Uncertainty law in ambient modal identification---Part II: Implication and field verification

• The asymptotic uncertainty laws derived in the companion paper are discussed qualitatively.
• Guidance on field test instrumentation are given based on the uncertainty laws.
• The uncertainty laws are verified using field test data.

This paper presents a qualitative analysis of the uncertainty laws for the modal parameters identified in a Bayesian approach using ambient vibration data, based on the theory developed in the companion paper. The uncertainty laws are also appraised using field test data. The paper intends to provide insights for planning ambient vibration tests and managing the uncertainties of the identified modal parameters. Some typical questions that shall be addressed are: to estimate the damping ratio to within 30% of posterior coefficient of variation (c.o.v), what is the minimum data duration? Will deploying an additional accelerometer significantly improve the accuracy in damping (or frequency)? Answers to these questions based on this work can be found in the Conclusions. As the Bayesian approach allows full use of information in the data for given modeling assumptions, the uncertainty laws obtained in this work represent the lower limit of uncertainty (estimation error) that can be achieved by any method (Bayesian or non-Bayesian).