HOME    »    PROGRAMS/ACTIVITIES    »    Annual Thematic Program
Talk Abstract
Uncertainty in Mode Shape Data and its Influence on the Comparison of Test and Analysis Models

Robert V. Lust and John A. Cafeo
Vehicle Analysis and Dynamics Lab
General Motors Research & Development and Planning
Warren, Michigan

In the design and development of automobiles, laboratory modal tests are typically used to improve and verify finite element analysis models that support design decisions. We know from experience that there is variation and uncertainty associated with both the test and the simulation results. This variation (or uncertainty) complicates the comparison of test and simulation results and, if not considered, can result in unnecessary attempts to reconcile the differences between the test and analysis models. It can also lead to erroneous design decisions.

In this paper we discuss a general view of variation and uncertainty as it relates to finite element analysis model correlation and reconciliation. Within this framework, and as a part of an experiment designed to quantify test-to-test variability in the modal analysis of automobiles, we propose a generic model for uncertainty in the components of experimental mode shapes. The uncertainty in the mode shape components is characterized in terms of a probability distribution and its associated parameters. After developing the uncertainty model we demonstrate its utility through an application to a typical test/analysis data correlation exercise. Specifically, we demonstrate and discuss the use the uncertainty model as a basis for comparing test and analysis mode shapes using the Modal Assurance Criteria (MAC).

Back to Workshop Schedule

Back to Decision Making under Uncertainty: Assessment of the Reliability of Mathematical Models

1999-2000 Reactive Flow and Transport Phenomena