An integrated approach to uncertainty and global sensitivity analysis in penstock structural modeling

Enhanced penstock structural models significantly advance hydropower engineering, yet their increasing complexity introduces challenges. As model interactions intensify, predictability and comprehensibility decrease, complicating the evaluation of model accuracy and alignment with operational performance metrics and safety standards. This issue is particularly pronounced in dynamic modeling, where knowledge gaps hinder straightforward validation via observational data. Traditional techniques for model calibration and validation are becoming impractical, necessitating a strategic approach to prioritize sources of uncertainty related to critical response variability. This study aims to advance our understanding and management of structural variabilities in penstock models by developing a comprehensive, step-by-step Global Sensitivity Analysis (GSA), designed to meet the specific needs of penstock modeling. Illustrated through a free vibration analysis model of a penstock span, this structured methodology begins with Uncertainty Analysis (UA) to identify variabilities, followed by a screening phase using the Morris method to enhance computational efficiency. Subsequent application of multi-method GSA ranks parameter sensitivities, assesses robustness, and provides a comparative evaluation, providing insights into the effective tools for conducting GSA on penstock models. The process culminates with Regional Sensitivity Analysis (RSA), targeting local sensitivities and enhancing understanding of local parameter influences, thereby supporting model adjustments and design optimization. Results from this application characterize model sensitivities for the most prominent mode shapes to specific structural parameters and indicate that these sensitivity outcomes are influenced by variability in parameter spaces and output sub-range definitions. This study provides a practical framework for addressing uncertainties in penstock design, enhancing model accuracy, prioritizing parameters, managing risks, and improving the reliability and efficiency of hydropower infrastructure.