1 Since the candidate DOF will be written to ASET cards, only those DOF that are not otherwise eliminated by Nastran’s normal pre-processing can be assigned to the candidate set. For example dependent DOF and SPC’d DOF cannot be in the candidate set since they are excluded from the analysis set by Nastran’s pre-processing. If any disallowed DOF are included on ASET cards Nastran will fail to solve.
2 The grid point kinetic energy is a standard Nastran output.
3 GA is compatible with any type of TAM reduction as long as the characteristic matrices are written in DMIG format to the punch file. However, because the Pseudo-Orthogonality calculation is performed at the FEM-size a Guyan back-expansion is used to expand the GA TAM-size modeshapes to the candidate TAM size. If a candidate TAM is loaded that is not a Guyan TAM it is recommended that a MAC cost function is used. For more details on the PO calculation see the GA OPTIONS FORM section.
4 In general the candidate set TAM should be a good enough representation of the full FEM so most mode numbers will correspond to each other; for example full FEM mode 12 will correspond to candidate TAM mode 12. However for modally dense models the one-to-one correlation may not hold. In these cases, for example, full FEM mode 12 may correspond to candidate TAM mode 11.
5 The Seed TAMs generated during the first generation are used to seed the initial population with accelerometer sets that are derived from standard deterministic algorithms. All other accelerometer sets in the first generation are randomly generated.
6 Full documentation of this alter can be found in ATA Engineering's TAMKIT documentation.