Statistical analysis of monitoring data for Daniel Johnson dam

Dam safety is always a high priority issue since failures result in severe property damage and loss of life. Many countries require that all existing dams need to be inspected and evaluated regularly. Monitoring instruments have long been an integral part of the dam safety process by providing information for detailed analyses as well as for monitoring the current condition of dams. In this thesis, statistical models are used to infer behavior characteristics of a multiple arch dam. First, a statistical model known as the Hydrostatic-Season-Time (H-S-T) model is fitted to each individual displacement component from pendulum data for the Daniel Johnson Dam, a multiple arch concrete dam located in Northern Quebec. Principal component analysis (PCA) was used with the data from pendulums. A modified PCA analysis was also performed after removing thermal effects. A C++ language program (DASOD) was developed for the estimation of the H-S-T model, visualization of results, and data preparation for the principal component analysis. (Abstract shortened by UMI.