Treatment of model uncertainties in structural dynamics

Abstract

The uncertainties related to the modeling of the dynamic behavior of a structure are analyzed using a probabilistic approach. First, the case of preliminary design is addressed, where the structure has not yet been built. A new efficient and accurate numerical method is proposed to investigate the resulting uncertainties in the structural response due to uncertainties in the modeling process, where engineering judgement is used to quantify the latter uncertainties. Second, the case where records of measured structural response are available to reduce the uncertainties in the structural models is addressed. The posterior probability distribution of the uncertain parameters is found to be very peaked at the values of some optimal parameters. These optimal parameters can be obtained by minimizing a positive- definite measure-of-fit function. A new efficient minimization algorithm is proposed to resolve difficulties in convergence of existing methods. The identifiability of the optimal parameters is also addressed. The problem of finding the whole set of the optimal models that have the same output at the observed degrees of freedom is resolved for the first time, by presenting an algorithm which methodically and efficiently searches the parameter space.

Files

Additional details