Professional Documents
Culture Documents
Kao R.stresses in Spherica - mar.1973.JSR
Kao R.stresses in Spherica - mar.1973.JSR
The m a x i m u m stresses are obtained for a spherical shell that is lifted or towed by a cable or any
mechanical power hoist. In view o f t h e highlylocalized nature o f t h e m a x i m u m stress induced in a
spherical shell due to local loading, the nonlinear (large deflection) shallow-shell theory is adopted
for the analysis. A nonlinear relaxation technique in conjunction with finite difference approxima-
tions is introduced for the numerical integration. Results obtained here are presented in the
graphic form that may be readily used by engineers in practical design.
I This work was supported in part by the Department of Navy __12 ( 1 - v) W. (U, - - xu q- 21 w ' 2 ) - - 4X4
rags(x) p = 0 (lb)
under IJrojeet THEMIS Grant 893, N000 14-68-A-0506.
2 l~esearch Assistant Professor, Department of Civil and Me-
chanical Engineering, The Catholic University of America, Wash- where u and w are dimensionless tangential and normal deflec-
ington, D. C. tions, respectively,
a Adjunct Professor of Mechanics, The Catholic University of
America; also, Director of Structural lVIechanics Program, Office
of Naval Research, Arlington, Virginia. V=() = (),, + ( 1 ) ( ) , , x 0 = r 0 / a
4 Numbers in brackets designate References at end of paper.
Manuscript received at SNAME Headquarters February 28,
1972. Revised manuscript received April 18, 1972. and
MARCH 1973 19
1.8
1.6
P /
1.4 /
/
1 II
1.2
~,=, /
~ LO / .i
~" r"
"o_ 0.8 -/~--~ Experimental w(0}, rola= I18, Ref. [9]
-.i Experimental w(0), rola = 116, Ref. [9]
/P=PR,E
---A Analytical W(ro), rola =118 Ref. [9]
o.e lF --i Analytical w(ro), rola=l/6~ Ref. [gJ
0.4 /~ ---m, Present result w(0), ro/a = I/8.25
Present result w(0), ro/a =1115.5
0.2/
0
IJ= ro/#-RS- 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 L6 1.8
deflection, wo
Fig. 2. Comparison of load-deflection curves o b t a i n e d in the p a p e r
for a shallow spherical shell, X ---- 5, v = -} with the results o b t a i n e d in,
reference [9] (Fig. 7)
28 -- !-- / 3.0
26 '~ 2.0
\\
1.0
~ , ~ , ~ present solution
~,= --- Result obtained
in Ref. [5]
18 U=I Yg]~ 0.5
II / / - .8 Wmax =. I
,,
w =#1 / ~x'~."x~."'~. /Result obtained by
::
max . ,,near t , e o r , , , ,
w < w
max
=.6 j ~ . ~ " ~. Ref. E6]
"~.....
"max -" ~ Wmax = ~ . ~ . ~ .
, / /// .Zu=,.0_ m"
o
.2 .4 .6 .8 1.0 1,2
Maximumdeflection.Wma~
Fig. 3 Load-deflection c u r v e s for a s p h e r i c a l shell 0.01 , i I i i , I I I I I I I I I I I I I I I I I
0.2 0.4 0.6 0.8 I0 12 1.4 1o6 1o8 20 2.2
U = ro/RJVh-
Fig. 4 Maximum stress v e r s u s l o a d e d a r e a p a r a m e t e r for d i f f e r e n t
_< p* _< 5) is obtained for each of these X values. The results so deflection Ievels
obtained indicate that the differences of the maxinmm stresses
among these X values are of the order of less than 2 percent while
the differences of the maximum deflections are less than 8 per-
cent. In addition, these results also show that deflection fields adopted in practical design. For example, if a U value of 0.5 is
die out quite rapidly from the center so that the boundary condi- selected, the maximum stress (a' ..... ) obtained here associated
tions introduced in the spherical cap have only minor effect. with a deflection level of w0 = 0.6 is only 74 percent of the value
With these preliminary test results at hand, we can proceed to obtained via linear theory in reference [6].
use one or two arbitrary X values to obtain the maximum stresses It is hoped that results presented in these two figures might
for different local levels (different P*) distributed over various serve as general guidelines in practical design. For example,
areas (different, U). However, it should be noted that the loaded considering a shell (R and h are given) under load P acting over
area must be kept relatively smaller than the spherical cap so that area of radius r0, one can always determine the corresponding
the edge effects introduced by using spherical cap theory could be maximum deflection from Fig. 3. With this information avail-
neglected. able, the maximum stress can be readily determined from Fig. 4.
Two cases of X values of 7.45 and 14.82 are examined. Again, On the other hand, if the design is required to estimate the shell
differences in the solutions between these two X values are about thickness, a trial-and-error process may be involved because of
of the order of errors as mentioned earlier; therefore, only the the shell thickness presented with different order in different scale
results of X = 7.45 are reported and presented in Figs. 3 and 4. parameters.
In Fig. 3, the curves of the external loadings versus the maxi-
mum deflections are plotted for different loading areas of U Discussion and conclusions
ranging from.0.2 to 2.2 (this is the range of practical interest); h/[arguerre equations governing the axisymmetric response of
the maximum deflection is up to an order of one shell thickness. shallow spherical shells are utilized to estimate the local stresses
From this figure it is demonstrated again that for a given magni- in a spherical shell under external loading. Numerical solutions
tude of externaI load the corresponding maximum deflection is are obtained using a nonlinear relaxation technique to solve the
higher when the load is distributed in a smaller area. finite-difference equivalents of shell equations. The present
Figure 4 exhibits the curves of the maximum stresses versus solution in the small-deflection range appears to be in good agree-
loaded-area parameters in a semilogarithmic scale for different ment with the result obtained in reference [6] (see Fig. 4).
levels of the maximum deflections. Displayed in the same figure In order to assess the validity of the general computer program
is the result obtained by Wichman et al. [6], which actually is a developed, a particular geometry of spherical cap under local
summarized result based on Bijlaard's work of the linear shallow- loading is examined. Solutions obtained herein compare very
shell theory. A good correlation is observed between their result favorably with results in the literature [9].
and that of Wma~ = 0.1 obtained here for U values of between 0.2 For the purpose of demonstrating the highly localized phenom-
and 1. enon of the maximum stress in the spherical shell under local
From Fig. 4 it is apparent that for higher load or deflection loading, a shell is chosen such that R, h, and ro are fixed (say U
levels the ratios of the maximum stresses (a .... ) to loads (P) tend = r o / ~ v / ~ = 0.22) and from which, four different sizes of small
to be smaller; this suggests that in cases of higher load levels the portion (spherical cap) are selected. In each cap geometry, solu-
linear (or small deflection) theory will be too conservative to be tions are obtained (with U = 0.22) for different values of P*
M A R C H 1973 21
(up to 5) ; the deviations among them are of the order of less than going discussion, an important relation between loading area and
2 percent. This indicates that the maximum stress in the spheri- cap geometry parameter, U and X, may be taken as U/X < 0.27.
cal shell due to local load can be quite accurately estimated from For a more accurate analysis, a complete shell theory is de-
shallow-shell theory if the load is confined to a relatively small sirable, but much difficulty in the mathematical aspects will be
area. encountered.
Solutions are then obtained for )~ = 7.45 and 14.82 for different The effect of the highly localized maximum stress may be
loaded area parameters and load levels; again, differences be- minimized by providing a reinforcing pad in the spherical vessel
tween these two solutions are negligible. Results presented in or by increasing the contact area between the cable and sphere.
Fig. 3 are the local loads versus the maximum deflections for the
different load area parameters U ranging from 0.2 to 2.2; results
References
displayed in Fig. 4 are the maximum stresses versus load area
parameters for different levels of the maximum deflections Wm~ 1 Laura P. A. and Casare]la, M J., "A Survey of Publications
on Mechanical Cables and Cable S:stems, Rept. 68-1, Institute of
with values ranging from 0.1 to 1. Ocean Science and Engineering, The Catholic University of America,
It is hoped that results presented in Figs. 3 and 4 can readily be Washington, D. C.
used in practical design. For example, if the shell geometry 2 Casarella, M. J. and Parsons, M., "A Survey of Investiga-
(R and h) and loading conditions (P and r0) are given, then the tions on the Configuration and Motion of Cable Systems under
Hydrodynamic Loading," Marine Technology, Vol. 4, No. 3, July
corresponding maximum deflection can be determined from Fig. 1970.
3. With this information at hand, the associated maximum 3 Bijlaard, P. P., "Computation of the Stresses from Local
stress can be readily obtained from Fig. 4. However, if the shell Loading in Spherical Pressure Vessels on Pressure Vessel Heads,"
thickness is desired in the design, then the solution will require a Welding Research Council Bulletin No. 34, March 1957
4: Bijlaard, P. P., "Stresses in a Spherical Vessel from Radial
short effort of trial and error. Load Acting on a Pipe," Welding Research Council Bulletin No. 49,
Again it should be emphasized that for smaller U values (smaller 1-30, April 1959
loaded area) the results obtained from shallow-shell theory 5 Bijlaard, P. P., "Influence of a Reinforcing Pad on ~he Stresses
could provide a good approximation to the maximum stresses in in a Spherical Vessel Under :Local Loading," Welding Research
Council Bulletin No. 49, 63-73, April 1959.
a spherical shell under local loading, since edge effects of spherical 6 Wiehman, K. R., Hopper, A. G., and Mershon, J. L., "Local
caps would become negligible, and the response of the full shell is Stresses in Spherical and Cylindrical Shells Due to External Load-
indeed a highly localized phenomenon. In fact, most cable sys- ings," Welding Research Council Bulletin, No. 107, Aug. 1965.
tems in the area of current interest possess small U values, be- 7 Reissner, E., "Stress attd Small Displacements of Shallow
Spherical Shells," J. Math. Phys., Vol. 25, 1946, pp. 80-85 and pp.
cause the size of the cable is relatively much smaller than that of 279-300.
spherical structure. 8 Ashwell,D. G., "On the Large Deflection of a Spherical Shell
For the sake of completeness, it may be advisable to discuss with an Inward Point Load," Proc., I.U.T.S.M. Syrup. on the Theory
briefly the limitations regarding key parameters involved in the of Thin Elastic Shells, Delft, August, 1959; North Holland Publi-
cation, Amsterdam, 1960, pp. 43-63.
computation. A shell is called "shallow" if the ratio of its rise 9 Penning, F. A. and Thurston, G.A., The ~tabfllty of Shal-
at the center to the base diameter is less than, say 1/8; useful low Spherical Shells Under Concentrated Load," NASA CR-265,
values for the shallow-cap geometry parameter, X, associated 1965
with this limitation are in the range of 4 to 15. In addition, 10 Fitch, J. R. and Budianskv, B., "The Buckling and Post-
Buckling Behavior of Spherical Caps Under Axisymmetric Load,"
the assumption of the localized nature of the maximum stress AIAA Jourt+al, Vol. 8, No. 4, April 1970, pp. 686-693.
field obtained by utilizing shallow-cap theory will be valid if the 11 Perrone, N. and Kao, ~., Large Deflection Response and
load radius, r0, is significantly less than that the cap, a; a realistic Buckling of Partially and Fulty Loaded Spherical Caps," A I A A
ratio between these two quantities is suggested as ro/a <_ 0.5. Journal, Ve]. 8, No. 2, Dec. 1970.
12 Perrone, N. and Kao, R., "A General Nonlinear Relaxation
Of course, the smaller the value of ro/a, the better the expected Technique for Solving Nonlinear Problem in Mechanics," Journal
accuracy of the maximum stress. In accordance with the fore- of Applied Mechanics, Vol. 38, No. 2, June 1971, pp. 371-376.