NJH.PORTFOLIO
CUBESAT IMAGER DYNAMIC FEA
In support of a Cal Poly senior project, titled UV Imager Application for a CubeSat, FEA analysis was
conducted on a portion of the satellite proof-of-concept. The purpose of the analysis was to verify the
dynamic behavior of a portion of the optical assembly in the satellite with fragile components, which is
called the Lens Assembly, such that the dynamic response of the satellite to launch conditions can be
verified and studied through FEA.
Four Analyses were done to further develop the model and predict the behavior seen from accelerometer data when placed on shake table:
STATIC
This was a somewhat of a dummy analysis to get a mesh convergence and gives insight into what mesh sizes and types would be best for the dynamic analysis before running something so computationally intensive. The analysis had a 1MPa load on the circular donut face of the of the lens holder and fixed the base. It was run for varying mesh sizes and as shown in the convergence section, gave a convincing result around a mesh size of 2mm.
FREQUENCY EXTRACTION
This analysis step was both to add insight into the model itself and support future steps with modal information. It extracted all of the natural frequencies and mode shapes of the model using eigenvalue calculations. Lanczos method was chosen as the solver because of its robustness on all software architectures and is the default solver in Abaqus.
Shown is the 50th mode shape. It is at 1750 Hz and most likely associated with the peak response of the accelerometer on the donut face.
RESPONSE SPECTRUM
The initial intent of this analysis was to get a response of the model subjected to base excitation over the spectrum of frequencies of interest: 5-2000Hz, but instead gave the peak response from the entire spectrum. This is still useful information, but did not validate the data because the experimental data could have been showing a different mode than the peak mode.
STEADY-STATE DYNAMIC MODEL
Finally, a steady-state dynamic model was run, to give the output caused by a base excitation. Again, this was not data one would hope for as the base excitation could not be defined fully. Ideally an acceleration boundary condition with an amplitude would have been applied to the model, but a sinusoidal sweep is not easy to create as an amplitude in Abaqus. The easiest next solution involved a user-subroutine to load in the acceleration data of a sine sweep. No action was taken to further develop this analysis since this seemed to be beyond the scope of the project.
