You are on page 1of 23

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/292703551

A survey of the damping properties of hard coatings for turbine engine blades

Conference Paper · July 2007

CITATIONS READS
10 522

1 author:

Peter Torvik
Air Force Institute of Technology
82 PUBLICATIONS 7,919 CITATIONS

SEE PROFILE

All content following this page was uploaded by Peter Torvik on 11 August 2016.

The user has requested enhancement of the downloaded file.


Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

A SURVEY OF THE DAMPING PROPERTIES


OF HARD COATINGS FOR TURBINE ENGINE BLADES

Peter J. Torvik
Consultant, Universal Technology Corporation
Professor Emeritus, Air Force Institute of Technology
1855 Winchester Road, Xenia, OH 45385
(937) 374 0521 Torvik@att.net

Abstract: The application of a thin coating of a hard material to airfoil surfaces has been
proposed as a means of reducing the amplitude of resonant vibration of gas turbine
components. A variety of materials and processes have been proposed. Their
effectiveness, however, typically has been reported in terms of the response of coated
specimens to an excitation. The use of different substrate properties and dimensions,
different coating thicknesses, and different test procedures make comparative evaluations
challenging. Moreover, as the damping capacity of these materials is often amplitude
dependent, comparisons must be made at common values of strain.

But material properties, rather than system measures, must be known if coatings are to be
included in design analyses. It is the objective of this survey to compare the effectiveness
of coating materials described in the open literature. Reported values of system damping
and test parameters are used to estimate the loss modulus of each material. Included in
comparisons are such plasma sprayed ceramics as magnesium aluminate spinel, alumina,
and yttria stabilized zirconia; magnetoelastic materials, both plasma sprayed and bonded;
homogeneous metals, metallic mixtures, and ceramic-metallic laminates applied by
electron beam physical vapor deposition; and a ceramic infiltrated with a viscoelastic
material.

Key Words: Blade vibration, damping, hard coatings, loss modulus

Introduction: Measured values of damping are commonly reported in terms of several


different parameters. When determinations are made from the bandwidth of the
frequency response function for a coated sample, the system quality factor, Q, is often
reported. From this a system loss factor may be found from ηSYS = 1/Q. Other attributes
of the response function may also be used [1], but a system loss factor may be deduced
from each. When determinations are made from the transient response following an
initial excitation, results are often reported in terms of the logarithmic decrement of the
system, δSYS. Such measurements may be converted to a system loss factor through ηSYS
= δSYS/π. Results obtained by either method are sometimes reported as the damping
capacity, ψ, defined as the ratio of energy dissipated in the system per cycle to the peak
energy stored. Thus, ηSYS = ψSYS/2π. Each of these conversions of damping units to the
common basis of a system loss factor is strictly true only for a linear system with viscous
damping. However, for relatively small values of system damping, such as ηSYS < 0.1, in
This document is cleared for public release by AFRL/WS Public Affairs.
Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 1
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

mildly nonlinear systems, such conversions of units may be made without significant
error.

But each of these measures is a system property. While such values are useful in the rank
ordering of the damping ability of various materials when tested with similar specimens
on the same apparatus, they do not enable the evaluation of damping on an absolute basis.
To accomplish this, damping must be expressed as a material, rather than system,
property. Several choices are available. The most fundamental material property is the
unit damping, D(ε), defined as the total energy dissipated per cycle of fully reversed
oscillation by a unit volume of material at a homogeneous value of maximum strain and
temperature. A second means of representing the damping ability of a material is the
material loss factor, defined as the ratio of the unit damping per radian of oscillation to
the unit stored energy, that is, the strain energy added to and recovered from the unit
volume during oscillation.

However, the most relevant and useful measure of material damping for comparing the
damping ability of different materials, especially for applications as free layer coatings, is
the loss modulus, defined as the imaginary part of a complex modulus,

E * (ε , T ) ≡ E1 (ε , T )[1 + jη (ε , T )] ≡ E1 (ε , T ) + jE 2 (ε , T ) (1)

having dependence both on strain and temperature. Note that the loss modulus, E2, and
the storage (Young’s) modulus, E1, and the material loss factor, η, are related, through

E 2 = ηE1 (2)
if all values are at common values of strain and temperature. Modest nonlinearities in
stiffness may be accounted for by using an amplitude-dependent secant modulus, E1(ε).
Thus, for a unit volume of material with amplitude dependent damping at homogeneous
strain and temperature:

D(ε ) ≡ 2π η (ε ) U (ε ) = 2πη (ε ) E1 (ε ) ε 2 / 2 = π E 2 (ε ) ε 2 (3)

Results from some commercially available devices for measuring damping capacity are
reported in terms of a loss angle, ϕ, which is the measure (in radians) of the extent to
which a sinusoidal stress and sinusoidal strain in a dissipative material are out of phase.
Such values are material properties, if measured with a sample undergoing homogeneous
strain and temperature, and may be converted to a material loss factor by η= tan (ϕ). If
the storage modulus is also known, the loss modulus may then be formed from the
product.

Considerations of weight, thickness, and the influence on blade aerodynamics dictate that
a hard coating applied to an airfoil must be thin. In the case of a uniform thickness, t, of
a linear coating material, fully covering both sides of a structure of thickness, h, in
bending, it can be shown [2] that the simple relationship

This document is cleared for public release by AFRL/WS Public Affairs.


Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 2
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

E2 t
η SYS = 6( )( ) (4)
E h

then captures the essential features of a free layer damping treatment. It is notable that
the loss factor and storage modulus of the coating material as individual properties are
not required. Only the product, Eq. (2), is of relevance. If the thickness ratio is held low,
e.g. t/h = 0.05, the achievement of a system Q of 100 (i. e, ηSYS ~ 0.01) due to coating
damping alone on a titanium blade (E ~ 110 GPa), requires a loss modulus of about 3.7
GPa, or 0.53 Mpsi.

But the simple relationship of Eq. (4) is normally not adequate for the determination of
material properties from test, as thicker coatings are typically used to better separate the
influence of the coating from other sources of dissipation. For a beam of uniform width
and thickness h, coated on both sides at thickness t, the loss modulus of the coating, Ec2,
may be evaluated as a material property from measured values of the system loss factors
for fully coated and uncoated test beams, ηSYS and ηBARE. The result is:

hE BARE
E C 2 = ηE C 1 = M ( N , t / h, n)[η SYS (1 + RSE ) − η BARE ] (5)
6tT (2, t / h)

The strain energy ratio introduced in Eq. (5) is the ratio of ratio of stored energy in the
coating to that in the beam,

UC 6 E C 1t
RSE = = T (2, t / h) (6)
U BARE E BARE h

If the beam is coated on one side only, and if the neutral axis shift is negligible, then Eqs.
(4-6) may be used if the factor 6 is replaced by 3. The parameter N represents the
dependence of the unit damping on strain when expressed by the power law relationship
used by Lazan [4],

D = Jε N (7)

A development of these relationships and the definitions of the functions T (2, t/h) and
M (N, t/h, n) are given as an appendix. A more extensive discussion, including a
procedure for extracting values of the coating storage modulus from measured
frequencies of coated and uncoated beams, is available [3].

If the damping is amplitude dependent, the influence of a material nonlinearity appears


both through amplitude dependence of the observed system loss factor and through the
multiplicative factor M (N, t/h, n) in Eq. (5). For N < 3, the influence of n is negligible
for all modes above the first, i.e., n ≥ 2. It may be seen from the table of representative
values in the appendix that for relatively thick coatings (t/h ≥ 0.15) and modest
nonlinearities, (N ≤ 2.6) the influence of the nonlinearity remains less than 20%. But if
This document is cleared for public release by AFRL/WS Public Affairs.
Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 3
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

the strain distribution is not uniform over the length of the test section, an influence of the
nonlinearity will remain at about 50% for N = 3, even when t/h is only about 0.05. In the
case of a strictly linear material, the loss factor and loss modulus are both independent of
amplitude, N = 2 and M (N, t/h, n) ≡ 1. In the case of very thin coatings, the strain
energy ratio is small and may be neglected in Eq. (5) and thickness integrals (T) in Eqs.
(5-6) become unity.

For linear materials (i.e. loss factors independent of amplitude), or for very thin nonlinear
coatings with negligible strain variations, Eq. (5) takes the simple form:

hE BARE Ac
EC 2 = [η SYS − η BARE ] (8)
3N C t Atotal

In this special case, a loss modulus may be estimated using no more than the measured
loss factors, the modulus of the substrate beam, and dimensions, including the fraction of
surface area (Ac/Atotal) that is coated. NC represents the number of covered sides, 1 or 2.

Damping of Ceramics- Universal Technology Corp/Air Force Research Laboratory:


The facilities of the Turbine Engine Fatigue Facility (TEFF) at Wright Patterson Air
Force Base were used to measure the damping properties of several ceramic coating
materials and combinations of ceramics with other materials. All data was taken at the
resonances of base-driven cantilever beams with response functions observed with
frequencies decreasing from above resonance and at various levels of constant base
acceleration. The facility and methodology are described elsewhere [5].

Alumina: Two beams of Ti-6Al-4V, each 0.625 in wide, 6 inches long, and 0.090 in thick
were coated on both sides over the free length with a NiCrAlY bond coat, nominally 2-3
mils thick, and air plasma sprayed (APS) alumina, nominally 10 mils in thickness.
Because of the roughness at the beam-bond coat and bond coat – alumina interfaces, the
actual total coating thickness was found to total about 14-15 mils. System loss factors for
the second mode, as observed [6] for increasing and decreasing excitation amplitude, are
compared in Fig.1a.

The results are characteristic of ceramic coatings. When tested with increasing excitation
amplitude, the damping increases monotonically. When tested with decreasing
amplitudes, the damping is nearly constant until a threshold value of about 100 ppm is
reached, and then decreases slightly. A Hastelloy X beam, 0.75 wide and 0.063 in. thick
was also prepared with the same coating treatment. System loss factors for this beam in
cantilever modes 2, 3 and 4 were measured [6] with decreasing amplitudes of excitation.
Data is given as Fig.1b. The logarithmic plot emphasizes the bilinear relationship (log-
log) of loss factors observed with decreasing amplitudes that is also characteristic of
plasma sprayed ceramic coatings.

This document is cleared for public release by AFRL/WS Public Affairs.


Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 4
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

Titanium with - Alumina+NiCrAlY Hastelloy with Alumina on NiCrAlY


0.012 101-Inc 101-Dec 0.100
System Loss Factor 4B

System Loss Factor


0.01 102-Inc 102-Dec
3B
0.008 2B
0.006 0.010
0.004

0.002

0 0.001
0 100 200 300 400 500 600 10 100 1000
Maximum Strain-ppm Strain (ppm)-Decreasing

a. On Titanium b. On Hastelloy X
Figure 1. System Damping with Plasma Sprayed Alumina-NiCrAlY

The system loss factors measured before and after coating were used with other system
parameters and Eq. (5) to estimate the loss modulus. Results are shown in Fig.2.

Alumina-NiCrAly Coating Alumina+NiCrAlY


101-Dec 102-Dec Mode 4 Mode 3 Mode 2
1.2 1.4
Loss Modulus - GPa

Loss Modulus-GPa

1.2
1
1
0.8
0.8
0.6 0.6
0.4 0.4

0.2 0.2
0
0
0 200 400 600 800
0 100 200 300 400 500 600
M aximum Strain-ppm Max Strain (ppm)-Decreasing

a. Tested on Titanium b. Tested on Hastelloy


Figure 2. Loss Modulus of Plasma Sprayed Alumina-NiCrAlY Coating System

Weights and measures of the beam before and after coating were used to estimate the
coating density. The strain energy ratio was then determined from this and the change in
resonant frequency due to the addition of the coating. The storage modulus was found
with Eq. (6) to diminish from about 43 GPa to about 38 GPa over the range from 25 to
500 ppm. These values are consistent with values reported in the literature [7].

The apparent mode dependence seen in Fig. 2b is actually a history effect, as tests were
conducted in the order of mode 4, mode 3, and mode 2, all on the same specimen.
Repeated tests on the coated titanium specimens showed similar increases in damping at
the higher levels of strain. A loss factor may be developed from the ratio of loss and
storage modulii. Maximum values are about 0.025 - 0.03. A more extensive discussion

This document is cleared for public release by AFRL/WS Public Affairs.


Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 5
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

of these tests and the results obtained is given elsewhere [6]. Material properties as given
in Fig. 2 are the effective, or average, values for the ceramic-bond coat system.

Magnesium Aluminate Spinel: System loss factors for a coated and uncoated
cantilevered beam of Hastelloy X, 0.75 in wide and 0.095 in thick, covered on both sides
of the 8 in free length with 0.015 in of plasma sprayed Mg Spinel (without bond coat),
were also obtained using the TEFF facility at AFRL [8]. Composite results, formed
from tests in the 2nd, 3rd and 4th bending modes as obtained after significant load history
are given Fig. 3a The two slope relationship is seen again, with notable slope in the high
strain region as these test were conducted with increasing values of excitation amplitude.
The methodology of Eq. (5) was used to obtain an estimate of the loss modulus, Fig. 3b.

Mg Spinel on Hastelloy X Magnesium Spinel-3 Modes


0.01 2

E2 - Loss Modulus - GPa


System Loss Factor

1.6

1.2

0.8
Data
Low er
0.4
Upper

0
0.001
0 100 200 300 400 500 600
10 100 1000
Max Strain (ppm)-Increasing Strain - ppm

a. System Data b. Deduced Loss Modulus


Figure 3. Damping of Plasma Sprayed MgO.Al2 O3

Values of the storage modulus for mag spinel (41 GPa at low strain) were found to be
similar to those of the alumina-NiCrAlY system. A typical loss factor at intermediate
strains may be deduced to be about 0.03 - 0.04.

Damping of Ceramics – University of Sheffield/Rolls Royce: Damping measurements


at the University of Sheffield were conducted by shaker-exciting cantilevered specimens
at resonance and then measuring the decay of vibrations after termination of excitation.
Frequencies and logarithmic decrement were obtained over short time intervals to better
capture the amplitude dependence. Both fully and partially covered beams have been
considered. The equipment and methodology are described elsewhere [5]. A dependence
of damping on initial amplitude has been noted, analogous to that which may be inferred
from Fig. 1a.

Rokide® A (Alumina): Patsias, et. al. [9] measured the damping and natural frequency
of cantilevered beams 1.59 mm (62.5 mil) thick 182 mm (7.2 in) long (with fillet) coated
on both sides to a thickness of 0.28 mm (11 mil) with Rokide® A aluminum oxide,
combustion sprayed at room temperature. From these they deduced the values of storage
modulus and loss factor shown in Fig. 4.

This document is cleared for public release by AFRL/WS Public Affairs.


Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 6
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

a. Storage Modulus b. Loss Factor


®
Figure 4. Material Properties of Rokide (Patsias et. al. [9])

A loss modulus can be formed from these results by simply taking the product, Eq. (2).
The results are shown in Fig. 5.

Rokide TM A
3
Loss Modulus (GPa)

0
0 200 400 600 800
Strain-ppm

Figure 5. Loss Modulus of Rokide® (Alumina)

These values are seen to be approximately double those of Fig. 2b for air plasma sprayed
alumina-NiCrAlY system, but it is likely that the damping in that case comes primarily
from the 10 mil of alumina, and only a negligible amount from the 5 mil combination of
bond-coat and voids. Thus the loss modulus for alumina alone might be nearly 50%
greater. Both sets of results suggest a maximum dissipative capacity at strains of about
200 microinches/inch.

Magnesium Aluminate Spinel: Patsias and Williams [10] employed a partially covered
beam, coated over a 40mm length at the location of the interior maximum in strain for the
second cantilever bending mode. The substrate beam was of a C263 super alloy, 192 mm
(7.56 in) long, including a fillet. Frequencies and system loss factors were determined in
This document is cleared for public release by AFRL/WS Public Affairs.
Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 7
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

free decay from second mode resonant responses of various amplitudes. A finite element
procedure was used to deduce the required amplitude dependencies of the storage
modulus and loss factor that would lead to the observed amplitude dependent rates of
decay. Their results (normalized) are given as Fig. 6. A normalized loss modulus can be
formed from the product, and is shown in Fig. 7.

a. Storage Modulus b. Loss Modulus


Figure 6. Material Properties of Plasma Sprayed MgO.Al2 O3 (Patsias et. al. [10])

Al-Mg-Spinel
0.8
Loss Modulus-

0.6
(Unscaled)

0.4

0.2

0
0 100 200 300 400 500 600 700
Strain-ppm

Figure 7. Loss Modulus of Plasma Sprayed MgO.Al2O3

While unscaled, the results confirm that the damping increases significantly at low strain,
as was seen in Fig. 3b. The slight decrease, rather than a slight increase, at high strain is
the expected outcome for tests conducted with decreasing amplitude (vibration decay),
rather than with increasing levels of excitation.

Zirconia: Tassini, et. al., determined [11] the properties of 8 wt% yttria stabilized
zirconia deposited by air plasma spray (APS) and electron beam physical vapor
deposition (EB-PVD) methods. Specimens were of Inconel® 625 super alloy. In this
case, the coating was applied on both surfaces of the beam, but only over an 13 mm
length near the root of the 96.7 mm long (including fillet),12.4mm wide, and 1.59 mm
This document is cleared for public release by AFRL/WS Public Affairs.
Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 8
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

thick substrate. The coating as applied by air plasma spray was 0.32 mm (12.6 mil) in
thickness without bond coat; that applied by electron beam physical vapor deposition was
0.28 mm (11 mil) in thickness, with a 0.06 mm (2 mil) bond coat of NiCoCrAlY, applied
by air plasma spray. Material properties were extracted from system data, with results as
shown in Fig. 8.

a. Storage Modulus b. Loss Factor


Figure 8. Properties of 8 wt% Yttria Stabilized Zirconia (Tassini, et. al. [11])

Estimates of the material loss modulus may be found from the product of the storage
modulus and loss factor. The result is shown in Fig. 10.

8YS Zirconia
APS EPVD
1.2
Loss Modulus (GPa)

0.8

0.6

0.4

0.2

0
0 200 400 600
Strain - ppm

Figure 9. Loss Modulus of 8wt% Yttria Stabilized Zirconia

The method of deposition appears to have a significant influence on the damping ability,
both qualitatively in that the loss modulus with APS is significantly greater, and
qualitatively in that the loss modulus of the coating as formed by EPVD does not appear
to show a maximum in the strain dependence.

Damping of Magnetomechanical Materials - Fortis Technologies, Inc.: A


magnetostrictive particulate composite (MPC) damping coating was formed by
embedding magnetostrictive particles (Terfenol-D) in an epoxy base. Loss factors with a
small patch of the MPC at the root of cantilever beam and with coverage over 50% of

This document is cleared for public release by AFRL/WS Public Affairs.


Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 9
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

beam length were found [12] to be significantly higher than for the bare aluminum
substrate. In further testing, a Hastelloy beam 0.0635 in thick, 0.75in wide and 8 inches
in unsupported length was fully covered on both sides by Fortis with tiles of the MPC
material. The tiles were composed of 40% (by volume) magnetostrictive particles
embedded in a polymer known to be of much lower damping capacity than the tile
material and attached to the substrate beam with a very stiff adhesive applied at minimal
thickness. The effective thickness of the composite and bond was found to be 0.0125 in
or 0.32 mm. Testing was conducted by UTC personnel using the TEFF facility and
procedures as described elsewhere [3, 5]. Frequency measurements were used to deduce
the effective longitudinal stiffness of the tiles, in situ, to be 11.5 GPa. The loss modulus
of the MPC tile/bond system, treated as an equivalent homogenous layer, was found with
Eq. (5) to be as shown in Fig. 10.

Fortis Tiled Beam


0.5
Loss Modulus-GPa

0.4

0.3

0.2

0.1

0
0 200 400 600 800
Strain-ppm

Figure 10. Loss Modulus of a Magnetostrictive Particulate Composite

Damping of Magnetomechanical Materials – Dr. Herman Shen: The system damping


due to a magnetoelastic material applied as a thin (0.0055 in) surface coating has been
evaluated [13]. Observed frequency response functions (velocity vs. frequency) at
various levels of excitation for the 2nd cantilever mode at two temperatures were reported
to be as shown in Fig. 11. The responses at RT were similar to those at 300 F, those at
900 F similar to those at 600 F.

These frequency response functions at various levels of base-driven excitation were


obtained with the TEFF facilities at WPAFB and used to determine system quality factors
(Q= 1/ηSYS). The test specimen was a coated beam of Hastelloy X, 6.15 in long, 0.75 in
wide and 0.095 in thick; coated by air-plasma spray with a material identified as iron-
chromium [14]. A storage modulus for the coating material, constant at 20 Mpsi to a
strain of 1000 ppm, was reported. Bare beam loss factors were also given.

Digitized velocities taken from the published response functions (e.g. Fig. 11) were used
to estimate peak strains from the mode shape for the 2nd cantilever mode of a beam of the
given dimensions. Using the properties of the substrate beam and the given thickness of
the coating, a strain energy ratio was estimated at 0.273. Equation (5) was then applied
to estimate the loss modulus. The results are shown in Fig. 12.

This document is cleared for public release by AFRL/WS Public Affairs.


Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 10
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

a. Response at 300F c. Response at 600F


Figure 11. Response of Magneto-mechanical Coating (Shen [13])

Magnetomechanical Coating (Shen)


RT 300F 600F 900F
2.5
Loss Modulus - GPa

1.5

0.5

0
0 50 100 150 200
Strain - ppm
.
Figure 12. Deduced Loss Modulus for a Magneto-mechanical Coating

Damping of EBPVD Metallics and Ceramics - E. O. Paton Institute: Professors


Ustinov and Movchan and associates at the E. G. International Center for Electron Beam
Technologies (ICEBT) of the E. O. Paton Institute for Electric Welding (Ukraine) have
carried on a series of measurements of the damping capacity of coatings applied by
Electron Beam Physical Vapor Deposition. The damping has been reported for Ti-6Al-
4V beams with deposited coatings of (1) single metallic elements, (2) dispersed
compounds of hard and soft metallics, and (3) laminates of metallics and ceramics. In
this work, the damping capacity was measured by observing the logarithmic decrement of
partially (36%) coated beams in free vibration from a resonance established by shaker
excitation. Beams tested were of a trapezoidal plan-form to minimize, to some degree,
the variation of strain over the coated region. Reported strains are the average over the
This document is cleared for public release by AFRL/WS Public Affairs.
Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 11
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

coated area, indicated to be about 70% of the root strain. In all cases, specimens of 1.5-
2 mm thickness were coated on one side only, at thicknesses of (about) 5% of substrate
thickness, or much less. In consequence, the strain energy ratio is small, and a loss
modulus can be estimated from system loss factors without knowledge of the storage
modulus of the coating. Moreover, since the strain distribution is approximately uniform
throughout the coating, the influence of the coating nonlinearity is minimal, enabling
estimates of the loss modulus to be made with the use of Eq. (8).

Monolithic Metals: Logarithmic decrements at room temperature as a function of strain


were obtained [15] for EBPVD coatings of tin and of yttrium of various thicknesses on a
2mm substrate beam. Reported values are shown in Fig. 13. In the results for tin, line
codes 1, 2 and 3 denote coating thicknesses (one side only) of 4, 8 and 19 μm. In the data
for yttrium, line codes 1 and 2 denote coating thicknesses of 58 and 120 μm. In both
cases, the logarithmic decrements for the bare substrates are also given (Line code I).

a. Tin b. Yttrium
Figure 13. Logarithmic Decrements with Tin and Yttrium (Ustinov, et. al. [18])

These data, with log decrements converted to system loss factors with ηSYS = δ/ π, were
used with Eq. (8) to estimate the loss modulus as a function of average strain over coating
thickness for each material. Results are shown in Fig. 14.

Tin Yttrium
4 8 19 58 130
5 1.6
Loss Modulus - GPa
Loss Modulus-GPa

4 1.2

3
0.8
2
0.4
1

0
0
0 500 1000 1500 2000
0 500 1000 1500 2000 2500
Strain-ppm Strain-ppm

a. Tin b. Yttrium
Figure 14. Estimates of Loss Modulus for EBPVD Metallics
This document is cleared for public release by AFRL/WS Public Affairs.
Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 12
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

In the case of the tin coatings, the consistency between values obtained for the greater
thicknesses suggests credibility. However, it appears that the precision of measurement
is not sufficient for an accurate determination of loss modulus in the case of the 4 μm
coating (2% of beam thickness).

Bimetallic Coatings: Dispersed bimetallic alloys of two metallic components were also
considered [16]. The measured logarithmic decrements of Fig. 15 are for coatings
formed by EBPVD from cobalt with 20% (weight fraction) iron, and cobalt with 35%
(weight fraction) nickel, each applied to a 1.5 mm thick Ti -6Al-4V substrate.

a. Co-20% (weight) Fe b. Co-35% (weight) Ni


Figure 15. Log Decrements for EBPVD Dispersed Bimetallics (Ustinov, et. al. [19])

By using Eq. (8), with measured values of the logarithmic decrement converted to system
loss factors with ηSYS = δ/ π, the material loss modulus may be estimated for each of
these materials, at each thickness of coating. The results are shown in Fig. 16. Because
of the strong (and complex) nonlinearity, however, the estimates obtained for the Co-Fe
coating may be somewhat less reliable than those for the yttrium.

C0-20%Fe C0-35% Ni
25 55 95
45 60 120
4
LossModulus- GPa
LossModulus- GPa

3 3

2 2

1 1

0 0
0 500 1000 1500
0 500 1000 1500
Strain - ppm
Strain - ppm

This document is cleared for public release by AFRL/WS Public Affairs.


Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 13
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

a. Co-20% (weight) Fe b. Co-35% (weight) Ni1


Figure 16. Deduced Loss Modulus of EBPVD Dispersed Bimetallics

Alloys of cobalt and nickel and of cobalt and iron at these weight fractions are known to
be strongly magnetostrictive [4, 17]. With regard to Co0-35% Ni material, it is of interest
to note that data for homogeneous specimens of Nivco 10 (72% Co, 23% Ni) heat treated
for maximum damping show [17] very high damping, continuing to increase
monotonically to a shear stress of (at least) 20 ksi, and then diminishing.

Metallic-Ceramic Laminates: A graded coating was also formed [18] by EBPVD of 10


wt % Sn, 15 wt% Cr, and balance MgO. The damping was compared with that of a thin
(4μm) layer of a bimetallic Cr-Sn coating, also formed by EBPVD. Logarithmic
decrements shown in Fig. 17a were measured for the 1.8 mm Ti-6Al-4V specimen after
applying the thin bimetallic coating, and after applying the graded coating.

SnCrMgO

CrSn CrSnMgO

Loss Modulus-GPa 6

0
0 500 1000 1500 2000
Strain-ppm

a. Data ((Movchan and Ustinov [22]) b. Deduced Loss Modulus


Figure 17. Damping of Laminates Formed by EBPVD Coating

After converting logarithmic decrements to system loss factors, this data was used in Eq.
(8) with the system parameters and bare beam losses to produce estimates of the loss
modulus. The results are given in Fig. 17b. The estimated loss modulus of the EBPVD
CrSnMgO system (~3 GPa, Fig. 17b) appears to be somewhat higher that the loss modulii
estimated previously for other ceramic coatings at similar values of strain (~500 ppm).

In continuing work at the Paton Institute, the damping of numerous other coatings of all
three classes have been investigated at both room and elevated temperatures. As this
work was supported by the European Office of Aerospace Research and Development,
government contractors may be able to obtain further results of this work.

Damping of Infiltrated Ceramics -APS Materials, Inc./ Universal Technology Corp:

1
In the published version of this paper, the results for Yttrium were erroneously inserted in place of this
figure.
This document is cleared for public release by AFRL/WS Public Affairs.
Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 14
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

The damping of ceramics impregnated with a polymeric material has been evaluated at
room and elevated temperatures [19]. Coatings of mg spinel, alumina, yttria stabilized
zirconia, and chrome carbide of various thicknesses, each with the surface infiltrated with
a viscoelastic material identified as APS 600, was considered. In some cases, a NiCrAlY
bond coat of 2-3 mil thickness was included. The Ti-6Al-4V substrate beams of nominal
dimensions 0.75 in width and 0.093 in thickness were coated by APS Inc. on both sides
of the 8 in unsupported length. Testing was conducted with the facilities of the TEFF at
WPAFB. Previously described test procedures and data reduction methods were used.
Tests at elevated temperatures were conducted in a heat box.

The system loss factors for 10 and 15 mil coatings with the viscoelastic infiltrate were
found to be similar, while those for 5 mil coatings somewhat less. From this it was
concluded that the impregnate influenced the response beyond the first 5 mils of the
coating, but not beyond 10 mils [20].

The reported [19] quality factors obtained the 3rd mode resonant response at several
temperatures for 10 mils of alumina with NiCrAlY bond coat and surface infiltrate were
used to obtain the system loss factors (ηSYS = 1/Q) shown in Fig. 18. The loss factors for
a bare titanium beam at room temperature have been added for comparison.

10Mil Alumina+APS 600 VEM


0.1
System Loss Factor

RT
0.01 150F
200F
225F

0.001 250F
300F
Bare-RT

0.0001
10 100 1000
Strain - ppm

Figure 18. System Loss Factors for a VEM Impregnated Ceramic Coating

The storage modulus of the coating was evaluated from the increase in resonant
frequency resulting from the application of the coating. The impregnation of the ceramic
by a viscoelastic material was also found to (approximately) double the modulus (and the
strain energy ratio). Estimates of the loss modulus as a material property were developed
from this data and system parameters by using Eq. (5). Results are shown in Fig. 19.

This document is cleared for public release by AFRL/WS Public Affairs.


Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 15
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

10 Mil Alumina+ NiCrAlY + APS 600 VEM


RT 150F 200F 225F 250F 300F
6

Loss Modulus -
5
4

GPa
3
2
1
0
0 200 400 600 800
Strain-ppm

Figure 19. Loss Modulus for a VEM Impregnated Ceramic Coating


The impregnation of a hard coating with a material having the damping characteristics of
a viscoelastic solid shows promise of high damping. The process also provides the
opportunity to select the impregnate so as to tailor the coating in order to provide
maximum damping at a particular temperature of interest. In this work, the viscoelastic
infiltrate was chosen to maximize damping at 250F. Work is currently underway to
identify suitable infiltrates and explore the use of this concept at higher values of
temperature.

Summary: Damping measurements have been reported for several materials proposed
as coatings for gas turbine airfoils. A sampling of data from the open literature was
reviewed, and, when feasible, reduced to a common basis for comparison. As the
material property that determines the effectiveness of free layer damping treatment is the
loss modulus, this quantity was estimated for each material and used as the basis for
comparing the damping capability of the various coating materials. See Eq. (4).

In general, the loss modulus of ceramic coatings display a strong amplitude dependence,
with monotonic increases up to 100-200 ppm of strain and a maximum at a strain of
several hundred ppm. In some cases, a diminution is seen at higher levels of strain. It is
of interest to note, however, that a comparison of the loss modulus for the same material
applied by APS and by EBPVD suggests that the damping for the APS coating shows a
maximum, while that for an EBPVD coating does not. Typical maximum values for
ceramic coatings applied by APS appear to be in the range of 1-2 GPa.

Materials dissipating energy through magnetostriction have been considered as damping


coatings. When particles are embedded in an epoxy matrix, the resulting low storage
modulus appears to impede the development of the high levels of strain required for
effectiveness, leading to a low value of the loss modulus. When a magnetostrictive
metallic is applied as thin layer by air plasma spray, however, a loss modulii comparable
to that of plasma sprayed ceramics appears possible.

Some metallic, bimetallic and metallic-ceramic coatings formed by EBPVD appear to


provide significant damping. For such materials, the loss modulus appears to be on the
order of several GPa at a strain of 500 ppm. These values are comparable to and, in some
cases, even higher than that seen for ceramics. Thus it would appear that such coatings
This document is cleared for public release by AFRL/WS Public Affairs.
Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 16
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

have significant potential as damping additions. Moreover, while having comparable


values of loss modulus at intermediate strains, some coatings of this class appear to
display a loss modulus that increases monotonically at even higher levels of strain.

In general, the damping of ceramic coatings is not believed to show frequency


dependence, or to have significant temperature dependence up to 600 C. In the case of
magnetostrictive materials, the damping capacity is known to decrease significantly as
the Curie temperature is approached. In the case of metallic coatings with damping
attributable to plastic deformation some temperature dependence may be expected.

The incorporation of viscoelastic materials under a plasma sprayed constraining layer has
been considered [21], as well as through impregnation of a ceramic coating. In the latter
case, it is conjectured that the viscoelastic constituent is deformed by the relative motion
of the platelets formed in the spray process, thereby dissipating energy. While this
concept enables the generation of very high values of loss modulus and the “tuning” of
the coating by selecting the optimal viscoelastic constituent for a desired range of
operating temperatures, a frequency and temperature dependence not seen in metallic or
homogeneous ceramics can be expected.

Not considered in this comparison of the damping ability of coating materials, but
critically important, are the durability of coatings chosen for dissipative properties and
the extent to which the application of such coatings degrade the structural integrity of the
substrate.

Acknowledgements: The author gratefully acknowledges the effort and care exercised
by UTC staff members Mr. John Justice and Mr. Jason Hansel in conducting some of the
experiments described here. Recognition is also given to the staff of the AFRL for
making available the test facilities of the Turbine Engine Fatigue Facility (TEFF); to Dr.
John Henderson, UTC, for coordination of the efforts; and to the several investigators
who have made their work accessible by publication in the open literature.

References:

[1] Torvik, P. J., “On Evaluating the Damping Of A Non-Linear Resonant System,”
AIAA Paper No 2002-1306, 43rd AIAA/ASME/ASCE/AHS Structures, Structural
Dynamics and Materials Conference, Denver, CO, 2002.

[2] Torvik, P. J., “Analysis of Free-layer Damping Coatings,” Key Engineering


Materials: Layered, Functional Gradient Ceramics, and Thermal Barriers, Ed, Marc
Anglada, E. Jimenez-Piqué, and P. Hvizodoš, Vol. 333, pp, 195-214, 2007.

[3] Torvik, P. J., “Determining Material Properties of Nonlinear Damping Materials from
System Response Data,” Proceedings, 8th National Turbine Engine High Cycle Fatigue
(HCF) Conference, Monterey, CA, 14-16 April, 2003.

This document is cleared for public release by AFRL/WS Public Affairs.


Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 17
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

[4] Lazan, B. J., Damping of Materials and Members in Structural Mechanics, Pergamon
Press, Oxford, 1968.

[5] Torvik, P. J., S. Patsias and G. R. Tomlinson, “Characterizing the Damping Behaviour
of Hard Coatings: A Comparison from Two Methodologies,” Proceedings of the 7th
National Turbine Engine High Cycle Fatigue Conference, West Palm Beach, FL, May
2002.

[6] Torvik, Peter J. Material Properties of Plasma Sprayed Alumina-NiCrAlY Coatings,


prepared for Universal Technology Corporation, 2006.

[7] Ková ík, O., J. Siegl, J.Nohava, and P. Chráska, “Young's Modulus and Fatigue
Behavior of Plasma-Sprayed Alumina Coatings,” Journal of Thermal Spray Technology,
Volume 14, Number 2, pp. 231-238, June 2005.

[8] Torvik, Peter J.: Evaluation of Damping Properties of Coatings: Part I, Room
Temperature, prepared for Universal Technology Corporation, 2002.

[9] Patsias, S., C, Saxton and M. Shipton, “Hard Damping Coatings: an Experimental
Procedure for Extraction of Damping Characteristics and Modulus of Elasticity,
Materials Science and Engineering A, Vol. 370, pp. 412-416, 2004.

[10] Patsias, S. and R. J. Williams, “Hard Damping Coatings: Material Properties and
F.E. Prediction Methods,” Proceedings 8th HCF Conference, Monterey, CA, 13-16 April,
2003.

[11] Tassini, N., K. Lambrinou, I. Mircea, S. Patsias, O. Van der Biest, and R. Stanway
“Comparison of the Damping and Stiffness properties of 8wt% Yttria Stabilized Zirconia
Ceramic Coating Deposited by the APS and EB-PVD Techniques,” Smart Structures and
Materials 2005; Damping and Isolation, Ed. K-L Wang, Proc. SPIE Vol. 5760, SPIE,
Bellingham, WA, pp. 109-117, 2005.

[12] Pulliam, W., D. Lee, G. Carman, “Turbomachinery Fan Blade Damping Using Thin-
layer Magnetostrictive Composite Films,” Proceedings, 8th National Turbine Engine
High Cycle Fatigue (HCF) Conference, Monterey, CA, 14-16 April, 2003.

[13] Shen, Herman, “Free Layer Blade Damper by Magneto-mechanical Coating,


Proceedings, 10th National Turbine Engine High Cycle Fatigue (HCF) Conference, New
Orleans, March 8-11, 2005.

[14] Yen, Hsin-Yi, New Analysis and Design Procedures for Ensuring Gas Turbine
Blades and Adhesive Bonded Joints Structural Integrity and Durability, PhD
Dissertation, The Ohio State University, 2000.

[15] Ustinov, A.I., B. A. Movchan, and V. S. Skorodzievskii, “A Study of Damping


Ability of Tin- and Yttrium-Coated Flat Specimens of Ti–6%Al–4%V Titanium Alloy,”

This document is cleared for public release by AFRL/WS Public Affairs.


Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 18
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

Strength of Materials, Vol. 33, No. 4, 2001. (Translation from: Problemy Prochnosti, No.
4, p 55-61, 2001).

[16] Ustinov, A.I., B. A. Movchan, F. Lemke, and V. S. Skorodzievskii, “Damping


Capacity of Co-Ni and Co-Fe Coatings Produced by EBPVD Method,” Proceedings of
the 1st International conference, Problems of Dynamics and Strength in Gas
Turbocompressors Construction, Vibration in Technique and Technology, (Вібрації в
Τехніці таТехнологіях), No. 4, p 123, 2001. (See also No. 5, p 104-106, 2004).

[17] Cochardt, A. W., “High Damping Ferromagnetic Alloys,” Trans J. of Metals, Vol. 8
No. 10, sec. 2, pp 1295-1298. Also Trans Am Inst Min and Met. Eng, Vol. 226, p 1295,
1956.

[18] Movchan, B. and A. Ustinov, “Highly Damped Hard Coatings for Protection of
Titanium Blades,” Proceedings, RTO AVT Symposium on Evaluation, Control, and
Prevention of High Cycle Fatigue in Gas Turbine Engines for Land, Sea, and Air
Vehicles, Seville, Spain, 3-5 October 2005. Published in RTO-MP-AVT-121.

[19] APS Materials, Inc., Damping Coatings for Gas Turbine Compression System
Airfoils, Final Report for SBIR Phase I Project, Contract No. N68335-04-C-0129, 2004.

[20] Henderson, J. P., R. M. Wilson, D. W. Zabierek and J. A. Justice, “Modified Plasma


Sprayed Coatings for the Damping of Titanium Turbine Engine Airfoils,” Proceedings of
the 75th Shock and Vibrations Symposium, Atlantic Beach, VA, Oct 17-22, 2004.

[21] Rongong, J. A., A. A. Goruppa, V. R. Buravalla, G. R. Tomlinson, and F. R. Jones,


“Plasma Deposition of Constrained Layer Damping Coatings,” Proc. Instn. Mech. Engrs.,
Vol. 218, Part C, Journal of Mechanical Engineering Science, pp. 669-680, July 2004.

Appendix: Extracting the Loss Modulus for Nonlinear Coating Materials

The most fundamental material damping property is the unit damping, D (ε), the energy
dissipated per cycle by a unit volume of material subjected to a fully reversed strain of
amplitude ε. If the distribution of strain is not uniform, the total energy, DC dissipated in
a component having a distribution of strain must be determined from the unit damping by

the coating volume is determined from a sequence of measurements at various levels of


maximum strain, the total energy , Dc, must related to the unit damping by:

DC (ε o ) = ∫
VOL
D(ε )dvol (A1)

where εo is a convenient measure of the greatest value of strain amplitude. Thus, when
the strain varies throughout the volume, solution for the unit damping of the material as a
function of strain necessitates solving an integral equation. This may be accomplished if
the total damping is known over a range of maximum amplitudes. One approach to this
This document is cleared for public release by AFRL/WS Public Affairs.
Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 19
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

is to assume a specific functional dependence of the unit damping on the local amplitude
of cyclic strain. While other choices are possible, a form found to be appropriate in many
cases is that used by Lazan [4]: that the unit damping of most materials is proportional to
the amplitude of cyclic strain raised to a power somewhat greater than two.

D = Jε ε
N
(A2)

The material constants Jε and N are independent of strain, but may vary with temperature
and/or frequency. Note that this form implies a linear relationship when the logarithm of
the unit damping is plotted vs. the logarithm of strain. In some cases, it may be found
that the damping relationship, D (ε), has two (or more) distinct regions of different
slopes, N. In such cases, a piece-wise continuous relationship may be used, with
different parameters in each of the regions.

In the case of a fully covered beam, if the storage (Young’s) modulli of the beam and
coating do not vary greatly with strain, and if the coating is perfectly bonded to the beam,
the strain distribution throughout both substrate beam and the coating is

ε ( x, z ) = C z∂ 2 Z n ( x) / ∂x 2 = C zχ n ( x) (A3)

where Zn (x) is the distribution of transverse displacement in one of the normal modes of
a Bernoullli-Euler beam and χn(x) is the curvature (second derivative) of that mode
shape. The maximum interface strain is a convenient measure of strain amplitude. If the
unit damping relationship of Eq. (A2) applies throughout the entire range of strains of
interest, the total energy dissipated in the coating (of thickness t on both sides) of a fully
covered beam of length L, width W, and thickness h may be expressed in terms of the
curvatures of beam mode shape as:

h / 2+t L ε0
DC = ∫ J ε ε dvol = 2WJ ε ∫ z dz ∫ [ ] N [Cχ n ( x)] N dx (A4)
N N

Vol h/2 0
Chχ n (max) / 2
For all modes of the cantilever beam, the maximum curvature is at x = 0. For simplicity
of notation, it is convenient to represent the influence of the integral of the strain
distribution through the coating thickness and the integral of the curvature by two
dimensionless functions,

1+ 2 t / h
T ( N , t / h) =
1h
t2 ∫ (
2z N 2z
h
) d( ) =
h
h 1
2t N + 1
[
{ (1 + 2t / h) N +1 − 1 } ] (A5)
1
L
1
I ( N , n) =
L0∫ [ χ n ( x) / χ n (max)] N dx (A6)

While it is not feasible to evaluate the integral I (N, n) in closed form for non-integer
values of N, it may be evaluated numerically for each mode (n) from the mode shapes.
The function T (N, t/h) is always greater than unity and takes into account that the strain
This document is cleared for public release by AFRL/WS Public Affairs.
Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 20
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

in the coating is always greater than the local interface strain. The energy dissipated in
the coating then becomes

DC = 2WLtJ ε ε 0 T ( N , t / h) I ( N , n)
N
(A7)

But the total energy dissipated may also be evaluated from measured values of the loss
factor of the coated and uncoated beam at strain εo. If the energies stored in the coating
are UC and UBARE respectively, and all losses other than coating dissipation are DBARE (ε0),
then the system loss factor is

DC + DBARE DC DBARE (U BARE / U C )


ηS = = + (A8)
2π (U C + U BARE ) 2πU C (1 + U BARE / U C ) 2πU BARE (1 + U BARE / U C )

After substitution of the definition of the loss factor, η BARE = DBARE /( 2πU BARE ) , we find:

DC (ε 0 ) = 2π [(1 + RSE )η S (ε 0 ) − η BARE (ε 0 )]U BARE


(A9)
where RSE ≡ U C / U BARE is the strain energy ratio, which may also be a (typically weak)
function of maximum strain. The strain energy in the bare beam is
WLhE BARE ε 02
U BARE = I (2, n) (A10)
6
The integral I (2, n) has value ¼ for all modes of a cantilever beam. We then find that:

πE BARE h I (2, n) 2− N
Jε = [η S (1 + RSE ) − η BARE ]ε 0 (A11)
6tT ( N , t / h) I ( N , n)

By using this relationship and the measured values of the system loss factors, we may
solve for the parameters of the assumed form of the unit damping relationship, Eq. (A2).
Since the parameter Jε is, by definition, a constant, the right hand side of Eq. (A11) must
be independent of the representative strain, ε0. If we express the observed amplitude
dependent system loss factors with a “best fit” of the form:

[η S (1 + RSE ) − η BARE ] ≅ η 0 ε m (A12)

it follows from Eq. (A8) that the necessary material parameters are N = m + 2 and

πE BARE h I (2, n)
Jε = η (A13)
6tT ( N , t / h) I ( N , n) 0

The loss modulus may then be formed from Jε and the definition of the loss factor.

This document is cleared for public release by AFRL/WS Public Affairs.


Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 21
Citation: Integration of Machinery Failure Prevention Technologies into System Health Management,
Society for Machine Failure Prevention Technology (MFPT), Dayton, OH, pp 485-506, 2007.

1 Jε ε
2+ m
1 D
EC 2 = EC1η = EC1 = C1
E (A14)
2π U 2π EC1ε 2 / 2

Substitution of Eq. (A13) then gives the loss modulus in terms of measured loss factors
and system parameters. It is convenient to write

hE BARE
EC 2 (ε 0 ) = M ( N , t / h, n)[(1 + RSE )η S (ε 0 ) − η BARE (ε 0 )] (A15)
6tT (2, t / h)

where the consequence of the material damping nonlinearity (N ≠ 2) is further captured


through the use of a multiplicative factor that is independent of strain,

T (2, t / h) I (2, n)
M ( N , t / h, n) ≡ (A16)
T ( N , t / h) I ( N , n )

Representative values of the nonlinearity factor for cantilever beams for modes other than
the first are given in Table I. Values for the first mode differ slightly.

Table I. Representative Values of Damping Nonlinearity for Mode 2


and Higher of a Cantilever Beam.
N 2 2.1 2.2 2.3 2.4 2.5 2.6 2.8 3 4
t/h= 0.05 1 1.05 1.09 1.14 1.19 1.24 1.29 1.39 1.50 2.14
t/h= 0.1 1 1.04 1.08 1.12 1.16 1.21 1.25 1.34 1.43 1.93
t/h= 0.15 1 1.04 1.07 1.11 1.14 1.18 1.21 1.29 1.36 1.74
t/h= 0.2 1 1.03 1.06 1.09 1.12 1.15 1.18 1.24 1.29 1.57
t/h= 0.25 1 1.03 1.05 1.07 1.10 1.12 1.15 1.19 1.23 1.42
t/h= 0.3 1 1.02 1.04 1.06 1.08 1.10 1.11 1.15 1.18 1.29

Note that the evaluation of Eq (A15) presumes knowledge of the strain energy ratio, the
computation of which requires knowledge of the storage modulus of the coating. And if
the storage modulus of the coating varies with strain, so will the strain energy ratio.
However, if the coating is sufficiently thin, the strain energy ratio may be neglected.
A more complete discussion, including a procedure for extracting values of the storage
modulus from the frequency of coated and uncoated beams is given elsewhere [3].

This document is cleared for public release by AFRL/WS Public Affairs.


Disposition Date 2/23/2007/ Document AFRL-WS-07-0361
Page 22

View publication stats

You might also like