Real Flashcards
1) How do you determine where to place observation points? What should you think about when choosing measurement points? Should you excite close together or far apart(in time)? Is there any difference in method if you want to find eigenfrequencies or modes? How do you choose if you want to move the accelerometer a lot or the excitation point a lot?
The fundamental requirement is that the structure should be time invariant. If this is not the case it may destroy the possibilities to get reliable modal analysis results. If the more advanced modal parameter extraction procedures are used, even a small mass variation may cause divergent results. For this reason it is customary to use as many simultaneous accelerometers as possible to get consistent measurements and save valuable measurement time.
Roving excitation… gives 3 rows in FRF matrix. Risk of missing modes due to nodal lines in fixed response point is small. A modal analysis based on the measured FRFs only gives mode shape info perpendicular to the object surface.
Roving triax response… gives one FRF matrix col w/ 3x as many FRFs as in one of the 3 rows for alternative 1. Risk of missing modes with node lines at fixed excitation point is larger. Fill info on observation points response in all directions obtained though.
For complex geometries we cannot intuitively know where the modes are, so there is a risk of missing some mode therefore might be good w/ 2 analyses.
If only a dual channel analyser available…acc or excit must be moved to get all data needed. If acc moved the test structure mass distribution changes b/w the measurements …violation of the fundamental requirement that the structure remain time invariant(may ruin modal analysis results) small mass variation can cause divergent results….instead better to use simultaneous accs to get consistent measurements and save measurement time!
If interested in modes at high frequencies it can be good to excite in short pulses(slå tät, alltså slå ofta i en kort period) b/c
2) Difference b/w shaker and hammer?
Which is better when it comes to noise?
HAMMER: often fast and loading of object is kept at minimum (duration of pulse). First choice. Depending on the test structure you can use a soft or hard tip which determines the upper frequency limit. Different pulse times and different force-frequency spectrums. Heavy test structure requires heavy hammer. Hard to control excitation spectrum.
ELECTRODYN SHAKER: Flexible frequency range and excitation force. Its more complicated to use and more time consuming. Using white noise yields the optimum linear approximation to a non-linear relation between the excitation and the response.
Shaker is better when it comes to noise, otherwise excitation hammer is better. This because we will average out all our data after, the shaker will produce more averages than hammer simply because we can’t physically hit the structure as many times as the shaker can.
3) How do you choose how to suspend an object?
- Name different alternatives for suspension.
- Provide characteristics of the different alternatives you chose.(advantages/disadvantages)
- How is your choice of suspension dependent on the objects geometry?
…No external forces other than the one required by the BCs act on the structure. aka we do not want reaction forces! And we dont want damping from for example the mat.
…We want it to be able to move as freely as possible(we don’t want to suppress any modes)
…The structure should be vibrationally isolated from the environment. The structure must be supported so that the contribution from the reaction forces are negligible. Elastic strings or very weak elastic foundation or an elastic slab.
-Need to choose a good k value of our springs… Eigen freqs of the rigid body modes should be less than 1/3 of the lowest frequency you’re interested in
…ratio between the mass and the stiffness should be low
If acc moved the test structure mass distribution changes b/w the measurements …violation of the fundamental requirement that the structure remain time invariant(may ruin modal analysis results) small mass variation can cause divergent results….instead better to use simultaneous accs to get consistent measurements and save measurement time!
5) You have been requested to analyse a geometry for a company in FE
- What input data should you focus on?
- How do you determine number of acceleration positions?
- How are you sure you’ve got all the information you need? How much data is enough? …Hur många svars frekvens funktioner behöver du för att avgöra om du fått med alla moder?
- Do you need more or less accelerometers if you are interested in higher or lower frequencies?
-Should conduct testing to find resonance frequencies, damping and mode shapes. Use those as input for the FE model
-You can use roving excitation or roving triaxial response.
-Can only know if our measurements are NOT reliable/valid, so never sure if enough.
You can miss something if you excite a node line or if there is movement in several directions.
-Need more accelerometers for high frequencies! Something with the shape being smaller, not “caught” by only 1 acc, resolution in the geometry..
- complex geometry or unsure? go w/ several accs
- Can miss something you you hit on a node line or if there is movement in several directions(??)…it maaaay be enough with one acc(in 3 directions) for a complex geometry if you know about the node lines etc…or just one direction is even enough if it is something like a plate where movement really only comes up in one direction
- Make sure to hit several times in every excitation point to be able to take a good average later
6) How can you determine the reliability/accuracy of your modal estimation result
Checking COHERENCE is the first step; control over how linear the relationship between input data and output data is (aka the exciting force and the FRF). If it is equal to 1 ==> linear relationship, which is positive but still not necessarily a guarantee. (Get it by comparing mode shape vectors(?))…
reciprocity, auto mac, cross mac, animation(mode shape approx)
- REAPEAT the experiment
- also mention approximation and error measurement/ residuals as well!
- CALIBRATION before and after
For example if you have a dual accelerometer you need to move it in between tests and even small mass changes to the structure can have influence (structure no longer time invariant). Reliable FRF measurements require experience and deep knowledge. Do calibration. If the structure is part of a larger structure the BC’s may change, keep this in mind. What approximations have you made and what errors do you know?
- in almost all cases good results also require long experience from FRF measurements on similar structures.
- If the structure is part of a larger structure it may change the BC:s. Bending moment.
9) Calibration procedure of FRF… Why do you have to do it? When do you have to do it/when is it maybe not necessary? Vad har calibreringen för inverkan på mode shape vektorer, vad är det händer då?
Complete calibration implies calibration of all parts of the measurement system …provides absolute values of force and motion
Use a reference object(known mass), m should be equal to F/a
Important that the force is applied in the measured DOF only(Other directions would contribute to the measured excitation but not the response)…straight rod often chosen since its 1D character makes it difficult to excite rotation
for freqs lower than 30% of the first quasi-longitudinal eigen freq the mass behaves as a rigid mass, oscillating back and forth on its suspension
Perform at beginning and end to assure it’s reliable
other
- F and response must have same time history
- Transducers give electric output signal prop. to the force loading
- Strain gauges better for torque, bening, shear..
- Det. signal: good for nonlinear prop., accurate, noise/signal ratio better(less noise), no leakage probs
- 1,2 or both signals can contain noise…. need to estimate FRF/the linear relationship
- Force more controlled and more free from noise than response measurement
- Averaging retains only correlations b/w F and a (infinite avgs actually..)
- Exp window for transient signals… forces signal + derive to 0 by synthetically damping aka speed up reverberation time
- Component alone… different BCs… vibration props change… results not representative
- FRFs determined w/ multichannel FFT analysers… common time base needed, then antialias, A/D converter, weighting window, auto and cross spectrum… FRF
- A/D discretises signal… gives resolution error…resolve w/ a high pass filter (helps when there are contrib at low freqs from rigid body motion is large)
- not periodic… leakage errors. periodic… DONT window(will introduce leakage)
- freq res increases w/ measurement time (0-padding! (transient signals))
- seismic mass increase… F increases… sensitivity increases(limits useful freq range tho)(also mass loading is bad..)
- G_AB = 0… A&B not correlated
- Only periodic signals will be correctly transformed to the freq domain
- sampling/AD conversion… spectrum is folded around multiples of 1/2 the sampling frequency
