Vibration testing has become a routine task in the condition assessment of built civil structures. However, the uncertainties involved in the identified modal parameter (e.g., modal frequencies, damping ratios, mode shapes) are still mysterious to most practitioners. A systematic treatment of uncertainties in vibration test will be introduced in this talk to demystify the related concepts, including their computation, understanding, and management. For computation, the fast Bayesian FFT method will be discussed in detail for both known and unknown input cases. Following the Bayesian framework, it allows the uncertainty quantified via the posterior distribution of modal parameters and in an efficient manner. The recently developed theory of uncertainty law will then be briefly overviewed. It analytically explains how the identification uncertainty depends on the test configuration and provides the first chance to understand the identification uncertainty for both known and unknown input cases. Based on uncertainty laws, the uncertainty management is finally considered with the target to minimize it, by optimizing the test configuration, e.g., the number and location of sensors/actuators. It is formulated as a decision-making problem under uncertainty, and thus can be solved consequently. Various applications of high-rise buildings and long-span bridges are included in each section, illustrating the performance and validating the theory. Uncertainties provide a unique perspective to understand the vibration problem of civil structures and can serve the engineering practice from various ways.
