Project Title:

Methods for Evaluating the Predictive Accuracy of Structural Dynamic Models

04.10-2551
Methods for Evaluating the Predictive Accuracy
	of Structural Dynamic Models
Engineering Mechanics Associates, Inc.
3820 Del Amo Boulevard, Suite 318
Torrance, CA 90503
Timothy K. Hasselman  (213-370-2551)
JPL -- NAS7-1020

Abstract:

	 Large space structures, such as the NASA Space Station, which cannot be fully tested
in a ground test laboratory, require both ground testing and on-orbit identification
of as-built structures. This project addressed ground-testing procedures which are
critical to the success of on-orbit identification.  Innovative methods for evaluating
and improving the predictive accuracy of structural dynamics models were investigated
during the Phase I study. This effort identified a new methodology for evaluating
uncertainties in mass, stiffness, and damping and how these propagate forward and
backward in order to evaluate the accuracy of response predictions and the uncertainty
in physical parameters.
	Three different methods for propagating uncertain-ties--first order statistics,
fuzzy set theory, and Monte-Carlo simulation--were examined. An approach which combines
the first two appears to be an efficient, cost-effective approach for bounding the
level of modeling uncertainty. Key elements of the methodology were demonstrated
using a realistic model of the NASA Space Station "Block 1" configuration. Experimental
error due to substructure and sub-scale testing were simulated and compared using
the above methodology.

Potential Commercial Application:

Potential Commercial Application: Applications exist in all areas where analytical
models are relied upon to predict structural performance which cannot be directly
verified by testing. Examples include off-shore structures, nuclear power plants,
high-rise buildings, and numerous applications in the aerospace industry.