EARTHQUAKE ENGINEERING AND STRUCTURAL DYNAMICS

Earthquake Engng Struct. Dyn. 2009; 38:11351142

Published online 23 December 2008 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/eqe.883

SHORT COMMUNICATION

A note on the Hertz contact model with nonlinear damping

for pounding simulation
Kun Ye, , Li Li and Hongping Zhu
Hubei Key Laboratory of Control Structure, College of Civil Engineering and Mechanics, Huazhong University
of Science and Technology, Wuhan 430074, China

SUMMARY
Structural pounding during earthquakes has been recently investigated extensively by using different
models of impact force. In this paper, reexamination of the Hertz contact model with nonlinear damping
is made. Based on this reexamination, the formula used to determine the damping constant in terms of the
spring stiffness, the coefficient of restitution and relative approaching velocity of two colliding bodies is
found to be incorrect for pounding simulation in structural engineering. In order to correct this problem,
a more accurate approximating formula for the damping constant is theoretically derived. The correctness
of the derived analytical formula has been confirmed through numerical simulations. Copyright q 2008
John Wiley & Sons, Ltd.
Received 19 March 2008; Revised 8 October 2008; Accepted 17 November 2008
KEY WORDS:

structural pounding; earthquake; Hertz model; nonlinear damping; coefficient of restitution

1. INTRODUCTION
Structural damage due to earthquake-induced pounding between closely spaced structures, or
structural components, has been identified in several severe earthquakes. Structural pounding is
a complex phenomenon that makes the mathematical analysis of this type of problem difficult.
In order to study the global structural response due to pounding during earthquakes, different
pounding analytical models have been used or developed by some researchers. The most widely
used pounding analytical models include the restitution-based stereo-mechanical model [1], and
the contact force-based linear elastic model, nonlinear elastic model (Hertz model [2]), linear
viscoelastic model (Kelvin model [3]) and nonlinear viscoelastic model proposed by Jankowski [4].
Correspondence

to: Kun Ye, Hubei Key Laboratory of Control Structure, College of Civil Engineering and Mechanics,
Huazhong University of Science and Technology, Wuhan 430074, China.

E-mail: [email protected]

Copyright q

2008 John Wiley & Sons, Ltd.

1136

K. YE, L. LI AND H. ZHU

However, some drawbacks can be found in these pounding analytical models. Stereo-mechanical
approach is not valid if the impact duration is large enough so that significant changes occur in the
configuration of the system. Energy loss during impact is not taken into account in the linear elastic
and Hertz model. Kelvin model exhibits an initial jump of the pounding force values upon impact
due to the damping term. Furthermore, the damping force can cause negative pounding forces
before separation that pull the colliding bodies. In the case of nonlinear viscoelastic model, the
pounding force-time curve is not smoothly varied between the approaching and restitution phase
during the collision. Different from aforementioned pounding analytical models, Hertz model with
nonlinear damping (Hertz damp model), already used in other areas such as robotics and multi-body
systems [5], was first introduced by Muthukumar and DesRoches [6] to pounding simulation in
structural engineering. The contact force during impact of Hertz damp model can be expressed as
Fc = kh 3/2 +ch 

(1)

where kh is the nonlinear impact spring stiffness, ch is the nonlinear damping coefficient,  is
the penetration and  is the penetration velocity. The nonlinear damping coefficient is taken as
follows [5]:
ch = 3/2

(2)

where  is the damping constant; an expression for the damping constant  can be found in terms
of kh , the coefficient of restitution e and the relative approaching velocity of two colliding bodies
v1 v2 as follows [5]:
=

3kh (1e2 )
4(v1 v2 )

(3)

According to the classical theory of impact, the coefficient of restitution e can be obtained from
the equation
e=

v2 v1
v1 v2

(4)

where v1 and v2 are post-impact velocity of two colliding bodies. According to the statement of
Jankowski in [7], in the case of the nonlinear viscoelastic model, a value of  = 0 stands for a fully
elastic collision and a value of  stands for a fully plastic one. Although in the nonlinear
viscoelastic model the restitution period of collision is considered to be fully elastic, it is believed
that similar relations may also concern the Hertz damp model. However, the formula of damping
constant  in Equation (3) will provide the following surprising relationship:
e=1 =0
e = 0  =

(5)

If we reexamine the derivation of the formula for damping constant  in Hertz damp model,
we will find that the derivation is based upon the following two assumptions: (1) the energy
dissipated during impact is small compared with the maximum absorbed elastic energy; (2) the
penetration velocities during the compression and restitution phases are approximately equal. In
other words, the Hertz damp model is valid only for the case of e approximating one. According
to the conclusions made by Anagnostopoulous in [8], the coefficient of restitution used to simulate
Copyright q

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:11351142

DOI: 10.1002/eqe

A NOTE ON THE HERTZ CONTACT MODEL WITH NONLINEAR DAMPING

1137

real collision in structural engineering varies in the range of 0.5 and 0.75, which indicates that the
Hertz damp model using Equation (3) to determine the damping constant  is incorrect for the
pounding simulation in structural engineering.
In order to remedy the relationship in (5) and make the Hertz damp model valid for the pounding
analysis in structural engineering, the corrected expression for the damping constant  should be
derived again, which is presented in the following section.
CORRECTED DERIVATION OF EXPRESSION FOR DAMPING CONSTANT 
According to stereo-mechanical model, the energy loss E during the impact can be expressed
in terms of the coefficient of restitution e and the approaching velocities v1 v2 of two colliding
bodies as follows:
E =

1 m1m2
(1e2 )(v1 v2 )2
2 m 1 +m 2

(6)

where m 1 , m 2 are the masses of the colliding bodies. The energy dissipated by the damping force
can be evaluated as [5]


E = ch  d = 3/2  d
(7)

where refers to the integration around a hysteresis loop during impact.
In order to evaluate the energy loss from Equation (7), the penetration velocity  must be
expressed as a function of penetration  at any time t during the period of contact. The variation
of  with time is illustrated in Figure 1, where t , t max and t + denote initial time of contact, time
of maximum penetration max and the time of separation of two colliding bodies. At the end of
the compression phase, the two bodies move with a common velocity V .
Owing to the nonlinearity of the Hertz damp model, there is no exact formula for the penetration
velocity during impact. However, we can try to use approximate functions to describe the relation
between penetration velocity  and penetration .

Figure 1. Typical curve of penetration versus time [9].

Copyright q

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:11351142

DOI: 10.1002/eqe

1138

K. YE, L. LI AND H. ZHU

m1

(a)

v1

m2

v2

(b)

v1 - v2

k
c

Figure 2. Equivalent model of impact between two colliding bodies.

As a matter of fact, through the transformation of reference system, the impact between two
colliding bodies (shown in Figure 2(a)) can be equivalently modeled as response of single-degreeof-freedom (SDOF) system (depicted in Figure 2(b)) with the initial penetration displacement
0 = 0 and initial penetration velocity  0 = v1 v2 , the equation of motion of such an SDOF system
is written as

m +c+k
=0

(8)

where m is the mass of equivalent SDOF system (= m 1 m 2 /(m 1 +m 2 )), damping coefficient and
stiffness of equivalent SDOF system are assumed to be linear in order to conveniently search
the approximate relation between  and . Thus, the solution to Equation (8) is expressed as
 = er t

 0
sin(d t),
d

 0
 = er t  0 cos(d t)r er t
sin(d t)
d

(9)

where  (= k/m) is the radial frequency, r (= c/2m) is the damping ratio and d (=  1r 2 )
is the damped radial frequency. If the damping effect (i.e. r = 0) is further neglected, t max and
max can be easily determined as

 0
, max =
2

Based upon Equations (9) and (10), the following expression can be obtained:

  2
 2

+
=1
max
 0
t max =

(10)

(11)

According to Equation (11), the relation between  and  can be approximately considered to
be elliptic. Hence, the penetration velocity  can be related to  at the approaching period and
restitution period, respectively, as follows:


2




 2
 =  0 1
,  =  f 1
(12)
max
max
where  f (= v1 v2 <0) is the post-impact (final) relative velocity between two colliding bodies.
Substituting Equations (12) into (7), we can get


2



max
max

 2
3/2
3/2
E = E 1 +E 2 =
 0 1
d+
 |f | 1
d (13)
max
max
0
0
Copyright q

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:11351142

DOI: 10.1002/eqe

A NOTE ON THE HERTZ CONTACT MODEL WITH NONLINEAR DAMPING

1139

The first term on the right-hand side of Equation (13) E 1 is equal to the dissipated energy due to
the damping force during compression period and the second term E 2 stands for the dissipated
energy at the restitution period. After integral transformation, we have
1

5/2

E = (0 +|f |)max

x 3/2 1 x 2 dx
(14)
0

Applying the numerical integration to evaluate the value of the integral term in Equation (14),
we have
1

1
(15)
x 3/2 1 x 2 dx = 0.2497
4
0
Using the definition of coefficient of restitution and the numerical result from(15), we arrive at
1
5/2
E = 14 ( 0 +| f |)5/2
max = 4 (1+e)0 max

(16)

Considering the momentum and energy balance between the start and the end of the compression
phase, we have
2 1
2
1
2 m 1 v1 + 2 m 2 v2

= Um +E 1 + 12 (m 1 +m 2 )V 2

m 1 v1 +m 2 v2 = (m 1 +m 2 )V

(17)
(18)

where Um and E 1 are the stored maximum strain energy and the dissipated energy due to the
damping force during compression period, respectively. Their mathematical expressions are as
follows:
max
2
Um =
kh 3/2 d = kh 5/2
max
5
(19)
0
5/2
1

E 1 = h 0 
max

If the energy dissipated during impact is not neglected, the energy balance between the start and
the end of the compression phase will result in the reformulation of Equation (17)
5/2
2 1
2
2
1
2
1 5/2
1
2 m 1 v1 + 2 m 2 v2 = 5 k h max + 4 0 max + 2 (m 1 +m 2 )V

(20)

Combining Equations (18) and (20), the expression for 5/2

max can be obtained as
5/2
max =

20
1 m1m2
(v1 v2 )2

2 m 1 +m 2
50 +8kh

(21)

Substituting for 5/2

max from (21) and equating energy loss from (16) and (6), a corrected approximate
expression for the damping constant  can be found in terms of the spring stiffness, the coefficient
of restitution and the relative approaching velocity as well
=
Copyright q

2008 John Wiley & Sons, Ltd.

8 kh (1e)
5 e 0

(22)

Earthquake Engng Struct. Dyn. 2009; 38:11351142

DOI: 10.1002/eqe

1140

K. YE, L. LI AND H. ZHU

If we make an observation on formula (22), we can obtain the following results, which provide
the following relationship between  and e:
e=1 =0
e=0 =

(23)

NUMERICAL VERIFICATION
During the derivation of corrected expression for damping constant , in order to search
the approximating relationship between penetration velocity  and penetration , we have reduced
the collision of two bodies into an SDOF model, and the solution for the second-order equation (8)
is basically the undamped free vibration response of a system with linear stiffness. It seems that
considering c = 0 and assuming k linear are unreasonable, when the problem under consideration
has nonlinear stiffness and some damping representing energy loss. Therefore, it is necessary to
verify the correctness of this formula and the corresponding theoretical derivation.
Both the initial and corrected formula for damping constant  are related to the coefficient of
restitution e; therefore, the verification can be carried out by using the following procedures: (a)
selecting a case of pounding simulation; (b) pre-specifying a value of coefficient of restitution epre ;
(c) conducting the pounding simulation; (d) calculating the coefficient of restitution epost from the
results of the proceeding step; (e) evaluating the difference between epre and epost by computing
the relative error (= |epre epost |/|epre |). Theoretically, epost should be equal to epre . Repeating the
steps from (b) to (e), comparisons in the case of different epre can be made.
The impact between a falling ball and a stationary rigid surface, shown in Figure 3, is selected
as the case of pounding simulation. The dynamic equation of motion for such a model can be
written as [4]
m y + F = mg

(24)

where m is the mass of the ball, y its vertical acceleration, g stands for the acceleration of gravity
and F is the pounding force, which is equal to zero when yh (h is the drop height) and is defined
by Equation (1) when y>h, where penetration  is expressed as
 = y h

(25)

In the numerical analysis, the following parameters have been used:

m = 0.5 kg,

h = 0.5 m,

kh = 2.0109 N/m3/2

m
y
h

Figure 3. Model of a ball falling on a stationary rigid surface.

Copyright q

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:11351142

DOI: 10.1002/eqe

A NOTE ON THE HERTZ CONTACT MODEL WITH NONLINEAR DAMPING

1141

Table I. Comparison of results using different formulas for damping constant.

Based on corrected formula (22)

Based on initial formula (3)

epre

epost

Relative error (%)

epost

Relative error (%)

0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0

0.0687
0.1549
0.2556
0.3624
0.4700
0.5769
0.6829
0.7883
0.8938
1.0000

31.35
22.56
14.29
9.41
6.00
3.85
2.44
1.46
0.69
0.00

0.6650
0.6720
0.6840
0.7014
0.7250
0.7559
0.7958
0.8470
0.9131
0.0000

564.98
235.99
127.98
75.34
45.00
25.99
13.68
5.87
1.46
0.00

The predefined coefficient of restitution epre varies from 0.1 to 1.0 with an interval 0.1. Fourth-order
RungeKutta method with adaptive time step has been applied to solve the equation of motion
(24) numerically.
Comparison of results using initial and corrected formula for the damping constant  is listed
in Table I. It can be seen from the table that difference between epost and epre based on either the
corrected formula or initial formula increases with the decrease of epre ; however, epost obtained
by using corrected formula is much closer to epre . If using initial formula, with the decrease of
epre , difference between epost and epre becomes so great that results of pounding simulation will be
completely unreliable, which further proves incorrectness of formula (3). Difference between epost
and epre given by the corrected formula is relative small in the case of epre >0.4 and somewhat large
in the case of epre 0.4, which indicates that the approximating relationship between penetration
and penetration velocity is rational for the case of epre >0.4 and unsuitable for the case of epre 0.4.
According to the conclusions in [8], the coefficient of restitution used to simulate real collision
between structures varies in the range of 0.5 and 0.75. Thus, reliable results of pound simulation
in structural engineering can be provided by using the corrected formula. Although the assumption
that c is equal to zero and k is supposed to be linear in the equivalent SDOF system is unreasonable,
the target using the unreasonable assumption is to obtain approximating relationship between
penetration and penetration velocity. Through the numerical verification, it can be found that such
unreasonable assumption used in the theoretical derivation can be acceptable.

CONCLUSIONS
In the Hertz contact model with nonlinear damping, the expression for the damping constant is found
to be wrong not only by theoretical analysis but also through numerical verification. To correct this
error, a more accurate approximate formula for the damping constant  is theoretically derived.
Through numerical analysis, the correctness of formula (22) and its corresponding theoretical
derivation has been verified. More reliable results of pound simulation in structural engineering
can be provided by using the Hertz damp model with the corrected formula for damping constant.
Copyright q

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:11351142

DOI: 10.1002/eqe

1142

K. YE, L. LI AND H. ZHU

REFERENCES
1. Goldsmith W. Impact: The Theory and Physical Behaviour of Colliding Solids [M]. Edward Arnold: London,
England, 1960.
2. Davis RO. Pounding of buildings modelled by an impact oscillator. Earthquake Engineering and Structural
Dynamics 1992; 21(3):253274.
3. Anagnostopoulos SA. Pounding of buildings in series during earthquakes. Earthquake Engineering and Structural
Dynamics 1988; 16(3):443456.
4. Jankowski R. Non-linear viscoelastic modelling of earthquake-induced structural pounding. Earthquake Engineering
and Structural Dynamics 2005; 34(6):595611.
5. Lankarani HM, Nikravesh PE. Contact force model with hysteresis damping for impact analysis of multibody
systems. Journal of Mechanisms, Transmissions, and Automation in Design 1990; 112(3):369376.
6. Muthukumar S, DesRoches R. A Hertz contact model with non-linear damping for pounding simulation. Earthquake
Engineering and Structural Dynamics 2006; 35(7):811828.
7. Jankowski R. Analytical expression between the impact damping ratio and the coefficient of restitution in the
non-linear viscoelastic model of structural pounding. Earthquake Engineering and Structural Dynamics 2006;
35(4):517524.
8. Anagnostopoulos SA, Spiliopoulos KV. An investigation of earthquake induced pounding between adjacent
buildings. Earthquake Engineering and Structural Dynamics 1992; 21(4):289302.
9. Muthukumar S. A contact element approach with hysteresis damping for the analysis and design of pounding in
bridges. Ph.D. Dissertation, Department of Civil and Environmental Engineering, Georgia Institute of Technology,
2003.

Copyright q

2008 John Wiley & Sons, Ltd.

Earthquake Engng Struct. Dyn. 2009; 38:11351142

DOI: 10.1002/eqe