I am currently developing a numerical model to observe the effect of bolt preload on ultrasonic signal propagation across a bolted interface. Since, bolt load and PZT patch modeling is only available in Dynamic implicit analysis, I applied the bolt load in a general static model and then imported the deformed geometry to model ultrasonic signal propagation in a dynamic explicit model.
However, in the transmitted signal, I am observing a dominant low-frequency vibration along with the ultrasonic signal. I was able to extract the actual signal by applying a bandpass filter around the signal frequency.
For the excitation of the signal, I used pressure at the patch area, and the transmitted signal was recorded with out-of-plane displacements.
Do you have any insights into why I am observing this low-frequency component? This phenomenon was not present when I performed a similar analysis on a single plate.
For the boundary condition, I had fixed the ends of the plate with ENCASTRE constraints.
Attached is the transmitted signal and its filtered version.
I'm not completely clear on the query here, but it seems like the bolted surface boundaries may be causing turbulence, which then leads to a sort of amplitude modulation that results in lamb waves at the lower frequencies at lower harmonics, or a sort of turbulent resonant phenomena (as circular boundaries may create an omnidirectional dispersion path in all directions, hence causing the low frequency lamb waves)
Mostly speculation though. just my 2c
It kind of figured a way out. Since i imported the deformed geometry from the bolt preload model, it didnt have stress information at the information and the interface will vibrate during the signal propagation, so i fixed the contact surface between the bolt head and the plate similarly between the nut and the bottom plate with ENCASTRE boundary condition. Got rid of lower frequency, hope it still make sense for other preloads.
First, I would suggest recording the out-of-plane velocities instead of displacements. Then, the wave propagation modeling can also be finished with dynamic implicit steps, in this manner you needn't import the deformed geometry separately. Finally, just record the signals in the second analysis step. It is quite reasonable that there are mixing waves in the output signals since it's a two-step analysis process (most pre-load problems) and vibration from the first bolt load step would not vanish, so proper signal processing is required (as you mentioned getting rid of the low-frequency components) to explain and extract your results. Hope these suggestions would help.
- Badges
- Science topic
- Similar topics
- Physical Sciences
- Waves