I would like to compare the Solidworks Simulation software and reality on a frequency simulation. So I started by doing a hands-on experiment:
I placed a 73x99x1.66 (mm) aluminum plate on a vibrating pot plated on a base and I glued an accelerometer in its center to detect the natural frequencies of the aluminum plate. So following my experience, the plate vibrates for a frequency of about 1860 Hz.
I wanted to find this result on Solidworks simulation by a frequency analysis. After recreating the aluminum base and plate, I started my analysis but the results are not at all satisfactory. Indeed, the first eigenmode gives me a frequency of 2236 Hz. So I have a 20% gap between reality and simulation.
I tried to modify my materials, the dimensions of the plate, the fixings to the base but I don't fall back on the practical values and I'm out of ideas.
If some have already done this kind of study, could he tell me if their results were conclusive?
Thank you very much
A few clarifications:
- The plate is made of 5086h111 aluminum
- The vibrating pot used cannot go above 2000Hz because afterwards, the pot enters into resonance which distorts the measurements
I have already made the solidworks tutorials and the simulation tutorials of the site, I manage to launch to make my mesh correctly and to solve the analysis but the problem remains the same.
I tried to do the same simulation with the Comsol Multiphysics software and I came back to the same results with an accuracy of a few percent
I performed the same analysis with both software and I find a resonant frequency of 2236hz with solidworks and a frequency of 2204Hz with Comsol so for me, the problem comes either from the calculation by the two software or from experience.
I'm going to do the simulations again with another assembly to see if I find the source of the errors.
So I started my experiment again with another smaller case and with a different material.
Unfortunately, it's even worse... I have a 30% error between simulation and practice. For me, the practical experience is well done, it's at the level of simulation that it gets stuck.
I thought of something else, do you know if solidworks simulation performs these calculations in a vacuum or with air? Because it modifies the resonant frequencies so it may be the cause of my mistakes.
I searched in the solidworks options and I haven't yet found if we can modify the environment so if you know if we can change these conditions, it would be very useful to me.
There are always discrepancies between the experiment and the simulation, but generally only a few percent. 20% is a lot. In any case, with us, the design office often does model recalibration after test.
Just 2 leads that come to mind like this:
- When you say that the sheet metal is plated on the baseplate, does it mean that it is in contact on a whole side? Did you model this contact in your simulation as well?
- Your plate being very light, have you modeled the mass of the accelerometer? Since the frequency is a function of K/M, if the mass increases, the frequency decreases.