EMA

Question 1

What is EMA?

Answer 1

Experimental Modal Analysis is a collection of methods, tools and theories used to identify modal properties of a structures experimental measurements.

Question 2

Why do we perform EMA?

Answer 2

To predict the response of an unknow structure, to validate models of structures, to check for structural damages and to update FEA models.

Question 3

What are the four stages of EMA?

Answer 3

1. Experimental Design
2. Digital Signal Processing
3. Frequency Domain Analysis
4. Identification of Modal Properties.

Question 4

What is required within the physical setup of an EMA?

Answer 4

• Choice of Sensors
• Choice of Excitation
• Choice of Mounting for the
  structure.
• Control Variables.
• and often other test specific requirements.

Question 5

What is a tranducer?

Answer 5

Any device that converts one form on energy into another (for EMA, typically displacement, velocity, force or acceleration signals into output voltage)

Question 6

What 6 ways can sensors be classified?

Answer 6

1. Contact or non-contact.
2. Bandwidth.
3. Low frequency performance.
4. Dynamic range.
5. Budget/Cost.
6. Environmental Factor.

Question 7

Describe some advantages and disadvantages of contact sensors?

Answer 7

• Simple and robust to operate.
• They measure a definite point in
  the structure.
• Can add mass and possible
  stiffness/damping to the structure.
• Cheaper to operate.

Question 8

Describe some advantages and disadvantages of non-contact sensors.

Answer 8

• More complex to setup.
• Can be used in extreme
  environments.
• No intrusion on modal properties
  of the structure.

Question 9

What does sensor bandwidth mean?

Answer 9

The bandwidth is the range of the frequencies the sensor can detect.

Question 10

At what frequency range do piezoelectric sensors best operate? What frequencies do the reading drift at?

Answer 10

Excellent at high frequency, drifts at low frequency.

Question 11

What’s the definition of dynamic range?

Answer 11

The ratio of largest to smallest amplitude.

Question 12

Describe piezoelectric force transducers, including advantages and disadvantages.

Answer 12

1. Exploit piezoelectric effect to convert force into charge / voltage.
2. Good high frequency performance, poor at low frequency.
3. Very stable and accurate.
4. Many forms available so very versatile.
5. Requires charge amplifier.
6. Modern ICP / IEPE sensors solve some issues with electromagnetic sensitivity.

Question 13

Describe a piezoelectric force accelerometer.

Answer 13

Similar to a piezoelectric force sensor but with seismic mass added.

Question 14

What determines the upper frequency limit of a piezoelectric force accelerometer?

Answer 14

The resonant peak of mass-spring system formed seismic mass and piezo ‘spring’.

Question 15

Describe the laser doppler vibrometer with some advantages.

Answer 15

• Uses the doppler effect to
  measure velocity.
• Very accurate, good dynamic
  range and bandwidth.

Question 16

How can excitation signals be categorised?

Answer 16

Either as broadband (periodic, random or transient) or harmonic signal.

Question 17

For a broadband signal, what should an ideal frequency response across the frequency range of interest look like?

Answer 17

Flat / constant.

Question 18

Explain what stepped sine testing is?

Answer 18

A modal testing technique that measures a system’s frequency response functions by incrementally increasing or decreasing the frequency of a sine wave.

Question 19

Give advantages and disadvantages of stepped sine testing.

Answer 19

• Vital if non linearity is believed to
  exist.
• Very accurate.
• Usually very slow.

Question 20

Describe what ‘twanging’ is?

Answer 20

Its displacing a structure and releasing it, viewing its free vibration. Gives information only on the 1st node, without giving a force measurement.

Question 21

Describe impact testing?

Answer 21

This is where you hit the structure with a ‘hammer’ to deliver an impulse signal. It is important to have clean hits with no ongoing contact. Force input is measured by hammer tip transducer.

Question 22

Describe shaker testing.

Answer 22

This is where your structure is excited by a continuous shaker, able to deliver stepped sine, random noise or other types of input. The force leaves the shaker, travels through a ‘stringer’ which has a force transducer attached, vibrating the structure.

Question 23

How can you control the signal bandwidth for an electrodynamic shaker?

Answer 23

The bandwidth is increased through the use of a harder tip.

Question 24

What effect limit the use of accelerometers at low and high frequencies?

Answer 24

Accelerometer have poor performance at low frequency, because the charge generated by the piezoelectric material gradually leaks to earth over time.

The upper limit of frequency is usually encountered due to the fact that the mass and piezoelectric layer and spring form a mass spring system, which forms a large unwanted peak in the otherwise flat frequency response.

Question 25

Describe 3 other considerations when doing EMA test setup.

Answer 25

1.Mounting arrangement. 2. Sensors/ shaker placement. 3. EM interference.

Question 26

Explain why considerations for structure mounting is important in EMA.

Answer 26

The sensors used should have minimal effect on the structures vibration response. Moreover, the boundary conditions of your structure should be maintained after mounts are added. Any difference in response due to the mounting configuration leads to inaccurate results.

Question 27

If the true mounting conditions for a structure are not possible to replicate, what test configuration is usually adopted for the test and what are the key requirements of the supports in this case?

Answer 27

Aim for free boundary conditions. This is not generally achievable, but is approximated by isolating the structure using soft mount. The risk is that the degree of freedom introduced by the mounts introduce new modes that bias the test. The mounts should be sufficiently flexible so that the natural frequencies of the 'rigid body modes' of the structure under test is much lower that the flexible modes, say by a factor of 10.

Question 28

Explain why considerations for sensor / shaker placement is important in EMA.

Answer 28

Shaker should aim to excite all modes equally, they should avoid nodes and often antinodes.

Question 29

Why should EM interference be considered within EMA setup?

Answer 29

Sensors and cable can act as aerials, particularly if forming ground loops through metallics structures. Never trust a peak at 50 Hz since this is the frequency of mains electricity.

Question 30

Describe basically, what sampling is?

Answer 30

For a time signal y(t), reading its value at regular time intervals gives a series of reading (y0, y1, y2, .... etc). The time between signals is the sample period 'Ts' .

Question 31

How do you calculate sample rate / sample frequency (fs) from sample period (Ts)?

Answer 31

fs = 1/Ts

Question 32

What's the equation for the total duration of the sampled signal (T), when you know the number of sample N, and the sample period Ts.

Answer 32

T =(N+1) * Ts

Question 33

What is Nyquist Frequency?

Answer 33

The minimum rate at which a signal can be sampled without introducing errors, = fs/2.

Question 34

Describe aliasing?

Answer 34

Any frequency above the Nyquist frequency cannot be represented properly and instead appears at lower frequency levels, distorting the signal. This is aliasing.

Question 35

Describe the purpose of anti-aliasing?

Answer 35

Any frequency above the Nyquist Frequency which will distort the signal must be removed with low band pass before passing before sampling. This is anti-aliasing.

Question 36

What does DFT and FFT mean? (the can be used interchangeably)

Answer 36

Discrete Fourier Transform and Fast Fourier Transform.

Question 37

Given a signal with period T, what is the difference in frequency between one value and the next (Δf = ...). What is Δf known as too?

Answer 37

Δf = 1 / T and is known as the frequency resolution.

Question 38

Explain what is meant by leakage in data acquisition and explain its effect on the resulting frequency response function (FRF).

Answer 38

When a signal is not periodic within the sample window, the frequency component can be misrepresented and leaks into numerous other frequencies, where it is shown instead. This leads to errors / distortion in DFTS.

Question 39

How do windows applied to the time data help to reduce the effect of leakage?

Answer 39

For non-periodic samples which would leak and distort the DFT, windowing, where you zero the function at the start and end of the time period, forces the signal to be periodic and stops the signal leaking.

Question 40

Name 3 types of windowing function?

Answer 40

Boxcar (not to much use), Hanning window and Exponential Window.

Question 41

What equation does a hanning window function?

Answer 41

1/2 * (1 - cos (2πt / T)

Question 42

What is a displacement Frequency Response Function (FRF) also known as?

Answer 42

Receptance FRF

Question 43

Why is dB (decibel) scale used for FRF's?

Answer 43

Because the range of FRF amplitudes is very large.

Question 44

How do you calculate the FRF H(w), using the response function X(w) and the force F(w)?

Answer 44

H(w) = X(w) / F(w)

Question 45

Why can't you calculate the FRF through H(w) = X(w) / F(w), with broadband measurements?

Answer 45

Broadband measurements give low amplitudes of w and therefore have higher noise.

Question 46

What are power spectral density and cross spectral density and how can they lead to response estimator?

Answer 46

They are bother statistical functions that show how the power of two signals is distributed across a frequency spectrum. Looking at the ratios of power and spectral density, gives us two different estimator (H1 and H2) which can be compared.

Question 47

What is coherence?

Answer 47

Its is a measure of noise within FRF. By comparing the two separate response estimator H1 and H2 which are obtained through power and cross spectral density, the ratio of these signals gives us a coherence value between 0 and 1. A higher coherence value means the two estimators are identical and random noise is not affecting results.

Question 48

What are the two SDOF methods for acquiring a modal frequency and damping ratio within EMA?

Answer 48

1. Peak picking method 2. Circle Fit Method

Question 49

What does it mean when an EMA method is called SDOF method?

Answer 49

SDOF methods only operate on a single mode at a time, assuming that neat the modal frequency that mode dominates. SDOF methods depend on the assumption that modes are well spaced in terms of frequency. If two modes are similar in frequency, they will interact and these methods cannot be used.

Question 50

Given a receptance frequency response function graph (frequency on the x-axis and H(w) in dB on the y-axis), explain how to calculate the damping ratio using the peak picking method.

Answer 50

(Refer to given equation sheet): 1. Get wd by reading from the graph. (Its the point on the axis below the largest peak). 2. Find the half power points w1 and w2 by reading 3dB below the peak on the y-axis, and finding the corresponding x-axis points. 3. Use the damping ratio SDOF equation in the datasheet (ratio = (w2 -w1) / 2*wd )

Question 51

What difficulties can arise with the peak picking method for lightly damped modes when the frequency resolution is poor?

Answer 51

Lightly damped modes have very sharp peaks and the measured frequency point may not be exactly at resonance, leading to an error in estimating the value of the FRF peak.

Question 52

Given a Nyquist plot, explain how to calculate the damping ratio using the circle fit method.

Answer 52

You'll have a circle plot where the points are not evenly spaced. 1. Locate the 2 adjacent plotted points on the circle which are furthest apart from one another. These are as w_b and w_a. 2. Calculate the angle between these two points (theta), geometrically. 3. Calculate wd, as the mean average of w_b and w_a. 4. Use the damping ratio equation in the data sheet for the circle fir method ( ratio = (w_b - w_a) / (2*wd*tan(theta/4))).

Question 53

How do you read mode shape from a Nyquist Plot with multiple circle?

Answer 53

The amplitude of Nyquist plot circle is given by the circle diameter, therefore the mode shapes are simply vertical vectors of the different amplitudes.