You are on page 1of 11

Journal of Sound and Vibration 340 (2015) 343–353

Contents lists available at ScienceDirect

Journal of Sound and Vibration


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

Mode selection of modal expansion method estimating


vibration field of washing machine
B.K. Jung, W.B. Jeong n
School of Mechanical Engineering, Pusan National University, Busan 609-735, Republic of Korea

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

Article history: This paper is about a study estimating the vibration and radiated noise of a washing machine
Received 10 February 2014 by using a mode selection-applied modal expansion method (MEM). MEM is a technique that
Received in revised form identifies the vibration field from a portion of eigenvectors (or mode shapes) of a structure,
10 December 2014
and thus, the selection of the eigenvectors has a big impact on the vibration results identified.
Accepted 13 December 2014
However, there have been few studies about selecting the eigenvectors with respect to the
Handling Editor: H. Ouyang
Available online 6 January 2015 structural vibration and radiated noise estimation. Accordingly, this paper proposes the use of
a new mode selection method to identify the vibration based on the MEM and then calculate
radiated noise of a washing machine. The results gained from the experiment were also
compared. The vibration and noise results of numerical analysis using the proposed selection
method are in line with the measured results. The selection method proposed in this paper
corresponds well with the MEM and this process seems to be applicable to the estimation of
various structure vibrations and radiated noise.
& 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Advanced mechanical engineering has led to an improvement of people's attitude toward life and an increase in people's
interest in improving living conditions. Consequently, noise and vibration issues related to household electrical appliances
have been rising. The noise and vibration of home appliances are likely to primarily shorten the lifespan of products, and
secondly, bring physical and psychological damage to indoor residents and users. Against this backdrop, the home appliance
industry has been recently carrying out studies on noise and vibration reduction [1–4], and as a part of those studies, this
paper analyzed the noise and vibration of household washing machines.
There are many experimental or numerical studies about the vibration and noise characteristics of washing machine. Agnani
researched the dynamic characteristics of the machine by using the multi-body analysis and compared the numerical results
with the experiment results [5]. Chiariotti researched the noise source of washing machine by using sensor array technique and
near field acoustic holography [6]. Jung mentioned from his study that the noise of the washing machine can be divided into the
structure-borne noise caused by tub and cabinet vibration and the air-borne noise caused by fluid behavior within a drum and
drain pump [7]. Commonly, the air-borne noise is more dominant than the structure-borne noise in the washing machine.
However, the home appliance industry often makes an effort to reduce the structure-borne noise through the structural
improvement in the product development phase. Thus, this paper estimates the washer's vibration and structure-borne noise at
the time of dehydration without fluid by using a finite element method (FEM) and a boundary element method (BEM).

n
Correspondence to: Department of Mechanical Engineering, Pusan National University, Jangjeon-dong, Kumjung-ku, Pusan 609-735, Republic of Korea.
Tel.: þ 82 51 510 2337; fax: þ82 51 517 3805.
E-mail address: wbjeong@pusan.ac.kr (W.B. Jeong).

http://dx.doi.org/10.1016/j.jsv.2014.12.016
0022-460X/& 2014 Elsevier Ltd. All rights reserved.
344 B.K. Jung, W.B. Jeong / Journal of Sound and Vibration 340 (2015) 343–353

FEM refers to the calculation of approximate solutions numerically by dividing into finite small areas the natural
phenomenon influenced by a specific rule. This is one way to reproduce static and dynamic phenomenon. However, it is very
important to implement an accurate exciting force so as to reproduce the exact dynamic phenomenon. The exciting force
which causes the vibration and noise of a structure is significant information in dynamic analysis, but it is often impossible
to directly measure it owing to geometric limits. Accordingly, Otsuka [8] and Jung [9] were involved in a study that
estimated the exciting force from vibration signals, which are relatively easy to measure, based on a numerical analysis. But,
in the case of a study on the exciting force identification [10,11], it is important to implement the exact transfer function in
producing reliable exciting force. Considering this, it is very difficult to implement precisely the transfer function in the case
of a complex shaped structure composed of various single pieces. On the other hand, if it is properly materialized in a
geometric perspective, it will be relatively easier to reproduce the eigenvector of a structure compared to the transfer
function. As a result, there have been studies about the modal expansion method (MEM), which estimates the structural
vibrations by using the structure's eigenvectors and vibration response signals [12–14]. The MEM refers to a way of
estimating the modal participation factor (MPF), which is the level of contribution of the eigenvector, can express the
vibration field in a faster manner and has relatively less computation in comparison to the force identification method.
However, given that the MEM is needed to estimate the vibration by using some of the numerous eigenvectors, the MEM has
an issue with selecting the eigenvectors. One of the representative mode selections includes the MSC. Nastran's selection
method, which uses effective mass [15,16], but the method has not been verified in terms of MEM and radiated noise.
Accordingly, this paper intends to describe a new mode selection method that considers the radiated noise and the MEM,
and estimates, based on the MEM, the vibration field that occurs when a washing machine is operated at the maximum
dehydration rate of 1370 rev/min. To that end, in this paper, the vibrations on the surface of a washing machine were
measured through an experiment, and the eigenvalue problem of the FE model was performed numerically. Additionally,
principal eigenvectors were selected with the utilization of the proposed selection. Based on this, this paper is designed to
estimate the vibration results on the surface of a washing machine by using the MEM, and additionally, the radiated noise is
estimated with the utilization of the acoustic direct BEM at the time of dehydration. These numerical results were compared
with the experimental results.

2. Vibration and noise identification theory and mode selection

2.1. MEM theory

The MEM is a method to estimate the MPF by using eigenvectors and the measured signals obtained from the surface of a
structure. This MEM theory originated from Eq. (1), which is the linear vibration system of the multidegree of freedom.
M x€ þ C x_ þKx ¼ F (1)
where, M refers to the mass matrix of a structure, C the damping matrix, K the stiffness matrix, F the external force vector. x,
x_ and x€ mean vibration displacement, velocity and acceleration vector, respectively. In order to calculate the forced vibration
response of Eq. (1), the free vibration should be first calculated based on an eigenvalue analysis of

Kφr ¼ ω2r Mφr (2)

where, ωr refers to the r-th natural frequency, and φr to the r-th eigenvector. Using these factors, the eigenvector matrix Φ
is represented as
 
Φ ¼ φ1 ; φ2 ; …; φr (3)

When the equation solution of Eq. (1) is represented in a modal coordinate using the above factors, the result is as seen in
the following equation:
x ¼ Φu (4)
where, u called MPF which means the eigenvector's contribution level or displacement of the modal coordinates.
The process of calculating the forced vibration response by using these factors can be seen, as shown below. First, after
Eq. (4) is substituted into Eq. (1), the equation of the physical coordinates is converted into the modal coordinates.
 
 ω2 Φ M Φ þjωΦ C Φ þ Φ K Φ u ¼ Φ F
T T T T
(5)

The modal coordinates go through the matrix diagonalization and then are represented as independent coordinates as
below.
 
 ω2 I þ2jωζ r ωr þ ω2r u ¼ qr (6)

Where, I refers to the identity matrix, ω the frequency, and ζ r to r-th the damping ratio. Based on the above equation, the
MPF vector, u, is defined as
 1
u ¼  ω2 I þ2jωζ r ωr þ ω2r qr (7)
B.K. Jung, W.B. Jeong / Journal of Sound and Vibration 340 (2015) 343–353 345

MEM that calculates the MPF by using the known response signals and the eigenvector matrix is as follows.
xm ¼ Φm u (8)
Where, xm refers to the vibration response vector of the point measured during an experiment, and Φm the eigenvector
matrix corresponding to the measured points. The method of estimating u is shown in the following equation:
þ
u ¼ Φm xm (9)
In the above equation,þ refers to pseudo-inverse, indicating that Φm may not be a square matrix. Generally, in the case
where the pseudo-inverse is used for a problem, a more accurate solution can be obtained when the number of equations
outnumbers the unknowns. In other words, a more accurate and reliable u cannot be estimated until the number of sensors
measured during an experiment outnumbers the eigenvectors used.
Eq. (10) shows the process of calculating the normal direction velocity of all points for radiated noise calculation with the
utilization of the calculated MPF vector.
Vn ¼ jωTx (10)
where, Vn refers to the normal-directional velocity vector, and T represents the transformation matrix for converting x into a
normal-directional response vector.

2.2. Mode selection theory

MEM uses a limited number of eigenvectors in order to estimate the response of overall points from vibration responses
of several points. However, as the structure has a theoretically unlimited number of eigenvectors, there is an arising issue of
which eigenvectors should be selected and used. Generally, the more complex or bigger finite element model of the
structure, the bigger the degree of freedom. As a result, a number of eigenvectors exist within the frequency bands of
interest in the case of the eigenvector analysis of the finite element model. Therefore, it is necessary to select principal
eigenvectors for applying MEM.
A representative selection method is one that uses the effective mass offered by MSC.Nastran. This method is composed
of that shown in Eq. (11), based on a theory that any eigenvector with bigger effective mass makes a bigger contribution to
the vibration response.
X
6  
EMr ¼ φT Mφ 2 (11)
r D
D¼1

where, φD refers to the eigenvector corresponding to rigid-body motions. By using the above equation, EMr which is the r-th
effective mass, can be calculated. By aligning in the order of biggest to smallest based on the calculated value of EMr , the
principal eigenvectors with a higher participation factor can be selected. It is a suitable mode selection method for the
estimation of the vibration, but it is not verified for the estimation of the radiated noise. Accordingly, this paper proposed
new mode selection method.
The vibration response is composed of the superposition of modes (or eigenvectors). So, using global modes has an
advantage to represent the vibration field rather than using local modes. Also, the radiation efficiency of global modes is
higher than that of local modes. Especially, bending modes of global modes have very high radiation efficiency. The new
mode selection method was based on these facts mentioned above. The index is primarily split into global modes and local
modes by dividing the average value of the normal-directional vibration by the maximum value. The transformation matrix
T is used to transform it into a normal-directional vibration and to consider bending modes which have high radiation
efficiency. This can be expressed, as shown below
 
E Tφr 
GLr ¼    (12)
max Tφ  r

where, GLr refers to a value of the proposed index corresponding to the r-th eigenvector, ‘E’ to a process of averaging and
‘max’ to a function that seeks the maximum value. In the above equation, whose index value ranges from 0 to 1, if it is closer
to 0, the index means the local vibration that has less contribution to the noise. On the other hand, if it is closer to 1, the
index means the global vibration that has a large contribution level to the noise.
This paper used GLr and EMr to select the principal eigenvectors corresponding to Φm , and then MEM was implemented
to estimate the vibration noise of a washing machine.

2.3. Radiated noise theory

In this paper, acoustic BEM was used to analyze the external radiated noise of a washing machine at the time of
dehydration. The acoustic BEM is split into a direct method [17,18] and an indirect method [19]. For the radiated noise
analysis of a closed space, the direct method is used, while for the radiated noise analysis of an open space, the indirect
method is used. In this paper, the direct method was used to predict the radiated noise from overall vibrations of a washing
machine. The method utilizes the surface pressures and normal velocities to calculate the sound pressure away from the
346 B.K. Jung, W.B. Jeong / Journal of Sound and Vibration 340 (2015) 343–353

surface of the structure. It has the following governing equation:


Z

∂ψ ðx; yÞ
C ðxÞP ðxÞ ¼ P ðyÞ þjρωV n ðyÞψ ðx; yÞ dS (13)
S ∂n

In the above equation, x means any location of field point, y to one location on the surface of an acoustic grid, ω to frequency,
ρ the density of a medium, n the normal vector, S the surface, P the acoustic pressure, V n the normal velocity, and ψ ðx; yÞ the
Green function, as shown below

ejkR
ψ ðx; yÞ ¼ (14)
4π r
 
where, R refers to distance between x and y (R ¼ x  y), and k to ω=c with the meaning of a wavenumber. According to
Eq. (13), C ðxÞ means a function dependent on the location of field point. In the case of the external radiated noise issue, it is
defined as
8
>
> 1 ðin domainÞ
< R ∂ð1=4π RÞ
C ðxÞ ¼ 1  S ∂n dS ðon boundaryÞ (15)
>
>
:0 ðout domainÞ

This paper uses the MEM to calculate V n ðyÞ and then uses Eqs. (13)–(15) to obtain the surface pressure P ðyÞ. Additionally, the
radiated noise analysis was implemented to estimate the acoustic pressure P ðxÞ at the field point by using the estimated
surface pressures and normal velocities.

3. Vibration field visualization

3.1. Finite element model and eigenvector analysis of a washing machine

The finite element model was built in order to realize numerically the vibrations occurring at the dehydration of a
washing machine. In this case, the outer face of the structure was accurately modeled in order to properly realize the surface
vibrations important in calculation of radiated noise, while internal structures were simplified with equivalent mass. The
equivalent mass models of the internal structures were connected with outer face of the washing machine by using two
spring elements and three damper elements. Boundary condition of the finite element model was the clamped condition on
the legs of the washing machine. In the case of the finite element analysis, a commercial program, MSC.Nastran, was used
for the eigenvector analysis, and the numbers of nodes and elements used for the analysis were 101,766 and 100,642,
respectively. Fig. 1 displays the finite element model of a washing machine modeled based on what was previously
mentioned. Fig. 2 shows the comparison of the natural modes and Fig. 3 shows the comparison of the frequency response
functions between the FE analysis and the experiment. These figures represent the validation of the finite element model.

3.2. Frequency vibration response acquisition experiment

It is necessary to measure the frequency responses at some points of surface so that the vibration field of a washing
machine at the time of dehydration can be predicted based on MEM. Thus, LMS.Scadas, FFT analyzer, was used and
acceleration signals were measured at an interval of 1 Hz from 0 Hz to 2048 Hz, when a washing machine reached the top
level which runs at the maximum dehydration velocity of 1370 rev/min. Hanning-windowed response signals with 400
times averaging to reduce noise were gained. When it comes to the measurement location, a total of 34 points of
acceleration responses were measured with a composition of 8 points in front of the cabinet, 5 points on the right, 6 points
on the left, 10 points in the back, and 5 points on the top plate of a cabinet so as to properly represent the overall vibrations
of the structure. Fig. 4 shows those points. The two verification points also shown in the figure were designed to verify the
validity of the predicted analysis results.

3.3. Mode selection and MEM application

This paper sets the frequency bandwidth of interest from 63 to 1000 Hz in the form of a 1/3 octave band, where 719
eigenvalues were present. Then, the two mode selection methods previously introduced in Section 2.2 were used to select
30 principle eigenvectors. The numbers of 30 eigenvectors are attributable to the fact that more reasonable results can be
obtained when the number of response points (34 points) outnumbers the number of eigenvectors used in the MEM. Table 1
refers to the order of the principle eigenvectors selected based on GLr or EMr , as well as relevant index values.
When taking a closer look at the selected eigenvectors identified in Table 1, it can be seen that those eigenvectors below
the 30th are selected in a similar manner, but others show different tendencies. These selected principle eigenvectors and
measured acceleration signals at the response points were utilized in the MEM. As the MEM of Eq. (9) uses the displacement
B.K. Jung, W.B. Jeong / Journal of Sound and Vibration 340 (2015) 343–353 347

Fig. 1. The finite element model of a washing machine.

Fig. 2. Comparison of the natural modes of the cabinet of a washing machine: (a) numerical modes and (b) experimental modes.
348 B.K. Jung, W.B. Jeong / Journal of Sound and Vibration 340 (2015) 343–353

1
10

0
10

Accelerance [(m/s 2)/N]


-1
10

-2
10

-3
10
Experimental FRF
Predicted FRF
-4
10
50 100 150 200 250 300 350 400
Frequency [Hz]

Fig. 3. Comparison of the experimental and predicted FRF (accelerance) of a washing machine.

Fig. 4. Measurement and verification points.

vector, the relational equation shown in Eq. (16) was used for transforming from acceleration to displacement.
x€ m
xm ¼  (16)
ω2
Using the displacement based on the above equation and the principle eigenvectors, the MPF was calculated from the
equation shown in Eq. (9). Fig. 5 illustrates with the 5th and 9th MPF, which are commonly shown in the selection of GLr
and EMr , where a bar with solid pattern refers to the MPF calculated by GLr , and a bar with slanted-line pattern refers to
MPF calculated by EMr . The 5th and 9th MPF originated from EMr are generally bigger than the MPF from GLr in overall
frequencies.
Based on the estimated MPF vector of u, the acceleration responses of the all nodes were calculated, as shown in Eq. (17),
by using the eigenvector matrix Φ corresponding to the overall system of the washing machine.

x€ ðωÞ ¼  ω2 xðωÞ ¼  ω2 ΦuðωÞ (17)

Fig. 6 illustrates with the 1/3 octave band a comparison result between the experimental vibration results and the
acceleration responses at the verification points calculated by Eq. (17). The dB reference of acceleration used at this time is
110  6 m s  2. The bar with solid pattern in Fig. 6 refers to the acceleration response of the experiment, and bars with
B.K. Jung, W.B. Jeong / Journal of Sound and Vibration 340 (2015) 343–353 349

Table 1
Results of two mode selection methods.

Order of eigenvectors GLr Order of eigenvectors EMr

5 0.5729 5 0.3134
2 0.5600 11 0.2442
3 0.3031 10 0.2415
4 0.1860 14 0.2323
15 0.1072 9 0.2267
9 0.0992 19 0.2144
10 0.0825 15 0.2123
48 0.808 34 0.1964
35 0.0427 2 0.1698
13 0.0373 4 0.1686
6 0.0278 44 0.1666
50 0.0267 6 0.1556
11 0.0264 7 0.1543
47 0.0232 24 0.1540
30 0.0188 3 0.1472
14 0.0172 12 0.1354
58 0.0143 151 0.1339
85 0.0140 48 0.1326
54 0.0124 33 0.1325
19 0.0116 28 0.1296
46 0.0107 20 0.1245
61 0.0104 43 0.1243
40 0.0104 116 0.1228
12 0.0099 8 0.1215
62 0.0098 18 0.1174
38 0.0097 17 0.1147
92 0.0089 22 0.1113
27 0.0088 31 0.1094
28 0.0086 45 0.1088
37 0.0080 65 0.1073

Fig. 5. Comparison of the MPFs estimated by MEMs using the mode selections: (a) 5th MPF and (b) 9th MPF.
350 B.K. Jung, W.B. Jeong / Journal of Sound and Vibration 340 (2015) 343–353

Fig. 6. Comparison of the accelerations at verification points between measured and estimated results: (a) verification point # 1 and (b) verification point # 2.

Table 2
Errors and standard deviation of the acceleration at verification points.

Frequency Error at point.1 (dB) Error at point.2 (dB)

GLr EMr GLr EMr

63  5.23  11.79  1.45  0.48


80  1.19  13.75 0.17 0.04
100 0.81 0.27  8.14  6.83
125  4.39  19.35 3.77  6.07
160  2.66  2.33 0.42  3.43
200 0.06 0.04  1.47  2.63
250  1.40  2.37  2.88  3.80
315  2.46  5.80 5.39 5.31
400  1.24  6.27 6.89 7.33
500  0.53  0.69 3.27 4.80
630  0.55  3.88 3.29 3.20
800  2.19  1.99 6.29 7.11
1000  2.77  2.58 7.74 8.11
Standard deviation 2.45 7.91 4.73 5.18

slanted-line pattern and crisscross pattern refer to the acceleration responses in the MEM by using GLr or EMr , respectively.
For the numerical comparison, Table 2 describes the acceleration errors between the experiment and analysis and standard
deviation of those errors. The standard deviation σ was calculated by using Eq. (18) below.
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
u P
u N 2
t i ¼ 1 ðErrorÞi
σ¼ (18)
N
where, N means the number of center frequencies in a 1/3 octave band between 63 and 1000 Hz. The acceleration vibration
results estimated by the MEM based on the two mode selection methods corresponded well to the experimental results, as
shown in (a) and (b) of Fig. 6. However, according to the numerically implemented Table 2, the acceleration result from
B.K. Jung, W.B. Jeong / Journal of Sound and Vibration 340 (2015) 343–353 351

using GLr has relatively fewer errors and a less standard deviation in most frequencies, compared to the result of EMr ,
thereby being thought to be a better one in terms of reproducing the experimental values.

3.4. Radiated noise analysis and verification

Using all of the surface vibration responses at the time of dehydration of a washing machine, which were previously
estimated in Section 3.3, an external radiated noise analysis was conducted based on the LMS.Sysnoise's acoustic direct BEM
[20]. Fig. 7 illustrates the acoustic boundary element model of a washing machine. The field point related to the numerical

Fig. 7. The boundary element model of a washing machine and a reflection plane.

Fig. 8. Comparison of the acoustic pressure at the field point between measured and estimated results.

Table 3
Errors and standard deviation of the acoustic pressure at field point.

Frequency Error at the field point (dB)

GLr EMr

63 4.11  4.32
80  12.69  18.65
100 1.21  3.49
125  2.70  12.89
160 0.55  2.11
200 5.52  3.57
250  1.28  11.85
315  2.13  14.38
400  9.16  19.95
500 7.47  9.40
630 0.02  10.23
800  2.42  10.52
1000  3.69  11.50
Standard deviation 5.43 11.58
352 B.K. Jung, W.B. Jeong / Journal of Sound and Vibration 340 (2015) 343–353

analysis was placed 1.0 m ahead of the machine and above the ground. A reflection plane was also installed on the ground to
reproduce real experimental conditions. In addition, a chief point was temporarily set with an aim to eliminate the internal
resonance. In case of the experiment, the microphone was set at the same position of the field point in the numerical
analysis. The acoustic signal was measured by using LMS.Scadas at an interval of 1 Hz from 0 Hz to 2048 Hz. A-weighting
was applied to the signal. Fig. 8 illustrates a graph which compares with the experimental pressure result (solid pattern) and
the acoustic pressure at the field point calculated by the radiated noise analysis using GLr (slanted-line pattern) or EMr
(crisscross pattern). At this time, the acoustic pressure's dB reference is 2  10  5 N m  2. Fig. 8 indicates that the acoustic
pressure at the field point estimated by the mode selection GLr proposed in this paper represents well the experimental
tendency in comparison to that of the other selection method. Table 3 depicts the relative errors and standard deviation σ
mentioned above about the acoustic pressure at the field point.
According to Table 3, it can be verified that the acoustic pressure resulted from using GLr has fewer errors and less
standard deviation compared to that of EMr . In the case of the overall SPL shown in Fig. 8, GLr has an error of about 1 dBA
while EMr of about 10 dBA indicate a significant difference from the experiment. The selection method proposed by this
paper is more reasonable not only in the vibration estimation, but also in the radiated noise prediction compared to the
existing method. Therefore, this method can be regarded as a more suitable mode selection for estimating the vibration and
noise of a washing machine using the MEM.

4. Conclusion

This paper used the MEM and the acoustic direct BEM in order to identify the vibration and noise of a washing machine
at the time of the dehydration. The MEM is a method to calculate the modal participation factors by using the experimental
vibration responses and the eigenvectors obtained from the finite element analysis. The vibration results estimated by the
MEM are affected by the eigenvectors used for the calculation of the MPFs. Therefore, the suitable mode selection method is
needed. This paper used two types of the mode selection methods, the well-known EMr of MSC.Nastran and the new index
GLr proposed in this paper. EMr generally used in the vibration analysis was calculated based on the effective mass, but GLr
was calculated based on the index of global/local modes. In spite of this difference between two methods, both estimated
vibration results calculated by using the GLr and the EMr corresponded well to the experimental results at the verification
points. However, estimated noise result calculated by using the GLr showed the more accuracy prediction result than that of
the EMr . Because the GLr mainly considered the bending modes which generally have the high radiation efficiency. In the
case of the numeric comparison using the errors and standard deviation, the values of the GLr were less than those of EMr .
In other words, the mode selection proposed in this paper was found to display reasonable results regarding the vibration
and noise estimation of a washing machine based on the MEM and the radiated noise analysis. Additionally, this process is
expected to be utilized for the vibration and noise analysis and design improvement of a washing machine.

Acknowledgment

This research was supported by Basic Science Research Program through the National Research Foundation of Korea
(NRF) funded by the Ministry of Education (NRF-2014R1A1A2054372).

References

[1] N. Qi, B.M. Gibbs, Circulation pumps as structure-borne sound sources emission to semi-infinite pipe system, Journal of Sound and Vibration 264 (2003)
157–176.
[2] H.S. Kim, J.E. Oh, The analysis of noise contribution about drum washer under dehydrating condition using multidimensional spectral analysis,
Transactions of the Korean Society for Noise and Vibration Engineering 17 (11) (2007) 1056–1063.
[3] K.H. Kang, J.E. Oh, The analysis of structure-borne noise due to cabinet vibration from operational drum type washing machine using TPA, in:
Proceedings of the KIEE Annual Autumn Conference, 2008, pp. 295–296.
[4] H.W. Chen, Q.J. Zhang, S.Y. Fan, Study on steady-state response of a vertical axis automatic washing machine with a hydraulic balancer using a new
approach and a method for getting a smaller deflection angle, Journal of Sound and Vibration 330 (2011) 2017–2030.
[5] A. Agnani, F. Cannella, M. Martarelli, G. Merloni, E.P. Tomasini, Dynamic characterization of a washing machine: numerical multi-body analysis and
experimental validation, in: Proceedings of the IMAC XXVI. Orlando, USA, 2008.
[6] P. Chiariotti, M. Martarelli, Noise source localization on washing machines by conformal array technique and near field acoustic holography, in:
Proceedings of the IMAC XXVIII. Jacksonville, USA, 2010.
[7] J.E. Jung, J.E. Oh, Sound quality evaluation for laundry noise by a virtual laundry noise considering the effect of various noise sources in a drum
washing machine, Transactions of the Korean Society for Noise and Vibration Engineering 22 (6) (2011) 564–573.
[8] T. Otsuka, T. Okada, T. Ikeno, K. Shiomi, M. Okuma, Force identification of an outboard engine by experimental means of linear structural modeling and
equivalent force transformation, Journal of Sound and Vibration 308 (2007) 541–547.
[9] B.K. Jung, W.B. Jeong, Estimation of vibration source and sound radiation of a refrigerator fan by using measured acceleration signals, Transactions of
the Korean Society for Noise and Vibration Engineering 21 (9) (2011) 834–841.
[10] E. Parloo, P. Verboven, P. Guillaume, Force identification by means of in-operation modal modes, Journal of Sound and Vibration 262 (2003) 161–173.
[11] Y.E. Lage, A.M.R. Ribeiro, Force identification using the concept of displacement transmissibility, Journal of Sound and Vibration 332 (2013) 1674–1686.
[12] P. Guisset, M. Brughmans, Modal Expansion of Experimental Vibration Data for Acoustic Radiation Prediction, SAE Technical Paper 951090, 1995,
10.4271/951090.
[13] P. Avitabile, J. O'Callahan, Frequency response function expansion for unmeasured translation and rotation DOFs for impedance modeling applications,
Mechanical Systems and Signal Processing 17 (4) (2003) 723–745.
B.K. Jung, W.B. Jeong / Journal of Sound and Vibration 340 (2015) 343–353 353

[14] B.K. Jung, W.B. Jeong, Estimation of vibration and radiated noise of a compressor by using modal expansion method, in: Proceedings of the KSNVE
Annual Spring Conference, 2011, pp. 554–555.
[15] M. Razeto, J.C. Golinval, M. Geradin, Modal Identification and Determination of Effective Mass Using Environmental Vibration Testing on an Electra-
Dynamic Shaker, PTDC. XII IMAC. Honolulu, Hawaii, 1994, pp. 648–654.
[16] V.L. Tom, B. Marc, Using MSC/Nastran and LMS/Pretest to find an optimal sensor placement for modal identification and correlation of aerospace
structures, LMS International, 2010.
[17] R.J. Bernhard, B.K. Gardner, C.G. Mollo, Prediction of sound fields in cavities using boundary element methods, AIAA Journal 25 (1987) 1176–1183.
[18] Z. Tong, Z. Zhang, Dynamic behavior and sound transmission analysis of a fluid-structure coupled system using the direct-BEM/FEM, Journal of Sound
and Vibration 299 (2007) 645–655.
[19] J.P. Coyette, K.R. Fyfe, Solution of elasto-acoustic problems using a variational finite element/boundary element technique, numerical techniques in
acoustic radiation, in: Proceedings of the Winter Annual Meeting of the ASME, 1989, pp. 15–25.
[20] LMS, LMS Virtual.Lab REV9 acoustic training, LMS International (2011).

You might also like