P u r e Bending in the Plastic Range

HARRY A. WILLIAMSt
Stanford University

ABSTRACT SYMBOLS

This paper illustrates the possibilities of the graphical solution e = strain

of Saint-Vehant's method by using what is termed a plastic bend- em = strain in outer fibers of a beam
ing factor which is plotted against outer fiber stress. The ulti- e* =.. strain in a layer of fibers where width of cross section
mate resisting moment of a number of materials may be readily changes
obtained from curves which are shown. It is concluded also t h a t / = stress
the ultimate resisting moments of beams are approximately pro- fm — bending stress in outer fibers of a beam
portional to the ultimate tensile strengths of the materials in- f = bending stress corresponding to strain, ef
volved if the stress-strain curves are of the same general form. Ftv — proportional limit stress in a tension test
Downloaded by TEXAS A & M UNIVERSITY on March 28, 2015 | http://arc.aiaa.org | DOI: 10.2514/8.1416

The investigation is restricted to pure bending. Fty — yield strength in a tension test
Ftu — ultimate strength in a tension test
/ / = plastic bending factor for circular cross section
INTRODUCTION Jfu * value of / / for ultimate loading condition
Kf — plastic bending factor for cross sections composed of
P URE BENDING in the plastic range has been discussed
by a number of investigators. Attempts to assume
rectangles
Kfu = value of Kf for ultimate loading condition
that the stress-strain curve or the stress distribution M = bending moment
curve is a simple parabola have resulted in approximate Mu = ultimate bending moment
solutions because only a portion of the curve usually
followed such a law. When a more exact equation of PLASTIC BENDING OF A RECTANGULAR BEAM—TENSION
the curve has been introduced, mathematical complica- AND COMPRESSION STRESS-STRAIN CURVES IDENTICAL
tions have resulted. The recent introduction of an
equivalent trapezoidal area under the stress-strain curve Several investigators have shown experimentally that
offers good possibilities for practical purposes although cross sections remain plane during pure bending in the
in its present form the method has certain limitations plastic range. With this as a basis, it follows from
and results in small approximations in certain problems. Fig. la that a layer of fibers originally of length mn
In searching for a method that was basically sound has its length increased a distance ns by the action of
and involved no simplifying assumptions other than the couples M in bending the beam. Thus, from similar
that stress-strain relationships in a beam are identical triangles the strain is
with those of simple tension and compression, it seemed e = ns/mn = y/r (1)
that the theory of Saint-Venant, as presented by
Timoshenko, 1 offered the best possibilities. Further where y is the distance from the neutral axis to the
investigation revealed that the expression for resisting layer of fibers in question and r is the radius of curva-
moment could be modified so that the radius of curva- ture of the bent beam. The strain distribution from top
ture was eliminated and the integral divided by the to bottom of the beam is as shown in Fig. lc. If we
square of the outer fiber strain could be plotted against assume that the relation between stress and strain in a
outer fiber stress. The integral itself was then
evaluated by a conventional method of graphical
integration which can result in a degree of accuracy
approaching that of more exact mathematical methods U-A-*

t
if desired. The method, as finally proposed, becomes a —;
-c
modification and expansion of existing theory and is not
essentially a new approach.
In the presentation that follows, first the method is
described and expressions are derived for several typical 1
M S2J
1
1
f
m
ri
(C)

cross sections. Solutions for other cross sections are

then presented briefly in the Appendix.

Received December 4, 1946.

* This paper is based on an investigation by the author while
-"-. rf
"-¥Hh<&nc
employed as Stress Analyst in the Engineering Department of
Douglas Aircraft Company, Inc., Santa Monica, Calif., and is -r'-" m
published with the company's permission and cooperation. (€)

f Associate Professor of Civil Engineering. FIG. 1.

457
 458 JOURNAL OF THE A E R O N A U T I C A L S C I E N C E S —A U G U S T , 1947

A> X/0S(#£l.)
beam during bending is the same as in simple tension
and compression, Fig. le, we can plot the stress distri-
bution curve of Fig. Id.
Since the tensile and compressive forces acting on the
beam cross section must be equal, we can write the
relationship
f0+cfbdy = f0~cfbdy (2)
From Eq. (1)
y = re, dy = rde
and Eq. (2) becomes
rbf0+,»fde = rbf-""fde (3)
This equation shows that the tension and compression
Downloaded by TEXAS A & M UNIVERSITY on March 28, 2015 | http://arc.aiaa.org | DOI: 10.2514/8.1416

areas under the stress-strain curves in Fig. le must be FIG. 4. Computation sheet for Kf vs. fm curve.
equal. Since we have assumed identical curves in this
case, it follows that the neutral axis must be at the
(2) Plot a second curve OB by multiplying values of
centroid of the cross section.
strain, e> by corresponding values of stress, / .
The moment of the force acting on the elementary
(3) Determine the area under the curve OB. Suf-
area bdy is y fb dy. Hence
ficient accuracy is usually obtained by dividing the area
M = f-V y fbdy = 2bf0'f ydy into a few trapezoids and a triangle. In some cases
em the area is nearly triangular. Since the area of the
= 2br* f j ede (4) shaded element as fede, evidently the area under the
The derivation thus far follows Timoshenko's pres- curve is equal to Jlem fede in accordance with conven-
entation. 1 However, the radius of curvature may now tional methods of graphical integration.
be eliminated and the moment expressed in terms of the (4) Divide the above area by em2 and plot this value
strain in the outer fibers of the beam.* of Kf against fm (point D in Fig. 3). If a complete curve
From Eq. (1), when y = c, e = emy and r = c/em. is desired, the procedure is repeated for other selected
Substituting in Eq. (4) : values of fm and em. Below the proportional limit
Kf = l/zfm and Eq. (5) reduces to the conventional
bh2 flexure formula.
M = 2bc2 fede/en< = -Kf = Wc)Kf (5)
( / " For some materials, the upper portion of the curve in
where Fig. 3 approaches a straight line, thus indicating that,
for rectangular beams, the moment bears a straight-line
Kf= (femfede/ej) relationship to outer fiber stress in this region.
The term, Kf, will be referred to as the plastic bending The procedure outlined above is illustrated nu-
factor. It is equal to the moment of the area under one merically in Fig. 4. All necessary computations and
branch of the stress-strain curve divided by the square curves are shown in the figure. For example, when the
of the strain in the outer fibers of the beam. This fac- stress in the outer fibers i s / w = 25,600 lbs. per sq.in., the
tor can be evaluated as follows for any stress, fm, in the corresponding strain, em = 0.012 and the area under the
outer fibers: fe curve is 1.661 lbs. per sq.in. Then Kf = 1.661 -r-
(1) Plot the tensile stress-strain curve OA (Fig. 2). (0.012)2 = 11,540 lbs. per sq.in. and the point on the
Kf vs. fm curve is plotted as indicated. All curves
* Other authors have used this modification.
should be plotted to large scales if any high degree of
accuracy is desired for the values between the propor-
tional limit and yield strength.

n.a.. -rv2th

—13
( a ) CROSS SECTION (b) S T R A I N (c) S T R E S S

FIG. 2 (left). F I G . 3 {right). F I G . 5.

 P U R E B E N D I N G "IN THE P L A S T I C R A N G E 459

Plastic bending factor curves based on minimum

specified yalues for four materials are shown in Figs. em
10-13, inclusive. T h e tension and compression stress-
strain curves for these materials are similar although
not identical.

P L A S T I C B E N D I N G OF I - B E A M OR C H A N N E L - N E U T R A L
A X I S P E R P E N D I C U L A R TO W E B — T E N S I O N AND C O M - C<x)CPOSS SECTION (b) S T R A I N (CJSTRESS
PRESSION S T R E S S - S T R A I N C U R V E S IDENTICAL
F I G . 6.
T h e bending m o m e n t expression for an I-beam or
channel m a y be derived b y the method outlined above.
However, t h e same result m a y be obtained by con- Ftu
sidering the total m o m e n t as being the difference be-
tween the resisting moments of an outer and an inner fm<2m
rectangular beam. This approach is analogous to
Downloaded by TEXAS A & M UNIVERSITY on March 28, 2015 | http://arc.aiaa.org | DOI: 10.2514/8.1416

computing t h e m o m e n t of inertia of such a cross section

b y subtracting the inoment of inertia of the inner rec-
tangles from t h a t of the outer rectangle.
In Fig. 5, t h e outer beam has a width b and a depth h. F I G . 7 (left). F I G . 8 (right).
I t s resisting m o m e n t from E q . (5) is 3(Io/co)Kf. The
inner beam has a width b — b' and a depth h — 2t.
C I R C U L A R B E A M — T E N S I O N AND C O M P R E S S I O N S T R E S S -
I t s resisting m o m e n t is 3(Ii/Ci)K/.
STRAIN C U R V E S IDENTICAL
Hence
An expression for the plastic bending m o m e n t of a
beam of circular cross section can be solved in a manner
similar to t h a t for the rectangular beam b u t t h e plastic
bending factor takes a different form.
Vbh*Kr
•A —
(b-b')(h-2tPK/l •_ I t is .assumed t h a t tension and compression stress-
6 strain curves are identical and t h a t strain is linear as
shown in Fig. 6b. Then, from the rectangular beam

fM>-DO-?)'*'' (6) theory

e = y/r, em = a/r, y = re, dy = r de

where
T h e m o m e n t of the force acting on the element in
Kf - plastic bending factor corresponding to a
Fig. 6a is
stress fm and a strain em in the outer fibers
K/ = plastic bending factor corresponding to a y f 2x dy — 2 / y V a2 — y2 dy
s t r e s s / ' and a strain e' a t the inner surface
Hence
of a flange (Figs. 5b and 5c.)
M = ±f0af yVa2 - y2 dy
T h e solution of E q . (6) requires t h e use of t h e stress-
strain curve to evaluate the factor K/. If the outer or, from the above relationship, between y and e and
fiber stress, fm, is known, Kf and em can be determined between r and em
from the Kf and stress-strain curves, respectively.
M = 4r 2 Jlgmf eVa2 - r2e2 de =
T h e strain a t the inner surface of a flange is e' =
[1 — (2t/h)]em (Fig. 5b). F r o m the stress-strain curve Jle'nfeVl - (e/em)2 de
/ ' is then determined and K/ is the factor corresponding 4#3 = y 2 £ 3 / , (7)
to this stress. I t will be noted t h a t / ' is the outer fiber
stress in the inner rectangular b e a m mentioned above. where / / is the plastic bending factor for a circular b e a m
Obviously, if the bending m o m e n t is known, t h e and is equal to t h e integral divided b y t h e square of t h e
outer fiber stress can be determined from E q . (6) only strain in the extreme fibers.
b y trial. An approximate value can be readily deter- T h e factor, / / , is evaluated as follows (Fig. 7 ) :
mined as illustrated in the Appendix. However, (1) D r a w the stress-strain curve OA.
margins of safety m a y be based on ratios of m o m e n t s (2) Construct thefe curve OB as for the rectangular
rather t h a n on ratios of stresses. beam.
Numerical illustrations will be found in t h e Appendix (3) Construct the curve O C D b y multiplying values
together with bending moment expressions for other oifeby corresponding values of V I — (e/em)2. I t will
beams of a rectangular type. be noted t h a t the area of the shaded element is
 460 JOURNAL OF THE AERONAUTICAL SCIENCES-AUGUST, 194 7

fex 1 — (e/em)2 de and, hence, the area under the curve for the compression area bc22Kfc. Adding these values
O C D is equal to the integral.
M = b(Cl*Kft + cfKfe) ' (8)
(4) Divide the area under curve O C D b y em2 to
T h e same general approach m a y be used for a shape
obtain the value of J / corresponding to the extreme fiber
such as an /-section, but, in locating t h e neutral axis,
stress, fm.
parts of t h e stress-strain curve m u s t be expanded ver-
(5) Repeat the procedure for other values of fm and tically before balancing areas.* T h e resisting m o m e n t
em. Values of Jf m a y be plotted against fm as shown in of the tension and compression areas m u s t be treated
Fig. 8.- W h e n / W is below t h e proportional limit, it can separately and added.
be shown b y direct integration t h a t J / = (7r/16)/ m .
T h e procedure outlined above is numerically illus-
trated in Fig. 9 for three points on t h e curve. Values of A P P R O X I M A T E C O R R E C T I O N F A C T O R FOR T E S T R E S U L T S
/ / are given for four materials in Figs. 10-13, inclusive.
I t is a p p a r e n t from E q . (5) t h a t m o m e n t is directly
T h e solution of a tubular beam is given in the Ap-
proportional to the plastic bending factor when t h e
pendix.
beam is rectangular and the tension and compression
Downloaded by TEXAS A & M UNIVERSITY on March 28, 2015 | http://arc.aiaa.org | DOI: 10.2514/8.1416

stress-strain curves are identical. This relationship is

RECTANGULAR BEAM—TENSION AND COMPRESSION
only approximately true for other cross sections since
S T R E S S - S T R A I N C U R V E S N O T IDENTICAL
t h e m o m e n t expressions are more complex, b u t it holds
In this case, t h e neutral axis is n o t a t t h e centroid and reasonably well as t h e ultimate loading condition is
m u s t be located b y t h e m e t h o d discussed below. After approached. I t can b e demonstrated further t h a t ,
the neutral axis has been determined, t h e beam cross for two similar b u t n o t identical materials, t h e ultimate
section can be considered as composed of two half- plastic bending factors are approximately proportional
sections with different physical properties. to ultimate tensile stresses so t h a t ultimate bending
T h e neutral axis can be located as follows: moments are roughly proportional to ultimate stresses.
Let I t will be noted t h a t the Kf curves in Figs. 10-12,
inclusive, are approximately straight lines beyond t h e
d = distance from neutral axis to outer tensile yield point. This is true to a lesser extent for t h e 14-ST
fibers forging in Fig. 13. This results from t h e fact t h a t t h e
c2 = distance to outer compressive fibers fe curve in each of these cases approximates a parabola
emi and fmi = maximum tensile strain and stress, of t h e form y = axn and t h e area under t h e curve is
respectively equal to the base times the altitude divided b y n + 1.
em2 a n d / w 2 = maximum compressive strain and stress,.= Hence, the factor
respectively
Then Eq. (3) becomes -rr Jo JCdB jm&m Jm
2 2
em (n + l)em n + 1
rb frafde = rb f~e™fde
If the area under t h e stress-strain curve is rectangular,
and it is a p p a r e n t t h a t t h e area under t h e tension curve n = 1; if it is triangular n = 2. Hence, n will always
in Fig. 14d m u s t equal t h e area under the compression lie between these values for conventional stress-strain
curve. curves.
If the stresses are well beyond the yield point, a high Actually t h e variation in n is n o t large, particularly
degree of accuracy is n o t essential and the areas can be when t h e stress-strain curves are somewhat similar.
balanced as follows: For example, the values for the aluminum alloys 356-T6
(1) D r a w t h e two stress-strain curves as shown in and 195-T6 are 1.19'and 1.25, respectively. Even for
Fig. 15. quite different materials such as Dowmetal H and
(2) T h e maximum tensile or compressive stress and 14-ST forging, n equals 1.38 for the former and 1.11
strain will be known or cstn.be estimated. Then, using for the latter.
the relationship, T h e correction factor for a tested beam is determined
emi/em2 = ci/d = ci/(h — Ci) . , as follows:
In Fig. 16, let
assume C\ and m a k e the construction indicated.
OA = t h e stress-strain curve based on certain mini-
(3) T r y various values of C\ until t h e shaded areas
m u m specified values. T h e yield and
are approximately equal. I t will be noted t h a t the
ultimate strengths are as indicated.
neutral axis does not remain in the same position b u t
OB = t h e stress multiplied b y t h e strain a t a n y
shifts as the stresses v a r y with t h e applied moment.
point on curve OA.
T o obtain the bending moment, we first construct
O A ' and O B ' = similar curves for a test b e a m
Kf — fm curves for b o t h tension and compression.
coupon.
F r o m t h e development of E q . (5), the resisting m o m e n t
2
for t h e tension area of t h e cross section is bci Kft and * See Timoshenko's analysis of a Tee-beam, reference 1, p. 366.
 P U R E B E N D I N G I N T H E P L A S T I C R A N G E 461

Ofx I O 5 ( P . S . I . ) Kf 4 J, *l03(p.st)
5 lO 0 S 10 15
&• 31
30 ~T .... ^^

25

S20\
Vr
/ /
! Stress-Strain

1
£15 /

4, '1 FIGI2-RELATI0NSH/P BETWEEN

STRESS, S TRAIN f PLASTIC
JO BENDING FACTORS
BASED ON MIN SPEC VALUES FOR

<0 *
M DOW METAL H-HT COMP A
AN-QQ-M56

r 01 .03 .04 .05 06 01 -e.

STRAIN-e
.OI2 .01 <b .020 FIG. 12.
STRAIN e
Downloaded by TEXAS A & M UNIVERSITY on March 28, 2015 | http://arc.aiaa.org | DOI: 10.2514/8.1416

F I G . 9. C o m p u t a t i o n sheet for / / vs. /,„ curve showing method

of obtaining three points on curve.

Kf * Jf ' I03 (p.s.i)

10 Z0 30 10

60\
W /
> -7
z_
^ Stress- Strain

'i\
KftJfxl03(ps.i.) 50 Ft, /
,0 5 10 15

*40
•Strain

K30
1\l FIG. 13-RELATIONSHIP BETWEEN
STRESS, STRAIN4 PLASTIC
BENDING FACTORS

I
£20
BASED ON MIN SPEC. VALUES FOR
ALUMINUM ALLOY FORGING HST
AN-QQ-A-367
<o
FIG.IO-RELATIONSHIP BETWEEN
5 TRESS, STRAIN $ PL A STIC
BENDING FACTORS
BASED ON MIN SPEC. VALUES FOR
f 02 03 .04 -05
STRAIN ~e
.06 07 .06 .09

SAND CAST AL. ALLOY I95-T6

AN-QQ-A-39Q F I G . 13.

•024 .028.03--eu

I—*.
F I G . 10. I "gma -fma

1u AVA^
Cm* fa, -fms
(<L)aSOSS S£CT/ON (6)STBAW (c) STttSS (ct) Sr££SS-Sr*/)M CVA?F£S
Kf tJ,*/0*(p.sj.)
0 5_ 10 IS
Ffu'30

£
F I G . 14.
jj Stress-5train
'
*s\
*20

^15
Ftv^

Ftp,/ /
ft
1 /
^

TENSION

FIG.II-RELATIONSHIP BETWEEN\
II 1 COMPJeESStON
8? STRESS, STRAIN4 PLASTIC
BENDING FACTORS
&o BASED ON MIN SPEC. VALUES FOR
SAND CAST AL. ALLOY 35€-T6
/ AN-QQ-A-394

f .004 .008 .012 016 .020

5 TRAIN- e
.024
1
.028.03=ea.

F I G . 11.
 462 J O U R N A L OF THE A E R O N A U T I C A L S C I E N C E S - A U G U S T , 1947

TABLE 1
Plastic Bending Factor Ratios
Ftu KfU Jfii
Materials Compared Ratio Ratio Ratio
195-T6 vs. 356-T6 1.06 1.09 1.05
195-T6 vs. Dowmetal H 1.00 1.13 1.13
356-T6 vs. Dowmetal H 0.94 1.04 1.07
14-ST vs. 195-T6 2.04 2.13 2.22
14-ST vs. 356-T6 2.17 2.32 2.32
14-ST vs. Dowmetal H 2.04 2.38 2.50

appear t h a t t h e ultimate plastic bending factors for a

new material might be approximated from known values
if t h e materials concerned have comparable stress-
strain curves.
FIG. 16. T h e main advantages of t h e proposed correction fac-
tor are its simplicity and the fact t h a t a complete stress-
Downloaded by TEXAS A & M UNIVERSITY on March 28, 2015 | http://arc.aiaa.org | DOI: 10.2514/8.1416

T h e value of n will be of t h e same order of magnitude strain curve for test beam coupons is n o t required.
for the two curves, OB and O B ' . Hence, for t h e ulti- Admittedly, a correction factor t h a t includes yield
mate, loading condition, t h e plastic bending factors are strength, as well as ultimate strength, should be more
approximately proportional to t h e u l t i m a t e tensile accurate t h a n the one proposed. T h e factor suggested
stresses. b y Cozzone 3 meets this requirement. W h e n t h e pro-
Furthermore, m o m e n t is directly proportional to t h e posed correction factors are computed for the experi-
plastic bending factor for rectangular cross sections. mental values presented b y Cozzone, they are found to
Hence, differ from his factors b y 3 to 5 per cent. T h e difference
is on the conservative side.
Mu' = ~ M.u CMU
rtu A P P R O X I M A T E SOLUTIONS
where
Expressions for cross sections other t h a n rectangular
Mur = ultimate m o m e n t of b e a m b y test or circular contain two or more plastic bending factors.
Mu — ultimate m o m e n t of b e a m predicted from the . A step in the solution m a y be saved b y assuming a ratio
m i n i m u m specified d a t a between the factors. For example, the ratio K//Kf =
0.9 approximately in the case of an I-beam. A n y error
T h e bending m o m e n t expression for an I-beam
in such an assumed value is further reduced because it
[Eq. (6)] m a y be written
is multiplied by a q u a n t i t y less t h a n unity. T h e ratio
bh< can be selected with little error after some experience.
M =
M-(—rX'-r)'i] Rough values are given in the Appendix for each t y p e of
problem.
I t is a p p a r e n t t h a t m o m e n t is n o t directly propor-
tional t o t h e plastic bending factor in this case. How- U L T I M A T E M O M E N T FACTORS
ever, t h e ratio K//Kf will be approximately the same
for two beams for which cross sections are identical and T h e expression for t h e ultimate m o m e n t of an I-beam
physical properties are similar b u t n o t identical. m a y be written in t h e form
Hence, t h e above correction factor gives reasonable re- Mu •= RM2
sults for this t y p e of cross section.
T h e same method of correction can be used for cir- where Rt might be called an ultimate m o m e n t factor.
cular a n d tubular beams since t h e Jf curves also ap- T h e value of this factor can b e plotted against various
proximate straight lines. Bending m o m e n t is directly web-flange and flange thickness-depth ratios as shown
proportional to Jf for circular beams and nearly so for in Fig. 22a. Other cross sections m a y be treated in a
tubular beams. similar manner. T h e d a t a in Fig. 22 provide a rapid
solution for the ultimate m o m e n t of a n u m b e r of stand-
I t is interesting to note t h a t t h e ultimate plastic
ard shapes.
bending factors are roughly proportional to ultimate
stress in t h e outer fibers even when materials of quite
T E S T RESULTS
different characteristics are considered. T h e ratios
given in T a b l e 1 were obtained from the curves, Figs. A series of tests was m a d e in t h e Douglas laboratory
10-13, inclusive. on beams of sand-cast aluminum alloys 195-T6 and
T h e largest discrepancy, which is 18 per cent, is in 356-T6 a n d of Dowmetal H - H T . Square, rectangular
the Jf ratio, in t h e last line, where t h e materials com- (on edge and flatwise), I, circular, a n d tubular sections
pared have completely different properties. I t would of each material were tested to failure in pure bending.
 PURE B E N D I N G IN THE PLASTIC RANGE 463

* TABLE 2
Ratio of Actual Strength to Predicted Strength of Sand Cast Beams

Ftu Average for No. of Actual Ultimate Moments

M.S.V.* Coupons Specimens Ratio: Predicted Ultimate Moments
Beam Shape Material Lbs, per Sq.In. Lbs. per Sq.In. Tested Range Average
Square 195 T-6 32,000 32,700 4 1.11-1.32 1.24
356 T-6 30,000 28,600 4 0.97-1.08 1.00
Dow H 32,000 19,000 3 1.13-1.52 1.38
Rectangle (on edge) 195 T-6 32,000 34,600 . 4 0.86-0.94 0.91
356 T-6 30,000 25,300 5 1.01-1.33 1.19
Dow H 32,000 23,500 3 1.32-1.38 1.35
Rectangle (flatwise) 195 T-6 32,000 34,900 3 0.92-1.17 1.03
356 T-6 30,000 25,300 3 1.11-1.24 1.16
DowH 32,000 25,200 3 1.21-1.23 1.22
I-beam 195 T-6 32,000 39,000 3 1.09-1.12 1.10
356 T-6 30,000 28,800 3 0.96-1.05 1.00
DowH 32,000 27,700 3 1.03-1.08 1.06
Circular 195 T-6 32,000 31,200 3 0.90-1.07 0.96
356 T-6 30,000 26,800 3 1.06-1.18 1.11
DowH 32,000 28,800 4 0.99-1.08 1.06
Downloaded by TEXAS A & M UNIVERSITY on March 28, 2015 | http://arc.aiaa.org | DOI: 10.2514/8.1416

Tubular 195 T-6 32,000 40,000 3 0.80-0.93 0.88

356 T-6 30,000 29,400 4 m 0.93-1.13 1.04
DowH 32,000 27,700 4 0.76-1.24 1.00
Summary
All Shapes 195 T-6 20 0.80-1.32 1.02
356 T-6 22 0.93-1.33 1.09
Dow H 20 0.76-1.52 1.16
All Beams 62 0.80-1.52 1.09
* Minimum specified values.

There were three to five beams of each type for each of theoretical result, based, on" the fundamental method,
the materials. Approximate dimensions were as fol- was within 10 per cent of the test value. (Supple-
lows : mentary tests described below also indicate this degree
of accuracy is possible.) It was found also that stand-
Square— 1 / 2 in. by l/2 in.
ard y 2 in. diameter test specimens, cast with the beams
Rectangular— x / 2 in. by 1 in.
and tested with the skin undisturbed, were not repre-
I-beam—flanges 1 in. wide by y 4 in. thick; depth 2
sentative of the beam material. When the ratios of
in.; web 1/A in. thick
actual to predicted moment were computed on the basis
Circular—1/2 in. diameter
of these specimens, the spread in values was still large
Tubular—2 in. diameter, 5/32 in. thick
and the results were on the unconservative side for all
The results of this series of tests are shown in Table 2. but nine tests.
The predicted ultimate moments were determined from Results obtained from a few tests of beams machined
the charts of Fig. 22, which are based on minimum speci- from 24 ST extruded bar stock were as follows:
fied values, and then corrected on the basis of the ulti- Mu (Test)
mate strength of coupons as compared with the specified Beam / Specimen Ratio:
Shape No. Mu (Predicted)
values. Inspection of the data indicates that while gen- 1.04
Rectangle 1
eral averages show reasonable correlation, the range of 2 1.03
values is large. An examination of the distribution of I 1 0.93
2 0.93
results (not shown in the table) reveals that the ratio
Tee 1 1.03
was less than 0.90 for only three tests and was greater
than 1.20 for 15 beams, nine of Dowmetal and three of The correlation between predicted moments and test
each aluminum alloy. Two-thirds of the ratios were results for the complete plastic range are shown in Fig.
greater than 1.00 and, therefore, were on the conserva- 17 for an I-beam of Dowmetal H-HT. Points on the
tive side. straight line representing test results were computed by
A part of the discrepancy may be attributed to the simple statics. In order' to obtain the predicted
use of the correction factor and to the fact that the moment, strain measurements were made on top and
tension and compression stress-strain curves were not bottom flanges during the test. Thus, for a given load
identical. The balance results from the difficulty in the measured strain corresponded to a certain outer fiber
obtaining coupons that actually represent the material stress, fm, from which the moment could be calculated.
in a beam. Coupons from the same beam sometimes The latter was based on a stress-strain curve for coupons
showed differences in the order of 30 per cent. Full from a similar untested beam but the variation in cou-
stress-strain curves were obtained in a number of cases pons made close correlation impossible.
for coupons from the tested beams and from similar The effect of shift of the neutral axis in the plastic
untested beams. If the curves were nearly alike, the range was investigated for one cast aluminum-alloy
 464 JOURNAL OF THE A E R O N A U T I C A L S C I E N C E S — A U G U S T, 194 7

strain at failure for a rectangular 24-ST duralumin

beam, however, was found to be about 8 per cent, or
less than half the tensile specimen elongation. This
indicates that the beam fibers were not able to deform
as if they were in simple tension. The computed outer
fiber stress nevertheless was within 6 per cent of the
ultimate strength of the tensile specimen. This correla-
tion may be attributed to the fact that the stress-strain
curve was almost flat beyond 10 per cent strain so that
the area under the outer portion was practically a rec-
tangle. The plastic bending factor, Kf, is changed a
negligible amount by such an area.
The method uses no other approximations except
those involved in the graphical solution. Most of the
equations contain several terms, some of which must be
Downloaded by TEXAS A & M UNIVERSITY on March 28, 2015 | http://arc.aiaa.org | DOI: 10.2514/8.1416

evaluated by computing the strain at a certain distance

from the neutral axis and then determining the plastic
bending factor from the approximate curve. An ap-
O B 4 € proximate solution, which eliminates the latter step,
LOAD ON B£AM x /0 s
IB may be made by assuming a ratio between the plastic
FIG. 17.
bending factors involved. Suggested ratios are given
in the Appendix.
195-T6 T-beam and the results are shown in Fig. 18. Cross sections for which the neutral axis is not an
Curve (a) represents the moment computed from the axis of symmetry can be solved by the method. If the
test load by simple statics. The neutral axis was lo- tension and compression stress-strain curves are reason-
cated from electric strain gage readings at various stages ably alike, the available experimental evidence indicates
of loading. Using measured outer fiber strains and the that fair results can be obtained by assuming the neutral
true neutral axis, the moment for a given load was then axis through the centroid. Actually it constantly shifts
predicted. These values are shown by curve (b). In from this position after the proportional limit stress is
the same way moments were also computed, curve (c), exceeded in the outer fibers. The method can also be
assuming the neutral axis always remained through the applied to the above problem when the stress-strain
centroid of the section. The ratios of test moment to curves are not identical for tension and compression.
predicted moment are shown in the figure. It will be
noted that the discrepancy between curves (b) and (c)
is not large even though the neutral axis shifted almost
10 per cent of the depth toward the compression flange.
Strain remained nearly linear until the ultimate load
was approached. Discrepancies at that time were
probably due to nonlinear characteristics of the gages
or to plastic flow during the recording period.
An analysis was made of a number of production
castings that had failed in bending during inspection
testing. The predicted ultimate moment, based on
coupon results, was usually within 10 per cent of the
test value. Various types of cross sections were, in-
vestigated but all castings were of the cantilever type. z
LU
2
O
2
GENERAL DISCUSSION

The method discussed above is based on the assump-

tion that the relationship between stress and strain is
the same for beam fibers as it is for simple tension and
compression, and that this relationship holds for the
full range of the stress-strain diagram. It appears to
the author that the assumption is reasonable if the total I Z 3 4 5 6 7
strain is less than about 10 per cent. Maximum strains
measured on the outer fibers of the cast beams agreed LOAD ON BEAM X I0 3 -LB.
fairly well with tensile specimen strains. The measured FIG. 18.
 PURE BENDING IN THE PLASTIC RANGE 465

It sometimes happens that the ultimate strain in a beams but are riot identical for other types of cross
coupon from an actual casting is considerably less than sections. The approximation that Cozzone makes is
the specified value on which the plastic bending factor not large.
is based. The ultimate strength may agree reasonably
well, however, with the specified ultimate strength.
The plastic bending factor based on specified data still
Appendix
gives good results since it depends primarily on the ulti- DERIVATIONS AND NUMERICAL EXAMPLES
mate strength and not on the ultimate strain.
Bending moment expressions for various types of
It is difficult to apply Saint-Venant's approach to a cross sections are derived below unless previously dis-
cross section such as an I-beam with large fillets. If cussed. All numerical examples are for sand-cast
the fillets are neglected, the predicted result may be aluminum-alloy 356-T6 beams. Data for this material
unduly conservative. One might compensate for the will be found in Fig. 11. All solutions are based on
fillets by slightly increasing the actual flange and web identical stress-strain curves in tension and compres-
thickness. sion. Ultimate resisting moment may be com-
Those familiar with Cozzone's method 3 of solving puted from the derived expressions or from the data in
Downloaded by TEXAS A & M UNIVERSITY on March 28, 2015 | http://arc.aiaa.org | DOI: 10.2514/8.1416

problems in plastic bending will note that his stress inter- Fig. 22.
cept, /o, when written in terms of the plastic bending
factor, becomes Beam Shape No. 1-^-Rectangular Beam (Fig. lb):

/o = 6K, - 2/ m From Eq. (5): M = 3(I/c)Kf.

Example 1: b = 1 in., h = 2 in.,/ w =• Ftu = 30,000
where Kf is the factor corresponding to the stress fm. lbs. per sq.in.
His moment equation may then be written From Fig. 11, JK> == 13,800 lbs. per sq.in. for this
value of fm. MU = 3X 0.667 X 13,800 = 27,600 in.lbs.
Mc/I = fm(S - 2K) + 6Kf(k - 1) Alternative solution: From Fig. 22, R = 41,500 lbs.
per sq.in. Mu = RI/c = 41,500 X 0.667 = 27,700
This equation may be used to construct curves for vari- in.lbs.
ous values of k as Cozzone proposed. Cozzone's equa- Example 2: b = 1 in., h = 2 in., M = 20,000 in.lbs.
tion becomes identical with the author's when the cross Then Kf = 20,000/(3 X 0.667) = 10,000 lbs. per
section is a rectangle. The methods do not agree for sq.in.
other cross sections because of an approximation in From Fig. 11, fm = 23,500 lbs. per sq.in.
Cozzone's approach to these problems. For example,
an I-beam can be considered as made up of a rectangular Beam Shape No. 2—Circular Beam (Fig. 6):
web, extending the full depth of the beam, and the pro- From Eq. (7), M = 1/2Z>3Jr/.
jecting flanges. The resisting moment of the web when Problems are solved in the same manner as for the
computed separately from Cozzone's equation would be rectangular beam, except that the Jf curve is used in-
correct. However, that portion of the area under the stead of the Kf curve in Fig. 11.
stress-strain diagram which applies to the flanges should
be replaced with a small trapezoidal area whose moment Beam Shape No. 3—Rectangular Cross Section with Cen-
with respect to the origin is the same as that of the tral Hole (Fig. 22):
original area. The sloping line of this small trapezoid This case may be solved by subtracting the resisting
will not coincide with the corresponding line from/ m to moment of the inner rectangle from that of the outer
fo in the larger trapezoid. rectangle.
Another way of showing the approximation is to con- Thus
sider the total resisting moment of the I-beam as being
equal to the resisting moment of the outer rectangle
minus that of the inner rectangles. The trapezoid for
the former will have ordinates fo and fm, for the latter
they will be/</ a n d / m ' , the latter being the true stress
at the inner surface of the flange. A solution made in
this way will give results identical with those obtained where
by the author. Following the same reasoning, Coz- Kf = the plastic bending factor corresponding to
zone's method can be extended to shapes such as T- stress fm and strain em in the outer fibers
beams where the neutral axis is not an axis of sym^ K/ ?= the factor corresponding to stress f and
metry. strain ef at the boundary of the hole.
In the preceding discussion, it is not intended to de- Since strain is linear, ef = (d/h)em.
tract from Cozzone's method but rather to point out Example 1: b — 1 in., h = 2 in., d = 0.5 in., fm =
the reason why the two methods agree for rectangular Ftu = 30,000 lbs. per sq.in.
 466 JOURNAL OF THE AERONAUTICAL S C I E N C E S - A U G U S T , 194

From Fig. 11, Kf = Kfu = 13,800 lbs. per sq.in. Beam Shape No. 5—I-Beam (Fig. 5);
em = eu = 0.03; e' = (0.5/2.0) X 0.03 = 0.0075 From Eq. (6):
From Fig. 1 1 , / = 23,200 lbs. per sq.in., K/ = 10,300
lbs. per sq.in.
Substituting: Mu = 26,300 in .lbs.
Alternative solution: From Fig. 22, Rh = 6,600 lbs. Example 1: b = 1 in., V = 0.25 in., h = 2 in.,
per sq.in. for d/h = 0.25. ilfM = 2?»&ft2 = 6,600 X t = 0.25 in., fm = Ftu = 30,000 lbs. per sq.in.
1 X 4 = 26,400 in.lbs. For an outer fiber stress, fm = Fiu = 30,000 lbs. per
Example 2: Data are the same as above except that sq.in., the corresponding strain, em = 0.03, and Kf =
the bending moment, M, is known, and the outer fiber 13,800 lbs. per sq.in.
stress, fm, is required. For an approximate solution as- The strain at the inner surface of the flange (Fig. 5b)
sume a value for K//Kf. Ranges of values are roughly is e' = [1 - (2t/h)]em = 0.0225. Entering Fig. 11 with
as follows: this value of strain, the stress-strain curve shows the
d/h 0.1 0.4 0.8 corresponding stress to be 28,700 lbs. per sq.in. and
Kf'/Kf 0.3-0.6 0.7-0.9 0.9 K/ = 13,200 lbs. per sq.in.
Downloaded by TEXAS A & M UNIVERSITY on March 28, 2015 | http://arc.aiaa.org | DOI: 10.2514/8.1416

For the assumed ratio, compute Kf from the formula Substituting in the equation :
and determine fm from Fig. 11. The error can be de- Mu = 2[13,800 - 0.75 X (0.75)2 13,200]
termined by solving for the moment, M, corresponding = 14,500 in.lbs.
to this stress as in Example 1. Further trials can be
made if desired. Example 2: Data are the same as in Example 1 ex-
cept that 'M = 10,000 in.lbs. is given and the outer
Beam Shape No. 4—Tubular Beam (Fig. 22): fiber stress is required. An approximate solution may
be obtained as follows:
The resisting moment of a tubular beam is equal to The ratio, K//Kf will vary from about 0.7 for thick
the moment for the outer cylinder minus the moment flanges to 0.95 for thin flanges. Assuming a value of
for the inner cylinder. 0.90 and substituting in the equation: Kf = 8,060 lbs.
Hence per sq.in. and fm = 19,000 lbs. per sq.in. Checking,
M V*D*Jf - l
/*DtU/ the moment when computed for this stress as in Ex-
ample 1 is 10,200 in.lbs.
where D and Dt are the outer and inner diameters, re-
spectively. If t is the thickness of the tube, Dt = Beam Shape No. 6—Channel with Neutral Axis Parallel to
D - 2t and Web or Tee Beam with Neutral Axis Perpendicular to
Web:
X
M= /*D*{J,- [1 - (2t/D)]U/} It is assumed in this case that the neutral axis is
through the centroid of the area. Actually, it shifts
Jf = the plastic bending factor corresponding to from this position as plastic bending occurs. However,
stress fm and strain em in the outer fi- the resulting error in computing the ultimate moment
bers appears to be small. The problem has two solutions
Jt' — the plastic bending factor corresponding to depending on whether the neutral axis is in the web or in
stress / ' and strain ef at the inner sur- the flange.
face Case 1. Neutral Axis in Web {Fig. 19):
The resisting moment of the area below the neutral
axis is one-half of that for a rectangle of twice the
' - ( - !> depth or
Example 1: D = 2 in., / = 0.20 in., fm = Ftu =
30,000 lbs. per sq.in., em = 0.03, e' = [1 - (0.4/2)]0.03 /2 —Z K
A biC^Kf.
= 0.024.
From the Jf curve (Fig. 11): Jf = 9,000 lbs. per
sq.in., J/ = 8,700 lbs. per sq.in.
Substituting: Mu = 18,200 in.lbs. assuming crip-
pling is not a factor.
Alternative Solution (Fig. 22a): D/t = 10, Rt = • u A
2,300 lbs. per sq.in., Mu = RtDz = 18,400 in.lbs.
Example 2: If the moment is known and the stress 11 G2 * €m fz^Trrt
is required, follow the procedure of Example 2, Beam
Shape No. 3, assuming a value for the ratio J//Jf.
This ratio varies roughly from 0.7 to 0.95 for D/t ratios (a.) T - SECTION (6JSTRAIN (c) STRESS
from 5 to 20. FIG. 19.
 PURE B E N D I N G IN THE PLASTIC RANGE 467

The resisting moment of the area above the neutral,

axis is
bxcSKfl - (h - h)(d - t)2K/
Adding:
M = b2c<?Kn + hcfKfl. - (&i - ftaXd - t^K/
-H4K

where
m
= V!2 A-)K/t + Kn-[l
b2
£X-^]
(a.) T - SECTION (b)STRAIN

FIG. 20.
(c) STRESS

The resisting moment of the area below the neutral

Kf2 = factor corresponding to stress f2 and strain e2 axis is
Kn = factor corresponding to stress /i and strain ci
iT/ = factor corresponding to s t r e s s / and strain e' b2c»}Kn + {h - b2){t - Cl)*K/
Downloaded by TEXAS A & M UNIVERSITY on March 28, 2015 | http://arc.aiaa.org | DOI: 10.2514/8.1416

e\ = {c\/c<L)e<i = strain at top of flange Adding:

e2 = 6m = maximum strain at bottom of web
e' = [(d — /)/c2]62 = strain at bottom of flange M = hc^Kn + b2c2*Kn + (h - b2)(t - cxyK/
/i, f2} and / ' == stresses corresponding to strains
ei, e2, and e', respectively, in stress-strain
diagram
Ci , t — Ci
When the flange is in compression, obviously the e\ = - e2, e = — — e2
c2 c2
tensile stress, / 2 , in the web governs. However, it
should be emphasized that when the flange is in tension The nomenclature is the same as in the previous
and the load approaches the ultimate condition, the problem. Examples are solved as in Case 1,
compressive stress, / 2 , and the tensile stress, fh may have
Beam Shape No. 7—Cross or I-Beam Section with Neutral
the same order of magnitude and the latter will reach a Axis Perpendicular to Flanges (Fig. 22):
critical value when the compressive stress is below
"block compression" ultimate. This case may be solved by adding the resisting
For example, if the material is sand-cast aluminum moments of the separate rectangles.
alloy 356-T6 and cx/c2 = V*/* = 30,000 lbs. per sq.in. Thus
(compression), e2 = 0.03, then ex = l/2 X 0.03 = 0.015 = gg . (* - 2tWy
a n d / i = 26,700 lbs. per sq.in. (tension). 2 2
Hence, the tensile stress in the flange is not far from
its ultimate value even though the compressive stress is
quite conservative if it is assumed that crippling is
not a factor. where
-Mi-xr)v
Example 1: b = 1 in., h = 2 in., t = 0.25 in., b2 = Kf = factor corresponding to stress fm and strain
0.25 in., fm = Ftu = 30,000 lbs. per sq.in., em = eu =
0.03; flange in compression. K/ = factor corresponding to stress/' and strain e'
Assuming the neutral axis at the centroid: c\ = 0.76 ef = (b'/b)em = strain at face of web
in., c2 = 1.24 in., then • /' = stress corresponding to strain e' in stress-
62 = em = 0.03 f
2 = } m = 30,000 lbs. per sq.in. strain diagram
/l = 27,600 lbs. per sq.in.
ei = - 1 e2 = 0.018 /' = 25,600 lbs. per sq.in. Problems are solved as for Beam Shape No. 5.
= 13,800 lbs. per sq.in.
= 12,600 lbs. per sq.in. For approximate solution, the ratio K//Kf varies
e' = ^ZiJ = 0.012
C'2
K/ = 11,500 lbs. per sq.in. from about 0.6 to 0.9 for Vjb ratios between 1/s
and V2.
Substituting: M„ = 10,200 in.lbs.
The problem may be solved more readily from the
data in Fig. 22.
Example 2: When the moment is known and the
stress, fm, is desired, assume K/ = 0.9Kfh Kn = 0.9i£/2, _&;
and solve for Kn. Determine the corresponding fm
from Fig. 11. e*7 fm
Case 2. Neutral Axis in Flange (Fig. 20): (ACROSS SECTION (fc)STRAIN (c) STRESS
The resisting moment of the area above the neutral
axis is biCi2K*. F i e 21.
 468 JOURNAL OF THE AERONAUTICAL SCIENCES-AUGUST, 194 7

BEAM SHAPE ULTIMATE MOMENT ULTIMATE MOMENT FACTORS' FIG.22(a)- ULTIMATE MOMENTS VALUES
MATERIAL BASED ON MINIMUM SPECIFIED VALUES
Ro
M l N . SPEC. VALUES
195- T6 44,500 4,820 MATERIAL Ft6i(P.S.l.)| Ft-</(RS.I)| ELONG.(%)

356-T6 41,500 4,510 A N - Q Q - A - 3 9 0 S.C. AL.ALLOY I95-T6 32,000 18,000

DOW H 4,220 A N - Q Q - A - 3 9 4 S.C.ALALLOY 3 5 6 - T 6 30,000 20,000

14 - S T 95,100 10,500 AN-QQ-M56 DOW METAL H-HT (COMRA) 32,000 10,000

AN-QQ-A-367 AL.ALLOY FORGING 14-ST 65,000 50,000 10

BEAM SHAPE 5
hHj. hA-H J,
D US • H t i
M«=RCD3 n.a..— /> M t t =R/*A 2
34
4 . TUBE NOTE: VALUES T
GIVEN BELOW DO (a) I - B E A M (6) CHANNEL
NOT PROVIDE FOR
CRIPPLING FAILURE 7
14
Downloaded by TEXAS A & M UNIVERSITY on March 28, 2015 | http://arc.aiaa.org | DOI: 10.2514/8.1416

6
b -2» .5
13
jft^
5 •K
3.
*" 12
,,K w y
H *=T 3
4 2

396- T6
OIO
II •=^
CO /' to X /
~ 2 ^ •r o /
K to

x 7
0 .1 .2 .3 .4 .5
a:

? 8 / 7 *— 14" ST
ST
••o
I 1
* 6 2*
[h. .5"
a
7 /
S3ssM 5
.4-
.3
6
?t
Tu
.5-
.4
.3
4 5
, • * "

DOW H .2
3 4
s< 19 5- T6
Pb ** w
0 2 4 6 m
8 10 12 14 16 18 20l
2 3
.4 .5

Beam Shape No. 8—Symmetrical Rectangular Cross Sec- ULTIMATE MOMENT FACTORS
tion with Two Holes (Fig, 21): *
This case may be solved as a combination of a solid The ultimate moment expression for various beam
rectangular beam and Beam Shape No. 3 . shapes can be reduced to the simple forms given in
Thus Fig. 22 and the results plotted as shown. The computa-
tion is somewhat laborious for some of the shapes and is
bW , b{hfy scarcely worthwhile unless many beams are to be
M= -Kr K/ Kf
checked. However, the data of Fig. 22 may be used to
where good advantage for materials hot shown by introducing
the correction factor discussed previously. For ex-
Kf = factor corresponding to stress fm and strain em ample, the ultimate moment of the 24-ST extruded
K/ = factor corresponding to stress/ 7 and strain ef aluminum beams mentioned under "Test Results" may
Kf = factor corresponding to stress/" and strain e"
be computed from the data for 14-ST aluminum-alloy
ef = (h'/h)em = strain at outer boundary of a
forging since the stress-strain curves of these two ma-
hole
e" = (df/h)em = strain at inner boundary of a terials are of the same general shape. The average
hole ultimate tensile strength of the 24-ST was 68,400 lbs.
f and/ 77 = stresses corresponding to strain ef and en, per sq.in. compared to 65,000 lbs. per sq.in., the mini-
respectively, in stress-strain diagram mum specified value for 14-ST. Hence, the correction
factor is 1.05. When computed in this way, predicted
Example: b = 1 in., h = 2 in., d' = 0.50 in., d = moments have been found to be within 2 per cent of
0.25, A' = 1 in., fm = Ftu = 30,000 lbs. per sq.in.,
values determined from 24-ST diagrams.
em = 0.03, e' = 7 2 X 0.03 = 0.015, e" = (0.50/2) X
0.03 = 0.0075. It should be emphasized that the data in Fig. 22 are
From Fig. 11: Kf = 13,800 lbs. per sq.in., K/ = based on the assumptions that tension and compression
12,200 lbs. per sq.in., Kf = 10,300 lbs. per sq.in. stress-strain curves are identical and that crippling
Substituting: Mu = 22,800 lbs. per sq.in. stress is not a factor.
 PURE B E N D I N G IN T H E PLASTIC R A N G E 469

FIG.22(b) ULTIMATE MOMENT VALUES 1

1 BASED ON MINIMUM SPECIFIED VALUES |

BEAM SHAPE - 6 , 2 NOTE: VALUES O F R T BELOW ARE BASED ON ULTIMATE TENSILE STRESS AT I
Mu =R T M BOTTOM OF BEAM SHOWN. RESULTS ARE CONSERVATIVE IF BOTTOM IS IN COMPRESSION.
4 \ 1 r~|
J | |
iih-H| 7T = 1.0. 1 1 U T~
;7
' /? L
C
a. _ J i h
'7 _*-'•«!- 7 A 7 h I.O VI
— < 1 Ws
j•
1 _
ifw s
L^
"
TO
- L * ° 1—
9^ \N
\
>k ^8j
6
<"*" 1V v Q 5 N^k ia
—\^L
10
K^ s
\^ i \
(fl) TEE. 54 i ^ •4.
hSXNk
K
4 \ \ Is sl 4 T •Si
8
Ri
\n .el
1 'v T '
.G\ ^Z r\ 1
%K 1 XI
fc
IO
^ 3 4 r" 3 -^ •SL <o ?>
JL h - .-iA, 0 .4 vP^k. 1
-| •,2I— h ^2L 4] *
\r hil ^

s
c, x 3 t "*>*-L
E-
t ^
|\
a
1
1 •f i
1]
4
« 2
I 1 I 1 •J i 11 2
2 5 5 6 - 1r& »95-T<bP DOW H 14-ST
1 1 1 1 1
(b) CHANNEL
2 3 rr 7
4 5 Q> 7
0
a 3 M4 i l5 €> 2 3 4 5 <o 7
0
a 3 4- 5 <b 7I
ft
Downloaded by TEXAS A & M UNIVERSITY on March 28, 2015 | http://arc.aiaa.org | DOI: 10.2514/8.1416

BEAM SHAPE - 7 My=RctbZ 1

25 50
i 23 i I _ 1 1 1 1 1 /
3 5 6 - TG>{~ 195-1 <b i3 0 W H 4 - ST
/7
22 A -J 22 48
22
21
21
\/\A 4 21 /
/ A
' 1 / V
h 1H20
\lA
20
r-^- ^ / 20 44
' 1 II
V
19 VA ~1 '^ 42 V A
(a) CROSS to y \/\ 1 \/\
5-8 IB ,yi_ X / 1 '"/1
/ /*> Y
Is A ^y ^/ A A / ^
1 y ^ •?
/A
m
n i,7 38

IT
0 17
X 16 y 16
A.
ft W 36 ]/
fyf >T
o
UX'/\*>/0U
•^"T
&
'•p ^ -^ 34 y
IT" lb* L n
(5 H 15
1 4

14 <£= y Y 32 &/v.
\{b) I - B E A M
VVEa UAD17nMTAI
,3 13
L^ ^ r^ 30
<5 .1 .a .3 .4 <D .1 .2 .3 .4 <D .J .2 3 4
^

4
REFERENCES Beilschmidt, J. L., The Stresses Developed in Sections Sub-
jected to Bending Moment, Journal of t h e Royal Aeronautical So-
1
Timoshenko, S., Strength of Materials, Part I I , 2nd Ed., ciety, Vol. XLVI, No. 379, p p . 161-182, July, 1942.
5
Chapt. V I I I ; D. Van Nostrand Company, Inc., New York, Beilschmidt, J . L., Strength of Light-Alloy Components, Air-
1941. craft Engineering, Vol. X V I , No. 181, pp. 76-81, March, 1944.
2 6
Timoshenko, S., and MacCullough, G. H., Elements of MacCullough, G. H., An Experimental and Analytical Investi-
Strength of Materials; D. Van Nostrand Company, Inc., New gation of Creep in Bending, Journal of Applied Mechanics, Vol. 1,
York, 1940. , . No. 2, pp. 55-60, April-June, 1933.
3 7
Cozzone, Frank P., Bending Strength in the Plastic Range, Hrennikoff, A., Theory of Inelastic Bending, with Reference to
Journal of t h e Aeronautical Sciences, Vol. 10, No. 5, p. 137, May, Limit Design, Proc. A.S.C.E., Vol. 73, No. 3, p p . 255-289,
1943. March, 1947.

Letter to the Editor

Dear Sir: clear; but, after all, energy remains the same kind of thing
Consideration of an extreme or "theoretical" case is often the whether applied to pushing on an outside medium such as air or
most practical means of clarifying thought and bringing out on mass initially carried in the craft itself. T h e differences in
basic principles, such as Professor Williams has attempted in his treatment are mainly concerned with t h e kind of energy ratios
Letter to the Editor in the JOURNAL OF THE AERONAUTICAL desired for comparative analysis and with the frame of reference
SCIENCES, page 236, April, 1947, but apparently without full use chosen for the speed.
of t h e possibilities implied in his interesting approach. For simple momentum theory applied to propellers, this frame
Many writers have stressed the alleged inappropriateness of of reference is most naturally the medium through which the
ordinary efficiency conceptions to rocket propulsion. But why aircraft flies and on which the propeller acts to produce propulsive
set up arbitrary or conventional limitations in the use of such a force. Here the useful power can certainly be defined, if desired,
broadly versatile parameter? In any use of efficiency as the ratio as the product of thrust and speed, regardless of whether such
of useful to expended energy or power, each must, of course, be thrust is used for acceleration, for counteracting gravity, or ex-
denned for the purpose at hand and the basic assumptions made clusively for overcoming resistance to level flight a t constant