Hermman 1967
Hermman 1967
H E R R M A N N
Professor of Civil Engineering,
Response of Cylindrical Sandwich Shells
The Technological Institute,
Northwestern University,
Evanston, III. Mem. ASME
to Moving Loads1
This paper presents an analysis into the dynamic response of a long cylindrical sand-
E. H. B A K E R 2 wich shell under a moving axially symmetric ring load. The shell is assumed to be
Research Specialist,
orthotropic and subjected to an initial axial stress. The uniform velocity of the load is
Structural Sciences,
Space and Information Systems Division, prescribed and only the steady-state response is considered. Numerical results indicate
North American Aviation, Inc., the effects of various relevant parameters. The behavior of orthotropic sandwich cylinders
Downey, Calif. under initial stress is compared with tluit of homogeneous isotropic cylindrical shells
free of initial stress, and differences are pointed out.
Mi refers to the axial inertia and M to the radial inertia. elastic range, T would be of the order of magnitude of 1 0 - 2 or
smaller.
The form of the roots of equation (16) may easily be found for
General Solution a particular value of G by finding in which region the velocity
Fourier transforms will be used to determine the displacements parameter is located. The form of the roots in each region is
from equation (9), i.e., given in Table 2.
For the special case in which A2 and G satisfy the equation
/(s) /(r))e- i s 'd7 ) (10) A = 0 for et ^ 0, equation C(s) = 0 has equal roots. If the equal
roots are imaginary, the deflection is well defined. If the equal
roots are real, a condition of instability exists which will be dis-
.f(s)e ,s Vs (11) cussed in a later section.
'(i?)
REGION 1
r REGION II CURVE 1
R e g i o n II
The roots are of the form s = ±ji-f-im; ± n — im, and the
radial deflection is
REGION III
IF c~mi
[A' 3 n sin nrj + N2U cos nr\]H(r))
So ' Nin
TF
Table 2 Type of roots of the characteristic equation — = A V v sin » i r j H ( - r i ) + AV V sin ntyH(r]) (20)
So
Region Roots of C(s) = 0
I ± imi, ± im2 where
II ± n + im, ± ii — im
III ± imi, ± im2 CI' - C2'ih2
IV ± ill, ± Jl2 AriIV =
V ± im, ± n Mi(ni2 — n22)
VI ± inii, ± im2 Ci - C\'n22
VII ± im, ± ii AV V = fll > 0 >ln > 0 11-2 > III
VIII ± iii, ± n2 nzini2 — ?io2)
Region V
similar manner. Velocities greater than (E0 — v'2)/E0M will be In region V, two of the roots are on the real axis if clamping is
excluded from this investigation because they do not appeal- not present. For a small amount of damping, the roots have
practical and because the original assumption of steady-state imaginary parts. As the damping approaches zero, the deflection
response becomes questionable. Therefore, the discussion of the becomes
displacements in regions VI, VII, and VIII will be omitted.
]V_
= [ A W * + A V sin ni)]H{-rt) + Nive-mm(-q) (21)
R e g i o n s I a n d III So
The roots of C(s) = 0 in regions I and III are of the form s =
± imi, ± im2. The inverse transform was obtained by ex- where
panding W/qo into partial fractions and inverting the individual Ci + m2C2' Cj' - W2C;'
terms by transform (11). Ar,v = AV
2m(n2 -f- m2) n(n2 + m2)
W
— = [A'iIc-m" + N2Ie~m"l]H(i]) Equal I m a g i n a r y Roots
So
If X2 and G satisfy the equation of curve 1 or curve 2 and e2 > 0,
+ W s ' " " + jVi'e""']^-))) (IS) equation (16) has equal imaginary roots. It may be shown that
where ro is always greater than zero on curve 1. If terms of the order
h/R are neglected in comparison to 1, it may also be shown that
G(1 + T — Af,X 2 ) e2 > 0 on curve 2 if G < lfh{Ev - v'2)1/x. The equal imaginary
C,' =
(G + T - M\2)(Di - 7 r X 2 )( 1 + T - Di + Ir\2 - il/iX 2 ) roots are of the form s = ± im, ± im, and the radial deflection is
For parameters of physical significance, we find that the rota- and G > Dx/\Eo - c' 2 )' 7 '. Therefore, the first critical velocity
tory and axial inertia have only a small effect on the two lowest
roots of equation (25). If rotatory and axial inertia are neg- A^
lected, equation (25) becomes
-h[ 2A'A(#o - c'2)'/! + T - —'- (Eo - v'"-)
G
(35)
eg'X4 + e/X 2 + e6' = 0 (26) where G > D'''2(Eo — vInvestigation of the coefficients of
where equations (19) and (20) reveals that the deflection becomes un-
bounded as X2 approaches XCR,2. Inspection of equation (35)
e»' = G2M2 shows that for T approaching To, X c r i approaches zero. There-
2
- G(T - 2v'D1)} Further investigation of equation (25) showed that the real
part of the two higher roots was approximately 1/jl/, which is
e6' = { A [ # o ( l + T - A ) - v'2 - 2v'G] + GTj2 beyond the scope of this investigation.
- 4GDt(G + T)[E0(l + T) - v'2][l + T - A]
Let XCR52 T ) (36)
The roots of equation (26) are
v'A The deflection becomes unbounded whenever X2 approaches
XcRj in regions I, IV, or V. Therefore, XCR 2 2 is a critical velocity.
2
X022 < X 2 < XCR z 2 for G < Dl'\Eo - v' 2 )'^ for
Xoi2 < X2 < X022 for G > Go2 e,' = -XK^EoH + f - il/iX 2 ) - v'2]
e% = -X/sT [ —2f'(£>! - 7rX 2 ) + (1 + f - MiX2)(f - M\ 2 )]
Region III
m
The coefficients ei, e2, Co, and e are the same as the corresponding
X2 < Xo,2 for G > Goi
coefficients of equation (13).
Region IV The form of the equation for the deflected shape depends on the
type of roots of the characteristic equation C6(s) = 0. The types
X022 < X 2 < XCR, 2 for X 2 < XCR, 2 , and G > Dl/\Eo - F'2)1/! which may be encountered and the corresponding deflected shapes
Region V are given in reference [6],
If Kf, is very large, equation (40) becomes the transform of the
Xcn22 < X2 < XCR,2 radial deflection of a cylinder without transverse shear deforma-
tion. For the special case of a homogeneous isotropic cylinder
Double Imaginary Rools
without initial stress and neglecting rotatory inertia (I, — 0), the
X2 = Xoi2 or X2 = X022 where G < Dl/KE0 - i''2)7' transform of the radial deflection for very large Ii,, agrees ex-
actly with the corresponding result presented in reference [2]
Critical Velocities using Fliigge's equations. It was also found that the expression
for the undamped deflected shape in region IV is the same as
XCR,2 = X022 where G > D'\Eo - i>' 2 )' /!
equation (20) if the method of diminishing shear damping is
used.
XcRj2 = 77— XcRj2 "
M EoM
Numerical Examples
Homogeneous Isotropic Cylinder Numerical results were obtained to determine the accuracy of
The velocities of special interest for a homogeneous isotropic some of the approximate expressions and to establish the effect
cylinder are of transverse shear deformations, as well as transverse shear
damping on the radial deflection. The effects of initial stress
Eh were also investigated. Only an isotropic sandwich shell with
for = — equal facing sheets was considered. The following parameters
Port[3(1 - f*)]'/• + poh
were used in Figs. 1 through 5:
Eh T
CCRI + po/;.
PoB[3(l - r2)]1''- ^ -T (38) lh Ee E
0.1 v' = 0.3 E„ =
h Ex1 1 - v2
E E_
CCR; = CC R , '
Po 2(1 + v) Po Fig. 1 shows the curves that divide the first quadrant of the
X2 - G plane into regions for h/R = 1/300, f = 0.004. The
They are associated with the dimensionless velocity parameters parameter rp was 0.02, and it was assumed to be independent of G.
XOI, X C R I , XCR,, and XCR,, respectively; v is Poisson's ratio, while The equation of each curve is given in Table 1. Curves 1, 2, 3,
E is Young's modulus. Terms of the order h/R and T were and 4 represent X I , X 0 2 , XCR-,2, and XCRJ2, respectively, as a func-
0 2 2
neglected in comparison to 1. It can be seen that elastic initial tion of G. Curve 2 is a plot of XCR, when G > h/R)( 1 - v'2)^'.
stresses can have a great influence on c0i2 and C C R , . 2
Curve 2 is tangent to curve 3 at G ~ i(h/R)( 1 - v'2)'^. When
If T is zero, CCR,2 and COR,2 agree with the results presented in the values of X R and X 0 2 were computed from equation (25) and
0 2
reference [2], provided terms of the order h/R are neglected in compared with the approximate results obtained from equations
comparison to unity. The velocities c0i2 and ccn,- were not ob- (32) and (33), good agreement was indicated. The differences be-
tained in reference [2] because initial stress and shear distortions, tween the approximate velocities and the velocities predicted by
respectively, were not considered. The critical velocity CCR,2 was equation (25) were so small that they cannot be displayed 011 the
obtained for similar problems [3] when the basic analysis in- scale of Fig. 1. The values of G at which X 0 i 2 and X 0 2 are zero 2
cluded shear deformations. were compared with the values predicted by equations (30) and
(31). The diflerences were within 1 percent.
Structural Damping Fig. 2 shows the radial deflection of four sandwich cylinders
subjected to the same ring load moving at a velocity c = (8.1 X
The stress-strain relation for the material of the core is now
l0~3E/piEach cylinder has a different core material. The
assumed to be
remaining parameters are the same for all four cylinders. The
thickness-to-radius ratio is 1/30. The two cylinders for which
J xz — KJxi 2e. (39)
r p is 0.01 and 0.04 have a shear stiffness of G = 0.01, while G = 0.1
for the other two. When G = 0.01 and rp = 0.01, the given
where \x is the material damping coefficient. velocity is in region I and the deflection decays exponentially.
If the only damping in the system is the shear damping repre- For a heavier core, rp = 0.04, but for the same G, the velocity is
sented by equation (39), the transformed radial deflection may in region Y. The maximum deflection for this case is larger.
be written as The deflected shape is sinusoidal behind the load, while decaying
W_ _ C3s2 + Ctis + C5 exponentially in front of the load. The velocity is in region II
(40) when rp = 0.1 and G = 0.1. The deflection decreases exponen-
qo C 6 (s)
tially while oscillating. If G is still 0.1 but rp = 0.3, the velocity
where falls in region IV. The deflected shape is sinusoidal both in front
C, = (£>, - 7rX2)e of and behind the load. It can be seen from Fig. 2 that the type
Ct = -K„M 1 + T - AfiX2)
of core can have an important effect on the maximum deflection
t*Gx and the deflected shape, even if all other parameters are the
C6 = G(l + T - jliiX 2 ) Kp = same.
R(ExRPl)'/>
G - 0.004
10/
X2 • 9.6 x 10"3
h 1
R 30 v o/8
• 0.02
/6
K^ • 6.4 X 10~3 4
/
K = oo
ii 2
«0
o. 2 -a i 0 0.1 0.2
n
Fig. 5 Effect of transverse shear damping on deflected shape for
G = 0.004