A methodological framework for defining dynamic control limits in hydropower dam instrumentation systems

Hydropower dam monitoring is key to safety, the environment, and operational management. In this context, this study proposes a methodology to define dynamic control limits for time series data from concrete dam instrumentation, enhancing dam health monitoring. Traditional static limits often become outdated during dam operation. Thus, the proposed framework adjusts thresholds based on structural behavior over time. The proposed methodology consists of two modules. The first is devoted to performing time series analysis and forecasting for dam sensors, and the second, based on the most suitable forecasting model, is to obtain control charts with dynamic control limits. For this purpose, statistical and machine learning models are used in the first step. These charts include exponentially weighted moving average (EWMA) and cumulative sum (CUSUM) for residuals, and a shift detection EWMA (SD-EWMA) for forecast-based control. A case study using piezometer data from the F-05/06 block of the ITAIPU Dam demonstrated promising results, supporting broader application to other monitoring instruments. • Control process is proposed for hydropower dam monitoring. • Statistical and machine learning models are used. • Dynamic control charts are considered for forecasting. • Two real datasets are used to validate the proposed approach. • Operational safety and efficiency are support by the proposed framework.