Basics of Experimental Modal Analysis: C.Sujatha

BASICS OF EXPERIMENTAL
Indian Institute of Technology Madras
MODAL ANALYSIS
Chennai – 600 036, India
C.Sujatha
Department of Mechanical Engineering, IIT Madras,

Chennai, India
INTRODUCTION
!   Experimental modal analysis deals with

determination of modal parameters
! N
  atural frequencies, damping properties and
mode shapes
!  Through experiments
! W
  ith signal processing software & hardware,
it is a common tool readily available in
vibration toolboxes
April 6, 2013 C.Sujatha 2

INTRODUCTION (Contd..)
!  Modal parameters are to be estimated from the

measured frequency response functions (FRFs)

!  From the output response and input force

data
!  Experimental modal model is only as good as

the quality of the FRFs

Important Experimental Aspects
§ Boundary conditions of the test structure
§ Choice of exciters/shakers : mechanism

§ Choice of excitation location

§ Minimisation of exciter/test structure interaction
§ Excitation signal, frequency range and amplitude of

the excitation force
§ Selection of appropriate transducers and their

positioning and mounting

Important Experimental Aspects (Contd.)
§ Aspects related to signal processing

§ Problems in the measurement of excitation force

§ Difficulties encountered in impact testing
§ Entire process of obtaining the required data is

experimental in nature
§ Mathematical techniques are involved in the

computation of FRFs and modal parameter
extraction

Support Conditions of Test Structure
! The fixture location & point of application of

excitation force (hammer/shaker) important
! Fixture can distort the test input so badly that

the specimen is overtested at some frequencies

and undertested at others
! Find out fixity conditions in which the structure

is used under actual operating conditions
! Choose the appropriate fixtures to obtain the

boundary conditions

Support Conditions of Test Structure (Contd.)
! Wrong fixity affects subsequent analyses such as

structural modification and finite element
correlation.
(i) Free Condition

!  Example: The airplane or rocket in flight.
!  Analytically achieved very easily by allowing all

degrees of freedom
!  Physically, achieving a truly free support is not

realizable, since the structure must be held in
some manner

Support Conditions of Test Structure (Contd.)
!  In free condition, the structure exhibits rigid body
behaviour ideally at low frequencies close to zero
frequency and does not bend/flex.
!  In practice however, the rigid body frequencies (3

translational and 3 rotational) will not be zero.
!  T e s t i n g w i t h t h i s B C t h e r e f o r e a l l o w s
determination of the rigid body modes and hence
the mass and inertia properties of the structure.
!  Free condition achieved by suspending the
structure using very soft, highly elastic bands or
suspension cord.
!  Alternately the test structure is placed on very
soft sponge or cushion or springs.

(i) Free Condition
!  Due to such an arrangement, the structure will be

constrained to some extent and the rigid body
modes will therefore have low frequencies.
!  As a guideline in testing with free supports, the

frequency of the highest rigid body mode should be
less than one tenth that of the first flexible mode.

!  Achieved using a sufficiently resilient support.

(ii) Fixed/Grounded Condition
!  Fixed boundary condition implies that all motion:

displacements as well as rotations, at the clamped
edge/support, are forced to zero
!  Analytically achieved by arresting all DOF at the

support
!  This BC much more difficult to achieve than a free

BC since most structures have some amount of
flexibility at the clamps
!  The base of the structure is attached and will have

some motion of its own due to flexibility of the
bolted, riveted or welded connections.

Support Conditions of Test Structure
!  FRFs may be measured at the base over the
frequency range of interest and compared with
the corresponding response of the structure.
!  If the former is negligibly small, the structural

data may be taken to be accurate.

(iii) Actual Fixity Conditions At Site
!  From a practical point of view, the actual boundary
conditions chosen for a test may depend on the
weight and size of the structure.
!  For instance, it is not possible to support a large

structure weighing a few tonnes in a free test state.
!  Also it may not be convenient to fix certain

structures in the clamped condition.
!  There are also situations requiring both test

conditions as in the case of a space vehicle.
!  It is desirable to test this structure in the free

condition to simulate its operating environment in
space.
(iii) Actual Fixity Conditions At Site (Contd.)
!  However, it is also required to do the test in the

grounded condition to study dynamic behaviour in
launch environment.
!  Sometimes the boundary conditions can lead to

uncertainties and non-linear behaviour of the test

structures and therefore fixing conditions should
be carefully considered.
!  Besides, it is to be remembered that the clamping

mechanism also adds damping to the structure due
to movement in the fixture joints.

Choice of Exciters/Shakers
!  Large variety of excitation techniques for

obtaining structural response in vibration/modal
testing.
!  The choice of a particular excitation technique

depends on:
!   The size and fixity condition of the structure

!   The excitation signal to be imparted
!   The required frequency range
!   The sensing mechanism and
!   The data analysis procedure.
Choice of Exciters/Shakers(Contd.)
!   Basically excitation techniques can be divided into

two categories, contacting (intrusive) and non
contacting (non-intrusive).
!   Contacting (intrusive) excitation involves

connection of the exciter to the test structure
throughout the test.
!   In non-contacting (non-intrusive) excitation, the

exciter does not touch the test structure

The range of exciters available:

!   Mechanical shakers of the displacement type

!   Electrodynamics exciters
!   Slip tables
!   Modal exciters
!   Unbalance mass exciters
!   Inertial exciters
! Electrohydraulic shakers

!   Electromagnetic shakers
!   Impact hammers
! Piezoactuators
!   The measurement of forces transmitted to

the structures by these shakers also
requires careful consideration

Exciters
!  Electrodynamic exciters are usually used to directly
transmit force to the structures through pushrods/
stingers
!  Bulk-fixing of items to these shakers is also possible

as with electrohydraulic or mechanical exciters .
!  This type of testing is more applicable to fatigue life

and resonance behaviour studies of the system fixed
to the base of the exciter.

Exciters/Shakers
!   In the case of unbalance
mass exciters, the force
produced is meω2.
!   Since these exciters are

capable of producing large

forces, they are used to
test structures like
bridges, masts, towers,
dams, foundations, etc.
!   The shaker itself offers

negligible mass loading in
these cases.

Impact Excitation
!  In the non-contacting excitation technique, the
exciter is not in contact/is in contact for a very
short duration with the test structure , e.g. impact
hammer
!  Impact hammer used for imparting forces to light

weight structures
!  For modal analysis of machines at site
!  For modal analysis of rotating structures

Impact Excitation (Contd.)
!   Disadvantage: multiple impacts and poor

repeatability of impact conditions, namely the
force amplitude and direction.

!   Does not possess the ability to apply a force

input with a controlled time history to a point on
the structure, which is a drawback.

Impact Testing and Difficulties
!  Major problem with impact testing is generation of
multiple impacts while hitting. This can be eliminated
by the use of a force window or rectangular window.
!  Multiple impacts should be avoided.

!   In practice it is usually easier to simply discard a

record with a double hit from the ensemble of
records and this is done by most modal software
packages.
!  If impact forces are high, they can overdrive the

system and aggravate its non-linear response.

Impact Testing and Difficulties(Contd.)
!  Every effort should be made to linearise the
system when making FRF measurements using
impact testing
!  Static preloads applied with soft springs can

tighten clearances and constrain the system into a

comparatively linear state.
!  The low stiffness of preload in series with the

stiffness of the structure causes a negligible
effect on the system dynamics.
!  Impacting at locations which are not accessible

with a conventional impact hammer poses a
problem.
Impact Testing and Difficulties(Contd.)
!  A punch impactor (electrical hammer) may be

used to precisely locate impacts, to minimize the

variance of the impact location and to impact in a
skewed direction (non-orthogonal to the

coordinates).
!  The punch impactor should be fitted with a hard

tip, typically a metal or plastic tip.

Other Excitation Techniques
!   Electromagnetic shakers used for non-contact
excitation of rotating/lightweight structures
!   They do not have the problems associated with
impact testing using a hammer.
!   They can be used to transmit forces to rotating

disk Piezoactuators can be used to excite
lightweight structures
!   S, say, for vibration control or for estimation of
the stiffness and damping of bearings.
!   These exciters are also useful for evaluating the
dynamic response of impellers, circular saws, etc.

Inertial Shakers
!  An interesting development in the field of exciters
is that of inertial shakers (reaction mass shakers)
!  Inertial shakers are useful for obtaining the
response of a structure in field conditions within

the shortest test time possible, without the use of
any additional test fixtures.

!  They can very conveniently be used for modal
testing of a ship hull or automobile body which
requires fixing the exciter with arbitrary
orientations.

Operational Excitation/Ambient
Testing
!  For very large structures, such as dams, large
bridges and buildings, ambient testing is used.
!  Such testing done in the presence of the inherent

operating forcing functions.
!  Applied force functions are often immeasurable

and of distributed nature.
!  Only response and not FRF data can be got;

reliable forced vibration studies not possible with
data of this type.

Operational Excitation/Ambient (Contd.)
Testing
!   Operational excitation provides a viable

technique for lightweight structures also.
!   Since no external connections are required,

external loading effects are eliminated.
!   However, all the structural modes may not be

excited and hence may not be present in the
vibration data base.

Step Relaxation
!  Useful for dynamic testing of large structure
such as bridges and windmills
!  In step relaxation/free decay testing the system

is given an initial condition (displacement, velocity,
etc.) and the free decay response of the system is

measured.
!  An impact with sufficient energy is supplied to

excite the test structure uniformly.
!  Since most dynamic characteristics of interest in

large structures are of low frequency, the input
spectrum of the force has to be restricted.

Step Relaxation (Contd.)
!  The structure is preloaded at a point by a flexible

tendon. When either the tensile load in the tendon
or the displacement reaches a specified value, the

load is suddenly released.
!  This imparts an impulsive step function to the

structure; load cell used to measure the force time
history.
!  The load may also be provided by hanging a large

mass from the structure and then suddenly
releasing it.

Step Relaxation (Contd.)
!  The technique is easy to use and requires minimal

equipment.
!  The difficulties: tedious time consuming set up

procedures required to produce repeatable
impulses and the application of the tendon
fracture mechanism.
!  Technique has little direct control over the

excitation levels.

Minimizing Exciter/Test Structure Interaction
!   The shaker physically interacts with the structure

through the force transducer, altering the system
dynamics.
!   In light weight structures, the load cell adapter

leads to considerable mass loading, causing the
measured force to be greater than actually applied.
!   The shaker should transmit forces in line with the

main axis of the load cell; however, it is possible
that there is rotation about the other two axes.

Minimizing Exciter/Test Structure Interaction (Contd.)
!  To minimize this, the exciter is connected to the

load cell through a slender rod (stinger), which
allows the load to move freely in other

directions.
!   This stinger is essentially a rod with a strong

axial stiffness like a truss, but weak bending and
shear stiffness.

Sensing Techniques
!   Displacement, velocity or acceleration

transducers used, depending on whether
receptance, mobility or accelerance FRFs
required.
!   Mass loading due to the sensor is a severe

problem for highly flexible structures, especially
if the sensor mass is comparable to any modal
mass; dynamics turns out to be different from
what would be produced in the absence of
loading.
!   Mass loading can be reduced by choosing light
weight transducers or even better, non-intrusive
sensors
Sensing Techniques (Contd.)
!   Many transducers available which are suitable

for non-contact vibration response
measurement; many have the desirable small

target footprints.
!   Often miniature accelerometers and strain

gauges used for sensing; they are not truly non-
intrusive, but they hardly alter the system
dynamics.

Mass loading
!   For example, consider the read /write head and

assembly on a typical computer hard disk, the
dynamic response of which is critical to the
performance of the system.

!   The mass of the entire read /write mechanism

quite small (approx. 30–40 g), the mounting of
either an excitation or response transducer can
alter the system dynamics.
!   The mass loading of even the lightest

accelerometer (less than a gram) can significantly
alter the system response.

Mass loading (Contd.)
!  Many optical non-contact response measurement

techniques available; they have very good
resolution, dynamic range and flexibility

necessary for modal analysis applications.
!  Examples are fibre-optic sensors, laser

holography, laser Doppler velocimetry (LDV),
laser triangulation techniques and line scan
cameras.

Sensing
!   There are other non-optical techniques using
proximity probes of both inductive and capacitive
type and acoustic sensors.
!   Considerations regarding the choice, location and

orientation of transducer and transducer

mounting are important.
!   Local conditions of a structure may drastically

vary, influencing the local stiffness at the
transducer location.
!   The transducer should not be mounted at or close

to a node of any structural mode of interest.

Sensing(Contd.)
!   For point FRF measurements, the force &

response should be measured at the same point;
done using an impedance head, housing.
!   It is important to note the direction of the force

and response.
!   Most present modal analysis techniques use

translational DOF.
!   Bi-and triaxial transducers are also available to

perform detailed modal analysis.

Vibration Transducers
!  Do I need to measure displacement, velocity or
acceleration?
!  Can I use contact type of transducers without
introducing mass loading?

!  Do I need any additional signal conditioning
devices to go with the transducers?

!  What are the frequency range, dynamic range
and sensitivity required?
!  Is the choice of the transducer dictated by any
environmental constraints like dust,
electromagnetic fields, humidity, shock levels,
etc.?
!  What is my budget?

Common Transducers
!   LVDT; typically for low frequencies for

receptance
! Electrodynamic pick up for mobility

measurements; mid-frequency range
!   Piezoelectric accelerometer for accelerance; for

high frequencies

Selection of Excitation Signal
!   The choice of excitation system & excitation

function can make the difference between a good
and bad measurement.
!   After choosing the shaker, the next step involves

selecting an excitation function such as sine

sweep, random, transient, etc.
!   The excitation signal generally depends on the
type of test to be conducted.
!   Though the level of non-linearities can be
measured and characterized effectively with sine
sweeps, a random function is most suitable to
estimate the best linear model of a non-linear
system
Selection of Excitation Signal(Contd.)
!  The amount of damping and the density of the

modes within the structure can also dictate the
use of specific excitation functions. The exciter

control electronics must therefore have a signal
generator and a gain control, dependent on the

feedback from the test object. Computer-aided
vibration control systems which are simple to
operate and interactive are commercially available
for shaker control. They utilize highly developed
and powerful digital signal processing hardware
and/or software.

The excitation signal can be one of the following:

i.  Stepped sinusoidal
ii.  Slow sinusoidal sweep
iii. Periodic
iv. True random
!   Sine on random
!   Random on random

iv.  Pseudo random

v.  Transient
!   Chirp random
!   Sine chirp
!   Burst chirp
!   External pulse
!   Shock

Stepped Sine Excitation
!  Signal given to the exciter is a discrete sinusoidal

voltage or current with a fixed amplitude and
frequency.
!  The command signal frequency is stepped from

one discrete value to another in the entire
frequency range such as to provide the necessary
frequency resolution in the FRF plot.
!  Excitation frequency should not be changed

abruptly; sufficient time should be given for the
system to settle down to steady state conditions.

Stepped Sine Excitation (Contd.)
!  Especially near resonance, or if one is

measuring lightly damped or closely spaced
modes, care should be taken to avoid
transient effects.
!  Major advantage with stepped sine excitation:

gives flexibility in choosing the frequency
resolution.
!  This method useful for large, complex

structures and for cases with high
background noise.

!   Near resonance, tests can be done at a larger
number of frequencies and away from resonance,
at a few frequencies only.
!   Important to have closely spaced points around

resonances if one is interested in measuring
modal damping.
!   Since excitation is always imparted at a single
well-known frequency, it is possible to use simple
measuring techniques for obtaining FRFs instead
of FFT analysers.
!   Though sine testing gives the best signal-to-
noise ratio (SNR), it is extremely slow for
broadband measurements.

!   Stepped sine excitation is ideally suited for the

evaluation of non-linearities since it is very easy
to control amplitude, frequency and phase of the
excitations.
!   The presence of non-linearities in the system may

be found out by analysing the dependence of the
response level on the excitation level and the
manifestation of higher order harmonic
components.
!   For large structures requiring a multiple
excitation/response measurement, controller
required; this defines a phase/amplitude
relationship between the different shakers.

Slow Sine Sweep
!  Involves the use of a sweep oscillator to provide a

sinusoidal command signal, the frequency of which
is varied continuously and slowly throughout the
frequency range of interest.

!  Swept sine capability allows conventional sine

testing of non-linear structures where high
energy at a sing1e frequency is necessary.
!  As in the case of stepped sine excitation, it is

imperative in this case also to ensure that the
frequency sweep is done sufficiently slowly to
enable the system to settle down to steady-state
conditions before measurements are made.
Slow Sine Sweep(Contd.)
!  If the sweep rate is too high, then the FRF gets
distorted with the true resonance peak getting
shifted to the right (higher frequency) with
increasing sweep rate.


!  A good way to check whether a sweep rate is
alright, is to measure twice, once sweeping
upwards, and the second time downwards through
the frequency range.

!  Sweep rate is correct, if both the sweeps give

rise to the same FRF.
!  It is possible to prescribe an optimum sweep rate

for a given structure, taking into account its
damping levels.

!  A good practical approach to guarantee that we

can very closely approach the desired condition
is by using a logarithmic sweep rate.

!  The ISO standards prescribe the maximum

allowable linear and logarithmic sweep rates
through a resonance.

Slow Sine Sweep: Sweep Rate: ISO Standards
fr
ςr
ηr Linear sweep
Logarithmic sweep
Smax denotes the maximum allowable sweep rate, fr

the resonant frequency of the mode under study, ζr
and ηr the viscous damping ratio and loss factor.

True Random Excitation
!  True random signal is the most popular signal for

shaker excitation since measurement time is
minimal.
!  This signal may be obtained from signal

generators or random noise generators, which
are commercially available.
!  Its peak to RMS ratio (crest factor) and RMS

value can be controlled.

True Random Excitation (Contd.)
!I
  t also has a reasonably good
SNR.
!  Band-limited random noise

allows exciter energy to be
concentrated in a small
frequency range for
improved SNR.

!  A random function is most suitable for obtaining

the best linear model of a non-linear system.
!  Typically in testing, the true random signal is

sent out to shaker by the controller and the
structure's response is measured synchronously.
!  During the next time period , a different random

excitation record (with the same RMS level and
crest factor however) is sent out.

!  Since the excitation level at a given frequency is

different during each period , the test engineer
gets the best linear approximation of a non-linear

system by using an averaging process.
!  Averaging increases the total analysis time.
!  Uniform white noise: Ideal white noise has equal

power per unit bandwidth, resulting in a flat power
spectral density across the frequency range of
interest.

!   Pink noise: also called weighted white noise, contains
equal energy per measurement band and thus has a
1/1 or 1/3 octave band energy level which is
constant. Thus the lower frequencies have a higher
energy level.
!   Uniform white noise: the probability density function
(PDF) curve is flat

!   Gaussian white noise: bell shaped PDF
2
µ=0, ! =0.5
0.8 2
µ=0, ! =1.0
2
µ=0, ! =2.0
2
0.6 µ=1, ! =0.5
p (x) 0.4
0.2
0.0
-6 -4 -2 0 2 4 6
x

Pseudo-Random Excitation
!   Pseudo- or periodic-random noise (PRN) is a

superposition of sinusoidal signals with equal
energy at all frequencies in the bandwidth
considered and with random phases.

!   True random signal has the disadvantage that

there is leakage in the analysis window due to
the non-periodic nature of signal in this window.
!   This problem is overcome by using pseudo-

random excitation which is exactly periodic in
the analyser window considered.

Pseudo-Random Excitation (Contd.)
!   PRN is thus self-windowing and there is no need

to do windowing before spectral analysis.
!   PRN does not have a continuous PSD curve, as
does white noise, but has energy only at discrete

frequencies .
!   With PRN the FRF of a linear system may be

computed with just one time record, instead of
averaging a number of FRFs over several time
records.

Transient Excitation
!  Transient excitation poses the constraint that
the response signal should have died by the end
of the sampling window.
!   This type of excitation is essentially one of short

duration and requires that lightly damped
structures have a rather long sampling window,

limiting the frequency range that can be studied.
!  With transient excitation, as in the case of
random excitation, it is the practice to make a
large number of repeated measurements under
nominally identical conditions
!  Then averaging is done, resulting in averaged FRF
estimates.

Transient Excitation(Contd.)
!   Most analysers provide the user with a transient

excitation signal called the chirp signal
!   Chirp signals can be fed to an attached shaker.
!   Sine chirp signal has the advantage that it

offers greater controllability in terms of the

amplitude and frequency content of the input
and allows larger vibration energy
!   Burst noise excitation provides decreased test
time .

Sine on Random Control Profile Window
!  Pure broadband random has small energy at each

frequency
!  Besides in reality pure sinusoidal vibration is rare;

generally superimposed over broadband random
noise
!  SOR produces better simulation of the real world
!  (SOR) has high quality random signal with tones

are added
!  Many controllers allow as many as 10 independent

tones on a broadband random
Sine on Random Control Profile Window(Contd.)
!  Independent sweep rate and direction for each

tone
!  What makes SOR unique is its ability to add or
subtract the tones without changing the random

energy

Sine on Random(Contd.)
!   E.g.1. Sine on random profile for a turboprop

aircraft: Random signal encompassing the entire
frequency range with sinusoidal "spikes"

representing the rotating propellers, i.e. at the
propeller blade passage frequency and the first

few harmonics.
!   E.g.2. Sine on random profile for a helicopter so

as to have a single vibration test and avoid two
separate tests.


Sine on Random(Contd.)
1.  Random Profile: PSD level of 0.001 g2/Hz

from 10Hz to 50Hz, negligible level at all

other frequencies.
2.  Sinusoidal vibration: 15 Hz (+/- 1Hz) with peak

acceleration of 1.0g to incorporate the purely
sinusoidal excitation of the main rotor.

Obtaining FRFS
!   The measurement of an FRF involves
measurement of the input force and output
response as a function of frequency with all
the phase information intact.

!  .
!   Modal testing may involve single or multipoint

excitation
!   The input force may be imparted through a

shaker or an impact hammer.
!   FRF measurements have to be made at a

sufficiently large number of DOFs to get
accurate mode shapes.

Obtaining FRFS(Contd.)
!   All modal parameters are extracted from these

FRFs and are therefore only as good as the
FRFs, even if very elegant parameter

estimation techniques are used.
!   Modal parameters are subsequently extracted

from the FRFs using any of the curve-fitting
techniques available.
!   It has to be ensured that driving-point FRFs
(with the excitation and response
measurements made at the same DOF) are
included, in order to calibrate (scale) the modal
model.

Roving Hammer/ Impact Excitation
Excitation is imparted using an impact hammer; hammer is
moved around the structure to excite every DOF
Accelerometer response DOF is fixed and serves as a
reference DOF.

Roving Response Method:
!   With shaker excitation, the excitation DOF is
usually fixed, serving as the reference DOF.
!   We have roving response measurements with
response transducer being moved around the

structure.
!   This will cause a varying mass loading of the

structure and corrective action for this must be
taken while extracting modal parameters.
!   If instead of one roving accelerometer, there are
many accelerometers simultaneously fixed at all
the DOFs, the mass loading at each DOF will be
the same in all the FRFs, providing better
consistency of results & minimizing testing time.

April 6, 2013
C.Sujatha
Roving Response Method (Contd.)
72
Single input Single Output testing
Xj
With random input, the equation relating the input
force and the output response of a linear system is
given by
is the FRF relating the two
with the inputs forced to zero at all DOFs other

than j
:Sxx(ω) and Syy(ω) are i/p and o/p PSDs

Indian Institute of Technology Madras Multi-input multi-output testing(Contd.)
For uncorrelated inputs, the input output expression

is
Sxx(ω) is the matrix of the auto- and cross-spectral
densities of and between the different input DOFs
Sxy(ω) is the matrix of the cross-spectral densities
between inputs and outputs.
It is assumed that all the forces are non-zero and
that no pair of forces is fully correlated

Experimentally obtaining the matrix
(a)  single fixed response measurement, roving impact

excitation
(b)  single fixed shaker excitation, roving response
measurement.
Estimation of FRFs
!  Ideally the FRF should give a very accurate

representation of the actual structural response
!  But quite often the FRFs are distorted by noise in
the input/output/both
!  Hence lot of research has been done on obtaining
good FRFs.

April 6, 2013
C.Sujatha
Estimation of FRFs (Contd.)
77
Estimation of FRFs (Contd.)
H1 FRF is defined as

Indian Institute of Technology Madras Estimation of FRFs (Contd.)
H1 is seen to be sensitive to input noise;

Is not a good estimator around resonance

However it is very good at anti-resonance
H2 FRF is defined as
H2 is seen to be sensitive to output noise;

Is not a good estimator around anti-resonance
However it is very good at resonance

Coherence Function
Coherence is used to detect errors in estimating an
FRF and is defined as
The coherence function is always less than or equal to 1.0

Coherence Function(Contd.)
!   If the measurements are made accurately,
coherence must be very close to 1
!   The coherence function also gives an indication if

the measured response can be fully attributed to
the measured inputs.

!   A value of the coherence function less than 1
over a frequency range indicates that response
is only partially due to the measured inputs and
may be the result of unknown inputs.
!   If coherence is small, the FRF should be
discarded

!  Reasons for small coherence:
!  noise in either the excitation or the response

signals
!  One of the most likely sources of low

coherence in lightly damped structures is
poor frequency resolution in the analyser.
!  If the resolution is not fine enough to

describe adequately the rapid amplitude
changes around resonances and
antiresonances, the resulting FRF data are
poor.

!   Increasing the number of averages will not

increase coherence, but will only make it less
noisy.
!   The only way to improve the quality of

coherence functions is to take all possible

precautions to eliminate unknown sources.
!   This can be achieved by ensuring proper
boundary conditions, minimizing shaker–
structure interaction, etc.

Coherence Function (Contd.)
!   When the coherence is low in spite of all such

precautions, a zoom measurement may be made

to enhance the accuracy of the measurements
around resonance, but at the cost of increased

measurement time.
!   It is a good practice to make several

measurements and to accumulate a running
average of the corresponding FRF estimates and
the coherence.

Basics of Experimental Modal Analysis: C.Sujatha

Diunggah oleh

Informasi Dokumen

Deskripsi Asli:

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

Basics of Experimental Modal Analysis: C.Sujatha

Diunggah oleh

Hak Cipta:

Format Tersedia

BASICS OF EXPERIMENTAL

Indian Institute of Technology Madras

Department of Mechanical Engineering, IIT Madras,

! Experimental modal analysis deals with

April 6, 2013 C.Sujatha 2

! Modal parameters are to be estimated from the

measured frequency response functions (FRFs)

! From the output response and input force

! Experimental modal model is only as good as

April 6, 2013 C.Sujatha 3

§ Boundary conditions of the test structure

§ Choice of exciters/shakers : mechanism

§ Choice of excitation location

§ Minimisation of exciter/test structure interaction

§ Excitation signal, frequency range and amplitude of

§ Selection of appropriate transducers and their

April 6, 2013 C.Sujatha 4

§ Aspects related to signal processing

§ Problems in the measurement of excitation force

§ Difficulties encountered in impact testing

§ Entire process of obtaining the required data is

§ Mathematical techniques are involved in the

April 6, 2013 C.Sujatha 5

! The fixture location & point of application of

! Fixture can distort the test input so badly that

the specimen is overtested at some frequencies

! Find out fixity conditions in which the structure

! Choose the appropriate fixtures to obtain the

April 6, 2013 C.Sujatha 6

! Wrong fixity affects subsequent analyses such as

(i) Free Condition

! Example: The airplane or rocket in flight.

! Analytically achieved very easily by allowing all

! Physically, achieving a truly free support is not

April 6, 2013 C.Sujatha 7

! In practice however, the rigid body frequencies (3

April 6, 2013 C.Sujatha 8

! Due to such an arrangement, the structure will be

! As a guideline in testing with free supports, the

less than one tenth that of the first flexible mode.

April 6, 2013 C.Sujatha 9

! Fixed boundary condition implies that all motion:

! Analytically achieved by arresting all DOF at the

! This BC much more difficult to achieve than a free

! The base of the structure is attached and will have

April 6, 2013 C.Sujatha 10

! If the former is negligibly small, the structural

data may be taken to be accurate.

April 6, 2013 C.Sujatha 11

! For instance, it is not possible to support a large

! Also it may not be convenient to fix certain

! There are also situations requiring both test

! It is desirable to test this structure in the free

! However, it is also required to do the test in the

! Sometimes the boundary conditions can lead to

uncertainties and non-linear behaviour of the test

! Besides, it is to be remembered that the clamping

April 6, 2013 C.Sujatha 13

! Large variety of excitation techniques for

! The choice of a particular excitation technique

! The size and fixity condition of the structure

! Basically excitation techniques can be divided into

!   Experimental modal analysis deals with

!  Modal parameters are to be estimated from the

!  From the output response and input force

!  Experimental modal model is only as good as

§ Boundary conditions of the test structure

§ Choice of exciters/shakers : mechanism

§ Choice of excitation location

§ Minimisation of exciter/test structure interaction

§ Excitation signal, frequency range and amplitude of

§ Selection of appropriate transducers and their

§ Aspects related to signal processing

§ Problems in the measurement of excitation force

§ Difficulties encountered in impact testing

§ Entire process of obtaining the required data is

§ Mathematical techniques are involved in the

!  Example: The airplane or rocket in flight.

!  Analytically achieved very easily by allowing all

!  Physically, achieving a truly free support is not

!  In practice however, the rigid body frequencies (3

!  Due to such an arrangement, the structure will be

!  As a guideline in testing with free supports, the

!  Fixed boundary condition implies that all motion:

!  Analytically achieved by arresting all DOF at the

!  This BC much more difficult to achieve than a free

!  The base of the structure is attached and will have

!  If the former is negligibly small, the structural

!  For instance, it is not possible to support a large

!  Also it may not be convenient to fix certain

!  There are also situations requiring both test

!  It is desirable to test this structure in the free

!  However, it is also required to do the test in the

!  Sometimes the boundary conditions can lead to

!  Besides, it is to be remembered that the clamping

!  Large variety of excitation techniques for

!  The choice of a particular excitation technique

!   The size and fixity condition of the structure

!   Basically excitation techniques can be divided into

!   Contacting (intrusive) excitation involves

!   In non-contacting (non-intrusive) excitation, the

!   Mechanical shakers of the displacement type

!   The measurement of forces transmitted to

!  Bulk-fixing of items to these shakers is also possible

!  This type of testing is more applicable to fatigue life

!   Since these exciters are

!   The shaker itself offers

!  Impact hammer used for imparting forces to light

!  For modal analysis of machines at site

!  For modal analysis of rotating structures

!   Disadvantage: multiple impacts and poor

!   Does not possess the ability to apply a force

!  Multiple impacts should be avoided.

!   In practice it is usually easier to simply discard a

!  If impact forces are high, they can overdrive the

!  Static preloads applied with soft springs can

!  The low stiffness of preload in series with the

!  Impacting at locations which are not accessible

!  A punch impactor (electrical hammer) may be

!  The punch impactor should be fitted with a hard

!   They can be used to transmit forces to rotating

!  Such testing done in the presence of the inherent

!  Applied force functions are often immeasurable

!  Only response and not FRF data can be got;

!   Operational excitation provides a viable

!   Since no external connections are required,

!   However, all the structural modes may not be

!  In step relaxation/free decay testing the system

!  An impact with sufficient energy is supplied to

!  Since most dynamic characteristics of interest in

!  The structure is preloaded at a point by a flexible

!  This imparts an impulsive step function to the

!  The load may also be provided by hanging a large

!  The technique is easy to use and requires minimal

!  The difficulties: tedious time consuming set up

!  Technique has little direct control over the

!   The shaker physically interacts with the structure

!   In light weight structures, the load cell adapter

!   The shaker should transmit forces in line with the

!  To minimize this, the exciter is connected to the

!   This stinger is essentially a rod with a strong