.

l-l

NATIONAL ADVISORY COMMITTEE

FOR AERONAUTICS

TECHNICAL
No.

A UNIFIED

THEORY

Langley

1556

OF PLASTIC

OF COLUMNS
~

NOTE

Elbridge

BUCKLING

AND PLATES
Z. Stowell

Memorial Aeronautical
Lax@ey Field, Va.

... ... .
. -..

.j... .

Laboratory
-.

,., ..+ .=.,..

. . . .
-.:-----..

.
.

Washington
.

April

1948

HAMPTON,
VIRGINA

>

..

,.
..

..

..

.: . .
--

.. .. . ...

... ..

. =-

---

NATIONAL
ADVISORY COMMITTEE FOR AERONAUTICS

.,4

TECHNICAL NO@3 NO. I-756

A UNIFIED TEEORY OF PLASLIIC3T?2KLING

OF COLUMNS AND PH.TF-S
Elbrit&e Z. -St:mell

.,

SUMW3Y
On the basis of modern plasticity considerations, a Uni.fl.ed
theory of plastic buckling applicable to botl!colmns and plates
has been developed, For uniform compression, the theory shows
that longcolum& which tend without-ap=m6ciable twisting ~qufre
the tangent modulus and that loRg flan@ B which twist witho~yt
apprectable Ix5ndm require the secant modulus. Structures that
both bend and twi.s%when they buckle require a mo@Lus whfch ia
a combination of the secant modulus and the targen$ mo(lulu~,
.
.

INTRODWTION

..
.-

-.L-

. #
.

The calculation of the critical compressive stress of CO1%S

and of structures readsup of plates is an important problem in
-.
-.
aircr8Ft design. Formulas f011the critical compressive stre;s
have been worked out for a multltude of wines of both columm. and
plates, but these formulas are accurate ordy if the buckling takes
place within the elast~c range of the DM ~.erial. In present+lay
desi@s, most buckling occurs above the elastic range. The usual
. . .
~thod of handling thi+ problem is to retain all the formul+s
derived for the elastic case, but to try to discover ~effectiv&-,or reduced..modulus of elasticity which will give the correct result

when inserted into these formulas.

Column buckling was the first struc%rsl problem to be stud~ed
in the Tlastic ran=.
In the latter part of the nineteenth centu-y,
Engesser proposed use of the tangent mOd.l;LUS
as the reduced modulus
for columns. At almost the same time, in the beliefthab the
.
column would be strengthened by unloading on We copvex sif.le2__
Consid&e s~ested
that the effective modulus should lie between _.
the tangent modulus and Young?s modulus. -This concept was.subsequently refined b~ En&esser and by Ton K&r&n (reference 1) and
led to what is generally known as the double modulus.

..,
-.

-.

..

-
.

NACA TN No.
.
4

E~eriments have shown, however, that the von K&&n

double
nlodulusgives values that are too high for the column stren@h
(reference 2) and that the correct modulus is nrobably the
tangent modulus. Shanley (reference 3) has stated the situation
compactly as follows: If the tsmgent modulus ie use&directly
in the Euler formula, the reeulting criiical lo@ f.ssomewhat
lower than that Siven by the reduced modulus theory. This
Mmpler formula, originally proposed by Engesser,
Is ndw widely
used by engineers, since it gives valves that agree very well
with test data. Further careful tests hy Shanley (reference 4)
and also by I.an@ey structvzws research laboratory have shown
that the unloading on one side of the c{i-umu,poetul.atedby
von K&m&s, does not occurat buckling PM tha~ the correct modulus
for columns 3s actuslly the tan~ent modl?~.us.This conclusion also
has theoretical justification (references snd 4).
,.
Xn.tihecase of localor
plate huckj.ing,the reduced modutus
3.s
appreciably higher thszathe tangent rncib.il.us,
Tests of the
local bucld.ing.s,trqss
of aircraft-section columns have been made
by Gerard (reference ~), who has suggested.the use of the secant
modulus for this t~e ofbuckling. .Extunsive teets In the
_ey
st~~tuee reeearch laboratory cm similar aircraft seotlons made and reported over a period oiseveral years and
summarized in reference 6 have also shown that the reduced
modulus for ~lates is in the vtcinity of the secret modulus.
In pccrticular,tests of lon~.alumlnum+q.loy cruclfomsectlon
columnsj designed to buckle by twisting withnut appreciable
lendin[j,have been made in a manner similar to that described
for the aircraft-section columns in reference 6. The results
have shown that the re,ducedmodulus for pure twistin~ is very
close to the secant modulus.
.,
The present paper constitutes @ theoretical tnvestlCation
of the bucklin~ of plates .~e.yoncl
the elastic ren~, which includes
columns as a limiting case, Such em invewti~tion requl.resa
kncwled~e of the relatlons betweenthe stress and strain components
beyond the elastic ranee. These relations have not as yet been
conclusively determined. A reoent paper by Handelman and Prager
(reference 7) lmsecl on one Possi%le set of stress-strain relutions
led to results for.the bucklin~.of hin~d flan~~s in sharp disagreement with test results obtained at the Lea[~#.ey
structures research
laboratory, Another set of stress+straiu relations is flenerally
accepted by the.Russian investigators and has Imen [email protected] by
Ilyuehln (reference 8) to the stress conditions,in thin plates.
These results form the foundation of the present paper, which
assumes that in plates ,aswellas In columns, unloc+d.ing
during the,
early sta~es ofl,buclcklnfl
does not OCCUY, On this basis, a unified

.
.
.
.

.
.,

NACA TN HO. 1556

theory of plastic buckling applicable to both COhlItU2 and 10Cd

has %een developed. The results are presented in the
following section.

buckling

RESULTS AND CONCLUSIONS

Ilyushin (reference 8) has treated the stability of plates .
stressed above the-elastic linxltwith consideration of the three
possible zones that might result
from buclkling: (1) a purely
elastic zone, (2) a zone in which part of the material is in the
elastic and part is in the plastic state - Lhe Jeliasto+plastic
zone, and (3) a purely plastic zone in wh;.chall of the plate
is stressed beyond the elastic limit. A31 three zones my exist
simultaneously if the plate is not entiraly in the plastic state
before buckling or if the buckling is &l,l.owed
to proceed beyond
the initial stages.

.
.
.
.

If, however, the plate is uniformly loaded before buckling

so that all.parts of it aro initid.ly at the same point in the
plastic range end if, in addition, buckling and increase in load
are assumed to progress simultaneously, then the plate nay be
expected to remain in the puzzelyplastic state in the eerly stages
of buckling. This second aesumptlon is in agreement with the
corresponding condition that apparently holds for columns
(reference 4)

~pon the assumption that the plate remains in the purely

plastic state during buckling, Ilyushinis general relations for
this state have been used to derive the differential equation of
equilibrium of the plate under ccmibinedloads. Since critical
stresses ere generally simpler ta compute from energy expressions
than froma differential equaticm, the corresponding ener~ expression
were also found- These derivation are given in appendix A,
together with applicaticms to compressive buckling of various.typs
of plate=. A comparison with Uyushints treatment of the p&&tic.
buckling problem is given in appendix B*
.L .-

The results of mast interest in the present analysis are given

table as values of a _tity
q, the number by
which the critical stress computed for the eustic case must bo multiplied to give the oritical stress for the plastic case:
in the folhwing

NACA TN

NO.

1556
.

..

Curve

structure

(See fig. 1)
E

sec
E

Long flange, one

unloaded edge,
simply supported

Long flsnge, one

unloaded edge
clamped
:

Iang plates, both

unloaded edges
siuply suppotied

Long plate, both

unloaded edges
clamped

.
.
.
.

Short plate
loaded aa a
column

()-b1<<1

Square plate
loaded .asa

1-see
4-M4E

#taJl

.:

Q 114

SeC

~.fjfj

tti

column

1
- =
b

Long column ~>>1

b
()

.3!

tsn

--%-

A to G in figure 1
The~e values of q are plotted
as curves
for extruded 24S-JTaluminum s3.J.oy
for which the cxxnpre
ssive yield
stress was 46 ksi. Similar curves for q could readily be
prepared for any other material having a known stress-strain
relationship.

NACA TN No. 1556

The values of q given in the table were o%tained by dividi~

the critical stress of the structure in the _@astic region by the
critical stress that would be obtained on the assumption of perfect
elasticity. Since PoiseoBs r&tio has been taken as one-half in both
computations, errors fdom this cause will ordinarily be present In
both critical stresses. Most of these ez~ors will be eliminated,
however, in the Process of tivision to obtain ~j and, o~S~~&l~ly}
the values of q given ai-ebelieved tohe nearly correct.

..

When plate-buckling stresses in tineplastic range are to be cow

puted, the experimmtal value of Poissons ratiothat applies as
closely as possible.to the stressed mata ie.1,to~ether with the appmpriate value of q from this paper, Shot:kibe used in the platetuckling formula.
,,
The hi@est value of q which is -~.
oen be realized only &

there is negligible longitudinal bending ~as w5th a long hinged flange

tm
7Z-occurs whsn the londtudinal bendinrcpredominates over other t~es of
distortion (as with-e.long colum &d&r Euler buclgl.ing).The theory
implies that a change in ~he stress-stre.tncurve caused by prestressing of the material would alter the value of q in the first
case but not in the second, if the buckling stress is higher than tk
highest stress reached during the ope~tiqn ~f pre-stressing. If, cn
the other hand, the buokltig stress is lower than the highest stress
reached during the operation of pre.stressing, then q s 1 for each
case.

which bucld.esby twisting). The lowest value of

.
.

which iS

.-=

National Advisory Committee for Aeronautics

Langley Meaoxial Aeronautical Laboratory
Langley Field, Vs., July ~, Z347

..
.

8
.
,

NACA TN No. 1556

APPENDIX A
TBEORITZCAL DERIW!~IONS

DefinltfonR.- The Intensities of stress and strain are defined

in referenoe 8, respectively, as

ei.

2,
2

+q
2
+
6Y*
+
Exe,,
v
4
\/jv6X

(2)

where
stress in the x-directIon

ax

strain in the x-direotion

E*

strew

irlthe y~irection

strain in the y-direction

Y
T

shear stress

shear

slmain

Accor?iuq to the fundamental hypothesis of the theory of plasticlt~,

the intai~~i~yof ~tress
f is a uniquely defined, single-valued
function of the intensity of strain et for any given material
if

Ui

increases in magnitude (load3,& condition). If

deoreases (unloading oondltion), the relation %etween

%ecomes lines

.
.

al

Ui

and

ei

as in a purely elastio case.

Zn the equations of definition (1) and (2), the material ieItaken

to bo inccmpromitb

and Poismnts ratfo = ~,

The Qtross-utruin

r61ations compatible with the equations of definition (1) and (2) are:

.
m

.,

NACA TN NU. 1356

.

ax - Ady
s
2
x
= e =
x
E eec
E
sec

(3 j

,=A
see

. . .. . .

.
.

These relations imply isotropy of the material.

-0
Variations of strain and stress-- W.en buckling occurs, let
~x$ ~;$ and Y vary slightly from their values lefore buckling.
The variations 5CX, tjey, and 57 will arise partly from the
variations of middle-surface
tending; thus,

strains

and partly from strains due to

5EX =

&

G1

Y=

2X1

- ZX2

Ey = 2E - 22X
3
3
,,

in which
.

EL

and

e2

.-

1~~-.
(4}

are.piddle-surfacestrain variations and

is the middle-surface shear-strain variation, Xl

changes in curvature and

-.

and

X2

is the change in twist, and

distance out from the middle surface of the plate.

we

the

is the

NACA TN No. 1556

The corresponding variations

Y

and

5SX,

&Sy,

and

bT

in

SX,

must he computed. From equations (3),

SX SE sec

therefore

(5)

Now the variation of the work of the internal foroes is

so that
, ,.. i

,,..

.* ..+

5ei =
i

Uxq

a G
Y2

2TE

~-z~X~i-ax2+2Tx
(6)
al

Substitution of this value of

in equation (5) gives

bi

- z axXl + ay?$ + 2T
5)]

.-..

NACA TN NO. 1553

---

the coordinate of the surface for which ael = O (the neutr~

surface) be z = Zo. The expression for z. is obtained%y sitting

Let

bei = O

in equation (6);
(7)

z =
o

By introduction of this coordinate into the expression%or

dcfl as
i ~B E
by recognition of ~
q
ten
Sec -~

5SX

and

5S= = E ~e=(l - i)

+
.

x
E
u~ei sec -Etan )(%\ax

+ cfyX2+ 2Tx~

(8)

(Z-zo)
)

In a similar way it may be shown that

and.

,-

&r =

?&c(, - .x$

?E
3~iei

>.

sec -

E-WI)
(axxl + cry% + 2TX~(z)

ZO)

(10)

NACA TN No. 1556

10

.!

the variations of the

Var~ttons of forceeand mom nts.- 11or
impre~d
forces Tx, Ty, and TW
and the moments ~,
MY,

L:.
,

1
I

(U)

./.
I

h
,1
-

l.

5M
Xy =

2
h8TZdZ

-%?.

where

is the thicknesg of the plate.

.
1:
i

NACA TN No. 1~~

11

#
r

.
.

where
E
D? = Sec 3
g

--

,*
.

Similarly,

dz

..Dt{[_,(.22@j
.
.

(13)

(14)
In these expressions,,the integrations of

5SX,

6SY,

and

57

in

the plastic region h&e been taken over the optire ~hfckness of the
plate, with the implication that no part of the plate is beln$
unloaded, .
. .
Ectuatlonof eauilibriug,- If W(X,y) 1s the bending deflection
of the plate at buoklin~, and if no external
moments are applied to
the plate, then the equation of equilibrium of an element of the
plate may be written

in whioh the impressed forces

as given (ax and
l?, the ch~es

uy

aYh, uyh, and fh are considered

.
are positive for compression), In terms of

in curvatures are

&3 ,
.
.

(3J%}

and
(161))
The chanGe in tvist is
-

b2w
x

(160)

&&
in equations (12), (13),

and (14),.respectively, are differentiated as required bj-e~uation (15)

c.ndsubstituted in that equation, the Cenere.1differential.equation
of equilibrium for a plate in the pkstic state is [email protected] es follows:
.
.
.

(17)
In the elastic range, equation (17)
a%
ha
-.
D
:2
(
,

reduces to the umol

&

+ Y ?@

vhere
D=@
9

~ ,T

a%
G

).

fOrm

.-

14

NACA TN No. 1356

EnerIz y inte~al.q.- Equation (17)

results f~m
1

is the Euler equation that

a mfldmization of the integgal

#w s??
.
ax%xay+

(18)

..

which representsthe difference between the stratn ener~ in the

plate and the work.done on the plate bythe external forceq. The
Coefficientein thi8integraJ.are:
In the plastic region

,,.

,.

Crxr

C2=3-7

1
..

C3 a-l

...

()
l-F-

a a

tan

the elastic region

1=2
C2

sec

+ 2,2

%a)

--3 XY
4,
=i2

1
(

- E

i3eo
)

=0

c
,

=1

Ck=o

.,

,.,

.,

.,1

restraintof magnitade
this restraintitself is taken to

there ia a

G
%
Y.

is the ed.py mcoxiinate. (Sea refmemce 9 for fom

of eqnwss!on.)

in

expression (19), tba

stiffness
D1 ~EIam-d
to le the
mm
as that
in eguation
(U). If rest~ts exe present along
Tbeae terms may be added to
twoedfps, there till %s two terms similarto expref3sion
{19).
inte~
(18) as additional
strah
emmg.
. .

,.

i.nwhichthe vibes
of the C;s
infha @iC
ran@
are &ad.
: This
.e&wsOloi
atrese Intensitymay be mlnlmized as with the correapondlng elastic case.

,,

,,

ror

,,

the

crlticsl--

of the
If the values
critical-stress intensity

Cs
h

In
the elastic region are used in formula
the elastic region
~uJel
Is as fo~~:

(20), the

&

(21)

(&)

. ..

The theory till now be

Amlicatl ~ to Pl~tea commwmed in %~e x4l9.rectfoa.appliedto flat, rectangularplates wdfomily compressedin the x-direction. Values
of ~ @.11 M computedfor the followingoases:
I.

II.
m.

,.

,.
*

*,

.,

!
.

.,

>,

Wb9n

aY,=T

=0,

q = ax and tbe

plasticity

cceff icienlm reduce to

C2=C4.

Cc=
35

1.

end.

The Mffemntial

f3quat~anuf

equilibrium,

equation

(17), then becmas

(23)

and th
ImmlQaa

correepnttlmg enbrfg- expreadon

(20) for tliecrltlcd. ati+m

in theplagtio range

(24)

Cam

1:

RlasMcally

nmtral&d

fI.ange

IB the elaaticellynmtraimxl edge of the ~lenge and y = b ia the free

If y :0
e~, a deflectionwmface lmcwn to he good in the elaetlo range aml presumably saHfactory also lwycnd thlo range is (reference9)

1,

,,,

I-J

a)

J
where

<

a2

= 4.963
=

9.&j2

.53 = +.na
E
is the ma@tude
ad
in eqnatlon (24) givm

of the elastic restraint.

Suln3tltutlon

of this expressionfor

mere
c1

= 0.23694

= o.56712

C2 = 0.79546

C6 = 0.17564

= o.@395
C3
CL = 0.04286

C7 ~ 0.19736
c~
C9

-2.3168

4.*

G
y

,,
.
P

.
*

.,

,.

.,

lb orter to find the ml.nhum m.lue of

x% ~
=
a(+!)2
\. . .

.,

The

SdJlhml

Vslw

of

( U=)ml

1s tbmef ms

For the elastic case, the sfum3 expresslm

D?
replacai by D. Fran equat ion (24

is

obtafned

from

equation

(21)

with

Cl u 1

and

thkmef OR ,

.. .
-.

,.

-.!

...

,.

., -

1
,1

.
I

.-

(a)If the e&3

.y= O

fe Mrgaa,

G .0

and, fkm

eqmti&;

(&3),

Sec
l-l=-y-

(26)

figure 1. -TbB tidividual points re&ese~t

the NACA tests of the buokling
section colomm
for which the ctmdition
6 . 0 is t%lfil.led.
{b] If the ed@

y = O

iB clemped,

6 = 01 and, from equation

of crudkom-

(5),

or

,=L4330+b.6.&7zj-)
1
.

.
.

,.
,,

..
.

NACA TN NO. l~fi

This value of
in figure 1.
Case II:
If

is plotted as cke

forr2@T-~&inum

..

Plate e~asticallyrestmafnpd along two @oaded

Y=V2+ ~

alloy
edges

are the immovable unloaded.edges which are

. .

elastically restrained against rotation by restraids of magnitude ~,a

satisfactory &efloction m.lrfcmeis known to be (refermce 1.0)

Substitution of this expression for

.-

-.

w in equation (2k) gives

: i

..-.
,.

,..

where

.
.
.

0.0046U%

and

0.0947G

.,

.
f2(G) =

o.o114e2

+ 0.1894E

o.oo461e2 + 0.0947G

In order to.find the minim-value

of

which gfves

+ 1

Cr

-.

+ $

()x @
o.

22

NACA TN No. 1556

For the elastic case, the se~ression is obtained from

equatton 21) with Cl = 1 and D~ replaced by D. From
equation [22), therefore,
s
l-l

+ f2(~)
sec.2{C; \f@
E
2 fl(c) + fp(e)
..
r

(a) If the edges y = ~ ~

are hinGQd,

(28)

c = O,

fl(c) = 1,

f2(e) = 2, and, from equation-(28),

See
L!!z
-E-

(29)

2-

This value of
figure 1.

is plotted as curve C

(b) If the edges y = *

b
F

for 24S-T

sre clsmped,

e=

aluminum alloy in
~, fl(e) = 5.15,

fp( c) = 2,46, and, from equation (28),

This VdUe Of
in figure 1.

i.splottea as curve D

for 24-FATalumiruanalloy

23

IJACA

.
.
.

Case III : Plastic bucH@g

of columnf.

For the discussion of the plastic bucH.&g of columns, it ie

convenient to revert to the differential equation (23).
The plate,
when loaded as a column, has t.wo,
free ed~es descEibed by the conditions

..

. .

,.

,-

.,

A soiution of equation (2s) which identically satisfies the-first:

condition is
.
.
.
,

-.

where

a=ll

,..

i++s%i?-i)
<.

.-<

;.-..

..

,.

.
.-

.-

NACA TN

2b
In order that this
at the free edges, itis required
that

a2q

r- -&

-,-

10]
z

+P?T

a
5

l~x

.
. .

the second condition

..

,-

2tsmhg

p -

~0*

q+

)]

,m2tan
E
*=0
\ 2,
2

(31)

which Is the buckling criterion for the plate whe,nloatid as a

column. Let

(32)

where ~2 is a quantfty to be determined for three individual cases.

.,
By use of equatton (?2)
.
.
m
.

2
P -q=$

()

.,

and the buckling criterion given in equation (31) becomes

tanhg
~2q2 +
a
z

tan ~P
#p2

-=0
E

,
i

-.

25

NACA TN Ho< 1556

.

From equation
stress In the
. .

where

h
P = r 12

,.

The corresponding criticsl stress in th~ elastic range is

.

The reduction factor

is obtained from formula (22) as

g
i~= ()

s~c , ; E;en
~

(a) ,In order to investigate the case O? short c@~,

~y
approach zero. Then, by definition of ~j

(34)

5~.

.....
. -

NACA TN No. 1556

.

t~.J2
2
E

-+1

Sn addition,

The buckling criterion given i.n, equati.on

E2+
[(

(33)

therefore reduces to

&
j@~jy=o

The expression tn the brackets approaches & as

2
to satisfy the buckling criterion, therefore,

~~

O.

In order
.
.

tanh :

-+0
a

2
which mm be realized only if a itslargej that is, if ~
For ehort columm, therefore,
kj.mo
and,f~om

equmtion

(31+),

This value of
in figure 1.

is large.

..

see

--

J+ ~

+ 3%
4~

is plotted as curve E

(35)

fOr 2&T

&i.UminUlll
d-10y

.
.

NACA TN NO. 1~~

27

1,

-p+(i+(++j

w=.-

k2

~2 = o,~5375. Fromeqwtion

which is satisfied by

E
q = 0.114 -+o.

\-

[34),

tall
$86
E

(36)

This

WilLle of
q is plotted as curve F for 2&T
aluminum alloy
in figure 1.
.
.
(o) For long columns, a ~a
B become 80 small that

and the %uoK1.ingcriterion of equation (33) ,reducesto

=0
.
which is satisfied by

~2 = ~.

From eqmtion
,.

tan

i
.

=57

(34),
..
(37)

This vslue of q agrees with the experimental results of

references 2 ~
k and is plotted aa curve G for 24S-T aluminum
alloy infQure 1.

28

NACA m

NO.

1556
.

Al?PEmrx B
.

COMPARISON WITH ILYUSHIN?S STABILITY CAICTJLATIONS

The %asic difference between Ilyushins solution of the plasti~

buckling problem and that given in this paper is that Ilyushin con.
siders the plate to unload on one face es it-buckl.es~ The unloading
process results in the creation of an elasto-plastic zone in the
plate, and different eqpatiwq from those that apyly when the plaiie
remains ylastic during the bucklingprocess are required fol this
zone.
.
The differential equation foi}the buckling of a rectangular
plate when buckling is accompanied by unloading is given by Ilyushin
a= equation (3*43) of refeience 80 For simple co~iressi~
in
the
x-direction this equation is of the same form as eqution (23) of
the present paper, hut with the fo~lowing different constants:
D is used instead of D! and
.
.
.

is used instead

of

&

In the formula

for k, from equation (3,1)

of reference 8,

and, from equation (1,22) of reference 8,

When the values of

for k

and

are

inserted

into

the expression

.-

NACA TN NO. 1556

.,.

29

,:.

E&
is always larger them
, so that t~
.
sec
tan
use of k h place of
will result in appreciably higher values
.
E sec
of ~ then those given in the present ~~r.
Since Ilyushin uses
the elastic value D
there is no possibility of the solution
yieldfig a secant m&Lu&.
Curve, A to G in figure I, if computed from Ilyushin rs e qkatikn (3, 43), would sta-t with a horizontal
line at unity for curve A (Youngfs modulus) and end with curve G
expressingthe K&&
doulle modulus which is appreciab~ higher then
the tangent modulus of this paper. If D were substituted.for D
in Ilyushinrs eqyaticm (3.43), curve A would then represent the
secant modulus as it does in the present paper, lut curve G would
sti~ remain tie ==
dofiblemodulus
Therefole$
when the unl&ding
is
consi,dere~results are
of the plate durin$ the buckling Piocess
obtained which are not confirmedly expeiiment,
Computation shows that

.
.-

.-

.
.

NACA TN No. 1556

y3

,.

REmw$ms
,, , ..,, ,

1. von H&&,
Th. : Ufitersuohungeni%er tiickfestigkeit. Mitt, u,
Forschungaarb. d. V.D.I., Heft 81, 1909-.1910,
2.

R ., and Holt, Marshall: The Column Strength of

Osgood, Wi.llkll
Two ExtruiledAluminum-Alloy H-Sections, NACA Rep. 110.656;
1939,

.,!
$. Shanley, F.R.:
The Colmm l?arado~.
no. 12, .Dec. 1946, p. 678.

Jo&.

Aero.

soi.~

vol.

~~

4.-Sh&ley~F,R.: Inel.a&ticColumn Thec~*y, Jour. Aero. Soi.,

V61,14, no. 5, w
1947, PP: 262-567.

13,
..

Secant Modulus MethotlforDetermining Plate

,-~n&tab~li%yabove the Proportional Limit. Jour. Aero. So%.,
vol. 13,no. 1$ Jem. 1946, ~pa 3&Jk4 ~d 48,

.5: %~=d,ti~rse:

6i:Bimer~,
&orge J.i
Dete~ination ofPlate Compressive Stre@hs.
. !
,.
fiCA TN No.1480, X347
,

,
.

7. Handelmen, F. A., and Prager, W.: Plastic Bu&lfng of a Recta&ular

Plate under Edge Thrusts. NACA TN No.1530, 1948,

8. Ilyushin,

A. A.: The Elast@Plastic Stability of Plates. NLCA

TM No.1188, 1947.

9. Lundquist, Eugene E., and Stowell, Elbridge

Z.:
CrftfceJ.Compressive
Stress for Outstanding Flanges. NACA Rep. No. 734, 1942.

10.Lundquiet, Eugene E., ad Stowell, EllxrhlgeZ.: Critical Compressive

Stress for Flat Rectangular Plates Supported along All Edges
and Elastically Restrained against Rotation along the Unloaded
Edges. NACA Rep, No. 733,
.- 1942.
.

.+.
.

,.

-.

NACA

TN

No.

31

1556

a!

I.O
I

.,.

.8
1

\m

.6
Effective modulus
Elastic modulus

Curve

.4
.
.

.2

.
.

Hinged flange; (~~

Ctomped
Srucre flange E
w
Long plate, hinged edges

Long plate, clamped

Short

Square

Long

; fest

data, o

edges
F

column (~ (( l)
plate loaded

as a

column (* >1);

column (~=

t)

(*)

.,9

n
-

10

Compressive

40

30

20
stress,

50

ksi
NATIONAL ADVISORY
CONMTTEE ~
AEEWAUTKS

Figure 1. Computed
various

structures

A to G are

curves
of

showing

24S-T

variation

aluminum

alloy

of

scatter bond enclosing curve A shows the Iimits of

this value.)

stress

in compress~.

drawn far a material with a yield stress of 46

properties from

with

for
(Curves

ksi. The

variat-bn of specimen