Sei sulla pagina 1di 35

International Journal of Impact Engineering 31 (2005) 185219

Multiple impact of beam-to-beam


X. Teng, T. Wierzbicki*
Impact and Crashworthiness Laboratory, Department of Ocean Engineering, Massachusetts Institute of Technology,
Room 5-218, 77 Massachusetts Avenue, Cambridge, MA 02139, USA
Received 2 July 2002; accepted 27 August 2003
Abstract
Based on the rigid-plastic string model, the problem of a beam impacted sequentially by numerous beams
with a high velocity is investigated using the wave propagation approach. Attention is focused on the
response of the stricken beam. The closed-form expressions for deection and tensile strain are obtained in
the double impact case. In the multiple impact case, a general recursion formula for the tensile strain as a
function of the number of the striking beam is derived. By assuming the tensile necking failure mode, the
critical impact velocity to fracture the stricken beam is predicted with the impact number specied.
Alternatively, with the impact velocity given, the critical impact number of the striking beam is determined.
Asymptotic analyses for two limiting cases with innite and innitesimal time interval are performed.
r 2003 Elsevier Ltd. All rights reserved.
Keywords: Multiple impact; Beam-string; High velocity; Failure
1. Introduction
In the report on the September 11th attack released by Federal Emergency Management
Agency (FEMA) in May 2002, a detailed survey of the damage of the airplane to the outer facade
of the Twin Towers was made [1]. Based on photos and video clips, it was determined that 3136
exterior columns of the North Tower were destroyed over portions of a four-story range, Fig. 1.
When a scaled outline of the Boeing 767 is superimposed on the damaged area, it becomes clear
that the gash in the facade of approximately 31 m in length is shorter than the wing span of
47:6 m; Fig. 1. The wing tip was deected by approximately 8 m by the exterior columns during
the impact process. Therefore, both the exterior column and the airplane wing must have
undergone considerable plastic deformation before rupture. As shown in Fig. 2, the structural
ARTICLE IN PRESS
*Corresponding author. Tel.: +1-617-253-2104; fax: +1-617-253-1962.
E-mail address: wierz@mit.edu (T. Wierzbicki).
0734-743X/$ - see front matter r 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijimpeng.2003.08.006
ARTICLE IN PRESS
Nomenclature
2b width of the stricken beams
2b
i
width of the striking beams
c stress wave speed for the stricken beam
E
0
kinetic energy acquired by the stricken beam at the rst impact
DE kinetic energy acquired by the stricken beam at the second impact
h thickness of the stricken beam
h
i
thickness of the striking beams
I transverse momentum provided by the striking beams
I
s
considered momentum transferred to the stricken beam
DI
s
momentum loss in the rst impact
m mass per unit length of the stricken beam
M
0
rigid body mass
m
i
mass per unit length of the striking beams
n impact number
%
N plastic tensile force in the stricken beam
t time
t
0
time interval
v
0
impact velocity
v
cr
critical impact velocity to fracture the stricken beam
v
j
transverse velocity due to the jth impact
v

j
transverse velocity immediately after the jth impact
v
j
transverse acceleration due to the jth impact
Dv
j
velocity increment due to the jth impact
W plastic axial stretching energy of the stricken beam
w transverse deection
w
0
deection slope of the stricken beam
. w transverse acceleration
Dw
0
j
slope increment due to the jth impact
x axial coordinate for the stricken beam
0 @
@x
jump of a given quantity across the wave front
: @
@t
a mass ratio
b wave speed ratio
g
j
normalized slope after the jth impact
d
c
displacement of the impact area
e tension strain
e
max
maximum tensile strain
e
f
fracture strain for the stricken beam
e
n
tensile strain after the nth impact
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 186
ARTICLE IN PRESS
Z dimensionless time interval
y dimensionless parameter
x
j
the jth wave front location
r material density of the stricken beam
r
i
material density of the striking beam
s
0
plastic ow stress of the stricken beams
s
i
plastic ow stress of the striking beam
t time parameter
Fig. 1. Superimposition of a scaled outline of the Boeing 767 on the damaged facade of the north face of the North
Tower.
Fig. 2. Structural component arrangement inside an airplane wing.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 187
arrangement of a modern airplane wing is quite complicated, consisting of open section beams,
ribs, and a skin reinforced by stringers. Hence, the exterior column must have been impacted
multiply by the structural components of the wing when the airplanes crashed into the Twin
Towers.
The immediate objective of the present paper is to develop a mathematically tractable,
computational model for a multiple impact event among plastically deforming and fracturing
beams, where the striking beams represent the structural components of the wing, and the stricken
beam an exterior column of the Twin Towers. At the same time, the analysis is rather general and
can be applied to any beams of solid section made of same or different materials.
According to the FEMA report, the travelling velocities of the airplanes were 210 m=s (North
Tower, WTC 1) and 264 m=s (South Tower, WTC 2), respectively, when the airplanes crashed
into the building [1]. These are an order of magnitude higher than the velocity attained by drop
towers or horizontal sledge facilities. In fact, most of the ne works published over the years by
Jones and his coworkers dealing with mass impact on beams apply to the lower end of the velocity
spectrum up to 20 m=s [27]. In such cases, the bending response is dominant, and the deection
of the beam is smaller than the beam thickness. As the impact velocity of the projectile increases,
axial stretching becomes comparable to bending response. A coupled bending-tension solution
was developed by Yu and Stronge [8]. The solution indicates that when the deection of the beam
exceeds the beam thickness, axial stretching becomes dominant over bending response. All of the
above references deal with the impact problem of rigid mass-to-beam. The interactive impact
problem between two deformable beams was investigated by Yu and his coworkers [9,10], where
the bending response was taken into account while the axial stretching was neglected. Therefore,
their solution also apply to the lower end of the velocity spectrum.
For sufciently high velocity impact, a beam can be simplied as a string with exural rigidity
neglected. The problem of high velocity mass impact on a plastic string was formulated and solved
by Mihailescu et al. [11] using a rigorous shock wave formulation and by Wierzbicki and Hoo
Fatt [12] using the momentum conservation principle. Both solutions were shown to be identical
for an innite beam. No experiments are available in the literature to validate the closed-form
solution, but recently performed numerical simulations conrm the correctness of the approach
[13]. Meanwhile, there are an abundance of test results on projectile impact on plates. The
solution for the problem of mass impact on a beam was extended by Wierzbicki and Hoo Fatt [14]
to the case of a large circular plate. Good agreements of transient deection proles were obtained
with the experiments by Calder and Goldsmith [15]. The theoretically predicted ballistic limits
agree within 10% to the test results by Calder and Goldsmith [15] and the recently published
experiments by B^rvik et al. [16]. Hence, the validity of the wave propagation approach has thus
been rmly established.
Since the September 11th attack, the Impact and Crashworthiness Lab at MIT has been
involved in the study of the airplane wing cutting through the exterior columns of the Twin
Towers. A series of reports/papers was published on the response and failure of two or multiple
impacts over a large range of beam dimensions, materials, and impact velocities. In Ref. [17], the
wing was modeled as a rigid mass due to heavy fuel tanks while the exterior column as a plastic
beam. The critical impact velocity to cut through the columns was estimated. Then, the wave
propagation approach was extended to the single impact case of beam-to-beam [18], in which
both the wing and the exterior column were modeled as two plastically deformable strings. The
ARTICLE IN PRESS
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 188
closed-form solution for the deection prole and tensile strains were obtained. Five fracture
scenarios were identied. Two beams could be of different materials and cross sections.
Depending on the combination of various parameters, either of the beams will fail rst. In the
present paper, we made an assumption that the striking beams fail rst upon impact while the
stricken beam continues to deform. This assumption further restricts the applicability of our
solution, but still a class of beams satisfying the solution is very broad. In a more recent paper
[19], we have treated numerically in the impact of rigid mass-to-beam. This paper quanties the
range of the problem parameters, such as mass ratios, relative impact velocities, etc. for which
through thickness shear failure or tensile failure occurs.
In the present paper, the wave propagation approach is rst used to investigate the double
impact case, where three deforming and fracturing beams are involved. Then, the solution is
extended to a general case where a beam is impacted sequentially by arbitrary number identical
beams. The present paper applies to a restricted range of geometrical parameters and impact
velocities for which the beam indeed be treated as a string, and the deections are larger than the
thickness of the beam. Consequently, the bending phase of the response which is relevant for the
deection less than one thickness is neglected. It is assumed that all of the striking beams fail by
shear plugging immediately upon impact. Attention is focused on the response of the stricken
beam. Of interest is to determine the transient velocity, the deection prole, the tensile strain,
and the critical impact velocity to fracture the stricken beam.
2. Double impact
Consider a stationary plastic beam subjected to double normal impact by two identical plastic
beams moving with the velocity v
0
and separated by the time interval t
0
; Fig. 3. The width and
thickness of the beam are denoted by 2b
i
and h
i
for the striking beams, and 2b and h for the
stricken beam. The subscript i represents the striking beam. The mechanical properties of the
beam are dened by the mass density r
i
and the average plastic ow stress s
i
for the striking
ARTICLE IN PRESS
Fig. 3. Schematic of double impact of beam-to-beam.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 189
beams, r and s
0
for the stricken beam. From the above parameters, one can uniquely dene the
mass density per unit length m
i
2r
i
b
i
h
i
for the striking beams and m 2rbh for the stricken
beam, respectively. The impact conguration is shown in Fig. 4, which also denes the axial
coordinate x in the stricken beam. For convenience, the origin of the axial coordinate is dened at
the edge of the impact area. Due to symmetry of the impact response about the impact area, the
expressions for deection and plastic strain will be given only for the positive side of the axial
coordinate in the following derivation.
The impact velocity is high enough so that the response of the beam is governed by local inertia
and wave propagation. Far-eld boundary conditions are not involved for sufciently long
beams. Due to high impact velocities, the beams are subjected to moderately large deection so
that the bending resistance of the beam becomes insignicant. It is assumed that two inequalities
b
i
=hb1 and b=h
i
b1 are satised. Otherwise, the problem becomes three dimensional and can only
be treated by means of numerical methods.
2.1. Range of validity
Even an elastic impact of three bodies poses a considerable mathematical problem. The same is
even more true for an inelastic impact with possible fracture. Many impact scenarios could
develop depending on geometry and materials of two beams as well as impact velocities. Some
typical impact scenarios are shown in Fig. 5. Both beams can continue deforming plastically
without fracture (Case a), or the striking beam fails in shear while the stricken beam is deforming
without fracture (Case b). It is also possible that it is the stricken beam that fractures and the
impacting beam does not (Case c). The solution in this paper is restricted to the case when the
striking beam fails immediately upon impact whereas the stricken beam keeps deforming.
Now, the deforming beam for which the transient velocity and displacement elds were
determined in the previous publication by the present authors [18] is subjected to the second
ARTICLE IN PRESS
Fig. 4. Impact conguration denoting geometrical and mechanical parameters for the stricken and striking beams.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 190
impact. It is assumed that the second striking beam also fails in shear (Case d) and the history of
the deformation of the stricken beam is the subject of the present paper.
The question arises if the sequence of deformation mode to be treated in the paper is possible in
practical situations. To prove this point, we are showing results of numerical simulation of rigid
mass-to-beam impact that could generate three different deformation and failure modes, see Fig. 6
[19]. For low impact velocities below 100 m=s no failure of the beam takes place. In the
intermediate range of the impact velocity (100180 m=s), the response is governed by tensile
tearing failure. For the impact velocity higher than 180 m=s; an instantaneous shear failure is
observed. This latter case is considered in the present paper.
As an illustration, Fig. 7 shows the deformation and fracture mode of two beams impacted at
600 m=s [20]. This result was obtained using ABAQUS/Explicit with the fracture criterion of
accumulated plastic strain and a cut-off value of negative triaxiality.
Two situations may develop later. The remainder of the striking beams could stick to the
impacted beam and move with it. Alternatively, the broken piece will slide off the impact area and
will no longer interfere with the stricken beam. Both cases will be considered in this paper. For
simplicity, the latter case will be taken as an example to present problem formulation in the
following sections.
ARTICLE IN PRESS
Fig. 5. Four possible impact scenarios.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 191
2.2. Problem formulation
The problem after the second impact is formulated by following the procedure for the single
impact case of beam-to-beam [18], and for rigid mass impact on a plastic beam [12]. However,
different from the single impact and the rigid mass impact where the stricken beam is at rest upon
ARTICLE IN PRESS
Fig. 7. Deection of the single impact of beam-to-beam at t 48 ms: The initial impact velocity is V
0
600 m=s:
Fig. 6. Residual velocity versus initial impact velocity for various mass ratios.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 192
impact, in the double impact case, the second impact occurs at the stricken beam with the
transient elds of velocity and deection. Hence, it is an entirely new initial-boundary value
problem.
Consider the stricken beam after the second impact. The local equilibrium equation for the
stricken beam in the deformed region behind the rst wave front is given by

%
Nw
0

0
m v; 0pxpx
1
; 1
where
%
N 2s
0
bh is the axial tensile force; x
1
ct is the location of the rst wave front; v is the
transverse velocity; w is the transverse deection of the stricken beam; and the symbols prime and
dot denote differentiation with respect to the spatial and temporal coordinates, x and t;
respectively. Because s
0
is assumed to be constant, so is
%
N; and the governing equation can be
reduced to the familiar wave equation
c
2
w
00
. w; 2
where . w v; and c is the wave speed, dened by
c

%
N
m
_

s
0
r
_
: 3
Due to the discontinuity of the transverse velocity eld, two shock waves are generated in the
stricken beam upon the rst and second impact, respectively. The shock waves propagate away
from the impact area with the constant plastic wave speed c: All plastic deformations of the beam
take place at the wave front, where the material element is instantly stretched, rotated, and
imparted a transverse velocity. The deformed regions behind the wave front undergo rigid body
motion. Different from the single impact case, a material element in the stricken beam is subjected
twice to plastic deformations due to the rst and second shock waves. Hence, two regions undergo
rigid body motion with different transverse velocities denoted by v
1
and v
2
; respectively. Both
velocities are constant in space and vary with time. A typical prole of the velocity eld after the
second impact for the stricken beam and the striking beams is shown in Fig. 8.
The deection and the transverse velocity of the beam should satisfy the kinematic and dynamic
continuity conditions at the wave front [12]:
w cw
0
0;
%
Nw
0
mc w 0; 4
where the symbol denotes a jump of a given quantity across the wave front. In detail, the
kinematic and dynamic continuity conditions at the rst wave front, x x
1
; require
0 v
1
c0 w
0
j
xx

1
0;
%
N0 w
0
j
xx

1
mc0 v
1
0; 5
where zero initial conditions in front of the rst wave front are used. Similarly, the continuity
conditions at the second wave front, x x
2
ct t
0
; are specied as
v
2
v
1
cw
0
j
xx

2
w
0
j
xx

2
0;
%
Nw
0
j
xx

2
w
0
j
xx

2
mcv
2
v
1
0: 6
ARTICLE IN PRESS
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 193
The problem also satises the boundary condition at x 0
2
%
Nw
0
2mb
i
v; 7
where only the mass of the stricken beam at the impact area contributes to the inertia force, which
corresponds to the case where no parts of the striking beams remain in contact with the stricken
beam after the impact. Eq. (7) means that the transverse shear force resulting from the axial
stretch is equilibrated by the inertia force. It is not necessary to introduce interaction force
between the stricken beam and the striking beams in the above equation, because no more
momentum transfers among the beams after fracture of the striking beams. Without difculty,
one can derive boundary conditions for the other case where the remainder of the striking beam
moves with the stricken beam.
In terms of v
1
and v
2
; the governing equation (1) can be rewritten as

%
Nw
0

0
m v
2
; 0pxpx
2
;

%
Nw
0

0
m v
1
; x
1
pxpx
2
: 8
ARTICLE IN PRESS
Fig. 8. Transient velocity eld in three beams after the second impact.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 194
In each deformed region, both the slope and transverse velocity of the beam are continuous.
Based on the assumption of the uniform distribution of the transverse velocity in each deformed
region, integrating the above equations with respect to x in the respective regions yields
%
Nw
0
j
xx

%
Nw
0
j
x0
m v
2
ct ct
0
; 0pxpx
2
;
%
Nw
0
j
xx

%
Nw
0
j
xx

1
m v
1
ct
0
; x
1
pxpx
2
:
9
Substituting the corresponding boundary and dynamic continuity conditions, and combining the
above equations together gives the governing equation for the stricken beam with respect to the
transverse velocities
b
i
ct t
0
v
2
ct
0
v
1
cv
2
0: 10
Similar to the case of rigid mass impact and the single impact of beam-to-beam, the above
governing equation can also be directly obtained from the principle of linear momentum
conservation by differentiating the following equation with respect to time t:
2mb
i
ct t
0
v
2
2mct
0
v
1
I; 11
where I is a constant denoting the total momentum supplied by the striking beams during the
double impact process.
Note that the momentum transferred from the striking beam is a little underestimated, since
only the momentum in the impact area is taken into account. When both stricken beams are
sheared off, some momentum of both striking beams away from the impact area is transferred to
the stricken beam. Preliminary nite element simulations of the single impact of beam-to-beam
indicate that the failure duration is of the order of 100 ms [20], and thus the underestimation of the
momentum imparted by the striking beams would not be signicant. Estimate can be given on the
amount of momentum loss during the fracture process. Assuming the duration of the shear
fracture process, t
s
; is
t
s

h
i
3v
0
; 12
where the coefcient 1/3 in the expression for the duration of fracture comes from the property
that shear crack travels at a speed approximately equal three times higher than the impact velocity
[21]. The neglected momentum in the rst impact is approximately given by
DI
s
r
i
2b
i
h
i
2t
s
cv
0
v

1
; 13
where v

1
is the instantaneous velocity of the striking beam at the impact zone immediately upon
impact. The considered momentum transferred from the rst striking beam to the stricken beam,
I
s
; is
I
s
r
i
2b
i
h
i
2bv
0
v

1
: 14
The ratio of the neglected momentum to the considered momentum is
DI
s
I
s

1
3
h
i
b
c
v
0
: 15
ARTICLE IN PRESS
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 195
In order for the problem to be one dimensional rather than three dimensional, we have already
assumed that h
i
5b: Therefore, the term DI
s
can be indeed disregarded as long as the impact
velocity v
0
is of the same order as the transverse plastic stress wave speed c:
2.3. Assumptions
It can be observed that from the global equilibrium and the conditions of the kinematic and
dynamic continuity, only one equation is found for the two unknown velocities v
1
t and v
2
t:
Therefore, an additional condition is needed to solve the problem.
Different from the rst shock wave which propagates in the stricken beam with zero initial
conditions, the second wave propagates in the already deformed region of the stricken beam. As
stated earlier, the slope w
0
j
xx

2
is generated when the rst wave arrives at x x
2
: Since then, the
slope of the beam at x x
2
does not change with time until the second shock wave arrives there.
Hence, Dw
0
2
w
0
j
xx

2
w
0
j
xx

2
is the increment of the slope at x x
2
due to the second shock
wave. The kinematic and dynamic continuity conditions at the second wave front, x x
2
; Eq. (9),
can be recast in the incremental form
Dv
2
cDw
0
2
0;
%
NDw
0
2
mcDv
2
0; 16
where Dv
2
v
2
v
1
:
Now, we introduce an assumption that the incremental parts, Dv
2
and Dw
0
2
; satisfy the
governing wave equation at 0pxpx
2
;

%
NDw
0
2

0
mD v
2
; 0pxpx
2
; 17
where D v
2
v
2
v
1
: Then, the governing wave equation, Eq. (8), with regard to the velocity v
1
can be extended from the region between the two wave fronts, x
2
pxpx
1
; to the whole deformed
region, 0pxpx
1
;

%
Nw
0
1

0
m v
1
; 0pxpx
1
; 18
where the deection w
1
is generated due to the transverse velocity v
1
: The above equation indicates
that the rst shock wave will continue to propagate to the far eld after the second impact as if the
second impact never happened. In fact, both shock waves propagate with the common wave speed
in the same direction. It is impossible that the second shock wave would interfere or overtake the
rst one, and vice versa. Also, note that the governing wave equation is a homogeneous partial
differential equation such that the superposition of the solution is applicable.
Based on the assumption, there are two governing equations (17) and (18) with regard to Dv
2
and v
1
; respectively, both of which satisfy the continuity conditions at the corresponding wave
front, Eqs. (5) and (16). As shown in the preceding section, if the transverse velocity satises the
governing equation and the continuity conditions, the corresponding transverse momentum is
conserved during the propagation process. As shown in Fig. 9, the transverse velocity eld in the
stricken beam after the second impact is decomposed into two parts. Each part represents a shock
wave propagating with the wave speed, c; and the transverse momentum is conserved for each
part. It is convenient to derive the transverse velocity at any time from the principle of momentum
conservation.
ARTICLE IN PRESS
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 196
2.4. Velocity history
The solution to the velocity and deformation before the second impact is the same as the single
impact case of beam-to-beam, which was presented in Ref. [18]. Immediately upon the rst
impact, the velocity is
v

1

2m
i
b
2m
i
b 2mb
i
v
0
av
0
; 19
where the velocity with the superscript represents the instantaneous velocity upon impact;
a denotes the relative weight of the stricken beam and the striking beam at the impact area,
dened by
a
m
i
b
m
i
b mb
i
; 0oao1: 20
At a time before the second impact, the principle of linear momentum conservation gives
2mb
0
ctv
1
2mb
0
v

1
: 21
Note that there is no more momentum transfer between the striking beam and the stricken beam
after the impact. From the above equation, one obtains the velocity v
1
v
1
t
v
0

a
1 Zt=t
0
; 22
where the dimensionless parameter Z is dened by
Z
ct
0
b
i

1 b
1 a
t
0
t
; 23
where the parameter Z is related to the wave speed ratio, b; and the time parameter, t; both of
which were dened in the single impact case [18]
b
m
i
c
i
mc m
i
c
i
; 24
t
mb
i
m
i
b
mc m
i
c
i
: 25
By means of b and t; one can make a comparison between the double impact and the single
impact. Just before the second impact, t t
0
; the velocity v
1
becomes
v
1
t
0

v
0

a
1 Z
y: 26
For convenience, a new dimensionless parameter y is introduced in the above equation.
ARTICLE IN PRESS
Fig. 9. Velocity eld in the stricken beam after decomposition.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 197
Now, solve the velocity eld in the stricken beam after the second impact. Immediately upon
the second impact, the velocity v

2
can be obtained from the momentum conservation at the
impact area
2m
i
b 2mb
i
v

2
2m
i
bv
0
2mb
i
v
1
t
0
; 27
where the rst term on the right-hand side represents the momentum provided by the second
striking beam; the second term denotes the existing momentum of the stricken beam at the impact
area. Using the dimensionless parameters a and Z; it is convenient to write v

2
as
v

2
v
0
a 1
1 a
1 Z
_ _
: 28
From the momentum conservation for the incremental part as shown in Fig. 9, one gets the
velocity increment Dv
2
t
Dv
2
Dv

m2b
i
m2b
i
2ct t
0


1
1 Zt=t
0
1
; 29
where Dv

2
is the initial value of the incremental part, given by
Dv

2
v

2
v
1
t
0
av
0
1 y: 30
As stated earlier, the variation of the velocity v
1
with time is independent of the second impact, i.e.
the expression for the velocity v
1
after the second impact is the same as Eq. (22). The velocity v
2
is
obtained by adding v
1
and Dv
2
together
v
2
v
0
a
1
1 Zt=t
0

1 y
1 Zt=t
0
1
_ _
: 31
Plots of the velocity variation with time for specic values of a and Z are shown in Fig. 10.
ARTICLE IN PRESS
Fig. 10. Velocity variation with time for v
1
and v
2
in the double impact case.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 198
2.5. Deection and plastic strain
The transverse deection of the beams is calculated by integrating the transverse velocity with
respect to time. For the stricken beam, a point located at a distance x from the impact area
acquires displacement only after time t x=c: The transverse velocity at this point changes from
v
1
to v
2
at t t
0
x=c: Hence, the transverse deection of the stricken beam after the second
impact can be expressed as
w
_
t
0
x=c
x=c
v
1
dt
_
t
t
0
x=c
v
2
dt; 0pxpx
2
;
w
_
t
x=c
v
1
dt; x
2
pxpx
1
:
32
Substituting the expressions for the velocities, and integrating Eq. (32), one obtains the transverse
deection of the stricken beam,
w
b
i

av
0
c
ln
1 Zt=t
0
1 x=b
i
_ _

av
0
c
1 y ln
1 Z Zt=t
0
1 x=b
i
_ _
; 0pxpx
2
;
w
b
i

av
0
c
ln
1 Zt=t
0
1 x=b
i
_ _
; x
2
pxpx
1
:
33
The rst term on the right-hand side of Eq. (33) represents the deection due to the velocity v
1
;
and the second term denotes the deection due to the velocity increment Dv
2
: Plots of
instantaneous deection proles of the stricken beam are shown in Fig. 11.
From the intermediately large deection theory, the axial tensile strain in the stricken beam is
given by
e
1
2
@w
@x
_ _
2

1
2
av
0
c
_ _
2
2 y
2
1
x
b
i
_ _
2
; 0pxpx
2
;
1
2
av
0
c
_ _
2
1
x
b
i
_ _
2
; x
2
pxpx
1
:
_

_
34
Plots of the tensile strain variation along the beam for different values of Z are shown in Fig. 12. It
can be seen that there is a jump of the tensile strain at the second wave front, x x
2
: In the range
ARTICLE IN PRESS
Fig. 11. Deection proles of the stricken beam after the second impact and compared to the case of the single impact.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 199
of x
2
pxpx
1
; the tensile strain is the same as that in the case of the single impact. The maximum
tensile strain e
max
always takes place at the impact point, i.e. x 0;
e
max

1
2
av
0
c
_ _
2
2 y
2
: 35
When the time interval t
0
goes innity, Z-N; the maximum tensile strain approaches an
asymptote
e
max
2
2

1
2
av
0
c
_ _
2
; Z-N; 36
which means that the double impact becomes two separate, single impact. When the time interval
t
0
goes zero, the maximum tensile strain approaches another asymptotic value,
e
max

1
2
av
0
c
_ _
2
2 a
2
; Z-0: 37
2.6. Kinetic energy and plastic energy
Immediately upon impact, the stricken beam acquires the kinetic energy
E
0

1
2
2mb
i
v

1

2
mb
i
av
0

2
: 38
At a time, tot
0
; before the second impact, the kinetic energy of the stricken beam becomes
E
1
t
1
2
2mb
i
2mctv
2
1

E
0
1 ct=b
i
; tot
0
: 39
ARTICLE IN PRESS
Fig. 12. Strain variation along the stricken beam for different values of Z:
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 200
Upon the second impact, the newly acquired kinetic energy is
DE
1
2
2mb
i
v

2

2
v
2
1
t
0
E
0
1 2
1 a
1 Z

a
2
2a
1 Z
2
_ _
: 40
At a time, t; after the second impact, the total kinetic energy of the stricken beam becomes
E
2
t
1
2
2mb
i
2mct t
0
v
2
2
t
1
2
2mct
0
v
2
1
t
E
0
3 2y
1 Zt=t
0

1 y
2
1 Zt=t
0
1
_ _
: 41
The loss of the kinetic energy during the wave propagation is absorbed through plastic
deformation
Wt E
0
DE E
2
E
0
1 1 y
2
2
1 y
1 Z

3 2y
1 Zt=t
0

1 y
2
1 Zt=t
0
1
_ _
: 42
For a rigid-perfectly plastic string, the loss of the kinetic energy is completely converted to the
axial stretching energy. For a plastic beam, some of the loss of the kinetic energy is dissipated in
the form of plastic bending, besides plastic axial stretching.
2.7. Critical impact velocities
A beam under high velocity impact would fail by either tensile tearing or shear plugging. Here,
it is assumed that the stricken beam fails by tensile tearing. A constant critical plastic strain, e
f
; is
used to predict fracture of the stricken beam, due to its simplicity.
By setting the maximum plastic strain equal to the fracture strain, one can estimate the critical
impact velocity of the striking beam, v
cr
; to fracture the stricken beam,
v
cr
c

2e
f
_
a2 y


2e
f
_
1 a 1 bt
0
=t
a2 a1 a 21 bt
0
=t
: 43
The component of the bending strain is not taken into account in the estimation of the critical
impact velocity in the above equation.
It can be shown that the critical velocity always decreases as the time interval increases, Fig. 13,
no matter whether a > b or aob: When the time interval t
0
goes innity, the critical velocity
becomes half of that in the single impact:
v
cr
c

1
2

2e
f
_
a
: 44
In the preceding, the critical impact velocity is determined by differentiating the deection to
obtain the strain. In the following, we will show an alternative way to determine the critical
impact velocity.
Considering that the transverse velocity is a function of time, independent of the spatial
coordinate, the deection slope of the beam can be calculated by differentiating equation (32) with
ARTICLE IN PRESS
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 201
respect to x;
@w
@x

1
c
v
1
x
c
_ _
v
1
t
0

x
c
_ _
v
2
t
0

x
c
_ _ _ _
; 0pxpx
2
; 45
where Leibnizs differential rule has been used. Specically, the slope at the impact area, i.e. at
x 0; is given by
@w
@x

x0

1
c
v
1
0 v
1
t
0
v
2
t
0

1
c
v

1
v
1
t
0
v

2
: 46
Three instantaneous velocities, v

1
; v
1
t
0
; and v

2
; are involved to determine the slope at x 0;
which have been given in Eqs. (19), (26), and (28), respectively. Note that these velocities are
exactly obtained from the linear momentum conservation without introducing any additional
assumptions, hence the slope at x 0 is exactly determined. Substituting the expressions for these
velocities into Eq. (46), the same critical velocity as Eq. (43) can be obtained after similar algebra.
This method is more convenient, and thus will be used in the following cases.
2.8. Double impact by two identical rigid masses
Consider another case that both striking beams break immediately upon impact, the remainder
of the striking beams sticks to the stricken beam at the impact area and moves with it, Fig. 14. In
fact, this case is equivalent to the double impact event of a plastic beam by two identical rigid
projectiles with the mass M
0
2m
i
b: The problem can be completely solved in the same procedure
as described in the preceding sections. For simplicity, only the critical impact velocity to fracture
the stricken beam will be derived in the following.
The simple method based on Leibnizs differential rule will be used, in which only three
instantaneous velocities v

1
; v
1
t
0
; and v

2
; are involved to determine the tensile strain at x 0;
Eq. (46). These velocities can be exactly solved from the principle of momentum conservation.
ARTICLE IN PRESS
Fig. 13. Critical velocity variation with time interval in the double impact.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 202
Different from the rst case, the masses of the remainder of two striking beams or two rigid
bodies contribute to the transverse momentum in the stricken beam during the whole wave
propagation process. After the rst impact, the momentum conservation in the stricken beam
gives
2mb
i
ct 2m
i
bv
1
2mb
i
2m
i
bv

1
: 47
From the above equation, one can express the velocity v
1
t in terms of the mass ratio, a; and the
time parameter, Z;
v
1
t
v
0

a
1 1 aZt=t
0
: 48
Particularly, the velocity v
1
at t t
0
is given by
v
1
t
0

v
0

a
1 1 aZ
: 49
Upon the second impact at t t
0
; the momentum conservation can be expressed as
4m
i
b 2mb
i
v

2
2m
i
bv
0
2mb
i
2m
i
bv
1
t
0
; 50
which yields the velocity v

2
v

2
v
0

a
1 a
1
1
1 1 aZ
_ _
: 51
Substituting the expressions for v

1
; v
1
t
0
; and v

2
into Eq. (46), one can determine the slope and
further the tensile strain at x 0:
e
max

1
2
@w
@x
_ _
2

1
2
v
0
c
_ _
2
a
1 a
_ _
2
2 a
a
1 1 aZ
_ _
2
: 52
Plots of the maximum tensile strain versus the mass ratio a at various values of Z are shown in
Fig. 15. It appears that the maximum tensile strain does not increase monotonically with the mass
ratio for the time parameter Z in some range, which is different from the single impact event of a
ARTICLE IN PRESS
Fig. 14. Schematic of deformation and fracture of the beams in the double impact case.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 203
plastic beam by a rigid mass. The condition under which the extreme value of the maximum
tensile strain exists can be found by setting the differential of the maximum tensile strain with
respect to the mass ratio equal to zero,
@e
max
@a
0; 53
which after arrangement gives
Z
2
a
4
2Za
3
Z
2
2Za
2
2Z
2
2Za 2Z 1
2
0: 54
With the value of Z given, the above polynomial equation in a can be solved numerically, Fig. 16.
The root of the equation determines the location of the extreme value in Fig. 15. If Z 1:0; the
extreme value of the maximum tensile strain occurs at a 1:0: If Zo1; there is no extreme values
of e
max
; and the tensile strain always increases monotonically with the mass ratio. The extreme
value usually occurs at aE0:90 under the condition of Z > 1: In the case of the single rigid mass
impact, a heavy impacting body is always easy to break the stricken beam, as shown by
Wierzbicki and Hoo Fatt [12]. However, in the case of the double rigid mass impact, there is an
optimal mass ratio as a function of the time interval, with which the stricken beam is the easiest to
fracture.
Setting e
max
e
f
in Eq. (52), one can determine the critical impact velocity to fracture the
stricken beam,
v
cr
c


2e
f
_
1 a
a
2 a
a
1 1 aZ
_ _
1
; 55
Plots of the critical impact velocity versus the time interval are shown in Fig. 17.
ARTICLE IN PRESS
Fig. 15. Maximum tensile strains versus mass ratios for different values of Z in the case of double rigid mass impact.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 204
2.9. Double impact by two intact striking beams
In the preceding sections, the cases that both striking beams undergo fracture immediately upon
impact were solved. Now, consider the third case that both striking beams deform plastically and never
fracture during the impact process. The striking beams will always contact with the stricken beam.
Assume that the maximum slope in the stricken beam occurs at x 0 after the second impact.
The expression for the slope at the impact point is the same as Eq. (46), while the expressions for
two instantaneous velocities, v
1
t
0
and v

2
; will be different.
ARTICLE IN PRESS
Fig. 17. Critical velocity variation with the time interval for different values of a and b:
Fig. 16. Critical mass ratio a determining the extreme value of the maximum tensile strain.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 205
From the analysis of the single impact of beam-to-beam [18], the velocity v
1
t
0
is given by
v
1
t
0

v
0
b
a b
1 t
0
=t
: 56
The transverse momentum is conserved at the impact area immediately upon the second impact,
i.e. t t
0
;
2m
i
bv
0
2m
i
b 2mb
i
v
1
t
0
4m
i
b 2mb
i
v

2
; 57
which gives the velocity v

2
v

2
v
0

a
1 a

1
1 a
v
1
t
0

v
0
: 58
Substituting the expressions for v

1
; v
1
t
0
and v

2
into Eq. (46), and setting e
max
e
f
; one can
obtain the critical velocity v
cr
to break the stricken beam in the case of both striking beams
without fracture
v
cr
c


2e
f
_
1 a
a
1 t
0
=t
2 2 a bt
0
=t
: 59
Plots of the critical velocity variation with the time interval are shown in Fig. 18. There are three
cases depending on relative values of a and b: For the case of a b; the critical velocity is constant
and independent of the time interval. For the case of a > b; the critical velocity decreases with
the time interval, and vice versa. If the time interval t
0
goes innity, the critical velocity
ARTICLE IN PRESS
Fig. 18. Comparison of the critical velocity to fracture the stricken beam among three cases: (Case 1) both striking
beams fracture upon impact and no parts of the striking beams remain in contact with the stricken beam; (Case 2) both
striking beams fracture upon impact and the remainder of the striking beam at the contact area moves with the striking
beam, or this case can be thought as double impact by a rigid mass; (Case 3) both striking beams keep intact during
impact process.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 206
approaches an asymptote,
v
cr
c


2e
f
_
1 a
a
1
2 a b
: 60
Fig. 18 illustrates comparison of the critical velocity to fracture the stricken beam among the three
cases. It can be seen that the critical impact velocities always decrease with the time interval for
the three cases. The stricken beam is the most prone to fracture in the case where both striking
beams break immediately upon impact and no parts remain in contact with the stricken beam.
3. Multiple impact
3.1. Problem formulation
Consider a stationary beam subjected to multiple impact by n identical beams moving with the
velocity, v
0
; the time interval, t
0
; and the total spacing, l n 1v
0
t
0
; Fig. 19. Because totally
n 1 beams are involved during the impact process, there are many possible fracture scenarios
leading to complicated transverse velocity eld for every beam. Here, two simple cases will be
investigated. The rst case is that all of the striking beams break immediately upon impact; and no
parts of the striking beams still rest on the stricken beam after the impact (Case 1). The second is
that all of the striking beams break immediately upon impact; and the remainder of the striking
beams moves with the stricken beam, as shown in Fig. 14 (Case 2). Case 2 can also be considered
as the multiple impact by identical rigid bodies with the mass M
0
2m
i
b: In both cases, attention
is focused on the response of the stricken beam. For simplicity, Case 1 will be taken as an example
to present the problem formulation and the solution procedure.
ARTICLE IN PRESS
Fig. 19. Schematic of multiple impact of beam-to-beam.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 207
After a new impact, two new shock waves are generated at the impact area and propagates to
the far eld along both sides of the stricken beam. An additional plastic deformation in the
stricken beam is instantaneously generated at the new wave front. The region behind the new
wave front is subjected to rigid body motion again. Hence, similar to the case of the double
impact, the velocity and the acceleration between two wave fronts are constant in space and vary
with time. A typical transverse velocity eld in the stricken beam after the nth impact is shown in
Fig. 20.
Following the procedure for the double impact event, it can also be shown that from the global
equilibrium, the kinematic and dynamic continuity conditions, there is only one governing
equation for the problem of the multiple impact, which is related to n unknown velocities after the
nth impact. Hence, the problem cannot be solved without introducing additional conditions or
assumptions.
Prior to this, let us look at the kinematic and dynamic continuity conditions at the jth wave
front, which are given by
v
j
v
j1
c w
0
j
xx

j
w
0
j
xx

j
0;
%
Nw
0
j
xx

j
w
0
j
xx

j
mcv
j
v
j1
0; 61
where x
j
b
i
c t jt
0
is the location of the jth wave front. Both conditions can be recast in the
incremental form
Dv
j
cDw
0
j
0;
%
NDw
0
j
mcDv
j
0; 62
where Dv
j
v
j
v
j1
is the velocity increment; and Dw
0
j
is the slope jump at the jth wave front.
Eq. (62) indicates that the increment of the plastic slope is proportional to the velocity increment.
Similar to the case of the double impact, it is assumed that the velocity increment Dv
j
and the
corresponding plastic slope increment Dw
0
j
satisfy the governing wave equation:

%
NDw
0
j

0
mD v
j
; x
j1
pxpx
j
: 63
ARTICLE IN PRESS
Fig. 20. Schematic of the velocity eld in the multiple impact case.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 208
It can be shown that the transverse momentum is conserved with regard to the velocity increment
Dv
j
; if the governing equation, the kinematic and dynamic continuity conditions are satised. The
velocity eld in Fig. 20 can be decomposed into n parts, Fig. 21. Every part represents a shock
wave propagating with the same wave speed c: The transverse momentum for every part is
conserved during the shock wave propagation process. Thus, the problem becomes amenable.
3.2. Velocity history
To develop the solution procedure, assume that the velocity eld after the jth impact has been
determined. Then the solution process for the next impact corresponding to the j 1th impact is
typical and can be applied repetitively until the nth impact is completed.
At t jt
0
; the stricken beam is impacted by the j 1th striking beam. The transverse
momentum is conserved at the impact area upon impact, given by
m
i
2bv
0
m2b
i
v
j
jt
0
m
i
2b m2b
i
v

j1
; 64
where the subscript j represents the jth impact; v

j1
is the instantaneous velocity at the impact area
just upon the j 1th impact; and v
j
jt
0
is the velocity of the stricken beam at t jt
0
after the
jth impact. From the above equation, it is convenient to express v

j1
; in terms of the mass ratio, as
v

j1
av
0
1 av
j
jt
0
: 65
The transverse velocity increment due to the j 1th impact is dened by
Dv
j1
t v
j1
t v
j
t; tXjt
0
: 66
As assumed earlier, the transverse momentum is conserved for this velocity increment as the
j 1th shock wave propagates away from the impact area. Hence, Dv
j1
at a time, t; is given by
Dv
j1
t
Dv

j1
jt
0


m2b
0
m2b
0
m2ct jt
0


1
1 Zt=t
0
j
; tXjt
0
; 67
ARTICLE IN PRESS
Fig. 21. Decomposition of the velocity eld in the multiple impact case.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 209
where Dv

j1
is the velocity increment immediately upon impact, i.e. at t jt
0
;
Dv

j1
v

j1
v
j
jt
0
av
0
v
j
jt
0
: 68
The total velocity v
j1
at the region near the impact area is a summation of all of the velocity
increments:
v
j1
t v
j
t Dv
j1
t

j1
k1
Dv
k
t; 69
where Dv
1
t v
1
t: Particularly, at t jt
0
; the velocity v
j
jt
0
is given by
v
j
jt
0

j
k1
Dv
k
jt
0

Dv

1
1 jZ

Dv

2
1 j 1Z
?
Dv

j
1 Z
; 70
where Eq. (67) has been used.
Based on the above solution process, it is easy to develop a calculation routine to determine the
velocity of the stricken beam at any time. Plots of the velocity versus time for specic values of a
and Z are shown in Fig. 22.
ARTICLE IN PRESS
Fig. 22. Temporal variation of the velocity of the stricken beam in the multiple impact event.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 210
3.3. Deection
The deection of the stricken beam at a distance x from the impact area after the nth impact is
given by
w
_
t
0
x=c
x=c
v
1
dt
_
2t
0
x=c
t
0
x=c
v
2
dt ?
_
t
n1t
0
x=c
v
n
dt; 0pxpct n 1t
0
: 71
Similar expressions can be given for the deection in other regions. Specially, the displacement at
the impact area, i.e. at x 0; is
d
c

_
t
0
Dv
1
dt
_
t
t
0
Dv
2
dt ?
_
t
n1t
0
Dv
n
dt: 72
Fig. 23 shows comparison of the displacement history at the impact area among three cases.
3.4. Tensile strain and critical impact velocity
In this section, the simple method based on Leibnizs differential rule will be used to derive the
tensile strain at the impact area and further the critical impact velocity to break the stricken beam.
From Eq. (71), differentiating the deection with the spatial coordinate x; the slope of the beam at
x 0 after the nth impact is given by
@w
@x

tn1t
0

1
c
v
1
0 v
1
t
0
v
2
t
0
v
2
2t
0
? v
n
n 1t
0


1
c
v

1
v
1
t
0
v

2
v
2
2t
0
? v

n
: 73
ARTICLE IN PRESS
Fig. 23. Comparison of the displacement at x 0 among three cases.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 211
From Eq. (68), the above equation can be rewritten as
@w
@x

tn1t
0

1
c
Dv

1
Dv

2
? Dv

n

1
c

n
j1
Dv

j
; 74
where Dv

1
v

1
: The above equation indicates that the plastic slope of the stricken beam at x 0
is proportional to the summation of all of the velocity increments upon impact. This equation will
be used to derive a recursion formula for the plastic slope later on.
Substituting the expression, Eq. (68), for every term in the right-hand side of the above
equation, one gets
@w
@x

tn1t
0
n
av
0
c

a
c
v
1
t
0
v
2
2t
0
? v
n1
n 1t
0
: 75
From Eq. (70), the above equation can be expanded as
@w
@x

tn1t
0
n
av
0
c

a
c
Dv

1
1 Z
_ _ _

Dv

1
1 2Z

Dv

2
1 Z
_ _
?
Dv

1
1 n 1Z

Dv

2
1 n 2Z
?
Dv

n1
1 Z
_ __
:
76
Further, collecting common terms in the above equation, one obtains
@w
@x

tn1t
0
n
av
0
c

a
1 Z
1
c
Dv

1
Dv

2
? Dv

n1

a
1 2Z
1
c
Dv

1
Dv

2
?Dv

n2
?
a
1 n 1Z
1
c
Dv

1
: 77
As indicated in Eq. (74), the above equation can be rearranged as
@w
@x

tn1t
0
n
av
0
c

a
1 Z
@w
@x

tn2t
0

a
1 2Z
@w
@x

tn3t
0
?
a
1 n 1Z
@w
@x

t0
: 78
For simplicity, introduce a dimensionless parameter g
j
to denote the slope at x 0 after the jth
impact,
g
j

@w
@x

t j1t
0
c
av
0
: 79
In terms of the denition of g
j
; Eq. (78) becomes a recursion formula for the plastic slope
g
n
n
a
1 Z
g
n1

a
1 2Z
g
n2
?
a
1 n 1Z
g
1
; 80
ARTICLE IN PRESS
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 212
where g
1
1: For the case of Z > 2; an approximate closed-form solution for g
n
can be obtained by
neglecting high order terms with regard to Z
g
n
n a

n1
j1
n j
1 jZ
: 81
A short computational routine is developed to calculate the plastic slope with the dimensionless
parameters given.
In terms of the dimensionless slope g
n
; the tensile strain in the stricken beam after the nth impact
is given by
e
n

1
2
av
0
c
_ _
2
g
2
n
: 82
It is interesting to discuss two limiting cases: Z-N and 0, which are helpful to understand the
general case derived in the preceding. If the time interval Z goes innity, the transverse velocity in
the stricken beam becomes innitesimal until the next impact. That is, the velocity v
j
jt
0
in
Eq. (75) can be neglected. The tensile strain approaches an asymptote
e
n

1
2
av
0
c
_ _
2
n
2
; Z-N: 83
On the other hand, if the time interval becomes innitesimal, no shock waves have sufcient time
to propagate away from the impact area until the whole impact process is nished. In this case, the
momentum conservation in the impact area after the jth impact is expressed as
2m
i
bv
0
2mb
i
v

j
2m
i
b 2mb
i
v

j1
; 84
where the velocity in the impact area before the next impact is still v

j
: The velocity v

j1
is given by
v

j1
av
0
1 av

j
: 85
A recursion formula with respect to the difference in the velocities can be obtained from the above
equation,
v

j1
v

j
1 av

j
v

j1
: 86
This recursion formula gives the closed-form solution for v

j
v

j
v
0
1 1 a
j
: 87
Note that in this case,
Dv

j
v

j
v

j1
: 88
Substituting the above expression into Eq. (74), the slope of the stricken beam after the nth impact
depends only on the velocity upon the last impact,
@w
@x

tn1t
0

1
c
v

1
v

2
v

1
? v

n
v

n1

v

n
c
: 89
ARTICLE IN PRESS
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 213
Hence, the tensile strain becomes
e
n

1
2
v
0
c
_ _
2
1 1 a
n

2
; Z-0: 90
As shown in Figs. 24 and 25, the nal tensile strain increases with the mass ratio and the time
interval.
Similar to the double impact case, it is assumed that the stricken beam fails by tensile tearing,
and the stricken beam does not have any damage before the nal impact. Hence, the critical
velocity to fracture the stricken beam is given by setting e
n
e
f
in Eq. (82),
v
cr
c

2e
f
_
ag
n
: 91
ARTICLE IN PRESS
Fig. 25. Tensile strain versus time intervals for different mass ratios for Case 1.
Fig. 24. Tensile strains versus mass ratios for different time intervals for Case 1.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 214
Alternatively, with the impact velocity v
0
given, one can determine the critical impact number n
cr
to fracture the stricken beam, as shown in Fig. 26. The discrete points are calculated at specic
values of a and Z: These points are tted with two smooth solid curves for two cases.
3.5. Multiple rigid mass impact
In the preceding section, the case that no parts of the striking beams remain in contact with
the stricken beam after the impact was formulated and solved. Here, it is assumed that all of the
striking beams break immediately upon impact, and the remainder of the striking beams at the
contact area sticks to the stricken beam and moves with it. Similarly, this case is equivalent to
the case of the stricken beam impacted multiply by the identical rigid masses with the mass
M
0
2m
i
b:
The problem is solved in the same procedure as that in the preceding case. Assume that we have
solved the velocity eld after the jth impact. Now, we want to determine the velocity after the
j 1th impact. The momentum conservation upon the j 1th impact at the impact area is
expressed as
2m
i
bv
0
j2m
i
b 2mb
i
v
j
jt
0
j 12m
i
b 2mb
i
v

j1
; 92
which gives the instantaneous velocity, v

j1
;
v

j1
v
0

a
ja 1
1
a
ja 1
_ _
v
j
jt
0

v
0
: 93
Based on the assumption, the transverse momentum is conserved at any time with regard to the
velocity increment Dv
j1
:
2mb
i
j 12m
i
bDv

j1
2mb
i
j 12m
i
b 2mct jt
0
Dv
j1
t; 94
ARTICLE IN PRESS
Fig. 26. Tensile strain versus impact number.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 215
where Dv

j1
is the velocity increment upon the j 1th impact, dened by
Dv

j1
v
0

j1
v
0

v
j
jt
0

v
0

a
ja 1
1
v
j
jt
0

v
0
_ _
: 95
In terms of a and Z; Dv
j1
t can be expressed as
Dv
j1
t
Dv

j1

ja 1
ja 1 1 aZt=t
0
j
: 96
ARTICLE IN PRESS
Fig. 28. Comparison of tensile strains versus time interval between Case 1 and Case 2.
Fig. 27. Comparison of tensile strains versus mass ratio between Case 1 and Case 2.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 216
Particularly, at t j 1t
0
;
Dv
j1
j 1t
0

Dv

j1

ja 1
ja 1 1 aZ
: 97
With every velocity increment known, one can obtain the velocity v
j
t from Eq. (69), and further
the slope at x 0 can be obtained from Eq. (74). Different from the preceding case, no recursion
formula was found. A short computational routine was developed to calculate the tensile strain
with the mass ratio, a; the time parameter, Z; and the impact number, n; given. Numerical results
for specic cases are shown in Figs. 27 and 28, and compared to Case 1. It appears that the tensile
strains in Case 1 are much larger than those in Case 2. Similar to the double impact case, the
tensile strain does not increase monotonically with the mass ratio, Fig. 27.
Similar to the preceding case, two limiting cases are investigated as follows. First, if the time
interval t
0
goes innity, the velocity in the stricken beam becomes innitesimal until the next
impact. Hence, in this case,
Dv

j
v

j
: 98
The total mass at the contact area in the stricken beam increases by 2m
i
b after every impact. After
the jth impact, the velocity v

j
is obtained from the momentum conservation,
Dv

j
v
0

j
v
0

2m
i
b
j2m
i
b 2mb
i

a
j 1a 1
: 99
Substituting the expression for every velocity into Eq. (74), one can obtain the slope at x 0; and
further the tensile strain in the stricken beam after the nth impact
e
n

1
2
v
0
c
_ _
2
n
j1
a
1 j 1a
_ _
2
; Z-N: 100
As shown in Fig. 28, the tensile strain approaches to an asymptotic value as the time interval
increases.
Now, consider another limiting case that the time interval t
0
becomes innitesimal. No shock
waves propagate away from the contact area until the whole impact process is nished. Similar to
the limiting case with Z-0 in Case 1, the slope after the nth impact depends only on the velocity
upon the last impact, v

n
: v

n
can be obtained from the momentum conservation
v

n
v
0

n 2m
i
b
n 2m
i
b 2mb
i

na
n 1 a 1
: 101
Hence, the tensile strain e
n
after the nth impact is given by
e
n

1
2
v
0
c
_ _
2

na
n 1 a 1
_ _
2
; Z-0: 102
Fig. 29 shows the relations between the tensile strain and the mass ratio for different values of Z:
The critical mass ratio a corresponding to the extreme value of the tensile strain increases with the
time interval Z; and approaches unity as Z becomes innity.
ARTICLE IN PRESS
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 217
Note that no assumptions are introduced in the asymptotic analysis. Good agreements between
the asymptotic analysis and the general solutions for large values of Z verify the correctness of the
additional assumption introduced in Section 3.1.
4. Conclusions
The problem of multiple impact of beam-to-beam was formulated based on the rigid-plastic
beam-string model. In the case of the double impact, the closed-form solutions were obtained for
the deection prole, the transverse velocity, and the tensile strain. By assuming the failure mode
of tensile necking, the critical impact velocity to fracture the stricken beam was predicted. In the
case of the multiple impact, a recursion formula with respect to the impact number was found to
calculate the tensile strain. A computational routine was composed to solve the critical impact
velocity to fracture the stricken beam with the impact number given, and to determine the critical
impact number with the impact velocity given. Asymptotic analyses were performed for two
limiting cases that the time interval goes either innity or is kept innitesimal.
To the best of the authors knowledge, no experiments are available in the literature to validate
the closed-form solution in the present paper. Numerical simulations will be performed to
investigate the multiple impact problem using nite element methods.
References
[1] FEMA. World Trade Center building performance study: data collection, preliminarily observation, and
recommendation. Technical report FEMA 403, Federal Emergency Management Agency, New York, May 2000.
[2] Jones N. Plastic failure of ductile beams loaded dynamically. J Eng Ind 1976;98(1):1316.
[3] Liu JH, Jones N. Experimental investigation of clamped beams struck transversely by a mass. Int J Impact Eng
1987;3:30535.
ARTICLE IN PRESS
Fig. 29. Tensile strains versus mass ratios for different values of Z for Case 2.
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 218
[4] Liu JH, Jones N. Dynamic response of a rigid plastic clamped beam struck by a mass at any point on the span. Int
J Solids Struct 1988;24(3):25170.
[5] Jones N. Structural impact. New York: Cambridge University Press, 1997.
[6] Yu J, Jones N. Further experimental investigations on the failure of clamped beams under impact loads. Int J
Solids Struct 1991;27(9):111337.
[7] Jones N, Shen WQ. Criteria for inelastic rupture of metal beams subjected to large dynamic loads. In: Jones N,
Wierzbicki T, editors. Structural crashworthiness and failure. Amsterdam: Elsevier Applied Science; 1993.
[8] Yu TX, Stronge WJ. Large deections of a rigid-plastic beam-on-foundation from impact. Int J Impact Eng
1990;9(1):11526.
[9] Yu TX, Yang JL, Reid SR. Deformable body impact: dynamic plastic behavior of a moving free-free beam striking
the tip of a cantilever beam. Int J Solids Struct 2001;38:26187.
[10] Yang JL, Yu TX. Dynamic plastic behavior of a free rotating hinged beam striking a cantilever beam. Mech Struct
Mach 2001;29:393411.
[11] Mihailescu-Suliciu M, Sulicio I, Wierzbicki T, Fatt MH. Transient response of an impulsively loaded plastic string
on a plastic foundation. Q Appl Math 1996;LIV (2):32743.
[12] Wierzbicki T, Fatt MSH. Impact response of a string-on-plastic foundation. Int J Impact Eng 1992;12(1):2136.
[13] Teng X, Wierzbicki T. Large deformation of a long beam under high velocity impact: analytical vs. numerical
solution, submitted for publication.
[14] Wierzbicki T, Fatt MSH. Deformation and perforation of a circular membrane due to rigid projectile impact. In:
Geers T, Shin Y, editors. Dynamic response of structures to high energy excitation, vol. PVP, No. 225, Atlanta,
GA, USA: AMSE; 1991.
[15] Calder CA, Goldsmith W. Plastic deformation and perforation of thin plates resulting from projectile impact. Int J
Solids Struct 1971;7(7):86381.
[16] B^rvik T, Langseth M, Hopperstad O, Malo K. Perforation of 12 mm thick steel plates by 20 mm diameter
projectiles with at, hemispherical and conical noses. Part I: experimental study. Int J Impact Eng 2002;27:1935.
[17] Wierzbicki T, Teng X. How the airplane wing cut through the exterior columns of the World Trade Center. Int J
Impact Eng 2003;28:60125.
[18] Teng X, Wierzbicki T. Interactive failure of two impacting beams. J Eng Mech ASCE 2003;129(8):91826.
[19] Teng X, Wierzbicki T. Effect of fracture criteria on high velocity perforation of thin beams, submitted for
publication.
[20] Teng X, Wierzbicki T. Numerical simulations of the single impact of beam-to-beam. Technical report. Impact &
Crashworthiness Lab, Cambridge, MA: MIT; in preparation.
[21] Teng X, Wierzbicki T. Numerical study on crack propagation in high velocity perforation. Technical report.
Impact & Crashworthiness Lab, Cambridge, MA: MIT; 2003.
ARTICLE IN PRESS
X. Teng, T. Wierzbicki / International Journal of Impact Engineering 31 (2005) 185219 219

Potrebbero piacerti anche