Anda di halaman 1dari 43

if

4 a
c
1
FORAERONAUTICS
.
TECHNICAL NOTE 9726

COMPRESSIVE AND TORSIONAL BUCKLING


OF THIN-WALL CYLINDERS“INYIELD REGION
By GeorgeGerard
New YorkUniversity

Washington
J.,. August1956
“1
I
,,

“-

5.. . ..s. . . . .- ,.. - . . . . . . . . . . .. ---- . . . . .. . . .. . .._J I


----
NATIONAL
ADVISORY
COMIZTEE
!rEcHNIcmmcl?l
3726

COMPRESSIVE
ANDTORSIONAL
BUCKUNG
OFTEIN-WALG
CYEDUMMINYIEIDREGION
By George
Gerard

ExJmARY”
Basedonassumptions
whichham ledtothebestagreemexrt
between
theoryandtestdataoninelastic
buckling
offl.at
plates,a general
set
ofequilibrium
differential
equations
fortheplastic
bucklingofcylind-
ers hasbeenderived.Theseequations
havebeenusedto obt@nsolu-
tionsforthecompressive
andtorsional
buckling
oflongcylindersin
theyieldregion.
Testdataarepresented
whichindicate
satisfactory
agreement
with .
thetheoretical
plasticity-reduction
factorsinmst cases.Wherea
difference
inresults
exists,
testdatasxein substantially
betteragree-
mentwiththeresults byuseofthemaximum-shear
obtained lawrather
thantheoctahedral-shear
lawtotransform
axial.
stress-strain
datato
shearstress-strain
data.

INTRODUCTION
.
TJlel.astic BuckmlgofFlatPlates
Ccapressive

Thestateofknowledgeupto 1936concerning
inelastic
bucklingof
platesandshellshasbeensumarizedby Thnoshenko
inreference
1. The
maineffortswereconcerned
withattemptstomodify
thevariousbend5ng-
momenttermsoftheequilibrium
differential
egpations
bytheuseof
suitable
plasticitycoefficients
determined
frcme~xzbnentaldataon
columns
. Althoughsuchsemiempirical.
effotismetwitha reasonable
degree
ofsuccess,thetheoretics.
determination
ofplasticity-reduction
factors
forflatplateshasbeenachievedwithinrecent
yearsastheresult of
thedevelopment
ofinelastic-buckling
theory.Becausesuchdevelopments
arerecentandformthebackgroundfortheinelastic-buckling
theoryfor
shellsdevelo~dherein,thefollowing
discussion
concerning
theassump-
tionssudresultsofthevarioustheoriesispresented
insomedetail..
Different
investigators
haveuseddiffering
assumptions
inthe
development Themajorassm@ionsuuderl.ying
oftheirtheories. each
ofthesetheories
== giveninthefol.lm *bk. .

. ..—. .-- ...__ _ —.— - —-— . , ——. .. . _ — —- —. __


2 NACATN 3726

Investigator stress-strain
law Plasticity
lawBuckling
model
Bijlaard Incremental
and octahedral No strain
(ref.
2) deformation
typs, Shesx reversal
v instantaneous
Deformation
@p, octahedral Strain
7=Y3) ‘v = 0.5 Shesx reversal
Eandelman
andPrsger
Incremental
type, octahedralStrain
4)
(ref. v instmtaneous Sh!em? reversal
Stowell Deformation
@pe, octahedral No strain
(refs.
5 &a 6) v = 0.5 shear reversal

lHstoricaJQ, Bijlmrd(ref. 2)appears toham beenthefirstto


arrive at satisfactory theoretical solutions forinelastic-buckling
theories. Hisworkisthemostcomprehensive ofallthoseconsidered
inthathe comiders bothincremental anddeformation theories
andcon-
cludes thatthedeformation typeiscorrect sinceitleadsto lower
inelastic buckling loadsthanthoseobtained fqmmincreme
ntaltheories.
ELsworkwasfirstp&iLished in 1937.-t pa~r andlaterpublications
iI.lChJ.d.e i.Tleh3tiC-bUCkk@ problems
SOhI%iOIISto nEDy iIUpOrtSJIt . BbW-
ever,thisworkappeers tohaveremained unknowntomostofthelater
investigators.
Ilyushin(ref.3)brieflyreferred
toBijlaard’sworkandthenpro-
ceeded
to derivethebasicdifferential
equationforinelastic
buckling
offI&tplates according
tothestrain-reversal
model.Th&derivation
ofthisequationisratherelegantandwasusedby Stmel.1(ref.5),who,
hawever,
usedtheno-strain-reversal
model.Thedifferential
eqpation
obtained
byBijlaard reduces.
tothatderivedby Stowell
by setting
v 1/2 intheformer.Handelman andPrager(ref.4),duringthistime,
ob~inedsoltiionsto several
inelastic-buckling
problems
byuseof
incremental
theory.Testdataoncmpressed flangesandplatesindicate
thattheresultsofincrementaltheories
aredefinitely
unconservative
regsxdless
ofthebucklingmodel,whereasdefamation-type
theories
are
inrelativelygoodagreement.
Theproblem
ofplasticbuckling
hasalsobeenthesubjectofmuch
exp35metia3.
research.Theuseofthesecant-modulus-reduction
factor
wasfirstproposed
forplates loadsby Gerard(ref.
undercompressive 7)
onthebasisoftestsonZ-md channelsections.Later,Stowel.1
(ref.5)
proved
theoretically
thatuseofthesecantmodulusiscorrectforhinged
flanges
andthatforelastically
restrained
KLangesandplatesthe

.. —. .. —.———.. . .. -—-
NACATN 3726 3

plastici@-reduction
factorincludesa function
ofthetangentmodulus
inadditiontothesecant modulus.Forlongcolumus, thefactordepends
onlyuponthetangent modulus.Theplasticity-reduction
factorsproposed
by Stowell
forsimply supported
flanges andplates
ham receivedexcellent
experimental
confirmation
(refs.8 to10)andithasbeenwellknownfor
some50yearsthatthetangent modulus isingoodagreementwithtestdata
forcolumns. Thus,thetheoreticalplastic-buckknfactorsforplates
undercampressiva
loadsam= tobewellsubstantiated by ratherprecise
experimental
data.

Inelastic
shear
“BucUngofFlatPlates
Incontrastwithplasticcompressive
buckling,shearbucklingof
platesappearstobe onlesssubstantial
ground.As theresult ofa
seriesoftestsonlong2024-0 aluminwn-edl.oy
platesuudershear,wrard
(ref.U.)proposeduseoftheshearsecant modulusastheplasticity-
reduction
factorforthiscase.Thesbesrsecant modulusisdetermined
froma shearstress-strain
curve,which,accordingto reference
U, is
tobe derivedfromanaxialstress-strain
curveonthebasisofthe
maximl@-shear law. E%oweu(ref.
plastici@T 6)deriveda theoretic-
plasticity-reduction
factor
forshearwhichhasvirtudQthesamenmeri-
calvalueforallconditionsofelasticrestra4ntbetweensimplesupport
andc-d.
In comparing
thetestdataofreferenceU withthetheoretical
reduction
factor,
Stowe~useda shearstress-strain
curvederived
by
theOctahedral-shear law. TheSheSJ?@astic-buckling
plasticity test
datawerefoundtolieconsistently
belowthetheoreticalfactor.Fur-
thermore,
Stowel.1
attempted
toexplaintheagreementbetyeen
theshear
secant-modulus
methodproposed
inreference11andthetestdatatherein
onthebasisthatthestfiss-strain
curvefor2024-0 aluminum
alloycan
bewellappro-ted bya powerlaw.
Recently,ina seriesoftestsonlong,sqpare, 2014-T6aluminum-
alloytubesintorsion, Peters(ref.
10)presenteda newsetoftest
dataonplastic shearbuckling.Although
thestress-straincurveof
thismaterialcannotbe adequately
approximated
by a powerlaw,excellent
agreementwasfoundtoexistbetweenthenewtestdataandtheshear
secant
-modulusmethodproposed”
inreferenceU.. Thetheoreticalfactors
ofStowell (ref.6)andBijlaard(ref.2)werefoundtobe consistently
higherthanthetestdataby an orderofapproximately15percenttithe
bucklingstress
.
ization,
InSumlar then,theassumptionswhichleadtothebestagree-
mentbetween
theory
andtestdataoninelasticbucklingofaluminum-alloy
flatplates
undercompression
iOabg are deformation-type
stress-strain
laws,stress
andstrain
intensitiesdefined
bytbeoctahedral-shear
law,
4 NACATN 3726

andtheno-strain-reversal
modelofinelastic
budding.Although there
~ be.theoretical
objections
to deformation
theoriesasa chss andthe
useofa no-strain-reversal
modelinconjunction
withclassical
stability
concepbs,
testdatadosuggesttheuseofresultsobtained
froma theory
basedontheseassumptions.
Fortheinelastic
buckling
offlatplates undershearloading,
plastic-buckllng
theoryandthetestdata.arenotingoodagreemeti.
Theprincipal
difficulty
appears
‘tolieintheuseoftheoctahedral-
sheerlawtotransform
stress-strain
dataunderaxialloading
to sbesr
stress-strain
data.Thissituationisdiscussedfurther
hereinincon-
nection
withresults
obtained
f& torsionalbucklingofcylinders.

Inelastic
BuckMngof Cylinders
Timmhenko(ref.
1)haspresentedsomeattempts
to describe
the
inelastic
bucKUngofa cylinder
underaxialccmpessiveforces
subjected
to -symmetricbuckling.
Theseresults
arebasedontheintuitiyeuse
ofthereducedmodulus
inplaceoftheelasticmodulUs
wherethelatter
appearsintheelastic-buckling-stress
equation.
Bijlaard(ref. 12),insmextensionofhistheoryforInekstic
buddingofflatplates, hasconsidered
theinelastic
buckling
ofa
cylindersubjectto compression.
Boththe-symmetricandtheCi3?CUll-
ferentialmodesofbuck33.ng
wereconsidered
inthisanalysis.The
resultsarediscussedinsubsegmentsections
ofthisreport.
Inthispaper,a general
setofequilibrium
differential
eqyations
fortheplasticbuckling
ofcylindersisderived.
Thissetofeqaations
is~rfectlygeneralanda~liestoanyloadingsystem
leading
tobuckling.
Inparticular,
solutions
areobtdnedforcompressive
andtorsional
buck-
lingoflongcylindersintheyieldregion.
Thepl.asticim
termsappearingintheequilibriumequationsdepend
uponthechoiceofthebuckling model.Fortheno-strain-reversalmodel,
whichisusedinthisanalysis, thefactthattheaxialloadmustincrease
slightly
duringbucklinginorderthatnounloading.shouldoccurpresents
a mathematical
difficultywhenusingclassical
stabilityconceptsinwhich
theloadingremains during
COllStSJlt buckling.Tkh3 d.iff
ic~~ iS diS-
cussedandanattempttoremove itispresented.
Thisinvestigation
wasconducted
atNewYorkUniversity
underthe
sponsorship
andwiththefinancial
assistance
oftheNational
Advisory
Committee
forAeronautics.

.--—— .. —— - . — ..— ——
..—
NACATN 3726 5

. SYMEas

Ai plasticity
coefficients
defined (A13)
by equations

B axialrigidi~,Est(1
/ -,2)
b platewidth
D bending Est312(1- V2)
rigidity,
/

d diameter
E modulus
ofelasticity
ES secant
modulus

% tangent
modulus
ei strain
intensity
defined
by equation
(A2)

force
FxyFe~Fz
G shearelastic
modulus
G~ shearsecant
modulus
k~ shear
buckling
coefficient
z length
ofcylinder . .

M bending
moment
perunitwidth
m number
oflongitudinal
halfwavelengths
N loading
perunitwidth
n number
ofcircumferential
wavelengths
P external
pressure
R radius
ofcylinder
t thickness

. ..— ..— .— ._ ————. -— .—. —— -.—---—-


6 NACATN 3726

U,v,
w displacements
X,e,
!z coordinates

a= (q.,q~ - (%/%!]

B = t21=2
I
shearstrain
axialstrain
plasticity-reduction
factor

Poisson’s
ratio
elastic
valueofPoisson’s
ratio,
equalto 0.3
axialstress
stress
intensity
defined
by eqyation
(Al)
Shesxstress

(
9= ‘crl-v 2)/Es
x curvature
~4”” operator,
[@x2) + (#/R%’j12
$ = (V4)2

( )’ variations
whichariseduring
buckling,
suchas M‘ and N‘
Subscripts
:
c compression
cr critical
e elastic

——. .—. —
NACATN3726

f failing
.
s shear
%Y)w coordinate
orientation
for M, N, u,and e
1,2>3 whichariseduring
variations bucklingine and X

Theplsmfollowed
inthisreport
istopresentthetheoretical
deri.
vations
inappe*s. Thetheoretical results,
comparisons
withtest
data,anda discussion
ofthesignificance
oftheseresults
appesrin
themainbodyofthereport.
b appendix
A,theassumptions
oftheplasticity
theory
usedare
discussed
andthestress
andstrain
intensities
aredefined
according
law. Considerations
totheOctahedral-shesr involved
inthebuckling
modelarethenconsidered
sadtheincremental
forces
andmcummts
which
ariseduring
buckling
arepresented
basedontheno-stmin-reversal
model.
InappendixB,Donnell’s(ref.13)simplified
strain-displacement
andequilibrium
eqyations
derived originally
forcylinder
elastic-bu-
problems
arecombinedwiththeincremental
forcesmdmometirelations
of
appendix
A. Inthismanner, a complete
setofequilibrium
differential
equations
isobtainedforuseinthesolution ofcylinder
plastic-buckling
problems.Included
inappendix B isanattempt
toremovethedifficulty
ofusingequilibrium
equationsbasedonclassical
stabili~conce~sfor
inelastic-buckling
problemsinwhichtheno-strain-reversal
modelrequires
thattheloadmustincreaseslightlyduring
buckling.

AXIALCOMPRESSIVE
BUCKLING
OFA.LONGCUJNDER
Solution
ofProblem

~ appendix
C,theaxisymetric
buckling
ofa longcylinder
under
axialcompression
isconsidered.
Thecriti&.stress6bt&ed byuseof
theequilibrium
equations
derived
inappendix
B hasthefollowing
form:

crcr
= ~(1- $jl-’/2E#12t1Rt/R (1)

.—. . __ __ ._ .— —. -———— ———. -—.-— . .


8 NACATN3726

As intheinelastic
buckling
offlatplates,
sJleffects
ofexceedingo
theproportional
limitareincorporated
ina plasticity-reduction
factor ‘
defined
asfollows:

~c= 1(a=)
UCr e (2)

solutionfortheelastic
caseisobtainedby substituting
Es = E, Et/Es=ljand V=Ve in equation
(1):

(3)

Byuseofequation
(2),

(4)

Ingeneral,
theref
ore,

u= = o.6q&(v’R) (5)

Bijlaard (ref.
12)haspreviouslyobtained
resultseqyivsdent
to
equations(1)and(4).Illsresultsaremoreexact, infact,sincethe
variationofPoisson’s
ratiointheinelastic rangeisincludeddirectly
h theanalysis. Intheinterests
ofsimplicity,the.
presentanalysis
utilizes
thearttiicialdeviceoftakingv = 1/2 inboththeelastic
andplastic regions
andthenemploysanapproximatecorrection
(see
eq.(C15)) whichyields
exactsolutionsfortheelasticandplastic
rangesas13mits.Thismethod follows
a suggestionofStawell(ref.5).
!Depurpose
hereinobtainingthissolution
wastopresenta unified
approach
tothecompressive
= torsionalbuckling
ofcylinders
fromthe
setofequilibrium
equations
derivedinappendix
B. Furthermore,
equa-
tions(1)and(4)serveasa basisforinterpreting
testdatapresented
hereinontheinelastic
buckling
ofcykinders.
Inaddition
tothesxisymetric caseconsidez%d
here,Bi31aard
(ref.12)hasalsoconsidered
thecircumferential
buckling
modeofa
cylinder
underaxislcompression.
As intheelasticcase,thecritical .
obtained
stresses forbothbucklingmodesareessentially
thesame.
Bijlaard
haspointedoutthatformildsteel thecircumferential
mode
~ leadto a slightly
higherbucklingloadthantheaxisymetricmode.

. —.- .- ————-
2Q
NACATN 3726 9
.
Itwouldbe a relatively simplematter
to solvethecircumferential
buckling
casebyuseoftheequilibrium equations
ofappendixB. However,
intiewof.Bijlaard
isresults amdthefactthattheaxisymmetric
modeis
oftenobserved
intestsonroundcylinders whichbuckleplastically,
the
solution
forthisbuckling modewasnotpursued.

TestData
Osgood(ref.14)smdMooreandHolt(ref. 15)havepresented test
dataonthefailing strength
ofdrawncircular tubesunderaxialcom-
pression.Osgoodtested2017-!P4
aluminum-allay
tubesforwhichcompres-
sivestress-strain
curvesweregivenandalsochrome-mo~bdenum tubes
forwhich,unfortunately,
neitherthecompressivestress-strain
curves
norcompressive
yieldstresseswereobtained. Itisinteresting tonote
thatphotographs
ofthetestspecimens indicatetheappearanceofaxisym-
metricbuckling
insomecases.MooreandHolttested 6061-T6aluminum-
alloytubesforwhichthecompressiveyieldstressesweregivenalthough
ntithecompressivestress-strain
curves.A @iCal COIIQreSSiV3StIYX3S-
straincurveforthismaterisL,witha corresponding
yieldstress, was
.
taken?romreference16forcorrd.ationpurposes.
Toreducetheexperimental
dataforcompa@aonwiththeory,the
experimental
failing
strengthwasdividedbythecriticalelasticstress
(3))
(eq.. todetermine
theexperimental
plastici~-reduction
factors
givenintable1. Thetheoretical
valueof qc wasdeterminedineach
case~ useofthepertinentstress-strain
dataaccording (4).
to eq,,tion
ForPoisson’s
ratio,
thefolluwingvalueswereused:
Ve= 0.3 (6)
v = 0.5-(EIs/FI)(O.5
- %)
relation
forthevariationinPoisson’s
ratioh the@eld region
hasbeenshowntoapplyto isotropic,
plastically
incompressible
solids
by Gerard
andWildhorn
(ref.17). -
Thetheoretical
ande~rimentalvalues
of qc asa function
of
theinelastic-compressive-buckling
stressareplotted
infigure1. For
rangeoftestdataon 2017-Tk
thelimited and606LT6alumimm-alloy
tubesitcanbe obse-d thatgoodagreement
isobtained.Thisisremark-
ableinviewofthefactthatwhileittendsto coul?irm
therdationfor ‘
theplasticity-reduction
factor(eq.(4))italsoconfirms
theclassical
‘smal.1-def
lectionstability
theoryforcompressed
cylinders.

-. . ...— . . .—_____ .—. .—. — — .—. _... . . . —____ . ..__ .__. .__.
10 NACATN 3726

TORSIONAL
B’UU_GOFA LONGCYUNIER
Cylinders
ofExtreme
Length

InappendixD,t@ torsional
bucklingofa longcylinder
iscon-
sidered.If thecylinder
isofsufficientlength,
theboundsry
condi-
tionsattheendshavenegligibleinfluence
andtwo-lobe
buckling
occurs.
Thecriticalstress
forthiscaseobtainedbyuseof theequilibrium
of ap~ndixB hasthefollowing
egyations form:

‘Cr= 0.272(1
- v2)-3/4~s(@)3/2 (7)
In the elastic
case,equation
(7)becomes
-3/4 3]2
()
‘cre = 0.272(1
- Ve2) E(tfi) (8)

Donnell 13)hasshownthatequation
(ref. (8)applies
for

(9)
wherea = 42 forsimply
suppo~ed
endsand a = 60 forclsm~dends.“
By useoftbeequivah?nt
ofequation
(2)forthetorsionalbuckling
case
v 2 3/4
(lo)
‘s= ()
- ‘s/E
Ingeneral,
therefore,

(U)

Theplasticity-m=duction
factor
depends
primarily
uponthesecant
modulus,whichhasbeenfoundtobethecasewhenever
buckling
occurs
as
a twisting
action.Thishasbeenpreviously
observed
forcompressive
bucklingofhingedfI..anges
wherebuckling
occurs
asa twisting
action .,
andtheplasticity-reduction
factor
depends
primsrily
uponthesecant
modulus
.

.—— — .
NACA~ 3726 XL


h apwndixA,thesecant
modulus
isdefined
as

.- Es = aiei
/
‘ (12)

Iftheshesrstress-strain
curve
‘isknown(thiscanbe relatively
simply
obta’ined
bytorsiontestsontubesalthough
no suchdirect-xperimental
procedure
exists
forobtainingthesedataforfla.t
plates),thenthe
respective
stress
andstrainintensities
according
to equations
(Al)
and(A2)areasfollows:

(13]

ei= (3)-1/27 (14)

Byuseofegpa%ion
(1.2)

Es = 3T/y (15)
SinceT/y= Gg,fromequation
(15)
Gs = EB/3 (16)
Therefore,
egpation as
(10)canW interpreted

f
2\./4
1 - v.
% =\*) Gs /G (17)

incases inwhich theshearstress-strain


curveisavailab”k.
H, ontheotherhand,shearstress-straindatamsynotbe availa-
ble, thenitispossibleto construct
a shearstress-strain
curvefrom
simp~tensionandcompression
stress-strain
databy useofeither the
octahedral-
orthemaximum-shear
law. As indicated
inreference11for
materials
whichareanisotropic
asa resultofstraightening
(Bauschigr
effect
),itisprobably besttoutilizesmsxialstress-strain
curve
whichisanaverage ofthetensionandcompression
curveseachat 45°
withthedirectionoftheappliedshear.H theaxialstress h the
simpleaxialtestis ~ thenthestress andstrainintensities
according
totheoctahedral-shear lsw$are
andmsximum-shesx

. .. . .. .._—... ____ ___ ._ _— .—. ——. —————._ .. ___


12 NACATN 3726

q = ax
(18) “-
ei = Gx w
}

Thus,inbothcases,

Es = ~px (19)
However,
theshearstresses to ax aredifferent.
corresponding
Foroctahedral
she=

T = (3)-1/2% (20)
andformxdmumshear
T =ax2
/
(a) -
J.
Thus,thevalues of Es andtherefore qs fora givenvalueof ~
correspond lawas compared
tq a lowervalueof T for themaximum-shesr
withthatoftheoct-dral-shear -W.

Cylinders Iength
ofModerate
A solution
fortheinelastic ofcylinders
buckling ofmoderate
length}
inwhichcasetheboundsry I&e a decided
conditions influence
uponthebuck13.ng
stress,
hasnotbeenobtained
herein.However,
the
elastic
solution
isknownandhasbeengiveninthefollowing
formby
Batdorf
(ref.18):

Tcr = 0.7k7E(tfi)5’4(R/Z
)1’2 (22) “

for 50t/R< (Z/R)2


< 10R/t.
Forshortcykbders
defined
apprmdmately
by Z2/Rt
<1, thefht- ~
platesolution

(23)

— —-. . — -.. --.—- - —


NACATN 3726 13”,

applies
where
‘nfl
=Gs
/G (24)

Theplasticity-reduction
factorforthiscasehasbeenproposed
by Gerard
11)andisbasedonthemaximum-shear
(ref. lawtotransform
theaxial
stress-strain
datato shesxdata.
Sinceequation
(24)for”
shortcylindersdepends
primarily
uponthe
secant
modulusasdoesegpation(10)or (17)forlongcylinders,
it
appears
thatequation(10)or (17)
my beusedasa reasonable
approxi-
mation
of q~ forcylindersofmoderate length.~

.WstDa-ta
Stang, andBack(ref.
Rsniberg, 19),MooreandPaul.
(ref.
20),and
15)
MooreandHolt(ref. havepresented
testdataonthetorsional
failing
-. strength
oflongandmoderate-length
drawncircular
tubes.Inmostcases
Occurred
failure asa result
ofinelastic
buckling
inthetwo-lo~e
mode.
Stang,Ramberg,
andBacktested2017-T4andchrome-molybdenum
tubeB
forwhichrepresentative
shearstress-strain
curveswerepresented.In
thiscase,therefore,
itwaspossibleto correlate
theoryandexperiment
onthebasisofequation (17).Ofthelargemassoftestdatagivenin
reference
19,a relatively
smallsmount
wasuseful forcorrelation
pur-
poses.Thesedhtaaregivenintable2 andwereselected onthebasis
thatthetubeswerelonginthesenseofequation (9)(clsmpedends).
Furthermore,
theyieldstress
ofthe2017-T4 tubeswasapproximately
23ksi- thefacingstress waskss tti 26ksito COrreSpOIMIWith
thegiven shearstress-straindata.Forthechrome-molybdenm tubes,
t~ COrreSPOm VSJJES were 49w 58ksi} respecti=~. MSW ofthe
othertestdatawerebeyond
therangeofthe@ven stress-strain
data.
Incomputing
theexperimental
values
of TIS,thetorsional
failing
stresswasdivided
bythecritical
elastic
shearstress
ofegpation
(8).
Sincetheshesrstress-strain
dataweregiven,
thetheoretical
values
of qs werecalculated
byuseofequations
(17)and(6)forPoisson’s
ratiowith Gs/
G replacing
EslE inthelatter.
Tbetheoretical
andexperimental
values
of q~ areshowninfig-
ure2. Itcanbe observedthatgoodagreement
isobtained
forthe
2017-T4
data,whereastheagreement
isnotsogoodforthechrame-
molybdenum-tube
data.

. .. ... . ...— —— ——.. —- —-— ————— -—-—-—-


14 NACATN 3726

MooreandPaul(ref.20)tested6051-T6seemless
tubesofmoderate
length.Tensionstress-strain
dataarealsogiven.Tbetorsional failing “ ‘
stresses
arelistedintable3 together
withtheexperimental
valueof qs
calculated
byuseofegyation(22)fortubesofmoderatelength.The &
theoretical
values
of q~ weredeterminedbyuseofequations(10)
and(6),usingthetensionstress-strain
data._ infigure 3 isa
comparison
ofthe qs valuesbasedontheuseoftheoctahedral- and
mudmum-shear
lswstotransformtheaxialstress-strain
data.Eetter
correlation
isobtained
withthelatter,ell.though
thetestdataarebelow
thetheoretical
values
of qs This~ possibly
● reflecttherelatively
lergescatter
emongthetestdataorthefactthatthevalheof q~ given
by equation
(10)isapproximate
fortubesofmoderate
length.
RelativelylargescattercanalsQ be observedforthe6061-T6
aluminum-alloy
testdataofMooreandHolt(ref. 15;alsoEstedin
table3 andshowninfig.3). Thesetestswereconducted onbothlong
tubesandtubesofmoderate lengthandaresodesignated intable3 and,
figure3. Thetheoreticalvaluesof q~ werecomputed froma typical ,
stress-strain
curvefor6061-T6~uminum~oy giveninreference 16and
hatingtheseineyieldproperties
aslisted inreference15. Inthis
case,thetestdataagainfavertheuseofthemaximum-shear lawto
transform
thetial datawithapproximately equalscatterofthetest
pointsaboutthetheoreticalline.

DISCUSSION

Indiscussing
thecorrelation
between
thetheoretical
plasticity-
reduction
factors
andtheavailebk testdata,itis convenient
to sum-
merize
theresults
as shownintable4.

compressive
Buckling
Forthelimited emountoftestdataon2017-T4 and6061-T6
aluminum-
alloytubesincompression,itappears
thattheplasticity-reduction
factorgivenbyequation(4)isinsubstantiallygoodagreement
withtest
data.Considerablymoreweightmustbeplacedonthe2017-T4dataas
ccmparedwiththe6061-T6datasincecompression
stress-strain
datawere
givenfortheformer, whereas
suchdataforthelatter wereestimated
fromothersources. Thus,theagreement
obtainedforthe2017-T4data
canbe interpreted
asexcellentsupport
forthetheoreticalvalueof qc. f
As indicated thee~rimentalvalues
previously, of qc forthe
testdatawerecomputedusingequation(3)whichisbasedoncl.assicd.

-. ..-
NACATN 3726 15

stibili~concepts.Sincetestdata,
onelasticbuckling
of compressed
A
cyLinders
generaUyfallconsiderably
belowthistheoretical
value,it
appears
fruitful
to discusstheimplications
oftheapparentagreement
oftestdataoncylinderswhichbuckleinelastically
withequation(3)
afterinelastic
bucklingeffectshavebeenaccounted
forby useof qc.
Thecylindersusedfortheinelastic testsweredrawnsesmless tubes
with R/t tiueskSS than 50. Conseq@tly, thesetubesprobablycon-
tainedverysmallgeometrical imperfections
andwererelatively freeof
residualstresses. ~ contrast,testsonelasticbucklingweregenerally
conductedon c~tirs fabricated fromflatsheetswith R/t values
~ti from200to 3,000.MaIIY ofthesetestswereconducted oncyl-
indersofverythinsheetstockandtherefore thegeometricalimperfec-
tionswouJdbe expectedtobeverymuchgreater thanthoseindrawntubes.
Thus,itisprobably safeto conclude
thattheinitial tiperfections
for
theinelastic cyllnders
wereconsiderably
lessthanthosefortheelastic
cylinders.Provided theplasticity-reduction
factoriscorrect, theini-
tialimperfections
wereapparently ofsucha smallmagnitudeforthe
inelasticcylindersthatthebucklingstressisadegyatelypredicted~
classicalsmall-clef
lectiontheory.
Thisapparent
agreement
withclassical
theQry
forinelastic
cylind-
ers mayhavesomebearing
onthecurrently
heldviewsconcerning
the
U@Z ofagreement
oftestdataonelasticcylinders
withsmall-clef
lection
theory.
Accordi~tothe”
energy
criterion
ofbuckling
usedbyTsien(ref.
21)
forperfect
elasticcylinders
ina rigidscrew-powered
testing
machine
(thecondition
leading
tothehighest
buckling
stress)

am = o.37m/lR (25)
Itisthecontention
ofthistheory thatthesmallamount
ofenergy
nec-
essarytotrigger
thejumpto largedeflections
isavailable
inthevibra-
tionsofthetesting
madine,forexample.
Donnell
andWan(ref.22)havemaintained
l%a.t
thepresence
ofgeo-
metricalimperfections
andresidual
stresses
rounds
offthesharp
~ak
inthestressend-shortening
curveofI.arge-def
lection
theoryandthere=
forefailure
isobserved
atloadsconsiderably
belowtheclassical.
value.
.
It isessential
torealizethatthehighestbuckling
stressofthe
energycriterion
isgivenby equation
(25)andtherefore
thistheorydoes
notadmitthepossible
realization
ofa bucklingstress
ashighasthat
givenbyegpation(3).Ontheotherhand,astheimperfectionsbecome
verysmall,equation
(3)isapproachedasa limitinDonnell’s
interpre-
tation.Sinceequation
(3)wasapp&ently confirmed
bythetestdata
.

------ . ..— — —— -.. ..— —. ——- .- .—


16 NACATN3726

showninfigure
1,forwhichtheimperfections
wereprobably
very-, ●

itwouldappear
thatDonnell’s
interpretation
isdefi.nitely
favored
over
theenergy
criterion
inthiscase.
w

Torsional
Buckling
Of the testdataavailabletoevaluatethetheoretical
@asticity-
reductionfactorfortorsionalbuckling
giveninfigures2 and3 and
summarizedin‘table
4,itwouldappesr thatconsiderable
weightshould
beplaced onthe2017-T4 data.Thisis duetothefactthata widerange
oftestdataaswellas shearstress-strain datawasgiven.Forthe
chrome-molybdenum
tubesonlya smallamountoftestdatacouldbeused
becausetheyieldstresses andstress-strain
properties
variedconsider-
ablyfromthoseofthesheerstress-strain curvegiven.Tberef
ore,it
wouldappear thatexcellentconfirmation
of q~ wasobtainedforthe
2017-T4data,whereasthechrome-molybdenum
dataweretoofewandtoo
variableinstress-straincharacteristics
to~rmitanydefinitecon-
clusionstobedrawnforthismaterial.
TIE605>-ti and6061-T6datad f- 3 areuseful
inproviding a
meansofchecking thetheoretical
Valtle
of q~ utilizingthemsxbmml- !>
shearandoctahedral-shearlawstotransform thesxid.stress-strain
datato shearstress-straindata.Although theconclusions
tobe drawn
arehandicapped byrelativelywidescatteroftestdataand~ thenature
ofthestress-strain dataavailableforcorrelation(table
4),itwould
appesr thatthetitsem inbetter agreementwiththeuseofthemaximum-
ShS2? ~W than withthatOftheOCtShdr=-SheS2 hW tO tX’SJU3fO~th
axial data.

SUNMARY
OFRESULTS

Thefolluwing
conclusions
werederived
froma theoretical
andexperi-
mental
investigation
ofthecompressive
andtorsional
buckling
ofthin-
wallcylinders
intheyieldregion:
1.A general
setofequilibrium
cliff
eremtial
equations
forplastic
buckling
ofcircular
cylinders
hasbeenderivedbasedondeformation
stress-strain
relstions
andtheno-strain-reversal
buckling
model.Fur-
thermore,
anattempt
hasbeenmadetoremovea difficul~associated
with
usingtheno-strain-reversal
modelinconjunction
withclassical
stability“
concepts.
2.Theplasticity-reduction
factors
forinelastic
buckling
oflong
cylinders
undercompressive
ortorsional
loadings
ham beenderived.

—.. ...—. — -.
IQ
NACATN 3726 17

. It isshown
thatthesefactors
areinsatisfactory
agreement
withtest
datawhensatisfactory
compression
andshearstress-strain
dataare
available.
andoctahedral-shear
3.Boththemsximum-shear plasticitylawswere
usedintransforming
axialstress-strain
datato shearstress-strain
datafortorsional
bucklingofcy~ndersinconjunctionwitha theoreti-
callyderived
plasticity-reduction
factorbasedontheoctahedral-shear
IRw. Insuchcases,resultsobtained
byuseofthetransformed shear
lawareinbetter
databasedonthemaximum-shear agreement
withtest
datathanthosebasedontheuseoftheoctahedral-shearlaw.
4.Compression
testdataontubeswhichprobably
contained
small
geometrical~
rfections correlated
verywellwiththec,ritical
stress
predicted
by classical
small-deflection
theory
afterthetheoretical
correction
forinelastic
buckling
hadbeenincorporated.
Thiscorrel&-
tionisviewedasa factor
favoring
theimperfection
interpretation
(IMnell)oftestsoneksticcylindersovertheener~interpretation
(Tsien).
.

Research
Division,
College
ofEngineering,
NewYorkUniversi@,
14,1954.
NewYork,N.Y.,October

,
. ..

.. . . .. . . - ...—— .~-—— _ . —. .-_. —___


18 NACATN 3726

fWPENDIX
A

PLASTICITY
CONSIDERATIONS

In tlk foU6wingderivations,
assumptions
havebeenemployed
which
appear
tohaveresulted inthebestagreement
between
theoryandtest
dataoninelasticbucklingoffletplates
withvarious
geometrical
bound-
aryconditionsandtypesofloading.
A ~ti~ ~OtheSiS Of ph3StiCity _th!30ry
iS that t~ StI’eSS
intensityui isa uniquelydefined, single-valued function ofthestrain
intensityei fora givenmaterial.whenthestress intensity increases
(loadbg)andisehsticwhenitdecreases (mil.oading). Thedefinitions
ofthestressandstrain intensitiestheoretically canbe chosen froma
manifold
ofrotationallyinvariantfunctions. Twosuchfunctions, the
msximum-shear
andoctahedral-shearlaws,havebeenuseful.

Fortheoctahdral-shear
law,thestress
endstrain
intensities
cambe def3nedasfollows:

..= (o--x’
+Oj’- G&j+ #2 (Al)

(A’)

Withtheassum@ionthattheprincipal
axesofstress
andstrain
coincide
, thesecant
modulus
canbedefined
as

Es = Uiei (A3)
/
~heI’More,byuseM *fo~tion-type stress-strati
lawstogether
with
theassumption
ofplasticisotro~andtheidealization
thatPoisson’s
ratiois~qualto 1/2forboth& elasticandthepla;ticregion,
the
following
simplified
two-dimensional
stress-strain
hWS are obtained:
>
%=+& - (way’] (A4)
NACATN 3726 19

(A5)

(A6)

lhelastic-Buckling
Considerations
Alloftheforegoing
assumptions
formthebasisforsolution
of
plastici@problems
ingeneral.
Forthespecific
problemofinehstic
buckling,
itisnecessary
tomakeanadditional
assumption
concerning
thestressdistribution
attheinstant
ofbuckling.
lh?om
thestandpoint
ofclassicalstability
theory,
theequilibrium
differential
equations
areformulatedonthebasisthatatthebuckling
loadanexchange
ofstableequilibriumconfiguration
occursbetween
the
. straight
formandthesli@htlybentform. Sincethe‘load
remains
con-
stantduring
thisexchange,
a strainreversal
mustoccurontheconvex
side,and,therefore,
thebucklingmodelleadingtothereduced-modulus
concept
forcolumusiscorrecttheoretically.
Practiticolumns andplates
invariably.
contain
initial.
imperfec-
tionsandthereforeaxialloading
andbending proceed
simultaneously.
Sinceinthepresence ofrelatively
largeaxialcompressive
stresses
thebendingstresses~ generslldy
smal!d.,
no strain
reversalwouldbe
expected
to occurandtheincremental
bending stresses
intheinelastic
rangearegivenbythetangent-modulus
model.However, thebentform
istheonlystable configuration
inthiscaseandtherefore useofequi-
Ubrim equationsbasedonperfectcolumns,
plates,orshells is clearly
unjustified.
Partially
toremuvethisdifficulty,
StoweSlhasassumed thatthe
straight
formoftheplateorcolumn isstable untilbuckling occurs
(ref.5). Atbuckling,
infinitesimal.
bendingisassumed to proceed
simultaneously
witha corresponding
infinitesimal
increase insxiii
loading
sothattheplateisnotsub jettedto a strainreversaland
remains
inelastic.Againthismodelposesanessential difficultysince
classical
stability
theoryisbasedontheassumption thattheaxial
losding
ramdllsconstant
duringthebuckling
process.
lha~endixB, inwhichtheequilibrium
equationsare considered,
anattempt
ismadetoremove thisdifficulty
by ShOWiIlgthat the infin-
itesimal
increase
inloadassociatedwiththeno-strain-reversal model
contributes
higherordertermsthanthosegenerallyconsidered inthe
eqyilibrim
equation.Thisisbyvirtue oftpefactthattheaxialloads
aremul.tiplJed
byfirstorsecond derivatives
ofthedisplacements and

. . ... -—__ ___ _ ——. . —--- .. . ..


20 IIACA
!r!N
3726

therefore
products
oftheincremental.
loadincrease
andthesederiva- >
tivesresult
in second-order
terms.
.
Incremental
Forces
andMoments
Nhenbucld.ing
occurs,
thedisplacements
varyslightly
fromtheir
valuesbefore
buckling.Theresultingstrainvariations
arisepsrtly
franvariations
ofmiddle-surface
strains
md partlybecause
ofbending
strains.Theseresulting
variations
ofstresseshavebeenconsidered
by Ilyushin
(ref.3)smdStowell(ref.5). Usingtheassumption
that
nopartoftheplateisunloaded, %owellhasderivedthevariations
ofthemomentsduringthebum process.- -iatio~ tithefiu-
surfaceforces
canbe derived
directlyfranthiswork.
Whenthevariations
oftheforces aredenoted
andmoments by
primes(‘),thefollowing
relations
applyto~ plastic
platesdurin$
buckling:
1
= B 1!
I?x’ 161i-(1/2)A~e2
- (A7) ‘

%’ = B[J%2 + @/@A21% - (A8)

=B 363- (1/2)A51e1
‘xY’ ~ 1! - (WA3262 1 “ (A9)

%’ = -DAl%
[ -I-
(1/2)Au~- (no)

MY’= -DA21@+ (1/2)~Xl- (Ku)


[
- WM32*]
Mw’ =-~ 3X3- (1/2)A31Xl (Au)
1!

~ equations
(A7)to (A12),G1 a E2 aremiddle--acenormal
strainvariations
@ 63 isthemiddle-surface
shearstrain
variation; .
xlti~=thechangesin~- -~is*_efi
twist
. Furthermore,
theplasticity
coefficients
aredefinedasfo~ows: u

. .— —-. — . — ——
MICATN3726 21

Al = 1- (cLux2/&)

& = 1- (*74)

A3=1-crrz
1

A31= A13= aux’r

.
where

Thesxial
rigidi~is
B = &Est/3 (A14)

Thebending
rigidity
is
D = Est39 (~5)
I
thee~sticregion,cc= O smd,therefore,
~ Al = A2 =A3=
A= = 1 snd A13= %3 = O. Byreplacing thedefinitions ofequa+
tions(A14) and(A15)whicharefora fullyplastic plateby
B = Et/(l- v=’) and D = E%3/12(1
- V=2),respectively,sndreplacing
tk coefficierrt
(1/2)by Ve,equations (A7)to (A12)reducetothe
familiarrelationsfortheelasticplate.
.

,,

_ ... .. . ..—. _ —-— .- >- .—. — -. —


22 NACATN 3726

APPENDIX
B

E~ CONSIDERATIONS
Elastic
Buckling

Donnell’s
equations
(refs.
13 and18)forelastic
buckling
ofthin-
wdl circular
cylinders
havebeenusedwitha considerable
degreeof
success
inbuckling
problems.
Therefore,
inthisinvestigation
ofine-
lastic
buckling
oflongcirculsr
cylinders
undercompressive
andtorbional
loads,
anextension
ofDonnell’s
equations
is considered.
Themiddle-surface
strainvariations
andcurvature
changes
that
occurduring
buckling
ofa circular
cylinder
arerelated
tothedis-

(Bl)

Thefollowing
simplifiedequilibrium
egyations
asderived
byDonnell
(ref.13)neglect
certaintermswhichsreofsmallmagnitude
whenthe
Ciml.lur
crosssection
ofthecylinder isdistorted
during
buckling.In
casesinwhichthecrosssectionretains
itscircular
shapeduringbuck-
Lblgthenegl.ected
termsaregeneralJyofscmeimportance.
.
~Fx q aN ‘ =()
+J!L (B2) d
‘—ax R&I
d

(B3)

—- .. —
NACATN 3726 23

a% +P=o
(B4)
% &?#

Inequation (~), Nx, NV, Ny,and p areprescribed


external
10adiII&, with Nx and Ny positive
incompression.
Thetermscon-
taining a primearethevariations
associated
withbuckling.

Inelastic
Buckling
~ appendixA, itwasindicatedthatuseoftheno-strain-reversal
modelforinelasticbucklingposesthedifficulty
thattheexternalload
mustincrease
slightly duringthebuckling
process.ThisisatVariance
withclassical
stability concepts
whichrequire
theexternalloadto
remainconstant
during buckling.Therefore,
samejustification
isnec-
esssryinordertouseequilibrium equations
bas”edonconstant
external
loadsforinelastic
lnmkl.ingproblems
inwhichtheexternalloadmust
increase
slightly. An atte@ hasbeenmadeto clarifythispointwhich
hasbeenoverlookd byprevious investigators.
Itisassumedthat,intheinelastic-buckling
process,
theexternal
loadsincrease
slightly.Denoting
thisincrementby ~, theexternal
loadsareincreased
asfollows: Nx+ ~x,!NW +&NW, Ny+~, snd
p+ 8p. Thetermscontaining
a primeinequations(B2)to (l#k)
arethe
middle-surface
forceandbending-moment
variations
arising
franbending
andtwisting
oftheplateatbuckling. Therefore,
theslightincrease
inefiernal
loadrepresented
by 8N canhaveonlya negligibleinfluence
upontheprimedtermsinequations(B2)to (Ilk).
Inequation
(w),theexternalloadsNx, NW, ~, and p appear.
Iftheseloadsarereplacedby Nx+ 5NX,. . ., p + bp,thenterms
(
such as Mx a%/ax2)and tipappear
thanthosetermsappearing
inequation
which areclearly ofhigher
(~) endcanbe neglected.
order
Thus,
ita~earspermissible
to conclude
thattheslightincrease inload
requiredfortheplate,
to remaininelasticduri~buckling iscompatible
withtheuseofequilibriumequations
basedonclassical stability
. concepts.

.. . .— -. ——..— —..— . . .——.. — —---- -. —-


24 NACATN3726

Equilibri-
Equations
~ useoftheforceandmcmentvariations
(eqs.(A7)*O (A12)
)d
thestrain-displacement
relations
(eqs.(Bl)
) theequilibrium
relations .
(eqs.(B2)to (*))Cm bewritten
intermsofthedisplacementsu, v,
and w andtheirderivatives:

(B5)

i+w ~ + (AE+ ‘3) ahw


- A23— a4w +
R2a&?ae2
‘A1a-A13 R ax3a(3 R3& hg3
[

A2

Nx (B7) “

Eq@ions (B5)to (B7)consti- a basicsetofequilibriumdif-


ferentiql
eqqations
forplasticbuckling
ofcircular
cylinders. In
case,Al =A2= A3 .Au. 1 and A13=~3=0, snd,by
theelastic
properly
accounti~forPoisson’s
ratio,e~tions (B5)to (B7)reduce .
tothefOllowing:

.-. ..—--— . .— —— -— -- .— .— ..
lQ
NACATN 3726 25

a% 1-vt3i3%+1+vt3
—— — a% bW=o (B9)
R2ae2.+ 2 a# 2 m + R2ae

m+ B
w+ Fve auav
—+— ~ +Nx”~+2NWR
( ax Raf3+R)
~ae+Ny~+P=0
R be
(Blo)

By suitable
manipulation
of equations (B8)to (B1O),
theaboveset
ofequations
canbe reduced
toa single equation
indefl.ection
w known
.
asDonnelJ’s
equation:

. . . ..- .-— .—— ._ ___ .-— —. —. -.. -—---


26 NACATN 3726

c
APPENDIX

AXIALCOMPRESSIVE
BUCKLING
OF

Fora longcimcular
cylinder
subjected
toaxialcompression,
%=’= o. Thevalueoftheterm ~ which
appearsinequations
(A13)
isgivenbyeqpation(Al)asdiscussed
inreference
5 and,therefore,
forthiscase Ui= q. Thus,theplasticity
coefficients
reduce
to

#+$>
s
A2 = A3 =A12=l (cl)

%3=%3=0
1
egyilibrium
Consegyently, (B5)to (B7)
equations reduce
tothefollowing .
equations:

(C2)

a% +la%+~
-— a% aW .0 (C3)
R%(32 4 a# 4 R= + R2ae

Compressive
testsofcylinders
whichbuckleplastically
indicate
thatanaxisymetric formofbuckling
oftenoccurs.!l!heref
ore,ifthis
modeofinstabilityisassmed,thedisplacements
sxeindqendent
ofthe
e coordinate
andallderivativescontaininge vanish.Thus,eqpa-
tions(C2)to (C4)reducetothefoll.cnkhg
forms:
NACATN 3726 27

a%+~law=o
‘la# 2Rax (C5)

a2v ~
—= (c6)
ax?

(C7)

Byperforming apx onequation


theoperation (C7)andthenusingegpa-
tion(C5),
a single
equilibrium
equation
in w isobtained:
;

.
(c8)

A solution
toequation
(c8)canbewritten
inthefollowing
form:

w = wm sin(~/A) (C9)

where
h= Z/m

Uponsubstituting
theappropriate
derivatives
ofequation
(C9)into
equation(c8) andusingthedefinitions
of D, B, and Al givenby
equations(A15),(A14),and(Cl),
respectively,
andtherelation
Nx = awt,thefollowing nontrivial
solution
isobtained:

Est2~
‘cl?=-- ‘——
9 4+4E@ ()
3%2+ %
R ~+~%
n 4 4 .%3
A2
~
(Clo)

.- ——...--—.—. .—— —z — — —— ----- ._


28 NACATN 3726

Thecylinder
isconsidefid
tobe long,sothatmanywavesform
alongthelength
andtherefore
Ua canbe considered
asa continuous
function
of X. Byminimizing
equation withrespect
(C1O) tothewave
length,
l/2
%=S 2E SREB
t%
-()
(Cll)

halfwavelength
corresponding ofthebuckles

(c12)

Fortheelastic
case,thecorresponding
solutions
are:

“== [i+ - ‘.2)l-1’24i


- (C13) -

A= @t)l/2
l+ -
[
(C14)

whichappear
By cmpsringthecoefficients inequations(Cll)
(C13),it canbe observed
thatthecoefficientinequation(C1.1)
canbeobtainedby substituting
a VS2ueof1/2for Ve inequation (C13).
Thus,thefollowi.ng
relations
canbewritten whichsreexactinthe
elasticandfullyplastic
rangesandresultinanexcellentdegreeof
approximation
intheinelastic
range:

(C15)

(c16) -

.- —.
NACATN3726 29

APPENDIX
D

TORSIO~BUCHJNGOFA LONG~ -

Fora longcircularcylinder
subjected
totorsional
mcmenta
atthe
ends,ax= ~ = O. Thevalueof ui whichappears
inequations
(A13)
isgivenby equation
(Al).Forthiscase,

(ZJ_
= (3)% (Dl)

Theplasticity
coefficients
reduce
to

(D2)
A15=~3=0
I
J

A3 = E@ (D3)

Consemntly, equations
theequilibrium (eqs.(B5)to (B7)
) reduce
to
thefOllowing
e~ressions:

a% ,J?3R:?32+(2+A3)a% , aW
-— o (D4)
ax2 4 Rm+2R ax=

(D5)

a%
‘W RhxaO
.0

(D6)

. ..— ..-— .—— —. ..— —— ..— — .— .—— —- .


30 NACATN 3726

Bymultiplying
egpation
(D6)by R2/B andletting

(D7)

Equation
(I%) canbe reduced
tothemoreconvenient
form:

[ 1‘:*+~’w+-a=O
‘R4*+P+A$R2&’* “8)
Following
themethod
ofsolution
fortorsional
elastic
buckling
of
a cylinder
asgivenbyTimoshenlso”
(ref.1),thefollowing
relations
are
usedforthedisplacements:

1“
I

u=~cos*-n(3
()

v= v..cos* - ne (D9)
()

~=~ms<~-~e) ,

Theanglewhichthehelical
buckle
makeswiththeoriginal
generator
of
thecylinder
isgivenby

tanu = I/n (D1O)

Sincethecylinderisassumed
tobe long,theboundary
conditions
atthe
endsarerelatively_ortant and,therefore,
equations
(II9
) canbe
usedalthoughtheydonotsatisfy
theusualboundary
conditions
ofsimple
supportor clamping.

. —
NACATN 3726 31

Thesubstitution
oftheappropriate
derivatives
ofeq~tions(~)
intoeqpations
(I$k),
(D5),
and(D8)results
inthefollowing
matrix
equation:

()2+A3
—All
4.
-n Tm = o (D1O)

l+2@tl+fh4+
-[
-n 1 %lTl

A nontrivial
solution
requires
thedeterminant
tovanish
independently.
Uponexpanding
thedeterminant,
thefollowing
result
isobtained:

As intheelastic
case,
verysmallvalues
of A yieldthesmallest
valueof q).Furthm&e,by assumingj3tobe smaJl,
equation
(Dll)
simplifies
to

(D12)
anz
.

Notethat theterm Az CSJICelS (D12)


outofeqyation underthese
assumptions
and.,ther%ore, for q) isindependent
thesolution of
valueof A3 givenby eWtion (D3)
.
Fora longcylinder
a minimum for n = 2
Valueofq isobtained
whichcorresponds
tothefamil.iar
two-lobe
buckling.
Therefore,
eqpa-
tion(D12)b=comes
~ = (3/4)A4
+ 256fI (D13)
64A

.... ... .. . . .—. — — .—. . —.. - .— — — .- —— -- —--— -


“32 NACATN 3726

Bydifferentiating
withrespect
to 1, . ‘

(D14)

Uponsubstituting
e~tion (D14)
intoequation
(D13)
andSh?lp~,

(D15)

Byuseof equations
(D7)

4(3)
=cr=-
9 -lhs(tm
312

= 0.338ES
(t/R)3/2 (DZ6)

Thecorresponding
elastic canbe ObtlXhll?d
solution by substitute@
1- ve2 forthecoefficient
3/4 inegyations(D13)
and(D7):

= -41-ve2)-3’4E@fi)3/
“ (D17)

~= [2(1-ve2pt/R~’2 (D18)

— -—.. . . . — .—
iQ
NACATN 3726 33


By identifying
thecoefficient
4/3 whichappears
inequations
(D13)
and(D7)&S the limit- valueof’ (. - #
with , = 1/2,it is
thenpossibleto conibine (D16)
equation andequat
with(D17) ion(D14)
with(D18)
:

= 0.272(1
‘cl” - (D19)

= -,$’’%J-’2
~(1 (D20)

(D19)
Egyations and(D20)
areexactfortheelastic
andplastic
ranges
andconstitute
excellent
approximations
intheinelastic
range.

-- ... ...-— — —.-— .—. . ,_ . -- —. .-—


34 NACATN 3726

RmERENcEs

1. mmoshenko,s.: ofElastic
Theory stability.
Firsted.,McGraw-Hill.
BookCO., ~C .,1936.
2. Bijlaard,
P.P.: Theory
andTkstsonthePlastic Stability
of
PlatesandShells.Jour.Aero.Sci.,vol.16,no.9,Sept.
1949,Pp 529-541.

3.~h, A.A.: Thellasto-P@stic


Stability
ofPlates.NACA
‘IMU88,1947.
4.Handelman,
G.H.,andPrsger,
W.: Plastic
Stability
ofa Rectangular
PlateUnderEdgeThrusts.
NACARep.946,“1949.(Supersedes
NACA
TN 1530.
)
5.Stowdl,Elbridge
Z.: A UtiiedTheory
ofPlastic
Buckling
ofColmns
andPlates.NACARep.898,1948.(Supersedes
NACATN1556.
)
6.StuwelL,
E.lbridge
Z.: critical sheer stress
ofanInfinitely
Long
PlateinthePlastic
Region.NACATN1681,1948.
7*Gerard,
George:SecantModulus
Method
forDetermining
PlateInsta-
bility
AbovetheProportional Aero.Sci.,vol.13,
Limit.Jour.
no.1,Jan.1946,PP.38-44,
48.
8.Stowell,
E.lbridge
Z.: Compressive
Strength
ofFlanges.NACA
Rep. lo2g,1951.(Supersedes
NACATN 2020.
)
9.Pride> Richard
A.>andHeimerl,
George
J.: Plastic
Buckling
of
SimplySu~orted Plates.NAM TN 1817,
Compressed 1949.
10.Peters,
Roger
W.: Buckling
ofLongSquare
TllbeS
inCombined
Com-
pression
andTorsionandComparison
WithFlat-Plate
Buckling
Theories.
NACATN 3184,1954.
U. Gerard,
George:Critical
ShearStress
ofPlates
AbovethePropor-
tional
Limit.Jour.Appl.Mech.,
vul.15,no.1,Mar.1948,
m= 7-E.
12.Bijlaard,
P.P.: OnthePlastic
Stability
ofThinPlates
andShills.
Verb.~ninkli
jkeNederl.andsche
Akademie
vanWetenschappen,
vol.L,
no.7,Sept.1947,pp.765-775.
13.Donnell,
L.H.: Stability
ofThin-Walled
!lkibes
UhderTorsion.
NACA
Rep.479,1933.

—.
NACATN 3726 35

14.Osgood, R.: TheCrinkling Strength


Willlam andtheBending
Strength
ofRoundAircraft . NACA-P. 632,1938.
Tubing
15.Moore,
R.L., andHolt
, ~S~ : BesmandTorsion
!lksts
ofAluminum-
WOY 6N-T ~bhg. NACATN867,1942.
16.Anon.
: Strength
ofMetalAircraft JXNC-5,
Elements. Munitions
Board
Aircraft
Comittee,Revised
ed.,June1951.
.
17.Gersrd,
George, Sorrel:A StudyofPoisson’s
andWildhorn, Ratio
intheYieldRegion.NACA.TN2561,1952.
18.Batdorf,
S.B.: A Simplified
Method
ofElastic-Stability
AnsJysis
forThinCylindrical
Shells.I - Donnell’s
Equation.
NACATN 1341,
1947.
19.Stang,
Ambrose
H.,Ram..rg,
Walter,
andBack,Goldie:Torsion
Tests
ofTubes.NACARep.601,1937.
20.Moore,
R.L.,andPaul,D.A.: TorsionsJ.
stability
ofAluminum
Alloy
*SlikSSTubing.NACATN 696,1939.
a. Tsien,
Hsue-Shen:
A Theory
fortbeBuckling
of ThinShells.Jour.
Aero.Sci.,vol.9,no.10,Aug.1942,pp.373-384.
22.Dcmndl,L.H.,and.
Wan,C.C.: Effect
ofImperfections
onBuckldng
ofThinCylinders
andColmnsUnderAxialCompression.
Jour.Appl.
kch.,vol.17,no.1,M. 195o,pp.73-83.
.

. ——..—-.
36 NACATN3726

-1
TESTDATAONCOMPRESSIVE
STRENGTH
OFTUBES
(a) 2017-T4
aluminum-alloy
dataofreference
14

djt Csf, ksi

98.k 41.8 0.330


97.9 41.9 .327
79.2 45.2 .288
78.4 45.6 .287
78.1 43.6 .273
63.7 47.0 .241
63.7 47.4 .242
62.2 46.3 .230
61.5 46.0 .226
61.5 .228
61.4 % .226

(b) 6061-T6
aluminum-alloy 15
dataofreference

I
1.32 0.o16 40.8 0.281
.023 39.0 .187
.033 41.9 .140
.066 46.3 .o~
.132 52.7 .044
2.00 0.025 38.6 o:q7
.033 40.8
.o~ 43.6 .146
.100 43.6 .o~
.200 53*3 .044

--- —. —
NACATN 3726 37

-1 TABLE2
TESTDATAONTORSIONAL
STRENGTH
OFIQNGTUBESF!RCM
~ 19
,

(a) 2017-T4
.sWminurn-alloy
tubes

Z/d */a ‘rf,


ksi ~s
19.9 0.02235 23.1 0.792
19.9 .02516 23.4 .698
60.0 .02848 24.2 .580
19.9 .03231 24.9 .491
59.8 .03242 26.0 .513
39*9 .02195 22.8 .830
13.3 .02903 25.1 .596
.
39*9 .02896 24.1 .572
13.3 .03273 26.4 .512
.
4Q.O .03312 26.1 .506
~.o .03350 25.7 .491

(b)Chrme-molybdenm
tubs

Z/d t/d hi Tf,ksi


‘yield~ qs

25.3 0@M& 48,6 50.6 0.262


25.3 47.9 50.4 .261
7999 .04020 49.5 51.1 .264
~.1 .02330 47.2 43.8 .530
1.2.6.O* 49.8 47:4 .309

——. -.— ——.— — - -———. —.—. .— ———. — -—— —.------


38 NACATN 3726

TABLE3

TESTDATAONTORSIOIiAL
STRENGTH
OFTUBES
(a) 60~-T6aluminum-alloy
tubesofmoderate
length
fromreference
~
a/t
= 104 d/t= 139
l/d Tf, ksi
Tf,ksi rf,ksi %

1 2L.6 o.41.218.9 0.525 18.4 0.728


1 .418 19.5 .542 18.4 .728
2 -%: .580 18.8 .740 17.0 .950
21.9 .594 19.7 .765 16.5 .923
: 2L.2 .810 17.7 .984 12.6 1.00
4 ZL.3 .814 18.6 1.03 12.4 .982
20.1 1.09 13.9 1.09 9.2 1.035
: 19.2 1.04 14.0 1.10 1.035
16 13.5 1.03 9.8 log ;:; 1.02
16 12.5 ●957 9.6 1.08 . 1.02
(b) 6061-T6
alumimm-sdhy
t~s frcan
reference
15

Iength d/t 2/d -cf,


ksi
A -
M 80.6 5.7 19.1 0.816
M 80.6 17.0 10.4 .813
L 27.6 9.3 .795
M 2:2 U..5 15.4 ~.987
M 58.8 5.7 18.6 .543
L 58.8 17.0 17.4 9927’
L 58.8 27.6 15.2 .810
M 39.4 5*7 22.7 .403
L 39.4 17.0 22.3 .648
L 39.4 27.6 22.0 .640
M 60.6 II*5 20.2 .928 .

%, mcd.crate
length;
L,long.
.

—. ..
,

FQwe stress-strain
Material Loading n data
%llEU’kB

2017-T4 Ccmpremion 1 Eq. ‘(4) Ccaupremion Excellentagreement;


limitedteat data
@h-T6 Ccmlpremion 1 Eq. (4), a= Fair limited
~-ntj
w
teat data

201T-Th Torsion 2 Eq.(17) Shear Excellantagreement;


wide range of test data
cllrme- Torsion 2 Eq. (17) 8hear Theory&L@; limitedteat
molybdenum data

605L-T6 Torsion 3 Eq. (10) TenOion Fair agreemnt; moderate


lengthtubes
6061-T6 Torsion 3 Eq. (M)
a’% Good agreemntj moderate
and long tubes

%eneile and ccmpressiw yield stressgiven. Stress-strain


data obtainedfrom
ref. 16 for threeyte~ vaLues.
.
1

%
I
I

I
o
I
30 40 50
COM%ES5YVE S7i%59S, ksi

WW’e 1.- Pks*iciW-reductim factorsfor cqpressivabucklingof


. aJumimm-elL2ytubes.
.

4
t
I

w a w w au t
WEAR S7R!?SS, ksi “

Figure 2.- Plasticity-reduction


factorsfor ijomional.
buoklingof long
tubes.
I

I
6

- S7RESS, ksi !3
H.@H 3.-Pksticitiy-reductj.on
factom for tomiond h@cM.ng of tubeg.
3
3/4
11.
= (1- V=P/l- #) (ISJ$) .

, , I