Journal of Sound and Vibration 377 (2016) 169–184

Contents lists available at ScienceDirect

Journal of Sound and Vibration

journal homepage: www.elsevier.com/locate/jsvi

Influence of inhomogeneous damping distribution on sound

radiation properties of complex vibration modes
in rectangular plates
Oliver Unruh a,b,n
a
German Aerospace Center (DLR), Institute of Composite Structures and Adaptive Systems, Lilienthalplatz 7, 38108 Braunschweig,
Germany
b
Technische Universitaet Braunschweig, Institute of Adaptronic and Functional Integration, Langer Kamp 6, 38106 Braunschweig,
Germany

a r t i c l e i n f o abstract

Article history: In order to reduce noise emitted by vibrating structures additional damping treatments
Received 20 August 2015 such as constraint layer damping or embedded elastomer layers can be used. To save
Received in revised form weight and cost, the additional damping is often placed at some critical locations of the
2 May 2016
structure, what leads to spatially inhomogeneous distribution of damping. This inhomo-
Accepted 4 May 2016
geneous distribution of structural damping leads to an occurrence of complex vibration
Handling Editor: D. Juve
Available online 19 May 2016 modes, which are no longer dominated by pure standing waves, but by a superposition of
travelling and standing waves. The existence of complex vibration modes raises the
Keywords: question about their influence on sound radiation.
Structural acoustics
Previous studies on the sound radiation of complex modes of rectangular plates reveal,
Complex vibration modes
that, depending on the direction of travelling waves, the radiation efficiency of structural
Sound radiation
Inhomogeneous damping modes can slightly decrease or significantly increase. These observations have been made
using a rectangular plate with a simple inhomogeneous damping configuration which
includes a single plate boundary with a higher structural damping ratio. In order to
answer the question about the influence of other possible damping configurations on the
sound radiation properties, this paper addresses the self- and mutual-radiation effi-
ciencies of the resulting complex vibration modes. Numerical simulations are used for the
calculation of complex structural modes of different inhomogeneous damping config-
urations with varying geometrical form and symmetry. The evaluation of self- and
mutual-radiation efficiencies reveals that primarily the symmetry properties of the
inhomogeneous damping distribution affect the sound radiation characteristics. Especially
the asymmetric distributions of inhomogeneous damping show a high influence on the
investigated acoustic metrics. The presented study also reveals that the acoustic cross-
coupling between structural modes, which is described by the mutual-radiation effi-
ciencies, generally increases with the presence of travelling waves.
& 2016 Elsevier Ltd. All rights reserved.

n
Correspondence address: German Aerospace Center (DLR), Insitute of Composite Structures and Adaptive Systems, Lilienthalplatz 7, 38108
Braunschweig, Germany.
E-mail address: oliver.unruh@dlr.de

http://dx.doi.org/10.1016/j.jsv.2016.05.009
0022-460X/& 2016 Elsevier Ltd. All rights reserved.
170 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184

1. Introduction

The reduction of structural weight is one of the key goals on the way to environmentally friendly and cost efficient
transportation. It is well known that there is a contradiction between a consequent structural lightweight design and noise
and vibration requirements. Due to the lower mass and higher stiffness, high performance lightweight structures such as
carbon fibre reinforced plastics (CFRP) exhibit increased vibration amplitudes and noise levels. A possible solution is the
application of additional damping treatments such as constraint layer damping (CLD) or embedded elastomer layers, which
increases dissipation rates of vibration energy and improve vibro acoustic behaviour of the structure [1,2]. In order to save
weight and cost, the placement of these damping treatments can be derived from an optimisation process, which is based
on the modal strain energy [3] or structural intensity [4] consideration. The optimisation leads to a very local application of
damping treatments, focussed on some critical locations of the vibrating structure [5–7]. In this case the spatial distribution
of structural damping becomes inhomogeneous and it leads to an occurrence of complex vibration modes. Compared to real
modes, complex modes are no longer dominated by pure standing waves but by a superposition of travelling and standing
waves [8].
The influence of homogeneous damping on the sound radiation of vibrating plates is discussed by Xie et al. [9] or Fahy
and Gardonio [10]. With increasing homogeneous damping the radiation efficiency of the plate also increases in the fre-
quency range of edge and corner radiators. This occurs due to a strong coupling of neighbouring resonances with resulting
piston-like vibration shapes and reduced acoustic short circuit phenomena.
Wodtke and Lamancusa [11] investigated the influence of inhomogeneous damping distribution and stated that it has a
structural (loss factors) and an acoustic effect (radiation efficiency) on sound radiation. In this study modified vibration
shapes, that occur due to inhomogeneous damping distribution, are mentioned as a reason for changes in radiation effi-
ciency. Nevertheless, complex vibration modes with travelling wave components are not investigated here. Marburg [12]
considered complex normal modes of a fluid in external acoustics using an example of an open cavity. Torres et al. [13]
observed complex vibration modes in a top plate of a classic guitar. This is one of the very few publications that link the
structural dynamic behaviour of complex vibration modes to the sound radiation. However, this study is focussed only on
the methodology of recognising complex modes using frequency response functions calculated by the FE simulation. The
acoustic relevance of complex vibration modes, and their influence on the radiation characteristics of the guitar is not
investigated in this work.
Consideration of complex vibration modes regarding their acoustic properties is addressed in some recent publications
by Unruh et al. [14,15]. In these studies, the acoustic metrics, such as self-radiation efficiency, spatial distribution of acoustic
intensity and directivity of the radiated sound field are investigated and it is shown that travelling wave components of
complex vibration modes can significantly influence the sound radiation of rectangular plates. Both publications investigate
the case, where bending waves travel along one dominant direction of the plate. This configuration of travelling waves
occurs, when one of the plates boundaries features an increased structural damping. In [14] it is shown that structural
modes with even order in the main direction of travelling waves significantly increase their radiation efficiency below the
coincidence frequency, due to a better coupling in the first radiation mode. In the case of structural modes with odd order,
this coupling slightly decreases, which results in a lower radiation efficiency. The coupling in the first radiation mode
correlates to the volume velocity of the mode shape, which is zero for real modes of even–even or even–odd order. For
example, when an even–odd order mode is affected by travelling waves in the direction of the even order, the volume
velocity becomes non-zero and the coupling in the first radiation mode increases.
The presence of inhomogeneous damping in plates not only leads to an occurrence of travelling waves but also to the
redistribution of the vibration maxima and minima, which can affect the sound radiation. In order to separate these two
mechanisms, a simplified analytical model is introduced in [15]. This model allows the calculation of complex vibration
patterns with a variable amount of travelling waves without amplitude redistribution. The results show that this redis-
tribution has a minor importance for the acoustic metrics and the presence of travelling waves dominates the sound
radiation phenomena.
As it has already been mentioned, these two initial studies only address a simple case, where bending waves travel along
one dominant direction of the vibrating plate. This simple approach is suitable for the basic understanding of the phe-
nomena, which affects the sound radiation of complex vibration modes. Nevertheless, in real applications the spatial dis-
tribution of inhomogeneous damping and accordingly the configuration of travelling bending waves can be much more
complicated.
For this reason, the present paper addresses the important question about the influence of other possible distributions of
inhomogeneous damping on the acoustic properties of complex vibration modes. The presented research is focussed on the
identification of relevant geometrical properties of additional damping, which influence the sound radiation.
The first part of the paper describes some basic damping configurations, which can be probably found in real applica-
tions. Therefore, the geometrical forms as well as symmetry properties of additional damping are varied. In the next part,
the resulting complex vibration modes are characterised regarding their self-radiation efficiency using the elemental
radiator approach [10]. This allows the identification of damping configuration with the biggest impact on sound radiation,
which are investigated in more detail. The self-radiation efficiencies of complex vibration modes are considered and it is
shown how the geometrical properties of additional damping determine the influence on sound radiation below the
 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184 171

Fig. 1. Basic damping distributions with different shape and symmetrical properties (S – symmetric, B – bilateral symmetric, C – central symmetric, A –
asymmetric).

coincidence frequency. As a complement to the self-radiation efficiencies, the mutual-radiation efficiencies [10,16] are
investigated. Finally, the presented results are discussed and an outlook to future research is given.

2. Basic damping distributions in rectangular plates

In rectangular plates or in more complex structures, which can be subdivided in plate elements, different configurations
of inhomogeneous damping can occur. Despite the seemingly infinite number of variations it is possible to define a limited
number of fundamentally different configurations. These mainly differ in their geometrical shape and the symmetry of the
areas with higher damping. For example, one or more edges of the plate can have an increased damping. In real applications,
this can be observed in skin panels of stiffened structures, where some stiffeners can be riveted or manufactured integrally
and therefore exhibit a different amount of damping. Edges with higher damping can be distributed symmetrically or
asymmetrically with respect to the plates centre. For example, two adjacent or opposite, or even all four edges can have an
increased damping. It is also conceivable that some areas of the plate, for example the centre or the corners feature higher
damping. This can be the case when some additional damping treatments are applied at certain areas of the structure, or the
variation of damping results from different materials (e.g. different core material in a sandwich).
Based on these considerations, a total of twelve configurations of inhomogeneous damping, shown in Fig. 1 can be
derived. The additional damping is marked by a dark colour. The notation of the presented configurations refers to different
symmetry properties of additional damping and gives S1–S5 symmetrical, B1–B3 bilateral symmetrical, C1–C2 central
symmetrical and A1–A2 asymmetrical distributions. In order to ensure the comparability of different configurations, the
same total area of additional damping is defined.
Of course many other configurations of inhomogeneous damping are possible. However, in the author's opinion, the
presented distributions cover all important properties in terms of symmetry and the related configurations of travelling
waves in the resulting complex mode shapes. For instance B1 represents the case, where bending waves travel towards one
of the plates boundaries. In S1 and S2, travelling waves start at the centre of the plate and propagate towards respectively
two or four boundaries. The configuration S3 addresses the reversed case, where the waves are travelling from the
boundaries to the centre of the plate. The asymmetric configurations A1 and A2 cause bending waves travelling in diagonal
direction. Other cases have either direct similarity to the configurations shown here, or can be considered as a combination
thereof. For example, a plate with three damped boundaries has partially the properties of the plate S1 and B1.
172 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184

Fig. 2. Calculated 4
4 structural modes of different damping configurations.

3. Numerical simulation of inhomogeneously damped plates

3.1. Calculated complex vibration modes

For every configuration shown in Fig. 1, as well as for the reference plate without additional damping, a finite-element
model is created in the FE-Software ANSYS. All plates have a dimension of 0:9
0:6 m. The thickness and material of the
panel have no importance for the consideration of complex vibration modes because they only affect the eigenfrequencies,
but in order to obtain an eigenvalue solution, aluminium material properties (E ¼72 GPa, ν ¼ 0:3, ρ ¼ 2700 kg=m3 ) and
thickness of 0.005 m are defined. Nevertheless, all results presented in this paper assume that complex vibration modes can
occur at any possible frequency below the coincidence.
In accordance with the FE-model presented in [14], the reference Rayleigh damping of the unaffected plate material is
defined by damping ratios of ξ1  0:01. For the areas affected by additional damping treatments, ratio of ξ2  0:2 is defined.
This high damping ratio is needed in order to achieve a sufficient complexity of the resulting complex mode shapes.
Nevertheless, this value is not unrealistic and can be expected by an intensive use (e.g. multiple elastomer layers in the
composite layup of damping materials [17]).
Clamped boundary conditions are defined for plates with all investigated damping configurations. Preliminary studies
also addressed plates with simply supported boundaries and it is noticed that the type of boundary condition has no
fundamental effect on investigated phenomena. The type of boundary condition only affects the absolute values of e.g.
radiation efficiency of a mode.
The DAMP-Solver is used in the ANSYS-Software for the solution of the eigenvalue problem for the inhomogeneously
damped system. Depending on the damping configuration, 22 to 24 Eigenvalues and Eigenvectors are calculated in the
frequency range up to 1000 Hz. Fig. 2 shows the real and imaginary parts as well as amplitude and phase distribution of the
even–even 4
4 mode shape for the reference plate and for one example from every symmetry group. It is important to
 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184 173

notice that complex modes are only determined up to a arbitrary normalisation factor. This means that the picture shown in
Fig. 2 is not uniquely determined and real and imaginary parts can change depending on the used normalisation factor.
The 4
4 mode of the reference plate is dominated by pure standing waves, what can be seen on 01 and 1801 angles in
the phase distribution. In case of the damping distributions with inhomogeneous damping the mode shape includes
bending waves, which travel towards the higher damped area. This can be seen by phase ramps with angles aside of 01 and
1801. Travelling waves can also be noticed by blurred nodal lines in the amplitude distribution.

3.2. Determination of the acoustic metrics

In this section, a short description of the acoustic modelling is provided. The calculation of acoustic metrics, such as self-
and mutual-radiation efficiencies is conducted by the elemental radiator approach [10]. The total sound power W radiated
by a plate into the far-field is calculated using an array of normal velocities vn , sound pressures p and the elemental radiator
area Se as follows:
Se
H 
W¼ Re vn p ; (1)
2
where the superposed H denotes the Hermitian transpose (complex conjugate transpose). For the sound power evaluation
harmonic vibrations are assumed. This means that normal velocities vn and sound pressures p are space dependant and the
time dependence is provided by eiωt .
The sound pressure can be derived from the structural velocities vn and the frequency-dependent radiation impedance
matrix Z. The off-diagonal elements of the radiation impedance matrix are expressed by:

jωρ0 Se e  jkrij
Z ij ¼ ; (2)
2π r ij

where k ¼ ω=c0 is the acoustic wavenumber and rij is the distance between element i and element j. In order to estimate the
sound power radiated by a complex mode shape, the field point sound pressure is calculated on the vibrating surface. The
values for air density ρ0 ¼ 1:225 kg=m3 and speed of sound c0 ¼ 340 m=s are used in the simulation.
The diagonal elements of the impedance matrix Zii are singular because of r ii ¼ 0 and are replaced by the formulation
which corresponds to the impedance of a baffled piston [18]:
 pffiffiffiffiffiffiffi 
Z ii ¼ ρ0 c0 1  e  jk Se =π : (3)

Using the acoustic impedance and the structural velocities, the radiated sound power can be formulated as [19]:
Se
H 
W¼ Re vn Zvn ¼ vH
n Rv n : (4)
2
The matrix R ¼ S2e ReðZÞ is called radiation resistance matrix and represents the free field related part of the radiation

Fig. 3. First four radiation modes of the investigated rectangular plate.

174 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184

impedance matrix Z. The analytical formulation of matrix R for N elemental radiators is given by:
2 sin ðkr 12 Þ
3
1 kr 12
⋯ sinkrðkr 1N Þ

6 1N
7
ω2 ρ0 S2e 6 sin ðkr Þ
6 kr2121 1 ⋮ 7 7
R¼ 6 7: (5)
4π c0 6 ⋮ ⋱ ⋮ 7
4 5
sin ðkr N1 Þ
kr N1
⋯ ⋯ 1

The self-radiation efficiencies of the vibration modes, which are addressed in this paper can be calculated by the fol-
lowing equation:
W W
σ¼ ¼ ; (6)
W 0 ρ0 c0 S〈 vn 2〉

where W is the sound power of the complex vibration mode according to Eq. (4), when structural velocities vn are sub-
stituted by vibration modes ψ n . W0 is the sound power radiated by a rigid piston with the same averaged velocity 〈vn 2 〉 and
area S.
An eigenvalue decomposition can be applied to the radiation resistance matrix R in order to calculate the so-called
radiation modes Q and their efficiencies Λ:

R ¼ Q T ΛQ : (7)

In contrast to structural modes, radiation modes are frequency dependant. Fig. 3 exemplarily shows the first four
radiation modes of the investigated rectangular plate for the frequency of 70 Hz.
According to Eq. (4), the sound power can also be expressed with the use of radiation modes and their efficiencies as
follows [10]:

n Rv n ¼ vn Q ΛQv n :
H T
W ¼ vH (8)

The product between the radiation modes Q and the structural velocities vn or vibration modes ψ n delivers coupling factors
yn and yr :
yn ¼ Qvn and yr ¼ Q ψ n : (9)

These factors describe the coupling of the structural velocity patterns or vibration modes with the radiation modes.
Especially the high coupling in the first radiation modes indicates a good radiation efficiency of the structural velocity
patterns or vibration modes. Further in this paper, these coupling factors will be used in order to investigate the increased
sound radiation of complex modes.
It is important in the consideration of the radiation behaviour of structural modes that they do not radiate sound
independently from each other [10]. To show this, the structural velocities vn are transformed in modal domain using the
matrix of eigenvectors Ψn and modal coordinates q:

vn ¼ Ψn q: (10)

Substituting this equation in the formulation of the sound power in Eq. (4) delivers:

W ¼ qH Ψn RΨn q ¼ qH Πq; with Π ¼ Ψn RΨn :

H H
(11)

Considering the matrix Π it can be noticed that the diagonal elements Πii correspond to self-radiation efficiencies. The
acoustic interaction between different structural modes is specified by off-diagonal elements of the matrix Π. The mutual-
radiation efficiencies are finally given by a normalisation of the off-diagonal elements Πij as follows [10]:
8Π ij
σ ij ¼ : (12)
ρ0 c 0 S

3.3. Quantification of the modal complexity

In order to ensure the comparability of complex vibration modes regarding their impact on the sound radiation, the
modal complexity of the calculated modes must be quantified. Modal complexity can be understood as a metric which
quantifies the non-proportionality of structural damping using the properties of real and imaginary parts of complex
vibration modes. It is important to notice that there are several different metrics available for the quantification of modal
complexity [20,21]. These metrics use different properties of the complex mode, for example mean phase (MP) or mean
phase deviation (MPD) in order to quantify modal complexity. Besides mathematical properties of a complex mode, also
physical measures such as strain energy can be used for the quantification of modal complexity [22]. The modal collinearity
index (MCI) is one of the metrics that can be used in order to quantify the level of modal complexity [21]. It is based on the
 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184 175

Fig. 4. MCI values of the calculated complex modes.

Fig. 5. Modal radiation efficiencies of complex vibration modes.

correlation between real and imaginary parts of the eigenvector. It is defined as:

jReðψ~ n ÞT Imðψ~ n Þj
MCI ¼ 1  qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi; (13)
½Reðψ~ n ÞT Reðψ~ n Þ½Imðψ~ n ÞT Imðψ~ n Þ

with
ψn
ψ~ n ¼ ejπ =4 ; (14)
maxðjψ n jÞ

where ψ~ n is the vector that results from the 45° rotation and normalisation of the eigenvector ψ n [21].
The MCI indicates real mode shapes with values of zero and mode shapes with maximum complexity with values of one.
176 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184

Fig. 4 shows the MCI values of the complex vibration modes for the investigated damping configurations. First of all, the
MCI values indicate the increasing complexity of the higher order modes, due to increasing damping ratios defined by the
Rayleigh damping model [14]. Furthermore it can be noticed that configurations with the biggest continuous damping
surface such as S3, B1, B2, A1 and A2 show the highest MCI values of up to 0.93. Other configurations such as S2, S5 or B3
have much lower complexities of vibration modes with MCI o 0:2. This large variation of MCI values of complex modes
makes the direct comparison of different configurations difficult. In order to simplify this, the Complex Mode Radiation
Index (CMRI) is defined in Section 4.3 of this paper. It allows an identification of damping distributions with the highest
influence on sound radiation, normalised by the value of complexity. But first, the self-radiation efficiencies of complex
vibration modes are considered in the next section.

4. Acoustic characterisation

4.1. Self radiation efficiencies

In this part of the paper, self-radiation efficiencies of selected configurations from every symmetry group are considered.
From the preliminary studies presented in [14,15] it is known that complexity of vibration modes influences their sound
radiation well below the coincidence frequency. It is demonstrated that the order of the mode (even or odd) and the
corresponding direction of travelling waves have the biggest importance for self-radiation efficiencies. For that reason four
examples (3
2, 2
3, 4
4, 7
1) are selected in order to represent odd–even, even–odd, even–even and odd–odd order
groups of modes.
Considering, that the compared configurations should have similar MCI values of the investigated modes (see Fig. 4), four
examples S3, B1, A1, C1 are selected from every symmetry group. Fig. 5 shows the self-radiation efficiencies of the selected
vibration modes for a reference case without additional damping and for these four configurations.
In [14] it is shown that bending waves travelling in the direction of the even order of the complex mode significantly
increase its radiation efficiency at frequencies well below the coincidence by 20–50 dB. This increase can be observed even
at relatively low complexities of MCIo 0:2. Odd order modes slightly reduce their radiation efficiency in the entire frequency
range below the coincidence by a couple of dB. In this case much higher complexities of MCI 4 0:4 are needed for an
observable effect.
Considering the odd–even order mode 3
2 in Fig. 5 it can be noticed that, besides the configuration A1 with asymmetric
damping distribution, only a negligible influence of travelling waves on radiation efficiency occurs. Regarding the config-
uration B1 this observation correlates to the outcome of the previous study [14]. According to these results, bending waves
which are travelling in the direction of the odd order of three do not increase the radiation efficiency of the complex mode.
In this case, the volume velocity of the mode and therefore the coupling into the first radiation mode remains zero. In
contrast to that, the configuration A1 increases the radiation efficiency, due to the fact that bending waves travel diagonally
towards the areas with increased damping and affect the even order of two. This configuration of travelling waves leads to
higher efficiencies due to an increase of volume velocity and coupling in the first radiation mode. These relations are
discussed in detail in Section 4.2, but first other complex modes in Fig. 5 are considered.
The complex even–odd order 2
3 mode shows a similar impact on the sound radiation as the previously considered
3
2 mode. Also for the 2
3 mode, the symmetric and central symmetric configuration shows a negligible influence on
self-radiation efficiencies. Nevertheless, the biggest difference is the fact that besides the asymmetric, also the bilateral
symmetric configuration increases its radiation efficiency below the coincidence frequency. This can be explained by the fact
that in comparison to the 3
2 mode, the waves are now travelling along the even order of two and therefore the coupling
in the first radiation mode is increased.
The highest impact on the sound radiation can be observed at the even–even order 4
4 mode. In this case, the B1, A1
and C1 configuration shows a significant increase of the radiation efficiencies at low frequencies. This can also be explained
by an increased coupling in lower radiation modes. Remarkable is the fact that only the 4
4 mode of the symmetric
configuration S3 remains nearly unaffected in terms of self-radiation efficiency.
Finally, the odd–odd order 7
1 mode is considered. This mode is the most efficient radiator under the presented series
of modes. It can be noticed that none of the examined configurations indicate a significant impact on the self-radiation
efficiencies. The asymmetric configuration A1 reveals, due to reduced coupling in the first radiation mode, the highest
decrease of sound radiation capability by 2 dB.
In order to summarise the presented results it can be stated that the discussed phenomena are of general nature and
valid for all configurations specified by the symmetry properties and order of the vibration mode. All symmetric config-
urations S1–S5 reveal a negligible influence on the sound radiation of all orders of complex vibration modes. Following from
the main direction of travelling waves, the damping distributions with bilateral symmetry B1–B3 increase the radiation
efficiency of even order modes and slightly decrease in the case of odd order modes. Central symmetric configurations C1–
C2 only increase the radiation capability of even–even order modes. Finally, the asymmetric configurations A1–A2 affect the
highest number of vibration modes.
 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184 177

Fig. 6. Coupling factors yr in the first radiation mode.

4.2. Coupling of structural modes in radiation modes

In order to improve the understanding of the results in the previous section, the coupling factors defined in Eq. (9) are
considered in detail for the same group of complex vibration modes. It is well known from the literature [10,23] that at
lower frequencies well below the coincidence the radiated sound power is dominated by a handful of radiation modes. The
first radiation mode, already considered in Fig. 3, is the most efficient of them and corresponds to an in-phase oscillation of
elemental radiators which results in a monopole character of the radiated sound field. Odd–odd order modes radiate well
due to a high coupling in this first radiation mode. The second and third radiation modes are less efficient and represent the
dipole character with 01 and 1801 oscillation of two opposite regions of the vibrating structure. For instance, even–odd and
odd–even mode shapes of a plate have a high coupling coefficient in this group of radiation modes. Finally the fourth
radiation mode corresponds to a much less efficient quadrupole radiation and is coupled with even–even order
structural modes.
To explain the variation of the radiation efficiencies, Fig. 6 shows the coupling factors yr as a function of dimensionless
frequency kLx with acoustic wavenumber k and plate dimension in x-direction Lx. In this case the same group of modes as
presented in Fig. 5 is considered.
Considering the coupling factors of the odd–even 3
2 mode it can be noticed that in correlation to the results of the
self-efficiencies only the A1 asymmetric configuration reveals an increased coupling in the first radiation mode. The cou-
pling of other damping configurations remains zero. A similar coupling increase can be observed for the even–odd 2
3
mode. Besides the A1 also a B1 configuration increases the volume velocity and therefore the coupling in the first radiation
mode. It can be explained by bending waves that are travelling along the even order of the complex mode.
The almost negligible reduction of the radiation efficiencies of the odd–odd 7
1 mode can also be observed using the
coupling coefficients. Considering the A1 configuration it can be seen that the reduction of the coupling factors results in the
mentioned slight reduction of the self-radiation efficiencies by 2 dB.
Interesting observations can be made on the even–even 4
4 mode. According to the results of the self-radiation effi-
ciencies, the A1 and C1 configurations increase their radiation due to increased coupling in the first radiation mode. The
self-radiation efficiency of the configuration B1 also increases but in contrast to configurations A1 and C1 the coupling in the
178 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184

Fig. 7. Coupling factors yr of the 4
4 in the second and fourth radiation mode.

Fig. 8. CMRI values of different damping configurations.

first radiation mode remains zero. This can be explained considering the coupling in the second and fourth radiation mode
in Fig. 7.
The reason for the increased radiation of the B1 configuration is the improved coupling in the second radiation mode.
This radiation mode is less efficient than the first, which explains the lower increase of the self-radiation efficiencies of the
B1 configuration compared to A1 and C1, seen in Fig. 5. In case of the fourth radiation mode it can be noticed that, except the
symmetric S3 configuration, all other configurations reduce their coupling. Regarding the coupling of complex vibration
modes in the lower radiation modes and the resulting increased radiation efficiency it can be summarised:

 Complex modes which result from symmetric damping distributions have a negligible effect on the coupling into the
radiation modes.
 Bilateral symmetric configurations increase the coupling of even–odd order structural modes into the first radiation
mode. In case of even–even order modes the coupling in the second radiation mode is improved. The coupling of odd–
odd order modes in the first radiation mode is slightly reduced.
 Central symmetric damping distributions only affect even–even order structural modes and increase their coupling into
the first radiation mode.
 Asymmetric configurations always increase the coupling into the first radiation mode when an even order in any
direction of the plate is present.

In other words, travelling bending waves usually increase the coupling of structural modes with even order into the
lower radiation modes and therefore lead to increased radiation efficiency. In contrast to that, the observed reduction of
radiation efficiency of odd order modes is almost negligible.

4.3. Complex mode radiation index

In order to substantiate the statements of the previous sections, this part of the paper verifies the influence of complexity
on sound radiation with a new evaluation index. In the previous sections it is shown that self-radiation efficiencies indicate
the amount of influence with a single scalar metric. This motivates the use of self-radiation efficiencies for the definition of a
 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184 179

evaluation index which quantifies the impact on sound radiation normalised by the level of complexity in complex vibration
modes. This Complex Mode Radiation Index CMRI is calculated for every damping configuration and every complex
vibration mode and is introduced by the following equation:
Z fi
10log10ðjσ r ðf Þ  σ n ðf ÞjÞ
CMRI ¼ df ; (15)
0 MCI
where σr is the radiation efficiency of the real mode of the reference plate without additional damping, the σn is the
radiation efficiency of the complex vibration mode and MCI its level of complexity. The CMRI represents the integral of
absolute deviations between the radiation efficiency of the real mode and the corresponding complex vibration mode for
the frequencies fi below the coincidence, normalised by the MCI value. According to the results from previous studies,
frequencies below 10 percent of coincidence are taken into account. In other words, the CMRI indicates with higher values a
bigger deviation of complex modes from the real modes per MCI unit of complexity.
Fig. 8 shows the CMRI values of complex vibration modes for the investigated damping configurations. In this analysis
the lower 17 modes are taken into account, because of the noticeable amplitude redistribution of modes with higher order,
as discussed in Section 3.1. Too high redistribution of vibration amplitudes, due to a very high damping, do not allow a
reliable statement about the influence of pure complexity in form of travelling bending waves.
First it can be noticed that even within a single damping configuration, there is a remarkable variation in evaluation
coefficients depending on the number of the mode. For example the bilateral symmetric B1 configuration reveals the CMRI
from 6  100 to 1:5  103 . In average the CMRI values are approximately around 102, what corresponds to deviation of
radiation efficiencies of 1 to 2 dB. Regarding the acoustic impact of configurations with different symmetry it is found that
the symmetric damping distributions S1–S5 never exceed values of 3  102. In contrast, the bilateral configurations B1–B3
can reach values of 9  103. The highest observable values of up to 3  104 are reached at the asymmetric A1–A2 and central
symmetric C1–C2 configurations. It is remarkable that central symmetric distributions show very high CMRI values only for
four even–even order modes with numbers 5, 9, 15 and 17. In contrast, configurations with asymmetric damping affect
much more modes and also reveal a higher averaged level of CMRI values. To illustrate this, Fig. 9 shows the CMRI values for
every damping distribution, which are average over all considered modes of the corresponding configuration. For con-
venience they are additionally normalised by the maximum value, found in the A1 configuration.
The main conclusions, from Sections 4.1 and 4.2, can be confirmed by the average CMRI values. Symmetric damping
distributions show a negligible effect of travelling waves on the self-radiation efficiencies. Plates with bilateral symmetry
show three to ten times higher impact on sound radiation than symmetrical configurations. Finally, the central symmetric
and asymmetric damping distributions reveal the highest influence on self-radiation efficiency of complex vibration modes.
It seems that the application of inhomogeneous damping with bilateral, central symmetric and asymmetric properties can
lead to an increased sound radiation of structural resonances and should be avoided at low frequencies below 10 percent of
coincidence.

4.4. Mutual-radiation efficiencies

As it is mentioned in Section 3.2, structural modes do not radiate sound independently of each other. The acoustic
interference between structural modes is described by the mutual-radiation efficiencies, defined in Eq. (12). Especially
between the resonance frequencies, the dynamic response implies a superposition of different modes and therefore the
contribution of mutual-radiation efficiencies to the radiated sound power grows. It is important to notice that the cross-
coupling only occurs between structural modes with similar order. This means that for example even–even order modes
only influence even–even order modes. Accordingly, the mutual-radiation efficiencies of even–even with odd–odd, even–
odd or odd–even order modes are zero.
Fig. 10 shows the mutual-radiation efficiencies for the reference plate with real structural modes. Besides the mentioned
group-wise cross-coupling between the structural modes it can be noticed that modes with similar order and therefore
similar bending wavelengths such as 2
2 and 2
4 have higher mutual efficiencies than for example 2
2 and 6
8
modes. Furthermore, it can be seen that the mutual efficiencies as a function of frequency can be positive or negative. This

Fig. 9. Normalised and averaged CMRI values.

180 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184

Fig. 10. Mutual-radiation efficiencies of the real structural modes.

Fig. 11. Mutual-radiation efficiencies of real and complex modes with similar orders.

means that the acoustic interaction between neighbouring resonances can be constructive (increasing sound power) or
destructive (decreasing sound power). According to Eq. (11) it is important that the positive or negative contribution to the
radiated sound power of the plate is also depending on the modal coordinates q, which assign the amplitude and phase of
different modes to each other.
Mutual-radiation efficiencies of complex modes, which result from different damping distributions, can be calculated
similar to real modes using Eq. (12). Fig. 11 shows the mutual-radiation efficiencies of the odd–odd 1
1 and 3
1 modes, as
well as for the even–even 2
2 and 4
2 modes. Due to the fact that the mutual-radiation efficiencies of complex vibration
modes are also complex, they are presented as frequency dependant locus curves in the complex plane. The locus curves are
shown for the frequencies between 10 and 2500 Hz and start at the coordinate origin. As expected, both pairs of real
structural modes of the reference plate stay on the real axis. In contrast, the mutual-radiation efficiencies of inhomogen-
eously damped plates extend through the entire complex plane. Due to the fact that matrix Π is hermitian the mutual-
 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184 181

radiation efficiencies σij and σji are conjugate complex. Therefore the algebraic sign of the imaginary parts is determined by
the choice of either σij or σji.
The complexity of the mutual efficiencies complicates the interpretation of the associated physical phenomena. In order
to explain how complex mutual-radiation efficiencies contribute to the radiated sound power, Eq. (11) is considered using a
simple example of a system with two modes. Knowing that the matrix Π is hermitian the sound power of a simplified
system with two natural modes can be specified as follows:
"  #T " #" #
q1 Π 11 Π 12 q1
W¼ ; (16)
q2 Π 21 Π 22 q2
where Π11, Π22 are the contributions of the self-radiation efficiencies and Π12, Π21 are conjugate values that are related to
mutual-radiation efficiencies. The values q1 and q2 are the corresponding modal coordinates of the two investigated
structural modes. After the multiplication of matrix and vector, the equation rearranges to:
W ¼ q1 Π 11 q1 þq2 Π 22 q2 þq1 Π 12 q2 þq2 Π 21 q1 : (17)

The first two terms in this equation correspond to the contributions of the self-efficiencies to the radiated sound power. The
last two terms result from the cross-coupling of both mode shapes and represent the contribution of the mutual-radiation
efficiencies. Due to the fact that both cross-coupling terms q1 Π 12 q2 and q2 Π 21 q1 are conjugate complex, the overall con-
tribution of the mutual-radiation efficiencies results in:
W m ¼ q1 Π 12 q2 þq2 Π 21 q1 ¼ 2Reðq1 Π 12 q2 Þ: (18)

This equation shows that only the real parts of the mutual-radiation efficiencies contribute to the radiated sound power.
Important is that the quantity of the real part Reðq1 Π 12 q2 Þ depends on the phase angle given by the modal coordinates q1
and q2. This means that the locus curve given in Fig. 11 rotates around the coordinate origin depending on the phase angle
between two modes given by their modal coordinates. The projection of the locus curve on the real axis gives the absolute
value and the algebraic sign of the contribution to the radiated sound power.
For complex modes, it is required to investigate the influence of the eigenvector complexity on mutual-radiation effi-
ciencies of pairings between modes with different orders e.g. even–odd and odd–even. As a reminder, in the case of real

Fig. 12. Mutual-radiation efficiencies of real and complex modes with different orders.
182 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184

Fig. 13. Off-diagonal elements Πij at 400 Hz.

mode shapes these mutual-radiation efficiencies are equal to zero. Fig. 12 shows locus curves of mutual-radiation effi-
ciencies for some mode pairings of the reference plate and the configurations A1, B1, C1. As expected, the locus curve of the
real mode shapes of the reference plate remains in the coordinate origin at a value of zero. In contrast to that, the locus
curves of inhomogeneously damped configurations are non-zero. This means that travelling waves increase the mutual
interaction between complex mode shapes of different order. In order to consider this effect in more detail, Fig. 13 visualises
the logarithmic off-diagonal elements Πij of the matrix Π, defined in Eq. (11) for a frequency of 400 Hz. This frequency is
selected in order to visualise the maximum mutual-radiation efficiencies. The dominant diagonal elements Πii, which are
correlated to the self-radiation efficiencies, are set to zero in order to focus on the lower valued mutual efficiencies.
In Fig. 13 it can be seen that the reference plate with real modes has a little number of spots that are not equal to zero.
Compared to the reference plate, the symmetric damping distribution S3 reveals a marginal and almost negligible increase
of acoustic interaction between the complex vibration modes. In contrast to that, the bilateral symmetric configuration B1
shows an improved cross-coupling between complex modes, which can be seen by an increased number of non-zero
elements. A similar behaviour can also be observed for the configuration C1 with a central symmetric damping distribution.
The influence of complex modes on the mutual-radiation efficiencies is especially apparent considering the asymmetric
configuration A1. It can be noticed that in this case all modes interact with each other. The highest sensitivity of this
asymmetric configuration regarding the acoustic properties of present complex modes has already been observed using
self-radiation efficiencies and couplings into lower radiation modes in Sections 4.1 and 4.2. Due to the fact that the
asymmetric damping distribution induces bending waves that are travelling diagonally, all nodal lines of the mode shape are
also travelling in this direction. This travelling of the nodal lines blurres the opposite properties of even and odd order
modes and increases the acoustic interaction between them, which raises the mutual-radiation efficiencies.

4.5. Discussion

In this part of the paper, several important issues regarding technical relevance of the presented results are discussed.
One of the questions is whether it is important to consider the complex mode shapes in acoustic simulations, which are for
example conducted in order to design low-noise structures. Despite the fact that most of the observed effects address very
low frequencies below the coincidence frequency, the consideration of complex vibration modes seems to be important for
accurate predictions of sound relevant design parameters. This might be relevant for bilateral and central symmetries and
 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184 183

seems to be even more important for asymmetrical distributions of inhomogeneous damping. It is also important to notice
that if the placement of additional damping is carried out by conventional methods based on modal strain energy [3] or
structural intensity distribution [4], symmetrical damping configurations will be the result of this optimisation process. In
this case the consideration of complex mode shapes and their acoustic characteristics is not mandatory and can be limited to
real modes.
Nevertheless, in the case of asymmetric damping distributions with very high damping ratios, which result in vibration
modes with high complexity, the effects presented in this paper should be considered in order to optimise the acoustic
performance of the vibrating structure. Due the fact that an increase of the radiation efficiency of even order modes is
expected at frequencies lower than 10 percent of the coincidence frequency, especially plates with very low eigen-
frequencies (high mass and low stiffness) and small scales (small bending wavelengths and large distance to the coincidence
frequency) are affected by presented phenomena. This means that lightweight structures with high stiffness and low mass
are not seriously affected by effect presented in this paper.
The discussion of symmetry in acoustics is not a new topic. The influence of symmetry has already been investigated by
Constans et al. [24] who uses an optimisation tool for the placement of point masses in vibrating shells. Supplementary to
the state of the art the presented paper addresses the influence of symmetry from a different perspective by putting the
complex vibration modes into account. Regarding the results of this paper, the symmetric placement of damping treatments
must be aimed in order to avoid the increase of radiation efficiency of complex vibration modes. However, it must be taken
into account that perfectly symmetrical configurations can hardly be produced in experiment due to uncertainties as it has
already been stated e.g. by Marburg in [25].

5. Conclusions

This paper investigates the influence of different spatial damping configurations on the resulting complex mode shapes
and their sound radiation properties. It is shown that symmetric, bilateral symmetric, central symmetric and asymmetric
distributions of inhomogeneous damping cover the majority of cases that can probably be expected in real applications.
Using an example of five configurations with different damping properties, the self-radiation efficiencies of complex
vibration modes are considered in detail. First, it is shown that the symmetric damping distributions do not affect the sound
radiation of complex modes. For bilateral symmetric configuration it is shown that modes with even order in the direction
of travelling waves significantly increase and modes with odd order slightly decrease their radiation capability below the
coincidence frequency. An interesting observation is made regarding the central symmetric configuration, which affects only
even–even order modes. Other groups of modes seem to remain unaffected in their radiation. The highest influence on the
self-radiation efficiencies is observed for asymmetric damping distributions. According to these results, all structural modes
with even order in any direction of space increase their radiation.
A new Complex Mode Radiation Index (CMRI) is introduced, which allows the evaluation of the influence on self-
radiation efficiencies below the coincidence frequency, normalised by complexity level, estimated by MCI. Using this metric
it is shown that symmetric damping distributions have an almost negligible influence on sound radiation. Furthermore, it is
noticed that bilateral and central symmetric distributions affect radiation capability of certain groups of modes. Never-
theless, the highest influence on sound radiation of the majority of complex vibration modes is observed by the asymmetric
damping configuration.
The consideration of the mutual-radiation efficiencies reveals that in the case of complex vibration modes this metric
also becomes complex valued. It is shown that the contribution of cross-couplings to the radiated sound power is related to
the real part of the complex mutual efficiencies. The quantity of the real part and therefore the amount and algebraic sign of
the contribution depend on the amplitude and phase given by the modal coordinates. The most important result regarding
the mutual-radiation efficiencies is that, depending on the symmetry of the damping distribution, the acoustic coupling
between structural modes increases, even in the case of different orders. Especially the asymmetrically distributed inho-
mogeneous damping leads to an increase of all mutual-radiation efficiencies.
With regard to the application of additional damping treatments the results of this paper show that, in order to avoid an
increase of the self-radiation efficiencies of complex vibration modes, the damping should be placed symmetrically. Bilateral
symmetric, central symmetric and especially asymmetric distributions of additional damping may improve the radiation
capability of structural resonances and increase the emitted noise levels. Nevertheless this increase is expected well below
the coincidence frequency for plates with small scale and very low eigenfrequencies.
Future works will include the investigation of the influence of complex vibration modes on the sound power radiated by
the inhomogeneously damped plates. Furthermore, an important question is the coupling of complex vibration modes in the
fluid modes of an adjacent cavity. The coupling of different types of excitations such as plane acoustic waves and turbulent
boundary layers can also have an importance for inhomogeneously damped plates.
184 O. Unruh / Journal of Sound and Vibration 377 (2016) 169–184

References

[1] E.M. Kerwin Jr, Damping of flexural waves by a constrained viscoelastic layer, The Journal of the Acoustical Society of America 31 (1959) 952.
[2] H. Kishi, M. Kuwata, S. Matsuda, T. Asami, A. Murakami, Damping properties of thermoplastic-elastomer interleaved carbon fiber-reinforced epoxy
composites, Composites Science and Technology 64 (2004) 2517–2523.
[3] F. Curá, A. Mura, F. Scarpa, Modal strain energy based methods for the analysis of complex patterned free layer damped plates, Journal of Vibration and
Control 18 (9) (2011) 1291–1302.
[4] D.H. Kruger, J.A. Mann III, T. Wiegandt, Placing constrained layer damping patches using reactive shearing structural intensity measurements, The
Journal of the Acoustical Society of America 101 (1997) 2075.
[5] D.S. Nokes, F.C. Nelson, Constrained layer damping with partial coverage, Shock and Vibration Bulletin 38 (1968) 5–10.
[6] B. Khalfi, A. Ross, Transient response of a plate with partial constrained viscoelastic layer damping, International Journal of Mechanical Sciences 68
(2013) 304–312.
[7] L.-H. Chen, S.-C. Huang, Vibrations of a cylindrical shell with partially constrained layer damping (cld) treatment, International Journal of Mechanical
Sciences 41 (1999) 1485–1498.
[8] G. Lang, Demystifying complex modes, Sound and Vibration 23 (1989) 36–40.
[9] G. Xie, D. Thompson, C. Jones, The radiation efficiency of baffled plates and strips, Journal of Sound and Vibration 280 (2005) 181–209.
[10] F. Fahy, P. Gardonio, Sound and Structural Vibration: Radiation, Transmission and Response, Elsevier, Amsterdam, Boston, Heidelberg, London, New York,
Oxford, Paris, San Diego, San Francisco, Singapore, Sydney, Tokyo, 2007.
[11] H.-W. Wodtke, J. Lamancusa, Sound power minimization of circular plates through damping layer placement, Journal of Sound and Vibration 215 (1998)
1145–1163.
[12] S. Marburg, Normal modes in external acoustics. Part III: sound power evaluation based on superposition of frequency-independent modes, Acta Acustica
United with Acustica 92 (2006) 296–311.
[13] J. Torres, P. Rendón, R. Boullosa, Complex modes of vibration due to small-scale damping in a guitar topplate, Journal of Applied Research and Technology
8 (01) (2010).
[14] O. Unruh, M. Sinapius, H. Monner, Sound radiation properties of complex modes in rectangular plates: a numerical study, Acta Acustica United with
Acustica 101 (2015) 62–72.
[15] O. Unruh, Parametric study of sound radiation properties of complex vibration patterns in rectangular plates using an analytical model, Acta Acustica
United with Acustica 101 (2015) 701–712.
[16] W. Li, H. Gibeling, Determination of the mutual radiation resistances of a rectangular plate and their impact on the radiated sound power, Journal of
Sound and Vibration 229 (2000) 1213–1233.
[17] D.I. Jones, Handbook of Viscoelastic Vibration Damping, John Wiley & Sons, Ltd, Chichester, New York, Weinheim, Brisbane, Singapore, Toronto, 2001.
[18] M. Oude Nijhuis, Analysis Tools for the Design of Active Structural Control Systems, PhD Thesis, University of Twente, 2003 (PhD dissertation).
[19] K. Cunefare, G. Koopmann, Global optimum active noise control: surface and far-field effects, The Journal of the Acoustical Society of America 90 (1991)
365.
[20] M. Imregun, D. Ewins, Complex modes-origins and limits, in: Proceedings of SPIE, The International Society for Optical Engineering, pp. 496–496.
[21] K. Liu, M. Kujath, W. Zheng, Quantification of non-proportionality of damping in discrete vibratory systems, Computers & Structures 77 (2000) 557–570.
[22] H. Koruk, K.Y. Sanliturk, A novel definition for quantification of mode shape complexity, Journal of Sound and Vibration 332 (14) (2013) 3390–3403.
[23] S. Elliott, M. Johnson, Radiation modes and the active control of sound power, The Journal of the Acoustical Society of America 94 (1993) 2194.
[24] E. Constans, G. Koopmann, A. Belegundu, The use of modal tailoring to minimize the radiated sound power of vibrating shells: theory and experiment,
Journal of Sound and Vibration 217 (1998) 335–350.
[25] S. Marburg, H.-J. Beer, J. Gier, H.-J. Hardtke, R. Rennert, F. Perret, Experimental verification of structural-acoustic modelling and design optimization,
Journal of Sound and Vibration 252 (2002) 591–615.