Beginning the Process:
According to our motion model, removing weights has a complicated effect because it decreases both the mass and moment of inertia (see the page "super cool discoveries" for more detail). For the configurations we tested, the model predicted that the effect of the removal of the mass and the decreased moment of inertia would cancel and result in almost the same critical airspeed, which is what we observed in our experimental data.
We found in our experiments that for each of the configurations we tested, below a certain critical airspeed, oscillations had net positive damping and decayed to the trivially stable zero amplitude case after an initial disturbance. Above the critical airspeed, oscillations in the system would not decay because of negative net damping. This pattern of behavior is also what was predicted by the mathematical model.
For each of the three experimental tests, this critical flutter speed was 22 m/s. Our model predicted the speeds would be 28 m/s, giving an error of about 20%. Our model also predicted that the frequencies of these configurations would be between 37 and 45 Hz. In our experiments, we observed a frequency of 32 Hz.
Our model predicted the qualitative effects of changing the mass and moment of inertia of the configurations very well. Of note is that the model predicted the effects of mass and moment of inertia would cancel out such that the flutter speed is the same for each of the configurations, and this is what happened in the experimental results. Discrepancies in the quantitative predictions of flutter speed and frequency may be due to any of the limitations noted in the "model limitations" section of the "motion model" sub-page. The over-prediction in frequency may be because our model neglects large-displacement effects and higher-order damping effects (e.g. structural damping and aerodynamic drag), and the over-prediction of flutter speed may be because our model assumes all of the mass is subject to both pitch and plunge (which is accurate for a real aircraft, but less accurate for our experimental setup because of the mass of the linear sliders).
Next Steps
For potential (hypothetical) future testing, we would run the same types of tests as we did during this project, but with a larger range of masses. With the masses we used, we discovered that the frequencies between tests were all very similar, if not the exact same at specific points. This was great for validating our calculator, but did not allow us to see a large range of flutter phenomena at different frequencies.
We would also like to use a better accelerometer. The one we had worked in theory for the tests we ran, but it was only able to read up to 8g, and it was not able to record data as quickly as we personally would have liked. Therefore we would like to have an accelerometer with a larger range and quicker read time.
In respect to design changes for our setup, the overarching design was adequate; although, there are some design changes that could improve the setup. Firstly, we could make the rig design applicable for a larger range of masses in order to see more of the phenomena that our model predicted. For future research, we could also include updating the test rig and dealing with general wear-and-tear complications that have occurred naturally through age and use. This potential step would also include making the rig overall easier to use, since as of now it is somewhat hard to reach the bolts and nuts that are needed to change the weights on the metal rod.