NASTRAN OPTIMIZATION SENSITIVITY
Discrepancies between the predicted responses of a finite element analysis (FEA) and reference data from test results arise for many reasons. Some are due to measurement errors, such as inaccurate sensors, noise in the acquisition system or environmental effects. Some are due to analyst errors precipitated by a lack of familiarity with the modeling or solver software. Still others are introduced by uncertainty in the governing physical relations (linear versus non-linear behavior), boundary conditions or the element material/geometrical properties. It is the uncertainty effects introduced by this last group that this study seeks to redress. The objective is the obtainment of model improvements that will reduce errors in predicted versus measured responses. This technique, whereby measured structural data is used to correct finite element model (FEM) errors, has become known as “model updating”. Model updating modifies any or all of the mass, stiffness, and damping parameters of a FEM until an improved agreement between the FEA data and test data is achieved. Unlike direct methods, producing a mathematical model representing a given state, the goal of FE model updating is to achieve an improved match between model and test data by making physically meaningful changes. This study replaces measured responses by reference output obtained from a FEA of a small spacecraft. This FEM is referred to as the “Baseline” model. A “Perturbed” model is created from this baseline my making prescribed changes to the various component masses. The degree of mass variation results from the level of confidence existing at a mature stage of the design cycle. Statistical mean levels of confidence are assigned based on the type of mass of which there are three types: • Concentrated masses – nonstructural, lumped mass formulation (uncoupled) • Smeared masses – nonstructural mass over length or area, lumped mass formulation (uncoupled) • Mass density – volumetric mass, lumped mass formulation (uncoupled) A methodology is presented that accurately predicts the forces occurring at the interface between the spacecraft and the launch vehicle. The methodology quantifies the relationships between spacecraft mass variations and the interface accelerations in the form of sensitivity coefficients. These coefficients are obtained by performing design sensitivity /optimization analyses while updating the Perturbed model to correlate with the Baseline model. The interface forces are responses obtained from a frequency response analysis that runs within the optimization analysis. These forces arise due to the imposition of unit white noise applied across a frequency range extending up to 200 hertz, a cut-off frequency encompassing the lift-off energy required to elicit global mass response. The focus is on lift-off as it is characterized by base excitation, which produces the largest interface forces.
If this is your thesis or dissertation, and want to learn how to access it or for more information about readership statistics, contact us at STARS@ucf.edu
Catbas, F. Necati
Master of Science (M.S.)
College of Engineering and Computer Science
Civil and Environmental Engineering
Length of Campus-only Access
Masters Thesis (Open Access)
Braswell, Tom, "Spacecraft Loads Predictionvia Sensitivity Analysis And Optimization" (2007). Electronic Theses and Dissertations. 3095.