Accurately predicting damping in the design phase of high-rise buildings is essential for reliable assessments of occupant comfort. To improve damping prediction models for Dutch high-rise buildings, it is important to assess the reliability of damping estimates derived from measured vibration responses. This thesis aims to evaluate the performance of the Random Decrement Technique ranked by peak amplitude (Peak RDT) in reliably estimating damping, for systems exhibiting constant and amplitude-dependent damping, excited by stationary and non-stationary loading conditions. Since this assessment requires a known ground truth, a series of numerical studies are conducted, producing a distribution of damping estimates.

Under stationary loading conditions, the mean damping estimates are accurate for both systems. However, the precision of damping estimates is significantly reduced at low and high relative amplitude ranges. At low amplitudes, the reduced precision in damping estimates arises because the Random Decrement Signatures (RDS) fail to accurately capture the system’s free decay, as segments are sampled at very low response levels. At high amplitudes, the lack of precision in damping estimates is primarily due to insufficient segment counts in the computation of the RDS, which prevents proper isolation of the system’s free vibration response.

Moreover, RDS corresponding to the system with amplitude-dependent damping exhibit a non-uniform decay, causing fitting errors when using non-linear Least Squares Minimization to estimate a constant damping value. These errors can be mitigated by reducing the length of the segments sampled by the Peak RDT algorithm.

Three types of non-stationarity are introduced in the excitation: time-varying mean, time-varying variance, and their combination. While the second case yields level-stationary response signals complying with the underlying assumptions of the Peak RDT, the other two introduce time-dependent non-zero means in the response data, leading to significant accuracy errors in mean damping estimates for both linear and non-linear systems.

In conclusion, the Peak RDT yields reliable damping estimates for systems with constant and amplitude-dependent damping only under stationary loading conditions. However, it must be acknowledged that the degrees of non-stationarity considered in this thesis is rather large. The Peak RDT might still produce reliable damping estimates for mildly non-stationary vibration responses.