Modal identification of arch dams using balanced stochastic subspace identification

Tarinejad, R; Pourgholi, M

Tarinejad, R (reprint author), Univ Tabriz, Fac Civil Engn, Tabriz, East Azerbaijan, Iran.



Dynamic characteristics extracted from ambient and forced vibration tests are always associated with some level of uncertainties because of unknown nature of applied forces, existence of ambient noises as well as measurement errors. Stochastic Subspace Methods are among the most accurate and consistent methods within the domain of operational modal analysis. In this research, a new technique for Operational Modal Analysis is proposed using Stochastic Realization Theory and Canonical Correlation Analysis, which in comparison with previous methodologies, identification process are directly performed in the prediction space by extracting orthonormal vectors of data space. Considering its optimized nature, the proposed method is expected to have superior accuracy in terms of elimination of unstable poles as well as low time consumption analysis. To indicate the efficiency and accuracy of the proposed algorithm, it is applied to reanalyze the results of the forced vibration tests performed on the Shahid-Rajaee arch dam in northern Iran. These tests were conducted via steady-state sinusoidal stimulation method. More accurate natural frequencies are obtained compared to those of previously reported results, besides the fact that the first three modes of the structure were identified by the new approach, while they were not observed via the previous one. In order to examine the capabilities of the proposed method for processing of ambient vibration records, the dynamic characteristics of the Pacoima dam was identified using the recorded responses during 2001 earthquake in San Fernando, California. The results indicated good accuracy in the obtained frequencies and damping ratios compared to those obtained via data driven subspace method. Time consumption of identification process were reduced significantly (up to 50%) for both case studies indicating a faster convergence rate provided by the proposed method.

Download PDF

Full Text Link