Upload
others
View
7
Download
0
Embed Size (px)
Citation preview
Proceedings of the 7th International Conference on Mechanics and Materials in Design
Albufeira/Portugal 11-15 June 2017. Editors J.F. Silva Gomes and S.A. Meguid.
Publ. INEGI/FEUP (2017)
-313-
PAPER REF: 6801
EXPERIMENTAL EVALUATION OF THE DAMPING PROPERTIES
AND OPTIMAL MODELLING OF COATINGS MADE BY PLASMA
DEPOSITION TECHNIQUES
Stefano Amadori(*), Giuseppe Catania
Department of Industrial Engineering - Ciri MaM, University of Bologna, Bologna, Italy (*)
Email: [email protected]
ABSTRACT
Coating layers, applied by means of different deposition technologies, can be used to produce
composites with high damping properties. This is well known in literature and it is also
verified by some experimental results reported in this work. The constitutive equivalent
material model of coated specimens is identified and optimized by means of a generalized
Kelvin model of n-th order, defined by means of the ratio of polynomials in the frequency
domain. Dynamical measurements data obtained from coated single-layered and multi-layered
samples obtained with different deposition techniques, are used to identify the optimal
material model order and parameters. A robust identification technique that makes use of
Forsythe orthogonal polynomials is employed for the numerical identification of the model
parameters and a specific technique is introduced to eliminate the model non-physical
components generated by measurement and model noise.
Keywords: damping, model identification, dynamical measurement, optimal model fitting.
INTRODUCTION
There is a great demand in industrial, aerospace and automotive mechanical applications for
non-conventional composite materials suitable to design components showing high stiffness,
high resistance and high damping behaviour. Coating layer technology can be employed to
increase the global dissipative properties of an industrial component with limited influence on
the other mechanical properties (Tassini, 2006; Yu, 2005; Ustinov, 2001). One or more
coating layers can be deposited or grown in order to obtain a finished composite component
with specifically designed characteristics. Different techniques are known (Rongong, 2004;
Blackwell, 2007), for example plasma vapor deposition, plasma spray or the anodization
process. When designing or analyzing the properties of a single or multi-layer coating, e.g.
damping efficiency, corrosion resistance, or thermal insulation, many factors must be taken
into account: layer composition, layer or multi-layer structure, deposition or growing
technology and adhesion to the substrate.
The dissipative properties of a component can be significantly tailored by means of the
adoption of layers showing high internal hysteresis or by maximizing the frictional actions at
the interface between the different layers (Casadei, 2014; Wang, 2013; Kireitseu, 2007). A
considerable number of applications was dedicated to study the influence of the coating
characteristics, such as the coating material structure (Casadei, 2014; Guangyu 2014;
Ustinov 2008) or the interface structure (Colorado, 2006) on the coated component damping
behaviour and its temperature dependence (Khor, 2001).
Topic-B: Experimental Mechanics
-314-
Dynamic mechanical measurements can be used as an effective experimental technique for
estimating the coated component damping properties. Forced and free vibration experimental
methods were successfully employed to estimate the component dissipative properties by
means of comparing the coated and uncoated component dynamical response (Reed, 2008;
Pastias, 2004; Torvik, 2009).
Fig. 1 - Generalized Kelvin model of order n
Single or multi-layered coated materials dynamic behavior can be considerably complex and
classical material models (Zener, 1948; Nowick, 1972) may not accurately describe the coated
component constitutive relationship (Amadori, 2016). Different higher order constitutive
models can be adopted (Finley, 1989; Ramirez, 2007; Vasquez, 2010). Such models may
require the use of not trivial constitutive equations and as a consequence a robust
identification procedure must be considered for solving the generally ill-conditioned problem
of parameter identification (Richardson, 1982; Amadori, 2016).
In this work, components made by single-layer and multi-layer coatings applied on a metal
substrate by means of two different techniques, i.e. by an anodizing process and by a reactive
plasma vapor deposition (RPVD) process, are considered. Coated and uncoated specimen
dynamic mechanical measurement test results are compared and processed by means of a
robust parameter identification and model condensation technique.
IDENTIFICATION PROCEDURE
The procedure used to identify the composite constitutive relationship, defined by the authors
of this work in a previous paper (Amadori, 2016), is briefly outlined here. A generalized
Kelvin model (also cited in literature as the standard linear solid model) is chosen to model
the composite material constitutive equation. Since low order material models such as the
Zener and Kelvin-Voight models can show limitations in predicting the dynamic behaviour of
complex materials, high order models are taken into account in this work. A generalized
Kelvin model of order n, obtained by connecting n single Kelvin model units in series is
shown in Fig.1, and starting from the relationship between the Fourier transform ( )$ of σstress and ε strain of the single Kelvin unit in the frequency domain:
ˆˆ ( ) , 1, ... ,i i iE j i nσ ω β ε= + ⋅ ⋅ = . (1)
The constitutive relationship for the generalize Kelvin model is:
( ) ( )ˆ ˆˆ E E Cσ ω ε ω ε= ⋅ = ⋅ ⋅ (2)
where:
Proceedings of the 7th International Conference on Mechanics and Materials in Design
-315-
( )( )
( )
1
1
11
1
11
, =
1
ns
s ns
nr i i
r
r
C EE
α ωω
β ω
−
=−
=
=
+ ⋅
= + ⋅
∑∑
∑ (3)
Fig. 2 - Slender beam in flexural excitation experimental set-up
with clamped-sliding boundary conditions
The constant �� is the modulus of the material when the frequency value is zero and it can be
obtained from static experiments or extrapolated from dynamical measurements estimates as
the frequency goes to zero. C(ω) is a rational function with a n-degree numerator polynomial
and a n-1 degree denominator polynomial. The estimate value of C(ω) in a wide frequency
range can be obtained from dynamic mechanical measurements in a forced flexural excitation
experimental set-up.
The contribution of the inertial actions is also taken into account and for a slender beam in a
flexural experimental set-up with clamped-sliding boundary condition (Fig.2) and with a
periodic force ( ) ( )expq t Q j tω= ⋅ applied at the sliding end, the relationship in the frequency domain between the transverse displacement ( )ν̂ ω and applied force ( )q̂ ω at the beam sliding end is approximated by:
( )
( )
( )( )
2
4 3 21
2 1 cosh cos
sinh sinˆ ˆ
m
s s
ns s
s s
k k
k kv q
E C k I L Mω ω
ω ω=
− ⋅ + ⋅⋅ ⋅ ⋅ − ⋅∑� (4)
M is the beam mass, I is the beam section moment, L is the beam length, ks satisfy the
following equation:
sinh cos cosh sin 0k k k k+ = (5)
Using Eq.4 and the corresponding displacement and applied force data, the material
constitutive relationship Ci=C(ωi) at ωi excitation frequency can be estimated by finding the
solution of the non-linear equation Eq.6:
( )
( )( )( )
2
34 21
2 1 cosh cos
ˆsinh sin0
ˆ
m
s s
ns s i
i
s ii s i
k k
k k vC
qE C k I L M
ωωω=
− ⋅ + Θ = − =⋅ ⋅ ⋅ − ⋅
∑ (6)
L 0
q(t)
Topic-B: Experimental Mechanics
-316-
The solution is found by means of the Newton-Raphson method, the procedure showed to be
stable and only a few iterations are generally needed for convergence.
The Ci experimentally estimated values are used to identify the parameters of the constitutive
relationship by means of an approach based on the least square error technique. The rational
function C(ω) is defined in Eq.7 by means of the ratio of two polynomials expressed in a
basis of orthogonal polynomials ( )s jη ω and ( )r jµ ω :
( )( )
( )1
1
1
1
1
n
s s i
si i n
r r i
r
j
C C
j
γ η ωω
δ µ ω
=−
=
+ ⋅= =
+ ⋅
∑
∑ (7)
Starting from Eq.6 and applying the least square error method the unknown γs and δr
coefficients can be obtained as the solution of two uncoupled systems of linear equations
(Amadori, 2016). With the adoption of an orthogonal polynomial base, and in the specific
case of a Forsythe polynomial base (Forsythe, 1957), high accuracy for the identification of
both low and high order model can be obtained. The polynomial base coefficients γs and δr
can be unsuitable for further processing. It is more convenient to convert them to an
equivalent monomial base, this can be done by means of known algorithms (Kelly, 1967).
This procedure has proved to be robust and the material constitutive relationship C(ω) can be
obtained as the ratio of two polynomials defined by coefficients in a monomial base, as in
Eq.3.
MODEL CONDENSATION
Experimental measurements are affected by noise and the basic least-square error technique is
not able to distinguish between system poles and virtual, non-physical poles resulting from
such noise. High order model fitting estimates then generally result. This model can be further
condensed, by finding an optimized model order and by eliminating the non-physical model
components. The rational function C(ωi) is manipulated in order to outline the system poles
and residues. By inverting C(ωi) and defining it in standard partial fraction form:
( ) ( )11 n s
si i s
R
C jω ω ψ==
−∑ (8)
where residues Rs are computed from the rζ zeros and sψ poles of 1/C(ω). Since from Eq.2,8:
( )( )
( )1
1ˆ ˆ
ns
s s
R
E jε ω σ ω
ω ψ== ⋅ ⋅
−∑ (9)
computational and physical sψ pole and Rs residues subsets can be found by observing that poles and residues are expected to appear as complex conjugated pairs or real values. Poles
with positive real part can be considered as computational poles and can be discarded.
Couples of physical poles and residues are also assumed to present stability if the model order
is increased. Unstable poles with respect to model order n can be considered not physical and
discarded. The model, condensed to a subset of nr physical pole-residues sR% and sψ% , is
expressed in the form:
( )( )
( ) ( ) ( )1 0
1 + ;
nr mrs
m m r
s rs
Rj
C jω ω ω λ ω
ω ψ= =Λ Λ = ⋅
−∑ ∑%
� (8)
Proceedings of the 7th International Conference on Mechanics and Materials in Design
-317-
EXPERIMENTAL SET-UP AND TESTS
Specimens in the form of slender beams of rectangular cross section are tested with a dynamic
mechanical analyzer (Fig.3) in a forced flexural excitation experimental set-up with
clamped-sliding boundary conditions.
(a) (b)
Fig. 3 - Experimental apparatus. Dynamic mechanical analyzer (a) and specimen
mounted in the flexural excitation experimental set-up (b).
Λm(ω) is a m order (m=1,2 generally works well) polynomial function used to estimate the
contribution of the nc=n-nr eliminated computational poles-residue pairs. The calculation
procedure to obtain coefficients λr can be found in the authors previous work
(Amadori, 2016).
Two different coating solutions are considered in this work. In both cases coating layers are
applied on a Al alloy (Al 1000) beam, length 10-2
m, width 3×10-3
m, thickness 5×10-4
m and
density 2700 kg/m3. One set of specimens is obtained by means of the reactive plasma vapor
deposition (RPVD) technique. A dual-layer coating is applied on two opposite faces of the
beam. First a Ti metallic bond coat is deposited and then coated with an external hard coating
layer formed by TiO2 (Fig.4a)..
(a) (b)
Fig. 4 - Schematic representation of the dual-layered structure of Ti+TiO2 coating (a)
and of the single layer Al oxide coating (b).
Both the metallic bond coat and the hard coating layer thickness is 1µm. Other coated
specimens are obtained by means of an anodizing process, the Al 1000 beam is coated on the
two opposite faces of the beam by a single, 4µm thick, Al oxide layer (Fig.4b). Dynamic
Topic-B: Experimental Mechanics
-318-
mechanical test on the coated specimen sets and homogeneous Al beams are made at 0.01%
maximum strain over a [0.01-200] Hz frequency range and 35°C constant temperature.
RESULTS AND DISCUSSION
The ratio of the imaginary and real part of E(ω) (Eq.2) is considered as a measure of the
material dissipative properties (Nowick, 1972), and the results obtained for the uncoated and
the coated specimens are compared. Fig.5 shows the estimated Im(E(ω))/Re(E(ω)) ratio of the
dual-layered (Fig.4a) and of a homogeneous Al 1000 specimen, and the relative increase of
the same ratio.
Fig. 5 - Estimates of the ratio Im(E(ω))/Re(E(ω)) for dual-layer (Ti+TiO2) coated specimen (red)
and homogeneous Al 1000 specimen (green), and their Im(E(ω))/Re(E(ω)) relative difference % (blue).
Fig.6 shows the same results obtained for the single-layered (Fig.4b) specimen. It is shown
that the coated specimen Im(E(ω)/Re(E(ω)) ratio increases with respect to the uncoated case.
The dual-layer coated specimen only exhibits a meaningful effect at frequencies above 70 Hz,
while a significant increase at all measured frequencies can be observed for the single-layer
coated specimen. Since for both coating technologies a very small coating layer thickness,
compared to the beam substrate thickness, is adopted, it can be assumed that this results are
mainly due to the interface frictional actions acting during the imposed flexural test cycle.
The identification and the model condensation procedure are used to process experimental
measurements data, and the results are shown in Fig.7 for the dual-layer specimen and in
Fig.8 for a single-layer specimen. Condensed model fitting estimates of order n=11 are shown
for both specimen solutions and compared with the estimated measurement values. In both
cases a n=43 model order was taken into account as the starting point of the condensation
procedure and a n=11 condensed model is obtained by eliminating all of the poles with
positive real part and the ones not exhibiting stability with respect to the model order increase.
Stability diagrams used to evaluate poles stability are shown in Fig.9.
0 20 40 60 80 100 120 140 160 180 2000
0.05
0.1
0.15
0.2
Im(E
)/R
e(E
)
Frequency [Hz]
0 50 100 150 200
0
100
200
300
400
Frequency [Hz]
Im(E
)/R
e(E
)
Rel
ativ
e D
iffe
ren
ce %
Proceedings of the 7th International Conference on Mechanics and Materials in Design
-319-
Fig. 6 - Estimates of the ratio Im(E(ω))/Re(E(ω)) for single-layer (Al oxide) coated specimen (red)
and homogeneous Al 1000 specimen (green), and their Im(E(ω))/Re(E(ω)) relative difference % (blue).
Fig. 7 - Dual-layer (Ti+TiO2) coated specimen E(ω) estimates (green) and n=11 condensed
model fitting estimate (red).
0 20 40 60 80 100 120 140 160 180 2000
0.05
0.1
0.15
0.2Im
(E)/
Re(
E)
Frequency [Hz]
0 50 100 150 200
0
200
400
600
800
Frequency [Hz]
Im(E
)/R
e(E
)
Rel
ativ
e D
iffe
ren
ce %
0 20 40 60 80 100 120 140 160 180 2005.5
6
6.5
7
7.5
8x 10
10
Re(
E)
[Pa]
Frequency [Hz]
0 20 40 60 80 100 120 140 160 180 200
0
5
10
15
20x 10
9
Im(E
) [P
a]
Frequency [Hz]
Topic-B: Experimental Mechanics
-320-
Fig. 8 - Single-layer (Al oxide) coated specimen E(ω) estimates (green) and n=11
condensed model fitting estimate (red).
(a) (b)
Fig. 9 - Stability diagrams with selected poles (blue and red) of n=11 condensed models
of the dual-layered (Ti+TiO2) (a) and single-layered (Al oxide) (b) specimens.
0 20 40 60 80 100 120 140 160 180 200
6
7
8
9
x 1010
Re(
E)
[Pa]
Frequency [Hz]
0 20 40 60 80 100 120 140 160 180 2000
1
2
3
4x 10
10
Im(E
) [P
a]
Frequency [Hz]
0 500 10000
10
20
30
40
50
60
Poles absolute value [rad/s]
Mo
del
ord
er
0 500 10000
10
20
30
40
50
60
Poles absolute value [rad/s]
Mo
del
ord
er
Proceedings of the 7th International Conference on Mechanics and Materials in Design
-321-
To assess the effectiveness of the model condensation step, Figs.10-11 report the
computational model fitting estimate results of the same order (n=11). It can be observed that
in both the dual-layer and the single-layer solutions the condensed model is a significantly
more accurate approximation of the experimental estimates and that the un-condensed model
fail to accurately model the experimental estimates.
These results show that both coating solutions may affect the specimen dissipative properties.
The integrity of the specimen layer interface showed to be nevertheless affected by cyclic
loading, and preliminary results showed that damages and cracks may appear at the interface
between the different layers, making the solution partially non-effective. Different coating
process parameters and technologies will be investigated in future work to solve this specific
problem.
CONCLUSIONS
A composite solution adopting many coating layers of different materials, may strongly
increase the component structural damping, according to some engineering design
specification constraints.
Two simple solutions are investigated, a dual-layer obtained by means of a thin coating of Ti
and of Ti oxide and a single-layer obtained by means of a thin coating of Al oxide. Coated
specimens show improved damping behaviour with respect to uncoated ones for both the
solutions under study. Preliminary cyclic loading tests show that the component dissipative
properties may decrease due to the formation of damage and cracks at the layer interfaces.
The obtained results are positive enough to justify further investigations. Different
technological and processing solutions will be examined in order to obtain increased damping
behavior by means of frictional actions at the layer interfaces.
Fig. 10 - Dual-layer (Ti+TiO2) coated specimen coated specimen E(ω) estimates (green)
and computational n=11 model fitting (blue)
0 20 40 60 80 100 120 140 160 180 2005.5
6
6.5
7
7.5
8x 10
10
Re(
E)
[Pa]
Frequency [Hz]
0 20 40 60 80 100 120 140 160 180 200
0
5
10
15
20x 10
9
Im(E
) [P
a]
Frequency [Hz]
Topic-B: Experimental Mechanics
-322-
Fig. 11 - Single-layer (Al oxide) coated specimen E(ω) estimates (green)
and computational n=11 model fitting (blue)
ACKNOWLEDGMENTS
This study was developed within the MAM-CIRI with the contribution of the Regione
Emilia Romagna, progetto POR-Fesr-Tecnopoli.
REFERENCES
[1]-Amadori S, Catania G, “Robust identification of the mechanical properties of viscoelatic
non-standard materials by means of time and frequency domain experimental measurements",
Composites Structures, http://dx.doi.org/10.1016/j.compstruct.2016.11.029, (In Press).
[2]-Blackwell C, Palazzotto A, George TJ, Cross CJ, ”The evaluation of the damping
characteristics of hard coating on titanium”, Shock and Vibrations, 2007, 14, p. 37-51.
[3]-Casadei F, Bertoldi K, Clarke DR, “Vibration damping of thermal barrier coatings
containing ductile metallic layers", ASME Journal of Applied Mechanics, 2014, 81,
p. 101001_1-101001_10.
[4]-Colorado HA., Velez J, Salva HR, Ghilarducci AA, “Damping behavior of physical
vapor-deposited TiN coatings on AISI 304 stainless steel and adhesion determination”,
Material Science and Engineering A, 2006, 442, p. 514-418.
[5]-Finley WN, Lai JS, Onaran K, Creep and relaxation of nonlinear viscoelastic materials,
New York: Dover Publications Inc., 1989.
0 20 40 60 80 100 120 140 160 180 200
6
7
8
9
x 1010
Re(
E)
[Pa]
Frequency [Hz]
0 20 40 60 80 100 120 140 160 180 2000
1
2
3
4x 10
10
Im(E
) [P
a]
Frequency [Hz]
Proceedings of the 7th International Conference on Mechanics and Materials in Design
-323-
[6]-Forsythe GE, “Generation and use of orthogonal polynomials for data fitting with a digital
computer” J Soc. Indust. Appl. Math, 1957, 5(2), p. 74-85.
[7]-Guangyu D, Zhen T, Dechum B, Kun L, Qingkai H, “Damping properties of a novel
porous Mg-Al alloy coating prepared by arc ion plating”, Surface&Coatings Technology,
2014, 238, p. 139-142
[8]-Kelly LG, Handbook of Numerical Methods and Applications, Addison Wesley
Publishing Company 1967.
[9]-Kiretseu M, Hui D, Tomlinson G, “Advanced shock-resistant and vibration damping of
nanoparticle-reinforced composite material”, Composites: Part B, 2008, 39, p. 128-138.
[10]-Khor KA, Chia CT, Gu YW, Boey FYC, “High temperature damping behavior of plasma
sprayed NiCoCrAlY coatings”, Journal of Thermal Spray Technology, 2002, 11(3),
p. 359-364.
[11]-Nowick AS., Berry BS, Anelastic Relaxation in Crystalline Solids, Academic Press Inc.,
1972.
[12]-Pastias S, Saxton C, Shipton M, “Hard damping coatings: an experimental procedure for
extraction of damping characteristics and modulus of elasticity”, Material Science and
Engineering A, 2004, 370, p. 412-416
[13]-Ramirez LES, Coimbra CFM, “A variable order constitutive relation for viscoelasticity”,
Annalen der Physik, 2007, 16(7‐8), p. 543-552.
[14]-Reed SA, Palazzotto AN, Baker WP, “An experimental technique for the evaluation of
strain dependent material properties of hard coatings”, Shock and Vibration, 2008, 15,
p. 697-712.
[15]-Richardson MH, Formenti DL, “Parameter estimation from frequency response
measurements using rational fraction polynomials”, 1st IMAC Conference, 1982, Orlando,
Florida.
[16]-Rongong JA, Goruppa AA, Buravalla VR, Tomlinson GR, Jones FR, “Plasma deposition
of constrained layer damping coating”, Proc. Instn Mech. Engrs Vol. 218 Part C: J.
Mechanical Engineering Science, 2004, 218, p. 669-680.
[17]-Tassini N, Pastias S, Lambrinou K; “Ceramic coatings: A phenomenological modeling
for damping behavior related to microstructural features”, Material Science and Engineering
A, 2006, 442, p. 509-513.
[18]-Torvik PJ, “Determination of mechanical properties of non-linear coatings from
measurements with coated beams”, International Journal of Solids and Structures, 2009, 46,
p. 1066-1077.
[19]-Ustinov AI., Movchan BA., “A study of damping ability of tin-and yttrium-coated flat
specimens of Ti-6%Al-4%V titanium alloy”, Strength of Materials, 2001, 33(4), p. 339-343
Topic-B: Experimental Mechanics
-324-
[20]-Ustinov AI., Skorodzievskii VS, “A study of the dissipative properties of homogeneus
materials deposited as coatings part 2. Copper condensates with different microstructural
characteristics”, Strength of Materials, 2008, 40(2), p. 275-277
[21]-Vasquez CMA, Moreira RAS, Dias RJ, “viscoelastic damping technologies - Part 1:
Modeling and finite element implementations.”, Journal of Advanced Research in Mechanical
Engineering, 2010, 1(2), p. 76-95
[22]-Wang X, Pei Y, Ma Y, “The effect of microstructure at interface between coating and
substrate on damping capacity of coating systems”, Applied Surface Science, 2013, 282,
p. 60-66.
[23]-Yu L, Ma Y, Zhou C, Xu H, “Damping efficiency of the coating structure” International
Journal of Solids and Structures, 2005, 42, p. 3045-3058
[24]-Zener C, Elasticity and anelasticity of metals, University of Chicago Press; 1948.