You are on page 1of 6

Available online at www.sciencedirect.

com

Procedia Engineering 48 (2012) 543 – 548

MMaMS 2012

Modal Assurance Criterion


Miroslav Pastora, Michal Bindaa*, Tomáš Harþarika
a
Technical University of Košice, Faculty of Mechanical Engineering, Letná 9, 042 00 Košice, Slovak Republic

Abstract

This article reviews the using of the original Modal Assurance Criterion (MAC). The Modal Assurance Criterion is a statistical indicator
that is most sensitive to large differences and relatively insensitive to small differences in the mode shapes. This yields a good statistic
indicator and a degree of consistency between mode shapes. The MAC considers only modal shapes which mean that a separate
frequency comparison must be used in conjunction with the MAC values to determine the correlated mode pairs. The MAC is often to
used to pair modes shapes derived from analytical models with those obtained experimentally. It is easy to apply and does not require an
estimate of the system matrices. It is bounded between 0 and 1, with 1 indicating fully consistent mode shapes. It can only indicate
consistency and does not indicate validity or orthogonality. A value near 0 indicates that the modes are not consistent.
© 2012 The Authors. Published by Elsevier Ltd.
© 2012 Published by Elsevier Ltd.Selection and/or peer-review under responsibility of the Branch Office of Slovak Metallurgical Society at
Selection
Faculty of and/or peer-review
Metallurgy under
and Faculty responsibility
of Mechanical of the Branch
Engineering, Office
Technical of Slovak
University Metallurgical Society at Faculty of Metallurgy and
of Košice
Faculty of Mechanical Engineering, Technical University of Košice.

Keywords: Modal Testing, Modal Assurance Criterion, Experimental Modal Anlaysis

Nomenclature

{ϕ X }q test modal vector, mode q


{ϕ A }r compatible analytical modal vector, mode r

{ϕ X }q {ϕ X }q
T
transpose of

{ϕ A }r {ϕ A }r
T
transpose of

{ψ X }q eigenvector defined the complex damped mode of vibration

{ψ A }r eigenvector defined the complex damped mode of vibration

{ψ X }q {ψ X }q
*
complex conjugate of

{ψ X }q {ψ X }q
T
transpose of

{ψ A }r {ψ A }r
*
complex conjugate of

* Corresponding author. Tel.: +421 902 432 148.


E-mail address: michal.binda@gmail.com.

1877-7058 © 2012 Published by Elsevier Ltd.Selection and/or peer-review under responsibility of the Branch Office of Slovak Metallurgical Society at
Faculty of Metallurgy and Faculty of Mechanical Engineering, Technical University of Košice
doi:10.1016/j.proeng.2012.09.551
544 Miroslav Pastor et al. / Procedia Engineering 48 (2012) 543 – 548

1. Comparison of modal properties

Modal properties that are compared usually include natural frequencies, real mode shape vectors, modal masses, modal
kinetic and strain energies. For system with complex modes of vibration one can add modal damping ratios and complex
mode shapes. Comparison of modal vectors can be done at the reduced order or at the full order of the FEM. Reduction of
the physical mass matrix or expansion of test modal vectors bring inherent approximations in the comparison criteria. A
test-analysis comparison is meaningful only for matched modes. These are estimates of the same physical mode shape and
their entries correspond one-for-one with their counterparts. Mode matching is an essential step before any comparison can
be undertaken [1], [10].
It is useful to compare:
• Measured mode shapes against vectors determined by analytical model
• Estimates of the same test modal vector obtained from different excitation locations
• Estimates of the same modal vector obtained from different modal parameter identification processes using the same test
data
• One test mode shape before and after a change in the physical structure caused by a wanted modification [8], [10]

2. Modal assurance criterion

One of the most popular tools for the quantitative comparison of modal vectors is the Modal Assurance Criterion (MAC).
The development of the MAC was modeled after the development of the ordinary coherence calculation associated with
computation of the frequency response function. The MAC is a statistical indicator, just like ordinary coherence. This least
squares based form of linear regression analysis yields an indicator that is most sensitive to the largest difference between
comparative values and results in modal assurance criterion that is insensitive to small changes or small magnitudes.
The MAC was originally introduced in modal testing in connection with The Modal Scale Factor, as an additional
confidence factor in the evaluation of modal vector from different excitation locations. When an FRF matrix is expressed in
the partial fraction expansion form, the numerator of each term represents the matrix of residues or modal constants [5], [6].

The MAC is calculated as the normalized scalar product of the two sets of vectors {ϕ A } and {ϕ X } . The resulting
scalars are arranged into the MAC matrix

2
{ϕ A }r {ϕ X }q
T

MAC ( r , q ) =
({ϕ } {ϕ } ) ({ϕ }q {ϕ X }q )
, (1)
T T
A r A r X

where the form of coherence function can be recognized, indicating the casual relationship between {ϕ A } and {ϕ X } .
In this case the modulus in the numerator is taken after the vector multiplication, so that the absolute value of the sum of
product elements is squared. And equivalent formulation is

¦ {ϕ } {ϕ }
n
j =1 A X
MAC ( A, X ) =
j j

( ¦ {ϕ } ) ( ¦ {ϕ } )
. (2)
n 2 n 2
j =1 A j j =1 X j

The MAC has been used as a Mode Shape Correlation Constant to quantify the accuracy of identified mode shapes. For
complex modes of vibration

* 2
{ψ A }r {ψ X }q
T

MAC ( r , q ) =
({ψ A }r {ψ A }r ) ({ψ X }q {ψ X }q )
. (2)
T * T *

and is clearly a real quantity, even if the mode shape data are complex.
Miroslav Pastor et al. / Procedia Engineering 48 (2012) 543 – 548 545

The MAC takes value between 0 (representing no consistent correspondence) and 1 (representing a consistent
correspondence). Values larger than 0.9 indicate consistent correspondence whereas small values indicate poor resemblance
of the two shapes.
The MAC can take on value near zero for the following reasons:
• The system is non-stationary resulting from changes in the mass, stiffness and damping properties during testing
• The system is non-linear
• There is noise on the reference mode shape
• The parameter extraction technique is invalid for the measured data set
• The mode shapes are linearly independent. While the MAC is not a true orthogonality check since the mass or stiffness
matrices have not been included in the calculation, it can be used as an approximation of an orthogonality check
Obviously, if the first four reasons can be eliminated, the MAC can be interpreted in a similar way as an orthogonality
calculation.
The MAC can take on a value near unity for the following reasons:
• The number of response degrees of freedom is insufficient to distinguish between independent mode shapes
• The mode shapes are a result of unmeasured forces to the system
• The mode shapes are primarily coherent noise
• The mode shapes represent the same motion different only by a scalar
Therefore, if the first three reasons can be eliminated, The MAC can be used to indicate consistence shapes thereby
indicating the validity of the modal scale factor.
The ideal MAC matrix cannot be a unit matrix because the modal vectors are not directly orthogonal but mass
orthogonal. However, the MAC matrix indicates which individual modes from the two sets relate to the each other. If two
vectors are switched in one set, then the largest entries of the MAC matrix are no more on the leading diagonal and it
resembles a permutation matrix.
The MAC can only indicate consistency, not validity, so it is mainly used in pre-test mode pairing. The MAC is
incapable of distinguishing between systematic errors and local discrepancies. It cannot identify whether the vectors are
orthogonal or incomplete [5], [6].

2.1. MAC presentation formats

One of the big changes in the application of the MAC over the last years in the way the information is presented. Today
most computer systems routinely utilize color to present magnitude data like MAC using 2D and 3D plot, see Fig. 1. It is
important to remember that MAC is a discrete calculation and what appears as a color contour plot really only represents the
discrete mode to mode comparison. Nevertheless, a color plot does allow for more data to be presented in an understandable
form in minimum space [2].

Fig. 1 2D MAC plot example


546 Miroslav Pastor et al. / Procedia Engineering 48 (2012) 543 – 548

3. Experimental procedure and results

As a test case for the proposed procedure, the modal analysis of a steel beam with free edges has been performed, see
Fig. 2. The steel beam was excited with modal hammer type 8206 with plastic tip by Bruel&Kjaer at appropriate points.
An analysis was performed using analyzer Bruel&Kjaer Pulse 3560C and laptop PC running PULSE software. The
response to excitation of structure was measured in two steps. First, system response was measured by accelerometer
Bruel&Kjaer 4506 B. The second step, the system response was determined using by optical vibrometer PDV100.

Fig. 2 measurement chain

Natural frequencies of the steel beam were determined by using PULSE Reflex software. In the Fig. 2 there are
frequency response functions – inertances. Natural frequencies were estimated using software algorithms using method
Polyreference Time, see Table 1.

>P1Vð@ >P1Vð@





 

 

 

 


     N      N
(a) >+]@ (b) >+]@

Fig. 3. (a) the maximum frequency response function - maximum inertance measured by Bruel&Kjaer 4506 B (b) the maximum frequency response
function - maximum inertance measured by PDV100.
Miroslav Pastor et al. / Procedia Engineering 48 (2012) 543 – 548 547

Table 1. Natural frequencies

Number of Mode Frequency - Bruel&Kjaer Frequency – PDV100 (Hz)


4506 B (Hz)
1 27.242 28.802
2 76.130 76.135
3 138.403 138.399
4 225.096 225.095
5 335.860 335.867

3.1. AutoMAC criterion

AutoMAC quantitatively compares all the possible combinations of test and analysis mode shape pairs for only one set of
mode shapes. The calculation assigns a value of 1.0 to test and analysis mode shape pairs that exactly match. A value of 0 is
given to those pairs that are completely independent or unrelated, and can indicate a problem. There could be spatial-
aliasing, which indicates that not enough response points were measured to uniquely identify the individual modes. Also,
one of the modes could be computational and could, therefore, look very much like a structural mode. A matrix presents the
results of this MAC calculation showing which test mode shapes match which analysis mode shapes and the degree of
correlation. Values between 0 and 1.0 are proportional to the degree of correlation between the mode shapes.
An example of an original AutoMAC plot for a simple steel beam is shown in Table 2. The use of the different color-
filled to indicate the MAC values are shown to be quite useful.

Table 2. AutoMAC matrix

27.242 76.130 138.403 225.096 335.860


27.242 1.000 0.439 0.003 0.031 0.001
76.130 0.439 1.000 0.046 0.031 0.009
138.403 0.003 0.046 1.000 0.016 0.025
225.096 0.031 0.031 0.016 1.000 0.011
335.860 0.001 0.009 0.025 0.011 1.000

3.2. CrossMAC criterion

CrossMAC is similar to AutoMAC, however, it is determined by using two different sets of modes for the rows and
columns of the MAC matrix.
CrossMAC can be used to:
• Calculate the MAC between two different test models
• Calculate the MAC between a test model and a Finite Element Model (FEM)
• Calculate the MAC between two analysis models developed using two different modal analysis algorithms
Values close to one mean that the two mode shapes are highly correlated and almost identical.
Table. 3 shows the comparison of natural frequencies obtained by using of two different accelerometers (Bruel&Kjaer
4506 B vs. PDV100) via CrossMAC.

Table 3. CrossMAC matrix

27.242 76.130 138.403 225.096 335.860


28.802 0.368 0.217 0.023 0.076 0.335
76.135 0.270 0.467 0.048 0.071 0.411
138.399 0.178 0.217 0.453 0.068 0.243
225.095 0.201 0.096 0.028 0.666 0.309
335.867 0.127 0.165 0.004 0.080 0.936
548 Miroslav Pastor et al. / Procedia Engineering 48 (2012) 543 – 548

4. Conclusion

In this paper, we have reviewed a method for the comparison and quantitative correlation between the dynamic properties
measured in a modal test. It is possible to determine the correlated modal pair.
During last two decades, the Modal Assurance Criterion has been demonstrated as a simple statistical concept, which can
become an extremely useful tool in experimental modal analysis. The use of the Modal Assurance Criterion and the
development and use of a significant number of related criteria has been remarkable and is most likely due to the overall
simplicity of the concept. New uses of the MAC will be developed over the next years as users more fully understand the
limitations of the current criteria. In the next few years, the increased uses of other statistical methods as well as further
development of singular value/vector methods are related areas. Since the MAC gives no information about the frequency
correspondence and can sometimes show good correlation between modes that have significant frequency separation, a
natural frequency comparison plot is also required.

Acknowledgements

The paper has been created realization of project “Centrum výskumu riadenia technických enviromentálnych
a humánnych rizík pre trvalý rozvoj produkcie a výrobkov v strojárstve” (IMTS:26220120060), based on operating program
support Research and Development financed from European Regional Development Fund and by Scientific Grant Agency
VEGA MŠ SR for the support of this work under Projects No. 1/0289/11 and No.1/0265/10.

References

[1] Rades, M., 2010. Mechanical Vibrations II – Structural Dynamic Modelling. Editura PRINTECH.
[2] Randall, J., A. 2003. The Modal Assurance Criterion – Twenty Years of Use and Abuse, Sound and Vibration, August 2003.,pp. 14-20.
[3] Ewins, D., J. 2000. Model Validation: Correlation for updating. Sadhana, Vol. 25, part 3, June 2000, pp 221-234.
[4] Ewins, D., J. 2000. Modal Testing: Theory, Practice and Application, Research Studies Press Ltd., USA.
[5] Fotsch, D., Ewins, D., J. Application of MAC in the Frequency Domain, Mechanical Engineering Department, Imperial College of
Science,Technology and Medicine, London, UK .
[6] Avilés, R. 2009, The Modal Assurance Criterion (MAC). Bilbao.
[7] TrebuĖa, F., Šimþák, F. 2007. Príruþka experimentálnej mechaniky. Typopress, Košice.
[8] Bilošová, A. 2001. Experimentální modální analýza. Spectris Praha spol. s.r.o., Praha.
[9] Kožešník, J. 1979. Kmitání mechanických soustav. Academica.
[10] Miláþek, S. 1992. Modální analýza mechanických kmitú. ýVUT, Praha.

You might also like