Rebound Property in Low Velocity Impact of Two Equivalent Balls

Eijiro Inamura et al. Int. Journal of Engineering Research and Application www.ijera.com
ISSN : 2248-9622, Vol. 7, Issue 5, ( Part -4) May 2017, pp.05-12
www.ijera.com DOI: 10.9790/9622-0705040512 5 | P a g e
Rebound Property in Low Velocity Impact of Two Equivalent
Balls
Eijiro Inamura*, Hirofumi Minamoto**, Shozo Kawamura***
*(Tokyo Metropolitan College of Industrial Technology, Japan
Email: inamura@metro-cit.ac.jp)
** (Department of Mechanical Engineering, Shizuoka Institute of Science and Technology, Japan
Email: minamoto.hirofumi@sist.ac.jp)
*** (Department of Mechanical Engineering, Toyohashi University of Technology, Japan
Email: kawamura@me.tut.ac.jp)
ABSTRACT
The present study focuses on an impact phenomenon of two spheres and its rebound property. The collision of
two spheres is a fundamental problem with impact phenomenon. The coefficient of restitution characterizes the
property of impact phenomenon and has been estimated in general by experiment. In addition, it has been tried to
estimate the coefficient by analytical and numerical methods. Considering body deformation, the body is def-
lected rapidly in collision and the strain rate occurred in the body is significantly high. It is well known the yield
stress of the specific industrial material depends on the strain rate. However, it has not been proved the relation-
ship between the coefficient of restitution and the strain rate in details yet. The present paper discusses the coef-
ficient of restitution in low velocity impact of two equivalent balls. The impact experiments were conducted for
the balls with different diameters by a pendulum impact apparatus. And the balls are numerically analyzed based
on the finite element method considering the dependence of yield stress on strain rate as the material property of
the balls. In conclusion, the strain rate decreases with the increment of the ball diameter and, it causes the coeffi-
cient of restitution to decrease.
Keywords: Coefficient of restitution, Elastic collision, Impact velocity, Inelastic collision, Size effect
I. INTRODUCTION
The behaviors of contact and collision of
spheres are fundamental problems in mechanics of
elasticity and machine dynamics, and so on. They
have been interested from long ago [1]-[3] and in-
vestigated by using experimental, analytical and
numerical methods. The outcomes of the investiga-
tions have been applied many objects in shot peen-
ing [4][5], sports engineering[6][7], automotive [8],
and many others. The rebound properties are
represented by coefficient of restitution, impact time
and so on. They are changed with various causes
such as ball size, material property, impact speed,
and so on. Considering the causes, methodologies to
obtain impact model and the coefficient of restitu-
tion have been improved until now [9][10]. Specifi-
cally, as a research about ball size effect, Aryaei et
al. [11] fell balls 6 to12 mm in diameter on a plate at
0.7 to 3.2 m/s speed. As the experimental and nu-
merical results, they pointed out that the coefficient
of restitution was changed by ball size. Minamoto et
al. [12] collided two equivalent balls 25.4 mm, 13.5
mm and 6.75 mm in diameter at high speed, 10 to 33
m/s. They clarified that the coefficient of restitution
was changed by ball size and elastic wave propaga-
tion as the experimental and numerical results. How-
ever, there is no precise report about that of two
equivalent balls impact at lower speed than 1.0 m/s.
In the present study, the rebound property
of two equivalent balls at relative low speed is ex-
amined and the influences of heat treating of the
balls, the impact speed and the ball size on the re-
bound property are discussed. The ball diameters are
13.5 mm, 25.4 mm, 31.8 mm and 44.5 mm. The
balls are made of high carbon chromium bearing
steel. Some of them are in the state of factory ship-
ment. The others are annealed. In the experiment,
two balls are collided at 0.2 to 1.2 m/s velocity by
using the motion of pendulum. Moreover, static and
dynamic compressive material tests for the balls are
performed and the relationship of the yield stress and
the strain rate are obtained. Considering the depen-
dence of yield stress on strain rate, the finite element
analysis is executed. In order to separate the contri-
butions of impact velocity and ball diameter, the
coefficient of restitution is expressed as an exponen-
tial function of impact velocity and then the relation-
ship between coefficients of the function and ball
diameter are obtained. Comparing the experimental
and numerical results, the changes of the coefficient
of restitution and the mechanical energy, the stress
and the strain in the balls, and so on are discussed.
RESEARCH ARTICLE OPEN ACCESS

II. IMPACT TEST
The ball diameters used by the experiment
are 13.5 mm (17/32 in), 25.4 mm (1 in), 31.8 mm (1-
1/4 in) and 44.5 mm (1-3/4 in). The ball material is
high carbon chromium bearing steel (JIS SUJ2). The
impact tests were conducted for the balls in the state
of factory shipment and the annealed balls. The
balls in the state of factory shipment have been har-
dened by cold forming and surface quenching in the
manufacturing process. The annealed balls are sof-
tened at 850°C in nitrogen gas. In the present paper,
the ball in the state of factory shipment and the an-
nealed ball are called balls A and B, respectively.
Figure 1 shows the experimental apparatus.
Figures 1 (a) and 1(b) are the front and side views,
respectively. A thread is glued on a ball. The two
balls are hanged at the top plate with the thread in V
form. The distance between the two balls is the same
as the ball diameter. The two balls are in touch each
other before experiment. The radius of pendulum R
is determined as shown Table 1 in order that the
thread does not sag and the maximum impact veloci-
ty reaches about 1 m/s. The right and left balls are
called ball 1 and 2, respectively. The ball 1 collides
with the ball 2. 0 is an angle of the raised ball 1
before the collision. 1 and 2 are angles of the rising
balls 1 and 2 after the collision, respectively. These
angles are measured with the protractors which
(a) Front view
(b) Side view
Figure 1 pendulum impact apparatus
center is the same location as the center of the pen-
dulum and is fixed in a horizontal position. At the
beginning, the ball 2 is set at rest. Next, the ball 1 is
left by 0, which is changed from 5° to 35° by 5°.
And then the ball 1 is released to collide with the
ball 2. 1 and 2 are measured four time at the same
value of 0. The relative averages of 1 and 2 are
used to calculate velocities and so on described in
the next section.
Table 1 Ball diameter and pendulum radius
Ball diameter d mm Pendulum radius R mm
13.5 0.40
25.4 0.43
31.8 0.43
44.5 0.46

III. COEFFICIENT OF RESTITUTION
AND ENERGY LOSS
v0 is the velocity of the ball 1 just before the
collision, namely the impact velocity. v1 and v2 are
the velocities of the balls 1 and 2 just after the colli-
sion, respectively. Neglecting the air resistance of
the balls, the velocities v0, v1 and v2 are given from R,
0, 1 and 2 by the next equations:
0 0
2 (1 cos )v gR   (1)
1 1
2 (1 cos )v gR   (2)
2 2
2 (1 cos )v gR   (3)
where g is the gravity acceleration. The coefficient
of restitution e is obtained by the next equation:
2 12 1
0 0
1 co s 1 co s
1 co s
v v
e
v
 

  
 

(4)
Because the two balls have the same mass, the fol-
lowing relationship is satisfied by the law of conser-
vation of momentum:
0 1 2
v v v  (5)
Solving Eqns. (4) and (5) for v1 and v2, they are de-
noted by v0 and e as the next equations:
1 0
1
(1 )
2
v e v  , 2 0
1
(1 )
2
v e v  (6)
If e = 1, then v1 = 0, v2 = v0. After the collision, the
ball 1 freezes. Instead, the ball 2 is accelerated by
the same velocity as the impact velocity and rises by
the same angle as the ball 1 is left. By the way, if e <
1, then v1 > 0, v2 < v0. The ball 1 has some velocity.
Since the velocity of the ball 2 just after the collision
is lower than v0, 2 is smaller than 0.
0 is mechanical energy of the system be-
fore the collision. E is energy loss of the system
due to the collision. From the relations (1)~(3), the

ratio of the energy loss E to the energy 0, E/0,
is represented by the next expression:
2 2 2
0 1 2
2
0
0
1 1 1
2 2 2
1
2
m v m v m v
Δ E
E
m v
 
  
 

(7)
where, m is the mass of the ball. Considering the
dependence of the impact velocity, the coefficient of
restitution is approximated by the next exponential
function [6]:
0
ex p ( )
n
e v  (8)
Following Eqn. (8), Eqn. (7) is rewritten to the next
equation:
  0
0
1
1 ex p 2
2
nE
v
E

   (9)
where,  and n are coefficients determined from
experimental results by the least square method.
Considering the dependence of the coefficient of
restitution e on the ball size, the coefficients  and n
are related to the ball diameter.
IV. MATERIAL PROPERTY OF THE
BALL
In order to obtain the static relationship of
stress and strain of the ball material, cylindrical spe-
cimens were made of the balls and had 8 mm in di-
ameter and 20 mm long. The static compressive test
is carried out with a universal testing machine. The
strain rate is 0.05/s. To obtain the dynamic relation-
ship of stress and strain, the impact compressive test
is performed by the Hopkinson bar method [13]. By
using various sizes of specimen and loading condi-
tions, the stress - strain curves for the different strain
rates were measured [12][14].
V. FINITE ELEMENT ANALYSIS
Because the analytical object is axisymme-
tric, the axisymmetric numerical analysis of the two
balls are executed based on the finite element me-
thod. Figure 2(a) shows the entire finite element
subdivision and Fig. 2(b) is the wide view in the
vicinity of contact part which area was meshed fine-
ly. The other area is meshed roughly. The elements
are four-node axisymmetric elements. The material
model includes the dependence of yield stress on
strain rate. LS-DYNA [15][16] is a general dynamic
non-linear analysis software based on the finite ele-
ment method. The software is used for the analysis
of the balls.
(a) Entire subdivision (b) Fine meshed area
Figure 2 finite element subdivision
VI. RESULTS AND DISCUSSIONS
6.1 Material property of the ball
Figure 3 shows the relations between the
strain and its rate obtained from the six kinds of test
condition for the impact compressive test. The ver-
tical axis is strain rate and the horizontal axis is
strain. When the strain rate increases, wave appears
strongly in the measurement results. For each condi-
tion, the strain rate is regarded as constant shown as
the broken line. They are 600/s, 1000/s, 1750/s,
2300/s, 2800/s and 4100/s, respectively.
Figure 3 strain rate - strain relations of SUJ2
Figure 4 shows the relationship between the
true stress and the true strain obtained by the static
and dynamic compressive tests. The lines are corres-
ponded to the strain rates shown in Fig. 3. The ver-
tical axis is true stress and the horizontal axis is true
strain. The compressive yield stress increases with
the strain rate. Also, the true stress increases linearly
after the yield. From the result of the static compres-
sive test, Young’s modulus is 202 GPa, the static
compressive yield stress is 470 MPa. There are

straight parts of the strain rates 600/s and 1000/s
which are stead relatively after the yield. From their
slopes, the tangent coefficient Et is gotten by the
least square method. As results, the slopes of the
strain rates 600/s and 1000/s are 2.6 GPa and 2.2
GPa, respectively. Then it was determined that the
tangent coefficient was the average of them, 2.4
GPa.
Figure 4 true stress - true strain curves of SUJ2
Figure 5 shows the relationship between the
compressive yield stress and the strain rate obtained
from Fig. 4. The vertical axis is compressive yield
stress and the horizontal axis is strain rate. The static
compressive yield stress 470 MPa is plotted at 0.05/s
strain rate. Because the results of the impact com-
pressive test oscillate (Fig.3), the values of the strain
rate and the compressive yield stress are read with
increments of 50/s and 50 MPa, respectively. The
broken line is a third order curve fitting the experi-
mental data by the least square method. The com-
pressive yield stress increases monotonically with
the strain rate. When the strain rate is higher than
2500/s, the compressive yield stress becomes con-
stant 900 MPa.
Figure 5 variation of yield stress with strain rate
The material properties are listed in Table
2. They and the relationship shown in Fig. 5 were
inputted to the software, LS-DYNA as material data.
The numerical analysis based on the finite element
method was executed considering the dependence of
yield stress on strain rate.
Table 2 Material properties of SUJ2 inputted in LS-
DYNA
Density  kg/m3
7850
Young's modulus E GPa 202
Poisson's ratio  0.3
Static yield stress ys MPa 470
Tangent modulus Et GPa 2.4
6.2 Experimental and analytical results for im-
pact of the balls
The respective averages of the experimental
values of 1 and 2 obtained in the impact test are
computed. v1 and v2 are calculated from the aver-
ages. Figure 6 shows the relationships between v1
and v0 and Fig. 7 shows one between v2 and v0. The
horizontal axes in the both figures are v0. The vertic-
al axis in Fig. 6 is v1 and one in Fig. 7 is v2. Solid
and empty marks indicate the results about the balls
A and B, respectively. The rectangles, circles, up-
pointing triangles and down-pointing triangles indi-
cate 13.5 mm, 25.4 mm, 31.8 mm and 44.5 mm di-
ameters, respectively. The broken lines are second
order curves fitting the respective results by the least
square method. In the case of v0 = 1.28 m/s (0 =
35°), the thread separated with the ball B with a di-
ameter of 44.5 mm during the experiment and then
the result could not be obtained. v1 of the ball A is
almost 0 m/s irrespective of v0. By the way, v1 of the
ball B increases with v0. The higher v0, the higher v1
of the ball B. v2 of the ball A is almost the same
Figure 6 variation of velocities v1 of ball 1 after col-
lision with impact velocity v0

Figure 7 variation of velocities v2 of ball 2 after col-
lision with impact velocity v0
value as v0 irrespective of v0. On the other hand, v2
of the ball B is lower than v0. The higher v0, the larg-
er difference between v0 and v2 of the ball B. As dis-
cussed later, the tendency means that the ball A be-
haves when e is near 1 and the ball B acts when e is
relatively smaller than 1. For the relationship be-
tween v1, v2 and the ball diameter d, obviously, there
is no much change of v1 and v2 of the ball A with d.
However, the larger d, the higher v1 of the ball B and
the lower v2 of the ball B.
Figure 8 shows the relationship between e
and v0 calculated from Eqn. (4) for the balls A and
B. The vertical axis is e and the horizontal axis is v0.
The broken lines are exponential curves of Eqn. (8)
fitting the results of the ball B by the least square
method. e of the ball A is irrespective of v0 and is
about constant from 0.97 to 0.99. It is not found the
significant change of e with d. Therefore, it is consi-
dered that the collision of ball A is almost perfectly
elastic. On the other hand, it is seen that e of the ball
B tends to decrease with an increase in v0. Also, it is
turned out that there is a tendency that the larger d,
the smaller e.
Figure 8 variation of the coefficient of restitution e
of balls A and B with impact velocity v0
and v0 obtained by the finite element method (FEM)
in addition to ones of the ball B. The solid lines are
exponential curves of Eqn. (8) fitting e of FEM by
the least square method. As well as the ball B, e ob-
tained by FEM decreases with an increase of v0. The
larger d, the smaller e obtained by FEM. In the case
of the ball B, e of d = 31.8 mm and 44.5 mm are
almost the same. e of d =13.5 mm and 25.4 mm are
different from e of d = 31.8 mm and 44.5 mm. By
the way, in the case of FEM, there is not a change of
e with d when v0 is equal to and lower than 0.3 m/s.
And e of d = 13.5 mm is larger than e of the other
ball diameters when v0 is higher than 0.3 m/s. e of d
= 25.4 mm, 31.8 mm and 44.5 mm are almost the
same. As mentioned later, the cause is that the strain
rate and the high stress region in the ball of d = 13.5
mm are much larger than ones in the balls of the
other diameter. The difference of e with d increases
with an increment of v0. In the case of d =13.5 mm
and 25.4 mm, e of the ball B and e obtained by FEM
are different. In the case of d = 31.8 mm and 44.5
mm, they are much the same.
Figure 9 variation of the coefficient of restitution e
of ball B and FEM with impact velocity v0
and d in the cases of v0 = 0.5 m/s and 1.0 m/s. The
values of the ball B are calculated by the linear in-
terpolation for the experimental results in the vicini-
ty of v0 = 0.5 m/s and 1.0 m/s. The circular and tri-
angular marks indicate v0 = 0.5 m/s and 1.0 m/s, re-
spectively. The empty and solid marks indicate the e
of the ball B and FEM, respectively. In the case of v0
= 0.5 m/s, e of the ball B decreases with an incre-
ment of d. However, e obtained by FEM except d =
13.5 mm is constant. When v0 is lower than 0.5 m/s
and d is larger than 25.4 mm, the dependence of
yield stress on the strain rate scarcely influence the
results. Jackson et al. [9] compared several models
to experimental results of aluminum oxide spheres
impacting on a steel flat and pointed out that the

similar reason. The tendency of e obtained by FEM
is different from e of the ball B. In the case of v0 =
1.0 m/s, e of the ball B and one of FEM decrease
with an increment of d. The decrement rates reduce
with increasing d. Also. e obtained by FEM is small-
er than e of the ball B.
Figure 10 variation of coefficient of restitution e
with impact velocity v0
Figure 11 shows E/E0 calculated by Eqn.
(7) for the ball B and FEM. The vertical axis is
E/E0 and the horizontal axis is v0. The marks and
lines identify the respective results as well as Fig. 9.
E/E0 increases with increasing v0. When v0 is lower
than 0.2m/s, E/E0 increases rapidly with v0. When
v0 is higher than 0.2m/s, the increment rate of E/E0
decreases little by little. E/E0 of the ball B 13.5 mm
in diameter is slightly smaller than ones of the ball B
with other diameter, which are not much different
together and increase with d. E/E0 of d = 13.5 mm
and one of d = 25.4 mm obtained by FEM are differ-
ent from ones of the ball B. E/E0 significantly
changes with v0 when d is small.
Figure 11 relationship of energy loss with impact
velocity for ball B
The coefficients  and n appear in the ap-
proximate expressions (8) of e and (9) of E/E0.
Figure 12 shows the relationship between  and d.
Also, Fig. 13 shows the relationship between n and
d. The circular and triangular marks indicate e and
E/E0, respectively. The empty and solid marks in-
dicate the results of ball B and FEM, respectively. In
the case of ball B, the reason for the difference be-
tween the results of e and E/E0 is that v1 + v2 is not
exactly equal to v0. The average of (v1 + v2)/v0 is
0.990 and its range is 0.954 to 1.013. On the other
hand, in the case of FEM, the results of e and E/E0
are mostly coincident. In Fig. 12, increases with d
and the increment rate decreases with d little by lit-
tle. For the ball B,  of E/E0 is larger than  of e as
a whole. For the results of FEM,  of e and E/E0
approximate to of E/E0 of the ball B. In Fig. 13,
n of the ball B does not change monotonically and
has a tendency to decrease with d. However, n of
FEM increases monotonically and its tendency was
different from n of the ball B.
Figure 12 relationship of coefficient  and ball di-
ameter d
Figure 13 relationship of coefficient n and ball di-
ameter d

As the results of FEM, Table 3 is a list of
the maximum equivalent strain rate  max during the
collision and the maximum equivalent stress max
when the ball deforms maximally. Also, Fig. 14
shows contour maps of equivalent stress max in the
vicinity of the contact part when the ball deforms
maximally. The photo height is a fortieth of d. The
arrows in the photos are locations where the maxi-
mum equivalent stress occurs. The larger d, the
smaller  max.  max of d = 25.4 mm is two thirds of
one of d = 13.5 mm and  max of d = 44.5 mm is
smaller than a half of one of d =13.5 mm. max of d =
13.5 mm is 10 MPa larger than ones of the other
diameters, which are about the same. Although max
does not change significantly, the incremental rate of
yield stress decreases with a decrement of  max.
Consequently, it appears the decrement of e with d.
Table 3 Equivalent strain rate and maximum equiva-
lent stress (v0 =1 m/s)
Ball diameter d mm  max × 103
/s max MPa
13.5 30.8 
25.4 20.5 
31.8 18.6 
44.5 13.6 
(a) 13.5 mm (b) 25.4 mm
(c) 31.8 mm (d) 44.5 mm
Figure 14 Contour maps of equivalent stress at max-
imum deformation (v0 =1 m/s)
VII. CONCLUSION
The Impact test of two equivalent balls was
performed by utilizing the motion of pendulums.
Some of them were in the state of factory shipment.
The others were annealed. In addition, the numerical
analysis based on the finite element method was
executed considering the dependence of yield stress
on strain rate as the material property of the an-
nealed ball. As the results, the ball in the state of
factory shipment behaves perfectly elastic collision
and are independent of the impact velocity and the
ball diameter. The coefficient of restitution of the
annealed ball decreases with an increment of the
impact velocity and a decrement of the ball diame-
ter. The higher impact velocity, the larger decrement
of the coefficient of restitution. The smaller ball di-
ameter, the higher effect of ball size on the coeffi-
cient of restitution. The coefficients involved in the
functions for the coefficient of restitution changed
corresponding to the ball diameter. By using the
relationship, it is possible to estimate results for an
arbitrary ball diameter. In the numerical results by
the finite element method, because the strain rate
decreases with increasing the ball diameter, it is
found that the coefficient of restitution decreases.
REFERENCES
[1] A. E. Love, A Treatise on the Mathematical
Theory of Elasticity, 4th edition (Cambridge
University Press, 1934).
[2] S. Timoshenko and N. Goodier, Theory of
Elasticity, 3rd edition (McGRAW-HILL,
New York, 1970).
[3] W. J. Stronge, Impact Mechanics (Cambridge
University, 2004).
[4] T. Hong, J. Y. Ooi, B.A. Shaw, A numerical
study of the residual stress pattern from single
shot impacting on a metallic component, Ad-
vances in Engineering Software, 39, 2008,
743-756.
[5] C. Nouguier-Lehon, M. Zarwel, C. Diviani, D.
Hertz, H. Zahouani, T. Hoc, Surface impact
analysis in shot peening process, Wear, 302,
2013, 1058–1063.
[6] M. H. A. Hassan, Z. Taha, Finite Element
Analysis of Soccer Heading, Procedia Engi-
neering, 112, 2015, 46-51.
[7] T. Y. Pang, A. Subic, M. Takla, Finite ele-
ment analysis of impact between cricket ball
and cantilever beam, Procedia Engineering,
13, 2011, 258-264.
[8] M. Batista, G. Zovak, A RESTITUTION
MODEL OF TWO-CAR COLLINEAR
COLLISIONS, Promet - Traffic & Transpor-
tation, 19(1), 2007, 1-6.
[9] R. L. Jackson, I. Green, D. B. Marghitu,
Predicting the coefficient of restitution of im-
pacting elastic-perfectly plastic spheres, Non-
linear Dynamics, 60, 2010, 217-229.
[10] M. Ahmad, K. A. Ismail, F. Mat, IMPACT
MODELS AND COEFFICIENT OF RESTI-
TUTION: A REVIEW, ARPN Journal of En-
gineering and Applied Sciences, 11(10), 2016,
6549-6555.
[11] A. Aryaei, K. Hashemnia, K. Jafarpur, Expe-
rimetal and Numerical Study of Ball Size Ef-
fect on Restitution Coefficient in Low Veloci-
ty Impacts, International Journal of Impact
Engineering, 37, 2010, 1037-1044.

[12] H. Minamoto, K. Onodera, S. Kawamura, E.
Inamura, The Effect of Ball Size on the Re-
bound Property and Elastic Wave Propagation
in Two-Sphere Impact, Proc. 20th Interna-
tional Congress on Sound and Vibration,
Bangkok, Thailand, 2013, 419-426.
[13] B. Hopkinson, A Method of Measuring the
Pressure in the Deformation of High Explo-
sives or by the Impact of Bullets, Phil. Trans.
Royal Soc., A213, 1914, 437-452.
[14] H. Minamoto, Elasto/Visco-Plastic Impact of
Two Equivalent Spheres Made of SUJ2,
Transactions of the Japan Society of Mechan-
ical Engineers, Series C, 71 (701), 2005, 54-
57 (in Japanese).
[15] LS-DYNA KEYWORD USER'S MANUAL,
Version 971 (Livermore Software Technology
Corporation, 2007).
[16] LS-DYNA THEORY MANUAL (Livermore
Software Technology Corporation, 2006).
[17] The Japan Society of Mechanical Engineers
(Ed.), JSME Mechanical Engineers' Hand-
book, 4th revised edition (The Japan Society
of Mechanical Engineers, 1967) (in Japanese).

Rebound Property in Low Velocity Impact of Two Equivalent Balls

More Related Content

What's hot (20)

Similar to Rebound Property in Low Velocity Impact of Two Equivalent Balls (20)

Recently uploaded (20)

Rebound Property in Low Velocity Impact of Two Equivalent Balls