DTMB 1959 1324 PDF
DTMB 1959 1324 PDF
DTMB 1959 1324 PDF
/e , --
-
V393
.R46 I
AERODYNAMICS
Thomas E. Reynolds and William F. Blumenberg
STRUCTURAL
MECHANICS
- I _~
GENERAL INSTABILITY OF RING-STIFFENED CYLINDRICAL
SHELLS SUBJECT TO :.XT2~NAL HYDROSTATIC PRESSURE
by
Page
A B ST R A C T ...............................................................................
1
F UT U R E WOR K .......................................................................................................................... 11
RE FERE NC ES ............................................................................................................................ 11
LIST OF FIGURES
LIST OF TABLES
-r' I I ,_~ II I I 1 II I, I I I I I
ABSTRACT
Hydrostatic pressure tests were conducted on four machined ring-stiffened
cylinders of different lengths to investigate the influence of end fixity on the
elastic general-inktability pressure. Each cylinder was tested nondestructively
with several different end conditions. Pressure variations of as much as 65 per-
cent were observed for a particular cylinder.
Other tests were conducted with one cylinder reinforced by internal mov-
able bulkheads whereby buckling pressures were obtained as a function of
bulkhead spacing. The resulting experimental curve confirms the existence of
discontinuities which occur, according to Kendrick's theory, at the transition
between two circumferential buckling modes.
INTRODUCTION
A project designated "Deep Dish" was instituted at the David Taylor Model Basin for
the study of elastic general instability of ring-stiffened cylindrical shells under external hy-
drostatic pressure. This work has been reported in References 1 through 7.* A primary ob-
8 9
jective of the project has been the evaluation of Kendrick's theoretical work ' through tests of
small machined cylinders having systematic variations in geometric parameters. This report
summarizes background information on this project and presents results of recent studies.
4 1 -rB
contention, it still led investigators at the Model Basin to believe that if clamped ends could
be approximated in the laboratory, the experimental pressures would exceed Kendrick's pres-
sures by a considerable margin. The results of the first (DD-8) series agreed well with Ken-
drick's theory. However, this close correlation did not persist in the second and third series
5
of tests. Some of these cylinders collapsed at pressures well above the Kendrick pressures,
although not nearly as high as those given by Kaminsky's theory.
The results of the DD-8, DD8-1, and DD8-2 series appeared to be so inconsistent that
no definite conclusions regarding the validity of available theories could be made without
additional investigation. Furthermore, it seemed clear that any intelligent interpretation of
these or future test results would require some knowledge of the influence of the experimental
end conditions on the collapse pressure. In Reference 5 it was proposed that duplicates of
of the three cylinders of the DD8-2 series which showed poorest agreement with Kendrick's
theory be tested with several different end closure arrangements. The buckling pressures
would be determined nondestructively by means of the Southwell method. 6 In this way it was
also hoped that some closure arrangement giving close agreement with Kendrick's theory
could be found, since it was desirable to use such an arrangement with forthcoming studies of
cylinders with intermediate heavy stiffeners.
RECENT STUDIES
So far the three duplicate cylinders have been tested with five different end attachments,
and an additional cylinder (DD8-2-6), the shortest of the series, has also been tested. Fur-
thermore, the longest cylinder, 4-A, had been tested using a technique with movable insert
disks to provide a test section of varying length. Use of this arrangement should more closely
simulate the actual conditions existing in a submarine. Also, since the only rotational re-
straint at the ends of the test section would come from the continuity of the shell, this con-
dition would appear to approximate that of simple supports more closely than the closure
arrangements used previously, thereby providing a more valid test of Kendrick's theory. Since
experimental points could be obtained at small intervals of length, a true experimental curve
of buckling pressure versus length could be established. Of particular interest was the tran-
sition between two and three circumferential lobes, a point where, theoretically, a disconti-
nuity (cusp) should exist.
This method thus providesa rapid and inexpensive means of obtaining much additional
data through the use of one existing cylinder. Thus far, tests have been completed with ten
different compartment lengths. The results of these tests together with those obtained using
the various end closure arrangements are summarized briefly in this third progress report.
I
Figure 1 - Test Cylinders with Gages Installed
Cylinder dimensions are shown in Figure 2. As with the original series of cylinders, these
cylinders had rectangular stiffening rings and were accurately machined from steel tubing
whose high yield strength (85,000 psi) was sufficient to prevent inelastic action prior to the
onset of buckling. The various end closure arrangements are illustrated in Figure 3. Case IV,
4
the arrangement used with the original set of cylinders, employed a closure plate approxi-
mately 1in. thick. The 2-in. plates were chosen to find out whether a thicker plate would
affect the collapse pressure significantly. Case II closure was used to determine the effect
of replacing the distributed pressure over the surface of the thinner plate with a concentrated
load around its edge. This was done by means of pressure caps which are shown in Figure 4
along with the /-in. and 2-in. closure plates. All the closure plates were equipped with ma-
chined inserts fitting tightly within the cylinder wall to prevent radial deflections at the ends.
Each cylinder was instrumented with electrical resistance strain gages most of which
were confined to the exterior of the centermost stiffener and oriented circumferentially. The
6
buckling pressures were obtained by the Southwell method applied to the strain-pressure plots.
The buckling pressures obtained under the various end conditions are given in Table 1
together with the pressures for the original group of cylinders and the theoretical pressures
according to Kendrick Part III* and Kaminsky.
3 7
*Since it has been shown ' that the theory of Kendrick Part I is often unconservative, and sometimes may be
unsafe to use, attention is confined to the second solution of Kendrick Part III, which is simply a more rigorous
treatment of the same problem.
Figure 2 - Schematic Diagram Showing Dimensions of Machined Cylinders
All dimensions are in inches.
The same information is presented graphically in Figure 5. The close agreement between Cy-
linders 2 and 2-A, 3 and 3-A, and 4 and 4-A under Case IV in Table 1 indicates that the three
cylinders were accurately duplicated. Most striking, perhaps, is the very large increase in
strength realized in progressing from Case I to Case V. For Cylinder 3-A, this amounted to
a pressure increase of 65 percent. Also remarkable is the nearly perfect agreement with
Kendrick's theory for Case I. Another interesting result was that, in those cases (I-IV) where
one or both %-in. plates were used, the appearance of the characteristic lobar strain pattern
prior to buckling was influenced by the orientation of the end plates.. That is, with Case II,
Cylinder 3-A, for example, a rotation through 90 deg of the end plates produced a 90 deg rota-
tion of the circumferential strain pattern, along with a slight change in the buckling pressure,
.... M
" W11111
---- ~- ----- MINI1
Case I
Case -2
TABLE 1
Results of Tests with Varied End Conditions
1000- 100
Experimental Pressures
o Case I
* Case Il
900 A Case I
A Original se 11
_V Duplicate Case"
o Case YT
800 * End Plates Rotated 90 degrees
700 (3)
Komaminsky Theory
/(Clamped Support)
600 Reference 10
\ ~(3)
S(2) O(2)
-Kendrick Part IMrTheory 0(2)
(Simple Support) V(2)
300 Reference 9 A(2) 0(2)
100
Cylinder Cylinder Cylinder Cylinder Cylinder Cyinder
SNo.6- No.I No. 2 N .3 No .45
0 -- 1 1 1
8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50
Frame Spaces
as shown in Table 1. This cannot be attributed to any lack of circularity in the plate inserts,
which were more nearly circular than the cylinders, nor was there any obvious lack of uniform-
ity in the plate cross sections which would indicate a lack of symmetry in the bending rigidity.
Rotation of the 2-in. plates, on the other hand, had no effect on the orientation of the strain
pattern nor on the buckling pressure.
Cylinder 4-A, the longest of the duplicate group, was tested with movable insert disks
simulating bulkheads. These were circular disks with edges machined to a small radius to
~ __
r
18"
8.1
~/~//~,~/////~P~
I R
- ----- -~~ --
500
400- -- - .) 13
O(2)
Kendrick Port m Theory
50 - (Simple Support) - Reference 9
350
2300 -_
250
200_
Tests were conducted with ten different compartment length-liameter ratios ranging
from 2.40 to 4.66. Table 2 lists the compartment length-diameter ratios, the corresponding
experimental pressures, and the pressures given by the Kendrick Part III theory.* These re-
sults appear graphically in Figure 7. It is seen that, although the experimental points lie
somewhat above the theoretical curve, they follow a similar curve, proving the existence of
the discontinuity (cusp) formed at'the intersection of the curves for two and three circumfer-
ential lobes. This fact, not previously verified experimentally for plate and shell structures,
should be of more than mere academic interest, since it indicates that compartment lengths
must be chosen with care if the material is to be used efficiently. For example, the experi-
mental curve shows that a compartment comprising 31 frame spaces can be appreciably strength-
ened if it is shortened by two frame spaces. Beyond that point, however, little benefit is gained
unless a reduction of ten spaces or more is made. In comparing these results with the previous
tests, good agreement is found between the pressures for 25 and 31 frame spaces and those
under Case IV for the cylinders (2 and 3) of corresponding length.
*The calculated pressures for 25 and 31 frame spaces given in Table 2 do not agree exactly with those given
in Table 1 for Cylinders 2 and 3 because, for all cylinders, the end frame spacings are slightly shorter than
typical, whereas the test sections in Table 2 had uniform frame spacing.
9
61WM101 II 0 0 I io ,w IAlMlil mla rIYII I I
Le = KL
where L is the actual length of the cylinder,
Le is its effective length, and
K is a constant depending on the degree of fixity, being less than one for all cases
falling between simple support and clamped ends.
Such a relationship is valid for the buckling of a centrally compressed column and has been
11
utilized with some success by Arnold and Warburton in a study of the effect of boundary con-
ditions on the flexural vibrations of unstiffened tubes.
If this relationship were to hold for the case of general instability, all conditions of
edge fixity could be represented by a single curve of pressure versus effective length, pro-
vided K is known as a function of edge fixity. From this representation comes the interesting
result that the transition from one circumferential mode to another must occur at the same pres-
sure, regardless of the degree of fixity. Figure 7 shows just this sort of correspondence be-
tween Kendrick's solution for simple support and the experimental curve obtained from the
tests with internal bulkheads. The theoretical transition between two and three lobes occurs
at about 401 psi, while experimentally it appears at about 389 psi. This small difference in
pressures could result from variations in cylinder dimensions within the specified tolerances
or from slight inaccuracies arising from the use of nominal values for Young's modulus
(30 x 106 psi) and Poisson's ratio (0.3) in the calculations.
On this basis Kendrick's theory would appear to be very accurate. On the other hand,
Kaminsky's theory for clamped ends (Figure 5) shows the transition from two to three lobes at
about 520 psi, considerably higher than that found experimentally.
It is emphasized that this concept of effective length is presented merely as one inter-
pretation of the results thus far obtained. At the present time there are insufficient data for a
proper evaluation of the idea. It can only be said that none of the tests to date has shown it
to be invalid, and that one of the immediate objectives of Project Deep Dish should be to pro-
vide the additional data necessary for a proper evaluation.
~I I I I II I I L I r Illir
I_ _ 111=l1 1 11,
FUTURE WORK
In order to continue the line of study described in this report and to investigate other
related areas of interest, the following work is anticipated:
2. Efforts to design a test arrangement that will simulate the simple support condition,
assumed by Kendrick in his theoretical treatment.
3. Further tests of cylinders having one intermediate heavy stiffener, toinvestigate the
efficiency of deep frames in breaking up the overall length of a submarine compartment. The
size of the stiffener will be systematically reduced, and the collapse pressure will be deter-
mined at each stage by nondestructive testing.
REFERENCES
1. Slankard, R.C., et al, "An Experimental Investigation of the Effect of Radial Excitation
on the General-Instability Strength of Stiffened Cylindrical Shells Subjected to Hydrostatic
Pressure (Models 1A and 1K)," David Taylor Model Basin Report C-724 (Jan 1956)
CONFIDENTIAL.
2. David Taylor Model Basin CONFIDENTIAL letter C-SS/S11 Serial 0171 of 28 Feb 1955
to Bureau of Ships.
3. Slankard, R.C. and Galletly, G.D., "The Effect of Reinforcing Rings on the General-
Instability Strength of Machined Cylindrical Shells under External Hydrostatic Pressure,"
David Taylor Model Basin Report C-822 (Jun 1957) CONFIDENTIAL.
f3Cr~-
,,srrmneha**rrn~~**~Cw*riuruuorr~i~*ao
ilwl~
,1611,
1WI 6111116
11 ,,Ia
6. Galletly, G.D. and Reynolds, T.E., "A Simple Extension of Southwell's Method for
Determining the Elastic General Instability Pressure of Ring-Stiffened Cylinders Subject to
External Hydrostatic Pressure," Proceedings for the Society of Experimental Stress Analysis,
Vol. XIII, No. 2, p. 141 (1956).
7. Reynolds, T.E., "A Graphical Method for Determining the General Instability Strength
of Stiffened Cylindrical Shells," David Taylor Model Basin Report 1106 (Sep 1957).
8. Kendrick, S., "The Buckling under External Pressure of Circular Cylindrical Shells
with Evenly Spaced Equal Strength Circular Ring Frames-Part I," Naval Construction Research
Establishment Report NCREI/R.211 (Feb 1953).
9. Kendrick, S., "The Buckling under External Pressure of Circular Cylindrical Shells
with Evenly Spaced Equal Strength Circular Ring Frames-Part III," Naval Construction
Research Establishment Report NCRE/R.244 (Sept 1953).
10. Kaminsky, E.L., "General Instability of Ring-Stiffened Cylinders with Clamped Ends
under External Pressure by Kendrick's Method," David Taylor Model Basin Report 855
(Jul 1954).
11. Arnold, R.N. and Warburton, G.B., "The Flexural Vibrations of Thin Cylinders,"
Proceedings (A) for the Institution of Mechanical Engineers, Vol. 167, No. 1, pp. 62-80 (1953).
(y~--~-rrP~ulPII I I 1 13 ~
~I ~_
INITIAL DISTRIBUTION
Copies Copies
1 OPNAV, Op 373
CDR, USNOL
DIR, USNRL (TID)
1 NAVSHIPYD PTSMH
1 NAVSHIPYD MARE
1 SUPSHIPINSORD, Groton
1 SUPSHIPINSORD, Pascagoula
David Taylor Model Basin. Report 1324. David Taylor Model Basin. Report 1324.
1. Cylindrical shells GENERAL INSTABILITY OF RING-STIFFENED CYLINDRI- 1. Cylindrical shells
GENERAL INSTABILITY OF RING-STIFFENED CYLINDRI- (Stiffened) - Buckling -
(Stiffened) - Buckling - CAL SHELLS SUBJECT TO EXTERNAL HYDROSTATIC PRES-
CAL SHELLS SUBJECT TO EXTERNAL HYDROSTATIC PRES- Model tests.
SURE, by Thomas E. Reynolds and William F. Blumenberg. Model tests. SURE, by Thomas E. Reynolds and William F. Blumenberg.
UNCLASSIFIED 2. Cylindrical shells June 1959. ii, 13p. photos., tables, refs. UNCLASSIFIED 2. Cylindrical shells
June 1959. ii, 13p. photos., tables, refs. (Stiffened) - Buckling -
(Stiffened) - Buckling -
Graphical analysis. Hydrostatic pressure tests were conducted on four machined Graphical analysis.
Hydrostatic pressure tests were conducted on four machined 3. Cylindrical shells
3. Cylindrical shells ring-stiffened cylinders of different lengths to investigate the in-
Ling-stiffened cylinders of different lengths to investigate the in- (Stiffened) - Buckling -
(Stiffened) - Buckling - fluence of end fixity on the elastic general-instability pressure.
fluence of end fixity on the elastic general-instability pressure. Test methods (Non-
Test methods (Non- Each cylinder was tested nondestructively with several different
Each cylinder was tested nondestructively with several different destructive)
destructive) end conditions. Pressure variations of as much as 65 percent
end conditions. Pressure variations of as much as 65 percent 4. Submarine hulls -
were observed for a particular cylinder. 4. Submarine hulls - were observed for a particular cylinder.
Characteristics - Model Other tests were conducted with one cylinder reinforced by in- Characteristics - Model
Other tests were conducted with one cylinder reinforced by in- tests.
ternal movable bulkheads whereby buckling pressures were ob- tests. ternal movable bulkheads whereby buckling pressures were ob-
I. Reynolds, Thomas E. tained as a function of bulkhead spacing. The resulting experi- I. Reynolds, Thomas E.
tained as a function of bulkhead spacing. The resulting experi- II. Blumenberg, Williani F.
mental curve confirms the existence of discontinuities which II. Blumenberg, William F. mental curve confirms the existence of discontinuities which
III. NS 731 038 occur, according to Kendrick's theory, at the tran ;tion between III. NS 731 038
occur, according to Kendrick's theory, at the transition between
two circumferential buckling modes. two circumferential buckling modes.
David Taylor Model Basin. Report 1324.
GENERAL INSTABILITY OF RING-STIFFENED CYLINDRI- 1. Cylindrical shells
CAL SHELLS SUBJECT TO EXTERNAL HYDROSTATIC PRES- (Stiffened) - Buckling -
SURE, by Thomas E. Reynolds and William F. Blumenberg. Model tests.
June 1959. ii, 13p. photos., tables, refs. UNCLASSIFIED 2. Cylindrical shells
(Stiffened) - Buckling -
Hydrostatic pressure tests were conducted on four machined Graphical analysis.
ring-stiffened cylinders of different lengths to investigate the in- 3. Cylindrical shells
fluence of end fixity on the elastic general-instability pressure. (Stiffened) - Buckling -
Each cylinder was tested nondestructively with several different Test methods (Non-
end conditions. Pressure variations of as much as 65 percent destructive)
were observed for a particular cylinder. 4. Submarine hulls -
Other tests were conducted with one cylinder reinforced by in- Characteristics - Model
ternal movable bulkheads whereby buckling pressures were ob- tests.
tained as a function of bulkhead spacing. The resulting experi- I. Reynolds, Thomas E.
mental curve confirms the existence of discontinuities which II. Blumenberg, William F.
occur, according to Kendrick's theory, at the transition between III. NS 731 038
two circumferential buckling modes.
.____
I I -L
~JI I II r r rclbl C~ r ~ ~-
MAR 2 '
A 2 :1977
APE 21 off
DEC 2 3 I 7
UL 3 1979
MAb-Y
1g98
oc T 7
r- I