[wh
!1
Jou[nal
ofStrucLural
andConstruction
EngineerLng
UDC :624. o74. 43:624. o4 :s3g. 3s4(Transactlons
ofAIj)
No.
364,
June,
lgB6
Htseca\2msinLxas
cvasst
ag
364
e-vatu
61
ij
6
fi
THEORETICAL
STUDIES
OF
THE
BUCKLING
AND
NONLINEAR
DEFLECTION
BEHAVIOR
OF
CLAMPED
PARTIAL
CIRCULAR
CYLINDRICAL
SHELLS
UNDER
EXTERNAL
PRESSURE
by
KAZUO
UCHIYAMA'
and
SEISHI
YAMADA'",
Members
of
A,
I,
J.
1.
Introduction
Many
circular
cylindrical
roofs
have
collapsed
during
the
snow
in
Japan,
and most ofthem
seemto
have
buckled
generally]).
Thus
the
buckling
criteria
of
partial
circular
cylindrical shellshave
requiredfor
the
designs.
In
this
paper,
to
make clearthe
basic
general
buckling
behaviors
of cylindrical shell roofs, we shalltake
the
case ofpartial
cylindrical shellssubmittecl
to
the
action of uniform normalpressure.
For
complete
cylindrical shells under externalpressure,
the
bilckling
behavior
wasfully
exploredtoward
the
beginning
ofthe
1970
s2).On
the
otherhand,
for
the
partial
ones,it
does
not seemto
have
been
fully
looked
into
theoretically
in
particular.
The
strongest reason mightbe
that
the
partial
cylindrical shellshave
not onlythe
curved edgesbut
alsothe
straight edges.Uchiyama3)
experimentally clarifiedthe
buckling
behavior
and expressed an approximate method of estimation ofthe
buckling
stressfor
the
design,
Singer
etal`)
obtained
the
buckling
loads
of simple-supportedpartial
cylindricalshells
based
onDonnell's
linear
equatiens neglectedthe
prebuckling
deflection
in
a wide range ofthe
geometrical
parameters.
Yokoo
andNakamuraSj
theoretically
obtained some examples ofthe
symmetrical noniineardeflection
behavior
of
clamped
partial
cylindrical
shells
usingthe
finite
differential
method,And
theyM
carried out experimentson
the
pre-and
post-buckling
behavior
for
the
shells withfree
edges ancl represented an empiricalformula
for
the
buckling
loads.
Kokawa')
theoretically
analyzedthe
symmetrical and asymmetricalgeneral
buckling
behavioi
of reticulated cylindricalroofs
with
arbitrary
generator
edges.
Suzuki
and
OgawaS)
a]so analyzedthat
behavier
using
a simplified method whichdealt
bending
rigidity as rotational spring stiffnessof
memberjoints.
The
authors")・'")
showed
the
experimentaldata
ofthe
overalldistribution
ofinitial
geometrical
imperfections
anddeflections
for
many accurate specimens of clampedpartial
cylindrical shells, anddiscussed
the
buckling
behavior,
The
object ofthis
paper
is
to
clarifythe
buckling
loads
andthe
nonlineardeflection
behavior
of clampedpartial
cylindrical shells under externalpressure
in
a wide range ofthe
Batdorf
parameter
Z
andof
the
aspect ratio a systematically.We
obtain
notonly
the
symmetrical
snapthrough
buckling
loads
but
alsottie
asymmetricalbifurcation
ones
applying
the
Galerkin
procedure
to
Donnell]s
nonlineareqliations.
The
po$sibility
of asymmetricatbuckling,
whichhave
notexperimentaly
been
established,is
indicated
for
some shells.The
effects ofthe
parameters
on
the
deflection
behavior
andthe
buckling
modes are also shown, anclthe
characteristics arein
good
agreement withthe
authors',
experimental
results.2.
Method
of
Analysis
Let
us
consider
athin-walled
paitial
circular cylindrical shell oflongitudinal
length
L,
wallthickness
h,
radiusR
and central angleip,
whichis
made ofhomogellerous,
isotropic
elastic
materiat withYoung's
modulesE
andPoisson's
ratio
v.The
middle surface ofthe
shellis
refeTredto
cyiindrical coordinates as shownin
Fig,
1,
Distances
'
PTofessor,
Department
ofArchitecture,
Faculty
ofEngineering,
Tohoku
Univeisity,
Dr.
Eng.
#Reseafch
Associate,
Department
ofATchitecture,
(Manuseript
[eceivedOctober
11,
1985)
Architectural Institute of Japan
NII-Electronic Library Service
Arohiteotural エnstitute of Japan
from
the
middle surface are measuredby
a coordinate z,
positive
inward
.
Displacemeot
componentsin
the
x,
Y
,
i
.
andz
directions
,
respectively,
aredeno1ed
by
u
,
v
,
andw
.
The
shellis
consideredto
be
subjected
to
uniform externalpressure
q
(
positive
inward
)
.
On
the
basis
ofthe
Donnell
nonlineaf
theory
ofthe
cylindricaL shell,
the
basic
equations aregiven
asfo
}lowsZ
川 1 : ∂nx
∂
nxy
十 ;0
∂x
∂
y
∂πエ ン∂πり ∂
x
+可
一
=
o
−
・(
房
∂2 + ∂∂y
z・)
2 ・+争
・導
…巉
晶
…
穿
・・一 ・Q
Fig
.
1
Sign
convention and shellgeometry
.
・
………・
・
……・
…・
・
………・
…
;………・
(
1
)
Eh3
whereP =
12
(
1
一
ノ}
In
equation(
1
)
,
n
エ,
ny
,
and
nxy
are
the
stress
components
per
unitlength,
and
τ
シ η η 1 ー レ0
−
2 ん yE
一
1
レ ー01
,]
籌
+弖
儲
)
2・
ll
一
詈
+去
(
迦
∂y
)
z ∂u ∂v
∂w
∂w
∂y
+『
認
+蕊
一
冴
The
following
nondimensionalforms
areintroduced
ξ
= ユ:/
」し,
η
=
y
/
R
φ,
σ=
Ru
/
L
〜L
,
V
二v
/
φh ,
w
=
ω/
ん,
Z =
s
/
i
’
=」
L2
/
R
ん(
Batdorf
parameter12
〕)
α;
L
/
R
φ
(
aspect ratio)
Q
=
肌 ’q
/
π2D(
10ad
parameter
)
・
・
一・
・
一一・
・
・
・
・
・
・
・
・
・
・
・
・
…噛
曁
・
一
…
一
・
一
・
・
・
・
・
・
・
・
・
・
・
・
・
・
・
・
・
・
・
・
・
・
…
(
2
)
・
……・
…………・
………・
……・
…・
・
…・
・
…
(
3
)
The
nondimensio 耳albasic
equations areder
玉vedby
出eintrQduction
ofequatlons
(
2
)
and(
3
)
into
equation(1
)
,
asfollows
: ・x−
・
{
離
・・
1
−
・
喋
・・
1
+)
・
諾
一
・・剽
・
誓
{
礎
・面
・劉
・・
1
+・・
錯
諾
=OLy
−
・
{
標
・(
≒
の
ぢ
i
¥
+〔
1
+の
冪
昜
一2
誓
}
・
鰐
{
・襟
・・
1
−
・
訓
珊
・・
鐸
蠶
=
0
し
一誓
・…
β
・(
∂u
∂v
P∂ξ
+
∂η
w
)
一
去(
妾
・・券
)
2 ・ ・ ・β
[
離
・・α誓
笋
}
嬰
・・
1
−
・
・舞
器
・者
(
∂w
∂ξ
)
!一
・畔
+
{
・肇
+考劉
哥
刊一
・)言
嵩
弩
+氛
咢
)
w
鬱
]
・ ・[
離
・・聯
}
闇
+ ・・
1
−
・
・鑞
讐 哥
+・
(
Yge
+・穿
}
(
Th
)
2]
=0
…・
…・
…………・
(4
>
一 81 一
N工 工一
Eleotronio LibraryU=Z
£
AtJU}j=Z
£
AwsmminesinnJrrn
tj iJV=2]ZB"V}J=ZZBijsinmEngsinnjnn
IJ iJW=ZZC"VXJ=
£
£
Cvsinm`nesinnjnosinnesinnn
tJ IJwhere
mi
andnJ
arethe
longitudina}
wave number andthe
application of
Galerkin's
methodto
equation(
4
},
the
Galerkin
where a=a2 and
fi
=zlVi=-IJTThe
independent
paTameters
for
the
shellgeometry
in
equation(4)
arethe
Batdorf
parameter
Z
and
the
aspect ratio ainstead
ofthe
four
geometrical
vaLuesR,
h,
L
and
ip
in
equation(1),
The
bundary
conditions
are
aw
w==
ae=u=V==o,
at
e=O,i
l""
-"".
"""-.
.
..
..
.
(s)
W=aevro=U==V=o,
atn=o,1
l
The
displacement
functions
U,
V,
and
W
are
assumed
in
the
form
of
a
linear
combination
offunctions
multipliedby
unknown coefficientsA",
B:J,
andC"
which are not onlyindependent
of each otherbut
also satisfy equation(
s
},
andthey
are written as---・---・--・---・---・-(6)
circumferential wave number, respeetively.
By
system of equations
is
given
by
.('
f'
L.
u},dpto==
.ti
-('
L.
v,,dcao=.ti
Jf'
L.
Viljdedo=o
'''''''''''''''''''''''''''''''''''''''''''''''''''''''''(
7
)
By
substituting equation(
6
)
into
equation
(
7
),
the
nonlineai algebraic equations,involving
the
load
parameter
Q,
for
the
unknown coefficientsAiJ,
Bw,
andCw
are obtainedia),The
numerical selutions ofthe
algebraic equationsare
procured
step-by-stepby
both
alimear
incremental
methodand
the
Newton-Raphson
method,The
bifurcation
point
for
an asymmetricaldeflection
modeis
obtained asthe
point
at whichthe
determinant
of coefficient matrixin
the
linear
simultaneous algebraic equationsfor
the
supposed asymmetrical modeis
zero,or
regarded as zeroi`).The
computation
wasperformed
on adigital
computer(ACOS
99-NEAC
1000)
atTohoku
University
in
double
precision.
3.
Numerical
Results
First,
we checkedour
numerical results with somepTevious
theoretical
results.In
ref.[15]
,the
comparison withthe
results ofYokoo
etal.obtained
by
usingthe
finite
difference
method5i
wasdescribed,
FiguTe
2
also showsthe
good
agreement withCrisfield's
resultsobtained
by
usingthe
finite
element
method]6) asto
nonlineatiarge
deflection
behavior.
In
the
computation,parameters
mt,and
njin
equation
(6)
aregiven
asfollows:
for
symmetrical mode;15
fn,=2,2,2,2,2,4,4,4,4,6,6,6,8,8,10
Ul
nj==1,3,5,7,9,1,3,5,7,1,3,5,1,3,1
mt=1,3,5,7,9,1,3,5,7,1,3,5,1,3,1
vl
nJ==2,2,2,2,2,4,4,4,4,6,6,6,8,8,10
m,=1,1,1,1,1,3,3,3,3,5,5,5,7,7,g
wl
nJ=1,3,5,7.9,1,3,5,7,1,3,s,1,3,](Ns=3
×15)
for
asymmetrical mode;m,=2,2,2,2,4,4,6,8
vl
nJ
==2,
4,
6,
8,
n,2 n, n, nm,=1,],1,1,3,3.5,5
vl
nJ=1,
3,
5,
7,
n-1,
n+1,n-1,
n+1
mt=1,1,1,L3,3,5,7
Wl
nj=2,4,6,8,n,2n,n,n
(IVb=3
×8)
-82-Q
10 5 o PRESENTCRISFIELD .(FEM)
a=1.oZ=30.5
O
l
2 5Wc
Fig.2
Comparison
ofload
Q
versus centerdeflection
VVI
curves wlthCrisfield's
results.Architectural Institute of Japan
NII-Electronic Library Service
ArchitecturalInstitute of Japan
CENTER
ARCH
(g=O,5)
a=1.0Z=200
-1.0
AS'ttttts11tL-O.5
,ln'V'o
O.2'f,'fO.4ttl
KO,6no.s'''J-t-'xLO /-sKhLX
'
ab
11S XxxK'..i
NSVxN't-i--t
O.5,x-LX1Xdzlij-f/c
ls,it' -'1,o
'W,h..et--'''(a)
Fig.3
Example
ofpre-and
post-buckling
deflection
behavior
for
the(
a)
Circumlerential
distJibutions
ofdeflectien
W
along theCb)
Load
Q
versus centerdeflection
iS(L
curves.Sym.
]o
'-.
c
dAsym.
e
20
Q
10 Ocrb
aWc
-O.5
O O.5{b)
shell
having
a=l.O andZ=
200.
center arch at variousload
levels.
where
N.
andAl.
arethe
total
numbers ofterms
for
symmetrical and asymmetricaldisplacement
modes, respectively, and nis
an evennumber
of
circumferential
half-waves
of asupposed asymmetricaldeflection
mode,On
the
basis
ofthe
previous
experimental results9)' iO)'i7), asymmetricaldeflection
modesin
the
longitudinal
direction
ares]ighted
in
this
study.The
difference
of syrnmetricaL snap-throughbuckling
loads
calculatedby
two
adoptions which wereN.=3
×15
andtV.=3
×21
was aboutO.5
%
in
the
case ofth,e
shellhaving
Z==300
anda=1.o.
Poisson's'ratio
y wastaken
asO.3
in
this
study.
Next,
we shall elucidatethe
variation ofthe
distribution
ofpre-and
post-buckling
deflection
with appliedpressure
Q
for
clamped
partial
cylindrical shells.Because
of space consideration,the
resultsforonly
the
shellhaving
Z=20e
and a=1.0 are shownin
Fig.3
(a),
In
this
figure,
solid andclotted
lines
correspondto
the
circumferentialdistributions
ofdeflection
W
atpre-and
post-buckling
equilibriumpoints
a,b,
c,d,
and ein
Fig,3{b).
The
nonlineardeflection
behavior
is
remarkably recognizable not only afterbtickling
but
aLsobefore
buckling,
andit
is
iound
that
the
efiect oiprebuckling
nonlinearity can notbe
omittedin
this
Problem.
4.
Effects
of
the
Geometrical
Parameter
Z
.
The
geometrical
parameter
Z
is
widelyknown
asthe
Batdorf
parameter
for
a complete circular cylindrical shell`Z) andparticipates
the
aspec,t ratio ain
the
independent
geometrical
parameters
for
apartial
circular cylindrical shell as shownin
eqdation(
4
).
Calculations
weie carried outfor
Z$loOO.
Table
l
showsthe
comparisonbetween
the
present
buckling
loads
Q,.--RL2q,.ln:D
andYarnaki's
results of complete cylindrical shells subjectedto
lateral
pressurei8].
The
buckling
behavior
ofpartial
cylindrical shells seemsto
differ
from
that
ofthe
complete ones without straight edges.However,
it
is
consideredthat
the
dominant
component of
buckling
stressis
the
circurnferentialTable1
Buck]ing
loads
for
partiar
circular cylindricat shells. membrane componentfor
both
the
partial
ones andthe
cemplete ones, so
the
similarityis
anticipated whena
issmall
enough.
The
buckling
loads
Q,,
for
a=1.0
anda
=2,
O
lie
16-29
%
and
62-91
%
higher
than
those
ofthe
complete
shells,
respectively.The
variation ofQ,.
withZ
is
also shownin
Fig.4,
where n,.denetes
the
buckling
mode
number whichis
defined
asthe
circu'mferentialhalf-wave
nUmber ofthe
dominant
component ofthe
symmetricalor
asymmetric-aldeflectien
atthe
buckling
point
in
the
case ofthe
syrnmetrica] or asymmetrical
buckling,
respectively.The.
buckling
load
Q..
andthe
buckling
rT{ode numbern..
become
large
asZ
increases.
In
the
case ofa=1.
o
PRESF.Nl'RESULTS
.1a・=1,Oa=2
40----F----5013.72J-L
7017.oq2.a'1oo20.30.29'
15023.B736' f200
300
400
500
700100026.9232,3835.7741,
l547. 2756,27E4L49.53.S7.69,78.
.o
134240462362636062
UAMAKI,$QoCCOMPLETE)
10. 6311.63'13,44
, 15.77.1
9,O52L
8626, 6330, 6834,2B40.584a, 56 -83-NII-Electronic Mbrary1OOQcr
50
20
10
20Fig,4
o
Fig.7
Effects
O.150of
the
O.2100
geometrical
o.se
o.4200
parameter
Z
en os 200025
400 O.]O 700 O.50Longitudinal
thebuckling
eedist[ibutionspoints
in
theo
O.5W
LO
n=1.0Q=Ocrofdeflection
case of a=1W at.o60
a=2.0
oJa=1.o
so
s
500
1000
Zthe
bllckling
loads
Q.r.
40
50
20
10
oo
Fig,5Loadcurves
O,1
O.2
VOI.UME
CHANGE
VeQ
versus voturne changein
the case of a=1.0O.3
va
-O,5
O
O.5
LO
1.5
2.0
-to
7o,s
o
O.5
1,O1.5
2,O
CENTRAL
DEFLECTION We
(a)
CENTER
DEFLECT[ON
W,{b)
Fig.6
(a,
b)
Load
Q
versus centerdeflection
WZ
curvesin
the cases of a=1.0 and a==Z.O,respectively,
・
the
shellsfor
2<4s
do
notbuckle,
anclthe
non-buckling range expandsinto
Z<60
in
the
case ofa=2.
o.
The
discrepancy
between
the
authors'experiment9L'O}
denoted
by
the
triangle
symbols andthe
present
theory
mightbe
caused
by
the
initial
imperfections'9).
Figures
5,
6,
7
and8
arethe
variation ofthe
deflection
behavior
withZ.
In
Fig.
s,
the
initial
stepe ofthe
load
Q
-Architectural Institute of Japan
NII-Electronic Library Service
ArchitecturalInstitute of Japan oO.1O.2
nO.5
O.4O.51Z=700
...7..
a=1.o.-.'
Z=200
s''-LY.--Jrt!''Z=70
LZ=100
Q=Qcrl
'
'
CENTERARCHCC=O.5)z=4oe
(a)
HO.5
o O.5w1,Oo
.5
5o
5Fig.
8
(
a,b
)
Variatiens
ofcircumferentialdistributions
ofdeflection
along the center aTch withZ
in
the cases of a=1.
0
and a=2.
0,
respectiveLy,versus
the
volume changeV}
curveincreases
asZ
increases,
where,from
equations
(3)
and
<6),
V}=
v.lhLRip=Jgr]Jl']
wa6du=
C.!4
-・・---・---・-・-・・--:----・-・・・・-・・-・----・i・-・・-・----・-・(
8
)
'
The
circles andthe
triangles
representthe
syinmetricalbuckling
points
andthe
asymmetricalbifurcation
points,
[espectively.In
this
figure,
the
even nombers accompaniedby
the
triangles
arethe
bifurcation
mode numbers.When
a shellhas
an asymmetricalbifurcation
point
before
syinmetrical snapthroughbuckling,
it
is
commonfor
the
load
after
buckling
to
decrease
in
the
same manner asthe
curve
cein
Fig,
3(
b
),
However,
in
the
case
ofthe
shell withZ
=::40 anda
=:1,
O
whichdoes
hot
have
a,symmetricalbuckling
point,
the
secondary equilibriumpath
with
n;2
has
apositive
sLope,Therefore
we understoodthat
it
does
notbuckle.
According
to
the
results ofthe
calculation, asymmetricalbuckling
(n..=even)
occursfor
Z==
(80-400)
andZ=
(100-250,
700-1000>
in
the
cases ofa=1.0
anda=2.0,
respectively.And
it
shouldbe
more emphasizedthat
the
'
asymmetrical
buckling
points
on
the
load-deflection
curves are very closeto
the
symmetricalbuckling
points
in
all'
cases as shown
in
Figs.5
and6,
'
Figure
7
showsthe
distributions
ofW
in
the
longitudinal
direction
through
the
maximumpoints
atthe
buckling
loads
Q...
For
all
valuesof
the
geometrical
parameters,
the
dominant
component ofthe
longitudinaL
distributions
ofW
was onehalf
wave modein
this
study.It
is
in
good
agreement withthe
previous
experimenta] results9)・iO)・t7).On
the
otherhand,
the
distribution
ofVV
in
the
circumferentialdirection
atQ=Q..
becomes
of ahigh
order asZ
increases
as shown'in
Fig.8.
Figure9
indicates
that
the
distribution
of asymmetricalbuckling
deflection
in
the
¢
ircumferential
clirection
becomes,
of
a
high
order
as
Z
keeps
increasign.
5.
Effects
efthe
Geometrical
Parameter
aIn
this
chapter we shalldiscuss
the
'effect$
ofthe
geemetficat
parameter
a,
whichis
defined
asthe
aspect [atio/.'Lxx
a=2.0
'li
ttO.i
O,1
SLo,5x o.s T]s
O.4[
-o
,
O.5o
5.0
N
L
i
N s Z=200 ,
Z=700
'
sN
t
Ii L i ! r CENTERARcH
X
1
cf=o.s)
N...-../
ASYM.BUCKLINGMODESFig.9
Examples
of asymmetricatmodes
for
a=2.07
/Jlr',rO.5
w
1.0buck]ing
20
oox L uo10
2.0
4.0
O.5
1,O
a
Fig,
le
Effects
ofthe
geometrical
parameter
a onthe
buckling
Lead
Q,.
in
the case ofZ=zoo.
--Table2Comparison
between
the authors' experimentai results andthe
present
theoreticai fesultsfor
thebuckling
loads
and medes.EXPERIMENT THEORYQcr aoqcrQcrncrQerncvEXXTH dee・"Pg%
O.72so1.1729.0636.7579
O.82701,2731.5537.25S5 O.95601,2731.5539.45so L15501.2631.3442.2474 1,43401,6340.3347.6485 i,91301.9949.3254.63goZ=477.
60Q
50
40
h=O,3mm,
R=L=1Z=200
50mm,
Et5.
6GPe,
a=5.0 a=2.0fi
2
ASYM,BfFURCATIONS\M.
BUCKLING
a±
1.5
a=Lo4
atO.8'
: a=O.6' '
',/-KS"t.compLete
stselL
under LateraL PressureNG
TION
'50
2o
10
o'j'''''-O.1
O.2
O.3
VOLUME CHANGE V,
versus yolume change
VL
curvesin
ofz=zoo.
-1,O
-O,5
O
O.5
1.01.5
CENTER
DEFLECTiON
W,Fig.11
Loacl
Q
versus centerdef]ection
VVZ
curvesin
the
case ofZ=200,
-O.5
Z=200
a[o.s
o
o.1
e.2
o.s
n
o.4 osQ=Qer
CENTER ARCH
{C=O.5)
Fig.13
Variatien
of circumferentialdeflection
alongthe
the
case ofZ=ZOO.
o O.6
O.5
1.5w
].o1.0
distributions
centeT arch with ao s
o
ofino
Fig.12Load
Qthe
caseASYM.
BUCKLiNG
MODES
Hg.14
Examples
of asyrnmetricalfor
z=zoo.
O.5
w
.o
buckling
modesLIR
¢
in
equation{
3
),
onthe
buckling
load
andthe
deflection
behavior
onthe
basis
ofthe
resultsfor
Z=2oo
andZ
=477.In
this
study, a was variedfrom
O.6
to
4.0.
Figure
lo
shows
the
variation ofthe
buckling
load
witha.
The
ordinateis
the
ratioof
the
buckling
load
for
partial
cylindrical shellsQ..
to
those
for
complete cylindrical shellsQ,
obtainedby
Yamakit8).
The
buckting
load
Q..
increases
with anincrease
in
a, andthat
of
a=4.0
is
aboutthree
times
aslaTge
asQo.
As
akeeps
decreasing,
Q..
decreases
like
approachesQ,,
andthe
buckling
mode shows more and more wavesin
the
circumferentialdirection.
-Architectural Institute of Japan
NII-Electronic Library Service
ArchitecturalInstitute of Japan
In
Table
2,
the
present
results are compared withthe
previous
experimental results9)in
the
case ofZ;477.
The
present
buckling
loads
are10-26
%
higher
than
the
experirnental ones,HosyeveT,
the
agreemen,tbetween
the
theory
andthe
experiment asto
the
effects
of
theparmamete[
a onthe
buckling
loads
and mpdesis
obtained.The
load-deflection
behavior
is
shown
in
Fig.11.
In
the
initial
loading
stagethe
nonlinearities are not very recognizable,however,
these
are remarkably recognizablein
the
vicinity ofthe
buckling
load..
The
initial
slopes ofQ-
VL
curve
in
Fig.
12
have
a
constant
gradient
of
294fl
whenZ
=200,and
this
valueis
larger
than
the
gradient
26011
{denoted
by
abioken
line)
for
the
complete cylindrical shell.The
difference
is
causedby
the
clampedlongitudinal
straight edges ofthe
partiaa
cylindrical,shells.
The
results ofFigs,
5
and12
indic4te
that
the
initial
slope ofQ-
V,
curvedepends
not onthe
parameter
a
but
onthe
parameter
z.
Figure
13
showsthat
the
symmetricaldeflection
modein
the
circumferentialdirection
has
more and more waves as akeeps
decreasing.
The
circumferentialdistribution
of asymmetricalbuckling
deflection
becomes
ef ahigh
order asa
keeps
decreasing
as shgwnin
Fig.
I4.
6.
Conclusiens
,
'
ApplyLng
the
Ggterkin
procedure
to
Donnelt]s
nonlinear equations,the
buckLing
and nonlineardgflection
behavior
ofpartial
circular cylindrical shells under externalpressuTe
is
the6retically
clarified.We
treat
the
fully
clampedbeundary
conditions so asto
compare our results withthe
previousi
experimental results,Main
contents aresummarized
as
follows
:
(
1
)
The
load-deflection
relationshipsin
the
initial
loading
stage are almostlinear.
'However,
the
nonline'aritiesare remarkably recognizable
in
the
vicinity ofthe
buckling
point.
(
2
)
The
nondimensionalbuckling
load
Q..
increases
with an{ncrease
in
the
Batdorf
parmeter
g,
and with anincrease
in
the
aspect iatio a.This
is
in
good
agreement withthe
previous
experimental results.(3)
It
is
possible
to
buckle
asyrnmetrically as well a,s symmetrically.However,
whenthe
shellhas
anasymmetrical
bifurcation
point
before
symmetrical snap-throughbuckling,
the
bifurcation
point
is
veryclose
to
the
symmetricalbuckling
point.
'
'
(
4
)
The
initial
slopes ofthe
nondimensionaHoadQ
versusthe
nondimensional vDIume changeV,
curves
aTe
almost constant
irrespective
ofa
if
Z
is
fixed.
(
5
)
The
dominant
cornponent ofthe
longitudinal
distribution
ofdeflection
is
onehalf
wave modein
each case.The
circumferentialdistribution
ofbuckling
deflection
has
more and more waves with anincrease
in
Z
andwith a
decrease
in
a.
These
are alsoin
good
agreement withthe
previous
experimental results.This
study $howesthat,
by
usingthis
analytica] rnethod, reasonably accurate solutions canbe
systematically
ebtained
for
the
buckling
loads
andthe
deflectien
behayior
ofpartial
cylindrical shells,in
the
range of values ofZ
anda
analyzedin
this
paper.
The
present
parametical
resutts maybe
availablefor
the
design
ofgylindrical
shetl roofs.Reterences
・
1}
Uchiyarna,
K.,
et al:
Snow
DaTnage
toS{eel
Structures
in
Tohoku
andHokuTiku
District,
Dec.19BO-Feb,
]981,
Jollrfial
ofSociety
ofSteel
Construction
ofJapan,
Vel.17,
No.185,
pp.1-4g,
1981,
(in
Japanese}
2)
Yarnaki,
N.
:EIastic
Stabi]ity
ofCircular
Cylindrical
She[Is.
North-Holland
Series
in
App!ied
Mathematics
andMechanlcs.
Vol.27,
North-Holland.
Amsterdam,
1984.
3)
Uchiyarna,
k,
:
The
Buckling
ofCylimdricai
Shell
Roofs,
Laige
Span
SheLls,
Proceedings,
of theI.A.S.S,
Congress,
pp.743-758,
1968.
4)
Singer,
J.,
Meer,
A.
andBaruch,
M,
:
Buckling
ofCyLind[ical
Panels
underLateral
Pressure,
Aeronautical
Journal
ofRoyat
Aerenautical
Society,
Vol,73,
pp.169-172,
1969.
5)
Yokoo.
Y.,
Nakarnura,
T.
andKanoh,
A.
:Numerical
AnaLysis
en theSnap-Threugh
Buckling
ofClamped
Partial
Cylindrical
Shells,
Technical
Papers
ofAnnual
Meeting,
A.I.J.,
pp.1069-1068,
1971.
(in
Japanese)
6)
Yokeo,
Y.
,Nakamura,
T.
andMiyazuki,
J.
:
A
New
Method
ofExperimental
Buckting
ofCircular
Cylindrical
Roof
Shetl
Models
withFree
Edges
subjected toExternal
Pressure,
Int.
J,
Non-Linear
Mechanies,
Vol.6,
pp.633-647,
1971,
7)
Kokawa,
T.
:
Gene[al
Buckling
ofCyllndrical
Reticulated
Shell
Reofs
withArbitrary
Generator
Edges,
Part
1,
2
and3,
T[ans,
ofA.I.J.,
No.280,
pp.53-65,
1979;No.283,
pp,76-86,
]979;Ne,295,
pp.51-61,
1980.CinJapanese)
8)
Suzuki,
T.
andOgawa,
T,
:
Buckling
Analysis
ofReticulated
Cylindrica[
Shel[
Roofs,
Part1
and2,
Trans.
ofA.I.J.,
-87-9)
10)
]])
12)l3)
14)
l5)
16)
]7)
18}
19)
No.288,
pp.29-37,
1980:No.295,
pp.45-50,
1980,
(in
Japanese)
Uehiyama,
K.
andSuzuki,
A.
:
The
Buckling
oftheCLamped
Cylindrical
Sheii
Roofs
underExternal
Pressure,
Proceedings
ofthe
I.A.S.S.
Congress,
pp.133-147,
l979.
Uchiyama,
K.
Yamada,
S.
andMinami,
H.
:
Experimental
Studies
of theBuckling
efClamped
Cylindrical
Panels
underExtemaL
Pressure,
Technical
Papers
ofAnnuat
Meeting,
A.I.J.,
pp.1103-ll04,
1985.
(in
Japanese}
Donnell,
L.H.
:A
New
Theoiy
for
the
Buckling
efThin
Cylindrical
SheHs
unclerAxial
Compression
andBending,
A,S.
M.E,
Trans.,
Vol.
56,
pp,
795-806,
1934,
Batdorf,
S.B.
:
A
Simplified
Method
ofElastic
Stabllity
Analysis
for
Thin
Cylindrical
Shells,
NACA
Rep,879,
1947.
Yarnada,
S.
andYamada,
M.
:
Buckling
andPostbtckling
Behavior
ofHalf-Loaded
Shallow
Sphe[ical
Shetls,
Int.
J.
Non-Llnear
Mechanics,
Vol.20,
pp.239-248,
1985.
Uchiyama,
K.
andYamada,
M.
:
BuckLing
ofClamped
Impe[fect
Thin
ShaL]ow
Spherical
Sheils
underExternal
Pressure,
Partl
and2,
TechneL
Rep.
Toheku
University,
Vel.39,
pp.101-l32,
1974;
Vol,40,
pp.l-23,
l975.
Yamada,
S.
andUchiyama,
K.
:Theoretical
Studies
oftlte
Buckl{ng
ofClamped
CyLind[lca[
Panels
underExternal
P[essure,
Part
1,
Technical
Papers
ofAnnual
Meeting,
A.I.J.,
pp.ll05-ll06,
1985.
(in
Japanese)
Crisfield,
M.
A.
:
Solutien
Procedures
for
Non-Linear
StructuraL
Problems,
Recent
Advances
in
Non-Linear
Computational
Mechanics,
Pineridge
Press
Lirnited,
pp,2Z-23,
1982.
Yokoo.
Y.,
Nakarnu[a,
T.
andMatsuda,
A.
:
ExpeTirnental
lnvestigation
of theStabi[ity
ofClamped
Partial
Cifcular
Cylindrical
Shells
subjectedto
External
Pressure.
Recent
Researches
ofStructural
Mechanics,
Uno
Shoten,
Tokyo,
pp.333-343,
1968.
Yamaki,
N.
:
Influence
ofPrebucklingDeforrnations
onthe
Buckling
ofCircularCylindrica]
She]Ls
undefExternal
Pressure,
Rep.
Institure
ofHigh
Speed
Mechanics.
Tohoku
University,
Vol.21,
pp.81-104,
1970.
Yamada,
S,
,Uchiyama,
K.
andYamada,
M.
:
Experirnental
InyestLgatlon
of theBuckling
ofShal;ow
Spherical
Shells,
Int,
J.
Non-Llnear
Mechanics,
Vol,18,
pp.37-54,
1983,
-88-Architectural Institute of Japan
NII-Electronic Library Service
Arohiteotural エnstitute of Japan