Analysis of periodicities in long-term displacement time series in concrete dams

Abstract

The hydrostatic-season-time (HST) model is a widely used method in the safety assessment of dams and for the estimation of reversible and irreversible deformations due to different load scenarios. After implementing an optimal HST-model to observational data, the residual time series can still expose some underlying periodicities. In our contribution, the underlying periodicities and their contribution to residual minimisation are further analysed by the Lomb–Scargle normalised periodogram frequency method. The extended HST-model, obtained by adding additional sinusoidal terms for the statistically most significant frequencies, presents a slight statistical improvement on the optimal HST-model. Statistically significant frequencies may not be justified or correlated to some physical process of the dam, but they make it possible to identify a pattern in the analysed period. The analyses are performed for the radial direction of long-term displacement time series measured at the topmost reading station of an inverted pendulum system in the central cantilever of the Alqueva concrete arch dam