Zoned deformation prediction model for super high arch dams using hierarchical clustering and panel data

Abstract

Purpose The purpose of this study is to develop and verify a methodology for a zoned deformation prediction model for super high arch dams, which is indeed a panel data-based regression model with the hierarchical clustering on principal components. Design/methodology/approach The hierarchical clustering method is used to highlight the main features of the time series. This method is used to select the typical points of the measured ambient and concrete temperatures as predictors and divide the deformation observation points into groups. Based on this, the panel data of each zone can be established, and its type can be judged using F and Hausman tests successively. Then hydrostatic–temperature–time–season models for zones can be constructed. Through the comparative analyses of the distributions and the fitted coefficients of these zones, the spatial deformation mechanism of a dam can be identified. A super high arch dam is taken as a case study. Findings According to the measured radial displacements during the initial operation period, the investigated pendulums are divided into four zones. After tests, fixed-effect regression models are established. The comparative analyses show that the dam deformation conforms to the natural condition. The factors such as the unstable temperature field and the nonlinear time-dependent effect have obvious effects on the dam deformation. The results show the efficiency of the proposed methodology in zoning and prediction modeling for deformation of super high arch dams and the potential to mining dam deformation mechanism. Originality/value A zoned deformation prediction model for super high arch dams is proposed where hierarchical clustering on principal component method and panel data model are combined.