
"
15‘x m
from wdiich
k = 0.360.

The moment arm of the couple TC is

jd = <i - - 0.88(i.
O
The compressive force in the concrete above the neutral axis is (3tX))(0.36fi)(20) ==»

2l60(i. The resisting moment is (2j60d)(0.8Sd) and is equal to the given bending
moment of 2,000,000 in. -lb. Therefoie 1052 sq in., from which d « 32.6 in.
In balanced reinforcement, T = f'\ and hence

- 2160 (32.5)

from which

A„ fsq iU.

This area of steel can Ik .«upphcd by six 1 hitih reinforcing bars.

PftoBL VJM.S

1. For a reiniorced-conoH'u^ Otam, o r3 . d ^ lu in ,

ana the hxtb. oi tiiB
reinforcing rods is } sq in. Asej^me n - 3 5. - 6(K) usi, and o, ^ iS,tKX) pti
IMiai is the maximum bending moment in iu.-Ib ihao t.Ue beam can carry? A>^8.
M « 66,800 in.-ib.
154 STRESSES IN BEAMS: II

2. For a roinforced-concrete beam, 6 = 15 in., d = 28 in., and the total area of

the reinforcing rods is 3.02 sq in. Assume w 12,500 psi. = 20, = 450 psi, (Ts —
What is the maximum resisting moment
76,000 ft-lb. m ft-lb? .Ins.
3. For a reinforet*d-eonerete beam, 6 = 10 in., d = IS in., t(»tal area of rein-
forcing rods Ls i sq in., n = 15. If this beam is subjected to a bending moment of
10,000 ft-lb, what will be the maximum stresses s(‘t up in the concrete and the
steel? .Ins. (Tc = 250 p.si; tr, = 7500 psi.
In a certain beam, the stn‘sses produced in the concrete and steel are to be
4.
determined. 6 = 20 in., d = 36 in., area of reinforcing rods = 7.2 sq in., n = 18
and the applied bending moment is 2,500,000 in. -lb. /Ins. = 11,330 psi; (Tr
— 508
psi.
For a reinforced-concrete beam, 6 = 18 in., d = 27 in., n = 15, ac - 500 psi.
5.
Ifthe beam is to resist a bending moment of 800,000 in. -lb, how much area of steel
reinforcement must be used? What is the stress set up in the stecP /his. 1.593 sq
in. ; 20,450 psi.
6 . Solve Problem 5 if the bending moment is 1,200,000 in.-lb. Ans. 5.14 sq in.;

10,100 psi.
7. A and 8 in. v\ide, is to be made of reinforced con-
cantilever beam, 6 ft long
crete. be designed to carry a total uniformly distributed load of 6000 lb,
It is to
including its o\\n weight. What should be its effective depth at the wall'^ 3 ft from
the w'all? Indicate the position of the steel reinforcement. Assume n = 15, (Tc ~
650 = 16,000 psi. .hi.s. 15.84 in.; 7.92 in.
psi, (j,

A reinforced-concrete b(‘am is to carry a bending moment of 102,400 in.-lb

8.
If d = 1.56, what are 6, d, and the total area of reinforcing steel needed? .Assume

(Ts= 18,000 psi and = 600 psi; = 30 X

(Tc psi and Er = 2 X 10® psi. Ans. 10’’'

d = 12 in. A = 0.533 sq in.;

9. A simply supported beam 15 ft l(^ng earrie.s a renter load of 6500 lb. The
economic percentage of steel is to be used and d = 20 in. Find 6 and /h, assuming
cr, - 16,(X)0 psi, a, - 5(X) and n = 15. Take account of the w<‘ight of the beam,
ji.si,

assuming concrete to weigh 150 lb per cu ft. Allow 2 in. for dejith a below the re-
infureiiig bars. .hi^. 6 = 14.1 in ;
.1, =- 1.41 sq in.
10. In th(‘ beam of round bars and two i-m.srjuan'
Problem 9 thri'c |-in -diainet(‘r
bars an‘ used foi be assumed to be 6.39 in.
rdnforcement. The neutral axis may
below the top. What is the maximum shearing stress set up in th(^ eonereb' and
what IS the maximum bond stress between reinforcement and concrete*'' Ans. 22 5
psi; 32 psi.

6.4 Bending of Beams of Arbitrary Cross-Section

In developing the theory of bending of prismatic beams, x\rt. 5.3, it

was emphasized that the beam was assumed to have an axial plane of
symmetry and that the applied loads were acting in this plane. In general,
if the applied loads do not act in a plane of symmetry, bending of the beam

will not take place m the plane of loading and the theory devek^ped iiii

Art. 5.3 becomes invalid.

Consider, for example, the case of a beam of Z-section subjected to
bending couples M applied in the xy-plsLiie as shown in Fig. 6.15. Let it

be assumed that the beam also bends only in this plane, i.e., that Oz is
 BEAMS OF ARBITRARY CROSS-SECTION 155

the neutral axis of the (Toss-section. Then and comproKSsive stresses

tensile
in the flanges will be uniformly distributed over the widths b, and resultant

internal forces T and C will act at the mid-points of the flanges as shown.
Clearly these ecpial and opposite forces constitute an internal fxmding

moment about the /y-axis which is not balanced bv any corresponding

applied externa! monnait, tlie applied couples M being in the r//-plane.

Thus the assumed deformation (curvatun* otily in tlu* ,r//-pUui(‘) is imain-

sistant with thi*, conditions of (Mpnlibrium, and th(‘ assumption is proved
invalid. Unless external constraint against bending in the ./T-plane is

provided, we must c()nclude that tiending cooplCvN in the .n/-p\ane produce

also some bending in the .rz-plane.

VVe now proccinl to develop a more genin’al llie(>rv of l^ending of pris-

matic beams of arbitrary rroas-turtion. In iMg. (>.10, let such a beam be

subjected to pure bending and assume that plane cross-sections before
bending remain plane aft(‘r bending. 'Fhen for the conditions shown, all

longitudinal fibers above a certain line nn in the cross-soUion will be in

compression, and all fibers below this line will be in tension. The line
nn represtuits the neutral axis of the section. Through any point 0 on
this line as origin, we take coordinate axes x, //, z, such that Ox is parallel

to the axis of the beam and the plane xy is the plane in which the external
couples Mz are acting. We
have already observed that such couples will,
in general, produce bending of the Ix^am both in the xy-plano of loading and

Fig. 6.16
150 STRESSES IN BEAMS: II

in the perpendieiilar .i:^-plane. Let 1/p^ denote the (’urv'ature in the xy~
plane and l/p^, that in the r^-plane. Then plane cross-sections before
bendini^ remainiiip: plane after bending, the strain in any longitudinal
fiber, defined \)y coordinates //, 2 ,
will be

For elastic behavior, stress is proportional to strain so that

Ey ,
Ez
fj 1
(a)
Pz pu

This defines the stress distribution over the cross-section and since the
stress resultant must reduce to a couple M about the 2 -axis, we have

adA = 0, an dA = M., az dA = 0. (b)

j
Substituting the value of a from cq. (a) into these (‘(luilibrmm ecpiations,
we obtain.

—
Pz
Ajjc H —Pu
Azc 0, I

M.,
pz Py

Pz Pu

wh(^re

h 1
zhiA y
ly

are moments of inertia and product of inertia of the cross-section and

//c, Zcy are the coordinates of the centroid of its area A.
The first of these three ecjuationsby taking yc = Zc ~-= is satisfied 0\
hence we conclude that the neutral axis passes through the centroid of
the cross-section. From the last two equations, we obtain

1 _ MJy 1 MJy,
6 8)
P. Eilyl, - V)’ Py
“ E{Iy.^ ~ lyL)
( .

Tliese expressions define the curvatures in the sy and xz planes, respectively,

due to Ix^nding couples applied in the x^-plane as shown in Fig. 6.16 b.
We note tint when xy is a plane of symmetry, lyz = 0 and l/py = 0,
while 1/pa = Mz/EIj which coincides wdth eq. (5.4) of Art. 5.3.
 BEAMS OF AHBITKARV CROSS-SECTION

Substituting expressions (b.S) tor eurvatures into e((. (a), w't* obtain

^ yjAhn -z 0)
11 , -
((>.

wliich gives the bending stress ni any fiber the hu'ation of \vhi(*h is dehnod
by the coordinates ?y, 2 . Agaiti if xy i< a plane of symmidry, ==0 and
ec]. (b.9) reduces to a = M .y /,, which coincides with eip (.Vo) of Art. 5.3.
Setting or = 0 from e(i. (b.9), we obtain, tor th(' ecjuation of lh(‘ neutral
axis,

,
I,y - --=
0
or

~ tan
lu

Idiiis the neutral axis of the cross-section maki^s tlu* angle d — tan ’

(/,/,//„) with the z-axi> as shown in Tig b.lba.

I'or any (‘ross-scauion, it can be shown that there are ah\ ays two ori hog-
onal centroidai axc's in its })lane for \\hi(‘h th(‘ product of inertia J — 0.

These are called the principal axrs of the cross-se(‘tion (see Aptiendix B).
The corresponding axial planes of ihe beam are called the principal plain's
of bending. From the foregoing discussion, we conclude that for bending
moments applied in such a primapal plane, only bending in this plane will
take place and the usual theory of bending is valid. Thus tor a liearu
subjected to Viending moment that is not m a principal [ilaiie, we may
always resolve this bending moment into components (Coinciding with the
two principal planes of the be.im. Then, by .superp(jsition, thr' total bending
stress in, any liber will be obtained by adding algebraically the two stn^s.ses

produced separately by these components. This procedure is illustrated in

the first of the examples following.
Ail of the above discussion holds rigorously only for pure bonding. If a
beam is bent by transverse loads, there will be sh(‘aring stressc's and defor-
mations as well as bending stresses. In the case 01 b(*ams of solid section,
these shearing stresses do not greatly elTect the bending action and it is
satisfactory to calculate the liending stresses as above without considora*
t ion of the shear. In the case of beams of thin-walled profile section, such as
the Z-section shown in Fig. 0.15, the slu^aring str^^sses are of especial inter-
est, and this question will be considered further in Art. 0.5.

uxAMi’LK 1. A
sin.ply supported wood beam (roof purlin) of rectangular cios.s-
section ftirrk'a a uniform load of inteasitv to as shown in Fig. 6.17. The plane ot
symmetry x02 of the beam is inclined to the \'ertieal .^//-plane of loading by an
angle cY as shown. Calculate the maximum bending stress a-,,.,.* »f / ^ 10 ft, w = 2(X)

lb per ft, ft == 6 in., ft — 8 in., and tan a = J.

15S STRESSES IX BEAMS: 11

SOLUTION. TIk' maximum bonding inomont at the middle of the span is

_ ivr^ 200 X 120^

" ~ ~ 80,000 in.-lb.
8 12 X8
Tlio comj)OTionts of this bonding moment m
th(‘ two planes of symmetry of the beam
arc Ml - eos a = 30,000 X 0.910 - 28,500 in.-lb and 2
~ ^V„iax sin a M
= 30,000 X 0.310 = 9500 in.-lb. I'hus the maximum bending stress at A or B is

28,500 X r> 9.500 X 0

Cr.nRX ()13 psi.
0 X S'' 8 X t)-

Kx.\MPLL 2. Lo('iite the muitral axis of the eross-seetion for the beam loaded as
.‘^how ?i in Fig. 0.17.
solution. The finncipal moments of imTtia of the cross-seetion with respect to
axes of svmmetr}' 1 and 2 are

hb^
h - /n - 0. (0
iT'

From A{)p('ndL\ li, ue }la^'e then

,
fi^ I2 f.-h — 2(x + ,
7,2 Sin 2a,

(f)

Liz — 2a -f- 1 12 cos 2a.

Substituting. the \ .a]ues of /j, L>, 1 12 for b

~ 0 in. and h — S m. into eqs. (f), we
find, for a = tan 'Hi^ 18*^20', I., ^ 155.2 in.'‘, and ly^ = 33.0 in.'* Then from
eq. (() 10)

tan i3
= ^ 0.2)64
K)5.2

and — 12° 13^ mea.sured

i3 shoun in Fig. 0.17a.
LWMPLE 3. A (aiitilever beam of Z-seetion (Fig. 6.18; is 100 in. long and is
load(‘d at the free end by a force F - 400 lb which acts in the vertical pl^e of the
web Find the maximum bending stress (Tmax-
SOLUTION From a handbook of Z-seidions, we find for this section: /* = 34.4
In.**,}y ^ 12.9 in.\ ^ 14.4 in.^ The maximum bending moment is (Aff)max =
— PI - —40,000 in.-lb. The maximum bonding stress may occur at either point .1
 BEAMS OF ARBITHARV CBOSS-SECTloN 159

Fig. 6 . 1 H

or point B. For point A, = —3.00 in., z,j = —3.25 in ;

f(»r i>oii't B, jii,
~ —3.00
in., Zf, = -f*0.25 in. Fsin^ all tlu^so data in (‘q. fO.9), wo obtain
-40,000(- 12.9 X 3 -b J4.4 X 3.25)
-13S0 psi.
34.4 X 12.9 - (14.4)-

-40,00()(-12.9
4
X 3 - 14.4 X 0 25)
“}“7150 j)si.
34. "X 12.9 - (14.4)2

44nis the lars;est bendin^i: stn^ss oooiirs at point B. 4'his iruiioato.s that tlio luaitral
axir^ of tlio section has approximately th(‘ po.^ition shov\ri in Fig. O.lSh. Fn actually
locate^ this axis, wr use eq. (6.10) and find

- tan"' (1.1 16) - 4S°0S^

PHOBLKM.S
1. A cast-iron l)t\am of tiiangiilar crctss-section, Fig. A, is to subjia'tcd to })im‘
bending by e(aiples apjilii'd at its ends and acting in the .r,//-piano, so as to induce
compression at d Find tlie sale btmdirig moment if the working stre.sses in timsion
.

and compres.sion are, re.spectively, cr, 5,000 ])si, (Tc = 8000 psi. The* dimensions
of the cross-section are b — 3 in., /i 6 in. Ans, — 22,50<^ in.-lb. M

Fig. a Ida.

2,' A prismatic steel bar of semicirculai cross- t'ction, I'ig. M, is used as a canti
lever beam loaded parallel to the ^-axis at its free end. Hie lengtli I - 72 m.
r — 1 in., and = 18, (XX) psi. Calculate the .‘^ale load I*. Ans. P — 98 2 lb.
100 STRESSES IN BEAMS: II

3. A Wt length of “(iiiartf^r round” of radius r = 1 in., Fig. C, is supported at its

ends and loaded at the middle b> a vertical force P acting parallel to the ^-axis. If
tlie allowable working stress m bending isff„. = ](K)0 psi, what is the safe value of P?

P ~ 0.95 lb.

i. X 0 K 0 X i'in. v^t(‘('l angle section serves as a simply supported beam with an

S-ft span, Fig. D. A c,oncentrat(‘d load P ~
2000 lb acts at mid-span parallel to the
//-a.xis. Calculate tlie maximum tensile and compressh e stresses induced in the
beam 'J'lie moments of inertia of the section are I ^
— 28.2 in.'*, = -It) ti

in.* cr 1
-- 9420 ])si, tcaision; an — 7380 psi, i‘ompressioii.

5. Ilelc'rrmg to the wood beam of rectangular cross-section shown in Fig. 6.17,

let h
- 10 in.. /; ^ 3 in., / ^ 10 ft, a - 30®. Calculate the corresponding safe load
for the ])(\‘un if the working stress = 1015 psi. Am. iv = 100 lb per ft.

6.5 Shcariiigr .Stresses in IJeains of Thin- Walled Profile Section

111 the jneeediiig article we have seen that pure bending couples acting in
a principal plane of a Ix'am produces bending only in that plane. For such
pur(‘ bending, then' are no shearing stresses and the internal stress resultant
on any seel ion is a couple which equilibrates the externally applied couple
at cither end of the beam. When bending of a beam in a principal plane is

produced by transverse loads, there will also bi*. shearing stresses to consider,
as was done in Art. 5.5. In general, the resultant of these shear stresses on
any section will lie a force parallel to the plane of loading but not necessarily
in this plane. Tliis stiuaiion causes some twdsting action on the beam so

that cross-sections rotate about the longitudinal axis during l)ending. To

attain fiimple bending, i.e., bending without twist, it is necessary to apply
the external loads in the same axial plane as that in which the shear stress
resultants' act. The determination of the location of this plane will, of
counse, require a careful examination of the distribution of shear stresses
over the cross-section.
Let us l)vgin with the case of a beam of singly symmetric cross-section
loaded in a plane perpendicular to the plane of symmetry, as shown in
 BEAMS OF THIN-WALLED PROFILE SECTION 161

Fig. ().19a, and assume that the external load P is applied at, such distance c
from the principal plane xy that simple bending without torsion occurs.
The neutral axis of the cross-section then coincides with the z-axis and the
normal stress a at any point in the cross-.section is proportional to the
distance y from the neutral axis. From the discussion of Art. 5.5, we know
that the shear stresses r will be distributed according to a par:;.bolic law

Fig. 6.19

over the depths hi and h <2 of the flanges and that the shear stress in the thin
horizontal web will be negligible. Thus practi(‘,ally all of the shear w'ill be
(“arried by the two flanges. If w(‘ (‘onsider the twT) flanges as separate beams
having moments of inertia I\ and then their curvatures in bending will be
equal if the load P is distributed between them in the ratio and the
shearing forces V\ and Vi wdll then be in the same ratio. This condition wall
be satisfied if the transverse load P acts in the v(Tti(*al plane such that

Thus, to obtain simple bending, the load P must be applied in a plane that
liesbetween the centroid C and the stiffer flange. In the limiting ease, Fig.
6.19b, where there is only one flange, w^e may take /. = 0, and (‘onchuie
that b\ — 0. Thus, for simple bending of a T-sectioii perpendicular to its
plane of symmetry, the load P must be applied in the middle plane of the
flange as shown. That point 0 in the cross-sec turn, in each casi^, repre-
senting the point of application of the shear stress resultant, is called the
shear center of the cross-section.
In general, the location of the shear
center for a solid cross-section of arbitrary
shape (Fig. 6.20a) is a complicated problem.
Fortunately, it is not so important for
beams of such solid sections because they
have considerable torsional rigidity and
twist very little during bending by loads Fig. 6.20
162 STRESSES IN BEAMS: II

acting through the c(Mnroid C. Hence in our further discussion, we shall

consider Qiily the ciis(^ of an arbitrary thin-walled profile section as shown
in Fig. 6.20b.
We begin with a beam of such cross-section loaded parallel to one of its

two principal planes, say xy, so as toundergo simple bending in this plane,
Fig. 6.21. Then the 2 -axis becomes the neutral axis and the normal stress a
in any longitudinal fiber, the location of which is defined by coordinates
//, 2 ,
is simply
M.y
(T (b)
L
Isolating an element A f)f + dx^ and
the wall between cross-sections x and x
of arc length .s, we
normal stress resultants
see that and Nx + dNx will
act on its transverse edges as showm. The bending moment at x + dx being
larger than that at x, there will be a net force dNx in the positive direction of
the .r-axis, which induces shear stress t along the inner edge of the element.
Since the wall thickness t is assumed small, this shear stress can have no

transverse component and must act parallel to the x-axis. The outer edge
of the element is a free surface and carries no shear stress. Thus for equilib-
rium of the element in the x direction, we must have
T t dx = dNx* (c)

Now using eq. (b), we conclude that

 BEAMS OF THIN-WALLED PROFILE SECTION 163

This cxproasioii, analogous to o(j (5 0), givt's t}i(' shoar stu'^s at any point
in the wall distanee .s from the free edge. Since ihe integral thi^rein is a

Innetion of .s\ >ve eonclude that this horizontal sl)(‘ar stress between adjacent
fibers varies w ith tlie distance s-, Ixanga maximinn at I Ik' lamlral plane and
zero at ea<‘h free edg(‘,
hroni the requirement for equality of complementary sla^ar stresses (s(?e

p. 30), we may now- conclude that there must b(‘ tlu^ same shear stress
distribution in lh(‘ plaiu^ of the cross-section as shown in Fig. 0.2 lb. Again,
since the wall thickn(\ss t is assum(‘d small, these shear stresses in tluj plane
of the cross-section must act in the tangential dire(rtion at each point. The
shearing strt'ss per unit length of the center line AB of the section is

and this is sometimes called the shear flow.

Considering now' the equilibrium of that portion of the beam between any
(Toss-section :r and the free end, it is seen that the elemental shear forces
rids over any cross-section must reduce to a force Vx parallel to the i/-axis.

This shear force wdll act through a point 0 such that its moment about the
centroid C is equal to the sum of moments of the elemental forces about this
same point. This ivxitiirement will enable us to locate the shear center 0 for
any specific cross-section. The beam undergoes simple bending without
twist only if the load P at the free end acts through the shear center of th(*
end cross-section.
 STKK8SES IN BEAMS' JI

LXAMPLK I. A a thin-wallcd chanm^l 8oction as shown in Fig. 6.22a

l)e.‘un ha\'in^^

is loaded in a the web so as t<j produce simple bending in

v^erticnl plane' parallel to

this ])lane. Find the distance r defining the location of the shear center 0 of the
section.

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

Fio. 6.22

soLU'i roN. Since the* wall thickness t is small, the shear stresses in the flanges will
1 h* horizontal, and those in the web
shown Fig. 6.22b. The resultant
vertical, as m
shear force for the wt'b alom* will lie a vertical force coiniuding with the middle line
(d the web, and its magnitude must be equal to the total shear force I" on the entire
section. This is shown in Fig 6.22e. To find the n'sultant slu'ar force for eith'‘,r

flangi', \\T min t first find the intensity of shear stress at any point .<? of a flange.
Jieferring to Fig. 6.22b and using eip (6.11) this beeomes

Then the n'sultano shear force Q for (uther flange bi'cumes

(e)

TIk'sc horizontal slK'ar sfiess n^sulUints for the flanges act along their middle lines
as shown in Fig. f> 22c, and constitute a counter-clockwise coupli' of moimmt Qh.
This couple, togetlu'r with the vi'rtical shear fore(‘ V coiueiding with the web. are
statii’allyequi\ak‘nt to a vertical force I acting through a point 0 in the plane of
tlu‘ cross-section as .shown in Fig, 6.22d. Thi.s ])oint 0 is the reipiirial shear center

and its distance c ff-om the middle line of the web is found from thi' ndation
IV - Qh,
from which
^Qh ^ hViH
'

~ (0
V U
The beam simple bending in the principal plane piarallel to the web
will und(‘rgo
only if is removed from the plane of the web by this distance e.
the plane of loading
.As a specitii* example, consider tlie case of a standard 10L120 channel section.
Then from Table B.4 of Appendix B, we find h ~ 9.56 in., b — 2.74 in., t = 0.436
in., I = 78.5 in.'* Substituting these data into eq. (f),

(2.74) ^ (9. 56) ^(0. 436)

= 0.952 in.
4 X 78.5
 BEAMS OF THIN-WAEBE/) PROFILE SECTION J65

EXAMPLE A
boain hax'in^ the thin-wallo*! sfmiciroular cross-section shown in
2.

Fig. 6.23 is loadofi in a })rinc

i)X‘ii plane .?•// su as to ^jn^duce simph' bending in this

plane. Find the distance e d('finiiig th(* location of the shear center 0.
SOLUTION. The shear stress r at c^ach point = r0 along the middle line of radius
'j

r will be in the direction of th(‘ tangent to this line as shown. From eq. (6.11), its
magnitude is

V
aA =
.
r P 172
— sin
r = y- / 1/ T-. / r cos \p • trd\l/ = j

Wv nob^ that this slaair stress is a maximum when 0 - tt/ 2 and zero for and 0—0
for 0 =
TT. The (ior responding ehmamtal sheai force is rtds = Ttrd<p and its moment
about point C is rtr'dl(f). The sum of these* moments o\(t the entire cross-section
becomes

7^ - sill <t>a<t>
“ -•
-K.i:
-
y
riie horizontal ^lomponents (A the elemental shear force's rtrchp above the neutral
axis Caiiool th(' liorizontal components of thovse below the nc'utral axis; h(*nce, the
shear stress resultant is a verti(;al force equal to the shf*ar V at the .section. To
prodiH'e the twisting moment T, caleulatial above, this force must act through a
point 0 such that

2VrH
V{r + e) T-----
from which,

(g)

where

'^7'H
,
rH cos® 0a0
„ , , ,
=
2
166 STRESSES IN BEAMS: II

Substituting? this value' of f, into eq. (g), we find

1.

PROBLKMS
Provo that lor simple Ix'iidinji ol a beam of angle section with equal legs
(Fig. A), the shear center axis Ox is defined bv the lint* of intersection of the two
planes of the l(*.gs.

Fig. a Fio. B Fig. C

2. C^alculate the distance e from the plane of the web to the shear center 0 of the
channel section hi Fig. 6.22 if the section a standard 6lJ8 2.
is e — 0.780 in.
3. Prove* that the shear center 0 for the balanced Z-section shown in Fig. B
coincides with its centroid t\

i. Locate the shear center 0 for the unbalam'cd I-section shown in Fig. C, for

simple liending iti the plane of the web. Arw. e = 0.41 in.
5. Locate the shear center O for the C -section shown in Fig. D, for simple bending

in the plane of the web. c = 0.55 in.

6. Locate the shear center 0 for the circular sector shown in Fig. E.

^ sm Of — a cos a
Am. e 2r :
r.
a — sin Of cos Of
 RENDING STRESSES IN CURVED BEAMS 167

6.6 Bending Stresses in Curved Beams

In this section we consider briefly the theory of pure bending of an
initially curved bar within the elastic range, l^eferring to lug. r).24a, con-
sider a short portion of a curved bar acted upon by couples of moment in M
the plane of initial curvature. Such bending moment which tends to
decrease the initial curvature' will be considered as positive. Each cross-
section of the bar is assumed to have an axis of symmetry wliich lies in the
plane of initial curvature. The locus of tin; centroids then is a platan curve
(‘ailed the center line of the bar and its radius of curvatun' is denoted by K.

(a) (b) (c)

Fig. 6.24

In discussing the stress distribution produced by jnire bending of such a

curved bar, we make the same assumptions as m iIk* of straight, bars,
namely, that transvTr.se (Toss-sect.ions of tin* bar, onginall^’ plane and
normal to the (Tuter line, remain so after bending, bet ah and cd denote two
neighboring cros.s-sfM?tions of the bar and let d(t> dmiott^ the small angle
between them before bending. As a result of bending, the cross-section cd
rotates with respect to ab. Let Adtp denotf* ihe small angle of rotation.
Due to this rotation, the longitudinal fillers on the c'onvTx sid(‘ of th(‘, bar
are compressed and the fil>ers on the concave side are extended.
n-n If
denotes the neutral surface, the extension of any fiber at the distance y
from this surface is y(Ad4)) and the corresponding unit elongation is

(a)
(r - y)d<t>

where r denotes the radius of the neut.ral surface and the denominator in
eq. (a) is the length of the fiber between the adjacent cross-sections before
]G8 STRESSES IN BEAMS: II

b>ending. A.ssuniing that then* is no lateral pressure between the longitu-

dinal fibers, * the bending stress at a distance y from the neutral surface is

, . (b)
(r - y)d<i>

LOq. (b) shows that the stress distribution is no longer littoar as in the case of
straight bars, but that it fcdlows a liyperbolic law as slu)wn in Fig. 0.24c.
From the condition that the sum of the normal forces distributed over the
fToss-seetioii is zero, it can be concluded that the neutral axis is displaced
from the centroid of the cross-section towards the (*enter of curvature of the
bar.
In the case of a rerJnngaUu' cross-section, the shaded an^a (Fig. 0 24c) in
tension must ecpial that in compression; heiuv the greatest bending stress
acts on the concave side. In order to make the stresses in the most remote
fibers in tension and in compri'ssion equal, it is necc'ssarv to use sectional
shapes which have the centroKl nearer the concave side of tlu* i)ar.
Kq (b) contains two unknowns, th(‘ radius r of th(' neutral surface' and
the angle' whitdi r(‘pre.sents th(‘ angular displat'cmfuit due to bending
To determine them, we must use two equations (4 statics. The first e(|i:a-
tion is based on the condition that the sum of the normal fon'cs distributed
over a cross-section is eijual to zero. The si'cond equation is based on the
condition that the moment of these normal forces is equal to the bending
moment M. Thus

fadA
_ EiAd<f>)
-0, (c)
J 'i<i> J r
r
- y

<J »''''
dA 11
fr . ij. (d)
l<Ty f - y

The integratior in both equations i.s extended over the total area of the
cTos.s-section
Eq. (c) enables one to determine r and, in turn, the distance y (considered
a positive quantity) from the centroidal axis to the neutral axis of the cross-
section. Let V represent the distance from the center of curvature to any
element dA; then y - r — v, and e({ (c) can be written

-*•
v)dA
= 0 ,
V
from which

(6.13)

*The exact theory shows that there is a certain radisl pressure but that it has no sub-
on the stress a and can be neglected.
stantial effect
 BENDING STRESSES IN CURVED BEAMS 169

or

y = R- ( 6 14 )
.

where R is the initial radius of cuivature of the center line ol the bar.
Eq. (d) may be used to obtain a formula for the fiber stresses in terms of
the bending moment. The integral in eq. (d) is first simplified as follows:

(o)

The first integral on the right side of eq. (e) represent.s the moment of the
(Toss-sectional area with respect to the neulral axis, and the second, ns is

seen from eq. (c), is equal to zero. Hence

fi/ijA
Ay, (f)
J r - If

Eq. (d) then hei‘omes

7£(Ad0) M_
^
<i<t> A Tf
Substituting this in ecp (b),

Ay(r - y)

The stresses in the most, remote fib<»rs which an* the maximum stres.ses m
the bar are

Mh, Mhj
^ in fix and Cm n 6 15 )
Ay a (
I .

Aijc

in hi and
which are the distances from the neutral axis to the mo.st remote
fibers,and a and c are the inner and outer radii of the bar.
So far we have considered the case of pure bending wdiere the bar is
subjected to end couples only. In a more general case wdien a curved bar is
bent by transverse forces ac ting in its plane of symmetry, the forces acting
upon the portion of the bar to. one side of any cross-s(*ction may be reduced
to a couple and a force applied at the centroia of the cross-section. The
moment of this couple equals that of the external torces with respect to the
centroidal axis of the cross-sc^ction. The stres.ses produced by the couple are
then obtained as explained above. The force is resolved into tw^o compo-
nents, a longitudinal force N in the direction of the tangent to the center
line of the bar and a shearing force V in the plane of the cross-section. The
longitudinal force produces tensile or compressive stresses uniformly distrio-
17U STRESSES IN BEAMS: Tl

utf*d (A or tJio cross-sort ioii and equal to S / A. To g;et the total axial stress
acting in any fiber, this uniform stress is added algebraically to the stress
caused by the couple. The transviTse force 1' produces shearing stresses

and th(‘ distril.iution of these stresses over the cross-section can be taken the
same as for a straight bar.
KXvvMi’Lh 1. D(‘tcrrnine the numerical value of the ratio for the cawse of
:i curved beam of rectangular cross-section in |)urc bending if R = 5 in. and /i = 4 in.
souiTioN. Erorn cfis. (b.lo)

^inax h\C
^
(h)
^iiiiji h 2O

where — h/2 — y and //.- ~ h 2 T V (se(^ Fig. 6.24b). d'o calculate y, we use
(‘q. (6.14) in wliicli

Thus
hi,
y R - 5 — 4.72 = 0 2S in
O.H476

'I'hen hi - 2 — ().2S 1.72 in. and -= 2 4- 6.2S 2.28 in, A\'itli the.sc value.-'
of hi and //'j, (‘(|. (h) becoiiK's

l./2X^ ^

F:\AMrLi 2 \ ciir\ <‘d b(‘ain \s ith a circular center line has the invented T-section

shown in Fig. 6.2o, and is ^ubjeeb'd to pure biUKhng in its plane of symmetry. Find
the dimiaision h\ m ordor b) ha\e (‘(pial teiisil(‘ and comjire.ssive stn^sses in extreme
fibers.
soijjTioN. Sin(‘(‘ v\e re(iuire | |
- ]
aunn I
, it follows from cqs. ( 6 15 ) that
.

(I r
 BENDIXC; STRESSES IN CrRVKl) BEAMS 171

from which
b
h‘2

I’hen since Ai + //> - 4 in., we find hi = 1.2 in., //«.. = 2.S in. This locates the
neutral axis of th(' cross-section and its radius r « -f // . 3 + 1.2 = 4.2 in.
This radius of tli(* neutnil axis is also defined by ecj. (6.13) wherein

^ " ^
I (0 (0
so that
"h 3
= 4.2 in.
dA 0.28Sf>, -f 0.5G0
/'
V

This gives hi
— .3.09 in.

EXAMPLE 3. A semicircular curved bar is loaded as shown in P'ig. 6.26a and has
the trapezoidal cioss-section shown in Fig. 6.26c. (’alculate thi' tensile stress ai at
point 1 if a = ^ = 1 in. and I* = 1000 lb.
SOLUTION. Considering that portion (d the bar abov<' the section 1-2 (Fig.
6.266), wo see that the* stress resultant on this .si'ction consists of a force P acting
at the centroid of the section and a bending moment .1/ - PR where R is the radius
of tlie centroidal axis. The extreme fiber stress at point due to the moment 1 M
will be given by the first of ec^s. (6.15). and that due to the force P will be simply
P/A . ddius the total stress at point 1 is

PRhi P
<Ti = (i)
A [fa 1

For the gi\'en cros.s-scction (see Fig. 6.26c) we find

d = X /, 2<* -

R = hjl \bh
iMv
bbbzJbi
S)ch
14

dA h b.
^
1.
Ja V eja r. 2

14 1^'
= ft - '

9 o

-14
/ti ft - a - z/
- — - 1
- I

2“‘-
.

18

Substituting these numerical values into eq. (i), we obtain

= 1000 X Hi; XJ nm .
(Ti
X X 1
^
i /18
 172
STRESSES IN BEAMS: II

PROBLEMS
1. Determine the numerical v'aJue of the ratin /r f
beam of circular cross-section in
pure bendine if /?
^ curved
in
cross-section is 4 in. A ns. ’ diameter d of the
= l.s/
«ho\„^in ""‘‘7 trapezoidal eross-«.ction
rlg'.'^A'^aTd ^slbj/cief? 'T "
face 6. ia the e side ^7
of base widths so that the
btr'if f-Ti
'

'• ^
extreme fiber *” tension
will be numerically and compression
equal. Ans. b,/hi = I.S7

ioaderrwVrCB. ^L'^lSd^J St
^ pjrr-‘-"i£
"* “ *»" '» r*- c-
 CHAPTER VII

ANALYSIS OF PLANE STRESS AND

PLANE STRAIN

7.1 General Case of Plane Stress

/

In preceding discussions of beams bent by transverse loads, we have

seen how an element of material in the beam can be subjected to both
normal and shearing stresses on its edges as shown in Fig. 7.1a. A similar
situation will occur in the case of an element of a shaft subjected to axial
loads and twisting moments as shown in Fig. 7.1b. Such a state of stress

Fig. 7.1

on the edges of a rectangular elemiuit in which there are no stresses normal

to its face is called j)lanc stress. After Mich normal and shearing stresses
as those shown in Fig. 7.1 have been found, it is frequently necessary to
examine further the state of stress within the element to find the magnitude.s
and directions of maximum stresses.
I^et us consider now the general case of an element under plane stress
as showm in Fig. 7-2a. The normal stress in the .r direction is denoted
by <Txj that in the y direction by ay, and tension is considered positive.
The shear stresses on the edges of the element that are normal to the
x-axis are denoted by Txy while those on the edges normal to the t/-axis are
denoted by ry*. The shear streseses r^y, having a clockwise sense of rotation
about a point inside the element, are to be considered 'positive in accordance
with our previous rule (see p. 28). The shear stresses ry., having a counter-
173
174 PLANE STRESS AND PLANE STRAIN

clockwise 8ens<^ of rotation, are negatwe. From the requirement of equality

of complementary shear stresses (see p. 30), we have Tx// = —Tyx. Because
of this equality of orthogonal shear stresses, it is customary to use only

one notation Tj,y for these stresses without regard to order of subscripts,
but in so doing, it is necessary to remember that shear stresses giving
counterclockwise rotation are to be treated as negative.

niven the state of plaiu‘ stress shown in Fig. 7.2a, the normal stress
Gn and th(‘ shear stress r on any plaiu* whose normal n makes the angle <#>

with the .r-axis can easily he found from the equilibrium conditions of the
triangular el(*meiit shown in 1^'ig. 7.2b. l^et the area of the inclined face
of this element he cienoted hy d4„; then the areas of the other two faces
are fMx = d. „ cos 1 and dAy — dA,, sin 0. Multiplying the various
stresses hy the areas of the faces on which they act, the total foiX'es on the
triangular (‘lement will be as shown in Fig. 7 2h. Then for eciuilihrium in
the n direction, we must have

CndA n - GjiA n cos'^ 0 + G A n siiF 0 — 'Ij^ydA ,, cos 0 sin 0. (a)

Similarly, for equilibrium in the direction perpendicular to n, we rmist

have

rdAn = GjdAn cos 0 sin 0 — GydAn sin 0 cos 0 + rxyd^„(cos^0 — sinVh

(b)
Equations (a) and (h) are readily reduced to

Gn = Gx COS“ 0 + (Ty sin^ 0 — 2rry SlU 0 COS 0

~ ^ 4" Gy) + ^ (gx — Gy) cos 20 — Txy sin 20,

T — {gx — Gy) sin 0 cos 0 + 7xy(cos2 0 — sin^ 0)

=" ~ Gy) sin 20 + Txy cos 20,

which are analogous to eqs. (3.2) in Art. 3.2.
 GENERAL CASE OF PLANE STRESS 176

To find the lo(*ation of the plane of maximum normal we

fstress crn,

make the derivitive d(Tnld<i> = 0 from the first of eqs. (7.1) and obtain

— {oj — (Ty) sin 2<i>

— 2Txy cos 20 == 0, (c)

from which

tan '24 = (7.2)

C^x

This condition defines two values of 20 differing by 180° and hence two
values of 0 differing by 90°, For one of these values. a„ is a maximum
and for the other, a minimum.
Considering the second of eqs. (7.1) and setting the shear stress t equal
to zero, we again obtain eq. (c). From this, it may be (‘oiicluded that on
those planes where Cn is a maximum or a minimum the shear stress r
vanish(‘s. The corre.sponding normal stresses (anj,„ax and (a-n)min are
called 'principal stresses^ and the planes on w'hich they act are (tailed

principal planes of stress.

Ueferring to the second of eqs. (7 1) and s(‘tting dr/dtii = 0, we obtain

{<Tx i^v) COS 20 — 2Txy siii 20 = 0, (d)

from which

cot 2</> = f- - (7.3)

<Tx (Ty

Comparing this with eq. (7.2), w'e see that the maximum shear stresses
occur on orthogonal planes Vnsecting the angle between principal planes,
i.e., at 45° to the planes of principal stress as already concluded in Art. 3.2
To evaluate the maximum normal and shear stresses in the element,
it is necessary to replace 0 in eqs. f7.l ) by the values defined by eejs. (7.2)

and (7.3). Because of the transcendental character of the equations, this

becomes somewhat involved and, for this purpose, it will be simpler to
use Mohr\s circle.
The general case of plane stress is shown again in J-'ig. 7.3a. To construct
Mohr^s circle for this case, one proceeds as follows: Lay out first the
coordinate axes a and r with origin at 0 as shown in Fig. 7.3. b Then to
locate the point I) representing the stress conditions on the x-plane, i.e.,

the plane normal to the .r-axis, lay off the value of Ox as abs(dssa OF and
the shear stress Try as a positive ordinate FI). '>ext, locate the point Z>i,
representing the state of stress on the //-plane, by laying off the abscissa
OFi to represent the normal stress and the negative ordinate F\Di to
represent the shear stress — Txy. Since the x- and y-planes are orthogonal,
the corresponding points D and Dy on Mohr's circle are 180° apart and
176 PLANE STRESS AND PLANE STRAIN

represent the ends of a diameter. Connecting these points with a straight

C on the cr-axis and the circle can be drawn as shown.
line locates the center
The maximum and minimum normal stresses are represented in Fig.
7.3b by OA and respectively. These principal stresses are denoted
hy (^^n) max (Ti and {(Jn)nMu ^ ff2, as shown. To locate their directions in
Fig. 7.3a, we start with point D on the circle, corresponding to the j-plane
of the element, and label this point 0 = 0, as shown. Then to reach point
A on the circle, corresponding to the plane of maximum prinicpal stress,
it- is necessary to pass through the clockwise angle 20rt. Hence, the direction

of ai in Fig. 7.3a found by laying out, also clockwise, the angle <t>a from
is

the x-axis, as shown. The direction of <72 is then at right angles to that of
<Ti and the principal planes are located as shown.

Fig. 7.3

In general, any plane through the element whose normal n makes the
angle with the x-axis and the corresponding point E on the circle, repre-

senting the state of stress on this plane, are related in the same way,
namely: the angle DCE in Fig. 7.3b is always double the angle 0 in Fig.
7.3a and is to be measured in the same direction.
Expressions for the principal stresses o-iand 0-2 are easily found in terms
of (Ty, and Txy, from the geometry of Mohr’s circle, Fig. 7.3b, as follows

OA OC + CD ^ ^

c,^0B = 0C-CD = I

These are the same values which w^ould be found by substituting the value
of 20 from eq. (7.2) into the first of eqs. (7.1) on p 174,

EX^PLB 1. A ^iquare element of a thin plate subjected to plane stress is shown

in Fig. 7.4a. The given stresseson its mutually perpendicular faces are <r, * —500
 GENERAL CASE OF PLANE STRESS 177

psi, (Ty = -hl500 psi, Tjy = = +10(K) psi. Construct Mohr's circle for this
element and find :

(a) the magnitudes and directions of principal stresses <ti and a 2.

(b) the normal stress (Tn and shear stress t on the diagonal plane ab.
(c) the aspect of the plane of maximum positive shear stress Tma*-
SOLUTION. From the origin 0 of the or-plane in Fig. 7.4b, lay off OD' = —500 psi
and D'D = -|- 1 000 psi to locate point />, representing the given state of stress on
the x-plane. Also from 0, lay off 01)^ — -f 1500 psi and D'lDi = —1000 psi to
locate point Di, representing the given state of stress on the y-plane. Connect
points 1) and Di by a straight line to locate the center C of Mohr’s circle and draw
the circle as shown. Alsf) label CD with <t> — 0 to serve as a reference line. Then the

Fig. 7.4

measured angle DC A — 2 </)q = — 135°, negative to show that it is clockwise. Hence

== — 67°30' and the direction of the principal stress Oi is as shown in Fig. 7.4a;
0a
the direction of <T2 is, of course, at right angles to this. The magnitudes of a. and 0-2
ar(‘ represented, respectively, by OA = <ti ~ -}- 1910 psi and by OB = <72 = —910

psi, scaled directly from the drawing The same values can be obtained from cqs,
(7.4) by substituting therein the given numerical values of ctx, Oy, and Txy, with due
regard to signs.
The normal to the plane ab which bisects the angle between the x- and yy-planes
is make an angle 0 = —45° with the j-axis. Hence from the reference point
seen to
D in Fig. 7.4b, we measure off the angle DCE = 20 = —90° to locate point E,
representing the state of stress on this plane. Scaling the coordinates of point E, we
find (Tn = 4-1 500 psi and r = -h 1000 psi.

Reference to Fig. 7.4b shows that point F on Mohr’s circle reprcvsents the plane of
maximum positive shear stress in the element. Then since the angle DCF measures
20 = —45°, the plane of this maximum positive shear stress is defined by 0 =
— 22°30' as shown in Fig. 7.4c. The magnitude of this maximum shear stress is
Tmax = CF = 1410 psi, scalcd from the drawing. As a check, it will be noted that
the angle between planes of prirtcipal stress and planes of maximum shear stress is
07°30' — 22°30' = 45° as it should be.
EXAMPLE 2. one element of a plate in a state of plane stress
Fig. 7.5a represents
and Fig. Mohr’s circle. Point D on this circle corresponds
7.5b, the corresponding
to the x-plane, point Di to the 2/-plane, and points .1 and B, to the two principal
planes. Point D' is vertically below D and point D’l, vertically above Di. Show
 178 PLANK STRESS AND PLANE STRAIN

(a) (b)

Fk;. 7.5

that tho si(J(*a of tlio rectangle Aiy iBD' represent the directions ot p)rincipal stress
in the element.
SOLUTION. Since dR is it follows that AD'B and D'BIY
a diameter of the circle,
are right angles. iVlso, since the arc ALY
equal to the arc DA, the angle ABD^ i.s
is
equal t^) lialf the angle DC A I'ha.s the angle ABD' = 0^, which defines the direc-
.

tion of the principal .stress <7i, a.s shown. This simple construction furnishes a
convenient way to find the' directions of principal stresses from Mohr’s circle.

PROULKM.s

iT For the rectangular element .shown in Fig. 7.4, th(; following data are given:
Gx = 5000 j)si, (Ty = 3000 psi, and = 750 psi. Determine the magnitude's and
directions of th(' two principal stre\ss('s ay arui Am. 0i — “•18°20'; uy — 5250
psi.
Find the magnitudes and directions of principal stres.se.s for tlie element in
Fig. 7.4 if O’ r == —hfXlO psi, Oy — 3000 psi, and Txu — 1000 psi. Am. ay — 3120 psi;

0-2 -- —5120 psi: 0a ”

— 82®50'.
D)cate the planes of maximum shear stress for the element in Fig. 7.4 if
(Ti +5000 psi, or,/ = +3000 psi and t^u — 750 psi. What are the magnitudes of
normal and .shear stress on these planes? Arts. 0 =
26°34'; On — tr'n = 4000 psi;r,uax = 1250 psi.
4. Find the magnitudes of normal stress a,, and
shear str(*ss r on tlie 45°-piane ab of tho element in Fig.
7.4 if ar ~ (Ty ~ 3000 psi and r^y — 1000 psi. Ans.
an = 4000 i)si r = 0. ;

5. Construct Mohr’s circle for the element in Fig.

7.4 if (Ti = GTy = 0 and r^y == 1000 psi. What are the
jf
magnitude's and directions of principal stresses in this
case? Ans, ay = — co = 1000 psi;0a = —45®.
6. Construc'.t Mohr\s circle for the element in Fig. 7.4
if
— —ay — 2000 psi and T^y = 0. What is the
a,
magnitude of maximum shear stress rmax in tliis case?
Fig. a Ans. Tnia* = 2000 psi when 0 = +45®.
 GENERAL CASE OF PLANE STRESS 179

7. A hard rubber cube (ju = 0.5) is completely constrained in the x direction by

the walls of a trough but is free to expand or contract in the y and z directions as
shown in Fig. A. The upper face of the cube is subjected to uniform compressive
stress fjy
— —1000 psi as shown. Find the direction of maximum positive shear
stress Tmax and evaluate the magnitude of this stress. Hint: Sec eqs. (3.5) for stress-
strain relationships in biaxial stress. A ns. r„,ttx = 250 psi at 0 - 45®.

7.2 Principal Stresses in Beams

Consider in l"ig. 7.6 the case of a simply supported beam of rectangular

cross-section subjected to uniformly distributed transverse loading.
Then for any cross-section defined by the distance .x from the left support,
the shear force will be

ir
Fx = - WX

and the bending moment will be

wlx
A/x
2

both of which vary with the distance x defining the location of the cross-
section. The normal stress <rx and the shear stress r^y at any distance y
from the neutral surface can be readily calculated from the formulas
developed in Chapter V. Thus, from eq. (5.5)

(Tx (a)

and, from eq. (5.8),

Since Mr and F* both vary with x, these stresses are seen to vary con-
tinuously with both X and ij For any particular
throughout the l)eam.
cross-section, the bending stress cr* and the shear
varies linearly with y
stress Txy varies parabolically as shown in Fig. 7.6a. For a rectangular
element such as A or B, Fig. 7.6b, the shear stress vanishes and the cor-
responding tensile or compressive stress due to bending represents a
principal stress. For an element C situated at the neutral surface, the
normal stress vanishes and the element is in a state of pure shear. For
such conditions the directions of principal stress are inclined by ±45°
to the.x-axis. For any intermediate element D, there are both normal and
sliearing stresses to consider. These stress conditions for the element D
are shown separately in Fig. 7.6c.
180 PLANE STRESS AND PLANE STRAIN

Fig. 7.6

The state of plane stress shown in Fig. 7.6c is simply a particular case
of plane stress as discussed in Art. 7.1, one of the normal stresses in this
case being dy = 0, Thus to find the magnitudes of principal stress for
this element we use eqs. (7.4), which, with dy = 0, become

<ri,2 = ^
It is seen that in this case the maximum principal stress o-i will always
be positive, i.e., tension, while the minimum principal stress d 2 will always
be negative, i.e., compression.
The directions of these principal stresses will be found by using eq.
(7.2) which, with dy = 0, becomes

tan 2* = (7.0)

For known values of o-, and T*y, this gives two values of 0 which locate
the directions of principal stresses di and shown
d^ as in Fig. 7.6c.
Applying eq. (7.6) to elements A or 5, we have = 0 and obtain
tan 2^ “
0, yielding 0i =* 0, = 90®. Thus at the lower and upper free
surfaces of the beam, the directions of principal stress coincide with the
.T- and 2 /-axe 8 as anticipated above. For the element C at the neutral
surface, tr, = 0, and eq. (7.6) gives tan 20 = — yielding 0 = ±45®
,
 PRINCIPAL STRESSES IN BEAMS 181

Thus, for points in the neutral surface of the beam, the directions of
principal stress cross the x-axis at ±45° as already anticipated.
By calculating <t>
from eq. (7.6) for a number of points defined by co-
ordinates X and y, it is possible to construct two families of orthogonal
curves whose tangents at each point coincide with the directions of principal
stresses at these points. Such curves for the right half of the beam under
uniform load will be as shown in Fig. 7.6a. They are called 'principal
stress trajectories. The solid line curves represent tensile stress trajectories
and the dotted line curves, compressive stress trajectories. Both sets of
curves cross the x-axis at 45° and always cross each other at right angles.
They terminate in the upper and lower free surfaces normal to the axis
of the beam.
In beam wants the numerically maximum values of
design, one usually
normal stress o-„. From eq. (7.5) it can be seen that for the most remote
fibers in tension, where the .shear stress is zero, the longitudinal normal
stress (Tx becomes the principal stress. For fibers nearer to the neutral
axis, the longitudinal fiber stress ax is less than at the extreme fiber; how-
ever, we now have a shear stre.ss Txy also and the stresses a* and tzv acting
together at this point may produce a principal stress, given by eq. (7.5),
which will be imrnerically larger thari that at the extreme fiber. In the
case of beams which the shearing
oi rectangular or circular cross-section, in
stress Txy varies continuously over the depth of the beam, this is not usually
the case, that is, the stress ((rx)max calculated for the most remote fiber
at the section of maximum bending moment is the maximum stress acting
in the beam. However, in the case of an I-beam, where a sudden change

occurs in the magnitude of .shearing stre.ss at the junction of flange and

web (see p. 129), the maximum stre.ss calculated
at this junction from may be larger than
eq, (7.5)
the tensil(‘ stress (o'i)rriax in the most remote fiber
and .should be taken into account in design.
To illustrate, consider the case of a simple beam
carrying a concentrated load P at the middle,
assuming / 2 ft and P = 60,000 lb. The cross-
vsectioii is an T-section as shown in Fig. 7.7; h = 12
in., hi = 10.5 in., 6 = 5 in., and 6i = 0.5 in.,
/, = 286 in.^Then Mmax = 30,000 ft-lb, F^ax =
30,000 lb. From eq. (a), the tensile stress in the
most remote fiber is Fig. 7.7

,
30,000 X 12 X 6 .

(^x)max — 7550 psi.

286

Now for a point at the junction of flange and web, we obtain the following
182 PLANE STRESS AND PLANE STRAIN

values of normal and shearing stresses:

cr,

Txy
=
=
7550

4430
X
psi.
^ = 0610 psi;

Then from eq. (7.5), the principal stress is

{<Tn)nnxx = 8830 psi.

It is seen that (<Jn)inax at the junction betw(‘en flange and web is larger
than the tensile stress in the most remote fiber and therefore should be
considered in design.
PROBLKMS
1. For the simply supported I-beaiP disoussed on p. 181, dotennino the span
length I such that (<r„)max at the junction between flange and web will be equal to

(<rj)inax in an extreme fiber. Ans. I = 39.8 in.

2. A cantilever beam has a rectangular cross-section of width h — 4 in. and deptn
A = 8 in. and carries a load P = 2000 lb at its free end. Calculate the magnitude
and direction of the principal compressive stress for an element situated midway
between the neutral surface and the bottom free surface in a cross-section 3 ft from
the loaded end of the beam. Ans. a 2 = --850 psi; = 4®44'. </>

10”

3. The beam shown in Fig. A carries a load P= 10 kips. Construct Mohr's

circle for stress on an clement at A and find the magnitude and direction of the
principal tensile stress cri at this point. Ans. a i ~ 194 psi; = 31°30'. —
4. If the beam in Fig. A has an 8WF20 wide-flange section and the load P — 25
kips, calculate the maximum normal stress induced in the beam. Ans <ji = 21,800
psi.
5. Write the differential equation for the family of principal stress trajectories
for the cantilever beam of rectangular cross-section shown in Fig. B.
 COMBINED BENDING AND TORSION 183

7.3 Stresses Due to Combined Bending and Torsion

One of the most important applications of the theory of combined
stresses arises in the case of shafts subjected to both bending and torsion.
Consider, in Fig. 7.8, a portion of a shaft of circular cross-section acted
upon at its ends by bending couples M
in the x|/-plane and by twisting
(^ouples T about the .r-axis. Under such loading, an element A in the top

surface of the shaft will carry normal stress ax due to bending, and shearing
stresses Txz due to torsion. I'hese stresses are readily computed from eq.

(5.5) and from cq. (4.5). Thus, wo have

Me 32M
(a)
ird^

Tr ler
(b)
7rr/»’

where d is the diameter of the shaft.

The element A is seen to be in a state of plane stress and the principal
stresses will be found from eqs. (7.4) of Art. 7 Thus the principal normal
J .

stresses are

(7.7)

The maximum shearing stress, equal to half thc^ difference between principal
stresses, is

ITIOX (7.8)

In the design of a shaft to carry a given loading, allowable working

((»r compression) and in shear will usually be prescribed.
stresses in tension
It then becomes necessary to find out from eqs. (7.7) and (7.8) which of
184 PLANE STRESS AND PLANE STRAIN
those stresses governs the design and then to select the required diameter
d of the shaft accordingly. For a brittle material like cast-iron, the maximum
normal stress ax should be used, while for a ductile material like structural
steel, the maximum shear stress rmax is most commonly used.
To find the reciuired diameter d of the shaft, it is necessary to substitute
expressions (a) and (b) into cqs. (7.7) and (7.8). Doing this, we obtain

<7. = ^ {M + yiiP + r=) (7.7a)

and

r„.ax = ^ ylM^ + T\ (7.8a)

Then to select the diameter d on the basis of an allowable working stress

Ow in tension, we use eq. (7.7a) which gives

d = {I-- {M + (7.9)
\ TTCtif

{Similarly, to select the reciuircd diameter d on the basis of an allowable

working stress Tu. in shear, we use eq. (7.8a) and find

d = W—
\ TTTifi
(7.10)

If allowable stresses in both tension and .shear are prescribed, we calculate

d from both ecjs. (7.9) and (7.10) and select the larger value.
Usually the bending of a shaft will be produced by transverse forces
rather than by bending couples as shown in Fig 7.8. In such case, we have
to con.sider also shearing stresses associated with noii-uniforin bending.
As observed in Art. 7.2, the.se shearing stresses usually vanish at those
points where the normal stresses due to bending are a maximum so that
they do not influence the maximum principal stresses and can be dis-

Fig. 7.9
 COMBINED BENDING AND TORSION 185

regarded. However, when bending by transverse forces is combined

with torsion, some consideration of the shear stresses associated with
bending may be necessary. Consider, for example, the shaft loaded as
shown in Fig. 7.9. In such case, a surface element A on the front side of
the shaft will sustain shearing stress

(c)

due to the torque T and also shearing stress

^ ^ SA
due to the transverse load P. Thus the maximum shear stress Txy =
+ t'" Even though the normal stress Ox for this element vanishes,
it might be in a worse condition than an element B at the top surface.

EXAMPLE 1. A steel shaft supporti^d in bearings at A and B and carrying pulleys

at C and D, is to transmit 100 hp at 500 rpin from the drive pulley I) to the offtake
pulley C as shown in Fig. 7.10. The following numerical data are given: Pi = 2 P 2,

Q\ = 2 Q 2 Pd = 6 in., Pc = 8 in., = 4 ft, a = 1 ft, and the working stress in shear

,
/

is Tte = 6000 psi. Calculate the required diameter d of the shaft.

SOLUTION. From eq. (4.7), p. 73, the torque required to transmit 100 hp at
500 rpm is

500
Then since T = (Pi —P 2) Pd we find, with the given data,

Pi = 2P 2 = 4200 lb.

Fig. 7.10
186 PLANE STRESS AND PLANE STRAIN
Similarly, T - {Q\ —Q 2) from which

Q, ^ 2Q, = 3150 lb.

With these values of the loads determined, we turn our attention to the bending
moments in the shaft caused by the transverse loads. It will be noted that the shaft
undergoes bending both in the horizontal jr^-plane and the vertical jy-plane. The
corresponding l)eriding moment diagrams are shown in Fig. 7.10b. A study of these
diagr 3,ms shows that the maximum bending moment in the shaft will occur either at
cross-section B or at cross-section C. The bending moment at B is

Mu = (Pi + P 2) a = 6300 X 12 = 75,000 in.-lb.

At C, the bending moment in the vertical plane is

700 in.-lb,

while that in the horizontal plane is

These two bending moments in

^ X 24

orthogonal axial planes of the shaft

= 37,800 in.-lb.

may be added
vcctorially and we obtain

Mr = = V‘i6.44(10)« = 68,200 in.-lb.

This resultant ixmding moment at C being slightly smaller than that at B and the
torque T being the same at both these cross-sections, we conclude that an element
on the front side of the shaft at B is the critical element. Substituting = 75,600 M
in.-lb, T = 12,600 :n.-lb, and Tu- = 6000 psi into eq. (7.10), we obtain-

d
16 sj ie X 76,700
V(75, 600)2 4. (12,600)2 4.02 in.
OOOOtt \ eoooTT

EXAMPLE 2. A 2.5-in.-diameter shaft carries a 30-in. -diameter pull(\v wcigliing

500 Ib (Fig. 7.11). Talculate the principal tensile stress at the section mn if the belt
tensions aie as shown.

SOLUTION. At the section mn,

T = (1750 - 250) 15 = 22,500 in.-lb,

Af = 6V(500)2 + (2(X)0)2 = 12,370 in.-lb.

Fig. 7.11
 COMBINED BENDING AND TORSION 187

Then from eq. (7.7a)

<r, = -4^, (12,370 + >/(12.370)* + 22,500)*) = 12,400 psi.

PROBLEMS
steel shaft in bearings A and B at its ends carries a pulley at C as
supported
shown in Fig. A. applied by a torque T at A and taken off through a belt
Power is

overrunning the pulley, the tensions in the two branches of the belt being as shown.
The allowable working stresses for the shaft are = 12,000 psi and = 6000 psi.
Find the required diameter d of the shaft. Ans. d = 2 in.
Q. A vertical tube built-in at its lower end is acted upon by a horizontal force
P = 250 lb applied at the upper end as shown in Fig. B. Determine the principal
stress Cl and the shear stress if the section modulus for the tube is Z = 10 in.*

Ans. a -
= 7530 psi; Tnmx = 3780 psi.

Fig. C Fio. D
188 PLANE STRESS AND PLANE STRAIN
Figure C represents an instantaneous position of a portion of a crankshaft, the
3.
force P representing the action of a connecting rod on the crank pin. Find the safe
value of the magnitude of this force if working stresses for the shaft material are
prescribed as follows: Cu ~ 12.750 psi; = 10,200 psi. Ans. P = 660 lb.
4. A hollow steel pipe is to be used as a standard to support a signboard as shown

in Fig. D. The maximum wind pressure on the face of the board is assumed to be
50 Ib/sq ft. The standard is unsupported laterally, and its outside to inside diameter
ratio is 1.12. The allow’able working stress in shear is given as Tu, = 8400 psi.
Calculate the required outside diameter d of the pipe. Ans. d = 5.20 in.
5. A stepped shaft of circular cross-section is built-in at C and carries a pulley at
A as shown in Fig. E. The pulley is acted upon by a vertical force P applied
tangentially to its rim. The shaft is to be designed on the basis of maximum shear
stress. Find the ratio d\/d 2 such that maximum shear stresses will be equal at
sections B and C. Assume I = 12 in. and r = 4 in. A/w. di/d 2 = 0.830.

6.

A shaft of diameter d bent in the form of a semicircle AB of radius R is built-in

at and loaded at B by a force P perpendicular to the plane of the ring as shown in
A
Fig. F. Thus an 3 cross-section C of the ring is subjected to both bending and twist.
^

Assuming that the diameter d is small compared with the radius Ry so that the
theory of bending of straight bars may be used, find the value of 4> for which the
principal stress ai will be a maximum. Arw. </> == 120°.
 COMBINED BENDING AND TORSION 189

7. Referring to Fig, G, calculate the diameter d of the shaft AD

required to safely
transmit horsepower under the conditions shown. Assume P1/P2 = 3 and r„ =
6000 psi. Ans. d = 2.49 in.
8. Referring to Fig. H, determine the required diameter d of the shaft to transmit
60 hp at 250 rpm from pulley C to pulley D. The power is applied at C by a hori-
zontal belt and taken off at D through a vertical belt. Each pulley is 30 in. in diame-
ter and the ratio of belt tension on the tight side to that on the slack side is P1/P2 =
3. ylns. d = 2.64 in.

7.4 Plane Strain

Very often the strains induced in an element of material subjected to

plane stress are of practical interest. To completely define the state of
strain in the j-.v-plane of such an element, it is necessary to specify the
linear strains and ty in two perpendicular directions and the change in

angle betAveon these two directions (the shearing strain 7xi/). In Fig. 7.12a,
let OX and OF represent two such perpendicular lines of unit length
coinciding with two edges of the element before deh)rmation. During
deformation, point A' moves to A"' and point Y moves to F' relative to
the origin 0, A' and F' remaining in the plane AOF. Then the strain in
the neighborhood of 0 is completely defined by the two extensions = XX'
and €y = FF^ and by the change yxy in the original right angle A"OF

as it deforms into the final angle X'OY', Extensjonal strains are considered
positive and compressive strains, negative. The shear strain 7*^ will be
considered positive when the right angle XOY is increased. With these sign
conventions, positive strains will be seen to correspond to positive stresses,
as previously defined.
We now consider the following general question: When a state of plane
strain by given values of
is defined Cy, and 7*^, in orthogonal directions

OX dnd OF, what are the corresponding strains associated with the
mutually perpendicular directions OA and OB which make the angle 0
with OX and OF, respectively, as shown in Fig. 7.12b? To answer this
190 PLANE STRESS AND PLANE STRAIN

question, consider a rectangular element OX AY whose diagonal OA makes

the angle with OX as shown in Fig. 7.13. Let rf.r, dy^ denote the lengths
of its sides and ds the length of its diagonal. As a result of the given
strains Cx, €y, OX elongates by the amount OY elongates by the
amount tydy, and the original right angle YOX increases by the amount
Fof ckrity, these three deformations of the element are shown
separately in Figs. 7.13a, b, c. The corresponding changes in the length
oi llic diagonal OA are +€rdx cos -f sin and —yxydx sin 0, as shown
</>,

in the figures. The net change in OA is the algebraic sum of these three

quantities and the corresponding strain in the direction OA is obtained by

Fig, 7.13

dividing this sum by the length ds. Thus with the observation that dx/ds
= cos 0 and dy/ds — sin 0, we obtain

ta = €x cos^ 0 + sin^ 0 — 7^ sin 0 cos 0 (a)

The strain eb in the directum of OB may be (obtained simply by replacing

0 in expression (a) by 90° + 0, giving

C-, = tjr siir 0 + Cy cos'-^ 0 + yxy sin 0 cos 0. (b)

By making the trigemometric substitutions siii^ 0 — ^(1 — cos!20),

cos^ 0 = 2(f + ^0), ^^-nd sin0 cos 4> ~ \ sin 20 in expressions (a) and
(b), they may be written in the form

Fig. 7.14
 PLANE STRAIN 191

= i(cx + €y) + -J(€x

— ey) cos 2<t> — ^yxy sin 2^,1

C6 = i(^x + €y) + — «y) COS 2^ + ^7xv sio 20. j

A general expression for the shearing strain 70^ may be obtained in a

similar manner. Consider, in Fig. 7.14, two adjacent rectangular elements
so proportioned that their diagonals OA and OB are mutually perpendicular
Then from F'ig. 7.14a representing the influence of Ci alone, we see that the
change in the original right angle BOA is as follows:

dx
ZAOA^ + ZBOBi =
^ ds
sin 0 6x dx' cos
ds'
0

= €x COS 0 sin 0 + ex sin 0 cos 0 = Cr sin 20.

Similarly, from Fig. 7.14b, the change in the right angle BOA due to the
strain alone is

- ZAOAi AA, BBi ty d\j COS 0 ty dy sin 0

^BOBi = _
OA OB ds ds'

= — sin 0 cos (f)

— €y cos 0 sin 0 = — sin 20.

Finally, referring to llg. 7 I4c, the change in the right angle BOA caused
by the shearing strain jxy alone is

A' 1 BBi _ yxy dx cos 0 yxy dx' sin 0

ZzVOAi - ZBOBi
OA “ OB "Is ds'

= yxu cos2 0 - yxy sin2 0 == 7 xy cos 20.

The algebraic sum of the above angle chaiiges gives the net change in the

angle BOA, i.e., the desired shearing strain y^b as follows:

yob = (cz — €y) sin 20 + 7zw cos 20,

which can arbitrarily be written in the form

hob = hr - sin 20 + hry COS 20. (7.12)

Comparing the first of eqs. (7.11) and eq. (7.12) with eqs. (7.1) of Art.
7.1, we see that there is a complete analogy between a state of plane strain
and one of plane stress. The linear strains fx, and «« in eqs. (7.11) and
(7.12) correspond, respectively, to the stresses o-*, ay, and an in eqs. (7.1),

while the half shear strains 7x^/2 and yab/2 correspond to the shear stresses
Txy and T, respectively. This suggests that a Mohr’s circle for strains may

be constructed in the same manner as was previously done for stresses.

In Fig. 7.15a, let €* and Cy represent given linear strains in the x and y
directions and let y^y represent the corresponding shearing strain, i.e., the
amount that the original right angle XOY increases due to deformation.
This shearing strain is represented in Fig. 7.15a by a clockwise rotation of
 PLANE STRESS AND PLANE STRAIN

OX through the ajigle yxy/2 and by a counterclockwise rotation of OY

through the angle yxy/2. Clockwise rotation of a line is to be considered as
positive and counterclockwise rotation as negative.
Now to construct a Mohr’s circle for this state of plane strain, we begin
with the coordinate plane c, 7/2, as shown in Fig. 7.15b. Laying out the
extensional strain tx as a positive abscissa and the clockwise rotation

yxy/2 of the line OX as a positive ordinate, we obtain point D on the circle

corresponding to the x direction in the plane of the plate. Similarly, laying
out the extensional strain ty as a positive abscissa and the counterclockwise
rotation yxy/2 of the line OF as a negative ordinate, we obtain point Di
on the corresponding to the y direction in the plane of the plate.
circle

Drawing the diameter Z)Di, the center C is located and the circle can be
drawn as shown. This done, the strains €„, and yab/2 associated with the
orthogonal lines OA and OB which make any angle 0 with the lines OX and
07, respectively, can be found from the circle as follows: Since is a

counterclockwise angle from A" to A, we lay out from D on the circle the
counterclockwise angle 2<t> to locate point E representing the direction OA
in Fig. 7! 15a,and draw^ the diameter EEi as shown. The coordinates of
points E and E] represents the strains and yab/2 associated with the
directions OA and OB in Fig. 7.15a. The positive coordinates of point E
indicate that the line OA undergoes extension and rotates clockwise during
deformation, while the coordinates of Ei (positive abscissa and negative
ordinate) indicate that OB also undergoes extension but that it rotates
counterclockwise. This means that after deformation the angle AOB is

slightly greater than a right angle.

Of particular interest in the case of a state of plane strain defined by
given values of e^, e,,, and 7*^, is the direction of principal strains. That is,
the directions of two mutually perpendicular lines 01 aad 02 which remain
 PLANE STRAIN m
at right angles after deformation and along which the linear strains ci and
€2 are, respectively, a maximum and a minimum. From Mohr^s circle,
Fig. 7.15b, we see that these principal directions are represented by points
A and B, since these points have zero ordinates indicating no shearing
strain. The clockwise angle DC A = 20i in Mohr’s circle indicates that the
direction of the principal strain ei makes a clockwi.se angle <t>i with the
direction OX. Similarly the clockwise angle DyCB = 2<i>] indicates that the
direction of principal strain 62 makes
same clockwise angle 4>i with the
the
direction OY. Thus 02 will be oriented
the principal axes of strain 01 and
as shown in Fig. 7.15c. The magnitudes of ei and are .seen to be ei = OA
and 62 = OB on Mohr’s circle. They are also seen to be the maximum and
minimum linear strains in the plane of the plate.
Analytic expressions for the principal strains ei and 62 in terms of given

values of c*, ey, and yxy can easily be found from the geometry of Mohr’s
circle (Fig. 7.15b) as follows :

OA = OC + CA = OC + CD,]
y (c)
= OB = OC - BC == OC - CD.]

Then noting that OC = + fy) while CD = yjiCD'Y + (DD')*. where

CD' - Ke* ~ «») and DD' = ^yxy, eqs. (c) become

(7.13)

- !(.. + )’
+ (f
..
I

The clockwise angle </>! defining the direction of these principal strains with
reference to OX is given by the equation

tan = — DD' 7gy/2 _ Txy

(7.14)
2<^i
CD' Wx ~ Cy) e, - €y-

obtained from the trigonometry of the right triangle CDD' in Fig. 7.15b.
Equations (7.13) and (7.14) are seen to be analogous to eqs. (7.4) and (7.2)
of Art. 7.1 for the magnitudes and directions of principal stresses. They can
be used for the calculation of magnitude and direction of principal strain
without the need of constructing Mohr’s circle, if preferred.

PROBLEMS
1. A state of plane strain in a steel plate is defined by the following data: e,
=
+0.00050, ey = +0.00014, y,y +0.00036. Construct a Mohr’s circle and find
the magnitudes and directions of principal strains. Ans. €1 = 575 (10)~*; 62 =
66(10)-®;*! » »22**30'.
194 PLANE STRESS AND PLANE STRAIN
2. S(jlve Problem 1 if = -0.00014, 6^ = -0.00050, jry = +0.00036. Ans.
€i = -65(40) 62 - -575 (10)-«;«, = -22°30'
3, Using eqs. (7.13) and (7.14), find the magnitude's and directions of principal
strains for the states of plane strain defined bv €x = +0.00050, 6y = +0.0(X)30,
= -0.00105. Ans. e, = 935 (10)-«; e. = -135 (10)-«; 0, = +39°59'.

7.5 The Strain Rosette

In the case of a material subjected to plane stress, it is often desirable to

obtain the stresses by

measurement. Since stresses cannot be
direct
measured directly, this requires the mt'asurement of strains or deformations
which take place in the material during loading. Such measurements are
usually made with very sensitive strain gages attached to the surface of the
body before it is loaded so that they can record the amount of strain that
takes place during loading. The (jiiestion naturally arises as to how to
arrange these strain gages so as to get sufficient strain data to be able to
(‘ompute therefrom the corresponding stresses. An obvious answer to this
question would be to measure the linear strains ex and in any two per-
pendicular directions at a chosen point on the surface of the body and the
(ffiange in angle 7*v between these two directions. Then the principal strains

('ould he found as explained in the preceding article, and having these, the
principal stresses would be calculated from eqs. (3.6) of Art. (3.2). How-
ever, accurate measurement of the shearing strain 7iy is found to be very
difficult. It is easier and more a(*curatc to measure, in the neighborhood

of a chosen point on the surfuci* of the body, three linear strains €«, ct, €c,
in three different directions and then deduce from these measurements the
magnitudes and directions of the principal strains ei and Such a group
of strain gages is called a strain rosette. One standard arrangement of the
strain ros^d^e will now be discu.ssed in detail.*
7'he 45° Strain Hosette. and €c are measured along
If linear strains

lines oa, uh, and or, 45° apart as

shown in Fig. 7.16a, Mohr^s circle can be
constructed without ambiguity and the principal strains ci and €2 can Vje
fo\md therefrom. This Mohrts circle will be constructed in the following
manner; In the e, 7/ 2-plane (Fig. 7.16b) draw three verticals aa, 56, and cc,
having respectively, the abscissas €«, Ch, as shown. (If any of the meas-
ured strains were negative, the corresponding verticals would lie to the left
of the origin 0.) Locate the center C of the circle midway between the tv/o
verticals aa and cc as shown. If the intermediate vertical 66 lies to the left
of C, as assumed here, lay off on the vertical aa the positive ordinate
Z) 'D =* CF' and on the vertical cc, lay off the negative ordinate D\Di = CF\

*Eot additioiidi information on the strain rosette, see K. J. Bossart and G. A. Brewer,
“A Graphical Method of Rosette Analysis,” Prw: Sor. Exp. Stress Analysis Vol. IV,
No. I, p. 1.
 THE STRAIN ROSETTE m
F' being the point where bb iniersectti the e-axis. The line DDi, so ol)taincd,
is the diameter of the required Mohr's circle which can now be drawn. This
circle will cut the intermediate vertical bb at a point F such that CF will be
perpendicular to DDi, since by construction the triangles CDIX and FCF'
have been made congruent. Thus points i), F, and Di, having, respectively,
abscissa €«, and and being 90° apart in counterclockwise order in
e?,,

Fig. 7.16b, represent completely the strains asso(‘iated with the directions
oUf ob, and oc which are 45° apart in Fig. T.lOa. The principal strains are
€1 = OA and €2= OB. The direction of the principal strain ei makes the
angle <l>) - ^ Z DC A w'ith the direction on as shown in Fig. 7.16c.

Fig. 7.16

Knowing the principal strains ci and the principal stresses cri and
can be calculated from eqs. (*1.6); namely

CTl = — +—
(«i ^^e2)E
5—’ CTo = — +—
(eo n(:i)E

M
^ ^
(a)
^

1 M I

This assumes, of course, that the material has not been stressed beyond its

elastic limit.

KXAMPLi: 1 Data taken from a 45°-Rtrain rosette (see Fig. 7.17a) reads as follows,
.

fa = 750 microinches/in., €6 = — 110 microiiichcs/in., and Cc = 210 microinches/in.

Find the magnitudes and directions of prin<*ipal strains and the corresponding
principal stresses. Assume F = 30(10)® psi and m = 0.30.
SOLUTION. Beginning with the f, 7/2-plane, Fig. 7.1,7b, construct the verticals
oa, hh, cCy with abscissas Ca = 750* units, = —110 units, and = 210 units, as
shown. Locate center C on the f-axis midway between the verticals aa and cc.
Then since the vertical bb lies to the left of C, lay off the positive ordinate D'D =
CF' along the vertical aa which locates point D. Similarly, lay off the negative
ordinate D'lDi = CF' along the vertical cc to locate point Z>i. With C as a center
and DDi as a diameter, draw Mohr's circle as shown. This circle intersects the
ertical bb at F making CF perpendicular to DD^ as it should be. Scaling OA and
196 PLANE STRESS AND PLANE STRAIN

Fig. 7.17

OB from the diagram^ we find that the principal strains are ti = 1130 (10)~®,
•2 = —170(10)“*. The angle DC A is measured with a protractor and found to be
65®30'. Hence = 32®45', clockwise from oa to ol as shown in Fig. 7.17c. Sub-
stituting the scaled values of ti and €2 into eqs. (a) above, the corresponding
principal stresses are found to be (T\ = 35,700 psi and = 5900 psi. It will be
noted that although the strain €2 is negative, the corresponding stress o’ 2 is positive.

PROBLEMS
Referred to the directions shown in Fig. 7.1 6a, a 45®-strain rosette gives the
1.

following data: = 670 microinches/in.. = 330 microinches/in., €c = 150

microinches/in. Construct a Mohr's circle for this state of plane strain and find the
principal stresses tri and and the angle defining their directions with reference
to the direction oa. Use F = 30 (10)® psi and m « 0.30. Ans, (Ti = 23,800 psi:
cr 2 = 11,300 pai; 01 = -8®33'.

Solve Problem 1 if the measured strains are €« = 232, €&

2. = 123, Ct = —80,
allvalues being microinches per inch. Ans. ai =* 7030 psi; a 2 = —500 psi; 0i =
+8®23'.
3. Solve Problem 1 if the measured strains are €« = 815; 66 = —72, 6c = 165,
all values being microinches per inch. Ana. ai = 36,000 psi; (72 = 6(XX) psi; 0i -
~30W.
 CHAPTER VIII

DEFLECTION OF BEAMS

8.1 Differential Equation of the Elastic Line

In discussing pure pending of a prismatic bar in Art. 5.3, it was shown
by the f^quation
that the cur\ atiire of the neutral surface was given
1

p
_ ^
El
(a)

Thus, for pure b(*iiding, where M is constant along the length of the bar, its

axis bends in a circular arc. It is customary to call this curved axis of the
bar the elastic Line or the deflection carve.
In the case of a beam bent by transverse loads acting in a plane of
symmetry, the bending moment /!/ varies along the length of the Vieam, and
we represent this variation by a bending moment diagram. For such non-
uniform bending it is usually assumed that eq. (a) holds at each cross-
section. Thus the curvature 1/p is seen lo vary along the beam as the
bending moment varies and the elastic line
becomes a rather complicated curve, the shape
of which is defined by eq. (a). To express the
shape of this curve in rectangular coordinates,
we consider any portion of a bent beam as
shown in Fig. 8.1. Through any point 0 on
this elastic line, we take coordinate axes x and y
as shown, the a:-axis coinciding with the original
straight axis of the beam, positive to the right,
and the iy-axis positive downwards. At the ends
of any element ds of this elastic line, we construct
normals which intersect at C, thus defining the. Piq^ g,l
radius of curvature p of the element. Denoting
the angle between these two normals by dSj which is also the angle between

the tangents to the elastic line at the ends of the element, we have the
relationship ds = pdO, from which
cW 1
^ (b)
ds p
197
198 DEFLECTION OF BEAMS

For a beam which is bent within the elastic range of its material, the
elastic line'will usually be a very flat curve. Limiting the discussion to such
conditions, and referring to Fig. 8.1, we now introduce the approximations
dfs = dx and 9 = dyjdXj so that eq. (b) becomes

^dx^
^ I'
p
(c)

indicating that we consider only the absolute value of d^yjd.:^

Combining expressions (a) and (c), we have

ty =
dx^^
^ K
El
(d)

The sign in this expression must now be chosen such that it will be con-
sistaiit with the choice of coordinate axes in Fig, 8.1 and the definition of
positive bending moment
as that which produces curvature concave up-
wards For the coordinate axes as shown in Fig. 8.1 we see that
(see p. 90). ,

when the curvature is concave upwifrds, the slope dyjdx is algebraically

decreasing with x and hence d^yfdx' is negative Likewise, when the
curvature is concave downwards (negative bending moment), the slope
dy/dx is algebraically increasing with x and (Py/dx^ is positive. Thus d'^y/dx’^
is always opposite in sign to M
and w^e take expression (d) in the form

( 8 . 1)
dx* El

This is the differential tquafion of the elastic line for a beam subjected to
bending in a plane of symmetr3^ Its solution y = f{x) defines the shape of
the elastic line or the dtflectior\ carve as it is frequently called.
Equation (8.1) carries with it two important limitations resulting from
the assumptions made in its derivation. First, the moment-curvature re-
lationship (a), derived in Art. 5.3, assumes that stress is proportional to
strain, i.e., that Hookers law applies. Thus the equation is valid only for
beams that are not stressed beyond the elastic limit of their materials.
Second, since expression (c) assumes that the curvature is always small,
the equation is limited to the treatment of small deflections. Most beams
encountered in engineering practice will be well within the validity of these
two limitations. Furthermore, eq. (8.1), derived originally for pure bending
does not account for any deflection resulting from shear deformation of the
material. The additional deflections due to such deformation will be dis-
cussed in Art. 8.4. It will be shown there that for beams of ordinary pro-
portions, the deflections due to shear deformation are usually small and can
be neglected.
 iSQUATION OF ELASTIC LINE 199

If we twice differentiate eq. (8.1) with respect to i*, we obtain, with

reference to eqs. (5.1) and (5.2),

( 8 2)
.

dx^ dx
and
(W
w, (8.3)
dx
where V is the shear force at any cross-seetion and w is the intensity of
distributed load. Equation (8.3), for example, can be used to investigate
the deflection of a beam supported by an elastic foundation.
The application of eq. (8.1) to find the deflection curves for various
transverse loadings of prismatic beams will now be shown by several
examples.

EXAMPLE 1. A simply supported pris-

matic beam AB carries a uniformly dis-
tributed load of intensite w over its span I
as shown in Fig. 8,2. Develop the equa-
tion of the elastic line and find the maxi-
mum deflection 5 at the middle of the
span.
SOLUTION. Taking coordinate axes x and
y as shown, we have for the bending mo-
ment at any point x Fig. 8.2

wx^

and eq. (8.1) becomes

wlx WC‘‘
-y- + -y* (e)

Multiplying both sides by dx and integrating, wo obtain

dy _ w/x* ut’
’’ (0
(ix 4 fi

where C\ is an integration constant. To evaluate this constant, we note from

symmetry that when x = 1/2, dy/dx = 0. From this condition, we find

and eq. becomes

~ + -r+
(f)

wlx^ wx^ wB
(g)

Again multiplying both sides by dx and integrating,

wlx^ wx* wl^x

Ely " (h)
12

The integration constant C2 is found from the condition that y = 0 when x = 0.

200 DEFLECTION OF BEAMS
Thus C 2 = 0 and the required equation for the elastic line becomes

(i)

To find the maximum deflection at mid-span, we sot x — 1/2 in eq. (i) and obtain

3S4/t7

The maximum slope at the left end of the beam can be found by setting
X = 0 in eq. (g), which gives

(
\dx 2AEI
EXAMPLE 2. A prismatic cantilever beam A B carries a concentrated load P at its
free end B as shown in Fig. 8.3.
Find the equation of the deflection curve referred to
coordinate axes x and y as shown. Determine also the maximum deflection 6 and
maximum slope 6b at B.
SOLUTION. For axes as shown, A/^ = ~P{1 — x) and eq. (8.1) becomes

Multiplying both sides by dx and integrating, we obtain

wherein Ci must be taken equal to zero to satisfy the condition dy/dx = 0 when
j: = 0. Thus
Px‘^
EI^
dx
= Plx
2
' (j)

Again multiplying both sides by dx and integrating,

Plx^ Px^ ^
- T- +
,

Since ^ = 0 when x = 0, we have C 2 = 0 and the required expression for the

deflection curve becomes

_ Fx* / xN
(k)
^ 2E1\ 3/
- /, ^

Po

Fig. 8.4

Setting X - Im eq, (k), we obtain, for the deflection at

PP
b =
3E7
 EQUATION OF ELASTIC LINE 201

Likewise, setting x = Hn eq. (j), we find, for the slope at J?,

EXAMPLE 3. A simply supported prismatic beam AB carries- a concentrated load

P as shown in Fig. 8.4. Locate the point of maximum deflection on the elastic line
and find the value of this deflection.
SOLUTION. Choosing coordinate axes x and y as shown, we have for 0 < a: < a,

M. = y*,

while for a < X < I

Mr = — a; —P (x — a).

Substituting these expressions for bending moment into eq. (8.1), we obtain for the
two portions of the deflection curve, the following two differential equations

0 <x < a a < X < I

Pbx Pbx
El El -PPix- a)
dx^ r dx^

Successive integrations of these equations give

(1)
dx 21

Ely
Pbx^
— + Crx + Cj (ni)

(o)

where (7i, (72, Dx, D 2 are constants of integration. To find these four constants, we
^

have the following conditions:

1. At X = 0, y = 0. 1
2. At X = Z,
y = 0.

3. At I = a,
^ ^ at X = o.
dxdx
4. At X = a, y = y at x = a.

Using condition (1) in eq. (m), we find (72 *= 0. Using condition (3) in eqs. (1) and
(n), we find Cx = Dx. Using condition (4) in eqs. (m) and (o) and noting that

Cl = Dx while C = 0, we find that D 2 = 0.

2 Finally, using condition (2) in eq. (o)
and noting that a - Z — 6, we find

C. - ^ (P - 6>) = D,.

Using the constants as determined, eqs. (m) and ( 0 ) defining the two portions of the
elastic line of the beam become

Ely =
^ - x»)
(p)
I
E/p = ^ (X - o)* -I- (J*
- 6*)x - x«
j. (q)
202 DEFLECTION OF BEAMS
For a> b, the maximum deflection will occur in the left portion of the span, to
which eq. (p) applies. Setting the derivative of this expression equal to zero gives

- b\
X
3

which defines the abscissa of the point having a horizontal tangent and hence the
point ofmaximum deflection. Substituting this value of x into eq. (p), we find

Ph
//ill ft X - hy. (r)

KXAMPLE 4. A simpl}'' supported prismatic

beam AB is ac^ted upon by a couple Mq ap-
plied at an intermediate point C as shown
in Fig. 8.5. Derive the general equation for
th(‘ portion A C of the elastic line. Find alsf>

the deflection of point C and the slope at A .

SOLUTION. The reaction Ra — Mo/ and

the l>ending moments in the two portions of
Fig. S.f) the vspan become

0 < X < a a < X < I

M, ~
MqX If
Mr = nr
Miu

Substituting these ('xpressioiis into eq. (8.1), we obtain

il/ uX

~T + Mu.

Two integrations of each of th(‘se equatuins produce

El "b C\ (s)
dx 21

El y — —— \- C\X C 2 (t) Ely =

TlfoX^
1
A/ax2
^ b OiX +D 2. (v)

The integration constants are determined from the conditions

(1) 1/ = 0 at X = 0 I
(2) y = 0 at X = /

3) at X = a, ~
(IX dx
at X = a

4) at X = a, y = y f
&t X a.
 EQUATION OF ELASTIC LINE 203

vSubstituting these conditions into the above equations, we find

n Af ^ 0®* Ml Mta*
~ 3 21
'

C, = 0 D, =

With those values, oqs. (s) and (t) for the portion AC of the elastic line become

- 3(0* + x*) - (w)

^ 2/*],

Ely = ^[6af - 3a* - J* - 2/*]. (x)

6/

Setting X = a in oq. (x), we find for the deflection of point C

= ~m - 2a* - (*1. (y)

Setting j = 0 in oq. (w), wo find for the slope at A

», - - .W - 2/1. (z)

PROBLEMS
1. With reference to the (joordmabi axes x and y as shown in Fig. A, derive the
t*quation defining the deflection curve of a uniformly loaded cantilever beam. From
this,evaluate the deflection 5 at the free end. /Ins. 5 = wl*/HEI.
2. Repeat the solution of the previous problem for the case of a distributed load
the intensity of which increases uniformly from zero at the free -end to w at the
built-in end. /Ins. 8 = wl^/30EI.
Repeat the solution of Problem I for the case of a clockwise couple of moment
3.
Mo applied at the free end of the beam. Ans. 8 — MJ}/2EI,
4. V simply supported beam is acted upon by a counterclockwise couple of
moment Mb at the end B as shown in Fig. B. Derive the equation of the deflection

curv(i and find the maximum deflection. Am. i/ma* = 9^3 ^ ~

5. Repeat the solution of the preceding problem for the case of a distributed
vertical load, the intensity of which varies linearly from zero at A to lo at B. Am.
2/m a, = 0.00652 wl*/EI at x = 0,519/,

IT per unit length

1
-LILlU 1 1 U. M nUT *
1

^ -n
/

Fig. a Fig. B
204 DEFLECTION OF BEAMS

7,

6/ A simply supported beam AB carries a distributed load, the intensity of which

varies as shown in Fig. C. Derive the equation of the elastic line. Hint: Ikgin with

the fourth-order differential equation (8.3) instead of (8.1). Am. y = -“pj^sin -r-

simply supported beam AB consists of two wood planks having thicknesses

A
hi and laid one on top of the other and loaded as shown in Fig. D. Neglecting
/i 2 ,

friction between the planks, find the ratio of their maximum bending stresses ui and
O’ 2. Hint: Both planks have the same deflection curve: hence, the same curvature at
each section. Ans. a 02 = hi ‘.hi. i :

w per unit length

:sr
A B

Fig. D Fig. E

8. A cantilever beam AB of uniform flexural rigidity

is built-in at A and FA
carries a vertical load P At B As the beam
under increasing P, it
(Fig. E). deflects
comes in contact with a rigid circular foundation of large radius R as shown. Find
the distance x defining the position of point C where the beam breaks contact with

the foundation for a given value of P. Am. x = /

— El
PR
9. What kind of pressure distribution exists between the beam and the circular
foundation in the region AC in Fig. E?

8.2 The Moment-Area Method

We shall now discuss a semigraphical method of dealing with the pro>dem
of deflection of beams subjected to bending. Combining eqs. (a) and (b) of
the preceding article, we obtain

(a)
ds El
where dd is the angle subtended by an arc element ds and Af is the bending
moment to which the element is subjected. Then, as before, for flat deflec-
tion curves, wr may take ds ~ dx (see Fig. 8.1) and write

d9 « Mdx (b)
El
 MOMENT-AREA METHOD 205

This relationship can now be given a very simple graphical interpretation

with reference to the elastic line of thebeam and its bending moment
diagram. In Fig. 8.6, AH is any portion of the elastic line and AiBi is the
corresponding bending moment diagram. AO and BO are tangents at A
and B, intersecting at 0, and 6 is the angle betw^eon these tangents, as-
sumed small. The vertical distance B'B is the deflection of point B away
from the tangent at A, while A' is the deflection of A away from. the
tangent at B. All these quantities are understood to be very small.
Now let ds = djc be any element of the elastic line at the distance x from
B and note that the angle between tangents at its ends is dd. Then from
eq. (b), we conclude that this angle is equal to the area M dx of the shaded
strip of the bending moment diagram in Fig. 8.0 divided by El. Integrating
eq. (b) between A and £f, we obtain for the total angle B between tangents at
A and B

Since this integral represents the total area of the bending moment diagram
A}B} divided by El, we have the following conclusion:

Theorem 1. The angle 6 between tangents at any two points A and B on the
elastic line is equal to the total area of the corresponding portion of the bend--
ing moment diagram, divided by EL

I^t us now consider the deflection of point B relative to the tangent at A ,

i.e., the vertical distance Keeping in mind that the angles

B'B in Fig 8.6.

between these tangents are very small, we note from the figure that bending

Fig. 8.6 Fig. 8.7

206 DEFLECTION OF BEAMS

element ds contributes to this deflection by the amount

of the xd9. Then by
summation, the total distance B'B becomes

Since M dx is the area of the shaded strip of the bending moment diagram
and X is its distance from By we conclude that the right-hand side of eq. (8.5)
represents the statical moment with respect to B of the total bending moment
area between A and by EL Thus, we have the conclusion:
B, divided

Theorem II. Tlie deflection of B away from the tangent at A is equal to the

statical moment with respect

y
to By of the bending moment area between A and
By dwided by EL
This statical moment will be obtained simply as the product of the total
area of the bending moment diagram betw'een A and B multiplied by the
distance x to its centroid C (Fig. 8.6).
If there is an inflection point in the elastic line between A and B as
shown in Fig. 8.7, caution must be exercised in using the above theorems.
In such a case, the bending moment diagram divides itself into a positive
portion Aand a negativ^e portion />iBi, with centroids C\ and C2 re-
J)i ,

spectively. Then to find the angle B between tangents at A and B, we have

from eq. (8.4)
M dx Mdx '
(c)
Jf'
A ~W~ El
111 short, to obtain the angle 6 between tangents at A and B, we must take
the net area of the bending moment diagram between A and B, divided by

A/2- — -*

parabola>^
T" yrA
-i-l
^
E [j

^ J
^dflTI
"nn
'1^
1 L 1 L . »

Area = bh Areo = th/Z Area = 2AA/3

 MOMENT-AREA METHOD 207

EL Similarly, for the deflection of B away from the tangent at A, eq. (8.5)
becomes

^^Mxdx f^Mxdx (d)

A El Jn El
That is, the required deflection is obtained as the difi‘eren(*e between the

statical moments with respect to B of the positive and negative portions of

the bending moment diagram. This can also be expressed as follows:

6 = (e)

where and X2 are the centroidal distances shown in h'ig. 8.7.

xi
Since most of the bending moment areas with w'hich we have to deal are
simple rectangles, triangles, and parabolic segments, their areas and the
location of their centroids are easily found. A few of these are sum-
marized in Fig. 8.8 for ready reference. In the.se diagrams, C denotes the
(!entroid of the shaded area and F, the vertex of the bounding curve.
Applications of the moment-area method will now be illustrated by
several examples.

EXAMPLE 1 Determine the deflection b and the slope B at the

.
free end A of the
cantileverbeam AB loaded as shown in Fig. 8.9a.
SOLUTION. The bending moment diagram is shown in Fig. 8.9b. Since the tangent
to the elastic line at B coincides with the undeflected axis of the beam, the required
deflection 5 will be the deflection of A a>vay from the tangent at B. Thus, using
Theorem II, we have
2/
EL ^ Pl^
~ 3Ei (0
2EI^ 3

Likewise, the slojje at A is the angle between tangents at A and at B and from
Theorem 1, we have
Pli
$ = (g)
2EI

Fio. 8.9 Pig. 8.10

208 DEFLECTION OF BEAMS
EXAMPLE 2. A simply supported beam AB carries a concentrated load P at
pointD as shown in Fig. 8.10a. Find the deflection 5 of point D from the cord line
AB and the angle B between this cord line and the tangent at A
SOLUTION. The bending moment diagram is shown in Fig. 8.10b. The area of
this diagram is Pab/2 and the distance of its centroid C from B is }i(l + b) as shown.
Taking the statical moment of this area with respect to point B, we obtain the
deflection B'B of B away from the tangent at A Thus .

B'B =
Pab {l + b)

2FJ 3

Then noting from the figure that Ba ~ B'B I, we have

Pab
Ba
mi + (/ b). (h)

We sec also from the figure that th(‘ required deflection of point I) fioni the chord
line AB is

6 = qBa - 5', (1)

where 8' is the deflection of D away from the tangent at A, This deflection 8' can
be found by using Theorem 11 for the portion A iDioi the bending moment diagram.
The area of this is Pa^b/2l and its centroid is at the distance a/3 to the left of D.
Thus
^
5' — V (i)

Substitution of expressions (h) and (j) into eq. (i) gives

mr ^^ mi mi' (k)

EXAMPLE Locate the point of max-

3.
imum on the simple beam
deflection
discussed in the preceding example and
evaluate this deflection by the moment-
area method.
SOLUTION. The maximum deflection
occurs at point E (Fig. 8.11a) where
the tangent to the elastic line is hori-
zontal, i.e., parallel to the chord line
AB. Let X denote the distance of this
point from A. Then the angle between
tangents at A and E must be equal to
6a as already found in eq. (h) of the
preceding example. Thus the area of
that portion of the bending moment di-
agram between A and E (Fig. 8.11b)
must be such th^t Fig. 8.11

Pa6 Pbx^
A
mi + O' h)
21EI
 MOMENT-AREA METHOD 209

from which
la(l-hb)
(1 )

Then since 3^ ax is the deflection of A away from the tangent at E, we have

5max - by. (m)

2/£;/‘ 3 9^J3lEI
This seen to agree with eq. (r) on p. 202.
is

EXAMPLE4. A prismatic cantilever beam AB

carries a uniformly distributed
load over the portion b of its length as shown in Fig. 8.12a. Find the deflection S of
the free end A,
SOLUTION. The bending moment diagram is shown in Fig. 8.12b. Its area io6V6
and the position of its centroid C are found by reference to Fig. 8.8. From Theorem
1 1 the deflection 5 will be obtained by dividing by El the statical moment of this
area with respect to point Ai. Thus

wb‘^
6 (a -f ib). (n)

Fig. 8,12

EXAMPLE 5. A simply supported beam

AB is acted upon by a couple of moment
M applied at B as shown in Fig, 8.13a.
Locate the point of maximum deflection
E and find

SOLUTION. The bending moment dia-

gram is shown in Fig. 8.13b. The deflec-
tion BB' of point B away from the
tangent at A is obtained by taking the
statical moment of this diagram with
respect to Bj. Thus

^ ^
2EI 3
I m
6EI
Fig. 8.13
210 DEFLECTION OF BEAMS
Then noting from Fip. 8.13a that Sa = BB* /, we have

Ml
Oa (o)

The point of maximum E where the tangent to the elastic

defleetion is that point
line is horizontal. found by equating the shaded area of the
Hen(;e, the distance x is

bending moment diagram divided by El to Sa in eq. (o), i.e.,

Ml ^
(yEI “ 2lEl'

from which z — l/^jS. Then fimax, equal to the deflection of A away from the
tangent at E becomes

Ml ^
8 max (P)
6 El' 3 El

PROBLEMS
the moment ar(‘a method, find the slope 6 and the deflection 8 at tlu;
free end of a cantilever beam AB of length I loaded as shown in Fig. A. Am.
S = Ml/Ef, 8 = Mlh^2EL

Fig. a

Repeat the solution of the preceding problem if instead of the end moment M
the beam carries a uniformly distributed vertical load of intensity w over Its full
length Am. 6 — wP/ 6 EI; 8 = wl*/SEI.
1.

3: A
simply supported beam having a span I and a free overhang a carries a
uniformly distributed load of intensity w between the supports as shown in Fig. B.
Using the moment-area method, find the deflection 5i at the middle of the span AB.
Ans. 8 = 5wl*/3S^EI
\

4.' Referring again to the beam in Fig. B and using the moment-area method,
find the deflection 62 of the free end C of the overhang. Ans. 82 = wal^/2AEl.
 MOMENT-AREA METHOD 211

5.* A simply supported beam AB carries a distributed load the intensity of which
varies uniformly from zero at A to ii? at B (Fig. C). Using the moment-area method
find the slope Ba of the elastic line at A. Am. 6 a = 7wl^/3GOEI.

Fig. C

6; Referring again to the beam in Fig. C and using the moment-area method,
locate the point of 5max. Am. 6max = 0.00652 wl*/EI.
maximum deflection and find
7." simply supported beam of span I has an overhang AC on the left at the end
A
C of which a vertical load P is applied as shown in Fig. D. Using the moment-area
Fa-
method, find the deflection 5 of point C. 5 = (I a).
obi
+

8: A cantilever beam AB carries a sinusoidally distributed load having intensity

wa at the built-in end B (Fig. E} Find the deflection 5 at the free end A by the mo-
.

wd*
ment-area method. Am. b = 0.0479
,

El

8.3 Deflections by Superposition

The differential equation of the elastic line of a beam bent by transverse

loads (eq. 8.1) was derived in Art. 8.1. Since the equation is linear in y and
its derivatives, it follows that its solutions for various loading conditions
may be superimposed. This means, for example, that if we have yi(x) de-
fining the deflection curve of a cantilever beam carrying a concentrated
load P at its free end and y^ix) defining the deflection curve under uniforml^^
 212 DEFLECTION OF BEAMS

FORMULAS

DEFLECTION

BEAM

8.1

Table
DEFLECTIONS BY SUPERPOSITION 213

3x*)

lOi'j*

(71*
214 DEFLECTION OF BEAMS

distributed load, then the equation of the defleetion curve for the simul-
taneous action of both loadings is simply

y(i) = yi(x) -f y^ix). (a)

Specifically, the deflection at the free end under the load P is 6i = PI^/SEI.
Under the uniform load end is
of intensity w, the deflection at the free
62 = wl^/SEI. Hence, duo to the combined loading, the deflection at the
free end is simply

PP wl*
~ SEI SEl'
(b)

For ready referent^e, a table of deflections, slopes, and complete elastic

line equations for ten primary (jases of loading of cantilever and simply sup-
ported beams is presented in Table These various results have already
8. 1 .

been found either in the examples or problems of the two preceding articles.
Using this table and the method of superposition, it is possible to obtain
various required deflections and slopes forbeams and loadings comprising
various combinations of these primary cases. The idea of superposition can
be employed in a variety of ways and some of these will be illustrated by the
examples which follow.

EXAMPLE 1 . A uniform cantilever beam AB is loaded as shown in Fig. 8.14, Find

the deflection 5 of the free end B.
SOLUTION. Using Case 2 of Table 8.1, we have

As a particular case, if Pi = P2 = P, a — 1/3, and b — 21/3, this reduces to

^
2PP
^ ~ 9EI

Fio. 8.14 Fio. 8.15

 DEFLECTIONS BY SUPERPOSITION 215

EXAMPLE 2. A simply supported prismatic beam with span I and overhang a is

loaded as shown in Fig. 8.15a. Find the deflection at the end C of the overhang.
SOLUTION. Owing to the distributed load on the overhang, the portion A 5 of the
beam is in the condition of a simple beam subjected to a couple of moment Mb =
wa^/2 at the end B as shown in Fig. 8.15b. Hence using Case 9 of Table 8.1, we
have

_ _ MbI waH
^'*"3^7“

Due to this rotation of the tangent at B, the overhanging portion of the beam,
considered as absolutely rigid, would take the position BC" in Fig. 8.15a, and we see
that C' would have a deflection

waH
6i = adfi
Wl
In addition to this, the overhanging portion BC does bend due*to the distributed
loading and we ha\M% with reference to Case 3 of Table 8.1,

8E/

as the deflection C'C of point C awav from the tangent at B. Thus the total deflec-
tion of point C becomes
waH wl^
d = 8} 62
6^
EXAMPLE 3. A prismatic; cantilever beam of length I carries a uniform load of

intensity between x - a and x — I sls shown in Fig. 8.16a. Find the deflection 8
lo

at the free end of the beam,

SOLUTION. Consider the beam acted upon by just one element w dx of the dis-
tributed load as shown in Fig, 8.16b. Then from Case 2 of Table 8.1, wc' have

wx^ dx
d8 (3/ - X).
6EJ
216 DEFLECTION OF BEAMS
Now since each element of the distributed load between x = a and .r = / produces
a like increment of deflection at the free end, we obtain by summation

* w dx w
*-j. -jsr 517
If the load extends over the full length of the beam, a = 0 and this reduces to

which checks with Case 3 of Table 8.1, as it should.

EXAMPLE 4. A. simple beam AB of length I carries a uniformly distributed load of

intensity w
over a portion 6 of the span as shown in Fig. 8.17. Assuming b, a>
find the deflection 8c at the mid-point C of the span.
SOLUTION. We first consider just one element iwif of the load at the distance f
from B, Then using Case 7 of Table 8.1,

m— — P) = --- — 6 ^)

Setting X = //2 in this expression, we obtain

(
3/2 - 262).

For the particular case where 6 - //2, this becomes

/ bwV
_
'
1
\
2\mEIJ
as is to be expected from a consideration of Case 8 in Table 8.1.

w per unit length

 DEFLECTIONS BY SUPERPOSITION 217

EXAMPLE 5. A simply supported beam having a span I and overhangs of length

a at each end carries a uniformly distributed losifd of intensity w over its entire
length as shown in Fig. 8.18a. Find the ratio a/l in order that the tangents to the
elastic line at each end of the beam will be horizontal.
SOLUTION. Since both the beam and the loading are symmetrical about point C,
the tangent to the elastic line at C remains horizontal. Hence we may consider half
of the beam as a cantilever built in at C (Fig. 8.18b, c). Then from Case 3 of Table
8.1, the slope 6 at the free end due to the uniform loading alone (Fig. 8.18b) is
i

Likewise, due to the concentrated force Ra = w’(a + 1/2) acting alone, the slope 62
at the free end (Case 2 of Table 8.1) becomes

These slopes are of opposite sign; hence setting 61 — 62 we realize the required
condition of the problem. This yields the quadratic equation

-p a/ - = 0,

from which

Taking only the positive root as physically meaningful, we obtain

" = 0.366.

The standard meter stick in Paris is supported in this manner to insure that its two
end cross-sections will remain parallel.

PROBLEMS

1. The beam in Fig. A has uniform flexural rigidity El along its full length 3a.
Find the ratio P/Q to make dc = 0. Neglect weight of beam. Ans. P/Q = i.

Fig. a Fig. B
218 DEFLECTION OF BEAMS
2. The bar ABC in has uniform flexural rigidity El over its entire length
Fig. 13

21 and forms a Neglecting the weight of the bar and axial elonga-
right angle at B,
tion of J5C, find the vertical deflection of point A. Ans. 6^ = 2Pl^/3EI.
3. A prismatic beam ABC hinged at A and supported at B by a vertical wire BD
carries a load P = 100 lb at C as shown in Fig. C. Calculate the vertical deflection
of point C if the wire BD has such cross-sectional area A and modulus E that
AE = 3 (10)® lb, while for the bar AC the flexural rigidity is E/ = 3 (10)^ lb in.*
Ans, dc = 0.0443 in.
4. A steel shaft having flexural rigidity El from A to B and 2EI from B to C is
built-in at C and loaded at A as shown in Fig. D; Find the vertical deflection of
point A. Ans. 5 a = 3PZV16^^/-

Fig. C Fig. D

5. A simple beam A B carries three equal loads P, one at the middle of the span /
and one at each quarter point. Find the deflection 5 at the middle of the span.
Ans. 5 = 19P1V384E/.
6. A compound beam AE consisting of two identical portions AC and CE hinged
together at C is supported and loaded as shown in Fig. E. Find the vertical deflec-
tion 5e of point E. Ans. 5e - 4:Pa^/3EI.
7. Solve the previous problem if, instead of a concentrated load P at E, there is a
uniformly distributed vertical load of intensity iv between C and E. Ans. 5 b =
wa*/4EI,

Fig. E

8. A standard 81-23 steel beam 12 ft long is supported by a fulcrum at C and a

steel tie rod AD, 10 ft long and i in. in Calculate the deflection
diameti^r (Fig. F).
of point B P = 1 kip acting as shown. Ans. 5b = 0.211 in.
due to a vertical load
9. The beem shown in Fig. G has rigidly attached to it at point C an arm AC to
the free end A of which a vertical load P = 10 kips is applied. Compute the vertical
deflection of the end B of the beam if E = 30 (10)® psi and I ~ 339.2 in.® Ans.
5b = 0.0925 in., up.
 STRAIN ENERGY OF BENDING 210

Fig. F Fig. G

8.4 Strain Energy of Bending

Consider in Fig. 8.19a, a prismatic beam subjected to pure bending within

the elastic limit of the material. For such loading, the bending moment M
is constant along the length I of the beam and the elastic line is a circular
arc of curvature A//A7(see p. 197) and the angle 6 subtended by this arc is

6 = Ml (a)
El'

This represents a linear relationship between M and 6 as shown in Fig. 8.19b.

Hence we conclude that as the applied couples at the ends of the beam are
gradually increased from zero to any value M, the w’ork which they do is

represented by the shaded area OAB of the diagram in Fig. 8.19b. This
work, ecpial to the strain energy stored in the beam, has the magnitude

MB
U= 2 — f
(b)

which is analogous to the strain energy of torsion in a bar subjected to

twist [see eq. (d), p. 83].
Using expression (a), the strain energy (,b) itiay be written in either of the
following two forms:
MH EW 8 6)
.
(
220 DEFLECTION OF BEAMS

If we know’ the bending moment M, we use the first form to calculate the
strain energy. If we know the curvature, defined by 6/1, we use the second
form to calculate the strain energy.
It is scjnetimes iisc^ful to express the strain energy U in a bar subjected
to pure abending in terms of the maximum fiber stress (7,nax = Mc/I^ as
given by eq. (5.oa), p. 1 15. In the particular case of a beam of rectangular
cross-se(!tion of w'idth b and depth h, eq. (5.5a) gives M = ^bk^ o-mnx and
the first of ecjs. (8.6) becomes

u=
I
m (c)

This shows that for the saiiK' maximum liber stress, a bar in pure bending
can 8tor(‘ only f)ne-third as much strain energy as if it were in simple
tension.
In discussing the strain (uun-gy in a beam subjected to non-uniform
bending (Fig. 8.2()a), we shall neglect .strain energy of shear and (‘onsider
only .strain energy of Ixmding.* C'onsidering any tOement of the Ix^am of
length r/.r, the bending rnonumt Mx will b(‘ essentially coirstantover the
length of the element Hence e(|s. (8.6) apply to the element provided we
replace M by Mx, I by d.v, and 6 by (id = dx/p == {(lhj/dx^)(lx. Then
summing up such expressions for all elements of the beam, we obtain

= MxMx
IJ or (8.7)
i, 2t:i

To illustrate the use of the first of (‘qs. (8.7), ccmsider, for example, a
cantilever beam loaded as shown in Fig, 8.21a. The bending moment at
any cross-section destined by r is M =- Px and the first of e(ps. (8.7)
becomes

(d)
j„ 2E1 HE I'

If the beaih has a rectangular (To.ss-.section of width b and depth h,

the maximum bending stress at the built-in end is

ffmax - j
-

Expressed m tiuins of this maximum stress, eq. (d) becomes

= I
m (.)

•Strain energy due to shear will be small in comparison with that due to bending except
for short thickbeams. This (luestion is discussed further in the following article.
 Ffo. 8.21 Fig. 8.22

Comparing this with expression (c) above, we observe that for the same
maximum bending stress, the cantilever beam carries only one-third as
much strain (Miergy as the beam subjected to uniform bending.
To illustrate the use of the second of eqs. (8.7), (consider the simply
supported beam AB in Fig. 8.22. This beam is so loaded that its deflection
curve has the form of a half sine wave defined by the equation

= 8 sin
y
y • (f)

The mid-point deflection 6 being specified, it is desired to find the total

strain energy stored in the beam. Differentiating expression (f) twice with
respect to .r, we find

d^y biA . TTX

Substitution of this into the second of eqs. (8.7) gives

Fig. 8.23

EXAMPLE 1. Calculate the deflection d at the middle of a simply supported

beam AB, under the action of a load P as shown in Fig, 8.23.
prismatic
222 DEFLECTION OF BEAMS
SOLUTION. The l>ending moment at any cross-section defined by x is

= '
2

Then, using the first form of eq. (8.7), we obtain for the strain energy of bending in
the beam

U ~ (h)
8EI 96El'

Equating this strain energy to the work done by the applied load as it is gradually
increased from 0 to P during deflection:

PH\
2

=
PP
from whicli 5 (i)
48P/‘

EXAMPLE 2. A simply supported beam as shown in Fig. 8.24 is struck at its mid-
point C b}' a ball of weight W freely falling from a height h above the beam. Neglect-
ing the weight of the beam and assuming that it behaves elastically, find the total
deflection 6 that will be produced at point C.
SOLUTION. The gravity force W of the ball falling through the total distance
h -f 5 does work equal to

W(h-\-6). (j)

Now let P denote the force exerted by the ball on the beam in the extreme position
of maximum deflection 6. Then for this configuration the strain energy in the
deflected beam is

Pd
’
(k)
2

Equating this .strain energy (k) to the work (j), we obtain

y= + «),

from which P=
0
(A + 5).

Substituting this expression for P in eq, (i) above, we obtain

2Wl' h + 6
^ ~ 48P/ 5
( 1)

Noting that

represents the static deflection of the beam under the load eq. (/) can be written
in the form
5* - - 2W.t « 0. (m)
 STRAIN ENERGY OF BENDING 223

Taking only the positive root of this quadratic equation, we obtain, for the maxi-
mum deflection,
5 2^81- (n)

Fig, 8.24 Fig. 8.25

It is seen from this that the dynamic deflection 8 is always larger than the static?

deflection 5,t. Even when /i = 0, i.c., when the weight IF has no free fall but is

suddenly applied to the beam, eq. (n) gives 8 = 28nt. That is, the suddenly applied
load produces twice as much deflection as when gradually applied.
As another extreme, we may take h »
in which case the quantity Sgt® under

the radical maybe neglected and « = «.»+

EXAMPLE 3. The rim of a flywheel of weight W
and mean center-line radius r is
attached to a hub by four spokes as shown in Fig. 8.25 Each spoke has a rectangu-
lar cross-section of dimensions b X h and a length 1. The spokes are built in at the
hub and pinned at the rim. While the wheel rotates with constant angular velocity
0), the hub is suddenly locked. What maximum bending stress will be induced in
each spoke where it joins the hub? Neglect the weight of the spokes.
SOLUTION. During free rotation, the flywheel rim has kinetic energy

Zw* W ^
foi

When the hub suddenly locks, the rim continues to rotate through some angle 0
before coming to rest and the spokes bend enough to absorb the kinetic energy of
the rim in the form of strain energy of bending. Since each spoke is a cantilever
beam of rectangular cross-section loaded at its pinned end, eq. (e) applies and the
strain energy of bending in four spokes is

Equating kinetic energy (o) to strain energy (p),

I (bhl) ’
9 2E 9 2

3 jW E
from which (q)
224 DEFLECTION OF BEAMS
We see that this maximum stress is proportional to the initial circumferential speed
= ro? of the rim and inversely proportional to the square root of the volume of
material in the spokes.

PROBLEM.S

1. A wooden cantilever beam has length

6 ft and a circular cross-section of / =
diameter d = 6 in. 350 lb at the free end, what is the
If it carries a load P -
amount of strain energy stored? The modulus E = 1.2 (10)® psi. Ans. U = 99.8
in.-lb.
2. Repeat the solution to the preceding problem if the total load P = 350 lb is
uniformly distributed along the length of the beam. Ans. U - %o (99.8) = 15.0
in.-lb.
3 . Two cantilever beams of circular cross-section are identical in every respect
except that the diameters of their cross-sections are in the ratio di:d 2 = 5:6. What
is the ratio of their strain energies to U 2 when they are identically loaded?
Ans. Ui\U 2 = 2.075.
4. A simply supported wooden beam of rectangular cross-section and span
/ = 9 ft is struck at mid-span by a weight IT = 40 lb falling from a height h = 12 in.

Determine the required cross-sectional area A if the maximum bending stress is iK^t
to exceed 1000 psi and E = 1.5 (10)® psi. Ans. A = 120 sq in.
5. It is decided to give the beam in the preceding problem a cross-section 10 X 12

in. When the 12-in. dimension is vertical, the falling weight produces a maximum

deflection 5i. When the 10-in. dimension is vertical, the maximum deflection is 5 2 .

What is the ratio 5 1 82 ?: .4n.s. 3i:52 ~ 0.834.

6 . cantilever beam (61-12.5) projects 8 ft beyond the built-in end. A weight
A
W= 100 lb falling through a height /i = 6 in. strikes the free end of the beam.
Neglecting the weight of the beam, calculate the maximum bending stress produced
by the impact. crmax = 22.800 psi.
7. A solid circular rotor of weight
— 386 lb and radius r = 10 in. is supported
W
between two parallel beams each having a § X 6-in. rectangular cross-section as
shown in Fig. A, Initially the rotor turns at a constant angular speed of 30 rpm.
If the bearings C siiddcniy freeze so that the rotation is almost instantly stopped,
what dynamical reactions will be produced at A and B1 Ans. Ra — —Rb ~ 2930 lb.

6"

A • OC
t
1 1

k 3' 4^ 3 -
*1

Fig, A

8.5 Deflections Due to Shearing Strain

In the pieceding discussions of beam deflections, only the deformation

due to bending stresst^s was considered. In all cases of non-uniform bending
where there is a shear force at each cross-section, there will be some addi-
 SHEARING STRAIN 225

tional deflection of the beam due to deformation associated with the shear-
ing stresses. Due to the fact that these shear stresses are not uniformly
distributed from top to bottom of the beam, cross-sections become warped
as shown which shows curvature of the beam axis produced by
in Fig. 8.26,
shear deformation alone. Small rectangular elements on the neutral axis
become rhombuses, the vertical edges of which remain vertical during de-
formation. Thus the slope of the deflection (iurve due to shear at any cross-
section is simply equal to the shear strain 7 at the neutral axis. Denoting
then by yi the deflection due to shear, we write the following expression for
the slope of the elastic line

T max kV / \
^ If ^ AG

where V xj is the average shear stress in the (jross-section and k is a

numerical factor by which this average shear stress must he multiplied to
obtain the maximum shear stress at the neutral axis. For a n^et.angular
cross-section, k = 3/2 (see eq. 5.9, p. 128) and for a circular cross-section,
k = 4/3 (see eq. (g), p. 132). Owing to the fact that the flanges of an /-beam
carry very little vertical shear, the factor k may have a value of 2 or 3 in
this case depending on the proportions of the /-section.

Assuming some continuous distribution of transverse load of intensity Wx

on the beam, the shear force Vx will be a continuous function of x which may
be differentiated with respect to x. Then the curvature c>f the elastic line
becomes

dx^ AG dx AG
Analogous to eq. (8.1), this represents the differential equation of the elastic
line of the beam for shear deflection alone. The deflections calculated from
eq. (8.8) can be added directly to those calculated from eq. (8.1) to obtain
the total deflections of a beam subjected to non-uniform bending. Usually,
the deflections due to shear will be found small compared with those due to
226 DEFLECTION OF BEAMS
bending and the former may be disregarded. However, in the case of
short deep beams, the shear deflections may become significant.
To illustrate, let us consider the case of a simple beam uniformly loaded
as shown in Fig. 8.27. In this case, the shear force at any section defined
by X is

wl
F, =

and eq. (8.8) becomes

(Pyi
“
___
kw
(b)
“IG’
Integrating once

dyi _
~ kwx
(c)
dx

and again

(d)

w per unit length

I l-l-l 1 1 - 1 - I 1 I I I l-I 1 1

wjt
2 2

Fig. 8.27

Using the end conditions: ?/j = 0 at j: = 0, = 0 at x = we find

Ca = 0 and C\ = kv}l/2AG. Then eq. (d) becomes
kwlx /. x\
2/1 = __
(e)
2AG\ ~l/

Taking x = Z/2, this gives for the maximum deflection at the middle of
the span
k wP
“ 8 AG'
(f)

Introducing the notation I =

Ar^y where r is the radius of gyration of the
cross-section eq. (f) can be written in the form

5wl*
384£/ h(0'i] (g)
 SHEARING STRAIN 227

Taking, as a particular case, a steel beam of rectangular cross-section, we

have k = ;V2, K/(i = 2.5, and r* = h^/l2 for which eq. (g) becomes

This shows that for a depth-span ratio h/l = 1/10, the maximum deflection
due to shear is approximately 3 per cent of that due to bending. As the
depth-span ratio increases, the deflection due to shear becomes more
important
In the case of an /-beam. Fig. 8.28, eq. (5.7'), p. 129,
gives
b

Tmax- 4/'‘’
^ -7<[2\^4 8 J’

from which

Fig. 8.28

Then for a 24WF120 section (see Table B.2 of Appendix B) we lind A =

35.29 in.'^ / - 3035 in.-*, r = 10.15 in., t = 0.556 in., b = 12.09 in., h =
24.31 in., h\ == 22.45 in,, and eq. (h) gives k = 2.91. With this value of fc,

and with E/G = 2.5, eq. (g) becomes

This shows that for a span I = 12 ft, the maximum deflection due to
shear would be approximately 34 per cent of that due to bending, which is

by no means negligible.
In the case of a simply supported beam that carries a concentrated load P
at mid-span (Fig, 8.29), the shear force V=
dtF!2 is constant in magnitude
but changes sign at the middle of the beam. When V is constant, we see
from eq. (a) that the slope of the deflection curye due to shear deformation
is also constant. Thus it may be concluded in this case that the deflection
curve due to shear is represented by two straight lines AC and CB SiS shown
in Fig. 8.29c, where the constant slope of the line AC is
dyi _ _ kP
dx ^ AG 2AG

Hence the maximum shear deflection at C is simply

I dwi kPl
228 DEFLECTION OF BEAxMS

Again introducing the notation / = /Ir-, this may be written in the form

Taking again, as a particular case, a steel beam of rectangular cross-section,

we have h = 3/2, E/G — 2.5, and r- = E^/12, and ecp (j) hK^comes

^
48£/
It has been assumed throughout the foregoing discussion that all cross-

sections of the beam are free to warp as shown in Fig. 8.26. In the case of a
uniformly loaded beam, this condition is approximately satisfied. The shear
force at the middle of such a beam is zero and there will be no w^arping of
this cross-section. The warping increases gradually with the shear force as
we proceed along thebeam to the left or to the right of the middle. The
condition of synim(‘try of deformation with refereiu'e to the middle section
is therefore satisfied .

/I

Consider now the case of the simple beam with a concentrated load at the
middle, Fig. 8.29. From the condition of symmetry, the middle cross-
section must remain a plane section. At the same time, adjacent sections to
the left or to the right of the middle section carry a shear force equal to P/2
and warping However, from the
of these sections should take place.
condition of continuity of deformation, there can be no abrupt change
from a plane middle section to warped adjacent sections. There must be a
gradual increase in warping as we proceed along the beam in either direction
from the middle. Thus only at some finite distance from the middle can the
 SHEARING STRAIN 229

'warping attain the value compatible with the shear force on the section.
From this discussion, it must Yye concluded that in the neighborhood of the
middle section the normal stress distribution cannot be that predicted by
the elementary theory of bending (see p. 118). Warping will be partially
prevented and the additional deflection duo to shear will be somewhat less
than that predicted by eq. (j) above. A more detailed investigation of this
question shows that for a simply supported steel beam of rec^tangular cross-
section with a concentrated load at the middle, the maximum defle(‘tion due
to both bending and shear is

Additional deflection of a transversely loaded beam due to shear de-

formation may also be calculated by using the strain energy of shear. To
illustrate, let us consider again thF (‘ase of the simply supported beam of
rectangular cross-section loaded at the middle as shown in Fig. 8.29a. In
such case, the shear force at any cross-section between A and C is simply
V = P/2, and from eq. (5.8), p. 128, the shear stress on any element
situated at the distance y from the neutral surface is

Now using the first of eqs. (4.12), p. 82, the strain energy of shear in this one
element of volume bdydx is

dU = -
3 ^ 7? (^ 2/’)
bdydx, (m)

and the total strain energy of shear in the entire beam becomes

U~ ^ Jo ^ ~ SOGI (n)
J-1 S2GP (4
Equating this strain energy to the work P6i/2 done by the applied load as
it gradually increases from zero to the final value P through the deflection 5i
of its point of application C, we obtain
Plh^
^ (o)
40G/
which can arbitrarily be written in the form

Taking the particular case of a steel beam for which E/G == 2.5, this
becomes
PI* „
230 DEFLECTION OF BEAMS
Comparing eqs. (p') (jO, we see that there is a discrepancy between
and
the two results. Eq. (p'), obtained by considering the strain energy of shear,
gives a smaller deflection than eq. (j'), obtained by assuming that the de-
flection curve (Fig. 8.29c) was two straight lines having slopes equal to the
shear strain at the neutral surface of the beam. To explain this dis-
crepancy, we recall that in arriving at eq. (j'), we assumed that all cross-
sections were free to warp. However, as already observed, this assumption
violates the condition of symmetry at the middle of the beam. Since eq.
(p'), obtained on the basis of strain energy of shear, does not involve this
assumption, it must be considered as the better result. In tact this can be
seen to be the case b}’^ comparing both eqs. (j') and (p') with the more
rigorous i(‘sult represented by eq. (k)
 CHAPTER IX

STATICALLY INDETERMINATE BEAMS

9.1 Method of Superposition

Consider in Fig. 9.1 a beam .4 supported by a hinge at A and rollers at B

and C, and subjected to applied loads Pi and Po acting in a plane of symme-
try of the beam. Due to the action of these loads^ reactions will be induced
at the points of support and a free-body diagram of the beam will be as
shown in Fig. 9.1b. The reactions Yb and Fc at B and C will be vertical
because the rollers can offer no horizontal restraint. The reaction at
which can have any direction in the plane of the figure, is represented by its
horizontal and vertical components Xa and Va^
Considering now the static equilibrium of this free body, we have the
general case of a system of forces in one plane. For such a system there are
three equations of equilibrium, namely:

= 0 2F, = 0 2M» - 0. (a)

Since these three equations are insufficient to determine four unknowns, the
beam is said to be statically indeterminate. There is no way, from the stand-
point of the statics of a rigid body, to determine the reactions at the
supports.

Fig. 9,1

231
232 STATICALLY INDETERMINATE BEAMS

If the roller at C is taken away, we arc left with a simple hea.m AB

supported at its (mds. Since such supports arc sufficient for complete
constraint of the beani in one plane, we sa}'^ that the intermediate support at
C is redundant. If the roller at C is removed, we have a
B instead of that at
simple beam AC with overhang, which again is completely constrained in
one plane. Thus either the support at B or the one at C may be considered
as redundant. In general, any supports of a bi^am in excess of those both
nec^essary and sufficient for its complete (*.onstraint. in the plane of loading
are said io be redundant supports.
Let us now choose the roller at C as the redundant support. Then re-

moving this support, we consider the simple beam AB as our primary

system. In l^’ig. 11.2, we study the deflection curves of this beam under two
si^parate conditions of loading. In Fig. l).2a, only the applied forces Pi and
/*2 an* acting am] h* represiMits the downward deflection of point C when
(he support there is removed. In Fig. 9.21), only a vertical force Fc is acting
at point (\ and represents the upward deflection of this point. Using
th(^ methods of the preceding chapter, the directions and 5"c can be
calculated without difficulty. Superimposing the two statics of loading in
Fig. 9.2a and b, we conclude that the net deflection of point C is Sc = 5'r —
iV'r. Xt)w since the support at C allows no deflection, the true v^alue of 3c
IS zero and we com hide (hat

5',. ^ 5"^ = 0. ib)

When (he detlecnnn.s S\ and 5", ari' expr(\ssiMl in (errus of Pi, P2 and Yr,
,

condition d)) driuieb ( lu' (lue value ol V(‘ As soon as Yc is known, the

t*' per unit length

4
p rrairLgzmrroj n T'l g

'

u
I
— — i
'

Fig. 9.3
 METHOD OF SUPERPOSITION 23.S

remaining reactions Ya and Yu may be ff)und from oqs. (a).

To illustrate the pr()(*edure more specifically, we consider the beam
shown in Fig. D.o. Choosing the support C as redundant, we obtain as our
primary system a simple b(‘am AB of span 21. Under the action of the
uniform load, th(‘ downward deflection of point from C'as(‘ 8 of Table 8.1,
p. 213, will he

_ ow(2]y
_ iywV
" '
- 384W “ 'IW
(' due
Likewise^ from C'ase (i of Table 8. » , the iii)ward defk'ction of point t-o a
vertical force Yc will }»c

AHEI {\KT

Thus condition (b) above gives

24EI
from which

With thi^ \’aliu‘ of }'c, eqs (a) giv(‘ Y i

“ Yn — I'wl and the react ioijs are

completely determined.
The ab()ve example may be look(^d ai in a somewhat, did'eriait way i)y

noting that the tangent to the elasti(*, line at E remains horizontal because
of symiiu^try conditions. Thus we may regard (‘ac.h half of the b(\am avS a
cantilever built-in ai C and supported at Us free end. Consideiing the
portion C/?, \ve have then a so-calh'd ‘'proppt'd cantilever
^
as shown in

Fig. 9.4. When this beam deflects under the applied loading, rotation of the
tangent to the elastic line at C is prevented and a counterclockwise moment
Me is developed at C as shown. Removing this constraining moment as the
redundant reaction, we obtain a simple beam CB as our primary system.
Then under the action of the distributed load, the tangent at C rotates
through the angle 0\ as shown in Fig. 9.4b, while under the action of the
redundant moment Mr. at it rotates in the opposite direction through an

angle Fig. 9.4c. The true value of Mr is found by making O'c

- 0,

since this realizes the condition of a built-in end. From Case 8 of Table 8. J

we find
wV
e'e
^7
From Case 9, we find
Mri.
234 STATICALLY INDETERMINATE BEAMS

Equating these values in accordance with the condition that B'c ~ = 0,

we obtain

M. = ~ (c)

The positive value indicates that the assumed direction of Me in Fig. 9.4a
v/as correct; according to our sign convention for bending moment, it
represents a negative bending moment at the built-in end C of the beam.
Returning now to Fig. 9.4a and equating to zero the algebraic sum of
moments of all forces around point C, we obtain

YbI-‘^-\-Mc = 0 .

Substituting expression (c) for Mcj this gives

as found before.

Fig. 9.4 Fig. 9.5

As a final illustration of the method of superposition in dealing with

statically indeterminate beams, consider the beam in Fig. 9.5 having
 METHOD OF SUPERPOSITION 235

both ends built-in and carrying a cojiceiitrated load P, Removing the resist-
ance against rotation at each end of the beam as a redundant constraint, we
obtain, as our primary system, a simply supported beam AB. In this case
there are two redundant reactions, Ma
and we say that the beam
and M b,
with built-in ends is twice statically indeterminate. To find Ma and
consider first the primary system under the action of the applied load P,
Fig. 9.5h. Then from Case 7 of Table 8.1, we have, for the angles of
rotation of the end tangents

,
P6(/^ - h^) Pab{2L - b)

‘-—am- mi
Considering now the primary system under the action of end moments Ma
and Mb as shown in Fig. 9.5c, and using Case 9 of Table 8.1, we have

MaI Mul MaI

//
d A and 6" = MbI I'
3EI QEI 3EI 6E
Finally, since the net angle of rotation at each end of the lieam must be
zero, we have O'a — 6" = 0 and d’n — d"u = 0. Thus
Pb{P - 62) MaI MbI _ .
61£/ 3EI &E1
Pabi2l - b) MbI MJ _ f.
"•
61E1 3EI 6Ei

These two equations yield for Ma and Mb the following values:

PaV Po*6
P
and Mb = P
(d)

Having the end moments Ma and M b, the bending moment diagram for
the beam is easily constructed by superimposing moment
the bending
diagrams for the loadings shown in Figs. 9.5b and c. This superposition
is shown in Fig. 9.5d, where the triangle abc represents the bending moment

diagram for the load P and the trapezoid end moments.

abde, that for the
By plotting the negative ordinates due to and Mb oa the same side
Ma
of the base line ab as the positive ordinates due to P, the overlapping
portions automatically cancel and the net bending moment is given by the
ordinates of the shaded portions of the diagram.
Assuming b < a, the numerically largest negative bending moment will
be Afs as given by the last of eqs. (d). This has its greatest value when
6 = l/Z for which iMslmnic = 4Pf/27. The positive bending moment at
the- point of application of the load P, by reference to Fig. 9.5d is

Me 12
^_Ma + Mb _ Pab.
21
236 STATICALLY INDETERMINATE BEAMS

This has its greatest value when a = 6 = //2, so that \Mc\max = Pl/S,
which is seen to be slightly less than |A//?|max-

EXAMPLE 1. Two wood beams cut from the same timber are arranged as shown
in Fig. 9.6a, the free end of the cantilever DC being supported at the middle of the
simple beam AB. Both beams are horizontal and at right angles to one another.
Find the vertical deflection 5c at the point of contact due to a vertical load P applied
to theend of the cantilever as shown.
SOLUTION. Lt't X represent the magnitude of the for(*,e with which the two beams

interact as the point of contact. the cantilever has a net load P —

Then at its X
end, Fig. 9.6b, and the deflection of this point, from Case 1 of Table 8.1, be(;omes

Fig. 9.6 Fig. 9.7

At the same time the mid-point of {tie beam A B is subjected to a downward force X
and deflects by the amount

^
48F/
by Case 6 of Teble 8.1. Since these two deflections must be equal, we obtain
P_X^X^ '

3 3 6
 METHOD OF SUPERPOSITION 237

from which

2
X F.
3

Substituting this back into either of the deflection expressions gives

6'e
= = 5. = PF
9EI
This assumes, of course, that there was contact but no pressure between the two
beams at C prior to the application of the load P.
EXAMPLE 2. A cantilever beam ABC is built-in at A propped at B, and carries a
load P at the free end as shown in Fig. 9.7. Find the magnitude of the redundant
reaction at B.
soLUTioisr. Replacing the support at B by the reactive force Rh, we obtain as our
primary system, the cantilever beam AC loaded at B and C as shown in Fig. 9.7b.
From Case 1 of Table 8.1, the downw'ard deflection of point B due to the load P is
PP
[3 (1 T a) - /].

Likewise, from Cast' 1, th(‘ upward deflection of point B due to the force Rjj is

5 "/. =

Since the net deflection d'a — = 9, we obtain

from which

Using this value for Rn, we find for the positive bt'nding moment at the built-in
end of the beam

Ma = RJ -P(a + l) = ~
PROBLEMS
ir Find the redundant reaction Rb for the beam AB uniformly loaded as shown
in Fig. A by taking the cantilever built-in at A and free at B as the primary system.
Ans, Rb — iwL

Fig, a Fig. B
238 STATICALLY INDETERMINATE BEAMS

2: A propped cantilver is subjected to the action of a moment a at the end A

as shown in Fig. B. What reactive moment Mb will be induced at the built-in end
B? Am. Mb =
3. Referring again to Fig. B, find the maximum deflection 6m ax due to the applied

couple Ma> 6max = at a; = 1/3.

£1 til
An L-shaped bar ABC of uniform El is built-in at A and supported by a roller
4.

at B as shown in Fig. C. Find the reactions at A and B due to the action of a horizon-
tal load P at C. Am. Xa = Xb = 3Pa/2l; Ya = P; M a = Pa/2.

5. Referring again to Fig. C, find the horizontal deflection of point C. Neglect

Pa^
axial tension in A Am. dc =
\2EI
(31 + 4a).

A prismatic beam on three supports is loaded as shown in Fig. D. Find the

rNfflndant reaction Rc at C Arts. Rc = wP/Sa (I + a).
^TyFor the beam AD supported and loaded as shown in Fig. E, find the reactions
aVxhe four points of support if a = 1/3. Am. Ra - Rd ~ 9ic//44; Rb = Rc =
31?/?//44.

A(*
1
tr/uni1 length

DC
^ g| 1 1
M'/'unit length

i M M M Tic

k j — iL-— X
Fig. D Fig. E
8. A propped cantilever beam AB carries externally applied couples of moments
M and 2M at its third-points as shown in Fig. F. Calculate the redundant moment
reaction Mb at the built-in end. Mb = 0.
9. A propped cantilever beam with overhang is loaded as shown in Fig. G.
Construct shear and bending moment diagrams and select a suitable WF-beair. if
the working stress = 20,000 psi. Ans. 12WF27.

M M
A B px:::::-

J
VrfvTT 1

MH 1.-^

Fig.
3

F
1
j3 1
1" » •

20*

Fig.
O*

G
fc 1- ft* r 1

3'
 METHOD OF SUPERPOSITION 239

10, A beam A BC built-in at C rests at B on the mid-point of a simply

cantilever
cross-beam as shown in Fig. H. The cross-beam is 10 ft long and hks the
supix)rte<l
same El as the cantilever. When P = 0, the two beams are just in contact at B.
Calculate the magnitude of the force X
transmitted to the lower beam. Ana. = X
0.6P.
11. A
uniform beam rests on three equally spaced supports and carries a uniform
load of intensity over its entire length. During loading the middle support settles
vertically a distance 6. Evaluate the magnitude R of the middle reaction if each

Fig. H Fig. f

12. A beam with built-in ends carries a triangular distribution of load as shown in
'wl^ xvL^
Fig. 1. Find the end moment reactions M and Mb- Ans. M a = Mb =
13, A propped cantilever is loaded as shown in Fig. J. Select a suitable stahdard
1 -section for the beam if the working stnsss = 18,000 psi. Arw. 81-18.4,
14, Referring to Fig. .1, the support at A is replaced by a vertical steel tie-rod
8 ft long an<l 0.20 sq in. in cross-section. If the brnsion in this rod is zero before any
load is applied to the bi‘am, wdiat t^msion S will it carry after the beam is loaded as
shown*'* Arts. S = 4940 lb.
15. Two cantilever beams A B and DC arc arranged as shown in Fig. K; they have
the same El. What kind of loading will the upper beam exert on the lower one if a
vertical load P is applied at B?

D
,

Fig. J Fig. K

9.2 Theorem of Three Moments

In the case of a uniform continuous beam on many supports (Fig. 9.8a),
one support is usually considered as an immovable hinge while all the others
are treated as rollers. In such an arrangement, each intermediate support
represents just one redundant constraint and hence the continuous beam
is as many times statically indeterminate as there are intermediate supports.
240 STATICALLY INDETERMINATE BEAMS

Thus the beam in P'ig. 9.8a is five times statically indeterminate. In such
case, we might choose the five intermediate reactions J?i, /?2 Rb as ,
•
. ,

the redundants and then set the deflections 5i, 62 Ss of their points ,
.

of application equal to zero. These five equations together with the three
equations of statics for the entire beam would be sufficient to determine
the eight unknown reaction components. This, however, turns out to
be a cumbersome and involved set of equations. Instead, it is much simpler
to cut the beam at each intermediate support and introduce the bending
moments Ml, 2 M
Ms, over the supports as the redundants. In this
, .
,

way, the primary system becomes six simple beams, each carrying its
own externally applied loads together with two redundant moments at
its two ends. By dealing with these simple beams two at a time, the
complexity of the problem will be greatly reduced.
In Fig. 9.8b, we any two adjacent spans with supports n — 1
consider ,

n, and + 1 . Let and

denote those two span lengths and let
Zn^-l

Mn~i, Mn, and denote the three redundant bending moments at

these three supports. Whether these bending momenhs are positive or
negative will depend upon the conditions of external loading. We will
assume that they are positive as shown in the figure. Then if calculation
produces for any one of them a negative value, this will automatically
indicate a negative bending moment.

idm n t I I i I M |5

(0)
 THEOREM OF THREE MOMENTS 241

The some bending of the beam

loadings shown in Fig. 9.8b produce
in and Zn+i and we indicate rotation of the end tangents at the
spans In

support n by B' n for the left span and by S'* n for the right span. These
angles of rotation are considered to be positive when they agree in direction
with the end moments Mn- Therefore, as a condition of continuity of
slope of the uncut elastic line over the support n, we have

- -B\. (a)

This is the key to the sohition of the problem.

To express B'n terms of the applied loads and end moments
and B'^ n in
for each span separately, we use the moment-area method discussed in
Art. 8.2. The bending moment diagrams hjr the end moments are the
triangles shown in Fig. 9.8c. The bending moment diagrams for the applied
loads are represented by the shaded areas in the same figure. We denote
these areas by An and AnJt-u respectively, and their centroids by Cn and
Cn+i, the positions of which are defined by a„, bn, and i, bn+i, as shown.

Now considering the left span In and using superposition, we find

~
^ fXn I
n-lln ,
AnCtn
3 £/ OEI InEl'

Considering the right span we have

rJ'n+l jl/ n4-1^n4-l A n-f
B\ I

^
.

(C)
SEI 6EI In^iEI

Substituting (b) and (c) into eq. (a), we get

Mn-dn + 2Mn{L + h+l) + = - ^4^ - (Q.l)

This is called the three-moment equation; it can be written once for each
intermediate support of the continuous beam. Then from this system
of ^
simultaneous equations the bending moments at the intermediate
supports can be found.
Throughout the foregoing discussion, it has been assumed that the tw^o
extreme ends of the continuous beam were simply supported. If one or
both of these ends should be built-in, the number of redundancies will
exceed the number of intermediate supports. In such case, for each built-in
end, an additional equation expressing the condition that no rotation can
take place at such a support will be available. Suppose, for example,
that the extreme left end of the beam is built-in. Then referring to Fig.
9.8b And taking n = 1 ,
we have
Moil Aibi
242 STATICALLY INDETERMINATE BEAMS

where do is the angle of rotation of the tangent at the left support. Setting
this equal to zero, we obtain

Having found the bending moments at all of the supports of a con-

tinuous beam, there is no difficulty about finding the reactions. Taking
again any two adjacent spans, as shown in Fig. 9.8b, let R'n be the simple
beam reaction at n due to loads in the span In and the reaction at
n due to loads in the span In+i. In addition, there will be reactions produced
by the end moments M
n_i, il/n, and Afn+i. Taking the directions of these

moments as shown in Fig. 9.8b, the total additional reaction at the support
n will be
il/n— 1 n n -j- A/n^-1

In-^l

the first part coming from the span /,* and the second part from the span
/n^i. Adding to this, the reactions R'n and due to applied loads, we
obtain for the total reaction at any intermediate support

Afn-1 - Mn — il/ n + M n-i- 1

Rn = R'n + R"n +
f1

Having he reactions and the bending moments at all supports, the shear
t

force and bending moment diagram for the continuous beam may be
constructed. Applications of eqs. (9.1), (9.2) and (9.3) to seve'ral specific
cases of continuous beams will be illustrated in the following examples.

EXAMPLE 1. Construct bending moment and shearing force diagrams for the
continuous three-span beam under uniform load of intensity w as shown in Fig. 9.9a.
I'hc three spans are equal and the beam is of constant cross-section.
SOLUTION. For a simple beam under uniform load, the bending moment diagram
is a parabola with maximum ordinate tc/V8. The area of the parabolic segment is

^ 2, wl^ wl^

and its ccmtroid is at raid-span, so that a — b = 1/2. Considering now the first two
spans on the left and noting that Mo = 0, eq. (9.1) becomes

0 -I- (2Z) -H M 2I -
wP - —
wl^
•
(e)

From conditions of symmetry, it is evident that Mi = A/ 2 ,

and eq. (e) gives
Ml *= —wiy 10 — M 2- The complete bending moment diagram as shown in Fig.
9.9b can now be drawn.
From eq. (9.3) the reaction at support 1 becomes

wl wl wl 1 Iwl
 THEOREM OF THREE MOMENTS 243

w per unit length

rTTTM II TTHTTI Tl 1 1 M 1 1

Ao jL, 1^2 -A- 3

• JL
- 1 —
(a)

(b)

Fig. 9.9

L From symmetry, we conclude that = /?i while = Ri. Then since Rq 4-

f /2i + i22 + ^3 = Ro = Ri - Awl/ 10. Having the reactions, the shear diagram
^i^onstructed as shown in Fig. 9.9c.
EXAMPLE 2. Construct bending moment and shearing force diagrams for the
^ continuous three-span beam loaded as shown in Fig. 9.10a.
SOLUTION. For a simple beam with a concentrated load P at the middle, the
bending moment diagram is a triangle with maximum ordinate P^/4. The area of
this triangle is

4*2“ s'
and its centroid is at mid-span so that a — h — 1/2.

Writing eq. (9.1), once for the first two spans on the left and again for the last two
spans on the right, we have

0 -h 2iVi(2/) + = 0,

+ 2 ^ 2(20 + 0 = 0 --^'
These simultaneous r*q nations are easily solved for the b(‘nding moments M\ and
Ml and we find

n
Ml + 40 ' " 10
*
244 STATICALLY INDETERMINATE BEAMS

The corresponding bending moment diagram for the entire ix^am is sliown in Fig.
9.10b. The maximum bending moment occurs under the load P and has the value

1 2
/,..x - ^ 20“ 5*

Tsing c(|. (9.3), the reactions Rx and Ri

at the intermediate supports are found as
follows

p p n op
^ = « + «- 40 - 40 -T0-~r
p p p p 2Q
=
«^ 0 + 2+40+10 + Io
= W-
Then with the entire beam as a free body,
we find = 4- -P/40 and /?3 = -f-lfiP/40,
The shearing force diagram may now be
constructed as shown in Fig. 9.10c.
EXAMPLE 3. Construct bending mo-
ment and shearing force diagrams for the
continuous beam shown in Fig. 9.11a,
which overhangs the right-hand support
and carries a load P at the free end.
SOLUTION. Writing eq. (9.1) for the two
adjacent spans 01 and 12, wo have

Md 4- 2Mx(2l) ^ MJ = 0. (g)
 THEOREM OF THREE MOMENTS 245

Then since the left hand support is built-in and there are no external loads on the
firstspan, eq. (9.2) gives Mn = -Af i/2. It must be noted also that Mi = P^/2 —
since the beam freely t>verhangs the last support on the right. Substituting these
values into eq. (g), we find i M
= -\~Pl/7. Then Mo = —Pi/14 and 2 = —Pi/2. M
The corresponding bending moment diagram is shown in Fig. 9.11b. From statics,
the reactions are now found to be Po = +3P/ 14, Pi = — i2P/14, P = 2 4'23P/14,
and the shear diagram is as shown in Fig. 9.11c.

PROBLEMS
1. Find the bending moments Mi and Mi at the supports 1 and 2 of the three-

span continuous beam loaded as shown in Fig. A. Ans. = — 22Pi/405, Mi — M 1

-32PZ/405.
2. A uniform three-span continuous beam with overhanging ends carries a
uniformly distributed load as shown in Fig. B. Find the ratio a/l in order to make
the bending moments at the three supports all equal. Arts, a/l = l/\6 = 0.408.

j)er unit length

0 3
^
1

T5r -JT TCI inttr JILCLmil

V- -4-- J- u a J /- --*j« ^
Fig. A Fig. B

3. Referring again to the* beam in Fig. H, find the ratio a/l in order to make the
reactions at the three supports equal. Ans. a/l = 0.44.
4. A continuous beam with overhang at the right hand end is loaded as
tw’o-spaii
shown Calculate the reactions at the three points of support.
in Fig. C. Ana.
Po = 750 lb, up; Pi = 2500 lb, down; P 2 = 750 lb, up.
5. A two-span eoiitinuous beam built-in at the left end is loaded as shown in

Fig. D. Calculate the bending moments at the three supports, Ans. Mo = +5140
ft.-lb; Ml = -10,286 ft. -lb; Mi = 0.

lOOOlb/ft

Fig. D

6. The right-hand span of a continuous beam

carries a uniform load of intensity
w over half its Find the bending moments at the two
length as shown in Fig. E.
intermediate supports. Ans. M\ - IwP/W/^t Mi = — 28u;ZV^0-
7. Construct bending moment and shear force diagrams for the continuous

beam shown in Fig. F. Ana. Mi — +4080 ft-lh; Po = P3 “ 10,680 lb;

Mi^
= ^2 = -34801b.
246 STATICALLY INDETERMINATE BEAMS

Fig. E Fig. F

8. Calculate the bending moments and reactions at each support of the con-
tinuous beam in Fig. G. Ans, Mx = -1.54 kip-ft; 2 = -3.74 kip-ft; = M
-1.65kip-ft; /?o = -0.386 kips; if? = +2.69 kips /? = +6.22 kips;
1 +3.75 ;

kips; Ri = —0.275 kips.

9. Calculate the bending moment Mo and the reaction Ro at the built-in end of
the continuous beam shown in Fig. H. Ans, Mo = —5280 ft. -lb; Ro = 3080 lb.

A* 9*

Fig. G Fig. H

9.3 The Theorem of Casti^liano

In preceding chapters, general expressions for the strain energy stored
in an elastic bar subjected to tension, torsion, or bending have been
developed. This concept of elastic strain energy can be very useful in
the study of deflections of various points of a structure under load. Con-
sider, for example, a prismatic bar under simple tension as shown in Fig.
2.14a, p, 40. The strain energy in this case is, from eq. (2.4),
FH
U= (a)
2AE
By taking the derivative of this expression with respect to the applied
load P, we obtain
^
dP AE
~ A
•

Thus the derivative of the strain energy with respect to the applied load
P gives the deflection of its point of application in the direction of the
load.
Again, in the case of a shaft of circular cross-section subjected to torsion
as shown in Fig. 4.12, p. 82, the strain energy is, from eq. (4.13'),

(b)
 THEOKEM OE C\STlGElk^O 2M

The derivative of this expression with respect to the applied torque T

becomes

dU_Tl_.
dT GJ
which is the angle of twist of one end of the shaft with respect to the
other. If we interpret the pair of statically balanced torques at the ends
and the angle of twist between these
of the shaft as a ‘^generalized force*'
two ends as the “corresponding displacement,” we conclude again that
the “displacement” is given by the first derivative of the strain energy
with respect to the “force.”
Finally, for a cantilever beam bent by a transverse load P at the free
end, Fig. 8.21a, p. 221, the strain energy of bending is

P^l^
U= (c)
i]EI

The derivative of this expression with respect xo P is

dU ^ PP ^ ,

dP dEI *

which is s(^en to be the deflection of the end of the beam in the direction
of the applied load P.
Each of the foregoing cases is simply an
examples of a general theorem regarding strain
energy which is called the theorem of Castigliano. /?

We shall now proceed with a general derivation

of this theorem. In Fig. 9.12, let AB be any
elastic body or structure completely constrained
in space and subjected to applied forces Pi,
P2, P.3, .... If the material follows Hooke's
law and the deformations are small, the Fig. 9.12
displacements of the points of loading will
usually be linear functions of the loads, i.e., the principle of superposition
will hold.* In such cases, the strain energy stored in the loaded system will
be equal to the work done by the applied forces and independent of the
order in which they are applied. If, for example, the forces are applied
simultaneously and gradually increased in the same proportion, then the
work done will be

t/ =i (P,5, + P-A + PaS, + • •

), (9.4)

where 61, 62, ... are the deflections of points 1, 2, 3, . in the directions

*For e.xceptional caaes, see Timoshenko and Young Theory of Structures, p. 222.
248 STATICALLY INDETERMINATE BEAMS

of the corresponding forces Pi, P2, Ps, . It must be clearly understood

here that 5i, ^2, ^3, may not be the total deflections of the points 1 ^ 2,

3, . . but, in each case, only that component of the total deflection in

the direction of the corresponding force. From our assumption that the
displacements 6j, 62, ... are linear functions of the forces Pi, P2, .
,

we conclude that if we substitute for 61, 62, •

,
their expressions in terms
of Pi, P2, .
.
,
eci. (9.4) becomes a homogeneous quadratic function of
the forces.
Assume now that after the loads Pi have all been applied and the strain
energy in the system (9.4), we increase any load Pn
is U, as given by eq.
by the amount dPn. This will cause a slight change in the deformation
of the body and the stored strain energy will be increased slightly. This
increase inav be expressed as the rate of change of 1/ with respect to Pn
times the change made in P„. In this way, the new amount of strain energ\'
becomes

• dp„. (d)

Since ihe final strain energy does not depend upon the order in which
the forces are applied, we will now assume that dPn is applied first and
afterwards the forces Pi, 1\ P3, .... When dPn is applied first, it produces
only infinitesimal deflections and the corresponding strain energy, of
second order, can be ignored. Applying now the forces Pi, P2, P3, ,

it must be seen that their effect on the structure will be unaltered by the
presence of dP„ and they will store the same strain energy If as before.
However, during the application of Pi, P2, Pt^j the; force dPn, already .
,

applied earlier, will ride through th<' displacement 5n produced by Pi,

P2, P3, In so doing, it produ(‘es additional work of the amount
dPn and the final strain energy in this case will be
*

U + dPn •
5n. (e)

Equating this expression for final strain energy to expression (d) obtained
before, we have

u + di\d„ = +
from which

5n = ^ •
(9.5)

This states the theorem of Castigliano in general form. For any elastic
system which olxjys the law of superposition, the partial derivative of the
 THEOREM OF CASTIGLIANO 249

strain energy, represented as a quadratic fimction of the forces, with respect

to any one of these forces, gives the corresponding component of displacement
of the point of application of that force. The terms “force’' and “displace-
ment” must be taken in the generalized sense. If the “force” is a couple,
the corresponding displacement will be an angle of rotation in the same
sense as the couple. If it is a true force, the corresponding displacement
will be a linear deflection in the direction of that force. It must also l>e

appreciated that we speak only of statically independent forces, to any

one of which we can give an arbitrary increment 6Pn without affecting
any of the other forces. This means that reactions determined by statics
cannot be considered as independent forces.

EXAMPLE 1. A simply supported beam with overhang is loaded as shown in

Fig. 9.13 h.Using the theorem of Castigliano, find the ^TTtical deflection of point C.
SOLUTION. The bending moment diagram for the beam is shown in Fig. 9.13b.
Betwe(*n .i and B, a gen(*ral expression for bending moment at any distance x to
the right of A is

= - Fax (f)
I

Between B and C, the bending moment at any distance .r to the left of C is

M. = -Px. (g)

Using the first of eqs. (8.7), p. 220, the strain energy of bending in the bt^am becomes

U ~
f‘ PVx\
p-^aV ,
rPVdx
pi P^a\, , ,
(h)
Jo Jo-

Then from eq. (9.5)

,dU_ Pa^
dP~ m ,, , ,

EXAMPLE 2. The axis of a cantilever ring, built-in at B and loaded at the fre6
end A, forms a horizontal quarter circular arc of radius R, Fig. 9.14. Find the
250 STATICALLY INDETERMINATE BEAMS
vertical deflection d of the free end A, assuming the ring to have a circular cross-
section the diameter of which is small compared with the radius R of
center line. its
SOLUTION. Such a curved cantilever subjected to torsion as well as bending.
is

At any cross-section Z), defined by the angle d, we see that the bending moment is

Me - PR sin d, (i)

while the twisting moment on the same cross-section is

Te = PR (I — cos 6). (j)

The total strain energy due to the combined bending and torsion of the ring is

Jo 2EI Jo 2GJ
Then by eq. (9.5)

where, from eqs. (i) and (j),

dMe
R sin (i')
dP

— R (1 — cos ^). (i')

dP
Substituting (i), (j), (i'), (j')i into eq. (k) and performing the indicated integrations,
we find

,
irPR» ,
(3t - 8) PR^
4^/ 4GJ

EXAMPLE 3.A simple truss composed of two bars each of length I carries a
vertical load P at joint A as shown in Fig. 9.15. Find the horizontal and vertical
 THEOREM OF CASTIGLIANO 251

components of the total deflection S of point A. The bars are of the same material,
AB having a cross-sectional area A and AC, a cross-sectional area Ai = 2A.
SOLUTION. From statics, we see that the tensile force in the bar AB and the
compressive force in the bar .1(7 are each equal to P. Hence the strain energy in the
system is

rr
^
- II -H- =
2AiE
^
AAE
Then, for the vertical component of the deflection of A we have
,

dP ^ 2AE
""

To find the horizontal component of the deflection of joint A, we introduce at A

a fictitious horizontal force Q with respect to which we can differentiate the strain
energy expression. With this force acting in addition to P, the strain energy
becomes

^ «?/V3 + py‘l (Q/V3 - P)H

2AE 2(2A)E

Then from eq. (9.5)

a
(<3 +
dQ 2AE
For the case in v/hioh we are interested, Q = 0 and the liorizonta! component of
deflection becomes
PI

2^Js AE

PROBLEMS

1. Calculate the vertical deflection 5 of the free end of a cantilever beam under
uniform load as shown in Fig. A, by using tl'«e Castigliano theoreni. Hint: Introduce
a fictitious vertical force P at A as showm by the dotted vector. Ans. 6 = wl^/SEl.
2. Referring again to the cantilever beam in Fig. A, find the angle of rotation 6
of the free end tangent by using the Castigliano theorem. Hint; Apply a fictional

couple A/ at A. Ans. 6 =- wl^/6EI.

3. A cantilever beam AB has a solid circular cross-section as shown in Fig. B.
A vertical load P is applied to the end of a bracket rigidly attached at the free end B.

Fig. a Fig. B
252 STATICALLY INDETERMINATE BEAMS
Neglecting bending of the bracket, but considering both bending and torsion of the
PaH Pl^
-
beam, find the vertical deflection of point C. Am, 8 = -777 + 77
-

y/ (jJ onjJ
^A cantilever ring AB
having a circular axis of radius R is built-in at B and
carries a vertical load P its free end A as shown in Fig. C.
at Considering only
bending of the ring, find the vertical deflection of point A, Arw. 8v — irPR^/^EI.
Referring again to the cantilever ring in Fig. C, find the horizontal deflection
8h of the free end A. Ans. 8h = PR^/2EI.
semicircular arch ring supported in its own vertical plane carries a vertical
load P at its crown C as show^ in Fig. D. Considering only the strain energy of
bending, find the vertical deflection of point C. Ans. 8 = 0.178 PR^/EI.
7. Referring again to the arch ring in Fig. D, find the horizontal deflection of the
roller at B due to the load P applied at the crown. Ans. 8 b = PR^/2EI.
8 . A simple truss with pinned joints is supported and loaded as shown in Fig. E.
Each bar that is in compression has a cross-sectional area of 5 sq in. and each bar
that is in tension has a cross-sectional area of 2 sq in. The load P = 20,000 lb.
Calculate the vertical deflection of the joint C. Ans. 8c = 0.0747 in.
9. Referring again to the truss in Fig. E, calculate the horizontal deflection of the
roller at B due to the action of the load P at C. Ans. 8 b = 0.0427 in.

9.4 Applications of Castigliano^s Theorem to Statically Indeter-

minate Problems

The theorem of Castigliano very useful in the treatment of statically

is

indeterminate problems. Consider, for example, the case of a continuous

beam having several redundant supports. Denoting by X, F, Z, . . . the
statically indeterminate reactions, the strain energy in the beam will be
a function of these forces and the displacements of their points of applica-
tion will be obtained as dUldXy dUfdY, dU jdZ, .... However, these
displacements are known to be zero in accordance with the conditions
all

of constraint. Hence, we have

(9.6a)

There will always be as many of these equations as there are redundant

reactions so that the reactions can be found.
If the bending moments over the supports are taken as the redundant
 CASTTGLIANO’S THEOREM APPLIED 263

quantities in the case of a continuous beam, then the strain energy can
be expressed as a function of these bending moments Mi, Afo, Ma,
In such case the partial derivatives dU/dMi, dl'/dAU, dU/dM^, will
represent the relative rotations between tangents lo the elastic line on the
two sides of each support. However, from continuity of the elastic line
over each support, we know that these relative rotations are all zero.
Hence, agam

dU dU dU =
= 0 ,
= 0, 0 ,. (9.6b)
dM, dAh dMg
In general, to find the redundant forces in a statically indeterminate
system, we remove the redundant constraints and replace them by the
corresponding forces. I'hen expressing the strain energy of the system
in terms of the forces and applying the Castigliano theorem, we obtain
from which the redundant forces can be calculated.
eqs. (9.6)
To the (‘ase of a propped cantilever beam
illustrate, let us reconsider
under uniform load, Fig. 9.16. This is a system having one statically
indeterminate reaction. Choosing the vertical reaction X at B as the
redundant force, the l)endir>g moment at any point along the beam will be

M. = Xr - ~
wx^
(a)

and the strain energy is

U p MMr '

0 2EI
which is a function of X, since is a function of X. Thus, diffei’entiatiug
under the integral sign
dM,
M.. dx
dX
(b)
dX I El
wherein, from eq. (a),

di\U
(c)
dX

Fig. 9.16
254 STATICALLY INDETERMINATE BEAMS

Substituting expressions (a) and (e) into eq. (b),

IVrfonning the indicated integration, we obtain

3 8'
fi'omwhich X = Iwl, as obtained previously on p. 233.
Choosing the restraining couple Ma at the built-in end of the beam as
the redundant rea(*1ion, the bending moment at any cross-s(^ction becomes

= wlx MaX
-2
r “ T-
3'hen since the angle of rotation of the tangent at A is known to be zero,
the Casligliano theorem gives

.. UAJx ,

dU _
(e)
SMa Jo El

wherein, from eq. (d),

dM X __ X
Wa~ (f)

Substituting expressions (d) and (f) into e(| (e).

Performing the indicated integration, we find Ma = wP/S, as previously

o^^ined on p. 234.

^ EXAMPLE 1. Three bars each of length / and pinned at theii ends are arranged in

a vertical plane as shown in Fig. 11.17. The vertical bar has a cross-sectional area A
and each inclined bar has cross-sectional area A i. A vertical load P acts at joint C
and it is desired to find the ratio A \/ of cross-sectional areas to make the tension in
DC numerically equal to the compressive forces in AC and BC.
SOLUTION. Lot X represent the tensile force in DC, chosen as the redundant bar.
Then the compreKSsive force in each inclined bar is {P — X)/y[2. Thus the strain
energy of the system becomes

XH {P-X)H
2AF. 2AiE
In this case, the end D of the vertical bar must have a displacement equal to zero.
 CASTIGLIANO'S THEOREM APPLIED 256

Hence, from the Castigliano theorem,

dU XI (P - X)l
dX^AE AiE

from which X (g)

The statement of the problem requires that

(h)

Eliminating X between eqs. (g) and (h), we find A i I = V2.

EXAMPLE 2. Two wood beams of identical cross-section are supported at their
ends and cross at their mid-points as shown in Fig. 9.1H. When unloaded, they are
just in contact at C. What interactive force X will exist betwTcn the two beams at
C when a vertical load P is applied to the upper beam as shown?
SOLUTION. I'he net downward load on the beam AB is P — X; that on the beam
DE is X. The total strain energy in a simple beam loaded at the middle by a force
Q is, from eq. (h), p. 222,

Q2/>
U=
96E/

Thus the strain energy in the two beams becomes

(p - xyp
U=
9fyEI
+ 96EI
(i)
250 STATICALLY INDETERMINATE BEAMS

Regarding tlie pair of interactive forces A" as a generalized force and observing that
the corresponding disj)la,c('ment is the relative ’ displacement between the mid-
points of the two beams, which is zero because they remain in contact at C, we have
by the Castigliano theorem

^ _~ _
(IX
(P -
4S/s7
xw a;^ “
4SA7

PIS
from which A -

EXAMPLE 3. To reduce deflection, a vsimply supported wood beam AB, loaded at

the middle, is trussed by steel cables Al) and Bl) and a post Cl) arranged as shown
in Fig. 9.19. Neglecting axial shortening of both th(^ beam and the post, find the
compressive force A^ induced in the post. 'Lhe beam has flexural rigidity El and the
cables hava* cross-sectional area A and modulus of elasticity E\.
i

soi.UTioN. Th(^ net downward load on th(‘ middle of the beam is Q — P — X and^
the tension in eacdi cable is ^Si = A/2 sin a. Henc(‘. ni'glecting strain t'niTgy of
compression in the beam and post, the strain miergy of the system is

Substituting for Q and and noting that h — ]l sec «, this becomes

_ XH s(>e g
9GA7 8AiFisin^a
Now shorteming of the post is th(' gmieralized displac.ernent corresponding to the
compressive forces X
an<l we have specified that this is to be neglected, i.e., that it
is to be taken as zero. Henc(», the Castigliano theorem giv(\s

(P - X)P A7 sec g
^
(IX 4HEf 4 1iEiSin''^a

=
P
from which A^
12 sec a El
sin* a A \Ei
Taking a — 20° and EI/AiEiP — 0.01, this gives A" = 0.478P. In such case the
deflection of the beam will be reduced by approximately 50 per cent.

Fig. 9.19
 CASTIGLIANO’S THEOREM APPLIED 257

PROBLEMS
For a uniformly loaded beam AB with built-in ends, dcU^rmine the end
1.

moments by using the theorem of Castigliano, .4ns. M = = —wl^/\2.

2. A continuous beam of two equal spans / is uuiforml}' loaded over its entire
length. Find the magnitude X of the middle reaction by using the Castigliano
Theorem. Ans. X = 5ieZ/4.
3. Repeat the preceding problem if the beam carries a con(;('ntratod load P at the
middle of each span instead of the uniform load. Ans. A" — llP/'8.
4. A continuous beam of two equal spans I has unyielding supports at its ends
while the middle support is a coil spring having a spring constant k. The be^im
carries a uniform load of intensity w over its full length. What value should the
spring constant k have in order that all three reactions will bo equal*^ Solve by
^ 48E/
usmg the Castigliano theorem, Ans. k -
,

uniform continuous bar ABCD is built-in at A and laterally supported at B

A
as shownin Fig. A. Find the horizontal reactive for(‘e X at B due to the action of a
vertical load P at D as show’n. Neglect the effect of direct compression in the
vertical portion of the bar. Ans. X = ZPa/l.

6. two-hinged semicircular arch ring is loaded at its crown point C as shown in

A
Fig. B. Using the theorem of Castigliano, find the horizontal cornp()n(mt A" of the
reaction at A or B. Consider only strain energy of biuiding of the ring. Ans. X =
P/n.

cross-sections 3 X 4 in. and 4 X in. Find the magnitude X of the force trans-
mitted through the roller at C. Ans. X = 1346 lb.
258 STATICALLY INDETERMINATE BEAMS
8. A cantilever beam AB is built-in at B and supported at A by
an inclined guy
wire as shown in Fig. D. Find the tensile force N guy wire due to the
iftduced in the
applied load P. The guy wire has cross-sectional area A and modulus of elasticity
\

2P
Am. S = =
S^J3EI
^
AiEil^

9.5 Limit Analysis of Statically Indeterminate Beams

In all the foregoing di^eussious of statically indeterminate beams, it

has been assunied that the beam behaves elastically throughout. It is

also of interest to investigate the behavior of such beams under loads
which induce bending of the beam, as discussed in Art. 6.1 Consider,
plastic .

for example, a propped cantilever steel beam subjected to a load P as

shown in Fig. 9.20a. For any value of this load less than the elastic limit-
value Pe, the bending moment diagram will be as shown in Fig. 9.20b.
From this diagram, it is seen that points B and D are pointe of largest
bending moment. If P is increased beyond the value P®, a plastic hinge

will begin to form first at Z), since the elastic moment Md is numerically
larger than M b. After this plastic hinge is fully developed at D, the beam
behaves as a statically determinate system represented by a cantilever
beam DB acted upon at D by the load P and a constant moment Mp.
 LIMIT ANALYSIS 259

As the load P is further increased, the negative bending moment Mb

continues to grow until it also attains the limiting value Mp. When this
condition is reached, the team will not resist any further increase in load
and becomes a mechanism with hinges at A ,
Z), ai»d B, which will collapse
freely without further increase in load. The value of the load required
to produce this condition represents the limit load Pl for the beam. This
condition is represented in Fig. 9,20c.
To calculate the limit load, it is not necessary to trace the behavior
from beginning to end as outlined above. We simply assume plastic
hinges at B and at £), each offering constant bending moment Mp, and
then solve for the value of required to maintain equilibrium. Referring
to Fig. 9.20c, the reaction at A in the limit condition will be

p PlI) Mp

Then the bending moincnt at D becomes

M = —
Phab Mpa
— = Mp,
Md = itAa
j
^

from which
MAI + g) .
(a)
a(Z - a)
''

since b — I — a. Taking theworking load Pu = Pi/n, we attain

safe
a factor of safety n against complete collapse.
It will be noted from expression (a) that the limit load Pl is a function
of the ratio a /I defining the position of the load on the span. To find the
most critical position, i.e., the ratio a/l to make Pl a minimum, we set
dPLida = 0. This gives
a2 + 2la -P
from which
? = ^- 1 = 0.414.

Substituting this value of a/l into eq. (a), we obtain (Pl) min * 5.84 Mp/l
Another method for calculating the limit load Pl for the mechanism
in Fig. 9.20c is to use the principle of virtual work. We take the system in
the limit condition, i.e,, B and D
as a* mechanism with' plastic hinges at
(black circles) and a simple hinge at A (white circle) as shown in Fig. 9.21.
The load Pl and the plastic moments at B and D we consider as aciwe
fortes on this movable system. Then a virtual displacement of the sysTern
can be defined by a small vertical displacement By of the hinge D. At the
same time the portion AD will rotate by the angle By /a about the im-
260 STATICALLY INDETERMINATE BEAMS

movable hinge A and the portion BD will rotate by the angle hy/h about
the immovable hinge J5, as shown. During this virtual displacement,
the load Pi, produces positive virtual work Pi^y while the plastic moment
Mh at B produces negative virtual work Mpby/h and those at D, negative
work Mp {by /a + by/b). As a condition of equilibrium of the system, the
algebraic sum of these virtual works must be zero. Hence

Pl by - Mp by/b - Mp(by/a + by/b) = 0,

from which, on canceling

0
+ Mf.
o
(b)

Substituting b — I — a, this gives the .same result as expressed by eq. (a)

above.
Limit analysis of statically indeterminate beams as illustrated above
is often much simpler than elastic analysis, since by such procedure the
problem is always reduced to a statically determinate one. It is also
gaining favor among structural engineers because it leads to greater
ec'onomy in design. In dealing with more complicated statically indeter-
minate beams and frames, the question of selecting, in advance, the proper
positions of all plastics hinges is Sometimes
not always a simple one.
there will be more than one possible mode of collapse for the system and
each of these must be investigated separately to find the one that leads
to the lowest limit load. These questions belong more prop)erly to special
books on plastic analysis and limit design and will not be discussed here.*

EXAMPLE 1. A two-span continuous beam simply supported at A and Rand

built-in at 0 carries a uniformly distributed load of intensity w over its entire length
(Fig. 9.22a). The beam has a 12WF50 wide-flange section for which the elastic
section modulus Z = 64.7 in ^ Calculate the limit intensity wl of the load for com-
plete collapse of the system.
SOLUTION. As already worked out on p. 142, the plastic section modulus for this
beam is •

Z/. ^ I.IOZ - 1.10 X 64.7 = 71.2 in.’‘

Then assuming a yield stress cTv.p = 40,000 psi, the corresponding plastic bending
moment becomes

Mp = 40,000 = 237,000 ft.-lb.

In the completely elastic condition, the bending moment diagram for the beam
will be as shewn in Fig. 9.22b. From this, we see that plastic hinges are likely to

•See, for example, Plastic Design in Steel, Am. Inst, of Steel Construction, 1959.
 LIMIT ANALYSIS 261

form first at the supports C and

B. After these hinges form, the system, on further
increase in load, behaves as two simply supported beams AB and jBC, until another
plastic hinge forms, either at D or at E. Let us assume that D will become the next
plastic hinge. Then the right-hand span becomes a mechanism and the limit load
has been reached. At such time, we have for equilibrium

Md = - Mp = Mp.

from which

imp 16 X 237,000
Wl 26,400 Ib/ft.
12 X 12

With a plastic hinge at B, the reaction at A will be wl/2 — M p/I and the shear
any section distance x to the right of A becomes
force at

V
wl
—
Mp
-WX.

Then to find the section of maximum bending moment in the span AB, we set this
expression equal to zero and obtain

(c)
2 xvl

w lbs per ft

1
i ! 1 M i M 114 i ; I 1 M 1 1 M UTM; Yy
St-

* b : -12'
1

(a)
262 STATICALLY INDETERMINATE BEAMS
The corresponding bending moment at E is
wx^
Me T (d)

Substituting the value of x from eq. (c) into eq. (d), we obtain

Me
8 2 V
Taking w — 26,400 lb per ft, / = 8 ft, and Mp = 237,000 ft-lb, this gives Me =
109,300 ft-lb. .Since this is less than Mp = 237,000, we conclude that the right-
hand span does collapse first and that Wl = 26,400 Ib/ft is the limit load intensity
for the continuous beam.

PROBLEMS
1. A two-span continuous beam carries a uniformly distributed load as shown in
Fig. A. If Mp is the plastic moment for the cross-section, find the intensity wi of
the limit load. Ana. = ll.G^Mp/P.
2. The continuous beam in Fig. A has a rectangular cross-section 2 in. wide by
4 in. deep, and / = 6 ft. The material is structural steel w ith a yield stress try p =
40,000 psi. What is the numerical value of ic/,? Ans. wl — 8640 Ib/ft.
3. Calculate the limit value Pl for the statically indeterminate beam shown in
Fig. B, if the plastic moment Mp for the section is given. Ans, Pl = ^Mp/l.

w per unit length

TT TTTT
-Jt-
h-
Fig. a Fig. B

4. Calculate the limiting value Pl for the beam in Fig. B if the load at the end of
the overhang is P/6 instead of P/3. Am. Pl = 1.2Mp/l.
5. If the statically indeterminate beam in Fig. B carries a load P D bls shown
and a load aP at C, what is the value of a to make the limit value of P a maximum
and what is this maximum value? Ans, Pl =» SMp/l when a = i.
6. A solid steel shaft 8 ft long and 6 in. in diameter is buiit-in at A and supported
at P by a steel tie rod of diameter d, Fig. C, The yield stress for the steel (both

Fig. C
 LIMIT ANALYSIS 263

beam and tic rod) is a-y p = 36,000 pai. Calculate the minimum diameter d of the
tie rod required to develop the full collapse load Fr. for the beam and evaluate this
load. Ans. Pl = 81,000 lb with d > in.
7. Calculate the collapse load Pl for the building frame shown in Fig. D if each

vertical member has a plastic moment The ends of these columns are built-m
top and bottom and AB is to be considered as completely rigid. Ans, Pl — 6Mp/L

8. A thin steel ring of mean radius R is loaded as shown in Fig. E. The cross-
section of the ring is such that the plastic moment is M p. Find the limit load Pl for
the ring. Ans, Pl = ^Mp/R.

Fig. D Fig. E
 CHAPTER X
THEORY OF COLUMNS

10.1 Eccentric Loading of a Short Strut

Consider in Fig. 10.1a a short post or strut subjected to an eccentrically

applied compressive force P at its upper end. If such a strut is compar-
atively short and stiff, the deflection due to bending action of the eccentric
load will be negligible compared with the eccentricity e, and the principle
of supperposition applies. The strut is assumed to have a plane of sym-

Fig. 10.1

metry (the o^y-plane) and the load P lies in this plane at the distance e
from the centroidal axis Ox. Such loading may be replaced by its statical
equivalent of a centrally applied compressive force P (Fig. 10.1b) and a

264
 ECCENTRIC LOADING OF STRUT 265

couple of moment M ^ Pe (Fig. 10.1c). The centrally applied load

produces uniform compressive stress ai — P/ over each cross-section
as shown by the diagram in Fig. 10. lb. Likewise, the end moment
stress
M produces linearly varying bending stress <72 = My/I as shown in Fig.
10.1c. Then, by superposition, the total compressive stress in any fibre
due to combined bending and compression becomes

a ( 10 . 1 )
1

where A is the cross-sectional area and I is the moment of inertia of the

section about the if-axis. Also, since we are most concerned here with
compression, compressive stress is treated as positiv^e. The stress diagram
for this superposition of stresses is trapezoidal as shown in Fig. 10.1a.
Taking y = ci, the distance to the extreme fiber on the right, and
introducing the notation r = /A for radius of gyratiotf of the cross
section, eq. (10.1) becomes

Similarly, taking y = —c-i, for the extreme fiber on the left, eq. (10.1)
gives

.... - 3 (i - f
Usually we are only interested in the maximum compressive stress as
given by eq. (10.2a). However, it is of interest to note from eq. (10. 2b)
that if the ratio ec2 /r^ becomes greater than unity, there will be tensile
stresses in some fibers on the left side of the strut. For example,^ the
cross-section is a rffttangle of width b in the z direction and height h in the

y direction, we have Ci = C2 = h/2 and -- Ij = Then from eq.

(10.2b), we conclude that no tensile stress will be produced by eccentric
compression so long as the condition

(a)
6 ® 6

is satisfied.If condition (a) is not fulfilled so that tensile stress does occur,

it always be smaller than the compressive stress obtained from eq.

will

U0.2a), but even so, it may become important in the case of such materials
as brick or concrete, which are weak in tension.
Let us consider now the more genera! case of a strut under the action
of a compressive load P which does not lie in either of the two principal
planes of bending. In Fig. 10.2, let Oy and Oz be the principal axes through
266 THEORY OF COLUMNS
the centroid of the cross-section and
let A be the point of application of
the load P. Denoting the coordi-
nates of this point by m and n, the
^ moments of P about the axes Oy
and Oz will be Pn and Pm, respec-
Then, by superposition, the
tivel}'.

compressive stress a for any point in

the cross-section defined by coordi-
nates .r and y becomes

(7 = P
I+
,
“
Pmy Pnz
"'77’
,

(10.3)

analogous to eq. (10.1). Equating the right-hand side of this equation

to zero, we obtain an e(}uation for the locus of points of zero stress in the
cro8s-*secti(ui. With the notations r, = ^JIz/A and Vy ~ yjiy/A for radii
of gyration, this ecjuation of the line of zero stress becomes

=
^ + r^ +
f z I
y
i 0. (10.4)

which is seen tr) he the etpiation of a straight line, shown in Fig. 10.2
as the line nb. All longitudinal fibers situated in the unshaded portion of
t.he cross-section will he in eompression, while all those situated in the
>'>haded portion will he in lension. The iiitereepts ii and v of this line on
the 2/- and z-axps are found from ecj. (10.4) by setting first z and then y
equal to zero. Thus
rj
n — — V = - (b)
m n

)u)r a given eross-section, it is of interest to define a region around the

centroid within whieh the load P will induce compression over the entire
section. This region is called the core of the sf^ction. The determination
of this core for a rectangular section i.s discussed in the following example.
 ECCENTRIC LOADING OF STRUT 267

KXAMPLE 1. Find the equation of the boundary of the core for the case of an
eccentrically loaded strut of rectangular cross-section as shown in Fig. 10.3a.
SOLUTION. Let the point of application of the compressive force P be confined to
the lower right-hand quadrant (m and n both positive). Then to make the line of
zero stress pass through point 5, we set y = -~k/2 and ^ = —6/2 in eq. (10.4).
This gives

mh nb ^

which, with = 62/12 and becomes

6m fin
(c)
~k^T
This the equation of a straight line cd having intercepts 6/fi and 6/6 on the y- and
is

respectively, as shown in Fig. 10.3a. As long as the point of application of

2 -axes,

the load P is situated between this line and point P, the fiber stress at B will be
compression. Similar arguments, when the load P is confined to each of the other
three quadrants, can be made and we cimclude that the core of the section is the
shaded rhombus shown m Fig. 10.3a.
A similar argument can be made for the case of a circular ci oss-section of diameter
d and we find that the (‘ore of the section is a (drcle of diameter d/4 as shown in Fig.
10.3b.

PROBLEMS
1. A short strut of square cross-section has a notch c ut in its side as shown in Fig.
A. Calculate the maximum compressive stress cr at the section mn du(i to a centrally
applied load Ans. (Tmax = HP/aK
P
2. Solve preceding problem if the cross-section of the strut is a circle of diame-
thf'

ter a. Ans. 'Tmax = 9.lP/a2.

3. At the cross-section mn, the C-clamp shown in Fig. B has a rectangular cross-
section 1 X i ii^- Determine the maximum tensile stress a at n if the load P = 400
Ih and the dimension 6 = 3 in. Ans. a
— 15,200 psi.
4. A short pie(;e of 10LJ20 channel carries a compressive load the line of action of
which passes through the centroid of the w^eb. If the allowable maximum compres-
sive stress is 16,000 psi, calculate the safe load P. Ans. P = 61,000 lb.

Fig. a Fig. H Fig. C

268 THEORY OF COLUMNS
5. A
short piece of 6 X 6 X 1-in. angle iron carries a compressive load the line of
action of which coincides with the intersection of the middle planes of the legs. If
the maximum compressive stress is not to exceed ] 6 000 psi, what is the safe load P?
Ans, P = 37,500 lb.
6. Solve the preceding problem if the angle iron has an 8 X 4 X 1-in. cross-

section. Ans. P = 37,000 lb.

7. Show that the core of a circular cross-section is circular and find its radius if d
is the diameter of the section. Ans. d/8.

8 . A short strut loaded in compression has an equilateral triangular cross-section

with sides of length a (Fig. C). Prove that the core of the section is an equilateral
triangle with sides of length 3a/8 as shown in Fig. C.
9. A short strut has a hollow circular cross-section of outside diameter di and
inside diameter d 2 Prove that the core of the section is a circle having radius
.

r. - i(di + d^Vdi).

TO*2 Long Columns; Euler’s Column Formula

Let us consider now the case of a long slender column AB of length /

which is end A and subjected to a <‘ontrally applied

built-in at its lower
compressive load P at its upper end R as shown in Fig. 10.4. This column
is assumed to be perfectly straight and of lUiiform (‘ross-section. It is
also assumed that the material is. homogejieous and that it behaves elasti-
cally.
Experienceshows that when the
vertical load P
is small, such a com-

pressed column is laterally stable.

That is, if the upper end B is pushed
B —
I
slightly to one side by a lateral force,
the column will return to its straight
form as soon as this lateral force is
removed. However, as P is gradually
increased, we observe that at a certain
value of this load, the straight form of
equilibrium becomes unstable and
the column, if pushed to one side,
A. A\
— / stays there even after the lateral force
is removed. This instability phe-
Fig. 10.4
nomenon is called lateral buckling and
the value of the load at which it occurs is called the critical load, denoted
by Per.
To find the load Per which will cause buckling, we consider the column
bent configuration shown in Fig. 10.4b and calculate the
in the slightly
magnitude of the vertical load necessary to hold it there. Choosing
coordinate axes through point A, as shown in Fig, 10.4b, we denote the
deflection at any point x on the elastic line by y and the deflection of
 EULER’S COLUMN FORMULA 260

point B by 6. Then the bending moment at the cross-section defined by

X becomes

M. = P(b - y). (a)

Since we consider only a very small lateral deflection the relation between ^

bending moment and curvature is (see eq. 8.1, p.l98)

«s- ±M.
For the chosen axes x and y, d^y/dx^ is positive in this case and we use the
plus sign in this expression. Hence the differential equation of the elastic
line becomes

(b)

Dividing both sides of eq. (b) by El and introducing the notation

(c)

it may be written in the form

+ k^y = k^. (d)

This is a linear differential equation with constant coefficients. Its solution

y = two parts :(1) the general solution of the corresponding
f(x) consists of
homogeneous equation (zero on the right-hand side) and (2) a particular
solution of the complete equation. It can easily be verified by substitution
that a general solution of the homogeneous equation is

yi == Cl sin kx +C 2 cos fcx,

where Ci and C2 are arbitrary constants, while a particular solution of

the complete equation is

if
2/2
= 5 .

Thus the complete solution becomes y = yi + or

y — Cl sin kx +C 2 cos kx + 6. (e)

To find the integration constants Ci and C2 we have the boundary

,

conditions at the built-in end A \

(f)
270 THEORY OF COLUMNS
Ono differentiation of eq. fe) with respect to x gives

^
ox
= Cl k COR kx -- C^k sin kx. (g)

Now substituting conditions (f) into eqs. (e) and (g), we find

Cl =.0, Co = -6,

and the general solution (e) becomes

~ 5(1 — cos kx). (h)

This shows that the column bends so that its elastic line takes the form of
a cosine curve, the amplitude 5 of which is as yet undetermined.
To examine rhis question further, we note that when .r = /. v =
Substituting this condition into eq. (h), we obtain
5 = 5(1 — cos kl)

from which

5 cos Id = 0. (i)

From eq (i), we conclude that either 5 - 0 or cos A:/ = 0. If 5 = 0, the

c'olumn stands in the straight vertical (‘ontiguration of equilibrium and
no limitation is imposf'd on the magnitude ()f the load F. On the other
hand, if cos A7 — 0, i.e.. if

A7 - (j)

where n is an odd integer, tlien the column can also be in equilibrium

with any small deflection 5 The smallest value of P for whieh
at the top.
this can c»ccur will be obtained by taking n = in ecj. (j). Then, using I

the notation (c). w(' obtain

/ = IT,

4,Ef 2

from which

T^EI
Per = • (10.5)
4/2

This is Euler's column formula column built-in

for the case of a slender
at its base and free at the top. For all we must have
values of P <
5 = 0 and the straight configuration of equilibrium shown in Fig. 10.4a
is said to be stable. If point B is pushed to the side and then released,
the column will return to the vertical position. If P = P^^, equilibrium
can exist for any small value of 5 and the vertical configuration is said
 EULER'S COLUMN FORMULA 271

to be indifferent or neutral.In this case, if point B is pushed to the side

and then column simply remains in the slightly defle(‘ted
released, the
position. If P > Per, and point B is pushed slightly to the side, the deflec-
tion increases indefinitely and the vertical eciuilibrium configuration is
said to be unstable. Thus the value Per as defined by eq. (10.5) represents
the maximum load that the column can carry; it is sometimes called the
Euler load for the column.
Examination of eq. (10.5) shows that the critical load for a given column
is proportional to the flexural rigidity El and inversely proportional to
the square of the length I of the column. It will also be noted that the
critical load is independent of the compressive strength of the material.
Thus two geometrically identical slender columns, one of high-strength
alloy stc^l and the other of ordinary structural steel, will fail by buckling
at approximately the same value of the load P, since the moduli of elasticity
are approximately the same. It must therefore b(' realized that the
‘‘strength^' of a long column is in no way dependent upon the strength of
its material in compresvsion, but only on its geometry and the stiffness of

the material.
Equation(10.5) shows also that the strength of a long column may be
increased by increasing the moment of inertia / of its cross-section. With-
out increasing the cross-sectional area, this may be done by distributing
the material as fax as possible from the principal axes of the cross-section.
Hence tubular sections are more economical for compression members
than solid sections. By diminishing the wall thickness of such sections
and in(!reasing the transverse dimensions, the stability of the column may
be increased. There is a lower limit for the wall thickness, however, below
which the wall itself may become unstable and, instead of buckling of the
column as a whole, there will be buckling of its longitudinal elements
which brings about a corrugation of the wail, called local buckling.
If we dc not limit our attention to the case where n - 1, eq. (j) is seen

to define an infinite number of values for the critical load. Thus, in general.

_ n^i^EI
(k)

Taking n = 1, 3, 5, .
,
we see that the values of Per are in the ratio
1:9:25:. . . . Correspondingly, eq. (h) for the elastic liiie'becornes

2/ = (1 )

and we see that as n is increased, the deflection curve has more and more
inflection points. Two of these deflection curves, for ri -3 and n = 5,
pre shown in Fig. 10.5. Although they represent theoretically possible
272 THEORY OF COLUMNS

Fig. 10.5 Fig. 10.6

inodes of buckling for the columns, they are of no practical interest because
the column will always buckle in the finest mode, as represented in Fig.
10.4b, as soon as P reaches the first critical value given by eq. (10.5).
Critical loads for long columns having other end conditions than those
of the simple cantilever column in Fig. 10.4 can easily be obtained from
eq. (10.5). Consider, for example, the long column with pinned ends as
shown in Fig. 10,6a, l"rom symmetry, may
be concluded that in the
it

first mode have a vertical tangent at its

of buckling the clastic line will
mid-point C, Hence, each half of the column is in the same condition
as that in Fig. 10,4 and the critical load is obtained from eq. (10.5) simply
by replacing I by 1/2, Thus

( 10 6 )
.

Similarly, for a long column having both ends built-in. Fig. 10.6b, the
hrst buckling mode be a full cosine wave having inflection points at
will
D and E, Then each end portion behaves as a simple cantilever column
of length t/4 and we obtain, from eq. (10.5)

P or (10.7)
P

The case of a column built-in at one end and pinned against lateral deflec-
tion at the other end requires some further consideration and is discussed
m the following example.

EXAMPLE 1. A slender column of length I is built-in at its lower end A and

laterally feupported at its upper end B as shown in Fig. 10.7. Find the first critical

value of the compressive load P.

 EULER’S COLUMN FORMULA 273

SOLUTION. In this case, a horizontal reaction Q wil] be induced at the pinned end
B when buckling takes place. Then the bending moment at section x is

M, - Q{1 - x) - Py,

and the differential equation of the deflection curve becomes

+ <3 a- ^)-

The general solution of this equation is

Q
—
Cl cos kx -f- Cf sin kx {I x),

where
4 ,-

For determining the constants Ci and C 2 and the unknown

reaction Q,we have the following conditions at the ends

— Oj
- {y)x-m > it)
\ctx J 1^0

Then, from the above equation for y,

Cl +^
01
= 0; Cl cos kl + Cl sin kl = 0; kCi —0
— 0 .

Substituting the values of Ci and C 2 from the first and third of the above expressions
into the second one, there results

“ sin kl — I cos k1^ = 0.

As long as the critical load has not been reached, there is no buckling and Q is zero.
When and buckling is present, Q
the critical load has been reached is no longer zero

and hence the expression within the parentheses must be zero. This gives the
following transcendental equation for determining the critical load:

tan kl == kh

The smallest value of kly different from zero, and therefore of P, which satisfies this
equation is kl - 4.49. Since k^ = P/EI,

202EI v^EI
P„ - k^EI - *’ (m)
P ( 0 70
.

PROBLEMS
1. A 6-ft length of 61-12.5 standard I-beam is to be used as a pin-ended column.

Calculate the critical value of the load, Ans. P„ = 103,000 lb.

2. A 14WF78 wide-fiange steel beam is to be used as a column 45 ft long with the
lower end built-in and the upper end pinned. Calculate the maximum compressive
load that the column can carry. Ans, Pro«x = 428,500 ib.
274 THEORY OF COLUMNS
an equal-leg steel angle section to serve as a pin-ended column 10 ft long
3. Select
ifthe centrally applied compressive load P = 18,000 lb and a factor of safety
n = 2.5 against failure by buckling is required. ^ ns. 4 X 4 X J in.
4. Select a wide-flange steel section to be used as a column with both ends built-in
and of length I — 30 ft if the compressive load P = 80,000 lb and a factor of safety
n = 2.5 against failure by buckling is required. Ans. 8WF28.
5. A slender pin-ended aluminum column 6 ft long is to have a thin-walled circular
cross-section of outside diameter d = 2 in. Calculate the wall thickness t required
in order to attain a factor of safety n - 2 against failure by buckling if the actual
load P —
3000 lb. Ans, < = i in.
6. A
pin-ended slender column has a rectangular cross-section with dimensions
b and h. It is laterally supported in the thin direction by rollers at its mid-point C
as shown in Fig. A but is free to slide between the rollers in the direction normal to
the paper. Find the ratio b/h so that the resistance to buckling will be the same in
both of the principal planes. The load P is assumed to be centrally applied. Ans.
b/h = 2.
7. horizontal rigid bar AB of negligible weight is supported by two slender
A
steel columns as shown in Fig. B. The column at A has a circular cross-section 1 in.
in diameter and the column at B has a 1 X 2-in. rectangular section. The bar AB
is constrained to move only mthe plane of the paper. Find the critical value of the
load Q if X - 1.5 ft. Arts Qcr = 3152 lb.
8. Referring again to the system in Fig. B, find th(^ A alue of x to make Qct a
maximum. What is this maximum critical load? Ans. Qmax = 15,270 lb when
.r = 2,69 ft.

9.

10.

Fig. A Fio. B
A rigid bar AR is pin-supported by a vertical tie rod AE and a slender column
DC built-in at D as shown in Fig. C. The column is made of steel and has a X 1- J

in. square cross-section. If the load P = 1000 lb, find the maximum value of x

consistent with stability of the system. Neglect the proper weight of the bar AB.
Ans. Xm.x = 3.42 ft.
In Fig. D
the bars AB, AC, and ADare slender circular steel columns all having
the same flexural rigidity EL They have pinned ends at A and C and built-in ends
at B and D. Calculate the critical value of the vertical load P applied at A Assume
.

buckling in the plane of the figure. Ans. For = 3.65 —

tt^EI
 \
L
— -/
p_.
EULER'S COLUMN FORMULA

-
_ .
P
275

2'
1

E
TTTT S'

vm?//h/y

Fig. C

^.3 Further Discussion of Euler’s Column Formula

The case of a column with pinned ends (Fig. 10.6a) is most frequently
encountered and is called the fundamental mse. Dividing both sides of
eq. (10.6) for this case by the cross-sectional area A and introducing the
notation r = yff/A^ for radius of gyration, we obtain

where the ratio l/r is called the slenderness ratio of the column and o-or, the
critical compressive stress. The curve ABC in Fig. 10.8, plotted from eq.

(10.8) is called Euler's curve. For a column of any given slenderness ratio,
plotted as abscissa, this curve shows the corresponding value of the average
compressive stress (P/A)ori plotted as ordinates, for which the column
becomes laterally unstable.

Since it was assumed that the material behaves elastically at the begin-

ning of lateral buckling, we conclude that eq. (10.8) is valid only if it

gives a value of less than the proportional limit oTp.i of the material.
276 THEORY OF COLUMNS
Thus, by setting i ,
we obtain from eq, (10.8), a limiting value of
l/r below w^hieh Euler^s formula does not apply. This is represented by
point B on the curve in Fig. 10.8, and marks the boundary between so-
called short and long columns. Taking structural steel with a proportional
limit cTp.i. = 30,000 psi and a modulus of elasticity E = 30(10)® psi, for
example, we find l/r = \T()00^ =
Thus for l/r < 100, the average
100.
compressive stress in a pin-ended steel column will reach the proportional
limit before lateral buckling can occur and eq. Q0.8) is inapplicable. For
this reason, the portion AB of the Euler curve in Fig. 10.8 is shown by a
dotted line; only the portion BC is valid.

The Euler curve for the fundamental case of a column with pinned
ends can be used for columns with other end conditions by introducing
a modified length h. Comparing eq. (10.5) with eq. (10.6), we see that the
modified length of a cantilever column will be h = 2/, where I is the actual-
length of the column. from eq. (10.7), we see that the modified
Similarly,
length of a column with both ends built-in will be U = 1/2, while for a
column built-in at one end and pinned at the other, l\ = 0.7/ (see p. 273).
In general, then, eq. (10.8) may be used in the form

<rcT (10.8a)

We have st^en above that for columns having sufficiently high slenderness

ratios, buckling occurs before the average compressive stress can reach
the proportional limit of the material. In all such cases, stability of the
column governs ihe working load and we use the portion
selection of a safe
BC of the Euler curve as shown in Fig. 10.9 as a basis of design. As another
extreme, columns having very low slenderness ratios can be expected to
fail due to some kind of weakness of

the material before the critical com-

pressive stress can be reached. This
failure associated with the strength
of the material may result from
crushing, as in the case of concrete,
or yielding, as in the case of struc-
tural steel. In any case, some max-
imum compressive stress can be set
as a limit of strength and the safe
working load chosen accordingly.
we can represent in Fig. 10.9
This,
by a horizontal line DF drawn
through the chosen strength limit.
In between the ranges of short columns and long columns there will be a
range of slenderness ratios too small for true elastic instability to govern
 EULER’S COLUMN FORMULA 277

and yet too large for strength considerations alone to govern. Columns
falling within this
range of slenderness ratios require special consideration
and are called medium column^. Analysis of their behavior under centrally
applied compressive loads has been widely treated both experimentally and
theoretically, but a discussion of such behavior will not be given here.* A
simple and satisfactory procedure, in these cases, is simply to draw a
EB in Fig. 10,^ and arbitrarily let its ordinates represent the
straight line
maximum compressive stress for such medium-length columns. In this way
we obtain the broken line DEBC in Fig. 10.9, which can be used as a basis
of design for a column of any length.
As a specific example, let us consider the case of ordinary structural
steel for which E = 30(10)® psi, ap i
= 30,000 psi, and cTy p ==?=
40,000 psi,
Then for all practical purposes, the yield stress (jy p will represent the
strength limit of the material for very short columns. Furthermore,
experiments show that l/r == 60 is the upper limit of slenderness ratios for
which a ('olumn can sustain this average compressive stress without
buckling. Thus for slenderness ratios from 0 to 60, we take (Xnuix = p
=
40,000 psi as shown by the line DE in Fig. 10.10 For slenderness ratios
greater than 100, (7,nax = ai^d we use the portion BC of the Pmler
curve. For intermediate slenderness ratios (60 < l/r < 100), the straight
line EB is used. Thus the complete line DEBC in Fig. 10.10 determines

the value of <r,nrtx for any given pin-ended mild steel column.
In discussing working stresses for steel columns, we must consider the
stress given by the diagram in Fig. 10.10 as an ultimate stress. Such a
compressive stress brings the columns to complete failure, either by buck-
ling, yielding, or a combination of both, and the working stress should be

*See Timoshenko and Gere, Theory of Elastic Stability, McGraw-HiJl Book Co., Inc.,
New York, 1960, p. 175.
278 THEORY OF COLUMNS
taken as where n is the desired factor of safety. The value to be
chostm for this factor of safety depends upon unforeseen or accidental
increases in the load P and also on possible errors in the central application
of this load, as well as possible initial crookedness of the column. A
common n in structural work is 2.5.
value for
Experiments show that both errors in central application of the load
and initial crookedness of the column tend to increase with increasing
slenderness ratios. This suggests the use of a variable factor of safety
which is usually increased linearly with l/r. For example, some structural
specifications recommend using a factor of safety n = 2.0 + 0.015 l/r
for 0 < //r < 100, and n = 3.5 for l/r > 100.
In the whole realm of structural design, there is perhaps no situation
that presents so troublesome a question in regard to the proper choice
of a factor of safety as that of columns. As already indicated, this stems
from the fact that inherent inaccuracies in centering the load and in attain-
ing assumed end conditions, as well as initial crookedness of the column,
have a pronoun(;ed effect upon its behavior under compression. Many
thousands of columns have been tested over the years in an attempt to
resolve this question. The results of such tests when plotted as shown in
Fig. 10.11 form a wide band and leave the question of proper factor of

(P/A/kni A
1 \
\

^ .
• •
•

•• •
•••
•\

• ^•>k

'^C
^
0 ^0 ioO 150 200 Zbd ^

Fio. 10.11

safety very much unsettled. One point, however, is clearly brought out by
these test results; namely, the broken line D5C, constructed as already
explained, represents an upper bound for the test results. This substan-
tiates the theory and leaves the pronounced but indeterminate influence of
imperfections simply to be adequately provided for by the use of a generous
factor of safety.
 EULER^S COLUMN FORMULA 279

PROBLEMS
1. A steel bar of rectangular cross-section 1 X 2 in. is to be used as a column

with pinned ends. What is the shortest length I for which Euler’s equation applies
if E — 30 (1C)® psi and cTp \ = 30, (XX) psi? Ans. I = 28.9 in.
2. Calculate the critical compressive stress for the column described in the pre-
ceding problem if it is 4 ft long. Ans. cr^r ~ 10,700 psi.
3. An extruded nickel steel tube having outside diameter d = 2 in. and wall
thickness t = J in. is used as a pin-ended column 12 ft long. Calculate the critical
compressive stress if E = 30 (10)® psi. Ans. acr = 6300 psi.
4. What is the shortest length / of the column described in the preceding problem
for which Euler’s formula can be applied if <t,> = 160,000 psi?
i I
-- 28.5 in.

5. Calculate the critical compressive stress for the column described in Problem 3,
if the material is extruded magnesium for which E = 6.5 (10)® psi. Ans. acr ="
1360 psi.
6. A steel column with built-in ends is 6 ft long and has a standard 61-12.5 section.
Using the broken line DEBC in Fig. 10.10 and a factor of safetv n - 3, find the safe
load Pu. - 4S,2001b.
7. Solve the preceding problem if the column is built-in at one end and pinned at

the other. = 45,200 lb.

8. C’onstruct a curve similar to the line DEBC in Fig. 10.10 to be used as a basis
of design forcolumns of extruded aluminum alloy having E ~ 10 (10)® psi, ap = i

20,000 psi, and a.nux = 30,000 psi. Assume that // r = 30 is the upper short-
column limit and l/r = 70 is the lower iong-c'olumn limit.

10.4 The Secant Formula

Referring to Fig. 10.12, let lus consider the case of a perfectly straight
cantilever column subjected to a compressive load P applied with ec-

P X P

Fig. 10.12
280 THEORY OF COLUMNS
centricity e. This case differs from that discussed in Art. 10.1 only in the
fact that we now assume the column to be slender so that the lateral
deflections cannot always he considered as small compared with the
eccentricity e of the applied load. As the load P is gradually increased,
the column will start to bend slightly, and for any value of P the deflection
curve w’ill be as shown in Fig. 10.12b.
Choosing coordinate axes x and y as shown and denoting by 6 the deflec-
tion of the upper end, the bending moment at any cross-section defined
by X will be
M* = P(6 +c- y). (a)

Then for sm 4 ll deflections within the elastic limit of the material, the
differential equation of the elastic line becomes

= P(S+e-y). (b)

Dividing both sides of this equation by El and using the notatit>n

— P (c)
El'

it may be written in the form

=
g+ *»(« -f e). (d)

This equation is similar to eq. (d) on p. 269 and its solution is

y = Cl sin kx + cos kx +5+ e, (e)

Noting that when x = 0: y = 0, dy/dx = 0, we find

Cl = 0, C2 = - (6 + c)
and the solution (e) becomes

-y = (6 + e){ 1— cos kx). (f)

This shows that the deflection curve AB has the form of a quarter cosine
wave.
At the upper end B we have y — 6 w^hen x = Z. Substituting these
simultaneous values into eq. (f), we obtain
— cos kl \
5 (10.9)
cos kl )

For a given value of this equation expresses the relation between the
lateral deflection 5 and the load P, indirectly defined by fc = yjP/EI.
 SECANT FORMULA 281

Taking e = 0, we see that 5 = 0

any value of kl < ir/2, i.e., any
for
value of P < Per (see eq. 10.5, p. 270). But when kl - t/2, i.e., when
P = Por, the deflection becomes indeterminate; it can have any small
value. These observations agree with those already reached in Art. 10.2
for the case of a centrally loaded column. However, for any small value
of c > 0, eq. (10.9) gives a definite value for d so long as P < P^ and
shows that 6 increases without limit as P approaches Per. The load-
deflection relationship expressed by eq. (10.9) is seen to be nonlinear.
For a chosen value of the eccentricity e and a given column, eq. (10.9)
can be represented graphically as shown in Fig. 10.13. For e = 0, the
load-deflection curve is represented by the two straight lines OAB. No

deflection takes place for P < P^r and at P = Per the deflection is in-
determinate. For e > 0, but small, the load-deflection relation will be
represented by the curve OCB. For small loads, the deflection increases
slowly but as the load approaches the critical value, it increases rapidly
and tends' toward infinity.
For small values of kl « ir/2, i.e., for values of P« Per, we may take

cos fci s= 1 —
2 ^

in eq. (10.9). Then

_ _ {kiy/2
“ 1 - (kiy/2
~

or, since k = \P/EIy

*-iir (g)
282 THEORY OF COLUMNS
This is seen to agree with the deflection of a cantilever beam
loaded by a couple M = Pc at the free end (see Case 5 of
Table 8.1, p. 212). This justifies the treatment of eccentri-
cally loaded short struts as discussed in Art. 10.1.
In the present case of an eccentrically loaded long column,
it is important to note that there is no proportionality be-

tween the load P and the deflection 5 that it produces; hence

the method of superposition cannot be used. An axially ap-
plied load P when acting alone produces no lateral deflection,
but when acting in company with a bending couple Pe, it
produces also some bending and the resulting deflections are
greater than the bending couple alone would produce. This
magnification is especially pronounced if the axial force is
near the Euler value.
All of the foregoing discussion can be applied also to the
fundamental case column eccentrically loaded
of a pin-ended
as shown in Fig. 10.14. From symmetry conditions, the
tangent to the elastic line at (7 remains vertical and each
half of the column is in the same condition as the column in Fig. 10.12.
Hence, upon replacing I by 1/2 in cq. (10.9), we obtain, for the deflec-
tion 6 of the mid-point C of the elastic line,

where I is now the full length of the pin-ended column.

The maximum bending moment in Fig. 10.14 occurs at the mid-section C

and has the magnitude
Afmax = P(5 + e). (i)

Substituting expression (h) for 5, this becomes

M„„ = Pe sec ~ (j)

The corresponding maximum compressive stress on the concave side of the

column at C is

Substituting expression (j) for Mmmx and using the notation r » for
radius of gyration of the cross-section, this becomes

ffmM =J 1
+ ^ sec j- (10.10a)
 SECANT FORMULA 283

If k is replaced by yjP/EI, we have

(10.10b)
<Tn.ax = I 1
+ 72 VZ^)]
Equation (10.10) represents the so-called secant formula for an eccentrically
loaded slender column. It gives the maxinnim compressive stress in terms

of the average compressive stress P/A^ the eccentricity ratioec/r’^^ and the so-

called Euler o.ngle li/2. If the column is short, and the load P is small, we
have sec kl/2 = 1 ,
and eq. (10.10) reduces to the same form as eq. (10.2a)
of Art. 10.1.
If wc set some limit on say o-mAx = o-y p. in the case of structural

steel, we may then calculate the corresponding average compressive stress

which will first produ(‘e yielding in the most severely stre.ssed fibers. For
this purpose, eq. (10.10) i.< written in the form

( 10 . 11 )

For any chosen eccentricity ratio, we may solve eq. (10.11) by trial and
errorand plot a curve which shows, as a function of //r, the value of P /
at which yielding lirst begins in the most stressed fibers. A family of suc^h

curves for several values of the eccentricity ratio cc/r^ and tor ay p
=
40,000 psi is shown in Fig. 10.15.

We usually consider the value of Pi A for which yielding in an extrenu^

fiber begins as the limit load for a steel column. Having found this value
for a given column from the graph in Fig. 10.15, or troni eq. (10.11), the

(P/A )kps

Fio. 10.15
284 THEORY OF COLUMNS
safe average compressive stress will be obtained by dividing the obtained
value by a suitable factor of safety n.
If desired, the factor of safety can be incorporated directly into eq.
(10.11) as follows. Let denote the safe working load, and n the required
factor of safety. Then the compressive load which would bring the extreme
(10.11)
fiber stress to the yield point will be nP^,. Substituting this for in eq. P
,
we obtain

K ( 10 12 )
.

A i 1
^ ItiP10 \
1 +758®*=

Comparison of this equation with eq. (10.10) shows that the introduction of
nPu, in place of P in the denominator of eq. (10.10) takes proper account of
the fact that amax does not increase linearly with P.
The transcendental character of the secant formula makes it rather
difficult to solve, trial and error procedures usually being necessary. Such
procedure is illustrated in the following example.

EXAMPLE 1. A wide-flange steel section (Fig. 10.16a) is to be

selected for a long and having pinned ends. The
column 25 ft

column is to carry a centrally applied load Pi = 100 kips and a

bracketed load Pa = 50 kips, the latter applied on the principal
axis 2-2 of the section at a distance of 10 in. from axis 1-1,
Fig. 10.16b. Select a suitable section such that the column
will have a factor of safety n = 2.5 against failure by yielding
= 40,000 psi).
(o-y p.

SOLUTION. The two loads applied as shown in Fig. 10.16b

can be replaced by a single load P = 150 kips applied with
eccentricity e = 3.33 in. as shown in Fig. 10.16c. Then to use
the secant formula as a basis of selecting the required section,
we write it in the form
o-yp P Pe (I lnP\

where Z is the section modulus and 7 the moment of inertia of

Fig. 10.16
the section. By trial and error, we now look for a WF section
which will make the right side of eq. (1) equal to or less than
the leftside, i.e., equal to or less than cry,p,/n = 40,000 -s- 2.5 = 16,000 psi. Using
Table B.2 of Appendix B, we find
For a 14WF68 section: Right side = 13,500 psi.
For a 14WF61 section: Right side * 15,270 psi.
For a 14WF53 section: Right side = 18,160 psi.
This indicates that the 14WF61 section is suitable so far as bending in the principal
plane of loading is concerned. We assume that buckling in the plane of smaller
flexural rigidity is prevented.
 SECANT FORMULA 285

PROBLEMS
1. A 14WF142 section sU^el column is 25 ft long and has pinned ends. It carries a
centrally applied load Pi =
420,000 lb and a bracketed load Pz = 50,000 lb on axis
2-2, 19 in. from axis 1-1, see Fig. 10.16. Using the secant formula, calculate the
maximum extreme fiber stress. Ana, o-max “ 15,900 psi.
2. If the yield stress for steel is <Ty „ = 40,000 psi, what factor of safety does the
column in the preceding problem have against failure due to yielding? Ans. n = 2.4.
3. A 6-ft length of steel pipe having outside diameter d = 2^ in. and wall thick-
ness t = ^
in. is to be used as a pin-ended column. The compressive load P is
applied at the rim of the cross- section, i.e., with eccentricity e = 1.0 in. Calculate
the value of P which will produce yielding (<ry p. = 40,000 psi). Ans. P = 62001b.
4. Calculate the safe compressive load P for a steel column 6 ft long with pin-end
conditions if the cross-section is a 4 X 4 X 1-in. angle section and the load P is
applied along the line of intersection of the planes of the legs. The required factor of
safety against failure by yielding is n = 2 and cts .p. = 40,000 psi. Ans. P^ = 7250
lb.

5. A 12WF72 section is used for a pin-ended steel column 10 ft long. It carries a

centrally placed load Pi = 100 kips and an eccentric load Pi which acts on axis 1-1
and 4 in. away from Using the secant formula with <ry p =
axis 2-2, see Fig. 10.16.
40,000 psi and n = the safe value of Pi. Ans. Pi = 60,200 lb.
2.5, find
6. A 12WF45 section is used as a pin-ended steel column 16 ft long. It supports a
compressive load applied with an eccentricity ratio ec/r* = 0.2 in the plane of least
flexural rigidity. Calculate the safe load if n = 2 and ay = 36,000 psi. Ans.
Pu. = 136,400 lb.
7. An 8WF40 section used for a pin-ended steel column 16 ft long. The load P
is

acts on axis 2-2 with an = 3 in. Calculate the magnitude of P which

e(‘cen tricity c
will produce yielding if (7y p = 36,000 psi. Ans. Py.p == 190,500 lb.
8. A column made of steel pipe is 12 ft long, built-in at the bottom end and pinned
at the top. The outside diameter of the pipe is 3.5 in. and the inside diameter is 3.0
in. Assuming ay = 36,000 psi and an eccentricity ratio ec/r^ = 0.25, find the safe
load P, if the desired factor of safety against yielding is n = 2.0. Ans. Pv, = 27,600
lb.

10.5 Representation of Imperfections by Equivalent Eccentricity

In Art. 10.3 it has been pointed out that the behavior of a column under
compressive load is very much affected by imperfections such as slight

crookedness of the column axis, errors in centering the load, uncertainty as

to end conditions, etc. The extent of these imperfections varies sufficiently
from one column to another to produce the rather wide scatter of test
results when represented as shown in Fig. 10.11. To appreciate the in-
fluence of imperfections on the strength of a column, it must be observed
that they result in the line of action of the load not coinciding exactly with
the axis of the column, thus introducing more or less bending action in

addition to direct compression. On this basis it seems logical to conclude

that the behavior of a real imperfect column undfer load will be similar to
that of a perfectly straight ideal column loaded with a suitable eccentricity e.
286 THEORY OF COLUMNS
This suggests that the secant formula derived in Art. 10.4 could be used
als(' as a basis of design for supposedly straight centrally loaded columns,
simply by choosing an appropriate value of the eccentricity ratio eo/r-
to account for the effect of imperfections. While such a procedure still

leaves unanswered the question of the proper value of cc/r^, it offers a

rational means to represent the effect of imperfe(‘tions rather than to
simply allow for them by an overlarge factor of safety as suggested on p. 278.
To select a suitable value of eejr^ to account for imperfections, recourse
must be had to the results of tests made on actual columns. As a result of
such experiments, a value ec/r- = is commonly recommended for pin-
0.25
ended columns as used in ordinary structural work. This does not mean
necessarily that every column is expected to be this imperfect, but rather
that no column is likely to exceed this degree of imperfection. That is, if
the results for the columns tested were plotted on the diagram in Fig. 10.15*
the points would all lie above the curve labeled eejr^ = 0.3. The single
curve for ec/r^ = 0.25 and for Oy p. = 40,000 psi is shown agaiii in Fig.
10.17a. This curve may be used as a basis for the design of structural steel
columns. For such a column of any given slenderness ratio //r, the cor-
responding ordinate of the curve shows the value of the average compressive
stress P! for which yielding of the mosi stressed fibers in compression can
be expected to begin. The (‘orresponding safe value would be this P/A
divided by the desired factor of safety n. Since the curve already takes ac-
count of the effect of imperfections, a safety factor n = 2 may be con-
sidered as adecaiate.
Since lack of straightness of a column can be expected to increase with
the length, sometimes preferred to allow for imperfections by using an
it is

which increases linearly with the length I of the

eciuivalent eccentricity e
column. On the basis of test results on steel columns, a commonly recom-
mended value is e/l = 1/400. In such case the secant formula becomes

(a)

1 -f
—
^400
• -
r

where the ratio c/r depends on the shape of the cross-section. For a solid
circular cross-section c/r =
2, for a rectangular cross-section c/r
= for
a thin- walled hollow circular cross-section c/r = ^|2y and for an /-section
iKjnding in its plane of greatest flexural rigidity, c/r — 1. Two curves DE^
plotted from eq. (a) for c/r = 1 and c/r = 2, are shown in Fig. 10.17b
where the yield point cjy.p = 40,000 psi. These curves may be used as a
basis for design of steel columns in the same manner as was explained above
in connection with the curve in Fig. 10.17a.
 EQUIVALENT ECCENTRICITY 287

Fig. 10.17

The use of such curves as those in Fig. 10.17 based on the secant formula
represents the most rational approach to the design of centrally loaded steel
columns. We have seen that the load which such a column can safely carry
is influenced by the impenections, the strength of the material, and the
inherent aspect of instability. Each of these factors is rationally repre-
sented in the secant formula. The effect of imperfections is accounted for
by the chosen eciuivalent eccentricity of load; the strength of the mateiial
itself, by the chosen value of o-y.p; and the inherent aspect of instability,

by the appearance of P/A in the secant term. Thus the secant formula
blends the cases of short, medium, and long columns into one continuous
range and obviates the necessity to make these classifications as was done in
Art. 10.3.
288 THEORY OF COLUMNS
The strongest objection to the secant formula is its transcendental
character which necessitates solution trial and error. However, if such
curves as those in Fig. 10.17 are at hand,' the selection of a safe load for a
given column or of a suitable column to carry a given load becomes a very
simple matter.

EXAMPLE ] The column shown in Fig. 10.7, p. 273, has length / = 8 ft and a
.

square box section 4 X 4 in. outside dimensions and wall thickness r= J in. The
material is steel with Cy p. = 40,000 psi. Using the curves in Fig. 10.17b and a
factor of safety n = 2, find the safe value of the compressive load P.
SOLUTION. The geometric properties of the cross-section are A =3.75 in.^
I = 8.84 in.^ r = yil/A = 1 .54 in., c = 2.0 in. The effective length of the column
is /i = X 96 = 67,2 in. Thus l/r = 43.6 and c/r = 1.3. Interpolating between
0.7
the curves DE in Fig. 10.17b for c/r = 1.3, we find {P/A)^ p = 33,500 psi, from
which Py = 33,500 X 3.75 = 12.5,500 lb. Then with a factor of safety n = 2, we
p.

have P«, = 62,750 lb.

PROBLEMS
1.Using the design curves in Fig. 10.l7b and a factor of safety n = 2, find the
safe load for a piiMmded steel column 4 ft long and having a solid circular cross-
section of diameter d = 2 in. Ans. Pw = 28,300 lb.
2. A 3 X 3 X J-in. steel angle section is used as a column 10 ft long with built-in
ends. Using the design curves in Fig. 10.17b with a factor of safety n = 1.5, find
the allowable magnitude of the compressive load. Ans, Pu. = 15,900 lb.
3. A 12-ft length of steel tubing has a circular cross-section of 4 in. outside diame-
6.

ter and wall thickness t = in. Assuming pinned ends and using the curve in Fig.
10.17a, find the limiting average compressive stress P/A that the column can carry.
Ans, (P/A)y p = 20,000 psi.
4. A structural steel column 40 ft long is built up of

four 2 2X X angle irons held together with diag-

onal lacing to form the square cross-section shown in
Fig. A. Assuming that the lacing is adequate to make
the column perform as one with a solid section, and
that it has pinned ends, find the safe compressive load
Pu, on the basis of the curves in Fig. 10.17b with a fact
or of safety n = 2.5. Ans. Pu, = 18,800 lb.
5. Using the curve EF in Fig. 10.17a and a factor of
safety n = 2, find the safe compressive load Pu, for a
pin-ended steel column if it has a standard 81-23 section
is 12 ft long. Ans. Pu, = 28,200 lb.
and
What is the maximum length I that the pin-ended column in the preceding
problem can have tfj carry a compressive load P = 40,000 lb with a factor of safety
of 2? A 715 . / = 10 ft.

10.6 Empirical Column Formulas

Because of the objections to the transcendental nature of the secant
formula, many simpler but wholly empirical formulas have been proposed
 EMPIRICAL COLUMN FORMULAS 289

as substitutes for it. Such formulas usually give the allowable average
compressive stress {P/A)u, as a function of the slenderness ratio //r, with-
out specifically indicating the factor of safety that is used.
Straight Line Formula. One of the most commonly used empirical
column formulas is the straight line formula

(3).

where <Tw is an allowable working stress for the material in compression and
a is a numerical factor. For pin-ended, centrally loaded structural steel
columns, a Chicago Building Code recommends = 16,000 psi and a = 70
so that eq. (10.13) becomes

7 )/ w = 16,000 - 70 -• (10.13a)
A r

The use of this formula is limited to (30 < l/r < 120). For l/r < 30,
{P/A)u> = 14,000 psi and for l/r > 120, the Euler formula (eq. 10.8) should
be used with a factor of safety n = 2.7.
The Parabolic Formula. In order to make greater allowance for the effect
of imperfections for the more slender columns, a parabolic formula is often
used, such that

(j).
- ’ - i-r)’
where again an allowable working stress in compression and d is a
ctu- is

numerical factor.For centrally loaded pin-ended structural steel columns,

the American Institute of Steel Construction (1947) recommends =
17,000 psi and /3 = 0.485 so that

= 17,000 - 0.485 (^y- (10.14a)

The use of this formula is limited to (0 < l/r < 120). For l/r > 120, the
Euler formula (eq. 10.8) should be used with a suitable factor of safety.
Rankine-Gordon Formula. Still another empirical formula proposed for
pin-ended, centrally loaded columns is the Rankine-Gordon formula

(10.15)
290 THEORY OF COLUMNS
where is an allowable working stress in compression and y a numerical
factor. A common form of this formula for structural steel columns is

18,000
(10.15a)
1
^
H
—
18,000

to be used only for 0 < l/r < V20.

For purposes of comparison, each of the foregoing empirical formulas for
steel columns is represented graphically in Fig. 10.18, together with the
Euler curve (n = 2) and the secant formula (ay = .30,000 psi, ec/r^ == 0.2,
and n — 2). 'Faking the secant curve as a basis of comparison, we see that
is somewhat more conservative while
the straight line formula, eq. (10.13a),
both (he parabolic curve and the Rankine-Gordon curve are less con-
s(*rvati\e

Many empirical formulas for columns of materials other than structural

steel are available for design purposes. For example, a formula of the
straight line type intended for design of pin-ended aluminum alloy columns
in the aircraft industry is

Use of this formula is limited to the range 0 < l/r < 83. For l/r > 83, the
Euler formula is to be used.
For pin-ended cast-iron columns, the New Y'ork Building Code (1917)
an allowable average compressive stress
specifies

= 9000 - 40 1 (10.13c)

for 0 < l/r < 70. Owing to the weakness of cast-iron in tension due to
bending, it should not be used as a material for columns having l/r > 70.
 EMPIRICAL COLUMN FORMULAS 291

For wood columns of rectangular cross-section, the Forest Products

Laboratory has proposed a special formula based on the results of many
tests. For short columns^ having an unsupported length not greater than 10

times the smaller lateral dimension, the allowable average compressive

stress shall be

(10.16a)

where values of S for different varieties of timber are given in Table A.2 of
Appendix A, p. 343. For medium columns^

where K == E being the modulus of elasticity in bending as given

in the table. Use is limited to the range (10 < l/b < K).
of this formula
When l/b = if, we see that (P/A)u, = 25/3. For long columns (l/b > K),
Euler^s formula is to be used with a factor of safety n. = 3. Thus
P\ _ (10.16c)
^
A j. 36(i/6)’

Values of 5, P, and K
for common structural grades of timber, kept dry,
are to be found in the table mentioned above.
The empirical formulas discussed above are all intended for
centrally loaded columns, the reduction in allowable average com-
pressive stress with increasing l/r being intended only to allow^ for
imperfections. In the case of columns carrying bracketed loads or,
bending as well as compression, the Ameri-
in general, intentional
can Institute of Steel Construction has proposed an empirical
procedure which avoids the difficulties connected with the trial
and error solution of the secant formula. To explain this, we refer
to the pin-ended column in Fig. 10.19, where P is the axial load
and M is an end moment. Neglecting the effect of the lateral de-

flection 5 on the bending moment at C, the maximum compressive

stress would be ‘
Fiq, iq.iq

_ P ,
M ^

^ ^
where Z is the section modulus with respect to the principal axis about
which bending takes place Dividing both sides of this equation by crmay,
we obtain
P/A ^M/Z (a)
 292 THEORY OF COLUMNS
Now if only the centrally applied loads P were acting, i.e., if M * 0, we
could logically take (7max = o-c the allowable average compressive stress
given by one of the empirical column formulas, say the parabolic formula,
eq. (10J4a). On the other hand, if only the end moments M were acting,
i.e., if P = 0, we would take (Tmax = the allowable working stress for
the material in pure bending. For the combined loading, we arbitrarily
replace amax in the term on the left side of eq. (a) by (Tc, and (Tmax in the
first

second term by at, and then require that the sum of the two terms shall not
exceed unity. Thus

P/A M/Z
<Te
,

(Th
^ ^ (10.17)

The use of the lower working stress ae in the first term is presumed to correct •

for the additional bending moment P5. In calculating the value of (To from

the parabolic or other empirical formula, the least radius of gyration of the
cross-section should be used regardless of the plane of the end moments M.
EXAMPLE L Repeat the solution of Example 1 of Art. 10.4, using condition
(10.17) instead of the secant formula. The working stress in bending is ah * 20,000
psi,
SOLUTION. The total compressive load P =«« 160 kips and the end moments
M « 500 in.-kips. For a suitable section, we try first the 14WF68 section already
considered on p. 284. For this section « 20.0 in.*, Z * 103 in.*, and rmin * 2.46
in. Then l/r « 300 + 2.46 * 122 and the parabolic formula, eq. (10.14a), gives

a, = 17,000 - 0,486 (122)* « 9800 psi.

Accordingly the left side of expression (10.17 )become8

7500 4860
"
9800 20,000

This being so nearly equal to unity, we many consider the 14WF68 section satift-

factory.
EXAMPLE 2. Using the Forest Products Laboratory formulas, select a Douglas
Fir timber of rectangular cross-section to be used as a pin-ended column of length
Z * 14 ft if the axial load
60,000 lb. P*
SOLUTION. From Table A. 2 of Appendix A, p. 343, we find for Douglas fir:
8 « 880 psi and K
« 27.3. If the column were very short, the required cross-
sectional area would be A - P/S - 60,000/880 * 68.2 sq in. This then would
require an 8 X 8-in. cross-section, which suggests that for a long column we try
first a 10 X 10-in. section. Then from eq. (10.16b)

838 psi.
 EMPIRICAL COLUMN FORMULAS 293

With this working stress, the safe load P* 838 X 100 = 83,800 lb, which is too
large. For an 8 X 10-in. section, eq. (10.16b) gives

and the safe load is 777 X 80 * 62,200 lb. This indicates that the 8 X lO-in.
section is adequate.

PROBLEMS
1. WF-beam section to be used
Using the A.I.S.C. formula, eq. (10.14a), select a
as a fixed-end column 25 long if the axial load P * 200,000 lb. Arw. 12WF50.
ft

2. Using the A.I.S.C. column formula, eq. (10.14a), find the safe axial load P for
a steel pipe 16 ft long and having pinned ends if the outside diameter is 12.75 in. and
the inside diameter is 12.00 in. Ana. P = 234,000 lb.
3. Solve the preceding problem, using the straight line formula, eq. (10.13a).
Ana. P « 188,500 lb.
4. A 12WF45 beam section is to be used for a cantilever column built-in at the
lower end, free at the top, and 8 ft long. Using the Rankine-Gordon formula, find
the safe load P axially applied. Ana. P •= 154,000 lb.
5. Using eq. (10,13b), with a factor of safety n * 2.0, find the maximum length
I of a pin-ended aluminum alloy pipe (outside diameter 3 in. and wall thickness i in.)

to carry an axial load P = 15,000 lb. Ana. L.x = 48 in.

6. A steel pipe having a wall thickness « J in. is to be used as a pin-ended column
t

18 ft long to carry an axial load P «= 100 kips. Using the straight line formula, eq.
(10.13a), find the required outside diameter of the pipe. Ana. d = 7.14 in.
7. hollow cast-iron pipe with wall thickness ^ = 1 in. is to be used as a pin-
A
ended column 12 ft long and is to carry an axial load P =* 175,000 lb. Find the
required outside diameter of the pipe. Ana. d = 9 in.
8 . What is the safe load P for a 10 X 12-in. shortleaf pine column 25 ft long if it
has pinned ends? Ana. P « 58,500 lb.
9. What is the safe axial load P for an 8 X 10-in. spruce column if it is (a) 6 ft
long, (b) 12 ft long? Ana. (a) 51,200 lb.; (b) 48,200 lb.

10. A 12WF72 beam section serves as a pin-ended column 10 ft long. It carries

an axial lead Pi * 100 kips and an eccentric load Pa acting on axis 1-1, in. from 4
axis 2-2. Using condition (10.17), find the safe magnitude of the eccentric load Pa-
Ana. Pi - 77,200 lb.
11. I^lect a 12-in. WF steel section to serve as a pin-ended column 16 ft long if it

is to carry a load P
80,000 lb. with eccentricity e 6 in., bending in the plane of
the web. Use condition (10.17) with (Xc determined from the parabolic column
formula and o-t = 20.000 psi. Ana. 12WF45.
12. A 3U60 steel channel section is to be used as a pin-ended column 4 ft long.
The compressive load P is applied on axis 1-1 in the middle plane of the web.
Using eq. (10.17), with ae calculated from the Rankine-Gordon formula and ffh
18,000 psi, find the safe load P. Ana P— 8900 lb.
 CHAPTER XI
MECHANICAL PROPERTIES OF
MATERIALS

11.1 The Tensile Test

The preceding chapters dealt with the methods of analyzing the stress
distribution produced by various kinds of forces on structures. Knowing
the stresses, the designer must then select the material and the dimensions
of the structure in such a way that it will safely withstand various loading
conditions in service. For this purpose it is necessary to have information
regarding the elastic properties and strength characteristics of structural
materials under various stress conditions. The designer must know the limits
under which the material can be considered as perfectly elastic for various
stress conditions, and also the behavior of the material beyond those limits.
Information of this type can be obtained only by experimental investigations.
Material.s-testing laboratories are equipped with testing machines* which
produce certain typical deformations of test specimens, such as tension,
compres.sion, torsion, and bending.
Experiments show that test results are sometimes affected by the size
and shape of the test specimen. Thus to make the results of tests com-
parable, certain proportions for test specimens have been established and
are recognized as standard. The most widely used of all mechanical tests of
structural materials is undoubtedly the tension test. The standard tensile
test specimen in the United States Ls circular, with i-in. diameter and 2-in.
gage length, so that

= 4 or J = 4.5lV^,
^

where A — '7rd-/4 is the cross-sectional area of the specimen.

*For a description of materials-testing machines and a bibliography on the subject see

the article by J. Marin in M. Hetenyi ed., Handbook of Experimental Stress Analysts,
New York, 1950.
294
 THE TENSILE TEST 295

The length of the cylindrical portion of the specimen is always somewhat

greater than the gage length I and is usually at least, I + d. The ends of the

specimen are generally made with a larger cross-section in order to prevent

the specimen from breaking in the grips of the testing machine, where stress
conditions are more severe because of local irregularities in stress distribu-
tion. A cylindrical specimen with — I JOd is shown in Fig. 11.1, which also
shows the spherical seats in the grips of the machine, used to insure central
application of the load.
Tensile test machines are usually provided with a device which auto-
matically draws a tensile diagram representing the relation between the
test

load P and the extension the specimen. Such a diagram exhibits im-
5 of

portant characteristics of the material. Fig. 11.2, for example, shows a

series of tensile test diagrams for carbon steel with various contents of
carbon. It can be seen that as the carbon content increases, the ultimate
strength of the steel also increases, but at the same time the elongation
before fracture decreases and the material has less ductility. High-carbon
steel is relatively brittle. It follows Hooke's law to a high value of stress
and then fractures at a very small elongation. On the other hand, a mild
steel with a small carbon content is ductile and stretches considerably
before fracture.
Fig. 11.3 represents the tensile test diagram for mild structural steel.

From diagram the important characteristics such as yield

this point,
ultimate strength^ and amount of plastic elongation can be obtained.
296 MECHANICAL PROPERTIES OF MATERIALS
In determining the proportional limit sensitive extensometers are
necessary in order to detect the slightest deviation from a straight line in
the tensile test diagram. Obviously the position found for this limit
depends considerably on the sensitivity of the instruments. In order to
obtain greater uniformity in results, a specified amount of permanent aet
or a certain deviation from proportionality is often taken as the basis for
determining the proportional limit. The International Congress for Testing
Materials at Brussels (1906) defined the proportional limit as the tensile
stress at which the permanent set is 0.001 per cent.
The is a very important characteristic for structural steel. At
yield point
the yield point stress, the specimen elongates a considerable amount with-
out any increase in load. In the case of mild steel this elongation may be
more than 2 per cent. Sometimes yielding is accompanied by an abrupt
decrease in load, and the tensile test diagram has the shape shown in Fig.
11.3. In such a case the upper and lower limits of the load at a and 6,
divided by the initial cross-sectional area, are called the upper and lower
yield points, respectively. The position of the upper yield point is affected
by the speed of testing, by the form of the specimen and by the shape of the
cross section. The lower yield point as already discussed in Art. 2.2 is

usually considered a true characteristic of the material and therefore is used

as a basis for determining working stresses.
Owing to the relatively large stretching of the material at the yield point
it is not necessary to use sensitive extensometers to determine this point. It
can be determined with the simplest instruments or can be taken directly
from the tensde test diagram. For structural carbon steel the stress at the
yield point is about 55-60 per cent of the ultimate strength. Structural
steel with about 1 per cent silicon has a yield point stress about 70-80 per
cent of the ultimate strength. The ultimate strength for the silicon steel is

about the same as for the carbon Such a high value for the yield point
steel.

justifies the usual practice of taking higher working stresses for silicon steel.
A sharply defined yield point
is a characteristic not only of structural

steel but also of materials such as bronze and brass. There are other
materials, however, which do not have a pronounced yield point. For
these materials the stress at which the permanent set reaches the value 0.2
per cent is sometimes arbitrarily called the yield point. It must be kept in
mind that the yield point defined in this manner does not represent a
definite physical characteristic of the material but depends upon the
arbitrarily chosen permanent set.

The ultimate strength is usually defined as the stress obtained by dividing

the maximum load on the specimen (point c in Fig. 11.3) by the initial cross-
sectional area. This quantity also is often taken as a basis for determining
the working stresses.
 THE TENSILE TEST 297

The area under the tensile test diagram Oacde (Fig. 11.3) represents the
work required to produce fracture. This quantity is also used as a char-
acteristic property of the material and depends not only on the strength but
also on the ductility of the material.
The ductility of a metal is usually considered to be characterized by the
elongation of the gage length of the specimen during a tensile test and by the
reduction in area of the cross-section where fracture occurs. In the first
stage of plastic elongation, from a to c in Fig. 11.3, the specimen elongates
uniformly along its length. This uniform elongation is accompanied by a
uniform lateral contraction, so that the volume of the specimen remains
practically constant.* At point c the tensile force reaches a maximum
value, and further extension of the specimen is accompanied by a decrease
in the load. At this stage of plastic elongation the deformation becomes
localized and necking begins, the specimen taking the shape shown in
Fig. 1 1.4. It is difficult to determine accurately the moment when
necking begins and thereby establish separately the magnitude of
the uniform stretching and the magnitude of the elongation due
to necking. It is therefore customary to measure the total increase
in the gage length after the specimen has fractured. The elonga-
tion is then defined as the ratio of this total elongation of the gage
length to its initial length. In practice the elongation at fracture
is usually given in percentage. If I is the original gage length and
5 the total elongation, the elongation at failure in percentage is

X Fig. 11.4
€ = y 100. (a)

This elongation is usually taken as a measure of the ductility of the material.

Elongation obtained in this manner depends on the proportions of the
specimen. The increase in the gage length due to necking is a large part of
the total increase and is practically the same for a short gage length as for a
long gage length. Hence the elongation defined by eq. (a) becomes larger as
the gage length decreases For steel, the elongation obtained for specimens
with I = about 1.22 times the elongation for a specimen of the same
5d is

material with = lOd. Experiments also show that the shape of the cross-
I

section affects the local deformation at the neck and hence affects the
elongation of the specimen. This shows that comparable results With
respect to elongation can be obtained only by using geometrically similar
specimens.

*The small elastic deformation in which the volume does change can he neglected in
comparison with the comparatively large plastic deformation.
298 MECHANICAL PROPERTIES OF MATERIALS
The reduction in area at the cross-section of fracture is expressed as
follows:

Q = (b)

in which Ao is the initial cross-sectional area and the final cross-sectional

area at the section where fracture occurs.

11.2 Yield Point

The early portion of the tensile test diagram in Fig. 11.3 is show'n to a
larger scale in Fig. 11.5, as a stress-strain diagram. The shape of this
diagram depends noticeably on the mechanical arrange-'
at the yield point
ment of the testing machine. If extension of the specimen is produced by an
increase of distance between the grips of the machine moving at a uniform
speed, the sudden plastic stretching will somewhat decrease the tensile
force in the specimen, and a sharp peak A in the diagram will be obtained.
If an elastic spring is inserted in series with the specimen, the slope of the

curve AB of the diagram can be reduced as shown by the broken line AB\.
On the other hand if the tensile load is applied directly to the specimen, the
tensile force at yielding will be affected by the inertia of the load in sudden
motion, and small vibrations may appear on the diagram.
In order to study in more detail the de-
formations w^hich occur at the yield point,
specimens with polished surfaces have been
used. Such experiments show that at the
time the tensile stress drops from point A to
point B (Fig. 11.5) fine, dull lines begin to
appear on the surface of the specimen.
These lines are inclined about 45° to the
direction of tension and are called Lueders^
lines* (see Fig. 2.2, p. 27). With further
Fig. 11.5 stretching, the lines increase in width and
in number, and during stretching from B to
Bi they cover the entire surface of the specimen. Instead of polishing,
sometimes special paints (called stress coats) are used to indicate Lueders’
lines.The paints are brittle and cannot sustain large deformations; hence
they crack during loading and indicate the pattern of Lueders' lines.
Studies V ith a microscope show that Lueders' lines represent the inter-

•These lines were first described by W. Lueders, Dinglers Polytech. J., 1854
 YIELD POINT 299

sections with the lateral surface of the specimen of thin layers of material
in which plastic deformation has occurred while the adjacent portions of the
material remain perfectly clastic. By cutting the specimen and using a
special etching, the thin plastic layers in the interior of the specimen can be
made visible. Under a microscope it is seen that these layers consist of
crystals which have been distorted by sliding.
Experiments show that the values of the yield point stress and the yield
point strain depend upon the rate of strain. The curves in Fig. 11.6 show
stress-strain diagrams for mild steel for a wide range of rates of strain
{u = de/dt = 9.5 X 10~^ per second to = 300 per second). It is seen that
not only the yield point but also the ultimate strength and the total elonga-
tion depend greatly upon the rate of strain. In general, these quantities in-
crease as the rate of strain increases.
To explain the sudden stretching of steel at its yield point, it has been
suggested* that the boundaries of the grains consist of a brittle material
and form a which prevents plastic deformation of the grains
rigid skeleton
at low stress. Without such a skeleton the tensile test diagram would be
like that indicated in Fig. 11.7 by the broken line. Owing to the presence

Fig. 11.6 Fig. 11.7

*See P. Liidwik and Scheu, "Werkstoifanschuss,” Ver. deiU. Ing. Ber., No. 70, 1925.
300 MECHANICAL PROPERTIES OF MATERIALS
of the rigid skeleton, the material remains perfectly elastic and follows
Hooke’s law up to point A, where the skeleton breaks down. Then the
plastic grain material suddently obtains the permanent strain AB, after
which the material follows the usual curve BC for a plastic material. This
theory explains the condition of instability of the material at the upper
yield point. It also accounts for the fact that materials with small grain
size usually show higher values for the yield stress. As a result, such
materials undergo more stretching at the yield point, as defined by the
length of the horizontal line AB in Fig. 11.7. In addition, the theory
explains the fact that in high-speed tests the increase in 3nield point stress
is accompanied by an increase in the amount of stretching at yielding,
as shown by the curves in Fig. 11.6.

11.3 Stretching of Steel Beyond the Yield Point

During stretching of a steel specimen beyond the yield point, the material
hardens and the stress required for stretching the bar increases as shown
by the portion BC of the stress-strain diagram, in Fig. 11.8. Elongation
of the specifnen is combined with uniform reduction of the cross-sectional

I I \ I I J I c
0 10 20 30
Elongation (per cent)

Fig. 11.8

area so that the volume of the specimen remains practically constant.

The work done during stretching is transformed largely into heat, and
the specimen. becomes hot. Calorimetric measurements show that not all of
the mechanical energy is transformed into heat, however; part of it remains
in the specimen in the form of strain energy. Owing to differences in
orientation of the crystals, the stresses are not uniformly distributed
 STRETCHING BEYOND YIELD POINT 301

over the cross sections, and after unloading, some residual stress and a
certain amount of strain energy remain in the specimen.
If after unloading we load the specimen a second time, we will find
that its yield point stress is raised. This characteristic is shown in Fig.
11.9, which represents a tensile test diagram for mild steel. After stretching
the bar to the point C
was unloaded. During unloading, the material
it

followed approximately a straight-line law, as shown by the line CD on

the diagram. When the load was applied to the bar a second time, the
material again followed approximately Hooke’s law and the line DF wsu9
obtained. At point F, which corresponds to the previous loading at C,
the curve abruptly changed character and traced the portion FG, which can
be considered a prolongation of the curve BC. This represents a raising
of the 3deld point due to previous stretching of the material. If several
days are allowed to elapse after the first unloading, then upon reloading a
still higher yield point may be obtained, as indicated by the dotted line

at F'. Fig. 11.10 shows the results of a tensile test of die-cast aluminum.

Elongation ( por cent

Fig. 11.9 Fig. 11.10

302 MECHANICAL PROPERTIES OF MATERIALS
The initial proportional limit of the material was 5600 psi. After stretching
the specimen 2 per cent, the proportional limit upon reloading was found
to be 20,000 psi and the yield point about 21,000 psi.
More complete investigations show that the time which elapses between
unloading and reloading has a great influence on the stress-strain curve
during reloading. If reloading begins immediately after unloading, ac-
curate measurements show that there are
deviations from the straight-line law at
very low stress, and the proportional
limit is greatly lowered. But if a con-
siderable interval of time elapses between
unloading and reloading, the material
recovers its elastic properties completely.
Fig. 11.11 shows curves obtained for mild
steel which indicate that if reloading
follows in ten minutes after overstrain,
the material does not follow Hookers law,
but after five days it has partially re-
covered its elasticity and after twenty-
Fig, 11.11
one days it has almost completely re-
covered it.

Experiments also show that if the material is subjected to mild heat

treatment after unloading, say bath of 100°C, the recovery of elastic
in a

properties occuirs in a much shorter interval of time. Fig. 11.12 shows the
results of tests made on a steel bar. The initial tensile test is represented
by the curve A. Curve B represents reloading of the same bar ten minutes
after unlording, and considerable deviation from Hooke's law is noticeable.

Elongation ( per cent

Fig. 11.12
 STRETCHING BEYOND YIELD POINT 303

Curve C is the diagram obtained with the same bar after a second unloading
and after heat treating at 100®C for four minutes. In this case the material
has completely recovered its elastic properties.
The phenomenon of strain hardening due to plastic deformation is
encountered in many technological processes such as rolling bars and
drawing tubes and wires at low temperature, cutting sheet metal by shears
and drawing, and punching holes. In all these cases the part of the material
which undergoes plastic deformation becomes harder, and its ductility
is greatly reduced. To eliminate this undesirable effect of strain hardening
it is customary to anneal the material, which restores the initial ductility.

Sometimes the strain hardening of ductile materials is of practical use

in manufacturing. It is common practice to subject the chains and cables
of hoisting machines to a certain amount of overstrain in order to eliminate
undesirable stretching of these parts in service. The cylinders of hydraulic
presses are sometimes subjected to an initial internal pressure sufficient
to produce permanent deformation of the walls. The strain hardening
and the produced by this pressure prevent any permanent
residual stresses
set in service. Overstraining of the metal is sometimes u.sed in the manu-
facture of guns. By stretching the metal in the wall of a gun beyond the
initial yield point and afterwards subjecting it to a mild heat treatment,

the elastic properties of the material are improved and at the same time
initial stresses are produced which combine with the stresses produced by

the explosion to give a more favorable stress distribution. Turbine discs

and rotors are sometimes given an analogous treatment. By running these
parts at overspeed, a permanent set is obtained around the central hole,
which raises the yield point of the material and produces initial stresses
which are in a favorable direction. Die-cast aluminum fans are sometimes
subjected to overstrain at the bore to prevent any possibility of their
loosening on the shaft in service. Considerable plastic flow of the metal is
sometimes produced in pressing the hubs of locomotive wheels onto their
axles, and this has proved to have a favorable effect. Copper bars in the
commutators of electric machinery are subjected to considerable cold work
by drawing in order to give them the required strength.
In using overstrain in this manner to raise the yield point and improve
the elastic properties of a structure, it is necessary to keep in mind: (1) that
the hardening disappears if the structure is subjected to annealing tempera-
tures and (2) that stretching the metal in a certain direction, while making
it stronger with respect to tension in that direction, does not proportionately
improve the mechanical properties with respect to compression in the
same direction.
The fact that stretching a metal in a certain direction does not improve
the mechanical properties in compression in the same proportion as it
304 MECHANICAL PROPERTIES OF MATERIALS
does in tension must not be overlooked in cases in which the material is

subjected to reversal of stresses. mentioned that there

It should also be
are indications that material which has yielded in a particular region is
more sensitive in that region to chemical action, and there is a tendency
for corrosion to enter the metal along the surfaces of sliding. This phe-
nomenon is of particular importance in the case of boilers and other
containers subjected simultaneously to stress and to chemical action.
In constructing a tensile test diagram such as curve ABC in Fig. 11.8 the
tensile load is usually divided by the initial cross-sectional area Ao of the
specimen in order to obtain the conventional unit stress. But for large
stretching there will be a considerable reduction in cross-sectional area;
and to obtain the true stress the actual area A, instead of Ao, should be
used. From the constancy of the volume of the specimen we have

lA, A - ^ (a)

and the true stress is

,.5. £(,+.), (b)

To obtain the true stress diagram the ordinates of the conventional dia-
gram must be multiplied by 1 -f- €. In Fig. 11.8 such a diagram is shown by
the broken line. It extends as far as a vertical through point C, where the
load reaches its maximum value. On further stretching of the specimen,
local reduction of the cross-section (necking) begins and e is no longer
constant along the specimen. Then eq. (b) is no longer applicable, since
the stresses over the minimum cross-section are not uniformly distributed.
In such a case, eq. (b) gives only an average value of <r. The average unit
elongation € at the minimum section may be found from eq. (a), which gives

(0

Using the symbol q for the unit reduction of the cross-sectional area, eq. (b),

p. 298, we obtain

A Ao{l - q);

and eq. (c) gives

From eq. (d) the unit elongation at the minimum section can be readily
calculated if the reduction in area of that section is measured. This
quantity is called the effective elongation and is much larger than the
elongation c = S/I determined from the total elongation S of the gage length.
 FRACTURES IN TENSION 305

11.4 Types of Fractures in Tension

In discussing fractures, we distinguish between (1) brittle fracture as in

the case of cast iron or glass, and (2) shear fracture^ as in the case of mild
steel, aluminum and other metals. In the first case, fracture occurs practi-
cally without plastic deformation over a cross-section perpendicular to the
axis of the specimen. In the second case fracture occurs after considerable
plastic stretching and has the familiar cup~cone form, shown in Fig. 11.13.

Fio. 11.13

In discussing these two kinds of fracture, the theory has again been for-
warded that the strength of the material can be described by two char-
acteristics, the resistance to separation and the resistance to sliding. If the
resistance to sliding is greater than the resistance to separation, we have a
brittle material, and fracture will occur as a result of overcoming the
cohesive forces without any appreciable deformation. If the resistance to

Table 11.1 ULTIMATE STRENGTH OF CYLINDRICAL AND

GROOVED SPECIMENS
Ultimate Strength (lb per sq in.)

6 Carbon Steel 1
1
Nickel Chrome Steel
(inches) Computed Computed Computed Computed
from from '
from from
Original Reduced Original Reduced
Area Area Area Area
1/32 163,000 176,000 103,000
1/16 164,000 177,000 184,000
1/8 143,000 158,000 154,000
Norma) specimen 102,000 227,000 108,000 1
348,000
306 MECHANICAL PROPERTIES OF MATERIALS
separation is larger than the resistance to sliding, we have a ductile material.
Then sliding along inclined planes begins first, and the cup-cone fracture
occurs only after considerable uniform stretching and subsequent local re-
duction of the cross-sectional area (necking) of the specimen.
The preceding discussion refers only to tensile tests of
standard circular specimens of cylindrical shape. The results
obtained with other shapes of specimens are quite different,
I I

H N as illustrated by the grooved specimen shown in Fig. 11.14.

I I

1 1
During a tensile test, reduction of the cross-sectional area at
I I

the grooved section is partially prevented by the presence of

I I
IT the portions of larger diameter D. It is natural that this action
I I

should increase as the width 5 of the groove decreases. Table

11.1 gives the results of tests obtained with two different
materials: (1) carbon steel with proportional limit 56,000
psi, yield point 64,500 psi, ultimate strength 102,000 psi,
elongation 26i per cent, reduction in area 55 per cent; and (2)
Fig. 11.14
nickel chrome steel with proportional limit, 80,000 psi, yield
point 85,000 psi, ultimate strength 108,000 psi, elongation 27 per cent,
reduction in area 69 per cent. These figures were obtained from ordinary
tensile tests on normal cylindrical specimens with 1/2-in-diameter and
2-in.-gage length. The original cross-sectional area was used in calculating
the stresses. The grooved specimens of the type shown in Fig. 11.14 had
d = i in. and D — in.

The table shows that in all cases the breaking load for the grooved speci-
mens was larger than for the corresponding cylindrical specimens. With
the grooved specimens only a small reduction in area took place, and the
appearance of the fracture was like that of brittle materials. The true ulti-
mate strength of the cylindrical specimens was larger than for the grooved
specimens because fracture of the cylindrical specimens occurred after
considerable plastic flow. This resulted in strain hardening and increased
not only the resistance to sliding but also the resistance to separation.
Similar conditions are sometimes encountered in engineering practice.
An effect analogous to that of the narrow groove in Fig. 11.14 may be
produced by internal cavities in large forgings, such as turborotors. Ther-
mal stresses and residual stresses may combine with the effect of the stress
concentration at the cavity to produce a crack. The resulting fracture will
have the characteristics of a brittle failure without appreciable plastic flow,
although the material may prove ductile in the usual tensile tests.
Because most of the grooved specimen remains elastic during a tensile
test to failure, it will have a very small elongation, and hence only a small
amount of work is required to produce fracture, A small impact force
can easily supply the work required for failure. The specimeri is brittle
 FRACTURES IN TENSION 307

l)ecause of its shape, not because of any mechanical property of the material.
Illmachine parts subjected to impact all sharp changes in cross-section are
dangerous and should be avoided.

11.5 Compression Tests

The compression test is commonly used for testing brittle materials such
as stone, concrete, and cast-iron. The specimens used in the tests are
usually made in either cubic or cylindric shape. In compressing the speci-
mens between the plane surfaces of the testing machine it is normally as-
sumed that the compressive force is uniformly distributed over the cross-
section. The actual stress distribution is much more complicated, even if
the surfaces are in perfect contact and the load is centrally applied. Owing
to friction on the surfaces of conta(‘t between the specimen and the heads of
the machine, the lateral expansion which accompanies compression is pre-
vented at these surfaces and the material in this region is in a more favor-
able stress condition. As a result, the type of fracture obtained in a com-
pre.ssion test of a cubic specimen of concrete is as shown in Fig. 11. 15. The
nuiterial in contact with the machine reiniiins unaffected* w’hile the material
at the sides is crushed out.

Fig. 11.15

In order to obtain the true resistance to compression of a matiTial such as

concrete, the influence of friction at the surfaces of contact mu.st be elimi-
nated or minimized. this purpose A Foppl covered the surfaces
For
and found that the ultimate strength was then
of contact with paraffin
greatly reduced. The type of failure was completely different, and cubic
308 MECHANICAL PROPERTIES OF MATERIALS
specimens failed by subdividing into plates parallel to one of the lateral
sides. Another method of eliminating the effect of friction forces is to use
specimens in the form of prisms having a length in the direction of com-
pression several times larger than the lateral dimensions. The middle
portion of the prism then approaches the condition of uniform compression.
A very interesting method of producing uniform compression on cylindrical
specimens as developed in. the Kaiser- Wilhelm Institut is shown in Fig.
11.16. The head pieces of the testing machine and the ends of the cylindri-

Fig. 11.16

cal specimen are machined to conical surfaces with the angle a equal to the
angle of friction. Thus the effect of friction is compensated for by the
wedging action, and uniform compression results.
Compression tests of materials such as concrete, stone, and cast-iron
show that these materials have a very low proportional limit. Beyond the
proportional limit the deformation increases at a faster rate relative to the
load, and the compression test diagram has the shape shown in Fig. 11.17.
Sometimes it is desirable to have an analytical expression for such a dia-
gram. For these cases Bach proposed an exponential law given by the
equation

€ = (a)
e'

in which n is a number depending on the properties of the material. Bach

found the values n = 1 .09 for pure cement and n = 1.13 for granite.
Compression tests of ductile materials show that the shape of the diagram
depends on the proportions of the specimen. As the dimension in the direc-
tion of compression decreases, the effect of friction at the ends becomes more
pronounced and the compression test diagram becomes steeper. For
example, Fig. 11.18 shows the results of compression tests on copper
cylinders with various ratios d/h of the diameter to the height of the speci-
 COMPRESSION TESTS 309

men. In compression tests of such ductile materials as copper, fracture

isseldom obtained. Compression is accompanied by lateral expansion
and a compressed cylinder ultimately assuim's the shape of a flat disc.

11.6 Tests of Materials Under Combined Stresses

Leaving the discussion of simple tension and compression tests, let ns

now (U)nsider cases in which the materials are tested under combined
stresses. We begin with a discus.sion of materials tested under uniform
hydrostatic pressure.* Such tests show that, under uniform pressure,
homogeneous materials can sustain enormous compressive stresses and
remain elastic. Tests also show that large pressures produce only small
changes in volume.
Several attempts have been made to produce uniform hydrostatic
tension of materials, but up to now there has not been a satisfactory solution
to this interesting problem
Tensile tests of various steels combined with lateral pressure have show'n
that the pressure lias a great effect on the shape of the neck and on the
reduction in area at the minimum cross-section. Fig. 11 111 shows the
yoke arrangement by whicli tension was applied to specimens within the

Fio. 11.19

•The most comprehensive tests of this kind were made by P. W. Bridgman, who de-
veloped a technique for obtairiing enormous pressures, see his books, The Physics of High
VressurBy New York, 1931, and Studies in Large Plastic Flow and Fracture, New York,
1952. A new triaxial stress-testing machine was described by H. A. B. Wiseman and
Joseph Marin, Proc. Am. Soc. Test. Mat., Vol. 54, 1954.
 310 MECHANICAL PROPERTIES OF MATERIALS
pressure vessel, l^^igs. 11.20a and h illustrate fractures of medium carbon
steel (0.45 per cent carbon) at atmospheric pressure and at a lateral pressure
of 145,000 psi. Ill the first case the average true stress was 114,000 psi. In
the second case the corresponding value was 474,000 psi.It was also found
that with an increase in lateral pressure the relative extent of the flat part
at the bottom of the cup-cone fracture diminishes; at a certain pressure it

entirely disappears, and the fracture becomes entirely shear fracture.

(a)

(b)

Fig. 11.20

The combination of axial compression and lateral pressure was used by

Th. V. Kirmdn in compression tests of marble. These tests showed that
with increasing lateral pressure marble becomes more and more plastic, and
 TESTS UNDER COMBINED STRESSES Ml

initially cylindrical spK'cimons may obtain barreled f(»rms, as shown in

Fig. 11.21.
In studying two-dimensional stress conditions, thin-walled cylindrical
tubes have been tested. By subjecting a lube to axial tension combined
with internal pressure, the yield point stress for various ratios of the two
principal stresses was established for several materials, including iron,
copper, and nickel. The results obtained in this way were in satisfactory
agreement with the maximum distortion energy theory to be discussed in
Art. 11.7.

Fig. 11.21

In practical applications not only the yield point siress but also the
ductilityand strain hardening are of great importance in cases of combined
stresses. Unusual cases of failure, such as explosions of large spherical
storage tanks and sudden cracks in the hulls of welded cargo ships, have
recently called attention to these subjects. In both these types of failure,
low-carbon steel plates were used which showed satisfactory strength and
ductility in ordinary tensile tests. But the fractured surfaces of the plates
in the exploded pressure vessels and in the damaged ships did not shcrw
plastic deformation and had a brittle character. Most of these failures oc-
curred at low atmospheric temperatures and under two-dimensional stress
conditions.
312 MECHANICAL PROPERTIES OF MATERIALS
In order to determine the influence of temperature and two-dimensional
stress on the strength and ductility of low-carbon steel, a considerable
amount of experimental work has been done in recent times in various
laboratories. Thin-walled tubes were used to produce two-dimensional
stress (•()nditions. These tubes were subjected simultaneously to axial
tension and internal hydrostatic pressure, so that tensile stresses at in the
circumferential direction and aa in the axial direction could be produced in
any desired ratio a = at>’a,^. Using tubes of medium-carbon steel (0.23 per
cent carbon) with 1.450-in. outside diameter and 0.100-in. wall thickness,
E. A. Davis made tosts’*^ with five difTerfuit values of the ratio rn Fig. 11 .22
shows the types of fractures obtained f or the small values of the ratio n

Fig. 11.22

•See E. A. Davis (Westinghoiis^* Reasearch Laboratories), J. Appl. Mech. ^Vol. 12, p. 13,
1945,and Vol. 15, p. 216, 1048
 TESTS UNDER COMBINED STRESSES 31.3

the cracks were circumferential, and for the larger values they were
longitudinal. By making an additional series of tests it was established
that the transition from one type of failure to the other occurred at the
value n = 0.76. It was found that in the case of circumferential cracks the
fracture occurred along the planes of maximum shearing stress and at true
stresses of about the same magnitude as specimens pre-
in the case of flat
pared from the same material as the tubes. In the case of the longitudinal
cracks, rupture appeared to be more brittle. Failure usually started along
the planes of maximum shearing stress, but owing to high stress concentra-
tion at the crack ends, it continued as brittle fracture in the axial plane
without substantial plastic deformation. The maximum shearing stress at
which the longitudinal cracks began was always much smaller than in the
case of circumferential cracks. It seems that the differences in the two
fractures were due largely to the shape of the specimens. In the case of
circumferential cracks the material was much more free to neck down than
in the case of longitudinal cracks and therefore the latter occurred with
smaller local deformation and smaller decrease in load beyond the ultimate
strength.
In experiments at the Univ'ersity of California* tests were made at two
different temperatures using thin-walled tubes of low-carbon steel. The
diameter of the tubes was 5^ in. and the temperatures were 70® F and
— 138® F. The room temperature always gave a shear type of
tests at
fracture with considerable plastic deformation. The tests at low tempera-
ture (with n = 1) showed brittle fracture with very small plastic deforma-
tion. This brittleness was attributed to the local stresses at the welded
junctions of the tubes with the end connections.
After these tests with small tubes, large-size tubular specimens of 20-in.
outside diameter and 10-ft length, made of i~in. ship plate, were tested at
70° F and at —40° F, The tests at low temperature, especially with the
ratio n ~ 1, showed brittle fracture at stresses much smaller than those
obtained from tensile tests of ordinary cylindrical specimens made of the
same material.

11.7 Strength Theories

The mechanical properties of structural materials are normally deter-

mined by which subject the specimen to comparatively simple stress
tests
conditions. The strength of materials under more complicated stress
conditions has only been investigated in a few exceptional cases, such as
those discussed in the preceding article.

•See H. E. Davis and E. R. Parker, J. Appl. Mech ,

Vol. 15, p. 201, 1948.
314 MECHANICAL PROPERTIES OF MATERIALS

In order to determine suitable allowable stresses for the complicated

which occur in practic^al design, various strength theories
stress conditions
have been developed. The purpose of these theories is to predict failure
conditions under combined stresses, assuming that the behavior in a simple
tension or compression test is known. By failure of the material is meant
either yielding or actual rupture, whichev^er occurs first.

The most general state of stress which

can exist in a body
is always completely

determined by specifying the principal

stresses ai, <t 2
,
and <73 (Fig. 11.23). In
the following discaissioii, tension is con-
sidered positive and compression negative,
and the axes in Fig. 1 1 .23 are chosen so
that the relations between the algebraic
values of the principal stresses are

<7l > a2 > (a)

The maximum stress theorij considers the

maximum or minimum principal stress
as the criterion for strength. F or ductile materials this means that yielding
begins in an element of a stressed body when cither the maximum stress
reaches the yield point in simple tension or the minimum stress reaches
the yield point in simple compression. Thus the conditions for yielding are

(^l)y p ^ p.» ur ](o^3)y.p.! “<^3 1)

(IFl)

in which Cy p. and a\. „ are the yield point stresses in simple tension and
compression, res])ectiv(‘ly. Ther(‘ are many examples which contradicr
the maximum stress theory. It has already been pointed out that in simple
tension sliding occurs along planes inclined at 45° to the axis of the specimen
For these planes neither the tensile nor the compressive stresses are maxi-
mum, and failure is caused by shear stres^s instead. It has also been
pointed out that a homogeneous and isotropic material, even though weak
in simple compression, may sustain very large hydrostatic pressures with-
out yielding. This indicates that the magnitude of the maximum stress is

not sufficient to determine the conditions for yielding of the material or

its fracture.

A second strength theory is the maximum strain theory. In this theory it

is assumed that a ductile material begins to yield either when the maximum
strain (elongation) equals the yield point strain in simple tension or when
the minimum strain (shortening) equals the yield point strain in simple
compression.
 STRENGTH THEORIES 315

Observing that stress in one direction produces lateral deformation in the

other two perpendicular directions, and using superposition, we find for the
three principal strain components of the element in Fig. 1 1.23, the following
expression.s:

*2 = ~ m(<^i + o-s)], ( 11 2 )
.

€3 = ^[<rs - mCo*! + CTj)].

With condition fa), the first of these equations represents the maximum
strain and the third, the minimum strain. Substituting the yield point
strains o-y.p./J? in tension and in compression for €i and cs, the
criterion of failure according to the maximum strain theory becomes

<T\ — + (Ti) = CT-y.p.

I

or y (11.3)
1(73 - m(<^1 + <^ 2 )! = <^'y.p.J

There are many cases in w^hich the maximum strain theory may also be
shown to be invalid. For example, if a plate is subjected to equal tensions
in two perpendicular directions, the maximum strain theory indicates that
the tensile stress at yielding will be higher than the yield point in simple
tension. This result is obtained because the elongation in each direction is

decreased by the tension in the perpendicular direction. However, this

conclusion is not supported by experiments. Tests o<* materials under
uniform hydrostatic pressure also contradict this theory. For this case, the
second of eqs. (11.3) gives

_ ^y p> .

^3
I I

|y p. - j
_

in which era represents the hydrostatic pressure. Experiments show that

homogeneous materials under uniform compression can withstand much
higher stresses and remain elastic.
The maximum shear theory gives better agreement with experiments, at
least for ductile materials which have Uy p = o''y.p theory as.sumes -

that yielding begins when the maximum shear stress in the material becomes
equal to the maximum shear stress at the yield point in a simple tension
test. Since the maximum shear stress in the material is equal to half
316 MECHANICAL PROPERTIES OF MATERIALS
the difference between the maximum and minimum principal stresses,
and since the maximum shear stress, in a tension test, is equal to half
the normal stress, the condition for yielding becomes

(Ti — <Ts = (11.4)

This theory is in good agreement with experiments and is widely used in

machine design for ductile materials.
To compare the preceding strength theories let us consider the case of
pure shear. For this special case of two-dimensional stress the maximum
tensile, compressive, and shearing stresses are all numerically equal

(see p. 68) and we have

(7i
= — 0*3 = r, 0’2 = 0.

Assuming that the material has the same yield point in tension and com-
pression, the conditions for yielding according to the maximum stress
theory, maximum strain theory, and maximum shear theory, respectively,
are

r>.p. — O'y.p.,

1 +M (b)

T y,p. —

Taking /x == 0.3, we find the following results for pure shear:

Maximum stress theory . . .... . Tv ,>.

= o^y.D.

Maximum strain theory . Ty.p. = 0.77(ry.p.

Maximum shear theory ry.p. = 0.50<ry p.

It is seen that the difference between the various theories is considerable

in this particular case. In the design of a circular shaft in torsion, for
example, it is first necessary to assume an allowable value of working stress
in shear r«, =
* ry.pjn. Then the diameter of the shaft may be
Tma*
found from eq. (4.5), p. 73. Using the three theories discussed above,
the foUowing ratios of the diameters are obtained:

1 : 1.09 : 1.26,

In more recent times, consideration of the strain energy of deformation

per unit volume of the material has been used as a basis of selecting working
 STRENGTH THEORIES 317

stresses in machine design.* Considering the element in Fig. 11.23 and

applying the same reasoning as for simple tension, we find that the strain
energy per unit volume is

U = + t2<r2 + CafTs).

Substituting the values of the strain components from eqs. (11.2), this
becomes

w = — 2yL{<TiG2 + + O-jCTa)]. (11.5)

As already pointed out, we know that materials can withstand very large
hydrostatic pressure without failure. Hence it is reasonable to resolve
the total strain energy into two parts: (1) the strain energy of uniform
tension or compression and (2) the strain energy of distortion and use,
for determining the limiting stress condition, only this later part of the
strain energy, i.e., the distortion energy. To accomplish this separation,
we use eqs. (11.2). Adding these equations we obtain

ei + €2 + €3 = g^(o’l +0-2+ O-s), (c)

which states that the unit volume change is proportional to the summation
of the three principal stresses. If this summation is zero, the volume change
vanishes and the material is subjected only to the deformation of distortion.
If (71 = (72 = era = p, as in the case of hydrostatic pressure, we have

fi == €2 = ea = € = —-^—2m
1
p. (d)

There will be no distortion in this case and uniform tension or compression

exists alone.
For the general case, we introduce the notation

^ p (g)

and then divide each of the three principal stresses into two parts as
follows:

= p + (r'h = P + oil = P + o's. (f)

Summing up these three quantities and using eq. (e) we obtain

(t'i + + o's ~ 0.

•See R. E. Peterson, Strm Concentration Design Factors, New York, 1953.

318 MECHANICAL PROPERTIES OF MATERIALS

Since the summation of a'l, (r'2, and (t's vanishes, these stresses produce
only distortion, and eqs. (f) provide a means for dividing the given system
of stresses o-i, 0-2, and 0-3 into two systems ;(1) uniform tension or compression

p, producing only change of volume, and (2) the system of stresses cr'i,
(t'?, and cr'3, producing only distortion.
As an example of the application of eqs. (f) let us consider the case of
simple tension, I"ig. 11.24a. Substituting <t 2 = crs = 0 into eqs. (e) and
(f) we obtain

V (T ] 0 -
3 = 0[1
3’
:r

Simple tension in the x direction can thus be resolved into uniform tension
(I"ig. 11.24b) and a combination of pure shear in the xy- and X2-planes
(Fig. 11.24c). It can be seen that the work of the stresses producing only,
distortion (Fig. 11.24c) on the displacements produced by uniform tension
(Fig. 11.24b) vanishes. The strain energies of cases (b) and (c) are thus
independent of each other, and the total strain energy in simple tension is
obtained by adding together the strain energy of uniform tension and the
strain energy of distortion.

(a) (b) (c)

Fig. 11.24

This conclusion also holds in the general case when all three principal
stresses <t\, (J 2 , and az are acting. From this it follow^s that the straui energy
of distortion is obtained by subtracting the strain energy of uniform tension
from the total strain energy. Substituting

O’! = <72 = <rs

_
= cTi + (7'2

^
+ trs

into eq. (11.5) we obtain, for the strain energy of uniform tension alone,
the expression

1 - 2m
 STRENGTH THEORIES 319

Thus the strain energy of distortion in the general case is

+ cr2^ + — 2n{ai<T2 + cr20's + am)]

1 - 2m
6£
(ai + a^ + 0-3)"

= “ ^ 2 )^ + (ai - azY + (cTi - as)^]. (11.6)

This equation may now be taken as the basis for predicting failure of
ductile materials having a pronounced yield point stress ay,p, in simple
tension. According to this theory, for the general case of stresses <ri, (^2,
and 0-3, yielding begins when the distortion energy (eq. 11.6) reaches the
value of the distortion energy at the yield point in a simple tension test.

This latter quantity is obtained from eq. (11.6) by substituting

ai = (Ty.p., o’2 = az = 0,

which give.s

Then the condition for yielding based on the distortion energy theory is

(<ri — ()r
2 )“ + (a 2 — az)^ + {ai — as)^ = 2cry.p.^ (11.8)

In the particular case of two-dimensional stress we put 0-3 = 0 in eq.

(11.8) and the condition for yielding becomes

aY — am + a2^ = ay,p}. ( 11 - 9)

Considering, for example, combined axial tension and torsion of thin

tr*bes and denoting by a and t the corresponding stresses, the principal
stresses will be (see eq. 7.7, p. 183)

and the condition of yielding (eq. 11.9) becomes

+ ( 11 . 10 )

In the case of torsion alone we have cr = 0 and eq. (11.10) gives

Ty.p. 0.577(ry.p., ( 11 . 11 )

which is in good agreement with experimental results.

 320 MECHANICAL PROPERTIES OF MATERIALS
The condition of yielding given by eq. (11.8) is currently accepted as
valid for ductile materials, and it is assumed that the material begins
to yield when the strain energy of distortion reaches a definite value
(see eq. 11.7).

11.8 Impact Tests

Impact tests are used in studying the toughness of materials, i.e., the
ability of the material to absorb energy during plastic deformation. In
static tensile tests this energy is represented by the area under the tensile
test diagram, and it can be concluded that in order to have high toughness
the material must have high strength and at the same time large ductility.
have low toughness since they have only small plastic
Brittle materials
deformation before fracture. The use of such materials in structures is
dangerous since fracture may occur suddenly without any noticeable
deformation.
In discussing various kinds of fractures (see Art. 11.4) it was indicated
that the same material may behave as a brittle or a plastic material,
depending on the external conditions. Considering, for example, such
an important case as mild structural steel, we find that under ordinary
tensile tests, may have large plastic deformation, while if tested at some
it

lower temperature it may fracture entirely as a brittle material. Disastrous

examples of such fractures occurred during World War II in the numerous
failures of welded cargo ships.* Subsequent research work showed that the
brittleness temjxjrature of the steel plates used in the hulls of the ships
was in the same range as the service temperature.
To explain the transition from brittle to plastic fracture, A. F. Joffe
distinguished between two kinds of tensile stresses, (1) tensile stress an
producing brittle fracture by separation and (2) tensile stress a, correspond-
ing to the beginning of sliding. In Fig. 11.25 these two quantities are
represented as functions of the specimen temperature t. In Joffe^s exper-
iments the resistance to separation remained practically independent of
temperature, and in Fig. 11.25 the diagram for an is given by the horizontal
line.At the same time the resistance to sliding was influenced considerably
by the temperature of the specimen, and the ordinates of the curve for a,
decrease as the temperature increases. The point of intersection C of the
two curves defines the critical value ter of the temperature. If the tempera-
ture of testing is higher than ter, the resistance to sliding is smaller than the
resistance to separation and the specimen will yield plastically. For
temperatures lower than ter we have an < a,, and the specimen will fail
by a separation fracture without plastic deformation.
•See the paper by Finn Jonassen in W. M. Murray (ed.), Fatigue and Fracture of Metals,
1952.
 IMPACT TESTS 321

There are other important conclusions

which can be obtained on the basis of
the diagram in Fig. 11.25. Let us
consider the effect of speed of loading
on the test results. It is known that
with an increase of speed the resistance
of the material to sliding increases while
its resistance to separation remains
practically constant. As a result of this
the ordinates of the a, curve will in-
crease, and the curve will move to the
new position AiBi (Fig. 11.25) while the Tsmperature
line (Tn remains stationary. Thus the Fig. 11.25
intersection point of the two curves is

displaced to the right, indicating that with an increase in the speed of load-
ing the critical temperature increases. This conclusion is verified by
impact tests, which give brittle fractures at higher temperatures than in
static tests.
Now assume that the specimen is subjected to torsion. Yielding of the
specimen in shear will begin at about the same value of shearing stress as in
the tension tests, but the corresponding value of the maximum normal
stress oTn, equal in this case to the maximum shearing stress, will be about
one-half of the value of <r„ in a tension test. Hence in constructing, for
torsion tests, a diagram similar to Fig. 11.25, we must take values of the
ordinates of the a, curve about one-half the values for tension tests. As a
result the intersection point C of the curves will be displaced to the left,
and we conclude that in torsion tests the critical temperature must be
lower than in tensile tests. This conclusion is also inagreement with
experiments.
Considering further the influence of the state of stress on the value of
the critical temperature, let us assume that a uniform tension in all three
directions is superposed on simple tension, so that we obtain a three-
dimensional stress condition. It is known that such a superposition does
not affect the value of the maximum shearing stress at which yielding
begins. The value of o-, increases, however, and the ordinates of the cr,
curve in Fig. 11.25 increase and the intersection point C moves to the right.
Thus the critical temperature for the assumed three-dimensional stress
condition will be higher than for simple tension. Similar three-dimensional
stress conditions are produced at the notch in a grooved specimen. Such
specimens have higher values of than in the case of smooth specimens.*

•For more details on stresses at grooves see E. Orowaii’s article in W. M. Murray (ed.),

Fatigue and Fracture of Metals 1962.

322 MECHANICAL PROPERTIES OF MATERIALS
The fundamental ideas regarding the critical temperature at which the
transition occurs from
brittle to plastic fracture were extended by N. N.
Davidenkov and applied to various kinds of steel. Using a diagram similar
to Fig. 11.25, he was able to predict the influence of various factors on the
value of the critical temperature and showed by his experimental work
that the predictions were in satisfactory agreement with the experimental
facts. For determining the critical temperature, impact tests were used.
Since in the case of brittle fracture the amount of work required to produce
failure is many times smaller than for plastic fracture, the tests showed at
the critical temperature a sharp change in the amount of energy absorbed.
Fig. 1 1.26 represents the results of impact tensile tests of smooth cylindrical
steel specimens. It can be seen that a sharp change occurs in the energy
absorbed in the interval —ISO^C to — nO°C.
By changing the process of heat treatment the grain size of steel can be
varied considerably, and it is of practical interest to investigate the in-
fluence of grain size on the magnitude of the critical temperature. It is

known that with an increase in grain size the resistance of steel to separation
diminishes. Hence for coarse-grained steels the horizontal line for
(Fig. 11.25) will be lowered and the critical temperature will be higher
than for fine-grained steels. To verify this conclvision, specimens of coarse-
and fine-grained medium-carbon steel (0.23 per cent carbon) were tested
in impact and the results are shown in Fig. 11.27. It is seen that the

critical temperature for the coarse-grained steel was about —95*^0, while

in the case of the fine-grained steel it was — 160°C.

-160 -140 -120 -100

Temperature (®C) Temperature (®C)

Fig. 11.26 Fig. 11.27

*Soe Davidenkov’s books, Dynam^ical Testing of Metals, 1936, and Problems of Impact
in Metal Study, Ed. Acad. Science, Most^ow, 1938 ("in Russian). The results given in the
following discussion, if not specifically noted, are taken from the latter book.
 IMPACT TESTS 323

The effect of the size of the specimen on the value of the critical tempera-
ture has also l>een investigated. Hut here the simple diagram of lug. 1 1 .25
does not give a (*lear interpretation of the experimental results. With an
increase in dimensions we may expect a decrease in the resistance to brittle
fracture, since the probability of having critical imperfections increases
with volume. Hence for larger volumes the horizontal line for will be
lowered in Fig. 11.25, producing a displacement of point C to the right.
But it also appears that an increase in volume reduces the value of a-, and
the corresponding lowering of the a, curve in Fig. 11.25 results in a dis-
placement of C in the opposite direction. Thus the final result depends on

the relative importance of the two factors. Experiments with smooth

cylindrical specimens indicate that the lowering of the On line is more
important and point C movesshowing that the critical
to the right,
temperature increases with an increase in volume of the specimen. This
factor must be considered when applying the results of tests on small
specimens to the design of large-sized structures.
' we have considered only tensile tests of
In the preceding discussion
cylindrical specimens in which the stress distribution was uniform. Tii
practice, however, notched specimens are used in impact tests and stress
roncerdrations arc present. To investigate the effect of non-uniform stress
distril)ution on the magnitude of the critical temperature, let us Ix'gin
with the case of bending of a smooth cjdindrical specimen. I^xperiments
in bending with static loads indicate that yielding of the steel begins at a
much higher stress than in the case of uniform tension. The yield point
stress is first reached in the thin layer of fibers at the farthest distance
from the neutral axis, and the formation of planes of yielding in those

prevented by the pre.sence of the adjacent material at lower stress.

fibers is
The resulting increase in the value of the yield stress must be considered
in applying the diagram of Fig. 1 1 .25 to bending tests, The ordinates of

the (T, curve must be increased, which results in a displacement of the inter-
section point C to the right. The critical temperature, as obtained from
bending tests, will then be higher than the value obtained from tensile
tests. This conclusion agrees with experimental results.

Similar reasoning can be applied to cases of stress concentration produced

by grooves and notches (see p. 4G) and we may expect an increase in
for notched bars.
After this general discussion het us consider the type of impact test
w^hich should be used in practice to determine tcr- The correct determination
of is important avoid the dangerous situation in which the
in order to
criti('al temperature of the material is the same as the service temperature

of th(' structure. It is apparent that impact tests at room temperature are

not sufficient and in important situations a series of tests over a range of
324 MECHANICAL PROPERTIES OF MATERIALS
temperatures should be made. A transition curve, similar to Fig. 11.27,
should be constructed and from it determined. When the critical
temperature has been determined, and knowing the service temperature
to of the structure, Davidenkov recommends that the measure of safety

be taken as the ratio

~ (a)
1 n

illwhich To and T^r are the absolute temperatures corresponding to to and

4r. This ratio diminishes and approaches zero as To approaches Ter. The
result is a very dangerous situation in which small external impulses may
produce brittle fracture of the structure. On the other hand, the ratio
approaches unity as 7\r approaches absolute zero. In this case brittle
fractures will not occur and it is only necevssary to select the dimensions of
the structure so that it will be strong enough to (*arry the loads without
plastic deformation.
In selecting a reasonable value of the ratio (a) for use in design, the

conditions which actually exist in the structure must be considered.

Such stress raisers as sharp reentrant corners and imperfections in welding
contribute to an increase in ter . An increase in size of the stru(‘ture has the
same effect. To have sufficient safety and to keep the ratio (a) as large as
possible, materials with low values of t^r should be used. The critical

temperature can be lowered not only by changing the chemical content of

the material bid also by proper heat, treatment. A fine-grained steel has a
lower value of Lr than a coarse-grained steel. Considerable interest in
the brittle character of metals at 1o\n temperatures has developcnl recently
in this country, and we can expect an improvement in our knowledge of
this important subject.

11.9 Fatigue of Metals

Machine parts are frequently subjected to varying stresses and it is

important to .kiujw the strength of materials under such conditions. It
is well known that materials fail under repeated loading and unloading,

or under reversal of stress, at stresses smaller than the ultimate strength

of the material under static loads. The magnitude of the stress recjuired
to produce failure decreases as the number of cycles of stress inci'eases.
This phenomenon of the decreased resistance of a material to repeated
stresses is called fatigue, and the testing of a material by the application of
such stresses is called an endurance test.
If md o-jnin arc the maximum and minimum values of the repeated
stress, then the algebraic difference

Tt ~ <r,oax tTniiu (a)

 FATIGUE OF METALS :i25

is called the range of stress. The cycle is coriipietely detiiicd if the range
and the maximum stress are given. The average or mean stress is

“F ^^min) (h)

In the particular case of reversed stress <Jmin = — R = 2a,u&x. and

am = 0. Any cycles -of varying stress can be obtained by sui)erposing a
cycle of reversed stress on a steady average stress. The maximum and
minimum ^^alues of the varying stress are then gi\Tn i)y the following
formulas:

There are v^arious methods of applying the load in an endurance test.

The specimen can be subjected to direct bmsicjn and compression, to

bending, to torsion, or to some combination of these. The simplest way
is by reversed bending. A common cantilever form of fatigue test bar is
shown in Fig. 11.28. The cross-section of the sp(‘cimen is varied along the

- - 41/4' - --
I*

F'ig. 1 1 2S

length in such a manner that the maximum stress occurs belneen cro.ss-

.sections mn and miUi and is pra(‘ticaily constant within that region. The
effect of stres.s concentrations is eliminated by using a large fillet radius
and by increasing the diameter of the bar near the fill(‘t. The load P
is always downward and the specimen rotates at constant speed. The
stress therefore changes sign every half-revolution, and the number of
cycles of stress is equal to the number of revolutions of the inaclniH', Tlu'
stress is a completely reversed stress, the av(u*ag(' stress being Z(‘ro and
the range of stress twic(^ a^i.is.

By taking several specimens and testing them at various loads P, a

curve such as is .shown in rig. 11 29a can be obtained. ]ler(‘ is repre-
sented as a function of the number of cycles a reqiired to produce fracture.
The curve sho\\n was obtaiimd with mild steel. At the beginning o-.n^x

decreases rapidly as n increases, bur after about 4 million cycles there is no

326 MECHANICAL PROPERTIES OF MATERIALS

longer any appreciable change in <r,r.ux, and the curve approaches asymptot-
ically the horizontal lir^‘ = 27,000 j)si. The stress corresponding to
such an asymptote is called the endurance limit of the material. It is
now the usual practice in endurance tests to plot against log n. In
this manner the magnitude of the endurance limit is disclosed by a definite
discontinuity in the curve. An example of such a curve is shown in Fig.
11.29b.

®'mox

Number of cycles (millions)

Number of cycles
(b)

Fia. 11.29

There is a great differeru^e betweem the fractures of mild steel specimens

lested statically and those tested by alternating stresses. In the first
case considerable plastic flow precedes fracture, and the surfaces at the
ruptured .section show^ a silky, fibrous structure due to the great stretching
of the crystals. A fatigue crack, however, appears entirely different. A
crack begins at some point in the material owing to a local defect or to a
produced by an abrupt change in the cross-section.
stress concentration
Once formed, the crack spreads owing to the stress concentrations at its
ends. This spreading progresses under the action of the alternating stress
until the cross-section becomes so reduced in area that the remaining
portion fractures suddenly under the load.
Two zones can usually be distinguished in a fatigue fracture, one due to
the gradual development of the crack and the other due to sudden fracture.
The latter zone resembles the fracture of a tensile test specimen with a
 FATIGUE OF METALS :\27

deep, narrow groove (see p. 306) in which the shape of the specimen prevents
sliding, and therefore fracture occurs as a result of overcoming t he cohesive
forces. This fracture is of the brittle type even though the material is

ductile. In the case of cantilever test specimens (Fig. 11. 28) the maximum
stresses are at the outer fibers. Hence the fatigue crack usually starts at
the circumference and spreads towards the center. Where there are stress
concentration^ due to fillets, grooves, or holes, the crack usually starts
at themost highly stressed portion and spreads outward from this point.
In such cases the fracture surface shows concentric* rings with respect to
this starting point. This is a very common type of fracture in machine
parts which have been subjected to alternating stresses. It is thus evident
that the brittle type of fatigue fracture is due mechanism
to the peculiar
of fracture, not to crystallization of the material as was once thought.

Fig. 11.30

It is evident from the above discussion that the determination of the

endurance limit for a particular material requires a large number of tests
and considerable time. Hence it is of practical interest to establish rela-
tions between the endurance limit and other mechanical properties which
can be determined by static tests. The large amount of experimental
data accumulated has not yet made it possible to establish such a correla-
tion. As a rough estimate, the endurance limit for ferrous^ metals under
reversal of stresses can* be taken equal to 0.40 to 0.55 times the ultimate
strength obtained in the usual way from a tensile test. When working
with materials whose mechanical characteristics are very well known,
such as carbon steels, estimates of this type can be considered reliable.
Otherwise such estimates are likely to be misleading, and direct endurance
328 MECHANICAL PROPERTIES OF MATERIALS

tests should be used instead. Some results of endurance tests of steels

are givc'u in I 'ig. 1 1 .30 and also in Table A. 4 of Appendix A.

In the jnajority of eases, endurance tests are carried out for completely
reversed stresses (o-mux — ffimn), while in many cases in machine design
the stresses vary but are not completely reversed. It is necessary to know
the endurance limits under thes(‘ varying stresses. Wohler was the first

experimenter who studied the phenomenon of fatigue systematically.*

He showed that the range of stress R ne(*essary to produce fracture de-
creases as the mean stress or^, incn'ases. On th(‘ liasis of these tests and of

Bauschinger’s work, Cierber proposed a parabolic* law relating the range

of .stress R and the mean stress (Jm. This is illustrated by the parabolic
curves in Pig. 11.31, in which the mean stress and the range of stress are

expressed as fractions of the ultimate -ti*ength. The range is a maximum

when the stress is completely reversed [<j,n
~ 0) and it approaches zero
when the mean stress appnuiches the ultimate strength. If the endurance
limit tor rev(‘rsed stress and the ultimate strength are known, the endurance
limit for any varying stre.ss (‘an b(‘ obi aim'd from such curve's. Other
investigations show that there is no general law connecting the mean
stress and the range of stress, luir instance, there are materials for which
the relation between R and is ^'presented more accurately bv the broken
lines (doodman law) in Fig. 11.31 than by parabolas.

Fig. 11.31

The straight lines OA and OB in Fig. 1 1.31 have a slope of 2 and deter-
mine the region AOB in which the stress changes sign during a cycle,
Out.side this region the stress always remains tension or compression.
Experimentally determined values within the region AOB usually lie

*A. Wohler, Z, Bauursen, Vols. 8, 10, 13, 16, and 20, 1858-70, An account of this work
in English is given in Engineering Vol. 11, 1871; see also Unwin, The Testing of Materials
of Constrvetion, 3d ed., 1910.
 FATIGUE OF METALS 329

between the parabolas and the corresponding straight lines. When the
stress isalways tension or always compression, the values of the range /?,
as found by test, are sometimes not only below Gerber^ parabolas but
also below the corresponding straight lines.

11.10 Fatigue Under Combined Stresses

Most of our experimental information on the fatigue strength of materials

has been obtained under conditions of uniaxial stress, as in rotating bending-
test But in practical problems we frequently encounter cases of
specimens.
combined stress, and it is important to know the fatigue strength for
such conditions. To obtain the fatigue strength of various ductile materials
in pure shear, torsion tests were made in which the angle of twist was
reversed. The results of some of these tests are shown* in Fig. 11.32.
For purposes of comparison, the endurance limit in bending is taken as the
abscissa and the endurance limit in shear is plotted as the ordinate. It is

.seen that the ratio of these limits for all the materials tested is very nearly

Endurance limit in bending ( kips per sq in )

Fig. 11.32

*See, R, E. Peterson, Stress Concentration Design Factors^ New York, 1963.

330 MECHANICAL PROPERTIES OF MATERIALS

equal to yjs. This is the value given by the maximum distortion energy
theory for the ratio of the yield point stresses in bending and shear (see
eq, ILll).
Fatigue tests under combined stresses produced by the simultaneous
action of alternating bending and torsion have also been made and the
results are shown in I'ig Here again the test results are in good
11.33.
agreement with the maximum distortion energy theory, as might be
expected, since slip generally prect^des the development of a fatigue crack.

Fio. 11.33

To ijbtain a/i (quation for calculating the endurance limi( ft)r (‘ombined
bending and torsion, we have only to substitute into the corresponding
equation for yielding (eq, 11.10) the value of the endurance limit cte for
reversed bending, in place of ay p ,
which gives

<7^ + 3t^ = (Te^- (a)

The corresponding ellipse is shown in Fig. 11.33, and it is apparent that

the test results are in good agreement with the equation.
Other fatigue tests with biaxial tension or tension and compression,
and with the ratio ai/(X 2 remaining constant during a cycle, are also in
satisfactory agreement with the maximum distortion energy theory. Thus
we can use for determining the fatigue limit in the case of complete reversal
of stresses the following equation (see eq. (11.9) p. 319):

(Ti^ — 0-1(72 + 0-2* = (b)

 FATIGUE UNDER COMBINED STRESSES 331

in which as is the endurance limit for uniaxial stress conditions. Assuming

that ffi > 0-2 and using the notation 0-2 = aci, we obtain from eq. (b)

<71^1 — a + = (Je> i<^)

In the case of pulsating stresses in which the stress varies from zero to
some maximum value, the corresponding uniaxial pulsating stress o-max
should be substituted for (te in eq. (c).
In conclusion, it should be pointed out that the endurance limit for
various materials can be affected bymany extraneous factors. For example,
moderate cold-stretching of steel has been found to produce some increase
in its endurance limit. However, when this cold-working is overdone, the
endurance limit may be lowered.
Most fatigue tests are made at room temperature. Some experiments
with fatigue of steel specimens made at lower temperatures (--20°C) have
shown a slight increase in the endurance limit at this lower temperature.
However, tests at higher than room temperature (up to 300°C) showed
no appreciable efh'ct of temperature on the endurance limit.
Fatigue tests on specimens in the presence of various corrosive agents
such as salt water hav(‘ .shown that the enduranc(‘ limit may be greatly
reduced Iw the combined action of fatigue with corrosion. There are
many known cases of failures in service which can be attributed to such
corrosion fatigue such as marine propeller shafts, turbine blades, oil-well
pump rods, etc For this' reason, special corrosion-resistant materials
are fre([uently used in such cases. Protective (boatings and surface cold-
working have also l^een used successfully in guarding against such failures.

Jl.ll Fatigue and Stress C^oneentrations

In discussing the stress concentrations produced by sharp variations in

it was indicated that such
the cross-sections of bars and shafts (see Art. 2.5)
damaging in the case of varying stresses.
stress concentrations are especially
In machine parts, stress concentrations are always present due to fillets,
grooves, holes, keyways, etc., and experierufe shows that most fatigue
cracks in service begin at points of stress concentration.
Figure 11.34 shows the torsional fatigue failure of a shaft of a large
motor-generator set which unfortunately operated near resonance.’^ The
crack started at the keyway, where a high stress concentration took place,
and gradually developed along the helical path. Figure 11.35 represents a
torsion failure of the shaft of a Diesel-driven generator. A high stress con-

•Tiiese figures are taken from a paper by R. E. Peterson presented at the Conference on
Strength of Material Problems in Industry, at the Massachusetts Institute of Tech-
nology, July 1937.
332 MECHANICAL PROPERTIES OF MATERIALS

Fig, 11.34

centration at the small fillet resulted in several helical cracks, which when
joined together produced the saw-toothed appearance, I’^inally, Fig. 11.36
represents a characteristic fatigue failure of a heavy helical spring. The
crack started from the inside, as theory predicts (see p. 78), and again
fi)llowc‘d the direction of one of the principal stresses. All these pictures

clearly demonstrate the damaging action produced by stress concentration,

and it is clear that this factor must be seriously considered in the design of
machine parts subjected to alternating stress.
Early fatigue tests made with specimens having sharp changes of cross
section showed that there was a reduction in strength due to the stress
concentration, but this reduction was usually smaller than expected from
the magnitude of the calculated stress concentration factor. For instance,
in the case of flat steel specimens with small circular holes subjected to
direct stress, the theoretical factor of stress concentration is 3 (see p. 48).
If the magnitude of the peak stress is the controlling factor in endurance
would be expected that the tension-compression load required to
tests, it

produce fatigue failure of a specimen with a hole would be about one-third

 STRESS CONCENTRATIONS 333

Fig. 11.35

of the load for a spocimrii Avithom a hole. However, experinients showed

that in this ease the reduetion in strength due to the stress eoneentration is

small as compared with the cahuilated effect.

To explain this discrepancy and to give the necc'ss;j.ry information for
designers, a very extensive series of tests were made by R. E. Peterson at
the Westiiighouse Research Laboratories.* Geometrically similar canti-
lever test specimens varying in diameter from with a fillet
0,1 in. to 3 in.,

or with a transverse circular hole and were tested in

of different materials
special fatigue-testing machines. The results of these tests for specimens
with fillets are given in Fig. 1 1 .37. The smaller diameters of the specimens
are taken a.s ab.scissas while the ordinates represent the ratios kf of the
endurance test loads for plain specimens to the endurance test loads for the
corresponding specimens with stress concentrations. Similar results were
obtained for specimens with transverse holes.
The horizontal lines in Fig. 11,37 give the values of the stress concentra-
tion factors obtained for each fillet size by a direct measurement of strain at

*R. E. Peterson, J Appl. Meek., Vol,

. 1, pp.79 and 157, 1933; and R. E. Peterson and
A. M. Wahl, ibid., Vol. 3, p. 15, 1936.
334 MECHANICAL PROPERTIES OF MATERIALS

Fig. 11.36

the points of maximum stress coiicc ntratioii. These values are d(‘signated
by kt and are called theoretical values of stress concentration in the following
discussion. If the fatigue strength of the specimen depends only on the
peak stress, then ki must evidently be equal to kf.
On a basis of his tests, Peterson came to the following conclusions:
(1) In some cases fatigue results are quite close to theoretical stress
concentration values. This conclusion is of great practical importance,
since a general idea seems to exist, based on some early experiments, that
fatigue data for stress concentration cases are always well below theoretical
values, i.e., on the safe side for design purposes.

(2) Fatigue results for alloy steels and quenched carbon steels are
usually closer to the theoretical values than are the corresponding fatigue
results for carbon steels not quenched. It was expected in these tests that
the theoretical values of kt would be reached for all steels provided the
specimens wTre made large enough, but Fig. 11.37 shows that the curves
 STRESS CONCENTRATIONS 335

for normaliaed 0.45 per cent carbon steel are apparently asymptotic to
values considerably below the theoretical.

Oiometer rf (m )

Kkj. 11.37

(3) With a deca-ease in the size of the specimen, the reduction in fatigue
strength due to a fillet or hole be(‘,onies somewhat less; and for very small
fillets or holes the reduction in fatigue strength is comparatively vsmall.

This can be clearly seen from the curves in h^ig. 11.37.

It can be appreciated that the probkan of reducing the damaging effect

of stress concentrations primary importance to designers. Some lower-
is of
ing of stress concentrations can be obtained by a suitable change in design.
For example, a design can be improved considerably by eliminating sharp
reentrant corners and introducing fillets of generous radius, by designing
fiUets of proper shape, by introducing relieving grooves, etc. In Fig. 11.38
are shown methods for reducing the stress concentration at a shoulder of a
336 MECHANICAL PROPERTIES OF MATERIALS
shaft, while maintaining the positioning line AA, The stress can be re-
duced by cutting into the shoulder and introducing a fillet of larger radius
without developing interference with the fitted member, as shown in Fig.
11.38b. If the shoulder height is too small, a relief groove may be used as
shown in Fig. lj.38c.
In Fig 1 .39 two different bolt-and-nut designs are shown.
1 In Fig. 11.39a
the nut is in compression while the bolt is in tension. High stress concentra-
tion takes place at the bottom of the thread in the face of the nut, and
under the action of variable forces, fatigue fracture occurs in that plane.
In the lip design. Fig. 11.39b, the peak stress is somewhat relieved because
the lip is stressed in the same direction as the bolt. Fatigue tests show the
lip design to be about 30 per cent stronger.

Fig. 11.39

Somelimes these relieving measures are not sufficient to (diininate fatigue

failures. As an important example let us consider the typical failures which
occur at the wheel seals of locomotive and railroad-car axles, at the wheel or
bearing seats of automobile axles, at the pressed or fitt(‘d bits of long drill
rods in oil-well operations, etc. All these cases of fitted members subjected
to the action of variable stresses have been a constant source of fatigue
failures. Considering, for example, the case of a wheel hub pressed on an
axle, Fig. we can si^e that a high stress concentration combined
11.40a,
with friction produced at the reentrant corners m and n. During rolling
is

of the axle a reversal of .stress at points m and n takes place, and finally a
fatigue failure over the cross-section mn, may occur. Stress concentrations
can be so mewhat reduced by introducing raisc^d seats and fillets as shown
in Fig. 11.40b. A further improvement is obtained by introducing the
relief groove a. Although such changes are an improvement, they are not
 STRESS CONCENTRATIONS 337

sufficient in this ca.se. Experience shows that the mere press fit of a hub on
an axle, reduces the fatigiu^ s1reng;th of the axle to less than half of its

initial strength, while the changes shown in Fig. 11.40b raise the fatigue
strength of the axle perhaps no more than 20 per cent. Fo improve this
condition and eliminate fatigue failures, surface cold-rolling of the axle in
the region of stress concentration has beeii successfully applied.

11.12 PhysieaJ Properties of Metals at High Temperatures*

There are many ca.ses in which parts of engineering structures are

subjected simultaneously to the action of stresse s and of high temperatures.
Such conditir)ns are found, for instnn<*e, in [)o\\er plants, chemical indus-
tries, and in the missile industry. Owing to the modern tendency to increase

temperature and pressure in steam power plants and in the oil-refining

iiidusi ries, the (luestioii of t.hf* >lrength f)f matt‘rials at high temperature

has become of practical importance and a considerable amount of research

work has be(m done in this field. Experiments show that the yield point
and ultimate strength of metals in tension depend very much on the
teuiperature. Fig. 11.41 show’s how' thc'se as well as other common mechani-
ca\ properties of a medium carbon steel vary wiih the temperature.
For loads acting over a long pc^riod of time and at high temperatures as,

for instance, the weight of a structure or steam pressure in pow’cr plants, we

need additional information regarding the time effect. Experience shows
that under such conditions a continuous deformation, called creep, may
take place w’hich ismost important factor to lie considered in design.
the
Although a considerable amount of research worje in this direction has been
done and much more is 4n^w' in progress, the question of the behavior of
metals under high temperature and prolonged loading cannot be considered
completel]^ cleared.

*For further information, see papers ami Inhiiography presented in Symposium 07i the
Effect ofTemperature on Properties of Metals, issued jointly the A.8.T.M. and the
A.S.M.E., 1931.
338 MECHANICAL PROPERTIES OF MATERIALS

0 *5
SQ \n

1 -
i 1

Fig. 11.41

In most experiments of this kind, the gradual elongation of materials

under prolonged tension is studied. Tensile test specimens at high tempera-
ture are subjected to a certain constant load and temperature, and the
progressive creep under this load is investigated. The results of such an
experiment when plotted as a time-extension diagram give a curve of the
shape shown in Fig. 11.42. When the load is first applied, there is an im-
mediate elastic extension OA. The specimen then begins to stretch at a
decreasing rate as shown by the portion AB ol the curve. At BJthe rate of
extension reaches a value which remains substantially constant for some
time, that is* along the portion BC of the diagram. At C, the rate of exten-
sion begins to increa.se and fracture finally takes place at point D. For the
 METALS AT HIGH TEMPERATURES 339

stresses encountered in practice, the portion OB represents a comparatively

short time, while the entire lifetime of the specimen lies within the range
BC. The slope of the portion SC, representing the rate of extension at a
certain stress and temperature, is therefore of the utmost practical interest
because the life of the structure depends on the rate of this extension. If the
tensile stress is decreased, the slope of BC decreases, but there is no conclu-
sive evidence that it will ever become horizontal, that is, that there is a
limiting stress at which the specimen can indefinitely resist the stress and
high temperature. Hence in such cases the design must be based on the
assumption of a (certain duration of service of the structure and of a certain
amount of distortion which can be considered permissible. The working
stresses are chosen so that the distortion of the structure during its lifetime
willnot exceed a definite limit, depending on the typo of structure. For
example, in the design of moving parts such as steam turbines, the creep
should nevTr exceed 1 per cent in 100,000 hr (about 11 years) and generally
is limited to a fraction of 1 per cent. Creep rates as high as 1 per cent in
10,000 hr may be used for steam piping and boiler tubes.
The comparative w^orking stresses or creep strengths of certain alloy
steels over a wide range of temperature are shown in Fig. 11.43. These
values, however, may be much affected by variations in grain size, by heat
treatment, and by previous strain hardening and should be used with
(;autioii.

Fia. 11.42 Fig. 11,43

 APPENDIX A

PHYSICAL PROPERTIES FOR COMMON

STRUCTURAL MATERIALS

TABLE A.l. Average Physical Properties of ('ommoii Metals.

TABLE A.2. Average Physical Properties and Working Stresses of
Structural Timber.
TABLE A 3. Average Physical Properties of Building Stone, Brick and
Concrete.
TABLP] A.4 Mechanical Properties of Steels

341
342 APPENDIX A

8 ^
,S 8
.tJ eJ

{ S £
a

oj

Metals

Common

of

s’?
.5 3
Properties

c
rt |g
S£ S i
. £ >.
Physical
"
rt
C 5
w e
C §
^i:
9
Average
rS
!i^i
S'©
c o
^ fi
i^T> -S.|K
ts a
£ 0) •sJSaS
Sal"
II
s-a
Sg » SjS

t: g.

£ ^
ml
CxO)
Ib n
II *8

-g I s':!

IS «
 APPENDIX A 343

be
oS
Work- Str. 200 350 325 300 500 380 325 250 2,50 260 300 reduced
ing
may

2
a O be
^ o
700 800 800 600 900 700 550 700 700
values

Limit
2,200
Prop.
(!< should

above

1-1 ;coc5oecort •

K (See
Art. 91)
r-.*
*
QQ ^ kd ^ t ^ > values

The
psi
Materials.

! 2 2 S^ *2 *2 stress

Timber
B Modulus
Elas-

ticity
X ixxxxxxxxx specimen.

Compression,
of o '
o tS o «c 'e (M IN -li*
above

1 Testing

o
o the the
D245-33,

S Work-
ing
Str.
560 8S0 880 720 880
1,025
880 700 800 640
of of
for

Structural
most Specifications,
Society
etc.,

Dry
5.000 6.000 5,000 5,000 8,500 5,800 5,000 5.000 7.000 4,600
wet,

American

of seasoning,

Materials

Limit
4.000 4.500 3.500 4.500 5.000 4.000 3.500 3,800 5.500 3.500
Continuously
Prop.
continuously

the
Stresses quality,

of

Shear
64 80 72 60 110 103
88 68 56 68 76 Testing
in Beams Work- Str. or
Long. ing
the
Kept
§, permission

for
—
upon

Working
0 100 125 125 75 225 130 150
80 90 100
Work- Str. 200 occasiorally

ing

i i 1
Society
by
Grade

depending

400 500 500 400 900 800 700 400 350 800 400
and
P % where
Standards

Arnencan

Structural
ooooooooooo used

Modulus
Elas-

ticity XXXXXXXXXXX
W O
If see

O M M
Properties

p
S.T.M.

of «P b* «p CO
vary

-jsed
A
Common I Work-
ing
Str.
720
1,040 1,200 1,040 1,100 1,400 1,200
720 900 880 960
timber be
informat’on,
1933

Physical
1c of stork
the

4; of
detailed
6.500 7.500 8,000 8,000 7,500 from
1 lua Rup-
ture 10,-500 10,000 13.000 11.000 10,000 13,000
species

2 selected

more
reprinted

given if
Average Limit
5.500 6.500 6.500 6,300 9.500 6.500 5.000 6,300 4.500 8.000 7.500
Prop.
For
a cent

are
of
per
per

Density
it
Coopoop-^M'^N^I-I 25
table

(dry),
cu
lb properties one-third.

or
. . above

.
.
.
20
. .
..
. to
. Hemlock

physical
some
the
Pine..
Piite.

Cypress. Fir
Species oae-6fth of
Dak
Cedar..

Pine

Tamarack Parts
Longleaf Shortleai
Douglas Western Redwood
The increased
Spruce
White
Bald Soft
Red from
344 APPENDIX A

of in shear-

10* 10* 10* 10* 10* 10* 10* 10* 10*

Compression
psi
XX XXXX XXX a
Modulus
Elasticity
strengths,

7.5 8.4 8.2 3.3 2.0 3.1 3.3 3.5 with

140

break
28-<lay

of in

psi 800 550 600 700 may

1,600 1,200 1,500 1,500 8,000
Modulus Rupture Bending,

theee

of concrete

psi ttt the

one-quarter

Across
1,000 1,250 1.500
Ultimate
Strength 2.300 1,400 1.300 1,700
Sheanng

Grain,
where

as

tension

taken

Compressive psi 4,000 3,700 4,300 5,200

Ultimate 20,000 10,000 12,000 10,000 15,000
Strength,

be
diagonal

CO may

< per
teiew
Coef. Fahr.
OOOO
1

‘O III
ooo
1 1 1 .
*e
t

Linear T-1^ f— f— . rH H ..H

strengths involving

:a
< XXXXjX XXX strength

H Temp.
of
Expansion

Degree
oopc^
CQ CO Tt* lO
;o
• rf
ooo
ooo beam

Working

a compressive

in

SSJhShJhS
OQO
*?5ou0 old.
used

its
be
days of
not

28 cent

must

per
I g
B B concrete

10

"o *0
strength
to
for 5
eQ
C o sf a are

^ ^ from

2 s. s.
shearing

S g
c; . to
^ c>
._
“ values

-§
"S oS equal
rt ^ ^ direct

ti JO j3
G
,

o ^ ^ Strength

^ CO OJ fcf
This stress

I i oa c O O CG • t
a .:3 s CO ^ c3 5 ing
 APPENDIX A 346

.a
c**

X .a
« S
X
in.
1
5
Remarks
sq
X
.a in X
‘^Ic 5
*n in,
.£ .a

2 fN
,1 Bar Bar
<3

P. i i 1 ii A i i§ 1 § P.i
Endurance Limit
/in.*
oToT --—^ ro -r ^kO lO
^^
r»»
lb «N <N xr 5S s s'
-H-H -H 41 4141 -H -H 4141 41?^ g

Reduction
Area

% ^« <0
S W?
lo
tn
in

3
Steels

ae u>

of
fN M mm — rs| t*i eo
M 2 «er^ 00
g.s fM CM e>i f*5 !ii rs PM

Properties

II I 1
s SI r3 SS 2K o'
-s
?*
«>«
EJ?

Yield Point
§ 1 II i 1 III
Mechamical
s'
ib/in*

s oc
*. s'
*' R s's r- S!?S

*
Limit
/in o o § I O
A.4 Prop. 'C ao to
5 S r~ <N
m
rsi «>» »n M 11 $
lb r- ^5 ••• Ov

.
TABLE

o u U ou
u o o
ol o u
>

o Ii
o #3;j 2
^lol« So a a
a oI
I'Sd 5 oq 2 S^j Srj

-?- *31
ip S-2

II si
IP llllii ll III
- si|
55 u<z
a a
a 1 i ZS s* z
1 « '^a
*g'
u u y»-» U"»
a 9.
m;
m •**
'SS’

Source «Nt - fo I
1
 APPENDIX B

MOMENTS OF INERTIA OF PLANE

AREAS

B-1. Moment of Inertia of a Plane Area with Respect to an Axis

in Its Plane

In discussing the bending of beams, one enco\inters integrals of the type

= jy^dA (1)

in which each element of area dA is multiplied by the square of its distance

from the .r-axis and integration is extended over the cross-sectional area A
of the beam (Fig. Bl). Such an integral is called the moment of inertia
of the area A with respect to the x-axis.

Fig. B.l Fig. B.2

In simple cases, moments of inertia can readily be calculated analytically.

Take, for instance, a rectangle (Fig. B2). In calculating the moment of
inertia of this rec tangk with re.spect to the horizontal axis of symmetry x,
we can divide the rectangle into infinitesimal elements such as shown in the
figure by the shaded area. Then

^= fbdy = ^- ( 2)

346
 APPENDIX B 347

In the same manner, the moment of inertia of the rectangle with respect
to the 2/-axis is

hb^
x-hdx = ’

12

Eq. (2) can also be used for calculating h for the parallelogram shown
can be obtained from the rectangle
in Fig. B3, because this parallelogram
shown by dotted lines by a displacement parallel to the axis x of elements
such as the one shown. The areas of the elements and their distances
from the j-axis remain unchanged during such displacement so that Ir it^
the same as for the rectangle.

In calculating the moment of inertia of a triangle with respect to the

base (Fig, B4), the area of an element such as shown in the figure is

dA == b — d?/

and eq. (1) gives

The method of calculation illustrated by the above examples can be applied

toany area. The moment of inertia is obtained by dividing the figure into
infinitesimal strips parallel to the axis and then integrating as in eq. (1).
The calculation can often be simplified if the figure can be divided into
portions whose moments of kiertia about the Jixis are known. In such
case, the total moment of inertia is the sum of the moments of inertia of
all the parts.
From its definition, eq. (1), it follows that the moment of inertia of an
area with respect to an axis has the dimensions of a length raised to the
fourth power; hence, by dividing the moment of inertia with respect to a
348 APPENDIX B

certain axis by the cross-sectional area of the figure, the square of a certain
length is obtained. This length is called the radius of gyration with respect
to that axis. For the r- and ^/-axes, the radii of gyration are

rx = (3)

PBOBLEMS
1. Find the moment of inertia of the rectangle in Fig. B2 with respect to the base.
Ans, Ix* * 6/iV3.
2. Find the moment of inertia of the triangle ABC with respect to the axis X'
(Fig. B4). Ans. lx! »
hhy\2 = 6/iV4. -
3. Find /* for the cross sections shown in Fig. B5. Arts. For {a), /, * aV12 —
(a - 2A)V12; for (b) and (c), = ba^/l2 - [(6 - hO(a - 2/i)»]/12.
4. Find the moment of inertia of a square with sides a with respect to a diagonal..
Atm. I * 0^/12.

5. Find r, and r^ for the rectangle shown in Fig. B2. A/w. r^ = h/{2‘^); «
V(2V3).
6. Find r, for Fig. B5 (a) and (6),

Y Y
A
i i i
1 A A
r
1 1
T
t
f
0 — k r
O
1
X <
^
J^ X a
t
£ '
2
(
1 h I h u

(a) (b) (c)

Fig. B,5

B-2. Polar Moment of Inertia of a Plane Area

The moment of inertia of a plane area with respect to an axis perpendicu-

lar to the plane of the figure is called the polar moment of inertia with
respect to the point, where the axis intersects the plane (point 0 in Fig. Bl).
It is defined as the integral

J=j^j^dA, (4)

inwhich each element of area dA is multiplied by the square of its distance

from the axis and integration is extended over the entire area of the figure.
 APPENDIX B 349

Referring to Fig. Bl, = x* + and from eq. (4)

(x* 4* y^) dA = /i + ly (5)

-L
That is, the polar moment of inertia of a plane area with respect to any
point 0 is sum of the moments of inertia with
equal to the respect to two
perpendicular axes x and y through the same point.
Let us consider a circular croaS’Section. ,We encounter the polar moment
of inertia of a circle with respect to its center in discussing the twist of a
circular shaft (see Art. 4.1). If we divide the area of the circle into thin
elemental rings, as shown in Fig. B6, we have dA = 27rr dr, and from eq. (4)

Td^
J = r^dr = ^^ (6 )

We know from symmetry that in this case /* = 1^; hence, from eqs. (5)
and (6),
J ird'*
‘
(7)
2 64
The moment of inertia of an ellipse with respect to a principal axis x
(Fig. B7) can be obtained by comparing the ellipse with the circle shown
in the figure by the dotted line.
The height y of any element of the ellipse, such as the element shown
shaded, can be obtained by reducing the height 2/1 of the corresponding
element of the circle in the ratio b/a. From eq. (2), the moments of inertia
of these two elements with respect to the x-axis are in the ratio 6*/a*.
The moments of inertia of the ellipse and of the circle are evidently in the
same ratio; hence, the moment of inertia of the ellipse is

,
ir{2ay raV ...
350 APPENDIX B

In the same manner, for the vertical axis,

the polar moment of inertia of an ellipse is then, from eq. (5),

7r6a®
( 9)
* 4

PROBLEMS
7. Find the polar moment of inertia of a rectangle with respect to the centroid
(Fig, B2). Ans. J = bhy\2 + hby\2.
8 . Find the polar moin(Mits of inertia, with respect to their centroids, of the areas
shown in Fig. B5.

B-3. Parallel-Axis Theorem

If the moment of inertia of an area with respect to an x-axis through the
centroid (Fig. B8) i.s knotvn, the moment of inertia with respect to any
parallel x'-axis can he (*alrulated from the equation:

I,. = (10)

in which A is the area of the figure and d is the distance between the axes.
This can be proved as follows: from eq. (1)

Fig. Ji.8 Fio. B.9

The integral on the right side is equal to /*, the third integral is equal
first

to AeP, and the second integral vanishes because the x-axis passes through
the centroid; hen^, this equation reduces to (10). Eq (10) is especially
 APPENDIX B 351

useful in calculating moments of inertia of cross sections of built-up beams

(Fig. B9). The and the mo-
positions of the centroids of standard angles
ments of inertia of their cross sections with respect to an axis through their
centroid are given in handbooks. By use of the parallel-axis theorem, the
moment of inertia of such a built-up section with respect to the x-axis can
readily calculated.

PROBLEMS
9. By
use of the parallel-axis theorem, find the moment of inertia of a triangle
(Fig. B4) with respect to the axis through the centroid and paralh'l to the base.
Ans, I = 5AV36,
10 . Find the moment of inertia I of the section shown in Fig. B9 if h = 20 in.,
b = i in., and the angles have the dimensions 4 X 4 X i in. Ans.h - (i)20V12 -f
4[5.56 +3.75(10 - 1.18)2] == 1522 in.-*
11. Find the moment of inertia with respect to the i/-axis of the cross section of
the channel shown in Fig. B.5 if = 2 in., b = 10 in., a = 24 in., = 2 in. Ahh.
I, = 667 in.^

B-4. Product of Inertia^ Principal Axes

The integral

Ixy - ( xydA, (1 1)

in which each clement of the area dA is multiplied by the product of its

coordinates, and integration is extended over the entire area i4 of a plane
figure, is called the product of inertia of the figure. If a figure has an axis of
symmetry which is taken for the x- or y-axis (Fig. BIO), the product of
inertia is equal to zero. This follows from the fact that in this case for any
element such as dA with a positive x, there exists an equal and symmetrical-
ly situated element d.4' with a negative x. The corresponding elementary
products xydA cancel each other; hence integral (11) vanishes.

Fio. B.IO Fig. B.ll

352 APPENDIX B

In the gfeneral case, for any point of any plane figure, we can always find two
perpendicular axes such that the product of inertia for these axes vanishes. Take,
for instan(;e, the axes x and y. Fig. Bll. If the axes are rotated 90° about 0 in the
clockwise direction, the new positions of the axes are x' and y* as shown in the
figure. There is then the following relation between the old coordinates of an
element dA and its new coordinates:

x' = y; y' = -x.

Hence the product of inertia for the new coordinates' is

thus, durmg this rotation, the product of inertia changes its sign. As the product
of inertia changes continuously with the angle of rotation, there must be certain
directions for which this quantity becomes zero. The axes in these directions are*
called the principal axes. Usually the centroid is taken as the origin of coordinates
and the corresponding principal axes are then called the centroidal principal axes.
The centroidal principal axes are of importance since the moments of inertia are
maxima or minima with respect to these axes (see the next article). If a figure has
an axis of symmetry, this axis and an axis perpendicular to it are principal axes of
the figure, because the product of inertia with respect to these axes is equal to zero,
as explained above.

If the product of inertia of a figure is known for x- and 2 /-axes (Fig. B12)
through the centroid, the product of inertia for parallel j'- and i/'-axes can
be found from the equation:

/.V = /*., + (12)

The coordinates of an clement dA for the new axes are

x’ = x + a; /'
1 = 1
/ + h.

Fig. B.12 Fig. B.13

 APPENDIX B 353

Hence,

Ix'v* = xy dA = (x + a)(y + b)dA

= xy dA abdA bxdA ya dA.

j +j +J j
The last two integrals vanish because C is the centroid so that the equation
reduces to (12).

PROBLEMS
12. Find Ix'y> for the rectangle in Fig. B2. Ans. I
13. Find the product of inertia of the angle section with respect to the x- and
j/-axes. Do the same for the xr and yi-axes (Fig. B13).
SOLUTION. Dividing the figure into two rectangles and using eq. (12) for each of
these rectangles, we find

- 4 + ^

From the symmetry condition, « 0.

14* Determine the products of inertia hy of the sections shown in Fig. B5 if C is

the centroid.
SOLUTION. For Fig. B5 (a) and (6), Ixy = 0 because of symmetry. In the case of
Fig. B5(c), dividing the section into three rectangles, and using eq. (12), we find

L, = - 2(6 - h,)h~~-

B*5. Change of Direction of Axis. Determination of the Principal

Axes

Suppose that the moments of inertia

/. = /
y^dA; Iy= x^dA (a)
. Ja Ja

and the product of inertia

/xy = xj/ dA
^ (6)

are known, and it is required to find the same quantities for the new axes
Xi and yi (Fig. B14). Considering an elementary area dA, the new coor-
dinates from the figure are

xi = X cos 0 + y sin 0; yi ^ y cos ^ — x sin 0, (c)

354 APPENDIX B

ill which </> IS ihc angle between x and .ri. Then

/xi = dA == iy ens ~ x sin 0)^ dA ^

J I'
<t>

dA ^ dA,
+ j
X' sin^ 4}
—j 2xy sin 4> cos

or, l)y using (a) and f/)j,

~ lx COS" 4> + ly sin^ 4> ^ Ixy sin 2</>. (13)

In the same manner

— Ix sin-’ <t> A- 1 cos^ 0 + /tv sin 2 4>. (13')

l^y substituting the trigonometric identities, cos 4> == ^(1 + cos 20) and
sin' 0 = ](l — cos 20), eqs. (13) and (13') become

y,, = + k^y cos 2« - sin 2<^ (14)

and

/,,
= cos + /« sin 2«. (140

These equations are very us('ful for calculating /*, and ly^.
The value of 0 which makes /t. a maximum or a minimum may be
found by differentiating in eq. (14) with respect to 0 and setting the

derivative equal to zero. This gives

tan 20 = y—
JV I
( 15 )
 APPENDIX B 355

Substituting this value of 2<t> in eq. (14), one obtains

(ixi)max — 2
^ '^\ j
'
^ (lO)
min

For calculating we find

X\yi dA =j (x cos 0 + y mu c‘os 0 — .r sin <i>)dA

/,

if sin 0 COS dA - sin 0 cos (^dA xy (cos* 0 —

= I
j +j sin"^ <t>)dA

By using ("njand (h)

/x... = sin 20 + COS 20. (17)

The angle 0, locating the two porpcndic\iIar axes with respect to which the
product of inertia is zero, maj" be found })y setting = 0 in eci- (17).
The resulting value of tan 20 is that alread.v stated in eq. (15) and shows
that the moment of inertia is a maximum or minimum about the principal
axes (the axes about which the product of inertia vanishes).
The radii of gyration correspondiiig to ^he principal axes are called
principal radn of gyration.
Let us determine, as an example, the directions of the principal axes of a
rectangle through a corner of the rectangle (Fig. B2). In this case

lx' — ;
ly' — ; iz'u' — ,

hence, from eq. (15),

2(/i6V3 - bh^/3) 2(62 „

In the derivation of eq. (15), the angle 0 w’as taken as positive in the
counterclockwise direction (Fig. B14), so 0 must be taken in this direction
if it comes out positive. Eq. (d) gives two different values for 0 differing
by 90°. These are the two perpendicular directions of the principal axes.
Knowing the directions of the principal axes, the corresponding moments
of inertia can be found from oqs. (14) and (14').

PROBLEMS
15. Determine the directions of the centroidal principal axes of the Z section in
Fig.-B5rif6 = /ii = 1 in.; 6 = 5in.; a = 10 in. .ln«. 0 = 32'’ -05'.
16. Find the directions of the centroidal principal axes and the corresponding
principal moments of inertia for an angle section 5 in. X 21 in. X J in. Ana,
tan 20 = 0.547; = 9.36 in.^• /n.i, = 0.99 m.^
 APPENDIX B-l

TABLES

TABLE B.l. Formulas for Elements of Common Sections.

TABLE B.2. Elements of Wide Flange Sections.

TABLE B.3. Elements of American Standard I-Beam Sections.

TABLE B.4. Elements of American Standard Channel Sections.
TABLE B.5. Elements of Equal Angle Sections.
TABLE B.6. Elements of Unequal Angle Sections.

357
368 APPENDIX B-l

TABLE B-l
Elements or Common Sections

Rectangle, axis of moments through centroid.

A hh

U 12 6

h
Tz /xy = 0
VT2
Rectangle, axis of moments through base.

1h
h
I Vl "'"T
Triangle, axis of moments through centroid parallel
to base.

h
V
3

h -
c 24

V18
Triangle, axis of moments through base.

_
- iT

V6
Ellipse, axis of moments through diameter.

A = Trab

'"T
£b
A
 APPENDIX B-1 359

TABLE B-l {Continued)

Circular Section, axis of moments through diameter.

= "2“ “ *32

Hollow Circular Section, axis of moments through

diameter.

Semi-Circular Section, axis of motnc nts through

bounding diameter.
360 APPENDIX B-1

-
s
ssss ssaa sgas sjss
cocomm' eccocac^ coecc>)c) cococ^n cocoon
1

«
»-i CO N 00 CO O hi hi CO O » CO O
N d
!>.*

i<i<
UJ
o OO
13
^ o i>
rH «
'«r ^
55
CO CO uj
o CO
®
N w 05 oi so
13 N
9 «5
® -<1
*i4

CN| OS -O* ^ COI^iH-i^t OS i-( OW OS 0» <0 ^ <0 wo

Ki d
i-( C^OOcOCi
SSSS
OOtOWN t»iOr-(iH
8SS
SSSS 28S2 ^CON

W tr
Xe Q O 03 ^ W5
OS 00 »H 00 b- <0 <0
^ 00 »0 N X CO
h- 25 e5 CO NO X ^ N iH
cococoeo CsiwC^OQ ^-hi-iO 000
r-<» 0 <00 s i-4 «O^X OSC9 h-Os XOSWX sOh-iH
lOwoeoci ,-Iosx^ OsxosK cod®
2Sis Sisli SSS8 5SSS 5S58

lO iH
X ® so CO OS
SS
8|82
« ^ o SS 2 S
OS
ei
® b»
XX ®
i/5 r-i 5S
hi
3
ON
Sections

OS SO

»0 lO SO
«0 IN h>
O o so so o iO so so
h. UO iH cO 8 gg
SS
OS hi h* <0 S 3 S? h- CD CO so so 'ijS

Flange
ddod
0000
o5 O *0 01
® hi
hi
CO
® CO
<

Wide
S (N 04 OS 1 ®
-t h.

OF
38 S
X
2
so SO

X CD so t’- SO -if eo

OOCO 0000
0 O ’if O O so
NO O
^ hJ iM O
8 CO ® 2
CO cs »H IH
b
W O rH
-if C'5
ff4 T-(
CO iH
-i r-1
1

Xi X» X» X
N N S Si

^ 88
S ^^
fa b (C( fa fa fa

8888 8888 ^^ S4 S eS S
 APPENDIX B-1 361

M *H CO M^ NNWN» ^ *H «H *-< f>< ^ • ^^^< 44 <^^COCOOO«0 «OCO

a <0 ® O 00 «e NOO 0» «M« lO 04 ^^» W^N^ CO 00 KM ^ 00 'C* « O •"! « 00 N

S 532 !: S 5 S 822 SSSSSS**® SSSSSSSlitSSSS? (N »H i-H

8 SSS 8 S 8 SSS JSSeSSSS? SSSSSS??S 8 ?SJ:2

da> 0)^ dooQOodoooo ^ototocodddd Nddddddddcdtod

OSW-^ N<0<00t''^ • W^-^NONO C0 M 04 (N’-'K'^W ^'t’«i‘« 0SC0 t>*O-^e0 O.-'

248.9
QP 40 N 0> 00 Q CO
*>• o ^ «00
CSi 00 t'- 00 04 h. kO ^ ^ C N O r* 0» ^ Q5
h- cO «0 0> «0 OC
»H'<jcch 25 iON c^o<»‘ c^»-< 04 ao kON-^Owoor^iS Oo^^tf^N«o^^.-^oo^-^ w

f
OCO-^ »-<<OO)-«|<C000 O0i0r>-0104QOO>0 «OCCC400'^’^C0CO M»/iO<^«OOWOi-<«5»OOS
^p
co-^d d d
erjdooddd QP dd'^tJojccddd
CO 04 wNtoocodcocd d»-<*cc CQooddcocodco;D
S 04
5

N w
Pp p P w M P
$ pP lO O 04 p b^
^ CO Sk $
1

2.987.3
q6 «6
CO o
eo O
1

^ Oo O t» O
P CO CO
OW
o ^P^
cs
00
^
O 00
00 00
S2
04 O 0) 00
‘S c© ^
P P^ P 04 P P P O
£J
PP <N O 9

CSC4C4
04* 04
1 CO C4 rH »H ^
COofc^rHiHiH
04* C4*HrHiHfHiH
04*
xx rlrH O^C0C404M»-li-Hr-l*-iiH
I
POP5 a4b»PpPO
Pb»PpPP
PMI^app
p P^MPOOCOOQO '^OPeONOPp
8528SS8S
M a p p pi0r-4<Mc0OPP
N. 0(NPr|500C-<t04
r-
PPPPOOOpOpPp
f^.'^O^COPOOPrs.^PN
0.458
ppp-^^Tjc pppp'^'^eoco pp^'^'^ccocoM ooPMOPr^pPPP'^'^
] odd 666666 OPPOOOPO OOOPOOOP «T-iOPOOOpOO
PPPPOP
C4C404 ppppp P
PPC0P'^»H gi^p»Hpcowb. P P M
00 CO 00 «!S8S?SSSf3S8
CO 00 00 CO 00 eo CO 00 CO op oo
PP P
0.775
^^-Q0
OPPNNP
QOb-P POOCPI^PPP « PPP t>.
t>
0* b-
04 04
P P PJPPCMOOPOObop
CO «
Pd
'Ij*
cO
P OOP KP 04
CO
00

1
ppp •^ppppp pppppppp 66666666 cod^^^HF-i^’fHodpd
.-( pp 04 < CO 04
PP QQ
Pp
QC4 < P CO P CO P 04
04
pop^por^cop PPPPPPPPPPPP
ppojpppp'tpoj^p
12.000 S OPP
i-< CO
.H
<

<
CO
OO N
P?0
h- OO h- 0-
CO
P
<
‘ ppkftpTrPPP PCOPPb-PPNPPpP
ppp COPPPPP *PPPPl-vN>P ppppppp'^'^-^^'tp
I
S b» W
»Hb- PCOfHOPCO
b- P 04 ^ 04
i-'P04NPPPr-i
P P OI P P P ^ b^ t^04PP’3*P^P
PPPPPt^04P PPCOPPPPPPP^P
PPPPPPPPPPP
04
20.43

bi* 04* 04 00 00 CO d 00 — d 00 d P 04 d 00 b^ CO d d d N K d P P 04 b^ P
040404 ^W04N04iH PWOlO404»H,-<r^ C^C^N^r^fH^rX O4Pb-PP^^T0COWO4O4
I
PPP PPPPPP PPPPPPPP PPOPPOPP PPPPPPPPPPPP
d ^ d 04 d N b^ 04 d d -I P
PPP d
100.0
^Kd 04 04 04 CO 04*
xf P d P d Pd 00 00 OO d P bi
04 W ^P 00 bi P ^ p p 00 p P P
tx o6 b- b» CO ®
JJ ^ 55 ^ ^ § 2 2 2 ^ ^
I
Pp^
04 O P
pp
^ O
P
<«
fi q6 04
P
P P 04 P ^ P b* 04 C
CO tx CO O P O
P 04 P P oO P rx
W ix O P C4
04 P
oO
P P Q P CO
P
04 P P P
P P P P tx lb b. P b*
01 P
CO »-
pppppppp
24.00

d xl J d d 00 00 00 oo d b^ « 00 d b^ p* p p p d d d d d d
NCd404 WWN
CO -X* *x*
O404 C4 ^fX*x»x»Xxx»x»x f^TXi-<»HfX»xxx»x .H»x*x«H»x^»XxNxXfXpx
,-4

P CO P «
;s?
rx
r
00
i?:
b» xp b- p P ^
« x** »
^
«
x« x* x»
^
04
^
04
.H
N
PXPP
x* a X’* XXa a a Xa
o
a X*
^
» » » X» X’* « » »
^

SS 3 S xXfX
2 SSS 2 S 8 S ^Cq04PPxX«x«Xx4iH
=
pit pKi («4 &E< bt^pstbbhfapk pit Pet P^ h Pn kpMtepMhPMPBifi^PbfopitEN

C4PPP
S! i! 3 !!
x^ -x xx xx
PC4C4 04
^
C4C4
-x pppppppoo
rxxX*XxX^xX*XxX
PPPPPPPP
xXx4*Xx4x4xXxXxX
^^^PP^ ^
^^xXxX^lHxX^xX
-ii*
 362 APPENDIX B-1

u
w
1 u
C3

GC

I
A
ca

w
vJ
a
 APPENDIX B-1 363

2.67
SSSSiSSSSSSSSS 2 2SSSSS582 S§J5?5SS-S388{:S2?;SS
NNN NNNN ^^ pH r-i MMM p tP-HOOOOOOO^»-4FHO
I
^
o Cp pH
OOQ ^
CO
45.2
N W S (N OO CO CO CO 00 «c>
Q
0> w? P
COCOWWWWt-itHPH
»0 00 05 lO CO ^OOiCC4do>«j»iOeON
>^lJeOCON»HfJcQp.i-Hi-iCOU5Cod
1
CiOiUpC^C^MOipHp^CO «
(0«0'^<00MOJON05«3<Nl^Ni OOSOiCU50«OOi»Ot^ ?ocDxw5«50co«oo>obc^'^aj
235.4
ddcOpHdcococo-fdddd
pooio^NOOJ»o^co.-i»-i
oo-’fdddi^pHooooc
OOtH«5^Tf<COC^pH
codo6*^cod^ddpHdr^^d
pH ,H h4

I
pHiftpHcopr^^OMcoo ocD^*HP-pH04XC^copi-»iHr^
4.67 h-OCOOiO^'^’pf'^X O»?)-^05XWi0^CiM0>«U5»0
cowcoeowcococococo
I
Cv'^OpHNOX’^iO ’^OPil*0*H««t<eoxo»
iScO^SphniO^CO»0050
126.3
O
S«X
CO CO Q
'tl
Ch c5 lO
05 CM lO
^^
CO CO CM CM
O®^ p CM CO lO h»
CO lO ^ W X CM
i-«
CM CM
Q^^ r>i
1-* Coddd^'ddNOh^eOCMCMCM
PHPH CMCM -H IH CMCMhiih

I
O'M<CMNi0N05C0tHC5C0C5C0 XXChWiONXiOCM^ h-X'^C»'Hi0NC«05XO*HXCMf
718.7
d d CM d CM X d d CO d
ci .e0C0XpC^CM05CC 2 phQiO‘0005pHthphOC*<
XM^
CM-^iOCMXCSfHTt^ONiOCOO
jOkO^'M^COXCMCMCMpHiHpH.H
00I CM O
'IfX X 10 c:5 g ^XMO XXM
pH,-h
CM <

S hH

I
)»0 10 O N* X O Q X CM 05 CM p
3pHX'-'XCD'«fXfHOXX^
UOO X X X 00 X X X O
r>-r-iOCOpHOOOO^'M'X
«
ft
050X0000000 QX
X^Xc5^XXXXt'*X0>XtH
0.765
>X X X X X X X CM CM CM CM X X ^ X X CM CM CM CM CM ^ CM CM CM CM CM CM CM CM i-i ph eO X CM -*h

I
oocoooooooooo
(
^ O) ^
1.248
iSSSS
ooooooo
I
xxxor^xoxp^pcMO
N05r>.-icMpCM05Xa5XX
c^xtHr'-t^popoop
n.
oo CMrHt^CMP'^OxX
xpxoopoppo
pHpCMXpHpxpX^XXXXX 00

10.415 S CM w rH pH OOO 05 O tH rn CM MrHOOOXXCMCM

S 0000000005d>XX050
OOOOOOOXChNXXX xxxxxxxxxx xxddd'«fddcocodxdco
I
oxH4XocMXXgop
t>-OrHt^X»HCMOXC
32.92

dt^Tfp-Jddodt>^dd X'M^XXXX’M'XXCMIhXXX
i
OXXOOOOOXOOOOX
112.0
05h-CMX’'f05i0 05X05XpH d d X CM d CO X CM PH d d X* d pH
8 Xt'-h«XX^'M<XCOCMCMSM XX'M'TfXXCMCMCMrH ^iHihCMhnpHpHpHpHpHCMCMCMih
I
11.38
2SSS??5282SS?3SS 8SSS28SS2S 882S2SS8fe88888
PH000000005050005 oxxxxxxt^xx XXXCMCMXXX 05 t* 05 XXCM
1
10 X X -M* ^-M^XXXX
X3 4 » :i
X® « X=* ** X’* « « a X* X’* Xa XXa xx» xxxxxx
10 X X CM XX OXoXX CM

I
X X 00 ^o n OXXCMOXXXphOXOQOph
XX X C X 05 X pH
N X^P ^X^
05
05 CM 00
112
8
pH
X ^ X X CM CM CM Pif
5
X 05 X CM CM CM »H CMpHiHCM^pHfHFH^pHCMCMCMpH

Cm b ^
(S4 ^ ^ Pm ^ Pm
(2^ (X4 (Z4 (S4 Pm Pm ^^^^^ Em ^
WF
^ ^ ^ ^ ^ ^ ^ ^ ^ pi fS ^^^^^^^^^ ^^^npamACQpqnSSS Jr
10 oocoooooooooo xxxxxxxxxx XXXCMCMXXXOXOXX 12
364 APPENDIX B-l
 APPENDIX B-l 365

ssso s& ss Rn ss ss ss ss

00 CO 0» GO o (O kO ss SS
iC kO CO CO CO CO oi iH

OOOOkO kOOi '^OO cooo gJS

coeoooa ooo ^co eooi ci ^ •

SE:f2S SS S8 S 5? fe § SS 2 S
CO'^ coco NC4 cici ^ -Hi-J

eo«ooo oe^ o-^ t^eo G cc coo o^-

rH^
S^cScS SS 22 o6t>^ coco

o o> O 00 00 ^ o CS N o OO o^ h- o O kO
r-* 00 t^ kO
sg 5 c?
cd
W
*-i
C<l
kd N to to W N
S W ?3‘S

feSS
25’^'^
32
»0 CO
^ 1^
1 CO ^

fcsss
^010 0
uo lO »o kd coco coco coco WCSI MW

coco CS|r-t Okh- COrH

iO(»c5« t^eo OOrfJ OCO «oo ^<D
-^p-iooi dc^ dkd kd'^ kdcd dd d^

kO
o 00 o 00 O ''C< O CO a ko o »ft |> »C b-

S 5 c3 S as 82 dd dd d kd

SSS8 S8 88 88 SS 88^ 88 8S
dd dd do 00 00 dd to d »d kd ^ ed cd

00 00 CO ,kO t>. kO Oh

SSS CO 23 18 20 15 r-i
09
rH
^ o
rH ^
dd d kd
»—1 kH HH »-H k-H k-S kH kH k-H >-H M k-l H HH 1-4 )-4 HM
09
r-J 1-^ ^^ O
rH
O 00 00 O cO kO kO CO CO
 APPENDIX B-1

ooooo t^OOOCl
Oo»-i Kr>.(jo ;o2>b-t>-

COCOWOS OOTf<(N t>-iOCOOI

coeofo

iCf-tooo weccsi CAICOS o^ooco

o6i^»o»o ^oioo lo^co ^cooioi

fl^iOCD
OOCO^
C^-^cO
NWCDt-
TtHk«<DOO
Sections

^CCO^OCO COCOCOfO

CONb- OO)’^
-^oieo.-^ Oo6»rteo
Channel ^ f5 <D ?5cnd»

Standard

CO <f$ a>
'

•g o » f5 t-
<p ip fO

American

5
j3 <1)

"C
c
'ts
®
2P
*3

J
8ii-
o o
T-i I'- »r:»

of cococo cocoes eoNNN

OOOOOQOO
05 i-(CO'«»<
'TO© esNW
t^ffOC5
QCOOt-
OOCOOO'T
Elements

^lOCCCsi o6i^«o

oo>oor» ooo oooco

s&i S’^-jirtN Opco
25^co cJ»^P
coc5c5 C0WC5.-I

8S88 888 8^8 8888

J

.is SS«5! S§53 8882

|a
® 00000050 lOtOiO C<IC<»W oooo
 APPENDIX B-1 367

sss .57 .56 .58 88 3 .62 .50 .52 55 5$ 533

.60 .62 .63 .57 .58 .59 883 C5 Q

25
to lO
•v ^ ^
pH pH

05 CO
wo a> 0 .86 .79 .65 .57 .50 .45 .38 S8 27 24 21
pH pH pH

05 00 O «0 CO ^ w S 1 87 70 S5 38 31 25 20

W •H T-< w «H pH pH pH pH

?3 55 882 PH 05
to ic
W 283 88 tp CO
25 822
CO CO CO w w CO cs w w WWW p.^ pH pH pH pH pH

lO CO iO 05 O rH tp 05
q 00 O CO lO O CO 05 ^ W PH
CO pH
?-^ »—1
O
pH
b 05 QC b b b CO CO W PH pH pH »H

•X) i> CO 1> 00 CO pH pH tH CO p- O q qq 1H 00 CO

SS!J RS 53 pH
b
PH
CO
PH
00 b CO W HH P-<

9>
T— »—<pH
O t;
W
•«<
8 8 8 8
I ^
t

CO CS CO

8 S5
1/5
5 :S
tH-
8$2
^
lA
HHfH rHi-li-t

?88 8SS oSfe8 SS g5?2

lO^co '^eooi cocow w^’ Wth

9.8 9 ^ 9.0 6.7

7.25
5.4 q q iH
20.0 15.0 13.4
18.75 13.75
11.5
14.75 12.25

CO
pH PH
d
00 bbb

S88 888 888 888 88 88 888

05 05 ^ GO o6 00 IN bO CD b »o b CO CO CO

q q PH
O
W lO CO
pH pH
00 CO pH Hj<
pH pH
W 05 «©
pH pH
00 05 b fp IC b b b

05 05 05 00 GO 00 t» t. t. CD CO CO »D 5Q CO CO CO
APPENDIX B-1
 APPENDIX B-1 369

0) 0) O) 0)
SSSS .58 .58 .58 .59 .59 SSc? WWW .20 .19 .20

ssjsasg O O)
000)05
'-H
93 .89
87 .84 .82
1H COWO)
00t>.tFiCO C0iO»O^ S3^ Tt4>^ '(t* ScSeS
fH tH rH »H fH

O)^ wco »oot^

8g$s Ss;gg$ ffSSfcff lO CO CO CO W lO ^ ^rji Ft* fcSSSS
CO CO CO 8SS
1 H ^ tH tH 1— t fH fH »— rH( i-<

00 O) 00 ^iOO)CO 0 )-ff o COON

OO’^OlflCOrH 10 0)0)*^ th
CQ
001^55^ esgg fH fH fH fH
1-1 O s^s sss
WWM »-H .-H »-( 1—

•^^00
CO 0) lOO oooiooS wSSS 5SS2 g522 tH fH O ssz sss
;o 00 CO cooioici O rH fH 1-C fH

^ t-H wo o^ as228
cO^
WOb-'0|
F^ 00
gsg coco
>o Ff CO
o
lo-^r cooi coc)ci»H O W fH fH F-t W ^ »H fH

0)^10 0)c00
ko or C50 w CO 1-^10 0)00 t>.Oi»-HS t-fH-^cO 85 o So
00 10 0)0)00 CO
t-H rH fH
fh 00
r-1
oir^cd'^co r^cd'^cQ •^cd W WfH WfHF-I F^ FH F-t

VMNOPS^^

:« :3;
CO w C4 w F-* FH Fli4

X X X X X X X X X
X
m CO
:r
w w
ii?
F-l
:s: ii?
F^ FH
370 APPENDIX B-1
 APPENDIX B-1 371

^ 0 e 0 e o o e e o e e 0
O 0N <0 N.
0
*<5 X CM X X OS O CM OX 1^4 X bs X X rr os ff CM X X O CM ^
^S2SSS OS C7S OS OS
fH fH fH fH 22888
05 05
00 00 00 00 00 00 S S S X 00 o8 S8 g{i{:R lO X X X X X h»
h* h* h« h»
X X
XXXX
XXXX
X

oo^ 5P c? a>
»-i
o> X X 2S3S$3;S sesss: p X
O Cl
fh QO X rH
XX XX
fF
05 {2{28$S
^ fH ^^ .H

Oo^
O» ^ Ol X •HCMXXXCrt*
rH fH ^ fH
F-^ 1-^ i-H
97 .99 .00 01 SSSS88S ^X
XXXX s£
•—< fH rH fH rH FiH fH fH fH •-< F^ fH fH fH fH fF

^ O X 0> X 1-4 O X OS X Fi4 X CM O lO

X CM OS X rf CM OS »F O OS XX
X If
F-,

X X <N CM »-« X X CM CM fH *H fH fH f.^ f-^ CM fF fF fF fF fF ^ fF

XXXw
CM fh X t^ X X X OS CM X X CS X XX ff X CM h- CM XXOX rH
O O l> X IC »o OS X h- O lO -iH rH rH X CM CM X rH X X CM CM CM CM CM ff

lO X XX ssss XOX
XX
h- h*
ff
X X 0>
XXCO
i2K8$S
CM CM CM CM CM CM CM csi CM ^ fH fh ^ CM CM CM FH ^^^^^ ^ ^ fH fF fF

X
XXXX
XX0> Q
XX
1-4 fF p X -4
OS
XX X CM
X
S 8 S S S Sm CO
ei CM CM CM CM CM FF< fH fH fH fH fH fH fH fF fH fH rF fF fF fF fF fF fF fH fF fF fF ff

rH ^X^ CM XXXXX 00 CM X CM X hFOXX OS X OS X CM OS X

OX XX rH hi OX XX CM rH X CM CM rH X X CM CM FF ff CM CM CM fh ff

OS X rH X
i>- h- X 1-4 X X rH X OS OS OS c>O O OS X X rH X rH rH W -H
CM h- CM X X p h* ^ fH »0 X »H X CM O X X N O X h* X X CS X |F CO X
rH X X « CM CM <N CM 1-r fh fH fm fH fF fF

sssssss S SJ X X sasssss XW
ff X Q
|F X^ rH
OS
00 hi X X rH X bl X X rH rH X X rH X CM CM X rH rH X X CM CM CO ed CM CM FF

CM CM *H OS XX CM X O CM X X X X h. 00 OS 00 00 X O rH h. o X X X CM X
ssassass X fh OS h»
fH fH
OS X X O X tF
CM
fH fF ?F fF fF fF ff
XX
CM ff os

(0 <0

rH rH X X X
X X X X X
tr , X X X X
372 APPENDIX B-1

I
O
5

I?
 APPENDIX B-i 373

e e 0 o 0 O 0 o 0 0 0 o 0 o o e o 0 0 0 0 0 e 0 O 0 0
O) ^ CO QO C> !-< Cl Cl »0 X rH CD XOX X d CO o X 05 d 05 X -I*

iO CD cO
N Cl
O
M X 05 05 p
rH rH rH d 88888 3888

S353SS5 sssss WXXw .42 .42 .42 .43 888 d d Si

X Cl a» c*
25 »o »o ^
X CO rH
io 25 lO X .42 .40 .38 .35
tH 05 h*
CO X XX
X fH
eo X

ffjsss UO CD
^ O Cr X
I'r
iO ^
p rH rH Cl
.70 .72 .73 .74 lO .58 .58 .59 .60
^ HjJ
.43 .44 ,45 ,35 .36 .37

CD 0> vH Q p ^ ^ rH X
c- lO 25 .74 .58 .49 .40 ^ CO So 8 8 .36 .31 .25 .20 d rH rH tH rH rH O .10 .08 .05

41 94 .78 eo o 8 21 CD »0
fr, Ol rH
XX X X Cr
rH
Xp d S3 222 X dp
rH rH O ^oX
p o
rH 1-^ rH rH

p <o
Cl »H ssss sssss; .83 .81 .79 .76 88883 66 .64 62 8S88
rH ^ *^
1 rH rH rH

60'
01 IT Zl Cl
03
^ to X
0> 0) 0)
C7) tees .78 .79 .79 .80 88S
fH ^

41 93 75 OX
rH Oi CO
CO »0 oSSS5 .55 .47 38 .29 81:888 S22 X '! p
rH rH O
vH rH

rH 05 lO fH
Cl CD Cl 00 fH Cl 05 X X rH S 05 Iv. CO »0 SSS:g .32 .25 .17 82;r
CO ci ci ^ Cl fH FH rH rH rH fH fH

^ XXX
lO 00
^ SSS CO 8R?;28 OX
rH
to rH
82SK S88? 838
Cl Cl ifH r-l Cl rH fH rH Cl rH r- -H rH rH rH rH rH

d *H c» lO CO CD lO tr 05 O O XXd X
CO b- t.82i: t:2|! SS3
O CD ^ X CO »o C* lO X X X Xd XXd ci ci rH 01 rH rH

«0 « <0 « « « to to «9 to

f
is;
Cl ei Cl Cl rH fH

X X X X X X X
:r ii?
eo CO X Cl Cl €1 fH
 INDEX

Angle sections, elements of, 368 Critical load, 268, 270

Area-moment method, 204 Critical stress, 275
Axes, principal, 157, 351, 353 Critical temperature, 320
Axial strain, 9 Curved bars, 167
Axial stress, 3
Deflection of beams, 197
Beams, deflection of, 197 by
area-moment method, 204
deflection formulas for, 212 by
double integration method, 197
limit analysis of, 142 by
energy method, 246
of two materials, 145 by
superposition method, 211
of unsyinmetrical section, 154 due to shear, 224
plastic bending of, 139 formulas for, 212
reinforced concrete, 149 under dynamic load, 222
statically indeterminate, 231 Distortion energy, 317
stresses in, 114, 127 Ductility, 297
Bending, l:>eyond yield point, 139
Eccentric loading, of columns, 280
of curved bars, 167
of short struts, 264
Bending moment, 96
Eccentricity ratio, 265, 283
Bending moment diagrams, 104
Effective elongation, 304
Biaxial stress, 50, 55
Efficiency of riveted joints, 66
Brittle materials, 49
Elasticity, definition of, 9 .
Brittleness temperature, 320
modulus of, 10
Buckling of columns, 268
Elastic line, 197
Built-up beams, 133
Elements of sections, 358
Castigliano's theorem, 247 Endurance limit, 326
Center of shear, 101 Endurance test, 324
Centroids of areas, 358 Energy of strain, 40, 82, 219
Channel sections, elements of, 366 Euler’s column formula, 270, 275
Columns, buckling of, 208 Factor, of safety, 6
empirical formulas for, 288
of stress concentration, 49
Euler's formula for, 270, 275
Failure, theories of, 313
secant formula for, 283
Fatigue, 324, 329, 331
Combined stresses, 26, 55
Fatigue failures, 331
fatigue under, 329
Flexure, see Bending
tests under, 309
Flexural rigidity, 114
Complementary stresses, 30, 66 Fracture, brittle, 306
Compression tests, 307 cup-cone, 306
Concrete beams, 149
Continuous beams, 239 Generaliaed force, 247
'
Core of a section, 266 Gerber’s law, 328
Corrosion fatigue, 331 Gkiodman law. 328
Creep, 337 Gordon-Rankine column formula, 289
376
376 INDEX
Helical springs, 77 Principal stresses, 175
Hinge, plastic, 143 in bending, 179
Hooke’s law, 10 in bending and torsion, 183
for biaxial stress, 57 Product of inertia, 351
Hoop tension, 22 Proportional limit, 32, 296
Horizontal shear in beams, 125 Pure bending, 1 1
Horsepower, 73 Pure shear, 62

I-beam sections, elements of, 364 Radius, of curvature, 114, 197

of gyration, 348, 355
shearing stress in, 130
Impact tests, 320 Rankine -Gordon column formula, 289
Imperfections in columns, 285 Reduction of area, 298
Inertia, moments of, 346 Redundant constraints, 232

principal moments of, 354 Reinforced concrete beams, 149

products of, 351 Resilience, modulus of, 41
Rings, stresses in, 22
Joints, riveted, 65 Riveted joints, 65
welded, 67 Rosette, strain, 194

Limit, endurance, 326 Safety, factor of, 6

proportional, 32, 296 Secant formula, 283

Limit design, 35, 142, 258 Section modulus, 115, 121
Lenders’ lines, 28, 298 for plastic bending, 142

Lower yield point, 296 Shear center, 161

Shear flow, 163
Membrane stresses in shells, 50 Shear force diagrams, 104
Modulus of elasticity, 10 Shear modulus, 63
in shear, 63 Shearing stresses, 27
Modulus of resilience, 41 in bending, 127, 163
Mohr^s circle, for strain, 192 complementary, 30
for stress, 59, 175 Shrink fits, 23
Moments of inertia, 346 Slenderness ratio, 275
polar, 348 Spring, helical, 77
353
principal, Spring constant, 80
table 358
of, Statically indeterminate beams, 231
Moment-area method, 204 method of superposition, 231
strain energy method, 252
Neutral axis, 112
theorem of three moments, 241
for plastic bending, 141
Statically indeterminate trusses, 15
Neutral surface, 112
Strain, definition of, 9
Normal stress, 27, 114
biaxial, 57
Parabolic column formula, 289 shearing, 63
Parallel axis theorem, 350, 352 Strain energy, in bending, 219
Physical properties of materials, 341 in shear, 82
Plane strain, 189 in tension, 40
Plane stress, 173 in torsion, 83
Plastic analysis, 35, 139, 258 Strain hardening, 32
Plastic hinge, 143 Strain rosette, 194
Poisson’s raf o, 11, 57 Strength theories, 313
Polar moment of inertia, 348 Stress, definition of, 3, 5
Pressure vessels, 50 shearing, 5
Principal axes, 157, 351, 353 working, 5, 33
 INDEX 377

Stress coat, 298 Torsion, 70

Stress concentration, 46, 331 combined with bending, 183
Stress resultant, 3, 96 of a hollow shaft, 74
Stress-strain diagram, 10, 32, 298 of a rectangular shaft, 91
Stresses, combined, 26, 55 of thin-walled tubes, 86
complementary, 30, 56 Trajectories, stress, 181
principal, 175, 183 True stress, 304
Superposition, method of, 211, 231 Twist, see Torsion

Ultimate strength, 32, 295

Tensile test, 294 Upper yield point, 296
Tensile test diagram, 295
Tension, axial, 2, 9 Virtual work, 259
Theorem, of Castigliano, 247 Volume change, 11
of three moments, 241
Warping, 91, 225
Theories of failure, 313
Welded joints, 67
maximum distortion energy, 317
Wide flange sections, elements of. 360
maximum shear, 315
Working stress, 5, 33
maximum strain, 314
maximum stress, 314 Yield point, 28, 32, 296, 298