XV393

if'
'0

P3W-ThS-6"6 (Rev. 3--58)

 DEPARTMENT OF THE NAVY
L_ DAVID TAYLOR MODEL BASIN
WASHINGTON 7. D.C. IN REPLY REFER TO

9110/Subs
5605
(727:JAN: 1kg)
Ser 7 -47 6
28 December 1961
From: Commanding Officer and Director, David Taylor Model Basin
To: :Chief, Bureau of Ships (442) (in duplicate)

Suhj: The effect of stiffener size on the buckling strength of

cylindrical pressure hulls;,forwarding of report on

Encl: (1) DATMOBAS Report 1600 entitled "Investigation on the

Influence of Stiffener Size on the Buckling Pressures
of Circular Cylindrical Shells under Hydrostatic
Pressure" 2 copies

1. As part of its structural research on the elastic- and

plastic-buckling collapse of ring-stiffened cylindrical pres-
sure hulls made of strain-hardening materials, the David
Taylor Model Basin has undertaken the development of rigorous
mathematical analyses and the verification of such theories
through a systematic series of structural model tests. One
such project concerns the investigation of stiffener size on
the asymmetric (lobar) buckling strength of cylindrical shell
structures. The results of a theoretical analysis and some
limited test data are presented in enclosure (1).

2. The theory presented in enclosure (1) is applicable to

shells made either of strain-hardening or elastic-perfectly
plastic materials. Methods are given for the determination
of collapse pressures for both elastic and plastic interbay
lobar buckling by iteration and numerical minimization. It
is shown both theoretically and experimentally that variations
in stiffener size can influence the buckling pressures. Test
data from high-strength steel and aluminum cylinders are pre-
sented which show agreement between theory and experiment to
within approximately 6 percent.

E.E. JOSON
By direction

II - '- I -
 9110/Subs
5605
(727:JAN: ikg)
Ser 7-476
28 December i961
Copy to:
BUSHIPS 320), with 1 copy of encl (1)
335) with 3 copies of encl (1)
341A , with 1 copy of encl (1)
342A), with 1 copy of encl (
420), with 1 copy of encl (1)
421, with 1 copy of encl 1I
423 , with 1 copy of enol 1
440 , with 1 copy of encl 1
443 , with 1 copy of encl 1
(525 , with 1 copy of encl 1)
CHONR (439),.,with 1 copy of encl (1)
CNO (Op 07TB), with 1 copy of encl (1)
CDR, USNOL, with i copy of encl (1)
CDR, USNRL (2027), with 1 copy of encl (1)
CDR, USNOTS, China Lake, with 1 copy of encl (1)
CO, USNUOS, Newport, with 1 copy of encl (1)
S CO, USNUSL, New London, with 1 copy of encl (1)

CDR, ASTIA, with 10 copies of encl (1)

NAVSHIPYD PTSMH, with 2 copies of encl (1)

NAVSHIPYD MARE, with 1 copy of endl (1)
NAVSHIPYD-CHASN, with 1 copy of encl (1)
NAVSHIPYD NY MATL LAB (948), with I copy of encl (1)

SUPSHIP, Groton, with 1 copy of encl- (1)

EB Div, Genl Dyn Corp, with 1 copy of encl (1)
SUPSHIP, Newport News, with I copy of encl (1)
NNSB and DD Co, with 1 copy of encl (1)
SUPSHIP, Pascagoula, with 1 copy of enl (1)
Ingalls Shipbldg Corp, with 1 copy of encl (1)

DIR DEF RandE, Attn: Tech Lib,, with 1 copy of encl (1)
O, USNROTC and NAVADMINU, MIT, with 1 copy of encl (1)
0 in C, PGSCOL, Webb, with i copy of endl (1)
Dr. E. Wenk, Jr., The' White House, with 1 copy of encl (1)
Dr. R.C. DeHart, SW Res Inst, with 1 copy of encl (1)
Prof. J. Kempner, Polytech. Inst of Bklyn,with i: copy of encl ()
Dean V.L. Salerno, Fairleigh Dickinson Univ, with I copy of
encl (1)
Dr. G. Gerard, College of Engrg, NYU, with 2 copies of encl (1)

V2
 INVESTIGATION ON THE INFLUENCE OF STIFFENER' SIZE ON
THE BUCKLING PRESSURES OF CIRCULAR CYLINDRICAL
SHELLS UNDER HYDROSTATIC PRESSURE

by

James A. Nott

This report is based on a thesis

investigation submitted to the Faculty of
The College of Engineering of the George
Washington University in partial satisfac-
tion of the requirements for the degree
of Master of Science in Engineering.

December 1961 Report 1600

S-F013 03 02

i I I
 TABLE OF CONTENTS

Page
ABSTRACT ................................ ........ 00 .... IV

INTRODUCTION ......................................... i

CHAPTER I - THEORYOF BUCKLING ......................... 5

PLASTIC-BUCKLING THEORY ........................... 5
METHOD FOR COMPUTING PLASTIC-COLLAPSE PRESSURE..... 19
ELASTIC -BUCKLING THEORY ........ o ................ 20
THEORETICAL RESULTS ......
00 000000...........0000.00 00.......00....00000 21
CHAPTER II - EXPERIMENTAL INVESTIGATION .............. 24

DESCRIPTION OF MODELS ............................ ,24

INSTRUMENTATION .................................... 26
TEST PROCEDURE .................................. 27

STRESS-STRAIN PROPERTIES ......................... 28

CHAPTER III - TEST RESULTS ........................... 33

CHAPTER IV - DISCUSSION .............. ..... . 0 0.O.0.a 42
DISCUSSION OF EXPERIMENTAL RESULTS .. ...... a.O . ... 42
COMPARISON OF THEORY WITH EXPERIMENT ....... O...... 43
CONCLUSIONS .. ......
00..... ........ 49

ACKNOWLEDGMENTS ........................ . . 0OO .. ,

000 . * 51
NOTATION ............................. 52
APPENDIX A .............. ........ ..... 54
0 0 0 0000
0 0 0 0 0 0 0 0 00000000O
0 00

APPENDIX B ............... ............. 57

LIST OF REFERENCES ......... 00. 000 00...........00
OO00000OOO00O0
.... 00
.. 0
..
.
..
.
....
..
..
.
..
.
61

I II 1
 LIST "OF TABLES

Table Page

1. Loading Schedule.................................. 30

2. Comparison of Theoretical and Experimental

Collapse Preassures............................

.~-~----- -------- ---

---
--- ---- ------------ ------ ------*-- --------- -- ----- ----- ---------
------~------ ..-.-.- ~--. ~-~-......,
-~.~.__~_~ I~Jr
 LIST OF ILLUSTRATIONS

Figure Page
1. Coordinate System for Stiffened Cylindrical Shell.. 6
2. Effect of Frame Size on Shell-Buckling Pressures
of Steel Cylinders..........................,,. 23
3. Model Cross Section Showing Dimensions............ 25
4. Test Setup ................................... 29
5. Stress-Strain Curve for 7075-T6'Aluminum........... 32
6. Model i1 After Collapse. 36
6a. Outside View
6b. Inside View
7. Model 2 After Collapse. 37
7a. Outside View
7b. Inside View
8. Model 3 After Collapse. ....... ................... 38
8a. Outside View
8b. Inside View
9. Model 4 After Collapse. 39
9a. Outside View
9b. Inside View
10. Effect of Frame Size on Collapse Pressures
of Aluminum Cylinder s.................................. 4o
11. Effect of Frame Size on Axisymmetric Strains
of Aluminum Cylinders..................... t a l* 0

12. Graphical Comparison of Theoretical vs Epmental

Collapse Pressures for Steel Cylinders..... .... ••o 48
13. Function F ........ ......................... 55
14. Function F 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. .. .O 56

iii

I I
 ABSTRACT

0 A theoretical derivation is given for

elastic and plastic buckling of stiffened circular

cylindrical shells under external hydrostatic pres-

sure. The theory accounts for variable shell

stresses, as influenced by the circular stiffeners,

and critical buckling pressures are obtained for

simple support conditions at the shell-frame

junctures. Methods are given for the determina-

tion of collapse pressures for both elastic and

plastic asymmetric buckling by iteration and

. numerical minimization. The theory is applicable

to shells made either of strain-hardening or elastic-

perfectly plastic materials.

Using the theory developed in this

report it is shown that a variation in stiffener

size can change the buckling pressures. Test data

from high-strength steel and aluminum cylinders

are presented which show theoretical and experi-

mental collapse pressures to agree within approxi-
mately 6 per cent.

L I -I I I
 INVESTIGATION .-ON THE INFLUENCE
OF STIFFENER SIZE ON THE BUCKLING PRESSURES

OF CIRCULAR CYLINDRICAL SHELLS

UNDER HYDROSTATIC PRESSURE

INTRODUCTION
Since the USS HOLLAND was launched in 191A, the Navy
has been interested in the design of reinforced cylinders

for submarine structures. Various theoretical formulae have

S been established for the purpose of design to calculate a

collapse pressure for a specific geometry, and experimental

models have been constructed and tested under external hydro-

static pressure to check the theories.

Collapse pressures for various modes of failure must

be determined before the naval architect can arrive at a

rational design. The collapse of a cylindrical shell stif-

fened by circular frames may occur in one of three modes

depending upon its geometry. Considering a given shell-

thickness to shell-diameter ratio, failure may occur by

i. General instability,
2. Asymmetric shell buckling, or
3. Axisymmetric shell collapse.
General instability occurs when the size of the frames

t I
 is critical for a given frame spacing, resulting in collapse

of the frames together with the shell. Failure may occur

along several frames or it may occur over the entire length

of a compartment. Shell buckling occurs when frame size is

sufficient to prevent general instability, but the frame

spacing is critical. In this type of shell failure a series

of asymmetrical lobes form in the shell between frames.

Axisymmetric shell collapse occurs when the frame size is

sufficient to prevent general instability and the diameter-

. frame spacing ratio is sufficient to prevent shell buckling.

Failure occurs by yielding of the shell material, resulting

in an axisymmetric fold in the shell between frames.

Theoretical solutions for the elastic instability

of cylindrical shells have been derived by Mises (1) and
I1
Sanden and Tolke (2), and their solutions apply when stresses
in the shell are linear when buckling occurs. The problem
of plastic collapse has been recently treated by Reynolds (3)
for the asymmetric mode of failure and by Lunchick (4 and 5)

for the axisymmetric mode. In their solutions the nonlinear

effect of the stress-strain curve in the elastic-plastic

region is considered.

A subject of current interest to the naval architect

is that of the effect of the size of the reinforcing circular

frames on the asymmetric shell buckling of cylindrical shells

under external hydrostatic pressure. This problem becomes

important in the design of submarines, since it is advantageous

~---~.-~~~~~ ~~~~~~~-
-.-~-~.~ ~.~~~-~
..-~- ~. ~~ ...~~~. ------------- -------- ---- ---- -- ------- --- ----.-.
~
to have the structural material in the shell and frame so
distributed that it gives a-maximum collapse pressure for a

minimum weight.

In this report a theoretical analysis of the asymmet-

ric shell-buckling mode of a circular framed cylindrical shell

loaded under external uniform lateral and axial pressure is

presented. Gerard's (6) equations of equilibrium for plastic

buckling are solved using realistic expressions for stresses

in the shell determined by the Salerno-Pulos (7) theory, which

accounts for the effect of circular frames. Since the

stresses at the shell-frame/ junctures are in the elastic-

plastic range prior to plastic collapse, simple support is

assumed at the shell-frame junctures. A series of curves show-

ing the theoretical effect of frame size for specific geometries

of interest to the naval architect is also presented.

The analysis presented in this paper is an extension

of Reynolds' (3) solution. It differs from Reynolds' (3)

work in that the plasticity coefficients in Gerard's (6) equa-

tions of equilibrium are expressed in terms of variable shell

stresses determined by Salerno and Pulos (7). The feature

of variable shell stresses becomes important in this problem,

as a change in frame size will produce a change in shell

stresses.

Many present structures are being constructed from

ma tials such as high-strength steel and aluminum, which

I 1 11now
exhibit a nonlinear type stress-strain curve in the elastic-

plastic region. Therefore, this analysis is outlined for

strain-hardening materials. Experimental results are shown

for comparison with the theory.

 CHAPTER I
THEORY OF BUCKLING

Various investigators have considered the critical

buckling pressure of a circular framed reinforced cylinder

loaded under external uniform pressure. Mises (i) considered

the case of a cylindrical shell between two bulkheads without

intermediate stiffening rings, for which he derived an elas-

tic analysis assuming the prebuckling stresses (stresses pre-

ceding bifurcation of equilibrium) to be 6x pR and 6 pR

if
Later Sanden and Tolke (2) considered the effect of the stif-

fening rings and derived an elastic analysis for hydrostatic

loading. Reynolds' (3) analysis considered both elastic and

plastic buckling for the case with stiffening rings. In

8 R
Reynolds' (3) solution the stresses and - were used to
FF h
calculate Gerard's (6) plasticity coefficients. Reynolds (3)
I!
then used stresses, as shown by Sanden and Gunther (8), to
define prebuckling stresses in the equations of equilibrium.

Plastic-Buckling Theory

In the case of circular cylindrical shells loaded under

external lateral and axial uniform pressure (external hydro-

static pressure) the two principal axes are parallel and per-

pendicular to the longitudinal axis of the cylinder (see

Figure 1).
 z
C3

/' NX

Detail

Figure 1 - Coordinate System for Stiffened

Cylindrical Shell
Therefore, the shear stress is given by

- /V[
]

Using membrane stress theory, which considers only stresses

on the middle surface of the shell (neglecting bending), the

longitudinal membrane stress can be determined from the equa-

tion of equilibrium in the longitudinal direction:

[21

where Nx is the longitudinal force per unit length in the

longitudinal direction.

The circumferential membrane stress can be obtained

by the analysis of Salerno and Pulos (7) who express the

stress as follows:

where

In the theory of buckling, a certain stress condition

at a point in the shell is assumed to reach a limiting value

at the onset of collapse. The circumferential stress varies
with x and the stress condition is assumed to be most criti-

cal at midbay; therefore, the stress is taken at the midbay,

midplane fiber location. The function, 0, which determines

the axisymmetric stress at this location of a circular

framed cylindrical shell loaded under external hydrostatic

Xlsa3~~ r~~ *xl*I-r*.-

l-
pressure is given by the theory of Salerno and Pulos (7) and

expressed by Krenzke and Short (9) as follows:

w =I - ( - -)ct,Fz i

where a, is the ratio of frame area to shell area and is

expressed as
c :,= L,
AF
> [a4a

and P is the ratio of faying width of the stiffener to bay

length and is expressed as
[4b]
LI
and
A is the effective area of the frame obtained by
F multiplying the true area of the frame by
R/RF for internally framed cylinders and
(R/RF) 2 for externally framed cylinders,

h is the shell thickness,

L is the center to center frame spacing,

b is the effective faying width of the stiffener
in contact with the shell,

L is the clear span frame spacing, LF - b,

R is the radius to the midplane surface of the shell,

R F is the radius to the c.g. of the frame,

v is Poisson's ratio,

E is Young's modulus, and

p is external uniform pressure.

IliL1.,
..~-------
..l,,,,----,,, -rrrr~r*l~nnr~~nn~-~,~r,~
YIIIIIIllllllllll I IIIIIIIIIIII
IIII11111 _~_~__

The functions F, and F2 are defined as follows:

~(*)~~ GOSH 29 Ca~

+?z s7, [4c]

P2=

+t Caso
Co' , Si/a, ? 9 S/9.?N

in which 0 is the shell flexibility parameter and is

expressed as
L
V3 (/ - -v, [4d]
and

[4e]

where T~ is a measure of the beam-column effect and

expressed as
2
P
[4f]
2E /7
When y1 = 0 (no beam-column effect), the above expressions

for the Salerno-Pulos (7) stress at the midbay, midplane

location reduce to the Sanden-Gnther (8) theory.

^-c~~~rn*-*~-~ ~-
--- s~~
 L I

For simplification, curves are shown in Appendix A for the

evaluation of the functions, F1 and F 2 . It has been shown

by Krenzke and Short (9) that stresses computed using these

curves are within 0.2 per cent of those computed by precise

calculations.

In this analysis, the plasticity coefficients are

expressed in terms of 6x and 6s and the assumption of Reynolds

(3) that 6s = 2 6x is not made. Utilizing Gerard's (6) plas-

ticity coefficients and general differential equations of

equilibrium for the plastic region (see Appendix B) and using

Equation i, one will find that the coefficients involving the

shear stress vanish; i.e.

.3 = C 3 / = C 2 3 = 32 0. 1[5]

Therefore, the equilibrium equations can be written as

+_)3_
+ S-S- dZ
j Z 0

c 4)-2 (4

30- X T as[6]
*=
A

10

L""~m~9 -~ _r I I __ ~
 II _

4.W
D/-[ < -e; a~w* ddx )-

((7rj [6] cont'd

N. 2
2S
w o

It is obvious that an exact solution to Equations6 is not

readily obtained and an approximate solution must be sought.

If simple support conditions are assumed at the shell-

frame junctures, the boundary conditions which must be satis-

fied are

WIx=L
W [71

and

x=o Wa0
1% : [8]
2-0

Simple support implies that the frames offer no restraint to

longitudinal bending in the shell at the shell-frame junctures.

This assumption at the boundary may be justified by concluding

that, when plastic behavior begins in the shell at the shell-

frame junctures, the frames produce little restraint against

rotation of the shell. The general solution of Equations 6

satisfying the boundary conditions, Equations 7 and 8, can be

expressed as

11

~ cr~rc*lr~li-~ (~ma(~-nmlrmr-)l~lr~r~~~lr~Dll~r~ULI
P~C -rnrr*r~n l~y
 &4 = A. S/%Nks CosAx
v =Bo Cos As SIN Ax
[9]
w : Co 5// AS SIv AX 191

where the mode shape coefficients, k and X, are expressed as

k = n/R and ? = mr/L,

in which m and n are numbers of half-waves of the buckling

configuration in the longitudinal and circumferential direc-

tions, respectively. The quantities u, v, and w in Equations

9 represent small displacements in the x, y, and z directions,

respectively (see Figure 1). Substituting these displace-

ments into Equations6, we obtain three linear equations:

*g
(/ d
-dCo =o~

*~DL.V(/~zd4J~)t
.~ J ~ ~~~4 '+ do-)G AA24L)(/0c-)
, =" ) ''=
A}
- 7 .
4-1i
c
0',
-it

12

**"l'"~""llll~~~*X1n- I~~*e, *rle --------

-----
~
Equations 10 express the'displacements u, v, and w, and their

derivatives, in terms of the arbitrary mode shape parameters,

Ao, Bo, And Co. The stability criterion assumes the shell

will buckle when the displacements increase beypond limit. To

satisfy this criterion, the determinant of the coefficients

of Ao, Bo, and Co must be set equal to zero. Equating this

determinant to zero, we obtain the characteristic equation:

E O ') * I Aae(
~ ~ !s (-.)

whereE C
23 Z

13

~U~Ua~l
- IIIII~"-*I*Ir~"*-----CUUI1I-~
I~-rr~ll~-r~
Mls 10I 'Ml _ ~_.~- -

Equation 11ii implicitly expresses the pressure as a function

of 6xs' x, and N s. Noting that

2 [±2]

and
N4[13]
we can put Equation 11 in the form p =f (6x , s)

Nadai (10) and Hoffman and Sachs (11) show that an

effective stress, 6i, and an effective strain ,ican be

determined by the octahedral shear-stress theory:

and

a+
S A *EKEJ [151

If we utilize the ratio of the circumferential and longitu-

dinal stresses, we obtain

-[16]

For principal stresses and strains T = T = O, and Equation

14 with the use of Equation 16 can be written in the following

three forms:
 d.a= e a.

'From Appendix B the effective stress parameter is expressed

. as
. OC ., [17]

where the tangent and secant modull are

- [19]

and

[20]

respectively.

The values 61 and Ei in Equations 14, 15, 19, and 20 are

assumed to be the same as those in a uniaxial compression

test, and therefore, Et and E s can be obtained directly from

a stress-strain curve of the material.

15

- I-~c~"Fan~~~-~ ~---------~
------- ~~~
 Equations 17 and 18 can be written as

ax
= C AA,
[21]

and
= 3 CAM
2

where the coefficient, C, is

[21a]
,(,, '- 20 */)

and the moduli parameter, M, is

M = (I [21b]

Substituting Equations 21 into the characteristic-value equa-

tion, Equation 11, the plastic-buckling pressure, Pp can be

expressed after simplification as

C A)X ,+--(A)a
An= AO'qi- ] E~A, A'
2
(.k'-. ) - 3Mr [22]
.t.' 20) + 2'j

where *
'A2
- ,
X,
j
X.:l 6A
0 A75' 2 -484 .9Tjr
 1i W
A
illi IYYII
III I IINYIYIY
I IIIMA
IIlll
dMMI l tilillIIIMYII OMNIMIUM
I-

and
2 *- A A;).
Al - I(k('j
2)
( 2

A, A ibcnu2Tpa p nE o

A =
A, =L-Y( A,4.Aa)
A6 2z k (2/ -,)
A [23]
A, = 2A 2A%(2A- a2)(A -2 X)
A8 8kA(A 2 X2)
- A5 = I,/ "(k,L, 2).

Since 0 is a function of the pressure (see Equation 3),

Equation 22 represents a transcendental equation for the

pressure, p . The plastic-buckling pressure, pp, in Equation

22 defines a range of collapse ressures for different values

of 6i beyond the elastic limit. The flexural rigidity of the
shell, D, in Equation 22 is given by

D =
[24

where Poisson's ratio, v, in the elastic-plastic region is

shown by Gerard and Wildhorn (12) to be

I - E
: Ve )
Ve). [25]

17
.Wh I, W
IJ
II

Although the equilibrium equations were derived for a con-

stant Poisson's ratio of 1/2, Equation 25 is used in Equation

24 to account for the fact that Poisson's ratio increases from

its value v- in the elastic region to an upper limit of 1/2

e
for an isotropic, incompressible material. Poisson's ratio

in the elastic region is usually assumed to be 0.3 for most

structural materials. For a check, a rigorous comparison
shows that Equation 22 with 0 = 1 reduces identically to

Reynolds' [(3), po4] solution for 6x /s = 1/2.

Equation 14 can also be used to determine the rela-

tionship between the prebuckling stress condition in the shell

and the applied pressure. Solving Equation 14 for p, one

obtains:

R P2F 2_ I [26]

Since 0 is a function of the applied pressure (see Equation

3), Equation 26 represents a transcendental equation for the

pressure, ps.
Buckling of a cylindrical shell in the asymmetric
mode is assumed to occur when the applied pressure, Ps, equals

the plastic-buckling pressure, p . Therefore, the plastic-

p
collapse pressure, pc', which uniquely defines the plastic-

buckling pressure of the shell, is obtained by the simultan-

eous solution of Equations 22 and 26. As an analytical solution

18
to these equations would be quite tedious, if not impossible,
a graphical solution is recommended. Equation 22 can be

plotted in the form p versus 6i and Equation 26 in the form

ps versus 61. The intersection of these two curves then

defines the collapse pressure, p c

Method for Computing Plastic-Collapse Pressure

1. Assume a value for n and compute k and N by Equation

9. The value of m which will yield a minimum pc is

unity.

2. Calculate coefficients A,, A 2 , As, A 4 , As, A8 , A 7 ,

As , and As by Equations 23.

3. Compute E, and Et and E s by Equations 19 and 20 for

stresses, di, beyond the elastic limit.

4. Assume a value for p at a specific 6.
i and compute
p 1
0 by Equation 4.

5. Compute pp by Equation 22.

6. Repeat Steps 4 and 5 until the assumed value of p

in Step 4 is sufficiently close to the computed

value of Step 5.

7. Repeat Steps 1, 2, 4, 5, and 6 varying n at one point

on the stress-strain curve to determine the value of
n which will yield a minimum ppo

8. With the value of n thus obtained, repeat Steps 4, 5,

and 6 for various values of 6. between the elastic
limit
yield
and point.
limit and yield point.

19
 i iIIYIYY
YIill
l llHM IIIMMY~lIlI I Ulm

9. Plot p versus 6.i in the p - 6.i plane.

p 1 1
10. Assume a value for ps at a specific 6i and compute

0 by Equation 4.

11. Compute ps by Equation 26.

12. Iterate Steps 10 and 11 until the computed ps equals

the assumed ps.

13. Repeat Steps 10, 11, and 12 for various values of

6i beyond the elastic limit.

14. Plot p versus 6. in the p - 6. plane.

s1 1

15. The intersection of the two curves of Steps 9 and

14 gives the collapse pressure, p C

The above procedure, outlined for a strain-hardening

material, is greatly simplified for an elastic-perfectly

plastic material. As the value of 6i for an elastic-perfectly

plastic material is never greater than 6y

, a curve of 6 1
versus e i is a horizontal line in the plastic region, and a

plot of p versus 6 in Step 9 is the vertical line 6. 6

p 1 1 y

Elastic-Buckling Theory

When the geometry of the shell structure is such that

elastic buckling can occur, the intersection of p versus 6.i

p 1
and ps versus 6. occurs for a value of 6. less than 6 ,e the
S1 1 e
elastic limit of the material. In this case one sees that

Et/Es = I and Equation 22 reduces to Reynolds' solution

[(3), p. 5] for the elastic-buckling pressure, Pe' which can

20
.' 0 WIIII 1 IMMONYIYIMMIYYNYYIIl0

be written as follows:

e_ A"' (A12 [27]

A plot of pe versus 6 i is the horizontal line p = pe in the

p - 61 plane and, therefore, the critical buckling pressure

may be obtained directly from Equation 27. Equation 27 is

also a transcendental equation, and the elastic-collapse

pressure, p, must be determined by iteration.

Theoretical Results

Calculations have been carried out for a series of

geometries in the plastic-buckling range to show the effect

of frame size on the shell-buckling pressure, pc', according

to the developed theory. A strain-hardening steel with a

yield strength of 88,000 psi is used for demonstration pur-

poses, and the results are presented in graphical form in

Figure 2. As shown in the graph, the flexibility parameter,
e, has a limiting value of 4.0, for which an increase of the
relative frame size will not produce any increase in collapse

pressure. Thus, at this limit the ratio of frame area to

shell area need only be sufficient to prevent frame failure

together with shell failure. As e is a function of h and R

and is directly proportional to the bay length, L, it is seen

that for a constant h and R, e is totally dependent on L. For

this case, frame spacing is an important aspect on the effect

of frame size. The collapse pressures represented in Figure

2 are for an asymmetric collapse, and thus, for small values

of the ratio of frame area to shell area (close to zero) the

theory no longer applies, since collapse by general instabil-

ity would occur.

22
- I NNNO
Illf

_______________ I
900
-7 9: 2.00 I

800

-*

700 e-2.90

"4
a
600

500
0:4.00

r4
e
r4 400
0
00U) 9 is the Flexibility Parameter,
'3(l - V2 ) L/V
c>
300
. Experimental Value
S2.90 A/As=:.952
(See Chapter 1V)
200
Material: Strain-Hardening T-Steel
6 y = 8 8 ,000 psi

100

0
0.2 0.4 0.6 0.8 1.0

Frame Area/Shell Area

Figure 2 - Effect of Frame Size on Shell-Buckling

Pressures of Steel Cylinders
- I IIYilli , I, 0,11u
I WIMMIN
[I, 1Y N 1 NMI

CHAPTER II

EXPERIMENTAL INVESTIGATION

Description of Models
To determine experimentally what effect the circular

frames have on a cylindrical shell loaded under external hydro-

static pressure, four models were fabricated and tested in a

pressure tank. -As pioneering work is currently being con-

ducted in the use of aluminum for oceanographic research vehi-

cles, a high-strength aluminum alloy was chosen. The four

models were constructed of 7075-T6 extruded aluminum 5 1/2-

in. diameter round bar stock. Machined structural models

were favored as opposed to welded models to eliminate the

effects of initial deflections and residual stresses which

occur in a welded model. Lunchick and Short (13) and Krenzke

(14) have shown that, in welded models, the heating and cool-

ing process occurring when the webs of the frames are welded

to the shell causes an initial inward frame deflection for an

externally-framed cylinder. On the other hand, an initial

outward frame deflection occurs for an internally-framed

cylinder. These initial deflections cause residual stresses

and beam-column effects which can affect collapse pressures.

Each model had the same shell thickness, radius, and

typical bay lengths, and only the cross-sectional area of the

frames varied. Figure 3 illustrates the various geometries

 46,*5 L--PLM s t "
-. 50 La 3 spe. at .5104 - L24 *.50

.A A
Model
Dimension 1 2 3 4
End Bay Length L2 .4538 *4396 .4328 64211
Frame A Flange width w,, .0739 .1053 .1053 .1053
Frame A
Frame A Flange Thickness t,/ .0305 .0305 .0748 .1192
Frame A veb Depth dw, .0805 .1005 .1005 .1005
Frame B Flange \idth wv, 2 .0739 .1053 .1055 .10553
0
Frame B Flange Thickness t, .0533 .0482 .0946 .1420 to
2 Ch

Frame B Web Depth du .0805 .1005 .1005 .1005

LM 3. 439 P.410 3.397 3.373
Total Model Length
--- - ---
.0305
Note: Flexibility parameter, , is Frame B
2.5 for all models.

Figure 3 - Model Cross Section Showing Dimensiona

of the models. As is noted, the two end frames of each model

have been made larger in cross-sectional area than the two

typical frames at midspan, and the lengths of the end bays are

slightly changed from that of the three typical bay lengths.

These changes are necessary to reduce discontinuity stresses (15)

caused by the very rigid closure bulkheads. These end-bay

arrangements were designed by the "Optimum End Bay Design" of

Short and Bart (16) and, in effect, this design produces a

stress distribution in the three center typical bays, which

would be the same as that for a circular framed cylinder of

infinite length and under external hydrostatic pressure.

Model I had a frame area equal to 30 per cent of the

shell area. The frame area of Model 2 was 40 per cent of that

in the shell. Model 3 had a frame area 70 per cent of the

shell, and Model 4, 100 per cent of the shell area. Frame

size was varied in the size of the flange and depth of the web.

The shape of the frames on all four models was that of a T-

section, and the faying width of the webs was held constant in

order to hold the bay lengths the same. Dimensions of the

frames for each model are noted in tabular form in Figure 3.

Critical dimensions of each model were machined to a

tolerance of .0005 in., and were measured after fabrication

to check tolerances.

Instrumentation

Models I and 3 were instrumented with 38 foil-type

electrical resistance strain gages. Fifteen gages were placed

26

r+us -- -- - I I - 111
 __

on the inside shell surface of the center bay at midbay in the

circumferential direction, equally spaced around a 180-dego

generator. Fifteen additional gages were placed on the out-

side shell surface directly opposite the inside circumferential

midbay gages. Four gages were placed on the inside surface

directly beneath the centerline of the typical frames in the

circumferential direction, and four gages were placed perpen-

dicular to these for strain measurements in the longitudinal

direction.

Models 2 and 4 were instrumented with 34 strain gages.

The gages were oriented as in Models I and 3, except for the

four longitudinal gages utilized on Models I and 3 but omitted

on Models 2 and 4.

Test Procedure

The tests were conducted in the following order:

1. Model 2
2. Model 4
3. Model 3
4o Model 1

Each model was tested in a pressure tank under external hydro-

static pressure. The ends of the models were closed by a

closure bulkhead and each end was sealed by means of an "0"

ring. Each model was placed in the pressure tank and a pipe-

line was connected between the head of the tank and one clo-

sure bulkhead, venting the interior of the model to atmos-

pheric pressure. Oil was then poured into the interior of the

model to absorb energy expected to be released at the moment

-~I*ll-l-~(*gUI
as*-l~(
~I~-I Ilr*-rrr~arr~l ---------~II~---~-- CIIIC
 I _

of collapse. The tank head was then sealed, and the model was

ready for testing. Figure 4 illustrates the test setup.

Two pressure runs were made on each model. During the

first run, strains were recorded at various pressure incre-

ments and, when plastic action was observed, the pressure was

dropped to zero. Strains were then recorded at zero pressure

to determine if any permanent set had occurred. During the

second run, strains were also recorded at various pressure

increments, but the model was tested to collapse. Table I

shows the loading schedule and pressure increments at which

strains were recorded.

Pressures were applied by means of a hand-operated

hydraulic pump and were recorded with a Bourdon-Tube pressure

gage. Strains were measured by means of Baldwin Strain

Indicators.

Stress-Strain Properties

To apply the plastic-buckling theory, the stress-

strain curves of the material had to be accurately determined.

Sixteen test specimens were removed from the cylindrical

7075-T6 bar stock. At each end of the bar stock, four speci-
mens were taken in the longitudinal direction and four in the

circumferential direction, each at 90-deg. intervals. Each

specimen was taken at a 2-in. radius to coincide with the

location of the shell of the models. Each specimen was

machined into a solid cylinder 1/2 in. in diameter and 2 in.

long. Before testing, the diameter of each specimen was

~I_
 Bourdon-Tube

To Strain To Pump
Reeorder

Atmospherie
Pressure Pressure
Vent

Test Model Pressure

Tank

Figure 4 - Test Setup

29
 11

TABLE i

LOADING SCHEDULE

Pressure, in psi, At Which Strains Were Recorded

Model 1 Model 2 Model 3 Model 4

Run i Run 2 Run 1 Run 2 Run 1i Run,2 Run I Run 2

0 0 0 0 0 0 0 0

100 250 50 250 200 500 100 500

200 500 100 500 400 750 200 750

300 750 200 750 600 900 300 900

400 900 300 1000 700 1000 400 1000

500 1000 4oo00 1100 800 1100 500 O1100

600 1100 500 1200 900 1200 600 1200

700 1200 6oo00 1255 1000 1300 700 1300

800 1250 700 1300 0 1350 800 1350

900 13 00a 800 1355 1400 900 1390 c

1000 0 900 1375 #
142 0c 100 0

0 1000# 1400b 0 O1100

1100 0 1200
1200 0

# Inelastic action was first observed.

a Failure by general instabili ty occurred before strains

could be read.
b Failure by shell buckling oc curred before strains
could be read.
c Failure by shell yielding oc curred before strains
could be read.

30
measured to the nearest 0.0001 in. Each specimen was loaded

in compression by a 30,000-lb. testing machine to determine

the characteristic load-strain curves.o Load-strain curves

were obtained on fourteen specimens by means of an automatic

recording extensometer, and two circumferential specimens,

one from each end of the bar stock, were tested with a

Tuckerman strain gage. Elastic limits, together with an

assumed yield strength at the 0.2-percent offset of the

stress-strain curve, were computed for all sixteen specimens.

The elastic, tangent, and secant moduli were computed from

results of the Tuckerman tests. Figure 5 illustrates in

detail the properties of the 7075-T6 aluminum.

31
SAmININImmmWm
l.lMl l

qa

'0
*4I

O0
U-'

2 4 6 8 10

-3
Strain, 61 x 10 , in / in./in.

Figure 5 - Stress-Strain Curve for 7075-T6 Aluminum

32
 CHAPTER III

TEST RESULTS

Model 1, which had a cross-sectional frame area 30

per cent of the shell area, collapsed at a pressure of 1300

psi by plastic general instability. The frames were not of

sfficient size to prevent frame failure, and both frames and

shell failed simultaneously over the entire length of the

model in a single "deep dish" lobe. Width of the lobe was

approximately one-eighth of the circumference of the model.

Tearing of the shell from the end rings and frames occurred

throughout the lobe, and the two center frames buckled inward

(see Figure 6). Tearing of the model was a secondary effect,

and occurred after collapse because of the brittleness of

aluminum.

Model 2, which had a cross-sectional frame area 40

per cent of the shell area, collapsed at 1400 psi by plastic

asymmetric buckling. Failure occurred in all three typical

bays by a series of nonsymmetrical lobes accompanied by lat-

eral twisting of the frames. The length of the lobes was

approximately one-tenth of the circumference of the model.

In several places tearing occurred at the shell-frame

junctures, but this was not as pronounced as in Model 1. It

was observed in areas where tearing did occur, however, that

several lobes ran together, giving the appearance of a longer

than normal lobe (see Figure 7).

33
 Model 3, which had a cross-sectional frame area TO

per cent of the shell area, collapsed at 1420 psi by axisym-

metric shell yielding. Failure occurred in the first typical

bay from the end ring along a 180-deg. generator around the

circumference. Tearing occurred at the two frame-shell junc-

tures and at midbay (see Figure 8).

Model 4, which had a cross-sectional frame area 100

per cent of the shell area, collapsed at 1390 psi by axisym-

metric shell yielding similar to Model 3; however, the area of

collapse was more pronounced in Model 4. The length of the

failure in Model 4 extended over approximately 200 degrees.

S Failure occurred in the first typical bay from the end ring

and tearing of the shell at the hinge locations occurred as

in Model 3 (see Figure 9).

A graphical representation of the collapse pressures

is shown in Figure 10, together with various corresponding

theoretical formulae. The Hencky-Mises (17) yield criterion

at outside midbay assumes that failure occurs when the effec-

tive stress,6i, on an outside fiber at midbay reaches the

yield strength of the material. An extension of this theory

is that of Kempner and Salerno (18), in which failure is

assumed to occur when the stresses inside at the framefollowed

by stresses at outside midbay reach the yield strength.

Lunchick's (4) plastic-hinge theory for axisymmetric collapse

is for an elastic-perfectly plastic material and allows for an

amount of plastic reserve strength before failure occurs.

-- lll~-'~~~ru(7rrnr~ ~~yC"""
-lAD dliiYn

Strain data indicated that all models failed in the

plastic region. Slight permanent set occurred after the first

pressure run, and much more was observed after the models failed.

Strain sensitivities for Models i, 2, 3, and 4 measured during

tests are shown in Figure 11ii, together with the corresponding

theoretical strain sensitivities as determined by the Salerno-

Pulos (7) theory. Strain levels decreased in the circumfer-

ential direction both inside and outside in all four models

for an increase of per cent area of frame. However, strain

levels increased in the longitudinal direction on the inside

surface at the frame for an increase of per cent area of frameo

35

'-x"~"--~s~- YC
 Fig. 6a Outside View

Pig. 6b Inside View

Figure 6 - Model 1 After Collapse

56
 Fig. Ta Outside View

Fig. 7b Inside View

Figure 7 - Model 2 After Collapse

57
 Fig. 8a Outside View

Fig. 8b Inside View

Figure 8 - Model 3 After Collapse

38
 INCHES
I 1 1 0

Fig. 9a Outside View

Fig. 9b Inside View

Figure 9 - Model 4 After Collapse

mim
IVI IIIYY11
mmum m mmYI

1800 I
Lunchick Plastic-
Reynolds- (3) Hinge (4)
1600
Eqs. 22 & 26, pc

1400

1200
Hencky-Mises at Kempner-
Outside Midbay (17) Salerno (18)
1000
Type of Applicable
Model Symbol Failure Theory

800 Plastic
1 General hone
Instability
Plastic
600 2 Asymmetric (a)
Buckle

Axisymmetric (b)
400 3 Yield-Type (b)
Collapse

Axisymmetric (b)
4 Yield-Type (b)
200 _ _ _ _Collapse

(a) Theoretical Buckle Equations

(b) Theoretical Yield-Type Equations
0 , , ,7,

0.2 0.4 0.6 0.8 1.0

Frame Area/Shell Area

Figure 10 - Effect of Frame Size on Collapse Pressures

of Aluminum Cylinders

40
 -7.0

Experimental Inside
kidspan (Circumferential)

Experimental Outside
Midspan (Circumferential)

S-P* Mid s pan

-5.0 ------ (Circumferential)

S-P Inside
at -rrame
-4.0 I - -- -~ (Longitudinal)

t zz

/V
-2.0 Experimental Inside
at Frame (Longitudinal)

0
-20
I Experimental Inside
I
0
at Frame (Circumferential)

S-P* at Frame /
-1.0 I "
(Circumferential)
* Circumferential strains in Salerno-Pulos (7)
theory are same lor outside and inside
surfaces.
I I I I
0.2 0.4 0.6 0.8 1.0
Frame Area/Shell Area

Figure 11 - Effect of Frame Size on Axisymmetric

Strains of Aluminum Cylinders

41
 1/1

CHAPTER IV

DISCUSSION

Discussion of Experimental Results

The experimental results showed that an appreciable

increase in collapse pressure occurred from the 30-percent

frame area case to the 40-percent frame area case. This kind
of behavior can be explained. At the 30-percent frame area

a general instability failure occurred. At the 40-percent

frame area buckling of the shell occurred between frames. Only
a small increase in collapse pressure occurred between 40-

percent and 70-percent frame size. At 70-percent frame size

an axisymmetric yield-type failure occurred instead of asym-

metric buckling. Strains at the frame indicated that longitu-

dinal stresses grow with an increase of per cent frame size

(see Figure 11ii) which could cause premature yielding. A sub-

sequent increase to 100-percent frame size caused collapse at

a lower pressure than that of the 40-percent frame size. For
an increase in per cent frame size, the relative decrease in

circumferential strains at a frame was greater than the decrease

in circumferential strains at midbay. This shows that large

frames lower frame deflections, but increase bending of the

shell at the frames, thus causing relatively higher longitudinal

I ~'~~~~-P"
stresses in the shell at the frame locations. Therefore, in

the case of the 100-percent frame size, the bending stresses

in the shell at the frames could have adversely affected the

collapse pressure.

Comparison of Theory with Experiment

For the models tested, the asymmetric theory predicts

an increase in shell-buckling pressure for an increase in frame

size. As only Model 2 failed in this mode, it is difficult to

make a positive conclusion concerning the actual trend. How-

ever, it would seem reasonable to assume, from the much lower

collapse pressure of Model I and the higher pressure of Model

3, that the experimental buckling pressures also increase with

an increase of frame size to a point where axisymmetric col-

lapse occurs. This increase in buckling pressure for an increase

of frame size agrees with Equations 22 and 26, p c' as shown in

Figure 10. Using Equations 22 and 26, pc, and Lunchick's (4)

plastic-hinge theory for axisymmetric collapse, the transition

between asymmetric and axisymmetric collapse occurs for a

frame area 62 per cent of the shell area, which case is between

Models 2 and 3.
The solution of Equations 22 and 26 of this report,

Reynolds' (3) theory, and Lunchick's (4) plastic-hinge theory

all predict collapse pressures on the unconservative side of

the experimental values. Reynolds (3) does not completely

account for actual prebuckling stresses in the shell as influ-

enced by the frames, and the plastic-hingetheory is not strictly

applicable to a strain-hardening material.

-~C~r(-~Vllmn~-rrrMLlr~~-*~~~P"-L
I~-L
 iIYI I'--

When the Hencky-Mises (17) yield criterion is utilized

at outside midbay, theoretical collapse pressures are on the

conservative side of the experimental values. The theory of

Kempner and Salerno (18) shows collapse pressures slightly lower

than those given by the theory of Hencky and Mises (17). The

theory of Kempner and Salerno (18) predicts collapse to occur

when the effective stresses in the shell on the inside surface

at the frame, followed by stresses at outside midbay reach the

yield strength of the material.

The experimental results for aluminum indicate that

failure occurs at a point somewhere between the membrane and

outer-fiber stress criteria. This should only apply to the

S aluminum used in these tests, and possibly for other structural

materials with the same properties as 7075-T6 aluminum.

The over-all effect of frame size between frames with

an area 40 per cent and 100 per cent of the shell area is

shown by Lunchick's (4) plastic-hinge theory, which indicates

only a slight change in collapse pressure with an increase in

frame area. Once axisymmetric yielding occurs, it appears

evident that any positive effect of an increase in frame size

in increasing collapse pressures is offset by higher bending

stresses created both at the frame location and at midbay.

As the test data presented in this report is insuffici-

ent for an adequate comparison with the theoretical concepts

presented in Chapter I, other experimental data are studied.

Reynolds (3), in his comprehensive study of plastic buckling,

44
also reported the test results of seven steel models, five

of welded construction and two machined. Results of these

tests are compared with theoretical formulae in Table 2.

Equations 22 and 26, p c, are within an average of 1.9 per cent

on the conservative side of test results, the maximum devia-

tion being 3.5 per cent for welded steel construction; average

deviation for machined steel construction is 4.9 per cent on

the conservative side, the maximum being 6.0 per cent. Model

2 of machined aluminum construction shows a theoretical value

6.1 per cent on the unconservative side of experiment.

Table 2 also shows a comparison between the analysis

presented in this report and Reynolds' (3) theoretical results.

For the eight geometries considered, Equations 22 and 26, p c

and Equation 27, pe, for the plastic- and elastic-buckling

pressures, respectively, are approximately 3 per cent on the

conservative side of Reynolds# (3) comparable solutions.

Figure 12 gives a graphical representation of theoretical ver-

sus experimental collapse pressures for the steel cylinders

shown in Table 2. Equations 22 and 26, pc' and Reynolds' (3)

plastic equations are shown to agree within approximately 6

per cent of the experiments. The elastic equations, Equation

ft

27, pe' Mises (1), Sanden and Tolke (2), and Reynolds (3),

predict collapse pressures which are unconservative when com-

pared with the experimental results. This can be expected,

since all the test models collapsed plastically.

 The property parameter, defined as

h=R [28]
y

is shown superimposed on the graphs in Figure 12. When h/R

is relatively high and 6y/E is relatively small, a high value

of is obtained. This is the case for Model U-12, in which

h/R is 0.o0193 and 6y/E is 2.27X10- s for 0.488 frame area to shell

area ratio (see Table 2). Also, for small values of h/R and

large 6y/E a low ( is obtained, as shown for Models T-2A and

T-3. The trend of the ( curve in Figure 12 agrees favorably

with the trend of the elastic-buckling equations. This should
be expected, as for Model U-12 the large h/Rincreases the

theoretical elastic-buckling pressure, and the small 6y/E

lowers the experimental collapse pressure. Thus, for this

case, a high ratio of theoretical collapse to experimental

collapse is obtained. Conversely, for Models T-2A and T-3

the small h/R and large 6y/E produce more conservative values
for the ratio of theoretical collapse to experimental col-

lapse. The trend of the plastic-buckling equations is inverted

from the ( curve. This is explained by noting that Equation

22, which defines the plastic-buckling pressure, is a function

of the effective stress, i, in the elastic-plastic region.

For relatively small values of 6 /E the effective stresses are

y
also relatively small, and the intersection of Equations 22

and 26 occurs at a relatively lower pressure.

 11411

TABLE 2

COMPARISON OF THEORETICAL AND EXPERIMENTAL COLLAPSE PRESSURES

**
Model Number I - 2* T - 3* T- 6* T - 2A* T - 7A* U - 12 U - 22 2

Relative Frame Size** 0.952 0.694 0.609 0.418 0.289 0.488 0.576 0.400

6y/Ex 10 s 2.93 3.60 3.83 3.43 2.80 2.27 2.35 6.89

h/R x 102 0.679 0.669 0.953 0.653 0.979 1.093 0.873 1.525

Shape of Frame Tee Rectangular Tee

Material T - Steel Mild Steel 7075-T6

Construction Welded Machined

Exp. Collapse Press. 670 553 1005 680 770 975 735 1400
Eqs. 22 & 662 548 980 691 743 917 707 1485
26, PC
Inelastic T / E*** 0.988 0.991 0.975 1.016 0.965 0o.940 0.962 1.061
Buckling
Reynolds(3) 696 563 1016 705 748 938 734 1502
T / E 1.039 1.018 1.011 1.037 0.971 0.962 0.999' 1.073
Eq. 27, 878 603 1210 755 978 1895 988 1943
Pe
T / E 1.310 1.090 1.204 1.110 1.270 1.944 1.344 1.388
Reynolds 906 626 1259 756 1010 1907 1002 1976
(3)
Elastic T / E 1.352 1.132 1.253 1.112 1.312 1.956 1.362 1.411
Buckling
Sanden- 930 631 1258 773 1032 2014 1054 1977
Tolke (2)
T / E 1.388 1.141 1.252 1.137 1.340 2.066 1.434 1.412
Mises (1) 786 585 1180 705 995 1786 963 1815

T / E 1.173 1.058 1.174 1.037 1.292 1.832 1.310 1.296

* Data taken from reference 3

** Relative Frame Size is Frame Area/Shell Area
*** Theoretical Collapse/Experimental Collapse

47
"11WIIll
WOMEN=I

2.0

1.8 Illi \\ *

1. 8 i11ll
ill 8*0
o

1- 1* lltI11'
Il Sanden-T61ke (2)

j1,4 \Reynolds (3) 4.0

l~=t
53\\\ Eq. 27, Pe /
i \ ises (1)
1
V49, 2.0
Ili\ , '
o 1.6
410 0

o "\li -a' \ nol (2)

0 - O

IReynolds (3)
EH Eqs. 22 & 26, pc

0.8 T/E for Plastic-Buckling Equations

T/E for Elastic-Buckling Equations
Property Parameter
0.6 .

0 0.2 0.4 0.6 0.8 1.0

Frame Area/She ll Area

Figure 12 - Graphical Comparison of Theoretical vs

Experimental Collapse Pressures
for Steel Cylinders

48

-. 1
 Conclusions

The following conclusions can be made for stiffened

* cylindrical shells loaded under external hydrostatic pressure:

i. The theory presented by the author for asymmetric

buckling adequately predicts collapse pressures

for shell geometries constructed from

• ao high-strength steel (Reynolds' (3)

experimental data)' and

b. high-strength aluminum,

when the observed collapse is in the asymmetric

-4
mode.

2. For a cylinder made of 7075-T6 aluminum and'having

a shell flexibility parameter of 2.5, an increase

in relative frame size leads to

ao a change in the observed mode of failure

between 30-percent and 40-percent frame

size from plastic general instability to

plastic asymmetric buckling,

b. a change in the observed mode of failure

between 40-percent and 70-percent frame

size from plastic buckling to an axisym-

metric yield-type collapse,

c. a change in the predicted mode of failure

from asymmetric buckling to axisymmetric

@ 49
4

----
--
--
--
-
 IY- -- -

yielding at 62-percent frame size (Lunchick's

(4) plastic hinge and Equations 22 and 26,

PC ) ,

do an increase in the theoretical asymmetric

buckling pressures between 30-percent and

70-percent frame size,

e, an increase in the experimental and theo-

retical longitudinal bending strains at the

frame locations, and

f. a decrease in the experimental and theo-

retical circumferential strains at the

midbay and frame locations.

3o A decrease in the shell flexibility parameter, e,

leads to
a. an increase in the plastic asymmetric

buckling pressures, pc', for a specified

per cent frame size and
b. a higher rate of increase in the plastic

buckling pressures for an increase in per-

cent frame size.

50
 ACKNOWLEDGMENTS

This thesis was written under the supervision of

Professor R.A. Hechtman of The George Washington University,

and is based on a study carried out at the David Taylor Model

Basin under the direction of Mr. J. G. Pulos and-Dr. M. E.

Lunchick. The author wishes to express his gratitude to these

individuals for their assistance and suggestions. He is also

grateful to Mr. T. E. Reynolds for his valuable comments, and

to Mr. R. D. Short for checking the theoretical derivations.

51
 IIL -

NOTATION

AF Effective area of frame cross section, sq. in.

Ai Coefficients for plastic-buckling equation, in. -8

b Faying width of frame, in.

Cii Gerard's plasticity coefficients, dimensionless

D Bending rigidity of shell, Esh 3 /12(i -v 2 ), lb.-in.

E Young's modulus, psi

Es Secant modulus, psi

Et Tangent modulus, psi

h Shell thickness, in.

k Mode shape coefficient, n/R, in. -

LF Center to center spacing of frames, in.

L Unsupported length of cylinder, LF - b, in.

M Moduli parameter, i1 - Et/Es, dimensionless

m, n Numbers of half-waves of the buckling configura-

tion in axial and circumferential directions,
respectively, dimensionless

Nx, Ns, Nxs Forces per unit length, lbs. per in.

p Pressure, psi

Pe Elastic -buckling pressure, psi

Pp Plastic buckling pressure, Equation 22, psi

P Plastic collapse pressure, Equations 22 and 26, psi

R Radius of cylinder to midplane of shell, in.

RF Radius of cylinder to c.g. of frame, in.

52
u, v, w Shell displacements, in.

x, y, z Coordinates, dimensionless
-2
a Effective stress parameter, 3/16i2(1 - Et/Es), psi

yi Measure of beam-column effect, p/2E(R/h)2 i - v2,

dimensionless

S Property parameter, R, dimensionless

y
7y Shear strain, radians

Ex Cs Membrane strains, in. per in.

ei Effective strain, in. per in.

6 Shell flexibility parameter, [3(1 - v 2 )] / 4 L/Rh,

dimensionless
-A Mode shape coefficient, mW/L, in."-

v Poisson's ratio, dimensionless

V Elastic value of Poisson's ratio, dimensionless
e
6X" 6s Membrane stresses, psi
6i Effective stress, psi

6e Elastic limit stress, psi

6y Yield stress, psi

0 Stress function, dimensionless

T Shear stress, psi
 APPENDIX A

For simplification, the terms F, and F2 in the Salerno-

Pulos axisymmetric stress distribution expressions are shown

in Figures 13 and 14 where

GosH ,9 - Cos' 1,9

F=
CosH , Snv "?,9B Cos '? S/v ?a 9
-1- oX
[Al]
Cos/#9 SIiN + Coas,9

Cos~ ,9 J/V*,B Co,8 /N 8

- C,0 ,?, J4 61'a
in which

[A2]
e=i ( - ---
and

( [A3]
3-(/ --v 2

-----
~~ I- **ru
0.4

03

0.2

Figure13- Function F,
 ....
......

0.9

L.. . .
.. ........
.....

0.8,

4 - 0.7 Ii ..
...... ; J
+

: .IF
T

Ip Q5 :t: it i
: H i; l iI
i:
;:
4 7 tii44
P1 : t N I I

iii
i::

O4 " li

::;11W;tli:i l::[:l::;iti!!l::i!l
W11:t.t

HHHHAH1 ;
;l:hI:I:H+ 4HH
~I I:ITITTI
; ~~~~ 1.. ~ I

o..
- - L
I I
::
0.1

::t
I
.-
F
:: i
UY0

-0 ).1~l h P 1\ 1

=04-
31 iI:1 i I; 1 1 i 1i+I P I 1 jii+ ;'I1
l ' 1.
. : i IPi ItI I ; 1 ii'1; q A VA TI:il1 1 ! 1: F ti '/.fl Htq f 1! .1 i .1 1 1i 1i 1 1. i mil!i. M

~1 I
I
I
I IJI
I I ~
i 71*1
II f 1 0!iiiili

::[: . . '.1111.14
11 i .1, 1 .

TU'A
: i 1: i : : I IT
-0.4
-L-4- I

I' I~ I [[I 1'1 I I I 1*I I III I I I I III ~

77 7
LAI.
r, ! : I ;I ; III I i 1i H 14;I I .1 1 + :I 1'[4.. ~
l,;+ .A I ' I i.
I 'I i i
k li 1:1.
P 1 I. l iiil.Ii; l!ii!:ij

I i AiI I I t I

- ti, TF T
I, F '1I T B 7 8 II'
1
I' TMI'hiT IT , F '

.U
I! lL ith-Ii ..
!'i I-'. I/ I . .............,,........... ''i"
.. i.
.. 1F
-0.7
1/j

-09 .

-sC--+ -. -
-410
.... ...... ... ITI
-1. 5.

Figure 14 - Function F
 APPENDIX B

Gerard (6), in his derivation of the plastic-buckling

equilibrium equations, assumes that when buckling occurs the

displacements u, v, and w increase slightly from that of the

displaced .equilibrium position just before buckling. These

changes in displacements cause incremental changes in the forces

and moments, designated by primes, ('), and are defined as

follows:

IA; CI C1. 3 X3)

,NS C3JYI
cz6 3 c31
'Vx --'B(CjX
C"
;-~c,4x--7 3 6-j
-zx. 36x,3)
C, -C, [B-
Ak -(c'2X2 + cz, x, -2'-2. X3)

-2x 3- C1X L63)2

57

. .I..,., .. 1 IN I i101 =011..I.. 1 - . [. ...... .. .

 where

,=

3
- ----
Et
+- 8

X ----- >]2

and the plasticity

and the plasticity coefficients
coefficients are as follows
are as follows;
GV w

( =/ - -
22 ?

C33 c- 2.

c = C,z
C,,, = / - oc. [:B3
[B3

C32,C 23 OC CfSZ)

which the
in which
in effective stress
the effective parameter is
stress parameter is
3 E o
58

- I II I .
..
......... --------- I
The axial rigidity is

38= - [B41

and the bending rigidity is

EB51
and the effective stress, as defined by the octahedral shear

law, is
020 [B6]
In the above equations Gerard has made the assumption that

Poisson's ratio is 1/2, which is true for an isotropic, incom-

pressible material. This simplifies the derivation of the
equilibrium equations.

The equilibrium equations, as shown by Gerard, are

_Fx=i X 4 215 =0

a=AS +
-- =)-S 0
=xs

F- 4, ~a/1S m 2i [B7]
aAA
ZW *W

+W aw,,, =0

where Nx, Ns, and Nxs are loads per unit width, and p is external

pressure. By use of Equations B1 and B2 these equilibrium

equations can be written in terms of the displacements and their

derivativesc
59

Y. I.
 av
C43
+ --
d x2.
Ca3 a(4s Y?

ca
cit -) -S Cia
C23d 2
2
CZ 3

C23 C3 3 C, 3
a. *-
'a y
ClA + CA's

1( 2 y ~7 T 347
C;rz aw Ca3
73 ' m
[B8]

w
Df C, ax * -C /3
j
-

xa as
+ (C12 / C 3

a'w 2 2a ' s
-,C2 5 a ba

LB C,2 L C23 ua CZJ

T 34 s y bx

,LC2 2 * C22~)
Onvs
w
Vx x S2Njcs -, */M q5± = 0.

6o

&Z-7-
 LIST OF REFERENCES

1. Von Mises, R., The Critical External Pressure of Cylindridal

Tubes Under Uniformn'Radial and Axial Load, Experiment al
Model Basin Translation 366, August, 1933.
2. Von Sanden, K. and Tolke, F., On Stability Problems in Thin
Cylindrical Shells, David Taylor Model basin Translation
33, December, 1949.
3. Reynolds, T. E., Inelastic Lobar Buckling of Circular
Cylindrical Shells Under External Hydrostatic PesSure,
David Taylor Model Basin Report 1392, August, 1960.
4. Lunchick, M. E., Ph. D., Yield Failure of Stiffened Cylinders
Under Hydrostatic Pressure, David Taylor Model Basin Repoi
129i, January, 1959.
5. Lunchick, M. E., Ph. D., Plastic Axisymmetric Buckling of
Ring-Stiffened Cylindrical Shells Fabricated From Strain-
SHardening Materials and Subjected to External -Hydiostatic
Pressure,
i961 .
David Taylor Model Basin Report 1393, January,
-.

6. Gerard, G., Compressive and Torsional Buckling of Thin-Wall

Cylinders in Yield Region, National Advisory Committee for
Aeronautics Note 3726, August, 1956o
7. Salerno, V. L. and Pulos, J. G., Stress Distribution in a
Circular Cylindrical Shell Under Hydrostatic Pressure
Supported by Equally Spaced Circular Ring Frames,
Polytechnic Institute of Brooklyn, Department of
Aeronautical Engineering and Applied Mechanics, Report
171-A, June, 1951.
8. Von Sanden, K. and Gunther, K., The Strength of Cylindrical
Shells, Stiffened by- Frames and Bulkheads, Under Uniform
External Pressure on All Sides, David Taylor Model Basin
Translation 38, Mardh, 1952.
9. Krenzke, M. A. and Short, R. D., Jr., Graphical Method for
Determining Stresses in Ring-Stiffened Cylinders Under
External H1ydrostatic Pressure, David Taylor Model Basin
Report 1348, October, 1959.
10. Nadai, A., Theories of Strength, Journal of Applied Mechanics,
Vol. I, No. 3, Jul.-Sep., 1933, pp 111i-129.

61

"~-'"~..~.7~.~-~"I~^"~~
~"~F~. -~-----*.~~~ ..1-.7~--7-.-~~-.7-.
- _1-~_.
~-- ~_.__
~_.~
_.1
11.___1--.1__11.~_11
~._ ~~~
 S 11. Hoffman, 0. and Sachs, G., Introduction to the Theory of
* Plasticity for Engineers, New York, McGraw-Hill Book Co.,
Viao, 1953,.P. 7.
12. Gerard, G. and Wildhorn, S., A Study of Poisson's Ratio
in the Yield Region, National Advisory Committee for
Aeronautics Technical Note 2561, June, 1952.

13. Lunchick, M. E., Ph. D. and Short, R. D., Jr., Behavior of

Cylinders With Initial Shell Deflections, Journ'alof
Applied Mechanics, Vol. 24, No. cDic
D4, er 957,pp.
559'-564.
14. Krenzke, M. A., Effect of Initial Deflections and Residual
Welding Stresses on Elastic Behavior and' Collapse pirebsure
of Stiffened Cylinders Sub-jected to External Hydrobstatic
Pressure, David Taylor Model Basin Report 1327T, April;
1966..
15. Timoshenko, S. and Woinowsky-Krieger, S., Theory of Plates
and Shells, 2d ed., New York, McGraw-Hill Book nd.
n.,
-J 1959, ..483.

16.- Short, R. D., Jr. and Bart, R., Analysis for Determining
" •Stresses in Stiffened Cylindrical'Shells Near Structural
Discontinuities, "David Taylor Model Basin Report 10b5,
January, 1959.
17. Seely, F. B. and Smith, J. 0., Advanced Mechanics of
Materials, New York, John Wiley and Sons,- Inc'., 1955,
p.:-di."
18. Kempner, J. and Salerno, V. L., Analysis of Inelastic
Behavior of Transversely Reinforced Cylindrical Shells
Under Hydrostatic Pressure, Polytechnic Institute of
Brooklyn Aeronautical Laboratory',Report No. 172, August,
1950.

62

r I I I I ~q
 INITIAL DISTRIBUTION

* Copies
'14 CHBUSHIPS
* 2 Sci &Res Sec (442)
1 Lab Mgt (320)
3 Tech Info Br (335)
1 Ships Res Br (341A)
I Applied Sci Br (342A)
I Prelim Des Br (420)
1 Prelim Des Sec (421)
1 Ship Protec (423)
1 Hull Des Br (44o0)
i Struc Sec (443)
1 Sub Br (525)

1 CHONR (439)
1 CNO (Op 07TB)
i CDR, USNOL
1 CDR, USNRL (2027)
V I CDR, USNOTS, China Lake
i CO, USNUOS, Newport
I CO, USNUSL, New London

10 CDR, ASTIA
2 NAVSHIPYD PTSMH
i NAVSHIPYD MARE
I NAVSHIPYD CHASN
I NAVSHIPYD NY Mati Lab (948)
1 SUPSHIP, Groton
1 Elec Boat Div, Genl Dyn Corp
1 SUPSHIP, Newport News
1 NNSB & DD Co
1 SUPSHIP, Pascagoula
1 Ingalls Shipbldg Corp

1 DIR DEF R & E, attn: Tech Lib

1 CO, USNROTC &:NAVADMINU, MIT
1 0 in C, PGSCOL, Webb
1 Dr. E. Wenk, Jr., The White House
i Dr. R.C. De Hart, SW Res Inst
1 Prof. J. Kempner, Polytechn Inst of 'Bklyn
1 Dean V.L. Salerno, Fairleigh'Dickinson Univ
2 Dr. G. Gerard, College of Engin, New York Univ

63

'C I I
David Taylor Model Basin. Report 1600. David Taylor Model Basin. Report 1600.
INVESTIGATION ON THE INFLUENCE OF STIFFENER SIZE 1. Cylindrical shells (Stif- INVESTIGATION ON THE INFLUENCE OF STIFFENER SIZE 1. Cylindrical shells (Stif-
ON THE BUCKLING PRESSURES OF CIRCULAR CYLINDRICAL fened)--Pressure--Measure- ON THE BUCKLING PRESSURES OF CIRCULAR CYLINDRICAL fened)--Pressure--Measure-
SHELLS UNDER HYDROSTATIC PRESSURE, by James A. Nott. ment SHELLS UNDER HYDROSTATIC PRESSURE, by James A. Nott. ment
Dec 1961. ii, 63p. illus., graphs, tables, refs. UNCLASSIFIED 2. Cylindrical shells (Stif.- Dec 1961. ii, 68p. illus., graphs, tables, refs. UNCLASSIFIED 2. Cylindrical shells (Stif-
fened)--Buckling-- fened)--Buckling--
A theoretical derivation is ,..a:en for elastic and plastic buck- Measurement A theoretical derivation is given for elastic and plastic buck- Measurement
ling of stiffened circular cylindrical shells under external hydro- 3. Cylindrical shells (Stif- ling of stiffened circular cylindrical shells under external hydro- 8. Cylindrical shells (Stif-
static pressure. The theory accounts for variable shell stresses, fened)--Elasticity--Model static pressure. The theory accounts for variable shell stresses, fened)--Elasticity--Model
as influenced by the circular stiffeners, and critical buckling tests as influenced by the circular stiffeners, and critical buckling tests
pressures are obtained for simple support conditions at the shell- 4. Cylindrical shells (Stif- pressures are obtained for simple support conditions at the shell- 4. Cylindrical shells (Stif-
frame junctures. Methods are given for the determination of col- fened)--Failure frame junctures. Methods are given for the determination of col- fened)--Failure
lapse pressures for both elastic and plastic asymmetric buckling I. Nott, James A. lapse pressures for both elastic and plastic asymmetric buckling I. Nott, James A.
by iteration and numerical minimization. The theory is applicable II. S-F018 03 02 by iteration and numerical minimization. The theory is applicable II. S-F013 03 02
to shells made either of strain-hardening or elastic-perfectly to shells made either of strain-hardening or elastic-perfectly
plastic materials. plastic materials.

--^-

David Taylor Model Basin. Report 1600. David Taylor Model Besin. Report 1600.
INVESTIGATION ON THE INFLUENCE OF STIFFENER SIZE 1. Cylindrical shells (Stif- INVESTIGATION ON THE INFLUENCE OF STIFFENER SIZE 1. Cylindrical shells (Stif-
ON THE BUCKLING PRESSURES OF CIRCULAR CYLINDRICAl fened)--Pressure--Measure- ON THE BUCKLING PRESSURES OF CIRCULAR CYLINDRICAI fened)--Pressure--Measure-
SIIELLE UNDER HYDROSTATIC PRESSURE, by James A. Nott. ment SHELLS UNDER HYDROSTATIC PRESSURE, by James A. Nott. ment
Dec 1961. ii, 63p. illus., graphs, tables, refs. UNCLASSIFIED 2. Cylindrical shells (Stif- Dec 1961. ii, 43p. illus., graphs, tables, refs. UNCLASSIFIED 2. Cylindrical shells (Stif-
fened)--Buckling-- fened)--Buckling--
A theoretical derivation is given for elastic and plastic buck- Measurement A theoretical derivation is given for elastic and plastic buck- Measurement
ling of stiffened circular cylindrical shells under external hydro- 3. Cylindrical shells (Stif- ling of stiffened circular cylindrical shells under external hydro- 8. Cylindrical shells (Stif-
static pressure. The theory accounts for variable shell stresses, fened)--Elasticity--Model static pressure. The theory accounts for variable shell stresses, fened)--Elasticity--Model
as influenced by the circular stiffeners, and critical buckling tests as influenced by the circular stiffeners, and critical buckling tests
pressures are obtained for simple support conditions at the shell- 4. Cylindrical shells (Stif.- pressures are obtained for simple support conditions at the shell- 4. Cylindrical shells (Stif-
frame junctures. Methods are given for the determination of col- fened)--Failure frame junctures. Methods are given for the determination of col- fened)--Failure
lapse pressures for both elastic and plastic asymmetric buckling I. Nott, James A. lapse pressures for both elastic and plastic asymmetric buckling I. Nott, James A.
by iteration and numerical minimization. The theory is applicable II. S-F013 03 02 by iteration and numerical minimization. The theory is applicablE I. S-F018 08 02
to shells made either of strain-hardening or elastic-perfectly to shells made either of strain-hardening or elastic-perfectly
plastic materials. plastic materials.
 Using the theory developed in this report it is shown that a variation in stiffener size can Using the theory developed in this report it is shown that a variation in stiffener size can
change the buckling pressures. Test data from high-strength steel and aluminum cylinders change the buckling pressures. Test data from high-strength steel and aluminum cylinders
are presented which show theoretical and experimental collapse pressures to agree within are presented which show theoretical and experimental collapse pressures to agree within
approximately 6 percent. approximately 6 percent.

Using the theory developed in this report it is shown that a variation in stiffener size can Using the theory developed in this report it is shown that a variation in stiffener size can
change the buckling pressures. Test data from high-strength steel and aluminum cylinders change the buckling pressures. Test data from high-strength steel and aluminum cylinders
are presented which show theoretical and experimental collapse pressures to agree within are presented which show theoretical and experimental collapse pressures to agree within
approximately 6 percent. approximately 6 percent.

-%4_
b .
David Taylor Model Basin. Report 1600.
INVESTIGATION ON THE INFLUENCE OF STIFFENER SIZE 1. Cylindrical shells (Stif-
ON THE BUCKLING PRESSURES OF CIRCULAR CYLINDRICAL fened)--Pressure--Mea sure-
SHELLS UNDER HYDROSTATIC PRESSURE, by James A. Nott. ment
Dec 1961. ii, 48p. illus., graphs, tables, refs. UNCLASSIFIED 2. Cylindrical shells (Stif-
fened)--Buckling--
A theoretical derivation is given for elastic and plastic buck- Measurement
ling of stiffened circular cylindrical shells under external hydro- 8. Cylindrical shells (Stif-
static pressure. The theory accounts for variableshell stresses, fened)--Elasticity--Model
as influenced by the circular stiffeners, and critical buckling tests
pressures are obtained for simple support conditions at the shell- 4. Cylindrical shells (Stif-
frame junctures. Methods are given for the determination of col- fened)--Failure
lapse pressures for both elastic and plastic asymmetric buckling I. Nott, James A.
by iteration and numerical minimization. The theory is applicable II. S-F013 03 02
to shells made either of strain-hardening or elastic-perfectly
plastic materials.
 Using the theory developed in this report it is shown that a variation in stiffener size can
change the buckling pressures. Test data from high-strength steel and aluminum cylinders
are presented which show theoretical and experimental collapse pressures to agree within
approximately 6 percent.
 ~e rr I II I I -r I laa~___ I

DEC 23 1098

- --