Sunteți pe pagina 1din 9

Composite Stmctures 32 (1995) 187-195

0 1995 Elsevier Science Limited


Printed in Great Britain. All rights reserved
0263-8223/95/$9.50
ELSEVIER 0263-8223(95)00072-O

Impact dynamics of composite beams

A. S. Yigit & A. I! Christoforou


Department of Mechanical and Industrial Engineeting, PO. Box 5969, Safat 13060, Kuwait

The dynamics of a composite beam subject to transverse impact is


investigated. A linearized contact law based on an elastic-plastic contact is
shown to yield excellent results for the impact response. A dynamic ratio
which is defined as the ratio of the maximum impact force obtained from
the dynamic simulation to the one obtained from a half-space analysis (i.e.,
local contact) is used to characterize the type of impact response, i.e.,
whether it is locally dominated, quasi-static or dynamic. It is found that this
ratio depends on a single nondimensional parameter called ‘dynamic impact
number’, which also governs the initial impact response until the waves are
reflected back from the boundaries.

INTRODUCTION Based on the characteristics of the impactor


and the structure, the impact response can vary
A major issue of great concern with fiber-rein- in nature, i.e., quasi-static, half-space or
forced composite structures is that they have dynamic. The nature of the impact response, in
low damage tolerance to out-of-plane loads turn, influences the type of damage and the
such as those resulting from lateral impact by extent of structural degradation. Therefore, the
foreign objects. Accurate prediction of the identification of the physical parameters or
impact response is an important first step in properties that determine the nature of the
assessing the composite material’s sensitivity im act response is very important. Bucinell et
due to impact. Impact of composite structures is al.F used a lumped parameter model to inves-
a complicated event which includes physical tigate the type of impact response. They
phenomena such as structural response, contact concluded that it is the ratio of the local contact
and wave effects. The relative importance of frequency to the structural frequency of the
these effects depends upon the characteristics target that determines whether the response is
of the impactor and the structure. quasi-static or dynamic in nature. Recently,
The contact models used in most impact SwansonlO has developed a procedure for
studies are traditionally based on the Hertzian establishing the limits of the quasi-static approx-
contact law.le7 The experimental contact law imation for the calculation of the impact
obtained by Yang & Sun,4 which is widely used response of structures. The procedure was
was the first attempt to include damage effects based on the calculation of an equivalent
into the contact law. Recently, Yigit & Christo- lumped mass for the structure and considers the
forou’ proposed an analytical contact law, response quasi-static if the impactor mass is
which includes damage effects by using the elas- greater than 10 times the lumped mass.
tic-plastic indentation of a transversely-isotropic Although the above mentioned studies are
half-space. Due to the nonlinearity of the con- useful to assess whether the response is quasi-
tact law an analytical solution is not generally static, a more comprehensive study is still
possible and a consistent linearization that needed to cover a wider range of impact
works for a wide range is not straightforward. situations. In the current study, effective
The current study uses a linearized version of management of the parameters that determine
the elastic-plastic contact law developed in Ref. the type of impact response is achieved through
8. It is shown that the proposed linearization normalization of the governing differential
provides an excellent approximation to the non- equations. It is shown that a single parameter
linear solution. termed ‘dynamic impact number’, which is a
187
188 A. S. Y&it,A. P Christoforou

grouping of all the pertinent impact parameters, Phase I: Elastic loading


governs the initial impact response and deter-
F(a) =Kha3’2 Os~q, (6)
mines whether the response is locally domi-
nated, quasi-static, or dynamic in nature. Phase II: Elastic-plastic loading
F(a) =K,(~-MJ+K,&‘~ g,sa<a, (7)
ANALYSIS Phase III: Elastic unloading
F(a) =Kh(c13’2-~~2+a~‘2)+Ky(am-~) (8)
Governing equations
where CIis the local indentation defined as the
The equations of motion for a specially ortho- difference between the displacement of the
tropic laminated beam subject to lateral impact impactor and the deflection of the beam at the
including transverse shear deformation are contact point, i.e.,
given as
4) = r%(t) -W(&, t), (9)
Kh is the Hertzian stiffness, ay is the critical
bD aZICl+ kbA55
l1 ax2 indentation for ‘local yield’ to occur, and KY is
the linear stiffness of the elastic-plastic loading
phase. The Hertzian stiffness, &, depends on
kbAss($-s)+p(x.t)=pA$ (2) the material properties and the contact geome-
try, and in the case of a spherical body of radius
R in contact with a flat surface is given in Ref.
where Dll is the bending stiffness and As5 is the
11 as
shear stiffness defined as usual, b, A and I are
the width, cross-sectional area, and moment of & =$&* (10)
inertia of the beam, respectively, p is the mate-
where E* is given by
rial density, w is the transverse deflection, I,$ is
the shear rotation, p is the lateral load per unit 1 1-v: 1
length, k is the Mindlin shear correction factor, -=-
(11)
and x and t are the space and time variables. E” J% +E Z
The impact force can be obtained from the
where v1 and El are the Poisson ratio and
deceleration of the impactor and given as
Young’s modulus for the impactor, respectively,
UZiW;iti
= -F (t) (3) and & is the transverse Young’s modulus of the
laminate, which is usually taken to be equal to
where mi and wi are the mass and the displace-
that of the ply, i.e., E, = Ez2.
ment of the impactor, respectively, and F is the
The critical indentation LX_,,
can be obtained
impact force.
from the contact stress distribution using a max-
The initial conditions of the impact problem
imum shear failure criterion, and is given by
are
w(x, 0) = 0 ti(x, 0) = 0 2.72S;7c2R
(4)
(12)
E *2
5J=
Wi(O) = 0 tii (0) = 00 (5)
where u. is the initial impact velocity. where S, is the shear strength of the laminate.
The slope for the elastic-plastic loading phase
Contact law KY is equal to the slope of the elastic indenta-
tion curve at a = ay and is given as
For low velocity impacts, the contact models
KY = 1*5K, 6 (13)
used in the analysis are traditionally based on
the Hertzian contact law.5T” In this study a line- Due to the nonlinearity of the contact law an
arized form of the contact law developed in analytical solution is not generally possible and
Ref. 8 from the elastic-plastic indentation of a a consistent linearization that works for a wide
transversely-isotropic half-space is used. For range is not straightforward. One method of
completeness a brief description of the contact contact stiffness linearization is to calculate an
law follows: equivalent linear stiffness for a single degree-of-
Impact dynamics of composite beams 189

freedom lumped model which results in the malized equations and initial conditions:
same maximum contact force.’ Since the linear
stiffness depends on the maximum force expec- a21j aw
ted, this linearization may not be adequate for
situations where the details of the contact law
-+y2
ai ( --g-$
>=o (16)

are important. In addition, in cases where mul-


tiple impacts occur, a new linearization is (17)
required for each impact. Another method,
when permanent deformation is present each
phase of the contact law may be linearized sep- a2ki&, 7) a2a
arately with very good results.12 For most +- = -P (7) (18)
aT2 ar2
composite materials, however, damage occurs
early in the loading phase, and most of the local Pi+, 0) = 0 bqf,O)=O (19)
response is dominated by the second phase of
the contact law.8 Therefore, it will be worth-
cc(O)= 0 i(0) = 1 (20)
while to consider only the elastic-plastic phase where
which is already linear.
The linearized contact law can then be
written as (21)

F (a) =&cl (14) are the governing nondimensional impact para-


meters, and m is the mass of the beam. The
nondimensional force is given as
Normalization

E (7) = ~
F (9
Reliable assessment of structural degradation
i?Zi 2loO.l
(22)
due to impact requires accurate modeling of
impact event. The nature of the impact
the shear and bending related frequencies are
response can influence the type of damage and
defined as
the extent of degradation. However, the identi-
fication of the physical parameters or properties
2 _ kb&
that determine the type of the impact response cos --
pA12
(23)
has been difficult. Effective management of
these parameters in nondimensional form
would therefore be very useful. In this study,
this is achieved by normalization of the govern- (24)
ing equations.
The governing equations are normalized by and the normalized force per unit length is
defining the following nondimensional variables given as
as:
P(XY t)l
X
p(q 7) = . (25)
x=---, l?ZiVoCO
z = cot
1
Finally, the normalized linear contact law can
be written as
(15) P(a)=6 (26)
It can be shown that the three nondimen-
where I is the beam length, w = &,/mi is the
sional parameters II, ,Uand y are related by
contact frequency and a” = u,,/o is the maxi-
mum indentation obtained from the half-space
solution (i.e., with no structural response). p=zE (27)
Substituting the nondimensional variables in Y
eqns (l-5) along with eqns (9) and (14), and
where
neglecting rotary inertia yield the following nor-
190 A. S. Y&it, A. I? Christoforou

tions, (‘) denotes the spatial derivative, and


&= (12/y (Ep)(sJ
(28) qn(r) and m(~) are the corresponding modal
amplitudes. For simplicity and without loss of
and y is given in terms of physical para- generality the present paper deals with a simply
meters by supported beam of rectangular cross section.
A standard modal analysis procedure using
the above modal expansion yields
y = (12k)“2
W2CJ
(29) &&)+C&,(z) = &(z) ?z= 1, 2, 3, . . . (33)

where G, and E, are the transverse shear and where Co, are the normalized natural frequen-
bending modulus, respectively, and h is the cies of vibration given as
thickness of the beam.
A further re-grouping of these nondimen- (-3 _ A2W4
n- y2+(mc)2
(34)
sional parameters results in a single parameter
called ‘dynamic impact number’ which, as it will
be shown later, governs the initial impact The equation of motion for the impactor is
response until the waves are reflected back obtained by substituting eqn (31) into eqn (18):
from the boundaries, and is given as nE1 &(+$&) +h = -E (2) (35)
Assuming a concentrated impact loading at the
Dynamic impact number = mid-point, i.e., XC= l/2, the generalized modal
forces are obtained as
It is interesting to note, at this point, that the
dynamic impact number, which contains all the
w-9
physical parameters of the impact event, is in
fact the ratio of the inertia effects represented
by the mass ratio, CL,to the stiffness effects rep- For an N mode approximation, eqns (33) and
resented by s/y. As it is known, a small mass (35) constitute a set of (N+ 1) linear ordinary
impact of a fixed target, which is represented by differential equations which can be solved
a small dynamic number, will result in a local- analytically (e.g., Laplace transform13) or by
ized type of response. On the other hand, a numerical integration. In this study, the equa-
large mass impact, which is represented by a tions are solved by the latter method for the
large dynamic impact number, will result in a ease of implementation.
quasi-static type of response. The opposite is
true for the stiffness. In what follows the utility Half-space and quasi-static approximations
of the dynamic impact number in providing
guidelines on the type of impact response will In order to gain more insight to the problem
be demonstrated. two extreme cases are worthwhile to consider.
Namely, the half-space and the quasi-static
approximations. If the impactor mass is very
SOLUTION PROCEDURE small, and the target is fairly stiff, the impact
does not produce significant structural response
In order to discretize the equations of motion and can be approximated by impact on a half-
the displacement and the rotation of the beam space. On the other hand, if the impactor is
are represented by series expansions using the quite heavy the inertia effects of the impactor
normal modes of the associated eigenvalue will dominate the response and can be approxi-
problem as mated by a quasi-static analysis.
In the case where the vibration of the beam is
J+ = azl 4&)M) (31) negligible, the impact response is localized in
nature (i.e., half-space approximation) and the
(32) impact force can be obtained from eqn (35) by
neglecting the structural response as
where & are the normalized mode shapes of
the beam under appropriate boundary condi- Fh.9= miuOm sin ot (37)
Impact dynamics of composite beams 191

The quasi-static approximation involves treat- the beam the impact response can be approxi-
ing the impact problem as an equivalent static mated by the so-called quasi-static approach.‘,”
problem with a static load applied to the impact These types of impact behavior were investi-
site. By neglecting the mass of the beam the gated further by carrying out more simulations
system can be thought of as a single degree-of- using the nondimensional parameters obtained
freedom system with the contact (KY) and static earlier. For these simulations the material
stiffness (KS,) of the beam in series. The impact system was kept the same, (i.e., E is constant),
force can then be obtained as and the beam thickness and the impactor mass
were varied resulting in different y and ~1.
Fqz = mivooqS sin uqst (38) Figures 4 and 5 show the impact response of a
where thick beam with h = O-016 m impacted by a
heavy mass, mi = 15 kg, which result in
K&y
wp =
J (39)
p= 148.3 and y = 10. As can be seen Fig. 4
represents a typical quasi-static response. The
beam deflection (Fig. 5) however, differs from
is the frequency of the response. the quasi-static approximation in the magni-
tude. This is due to the local indentation, which
is not negligible because of the large beam stiff-
RESULTS AND DISCUSSION ness.
The impact response of a thinner beam with
In the following simulations a composite beam a similar mass ratio, p = 158, but with a larger
impacted by a steel spherical impactor was y = 40 is given in Figs 6 and 7 where h = 0.004
used. The material properties, geometry of the m and mi = 4 kg. As is seen this case can also
beam and the impact conditions are given in be represented by a quasi-static approximation.
Table 1. Unless otherwise stated these condi- The beam deflection and the impactor displace-
tions represent the nominal case of this study. ments follow closely the quasi-static solution,
In order to investigate the degree of accuracy which shows that in this case the local indenta-
of the proposed linearization of the contact law
simulations were carried out for the nominal
case using both nonlinear and linear contact
laws. As can be seen in Figs 2 and 3 the simula-
tion results using the linear contact law provide
an excellent comparison with the simulation
results using the nonlinear contact law. Figures
2 and 3 indicate that the impact response is
‘dynamic’, i.e., it is a combination of local and
structural effects. Note that the impact force
profile does not follow the deflection of the
beam.
When the impactor mass is much smaller
than the mass of the beam and/or the beam is
very stiff the impact response will be dominated
only by the local effects, i.e., half-space approx-
imation.14 On the other hand, when the mass of % Indentation nf

the impactor is much larger than the mass of Fig 1. Contact law.

Table 1. Properties of the composite beam and the impact parameters

Beam: [O/90/0/90/0], T300/934 carbon-epoxy, simply supported


beam span: 200 mm
beam width: 20 mm
beam thickness: h = 2.69 mm (0.269 mm/layer)
El1 = 120 GPa, Ez2 = 7.9 GPa, Glz = Gz3 = 5.5 GPa
q2 = vz3 = 0.3, p = 1580 kg/m3, S, = 101 MPa
Impactor: 12.7 mm diameter spherical, p = 7960 kg/m3, steel mass = m = 8537 g, impact velocity = 3.0 m/s
192 A. S. Y&it, A. l? Christoforou

16.0-

0 50 100 150 200 250 300 350 400 0:o 0.002 0.604 O.dO6 0.008 0
Time,(s)
'lo6
Time,(s) Fig. 4. Comparison of the impact force (h = 0.016 m,
mj = 15 kg).
Fig. 2. Comparison of the impact force for the nominal
case.

0.009

'105
2.0

1.5

E
i
p
9 1.0
0"
E - - - kllpactsr
asp.
:
al

0.5

0.0 0.002 0.004 0.006 0.006 C


Time,(s)
0.0
0 500 1000 1500 2000 Fig. 5. Comparison of the beam deflection and impactor
'lo= displacement (h = 0.016 m, mi = 15 kg).
Time,(s)

Fig. 3. Comparison of the beam deflection for the


nominal case. desirable to have a more inclusive quantitative
measure to characterize the type of impact
response.
tion is negligible. However, the quasi-static Swanson” used the ratio of dynamic to quasi-
solution provides an approximation for the static solution as a function of ratio of the
impact force in an average sense. It is seen that impact mass to equivalent lumped structural
due to the excitation of higher modes a series of mass to investigate the limits of the quasi-static
distinct impacts can take place. More distinct solution. In this case, a ratio close to unity
impacts are expected as the thickness decreases. would represent a quasi-static type response.
For such systems what is normally regarded as a Although this ratio is adequate to assess
single impact is, in fact a series of impacts in whether a quasi-static solution is applicable for
rapid succession? Although for very large a certain impact situation it is certainly not
impact mass, it might suffice to specify the mass enough to distinguish different types of impact
ratio to characterize the type of impact responses. Another possible measure, proposed
response, it is clear that the stiffness of the here, is the ratio of dynamic to the half-space
structure plays a major role as well. Thus, it is solution, and is called ‘the dynamic ratio’. In
Impact dynamics of composite beams

0.015 10’ IO’ lo3


iOO 10’
Time, (s) Mass Ratio

Fig. 6. Comparison of the impact force (h = 0.004 m, Fig. 8. Effect of mass ratio on the dynamic ratio for two
mi = 4 kg). different beams.

0.025-

z
4 0.02-
6
@I
u
go:o.or5-
E
-E
; o.oi-
g
al

0.005-

0.0 y v
0.0
I
0.005
I
0.01
I
0.015
I
0.02
I
0.025 0.03
-1
0.5 1.0 5.0 10.0 50.0 100.0
Time, (s) Dynamic Impact Number

Fig. 7. Comparison of the beam deflection and impactor Fig. 9. Effect of dynamic impact number on the dynamic
displacement (h = O-004 m, mj = 4 kg). ratio for two different beams.

general, the impact response contains multiple can be seen both nondimensional parameters ,U
impacts or impact forces with multiple local and y affect the type of response, which leads to
maxima. In this study the dynamic ratio is the conclusion that the mass ratio is not enough
defined as the ratio of the first local maximum to characterize the type of response. Note that
in the impact force history obtained from the the material properties which have not been
dynamic simulation to the one obtained from varied in these simulations will also affect the
the half-space solution (i.e., local contact). response.
A series of simulations were carried out to The dynamic impact number defined earlier
obtain Fig. 8 where both beams of Figs 4-7 are can now be used since it is a grouping of all the
impacted by different masses. In these cases a pertinent parameters. Figure 9 shows the results
dynamic ratio close to unity signifies a locally of all the simulations of Fig. 8 using the
dominated response (e.g., single impacts), dynamic impact number instead of the mass
whereas a small dynamic ratio signifies a more ratio. It is seen that the two curves of Fig. 8 are
complicated local-global response. As the ratio now mapped into a single curve. Potentially,
approaches zero the response approaches the many impact situations can be represented by
case where a quasi-static analysis is valid. As each point on this curve. It is clear that the
194 A. S. Yigit, A. I? Christoforou

3.0
dynamic impact number alone characterizes the
type of response, i.e., quasi-static, dynamic,
locally dominated, single or multiple impacts. 2.5-

For instance, based on the numerical experi-


ments performed, it is found that a dynamic .$2.0-
P
number less than about 0.8 results in single %
n
impacts which are locally dominated. For u1.5-
=w
dynamic numbers between 0% and 3-O there
E
exist multiple distinct impacts. More compli- q1.0-
cated impact responses (i.e., impact forces with
multiple local maxima) are obtained for 0.5- . . 1___-__
dynamic impact numbers between 3 and 20. For *.
. .
dynamic impact numbers greater than 20 the 0.0 I I I I I I
impact response becomes quasi-static in nature. 0 1 2 3 4 5 6 7
Since the multiple curves of Fig. 8 are Normalized Time

mapped into a single one of Fig. 9, some Fig. 11. Dynamic similarity of normalized beam deflection
dynamic similarity should exist for the same for two different beams with the same dynamic impact
number.
dynamic impact number. In order to demon-
strate this similarity the normalized impact
force and deflections for the two different
beams (Beam 1: h = 0*004 m, Beam 2: h = O-016 and stiffness effects for the whole system and
m), but with the same dynamic impact numbers determines the type of response.
(0.852) are shown in Figs 10 and 11. The impac-
tor masses were 0.005 kg and 0.080 kg for Beam
1 and 2, respectively. As expected there are CONCLUSIONS
multiple but distinct impacts for both beams,
and dynamic similarity exists for the first The dynamic behavior of a composite beam
impact. It can be concluded that a dynamic sim- subject to transverse impact has been investi-
ilarity is valid until the waves are reflected back gated. A linearized contact law based on an
from the boundaries. Thus, the dynamic impact elastic-plastic contact was used which yielded
number governs the initial impact response and excellent results for the impact response. The
is conceptually similar to Reynolds number in effect of impact parameters on the type of
fluid mechanics in that it compares the inertia impact response has been examined through
numerical experiments with normalized equa-
tions. Normalization of the governing equations
leads to three nondimensional parameters
which characterize the whole impact event. Fur-
ther regrouping of these parameters resulted in
a single parameter called ‘dynamic impact num-
ber’, which governs the initial impact response
until the waves are reflected back from the
boundaries. Furthermore, it has been shown
that for the same dynamic impact number
dynamic similarity exists. The dynamic impact
number is conceptually similar to Reynolds
number in fluid mechanics in that it compares
the inertia and stiffness effects for the whole
system, and determines the type of response,
(i.e., quasi-static, dynamic, locally dominated,
0 1 2 3
single or multiple impacts). The type of
4 5 6 7 a
Normalized Time response was represented by the dynamic ratio
which is defined as the ratio of the maximum
Fig. 10. Dynamic similarity of normalized impact force for
two different beams with the same dynamic impact impact force obtained from the dynamic simula-
number. tion to the one obtained from a half-space
Impact dynamics of composite beams 195

analysis (i.e., local contact). It has been found plates. .I. Appl. Mech., 52 (1985) 6-12.
6. Sun, C. T. & Chen, J. K., On the impact of initially
that this ratio depends only on the dynamic stressed composite laminates. .I. Comp. Mat., 19
impact number. (1985) 490-504.
7. Cairns, D. S. & Lagace, P. A., Transient response of
graphite/epoxy and kevlar/epoq laminates subjected
to impact. AZAA J., 27 (1989) 1590-96.
ACKNOWLEDGEMENT 8. Yigit, A. S. & Christoforou, A. P., On the impact of a
spherical indenter and an elastic-plastic transversely
This work was supported by Kuwait University isotropic half-space. Comp. Engng, 4 (1994) 1143-52.
9. Bucinell, R. B., Nuismer, R. J. & Koury, J. L.,
Research Administration, under Project No. Response of composite plates to quasi-static impact
EM-089. events. In Composite Materials: Fatigue and Fracture,
(Third Volume), ASTM STP 1110, ASTM, Philadel-
phia, PA, 1991, pp. 528-49.
10. Swanson, S. R., Limits of quasi-static solutions in
REFERENCES impact of composite structures. Comp. Engng, 2
(1992) 261-67.
1. Sun, C. T., An analytical method for evaluation of 11. Johnson, K. L., Contact Mechanics, Cambridge Uni-
impact damage energy of laminated composites. In versity Press, Cambridge, 1985.
Composite Materials: Testing and Design (Fourth Con- 12. Yigit, A. S. & Christoforou, A. P., On the impact
ference), ASTM STP 617, ASTM, Philadelphia, PA, between a rigid sphere and a thin composite laminate
1977, pp. 427-40. supported by a rigid substrate. Comp. Struct., 30
2. Dobyns, A. L., Analysis of simply-supported ortho- (1995) 169-77.
tropic plates subject to static and dynamic loads. 13. Christoforou, A. P. & Swanson, S. R., Analysis of
A&I/l J., 19 (1981) 642-50. impact response in composite plates. Znt. J. Solids &
3. Greszczuk, L. B., Damage in composite materials due Strut., 27 (1991) 161-70.
to low velocity impact. In Impact Dynamics, John 14. Poe, C. C., Jr. & Illg, W., Strength of a thick graphite/
Wiley and Sons, Inc., New York, 1982, pp. 55-94. epoxy rocket motor case after impact by a blunt
4. Yang, S. H. & Sun, C. T., Indentation law for com- object. In Test Methods for Design Allowables for
posite laminates. In Composite Materials: Testing and Fibrous Composites (Second Volume), ASTM STP
Design (Sixth Conference), ASTM STP 787, ASTM, 1003, ASTM, Philadelphia, PA, 1989, pp. 150-79.
Philadelphia, PA, 1982, pp. 425-49. 15. Barkan, P., Impact. In Mechanical Design and Systems
5. Tan, T. M. & Sun, C. T., Use of statical indentation Handbook, 2nd Ed., McGraw Hill Inc., New York,
laws in the impact analysis of laminated composite 1985, pp. 15.3-15.43.

S-ar putea să vă placă și