archives of civil and mechanical engineering 14 (2014) 354–359

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: http://www.elsevier.com/locate/acme

Original Research Article

Method of identifying nonlinear characteristic of

energy dissipation in dynamic systems with one
degree of freedom

M. Bocian *, M. Kulisiewicz
Wroclaw University of Technology, Institute of Materials Science and Applied Mechanics, Smoluchowskiego 25 Str.,
50-370 Wroclaw, Poland

article info abstract

Article history: A method of determining the shape of the vibration damping characteristic in systems with
Received 15 May 2013 one degree of freedom in the case when the characteristic depends not only on the velocity
Accepted 13 November 2013 but also in an unknown way depends on the displacement has been developed. The method
Available online 15 December 2013 is intended for determining the specific form of the mathematical function describing this
dependence. The method utilizes an appropriate analysis of experimentally determined
Keywords: traces of free vibrations of the system. The method has been verified on a few selected
Nonlinear mechanical systems computer systems.
Dynamics # 2013 Politechnika Wrocławska. Published by Elsevier Urban & Partner Sp. z o.o. All
Vibrations rights reserved.

materials is not linear. This is observed already for quasi-static

1. Introduction
The dynamic properties of most construction materials are
usually described by a stiffness parameter and a viscous
damping parameter determining the rate of energy dissipa-
tion. The parameters stem from the traditionally adopted
rheological Kelvin model [1], which was the basis for
developing, among other things, the well-known method of
analyzing dynamic mechanical systems, called experimental
modal analysis [2–4]. However, the analytical description of
the dynamic properties of novel construction materials as well
as biological materials (biomaterials) often poses difficulties
due to the fact that the strain–stress dependence in these
loads inducing deformations with given constant velocities.
Moreover, the strain–stress dependences obtained at constant
velocities often depend on the velocity with which specific
strength tests (tensile, compressive, torsional, etc.) are con-
ducted while the moduli of elasticity determined in this way
often significantly depend on the rate of deformation and
change depending on the deformation level (nonlinear
dependence). This is observed not only in the case of many
non-metallic construction materials (plastics and composites),
but also for the deformation of biological materials, e.g. human
and animal bones [5–9]. This indicates that the linear Hooke
model which is traditionally adopted in the mechanics of
materials is unreliable for a very wide range of materials.

* Corresponding author. Tel.: +48 71 320 27 54.

E-mail address: miroslaw.bocian@pwr.wroc.pl (M. Bocian).
1644-9665/$ – see front matter # 2013 Politechnika Wrocławska. Published by Elsevier Urban & Partner Sp. z o.o. All rights reserved.
http://dx.doi.org/10.1016/j.acme.2013.11.006
 archives of civil and mechanical engineering 14 (2014) 354–359
Fig. 1 – Example of dynamic system in which tested material
acts on concentrated mass m in accordance with Eq. (1).
Hence there is a need to look for new rheological models of
materials and to develop new identification methods for
them.
The idea put forward by the authors of this paper is to build
such a (possibly simplest) dynamic system whose elasto-
damping element would be wholly made of the tested material
while the movement of concentrated mass m attached in a
specified point to this element would be described by a simple
differential equation in the form
€ þ sð?Þ ¼ pðtÞ
mx (1)
where x stands for time function x(t) describing the movement
of mass m, and p(t) is the exciting force. Moreover, it is assumed
that the action of the tested elasto-damping element on mass
m is described by unknown force S which is a function of
displacement x and velocity v. An example of such solution
can be a beam made of the tested material that is rigidly
constrained at one end and a concentrated mass m is attached
to the other end (Fig. 1). If mass m jest is sufficiently large in
comparison with the mass of the beam, then one can assume
that Eq. (1) sufficiently accurately describes the vibrations of
mass m. In this paper it is assumed, a priori, that force S(x,v)
has the form
Sðx; vÞ ¼ ½k þ kðxÞv þ f s ðxÞ (2)
where k(x) is a certain unknown displacement function satis-
fying the condition
kðx ¼ 0Þ ¼ 0 (3)
It is apparent that for k(x) = 0 and fs(x) = cx relation (2) defines
force S in the Kelvin model, i.e. the model commonly used in
engineering practice to describe the mechanical vibrations of
dynamic systems. Term k(x) (in any form) was introduced
because damping properties often depend on the level (state)
of material deformation. For example, the resistance of the
fluids moving in the canaliculi of a deformed bone is greater
than in an undeformed bone since the canaliculi in the former
bone are narrowed.
To sum up, a dynamic model of a system described by the
equation
€ þ ½k þ kðxÞx_ þ f s ðxÞ ¼ pðtÞ
mx (4)
was adopted in this paper. A scheme of the model is shown in
Fig. 2.
The aim of the method described below is to determine
constant k and function k(x) for arbitrarily nonlinear elasticity
characteristic fs(x) describing pure elastic interactions. Eq. (4)
355
Fig. 2 – Scheme of adopted model.
leads to a qualitatively new process modeling of vibration of
mechanical systems similarly as it have place today in
mechatronics [10,11].
2. Theoretical basis of the method
Differential equation (4) describes the motion of concentrat-
ed mass m in the adopted dynamic model having the form
shown in Fig. 2. In the case of nonlinear and unknown
functions k(x), fs(x), it is impossible to obtain solution x(t) of
this equation. One can notice, however, that for oscillating
motion there exist such instants ti in which acceleration x €ðtÞ
is equal to zero. In such instants velocity v ¼ x, _ regardless of
the form of the solution, must reach extreme values since
€ ¼ dv=dt.
x
In order to facilitate the analysis let us assume that p(t) = 0,
which corresponds to free vibrations. Thus for instant t = ti one
gets:
€ðti Þ ¼ 0; xðt
x _ i Þ ¼ vðti Þ ¼ vi ; xðti Þ ¼ xi (5)
According to differential equation (4), values vi, xi for p(t) = 0
must satisfy the following algebraic equation:
½k þ kðxi Þvi þ f s ðxi Þ ¼ 0 (6)
Hence the following relation is obtained:
f s ðxi Þ
 ¼ k þ kðxi Þ ¼ yðxi Þ (7)
vi
where y(x) is a certain unknown function of variable x. As one
can notice, the graph of this function corresponds to that of the
characteristic k(x) shifted parallel by constant k.
Relation (7) can be used to determine constant k and
function k(x) in a given interval of variable x if:
 a certain set of data xi, vi is experimentally determined (by
measurement) for values sufficiently densely filling the
given interval of variable x,
 function fs(x) is known (e.g. has been determined by static
measurements) in the given interval of variable x.
356 archives of civil and mechanical engineering 14 (2014) 354–359

Knowing the shape of the diagram of yi versus xi pðxÞ

f s ðxÞ ¼ (12)
½k þ kðxÞv0
f ðx Þ
yi ¼  s i (8)
vi
It is apparent that if the form of function k(x) is unknown,
one can adopt a certain class of function k(x) (a linear function, function fs(x) cannot be determined from quasi-static mea-
a polynomial of any degree, an exponential function, etc.) and surements. However, the lower the given velocity v0, the more
approximate it for the experimental data in the given interval similar the relation p(x) to sought function fs(x) since according
of variable x. to relation (11) the following holds true:

limv0 ! 0 pðxÞ ¼ f s ðxÞ (13)

3. Description of the application of the method
Hence the best solution for estimating characteristic fs(x) in
this case seems to be to use the static method and relation (9).
The basis for the use of the described method is the precise Having obtained the static characteristic fs(x) one should
determination of static characteristic fs(x) in the given interval make the system vibrate freely by setting different (e.g.
of changes in displacement x. In the case of static tests randomly selected) initial conditions:
consisting in successively applying static loads pv = const
xð0Þ ¼ x0 ; vð0Þ ¼ v0 (14)
(v = 1, 2, . . ., n) and recording the constant displacements xv at
which x_ v ¼ 0, one directly gets and then from the time functions x €ðtÞ; xðtÞ;
_ xðtÞ recorded for
them one should select the values of vi, xi in time instants ti, for
f s ðxv Þ ¼ pv (9)
which x €ðti Þ ¼ 0, to obtain the following two-column matrix:
whereas in the case of quasi-static tests in which such load p(t) 2 3
v1 x1
is applied at which displacements increasing at constant rate 6 v2 x2 7
v0 are obtained (as in typical strength tests), one gets: E0 ¼ 6 7
4 . . . . . . 5 ¼ ½vi ; xi  ði ¼ 1; 2; . . . ; sÞ (15)
vs xs
xðtÞ ¼ v0 t (10)

(see Fig. 3). Substituting the values in the successive rows of

½k þ kðxÞv0 þ f s ðxÞ ¼ pðxÞ (11)
this matrix into relation (8) one gets appropriate values yi of
hence variable y, forming the following matrix Ec

Fig. 3 – Illustration of method of determining values xi, vi from free vibration diagrams.
 archives of civil and mechanical engineering 14 (2014) 354–359 357

Fig. 4 – (a) Exemplary free vibration diagrams for System 1: x0 = 1.5, c3 = 0. (b) Exemplary free vibration diagrams for System 2:
x0 = 1.5, c3 = 10,000. (c) Exemplary free vibration diagrams for System 3: x0 = 1.5, c3 = 0, ABS(kappa). (d) Exemplary free
vibration diagrams for System 4: x0 = 1.5, c3 = 10,000, ABS(kappa).

2 3
y1 x1
6y x2 7
Ec ¼ 6 2
4...
7 ¼ ½y ; xi  ði ¼ 1; 2; . . . ; sÞ (16)
...5 i 4. Computer verification of the method
ys xs

Examining relation yi(xi) in the form of a two-dimensional
diagram of dependent variable y versus independent variable
x one should:
a. select a particular shape of function k(x),
b. calculate the constant parameters of the selected function
(including parameter k), using the well-known methods of
parametric optimization (e.g. nonlinear regression analysis
[12]).
The practical functioning of the above method was tested on
four dynamic systems described by Eq. (4). The verification
tests for the systems were carried out for both linear and
nonlinear functions fs(x), k(x) and the numerical data are
shown in Table 1.
For the tests it was assumed that the elasticity character-
istics for all the systems were precisely determined. The aim of
the tests was to determine constant k and function k(x). The
systems were built using the Matlab Simulink software.

Table 1 – Numerical data on tested systems.

m [kg] c1 [N/m] c3 [N/m3] k [Ns/m] k0 k(x)
System 1 20 800 0 40 600 k0 x v
System 2 20 800 0 40 600 k0 x jxj v
System 3 20 800 10,000 40 600 k0 x v
System 4 20 800 10,000 40 600 k0 x jxj v

Table 2 – Results of identification.

k [N/m] k0
System 1 (c3 = 0) 40.00  0.01 599.98  0.13
System 2 (c3 = 0) 40.00  0.01 603.12  14.79
System 3 40.00  0.01 600.07  0.16
System 4 40.00  0.01 590.87  16.24
358 archives of civil and mechanical engineering 14 (2014) 354–359
Fig. 5 – Relations yi(xi) for tested systems and their approximations (continuous lines).
Different initial conditions for displacement x0 in the systems
were set at v0 = 0 and p(t) = 0. In total, 15 different initial
conditions for x0 ranging from x0 = 0.01 to x0 = 0.15 at a step of
0.01 were set for each of the systems. Exemplary time
diagrams for selected initial conditions for all the tested
systems are shown in Fig. 4.
Four values of xi in instants ti determined in the previous
section of this paper were selected for each of the initial
condition, using a dedicated software. In this way about 60
values xi (i = 1, 2, . . ., 60) and extreme velocity values vi were
obtained for which the corresponding values of yi were
calculated from relation (8). Diagrams of relation yi(xi) for
each of the systems are shown in Fig. 5. By approximating
these relations by means of the regression analysis method
the results presented in Table 2 were obtained.
5. Conclusions
The tests have shown that the proposed method functions
properly. The values obtained for the systems with linear
characteristics k(x) are practically equal to the given values,
regardless of the shape of function fs(x). Slightly larger scatter
of results was obtained for the systems with nonlinear
characteristics k(x) (systems 2 and 4). It should be noted that
the precise synchronization of the displacement, velocity and
acceleration traces in time is critical for the accuracy of the
obtained results. Such synchronization can be difficult in the
case of real measurements performed on real objects, which
requires further research. It would be also interesting to
extend the developed method for the so-called degenerated
systems which are used, among other things, in the process of
piercing ballistic shields [13].
references
[1] H.A. Barnes, J.F. Hutton, K. Walters, An Introduction to
Rheology, Elsevier Science B.V., Amsterdam, Lausanne,
New York, Oxford, Shannon, Singapore, Tokyo, 1989.
[2] D.J. Ewins, Modal Testing, Theory, Practice and Application,
second ed., Research Studies Press Ltd., Baldock, Hertfordshire,
England, 2000.
[3] Structural Testing. Part 1. Mechanical Mobility
Measurements, Bruel & Kjaer, 1988.
[4] Structural Testing. Part 2. Modal Analysis and Simulation,
Bruel & Kjaer, 1988.
[5] R. Skalak, S. Chien, Handbook of Bioengineering, McGraw-
Hill Book Company, New York, 1979.
[6] F.G. Evans, Mechanical Properties of Bone, Charles C.
Thomas, Springfield, IL, 1973.
[7] D.T. Reilly, A.H. Burstein, V.H. Frankel, The elastic modulus
for bone, Journal of Biomechanics 7 (3) (1974) 271–275.
[8] J.H. McElhaney, Dynamic response of bone and muscle tissue,
Journal of Applied Physiology 21 (1966) 1231–1236.
 archives of civil and mechanical engineering 14 (2014) 354–359 359

[9] S.S. Sarva, S. Deschanel, M.C. Boyce, C. Weinong, Stress–
strain behavior of polyurea and polyurethane from low to
high strain rates, Polymer 48 (2007) 2208–2213.
[10] A. Buchacz, M. Płaczek, Development of mathematical model
of a mechatronic system, Solid State Phenomena 164 (2010)
319–322.
[11] A. Buchacz, D. Galeziowski, Synthesis as a designing of
mechatronic vibrating mixed systems, Journal of
Vibroengineering 14 (June (2)) (2012) 553–559 (no. of
publication 781).
[12] J.J. Johnston, Econometric Methods, McGraw-Hill Book
Company Inc., New York, 1963.
[13] K. Jamroziak, Analyis of a degenerated standard model in the
piercing process, Journal of Achievements in Materials and
Manufacturing Engineering 22 (1) (2007) 49–52.