A shift in mode sequence can happen in even flat plates. So this is normal. If the model and the object show different behaviour, then the model data or type of elements may be incorrect. A twiseted blade is probaly a shell structure. So is the shell nature included in the model?

Anders Buen thanks a lot for the feedback, yes the shell nature is included in the model, the only matter that I found it weird is that the 3rd mode associated amplitude is higher than the second one, while what I now and was expecting is due to energy it should be the inverse so as long as frequency increase the amplitude decrease. the error between the numerical and experimental is 2% only

Dear Mohammed,

1) Can you upload the first, second, and third mode shape analysis images (vertical, horizontal, twisting, longitudinal?)

2) I understand that you are doing finite element analysis. In the FEA analysis, what did you enter for the support stiffness, unit volume weight, elastic modulus, and what is the unit as amplitude? What is amplitude unit in experimental modal analysis?

Depending on the geometry of the structural element you are analyzing, the amplitude of any mode shape (i.e., FRF) will differ from the other anyway. The damping it shows against vertical bending is different, the damping it shows against horizontal bending is different, and the effect it shows on twisting is different. Therefore, the FRF magnitude decreases as damping increases, and vice versa...

Mohammed Lamine Mekhalfia The amplitude of a resonance mode is dependant on the damping and the effective mass. The damping can be different from mode to mode. Damping is one of the factors that usually is input to the model, unless a model for the damping is included. Then either the damping may be lower for your third mode, or the modal mass can be lower or both.

Anders Buen I wanted to share some recent findings from my testing involving accelerometers on a concrete beam. Specifically, I utilized the roving accelerometer method. Following the importation of acceleration and hammer data into Artemis Pro software and the subsequent acquisition of the first mode shapes, I have noticed that the mode shapes are not as smooth as desired.

For this particular 2-meter beam, I employed 22 measurement points. Considering my focus on damage detection, I am uncertain whether the observed noise stems from the measurement process or is indicative of structural damage. Consequently, I am contemplating whether it would be advisable to remove the noise before importing the data into the Modal analysis software (Artemis).

Your expertise and guidance in this matter would be greatly appreciated.

Fahime Sokhangou If there are cracks in your beam, it may reveal itself as increrased damping for certain modes. This was the primary test for roof tiles if they had cracks or not. Tapping them could reveal that. If you have ever experienced a coffee cup with a crack in it, and you can compare their tapped sounds, you may hear that the cracked one may ring less than the healthy.

The mode shapes may also be affected. You could compare the data from a perfect beam and one ,or several, with defects. You could add the defects mechanically in places they use to appear.

My experience with modal analysis is that sometimes the signal can be affected by sharp peaks of power supply harmonics, 50Hz or 60 Hz and harmonics. So I run the computer on its battery to avoid that. Modern power supplies can be really "dirty" and noisy. And "ground noise" loops can be difficult to remove with many cables in a room.

Anders Buen Thank you so much. I think you are talking about electronic noise in the system that sometimes we see it as a peak in PSD plot.

Do you have any experience in the Roving hammer method? how many hits do you do for exciting each point?

1- one hit for each point

2- 5 or 6 hits

which technic you use for filtering the noise?

Fahime Sokhangou i have experience with the rowing hammer method using miniature transducers on violins and smaller plates. I use five hits usually.

I do not filter any noise intnetionally, I make sure there is no noise entering the files. By turning off the power supply and run the computer and system on a battery.

Regarding filterning, it may make sense to cut the highs you do not need to avoid aliasing. That is: a low pass filter. Now, we rely on audio sound so the soundcard and software usually takes care of this.

Anders Buen I do not have an updated data acquisition system,

the instruments that I have are 2 accelerometer and one modal hammer but I am sure that I do not have electronic noise.

May I know which software you use for the modal analysis and if it is MATLAB code, is it possible to share it?

Thanks

Fahime

Fahime Sokhangou there is always a noise floor in any measurement system. It is rather important to know at what level it is.

Anders Buen May I know which software you used for the experimental modal analysis?

Fahime Sokhangou I use George Stoppanis modal analysis and vibration asessment programs. It is developed for use on violins and musical instruments and is kind of freeware. Stoppani is a violin maker and have good contact with Jim Woodhouse former Cambridge professor and Colin Gough, a professor emiritus in experimental physics. They run courses on the progrm yearly (in june) at the VSA Oberlin Violin Acoustics workshop in Oberlin Ohio, USA.

Anders Buen Thanks a million for such interesting information.

http://www.stoppani.co.uk/Technical.htm

I found his website.

You used the same program?

Does it work or my case that it is roving hammer, with 2 reference accelerometer?

Thanks in advance

Fahime

A presentation by George on modal analysis: https://youtu.be/5Gk7SkGVcQw

Anders Buen Thanks a million

Mohammed Lamine Mekhalfia

A (continuous) structure has theoretically an infinity of modes. The importance of a mode should be based on its modal gain rather than on its frequency. The modal gain represents the energy of a mode in response to a pulse excitation. A good finite element program delivers not only the frequencies but also their modal gains (calculated from the modeshapes and the mass matrix). See the literature for this. Good luck in your research.