Candidate Set DOF

 

The first step of model preparation is to select a candidate set of degrees-of-freedom (DOF) that will be used to create a Test-Analysis Model (TAM). The candidate set of DOF is the set of all potential accelerometer locations from which GA will select a subset to be the final set of accelerometer locations.

All DOF in the candidate set should be both accessible and measurable for the test. Since only translational accelerometers are typically used in modal testing all rotational DOF should be excluded from the candidate set. Additionally, some FEM DOF may correspond to locations on the structure that are not accessible for accelerometer placement so these inaccessible DOF should also be excluded from the candidate set. Nastran ASET cards should be created for all DOF in the candidate set.[1]

Except for FEMs with the smallest number of DOF, including all accessible and measurable DOF in the candidate set is not good practice. Having too many DOF will not only be extremely computationally expensive, but it will detrimentally affect the quality of the accelerometer set derived by the GA. The candidate set should be limited to DOF that are “good” potential accelerometer locations: those locations with either large mass or large response for the target modes. This ensures that GA’s randomly selected accelerometer locations are picked from a pool of “good” locations, which would more likely represent the target modes as being linearly independent. As a rough rule of thumb the candidate set should be about 5-10 times larger than the desired number of accelerometers, although this range is only an approximate guideline. Currently it is suggested to keep candidate sets to less than 2000 DOF, although this limitation is based on the memory available on the computer running Matlab.

A simple method to rank the importance of DOF for a particular set of modes is to examine the grid point kinetic energy.[2] Other more sophisticated Nastran-based methods can be used. An example is the Iterative Residual Kinetic Energy (IRKE) process which starts with a small set of DOF and iteratively adds DOF to minimize the residual kinetic energy while providing a measure of the effectiveness of each iteratively added DOF. This process is implemented as DMAP alter fast_iter_rke and is available with TAMKIT. A more detailed description of the IRKE implementation is available in the DEMO section.