An efficient method for transient data recovery from large DOF FE models

A direct result of the increased computing abilities of desktop engineering workstations is the higher density of nodes with which a finite element model of a spacecraft is defined. Spacecraft usually have more sensitive appendages as compared to the launch vehicle and therefore require more definition for accuracy in the solution. A drawback to higher fidelity models is the increased amount of CPU time, memory, and data storage required to perform any dynamic analyses: specifically coupled loads analysis. UAI/Nastran partially overcomes this problem by its substructuring tool where large models can be dynamically reduced down to their boundary grids including internal points where data may be required. The drawback to this method is that if element quantities such as force and stress are required, Nastran requires that the nodal data at every time step for the unreduced model be saved for element data extraction. In this paper, a system model containing a spacecraft and launch vehicle totaling over 130,000 degrees of freedom is run through a transient solution containing 8001 time steps. For the substructuring method to work, the disk storage requirement would have to be in excess of fifty gigabytes to contain all nodal data. A specialized code using the DMAP language for UAI/Nastran and based on the Craig-Bampton methodology provides the required data without such large storage requirements.